Datasets:

gabrielaltay
/

pmcoa

Dataset card Data Studio Files Files and versions Community

Split (2)

train · 4.94M rows

text stringlengths 87 880k	pmid stringlengths 1 8	accession_id stringlengths 9 10	license stringclasses 2 values	last_updated stringlengths 19 19	retracted stringclasses 2 values	citation stringlengths 22 94	decoded_as stringclasses 2 values	journal stringlengths 3 48	year int32 1.95k 2.02k	doi stringlengths 3 61	oa_subset stringclasses 1 value
==== Front J Med Imaging Radiat Sci J Med Imaging Radiat Sci Journal of Medical Imaging and Radiation Sciences 1939-8654 1876-7982 Published by Elsevier Inc. S1939-8654(22)00539-2 10.1016/j.jmir.2022.10.160 Article ASCERTAINING THE RELATIONSHIP BETWEEN THE RADIOLOGIC TECHNOLOGISTS’ STANDARD PRECAUTION PRACTICES AND ATTITUDES IN HANDLING COVID-19 PATIENTS Torio Mark Anthony G. 1 Rillera Arnold D. 2 Codizal Emmanuel L. 3 1 College of Radiologic Technology, University of Perpetual Help System DALTA- Molino Campus, Bacoor City, Cavite 4102 Philippines 2 College of Radiologic Technology, University of Perpetual Help System DALTA- Molino Campus, Bacoor City, Cavite 4102 Philippines 3 College of Radiologic Technology, University of Perpetual Help System DALTA- Molino Campus, Bacoor City, Cavite 4102 Philippines 2 12 2022 12 2022 2 12 2022 53 4 S49S49 Copyright © 2022 Published by Elsevier Inc. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Introduction Standard precautions are part of the infection control basis to protect radiology technologists from COVID-19 and other infections to prevent transmission from patient to patient and from healthcare workers to healthcare workers. This study ascertains the relationship between the standard practices and attitudes of radiologic technologists in handling COVID-19 patients. Methodology A quantitative-correlational design as used to specifically determine the standard precaution practices and attitudes of 46 radiologic technologists in the affiliated hospitals of the university, and the relationship of the two variables. Google form was used to gather data using the WHO Standard Precautions Protocol, which underwent pilot testing and a Cronbach alpha value of .804. Data analysis included descriptive statistics, Analysis of Variance (ANOVA) and Pearson-r correlation. Results The study revealed that radiologic technologists’ have high level of practices (μ=2.97 ±.0694) and attitude (μ=2.97 ±.824) in standard precaution. No significant difference was revealed in the standard precaution practices of radiologic technologists when grouped according to gender, years of service, age, and hospital type, Furthermore, no significant difference was revealed in the attitudes of radiologic technologist when grouped according to years of service, age, and hospital type, however, a significant difference was observed when the radiologic technologists were grouped according to gender (t (46) =2.04, p=.048) with female radiologic technologists having higher attitude (x̄=2.99, SD=.0253) on standard precautions than male radiologic technologists (x̄=2.94, SD=.0118). Overall, a high-positive correlation (r= .855, p=0.001) was revealed between the standard precaution practices and attitudes of radiologic technologists in handling COVID-19 patients. Conclusion There exist a good practice and good attitude in standard precaution of Radiologic technologists in handling COVID-19 patients and a very high-positive relationship between the two variables. Keywords Radiologic technologist standard precaution practice attitude COVID-19 Philippines ==== Body pmc	0	PMC9716011	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S57	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.187	oa_other
==== Front Food Control Food Control Food Control 0956-7135 0956-7135 Elsevier Ltd. S0956-7135(21)00490-4 10.1016/j.foodcont.2021.108352 108352 Article Did the COVID-19 lockdown affect consumers’ sustainable behaviour in food purchasing and consumption in China? Li Shanshan a∗ Kallas Zein bc∗∗ Rahmani Djamel b a Institute for Research in Sustainability Science and Technology (IS-UPC), Polytechnic University of Catalonia, 08034, Barcelona, Spain b CREDA-UPC-IRTA, Centre for Agro-food Economy & Development, 08860, Castelldefels, Spain c DEAB (Department of Agrifood Engineering and Biotechnology-Universitat Politècnica de Catalunya), Castelldefels, Spain ∗ Corresponding author. Institute for Research in Sustainability Science and Technology (IS-UPC), Polytechnic University of Catalonia, 08034, Barcelona, Spain. ∗∗ Corresponding author. CREDA-UPC-IRTA, Centre for Agro-food Economy & Development, 08860, Castelldefels, Spain. 15 6 2021 2 2022 15 6 2021 132 108352108352 7 4 2021 31 5 2021 13 6 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The COVID-19 pandemic situation has altered consumers’ behaviour in food purchasing and consumption. This study, as a first attempt, assesses how the COVID-19 lockdown affects Chinese consumers’ purchasing and consumption behaviour from a sustainability point of view. To reach this objective, a semi-structured questionnaire is designed, collecting data from 1006 participants. The food purchasing behaviour towards the importance of sustainable attributes (P), sustainable and healthy diets (D), and food waste (W) as three dependent variables are measured, and three binary logistic regressions are estimated. The results suggest that gender and age are relevant factors affecting sustainable behaviour. Household size has a significant effect on the healthy diet shift and food waste reduction. Risk attitude has a negative and significant impact on the sustainable purchase decision. In addition, consumers’ food security, financial, and health risk perceptions are highly important factors in understanding consumers’ sustainable purchasing and consumption behaviour. Consumers’ subjective and objective knowledge levels regarding COVID-19 influence consumers’ sustainability shift during the lockdown. The findings provide some practical implications for policymakers and stakeholders to carry out more socially acceptable policy actions that ensure consumers’ sustainable purchasing and consumption behaviour during the COVID-19 pandemic. Keywords COVID-19 lockdown Food purchasing behaviour Food consumption behaviour Sustainable food consumption Binary logistic regression Food waste Healthy diets ==== Body pmc1 Introduction A novel coronavirus (SARS-CoV-2) caused a newly emerged respiratory disease named by the World Health Organization as COVID-19 (COronaVIrus Disease 2019), which was first reported in Hubei Province, China in December 2019 and then spread across China and worldwide (Velavan & Meyer, 2020). Many countries began to conduct restrictions starting by washing hands with alcohol-based hand sanitisers, wearing face masks, decreasing social activities and ending with a complete lockdown. As a result, citizens were asked to stay home, and mobility was only justified for essential journeys, such as going to medical centres, buying food (e.g., going to grocery stores) or going to essential work, which affected their food buying and consumption behaviour (Ruiz-Roso et al., 2020). The pandemic situation altered consumers’ dietary habits, and the uncontrollable stress caused by the pandemic played an important role in affecting consumers’ eating patterns (Yau & Potenza, 2013). For example, stress made people towards overeating, especially “comfort food” high in sugar, defined as “food craving” (Rodríguez-Martín & Meule, 2015) due to their effect in reducing stress (Ma et al., 2017). However, many consumers switched to a healthier and balanced diet during the COVID-19 lockdown in order to maintain a correct nutrition status and reduce health risks (Di Renzo et al., 2020; Sidor & Rzymski, 2020). For instance, in Spain, consumers tended to have healthier dietary habits during the COVID-19 pandemic (Rodríguez-Pérez et al., 2020). Also, some Chinese adults shifted to a healthier diet by increasing their consumption of vegetables and fruit than the situation before the restrictions (Wang, Lei, et al., 2020). This dietary change is also related to the time available for preparing meals. The COVID-19 situation brings individuals to spend more time cooking and trying new recipes (Sidor & Rzymski, 2020). However, although many restaurants were closed during the lockdown, restaurant-to-consumer delivery was still optional for consumers, and the order could be made directly through the restaurant’s online platform (e.g., KFC, McDonald’s, and Pizza Hut) or via a third-party platform (e.g., Eleme platform in China) (Li et al., 2020). Previous studies have shown that food and dietary choices can affect the environment in different ways, such as climate change (GHGE greenhouse gas emissions), land, water and energy use, and biodiversity (Macdiarmid, 2013). Meat and dairy products contributed the most to these emissions (Garnett, 2008), while fruit and vegetables contributed less (low environmental impact). The 2010 Food and Agriculture Organization (FAO) defined sustainable diets as those diets with low environmental impacts which contribute to food and nutrition security and healthy life for present and future generations (Food and Agriculture Organization of the United Nations, 2010). According to their recommendation, a more sustainable diet included consumption of fresh ingredients, more seasonal foods, especially fresh fruit and vegetables, and less red and processed meat and salt. The study of Duchin (2005) showed that a healthy diet is rich in fresh fruit and vegetables, low in meat, and low in added sugar, salty snacks and saturated fatty acids. Thus, an increase in a healthier diet was regarded as more sustainable consumption in this research. In fact, there have been many studies showing a trend for consumers’ purchasing behaviour towards more sustainable attributes of food before the COVID-19 pandemic. These include consumers seeking more local, animal welfare, fair-trade, organic, seasonal, and carbon footprint food products (Codron et al., 2006). There is no doubt that the crisis caused by the COVID-19 pandemic has changed consumers’ behaviour to buy even more of these food products with sustainable attributes. Previous research conducted in Catalonia (Spain) indicated that consumers’ preferences for the local food products were enhanced during the economic crisis from 2008 (Escobar et al., 2018). Therefore, exploring the rationale for consumers’ purchasing behaviour towards food products with sustainable attributes during the COVID-19 lockdown will allow policymakers and multi-agents stakeholders to carry out and design more socially acceptable policy actions that ensure the production and retail of food with sustainable attributes during the COVID-19 pandemic. Consumers’ food behaviour consists of the food product journeys: planning, purchasing, storage, preparation, and consumption of food, and food waste is an outcome of the way households deal with these different stages (Kim et al., 2019). At the buying stage, consumers often rely on food shopping routines and admit to regularly buying more food than needed (Evans, 2011) or buying food products they never use or seldom use, thus increasing the likelihood of food being spoiled and discarded as waste. In fact, one-third of the food produced globally for human consumption is lost or wasted (Food and Agriculture Organization of the United Nations, 2011). Food waste has negative environmental, economic, and social consequences for the sustainability of the food sector (Aschemann-Witzel et al., 2015; Garnett, 2011). More food that is wasted is a measure of less sustainability. The COVID-19 lockdown measures forced consumers to stay at home, resulting in a higher possibility of panic buying at grocery stores, which may cause food waste (Pappalardo et al., 2020). In addition, food waste increased due to the broken supply chains (e.g., absence of labour and food items getting stuck on the road due to restrictions on vehicle movements), which caused millions of products to rot in the fields (Sharma et al., 2020). One of the major goals of the European societies is their quest for sustainable development (Abeliotis et al., 2010), and proper food production and consumption is a major component of overall achievable sustainability (Annunziata & Scarpato, 2014). A growing number of consumers are beginning to realize the importance to have more sustainable consumption as it can be an effective way to solve sustainability problems (Annunziata & Scarpato, 2014; de Bakker & Dagevos, 2012). In the last decades, many researchers have studied consumers’ sustainable consumption (de Bakker & Dagevos, 2012; Gallenti et al., 2016; Lorek & Spangenberg, 2014). Because there are fewer studies on consumer sustainable consumption in Asian (emerging economy) countries compared with those in European (developed economy) markets, it seems appropriate to understand consumers’ sustainable behaviour in China in general, and specifically to analyze how the COVID-19 lockdown has affected such behaviour. We believe that the following research is the first study to explore such an impact. In order to examine Chinese consumers’ consumption and purchasing behaviour from a sustainability point of view, several specific objectives were proposed as follows: a) changes in buying food products with sustainable attributes as a sustainable purchasing behaviour; b) the effect of the lockdown on the sustainability of behavioural decisions using changes in consumers’ diets (whether they adopted a healthier food choice or not) as a sustainable consumption shift; and c) how food waste behaviour changed during the COVID-19 lockdown. 2 Material and methods 2.1 Data collection Data were collected from 1006 consumers in China using a semi-structured questionnaire in an online environment (Wenjuanxing platform, similar to Qualtrics) in June 2020, two months after the lifting of the Wuhan lockdown (April 8, 2020), when the sanitary situation was gradually returning to be normal. The starting point of our sampling procedure was using a non-probabilistic sampling method (Quota sampling) where we divided the sample size into mutually exclusive subgroups based on known quotas of gender, and a selection criterion to be eligible was selected. Thus, data were collected by promoting the link to the questionnaire in several electronic media following the snowballing procedure by sending it to students’ contacts (e.g., family and relatives), consumer associations or organizations, local municipality contact and public institutions (e.g., consumers/citizens issue office), and personal and institutional social network (WeChat, similar to WhatsApp and Facebook; Weibo, similar to Twitter; and other Chinese social networks) and asked them to share with their contacts (e.g., family and relatives). Snowball sampling could reduce costs and time, achieve higher response rates, and expand samples with different professions, places, genders, and ages (Baltar & Brunet, 2012; Benfield & Szlemko, 2006). Only consumers who were totally or in part responsible for food purchasing were included in the study. Respondents who volunteered to participate in the survey and received an explanation of the objectives of the study were told that the information received would be exclusively used for research and that their confidentiality would be honoured. At the same time, we offered incentives (5 RMB) to the participants through the Wenjuanxing platform to boost completion rates. That is, when they complete the questionnaire (finish answering the last question of the questionnaire), they will see and get incentives to thank them for participating in the survey. The questionnaire was approved by the Ethics Committee of the Centre for Agro-food Economy and Development and was conducted according to the ethical principles in social science studies. 2.2 Measuring sustainability of consumers’ purchasing and consumption behaviour In order to measure how the COVID-19 lockdown influences consumers’ purchasing and consumption behaviour from a sustainability point of view in China, three dependent variables were defined, consisting of identifying sustainable purchasing behaviour ( P ) (purchasing food products with sustainable attributes), sustainable and healthy diets adoption ( D ), and food waste behaviour change ( W ). Consumers’ change in purchasing behaviour ( P ) towards sustainable attributes was identified, based on the following food selections: local, animal welfare, fair-trade, and organic food. Changes were measured to determine whether consumers’ purchasing behaviour during the COVID-19 lockdown became more or less sustainable. Respondents were asked about the change in sustainable attributes of food purchasing behaviour ( P ) scoring from “-3” (decreased a lot) to “+3” (increased a lot) (How has the importance of the following attributes changed for you during COVID-19? - Local; animal welfare; fair-trade; organic). Regarding sustainable consumption, it was measured using a proxy variable representing changes in a healthy diet adoption ( D ). An increase in a healthier diet was considered a tendency to more sustainable consumption. As previously noted, according to the FAO and previous study (Duchin, 2005), a more sustainable diet may include consumption of fresh ingredients, more fruit and vegetables items, and less salt. Thus, changes in the consumption of these food items were considered as a proxy for a sustainable and healthy diet in this research. A synthetic index was created reflecting the increase in the consumption of healthy food products. Individual scores from “-3” (decreased a lot) to “+3” (increased a lot) of healthy food items (fruits, vegetables, fresh food, less processed food, low in calories, low in fats, low in sugar, low in salt, high in fibre, and high in calcium) were summed up. Furthermore, individual scores of unhealthy food items (sweets, chocolate, candies, and snacks) were also summed up after reversing the scale. Finally, the food waste ( W ) variable was measured by asking respondents about the change in food waste at home scoring from “-3” (decreased a lot) to “+3” (increased a lot) (How has COVID-19 impacted the amount of food waste in your home?). In the case of the mentioned changes to be investigated ( P, D, and W ), the binary logistic model was used to analyze the behavioural changes before and during the lockdown. The first one comprised respondent who stated that the lockdown caused an increasing change in behaviour analyzed ( Y = 1 ) compared to those they didn’t ( Y = 0 ). Previous research demonstrated that risk attitude and risk perception could affect consumers’ purchasing and consumption behaviour (Zhu & Deng, 2020). The preponderance of the literature concludes that consumers’ knowledge influences their food purchasing behaviour. For example, some studies show that knowledge plays an important role in the purchase of organic products (Hill & Lynchehaun, 2002; Wang et al., 2019). In addition, health concerns and food security concerns influence consumers’ attitude and ultimately influence purchasing behaviour towards organic food (Basha et al., 2015). Consumers’ buying and consumption intentions and behaviour may change due to their experiences during the COVID-19 lockdown. Therefore, the independent variables were those noted as potentially relevant factors and were presented as follows:(1) Socio-demographic variables presented in Table 1 ;Table 1 Socio-demographic variables in the research. Table 1Socio-demographic variables Numbers (%) Gender Male 507 (50.4) Female 499 (49.6) Age 18–39 years 631 (62.7) 40–59 years 367 (36.5) More than 60 years 8 (0.8) Average age (years) 36.5 Monthly household income (before the lockdown) <5,000 RMB 149 (14.8) 5,001–15,000 RMB 371 (36.9) >15,001 RMB 279 (27.7) Missing 207 (20.6) Monthly household income (during the lockdown) <5,000 RMB 286 (28.4) 5,001–15,000 RMB 335 (33.3) >15,001 RMB 176 (17.5) Missing 209 (20.8) Stated health status Unhealthy 272 (27.0) Healthy 734 (73.0) Household size 1 person 22 (2.2) 2 persons 67 (6.7) 3 persons 241 (24.0) 4 persons 326 (32.4) 5 persons 186 (18.5) 6 persons or more 164 (16.3) Sample size 1006 Note: 1 RMB = 0.13 euro = 0.14 US dollar (at the time of writing this paper). (2) Multiple Price List (MPL) stated risk attitude; (3) Risk perception (including health, financial, and food security risks); (4) Consumers’ health concerns level about COVID-19; (5) Experience (food shortages, food price increases, and neither); (6) Subjective knowledge level regarding COVID-19; (7) Objective knowledge level regarding COVID-19. Fig. 1 summarizes the different determinant factors finally included in this study to understand consumers’ shift in the sustainability behaviour of food consumption and purchasing in China.Fig. 1 Factors affecting consumers’ sustainable behaviour. Fig. 1 2.2.1 Multiple Price List stated risk attitude: the lotteries approach Risk perception and risk attitude can exert an influence on behavioural intention (Zhu & Deng, 2020). In this context, the consumers’ stated risk attitude is measured by adopting a multiple price list (MPL), known as “Lotteries”, introduced by Holt and Laury (H&L) (Drichoutis & Lusk, 2016). All definitions of “risk” included two characteristics. One was related to uncertainty, and the other one was its consequences. The simplest definition of risk was “uncertainty that matters”, since uncertainty without consequence poses no risk (Hillson & Murray-Webster, 2006). “Risk attitude” was defined as “consumers’ consistent choice tendency to face different risk levels” or “consumers’ willingness to accept risks” (Schroeder et al., 2007). Different risk attitude elicitation techniques were employed in previous research (Pennings & Garcia, 2001; Smidts, 1997), and the MPL was very popular in some experimental studies in psychology and economics (Harrison et al., 2007). The MPL is a relatively simple procedure for eliciting values from a subject (Anderson et al., 2007), and it is based on the economic theory of the expected utility (Orduño et al., 2019). Tversky (1995) indicates that one of the fundamental assumptions of the economic analysis of risk that is built into portfolio theory is the assumption of risk aversion. Analysts assume that, holding expected value constant, people would rather have a certain return than an uncertain return, and people need to be compensated for bearing risk and people exhibit inconsistent attitudes towards risk. Moreover, risk aversion is not always valid, especially in the domain of losses, where risk-loving is frequently observed. The point at which an individual switches from choosing one outcome over the other is often used as a measure of risk aversion (Drichoutis & Lusk, 2016). In the Lottery Game of this research, question 1: Option A is that individuals can be sure to get 200 RMB, while option B is flipping a coin. If the coin comes out HEADS, they will get 200 RMB, but if TAILS comes out, they will get nothing. Question 2: Option A is that individuals can be sure to get 190 RMB, while option B is flipping a coin. If the coin comes out HEADS, they will get 200 RMB, but if TAILS comes out, they will get nothing. By that analogy, 20 questions (until option A is 10 RMB, option B remains unchanged) are asked to measure consumers’ risk attitudes. This part of the questionnaire about risk attitudes will be over when respondents choose option B at any time. In the 20 questions, the payoff associated with option A declines systematically, while the payoff for option B remains unchanged (Brick et al., 2012). Table 2 displayed the payoff matrix from the risk attitude experiment in this research. Previous literature indicated that participants who were risk-loving would choose option B in the first lottery task, while risk-averse participants would choose option A in the second last row. A risk-neutral subject should switch from choosing A to B when the expected values (EV) of each are approximately the same, so a risk-neutral subject would choose A for the first four rows and B thereafter (Andersen et al., 2006; Brick et al., 2012; Harrison et al., 2005). In this research, a risk-loving person would choose option B in the first task, and a risk-neutral participant would choose option B from A in the eleventh task, meaning that a risk-neutral person would choose A ten times before switching to B. Risk-averse subjects would choose option A in the twentieth task. The number of “safe choices” (choosing option A) or the switching point from choosing A to B is often used to describe risk attitude (Lusk & Coble, 2005). According to expected utility theory, people should choose A from task 1 to 10 and choose B from task 11 to 20. The safe choices number of risk-loving people should be below or equal to 9, and the safe choices number of risk-neutral ones should be equal to 10. The safe choices number of risk-averse people should be more than or equal to 11. This research used this method to analyze the risk attitude variable.Table 2 Payoff matrix from the risk attitude lottery experiment. Table 2Task No. Option A Option B Expected values and difference P (￥) P (200￥) P (0￥) EVA (￥) EVB (￥) Difference (￥) 1 1 (200￥) 0.5 (200￥) 0.5 (0￥) 200 100 100 2 1 (190￥) 0.5 (200￥) 0.5 (0￥) 190 100 90 3 1 (180￥) 0.5 (200￥) 0.5 (0￥) 180 100 80 4 1 (170￥) 0.5 (200￥) 0.5 (0￥) 170 100 70 5 1 (160￥) 0.5 (200￥) 0.5 (0￥) 160 100 60 6 1 (150￥) 0.5 (200￥) 0.5 (0￥) 150 100 50 7 1 (140￥) 0.5 (200￥) 0.5 (0￥) 140 100 40 8 1 (130￥) 0.5 (200￥) 0.5 (0￥) 130 100 30 9 1 (120￥) 0.5 (200￥) 0.5 (0￥) 120 100 20 10 1 (110￥) 0.5 (200￥) 0.5 (0￥) 110 100 10 11 1 (100￥) 0.5 (200￥) 0.5 (0￥) 100 100 0 12 1 (90￥) 0.5 (200￥) 0.5 (0￥) 90 100 −10 13 1 (80￥) 0.5 (200￥) 0.5 (0￥) 80 100 −20 14 1 (70￥) 0.5 (200￥) 0.5 (0￥) 70 100 −30 15 1 (60￥) 0.5 (200￥) 0.5 (0￥) 60 100 −40 16 1 (50￥) 0.5 (200￥) 0.5 (0￥) 50 100 −50 17 1 (40￥) 0.5 (200￥) 0.5 (0￥) 40 100 −60 18 1 (30￥) 0.5 (200￥) 0.5 (0￥) 30 100 −70 19 1 (20￥) 0.5 (200￥) 0.5 (0￥) 20 100 −80 20 1 (10￥) 0.5 (200￥) 0.5 (0￥) 10 100 −90 Note: The last three columns in this table, which showed the expected values (EV) of the lotteries and their difference, were not shown to the participants. 2.2.2 Risk perception Risk perception refers to people’s judgments and evaluations of hazards they (or environments) are or might be exposed to. Such perceptions steer decisions about the acceptability of risks and have a crucial impact on behaviour before, during, and after a disaster (Rohrmann, 2008). As a consequence, three types of perceived risks, including health risk, financial risk, and food security risk, are measured in this research. Participants’ health risk perception is elicited using a 10-point Likert scale ranging from 1 (not serious at all) to 10 (very serious) (In case you contract COVID-19 in the next six months, how serious do you think your health condition will be?) and a 5-point Likert scale that ranges from 1 (very unlikely) to 5 (very likely) (How likely do you think it is that you will develop or contract COVID-19 in the next six months?). Respondents are asked to indicate the feelings about their current financial situation, including uncertainty, at risk, threatened, worried about it, and think about it, to measure their financial risk perception via a 5-point Likert scale ranging from 1 (not at all) to 5 (a great deal) (Please indicate how you feel about your current financial situation?). A higher point indicates a higher financial risk perception. In addition, consumers’ perceived food security risk is elicited using a 7-point Likert scale ranging from 1 (very unlikely) to 7 (very likely), and the questions are the possibility of perceived food shortages and food price increases in the next six months (Do you think the following scenarios are likely or unlikely in the next six months? -Food shortages; Food price increases). 2.2.3 Subjective and objective knowledge level regarding COVID-19 Knowledge is divided into what individuals perceive they know (subjective knowledge) and what they actually know (objective knowledge) (Brucks, 1985). Peschel et al. (2016) demonstrated irrespective of product or country under investigation, consumers who have higher subjective and objective knowledge levels tend to have a more environmentally sustainable food choice. Taufique et al. (2017) found that environmental and eco-label knowledge is positively associated with attitudes towards the environment and affects their pro-environmental consumer behaviour. As a result, consumers’ subjective and objective knowledge levels are measured to test their influence on consumers’ purchasing and consumption behaviour in this research. Specifically, respondents are asked to respond to their perceived subjective knowledge level via a 7-point Likert scale ranging from 1 (not knowledgeable at all) to 7 (very knowledgeable), and its result is presented in percentage terms ranging from 0 (not knowledgeable at all) to 100 (very knowledgeable) (Please indicate how knowledgeable you feel with regard to COVID-19). Respondents’ objective knowledge is measured by asking them to judge whether the symptoms of COVID-19 are right or false by including several existing symptoms and non-existent symptoms (True or False? These are common symptoms of COVID-19). Objective knowledge is expressed as the percentage of correct answers to questions of knowledge on seventeen statements. In addition, respondents’ discrepancy intensity between subjective and objective knowledge is also explored in this research. Knowledge discrepancy has two aspects: subjective knowledge level is higher than objective knowledge (overestimation), or subjective knowledge level is lower than objective knowledge (underestimation) (Khan et al., 2017). 2.2.4 Health concerns about COVID-19 and experience during the COVID-19 outbreak The COVID-19 pandemic is a challenge for global food supply chains. It may result in food shortages and food price increases in developing countries (Reardon et al., 2020). Consumers’ behaviour is sometimes designed to mitigate against the risk of not being able to purchase food, or indeed other items, at a later date for those who have experienced food shortages or food price increases during the COVID-19 outbreak (Power et al., 2020). Therefore, experience as an explanatory variable is elicited in this research (Did you experience the following scenarios? - You faced food shortages in your area during the COVID-19 outbreak; You experienced an increase in food prices; You experienced neither). During the COVID-19 pandemic, there are concerns about food security and health (Pu & Zhong, 2020), and they may influence consumers’ food purchasing and consumption behaviour. As a consequence, respondents’ level of health concerns about COVID-19 is also measured using a 7-point Likert scale ranging from 1 (not concerned at all) to 7 (extremely concerned) (Please indicate your level of health concern about COVID-19). 2.3 Empirical application Methodologically, this analysis is based on a binary logistic regression model using the IBM SPSS v.24 software. This model is often used when the dependent variable is a dichotomous variable to check out the factors that influence the odds ratio of the dependent variable (Serrano-Cruz et al., 2018). The logit model has the formula (Osborne & King, 2011):(1) Logit (P) = Log [Pi / (1 − Pi)] Where Pi is the probability of the event occurring (the probability of increasing food purchase and consumption in this research). 1 – Pi refers to the probability that respondents do not increase their food purchase and consumption. The odds ratio (OR) is the ratio of both previous probabilities. In this research, the logistic model of the relationship between the variable of food increasing or not and its explanatory variables is specified as follows:(2) ln [Pi / (1 − Pi)] = β0 +β1X1i + β2X2i + … +β14X14i Where the subscript i denotes the i-th observation in the sample. P is the probability of the outcome. X1, X2, X3, …, X14 are independent variables. β0 is the intercept term, and β1, β2, β3, …, β14 are the coefficients associated with each independent variable. The coefficients do not directly indicate the effect of changes in the corresponding explanatory variables on the probability (P) of the outcome occurring. Rather, the coefficients reflect the effect of individual explanatory variables on the OR of the dependent variable (Zakari et al., 2014). Thus, the model can be written in terms of OR as follows:(3) Pi /(1-Pi) = exp (β0 +β1X1i + β2X2i + … +β14X14i) 3 Results and discussion 3.1 Characteristics of the study participants A total of 1006 adults completed the questionnaire. As shown earlier, Table 1 displayed the characteristics of the respondents. Table 3 also presented some characteristics of the sample. The majority of the respondents were male (50.4%), 18–39 years old (62.7%), healthy (73.0%), with an average monthly household income of 5,001–15,000 RMB (36.9% and 33.3%), risk-averse (57.1%), with 4 persons in a household (32.4%), and who experienced an increase in food prices (51.3%). According to the gender distribution, the sample reflected the population of China.Table 3 Results of the factors affecting the sustainability of consumers’ behaviour. Table 3Variables Percentage Scale Knowledge regarding COVID-19 Subjective knowledge level 71.97% 0–100% Objective knowledge level 55.78% 0–100% Discrepancy intensity between knowledge 16.19% Experience Food shortages 8.40% Food price increases 51.30% Experienced neither 40.30% Risk attitude Risk-loving 32.90% Risk-neutral 10.00% Risk-averse 57.10% Mean(SD) Concerns about COVID-19 5.56 (1.42) 7-point Likert scale Health risk perception The severity of health condition will be if contract COVID-19 in the next 6 months 6.19 (2.66) 10-point Likert scale The probability of contracting COVID-19 1.99 (0.92) 5-point Likert scale Food security risk perception The probability of facing food shortages in the next 6 months 3.06 (1.62) 7-point Likert scale The probability of facing food price increases 3.78 (1.75) 7-point Likert scale 3.2 Factors affecting the sustainability of consumers’ purchasing and consumption behaviour Table 3 and Fig. 2 presented the results of the factors affecting the sustainability of consumers’ behaviour. The results demonstrate a high level of subjective and objective knowledge in China, with the values being above average (71.97% > 50.00% and 55.78% > 50.00%). In addition, the discrepancy intensity between knowledge is 16.19%, indicating that consumers believed that they know more than they really know (overestimation). This may be related to the first outbreak of the COVID-19 in Wuhan, China, and China has released sufficient information about COVID-19 to the society, increasing individuals’ confidence that led them to believe that they know more than they really know. This is supported by Pejman et al. (2019), who proved that when respondents receive sufficient information, their perceived knowledge will increase.Fig. 2 Factors affecting the sustainability of consumers’ behaviour. Fig. 2 As shown in Fig. 2, respondents’ concern level about COVID-19 is above average (5.56 > 3.5 points on a 7-point scale). With regard to the severity of the perceived health risk, it shows that the severity is above average (6.19 > 5 points on a 10-point scale). As for the probability of contracting COVID-19 in the next 6 months, the result indicates that Chinese consumers perceive a low likelihood (1.99 < 2.5 points on a 5-point Likert scale). This is because the Chinese are very confident in the measures adopted by the Chinese government and perceive a very low likelihood of contracting COVID-19, a very high likelihood of survival, and a high level of satisfaction with health information (Wang, Pan, et al., 2020). The value of the probability of facing food shortages in the next 6 months is below average (3.06 < 3.5 points on a 7-point scale), while the value of the probability of facing a price increase is above average (3.78 > 3.5 points on a 7-point scale). These are in line with the results of experience in this research that 51.30% of respondents experienced a price increase, while only 8.40% of them experienced food shortages, and thus they perceived a higher risk of a price increase. 3.3 Results of consumers’ sustainable purchasing and consumption behaviour 3.3.1 Changes in purchasing food products with sustainable attributes during the COVID-19 lockdown The results were presented as the coefficient (β), significance (Sig.), and Exp (β). Table 4 (model 1) listed the results of purchasing behaviour towards food with sustainable attributes. The fit was acceptable as indicated by Hosmer-Lemeshow’s goodness of fit measures and the percentage of correct classification. The result implies that, during the lockdown, females were 1.517 times more likely to increase the purchase of food with sustainable attributes than males when compared to the situation before the lockdown. It is consistent with the previous research which indicated that females are more proactive in the purchase and consumption of organic food than males due to their lifestyle (Olivas & Bernabéu, 2012; Ureña et al., 2008). It also demonstrates that respondents aged 40–59 years were more likely to purchase more food with sustainable attributes than those aged 18–39 years. It is in line with a study which showed that Chinese consumers aged more than 36 were more likely to buy certified organic food (McCarthy et al., 2014).Table 4 Logit model of purchasing food with sustainable attributes (P) (model 1). Table 4Variables B Sig. Exp (B) Gender Female 0.417 0.035 1.517 Age 40–59 years old 0.520 0.011 1.682 Risk attitude Risk-neutral −0.843 0.020 0.430 Risk-averse −0.404 0.056 0.667 Financial risk perception Think moderately about the current finance −1.453 0.054 0.234 Food security risk perception Neutral likely to face food shortages 0.707 0.097 2.028 A lot likely to face food shortages 1.009 0.093 2.743 Percentage of correct classification 68.8% Hosmer-Lemeshow’s goodness of fit 0.367 In addition, risk-neutral and risk-averse people were less likely to increase their purchase of food with sustainable attributes during the lockdown. One possible reason is that the purchase of food with sustainable attributes (e.g., organic food) is considered a risky choice, as consumers lack information and some are unfamiliar compared to conventional ones, such that they prefer the certainty of conventional products to the uncertainty of sustainable ones (Anderson et al., 2006; Smith & Paladino, 2010). Therefore, information campaigns in China could play an important role in promoting consumers’ current sustainable attributes information level, where the domestic market for food products with sustainable attributes is still at an early stage (von Meyer-Höfer et al., 2015). The result also demonstrates that respondents who perceived a higher financial risk (got a higher point on a 5-point Likert scale for the financial risk variable) were less likely to purchase more food products with sustainable attributes. Not surprisingly, food products with sustainable attributes were more expensive than conventional food (Bhaskaran et al., 2006), so consumers purchased less of these food products when they perceived a higher financial risk, and they would be more cautious about spending money during the COVID-19 lockdown. In addition, individuals who perceived a higher food security risk were more likely to purchase more food products with sustainable attributes. 3.3.2 Changes in the sustainable and healthy diets during the COVID-19 lockdown In Table 5 (model 2), the model had a percentage of correct predictions of 72.4%, and the Hosmer-Lemeshow’s goodness of fit was equal to 0.407. The null hypothesis was accepted, indicating that there were no differences between observed and model-predicted values (Maharjan & Joshi, 2011). Both tests pointed out that the model fitted well. The result reveals that females were 1.617 times more likely to increase sustainable and healthy diets than males when compared to the situation before the lockdown. This may be related to women’s attention to their weight that females seem to be more influenced by the current ideal of slimness, and thus they always attempt to reduce weight more often than males (Kiefer et al., 2000). Consequently, females tend to have a healthier diet than males. It also indicates that consumers who stated that they are in healthy conditions were less likely to increase sustainable and healthy diets than those who stated that they are unhealthy during the lockdown. The severe COVID-19 threatened consumers’ health ranges from asymptomatic infection to life-threatening and fatal disease (Del Rio et al., 2020), especially for unhealthy people. As a result, unhealthy consumers were more likely to have a healthier diet to boost their immune system and reduce the health risk. The result also implies that households with 4 members were 3.887 times more likely to increase sustainable and healthy diets than those living alone during the lockdown. Individuals living in larger households exhibited a higher possibility of adopting a healthier and more sustainable diet, especially those living with children and elderly people, and they had an increased likelihood of contracting COVID-19 (He et al., 2020).Table 5 Logit model of sustainable and healthy diets (D) (model 2). Table 5Variables B Sig. Exp (B) Gender Female 0.481 0.010 1.617 Stated health status Healthy −0.566 0.011 0.568 Household size Households with 4 members 1.358 0.061 3.887 Health risk perception Very likely to contract COVID-19 in the next six months 0.370 0.093 1.447 Food security risk perception A little likely to face food shortages 0.939 0.043 2.557 Neutral likely to face food price increases 0.715 0.081 2.044 Financial risk perception Feel uncertain slightly about the current finance 0.615 0.077 1.851 Feel uncertain moderately about the current finance 0.628 0.071 1.873 Think considerably about the current finance 1.034 0.089 2.812 Knowledge regarding COVID-19 Subjective knowledge level 2.061 0.001 7.853 Objective knowledge level 0.979 0.013 2.663 Percentage of correct classification 72.4% Hosmer-Lemeshow’s goodness of fit 0.407 In addition, people who perceived a higher health risk were more likely to have a more sustainable and healthy diet than those with the lowest health risk in order to boost immunity and reduce their health risk. Regarding food security risk perception, the estimates indicate that individuals who perceived a higher food security risk were more likely to increase the consumption of a sustainable and healthy diet than those who perceived a lower one. The reason may be that fruit and vegetables (healthy and sustainable food items) are much cheaper than meat and dairy products in China (Yu & Abler, 2009). Therefore, people who perceived a higher food security risk (perceived a higher likelihood of food price increases or food shortages in the next six months) may tend to spend less money on food (mainly buy fruit and vegetables) during the lockdown to prevent food price increases in the next six months and thus meet the needs of food spending in the future. Hence, they were more likely to adopt a healthier and sustainable diet. As for financial risk perception, consumers who perceived a higher financial risk (got a higher point on a 5-point Likert scale for the financial risk variable) were more likely to increase the consumption of healthy food than people who perceived a lower financial risk. Similarly, fruit and vegetables have a lower price than meat, resulting in an increase in consumption of fruit and vegetables and a decrease in consumption of red and processed meat, exhibiting an increase in a healthier diet, for those who perceived a higher financial risk. The result also reveals that consumers with higher levels of objective knowledge and subjective knowledge were more likely to increase the consumption of a healthy diet than those with lower levels of objective knowledge and subjective knowledge. It was expected that the more knowledge consumers had, the more severity about COVID-19 they perceive, such that they tend to increase the consumption of a healthy diet to reduce the risk of contracting COVID-19. 3.3.3 Changes in the total food waste during the COVID-19 lockdown Table 6 (model 3) presented the results of total food waste. The percentage of correct classification was 71.6%, and the value of Hosmer-Lemeshow’s goodness of fit was 0.058. It guided us to accept the null hypothesis, which meant that there was no difference between observed and model-predicted values. The result shows that consumers who stated that they are in healthy conditions were less likely to increase food waste than those who stated that they are unhealthy during the COVID-19 lockdown, which could be explained by the fact that unhealthy people may tend to over-shop to reduce the risk of contracting COVID-19, resulting in more food that spoils and is discarded as waste. The result demonstrates that households with 2, 3, and 4 members were less likely to increase food waste than those with one member, which is in accordance with the research of Fonseca (2013) conducted in Portugal showing that single consumers wasted more food. It also indicates that respondents who perceived a higher health risk and food security risk were more likely to increase food waste. Similarly, it seems that these individuals tended to purchase and stockpile more food during the lockdown, on the one hand, to prevent food shortages or rising food prices (food insecurity) in the future, making them unable to buy or afford the food they need. On the other hand, they could minimize trips to the grocery store to reduce the risk of contracting COVID-19 (health risk). However, stockpiling more food could lead to a lot of food that spoils and is discarded as waste. The findings also suggest that people with a higher financial risk perception were less likely to increase food waste. This ties in with Graham-Rowe et al. (2014) finding that many household food purchasers avoid wasting food for financial reasons because they think throwing away food is a waste of money.Table 6 Logit model of total food waste (W) (model 3). Table 6Variables B Sig. Exp (B) Stated health status Healthy −0.411 0.041 0.663 Household size Households with 2 members −1.731 0.023 0.177 Households with 3 members −1.324 0.058 0.266 Households with 4 members −1.364 0.048 0.256 Health risk perception Neutral likely to contract COVID-19 in the next 6 months 0.538 0.059 1.713 A lot likely to contract COVID-19 in the next 6 months 0.732 0.079 2.079 Food security risk perception A lot unlikely to face food shortages 0.719 0.047 2.053 Neutral likely to face food shortages 0.657 0.090 1.929 Financial risk perception Feel uncertain considerably about the current finance −0.679 0.095 0.507 Concerns about COVID-19 A little concerned −2.429 0.039 0.088 Extremally concerned −2.017 0.070 0.133 Knowledge regarding COVID-19 Subjective knowledge level 1.540 0.008 4.667 Percentage of correct classification 71.6% Hosmer-Lemeshow’s goodness of fit 0.058 Additionally, individuals who were more concerned about COVID-19 were less likely to increase their total food waste. In addition, people who think they know more about COVID-19 (with a higher level of subjective knowledge) were more likely to increase food waste. One possible reason is that the more consumers think they know about the virus, the more aware they are of the severity of the COVID-19, so they will reduce the health risk by buying more food, which probably leads to more food that spoils and is discarded as waste. 3.4 Summary of all models and practical implications Fig. 3 was drawn to make the results of all models easier to observe.Fig. 3 Results of all models Note: The y-axis represents the significant factors influencing consumers’ behaviour, and the x-axis shows each dependent variable. On the right side of the scale line of 0, it means a positive relationship between the independent and the dependent variables, while on the left side it means a negative relationship. Fig. 3 The complex issue of food consumption and purchasing behaviour requires a combination of several actions to be taken. Results demonstrate that a shift towards more sustainable behaviour is realistic and likely to occur. Thus, targeting measures according to different consumers’ characteristics and profiles can be designed, promoted, and applied not only during the pandemic situation, but also after the sanitary crisis to maintain the identified sustainable behaviour. The results indicate that females and consumers aged 40–59 years purchase more food products with sustainable attributes during the COVID-19 lockdown, reminding producer marketing tools to be focused on increasing purchase by improving sustainable food product availability and consumers’ access, especially for females and those aged 40–59 years. According to the findings, younger consumers aged 18–39 years are less likely to increase their purchase of food with sustainable attributes. As a result, it is necessary to increase these consumers’ knowledge about sustainable food products and consider how to differentiate them in the market during the lockdown. As for sustainable and healthy diets, males, healthy consumers, and people living alone are less likely to increase the consumption of a healthy diet, so the government can recommend through public communication campaigns with a specific focus on males, healthy consumers, and people living alone. Regarding food waste behaviour, Chinese consumers, the Chinese government, and stakeholders within the food chain should work together to reduce food waste. The results show that the household with 1 member wastes more food during the COVID-19 outbreak. Retailers can play an important role by offering food in small packages (suitable for a single person) to reduce waste. The results also suggest that people with a higher health risk and food security risk perception waste more food when compared to the situation before the lockdown. Therefore, more information about COVID-19 and food security should be delivered to consumers to reduce the perception of the risk of consumers’ health and food security (e.g., information on food availability and price stability), which could reduce their panic buying and food waste. In addition, consumers should also take the initiative to improve their awareness of environmental protection and avoid food waste. 4 Conclusion This research explored the factors in consumers’ sustainable purchasing and consumption behaviour during the lockdown in China and provided a reference for academic research. Monthly household income and experience (food shortages, food price increases, and neither) were not statistically significant factors affecting the food sustainable behaviour defined in this study. Females were found to increase their purchase of food with sustainable attributes and consumption of a healthy diet than males when compared to the situation before the lockdown. Age was only found to have a significant association with sustainable purchasing. People aged 40–59 years were more likely to purchase more food with sustainable attributes than those aged 18–39 years during the lockdown. Consumers who stated that they are in healthy conditions consumed less healthy diets and had low food waste during the lockdown. In addition, household size was found to have a significant effect on a healthy diet and food waste, which indicated that households with 4 members consumed a healthier diet and had less food waste than those living alone when compared with the situation before the lockdown. Risk attitude had a negative and significant impact on sustainable purchasing behaviour. Compared with the situation before the lockdown, the lockdown made risk-averse and risk-neutral consumers exhibit less sustainable attributes of food purchasing behaviour. Regarding health risk perception, consumers with a higher health risk perception increased their healthy diets and food waste than the situation before the lockdown. Consumers who perceived a higher food security risk tended to purchase more food with sustainable attributes, have a healthier diet, but with increased food waste behaviour due to the lockdown. Consumers who perceived a higher financial risk were less likely to increase the purchase of food with sustainable attributes and food waste, but more likely to increase sustainable and healthy diets when compared to the situation before the lockdown. Respondents who had low food waste during the lockdown exhibited higher health concerns about COVID-19. As for consumers’ knowledge regarding COVID-19, when compared to the situation before the lockdown, healthy diets and food waste increased with a higher subjective knowledge level, and healthy diets increased with the rising objective knowledge level. The main limitation of the research is that the sample size of people aged over 60 is relatively low (only 0.8%) due to the lack of access to smartphones or computers, which indicates that its result should be explained cautiously. With the COVID-19 pandemic going on, the results need to be further confirmed and investigated with a larger number of samples in future research. Future research could focus on consumers who are not familiar with online surveys. However, it enabled us to get data in a rapid and efficient way from different areas in China, avoiding face-to-face surveys due to the COVID-19 limitations. Author Contributions S.L: Investigation, Data curation, Formal analysis, Software, Visualization, Writing – original draft; Conceptualization, Methodology. Z.K: Validation, Writing – review & editing, Supervision, Project administration, Conceptualization, Methodology. D.R: Writing – review & editing, Conceptualization, Methodology. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declaration of competing interest None. Acknowledgements The authors thank two anonymous reviewers for their valuable comments which helped to considerably improve the manuscript. ==== Refs References Abeliotis K. Koniari C. Sardianou E. The profile of the green consumer in Greece International Journal of Consumer Studies 34 2 2010 153 160 10.1111/j.1470-6431.2009.00833.x Andersen S. Harrison G.W. Lau M.I. Rutström E.E. Elicitation using multiple price list formats Experimental Economics 9 4 2006 383 405 10.1007/s10683-006-7055-6 Anderson S. Harrison G.W. Lau M.I. Rutstrom E.E. Valuation using multiple price list formats Applied Economics 39 6 2007 675 682 10.1080/00036840500462046 Anderson J.C. Wachenheim C.J. Lesch W.C. Perceptions of genetically modified and organic foods and processes AgBioforum 9 3 2006 180 194 Annunziata A. Scarpato D. Factors affecting consumer attitudes towards food products with sustainable attributes Agricultural Economics 60 8 2014 353 363 10.17221/156/2013-agricecon Aschemann-Witzel J. de Hooge I. Amani P. Bech-Larsen T. Oostindjer M. Consumer-related food waste: Causes and potential for action Sustainability 7 6 2015 6457 6477 10.3390/su7066457 de Bakker E. Dagevos H. Reducing meat consumption in Today's consumer society: Questioning the citizen-consumer gap Journal of Agricultural and Environmental Ethics 25 6 2012 877 894 10.1007/s10806-011-9345-z Baltar F. Brunet I. Social research 2.0: Virtual snowball sampling method using Facebook 2012 10.1108/10662241211199960 Internet Research Basha M.B. Mason C. Shamsudin M.F. Hussain H.I. Salem M.A. Consumers attitude towards organic food Procedia Economics and Finance 31 2015 444 452 Benfield J.A. Szlemko W.J. Internet-based data collection: Promises and realities Journal of Research Practice 2 2 2006 Bhaskaran S. Polonsky M. Cary J. Fernandez S. Environmentally sustainable food production and marketing: Opportunity or hype? British Food Journal 108 8 2006 677 690 10.1108/00070700610682355 Brick K. Visser M. Burns J. Risk aversion: Experimental evidence from South African fishing communities American Journal of Agricultural Economics 94 1 2012 133 152 10.1093/ajae/aar120 Brucks M. The effects of product class knowledge on information search behavior Journal of Consumer Research 12 1 1985 1 16 10.1086/209031 Codron J.M. Siriex L. Reardon T. Social and environmental attributes of food products in an emerging mass market: Challenges of signaling and consumer perception, with European illustrations Agriculture and Human Values 23 3 2006 283 297 10.1007/s10460-006-9000-x Del Rio C. Collins L.F. Malani P. Long-term health consequences of COVID-19 Jama 324 17 2020 1723 1724 33031513 Di Renzo L. Gualtieri P. Pivari F. Soldati L. Attinà A. Cinelli G. Cinelli G. Leggeri C. Caparello G. Barrea L. Scerbo F. Esposito E. De Lorenzo A. Eating habits and lifestyle changes during COVID-19 lockdown: An Italian survey Journal of Translational Medicine 18 1 2020 1 15 10.1186/s12967-020-02399-5 31900168 Drichoutis A.C. Lusk J.L. What can multiple price lists really tell us about risk preferences? Journal of Risk and Uncertainty 53 2–3 2016 89 106 10.1007/s11166-016-9248-5 Duchin F. Sustainable consumption of food: A framework for analyzing scenarios about changes in diets Journal of Industrial Ecology 9 1‐2 2005 99 114 Escobar C. Kallas Z. Gil J.M. Consumers’ wine preferences in a changing scenario British Food Journal 120 1 2018 18 32 10.1108/BFJ-02-2017-0070 Evans D. Blaming the consumer - once again: The social and material contexts of everyday food waste practices in some English households Critical Public Health 21 4 2011 429 440 10.1080/09581596.2011.608797 Fonseca J.R.S. A latent class model to discover household food waste patterns in Lisbon city in support of food security, public health and environmental protection International Journal on Food System Dynamics 4 3 2013 184 197 Food and Agriculture Organization of the United Nations International Scientific Symposium. Biodiversity and sustainable diets – united against Hunger 2010 FAO Headquarters Rome Food and Agriculture Organization of the United Nations Global food losses and food waste 2011 http://www.fao.org/3/mb060e/mb060e00.htm Gallenti G. Troiano S. Cosmina M. Marangon F. Ethical and sustainable consumption in the Italian coffee market: A choice experiment to analyse consumers’ willingness to pay Rivista di Economia Agraria 71 2 2016 153 176 10.13128/REA-20077 Garnett T. Cooking up a storm: Food, greenhouse gas emissions and our changing climate Food climate research network 2008 Centre for Environmental Strategy, University of Surrey Garnett T. Where are the best opportunities for reducing greenhouse gas emissions in the food system (including the food chain)? Food Policy 36 2011 S23 S32 10.1016/j.foodpol.2010.10.010 Graham-Rowe E. Jessop D.C. Sparks P. Identifying motivations and barriers to minimising household food waste Resources, Conservation and Recycling 84 2014 15 23 10.1016/j.resconrec.2013.12.005 Harrison G.W. Lau M.I. Rutström E.E. Estimating risk attitudes in Denmark: A field experiment The Scandinavian Journal of Economics 109 2 2007 341 368 10.1111/j.1467-9442.2007.00496.x Harrison G.W. Lau M.I. Rutström E.E. Sullivan M.B. Eliciting risk and time preferences using field experiments: Some methodological issues Field Experiments in Economics 125–218 2005 10.1016/S0193-2306(04)10005-7 He S. Chen S. Kong L. Liu W. Analysis of risk perceptions and related factors concerning COVID-19 epidemic in chongqing, China Journal of Community Health 46 2 2020 278 285 Hill H. Lynchehaun F. Organic milk: Attitudes and consumption patterns British Food Journal 104 7 2002 526 542 10.1108/00070700210434570 Hillson D. Murray-Webster R. Managing risk attitude using emotional literacy Proceedings of the PMI Congress 2006 EMEA, presented in Madrid, Spain, 9 May 2006 Khan M.N. Rothwell D.W. Cherney K. Sussman T. Understanding the financial knowledge gap: A new dimension of inequality in later life Journal of Gerontological Social Work 60 6–7 2017 487 503 28398136 Kiefer I. Leitner B. Bauer R. Rieder A. Body weight: The male and female perception Sozial- und Präventivmedizin 45 6 2000 274 278 10.1007/BF01591690 11210598 Kim J. Lambrechts W. Annet van Osch J.S. How consumer behavior in daily food provisioning affects food waste at household level in The Netherlands Foods 8 10 2019 428 31547123 Li C. Mirosa M. Bremer P. Review of online food delivery platforms and their impacts on sustainability Sustainability 12 14 2020 5528 Lorek S. Spangenberg J.H. Sustainable consumption within a sustainable economy - beyond green growth and green economies Journal of Cleaner Production 63 2014 33 44 10.1016/j.jclepro.2013.08.045 Lusk J.L. Coble K.H. Risk perceptions, risk preference, and acceptance of risky food American Journal of Agricultural Economics 87 2 2005 393 405 Macdiarmid J.I. Is a healthy diet an environmentally sustainable diet? Proceedings of the Nutrition Society 72 1 2013 13 20 23186839 Maharjan K.L. Joshi N.P. Determinants of household food security in Nepal: A binary logistic regression analysis Journal of Mountain Science 8 3 2011 403 413 10.1007/s11629-011-2001-2 Ma Y. Ratnasabapathy R. Gardiner J. Carbohydrate craving: Not everything is sweet Current Opinion in Clinical Nutrition and Metabolic Care 20 4 2017 261 10.1097/MCO.0000000000000374 28375878 McCarthy B. Liu H.-B. Chen T. Trends in organic food consumption in China:opportunities and challenges for regional Australian exporters ResearchOnline@JCU 2014 8 10 Olivas R. Bernabéu R. Men's and women's attitudes toward organic food consumption. A Spanish case study Spanish Journal of Agricultural Research 2 2012 281 291 10.5424/sjar/2012102-507-11 Orduño Torres M.A. Kallas Z. Ornelas Herrera S.I. Analysis of farmers' stated risk using lotteries and their perceptions of climate change in the Northwest of Mexico Agronomy 9 1 2019 4 Osborne J. King J.E. Binary logistic regression Best Practices in Quantitative Methods 2011 358 384 10.4135/9781412995627.d29 Pappalardo G. Cerroni S. Nayga R.M. Yang W. Impact of covid-19 on household food waste: The case of Italy Frontiers in Nutrition 7 2020 291 10.3389/fnut.2020.585090 Pejman N. Kallas Z. Dalmau A. Velarde A. Should animal welfare regulations be more restrictive? A case study in eight European Union countries Animals 9 4 2019 195 10.3390/ani9040195 31027232 Pennings J.M.E. Garcia P. Measuring producers' risk preferences: A global risk-attitude construct American Journal of Agricultural Economics 83 4 2001 993 1009 Peschel A. Grebitus C. Steiner B. Veeman M. How does consumer knowledge affect environmentally sustainable choices? Evidence from a cross-country latent class analysis of food labels Appetite 106 2016 78 91 10.1016/j.appet.2016.02.162 69864 26944229 Power M. Doherty B. Pybus K. Pickett K. How COVID-19 has exposed inequalities in the UK food system: The case of UK food and poverty Emerald Open Research 2 2020 11 Pu M. Zhong Y. Rising concerns over agricultural production as COVID-19 spreads: Lessons from China Global Food Security 26 2020 100409 10.1016/j.gfs.2020.100409 32834954 Reardon T. Bellemare M.F. Zilberman D. How COVID-19 may disrupt food supply chains in developing countries IFPRI Book Chapters 2020 78 80 Rodríguez-Martín B.C. Meule A. Food craving: New contributions on its assessment, moderators, and consequences Frontiers in Psychology 6 2015 21 10.3389/fpsyg.2015.00021 25657636 Rodríguez-Pérez C. Molina-Montes E. Verardo V. Artacho R. García-Villanova B. Guerra-Hernández E.J. Ruíz-López M.D. Changes in dietary behaviours during the COVID-19 outbreak confinement in the Spanish COVIDiet study Nutrients 12 6 2020 1 19 10.3390/nu12061730 Rohrmann B. Risk perception, risk attitude, risk communication, risk management: A conceptual appraisal 15th Internaional Emergency Management Society (TIEMS) Annual Conference (Vol. 2008) 2008 Ruiz-Roso M.B. Knott-Torcal C. Matilla-Escalante D.C. Garcimartín A. Sampedro-Nuñez M.A. Dávalos A. Marazuela M. Covid-19 lockdown and changes of the dietary pattern and physical activity habits in a cohort of patients with type 2 diabetes mellitus Nutrients 12 8 2020 1 16 10.3390/nu12082327 Schroeder T.C. Tonsort G.T. Pennings J.M.E. Minter J. Consumer food safety risk perceptions and attitudes: Impacts on beef consumption across countries The B.E. Journal of Economic Analysis & Policy 7 1 2007 10.2202/1935-1682.1848 Serrano-Cruz M.R. Espinoza-Ortega A. Sepúlveda W.S. Vizcarra-Bordi I. Thomé-Ortiz H. Factors associated with the consumption of traditional foods in central Mexico British Food Journal 120 11 2018 2695 2709 10.1108/BFJ-11-2017-0663 Sharma H.B. Vanapalli K.R. Cheela V.S. Ranjan V.P. Jaglan A.K. Dubey B. Goel S. Bhattacharya J. Challenges, opportunities, and innovations for effective solid waste management during and post COVID-19 pandemic Resources, Conservation and Recycling 162 2020 105052 10.1016/j.resconrec.2020.105052 32834486 Sidor A. Rzymski P. Dietary choices and habits during COVID-19 lockdown: Experience from Poland Nutrients 12 6 2020 1 13 10.3390/nu12061657 Smidts A. The relationship between risk attitude and strength of preference: A test of intrinsic risk attitude Management Science 43 3 1997 357 370 Smith S. Paladino A. Eating clean and green? Investigating consumer motivations towards the purchase of organic food Australasian Marketing Journal 18 2 2010 93 104 10.1016/j.ausmj.2010.01.001 Taufique K.M.R. Vocino A. Polonsky M.J. The influence of eco-label knowledge and trust on pro-environmental consumer behaviour in an emerging market Journal of Strategic Marketing 25 7 2017 511 529 10.1080/0965254X.2016.1240219 Tversky A. The psychology of decision making Behavioural finance and decision theory in investment management Vol. 4 1995 AIMR Publication 2 6 10.1016/B978-0-12-800278-0.00006-3 Ureña F. Bernabéu R. Olmeda M. Women, men and organic food: Differences in their attitudes and willingness to pay. A Spanish case study International Journal of Consumer Studies 32 1 2008 18 26 10.1111/j.1470-6431.2007.00637.x Velavan T.P. Meyer C.G. The COVID-19 epidemic Tropical Medicine and International Health 25 3 2020 278 280 10.1111/tmi.13383 32052514 Wang X. Lei S.M. Le S. Yang Y. Zhang B. Yao W. Gao Z. Cheng S. Bidirectional influence of the COVID-19 pandemic lockdowns on health behaviors and quality of life among Chinese adults International Journal of Environmental Research and Public Health 17 15 2020 1 17 10.3390/ijerph17155575 Wang X. Pacho F. Liu J. Kajungiro R. Factors influencing organic food purchase intention in developing countries and the moderating role of knowledge Sustainability 11 1 2019 209 von Meyer-Höfer M. Juarez Tijerino A.M. Spiller A. Sustainable food consumption in China and India 2015 GlobalFood Discussion Papers, No. 60, Georg-August-Universität Göttingen, Research Training Group (RTG) 1666 - GlobalFood, Göttingen Wang C. Pan R. Wan X. Tan Y. Xu L. McIntyre R.S. Choo F.N. Tran B. Ho R. Sharma V.K. Ho C. A longitudinal study on the mental health of general population during the COVID-19 epidemic in China Brain, Behavior, and Immunity 87 2020 40 48 10.1016/j.bbi.2020.04.028 32298802 Yau Y.H.C. Potenza M.N. Stress and eating behaviors Minerva Endocrinologica 2013 255 Yu X. Abler D. The demand for food quality in rural China American Journal of Agricultural Economics 91 1 2009 57 69 Zakari S. Ying L. Song B. Factors influencing household food security in West Africa: The case of southern Niger Sustainability 6 3 2014 1191 1202 10.3390/su6031191 Zhu H. Deng F. How to influence rural tourism intention by risk knowledge during COVID-19 containment in China: Mediating role of risk perception and attitude International Journal of Environmental Research and Public Health 17 10 2020 3514 32443430	36474958	PMC9716012	NO-CC CODE	2022-12-03 23:20:52	no	Food Control. 2022 Feb 15; 132:108352	utf-8	Food Control	2,021	10.1016/j.foodcont.2021.108352	oa_other
==== Front Socioecon Plann Sci Socioecon Plann Sci Socio-Economic Planning Sciences 0038-0121 0038-0121 Published by Elsevier Ltd. S0038-0121(22)00028-3 10.1016/j.seps.2022.101250 101250 Article Integrating GIS in reorganizing blood supply network in a robust-stochastic approach by combating disruption damages Gilani Larimi Niloofar a Azhdari Abolghasem b Ghousi Rouzbeh c∗ Du Bo de a Gustavson School of Business, University of Victoria, Victoria, British Columbia, Canada b School of Earth and Environmental Sciences, The University of Queensland, Brisbane, Queensland, Australia c School of Industrial Engineering, Iran University of Science and Technology, Tehran, Iran d SMART Infrastructure Facility, University of Wollongong, Wollongong, NSW, Australia e School of Civil, Mining and Environmental Engineering, University of Wollongong, Wollongong, NSW, Australia ∗ Corresponding author. 29 1 2022 8 2022 29 1 2022 82 101250101250 19 4 2021 25 12 2021 22 1 2022 © 2022 Published by Elsevier Ltd. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. As supplying adequate blood in multiple countries has failed due to the Covid-19 pandemic, the importance of redesigning a sensible protective-resilience blood supply chain is underscored. The outbreak-as an extensive disruption-has caused a delay in ordering and delivering blood and its by-products, which leads to severe social and financial loss to healthcare organizations. This paper presents a robust multi-phase optimization approach to model a blood supply network ensuring blood is collected efficiently. We evaluate the effectiveness of the model using real-world data from two mechanisms. Firstly, a Geographic Information System (GIS)-based method is presented to find potential alternative locations for blood donation centers to maximize availability, accessibility, and proximity to blood donors. Then, a protective mathematical model is developed with the incorporation of (a) blood perishability, (b) efficient collation centers, (c) multiple-source of suppliers, (d) back-up centers, (e) capacity limitation, and (f) uncertain demand. Emergency back-up for laboratory centers to supplement and offset the processing plants against the possible disorders is applied in a two-stage stochastic robust optimization model to maximize the level of hospitals' coverage. The results highlight the fraction cost of considering back-up facilities in the total costs and provide more resilient decisions with lower risks by examining resource limitations. Keywords Blood supply chain Alternative collection facilities Disruption management Robust optimization Geographic information system Pandemic ==== Body pmc1 Introduction One of the vital roles of national healthcare systems is to transfer blood and its components from donation centers to demand zones by ensuring quality and safety [1]. The Coronavirus outbreak (Covid-19), as one of the severest disruptions due to its long term [2,3], increases the risk of unexpected failures, for it has had adverse influences on both the number of potential donors and physicians in the supply chain [4,5]. As a highly perishable and time-sensitive product, blood requires different strategies from those used in business supply chains [6,7]. Red blood cells or Red Packed cells (RPC) - with a lifespan between 35 and 42 days - are one of the vital components of whole blood, making up about 45% of whole blood [8,9]. The only way to achieve the availability of this perishable product is to sufficiently prepare a situation to increase the number of potential blood donors. Notably, donating blood in most countries is free, and blood transfusion organizations do not pay the donors [10]. In addition, various items -such as blood extraction technologies, different types of cross-matching, the distance between located donation centers and processing plants, and a high rate of perishability-boost the importance of the blood supply network. This research employs a multiphase approach to address this issue by integrating a GIS-based location method and a two-stage stochastic robust optimization model. Regarding this, the concern of blood supply in pandemics and gap analysis are described as follows. 1.1 Background and context of real study In the light of the COVID-19 pandemic, individuals use public transportation less than usual because of coronavirus dangers. This situation also affects the number of donors in the system, and donors are unwilling to travel a long distance to donate their blood. For example, in India, the Covid-19 outbreak creates many challenges in blood transfusion [11], including finding healthy blood, inventory management, and staff safety. Therefore, a strategic plan is needed to curb the demand dissatisfaction rate, outdated blood units, and supply cost. In a sprawling metropolis like Tehran, easy access to donation centers, processing plants, and medical centers plays a vital role in the supply chain. Accordingly, the importance of donation centers' geographic location to manage both supply and demand in the network is being considered to face this problem. Spatial differences in maximum coverage distance, availability to potential donors in a city, proximity to hospitals, and accessibility to the road network make these candidate locations distinguished. Moreover, the outbreak creates a worse condition and increases complexity in hospitals' blood supplies. After business closures in March 2020 due to the coronavirus pandemic, some donation centers were closed, and blood supply decreased abruptly in the United States [12]. This situation can be even more complicated in developing countries due to less stability and capacity limitation in processing plants. Besides that, the number of medical staff may be fluctuated by being infected with the coronavirus [13]. Therefore, to have a reliable system, it is crucial to manage data uncertainty and the processing plant's vulnerability [[14], [15], [16]]. Apart from the critical issues in the blood supply chain (BSC) that have arisen due to pandemic conditions, the blood supply chain's importance is increased by its unique features, such as blood extraction technologies and different types of cross-matching. Cross-matching is chosen depending on the time of blood transfusion and whether the injection should be given immediately, which affects managing inventory and patients' demands [10,17]. With this in mind, an efficient redesign of blood supply is crucial as ignoring these issues increases the mismatch between demand and capacity and escalates the possibility of irreversible occurrence in healthcare organizations. Fig. 1 shows the different layers of the supply chain and their connections after a massive disruption. The green flow shows the transportation of blood and its sub-products to hospitals in the case of disruption in processing plants. The dash-dot red arrows connect the donors to candidate donation centers, which need to be established first, and the red arrows show the available centers for donor arrival.Fig. 1 Conceptual representation of designing a blood supply network in a massive disruption or pandemics. Fig. 1 1.2 Gap analysis and contributions Failures in processing plants and demanding access to donation centers have unfavorable consequences for the system, especially in hospitals and medical centers. The unfavorable consequences include chronic shortage, delay in order, poor quality, and a high risk of cross-matching transfusion-transmitted infections. The Covid-19 pandemic has increased the risk of this situation. Accordingly, motivated by the real case, in which “back-up facilities” help the system not only to be reliable and resilient but also be protective against potential disruption and additional costs, we aim to generalize the case in analytical modeling with the following features:a) Efficient potential donation centers, which help decision-makers to establish the ones with higher efficiency and limitations. b) Perishability of RPC and rate of inappropriate production based on cross-matching type (based on patients' demand and laboratory product, Anti-globulin Cross-match (AC), and Immediate spin Cross-match (IC)). c) Different types of collection centers as suppliers (mobile and fixed), and the importance of availability, accessibility, and proximity in finding the best candidate location for donation centers to control donors' fluctuation d) A disruption in laboratories' supply capacity, which shows the importance of a back-up plan to test and analyze the collected whole-blood units. e) The coverage distance limitation for allocating donors due to the pandemic. f) Uncertainty in the number of arrival patients The purposes of the presented mixed-integer linear programming model are as follows:• Taking accurate and micro-scale vector data into account based on a GIS method to find the optimal locations for blood donation centers to maximize coverage and provide a boosted and more precise input for the optimization model. • Developing a two-stage stochastic framework in analytical mathematical modeling to optimize the BSC network, considering different types of cross-matching and protective approaches for processing plants. • Evaluating the demand fluctuation and disruption conditions in processing plants in the mathematical model. • Locating back-up facilities to support the processing plants and increasing the resiliency in the healthcare system. • Assessing the conservative level in the system by considering different amounts of the price of robustness to calculate both penalty value and unexpected penalty costs. This paper is organized as follows. Section 2 shows the relevant literature. The problem statement is presented in Section 3. In Section 4, the case description and solution methodology are discussed. Computational results and managerial insights are respectively defined in Sections 5, 6. Finally, conclusions and future opportunities are mentioned in Section 7. 2 Literature review Many studies have worked on city logistics and disruption conditions, and operations research models have become powerful approaches to manage logistics in emergencies [18]. Several related surveys on blood supply networks and healthcare management have been reviewed to highlight the research gaps and contributions better. Osorio et al. [19], Pirabán et al. [20], and Williams et al. [21] are the three important review papers in the field of BSC. Osorio et al. [19] reviewed the papers up to 2014 and categorized them based on the BSC echelons. Recently, Pirabán et al. [20] investigated documents up to 2019 based on network design, decision makings, processes, management of inventory, and data characteristics. Williams et al. [21] reviewed 46 papers on collection echelon and categorized them in appointment scheduling, collection policy, crisis, donor demographics, location/clinic planning, staff utilization, and vehicle routing. In the following, Section 2.1 highlights the concept of healthcare management and blood supply network in relevant essays by considering different features –such as perishability, cross-matching, variable demands, and capacity limitation. Section 2.2 categorized literature undertaking disastrous situations. The importance of epidemic outbreaks in healthcare analysis is mentioned in Section 2.3. 2.1 Healthcare management and blood supply chain One of our contributions is considering RPC perishability and rate of production based on cross-matching type. Limited essays have paid attention to testing and processing, especially cross-matching and ABO-RH(D) compatibility issues [20]. To describe briefly, the antigens' absence or existence classifies RPC into four different groups, A, B, AB, and O, subdivided into two classifications based on blood antibodies, named RH+ and RH_. Testing before RPC transfusion includes three steps, (1) Testing the ABO group of blood, (2) evaluating RH (D) matching, and (3) Cross-matching [22,23]. The third one is identifying and preventing harmful interactions between donors and recipients, including two different types, Antiglobulin Cross-match (AC) and Immediate spin Cross-match (IC). According to Achmadi & Mansur [24], the severity of cross-matching in the supply chain is 9 out of 9, illustrating the importance of this issue in the system. Most of the related works focused on cross-matching by healthcare analysis and management viewpoint [[25], [26], [27]], and a few are from analytical modeling and supply chain points of view. Gunpinar & Centeno [28] presented a model as a stochastic integer programming framework for RPC and platelets by considering two different types of demand based on the age of the product. Also, the cross-match-to-transfusion ratio was taken into account. Zahiri & Pishvaee [29] outlined a bi-objective mixed-integer linear programming model to design a BSC network. Their model aimed to minimize shortage and total costs by making a trade-off between these two objective functions. Baş Güre et al. [30] focused on a donation scheduling problem and balanced supply and demand using a robust possibilistic programming approach. Hosseini-Motlagh et al. [31] proposed a BSC network design by considering ABO-RH(D) compatibility to minimize the system's total costs. They used a robustness approach to tackle data uncertainty in their model. Hosseini-Motlagh, Samani, & Cheraghi [32] presented a location-allocation model for BSC to increase donors. Advertisement, education, and medical credits are motivational initiatives to boost blood donation. Our mathematical model, which distinguishes from the previous studies, tries to redesign a patient-centered BSC to manage a sudden massive disruption, such as the Covid-19 pandemic, simultaneously minimizing total costs and increasing healthcare resilience by considering different types of blood cross-matching. 2.2 Healthcare management in disastrous situations Disruption and disastrous conditions are concerns in production and supply chain environments [33]. An organization's entire plan can be distorted due to disruption, causing a deficiency in goods and unfulfilled customer requests. The development of fit policies can support minimizing losses and maintaining the goodwill of a company [34]. Several studies considered emergency response plans in the wake of disruptions. However, limited essays used mobility in facilities to manage the disastrous conditions. Acar and Kaya [35] presented a model based on a real case in Turkey, in which mobile hospitals are deployed to cover patient demand better. A scenario-based model was proposed by Li et al. [36] to manage uncertain data and transportation time in a combination of disasters. Blood is different from business models due to its prominent features, such as time sensitivity, perishability, and irreplaceability. Dealing with shortages and wastages are important issues in a blood network, especially in disruption conditions. Some researchers investigated the effect of disruptions in the BSC. Using a Lagrangian relaxation algorithm, Rahmani [37] presented a robust BSC network design to simultaneously control the solution against the disruption risks and uncertainty (in cost and demand). Yaghoubi et al. [38] used a two-objective mathematical model to make a trade-off between minimizing the total costs and delivery time for different types of platelets by assuming various types of production methods. They considered both partial and complete disruptions in their formulation. Haghjoo et al. [6] presented a scenario-based model of designing a BSC by considering the risk of uncertainty in demand and temporary facilities' disruption. The model aims to fulfill the demand by minimizing the total cost and considering the budget limitation. Hosseini-Motlagh, Samani, & Homaei [39] presented a fuzzy stochastic robust programming model in a BSC to simultaneously minimize shortage and cost in a disastrous situation. Patra and Jha [40] focused on the lack of replenishment opportunities after a disaster by formulating a prepositioning problem in a two-period newsvendor model. Our study, inspired by the previous articles in the crisis area, presents a robust-stochastic mixed-integer programming model to respond to emergencies with lower penalty costs and higher satisfaction rates for the Covid-19 patients. Partial or complete disruption in processing centers and demand fluctuations are the two primary uncertain data in this study. Thus, our approach finds an optimized solution for providing sufficient blood units for patients by considering simultaneously high penalty costs for shortage, and potential backup plans for processing centers and mobile clinics for donation. 2.3 Pandemic outbreak in healthcare management The roles played by the Covid-19 outbreak in supply chain designs are, of course, a significant disruption [41]. Blood -as a highly perishable product-needs to be supplied in a short interval period. However, many existing potential donors are unwilling to participate in donations because of the risk of disease transmission [42]. Choi [43] shows the effects of mobile centers in absorbing more customers in a business during a pandemic, showing that increasing and reorganizing the available donation centers around a region will positively affect the number of available donors in the supply network. It is because donors do not need to travel a long distance to reach the donation centers. For one thing, production procedures and donors' absorption are critical issues in supplying blood. For another, social distancing is vital to avoid contact during pandemics, similar to the COVID-19 [44]. Finding a way to increase donors' attraction to the system to produce sufficient blood can provide patients and healthcare organizations with an excellent opportunity to handle this critical situation better. Geographic information systems (GIS) is a method that allows the decision-makers to gather and manage data accurately, quickly, and precisely, which can be used in BSC network design [45]. Mollalo et al. [46] developed a GIS-based multiscale weighted regression model to investigate the interaction between several socio-demographic factors and the COVID-19 pandemic. After evaluating the effectiveness of the COVID-19 measures in India using the Bayesian probability model, Bherwani et al. [47] applied the GIS-based Voronoi approach to identify high-risk areas. Franch-Pardo et al. [48] also conducted a literature review on scientific articles that employed statistical analysis and GIS to study the geographical dimension of COVID-19. Moreover, a considerable number of related healthcare studies considered the optimal design in a strategic decision by integrating GIS. GIS technology not only analyze as well as visualize spatial data, it also has contributed to the widespread use of location models due to its user-friendly nature and low cost [49,50]. Delen et al. [45] presented a model based on operations research, data mining, and GIS approach to manage the blood supply chain and increase its availability. Kaveh & Mesgari [51] presented a location-allocation based on improved biogeography-based optimization to maximize the coverage of emergency centers. Comber [52], Dell'Ovo [53], and Özceylan [54] are the relevant papers, which utilized a GIS method to find the candidate locations. Concerning social aspects and reliability of donation centers is a critical topic in the blood supply network, especially in pandemic situations that availability, accessibility, and proximity are of utmost importance for both the donors and the blood transportation system. Given the power of GIS in analyzing spatial data, GIS-based location models in commercial GIS packages such as ArcGIS have been a significant part of optimization modeling [50]. However, in order to solve location problems more efficiently to reduce computing time and resources, commercial GIS packages employ heuristic methods to carry out faster analysis [49]. Given there is no need to use a solver that models the exact location in heuristic methods, the reliability of location models in commercial GIS packages is, therefore, questionable, and they are likely to render no optimal or sometimes even good solutions [50,55]. Another problem with location models in GIS packages is the assumption that the capacity of the facilities is unlimited, and that each can meet demand as much as possible. As most facilities are limited in terms of their service capacities, this problematic assumption can lead to questionable results [56]. Nonetheless, GIS can still be an effective tool in location modeling, especially for large datasets. For example, in the case of this study, which aims to serve the entire population of Tehran with more than eight million residents, the power of GIS can be used. The population of a city is not evenly distributed and significant demographic variations can be seen in a city's regions or statistical divisions. GIS is able to take into account such significant demographic dispersions in large datasets, while such a problem renders mathematical models impractical. Therefore, to use the power of both exact and heuristic methods, we used a GIS-integrated multi-phase mathematical model to find the donation centers' optimal location. The summary of the reviewed literature is shown in Table A1 (see in Appendix A) to highlight the research gaps. To the best of our knowledge, many of the papers have paid attention to the disruption condition in the first layer (collection procedure), and none of the mentioned papers present an alternative to cope with the effects of disruption in distribution and laboratories endangering patients' safety. Noteworthy, the number of processing plants (responsible for distributing the blood to hospitals on time) in a city is much lower than the donation centers (i.e., Tehran only has one processing plant, but six fixed donation centers). By this motivation, we utilized a GIS-based analytical model with the combination of back-up plans to better respond to the demands in the wake of disruption in pandemic conditions. Therefore, to maximize availability, accessibility, and proximity for the blood-donors to timely response in these emergencies, we reorganized more donation centers, as mobile and fixed, which helps us to increase potential donors and meet demands on time. Also, leveraging the back-up facility strategy implements low vulnerability and high reliability for the supply network against disorders. A robust analytical mathematical model with a service-oriented and cost-efficient approach is presented to combat demand variations and manage blood flow. 3 Problem statement This paper redesigned a four-echelon BSC network to prepare sufficient RPC units from donor groups to demand zones. Donor groups can donate their blood in three places: main donation centers, mobile facilities, and blood processing plants. The extracted blood from main donation centers and mobile facilities are sent to blood processing plants for testing, analyzing the quality of the product, and cross-matching the RPC based on hospitals' requirements. At the last echelon, the appropriate RPC will be distributed to demand zones. Fig. 2 shows a schematic summary of the phases in the proposed supply chain.Fig. 2 Flow chart of the main phases in the proposed BSC. Fig. 2 3.1 Location-allocation analysis in ArcGIS The three main factors, availability, accessibility, and proximity, play an important role in finding the best location for blood donation in a city [45]. The closer the blood donation centers are to the city's population centers, the more available blood donation centers are to donors, and the more likely residents are to donate blood [57]. Accessibility of these centers to the city's main routes also plays an essential role in the timely delivery of blood to hospitals and donors' access. The proximity of blood donation centers to city hospitals is also crucial, particularly in times of crisis. The need for blood in hospitals increases in times of crisis and blood can be used for a maximum of 8 h. Thus, the proximity of blood donation centers to hospitals in improving crisis response is essential. GIS is effective in studies that aim to find the optimal location for facilities to maximize services. Location-Allocation Analysis is one of the Network Analyst tools in ArcGIS that finds the optimal location for facilities based on the demand generated by different variables. In this method, candidate locations are selected to maximize coverage, or minimize costs, or maximize market share. Therefore, using ArcGIS 10.5, Location-Allocation Analysis includes demand points, facilities, and the network database, and it was set to cover the most demand. This analysis aims to provide the most coverage of demand points by facilities in the network database. The demand points of the analysis were the main hospitals as well as the population of the city. The population of census tracts had to be converted to population centroids to be used as demand points. A census tract consists of several adjacent urban blocks to form a homogeneous neighborhood, i.e., a census tract. Before that, the city's road network, which included the maximum speed limit and the direction of traffic on the street, was used as the input for the network dataset. The existing and the proposed facilities should also be identified in Location-Allocation Analysis, and they are referred to as required facilities and candidate facilities, respectively. Tehran has 22 districts whose activity centers can potentially be the best locations to build new blood collection centers. Thus, the existing blood donation centers and the activity centers of the districts were used as required facilities and candidate facilities, respectively. 3.2 Supply chain resiliency strategy with protective approach Using analytical mathematical modeling as a mixed-integer linear programming (MILP) framework, this study aims to combat the uncertain input data and possible disruption based on non-resiliency, partial-resiliency, and complete resiliency. Regarding this, based on the conservative robustness approach, the non-resilience model has considered no level of conservatism. In the partial-resilience model, this approach is less risky than the former to face unexpected penalty costs, although the highest level of conservatism has not been considered. The complete-resilience approach manages the system with the highest conservative level. Noteworthy, although the total cost of the supply chain might be increased in general conditions, it would significantly lessen the unexpected expenses in disastrous situations. Moreover, the flow of cross-matched blood between the echelons and the importance of each located donation center for absorbing donors and transshipping the extracted blood to both the processing plant and demand zones (in the case of disruption in the processing plant) need to be managed. In specific terms, the assumptions of the mathematical models are as follows:• Six main donation centers have already been established in Tehran. • Disruption occurs in processing plants in both partial and complete scenarios. Main donation centers and mobile facilities are considered as reliable centers in the supply chain. • Only located main donation centers can be considered as back-up facilities in the system. • The capacity of main donation centers, mobile facilities, and processing plants is fixed and bounded. • After transferring whole blood to processing plants, the process of extracting RPC begins. • Based on patients' demands, the extracted RPC divides into two types of cross-matching, AC and IC. • RPC perishability and inappropriate production rate are considered in the model, as the outdated RPC in processing plants or hospitals has wastage costs. The following indices, parameters, and variables are introduced in Table 1 to present the MILP model in a two-stage stochastic programming framework, in which patients' demand parameter is considered deterministically. Noteworthy, the three main disruption conditions incorporated in the formulation, can be dedicated as follows:• Condition 1: In the first condition, the processing plant is the only option for testing and analyzing the extracted blood and transferring them to hospitals. The blood will be extracted in donation centers, mobile facilities, and the processing plant. Then, all the extracted blood in donation centers and mobile facilities will be delivered to the processing plant as the only place equipped with expert staff and required technologies. • Condition 2: In the second condition, partial disruption occurs in the processing plant. Therefore, both the processing plant and donation centers can test and then transfer the extracted blood to the hospitals. Noteworthy, Condition 2 is the same as Condition 1 if the number of total demands is less than the non-disrupted capacity of the processing plant. In this situation, there is no need to assign back-up facilities in other centers. • Condition 3: In the third condition, the processing plant completely fails. Therefore, the only options to deliver the required blood units to the hospitals are the equipped main donation centers. It is noteworthy to mention that lateral transshipment between the located donation centers is not allowed, for it makes it hard to check the blood flow path. Table 1 Notations applied in the mathematical formulation. Table 1Indices I index of available donors in each region J index of possible main donation centers U index of possible mobile facilities B index of blood processing plants H index of public therapy centers R index of cross-matching types T index of time-periods S index of probable scenarios (probable conditions) Parameters a Holding cost of one blood unit c Cost of blood extraction through blood donors drht Demand of RPC cross-matched type r at public therapy center h in time-period t do'it Available blood donors in area i in time-period t Ebh Moving cost of per RPC unit from processing plant b to public therapy center h E'jh Moving cost of per RPC unit from the located main donation center j to public therapy center h (In the case of disruption in the processing plant) Fj Fixed cost of establishing one main donation center in location j ctjb& ctub' Transportation fee of one unit of RPC from location j (u forct') to processing plant b gr Rate of appropriate RPC units cross-matched type r produced at the processing plant Ku'u″ Cost of moving a mobile facility from location u1 to location u2 Luj Transportation cost of one RPC unit from mobile facility at location u to main donation center j M A big number Nbs Disruption parameter in each scenario, 0: if a disruption does not occur in processing plant b or Shortage penalty per unit of RPC demand cross-matched type r ps Probability of each scenario qij The full length between donor group i and main donation center j q'iu The full length between donor group i and mobile facility u q''ib The full length between donor group i and processing plant b vb Maximum vacancy of the blood bank in processing plant b v'b Maximum vacancy of processing plant b to accept blood donors wr Wastage penalty per unit of RPC demand cross-matched type r x,x',andx″ Coverage distance of each main donation center, processing plant, and mobile facility, respectively yj Cost of considering main donation center j as a back-up facility zj, zu' Maximum vacancy of main donation center j and temporary facility u, respectively Positive variables ccbrhts, ccjrht's Fraction of maximum RPC requirement cross-matched type r at public therapy center h in time-period t supplied by processing plant b and main donation center j, respectively IPbrts Inventory level of RPC units cross-matched type r at processing plant b in each scenario IPhrt's Inventory level of RPC units cross-matched type r at public therapy center h in each scenario NB' Number of required mobile blood facilities qpijts Quantity of RPC extracted through donor area i at main donation center j in time-period t in each scenario qp'iuts Quantity of RPC extracted through donor area i at mobile facility u in time-period t in each scenario qp''ibts Quantity of RPC extracted through donor area i at processing plant b in time-period t in each scenario φujts Quantity of RPC delivered from a mobile facility located in u to a main facility at location j in time-period t in each scenario φjbt's Quantity of RPC delivered from a main facility located in j to processing plant b in time-period t in each scenario φubt's Quantity of RPC delivered from a mobile facility located in u to processing plant b in period t in each scenario δhrts Shortage level of RPC units cross-matched type r at public therapy center h in time-period t in each scenario Δbrs Wastage level of RPC units cross-matched type r at processing plant b in each scenario Δhr's Wastage level of RPC units cross-matched type r at public therapy center h in each scenario Binary variables bjs' 1: if main donation center j assigned as a back-up facility in scenario s; 0 otherwise ffu'u″t's 1: if a mobile facility moved from location u1 to location u2 in period t in scenario s; 0: otherwise ffj 1: if a main donation center is established in location j; 0: otherwise xxjuts 1: if mobile facility u is dedicated to main facility j in time t in scenario s; 0: otherwise zzijts 1: if donor area i donates in main donation center j in time t in scenario s; 0: otherwise zz'iuts 1: if donor area i donates in mobile facility u in time t in scenario s; 0: otherwise zz''ibts 1: if donor area i donates in processing plant b in time t in scenario s; 0: otherwise Objective function:(1) Min∑jffjFj+∑sps(∑ju'u″tffu'u″t'sKu'u″+∑jutφujtsLuj+∑bhrt(IPbrts+IPhrt's)a+∑jbtφjbt'sctjb+∑ubtφubt'sctub'+∑hrtδhrtsor+∑jbjs'yj+∑br(Δbrs+Δ′hrs)wr+∑brhtccbrhtsdrhtEbh+∑jrhtccjrht'sdrhtE'jh+∑iujbt(qp'iuts+qpijts+qp″ibts)c) The objective function aims to minimize the overall costs in the supply chain, including costs of inventory in hospitals and the processing plant, shortage and wastage costs of each type of patient, costs of establishing main donation centers and mobile facilities, transportation costs (i.e., the amount of blood that moved from one echelon to another; and moving mobile facilities), and considering an established main donation center as a back-up facility. The constraints are divided into seven groups, mentioned in Appendix B. Equations (B.2) to (B.8) refer to assigning blood donors to mobile facilities. Equations (B.9) to (B.14) refer to the main donation centers' constraints. The blood processing plant constraints are mentioned in equations (B.15) to (B.17). Constraints (B.18) to (B.22) are the inventory constraints. The back-up plan in the mathematical model has mentioned in Constraints (B.23) to (B.26). Constraints (B.27) and (B.28) refer to the wastage amount in the model. The last two constraints show the domain of decision variables. 3.3 Robustness approach for blood demand tailoring problem Fluctuation in blood demands in different periods is inseparable in reality, as inaccurate estimation poses a delay in its access, which may cause an escalating rate of unsatisfied demand in the system [58]. Regarding this, a conservative robustness approach is applied in this study [[59], [60]] (Refer to Appendix C). 4 Data collection This section demonstrates the data used for the GIS approach and the mathematical model from a realistic case study to evaluate its possibility and practicality. Tehran is the capital of Iran and its largest city with the best equipment and treatment staff. The demand points for the Location-Allocation Analysis include 18 main hospitals of Tehran, whose beds vary from 1300 to 119, representing the weight of the demand points (Fig. 3 ). Information on the hospitals and the number of their beds were obtained from the Iran Blood Transfusion Organization (IBTO) and the hospitals, respectively. Also, Tehran's census tracts were the rest of the demand points for analysis obtained from the last Iranian national census in 2016. Tehran is made up of 4330 census tracts, and they were converted to population centroids. Thus, there are 4330 population centroids or demand points used in this article, the population of which represented their weight in the analysis. The road network, which was extracted from the vector maps of Tehran's master plan, formed the network database, in which the maximum speed limit and the direction of traffic on the street were included. Also, facilities included the existing main donation centers as required facilities and 29 activity centers of Tehran's districts as candidate facilities. There are 22 districts in Tehran; however, some districts had two activity centers due to their size or large population. According to IBTO and hospitals' reports, the rate of producing cross-matched unit type AC and IC are 70% and 30%, respectively. The reason for the lower use of the second type is that the ABO-RH (D) of the donor and recipient must be equal as a prerequisite. The rest of the parameters are gained through experts' knowledge, IBTO, and relevant papers [[61], [62], [63]] (Refer to Appendix D).Fig. 3 Major roads in 22 districts of Tehran, as well as the geographical distribution of activity centers, existing main donation centers, and main hospitals. Fig. 3 5 Results and discussion The applicability and sensitivity analysis of the proposed model and the solution techniques are presented in this section. All the models are solved with GAMS 24.1.2 using CPLEX solver on a PC with CORE i7 and 8 GB memory. First, the candidate location of the main donation centers is calculated by the GIS approach. Then, the resilience model with a protective approach against the processing plant disruption is analyzed in two steps. In the first step, the protective model is applied in different disruption scenarios. In the second step, the protective model is validated in a robustness approach, in which the analytical approach aims to combat both demand fluctuations and possible disruption. 5.1 Locating main donation centers Assuming that each person is willing to spend 30 min to reach the blood donation centers, location-allocation analysis located 22 potential blood donation centers from 36 existing blood collection centers and activity centers (Fig. 4 ). Since the population of census tracts and main hospitals with different bed capacities have not been evenly distributed across Tehran, districts 4, 5, and 12 each had two potential locations for blood collection centers. On the other hand, the analysis proposed no potential blood collection center in the 22nd district of Tehran, which has a low population density. These potential blood collection centers were then used in the optimization model as input data to eventually determine the optimum blood collection centers and the back-up facilities needed in times of crisis.Fig. 4 Proposed blood donation centers located by Location-Allocation Analysis. Fig. 4 5.2 The impact of coverage distance on logistics costs Due to the pandemic and to increase the available donors in the system, this model considered restricted coverage distance for each mobile and fixed donation center (see Table 2 ). For one thing, this limitation would increase the logistics cost in comparison to normal situations. For another, this decision increases donors' satisfaction rate and makes a proper condition to donate blood by traveling a short distance. Table 3 shows different distance limitations in the system, which would help the decision-makers (DMs) choose the best-fit decision based on the budget limitation. In the optimistic case, DMs can ignore adding one more donation center (DC) and two mobiles, and in the pessimistic point of view, adding two more DCs and using all mobile DCs are recommended. Also, the existing facility 1 is useless for its location cannot cover considerable donors. This result also highlights the impact of mobile centers in pandemics, as if the DMs prefer to limit each collection center coverage, adding more mobile facilities is preferred.Table 2 The acronym of each donation center. Table 21 Vesal 6 Piroozi 11 Al-Qadir Sq. 16 Shoosh Sq. 21 Tehransar 2 Sattari 7 Imam Khomeini SW 12 Enghelab Sq. 17 Nabard Sq. 22 Shahed Sq. 3 Tehranpars 8 Tajrish Sq. 13 Azadi Sq. 18 Basij Sq. 4 Shahre Ray 9 Sanaat Sq. 14 Shamshiri Sq. 19 Yaftabad 5 Sadr 10 Resalat Sq. 15 Imam Khomeini Sq. 20 Sarvi Sq. Table 3 The effect of distance limitation in total costs and number of DCs. Table 3Coverage distance Total cost number of located DCs number of located mobile DCs Inefficiency in existed DCs Mobile facility Fixed DCs 1500 1500 3330000 15 4 facility 1 1500 3500 1570000 15 4 facility 1 1500 4000 1350000 15 4 facility 1 2250 5250 825150 14 2 – 3000 7000 779000 13 2 – 3375 7875 701000 13 1 – 4500 10500 645000 12 0 – 5.3 Back-up facility and resiliency sensitivity analysis As stated before, IBTO has considered several methods to protect the model against disruption in the processing centers. For this purpose, it has been considered that the processing center's capacity in Tehran should be twice as big as the expected demand. Therefore, failures over 50% need back-up facilities. The first scenario (S = 0) demonstrates that the model is reliable and in stable conditions, in which disruption cannot affect the system. The second and third scenarios show the model in partial disruption, and the fourth scenario (S = 4) is when the processing center has wholly failed. Fig. 5 illustrates the number of main donation centers and back-up centers in different scenarios. As can be seen, the number of back-up facilities has a direct relationship with the failure rate. Moreover, the number of considered main donation centers and mobile facilities has increased due to the disruption in the supply chain. Noteworthy, to fulfill all blood demands in the system, at least seven more donation centers need to be established in non-critical situations.Fig. 5 Number of located main donation center and mobile facilities in deterministic form of demand. Fig. 5 Table 4 demonstrates the logistics costs, back-up facility costs, transportation costs, and inventory costs in all periods in the healthcare system. Since the proposed blood supply network is service-based, we investigate different incident scenarios and a two-stochastic approach. The first row (S = 0) shows the supply chain in the deterministic condition. Rows two to four respectively reveal the outcome costs in 50%, 70%, and 100% disruption in the processing plant, and the fifth row illustrates the model based on the two-stage stochastic approach. The last row, titled “Gap%," points to the solutions gained through the MILP approach, which means the difference between the current solution and the best lower bound (optimal answer) when the problem-solving process ends. The columns are divided into three groups. The first category reveals the outputs by considering back-up facilities for increasing the resiliency of the system. The second group shows the costs without a protection strategy for the processing plant with the exact probable disruption. The last column of the table shows the fraction cost of the back-up facility in the total logistics costs in different disruption situations, which fluctuate in the range of 21%–67% based on the level of destruction in processing plants. As can be seen, the total costs increase under disruption occurrence in the processing plant. However, this cost is much more if no resiliency strategy applies to the system. Based on Table 4, if the system is protected for 50% disruption, the total costs will increase by 11%. The total logistics costs in scenarios 2, 3, and the stochastic model are 2, 3.84, and 2.13 times higher than the deterministic one. On the other side, if the network faces disruption without any prior management, not only do higher logistics costs incur to the system, but the cost of patient dissatisfaction will also be added. Noteworthy, not assigning back-up facilities in complete disruption in the processing plant gives infeasible results, as no other supplier is considered to satisfy hospitals' demands. By applying the stochastic model, not only the proposed supply chain is protected against partial or complete failure, but the total logistics costs will also be decreased to 1663100 IRR (Rials/unit). The required resilient conditions will be met if costs increase by 36% to allocate support centers and back-up facilities.Table 4 Logistics costs' reports (Rials/unit). Table 4Disruption Scenarios with protective centers without protective centers Costb/t Total logistics costs Back-up facility costs Transportation costs Inventory costs Total logistics costs Shortage costs S = 0 778760 0.0 35642.4 743.604 778760 0.0 0.0 S = 1 881220 192500 39866.2 9.5 5696940.1 5285700 0.21 S = 2 1572400 927850 70287.1 22 24297558.7 24276420 0.59 S = 3 2991600 2020287 201584.4 0 infeasible infeasible 0.67 TSS 1663100 613112.5 49412.7 116.8 – 0.0 0.36 Gap%: 0.00 Gap%: 0.00 TSS: Two-stage stochastic model. Costb/t: The fraction cost of back-up facility in the total logistics costs. Fig. 6 illustrates the percentage of donation participation in each region in the deterministic situation. As can be seen, activity centers 4, 8, and 15 are the three critical points in supplying the blood demand. Therefore, if healthcare organizations and IBTO wish to persuade other regions and donation centers to increase donations, rewarding these three donor groups can be beneficial. These areas are very close to the processing plant, illustrating the importance of transportation in locating donation centers.Fig. 6 Blood donation in each area in the deterministic form of demand without disruption in all periods. Fig. 6 Fig. 7 explicitly demonstrates the effect of incident scenarios in transferring extracted blood units from donation centers to the demand zones. As can be seen, the higher the failure rate in the processing plant, the more donation centers will be considered as back-up centers. Since lateral transshipment between donation centers cannot be considered due to the rules of blood transshipment according to IBTO, all the located donation centers are considered back-up centers if a complete disruption occurs in the processing plant. Noteworthy, the number of located donation centers depends on various factors, such as limited capacity, coverage distance, available blood donors, establishment cost, implemented in the problem formulation with two mechanisms. Firstly, the GIS approach is used to find the optimal candidate centers, and then, the proposed MILP model is presented to find the best fit. According to the Figure, alternative facilities 6, 15, and 16 are the three most important ones in collecting and transferring blood units to hospitals in the situation of high rate or complete failure in the supply chain. This result also shows that among all DCs, DCs 14, 19, and 20 have the least role. It means that in case of budget limitations, it would be better to eliminate them from the candidate DCs.Fig. 7 Effect of each main donation center in transferring blood to hospitals in different scenarios. Fig. 7 The importance of each mobile facility is illustrated in Table 5 . As can be seen through the results of assessment metrics, mobile centers are moved among activity areas 7, 10, and 18 and preferred not to meet the two other facilities. The main reasons for this result are that the donation rate of activity centers 22 and 16 is lower than the other areas, and the transportation costs to reach the destination are not affordable. Therefore, DMs prefer to lose the available donors in those areas (see Table 6 ).Table 5 The movement among candidate mobile areas in different periods. Table 5Period 1 10 → 18→2 Period 2 18 → 10→2 Period 3 7 → 10→18 Period 4 18 → 10→7 Period 5 7 → 10→7 Total number of located mobile facilities 2 Table 6 Evaluation of the expected value solution based on different weights of each scenario. Table 6Weight of each scenario Total cost Transportation cost Back-up cost DAM DAF DAP S0=0.8,S1=0.2,S2&3 = 0 805119 37639 41195 88 7489 1382 S0=0.7,S1&2&3 = 0.1 1133075 56382 310598 74 7784 1240 S0=0.6,S1=0.2,S2&3 = 0.1 1159684 59766 338703 124 7802 1132 S0=0.5,S1&2=0.2,S3=0.1 1236126 62517 416473 162 7374 1291 S0=0.4,S1=0.3,S2=0.2,S3=0.1 1246704 61742 436878 172 7358 1272 DAM: Donor absorption to mobile centers. DAF: Donor absorption to fixed collection centers. DAP: Donor absorption to processing plants. 5.4 The impact of price of dissatisfaction rate in the supply network The shortage cost parameter is given from Ensafian & Yaghoubi [61], which is 1500 $/unit (6300 × 103 Toman/unit or 6300 × 104 Rials/unit). This large number is considered to minimize and avoid shortages in the proposed network. Fig. 8 reveals the deficiency rate changes in all periods based on different unfulfillment costs for each cross-matching type. S and S′ indicate the shortage level of AC and IC cross-matching types in different scenarios, respectively. As shown in Fig. 8, the deficiency amount-especially the IC type-will escalate considerably by decreasing the importance of facing shortages in the supply chain. The ratio of deficit cost to the total cost (RCT) varies in different shortage costs, so the lower the shortage cost, the higher RCT. Calculating RCT, the cost of the deficit is divided by the total logistics costs. RCT gives its higher amount if the shortage cost is considered its least value (130 × 103 Toman/unit), about 90% of logistics cost. It means that if patients are not in emergency status, DMs may decide not to respond to the hospital's demand due to limited supply and logistics costs. It would be logical for DMs not to add extra DCs when the price of dissatisfaction rate downs to 3 times than the emergency one.Fig. 8 Amount of shortage of each cross-matched RPC based on different values of shortage costs. Fig. 8 5.5 Analyzing the quality of the expected value solution Different weights for each scenario are considered, and the value of each variable is assessed [64]. As can be seen, as the weights of partial or complete disruption scenarios increase, the total logistics costs also escalate. This change also reveals an upward trend in the number of donors to mobile centers and transportation costs. 5.6 Sensitivity of robust-stochastic approach and realization One of the critical parameters, which is the inseparable part of the blood supply, is demand fluctuation. The scenario-based approach is a risk-averse method to combat and control the adverse effects of demand uncertainty in the healthcare system. Table 7 analyses the model's robustness when the model faces uncertainty in both demand and probable processing center's disruption. The first row of the table shows different budget levels of uncertainty in the system, and the first column shows the results in different scenarios. The last part of the table reveals the optimal data in the two-stage stochastic approach. The results in the deterministic form of both demand and disruption can be seen in κ=0%. The uncertainty effects will be escalated as the values of budget of uncertainties and S are close to one and three, respectively (The last column of the fourth row). The results reveal that by increasing uncertainty in the supply network, the expected logistics costs – to control and manage-will intensify. Moreover, as can be seen from Fig. 9 , by considering both fluctuations in demand and failure in the processing center, the computational time rises significantly. However, by combining the TSS method to protect the system, the computational time only increased up to 1.47 times than the worst scenario time.Table 7 The costs result of the model in deterministic, robust, and stochastic framework (℧1=℧2=1). Table 7Budget of uncertainty κ=0% κ=20% κ=40% κ=60% κ=80% κ=100% S=0 Logistic costs 778760 887230 993100 1115700 1256200 1402100 Back-up facility costs 0.0 0.0 0.0 0.0 0.0 0.0 S = 1 Logistic costs 881220 1042000 1225400 1416300 1617700 1825100 Back-up facility costs 192500 274312 351312 442750 534187 620812 S = 2 Logistic costs 1794360 1881091 1921200 2105800 2309800 2505200 Back-up facility costs 927850 1020250 1091475 1178100 1267612 1339800 S = 3 Logistic costs 2991600 3152300 3315600 3439700 3613200 3802300 Back-up facility costs 2020287 2111725 2150225 2265725 2353312 2444750 TSS Logistic costs 1663100 1784165 1813441 1910350 2074377 2238993 Back-up facility costs 613112 835714 951440 1001371 1117941 1199899 TSS = Two-stage stochastic form. Fig. 9 Computational time comparison of the proposed solution techniques based on different conservative levels. Fig. 9 We proposed ten different realizations to examine the applicability of the robustness approach. The impact of different budgets of uncertainty levels between the deterministic and robustness strategy is described in Table 8a, Table 8b . The penalty cost (PC) and unexpected penalty cost (UPC) have been highlighted in each realization. Symbol Re shows the realization number.Table 8a The model effectiveness under different realization. Table 8aRe Deterministic Robustness approach κ=0% κ=20% κ=40% PC UPC PC UPC PC UPC 1 6233268 4570168 4654885 2870720 4143263 2329822 2 5025164 3362064 4959105 3174940 4239070 2425629 3 6222866 4559766 4917935 3133770 4229249 2415808 4 5217873 3554773 4937357 3153192 4079985 2266544 5 5342685 3679585 4535473 2751308 4035069 2221628 6 5710973 4047873 4904441 3120276 4075860 2262419 7 5757660 4094560 4816854 3032689 4035178 2221737 8 5267115 3604015 4463038 2678873 4245374 2431933 9 6292726 4629626 4668846 2884681 4113709 2300268 10 5069447 3406347 4995066 3210901 4252205 2438764 AVG 5613978 3950878 4785300 3001135 4144896 2331455 SDV 473185 473185 180817 180817 84646.28 84646.28 Table 8b The model effectiveness under different realization. Table 8bRe Robustness approach κ=60% κ=80% κ=100% PC UPC PC UPC PC UPC 1 3617836 1707486 2221392 147015 1470148 0 2 3513031 1602681 1904442 0 1196958 0 3 3813999 1903649 2278453 204076 1802371 0 4 3666908 1756558 2265369 190992 2142477 0 5 3246586 1336236 2156937 82560 1174362 0 6 3136721 1226371 1923605 0 1365288 0 7 3254834 1344484 1989375 0 1367795 0 8 3596658 1686308 2384515 310138 1854911 0 9 3546673 1636323 1904551 0 1326835 0 10 3083708 1173358 2224493 150116 1140194 0 AVG 3447695 1537345 2125313 108489.7 1484134 0 SDV 235428.9 235428.9 169412.6 103718.2 319607.1 0 As can be observed, the optimal value of the objective function in the robustness approach is notably better than the deterministic one. The PC, UPC, and average also take lower values in the robust optimization. Besides that, the average and standard deviation of the penalty and unexpected penalty costs decrease considerably by increasing the conservatism level. When the level of conservatism escalates, the price of robustness would be more than the deterministic status. On the other hand, the PC and UPC would be controlled significantly when the system meets a fluctuation in the number of unexpected new patients. By enhancing the conservatism level from 0 to 0.2, the robustness price would be 121,065 × 104 IRR, but the system would save 1,699,448 × 104 IRR when there is an extra patient in a period. Moreover, by extending the conservatism level from 0.2 to 1, the robustness price will be only 25%. However, the system can reduce its cost by 68% if newly sudden patients are in the supply network. 6 Managerial insights According to the WHO, the COVID-19 pandemic has caused blood supply in emergencies, and the healthcare systems should be reliable enough to work properly during this pandemic. With this in mind, a resilient and optimized strategic-tactical decision-making process must be evaluated by considering BSC features to manage the supply network entirely. The results provided the following findings as managerial implications:(a) Perishability consideration can combat with wastage amount and its cost in the system. Regarding this, the amount of useable blood units increases in the blood banks. Moreover, perishability forces the system not to hold the blood unit up to the last lifetime period, bringing about lower inventory cost for the supply chain and higher quality of blood for patients. (b) The population of a city is never evenly distributed across the city. Furthermore, the quality of the road network has a significant effect on access to facilities. In other words, while each component of the road network has a different maximum speed, traffic flow, and traffic direction, the Euclidean distance between different variables in conventional models does not give an accurate picture of accessibility between facilities and demand points (i.e., potential donors and hospitals). GIS addresses this issue by employing accurate and micro-scale data. By conducting location-allocation analysis in ArcGIS 10.5, blood donation centers were optimally located to have better access to the road network, are available to potential donors, and are proximate to hospitals. (c) The rate of collecting and delivering RPC (in the case of disruption and by considering the center as a back-up) in donation centers helps us find the importance of each donation center in the system. Factors –such as coverage distance, available potential donors, and the establishment and transportation costs-endow the decision-makers to choose a candidate center to be located. Therefore, in the case of budget limitations, DCs with the highest contribution should be selected. (d) The fraction cost of the back-up facility in the total logistics costs gives a logical view to decision-makers to choose the best decision for the system by considering the budget limitation. It is notable to mention that due to the pandemic, medical staff is endangered by the threats of being infected, which means that a portion of them might become deactivated. According to our results, by increasing almost 36% of the expected total costs, the supply network can be supported entirely by the potential disruption. (e) The decision-makers can reorganize the proposed BSC for applying a protective network in the healthcare system. For this purpose, the maximum capacity of the service level in the processing plant plays a leading role in disruption situations. The current Tehran processing plant can support up to 50% of the disruption without back-up in the system. This percentage can be varied in different supply chain networks, in which the lower value requires higher back-up importance in the healthcare organization. (f) Due to the natural uncertainty in demand and lack of information on possible disruption in the supply chain, sensitivity analysis should be implemented in the uncertainty budget based on the probability of each scenario. This study evaluated the total logistics cost and RPC production in a demand-protection model to face possible data fluctuations. Regarding this, PC and UPC decline by raising the conservative level, the average and standard deviation of the objective function. Moreover, a minor increase in the system's total cost can protect the blood supply network from failure by considering fluctuations in the number of new patients. By increasing the price of robustness from 0 to 0.2, the system can save 23% of the penalty costs, and by increasing the level of conservatism from 0.2 to 0.4, this value will grow significantly by only a 1.6% increase in total costs. By turning the value of robustness to 1, the price of UPC would reach almost 0. This point can turn on lights for DMs to choose the best decision for the blood supply chain. 7 Conclusion The outbreak of COVID-19 has caused a substantial reduction in blood transfusion activities, including blood donations. Under the guidance of blood transfusion and organization, several visible systemic issues (i.e., RPC cross-matching type, blood perishability, multiple-source of suppliers, back-up centers, capacity limitation, and pandemic situations) have been found in the blood supply network, which ushered us to incorporate the GIS approach with an analytical mathematical model. By considering three different conditions, namely non-disrupted, partial disrupted, and complete failure, the application of our MILP model presented an incentive mechanism to minimize both the total logistics costs and RPC flow disorder. The summary of this paper is provided as follows:• To come closer to reality, availability, accessibility, and proximity, as the three main factors, are considered and evaluated by the GIS approach. By applying a GIS-based method, 22 potential blood donation centers were located to be optimally available to potential donors throughout the city, most accessible through the city's road network, and finally proximate to the city's main hospitals. • Then, a protective blood supply network designed to combat the uncertain data and failures in the system based on a real case. The proposed study includes four leading echelons from donor arrival to hospitals for collecting, testing, distributing, and consuming the extracted blood. In this study, the possible disorders were managed in three statuses. (a) As during the Covid-19 pandemic, many potential donors prefer to travel the shortest distance for donation, we assessed the effectiveness of existing DCs based on limited coverage distance. The results illustrated that even some existing DCs might need to be closed for the donation rate at negligible areas. Also, the outcome revealed the importance of mobility in facilities when the maximum coverage distance becomes more limited. (b) The test and distribution echelon may face disruption, where the extracted blood units need to be analyzed and subdivided into two cross-matching types (i.e., AC and IC). Besides that, the number of processing plants in the blood supply network is limited. Therefore, some DCs as alternative facilities were incorporated into the analytical model to protect the system. Among different disruption possibilities, four situations were considered in a scenario-based approach to handle the blood flow and control the shortage level. (c) There is a fluctuation in the number of patients in the healthcare system. Therefore, a robust optimization approach was presented at first, and then, ten different realizations were provided to show the price of robustness effectiveness in the supply network. The conservative level of each budget of uncertainty has shown, and the penalty costs and unexpected penalty costs are dedicated to shedding some light on this subject better. The practical management insights through the findings of the proposed model can be seen in the preceding section. Future works can focus on dynamic routing strategies to absorb more donors in disruption and disastrous conditions. As during an outbreak, the number of available blood donors would decrease, practitioners can manage this issue by mobile facilities and absorbing donors from farther areas. Besides that, the high rate of perishability in the system as a distance limitation and the other type of mobile facility (without extraction technology, which is used only for transferring blood to the laboratory centers) need to be considered. Further research can be focused on different types of production procedures for blood extraction. In the end, evaluating the concept of resource sharing in hospitals and its efficiency is another direction on this subject of interest. Appendix A Table A1 An overview of related studies Table A1References Product type Hierarchical level Donation mode Planning horizon Goal and objective Time period Back-up facility Solution approach Cross-matching ABO-RH compatibility Real case Uncertainty Different patients Network design Perishability Disaster/disruption Other feature Processing plant Donation center Larimi and Yaghoubi [10] Platelet Integrated Apheresis-whole blood Tactical Min Cost& Max donors Multi – Robust-stochastic – Yes Demand Yes – Yes – – – Hosseini-Motlagh et al. [31] Blood Integrated whole blood Strategic- tactical Min cost& max service Multi – Robust – Yes Yes Demand – Yes Yes – – – Ensafian et al. [23] Platelet Integrated Apheresis-whole blood Tactical-operational Min cost Multi – Robust-stochastic – Yes Yes Demand Yes – Yes – – – Gunpinar and Centeno [28] Blood-platelet Distribution whole blood Tactical-operational Min cost Multi – Stochastic Yes – – Demand Yes – Yes – – – Zahiri and Pishvaee [29] Blood Collection whole blood Strategic- tactical Min cost& shortage Multi – Fuzzy-robust – Yes Yes Demand – Yes – – – – Beheshtifar and Alimoahmmadi [68] – Network design – Strategic Min cost& max service single – exact – – Yes – – Yes – – – GIS-approach for locating hospitals Kizito andWilliam [65] – Network design – Strategic Min non-standard hospitals single – exact – – Yes – – Yes – – – GIS-approach for locating hospitals Yaghoubi et al. [38] Platelet Inventory- distribution whole blood Strategic- tactical Min cost& time delivery Multi – Robust& p-robust – – Yes Demand Yes Yes Yes – Yes Samani et al.[66] Blood Inventory- distribution Not-mentioned tactical Min cost& max service Multi – Robust – – Yes Demand, supply, costs – Yes Yes – – – Rahmani [37] Blood Collection Not-mentioned Strategic Min cost Multi – Robust& lagrangian relaxation – – – Demand& cost – Yes – – Yes – Hosseini-Motlagh et al. [39] Plasma Integrated Apheresis-whole blood Strategic- tactical Min cost Multi – Robust& p-robust – – Yes Demand& cost Yes Yes Yes – Yes – Haghjoo et al. [6] Blood Collection Not-mentioned Strategic Min cost Multi – stochastic – – Demand& failure – Yes – – Yes – Liu et al. [16] Blood-platelet Inventory- distribution Not-mentioned Tactical Min cost Multi – decomposition-based algorithm – – – – – – Yes – – – Achmadi & Mansur [24] Blood products Inventory Not-mentioned – Cross-matching – – Operations management Yes – – – – – – – – – Baş Güre et al. [30] Blood Donation – Operational Min time Single – Robust-possiblistic – – – Demand – – – – – Scheduling Hosseini-Motlagh et al. [32] Blood Network design – Strategic Min costs Multi – Fuzzy – – Yes Demand – Yes – – – Donor motivation, advertisement Acar and Kaya [35] – Network design – Tactical Max satisfaction rate – – Scenario-based – – Yes Demand- transportation time Yes Yes – – – Disastrous situation in hospitals Patra and Jha [40] – Network design – Tactical-operational max service Multi – Scenario-based – – Yes – – – – – Yes lack of replenishment opportunities Choi [43] – Network design – Tactical Min costs Multi – Scenario-based – – – – – – – – Yes Mobile centers considered Mollalo et al. [46] – Supplier – Strategic max service – – – – – Yes – – – – – Yes GIS-based approach Bherwani et al. [47] – Supplier – Strategic max service – – Bayesian probability model – – – – – – – – – GIS-based approach in Pandemics Delen et al. [45] Blood Collection – Strategic max service – – data mining – – Yes – – – – – – Max availability Kaveh & Mesgari [49] – Network design – Strategic- tactical Min costs – – – – – – – – Yes – – – Max coverage in emergencies Samani et al. [69] blood Integrated – Strategic- tactical Min cost-Max quality Multi-period – Fuzzy – – Yes Costs- demand- quality – Yes Yes – – – This research Blood Integrated whole blood Strategic- tactical Min cost& max protect Multi-period Yes Robust & scenario-based Yes No Yes Demand& failure No Yes Yes Yes – GIS-approach for finding the candidate donation centers' location Appendix B Subject to: Constraints of assigning blood donors to mobile facilities:(B2) ∑u″ffu'u″t's≤1∀u',t,s (B3) ∑u'≠u″ffu'u″t's=NB'∀t,s (B4) ∑u″ffu'u″t's≤∑u″ffu'u″t−1's∀u',t,s (B5) q'iuzz'iuts≤x″ffu'u″t's∀i,u,t,s (B6) qp'iuts≤zz'iutsM∀i,u,t,s (B7) ∑ieeiuts≤zu'∀u,t,s (B8) ∑jzzijts+∑uzz'iuts+∑bzz''ibts≤1∀j,s,t Constraint (B.2) assures that at most one mobile facility can move from other locations to a specific area in each period. The third constraint points to the number of open mobile facilities in each time period. Constraint (B.4) guarantees that a mobile facility can move to another location only if it is first allocated in location u. Constraint (B.5) points to the coverage distance limitation for assigning donor groups to mobile facilities. Constraint (B.6) makes sure that donors can donate their blood in a mobile facility only if it has been established. Constraint (B.7) refers to the capacity limitation in each mobile facility. Constraint (B.8) ensures that each group of donors can donate in only one of the three centers of mobile facilities, main donation centers, or the processing plant. Constraints of main donation centers:(B9) qijzzijts≤xffj∀i,j,t,s (B10) qpijts≤zzijtsM∀i,j,t,s (B11) φujts≤xxujtsM∀j,u,t,s (B12) xxujts≤∑u'ffu'u″t's∀j,u,u″,t (B13) xxujts≤ffjM∀u,j,t,s (B14) ∑iqpijts+∑uφujts≤zj∀j,t,s Constraints (B.9) and (B.10) refer to the point that each donor group only can assign to a main donation center if the donation center has been founded before in its coverage distance. Constraints (B.11) and (B.13) ensure that each mobile facility can assign to a main donation center if both the donation center and mobile facility have been established. Constraint (B.14) limits the maximum capacity in each main donation center. Constrains of processing plant:(B15) q''ibzz''ibts≤x'∀i,b,t,s (B16) qp''ibts≤zz''ibtsv'b∀i,b,t,s (B17) ∑iqp''ibt+∑jφjbt's+∑uφubt's≤vb×(1−Nbs)∀b,t,s Constraint (B.15) refers to the coverage distance for each processing plant. Constraint (B.16) limits the number of donors that can be accepted to donate their blood in the processing plant. Constraint (B.17) ensures that amount of blood units in the processing plant is not more than its capacity. (1−Nbs) on the left side of the equation refers to the probable disruption situation in the processing plant, which decreases the processing plant's capacity. Flow balance and inventory constraints:(B18) ∑bqp''ibts+∑uqp'iuts+∑jqpijts≤do'it∀i,t,s (B19) ∑iqp'iuts=∑jφujts+∑bφubt's∀u,t,s (B20) ∑bφjbt's+∑rhccjrht'sdrht=∑uφujts+∑iqpijts∀j,t,s (B21) IPbrts=IPbr,t−1s+(∑jφjbt's+∑uφubt's+∑iqp''ibts)gr−∑hccbrhtsdrht∀b,r,t,s (B22) IPhrt's=IPhr,t−1's+∑bccbrhtsdrht+∑jccjrht'sdrht−drht+δhrts∀h,r,t,s Constraint (B.18) limits the maximum supply achieved by donors. The flow balance for the mobile facilities in different periods has shown in constraint (B.19). Constraint (B.20) illustrates the flow balance for each main donation center. Constraints (B.21) and (B.22) evaluate the inventory level of each kind of cross-matched product in each processing plant and hospital, respectively. Back-up facilities and protective approach constraints:(B23) ∑jccjrht's+∑bccbrhts≤1∀h,r,t,s (B24) ccjrht's≤bjs'×M∀j,h,r,t,s (B25) bjs'≤∑bnbs×M∀j,t (B26) bjs'≤fj×M∀j,s Constraint (B.23) ensures that the delivery rate of a hospital cannot be more than one. It means that a hospital can gain at most 100% of its demand from both the back-up facilities and processing plant. Constraint (B.24) ensures that a main donation center can directly transfer their blood units to hospitals only if considered a back-up facility. Constraint (B.25) guarantees that none of the main donation centers can be regarded as a back-up center unless there is a disruption in the processing plant. Constraint (B.26) states that only a located main donation center can be considered as a back-up center. Constraints of wastages amount(B27) IPbrts=Δbrs∀b,r,s,t=6 (B28) IPhrt's=Δhr's∀h,r,s,t=6 Constraints (B.27) and (B.28) calculate the number of wastages units of both types of cross-matched blood in the processing plant and hospitals, respectively. Domain of decision variables(B29) zzijts,zz'iuts,zz''ibts,xxjuts,ffu1u2t's,ffj,bjs'ε{0,1}∀i,j,u,b,s,t (B30) ccbrhts,ccjrht's,IPbrts,IPhrt's,NB',qpijts,qp'iuts,qp''ibts,φujts,φjbt's,φubt's,δhrts,Δbr,Δhr'ss≥0∀i,j,u,b,h,r,s,t Appendix C Based on Bertsimas & sim [59] robustness approach, the following MILP model is considered, where N',c,andb are respectively convex polyhedron, the n-vector of objective coefficients, and right-hand side constraints:(C31) min∑jcjn'j (C32) subjectto:∑ja˜ijn'j≤b˜i∀i (C33) n'∈N' If a˜ij is considered as the uncertain parameter, which can be changed in the range of [aij−aˆij,aij+aˆij], where aˆij refers to the deviation of aij, its uncertain formulation can be written as follows,(C34) ψ(N',Γi)=maxaij∈Ji(∑jaˆijnj')∀i (C35) ∑jaijnj'+ψ(N',Γi)≤bi∀i In which, Ji and Γi are an uncertainty set and budget of uncertainty for each constraint i, respectively. Γi can take a value between [0, \|Ji\|]. The lower bound of the budget of uncertainty illustrates the model in the lower protected situation, as Γi=0 means the model cannot be protected against demand fluctuation. Regarding this, Γi=\|Ji\| shows the fully protected system against uncertainty. The duality of the robust model can be formulated as:(C36) min∑jcjn'j (C37) ∑jaijnj'+ZiΓi+∑jPij≤bi∀i (C38) Zi+Pij≥aijnj'∀i,j (C39) Pij≥0∀i,j (C40) Zi≥0∀i (C41) n'j≥0∀j Similar to the above formulation, if b˜i is the uncertain parameter, which can be changed in the range of [bi−bˆi,bi+bˆi], where bˆi refers to the deviation of bi, its uncertain formulation can be written as follows [70],(C42) min∑jcjn'j (C43) subjectto:AN'≤bi+bˆiΓ′i∀i Accordingly, the conservatism level (κ) in this model is in the range of [0, 100%]. As this value goes to the upper bound [100%], the level of protection of the model increases. If the uncertain demand varies in the range of [drht‾−drht˜,drht‾+drht˜], the revision of equation (1), (B.20), (B.21), and (B.22) can be seen in the following:(C44) Min∑jfjaj'+∑jutspsfu1u2t'sgu1u2+∑jutspsφujtsguj″+∑bhrtsps(qbrts+qhrt's)a+∑jbtspsφjbt'sgjb'+∑ubtspsφubt'sggub'+∑hrtspsδhrtsor+∑jspsbjs'bej+∑brsps(Δbrs+Δhrs)wr+∑brhtspscbrhtsd‾rhtEbh+∑jrhtspscjrht'sd‾rhtE'jh+∑iujbtsps(eeiuts+eijts+e'ibts)ro+ℶ1℧1+∑brhtspsℷbrhts+ℶ2℧2+∑brhtspsℷ'brhts (C45) cbrhtsd‾rht≤ℶ1+ℷbrhts∀b,r,h,t,s (C46) cjrht'sd‾rht≤ℶ2+ℷ'brhts∀j,r,h,t,s (C47) ∑bφjbt's+∑rhcjrht'sd‾rht+ΓtΖt+∑rhΨrht=∑uφujts+∑ieijts∀j,t,s (C48) ∑rhΨrht+Ζt≥∑rhcjrht'sd‾rht∀t (C49) qbrts=qbr,t−1s+(∑jφjbt's+∑uφubt's+∑ie'ibts)brrbir−[∑hcbrhtsd‾rht+Γ′rtΖ′rt+∑hΨ′rht]∀b,r,t,s (C50) ∑hΨ′rht+Ζ′rt≥∑hcbrhtsd‾rht∀r,t (C51) qhrt's=qhr,t−1's+[∑bcbrhtsd‾rht+Γ″rhtΖ″rht+Ψ″rht]+[∑jcjrht'sd‾rht+Γ″rhtΖ″rht+Ψ″rht]−[d‾rht+Γ‴drht˜]+δhrts∀h,r,t,s (C52) Ψ″rht+Ζ″rht≥∑bcbrhtsd‾rht∀r,h,t (C53) Ψ″rht+Ζ″rht≥∑jcjrht'sd‾rht∀r,h,t (C54) ℶ1,ℶ2,Ζt,Ψrht,Ψ′rht,Ζ′rt,Ζ″rht,Ψ″rht≥0 where ℶ1,ℶ2,ℷbrhts, and ℷ'brhts are dual auxiliary variables. To find the maximum deviation of drht˜, an index named perturbation level (Λ) is defined as a given data to take a value between [0,1]. With this in mind, drht˜ can be calculated as Λdrht. Budget of uncertainty Γ can be changed between [0,\|t\|], Γ′ is in the range between [0,\|t×r\|], Γ″ can be changed between [0,\|t×r×t\|]. ℧1 and ℧2 can be changed between [0,\|t×r×t\|]. For further information, readers can see Refs. [36,[66], [67]]. Appendix D Table. D1 shows the sources of the input data. Table. D1The value of input data Parameter Reference Parameter Reference i Real case, Tehran is divided into 22 areas s Decision-makers, divided into four different scenarios, no disruption, 50% disruption, 70% disruption, and complete disruption. j Based on the IBTO website ps Decision-makers, 0.4 for s = 0, 0.2 for the rest scenarios u Based on the IBTO website Cost parameters Based on experts' knowledge, and relevant literature [7,38,61,71] b Real case, Tehran has one processing plant Distance parameters Based on the real case by GIS approach h Number of main consumers of blood, based on the IBTO website Capacity parameters The real data, based on IBTO experts' knowledge and documents r Based on medical facts brr Real case, based on IBTO reports and related studies [58] t Real case, the weekly period up to blood lifespan bir Relevant articles [23] and IBTO reports The description of each area in Tehran is illustrated in Table D2, including population, number of available donation centers, number of potential donation centers as alternative care facilities, and donation rate.Table D2 The data of each activity center in Tehran Table D2Activity center Population (π) No. of available DC No. of Potential DC No. of Potential mobile clinics Available donation rate (π×0.1%) 1 381003 1 1 – 1905 2 608918 – 1 – 3045 3 655718 – 1 – 3279 4 856072 1 2 – 4280 5 687377 1 2 – 3437 6 240751 1 – – 1204 7 310184 – – 1 1551 8 379036 – 1 – 1895 9 167000 – 2 – 835 10 316620 – – 1 1583 11 275595 – 1 – 1378 12 250188 – 2 – 1251 13 246407 1 1 – 1232 14 690609 – 1 – 3453 15 655277 – 1 – 3276 16 315115 – – 1 1576 17 256439 – 1 – 1282 18 322656 – – 1 1613 19 255027 – 1 – 1275 20 336339 1 1 – 1682 21 160107 – 1 – 801 22 112711 – – 1 564 *Note that DC is the acronym of donation center. Acknowledgement We would like to show our gratitude to the: (1) Bita Hadinejad, Head-nurse (2) Mahtab Moradi Koohbad, physician; And (3) Naser Gilani Larimi, physician for sharing their knowledge and information with us during this research study. We sincerely thank the Editor and two anonymous reviewers for their kind and helpful comments on this paper. ==== Refs References 1 Williamson L.M. V Devine D. Challenges in the management of the blood supply Lancet 381 9880 2013 1866 1875 23706803 2 Govindan K. Mina H. Alavi B. A decision support system for demand management in healthcare supply chains considering the epidemic outbreaks: a case study of coronavirus disease 2019 (COVID-19) Transport Res Part E Logist Transp Rev 138 2020 101967 3 Verikios G. The dynamic effects of infectious disease outbreaks: the case of pandemic influenza and human coronavirus Socioecon. Plann. Sci. 71 2020 100898 32834133 4 Politis C. Strategies for blood collection and optimization of the blood supply chain during the COVID‐19 pandemic in Greece ISBT Sci Ser 15 4 2020 386 392 5 Leung J.N.S. Lee C.-K. Impact of the COVID‐19–a regional blood centre's perspective ISBT Sci Ser 15 4 2020 362 364 6 Haghjoo N. Tavakkoli-Moghaddam R. Shahmoradi-Moghadam H. Rahimi Y. Reliable blood supply chain network design with facility disruption: a real-world application Eng Appl Artif Intell 90 2020 103493 7 Hosseini-Motlagh S.-M. Larimi N.G. Nejad M.O. A qualitative, patient-centered perspective toward plasma products supply chain network design with risk controlling Oper Res 2020 1 46 8 Hamdan B. Diabat A. Robust design of blood supply chains under risk of disruptions using Lagrangian relaxation Transport Res Part E Logist Transp Rev 134 2020 101764 9 Jafarkhan F. Yaghoubib S. Gilani Larimi N. Farhang Moghadam B. The inventory–routing problem for distribution of red blood cells considering compatibility of blood group and transshipment between hospitals Adv. Ind. Eng. 53 3 2020 31 44 10 Larimi N.G. Yaghoubi S. A robust mathematical model for platelet supply chain considering social announcements and blood extraction technologies Comput Ind Eng 137 2019 106014 11 Dhiman Y. Patidar G.K. Arora S. Covid‐19 pandemic‐response to challenges by blood transfusion services in India: a review report ISBT Sci Ser 15 4 2020 365 373 12 Gniadek T.J. Expansion of hospital‐based blood collections in the face of COVID‐19 associated national blood shortage Transfusion 60 7 2020 1470 1475 32445417 13 Teng Z. Psychological status and fatigue of frontline staff two months after the COVID-19 pandemic outbreak in China: a cross-sectional study J Affect Disord 275 2020 247 252 32734915 14 Ratick S. Meacham B. Aoyama Y. Locating backup facilities to enhance supply chain disaster resilience Growth Change 39 4 2008 642 666 15 Jabbarzadeh A. Fahimnia B. Seuring S. Dynamic supply chain network design for the supply of blood in disasters: a robust model with real world application Transport Res Part E Logist Transp Rev 70 2014 225 244 16 Liu K. Li Q. Zhang Z.-H. Distributionally robust optimization of an emergency medical service station location and sizing problem with joint chance constraints Transp Res part B Methodol 119 2019 79 101 17 Lowalekar H. Ravi R.R. Revolutionizing blood bank inventory management using the TOC thinking process: an Indian case study Int J Prod Econ 186 2017 89 122 18 Caunhye A.M. Nie X. Pokharel S. Optimization models in emergency logistics: a literature review Socioecon. Plann. Sci. 46 1 2012 4 13 19 Osorio A.F. Brailsford S.C. Smith H.K. A structured review of quantitative models in the blood supply chain: a taxonomic framework for decision-making Int J Prod Res 53 24 2015 7191 7212 20 Pirabán A. Guerrero W.J. Labadie N. Survey on blood supply chain management: models and methods Comput Oper Res 112 2019 104756 21 Williams E.P. Harper P.R. Gartner D. Modeling of the collections process in the blood supply chain: a literature review IISE Trans Healthc Syst Eng 10 3 2020 200 211 22 Bandarenko N. King K. Blood transfusion therapy: a physician's handbook eleventh ed. 2014 American Association of Blood Banks (AABB) Bethesda, Maryland 23 Ensafian H. Yaghoubi S. Yazdi M.M. Raising quality and safety of platelet transfusion services in a patient-based integrated supply chain under uncertainty Comput Chem Eng 106 2017 355 372 24 Achmadi R.E. Mansur A. Design mitigation of blood supply chain using supply chain risk management approach Proceedings of the international conference on industrial engineering and operations management 2018 1763 1772 25 Bozorgi A. Najafi M. Improving blood bank inventory management using double cross-match and hybrid issuance policy 2020 IEEE 7th international conference on industrial engineering and applications 2020 ICIEA) 819 826 26 Dumas M.B. Rabinowitz M. Policies for reducing blood wastage in hospital blood banks Manag Sci 23 10 1977 1124 1132 27 Katsaliaki K. Cost-effective practices in the blood service sector Health Pol 86 2–3 2008 276 287 28 Gunpinar S. Centeno G. Stochastic integer programming models for reducing wastages and shortages of blood products at hospitals Comput Oper Res 54 2015 129 141 29 Zahiri B. Pishvaee M.S. Blood supply chain network design considering blood group compatibility under uncertainty Int J Prod Res 55 7 2017 2013 2033 30 Baş Güre S. Carello G. Lanzarone E. Yalçındağ S. Unaddressed problems and research perspectives in scheduling blood collection from donors Prod Plann Control 29 1 2018 84 90 31 Hosseini-Motlagh S.-M. Samani M.R.G. Homaei S. Blood supply chain management: robust optimization, disruption risk, and blood group compatibility (a real-life case) J Ambient Intell Hum Comput 11 3 2020 1085 1104 32 Hosseini-Motlagh S.-M. Samani M.R.G. Cheraghi S. Robust and stable flexible blood supply chain network design under motivational initiatives Socioecon. Plann. Sci. 70 2020 100725 33 Fan D. Lo C.K.Y. Zhou Y. Sustainability risk in supply bases: the role of complexity and coupling Transport Res Part E Logist Transp Rev 145 2021 102175 34 Qiu R. Sun Y. Sun M. A distributionally robust optimization approach for multi-product inventory decisions with budget constraint and demand and yield uncertainties Comput Oper Res 126 2021 105081 35 Acar M. Kaya O. A healthcare network design model with mobile hospitals for disaster preparedness: a case study for Istanbul earthquake Transport Res Part E Logist Transp Rev 130 2019 273 292 36 Li Y. Zhang J. Yu G. A scenario-based hybrid robust and stochastic approach for joint planning of relief logistics and casualty distribution considering secondary disasters Transport Res Part E Logist Transp Rev 141 2020 102029 37 Rahmani D. Designing a robust and dynamic network for the emergency blood supply chain with the risk of disruptions Ann Oper Res 283 1 2019 613 641 38 Yaghoubi S. Hosseini-Motlagh S.-M. Cheraghi S. Larimi N.G. Designing a robust demand-differentiated platelet supply chain network under disruption and uncertainty J Ambient Intell Hum Comput 11 8 2020 3231 3258 39 Hosseini-Motlagh S.-M. Samani M.R.G. Homaei S. Toward a coordination of inventory and distribution schedules for blood in disasters Socioecon. Plann. Sci. 72 2020 100897 40 Patra T.D.P. Jha J.K. A two-period newsvendor model for prepositioning with a post-disaster replenishment using Bayesian demand update Socioecon. Plann. Sci. 2021 101080 41 Khan S.A.R. Razzaq A. Yu Z. Shah A. Sharif A. Janjua L. Disruption in food supply chain and undernourishment challenges: an empirical study in the context of Asian countries Socioecon. Plann. Sci. 2021 101033 42 Stanworth S.J. Effects of the COVID-19 pandemic on supply and use of blood for transfusion Lancet Haematol 7 10 2020 756 764 43 Choi T.-M. Innovative ‘bring-service-near-your-home’ operations under Corona-virus (COVID-19/SARS-CoV-2) outbreak: can logistics become the messiah? Transport Res Part E Logist Transp Rev 140 2020 101961 44 Jorfi M. Diagnostic technology for COVID-19: comparative evaluation of antigen and serology-based SARS-CoV-2 immunoassays, and contact tracing solutions for potential use as at-home products,” medRxiv 2020 45 Delen D. Erraguntla M. Mayer R.J. Wu C.-N. Better management of blood supply-chain with GIS-based analytics Ann Oper Res 185 1 2011 181 193 46 Mollalo A. Vahedi B. Rivera K.M. GIS-based spatial modeling of COVID-19 incidence rate in the continental United States Sci Total Environ 728 2020 138884 32335404 47 Bherwani H. Understanding COVID-19 transmission through Bayesian probabilistic modeling and GIS-based Voronoi approach: a policy perspective Environ Dev Sustain 23 4 2020 5846 5864 32837277 48 Franch-Pardo I. Napoletano B.M. Rosete-Verges F. Billa L. Spatial analysis and GIS in the study of COVID-19. A review Sci Total Environ 739 2020 140033 32534320 49 Murray A.T. Contemporary optimization application through geographic information systems Omega 99 2021 102176 50 Murray A.T. Xu J. Wang Z. Church R.L. Commercial GIS location analytics: capabilities and performance Int J Geogr Inf Sci 33 5 2019 1106 1130 51 Kaveh M. Mesgari M.S. Improved biogeography-based optimization using migration process adjustment: an approach for location-allocation of ambulances Comput Ind Eng 135 2019 800 813 52 Comber A. Dickie J. Jarvis C. Phillips M. Tansey K. Locating bioenergy facilities using a modified GIS-based location–allocation-algorithm: considering the spatial distribution of resource supply Appl Energy 154 2015 309 316 53 Dell'Ovo M. Capolongo S. Oppio A. Combining spatial analysis with MCDA for the siting of healthcare facilities Land Use Pol 76 2018 634 644 54 Özceylan E. Çetinkaya C. Erbaş M. Kabak M. Logistic performance evaluation of provinces in Turkey: a GIS-based multi-criteria decision analysis Transport Res Part A Policy Pract 94 2016 323 337 55 Tong D. Murray A.T. Spatial optimization in geography Ann Assoc Am Geogr 102 6 2012 1290 1309 56 Xu J. Murray A. Wang Z. Church R. Challenges in applying capacitated covering models Trans GIS 24 2 2020 268 290 57 Piersma T.W. Bekkers R. de Kort W. Merz E.-M. Altruism in blood donation: out of sight out of mind? Closing donation centers influences blood donor lapse Health Place 67 2021 102495 33341080 58 Larimi N.G. Yaghoubi S. Hosseini-Motlagh S.-M. Itemized platelet supply chain with lateral transshipment under uncertainty evaluating inappropriate output in laboratories Socioecon. Plann. Sci. 68 2019 100697 59 Bertsimas D. Sim M. Robust discrete optimization and network flows Math Program 98 1 2003 49 71 60 Hanks R.W. Lunday B.J. Weir J.D. Robust goal programming for multi-objective optimization of data-driven problems: a use case for the United States transportation command's liner rate setting problem Omega 90 2020 101983 61 Ensafian H. Yaghoubi S. Robust optimization model for integrated procurement, production and distribution in platelet supply chain Transport Res Part E Logist Transp Rev 103 2017 32 55 62 Cheraghi S. Hosseini-Motlagh S.-M. Responsive and reliable injured-oriented blood supply chain for disaster relief: a real case study Ann Oper Res 291 1 2020 129 167 63 Wang C. Chen S. A distributionally robust optimization for blood supply network considering disasters Transport Res Part E Logist Transp Rev 134 2020 101840 64 Maggioni F. Wallace S.W. Analyzing the quality of the expected value solution in stochastic programming Ann Oper Res 200 1 2012 37 54 65 Kizito A. William C.M.P. Geospatial location-allocation approach to optimize healthcare facilities in peri-urban dwellers in mbeya city, Tanzania MUST J Res Dev 1 1 2020 56 78 66 Samani M.R.G. Hosseini-Motlagh S.-M. Ghannadpour S.F. A multilateral perspective towards blood network design in an uncertain environment: methodology and implementation Comput Ind Eng 130 2019 450 471 67 Paul J.A. Wang X.J. Robust location-allocation network design for earthquake preparedness Transp Res part B Methodol 119 2019 139 155 68 Beheshtifar S. Alimoahmmadi A. A multiobjective optimization approach for location‐allocation of clinics Int Trans Oper Res 22 2 2015 313 328 69 Samani M.R. Hosseini-Motlagh S.M. Sheshkol M.I. Shetab-Boushehri S.N. A bi-objective integrated model for the uncertain blood network design with raising products quality Eur J Ind Eng 13 5 2019 553 588 70 Hatefi S.M. Jolai F. Robust and reliable forward–reverse logistics network design under demand uncertainty and facility disruptions Appl Math Model 38 9–10 2014 2630 2647 71 Rajendran S. Ravindran A.R. Platelet ordering policies at hospitals using stochastic integer programming model and heuristic approaches to reduce wastage Comput Ind Eng 110 2017 151 164	36475013	PMC9716013	NO-CC CODE	2022-12-03 23:20:52	no	Socioecon Plann Sci. 2022 Aug 29; 82:101250	utf-8	Socioecon Plann Sci	2,022	10.1016/j.seps.2022.101250	oa_other
==== Front J Health Soc Behav J Health Soc Behav HSB sphsb Journal of Health and Social Behavior 0022-1465 2150-6000 SAGE Publications Sage CA: Los Angeles, CA 35164599 10.1177/00221465221074915 10.1177_00221465221074915 Article Racial-Ethnic Residential Clustering and Early COVID-19 Vaccine Allocations in Five Urban Texas Counties https://orcid.org/0000-0002-7425-8923 Anderson Kathryn Freeman 1 Ray-Warren Darra 1 1 University of Houston, Houston, TX, USA Kathryn Freeman Anderson, Department of Sociology, University of Houston, 3551 Cullen Blvd, PGH Building, Room 450, Houston, TX 77204-3012, USA. Email: [email protected] 12 2022 12 2022 12 2022 63 4 472490 © American Sociological Association 2022 2022 American Sociological Association This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Previous research has indicated that racial-ethnic minority communities lack a wide variety of health-related organizations. We examine how this relates to the early COVID-19 vaccine rollout. In a series of spatial error and linear growth models, we analyze how racial-ethnic residential segregation is associated with the distribution of vaccine sites and vaccine doses across ZIP codes in the five largest urban counties in Texas. We find that Black and Latino clustered ZIP codes are less likely to have vaccine distribution sites and that this disparity is partially explained by the lack of hospitals and physicians’ offices in these areas. Moreover, Black clustering is also negatively related to the number of allocated vaccine doses, and again, this is largely explained by the unequal distribution of health care resources. These results suggest that extant disparities in service provision are key to understanding racial-ethnic inequality in an acute crisis like the COVID-19 pandemic. COVID-19 health care race-ethnicity residential segregation vaccines typesetterts1 ==== Body pmcThe COVID-19 pandemic has been one of the most disruptive world events in modern times in terms of health and fundamentally reshaping the economy and people’s social lives. Moreover, this impact has not been equally shared. Some people, because of existing social and economic resources, have been better able to shield themselves from transmission of the disease and the various negative consequences reverberating from the pandemic. In particular, Black, Latino, and Native American populations in the United States have experienced higher rates of infection and, when infected, have been more likely to experience severe disease and death (Millett et al. 2020; Peek et al. 2021; Ramos and Zamudio 2020). Other work has shown how the effects of the pandemic are also spaced and placed, with poor and minority neighborhoods suffering disproportionately from the disease and having limited access to care sites and testing locations (McMinn et al. 2020; Yang, Choi, and Sun 2021). In this analysis, we focus on racial-ethnic disparities in access to the vaccine in the United States specifically as it relates to racial-ethnic residential segregation. On December 11, 2020, the U.S. Food and Drug Administration issued the first emergency use authorization for the Pfizer-BioNTech vaccine for COVID-19 in individuals 16 years of age and older in the United States. Eight days later, a second vaccine, Moderna, was approved for use in individuals ages 18 and older. However, distribution of the vaccine has been a logistical hurdle necessitating extremely low temperature cold storage and personnel needs, all while having to maintain COVID-19 safety protocols. For example, unpunctured vials of the vaccines must be kept between −90°C and −60°C for the Pfizer-BioNTech vaccine and between −50°C and −15°C for the Moderna vaccine, well below the temperature of a standard freezer. An obvious question that has emerged is how to do this equitably, especially when vaccine supply is low. Roughly following guidelines from the Centers for Disease Control and Prevention (CDC; 2020a), in the case of Texas, which is focus of this analysis, Phase 1A of the rollout was reserved for frontline health care workers, residents of long-term nursing facilities for the elderly, and workers in those same facilities. Phase 1B included people ages 65 and above and people ages 16 and above with at least one comorbidity defined by the state as being particularly vulnerable for the disease, such as cancer, type 2 diabetes, obesity, chronic heart disease, and so on. Although the goal was to equitably distribute vaccine doses to reach the vulnerable, the question remains whether this goal was met. This is the overarching question that we tackle with this analysis. We focus only on the Phase 1A and 1B periods to examine who got early access while vaccine supply and eligibility were still limited. Obviously, many considerations are at play for how any individual can get the vaccine, and early media reports indicate that access to reliable Internet and transportation play a major role in people’s ability to find an appropriate appointment time and travel to the location (Menchaca and Agnew 2021; Prescott and Prescott 2021; Stone 2021). Here, we focus on the structural element of this story to examine where these vaccine allocations went and whether certain types of neighborhoods were advantaged or disadvantaged in the rollout. We ask: What neighborhood characteristics are related to the distribution of COVID-19 vaccines? Specifically, are racial-ethnic minority communities less likely to have vaccine supply? Furthermore, how do these patterns relate to extant inequalities in health care provision? We posit that racial-ethnic minority areas will be less likely to receive vaccine doses and that this will be explained away or mediated by existing disparities in service provision across neighborhoods given the difficult infrastructure requirements of distributing the two vaccines. Background Segregation and Access to Health Care Research has long documented important health and health care outcomes related to racial-ethnic residential segregation for Black and Latino communities in the United States. For instance, analysts have documented a relationship between segregation and racial differences in mortality for several causes of death (Collins 1999; Collins and Williams 1999; Hart et al. 1998), including infant mortality (Grady 2006; McFarland and Smith 2011). In addition, other studies have examined overall physical health, mental health, and functional disability and found that minority residents of racially segregated neighborhoods are more likely to report experiencing overall poorer physical and mental health and disability (Acevedo-Garcia 2000; Acevedo-Garcia et al. 2003; Anderson and Fullerton 2014; Lee 2009). Finally, a few studies on access to health care demonstrate an association between racial-ethnic segregation and diminished access to health care coverage, having a personal physician, and health care utilization (Anderson and Fullerton 2014; Gaskin et al. 2009; Rodriguez et al. 2007). Although there is robust literature demonstrating the association between segregation and several health and health care outcomes, less attention has been given to understanding and testing the mechanisms that could link these two. A notable exception is Williams and Collins’s (2001) piece in which they describe racial-ethnic minority segregation as a “fundamental cause” of health and detail how this form of segregation can be fundamentally linked to health outcomes. They suggest that racial-ethnic minority segregation, with its intimate ties to socioeconomic status and overall life chances, can be thought of as a fundamental cause of poor health/health care outcomes in these communities (Williams and Collins 2001). Williams and Collins (2001) describe several specific mechanisms by which racial-ethnic minority segregation may limit life opportunities that impact health/health care outcomes. One such mechanism argues that racial-ethnic minority residential segregation may be related to health outcomes because it can constrain access to a variety of key community resources, including daily necessities such as food, recreation, and public services, and critical resources, such as health care. This perspective focuses on the unequal distribution of resources throughout urban areas and how that inequity may lead to diminished health outcomes for these residents. This is the approach utilized in the study presented here. Recently, there has been an explosion of interest in this mechanism, and such work has shown how different neighborhood characteristics, mainly differentiated by race and class, relate to service providers and community organizations. The bulk of this research has focused on food resources, known as the “food deserts” or “food swamps” literature (Beaulac, Kristjansson, and Cummins 2009; Cooksey-Stowers, Schwartz, and Brownell 2017). This literature has demonstrated that disproportionately poor and minority neighborhoods lack food resources and have demonstrated important differences in quality and price across neighborhoods (Beaulac et al. 2009; Cooksey-Stowers et al. 2017; Moore and Diez Roux 2006; Walker, Keane, and Burke 2010). Although food is the focus of much of this literature, other work has shown a similar pattern for physical fitness centers, park and green space, retail establishments, nonprofit associations, and social services (Allard 2009; Anderson 2017; Gordon-Larsen et al. 2006; Marwell and Gullickson 2013; Small and McDermott 2006). Whereas the work on food and retail has been well studied, there are relatively few sociological studies on the case of health care specifically, which is most relevant to our present study on the COVID-19 vaccine distribution. The studies that exist on health care have generally shown a similar pattern in that Black and Latino segregated areas are less likely to have a wide variety of health care establishments in terms of density or distance to facilities (Anderson 2017; Dai 2010; Dinwiddie et al. 2013; Gaskin et al. 2012; Ko et al. 2014; Rodriguez et al. 2007). Specifically, this work has demonstrated that racial-ethnic minority segregation and socioeconomic variables are related to a lower incidence of physicians’ offices, primary care providers, mental health practitioners, urgent care facilities, auxiliary health care practitioners, surgical centers, and dialysis facilities (Anderson 2017; Dai 2010; Dinwiddie et al. 2013; Gaskin et al. 2012; Ko et al. 2014; Rodriguez et al. 2007). Some limited work has also linked the distribution of health care locations to health and health care outcomes. For example, Dai (2010), in a study of Detroit area neighborhoods, found that Black residents of segregated neighborhoods had fewer facilities that provided mammography services and consequently had higher rates of late-stage breast cancer diagnosis. Another study found that racial-ethnic residential segregation was linked to a greater likelihood of patients seeing nonpsychiatrists for mental health needs (Dinwiddie et al. 2013). A study of the Phoenix area found that the lack of health care service provision in Latino-segregated neighborhoods was linked to a lower likelihood of seeing a personal physician for pediatric care and greater use of clinics (Anderson 2020). Chan et al. (2012) also found that although Black and Latinos living in highly segregated areas had adequate spatial access to facilities, they were less likely to receive certain services, especially specialist care. This work provides some initial evidence for a disparity by racial-ethnic residential segregation in health care service provision. In this analysis, we aim to extend this literature to examine how this lack of adequate health care provision relates to the COVID-19 pandemic and the early rollout of the vaccine. There may be myriad reasons that affect where and how people choose to procure provider services, with physical proximity/convenience being only one of them that may be important to people of limited means, who lack of access to transportation or are elderly/homebound and may find travel difficult. Distance from providers does not create an impenetrable barrier to access (in this case, vaccine access), but it can increase the “friction of distance,” which may be difficult to overcome absent other kinds of individual or household resources (Tobler 1970). Therefore, with this analysis, we are not assuming that people can only or will only seek vaccine access within their local communities but, rather, that having access in close proximity would facilitate their ability to get it, especially for those who are more vulnerable. Segregation and the COVID-19 Pandemic At this stage, limited work has been conducted in a systematic fashion on the COVID-19 pandemic and, more specifically, the vaccine rollout. However, the extant work published over the past year, including media reporting, has demonstrated vast inequalities by race-ethnicity throughout the pandemic. This work has shown that racial-ethnic minority populations have been more vulnerable to COVID-19 disease in terms of rates of infection and mortality (Millett et al. 2020; Ramos and Zamudio 2020). As it relates to segregation specifically, new work has shown that areas with higher numbers of racial-ethnic minorities were more likely to have higher rates of infections and that this pattern was further exacerbated where residential segregation was also high (Yang et al. 2021). Various news reports have also indicated that minority-segregated areas lack vital resources to contend with the spread of the virus, such as testing facilities and health care resources (Garnam and Cai 2021; Godoy and Wood 2020; Martinez 2021; McMinn et al. 2020). Moreover, what is happening now is clearly part of a broader pattern of inequality in infectious disease as it relates to racial-ethnic residential segregation. Previous work has shown important disparities by racial-ethnic residential segregation in terms of infectious disease, such as tuberculosis and sexually transmitted diseases, among others (Acevedo-Garcia 2000, 2001; Biello et al. 2012; Strully 2011). The limited literature that examines past pandemics and social inequality has focused primarily on race-ethnicity and socioeconomic status (Økland and Mamelund 2019; Roberts and Tehrani 2020; Strully 2011). From this limited early work on the topic at hand and the more extensive literature on broader inequalities in health care by segregation, we expect these inequalities by segregation to carry over to the early COVID-19 vaccine rollout. Throughout fall 2020, when the possibility of a vaccine was becoming a reality, public health and medical practice journals published numerous editorials with recommendations for how to effectively and equitably roll out the vaccine (Persad, Peek, and Emanuel 2020). However, states must work within the existing public health and health care infrastructure to provide the vaccine in a manner that fulfills more stringent requirements than most vaccines. Given the existing unequal infrastructure, it would be difficult to imagine that the vaccine distribution could be done equitably. The first weeks of vaccine distribution saw reporting from media outlets questioning the equity of allocation locations (Harper 2021; Oladipo 2021). This remains an empirical question, though, because no systematic work has been conducted on this topic concerning the COVID-19 pandemic. Conceptual Framework and Hypotheses From the theory and extant work, we thus have several hypotheses for this analysis. Given the documented inequalities in community establishments, we expect that the gap in service provision in segregated communities will extend to the case of the COVID-19 vaccine rollout. We examine this in two ways—the number of vaccination sites and the allocation of doses of vaccine. We also examine this at the neighborhood level by examining the geographic clustering of groups in space (details in the following). Thus, we have the following first hypothesis: Hypothesis 1: A higher degree of racial-ethnic minority clustering across urban areas will be related to fewer vaccination sites and fewer vaccine doses. However, we also expect that this will be related to the extant lack of health care provision within these neighborhoods. Thus, we expect that the density of existing health care organizations within urban neighborhoods will attenuate this association. This leads us to our second hypothesis: Hypothesis 2: The density of health care service providers across urban areas will attenuate the association between the racial-ethnic minority clustering and vaccine provision. We test these two hypotheses in a study of early vaccine allocations over a 9-week period in the five largest urban counties in Texas. Data and Methods Data To examine the association between racial-ethnic minority clustering and the vaccine rollout, we combined several sources of area-level data measured at the ZIP (Zone Improvement Plan) area unit of analysis (N = 431). First, for data on vaccine allocations, we used data from the Texas Department of State Health Services. Since the first shipment of the Pfizer-BioNTech vaccine on December 14, 2020, the state of Texas has released information on a weekly basis on the location and quantity of doses allocated across the state by county. We collected these data each week for the first 10 weeks of the vaccine rollout (December 14, 2020, to February 15, 2021). However, we decided to exclude the first week from our analysis because these allocations went only to major hospital sites and were exclusively distributed to first-tier health care workers under Texas’s Phase 1A of the vaccine distribution. We chose these weeks for several reasons. We wanted to focus on the early vaccine rollout when supplies were limited and access was difficult to examine which areas got early access to this privilege. Second, we chose this specific set of weeks due to several circumstances in the weeks that followed. In the subsequent week (Week 11 of the rollout), most of the state of Texas experienced a catastrophic winter storm that affected vaccine distributions both in terms of the ability to make shipments to vaccine sites and the ability to keep those doses properly chilled because of widespread power outages. Following this event, in March, the state began to open up eligibility, and the federal government allocated more funding to the vaccine rollout that dramatically increased vaccine supplies across the state. To analyze the spatial distribution of these sites, an address location was provided for each site, and we geocoded these to a point-level location in ESRI’s ArcMap. Furthermore, because we were primarily interested in how these inequalities relate to urban dynamics, not differences between urban and rural locales, we focused on the five largest counties in the state, which are also the core counties of the five largest cities in the state. These included Bexar County (San Antonio), Dallas County (Dallas), Harris County (Houston), Tarrant County (Fort Worth), and Travis County (Austin). We used counties rather than cities or metropolitan areas because we paired the vaccine data with data on infection rates by ZIP code, which were provided by county health departments. We recognize that other states have provided similar data about their vaccine allocations. However, we chose to focus on a single state because each state has implemented its own criteria for how vaccines should be allocated and who should be eligible to receive them. Texas, in particular, allocated vaccines to sites based on two pieces of information: an advisory panel, which made recommendations for allocations, and a registration system for vaccine providers to determine eligibility and feasibility. Texas appointed a team of subject matter experts onto an Expert Vaccine Advisory Panel to develop vaccine allocation strategies as recommendations to the Texas Commissioner of Health. Information from the CDC and an appointed Advisory Committee on Immunization Practices helped develop these strategies. Vaccine equity was one of the stated goals of this panel, but the exact algorithm used by the panel is not publicly available. On the supply side of this question, vaccine provider registration data aided in determining the physical locations and quantities for distribution. According to the CDC provider agreement, registration data included licensure information, patient population numbers, and other logistical details required to ensure each facility’s ability to store and administer the vaccines. Although we limited our analysis to this one state, we argue that Texas serves as a good test case. Part of this rationale is practical—Texas provided detailed information on the location and quantities of vaccines in a publicly available format. It is also the only state with that many major U.S. cities (5 in the top 20 largest cities in the United States) and that also has a high degree of racial-ethnic diversity across all three of the largest racial-ethnic groups in the United States. Although Texas cities do not have the highest rates of residential segregation, certain Texas cities are fairly segregated, and there is quite a bit of variation across cities to allow for comparison. Full descriptive statistics for variables used in our analysis and the racial-ethnic breakdown for each county (divided by county) can be found in Appendix A in the online version of the journal for reference. We combined these data with sociodemographic data and data on establishments from two census products. First, for sociodemographic data, including our measures of racial-ethnic clustering, we used the 2014 to 2018 American Community Survey 5-year estimates at the ZIP code tabulation area level. The Census Bureau only provides data at this small geographic unit using 5-year aggregates because the data are not representative for small units of analysis like the ZIP code for a single year. We also combined this with data on establishments from the 2016 County Business Patterns (CBP) ZIP Code Industry Detail File. The CPB uses IRS tax records to provide counts of establishments by ZIP code and by industry type using the North American Industry Classification System (NAICS). We included several industry codes related to health care, which we hypothesized may be related to higher vaccine allocations (details in the following). Although 2016 is not the most recent year of data publicly available (2018), starting in 2017, the Census Bureau stopped releasing counts of establishments where the count for the ZIP code was less than three to further deidentify the information. Because we included several industry classifications that are relatively rare, such as hospitals, we opted to use the older version of the data to get better estimates of the available resources in neighborhoods.1 Across all models, we also included a control variable for the cumulative number of infections in the ZIP code for all five of our counties. These came from publicly available sources from each county’s public health department. These were collected in each of the five counties on February 9, 2021, which was the second to last week of our vaccine rollout time frame such that it could have affected allocation decisions. We included this as a covariate in the event that state health officials were using the infection rate as the basis by which vaccines were being allocated by factoring in local vulnerability. We included two dependent variables, each meant to capture a different facet of the vaccine rollout across urban areas in the state. First, we examined the number of vaccine sites per 100,000 people in a ZIP code. This came from the geocoded locations for vaccine allocations and reflected a simple count of the locations providing COVID-19 vaccines. We also divided this number by the population size and multiplied by 100,000 to derive a rate per 100,000 people. Beyond the mere availability of a vaccine site, we also examined the distribution of vaccine doses per 10,000 people to each site for each week of the 9-week period. The state provided data on the week-by-week allocations to each of the geocoded locations. To derive a rate, we divided this number by the population and multiplied by 10,000. Our main independent variables in this analysis included a set of variables for racial-ethnic clustering. Although much of the literature on racial-ethnic segregation is focused on measuring global segregation across a large area, such as the county or metropolitan area (Massey and Denton 1988), in this study, we were interested in coding for which areas within the county have disproportionately high numbers of certain groups. Typically, studies that examine racial-ethnic concentration at a smaller geographic unit of analysis use composition scores, which reflect simply the percentage of a group over a certain area.2 However, this approach is aspatial and ignores the role of geographic clustering across space and how adjacent areas may influence each other (Reardon and O’Sullivan 2004; Roberto 2018). Thus, for this analysis, we used a geographic clustering score that considered two pieces of information: the concentration of a group in an area (ZIP codes) and the extent to which that group is geographically clustered. To be more precise, we refer to this as clustering rather than segregation throughout the discussion of this analysis, and we conceptualized this as a neighborhood-level measure of segregation. We used the following formula: Ci=xi∑j=1,j≠inwijxj, where xi is the variable for ZIP code i, xj is the variable for ZIP code j, and wij is the spatial weight between ZIP codes i and j (Anderson 2017). The measure is essentially the product of the percentage of a certain group in a ZIP code and the spatial weight of its neighborhoods (row standardized). We used a queen contiguity matrix to calculate the spatial weight. This produces a theoretical range of 0 to 10,000. For example, a ZIP code could have a score of 10,000 if that ZIP code contained 100% of its residents from a certain group and all adjacent neighborhoods also had a population composition of 100% of the same group. In practice, no ZIP code in Texas has this high of a score for any group, and the scores vary considerably depending on the county and group in question. For this reason, we also group mean centered (to the county) each of these scores to make them relative to the population sizes of the county. Otherwise, these measures might reflect differences in the relative sizes of these groups across areas rather than differences within a particular county in terms of how these groups are spatially patterned. For example, San Antonio is 64.2% Latino, whereas Austin is only 33.9% Latino. In each of these contexts, what might be considered a disproportionately Latino community would be different, and group mean centering would contextualize this difference. This is an approach used in previous work examining neighborhoods (Sampson, Raudenbush, and Earls 1997). For this study, we included three scores: clustering measure for percentage Black (non-Latino), clustering measure for percentage Latino (of any race), and clustering measure for percentage Asian (non-Latino).3 To examine whether these patterns relate to existing disparities in resource distribution, we also included four variables for counts of organizations that are more likely to receive vaccine allocations. These variables came from the CBP data set, which classifies establishments by industry code using NAICS codes. In this analysis, we included counts by ZIP code of general hospitals (622///), physicians’ offices (6211//), pharmacies (446110), and retirement and assisted living communities for the elderly (6233//).4 We also included several control variables to account for populations that are eligible to receive the vaccine earlier on the priority list, vulnerability to the disease, and other sociodemographic factors. These included population density, percentage age 65 and above, percentage of people employed in service occupations, median family income, percentage of people with a bachelor’s degree or higher, percentage of households with no private vehicle, and cumulative number of infections in a ZIP code. Descriptive statistics for all variables can be found in Table 1. Table 1. Descriptive Statistics for Variables Used in Statistical Models. Variable Name Mean SD Range Description Dependent variables Vaccine sites 7.92 17.95 0 to 169.71 Number of vaccine sites per 100,000 people in a ZIP code Vaccine doses 1,199.01 8,295.32 0 to 238,415.58 Number of vaccine doses allocated to ZIP code per 10,000 people Independent variables Black clustering 0 648.99 −700.47 to 4,150.46 Clustering measure of % Black Latino clustering 0 1,611.93 −2,986.80 to 5,851.91 Clustering measure of % Latino Asian clustering 0 84.01 −73.08 to 812.42 Clustering measure of % Asian Population density 3,291.17 2,359.18 6.83 to 16,811.43 Population density per square mile Population age 65+ 10.94 4.13 0 to 36.60 % of population age 65 and above % Service work 17.01 6.43 0 to 43.40 % of employed population in service occupations Household income 68,094.12 31,467.35 17,798 to 240,417 Median household income % Bachelor’s degree 32.96 21.30 2.60 to 90.00 % of population over 25 with at least a bachelor’s degree % No vehicle 6.10 5.38 0 to 39 % of households with no private vehicle Infections 2,127.59 1,652.52 0 to 9,906 Number of positive SARS-CoV-2 infections Hospitals .61 1.46 0 to 15 Number of general hospitals Physicians’ offices 26.84 41.99 0 to 294 Number of physicians’ offices Pharmacies 3.89 3.39 0 to 18 Number of pharmacies Retirement communities 1.41 1.88 0 to 14 Number of retirement and assisted living facilities for the elderly Note: N = 431. Data come from the Texas Department of State Health Services, the 2014–2018 American Community Survey, and the 2016 County Business Patterns. Methods To model these two different dependent variables, we present two sets of models, one for each outcome. First, for the number of vaccine distribution sites, we estimated a series of spatial error models. We calculated univariate global Moran’s I statistics for our key independent variables that suggested that spatial autocorrelation was a problem and would therefore not meet the assumptions of ordineary least squares (OLS) regression. Furthermore, the LaGrange multiplier statistics indicated that the spatial error model was the most appropriate method to contend with this autocorrelation (Anselin, Florax, and Rey 2004). Specifically, we used a queen spatial weight matrix because this was found to best maximize global Moran’s I for each of the key variables used in the models (Anselin 1995; Anselin, Florax, and Rey 2004). We also included Kelejian and Prucha (2010) robust standard errors to account for significant heteroscedasticity. However, of note, the term for lambda is not significant in the models presented in Table 2, suggesting that correlated errors in omitted variables may not be particularly a problem here. The OLS results were also virtually identical to what is presented here in terms of the sign, significance, and relative effect sizes for each of our variables. Because spatial autocorrelation was significant in our preliminary analyses, though, we chose to present the results as spatial error models. These results can be found in Table 2. Table 2. Coefficients and Z-Ratios from Spatial Error Models of Vaccination Sites per 100,000 People. Variable Name Model 1 Model 2 Model 3 Model 4 Model 5 Black clusteringa −.535* (−3.299) −.503 (−3.044) −.461 (−2.819) −.516 (−3.091) −.547*** (−3.326) Latino clusteringa −.137* (−2.185) −.101 (−1.530) −.104 (−1.625) −.120† (−1.793) −.149* (−2.315) Asian clusteringa .474 (.383) .198 (.226) .100 (.097) .230 (.187) .565 (.446) Population density −.001 (−2.729) −.001 (−2.687) −.002 (−3.033) −.002 (−2.845) −.001 (−2.680) Population age 65 and up .099 (.353) .173 (.594) .085 (.291) .105 (.377) .204 (.691) % Service work −.426† (−1.767) −.391† (−1.697) −.406† (−1.711) −.418† (−1.723) −.413† (−1.713) Median household incomea −.073 (−1.380) −.029 (−.629) −.053 (−1.137) −.068 (−1.318) −.081 (−1.482) % Bachelor’s degree .172† (1.826) .102 (1.206) .100 (1.069) .164† (1.757) .187† (1.897) % No vehicle 2.193* (4.175) 2.014* (3.726) 2.040* (3.887) 2.180* (4.122) 2.153* (4.132) Infectionsa .004 (.100) −.015 (−.386) −.024 (−.598) −.030 (−.754) .022 (.533) Hospitals 3.290** (2.866) Physicians’ offices .097* (2.383) Pharmacies .419† (1.677) Retirement communities −.719* (−2.387) Lambda −.935 (−.765) −1.263 (−1.000) −.740 (−.603) −.775 (−.640) −.660 (−.540) Pseudo R2 .345 .408 .384 .349 .350 Note: N = 431. Data come from the Texas Department of State Health Services, the 2014–2018 American Community Survey, and the 2016 County Business Patterns. a Coefficient multiplied by 100 for the ease of presentation. † p < .1, p < .05, p < .01, p < .001 (for two-tailed test). For the second set of models for the week-by-week allocation of vaccine doses, we present a series of hierarchical linear growth models (Raudenbush and Bryk 2002; Singer and Willett 2003) with a correction for spatial dependency using a queen contiguity spatial weight matrix (Savitz and Raudenbush 2009). Given the nested structure of the data, weekly allocations per ZIP code, we used this approach to model the cumulative change over time in the number of vaccine doses per ZIP code. However, because the data were still organized by physically adjacent spatial units of analysis at Level 2, we used Savitz and Raudenbush’s (2009) routine to account for spatial dependency using HLM 8.1, which is an approach used in similar health research (O’Connell 2015). These results can be found in Table 3. Table 3. Coefficients and T-Ratios from Linear Growth Models of Weekly Vaccination Allocations per 10,000 People. Variable Name Model 1 Model 2 Model 3 Model 4 Model 5 Fixed effects Week 312.927* (8.553) 312.927* (8.553) 312.927* (8.553) 312.927* (8.553) 312.927* (8.553) Black clustering −1.014* (−1.984) −.820† (−1.785) −.593 (−1.215) −.907† –1.771) −1.036* (−2.029) Latino clustering −.424† (−1.657) −.204 (−.886) −.250 (−1.026) −.345 (−1.337) −.465† (−1.810) Asian clustering 4.731 (1.256) 3.212 (.949) 2.669 (.745) 3.597 (.947) 4.956 (1.317) Population density −.376* (−2.355) −.347* (−2.415) −.467 (−3.072) −.437 (−2.687) −.352* (−2.190) Population age 65 and up −141.990† (−1.841) −90.095 (−1.298) −150.614* (−2.060) −140.289† (−1.825) −109.474 (−1.364) % Service work −154.071† (−1.941) −138.248† (−1.939) −138.666† (−1.842) −148.326† –1.873) −147.523† (−1.858) Median household income −.038* (−2.109) −.012 (−.752) −.027 (−1.608) −.035* (−1.970) −.040* (−2.250) % Bachelor’s degree 42.995 (1.348) 2.743 (.095) 6.362 (.207) 39.659 (1.245) 48.578 (1.513) % No vehicle 485.960* (6.382) 384.743* (5.568) 401.661* (5.487) 478.015* (6.288) 471.639* (6.149) Infections −.223 (−1.132) −.323† (−1.826) −.354† (−1.891) −.371† (−1.759) −.156 (−.774) Hospitals 1855.981* (10.099) Physicians’ offices 50.823*** (6.954) Pharmacies 192.471† (1.908) Retirement communities −241.504 (−1.440) Random effects Level 2 error variance 30,132,845.834 23,546,583.226 26,689,874.927 29,920,526.251 30,046,250.421 Level pseudo R2 .146 .333 .244 .152 .149 Note: Level 1 N = 3,879. Level 2 N = 431. Data come from the Texas Department of State Health Services, the 2014–2018 American Community Survey, and the 2016 County Business Patterns. The Level 2 pseudo R2 is calculated from the proportional reduction in error variance from a model with no Level 2 variables (τ = 35,296,518.409). † p < .1, p < .05, p < .01, **p < .001 (for two-tailed test). For each set of models, we first present a model with only the clustering scores and ZIP-code-level control variables included (Model 1). Then, we add the health care resource variables one by one in the model (Models 2–5) to avoid the problems of multicollinearity and to isolate the effect of each organizational type as health care organizations tend to agglomerate. To our knowledge, no formal mediation test exists that can account for the spatial dependencies in the models and the multilevel structure of the second set of models. Therefore, we used an informal approach and examine change in the effect sizes with the inclusion of certain variables. Results From the results in Table 2, we can see that the racial-ethnic clustering scores have a significant relationship to the density of vaccine sites. Specifically, for Black and Latino clustering, these scores are significant and negative, meaning that as the clustering of these two groups increases, the number of vaccine sites per 100,000 people decreases. Essentially, the higher the concentration and clustering of these two groups, the lower the number of vaccine sites. These effect sizes are also large. We discuss these coefficients in terms of standard deviation changes because the scale of each variable is so different, although the original coefficients are available in the tables. In the case of Black clustering, a 1 SD (648.99) increase in Black clustering relates to 3.47 decrease in the number of vaccine sites per 100,000 people. Given that the average number of sites per 100,000 people is only 7.92, this is a notable change. For Latino clustering, a 1 SD (1,611.93) increase in Latino clustering is related to a 1.59 decrease in the number of vaccine sites per 100,000 people. These are both sizable coefficients and indicate that Black- and Latino-clustered areas are less likely to have vaccine sites. Several of the other area-level coefficients are significant as well. Population density is significant and negative, the percentage of households with no car is significant and positive, and two others, percentage college educated and percentage in service work, are significant at the .1 level. The largest effect comes from the percentage of households with no car, where a 1 SD increase (5.38) relates to an increase of 11.8 vaccine sites. This seems to run counter to what we would expect from the literature because private vehicle ownership may relate to socioeconomic status, but it may also be an indication of central city location where residents may perceive less of a need to own a vehicle. Taken together, from the pseudo R2 value, these sociodemographic characteristics of ZIP codes explain 34.5% of the variation in vaccine sites per 100,000 people, which is sizable. In Models 2 to 5, when we add the health care establishment variables, this pattern changes somewhat. First, when we include two such health care establishments, the number of hospitals and the number of physicians’ offices, these are both significant and positive. This indicates that having more health care establishments in a ZIP code relates to a greater number of vaccine distribution sites. This is a notable increase as well, where each additional hospital is related to an increase of 3.29 vaccination sites and each additional physician’s office is related to a .10 increase in sites. This is expected because hospitals and clinics are where most of the vaccine doses were allocated. The addition of these two variables also reduces the size of the coefficients for the clustering variables, suggesting that the lower number of vaccine sites in these areas is a function of their lack of health care resources. In the case of Latino clustering, the coefficients drop to nonsignificance with the inclusion of hospitals and physicians’ offices, and these changes relate to a 26.48% and a 23.99% reduction, respectively, in the size of the coefficients for Latino clustering. This percentage change in the coefficient size is somewhat smaller for Black clustering at 9.09% and 13.77%, respectively. This is reflected in their pseudo R2 values as well, where hospitals alone explain an additional 6.33% of the variation in vaccine sites and physicians’ offices explain 3.96%. This suggests that existing health care resources in areas may explain some part of the negative association between minority clustering in ZIP codes and the number of vaccine sites. However, these findings are only significant at the .1 level for pharmacies and significant and negative for retirement communities. Therefore, the bulk of this effect seems to be driven by hospitals and physicians’ offices rather than other kinds of facilities that may offer the vaccine. We also illustrate these patterns graphically in a series of maps for each county. These can be found in Figures 1 to 5. In the background of each map is a choropleth quintile map of the two clustering scores with the location of the vaccine sites overlaid on top. Choropleth maps use shading to map patterns across a polygonal area. Here, we use a quintile map, meaning that the range of mapped values in the polygons include an equal number of ZIP codes for each shade. Note that because these are evenly distributed quintile maps, the scale is different for each group and location. From these maps, it is clear that minority clustered areas are less likely to have vaccine sites, but these patterns differ somewhat by county, with more stark patterns in certain counties over others and depending on the group in question. For example, Harris and Travis counties each have a clearer clustering pattern to them where the vaccine sites appear to be less likely to be located in racial-ethnic minority areas. By contrast, Tarrant County (Fort Worth) has a more scattered pattern to the location of the facilities. Figure 1. Vaccine Distribution Sites over Racial-Ethnic Clustering Scores in Bexar County (San Antonio), Texas. Note: Data come from the Texas Department of State Health Services and the 2014–2018 American Community Survey. Figure 2. Vaccine Distribution Sites over Racial-Ethnic Clustering Scores in Dallas County (Dallas), Texas. Note: Data come from the Texas Department of State Health Services and the 2014–2018 American Community Survey. Figure 3. Vaccine Distribution Sites over Racial-Ethnic Clustering Scores in Harris County (Houston), Texas. Note: Data come from the Texas Department of State Health Services and the 2014–2018 American Community Survey. Figure 4. Vaccine Distribution Sites over Racial-Ethnic Clustering Scores in Tarrant County (Fort Worth), Texas. Note: Data come from the Texas Department of State Health Services and the 2014–2018 American Community Survey. Figure 5. Vaccine Distribution Sites over Racial-Ethnic Clustering Scores in Travis County (Austin), Texas. Note: Data come from the Texas Department of State Health Services and the 2014–2018 American Community Survey. In Table 3, we present the results for the number of vaccine allocations over the 9-week period using a series of linear growth models. The results with this dependent variable reflect a similar pattern to the one described previously with some notable differences. Again, in Model 1, the coefficient for Black clustering is significant and negative, indicating that the higher the degree of concentration and clustering of Blacks across these five counties, the smaller the vaccine allocations week to week. Specifically, every 1 SD increase in Black clustering (648.99) relates to 658.08 fewer vaccine doses per 10,000 people in a ZIP code. The same figure for a 1 SD change in Latino clustering (1,611.93) is 492.66 vaccine doses per 10,000 people, which is not an inconsequential amount, although this coefficient is only significant at the .1 level for Latino clustering. In all subsequent models, the results for Latino clustering are not significant. Concerning the control variables, the results are similar to previous models except that median household income is significant and negative here. When we add the different types of health care establishments to the baseline model, the results are similar in pattern to the results previously described. Once again, the distribution of hospitals and physicians’ offices is significant and positive, meaning that their presence in an area is related to a higher allocation of vaccine doses per 10,000 people. Moreover, the inclusion of these variables weakens the association between the clustering scores and the number of vaccine doses per 10,000 people, especially in the case of Black clustering. Black clustering is only significant in Model 1 and Model 5, and the inclusion of hospitals and physicians’ offices rather substantially reduces the size of those coefficients. These reductions are percentage changes of 19.13%, 41.52%, and 10.55% for hospitals, physicians’ offices, and pharmacies, respectively. Hospitals and physicians’ offices appear to explain away a substantial portion of the differential allocations across areas related to Black clustering. We also see changes in the size of the coefficients for Latino clustering, but this score was only significant at the .1 level in Model 1. Discussion In this study, we aim to understand how neighborhood sociodemographic characteristics relate to the early COVID-19 vaccine rollout. We hypothesize that racial-ethnic minority clustered areas will be less likely to have vaccination sites, and we surmise that this is primarily a function of a lack of key health care sites prior to the pandemic. We test these hypotheses across the five largest urban counties in the state of Texas during the first 10 weeks of the vaccine rollout in Texas (excluding the first week). Moreover, we test this using two ways of capturing access to the vaccine for neighborhoods—the number of vaccine sites and the number of doses allocated to each site. For the first outcome, examining the number of vaccine sites across these urban counties, we find that a higher concentration and clustering of Black and Latino residents in ZIP codes is associated with fewer vaccine sites per 100,000 people. These results provide support for Hypothesis 1 for Black and Latino clustering (not Asian clustering). Moreover, this negative association is partially explained by the existing distribution of health care resources across communities, especially hospitals and physicians’ offices. The presence of these establishments is associated with a higher number of vaccine sites, so the lack of these sites in segregated communities means fewer vaccine sites for these areas. However, we do not observe the same association for pharmacies and retirement communities. Indeed, the association for retirement communities is actually significant and negative and appears to have a suppression effect with the racial-ethnic clustering scores. We speculate that this may be because these groups are less likely to use retirement homes for elderly relatives as other work suggests (Dilworth-Anderson, Williams, and Gibson 2002), although we cannot directly test this assertion with the data here. These findings provide partial support for Hypothesis 2, with stronger evidence for the case of Latino clustering. For the second outcome, we find a similar pattern, although with somewhat weaker effects. Black clustering is again significant and negatively associated with the number of vaccine doses per 10,000 people in a ZIP code over a 9-week period (although this is only significant for Latino clustering at the .1 level), meaning that minority segregated communities were less likely to receive doses of vaccine over this period. This provides further partial support for Hypothesis 1 in the case of Black clustering. And, again, these results are somewhat attenuated with the inclusion of the number of hospitals and physicians’ offices. Both these types of establishments are strongly associated with a higher number of vaccine doses, and they reduce the size of the coefficient for Black clustering. Thus, we find partial support for Hypothesis 2. These findings are fitting with previous literature on segregation and the distribution of resources more broadly that demonstrates poor access to a wide variety of community establishments that would support well-being (Anderson 2017; Dinwiddie et al. 2013; Gaskin et al. 2012; Ko and Ponce 2013). We find this for Black and Latino clustering but not for Asian clustering. However, the existing work on Asian segregation is limited and has not found the same inequalities as Black and Latino segregations (Anderson 2017). The findings presented here also point to a much stronger race story than one about socioeconomic dynamics, which is a prominent theme in much of the previous literature on resource allocation across neighborhoods (Beaulac et al. 2009; Ko et al. 2014). In our study, area-level median income was not significant in the first analysis, and in the second analysis, it was significant and negative, meaning that higher income areas were less likely to receive doses. Thus, the story seems to be more about race specifically rather than class, which limited research has also shown (Anderson 2017; Small and McDermott 2006). Given the lack of detail provided by the state in terms of how specifically the vaccine allocation algorithm was designed, we are not necessarily suggesting that this is an act of blatant and purposive racism on the part of the advisory board responsible for vaccine allocation decision-making. Rather, because the infrastructure in terms of health care resources already disadvantages racial-ethnic minority communities, this represents a case of structural discrimination. Thus, we extend this literature by demonstrating yet another empirical instance of a racial disparity, especially for a hugely impactful event like the current COVID-19 crisis. Some limited findings on the COVID-19 pandemic, including media sources, has revealed important disparities by race-ethnicity in infection rates and access to medical services, such as testing sites and ICU facilities (CDC 2020b; McMinn et al. 2020; Miller, Peek, and Parker 2020; Millett et al. 2020; Ross 2021). The vaccine rollout appears to fit within this pattern. This is relevant to the current inequities we have seen throughout the pandemic and has implications for the next crisis. Despite these advances to the literature, the study has several limitations. First, the study only relates the sociodemographic characteristics of neighborhoods with the distribution of vaccine sites and doses and not who is being vaccinated at these sites. Media reporting on access to the vaccine has highlighted how differential access to the Internet to search for vaccine appointments, time to spend searching for open appointments, and adequate transportation have led to important disparities in early access to the vaccine, even for eligible populations (Garnam and Cai 2021; Harper 2021; Menchaca and Agnew 2021; Oladipo 2021; Ross 2021; Stone 2021). This study only factors in disparities in the built environment in terms of the location of vaccine sites without accounting for who is going to these sites. Second, the data from the Texas Department of Health Services only includes address locations for where vaccine doses were shipped. Although the vast majority of shipping locations were the same as where the vaccines were administered, this approach fails to consider mobile units that some city and county health departments have utilized to reach special populations, like mobility-challenged patients in elder care facilities. These types of activities were not possible to systematically track over time. Future work should consider these limitations. With this study, we contribute to the literature on the unequal distribution of resources across neighborhoods, particularly by racial-ethnic minority clustering. Moreover, we demonstrate that these patterns are not a neutral fact. Not having health care infrastructure in place means that when confronted with a public health catastrophe, the existing inequalities in our health care system are deepened. The COVID-19 pandemic is, for many people, one of the most disruptive and challenging public health events of our lifetimes. Vaccination in this context represents a lifeline to spare further human suffering and loss of life and a potential return to normalcy. However, this valuable resource was not distributed evenly across urban areas, with limited access to populations already at risk for complications from the virus. Although state public health officials implemented eligibility systems that prioritized health care workers, the elderly, and those with medical comorbidities, the geographic component of the vaccine rollout and allocation of the vaccine over time has not been equal. This highlights the necessity of creating more equitable access to care broadly so that in crisis times, the infrastructure is available to equitably meet the needs of the affected communities. Supplemental Material sj-docx-1-hsb-10.1177_00221465221074915 – Supplemental material for Racial-Ethnic Residential Clustering and Early COVID-19 Vaccine Allocations in Five Urban Texas Counties Click here for additional data file. Supplemental material, sj-docx-1-hsb-10.1177_00221465221074915 for Racial-Ethnic Residential Clustering and Early COVID-19 Vaccine Allocations in Five Urban Texas Counties by Kathryn Freeman Anderson and Darra Ray-Warren in Journal of Health and Social Behavior Author Biographies Kathryn Freeman Anderson is an associate professor in the Department of Sociology at the University of Houston. Her research focuses on understanding the social sources of health disparities in the United States. In particular, she examines the role of race-ethnicity and urban neighborhood dynamics to analyze how these factors may affect individual health. Her recent work has been published in Social Problems, City & Community, and Race and Social Problems. Darra Ray-Warren is a graduate student in the Department of Sociology at the University of Houston. Her research interests are in the socioeconomic determinants affecting mental and physical health, social status, and the public policies that may impact human flourishing. Her current research focuses on criminal law and punishment rationale, emerging progressive prosecutorial and indigent defense styles, and the social factors that drive criminal case outcomes. ORCID iD: Kathryn Freeman Anderson https://orcid.org/0000-0002-7425-8923 Supplemental Material: Appendix A is available in the online version of the article. 1. We also ran the same models using the 2018 version of the data, which is the most recent version available but likely has undercounts of establishments in many areas. The results generally followed the same pattern as what is presented here but with different effect sizes in some cases. As an additional sensitivity check, we also checked previous waves of the data to see how correlated the counts are over time to justify the use of an earlier date. Using the 2012 CBP file (being four years before our data, which is the same distance from 2016 to 2020), we found that all four of our organizational types were highly correlated. From 2012 to 2016, hospitals, physicians’ offices, pharmacies, and retirement communities had a correlation of .93, .98, .86, and .86, respectively, indicating that these organizational resources are relatively stable over time, especially health care provision. 2. Our analysis focused on these clustering scores as a more geographically informed way of examining the problem. To relate these findings to the broader literature that typically uses composition scores, we ran the same models using percentage non-Latino Black, percentage Latino, percentage non-Latino Asian, and another version of these models’ variables that also included their spatial W lags (which is the other term in the clustering equation). The results were similar to what we found in the present study, with some differences in the effect sizes (results available on request). We chose to present the results with the clustering score as indicated by the formula because we think this best captures the spatial dynamics of neighborhood-level segregation in a manner that accounts for both the composition of groups in an area and the extent to which they are spatially clustered, which is one of the main ways that segregation is theorized and conceptualized in the literature (Massey and Denton 1988). 3. The distribution of Whites also plays a role in the segregation level of an area. However, due to multicollinearity, we could not include all four clustering scores for each of these groups in a single model. Instead, we focused on the distribution of racial-ethnic minority populations across these counties. As a check on this choice of approach, we also ran the same models using the clustering score for percentage White. This score for Whites was not significant in any of the models, suggesting that the distribution of Whites across areas is unrelated to the distribution of vaccine sites and allocations. 4. We also tested several other organizational types that might have received vaccine allocations, including general stores (452///), supermarkets (4451//), freestanding ambulatory care facilities (621493), and other (nonelderly) nursing and residential care facilities (623///). We only present the results for those that were most relevant and had significant results. ==== Refs References Acevedo-Garcia Dolores . 2000. “Residential Segregation and the Epidemiology of Infectious Diseases.” Social Science & Medicine 51 (8 ):1143–61. Acevedo-Garcia Dolores . 2001. “ZIP Code-Level Risk Factors for Tuberculosis: Neighborhood Environment and Residential Segregation in New Jersey, 1985–1992.” American Journal of Public Health 91 (5 ):734–41. Acevedo-Garcia Dolores Lochner Kimberly Osypuk Theresa Subramanian S. V. 2003. “Future Directions in Residential Segregation and Health Research: A Multilevel Approach.” American Journal of Public Health 93 (2 ):215–21. Allard Scott W. 2009. Out of Reach: Place, Poverty, and the New American Welfare State. New Haven, CT: Yale University Press. Anderson Kathryn Freeman . 2017. “Racial Residential Segregation and the Distribution of Health-Related Organizations in Urban Neighborhoods.” Social Problems 64 (2 ):256–76. Anderson Kathryn Freeman . 2020. “Residential Segregation, Neighborhood Health Care Organizations, and Children’s Health Care Utilization: The Case of the Phoenix Urbanized Area.” City & Community 19 (3 ):771–801. Anderson Kathryn Freeman Fullerton Andrew . 2014. “Residential Segregation, Health, and Health Care: Answering the Latino Question.” Race and Social Problems 6 (3 ):262–79. Anselin Luc . 1995. “Local Indicators of Spatial Association—LISA.” Geographical Analysis 27 (2 ):93–115. Anselin Luc Florax Raymond Rey Sergio . 2004. Advances in Spatial Econometrics: Methodology, Tools and Applications. Berlin: Springer-Verlag. Beaulac Julie Kristjansson Elizabeth Cummins Steven . 2009. “A Systematic Review of Food Deserts, 1966–2007.” Preventing Chronic Disease 6 (3 ):1–10. Biello Katie Kershaw Trace Nelson Robert Hogben Matthew Ickovics Jeannette Niccolai Linda . 2012. “Racial Residential Segregation and Rates of Gonorrhea in the United States, 2003–2007.” American Journal of Public Health 102 (7 ):1370–77. Centers for Disease Control and Prevention. 2020a. “CDC’s COVID-19 Vaccine Rollout Recommendations.” https://www.cdc.gov/coronavirus/2019-ncov/vaccines/recommendations.html. Centers for Disease Control and Prevention. 2020b. “COVID-19 in Racial and Ethnic Minority Groups.” https://www.cdc.gov/coronavirus/2019-ncov/need-extra-precautions/racial-ethnic-minorities.html. Chan Kitty Gaskin Darrell Dinwiddie Gniesha McCleary Rachael . 2012. “Do Diabetic Patients Living in Racially Segregated Neighborhoods Experience Different Access and Quality of Care?” Medical Care 50 (8 ):692–99. Collins Chiquita . 1999. “Racism and Health: Segregation and Causes of Death Amenable to Medical Intervention in Major U.S. Cities.” Annals of the New York Academy of Sciences 896 :396–98. Collins Chiquita Williams David . 1999. “Segregation and Mortality: The Deadly Effects of Racism?” Sociological Forum 14 (3 ):495–523. Cooksey-Stowers Kristen Schwartz Marlene Brownell Kelly . 2017. “Food Swamps Predict Obesity Rates Better Than Food Deserts in the United States.” International Journal of Environmental Research and Public Health 14 (11 ):1366–86. Dai Dajun . 2010. “Black Residential Segregation, Disparities in Spatial Access to Health Care Facilities, and Late-Stage Breast Cancer Diagnosis in Metropolitan Detroit.” Health & Place 16 (5 ):1038–52. Dilworth-Anderson Peggye Williams Ishan Canty Gibson Brent . 2002. “Issues of Race, Ethnicity, and Culture in Caregiving Research: A 20-Year Review (1980–2000).” The Gerontologist 42 (2 ):237–72. Dinwiddie Gniesha Gaskin Darrell Chan Kitty Norrington Janette McCleary Rachel . 2013. “Residential Segregation, Geographic Proximity and Type of Services Used: Evidence for Racial-Ethnic Disparities in Mental Health.” Social Science & Medicine 80 :67–75.23312305 Garnam Juan Pablo Cai Mandi . 2021. “Advocates Worry Vaccines Will Be out of Reach for Black and Hispanic Neighborhoods Devastated by COVID-19.” The Texas Tribune, January 9. https://www.texastribune.org/2021/01/09/texas-coronavirus-vaccine-racial-inequality/. Gaskin Darrell J. Dinwiddie Gniesha Chan Kitty McCleary Rachael . 2012. “Residential Segregation and the Availability of Primary Care Physicians.” Health Services Research 47 (6 ):2353–76. Gaskin Darrell J. Price Adrian Brandon Dwayne LaVeist Thomas . 2009. “Segregation and Disparities in Health Services Use.” Medical Care Research and Review 66 (5 ):578–89. Godoy Maria Wood Daniel . 2020. “What Do Coronavirus Racial Disparities Look like State by State?” National Public Radio, May 30. https://www.npr.org/sections/health-shots/2020/05/30/865413079/what-do-coronavirus-racial-disparities-look-like-state-by-state. Gordon-Larsen Penny Nelson Melissa Page Phil Popkin Barry . 2006. “Inequality in the Built Environment Underlies Key Health Disparities in Physical Activity and Obesity.” Pediatrics 117 (2 ):417–24. Grady Sue . 2006. “Racial Disparities in Low Birthweight and the Contribution of Residential Segregation: A Multilevel Analysis.” Social Science & Medicine 63 (12 ):3013–29. Harper Karen Brooks . 2021. “As Vaccine Eligibility Widens, Some Vulnerable Texans Are Still Fighting for Access.” The Texas Tribune, March 31. https://www.texastribune.org/2021/03/31/vaccine-eligibility-texas/. Hart Kevin Kunitz Stephen Sell Ralph Mukamel Dana . 1998. “Metropolitan Governance, Residential Segregation, and Mortality among African Americans.” American Journal of Public Health 88 (3 ):434–38. Kelejian Harry Prucha Ingmar . 2010. “Specification and Estimation of Spatial Autoregressive Models with Autoregressive and Heteroskedastic Disturbances.” Journal of Econometrics 157 (1 ):53–67. Ko Michelle Needleman Jack Derose Kathryn Pitkin Laugesen Miriam Ponce Ninez . 2014. “Residential Segregation and the Survival of U.S. Urban Public Hospitals.” Medical Care Research and Review 71 (3 ):243–60. Ko Michelle Ponce Ninez . 2013. “Community Residential Segregation and the Local Supply of Federally Qualified Health Centers.” Health Services Research 48 (1 ):253–70. Lee Min-Ah A. 2009. “Neighborhood Residential Segregation and Mental Health: A Multilevel Analysis on Hispanic Americans in Chicago.” Social Science & Medicine 68 (11 ):1975–84. Martinez A. 2021. “The Racial Disparities, Systemic Racism Behind Who Has Received Vaccines.” National Public Radio, March 18. https://www.npr.org/2021/03/18/978496045/the-racial-disparities-systemic-racism-behind-who-has-received-vaccines. Marwell Nicole Gullickson Aaron . 2013. “Inequality in the Spatial Allocation of Social Services: Government Contracts to Nonprofit Organizations in New York City.” Social Service Review 87 (2 ):319–53. Massey Douglas Denton Nancy . 1988. “The Dimensions of Residential Segregation.” Social Forces 67 (2 ):281–315. McFarland Michael Smith Cheryl . 2011. “Segregation, Race, and Infant Well-Being.” Population Research and Policy Review 30 (3 ):467–93. McMinn Sean Carlsen Audrey Jaspers Bret Talbot Ruth Adeline Stephanie . 2020. “In Large Texas Cities, Access to Coronavirus Testing May Depend on Where You Live.” National Public Radio, May 27. https://www.npr.org/sections/health-shots/2020/05/27/862215848/across-texas-black-and-hispanic-neighborhoods-have-fewer-coronavirus-testing-sit. Menchaca Megan Agnew Duncan . 2021. “Texas’ Decentralized Internet Reliant System for Vaccine Appointments Leaves Many Eligible People Unable to Access a Shot.” The Texas Tribune, March 20. https://www.texastribune.org/2021/03/20/texas-coronavirus-vaccine-difficult/. Miller William Peek Monica Parker William . 2020. “Scarce Resource Allocation Scores Threaten to Exacerbate Racial Disparities in Health Care.” CHEST 158 (4 ):1332–34. Millett Gregorio Jones Austin Benkeser David Baral Stefan Mercer Laina Beyrer Chris Honermann Brian , et al 2020. “Assessing Differential Impacts of COVID-19 on Black Communities.” Annals of Epidemiology 47 :37–44.32419766 Moore Latetia Diez Roux Ana . 2006. “Associations of Neighborhood Characteristics with the Location and Type of Food Stores.” American Journal of Public Health 96 (2 ):325–31. O’Connell Heather . 2015. “Where There’s Smoke: Cigarette Use, Social Acceptability, and Spatial Approaches to Multilevel Modeling.” Social Science & Medicine 140 :18–26.26188587 Oladipo Gloria . 2021. “How Chicago’s Vaccine Rollout Is Inhibited by Longstanding Inequality.” The Guardian, February 5. https://www.theguardian.com/us-news/2021/feb/05/chicago-blacks-latinos-vaccine-distribution?utm_term=2bd88cc02b2faffa886b221bdd480dc4&utm_campaign=GuardianTodayUS&utm_source=esp&utm_medium=Email&CMP=GTUS_email. Økland Helene Mamelund Svenn-Erik . 2019. “Race and 1918 Influenza Pandemic in the United States: A Review of the Literature.” International Journal of Environmental Research and Public Health 16 (14 ):2487–505. Peek Monica Simons Russell Parker William Ansell David Rogers Selwyn Edmonds Brownsyne Tucker . 2021. “COVID-19 among African Americans: An Action Plan for Mitigating Disparities.” American Journal of Public Health 111 (2 ):286–92. Persad Govind Peek Monica Emanuel Ezekiel . 2020. “Fairly Prioritizing Groups for Access to COVID-19 Vaccines.” JAMA 324 (16 ):1601–1602.32910182 Prescott Gina Marie Prescott William Allan . 2021. “Health Information Technology Utilization and Impact on COVID-19 Vaccination.” Journal of the American Pharmacists Association 61 (40 ):E230–32. Ramos Elliot Zamudio María Inés . 2020. “In Chicago, 70% of COVID-19 Deaths Are Black.” WBEZ Chicago, April 5. https://www.wbez.org/stories/in-chicago-70-of-covid-19-deaths-are-black/dd3f295f-445e-4e38-b37f-a1503782b507. Raudenbush Stephen Bryk Anthony . 2002. Hierarchical Linear Models: Applications and Data Analysis Methods. Thousand Oaks, CA: SAGE Publications. Reardon Sean O’Sullivan David . 2004. “Measures of Spatial Segregation.” Sociological Methodology 34 :121–62. Roberto Elizabeth . 2018. “The Spatial Proximity and Connectivity Method for Measuring and Analyzing Residential Segregation.” Sociological Methodology 48 (1 ):182–224. Roberts Jennifer Tehrani Shadi . 2020. “Environments, Behaviors, and Inequalities: Reflecting on the Impacts of the Influenza and Coronavirus Pandemics in the United States.” International Journal of Environmental Research and Public Health 17 (12 ):4484–511. Rodriguez Rudolph Sen Saunak Mehta Kala Moody-Ayers Sandra Bacchetti Peter O’Hare Ann . 2007. “Geography Matters: Relationships among Urban Residential Segregation, Dialysis Facilities, and Patient Outcomes.” Annals of Internal Medicine 146 (7 ):493–501.17404351 Ross Janell . 2021. “When a Texas County Tried to Ensure Racial Equity in COVID-19 Vaccinations, It Didn’t Go as Planned.” Time Magazine, March 2. https://time.com/5942884/covid-19-vaccine-racial-inequity-dallas/. Sampson Robert Raudenbush Stephen Earls Felton . 1997. “Neighborhoods and Violent Crime: A Multilevel Study of Collective Efficacy.” Science 277 (5328 ):918–24. Savitz Natalya Verbitsky Raudenbush Stephen . 2009. “Exploiting Spatial Dependence to Improve Measurement of Neighborhood Social Processes.” Sociological Methodology 39 (1 ):151–83. Singer Judith Willett John . 2003. Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. New York, NY: Oxford University Press. Small Mario McDermott Monica . 2006. “The Presence of Organizational Resources in Poor Urban Neighborhoods: An Analysis of Average and Contextual Effects.” Social Forces 84 (3 ):1697–724. Stone Will . 2021. “‘Just Cruel’: Digital Race For COVID-19 Vaccines Leaves Many Seniors Behind.” National Public Radio, February 4. https://www.npr.org/sections/health-shots/2021/02/04/963758458/digital-race-for-covid-19-vaccines-leaves-many-seniors-behind. Strully Kate . 2011. “Health Care Segregation and Race Disparities in Infectious Disease: The Case of Nursing Homes and Seasonal Influenza Vaccinations.” Journal of Health and Social Behavior 52 (4 ):510–26. Tobler Waldo . 1970. “A Computer Movie Simulating Urban Growth in the Detroit Region.” Economic Geography 46 (Suppl. 1 ):234–40. Walker Renee Keane Christopher Burke Jessica . 2010. “Disparities and Access to Healthy Food in the United States: A Review of Food Deserts Literature.” Health & Place 16 (5 ):876–84. Williams David Collins Chiquita . 2001. “Racial Residential Segregation: A Fundamental Cause of Racial Disparities in Health.” Public Health Reports 116 (5 ):404–16. Yang Tse-Chuan Choi Seung-won Emily Sun Feinuo . 2021. “COVID-19 Cases in U.S. Counties: Roles of Racial-Ethnic Density and Residential Segregation.” Ethnicity & Health 26 (1 ):11–21.33059471	35164599	PMC9716049	NO-CC CODE	2022-12-03 23:20:53	no	J Health Soc Behav. 2022 Dec; 63(4):472-490	utf-8	J Health Soc Behav	2,022	10.1177/00221465221074915	oa_other
==== Front Nutr Health Nutr Health NAH spnah Nutrition and Health 0260-1060 2047-945X SAGE Publications Sage UK: London, England 35360990 10.1177/02601060221090293 10.1177_02601060221090293 Special Section on ‘COVID 19’ Original Articles Vitamin D in the news: A call for clear public health messaging during Covid-19 https://orcid.org/0000-0002-6094-6389 Heer Randeep S 1 Sandhu Preeti 1 Wenban Charlotte 1 Mandal Amit K J 1 Missouris Constantinos G 12 1 156548 Wexham Park Hospital , Frimley Health NHS Foundation Trust, UK 2 486462 University of Nicosia Medical School , Nicosia, Cyprus Constantinos G Missouris, Department of Medicine, Wexham Park Hospital, Wexham Street, Slough, SL2 4HL, United Kingdom. Email: [email protected] 12 2022 12 2022 12 2022 28 4 733739 © The Author(s) 2022 2022 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background: The public are increasingly reliant on the internet and media to access healthcare related information during the Covid-19 pandemic. Vitamin D is essential for musculoskeletal and immune health, with daily supplementation advised by public health bodies. Several studies assessing a possible link between vitamin D and Covid-19 severity have arrived at conflicting results and news articles have been rapidly disseminating such research to the public. There has been little focus on studying the quality of information available. Aim: To identify if online search interest in vitamin D increased with pandemic burden and analyse the accuracy of public health messaging relating to vitamin D in online news articles. Methods: Online search interest data for vitamin D was correlated with pandemic burden, defined as the number of Covid-19 deaths. Online news articles discussing vitamin D and Covid-19 were analysed using qualitative coding. Results: Online search interest for vitamin D increased as pandemic burden increased (p < 0.0001, Spearman's rank). Of the 72 articles identified, most (50%) offered a mixed opinion on the benefit of vitamin D in Covid-19. From articles making a recommendation for vitamin D supplementation, 23% of articles gave mixed messaging or advised against supplementation. 16% of articles recommended a dose which exceeded the safe limit of 4000 IU/day, risking toxicity. Conclusion: A significant number of articles provided mixed messaging or incorrectly advised supratherapeutic doses. This study highlights an opportunity for public health bodies to utilise the increased interest in vitamin D during the pandemic to raise awareness with accurate information. typesetterts19 ==== Body pmcIntroduction Interest in a putative link between vitamin D deficiency and severity of Coronavirus disease 2019 (Covid-19) has led to several studies being published showing conflicting results (Murdaca et al., 2020). In the United Kingdom (UK), a demand for clarity amongst the medical community was met by the publication of a rapid review by the National Institute for Health and Care Excellence (NICE) in December 2020 which concluded that there is currently insufficient evidence to recommend vitamin D supplementation for the prevention or treatment of Covid-19 (NICE, 2020a). Since this guidance, further reviews of high-quality studies have confirmed this lack of evidence (Bassatne et al., 2021). Although these publications and guidelines are targeted towards healthcare professionals, during the Covid-19 pandemic the public have been increasingly exposed to scientific literature which has been rapidly cited by major news outlets prior to peer review (Nadella and Navathe, 2020). Without a clear strategy for public health messaging, conflicting reports disseminated through these outlets may lead to confusion amongst the public, eventually leading to reduced engagement with health promoting behaviours and public health recommendations (Carpenter et al., 2015). Reduced face to face contact with healthcare professionals during the pandemic has encouraged many people to rely on the internet and media to make healthcare related decisions (Tsao et al., 2021). This is particularly important within the context of vitamin D where the public may be discouraged from taking supplements after exposure to news articles reporting unclear or unelucidated benefits in Covid-19. Regardless of any putative benefit related to Covid-19, adequate vitamin D levels are essential for musculoskeletal health and immune function (Nair and Maseeh, 2012). Prior to the pandemic, the prevalence of vitamin D deficiency in the UK was as high as 40%, a figure which may have increased during ‘lockdown’ restrictions because of reduced sunlight exposure (SACN, 2016). Adults in the UK are advised to take 400 IU/ day of vitamin D supplementation during winter months when sunlight is reduced and those at increased risk of deficiency are encouraged to supplement throughout the year (NICE, 2020a). Whilst doses up to 4000 IU/ day are considered safe (SACN, 2016; EFSA NDA Panel, 2016), there have been reports of ‘mega-doses’ of vitamin D being recommended, which could result in toxicity when taken unsupervised (Lanham-New et al., 2020). During times of heightened pandemic burden, healthcare behaviours of the public may be altered. Aspects of the Health Belief Model (HBM) such as perceived risk may be increased due to increased fear from Covid-19 (Schneider et al., 2021). This could in turn lead to an increased interest in possible protective measures against the disease such as vitamin D supplementation. When combined with recommendations of ‘mega-doses’, misleading information that suggests a clear benefit of vitamin D in Covid-19 may result in over-consumption of vitamin D due to the public's fear of the disease. The aims of this study were twofold. Firstly, to identify if public interest in vitamin D measured through online searches increased as the burden of the pandemic increased in the UK. And secondly, to characterise the reporting of vitamin D research in the midst of Covid-19 and analyse the accuracy of public health messaging within online news articles. Methods Public interest in vitamin D Public interest in vitamin D during the pandemic was measured through internet search demand from the popular search engine Google. Data was extracted from Google Trends (https://trends.google.com/trends) for searches performed from the UK using the term “vitamin D” between 5th January 2020 and 10th January 2021 (the height of the second wave of the pandemic). Each data point represents relative search volume (RSV), where the number of searches performed in a given week is normalised against the highest number of searches over the time period studied to generate a number between 0 and 100. The burden of the Covid-19 pandemic in the UK was measured through the number of deaths recorded within 28 days of a positive Covid-19 test. Raw data was extracted as a comma-separated values file from the UK government website for data and insights on Covid-19 (https://coronavirus.data.gov.uk) and total deaths for each week were calculated. RSV was also compared to the dates of national ‘lockdown’ restrictions in the UK (Institute for Government, 2021). To identify a correlation between RSV and Covid-19 deaths, Spearman's rank analysis was performed using GraphPad Prism for MacOS where p < 0.05 was deemed statistically significant. News article analysis Online news articles were identified through a search performed on Google News using a cleared browser on 9th January 2021 with no time filters using the searches [vitamin D AND Covid] and [vitamin D AND coronavirus]. To identify articles focusing on vitamin D as the main topic, only articles which included these phrases in their titles were included. To focus on content designed for the general public, PubMed cited articles and articles issued directly by government agencies were excluded. Websites requiring a paid subscription or membership as a healthcare professional for viewing were also excluded. The articles were downloaded and imported into NVivo for MacOS (QRS International), where qualitative analysis was performed. Two investigators independently performed line-by-line coding and classified each article using the criteria shown in Table 1. Articles containing videos were transcribed and also analysed using the same methodology. To assess whether articles were based directly off scientific information, articles were analysed to identify any direct citations to PubMed cited articles, articles on pre-print servers or documents from national guidelines. If articles suggested a daily vitamin D dose including a short ‘loading’ course, the maximum dose suggested and the evidence cited to support this dose were recorded. Doses mentioned in the setting of future clinical trials were excluded. All doses stated in articles were converted to equivalent IU/ day. Articles containing hyperlinks to official sources for further information regarding vitamin D defined as government websites, learned medical societies or medical charities were recorded. When there was a disagreement between the classifications made by the initial two investigators, the article was randomly assigned to one of the remaining investigators to serve as a tiebreak. Initial agreement in article classification between the two initial investigators was high, with only 6% of articles sent to a third investigator. Table 1. Criteria used to classify online news articles. Description and examples Benefit of vitamin D in Covid-19 Beneficial - Vitamin D supplementation reduces the risk of catching Covid-19 or the risk of severe disease. - Low vitamin D levels are associated with higher rates of Covid-19 or severe disease. Mixed - Articles containing both ‘beneficial’ and ‘no benefit’ statements. No benefit - There is no benefit to vitamin D supplementation in Covid-19. - Vitamin D deficiency is not associated with higher rates of infection or severe disease. - Criticism of research studying vitamin D and Covid-19. Recommendation for vitamin D supplementation Recommended - Statements recommending vitamin D supplementation for any reason. (Supplementation through food or sunlight is not included). - Personal accounts of people taking vitamin D supplementation for any reason. Mixed - Contains statements both recommending and not recommending vitamin D supplementation. Not recommended - Advises against or advises caution taking vitamin D supplementation. - Emphasises the harms of vitamin D supplementation without recommending supplementation. Not stated - No recommendation made. Results Public interest in vitamin D During the winter months prior to the first recorded Covid-19 death in the UK, RSV for vitamin D remained below 50 (Figure 1). As deaths increased during the first wave of the pandemic, an increase in RSV followed. In the summer months when the pandemic burden was low, RSV fell to its lowest point. As deaths increased again during the second wave, RSV also increased again and appears to be on an upward trend entering 2021. Upon applying Spearman's rank analysis, there was a positive correlation between RSV and Covid-19 related deaths (ρ = 0.62, p < 0.0001). RSV also appeared to increase during national ‘lockdown’ restrictions. Figure 1. Interest in vitamin D correlates with pandemic burden. Relative search volume (RSV) for the term ‘vitamin D’ (solid blue line) correlates with both the number of deaths in the UK within 28 days of a positive Covid-19 test (dotted red line) and dates of national ‘lockdown’ restrictions (rectangular boxes). News article analysis The initial search returned 120 unique results of which 72 articles were included for final analysis (Figure 2). Most articles (50%) gave a mixed opinion on the benefit of vitamin D in Covid-19, followed by 35% of articles being supportive of a benefit of vitamin D and 15% of articles stating no benefit of vitamin D (Table 2). Around 86% of articles made a recommendation surrounding vitamin D supplementation. Of these, the majority of articles (77%) recommended taking vitamin D supplementation with 23% of articles giving mixed messages or advising against taking supplementation. The majority of news articles (61%) were based directly off scientific information of which, 22% were based on information published on pre-print servers. Only 11 out of 72 news articles contained hyperlinks to further information for the public from official sources such as government websites or medical charities. Figure 2. CONSORT flow diagram representing the selection of news articles included for analysis. Table 2. Interpretations of the benefit of vitamin D in Covid-19 and recommendations for supplementations in online news articles. Number of articles Benefit of vitamin D in Covid-19 Beneficial 25 Mixed 36 No benefit 11 Recommendation for vitamin D supplementation Recommended 48 Mixed 11 Not recommended 3 Not stated 10 There were 44 articles which stated a recommended vitamin D dose of which the majority (55%) suggested the vitamin D dose recommended by NICE of 400 IU/ day (Table 3). The majority of articles also did not exceed the maximum safe dose described by public health bodies of 4000 IU/ day. However, there were seven articles which exceeded this recommendation and six of these described doses of 10,000 IU/ day and above. Three of these articles did not specify a duration for such a high dose. The evidence cited for super-therapeutic dosing included expert opinion (six counts), PubMed cited articles (two counts) and anecdote (two counts). Table 3. Grouped frequency table for vitamin D doses (d) described in online news articles. All doses were converted to equivalent IU/day. Vitamin D dose (d) Number of articles 0 IU < d ≤ 400 IU/day 24 400 IU < d ≤ 4000 IU/day 13 4000 IU < d ≤ 10,000 IU/day 6 10,000 IU < d ≤ 30,000 IU/day 1 Discussion The Covid-19 pandemic has changed the way many aspects of healthcare are delivered with reduced face-to-face contact between patients and healthcare providers and increasing numbers of virtual consultations (Webster, 2020). This has likely resulted in the public having reduced exposure to materials traditionally used in public health messaging such as leaflets and posters displayed at a healthcare setting or information given by healthcare professionals during consultations. Many people may instead be relying on other sources of information such as the internet. This study has confirmed that during times of heightened pandemic burden, online search interest in vitamin D increased. Whilst it is not possible to determine the intention of these internet searches, this association may be explained by the HBM (Janz and Becker, 1984). With the public now having readily accessible information on Covid-19 and vitamin D supplementation through online materials, the accessing individual can independently assess their susceptibility to Covid-19 and likelihood of vitamin D deficiency. This would be combined with the individual learning about the benefits of supplementation, their self-efficacy in performing such a behaviour change and finally their exposure to any cues to action, such as experience of others affected from Covid-19. In the previous 2009 H1N1 influenza pandemic, certain behavioural responses such as fear and health protective behaviours correlated with pandemic burden which could increase the public's perceived risk of the disease (Wong and Sam, 2010). A previous study demonstrated Covid-19 pandemic burden to also be a predictor of perceived risk, although this was less important than other psychological factors (Schneider et al., 2021). Further study is required to confirm this putative association. Media outlets and social media could also accentuate changes to these constructs of the HBM during the Covid-19 pandemic for example through altering behaviours and the public's perceived threat from Covid-19 (Raamkumar et al., 2020). Therefore, our findings of an increased search interest related to vitamin D supplementation during heightened pandemic burden may reflect the public's attempts to minimise the threat of Covid-19 due to increases in perceived risk from the disease coupled with understanding the potential perceived benefits of vitamin D supplementation. Alternatively, this association may represent people seeking further information about a possible link between vitamin D and Covid-19 or due to public awareness of reduced sunlight exposure leading to vitamin D deficiency during national ‘lockdown’ restrictions. Further study surveying public opinion regarding vitamin D during the Covid-19 pandemic may be helpful in better understanding the reasons for increased search interest. At the time of writing this article, there is currently no national public health campaign in the UK for vitamin D supplementation. Regardless of a possible benefit in Covid-19, adults in the UK are advised to take 400 IU/ day of vitamin D supplementation during winter months and those at higher risk of deficiency are advised to take supplements throughout the year (NICE, 2020a). However, the panel producing the rapid vitamin D guideline for Covid-19 expressed concern over the lack of public awareness of these national guidelines (NICE, 2020a). Additionally, previous qualitative studies have revealed a low level of satisfaction on the quality of information currently available to the public (Day et al., 2019). This allows for other forms of media to fulfil the role of providing health-related information to the public. Online news articles are a major influential source of information obtainable through internet searches and previous studies have highlighted instances of inaccurate information present in these articles during the Covid-19 pandemic (Islam et al., 2020). There is therefore a need to assess their role in public health messaging as they can improperly influence public opinion and distract from evidence-based measures against the disease. Whilst the benefits of vitamin D supplementation on musculoskeletal health are clear, its benefits in Covid-19 remain disputed with multiple studies performed showing conflicting results (Murdaca et al., 2020). There is, therefore, insufficient evidence currently to recommend the use of vitamin D supplementation to prevent or reduce the severity of Covid-19. The majority of articles report a mixed benefit of vitamin D in Covid-19, likely reflective of the conflicting results seen in scientific articles. Given the number of articles making a recommendation for vitamin D supplementation, it is clear that online news articles are providing a role in public health messaging when national campaigns do not exist Additionally, as most articles were based directly off scientific information, it is evident that news articles are rapidly disseminating scientific research to the public as initially suspected. Many articles directly disseminated research from pre-print servers which had not undergone the peer review process. The general public may not be aware of the limitations of such sources unless explicitly stated in articles, meaning information from poor quality studies could improperly influence public opinion (Nadella and Navathe, 2020). Although the majority of articles recommended vitamin D supplementation, one in five articles gave mixed or conflicting messaging or advised against supplementation which can be detrimental to public health messaging. Conflicting health information can be defined as two or more healthcare-related propositions that are inconsistent with one another (Carpenter et al., 2015). There have also been previous instances of conflicting information during the pandemic for example the use of face coverings in reducing the risk of transmission of the virus (Breslow, 2021). Such conflicting information could arise due to research showing conflicting results, differences in the way research is interpreted and disagreements in recommendations from various professional bodies or stakeholders (Nagler et al., 2020). In the context of vitamin D and Covid-19, it is likely that the abundance of conflicting research on this topic is a major reason for the extent of conflicting information present in online news articles (Murdaca et al., 2020). Conflicting health information has been shown to cause confusion and induce negative emotional reactions to the reader including distress and frustration (Chang, 2015; Nagler et al., 2018). Similar cognitive changes have been demonstrated in the field of nutrition whereby conflicting messages regarding nutrition led to ‘backlash’- described as negative feelings towards dietary recommendations (Lee et al., 2017). This suggests that there is a strong possibility that conflicting messaging regarding vitamin D as a nutritional supplement could therefore exert a similar effect. Additionally, when people encounter conflicting information, they may employ strategies for clarification such as seeking advice from a healthcare professional which is now less readily available due to reduced face-to-face clinical contact (Elstad et al., 2012). An alternative strategy to mitigate confusion in the absence of contact with healthcare professionals can be the use of hyperlinks (clickable links to other websites) to enable the reader to gain clarity and easily access verified information (Stroobant, 2018). However, only 15% of articles analysed contained such hyperlinks to official sources of information and this possibly reduced the intelligibility and credibility of these articles. Ultimately, these negative cognitive responses can translate into detrimental behavioural changes such as reduced engagement with health promoting behaviours and reduced trust in public health recommendations (Carpenter et al., 2015). Importantly in Covid-19, this may also result in reduced engagement with other behaviours essential in mitigating the spread of the virus including handwashing and social distancing. Given that the public are readily accessing online information about vitamin D in the UK where deficiency rates are high, and many may be incorrectly discouraged from taking vitamin D, the need for accurate and consistent public health messaging amongst news articles is highlighted. Whilst there is no international consensus on the recommended daily vitamin D dose, public health bodies from the UK, America and Europe are in agreement that daily vitamin D doses should not exceed 4000 IU/ day (SACN, 2016; EFSA NDA Panel, 2016; Office of Dietary Supplements, 2021). We found multiple articles which recommended doses in excess of 4000 IU/ day and in one article, an extraordinary dose of 30,000 IU/ day was recommended. Although most articles recommended vitamin D supplementation, when written in the context of excessive doses, the risk of overconsumption may be increased due to the public's fear of the disease altering the ‘perceived risk’ aspect of the HBM. Although high doses may be utilised by clinicians to correct severe deficiency, without medical supervision including prior assessment of serum calcium levels, such doses can lead to vitamin D toxicity. Other authors have also expressed concerns about overpromotion of very high doses of vitamin D during the pandemic especially when there is a lack of supporting evidence (Lanham-New et al., 2020). In fact, the level of evidence used to support high dose vitamin D supplementation in these articles was largely ‘expert opinion’ or anecdotal and lacking in scientific-based rationale. Prescription duration of high dose supplementation should generally not exceed 8 weeks (NICE, 2020b); however, in three of the articles advocating doses in excess of 4000IU/ day, no duration was specified and the reader could easily assume these doses should be taken indefinitely. In parallel, reports of vitamin D toxicity related to high dose over-the-counter supplements are growing (Taylor and Davies, 2018). Conclusion It is undisputed that vitamin D has multiple health benefits, and its deficiency is readily correctable with cost-effective and safe intervention. Public interest in vitamin D measured through online search interest increased during times of heightened pandemic burden. News articles accessible through internet searches provided varying interpretations of research around the subject and provided a role in public health messaging when national campaigns did not exist. Whilst the majority of articles supported vitamin D supplementation in line with national guidelines, one in five articles provided mixed messaging or discouraged supplementation and a significant number of articles advocated supratherapeutic dosing regimens. The increased public interest in vitamin D during the pandemic may have represented a missed opportunity for public health bodies to raise awareness about vitamin D supplementation with clear messaging. Misleading information present in news articles may divert public attention away from evidence-based measures to protect against the disease. Going forward, we propose that media agencies collaborate with public health agencies to disseminate accurate information from official and validated sources. Supplemental Material sj-docx-1-nah-10.1177_02601060221090293 - Supplemental material for Vitamin D in the news: A call for clear public health messaging during Covid-19 Click here for additional data file. Supplemental material, sj-docx-1-nah-10.1177_02601060221090293 for Vitamin D in the news: A call for clear public health messaging during Covid-19 by Randeep S Heer, Preeti Sandhu, Charlotte Wenban, Amit K J Mandal and Constantinos G Missouris in Nutrition and Health Limitations: We acknowledge that not all members of the public are able to access the internet therefore findings described in this study may only be applicable to those who have internet access. Older adults and the institutionalised, who are inherently at risk of vitamin D deficiency, are also less likely to be familiar with the internet and to be guided by media outlets. This study does not include news disseminated on radio, television or print only media. Internet search results can vary based on location therefore findings may not be applicable to all regions of the world. Availability of data and materials: Data used to calculate online search interest and pandemic burden are publicly available at https://trends.google.com/trends and https://coronavirus.data.gov.uk respectively. Competing interests: All authors understand the policy of declaration of interests. RSH, PS, CW, AKJM and CGM all declare that they have no competing interests. Contributorship: All five authors contributed to the manuscript. RSH and PS collected and analysed the data. RSH, PS, CW and AKJM wrote the manuscript and all authors were involved in its final approval. AKJM and CGM are acting as guarantors of the submitted work. Statement of ethics committee approval: As only publicly available data was used in this study, ethical approval was not required. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ORCID iD: Randeep S Heer https://orcid.org/0000-0002-6094-6389 Supplemental material: Supplemental material for this article is available online. ==== Refs References Bassatne A Basbous M Chakhtoura M , et al. (2021) The link between COVID-19 and Vitamin D (VIVID): a systematic review and meta-analysis. Metabolism: Clinical and Experimental 119 : 154753.33774074 Breslow J (2021) Mixed messaging on masks set U.S. public health response back. Available at: https://www.npr.org/sections/health-shots/2020/07/01/886299190/it-does-not-have-to-be-100-000-cases-a-day-fauci-urges-u-s-to-follow-guidelines (accessed 9 March 2021). Carpenter D Geryk L Chen A , et al. (2015) Conflicting health information: a critical research need. Health Expectations 19 (6 ): 1173–1182.26709206 Chang C (2015) Motivated processing. Science Communication 37 (5 ): 602–634. Day R Krishnarao R Sahota P , et al. (2019) We still don’t know that our children need vitamin D daily: a study of parents’ understanding of vitamin D requirements in children aged 0-2 years. BMC Public Health 19 (1 ): 1119. EFSA NDA Panel (EFSA Panel on Dietetic Products, Nutrition and Allergies) (2016) Scientific opinion on Dietary Reference Values for vitamin D. EFSA J 14 : 4547. Elstad E Carpenter D Devellis R , et al. (2012) Patient decision making in the face of conflicting medication information. International Journal of Qualitative Studies on Health and Well-Being 7 (1 ): 18523. Institute for Government (2021) Timeline of UK coronavirus lockdowns. Available at: https://www.instituteforgovernment.org.uk/sites/default/files/timeline-lockdown-web.pdf (accessed 9 March 2021). Islam M Sarkar T Khan S , et al. (2020) COVID-19–related infodemic and its impact on public health: a global social media analysis. The American Journal of Tropical Medicine and Hygiene 103 (4 ): 1621–1629.32783794 Janz N Becker M (1984) The health belief model: a decade later. Health Education Quarterly 11 (1 ): 1–47.6392204 Lanham-New S Webb A Cashman K , et al. (2020) Vitamin D and SARS-CoV-2 virus/COVID-19 disease. BMJ Nutrition Prevention & Health 3 (1 ): 106–110. Lee C Nagler R Wang N (2017) Source-specific exposure to contradictory nutrition information: documenting prevalence and effects on adverse cognitive and behavioral outcomes. Health Communication 33 (4 ): 453–461.28151010 Murdaca G Pioggia G Negrini S (2020) Vitamin D and COVID-19: an update on evidence and potential therapeutic implications. Clinical and Molecular Allergy 18 (1 ): 23. Nadella P Navathe A (2020) The media needs to incorporate principles of research communication to improve COVID-19 reporting. Healthcare 8 (4 ): 100473.32992110 Nagler R Vogel R Gollust S , et al. (2020) Public perceptions of conflicting information surrounding COVID-19: results from a nationally representative survey of U.S. adults. PLOS ONE 15 (10 ): e0240776.33085719 Nagler R Yzer M Rothman A (2018) Effects of media exposure to conflicting information about mammography: results from a population-based survey experiment. Annals of Behavioral Medicine 53 (10 ): 896–908. Nair R Maseeh A (2012) Vitamin D: the “sunshine” vitamin. Journal of Pharmacology & Pharmacotherapeutics 3 (2 ): 118–126.22629085 National Institute for Health and Care Excellence (NICE) (2020a) COVID-19 rapid guideline: vitamin D. Available at: https://www.nice.org.uk/guidance/ng187/resources/covid19-rapid-guideline-vitamin-d-pdf-66142026720709 (Accessed 9 March 2021). National Institute for Health and Care Excellence (NICE) (2020b) Vitamin D deficiency in adults - treatment and prevention. Available at: https://cks.nice.org.uk/topics/vitamin-d-deficiency-in-adults-treatment-prevention/management/management/ (accessed 9 March 2021). Office of Dietary Supplements (2021) Vitamin D. Available at: https://ods.od.nih.gov/factsheets/VitaminD-HealthProfessional/ (accessed 9 March 2021). Raamkumar A Tan S Wee H (2020) Use of health belief model–based deep learning classifiers for COVID-19 social Media content to examine public perceptions of physical distancing: model development and case study. JMIR Public Health and Surveillance 6 (3 ): e20493.32540840 Schneider C Dryhurst S Kerr J , et al. (2021) COVID-19 risk perception: a longitudinal analysis of its predictors and associations with health protective behaviours in the United Kingdom. Journal of Risk Research 24 (3-4 ): 294–313. Scientific Advisory Committee on Nutrition (SACN) (2016) Vitamin D and health. Available at: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/537616/SACN_Vitamin_D_and_Health_report.pdf (accessed 9 March 2021). Stroobant J (2018) Finding the news and mapping the links: a case study of hypertextuality in Dutch-language health news websites. Information, Communication & Society 22 (14 ): 2138–2155. Taylor P Davies J (2018) A review of the growing risk of vitamin D toxicity from inappropriate practice. British Journal of Clinical Pharmacology 84 (6 ): 1121–1127.29498758 Tsao S Chen H Tisseverasinghe T , et al. (2021) What social media told us in the time of COVID-19: a scoping review. The Lancet Digital Health 3 (3 ): e175–e194.33518503 Webster P (2020) Virtual health care in the era of COVID-19. The Lancet 395 (10231 ): 1180–1181. Wong L Sam I (2010) Temporal changes in psychobehavioral responses during the 2009 H1N1 influenza pandemic. Preventive Medicine 51 (1 ): 92–93.20403375	35360990	PMC9716052	NO-CC CODE	2022-12-03 23:20:53	no	Nutr Health. 2022 Dec; 28(4):733-739	utf-8	Nutr Health	2,022	10.1177/02601060221090293	oa_other
==== Front Nutr Health Nutr Health NAH spnah Nutrition and Health 0260-1060 2047-945X SAGE Publications Sage UK: London, England 33843324 10.1177/02601060211009033 10.1177_02601060211009033 Special Section on ‘COVID 19’ Original Articles How will the way we live look different in the wake of the COVID-19 pandemic? A nutrition survey in Greece https://orcid.org/0000-0001-8583-153X Panagiotakos Demosthenes 1 Kosti Rena I. 2 Pitsavos Christos 3 1 School of Health Sciences and Education, 68996 Harokopio University , Kallithea, Athens, Greece 2 Department of Nutrition and Dietetics, School of Physical Education, Sports and Dietetics, 37786 University of Thessaly , Trikala, Greece 3 School of Medicine, 68993 University of Athens , Athens, Greece Demosthenes Panagiotakos, School of Health Sciences and Education, Harokopio University, Kallithea, Athens, 70 El, Venizelou Avenue, 17676, Athens, Greece. Email: [email protected] 12 2022 12 2022 12 2022 28 4 677683 © The Author(s) 2021 2021 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background: As we move towards a post-pandemic society, a question arises: How will the way we live look different in the wake of the COVID-19 pandemic? Aim: The aim of this survey was to evaluate changes in eating habits and other lifestyle behaviours (i.e., exercise and smoking) of people of almost all ages, who live in Greece, during the COVID-19 pandemic. Methods: A web-based survey using conventional sampling was conducted from during December 2020, in Greece. A total of 2258 individuals, aged 17 years and older voluntarily participated (912 (40%) men). Results: 89 (3.94%) of the participants reported that they had, or currently have been diagnosed with COVID-19. Moreover, 36.4% of the participants reported that they have changed their dietary habits during the pandemic towards a healthier diet – those participants had median age of 35 years, were of both sexes, 17% had co-morbidities and 69% with higher education level; moreover, 19% of those participants have started or increased the frequency of receiving dietary supplements that enhance the immune system, 34% of the participants reported that they gained weight during the pandemic period, whereas 19.8% reported that they have lost weight, and 37% of the participants reported that they have started or increased, as compared to the pre-pandemic time, their frequency of physical activities. Conclusions: The COVID-19_pandemic seems to have forced people to discover again habits and traditions towards a more natural and healthier way of living. Long-term consequences and the evolution of these lifestyle changes after the COVID-19 pandemic have to be evaluated relevant to their implications in public health. Lifestyle diet behaviour exercise COVID-19 typesetterts19 ==== Body pmcIntroduction The first pandemic of the 21st century due to the new coronavirus SARS-CoV-2 (COVID-19) has caused unprecedented geopolitical, economic, and health consequences, and, without any doubt, has changed our lives (World Economic Forum, 2020). As we move toward a post-pandemic society, several surveys have been conducted in many countries by global thought leaders, and thinkers, in order to evaluate the effect of COVID-19 on people’s lives. Since the beginning of the pandemic, empty restaurants and cafés, have been a foretokening the crisis that was coming, leading to the imposed lockdown of our lives. Early studies on the COVID-19 pandemic had pointed out the upcoming mental health issues in the population (Torales et al., 2020), whereas these worries were confirmed later, and in particular among adults at high risk of severe illness from COVID-19 (Flint et al., 2020). It seems that many people have changed their lifestyle habits during the pandemic. In the past months, researchers have begun to systematically study people’s eating habits and behaviours during the pandemic. Although there is no robust data yet, findings from an online international survey demonstrate shifts towards a health-compromising direction (Ammar et al., 2020). However, whether eating habits have changed towards a healthier pattern as a result of the anticipated increased consumption of homemade meals due to home confinement is still a matter of concern since the literature shows inconsistent results (Alhusseini and Alqahtani, 2020; Cheikh Ismail et al., 2020; Meister, 2020; Di Renzo et al., 2020; Rodríguez-Pérez et al., 2020; Rossinot et al., 2020; Sánchez-Sánchez et al., 2020; Sidor and Rzymski 2020; Tárraga López et al., 2020). Regarding physical activities (PAs), the results of the majority but not all of earlier studies conclude a decreasing trend, especially among middle-aged and older adults (Alomari et al., 2020; Amini et al., 2020; Antunes et al., 2020; Caputo and Reichert, 2020; Cheikh Ismail et al., 2020; Di Renzo et al., 2020; Rossinot et al., 2020; Sánchez-Sánchez et al., 2020). The lockdown established during the pandemic in the majority of the countries, forced people to reduce considerably their mobility and motor activities, which has led to an increase in adopting a sedentary lifestyle contributing to higher anxiety levels (Chen et al., 2020). Although public health authorities in several countries, as well as the World Health Organization which has started the campaign “#HealthyAtHome: Healthy Diet”, have tried to inform people on the way they should behave regarding diet and exercise, during the pandemic, the effect of these recommendations on people’s health is unknown (World Health Organization, 2020). Hence, a crucial question arises: How will the way we live look different in the wake of the COVID-19 pandemic? It is difficult to say that we know the answer. Thus, the aim of this survey was to evaluate self-perceived changes in eating habits and other lifestyle behaviours (i.e., exercise and smoking) during the COVID-19 pandemic, of adult people who live in Greece. Methods Design A web-based survey based on conventional sampling was conducted. The study’s investigators invited people aged 17 years old and older to participate in the study by providing them the link to the anonymized, structured questionnaire that has been designed for the survey. In the case that the participants could not have Internet access, the study’s investigators provided them a printed questionnaire and asked to them to complete it. Setting The survey took place in all seven Greek regions, between 1 and 31December 2020. Sample A total of 2258 individuals voluntarily participated in the survey; of these 912 (40%) were men with mean age of 36 (standard deviation (SD) 18) years old (range 17–101 years) and 1346 (60%) were women with mean age of 33 (SD 18) years old (range 17–92 years). The age–sex distribution of the working sample was slightly different from the age–sex distribution of the reference population (p <0.05); in particular the participants of this survey were younger (average age of Greece = 45.6 years) and more were females, as compared to the total Greek population. The study’s sample represents (by age and sex) 0.02% of the total Greek population and it is considered adequate (i.e., statistical power 99%) for the evaluation of 10% differences between levels of the investigated measurements at 0.05 significance level of two-sided hypotheses. Measurements The survey’s questionnaire included socio-demographic questions about age, sex, region of residence, and the higher education level achieved by the adult participants (in years of school) which was categorized in three classes: lower education (<6 years or technical training); medium (7–12 years); and higher education (university degree). At this point it should be noted that the education level of the participants was higher than the median level of the Greek population (data from the Organization for Economic Co-operation and Development, accessed at https://www.oecd.org/education/education-at-a-glance/EAG2019_CN_GRC.pdf). The medical history of diagnosed chronic diseases, that is, cardiovascular, cancer, chronic obstructive pulmonary disease, hypertension, dyslipidaemia, nephropathy and type 2 diabetes mellitus, as well as their management, was also recorded. Moreover, participants were asked whether they have been infected by the new coronavirus SARS-CoV-2 (through reverse transcription polymerase chain reaction testing). Body weight and height were reported, and body mass index (BMI) was then calculated: overweight was defined as BMI >24.9 kg/m2; and obesity as BMI >29.9 kg/m2. Smoking habits were also recorded, and participants were categorized as current smokers (those who smoked at least one cigarette per day during the preceding year), former smokers (those who have stopped smoking for at least one year) and never smokers. Lifestyle changes that occurred after the onset of the COVID-19 pandemic in Greece (i.e., 26 February 2020) were carefully recorded using a structured questionnaire including specific Likert-type questions (e.g., disagree, no change, and agree). These questions evaluated any change in participants’ eating habits and participation in PAs. In particular, participants were asked to report whether they have changed their frequency of consumption of all main food-groups and beverages, that is, cereals, legumes, vegetables, salads and green vegetables, fish and fisheries, meat and processed meat, drinking water, beverages and alcohol, as well as use of any dietary supplements that strengthen the immune system. Moreover, a healthy diet was defined as a pattern rich in fruits, salads, green vegetables, vegetables, cereals, legumes, use of olive oil instead of other added lipids, and less meat, especially processed meat. Engagement in any type of organized (i.e., group exercise) or not (walking, running, swimming, etc.) PAs was recorded and compared to the periods before and after the COVID-19 pandemic. Participants were also asked to report their level of satisfaction (0: unacceptable, to 5: excellent) regarding the provided lifestyle guidelines by the public health authorities. Statistical analysis Categorical variables are presented as frequencies and relative frequencies. Continuous variables are presented as mean and SD. Incidence of COVID-19 was estimated using weighted proportions by sex and age of the countrywide distribution of the diagnosed cases. Associations between categorical variables were evaluated using Pearson’s Chi-square test, while mean values’ group comparisons were performed using the t-test since the continuous variables were normally distributed as evaluated using symmetry plots (quantile-quantile plot). Multiple correspondence analysis (MCA) was applied in order to evaluate the changes in the lifestyle patterns of the participants. Logistic regression analysis (odds ratios and 95% confidence intervals (CIs)) was applied to evaluate participants’ characteristics towards the lifestyle changes that occurred during the COVID-19 pandemic. STATA version 16 statistical software was used for the data analyses (TStat, S.r.l., 67039 Sulmona, Italy). Results In total, 89 (3.94%) of the participants reported that they had, or currently have been diagnosed with COVID-19; of these, 36 were men (3.95%, mean age 33 (SD 13) years) and 53 were women (3.94%, age 32 (SD 16) years) (p-value for sex comparisons = 0.82). The incidence of COVID-19 observed in this survey is in line (adjusted for period effect) with reports from seroepidemiological studies conducted in Greece during the past months (Bogogiannidou et al., 2020; Tsitsilonis et al., 2020). In Table 1, participants’ characteristics and certain lifestyle behaviours during the pandemic time are presented, separately for men and women. Almost one out of three of the participants (36.4%) reported that they have changed their dietary habits during the COVID-19 pandemic towards a healthier diet (i.e., a pattern rich in fruits, vegetables, salads, green vegetables, cereals, legumes, use of olive oil instead of other added lipids, and less meat, especially processed meat). Those participants were of median age 35 years (1st and 3rd quartiles 31, 45), were of both sexes (34% of men vs. 38% of women, p = 0.09), 17% had co-morbidities and 69% with higher education level. In addition, 19% of the participants have started or increased the frequency of receiving dietary supplements (vitamins and minerals) that enhance the immune system (p <0.001). Table 1. Socio-demographic and clinical characteristics, and lifestyle changes during the COVID-19 pandemic period, among 2258 men and women, aged 17–101 years old, from Greece (web-based survey conducted during December 2020, in all Greek regions). Participants’ characteristics Men Women p-value (n = 912) (n = 1346) Age, years (mean (standard deviation) 36 (18) 33 (18) 0.0003 Higher education, n (%) 566 (62%) 928 (69%) <0.001 Smoking <0.001 Current, n (%) 229 (26%) 285 (21%) Former, n (%) 184 (20%) 132 (10%) Diagnosed cases of COVID-19, n (%) 36 (3.95) 53 (3.94) 0.99 History of hypertension, n (%) 134 (15%) 97 (8%) <0.001 History of diabetes mellitus, n (%) 50 (6%) 48 (4%) 0.007 History of dyslipidaemia, n (%) 150 (17%) 139 (11%) 0.001 History of renal disease, n (%) 29 (3%) 18 (1.4%) <0.001 Overweight/obesity, n (%) 477 (52%) 386 (29%) <0.001 Lifestyle changes during the COVID-19 pandemic period Dietary habits change to healthier pattern, yes (%) 313 (34%) 509 (38%) 0.09 Legumes and cereals, increase (%) 212 (24%) 358 (27%) 0.24 Fruits and vegetables, increase (%) 287 (32%) 515 (38%) 0.004 Salads and green vegetables, increase (%) 214 (24%) 356 (26%) 0.25 Fish and fisheries, increase (%) 135 (15%) 262 (20%) 0.005 Meat and processed meat products, <0.001 Reduction (%) 159 (17%) 300 (22%) Increase (%) 216 (24%) 226 (18%) Drinking water, glasses/day 5.7 (2.3) 4.6 (2.1) 0.03 Dietary supplements, increase (%) 142 (16%) 297 (22%) 0.001 Habitual physical activity (PA), yes (%) 722 (79%) 1,009 (75%) 0.001 Frequency of PAs, times/week 3.3 (1.5) 2.09 (1.9) <0.001 PA, increase (%) 325 (36%) 509 (38%) 0.001 Body weight change 0.001 Loss, n (%) 142 (16%) 303 (23%) Gain, n (%) 300 (33%) 474 (35%) Overall, 34% of the participants reported that they gained weight during the COVID-19 pandemic, whereas 19.8% reported that they have lost weight. Moreover, 40% of the participants with obesity reported that they have gained weight as compared to only 16.9% of them who reported weight loss (p = 0.001). A substantial proportion, that is, 37% of the participants reported that they have started or increased the frequency of engagement in PAs as compared to the pre-pandemic period. Strong associations were revealed between certain lifestyle changes and the medical history of the participants, after adjusting for age, sex, and education level. Individuals with one or more morbidities were more likely to adopt lifestyle changes as compared to the others. Specifically, individuals with hypertension were 1.42-times (95% CI 1.21, 1.78) more likely to have changed their dietary habits during the COVID-19 pandemic towards a healthier pattern as compared to those who are normotensive. Similarly, individuals with diabetes were 1.51-times (95% CI 1.20, 1.92) more likely to have changed their dietary habits to a healthier pattern, individuals with dyslipidaemia were 1.52-times (95%CI 1.28, 1.78) more likely to have changed their dietary habits to a healthier pattern, and individuals with renal disease were 1.92-times (95% CI 1.28, 2.85) more likely to have changed their dietary habits towards a healthier pattern, after taking into account age, sex and education status. However, no associations were observed between history of COVID-19 and lifestyle changes during the pandemic period (for diet changes p = 0.86; and for PA changes p = 0.46). Regarding PAs, a considerable proportion of people reported that they have started or increased the frequency of participating in PAs during the COVID-19 pandemic period (37.17%); however, no significant associations were observed between PAs changes and participants’ health status (all p >0.35). An additional analysis was performed focused on those who changed to healthier pattern of dietary habits and also increased their PAs. It was observed that 26.02% of the participants reported increasing their engagement in PAs without improving their diet, whereas 25.49% reported that they have improved their dietary habits without being more engaged in PAs. In addition, 37.34% reported no improvement either on dietary habits or on PA status, and only 11.14% reported an improvement in both dietary habits and PA status. Data analysis revealed that participants who changed only their dietary habits but not their PAs status were older (35 vs., 30 years old, p = 0.002), similarly distributed in both sexes (p = 0.37), and with similar education level (p = 0.99). Moreover, changes to a healthier dietary pattern were more common among never smokers as compared to ex-smokers or current smokers (67% vs., 14%, 19%, respectively, p <0.001). The MCA revealed a dominant “healthy” behavioural pattern of participants’ lifestyle habits, as compared to the pre-COVID-19 pandemic period, that explained 67% of the total information of the collected data. This pattern was characterized by participants who adopted healthier eating behaviours, such as eating more fruits, vegetables, salads and green vegetables, fish, legumes, and cereals, and less meat and processed meat, drinking more than six glasses of water per day, and use of dietary supplements, in conjunction with increased PAs, as compared to the pre-COVID-19 pandemic period. This group had a median age of 41 years, was similarly – to the overall sample – distributed in women and men (i.e., 59%/41%), and 65% had higher education level and were more likely to have more than one morbidity (2.4 (0.9) vs. 0.8 (0.2) co-morbidities, p <0.001). The majority of the participants (i.e., 1684, 75%) urged for dietary guidelines and the need for better information from the public health authorities during the COVID-19 pandemic; moreover, 43% of men and 39% of women reported that the information received from the authorities and other specialists regarding their diet during the COVID-19 pandemic was poor and limited. Discussion The potential changes of eating habits and PAs that have been brought on by the COVID-19 pandemic in Greece were explored through a nationwide survey of 2258 adult individuals. The survey was conducted during the second phase of the COVID-19 pandemic, that is, December 2020, in which Greece experienced a much higher outbreak of COVID-19 cases and related deaths, compared to the first phase in spring 2020. Despite the inherent limitations of the present observational survey, an answer to the question “How will the way we live look different in the wake of the COVID-19 pandemic?” could be given, at least regarding the participants’ dietary habits and PAs. The lockdown seems to have forced people to discover again old habits and traditions towards a more natural and healthier way of living, since the factors that were responsible for the nutrition transition to more Westernized dietary patterns may well have been less prevalent and, thus, people “re-invented” their dietary origins. Food-related behaviour is to a great degree subject to habits and routines. Changes in eating patterns are normally occurring rather slowly over long periods of time. Many factors have been proposed that affect human eating behaviour. These factors include cultural, evolutionary, social, family, financial and psychological characteristics of individuals. It has been suggested that people use food as a coping mechanism to deal with emotions such as stress, boredom or anxiety, or even to prolong feelings of joy (Arora and Grey, 2020; Auestad and Fulgoni, 2015; Carins and Rundle-Thiele, 2014; Olson, 2016). Besides which, we know that food consumption is largely influenced not only by an individual’s preferences but also by where they eat and with whom. Since the beginning of the COVID-19 pandemic, society has experienced an unprecedented case that many people have spent much more time at home. That also means many people have eaten more meals at home than before the COVID-19 pandemic. Indeed, in the present study, more than three out of 10 of the participants reported weight gain during the COVID-19 pandemic period, whereas one out of five reported weight loss with more prominent similar trends in participants with obesity (40% vs. 16.9%, respectively), which was in line with other studies, too (Haddad et al., 2020; López-Moreno et al., 2020; Robertson et al., 2020). In particular, the literature suggests that due to home-confinement and in response to this stressful situation, people may change their everyday eating behaviour since they experience higher boredom, higher anxiety, higher fear, and higher anger. Research findings from a cross-sectional web-based online survey conducted in Lebanon showed that the fear of COVID-19 was correlated with more eating restraint, weight, and shape concerns (Haddad et al., 2020). Similar findings were demonstrated in another cross-sectional online survey in Spain, where the authors concluded that 38.8% of the respondents experienced weight gain while 31.1% lost weight during confinement and the prevalence of emotional eating was also high (López-Moreno et al., 2020). In the same context, in a United Kingdom online survey, large differences in perceived changes in eating, exercise, and body image during the COVID-19 pandemic period were revealed. Women were more likely to report increasing struggles with regulating eating, preoccupation with food, and worsening body image, compared to men, and this was more prominent among those with a current or past diagnosis of eating disorders (Robertson et al., 2020). Relevant to the observed self-perceived shifts towards a more balanced or a Mediterranean diet-oriented pattern, our findings are in line with other studies conducted in the Mediterranean region (Di Renzo et al., 2020; Rodríguez-Pérez et al., 2020; Rossinot et al., 2020; Sánchez-Sánchez et al., 2020; Tárraga López et al., 2020). Conversely, other web-based surveys conducted in Denmark, United Arab Emirates, Poland, and Saudi Arabia, including the International Online Survey, albeit confirming changes in eating habits during the COVID-19 pandemic, suggested an overall deterioration of diet quality (Alhusseini and Alqahtani, 2020; Ammar et al., 2020; Cheikh Ismail et al., 2020; Meister, 2020; Sidor and Rzymski, 2020). A probable explanation for the observed discrepancy between the Mediterranean countries and the rest could be attributed to the idea of a shared Mediterranean culture mostly related to ancient cultural heritage (Helly, 2018). Hence, probably Mediterranean citizens re-invented instinctively the unique weapon, they had available to cope with the “fear of COVID-19”: the Mediterranean diet. Indeed, recent research pointed out that Mediterranean dietary patterns and better diet quality were all positively correlated with higher psychological resilience, unlike Western-type diets (Bonaccio et al., 2018). This speculation is further reinforced by the fact that in our survey the participants, and in particular participants with underlying diseases, were found to be more likely to adopt healthier lifestyle habits. These findings are justified by other studies that investigated the impact of COVID-19 on people at high risk of severe illness concluding that the “fear of COVID-19” contributes to a negative psychosocial impact in most vulnerable participants (Flint et al., 2020; Grannell et al., 2020). Following the above conjecture, in our survey one out of five participants have started or increased the frequency of receiving dietary supplements as enhancers of the immune system. One could speculate that this observation is presumably interpreted as an attempt to find an easy and quick solution “under the threat of the enemy” although evidence evaluating these supplements in COVID-19 patients is lacking (Adams et al., 2020). Moreover, our findings revealed that almost four out of 10 of the participants engaged in PAs during the COVID-19 pandemic, and almost one out of 10 reported that they have improved both dietary habits and PAs status. The observation that the engagement in PAs conforms with healthier food choices is in line with the findings of a longitudinal observational study conducted in Italy where the authors suggested that PA was positively correlated with fruit, vegetables and fish consumption, thereby mediating the effects of mood states (Amatori et al., 2020). Furthermore, as it was observed here people who changed their dietary habits and PA level to a healthier status, compared to the pre-COVID-19 pandemic period, were mainly of high education level. However, overall, the education level of the participants of this survey was higher than the median level of the total Greek population, a fact that may partially explain the tendency in the study participants towards adopting a healthier lifestyle, since it has already been reported that people with higher education level are more likely to adopt healthier lifestyle changes (Cohen and Syme, 2013). Another possible explanation could be that during the COVID-19 pandemic some of the factors that were responsible for the nutrition transition to more Westernized dietary patterns (i.e., limited time for cooking, gathering with the family, work-related stress, etc.) were less prevalent and thus, people re-invented their dietary origins. Nevertheless, to better understand what consequences these changes in the context of lifestyle have had, for example, in terms of how balanced or unbalanced the diets have been or whether people’s PA habits have improved, we need a number of large-scale population-based nutrition/lifestyle studies. The COVID-19 pandemic has undoubtedly changed how we work, learn, interact and communicate, as social distancing has led to a more virtual existence. Whether the COVID-19 pandemic has changed how people approach their health, as well as health-related behaviours, it is too early to understand. Though much of the world has come to a “stop” or “delay” during the COVID-19 pandemic, the need for healthier eating has not. Throughout the COVID-19 pandemic, there have arisen both benefits and drawbacks of being cooped up with family for long periods of time. Adopting healthier eating habits and more exercise together with other family members during these “escapes” from the restrictions due to the lockdown, may be some of these benefits. The emerging need for nutrition and exercise guidelines during the COVID-19 pandemic is something that governments and health organizations should pay much more attention to, as COVID-19 seems that it will be present for some time, at least the first six months of the new year, 2022. Acknowledgements The authors thank the study investigators and all participants in the online survey for their voluntary enrollment. Authors’ contributions: DP and CP collected the data. DP performed the data analyses and interpreted the results. DP and RK contributed to writing and CP in reviewing the manuscript. All authors read and approved the final manuscript. Availability of data and materials: The dataset used and analysed during the current study is available from the corresponding author on reasonable request. The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Ethical approval: The study was approved by the Ethical Review Board of the University of Thessaly (December 2020). Approval no. authenticating the statement is : 2/25-01-2021. Funding: The authors received no financial support for the research, authorship, and/or publication of this article. ORCID iD: Demosthenes Panagiotakos https://orcid.org/0000-0001-8583-153X ==== Refs References Adams KK Baker WL Sobieraj DM (2020) Myth busters: Dietary supplements and COVID-19. The Annals of Pharmacotherapy 54 (8 ): 820–826.32396382 Alhusseini N Alqahtani A (2020) COVID-19 pandemic’s impact on eating habits in Saudi Arabia. Journal of Public Health Research 9 (3 ): 1868.33024727 Alomari MA Khabour OF Alzoubi KH (2020) Changes in physical activity and sedentary behavior amid confinement: The BKSQ-COVID-19 Project. Risk Management and Healthcare Policy 13 : 1757–1764.33061709 Amatori S Donati Zeppa S Preti A , et al. (2020) Dietary habits and psychological states during COVID-19 home isolation in Italian college students: The role of physical exercise. Nutrients 12 (12 ): 3660.33260711 Amini H Isanejad A Chamani N , et al. (2020) Physical activity during COVID-19 pandemic in the Iranian population: A brief report. Heliyon 6 (11 ): e05411.33163638 Ammar A Brach M Trabelsi K , et al. (2020) Effects of COVID-19 home confinement on eating behaviour and physical activity: Results of the ECLB-COVID19 International Online Survey. Nutrients 12 (6 ): 1583.32481594 Antunes R Frontini R Amaro N , et al. (2020) Exploring lifestyle habits, physical activity, anxiety and basic psychological needs in a sample of Portuguese adults during COVID-19. International Journal of Environmental Research and Public Health 17 (12 ): 4360.32570737 Arora T Grey I (2020) Health behaviour changes during COVID-19 and the potential consequences: A mini-review. Journal of Health Psychology 25 (9 ): 1155–1163.32551944 Auestad N Fulgoni VL III (2015) What current literature tells us about sustainable diets: emerging research linking dietary patterns, environmental sustainability, and economics. Advances in Nutrition 6 (1 ): 19–36.25593141 Bogogiannidou Z Vontas A Dadouli K , et al. (2020) Repeated leftover serosurvey of SARS-CoV-2 IgG antibodies, Greece, March and April 2020. Eurosurveillance 25 (31 ): 2001369.32762796 Bonaccio M Di Castelnuovo A Costanzo S , et al. (2018) Mediterranean-type diet is associated with higher psychological resilience in a general adult population: findings from the Moli-sani study. European Journal of Clinical Nutrition 72 (1 ): 154–160.28952609 Caputo EL Reichert FF (2020) Studies of physical activity and COVID-19 during the Pandemic: A scoping review. Journal of Physical Activity & Health. Epub ahead of print 3 November 2020. DOI:10.1123/jpah.2020-0406. Carins JE Rundle-Thiele SR (2014) Eating for the better: A social marketing review (2000–2012). Public Health Nutrition 17 (7 ):1628–1639 23711161 Cheikh Ismail L Osaili TM Mohamad MN , et al. (2020) Eating habits and lifestyle during COVID-19 lockdown in the United Arab Emirates: A cross-sectional study. Nutrients 12 (11 ): 3314.33137947 Chen P Mao L Nassis GP , et al. (2020) Coronavirus disease (COVID-19): The need to maintain regular physical activity while taking precautions. Journal of Sport and Health Science (2 ): 103–104.32099716 Cohen AK Syme SL (2013) Education: A missed opportunity for public health intervention. American Journal of Public Health 103 (6 ): 997–1001.23597373 Di Renzo L Gualtieri P Pivari F , et al. (2020) Eating habits and lifestyle changes during COVID-19 lockdown: An Italian survey. Journal of Translational Medicine 18 (1 ): 229.32513197 Flint SW Brown A Tahrani AA , et al. (2020) Cross-sectional analysis to explore the awareness, attitudes and actions of UK adults at high risk of severe illness from COVID-19. BMJ Open 10 (12 ): e045309. Grannell A le Roux CW McGillicuddy D (2020) “I am terrified of something happening to me.” The lived experience of people with obesity during the COVID-19 pandemic. Clinical Obesity 10 (6 ): e12406.32885595 Haddad C Zakhour M Bou Kheir M , et al. (2020) Association between eating behavior and quarantine/confinement stressors during the coronavirus disease 2019 outbreak. Journal of Eating Disorders 8 (40 ). Helly D (2018) Cultural heritage protection in the Mediterranean region: Beyond resilience. In: Mediterranean Yearbook 2018. Available at: https://www.iemed.org/observatori/arees-danalisi/arxius-adjunts/anuari/med.2018/Cultural_Heritage_Protection_Damien_Helly_Medyearbook2018.pdf (assessed 4 January 2020). López-Moreno M López MTI Miguel M , et al. (2020) Physical and psychological effects related to food habits and lifestyle changes derived from COVID-19 home confinement in the Spanish population. Nutrients 12 (11 ): 3445.33182816 Meister M (2020) What effect is the corona pandemic having on what Danes eat? In: DTU Food, 23 April 2020. Available at: https://www.food.dtu.dk/english/news/2020/04/measuring-the-effect-of-the-corona-pandemic-on-what-danes-eat?id=134e777a-853d-496c-80ad520cd6ed4546 (accessed 30 December 2020). Olson CM (2016) Behavioral nutrition interventions using e- and m-health communication technologies: A narrative review. Annual Review of Nutrition 36 : 647–664. Robertson M Duffy F Newman E , et al. (2020) Exploring changes in body image, eating and exercise during the COVID-19 lockdown: A UK survey. Appetite 159: 105062 . Rodríguez-Pérez C Molina-Montes E Verardo V , et al. (2020) Changes in dietary behaviours during the COVID-19 outbreak confinement in the Spanish COVIDiet Study. Nutrients 12 (6 ): 1730.32531892 Rossinot H Fantin R Venne J (2020) Behavioral changes during COVID-19 confinement in France: A web-based study. International Journal of Environmental Research and Public Health 17 (22 ): 8444.33202686 Sánchez-Sánchez E Ramírez-Vargas G Avellaneda-López Y , et al. (2020) Eating habits and physical activity of the Spanish population during the COVID-19 pandemic period. Nutrients 12 (9 ): 2826.32942695 Sidor A Rzymski P (2020) Dietary choices and habits during COVID-19 lockdown: Experience from Poland. Nutrients 12 (6 ): 1657.32503173 Tárraga López PJ Panisello Royo JM Carbayo Herencia J , et al. (2020) Cambios observados en la adherencia a la dieta mediterránea en una población española durante el confinamiento debido a la pandemia ocasionada por el SARS-CoV-2 [Changes in adherence to the Mediterranean diet observed in a Spanish population during confinement for the SARS-CoV-2 pandemic]. Nutricion Hospitalaria 10.20960/nh.03275. [In Spanish.] Torales J O’Higgins M Castaldelli-Maia JM , et al. (2020) The outbreak of COVID-19 coronavirus and its impact on global mental health. International Journal of Social Psychiatry 66 (4 ): 317–320.32233719 Tsitsilonis OE Paraskevis D Lianidou E , et al. (2020) Seroprevalence of antibodies against SARS-CoV-2 among the personnel and students of the National and Kapodistrian University of Athens, Greece: A preliminary report. Life (Basel) 10 (9 ): 214.32967110 World Economic Forum (2020) COVID-19: What you need to know about the coronavirus pandemic on 29 September. Available at: https://www.weforum.org/agenda/2020/09/covid-19-coronavirus-pandemic-29-september/ (accessed 28 September 2020). World Health Organization (2020) #HealthyAtHome: Healthy Diet. Available at: https://www.who.int/campaigns/connecting-the-world-to-combat-coronavirus/healthyathome/healthyathome---healthy-diet (assessed 30 December 2020).	33843324	PMC9716055	NO-CC CODE	2022-12-03 23:20:53	no	Nutr Health. 2022 Dec; 28(4):677-683	utf-8	Nutr Health	2,022	10.1177/02601060211009033	oa_other
==== Front Nutr Health Nutr Health NAH spnah Nutrition and Health 0260-1060 2047-945X SAGE Publications Sage UK: London, England 36066026 10.1177/02601060221124068 10.1177_02601060221124068 Special Section on ‘COVID 19’ Original Articles Association of food habit with the COVID-19 severity and hospitalization: A cross-sectional study among the recovered individuals in Bangladesh https://orcid.org/0000-0002-1944-7756 Ganguli Sumon 12 https://orcid.org/0000-0002-9593-1697 Howlader Sabbir 2 Dey Kamol 2 Barua Suman 2 Islam Md. Nazrul 23 Begum Afroza 4 Sobahan Md. Abdus 2 Chakraborty Rivu Raj 5 Hawlader Mohammad Delwer Hossain 6 Biswas Paritosh Kumar 7 1 Biomaterials Research Laboratory, Department of Applied Chemistry and Chemical Engineering, Faculty of Science, 54493 University of Chittagong , Chattogram, Bangladesh 2 Department of Applied Chemistry and Chemical Engineering, Faculty of Science, 54493 University of Chittagong , Chattogram, Bangladesh 3 School of Pharmacy, The University of Queensland, Queensland, Australia 4 Department of Statistics, 54493 University of Chittagong , Chattogram, Bangladesh 5 Department of Surgery, Rangamati Medical College and Hospital, Rangamati, Bangladesh 6 Department of Public Health, School of Health and Life Sciences, 54495 North South University , Dhaka, Bangladesh 7 Department of Microbiology and Veterinary Public Health, Chattogram Veterinary and Animal Sciences University, Chattogram, Bangladesh Sumon Ganguli, Biomaterials Research Laboratory, Department of Applied Chemistry and Chemical Engineering, Faculty of Science, University of Chittagong, Chattogram-4331, Bangladesh. Emails: [email protected]; [email protected] 12 2022 12 2022 12 2022 28 4 771782 © The Author(s) 2022 2022 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background: It was assumed that dietary habits might influence the status of COVID-19 patients. Aim: We aimed at the identification of association of dietary habits with the COVID-19 severity and hospitalization. Methods: It was a retrospective cross-sectional study (n = 1025). We used bivariate and multivariate analyses to correlate the association between self-reported dietary patterns and COVID-19 severity and hospitalization. Results: Dietary habits (black tea, milked tea, pickles, black caraway seeds, honey, fish, fruits, vegetables, garlic, onion and turmeric) were identified with lower risk of COVID-19 severity and hospitalization. Interestingly, the consumption frequency (one-, two- or three-times/day) of rice - the staple food in Bangladesh - was not associated with COVID-19 severity and hospitalization for comorbid patients. In contrast, a moderate rice-eating habit (two times/day) was strongly associated with the lower risk of severity and hospitalization for non-comorbid patients. However, for both comorbid and non-comorbid patients, consumption of black tea, milked tea, pickles and honey were associated with a lower likelihood of severity and hospitalization. Overall, a high consumption (three-times/day) of fish, fruits and vegetables, a moderate consumption of garlic, onion and turmeric spices and a daily intake of black/milked tea, and honey were associated with reduced risk of COVID-19 severity and hospitalization. Conclusions: To reduce the severity of COVID-19, a habitual practice of intaking black tea, milked tea, black caraway seeds and honey along with dietary habit (rice, fish and vegetables) and with a moderate consumption of ginger, garlic, onion, mixed aromatic spices (cinnamon + cardamom + cloves) and turmeric might be suggested. COVID-19 severity hospitalization dietary habits foods spices Research and Publication Cell, University of Chittagong 352/Res/Pln/Pub/Cell/CU/2021 typesetterts19 ==== Body pmcIntroduction COVID-19 – a disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2 - has been a buzzing word across the world posing a severe stress to the global healthcare system. Since the first report in December 2019 in Wuhan, China, COVID-19 has caused over 265 million cases and over 5.24 million deaths globally as of 4th December 2021 (Andersen et al., 2020; WHO, 2021). On the other hand, an unprecedented speed in developing COVID-19 vaccines around the world resulted in a handful of successful vaccine candidates; however, some big concerns still remain such as lasting of vaccine-imparted immunity, vaccine effectiveness against new variant or even facing more infectious variant and new pandemic (Dodd et al., 2021; Forni et al., 2021; Lai et al., 2021; Li et al., 2021). Additionally, the solid evidence of vaccine in terms of longer-term effectiveness, safety, and protection against severe COVID-19 is still missing (Krause et al., 2020). Of note, Delta variant is the most common variants available recently infected people in Bangladesh (Ghosh et al., 2021; Moona et al., 2021). Recently, a new variant designated as Omicron has surged over the world while Bangladesh is experiencing a rapid spreading of third wave from January 2022 (TBS, 2022). So far, scientists and physicians have noticed that people exposed to SARS-CoV-2 are not equally infected with COVID-19 disease (asymptomatic to mild to severe to death) (Kim et al., 2020; Shi et al., 2020). Notable that, several factors have been found to be associated with the severity and mortality rate which may increase the risk in COVID-19 patients, including aged person (>60 years) and having comorbidities (Akbar and Gilroy, 2020; Martins-Filho et al., 2020; Richardson et al., 2020). However, elderly-aged and comorbidities are not enough to explain the vulnerability to this severe infection. Additionally, pre-existing dietary habits of people appear to be linked to variability in COVID-19 symptoms (Butler and Barrientos, 2020; Hull et al., 2020; Whittemore, 2020; Zbinden-Foncea et al., 2020). Nevertheless, it is poorly known exactly why some people are more vulnerable whilst others less. It is obvious that balanced nutritional status of the individuals can fight against any disease. Recently, Alam et al. (2021) has carried out a comprehensive review where strong nutritional interventions serve as a therapeutic tool against COVID-19 (Alam et al., 2021). Abdulah and Hassan (2020) explored the relation between dietary factors and infection/mortality rates globally; briefly, higher intake of fruits/sugar-sweetened products and lower intake of legumes/beans had a positive and negative effect on infection and mortality rates (Abdulah and Hassan, 2020). Several other review and original articles forecasted that nutritional management/supplement/interventions could be considered as a therapeutic gun against pandemic (Brugliera et al., 2020; Cobre et al., 2021; Grant et al., 2020; Iddir et al., 2020; Lidoriki et al., 2020; Liu et al., 2020; Moscatelli et al., 2021; Singh et al., 2020). Particularly, very recently, Kim et al. (2021), Salazar-Robles et al. (2021), Tavakol et al. (2021), and Merino et al. (2021) reported that increased level of physical activity and some dietary patterns such as consuming plant-based diets, fruits and poultry were associated with less severity of COVID-19 and disease duration (Kim et al., 2021; Merino et al., 2021; Salazar-Robles et al., 2021; Tavakol et al., 2021). However, dietary habit considerably varies across cultures as well as within culture among people. All the reports mentioned earlier mainly focused on the Western and Mediterranean diets which are typically less spicy, though various spices are regarded as hidden treasures of numerous therapeutic components and related health benefits (Peter, 2006; Sharif et al., 2018). Consequently, it is critical to decipher the impact of dietary habits including spices consumption on the susceptibility to COVID-19, hospitalization risk and recovery. In this scenario, various predictors such as pre-existing food habits of infected people, symptoms ranging from asymptomatic, mild to moderate to severe can be considered for the study. No doubt, the eating habit of Bangladeshi people is a very complex issue. Keeping this in mind, we aimed to focus on the specific food habits such as various spices and drinks, dietary carbohydrates, proteins and fats. Furthermore, biological attribution of variation in age and gender were considered to find out their association with COVID-19 severity and hospitalization (Hu et al., 2021; Kushwaha et al., 2021). The various benefitting predictors, of course will not terminate COVID-19 infection rate, but may help to reduce the pain and mortality rate of the patients. Therefore, this study will help to examine this association (beneficiary or worsening) so that a new combo approach can be developed to relieve patients from COVID-19 sufferings and saves as many lives as possible via reducing severity and prompting recovery during this deadly contagious disease. Though a few investigations have been done around the world (Bousquet et al., 2020; Ingram et al., 2020; Jayawardena and Misra, 2020; Maiorino et al., 2020; Martinez-Ferran et al., 2020), to our knowledge, this is the first survey-based findings in Bangladesh which deals with the investigation on the association between pre-existing food habits with COVID-19 severity and hospitalization. Methods Study design, participants, sites and data collection The study included laboratory-confirmed COVID-19 patients who were diagnosed positive and later found to be negative at the COVID-19 test laboratory by RT-PCR assay. The risk factors such as age, sex, dwelling and comorbidities were considered to quantify the association between COVID-19 severity and hospitalization status of the recovered individuals. By following the guidelines from Directorate General of Health Services (DGHS), Ministry of Health and Family Welfare (MOHFW), Bangladesh and World Health Organization other than clinical data, the symptoms of recovered patients were divided into three categories (we did not find any moderate case among the respondents): asymptomatic, mild and severe (DGHS, MOHFW, 2020). One might raise a valid question about why the COVID-19 patients enrolled for the study were not clinically assessed to ascertain mild, moderate, and severe cases. Inclusion of clinical examination of cases could have been more ideal for this study, but was impracticable for recruiting a fairly large number of cases from a South Asian country like Bangladesh where access to quality health services (proper documentation, electronic medical data archive etc.) for the common people is insufficient. To overcome the limitation, we compared the patients self-descried reports, as received during the telephonic interviews, with the guidelines provided by the DGHS (Bangladesh) aligned with the COVID-19 categorizations described by WHO (DGHS, MOHFW, 2020), to discriminate among the severity of the cases enrolled. In our previous study, we followed the similar way to ascertain the epidemiological and symptom data (Ganguli et al., 2022). Of note, it was confirmed that the attending doctors had already diagnosed the patients during their infection period and the same was reported by the respondents (patients) through telephonic interviews. Additionally, physicians directed the patient’s severity (asymptomatic, mild, severe) of COVID-19 disease upon examination of laboratory reports based on the guidelines of DGHS, MOHFW and WHO in order to avoid unnecessary hospital admission load caused by the COVID-19. Huang C. (2020) and his group also directly communicated with patients or their relatives to ascertain the epidemiological and symptom data which were not available from electronic medical records (Huang et al., 2020). For asymptomatic, mild, and moderate cases, interviews were conducted over telephone following the list collected from city corporation booths in Dhaka, Bangladesh by Hawlader M. D. H. and his group (Mohsin et al., 2021). COVID-19 severities and hospitalization were compared between the recovered patients with or without comorbidity status. We randomly included 1025 recovered individuals who provided completed data (See Supplementary Information for details). The study sites were two districts (Dhaka and Jhenaidah) of Bangladesh. All data were collected through telephone interviews and the same were recorded in an electronic form. The data collection period was November 2020 to March 2021. Statistical analysis Bivariate and multivariate analyses were done to identify the association of the selected parameters with COVID-19 status. We used graphical presentation and tabular format to display the distribution of age, sex and dwelling for identifying the pattern of the patients who suffered from COVID-19. We performed the bivariate analysis to compare age, hospitalization, and comorbidities between male and female. Chi-square test was performed to test the association between the categorical variables (sex, dwelling, consumption habits, dietary foods and spices) while Student’s t-test was used to explore the significance of the mean difference of the continuous variables (age). In multivariate analysis, we performed three regression models to assess the influencing factors of two outcome variables such as hospitalization status (model 1) and degree of severity (model 2). The binary logistic regression model was used to estimate each model of interest It is noted that the outcome variable “degree of severity” was defined as “severe” if the patients reported the severity level of the symptoms as “severe” and as “not severe” if the patients reported the severity level as “mild” or they were asymptomatic. The explanatory variables were age (in years), sex (male vs. female), dwelling status (urban vs. village), consumption habits (black tea, milked tea, coffee, betel leaf, honey, black caraway seeds and pickles; consumption vs non-consumption), dietary foods (rice fish, meat, vegetables and fruits; consumption-once/twice/thrice per day vs non-consumption) and spices (ginger, garlic, onion, turmeric, bay leaf and mixed aromatic spices (cinnamon + cardamom + cloves); consumption vs non-consumption). To identify consumption vs non-consumption habits, at least 3 months period prior to COVID-19 infection was considered for analyzing risk association of dietary habits to hospitalization and severity. We considered the odds ratios (ORs), 95% confidence intervals (CI), and p values for the variables in the models. In the case of all statistical analyses, we assumed significance only if p < 0.05 (two-tailed). We performed the statistical analysis using SPSS (version 25). Results Demographic data Demographic data suggested that the non-comorbid patients differed from the comorbid patients. Of the patients enrolled for the study, 45.6% (n = 467) had at least one comorbidity (Supplementary Information (SI) Figure 1). Overall, comorbid patients required more hospitalization (12.6%) (SI Figure 2). In addition, comorbidity might have influenced the severity of the diseases. Though mild symptoms prevailed mostly for both types of patients, severe symptoms were found comparatively higher in the case of comorbid patients (Figure 1). Median age of the non-comorbid patients was 31 ± 10.7 year which was much less than that of comorbid patients (45 ± 12.8 year) (SI Figure 3). It is worth noting that male and urban patients outnumbered their counterparts in terms of numbers (SI Figure 4). On the other hand, regardless of gender and comorbidity status, a higher percentage of urban patients suffered from severe symptoms and required a greater rate of hospitalization compared to those of dwelling in villages (Figure 1). For instance, comorbid urban dwellers experienced a greater degree of disease severity (urban 38.6% vs village 32.1%) as well as a higher rate of hospitalization (urban 13.2% vs village 9.9%) than the villagers. A similar trend also was observed for non-comorbid patients (severity: urban 20.7% vs village 14.0% and hospitalization: urban 7.2% vs village 4.4%). The results of chi-square test showed that significant association between severity and socio-demographic characteristics (sex, dwelling). Alike severity, similar results were obtained in case of hospitalization as well (Supplementary Table 1). Figure 1. Demographic data of gender (a, b) and habitation (c, d) in terms of hospital admission and degree of severity. Comparison among the consumption habits of tea, honey, black caraway seeds, pickles and betel leaf Consumption of black tea was found as most popular habit in comparison to milked tea and coffee for both types of patients (SI Figure 5(a) and 5(b)). Of the patients with comorbidity, 88.4% consumed black tea (SI Figure 5(a)). Betel leaf was the least consumed of all the regular habits (28.8% for non-comorbid and 40.1% for comorbid), while honey, black caraway seeds, and pickles were consumed by more than 70% of non-comorbid patients (SI Figure 5(b)). Their corresponding consumption rates were a little lower among the comorbid patients (68–82%). As rice is the staple food of Bangladesh, a large number of patients consumed rice twice or thrice a day (SI Figure 5(c) and 5(d)). Fish are found adequately and cheaply compared with another source of animal protein in Bangladesh. Consequently, two times consumption of fish were the frequent phenomenon because of the traditional habit where they mostly consumed meat once a day. Vegetables (leafy or fleshy; fiber and vitamin-rich) were evenly popular for both types of the patients. Surprisingly, fruits were consumed thrice per day by non-comorbid patients (55.9%) but much less by comorbid patients (22.9%). The percentages of patients with moderate consumption of spices (70–79%) were higher than those of low and high consumption of spices for both types of the patients (SI Figure 5(e) and 5(f)). Impact of dietary habits towards hospitalization and severity Table 1 shows the correlation among food habits, severity and hospitalization for comorbid and non-comorbid patients. The trend of degree of severity and the hospitalization rate for non-comorbid patients were found decreasing with consumption of black and milked tea, honey, black caraway seeds and pickles. Hospitalization rates were less than 7%. Besides, severity decreased along with regular consumption of these habitual foods. However, mixed results were observed for the comorbid patients. Less hospitalization was observed for comorbid patients who had consumed pickles (12.1%), black tea (11.9%), milked tea (11.3%) and honey (11.3%). However, opposite trends for hospitalization of comorbid patients were found for coffee, betel leaf and black caraway seeds. The least severity was found in case of comorbid black tea consumers (29.7%) where the severity was increased up to 35.6% for black caraway seeds consumers. On the other hand, severity did not exceed 20% for non-comorbid patients when they consumed these foods (Table 1). Table 1. Correlation between food habit, severity and hospitalization. Comorbid Non-comorbid Hospitalization Severity Hospitalization Severity No Yes Asymptomatic Mild Severe No Yes Asymptomatic Mild Severe Black tea Not Consumed 82.8% 17.2% 6.0% 55.3% 38.7% 92.6% 7.4% 11.3% 69.2% 19.5% Consumed 88.1% 11.9% 9.4% 60.9% 29.7% 93.5% 6.5% 12.3% 69.1% 18.5% Milk tea Not consumed 83.7% 16.3% 6.1% 54.9% 39.0% 84.6% 15.4% 11.4% 67.5% 21.1% Consumed 88.7% 11.3% 7.3% 59.3% 33.3% 95.9% 4.1% 11.5% 69.7% 18.9% Coffee Not consumed 88.1% 11.9% 4.9% 55.3% 39.8% 91.8% 8.2% 10.7% 69.5% 19.8% Consumed 86.8% 13.2% 8.5% 57.2% 34.3% 94.2% 5.8% 12.9% 68.6% 18.6% Betel leaf Not consumed 88.7% 11.3% 5.1% 55.1% 39.8% 92.5% 7.5% 13.4% 68.8% 17.8% Consumed 85.2% 14.8% 7.2% 56.7% 36.1% 95.5% 4.5% 10.7% 69.3% 20.0% Honey Not consumed 84.5% 15.5% 5.0% 56.1% 38.9% 89.9% 10.1% 11.6% 67.4% 21.0% Consumed 88.7% 11.3% 9.5% 56.1% 34.5% 94.5% 5.5% 11.4% 69.8% 18.8% Black caraway seeds Not consumed 89.9% 10.1% 5.0% 57.6% 37.4% 92.1% 7.9% 15.9% 63.4% 20.7% Consumed 86.3% 13.7% 7.0% 55.5% 37.5% 93.9% 6.1% 9.6% 71.6% 18.8% Pickles Not consumed 85.4% 14.6% 6.5% 54.4% 39.1% 91.2% 8.8% 10.3% 69.4% 20.3% Consumed 87.9% 12.1% 6.3% 62.5% 31.3% 93.8% 6.2% 17.6% 68.1% 14.3% Ricea Once 87.1% 12.9% 4.3% 58.6% 37.1% 89.2% 10.8% 2.7% 59.5% 37.8% Twice 86.7% 13.3% 7.6% 54.3% 38.1% 97.3% 2.7% 13.3% 70.1% 16.7% Thrice 88.2% 11.8% 5.9% 57.2% 36.9% 89.9% 10.1% 10.9% 69.6% 19.5% Fisha Once 89.3% 10.7% 7.1% 46.4% 46.4% 91.6% 8.4% 13.7% 59.5% 26.7% Twice 86.7% 13.3% 6.8% 59.0% 34.1% 95.0% 5.0% 12.2% 69.4% 18.3% Thrice 86.8% 13.2% 4.7% 59.4% 35.8% 91.9% 8.1% 8.1% 77.2% 14.8% Meata Once 85.8% 14.2% 7.1% 50.0% 42.9% 92.6% 7.4% 14.8% 59.3% 25.9% Twice 87.8% 12.2% 6.8% 63.3% 29.9% 93.7% 6.3% 9.4% 75.4% 15.2% Thrice 91.3% 8.8% 3.8% 61.3% 35.0% 94.4% 5.6% 8.1% 79.0% 12.9% Vegetablesa Once 89.9% 10.1% 9.0% 32.6% 58.4% 88.4% 11.6% 11.2% 62.9% 25.8% Twice 87.4% 12.6% 5.7% 58.3% 36.0% 96.1% 3.9% 13.2% 68.0% 18.9% Thrice 86.2% 13.8% 5.9% 64.5% 29.6% 100.0% 0.0% 10.0% 72.6% 17.4% Fruitsa Once 87.2% 12.8% 5.8% 52.9% 41.2% 90.7% 9.3% 12.7% 63.3% 24.1% Twice 88.4% 11.6% 7.0% 59.3% 33.7% 95.2% 4.8% 15.0% 67.1% 18.0% Thrice 86.9% 13.1% 7.5% 61.7% 30.8% 100.0% 0.0% 9.3% 71.8% 18.9% Low 77.6% 22.4% 7.5% 56.7% 35.8% 90.1% 9.9% 16.0% 59.3% 24.7% Medium 89.6% 10.4% 7.0% 57.7% 35.4% 94.5% 5.5% 11.4% 70.1% 18.5% High 85.5% 14.5% 1.8% 45.5% 52.7% 87.2% 12.8% 2.6% 79.5% 17.9% Ginger Low 78.3% 21.7% 0.0% 48.1% 51.9% 87.2% 12.8% 16.7% 59.0% 24.4% Medium 89.9% 10.1% 7.2% 56.9% 35.8% 94.5% 5.5% 11.3% 70.3% 18.4% High 82.7% 17.3% 7.2% 58.0% 34.8% 90.1% 9.9% 4.4% 75.6% 20.0% Onion Low 78.8% 21.2% 0.0% 55.3% 44.7% 87.5% 12.5% 15.8% 60.5% 23.7% Medium 89.8% 10.2% 7.1% 55.6% 37.3% 94.6% 5.4% 11.0% 70.4% 18.5% High 80.9% 19.1% 7.6% 59.1% 33.3% 90.8% 9.2% 8.9% 71.4% 19.6% Turmeric Low 79.7% 20.3% 0.0% 55.8% 44.2% 86.8% 13.2% 17.5% 61.3% 21.3% Medium 89.6% 10.4% 7.0% 56.1% 36.9% 94.1% 5.9% 11.1% 69.8% 19.1% High 81.4% 18.6% 7.2% 56.5% 36.2% 92.5% 7.5% 2.6% 78.9% 18.4% Bay leaf Low 77.8% 22.2% 0% 51.1% 48.9% 86.5% 13.5% 14.5% 59.1% 26.4% Medium 90.7% 9.3% 7.5% 56.3% 36.1% 94.2% 5.8% 11.4% 71.0% 17.5% High 82.2% 17.8% 5.6% 57.8% 36.7% 92.7% 7.3% 2.7% 78.4% 718.9% Mixed Aromatic Spices Low 78.9% 21.1% 0% 56.1% 43.2% 88.9% 11.1% 15.4% 59.0% 25.6% Medium 90.4% 9.6% 7.5% 56.8% 36.6% 94.1% 5.9% 11.1% 71.4% 17.5% High 81.8% 18.2% 5.6% 56.7% 37.8% 92.3% 7.7% 2.8% 77.8% 19.4% a Per day. In case of non-comorbid patients, moderate (twice) to high (thrice) consumption of rice, fish, meat, vegetables, and fruits were associated with mild form of disease and lower rate of hospitalization. For instance, three times consumption of fish per day reduced severity from 26.7% (once/day) to 14.8%. Whereas two times consumption of fish per day reduced the rate of hospitalization from 8.4% (once/day) to 5.0%. Overall, a similar trend was observed for the patients with comorbidity. Interestingly, no hospitalization was observed in case of the non-comorbid consumers of vegetables and fish (thrice/day) where rice, fish and meat were associated with 5–10% hospitalization for the non-comorbid patients. On the contrary, hospitalization percentages (10–13%) for comorbid patients did not differ much along with the frequency of the consumption of these food. In case of ginger, garlic, onion, turmeric and mixed aromatic spices (MAS) (cinnamon + cardamom + cloves), reduced degree of severity and hospitalization were observed in case of the moderate consumers of these spices for both types of patients (Table 1). For example, moderate consumption of ginger reduced the hospitalization rate to 10.1% from 21.7% (low consumption) and 17.3% (high consumption) for the comorbid patients, whereas to 5.5% from 12.8% (low consumption) and 9.9% (high consumption) for the non-comorbid patients. Overall, moderate consumption of spices was associated with least hospitalization for comorbid patients (<10%) which were quite similar to the trend for non-comorbid patients. Risk association of dietary habits to hospitalization and severity The results of chi-square test indicated the significant association between severity and food habits (consumption habits, dietary foods, or spices) and similar results were obtained in cases of hospitalization as well (Supplementary Table 1). Table 2 displays the odds ratio of the dominating factors of hospitalization and severity of symptoms due to COVID-19. It is manifest from a cursory glance at the habitual consumption of the respondents, those who did not consume any kinds of hot drinks such as black or milk tea or coffee, natural remedies such as honey or black caraway seeds had more likelihood of being hospitalized due to COVID-19 compared to the counterpart. Similar evidence has been observed for both non-comorbid and comorbid conditions. Table 2. Risk association for hospitalization and severity. Hospitalization Severity Non-comorbid Comorbid Non-comorbid Comorbid Category OR (95%CI) p-valuea OR (95%CI) p-valuea OR (95%CI) p-valuea OR (95%CI) p-valuea Consumption Habits Black tea Not consumed 1.167 (.348–3.915) .061 1.146 (.488–2.687) .045 .602 (.270–1.340) .014 1.271 (.644–2.511) .489 Milked tea Not consumed 3.317 (1.286–8.556) <.001 1.597 (.777–3.280) .003 1.129 (.619–2.060) .032 1.191 (.689–2.031) .022 Coffee Not consumed 2.153 (.798–5.805) .063 1.357 (.677–2.720) .039 1.054 (.622–1.787) .045 1.267 (.794–2.022) .044 Betel leaf Not consumed 2.407 (.786–7.371) .197 .715 (.380–1.347) .300 1.247 (.718–2.166) .433 1.544 (.984–2.422) .283 Honey Not consumed 1.569 (.583–4.221) .046 1.750 (.869–3.527) .017 1.028 (.558–1.894) .030 1.265 (.746–2.126) .122 Black caraway seeds Not consumed 1.295 (.469–3.578) .027 2.238 (.958–5.227) .043 1.031(.488–1.95) .028 1.029 (.583–1.814) .085 Pickles Not consumed 1.583 (.558–4.489) .065 1.016 (.453–2.279) .269 1.605 (.783–3.292) .197 1.232 (.680–2.232) .047 Dietary Foods b Rice Once .002 .399 .003 .521 Twice .386 (.073–2.046) <.001 1.242 (.457–3.378) .671 .242 (.103-.571) .001 .785 (.411–1.500) .108 Thrice .980 (.207–4.641) .002 1.522 (.672–3.448) .314 .465 (.193–1.123) .089 1.157 (.571–2.344) .086 Fish Once .318 .257 .710 .442 Twice .311 (.095–1.019) .131 .342 (.095–1.230) .100 1.344 (.654–2.764) .421 1.326 (.700–2.514) .387 Thrice .602 (.149–2.433) .415 .547 (.183–1.639) .282 1.090 (.339–3.507) .884 1.920 (.685–5.383) .215 Meat Once .791 .052 .012 .063 Twice 3.508 (.847–14.526) .517 5.003 (1.282–19.529) .020 .995 (.855–8.010) .017 .566 (.308–1.039) .066 Thrice 4.590 (.919–22.910) .812 2.532 (1.672–9.537) .170 .917 (.282–3.506) .013 .569 (.205–1.578) .279 Vegetable Once <.001 .017 .008 <.001 Twice .394 (.128–3.524) .006 .751 (.249–2.271) .012 .927 (.423–11.722) .045 .444 (.225-.878) .020 Thrice .182 (.020–1.690) .034 .798 (.364–1.752) <.001 .867 (.911–5.642) .039 .201 (.091-.442) <.001 Fruit Once .006 .049 .186 .080 Twice .893 (.324–1.723) .044 .785 (.325–1.895) .014 1.028 (.233–4.527) .071 .958 (.516–1.778) .089 Thrice .346 (.128–3.695) .004 .755 (.293–1.945) .036 2.081 (.428–10.130) .064 .813 (.435–1.520) .117 Hospitalization Severity Non-comorbid Comorbid Non-comorbid Comorbid Category OR (95%CI) p-valuea OR (95%CI) p-valuea OR (95%CI) p-valuea OR (95%CI) p-valuea Spices consumption Ginger Low .048 .024 .035 .199 Medium .823 (.001–14.330) .037 .420 (.037–4.729) .016 .720 (.025–7.152) .049 .226 (.024–2.136) .194 High .774 (.621–3.249) .014 .195 (.006–5.967) .650 .683 (.324–4.287) .027 .588 (.040–8.604) .068 Garlic Low .005 .017 .011 .014 Medium .821 (.056–1.166) .016 1.059 (.017–7.555) .014 .957 (.554–11.166) .013 .945 (.008–2.807) .020 High .329 (.192-.736) .002 .656 (.054–7.978) .006 .859 (.099–7.392) .030 .856 (.196–7.664) .046 Onion Low .043 .017 .046 .333 Medium .915 (.082–3.219) .036 .315 (.006–16.438) .024 .864 (.568–7.256) .049 7.152 (1.440–16.316) .167 High .955 (.469–2.008) .022 .746 (.058–9.586) .005 .936 (.268–5.773) .044 3.669 (.070–9.106) .520 Turmeric Low .214 .036 .094 .031 Medium .646 (.109–2.563) .186 .277 (.005–15.268) .038 .260 (.193–2.430) .041 2.220 (.361–13.645) .038 High .869 (.571–3.438) .094 .442 (.013–15.163) .010 .988 (.307–1.811) .080 .951 (.008–8.207) .043 Bay leaf Low .190 .097 .013 .099 Medium 1.670 (.050–16.143) .209 3.011 (.031–295.372) .082 .643 (.040–2.963) .066 1.138 (.187–6.916) .083 High .864 (.559–7.326) .082 .794 (.009–72.150) .051 .770 (.593–1.038) 045 .843 (.573–3.227) .089 Mixed Aromatic Spices Low .026 .037 .013 .037 Medium .587 (.013–2.599) .076 .637 (.008–48.027) .089 .603 (.084 4.346) .003 1.414 (.216–9.259) .018 High .739 (.318–6.544) .064 1.017 (.016–66.299) .042 .898 (.495–1.428) .002 .648 (.187–1.054) .045 a Two-tailed independent t-test, bPer day, OR: odds ratio. COVID-19 patients without comorbidity had 57% (OR 1.57; p < 0.05) and 30% (OR 1.30; p < 0.05) higher risk of admitting hospital when they did not consume honey and black caraway seeds respectively than those patients who consumed these items. While these figures were 75% (OR 1.75; p < 0.05) and 124% (OR 2.24; p < 0.05) for COVID-19 patients with comorbid condition. The risk of hospitalization was surprisingly 232% (OR 3.32; p < 0.01) and 60% (OR 1.60; p < 0.01) higher respectively in case of non-comorbid and comorbid milked tea consumer. However, habit of milked tea consumption has less benefit compared to black tea. These figures were 13% and 19% respectively for non-comorbid and comorbid patients in developing the risk of severe symptoms. Honey (OR 1.028; p < .05) and black caraway seeds (OR1.038; p < .05) also had a significant effect on the occurrence of severe symptoms in COVID-19 patients without comorbidity; however, surprisingly, they had no significant effect (p > .05) on comorbid patients. There was less likelihood of hospitalization for the non-comorbid patients who had consumed rice, vegetables, and fruits two or three times daily. The likelihood of being hospitalized for the non-comorbid patients decreased by 61 percent (OR 0.39; p < 0.01) with increased (twice daily) consumption of rice. For comorbid patients, there was no significant effect (p > .05) of rice consumption on their hospitalization and degree of severity. Moreover, patients without comorbidity who ingested more (2 times/day) vegetables and fruits were 61% and 11% respectively less likely to be hospitalized compared to those who did not take such foods once a day. If these patients increased their daily consumption level of such foods, i.e., three times/day, the corresponding figures also increased to 82% and 65%, respectively. Vegetables and fruits consumption also reduced the likelihood of hospitalization and severity of the patients with comorbidity. Interestingly, the comorbid patients who took meat two times daily had 400% more likelihood of being hospitalized due to COVID-19 compared to the counterpart though the likelihoods for developing severe symptoms were slightly decreased for both two- and three-time consumption (OR 0.995 and 0.917 respectively). The likelihood of having severe symptoms decreased with an increased intake of vegetables. Specially, three-time consumption per day was associated with the least likelihood (with respect to one-and two-time consumers) of hospitalization and severe symptoms both for comorbid and non-comorbid patients. According to the amount of spices intake, more consumption of ginger, garlic, and onion reduced the hospitalization risk for both comorbid and non-comorbid patients. The likelihood of being hospitalized due to COVID-19 for the non-comorbid patients decreased by 23% (OR 0.77; p < 0.01) and 67% (OR 0.33; p < 0.01) with the increased (high) levels of ginger and garlic intake, respectively. In addition to these spices, other spices such as turmeric and bay leaf also played a significant role in reducing the risk of hospitalization for patients with comorbidity. The comorbid patients who consumed a medium and high levels of turmeric had 72% (OR 0.28; p < 0.05) and 56% (OR 0.44; p < 0.01) respectively less likelihood of hospitalization. Medium and high levels of consumption of all the aforementioned spices had a significant effect on reducing the risk of severe symptom development for the patient without comorbidity. In the case of comorbid patients, only high-level intake of garlic (OR 0.86; p < 0.05), turmeric (OR 0.95; p < 0.05), and mixed aromatic spices (OR 0.65; p < 0.05) had a significant effect on developing the severity symptom, however, surprisingly, those patients who consumed medium level of turmeric (OR 2.22; p < 0.05), and mixed aromatic spices (OR 1.41; p < 0.05) had the opposite effect. Discussion Hospitalization requirements were relatively higher in case of comorbid patients which might be due to less immunity (Callender et al., 2020; Castle et al., 2005). In this study, comorbid patients experienced higher severity and had the fewest asymptomatic cases which are coherent with prior investigations (Davies et al., 2020). According to our data, males prevailed over females in number of cases where most of the patients were urban dwellers which also reflected the previous reports (Bwire, 2020; Girdhar et al., 2021; Peckham et al., 2020). Of note, hospitalization and severe symptoms were observed in higher percentages for comorbid and non-comorbid female patients than their male counterparts which might be observed for the first time in the South-Asian country. Females, specially, aged ones usually suffer from different post-menopausal difficulties which might trigger worseCOVID-19 outcomes (Colditz et al., 2010; Costeira et al., 2021). In addition, Delta variant of coronavirus (SARS COV-2) were predominant during our investigation, therefore, possibly females were vulnerable against Delta variant (Chen et al., 2021; Rangchaikul and Venketaraman, 2021).However, villagers required less hospitalization and suffered mostly from mild symptoms. This phenomenon can be explained by the fact that they were accustomed to more physical activities, greener abode and contamination-free foods than the urban peoples (Chen et al., 2017; Riva et al., 2009). However, easy access to hospital for urban peoples cannot be ruled out at this moment (Bain et al., 2014; Chatterjee and Sarkar, 2021; Lee et al., 2015). We found that consumption of black tea, milked tea, honey and pickles were responsible for decreasing the degree of severity and hospitalization requirement due to COVID-19 (Table 1). Particularly, consumption of black tea appeared to be liked with lesser severity (non-consumer 38.7% vs consumer 29.7%) and lower hospitalization (non-consumer 17.2% vs consumer 11.9%) for comorbid patients. It is worthwhile to mention that the role of natural products for defending against infectious disease is a widespread practice from the earlier days. Because of their preventative properties against different microbial infection, they are being suggested to use in case of COVID-19 as well, somewhere, they have been proved to be effective (Aman and Masood, 2020; Ayivi et al., 2021; Gasmi et al., 2021). It is well known that tea and coffee contain antioxidants which are regarded as the immune promoters (Açıkalın and Sanlier, 2021; Bhattacharyya et al., 2003; Hamer, 2007; Turkmen et al., 2007). Even tea has been reported as a bioactive modulator of innate immunity in cases of COVID-19 (Chowdhury and Barooah, 2020). Though coffee is considered as an effective beverage against different diseases, its impact on the recovery from COVID-19 is yet to be concluded (Belaroussi et al., 2020; Kennedy et al., 2021; Wierzejska, 2017).Black caraway seeds and pickles were also reported for their high medicinal values as they are well known for their anti-microbial actions (Chakraborty and Roy, 2018; Forouzanfar et al., 2014). Taken collectively, it can be recalled that due to antimicrobial, antiviral, antioxidant properties and having vitamins/minerals into tea, honey and pickles might play a role against COVID-19. On the other hand, increased consumption of rice, fish, meat, fruits and vegetables were found to be associated with less severity and hospitalization (Table 1). Though direct relationship between rice consumption and COVID-19 severity and hospitalization is not evident, but, rice consumption might be beneficiary for avoiding massive damage by this disease. As carbohydrates are closely associated with immune components at molecular level, carbohydrate rich food is suggested for COVID-19 patients in different studies (Kumbhar et al., 2021). Fish and meat both contain protein which might be the factor for less hospitalization and severity among the consumers. This result is coherent with previous studies (Batiha et al., 2021; Fan et al., 2020). Fruits and vegetables contain different vitamins and minerals including vitamin-C which assist the immune system (Carr and Maggini, 2017). For this reason, most of the dietary guidelines recommended for taking vegetables and fruits to prevent COVID-19 (Jayawardena and Misra, 2020). Besides, moderate consumption of spices (ginger, garlic, turmeric, onion, mixed aromatic spices and bay leaf) was linked with less hospitalization and severity comorbid and non-comorbid patients. Ginger was reported to modulate oxidative stress and prostaglandins as harmful factors in COVID-19 (Mashhadi et al., 2013; Mohamed et al., 2015). In addition, it is capable of modulating improper effect or T cell responses in COVID-19 (Jafarzadeh et al., 2021). Garlic is a potential therapeutic as it contains organosulfur and flavonoid compounds which can be used to fight with COVID-19 (Khubber et al., 2020). Turmeric contains curcumin which is regarded as a treatment for COVID-19 infection (Babaei et al., 2020). Meanwhile, onion and bay leaf have also high medicinal values which might be the reason for less damage among the users. High and low consumption of these spices were observed with higher hospitalization and severity, moderate consumption of these spices were mostly favorable. Significant correlation between food habits with severity and hospitalization for comorbid and non-comorbid COVID-19 patients were observed. Among the pre-existing habits, consumption of black tea, milked tea, pickles, honey, and black caraway seeds were identified as positive factors of reducing the degree of severity and hospitalization for COVID-19 patients. Two- or three-times/day consumption of fish, fruits and vegetables may reduce severity and hospitalization requirements. On the other hand, comorbid patients with a daily meat intake (twice daily) were at a very high risk of hospitalization. Moderate consumption of ginger, garlic, onion, turmeric and MAS was found as favorable and hence recommended. Therefore, we hypothesized that a combo approach of nutritional management with the necessary medical diagnose and treatment might be suggested against COVID-19. However, cautions should be considered as it is first time survey-based reports among the recovered individuals in Bangladesh and sixth report in world. Limitations This study included recovered patients from only two selected cities in Bangladesh. No specific community population involvement in the present study. Therefore, the diversity of the sample size is questionable. We do not have clinical data such as oxygen level during infection, CBC, SGPT, blood glucose, ferritin, D-dimer, ESR, Creatinine, HbA1C, etc. Moreover, we could not record the duration of the underlying conditions. In addition, we collected data through a telephone call; there could be misinformation or biased information from the interviewer and participants. However, this study reconfirms the global findings that preexisting nutritional habits associated with low severity and hospitalization. Supplemental Material sj-docx-1-nah-10.1177_02601060221124068 - Supplemental material for Association of food habit with the COVID-19 severity and hospitalization: A cross-sectional study among the recovered individuals in Bangladesh Click here for additional data file. Supplemental material, sj-docx-1-nah-10.1177_02601060221124068 for Association of food habit with the COVID-19 severity and hospitalization: A cross-sectional study among the recovered individuals in Bangladesh by Sumon Ganguli, Sabbir Howlader, Kamol Dey, Suman Barua, Md. Nazrul Islam, Afroza Begum, Md. Abdus Sobahan, Rivu Raj Chakraborty, Mohammad Delwer Hossain Hawlader and Paritosh Kumar Biswas in Nutrition and Health Acknowledgements We are thankful to the authority of the Department of Applied Chemistry and Chemical Engineering, University of Chittagong, Bangladesh for their logistic support. Author contributions: (I) Conception and design: S Ganguli, S. Howlader, S Barua, K Dey, MN Islam (II) Administrative support: S. Ganguli, S. Barua, K Dey (III) Provision of study materials or patients: K Dey and RR Chakraborty (IV) Collection and assembly of data: S. Howlader, MDH Hawlader, A. Begum and MA Sobahan (V) Data analysis and interpretation: S Ganguli, MDH Hawlader, RR Chakraborty and A Begum (VI) Manuscript writing: All authors (VII) Final approval of manuscript: All authors Availability of data and materials: Available upon request. Consent for publication: Yes. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Research and Publication Cell, University of Chittagong, Chattogram (grant number 352/Res/Pln/Pub/Cell/CU/2021). Ethical approval: The protocol was approved by the ethics committee, Chattogram Veterinary and Animal Sciences University (CVASU), Chattogram, Bangladesh (CVASU/DIR/(R & E)/EC/2020.191/1). ORCID iDs: Sumon Ganguli https://orcid.org/0000-0002-1944-7756 Sabbir Howlader https://orcid.org/0000-0002-9593-1697 Supplemental material: Supplemental material for this article is available online. ==== Refs References Abdulah DM Hassan AB (2020) Relation of dietary factors with infection and mortality rates of COVID-19 across the world. The Journal of Nutrition, Health & Aging 24 (9 ): 1011–1018. Açıkalın B Sanlier N (2021) Coffee and its effects on the immune system. Trends in Food Science & Technology 114 : 625–632. Akbar AN Gilroy DW (2020) Aging immunity may exacerbate COVID-19. Science (New York, N.Y.) 369 (6501 ): 256–257.32675364 Alam S Bhuiyan FR Emon TH , et al. (2021) Prospects of nutritional interventions in the care of COVID-19 patients. Heliyon 7 (2 ): e06285.33615017 Aman F Masood S (2020) How nutrition can help to fight against COVID-19 pandemic. Pakistan Journal of Medical Sciences 36 (COVID19-S4 ): S121.32582329 Andersen KG Rambaut A Lipkin WI , et al. (2020) The proximal origin of SARS-CoV-2. Nature Medicine 26 (4 ): 450–452. Ayivi R Ibrahim SA Colleran H , et al. (2021) COVID-19: Human immune response and the influence of food ingredients and active compounds. Bioactive Compounds in Health and Disease 4 (6 ): 100–148. Babaei F Nassiri-Asl M Hosseinzadeh H (2020) Curcumin (a constituent of turmeric): New treatment option against COVID-19. Food Science & Nutrition 8 (10 ): 5215–5227.33133525 Bain RES Wright JA Christenson E , et al. (2014) Rural:Urban inequalities in post 2015 targets and indicators for drinking-water. Science of The Total Environment 490 : 509–513.24875263 Batiha GE-S Alqarni M Awad DAB , et al. (2021) Dairy-Derived and egg white proteins in enhancing immune system against COVID-19. Frontiers in Nutrition 8 : 629440.34322507 Belaroussi Y Roblot P Peiffer-Smadja N , et al. (2020) Why methodology is important: Coffee as a candidate treatment for COVID-19? Journal of Clinical Medicine 9 (11 ): 3691.33213035 Bhattacharyya A Sa G Das T , et al. (2003) Black tea-induced cellular survival: Evidence for reduced toxicity and enhanced immunity in mice under stress. International Journal of Tea Science 2 (2 ). https://doi.org/10.20425/ijts.v2i1and2.4587. Bousquet J Anto JM Iaccarino G , et al. (2020) Is diet partly responsible for differences in COVID-19 death rates between and within countries? Clinical and Translational Allergy 10 (1 ): 16.32499909 Brugliera L Spina A Castellazzi P , et al. (2020) Nutritional management of COVID-19 patients in a rehabilitation unit. European Journal of Clinical Nutrition 74 (6 ): 860–863.32433599 Butler MJ Barrientos RM (2020) The impact of nutrition on COVID-19 susceptibility and long-term consequences. Brain, Behavior, and Immunity 87 : 53–54.32311498 Bwire GM (2020) Coronavirus: Why men are more vulnerable to COVID-19 than women? Sn Comprehensive Clinical Medicine 2 : 7. Callender LA Curran M Bates SM , et al. (2020) The impact of pre-existing comorbidities and therapeutic interventions on COVID-19. Frontiers in Immunology 11 : 1991.32903476 Carr AC Maggini S (2017) Vitamin C and Immune Function. Nutrients 9 (11 ): 1211. Castle SC Uyemura K Rafi A , et al. (2005) Comorbidity is a better predictor of impaired immunity than chronological age in older adults. Journal of the American Geriatrics Society 53 (9 ): 1565–1569.16137288 Chakraborty R Roy S (2018) Exploration of the diversity and associated health benefits of traditional pickles from the Himalayan and adjacent hilly regions of Indian subcontinent. Journal of Food Science and Technology 55 (5 ): 1599–1613.29666512 Chatterjee S Sarkar K (2021) Appraisal of urban–rural disparities in access to health care facilities and exposure to health risk factors: A case study of Durgapur Industrial region, India.2021: 1–18. Chen H Liu Y Zhu Z , et al. (2017) Does where you live matter to your health? Investigating factors that influence the self-rated health of urban and rural Chinese residents: Evidence drawn from Chinese general social survey data. Health and Quality of Life Outcomes 15 (1 ): 1–11.28069015 Chen S Liu T Li X , et al. (2021) Health QR code application in the novel containment strategy and healthcare plan for pregnant women and children under quarantine during the summer outbreak of SARS-CoV-2 delta variant in Chengdu, China: An observational study. Risk Management and Healthcare Policy 14 : 4499.34764708 Chowdhury P Barooah AK (2020) Tea bioactive modulate innate immunity: In perception to COVID-19 pandemic. Frontiers in Immunology 11 (1 ): 590716. Cobre AF Surek M Vilhena RO , et al. (2021) Influence of foods and nutrients on COVID-19 recovery: A multivariate analysis of data from 170 countries using a generalized linear model. Clinical Nutrition S0261-5614 (21 ): 00157–6. Colditz GA Willett WC Stampfer MJ , et al. (2010) Menopause and the risk of coronary heart disease in women. The New England Journal of Medicine 316 (18 ): 1105–1110. Costeira R Lee KA Murray B , et al. (2021) Estrogen and COVID-19 symptoms: Associations in women from the COVID symptom study. PLOS ONE 16 (9 ): e0257051.34506535 Davies NG Klepac P Liu Y , et al. (2020) Age-dependent effects in the transmission and control of COVID-19 epidemics. Nature Medicine 26 (8 ): 1205–1211. DGHS, MOHFW (2020) Novel Coronavirus (COVID-19) Guidelines. Available at: https://dghs.gov.bd/index.php/en/home/5376-novel-coronavirus-covid-19-guidelines (accessed 21 January 2022). Dodd RH Pickles K Nickel B , et al. (2021) Concerns and motivations about COVID-19 vaccination. The Lancet Infectious Diseases 21 (2 ): 161–163.33338440 Fan H Hong B Luo Y , et al. (2020) The effect of whey protein on viral infection and replication of SARS-CoV-2 and pangolin coronavirus in vitro. Signal Transduction and Targeted Therapy 5 (1 ): 1–3.32296011 Forni G Mantovani A Forni G , et al. (2021) COVID-19 vaccines: Where we stand and challenges ahead. Cell Death & Differentiation 28 (2 ): 626–639.33479399 Forouzanfar F Fazly Bazzaz BS Hosseinzadeh H (2014) Black cumin (Nigella sativa) and its constituent (thymoquinone): A review on antimicrobial effects. Iranian Journal of Basic Medical Sciences 17 (12 ): 929.25859296 Ganguli S Howlader S Dey K , et al. (2022) Association of comorbidities with the COVID-19 severity and hospitalization: A study among the recovered individuals in Bangladesh. International Journal of Health Sciences 16 (4 ): 30–38. Gasmi A Chirumbolo S Peana M , et al. (2021) The role of diet and supplementation of natural products in COVID-19 prevention. Biological Trace Element Research 200 : 27–30.33630276 Ghosh AK Kaiser M Molla MMA , et al. (2021) Molecular and serological characterization of the SARS-CoV-2 delta variant in Bangladesh in 2021. Viruses 13 (11 ): 2310.34835116 Girdhar A Kapur H Kumar V , et al. (2021) Effect of COVID-19 outbreak on urban health and environment. Air Quality, Atmosphere and Health 14 (3 ): 389–397. Grant WB Lahore H McDonnell SL , et al. (2020) Evidence that vitamin D supplementation could reduce risk of influenza and COVID-19 infections and deaths. Nutrients 12 (4 ): 988.32252338 Hamer M (2007) The beneficial effects of tea on immune function and inflammation: A review of evidence from in vitro, animal, and human research. Nutrition Research 27 (7 ): 373–379. Hu D Lou X Meng N , et al. (2021) Influence of age and gender on the epidemic of COVID-19: Evidence from 177 countries and territories—an exploratory, ecological study. Wiener Klinische Wochenschrift 133 : 321–330.33547492 Huang C Wang Y Li X , et al. (2020) Clinical features of patients infected with 2019 novel coronavirus in Wuhan. China. The Lancet 395 (10223 ): 497–506. Hull JH Loosemore M Schwellnus M (2020) Respiratory health in athletes: Facing the COVID-19 challenge. The Lancet Respiratory Medicine 8 (6 ): 557–558.32277869 Iddir M Brito A Dingeo G , et al. (2020) Strengthening the immune system and reducing inflammation and oxidative stress through diet and nutrition: Considerations during the COVID-19 crisis. Nutrients 12 (6 ): 1562.32471251 Ingram J Maciejewski G Hand CJ (2020) Changes in diet, sleep, and physical activity are associated with differences in negative mood during COVID-19 lockdown. Frontiers in Psychology 11 : 2328. Jafarzadeh A Jafarzadeh S Nemati M (2021) Therapeutic potential of ginger against COVID-19: Is there enough evidence? Journal of Traditional Chinese Medical Sciences 8 (4 ): 267–279. Jayawardena R Misra A (2020) Balanced diet is a major casualty in COVID-19. Diabetes & Metabolic Syndrome 14 (5 ): 1085–1086.32652495 Kennedy OJ Fallowfield JA Poole R , et al. (2021) All coffee types decrease the risk of adverse clinical outcomes in chronic liver disease: A UK biobank study. BMC Public Health 21 (1 ): 970.34154561 Khubber S Hashemifesharaki R Mohammadi M , et al. (2020) Garlic (allium sativum L.): A potential unique therapeutic food rich in organosulfur and flavonoid compounds to fight with COVID-19. Nutrition Journal 19 (1 ): 124.33208167 Kim GU Kim MJ Ra SH , et al. (2020) Clinical characteristics of asymptomatic and symptomatic patients with mild COVID-19. Clinical Microbiology and Infection 26 (7 ): 948.e1–948.e3. Kim H Rebholz CM Hegde S , et al. (2021) Plant-based diets, pescatarian diets and COVID-19 severity: A population-based case–control study in six countries. BMJ Nutrition, Prevention & Health 4 (1 ). doi: 10.1136/ bmjnph-2021-000272. Krause P Fleming Thomas R Longini I , et al. (2020) COVID-19 vaccine trials should seek worthwhile efficacy. The Lancet 396 (10253 ): 741–743. Kumbhar PS Pandya AK Manjappa AS , et al. (2021) Carbohydrates-based diagnosis, prophylaxis and treatment of infectious diseases: Special emphasis on COVID-19. Carbohydrate Polymer Technologies and Applications 2 : 100052. Kushwaha S Khanna P Rajagopal V , et al. (2021) Biological attributes of age and gender variations in Indian COVID-19 cases: A retrospective data analysis. Clinical Epidemiology and Global Health 11 : 100788.34079918 Lai CC Chen IT Chao CM , et al. (2021) COVID-19 vaccines: Concerns beyond protective efficacy and safety. Expert Review of Vaccines 20 (8 ): 1013–1025.34180347 Lee JA Park JH Kim M (2015) Social and physical environments and self-rated health in urban and rural communities in Korea. International Journal of Environmental Research and Public Health 12 (11 ): 14329–14341.26569279 Li Y Tenchov R Smoot J , et al. (2021) A comprehensive review of the global efforts on COVID-19 vaccine development. ACS Central Science 7 (4 ): 512–533.34056083 Lidoriki I Frountzas M Schizas D (2020) Could nutritional and functional status serve as prognostic factors for COVID-19 in the elderly? Medical Hypotheses 144 : 109946.32512494 Liu G Zhang S Mao Z , et al. (2020) Clinical significance of nutritional risk screening for older adult patients with COVID-19. European Journal of Clinical Nutrition 74 (6 ): 876–883.32404899 Maiorino MI Bellastella G Longo M , et al. (2020) Mediterranean Diet and COVID-19: Hypothesizing potential benefits in people with diabetes. Frontiers in Endocrinology 11 : 574315.33042027 Martinez-Ferran M de la Guía-Galipienso F Sanchis-Gomar F , et al. (2020) Metabolic impacts of confinement during the COVID-19 pandemic due to modified diet and physical activity habits. Nutrients 12 (6 ): 1549.32466598 Martins-Filho PR Tavares CSS Santos VS (2020) Factors associated with mortality in patients with COVID-19. A quantitative evidence synthesis of clinical and laboratory data. European Journal of Internal Medicine 76 : 97–99.32345526 Mashhadi NS Ghiasvand R Askari G , et al. (2013) Anti-oxidative and anti-inflammatory effects of ginger in health and physical activity: Review of current evidence. International Journal of Preventive Medicine 4 (Suppl 1 ): S36.23717767 Merino J Joshi AD Nguyen LH , et al. (2021) Diet quality and risk and severity of COVID-19: A prospective cohort study. Gut 70 (11 ): 2096–2104.34489306 Mohamed OI El-Nahas AF El-Sayed YS , et al. (2015) Ginger extract modulates Pb-induced hepatic oxidative stress and expression of antioxidant gene transcripts in rat liver. Pharmaceutical Biology 54 (7 ): 1164–1172.26079851 Mohsin FM Tonmon TT Nahrin R , et al. (2021) Association between smoking and COVID-19 severity: Evidence from Bangladesh. Journal of Multidisciplinary Healthcare 14 : 1923–1933.34326643 Moona AA Daria S Asaduzzaman M , et al. (2021) Bangladesh reported delta variant of coronavirus among its citizen: Actionable items to tackle the potential massive third wave. Infection prevention in practice 3 (3 ): 100159. Moscatelli F Sessa F Valenzano A , et al. (2021) COVID-19: Role of nutrition and supplementation. Nutrients 13 (3 ): 976.33803015 Peckham H de Gruijter NM Raine C , et al. (2020) Male sex identified by global COVID-19 meta-analysis as a risk factor for death and ITU admission. Nature Communications 11 (1 ): 6317. Peter KV (2006) Handbook of Herbs and Spices, 1st ed. Swaston: Woodhead Publishing. Rangchaikul P Venketaraman V (2021) SARS-CoV-2 and the immune response in pregnancy with delta variant considerations. Infectious Disease Reports 13 (4 ): 993–1008.34940401 Richardson S Hirsch JS Narasimhan M , et al. (2020) Presenting characteristics, comorbidities, and outcomes among 5700 patients hospitalized with COVID-19 in the New York city area. Journal of American Medical Association 323 (20 ): 2052–2059. Riva M Curtis S Gauvin L , et al. (2009) Unravelling the extent of inequalities in health across urban and rural areas: Evidence from a national sample in England. Social Science &Medicine 68 (4 ): 654–663.19108940 Salazar-Robles E Kalantar-Zadeh K Badillo H , et al. (2021) Association between severity of COVID-19 symptoms and habitual food intake in adult outpatients. BMJ Nutrition 4 : e000348. Sharif MK Rebia E Imran P (2018) Nutritional and therapeutic potential of spices. In: Therapeutic, Probiotic, and Unconventional Foods. Cambridge, MA: Academic Press, pp. 181–199. Shi Y Wang Y Shao C , et al. (2020) COVID-19 infection: The perspectives on immune responses. Cell Death & Differentiation 27 (5 ): 1451–1454.32205856 Singh P Tripathi MK Yasir M , et al. (2020) Potential inhibitors for SARS-CoV-2 and functional food components as nutritional supplement for COVID-19: A review. Plant Foods for Human Nutrition 75 (4 ): 458–446. Tavakol Z Ghannadi S Tabesh MR , et al. (2021) Relationship between physical activity, healthy lifestyle and COVID-19 disease severity; a cross-sectional study. Journal of Public Health: From Theory to Practice. 10.1007/s10389-020-01468-9. TBS (2022) Bangladesh braces for third wave as Omicron spreads. Available at https://www.tbsnews.net/coronavirus-chronicle/covid-19-bangladesh/bangladesh-braces-third-wave-omicron-spreads-354526 (accessed 3 February 2022). Turkmen N Velioglu YS Sari F , et al. (2007) Effect of extraction conditions on measured total polyphenol contents and antioxidant and antibacterial activities of black tea. Molecules 12 (3 ): 484–496.17851405 Whittemore PB (2020) COVID-19 fatalities, latitude, sunlight, and vitamin D. American Journal of Infection Control 48 (9 ): 1042–1044.32599103 WHO (2021) Coronavirus Disease (COVID-19) Situation Reports. Available at: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports (accessed 11 December 2021). Wierzejska R (2017) Can coffee consumption lower the risk of Alzheimer’s disease and Parkinson’s disease? A literature review. Archives of Medical Science: AMS 13 (3 ): 507–514.28507563 Zbinden-Foncea H Francaux M Deldicque L , et al. (2020) Does high cardiorespiratory fitness confer some protection against proinflammatory responses after infection by SARS-CoV-2? Obesity 28 (8 ): 1378–1381.32324968	36066026	PMC9716059	NO-CC CODE	2022-12-03 23:20:53	no	Nutr Health. 2022 Dec; 28(4):771-782	utf-8	Nutr Health	2,022	10.1177/02601060221124068	oa_other
==== Front Nutr Health Nutr Health NAH spnah Nutrition and Health 0260-1060 2047-945X SAGE Publications Sage UK: London, England 34041988 10.1177/02601060211019676 10.1177_02601060211019676 Special Section on ‘COVID 19’ Original Articles Diet diversity of urban households in India during the COVID-19 lockdown https://orcid.org/0000-0001-9154-6188 Aneesh Mitravinda 1 Patil Rita S 2 1 Department of Nutrition and Dietetics, Mount Carmel College Autonomous, India 2 Department of Food and Nutrition, Maniben Nanavati Women’s College, India Mitravinda Aneesh, Department of Nutrition and Dietetics, Mount Carmel College Autonomous, 58 Palace Road, Bengaluru 560052, India. Email: [email protected] 12 2022 12 2022 12 2022 28 4 685691 © The Author(s) 2021 2021 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background: The nature of the COVID-19 pandemic forced several nations to impose country-wide lockdowns. The lockdown impacted several aspects of life including the economy. Food security became a growing concern for many households. Aim: The aim of the study was to explore the diet diversity of urban households in India during the nationwide COVID-19 lockdown. Methods: Information regarding socioeconomic status (SES), family size and information regarding availability and access to food were gathered from 450 households. Diet diversity was assessed using a 69-item food frequency questionnaire. Food variety scores (FVS) were computed for individual food groups and overall. Results: The majority of the households (86.4%) belonged to the upper-middle or upper SES. Households did not experience any constraints in accessibility and availability of food except the meat group. Overall, 84% of the households had low FVS for most of the food groups except for sugar and milk and milk products. The household SES score was positively associated with the milk FVS (B = 0.039, p = 0.020) and negatively with the fat FVS (B = −0.062, p < 0.001). The number of adults (B = 6.773, p < 0.001) in the household positively predicted the FVS of cereal, vegetable, fruit, fat and total FVS. Conclusions: The higher SES households in urban India did not experience food insecurity. Despite this, their poor diet diversity is a serious cause for concern, especially in the wake of the evolving pandemic. This highlights the need to promote consumption of a diverse variety of foods. Food Variety Score diet diversity lockdown COVID-19 pandemic typesetterts19 ==== Body pmcIntroduction In January 2020, the outbreak of the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) was confirmed (World Health Organization, 2019). Soon after, several countries reported cases of SARS-CoV-2. In India, the first case was reported on 30 January 2020. After the initial few cases, a surge in reported cases began in early March 2020. The Government of India took cognizance of the situation and imposed a nationwide lockdown from 25 March 2020, which was extended until 31 May 2020. After this, the government brought in a few zone-wise relaxations while the lockdown continued in the containment areas. The lockdown period witnessed a number of changes. COVID-19 not only emerged as a public health crisis but also impacted the different economic sectors. Travel and tourism, hospitality, manufacturing and transportation were adversely affected (Saraswathy, 2020). The micro, small and medium enterprises were affected the most (Rakshit and Basistha, 2020). Migrant labourers were hit the worst, with no wages, no food and no means to travel back to their home towns. On the other hand, the workforce from the service sector such as IT firms moved to a work-from-home mode of working. To bear the economic crisis, many large companies started to lay off their employees (Ray, 2020). With the economic crisis, food security became a cause for concern. The lockdown impacted the agri-food supply chain (ETGovernment, 2020). During this period, only essential services such as groceries, pharmacy and emergency medical services were available. Consumers were anxious about the availability, quality and prices of the fresh foods and food products. This led to some consumers purchasing huge stocks of foods. On the other hand, the government also made efforts to step up the public distribution system to ensure food availability to the poorer sections of the population (Pathak et al., 2020). Studies in Kenya, Uganda, Bangladesh and Iran reported a rise in food insecurity during the lockdown period (Kansiime et al., 2021; Kundu et al., 2021; Pakravan-Charvadeh, et al., 2021a, 2021b). In India too, with the rising cases and the extended lockdown, the household food purchasing power, food choices and, therefore, the diet diversity seem to have been affected (Headey and Ruel, 2020). Dietary diversity is an effective tool to assess the diet quality in an emergency situation where resources are limited. At the household level, diet diversity is a reflection of the economic ability to access food (Kennedy et al., 2011). Household food variety scores (FVS) are an extension of household diet diversity which give a glimpse of the variety of foods that the household has access to (Badari et al., 2012). In view of the above-mentioned concerns, we undertook the present study with an aim to assess the diet diversity of urban Indian households during the lockdown period. The specific research questions that we explored were (a) did the lockdown affect the food access and availability?; (b) was food availability associated with diet diversity?; and (c) did the socioeconomic status (SES) of the household predict the diet diversity? Methods Study design: It was a cross-sectional study conducted in the urban areas of India during April–June 2020, when the lockdown was in place in most parts of the country. Sample selection and size: Four hundred and fifty households across Indian cities such as Mumbai, Pune, Bengaluru, Hyderabad, Chennai, Delhi and Kolkata participated in the study via convenience sampling technique. The households were selected based on the inclusion and exclusion criteria. Inclusion criteria: households in the jurisdiction of urban areas/cities were included. Quarantined households were eligible to participate in the study post their quarantine period. During the quarantine period, food access could have been affected due to restrictions in movement. Therefore, such households were eligible to participate post their quarantine. Households with at least one person having access to the internet (via smartphones or other devices) and who was able to read and write English language were eligible to participate in the study. Individuals or families residing in hostels, community-living centres, institutional homes and migrant camps were excluded from the study. Data collection: During the lockdown and with measures of social distancing in place, collecting data on a one-to-one basis was not feasible. Thus, the electronic forms were created on Google. These were electronically shared with the eligible participants via email and social media (Facebook, WhatsApp and LinkedIn). The participants were informed about the objectives of the study, the nature of information required, confidentiality of the data and the average time required to complete the survey. The privacy of the data was ensured by access being only with the authors. The information sheet was shared with the participants and electronic consent was obtained from them. Measurements Diet diversity – Diet diversity of the households was assessed using FVS. FVS is a reflection of the diet quality and the diversity of the diet (Choi and Bae, 2014; Evans et al., 2018; Sonandi et al., 2017). We assessed the FVS of the household using a 69-item food frequency questionnaire. FVS was defined as the number of different food items eaten in a week during the lockdown period. The FVS of the households was calculated for commonly eaten foods listed in nine groups: (a) cereals and millets; (b) pulses and legumes; (c) nuts; (d) milk and milk products; (e) meat; (f) vegetables; (g) fruits; (h) sugar; and (i) fats. The scoring was done as follows – daily intake was given a score of 7, three to six times a week was score 4, once a week was 1, and less than once a week was 0 (Badari et al., 2012). The FVS was calculated for each food group, which helped us to understand the number and types of food eaten in that group. The FVS of individual food groups was summed up to arrive at the total FVS. In order to classify the FVS as ‘low’ or ‘high’, we calculated the minimum and maximum scores for each food group and the overall score. The score above mid-point was classified as ‘high FVS’ and those below as ‘low FVS’. SES – The Updated Modified Kuppuswamy SES for the year 2019 was used to assess the SES of the households (Mohd Saleem, 2019). The scale included information about the education and occupation of the head of the household along with the monthly household income (in INR). The total scores of the scale ranged from 3 to 29. Based on this, the households were classified into five SES groups – ‘low’ (< 5), ‘upper-lower’ (5–10), ‘lower-middle’ (11–15), ‘upper-middle’ (16–25) and ‘upper’ (26–29) SES. Other information regarding household size along with number of adults, senior citizens, adolescents and children was collected. We also included COVID-related information – if the household was located in a containment zone, food availability (if any food or food group was not available during lockdown), accessibility (how far did they have to travel to purchase their food supply), if they purchased food online and if any of the foods had become expensive during the lockdown. Statistical analysis: The data were analysed using IBM SPSS 20. Mean ± SD and frequency distribution were carried out for the FVS and sociodemographic characteristics respectively. Binary logistic regression was carried out to find out the variables that influence individual and total FVS. For all the individual and total FVS, the scores below median and above median were considered as dependent variables. The independent variables considered were number of adults, adolescents in the household and SES score. In the regression model for meat variety score, we included the availability of meat (0 = yes, 1 = no) and controlled for the food preferences of the households – vegetarian and non-vegetarian. Further, in the regression models for vegetable and fruit groups, we tested if the rise in the cost (0 = no change, 1 = expensive) predicted their FVS. Results Four hundred and fifty households participated in this study. The size of the household varied between a minimum of one to a maximum of 32 members. The mean size of the household was 4.06 ± 2.20 members. The mean number of adults in the household was 3.24 ± 1.56 (1–11). Nearly 56% of the households had at least one senior citizen residing with them. One-quarter of the households had at least one adolescent (mean: 0.34 ± 0.65). Almost 22% of the households had at least one child below 10 years of age (mean: 0.27 ± 0.57). SES of the households The majority of the households (53.1%, n = 239) belonged to the upper SES class. One-third of the households (33.3%, n = 150) were from the upper-middle SES. The remaining 13.6% (n = 61) of households belonged to the lower-middle and upper-lower SES. In over 80% of households, the head of the family was either a graduate (41.3%, n = 186) or had a professional degree (41.3%, n = 186). The rest of the head of households (18.4%, n = 78) had completed either middle school or high school or held a diploma. Over half the head of households were employed as professionals (56%, n = 252), such as engineers, chartered accountants, teachers and so on. Over 16% (n = 74) of the head of households held senior managerial positions or were legislators. Nearly 9% (n = 40) of the heads were unemployed. The rest of the heads (19%, n = 85) were involved in technical or clerical or sales-related jobs. Nearly half the households (47.1%, n = 212) had a monthly income of over INR 78,062 and 23% of the households (n = 105) earned a monthly income between INR 39,033 and 78,062. Over 16% of the households (n = 75) earned between INR 19,516 and 39,032 per month. The remaining households (12.9%, n = 58) received an income of less than INR 19,515 per month. Food purchases during lockdown Over two-thirds of the households (n = 314) preferred vegetarian foods while the rest (30.2%, n = 136) chose non-vegetarian foods. During the lockdown, nearly half of the households (n = 220) were able to purchase food items within 500 metres of their residence. One-quarter of the households (n = 116) purchased food items within a distance of 500–1000 metres from their homes. About 3% of families (n = 12) purchased only online. The remaining households travelled more than a kilometre to buy food items. About 9% of the families (n = 39) preferred purchasing online and physical buying. Over half the respondents (n = 240) felt that the food items had become expensive during the lockdown while 17% (n = 76) did not feel so. The remaining 30% (n = 134) did not know about the prices of the food items. About one-third of the respondents (n = 152) expressed that the cost of vegetables and fruits had risen during the lockdown. This was followed by non-vegetarian foods (7.6%, n = 34), cereals and pulses (7.3%, n = 33) and packaged and online ordered foods (4.4%, n = 20). About 19% (n = 85) of respondents felt that all the food items had become expensive. The majority of the participants (93%, n = 418) expressed that almost all foods were available. Of the participants who consumed flesh foods, 23.5% (n = 32) mentioned that sea foods and other flesh foods were not available. FVS The mean ± SD of the FVS is presented in Table 1. Overall, 84% of the households reported low total FVS. The majority of the households were seen to have low FVS for almost all the food groups except for milk and milk products and sugar. Table 1. Mean ± SD and percentage of households with low and high FVS. Food group Mean (min–max) SD Low FVS % (n) High FVS % (n) Cereals 21.48 (2–54) 7.7 87.7 (395) 12.3 (55) Pulses 15.68 (0–66) 10.8 96.0 (432) 4.0 (18) Nuts 7.29 (0–21) 5.6 73.6 (331) 26.4 (119) Milk and products 12.99 (0–21) 3.4 21.4 (96) 78.6 (354) Meat 4.19 (0–23) 5.0 98.0 (441) 2.0 (9) Roots and tubers 14.08 (0–35) 6.0 76.7 (345) 23.3 (105) Green leafy vegetables 6.18 (0–35) 6.0 94.2 (424) 5.8 (26) Other vegetables 21.70 (0–70) 10.2 88.9 (400) 21.1 (50) Total vegetables 41.97 (4–140) 18.1 93.6 (421) 6.4 (29) Fruits 20.1 (0–84) 13.3 92.2 (415) 7.8 (35) Sugars 8.75 (7–14) 2.0 - 100 (450) Fats 10.96 (7–21) 3.3 53.6 (241) 46.4 (209) Total FVS 185.36 (57–483) 58.0 84 (378) 16.0 (72) Factors associated with FVS We carried out binary logistic regression to find out the predictors of FVS. We did this individually for all the food groups and for the total FVS (Tables 2 and 3). The number of adults in the household were seen to positively predict the FVS of cereals, roots and tubers, other vegetables, overall vegetables, fruits, fat and total FVS (p < 0.05). Table 2. Binary logistic regression of cereal, pulses, milk and meat FVS with family members and SES variables. Cereal FVS Pulses FVS Variables B (SE) OR p B (SE) OR p Number of adults 0.150 (0.064) 1.161 0.020 0.999 (0.063) 1.104 0.117 Number of adolescents 0.261 (0.151) 1.298 0.084 0.228 (0.151) 1.334 0.057 SES score 0.020 (0.016) 1.020 0.217 −0.002 (0.016) 0.998 0.998 Milk FVS Meat FVSa Variables B (SE) OR p B (SE) OR p Number of adults 0.044 (0.063) 1.045 0.483 0.047 (0.076) 1.048 0.534 Number of adolescents 0.027 (0.150) 1.316 0.067 0.303 (0.182) 1.354 0.095 SES score 0.039 (0.017) 1.040 0.020 −0.010 (0.020) 0.990 0.616 Availability of meat (1)b - - - −1.663 (0.650) 0.190 0.011 B: unstandardized regression co-efficient; SE: standard error; OR: odds ratio. a Controlled for preference for non-vegetarian foods. b Availability of meat (0 = yes, 1 = no). Table 3. Binary logistic regression of vegetables, fruits, sugar, fat and total FVS with family members and SES variables. Roots and tubers Green leafy vegetables Other vegetables Overall vegetable FVS Variables B (SE) OR p B (SE) OR p B (SE) OR p B (SE) OR p Number of adults 0.233 (0.067) 1.262 <0.001 0.116 (0.064) 1.123 0.071 0.145 (0.064) 1.156 0.024 0.235 (0.067) 1.265 <0.001 Number of adolescents 0.109 (0.150) 1.115 0.469 0.272 (0.152) 1.313 0.073 −0.102 (0.149) 0.903 0.493 0.034 (0.150) 1.034 0.823 SES score 0.010 (0.016) 1.010 0.559 −0.027 (0.016) 0.973 0.095 −0.004 (0.016) 0.996 0.817 0.015 (0.016) 1.015 0.376 Cost (1)# 0.176 (0.206) 1.192 0.393 −0.037 (0.206) 0.964 0.859 0.057 (0.203) 1.058 0.780 −0.253 (0.205) 0.777 0.218 Fruits FVS Sugar FVS Fat FVS Total FVS Variables B (SE) OR p B (SE) OR p B (SE) OR p B (SE) OR p Number of adults 0.185 (0.065) 1.203 0.004 0.062 (0.067) 1.064 0.357 0.157 (0.066) 1.169 0.018 0.357 (0.073) 1.429 <0.001 Number of adolescents 0.109 (0.150) 1.115 0.465 0.076 (0.158) 1.079 0.629 0.087 (0.154) 1.091 0.573 0.104 (0.154) 1.109 0.501 SES score 0.012 (0.016) 1.013 0.449 −0.022 (0.017) 0.979 0.210 −0.062 (0.017) 0.940 <0.001 0.002 (0.017) 0.970 0.900 Cost (1)a 0.157 (0.206) 0.999 0.780 - - - - - - - - - B: unstandardized regression co-efficient; SE: standard error; OR: odds ratio. a Cost (0 = no change, 1 = expensive). Further, the socioeconomic variables also predicted the FVS of certain food groups. A higher household SES score was noted to predict lower FVS for fat (B = −0.062, p < 0.001). On the other hand, a higher household SES score positively predicted the FVS of milk and milk products (B = 0.039, p = 0.020). The non-availability of meat foods negatively predicted the meat FVS (p = 0.011). As all the respondents reported 100% availability of other food groups, the variability was constant. As a result, the constant parameter was removed from the analysis. Discussion We carried out the present study with an aim to understand the household diet diversity among urban Indian households during the COVID-19 lockdown. The findings suggest that access and availability of the food items remained largely unaffected in the urban areas during the lockdown. However, households reported poor diet diversity. This was seen despite the fact that the majority belonged to upper-middle and upper SES. Impact of lockdown on food access and food availability The study participants did not report any challenges in accessibility and availability of food items except sea foods and flesh foods. Cariappa et al. (2020) also reported uninterrupted access to markets in India. It is noteworthy that the present study was restricted only to the urban areas. Rural India, on the other hand, may present a completely varied scenario. A sizeable proportion of our participants expressed an increase in the cost of certain food items, particularly vegetables and fruits. In congruence with this, Tata-Cornell Institute (2020) reported that, during the months of March to May 2020, the prices of non-cereal foods such as vegetables and fruits, pulses and eggs increased. Further, a general upward trend was noted in the retail food prices as suggested by the consumer food price index (Cariappa et al., 2020). Diet diversity during the lockdown Majority of the households had poor diet diversity. Most of the households consumed not more than one to two staple cereals and pulses. Among cereals and millets, almost all the households reported consumption of rice and wheat. About 30 to 40% of the households consumed millets such as pearl millet, sorghum and finger millet at least once a week. Like millets, legumes seemed to be consumed less than pulses. Among the pulses, red gram dal (94%) and green gram dal (89.6%) emerged as the most commonly consumed by the households. About 32–57% of the households reported consumption of legumes such as whole lentils, rajmah (kidney beans), cow pea, moth beans, horsegram and field beans at least once a week. Households seemed to prefer processed cereal alternatives such as bread (77.1%) and breakfast cereals (62.2%) rather than millets and legumes. Over 90% of the households reported daily intake of milk, and two-thirds of households consumed curd daily. Further, nearly 64% of the households consumed paneer or cottage cheese at least once a week. In India, tea with a small quantity of milk is a popular beverage which is consumed in almost every household irrespective of the SES. In addition to this, curd is an inseparable part of the meal in many parts of the country. This may explain the higher FVS for milk. The FVS of vegetables and fruits was particularly low. The most commonly consumed vegetables on a daily basis were tomatoes, onions and potatoes. Many of the Indian curries and gravies have onion and tomato as the base. Potato is an integral part of cooking in several Indian households across the SES spectrum. The remaining vegetables were consumed by 60–70% of the households. However, their frequency of consumption was not more than once a week. Likewise, the majority of the fruits were consumed in 75–80% of the households at least once a week. Our findings of poor diet diversity resonate with those of previous studies. Gaiha et al. (2012) studied the diet quality of India between 1993 to 2009. They documented a drop in the urban consumption of cereals and pulses. Along with this, they observed a rise in the urban intake of milk and its products as well as fat. Kulkarni and Gokhale (2014) also noted that 90% of the participants had the lowest mean diversity for vegetables and fruits in higher socioeconomic areas of Mumbai. Likewise, Popkin et al. (2009) reviewed Indian diets and noted a rise in the availability and consumption of sugar. A recent survey among college students across India showed that about three-quarters of the participants had a poor dietary diversity score (Kumar et al., 2020). Association of food availability and diet diversity Our respondents reported 100% availability of all the food items except meat foods. Our analysis suggested that the non-availability of the meat food group resulted in significantly lower FVS. During the early days of the pandemic, there was immense uncertainty looming over the source of the coronavirus. Several reports on social media were suggestive that the virus did spread through animal markets and foods. Such reports adversely hit the availability and the sales of meat and its products. As a result, the local meat shops were closed in the study period (March to May) and opened in late June. Although nearly one-third of our participants reported an increase in the price of fruits and vegetables, this did not seem to impact the diet diversity of these foods. As mentioned earlier, Kulkarni and Gokhale (2014) reported poor diversity of vegetables and fruits among urban Indians belonging to higher economic groups. This is much in line with our study. Association of SES and diet diversity The diversity of the diets consumed in households has been closely associated with the SES of the families. Studies carried out across the globe suggest that higher income and literacy levels were associated with greater FVS (Badari et al., 2012; Keding et al., 2012; Sonandi et al., 2017). Contrary to these findings, our current study indicated an inverse relationship between SES score and FVS of fat (B = −0.062, p < 0.001). In the present study, over 85% of the households were from the upper and upper-middle SES. The majority of head of households were graduates or had professional degrees. The higher socioeconomic groups are considered to be more health conscious. Obesity and other comorbid conditions have been known to increase the risk of complications of COVID-19. This could have been a reason for the lower FVS for fat consumption in the upper and upper-middle SES households. Further, a positive correlation was noted between the FVS for milk and milk products and SES. In India, milk is considered to be a wholesome food that promotes growth during childhood and adolescence. During the pandemic, milk with turmeric was being promoted as an immunity-building home remedy. Besides this, turmeric milk and curd have been integral immune-boosting foods in a traditional Indian diet. Curd, being a probiotic, plays a vital role in promoting healthy gut microbiota. In a study conducted in Tehran, Iran during the lockdown, personal savings, occupation status of the head of the family and number of educated members positively influenced the diet diversity (Pakravan-Charvadeh et al., 2021a, 2021b). These factors could probably explain the relation observed. Association of number of adults and diet diversity Regression analysis showed that the FVS of cereals, vegetables, fruits, fat and overall food groups was strongly associated with the number of adults in the family. Adults are one of the main consumers of food in the family. Thus, the more adults there are, the greater could be the variety of food consumed in households belonging to upper SES. Studies conducted in Kenya, Uganda, Bangladesh and Iran during the lockdown reported an increase in food insecurity (Kansiime et al., 2021; Kundu et al., 2021; Pakravan-Charvadeh et al., 2021a, 2021b). Like India, all the mentioned nations are classified as developing economies. Unlike these studies, our study did not highlight food insecurity as a concern in urban India. The findings of the present study should be considered in the light of a few concerns. Firstly, owing to the restrictions imposed during the pandemic, we used convenience sampling. Secondly, we had to use electronic forms for collecting data. These forms were circulated via social media platforms. This restricted our participants to the upper-middle and upper SES. We were unable to include vulnerable groups such as below poverty line families, migrant workers and daily wage workers, who were impacted the most in our sample. As our participants were from the higher SES in urban areas, they did not experience any form of food insecurity. Despite this, it is alarming to find such poor diet diversity among the higher economic urban participants. Conclusions To the best of our knowledge, this is one of the earliest studies from India regarding household diet quality during the pandemic. Though the upper SES households of urban India did not face challenges related to food insecurity, their diets during the national lockdown were less diverse. Studies conducted prior to the pandemic suggest the same. The lockdown did not seem to impact the access and availability of essential food items except meat. However, poor diversity of the key food groups such as pulses, vegetables and fruits can adversely influence the nutritional status. Consumption of healthy diverse diets needs to be promoted more than ever before to enhance our nutritional status and immunity. Acknowledgement We acknowledge all the respondents for willingly participating in our study. Authorship: Mitravinda Aneesh and Rita S Patil contributed to the study conception, design, material preparation and data collection. Statistical analysis was performed by Mitravinda Aneesh. The manuscript was prepared by both the authors. Both the authors read and approved the final manuscript. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Ethical statement: This study was approved by the Institutional Ethics Research Committee of Maniben Nanavati Women’s College, Mumbai, India (ERC/2020/Apr/01). Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ORCID iD: Mitravinda Aneesh https://orcid.org/0000-0001-9154-6188 ==== Refs References Badari SAZ Arcot J Haron SA , et al. (2012) Food variety and dietary diversity scores to understand the food-intake pattern among selected Malaysian households. Ecology of Food and Nutrition 51 (4 ): 265–299.22794127 Cariappa AGA Acharya KK Adhav C , et al. (2020) Pandemic led food price anomalies and supply chain disruption: Evidence from COVID-19 incidence in India . ID 3680634, SSRN Scholarly Paper, 27 August. Rochester, NY: Social Science Research Network. Choi M-K Bae Y-J (2014) Evaluation of nutrient intake and food variety in Korean male adults according to Framingham Risk Score. The Korean Journal of Food And Nutrition 27 (3 ): 484–494. ETGovernment (2020) Opinion: Impact of Covid-19 pandemic on food security of India. ETGovernment.com, 19 July. Available at: https://government.economictimes.indiatimes.com/news/policy/opinon-impact-of-the-covid-19-pandemic-on-food-security-of-india/77048453 (accessed 20 October 2020). Evans CEL Hutchinson J Christian MS , et al. (2018) Measures of low food variety and poor dietary quality in a cross-sectional study of London school children. European Journal of Clinical Nutrition 72 (11 ): 1497–1505.29391590 Gaiha R Kaicker N Imai K , et al. (2012) Dietary shift and diet quality in India: An analysis based on 50th, 61st and 66th rounds of NSS. ASARC Working Paper 2012/17. Available at: https://acde.crawford.anu.edu.au/publication/asarc-working-papers/8961/dietary-shift-and-diet-quality-india-analysis-based-50th-61st Headey D Ruel M (2020) The COVID-19 nutrition crisis: What to expect and how to protect. In: International Food Policy Research Institute Blog. Available at: https://www.ifpri.org/blog/covid-19-nutrition-crisis-what-expect-and-how-protect (accessed 1 October 2020). Kansiime MK Tambo JA Mugambi I , et al. (2021) COVID-19 implications on household income and food security in Kenya and Uganda: Findings from a rapid assessment. World Development 137: 105199 . Keding GB Msuya JM Maass BL , et al. (2012) Relating dietary diversity and food variety scores to vegetable production and socio-economic status of women in rural Tanzania. Food Security 4 (1 ): 129–140. Kennedy G Ballard T Dop MC , et al. (2011) Guidelines for measuring household and individual dietary diversity. Report, Food and Agriculture Organization of the United Nations, Rome. Kulkarni N Gokhale C (2014) Vegetable and fruit diversity in an Indian community. International Journal of Food, Nutrition and Public Health 7 (1 ): 13–24. Kumar A Ayub A Roy R , et al. (2020) Assessment of diet diversity and eating pattern of undergraduate students: A pan India study. International Journal of Medicine and Public Health 10 (1 ): 46–51. Kundu S Banna MHA Sayeed A , et al. (2021) Determinants of household food security and dietary diversity during the COVID-19 pandemic in Bangladesh. Public Health Nutrition 24 (5 ): 1079–1087.33317657 Mohd Saleem S (2019) Modified Kuppuswamy socioeconomic scale updated for the year 2019. Indian Journal of Forensic and Community Medicine 6 (1 ): 1–3. Pakravan-Charvadeh MR Mohammadi-Nasrabadi F Gholamrezai S , et al. (2021a) The short-term effects of COVID-19 outbreak on dietary diversity and food security status of Iranian households (A case study in Tehran province). Journal of Cleaner Production 281: 124537 . Pakravan-Charvadeh MR Savari M Khan HA , et al. (2021b) Determinants of household vulnerability to food insecurity during COVID-19 lockdown in a mid-term period in Iran. Public Health Nutrition 24 (7 ): 1–10.32662385 Pathak P Gope T Bader N (2020) Effect of COVID-19 on public distribution system in India. International Journal of Community Medicine and Public Health 7 (6 ): 2411–2415. Popkin BM Horton S Kim S , et al. (2009) Trends in diet, nutritional status, and diet-related noncommunicable diseases in China and India: The economic costs of the nutrition transition. Nutrition Reviews 59 (12 ): 379–390. Rakshit B Basistha D (2020) Can India stay immune enough to combat COVID-19 pandemic? An economic query. Journal of Public Affairs 20 (4 ): 1–7. Ray A (2020) Lockdown impact: 5 million salaried Indians lost their jobs in July, show data. Mint, 18 August. Available at: https://www.livemint.com/news/india/covid-19-five-million-salaried-people-lost-their-jobs-in-july-shows-cmie-data-11597752797552.html (accessed 19 September 2020). Saraswathy (2020) 10.8 million and counting: Take a look at how many jobs Covid-19 has wiped out. Moneycontrol, 15 August. Available at: https://www.moneycontrol.com/news/business/economy/10-8-million-and-counting-take-a-look-at-how-many-jobs-covid-19-has-wiped-out-5704851.html (accessed 13 April 2021). Sonandi A Zwane EF Niekerk JAV (2017) Feeding practices, food variety, and dietary diversity – indicators of nutritional status among historically disadvantaged agri-business families, South Africa. Journal of Nutrition and Food Security 2 (4 ): 308–317. Tata-Cornell Institute (2020) Pandemic prices: COVID-19 price shocks and their implications for nutrition security in India. 16 July. Tata-Cornell Institute. Available at: https://tci.cornell.edu/news/publications/pandemic-prices-covid-19-price-shocks-and-their-implications-for-nutrition-security-in-india/ (accessed 19 April 2021). World Health Organization (2019) Archived: WHO Timeline – COVID-19. Available at: https://www.who.int/news/item/27-04-2020-who-timeline---covid-19 (accessed 24 October 2020).	34041988	PMC9716060	NO-CC CODE	2022-12-03 23:20:53	no	Nutr Health. 2022 Dec; 28(4):685-691	utf-8	Nutr Health	2,022	10.1177/02601060211019676	oa_other
==== Front Nutr Health Nutr Health NAH spnah Nutrition and Health 0260-1060 2047-945X SAGE Publications Sage UK: London, England 36441515 10.1177/02601060221137176 10.1177_02601060221137176 Special Section on ‘COVID 19’ Editorial Editorial: Changes in eating habits, nutrition trends, and lifestyle-associated factors in the wake of the COVID-19 pandemic https://orcid.org/0000-0002-2337-6612 Asghar Waqas 1 https://orcid.org/0000-0002-8045-199X Khalid Nauman 1 1 School of Food and Agricultural Sciences, University of Management and Technology, Lahore Pakistan Nauman Khalid, School of Food and Agricultural Sciences, University of Management and Technology, Lahore 54000 Pakistan. Email: [email protected] 12 2022 12 2022 12 2022 28 4 633634 © The Author(s) 2022 2022 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. typesetterts19 ==== Body pmcWe are living in unprecedented times. The COVID-19 pandemic, from its onset in late 2019, can be regarded as a transformative period in terms of changes it brought about in the lives of people worldwide, in particular, in the context of food, nutrition, and health. Widespread disruptions of food systems, instigated by logistical restrictions amid fears of spread of infection led to impairment of supply chains, and potential shortages of food (Galanakis, 2020). Moreover, partial, or complete lockdown procedures led to closures of establishments such as restaurants and hotels, schools, universities, and workplaces, as well as travel restrictions, changed the way the people gained access to food, as well as how, and where the food was prepared (Janssen et al., 2021). Additionally, many COVID-19-associated psychological factors could have significantly influenced dietary behaviors, in particular, in people staying at home owing to a variety of reasons including work-from-home, digital education, and even self-isolation (Bennett et al., 2021). Frequent and extensive communication related to risk factors associated with COVID-19 might also have induced emotional stress, resulting in overconsumption, especially of the so-called “comfort foods,” for instance, chocolate, snacks, and chips, essentially processed foods rich in sugar and calories, a phenomenon that can be termed as “food craving,” or “emotional eating” (Salazar-Fernández et al., 2021). The metabolic impact of these high glycemic index foods, in turn, presents additional health risks, such as obesity and cardiovascular diseases beyond the chronic state of inflammation, increasing the risk for more severe complications associated with COVID-19 (Di Renzo et al., 2020). This is particularly significant in the context that obesity-associated conditions have the potential to impair immune responses, predisposing severely obese individuals to higher risks of COVID-19-associated complications. These complications potentially further hampered the ability of individuals to maintain a healthier lifestyle, involving healthy, varied diets, coupled with adequate physical activity levels. Conversely, negative experiences stemming from confinement at home may also culminate in restricted eating habits, which can be attributed to physiological stress reactions mimicking the internal sensations linked with feeding-induced satiety (Yang et al. 2022). Finally, home confinement significantly increased the likelihood of sedentary behaviors, such as low physical activity levels (associated with both body fat and diet dysregulation), smoking, and sleep disturbances, further exacerbating anxiety and stress-induced behaviors (Dietz & Santos-Burgoa 2020). Given that the restrictions related to COVID-19 have eased considerably, it is time to get our dietary behaviors back on track. We urge our readers to observe healthy eating habits and make appropriate adjustments to lifestyles, ensuring prolonged health benefits and longevity. It is also high time to take lessons from this pandemic and make our food systems much more resilient, effective, and inclusive with increased access to more susceptible and vulnerable demographics of our population, preventing a repeat of the “dominos effect” these disrupted food systems created in the context of health and disease. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article ORCID iD: Waqas Asghar https://orcid.org/0000-0002-2337-6612 Nauman Khalid https://orcid.org/0000-0002-8045-199X ==== Refs References Bennett G. Young E. Butler I. , et al. (2021) The impact of lockdown during the COVID-19 outbreak on dietary habits in various population groups: A scoping review. Frontiers in Nutrition 8 , Article number: 626432. 10.3389/fnut.2021.626432 Dietz W. Santos-Burgoa C. (2020) Obesity and its implications for COVID-19 mortality. Obesity 28 (6 ): 1005–1005.32237206 Di Renzo L. Gualtieri P. Pivari F. , et al . (2020) Eating habits and lifestyle changes during COVID-19 lockdown: An Italian survey. Journal of Translational Medicine 18 , Article number: 229. 10.1186/s12967-020-02399-5 Galanakis C . (2020) The food systems in the era of the coronavirus (COVID-19) pandemic crisis. Foods (Basel, Switzerland) 9 (4 ), 23. Janssen M. Chang B. Hristov H ., et al. (2021) Changes in food consumption during the COVID-19 pandemic: Analysis of consumer survey data from the first lockdown period in Denmark, Germany, and Slovenia. Frontiers in Nutrition 8, Article number: 635859. 10.3389/fnut.2021.635859 Salazar-Fernández C. Palet D. Haeger P. A. , et al. (2021) The perceived impact of COVID-19 on comfort food consumption over time: The mediational role of emotional distress. Nutrients 13 (6 ), 1910.34199403 Yang C. C. Chen Y. S. Chen J . (2022) The impact of the COVID-19 pandemic on food consumption behavior: Based on the perspective of accounting data of Chinese food enterprises and economic theory. Nutrients 14 (6 ), 1206.35334868	36441515	PMC9716061	NO-CC CODE	2022-12-03 23:20:53	no	Nutr Health. 2022 Dec; 28(4):633-634	utf-8	Nutr Health	2,022	10.1177/02601060221137176	oa_other
==== Front Neurosci Lett Neurosci Lett Neuroscience Letters 0304-3940 1872-7972 Elsevier B.V. S0304-3940(21)00451-1 10.1016/j.neulet.2021.136073 136073 Introduction Introduction to the Special Issue on New Developments in Undergraduate Neuroscience Education Pollack Alexia E. Department of Biology, University of Massachusetts Boston, Boston, MA 02125, United States Parfitt Karen D. Department of Neuroscience, Pomona College, Claremont, CA 91711, United States 18 6 2021 10 8 2021 18 6 2021 759 136073136073 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcWhen we began planning this Special Issue on Undergraduate Neuroscience Education in November 2019, none of us could have anticipated the dramatic shift to online teaching and learning in March 2020 due to the Covid-19 pandemic. Before this happened, faculty were focused on the growing numbers of undergraduate students studying neuroscience, and how to accommodate these burgeoning interests. At the same time, curricular changes were needed in order to prepare students for the 21st century. Consistent with the recommendations of the American Association for the Advancement of Science’s Vision and Change in Undergraduate Biology Education document [1], as well as the National Research Council’s BIO2010 document [2], most neuroscience faculty were moving away from content-driven courses and toward inquiry-driven pedagogy. The mini-reviews contained in this Special Issue provide examples of these shifts in undergraduate neuroscience education. We hope that you are inspired by these mini-reviews to adapt new approaches in your own classrooms in order to train the next generation of neuroscientists and community leaders. In the first mini-review, Ramirez, the founding president of the Faculty for Undergraduate Neuroscience (FUN), reflects on the state of undergraduate science education in the Covid-19 era and reviews undergraduate neuroscience education initiatives since the 1990s, including the curricular “Blueprints.” He closes with suggestions for ways in which colleges and universities can nurture the success of undergraduate neuroscience curricula. A theme throughout this Special Issue is the shared goal of enhancing diversity, equity and inclusion. In their mini-review, Basu and colleagues at College of the Holy Cross emphasize the strength of providing an integrative neuroscience core curriculum that involves faculty collaboration. Promotion of broad STEM proficiency in their students, particularly through their introductory neuroscience course, has been a successful strategy in promoting equity of access and inclusion of all students, regardless of their backgrounds. Discipline-based education research has demonstrated that active learning enhances the success of all students, particularly those who are underrepresented in STEM. In their mini-review, Williams and O’Dowd describe seven strategies to incorporate active learning, and they share evidence that these activities enhance learning outcomes in classrooms both large and small. In their mini-review, Penner and colleagues are similarly focused on increasing access of minoritized students to neuroscience coursework through innovative approaches. They provide advice on how to develop two different kinds of courses: makerspace and course-based undergraduate research experiences (CUREs), which can be integrated into the introductory as well as upper-level curricula. They discuss the welcoming nature of such courses, as well as their potential for attracting and retaining diverse students. A key skill that undergraduate students should develop is the ability to read, evaluate and understand how primary literature fits into the field at-large. In their mini-review, Pugh-Bernard and Kenyon describe the use of the CREATE method (Consider Read, Elucidate the hypotheses, Analyze and interpret the data, and Think of the next Experiment) to enable students to think like scientists, whether they are learning at the introductory level or in upper-division courses. In another strategy, the mini-review by Khan and colleagues at the University of Texas-El Paso takes a historical perspective and compares their CURE, focused on Brain Mapping and Connectomics, to the neuroanatomy teaching approaches used more than one hundred years ago at Johns Hopkins University Medical School. They highlight the success of these research-intensive approaches in enhancing the retention of diverse students. Success in 21st century neuroscience also depends on developing strong written communication and quantitative skills. To that end, in their mini-review Petersen and colleagues at Kenyon College outline effective strategies to teach undergraduate students writing within their neuroscience courses. In his commentary, Hoy stresses the importance of strong quantitative skills for neuroscience majors, and how students (and faculty) can develop these critical skills through a variety of means, many of which take advantage of a multitude of online resources. Of course, much of neuroscience education occurs in the laboratory, but due to the Covid-19 pandemic, lab work came to a complete halt for many of us. You may have noticed the limited mention of laboratory-based neuroscience teaching included in the Special Issue. With the onset of the pandemic, potential contributors to this issue were consumed with pivoting from in-person to virtual labs and were unable to write about the lab activities that their students have found so rewarding. As with other changes in “classroom” teaching, virtual online labs will likely give rise to new means for developing critical thinking and experimental design skills that go beyond the need for physical laboratory spaces. With more than a year of online or hybrid instruction during the Covid-19 pandemic, our teaching has changed in more ways than could ever be encapsulated in a single Special Issue. We have learned and adopted new technologies, incorporated student feedback, and increased our efforts to make materials more inclusive. When we return to face-to-face teaching, we will keep the best of these approaches and gladly discard the rest. However, what seems abundantly clear is that the days of the 50-minute lecture are behind us. We hope that the mini-reviews in this Special Issue provide enriching alternatives. ==== Refs References 1 American Association for the Advancement of Science, Vision and change in undergraduate biology education: A call to action. http://visionandchange.org/finalreport/, 2011 (accessed March 9, 2021). 2 National Research Council BIO2010: Transforming Undergraduate Education for Future Research Biologists 2003 National Academies Press Washington, D.C. https://doi.org/10.17226/10497	34147537	PMC9716120	NO-CC CODE	2022-12-03 23:20:54	no	Neurosci Lett. 2021 Aug 10; 759:136073	utf-8	Neurosci Lett	2,021	10.1016/j.neulet.2021.136073	oa_other
==== Front Neurosci Lett Neurosci Lett Neuroscience Letters 0304-3940 1872-7972 Elsevier B.V. S0304-3940(21)00451-1 10.1016/j.neulet.2021.136073 136073 Introduction Introduction to the Special Issue on New Developments in Undergraduate Neuroscience Education Pollack Alexia E. Department of Biology, University of Massachusetts Boston, Boston, MA 02125, United States Parfitt Karen D. Department of Neuroscience, Pomona College, Claremont, CA 91711, United States 18 6 2021 10 8 2021 18 6 2021 759 136073136073 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcWhen we began planning this Special Issue on Undergraduate Neuroscience Education in November 2019, none of us could have anticipated the dramatic shift to online teaching and learning in March 2020 due to the Covid-19 pandemic. Before this happened, faculty were focused on the growing numbers of undergraduate students studying neuroscience, and how to accommodate these burgeoning interests. At the same time, curricular changes were needed in order to prepare students for the 21st century. Consistent with the recommendations of the American Association for the Advancement of Science’s Vision and Change in Undergraduate Biology Education document [1], as well as the National Research Council’s BIO2010 document [2], most neuroscience faculty were moving away from content-driven courses and toward inquiry-driven pedagogy. The mini-reviews contained in this Special Issue provide examples of these shifts in undergraduate neuroscience education. We hope that you are inspired by these mini-reviews to adapt new approaches in your own classrooms in order to train the next generation of neuroscientists and community leaders. In the first mini-review, Ramirez, the founding president of the Faculty for Undergraduate Neuroscience (FUN), reflects on the state of undergraduate science education in the Covid-19 era and reviews undergraduate neuroscience education initiatives since the 1990s, including the curricular “Blueprints.” He closes with suggestions for ways in which colleges and universities can nurture the success of undergraduate neuroscience curricula. A theme throughout this Special Issue is the shared goal of enhancing diversity, equity and inclusion. In their mini-review, Basu and colleagues at College of the Holy Cross emphasize the strength of providing an integrative neuroscience core curriculum that involves faculty collaboration. Promotion of broad STEM proficiency in their students, particularly through their introductory neuroscience course, has been a successful strategy in promoting equity of access and inclusion of all students, regardless of their backgrounds. Discipline-based education research has demonstrated that active learning enhances the success of all students, particularly those who are underrepresented in STEM. In their mini-review, Williams and O’Dowd describe seven strategies to incorporate active learning, and they share evidence that these activities enhance learning outcomes in classrooms both large and small. In their mini-review, Penner and colleagues are similarly focused on increasing access of minoritized students to neuroscience coursework through innovative approaches. They provide advice on how to develop two different kinds of courses: makerspace and course-based undergraduate research experiences (CUREs), which can be integrated into the introductory as well as upper-level curricula. They discuss the welcoming nature of such courses, as well as their potential for attracting and retaining diverse students. A key skill that undergraduate students should develop is the ability to read, evaluate and understand how primary literature fits into the field at-large. In their mini-review, Pugh-Bernard and Kenyon describe the use of the CREATE method (Consider Read, Elucidate the hypotheses, Analyze and interpret the data, and Think of the next Experiment) to enable students to think like scientists, whether they are learning at the introductory level or in upper-division courses. In another strategy, the mini-review by Khan and colleagues at the University of Texas-El Paso takes a historical perspective and compares their CURE, focused on Brain Mapping and Connectomics, to the neuroanatomy teaching approaches used more than one hundred years ago at Johns Hopkins University Medical School. They highlight the success of these research-intensive approaches in enhancing the retention of diverse students. Success in 21st century neuroscience also depends on developing strong written communication and quantitative skills. To that end, in their mini-review Petersen and colleagues at Kenyon College outline effective strategies to teach undergraduate students writing within their neuroscience courses. In his commentary, Hoy stresses the importance of strong quantitative skills for neuroscience majors, and how students (and faculty) can develop these critical skills through a variety of means, many of which take advantage of a multitude of online resources. Of course, much of neuroscience education occurs in the laboratory, but due to the Covid-19 pandemic, lab work came to a complete halt for many of us. You may have noticed the limited mention of laboratory-based neuroscience teaching included in the Special Issue. With the onset of the pandemic, potential contributors to this issue were consumed with pivoting from in-person to virtual labs and were unable to write about the lab activities that their students have found so rewarding. As with other changes in “classroom” teaching, virtual online labs will likely give rise to new means for developing critical thinking and experimental design skills that go beyond the need for physical laboratory spaces. With more than a year of online or hybrid instruction during the Covid-19 pandemic, our teaching has changed in more ways than could ever be encapsulated in a single Special Issue. We have learned and adopted new technologies, incorporated student feedback, and increased our efforts to make materials more inclusive. When we return to face-to-face teaching, we will keep the best of these approaches and gladly discard the rest. However, what seems abundantly clear is that the days of the 50-minute lecture are behind us. We hope that the mini-reviews in this Special Issue provide enriching alternatives. ==== Refs References 1 American Association for the Advancement of Science, Vision and change in undergraduate biology education: A call to action. http://visionandchange.org/finalreport/, 2011 (accessed March 9, 2021). 2 National Research Council BIO2010: Transforming Undergraduate Education for Future Research Biologists 2003 National Academies Press Washington, D.C. https://doi.org/10.17226/10497	36459354	PMC9716122	NO-CC CODE	2022-12-03 23:20:54	no	Sci China Life Sci. 2022 Nov 29;:1-4	latin-1	Sci China Life Sci	2,022	10.1007/s11427-022-2192-5	oa_other
==== Front Curr Psychol Curr Psychol Current Psychology (New Brunswick, N.j.) 1046-1310 1936-4733 Springer US New York 4080 10.1007/s12144-022-04080-0 Article Seeing the big picture during the COVID-19 pandemic: the spillover effects of visionary leadership on employees’ work-to-family conflict http://orcid.org/0000-0002-7807-446X Wang Haibo Zhang Huiying http://orcid.org/0000-0002-4620-053X Xie Jun [email protected] Zheng Jia grid.440718.e 0000 0001 2301 6433 School of Business, Guangdong University of Foreign Studies, No. 178 Waihuan Dong Road, Panyu District, Guangzhou, China 2 12 2022 112 24 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. While an expanding body of research has revealed the beneficial impacts of visionary leadership on employees’ work-related outcomes, little is known about its spillover effects on nonwork domains. Drawing upon work–home resources theory, we investigated the impacts of visionary leadership on employees’ work-to-family conflict (WFC). Utilizing three-wave data from 268 employees, the results indicate that visionary leadership promotes follower relational energy, which in turn reduces WFC. Furthermore, perceived COVID-19 crisis disruption was found to strengthen the negative indirect link between visionary leadership and employees’ WFC. Our research broadens our understanding of the potential positive spillover effects of visionary leadership in the nonwork domain through relational energy, and the accentuating effect of perceived crisis disruption on the work–family interface. The theoretical and managerial implications of these findings are discussed. Keywords Visionary leadership Relational energy Work-to-family conflict Perceived COVID-19 crisis disruption Work-home resources model http://dx.doi.org/10.13039/501100001809 National Natural Science Foundation of China 72102051 Wang Haibo The Project of Guangdong Provincial Humanities and Social Sciences Key Research Base18JD07 Xie Jun ==== Body pmcIntroduction In today’s competitive and uncertain socioeconomic environment, there is a pressing need for visionary leadership to galvanize collective action (Carton & Lucas, 2018). In particular, visionary leadership can play a critical role in helping employees navigate the uncertainty associated with crises like the COVID-19 pandemic (Stam et al., 2018). Visionary leadership, which refers to a leader’s communication of an idealized future to motivate followers to strive for its realization (Stam et al., 2014), is considered one of the most influential leadership styles (Kearney et al., 2019). Examples of visionary leadership include clearly stating where the collective is going and communicating frequently about what the future should look like (Carton et al., 2014; Kearney et al., 2019). The wide-ranging beneficial effects of this approach highlight the importance of visionary leadership research. Previous studies have demonstrated that visionary leadership improves followers’ organizational commitment, job satisfaction, positive emotions, intention to stay, and job performance (e.g., Kearney et al., 2019; Kohles et al., 2012; Mayfield & Mayfield, 2007). Despite extensive evidence of the impacts of visionary leadership in the work domain, the potential spillover effects on employees’ nonwork domains (e.g., life and family) have been largely ignored in the literature. Such an omission is unfortunate, as this style of leadership not only impacts the work of followers but also has far-reaching implications for other parts of their lives (Carton & Lucas, 2018). Today, as the COVID-19 pandemic has led to more people working from home, employees experience a substantial integration of work and family demands (Fisher et al., 2020; Kerman et al., 2021), meaning that the beneficial effects of leaders’ visionary communication can more readily spill over into the home domain and reduce work-to-family conflict (WFC), defined as employees’ perception of the extent to which role pressures arising from work domains lead to inter-role conflict and clashes in family domains (Allen et al., 2000; Greenhaus & Beutell, 1985). The main reason for such omissions may lie in the limitations of the theoretical perspectives used in prior research. To date, previous studies have largely relied on either the motivational perspective (e.g., Eseryel et al., 2021) or the self-conception perspective (e.g., Stam et al., 2014) to explain the beneficial effects of visionary leadership. The motivational perspective argues that visionary communication stimulates and aligns employee work motivation, leading to improved job-related outcomes (e.g., Berson et al., 2015; Maran et al., 2022; Van Knippenberg & Sitkin, 2013), while the self-conception perspective posits that visionary communication improves employees’ work performance by arousing their ideal self (e.g., Jim et al., 2001; Stam et al., 2010). These two research perspectives have clear limitations in their scope of analysis. The motivational perspective focuses on job motivation reinforced by visionary leadership, while the self-conception perspective emphasizes the role that visionary leadership may play in establishing the ideal self of followers. Their analytical logic means that both can only be applied within the scope of the work domain and cannot account for the cross-domain effects of visionary leadership. In the present study, we draw upon the work–home resources (W-HR) model (Ten Brummelhuis & Bakker, 2012) to examine whether and how visionary leadership has an impact on employees’ family domain outcomes. The W-HR model, building on Hobfoll’s (1989) conservation of resources (COR) theory, posits that work–home enrichment is a process through which resources accumulated in one domain lead to an increase in personal resources that facilitate better performance in another domain (Ten Brummelhuis & Bakker, 2012). Based on this logic, we argue that visionary leadership can act as a kind of contextual resource for employees, as it involves proactive vision communication initiated by supervisors and stimulates cognitive and emotional resources in employees. These personal resources, generated through interaction with visionary leaders, can be captured by relational energy, defined as “a heightened level of psychological resourcefulness generated from interpersonal interactions that enhances one’s capacity to do work” (Owens et al., 2016, p. 37). This, in turn, could reduce followers’ WFC, which is a construct reflecting the extent to which one’s work roles interfere with family roles. Drawing on the same theoretical lens, we further introduce COVID-19 crisis disruption as critical conditional factor in the association between visionary leadership and follower WFC. In adverse situations, individuals generally have stronger motivation to seek resources from their external environment (Stam et al., 2018). In this sense, employees with a stronger perception of COVID-19 crisis disruption, which is defined as employees’ perceptions of the extent to which COVID-19 has affected their capacity to work (Morgeson & Frederick, 2005), are more likely to benefit from visionary leadership, as they are most in need of the inspiring future communicated by their leaders (Jones & Comfort, 2020). Therefore, we propose that perceived COVID-19 crisis disruption can enhance the effects of visionary leadership on followers’ WFC through their relational energy. Our theoretical model is depicted in Fig. 1. Fig. 1 Research model The present study makes contributions to the existing literature in three key ways. First, by drawing on a resource perspective to explore the cross-domain effects of visionary leadership on followers’ WFC, we not only extend the literature by shifting the focus from the work domain to the nonwork domain but also provide a new perspective that compensates for the limitations of existing motivation and self-conception perspectives. Second, our study reveals a new mechanism linking visionary leadership and employee WFC by conceptualizing relational energy as a kind of personal resource generated from interpersonal interactions with leaders. Third, by investigating the moderating role of perceived COVID-19 crisis disruption, we identify a critical boundary condition for the cross-domain effects of visionary leadership. Hypothesis development Relationship between visionary leadership and follower relational energy According to the W-HR model, contextual work resources in the work and home domains can help individuals develop personal resources (Ten Brummelhuis & Bakker, 2012). Following this logic, we propose that visionary leadership can act as a kind of contextual resource that helps increase followers’ relational energy in cognitive and emotional ways. By communicating an image of the future, whether it is a desirable end-state to reach or an undesirable situation to avoid (Stam et al., 2010), visionary leaders form a shared understanding of future situations and highlight how followers can be part of the bigger picture (Kearney et al., 2019). Such communication helps followers predict and evaluate the potential future, and cognitively motivates them to strive for the vision (Stam et al., 2014). Thus, interactions with visionary leaders can engender cognitive resources that help employees develop a clearer understanding of their work goals and complete their work more efficiently, which can be represented by relational energy. Visions communicated by visionary leaders are typically ambitious, defying conventional wisdom, challenging existing norms and policies, conveying expectations of strong performance, and instilling confidence in followers that they can achieve the vision (Stam et al., 2010). Such rhetoric can effectively invoke a mental image of this vision among followers, making it feel realistic and attractive (Carton & Lucas, 2018). By encouraging followers to maintain a compelling vision of the organization and set challenging goals, visionary leadership can arouse their enthusiasm and provide them with a sense of meaning and purpose (Kearney et al., 2019). These enhanced emotional resources can act as an intrinsically motivating force that encourages individuals to perform better in their work roles (Owens et al., 2016). Employees with frequent interactions with visionary supervisors can attain high levels of relational energy with respect to their jobs. Thus, we propose the following: Hypothesis 1: Visionary leadership is positively related to followers’ relational energy. Relationship between relational energy and WFC According to the W-HR model, the personal resources developed in one domain (e.g., work) can subsequently promote performance in another (e.g., home; Ten Brummelhuis & Bakker, 2012). Based on this logic, we explain how relational energy stemming from interactions with visionary leaders can help employees reduce WFC. WFC can be divided into three forms: time-based conflict, strain-based conflict, and behavior-based conflict (Greenhaus & Beutell, 1985). Time-based WFC occurs when the time invested in a work role makes it difficult for employees to take on family responsibilities. Strain-based WFC takes place when the strain stemming from the work domain spills over into the family domain. Behavior-based WFC occurs when employees cannot modify their behavioral patterns at work to meet the needs of family life. The relational energy gained from interactions with visionary leaders allows employees to function optimally and manage work demands more effectively (Fan et al., 2021). This means that they can conserve more time and energy for family life, thereby alleviating the conflict between work and family. Followers with low relational energy are less likely to manage work demands effectively and may become preoccupied with work-related matters (Mao et al., 2022). They tend to spend more time and effort on work and have less time and energy for family life, leading to WFC. Moreover, individuals high in relational energy feel enthusiastic and inspired (Yang et al., 2019), and are more likely to take their positive mood home, thus reducing strain-based WFC. Individuals with high relational energy have a greater willingness and ability to show care and warmth to family members (Kwan et al., 2022). Energized individuals have ample resources that allow them to adjust their behavior to meet family expectations, which can help reduce WFC. Accordingly, we propose the following: Hypothesis 2: Followers’ relational energy is negatively related to their WFC. Indirect effects of visionary leadership on follower WFC This theoretical logic indicates that visionary leadership has an indirect effect on follower WFC via follower relational energy. Visionary leaders can provide followers with meaning and goals in their work roles by communicating an attractive future (Venus et al., 2019). Employees who regularly interact with visionary leaders obtain higher levels of relational energy at work, which enables them to better manage their work roles. As a result, they can spend more time and energy fulfilling family obligations, which helps reduce WFC. Therefore, we propose the following: Hypothesis 3: There is an indirect negative relationship between visionary leadership and followers’ WFC through followers’ relational energy. Moderating effect of perceived COVID-19 crisis disruption The W-HR model further suggests that the macro context serves as an important boundary condition for the work–home process. Following this logic, we investigate how COVID-19 crisis disruption, conceptualized as the extent to which individuals perceive that COVID-19 has affected their capacity to complete work tasks (Morgeson & Frederick, 2005), can serve as a crucial boundary condition for understanding the function of visionary leadership. Visionary leadership, acting as a kind of contextual work resource, can facilitate the development of employees’ relational energy because, through interactions with visionary leaders, employees can acquire cognitive resources (e.g., understanding of the bigger picture and work goals) and emotional resources (e.g., enthusiasm and sense of meaning) that help them complete their work efficiently. However, people may differ in their need to seek contextual resources in different situations. In adverse situations, rather than favorable ones, people have stronger motivation to seek resources from their external environment, thus benefiting more from available contextual resources (Stam et al., 2018). In the current work context, perception of COVID-19 crisis disruption can be an indicator of such exogenous shocks. Disruptive events such as the COVID-19 pandemic can diminish certainty about present and future situations (Paredes et al., 2021), as its impacts are uncontrollable and its outcome unpredictable (Liu et al., 2020). The turbulence and uncertainty caused by COVID-19 can create unprecedented confusion about the future of organizations and individuals (Paredes et al., 2021), and have a profound psychological impact, magnifying anticipatory fear and emotional distress. During the pandemic, the cognitive and emotional problems associated with the crisis create a greater need for insightful guidance and encouragement from supervisors. Thus, employees with a stronger perception of COVID-19 crisis disruption are more likely to benefit from visionary leadership, as they are most in need of the clear future communicated by their leaders and the enthusiasm conveyed in the process (Jones & Comfort, 2020). By strategically increasing their interactions with visionary leaders, employees with strong perceptions of COVID-19 crisis disruption can generate psychological resourcefulness and improve their ability to accomplish work tasks. On the contrary, as employees with low levels of perceived crisis disruption are less affected by the changes and uncertainty arising from external shocks, they may be less motivated to gain energy through interaction with visionary leaders. Thus, they are less likely to benefit from visionary leadership that focuses on hope and belief in a promising future. Accordingly, we propose the following: Hypothesis 4: Perceived COVID-19 crisis disruption moderates the positive association between visionary leadership and followers’ relational energy, such that the association is stronger when employees’ perceived COVID-19 crisis disruption is higher. The aforementioned hypotheses suggest an integrated framework in which follower relational energy serves as a mediator in the relationship between visionary leadership and follower WFC, and perceived COVID-19 crisis disruption moderates the link between visionary leadership and followers’ relational energy. Thus, we expect that perceived COVID-19 crisis disruption also moderates the mediating effects of relational energy in the link between visionary leadership and WFC (Edwards & Lambert, 2007). As we predict a stronger association between visionary leadership and relational energy among individuals with high perceived COVID-19 crisis disruption, the negative indirect impact of visionary leadership on follower WFC should be stronger among employees with high perceived COVID-19 crisis disruption. Thus, we propose the following: Hypothesis 5: Perceived COVID-19 moderates the negative and indirect effects of visionary leadership on followers’ WFC, such that the indirect effect is stronger when followers’ perceived COVID-19 crisis is higher. Methods Sample and procedures To test the hypotheses proposed in our research, we collected data from five service-oriented companies located in Guangzhou, China. After obtaining the business owners’ approval, we assured the participants that their responses would remain anonymous and would be used for academic purposes only. With the assistance of the human resources manager, we coded volunteers by the last four digits of their phone numbers. The surveys were conducted online and completed during work time. To reduce the interference of common method bias on the results of our study (Doty & Glick, 1998), the data were collected in three waves over a one-month period. Data collection started on 24 February 2021 (first wave), continued on 9 March 2021 (second wave), and ended on 23 March 2021 (third wave). During this period, parts of Guangzhou were under lockdown due to intermittent outbreaks of COVID-19, and some respondents may have needed to work from home during the investigation. In the first wave, we sent questionnaires to 362 randomly selected participants, who were asked to answer questions related to their demographic information, control variables, visionary leadership, and perceived COVID-19 crisis disruption. We received responses from 331 employees (91.4% response rate). Two weeks later, these 331 participants were asked to complete the second questionnaire, which contained a measure of relational energy. We received responses from 296 participants (89.4% response rate). Two weeks later, a third questionnaire was sent to these 296 participants, asking them to rate their WFC. A total of 271 responses were received (91.6% response rate). Some of these responses were excluded due to incomplete answers, and our final sample consisted of 268 respondents. Of these, 45.6% were male, and most were aged under 40 (39.8% were 21–30 years of age, and 42.3% were 31–40). Their average job tenure was 2.5 years, and they had worked with their current leader for an average of 2.2 years. The educational backgrounds of the participants were relatively evenly distributed, with approximately one-third of the participants each having completed high school or below (34.8%), junior college (28.7%), and a bachelor’s degree or above (33.0%). Regarding family status, 56.0% were married, and 45.5% had underage children. Measures Apart from the demographics, all key variables were measured using five-point response categories (1 = strongly disagree to 5 = strongly agree). The questionnaires used were translated from English to Chinese, following the procedures suggested by Brislin (1970), and all surveys were presented in Chinese. Visionary leadership A five-item measure developed by Kearney et al. (2019) was utilized to measure leader vision communication. A sample item is “My direct supervisor communicates a clear idea about what should be accomplished” (α = 0.91). Perceived COVID-19 crisis disruption We used the five-item measure developed by Morgeson (2005) to assess perceived COVID-19 crisis disruption. A sample item is “To what extent does COVID-19 disrupt your ability to get work done?” (α = 0.91). Relational energy We measured followers’ relational energy using Owens et al.’s (2016) five-item measure. A sample item is “I feel invigorated when I interact with my direct supervisor” (α = 0.90). Work-to-family conflict We assessed WFC using a nine-item measure developed by Carlson et al. (2000). A sample item is “The stress from my job often makes me irritable when I get home” (α = 0.92). Control variables To eliminate the potential impacts of demographic variables on the work–family interface (e.g., Kwan et al., 2022), we controlled for participants’ gender (1 = male and 2 = female), age (years), education background, job tenure (years), time spent working with direct manager (years), marital status (1 = married, 2 = single), raising a child under the age of 18 (1 = yes, 2 = no), role as a manager (1 = yes, 2 = no), and industry. Analytic strategy Prior to testing the hypotheses proposed in this study, we first conducted confirmatory factor analysis (CFA) to assess the distinctiveness of the four variables included in the current study and then use two commonly applied methods to evaluate potential common method bias. Next, using Mplus 8.0 (Muthén & Muthén, 2017), we performed structural equation modeling (SEM) to test our hypotheses in two steps (Cole et al., 2008; Mueller & Hancock, 2010). We first tested the mediation model that included visionary leadership, relational energy, WFC, and control variables, the results of which are shown in Models 1 and 3 in Table 3. In the second moderated mediation model, perceived COVID-19 crisis disruption was also included, and the results of which are illustrated in Model 2 in Table 3. Confidence intervals (CIs) were then generated by bootstrapping with 2000 iterations and used to estimate the strength of both indirect and conditional indirect effects. Results Measurement model testing First, we ran a series of CFAs to assess the discriminant validity of the four variables (visionary leadership, relational energy, perceived COVID-19 crisis disruption, and WFC). We assessed the measurement model with all possible combinations of the four variables using commonly accepted cutoff values (Marsh et al., 2004). The results shown in Table 1 suggest that the hypothesized four-factor model was a good fit for the data (χ2/df = 2.46, TLI = 0.91, CFI = 0.92, RMSEA = 0.07, SRMR = 0.06). This four-factor model provided a better fit than the alternative three-factor models, two-factor models, and one-factor model. Thus, the CFAs’ results confirmed the discriminant validity among key constructs. Table 1 Comparison of measurement models Model Descriptions χ2 df χ2 /df RMSEA SRMR CFI TLI Model 1 Four factors: VL, RE, PCD, WFC 604.33 246 2.46 0.07 0.06 0.92 0.91 Model 2 Three factors: VL + RE, PCD, WFC 1432.99 249 5.75 0.13 0.10 0.74 0.71 Model 3 Three factors: VL + PCD, RE, WFC 1372.72 249 5.51 0.13 0.14 0.75 0.72 Model 4 Three factors: VL + WFC, RE, PCD 1946.76 249 7.82 0.16 0.18 0.62 0.58 Model 5 Three factors: RE + PCD, VL, WFC 1683.00 249 6.76 0.15 0.17 0.68 0.65 Model 6 Three factors: RE + WFC, VL, PCD 1900.52 249 7.63 0.16 0.18 0.63 0.59 Model 7 Three factors: PCD + WFC, VL, RE 1321.90 249 5.31 0.13 0.13 0.76 0.74 Model 8 Two factors: VL + RE, PCD + WFC 2148.41 251 8.56 0.17 0.15 0.58 0.54 Model 9 Two factors: VL + PCD, RE + WFC 2642.27 251 10.53 0.19 0.21 0.47 42 Model 10 Two factors: VL + WFC, PCD + RE 3016.11 251 12.02 0.20 0.24 0.39 0.33 Model 11 One factor: VL + RE + PCD + WFC 3356.57 252 13.32 0.21 0.22 0.31 0.25 n = 268. VL = visionary leadership, RE = followers’ relational energy, PCD = perceived COVID-19 crisis disruption, WFC = work-to-family conflict Second, we used two commonly used methods to evaluate possible common method bias. We first conducted Harman’s single-factor test using SPSS 22.0. The results of the exploratory factor analysis (EFA) showed that there were four factors captured 27.83%, 20.92%, 11.44%, and 9.05% of the variance in the data, respectively. None of them exceeded the threshold of 40% suggested by Podsakoff et al. (2003). In addition, a CFA was performed using Mplus 8.0, with every item being allowed to load on its respective theoretical construct and a latent variable (common method factor). The results indicated that the variance explained by the common method factor was 12%, which is below the 25% average reported in previous studies (Perry et al., 2010; Williams et al., 1989). Hence, common method bias was not a serious problem in our study. Descriptive statistics Descriptive statistics are shown in Table 2, including the Pearson correlations, means, standard deviations, and Cronbach’s alpha for the variables. As expected, visionary leadership was positively correlated with relational energy (r = 0.45, p < 0.01), while relational energy was negatively correlated with WFC (r = -0.16, p < 0.01). Table 2 Descriptive statistics and correlation matrix Variablesa M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Genderb 1.54 0.50 2.Age 2.72 0.73 0.01 3.Education 2.01 0.88 0.09 − 0.07 4.Marriage 1.42 0.49 − 0.04 − 0.51 0.00 5.Children 1.53 0.50 0.05 − 0.43 0.04 0.72 6.Tenure 2.52 1.25 − 0.01 0.44 0.18 − 0.39 − 0.31 7.Time with supervisor 2.20 1.03 − 0.08 0.39 − 0.03 − 0.30 − 0.20 0.70 8.Manager 1.76 0.43 0.09 − 0.25 − 0.19 0.20 0.18 − 0.32 − 0.32 9.Industry 2.66 1.45 − 0.27 − 0.08 0.15* 0.06 0.07 0.12 − 0.01 − 0.17** 10.VL 3.90 0.84 − 0.07 0.04 − 0.07 − 0.05 − 0.06 0.00 0.06 − 0.15* 0.00 (0.91) 11.RE 3.68 0.83 − 0.14* − 0.04 − 0.02 0.02 0.10 − 0.03 0.06 − 0.03 − 0.07 0.42** (0.90) 12.PCD 3.13 1.36 − 0.02 − 0.01 0.06 0.10 − 0.01 − 0.08 − 0.10 − 0.04 0.10 0.11 0.15* (0.91) 13.WFC 2.79 0.74 − 0.10 − 0.07 − 0.06 0.11 0.09 − 0.06 − 0.05 0.05 0.08 − 0.16 − 0.20 0.22 (0.92) n = 268. Coefficient alphas are given in parentheses on the diagonal a VL = visionary leadership, PCD = perceived COVID-19 crisis disruption, RE = followers’ relational energy, WFC = work-family conflict b Males were coded as 1, and females were coded as 2 p < 0.01; p < 0.05 (two-tailed) Hypothesis testing Hypothesis 1 predicted a positive relationship between visionary leadership and follower relational energy. As indicated in Table 3, visionary leadership had a positive effect on follower relational energy (β = 0.37, p < 0.01, Model 1). Thus, Hypothesis 1 was supported. Hypothesis 2 posited a negative relationship between relational energy and WFC. As indicated in Table 3, this hypothesis was also supported (β = -0.17, p < 0.01, Model 3).Table 3 Results of structural equation modelling for the mediation and moderated mediation models Outcome variables Relational energy WFC Model 1 Model 2 Model 3 Control variables Gender − 0.18 − 0.15 − 0.13 Age − 0.06 − 0.07 − 0.01 Education 0.05 0.04 − 0.04 Marriage 0.09 0.12 0.23 Children − 0.11 − 0.18 − 0.16 Tenure − 0.04 − 0.03 − 0.02 Time with supervisor 0.08 0.08 0.01 Manager 0.06 0.08 0.02 Industry 0.05 − 0.04 0.02 Predictor variables Visionary leadership 0.37 0.34 − 0.09 Relational energy -17** Moderating effects PCD 0.07* Visionary leadership x PCD 0.13 R 2 0.17 0.65 0.10 n = 268 employees. For the R2, all models were compared to the corresponding null models. PCD = perceived COVID-19 crisis disruption; all coefficients reported were unstandardized; p ≤ 0.01, * p ≤ 0.05 (two-tailed) Hypothesis 3 predicted that follower relational energy would mediate the relationship between visionary leadership and follower WFC. The results of resampling-based bootstrapping indicated that visionary leadership had significant negative effects on follower WFC through follower relational energy (indirect effect = -0.07, 95% CI [-0.13, -0.02]). Thus, Hypothesis 3 was supported. Hypothesis 4 predicted that perceived COVID-19 crisis disruption would act as a moderator in the link between visionary leadership and follower WFC. Consistent with Hypothesis 4, the interaction between visionary leadership and perceived COVID-19 crisis disruption was positively associated with follower relational energy (β = 0.13, p < 0.01, Model 2), as shown in Table 3. Following the procedure proposed by Aiken and West (1991), the results of simple slope tests further revealed that the impact of visionary leadership on follower relational energy was stronger when employees’ perceived COVID-19 crisis disruptions were high (+ SD; simple slope = 0.50, p < 0.01), compared with low perceived COVID-19 crisis disruption (- SD, simple slope = 0.19, p < 0.05). Therefore, Hypothesis 4 was supported. The simple slopes are illustrated in Fig. 2.Fig. 2 The moderating effect of perceived COVID-19 crisis disruption Hypothesis 5 predicted a moderated mediation model in which perceived COVID-19 crisis disruption would increase the indirect effects of visionary leadership on follower WFC through follower relational energy. Results of bootstrap analyses indicated that the indirect effects of visionary leadership on WFC were stronger when employees’ perceived COVID-19 crisis disruption was high (+ SD; indirect effect = -0.11, 95% CI [-0.19, -0.05]) compared with low perceived COVID-19 crisis disruption (- SD; indirect effect = -0.04, 95% CI [-0.10, -0.01]), as shown in Table 4. In addition, the difference in the indirect effect of visionary leadership on WFC was significant (difference = -0.07, 95% CI [-0.14, -0.02]). Therefore, Hypothesis 5 was supported. Table 4 Results of moderated path analysisa Visionary leadership (X) → Relational energy (M) → WFC (Y) Pathsb First Stage PMX Second Stage PYM Direct Effects PYX Indirect Effects PYMPMX Total Effects PYX + PYMPMX Simple paths for high PCD 0.50 − 0.21 − 0.10 − 0.11 − 0.21** Simple paths for low PCD 0.19* − 0.21 − 0.10 − 0.04 − 0.14 Differences 0.31 0.00 0.00 − 0.07* − 0.07* an = 268 b PMX is the path from visionary leadership to follower relational energy; PYM, the path from follower relational energy to WFC; PYX, the path from visionary leadership to WFC. Low/ high perceived COVID-19 crisis disruption refers to one standard deviation below/above the mean of perceived COVID-19 crisis disruption. Tests of differences for the indirect and total effect were based on bias-corrected confidence intervals derived from bootstrap estimates All coefficients reported were unstandardized; ** p ≤ 0.01, * p ≤ 0.05 (two-tailed) Discussion Theoretical implications The current study makes three important theoretical contributions to the existing literature. First, by introducing a resource perspective to theoretically and empirically integrate visionary leadership with employees’ family life outcomes, our research extends the understanding of the cross-domain effects of visionary leadership. Prior research has predominantly focused on investigating the impacts of visionary leadership on followers’ work domain outcomes (e.g., Kearney et al., 2019). While this is an important issue to investigate, the potential cross-domain effects of visionary leadership remain largely unexplored. This is largely due to limitations in the scope of existing theoretical perspectives (e.g., motivation and self-conception perspectives), whose analytical logic can only be applied to work-related outcomes. In the current study, we introduced a new perspective (i.e., the resource perspective) that enabled us to go further by shifting away from the predominant focus on work-related outcomes and looking instead at home domain outcomes. In doing so, we offer a timely response to Li et al.’s (2017) call for more research on the cross-domain impacts of visionary leadership. We hope our study can provide a springboard for future studies and inspire further research on visionary leadership’s impact on multiple domains, thus leading to a more complete picture of how classic leadership types facilitate the enrichment of employees’ work–family interface. Second, the mediating role of relational energy proposed in our study can enrich our knowledge of the influencing mechanisms of visionary leadership. When investigating the impacts of visionary leadership, previous research has largely focused on job-related process factors, such as team members’ strategic consensus (Ates et al., 2020) and the adequacy of team communication (Greer et al., 2012). However, little is known about the role that employees’ personal resource-related factors may play in explaining the influence of visionary leadership. Driven by the W-HR model (Ten Brummelhuis & Bakker, 2012), we identified relational energy as a key personal resource that connects visionary leadership and the cross-domain interface. Our results suggest that visionary leadership offsets WFC by facilitating the accumulation of employees’ relational energy. By proposing this novel explanatory mechanism, we address the critical question of whether and how visionary leaders have an impact on followers’ WFC, and provide theoretical and empirical evidence for the cross-domain effects of visionary leadership. Third, by exploring the moderating role of perceived COVID-19 crisis disruption in the association between visionary leadership and followers’ WFC, the current research contributes to the visionary leadership literature by identifying a critical situational contingency for its cross-domain consequences. Previous studies have mainly focused on individual traits (e.g., political skill; Kwan et al., 2022) or organizational characteristics (e.g., leader–follower spatial distance; Yair et al., 2015) when investigating the boundary conditions of the impacts of visionary leadership. Although these are important issues to explore, the role of employees’ perceptions of the macro environment has been largely ignored. Drawing on the W-HR model (Ten Brummelhuis & Bakker, 2012), we introduced perceived COVID-19 crisis disruption as a key situational factor that creates a greater need for insightful guidance and encouragement from supervisors, thereby accentuating the beneficial effects of visionary leadership on followers’ relational energy and, in turn, their WFC. Our findings extend the understanding of macro factors that may influence the spillover effects of visionary leadership. Practical implications This research provides some meaningful insights for management practice. First, our work highlights the significant role of visionary leadership in reducing employee WFC. Visionary leadership can provide subordinates with energetic activation by helping them maintain a compelling vision of the organization, which in turn encourages them to modify their behaviors to meet family expectations and reduce WFC. Thus, our results encourage organizations to select leadership candidates with strong visionary traits. Training programs should be provided for leaders to cultivate their awareness of and ability to apply visionary leadership. Second, by revealing the mediating role of relational energy in the link between visionary leadership and the WFC of followers, our study indicates a new way to promote the potential positive cross-domain effects of visionary leadership in management practice. Our results indicate that visionary leadership can generate greater relational energy among followers, which can be utilized to reduce WFC. Therefore, managers should pay attention to employees’ relational energy when striving to help them achieve work–life balance. Because interactions between leaders and followers are often limited by time and physical distance, leaders should endeavor to improve the quality of their communication and develop interpersonal bonds with followers (Mao et al., 2022). Highlighting communication transparency might also be a potential strategy for improving leader–subordinate dyadic relationships. Third, our research highlights the necessity of visionary leadership in the context of the COVID-19 pandemic. Our research reveals that employees who perceive high levels of COVID-19 pandemic disruption tend to benefit more from the cross-domain effects of visionary leadership, mainly because they suffer more from the changes and uncertainty arising from the pandemic. For organizations that are dedicated to the development of visionary leadership in the context of the crisis, our research illustrates that visionary leadership addresses crisis response by communicating values and setting a direction for the future. The COVID-19 pandemic is disruptive and complex, and has interrupted the established routines of organizations and caused significant anxiety and pressure among employees (Collings et al., 2021). In times of crisis, our findings suggest that visionary leadership is instrumental and uniquely positioned to lead employees through potential confusion. It is advisable for organizations to implement assistance programs designed to help employees recover from the problematic states associated with the pandemic. For example, various types of fun team development activities can build internal ties and enhance mutual support among team members, which may help them cope with the COVID-19 crisis more effectively. Limitations and future directions Some limitations are worth noting in this study. First, all our data came from five enterprises in China, which may undermine the external validity of our findings. Prior research has posited that, compared with people in Western countries, Chinese people may pay more attention to their family roles due to cultural familism, and that the boundaries between their family and work roles are often blurred (Zhang & Tu, 2018). Therefore, the effects of visionary leadership on followers’ home domain outcomes may vary in other cultural contexts. We suggest that future research use samples from different cultures to further validate our theoretical framework. Second, although we investigated perceived COVID-19 pandemic disruption as a critical boundary factor for understanding the cross-domain consequences of visionary leadership, our analytical scope could still be limited. The W-HR framework suggests that key resources, such as personality traits, are a type of management resource that facilitates the selection, distribution, and function of other kinds of resources (Wang et al., 2022). These key resources may interact with macro factors and collectively affect the work–home resources process (Ten Brummelhuis & Bakker, 2012). Future research could go a step further by examining the interaction effects of contextual factors and individual personalities when exploring the cross-domain consequences of visionary leadership. In addition, based on our findings, it is important for future work to further explore potential moderators in the relationship between relational energy and WFC. Third, although we used WFC as a key variable to reflect family domain outcomes, there are other important variables that can reflect employees’ quality of family life from different angles, such as family satisfaction, marital satisfaction, family stress, and work–family enrichment (Kim et al., 2021; Li et al., 2017). To better understand the potential spillover effects of visionary leadership on employees’ nonwork domains, future studies could explore the underlying mechanisms of the relationships between visionary leadership and these family-related outcomes. Fourth, our study did not control for the COVID-19-related lockdown context or whether the respondents were working from home or from the office (or both) during the data collection. Previous research suggests that COVID-related worry and psychological distress tend to increase during lockdown (e.g., Fernández et al., 2022; O’Connor et al., 2022). To provide a more nuanced understanding of the visionary leadership–WFC association in the COVID-19 context, future studies could further validate our findings by controlling for COVID-19-related factors. Conclusion Drawing upon the W-HR model (Ten Brummelhuis & Bakker, 2012), we proposed a moderated mediation model and explained how and when visionary leadership has a spillover effect on the WFC of subordinates. In support of our model, we found that visionary leadership had beneficial effects of aiding followers’ relational energy, which in turn negatively affected WFC. Moreover, perceived COVID-19 crisis disruption moderated the negative indirect relationship between visionary leadership and followers’ WFC, such that the relationship was stronger when followers perceived a high level of COVID-19 crisis disruption. Funding This research was funded by the National Natural Science Foundation of China (72,102,051), Guangdong Basic and Applied Basic Research Foundation (2019A1515010727, 2020A151 5,110,530), Guangdong Planning Office of Philosophy and Social Science (GD19CGL15), and the Project of Guangdong Provincial Humanities and Social Sciences Key Research Base (18JD07). Data availability The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethical approval All procedures performed in the study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Conflict of interest All authors declare no conflicts of interest with respect to the authorship or the publication of this article. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Aiken, L.S., & West, S.G. (1991). Multiple regression: Testing and interpreting interactions. Sage Allen TD Herst DEL Bruck CS Sutton M Consequences associated with work-to-family conflict: A review and agenda for future research Journal of Occupational Health Psychology 2000 5 2 278 308 10.1037/1076-8998.5.2.278 10784291 Ateş NY Tarakci M Porck JP van Knippenberg D Groenen PJF The dark side of visionary leadership in strategy implementation: Strategic alignment, strategic consensus, and commitment Journal of Management 2020 46 5 637 665 10.1177/0149206318811567 Berson Y Halevy N Shamir B Erez M Leading from different psychological distances: A construal-level perspective on vision communication, goal setting, and follower motivation Leadership Quarterly 2015 26 2 143 155 10.1016/j.leaqua.2014.07.011 Brislin RW Back-translation for cross-cultural research Journal of Cross-Cultural Psychology 1970 1 3 185 216 10.1177/135910457000100301 Carlson DS Kacmar KM Williams LJ Construction and initial validation of a multidimensional measure of work-family conflict Journal of Vocational Behavior 2000 56 2 249 276 10.1006/jvbe.1999.1713 Carton AM Lucas BJ How can leaders overcome the blurry vision bias? Identifying an antidote to the paradox of vision communication Academy of Management Journal 2018 61 6 2106 2129 10.5465/amj.2015.0375 Carton AM Murphy C Clark JR A (blurry) vision of the future: How leader rhetoric about ultimate goals influences performance Academy of Management Journal 2014 57 6 1544 1570 10.5465/amj.2012.0101 Cole MS Walter F Bruch H Affective mechanisms linking dysfunctional behavior to performance in work teams: A moderated mediation study Journal of Applied Psychology 2008 93 5 945 958 10.1037/0021-9010.93.5.945 18808218 Collings DG Nyberg AJ Wright PM Mcmackin J Leading through paradox in a covid world: Human resources come of age Human Resource Management Journal 2021 31 4 819 833 10.1111/1748-8583.12343 Doty DH Glick WH Common methods bias: Does common methods variance really bias results? Organizational Research Methods 1998 1 4 374 406 10.1177/109442819814002 Edwards JR Lambert LS Methods for integrating moderation and mediation: A general analytical framework using moderated path analysis Psychological Methods 2007 12 1 1 22 10.1037/1082-989X.12.1.1 17402809 Eseryel UY Crowston K Heckman R Functional and visionary leadership in self-managing virtual teams Group & Organization Management 2021 46 2 424 460 10.1177/1059601120955034 Fan J Wei X Ko I How do hotel employees’ feeling trusted and its differentiation shape service performance: The role of relational energy International Journal of Hospitality Management 2021 92 102700 10.1016/j.ijhm.2020.102700 Fernández RS Crivelli L Guimet NM Allegri RF Picco S Pedreira ME Psychological distress and mental health trajectories during the COVID-19 pandemic in Argentina: A longitudinal study Scientific Reports 2022 12 1 1 10 10.1038/s41598-022-09663-2 34992227 Fisher J Languilaire JC Lawthom R Nieuwenhuis R Petts RJ Runswick-Cole K Yerkes MA Community, work, and family in times of COVID-19 Community Work & Family 2020 23 3 247 252 10.1080/13668803.2020.1756568 Greer LL Homan AC De Hoogh AHB Den Hartog DN Tainted visions: The effect of visionary leader behaviors and leader categorization tendencies on the financial performance of ethnically diverse teams Journal of Applied Psychology 2012 97 1 203 213 10.1037/a0025583 21942407 Greenhaus JH Beutell NJ Sources of conflict between work and family roles Academy of Management Review 1985 10 1 76 88 10.5465/amr.1985.4277352 Hobfoll SE Conservation of resources: A new attempt at conceptualizing stress American Psychologist 1989 44 3 513 524 10.1037/0003-066X.44.3.513 2648906 Jones P Comfort D The COVID-19 crisis and sustainability in the hospitality industry International Journal of Contemporary Hospitality Management 2020 32 10 3037 3050 10.1108/IJCHM-04-2020-0357 Kerman K Korunka C Tement S Work and home boundary violations during the COVID-19 pandemic: The role of segmentation preferences and unfinished tasks Applied Psychology 2021 71 3 784 806 10.1111/apps.12335 34548734 Kearney E Shemla M Knippenberg DV Scholz FA A paradox perspective on the interactive effects of visionary and empowering leadership Organizational Behavior and Human Decision Processes 2019 155 20 30 10.1016/j.obhdp.2019.01.001 Kim SY Shigemoto Y Neduvelil A Grzywacz JG Longitudinal stability of work–family enrichment and its association with well-being and personality traits The Counseling Psychologist 2021 49 6 907 939 10.1177/00110000211015909 34381249 Kohles JC Bligh MC Carsten MK A follower-centric approach to the vision integration process The Leadership Quarterly 2012 23 3 476 487 10.1016/j.leaqua.2011.12.002 Kwan HK Chen H Chiu RK Effects of empowering leadership on followers’ work–family interface The International Journal of Human Resource Management 2022 33 7 1403 1436 10.1080/09585192.2020.1762701 Li A McCauley KD Shaffer JA The influence of leadership behavior on employee work–family outcomes: A review and research agenda Human Resource Management Review 2017 27 3 458 472 10.1016/j.hrmr.2017.02.003 Liu, W., Zhang, Q. I., Chen, J., Xiang, R., Song, H., Shu, S., & Liu, Y. (2020). Detection of Covid-19 in children in early January 2020 in Wuhan, China. New England Journal of Medicine, 382(14), 1370–1371. http://hdl.handle.net/10722/283675 Mao JY Li Y Guo L Shi Y Wang J Diminished relational energy: How and when co-worker incompetence accusations hinder employee creativity European Journal of Work and Organizational Psychology 2022 31 3 383 394 10.1080/1359432X.2021.1979963 Maran TK Baldegger U Klösel K Turning visions into results: Unraveling the distinctive paths of leading with vision and autonomy to goal achievement Leadership & Organization Development Journal 2022 43 1 133 154 10.1108/LODJ-06-2021-0268 Marsh HW Hau KT Wen Z In search of golden rules: Comment on hypothesis-testing approaches to setting cutoff values for fit indexes and dangers in overgeneralizing hu and bentler’s (1999) findings Structural Equation Modeling 2004 11 3 320 341 10.1207/s15328007sem11032 Mayfield J Mayfield M The creative environment’s influence on intent to turnover Management Research News 2007 31 1 41 56 10.1108/01409170810845949 Morgeson Frederick P The external leadership of self-managing teams: Intervening in the context of novel and disruptive events Journal of Applied Psychology 2005 90 3 497 508 10.1037/0021-9010.90.3.497 15910145 Mueller RO Hancock GR Hancock GR Mueller RO Structural equation modeling The reviewer’s guide to quantitative methods in the social sciences 2010 Routledge 371 384 Muthén, L. K., & Muthén, B. O. (1998–2017). Mplus user’s guide (8th ed.). Muthén & Muthén. O’Connor, D. B., Wilding, S., Ferguson, E., Cleare, S., Wetherall, K., McClelland, H., & O’Connor, R. C. (2022). Effects of COVID-19-related worry and rumination on mental health and loneliness during the pandemic: longitudinal analyses of adults in the UK COVID-19 mental health & wellbeing study. Journal of Mental Health, 1–12. 10.1080/09638237.2022.2069716 Owens BP Baker WE Sumpter MD Cameron KS Relational energy at work: Implications for job engagement and job performance Journal of Applied Psychology 2016 101 1 35 10.1037/apl0000032 26098165 Paredes RM Apaolaza V Fernandez-Robinr C Hartmann P Yañez-Martinezc D The impact of the covid-19 pandemic on subjective mental well-being: the interplay of perceived threat, future anxiety and resilience Personality and Individual Differences 2021 170 110445 10.1016/j.paid.2020.110455 Paul, J., Costley, D. L., Howell, J. P., et al. (2001). The effects of charismatic leadership on followers’ self-concept accessibility. Journal of Applied Social Psychology, 31(9), 1821–1842. 10.1111/j.1559-1816.2001.tb00206.x Perry SJ Witt LA Penney LM Atwater L The downside of goal-focused leadership: the role of personality in subordinate exhaustion Journal of Applied Psychology 2010 95 6 1145 1153 10.1037/a0020538 20718524 Podsakoff PM MacKenzie SB Lee J Common method biases in behavioral research: a critical review of the literature and recommended remedies Journal of Applied Psychology 2003 88 5 879 903 10.1037/0021-9010.88.5.879 14516251 Stam D Lord RG Knippenberg DV Wisse B An image of who we might become: Vision communication, possible selves, and vision pursuit Organization Science 2014 25 4 1172 1194 10.1287/orsc.2013.0891 Stam D Van Knippenberg D Wisse B Focusing on followers: The role of regulatory focus and possible selves in visionary leadership The Leadership Quarterly 2010 21 3 457 468 10.1016/j.leaqua.2010.03.009 Stam D Van Knippenberg D Wisse B Nederveen Pieterse A Motivation in words: Promotion-and prevention-oriented leader communication in times of crisis Journal of Management 2018 44 7 2859 2887 10.1177/0149206316654543 Ten Brummelhuis LL Bakker AB A resource perspective on the work–home interface: The work–home resources model American Psychologist 2012 67 7 545 556 10.1037/a0027974 22506688 Van Knippenberg D Sitkin SB A critical assessment of charismatic-transformational leadership research: Back to the drawing board? The Academy of Management Annals 2013 7 1 60 10.1080/19416520.2013.759433 Venus M Johnson RE Zhang S Wang X Lanaj K Seeing the big picture: A within-person examination of leader construal level and vision communication Journal of Management 2019 45 7 2666 2684 10.1177/0149206318761576 Wang, H., Zhu, Y., Liang, Y., & Xie, J. (2022). Does customer mistreatment hinder employees from going the extra mile? The mixed blessing of being conscientious. International Journal of Hospitality Management, 102, 103155. Advanced published online. 10.1016/j.ijhm.2022.103155 Williams LJ Cote JA Buckley MR Lack of method variance in self-reported affect and perceptions at work: Reality or artifact? Journal of Applied Psychology 1989 74 3 462 468 10.1037/0021-9010.74.3.462 Yair B Nir H Boas S Miriam E Leading from different psychological distances: A construal-level perspective on vision communication, goal setting, and follower motivation The Leadership Quarterly 2015 26 2 143 155 10.1016/j.leaqua.2014.07.011 Yang F Liu J Wang Z Zhang Y Feeling energized: A multilevel model of spiritual leadership, leader integrity, relational energy, and job performance Journal of Business Ethics 2019 158 4 983 997 10.1007/s10551-017-3713-1 Zhang S Tu Y Cross-domain effects of ethical leadership on employee family and life satisfaction: The moderating role of family-supportive supervisor behaviors Journal of Business Ethics 2018 152 4 1085 1097 10.1007/s10551-016-3306-4	36474483	PMC9716123	NO-CC CODE	2022-12-03 23:20:54	no	Curr Psychol. 2022 Dec 2;:1-12	utf-8	Curr Psychol	2,022	10.1007/s12144-022-04080-0	oa_other
==== Front Pediatr Nephrol Pediatr Nephrol Pediatric Nephrology (Berlin, Germany) 0931-041X 1432-198X Springer Berlin Heidelberg Berlin/Heidelberg 5813 10.1007/s00467-022-05813-w Original Article Humoral and cellular immune response to SARS-CoV-2 mRNA BNT162b2 vaccine in pediatric kidney transplant recipients compared with dialysis patients and healthy children http://orcid.org/0000-0001-5504-3516 Gulmez Ruveyda 1 http://orcid.org/0000-0002-0596-1551 Ozbey Dogukan 2 http://orcid.org/0000-0002-3658-8622 Agbas Ayse [email protected] 1 http://orcid.org/0000-0003-3274-8024 Aksu Bagdagul 34 http://orcid.org/0000-0001-6805-5313 Yildiz Nurdan 5 http://orcid.org/0000-0002-3849-3280 Uckardes Diana 6 http://orcid.org/0000-0002-2424-6959 Saygili Seha 1 http://orcid.org/0000-0002-6097-657X Yilmaz Esra Karabag 1 http://orcid.org/0000-0003-2891-2231 Yildirim Zeynep Yuruk 3 http://orcid.org/0000-0002-5579-6339 Tasdemir Mehmet 7 http://orcid.org/0000-0001-5821-3963 Kiykim Ayca 8 http://orcid.org/0000-0002-0086-3936 Cokugras Haluk 8 http://orcid.org/0000-0002-3420-9756 Canpolat Nur 1 http://orcid.org/0000-0002-3357-9237 Nayir Ahmet 39 http://orcid.org/0000-0003-1072-3846 Kocazeybek Bekir 2 http://orcid.org/0000-0002-3316-8032 Caliskan Salim 1 1 grid.506076.2 0000 0004 1797 5496 Department of Pediatric Nephrology, Cerrahpasa School of Medicine, IU-Cerrahpasa, Istanbul, Turkey 2 grid.506076.2 0000 0004 1797 5496 Department of Microbiology, Cerrahpasa School of Medicine, IU-Cerrahpasa, Istanbul, Turkey 3 grid.9601.e 0000 0001 2166 6619 Department of Pediatric Nephrology, Istanbul University School of Medicine, Istanbul, Turkey 4 grid.9601.e 0000 0001 2166 6619 Institute of Child Health, Istanbul University, Istanbul, Turkey 5 grid.16477.33 0000 0001 0668 8422 Department of Pediatric Nephrology, Marmara University School of Medicine, Istanbul, Turkey 6 grid.411776.2 0000 0004 0454 921X Department of Pediatric Nephrology, Medeniyet University School of Medicine, Istanbul, Turkey 7 grid.459708.7 0000 0004 7553 3311 Department of Pediatric Nephrology, Istinye University School of Medicine, Liv Hospital, Istanbul, Turkey 8 grid.506076.2 0000 0004 1797 5496 Department of Pediatric Immunology and Allergy, Cerrahpasa School of Medicine, IU-Cerrahpasa, Istanbul, Turkey 9 Department of Pediatric Nephrology, Memorial Hospital, Istanbul, Turkey 2 12 2022 110 11 7 2022 10 10 2022 31 10 2022 © The Author(s), under exclusive licence to International Pediatric Nephrology Association 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Background Compared with the general population, the immune response to COVID-19 mRNA vaccines is lower in adult kidney transplant recipients (KTRs). However, data is limited for pediatric KTRs. In this study, we aimed to assess humoral and cellular immune responses to the COVID-19 mRNA vaccine in pediatric KTRs. Methods This multicenter, prospective, case–control study included 63 KTRs (37 male, aged 12–21 years), 19 dialysis patients, and 19 controls. Humoral (anti-SARS-CoV2 IgG, neutralizing Ab (nAb)) and cellular (interferon-gamma release assay (IGRA)) immune responses were assessed at least one month after two doses of BNT162b2 mRNA vaccine. Results Among COVID-19 naïve KTRs (n = 46), 76.1% tested positive for anti-SARS-CoV-2 IgG, 54.3% for nAb, and 63% for IGRA. Serum levels of anti-SARS-CoV-2 IgG and nAb activity were significantly lower in KTRs compared to dialysis and control groups (p < 0.05 for all). Seropositivity in KTRs was independently associated with shorter transplant duration (p = 0.005), and higher eGFR (p = 0.007). IGRA titer was significantly lower than dialysis patients (p = 0.009). Twenty (43.4%) KTRs were positive for all immune parameters. Only four of 11 seronegative KTRs were IGRA-positive. COVID-19 recovered KTRs had significantly higher anti-SARS-CoV-2 IgG and nAb activity levels than COVID-19 naïve KTRs (p = 0.018 and p = 0.007, respectively). Conclusions The humoral and cellular immune responses to SARS-CoV-2 mRNA BNT162b2 vaccine are lower in pediatric KTRs compared to dialysis patients. Further prospective studies are required to demonstrate the clinical efficacy of the mRNA vaccine in KTRs. This prospective study was registered in ClinicalTrials.gov (NCT05465863, registered retrospectively at 20.07.2022). Graphical abstract A higher resolution version of the Graphical abstract is available as Supplementary information. Supplementary information The online version contains supplementary material available at 10.1007/s00467-022-05813-w. Keywords mRNA vaccine BNT162b2 COVID-19 Immune response Anti-SARS-CoV-2 IgG IGRA Neutralizing antibody Scientific Research Projects Coordination Unit of Istanbul University-Cerrahpasa35866 Caliskan Salim ==== Body pmcIntroduction Coronavirus 2019 (COVID-19) infection is associated with higher morbidity and mortality in adult patients on dialysis and kidney transplant recipients (KTRs) [1–4]. Pediatric KTRs develop asymptomatic or mild COVID-19 disease with a favorable outcome [5]. An increased risk of subclinical acute kidney injury (AKI) is associated with mild to moderate COVID-19 in children; however, less is known about the transplanted kidney [6]. Given KTRs are immunosuppressed and vulnerable to infection and SARS-CoV-2 appears to have an affinity for the kidney, vaccination of this population is important. It is well established that vaccine response (to attenuated, conjugated, or recombinant) is lower in pediatric dialysis patients and KTRs when compared with the general population [7, 8]. This is attributed to uremia and immunosuppressant medications [9]. The new mRNA vaccine technology is being used worldwide, including in children and adolescents during the pandemic. Studies have demonstrated a lower immune response to the new SARS-CoV-2 mRNA vaccine in adult KTRs [10–17]. However, there are limited data on the immune response elicited by the vaccine in children and adolescents with kidney replacement therapy [18, 19]. The aim of this study was to investigate both humoral and cellular immune responses to two-doses of BNT162b2 mRNA COVID-19 vaccine in pediatric KTRs compared with dialysis patients and healthy controls. The humoral immune response was assessed using anti-SARS-CoV-2 immunoglobulin G (anti-SARS-CoV-2 IgG) and SARS-CoV-2 neutralizing antibody (nAb). The cellular immune response was assessed using the SARS-CoV-2-specific interferon-ɣ-release assay (IGRA). Material and methods Study design This prospective, multicenter case–control study was conducted with the participation of five pediatric nephrology centers in Istanbul between September 2021 and March 2022. The centers were asked to report all dialysis and kidney transplant patients between the ages of 12 and 21 years to be vaccinated against COVID-19. Patients were informed about the vaccine according to local vaccination schedule and requested to make an appointment for vaccination. The BNT162b2 mRNA COVID-19 vaccine (Pfizer-BioNTech®) was administered intramuscularly into the deltoid region of all participants who agreed to participate in the study. At least one month after the second vaccine dose, serum and whole blood samples were collected from all patients and controls to analyze the humoral and cellular immune response to the vaccine. All samples were stored at –20 °C until assayed. SARS-CoV-2 PCR test results were collected retrospectively to determine natural SARS-CoV-2 infection. Patients with a history of positive SARS-CoV-2 PCR were defined as “COVID-19 recovered,” and the remaining were defined as “COVID-19 naïve.” The control group consisted of 19 age- and gender-comparable healthy children. See Fig. 1 for the flow diagram of the study design. All patients were recommended a third dose of the vaccine, but only 13 of 63 KTRs received a third dose.Fig. 1 Flow diagram of the study design. KTRs, kidney transplant recipients; HD, hemodialysis; PD, peritoneal dialysis Assessment of immune response to SARS-CoV-2 vaccine Humoral response Humoral immune response was assessed with anti-SARS-CoV-2 IgG quantification (anti-SARS-CoV-2 IgG) and neutralization test (nAb activity). Anti-SARS-CoV-2 IgG antibody titers, which prevent the binding of SARS-CoV-2 S1/RBD region to ACE2 receptors, were determined by SARS-CoV-2 QuantiVac ELISA (IgG) (Euroimmun AG, Lübeck, Germany). Neutralization capacities of these antibodies were also determined by SARS-CoV-2 NeutraLISA (Euroimmun AG, Lübeck, Germany). Antibody titers were obtained in relative units/mL (RU/mL) (1 RU/mL * 3.2 = 1 Binding Antibody Unit per mL (BAU/mL)). Antibody titer values below 8 RU/mL (25.6 BAU/mL) were interpreted as “seronegative,” and values above 11 RU/mL (35.2 BAU/mL) were interpreted as “seropositive,” according to manufacturer’s guidelines. Serum samples that exceeded assay measuring range (> 120 RU/ml) were diluted by a 1:10 factor and tested again to obtain more accurate results. Neutralizing antibody responses were assessed as percent inhibition (%IH). Percent inhibition values below 20% were considered “nAb-negative,” and values above 35% were considered “nAb-positive,” according to manufacturer’s guidelines. Cellular immune response Cellular immune response was assessed with IGRA. A specific stimulation of T-cells by the spike protein of SARS-CoV-2 was performed using the Quan-T-Cell SARS-CoV-2 (Euroimmun AG, Lübeck, Germany) to determine the amount of IFN-γ released by immune cells. IFN-y responses were then measured using the Quan-T-Cell ELISA (Euroimmun AG, Lübeck, Germany) Interferon Gamma Release Assay. Results were obtained in milli-international units per milliliter (mIU/mL) in accordance with the manufacturer’s instructions. Values below 100 mIU/mL were interpreted as “IGRA-negative,” and values above 200 mIU/mL were interpreted as “IGRA-positive.” Statistical analyses The Statistical Package for Social Sciences (SPSS) for Windows version 20.0 (SPSS, IBM Corporation, Chicago) was used for analysis. GraphPad Prism version 9.4.0 (GraphPad Software, San Diego, CA) was used for figures. Normality of the data were tested with the Kolmogorov–Smirnov test. Continuous data were expressed as mean ± standard deviation (SD) in the case of a normal distribution and analyzed with the Student t-test or one-way ANOVA test. In the case of a non-normal distribution, data were expressed as the median (interquartile range, 25th; 75th percentiles) and analyzed using the Mann–Whitney U test or the Kruskal–Wallis test. Categorical variables were expressed as n (%) and analyzed with Chi-Square test. Bonferroni correction was applied when appropriate. The correlation between anti-SARS-CoV-2 IgG, nAb, and IGRA levels were analyzed using the Spearman correlation test. Variables with a p value of < 0.1 were analyzed with backward multivariate logistic regression analysis to determine the independent predictors of anti-SARS-CoV-2 IgG and nAb positivity. A two tailed p value < 0.05 was defined as significant. Results A total of 63 KTRs, 19 dialysis patients (15 HD and 4 PD), and 19 healthy controls were included in the study (Fig. 1 and Table 1). The median time between second vaccine dose and the assessment of immune response was 8 weeks (7;14 weeks) in the KTRs. The difference was not statistically significant from dialysis and control groups. There was no statistically significant difference in age, gender, or SARS-CoV-2 PCR positivity between the three groups (Table 1). Forty-nine of the KTRs (78%) were on standard triple immunosuppressive therapy (prednisolone, tacrolimus, and mycophenolate mofetil/mycophenolic acid (MMF/MPA)), five were on triple treatment with prednisolone, mTORi, and MMF/MPA, and four were on tacrolimus/cyclosporin and MMF/MPA. Fifty-five KTRs (82.5%) received a kidney from a living donor. Acute rejection history was present in six KTRs; none of which occurred within 6 months of the study. Median time posttransplantation was 77 months with all participants having received a transplant greater than 12 months prior to the study. A total of 17 KTRs had a history of COVID-19 before vaccinations; the median time between COVID-19 and evaluation of immune response to vaccine was 47 weeks.Table 1 Characteristics of study population KTRs (n = 63) Dialysis patients (n = 19) Controls (n = 19) p value Clinical characteristics Age, years 15.9 ± 2.86 17.1 ± 1.90 15.9 ± 2.27 0.262 Male sex, n (%) 37/63 (58.7) 8/19 (42.1) 9/19 (47.4) 0.373 Time on dialysis, months 1.0 (0–31) 29 (12–48) 0.001 Time on transplantation, months 77 (55.5–103) - - Blood tests White blood cells, × 103 U/µl 7.4 (6.1–10.3) 6.1 (5.0–7.9) NA 0.013 Lymphocytes, × 103 U/µl 2.4 (2.0–2.9) 2.3 (1.6–2.7) NA 0.135 Vaccination and immunity related features SARS-Cov-2 PCR positivity, n (%) 17/58 (29.3) 8/16 (50.0) 8/18 (44.4) 0.218 Time after PCR ( +), weeks 47 (33–58) 56 (48–63) 52 (38–64) 0.665 Time after 2nd vaccine dose, weeks 8.0 (5.5–10.5) 8.0 (7.0–14.0) 15.0 (4.0–22.5) 0.448 Anti-SARS-CoV-2 IgG titers, RU/ml 306.3 (28.2–1200)a 1052 (369–1200)b 731 (394–1200)b 0.014 Neutralizing antibody, % activity 91.4 (4.93–99.4)a 99.4 (97.6–99.5)b 99.4 (98.9–99.6)b < 0.001 IGRA titer, mIU/ml 304 (138–1438)a 1830 (1024–1985)b 841 (421–2057)a,b 0.003 Anti-SARS-CoV-2 IgG positivity, n (%) 51/63 (81)a 19/19 (100)a 19/19 (100)a 0.015 Neutralizing antibody positivity, n (%) 41/63 (65.1)a 16/19 (84.2)b 18/19 (94.7)b 0.023 IGRA positivity, n (%) 41/63 (65.1)a 19/19 (100)b 17/19 (89.5)a,b 0.001 COVID-naïve study population (n = 46) (n = 11) (n = 11) Anti-SARS-CoV-2 IgG titers, RU/ml 235 (14.8–746)a 526 (369–1200)b 992 (394–1118)b 0.012 Neutralizing antibody, % 48.9 (3.9–98.8)a 99.4 (97.4–99.5)b 99.1 (98.8–99.5)b 0.001 IGRA titer, mIU/ml 282 (85.7–1288)a 1776 (1024–1985)b 723 (366–1793)a 0.021 Anti-SARS-CoV-2 IgG positivity, n (%) 35/46 (76.1)a 11/11 (100)a 11/11 (100)a 0.048 Neutralizing antibody positivity, n (%) 25/46 (54.3)a 9/11 (81.8)a 11/11 (100)b 0.005 IGRA positivity, n (%) 29/46 (63.0)a 11/11 (100)b 9/11 (81.8)a,b 0.029 COVID recovered study population (n = 17) (n = 8) (n = 8) Anti-SARS-CoV-2 IgG titers, RU/ml 906 (177–1200) 1200 (380–12,000) 703 (425–1200) 0.765 Neutralizing antibody, % 99.2 (93.8–99.5) 99.4 (97.6–99.5) 99.4 (98.4–99.5) 0.318 IGRA titer, mIU/ml 523 (162–1984) 1886 (730–2023) 1646 (583–2060) 0.244 Anti-SARS-CoV-2 IgG positivity, n (%) 16/17 (94.1) 8/8 (100) 8/8 (100) 1.000 Neutralizing antibody positivity, n (%) 16/17 (100) 7/8 (87.5) 7/8 (87.5) 1.000 IGRA positivity, n (%) 12/17 (70.6) 8/8 (100) 8/8 (100) 0.053 Data are presented as mean (SD), median (25th; 75th percentile) or n/n (%). Continuous data were analyzed by the Mann–Whitney U test for two groups and the Kruskal–Wallis test or one-way ANOVA for three groups. Chi-square test or Fischer’s Exact test, where appropriate for categorical variables. Superscripts demonstrate the pairwise comparisons by Mann–Whitney U test or Chi-square test, which are given in detail for COVID-19 naïve study population in Fig. 2. P values lower than 0.05 are given in bold KTRs, kidney transplant recipients; NA, not available; anti-SARS-CoV-2 IgG, anti-SARS-CoV-2 immunoglobulin G; IGRA, interferon gamma release assay Immune response to SARS-CoV-2 vaccine in COVID-19 naïve study population The immune responses to the vaccine in COVID-19 naïve study populations are given in Table 1 and Fig. 2. KTRs had significantly lower anti-SARS-CoV-2 IgG titer levels than both dialysis (p = 0.020) and control (p = 0.023) groups (Fig. 2A). KTRs also had lower anti-SARS-CoV-2 IgG positivity (76.1%) but the difference did not reach statistical significance, either for dialysis (100%) or control (100%) groups (p = 0.099 for both, Fig. 2B). Furthermore, KTRs had significantly lower nAb activity levels than both dialysis (p = 0.006) and control (p = 0.002) groups (Fig. 2C). KTRs had also lower nAb positivity (54.3%) than dialysis (81.8%) and control (100%) groups, but the difference was statistically significant only between KTRs and controls (p = 0.004) (Fig. 2D).Fig. 2 Comparisons of humoral and cellular immune responses between kidney transplant recipients (KTRs), dialysis patients, and control subjects, among COVID-19 naïve study population. A Anti-SARS-CoV2 IgG titer, B anti-SARS-CoV2 IgG seropositivity rate, C neutralizing antibody (nAb) activity, D nAb positivity rate, E interferon gamma release assay (IGRA) titer, and F IGRA positivity rate. Only the differences between groups with a p value < 0.10 are shown in the figure The prevalence of IGRA positivity in KTRs, dialysis, and control groups were 63% (29/46), 100% (11/11), and 81.8% (9/11), respectively. KTRs also had lower IGRA titer levels compared to other two groups. However, the difference was statistically significant only between KTRs and dialysis group for both positivity (p = 0.024) and titer level (p = 0.009) (Fig. 2E and F). None of the humoral (anti-SARS-CoV-2 IgG, nAb) and cellular (IGRA) immune parameters demonstrated statistical significance between dialysis and control groups (Table 1). Factors affecting immune response to SARS-CoV-2 vaccine in COVID-19 naïve KTRs KTRs with a positive anti-SARS-CoV-2 IgG had significantly shorter time on transplantation (p = 0.005) and higher eGFR (p = 0.007) compared to seronegative KTRs (Table 2). Three of the 11 seronegative KTRs had a history of rituximab due to acute rejection, while none of the seropositive group had such a history (p = 0.012). These three KTRs had an eGFR < 50 ml/min/1.73 m2 and one of them had hypogammaglobulinemia. In multivariate logistic regression analysis, only shorter time on transplantation and higher eGFR were independently associated with a positive anti-SARS-CoV-2 IgG (ß: −0.586, OR: 0.961, 95% CI: 0.924–0.998 and ß: 0.079, OR: 0.1.082, 95% CI: 1.014–1.155, respectively). KTRs with a positive nAb activity had higher levels of tacrolimus dose, but the difference did not reach statistical significance (p = 0.063, Table 2). There was no statistical significance between IGRA-positive (n = 29) and IGRA-negative KTRs (n = 17) in terms of clinical or laboratory parameters (Table 2).Table 2 Comparison of COVID naive KTRs in terms of humoral and cellular immune response to SARS-CoV-2 vaccine Anti-SARS-CoV-2 IgG (-) (n = 11) Anti-SARS-CoV-2 IgG ( +) (n = 35) P value nAb (-) (n = 21) nAb ( +) (n = 25) p value IGRA (-) (n = 17) IGRA ( +) (n = 29) p value Clinical characteristics Age, years 14.0 (4.0) 16.0 (4.0) 0.221 14.0 (3.5) 16.5 (5.2) 0.089 16.0 (4.0) 16.0 (5.0) 0.604 Male sex, n (%) 7/11 (63.6) 24/35 (68.6) 1.000 12/21 (57.1) 19/25 (76.0) 0.174 10/17 (58.8) 21/29 (72.4) 0.516 Time on transplantation, months 102 (14) 69 (65) 0.005 92.5 (38.7) 71.5(65.5) 0.732 101 (55) 74 (59) 0.909 Living related donor, n (%) 9/11 (81.8) 32/35 (91.4) 0.580 18/21 (85.7) 23/25 (92) 0.648 16/17 (94.1) 25/29 (86.2) 0.637 Viral infection history (EBV, CMV and/or BKV history), n (%) 3/10 (30.0) 4/34 (11.8) 0.322 4/20 (20.0) 3/24 (12.5) 0.684 1/17 (5.9) 6/27 (22.2) 0.220 Time after 2nd vaccine, months 6.0 (3.0) 9.0 (7.5) 0.454 6.0 (6.2) 9.0 (7.2) 0.222 8.0 (12) 8.0 (5) 0.556 Immunosuppressive treatment Induction therapy, (no/anti IL2 R/ATG) 0/9/2 2/26/7 1.000 2/14/5 0/21/4 0.200 2/11/4 0/24/5 0.130 Acute rejection history, n (%) 3/11 (27.3) 0/35 (0) 0.012 3/21 (14.3) 0/23 (0) 0.100 2/17 (11.8) 1/27(3.7) 0.549 Rituximab treatment, n (%) 3/11 (27.3) 0/35 (0) 0.012 3/21 (14.3) 0/25 (0) 0.100 2/17 (11.8) 1/29 (3.4) 1 Total ATG treatment, n (%) 2/11 (18.2) 7/35 (20.0) 1.000 5/21 (23.8) 4/25 (16.0) 0.711 4/17 (23.5) 5/29 (17.2) 0.707 Maintenance immunosuppression Steroid, (no/daily/alternate day) 0/4/7 2/6/27 0.339 0/6/15 3/4/19 0.412 0/4/13 2/6/21 0.740 MMF/MPA, n (%) 11/11 (100) 34/35 (97.1) 1.000 21/21(100) 24/25 (96.0) 1.000 16/17 (94.1) 29/29 (100) 0.370 MMF dose, mg/m2/day 654 (144) 666 (177) 0.687 655 (61) 666 (244) 0.333 650 (411) 671 (159) 0.661 Tacrolimus dose, mg/kg/day 0.05 (0.06) 0.07 (0.04) 0.963 0.08 (0.05) 0.06 (0.05) 0.063 0.08 (0.06) 0.07 (0.05) 0.847 Tacrolimus through level, ng/ml 5.8 (2.0) 5.3 (1.6) 0.646 5.4 (1.9) 5.3 (1.3) 0.708 5.8 (2.1) 5.3 (1.4) 0.162 Tacrolimus, n (%) 10/11 (90.9) 30/35 (85.7) 1.000 20/21 (95.2) 20/25 (80.0) 0.198 16/17 (94.1) 24/29 (82.8) 0.390 Blood tests White blood cells, × 103 U/µl 5.6 (3.9) 6.9 (3.7) 0.547 6.8 (4.7) 6.7 (2.1) 0.854 7.5 (5.0) 6.6 (2.5) 0.950 Lymphocytes, × 103 U/µl 2.4 (1.1) 2.4 (0.8) 0.255 2.4 (1.0) 2.2 (0.8) 0.845 2.6 (1.5) 2.4 (0.7) 0.753 Creatinine, mg/dl 1.16 (0.4) 1.14 (0.4) 0.014 1.0 (0.4) 0.97 (0.4) 0.366 1.0 (0.3) 0.95 (0.4) 0.539 eGFR, ml/min per 1.73 m2 55.8 (28.8) 69.5 (22.1) 0.007 63.0 (17.3) 68.2 (26.7) 0.264 59.4 (19.2) 69.5 (26.8) 0.301 Data are given as median (interquartile range) and analyzed with the Mann–Whitney U test and categorical data are given as n/n (%) and analyzed with the Chi-square test or Fischer’s Exact test, where appropriate. Bold values indicate statistically significant p values (p < 0.05). nAb, neutralizing antibody; ATG, antithymocyte globulin; MMF, mycophenolate mofetil; MPA, mycophenolic acid; eGFR, estimated glomerular filtration rate. P values lower than 0.05 are given in bold The relationship between humoral and cellular immunity among COVID-19 naïve KTRs The distribution of humoral and cellular immune responses in the COVID-19 naïve KTRs is shown in Fig. 3. Out of 35 anti-SARS-CoV-2 IgG seropositive KTRs, 10 (28.5%) were nAb-negative. Four out of 11 anti-SARS-CoV-2 IgG seronegative KTRs were IGRA-positive. A complete immune response (positive anti-SARS-CoV-2 IgG, nAb, and IGRA) was observed in 20 KTRs (43.4%), whereas 7/46 KTRs (15.2%) showed no immune response at all. All immune parameters—anti-SARS-CoV-2 IgG, nAb activity, and IGRA titer levels—were significantly correlated with each other (p < 0.001 for all, Fig. 4).Fig. 3 The distribution of COVID-19 naïve kidney transplant recipients (KTRs) according to humoral and cellular immune responses to the mRNA vaccine. IGRA, interferon gamma release assay; nAb, neutralizing antibody Fig. 4 The correlations between A anti-SARS-CoV2 IgG titers, B neutralizing antibody (nAb) activities, and C interferon gamma release assay (IGRA) levels in COVID-19 naïve kidney transplant recipients. All parameters were significantly correlated with each other Comparison of immune response to SARS-CoV-2 between COVID-19 naïve and recovered KTRs COVID-19 recovered KTRs had significantly higher titers of both anti-SARS-COV-2 IgG and nAb compared to COVID-19 naïve KTRs (p = 0.018 and p = 0.007 respectively, Table 1). In terms of positivity, COVID-19 recovered KTRs had significantly higher nAb positivity (100% vs. 54.3% respectively, p = 0.003), but anti-SARS-CoV-2 IgG did not differ significantly (94.1% vs. 76.1% respectively, p = 0.155). Neither IGRA titer level nor IGRA positivity differed significantly according to COVID-19 history (p > 0.05 for both). During the study period, six out of 63 KTRs had COVID-19 after two doses of SARS-CoV-2 vaccine. Two of them had no humoral or cellular immune response to the vaccine. None of them experienced a severe disease needing hospitalization. Thirteen of 63 KTRs received a third dose of the vaccine. Only one of four anti-SARS-COV-2 IgG-seronegative KTRs and two of five IGRA-negative KTRs had a positive result after the third dose of the vaccine. Discussion In this prospective multicenter study, both humoral and cellular immune responses to the two doses of BNT162b2 mRNA COVID-19 vaccine were assessed in pediatric KTRs and compared with pediatric dialysis patients and healthy controls. The main findings of our study were that COVID-19 naïve KTRs have significantly lower levels of anti-SARS-CoV-2 IgG titer and nAb activity compared to both dialysis and control groups, demonstrating lower vaccine-induced humoral immunity among KTRs. Shorter time on transplantation and higher eGFR were independently associated with anti-SARS-CoV-2 seropositivity in the KTRs. Furthermore, COVID-19 naïve KTRs had significantly lower IGRA levels than dialysis patients. They also demonstrated a trend toward lower IGRA levels than controls, but the difference did not reach statistical significance. COVID-19 recovered KTRs had significantly higher anti-SARS-CoV-2 IgG titer and nAb activity levels compared to COVID-19 naïve KTRs, but IGRA titers did not differ significantly. These findings demonstrated the booster effect of natural SARS-CoV-2 infection on humoral immunity, but not on cellular immunity. Dialysis patients demonstrated similar humoral and cellular immune response to the SARS-CoV-2 mRNA vaccine compared to the healthy individuals, which is similar to data in adult cohorts [15, 16, 20]. The reported prevalence of a positive humoral immune response to SARS-CoV-2 mRNA vaccines varies widely in adult KTRs due to differences in study protocols, established cut-off values, and sensitivity of different assays. The prevalence of seroconversion in KTRs has been reported to range from 36 to 63%, which is significantly lower compared to both CKD, dialysis, and healthy individuals [10–16]. Younger age, lower MMF dose, low tacrolimus trough level, and higher eGFR have been reported to be associated with improved humoral immune response [10–12, 15–17], whereas shorter time on transplantation, especially post-transplant first year, has been associated with a negative humoral immune response [15, 16]. There are few studies conducted in pediatric KTRs. Haskin et al. [18] demonstrated a 63% seroconversion rate in 38 adolescents and young adults with a mean age of 18 years after two doses of BNT162b2 mRNA vaccine. The authors reported that seroconverted KTRs had a significantly lower use of rituximab and a longer time after the second vaccine dose compared to seronegative KTRs. Crane et al. [19] have reported a 52% seroconversion rate after two doses of BNT162b2 mRNA vaccine in 25 adolescents with a median age of 19 years. The authors reported a higher number of KTRs on MMF and higher doses of MMF use in the non-responders. In the present study, the seroconversion rate was 76.1% in COVID-19 naïve KTRs. This higher prevalence of seroconversion compared to previous studies among adolescents and adults may be partly explained by younger age of our cohort or by the different assays used in these studies. Consistent with the previous studies, anti-SARS-CoV-2 IgG seropositivity in our cohort was associated with higher eGFR in KTRs. Lower eGFR does not explain lower immunity against SARS-CoV-2 vaccination as dialysis patients with much lower eGFR had significantly higher anti-SARS-CoV-2 IgG than KTRs. Three of the eleven anti-SARS-CoV-2 seronegative KTRs were given rituximab due to acute rejection. These patients had low eGFR and one of them had hypogammaglobulinemia. Although rituximab or acute rejection history did not remain in the regression analysis, they may have an effect on eGFR and seropositivity association. The seronegative group was small; larger cohorts are needed to assess this association. In contrast to the published reports, shorter time on transplantation was associated with seropositivity in the present study. However, it is important to note that in our cohort none of the KTRs had a shorter transplant duration than 12 months. Although the seroconversion rate seems higher than previous studies, COVID-19 naïve KTRs had still significantly lower anti-SARS-CoV-2 IgG titers compared to both dialysis and control groups. Neutralizing antibody activity indicates the functional antibodies that can inhibit SARS-CoV-2 infection; in other words, it represents the clinical efficacy of vaccine-induced measured antibodies [21]. Lower nAb titers have been reported in adult KTRs compared with healthy controls [16, 22], but pediatric data is not yet available. The prevalence of nAb positivity has been reported to range from 31 to 65.8% in adult studies including all KTRs. On the other hand, in the studies including only seropositive adult KTRs, this prevalence has been reported as high as 79.3%, which is still lower than in healthy controls [13, 16, 22]. In our cohort, COVID-19 naïve KTRs had significantly lower nAb activity compared to both dialysis and control groups. The prevalence of nAb positivity was also significantly lower in COVID-19 naïve KTRs than controls (54.3% vs. 100%). In line with the literature, there was a strong correlation between the titer of anti-SARS-CoV-2 IgG and nAb activity [16]. The frequency of a negative nAb among anti-SARS-CoV-2 IgG seropositive KTRs was about 30% in the present study, which has been reported as 10% by Pedersen et al. [22]. These findings demonstrate that not only seropositivity but also titer levels of anti-SARS-CoV-2 IgG are important to predict protection from COVID-19 in KTRs. It is known that repeated vaccination may not elicit a humoral response but a cellular immune response. The prevalence of a positive cellular immune response to an mRNA vaccine has been reported to range between 16.2 and 60% in adult KTRs [15, 16, 23, 24]. The current study was the first to investigate the cellular immune response in pediatric KTRs after SARS-CoV-2 vaccination. The prevalence of a positive cellular response was 63% in COVID-19 naïve KTRs, which was quite low compared with both dialysis patients (100%) and healthy controls (81.8%). Both the positivity and titer levels of IGRA were significantly lower compared to dialysis patients, but not from controls. However, the number of COVID-19 naïve healthy controls was low. Four of eleven seronegative KTRs had positive cellular immunity. We could not detect any clinical or laboratory factors affecting cellular immune response. The effect of natural COVID-19 on immunity in vaccinated KTRs was assessed in adults by Magicova et al. [25]. They demonstrated a higher prevalence of seroconversion after two doses of BNT162b2 or Moderna mRNA-1273 vaccine, in KTRs with a COVID-19 history, than COVID-19 naïve vaccinated KTRs (97% vs. 40%). In our study, COVID-19 recovered KTRs had significantly higher titers of anti-SARS-CoV-2 IgG and nAb activity compared to COVID-19 naïve KTRs. These results demonstrate the booster effect of natural infection on humoral immunity. Although the IGRA positivity was higher in COVID-19 recovered KTRs than COVID-19 naïve KTRs (100% vs. 81%), the difference did not reach statistical significance. Magicova et al. [25] also demonstrated a better cellular immune response in previously infected, vaccinated adult KTRs compared with naïve vaccinated KTRs (90% vs. 9%). This study demonstrated a trend toward an improved cellular immune response at higher anti-SARS-CoV-2 IgG titers. The seroconversion rate appears to be high, but a complete immune response, i.e., positive nAb and cellular immune response in addition to seroconversion, was present in about 40% of the KTRs. This may be indicative of the lower clinical efficacy of the SARS-CoV-2 mRNA vaccine in pediatric KTRs. These results suggest that booster vaccination and/or possibly an increase in vaccine dose is needed, similar to HBV vaccination in CKD patients. In our cohort, only 13 KTRs received a third dose of the vaccine during the study period. One out of four anti-SARS-CoV-2 IgG seronegative and two out of five IGRA-negative KTRs were positive after a third dose of vaccine. It is difficult to draw significance from this, given the small sample size. Nevertheless, results from the adult studies demonstrate an enhanced humoral and cellular immune response after the third dose of mRNA vaccine in KTRs [23, 26]. Our study has several limitations. Sampling was planned after 4 weeks following the second vaccine dose; however, the delay in study approval resulted in heterogenous timing of blood sampling with a median 8 weeks in KTRs. Nevertheless, timing was not significantly different from dialysis or control groups. Secondly, due to this delay in the study start, we missed the prevaccination sampling to measure anti-SARS-CoV-2 IgG to determine natural SARS-CoV-2 infection. Therefore, we defined natural SARS-CoV-2 infection with a previously positive PCR test, which may result in asymptomatic cases being missed. Lastly, the number of dialysis and control groups was small. Although KTRs had lower IGRA levels and positivity than controls, the difference was not statistically significant. This finding can be explained by the small sample size for controls and the sensitivity of the assay. The strength of our study is that we assessed not only seroconversion but also detailed immune analyses, including SARS-CoV-2-specific nAb and cellular immune responses to mRNA vaccination, in a relatively high number of KTRs. In conclusion, the humoral and cellular immune response after two doses of SARS-CoV-2 mRNA vaccination appears to be better in pediatric KTRs than in adult KTRs, whereas the immune response is still lower compared to healthy children. In particular, KTRs with longer transplant duration and lower eGFR have a lower humoral immune response, whereas natural SARS-CoV-2 infection has a booster effect on the humoral immune response. Although seroconversion prevalence appears to be high, only about 40% of the KTRs have both a positive nAb and T-cell immune response in addition to seroconversion, which may demonstrate the need for booster doses or an increase in vaccine dose. Further prospective studies are required to demonstrate the clinical efficacy of the SARS-CoV-2 mRNA vaccine-induced immune response in KTRs. Supplementary information Below is the link to the electronic supplementary material.Graphical Abstract (PPTX 181 kb) Acknowledgements We are thankful to Dr Omer Deniz Ocakli for assistance with data entry. Author’s contribution Conceptualization: Salim Caliskan; methodology: Salim Caliskan, Bekir Kocazeybek, Ayca Kıykım; data collection: Ruveyda Gulmez, Bagdagul Aksu, Nurdan Yıldız, Diana Uckardes, Seha Saygılı, Esra Karabag Yılmaz, Zeynep Yuruk Yıldırım, Mehmet Tasdemir, Ahmet Nayır; material preparation and analysis were performed by Ruveyda Gulmez, Dogukan Ozbey; statistical analysis: Ayse Agbas; writing—original draft preparation: Ruveyda Gulmez, Ayse Agbas; writing—review and editing: Nur Canpolat, Salim Caliskan; supervision: Salim Caliskan, Bekir Kocazeybek, Haluk Cokugras. All authors read and approved the final manuscript. Funding This work was supported by Scientific Research Projects Coordination Unit of Istanbul University-Cerrahpasa (35866). Data availability All data generated or analyzed during this study are included in this published article and its supplementary information files. Declarations Ethics approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Istanbul University-Cerrahpasa, Cerrahpasa School of Medicine (2021–70493). Consent Written informed consent was obtained from the parents and participants where available. Competing interests The authors declare no competing interests. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Chung EYM Palmer SC Natale P Krishnan A Cooper TE Saglimbene VM Ruospo M Au E Jayanti S Liang A Jie Deng DJ Chui J Higgins GY Tong A Wong G Teixeira-Pinto A Hodson EM Craig JC Strippoli GFM Incidence and outcomes of COVID-19 in people with CKD: a systematic review and meta-analysis Am J Kidney Dis 2021 78 804 815 10.1053/j.ajkd.2021.07.003 34364906 2. Jager KJ Kramer A Chesnaye NC Couchoud C Sanchez-Alvarez JE Garneata L Collart F Hemmelder MH Ambuhl P Kerschbaum J Legeai C Del Pino YPMD Mircescu G Mazzoleni L Hoekstra T Winzeler R Mayer G Stel VS Wanner C Zoccali C Massy ZA Results from the ERA-EDTA Registry indicate a high mortality due to COVID-19 in dialysis patients and kidney transplant recipients across Europe Kidney Int 2020 98 1540 1548 10.1016/j.kint.2020.09.006 32979369 3. Ng JH Hirsch JS Wanchoo R Sachdeva M Sakhiya V Hong S Jhaveri KD Fishbane S Northwell COVID-19 Research Consortium and the Northwell Nephrology COVID-19 Research Consortium Outcomes of patients with end-stage kidney disease hospitalized with COVID-19 Kidney Int 2020 98 1530 1539 10.1016/j.kint.2020.07.030 32810523 4. Seidel M Holzer B Appel H Babel N Westhoff TH COVID Dialysis Working Group Impact of renal disease and comorbidities on mortality in hemodialysis patients with COVID-19: a multicenter experience from Germany J Nephrol 2020 33 871 874 10.1007/s40620-020-00828-8 32804355 5. Canpolat N Yildirim ZY Yildiz N Tasdemir M Goknar N Evrengul H Gulmez R Aksu B Dursun H Ozcelik G Yavascan O Cicek RY Tulpar S Hacihamdioglu DO Nayir A Alpay H COVID-19 in pediatric patients undergoing chronic dialysis and kidney transplantation Eur J Pediatr 2022 181 117 123 10.1007/s00431-021-04191-z 34218318 6. Saygili S Canpolat N Cicek RY Agbas A Yilmaz EK Sakalli AAK Aygun D Akkoc G Demirbas KC Konukoglu D Cokugras H Caliskan S Sever L Clinical and subclinical acute kidney injury in children with mild-to-moderate COVID-19 Pediatr Res 2022 10.1038/s41390-022-02124-6 35681092 7. Gasim GI Bella A Adam I Immune response to hepatitis B vaccine among patients on hemodialysis World J Hepatol 2015 7 270 275 10.4254/wjh.v7.i2.270 25729482 8. Nailescu C Xu X Zhou H Hall H Wilson AC Leiser JD Chand DH Valentini RP Hebert D Mahan JD Influenza vaccine after pediatric kidney transplant: a Midwest Pediatric Nephrology Consortium study Pediatr Nephrol 2011 26 459 467 10.1007/s00467-010-1729-1 21181206 9. Reddy S Chitturi C Yee J Vaccination in chronic kidney disease Adv Chronic Kidney Dis 2019 26 72 78 10.1053/j.ackd.2018.10.002 30876620 10. Benotmane I Gautier-Vargas G Cognard N Olagne J Heibel F Braun-Parvez L Martzloff J Perrin P Moulin B Fafi-Kremer S Caillard S Low immunization rates among kidney transplant recipients who received 2 doses of the mRNA-1273 SARS-CoV-2 vaccine Kidney Int 2021 99 1498 1500 10.1016/j.kint.2021.04.005 33887315 11. Marinaki S Adamopoulos S Degiannis D Roussos S Pavlopoulou ID Hatzakis A Boletis IN Immunogenicity of SARS-CoV-2 BNT162b2 vaccine in solid organ transplant recipients Am J Transplant 2021 21 2913 2915 10.1111/ajt.16607 33864722 12. Rozen-Zvi B Yahav D Agur T Zingerman B Ben-Zvi H Atamna A Tau N Mashraki T Nesher E Rahamimov R Antibody response to SARS-CoV-2 mRNA vaccine among kidney transplant recipients: a prospective cohort study Clin Microbiol Infect 2021 27 1173 e1171 1173 e1174 10.1016/j.cmi.2021.04.028 13. Stumpf J Siepmann T Lindner T Karger C Schwobel J Anders L Faulhaber-Walter R Schewe J Martin H Schirutschke H Barnett K Huther J Muller P Langer T Pluntke T Anding-Rost K Meistring F Stehr T Pietzonka A Escher K Cerny S Rothe H Pistrosch F Seidel H Paliege A Beige J Bast I Steglich A Gembardt F Kessel F Kroger H Arndt P Sradnick J Frank K Klimova A Mauer R Grahlert X Anft M Blazquez-Navarro A Westhoff TH Stervbo U Tonn T Babel N Hugo C Humoral and cellular immunity to SARS-CoV-2 vaccination in renal transplant versus dialysis patients: a prospective, multicenter observational study using mRNA-1273 or BNT162b2 mRNA vaccine Lancet Reg Health Eur 2021 9 100178 10.1016/j.lanepe.2021.100178 34318288 14. Eckerle I, Rosenberger KD, Zwahlen M, Junghanss T (2013) Serologic vaccination response after solid organ transplantation: a systematic review. PLoS ONE 8:e56974. 10.1371/journal.pone.0056974 15. Crespo M Barrilado-Jackson A Padilla E Eguia J Echeverria-Esnal D Cao H Faura A Folgueiras M Sola-Porta E Pascual S Barbosa F Hurtado S Ribera L Rio-No L Perez-Saez MJ Redondo-Pachon D Pascual J Negative immune responses to two-dose mRNA COVID-19 vaccines in renal allograft recipients assessed with simple antibody and interferon gamma release assay cellular monitoring Am J Transplant 2022 22 786 800 10.1111/ajt.16854 34551187 16. Sanders JF Bemelman FJ Messchendorp AL Baan CC van Baarle D van Binnendijk R Diavatopoulos DA Frolke SC Geers D GeurtsvanKessel CH den Hartog G van der Heiden M Imhof C Kho MML Koopmans MPG Malahe SRK Mattheussens WB van der Molen R van Mourik D Remmerswaal EBM Rots N Vart P de Vries RD Gansevoort RT Hilbrands LB Reinders MEJ RECOVAC Collaborators The RECOVAC immune-response study: the immunogenicity, tolerability, and safety of COVID-19 vaccination in patients with chronic kidney disease, on dialysis, or living with a kidney transplant Transplantation 2022 106 821 834 10.1097/TP.0000000000003983 34753894 17. Boyarsky BJ Werbel WA Avery RK Tobian AAR Massie AB Segev DL Garonzik-Wang JM Antibody response to 2-dose SARS-CoV-2 mRNA vaccine series in solid organ transplant recipients JAMA 2021 325 2204 2206 10.1001/jama.2021.7489 33950155 18. Haskin O Ashkenazi-Hoffnung L Ziv N Borovitz Y Dagan A Levi S Koren G Hamdani G Levi-Erez D Landau D Alfandary H Serological response to the BNT162b2 COVID-19 mRNA vaccine in adolescent and young adult kidney transplant recipients Transplantation 2021 105 e226 e233 10.1097/TP.0000000000003922 34381004 19. Crane C Phebus E Ingulli E Immunologic response of mRNA SARS-CoV-2 vaccination in adolescent kidney transplant recipients Pediatr Nephrol 2022 37 449 453 10.1007/s00467-021-05256-9 34522992 20. Swai J Gui M Long M Wei Z Hu Z Liu S Humoral and cellular immune response to severe acute respiratory syndrome coronavirus-2 vaccination in haemodialysis and kidney transplant patients Nephrology (Carlton) 2022 27 7 24 10.1111/nep.13974 34510645 21. Khoury DS Cromer D Reynaldi A Schlub TE Wheatley AK Juno JA Subbarao K Kent SJ Triccas JA Davenport MP Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection Nat Med 2021 27 1205 1211 10.1038/s41591-021-01377-8 34002089 22. Pedersen RM Bang LL Tornby DS Kierkegaard H Nilsson AC Johansen IS Bistrup C Jensen TG Justesen US Andersen TE The SARS-CoV-2-neutralizing capacity of kidney transplant recipients 4 weeks after receiving a second dose of the BNT162b2 vaccine Kidney Int 2021 100 1129 1131 10.1016/j.kint.2021.09.006 34547366 23. Bertrand D Hamzaoui M Lemee V Lamulle J Laurent C Etienne I Lemoine M Lebourg L Hanoy M Le Roy F Nezam D Farce F Plantier JC Boyer O Guerrot D Candon S Antibody and T-cell response to a third dose of SARS-CoV-2 mRNA BNT162b2 vaccine in kidney transplant recipients Kidney Int 2021 100 1337 1340 10.1016/j.kint.2021.09.014 34619232 24. Boedecker-Lips SC Lautem A Runkel S Klimpke P Kraus D Keil P Holtz S Tomalla V Marczynski P Boedecker CB Galle PR Koch M Weinmann-Menke J Six-month follow-up after vaccination with BNT162b2: SARS-CoV-2 antigen-specific cellular and humoral immune responses in hemodialysis patients and kidney transplant recipients Pathogens 2022 11 67 10.3390/pathogens11010067 35056015 25. Magicova M Zahradka I Fialova M Neskudla T Gurka J Modos I Hojny M Raska P Smejkal P Striz I Viklicky O Determinants of immune response to anti-SARS-CoV-2 mRNA vaccines in kidney transplant recipients: a prospective cohort study Transplantation 2022 106 842 852 10.1097/TP.0000000000004044 34999659 26. Stumpf J Tonnus W Paliege A Rettig R Steglich A Gembardt F Kessel F Kroger H Arndt P Sradnick J Frank K Tonn T Hugo C Cellular and humoral immune responses after 3 doses of BNT162b2 mRNA SARS-CoV-2 vaccine in kidney transplant Transplantation 2021 105 e267 e269 10.1097/TP.0000000000003903 34342963	36459243	PMC9716124	NO-CC CODE	2022-12-03 23:20:54	no	Pediatr Nephrol. 2022 Dec 2;:1-10	utf-8	Pediatr Nephrol	2,022	10.1007/s00467-022-05813-w	oa_other
==== Front Asian J Bus Ethics Asian Journal of Business Ethics 2210-6723 2210-6731 Springer Netherlands Dordrecht 162 10.1007/s13520-022-00162-1 Article Gamification and customer experience in online retail: a qualitative study focusing on ethical perspective Sheetal [email protected] 1 Tyagi Rimjim 2 Singh Gursimranjit 3 1 grid.512249.9 0000 0004 1764 8954 Jaipuria Institute of Management, Uttar Pradesh, Ghaziabad, Lucknow, India 2 grid.449005.c Mittal School of Business, Lovely Professional University, Phagwara, Punjab India 3 grid.444475.2 0000 0004 1767 2962 Dr. B. R. Ambedkar National Institute of Technology, Jalandhar, Punjab India 2 12 2022 121 2 9 2022 16 11 2022 © The Author(s), under exclusive licence to Springer Nature B.V. 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This paper aims to investigate the effect of gamification in engaging and motivating consumers for online shopping and also the use of gamification to enhance sales. Moreover, this study has also explored the ethical concerns in gamified marketing. This is a qualitative study to investigate the effect of gamification during online shopping and the ethical issues involved in gamified marketing. Semi-structured interviews with ten gamification experts are conducted and analyzed through NVivo. The themes that emerged from qualitative analysis are the applicability of gamification in online retail, consumer experiences in gamified retail, and ethics and challenges in gamification. Semantic analysis is performed, and as per the viewpoint of the gamification experts, it was found that the perception of ethics in gamification is negative, which shows that there are many unethical practices in gamified marketing. This paper shows that by focusing on every relational aspect of consumer engagement, retailers can build trust and retain their most valuable stakeholders — the customers, thereby addressing the crucial negative concerns of gamified marketing. This research is one of its types to explore the significant ethical issues that affect consumers in the retail context. The undertaking of this study in an emerging economy adds further insight into gamified retail literature by generalizing the applicability of gamified studies across geographic contexts. Keywords Gamification Covid-19 Retail Customer engagement Motivation Marketing Ethics ==== Body pmcIntroduction Covid-19 has dramatically disrupted the retail sector, with the shock differing massively between brick-and-mortar stores versus online shops, essential versus non-essential stores, and small versus large retailers (Ali, 2020). The virus gradually slows down the economy and leads to recession. Social distancing is the only effective way to eliminate the coronavirus pandemic. The business environment is changing rapidly under the threat of Covid-19, and it will continue until the virus is not eliminated. The pandemic causes many problems for retailers. Also, Covid-19 forced consumers to slow their spending ability. Hence, the retail market faced many challenges as the retail market is hammered by both sides, the consumers and the government (Szász et al., 2022). Retailers are facing the struggle to cope with the ongoing crisis and competition (Thukral, 2021). E-commerce is now the best hope for retailers to survive the Covid-19 crisis. The pandemic and work-from-home concept changed consumers’ purchasing habits and shopping manners as well (Sheth, 2020). Online shopping was not everyone’s choice before the pandemic, but now they are forced to do so. With the threat of this crisis, consumers want to shop from the comfort of their homes, but retailers need to give them the experience to stay in the market. The Covid-19 impact on the retail sector has largely been dependent on lockdown and social distancing (Donthu & Gustafsson, 2020). According to a Nielsen report, the e-commerce market share increased to almost 10% of sales compared to 6% less in 2019. Due to recent technological advancements, gamification has emerged: non-monetary incentives take a boost in every sphere of human life which is generally a motivational technique using video game elements, such as digital points, and badges. Moreover, it is an innovative approach to using game mechanics in a non-gaming context and, in retailing also, gamification is being used (Deterding et al., 2011; Insley & Nunan, 2014). Gamification in online retail draws the attention of the consumer. According to Harvard Business Review, a 5% increase in customer attention leads to a 25–30% rise in earnings. Gamification is a valuable strategy for enticing customers, and over 87% of merchants want to employ it in the next 5 years. Gamification makes shopping fun and attractive by increasing motivation, engagement, and loyalty (Insley & Nunan, 2014; Karać & Stabauer, 2017). Now, retailers started to enhance the online customer experience by implementing game elements and mechanics in e-commerce to increase customer engagement on their sites (Insley & Nunan, 2014). Gamification is assumed to be a critical factor in market strategies (Hamari et al., 2014). Every e-commerce website and retailer have several opportunities to use gamification. Through gamification, feeling, satisfaction, and relatedness are promoted and they significantly enhance engagement (Pasca et al., 2021). Because games are intended to be entertaining, they may provide good experiences and encourage users to stay engrossed in the game for an extended amount of time (Deterding et al., 2011). In order to create such a condition, marketers would greatly profit from properly incorporating elements of game design into their product/service. However, researchers and marketers have struggled to provide an effective response addressing when and how gamification should be employed since the notion of gamification was created (Gatautis et al., 2016). Although there are numerous examples of businesses that have successfully incorporated game elements into their value propositions — simple loyalty programs like Starbucks and more elaborate gamified systems like Foursquare come to mind — there is no single solution or framework that can reliably deliver on this promise to replicate these successes (Hamari et al., 2014; Nacke & Deterding, 2017). Before the pandemic, many did online shopping, but they did not believe it was convenient. Gamification can add several features to give great experience like try-on, awareness, transparency, and many more which will significantly increase the visit rate on the retailer’s website (De Canio et al., 2021a, 2021b). In addition to this, gamification also motivates users to perform a specific activity which brings enhanced customer experience (Minge & Cymek, 2020). According to Harvard Business Review, a rise in customer attention by 5% generates a 25–95% increase in profit. Gamification appears to be an excellent tactic to enhance sales, brand awareness, and customer loyalty. The introduction of gamification in the professional environment has paternalistic qualities that have the same challenges and dangers as paternalism in digital nudging, resulting in the same problems and restricting the users’ autonomy and freedom of choice (Hassan & Hamari, 2020). The purpose of this study is to identify the ethical concerns involved with gamification originating in gaming and video games; the dynamics and hazards of these settings are partially translated to gamification. Moreover, the word “gamification” is still relatively new, so there is paucity of literature, particularly at the online retail level. This study contributes to the investigation of ethical concerns by demonstrating several unethical activities in gamification marketing that undermine customers’ trust and experiences. This paper aims at strengthening the literature related to retailing and marketing as companies marketing strategies are increasingly incorporating fast-trending gamification (Yang et al., 2017). Therefore, the study aims to develop a conceptual framework that focuses on exploring the effect of gamification on consumer engagement in the context of online retail during the coronavirus pandemic. The viewpoint of gamification experts about unethical gamification is vital as they enlighten designers and marketers about when and how this affects them, and, perhaps more significantly, when they provide these kinds of experiences how consumers react to a particular game. This might result in the more successful deployment of gamification from the standpoints of both the consumer and the marketers. Literature review Gamification mechanics Gamification is a marketing and business strategy applied to increase customer engagement and loyalty, and gamification can influence customer engagement and encourage behaviors, supporting and motivating users toward tasks (Hsu & Chen, 2018). First, when merchants want to include gamified mechanisms in their mobile retailing applications, they should do it through challenge levels, just as in video games (Aydınlıyur et al., 2021). The app, like WeChat, is gradually expanding functions; this strategy has enabled the app to become China’s first multipurpose app, with users spending more than 360 min each day on it. Gamification is a marketing and business strategy applied to increase customer engagement and loyalty, and gamification can influence customer engagement and encourage behaviors, supporting and motivating users to the task. Gamification enhanced consumers’ knowledge, attitudes, behavior, and reward-based game elements to enhance sustainable behavior outcomes. Organizations can use serious game elements to promote desirable and sustainable behavior (Shevchuk et al., 2019). Gamification technology can change desired behavior by increasing consumer loyalty awareness and developing an eco-friendly mindset. Game design enhances the efficiency to change behavior. Sometimes, if game design is inappropriate, it reduces replayability or brings a lack of fun. For applying games, every aspect of context and mechanism needs to consider to achieve desired behavior change outcomes (Haque et al., 2014). For designing a game, a person who needs the context and target players who game well should be tested in the real world to lead to better outcomes. For engagement, consider the fun and engagement of players because, without engagement, there will be no desired result (Epstein et al., 2021). A gamification platform is developed to motivate behavioral change by increasing awareness and consumer engagement using a pervasive application that analyzes context, sends personalized messages, and manages gamification peer competition and feedback (Soares et al., 2021). Gamification contributes to the value creation of customers in the retail context. When gamification applies in an activity, it affects the hedonic value, which is positive. Gamified motivators lead to a psychological outcome that changes consumer behavior (Gatautis et al., 2021). The satisfaction of hedonic value is better than the reward. Hedonic value provides better-continued engagement. Gamification with continued engagement is positively associated with brand engagement (Hogberg et al., 2019). Gamification and retail experiences Gamification in online retail contains the potential to affect a meaningful set of outcomes for service firms throughout the consumer decision process (Hofacker et al., 2016). The retail landscape adopting innovative technologies to stay ahead as aggressive competition prevails within the industry. Recently, retail firms started to employ several technologies including gamification. Gamification among others carries out marketing efforts efficiently (Jayasooriya et al., 2020). Encouraging customers to enter a data disclosure process is a challenge for retailers. Retailers facing customers with low trust levels need to enhance the design of their data requests. Relevance illustrating game elements amplify the positive effect and increase hedonic and meaningful engagement (Bidler et al., 2020). Consumer brand stickiness will grow if gamified marketing is executed successfully. Fun is a crucial component of gamified marketing; in the activities thereof, the information and prizes provided by salespeople must pique customers’ initial attention before allowing them to participate in the activities. The mechanisms design generates incentives such as enjoyment and attraction and gives distinct feedback so that customers may enjoy a fascinating game and join into the competition unwittingly in quest of glory (Lu & Hu, 2020). Gamification is a motivating strategy for increasing user involvement. Indeed, the key to gamification success is emotionally engaging individuals and inspiring them to attain their goals. Human motivation can be classified into two types: intrinsic and extrinsic. Extrinsic motivation is driven by an external element that motivates a person to perform something in the expectation of receiving a reward. On the other hand, intrinsic motivation refers to action driven by internal incentives such as enjoyment or pleasant sensations (Ushakov & Shatila, 2021). When combined with monetary rewards, gamification loses its efficiency and positive effect because players do not get satisfaction and enjoyment, but the extrinsic motivation to receive a discount shopping-related game increases satisfaction with the retailer and loyalty (Bauer et al., 2020b). If gamification and social interaction combined with marketing show much potential, they help promote user engagement and influence user behavior. Gamification and social networks strengthen the bond between consumers and merchants. Gamification can increase user participation in online shopping. Social interaction motivates users to participate in the use of the gamified system (Gajanova & Radisic, 2021). Nowadays, the gamification approach sparked interest as a new trend of increasing customer engagement in business-to-consumer contexts like online retail. They identify that many game mechanisms used in designing games, among which are three key factors, feedback, gift, and challenges, are more effective in customer engagement. The access platform and gaming device during gamification play an important role. Most people play games on their smartphones. The interaction between accessing the games and gaming devices affects the brand attitude and memory of the consumer differently. However, consumer engagement and flow experience mediate access platform and gaming device effect on brand memory and attitude (Sreejesh et al., 2021). Advertising gamification increases consumer engagement and motivation. An interactive gamified experience increases attitude, brand image, and brand attitude (Xu et al., 2020). In this regard, as noted in the research (Insley & Nunan, 2014), harnessing customers’ inherent motives might lead to obtaining consumers’ preferences. Furthermore, by implementing communication campaigns to clarify and explain how consumers may purchase online, businesses can increase the level of online channel knowledge of their current and prospective customers, reinforcing their engagement, enriching previous online shopping experiences, and increasing their proneness to online purchases via mobile devices (Dwivedi et al., 2021). Mobile apps with gamified components can thus be more versatile, convenient, dynamic, and entertaining. Indeed, gamification is part of a larger company’s plan to increase platform-user engagement, and social and gamified cues should be deployed consistently by the platform to have an evident influence on end users. On the other hand, gamification may result in a shift in marketing techniques (Canio et al., 2021a, 2021b). Implementation of the gamification approach gives benefits to support behavior change. A survey shows that gamification leads to a positive effect on behavior change (Zain et al., 2021). User engagement meant a desirable and essential human response to computer-mediated activities like gamification. The user engagement concept has two views, deep engagement and meaningful engagement (Suh et al., 2017). Casual games motivate learners to engage in the online learning platform. They use two groups of participants, one of whom can play a casual game for a maximum of 5 min every time they log in (Gooch et al., 2016). The other group did not have the option to play a game at any time. They are playing casual games to increase their level of engagement. Also, the group that played the game answered most answers correctly and logged in significantly instead to those in the non-game condition. It concludes that gamified activity motivates learners to come back to play a casual game (Kapp et al., 2020). Dynamic pricing is growing due to various factors: demand data, technology, and decision support tools. A survey was conducted to understand retail and consumer perspectives on dynamic pricing and gamification. Retail companies have to prepare for the changes in logistics due to dynamic pricing in traditional ways. Retail stores are ensuring the supply chain according to demands (Guvenc et al., 2020). Prior service-related research has demonstrated that gamification is an effective technique for increasing customer loyalty, encouraging positive word of mouth, and increasing engagement with the given service. Furthermore, providing hedonic values (one of gamification’s key foundations) has proven to be a successful technique for engaging customers and encouraging repeat purchase intentions (Shi et al., 2022). Rewards, competition, achievement, challenges, and gaining knowledge represent consumers’ motives for using the app. The use of gamified applications influenced users’ attitudinal loyalty toward the core service (Kunkel et al., 2021). Ethics There is a huge debate among researchers about moral issues and business ethics in marketing gamification. Ethics is defined as a “collection of beliefs and rules that assist us in judging whether behavior helps or damages sentient creatures” in this context (Paul et al., 2003). The application of ethics by Andrzej Marczewski to gamification designers and the systems that employ such applications are also beneficial (Marczewski, 2017). He further enunciated that the focus here is on the aim of the gamification designer to develop systems that benefit rather than hurt others while defining damage is somewhat subjective. The ethics of gamification has been studied more as a practice-based approach to persuasion in the extant literature (Deterding et al., 2011). Thus, we choose a normative approach, which we think is best suited to investigating ethics in action and may help organizations choose ethical courses of conduct (Michalos, 1995). Personalized digital marketing, for example, works by tracking internet surfing histories via cookies. However, it raises concerns about how informed viewers are about this process (Álvarez-Bermejo et al., 2016). Gamification ethics can be intellectually located within this tradition of technology ethics as part of a further sub-discipline called applied ethics, which seeks to apply theories, normative standards, concepts, and methods developed within ethics to, for example, inquiries about specific technologies. Gamification ethics is a new yet fast-expanding discipline that seeks to investigate ethical challenges presented by gamification (Hyrynsalmi et al., 2017). In unethical gamification, the consumers are deceived for the greatest profit rather than persuading people into doing things, and this coercion is generally figured out by the customers when they are exposed to such an environment (Goethe, 2019). They would then struggle to remove the gamified environment and such offensive design and also prefer to communicate their displeasure with the design in a very public way (Goethe, 2019; Lowry et al., 2013). It is critical to take an ethical approach to gamification and imagine yourself as the user of what you are developing, and the design should embrace the intelligence of the users. After this, your users will automatically discuss the design with the rest of the world, and brand loyalty will emerge (Goethe, 2019). Gamification and ethics The introduction and use of persuasive technology to encourage a change of attitude and behavior, like other technologies, may have both good and bad consequences for the user (Thorpe & Roper, 2019). The ethical implications of gamification are rarely discussed. Gamification combines the entertaining world of games with the serious world of business, and as we will see, the collision of the two realms creates a number of normative tension points (Kagan, 2018; Thorpe & Roper, 2019). On a number of grounds, critics have called into doubt the moral validity of gamification. Given that gamification is one of the most rapidly disseminating behavioral techniques in business, there has been less serious research into the ethical implications of gamification in business than one might assume (Kim & Werbach, 2016; Goethe, 2019; Thorpe & Roper, 2019). Extrinsic rewards must be offered often when gamifying and employing extrinsic rewards to maintain motivation levels high. According to research, persons who had an inherent motivation from the start might be demotivated by gamification and felt like guinea pigs forced to respond to stimuli. Kim and Werbach (2016) stated that the ethical status of a gamification practice is mainly, but not only, decided by the extent to which the practice is exploitative, is manipulative, is purposefully or inadvertently harmful to the people involved, and has a socially unacceptable level. Manipulation Manipulation is defined as any behavior that infringes on the autonomy of users. It entails elements such as openness, consent, and self-reflection (Kim, 2021; Kim & Werbach, 2016). Manipulation is demonstrated by the following examples: companies/brands that do not reveal the content and aims of a gamified system because consumers would otherwise not engage (Kim, 2021; Kim & Werbach, 2016). Sometimes, the users may not consent to data gathering if online service providers do not explain privacy conditions or games that take use of players’ additive or obsessive habits in order to make it tough for users to quit playing (Kim, 2021; Kim & Werbach, 2016; Arora & Razavian, 2021; Kriz et al., 2022). Lastly is when gamification features and procedures are hidden from individuals to whom they are applied and when gamification tactics limit rational self-reflection and weaken autonomy in unacceptable ways (Kim, 2021; Vashisht & Sreejesh 2015; Kim & Werbach, 2016; Arora & Razavian, 2021; Kriz et al., 2022). Exploitation This concept refers to the employment of gamified components to push people to accomplish more than their position requires. It describes circumstances where an enterprise exploits the social environment to get more work done without a tangible reward for the customer (Kim, 20215; Kim & Werbach, 2016; Kriz et al., 2022). Business ethicists have started discussing the moral debates about the gamification of labor (Kim, 2021; Kim & Werbach, 2016; Vdov, 2020; Kriz et al., 2022). According to the lower part of the Octalysis framework: black hat, exploitation in gamification means manipulating the users, and the drivers of dissatisfaction and negative customer experience are scarcity, avoidance, and unpredictability (Vdov, 2020). Generally, a poorly designed gamified system is the biggest negative motivator and the reason behind this dissatisfaction (Kim & Werbach, 2016; Vdov, 2020). Harms Gamification providers have no intention of causing physical or psychological harm to participants. However, this might happen inadvertently when it comes to damaging behavior, the player may have the major ethical responsibility (Kim, 2021; Kim & Werbach, 2016; Kriz et al., 2022). Nonetheless, designers should try to predict or react to unforeseen negative events wherever feasible (Vdov, 2020). While these ethical concerns are not comprehensive, they are illustrative of the negative consequences that creators and users of gamification programs may have on users (Weiss, 2019). For example, within a game, it may be permissible to manipulate or deceive someone, but such practices are scarcely acceptable in the real world, and if a gamification approach that transposes a game-world norm to the real world is used, ethical difficulties may develop (Arora & Razavian, 2021). This study is an attempt to offer an analytical overview of the existing research in the field of gamification and customer experience in online retail with a special focus on the ethical perspective. Besides that, the present study is aimed to propose a conceptual framework to conduct empirical research to study the relationship between gamified mechanics and customer engagement and experience. On the basis of the literature review the following research questions are framed:o RQ-1- > How do gamified mechanics engage and motivate consumers in context to online retail? p RQ-2- > How do consumers react to gamified retailing experiences? q RQ-3- > What are the ethical concerns in the field of gamification? Research methodology This study used a qualitative method to understand online shopper behavior using gamified retail and consumer experience. Data was collected through semi-structured in-depth interviews by developing an extensive interview guide based on a literature review. Although it provided some structure for the interviews, participants could share other issues and ideas that were not included in the interview questions. The interviews were conducted with open-ended questions as shown in the Appendix. The guide consisted of different themes regarding how to use gamification themes including engagement, including retention, personalization, barriers, and duration of visits. After discussing their experiences, and what they liked and disliked, we asked for information about the gamification practices they used. We prompted them to share their thoughts about the game challenge design, including their perceived impacts. Judgment sampling, a form of non-probability sampling, was used to select the sample. The sample consisted of 10 participants for the interview. The participants were invited through LinkedIn with a written request to participate in the study mentioning the topic of the study. All the participants willingly gave their consent for the interview. The inclusion criterion for expert interviews was prior experience in gamification in connection to online retail apps in one or more of the following fields: game design, web design, user experience design (UX), and research. Interviews were taken online through voice and video calls. Interviews took an average of 30 min. The interviewees (a) were highly educated and (b) had experience with technology in general and gamification in particular. Participants were guaranteed strict confidentiality. The details of participants’ age, gender, and education level were obtained during the interview as shown in Table 1.Table 1 Demographics of respondents ID Age Date Education level Gender Place Profession R1 33 12/01/2022 Doctorate Female Lucknow UX Researcher R2 35 19/12/2021 Post graduate Male Karnataka Program Manager R3 24 19/12/2021 Post graduate Female Lucknow Associate Product Manager R4 30 14/12/2021 Under graduate Male Bihar CEO-Gamification Labs R5 45 18/10/2021 Post graduate Male Mumbai Head Gamification R6 30 14/10/2021 Post graduate Female Mumbai Social Media Marketing Specialist R7 38 21/12/2021 Post graduate Male Colombia Gamification Designer R8 40 20/12/2021 Post graduate Male Hungary Business Stimulation R9 41 15/12/2021 Post graduate Male Delhi Principal Product Manager R10 26 05/10/2021 Post graduate Female Karnataka Product Manager After completion, the transcripts were analyzed using NVivo. NVivo is one of the most widely used qualitative data analysis software tools (Bonello & Meehan, 2019). The NVivo 12 software was used to simplify the coding process, and audit trail, making the results credible and dependable (Soni et al., 2019; Richards & Hemphill, 2018; Mustafa et al., 2020; van Mastrigt et al., 2015). Furthermore, utilizing the NVivo 12 program and text analysis techniques based on word similarity, the findings of the interviews were automatically collated and coded. Sentiment analysis was used to comprehend and categorize emotions through characteristics and phases that will earn favorable, neutral, or unfavorable evaluations from interviewees to increase interpretation. NVivo 12 has a function for automatically identifying feelings in text that also reduces the subjectivity of human analysis. Sentences can be classified as moderately positive, very positive, moderately negative, or very negative. Comments that are not categorized into one of these four categories are regarded as neutral (Pudaruth et al., 2018). NVivo’s auto-coding function does not attempt to categorize an entire comment as positive or negative; instead, it examines words in isolation. As a result, some comments are labeled as both moderately positive and moderately positive, or extremely negative positive and very negative. These findings from sentiment auto-coding are depicted visually in Fig. 1 above.Fig. 1 Semantic auto coding analysis using NVivo Results and discussion Gamification implementation delivers a more engaging experience by engaging users in adopting certain behaviors by reducing boredom and overcoming weariness. Furthermore, gamification is not just a philosophy that allows for behavior modification, but it also has a technological component. The gamification model’s possible use is thought to produce emotions to relieve stress on engagement, motivation, friendly competition, cooperation, and behavior modification in many situations, including activities, customer loyalty, and consumer experience (Saputra & Rahmatia, 2021). The data analysis reveals several critical themes grounded empirically and well supported by evidence gathered from interview quotes. Applicability of gamification in online retail Gamification has shown to be a beneficial tactic in online retail. According to the findings of this study, the game components inspire a customer to spend more time on a gamified app, which increases traffic and screen time for shops.“Gamification is actually for me, it’s, it’s something that is subconsciously always there” (R1). Another gamification expert says that “gamification is an exciting tool to make the process more interesting and interactive” (R6). Gamification increases fun in its use as shown under the sub-themes in Fig. 2. This occurs due to the dynamics process’s influence on the player’s mental state while expressing strategic behaviors and actions (Saputra & Rahmatia, 2021).“Add fun to the activity or, you know, it is encouraging or entertaining because of the complicated nature” (R5). Fig. 2 Theme 1 and subthemes describing the applicability of gamification The next several years will see significant advancements in mobile technology, including the widespread deployment of 5G in major metropolitan regions. Retail applications, particularly those with physical stores, should explore ways to gamify their app services using this cutting-edge technology. Gamification employs a variety of mechanisms such as points, levels, challenges, leader boards, and many others to improve the user experience. Retail profits increase as a result of gamification. This is because, as a psychologist, the human mind is constantly competitive, and people enjoy gamification in retail which has various advantages, including increased customer retention and loyalty as also shown in Fig. 2. Customer retention is the ultimate objective for merchants since it aids in profit growth. Gamified retail may also aid in communicating with customers if correctly and efficiently designed. Gamification allows customers to interact with brands, which increases a customer’s loyalty to the brand. Retailers find it tough to provide an enriching in-store experience in the current environment. Online shopping has become a must, and gamified applications provide. consumers with an unforgettable experience—playing games. Gamification is already being applied in various apps in unique and efficient ways. However, it promises to have the most influence on the consumer experience in retail. Vendors may increase consumer loyalty and engagement, brand exposure, and income by embracing gamification and finding new methods to include it in their apps. Figure 2 also demonstrates that gamification enhances the overall experience of consumers. It improves the quality of time spent on these retail websites. It makes the experience fun and worth it. Consumers value avoiding the hassle of stores and being able to shop “whenever you want, there are no set opening hours,” enabling shopping activities to be an option during time slots that might have previously been reserved for entertainment activity as also discussed by Insley and Nunan (2014). Consumer experiences in gamified retail Consumers get a playful, pleasant, and enjoyable experience using gamification. Respondents talk about gamification mechanics, which help increase engagement and enhance the experiences. Points, leaderboards, competitions, mini-games, and badges are primary mechanics.“Gamification is more about exclusivity, like, for example, if there are like three dresses that are super exclusive and they are, they are cheap, they are not costly, but you, but you can only unlock them when you reach a level, right? Or a level of loyalty. Moreover, I think that then becomes a nice motive for me actually to play the game or be a part of it” (R3). Gamification is a valuable technique for improving the customer experience. To engage customers, retailers employ a variety of gamified technologies. This is the digital era, and great consumer experiences may lead to better earnings.“It is a creative call, new different experience.” (R6). Retailers recognize this and are using gamification as an engagement tactic to attract customers without deviating from the central premise of the business, resulting in increased sales. On the other hand, gamification must be deliberately implemented as part of a bigger company strategy to influence customer behavior. Several merchants utilize these approaches to demonstrate the effectiveness of gamification. Nike + is one such example. It is a fitness app that has been gamified. They monitor their customers’ activities and gather their data. After tracking the data, they provide it based on the points obtained. They employ game mechanics such as points, badges, and leaderboards, and uses gamification as a market tactic to improve revenues. ASOS is a prominent example in the fashion retail business. They allow their customers to create their outfits, which allows them to be creative and, as a result, enhances engagement and improves the ASOS customer experience.“Gamification, which is everything about positive reinforcement, it is about. Happy moments in life. It is about, I mean, that is the reason why people still use fees for competent profits. That is the reason why people upload so many photographs on Instagram, or they come to life, and they move up the value chain because they are all positive reinforcements.” (R9). 2arConsumer experience is a goal that may be attained via correct and successful gamification implementation in retail. Gamifying your loyalty program improves your customer experience by making it more enjoyable — and worthwhile — to participate in the program as shown in Fig. 3. There is something in it for your customers, even if it is simply a delightful game.Fig. 3 Theme 2 and subthemes describing the consumer experience in gamified retail Levels, progress indicators, customizable avatars, and branded games not only improve the user experience of your loyalty program but also may increase its efficacy. They make it enjoyable to interact with your program regularly. Ethics and challenges in gamification According to the literature, there is no regulation in gamified marketing of government. Various retailers do some unethical practices using gamification. Some retailers do not disclose their primary goals and intention. There is a lack of transparency in gamified marketing, and the respondents talk about some of the unethical practices they come across using gamification and during online shopping. The participants unanimously agreed that integrating gamification into a commercial setting might raise ethical concerns. If contextual variables such as user culture, conventions, and personality are not taken into account throughout the design process, gamification may cause difficulties such as increased stress and pressure on individuals, compelling them to give up privacy, or creating clusters of users while excluding others. When asked about ethical concerns, they share their personal experiences. One ethical concern may be that it becomes essential to the gamified application. So, for example, I open an app, download an app, and access a webpage for a problem that I need to fix. Yes. So I might have an end objective in mind, such as finishing my shopping. I need to take charge of a refund. The policy is on the website, and you are making me go through numerous steps to play a game, which is like, you know, I do not find that anywhere. Change the purpose of your customers, you know. From that perspective, I believe it is unethical. (R1). So there are many issues in gamified marketing, whether it is a lack of transparency, manipulation, hiding purposes and gaining too much personal information as shown in Fig. 4 as well. Gamification design and implementation can be immoral at times. The question of ethics in gamification is constantly a matter of discussion. Some unethical behaviors exist, and.Fig. 4 Theme 3 and subthemes describing the ethical issues and challenges in gamification the negative repercussions of gamification are prevalent. According to a 2013 poll on the use and impression of gamification, consumers are not always aware of the game aspects with which they engage. Because there is no set of standard policies on ethics in gamification and no guidelines defining how to use it, certain unethical acts occur in gamified systems. Consumers also feel that sometimes there is a breach of security because of the no regulation.“They try to become someone they are not and get that information. One day I got a message asking if I wanted to play this game. After playing those games, you will get certain vouchers, but I found out that it is not Adidas. It does not. It is someone else.” (R7). Gamification users suffer issues with a lack of transparency, and they are sometimes deceived and exploited. Despite having been employed in various settings for some years, gamification is still in its infancy. The negative impacts of gamification and its harm make the topic of ethics in gamification more debatable. The obstacles and concerns develop due to the design of the gamification framework. A lack of a defined framework exacerbates the negative impacts of gamification. Sometimes, some gamified platforms take so much information about the consumer. A respondent talked about it:“if the customers see the experiences that they have been through. They do not have authority about the information about them at the company, which can be a negative experience and cause distrust.” (R3). Given that gamification is one of the fastest disseminating behavioral techniques in business, there has been a less severe examination of the ethical implications of gamification in business than one might assume. We are concerned that this lack of attention is a precursor to what business ethicist Thomas Donaldson refers to as a phenomenon in which a severe moral and social failure occurs as a result of a significant lag time between the development of new technologies and the development of adequate normative frameworks to assess involved ethical issues. These difficulties occur because gamification suppliers may not be moral individuals. To maximize their earnings and traffic, they employ some heinous and wicked methods of attracting customers to their brand. As a result, they should not overlook these types of considerations for gamified marketing. Gamification providers should fully explore gamification frameworks, and they must incorporate morals, ethics, and overcoming hurdles. Conclusion This study explores the perspectives and gains insights into the gamification of online retail, focusing on its applicability, potential benefits, and obstacles. The themes that emerged from qualitative analysis are the applicability of gamification in online retail, consumer experiences in gamified retail, and ethics and challenges in gamification as shown in Table 2 also. Semantic analysis is performed, and as per the viewpoint of the gamification experts, it was found that the perception of ethics in gamification is negative, which shows that there are many unethical practices in gamified marketing. This study’s result should help retailers enhance their understanding of customers’ perceptions. The Covid-19 pandemic changed the retail scenario and caused many losses to the retail industry. So in 2020, many big retail brands went into administration. The whole pattern changes due to the closing of stores, social distancing, lockdowns, and work from home concept. However, still, consumers need retail experience. They want surety about the quality of the product. A retailer needs to give the consumers a reason to visit their website and app and convert their visit into a purchase. Gamification digitally engages people and motivates them to a particular behavior. Gamification effectively increases brand loyalty, improves brand recognition, enhances revenues, and attracts new customers. A retailer needs a clear goal and a well-designed and developed gamification app and website. Due to the pandemic, e-commerce activities have increased, and gamification as a digital engagement model enhanced the online retail experience and makes online shopping convenient, fun, and engaging. In order to create such a condition, a game should be designed in such a way that consumers/participants should not be harmed and manipulated. Moreover, a game should be tested in the real world to lead to better outcomes. The findings from the current study provide some valuable insights, thereby providing a rational basis for potentially fruitful future research in this area of growing interest.Table 2 Summary of qualitative analysis and emergence of themes Applicability of gamification in online retail Gamification Make the process more interesting and interactive It involves gratification and increases customer engagement ••••••••It makes the shopping experience enjoyable and fun ••••••••Gamification technology is an emerging phenomenon that impacts purchase intention during online shopping Gamified marketing It enhances traditional marketing tools using the playful experience it is good for branding ••••••••Marketers need to improve traditional marketing techniques Application It uses game mechanics to gather customer preferences and details and then generate rewards ••••••••Gamification helps in enhancing brand management by providing intrinsic and extrinsic motivation both Consumer experiences in gamified retail Engagement Creates engagement using financial rewards and enables consumers to achieve something ••••••••Personalization emerged to be an alternative to improve gamification effectiveness Motivation Motivates customers to perform certain activities for leaderboards Motivates consumers to stick to your brand ••••••••Consumers are not always aware of the game aspects with which they engage Experience Personalized the experience to provide a sense of accomplishment ••••••••Users suffer issues with a lack of transparency, and they are sometimes deceived Ethics and challenges in gamification Transparency The information gathered by the company from the consumer should be using a transparent process ••••••••They should not overlook these types of considerations for gamified marketing and they must incorporate morals, ethics, and overcoming hurdles Morality Try to change the purpose of consumer disguise like somebody who they are not and try to get information Ethical concerns There are no government rules and regulations Creates a sense of addiction and trying to get personal information Source: author compilation on the basis of qualitative interviews. Appendix Did you like to play games? What is your opinion about Gamification? If you get to play games and get rewards during online shopping, will it increase your purchase intention? (Bauer et al., 2020a) What motivates you to use a gamified shopping app? Does playing game on the online retail platform motivate you to spend more time on the platform? Why or why not? (Przybylski et al., 2010) Do you purchase more if the app is gamified? What do you think are the critical elements of gamification in retail? (Menon et al., 2022) Do you think that gamification help in the personal growth of consumers? (Berger et al., 2014) Did gamification impact your emotions? (Korn et al., 2015) According to you, is the effect of gamification long-term or short-term? (Kim & Castelli, 2021) Did gamified experience help in improving your trust? (Muskan Agarwal et al., 2021) 10) What do you prefer a non-gamified shopping site or a gamified shopping site? 11) Is gamification help in increasing screen time or traffic on the retailer’s website? 12) As a consumer do you think that the gamified experience should be enhanced and changed according to consumers? (Mustafa & Karimi, 2021) 13) What motivates you more the rewards and prizes you get or the fun and satisfaction you feel during shopping? (Fenton et al., 2022) 14) Did Gamification enhance your overall online retailing experience of yours? (Insley & Nunan, 2014) 15) What are the main barriers you came across using a Gamified app? 16) Please discuss any ethical issues in Gamified Marketing? (Thorpe & Roper, 2019) Declarations Conflict of interest The authors declare no competing interests. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Ali BJ Impact of COVID-19 on consumer buying behavior toward online shopping in Iraq Economic Studies Journal 2020 18 42 267 280 Agarwal M Goyal R Verma E Goyal H Ahmed G Singhal S Predictive sentimental analysis of spam detection using machine learning 2021 International Conference on Computing, Communication and Green Engineering (CCGE) 2021 IEEE 1 6 Álvarez-Bermejo JA Belmonte-Ureña LJ Martos-Martínez A Barragán-Martín AB del Mar Simón-Marquez M System to detect racial-based bullying through gamification Frontiers in Psychology 2016 7 1791 10.3389/fpsyg.2016.01791 27933006 Arora C Razavian M Ethics of gamification in health and fitnesstracking International Journal of Environmental Research and Public Health 2021 18 21 11052 10.3390/ijerph182111052 34769570 Aydınlıyurt ET Taşkın N Scahill S Toker A Continuance intention in gamified mobile applications: A study of behavioral inhibition and activation systems International Journal of Information Management 2021 61 102414 10.1016/j.ijinfomgt.2021.102414 Bauer, J. C., Linzmajer, M., Nagengast, L., Rudolph, T., & D'Cruz, E. (2020a). Gamifying the digital shopping experience: Games without monetary participation incentives increase customer satisfaction and loyalty. Journal of Service Management. Bauer KN Garcia-Marquez C Gandara DA Teaching with games and gamification: best practices and future research needs Handbook of teaching with technology in management, leadership, and business 2020 Edward Elgar Publishing Berger, V., Miesler, L., & Hari, J. J. (2014). The potential of gamification in changing consumer behaviour towards a more sustainable nutrition behaviour. In Academy of Marketing Conference 2014, Bournemouth, UK. Bidler M Zimmermann J Schumann JH Widjaja T Increasing consumers’ willingness to engage in data disclosure processes through relevance-illustrating game elements Journal of Retailing 2020 96 4 507 523 10.1016/j.jretai.2020.10.001 Bonello M Meehan B Transparency and coherence in a doctoral study case analysis: Reflecting on the use of NVivo within a'Framework'approach The Qualitative Report 2019 24 3 483 498 De Canio, F., Fuentes-Blasco, M., & Martinelli, E. (2021b). Engaging shoppers through mobile apps: the role of gamification. International Journal of Retail & Distribution Management. De Canio F Fuentes-Blasco M Martinelli E Engaging shoppers through mobile apps: The role of Gamification International Journal of Retail and Distribution Management 2021 10.1108/IJRDM-09-2020-0360 Deterding, S., Dixon, D., Khaled, R., & Nacke, L. (2011, September). From game design elements to gamefulness: Defining "gamification." Proceedings of the 15th International Academic MindTrek conference: Envisioning future media environments, 9–15.10.1145/2181037.2181040 Donthu N Gustafsson A Effects of COVID-19 on business and research Journal of Business Research 2020 117 284 289 10.1016/j.jbusres.2020.06.008 32536736 Dwivedi, Y. K., Ismagilova, E., Hughes, D. L., Carlson, J., Filieri, R., Jacobson, J., ... & Wang, Y. (2021). Setting the future of digital and social media marketing research: Perspectives and research propositions. International Journal of Information Management, 59, 102168 Epstein, D. S., Zemski, A., Enticott, J., & Barton, C. (2021). Tabletop board game elements and Gamification interventions for health behavior change: Realist review and proposal of a game design framework. JMIR Serious Games, 9(1), e23302.10.2196/23302 Fenton A Parry KD Chadwick S Guimarães G Aeron V Digital innovation in sport–barriers and opportunities for branded fitness apps for fans Entrepreneurial innovation 2022 Springer 25 42 Gajanova, L., & Radišić, M. (2021). Self-determination theory as mediator in the nexus of gamification and customer purchasing behaviour. In SHS Web of Conferences (Vol. 90, p. 01005). EDP Sciences. Gatautis R Vitkauskaite E Gadeikiene A Piligrimiene Z Gamification as a mean of driving online consumer behaviour: SOR model perspective Engineering Economics 2016 27 1 90 97 10.5755/j01.ee.27.1.13198 Gatautis, R., Banytė, J., Kuvykaitė, R., Virvilaitė, R., Dovalienė, A., Piligrimienė, Ž., Gadeikienė, A., Vitkauskaitė, E., & Tarutė, A. (2021). The conceptual model of gamification-based consumer engagement in value creation10.1007/978-3-030-54205-4_5 Goethe, O. (2019). Gamification mindset. Springer International Publishing. Gooch D Vasalou A Benton L Khaled R Using gamification to motivate students with dyslexia. In Proceedings of the 2016 CHI Conference on human factors in computing systems 2016 969 980 Guvenc, O., Kazybayeva, A., & Abeshev, K. (2020). Dynamic prices in retail and its impacts on logistics. In VEHITS (pp. 659–666). Hamari, J., Koivisto, J., & Sarsa, H. (2014, January). Does gamification work?--A literature review of empirical studies on gamification. In 2014 47th Hawaii international conference on system sciences (pp. 3025–3034). Ieee. Haque, M. F., Haque, M. A., & Islam, M. (2014). Motivational theories-A critical analysis. ASA University Review, 8(1). Hassan L Hamari J Gameful civic engagement: A review of the literature on gamification of e-participation Government Information Quarterly 2020 37 3 101461 10.1016/j.giq.2020.101461 Hofacker CF De Ruyter K Lurie NH Manchanda P Donaldson J Gamification and mobile marketing effectiveness Journal of Interactive Marketing 2016 34 25 36 10.1016/j.intmar.2016.03.001 Högberg J Ramberg MO Gustafsson A Wästlund E Creating brand engagement through in-store gamified customer experiences Journal of Retailing and Consumer Services 2019 50 122 130 10.1016/j.jretconser.2019.05.006 Hsu CL Chen MC How gamification marketing activities motivate desirable consumer behaviors: Focusing on the role of brand love Computers in Human Behavior 2018 88 121 133 10.1016/j.chb.2018.06.037 Hyrynsalmi S Smed J Kimppa K The dark side of gamification: how we should stop worrying and study also the negative impacts of bringing game design elements to everywhere GamiFIN 2017 96 104 Insley V Nunan D Gamification and the online retail experience International Journal of Retail and Distribution Management 2014 42 5 340 351 10.1108/IJRDM-01-2013-0030 Jayasooriya S Alles T Thelijjagoda S Demystifying the concept of IoT enabled gamification in retail marketing: An exploratory study In 2020 International Research Conference on Smart Computing and Systems Engineering (SCSE) 2020 IEEE 234 241 Kagan S Normative ethics 2018 Routledge Kapp KM Valtchanov D Pastore R Enhancing motivation in workplace training with casual games: A twelve month field study of retail employees Educational Technology Research and Development 2020 68 5 2263 2284 10.1007/s11423-020-09769-2 Karać, J., & Stabauer, M. (2017, July). Gamification in E-commerce. In International Conference on HCI in Business, Government, and Organizations (pp. 41–54). Springer. Kim S How a company’s gamification strategy influences corporate learning: A study based on gamified MSLP (Mobile social learning platform) Telematics and Informatics 2021 57 101505 10.1016/j.tele.2020.101505 Kim J Castelli DM Effects of gamification on behavioral change in education: A meta-analysis International Journal of Environmental Research and Public Health 2021 18 7 3550 10.3390/ijerph18073550 33805530 Kim TW Werbach K More than just a game: Ethical issues in gamification Ethic Inf. Technol. 2016 18 157 173 10.1007/s10676-016-9401-5 Korn O Boffo S Schmidt A The effect of gamification on emotions-the potential of facial recognition in work environments International Conference on Human-Computer Interaction 2015 Springer 489 499 Kriz WC Kikkawa T Sugiura J Manipulation through gamification and gaming Gaming as a cultural commons 2022 Springer 185 199 Kunkel T Lock D Doyle JP Gamification via mobile applications: A longitudinal examination of its impact on attitudinal loyalty and behavior toward a core service Psychology & Marketing 2021 38 6 948 964 10.1002/mar.21467 Lowry PB Gaskin J Twyman NW Hammer B Roberts TL Taking ‘fun and games’ seriously: Proposing the hedonic-motivation system adoption model (HMSAM) Journal of the Association for Information Systems 2013 14 11 617 671 10.17705/1jais.00347 Lu HP Ho HC Exploring the impact of gamification on users’ engagement for sustainable development: A case study in brand applications Sustainability 2020 12 10 4169 10.3390/su12104169 Marczewski A The ethics of gamification ACM Interactions IX, Fall 2017 24 1 56 Menon, A., Khator, D., Prajapati, D., & Ekbote, A. (2022). IPL prediction using machine learning. Indian Journal of Computer Science, 7(3). Michalos AC A pragmatic approach to business ethics 1995 Sage Minge M Cymek DH Investigating the potential of gamification to improve seniors' experience and use of technology Information 2020 11 5 249 10.3390/info11050249 Mulcahy, R. F., McAndrew, R., Russell-Bennett, R., & Iacobucci, D. (2021). "Game on!" Pushing consumer buttons to change sustainable behaviour: A gamification field study. European Journal of Marketing. 10.1108/EJM-05-2020-0341 Mustafa AS Karimi K Enhancing gamified online learning user experience (UX): A systematic literature review of recent trends Human-Computer Interaction and Beyond-Part 2021 I 74 99 Mustafa J Hodgson P Lhussier M Forster N Carr SM Dalkin SM ‘Everything takes too long and nobody is listening’: Developing theory to understand the impact of advice on stress and the ability to cope PloS One 2020 15 4 e0231014 10.1371/journal.pone.0231014 32324749 Nacke, L. E., & Deterding, C. S. (2017). The maturing of gamification research. Computers in Human Behaviour, 450–454. Pasca, M. G., Renzi, M. F., Di Pietro, L., & Mugion, R. G. (2021). Gamification in tourism and hospitality research in the era of digital platforms: A systematic literature review. Journal of Service Theory and Practice. B. R.Paul and L. Elder, L. Miniature guide to ethical reasoning. Foundation for critical thinking, 2003. https://www.criticalthinking.org/files/Concepts&Tools.pdf. Przybylski A Rigby S Ryan R A motivational model of video game engagement Review of General Psychology 2010 14 154 166 10.1037/a0019440 Pudaruth, S., Moheeputh, S., Permessur, N., & Chamroo, A. (2018). Sentiment analysis from Facebook comments using automatic coding in NVivo 11. Richards KAR Hemphill MA A practical guide to collaborative qualitative data analysis Journal of Teaching in Physical Education 2018 37 2 225 231 10.1123/jtpe.2017-0084 Saputra AD Rahmatia A Gamification model as a business strategy for MSMEs in Indonesia Journal of Accounting and Strategic Finance 2021 4 1 91 107 10.33005/jasf.v4i1.162 Sheth J Impact of Covid-19 on consumer behavior: Will the old habits return or die? Journal of Business Research 2020 117 280 283 10.1016/j.jbusres.2020.05.059 32536735 Shevchuk, N., Degirmenci, K., & Oinas-Kukkonen, H. (2019). Adoption of gamified persuasive systems to encourage sustainable behaviours: Interplay between perceived persuasiveness and cognitive absorption. In Proceedings of the 40th International Conference on Information Systems, ICIS 2019 (pp. 1–17). Association for Information Systems. Shi S Leung WK Munelli F Gamification in OTA platforms: A mixed-methods research involving online shopping carnival Tourism Management 2022 88 104426 10.1016/j.tourman.2021.104426 Soares, F., Madureira, A., Pagès, A., Barbosa, A., Coelho, A., Cassola, F., Ribeiro, F., Viana, J., Andrade, J., Dorokhova, M., Morais, N., Wyrsch, N., & Sørensen, T. (2021). Feedback: An ICT-based platform to increase energy efficiency through buildings' consumer engagement. Energies, 14(6). 10.3390/en14061524 Soni N Gleaves S Neff H Morrison-Smith S Esmaeili S Mayne I Anthony L Do user-defined gestures for flatscreens generalize to interactive spherical displays for adults and children? Proceedings of the 8th ACM International Symposium on Pervasive Displays 2019 1 7 Sreejesh S Ghosh T Dwivedi YK Moving beyond the content: The role of contextual cues in the effectiveness of gamification of advertising Journal of Business Research 2021 132 88 101 10.1016/j.jbusres.2021.04.007 Suh A Cheung CM Ahuja M Wagner C Gamification in the workplace: The central role of the aesthetic experience Journal of Management Information Systems 2017 34 1 268 305 10.1080/07421222.2017.1297642 Szász L Bálint C Csíki O Nagy BZ Rácz BG Csala D Harris LC The impact of COVID-19 on the evolution of online retail: The pandemic as a window of opportunity Journal of Retailing and Consumer Services 2022 69 103089 10.1016/j.jretconser.2022.103089 Thorpe AS Roper S The ethics of gamification in a marketing context Journal of Business Ethics 2019 155 2 597 609 10.1007/s10551-017-3501-y Thukral E COVID-19: Small and medium enterprises challenges and responses with creativity, innovation, and entrepreneurship Strategic Change 2021 30 2 153 158 10.1002/jsc.2399 Ushakov D Shatila K The impact of workplace culture on employee retention: An empirical study from Lebanon The Journal of Asian Finance, Economics and Business 2021 8 12 541 551 van Mastrigt GA Paulus AT Aarts MJ Evers SM Alayli-Goebbels AF A qualitative study on the views of experts regarding the incorporation of non-health outcomes into the economic evaluations of public health interventions BMC Public Health 2015 15 1 1 17 25563658 Vashisht D Sreejesh S Effects of brand placement strength, prior game playing experience and game involvement on brand recall in advergames Journal of Indian Business Research 2015 7 3 292 312 10.1108/JIBR-11-2014-0082 Vdov K The effect of gamification on customer experience in the digital environment 2020 Weiss, J. (2019, January). Gamification and scholarly ethical perspectives on industries, a bibliometric analysis. In Proceedings of the 52nd hawaii international conference on system sciences. Yang Y Asaad Y Dwivedi Y Examining the impact of Gamification on intention of engagement and brand attitude in the marketing context Computers in Human Behavior 2017 73 459 469 10.1016/j.chb.2017.03.066 Xu Y Chen Z Peng MYP Anser MK Enhancing consumer online purchase intention through gamification in China: Perspective of cognitive evaluation theory Frontiers in Psychology 2020 11 581200 10.3389/fpsyg.2020.581200 33329236 Zain NHM Johari SN Aziz SRA Teo NHI Ishak NH Othman Z Winning the needs of the Gen Z: Gamified health awareness campaign in defeating COVID-19 pandemic Procedia Computer Science 2021 179 974 981 10.1016/j.procs.2021.01.087	0	PMC9716125	NO-CC CODE	2022-12-03 23:20:54	no	Asian J Bus Ethics. 2022 Dec 2;:1-21	utf-8	null	null	null	oa_other
==== Front Immunol Res Immunol Res Immunologic Research 0257-277X 1559-0755 Springer US New York 9346 10.1007/s12026-022-09346-0 Original Article Immunoinformatic-guided designing of multi-epitope vaccine construct against Brucella Suis 1300 Jalal Khurshid 1 Khan Kanwal 2 http://orcid.org/0000-0002-9928-4482 Uddin Reaz [email protected] 2 1 grid.266518.e 0000 0001 0219 3705 HEJ Research Institute of Chemistry International Center for Chemical and Biological Sciences, University of Karachi, Karachi, Pakistan 2 grid.266518.e 0000 0001 0219 3705 Lab 103 PCMD Ext. Dr. Panjwani Center for Molecular Medicine and Drug Research, International Center for Chemical and Biological Sciences, University of Karachi, Karachi, 75270 Pakistan 2 12 2022 120 20 6 2022 20 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Abstract Brucella suis mediates the transmission of brucellosis in humans and animals and a significant facultative zoonotic pathogen found in livestock. It has the capacity to survive and multiply in a phagocytic environment and to acquire resistance under hostile conditions thus becoming a threat globally. Antibiotic resistance is posing a substantial public health threat, hence there is an unmet and urgent clinical need for immune-based non-antibiotic methods to treat brucellosis. Hence, we aimed to explore the whole proteome of Brucella suis to predict antigenic proteins as a vaccine target and designed a novel chimeric vaccine (multi-epitope vaccine) through subtractive genomics-based reverse vaccinology approaches. The applied subsequent hierarchical shortlisting resulted in the identification of Multidrug efflux Resistance-nodulation-division (RND) transporter outer membrane subunit (gene BepC) that may act as a potential vaccine target. T-cell and B-cell epitopes have been predicted from target proteins using a number of immunoinformatic methods. Six MHC I, ten MHC II, and four B-cell epitopes were used to create a 324-amino-acid MEV construct, which was coupled with appropriate linkers and adjuvant. To boost the immunological response to the vaccine, the vaccine was combined with the TLR4 agonist HBHA protein. The MEV structure predicted was found to be highly antigenic, non-toxic, non-allergenic, flexible, stable, and soluble. To confirm the interactions with the receptors, a molecular docking simulation of the MEV was done using the human TLR4 (toll-like receptor 4) and HLAs. The stability and binding of the MEV-docked complexes with TLR4 were assessed using molecular dynamics (MD) simulation. Finally, MEV was reverse translated, its cDNA structure was evaluated, and then, in silico cloning into an E. coli expression host was conducted to promote maximum vaccine protein production with appropriate post-translational modifications. These comprehensive computer calculations backed up the efficacy of the suggested MEV in protecting against B. suis infections. However, more experimental validations are needed to adequately assess the vaccine candidate’s potential. Highlights • Subtractive genomic analysis and reverse vaccinology for the prioritization of novel vaccine target • Examination of chimeric vaccine in terms of allergenicity, antigenicity, MHC I, II binding efficacy, and structural-based studies • Molecular docking simulation method to rank based vaccine candidate and understand their binding modes Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1007/s12026-022-09346-0. Keywords Brucella suis Chimeric vaccine Reverse vaccinology Vaccine target Epitope prediction ==== Body pmcIntroduction Brucellosis is an endemic disease, also known as Malta fever, Mediterranean fever, undulant fever, and Bang’s disease caused by Brucella genus belong to a family of Brucellaceae, class Alphaproteobacteria infecting both animals and humans [1]. Species of this genus are Gram-negative intracellular facultative pathogens. Based on specific phenotypes with the host and environmental adaptation, till now more than ten known species of Brucella are identified [2]. Brucellosis is characterized as acute fever illness [3], associated with various symptoms in human such as liver and spleen disorders, reproductive abnormalities, neurological problems, heart-related problems, and also have been classified as a potential bioterrorism agent [4, 5]. Brucellosis remains endemic in various emerging countries in Asia, Africa, Middle East, and South America, where screening of livestock and vaccination fails to control and exterminate the disease [6]. The World Health Organization (WHO), Food and Agriculture Organization (FAO), and World Organization for Animal Health (OIE) consider brucellosis as one of the most contagious zoonotic disease worldwide [7], yet human brucellosis remains the commonest zoonotic disease around the world [7]. Brucellosis is transmitted to humans via several routes including, eating raw dairy products from infected animals, aerosol inhalation of an infected animal in clinical lab and slaughterhouse, handling, and exposure to tissue and body fluid of infected animals without Personal Protective Equipment (PPE) [8]. According to Hull [9], there are 500,000 cases of human brucellosis reported per year around the globe due to their ability to survive and multiply within the host phagocytotic and non-phagocytotic cell. Surprisingly, Brucella did not show classical virulence mechanisms such as producing cytolysin, plasmids, fimbria, exotoxin, exoenzymes, and drug resistant forms. However, Brucella is having major virulence factors such as lipopolysaccharide (LPS) [10], β-cyclic glucan [11], outer membrane proteins (Omps) [12], MucR [13], T4SS secretion system, and BvrR/BvrS system which permit the Brucella to interact with the host cell [14]. Similarly, T4SS-based VirB proteins complex their 5 effectors are an essential part of Brucella pathogenesis that regulate the host cell inflammation response and vesicle trafficking [15]. Due to the limited knowledge regarding the genomics of Brucella suis, a new species-specific therapeutic compound and vaccine candidates are difficult to design experimentally [16]. The immune system prophylactic and designing of new therapeutic approaches are of significant interest to combat the antimicrobial resistance. The peptide-based chimeric vaccine, expressed by the pathogenic strain of Brucella, would be an appropriate alternative to its resistance. Nevertheless, evaluation of thousands of macro-molecules and their subsequent in vivo assays in the wet lab need a lot of time and cost for vaccine design. On the other hand, development in computational biology and various other bioinformatics fields have made wonderful progress leading to a great reduction in consumption of time and associated expenses [17–19]. Bioinformatics analysis typically employs alternative approaches for finding novel drug targets, identifying vaccine or drug candidates, elucidating the host–pathogen interactions, designing structure-based drugs, allowing genome-based comparative study, and so on. It is thereby reducing the conventional laboratory-based experimental practice [20–22]. The subtractive genomic and reverse vaccinology are one of the most used computational approaches to evaluate the suitability of the vaccine target among the previously known drug target based on selectivity and specificity. Numerous studies had widely reported the use of subtractive genomics and reverse vaccinology approaches against various pathogenic strains for the identification of novel species-specific therapeutic targets [22–25]. Hence, in the current study, we used the subtractive genomic analysis and reverse vaccinology approach against the whole proteome of the Brucella suis strain to shortlist the vital proteins as vaccine targets. The results of this study suggest that our resultant proteins may be considered the best vaccine targets along with identified multi-epitopes chimeric vaccine that can be used for establishing a universal vaccine, which may provide a basic pipeline against Brucellosis. Material and methods The current study utilized a subtractive genomic and reverse vaccinology approach for the identification of Brucella suis specific vaccine targets [26]. Data retrieval The complete proteome of Brucella suis was retrieved from the National Centre for Biotechnology Information (NCBI) database [27] along with the complete proteome of Human from UNIPROT database (Table 1) [28]. The Database of Essential Gene (DEG) was used to identify the essentiality of proteins, whereas KEGG sever was used to retrieve human and Brucella metabolic pathways.Table 1 Complete proteome of human-host and Brucella suis Strain ID Strain name Proteins GCF_0000007505.1 1330 2974 9606 Human ~ 20,0000 Prioritization of pathogen-specific metabolic pathways The KEGG [29] and Automatic Annotation Server (KAAS) metabolic pathway database were used for the analysis of human-host and Brucella suis metabolic pathways. We retrieved the metabolic pathway ID with related information from the KEGG database. Those pathways which were present in both considered a common metabolic pathway, while the remaining considered unique metabolic pathways. We retrieved the FASTA sequence of proteins involved in unique metabolic pathways of the Brucella suis from the NCBI database. Non-homologous protein identification Unique metabolic pathways proteins were subjected to evaluate through a BLASTp against the whole proteome of a human using the cut-off value of 0.0001 (E-value 10–3). Those proteins which had a high sequence similarity (> 80%) with human proteome were excluded and the remaining proteins having no similarities were retrieved and further analyzed in the next step of subtractive genomic analysis. Prioritization of essential proteins among the whole proteome of Brucella suis In order to prioritize the essential proteins of the Brucella suis the proteins were analyzed via BLASTp with a threshold value of 10–5, against the Database of Essential Gene (DEG) [30] consisted of proteins responsible for the survival of the organisms. Those proteins which had sequence similarity with essential protein of DEG were further analyzed and the non-essential proteins were excluded. Determination of virulence factor (proteins) The virulence of proteins helps bacteria to destroy the host-immune system with the help of colonization and invading the host immune cell and as a result, the disease caused. For the determination of virulence of proteins, VFDB (virulence factor of pathogenic bacteria) online database [31] was used. The shortlisted proteins of Brucella suis were subjected to BLAST against the VFDB. Identification of resistance proteins The Antibiotic Resistance Gene-ANNOTation V6 (ARG-ANNOT V6) tool was used for the prediction of novel resistance protein sequences from whole-genome and proteome of a pathogen. All the resistant proteins data were collected and analyzed from different experimental published work, different online sources, and protein sequences were retrieved from NCBI database. The shortlisted proteins FASTA sequence were then subjected to BLAST against the resistance proteins of the ARG-ANNOT V6 database with a threshold of 10–5 [32]. Subcellular localization of shortlisted proteins All the shortlisted proteins were then subjected to PSORTb version 3.0.2 [33] and Cello v.2.5 [34] online tool in order to identify subcellular localization. The main principle of subcellular localization (SCI BLAST) is to BLAST all the shortlisted proteins against the PSORTb and Cello v.2.5 online servers. The subcellular localization consists of cytoplasm, cytoplasmic membrane, periplasmic membrane, and extracellular space, and some are unknown. Non-homolog gut protein identification Nonhomologous essential proteins selected as vaccine candidates and novel drug targets were further subjected to standalone BLASTp with an E value cutoff score of 0.001 against the data set present in the Human Microbiome Project server (https://www.hmpdacc.org/hmp/) “28,331” (BioProject) retrieved as gastrointestinal tract from HMP [35] i.e., (https://hmpdacc.org/hmp/catalog/grid.php?dataset=genomic&hmp_isolation_body_site=gastrointestinal_tract) having 26,295 proteins (accessed on 1st Sep, 2022). The HMP sampled microbes from a healthy adult population of 239 people, taking samples from 18 different sites around the body (the mouth, skin, nose, intestines, and genitourinary tract). This yielded ~ 823 unique gut microbial species compositions [36]. Novel therapeutic targets and vaccine candidates were found for proteins with a no similarity threshold. Based on these findings, precautions may be taken to ensure that human microflora proteins are not inadvertently inhibited or blocked. Prediction of antigenic protein Outer membrane protein shortlisted from the above subtractive genomic analysis was selected that contain the potency for vaccine development. A bioinformatics approach was applied to that protein to identify the epitopes to boost the immune response of the host. For the prediction of antigenic outer membrane protein VaxiJen webserver [37] was used with a default parameter of 0.5 considered potent antigenicity. Prediction of MHC class I T-cell epitope The MHC I epitopes for the selected protein was predicted by NetCTL server [38]. The predicted epitopes were chosen based on a high score. The default parameter of 0.75 was used for the prediction of potent T cell MHC I epitopes. Prediction of MHC I epitope immunogenicity A bioinformatics tool IEDB server [39] was used for MHC I antigenicity prediction. The MHC I epitopes should have the strength to evoke a host immune response. A default parameter was used to predict the immunogenicity of MHC I epitopes. Those epitopes having positive antigenic values were selected for further analysis. Antigenicity, conservancy, and toxicity analysis The immunogenic epitopes obtained from the IEDB server were analyzed for the antigenic capability using Vaxijen version 2.0 server with a threshold value of 0.5. The IEDB conservancy analysis [40] tool was used for the assessment of the conserved sequence among all the genotypes of MHC I epitopes. The conserved sequence identity parameter was set as a default. This analysis showed the conserved epitopes within the given protein sequence [41]. The assessment of toxicity level was predicted by an online tool called ToxinPred. The parameter was set on default. The toxicity level confirms that the specific host immune response will only target the bacteria rather than host cell itself [42]. MHC II epitope prediction The IEDB-AR server was used for the identification of MHC class 2 epitopes. It applies the consensus-based prediction approach of both the average relative binding matrix method and stabilization matrix alignment method [43]. MHC I- and II-restricted allele cluster analysis To confirm the predicted T-cell epitopes, the MHCcluster v2.0 [44] was used to find the cluster of MHC-restricted alleles with their appropriate peptides. This server crosschecks the MHC-restricted allele analysis from the IEDB analysis resources. It results in the static heat map and phylogenetic tree analyzing the functional relation between peptides and HLAs. Prediction of B-cell epitopes The B-cell epitope prediction was performed by online bioinformatics server BCPred [45] and FBCpred server [45]. The BCPred works on five different kernel methods whereas, FBCpred is based on consequent kernel methods. The cut-off score of 0.8 was used for B-cell epitopes identification via BCPred [46]. The IEDB B-cell epitope prediction server was used for biochemical properties analysis such as hydrophobicity, surface accessibility, hydrophilicity, amino acid composition, and secondary structure to predict the linear B-cell epitopes. Construction of model vaccine Different combinations of the shortlisted T-cell and B-cell epitopes were conjugated sequentially to model a vaccine construct with low toxicity, allergenicity, and high immunogenicity. During the vaccine construction, four different epitopes sequence were added in different combinations. Different amino acid linkers such as GGGS were used between these sequences. To enhance the antigenicity of the vaccine, PADRE (Pan HLA-DR reactive epitope) along with four different adjuvants such as HBHA, HBHA conserved, ribosomal, and beta-defensin were used in each vaccine construct, respectively. The PADRE sequence induced CD4 + T-cells that improve efficacy and potency of peptide vaccine [47]. The adjuvant HBHA and ribosomal adjuvants sequence are agonists of the toll-like receptor 4 (TLR4) while beta-defensin adjuvant is an agonist of the TLR1, TLR2, and TLR4. The sequence-fused vaccine construct was used for further analysis. Evaluation of allergenicity, antigenicity, and solubility for vaccine construct The allergenicity was determined by online tool AlgPred program (Saha & Raghava, 2006). The AlgPred tool uses all these parameters (IgE epitope + MAST + ARPs BLAST + SVM) combined to predict the allergenicity of vaccine construct with a threshold value of − 0.40 prediction score. The ANTIGENpro program [48] was used for the determination of vaccine antigenicity while SOLpro was used for the solubility of a vaccine. Physicochemical properties of vaccine construct The physicochemical properties of vaccine sequence were observed by the Expasy ProtParam online server [49]. This tool is used for the identification of different physicochemical properties such as molecular weight, number of amino acids, PI values, hydropathicity GRAVY values, aliphatic index, instability index, and estimated half-life of the protein of a generated protein model. This software can analyze various physicochemical properties based on pKa values of different amino acids sequence of a vaccine. The instability index of protein forecasts whether our vaccine is stable or unstable. The aliphatic index of a protein referred to the volume occupied by the aliphatic side chains of amino acid. Secondary structure modeling The modeling of vaccine construct was performed by the online tool Swiss modeler [50]. The homology models are constructed using the automated SWISS-MODEL homology modeling server pipeline, and experimental resolve crystal structures of proteins are mapped between the PDB database and UniProtKB using SIFTS [51]. The PSIPRED and PROCHECK were used for the model structure evaluation. The Psi-BLAST selects the relating sequence which had similarity > 80.6% with reported proteins. The best model for each protein was selected and subjected to further analysis. Molecular docking and molecular dynamic simulation The molecular docking of V1 with 6 different human leukocyte antigen (HLA) alleles was performed via online tool PatchDock [52] to show the interaction of V1 with HLA alleles. Six different HLA alleles were downloaded from the PDB database with its PDB ID 2FSE (HLA-DRB101:01), 3C5J (HLA-DR B302:02), 1H15 (HLA-DR B501:01), 2Q6W (HLA-DR B301:01), 2SEB (HLA-DRB104:01), and 1A6A (HLA-DR B103:01). The FireDock (Fast Interaction Refinement in Molecular Docking) was used for further validation and more refinement of interactions [52]. Similarly, docking of vaccine (V1) with TLR4 was performed by the GRAMMX [53]. Finally, molecular dynamic simulation of vaccine and TLR4 complex was performed using GROMACS [54]. The final vaccine solvation was executed with an SPC water model in a cubic box with energy minimization using the steepest algorithm. Here, the system was firstly equilibrated using NVT ensemble followed by NPT ensemble. Finally, the vaccine’s molecular dynamic simulation was performed for 10 ns. Furthermore, molecular dynamic simulation of docked complex (vaccine with TLR4) was performed via iMODs server [55]. The iMODs define and calculate the flexibility of protein complex and can be accessed freely. The server asses the direction and extent of the immanent motions of the complex in terms of B-factors, deformability, covariance, and eigenvalue. Optimization and in silico cloning of final vaccine model The amino acid sequence of vaccine was reversed translated into DNA sequence to optimized and enhance the expression of chimeric vaccine proteins in the E. coli system through JCAT (Java Codon Adaptation Tool) [56]. The JCAT tool resulted in the prediction of GC content and codon Adaption Index score (CAI) for DNA sequences while avoiding the cleavage site for restriction enzymes Rho-independent termination of transcription and prokaryotic ribosome binding sites. Snapgene tool was used for the insertion of an optimized amino acid sequence of vaccine construct in E. coli pET-28a( +) expression system. Results Unique metabolic pathway analysis Complete metabolic pathways of both Brucella suis (109 pathways) and human host (330 pathways) were downloaded from the KEGG (Kyoto encyclopedia of gene and genome) server. We compared both the human and Brucella suis metabolic pathways manually to find both the common and unique metabolic pathways. The results showed that 71 pathways were common (Table S1) both in humans and Brucella suis and the 38 pathways (Table 2) were unique to Brucella suis. Among these 38 pathways, two-component system, quorum sensing, ribosome, flagellar assembly, and bacterial secretion pathways was observed as highly unique pathways to brucella with 64, 114, 55, 35, and 27 unique proteins. Two-component system (TCS) and quorum sensing regulate the expression of virulence genes, enable the pathogen to detect changes in its external environment and adapt its gene expression appropriately [57]. For example, the quorum-sensing (QS) regulator VjbR, which triggers virB expression, is encoded by the gene encoding the two-component regulator BvrRS, which enables the brucellae to sense the acidic pH and food deprivation they face in the eBCV [57]. In Brucella, a non-motile bacterium, the flagellum, is involved in virulence, infectivity, cell proliferation, and biofilm formation. Brucella’s flagellar proteins are expressed, providing evidence for an evolutionary scenario in which a free-living bacterium acquired flagellar genes from environmental microorganisms, conferring on it the ability to reach other hosts (mammals), and, under selective pressure from the environment, can express these genes, helping it evade the immune response [58]. Likewise, the bacterial secretion pathways and ribosomal pathways help brucella to evade certain non-favorable environment by the secretion of various toxins and proteins. Additionally, Methane metabolism, Lysine biosynthesis, Lipopolysaccharide biosynthesis, O-Antigen nucleotide sugar biosynthesis, Peptidoglycan biosynthesis, Benzoate degradation, Beta-Lactam resistance, and Cationic antimicrobial peptide (CAMP) resistance pathways were consisting of 10–20 unique proteins. These metabolic pathways enable brucella to modulate the cell wall composition to evade the antibiotic environment. Moreover, other pathways such as Vancomycin resistance, d-Alanine metabolism, Carotenoid biosynthesis, Limonene and pinene degradation, Geraniol degradation, Polyketide sugar unit biosynthesis, Carbapenem, Monobactam, Streptomycin, and Novobiocin biosynthesis having less than 10 unique proteins respectively. Total unique metabolic pathways consisted of 503 proteins (Fig. 1).Table 2 Unique metabolic pathways of Brucella suis S. no Unique metabolic pathways Pathway ID Number of proteins 1 C5-Branched dibasic acid metabolism Bms00660 8 2 Methane metabolism Bms00680 16 3 Secondary bile acid biosynthesis Bms00121 3 4 Lysine biosynthesis Bms00300 13 5 Cyanoamino acid metabolism Bms00460 3 6 d-Alanine metabolism Bms00473 4 7 Lipopolysaccharide biosynthesis Bms00540 12 8 O-Antigen nucleotide sugar biosynthesis Bms00541 15 9 Peptidoglycan biosynthesis Bms00550 19 10 Carotenoid biosynthesis Bms00906 1 11 Limonene and pinene degradation Bms00903 7 12 Geraniol degradation Bms00281 4 13 Polyketide sugar unit biosynthesis Bms00523 2 14 Biosynthesis of siderophoro group nonribosomal peptides Bms01053 7 15 Carbapenem biosynthesis Bms00332 2 16 Monobactam biosynthesis Bms00261 7 17 Streptomycin biosynthesis Bms00521 9 18 Novobiocin biosynthesis Bms00401 4 19 Benzoate degradation Bms00362 20 20 Aminobenzoate degradation Bms00627 5 21 Chloroalkane and chloroalkene degradation Bms00625 10 22 Chlorocyclohexane and chlorobenzene degradation Bms00361 3 23 Xylene degradation Bms00622 1 24 Nitrotoluene degradation Bms00633 1 25 Styrene degradation Bms00643 3 26 Caprolactam degradation Bms00930 5 27 Naphthalene degradation Bms00626 6 28 Polycyclic aromatic hydrocarbon degradation Bms00624 3 29 Ribosome Bms03010 55 30 Phosphotransferase system Bms02060 4 31 Bacterial secretion Bms03070 27 32 Two-component system Bms02020 64 33 Quorum sensing Bms02024 114 34 Bacterial chemotaxis Bms02030 7 35 Flagellar assembly Bms02040 35 36 Beta-Lactam resistance Bms01501 17 37 Vancomycin resistance Bms01502 6 38 Cationic antimicrobial peptide (CAMP) resistance Bms01503 12 Fig. 1 Unique metabolic pathways. Schematic representation of unique metabolic pathways found in Brucella suis along with number proteins identified in it Prioritization of non-homologous proteins We subjected 503 unique metabolic pathways of Brucella suis to run a BLASTp with a cutoff value 10–3 against the whole proteome of humans to identify only non-homologous proteins for novel drug targets prioritization. We selected only non-homologous proteins to avoid the undesirable side effects of the drug. The BLASTp results showed 82 proteins were homologous to human host with high similarity with human proteome and were excluded. The remaining 421 non-homologous proteins were analyzed in next step. Identification of the essential proteins The Database of Essential Gene (DEG) provides complete information of the essentiality of proteins of the bacteria determined from the experimental analysis. Those proteins having high similarity with proteins of DEG were selected as essential proteins. We ran a BLASTp of non-homologous 421 proteins against the DEG database with a default parameter 60% sequence identity. The results showed that 350 proteins were essential for the survival of Brucella suis. Predicting the virulence factor (proteins) of B. suis The VFDB database comes up with complete information of protein virulence. The VFDB explored the importance of virulent proteins in disease progression. The VFDB results revealed that 45 proteins out of 350 (shortlisted proteins after the analysis of metabolic pathways associated proteins) were correlated with virulence of B. suis. However, these 45 proteins can be used as a novel and potent drug target against B. suis strain 1330. Identification of the resistance proteins The pathogens resistant to drugs are more hassle to treat the disease, which requires a higher dose that shows a diverse effect in patients. The pathogen acquires with a drug resistance is due to the continuous exposure to drug or the drug used in a higher dose. We subjected the FASTA sequence of shortlisted virulent proteins to ARG-ANNOT V6 online tool. The results showed that 42 out of 45 proteins were correlated with the resistivity of a pathogen. These 42 proteins are mostly involved in the degradation and efflux of numerous drugs. However, these 45 proteins can be used as a potential drug target. Prediction of subcellular localization All proteins require a specific location for their optimal function. Transporting proteins to unspecified region may result in server diseases [59]. Based on the subcellular localization of proteins, we design vaccine and drug against the specific localized protein target. The cytoplasmic proteins may act as drug target and outer membrane proteins may act as vaccine target. The results of current study showed ~ 23 cytoplasmic, ~ 14 periplasmic, ~ 6 outer membrane, and ~ 4 inner membrane proteins. Figure 2 represents the graphical sketch of sub cellular localization.Fig. 2 Sub-cellular localization. Quantitative representation of sub-cellular localization of shortlisted essential, druggable, pathogen-specific proteins predicted through PSORTb (A), and Cello2 (B) Non-homolog gut protein identification Different beneficial activities of the human microbiome were reported, and the link between gut flora and humans is not only commensal but symbiotic, mutualistic [60]. The host may experience negative consequences if proteins in this microbiota are blocked or inhibited inadvertently. Therefore, the BLATp was performed against the human gut microbial strains included in the HMP server, and the results indicated that only 2 of 42 proteins exhibited no similarity i.e., sn-glycerol-3-phosphate ABC transporter substrate-binding protein UgpB (WP_004686048.1, Periplasmic Protein), and multidrug efflux RND transporter outer membrane subunit BepC (WP_011068960.1, outermembrane). Because these proteins are not part of shared host–pathogen pathways and have no homology with human “anti-targets,” they are an appropriate target. Prediction of antigenic protein The antigenicity of identified outer membrane proteins (n = 6) were predicted to prioritize potential vaccine target against B. suis i.e., d-alanyl-d-alanine carboxypeptidase, Amino acid ABC transporter substrate-binding protein, and multidrug efflux RND transporter outer membrane subunit BepC. We uploaded these protein sequences to VaxiJen web server for the identification of antigenicity of outer membrane protein. The Vaxijen results characterized BepC as the most antigenic protein on the basis of antigenic prediction score which is 0.6511 making protein probable antigenic in nature. It was the only protein fully characterized as outermembrane protein from both the tools (Psortb and cello2) (Table 3). Out of six shortlisted proteins from the subtractive genomic analysis approach, we selected the outer-membrane protein Multidrug efflux Resistance-nodulation-division (RND) transporter outer membrane subunit (BepC gene) for the designing of multi-epitope vaccine. Hence, we may propose this protein for the designing of vaccine.Table 3 Identified six outermembrane shortlisted vaccine candidates against B. suis S. no Protein name Protein ID Psort subcellular localization Cello2 subcellular localization Antigenic score via VaxiJen 1 WP_002964109.1 d-alanyl-d-alanine carboxypeptidase Cytoplasmic Membrane 9.97 Outermembrane 0.3749 (probable non-antigen) 2 WP_002965100.1 d-alanyl-d-alanine carboxypeptidase Cytoplasmic Membrane 9.97 Outermembrane 0.6064 (probable antigen) 3 WP_004689042.1 amino acid ABC transporter substrate-binding protein Periplasmic 9.44 Outermembrane 0.5095 (probable antigen) 4 WP_004690853.1 d-alanyl-d-alanine carboxypeptidase Cytoplasmic Membrane 9.97 Outermembrane 0.5432 (probable antigen) 5 WP_006190489.1 d-alanyl-d-alanine carboxypeptidase Cytoplasmic Membrane 9.82 Outermembrane 0.4966 (probable non-antigen) 6 WP_011068960.1 multidrug efflux RND transporter outer membrane subunit BepC Outer Membrane 10.00 Outermembrane 0.6511 (probable antigen) Prediction of T-cell MHC-I epitopes The prediction of T-cell MHC-I was achieved by bioinformatics online tool NetCTL server by uploading the FASTA sequence of BepC protein into it. The NetCTL server predicted 448 T-cell epitopes in BepC protein. We selected 83 T-cell epitopes on the basis of high scores than a threshold of 0.2. Epitope’s sequence having a high predicted score representing the higher capability. These 83 predicted epitopes were then subjected to IEDB server to predict the binding affinity of these epitopes to MHC Class I. It resulted in the identification of 35 peptides with binding interaction with MHC1 molecules (Table S2). MHC I epitope immunogenicity prediction The epitopes present on the MHC molecules are recognized by CD + 8 to detect the aberrancies such as an infection. Several studies have shown that some peptides are more immunogenic than another peptide because of their amino acid sequence such as peptide having more aromatic amino acids are more immunogenic than other. The strength of the interaction between the peptide-MHC complexes (pMHC) TCRs depends both on the MHC I molecules and the presented peptide. The capability of epitope to stimulate T-cell responses depends on the level of immunogenicity score. The resultant epitopes from the above analysis were further subjected to IEDB server to determine the immunogenicity with a cut-off value of the positive score. The IEDB results showed that out of 35 epitopes 15 epitopes (after the removal of redundant sequences) were most immunogenic. Hence, we selected these immunogenic epitopes for further analysis as shown in Table 4.Table 4 predicted MHC-I epitopes and class-I immunogenicity analysis using IEDB S. no Protein ID Protein name Peptide MHC-I immunogenicity 1 WP_011068960.1 multidrug efflux RND transporter outer membrane subunit BepC KELVAAAVL DVKTAEATY NVAAAETQV KYAVNAAGY MLFDGFQTR ALSETLTGA YQLRQIAAL VAAAETQVF QLRQIAALR AMNEQVRAA HPGILATKY AVKDKWFGL NTASIGVGV ASRSTAIAA RGKTPATDY 0.17001 0.15711 0.1495 0.13759 0.12844 0.12697 0.11684 0.11247 0.09321 0.09125 0.07771 0.06943 0.06237 0.06014 0.04081 Antigenic, conservancy, and toxicity analysis For the assessment of toxicity, the level of toxicity was predicted by an online tool called as ToxinPred. The results showed that all selected 15 epitopes were not toxic to the host cell. Hence, we selected these epitopes for further analysis. Similarly, IEDB conserved sequence analysis tool was used for the assessment of the conserved sequence among all the genotype of MHC I epitopes. The conserved sequence identity parameters were set as a default. This analysis showed the conserved epitopes within the given protein sequence. The epitopes which showed 50% conserved sequence were selected as a conserved epitope. The result showed that all the 15 epitopes were shown 100% conserved sequence. Non-toxic and most conserved epitopes as a result of the ToxinPred and IEDB tool respectively were further analyzed for their antigenicity through VaxiJen with a cut-off value of 0.5. The results of Vaxijen server showed that 6 epitopes were found depicting the most antigenicity (Table 5). Hence, we selected these 6 epitopes as MHC I epitopes and these are DVKTAEATY, NVAAAETQV, MLFDGFQTR, ALSETLTGA, AMNEQVRAA, and NTASIGVGV.Table 5 Antigenicity, toxicity, and conservancy predicted for MHC I peptides S. no Protein Peptide Antigenicity Toxicity Conservancy Multidrug efflux RND transporter outer membrane subunit BepC (WP_011068960.1) DVKTAEATY NVAAAETQV MLFDGFQTR ALSETLTGA AMNEQVRAA NTASIGVGV 1.7278 1.1697 0.5787 0.5550 0.7509 1.3649 Non-toxic Non-toxic Non-toxic Non-toxic Non-toxic Non-toxic 100.00% 100.00% 100.00% 100.00% 100.00% 100.00% Prediction of MHC II epitopes In addition to MHC Class I epitope prediction, the BepC protein was used to predict MHC Class II using IEDB server. The epitopes having binding affinity < 200 nM and percentile ranks < 0.2 were shortlisted and used for further analysis. The results showed that total 11,934 epitopes were generated. We further shortlisted 10 epitopes by using the cut off value of 0.2 i.e., RSTAIAALNAARADV, SRSTAIAALNAARAD, STAIAALNAARADVK, TAIAALNAARADVKT, AIAALNAARADVKTA, ASRSTAIAALNAARA, CKELVAAAVLLSGTV, KELVAAAVLLSGTVL, ACKELVAAAVLLSGT, and KACKELVAAAVLLSG as shown in Table 6.Table 6 MHC-II epitopes predicted through IEBD Server S. no Protein Peptide MHC-II alleles 1 multidrug efflux RND transporter outer membrane subunit BepC (WP_011068960.1) RSTAIAALNAARADV HLA-DRB101:01 SRSTAIAALNAARAD HLA-DRB101:01 STAIAALNAARADVK HLA-DRB101:01 TAIAALNAARADVKT HLA-DRB101:01 AIAALNAARADVKTA HLA-DRB101:01 ASRSTAIAALNAARA HLA-DRB101:01 CKELVAAAVLLSGTV HLA-DQA101:02 KELVAAAVLLSGTVL HLA-DQA101:02 ACKELVAAAVLLSGT HLA-DQA101:02 KACKELVAAAVLLSG HLA-DQA101:02 MHC restriction cluster analysis of shortlisted epitopes Clusters of MHC-restricted allele and their appropriate peptides were analyzed through MHCcluster v2.0 for the confirmation of the predicted T-cell epitopes on the basis of the IC50 value. Moreover, the interacted alleles were re-evaluated by cluster analysis, and results are shown as a heat map (Fig. 3) and phylogenetic tree (Fig. S1) of MHC-1 and MHC II, respectively. Epitopes clustered are formed based on their interaction with the human leukocyte antigen (HLA). The yellow color shows weaker interactions whereas red color represents strong interactions with proper annotation.Fig. 3 Clustering analysis for MHC I and II epitopes. The cluster analysis of MHC molecules and HLA alleles (A), MHCI clustering alleles, (B) MHCII clustering alleles. Red color indicating strong interaction while the yellow zone indicates the weaker interaction B-cell epitope prediction in BepC In order to inflict humoral immunity, apart from the MHC-I and MHC-II epitopes (cellular immunity), B-cell epitopes were also identified using BepC protein sequences. For the elimination of pathogen, humoral immunity is also needed along with cellular immunity. The B-cell epitope prediction and classification plays a vital role in vaccine designing, antibody production, and immunodiagnostic tests. Identifying the B-cell epitopes experimentally is an expensive and time-consuming process while computational methods are highly desirable to predict the B-cell epitopes. Hence, the prediction of B-cell epitopes was performed using BCPreds, FBCpred, and ABCpred tools. The results showed that five epitopes were generated via BCPred server, twelve epitopes were generated through the FBCpred tool and twenty-eight epitopes were predicted via ABCpred server (Table S3). Predicted epitopes comparison for vaccine construct The predicted B-cell epitopes were manually compared and aligned against MHC I-II epitopes for the construction of the final chimeric vaccine designing. For the final vaccine models, the epitopes having sequences consists of overlapping T- and B-cell epitopes were selected. Finally, we shortlisted four peptides on the basis of their similarities among MHC-I, MHC-II, and B-cell epitopes i.e., GIQLNQMLFDGFQTR, DTIAGTDMGDGNTASIGVGV, TVFKACKELVAAAVL, and AQAEASRSTAIAALNA (Table 7).Table 7 Final predicted B-epitopes with comparison to MHC 1 and MHC II epitopes S. no B-cell epitopes MHC-I epitopes MHC-II epitopes 1 GIQLNQMLFDGFQTR MLFDGFQTR – 2 DTIAGTDMGDGNTASIGVGV NTASIGVGV – 3 YTVFKACKELVAAAVL – CKELVAAAVLLSGTV KELVAAAVLLSGTVL ACKELVAAAVLLSGT KACKELVAAAVLLSG 4 AQAEASRSTAIAALNA – ASRSTAIAALNAARA SRSTAIAALNAARAD ASRSTAIAALNAARA AIAALNAARADVKTA TAIAALNAARADVKT TAIAALNAARADVKT SRSTAIAALNAARAD RSTAIAALNAARADV Construction of final vaccine model An adjuvant and PADRE sequence (AKFVAAWTLKAAA) are the most significant component of a multi-epitope vaccine that inflict strong immune response in the host body [61, 62]. The PADRE sequences were chosen for the construction of vaccine in order to examine the adjuvants effects on the antigenicity and allergenicity and to combat the polymorphism complications of HLA-DR molecules caused in the global population. A total of sixteen vaccine constructs were joined with the EAAAK linker and respective adjuvants (HBHA, HBHA conserved, ribosomal, and beta defensin). The MHC Class-I, MHC Class-II, and B-cell epitopes were linked with GGGS, HEYGAEALERAG, and PADRE sequence. A study reported that all the epitopes were joined together with these two linkers do not change the conformation of designed vaccine construct [63] (Table S4). Prediction of antigenicity, allergenicity, and solubility for vaccine models The prediction of non-allergic vaccine constructs was performed by AlgPred online server. The constructed vaccine should not be allergenic, because allergic vaccines can induce a cross-reaction in the body. The allergenicity of all the sixteen vaccine constructs was checked and only those vaccines were selected which showed a value > 0.7. Similarly, the antigenicity and solubility of the designed multi-epitope vaccines were also determined by ANTIGENpro and SOLpro, respectively. The solubility of all the vaccine constructs was > 0.8. Finally, we selected only one vaccine to construct on the basis of allergenicity, antigenicity, and solubility for further analysis (Table S5). Physicochemical analysis of vaccine constructs The physicochemical properties of vaccine constructs were assessed by ProtParam server which includes a number of amino acids, molecular weight, aliphatic index, PI value, hydropathicity index, and instability index of the shortlisted vaccine constructs. The ProtParam results showed that the molecular weight of molded vaccine ~ 33 kDa, PI was calculated to be 5.03, having stability with 26.98 scores while GRAVY index was calculated to be −0.160 (Table S6). Construction of 3D structure of V1 We needed the 3-dimensional (3D) structure of vaccine construct to be functional. Hence, we modeled the 3D structure of vaccine via online tool Phyre2. We uploaded the FASTA sequence of vaccine to Phyre2. It compares the FASTA sequence of vaccine with earlier reported 3D structures in PDB database. The Phyre2 results provided us different modeled structures against different template proteins. Finally, we selected the modeled protein against the lipid-binding protein Ce-FAR-7 (2w9y) on the basis of percent identity and confidence score and modeled the vaccine structure (Fig. 4).Fig. 4 Vaccine structure modeling and validation. A The 3D model of a multi-epitope vaccine was obtained by Swiss model and B vaccine sequence Structure validation through Ramachandran plot (PROCHECK) The constructed 3D structure of the vaccine was validated by online server Procheck. The Procheck results revealed that 93.2% of residues found in the most accepted regions, and 6.8% residues found at additional allowed regions (Fig. 5a). The secondary structure (β-sheets, α-helices, and random coils) of constructed vaccine was analyzed PSIPRED. The results showed the same number of β-sheets, and α-helices in vaccine construct as modeled by Pyre2 tool (Fig. 5b).Fig. 5 Structure evaluation through PSIPRED and PROCHECK. A Shows structure confirmation for final vaccine construct generated through PSIPRED nearly same position of helixes and beta sheets as modeled structure whereas B modeled structure validation through Ramachandran plot using PROCHECK Molecular docking and molecular dynamic simulation A vaccine construct may have potency to build immune response against a different number of epitopes recognized by HLA allele’s proteins. Therefore, molecular docking of V1 with six different alleles such as 2FSE (HLA-DRB101:01), 3C5J (HLA-DR B302:02), 1H15(HLA-DR B501:01), 2Q6W (HLA-DR B301:01), 2SEB (HLA-DRB104:01), and 1A6A (HLA-DR B103:01) was performed through PatchDock and re-defined by FireDock analysis as shown in Table 8. Similarly, docking of TLR4 and design vaccine was performed by GRAMMX tool. The GRAMXX results showed different interactions of vaccine amino acids with TLR4 protein as shown in Fig. 6.Table 8 Docked score of HLA and vaccine (V1) model Vaccine construct HLA alleles (PDB: ID) SCORE AREA Hydrogen bond energy Global energy ACE V1 1A6A 16,028 1951.40 − 0.63 − 5.15 4.32 3C5J 14,778 2271.80 − 1.41 − 3.16 − 1.37 1H15 16,530 2338.40 − 3.10 − 4.67 8.67 2FSE 16,376 2078.00 − 5.14 − 21.75 10.77 2Q6W 16,080 2607.10 − 0.78 − 13.64 3.96 2SEB 15,476 1912.00 − 1.77 1.67 9.23 2Z65 15,970 2499.60 − 2.55 − 6.69 8.55 Fig. 6 Docked vaccine construct with TLR4/MD. A Docked complex of vaccine (red) and TL4/MD (purple) (B), interaction occurs between the vaccine model and TLR4/MD protein. Interacting residues of vaccine are represented in orange color, while protein-interacting residues highlighted in blue color, C all interactions found between the docked complexes i.e., blue lines represent hydrogen bonding, red color represents salt bridges Finally, molecular dynamic simulation of V1 performed by GROMACS tool showed the stability of vaccine models at 7 ns (Fig. 7a).Fig. 7 Molecular dynamics simulation of V1. A Root mean square deviation (RMSD) of the protein backbone, B potential energy of vaccine model, and C plot of the radius of gyration vs time during MDS Moreover, NMA stability of complex simulation resulted in the deformation graph illustrating the peaks (Fig. 8a). The eigenvalue detected for the complex was 1.08615e − 04 as shown in (Fig. 8b). The cumulative variance (green colored) and individual variance (red colored) are displayed by variance and B-factor graph visualizing the relation of the docked complex (Fig. 8C). The covariance map represents the motion between pair of residues of complex, where red color indicates the correlated motion, white color represents uncorrelated motion, and anti-correlated motion is represented by blue color (Fig. 8D). The complex’s elastic map shows the relation between the atoms and darker gray regions, indicating stiffer regions (Fig. 8E).Fig. 8 The results of molecular dynamics simulation of vaccine construct and TLR4/MD docked complex. A Deformability, B eigenvalues, C variance (red color indicates individual variances and green color indicates cumulative variances), and D co-variance map (correlated (red), uncorrelated (white), or anti-correlated (blue) motions) Codon optimization and cloning of chimeric vaccine construct (V1) The codon optimization of chimeric vaccine was performed by online tool JCAT. The vaccine FASTA sequence was reverse translated into a 950 base DNA sequence. The JCAT calculated the CAI score which was 1 and the GC content was 70.78% indicating the higher level of expression. The pET28a vector was used for its heterologous cloning and expression in E. coli by the Snapgene tool (Fig. 9).Fig. 9 Codon optimization and in-silico cloning of vaccine model. In silico restriction cloning of the multi-epitope vaccine sequence into the pET28a ( +) expression vector using Snapgene software, the red part represents the vaccine’s gene coding, and the black circle represents the vector backbone Discussion Understanding the proteome of a pathogen is important as it facilitates further comprehensive analysis of proteins in various biochemical and pathological pathways that help in the identification of novel drug targets. The development of a novel therapeutic target and vaccination is a significant scientific challenge to combat Brucella suis-related brucellosis [64]. Peptide-based vaccines are now possible to design due to the advancement of sequence-based technologies, computational analysis, and the abundance of genomes and proteomics data for many diseases [61, 62]. The subtractive genomics along with reverse vaccinology can aid in arraying the vast information regarding genomics and proteomics of various pathogens providing acceleration in drug and vaccines designing and pharmacogenomics in the treatments of bacterial infection. The “vaccinomics/Reverse Vaccinology” [65] approach has been proved to be promising approach widely used against Meningococcus B (MenB) [66], antibiotic-resistant Staphylococcus aureus [67], Chlamydia [68], group A Streptococcus [69], and Streptococcus pneumonia [70]. Brucellosis is characterized as acute fever illness associated with various symptoms in human such as liver and spleen disorders, reproductive abnormalities, neurological problems, heart-related problems, and also have been classified as a potential bioterrorism agent. There are 500,000 cases of human brucellosis reported per year around the globe due to their ability to survive and multiply within the host phagocytotic and non-phagocytotic cell. Due to the limited knowledge regarding the genomics of Brucella suis, new species-specific therapeutic compound and vaccine candidates are difficult to design experimentally. Therefore, the development of new vaccines model against Brucella suis is necessary. In the present study, we have applied reverse vaccinology and subtractive genomics based computational scheme to screen whole proteome of Brucella Suis for the identification of novel drug target and multi-epitope vaccine construction. We selected unique metabolic pathways composed of proteins unique to Brucella suis and excluded the common metabolic pathways present both in brucella and human-host since unique proteins are of interest. Total unique metabolic pathways consisted of 503 proteins followed by the foretelling of non-homologous protein from the complete proteome of Brucella Suis. Similarly, the essential proteins from the complete proteome of Brucella suis were determine which can be used as a potential drug target. The DEG analysis led to the identification of essential protein required for the survival of Brucella suis followed by the determination of virulence protein which are responsible for infection in human. Likewise, resistance protein identification was also performed. Resistance proteins are those which are responsible for the efflux of various antibiotic from the bacterial cell to counteract the drug action [71]. The subcellular localization of shortlisted proteins was also predicted for the identification of cytoplasmic protein for the drug targets and cell membrane protein for the construction of multi-epitope vaccine [72]. We shortlisted one protein namely Multidrug Efflux RND Transporter Outer Membrane Subunit BepC (involve in LPS metabolic pathway) for vaccine construction. Since this metabolic pathway is essential for bacteria but is absent in humans, BepC is being investigated as a possible vaccine candidate [73]. Surprisingly, Brucella did not show classical virulence mechanisms such as producing cytolysins, plasmids, fimbria, exotoxin, exoenzymes, and drug resistant forms. However, Brucella is having major virulence factors such as, lipopolysaccharide (LPS) [10]. This research employed a combination of prediction techniques to identify putative B- and T-cell epitopes that might trigger humoral or cell-mediated responses. Antibodies, also called immunoglobulins, are produced in large part by a kind of cell called a B-lymphocyte. Each of these epitopes takes on a linear or conformational shape [74]. Only in the protein’s basic structure can the linear B-cell fold be found in its entirety. It is protein folding that brings together discontinuous or conformational B-cell epitopes. This is an essential factor while developing vaccines [75, 76]. As a result of their ability to recognize and interact with MHC (Major Histocompatibility Complex) linked to antigen-presenting cells, T cells are classified as CD4 + and CD8 + cells, respectively (APCs). In summary, these APCs have surface antigens that are recognized by T-cell receptors [73]. The predicted peptide in present study were combined with the help of different linkers and adjuvant to construct chimeric based vaccine against Brucella suis. The addition of used adjuvant increased the immunogenicity of the vaccine formulation. Because of its immunomodulatory characteristics, this adjuvant is increasingly being employed in multiepitope vaccines [72]. Eventually, 16 vaccine constructs were designed against Brucella suis which were comprehensively investigated for toxicity profile, allergenicity pattern immunogenicity, and conservancy analysis resultant in only one final vaccine construct. The interaction of modeled vaccine with human leukocyte antigen (HLA) allele to interpret effective immune response was examine using molecular docking simulation studies with GRAMAXX and GROMACS tools. Protein–protein docking is frequently used in reverse vaccinology to identify the most promising vaccine design [72]. However, the results of docking analysis showed the binding affinity of promiscuous epitopes with different HLA alleles. An important TLR in mammals, TLR2/4 can identify lipoproteins from bacteria, viruses, fungi, and parasites [77]. Therefore, the shortlisted vaccine construct was used for the binding affinity and stability estimation of the of the vaccine and TLR4 complex. The JCAT tool was used for codon optimization. The JCAT calculated the CAI score which was 1 and the GC content was 70.78% indicating the higher level of expression. The pET28a vector was used for its heterologous cloning and expression in E. coli by the Snapgene tool. In order to confirm the findings of this research, it is recommended that the vaccine candidate be expressed in bacteria for further investigation. The inquiry would next be guided by pre-clinical studies, including tissue-culture or cell-culture systems, and animal experimentation. Conclusion The current study applied the subtractive genomics and reverse vaccinology approach for the prioritization of potent vaccine targets against Brucella suis 1300 strain. It smears multiple essential analyses at different stages i.e., non-homologs, essential, and unique to pathogens proteins, and sub cellular localization. In this study numbers of proteins along with Multidrug efflux RND transporter outer membrane subunit BepC was shortlisted as a novel vaccine target against Brucella suis. Consequently, the shortlisted essential proteins may be further study and used as a therapeutic vaccine candidate for Brucella Suis. Furthermore, immunogenicity prediction, allergenicity identification, and tertiary structure analysis of vaccine proposed it as a potent chimeric vaccine against Brucella suis. The molecular docking simulation and codon optimization were also found as satisfactory hence giving confidence to this study. It was found that the designed MEV in the present study can make stable interactions with human immune receptors and able to stimulate an efficient host immune system response. However, experimental validation with computational approaches is required for further analysis to improve the efficacy of predicted MEV. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 196 KB) Supplementary file2 (DOCX 22 KB) Acknowledgements The authors would like to acknowledge the Higher Education Commission of Pakistan for providing financial support under National Research Program for Universities. Data availability All data generated or analyzed during this study are included in this published article (and its supplementary information files). Declarations Ethical approval This article does not contain any studies with human participants and animals performed by any of the authors. Conflict of interest The authors declare that they have no conflict of interest. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Moreno E Cloeckaert A Moriyón I Brucella evolution and taxonomy Vet Microbiol 2002 90 1–4 209 227 10.1016/S0378-1135(02)00210-9 12414145 2. Pappas G The new global map of human brucellosis Lancet Infect Dis 2006 6 2 91 99 10.1016/S1473-3099(06)70382-6 16439329 3. Posadas DM BmaC, a novel autotransporter of Brucella suis, is involved in bacterial adhesion to host cells Cell Microbiol 2012 14 6 965 982 10.1111/j.1462-5822.2012.01771.x 22321605 4. Gyuranecz M Detection of Brucella canis–induced reproductive diseases in a kennel J Vet Diagn Invest 2011 23 1 143 147 10.1177/104063871102300127 21217047 5. Doganay GD Doganay M Brucella as a potential agent of bioterrorism Recent Pat Antiinfect Drug Discov 2013 8 1 27 33 10.2174/1574891X11308010006 22934672 6. Seleem MN Boyle SM Sriranganathan N Brucellosis: a re-emerging zoonosis Vet Microbiol 2010 140 3–4 392 398 10.1016/j.vetmic.2009.06.021 19604656 7. Schelling E Brucellosis and Q-fever seroprevalences of nomadic pastoralists and their livestock in Chad Prev Vet Med 2003 61 4 279 293 10.1016/j.prevetmed.2003.08.004 14623412 8. Fiori PL Brucella abortus infection acquired in microbiology laboratories J Clin Microbiol 2000 38 5 2005 2006 10.1128/JCM.38.5.2005-2006.2000 10790142 9. Hull NC Schumaker BA Comparisons of brucellosis between human and veterinary medicine Infect Ecol Epidemiol 2018 8 1 1500846 30083304 10. Lapaque N Brucella lipopolysaccharide acts as a virulence factor Curr Opin Microbiol 2005 8 1 60 66 10.1016/j.mib.2004.12.003 15694858 11. Martirosyan A Brucella β 1, 2 cyclic glucan is an activator of human and mouse dendritic cells PLoS Pathog 2012 8 11 e1002983 10.1371/journal.ppat.1002983 23166489 12. Lim JJ Protective effects of recombinant Brucella abortus Omp28 against infection with a virulent strain of Brucella abortus 544 in mice J Vet Sci 2012 13 3 287 10.4142/jvs.2012.13.3.287 23000585 13. Mirabella A Brucella melitensis MucR, an orthologue of Sinorhizobium meliloti MucR, Is involved in resistance to oxidative, detergent, and saline stresses and cell envelope modifications J Bacteriol 2013 195 3 453 465 10.1128/JB.01336-12 23161025 14. Głowacka P Brucella–virulence factors, pathogenesis and treatment Pol J Microbiol 2018 67 2 151 10.21307/pjm-2018-029 30015453 15. Ke Y Type IV secretion system of Brucella spp. and its effectors Front Cell Infect Microbiol 2015 5 72 10.3389/fcimb.2015.00072 26528442 16 Henriquez-Camacho C Proteins of Bartonella bacilliformis: candidates for vaccine development Int J Pept 2015 2015 1 5 10.1155/2015/702784 17. Barh D A novel strategy of epitope design in Neisseria gonorrhoeae Bioinform Biol Insights 2010 5 2 77 18. Lin X Li X Lin X A review on applications of computational methods in drug screening and design Molecules 2020 25 6 1375 10.3390/molecules25061375 32197324 19. Hasan M Reverse vaccinology approach to design a novel multi-epitope subunit vaccine against avian influenza A (H7N9) virus Microb Pathog 2019 130 19 37 10.1016/j.micpath.2019.02.023 30822457 20. Zhu H Big data and artificial intelligence modeling for drug discovery Annu Rev Pharmacol Toxicol 2020 60 573 589 10.1146/annurev-pharmtox-010919-023324 31518513 21. Lakshmi P Ramyachitra D Review about bioinformatics, databases, sequence alignment, docking, and drug discovery Statistical Modelling and Machine Learning Principles for Bioinformatics Techniques, Tools, and Applications 2020 Springer 11 23 22. Hasan M Vaccinomics strategy for developing a unique multi-epitope monovalent vaccine against Marburg marburgvirus Infect Genet Evol 2019 70 140 157 10.1016/j.meegid.2019.03.003 30849525 23. Uddin R Jamil F Prioritization of potential drug targets against P. aeruginosa by core proteomic analysis using computational subtractive genomics and Protein-Protein interaction network Comput Biol Chem 2018 74 115 122 10.1016/j.compbiolchem.2018.02.017 29587180 24. Sarangi AN Subtractive genomics approach for in silico identification and characterization of novel drug targets in Neisseria Meningitides Serogroup B J Comput Sci Syst Biol 2009 2 5 255 258 25. Mondal SI Identification of potential drug targets by subtractive genome analysis of Escherichia coli O157: H7: an in silico approach Adv Appl Bioinform Chem: AABC 2015 8 49 26677339 26. Uddin R Computational identification of potential drug targets against Mycobacterium leprae Med Chem Res 2016 25 3 473 481 10.1007/s00044-016-1501-6 27. Tatusova TA Karsch-Mizrachi I Ostell JA Complete genomes in WWW Entrez: data representation and analysis Bioinformatics (Oxford, England) 1999 15 7 536 543 10.1093/bioinformatics/15.7.536 10487861 28. UniProt: the universal protein knowledgebase in 2021. J Nucleic Acids Res. 2021; 49, no. D1 (2021): D480–D489. 29. Kanehisa M KEGG: integrating viruses and cellular organisms Nucleic Acids Res 2021 49 D1 D545 D551 10.1093/nar/gkaa970 33125081 30. Luo H DEG 15, an update of the Database of Essential Genes that includes built-in analysis tools Nucleic Acids Res 2021 49 D1 D677 D686 10.1093/nar/gkaa917 33095861 31. Liu B VFDB 2019: a comparative pathogenomic platform with an interactive web interface Nucleic Acids Res 2019 47 D1 D687 D692 10.1093/nar/gky1080 30395255 32. Gupta SK ARG-ANNOT, a new bioinformatic tool to discover antibiotic resistance genes in bacterial genomes Antimicrob Agents Chemother 2014 58 1 212 220 10.1128/AAC.01310-13 24145532 33 Yu NY PSORTb 3.0: improved protein subcellular localization prediction with refined localization subcategories and predictive capabilities for all prokaryotes Bioinformatics 2010 26 13 1608 1615 10.1093/bioinformatics/btq249 20472543 34. Yu CS Lin CJ Hwang JK Predicting subcellular localization of proteins for Gram-negative bacteria by support vector machines based on n-peptide compositions Protein Sci 2004 13 5 1402 1406 10.1110/ps.03479604 15096640 35. Peterson J The NIH human microbiome project Genome Res 2009 19 12 2317 2323 10.1101/gr.096651.109 19819907 36. Consutium A framework for human microbiome research Nature 2012 486 7402 215 221 10.1038/nature11209 22699610 37. Doytchinova IA Flower DR VaxiJen: a server for prediction of protective antigens, tumour antigens and subunit vaccines BioMed Central Bioinformatics 2007 8 1 1 7 17199892 38. Larsen MV Large-scale validation of methods for cytotoxic T-lymphocyte epitope prediction BioMed Central Bioinformatics 2007 8 1 1 12 17199892 39. Calis JJ Properties of MHC class I presented peptides that enhance immunogenicity PLoS Comput Biol 2013 9 10 e1003266 10.1371/journal.pcbi.1003266 24204222 40. Bui H-H Development of an epitope conservancy analysis tool to facilitate the design of epitope-based diagnostics and vaccines BioMed Central Bioinformatics 2007 8 1 1 6 17199892 41. Dash R In silico-based vaccine design against Ebola virus glycoprotein Adv Appl Bioinform Chem: AABC 2017 10 11 28356762 42. Karthik L Protease inhibitors from marine actinobacteria as a potential source for antimalarial compound PLoS ONE 2014 9 3 e90972 10.1371/journal.pone.0090972 24618707 43. Webb B, Sali A. Comparative protein structure modeling using MODELLER. Curr Protoc Bioinformatics. 2016;54(1):5.6. 1–5.6. 37. 44. Thomsen M MHCcluster, a method for functional clustering of MHC molecules Immunogenetics 2013 65 9 655 665 10.1007/s00251-013-0714-9 23775223 45. El-Manzalawy Y Dobbs D Honavar V Predicting flexible length linear B-cell epitopes Computational Systems Bioinformatics: (Volume 7) 2008 World Scientific 121 132 46. Barh D In silico subtractive genomics for target identification in human bacterial pathogens Drug Dev Res 2011 72 2 162 177 10.1002/ddr.20413 47. Ghaffari-Nazari H Improving multi-epitope long peptide vaccine potency by using a strategy that enhances CD4+ T help in BALB/c mice PLoS ONE 2015 10 11 e0142563 10.1371/journal.pone.0142563 26556756 48. Magnan CN High-throughput prediction of protein antigenicity using protein microarray data Bioinformatics 2010 26 23 2936 2943 10.1093/bioinformatics/btq551 20934990 49. Gasteiger E, Hoogland C, Gattiker A, Wilkins MR, Appel RD, Bairoch A. Protein identification and analysis tools on the ExPASy server. The proteomics protocols handbook; 2005, pp 571–607. 50. Waterhouse A SWISS-MODEL: homology modelling of protein structures and complexes Nucleic Acids Res 2018 46 W1 W296 W303 10.1093/nar/gky427 29788355 51. Bienert S The SWISS-MODEL Repository—new features and functionality Nucleic Acids Res 2017 45 D1 D313 D319 10.1093/nar/gkw1132 27899672 52. Mashiach E, et al. FireDock: a web server for fast interaction refinement in molecular docking. Nucleic Acids Res. 2008;36(suppl_2): W229–W232. 53. Tovchigrechko A, Vakser IA. GRAMM-X public web server for protein–protein docking. Nucleic Acids Res. 2006;34(suppl_2): W310–W314. 54 Pronk S GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit Bioinformatics 2013 29 7 845 854 10.1093/bioinformatics/btt055 23407358 55. López-Blanco JR iMODS: internal coordinates normal mode analysis server Nucleic Acids Res 2014 42 W1 W271 W276 10.1093/nar/gku339 24771341 56. Grote A, et al. JCat: a novel tool to adapt codon usage of a target gene to its potential expression host. Nucleic Acids Res. 2005;33(suppl_2):W526–W531. 57. Roop RM Uncovering the hidden credentials of Brucella virulence Microbiol Mol Biol Rev 2021 85 1 e00021 e119 10.1128/MMBR.00021-19 33568459 58. Coloma-Rivero RF Brucella and its hidden flagellar system Microorganisms 2021 10 1 83 10.3390/microorganisms10010083 35056531 59. Shurety W Localization and post-Golgi trafficking of tumor necrosis factor-alpha in macrophages J Interferon Cytokine Res 2000 20 4 427 438 10.1089/107999000312379 10805378 60. Khan MT, Mahmud A, Hasan M, Azim KF, Begum MK, Rolin MH, Mondal SI. Proteome exploration of legionella pneumophila to identify novel therapeutics: a hierarchical subtractive genomics and reverse vaccinology approach. Microbiol Spectr. 2022;10(4):e00373-22. 61. Jalal K Identification of vaccine and drug targets in Shigella dysenteriae sd197 using reverse vaccinology approach Sci Rep 2022 12 1 1 19 10.1038/s41598-021-03988-0 34992227 62. Khan K Jalal K Uddin R An integrated in silico based subtractive genomics and reverse vaccinology approach for the identification of novel vaccine candidate and chimeric vaccine against XDR Salmonella typhi H58 Genomics 2022 114 2 110301 10.1016/j.ygeno.2022.110301 35149170 63. Solanki V Tiwari V Subtractive proteomics to identify novel drug targets and reverse vaccinology for the development of chimeric vaccine against Acinetobacter baumannii Sci Rep 2018 8 1 1 19 10.1038/s41598-018-26689-7 29311619 64. Scoffone VC Vaccines to overcome antibiotic resistance: the challenge of Burkholderia cenocepacia Trends Microbiol 2020 28 4 315 326 10.1016/j.tim.2019.12.005 31932141 65. Donati C Rappuoli R Reverse vaccinology in the 21st century: improvements over the original design Ann N Y Acad Sci 2013 1285 1 115 132 10.1111/nyas.12046 23527566 66. Mora M Reverse vaccinology Drug Discovery Today 2003 8 10 459 464 10.1016/S1359-6446(03)02689-8 12801798 67. Sharma S. Identificationof vaccine candidates against Staphylococcus Aureus: an in silico reverse vaccinology approach (Doctoral dissertation). 2019. 68. Adu-Bobie J Two years into reverse vaccinology Vaccine 2003 21 7–8 605 610 10.1016/S0264-410X(02)00566-2 12531326 69. Seib KL Zhao X Rappuoli R Developing vaccines in the era of genomics: a decade of reverse vaccinology Clin Microbiol Infect 2012 18 109 116 10.1111/j.1469-0691.2012.03939.x 22882709 70. Talukdar S Identification of potential vaccine candidates against Streptococcus pneumoniae by reverse vaccinology approach Appl Biochem Biotechnol 2014 172 6 3026 3041 10.1007/s12010-014-0749-x 24482282 71. Basharat Z Differential analysis of Orientia tsutsugamushi genomes for therapeutic target identification and possible intervention through natural product inhibitor screening Comput Biol Med 2022 141 105165 10.1016/j.compbiomed.2021.105165 34973586 72. Alotaibi G Pan genome based reverse vaccinology approach to explore Enterococcus faecium (VRE) strains for identification of novel multi-epitopes vaccine candidate Immunobiology 2022 227 3 152221 10.1016/j.imbio.2022.152221 35483110 73. Jalal K Pan-genome reverse vaccinology approach for the design of multi-epitope vaccine construct against Escherichia albertii Int J Mol Sci 2021 22 23 12814 10.3390/ijms222312814 34884620 74 Sanchez-Trincado JL Gomez-Perosanz M Reche PA Fundamentals and methods for T-and B-cell epitope prediction J Immunol Res 2017 2017 1 14 10.1155/2017/2680160 75. Jespersen MC, et al. BepiPred-2.0: improving sequence-based B-cell epitope prediction using conformational epitopes. Nucleic Acids Res. 2017;45(W1):W24–W29. 76. Moodley A, et al. Reverse vaccinology approach to design a multi-epitope vaccine construct based on the Mycobacterium tuberculosis biomarker PE_PGRS17. Immunol Res. 2022:1–17. 77. Ardestani H In silico and in vivo approaches to recombinant multi-epitope immunogen of GroEL provides efficient cross protection against S. Typhimurium, S. flexneri, and S. dysenteriae Mol Immunol 2022 144 96 105 10.1016/j.molimm.2022.02.013 35217247	36459272	PMC9716126	NO-CC CODE	2022-12-03 23:20:54	no	Immunol Res. 2022 Dec 2;:1-20	utf-8	Immunol Res	2,022	10.1007/s12026-022-09346-0	oa_other
==== Front Atmos Pollut Res Atmos Pollut Res Atmospheric Pollution Research 1309-1042 Turkish National Committee for Air Pollution Research and Control. Production and hosting by Elsevier B.V. S1309-1042(22)00301-4 10.1016/j.apr.2022.101620 101620 Article Lessons learnt for air pollution mitigation policies from the COVID-19 pandemic: The Italian perspective D'Isidoro Massimo a1 D'Elia Ilaria a∗1 Vitali Lina a1 Briganti Gino a Cappelletti Andrea a Piersanti Antonio a Finardi Sandro b Calori Giuseppe b Pepe Nicola b Di Giosa Alessandro c Bolignano Andrea c Zanini Gabriele a a ENEA – Italian Agency for New Technologies, Energy and Sustainable Economic Development, Bologna, Italy b Arianet s.r.l., Milano, Italy c ARPA-Lazio Environmental Protection Agency of the Lazio Region, Rome, Italy ∗ Corresponding author. 1 Authors equally contributed to the paper. 2 12 2022 12 2022 2 12 2022 13 12 101620101620 8 8 2022 1 12 2022 1 12 2022 © 2022 Turkish National Committee for Air Pollution Research and Control. Production and hosting by Elsevier B.V. All rights reserved. 2022 Turkish National Committee for Air Pollution Research and Control Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Policies to improve air quality need to be based on effective plans for reducing anthropogenic emissions. In 2020, the outbreak of COVID-19 pandemic resulted in significant reductions of anthropogenic pollutant emissions, offering an unexpected opportunity to observe their consequences on ambient concentrations. Taking the national lockdown occurred in Italy between March and May 2020 as a case study, this work tries to infer if and what lessons may be learnt concerning the impact of emission reduction policies on air quality. Variations of NO2, O3, PM10 and PM2.5 concentrations were calculated from numerical model simulations obtained with business as usual and lockdown specific emissions. Both simulations were performed at national level with a horizontal resolution of 4 km, and at local level on the capital city Rome at 1 km resolution. Simulated concentrations showed a good agreement with in-situ observations, confirming the modelling systems capability to reproduce the effects of emission reductions on ambient concentration variations, which differ according to the individual air pollutant. We found a general reduction of pollutant concentrations except for ozone, that experienced an increase in Rome and in the other urban areas, and a decrease elsewhere. The obtained results suggest that acting on precursor emissions, even with sharp reductions like those experienced during the lockdown, may lead to significant, albeit complex, reduction patterns for secondary pollutant concentrations. Therefore, to be more effective, reduction measures should be carefully selected, involving more sectors than those related to mobility, such as residential and agriculture, and integrated on different scales. Graphical abstract Image 1 Keywords Air pollution COVID-19 Lockdown Air quality policies Air quality modelling ==== Body pmc1 Introduction Air pollution represents the biggest environmental threat to human health together with climate change. The World Health Organization (WHO) estimates 7 million of premature deaths every year due to air pollution exposure and in September 2021 reviewed and updated its Air Quality Guidelines lowering the recommended limit values of air pollutants for public health protection (WHO, 2021). Efforts to improve air quality have also benefits in contrasting climate change, and the adoption of policies and measures reducing greenhouse gases emissions can in turn improve air quality. When in early 2020 the world had to face the spread of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), different lockdown measures were put in place that lead to an unforeseen decrease in air pollutant and greenhouse gases emissions (Gkatzelis et al., 2021). From this point of view, the tragic coronavirus disease 2019 (COVID-19) pandemic, which led governments to act with drastic interventions to limit its spread, can be used as a test bench to assess the effectiveness of policies based on anthropogenic emission reduction. Indeed, during the COVID-19 induced lockdowns, some human activity sectors were affected rather than others, with very important emission variations in just a few weeks. By estimating those variations, simulating their effects on air quality, and checking them against monitoring data, we can quantify the expected impact on air quality of hypothetical long-term policies, acting on those sectors, with similar variations. Therefore, the first wave of the pandemic offers a unique opportunity, although as an unforeseen and side effect, to understand how air quality can be modified if measures to reduce air pollution emissions are adopted at different scales, from regional to local ones (Sokhi et al., 2021). On the other hand, this unprecedented real-world experiment raised the interest of many researchers from different fields in studying the link between air pollution and COVID-19. As Heederik et al. (2020) pointed out, some papers were accepted for publication after a very short review period abandoning the rigorous peer review process, rushing the dissemination of several findings (Villeneuve and Goldberg, 2020) and suggesting that sometimes the observed link is a correlation rather than a causation (Riccò et al., 2020). If the evaluation of the relations between air pollution and SARS-CoV-2 is still in progress, with new researches showing more rigorous analyses of previous papers (e.g. Kogevinas et al. (2021)), many studies have demonstrated that the unprecedented national lockdowns, travel restrictions and the closing of national borders during the first wave of the pandemic in 2020 caused short-term pollutant concentration reductions. Several analyses regarding the effect of lockdowns on air quality were conducted at different scales using various approaches such as machine learning, statistical analysis, satellite data and numerical modelling. A review can be found in Silva et al. (2022) and Gkatzelis et al. (2021) and at the following link https://amigo.aeronomie.be/covid-19-publications/peer-reviewed. At global level, Sokhi et al. (2021) evaluated air quality changes comparing 2020 with the period 2015–2019 using ground-based air quality observations over 63 cities, covering 25 countries, and observing a decrease up to 70% for nitrogen dioxide (NO2), between 30% and 40% for particulate matter with a diameter of 2.5 μm or less (PM2.5) even if the signal resulted to be rather complex in different cities and among the same regions, while ozone (O3) showed no or small increases. In Europe, Barré et al. (2021) estimated the lockdown effects on NO2 concentrations using satellite data, surface site measurements and regional ensemble air quality modelling simulations from the Copernicus Atmosphere Monitoring Service (CAMS), with an estimated NO2 concentration reductions of 23%, 43% and 32%, respectively; Schneider et al. (2022) quantified the effect of different lockdown measures on concentrations of NO2, O3, particulate matter with a diameter of 10 μm or less (PM10) and PM2.5 across 47 European cities using CAMS modelling data, showing a general decrease in NO2 and PM concentrations with a different behaviour for O3 whose concentrations presented limited reductions and a slight increase in April, and also estimating the number of avoided deaths attributable to single activity reductions. Numerous studies were also conducted in specific European countries, for example Velders et al. (2021) quantified the lockdown effects on PM and gaseous pollutants in the Netherlands using a numerical model and a random forest technique, estimating a reduction range of 8–30% for NO2 and 5–20% for PM2.5; von Schneidemesser et al. (2021) estimated the effects of traffic lockdown measures on NO2 levels using both observations and models in Berlin with the median observed NO2 concentration reduction of around 40%, while in Spain Querol et al. (2021) studied the effects of COVID-19 on air quality for different pollutants in 11 cities, finding a marked decreased in NO2 concentrations, low decrease in PM levels, and different urban O3 behaviour with a slight decrease in rural areas. In general, all the studies agreed on the overall significant reduction of NO2 concentrations, leading to a slight increase of O3 levels in urban areas, while the PM signal is rather complex among different cities and places. In Italy, one of the countries most affected by the first wave of the pandemic, many studies to assess the impact of lockdown on air pollutant concentrations were performed: Campanelli et al. (2021) used in-situ observations, aerosol and gas column measurements and satellite data to analyse NO2 air concentrations at five urban sites distributed over the whole Italian territory; Putaud et al. (2021) analysed the impact of COVID-19 lockdown on air quality at regional and urban background sites in northern Italy; Bontempi et al. (2022) in the city of Brescia (Lombardy region, northern Italy), while Cucciniello et al. (2022) in the city of Avellino (southern Italy). Whereas previous studies were mainly limited to regions or cities, our work represents the first study covering the entire Italian territory, where a high-resolution chemical transport model (CTM), validated with ground-based observations, was applied to integrate different scales. Moreover, the use of a CTM enabled us to evaluate the emission variations of each lockdown restriction measure and to estimate their effects in terms of air pollutant concentrations. Our aim is to infer from the pandemic lockdowns what messages we can take home in defining air pollution control programs to provide useful information to policymakers for the definition of future abatement policies. In this respect, our approach is to relate COVID-19 induced emission reductions in some sectors with air pollutant concentrations, aiming at assessing the effectiveness of emission reduction policies in air quality regulations. To this scope, we used the case study of Italy national lockdown between February and May 2020. Numerical simulations at different horizontal spatial resolutions were conducted to evaluate the effects on air pollution: at national level with a resolution of 4 km and in the case of the capital city Rome with a resolution of 1 km. The model evaluation performed with pollutant concentrations, observed before and during the lockdown, provides a validation of the chemical transport model as a tool to estimate the air quality impact of emission reduction measures in air quality plans, that is not achievable through feasible real-world experiments. Data and methods are presented in Section 2, the results are shown in Section 3 with a brief discussion, and some conclusions are drawn in Section 4. 2 Data and methods To evaluate the changes in NO2, O3, PM10 and PM2.5 concentrations from the emission reductions caused by COVID-19 lockdown measures enforced in Italy, we used an approach based on three-dimensional atmospheric modelling. Air quality simulations were performed from January to May 2020 using the Flexible Air quality Regional Model (FARM; Gariazzo et al., 2007; Silibello et al., 2008; Kukkonen et al., 2012) Chemical Transport Model (CTM) with different setups over two domains, at lower (4 km) and higher (1 km) horizontal resolution, covering the whole Italy and the city of Rome, respectively. Modelling domains are depicted in Fig. 1 .Fig. 1 Domains considered in the present study: Italian national domain with a resolution of 4 km (on the left, green border) and domains of the Lazio regional system with the inner Rome domain having a resolution of 1 km (on the right, red border, the black star indicating the city location). Fig. 1 2.1 Simulation over Italy The study at national scale was carried out by means of the Atmospheric Modelling System of the MINNI project (AMS-MINNI; Mircea et al., 2014; 2016; D'Elia et al., 2021) whose main components are the Weather Research and Forecasting (WRF, Version 4.1.2, Skamarock et al., 2019) meteorological model, the EMission Manager (EMMA; ARIA/ARIANET, 2013) emission processor, and the FARM CTM. Two different air quality simulations were conducted, covering the period from January to May 2020, over the 4 km resolution grid: the first one (BASE) made use of the business-as-usual anthropogenic emission inventory, while the second one (LOCK) was set up to include the lockdown-induced emission reductions in different sectors. The same meteorological conditions, described in the following Section, were used to drive the two simulations. Two different AMS-MINNI air quality simulations, produced within CAMS over Europe and consistently referred to business-as-usual and lockdown scenarios, were used as initial conditions and to force pollutant concentrations at lateral boundaries. More details on emissions and boundary conditions can be found in Sections 2.1.2 and 2.1.3. 2.1.1 Meteorology Hourly meteorological fields over Italy for the period of interest were reconstructed using the WRF model, considering two nested domains (2-way mode) at 12 and 4 km resolution, respectively, the former over Central Europe and the latter covering Italy. ERA5 (Hersbach et al., 2020) reanalyses fields were used as initial and boundary conditions for the coarser grid. Details on WRF setup are summarised in Table 1 .Table 1 WRF setup used for the simulation over Italy. Table 1Model version WRF v4.1.2 (Skamarock et al., 2019) Domain Central Europe Italy Horizontal resolution 12 km 4 km Number of grid points in x 204 388 Number of grid points in y 190 418 Number of vertical sigma-hybrid levels 35 35 Microphysics WRF Single Moment 6-class scheme (Hong and Lim, 2006) Cumulus Parameterization Kain–Fritsch Convective Parameterization (Kain, 2004) Explicit convection PBL Scheme Mellor Yamada Jancic (MYJ; (Janjić, 1994) Surface layer Monin-Obukhov/Janjic Eta (Janjić, 1996) Land Surface Noah LSM (Land Surface Model; (Liu et al., 2006) Longwave Radiation RRTMG (Iacono et al., 2008) Shortwave Radiation RRTMG (Iacono et al., 2008) 2.1.2 Emissions Many studies analysed the impact of COVID-19 restrictions on air quality using different approaches, from in situ observations to satellite data. The estimation of those impacts using an air quality modelling system requires the quantification of emission variations per source and pollutant, that is also important to provide significant information to policymakers in designing future abatement policies and measures. Quantifying an emission variation during the lockdown period is not a straightforward task which is affected by large uncertainties, due to types and timing of restrictions in different areas, or different individual responses, among countries and regions (Guevara et al., 2021; Rodríguez-Sánchez et al., 2022). In this study, we followed the approach defined in Guevara et al. (2021, 2022) and Rodríguez-Sánchez et al. (2022). We firstly estimated a baseline emission inventory for the year 2020 (BASE), developed from the official national emission inventory elaborated by the Italian Institute for Environmental Protection and Research (ISPRA; Taurino et al. (2022)) with a topdown approach on a provincial level (NUTS3 level, where NUTS stands for Nomenclature of Territorial Units for Statistics, https://ec.europa.eu/eurostat/web/nuts/background). The topdown emission inventory was compared and harmonized with the regional emission inventories made available by the EU Life Prepair project (https://www.lifeprepair.eu/) for the Po Valley Regions (northern Italy) and by ARPA Lazio for the Lazio Region (central Italy; https://www.arpalazio.it/ambiente/aria/inventario-regionale-delle-emissioni-in-atmosfera), producing for that regions an inventory with a bottomup approach. Concerning the dust emissions from road transport resuspension, to be coherent on the entire 4 km domain, the methodology of the EPA algorithm AP-42 (http://www.epa.gov/ttn/chief/ap42/ch13/bgdocs/b13s0201.pdf) and the emission factors from Amato et al. (2012) were used, not including the regional estimates. The emission inventory is referred to 2017, being the most recent year with available official emission estimates both at national and regional level, and we assumed that it could represent BASE emissions for 2020. An exception was the residential sector, for which the heating degree days sum of 2017 and 2020 was used to update the emissions to the year 2020, accounting for the effect of the different meteorology. To be used in FARM CTM, the provincial emission inventory was then spatially disaggregated to the 4 km grid, hourly modulated and speciated by EMMA. To quantify the emission variation due to the COVID-19 restrictions put in place all over Italy, emission adjustment factors were built following a data-driven approach (Guevara et al., 2021, 2022; Rodríguez-Sánchez et al., 2022). As in the studies already cited, we assumed that changes in emissions follow changes in the main activity data. Data from different sources were analysed and compared. For each of the following sector, classified considering the European Selected Nomenclature for Air Pollution (SNAP), we prepared a dataset of adjustment factors to be applied to BASE emission inventory in order to build the LOCK emission inventory: energy industry (SNAP 01), manufacturing industry (SNAP 03 and 04), civil sector (SNAP 02, divided in commercial/institutional and residential sub-sectors), road transport (SNAP 07, divided in light and duty vehicles), maritime transport (SNAP 0804) and aviation (SNAP 0805). A summary of the activity data considered for each sector, with temporal variation (daily, weekly, or monthly) and data source is reported in table S1 of the Supplementary Material (SM). 2.1.3 Air quality The BASE and LOCK air quality simulations were performed running FARM on the target domain reported in Fig. 1, having a horizontal resolution of about 4 km (0.045° latitude x 0.035° longitude). FARM was forced by the meteorological fields and anthropogenic emissions produced as described in the previous Sections, interpolated on the model grid. The 0.15° × 0.1° horizontal resolution concentration fields over Europe from AMS-MINNI, one of the models participating in CAMS71 COVID19 exercise (Schneider et al., 2022), were used as boundary and initial conditions. FARM operates off-line, downstream with respect to meteorology, taking as input the WRF hourly fields to calculate transport, diffusion, and transformations of the atmospheric pollutants. The meteorological post-processor SURFPRO (SURFace-atmosphere interface PROcessor; AriaNet srl, 2011) was used to interface the two models, using WRF simulated quantities to produce a more detailed description of boundary layer structure and to estimate micrometeorological parameters (turbulence scales like Reynolds stresses and scale temperature), horizontal and vertical diffusivities, deposition velocities for gaseous species and natural emissions (biogenic, soil dust and marine aerosols) on the simulation grid. Table 2, Table 3 summarize the settings adopted for FARM and SURFPRO, respectively.Table 2 FARM setup used for the simulation over Italy. Table 2Version FARM v5.1 Domain Italy (see Fig. 1) Horizontal resolution ∼4 km (0.045° × 0.035° lat-lon) Number of grid points in x 388 Number of grid points in y 418 Number of fixed terrain-following vertical levels 14 (from 20 m up to 6290 m) Advection scheme Blackman cubic polynomials (Yamartino, 1993) Vertical levels 16 fixed terrain following layers with a quasi-logarithmic distribution. Meteorological fields are provided on the centre of the cells. Lower layer thickness 40 m Concentrations are provided at this height. Domain top 10 km Chemical mechanism SAPRC-99 (Carter, 2000) Aerosol model AERO3 (Binkowski and Roselle, 2003) Dynamics is described by three growing and interacting lognormal modes: nucleation (Aitken), accumulation and coarse. Cloud chemistry Aqueous SO2 simplified chemistry (Seinfeld and Pandis, 1998) Inorganic chemistry ISORROPIA v1.7 (Nenes et al., 1998) Organic chemistry SORGAM (Schell et al., 2001) Boundary conditions From MINNI/CAMS COVID-19 exercise Table 3 SURFPRO setup used for the simulation over Italy. Table 3Version SURFPRO v3.3 Land use Corine Land Cover 2006 with 22 classes Biogenic VOC MEGAN 2.04 (Guenther et al., 2006) NOx from soil MEGAN 2.04 (Guenther et al., 2006) Sea salt emissions Zhang et al. (2005) Windblown dust Vautard et al. (2005) Vertical diffusivity Kz-closure approach following Lange (1989) Horizontal diffusivity combination of two methods: Smagorinsky (1963) and scale function depending on the local stability class and wind speed (sum of the two values). Dry deposition Resistance model based on (Wesely, 1989) Wet deposition for both gases and aerosols species In-cloud and sub-cloud scavenging coefficients following EMEP (Simpson et al., 2003) Mixing height computation scheme Bulk Richardson number method (with fixed Ric = 0.25) 2.2 High resolution simulation over rome The study over Rome shared the air quality modelling system used in the national scale analysis, including FARM CTM, EMMA and SURFPRO. Moreover, the air quality boundary conditions were extracted from the national scale simulations as the best available option to describe the long-range transport of pollutants from other areas of the country and from further European regions. The local scale simulation differed from the national scale one for what concerns meteorology and emission reconstruction, consistently performed according to what employed in ARPA Lazio operational air quality forecast system configuration over Lazio region and Rome metropolitan area (https://qa.arpalazio.net/, in Italian), also used in year-by-year air quality assessments. The meteorology was reconstructed by the Regional Atmospheric Modelling System (RAMS, Cotton et al., 2003), while emission estimates were based on Lazio Region emission inventory for the reference year 2017. Two different air quality simulations were performed on a two-way nested modelling system covering the Lazio Region with 4 km resolution, and an inner area spanning 60 × 60 km centred over Rome city with a resolution of 1 km (Fig. 1 ). Adopting the same approach described for the national scale analysis, the first simulation (BASE) made use of business-as-usual anthropogenic emissions described by the regional emission inventory, while the second one (LOCK) included the lockdown-induced emission reductions. The two corresponding national scale AMS-MINNI air quality simulations were used to define pollutant concentrations at lateral boundaries. The simulations covered the whole year 2020 to support the ARPA Lazio institutional activities in the yearly air quality assessment, while the analysis presented here is focused on the period from 1st January to May 3, 2020 to highlight lockdown related emission reduction impact. More details on meteorology and emissions are given in the following Sections.Fig. 2 Top: contribution of each sector (affected by the restrictions) to total BASE emissions over the entire period (Feb–May 2020). Middle: total emission variations (LOCK-BASE)/BASE [%] due to lockdown measures. Bottom: emission variations over the entire period for each sector. Fig. 2 Fig. 3 NO2 concentration differences (in μg/m3) between LOCK and BASE in April 2020: Italian domain at 4 km resolution (left) and zoomed over Rome (top, right); 1 km resolution simulation over Rome (bottom, right). Gray lines represent Rome main road network. Fig. 3 Fig. 4 As in Fig. 3 but referred to O3 concentration differences. Fig. 4 Fig. 5 As in Fig. 3 but referred to PM10 concentration differences. Fig. 5 Fig. 6 As in Fig. 3 but referred to PM2.5 concentration differences. Fig. 6 2.2.1 Meteorology Meteorological fields over Lazio Region and Rome were reconstructed by the application of the RAMS model on four two-way nested domains at 32, 16, 4 and 1 km resolution, to downscale the US National Centers for Environmental Prediction GFS (Global Forecast System (https://www.ncei.noaa.gov/products/weather-climate-models/global-forecast); global meteorological forecast over the target urban area of Rome. Details on the RAMS setup are summarised in Table 4 .Table 4 RAMS setup used for the simulation down to Lazio Region and Rome. Table 4Model version RAMS v6 (Cotton et al., 2003) Domain Central Mediterranean Italy Lazio Region Rome Horizontal resolution 32 km 16 km 4 km 1 km Number of grid points in x 54 58 66 70 Number of grid points in y 54 58 58 70 Number of vertical sigma-hybrid levels 32 32 32 32 Microphysics Two-moment bulk scheme (Meyers et al., 1997; Walko et al., 1995) Cumulus Parameterization Modified Kuo – (Tremback, 1990) Modified Kuo – (Tremback, 1990) Explicit convection Explicit convection PBL Scheme Mellor-Yamada level 2.5 scheme – (Mellor and Yamada, 1982) Surface layer Surface layer similarity theory (Louis, 1979) Land Surface Soil – vegetation – snow parameterization (LEAF-3) (Walko et al., 2000) Longwave Radiation Harrington (1997) long/shortwave model – two stream scheme interacts with liquid and ice hydrometeor size spectra and with dust particles Shortwave Radiation Harrington (1997) long/shortwave model – two stream scheme interacts with liquid and ice hydrometeor size spectra and with dust particles 2.2.2 Emissions The emission input for 4 and 1 km resolution air quality simulations was prepared using the same emission processor explained in Section 2.1.2, starting from the data of the ARPA Lazio inventory for the reference year 2017 (https://www.arpalazio.it/ambiente/aria/inventario-regionale-delle-emissioni-in-atmosfera), available at municipality level. Dust emissions from road transport resuspension were estimated using the EPA algorithm AP-42 (https://www.epa.gov/sites/default/files/2020-10/documents/13.2.1_paved_roads.pdf). As done for the national scale simulations, the inventory data were used for the BASE case, and the LOCK case was generated by applying daily based emission reductions per sector, also using the same adjustment factors of the national scale. 3 Results and discussion In the following paragraphs, we firstly discuss the effect of lockdown measures on emissions (Section 3.1) and then on air quality (Section 3.2) both on national and Rome domain. 3.1 Emissions On a national level, the comparison between LOCK and BASE emission scenarios (reported in Section S2 of the SM, Fig. 1, Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6) shows a significant variation in nitrogen oxides (NOx), sulphur oxides (SOx) and non-methane volatile organic compound (NMVOC) emissions from the second part of March to the end of April with a daily reduction that reaches 30%, while in May, when the restrictions were less stringent, a slight emission recovery is foreseen. Minor variations are estimated for PM2.5 (similar considerations can be extended also to PM10) and carbon monoxide (CO), while no significant variations are estimated for ammonia (NH3), as the agricultural sector, the main contributor for this pollutant, was not interested by the restrictions. Overall, in the entire period, we estimated an average emission reduction of 15% for SOx, 11% for NOx, 10% for NMVOC, 4% for CO, 2% for PM2.5, and no reduction for NH3 (Fig. 2, middle). For each pollutant, Fig. 2 shows the contribution to the total BASE emissions coming from all the sectors affected by the restrictions (top histogram) and the estimated emission variation (LOCK-BASE)/BASE during the first wave of the pandemic (middle). The variation occurred in the entire period in each sector for each pollutant are presented as well (bottom). For NOx emission, the sectors affected by the restrictions contributed to nearly 60% of total BASE emissions (Fig. 2, top), with the road transport sector being the most contributing one, followed by industrial, maritime and energy sectors, whose reductions respect to their sectorial BASE value are −22%, −25%, −20% and −16%, respectively (bottom panel). Indeed, industrial, maritime and energy sectors show high sectoral variation, but their contribution to the total emission variation is limited to a few percentage points (middle panel). The emission variations for PM2.5, CO and NH3 are low (middle), as their respective dominant sectors (top panel) were either not affected (agriculture for NH3) or increased their activity rates (residential heating for PM2.5 and CO, bottom panel). The sectors involved by the COVID-19 measures contributed to 53% of total NMVOC emissions (top panel), with the industrial sector reduced by 24% (bottom panel) and driving the overall NMVOC reduction (middle panel). 3.2 Air quality An in-depth evaluation of the LOCK simulation was firstly carried out. Different statistical indices were calculated as suggested by literature on model validation (Chang and Hanna, 2004), to quantify the agreement between modelled and observed values. The statistical metrics were calculated for the background monitoring stations satisfying the criteria of 75% of valid data per month for all the period (Jan–May 2020). Valid stations for the three zones urban (UB), suburban (SB) and rural (RB) and for each pollutant were grouped by 9 climatic zones and depicted in Figure SF3.1 of the SM. The results for mean bias (MB), root mean square error (RMSE) and the correlation coefficient (corr) are reported in Section S3 of the SM Figures SF3.2-9, along with their formulations. The analysis shows that, regardless of the station zone, for both NO2 and O3 the highest correlation values are found in the Po Valley (Pianura Padana and Alto Adriatico, the latter being the coastal area of the Po Valley), where also the higher PM10 and PM2.5 values are observed. This result is noteworthy as Po Valley, representing a hotspot for air pollution in Italy, was particularly affected by lockdown measures, and good model skills in correlation values indicate the robustness of the temporal modulation adopted in the emissions and in the applied reductions. Indeed, the comparison between observed and simulated time series of daily average concentrations (Section 4 in the SM) underlines as the modelling systems, with few exceptions occurring in less polluted areas, are in good agreement with the observed time evolution over Italy (4 km resolution, Figures SF4.1.1-12) as well as in Rome (4 and 1 km resolutions, Figures SF4.2.2-9). This behaviour, during the lockdown period when substantial and rapid emission variations occurred, confirms that the modelling systems as a whole can be valuable tools in analysing the effects of reduction measures implied by air quality policies, assessing cross-sectoral synergies and trade-offs. This assessment is a peculiar result enabled by the sharp emissions reduction occurred during the lockdown that cannot be obtained from model evaluation using observations gathered by feasible real-world experiments. Once validated the model results, simulated concentrations of NO2, O3, PM10 and PM2.5 are presented here focussing only on the month of April 2020, when the most stringent restrictions were fully in place all over Italy. This choice allows us to avoid possible misinterpretation in our results by considering a larger period, only partially interested by restrictions, and when the highest air pollutant concentration variations were observed. As a reference for the whole simulated months, the reader can refer to the SM (Section S4). In the following paragraphs, for each pollutant we present the results in terms of average concentration difference between LOCK and BASE simulated concentrations over Italy and over Rome at 4 and 1 km resolution, respectively. 3.2.1 NO2 In Section S4.1.1 (SM), a detailed comparison between NO2 observed and simulated (LOCK) concentrations is presented at national level, as the daily time series computed as the median of all the background stations data, grouped by climatic zone (Figure SF4.1.1). The modelling system gives reasonably good results in most areas, except mainly in the mountains (Arco Alpino and Appennino, i.e., Alps and Apennines) where the 4 km horizontal resolution is not enough to resolve the effects induced by the complex orography, as well as in some southern areas of the country where the number of available stations is too limited to provide an extensive description of background concentrations. As mentioned before, it is worth noting that the model performs reasonably well in the Po Valley. A similar behaviour is observed over the Rome domain (see S4.2.1 in SM for reference), where both simulations tend to well reproduce the time evolution (SF4.2.2), but some differences can be observed: if we focus on April (SF4.2.3), high resolution simulation tends to produce lower NO2 concentrations than the 4 km resolution one at the selected sites, in some cases underestimating the observed values. Fig. 3 shows the average NO2 concentration difference (LOCK-BASE) in April 2020 for the 4 km resolution simulations at national level (left) and zoomed over Rome area (top, right), as well as the same difference related to 1 km resolution domain (bottom, right). At Italian level, a reduction of NO2 concentrations is observed, with the highest differences exceeding 12 μg/m3 in northern Italy (mainly the Po Valley) and in general within the major urban areas and along the main road axes; far from these zones, the concentration variations are very small or negligible. A similar pattern is shown both in the 4 and 1 km resolution over the Rome domain (top right and bottom right in Fig. 3, respectively): in this case, increasing the resolution does not dramatically change the absolute magnitude of the variations, but highlights spatial details reflecting the most urbanised areas of the city and its main road network. 3.2.2 O3 In Section S4.1.2 (SM), a detailed comparison between observed and simulated (LOCK) ozone concentrations is presented at national level for daily maximum 8 h average (MDA8) time series, as the median value of all the background stations grouped by climatic zone (Figure SF4.1.4). Excluding the Alps and the Apennines where, as noted before, the model resolution is not adequate for resolving complex orography, a good agreement with measurements is generically shown, especially where high pollution levels are observed (e.g., Po Valley). Moreover, from figure SF4.1.4 (in SM) it is worth noting that the ozone concentration increase observed in April is well reproduced by the model. Similar behaviour is noted in Rome (Section S4.2.2 in SM), where the two simulations give similar results, well reproducing the ozone trend (SF4.2.4–5). The average ozone air concentration difference for April 2020 is depicted in Fig. 4. At the national scale (left panel), positive values in the map indicate that lockdown measures led to an increase of ozone concentrations in main urban areas and close to the main roads, while a reduction is shown in rural areas. The same behaviour is also evident in the zoom over Rome. As in the case of NO2, also in the case of ozone the 1 km resolution simulation (bottom, right in the Figure) highlights more details of the urban area with respect to the 4 km resolution one (top, right). In both cases, maximum increments reach values of 10–12 μg/m3. These values are compatible with the estimate obtained comparing 2020 observations with 2015–2019 monthly mean concentrations (as e.g. Sokhi et al., 2021) at six urban background stations, that provided values ranging from +4 to +16 μg/m3. The observed increment of ozone concentrations in urban areas at both resolutions, combined to the corresponding NO2 decrease, reflects the VOC-limited regime occurring in European urban areas, as showed for example by Beekmann and Vautard (2010). The decrease of NOx associated to an increase of ozone was also pointed out in other works investigating air pollution related to COVID-19 pandemic (e.g. Brancher, 2021; Kroll et al., 2020). As also highlighted in Grange et al. (2021), the NOx-limited or VOC-limited regime is a crucial aspect to be considered in planning effective strategies for ozone pollution control. In this regard, numerical modelling is the only tool capable of providing a comprehensive picture at national scale to support policymakers. Our modelling analysis confirms that the increase of ozone concentration observed during the lockdown in many urban locations (see e.g. Gkatzelis et al., 2021; Sokhi et al., 2022) is limited to the cities, major roads and large conurbation areas, while rural background areas show different behaviours, with prevailing ozone concentration decrease, in agreement with the observational analyses provided by Cristofanelli et al. (2021) and Steinbrecht et al. (2021) and with model simulation results obtained for central to southern Europe by Matthias et al. (2021). 3.2.3 PM In SM, a detailed comparison between observed and simulated PM10 and PM2.5 concentrations is presented over Italian (SM Sections S4.1.3, S4.1.4) and Rome (S4.2.3, S4.2.4) domains for daily time series, at all the background stations. Generally, the modelled concentrations show a good agreement with measurements at both resolutions, concerning the temporal variation and timing of the peaks. In general, over Rome the 4 km simulation produces lower concentrations than the high resolution one, especially during winter months (SF4.2.6, SF4.2.8). It is worth noting that an outlier is present in the observations at the end of March, not captured by the model neither at national level nor over Rome (Figures SF4.1.7 and SF4.2.6). This event was caused by long-range dust transport from the Caspian Sea area (Campanelli et al., 2021), a contribution that was not included at the model domain boundary as not described through boundary conditions, while the concentration peak observed in mid-May in central to southern Italian regions, and contributed by Sahara dust advection (see e.g. https://dust.aemet.es/) is correctly reproduced by the model (Figures SF4.1.7). In the city of Rome, larger differences occur in the first part of April, when the model underestimates observed PM values. In this case, the difference can be explained by intense fires in the Balkan region. This is confirmed by back-trajectories calculated by means of M-TrraCE (Vitali et al., 2017) from 4th to 14th of April using the hourly 4 km meteorological fields produced by WRF. As an example, Figure SF4.2.10 in SM shows the 8th of April as the most representative day of the whole event. Moreover, the descent of the airmass from the Balkans, showed in the Figure, is consistent with the synoptic pattern in the area with the presence of a high-pressure system favouring subsidence conditions, as highlighted by Campanelli et al. (2021). Concerning PM10 fields, the difference of average simulations for April 2020 highlights a general reduction at national level (Fig. 5, left panel), with maximum values in the Po Valley, varying between 4 and 6 μg/m3. In other parts of the country, the difference is much less pronounced, except in few major urban areas like Rome (top, right panel) where an average reduction of 2–4 μg/m3 is reached. In the 1 km simulations (bottom, right panel), the area within the reduction range 2–4 μg/m3 is wider, and in the city centre and surrounding main roads the concentration reduction exceeds 6 μg/m3. Looking at the difference of average simulated PM2.5 fields for April 2020 in Fig. 6, it is worth noting that the level of concentration reductions is close to that found for PM10, suggesting the predominance of the fine component over the coarse one in the modelled concentrations, as also confirmed by the comparison with observations (see SM Section S4). 3.3 Discussion The results presented in this Section highlight the connection between emission reductions in some sectors and the air quality in urban and rural areas. By using a numerical modelling system, we showed that the effect of emission reductions on air pollutant concentrations may change dramatically when looking at different pollutants and/or at urban or rural areas. In the case of COVID-19 induced lockdown, the road transport sector experienced the highest variation, leading the 11% total NOx emission reduction. The effects of such reductions were largely visible on the closely interrelated NO2 and O3 ambient concentrations showing opposite behaviour in urban areas, due to the occurring VOC-limited chemical regime, a crucial aspect to be considered when elaborating emission control policies. Conversely, for other pollutants, such as PM, changes in modelled concentrations were less pronounced, confirming the prominent role of the emissions from the heating sector, only slightly affected by changes during the lockdown, and from the agricultural sector not affected by the restrictions. The impact of long-range transport of desert dust and wildfires was also identified during short term episodes occurred during the studied period. It is worthwhile considering that long-range transport and secondary pollutant formation cannot be prevented or controlled by local policies but require measures coordinated over larger scales. Investigating the model response to emission scenarios at different resolutions, 4 km over Italy and 1 km over Rome, revealed a substantial coherence in the sign of the concentration variations, their spatial patterns, and their magnitude, even though at higher resolution more spatial details were obtained. Comparison with observed time series over Rome showed that the model performs well in terms of temporal evolution and timing of the peaks at both resolutions, while at some sites, on specific periods/days, the two simulations can differ. This different behaviour could be explained by the use of two different meteorological drivers, as illustrated by Adani et al. (2020). It is worth noting that the model best performances compared to observations were obtained in the Po Valley, a hotspot for atmospheric pollution in Italy, where the lockdown effects on air quality were larger and where we used a bottom-up emission inventory (see Section 2.1.2). We highlight the crucial role played by state-of-the-art air quality modelling systems that allow to study and quantify the impact of emissions on air concentration variations. This feature is of the utmost importance in supporting policymakers when designing future emission abatement policies over large geographical areas and at different scales, from national down to city level. Such results, even if affected by uncertainties, can only be reached operating these comprehensive tools that integrate atmospheric dynamics, thermodynamics, and chemistry. 4 Conclusions Many papers have been written to investigate the effects of the COVID-19 pandemic on air quality, but a study covering the whole of Italy, one of the most affected areas during the first wave of the pandemic, was missing. In our work, we presented two different numerical simulations, related to the business as usual and the lockdown-specific emissions, with a horizontal spatial resolution of 4 km over Italy and of 1 km over a smaller domain around Rome. The measures adopted by the different governments to contrast the SARS-CoV-2 spread represent an unprecedented and unique experiment to better understand the effects on air quality of substantial emission reductions, the peculiarities of the different pollutants, the key factors in the definition of air pollution mitigation policies. The first question we tried to answer was whether air quality modelling systems could well reproduce the effects of large and rapid emission variations occurred in a relatively short period (two months in our analysis). Our results showed a good agreement with observations both in terms of statistical scores and time series behaviour, indicating that models can be a valuable tool in supporting the analysis and the screening of air quality policies. We also investigated the effects of the different lockdown measures on the variation of air pollution levels and how these measures affect pollutants whose formation derives from complex physical and chemical transformations, such as PM and O3. To this aim, we elaborated and compared two different simulations, the BASE and LOCK case, sharing the same input data, but emissions and boundary conditions. The highest emission reductions occurred in the mobility sector, where road transport is the main contributor to NOx emissions. This caused an estimated reduction of 11% of total NOx emissions, that in the month of April 2020 led to a maximum reduction of 12 μg/m3 in NO2 concentrations, especially in urban areas and along the main roads. The NO2 reduction in urban areas led to an increase in O3 concentrations due to the titration process, underlining that, when defining O3 reduction measures, the prevalent chemical regime should be carefully considered. The industrial sector mostly affected SOx and NMVOC emissions but with moderate absolute variations, while the much less pronounced PM concentration reduction was linked to the increase in the emissions from the residential sector, especially during the month of March and in the first half of April, and to the lack of NH3 emission variations, mostly caused by the agricultural sector not affected by the restrictions. This study also confirms the direct link between NO2 concentration variations and the reduction in the emission precursors (NOx), while the same is not true for PM, being its concentrations linked to a broader set of precursors (like ammonia from the agricultural sector). The results from the higher resolution simulations over Rome, broadly showing the same patterns of the coarser simulations but better detailing their shape over and around the urban area, evidenced as synergies at different scales are required also in the policy definition. This study demonstrated that even considerable actions in few sectors, like the transport or industrial ones, may have distinct effectiveness in abating concentrations of different air pollutants. Moreover, measures limited in time jeopardize all the efforts: Bartoňová et al. (2022), for example, demonstrates that, when averaged over the year 2020, the effects on air quality of the pandemic measures, that were gradually lifted after the first wave, will be substantially smaller than those of April 2020. The experimented abatement of traffic related emissions can be considered representative of local measures, like e.g., the urban traffic electrification, that can be expected to strongly reduce NO2 concentrations (Piersanti et al., 2021), but slightly impact pollutants influenced by long-range transport and secondary formation, which require large scale international emission reduction and prevention policies. An integrated approach should be adopted by policymakers in defining reduction measures where policies in different fields, such as energy, climate and air quality, and different aspects, from economic to societal, must be considered in a holistic way. Author contribution statement Massimo D’Isidoro: Conceptualization; Methodology; Software; Validation; Formal Analysis; Investigation; Data Curation; Writing – Original Draft; Writing – Reviewing and Editing; Visualization; Ilaria D’Elia: Conceptualization; Methodology; Software; Validation; Formal Analysis; Investigation; Data Curation; Writing – Original Draft; Writing – Reviewing and Editing; Visualization; Lina Vitali: Conceptualization; Methodology; Software; Validation; Formal Analysis; Investigation; Data Curation; Writing – Original Draft; Writing – Reviewing and Editing; Visualization; Gino Briganti: Software; Validation; Formal Analysis; Investigation; Data Curation; Writing – Reviewing and Editing; Visualization; Andrea Cappelletti: Software; Data Curation; Writing – Reviewing and Editing; Antonio Piersanti: Conceptualization; Formal Analysis; Investigation; Writing – Reviewing and Editing; Sandro Finardi: Conceptualization; Methodology; Software; Validation; Writing – Original Draft; Writing – Reviewing and Editing; Giuseppe Calori: Software; Validation; Investigation; Writing – Reviewing and Editing; Nicola Pepe: Software; Validation; Investigation; Writing – Reviewing and Editing; Alessandro Di Giosa: Writing – Reviewing and Editing; Andrea Bolignano: Software; Validation; Investigation; Writing – Reviewing and Editing; Gabriele Zanini: Writing – Reviewing and Editing; Funding Acquisition Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Supplementary data The following is the Supplementary data to this article:Multimedia component 1 Multimedia component 1 Acknowledgements We would like to express out thanks to the ENEA colleagues Dr. Mario Adani for providing the boundary conditions for the national air quality simulations, and Dr. Luisella Ciancarella, Dr. Giandomenico Pace and Alcide Di Sarra for sharing with us their useful ideas and comments. We acknowledge the PULVIRUS Project that funded this activity (http://www.pulvirus.it/) and the colleagues from the National and Regional Environmental Agencies involved in the Project. The computing resources and the related technical support used for the national simulations were provided by the CRESCO/ENEAGRID High Performance Computing infrastructure and its staff (Iannone et al., 2019). The infrastructure is funded by 10.13039/501100022400 ENEA , the Italian National Agency for New Technologies, Energy and Sustainable Economic Development and by Italian and European research programmes (http://www.cresco.enea.it/english). Peer review under responsibility of Turkish National Committee for Air Pollution Research and Control. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.apr.2022.101620. ==== Refs References Adani M. Piersanti A. Ciancarella L. D'Isidoro M. Villani M.G. Vitali L. Preliminary tests on the sensitivity of the FORAIR_IT air quality forecasting system to different meteorological drivers Atmosphere 11 2020 10.3390/atmos11060574 Amato F. Karanasiou A. Moreno T. Alastuey A. Orza J.A.G. Lumbreras J. Borge R. Boldo E. Linares C. Querol X. Emission factors from road dust resuspension in a Mediterranean freeway Atmos. Environ. 61 2012 580 587 10.1016/j.atmosenv.2012.07.065 ARIA/ARIANET Emission Manager - processing system for model-ready emission input - user's guide R2013 Milano 19 2013 No. R2013.19) AriaNet srl SURFPRO3 User's Guide (SURFace-Atmosphere Interface PROcessor 2011 Version 3) Barré J. Petetin H. Colette A. Guevara M. Peuch V.-H. Rouil L. Engelen R. Inness A. Flemming J. Pérez García-Pando C. Bowdalo D. Meleux F. Geels C. Christensen J.H. Gauss M. Benedictow A. Tsyro S. Friese E. Struzewska J. Kaminski J.W. Douros J. Timmermans R. Robertson L. Adani M. Jorba O. Joly M. Kouznetsov R. Estimating lockdown-induced European NO2 changes using satellite and surface observations and air quality models Atmos. Chem. Phys. 21 2021 7373 7394 10.5194/acp-21-7373-2021 Bartoňová A. Colette A. Zhang H. Fons J. Liu H.-Y. Brzezina J. Chantreux A. Couvidat F. Guerreiro C. Guevara M. Kuenen J. Solberg S. Super I. Szanto C. Tarasson L. Thornton A. Ortiz A.G. The Covid-19 Pandemic and Environmental Stressors in Europe: Synergies and Interplays 2022 ETC/ATNI Report Kjeller 2021/16 Binkowski F.S. Roselle S.J. Models-3 community multiscale air quality (CMAQ) model aerosol component 1. Model description J. Geophys. Res. 108 2003 10.1029/2001jd001409 Bontempi E. Carnevale C. Cornelio A. Volta M. Zanoletti A. Analysis of the lockdown effects due to the COVID-19 on air pollution in Brescia (Lombardy) Environ. Res. 212 2022 113193 10.1016/j.envres.2022.113193 Brancher M. Increased ozone pollution alongside reduced nitrogen dioxide concentrations during Vienna's first COVID-19 lockdown: significance for air quality management Environ. Pollut. 284 2021 117153 10.1016/j.envpol.2021.117153 Campanelli M. Iannarelli A.M. Mevi G. Casadio S. Diémoz H. Finardi S. Dinoi A. Castelli E. di Sarra A. Di Bernardino A. Casasanta G. Bassani C. Siani A.M. Cacciani M. Barnaba F. Di Liberto L. Argentini S. A wide-ranging investigation of the COVID-19 lockdown effects on the atmospheric composition in various Italian urban sites (AER – LOCUS) Urban Clim. 39 2021 100954 10.1016/j.uclim.2021.100954 Carter W.P.L. Documentation of the SAPRC-99 chemical mechanism for VOC reactivity assessment Final Report to California Air Resources Board, Contract 92–329 2000 Contract 95–308 Chang J.C. Hanna S.R. Air quality model performance evaluation Meteorol. Atmos. Phys. 87 2004 167 196 10.1007/s00703-003-0070-7 Cotton W.R. Pielke R.A. Sr. Walko R.L. Liston G.E. Tremback C.J. Jiang H. McAnelly R.L. Harrington J.Y. Nicholls M.E. Carrio G.G. McFadden J.P. RAMS 2001: current status and future directions Meteorol. Atmos. Phys. 82 2003 5 29 10.1007/s00703-001-0584-9 Cristofanelli P. Arduni J. Serva F. Calzolari F. Bonasoni P. Busetto M. Maione M. Sprenger M. Trisolino P. Putero D. Negative ozone anomalies at a high mountain site in northern Italy during 2020: a possible role of COVID-19 lockdowns? Environ. Res. Lett. 16 2021 074029 10.1088/1748-9326/ac0b6a Cucciniello R. Raia L. Vasca E. Air quality evaluation during COVID-19 in Southern Italy: the case study of Avellino city Environ. Res. 203 2022 111803 10.1016/j.envres.2021.111803 D'Elia I. Briganti G. Vitali L. Piersanti A. Righini G. D'Isidoro M. Cappelletti A. Mircea M. Adani M. Zanini G. Ciancarella L. Measured and modelled air quality trends in Italy over the period 2003-2010 Atmos. Chem. Phys. 21 2021 10825 10849 10.5194/acp-21-10825-2021 Gariazzo C. Silibello C. Finardi S. Radice P. Piersanti A. Calori G. Cecinato A. Perrino C. Nussio F. Cagnoli M. Pelliccioni A. Gobbi G.P. Di Filippo P. A gas/aerosol air pollutants study over the urban area of Rome using a comprehensive chemical transport model Atmos. Environ. 41 2007 7286 7303 10.1016/j.atmosenv.2007.05.018 Gkatzelis G.I. Gilman J.B. Brown S.S. Eskes H. Gomes A.R. Lange A.C. McDonald B.C. Peischl J. Petzold A. Thompson C.R. Kiendler-Scharr A. The global impacts of COVID-19 lockdowns on urban air pollution: a critical review and recommendations Elementa: Science of the Anthropocene 9 2021 00176 10.1525/elementa.2021.00176 Grange S.K. Lee J.D. Drysdale W.S. Lewis A.C. Hueglin C. Emmenegger L. Carslaw D.C. COVID-19 lockdowns highlight a risk of increasing ozone pollution in European urban areas Atmos. Chem. Phys. 21 2021 4169 4185 10.5194/acp-21-4169-2021 Guenther A. Karl T. Harley P. Wiedinmyer C. Palmer P.I. Geron C. Estimates of global terrestrial isoprene emissions using MEGAN (model of emissions of gases and aerosols from nature) Atmos. Chem. Phys. 6 2006 3181 3210 10.5194/acp-6-3181-2006 Guevara M. Jorba O. Soret A. Petetin H. Bowdalo D. Serradell K. Tena C. Denier van der Gon H. Kuenen J. Peuch V.-H. Pérez García-Pando C. Time-resolved emission reductions for atmospheric chemistry modelling in Europe during the COVID-19 lockdowns Atmos. Chem. Phys. 21 2021 773 797 10.5194/acp-21-773-2021 Guevara M. Petetin H. Jorba O. Denier van der Gon H. Kuenen J. Super I. Jalkanen J.-P. Majamäki E. Johansson L. Peuch V.-H. Pérez García-Pando C. European primary emissions of criteria pollutants and greenhouse gases in 2020 modulated by the COVID-19 pandemic disruptions Earth Syst. Sci. Data 14 2022 2521 2552 10.5194/essd-14-2521-2022 Harrington J.Y. The Effects of Radiative and Microphysical Processes on Simulated Warm and Transition Season Arctic Stratus 1997 Colorado State University Heederik D.J.J. Smit L.A.M. Vermeulen R.C.H. Go slow to go fast: a plea for sustained scientific rigour in air pollution research during the COVID-19 pandemic Eur. Respir. J. 56 2020 10.1183/13993003.01361-2020 Hersbach H. Bell B. Berrisford P. Hirahara S. Horányi A. Muñoz-Sabater J. Nicolas J. Peubey C. Radu R. Schepers D. Simmons A. Soci C. Abdalla S. Abellan X. Balsamo G. Bechtold P. Biavati G. Bidlot J. Bonavita M. De Chiara G. Dahlgren P. Dee D. Diamantakis M. Dragani R. Flemming J. Forbes R. Fuentes M. Geer A. Haimberger L. Healy S. Hogan R.J. Hólm E. Janisková M. Keeley S. Laloyaux P. Lopez P. Lupu C. Radnoti G. de Rosnay P. Rozum I. Vamborg F. Villaume S. Thépaut J.-N. The ERA5 global reanalysis Q. J. R. Meteorol. Soc. 146 2020 1999 2049 10.1002/qj.3803 Hong S.Y. Lim J.-O.J. The WRF single-moment 6-class microphysics scheme (WSM6) J. Kor. Meteorol. Soc. 42 2006 129 151 Iacono M.J. Delamere J.S. Mlawer E.J. Shephard M.W. Clough S.A. Collins W.D. Radiative forcing by long-lived greenhouse gases: calculations with the AER radiative transfer models J. Geophys. Res. Atmos. 113 2008 10.1029/2008JD009944 Iannone F. Ambrosino F. Bracco G. De Rosa M. Funel A. Guarnieri G. Migliori S. Palombi F. Ponti G. Santomauro G. Procacci P. CRESCO ENEA HPC clusters: a working example of a multifabric GPFS Spectrum Scale layout 2019 International Conference on High Performance Computing Simulation (HPCS). Presented at the 2019 International Conference on High Performance Computing Simulation 2019 HPCS) 1051 1052 10.1109/HPCS48598.2019.9188135 Janjić Z.I. The Surface Layer in the NCEP Eta Model. Presented at the Eleventh Conference on Numerical Weather Prediction 1996 Amer. Meteor. Soc. Norfolk, VA 354 355 Janjić Z.I. The step-mountain eta coordinate model: further developments of the convection, viscous sublayer, and turbulence closure schemes Mon. Weather Rev. 122 1994 927 945 10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO 2 Kain J.S. The kain–fritsch convective parameterization: an update J. Appl. Meteorol. Climatol. 43 2004 170 181 10.1175/1520-0450(2004)043<0170:TKCPAU>2.0.CO 2 Kogevinas M. Castaño-Vinyals G. Karachaliou M. Espinosa A. de C.R. Garcia-Aymerich J. Carreras A. Cort és B. Pleguezuelos V. Jimenez A. Vidal M. O'Callaghan-Gordo C. Cirach M. Santano R. Barrios D. Puyol L. Rubio R. Izquierdo L. Nieuwenhuijsen M. Dadvand P. Aguilar R. Moncunill G. Dobaño C. Tonne C. Ambient air pollution in relation to SARS-CoV-2 infection, antibody response, and COVID-19 disease: a cohort study in catalonia, Spain (COVICAT study) Environ. Health Perspect. 129 2021 117003 10.1289/EHP9726 Kroll J.H. Heald C.L. Cappa C.D. Farmer D.K. Fry J.L. Murphy J.G. Steiner A.L. The complex chemical effects of COVID-19 shutdowns on air quality Nat. Chem. 12 2020 777 779 10.1038/s41557-020-0535-z 32843750 Kukkonen J. Olsson T. Schultz D.M. Baklanov A. Klein T. Miranda A.I. Monteiro A. Hirtl M. Tarvainen V. Boy M. Peuch V.-H. Poupkou A. Kioutsioukis I. Finardi S. Sofiev M. Sokhi R. Lehtinen K.E.J. Karatzas K. San José R. Astitha M. Kallos G. Schaap M. Reimer E. Jakobs H. Eben K. A review of operational, regional-scale, chemical weather forecasting models in Europe Atmos. Chem. Phys. 12 2012 1 87 10.5194/acp-12-1-2012 Lange R. Transferability of a three-dimensional air quality model between two different sites in complex terrain J. Appl. Meteorol. Climatol. 28 1989 665 679 10.1175/1520-0450(1989)028<0665:TOATDA>2.0.CO 2 Liu Y. Chen F. Warner T. Basara J. Verification of a mesoscale data-assimilation and forecasting system for the Oklahoma city area during the joint urban 2003 field project J. Appl. Meteorol. Climatol. 45 2006 912 929 10.1175/JAM2383.1 Louis J.-F. A parametric model of vertical eddy fluxes in the atmosphere Boundary-Layer Meteorol. 17 1979 187 202 10.1007/BF00117978 Matthias V. Quante M. Arndt J.A. Badeke R. Fink L. Petrik R. Feldner J. Schwarzkopf D. Link E.-M. Ramacher M.O.P. Wedemann R. The role of emission reductions and the meteorological situation for air quality improvements during the COVID-19 lockdown period in central Europe Atmos. Chem. Phys. 21 2021 13931 13971 10.5194/acp-21-13931-2021 Mellor G.L. Yamada T. Development of a turbulence closure model for geophysical fluid problems Rev. Geophys. 20 1982 851 875 10.1029/RG020i004p00851 Meyers M.P. Walko R.L. Harrington J.Y. Cotton W.R. New RAMS cloud microphysics parameterization. Part II: the two-moment scheme Atmos. Res. 45 1997 3 39 10.1016/S0169-8095(97)00018-5 Mircea M. Ciancarella L. Briganti G. Calori G. Cappelletti A. Cionni I. Costa M. Cremona G. D'Isidoro M. Finardi S. Pace G. Piersanti A. Righini G. Silibello C. Vitali L. Zanini G. Assessment of the AMS-MINNI system capabilities to simulate air quality over Italy for the calendar year 2005 Atmos. Environ. 84 2014 178 188 10.1016/j.atmosenv.2013.11.006 Mircea M. Grigoras G. D'Isidoro M. Righini G. Adani M. Briganti G. Ciancarella L. Cappelletti A. Calori G. Cionni I. Cremona G. Finardi S. Larsen B.R. Pace G. Perrino C. Piersanti A. Silibello C. Vitali L. Zanini G. Impact of grid resolution on aerosol predictions: a case study over Italy Aerosol Air Qual. Res. 16 2016 1253 1267 10.4209/aaqr.2015.02.0058 Nenes A. Pandis S.N. Pilinis C. ISORROPIA: a new thermodynamic equilibrium model for multiphase multicomponent inorganic aerosols Aquat. Geochem. 4 1998 123 152 10.1023/A:1009604003981 Piersanti A. D'Elia I. Gualtieri M. Briganti G. Cappelletti A. Zanini G. Ciancarella L. The Italian national air pollution control programme: air quality, health impact and cost assessment Atmosphere 12 2021 196 10.3390/atmos12020196 Putaud J.-P. Pozzoli L. Pisoni E. Martins Dos Santos S. Lagler F. Lanzani G. Dal Santo U. Colette A. Impacts of the COVID-19 lockdown on air pollution at regional and urban background sites in northern Italy Atmos. Chem. Phys. 21 2021 7597 7609 10.5194/acp-21-7597-2021 Querol X. Massagué J. Alastuey A. Moreno T. Gangoiti G. Mantilla E. Duéguez J.J. Escudero M. Monfort E. Pérez García-Pando C. Petetin H. Jorba O. Vázquez V. de la Rosa J. Campos A. Muñóz M. Monge S. Hervás M. Javato R. Cornide M.J. Lessons from the COVID-19 air pollution decrease in Spain: now what? Sci. Total Environ. 779 2021 146380 10.1016/j.scitotenv.2021.146380 Riccò M. Ranzieri S. Balzarini F. Bragazzi N.L. Corradi M. SARS-CoV-2 infection and air pollutants: correlation or causation? Sci. Total Environ. 734 2020 139489 10.1016/j.scitotenv.2020.139489 Rodríguez-Sánchez A. Vivanco M.G. Theobald M.R. Martín F. Estimating the effect of the COVID-19 pandemic on pollutant emissions in Europe Atmos. Pollut. Res. 13 2022 101388 10.1016/j.apr.2022.101388 Schell B. Ackermann I.J. Hass H. Binkowski F.S. Ebel A. Modeling the formation of secondary organic aerosol within a comprehensive air quality model system J. Geophys. Res. Atmos. 106 2001 28275 28293 10.1029/2001JD000384 Schneider R. Masselot P. Vicedo-Cabrera A.M. Sera F. Blangiardo M. Forlani C. Douros J. Jorba O. Adani M. Kouznetsov R. Couvidat F. Arteta J. Raux B. Guevara M. Colette A. Barré J. Peuch V.-H. Gasparrini A. Differential impact of government lockdown policies on reducing air pollution levels and related mortality in Europe Sci. Rep. 12 2022 726 10.1038/s41598-021-04277-6 35082316 Seinfeld J.H. Pandis S.N. Atmospheric Chemistry and Physics 1998 John Wiley and Sons Silibello C. Calori G. Brusasca G. Giudici A. Angelino E. Fossati G. Peroni E. Buganza E. Modelling of PM10 concentrations over Milano urban area using two aerosol modules Environmental Modelling & Software, New Approaches to Urban Air Quality Modelling 23 2008 333 343 10.1016/j.envsoft.2007.04.002 Silva A.C.T. Branco P.T.B.S. Sousa S.I.V. Impact of COVID-19 pandemic on air quality: a systematic review Int. J. Environ. Res. Publ. Health 19 2022 1950 10.3390/ijerph19041950 Simpson D. Fagerli H. Jonson J.E. Tsyro S. Wind P. Tuovinen J.-P. Transboundary Acidification, Eutrophication and Ground Level Ozone in Europe PART I Unified EMEP Model Description (EMEP Status Report 2003) 2003 Skamarock W.C. Klemp J.B. Dudhia J. Gill D.O. Liu Z. Berner J. Wang W. Powers J.J. Duda M.G. Barker D. Huang X.-Y. A DESCRIPTION OF THE ADVANCED RESEARCH WRF MODEL VERSION 4.1 (No 2019 NCAR/TN-556+STR Smagorinsky J. General circulation experiments with the primitive equations: I. The basic experiment Mon. Weather Rev. 91 1963 99 164 Sokhi R.S. Moussiopoulos N. Baklanov A. Bartzis J. Coll I. Finardi S. Friedrich R. Geels C. Grönholm T. Halenka T. Ketzel M. Maragkidou A. Matthias V. Moldanova J. Ntziachristos L. Schäfer K. Suppan P. Tsegas G. Carmichael G. Franco V. Hanna S. Jalkanen J.-P. Velders G.J.M. Kukkonen J. Advances in air quality research – current and emerging challenges Atmos. Chem. Phys. 22 2022 4615 4703 10.5194/acp-22-4615-2022 Sokhi R.S. Singh V. Querol X. Finardi S. Targino A.C. Andrade M. de F. Pavlovic R. Garland R.M. Massagué J. Kong S. Baklanov A. Ren L. Tarasova O. Carmichael G. Peuch V.-H. Anand V. Arbilla G. Badali K. Beig G. Belalcazar L.C. Bolignano A. Brimblecombe P. Camacho P. Casallas A. Charland J.-P. Choi J. Chourdakis E. Coll I. Collins M. Cyrys J. da Silva C.M. Di Giosa A.D. Di Leo A. Ferro C. Gavidia-Calderon M. Gayen A. Ginzburg A. Godefroy F. Gonzalez Y.A. Guevara-Luna M. Haque SkM. Havenga H. Herod D. Hõrrak U. Hussein T. Ibarra S. Jaimes M. Kaasik M. Khaiwal R. Kim J. Kousa A. Kukkonen J. Kulmala M. Kuula J. La Violette N. Lanzani G. Liu X. MacDougall S. Manseau P.M. Marchegiani G. McDonald B. Mishra S.V. Molina L.T. Mooibroek D. Mor S. Moussiopoulos N. Murena F. Niemi J.V. Noe S. Nogueira T. Norman M. Pérez-Camaño J.L. Petäjä T. Piketh S. Rathod A. Reid K. Retama A. Rivera O. Rojas N.Y. Rojas-Quincho J.P. San José R. Sánchez O. Seguel R.J. Sillanpää S. Su Y. Tapper N. Terrazas A. Timonen H. Toscano D. Tsegas G. Velders G.J.M. Vlachokostas C. von Schneidemesser E. Vpm R. Yadav R. Zalakeviciute R. Zavala M. A global observational analysis to understand changes in air quality during exceptionally low anthropogenic emission conditions Environ. Int. 157 2021 106818 10.1016/j.envint.2021.106818 Steinbrecht W. Kubistin D. Plass-Dülmer C. Davies J. Tarasick D.W. von der Gathen P. Deckelmann H. Jepsen N. Kivi R. Lyall N. Palm M. Notholt J. Kois B. Oelsner P. Allaart M. Piters A. Gill M. Van Malderen R. Delcloo A.W. Sussmann R. Mahieu E. Servais C. Romanens G. Stübi R. Ancellet G. Godin-Beekmann S. Yamanouchi S. Strong K. Johnson B. Cullis P. Petropavlovskikh I. Hannigan J.W. Hernandez J.-L. Diaz Rodriguez A. Nakano T. Chouza F. Leblanc T. Torres C. Garcia O. Röhling A.N. Schneider M. Blumenstock T. Tully M. Paton-Walsh C. Jones N. Querel R. Strahan S. Stauffer R.M. Thompson A.M. Inness A. Engelen R. Chang K.-L. Cooper O.R. COVID-19 crisis reduces free tropospheric ozone across the northern hemisphere Geophys. Res. Lett. 48 2021 10.1029/2020GL091987 Taurino E. Bernetti A. Caputo A. Cordella M. De Lauretis R. D'Elia I. Di Cristofaro E. Gagna A. Gonella B. Moricci F. Peschi E. Romano D. Vitullo M. Italian Emission Inventory 1990-2020. Informative Inventory Report 2022 (No. 361/2022), ISPRA, Rapporti 2022 ISPRA Tremback C.J. Numerical Simulation of a Mesoscale Convective Complex: Model Development and Numerical Results (PhD. Dissertation). Atmos. Sci. Paper No. 465 1990 Colorado State University, Dept. of Atmospheric Science Fort Collins, CO 80523 Vautard R. Bessagnet B. Chin M. Menut L. On the contribution of natural Aeolian sources to particulate matter concentrations in Europe: testing hypotheses with a modelling approach Atmos. Environ. 39 2005 3291 3303 10.1016/j.atmosenv.2005.01.051 Velders G.J.M. Willers S.M. Wesseling J. den Elshout S. van van der Swaluw E. Mooibroek D. van Ratingen S. Improvements in air quality in The Netherlands during the corona lockdown based on observations and model simulations Atmos. Environ. 247 2021 118158 10.1016/j.atmosenv.2020.118158 Villeneuve P.J. Goldberg M.S. Methodological considerations for epidemiological studies of air pollution and the SARS and COVID-19 coronavirus outbreaks Environ. Health Perspect. 128 2020 095001 10.1289/EHP7411 Vitali L. Righini G. Piersanti A. Cremona G. Pace G. Ciancarella L. M-TraCE: a new tool for high-resolution computation and statistical elaboration of backward trajectories on the Italian domain Meteorol. Atmos. Phys. 129 2017 629 643 10.1007/s00703-016-0491-8 von Schneidemesser E. Sibiya B. Caseiro A. Butler T. Lawrence M.G. Leitao J. Lupascu A. Salvador P. Learning from the COVID-19 lockdown in berlin: observations and modelling to support understanding policies to reduce NO2 Atmos. Environ. X 12 2021 100122 10.1016/j.aeaoa.2021.100122 Walko R.L. Band L.E. Baron J. Kittel T.G.F. Lammers R. Lee T.J. Ojima D. Pielke R.A. Taylor C. Tague C. Tremback C.J. Vidale P.L. Coupled atmosphere–biophysics–hydrology models for environmental modeling J. Appl. Meteorol. Climatol. 39 2000 931 944 10.1175/1520-0450(2000)039<0931:CABHMF>2.0.CO 2 Walko R.L. Cotton W.R. Meyers M.P. Harrington J.Y. New RAMS cloud microphysics parameterization part I: the single-moment scheme Atmos. Res. 38 1995 29 62 10.1016/0169-8095(94)00087-T Wesely M.L. Parameterization of surface resistances to gaseous dry deposition in regional scale numerical models Atmos. Environ. 23 1989 1293 1304 WHO WHO Global Air Quality Guidelines: Particulate Matter (PM2.5 and PM10), Ozone, Nitrogen Dioxide, Sulfur Dioxide and Carbon Monoxide 2021 World Health Organization Yamartino R.J. Nonnegative, conserved scalar transport using grid-cell-centered, spectrally constrained blackman cubics for applications on a variable-thickness mesh Mon. Weather Rev. 121 1993 753 763 10.1175/1520-0493(1993)121<0753:NCSTUG>2.0.CO 2 Zhang K.M. Knipping E.M. Wexler A.S. Bhave P.V. Tonnesen G.S. Size distribution of sea-salt emissions as a function of relative humidity Atmos. Environ. 39 2005 3373 3379 10.1016/j.atmosenv.2005.02.032	36474671	PMC9716127	NO-CC CODE	2022-12-09 23:15:16	no	Atmos Pollut Res. 2022 Dec 2; 13(12):101620	utf-8	Atmos Pollut Res	2,022	10.1016/j.apr.2022.101620	oa_other
==== Front Lang Resour Eval Lang Resour Eval Language Resources and Evaluation 1574-020X 1574-0218 Springer Netherlands Dordrecht 9626 10.1007/s10579-022-09626-z Book Review Book Review: the Routledge Handbook of Translation and Ethics http://orcid.org/0000-0002-9126-7530 Li Hongzheng [email protected] 12 Wang Ruojin [email protected] 12 1 grid.43555.32 0000 0000 8841 6246 School of Foreign Languages, Beijing Institute of Technology, Beijing, 102488 China 2 grid.424018.b 0000 0004 0605 0826 Key Laboratory of Language, Cognition and Computation, Ministry of Industry and Information Technology, Beijing, 102488 China 2 12 2022 14 27 10 2022 © The Author(s), under exclusive licence to Springer Nature B.V. 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Translation always faces various ethics issues. Ethics in translation has attracted widespread attention from researchers around the world. The Routledge Handbook of Translation and Ethics published in 2021 comprehensively discusses various ethics issues in translation and translation studies, including but not limited to the fundamental ethical theories, ethics of translators and translation industry in different contexts, as well as the new challenges and trends with the development of globalization and technology, and provides new perspectives and critical thoughts for the translation ethics research. Keywords Book review Translation Ethics http://dx.doi.org/10.13039/501100001809 National Natural Science Foundation of China 61902024 Li Hongzheng http://dx.doi.org/10.13039/501100012236 Beijing Institute of Technology Research Fund Program for Young Scholars ==== Body pmcAs a complex and multi-dimensional human cognition and information processing activity, translation has different forms and always inevitably involves various elements such as interpersonal relationship and cultural environments. Translation ethics has been discussed and attracted growing attention since it was proposed more than 30 years ago (Berman, 1992). With the intensification of globalization and the development of technology and informatization, ethics issues and challenges in translation have become increasingly prominent. Correspondingly, academic research on translation ethics is necessary. The Routledge Handbook of Translation and Ethics (edited by Kaisa Koskinen and Nike K. Pokorn, published in 2021), one of the series of Routledge Handbooks in Translation and Interpreting Studies, provides a comprehensive overview of ethics in translating and interpreting, in a metaethical way. The handbook is composed of 31 chapters. The first chapter is the introduction, framing the definition of ethics and ethics in translation studies and giving an outline of the book. The remaining 30 chapters are divided into four parts and each of them focuses on one topic. Part I introduces the most influential ethical theories in the field of translation. Part II focuses on the responsibilities of agents in different contexts and the ethical dilemmas they face. Part III takes a critical look at institutions, collectives, and individuals to (re)construct ethics in translator and interpreters’ education. The last part, Part IV discusses some special issues and new challenges in practical translating and interpreting scenarios, and signals new directions for further study. Each chapter in the four parts is organized in a similar structure, including introduction, history, core issues, new challenges, and conclusion, as well as recommended further reading at the end. Each part is introduced and discussed in more depth below. Part I (Chapter 2 to Chapter 11) shows a full map of the fundamental ethical theories and traditions, providing an insight into the development of ethical thinking. Chapter 2 discusses important and central virtue ethics in translation, including but not limited to fidelity, loyalty, integrity and honesty. Chapter 3, aiming at elucidating Chinese tradition, explores Chinese discourse on translation ethics, which is deeply affected by Confucian ethics. Chapter 4 reviews socialist translation theories in historical and political contexts and provides an overview of the emergence of socialist translation ethics. Chapter 5 discusses the translation theories and ethics from the perspective of functionalist approaches. Chapter 6 mainly focuses on the ethics and contributions of two French thinkers: Antoine Berman (1942–1991) and Henri Meschonnic (1932–2009). Chapter 7 centers on the linguistic hospitality and untranslatability ethics theories of two philosophers: Jacques Derrida (1930–2004) and Paul Ricœur (1913–2005). Chapter 8 discusses postcolonial translation ethics with the emphasis on decolonization. Chapter 9 depicts the relations between feminisms and feminist translation ethics, emphasizing the urgency of developing ethical framework for decolonization. Chapter 10 introduces a famous scholar in the field of translation studies, Lawrence Venuti and his theoretical contributions: ethics of location and ethics of difference. Chapter 11 analyzes the translator ethics developed by Anthony Pym, which focuses on the people involved in translation activities rather than on the texts. Part II (Chapter 12 to Chapter 19) looks at the translator ethics in various contexts that can guide the ethical demands on translators and interpreters. The professional translator ethics (Chapter 12) formulates the definition of professional translators and discusses the ethical questions they may encounter. Chapter 13 reviews key approaches of literary translator ethics such as Meschonnic’s poetic approach (Meschonnic, 2011), Steiner’s hermeneutic approach (Steiner, 1998), and Descriptive translation studies (DTS) (Toury, 1995). Conference interpreter ethical virtues like competence, integrity, confidentiality, neutrality, and fidelity are discussed in Chapter 14. Chapter 15 then discusses ethics in public service interpreting of both spoken and sign language. Chapter 16 brings up ethical issues related to volunteering in translation and interpreting, followed by ethics of the activist translation and interpreting that supports political agendas and struggles at both local and global level (Chapter 17). Chapter 18 turns to the relations between ethics and modern translation technology in the Artificial Intelligence (AI) era, such as machine translation (MT), computer-aided translation (CAT) and translation memory. Chapter 19, Translation and Posthumanism, focuses on the ethical implications of translation technology in various forms of posthuman, including globotics, patronage, labour and data extractivism. Part III (Chapter 20 to Chapter 24) covers the ethical topics in translation industry or institutions and organizations. Chapter 20 focuses on different ethics codes related to translation practices and expectations for translators and interpreters. Chapter 21 discusses ethical issues in translation industry such as disparities of power, ownership of resources and the challenge of crowdsourcing. Chapter 22 concentrates on the Ethics in translator and interpreter(T&I) education, emphasizing the translator’s responsibilities in the process of translation teaching. The next Chapter 23 addresses the Ethics of T&I education: how the T&I educational institutions and their educational practices should understand ethics. And Chapter 24 provides an overview of research ethics in translation and interpreting studies. As the final section of this handbook, Part IV (Chapter 25 to Chapter 31) presents the emerging challenges and trends in translator and interpreter ethics. This part starts with the ethics in child language brokering (Chapter 25) which refer to the process whereby children or young people translate or interpret from one language to another for adult family members to enable intercultural and inter-linguistic communication. The following chapters respectively illuminate ethics of translating and interpreting in conflict and crisis associated with mass migration, humanitarian emergency or gender violence (Chapter 26); ethical stress in translation and interpreting that may result in nefarious consequences for individuals (Chapter 27); linguistic first aid to overcome linguistic barriers in various situations (Chapter 28); ethics of translation of sacred texts with three key ethical considerations (Chapter 29); ethics of collaboration and control in literary translation (Chapter 30); and accessibility and linguistic rights in translating and interpreting (Chapter 31) that aim at guaranteeing equal commutation for all the people (healthy and disabled). As a researcher devoted to machine translation and modern translation technology for many years, I am especially impressed and inspired by Chapter 18 and Chapter 28. Though improving translation quality and saving time and efforts, new translation technologies and approaches have also created many ethical problems and challenges that we must face and try to solve. Some of these challenges, like data confidentiality and privacy, are not only closely related to translation, but also related to the whole society of AI (Zhang et al., 2022). The detailed discussion of the core and emerging issues in Chapter 18 provides a new insight for us to balance the relationship between technology and ethics. Linguistic first aid in Chapter 28 is also one of the most popular topics in recent years, especially under the global influence of COVID-19. For example, China has been providing multilingual translation service for foreigners in China to help them accurately understand the policies and requirements of epidemic prevention, and these practices have achieved good results (Li et al., 2020). In many emergency situations, providing timely and accurate language services is essential to overcome the language barriers and has already attracted many countries and regions’ attention around the world, especially those with multilingual communities (O’Brien, 2020). In conclusion, the well-structured handbook provides a comprehensive and full landscape on translation and ethics studies from various perspectives and inspires critical ideas. With its many unique features, this handbook is highly recommended for students majoring in (machine) translation, translation researchers and scholars, practitioners such as translators and interpreters, as well as other readers who are interested in translation, to build up a better and in-depth understanding of translation ethics. Acknowledgements This research is funded by National Natural Science Foundation of China(61902024) and Beijing Institute of Technology Research Fund Program for Young Scholars. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Berman A The experience of the foreign: Culture and translation in romantic Germany 1992 SUNY Press Li Y Rao G Zhang J Li J Conceptualizing national emergency language competence Multilingua 2020 39 5 617 623 10.1515/multi-2020-0111 Meschonnic H Ethics and politics of translating 2011 John Benjamins Publishing Company O’Brien S O’Hagan M Translation technology and disaster management The Routledge handbook of translation and technology 2020 Routledge Steiner G After Babel: Aspects of language and translation 1998 3 Oxford University Press Toury G Descriptive translation studies and beyond, revised edition 2012 1995 John Benjamins Publishing Company Zhang, D., Maslej, N., Brynjolfsson, E., Etchemendy, J., Lyons, T., Manyika, J. & Perrault, R. (2022). The AI Index 2022 Annual Report. Retrieved from http://arxiv.org/abs/2205.03468	0	PMC9716144	NO-CC CODE	2022-12-03 23:20:54	no	Lang Resour Eval. 2022 Dec 2;:1-4	utf-8	Lang Resour Eval	2,022	10.1007/s10579-022-09626-z	oa_other
==== Front J Anesth J Anesth Journal of Anesthesia 0913-8668 1438-8359 Springer Nature Singapore Singapore 3149 10.1007/s00540-022-03149-1 Editorial Pandemic and infodemic: the role of academic journals and preprints http://orcid.org/0000-0001-9202-9932 Asai Takashi [email protected] grid.416093.9 Department of Anesthesiology, Dokkyo Medical University Saitama Medical Center, 2-1-50 Minami-Koshigaya, Koshigaya, Saitama 343-8555 Japan 2 12 2022 14 26 10 2022 29 11 2022 © The Author(s) under exclusive licence to Japanese Society of Anesthesiologists 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Keywords COVID-19 Pandemic Infodemic Preprints Peer-review journals ==== Body pmc "Elle ne pouvait être que le témoignage de ce qu'il avait fallu accomplir et que, sans doute, devraient accomplir encore, contre la terreur et son arme inlassable,…Car il savait ce que cette foule en joie ignorait, et qu'on peut lire dans les livres, que le bacille de la peste ne meurt ni ne disparaît jamais, qu'il peut rester pendant des dizaines d'années endormi….la peste réveillerait ses rats et les enverrait mourir dans une cité heureuse." (Albert Camus "La Pestes", 1947) “It could be only the record of what had had to be done, and what assuredly would have to be done again in the never ending fight against terror and its relentless onslaughts,…He knew what those jubilant crowds did not know but could have learned from books: that the plague bacillus never dies or disappears for good; that it can lie dormant for years and years…it would rouse up its rats again and send them forth to die in a happy city.” ("The Plague", translation by Stuart Gilbert) In March 2020, the World Health Organization (WHO) has declared coronavirus infectious disease-2019 (COVID-19) outbreak a pandemic. It had soon become clear that this virus was highly contagious and the mortality rate was high, so that reliable information became urgently required to treat patients and to prevent the risk of cross-transmission of infection to other patients and to healthcare workers. We healthcare workers had taken all possible efforts to save patients, but it was not an easy task, because the flames of fear about the epidemic were being fanned by “infodemic”: rumors and misinformation disseminated by mass media, social media, and even by governments. Infodemic, a portmanteau of “information” and “epidemic”, refers to the rapid spread of information—both accurate and inaccurate—about something, such as a disease. This word was coined by David J. Rothkopf during the outbreak of severe acute respiratory syndrome (SARS) and first used in his following commentary made in The Washington Post in 2003 [1].“SARS is the story of not one epidemic but two, and the second epidemic, the one that has largely escaped the headlines, has implications that are far greater than the disease itself. That is because it is not the viral epidemic but rather an "information epidemic" that has transformed SARS,…into a global economic and social debacle…What is more, the information epidemic—or "infodemic"—has made the public health crisis harder to control and contain.” (Rothkopf DJ: The Washington Post. 11th May, 2003) [1] Tedros Adhanom Ghebreyesus, the Director-General of the WHO, stated shortly before declaring pandemic (on 15th February, 2020) that “…we’re not just fighting an epidemic; we’re fighting an infodemic” [2]. This was followed by the statement that “[f]ake news spreads faster and more easily than this virus, and is just as dangerous”. Nevertheless, since the WHO's declaration of the pandemic, some mass media advocated PCR testing to everyone, and promoted numerous “effective” preventative and treatment methods of COVID-19, and some social media platforms discouraged vaccination, but these were largely not evidence-based [3]. So, how healthcare workers have managed infodemic? Source of information included scientific articles which have gone through peer-review, those not peer-reviewed, textbooks, governmental statements, mass media reports, and social media information. To practice evidence-based clinical decision making, it was necessary to gain access to reliable sources of information, such as peer-reviewed randomized controlled studies, meta-analyses, and systematic reviews. One major practical problem with obtaining reliable information had been that considerable time was required until randomized controlled studies and systematic reviews were published in peer-review journals. Therefore, until these have become available, it was necessary to obtain information from statements made by academic societies, or reports made by healthcare workers on the social media. Nevertheless, caution was required in incorporating these information to aid clinical decision making. For example, one anesthesiologist designed an acrylic “Aerosol box” (or aerosol containment device) as a part of personal protective equipment (PPE) during airway management, and shared his idea on social media. This idea was picked up by an influential medical journal [4], and the use of such a device has rapidly spread worldwide, without being formally assessed for its effectiveness, efficacy and safety. However, subsequent studies have indicated that any “aerosol containment device” would make airway management more difficult, and the use of the device may not decrease, but even increase, the risk of healthcare workers being exposed to a high concentration of viral aerosols [5]. This example clearly indicates that healthcare workers have needed to keep updating their knowledge with new studies published in peer-review journals, to make sure that daily clinical practice is based on the currently available evidence. So, what efforts have clinical researchers made to accelerate publication of medical articles during the pandemic? One effective method heavily used during this pandemic was publicizing their studies as preprints. Preprints are scientific manuscripts which the authors posted online, before submitting them to peer-review journals. Paul Ginsparg, a physicist, was the first to launch a preprint server, aiχiv.org (pronounce it “archive”), in 1991 [6]. This was well accepted by physicists who previously distributed paper copies of their drafts by post to peer researchers. In contrast, before the outbreak of COVID-19 pandemic, clinical researchers were generally reluctant to adopt widespread sharing of preprints, probably because of concern that the potential harm that could result to patients, if medical treatment is based on findings that have not been vetted by peer reviewers. For example, the BMJ group opened a preprint server (ClinMedNetPrints.org) in 1999, but was closed in 2008, because only around 80 submissions were posted during this period [7]. The BMJ group, together with Cold Spring Harbor Laboratory and Yale University launched a new server, bioRχiv in 2013, and medRχiv in 2019 [7], but they were not actively used. Outbreak of COVID-19 pandemic triggered clinical researchers to use actively preprint servers, and during the initial few years of the COVID-19 pandemic, more than 35,000 preprints, mainly related to COVID-19, have been posted to medRχiv. This marked increase in the posting of preprints indicates that clinical researchers have found benefits of preprints in the era of COVID-19 pandemic: research outcomes can be disseminated quickly, potentially speeding up research that may lead to the development of vaccines and treatments; quality of the draft can be improved by receiving feedback from a wider group of readers; the authors can claim priority of their discovery; and unlike articles published in subscription-based journals, all the preprints are freely available to anyone. Because of these reasons, preprints have become potentially useful source of information. Nevertheless, healthcare workers should not, in principle, have use preprints for clinical decision making and policy formation. The medχRiv, for example, states in bold letters that “[t]hey should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information” (medχRiv Submission guide). In fact, a considerable number of preprints has already been retracted [8]. One possible problem for clinical researchers who uploaded their drafts to preprint servers is that some journals may refuse to publish the drafts, if they have already been published as preprints. For example, in anesthesia field, the Journal of Anesthesia [9] and the British Journal of Anaesthesia [8] do not regard preprints as prior publication, but the Anesthesiology does [10]. What then traditional peer-review journals have done to address the issue of the “two epidemics”? Journals, including the Journal of Anesthesia, have taken much efforts to accelerate the review process, to make a quick decision to accept or not accept for publication, and to publish accepted articles instantly as Epub ahead of prints. In addition, subscription-based journals, in principle, have made all the articles related to COVID-19 fee to read, download and share. The Journal of Anesthesia has actively published articles related to COVID-19, and published an issue containing “Special Feature on Anesthesia in the Time of COVID-19”. Some of these articles have been frequently cited (Table 1) [3, 5, 11–18], indicating that clinicians have regarded these articles as reliable source of information, for performing evidence-based treatment of patients with COVID-19.Table 1 Ten most frequently cited articles related to the COVID-19 First author Title Reference number Hasanin A. Evaluation of fluid responsiveness during COVID-19 pandemic: what are the remaining choices? [11] Zhang L. Summary of 20 tracheal intubation by anesthesiologists for patients with severe COVID-19 pneumonia: retrospective case series [12] Hotta K. Regional anesthesia in the time of COVID-19: a minireview [13] Saito T. Aerosol containment device for airway management of patients with COVID-19: a narrative review [5] Hirota K. Air contamination with SARS-CoV-2 in the operating room [14] Gai N. Unique challenges in pediatric anesthesia created by COVID-19 [15] Yamakage M. Anesthesia in the times of COVID-19 [16] Wong P. Aligning difficult airway guidelines with the anesthetic COVID-19 guidelines to develop a COVID-19 difficult airway strategy: a narrative review [17] Asai T. COVID-19: accurate interpretation of diagnostic tests—a statistical point of view [3] Burnett GW. Managing COVID-19 from the epicenter: adaptations and suggestions based on experience [18] During the last few years, numerous studies related to COVID-19 have been published in peer-review journals, so it has increasingly been difficult for clinicians to find suitable articles to aid clinical decision making. Systematic reviews and clinical practice guidelines were urgently required, but those were not instantly available. To address this issue, some newer types of guidelines have been developed [19]: focused clinical practice guidelines, living clinical practice guidelines, or consensus statements. Journals have also actively published editorials as mini-review articles, and have actively used social media to pick up published articles which should be useful for clinicians. So, have we managed to control pandemic and infodemic, and have the journals played crucial role in actively providing reliable information? A definite answer is yet to be provided, but it would be yes, as we are coming back to normal life. Should the journals now end their role in publishing articles related to COVID-19? The answer is definitely no. This is because another epidemic or pandemic will certainly occur, 100 years, 10 years, or even a few years later. During the initial period of the outbreak of COVID-19, healthcare workers could use the knowledge that obtained during the SARS occurring in 2003 [20, 21]. Therefore, as Albert Camus wrote in his "La Pestes" ("The Plague"), the journals should keep publishing useful articles, to record what had had to be done, and what assuredly would have to be done in managing pandemic and infodemic. By doing so, healthcare workers would be able to utilize lessons from what we have achieved during the COVID-19 pandemic, to control effectively future pandemic and infodemic. Data availability There are no data obtained for this report. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Rothkopf DJ. When the Buzz Bites Back. The Washington Post. May 11, 2003. https://www.washingtonpost.com/archive/opinions/2003/05/11/when-the-buzz-bites-back/bc8cd84f-cab6-4648-bf58-0277261af6cd/. Accessed 2 Dec 2022 2. WHO Director-general speeches: Munich Security Conference on 15th February 2020. https://www.who.int/director-general/speeches/detail/munich-security-conference. Accessed 2 Dec 2022 3. Asai T Covid-19: accurate interpretation of diagnostic tests-a statistical point of view J Anesth 2021 35 328 332 10.1007/s00540-020-02875-8 33306137 4. Canelli R Connor CW Gonzalez M Nozari A Ortega R Barrier enclosure during endotracheal intubation N Engl J Med 2020 382 1957 1958 10.1056/NEJMc2007589 32243118 5. Saito T Asai T Aerosol containment device for airway management of patients with COVID-19: a narrative review J Anesth 2021 35 384 389 10.1007/s00540-020-02879-4 33226519 6. Ginsparg P ArXiv at 20 Nature 2011 476 145 147 10.1038/476145a 21833066 7. Rawlinson C Theodora BT New preprint server for medical research. Announcing the launch of medRxiv for faster access to better evidence BMJ 2019 365 l2301 10.1136/bmj.l2301 31167753 8. Pearse RM Ackland GL Asai T Hemmings HC Jr Preprints in perioperative medicine: immediacy for the greater good Br J Anaesth 2021 126 915 918 10.1016/j.bja.2021.02.024 33795134 9. Yamakage M Update to the handling of "Preprints" by the Journal of Anesthesia J Anesth 2021 35 1 2 10.1007/s00540-020-02885-6 33320281 10. Kharasch ED Avram MJ Clark JD Peer review matters: research quality and the public trust Anesthesiology 2021 134 1 6 10.1097/ALN.0000000000003608 33395468 11. Hasanin A Mostafa M Evaluation of fluid responsiveness during COVID-19 pandemic: what are the remaining choices? J Anesth 2020 34 758 764 10.1007/s00540-020-02801-y 32451626 12. Zhang L Li J Zhou M Chen Z Summary of 20 tracheal intubation by anesthesiologists for patients with severe COVID-19 pneumonia: retrospective case series J Anesth 2020 34 7 599 606 10.1007/s00540-020-02778-8 32303885 13. Hotta K Regional anesthesia in the time of COVID-19: a minireview J Anesth 2021 35 341 344 10.1007/s00540-020-02834-3 32712704 14. Hirota K Air contamination with SARS-CoV-2 in the operating room J Anesth 2021 35 333 336 10.1007/s00540-020-02814-7 32562137 15. Gai N Maynes JT Aoyama K Unique challenges in pediatric anesthesia created by COVID-19 J Anesth 2021 35 345 350 10.1007/s00540-020-02837-0 32770277 16. Yamakage M Anesthesia in the times of COVID-19 J Anesth 2021 35 325 327 10.1007/s00540-020-02798-4 32451627 17. Wong P Lim WY Aligning difficult airway guidelines with the anesthetic COVID-19 guidelines to develop a COVID-19 difficult airway strategy: a narrative review J Anesth 2020 34 924 943 10.1007/s00540-020-02819-2 32642840 18. Burnett GW Katz D Park CH Hyman JB Dickstein E Levin MA Sim A Salter B Owen RM Leibowitz AB Hamburger J Managing COVID-19 from the epicenter: adaptations and suggestions based on experience J Anesth 2021 35 366 373 10.1007/s00540-020-02860-1 33006071 19. Rong LQ Audisio K O'Shaughnessy SM Guidelines and evidence-based recommendations in anaesthesia: where do we stand? Br J Anaesth 2022 128 903 908 10.1016/j.bja.2022.02.025 35314064 20. Peng PWH Ho PL Hota SS Outbreak of a new coronavirus: what anaesthetists should know Br J Anaesth 2020 124 497 501 10.1016/j.bja.2020.02.008 32115186 21. Chen Q Lim B Ong S Wong WY Kong YC Rapid ramp-up of powered air-purifying respirator (PAPR) training for infection prevention and control during the COVID-19 pandemic Br J Anaesth 2020 1254 e171 e176 10.1016/j.bja.2020.04.006	36459231	PMC9716145	NO-CC CODE	2022-12-03 23:20:54	no	J Anesth. 2022 Dec 2;:1-4	utf-8	J Anesth	2,022	10.1007/s00540-022-03149-1	oa_other
==== Front Curr Pain Headache Rep Curr Pain Headache Rep Current Pain and Headache Reports 1531-3433 1534-3081 Springer US New York 1091 10.1007/s11916-022-01091-1 Chronic Pain Medicine (O Viswanath, Section Editor) Pill Counting as an Intervention to Enhance Compliance and Reduce Adverse Outcomes with Analgesics Prescribed for Chronic Pain Conditions: A Systematic Review Gill Benjamin 1 Obayashi Kotomi 1 Soto Victoria B. 2 Schatman Michael E. 34 Abd-Elsayed Alaa [email protected] 5 1 grid.134936.a 0000 0001 2162 3504 Physical Medicine and Rehabilitation, University of Missouri, Columbia, MO USA 2 Law Office of Victoria Soto, Austin, TX USA 3 grid.137628.9 0000 0004 1936 8753 Department of Anesthesiology, Perioperative Care, NYU Grossman School of Medicine, & Pain Medicine, New York, NY USA 4 grid.137628.9 0000 0004 1936 8753 Division of Medical Ethics, Department of Population Health, NYU Grossman School of Medicine, New York, NY USA 5 grid.28803.31 0000 0001 0701 8607 Department of Anesthesiology, School of Medicine and Public Health, University of Wisconsin, Madison, WI USA 2 12 2022 15 26 9 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Purpose of Review Appropriate use of opioid analgesics is a key concern within the field of pain medicine. Several methods exist to discourage abuse and facilitate effective treatment regimens. Pill counting is often cited as one such method and frequently employed in varying fashions within clinical practice. However, to date, there is no published review of the evidence to support this practice. This was a comprehensive review of the available literature that was conducted with analysis of the efficacy and practical application of pill counting during treatment of chronic pain conditions. Recent Findings There is paucity in data regarding pill count importance in pain management. Pill count is a very important tool to monitor compliance of opioids use which in turn can prevent several complications associated with opioid misuse. Summary Pill counting may be used in conjunction with other abuse deterrents, although increased support for this practice requires standardized methods of pill counting and further analysis of its effectiveness. Keywords Pill count Opioids Analgesics Abuse deterrent ==== Body pmcIntroduction The risks of opioid overdose and abuse continue to be preeminent concerns in the field of pain medicine. Although those committed to eradicating opioid analgesia from the pain management armamentarium intentionally conflate prescription opioid overdose deaths with those due to illicit fentalogues and recent data indicating that involuntary tapers of opioids actually increase the likelihood of death from overdose and suicide, the need for continued diligent focus on prescription practices by clinicians on behalf of their patient population remains a best practice [1, 2]. Multiple strategies have been proposed to mitigate the risks of inappropriate medication use both wittingly and unwittingly. Methods today may include urine drug testing (UDT), pain diaries, patient contracts or agreements, governmental prescription monitoring programs, electronic container monitors, and ingestible markers [3, 4]. Pill counting is often mentioned in studies on medication effectiveness or adherence with the goal of verifying the actual medication intake pattern. This method of medication monitoring has been implemented with pharmacotherapies for multiple pathologies, including lipid-lowering, HIV treatment, contraception, depression, and sickle cell disease [5–9]. Pill counting may involve a variety of approaches, including the requirement of the patient to present their current medication supply for verification at the clinic, at their home, at a local pharmacy, or via telephone or video calls over scheduled or variable intervals. It is also often mentioned in guidelines aimed at the prevention of opioid abuse in chronic cancer and nonmalignant pain [10, 11]. This review seeks to clarify the literature regarding pill counting as an effective and practical method to enhance compliance with analgesics prescribed for chronic pain conditions. Methods This study reviewed the literature to appraise the rationale and efficacy of pill counting in the management of outpatient oral analgesic regimens. To maintain transparent and detailed standards, Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed throughout the search and review. Searches were performed of the electronic databases PubMed, Scopus, MEDLINE, and the Cochrane Database for manuscripts indexed from the inception of each database until September 2021. The search terms included “pill count*,” filtered by “opioid,” “narcotic,” “chronic,” and “analgesic” in each respective database. Identified studies were uploaded into Mendeley, and duplicates were removed. Abstracts were screened to identify meta-analyses, randomized controlled trials (RCTs), case–control and cohort studies, quasi-experimental studies, observational studies, case series, and case reports. Selection criteria were applied by two independent reviewers, and discrepancies were resolved by a third author. Inclusion criteria focused on English-language human studies that investigated the efficacy of pill counting as a risk mitigation tool in chronic pain conditions. This included outcomes such as misuse of opioids or clinical pain score changes, objective functional results, and/or depression and anxiety scales. To investigate the efficacy of pill counting as an adherence monitoring instrument, studies were excluded if there was non-medical opioid use (i.e., use of opioid medication without a prescription or use with a prescription but not as prescribed). Studies were also excluded if pill counting was not used as an outcome measure or was a non-principal component of the study (Fig. 1).Fig. 1 Flow chart of article search and inclusion process Results A total of 82,718 manuscripts were identified in the initial search, and 2277 duplicates were removed. After filtering by keyword, 336 articles remained. Titles and abstracts were reviewed for pill counting as a component of chronic pain treatment. Of the 24 identified full-text manuscript with potential eligibility, 20 lacked a focus on pill counting efficacy, and one study used medication that was not prescribed as an analgesic. Three articles met inclusion and exclusion criteria and were reviewed (Table 1). All studies were prospective in nature using multicomponent interventions including controlled substance agreements, random drug screening through UDT, and pill counts. No study exclusively isolated pill counts as an intervention but rather used it as a component of the study interventions. Outcome measures included opioid pill confiscations by law enforcement and proportion of patients with aberrant drug-related behaviors [12–14].Table 1 Analysis of articles meeting inclusion criteria Study Focus Design/methodology Sample size Participant characteristics Condition Analgesic Study duration Bujold et al. [12] To describe the effects of practice guidelines on reducing the diversion of prescription drugs Prospective cohort study Participants: n = 27 (primary care clinicians) Primary care clinicians treating patients in a rural setting Chronic nonmalignant pain Non-specific opioids 24 months Brown et al. [13] To evaluate divergent drug-related behaviors among patients with chronic pain and investigator compliance with universal precautions Open-label, nonrandomized, non-comparative, multicenter, prospective Participants: n = 1487 (patients), n = 281 (investigators) Mean age of patients: 53, 57% female, 92% with chronic pain > 1 year Chronic, moderate-to-severe pain ≥ 3 months Morphine sulfate-extended release Not clear Manchikanti et al. [14] To evaluate controlled substance abuse after implementation of a monitoring program Prospective with historical controls Participants: n = 500 Mean age of patients: 48.5, 59% female; duration of pain mean: 10.7 years Chronic pain Hydrocodone, oxycodone, methadone, morphine Not clear Bujold et al. studied the implementation of practice guidelines intended for primary care clinicians in a rural community and their effects on opioid pill confiscations by law enforcement [12]. The guidelines encouraged patient-signed pain contracts, use of random UDT, and random pill counts. A survey was sent to 35 primary care clinicians with a subsequent 77% response rate: 90% indicated that they used the guidelines developed by the task force. Over a 2-year period, confiscation of opioid pills decreased by 300%. Brown et al. analyzed the feasibility and compliance of a universal precautions (UP) approach to determine the risk of aberrant drug-related behavior and to guide management of patients with chronic, moderate-to-severe pain being treated with morphine sulfate [13]. Components of the UP approach included treatment agreements, tracking of prescriptions, use of the Screener and Opioid Assessment for Patients with Pain®-Revised questionnaire, pill counts, pain patient follow-up tools, investigator assessment/plan, and urine drug screens. Using the UP approach, 1487 patients in a primary care setting were assigned a risk level for opioid abuse/misuse at each of the five visits with their care provider. Following the completion of the study questionnaire, 80% of the investigators who responded opined that pill counts were a useful or very useful utility and 58% indicated that they would continue to utilize pill counts in their practices. Aberrant drug-related behavior was detected throughout the study, which supported the feasibility of a UP approach for assessing aberrant drug-related behavior. Manchikanti et al. aimed to identify controlled substance abuse following the implementation of a controlled substance agreement [14]. Five hundred patients who were receiving stable doses of hydrocodone, oxycodone, methadone, or morphine were followed prospectively. All participants signed consent and controlled substance agreements. This agreement allowed the investigators to review charts and collect information regarding controlled substance intake, which included random drug screening, pill counts, and education. Overall, 9% of participants were determined to be abusing prescription drugs. When compared to historical controls, this demonstrated a 50% reduction in opioid abuse. Discussion The optimal use of opioid analgesics remains a crucial focus of pain management principles. The emphasis on a clinician’s duty to reduce abuse raises the question of optimal prescription monitoring. Despite pill counting being frequently mentioned throughout the literature as a method for promoting adherence to pain medication regimens, it is poorly defined, sparsely studied, and inconsistently implemented. This systematic review serves to determine the utility of pill counting as an intervention to improve compliance and safety in patients who are prescribed opioid medication for chronic pain conditions. In addition, this review demonstrates that the prescribing clinician’s compliance tools and methods are respectfully utilized to the fullest and assist him/her in efforts to utilize this risk mitigation tool. Based on the review’s data, the clinician will ideally become a safer and thus more effective prescriber through pill counting. Following extensive analysis, there is a dearth of high-quality studies that focus on pill counting as an intervention. There were no studies that exclusively isolated pill counting as an intervention despite the existence of many studies in which the authors posit on the perceived utility of pill counting. These studies, however, did so without actual supportive data. Rather than being incorporated as an intervention, pill counting was included as a component in a monitoring program or practice guideline, although without analysis of efficacy. Several studies mention the utility of pill counting as a component of opioid management; however, few include specific citations for these statements. Of the 24 full-text manuscripts assessed with potential eligibility and not included in the final analysis, only nine specifically reference the utility of pill counting [15–23]. Colasanti et al. cited an article by Chou and colleagues, in which the authors state, “Because patient self-report may be unreliable for determining amount of opioid use, functionality, or aberrant drug-related behaviors, pill counts…can be useful supplements” (p. 123) [15, 24]. Likewise, Viscomi et al. cited guidelines by Trescot and colleagues, in which the authors state, “Adherence monitoring is crucial to avoid abuse of the drugs and at the same time to encourage appropriate use, and involves the initiation of drug screening, pill counts, and patient care agreements” (p. 41) [16, 25]. In general, these statements regarding the utility of pill counting in opioid monitoring appear to be consensus statements and author opinions rather than empirically derived. These articles do not refer to specific analyses or trials of pill counting utility in opioid management. The simple count of current pill supply may be a more sensitive and objective measure than other methods to detect aberrant behavior for purposes of therapeutic adjustment or signaling warnings. Logical reasoning suggests that an inappropriately low supply could be related to abusive medication patterns or inadequate coverage of pain at a given dosage. However, it also stands to reason that patients may forget to bring medications for verification, discard unused doses (“pill dumping”), or share medication with other individuals [3]. Excessively high supply may signal memory problems in adhering to a regimen or the need to decrease prescription quantities. Logical reasoning affirms this method’s limitation outside of discrete formulations such as capsules or tablets. Furthermore, the pro re nata (PRN) use of medications will result in a more variable representation [26]. Since pill counting is inherently calculated on the quantity of dispensed medications, it is not able to account for ingestion of medication from non-prescription or other sources, such as previous surplus, those obtained from another provider, or those obtained through diversion. Ultimately, feasible implementation of pill counting, in and by itself, seems to have limited utility aside from representing a potentially problematic snapshot of current medication stock. If the concern is patient disuse of medication, excess supply may be artificially decreased by simply not providing all available medications or disposing of unused pills. There is no guaranteed relationship between the number of pills remaining and those actually ingested by the patient. If the concern is patient excessive use of medication, pill counting is again fraught with issues of supply inputs or medication obtained from additional sources and overall remains an unproven system for promoting safe medication practices. The resources expended by the physician and clinical staff to enforce pill counting provides additional concern for the implementation of this measure. If technology is leveraged to assist with the process, such as a mobile telephone application which would allow for instantaneous verification of current pill supply via photo on the device, then it may provide greater utility for tracking medication use and providing clinicians with additional data necessary to adjust prescriptions. Furthermore, placing the verification in the patient’s hands may attenuate feelings of distrust between the provider and patient. All of this may occur with pill counting implemented in the clinic setting. Regarding limitations of this review, the most notable one is the limited number of studies that met the inclusion criteria limiting the strength of the conclusions that we have drawn. Hopefully, this review will inspire more high-quality prospective studies on the efficacy of pill counts as a risk mitigation tool and consequently increase the strength of the support for pill counting in risk mitigation platforms. On a final note regarding ethical issues, social media is becoming rife with claims and copies of actual opioid agreements, indicating that certain prescribers are requiring patients to appear at their physicians’ offices within 2 h of notification in order to undergo a random pill count. As a result, patients in these practices are “handcuffed” by this requirement and are not allowed to travel whatsoever due to the risk of not being able to show up for the pill count within the given time frame. Particularly, given that many patients were unable to visit family members who lived out of state for more than 2 years during the throes of COVID-19, such draconian policies are highly unethical, representing a clear assault on the bioethical principle of respect for patient autonomy. Any physicians who are engaged in such problematic practices need to reconsider their practices of “chaining” their patients to the practices physically. Conclusion The clinical utility of pill counting remains limited due to feasibility concerns as presented above. Although frequently suggested as a component of a monitoring program or practice guideline, there is limited evidence to support the use of pill counting as a stand-alone approach to risk mitigation. Pill counting in the form in which it has been utilized thus far may be helpful when used in conjunction with other methods such as random urine drug screens. There is a great need for more feasible and precise methods for pill counting to improve the validity and reliability of its utilization. Engaging in more sophisticated pill counting approaches supports the notion that practicing clinicians are doing all that they can to treat their pain patients in the safest, most effective ways available to them, which equates to improved overall quality in the practice of pain medicine. Compliance with Ethical Standards Conflict of Interest The authors declare no competing interests. This article is part of Topical Collection on Chronic Pain Medicine Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Oliva EM et al. Associations between stopping prescriptions for opioids, length of opioid treatment, and overdose or suicide deaths in US veterans: Observational Evaluation. BMJ. 2020;368. 2. Persico AL Wegrzyn EL Fudin J Schatman ME Fentalogues J Pain Res 2020 13 2131 10.2147/JPR.S265901 32904462 3. Lehmann A Assessing medication adherence: options to consider Int J Clin Pharm 2014 36 1 55 69 10.1007/s11096-013-9865-x 24166659 4. Jones MR Government legislation in response to the opioid epidemic Curr Pain Headache Rep 2019 23 6 1 7 10.1007/s11916-019-0781-1 30637489 5. van Driel ML, Morledge MD, Ulep R, Shaffer JP, Davies P, Deichmann R. Interventions to improve adherence to lipid‐lowering medication. Cochrane Database Syst Rev. 2016;12. 6. Rueda S, et al. Patient support and education for promoting adherence to highly active antiretroviral therapy for HIV/AIDS. Cochrane Database Syst Rev. 2006;3. 7. Mack N, et al. Strategies to improve adherence and continuation of shorter‐term hormonal methods of contraception. Cochrane Database Syst Rev. 2019;4. 8. Hetrick SE, et al. New generation antidepressants for depression in children and adolescents: A network meta‐analysis. Cochrane Database Syst Rev 2021;5. 9. Rankine-Mullings AE, Owusu-Ofori S. Prophylactic antibiotics for preventing pneumococcal infection in children with sickle cell disease. Cochrane Database Syst Rev. 2021;3. 10. Amaram-Davila JS Arthur J Reddy A Bruera E Managing nonmedical opioid use among patients with cancer pain during the COVID-19 pandemic using the CHAT model and telehealth J Pain Symptom Manage 2021 62 1 192 196 10.1016/j.jpainsymman.2021.01.005 33515659 11. Atluri S, Akbik H, Sudarshan G. Prevention of opioid abuse in chronic non-cancer pain: an algorithmic, evidence based approach. Pain Physician. 2012;15(3S). 12. Bujold E Huff J Staton EW Pace WD Improving use of narcotics for nonmalignant chronic pain: a lesson from Community Care of North Carolina J Opioid Manag 2012 8 6 363 367 10.5055/jom.2012.0136 23264314 13. Brown J Assessment, stratification, and monitoring of the risk for prescription opioid misuse and abuse in the primary care setting J Opioid Manag 2011 7 6 467 483 10.5055/jom.2011.0088 22320029 14. Manchikanti L Manchukonda R Damron KS Brandon D McManus CD Cash K Does adherence monitoring reduce controlled substance abuse in chronic pain patients? Pain Physician 2006 9 1 57 10.36076/ppj.2006/9/57 16700282 15. Colasanti J Chronic opioid therapy in people living with human immunodeficiency virus: patients’ perspectives on risks, monitoring, and guidelines Clin Infect Dis 2019 68 2 291 297 10.1093/cid/ciy452 29860411 16. Viscomi CM Covington M Christenson C Pill counts and pill rental: unintended entrepreneurial opportunities Clin J Pain 2013 29 7 623 624 10.1097/AJP.0b013e31826f9a24 23328335 17. Ulker E Del Fabbro E Best practices in the management of nonmedical opioid use in patients with cancer-related pain Oncologist 2020 25 3 189 196 10.1634/theoncologist.2019-0540 31872911 18. Scarborough BM Smith CB Optimal pain management for patients with cancer in the modern era CA Cancer J Clin 2018 68 3 182 196 10.3322/caac.21453 29603142 19. Rager JB Schwartz PH Defending opioid treatment agreements: disclosure, not promises Hastings Cent Rep 2017 47 3 24 33 10.1002/hast.702 28543422 20. Vadivelu N Lumermann L Zhu R Kodumudi G Elhassan AO Kaye AD Pain control in the presence of drug addiction Curr Pain Headache Rep 2016 20 5 1 8 10.1007/s11916-016-0561-0 26695061 21. Alford DP Chronic back pain with possible prescription opioid misuse JAMA 2013 309 9 919 925 10.1001/jama.2013.522 23462788 22. Jan S. Patient perspective, complexities, and challenges in managed care. J Manag Care Pharm Supp B. 2016;16(1);22–25 23. Lafleur J Oderda GM Methods to measure patient compliance with medication regimens J Pain Palliat Care Pharmacother 2004 18 3 81 87 10.1080/J354v18n03_09 15364635 24. Chou R Clinical guidelines for the use of chronic opioid therapy in chronic noncancer pain J Pain 2009 10 2 113 130 10.1016/j.jpain.2008.10.008 19187889 25. Trescot AM Opioids in the management of chronic non-cancer pain: an update of American Society of the Interventional Pain Physicians?(ASIPP) Guidelines Pain Physician 2008 11 2S S5 18443640 26. Vik SA Maxwell CJ Hogan DB Measurement, correlates, and health outcomes of medication adherence among seniors Ann Pharmacother 2004 38 2 303 312 10.1345/aph.1D252 14742770	36459370	PMC9716148	NO-CC CODE	2022-12-03 23:20:54	no	Curr Pain Headache Rep. 2022 Dec 2;:1-5	utf-8	Curr Pain Headache Rep	2,022	10.1007/s11916-022-01091-1	oa_other
==== Front Am J Psychoanal Am J Psychoanal American Journal of Psychoanalysis 0002-9548 1573-6741 Palgrave Macmillan UK London 9374 10.1057/s11231-022-09374-7 Article The Pandemic, the Protests, the Chaos: A Destabilizing Effect on the Analyst Analyst Involvement in Protests During the Pandemic and Its Effect on a Treatment Frankfeldt Valerie R. [email protected] 30 East 9th St., #2K, New York, NY 10003 USA 2 12 2022 122 © Association for the Advancement of Psychoanalysis 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. In May 2020, within the cultural and emotionally regressive chaos of the pandemic, the analyst witnessed a violent Black Lives Matter protest. Myriad unprocessed feelings subsequently impacted her handling the treatment of a patient who abruptly left a session to attend a protest herself. The analyst describes her own personal experience and the cascade of events that affected the treatment. She suggests that analysts can be armed with the awareness that enactments are more likely to happen when the analyst, as well the patient, are under extreme duress as is the case in the time of Covid. She describes some of the forces that were specific to this case and her own personal embroilment. She then broadens the discussion to other analysts’ reports of overwhelming pandemic experiences and the corresponding effect on the work. She also elucidates the importance of the frame for therapeutic work. Keywords Pandemic Covid frame countertransference mentalization destabilization dysregulation enactment acting out ==== Body pmcIntroduction Life can be hard at the best of times. But especially since 2020, analysts and patients alike have lived through excessive turmoil evoking fear, grief and uncertainty that is unparalleled in our lifetime. The pandemic wreaked havoc with our thinking, feelings, spirit, behavior, and sense of security. As analysts, we may have initially underestimated the impact of these traumatic events on ourselves because we are used to sitting quietly in the face of overwhelming feelings. We know how to experience, observe, and make sense. We expect and assume that our observing egos will be able to take over and help us achieve objectivity. Zerbe noted that during the pandemic, analysts’ needs for self-care may have been neglected representing a “mind-body split [that] is culturally sanctioned and professionally overlooked within our field” (in press). At this time, we were all forced to react and respond on many different levels. It was personal. It was lived, not represented. Our trusted analytic protocols were not sufficient during this period. We were awash with unprecedented experiences and feelings in response to events that affected everyone—the global culture and environment. What was the effect on the psychoanalytic stance when the analytic dyad was operating within the equivalent of a building on fire? Are there ways it exerted a regressive force? This paper focuses on a piece of acting out between analyst and patient in early 2020 that was related to the Black Lives Matter protests in the context of the pandemic and impacted an analytic treatment. The issues involved in this specific case example can be generalized to provide learning for all therapists going forward in terms of ourselves and our work. The unmentalized components of the catastrophe represented by the pandemic has had wide ranging implications for all of us. The first force—on the macro level—was our ubiquitous sense of being thrown off balance due to the pandemic itself. The treatment frame was wildly stretched insofar as we had to immediately distance ourselves from our patients on teletherapy and all the disorganization and discombobulation that represented. Secondly, and more personally, on a micro level, this analyst was inadvertently thrust into a Black Lives Matter (BLM) protest that transmogrified into violence outside her window the next day. The day after that the hyperstimulation from the protests plus the continuous invisible stress running in the background of the pandemic spilled over into her mishandling of a moment in a treatment in which the patient precipitously left a session to join a protest herself. The case had its own complications apart from these specific events, and all of the above conspired to influence the analyst to break the frame, which produced ripple effects in the rest of the therapy. What follows is an attempt to lay out the perplexing disturbance of a case caused by this traumatic cultural phenomenon. The study is intended to be primarily about the analyst’s reactions rather than a focus on the specifics of the patient. Part I recounts the analyst's personal experience in the protests; Part II presents an overview and reflections on how reactions to the pandemic undermined the author’s ego functions in relation to a particular patient and her process. Part III discusses the impingements of living and working during Covid on psychoanalysts as a group. Our primary holding mechanism, the frame, is discussed in light of the extraordinary changes we’ve had to make to it and how that confounds treatment. Part I: My Personal Experience Sunday in the Park with a Protest From my journal Sunday, 5/31/2020 Today I went to Washington Square Park to read. It was a beautiful warm day in the midst of the Coronavirus pandemic. It was hard to find a quiet spot six feet from other people, but the atmosphere was peaceful and serene. Suddenly, Black Lives Matter protesters, catalyzed by the murder of George Floyd three days earlier, poured into the park. They chanted and punched their fists in the air: “SILENCE IS VIOLENCE!” Most were white. The crowd’s energy kindles an abrupt and unexpected reaction within me. I want to jump in and join them; I am suffused with nameless rage. I have gone from zero to sixty for no apparent reason. I would love to let go and scream and chant, too. I’m not the only one. Everyone claps and raises both arms in solidarity with the “don't shoot” meme. Then it hits me—because of Covid and my being in a high-risk group due to age, I am in peril. Many in this very tight throng are yelling and are not wearing masks. I must get out. But first I have to move through the crowd. I try to hold my breath and I flash on how George Floyd and Eric Garner didn’t have that choice. Violence is very much in the air, so I also have to get out because I don’t want to be caught in the midst of this if it goes out of control. Riots It reminds me of joining riots back in 1969 in Berkeley and 1970 at Columbia. The group contagion is instant, overwhelming. It overtakes you. The danger evokes wild feelings. I don't want to get hurt or infected, so I drag myself home. I must resign myself to watching the crowds from my window overlooking University Place. Hordes of people chant and march toward the park; some wear masks, many also video with their phones. This is freaky! They turn over trash cans on the street corners that border my building, push them into the middle of the street and set them on fire right in front of my building!!!! The smell of smoke seeps in my open window. But it’s OK; it enables me to be part of it. I am full of righteous bloodlust but am also fearful about a bullet coming in through the window. The police are mobilizing. I want to run downstairs and join but I am afraid I'll get sick or killed. People run south toward the park and then inexplicably charge north yelling triumphantly. They chase the police away in front of them. Some guys have boards ripped from NYPD barricades that they hurl like javelins at the police. A bus stops in the middle of the crosswalk on University Place which effectively blocks the entire street. The crowd cheers. I’m momentarily thrilled; has the bus deliberately stopped there to protect the protesters from the police? But then the bus moves on. The group flees south once more. I can’t tell if things are escalating or de-escalating because it’s happening in waves. Here comes a phalanx of police—walking, in cars, and on bikes. The officers who walk have batons clenched firmly in front of them in both hands. Fire My 20-something woke nephew, Charlie, is staying with me. We watch together from the window. A fire on the street blazes steadily, increasing in size. I become frightened that a building will catch on fire. Mine! A car might be on fire on 8th street, but I can’t tell for sure because my sight line from the window is partly obscured by my air conditioner. Charlie and I debate whether to call 911 about the fires. Charlie doesn’t want to bring in more cops and I agree because we don’t trust them. The fire department seems safer, so I call 911. They patch me through to the fire department who say they will come. An FDNY truck has already driven by, but it keeps on going, siren shrieking. A few minutes later, the firefighters show up and appear to be walking around the fires. Then another fire truck comes but again my vision is blocked from seeing what is happening. A huge number of people, mostly guys, show up on Citibikes, or their own bikes or on skateboards. This is a diverse mix of people. Usually, my neighborhood is filled with NYU students, especially in protests, but because of Covid all the NYU students are long gone. I find pictures on Twitter of the corner at 12th street: NYPD vans are on fire. I send the pictures to my friends. I need to reach out. Blankness A month into lockdown, I have found myself dismayed and perplexed by my absence of feelings. I assume it is some kind of muffling defense. It's not dissociation—that feels different. It’s not the sensation of my head being disconnected from my body as it was on 9/11 looking straight down Fifth Avenue and watching the World Trade Center buildings burn—until they disappeared. It scares me now because I am afraid my feelings won’t come back. What might result from being so emotionally stifled? Will I develop physical symptoms or even go crazy? I wonder how it might affect my work with patients, which on the surface seems to continue to be going well. But I'm not sure I'm able to make reasonable observations at this time since I don't have my feelings to guide me. Feelings Erupt However, now I am assailed by feelings. The whole time I am immersed in the uprising all I feel is rage, excitement, and exhilaration. Charlie points out that because everyone has been imprisoned for three months while terrified of disease, perhaps more fuel has been thrown on the sparks of longstanding smoldering wrath about police brutality. The energy of the crowd generates a mood of release and freedom. I feel ashamed that I do not have a whit of fear. I am appalled to find that I am clearly waiting and hoping for things to escalate into more chaos and destruction, and, further, that the thought keeps intruding into my mind that this could not be happening here, in Greenwich Village! Humiliatingly: “This is the land of the gentry!” Likewise, the same voice of shame is running continuously in the back of my head berating me by intoning, “White privilege.” At the same time, I feel proud to live where The People are having their voice. It’s as if I represent the protests because I live here, and Greenwich Village is historically where the Voice of the People has been raised and heard. Aftermath The next morning, I look outside to the street. There is not a trace of last night’s destruction. All is calm. A few people are out and about tending to their errands. I take a walk. Two blocks away there are broken windows in various stages of repair. The Verizon store, Chase and TD Banks, and the classy gelato shop are attended by workmen boarding up the empty window frames. I see police cars—their windows are smashed, their doors are hanging open, and glass is sprayed all over the street. An empty space holds the remaining ashes of a charred police car. Tourists are standing around taking pictures and videos. “Fuck the NYPD” graffiti is sprayed on posh building facades. My building sends out a notice to the residents to stay inside due to the 8:00 pm curfew mandated by Mayor de Blasio. The front door to the building will be locked. This has never happened in my 30 years of living here. I just cannot begin to wrap my mind around what is happening. Breaking Glass Two days later, Tuesday, 6/2/20 A friend sends me “Pundemic” jokes. It is hard to laugh because it is so scary here. It seems that fear and exhilaration seesaw back and forth. Last night while I was sitting in front of my computer screen trying to conduct a therapy session, outside people were on a rampage. They crashed into the grocery store directly below my second-floor apartment. I saw and heard it from my window. I didn’t know whether to run downstairs and join the frenzy or hide in a closet. Stores on my street had been boarded up, some to prevent being broken into and some already having been broken into. Helicopters flew around day and night. My head reverberated with the memory of those imposing black military choppers in Vietnam in the 1970s accompanied by Wagner’s (1856) “The Ride of the Valkyries” in Coppola’s film Apocalypse Now (1979). I was hungry and I wanted to dash out and get some sushi after my last Zoom therapy session, but it was too late. Post curfew. An alert had gone off on both my and my Brooklyn patient’s phones during the session. With six minutes to go, I heard protesters in the street yelling and chanting. Did my patient notice? Should I have asked for her reaction? She appeared to be absorbed in her personal narrative, so I did not interrupt. I was frightened about whether my nephew, Charlie, and I, were in danger in the apartment. Totally distracted, I contemplated whether to end the session, excuse myself for a moment and look outside, tell the patient what was going on, not tell her, take my computer into a different room, or what? I was paralyzed. There was no playbook for this situation! The session thankfully came to an end. With my preoccupation split between the events outside and trying to focus on my patient’s inner concerns, I wondered, is this psychoanalysis? Part II: Effect on Treatment: Sheila This is a story about a short, six-month treatment. Sheila began treatment during a particularly vulnerable time for me in lockdown. Despite years of training and experience, I behaved impulsively, wiping out the frame. This was singularly noteworthy to me given my extensive training in both Modern Psychoanalysis that focuses on pre-oedipal phenomena and my professional origins in the field of addictions. I have a special interest in studying and understanding primitive defenses characterized by going into action rather than being able to mentalize and putting thoughts and feelings into words. What happened to me and why? How did the patient’s dynamics contribute? Is This a Patient? April 2020 Sheila, a 57-year-old woman, started teletherapy about 2 months into lockdown. She had no prior treatment. Only in retrospect have I been able to unearth underlying factors in her abrupt ending of the treatment. Sheila’s presenting problem was high anxiety which she said “normally isn’t really a problem” but which frightened her during her partner’s recent illness with the Coronavirus. This also came up around public speaking. At this time, she and her 58-year-old partner, Roberta (especially Roberta) thought this would be a good time for Sheila to start therapy. I missed noting the major resistance—that she came to treatment primarily to satisfy her partner. Roberta convinced her to initiate therapy by tapping into Sheila’s own operating principle that there was something wrong with her if she experienced a strong feeling. Furthermore, early on she had explained that if she hadn’t had the convenience of Zoom, she wouldn’t be doing therapy. Nevertheless, at the time, I bought into the myth that she was a committed patient perhaps because of my anxiety about maintaining my caseload during the pandemic. I was hounded by a subliminally hovering fear of potential loss, influenced by the horrific news and never-ending wailing of ambulance sirens. Sheila seemed so earnest about wanting help to understand her overwhelming and disorienting anxiety when her partner was sick. In addition, she was charming, and I was fascinated by the glimpses of rich analytic material she let slip regarding her unpredictably hysterical and demonic sounding father. Moreover, there was a corresponding burgeoning negative transference about my “incompetence” that was tantalizing analytically. Under her extremely polite presentation there were veiled intimations about my not being good enough. Her ongoing seeming disappointment once the treatment unfolded paralleled her description of her feelings toward her father. I was also hooked by my interest in her brilliant and eloquent deconstructions of the differences between American culture and that of her Eastern European heritage. For example, she pointed out that she would only want treatment to be conducted in English. If it were in her “mother” tongue, she would experience the situation as more judgmental. Acting Out Early June 2020 In our eighth session, the day after the extraordinary violence in my neighborhood, Sheila’s partner materialized onto the Zoom screen, whispered something, and zipped out. Sheila looked up at me, saying, “I have to leave right now to attend a rally at the mayor’s mansion.” I responded precipitously and, in effect, agreed with her immediate leave taking: “Do you want to postpone the session for later in the week?” Sheila nodded and we rescheduled for a totally anomalous hour I created for the following Sunday. She took off. I said to myself that these were extenuating circumstances. This was a rationalization. I cannot imagine such rescheduling during “normal” times. In the Sunday session she recounted having been caught in a protest that became violent. The police hemmed people in, beat them up and arrested many. She was quite concerned because she had a misdemeanor on her record related to marijuana and was afraid of the consequences of a second black mark in her police file. Fortunately, she and Roberta were able to run away. She was agitated as she talked. I felt like I was back in it myself. At the end of that Sunday session, she canceled the next regular session saying the sessions were too close in time. Instantly, I regretted having offered her the odd hour. In the next usual treatment hour, I broached, “You know, it was my idea to have an impromptu session on a Sunday. Maybe you didn’t want it in the first place?” Sheila waved her hand as if dismissing the whole thing, “Oh it’s fine; it’s in the past.” I gamely plowed on, “May I ask what’s the problem with having two sessions close in time?” Sheila explained, “As time goes on and we re-enter real life I will want to cut down on sessions because I will feel constrained. I’ll feel like, ‘Oh it’s Thursday, my therapy night and I HAVE to be there.’ I’d rather go out dancing with my friends. I know if I called to cancel, you’d ask me ‘how about rescheduling?’ I wouldn’t want that. I know myself. It’s not your fault, that’s just the way I am.” I proceeded to act out feelings of anger and impotence, while wondering to myself, does this person even want treatment, saying “If you have a plan to cancel sessions to avoid feeling constrained, we might as well stop now.” Harsh! I thought she looked stunned, though she said nothing. I managed to cool down and tried to recoup by explaining the theoretical rationale accompanying my outburst. “Consistency in therapy is the foundation for growth. It can be helpful to talk rather than act when a person wants to skip a session.” Sheila again was quiet. At this stage of treatment and at a time when feelings were so high my explanation was completely unrelated to her state of being and flew over her head. I was paralyzed from thinking creatively and empathically of a way to reconnect. Loss of Analytic Stance I was impacted by my own experience and the feelings that overpowered my analytic mind. I missed the present multi-determined symbolic message behind her aversion to feeling constricted. In reality, being constricted day after day was something we were all afflicted by. Further, I had behaved in a constricting manner, re-enacting her relationship with her father in the transference. It seems that an overpowering mutual regression created a contagion in which I “became” the father who traumatized her. Rethinking the rescheduling, perhaps my impulsivity was contaminated by hers. More likely it stemmed from my own fear, excitement, and primordial fury triggered by the carnage outside my window the previous night. Twenty-four hours later, I was still reverberating with feelings too hot to handle that catapulted me into action. If I’d gone ahead and joined a protest, it might have enabled me to vent my own primal wrath. I’d been inside for so long. I was deprived of my in-person work, of seeing friends, and of the pleasure and normalcy of going to restaurants and shows. It now becomes evident how everyday events had provided stabilizing pillars to our lives, at the very least to our ego functions, thought processes, and sense of object constancy. On a deeper level the violence of the protests stirred up rage related to my abusive father and a long-repressed desire to get back at him with a show of force. Every hyper-uniformed and over-weaponized policeman I saw from my window easily stood in for the other brutal figure in my life. What is it going to take to overcome the strong getting to overpower the weak? All these forces roiled inside me. Circumstances beyond my control and my professional experience swayed me to lose my ingrained analytic stance. Slochower (2019) has written eloquently about this phenomenon. I went along with the patient’s abrupt leave-taking in favor of a protest and prompted her to take another session at an hour I would never would have normally given. She, in turn, was swept away by her own overdetermined motivation to cancel. Sheila running out of the session provoked abandonment anxiety in me. The frame, and work during this time of loss, chaos and danger set up a state of insecurity that made some vestige of keeping the frame (in this case a “make-up” appointment), become more important than honoring her agency around canceling. Nature of the Acting Out I speculated that the pressures set off in her were twofold: it was very early in treatment, and she was unfamiliar with therapeutic work as an inexperienced patient. While not yet explicit, she did not want to be there in the first place. It was her partner Roberta who wanted her there for unknown reasons. Maybe the rally was important to her in ways neither she nor I yet understood. She did what her partner wanted, and evidently Roberta wanted her to skip out on the session and go to the rally. Or perhaps Sheila was looking for some excitement to break up the tedium of working from home while missing her co-workers. Prior to this she had not voiced any identification with the BLM movement or interest in political activism. I was informed by my understanding of the forces that give rise to impulsivity. On one hand, there are certain patients who are fixated at a pre-verbal level of development in which action is a primary means of communication as well as defense against intolerable feelings and also a form of discharge of affects that are too much to handle (Abrams, 1976; Goldwater, 1978). Additionally, extreme stress at any stage of life can plummet an otherwise more integrated person into a regressive state of black and white thinking and corresponding impulsivity (Music, 2021). I believe it was this state of affairs that drove both Sheila and me into a more action-oriented mode. Abortive Termination October 2020 Sheila emailed to cancel 40 minutes before the session citing an opportunity to attend a Zoom meeting for a new volunteer organization. She added a few days and times when she could come but noted she was not free on the weekend. I was confounded. She knew my 24-hour cancellation policy and she did not send the usual payment through her phone app. I understood that she was influenced by the one-time experience in which I abandoned my cancellation policy. Based on my having changed the frame, she now offered some regular and irregular rescheduling possibilities (the weekend). Now I had to do damage control wreaked by my veering so far off course. I was stuck with having to address this intrusive, unexpected issue that had popped into our work! I did not want to get into money or policy issues by email outside the session. I had to deal with having unwittingly conspired with her resistance and had thus sacrificed the frame. Of course, in addition to my countertransference-powered enactment secondary to my responses to the protests, Sheila’s original ambivalence still held sway, as she’d been effectively mandated by her partner for reasons that didn’t quite make sense to her. My head was spinning. In fact, not knowing what to do, I rapidly developed a headache upon taking in her email aborting her session. Ultimately, I decided to text back, “Let’s just leave it till next week.” In the next session I asked, “How did the Zoom go last week?” She basically reiterated, “At the last minute my friend Ruth called and I couldn't get out of it.” Plowing on, I uncomfortably pointed out, “I notice that you didn’t pay for the missed session. I wonder if I’ve been confusing about my cancellation policy.” A reasonable person, Sheila replied, “No, I know what your 24-hour policy is.” I said, “So, unfortunately, I need to charge you for the session. I realize it must feel unfair given that you canceled through no fault of your own.” Sheila, again characteristically waved it (and me) away with her hand, replying, “Oh, no, it’s fine—I get it.” She went on to have her usual repetitive session focused on surface issues related to work. There was also a subliminal sense that something was amiss in the treatment which I hadn’t figured out yet. And of course, she must have been angry about being charged. In the next session, she showed up in order to quit. Charging her obviously reinforced the resistance she already had to treatment. My efforts to access her anger about that charge which were bound to feel unfair were unsuccessful. However, I believe this early termination was in the wind for some time in addition to the confrontational incident I instigated that led to her quitting. Only in retrospect have I been able to unearth more fully underlying factors in her abrupt ending of the treatment. Reflection I went along with her ending as I agreed with her that the sessions’ content was repetitive and perfunctory, and I hadn’t helped deepen it. I put the onus on myself to obviate any feeling she might have that I was blaming her. I also noted silently that she had succeeded in letting me know how incompetent and inadequate I was, just like her father, but we were worlds away from bringing that to awareness. I do wish I had circled back to try to unearth the multiple pieces of acting out. For example, I might have asked:"I know we’re ending, but I have a question or two that might interest you. May I ask, was this treatment more your partner’s idea than yours? Is there anything you might have wanted to get out of it?" Or: "When we first started you mentioned the anxiety you had when Roberta had Covid and when you encountered unexpected turbulence on a plane. Was your idea that there was something wrong with your feelings? And that therapy should eradicate such feelings?" There was at least the shift in the direction of Sheila’s increased autonomy. Characterologically acquiescent to authority figures, she came to assert herself with me, a woman in authority, in this short treatment. To have been able to metaphorically and transferentially kill off the father toward whom she felt ineffectual, may have been exceedingly satisfying. Case Conclusions This is a case in which my rational thinking was hijacked at two points. A patient made a seemingly reasonable request for therapy, but I overlooked that she was coming at the behest of her partner. My rational mind went offline at the outset. Secondly, I was overwhelmed by the surrounding political and cultural environment within the life-threatening worldwide pandemic. We all felt unsafe and under siege. The violence, both in the air and in reality, put me into a regressive hypervigilant mode in which I was more action oriented than able to mentalize. In addition, I had subsequently had a wildly intense experience that overlapped with the patient’s, which led to my suggesting a rescheduled session instead of waiting for the request to come from her. I didn’t have the opportunity at the time to consult with colleagues to gain the control and objectivity that come from processing a difficult situation. There were at least five levels of destabilizing impact contributing to the enactment in the case: the overarching global pandemic, the protests in my neighborhood, the issues from the patient’s history and current life, the dynamic in the analytic dyad, and the triggering of my own early attachment issues. The latter three are comprised of:the patient’s characterologically avoidant defensive structure leading to a push-pull dynamic in the therapy setting; the trauma she suffered having had to face the possible loss of her very ill partner over a four-week period in the beginning of Covid. NYC was the epicenter of Covid-19 during the Spring of 2020, with 4 to 5 thousand deaths a day. Sheila experienced terror, helplessness, and uncertainty from the beginning. She must have feared that she could get sick as well. Her having been pushed into treatment and her desire to leave in a way that wouldn’t hurt my feelings made dashing out to a protest acceptable to her under the circumstances. A dynamic in the treatment in which Sheila’s unspoken wanting to leave triggered my counter-response to chase. This created an avoidant/anxious attachment dynamic between us. My own early anxious attachment wound was triggered by the similar overarching fear and helplessness that everyone was subject to in pandemic life. This became acutely activated by an at first subliminally, but then overtly, fleeing object. Regressive Dynamics I was subject to the protests in my neighborhood and corresponding immediate threat to personal safety. At the same time, I had a conflict as I wished to join in. I was deluged with the contagious rage in both the here and now and from my history. I was also angry and frightened about both the political situation and the pandemic. I went into an action-oriented mode that precluded any possibility of mentalizing and containing. This is the kind of regressive motion, much referenced in trauma studies, that catapults the individual into concrete thinking that can lead to action, and which freezes the capacity to mentalize (van der Kolk, McFarlane & Weisaeth, 1996; Hill, 2015; Schore, 1994). Precipitous action tends to move in a destructive direction. “To create—whether it’s a new watch or a new life—takes longer than to destroy” (Goldwater, 1994, p. 21).Some old and new humbling lessons emerge from this experience that inform my teaching and consultation practice:Don’t go into action when feelings are high. Keep in mind that when a patient comes in to satisfy someone else, they are entering into treatment with a built-in destructive resistance. They already have one foot out the door. Focus as soon as possible on that. The analyst is fallible; sometimes we’re going to make mistakes, and that’s how we learn; we are all subject to creating enactments or acting out when overstimulated. Most importantly in this case was the overarching impact, conscious and unconscious, of doing therapy in the time of Covid. “It therefore leads to the expanding literature on how Covid- 19 affected our practice and attunes the clinician to slow down and expect enactments when a worldwide cataclysm occurs” (Zerbe, personal communication, March, 2022). We can't minimize the impact of the stress we’ve all been subject to during this awful time. But we can use it to learn by reflecting retrospectively and fortifying ourselves with anticipation of what could happen. Part III: General Considerations Regarding Treatment in the Time of the COVID-19 Pandemic Pandemic Fatigue How can we make sense of the incomprehensible? What, exactly, was (and at the time of this writing, still is) the inchoate intrapsychic and somatic impact of living in a chaotic dangerous environment and working in a virtual world with a dramatically changed frame, and within the reality of the trauma we shared with our patients? How did pandemic induced stress, also dubbed “pandemic fatigue” (Zerbe, 2020), manifest? Some of it has been more tangible, like the cognitive problems many complained of, such as forgetting things, making mistakes, having trouble finding words, distractibility and not knowing what day it was. Other regressive manifestations have occurred in the body and the psyche—that of experiencing anxiety, depression, suffering accidents, pain, or actual illness. Zerbe (in press) noted, within her treatment and supervision practice, reports of changes in “sleep, energy level, exercise and eating patterns and somatic reactivity.” Anecdotal reports from colleagues and patients included attacks of colitis, non-Covid respiratory issues, muscle spasms, falling and breaking bones or teeth, and more. There was also hypochondriasis, focusing in on any possible symptom, out of terror that it was incipient Covid, especially before the vaccines. Some reported unexplained surges of rage that leaked out into fights and arguments with friends and family. It wasn’t unusual to have an extreme reaction to minor technological quirkiness such as the screen freezing, looking up the other’s nose seemingly a few inches away, hearing clicking or rustling noises secondary to the other’s mic rubbing against their necklace, getting seasick when the patient Zooming from their phone is moving their screen around, etc. Harris vividly depicted our “new normal,” a kind of physical regression of sorts in which she wasnoticing and tracking all too frequent falls and tumbles…. Bones cracked, bruises, unsettling moments of disarray and confusion. I think this is one of the unexpected results of our long hours of sitting, our immobility, the work on screens that reduces our sense of dimensionality and movement in space (Harris, 2021, p.100) We are now mutually confronted with the inevitability of death as real and immediate. It is no longer possible to deny or avoid whatever we’ve thought death was; it’s here. Rose points this out while noting that our former ability to tune it out “which may seem to be the condition of daily sanity—has been revealed for the delusion it always is.” This is a pretty scary thought! She colorfully elaborates that “In the midst of a pandemic, death cannot be exiled to the outskirts of existence. Instead, it is an unremitting presence that seems to trail from room to room” (Rose, 2021, para. 3). Long before Covid, analytic writers were already documenting the difficulties of working remotely (Essig, Turkle & Russell, 2018; Bayles, 2016, to name a few). Now there was no choice. These are some of the burdens that therapists had to shoulder while working (See Harris et al., 2021). How could we contain our own engulfment as well as that of each idiosyncratic patient? Changing the Frame In order to unpack what happened in the case resulting from numerous breaks in the frame, it will be helpful to consider the meaning and use of the frame. This case was impacted by a number of changes to the traditional frame—one being pandemic induced--zooming from the therapist’s and patient’s homes, instead of sitting together in an office, and the other, even more outside the box, both therapist and patient unexpectedly and coincidentally being swept away, albeit each quite differently, by involvement in Black Lives Matter protests within the larger context of the pandemic. It has been said that the set-up of the therapy process—in a room together, at a stated time, with a stated frequency, a fee, and a mutually agreed upon goal—provides hearty leverage for the work of therapy. That form, the frame, plus mutually positive intentions, creates a space for the work to happen almost irrespective of the content of what is said (R. Unger, personal communication, March 17, 2019, as noted in Frankfeldt, 2020). The frame is at once a sturdy yet pliable mechanism for enabling analytic work to take place. Many forces come to bear on its form and flexibility. Prior to Covid, much had been written on the importance of the frame even before it was drastically altered by our not being able to meet in person. A quick search in PEP-Web yielded 12,803 results of its being mentioned. It is a complex entity which will be helpful to examine in more detail in order to think about the effect of its alteration during Covid in general and the acting out in this case, specifically. We think of the frame as the structure that establishes an atmosphere of safety for both patient and analyst necessary for promoting the therapeutic relationship. Its rules are both implicit and explicit. The contract between therapist and patient is explicit and consists of an agreement about frequency, fee, when the fee is to be paid, and length of the session. Sooner or later, depending on the therapist’s judgment, confidentiality and cancellation policies will be discussed. It may or may not include a plan for use of the couch. Until more recently it was understood that sessions took place in the therapist’s office. The office was a crucially significant part of the frame. Additionally, and concretely, the frame, as exemplified by a brick-and-mortar office, creates an implicit boundary between the inner (analytic) and outer (reality/environment) worlds. The space provided by the frame is conducive to a safe regression to intrapsychic, early developmental states which can then be experienced and put into words. It is through this means that we have an avenue to making the unconscious conscious. Another aspect of the analytic situation enfolded within the frame is the emotional power differential created by one person asking for help and the other taking on that facilitative role. The analyst defines under what circumstances this will happen: use of the office, payment of a fee set by the analyst, use of a time within the analyst’s working hours and sometimes the analyst’s sitting up while the patient is lying down. The tension created by the asymmetrical balance of power provides another vehicle for the fostering and analysis of transference, countertransference, and resistance as they develop. It is this space that allows for enactments which provide more opportunity for studying and understanding the patient’s and analyst’s co-created dynamics. Until recently, the frame, and especially the holding space provided by the therapist’s office, further established an implicit as well as real separation of the outside world from the inside world. As such, the office was beautifully positioned to invite the patient to move from an external/reality focus to an internal/transference-oriented focus uncontaminated by everything going on in the outside world. Theoretically, at least. Of course, we know there is no such thing as complete lack of contamination, but there is still a big difference in the conduciveness to free associate by being in a quiet, non-stimulating office atmosphere versus sitting in a car by the side of the road with traffic whizzing by, neighbors stopping to say hello through the window, while notifications are going off on the screen. What, in fact, was the effect of the plethora of changes, violations even, of the frame on treatment? We have done it as a matter of course; we had no choice. But given it is precisely the structure of the frame and the therapeutic leverage afforded by analyzing changes to the frame, that is the bottom line of treatment, how do we assess what is lost when the patient is sitting on the floor of their relative’s bathroom, worried that their spouse or parent or child can hear their bitter complaints, while we are on the other side of the Zoom room, perhaps also worried about who in our house can hear what is going on in the session? In an ideal world, the analyst is unencumbered by personal problems, emergencies, physical pain, etc., so that we can focus on the patient’s communications and our corresponding associations and fantasies. Maybe this is rarely the case! But we do have an assigned role and purpose. We strive to bring our best effort to look at our mutually created dynamics and what that tells us about the patient for the benefit of the patient's growth. Extra Analytic Self-Disclosure Analysts found themselves breaking the transference by engaging in self disclosure as a matter of course. When physical and/or psychic survival are at stake, more existential questioning goes by the wayside. We were faced with myriad interactions in which we had to make a choice. Should we be dealing on a reality level or on one which would access an underlying historically-based conflict or anxiety? Under “normal” circumstances the choice would be easier depending on factors involving the patient’s developmental level and time in treatment, but the Covid lifestyle introduced an added feature that could make delving crass and inappropriate. How would we make these decisions and how would we know to what extent they were made based on our own active anxiety? In a discussion, Dr. S. Sherman postulated that being “all in it together” prompted therapists’ reasonable urges to self-disclose (S. Sherman, personal communication, February 2022). In an email communication M. Cohen, LCSW, added thatWe therapists are more deprived of our usual friend and family social interactions because of Covid isolation. We are more likely to “use” our patients to satisfy our social needs as we ourselves are hungrier for connection. We are also working harder to foster and maintain a connection with our patients through the medium of a computer screen (M. Cohen, personal communication, December 2021). Previously, she wrote “We are striving to have three dimensional relationships in a two-dimensional space.” (Cohen, 2020, p.17). Effect of Zoom on the Frame The switch from face-to-face talking therapy to e-therapy has hardened the therapist’s ability to create a safe and containing space, as Mateescu (2021) reminds us. In online therapy, the physical therapeutic presence, which is regarded as a critical element in therapy efficacy, (Geller & Greenberg, 2002) needs to be rethought. Along with these elements, the self is negatively affected in online therapy, as it becomes a “disembodied self” (Weinberg & Rolnick, 2020, p. 6). Conversely, Lemma (2017) argues for an embodied presence/self in online, mediated therapy, the difference lying only in the way we perceive and experience it (Mateescu, 2021, p. 114). What happens when the frame is drastically changed and takes place within the shared experience of the patient and the analyst in the context of chaos and danger in the world? What is the effect of working from one’s bedroom within the eeriness of a virtual background? How does it affect the patient to see the analyst’s hair or earrings disappearing and appearing if we move our head? A colleague labeled this phenomenon “psychotic” (N. Stiefel, personal communication, June, 2021). Are patients supposed to be able to put into words what that does to them? Would it be disruptive to ask? Generally, we would wait until the patient brings it up, but in the two years of working on Zoom no patient in my experience has commented on the visual peculiarity of “Zoomness” in treatment. Why is this? Chaos A quick Google search (Oxford Languages, n.d.) of “chaos” produced: “complete disorder and confusion.” This aptly describes our collective experience during the first few months of the pandemic given the political situation, the invisible but ever-present disease and the BLM movement. Our societal infrastructure, the larger “frame,” was shattered to a great degree. Some reacted by isolating psychologically as well as physically, while others of us reacted by wanting, by any means, to move closer to others. This produced both a tension and a tension-relieving mechanism in online therapy. The hierarchy that previously existed, for better or worse, was also eliminated with respect to our shared circumstances. This, again, could be experienced paradoxically as both destabilizing and reassuring. But how to parse this as a clinician?! On one hand it meant making concessions within the frame and our accustomed discourse and on the other it meant attempting to maintain what we could of a frame that would still enable analytic work to take place. The internal tension and feeling of helplessness, one of the most difficult and painful feelings that we, as helping people, can undergo (Hoffer & Buie, 2016), produced more stress. I believe (and based on anecdotal evidence in addition to my own experience) that this created the potential for more enactments to take place than would be true in normal times. On the other hand, this struggle opened up the possibility for creativity between analyst and patient to co-create a generative space. New meanings would be bound to result from observing, reflecting and mentalizing together about the larger environment. Coping Constructively We can't minimize the impact of the stress we’ve all been subjected to during this awful time. But we can learn from it by reflecting retrospectively and fortifying ourselves with anticipation of what can happen moving forward. Bion (1970) pointed out that “catastrophic change” occurs after a crisis in which there is an unexpected loss of something we rely upon. What does this change look like? Is it good or bad; constructive and/or destructive? As difficult as it has been, it is also possible that as individuals and as a group we have opened up new tracks to treatment, just like new neuronal pathways. We may still be too much in the midst of the storm to be objective. This is an evolving process. People are regularly sharing and conducting webinars on the value of mindfulness, walks in quiet, lush natural settings, taking up hobbies, taking a break from the news, learning how to start new healthy habits, and curtailing counterproductive old habits. One thing that differs in these times is the proliferation of sites proffering advice about self-care. “Intensive self-care” might be more appropriate, especially for those in the mental health field. Garnering support from colleagues, friends and family is imperative; providing support can be strengthening as well. Activism can boost feelings of efficacy. Each of us finds our own favored methods. At times it may be easier to withdraw and isolate to protect oneself, and sometimes this is necessary. Yet in the long run this is antithetical to our work. Additionally, it is all too easy to “use” the patient in our own interest; although, as has been said, it is natural, human, and often therapeutic to acknowledge with patients our shared situation—there is a fine line to consider in making that judgment. The most powerful antidote to dealing with the internal chaos the real-life situation has imposed upon us is, as always, talking. Individually and in groups, sharing our conflicts as practitioners helps. Many groups both leader-led and peer-led have sprung up to provide support. No less important is discussing with the patient the effect of any confusion, enactment, or outright acting out on the therapist’s part on the treatment. This is invaluable and growth enhancing for both. Conclusion In Part I, I describe my personal experience in a Black Lives Matter protest that became violent at the very beginning of the Covid-19 pandemic. Part II recounts how this confluence of overwhelming events had the effect of overriding my efforts to hold the frame with a patient who was ambivalently in treatment. Part III elucidates more generally how analysts were impacted by the chaos affecting the way treatment was conducted during Covid. Of particular interest, and highlighted throughout, is the way the psychoanalytic frame was muddied by moving to online therapy. I point out the myriad ways that the frame has been distorted out of necessity and how that can add to confusion in an analysis. This is a stressor that we as analysts must contend with in everyday life and work. I talked about some ways we might take care of ourselves. So far, the analytic community has been able to put into words external manifestations of how we have been impacted as well as venturing into the somatic (Zerbe, in press). There is so much more that is as yet unknown. Perhaps I have raised more questions than I have answered. We’re currently in the process of studying and evaluating as we evolve through this ever-changing landscape, inventing new ways of functioning within a new analytic form. Will some good come of this? Might we track post-traumatic growth? (M. Cohen, personal communication, February, 2022). It is going to take time to process and understand what we’ve been through and how it has shown itself in each of us. We have the tools of awareness and self-analysis in addition to that of leaning on colleagues and mentors for help. It is essential that we keep striving to make sense of our experience for our own sakes as well as that of our clients’. Note Valerie R. Frankfeldt, LCSW, PhD, is the former Director of Training and current faculty member, supervisor and training analyst at the Psychoanalytic Psychotherapy Study Center. She is a Certified Imago Relationship Therapist and Modern Psychoanalyst. Dr. Frankfeldt is a graduate of the New Directions psychoanalytic writing program, focusing most recently on the dilemmas posed by the intersection of technology and psychoanalytic treatment. She is in private practice in Greenwich Village, working with individuals and couples and providing clinical case consultation. Valerie R. Frankfeldt, LCSW, PhD is a faculty member, supervisor and training analyst at the Psychoanalytic Psychotherapy Study Center. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abrams EK The narcissistic transference as a resistance Modern Psychoanalysis 1976 1 218 230 Bayles, M. (2016). A review of screen relations: The limits of computer-mediated psychoanalysis and psychotherapy by Gillian Isaacs Russell. London: Karnac Books, 2015. 224 pp. Contemporary Psychoanalysis, 52(4), 653–659. Bion W Attention and interpretation: A scientific approach to insight in psycho-analysis and groups 1970 Tavistock Coppola, F. F. (Director & Producer). (1979). Apocalypse now. [Motion Picture]. Omni Zoetrope Studios [aka American Zoetrope]. San Francisco: USA. Cohen M Deprivation and abundance in the time of Coronavirus The Clinician 2020 51 1 16 18 Essig, T., Turkle, S. & Russell, G. I. (2018). Sleepwalking towards artificial intimacy: How psychotherapy is failing the future. Forbes Magazine Online. June 7. New York: Forbes Media, LLC. Frankfeldt V Digital communication in psychoanalysis: An oxymoron? Psychoanalytic Social Work 2020 27 1 1 16 10.1080/15228878.2019.1625791 Geller SM Greenberg LS Therapeutic presence: Therapists’ experience of presence in the psychotherapy encounter Person-Centered and Experiential Psychotherapies 2002 1 1–2 72 73 Goldwater E Impulsivity, aggression, fantasy, space, and time Modern Psychoanalysis 1994 19 1 19 26 Goldwater WE A model for understanding and treating the impulsive patient Modern Psychoanalysis 1978 3 173 196 Harris A Working in the shadow of COVID-19 Psychoanalytic Psychology 2021 38 2 99 100 10.1037/pap0000356 Harris A Csillag V Cutner N Freeman-Carroll N Mayson SJ Rufino M Clinical life in the context of the pandemic American Journal of Psychoanalysis 2021 81 3 361 394 10.1057/s11231-021-09308-9 34341465 Hill D Affect regulation theory: A clinical model 2015 W. W. Norton & Co Hoffer A Buie DH Helplessness and the analyst’s war against feeling it American Journal of Psychoanalysis 2016 76 1 1 17 10.1057/ajp.2015.56 26912242 Lemma, A. (2017). The digital age on the couch. Psychoanalytic practice and new media. London & New York: Taylor & Francis Group. Oxford Languages, (n.d.). Retrieved from https://www.google.com/search?q=chaos&oq=chaos&aqs=chrome..69i57j46i433i512j0i20i263i512j0i433i512l2j46i131i433i512j0i131i433i512j0i433i512j0i131i433i512j0i512.8403j0j7&sourceid=chrome&ie=UTF-8 Mateescu VM Distance and boundaries issues in the transition from face-to-face talking therapy to online therapy in the time of Covid-19 Studia Universitatis Babes-Bolyai Sociologia 2021 66 2 113 123 10.2478/subbs-2021-0010 Music, G. (2021). Attachment, deadness, desire, and addictive buzz traps. Attachment: New Directions in Relational Psychoanalysis and Psychotherapy, 15(2), 198–213. Rose, J. (2021). Life after death: How the pandemic transformed our psychic landscape. The Guardian Online. December 7. Schore AN Affect regulation and the origin of the self: The neurobiology of emotional development 1994 Lawrence Erlbaum Associates Inc Slochower J Uncanny resonances: A discussion of “did you say he has a Hitler mustache?: Analytic vulnerability and destabilization in the face of trauma” Psychoanalytic Dialogues 2019 29 1 28 34 10.1080/10481885.2018.1560872 van der Kolk, B. A., McFarlane, A. C. & Weisaeth, L. (1996). Traumatic stress: The effects of overwhelming experience on mind, body, and society. New York & London: The Guilford Press. Wagner, W. R. (1856). Ride of the Valkyries. [Musical Composition]. From the opera, Der Ring des Nibelungen. Leipzig, Germany: Schott Publishing. Weinberg, H. & Rolnick, A. (2020). Introduction. In Theory and practice of online therapy: Internet-delivered interventions for individuals, groups, families, and organizations. H. Weinberg & A. Rolnick, (Eds.). (pp 1-10). New York and London: Routledge. Zerbe K Pandemic fatigue: Facing the body’s inexorable demands in the time of COVID-19 Journal of the American Psychoanalytic Association 2020 65 3 475 478 10.1177/0003065120938774 Zerbe, K. (in press). The analyst’s self-care: Further reflections after the pandemic on cultivating resilience and attending to the body-mind. Psychodynamic Psychiatry.	36460894	PMC9716150	NO-CC CODE	2022-12-03 23:20:54	no	Am J Psychoanal. 2022 Dec 2;:1-22	utf-8	Am J Psychoanal	2,022	10.1057/s11231-022-09374-7	oa_other
==== Front Curr Urol Rep Curr Urol Rep Current Urology Reports 1527-2737 1534-6285 Springer US New York 1122 10.1007/s11934-022-01122-9 Education (G Badalato and E Margolin, Section Editors) A Review of Mentorship in Urology: Are We Satisfied? Chen Annie Harnett Joseph Kothari Pankti http://orcid.org/0000-0003-2874-6528 Ernst Michael [email protected] grid.36425.36 0000 0001 2216 9681 Department of Urology, Stony Brook University, 101 Nicholls Rd, Stony Brook, New York, NY 11795 USA 2 12 2022 110 8 10 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Purpose of Review To evaluate the state of mentorship in the field of urology. Recent Findings Mentorship has been shown to decrease burnout, increase recruitment of underrepresented minority groups, and have a positive influence on the career trajectory of mentees. Approximately half of surgical residency programs have mentorship programs. The current literature supports the idea that formal mentorship programs are successful based on level 1 satisfaction scores. However, studies are sparse and of low quality. Mentorship program success is rarely objectively measured. Summary Structured mentorship programs appear to be beneficial, but require serious planning, evaluation, and ongoing support without which the programs can fail. Future research should be focused on objective and measurable metrics of success. Keywords Mentorship Urology Medical education Mentee Mentor ==== Body pmcIntroduction To all of the mentors in urology, a toast to you and the hope that you provide. Mentorship is a collaborative interpersonal relationship that demands time and energy from both mentor and mentee and should be an integral part of training programs. Mentorship is in the service of continued learning and has been identified as a key driver for many different domains including professional development, diversity and equity, wellness, productivity, and career satisfaction [1]. Despite the importance of mentorship in the medical field, its study often lacks scientific rigor. Mentors are often likened to “coaches” or “sponsors,” but these titles are not interchangeable. A coach’s objective is to develop or improve a distinct skill. Their role may also be to integrate a person into a team in order to achieve a collective goal. A sponsor is typically a well-regarded individual that leverages their own power to influence the trajectory of a pupil’s career. A mentor, however, typically knows the mentee on an individual level and aids in achieving the mentee’s unique goals [2]. There are many different types of mentorship relationships [3]. The classic relationship, dyad, pairs one more senior mentor with a junior mentee. Other types include “speed mentorship” where a mentee will briefly be paired with multiple mentors. This type of mentorship works well when it is desirable to cover multiple topics or obtain multiple opinions, with the hope that a longer relationship could develop. “Functional mentorship” is similar to coaching in that it is a relationship structured around the learning of a specific skill. “Peer mentorship” pairs a group of more junior people and enables them to work together to reach their goals. This works well when there is a paucity of senior mentors. Similarly, “group mentorship” works well with limited mentor availability by pairing multiple mentees with a single mentor. Peer mentorship and group mentorship can be combined in “facilitated peer mentorship” [3]. Within a formal program, the mentorship relationship type can be chosen to meet an individual’s and institution’s goals. Historically, the classic model of mentorship has been a part of surgical training. As surgical training has evolved, so too has the relationship between mentor and mentee. Many questions regarding the best forms of mentorship within training programs remain unanswered. Are formal and structured programs superior to informal curriculums? What is the best way to design an effective mentorship program for urology residents, fellows, and faculty members? The goal of this review is to describe the current published literature on mentorship within the field of urology and other surgical specialties, summarize the best practices on mentorship program development, and identify areas where mentorship programs could be used to advance growth and development within the field of urology. Does Formal Mentorship Work? Formal mentorship programs are those that have established an infrastructure for mentoring, though the exact process to achieving such a foundation and the components themselves vary from program to program. Most successful programs share several key components: (1) mentor preparation, (2) planning committees, (3) contracts, (4) pairing of mentors/mentees, (5) mentorship activities, and (6) formal curricula for mentees [3]. Some programs have also provided compensation in the form of time/funding/Continuing Medical Education (CME) credits as motivation for participation. Formal mentorship programs have been shown to be successful by multiple systematic literature reviews [3–10]. Mentorship programs for physicians have been shown to increase clinical productivity [11], professional development [12–14], academic accomplishment [13, 15, 16], and self-confidence in a career [13]. Mentorship programs have led to increased faculty retention [12, 17, 18] and faculty remaining in academic medicine[12]. In many studies, the participants’ satisfaction with the mentorship program exceeded 90% [14, 19–22]. Additionally, mentorship programs have helped medical students to engage in self-reflection [23] and achieve success on clerkships [24]. Specific departments have used mentorship programs to successfully increase applications to obstetrics/gynecology [25], neurology [26], and primary care [27]. Mentorship programs geared toward those in training have shown participant satisfaction whether the mentees are matched with senior faculty [28], near peers [4], or peers[9]. Many would argue that a mentorship relationship that develops organically or informally is the ideal and preferable to structured mentorship programs. A 2015 study comparing formal and informal mentorship showed that a self-identified “actual mentorship relationship” developed in 70% of cases where a formal mentor was assigned. Furthermore, there was no difference in the number of meetings and subjective “investment in mentorship” between formal mentorship relationships compared with informal ones. This study demonstrated that a formal-organized mentorship intervention increased the prevalence of mentorship without decreasing the quality of mentorship from the perspective of mentees. This is a strong argument for the benefits of mentorship program creation [29•]. A challenge in reviewing the academic literature concerning mentorship program creation is that objective outcomes are very rarely measured, and this “success” is often defined as participant satisfaction. Two systematic reviews by Farkas et al. found that objective outcomes were only measured in 8/20 and 7/30 mentorship programs for women and medical students, respectively [7, 8•]. Similarly, a review by Chua et al. of structured mentoring in medicine and surgery found that only 30 of 71 included articles described an evaluation of the mentoring program [5••]. This is a weakness that is seen throughout the mentorship program evaluation literature. Scientific rigor is lacking from the description of mentorship programs, and no conclusions can be made regarding the effect of individual components of the programs on mentoring outcomes. Major limitations in the existing literature on mentorship programs include not just the lack of objective data but also a lack of long-term evaluation of the impact of programs on individuals’ careers. There is a need to bolster claims that have been made by subjective and unvalidated studies published in the literature thus far. Newer literature recommends analyzing outcomes such as research grants, publications, validated mentorship evaluations, quality improvement measures, academic advancement, and career satisfaction in a standardized fashion, in addition to more subjective measures [30•]. The Current State of Mentorship in Urology The current state of mentorship in urology training is variable. In a mentorship survey of 64 urology residency program directors (response rate 54%), 75% approved formal mentorship programs, and 58% had an established program. The most likely reason for not having an established program was the consensus that an informal mentorship program was sufficient. Interestingly, only 5% of residency programs had official training courses for faculty mentors, and 20% had career development courses for trainees. Thirty-eight percent of programs did not have requirements for the frequency of meetings between mentor and mentee [31••]. So even in “formal” mentorship models, there is great variability in curriculum, rigor, and structure across residency programs, and the optimal curriculum, structure, and timeframe for meetings are not well elucidated. Historically, trainees have sought their own mentors, and this has “worked” since the apprenticeship model of Halsted. However, informal mentorship may self-select for trainees that are more outgoing or have socially dominant traits, possibly excluding more introverted trainees. Additionally, there is a significant risk of perpetuating the inequities already present within medicine and urology if we are not deliberate about how we distribute the valuable resource of mentorship time and commitment. When considering the other attitudes that exist within the field of urology, mentorship certainly appears to be influential and desired. In a survey of 73, recent and current pediatric urology fellows (response rate of 58%), over 90% had an influential mentor. When choosing where to apply for pediatric urology fellowship, advice from mentors was rated as the most important factor [32]. Similarly, participants in another survey of 111 recent and current pediatric urology fellows (75% response rate) ranked the potential for mentorship as the most important factor when choosing a job after fellowship [33]. Across all urology subspecialties, mentorship was ranked as the second most important factor impacting the desire to pursue a fellowship by graduating residents. In fact, residents with a mentor were 20 times more likely to pursue a fellowship [34]. A separate survey-based study of 1149 successfully matched urology applicants identified strong mentorship as one of the reasons these students specifically elected for urology as a specialty. Even at the medical school level, mentorship appears to be influential in a student’s pursuit of urology [35]. Successful Mentorship Models in Urology At the medical school level, resident-led mentorship models have shown promise in urology. The UReTER program, led by residents at the University of California San Francisco (UCSF), was created to mentor Black, Indigenous, and LatinX medical students applying for urology residency during the first year of the COVID-19 pandemic in 2020. Mentors (current urology fellows or residents) and mentees (students applying to urology residency) were recruited using social media and email list serves. Each mentor would be assigned up to three mentees and asked to meet at least two times during the course of the application cycle. A total of 101 mentors and 71 mentees were recruited during the initial year of the program, resulting in 71 mentor–mentee pairs of the dyad mentorship model. Follow-up data was collected via a survey sent out on Match Day. Of the 16 survey respondents who participated in the 2021 Urology Match, 94% (15) were matched into a urology program, and 38% (6) felt their match was directly attributable to the program [36]. Another program, #UroStream, paired 111 individuals, including medical students and some graduate students, interested in applying to urology residency in 2020 with 93 urology resident mentors via a social media platform during the COVID-19 pandemic. Among all 111 students, they found that 19% of mentees lacked any affiliation with a urology department, 24% had no urology mentor, and 32% had no exposure to urology at the time of enrollment. Program success was quantified by using the MEMeQ, a validated survey sent out at the conclusion of residency interviews and after Match Day. Among the 29 students (26%) who completed the full MEMeQ, the overall satisfaction was 6.1/7, “very satisfied,” and students identified obtaining guidance on ERAS application and help with residency program selection as their key accomplished goals [37]. These formal mentorship models have shown a measurable positive impact on students. Mentorship programs have also been described within urology residency programs themselves. Baylor College of Medicine created a 5-component resident wellness curriculum for their urology residency program that integrated structured mentorship, funding for social events, and wellness education [38]. They employed both dyad and group structures of mentorship in their program. They analyzed outcomes on burnout using validated questionnaires. High resident burnout rates were noted at baseline; however, implementation of the wellness program did result in a significant improvement in depersonalization and decreased the level of distress. Though these results were seen after the implementation of a multifactorial system, they provide further evidence that mentorship programs in urology residency can have a favorable impact on resident wellness and burnout. Mentorship programs also exist regionally and nationally outside of individual institutions. The American Urological Association (AUA) has organized many mentorship programs for urologists at points throughout their careers. For example, the USMART academy provides research mentorship, the AUA Leadership Program provides organizational leadership training, the Young Urologist Speed Mentorship program is run for residents at the annual meeting, and the upcoming Global Residents Leadership Retreat will help residents become more involved with the AUA [39]. Additionally, the Society for Women in Urology and several of the subspecialty organizations run mentorship programs for interested medical students, residents, fellows, and/or faculty; however, there is a paucity of published data on outcomes beyond participation numbers. Mentorship Programs Across Other Surgical Residency Programs While mentorship is becoming more common in urology programs, it has also been studied in other surgical specialties. Time is the most difficult barrier to the implementation of formal mentorship programs. This is especially evident for surgical residences where long hours in the operating room are routine, making meetings difficult to arrange. Only about half of surgery residency programs in the USA have established mentorship programs [40]. The majority of programs are a dyad model where each mentee is assigned to one mentor. One general surgery residency mentorship program that required at least three mentor meetings annually and two social events showed that residents reported an improved perception of faculty involvement and support after the implementation of the program [41]. Another general surgery program described a “Mentor Match” design where both residents and attendings filled out surveys based on the six ACGME clinical competencies. Residents reported perceived weaknesses and were matched with an attending that reported their weaknesses as strengths. One year after implementation, 75% of survey responses indicated they felt as if “Mentor Match” was an effective tool in assigning a mentor, and 83% felt they improved on their weaknesses over the year [42]. Otolaryngology residency programs in both the USA and Canada have published studies about their mentorship programs. One study by Grugel et al. in 2010 found a low percentage of otolaryngology residents had a formal mentor and only 26/71 of respondent residency programs had a formal mentorship program. The authors suggested that the need for formal mentors may be diminished in residencies with fewer residents since they already work very closely with all faculty members [43]. However, a 2010 study of otolaryngology chief residents showed positive benefits of formal mentorship programs. This study assessed the chief residents’ experience with mentorship with 38% of respondents reporting having an official faculty mentor. Most respondents found mentorship outside of an assigned mentor; however, those with assigned mentors reported higher satisfaction with their mentorship. Furthermore, assigned mentorship was reported to influence satisfaction with the residency program and career decision-making [44]. This was largely attributed to research opportunities, as assigned mentors were more likely involved with research and could aid the mentee in those pursuits. With the time constraints of surgical residency, having a research mentor is valuable for interested residents. An otolaryngology program in Canada implemented a formal mentorship program for 1 year and prospectively studied the influence of mentorship on resident wellness and burnout. In this program, residents between PGY-2 and 5 chose their own mentor, while PGY-1 residents were assigned a mentor based on de-identified personality surveys. The PGY-1 residents were given the option of choosing a new mentor at the end of 1 year. Mentors and mentees were encouraged to meet every 3 months. The results of the study demonstrated improved quality of life, stress, and burnout metrics for the residents. However, the study was limited by low power and a short follow-up period of only 1 year. The study did not specify whether the PGY-1 residents retained their mentors after the year [45]. Compared to otolaryngology residencies, orthopedic surgery residences are often larger, with between 5 and 7 residents per year in most programs. A 2018 survey of orthopedic surgery residents demonstrated over two-thirds of the programs had either a formal or informal mentorship program. Most of the respondents (52%) obtained their mentor on their own, and most were obtained during PGY-1 year (51%). Despite the higher percentage of mentorship in this survey, 31% of residents still reported burnout. Satisfaction with mentorship was positive at 77%. The authors believed an organic and non-required mentorship program to be the most effective form due to residents’ ability to choose a relatable mentor in line with their interests. However, they further discuss that it may be beneficial to have a formal, assigned mentorship model during the early, formative years of residency with the flexibility to pursue other mentors later in residency. This allows residents to have structured guidance early on in training while not limiting them to a mentor they may not relate as well to [46]. A 2020 survey of neurosurgery residency programs described a robust mentorship culture in neurosurgery with 65% of survey respondents reporting formal mentorship programs. For most programs, a mentor was assigned to a mentee based on their career or research interests. Their study described better ACGME outcomes for residents in mentorship programs that had been established for more than 5 years compared to programs that were less than 5 years old. These outcomes included ACGME survey results, oral and written board exam pass rates, faculty interest, and publication output [47•]. In ophthalmology, a study from Canada described that most residents reported not being aware of a formal mentorship program at their institution, and 52% of ophthalmology residents lacked a mentor [48]. Overall, although the data is of low quality and volume, surgical specialties have had similar experiences with mentorship thus far. While few programs have a formal program, those that do have yielded positive results for their residents. The difficulty among all specialties is the time required for implementation. Other specialties have also echoed the belief that smaller residency programs may have less need for formal mentorship, and informal mentorship will suffice, due to the already more intimate relationship between residents and attendings. Failure of Mentorship Program Implementation The most common objectives for mentorship programs include professional or career development, academic success, networking, faculty retention, and increased diversity/inclusion [3]. Program failure, then, can be characterized as failure to meet one or more of the program objectives. Programs fail for a variety of reasons. This failure can be at either the program level or the individual level. Examples at the program level that can lead to failure include misaligned program focus (i.e., focus on launch or matching pairs instead of content and structure of program), lack of institutional support, and inadequate mentor/mentee time availability [49, 50]. A 2022 commentary about mentorship in urology suggested that possible causes of mentorship program failure include lack of clear communication about expectations within mentorship programs and formal commitment at the institutional level [50]. For example, lack of protected time for mentoring could result in a perception that the institution does not see mentorship as an important academic activity. Recent literature supports the notion that strong institutional support for mentorship at all levels is key. A commentary published in 2019 recommends the use of incentives such as research grants, awards, and inclusion of mentorship involvement during consideration for promotion to establish a broad culture in support of mentorship [30•]. The actual cost of implementing a mentorship program for an institution is worthy of further study. Programs can also fail if the mentor–mentee relationship fails. Ultimately, mentorship is about the relationship between two or more people, and this relationship is subject to the same positive and negative influences as any human relationship. The relationship is strengthened by shared values, clear expectations and goals, and setting boundaries [51]. Prior to entering a mentorship relationship, both the mentor and mentee must do a self-evaluation. A mentor may ask themselves: “Do I want to do this? Do I have the skills to do this? Do I have the time to do this?” If the answer to any of these is, “no”, then it is not the right time to take on a mentoring role. Mentorship to Increase Diversity The field of urology remains heavily underrepresented by female, Black, and Latinx physicians. While representation is improving for some groups, significant barriers to advancement for women and underrepresented racial/ethnic minorities continue to exist. Mentorship has been identified as a key contributor to improving the representation of these groups in urology and other areas of medicine. While women have accounted for 50% of medical students since 2003, they represent only 22% of full-time professors, 16% of department chairs, and 17% of medical school deans. This demonstrates a lack of representation in leadership roles in academic medicine. Women are also more likely to leave academic medicine and not be promoted to leadership positions. Mentorship has been proposed as being critical in providing the support needed to overcome this gender bias. Despite the known benefits of mentorship, women have been shown to be statistically less likely to have a mentor compared to male colleagues [52]. A 2019 systematic review by Farkas et al. aimed to identify and describe current mentorship programs across medical residencies that were specific for women. Their results showed an overall positive impression of mentorship programs, as they were all highly rated by participants. Interestingly, although subsets of participants valued gender concordance, there was no overall difference in mentorship satisfaction among gender-discordant and concordant pairs. This indicates that female mentorship can still be implemented if there is a lack of female senior faculty [8•]. Underrepresented in medicine (URiM) racial and ethnic groups may also benefit from mentorship programs. Numerous studies have emphasized the importance of expanding diversity in medicine and the potential benefits for patient care. Despite the known benefits of mentorship for URiM groups, studies have shown they are less likely to have a mentor both as a trainee and as faculty [53]. A systematic review in 2021 of mentorship programs for URiM demonstrated positive results for all types of mentorship models, although the dyad model was the most common. They identified several themes across their review including the importance of institutional support, using resources effectively, and utilization of both non-URiM and URiM faculty mentors. They found that racial/ethnic concordance between mentor and mentee did not impact satisfaction with mentorship [54]. For urology specifically, Black physicians only make up 2% of all urologists. Several programs have emerged in recent years to provide support for Black and other URiM groups including establishing mentorship programs. The nonprofit group urology unbound developed the R. Frank Jones Urology Interest group aimed at providing a pipeline for URiM groups in urology. The program provides mentorship, research opportunities, and professional development. In 2021, 31 of the 39 group members applying for urology residency matched [55]. The previously mentioned UReTER program was a successful pilot project for Black, LatinX, and indigenous students interested in urology. Other mentorship programs aimed specifically for URiM groups have demonstrated early success from the University of California Los Angeles (UCLA) and the University of Michigan [56, 57]. As women and URiM groups are still grossly under-represented in medicine, especially in Urology, mentorship represents a crucial element in the goal of expanding diversity. Despite a lower percentage of these groups as established urology faculty, it is evident that mentorship can be successful regardless of gender or race/ethnicity. These early studies emphasize the heightened value of mentorship programs for women and URiM groups. Mentorship and Burnout Access to structured mentorship programs may mitigate the effects of physician burnout. Physician burnout appears to be a multifactorial issue that can start as early as medical school. In a national study of urology trainees, 17.6% experienced depression, and 11% endorsed suicidal ideation. Self-reported burnout was predictive of suicidal ideation (OR 7.6 [95% CI 2.5–23]), with access to mental health services being protective (p = 0.016) [58]. The implications of burnout are well understood by Urology residency program directors, considering 87% of 72 program directors surveyed agree that residents should be screened periodically for burnout [59•]. In a national survey of 211 urology residents (of 1011 contacted), access to structured mentorship programs was shown to be associated with decreased burnout (60% versus 75%, p = 0.02) [60••]. Lack of access to mental health services was also a key factor in predicting burnout (OR 5.4, p < 0.001). Residents who did not have a structured mentorship program had both decreased access to mental health services and were more likely to be working more than 80 h per week [60••, 61•]. Collectively, these studies highlight the importance of mental health resources and formal mentorship in mitigating the ill effects of burnout in light of the high prevalence of burnout in urology. The fact that access to mental health resources is correlated with the existence of a formal mentorship program suggests a possible global trainee support issue underlying both. A formal mentorship program could potentially be used to assess for burnout and identify at-risk trainees. Mentorship Program Design While the benefits of mentorship are well established, the creation of a formal mentorship program should not be undertaken without serious planning and consideration. Mentorship programs are not unique to medicine, and when thinking about their design, it is helpful to borrow from business [49]. While there are an unlimited number of ways to go about this process, it is useful to walk through the general process of mentorship program design using the AXLES model (Align, eXperience, Launch, Effectiveness, Support), a stepwise process proposed in “Mentoring Programs That Work” by Jenn Labin (Table 1).Table 1 AXELS Model for stepwise mentorship program creation [28]. The Align and Experience steps together can form written a ‘Program Charter’ Align to a purpose Key aspects/decisions: • Mentoring purpose statement • Program objectives • Stakeholders • Participants (learners and mentors) • Benefits to mentors and mentees • Success measures Taking in to account talent needs and institutional culture a “Purpose Statement” can be written Design the eXperience Five key design decisions must be made. These are: 1. Structure 2. Schedule 3. Matching 4. Learner Participation 5. Mentor Participation For learner and mentor participation this includes how they will enter and exit the program as well as expectations for participation Launch • Program launch: virtual vs. in person, big event vs. soft launch, guest speakers, panel discussion • Welcome guide for participants Evaluate Effectiveness • Four levels (New World Kirkpatrick Model) • Frequent and Consistent Check-ins • Evaluation Plan Support Mentor preparation • Explain expectations • Provide practice • Review resources, mentor skillset/toolbox • Communicate Learner Resources and Support Participant Community The first step is to identify a talent need or gap that the mentorship program will address. Possible talent needs include attrition of top talent, ineffective recruiting and onboarding, long time to succeed in critical roles, lack of leadership in the talent pipeline, loss of institutional knowledge, disengaged employees, diversity and inclusion gaps, job performance and skills gaps, and limited internal networks. While there may be many needs, it is important to identify 2–3 that are most critical and if not addressed will result in disaster. Next, the “experience” of the mentorship program can be designed. This includes the structure, schedule, matching, and entry/exit of participants from the program. Each one of these can be tailored to the needs of an individual institution. Will the program run annually with new participants? Or will participants enter and then be paired with a mentor indefinitely? Do they switch mentors every 3 months? Should the number of meetings be assigned or be left up to the mentor/mentee pairs? The experience is the key area where changes can be made based on feedback and evaluation of the program. Planned alignment to institutional goals and design of experience should happen before the program is officially launched. For programs that will run on a cyclic basis, it can be helpful to have a “launch event” where mentors and mentees gain some of the skills or knowledge necessary to be successful in the program, and key program information is disseminated. For virtual mentorship programs, the program launch may be done asynchronously or via a virtual platform. After a mentorship program is launched, it must be evaluated for effectiveness. There are many ways to evaluate a program; however, fundamentally, it is important to collect information from both mentors and mentees. Feedback from this program assessment can then be used to improve the program experience for mentors and mentees. One method of evaluation, the New World Kirkpatrick Model, is discussed below [62]. Lastly, the mentors and mentees in a program require ongoing support. This can include resources to help develop mentor skill, frequent communication from program leadership, and a welcome guide that is easily accessible. By supporting the participants in the mentorship program, it helps to build community, long-term buy-in from mentors, and ultimately program success. This support and program management can create significant administrative needs, which is why it is important to consider the ongoing support needs during the program design. Mentorship programs that are launched without the infrastructure to be evaluated, iteratively adapt, and have long-term support are destined to fail. As discussed earlier, within the medical education, literature oftentimes the only measure of success for a mentorship program is participant satisfaction. The New World Kirkpatrick Model [62] is one method of evaluation that can help expand and improve mentorship program evaluation. This model includes 4 levels of achievement, and the evaluation will incorporate information from all 4 levels (IMAGE). The model starts with the desired outcome, level 4, and then works backward to behaviors (level 3), learning (level 2), and reactions (level 1) that should happen to achieve the level 4 outcome. For example, consider a mentorship program that is designed to increase the diversity of the urology residency applicant pipeline. The level 4 outcome could be measured by calculating the number of underrepresented minorities applying to urology after implementing the program. Level 3 involves the demonstration of “behaviors” that will lead to level 4 results. In the example, this could be a demonstration of skills that would make a student successful in a urology clerkship, production of published research papers, or leadership in a urology interest group. Level 2 is “learning” gained from the mentorship program that will lead to the behaviors above. In the example, this could be demonstrating knowledge about the field of urology, relevant anatomy and pathology, and the application process. Lastly, level 1 is “reaction” or satisfaction with the program. Many published mentorship programs only collect level 1 data. It is important to define these strategic results that will translate to the success of a program. In the example, a successful program would generate a positive reaction to the mentorship program leading to learning the appropriate information about urology. This learning would then translate success in important parts of the urology residency application process (i.e., clerkships, research, commitment to the field). These behaviors then translate to successful application and match into urology residency. The lower-level results can be measured to evaluate if the program is progressing toward the original goal and then changes made in an iterative process to improve the program. Conclusions Formal mentorship programs can and do work in urology and other surgical specialties; however, more robust and rigorous data collection is needed. The benefits of mentorship for professional development, well-being, and productivity are well described, and the impact of mentorship on increased diversity and equity within medicine cannot be ignored. An initial investment of resources and planning is required when creating a formal mentorship program, and this investment should not be treated casually. The creation of a mentorship program is an iterative process, and programs can progress based on concrete outcomes and measurement of validated metrics. Compliance with Ethical Standards Conflict of Interest The authors declare no competing interests. Human and Animal Rights and Informed Consent This article does not contain any studies with human or animal subjects performed by any of the authors. This article is part of the Topical Collection on Education Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Papers of particular interest, published recently, have been highlighted as: • Of importance •• Of major importance 1. Khan NR Rialon KL Buretta KJ Deslauriers JR Harwood JL Jardine DA Residents as mentors: the development of resident mentorship milestones J Grad Med Educ 2017 9 551 554 10.4300/JGME-D-17-00415.1 28824785 2. Will You Be My Mentor?—Four archetypes to help mentees succeed in academic medicine \| Child Development \| JAMA Internal Medicine \| JAMA Network [Internet]. [cited 2022 Sep 13]. Available from: https://jamanetwork.com/journals/jamainternalmedicine/article-abstract/2664070. 3. Kashiwagi DT Varkey P Cook DA Mentoring programs for physicians in academic medicine: a systematic review Acad Med 2013 88 1029 1037 10.1097/ACM.0b013e318294f368 23702518 4. Akinla O Hagan P Atiomo W A systematic review of the literature describing the outcomes of near-peer mentoring programs for first year medical students BMC Med Educ 2018 18 98 10.1186/s12909-018-1195-1 29739376 5. •• Chua WJ, Cheong CWS, Lee FQH, Koh EYH, Toh YP, Mason S, et al. Structuring mentoring in medicine and surgery. A systematic scoping review of mentoring programs between 2000 and 2019. J Contin Educ Health Prof. 2020;40:158–68. A systematic review of mentorship programs in medicine and surgery programs in order to design a consistent structured approach for implementation of formal curricula. 6. Entezami P Franzblau LE Chung KC Mentorship in surgical training: a systematic review HAND 2012 7 30 36 10.1007/s11552-011-9379-8 23448749 7. Farkas AH Allenbaugh J Bonifacino E Turner R Corbelli JA Mentorship of US medical students: a systematic review J Gen Intern Med 2019 34 2602 2609 10.1007/s11606-019-05256-4 31485967 8. Farkas AH Bonifacino E Turner R Tilstra SA Corbelli JA Mentorship of women in academic medicine: a systematic review J Gen Intern Med 2019 34 1322 1329 10.1007/s11606-019-04955-2 31037545 9. Pethrick H Nowell L Paolucci EO Lorenzetti L Jacobsen M Clancy T Peer mentoring in medical residency education: a systematic review Canadian medical education journal 2020 11 e128 e137 33349761 10. Skjevik EP Boudreau JD Ringberg U Schei E Stenfors T Kvernenes M Group mentorship for undergraduate medical students—a systematic review Perspectives on Medical Education 2020 9 272 280 10.1007/s40037-020-00610-3 32820416 11. Benson CA Morahan PS Sachdeva AK Richman RC Effective faculty preceptoring and mentoring during reorganization of an academic medical center Med Teach 2002 24 550 557 10.1080/0142159021000002612 12450479 12. Wingard DL Garman KA Reznik V Facilitating faculty success: outcomes and cost benefit of the UCSD National Center of Leadership in Academic Medicine Acad Med 2004 79 S9 11 10.1097/00001888-200410001-00003 15383376 13. Tracy EE Jagsi R Starr R Tarbell NJ Outcomes of a pilot faculty mentoring program Am J Obstet Gynecol 2004 191 1846 1850 10.1016/j.ajog.2004.08.002 15592263 14. Fleming M House S Hanson VS Yu L Garbutt J McGee R The mentoring competency assessment: validation of a new instrument to evaluate skills of research mentors Acad Med 2013 88 1002 1008 10.1097/ACM.0b013e318295e298 23702534 15. Files JA Blair JE Mayer AP Ko MG Facilitated peer mentorship: a pilot program for academic advancement of female medical faculty J Womens Health (Larchmt) 2008 17 1009 1015 10.1089/jwh.2007.0647 18681821 16. Varkey P Jatoi A Williams A Mayer A Ko M Files J The positive impact of a facilitated peer mentoring program on academic skills of women faculty BMC Med Educ 2012 12 14 10.1186/1472-6920-12-14 22439908 17. Kosoko-Lasaki O Sonnino RE Voytko ML Mentoring for women and underrepresented minority faculty and students: experience at two institutions of higher education J Natl Med Assoc 2006 98 1449 1459 17019912 18. Pololi L Knight S Mentoring faculty in academic medicine J Gen Intern Med 2005 20 866 870 10.1111/j.1525-1497.2005.05007.x 16117759 19. Eisen S Sukhani S Brightwell A Stoneham S Long A Peer mentoring: evaluation of a novel programme in paediatrics Arch Dis Child 2014 99 142 146 10.1136/archdischild-2013-304277 24152570 20. Thorndyke LE Gusic ME Milner RJ Functional mentoring: A practical approach with multilevel outcomes J Contin Educ Heal Prof 2008 28 157 164 10.1002/chp.178 21. Alleyne SD Horner MS Walter G Hall Fleisher S Arzubi E Martin A Mentors’ perspectives on group mentorship: a descriptive study of two programs in child and adolescent psychiatry Acad Psychiatry 2009 33 377 382 10.1176/appi.ap.33.5.377 19828850 22. Andre C Deerin J Leykum L Students helping students: vertical peer mentoring to enhance the medical school experience BMC Res Notes 2017 10 176 10.1186/s13104-017-2498-8 28464902 23. Duke P Grosseman S Novack DH Rosenzweig S Preserving third year medical students’ empathy and enhancing self-reflection using small group “virtual hangout” technology Med Teach 2015 37 566 571 10.3109/0142159X.2014.956057 25189277 24. Day KM Schwartz TM Rao V Khokhar MT Miner TJ Harrington DT Medical student clerkship performance and career selection after a junior medical student surgical mentorship program Am J Surg 2016 211 431 436 10.1016/j.amjsurg.2015.10.007 26709047 25. Sobbing J Duong J Dong F Grainger D Residents as medical student mentors during an obstetrics and gynecology clerkship J Grad Med Educ 2015 7 412 416 10.4300/JGME-D-14-00667.1 26457148 26. Zuzuárregui JRP Hohler AD Comprehensive opportunities for research and teaching experience (CORTEX): a mentorship program Neurology 2015 84 2372 2376 10.1212/WNL.0000000000001663 25957332 27. Indyk D Deen D Fornari A Santos MT Lu W-H Rucker L The influence of longitudinal mentoring on medical student selection of primary care residencies BMC Med Educ 2011 11 27 10.1186/1472-6920-11-27 21635770 28. Coates WC Crooks K Slavin SJ Guiton G Wilkerson L Medical school curricular reform: fourth-year colleges improve access to career mentoring and overall satisfaction Acad Med 2008 83 754 760 10.1097/ACM.0b013e31817eb7dc 18667890 29. Cohee BM Koplin SA Shimeall WT Quast TM Hartzell JD Results of a formal mentorship program for internal medicine residents: can we facilitate genuine mentorship? J Grad Med Educ 2015 7 105 108 10.4300/JGME-D-14-00315.1 26217434 30. Choi AMK Moon JE Steinecke A Prescott JE Developing a culture of mentorship to strengthen academic medical centers Acad Med 2019 94 630 633 10.1097/ACM.0000000000002498 31026234 31. Bortnick E Stock J Simma-Chiang V Mentorship in urology residency programs in the United States Urology 2020 136 58 62 10.1016/j.urology.2019.09.053 31738942 32. Ernst M Ebert K Rehfuss A McLeod D Alpert S A survey of pediatric urology fellowship applicants: past priorities to guide future possibilities J Pediatr Urol 2022 S1477–5131 22 00280 287 33. Wang LC Mittal AG Puttmann K Janzen N Palmer LS Yerkes EB The changing gender landscape of pediatric urology fellowship: results from a survey of fellows and recent graduates Journal of Pediatric Urology Elsevier 2019 15 51 57 10.1016/j.jpurol.2018.09.001 34. Freilich DA Nguyen HT Phillips JL Factors influencing residents’ pursuit of urology fellowships Urology 2011 78 986 992 10.1016/j.urology.2011.05.068 22054361 35. Kutikov A Bonslaver J Casey JT Degrado J Dusseault BN Fox JA The gatekeeper disparity–why do some medical schools send more medical students into urology? J Urol 2011 185 647 652 10.1016/j.juro.2010.09.113 21168862 36. Zheng MY, Overland M, Escobar D, Fakunle M, Li Y, Chu C, et al. Formal mentorship as an opportunity to expand the urology pipeline: Under Represented Trainees Entering Residency (UReTER) Program Evaluation 2020–2021. Urology [Internet]. 2021; Available from: https://www.sciencedirect.com/science/article/pii/S0090429521008323. 37. Ernst M, Badkhshan S. #UroStream101: Social Media as a Medium for Mentorship in Urology. Urology. 2021;158:39–44. 38. Anaissie J Popat S Mayer WA Taylor JM Innovative approaches to battling resident burnout in a urology residency program Urology Practice WoltersKluwer 2021 8 387 392 10.1097/UPJ.0000000000000215 39. AUA Institute for Leadership & Business [Internet]. [cited 2022 Jul 24]. Available from: https://www.auanet.org/Institute-for-Leadership-Business/. 40. Kibbe MR Pellegrini CA Townsend CM Helenowski IB Patti MG Characterization of mentorship programs in departments of surgery in the United States JAMA Surg 2016 151 900 906 10.1001/jamasurg.2016.1670 27383863 41. Chan K-M Hsu J-T Yeh C-N Yeh T-S Lee W-C Lien H-Y Application of mentorship program for another aspect of surgical residency training Medicine (Baltimore) 2021 100 e26939 10.1097/MD.0000000000026939 34397945 42. Ullrich LA Jordan RM Bannon J Stella J Oxenberg J The mentor match: a new approach to implementing formal mentorship in general surgery residency Am J Surg 2020 220 589 592 10.1016/j.amjsurg.2020.01.011 31996294 43. Gurgel RK Schiff BA Flint JH Miller RA Zahtz GD Smith RV Mentoring in otolaryngology training programs Otolaryngol Head Neck Surg 2010 142 487 492 10.1016/j.otohns.2009.12.002 20304265 44. Hsu AK Tabaee A Persky MS Mentorship in otolaryngology residency: the resident perspective Laryngoscope 2010 120 1263 1268 10.1002/lary.20858 20513049 45. Zhang H Isaac A Wright ED Alrajhi Y Seikaly H Formal mentorship in a surgical residency training program: a prospective interventional study J Otolaryngol Head Neck Surg 2017 46 13 10.1186/s40463-017-0186-2 28193248 46. Oladeji LO Ponce BA Worley JR Keeney JA Mentorship in orthopedics: a national survey of orthopedic surgery residents J Surg Educ 2018 75 1606 1614 10.1016/j.jsurg.2018.04.007 29685787 47. Khan NR Derstine PL Gienapp AJ Klimo P Barbaro NM A Survey of neurological surgery residency program mentorship practices compared to accreditation council for graduate medical education resident outcome data Neurosurgery 2020 87 E566 E572 10.1093/neuros/nyz479 31748813 48. Nassrallah G Arora S Kulkarni S Hutnik CML Perspective on a formal mentorship program in ophthalmology residency Can J Ophthalmol 2017 52 321 322 10.1016/j.jcjo.2017.03.005 28774510 49. Labin J. Mentoring Programs That Work. Alexandria, VA: Association for talent development; 2017. 50. Patel HV, Shaw NM, Breyer BN. Building the pipeline — mentoring success in urology. Nature Reviews Urology [Internet]. 2022; Available from: 10.1038/s41585-021-00559-w. 51. Maloney ME Ethical mentorship: the dilemma of success or failure Clin Dermatol 2012 30 210 215 10.1016/j.clindermatol.2011.06.009 22330665 52. The state of women in academic medicine [Internet]. AAMC. [cited 2022 Jul 19]. Available from: https://www.aamc.org/data-reports/faculty-institutions/report/state-women-academic-medicine. 53. Nivet MA, Taylor VS, Butts GC, Strelnick AH, Herbert-Carter J, Fry-Johnson YW, et al. Diversity in academic medicine no. 1 case for minority faculty development today. Mt Sinai J Med. 2008;75:491–8. 54. Bonifacino E Ufomata EO Farkas AH Turner R Corbelli JA Mentorship of underrepresented physicians and trainees in academic medicine: a systematic review J Gen Intern Med 2021 36 1023 1034 10.1007/s11606-020-06478-7 33532959 55. Wallace NO Pittman AB Wilson SN The R Frank Jones urology interest group: an intentional and strategic pipeline program to increase diversity in urology Urology 2022 162 27 32 10.1016/j.urology.2021.08.047 34666122 56. Underrepresented Resident Mentorship Program - UCLA Urology, Los Angeles, California (CA) [Internet]. [cited 2022 Jul 19]. Available from: https://www.uclahealth.org/urology/underrepresented-resident-mentorship-program. 57. Michigan Urology Academy (MUA) \| Urology \| Michigan Medicine [Internet]. Urology. 2020 [cited 2022 Jul 19]. Available from: https://medicine.umich.edu/dept/urology/education/medical-students/michigan-urology-academy-mua. 58. Marchalik D Goldman C Alger J Rodriguez A Catomeris A Lynch JH The impact of gender and institutional factors on depression and suicidality in urology residents Can J Urol 2020 27 10471 10479 33325351 59. Glick H Ganesh Kumar N Olinger TA Vercler CJ Kraft KH Resident mental health and burnout: current practices and perspectives of urology program directors Urology 2022 160 40 45 10.1016/j.urology.2021.09.043 34863792 60. Marchalik D Brems J Rodriguez A Lynch JH Padmore J Stamatakis L The impact of institutional factors on physician burnout: a national study of urology trainees Urology 2019 131 27 35 10.1016/j.urology.2019.04.042 31199967 61. Nauheim J North AC An updated review on physician burnout in urology Urol Clin North Am 2021 48 173 178 10.1016/j.ucl.2021.01.003 33795050 62. Introduction-to-the-Kirkpatrick-New-World-Model.pdf [Internet]. [cited 2022 Jul 22]. Available from: https://www.kirkpatrickpartners.com/wp-content/uploads/2021/11/Introduction-to-the-Kirkpatrick-New-World-Model.pdf.	36459377	PMC9716155	NO-CC CODE	2022-12-03 23:20:54	no	Curr Urol Rep. 2022 Dec 2;:1-10	utf-8	Curr Urol Rep	2,022	10.1007/s11934-022-01122-9	oa_other
==== Front J Chin Polit Sci J Chin Polit Sci Journal of Chinese Political Science 1080-6954 1874-6357 Springer Netherlands Dordrecht 9840 10.1007/s11366-022-09840-0 Review Essay Sino-US Competition: Is Liberal Democracy an Asset or Liability? http://orcid.org/0000-0002-1331-7221 Xia Ming [email protected] Professor Ming Xia Ming Xia is a Professor of Political Science at the Department of Political Science and Global Affairs, the College of Staten Island, the City University of New York and a doctoral faculty member at the CUNY Graduate Center. He received his degrees from Fudan University and Temple University. He once taught at Fudan University and served as a residential fellow at the Sigur Center for Asian Studies at the George Washington University(2003), the Woodrow Wilson International Center for Scholars (2004), the East Asian Institute at the National University of Singapore (2004 and 2011) and the Asian Research Institute at NUS (2012). He is the author of The Dual Developmental State (2000), The People’s Congresses and Governance in China (2008), Political Venus (2012, in Chinese) and Empire of the Red Sun (2015 in Chinese). He is a co-editor of The Crown of Thorn: Liu Xiaobo and the Nobel Peace Prize (in Chinese 2010) and the editor of Chen Ping’s The Age of Plunder: The 2008 Economic Crisis as a Turning Point in Chinese History and World Civilization (2016). He is a co-producer of an HBO Oscar-nominated documentary movie, "China’s Unnatural Disaster" (2009) and the historical advisor/translator for the documentary movie Dream against the World: Mu Xin (2015). He was included consecutively to the "Top 100 Chinese Public Intellectuals” from 2009 to 2013, then in 2015 and 2017. Most Recent Publications: “Triangulating Human Political Conditions and Reorienting Political Development in China,” Journal of Chinese Governance, Vol. 1, Issue 3, 2016, pp. 405–426. “Movement and Migration” in Handbook on Human Rights in China, ed. by Sarah Biddulph and Joshua Rosenzweig, Cheltenham, UK and Northampton, MA: Edward Elgar Publishing, 2019. “A China Scholar’s Rendezvous with Islam,” in Choosing Asian-America: A New York Reader, edited by Russell Leung, 2017 September CUNU Forum, Asian American and Asian Research Institute, the City University of New York. Book Editor for CHEN Ping, The Age of Plunder: The 2008 Economic Crisis as a Turning Point in Chinese History and World Civilization (Hong Kong: iSun Affairs Limited, 2016, pp. 287). “Media Control as Stability Maintenance: The Case of Sichuan Earthquake” in Media at Work in India and China: Discovering and Dissecting (edited by Robin Jeffrey and Ronojoy Sen), New Delhi, India: Sage, 2015, pp. 245–270. “Communist Oligarchy and Oligarchic Transition in China,” in Guoguang Wu and Helen Lansdowne, ed., China’s Transition from Communism: New Perspectives (New York: Routledge, 2015), pp. 34–55, 2015. Book review: The Politics of Controlling Organized Crime in Greater China by Sonny Shiu-hing Luo, The China Quarterly, Volume 226 / June 2016, pp 571–573. Review Essay:” See No Evil: A Tailored Reality?” A Review of Selling Sex Overseas by Ko-lin Chin and James Finchenauer, CUNY Forum: Asian American /Asian Studies, Vol. 1:3, Fall & Winter 2015–2016, August 2015, pp. 101–106. Book Review: Chinese Politics and Government: Power, ideology, and organization by Sujian Guo (New York and London: Routledge, 2013), Jan. 2016, Journal of Chinese Political Science. Personal Homepage: www.dr-ming-xia.org . Office Phone: 1-718-982-3197. Email: [email protected]. grid.254498.6 0000 0001 2198 5185 Department of Political Science and Global Affairs, College of Staten Island, The City University of New York, New York, NY USA 2 12 2022 113 7 6 2022 15 11 2022 18 11 2022 © Journal of Chinese Political Science/Association of Chinese Political Studies 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This review essay covers five recent books on US-China relations, in particular addressing the rising challenge from China to the United States. These books examine US-China rivalry and advocate for changes, more or less, in US foreign policy. The essay offers a new synthesis by referring to lessons in US history and theoretical inspirations on flexible network. It evaluates the importance of liberal democracy for the United States to formulate its strategy and policy in response to China’s rising authoritarianism. Keywords US-China relations Liberal democracy Strategy China Authoritarian advantage ==== Body pmcHeated Debate on US-China Rivalry For the past three years, the American debate on China has become heated. For an updated understanding of this ongoing debate, this review essay selects the following five books: The Longer Game: China’s Grand Strategy to Displace American Order by Rush Doshi (2021); Stronger: Adapting America’s China Strategy in an Age of Competitive Interdependence by Ryan Haas (2021); The World According to China by Elizabeth C. Economy (2022); An Open World: How America Can Win the Contest for Twenty-First Century Order by Rebecca Lissner and Mira Rapp-Hooper (2020); and The Return of Great Power Rivalry: Democracy Versus Autocracy from the Ancient World to the US and China by Matthew Kroenig (2020). Although the selected titles do not all focus on the same research question with equal emphasis, this essay will examine the authors’ arguments about the value of liberal democracy in the current US-China rivalry. To summarize succinctly, these authors differ in how they view the “authoritarian advantage” or the “democratic advantage” in the Sino-American rivalry, and to what degree they are willing to compromise democratic values in the face of China’s threat. These books were selected because these authors represent a convergence toward China policy among Democrats and moderate Republicans and many consult with the President. For example, Doshi is the Director for China on the National Security Council (NSC); Lissner is the Director for Strategic Planning at the NSC; Rapp-Hooper is a senior advisor on China at the State Department; and Economy is the advisor in the Commerce department under the current Biden administration. Additionally, Haas held the position of the Director for China under the Obama administration, and in slight contrast, Kroenig is associated with Republican presidential candidates and the Scowcroft Center. By examining these five books, we may assess the shared understanding of US China strategy as one of the few policy areas with bipartisan consensus in Washington. This review essay will use the theme of the “value of democracy” to critically cross-examine the authors’ central arguments, hidden assumptions, and possible inferences in order to identify potential misdiagnosis in American strategic thinking. Although these problems might be minute at the first glance, the devil is in the details. They warrant our careful and critical reflection in order to detect misunderstandings before they are encapsulated in policy and lead to misguided decision. China as a Challenger to the American Democracy Model In 1991, at a low point for China’s political development after the 1989 Tiananmen Massacre, my former student at Fudan University presented me a book written by Wang Huning, entitled America against America (1991). Since my former student was working in the propaganda department under the university Party committee, obviously the free promotional copy was part of the ongoing ideological campaign against the “Bourgeois Liberalization.” Wang then served three Chinese top leaders for three decades so his ideas are popular with Chinese leadership. Inspired by the Hollywood blockbuster movie, Empire of the Sun (1987), Wang predicted China’s ultimate showdown with the US and its outcome: with its collectivism, authoritarianism and self-sacrifice (or altruism), China would ultimately win over the US founded on individualism, democracy and hedonism [26:389–390]. Wang could be credited as the earliest messenger for the couplet of “China is rising and US is declining,” which evolved into Xi Jinping’s mantra of today: “the rise of the East and the decline of the West.” Since taking power in 2012-13, Xi, the most powerful and self-assured leader since Mao Zedong under the PRC rule, has thrown down the gauntlet in front of the US and its democratic allies. Betting on the largest population, the second largest economy (China’s economy is already 25% larger than the US economy based upon the relative price of goods, according to Doshi [2:6]), and the third (or fourth, depending upon which method you choose, Chinese or American, inclusion or exclusion of Taiwan) largest territory, Xi has created a clear and present challenge to the US. This aggression was unequivocally epitomized in his de facto alliance with Putin through a joint declaration of the two Eurasian autocracies at the Beijing winter Olympics in February 2022 [14]. Since 2016 in the US, we have observed a similar paradigm shift from engagement to decoupling/deterrence/containment. Out of the five books analyzed, Economy offers the most comprehensive assessment about China’s ambition and challenge to the US. Economy’s book title, The World according to China, is dubious, if we consider that Xi Jinping is not equivalent to China. From the Chinese perspective, this is not a negligible difference and neither is the separation of the CCP from the Chinese. In fact, Haas disagreed with this approach to “clumsily attempting to drive a wedge between the Communist Party leadership and the Chinese people” [9:95]). Economy details some significant achievements China has made both at home and abroad. According to her, “China’s economy will soon surpass that of the United States” [6:5]. China has capabilities and developed strategies to use its hard, soft, and sharp powers against the US in an existential competition. She describes how China has tried to export its authoritarian model, if not communism, to the world, to rewrite the rules of the game, and to “reorder the world order” for gaining the global centrality for itself on the world stage. Xi has clearly identified the US as the “biggest threat” to China’s development and security [6:27–28]. However, after carefully assessing the Belt and Road Initiative (BRI), technology (e.g., “Made in China 2025”), diplomacy, alliance, and the domestic strains at home, Economy concludes that “China does not appear prepared to supplant the United States as the world’s sole superpower” [6:27]. Economy intends to warn the complacent American public in a similar way to Thomas Friedman, Michael Mandelbaum and Peter Navarro [7; 21]. However, an alarmist style could easily fall into the trap of sensationalism in Chinese popular writings, which, besides commercial interest in an overcharged nationalistic environment, could be viewed as an often-used stratagem in Chinese statecraft: the empty fortress [19, 20]. Between Scylla and Charybdis, she has to phrase a strong message without being misperceived as defeatist and formulate a confident policy without inflating American-centric arrogance. In her evaluation of BRI policy, Economy focuses on the “shelling out RMB” diplomacy (in Chinese it has a vulgar and unfortunate inference) encapsulated by the BRI, covering more than 140 countries in four continents and expanding to Digital Silk Road, Polar Silk Road and space roads [2:242]. If we apply cost-effective calculus, we see a “vanity project” [2:242] or a reckless money-losing political project as Kroenig also finds [14:190]). This egregious mismatch between China’s strategic goals and resources is especially risky for a country to indulge in an imperial overstretch before becoming a global empire [13]. Furthermore, Economy identifies China’s weaknesses from production chains, in which the Chinese companies, such as Huawei, are forced to pay a dear price or risk being shut out of the global market due to US prioritizing geostrategic interests over commercial ones. Although she does not encourage a complete decoupling between the US and China, Economy anticipates some amount of political, ideological, economic and technological decoupling. For others, like Haas, decoupling is unthinkable. China as a Frenemy Haas criticizes Trump administration’s “omnidirectional confrontation” posture as a losing strategy and proposes a “more constructive approach,” not a “binary ‘good vs. evil’ approach” [9:9]. He advises the US not “to stand in China’s way” [9:6] or defensively “blunt China’s progress” [9:7], but “both sides need to maintain an ability to co-exist within a heightened state of competition” [9:7]. Differing from the Trump administration, Haas does not think of Xi’s China as an existential threat to US but rather, as a potential partner [9:16]. He wants to convince us that the US has enough enduring strengths to be confident (a “confident strategy”), but should concede to China’s primacy in East Asia, abide by non-intervention to China’s domestic political system, and finally give up Pax Americana. He proposes “competitive interdependence” for the US-China relationship and advises US to look inward “to concentrate more on nurturing its own progress”: “America cannot choose China’s path; America can only control its own.” [9:7] Haas warns against the “ten-foot-tall syndrome” that exaggerates China’s strength; however, he is ready to concede “China’s rise as a peer competitor poses the most direct test of American policy in decades” [9:3–5]: “The United States needs to learn to coexist with a powerful China. At the same time, the United States needs to help China realize the boundaries of where it can pursue a greater role for itself without coming into conflict with vital American interests and values, and where it cannot.” [9:65] This advice to not challenge China’s domestic political system [9:86] is to some extent ignoring the heterogenous and incompatible governance and values between the two countries as well as the militaristic nature of the totalitarian system. Politically, it also provides ample ground for the Trump supporters to critique the “engagement policy,” “responsible stake holder” and G2 Condominium, important ideas under various administrations before Trump. There is perhaps wishful thinking in the book as seen with these phrases: “symbiotic relationship”, “virtuous cycle”, “new equilibrium for the relationship” and “affirmative agenda with Beijing”, etc. Understandably, for some critics, this is simply the continuation of an appeasement policy. The deep contradiction of Haas’ proposal lies in that on the one hand, he intends to defend American credibility abroad and build an American “united front” to resist China’s aggressive demands; on the other hand, his proposed “balance of power” policy would make some countries a pawn on the American strategic chessboard and sow doubts in the mind of foreign leaders, especially those from small Asian countries, about US reliability when discussing giving up its military primacy in East Asia. Haas tries to be balanced by walking the middle road; however, his spatial distance from his perceived two ends does not guarantee his arrival at a better perspective, because a new synthesis often draws positive ingredients from the opposing sides and transcends and reconciles both the original thesis and its anti-thesis (See, for example, Doshi’s discussing “oppositional” and “dialectical” unity [9:177]). Instead, his superficially balanced neutrality leads to a moral equivalence: He suggests that the US and China need to understand each other’s ambition, be ready to accept each other’s “progress,” and the US has to see its own faults even as it has the urge to criticize China’s failings. Haas treats the US and China as in the exclusive G2 Club (although he denies the G2 concept) above other middle great powers (including American democratic allies such as U.K., Germany, France and Japan), which gives China a false sense of parity with the US that Xi has been calling for a “new type relationship of great powers.” The following argument from Haas certainly would not be easily accepted by many people: “They both likely will rise or fall in tandem” [9:41]. Nevertheless, the engagement policy has a strong lingering pull. [19, 20] China’s Grand Strategy Haas acknowledges China’s strategic ambitions, but limits them to three “core interests”: sovereignty, security, and development [9:54], downplaying its ideological nature [9:51–52]. He wants to “keep China inside the tent” [9:88], not making an enemy out of a competitor [9:63]. In contrast, Economy’s “bigger tent” [6:219], which is bigger than American traditional allies, has expanded to include “a broader range of partners and potential allies”, looking for support and allies in the developing countries. Obviously, this bigger tent does not include China, who is its target. Rush Doshi goes one step further by arguing that China has viewed the US as “the greatest threat” and “main adversary” and formulated a “grand strategy to displace American order.” In his book entitled The Long Game, he sees China following a “template” or a “coherent scheme” to execute strategies of displacement in three steps: the first one (1989–2008) was “to quietly blunt American power over China,” the second (2008–2016) “sought to build the foundation for regional hegemony in Asia,” and the third (2016 to present) “expands its blunting and building efforts worldwide to displace the United States as the global leader” [2:4, italic original]. While Doshi’s highlighting of heightened antagonism from China toward the United States is more sober than Haas’s assessment, it is still plagued by a common misperception seen in Economy and Haas, namely taking Chinese commentators’ sensationalism (often verging on yellow journalism such as in The Global Times) at its face value. The perceived long game of China could ultimately be a cognitive blind spot. Doshi defines “grand strategy as a state theory of how it can achieve its strategic objectives that is intentional, coordinated, and implemented across multiple means of statecraft—military, economic, and political.” As a believer of China’s grand strategy, Doshi tries to convince the skeptics by “applying a unique social-scientific approach” and relying on “a large trove of rarely cited or previously inaccessible Chinese texts” [2:8]. While not denying the value of official documents and speeches from Chinese leaders, as a scholar who spent his first half of life in China, I cannot easily ignore the persistent gap between promise and performance in Communist official ideological propaganda. If fact, Chinese-style Gargantuan-mania (being crazy for big) is deeply rooted in Marxist ideology of historical and dialectical materialism. Doshi is aware of the ideological origin of China’s strategic thinking from Leninism [2:26–27]. However, under Chap. 2, “The Party Leads Everything,” he has argued: “Together, with leadership at the top and institutional penetration virtually all the way through to the bottom, the Party has the ability not only to coordinate and direct state behavior but, in many cases, to monitor it. This is by design” (p. 36). Tragically, the CCP’s confidence in grasping the “red lining” of historical law, teleology, determinism, and historicism often ends up with Quixotism, utopianism and fanaticism in the Communist grand social engineering (the Great Leap Forward, the Great Famine, the Family Planning, and Xi’s abrupt reversal of Reform and Opening are examples). Doshi compares today’s debate on China’s grand strategy to Crowe’s discussion on the German strategy a century ago [2:16], of which Haas is well aware [9:126–127]. Both Doshi and Haas do have a point to warn the world about the imminent danger from China, but I must point out that, as in the German case, “the politics of cultural despair” (the namesake for Fritz Stern’s book on “a study in the rise of the Germanic ideology” [24]) has also played an important role in China’s restless and unsuccessful search for identity. To be fair, Economy, Haas and Doshi have seen China’s perennial internal crises and recent setbacks on the world stage; but it is problematic for Doshi to credit some of China’s obvious failures as rationality and strategic planning. For example, Doshi believes that the “traumatic trifecta of Tiananmen Square, the Gulf War, and the Soviet collapse gave rise to China’s first displacement strategy” [2:44]; he also believes that the new traumatic trifecta of the 2008 financial crisis, Trump’s trade war and COVID-19 pandemic has worked to China’s favor. We can argue that all these six events shocked the regime to its core and caused everlasting legitimacy crisis (the ghosts of Tiananmen massacre have been haunted the regime for more than three decades). As a matter of fact, the 2008 financial crisis caused the Chinese government to pump trillions of yuan into the economy and had created bubbles in the stock market and real estate [1]. While the West has acquired more instruments to cope with the pandemic, China has been struggling with vaccinations and its ongoing lockdowns have been unscientific, chaotic, brutal and unsustainable. Thus, the argument that China has a coherent grand strategy does not have much evidence to support it. The Power of Liberal Democracies and Open Order Doshi [2:332–333] does have made effort to balance his alarmist argument by concluding his book with a discussion on the rise and fall of five rounds of American “declinism”. He (2:334) wrote: “[P]olicymakers must resist the common declinist tendency to see US competitors as ten feet tall and instead calibrate a response that spurs innovation without stoking fear and prejudice.” At this point, we will bring three generalists, Kronenig, Lissner and Rapp-Hooper into my discussion of if we can still find confidence in the American system and its ability to defend liberal democracy, both at home and on the world stage, against the techno-authoritarian challenge of China. Lissner and Rapp-Hooper in their book An Open World: How America Can Win the Contest for the Twenty-First-Century Order argue that: “The liberal international order will not survive the tectonic movements under way” [18:61]. They claim: “The drawbacks of a persistently liberal-universalist strategy would likely come in the form of overreach, as policymakers devote resources to reshaping states whose characters cannot be transformed in Washington” [18:118]. They continue, “In all, the United States and its allies can secure a favorable and accessible future order, despite the dusk of the liberal international order. Guaranteeing an open order despite opposing forces of closure, however, will require a lucid American vision alongside shrewd statecraft” [18:63]. They prescribe “the possibility for an American strategy that promotes openness—free access to the global commons, economic exchange, information flow, and security cooperation—without insisting upon the spread of liberal democratic governance as a prerequisite” [18:28]. Sadly, with a belief that “China is a power in unequivocal ascendance” [18:58] and the US acknowledges “sundown on the unipolar era” [18:119], their retreat from a liberal global order to an open system is an unnecessary concession to China by renouncing regime type as a requirement for open governance and policy of regime change [18:101]. In fact, the authors point out with an error that: “Advanced economies with significant technological market share, the United States and its allies possess more than 28 times (sic!) China’s overall GDP and exceeds its per capita wealth by a multiple of 4.5” [18:136]. The fundamental flaw in Lissner and Rapp-Hooper’s book lies in the artificial separation of openness from liberalism, openness as “a model for international, not domestic governance” [18:98]. As argued by Karl Popper, at the regime level or global governance, openness or “open society” is an essential part and parcel of liberalism [23]. If one tries to downplay liberalism and emphasize openness as a “shrewd statecraft,” it has overshadowed the holistic vision of liberalism. To phrase US-China competition/confrontation as openness vs. closure, we can easily lose sight of the expansive and dynamic nature of China’s global reach, for example, the BRI. To advocate global openness with “exceptions” (e.g., migration in Lissner and Rapp-Hooper [18:116]) and not to support domestic liberal regimes in the world can be a contradiction at best and a hypocrisy at worst. Lissner and Rapp-Hooper [18:117] explain that: “The magnitude of American power affords some space for hypocrisy”. Contrary to their realist assessment, at this historical juncture, we may actually have been witnessing the ongoing transition from a “liberal international order” to a “liberal global order”: the former applies to the US-led “free world”/ “democratic alliance,” the latter will be the lingua franca for an emerging new world. At the dawn of post-WWII liberal international order, as described by the historian Tuchman [25], China under Chiang Kai-shek failed to live up to FDR’s expectation to become a key contributor in the new world order as one of the “Four Policemen.” Now at the twilight (not “dusk” used by Lissner and Rapp-Hooper, but the moment before sunrise) of a liberal global order, China’s oligarchic leadership ignored the second invitation from four American administrations to work with the US (such as G2) on liberalization and democratization. Although China has been challenging the American-led liberal order, but it will not displace the liberal international order. In fact, my research has convinced me that Xi’s power has already started to diminish after the crash of the stock market in the summer of 2015. For example, as Doshi has described, China under Jiang Zemin started “blue sea navy” and “sea control” strategy but later followed “sea denial” strategy [2:11; also see: 6:51]. By contrast, Xi’s continental, inward-looking BRI has been a proactive strategy in response to the containment of maritime countries [Haas 9:32–33], especially the reaction and consolidation of the first island chain (in particular Japan and Taiwan with American intensifying support), rather than a trump card against the US. On this point, I agree with Kroenig [14:190], who said: “BRI is in part a sign of Chinese weakness, not strength. Boxed in from expanding in the East by America and its allies, the expansionist CCP had no choice but to look west to Central Asia.” The American global geostrategic realignment completed with the Indo-Pacific strategy has created a donut-structure in which the maritime ring of democratic countries and their allies has successfully surrounded the Eurasian authoritarian core based upon the Shanghai Cooperation Organization. This new realignment and the formulation of an antagonistic rivalry has become the important foundation for our understanding of US-China interactions. The Democratic Advantage and Democracy’s Neglected Vulnerability Kroenig shows us the evolution from “China as a global superpower,” a power transition from US to China, and a “post-American world” and debunks the authoritarian advantage in foreign policy arguing: “[D]emocracies enjoyed a built-in advantage in long-run geopolitical competitions.” [14:3] He continues to argue that: “[D]emocratic countries are better able to amass power, wealth, and influence on the world stage than their autocratic competitors. Democracy is a force multiplier that helps states punch above their weight in international geopolitics” [14:4]. In his conclusion, he [14:215] states: “Democracy is the best machine ever invented for generating enormous state wealth, influence, power, and prestige on the international stage. Indeed, it is difficult if not impossible to achieve lasting global mastery without it.” In addition to its instrumental values and premium values (for example, brain gain for America from other countries’ brain drain, and capital gain out of capital flight from autocratic states), a liberal democratic republic has the most important intrinsic value as an effective organizational mechanism for political governance: to overcome both the prisoners’ dilemma (to end anarchy) and the dictators’ dilemma (to end autocracy and keep power under check) [15, 16]. He rejects the “long game” argument by explaining: “[P]roponents of autocratic advantage theory state that autocracies can make long-term plans and stick to them. But theory and history suggest that autocracies are erratic strategic decision-makers” [14:193]. Kroenig offer us an even longer perspective from the vantagepoint of democratic development: “[T]he history of Western civilization can be thought of as the passing of the torch of liberal hegemony from Athens to Rome, to Venice, Amsterdam, and London, and on to its current resting place in Washington, DC” [14:4]. The US victories in the World War II and the Cold War, the ensuing American “unipolar moment” and the Third Wave of Democracy must have astounded China, one of the five surviving communist countries, with a strong message: “the end of history” [also see: 11; 8; 19]. Stepping into the shoes of Chinese leaders with heightening anxiety in the era of democratization and globalization, we can easily understand that they believe a long, longer game had been concocted by the US to subjugate China. To build his theory of democratic advantage, Kroenig [14:8] reviews seven cases of democracy vs. autocracy in the history of more than two thousand years (his democracy is more an open system which covers both democracy and republic in content) [14:18]. If we can distinguish these two forms of system under the terminology of Aristotle, after the creation of the representative democracy in the American revolution, the rise of market economy, and the expansion of enfranchisement, a liberal democracy has become a new synthesis. Like Doshi, Kroenig follows the social scientific research design and builds a falsifiable theory by betting on the highly likely collapse of Chinese communist regime [14:194] to be the next validating case. Even if history follows the course predicted by Kroenig, his proposition still needs to stand the test of falsification. Instead of waiting for the “black swan,” we can move one step closer to a scientific theory by examining a counter example that a democracy was defeated by an autocracy of comparable power, for example, France vs. Germany in 1940 during the World War Two. Due to space limitation, I can only point out that the immobilisme resulting from a parliamentary system plus multi-party system reminds us the vulnerability of democracy in maintaining the public good (such as national defense). Under American presidentialism, Le Mal Français is called deadlock, gridlock, institutional sclerosis [22] or democratic sclerosis [4]. In reflecting upon American domestic governance, Haas [11:39–40] does not realize this danger. His policy prescription needs to appreciate of the pluralistic and diverse American society with talented immigrants; and to invest in education, including international/area studies, for Americans’ better understanding of the world. Even though Lissner and Rapp-Hooper had a better understanding of human capital [18:122–123], as they were suggesting for “building strength at home” [18:5–6], they failed to ask how we can muster strength to build. For example, they did not quite reconcile the tensions between political independence of nations and global openness, such as the advocacy for openness/transparency and the insulation of US foreign policy, and the close techno-state partnership being a strength for China but a failure for the US. Kroenig makes a similar mistake focusing too much on the hardware, especially weapons [14:132–133]. As clearly detected by Haas [9:186–187], some American strategists might have already been entrapped by the Chinese deceptive stratagem of “empty fortress” (“rational irrationality” in Western parlance) to demand more guns at the cost of butter, namely infrastructure and human infrastructure (e.g., education and health) in the US. Economy [6:28] rightly warns the United States not to fall into a dyadic zero-sum game with China that could isolate the US from its allies and partners. “Instead,” she argues, “the fundamental challenge presented by China is to the broader values, norms, and institutions that underpin the current rule-based order.” Unfortunately, Xi has been advocating for “a global community of shared future,” a “win-win cooperation”; and “as world powers with rich cultural and historical heritage” claiming “long-standing traditions of democracy” with Russia [11]. This advocacy for “democratizing international relations” allows Xi to occupy the moral high ground in the global discourse [6:8–9]. In contrast, many American leaders and thinkers are defending the status quo, being restricted by an American-centric straightjacket. For example, Lissner and Rapp-Hooper [18:11] believe that “The term ‘liberal international order’ has always served as a shorthand for a fairly benign form of US hegemony.” Such a historic myopia writes off Pax Britannica too easily. The same ignorance about the “Cold War consensus”, “particularly during the early decades of US-USSR rivalry” overemphasized the “unity” without adequate attention to McCarthyism that left deep wounds in American society. Similar to McCarthyism [3:22, 49, 66], under Trump, the debate on China risked going astray into paranoia. The veteran diplomat George Kennan [12:228] deplored that the witch hunt under McCarthyism caused the purge of a generation of “China Hands” and implanted “a lasting doubt” into his consciousness about “the adequacy of our political system”. Today we have similar weaknesses with both historical myopia and American-centric analysis. For example, when the US has failed to ratify UNCLOS for four decades, it is counterproductive to use it over South China Sea dispute [18:142–143]. Furthermore, Doshi’s terminology of “American liberal order,” “American order,” “US order,” and “US hegemony” loses the last thin internationalist veneer and refocuses the urgent issue of “China’s subverting the liberal order” as the “US order being displaced by the China order” [2:329]. Similarly, Kroenig uses the terms “American leadership”, American “benevolent hegemony,” ‘the global pecking order,” “global mastery,” “liberal leviathan,” and “American era”. This is not the best strategy to spread American soft power, and the Chinese propogandists love to exploit this for their “sharp power.” Conclusion Based upon the track record, while its democracy and strategic position has gone through numerous crises, the United States has emerged stronger and smarter. As Walter Lippmann [17] assured Americans, the good society is a free society which will not be vanquished by the “providential state” with a command economy and repressive politics. Given a reviving authoritarian/totalitarian political economy, many Americans seem inclined toward a direction of moral equivalence and self-doubt. Liberal democratic political economy is in danger or without strong advocates. Furthermore, a schizophrenia between self-aggrandizement and self-defeatism hampered many analysts from understanding the internal dynamics within China and the defensive/paranoid nature of China’s policy. Kroenig has correctly realized that autocracies’ “biggest security threat comes from their own people” [14:222], as evidenced by China’s bigger spending on internal security (“stability-maintenance”) than national defense. However, the profiled authors have given up on efforts to democratize China’s regime, which if we can draw a lesson from the US victory over USSR in the Cold War, such an abandonment may lose an effective strategy. At Biden’s Summit for Democracy (2021), no representative was invited from the Chinese overseas democracy movement. This lack of Chinese participation shows up again when Doshi made several Chinese transliteration errors and cited two wrong cases: the submarine incident on [2:83] was nothing more than a blunder due to lack of training; and Zheng Min and Zheng Ming are the same person [2:96–97]. The ultimate goal for the United States is a liberal, democratic, just, progressive and sustainable global order, which cannot easily be categorized as an American Order versus the China Order, the latter as a challenger and replacement. Almost all the profiled authors have emphasized that to achieve the public good of global governance and order, American allies are essential. However, it would be wrong-headed to pursue a “single organization” or “standing organization,” namely democratic allies and partners form ‘a new Alliance of Free Nations” [14:221]. To consolidate a democratic alliance against the autocracies in a trench war is not a way forward, either. To tear down the Bamboo Curtain and to penetrate the bamboo network [27], the starting point lies in the idea of “rhizome” and “rhizomatic network.” According to Deleuze and Guattari [5:21–22], a rhizome as subterranean stem is different from roots and radicles as in a tree (which is dominant and hierarchical, as advocated by American-centric strategists). Its principles of connection, heterogeneity, and multiplicity open up multiple entryways and constitute a contrast to unity and “either-or dichotomy.” Although Lissner and Rapp-Hooper [18:87] do not agree that “autocracies vs. democracies” is the leitmotif of global rivalry, and Haas does not see an ideological competition, President Biden’s characterization of “democracy versus autocracy’ as the major battle of the 21st century is more accurate. However, in this human progress, the US must uphold strong democratic ideals and use this strength to build a strong network of allies. As the whole world has been thrown into a once-a-century comprehensive crisis (a mixture of pandemic, recession and war), we will soon see which system, democracy or autocracy, will survive better in this humongous stress test. Disclaimer I hereby disclose that there is no conflict of interest in writing this article. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Chen, Ping. 2016. The Age of Plunder: The 2008 Economic Crisis as Turning a Point in Chinese History and World Civilization, edited by Ming Xia. Hong Kong: The iSun Affairs Limited. 2. Doshi, Rush. 2021. The longer game: China’s Grand Strategy to Displace American Order. New York, NY: Oxford University Press. 3. Cohen, Michael. 2020. Disloyal: the true story of the former Personal Attorney to President Donald Trump. New York: Skyhorse Publishing. 4. Corcoran, Edward. 2022. “Democratic Sclerosis”. Website Data at: https://www.huffpost.com/entry/democratic-sclerosis_b_1190539. Accessed on 08 May 2022. 5. Deleuze, Gilles and Felix Guattari. 1987. A Thousand Plateaus. Minneapolis, Minnesota: University of Minnesota Press. 6. Economy, Elizabeth C. 2022. The World according to China. Oxford, UK: Polity. 7. Friedman, Thomas, and Michael Mandelbaum. 2011. That used to be us: how America fell behind in the World it invented and how we can come back. New York, NY: Farrar, Straus and Giroux. 8. Fukuyama, Francis. 2002. The end of history and the last man. New York, NY: Perennial. 9. Haas, Ryan. 2021. Stronger: adapting America’s China Strategy in an age of competitive interdependence. New Haven, CT: Yale University Press. 10. Huntington, Samuel. 1991. The Third Wave: Democratization in the Late twentieth Century. Norman and London: The University of Oklahoma Press. Joint Statement of the Russian Federation and the People’s Republic of China on the International Relations Entering a New Era and the Global Sustainable Development, 4 February, 2022. Web data at: http://en.kremlin.ru/supplement/5770. Accessed on 2 June 2022. 11. Joint Statement of the Russian Federation and the People’s Republic of China on the International Relations Entering a New Era and the Global Sustainable Development,4 February, 2022. Web data at: http://en.kremlin.ru/supplement/5770. Accessed on 2 June 2022. 12. Kennan, George. 1972. Memoirs 1950–1963. New York, NY: Pantheon Books. 13. Kennedy, Paul. 1987. The rise and fall of the Great Powers. New York, NY: Vintage Books. 14. Kroenig, Matthew. 2020. The return of Great Power Rivalry: Democracy Versus Autocracy from the Ancient World to the US and China. New York, NY: Oxford University Press. 15. Lichbach, Mark Irving. 1995. The rebel’s Dilemma. Ann Arbor: University of Michigan Press. 16. Lichbach, Mark Irving. 1996. The Cooperator’s Dilemma. Ann Arbor: University of Michigan Press. 17. Lippmann, Walter. 1936. The Good Society. New York, NY: Grosset & Dunlap. 18. Lissner, Rebecca, and Mira Rapp-Hooper. 2020. An Open World: how american Can Win the Contest for Twenty-First Century Order. New Haven, CT: Yale University Press. 19. Mann, James. 2007. The China Fantasy: why capitalism will not bring democracy to China. New York, NY: Penguin Books. 20. Nathan, Andrew J., and Robert S. Ross. 1998. The Great Wall and the empty fortress: China’s search for security. New York, NY: W.W. Norton. 21. Navarro, Peter. 2008. The coming China Wars: where they will be fought and how they can be Won. Upper Saddle River, NJ: FT Press. 22. Olson, Mancur. 1982. The rise and decline of nations: economic growth, stagflation, and social rigidities. New Haven, CT: Yale University Press. 23. Popper, Karl. 1972. The Open Society and its enemy. Princeton, NJ: Princeton University Press. 24. Stern, Fritz. 1965. The politics of Cultural despair: a study in the rise of the Germanic ideology. New York, NY: Anchor Books. 25. Tuchman, Barbara W. 1970. Stilwell and the american experience in China, 1911-45. New York, NY: Grove Press. 26. Wang, Huning. 1991. America against America. Shanghai: Shanghai People’s Publishing House. 27. Weidenbaum, Murray, and Samuel Hughes. 1996. The Bamboo Network: how Expatriate Chinese Entrepreneurs are creating a New Economic Superpower in Asia. New York, NY: Martin Kessler Books.	36474773	PMC9716157	NO-CC CODE	2022-12-03 23:20:54	no	J Chin Polit Sci. 2022 Dec 2;:1-13	utf-8	J Chin Polit Sci	2,022	10.1007/s11366-022-09840-0	oa_other
==== Front J Food Sci Technol J Food Sci Technol Journal of Food Science and Technology 0022-1155 0975-8402 Springer India New Delhi 5634 10.1007/s13197-022-05634-7 Original Article Effect of addition of Syrian thyme (Thymus syriacus) on physiochemical and sensory quality of Sudanese Mudaffara cheese during storage El gabali Tasneem M. M. A. [email protected] Jadain Osman A. M. [email protected] http://orcid.org/0000-0001-8173-7693 El Zubeir Ibtisam E. M. [email protected] [email protected] grid.9763.b 0000 0001 0674 6207 Department of Dairy Production, Faculty of Animal Production, University of Khartoum, P. O. Box 321, Khartoum, Sudan 2 12 2022 111 20 5 2022 11 10 2022 © Association of Food Scientists & Technologists (India) 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This study was designed to introduce Syrian thyme (Thymus syriacus) as a new additive flavor for Mudaffara cheese. Mudaffara cheese was prepared from cow’s milk using the commonly used black cumin as control and Syrian thyme (0.3 and 0.5%) as treatment. The physiochemical properties and the sensory attributes were evaluated. The results indicated that Mudaffara cheese samples flavored with 0.3% Syrian thyme were significantly (P < 0.05) higher in protein and acidity content compared to the other cheeses. During the storage period, significant (P < 0.05) differences were obtained for all the studied physicochemical parameters except the ash content. Also the interaction of additives and storage period showed significant (P < 0.05) effect on the protein and fat content of Mudaffara cheese samples. However the additives had no significant effect on all sensory characteristics except the general acceptability. According to the panelist test, the overall acceptability of Mudaffara cheese sample flavored with 0.5% Syrian thyme showed the highest numerical score compared to the others Mudaffara cheese samples. During the storage period, Mudaffara cheese samples revealed significant (P < 0.05) variations in texture, acidity, flavor, taste and general acceptability scores. This study concluded that Mudaffara cheese can be flavored with Syrian thyme at a rate of 0.3 and 0.5%. Supplementary Information The online version contains supplementary material available at 10.1007/s13197-022-05634-7. Keywords Aromatic plants Braided cheese Proximate analysis Acidity Sensory characteristics Ripening ==== Body pmcIntroduction Cheese is a concentrated form of milk with the benefit of prolonged shelf life and an important health contribution to human. It is primarily a rich source of essential nutrients such as proteins, bioactive peptides, amino acids, fat, fatty acids, vitamins and minerals (Walther et al. 2008). Cheese is a dairy product that has played a key role in human nutrition for centuries and it is the most popular dairy product in Sudan. The braided; semi hard cheese (Mudaffara) is among the most popular types of cheese in Sudan (Mohammed Salih et al. 2011; Harun et al. 2015; Farah and El Zubeir 2020). Mudaffara cheese is a braided, unmatured semi hard cheese, which originated in the Mediterranean area and now is widely produced and consumed in Sudan (Mohammed Salih et al. 2011; Abdalla and Gamer Eldin 2018; Farah and El Zubeir 2020). Mudaffara cheese is a pickled semi-hard cheese braided into a characteristic shape, to which spices such as black cumin (Nigella sativa) are added (Abdalla et al. 2019). Mudaffara cheese production in Sudan is a small business that each producer develops and adopted his own procedure for its production (Abdel-Razig et al. 2014). Harun et al. (2015) reported 17.2 ± 2.13% for fat, 21.49 ± 2.53% for protein, 51.89 ± 12.25% for total solids, 10.96 ± 9.32% for ash and 0.48 ± 0.06% for acidity content of Mudaffara cheese. Sudanese braided cheese (Mudaffara) samples had a fat content of 26–37%, while 51% had low protein content (7–10%), moreover 83% of the samples had a total solids content of 55–70%, 70% of samples had an ash content of 3–5.99%, and 39% of the samples revealed highly acidic of 1.0–1.5% (Abdalla et al. 2019). The means reported for Mudaffara cheese by Farah and El Zubeir (2020) were 61.69 ± 0.12% total solids, 30.47 ± 0.31% protein, 22.00 ± 0.28% fat and 4.17 ± 0.06% ash and 0.46 ± 0.29% acidity. Spices and herbs have been used as flavor, colour, aroma, enhancing agents and for the preservation of foods and the addition of herbs and spices or their extracts to different dairy products make these products act as a carrier for nutraceuticals (El-Sayed and Youssef 2019). Thyme is an aromatic plant and is widely distributed over the Mediterranean area. The generic name comes from the Greek verb Thym, which translates to perfume, in allusion to the intense and pleasant aroma of the plant (Nieto, 2020). Thyme (Thymus vulgaris) is among the aromatic plants commonly used in Turkish traditional yoghurt drink, yoghurt and various cheese types. In addition to their taste and aromatizing uses in dairy products, bioactive molecules extracted from medicinal plants can be used as natural preservative additives due to their antimicrobial and antioxidant activity (Kaptan and Sivri 2018). T. syriacus is used as herbal tea and condiment, fresh leaves are used for the aromatization of home-made jams, candies and similar confections, it is also known to have positive results for coughs and other respiratory complaints, as well as some cases of gastrointestinal disorders (Al Mariri et al. 2013). Mudaffara cheese is commonly flavored with black cumin (N. sativa) seed. Due to some common flavored of Syrian thyme to black cumin, this study was designed to introduce it as a new additive for flavoring Mudaffara cheese. It is also meant to evaluate the effect of addition of Syrian thyme on the physiochemical and sensory quality of Sudanese braided (Mudaffara) cheese during storage. Materials and methods This experiment was conducted at the laboratory of Diary Technology, Faculty of Animal Production, University of Khartoum from 9/2/2020 to 24/3/2020. In this study two different favoring additives (0.5% black cumin and Syrian thyme with two levels: 0.3 and 0.5% were added to the Mudaffara cheese. chemical compositions of cheese were determined weekly for 4 weeks, while the sensory evaluations were carried out for 2 weeks. The fresh raw cow’s milk used in the processing of Mudaffara cheese in this experiment was obtained from the University of Khartoum farm on 9/2/2020. Syrian thyme (T. syriacus), black cumin (N. sativa L.) and starter culture were obtained from the local market. Rennet sticks (Christen Hansen’s laboratories—Copenhagen, Denmark was used; 1 stick coagulate 50 kg milk). Both Syrian thyme and black cumin were cleaned from impurities using hot water and dried before use. Thyme was crushed into powder form, while the whole seed of black cumin before added to the cheese paste. Manufacture of mudaffara cheese The steps for cheese processing were described previously by Farah and El Zubeir (2020). After the milk was filtered and heated to 40 °C, 1% of the starter culture (Lactobacillus bulgaricus and Lactobacillus thermophilus) and 1% of dissolved rennet powder were added and stirred for 2–3 min to ensure uniform distribution of the rennet. The milk mixture was left undisturbed to coagulate into a curd. After about 40 min, coagulation was complete and the whey was drained from the curd that was cut into slices for more whey drainage. The curd was then incubated at 40 °C for 1.5 h to reach the required elasticity and acidity (0.54–0.60%). Every half an hour during this incubation, the inflatable of the curd was measured by dipping a small piece of curd (about 20 g) into hot water (85 °C), then holding it in hands, kneading and pulling it to form a cord of 2 m long. If the curd breaks before reaching this length, ripening is considered incomplete. When the curd became a smooth paste that showed satisfactory stretching to a rope of more than 4 m long and elastic, the curd was cut into strips. Then 4–5 pieces were taken at a time and put in the hot water (65–75 °C) for 3–5 min using wooden paddles until the curd became smooth. The curd was divided into three parts; 700 g each; for the two different flavoring additives (0.5% black cumin and Syrian thyme with the two levels: 0.3 and 0.5%. After that the additives were added to the hot paste before braiding. The curd was then hand worked and drawn to form a long rope which was then braided and washed by immersing it in cold water. The three types of cheese were taken after the cooking process and put in steel buckets. All the braided cheeses were immersed in sterilized salted whey (3% NaCl) for 12 h. Then all samples of Mudaffara cheese were stored in the refrigerator at 4 °C for 4 weeks. Analysis of Mudaffara cheese The chemical composition of Mudaffara cheeses was conducted at a week interval for 4 weeks. The fat content (Gerber method), the protein content (Kjeldahl method), total solids content, ash content and the titratable acidity were determined according to the AOAC (2000). The scores of sensory attributes of Mudaffara cheese samples that were stored in the refrigerator were evaluated by 10 semi-trained panelists; who are familiar with Mudaffara cheese. The judgment was done for the appearance, color, flavor, texture, taste, acidity and overall acceptability. The evaluation was based on on the five-point scale according to the method described by Lim (2011) using sensory evaluation sheet (excellent = 5, very good = 4, good = 3, fair = 2 and poor = 1). Statistical analysis The data of the present study were analyzed using the SPSS (Statistical Package for Social Sciences) (version. 16). General Linear Model was used to determine the effect of treatment and storage period on the physicochemical and sensory properties of Mudaffara cheese. The Least Significant Difference test was used for the mean separation between the treatments. The level of significance (P ≤ 0.05) was used to consider the variations among the different treatments to get 95% countenance os accuracy. The graphs were plotted using Microsoft Office Excel 2007. Results and discussion Effect of additives on physicochemical characteristics of Mudaffara cheese Total solids content The total solids content of Mudaffara cheese was not affected significantly (P > 0.05) by the addition of 0.3 or 0.5% Syrian thyme (48.5 ± 4.5 and 46.0 ± 5.6%, respectively) and black cumin (47.7 ± 4.8%) as shown in Table 1. However Slman et al. (2020) found significantly higher total solids content of Syrian white cheese to which thyme powder was added at the beginning of the storage period and up to 45 days of the storage. Nonetheless, Slman et al. (2020) used higher inclusion levels (1 and 2%), whereas in this study lower levels were used (0.3 and 0.5%), which may explain the differences in findings.Table 1 Effect of flavored additives on physicochemical characteristics of Mudaffara cheese (M ± SD) Treatment Physicochemical composition (%) Total solids Fat Protein Ash Acidity Black cumin (0.5%) 47.7 ± 4.8 20.6 ± 4.1 20.9 ± 1.6 c 3.2 ± 0.43 0.60 ± 0.12 b Syrian thyme (0.3%) 48.5 ± 4.5 21.0 ± 3.9 24.1 ± 7.7 a 3.1 ± 0.30 0.74 ± 0.11a Syrian thyme (0.5%) 46.0 ± 5.6 22.7 ± 3.5 22.6 ± 6.3 b 3.2 ± 0.51 0.71 ± 0.90 a LS NS NS * NS a,b,cMeans bearing the same superscripts letters in the same column are not significantly different (P > 0.05) M mean value, SD standard deviation, LS level of significant, NS not significant *P < 0.001, P < 0.01 Harun et al. (2015) and Abdalla et al. (2019) reported a higher total solids content (51.89 ± 12.25 and 58.19 ± 4.21%, respectively) for Mudaffara cheese than the one reported in this study (48.5 ± 4.5%). The differences reported in total solids content could be due to differences in moisture content as a result of the method of manufacture and packaging used in each study. However the total solids content of Mudaffara cheese was significantly (P < 0.05) affected by the storage period (Table 2). It increased gradually at week 2 (51.0% ± 3.4) and then decreased at the end of the storage period. This might be because the cheese was immersed in its brine during the first 2 weeks and the decrease was the result of ripening. After week 2, the cheeses were kept in the refrigerator; this situation might have influence on ripening process. Similarly fluctuation of the total solids content of Mudaffara cheese during the storage was reported previously by Harun et al. (2015) and Abdalla and Gamer Eldin (2018). However slight decrease in the total solids of Mudaffara cheese during the storage was found by Farah and El Zubeir (2020). The difference in the total solids content of Mudaffara cheese may be due to several factors, including the close correlation between moisture content and the total solids, the preservation methods applied and the storage temperature (Abdalla and Gamer Eldin 2018 and Farah and El Zubeir, 2020). Also the expulsion of moisture from cheese curd is the reason for the increase in the total solids content of Mudaffara cheese (Harun et al. 2015). The increase in the total solids contents could be also attributed to a decrease in the moisture content as a result of lactic acid developments, which caused curd contraction (Abdalla et al. 2019). However the decrease in total solids content was due to the degradation of protein and lipolytic activity on the fat content. The refrigerator storage of Mudaffara cheeses after the second week might also be a reason for suppressing the microflora o the cheese and hence reducing the rate of protein and fat degradation.Table 2 Effect of storage period on physicochemical characteristics of Mudaffara cheese (M ± SD) Storage periods Physicochemical composition (%) Total solids Protein Fat Acidity Ash Week 1 42.5 ± 3.1b 19.4 ± 2.3c 17.0 ± 2.5c 0.58 ± 0.09b 3.5 ± 0.37 Week 2 51.0 ± 3.4a 28.9 ± 7.3a 22.0 ± 3.2b 0.79 ± 0.05a 2.9 ± 0.42 Week 3 49.4 ± 4.2a 24.0 ± 1.6b 22.7 ± 0.8ab 0.73 ± 0.08a 3.3 ± 0.37 Week 4 46.7 ± 4.8ab 17.9 ± 1.2d 25.0 ± 2.7a 0.62 ± 0.12b 3.1 ± 0.31 LS * * * * NS a,b,cMeans bearing the same superscripts letters in the same column are not significantly different (P > 0.05) M mean value, SD standard deviation, LS level of significant, NS not significant P < 0.001, P < 0.05 The interaction of the used additives and the storage period showed no significant (P > 0.05) effect on the total solids content of Mudaffara cheese (Fig. 1a). However, the cheese samples flavored with 0.3% Syrian thyme generally had higher total solids content during the storage period in comparison with cheese samples flavored with 0.5% Syrian thyme and black cumin. The reason could be due to the photolytic effect of Syrian thyme and that the Syrian thyme should be used at a rate of up to 0.3% and not more than 0.5%. This supported the findings drawn by Slman et al. (2020) who found that adding 2% of the thyme powder compared to that using 1% increases the percentage of the total solids in the cheese mass as well as the high concentration of the brine. This is mainly because Mudaffara cheese flavored with 0.3% Syrian thyme revealed higher protein content compared to other cheeses (Table 1). All types of Mudaffara cheese obtained during the present study comply with SSMO for the total solids content (SSMO 2007).Fig. 1 a Comparison of the total solids content of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. b Comparison of the protein content of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. c Comparison of the fat content of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. d Comparison of the ash content of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. e Comparison of the acidity content of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage Fat content Data in Table 1 revealed that the fat content of Mudaffara cheese flavored with black cumin (20.6 ± 4.1%) was slightly lower compared to that flavored with Syrian thyme at 0.3% (21.0 ± 3.9%) and 0.5% (22.7 ± 3.5%). This suggested that the fat content is high in Syrian thyme compared to the black cumin. Also Jemaa et al. (2017) reported that for treated milk, in addition to casein antioxidant activity, T. capitatus essential oils also had contributed to the protection of milk fat from oxidation. This is because the antioxidants molecules from T. capitatus essential oils work against oxidation by donating their hydrogen atom to the lipid free radicals to stop the chain reaction from proceeding further. Lower values were obtained by Harun et al. (2015) who obtained 17.2 ± 2.13% fat content for Mudaffara cheese. Also Abdalla and Gamer Eldin (2018) found that the fat content for Mudaffara cheese brined using 15% salt (19.38%). However Abdalla et al. (2019) reported a higher mean of fat in Mudaffara cheese (26.46 ± 4.94%). The fat content of the cheese was not affected significantly (P > 0.05) by the addition of Syrian thyme and black cumin. This is because both of the flavored additives used are aromatic plants. Moreover various spices, herbs and their volatile oils are included and often identified as being among the most effective natural antioxidants (Bhat et al. 2014). Also Nieto (2020) stated that the development of functional foods could be enhanced by the inclusion of thyme because of its value-added properties, which could be of great interest to both the scientific community and the food industry. In particular, the T. syriacus plant extracts are found to contain various phytochemicals with biological activity that is attributed to the medicinal characteristics of this plant, which can be of valuable therapeutic uses (Tamim and Akrama 2020). Moreover the chemical composition and high-level of antimicrobial activity of its essential oil and major constituents of T. syriacus are efficient against some Gram-negative human pathogenic bacteria (Al Mariri et al. 2013). The fat content of Mudaffara cheese was affected significantly (P < 0.001) during the storage period, it increased significantly with the advancement of the storage period (Table 2). The fat content of Mudaffara cheese was significantly (P < 0.05) affected by the interactions of the used additives and storage period. The highest value of fat content was found in cheese samples flavored with 0.3% Syrian thyme at week 4 (26.5% ± 2.1), however the lowest value of fat content was found in cheese samples flavored with black cumin (Fig. 1b). The increase in fat content of Mudaffara might be due to high moisture loss during the storage (Harun et al. 2015 and Farah and El Zubeir 2020). The increase in fat content till day 21 might be due to the high moisture loss during the storage, while the decrease in the fat content at end of the storage period may be attributed to breakdown of fat by microorganisms and their loss from cheese (Harun et al. 2015). Protein content As shown in Table 1, the protein content of Mudaffara cheese was significantly (P < 0.001) affected by the addition of Syrian thyme. The cheese samples flavored with 0.3% Syrian thyme (24.1 ± 7.7%) showed the highest value of protein content in comparison with the cheese samples flavored with 0.5% Syrian thyme (22.6 ± 6.3%) and black cumin (20.9 ± 1.6%). Abdalla et al. (2019) reported a lower mean for protein (14.72 ± 8.14%) content of Mudaffara cheese in comparison to the current findings. Also Harun et al. (2015) reported 21.49 ± 2.53% for the protein content of Mudaffara cheese. The present findings agree with El-Sayed (2017) who reported that the higher ratios of thyme powder in the blend of Ras processed cheese could lead to the loss in protein network, which is due to the decrease that occurred in the intact casein of the blend with increasing the thyme powder. More explanation was also stated by Tarakçı and Deveci (2019) who found that the cheese incorporated with black cumin possessed higher ripening rates than others added spices and the highest decrease in the amount of β-casein was observed in the thyme-added cheese, followed by the black cumin-added cheeses. They added that thyme was also more effective on αs1-casein fraction. On the other hand, when considering the starters’ contribution to the proteolysis of aS1-casein it would be possible to suggest that the antimicrobial activities of tested spices inhibited the microbial protease activities (Tarakçı and Deveci 2019). As shown in Table 2, the protein content of Mudaffara cheese was significantly (P < 0.001) affected by the storage period. It increased gradually at week 2 (28.9 ± 7.3) and then decreased at the end of the storage period. The reason for the increase could be due to the loss of the moisture content from the cheese, while the reduction that occurred after week 2 might because of refrigerating storage of the cheese. However Altahir et al. (2014) and Farah and El Zubeir (2020) found that the protein content of Mudaffara cheese was found to decrease significantly during storage. The reduction in protein content was possibly due to the activity of proteolytic microorganisms leading to protein degradation. However the expulsion of moisture from cheese curd could be another reason for the increase in protein content of Mudaffara cheese (Harun et al. 2015). The interactions of the additives used and the storage period showed a significant (P < 0.001) effect on the protein content of Mudaffara cheese (Fig. 1c). The highest value of protein content was found in the cheese samples flavored with 0.3% Syrian thyme at week 2 (35.2% ± 0.00), while the lowest value of protein content was found in the cheese samples flavored with 0.5% Syrian thyme at week 4 (16.9% ± 0.39). Storing of Mudaffara cheese in the refrigerator after week 2 might be the reason of these contradictory findings. Hence further futures research should be conducted in order to verify and explain the various reasons that might involved during the storage of the cheeses to which some aromatic plants are added. The decrease in the protein content during pickling was a direct result of protein degradation leading to the formation of water-soluble compounds, some of which were lost in the pickling (Farah and El Zubeir 2020). Tarakçı and Deveci (2019) reported that more decrease in the amount of β-casein was found in the thyme-added cheese compared to the cheese incorporated with the black cumin and the increase in parameters of proteolysis leading to protein degradation during cheese ripening is closely related to the increase in bacterial activity and enzyme activity over time. Ash content More or less similar values of ash content of Mudaffara cheese flavored with black cumin, 0.3 and 0.5% Syrian thyme were obtained (3.2 ± 0.43, 3.1 ± 0.30 and 3.2 ± 0.51%, respectively). Moreover the results in Table 1 showed that the Syrian thyme and black cumin did not show a significant (P > 0.05) effect on the ash content of Mudaffara cheese. The small variations in the obtained values might be due to the mineral content of the used concentration of the spice added. Thyme has a high level of iron, manganese and copper (Ozkan et al. 2007). The rate of salting the cheese during this study was 3% for all types of Mudaffara cheeses. Similarly the ash content of Mudaffara cheese reported by Abdalla et al. (2019) was 3.52 ± 1.07%. However Harun et al. (2015) reported a higher value (10.96 ± 9.32%) for the ash content of Mudaffara cheese. The variation is an indication of the different level of sodium chloride used for salting the cheese. The storage period had no significant effect (P > 0.05) on the ash content of Mudaffara cheese. However this study showed a decrease in the ash content of Mudaffara cheese during the storage (Table 2). The results are in disagreement with those who reported that the ash content of cheese increased as the storage period progressed (Altahir et al. 2014). The results in Fig. 1d showed that the interactions and the used additives and storage period had no significant (P > 0.05) effect on the ash content of Mudaffara cheese. The highest value of the ash content was found in the cheese samples flavored with black cumin at week 1 (3.8 ± 0.05%), whereas the lowest value of the ash content was found in the cheese samples flavored with 0.5% Syrian thyme. This might be because the minerals content of Syrian thyme was low compared to those in black cumin seeds. Similarly Farah and El Zubeir (2020) found that the ash content of Mudaffara cheese was significantly (P < 0.001) higher on day 0 and 7 (4.17 ± 0.06 and 4.08 ± 0.06%, respectively and significantly (P < 0.01) lower value was obtained at day 35 (2.93 ± 0.06%). The ash content in curd and whey was very much affected by the diffusion of salt from curd into the whey. Acidity content Table 1 illustrated that highest value of acidity content was found in Mudaffara cheese samples flavored with Syrian thyme at 0.3% (0.74 ± 0.11%) and 0.5% (0.71 ± 0.9%), while the lowest value was found in Mudaffara cheese samples flavored with black cumin (0.6 ± 0.12%). However lower value was reported by Harun et al. (2015) for acidity content (0.48 ± 0.06%) of Mudaffara cheese. The acidity of Mudaffara cheese revealed 0.978 ± 0.294% (Abdalla et al. 2019). The variations could be due to the antimicrobial properties of Syrian thyme and black cumin. The acidity content of Mudaffara cheese was significantly (P < 0.01) affected by the addition of Syrian thyme (Table 1). Generally higher acidity content was found for Mudaffara cheese flavored with Syrian thyme compared to that flavored with black cumin. The higher level of the acidity in Mudaffara cheese flavored with 3% Syrian thyme compared to that using 5% Syrian thyme is indicative of the antimicrobial activity of the thyme (Al Mariri et al. 2013; Kaptan and Sivri 2018; Tamim and Akrama 2020) as well as black cumin (El Zubeir et al. 2005; Abdel-Gadir et al. 2013). El-Sayed (2017) found that increasing the thyme powder (0.1, 0.2, 0.3, 0.4 and 0.5%) in the base blend of Ras cheese samples decreased the pH values. Moreover they concluded that thyme powder concentrations especially 0.50% thyme; showed the highest effect on the total bacterial count, which suggested its antimicrobial effect on cheese microflora. Similarly Slman et al. (2020) observed higher acidity in the white cheese using 2% thyme powder compared to that using 1%, they attributed this to the presence of some components in thyme that caused the acidity of the product to rise when the added quantities are increased. This could also be due to the high pH value of thyme powder used in all treatments compared to the pH of the control Ras cheeses (El-Sayed, 2017). The titratable acidity was significantly (P < 0.001) affected by the storage period (Table 2). It increased gradually until week 2 (0.79% ± 0.05), then it decreased at the end of the storage period (0.62% ± 0.12). This might be because Mudaffara cheese was kept in its whey for 2 weeks and was removed from it during the rest of the storage that was done in the refrigerator. These results are in line with those reported that the acidity content of Mudaffara cheese increased gradually until day 21, then it decreased at the end of the storage period (Harun et al. 2015; Farah and El Zubeir 2020). The increase in the titratable acidity may be due to the growth of lactic acid bacteria that leading to the increase of lactic acid content (Abdalla et al. 2019). Data in Fig. 1e revealed that the acidity content of Mudaffara cheese did not affected significantly (P > 0.05) by the interaction of the additives used and the storage period. Generally, the cheese samples flavored with black cumin showed lower acidity content during the storage in comparison to the cheese samples flavored with 0.03 and 0.5% Syrian thyme. The antimicrobial and preservative properties of the black cumin were the reasons (El Zubeir et al. 2005; Abdel-Gadir et al. 2013). Also the efficiency of thyme in preventing the growth of some bacteria was proved (El-Sayed, 2017; Kaptan and Sivri 2018; Al-Obaidi and Hussein 2019; Tamim and Akrama 2020). Effect of additives on the sensory characteristics of Mudaffara cheese Data in Table 3 revealed that the sensory characteristics of Mudaffara cheese did not affected significantly (P > 0.05) by the additives used except for the overall acceptability. Bhat et al. (2014) reported that spices and herbs are commonly used for seasoning and increasing the shelf life of food and restoring health.Table 3 Effect of additives on sensory characteristics of Mudaffara cheese (M ± SD) Treatment Sensory characteristics Appearance Color Texture Acidity Flavor Taste Overall acceptability Black cumin (0.5%) 3.6 ± 1.20 3.9 ± 1.30 4.4 ± 1.01 4.1 ± 0.92 4.1 ± 1.02 4.4 ± 0.48 4.4 ± 0.49a Syrian thyme (0.3%) 3.7 ± 1.40 3.7 ± 1.40 4.0 ± 0.96 4.0 ± 0.96 4.0 ± 1.20 3.9 ± 1.30 3.9 ± 0.95b Syrian thyme (0.5%) 4.4 ± 0.93 4.6 ± 0.76 4.3 ± 0.99 4.4 ± 0.36 4.4 ± 1.01 4.4 ± 1.01 4.6 ± 0.63a LS NS a,b,cMeans bearing the same superscripts letters in the same column are not significantly different (P > 0.05) M mean value, SD standard deviation, LS level of significant, NS not significant P < 0.01 The appearance (4.4 ± 0.93) and color (4.6 ± 0.76) of Mudaffara cheese samples flavored with 0.5% Syrian thyme showed the highest numerical scores compared to those flavored with 0.3% Syrian thyme and 0.5% black cumin (Table 3). However El-Sayed (2017) reported that there were no changes in the appearance of processed cheese after adding thyme powder up to 0.2%. Increasing the ratio of Syrian thyme up to 0.3% was found to affect slightly the appearance, while with 0.5% the block of the processed cheese showed a slightly off color. Similar to the present study, non significant differences were obtained in the color of the plain soft cheese and those to which 1 or 2 ml of thyme oil/1000 ml milk were added, while a significant difference was found when adding 3 ml of thyme oil/1000 ml milk (Al-Obaidi and Hussein 2019). The appearance and color of Mudaffara cheese samples showed non significant (P > 0.05) effect during the 2 weeks of the storage (Table 4 and Fig. 2a and 2b, respectively). Also non significant variation was reported in the color of the surface of the cheese slices after 30 days of storage at a temperature of 4 °C when adding Thymus algeriensis (Bukvicki et al. 2018).Table 4 Effect of storage period on sensory characteristics of Mudaffara cheese (M ± SD) Storage periods Sensory characteristics Appearance Color Texture Acidity Flavor Taste Acceptability Week 1 4.1 ± 1.30 4.4 ± 1.30 4.7 ± 0.46a 4.5 ± 0.75a 4.6 ± 0.67a 4.6 ± 0.59a 4.6 ± 0.51a Week 2 3.7 ± 1.10 3.8 ± 1.04 3.8 ± 1.10b 3.8 ± 0.94b 3.7 ± 1.20b 3.8 ± 1.20b 4.0 ± 0.89b LS NS NS *** * a,b,cMeans bearing the same superscripts letters in the same column are not significantly different (P > 0.05) M mean value, SD standard deviation, LS level of significant, NS not significant P < 0.001, P < 0.01, P < 0.05 Fig. 2 a Comparison of the appearance scores of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. b Comparison of the color scores of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. c Comparison of the texture scores of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. d Comparison of the acidity scores of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. e Comparison of the flavor scores of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. f Comparison of the taste scores of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage. g Comparison of the overall acceptability scores of Mudaffara cheese flavored by black cumin and Syrian thyme during the storage The highest scores for texture was obtained in Mudaffara cheese samples flavored with black cumin (4.4 ± 1.01) followed by that flavored with 0.5% Syrian thyme (4.3 ± 0.99) compared to the cheese flavored with 0.3% Syrian thyme (4.0 ± 0.96). Similarly El-Sayed (2017) found that with increasing the percentage of thyme powder (0.1, 0.2, 0.3, 0.4 and 0.5%) in the base blend of Ras cheese samples, the oil separation and the firmness were found to increase. However Al-Obaidi and Hussein (2019) reported that the texture score showed non significant differences between the control cheese and those to which different concentrations of thyme oil were added. The texture of Mudaffara cheese was affected significantly (P < 0.001) during the second week of storage compared to that examined in week 1 (Table 4). The texture score was reported to decrease from 4.7 ± 0.46 in week 1 to 3.7 ± 1.1 in week 2 (Fig. 2c). This result was in line with those reported by Farah and El Zubier (2019) who found that the texture scores changed during the storage period with a decreasing trend. Similar to the present study, El-Sayed (2017) reported that the addition of thyme powder in the blend of 0.1 up to 0.3% improved the body and texture of the block cheese samples during the storage period. However Abdel-Razig et al. (2014) reported a lower value (3.32) for the texture score of the braided cheese at the beginning of the storage compared to the highest score (4.48), which was recorded at the end of the storage period. According to Bukvicki et al. (2018), the result of the sensory evaluation of cheese sprayed with essential oils of T. algeriensis revealed no changes in the texture of the surface of cheese slices after 30 days of storage at a temperature of 4 °C. These differences might be due to the fact that the sensory evaluation of Mudaffara cheese in the present study was evaluated only during week 1 and 2 because we missed the postgraduate students who act as panelist because the University was closed due to the announcement of the prevalence of some cases of covid 19 in Sudan. The highest value of acid score was found in Mudaffara cheese flavored with Syrian thyme at 0.5% (4.4 ± 0.36) compared to that flavored with 0.3% Syrian thyme (4.0 ± 0.96) or black cumin (4.1 ± 0.92). This indicated that the concentration of the added spices has a role in the acid taste of Mudaffara cheese. The obtained data was not significantly (P > 0.05) different. This matches with the achieved result shown in Fig. 1d that the lactic acid content of Mudaffara cheese did not affected significantly (P > 0.05) by the use of the different additives flavors used in this study. However the acid scores of Mudaffara cheese were significantly (P < 0.05) affected by the storage period (Table 4). The score was decreased from 4.5 ± 0.75 in week 1 to 3.8 ± 0.94 in week 2. The best flavor score (4.4 ± 1.01) was obtained for Mudaffara cheese flavored with 0.5% Syrian thyme followed by that flavored with black cumin (4.1 ± 1.02) compared to that using 0.3% Syrian thyme (Table 3). The flavor of the processed cheese was enhanced and became more preferable to panelists by incorporating thyme powder in the blend up to 0.2–0.3% compared to the control (El-Sayed, 2017). A significant difference was only found in the flavor score of the soft cheese when increasing the level of essential oil of thyme to 3 ml oil/1000 ml milk, while with the addition of 1 or 2 ml of thyme oil/1000 ml milk non significant variation was found for the flavor score of the soft cheese (Al-Obaidi and Hussein 2019). Moreover the aromatic plants that are used in imparting flavor and aroma in dairy products are also effective in increasing the nutritional value of the product and in extending its shelf life (Kaptan and Sivri 2018). As shown in Table 4 and Fig. 2e, the score flavor of Mudaffara cheese was significantly (P < 0.01) affected by the storage period. The score was found to decrease from 4.6 ± 0.67 in week 1 to 3.7 ± 1.2 in week 2. However Farah and El Zubier (2019) reported increasing trend for the flavor score during ripening and attributed the variations to differences in the cheese types and their processing conditions. Similarly El-Sayed (2017) reported that the addition of thyme powder in the blend using 0.1 up to 0.3% enhanced the flavor of the block cheese samples during the storage period. The improvement in the flavor was mainly due to the development of lactic acid bacteria, which controls the growth of undesirable microorganisms (Papetti and Carelli 2013). Shuang et al. (2014) also reported that it is important to perfect the flavor by blending and seasoning to improve the market acceptance of the cheese. Moreover the cheese was widely popular worldwide due to its good taste and the nice diverse flavor (Walther et al. (2008). According to Nieto (2020), the main objective of the use of thyme in food is to extend its shelf life; however the main limiting aspect for the use of the essential oil and plant extract of thyme is the development of negative organoleptic characteristics in foods that contributing to an unpleasant odor and taste. Data in Table 4 and Fig. 2f revealed that the taste scores of Mudaffara cheese were significantly (P < 0.01) affected during the storage period. The taste score was found to decrease from 4.6 ± 0.59 in week 1 to 3.8 ± 1.2 in week 2. The taste scores of Mudaffara cheese made by Farah and El Zubier (2019) also showed high significant (P < 0.01) differences during day 0, 28 and 35 of the storage, with a slight decrease during the rest of the storage period. The variation shown in Table 4 and Fig. 2f might be because of the rich content of vitamin A and C, iron, manganese and copper of thyme in addition to its characteristic sharp aroma that results from the essential oils, which give it the characteristic taste and medical properties as thyme had potent antioxidant effects originating from the thymol, carvacrol and p-cymene content of the plant (Ozkan et al. 2007). Moreover Thymus species are well-known culinary spices widely used all around the world, with pleasant taste and flavor. This in addition to its high antifungal and antiradical abilities, their essential oils could be recommended as natural antimicrobial additives for the shelf-life prolongation of the soft cheese (Nieto, 2020). According to the panelist, the cheese samples flavored with 0.5% Syrian thyme (4.6 ± 0.63) and black cumin (4.40.49) showed significantly (P < 0.01) higher numerical scores in the overall acceptability compared to cheese samples flavored with 0.3% Syrian thyme (3.9 ± 0.95). Incorporation of 0.5% Syrian thyme showed the best overall acceptability score which supported Al-Obaidi and Hussein (2019) who reported that the high concentration of thyme oil revealed positive effect on the general acceptance of the cheese. Similarly El-Sayed (2017) reported that total scores of the processed cheeses were acceptable and the addition of thyme powder in the blend up to 0.1–0.4% gave a better organoleptic quality compared to the control cheese and that the treatments with 0.6% thyme exhibited significantly lower quality attributes compared to other treatments. Also the higher score reported for the overall acceptability of Mudaffara cheese flavored with black cumin might be because the panelists are used to the flavor of black cumin in Mudaffara cheese. Mudaffara cheese is characterized by a close texture, yellow colour and slightly acidic taste (Abdel-Razig et al. 2014). It could also be attributed to the essential oil of black cumin, which increased sensorial attributes (Mishra et al. 2020). Also black cumin showed a lot of functional properties (Butt and Sultan 2010) and has significant preservative effect in improving the keeping quality of the cheese (El Zubeir et al. 2005; Abdel-Gadir et al. 2013). Results in Table 4 and Fig. 2g showed that the overall acceptability of Mudaffara cheese was significantly (P < 0.01) affected by the storage period. The score was decreased from 4.6 ± 0.51 in week 1 to 4.0 ± 0.89 in week 2. Similarly Abdel-Razig et al. (2014) reported that the storage period affected significantly the acceptability of Mudaffara cheese as the best acceptability score was obtained at day 30 compared to that obtained after 60 days of storage. However El-Sayed (2017) added that the storage of cheese products for up to 3 months slightly lowered the total quality attributes and this effect was more marked in the cheese samples stored at room temperature. Results in Fig. 2a–g showed that the interactions of the used additives and the storage period showed non significant (P > 0.05) effect on the sensory characteristics of Mudaffara cheese. Generally, Mudaffara cheese samples flavored with 0.5% Syrian thyme showed the highest numerical scores in all sensory parameters during the storage period except for texture and taste scores. Similarly the meltability and the sensory quality of Ras cheese samples were found to decrease with increasing of the thyme powder (0.1, 0.2, 0.3, 0.4 and 0.5%) in the base blend (El-Sayed, 2017). This suggested that a concentration of less than 5% should be used. Moreover it is better to consume Mudaffara cheese when fresh because all the sensory characteristics scores of Mudaffara cheese decreased from week 1 to week 2. Similarly Farah and El Zubeir (2019) found that the sensory evaluation scores for Mudaffara cheese showed decreasing values during the storage period. Ripening of cheese during the storage has a great influence on its compositional and acceptability as shown in Figs. 1 and 2. Conclusion From the results attained in this study, it can be concluded that there is a possibility of using Syrian thyme in addition to the commonly used black cumin in flavoring Mudaffara cheese. Syrian thyme improved physicochemical content at a rate of 0.3% and sensory characteristics at a rate of 0.5% when used as flavoring additives for Mudaffara cheese. The storage period revealed non significant variations for Mudaffara cheese. Consequently, the braided (Mudaffara) cheese industry is an applied method to save the milk and to give a product with high nutritional value with a long shelf life if there are a proper storage conditions. Hence it was recommended that the processing and consumption of braided (Mudaffara) cheese flavored with Syrian thyme should be processed and practiced at a large scale. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (XLSX 10 kb) Acknowledgements The technical help provided by the staff of the Department of Dairy Production Faculty of Animal Production, U of K during the processing of the cheese and its analysis is appreciated with thanks. Author contributions TMMAE: conceptualization, investigation, data curation, methodology, writing the original draft. OAMJ: data curation, validation, writing–review and editing. IEME: supervision, validation, writing–review and editing. Funding Not applicable. Data availability All the generated data obtained during this study are included with the submission of the paper. Declarations Conflict of interest The authors declare that they have no competing financial interests for this work. Consent to participate Not applicable. Ethical approval Not applicable in this study. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abdalla MOM Gamer Eldin HEA Effectof milk type and salt concentration on the physicochemical characteristics of Mudaffara (Braided) cheese Adv Res 2018 13 5 1 9 10.9734/AIR/2018/33121 Abdalla MOM Elsiddig HM Elhaj NM Suleiman TA Chemical and microbiological evaluation of Sudanese braided cheese (Mudaffara) Univ Khartoum J Agric Sci 2019 21 2 253 268 Abdel-Gader MKH El-bakri JM El Zubeir IEM The physicochemical and microbiological quality of the Sudanese yoghurt Mish produced by traditional and modernized methods Int J Dairy Technol 2013 66 2 273 278 10.1111/1471-0307.12010 Abdel-Razig AK Kunna MA Mohammed AS Effect of levels of black cumin seeds (N. sativa) and storage period on biochemical properties and acceptability of Sudanese braided cheese Sudan J Agric Sci 2014 1 55 62 Al-Mariri A Swied G Oda A Al Hallab L Antibacterial activity of T. syriacus boiss essential oil and its components against some Syrian gram-negative bacteria isolates Iran J Med Sci 2013 38 2 180 186 24031109 Al-Obaidi LFH Hussein HA Effect of adding different concentrations of thyme oil on reducing the total microbial count in the soft cheese to increase the shells life Biochem Cell Arch 2019 19 1 539 542 10.35124/bca.2019.19.1.539 Altahir MOE Elgasim EA Mohamed Ahmed IA Ripening of Sudanese braided (Muddaffara) cheese manufactured from raw or pasteurized milk: effect of heat treatment and salt concentration on mineral content Innov Romanian Food Biotech 2014 14 26 36 Bhat S Kaushal P Kaur M Sharma HK Coriander (Coriandrum sativum L.): processing, nutritional and functional aspects Afr J Plant Sci 2014 8 1 25 33 10.5897/AJPS2013.1118 Bukvicki D Giweli A Stojkovic D Vujisic L Tesevic V Nikolic M Sokovic M Marin PD Cheese supplemented with T. algeriensis oil, a potential natural food preservative J Dairy Sci 2018 101 5 3859 3865 10.3168/jds.2017-13714 29477526 Butt MS Sultan MT Nigella sative: reduces the risk of various maldies Crit Rev Food Sci Nutr 2010 50 654 665 10.1080/10408390902768797 20694927 El Zubeir IEM Abdalla WM El Owni OAO Chemical composition of fermented milk (roub and mish) in Sudan J Food Control 2005 16 7 633 637 10.1016/j.foodcont.2004.07.003 El-Sayed HA Effect of thyme on the quality and shelf life of block processed cheese J Food Dairy Sci 2017 8 8 335 340 10.21608/jfds.2017.38894 El-Sayed SM Youssef AM Potential application of herbs and spices and their effects in functional dairy products Heliyon 2019 5 6 e01989 10.1016/j.heliyon.2019.e01989 31338458 Farah NAM El Zubeir IEM Comparison on the acceptability of mudaffara cheese using black cumin and sesame seeds during storage Ann Food Sci Technol 2019 20 4 721 727 Farah NAM El Zubeir IEM Effect of addition of sesame seeds on physiochemical quality of Mudaffara cheese during storage Biotech Res Biochem 2020 3 8 1 7 Harun KI Abdalla MOM El Owni OAO Chemical composition of Mudaffara cheese manufactured by Solanum dubium coat extract and chymosin J Vet Med Anim Prod 2015 6 1 27 37 Jemaa MB Falleh H Neves MA Isoda H Nakajima M Ksouri R Quality preservation of deliberately contaminated milk using thyme free and nanoemulsified essential oils Food Chem 2017 217 726 734 10.1016/j.foodchem.2016.09.030 27664691 Kaptan B Sivri GT Utilization of medicinal and aromatic plants in dairy products J Adv Plant Sci 2018 1 1 6 Lim J Hedonic scaling: a review of methods and theory Food Qual Prefer 2011 22 8 733 747 Mishra AP Devkota HP Nigam M Adetunji CO Srivastava N Saklani S Khaneghah AM Combination of essential oils in dairy products: a review of their functions and potential benefits LWT Food Sci Technol 2020 133 110116 10.1016/j.lwt.2020.110116 Mohammed Salih AM El Sanousi SM El Zubeir IEM A review on the Sudanese traditional dairy products and technology Int J Dairy Sci 2011 6 4 227 245 10.3923/ijds.2011.227.245 Nieto G A review on applications and uses of thymus in the food industry Plants 2020 9 8 961 10.3390/plants9080961 32751488 Ozkan G Simsek B Kuleasan H Antioxidant activities of Satureja cilicia essential oil in butter and in vitro J Food Eng 2007 79 1391 1396 10.1016/j.jfoodeng.2006.04.020 Papetti P Carelli A Composition and sensory analysis for quality evaluation of a typical Italian cheese: influence of ripening period Czech J Food Sci 2013 31 5 438 444 10.17221/447/2012-CJFS Shuang TX Fei L Jian DB Yue ZC Jing M Yun L Jing LX Review on the blending and seasoning technology of the flavored cheeses J Food Saf Qual 2014 5 5 1398 1404 Slman F Sultana A Ali AQ Halabi C Effect of thyme on the quality and shelf life of white Syrian cheese SSRG Int J Agric Environ Sci (SSRG-IJAES) 2020 7 1 36 38 SSMO (2007) Standard for Mudaffara cheese. Sudanese Standard and Metrology Organization. Specification No. 1429/2007. Sudanese Standards and Metrology Organization, Khartoum, Sudan Tamim A Akrama H Chemical composition of ethanolic extracts of T. Syriacus boiss of three sites of Syrian coast by using gas chromatography (GC/MS) Chem Res J 2020 5 4 21 28 Tarakçı Z Deveci F The effects of different spices on chemical, biochemical, textural and sensory properties of white cheeses during ripening Mljekarstvo 2019 69 1 64 77 10.15567/mljekarstvo.2019.0106 Walther B Schmid A Sieber R Wehrmuller K Cheese in nutrition and health Dairy Sci Technol 2008 88 389 405 10.1051/dst:2008012	36474963	PMC9716158	NO-CC CODE	2022-12-03 23:20:54	no	J Food Sci Technol. 2022 Dec 2;:1-11	utf-8	J Food Sci Technol	2,022	10.1007/s13197-022-05634-7	oa_other
==== Front Pediatr Nephrol Pediatr Nephrol Pediatric Nephrology (Berlin, Germany) 0931-041X 1432-198X Springer Berlin Heidelberg Berlin/Heidelberg 5809 10.1007/s00467-022-05809-6 Brief Report Family functioning and quality of life among children with nephrotic syndrome during the first pandemic wave Aman Nowrin F. 12 Fitzpatrick Jessica 13 de Verteuil Isabel 1 Vasilevska-Ristovska Jovanka 12 Banh Tonny Hue Minh 1 Korczak Daphne J. 45 http://orcid.org/0000-0001-6313-5752 Parekh Rulan S. [email protected] 12678 1 grid.42327.30 0000 0004 0473 9646 Child Health and Evaluative Sciences, Research Institute, The Hospital for Sick Children, Toronto, ON M5G 1X8 Canada 2 grid.417199.3 0000 0004 0474 0188 Academics, Women’s College Hospital, Toronto, ON M5S 1B2 Canada 3 grid.266102.1 0000 0001 2297 6811 Department of Medicine, University of California, San Francisco, CA 94143 USA 4 grid.42327.30 0000 0004 0473 9646 Department of Psychiatry, Hospital for Sick Children, Toronto, ON Canada 5 grid.17063.33 0000 0001 2157 2938 Department of Psychiatry, Faculty of Medicine, University of Toronto, Toronto, ON Canada 6 grid.42327.30 0000 0004 0473 9646 Division of Nephrology, The Hospital for Sick Children, 555 University Avenue, Toronto, ON M5G 1X8 Canada 7 grid.17063.33 0000 0001 2157 2938 Department of Pediatrics, Faculty of Medicine, University of Toronto, 27 King’s College Circle, Toronto, ON M5S 1A1 Canada 8 grid.417199.3 0000 0004 0474 0188 Women’s College Hospital, 76 Grenville Street, Toronto, ON M5S 1B2 Canada 2 12 2022 16 27 4 2022 29 9 2022 24 10 2022 © The Author(s), under exclusive licence to International Pediatric Nephrology Association 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Background During the SARS-CoV-2 global pandemic, one of the longest lockdowns worldwide occurred in Ontario, Canada, during the first wave. For parents and children managing care at home and at risk for COVID-19, the impact on their psychosocial functioning is unknown. Methods A total of 122 families of children aged 2–18 years were enrolled as part of the prospective cohort of childhood nephrotic syndrome and completed a survey during the first wave of the pandemic (August 21–December 10), 2020. In a subset, 107 families had data available pre-pandemic to assess change. Validated measures included the McMaster Family Assessment Device (FAD) for parents and children ≥ 12 years for family functioning, the Patient Health Questionnaire for Depression and Anxiety (PHQ-4) for both parent and child, and Pediatric Quality of Life Inventory (PEDSQL™-V4) for children only. Scores were compared using Student’s t-test or the Mann–Whitney U test, as appropriate. Results Among the 107 children, 71% were male with a mean age of 9 years old at the time of questionnaire completion, and the mean age of parents was 41 years old. Parents and children reported that family functioning improved during COVID (parent: p < 0.01; child: p = 0.05). Children’s overall HRQOL declined (p = 0.04), specifically increased sleep disruption (p = 0.01). Increasing child age was associated with a greater sleep disruption (β = − 1.6 [IQR: − 2.6, − 0.67]) and a related decrease in QOL (β = − 1.0 [IQR: − 1.7, − 0.2]), adjusted for sex. Conclusions Despite the positive effects of family dynamics during the first wave, there were negative effects of sleep disruptions and reduced quality of life in children, especially among older children. Graphical abstract A higher resolution version of the Graphical abstract is available as Supplementary information Supplementary Information The online version contains supplementary material available at 10.1007/s00467-022-05809-6. Keywords Nephrotic syndrome Chronic disease Well-being Quality of life COVID-19 Family functioning ==== Body pmcIntroduction In March 2020, the World Health Organization declared a global pandemic of the novel virus, severe acute respiratory syndrome coronavirus (SARS-CoV-2) [1]. During the first wave of the global pandemic, the province of Ontario, Canada, was under 5 months of lockdown with a slow opening through the recovery phase. This time of crisis led to significant stress for people with months of lockdown, isolation, school closures, and strict social distancing measures in place [2–4]. Children with nephrotic syndrome are more susceptible to infections due to the underlying condition as well as the use of immunosuppressives and are more likely to experience relapse episodes during infection [5]. The susceptibility to infection and relapses increased the fears and uncertainty if exposed to COVID-19 infection. Empirical clinical guidelines and recommendations were available for the pandemic care of children with glomerular diseases [5]. Therefore, we aimed to explore the impact of COVID-19 on parental stress and family functioning in children with nephrotic syndrome [6, 7]. Although several studies have assessed the well-being of children and their caregivers during the pandemic, there are limited data on the mental, physical, and family well-being of both parents and children with a chronic disease, such as nephrotic syndrome, potentially at risk for COVID-19 [3, 6, 7]. Among children with nephrotic syndrome, health-related quality of life (HRQOL) is consistently decreased compared to healthy children or for those on steroids or steroid-sparing agents [8]. Thus, we wanted to assess if children’s mental health status, fatigue, and parental well-being worsened during the pandemic [8]. This study is designed to understand the family functioning and parental and child quality of life during the pandemic compared to pre-pandemic assessment. Methods Study design and population Insight into Nephrotic Syndrome: Investigating Genes, Health, and Therapeutics (INSIGHT) is a prospective cohort study of children with nephrotic syndrome, ages 6 months–18 years, conducted at the Hospital for Sick Children in Toronto, Ontario, approved by the Research Ethics Board (ClinicalTrials.gov Identifier: NCT01605266)[9]. Per protocol, families complete questionnaires at baseline and annual follow-ups using the same validated measures of well-being as used in this study [9]. Among the recruited children in INSIGHT, 122 families completed a survey to assess parental and child well-being during the first wave of the COVID-19 pandemic from August 21 to December 10, 2020. In a subset (n = 107 families), we compared the available pre-pandemic data prior to February 29, 2020, and assessed the changes during the pandemic. Questionnaires were completed by emails, phone calls, or during clinic visits. Outcomes Validated measures of well-being for parents included the General Functioning Scale of the FAD and PHQ-4 [10, 11]. A validated measure of well-being in children included the age-specific PEDSQL™-V4 [12]. Family functioning The general functioning subscale of family assessment is a 12-item parent-report scale that measures family functioning. The scale is composed of six positive items that assess healthy family functioning and six items that assess unhealthy functioning [10]. Scoring for each of the items is on a 4-point Likert scale ranging from 1 = strongly agree to 4 = strongly disagree, with the scale for negatively worded items reversed, and the total score is divided by the number of items to provide a score ranging from 1.00 to 4.00, with a cut-off score of < 2.00 indicating better family functioning. Children ≥ 12 years also completed the McMaster Family Assessment Device [10]. Parental anxiety and depression The PHQ-4 is a 4-item composite screening tool for assessing anxiety and depression in parents [11]. Parents were asked to rate their frequency of 2-item anxiety and 2-item depressive symptoms over the past 2 weeks on a 4-point Likert scale (0–3; 0 = not at all, 1 = several days, 2 = more than half the days, and 3 = nearly every day). A cut-off score of ≥ 3 in each anxiety or depression category was used for an indication of mental health conditions [11]. Health-related quality of life in children The PEDSQL™-V4 is a 41-item parent report or self-report measure that assesses children’s HRQOL and fatigue for the past 1 month [12]. It is composed of a 23-item generic core scale comprising physical functioning (8 items), emotional functioning (5 items), social functioning (5 items), and school functioning (5 items), and an 18-item multidimensional fatigue scale comprising three 6-item domains of general fatigue, sleep/rest fatigue, and cognitive fatigue [12]. Responses were recorded on a 5-point Likert scale (0–4; 0 = not at all a problem; 4 = almost always a problem), and items on each domain are reverse scored and transformed to a scale of 0–100 (0 = 100, 1 = 75, 2 = 50, 3 = 25, and 4 = 0), with higher scores indicating better HRQOL and lower fatigue symptoms [12]. For reference, healthy children have a mean score of 83 ± 15.40 [13]. The school functioning domain is not administered during the summer months as per standard protocol. Data analyses Parent and child characteristics were summarized using descriptive statistics, and well-being scores were compared pre- and post-COVID-19 outbreak using the Student’s t-test or the Mann–Whitney U test, as appropriate. We examined the association of demographic characteristics with PEDSQL scores that were found to be statistically different by t-test or Mann–Whitney U test using linear regression. Results Among the 107 children, 71% were male; the median age at diagnosis was 3.4 years old, and the mean age at survey completion was 9 years old (Supplementary Table S1). Among the parents (n = 107), 77% were employed, and the mean parental age was 41 years old. The median time between the annual questionnaire completed before the pandemic and the COVID-19 survey during the pandemic was 16.9 months. Parents and children reported that family functioning during COVID improved (Fig. 1A and B) compared with pre-COVID (parent 2.03 ± 0.19 vs. 1.65 ± 0.45; p < 0.01; child 2.10 ± 0.21 vs. 1.80 ± 045; p = 0.05). There was no difference in family functioning by age or sex of children. The parent’s depression and anxiety level were affected during COVID but did not show any significant difference.Fig. 1 Assessment of family functioning and quality of life before and during wave 1 of the pandemic. Family functioning was assessed by the General Functioning Scale of McMaster Family Assessment Device by parents (A) and children ≥ 12 years (B), with a cut-off score of < 2.00 indicating better family functioning. Quality of life was assessed in children by total PEDSQL fatigue score (C) and PEDSQL psychosocial summary score (D). The total fatigue score is calculated as the mean of all the domains in the multidimensional fatigue scale, and the PEDSQL psychosocial summary is the mean of emotional, social, and school functioning domains. Quality of life assessed in children by total PEDSQL scores by age (E) and by sex (F) of children Children’s overall HRQOL declined (p = 0.04), specifically increased sleep disruption and decreased energy (p = 0.01) during the pandemic. All other domains were stable across study time points. The total fatigue score (p = 0.09) and psychosocial summary score (p = 0.70) did not show any significant difference (Fig. 1, panels C and D), regardless of medication status (Supplementary Table S2). Further analysis showed that increasing child age was associated with greater sleep disruption (β = − 1.6 [IQR: − 2.6, − 0.67]) and a decrease in overall QOL (β = − 1.0 [IQR: − 1.7, − 0.2]), adjusted for sex. Discussion Our results highlight that family functioning improved in families with childhood nephrotic syndrome during the first wave in Ontario, in contrast to pre-pandemic assessment, with no significant change in parental anxiety and depression level. The pandemic, however, did impact the overall quality of life in children, with significant sleep disruption, especially among older children. Improved family dynamics and cohesion, as reported by families and older children, could perhaps be attributed to families spending more time at home together, engaging in family activities, and better communication [14]. These results presented are supported by comments made in the survey by participating families, such as “It is been summer holidays for us, so our family has had a very positive time this last 2 months.” Although parental mental health status has been negatively impacted during the pandemic, as reported in several studies, our study does not show any significant levels of parental stress and anxiety [6]. Overall HRQOL had significantly declined in children, although no physical or psychosocial domains other than sleep fatigue were significantly impacted over the pandemic with lockdown measures in place. This adds to the existing evidence of already reduced scores in HRQOL in children with nephrotic syndrome receiving steroid and steroid-sparing treatments [8]. We demonstrated that children experience greater sleep disruption, related to difficulty maintaining sleep, daytime tiredness, spending more time in bed, and decreased energy, possibly in the context of loss of routines, reduced outdoor activities, virtual schooling, more screen time, and fear of COVID-19 infection [15]. We specifically found that decreased HRQOL and sleep disruption are seen more among older children, which could be related to stress, change in routine, and increased screen time [15]. Single parent–reported responses, a small sample size of a selected group of participants, limited online survey completion, and reporting bias are some limitations of this study. Nonetheless, the study results give a unique opportunity to highlight the importance of studying both the potential positive impact on family functioning as well as negative impacts on the overall quality of life and sleep disruption during the pandemic. Our results highlight the need to develop effective strategies to improve the overall mental health and well-being of both children with nephrotic syndrome and their parents. Supplementary Information Below is the link to the electronic supplementary material.Graphical Abstract (PPTX 153 KB) Supplementary file2 (DOCX 21 KB) Acknowledgements We thank the participants and their families for their time and effort, as well as the nurses and staff from the Nephrology Clinic at The Hospital for Sick Children. Author contribution Nowrin F. Aman conceptualized and designed the study, collected data, drafted the initial manuscript, interpreted the data, and critically reviewed and revised the manuscript. Jessica Fitzpatrick-Collins carried out the data analyses and critically reviewed and revised the manuscript. Isabel de Verteuil conceptualized and designed the study, designed the data collection instruments, collected data, carried out the initial analyses, and critically reviewed and revised the manuscript. Jovanka Vasilevska-Ristovska conceptualized and designed the study, designed the data collection instruments, interpreted the data, and critically reviewed and revised the manuscript. Tonny Hue Minh Banh designed the data collection instruments, analyzed the data, interpreted the data, and critically reviewed and revised the manuscript. Daphne J Korczak conceptualized and designed the study, coordinated and supervised data collection, interpreted the data, and critically reviewed the manuscript for important intellectual content. Rulan S. Parekh conceptualized and designed the study, coordinated and supervised data collection, interpreted the data, and critically reviewed and revised the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request. Declarations Ethics approval The questionnaire and methodology for this study were approved by the Research Ethics Board of the Hospital for Sick Children, in Toronto, Ontario (ethics approval number: 1000021384). Consent to participate Informed consent was obtained from all parents/legal guardians of individual participants included in the study. Consent for publication The authors affirm that human research participants provided informed consent for the publication of the data. Competing interests The authors declare no competing interests. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. World Health Organization (WHO) (2020) Coronavirus disease (COVID-2019) situation reports. https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports. Accessed July 14, 2020 2. Brooks SK Webster RK Smith LE Woodland L Wessely S Greenberg N Rubin GJ The psychological impact of quarantine and how to reduce it: rapid review of the evidence Lancet 2020 395 912 920 10.1016/S0140-6736(20)30460-84 32112714 3. Fegert JM Vitiello B Plener PL Clemens V Challenges and burden of the Coronavirus 2019 (COVID-19) pandemic for child and adolescent mental health: a narrative review to highlight clinical and research needs in the acute phase and the long return to normality Child Adolesc Psychiatry Ment Health 2020 14 20 10.1186/s13034-020-00329-3 32419840 4. Viner R Russell S Saulle R Croker H Stansfield C Packer J Nicholls D Goddings AL Bonell C Hudson L Hope S Ward J Schwalbe N Morgan A Minozzi S School closures during social lockdown and mental health, health behaviors, and well-being among children and adolescents during the first COVID-19 wave: a systematic review JAMA Pediatr 2022 176 400 409 10.1001/jamapediatrics.2021.5840 35040870 5. Robinson C Ruhl M Kirpalani A Alabbas A Noone D Teoh CW Langlois V Phan V Lemaire M Chanchlani R authors of the CCRRT recommendations for adults with glomerulonephritis Management of Canadian pediatric patients with glomerular diseases during the COVID-19 pandemic: recommendations from the Canadian Association of Pediatric Nephrologists COVID-19 rapid response team Can J Kidney Health Dis 2020 7 2054358120970713 10.1177/2054358120970713 33240518 6. Ademhan Tural D Emiralioglu N Tural Hesapcioglu S Karahan S Ozsezen B Sunman B Nayir Buyuksahin H Yalcin E Dogru D Ozcelik U Kiper N Psychiatric and general health effects of COVID-19 pandemic on children with chronic lung disease and parents’ coping styles Pediatr Pulmonol 2020 55 3579 3586 10.1002/ppul.250827 32946202 7. Serlachius A Badawy SM Thabrew H Psychosocial challenges and opportunities for youth with chronic health conditions during the COVID-19 pandemic JMIR Pediatr Parent 2020 3 e23057 10.2196/23057 33001834 8. Khullar S Banh T Vasilevska-Ristovska J Chanchlani R Brooke J Licht CPB Reddon M Radhakrishnan S Piekut M Langlois V Aitken-Menezes K Pearl RJ Hebert D Noone D Parekh RS Impact of steroids and steroid-sparing agents on quality of life in children with nephrotic syndrome Pediatr Nephrol 2021 36 93 102 10.1007/s00467-020-04684-3 32671615 9. Hussain N Zello JA Vasilevska-Ristovska J Banh TM Patel VP Patel P Battiston CD Hebert D Licht CP Piscione TD Parekh RS The rationale and design of insight into nephrotic syndrome: investigating genes, health and therapeutics (INSIGHT): a prospective cohort study of childhood nephrotic syndrome BMC Nephrol 2013 14 25 10.1186/1471-2369-14-25 23351121 10. Byles J Byrne C Boyle MH Offord DR Ontario Child Health Study: reliability and validity of the general functioning subscale of the McMaster Family Assessment Device Fam Process 1988 27 97 104 10.1111/j.1545-5300.1988.00097.x 3360100 11. Kroenke K Spitzer RL Williams JB Löwe B An ultra-brief screening scale for anxiety and depression: the PHQ-4 Psychosomatics 2009 50 613 621 10.1176/appi.psy.50.6.61312 19996233 12. Varni JW Seid M Kurtin PS PedsQL 4.0: reliability and validity of the Pediatric Quality of Life Inventory version 4.0 generic core scales in healthy and patient populations Med Care 2001 39 800 812 10.1097/00005650-200108000-00006 11468499 13. Varni JW Limbers CA Burwinkle TM Impaired health-related quality of life in children and adolescents with chronic conditions: a comparative analysis of 10 disease clusters and 33 disease categories/severities utilizing the PedsQL 4.0 generic core scales Health Qual Life Outcomes 2007 5 43 10.1186/1477-7525-5-43 17634123 14. Tam W Poon SN Mahendran R Kua EH Wu XV Impacts of COVID-19 and partial lockdown on family functioning, intergenerational communication and associated psychosocial factors among young adults in Singapore BMC Psychiatry 2021 21 595 10.1186/s12888-021-03599-z 34836516 15. Dondi A Fetta A Lenzi J Morigi F Candela E Rocca A Cordelli DM Lanari M Sleep disorders reveal distress among children and adolescents during the COVID-19 first wave: results of a large web-based Italian survey Ital J Pediatr 2021 47 130 10.1186/s13052-021-01083-8 34088351	36459245	PMC9716160	NO-CC CODE	2022-12-03 23:20:54	no	Pediatr Nephrol. 2022 Dec 2;:1-6	utf-8	Pediatr Nephrol	2,022	10.1007/s00467-022-05809-6	oa_other
==== Front Nat Hazards (Dordr) Nat Hazards (Dordr) Natural Hazards (Dordrecht, Netherlands) 0921-030X 1573-0840 Springer Netherlands Dordrecht 5731 10.1007/s11069-022-05731-y Original Paper Effect of individual characteristics, risk perception, self-efficacy and social support on willingness to relocate due to floods and landslides http://orcid.org/0000-0002-3751-131X Mızrak Sefa [email protected] 1 http://orcid.org/0000-0002-0588-2191 Turan Melikşah [email protected] 2 1 grid.448936.4 0000 0004 0369 6808 Department of Emergency Aid and Disaster Management, Faculty of Health Sciences, Gümüşhane University, Gümüşhane, Turkey 2 grid.448691.6 0000 0004 0454 905X Department of Emergency Aid and Disaster Management, Faculty of Health Sciences, Erzurum Technical University, Erzurum, Turkey 2 12 2022 123 15 6 2022 22 11 2022 © The Author(s), under exclusive licence to Springer Nature B.V. 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. People may have to leave their home, environment, region and country because of disasters or disaster risks. Effective and efficient disaster risk reduction activities involving the community can reduce disaster risks and enable people to reside more safely and peacefully in their environment. The objective of this study was to investigate whether individual characteristics, risk perception, self-efficacy and perceived social support were correlated with the willingness to relocate due to floods and landslides. The data were collected from 947 people residing in Gümüşhane Province (Türkiye) using a survey. In the study, a total of ten models were tested with the help of ordinal logistic regression analysis. Consequently, the participants' willingness to relocate due to landslides was determined to be higher than the willingness to relocate due to floods. University students and people with chronic diseases and flood and landslide experiences had a greater willingness to relocate. Residence duration and informal social support were negatively correlated with relocation willingness. Those who believed that they could protect themselves in the event of a flood and landslide were more likely to relocate. Among risk perceptions, probability increased relocation willingness mostly due to floods, while fear increased relocation willingness mostly due to landslides. This study attempted to provide policy makers and scientists insight into disaster risk reduction and disaster risk communication related to relocation. Keywords Risk perception Self-efficacy Social support Relocation willingness ==== Body pmcIntroduction Natural disasters endanger the sustainability of the community and the environment by causing death, injury, and economic, social, cultural and psychological damage. Flood, which was the most common type of natural disaster in the world between 2000 and 2019, affected 1.6 billion people and caused 104 614 deaths (United Nations Office for Disaster Risk Reduction 2020). Between 1994 and 2014, a total of 3876 landslides occurred in 128 countries, and these landslides caused 177 536 casualties (Haque et al. 2019). In addition, due to the possibility of a disaster or disasters that may occur, people may have to leave their home, environment, region or country permanently or temporarily in the long or short term and relocate in other settlements. The International Displacement Monitoring Centre (2020) reported that 23.9 million people were displaced by weather-related disasters in 140 countries and territories in 2019. Badong County, China, was relocated three times due to natural hazards (Gong et al. 2021). The study conducted in two regions affected by the earthquake found that three-quarters of the participants wanted to relocate because of earthquakes (Xu et al. 2020). Relocation is a risk coping behavior (Qing et al. 2022), investment in resilience (Pinter and Rees 2021), adaptation strategy (Seebauer and Winkler 2020) and risk reduction policy (Thaler and Fuchs 2020). However, people who have to relocate due to disasters encounter social, economic, mental and physical problems. For example, the 2011 Great East Japan Earthquake and Tsunami worsened the socioeconomic, mental and physical conditions of the displaced people (Takahashi et al. 2016). The unsuccessful and unplanned management of forced relocations caused by a typhoon in Taiwan adversely affected the security, housing, agricultural activities, occupations and emotional adjustment of the community (Taiban et al. 2020). Post-disaster relocation may cause deterioration of family structure and negative behaviors of family members toward each other (Samonte 2021). After Hurricane Katrina, the trauma symptoms of children who were permanently relocated were higher than those of children who returned to their previous residence (Hansel et al. 2013). Financial resources, such as government aid, private aid and insurance payments, are also used to accelerate recovery in the relocation process (Thaler and Fuchs 2020). Post-flood relocation in Iran reduced production and employment and increased the debts of the relocated people (Garakani et al. 2020). Relocation needs of people due to disasters can be eliminated by reducing vulnerability to disasters, and as a result, the social and cultural structure of the community can be protected. Scientific studies were conducted in order to remove and reduce the problems encountered during and after the relocation process. Garakani et al. (2020) investigated the impact of relocation on production, economy, social structure, family order, environment, population and urbanization in the region where eleven villages were relocated and in three villages that were relocated to their previous locations after the devastating flood in Iran. Kılıc et al. (2006) investigated the effects of earthquake experience and socioeconomic factors on traumatic stress and depression in people who moved to a low-earthquake-risk region due to the 1999 Marmara Earthquake. Iuchi and Mutter (2020) comparatively reviewed the social and economic impacts of relocation strategies adopted after earthquake and tsunami, storm surge and volcanic eruption. Sipe and Vella (2014) evaluated policies applied to the post-flood relocation community in Australia in terms of best practices and argued whether practices for relocation could be adapted to a situation of the same or larger scale. As a case study, stakeholders and people affected by the flood were interviewed about the strategies implemented throughout the process for the households relocated due to the floods in Germany (Mayr et al. 2020). The factors affecting people's thoughts and attitudes about relocation were investigated through semi-structured interviews, focus group discussions and questionnaires in landslide-prone areas in Cameroon (Baert et al. 2020). Rey-Valette et al. (2019) investigated the effect of socio-demographic characteristics, life satisfaction, early warning awareness, home location and home ownership on resistance against relocation in flood-prone coastal areas via a composite index combining place attachment, housing mobility and risk perception. Rashid et al. (2007) examined whether people residing in slums in areas with high flood risk would relocate to areas without flood risk in exchange for employment, new land, loans and grants. It is important for disaster risk reduction to reveal people's perceptions and vulnerabilities toward relocation and the factors that affect the relocation decision. This study aimed to reveal the factors affecting people's willingness to relocate due to floods and landslides. Ordinal logistic regression analysis revealed whether individual characteristics, risk perception, self-efficacy and social support were correlated with the willingness to relocate. Except for individual characteristics, all other variables were measured separately for floods and landslides. This study aimed to contribute to disaster resilience and risk reduction policies by investigating the relocation willingness of people in the region where there is a high risk of flood and landslide, as well as rapid population growth and urbanization. In addition, it explored how and to what extent individual characteristics, risk perception, social support and self-efficacy variables were correlated with people's attitudes toward both flood and landslide risk, thus providing a better understanding of the type of disaster and the relationship between these variables. The results may benefit scientists and decision makers who contribute to disaster risk reduction, disaster preparedness and relocation policies. Theoretical background, research model and hypotheses The independent variables utilized in this study were created based on the results of empirical studies. This study tested four hypotheses to understand which factors were correlated with relocation willingness due to both floods and landslides. The variables in the models tested in this study are presented in Fig. 1.Fig. 1 Research model Individual characteristics, such as gender, marital status, income, age and education, are important variables that enable people to cope with disasters. Scientists mostly investigated the effects of education, income, disaster experience and occupation variables among individual characteristics that affect community resilience to disasters (Cai et al. 2018). Gender, demographic characteristics, economic status and disability and the special needs of individuals were the main indicators that affect social vulnerability to disasters (Fatemi et al. 2017). Therefore, some groups are considered as vulnerable groups to disasters. For example, women's disaster resilience is lower than men's disaster resilience (Drolet et al. 2015). Marital status (Tekeli-Yeşil et al. 2011) and number of children (Ozdemir and Yilmaz 2011) affect people's attitude toward disaster risks. People with high income (Strömberg 2007) and education (Garbero and Muttarak 2013; Ma et al. 2021) levels have a higher capacity to cope with disasters. People experiencing floods have higher awareness about floods and care more about flood risks (Bera and Daněk 2018). People with chronic diseases have a low level of disaster preparedness (Bethel et al. 2011) and confront a lot of problems after disasters because of their chronic diseases (Jhung et al. 2007). Using data from 175 countries between 1960 and 2015, a study reported that disasters caused more deaths and economic damage in countries with high unemployment rates (Tselios and Tompkins 2019). Compared to young people, older people need more help after disasters because of their physical and mental problems (Ahmadi et al. 2018; Malak et al. 2020). University students constitute a highly vulnerable part of community (Mızrak and Aslan 2020). Studies conducted in regions with high disaster risk found that there was a positive relationship between residence duration and sense of place (Anton and Lawrence 2014; Anacio et al. 2016). Moreover, scientists scrutinized whether individual characteristics were correlated with the willingness to relocate due to disasters (Rashid et al. 2007; Vlaeminck et al. 2016; Xu et al. 2017; Chao 2017; Shao et al. 2017; Song and Peng 2017; Rey-Valette et al. 2019; Baert et al. 2020; Holley et al. 2022). Furthermore, some scientists employed individual characteristics as control variables while investigating the factors affecting the willingness to relocate due to disasters (Xu et al. 2017; Chao 2017; Shao et al. 2017; Zhou et al. 2021). Based on these studies, Hypothesis 1 was proposed. Hypothesis 1 (H1) Vulnerable groups will be more likely to relocate due to floods and landslides (women, married people, people with low income and education levels, families with children, those with disaster experiences and chronic diseases, the unemployed, university students, the elderly and people residing in a high-risk area for less time). Risk perception is an important factor that affects people's behaviors and thoughts before, during and after a disaster. People with high flood and hurricane risk perception were more supportive of public policies for relocation and evacuation (Shao et al. 2017). Xu et al. (2017) found that risk perception was the strongest variable that increased the willingness to relocate due to landslide hazard, and suggested that people's perceptions of probability and threat to disasters should be understood in order to reduce problems between government and the public when establishing relocation policies. The possibility of landslides occurrence in Uganda increased household heads' willingness to relocate (Vlaeminck et al. 2016). Individuals with low-risk perception residing in high-landslide-hazard areas in Bangladesh had less preparedness and willingness to settle in a safer area (Alam 2020). People with continuous negative psychological effects of earthquake and people who felt sensitive to earthquake tended to evacuate more (Ao et al. 2020). People who did not feel safe in their homes in New Orleans against disaster were more likely to evacuate (Burnside et al. 2007). People's perception of disaster risk may vary according to the type of disaster, and people's feelings, thoughts, behaviors and approaches to disasters may also affect their relocation decisions differently. Therefore, in order to better understand the impact of disaster risk perception on relocation willingness, risk perception should be comprehensively addressed. In this study, Hypothesis 2 was proposed to test whether the perception of flood and landslide risk was correlated with the willingness to relocate. Hypothesis 2 (H2) The four dimensions of risk perception (severity, possibility, fear and uncontrollable) will be significantly and positively correlated with the willingness to relocate due to floods and landslides. Meta-analysis studies revealed that high perceived self-efficacy was a strong motivational factor that increased adaptation behaviors toward climate change (van Valkengoed and Steg 2019) and floods (Bamberg et al. 2017). Samaddar et al (2014) found that people with high self-efficacy to cope with floods were more confident in their own abilities and capacities and had more flood preparedness intention. As the self-efficacy of people living in flood-prone areas increased, their flood damage mitigation measures also increased (Botzen et al. 2019). Self-efficacy was positively and significantly correlated with the willingness to participate in adaptive measures related to geological hazards (Hu et al. 2022). The self-efficacy of earthquake survivors was negatively and significantly correlated with the level of post-traumatic stress disorder (Guerra et al. 2014). While the self-efficacy belief of people residing in earthquake-prone area was not significantly correlated with their relocation intention, it was positively and significantly correlated with their evacuation intention (Qing et al. 2022). Self-efficacy increases people's resilience to disasters, and people with high self-efficacy against any disaster may want to stay in the area they live in because they think that they can protect themselves in case of a disaster. Hypothesis 3 was proposed to reveal the effect of self-efficacy on the willingness to relocate due to floods and landslides. Hypothesis 3 (H3) Self-efficacy will be significantly and negatively correlated with the willingness to relocate due to floods and landslides. Strong social relationships in the community facilitate the reduction of the effects of disasters. Perceived financial support decreased the depression and anxiety of the victims affected by the 2013 flood in Germany (Daniel and Michaela 2021). Social support increased the resilience level and quality of life of earthquake survivors (Xu and Ou 2014). Babcicky and Seebauer (2020) found that social support decreased the fear of floods and increased self-efficacy against flood risks. The study conducted on the people affected by the 2010 Chile earthquake revealed that expected support was positively and significantly correlated with the collective efficacy (Drury et al. 2016). The greater number of reliable relatives and membership in informal groups reduced the household’s post-flood recovery speed in central Vietnam (Dinh et al. 2021). A qualitative study conducted in disaster-prone communities reported that external support from government and other partners was an effective strategy to reduce the damage of landslides and floods (Osuret et al. 2016). Xu et al. (2017) investigated whether social support, measured by the number of people who could receive financial aid, was significantly and negatively correlated with the willingness to relocate in the region under the threat of landslides. Social support from people or institutions enables people to better prepare for disasters and to recover from the effects of disasters quickly. Individuals may not want to move from a disaster-prone area because they think that they will be less affected by disasters thanks to social support. Based on this, Hypothesis 4 was proposed. Hypothesis 4 (H4) Informal and formal social support will be significantly and negatively correlated with the willingness to relocate due to floods and landslides. Methods Study area Gümüşhane province, located in the Eastern Black Sea region of Türkiye, is on a deep valley in a mountainous land, and the majority of settlement is on the mountain slopes and by a river called the Harşit River flowing through the city (Figs. 2, 3a–e). The central population of Gümüşhane was 37 856 in 2008 and 54 108 in 2021 (Turkish Statistical Institute 2022). Gümüşhane ranks 64th among the 81 provinces in Türkiye according to the social and economic development index calculated with 52 variables such as population, education, employment, health, accessibility and quality of life (Acar et al. 2019). On the one hand, Gümüşhane University, which was established in 2008, has made a great contribution to the development of the province. On the other hand, the university has led to an increase in the population of the province and rapid urbanization.Fig. 2 Türkiye disaster risk map (Ministry of Interior of the Republic of Türkiye 2022). Note: The red color on the map indicates the earthquake hazard Fig. 3 Study area The mountainous terrain and climatic conditions in the province make human life difficult. There are many tunnels on the entrance and exit roads of the province, and the traffic flow on the highways and in the center is disrupted due to rockfalls and landslides. Many retaining walls have been built in the city center to prevent landslides and rockfalls, and there are wire fences in some places. Rain-induced flooding causes disruption of transportation on the highway, which is the only one in many parts of the city (Fig. 3e). Gümüşhane has ground and environmental problems for building construction (Tudes et al. 2012). Besides, due to rapid urbanization in Gümüşhane, excavations for building construction trigger landslides (Alemdag et al. 2014; Kaya et al. 2016). Moreover, structures on the banks of the Harşit river are highly vulnerable to floods (Fig. 3d). Furthermore, the risk of flood, landslide and forest fire is high in the provinces around Gümüşhane (Fig. 2). The Gümüşhane Disaster and Emergency Directorate reported that Gümüşhane was the 21st province with the highest number of disasters among the 81 provinces in Türkiye. The most common disasters in the province are landslides (%49), rockfalls (%38), floods (%9) and avalanches (%3). In Gümüşhane, 39 132 residences, 1573 public buildings and 838 workplaces are located in a very high- and high-landslide-risk area. The Gümüşhane Disaster and Emergency Directorate requested 3,337,000 Turkish Liras from the national disaster management authority to cover the damage of 52 floods between 2016 and 2021 (Gümüşhane Disaster and Emergency Directorate 2021). Participants In the study, the data were collected through a survey using the convenience sampling method in March 2022. In the convenience sampling method, which is one of the nonprobability sampling strategies, people who are easily accessible and close are selected to save time and money. On the other hand, it is a disadvantage that the results of studies using this sampling method cannot be generalized to the whole study area (Bornstein et al. 2013). The COVID-19 outbreak was ongoing at the time of the data collection; therefore, this sampling method was preferred to protect both researchers and participants because the number of people who could be in every workplace, public institution and open spaces at the same time was limited in order to prevent the spread of the epidemic and fines were imposed on those who did not comply with the rules. In addition, people did not allow strangers to enter their homes due to fear of illness. This situation made it difficult to reach the participants, and therefore, suitable areas and people were preferred to administer the survey. The survey was first applied online through social media platforms, and the participants were asked to send the survey link to other people residing in Gümüşhane. However, only 433 people participated in the study using the online method. For population sizes of 50,000 and 75,000, a sample size of 381–382 is sufficient (Krejcie and Morgan 1970). Since the convenience sampling method was used in order to increase the scope of the results, it was desired to reach more participants. Five trained interviewers and the authors of this study reached people on the streets and in the workplaces and administered the survey to 514 people. Finally, a total of 947 people voluntarily participated in this research. Instrument The survey used in this study consists of three parts. The first part of the survey determined individual characteristics, while the second part related to floods and the third part related to landslides measured risk perception, self-efficacy, social support and relocation willingness. The individual characteristics included gender, age, marital status, monthly income, number of children, education, chronic disease, occupation, residence duration in the province, flood experience and landslide experience. Risk perception, self-efficacy, social support and relocation willingness were determined separately for floods and landslides with Likert type scales ranging 0–4 (see Table 2). Flood and landslide risk perception was measured via severity, possibility, fear and uncontrollable sub-dimensions, using studies investigating disaster risk perception (Bubeck et al. 2012; Xu et al. 2016, 2020; Peng et al. 2017, 2019). Self-efficacy addressed according to previous studies (Samaddar et al. 2014; Babcicky and Seebauer 2020; Yu et al. 2022) was determined according to the level of knowledge that people perceive to protect themselves in case of a flood and landslide. Questions measuring social support were formed according to the approaches of studies investigating the relationship between disaster and social support (Chao 2017; Xu et al. 2017; Yu et al. 2022). Perceived social support in the face of flood or landslide damage was determined as formal and informal social support. The willingness to relocate was determined by how often people thought of moving from their city to another place due to floods and landslides. Data analysis The data were analyzed with the help of the SPSS program. The frequency and percentage values of the categorical variables and the mean and standard deviation (SD) values of the continuous variables were presented. The willingness to relocate due to floods and landslides was employed as dependent variables, and the relationships between the dependent and independent variables (individual characteristics, risk perception, self-efficacy, social support) were revealed by ordinal logistic regression analysis. Ordinal logistic regression was constructed since the dependent variables were an ordered multiclassification variable and the independent variables were both categorical and continuous variables. A total of 10 models were tested in this study. Before the regression analysis, the control of multicollinearity problem among the independent variables was checked with variance inflation factors (VIF) and tolerance values. In order to avoid multicollinearity problem among the independent variables, the VIF value should be below 4 (O’brien 2007) and the tolerance value should be above 0.2 (Hosmer et al. 2008). All VIF values below 4 and all tolerance values above 0.2 in this study showed that there was no multicollinearity problem for regression analyses. First, in Model 1–5, the effect of the individual characteristics, flood risk perception, self-efficacy and social support sub-dimensions on the willingness to relocate due to floods were tested. Second, in Model 6–10, the effects of the individual characteristics, landslide risk perception, self-efficacy and social support sub-dimensions on the willingness to relocate due to landslides were tested. After creating a separate model for each set of independent variables (Model 1, 2, 3, 4, 6, 7, 8, 9), all independent variables were included in the final models (Model 5, 10). The results of the analyses were interpreted with estimates, standard errors and R square (R2). For the education and occupation variable, undergraduate or higher and others options were taken as reference categories, respectively, and the relocation willingness of the other groups were compared according to these reference categories. The other independent variables were interpreted according to the estimation values indicating the change in the dependent variable caused by a unit increase or decrease in the independent variable. Statistical significance levels of .05, .01 and .001 were used to indicate the 95%, 99% and 99.9% confidence levels of the regression results, respectively. Ethical consideration People participated in this study voluntarily and without expecting anything in return. The Scientific Research and Publication Ethics Committee of Gümüşhane University approved the method and rationale of this research scientifically and ethically on February 2022. Results Descriptive statistical analysis Table 1 presents the characteristics of the individuals participating in this study. 50.7% of the participants were male and 35.5% were married. According to the income group, the middle-income group was the highest with 52% and the least participant was in the very high-income group with 4%. 68.6% of individuals did not have children and 60.7% were university graduates. While 93.5% of them had no flood experience, 91.9% had no landslide experience. 82.7% of them did not have a chronic disease and 38.1% were university students. The oldest participant was 65 years old, and the participant with the longest residence duration in the province had been living in the province for 63 years.Table 1 Individual characteristics of participants Variable Groups Frequency Percent Gender Male 480 50.7 Female 467 49.3 Marital status Single 611 64.5 Married 336 35.5 Income Very low 117 12.4 Low 170 18.0 Middle 492 52.0 High 130 13.7 Very high 38 4.0 Number of children No child 650 68.6 1 child 109 11.5 2 child 143 15.1 More than 2 45 4.8 Education level Literate 31 3.3 Primary school 40 4.2 Elementary 43 4.5 High school 258 27.2 Pre-undergraduate 228 24.1 Undergraduate or above 347 36.6 Flood experience No 885 93.5 Yes 62 6.5 Landslide experience No 870 91.9 Yes 77 8.1 Chronic disease No 783 82.7 Yes 164 17.3 Occupation Unemployed 72 7.6 University student 361 38.1 Self-employed 120 12.7 Officer 174 18.4 Tradesmen 65 6.9 Others 155 16.4 Variable Minimum Maximum Mean SD Age 18 65 29.15 9.04 Residence duration (years) 1 63 14.92 14.2 Table 2 demonstrates the survey questions and their means and standard deviations. The ranking of the flood risk perception sub-dimensions from the highest mean to the lowest mean was severity, uncontrollable, possibility and fear. For the landslide risk perception, this ranking was from highest to lowest severity, probability, uncontrollable and fear. The perceived self-efficacy and social support for floods were higher than the perceived self-efficacy and social support for landslides. The participants' willingness to relocate due to landslides was higher than their willingness to relocate due to floods. Generally speaking, for both floods and landslides, the participants' perceptions of risk, self-efficacy and social support were moderate, while their willingness to relocate was low.Table 2 Survey questions, means and standard deviations Variables related to flood Mean SD Risk perception Severity = If a flood occurred in Gümüşhane, how much would the damage be? 2.50 .95 Possibility = What is the possibility of a flood that can cause a disaster in Gümüşhane within 2 years? 1.97 .99 Fear = How much does the possibility of flooding in Gümüşhane frighten you? 1.83 1.26 Uncontrollable = How adequate are the measures taken to prevent floods that may cause disasters in Gümüşhane? 2.25 1.07 Self-efficacy = Do you have enough knowledge to protect yourself in the event of a flood? 1.91 1.14 Social support Informal Social Support = If you were damaged by a flood in Gümüşhane, how much would people around you help you? 2.26 1.16 Formal Social Support = If you were damaged by a flood in Gümüşhane, how much would public institutions help you? 2.16 1.12 Relocation Willingness = Are you considering moving from Gümüşhane because of flood? 1.42 1.35 Variables related to landslide Mean SD Risk perception Severity = If a landslide occurred in Gümüşhane, how much would the damage be? 2.62 1.09 Possibility = What is the possibility of a landslide that can cause a disaster in Gümüşhane within two years? 2.33 1.04 Fear = How much does the possibility of landslide in Gümüşhane frighten you? 2.05 1.23 Uncontrollable = How adequate are the measures taken to prevent landslides that may cause disasters in Gümüşhane? 2.20 1.05 Self-efficacy = Do you have enough knowledge to protect yourself in the event of a landslide? 1.86 1.15 Social support Informal Social Support = If you were damaged by a landslide in Gümüşhane, how much would people around you help you? 2.24 1.13 Formal Social Support = If you were damaged by a landslide in Gümüşhane, how much would public institutions help you? 2.15 1.11 Relocation Willingness = Are you considering moving from Gümüşhane because of landslide? 1.53 1.36 Response options: Severity (0 = No damage, 4 = There would be a lot of damage), Possibility (0 = Very low, 4 = Very high), Fear (0 = It does not fright at all, 4 = It frightens a lot), Uncontrollable (0 = Very adequate, 4 = Very inadequate), Self-efficacy (0 = I have no knowledge, 4 = I have a lot of knowledge), Social support (0 = They never help, 4 = They help a lot), Relocation willingness (0 = Never, 4 = Always) Model results Table 3 shows the results of the models (Model 1–5) that tested the effect of the independent variables on the willingness to relocate due to floods. According to Model 1, which shows the effect of individual characteristics on the dependent variable, gender, marital status, income, number of children and age were not significantly correlated with the willingness to relocate due to floods. Literate people and primary and secondary school graduates were more likely to relocate due to floods compared to those with undergraduate or higher education levels. Flood experience and chronic disease were positively and significantly correlated with the willingness to relocate because of floods. Compared to other occupational groups, only being a university student among occupational groups was positively and significantly correlated with the willingness to relocate because of floods. Residence duration was negatively and significantly correlated with the willingness to relocate because of floods. Severity was not significantly correlated with the dependent variable, and the strongest flood risk perception that was positively and significantly correlated with the dependent variable was possibility, fear and uncontrollable, respectively (Model 2).Table 3 Ordinal logistic regression results of the effect of independent variables on willingness to relocate due to floods Independent variable Model 1 Model 2 Model 3 Model 4 Model 5 Gender − .151 (.126) − .113 (.133) Marital status − .029 (.203) − .213 (.216) Income .033 (.071) .090 (.074) Number of children .163 (.11) .292 (.117)* Education Literate .929 (.35)** .840 (.358)* Primary school .977 (336)** .697 (.347)* Elementary .916 (.316)** .485 (.333) High school .198 (.162) .273 (.168) Pre-undergraduate .155 (.16) .208 (.166) Undergraduate or abovea Flood Experience .632 (.25)* .024 (.258) Chronic Illness .596 (.165)* .448 (.170) Occupation Unemployed .379 (.268) .268 (.280) University Student .738 (.226) .628 (.237) Self-employed .18 (.226) − .136 (.241) Officer − .039 (.223) − .141 (.233) Tradesmen − .058 (.28) − .212 (.290) Othersa Age − .01 (.011) − .020 (.012) Residence duration − .019 (.006)** − 015 (.007)* Risk perception = Severity − .029 (.079) − .034 (.082) Risk perception = Possibility .671 (.08)* .667 (.084)* Risk perception = Fear .306 (.058)* .323 (.061)* Risk perception = Uncontrollable .160 (.06) .118 (.063) Self-efficacy .302 (.052)* .271 (.060)*** Informal social support − .132 (.059)* − .198 (.065)** Formal social support − .093 (.061) − .163 (.066)* Pseudo-R square Cox and Snell .10 .21 .03 .01 .31 Nagelkerke .11 .23 .03 .01 .33 McFadden .03 .08 .01 .005 .12 Model fitting information − 2 Log Likelihood 2585.628 1428.292 153.519 470.202 2490.847 Chi-square 108.926* 234.384* 31.706* 13.411 360.575*** a = Reference category. Robust standard errors in parentheses. Outside the parentheses are the estimates. Results significant at the p < .05; p < .01; *p < .001 levels Model 3 stated that self-efficacy against floods was positively and significantly correlated with the willingness to relocate. The perceived informal social support was negatively and significantly correlated with the willingness to relocate due to floods, whereas formal social support was not significantly correlated with the willingness to relocate due to floods (Model 4). In Model 5, in which individual characteristics, flood risk perception, social support and self-efficacy were applied as the independent variables, gender, marital status, income, flood experience and age were not significantly correlated with the dependent variable. Model 5 demonstrated that the number of children, education, chronic illness, being a university student, possibility, fear and self-efficacy were positively and significantly correlated with the willingness to relocate because of floods. In model 5, residence duration and informal and formal social support were negatively and significantly correlated with the dependent variable. According to the R square values, the best model explaining the variation in the willingness to relocate due to floods was Model 5 (all independent variables), Model 2 (flood risk perception), Model 1 (individual characteristics), Model 3 (self-efficacy) and Model 4 (social support), respectively. Table 4 demonstrates the results of the models (Model 6–10) that tested the effect of the independent variables on the willingness to relocate due to landslides. Model 6 shows that the variables of gender, marital status, income, number of children and age were not significantly correlated with the dependent variable. On the other hand, being a primary school graduate, landslide experience, chronic illness and being a university student were positively and significantly correlated with the willingness to relocate due to landslides. Residence duration was negatively and significantly correlated with the willingness to relocate due to landslides (Model 6). From the landslide risk perception sub-dimensions, while severity was not significantly correlated with the dependent variable, the strongest independent variable that was positively and significantly correlated with the dependent variable were fear, possibility and uncontrollable, respectively (Model 7).Table 4 Ordinal logistic regression results of the effect of independent variables on willingness to relocate due to landslides Independent variable Model 6 Model 7 Model 8 Model 9 Model 10 Gender − .168 (.125) − .067(.130) Marital status − .014 (.202) − .160(.210) Income − .008 (.070) − .001(.073) Number of children .147 (.110) .233(.114) Education Literate .401 (.351) .372 (.358) Primary school .798 (.337)* .715 (.348)* Elementary .466 (.319) .297 (.333) High school .082 (.161) .272 (.167) Pre-undergraduate − .014 (.159) .096 (.165) Undergraduate or abovea Landslide experience .947 (.235)* .437 (.243) Chronic Illness .667 (.166)* .527 (.171)** Occupation Unemployed .311 (.266) .321 (.276) University student .556 (.225)* .494 (.233)* Self-employed .140 (.226) − .136 (.236) Officer − .087 (.222) − .102 (.229) Tradesmen − .159 (.281) − .048 (.291) Othersa Age − .015 (.011) − .013 (.011) Residence duration − .029 (.006)* − .024 (.007)* Risk perception = Severity − .008 (.078) − .005 (.08) Risk perception = Possibility .403 (.077)* .396 (.08)* Risk perception = Fear .567 (.063)* .489 (.065)* Risk Perception = Uncontrollable .238 (.060)* .208 (.063) Self-efficacy .332 (.051)* .297 (.059)* Informal social support − .112 (.063) − .203 (.069) Formal social support − .040 (.064) − .065 (.068) Pseudo-R square Cox and Snell .13 .25 .04 .007 .33 Nagelkerke .14 .26 .04 .007 .35 McFadden .04 .09 .01 .002 .13 Model fitting information − 2 Log Likelihood 2616.052 1327.980 145.975 438.061 2527.634 Chi-square 137.985* 271.754* 39.782* 6.542* 390.072** a = Reference category. Robust standard errors in parentheses. Outside the parentheses are the estimates. Results significant at the p < .05; p < .01; **p < .001 levels Model 8 shows that self-efficacy against landslides was positively and significantly correlated with the willingness to relocate. Perceived social support in case of a landslide damage was not significantly correlated with the willingness to relocate due to landslides (Model 9). When individual characteristics, landslide risk perception, self-efficacy and social support were included in Model 10 as independent variables, gender, marital status, income, landslide experience, age, severity and formal social support were not significantly correlated with the dependent variable. The number of children, chronic illness, being a primary school graduate, possibility, fear, uncontrollable and self-efficacy were positively and significantly correlated with the willingness to relocate due to landslides. Informal social support and residence duration were negatively and significantly correlated with the willingness to relocate due to landslides. The R square values revealed that the model that best explained the variation in the willingness to relocate due to landslides was Model 10 (all independent variables), Model 7 (landslide risk perception), Model 6 (individual characteristics), Model 8 (self-efficacy) and Model 9 (social support), respectively. Discussion This study revealed how and to what extent individual characteristics, risk perception, self-efficacy and perceived social support were correlated with the willingness of people residing in Gümüşhane (Türkiye) to relocate due to floods and landslides via ordinal logistic regression analysis. Risk perception (severity, possibility, fear and uncontrollable), self-efficacy and perceived social support were determined separately for floods and landslides, and five models for floods and five models for landslides were tested. The independent variables were included in the analysis one by one and finally all together. It is thought that the results of the research will theoretically and practically contribute to disaster management. This study focused on the pre-disaster period for better disaster risk management and compared the results according to floods and landslides. It presented the results of a study conducted in Gümüşhane, where flood and landslide risks are high and the socioeconomic level is low. It also revealed that risk perception, self-efficacy, perceived social support, and relocation willingness differed for floods and landslides and the strength of the relationship between independent variables and relocation willingness varied according to floods and landslides. In short, this study emphasized that disaster risk reduction studies should be conducted separately for each disaster type to increase regional resilience and to plan better disaster preparedness. The design and results of the study can be a reference for scientists and those who implement public policy for disaster risk reduction in settlements in the world where there are many hazards. Inconsistent with hypothesis H1, gender, marital status, income and age were not significantly correlated with the willingness to relocate due to both floods and landslides. When included in the analysis with all independent variables, consistent with hypothesis H1, the number of children was positively and significantly correlated with the willingness to relocate due to floods and landslides. Similarly, Seebauer and Winkler (2020) revealed that children affected families' relocation decisions. On the other hand, Holley (2022) found that the presence of children and grandchildren at home was not correlated with the relocation decision. Families with a large number of children may want to move from the unsafe area, as they are more worried about their children in case of floods and landslides. Consistent with hypothesis H1, people with lower education levels were more likely to relocate. Education level is an important factor affecting people's resilience to disasters (Cai et al. 2018); therefore, people with low education level may want to relocate, as they feel more vulnerable to floods and landslides. Consistent with hypothesis H1, people who experienced floods and landslides were more likely to relocate. Disaster experience is an important factor affecting people’s attitudes and behaviors toward disasters. For example, in France, the severity of flood experience was positively and significantly correlated with the flood mitigation behavior (Richert et al. 2017). People who were more damaged by flooding thought that it was less possible to prevent floods and reduce flood damage, and had higher-flood-risk perceptions (Hudson et al. 2020). Flood experience of people residing in flood risk area increased perceived probability of a further flood risk (Bustillos Ardaya et al. 2017). A meta-analysis study revealed that flood experience negatively affected trust in public flood protection, but positively affected threat appraisal (Bamberg et al. 2017). The severity of disasters experienced was significantly and positively correlated with perceived threat and responsive efficacy (Xue et al. 2021). People experiencing floods and landslides may want to leave the area where they live because they are worried about future disasters. Future studies can provide a better understanding of the past disaster experiences of these people with the help of qualitative methods and therefore produce solutions against flood and landslide hazards. Consistent with hypothesis H1, the chronic disease variable was positively and significantly correlated with the willingness to relocate due to floods and landslides. Scientific studies showed that people with chronic diseases or any disability confronted many problems in the fight against disasters. For instance, those with poor general health, the disabled and those with three or more chronic diseases had lower disaster preparedness levels than those without any health problems (Bethel et al. 2011). Tomio et al. (2010) found that it became more and more difficult for people with chronic diseases to obtain their medicines after the flood. Students with chronic diseases thought that disasters would be more difficult to control in the university campus (Mızrak and Aslan 2020). People with chronic diseases participating in this research may want to move from the hazardous area because they feel vulnerable to floods and landslides. For this reason, disaster management plans should be prepared according to the needs of people with chronic diseases and disabilities, and the demands and suggestions of these people for protection from disasters should be understood. In this way, the resilience of people with chronic diseases and disabilities against disasters is increased, and they are provided to live in a safer environment. Consistent with hypothesis H1, university students had more intention to leave the city due to floods and landslides. University students have low awareness of the environment (Simms et al. 2013) and disaster preparedness (Lovekamp and Tate 2008; Wu et al. 2017; Hasan et al. 2022), and need more external assistance (Tanner and Doberstein 2015). Flood and landslide risk perceptions of university students should be understood, and their awareness related to environmental hazards and institutions and stakeholders should be increased by collaborating with the university administration. Consistent with hypothesis H1, residence duration in the province was positively and significantly correlated with the willingness to relocate due to floods and landslides. People with a long period of residence in the province may have higher material and moral dependence on the environment and other people, which may increase their intention to stay in the province. Consistent with H2, risk perception was positively and significantly correlated with willingness to relocate due to both floods and landslides. However, inconsistent with H2, severity, which is one of the risk perception sub-dimensions, was not positively and significantly correlated to the willingness to relocate due to both floods and landslides. In addition, risk perception was the strongest independent variable explaining the change in the willingness to relocate and the effect size of the risk perception sub-dimensions on willingness to relocate changed. The willingness to relocate due to floods was most correlated with possibility, while the willingness due to landslides was most correlated with fear. Similarly, other scientific studies revealed that risk perception factors had different effects on willingness to relocate. For instance, uncertainty about flooding was a primary factor influencing people's relocation decisions in Austria (Seebauer and Winkler 2020). Among the sub-dimensions of landslide risk perception, worry and unknown were not significantly correlated with relocation willingness, while probability was positively and significantly correlated with relocation willingness more than threat. Moreover, controllability was negatively and significantly correlated with relocation willingness (Xu et al. 2017). In another study, probability was not significantly correlated with the intention to relocate elsewhere in or out of the state due to floods; however, perceived consequences were positively and significantly correlated with the intention to relocate (Holley et al. 2022). These results highlight the need for a comprehensive study of people's risk perceptions and the importance of risk perception on relocation policies. Inconsistent with H3 and Seebauer and Winkler (2020), self-efficacy was positively and significantly correlated with the willingness to relocate due to both floods and landslides. Seebauer and Winkler (2020) found that people with strong self-efficacy regarding flood preparedness and prevention tended to stay in the flood risk area. On the other hand, Song and Peng (2017) found that high awareness of the effects of sea level rise increased relocation willingness. Likewise, the perceived self-efficacy to avoid flood risk was positively and significantly correlated with the intention to move to a new place in Louisiana and to move from Louisiana to elsewhere (Holley et al. 2022). Although the participants in this study believed that they could protect themselves in case of floods and landslides, the reason why they wanted to move from the province might be that they were worried about their families and their welfare. Persons responsible for disaster management in the province should be provided with the necessary training and practices so that people can acquire the knowledge and skills to protect other people in case of a flood and landslide and to take the necessary protection measures before floods and landslides. When all independent variables were included in the analysis, consistent with H4, formal and informal social support were negatively and significantly correlated the willingness to relocate due to floods. However, consistent with H4, only informal social support was negatively and significantly correlated the willingness to relocate due to landslides. Social support can reduce the willingness to relocate because it helps people cope with disasters. For example, social support perceptions of the elderly increased community cohesion and residential satisfaction after typhoon-induced relocation (Chao 2017). People with high social commitment stated that it was more possible to be protected from flood damages (Hudson et al. 2020). Moreover, this study showed that the variable that least explained the change in willingness to relocate was social support. Since formal social support did not affect the relocation decision due to landslides, institutions should interact more with the public in their work on landslides. In this study, the model that best explained the willingness to relocate was the models in which all independent variables were included in the analysis. Therefore, the factors affecting people's decision to relocate due to disasters should be addressed more comprehensively. Conclusion and recommendations This study investigated the factors affecting the relocation willingness of people residing in areas with high flood and landslide risks. In addition to individual characteristics, risk perception, self-efficacy and social support measured separately for floods and landslides were utilized as independent variables, and the data were analyzed separately according to floods and landslides. Except for the uncontrollable sub-dimension, the landslide risk perception was higher than the flood risk perception in the other sub-dimensions (severity, possibility, fear). The perceived informal and formal social support and self-efficacy for floods were higher, and the informal social support was higher than the formal social support for both floods and landslides. The willingness to relocate due to landslides was higher than the willingness to relocate due to floods. People with more children, people with disaster experience and chronic diseases, and university students were more likely to relocate because of both floods and landslides. Compared with graduates of undergraduate or higher education, literate people and primary and secondary school graduates were more likely to relocate due to floods, while primary school graduates were more likely to relocate due to landslides. Those who resided longer in an area with high flood and landslide risks had less willingness to relocate. The possibility and fear of floods and landslides and the thought that the measures taken against these disasters were inadequate triggered the willingness to relocate. In the event of a landslide or flood, people's high level of knowledge to protect themselves increased their willingness to relocate. People who thought that they could get help from people and institutions around them in case of a flood were less likely to relocate because of floods. People's thoughts that institutions could help them in case of a landslide were not correlated with their decision to relocate because of the landslides. People's thoughts, attitudes and behaviors toward different types of disasters should be monitored spatially and temporally because the type, severity and probability of disasters vary from region to region. If the factors that cause people to be affected by disasters are reduced or removed, people will live in a more sustainable community and safer environment. In order to control the relocation caused by disasters and to protect the social and environmental structure, people's place preferences for relocation may be investigated. Policy makers and scientists should support the disaster management strategies of institutions more and increase the cooperation between institutions and the public in order to increase the disaster resilience of the community. All disaster risks in the region should be considered while creating disaster management policies, and disaster plans should be customized according to the type of disaster. Limitations This study has limitations that should be considered. This study only discussed the willingness to relocate and not the willingness to evacuate. The factors affecting the relocation and evacuation decision can be investigated together, and the results can be compared. The data in this study were obtained as cross-sectional and generally reflected the thoughts of people in the city center of Gümüşhane related to floods and landslides. However, floods may cause more damage to those structures close to the Harşit River, while landslides may damage settlements in areas with high slopes. Future research should focus on people living in the high-risk areas of the city to gain deeper knowledge. If the coronavirus pandemic had not occurred during the data collection process, more participants could have been reached, the probability sampling method could have been used or more specific data could have been collected from households. Author contributions The authors read, reviewed and approved the final manuscript. Funding This research did not receive any funding. Declarations Conflict of interest No conflict of interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Acar S Bilen Kazancık L Meydan MC Işık M Socio-economic development of provinces and regions ranking survey SEGE-2017 2019 General Directorate of Development Agencies, Ankara Republic of Türkiye Ministry of Industry and Technology Ahmadi S Khankeh H Sahaf R Daily life challenges in an earthquake disaster situation in older adults a qualitative study in Iran J Clin Diagnostic Res 2018 12 08 12 10.7860/JCDR/2018/32699.11377 Alam E Landslide hazard knowledge, risk perception and preparedness in Southeast Bangladesh Sustainability 2020 12 6305 10.3390/su12166305 Alemdag S Akgun A Kaya A Gokceoglu C A large and rapid planar failure: causes, mechanism, and consequences (Mordut, Gumushane, Turkey) Arab J Geosci 2014 7 1205 1221 10.1007/s12517-012-0821-1 Anacio DB Hilvano NF Burias IC Dwelling structures in a flood-prone area in the philippines: sense of place and its functions for mitigating flood experiences Int J Disaster Risk Reduct 2016 15 108 115 10.1016/j.ijdrr.2016.01.005 Anton CE Lawrence C Home is where the heart is: the effect of place of residence on place attachment and community participation J Environ Psychol 2014 40 451 461 10.1016/j.jenvp.2014.10.007 Ao Y Huang K Wang Y Influence of built environment and risk perception on seismic evacuation behavior: evidence from rural areas affected by Wenchuan earthquake Int J Disaster Risk Reduct 2020 46 101504 10.1016/j.ijdrr.2020.101504 Babcicky P Seebauer S Collective efficacy and natural hazards: differing roles of social cohesion and task-specific efficacy in shaping risk and coping beliefs J Risk Res 2020 23 695 712 10.1080/13669877.2019.1628096 Baert M Kervyn M Kagou AD Resettlement preferences from landslide prone areas in Cameroon: willingness to move, reasons to stay Land Use Policy 2020 10.1016/j.landusepol.2019.04.036 Bamberg S, Masson T, Brewitt K, Nemetschek N (2017) Threat, coping and flood prevention – A meta-analysis. J Environ Psychol 54:116–126. 10.1016/j.jenvp.2017.08.001 Bera MK Daněk P The perception of risk in the flood-prone area: a case study from the Czech municipality Disaster Prev Manag Int J 2018 27 2 14 10.1108/DPM-01-2017-0004 Bethel JW Foreman AN Burke SC Disaster preparedness among medically vulnerable populations Am J Prev Med 2011 40 139 143 10.1016/j.amepre.2010.10.020 21238861 Bornstein MH Jager J Putnick DL Sampling in developmental science: situations, shortcomings, solutions, and standards Dev Rev 2013 33 357 370 10.1016/j.dr.2013.08.003 25580049 Botzen WJW Kunreuther H Czajkowski J Moel H Adoption of individual flood damage mitigation measures in New York City: an extension of protection motivation theory Risk Anal 2019 39 2143 2159 10.1111/risa.13318 31021457 Bubeck P Botzen WJW Aerts JCJH A review of risk perceptions and other factors that influence flood mitigation behavior Risk Anal 2012 32 1481 1495 10.1111/j.1539-6924.2011.01783.x 22394258 Burnside R, Miller DMS, Rivera JD (2007) The impact of information and risk perception on the Hurricane evacuation decision-making of greater New Orleans residents. Sociol Spectr 27:727–740. 10.1080/02732170701534226 BustillosArdaya A Evers M Ribbe L What influences disaster risk perception? Intervention measures, flood and landslide risk perception of the population living in flood risk areas in Rio de Janeiro state, Brazil Int J Disaster Risk Reduct 2017 25 227 237 10.1016/j.ijdrr.2017.09.006 Cai H Lam NSN Qiang Y A synthesis of disaster resilience measurement methods and indices Int J Disaster Risk Reduct 2018 31 844 855 10.1016/j.ijdrr.2018.07.015 Chao S-F Social support, coping strategies and their correlations with older adults’ relocation adjustments after natural disaster Geriatr Gerontol Int 2017 17 1006 1014 10.1111/ggi.12807 27380717 Daniel A Michaela C Mental health and health-related quality of life in victims of the 2013 flood disaster in Germany—a longitudinal study of health-related flood consequences and evaluation of institutionalized low-threshold psycho-social support Int J Disaster Risk Reduct 2021 10.1016/j.ijdrr.2021.102179 Dinh NC Ubukata F Tan NQ Ha VH How do social connections accelerate post-flood recovery? Insights from a survey of rural households in central Vietnam Int J Disaster Risk Reduct 2021 10.1016/j.ijdrr.2021.102342 Drolet J Dominelli L Alston M Women rebuilding lives post-disaster: innovative community practices for building resilience and promoting sustainable development Gend Dev 2015 23 433 448 10.1080/13552074.2015.1096040 Drury J Brown R González R Miranda D Emergent social identity and observing social support predict social support provided by survivors in a disaster: solidarity in the 2010 Chile earthquake Eur J Soc Psychol 2016 46 209 223 10.1002/ejsp.2146 Fatemi F Ardalan A Aguirre B Social vulnerability indicators in disasters: findings from a systematic review Int J Disaster Risk Reduct 2017 22 219 227 10.1016/j.ijdrr.2016.09.006 Garakani SA Lak A Niyasati M Toward sustainable development in post-flood relocation of rural settlements in Iran Int J Disaster Resil Built Environ 2020 11 359 377 10.1108/IJDRBE-05-2019-0020 Garbero A Muttarak R Impacts of the 2010 droughts and floods on community welfare in rural Thailand: differential effects of village educational attainment Ecol Soc 2013 18 art27 10.5751/ES-05871-180427 Gong W Juang CH Wasowski J Geohazards and human settlements: lessons learned from multiple relocation events in Badong, China–Engineering geologist’s perspective Eng Geol 2021 285 106051 10.1016/j.enggeo.2021.106051 Guerra C Cumsille P Martínez ML Post-traumatic stress symptoms in adolescents exposed to an earthquake: association with self-efficacy, perceived magnitude, and fear Int J Clin Heal Psychol 2014 14 202 207 10.1016/j.ijchp.2014.05.001 Gümüşhane Disaster and Emergency Directorate (2021) Provincial Disaster and Risk Reduction Plan. https://gumushane.afad.gov.tr/il-planlari. Accessed 8 June 2022 Hansel TC Osofsky JD Osofsky HJ Friedrich P The Effect of long-term relocation on child and adolescent survivors of Hurricane Katrina J Trauma Stress 2013 26 613 620 10.1002/jts.21837 24115291 Haque U da Silva PF Devoli G The human cost of global warming: deadly landslides and their triggers (1995–2014) Sci Total Environ 2019 682 673 684 10.1016/j.scitotenv.2019.03.415 31129549 Hasan MK Moriom M Shuprio SIM Exploring disaster preparedness of students at university in Bangladesh 2022 Netherlands Springer Holley JR McComas KA Lambert CE Responding to flood risk in Louisiana: the roles of place attachment, emotions, and location Nat Hazards 2022 10.1007/s11069-022-05316-9 Hosmer DW, Lemeshow S, May S (2008) Applied survival analysis: regression modeling of time to event data, Second edition. Wiley, Ottawa Hudson P Hagedoorn L Bubeck P Potential linkages between social capital, flood risk perceptions, and self-efficacy Int J Disaster Risk Sci 2020 11 251 262 10.1007/s13753-020-00259-w Huyu S Yu M Que T Individual willingness to prepare for disasters in a geological hazard risk area: an empirical study based on the protection motivation theory Nat Hazards 2022 110 2087 2111 10.1007/s11069-021-05026-8 International Displacement Monitoring Centre (2020) Global report on internal displacement. In: Nor Refug Counc https://www.internal-displacement.org/sites/default/files/publications/documents/2020-IDMC-GRID.pdf. Accessed 13 Feb 2022 Iuchi K Mutter J Governing community relocation after major disasters: an analysis of three different approaches and its outcomes in Asia Prog Disaster Sci 2020 6 100071 10.1016/j.pdisas.2020.100071 Jhung MA Shehab N Rohr-Allegrini C Chronic disease and disasters. Medication demands of Hurricane Katrina Evacuees Am J Prev Med 2007 33 207 210 10.1016/j.amepre.2007.04.030 17826580 Kaya A Alemdağ S Dağ S Gürocak Z Stability assessment of high-steep cut slope debris on a landslide (Gumushane, NE Turkey) Bull Eng Geol Environ 2016 75 89 99 10.1007/s10064-015-0753-6 Kılıc C Aydın İ Taşkıntuna N Predictors of psychological distress in survivors of the 1999 earthquakes in Turkey: effects of relocation after the disaster Acta Psychiatr Scand 2006 114 194 202 10.1111/j.1600-0447.2006.00786.x 16889590 Krejcie RV Morgan DW Determining sample size for research activities Educ Psychol Meas 1970 30 607 610 10.1177/001316447003000308 Lovekamp WE Tate ML College student disaster risk, fear and preparedness Int J Mass Emerg Disasters 2008 26 70 90 Ma Z Guo S Deng X Xu D Community resilience and resident’s disaster preparedness: evidence from China’s earthquake-stricken areas Nat Hazards 2021 108 567 591 10.1007/s11069-021-04695-9 Malak MA Sajib AM Quader MA Anjum H “We are feeling older than our age”: vulnerability and adaptive strategies of aging people to cyclones in coastal Bangladesh Int J Disaster Risk Reduct 2020 48 101595 10.1016/j.ijdrr.2020.101595 Mayr B Thaler T Hübl J Successful small-scale household relocation after a millennial flood event in Simbach, Germany 2016 Water (Switzerland) 2020 10.3390/w12010156 Ministry of Interior of the Republic of Türkiye (2022) DEMA draws Türkiye’s disaster risk map. https://www.icisleri.gov.tr/afad-turkiyenin-afet-risk-haritasini-cikardi Mızrak S Aslan R Disaster risk perception of university students Risk, Hazards Cris Public Policy 2020 11 411 433 10.1002/rhc3.12202 O’brien RM, A caution regarding rules of thumb for variance inflation factors Qual Quant 2007 41 673 690 10.1007/s11135-006-9018-6 Osuret J Atuyambe LM Mayega RW Coping strategies for landslide and flood disasters: a qualitative study of Mt. Elgon region, Uganda PLoS Curr 2016 10.1371/currents.dis.4250a225860babf3601a18e33e172d8b Ozdemir O Yilmaz C Factors affecting risk mitigation revisited: the case of earthquake in Turkey J Risk Res 2011 14 17 46 10.1080/13669871003782751 Peng L Lin L Liu S De XuD Interaction between risk perception and sense of place in disaster-prone mountain areas: a case study in China’s Three Gorges Reservoir Area Nat Hazards 2017 85 777 792 10.1007/s11069-016-2604-6 Peng L Tan J Lin L Xu D Understanding sustainable disaster mitigation of stakeholder engagement: risk perception, trust in public institutions, and disaster insurance Sustain Dev 2019 27 885 897 10.1002/sd.1948 Pinter N Rees JC Assessing managed flood retreat and community relocation in the Midwest USA Nat Hazards 2021 107 497 518 10.1007/s11069-021-04592-1 Qing C Guo S Deng X Stay in risk area: place attachment, efficacy beliefs and risk coping Int J Environ Res Public Health 2022 19 2375 10.3390/ijerph19042375 35206573 Rashid H Hunt LM Haider W Urban flood problems in Dhaka, Bangladesh: slum residents’ choices for relocation to flood-free areas Environ Manag 2007 40 95 104 10.1007/s00267-006-0233-7 Rey-Valette H Robert S Rulleau B Resistance to relocation in flood-vulnerable coastal areas: a proposed composite index Clim Policy 2019 19 206 218 10.1080/14693062.2018.1482823 Richert C Erdlenbruch K Figuières C The determinants of households’ flood mitigation decisions in France-on the possibility of feedback effects from past investments Ecol Econ 2017 131 342 352 10.1016/j.ecolecon.2016.09.014 Samaddar S Chatterjee R Misra B Tatano H Outcome-expectancy and self-efficacy: reasons or results of flood preparedness intention? Int J Disaster Risk Reduct 2014 8 91 99 10.1016/j.ijdrr.2014.02.002 Samonte P Understanding the impacts of post-disaster relocation on family dynamics and resilience Int J Disaster Resil Built Environ 2021 10.1108/IJDRBE-03-2021-0026 Seebauer S Winkler C Should I stay or should I go? Factors in household decisions for or against relocation from a flood risk area Glob Environ Chang 2020 60 102018 10.1016/j.gloenvcha.2019.102018 Shao W Xian S Lin N Small MJ A sequential model to link contextual risk, perception and public support for flood adaptation policy Water Res 2017 122 216 225 10.1016/j.watres.2017.05.072 28601034 Simms JL Kusenbach M Tobin GA Equally unprepared: assessing the hurricane vulnerability of undergraduate students Weather Clim Soc 2013 5 233 243 10.1175/WCAS-D-12-00056.1 Sipe N Vella K Relocating a flood-affected community: good planning or good politics? J Am Plan As 2014 80 400 412 10.1080/01944363.2014.976586 Song J Peng B Should we leave? Attitudes towards relocation in response to sea level rise Water 2017 9 941 10.3390/w9120941 Strömberg D Natural disasters, economic development, and humanitarian aid J Econ Perspect 2007 21 199 222 10.1257/jep.21.3.199 Taiban S Lin HN Ko CC Disaster, relocation, and resilience: recovery and adaptation of Karamemedesane in Lily Tribal Community after Typhoon Morakot, Taiwan Environ Hazards 2020 19 209 222 10.1080/17477891.2019.1708234 Takahashi S Nakamura M Yonekura Y Association between relocation and changes in cardiometabolic risk factors: a longitudinal study in tsunami survivors of the 2011 Great East Japan earthquake BMJ Open 2016 10.1136/bmjopen-2016-011291 Tanner A Doberstein B Emergency preparedness amongst university students Int J Disaster Risk Reduct 2015 13 409 413 10.1016/j.ijdrr.2015.08.007 Tekeli-Yeşil S Dedeoğlu N Braun-Fahrlaender C Tanner M Earthquake awareness and perception of risk among the residents of Istanbul Nat Hazards 2011 59 427 446 10.1007/s11069-011-9764-1 Thaler T Fuchs S Financial recovery schemes in Austria: how planned relocation is used as an answer to future flood events Environ Hazards 2020 19 268 284 10.1080/17477891.2019.1665982 Tomio J Sato H Mizumura H Interruption of medication among outpatients with chronic conditions after a flood Prehosp Disaster Med 2010 25 4 6 10.1017/S1049023X00007652 20405454 Tselios V Tompkins EL What causes nations to recover from disasters? An inquiry into the role of wealth, income inequality, and social welfare provisioning Int J Disaster Risk Reduct 2019 33 162 180 10.1016/j.ijdrr.2018.10.003 Tudes S Ceryan S Bulut F Geoenvironmental evaluation for planning: an example from Gumushane City, close to the North Anatolia Fault Zone, NE Turkey Bull Eng Geol Environ 2012 71 679 690 10.1007/s10064-012-0439-2 Turkish Statistical Institute (2022) Address based census registration system results. https://biruni.tuik.gov.tr/medas/?kn=100&locale=tr. Accessed 6 Aug 2022 United Nations Office for Disaster Risk Reduction (2020) Human cost of disasters. An overview of the last 20 years the last 20 years 2000–2019. In: Centre for Research on the Epidemiology of Disasters https://www.undrr.org/media/48008/download. Accessed 19 Oct 2021 van Valkengoed AM Steg L Meta-analyses of factors motivating climate change adaptation behaviour Nat Clim Chang 2019 9 158 163 10.1038/s41558-018-0371-y Vlaeminck P Maertens M Isabirye M Coping with landslide risk through preventive resettlement. Designing optimal strategies through choice experiments for the Mount Elgon region Uganda Land Use Policy 2016 51 301 311 10.1016/j.landusepol.2015.11.023 Wu H-C Greer A Murphy HC Chang R Preparing for the new normal: students and earthquake hazard adjustments in Oklahoma Int J Disaster Risk Reduct 2017 25 312 323 10.1016/j.ijdrr.2017.09.033 Xu J Ou L Resilience and quality of life among Wenchuan earthquake survivors: the mediating role of social support Public Health 2014 128 430 437 10.1016/j.puhe.2014.03.002 24792190 Xu D Peng L Su C Influences of mass monitoring and mass prevention systems on peasant households’ disaster risk perception in the landslide-threatened Three Gorges Reservoir area, China Habitat Int 2016 58 23 33 10.1016/j.habitatint.2016.09.003 Xu D Peng L Liu S Influences of sense of place on farming households’ relocation willingness in areas threatened by geological disasters: evidence from China Int J Disaster Risk Sci 2017 8 16 32 10.1007/s13753-017-0112-2 Xu D Qing C Deng X Disaster risk perception, sense of pace, evacuation willingness, and relocation willingness of rural households in earthquake-stricken areas: evidence from sichuan province, china Int J Environ Res Public Health 2020 10.3390/ijerph17020602 Xue K Guo S Liu Y Social networks, trust, and disaster-risk perceptions of rural residents in a multi-disaster environment: evidence from Sichuan, China Int J Environ Res Public Health 2021 18 1 25 10.3390/ijerph18042106 Yu J Sim T Qi W Informal social support and disaster preparedness: Mediating roles of perceived collective efficacy and self-efficacy Int J Disaster Risk Reduct 2022 10.1016/j.ijdrr.2021.102734 Zhou W, Ma Z, Guo S et al (2021) Livelihood capital, evacuation and relocation willingness of residents in earthquake-stricken areas of rural China. Saf Sci 141:105350. 10.1016/j.ssci.2021.105350	36474522	PMC9716163	NO-CC CODE	2022-12-03 23:20:54	no	Nat Hazards (Dordr). 2022 Dec 2;:1-23	utf-8	Nat Hazards (Dordr)	2,022	10.1007/s11069-022-05731-y	oa_other
==== Front Biophys Rev Biophys Rev Biophysical Reviews 1867-2450 1867-2469 Springer Berlin Heidelberg Berlin/Heidelberg 1020 10.1007/s12551-022-01020-x Review Functional dynamics of SARS-CoV-2 3C-like protease as a member of clan PA http://orcid.org/0000-0002-3085-9540 Kidera Akinori [email protected] 1 Moritsugu Kei 12 Ekimoto Toru 1 Ikeguchi Mitsunori 1 1 grid.268441.d 0000 0001 1033 6139 Graduate School of Medical Life Science, Yokohama City University, 1-7-29 Suehiro-Cho, Tsurumi, Yokohama 230-0045 Japan 2 Present Address: Graduate School of Science, Osaka Metropolitan University, 1-1 Gakuen-Cho, Nakaku, Sakai, Osaka 599-8570 Japan 2 12 2022 113 13 9 2022 17 11 2022 © International Union for Pure and Applied Biophysics (IUPAB) and Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. SARS-CoV-2 3C-like protease (3CLpro), a potential therapeutic target for COVID-19, consists of a chymotrypsin fold and a C-terminal α-helical domain (domain III), the latter of which mediates dimerization required for catalytic activation. To gain further understanding of the functional dynamics of SARS-CoV-2 3CLpro, this review extends the scope to the comparative study of many crystal structures of proteases having the chymotrypsin fold (clan PA of the MEROPS database). First, the close correspondence between the zymogen-enzyme transformation in chymotrypsin and the allosteric dimerization activation in SARS-CoV-2 3CLpro is illustrated. Then, it is shown that the 3C-like proteases of family Coronaviridae (the protease family C30), which are closely related to SARS-CoV-2 3CLpro, have the same homodimeric structure and common activation mechanism via domain III mediated dimerization. The survey extended to order Nidovirales reveals that all 3C-like proteases belonging to Nidovirales have domain III, but with various chain lengths, and 3CLpro of family Mesoniviridae (family C107) has the same homodimeric structure as that of C30, even though they have no sequence similarity. As a reference, monomeric 3C proteases belonging to the more distant family Picornaviridae (family C3) lacking domain III are compared with C30, and it is shown that the 3C proteases are rigid enough to maintain their structures in the active state. Supplementary Information The online version contains supplementary material available at 10.1007/s12551-022-01020-x. Keywords Protein Data Bank Comparison of protein structures SARS-CoV-2 3C-like protease Chymotrypsin fold Clan PA http://dx.doi.org/10.13039/100009619 Japan Agency for Medical Research and Development JP22ama121023 Ikeguchi Mitsunori ==== Body pmcIntroduction The COVID-19 pandemic has had a significant impact on every sector of society worldwide. The extensive response to the pandemic from the biological science community is evident in the enormous number of publications on COVID-19 and SARS-CoV-2. More than 283,000 articles have been collected in the literature hub of COVID-19, LitCovid (08/2022; Chen et al. 2021), since the outbreak. This review focuses on SARS-CoV-2 3C-like protease (3CLpro, also known as main protease), which has been the subject of over 1860 publications. SARS-CoV-2 3CLpro is a cysteine protease that has an important role in viral replication by cleaving the replicase polyprotein to release functional proteins. As such, it is considered a potential target for the development of antiviral therapeutics (Ullrich and Nitsche 2020; Banerjee et al. 2021; Owen et al. 2021; Unoh et al. 2022). Because of the extensive efforts focused on structure-based drug discovery of an inhibitor, more than 580 crystal structures of SARS-CoV-2 3CLpro, most of which are complexed with various drug candidates, have been deposited into the Protein Data Bank (PDB; 08/2022; Berman et al. 2003; Kinjo et al. 2017). This vast amount of the structural information accumulated in the PDB is an invaluable resource, not only for providing the binding poses of various ligands (Gilson et al. 2016; Wang et al. 2020a, b), but also for constituting a basis for the functional dynamics (Best et al. 2006; Kidera et al. 2021). Interactions with a bound ligand alter the structure of the receptor protein in different ways to produce structural variations in the protein depending on their binding poses (Boehr et al. 2009; Feixas et al. 2014). Considering that crystal packing and amino acid mutations also affect protein structure (Andrec et al. 2007; Cavasotto and Phatak 2009), many crystal structures are affected differently by these factors and constitute a structural ensemble, which is termed “crystal structure ensemble” (Kidera et al. 2021). As the theories of protein dynamics, the conformational selection (Ma et al. 1999) and the linear response theory (Ikeguchi et al. 2005) state that the structural change occurring as a response to external stimulation is a reflection of its intrinsic dynamics, and the crystal structure ensemble can be regarded as a sampled subset of the native structural ensemble (Best et al. 2006; Kidera et al. 2021). Based on the crystal structure ensemble consisting of 343 PDB entries (490 independent chains) of SARS-CoV-2 3CLpro (PDB version 07/21), together with those of highly homologous SARS-CoV 3CLpro (96% identity with SARS-CoV-2 3CLpro; SARS-CoV is the etiological agent of SARS in 2002), we examined the functional dynamics of SARS-CoV-2 3CLpro (Kidera et al. 2021). In this review, we extend the structural analysis of SARS-CoV-2 3CLpro to the structures of various proteases in clan PA, classified by the MEROPS peptidase database (Rawlings et al. 2018). SARS-CoV-2 3CLpro adopts a highly ubiquitous chymotrypsin fold that is the hallmark of clan PA and belongs to the protease family C30 (Rawlings et al. 2018). As shown in the phylogeny of clan PA (Fig. 1), the chymotrypsin fold ubiquitously appears from animal to virus and has either a serine or cysteine residue as the nucleophile in the catalytic triad/dyad. This versatility of the chymotrypsin fold is a key to understanding the function of SARS-CoV-2 3CLpro and hints at the necessity of comparing SARS-CoV-2 3CLpro with other members of clan PA. Here the crystal structures of the protease families C30, C107, and S32, which belong to order Nidovirales (vertebrate and invertebrate host; Gorbalenya et al. 2006), are summarized in terms of the activating dimerization tightly coupled with the C-terminal α-helical domain III (the crystal structures are listed in Table S1A and the representative structures are shown in Fig. 2). Domain III in C30 of family Coronaviridae has a key role in allosterically activating dimerization (the superimposed structures taken from each species are shown in Fig. S1; Goyal and Goyal 2020). C107 of family Mesoniviridae also forms a homodimer with domain III of the same fold as that of C30 (Fig. 2b), whereas S32 of family Arteriviridae is monomeric with a half-sized C-terminal domain (Fig. 2c and d). Through a comparison of these proteases, we discuss the role of domain III in dimerization. As a reference, we also compared two families of clan PA with no extra C-terminal domain, including family S1 of the animal digestive enzyme and family C3 of cysteine proteases [3C proteases (3Cpro)]. The latter belongs to order Picornavirales (uni- and multi-cellular eukaryote host) whose name 3Cpro is the origin of the name 3CLpro of C30 and is focused on the activation mechanism (the list of the crystal structures are summarized in Table S2; note that 3C indicates the genome position and C3 is the family name in MEROPS). Family S1 is activated by the zymogen-enzyme transformation, which suggests a similarity to the allosteric dimerization activation in C30, whereas monomeric C3 does not have such an activation mechanism. Based on these crystal structures and the relevant literatures, the functional implication of SARS-CoV-2 3CLpro is reviewed. Here, it is noted that both the classification for biological organisms (the taxonomic genera/families) and the classification for molecules (the MEROPS families of proteases) are used concurrently. The former is written in italics, and the latter is designated by the format: “S/C + number” of MEROPS (Rawlings et al. 2018).Fig. 1 a Taxonomy of viruses belonging to clan PA. The hierarchical classification covers from kingdom/domain/realm to genus, where viruses are under the realm Riboviria. The numbers after the names are the family names of the proteases according to the MEROPS database (Rawlings et al. 2018). In the text, both the virus taxonomy and the protease classification of MEROPS are used. The names in blue are those discussed in this review. The two phyla of virus are positive-sense single-stranded RNA (+ ssRNA) viruses. The members of Nidovirales are given the chain length of 3CLpro listed in MEROPS, along with the variation in each species. Coronaviridae C30 is connected to the detailed phylogenetic tree in (b). At the top of the tree, a cartoon picture of the chymotrypsin fold (chymotrypsinogen, PDB: 1chg) is shown, as the hallmark for clan PA. b Phylogenetic tree of Coronaviridae C30 proteases calculated by the sequences of 3CLpro belonging to Coronaviridae using COBALT (Papadopoulos and Agarwala 2007) and the neighbor joining method. The proteases are those listed in Table S1A whose 3D structures are deposited to the PDB. The subfamilies are designated by the names in blue, Alpha-, Beta-, Gamma-, and Deltacoronavirus Fig. 2 a Homodimer of SARS-CoV-2 3CLpro (PDB: 6m03; family C30). b Homodimer of Cavally virus 3CLpro (5lac; family C107). c Equine arteritis virus 3CLpro (1mbm; family S32). d Porcine reproductive and respiratory syndrome virus 3CL.pro (5y4l; S32). The catalytic dyad for C30 and C107 and the catalytic triad for S32 are drawn in stick. These structures have the chymotrypsin fold (domains I and II) at the N-terminus and domain III at the C-terminus. The size of domain III C30 and C107 is twice that of S32 Catalytic activation allosterically induced by dimerization and ligand binding in SARS-CoV-2 3CLpro The phylogenetic tree of C30 3CLpro is shown in Fig. 1b. The first crystal structure was solved in 2002, which was 3CLpro of porcine transmissible gastroenteritis coronavirus (TGEV) belonging to genus Alphacoronavirus (PDB:1lvo; Anand et al. 2002). Figure S1 shows the representative structures of the C30 proteases. Anand et al. (2002) revealed that 3CLpro forms a homodimer with the N-terminal Ser1 located near the active site of the partner protomer. These observations were supported by mutational studies of TGEV 3CLpro, in which deletion mutants of domain III (Δ200‒302) and of the five N-terminal residues (Δ1‒5) nearly abolished the proteolytic activity, suggesting that the activation requires domain III and the N-terminal residues (later called the N-finger; Yang et al. 2003), which plays a central role in dimerization (Anand et al. 2002). Shi et al. (2004) confirmed in SARS-CoV 3CLpro that proteolytic activity requires dimerization based on the experiments using dynamic light scattering. Moreover, the importance of the N-finger in dimerization was demonstrated by Hsu et al. (2005a, b) using an ultracentrifuge experiment on SARS-CoV 3CLpro. The importance of domain III to the protease activity was also demonstrated by deletion mutants of human coronavirus 229E 3CLpro (Alphacoronavirus; Ziebuhr et al. 1997), murine coronavirus 3CLpro (Betacoronavirus; Lu and Denison 1997), and avian coronavirus 3CLpro (Gammacoronavirus; Ng and Liu 2000). The importance of the N-finger was also found in porcine epidemic diarrhea virus 3CLpro (Alphacoronavirus; Ye et al. 2016). Therefore, the role of domain III and the N-finger in dimerization and catalytic activity are common to all member of family C30. The molecular mechanism of the activation became clear when the first crystal structure of SARS-CoV 3CLpro was solved (PDB:1uj1; Yang et al. 2003). The protomers of the dimer in this crystal structure assume different structures (i.e., chain A is in the active state and chain B is in the inactive collapsed state; Fig. 3a), because it is in a highly heterogeneous environment of the crystal in the space group P 1 21 1 (Kidera et al. 2021). For chain A, the inter-protomer hydrogen bond (HB) is formed between Ser1 of chain B and Phe140 of chain A to induce the cooperative formation of many HBs, G138-H172, S139-Y126, L141-Y118, G143-N28, and C145-N28 (Fig. 3a, also see Table S3), which stabilize the active conformation of the residues 138–145 [called the C-loop as it contains the nucleophile Cys145 of the catalytic dyad (Kidera et al. 2021), also called the oxyanion hole loop named after the corresponding loop in lipase (van Tilbeurgh et al. 1993) and the L1 loop by Tan et al. (2005)]. For chain B, the C-loop is collapsed with no HB to the other part of the protein except HB_C145O-N28N (HB between atom O of Cys145 and atom N of Asn28). Of note, the N-finger of chain A is disordered at Ser1 and Gly2 and fails to form the inter-protomer interaction with the C-loop of chain B. HB_C145O-N28N is maintained in both chains, whereas HB_G143O-N28ND2 is only formed in the active chain A (because of this finding, HB_G143O-N28ND2 is regarded as a marker of the active state (Kidera et al. 2021)), resulting in the formation/collapse transition of the oxyanion hole (Fig. 3a; the NH groups of Gly143 and Cys145 together stabilize the oxyanion at Gln in the substrate, which occurs as a reaction intermediate of the proteolytic reaction; Otto and Schirmeister 1997). Clear evidence of the coupling between the proteolytic activity and dimerization was presented in the monomeric crystal structures of dimerization-deficient mutants of SARS-CoV 3CLpro [PDB:2pwx (G11A), 2qcy (R298A), 3f9e (S139A), and 3m3t (R298A); Chen et al. 2008; Shi et al. 2008; Hu et al. 2009]. These monomeric structures contain a collapsed C-loop and lose all HBs listed above, except for HB_C145O-N28N. The cooperative structural transition of the C-loop allosterically triggered by HB_F140O-S1N* (* indicates the other protomer) was shown to be a common feature in SARS-CoV-2 and SARS-CoV 3CLpro (Kidera et al. 2021).Fig. 3 a The structure of the C-loop in SARS-CoV 3CLpro. Active state (left; PDB: 1uj1_A) and collapsed inactive state (right; 1uj1_B). The active state is defined by the presence of the oxyanion hole (the main-chain NH groups of Gly143 and Cys145 are ready to catch an oxyanion as the reaction intermediate of the substrate). This is monitored by the main-chain oxygen atoms of Gly143 and Cys145, both of which have HBs with Asn28 in the active state and HB_G143-N28 is lost in the collapsed state. Since HB_C145-N28 is maintained even in the collapsed state, the main-chain NH group of Gly143 goes down to the N-terminal side of the C-loop, indicated by the broken lines, to collapse the oxyanion hole. b The structure of the C-loop in chymotrypsin/chymotrypsinogen. Active state (left; PDB: 4cha, chymotrypsin) and collapsed state (right; 1ex3, chymotrypsinogen). Zymogen activation causes a cleavage at Ile16 to produce the positive charge at the N-terminus. The charged Ile16 forms a salt bridge with the carboxyl group of Asp194. In the zymogen state, the side-chain Asp194 forms a polar contact with His40, and salt bridge formation in the activation causes a large flip of the side-chain of Asp194 to make the conformational change in the C-loop to form HB_G193_H40. This bond enables Gly193 and Cys195 to form the oxyanion hole Another factor that contributes to the activation of 3CLpro is substrate/ligand binding. Cheng et al. (2010) demonstrated substrate-induced dimerization by showing that the dimerization-deficient mutant R298A/L of SARS-CoV 3CLpro shifts the equilibrium to the dimer side when the concentration of the substrate peptide is increased. The crystal structure of the R298A mutant of SARS-CoV 3CLpro (PDB:4hi3; Wu et al. 2013) crystalized with a high concentration of a substrate molecule shows a homodimeric structure, in which the C-loop recovered the active state including HB_F140O-S1N, although the electron density of the substrate was not observed. The recovery of the active C-loop is clearly explained by substrate recognition at the subsite S1 including the oxyanion hole defined by Schechter and Berger (1967). The stabilization of the dimeric form by substrate binding is not straightforward, because the substrate does not directly contribute to the inter-protomer interactions. The following scenario may be considered: the bound substrate induces the active C-loop, which in turn enables HB_F140O-S1N to form. This HB then sets the proper position of the two protomers to trigger the formation of the other inter-protomer interactions. Many cases of ligand-induced activation were found in the crystal structure ensemble (Kidera et al. 2021). Some amino acids appended to the N-terminus eliminate the positive charge at the N-terminal Ser1 to weaken the interaction with Phe140, although these crystal structures maintain the homodimeric structure (Lee et al. 2005; Xue et al. 2007). As a result, the C-loop conformation tends to be collapsed in the ligand-free state with a frequency of 13/25 (there are 25 ligand-free chains in 3CLpro having some amino acids appended to the N-terminus. Of these, 13 chains have a collapsed C-loop conformation); however, this frequency decreases almost ten-fold to 7/91 in the ligand-bound structures (Kidera et al. 2021). The influence of ligand interactions is a clue to understanding the maturation process of 3CLpro in the replicase polyprotein, in which 3CLpro in a polyprotein, having the extension at both termini, autoclaves itself with the aid of ligand-induced activation to produce the mature 3CLpro (Hsu et al. 2005a, b; Li et al. 2010; Xia and Kang 2011). Comparison of SARS-CoV-2 3CLpro with family S1: activation and ligand interactions The zymogen-enzyme transformation is the activation mechanism of S1 proteases (Stroud et al. 1977; S1 is the largest family of proteases, and its nucleophile in the catalytic triad is serine, including trypsin and chymotrypsin). Figure 3b shows a representative case in which the active form of the C-loop of α-chymotrypsin (residues 188–195 containing the nucleophile Ser195 of the catalytic triad; PDB:4cha; Tsukada and Blow 1985) is compared with the inactive form of chymotrypsinogen (the zymogen of chymotrypsin; PDB:1ex3; Pjura et al. 2000). Because chymotrypsinogen is cleaved into three chains, residues 1‒13, 16‒146, and 149‒149, for conversion into α-chymotrypsin, the N-terminal positive charge NH3+ of Ile16 forms a salt bridge with the carboxylate of Asp194 and induces a flipping motion of the Asp194 side-chain from the position hydrogen bonded to His40 in chymotrypsinogen. The dissociation between Asp194 and His40 allows the oxygen atom of Gly193 to form an HB to His40 and to complete the oxyanion hole consisting of the nitrogen atoms of Gly193 and Ser195, whereas Ser195 stably maintains an HB with Gly43 during the whole transformation process (Fig. 3b). The active form has a number of other HBs to the C-loop (S189-V17, M192-L143, D194-G142, and S195-H57, where His57 is the base of the catalytic triad). Another unique point is the disulfide bridge between Cys191 and Cys220, which fixes the N-terminal position of the C-loop, although this disulfide bridge occurs only in the subfamily S1A (family S1 contains the subfamilies S1A-S1F in MEROPS (Rawlings et al. 2018)). Comparing the conformational changes occurring during the activation process between SARS-CoV 3CLpro and chymotrypsinogen/chymotrypsin, we noticed similarities between the two proteases including the salt bridge with the N-terminal residue as the trigger for activation (Ser1 in another protomer of 3CLpro vs. Ile16 of the cleaved N-terminus in chymotrypsin), the stable position of the catalytic residue before and after activation (HB_C145O-N28N vs. HB_S195O-G43N), the switchable HB of the glycine residue constituting the oxyanion hole (HB_G143O-N28ND2 vs. HB_G193O-H40NE2), and many HBs cooperatively formed to stabilize the active form of the C-loop. Because there is no significant sequence similarity between the two proteases, the similarity in the activation mechanism should be the consequence of the strong evolutional constraint under the environment of the same chymotrypsin fold. However, the implication of the similar regulation for the catalytic activity of SARS-CoV-2 3CLpro remains obscure, whereas the necessity of the strict regulation for the cleavage activity in chymotrypsin is well understood (Stroud et al. 1977). In addition to the structural changes during activation, another important functional dynamics upon ligand binding occurs in the β-turn at residues 166–178 in SARS-CoV-2 3CLpro, called the E-loop (named after the N-terminal residue of Glu166 at the subsite S3, also called L2 by Tan et al. (2005)). According to the crystal structure ensemble of 3CLpro, the E-loop sensitively responds to the ligand interactions to cause a ligand size-dependent conformational change. A large ligand shifts down the E-loop to make a larger space for the ligand, whereas a small ligand shifts up the E-loop to maintain the interactions (Kidera et al. 2021). The simulation of the binding process of a substrate peptide indicated that the dynamics of the E-loop play an important role in leading the peptide from the initially encountered position at the protein surface to the fully bound state deep inside of the cleft by susceptibly changing its conformation (Moritsugu et al. 2022). With respect to family S1, thrombin shows a highly dynamic structure at a β-loop-β segment (residues 213–226; known as the Na+ loop containing the Na+ binding sites, Arg221a and Lys224) corresponding to the E-loop in SARS-CoV-2 3CLpro (Gohara and Di Cera 2011; Lechtenberg et al. 2012). Within the fluctuation range, it contains the two conformational states of the slow and fast forms (also called the E* and E forms), which regulate the catalytic activity responding to the inter-molecular interactions including Na+ binding. This dynamic feature in this segment is believed to be shared by various members of family S1 (Gohara and Di Cera 2011; Lechtenberg et al. 2012). A functionally relevant dynamics is also found in the high temperature requirement A (HtrA) proteases, in which the loop corresponding to the E-loop of SARS-CoV-2 3CLpro allosterically changes conformation in the trimeric structure to induce the active structure upon binding of a signal peptide to the C-terminal PDZ domain (Wilken et al. 2004; Krojer et al. 2010; Sawa et al. 2011). These coincidences between the E-loop of SARS-CoV-2 3CLpro and the corresponding loop of the members of family S1 are also considered to be resulted from the strong evolutionary constraint of the chymotrypsin fold. SARS-CoV-2 3CL.pro within family C30 (family Coronaviridae) SARS-CoV-2 3CLpro belongs to the protease family C30 (Rawlings et al. 2018), which consists of the four genera of family Coronaviridae, Alpha-, Beta-, Gamma-, and Deltacoronavirus (Woo et al. 2010). In Fig. 1b, the phylogenetic tree of C30 3CLpro was constructed for the species having the PDB entries listed in Table S1A. The classification of 3CLpro is consistent with that of the four genera and the subgenera of viruses defined by the whole genome. SARS-CoV-2 3CLpro is correctly classified in Betacoronavirus. The representative structures of C30 are shown in Fig. S1a after superimposition at the chymotrypsin fold of chain A (N-terminal domains I and II), because domain III and chain B fluctuate largely and randomly against domains I and II of chain A, which primarily depends on the crystal environment. These structures clearly show that the C30 proteases share the same homodimeric form as that of SARS-CoV-2 3CLpro. The significant structural variation in the chymotrypsin fold is found in the E-loop, of which dynamics is relevant to the functional motion in SARS-CoV-2 3CLpro as described above. The other loops having large fluctuations (residues 46–49 and 70–73) are caused mostly by amino acid insertion/deletion. When the C-terminal domain III is superimposed (Fig. S1b), it is found that all 3CLpro have the same fold with large fluctuations at the loop regions (residues 214‒215 and 243‒247), which are also caused by amino acid insertion/deletion. More detailed structural comparisons were done separately for N-terminal domains I and II and for domain III. The structures of family C30 were subjected to hierarchical clustering based on the Cα RMSD after structural alignment by the CE algorithm (Shindyalov and Bourne 1998). The structural classification of domains I and II (Fig. S2a) is mostly consistent with the sequence classification shown in Fig. S2c, in which the four genera and their subgenera are separately clustered. In contrast in the comparison of domain III, Fig. S2b shows that SARS-CoV-2 is classified outside of the cluster of Betacoronavirus because more significant mutations occurred in domain III of SARS-CoV-2 3CLpro compared with the other members (Fig. S2d). This is more clearly shown in the plot of the percentage of identical amino acids in domain III versus that of domains I and II (Fig. S2e). This figure also shows that 70% of the alignments revealed that domain III accumulates more mutations compared with domains I and II, particularly in SARS-CoV-2. This is because domain III has a larger mutational space under the weaker evolutional constraint probably because of the α-helical structure and the distant location from the active site. We examined the HB pattern of the C-loop in 3CLpro of family C30, the loop containing the oxyanion hole regulating the catalytic activity. For SARS-CoV-2 3CLpro, the five HBs are cooperatively formed between the main-chain atoms of the C-loop and the side-chain atoms of the surrounding residues except for HB_F140O-S1N*, when allosterically induced by dimerization. In contrast, HB_C145O-N28N is kept both in the active and collapsed states. This pattern of the HBs between the main-chain and side-chain atoms indicates that the transition between the active and the collapsed states occurs exclusively in the C-loop, but the main-chain of the surrounding residues stays in almost the same position throughout the transition, except for the N-finger. Table S3 shows the HB patterns of the C-loop for the members of C30 listed in Table S1A. The same behaviors as those of SARS-CoV-2 3CLpro were observed for the members of C30. The main-chain atoms of the C-loop bind to the side-chain atoms of the surrounding residues in the active state. The transition between the active and collapsed forms is evident in the comparison between the monomeric collapsed form of 2q6d and the ligand-bound dimer of 2q6f (3CLpro of avian infectious bronchitis virus (IBV) belonging to Gammacoronavirus) (Xue et al. 2008). The former has a collapsed C-loop with none of the five HBs formed, whereas the latter is in the active state with all five HBs formed. However, in the other coronavirus 3C-like proteases, there is a marked distinction from SARS-CoV-2 3CLpro, that is, these proteases have, at most, four HBs instead of five except for porcine epidemic diarrhea virus 3CLpro having five HBs. The decrease in the number of the HBs is resulted from the mutation from Tyr to Phe disabling HB formation at the side-chain (Phe125 in human coronavirus 229E, human coronavirus NL63, and Phe129 in Betacoronaviruses other than SARS-CoV-2) and the mutation from Ala/Ser to Thr at position 143, which causes a steric hindrance at the side-chain methyl group to disrupt HB_I140N-Y117OH (porcine transmissible gastroenteritis coronavirus 3CLpro and feline coronavirus 3CLpro). The reason why the active form is maintained in these proteases despite the smaller number of HBs is because ligand binding assists to stabilize the active form of the C-loop, that is, the ligand-induced activation occurs. Tomar et al. (2015) found that MERS-CoV is a weakly associated dimer requiring ligand binding for activation. It is also notable that the PDB entry of 2q6d mentioned above has a monomeric protein in the asymmetric unit, suggesting that IBV 3CLpro has a large population of the monomeric protein even at high concentration during crystallization (Xue et al. 2008). The presence of ligand-induced activation can also be postulated based on the fact that 70% of the PDB entries in Table S1A are in the ligand-bound state. SARS-CoV-2 3CLpro as a member of Nidovirales: domain III As shown in Fig. 1a, SARS-CoV-2 belongs to order Nidovirales, whose crystal structures of 3CLpro in the families S32 (PDB:1mbm; Barrette-Ng et al. 2002, 3fan, 3fao; Tian et al. 2009, and 5y4l; Shi et al. 2018), C30 (Table S1A), and C107 (5lac and 5lak; Kanitz et al. 2019) contain the C-terminal domain III in addition to the chymotrypsin fold of domains I and II (Fig. 2). The presence of domain III was also reported in families C62, S65, and S75 based on the sequence analysis (Ziebuhr et al. 2003; Smits et al. 2006; Ulferts et al. 2011). Therefore, the C-terminal domain III is a conserved feature of Nidovirales 3CLpro (Gorbalenya et al. 2006). The most striking finding regarding Nidovirales is that C107 of family Mesoniviridae, infecting mosquito, is a homodimer of which fold architecture is essentially the same as that of C30 (family Coronaviridae) as shown in Fig. 2b, although there is no significant sequence similarity between the two families (Kanitz et al. 2019). However, a close examination shows that these distantly related proteases exhibit significant structural variations caused by insertions and deletions (Fig. S3). Particularly in domain III, the structural correspondence is not easy to trace. The HB pattern of the C-loop listed in Table S3 indicates that C107 has fewer HBs stabilizing the C-loop, and both of the ligand-free and ligand-bound structures (PDB: 5lac and 5lak, respectively; Kanitz et al. 2019) are in the active state according to the HB pattern for the residues contributing to the oxyanion hole (HB_G151O-R35N and HB_G153O-R35NH1), although neither of them has an HB with S1 of the partner protomer. These observations suggest that the role of HBs is different from that of C30, although the two structures may not be sufficient to draw a definitive conclusion. Family S32 is monomeric (Fig. 2c and d) and has domain III of half the size of domain III in families C30 and C107. The HB pattern at the oxyanion hole (HB_S120O-T22N (or HB_S118O-S22N) and HB_G118O-T22OG1 (or HB_G116O-S22OG)) shown in Table S3 indicates that the two structures are in the active state (PDB: 1mbm; Barrette-Ng et al. 2002 and 5y4l; Shi et al. 2018), whereas the other two are collapsed (3fan and 3fao; Tian et al. 2009). This suggests that S32 retains the activating transition of the C-loop as is the case for the dimeric C30. This is a marked difference from the monomeric 3Cpro of family C3 (Picornaviridae) discussed below. To discuss the structures of the members with unknown 3D structures, families C62, S65, and S75, the families in order Nidovirales are roughly classified into two groups: (I) S32, S65, and S75 and (II) C30, C107, and C62, based on the following three pieces of information: (1) The phylogenetic tree of Nidovirales classifies the families into these two groups (Fig. 3 in Gulyaeva and Gorbalenya (2021)); (2) the nucleophile of the proteolytic reaction separates them into the two groups of (I) serine and (II) cysteine groups; (3) domain III has the three different sizes including small (S32), medium (S65 and S75), and large (C30, C107 and C65) as indicated in Fig. 1a. Based on this classification, it is hypothesized that S65 and S75 are monomeric like S32. Xu et al. (2020) successfully built a homology model of S65 using the crystal structure of S32 (PDB: 1mbm) as a template. C62 can be classified into the dimeric C30 and C107 groups, although there is no literature that discusses whether C62 is a homodimer or not. We further applied AlphaFold 2 and AlphaFold-Multimer (Jumper et al. 2021; Evans et al. 2021) to the structural prediction of the representative members of families C62, S65, and S75. Fig. S4 shows the prediction results. When the predicted monomer structures are compared with the structure of SARS-CoV-2 3CLpro (PDB: 6m03), we found that the chymotrypsin fold are correctly predicted together with the positions of the catalytic residues as indicated by the high confidence level (pLDDT, predicted Local Distance Difference Test) despite no significant sequence similarity between these families and the other members of clan PA. These results indicate that the machine learning of the chymotrypsin fold is at a high level due to the unique 3D structure with the diverse sequences and enables us to predict the chymotrypsin fold for any member of clan PA with high precision. On the other hand, the prediction of domain III was not satisfactory as shown in the right figures and the low confidence level (Fig. S4a-c). Domain III has a unique α-helical fold (InterPro: Peptidase_C30_dom3_CoV; URL: www.ebi.ac.uk/interpro/) and is found only in the limited protease families of C30 and C107. These conditions made the prediction of domain III difficult when there is no significant sequence similarity with C30 or C107. In the prediction of the homodimer, only family S75 gave a significant result with the low error level (PAE, predicted aligned error) of the inter-protomer arrangement, whereas families C62 and S65 did not provide a reasonable prediction (data not shown). Unexpectedly, the prediction of S75 closely resembles the dimeric structure of C30 (Fig. S4d) despite the low confidence level in domain III. These prediction results appear to suggest that S75 is dimeric, whereas C62 and S65 are monomeric, which are incompatible with the expectation in the above classification. The prediction of the homodimer structures was repeated with the sequences with domain III removed, and basically the same results were obtained. This indicates that the interface between the two chymotrypsin folds is the determinant in the dimer prediction. However, we need a further study on the structures of C62, S65, and S75 to draw a definitive conclusion. Finally, the role of domain III is discussed by quoting the work of van Aken et al. (2006). They carried out a mutagenesis study on equine arteritis virus 3CLpro (the same protein as PDB:1mbm belonging to family S32) to identify the role of domain III. They demonstrated the importance of not only domain III but also the linker connecting domains II and III (they call it the hinge region) in the proteolytic processing of the replicase polyprotein using mutants of the linker as well as a deletion mutant of domain III. The linker binds the N-terminal part of the substrate by adaptively changing the conformation, whereas domain III does not have a direct interaction with the substrate. Therefore, van Aken et al. (2006) concluded that the linker has an important role in the proteolytic reaction, and domain III works to situate the linker at an appropriate position for catalysis. This point is also argued by Anand et al. (2002). This speculation is supported by the comparison between family C30 containing domain III and family C3 lacking domain III. Figure S5 compares SARS-CoV 3CLpro (C30; PDB: 2q6g, Xue et al. 2008) with coxsackievirus A16 3Cpro, a representative member of C3 belonging to family Picornaviridae (PDB: 3sj9; Lu et al. 2011; also see Fig. 1a). The recognition site at the linker in SARS-CoV 3CLpro is replaced by the elongated loop located at the N-terminus to the C-loop (known as the β-ribbon; Sweeney et al. 2007) in coxsackievirus A16 3Cpro. The β-ribbon of C3 can be stably maintained by itself, whereas the flexible linker of C30 is regulated by domain III. This difference reflects in the cleavage site specificity. According to the substrate specificity data in MEROPS (Rawlings et al. 2018) where the substrate sequences are compiled, the amino acid preference at the peptide site P4 (the substrate amino acid position corresponding to the subsite S4; Schechter and Berger 1967) recognized by the β-ribbon in C3 is more specific compared with the P4 site recognized by the linker in C30. The most predominant amino acid appearing at site P4 accounts for 60% (35/58) of the total amino acid occurrence observed in C3, whereas in C30, it accounts for only up to 36% (43/119) (Rawlings et al. 2018). This suggests that the flexible linker of C30 recognizes a larger variety of amino acids at the P4 site compared with the rigid β-ribbon of C3. However, the linker in C30 is not freely fluctuating, but dimerization restricts the conformational freedom of the linker to a certain level and enables it to susceptibly respond to various substrates. 3C proteases of family C3 (family Picornaviridae) Family C3, 3Cpro of family Picornaviridae belonging to order Picornavirales (Fig. 1), has the chymotrypsin fold but lacks domain III; thus, it is monomeric (Fig. S6; Sun et al. 2016; Yi et al. 2021). Although 3CLpro of C30 is named after 3Cpro of C3, these two families belong to different orders and have no sequence similarity. Unlike family S1 or C30, which are under allosteric regulation of the catalytic activation, in C3 such regulation is not known. Therefore, C3 likely has a different HB pattern in the C-loop compared with that of C30. Table S4 summarizes the HB pattern of the C-loop in the protease family C3. Compared with the HB pattern in C30 shown in Table S3, the number of HBs is smaller in C3. The HBs are simply classified into either of the following two types. The first are HBs between the C-loop and the E-loop (HB pairs 1 and 2 in Table S4), which are stably formed between the main-chain atoms even in the collapsed C-loop (PDB: 3zz4 and 3osy; Cui et al. 2011). These HBs are considered to increase the rigidity of the E-loop in C3 compared with the flexible E-loop in C30 (Fig. S1). Recalling the discussion in the previous section that the β-ribbon in C3 was more rigid compared with the linker in C30, we conclude that monomeric C3 is more rigid. The second type of HBs is associated with the main-chain oxygen atoms in the oxyanion hole-forming residues (HB pair 3 and 4 in Table S4). These HBs in C3 exhibit various patterns for each protease subfamily. Subfamily C3E has the same HB pattern as C30 (i.e., HB_C172O-N30N and HB_C170O-N30ND2). Subfamily C3C has HB_C163O-C32N and HB_C161O-N121ND2, in which the HB is formed, not with Cys32, but with Asn121 in the β-turn located on the N-terminus to the β-ribbon, because the side-chain of Cys32 does not provide a hydrogen donor. Subfamily C3A has an HB between Cys147 and Thr26 in only a half of the entries, and the HB to Gly145 is scarcely formed because the hydroxyl group of Thr26 is not strong enough to compete with water for HB formation. Subfamily C3B does not contain these HBs simply because it lacks the secondary structure to generate the HB to the catalytic cysteine residue (Fig. S6). Even though these various HB patterns exist, there are only two entries exhibiting a completely collapsed C-loop (PDB: 3zz4 and 3osy; in 3osy, the β-ribbon in another molecule in the crystal takes the open form to interact with the C-loop and to collapse it). Therefore, these HBs are not required to maintain the C-loop conformation in the active state. These observations suggest that C3 is rigid enough to constantly maintain the structure in the active state. To summarize the survey on the proteases in clan PA, we found various types of common features originating from the chymotrypsin fold, whereas the features specific to SARS-CoV-2 3CLpro were also found. Among the common features, the ligand molecules bound by proteases of multiple species/families should be remarked here. Tables S1A, B, and S2 cite the ligand names in the crystal structures of family C30 and C3, respectively, where we marked the names when their ligands are shared with SARS-CoV-2/SARS-CoV 3CLpro. The result indicates that the majority of the PDB entries contain ligand molecules which also appear in the crystal structures of SARS-CoV-2/SARS-CoV 3CLpro: 17 entries out of 26 entries in family C30 and 29 entries out of 56 entries in family C3A share the same ligand with the entries of SARS-CoV-2/SARS-CoV 3CLpro. This finding suggests not only the 3D structural similarity of the ligand binding site within the members of clan PA, but also possible contributions of these evolutional relation to the drug discovery problem for SARS-CoV-2 3CLpro and to the development of broad-spectrum inhibitors covering various species and variants belonging to clan PA (Wang et al. 2020a, b; Jukič et al. 2021; Luttens et al. 2022; Ullrich et al. 2022; Uraki et al. 2022). Concerning the feature specific to SARS-CoV-2 3CLpro, we hypothesize that SARS-CoV-2 3CLpro utilizes the machinery available in the chymotrypsin fold and domain III optimally to achieve the most susceptible regulation of proteolytic function compared with other proteases in clan PA. However, we have not understood the details how and why this susceptible regulation is employed in the proteolytic processing of the replicase polyprotein or in the cleavage of host proteins. Supplementary Information Below is the link to the electronic supplementary material. Supplementary file1 (XLSX 57 KB) Supplementary file2 (DOCX 24.1 MB) Author contribution The computational analyses were performed by all authors. The first draft was written by AK, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding This research was supported by the Research Support Project for Life Science and Drug Discovery (project ID: JP22ama121023) from the Japan Agency for Medical Research and Development (AMED). Declarations Ethical approval Not applicable. Consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Anand K Palm GJ Mesters JR Siddell SG Ziebuhr J Hilgenfeld R Structure of coronavirus main proteinase reveals combination of a chymotrypsin fold with an extra alpha-helical domain EMBO J 2002 21 3213 3224 10.1093/emboj/cdf327 12093723 Andrec M Snyder DA Zhou Z Young J Montelione GT Levy RM A large data set comparison of protein structures determined by crystallography and NMR statistical test for structural differences and the effect of crystal packing Proteins 2007 69 449 465 10.1002/prot.21507 17623851 Banerjee R Perera L Tillekeratne LMV Potential SARS-CoV-2 main protease inhibitors Drug Discov Today 2021 26 804 816 10.1016/j.drudis.2020.12.005 33309533 Barrette-Ng IH Ng KK Mark BL Van Aken D Cherney MM Garen C Kolodenko Y Gorbalenya AE Snijder EJ James MN Structure of arterivirus nsp4. The smallest chymotrypsin-like proteinase with an alpha/beta C-terminal extension and alternate conformations of the oxyanion hole J Biol Chem 2002 277 39960 39966 10.1074/jbc.M206978200 12163505 Berman H Henrick K Nakamura H Announcing the worldwide Protein Data Bank Nat Struct Biol 2003 10 980 10.1038/nsb1203-980 14634627 Best RB Lindorff-Larsen K DePristo MA Vendruscolo M Relation between native ensembles and experimental structures of proteins Proc Natl Acad Sci USA 2006 103 10901 10906 10.1073/pnas.0511156103 16829580 Boehr DD Nussinov R Wright PE The role of dynamic conformational ensembles in biomolecular recognition Nat Chem Biol 2009 5 789 796 10.1038/nchembio.232 19841628 Cavasotto CN Phatak SS Homology modeling in drug discovery current trends and applications Drug Discov Today 2009 14 676 683 10.1016/j.drudis.2009.04.006 19422931 Chen S Hu T Zhang J Chen J Chen K Ding J Jiang H Shen X Mutation of Gly-11 on the dimer interface results in the complete crystallographic dimer dissociation of severe acute respiratory syndrome coronavirus 3C-like protease crystal structure with molecular dynamics simulations J Biol Chem 2008 283 554 564 10.1074/jbc.M705240200 17977841 Chen Q Allot A Lu Z LitCovid: an open database of COVID-19 literature Nucleic Acids Res 2021 49 D1534 D1540 10.1093/nar/gkaa952 33166392 Cheng SC Chang GG Chou CY Mutation of Glu-166 blocks the substrate-induced dimerization of SARS coronavirus main protease Biophys J 2010 98 1327 1336 10.1016/j.bpj.2009.12.4272 20371333 Cui S Wang J Fan T Qin B Guo L Lei X Wang J Wang M Jin Q Crystal structure of human enterovirus 71 3C protease J Mol Biol 2011 408 449 461 10.1016/j.jmb.2011.03.007 21396941 Evans R, O'Neill M, Pritzel A, Antropova N, Senior A, Green T, Žídek A, Bates R, Blackwell S, Yim J, Ronneberger O, Bodenstein S. Zielinski M. Bridgland A, Potapenko A, Cowie A, Tunyasuvunakool K, Jain R, Clancy E, Hassabis D (2021) Protein complex prediction with AlphaFold-Multimer. bioRxiv 2021.10.04.463034. 10.1101/2021.10.04.463034. Feixas F Lindert S Sinko W McCammon JA Exploring the role of receptor flexibility in structure-based drug discovery Biophys Chem 2014 186 31 45 10.1016/j.bpc.2013.10.007 24332165 Gilson MK Liu T Baitaluk M Nicola G Hwang L Chong J BindingDB in 2015: a public database for medicinal chemistry, computational chemistry and systems pharmacology Nucleic Acids Res 2016 44 D1045 D1053 10.1093/nar/gkv1072 26481362 Gohara DW Di Cera E Allostery in trypsin-like proteases suggests new therapeutic strategies Trends Biotechnol 2011 29 577 585 10.1016/j.tibtech.2011.06.001 21726912 Gorbalenya AE Enjuanes L Ziebuhr J Snijder EJ Nidovirales evolving the largest RNA virus genome Virus Res 2006 117 17 37 10.1016/j.virusres.2006.01.017 16503362 Goyal B Goyal D Targeting the dimerization of the main protease of coronaviruses: a potential broad-spectrum therapeutic strategy ACS Comb Sci 2020 22 297 305 10.1021/acscombsci.0c00058 32402186 Gulyaeva AA Gorbalenya AE A nidovirus perspective on SARS-CoV-2 Biochem Biophys Res Commun 2021 538 24 34 10.1016/j.bbrc.2020.11.015 33413979 Hsu MF Kuo CJ Chang KT Chang HC Chou CC Ko TP Shr HL Chang GG Wang AH Liang PH Mechanism of the maturation process of SARS-CoV 3CL protease J Biol Chem 2005 280 31257 31266 10.1074/jbc.M502577200 15788388 Hsu WC Chang HC Chou CY Tsai PJ Lin PI Chang GG Critical assessment of important regions in the subunit association and catalytic action of the severe acute respiratory syndrome coronavirus main protease J Biol Chem 2005 280 22741 22748 10.1074/jbc.M502556200 15831489 Hu T Zhang Y Li L Wang K Chen S Chen J Ding J Jiang H Shen X Two adjacent mutations on the dimer interface of SARS coronavirus 3C-like protease cause different conformational changes in crystal structure Virology 2009 388 324 334 10.1016/j.virol.2009.03.034 19409595 Ikeguchi M Ueno J Sato M Kidera A Protein structural change upon ligand binding: linear response theory Phys Rev Lett 2005 94 078102 10.1103/PhysRevLett.94.078102 15783858 Jukič M Škrlj B Tomšič G Pleško S Podlipnik Č Bren U Prioritisation of compounds for 3CLpro inhibitor development on SARS-CoV-2 variants Molecules 2021 26 3003 10.3390/molecules26103003 34070140 Jumper J Evans R Pritzel A Green T Figurnov M Ronneberger O Tunyasuvunakool K Bates R Žídek A Potapenko A Bridgland A Meyer C Kohl SAA Ballard AJ Cowie A Romera-Paredes B Nikolov S Jain R Adler J Back T Petersen S Reiman D Clancy E Zielinski M Steinegger M Pacholska M Berghammer T Bodenstein S Silver D Vinyals O Senior AW Kavukcuoglu K Kohli P Hassabis D Highly accurate protein structure prediction with AlphaFold Nature 2021 596 583 589 10.1038/s41586-021-03819-2 34265844 Kanitz M Blanck S Heine A Gulyaeva AA Gorbalenya AE Ziebuhr J Diederich WE Structural basis for catalysis and substrate specificity of a 3C-like cysteine protease from a mosquito mesonivirus Virology 2019 533 21 33 10.1016/j.virol.2019.05.001 31078932 Kidera A Moritsugu K Ekimoto T Ikeguchi M Allosteric regulation of 3CL protease of SARS-CoV-2 and SARS-CoV observed in the crystal structure ensemble J Mol Biol 2021 433 167324 10.1016/j.jmb.2021.167324 34717972 Kinjo AR Bekker GJ Suzuki H Tsuchiya Y Kawabata T Ikegawa Y Nakamura H Protein Data Bank Japan (PDBj) updated user interfaces, resource description framework, analysis tools for large structures Nucleic Acids Res 2017 45 D282 D288 10.1093/nar/gkw962 27789697 Krojer T Sawa J Huber R Clausen T (2010) HtrA proteases have a conserved activation mechanism that can be triggered by distinct molecular cues Nat Struct Mol Biol 2010 17 7 844 852 10.1038/nsmb.1840 20581825 Lechtenberg BC Freund SM Huntington JA An ensemble view of thrombin allostery Biol Chem 2012 393 889 898 10.1515/hsz-2012-0178 22944689 Lee TW Cherney MM Huitema C Liu J James KE Powers JC Eltis LD James MN Crystal structures of the main peptidase from the SARS coronavirus inhibited by a substrate-like aza-peptide epoxide J Mol Biol 2005 353 1137 1151 10.1016/j.jmb.2005.09.004 16219322 Li C Qi Y Teng X Yang Z Wei P Zhang C Tan L Zhou L Liu Y Lai L Maturation mechanism of severe acute respiratory syndrome (SARS) coronavirus 3C-like proteinase J Biol Chem 2010 285 28134 28140 10.1074/jbc.M109.095851 20489209 Lu Y Denison MR Determinants of mouse hepatitis virus 3C-like proteinase activity Virology 1997 230 335 342 10.1006/viro.1997.8479 9143289 Lu G Qi J Chen Z Xu X Gao F Lin D Qian W Liu H Jiang H Yan J Gao GF Enterovirus 71 and coxsackievirus A16 3C proteasesbinding to rupintrivir and their substrates and anti-hand, foot, and mouth disease virus drug design J Virol 2011 85 10319 10331 10.1128/JVI.00787-11 21795339 Luttens A Gullberg H Abdurakhmanov E Vo DD Akaberi D Talibov VO Nekhotiaeva N Vangeel L De Jonghe S Jochmans D Krambrich J Tas A Lundgren B Gravenfors Y Craig AJ Atilaw Y Sandström A Moodie LWK Lundkvist Å van Hemert MJ Neyts J Lennerstrand J Kihlberg J Sandberg K Danielson UH Carlsson J Ultralarge virtual screening identifies SARS-CoV-2 main protease inhibitors with broad-spectrum activity against coronaviruses J Am Chem Soc 2022 144 2905 2920 10.1021/jacs.1c08402 35142215 Ma B Kumar S Tsai CJ Nussinov R Folding funnels and binding mechanisms Protein Eng 1999 12 713 720 10.1093/protein/12.9.713 10506280 Moritsugu K, Ekimoto T, Ikeguchi M, Kidera A (2022) Binding and unbinding pathways of peptide substrate on SARS-CoV-2 3CL protease. bioRxiv 2022.06.08.495396. 10.1101/2022.06.08.495396. Ng LF Liu DX Further characterization of the coronavirus infectious bronchitis virus 3C-like proteinase and determination of a new cleavage site Virology 2000 272 27 39 10.1006/viro.2000.0330 10873746 Otto HH Schirmeister T Cysteine proteases and their inhibitors Chem Rev 1997 97 133 172 10.1021/cr950025u 11848867 Owen DR, Allerton CMN, Anderson AS, Aschenbrenner L, Avery M, Berritt S, Boras B, Cardin RD, Carlo A, Coffman KJ, Dantonio A, Di L, Eng H, Ferre R, Gajiwala KS, Gibson SA, Greasley SE, Hurst BL, Kadar EP, Kalgutkar AS, Lee JC, Lee J, Liu W, Mason SW, Noell S, Novak JJ, Obach RS, Ogilvie K, Patel NC, Pettersson M, Rai DK, Reese MR, Sammons MF, Sathish JG, Singh RSP, Steppan CM, Stewart AE, Tuttle JB, Updyke L, Verhoest PR, Wei L, Yang Q, Zhu Y (2021) An oral SARS-CoV-2 Mpro inhibitor clinical candidate for the treatment of COVID-19. Science 374:1586–1593. 10.1126/science.abl4784 Papadopoulos JS Agarwala R COBALT: constraint-based alignment tool for multiple protein sequences Bioinformatics 2007 23 1073 1079 10.1093/bioinformatics/btm076 17332019 Pjura PE Lenhoff AM Leonard SA Gittis A Protein crystallization by design chymotrypsinogen without precipitants J Mol Biol 2000 300 235 239 10.1006/jmbi.2000.3851 10873462 Rawlings ND Barrett AJ Thomas PD Huang X Bateman A Finn RD The MEROPS database of proteolytic enzymes, their substrates and inhibitors in 2017 and a comparison with peptidases in the PANTHER database Nucleic Acids Res 2018 46 D624 D632 10.1093/nar/gkx1134 29145643 Sawa J Malet H Krojer T Canellas F Ehrmann M Clausen T Molecular adaptation of the DegQ protease to exert protein quality control in the bacterial cell envelope J Biol Chem 2011 286 30680 30690 10.1074/jbc.M111.243832 21685389 Schechter I Berger A On the size of the active site in proteases I Papain Biochem Biophys Res Commun 1967 27 157 162 10.1016/s0006-291x(67)80055-x 6035483 Shi J Wei Z Song J Dissection study on the severe acute respiratory syndrome 3C-like protease reveals the critical role of the extra domain in dimerization of the enzyme defining the extra domain as a new target for design of highly specific protease inhibitors J Biol Chem 2004 279 24765 24773 10.1074/jbc.M311744200 15037623 Shi J Sivaraman J Song J Mechanism for controlling the dimer-monomer switch and coupling dimerization to catalysis of the severe acute respiratory syndrome coronavirus 3C-like protease J Virol 2008 82 4620 4629 10.1128/JVI.02680-07 18305031 Shi Y Lei Y Ye G Sun L Fang L Xiao S Fu ZF Yin P Song Y Peng G Identification of two antiviral inhibitors targeting 3C-like serine/3C-like protease of porcine reproductive and respiratory syndrome virus and porcine epidemic diarrhea virus Vet Microbiol 2018 213 114 122 10.1016/j.vetmic.2017.11.031 29291994 Shindyalov IN Bourne PE Protein structure alignment by incremental combinatorial extension (CE) of the optimal path Protein Eng 1998 11 739 747 10.1093/protein/11.9.739 9796821 Smits SL Snijder EJ de Groot RJ Characterization of a torovirus main proteinase J Virol 2006 80 4157 4167 10.1128/JVI.80.8.4157-4167.2006 16571831 Stroud RM Kossiakoff AA Chambers JL Mechanisms of zymogen activation Annu Rev Biophys Bioeng 1977 6 177 193 10.1146/annurev.bb.06.060177.001141 17350 Sun D Chen S Cheng A Wang M Roles of the picornaviral 3C proteinase in the viral life cycle and host cells Viruses 2016 8 82 10.3390/v8030082 26999188 Sweeney TR Roqué-Rosell N Birtley JR Leatherbarrow RJ Curry S Structural and mutagenic analysis of foot-and-mouth disease virus 3C protease reveals the role of the beta-ribbon in proteolysis J Virol 2007 81 115 124 10.1128/JVI.01587-06 17065215 Tan J Verschueren KH Anand K Shen J Yang M Xu Y Rao Z Bigalke J Heisen B Mesters JR Chen K Shen X Jiang H Hilgenfeld R pH-dependent conformational flexibility of the SARS-CoV main proteinase (M(pro)) dimer: molecular dynamics simulations and multiple X-ray structure analyses J Mol Biol 2005 354 25 40 10.1016/j.jmb.2005.09.012 16242152 Tian X Lu G Gao F Peng H Feng Y Ma G Bartlam M Tian K Yan J Hilgenfeld R Gao GF Structure and cleavage specificity of the chymotrypsin-like serine protease (3CLSP/nsp4) of Porcine Reproductive and Respiratory Syndrome Virus (PRRSV) J Mol Biol 2009 392 977 993 10.1016/j.jmb.2009.07.062 19646449 Tomar S Johnston ML St John SE Osswald HL Nyalapatla PR Paul LN Ghosh AK Denison MR Mesecar AD Ligand-induced dimerization of Middle East respiratory syndrome (MERS) coronavirus nsp5 Protease (3CLpro): implication for nsp5 regulation and the development of antivirals J Biol Chem 2015 290 19403 19422 10.1074/jbc.M115.651463 26055715 Tsukada H Blow DM Structure of alpha-chymotrypsin refined at 1.68 A resolution J Mol Biol 1985 184 703 711 10.1016/0022-2836(85)90314-6 4046030 Ulferts R Mettenleiter TC Ziebuhr J Characterization of Bafinivirus main protease autoprocessing activities J Virol 2011 85 1348 1359 10.1128/JVI.01716-10 21068254 Ullrich S Nitsche C The SARS-CoV-2 main protease as drug target Bioorg Med Chem Lett 2020 30 127377 10.1016/j.bmcl.2020.127377 32738988 Ullrich S Ekanayake KB Otting G Nitsche C Main protease mutants of SARS-CoV-2 variants remain susceptible to nirmatrelvir Bioorg Med Chem Lett 2022 62 128629 10.1016/j.bmcl.2022.128629 35182772 Unoh Y Uehara S Nakahara K Nobori H Yamatsu Y Yamamoto S Maruyama Y Taoda Y Kasamatsu K Suto T Kouki K Nakahashi A Kawashima S Sanaki T Toba S Uemura K Mizutare T Ando S Sasaki M Orba Y Sawa H Sato A Sato T Kato T Tachibana Y Discovery of S-217622, a noncovalent oral SARS-CoV-2 3CL protease inhibitor clinical candidate for treating COVID-19 J Med Chem 2022 65 6499 6512 10.1021/acs.jmedchem.2c00117 35352927 Uraki R Kiso M Iida S Imai M Takashita E Kuroda M Halfmann PJ Loeber S Maemura T Yamayoshi S Fujisaki S Wang Z Ito M Ujie M Iwatsuki-Horimoto K Furusawa Y Wright R Chong Z Ozono S Yasuhara A Ueki H Sakai-Tagawa Y Li R Liu Y Larson D Koga M Tsutsumi T Adachi E Saito M Yamamoto S Hagihara M Mitamura K Sato T Hojo M Hattori SI Maeda K Valdez R IASO study teamOkuda M Murakami J Duong C Godbole S Douek DC Maeda K Watanabe S Gordon A Ohmagari N Yotsuyanagi H Diamond MS Hasegawa H Mitsuya H Suzuki T Kawaoka Y Characterization and antiviral susceptibility of SARS-CoV-2 Omicron BA.2 Nature. 2022 607 119 127 10.1038/s41586-022-04856-1 35576972 van Aken D Snijder EJ Gorbalenya AE Mutagenesis analysis of the nsp4 main proteinase reveals determinants of arterivirus replicase polyprotein autoprocessing J Virol 2006 80 3428 3437 10.1128/JVI.80.7.3428-3437.2006 16537610 van Tilbeurgh H Egloff MP Martinez C Rugani N Verger R Cambillau C Interfacial activation of the lipase-procolipase complex by mixed micelles revealed by X-ray crystallography Nature 1993 362 814 820 10.1038/362814a0 8479519 Wang YC Yang WH Yang CS Hou MH Tsai CL Chou YZ Hung MC Chen Y Structural basis of SARS-CoV-2 main protease inhibition by a broad-spectrum anti-coronaviral drug Am J Cancer Res 2020 10 2535 2545 32905393 Wang Z Sun H Shen C Hu X Gao J Li D Cao D Hou T Combined strategies in structure-based virtual screening Phys Chem Chem Phys 2020 22 3149 3159 10.1039/c9cp06303j 31995074 Wilken C Kitzing K Kurzbauer R Ehrmann M Clausen T Crystal structure of the DegS stress sensor. How a PDZ domain recognizes misfolded protein and activates a protease Cell 2004 117 483 494 10.1016/s0092-8674(04)00454-4 15137941 Woo PC Huang Y Lau SK Yuen KY Coronavirus genomics and bioinformatics analysis Viruses 2010 2 1804 1820 10.3390/v2081803 21994708 Wu CG Cheng SC Chen SC Li JY Fang YH Chen YH Chou CY Mechanism for controlling the monomer-dimer conversion of SARS coronavirus main protease Acta Crystallogr D Biol Crystallogr 2013 69 747 755 10.1107/S0907444913001315 23633583 Xia B Kang X Activation and maturation of SARS-CoV main protease Protein Cell 2011 2 282 290 10.1007/s13238-011-1034-1 21533772 Xu S Zhou J Chen Y Tong X Wang Z Guo J Chen J Fang L Wang D Xiao S Characterization of self-processing activities and substrate specificities of porcine torovirus 3C-like protease J Virol 2020 94 e01282 e1320 10.1128/JVI.01282-20 32727876 Xue X Yang H Shen W Zhao Q Li J Yang K Chen C Jin Y Bartlam M Rao Z Production of authentic SARS-CoV M(pro) with enhanced activity: application as a novel tag-cleavage endopeptidase for protein overproduction J Mol Biol 2007 366 965 975 10.1016/j.jmb.2006.11.073 17189639 Xue X Yu H Yang H Xue F Wu Z Shen W Li J Zhou Z Ding Y Zhao Q Zhang XC Liao M Bartlam M Rao Z Structures of two coronavirus main proteases: implications for substrate binding and antiviral drug design J Virol 2008 82 2515 2527 10.1128/JVI.02114-07 18094151 Yang H Yang M Ding Y Liu Y Lou Z Zhou Z Sun L Mo L Ye S Pang H Gao GF Anand K Bartlam M Hilgenfeld R Rao Z The crystal structures of severe acute respiratory syndrome virus main protease and its complex with an inhibitor Proc Natl Acad Sci USA 2003 100 13190 13195 10.1073/pnas.1835675100 14585926 Ye G Deng F Shen Z Luo R Zhao L Xiao S Fu ZF Peng G Structural basis for the dimerization and substrate recognition specificity of porcine epidemic diarrhea virus 3C-like protease Virology 2016 494 225 235 10.1016/j.virol.2016.04.018 27128350 Yi J Peng J Yang W Zhu G Ren J Li D Zheng H Picornavirus 3C - a protease ensuring virus replication and subverting host responses J Cell Sci 2021 134 jcs253237 10.1242/jcs.253237 33692152 Ziebuhr J Heusipp G Siddell SG Biosynthesis, purification, and characterization of the human coronavirus 229E 3C-like proteinase J Virol 1997 71 3992 3997 10.1128/JVI.71.5.3992-3997.1997 9094676 Ziebuhr J Bayer S Cowley JA Gorbalenya AE The 3C-like proteinase of an invertebrate nidovirus links coronavirus and potyvirus homologs J Virol 2003 77 1415 1426 10.1128/jvi.77.2.1415-1426.2003 12502857	36474932	PMC9716165	NO-CC CODE	2022-12-03 23:20:54	no	Biophys Rev. 2022 Dec 2;:1-13	utf-8	Biophys Rev	2,022	10.1007/s12551-022-01020-x	oa_other
==== Front Talanta Talanta Talanta 0039-9140 1873-3573 Elsevier S0039-9140(20)31112-7 10.1016/S0039-9140(20)31112-7 121821 Article Outside Front Cover 6 11 2020 15 1 2021 6 11 2020 222 121821121821 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc	0	PMC9716167	NO-CC CODE	2022-12-10 23:20:30	no	Talanta. 2021 Jan 15; 222:121821	utf-8	Talanta	2,020	10.1016/S0039-9140(20)31112-7	oa_other
==== Front ZDM ZDM Zdm 1863-9690 1863-9704 Springer Berlin Heidelberg Berlin/Heidelberg 1447 10.1007/s11858-022-01447-2 Original Paper Relevant mathematical modelling efforts for understanding COVID-19 dynamics: an educational challenge http://orcid.org/0000-0001-8002-7075 Meyer J. F. C. A. [email protected] http://orcid.org/0000-0002-8520-4306 Lima M. [email protected] Department of Applied Mathematics, IMECC-UNICAMP, Campinas, SP Brazil 2 12 2022 114 21 10 2022 © FIZ Karlsruhe 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The purpose of the work described in this paper is to emphasize the importance of using mathematical models and mathematical modelling in order to be able to understand and to learn possible behaviours in epidemic situations such as that of the COVID-19 pandemic, besides suggesting modelling techniques with which to evaluate certain sanitary decisions and policies which do, in fact, affect society as a whole. The mathematical tools that are used derive from nonlinear systems of difference equations (possibly viable at a high school level, using spreadsheets or adequate software) as well as nonlinear systems of ordinary differential equations (therefore using mathematical tools and software well within the reach of undergraduate students of many courses). This purpose is accomplished by motivating students and learners to study existing SIR-type models and modifying them in order to have a fully understandable translation of dynamics for infectious diseases such as COVID-19 in several different realistic scenarios, that is to say, situations that consider social distancing policies, widespread vaccination programmes, as well as possible and even probable results when in the presence of negationist postures and attitudes. Several modelling choices referring to real-life situations are shown and explored. These models are analysed and discussed, implicitly proposing similar attitudes and evaluations in learning environments. Conclusions are drawn, stimulating further work using the described mathematical tools and resources. Supplementary Information The online version contains supplementary material available at 10.1007/s11858-022-01447-2. Keywords COVID-19 Mathematical modelling SARS-Cov-2 variants Mathematical epidemiology Nonlinear systems of ODE ==== Body pmcIntroduction Ubiratan D’Ambrósio (1998), in a characteristically provocative statement, wrote that the “Mathematics we teach in school is obsolete, boring and useless” (p. 28, informal translation by the authors). This statement has often been referred to in relation to the need for continuing to change how, why, and what for, mathematics is taught—and sometimes learned. He did not even need to mention examples present in mathematics syllabi at practically all levels, nor to mention our didactic practices in which so many teachers prefer paper and pencil (and, consequently, twelfth century technology) to modern software and databases (technologies of the present century), and also choose memorization instead of the ability to find information on-line and to think. According to Clements (2013) and others, the resource of technology plays a fundamental part in learning processes, in fact, as Williams and Goos (2013) stated, “This notion situates ‘technology’—and mathematics, also—as an essential part or ‘moment’ of the whole activity, alongside other mediational means; thus it can only be fully understood in relation to all the other moments” (p. 549). Now these movements involve the whole procedure of the study, comprehension, and expression of real-life problems, besides the mathematical efforts that follow, as well as evaluation and criticism. Sometimes mathematical concepts and operations can be introduced playfully with games during which certain mechanical calculation results are remembered, but the ongoing challenge is to enable students to become full conscious citizens—and, for this, the mathematics to which Ubiratan D'Ambrósio referred is unfortunately quite irrelevant. An implicit corollary is that, as in much of the effective work with mathematics in the past centuries, mathematical development has also been based upon the need to change the world and its realities, besides its theoretical challenges. The authors believe that an immediate change of attitude in mathematics teaching is needed, in order to accomplish a reversal of Ubiratan D'Ambrósio's phrase. All this is unfortunately true, not only in teaching and learning mathematics but in so many other areas as well. D'Ambrósio's words reflect a universe that is slowly (too slowly, for the authors) being relegated to the past in worldwide teaching experiences, but much of it still thrives in many schools. And it must be said that this is sadly true at all levels and in practical as well as theoretical teaching and learning situations. In implicit agreement with Ubiratan D'Ambrósio, Freire (1967) stated that the first step in an educational process is (in his terms) “to 'read' the world”. But this is not so easily done, even if it is simple to say. There are many ethical requirements, many historical and cultural contexts and a quasi-infinite list of interpretations of facts that are discussed in a school environment, as in society. Paulo Freire (1967) went on to mention that the second step is to construe knowledge with dialogue, collectively, with frequent back-tracking, modifications, and the acceptance of an enormous range of all kinds of differences. The third step is to criticize and evaluate the knowledge thus obtained. Again, for this, there must be a combination of ethnic, cultural, historical and social contexts and environments. And there is also the transdisciplinary dialogue that is necessary for a critical understanding of what has been learnt, but not only this: in addition it must be asked, for how long do these concepts that were understood remain relevant in a learner’s daily activities? For some, these three steps complete the learning cycle, but Freire has another 4th step: that of putting the acquired knowledge to use in serving communities, regions, groups and even whole towns. Choosing to work with the COVID-19 pandemic in this pedagogical context inverts Ubiratan D'Ambrósio's statement: solving a social or natural problem using several different scientific fields with the help of mathematical modelling tools brings out mathematics which is interesting, useful (and necessary) as well as up to date, both theoretically and technologically. We do not claim that working in this way is easy—it most certainly is not! Modelling a social situation like the successive dynamics of COVID-19 with its several variants succeeding each other in affecting human beings (as well as animals), leaves those responsible for this study in learning environments with some difficult challenges: there are no ‘correct’ answers, only adequate approximations; different ideas and techniques, with the use of many different softwares, is undertaken with the certainty that to disagree is not only accepted, but stimulated, and welcomed. In general this attitude may not only go beyond existing syllabi, but it presents a pedagogical problem: many times there is only the problem in itself—and no mathematical questions are asked, as in many mathematics textbooks, as well as most textbooks in general. Questions must be provided by the modellers themselves so that, consequently, approximations of the desired solutions must inevitably include the collective and dialogical evaluation both from a mathematical point of view as well as from a social stance.In fact, Stillman, Brown and Geiger (2015), provided this challenging reference: Finding or generating a problem from a real world messy situation is a crucial cognitive step (Getzels, 1979). Besides this “mathematics of 'messy' needs”, Wolff-Michael Roth (2007)—referring to mathematical modelling and mathematics education—calls our attention to the fact that “... emotions are central to decision-making, 'motivation', learning and identity” which he presents as inherent to the modelling of real-world situations and, consequently, one can “no longer theorize mathematical modelling independent of emotions and emotional valence” (Roth, 2007). (Leash et al., p. 95) For the authors, in agreement with vast literature sources, this is especially true when, instead of proposed problems by the teacher whose main purpose is to teach a mathematical concept, it becomes necessary to face those problems that affect student, their communities or towns and countries, or, as is the case with the COVID-19 pandemic, the whole world. In this paper, we present the results of a modelling effort to describe successive models, studied and modified in order to learn about the dynamics of this disease in diversified societies with their dramatically different histories, cultures, governments, beliefs and resources. This activity was undertaken with a heterogeneous group of mathematicians and mathematics students at several levels of their courses, ranging from undergraduate to post-doctoral participants, in the search for an understanding of the multifaceted problem of the spread of the pandemic in so many different societies with so many different results. In Sect. 2, we very briefly mention the history of the kind of model that was (and still is) used for the description and simulation of the pandemic in a society, mathematically working out the details. It is necessary to mention that, in all this, technology plays a very important part. Historically, as described by Murray (2002), models had to be much simpler in order for it to be possible to use almost exclusively analytical tools. Nevertheless, this type of modelling effort was not only revolutionary, it also allowed professionals to understand some of the basic concepts, limitations, thresholds and possibilities, some of which are still valid. In Sect. 3, we present some modifications of the original modelling efforts in order to include other situations, such as the asymptomatic transmission of the SARS-CoV-2 virus, adopted social distancing policies, as well as widespread vaccination campaigns. The results are evaluated critically and we hope that a challenge will emerge: new and improved models, using better software, must continue to be developed, so that decision-makers may have an auxiliary tool in mathematics, mathematical modelling and technological resources. In Sect. 4, we present our conclusions, hoping that, in fact, these so-called ‘conclusions’ may serve as a starting point for other mathematicians. Historical background In 1927, 1932 and 1933, and building upon ideas of Ronald Ross and Hilda Hudson, A. G. McKendrick and W. O Kermack developed a theory that led to the use of compartmental models in the description and study of endemic and epidemic diseases. These models established an important trend in mathematical epidemiology. Their initial studies focused on cholera spread in the nineteenth century, which from the Ganges river delta, spread around the world. Besides studying the spread of the Spanish Flu, they also studied some aspects of several phenomena of the Bubonic plague, also called the Black Death, that entered Europe in the seventeenth century with almost unbelievable effects in terms of painful deaths throughout the entire continent (Kermack & McKendrick, 1927, 1932, 1933; Ross & Hudson, 1917). Possible similarities in the dynamics and the geographical spread of infectious diseases led us to the modelling of the behaviour of this kind of disease using previous studies by Ross and Hudson (1917). The model they developed needed relatively more complex mathematical tools since it used the ages of susceptible individuals in identifying incidence of infection as well as removal and transmission rates. These mathematical tools evolved and became what is identified today as the model of Kermack and McKendrick (1927, 1932, 1933). This is much more than a model: it opens up a theory with which many diseases can be analysed, studied, understood and simulated (Edelstein-Keshet, 2005). Whereas initial mathematical considerations focused only on the infected individual's etiology and evolution, Kermack and McKendrick's model considered society as a whole. The fundamental idea was that of dividing a population into separate compartments, acknowledging the fact that any individual must be in only one compartment at a time; and the movement from one of these compartments to another is used to describe, for example, the infection of a susceptible individual. In the initial model, the considered population was divided into three compartments, namely, that of the Susceptible (identified as S), one for Infected individuals (denoted I) and the third compartment of Removed individuals (denoted by R). In some instances, this third compartment refers to those individuals who were removed to a cemetery and, in other situations and models, the R stands for permanently resistant (as in measles, for example) or temporarily resistant (as in cases of the common flu). This can be illustrated with the following diagram (Fig. 1), which indicates that, in contact with Infected individuals, Susceptible ones also become Infected, and, after a certain time period, Infected individuals become removed, resistant or partially resistant.Fig. 1 Flow chart for the SIR model The causes for the movement from one compartment to another are based on the transmission through individual contact and the time-periods during which individuals remain infected, or the time-period of a temporary resistance. For university students, the Kermack and McKendrick model may be presented as a nonlinear system of differential equations, whereas at a high school level, a system of difference equations can be used, with which a discrete mathematical approach is not only completely adequate and feasible in terms of mathematics applications but also desirable in the concept of a trans-disciplinary educational practice. Effective modelling Schematically, Fig. 1 presents a visual model which indicates, albeit qualitatively, the spread of a disease in a certain population. It indicates that, as time passes, susceptible can become infected, infected become removed, which eventually become resistant, totally or partially. Using discrete difference equations, the dynamics of a disease in a society can be described identifying the movement individuals undertake when they stop belonging to one compartment and become individuals in another one. By using these difference equations, therefore, and assuming that a set of parameters can describe how this happens, we can describe the disease in a society using the system of equations given by the following:1 Sn+1-Sn=-βSI.Sn.In,In+1-In=βSI.Sn.In-γIR.In,Rn+1-Rn=γIR.In. Here, βSI and γIR respectively represent how contagion moves a susceptible individual into the compartment of infected individuals in the context of 1/(time.person) whereas γIR is the inverse of the time during which an individual remains infected. After this time-period an infected individual becomes removed or permanently resistant. A similar system can model a disease where resistance is temporary, a disease to which individuals have resistance for a period of time:2 Sn+1-Sn=-βSI.Sn.In+δSR.R(n),In+1-In=βSI.Sn.In-γIR.In,Rn+1-Rn=γIR.In-δSR.R(n). In this situation, δSR (when staying in compartment R is temporary) stands for the inverse of the time during which the temporary resistance remains. Some remarks are in order:In both of these systems, the considered disease has a short duration, so the number of individuals in the entire population remains constant, say, N. Then,Sn+In+Rn=N. A disease is considered endemic when we have, for whatever moment n in time after an initial period, S(n + 1) = S(n) = S, I(n + 1) = I(n) = I and R(n + 1) = R(n) = R, where the constant numbers S, I and R respectively represent the number of susceptible, infected and resistant individuals in the studied society for any given time. Due to the equation in the first remark, one has R(n) = N − S(n) − I(n), and the three-dimensional system can be reduced to a bi-dimensional one; in which casa the algebraic work becomes somewhat simpler. Working with these systems can be quite easy using adequate spreadsheet software. Also, a phase diagram can be used to illustrate different situations by using the x-axis for successive values of S(n) and the y-axis for successive values of I(n), and with these graphs, to analyse how different behaviours—which means different values for the parameter βSI, representing mask use, for example, lockdowns or vaccination procedures—affect a disease dynamic. In general, the necessary initial conditions that are used take into account the beginning of a disease, with only one infected individual: S(0) = N − 1, I(0) = 1 and R(0) = 0. In the first months of the COVID-19 pandemic, discrete equations (as in systems (1) and (2)) were very useful but, as time went by, this disease showed itself to last much longer. This led to a modification in the initial models and mathematicians resorted naturally to the use of ordinary differential equations in which, instead of successive discrete time steps, a continuous variable is used for time: the derivatives of S(t), I(t) and R(t) are used instead of the differences S(n + 1) − S(n), I(n + 1) − I(n) and R(n + 1) − R(n). The second model described above (for temporary resistance) becomes the following: For S(t), I(t), R(t), given S(0), I(0), R(0), and for t varying in the interval [0,T],3 dS(t)dt=-βSI.St.It+δSR.R(t),dI(t)dt=βSI.St.It-γIR.It,dR(t)dt=γIR.It-δSR.Rt. Here, the first model is obtained simply making parameter δSR = 0. In a mathematical sense, the first equation of system (3) describes the instantaneous change in the rate with which susceptible individuals become infected ones. The same with the second and third equations. These equations describe the movement per time unit from S to I, from I to R and from R back to I given by the dynamics of the disease, as presented in Fig. 2.Fig. 2 Flow chart for the SIR model The chart presented in Fig. 2 indicates that S, the number of susceptible individuals decreases as I, the number of infected individuals increases, and at the same rate, βSI. Analogously, the number of infected individuals decreases as the number of removed or resistant individuals increases, also by the same rate, δSR. As mentioned above, it is necessary to note that no increase in the total population is considered in this model, used for an illness that has a relatively short duration, such as measles, and not as in the case of Chagas’ or Hansen’s diseases, which are generally long-lasting phenomena. Considering (3) as the continuous form of (2) will permit us to analyse the equations that make up both systems and to obtain conclusions about the dynamics of the considered disease—and this can be done numerically and graphically, on the following two levels: systems (1) and (2) for a high-school environments and system (3) for university undergraduate students. But rather than trying to solve these non-linear systems (a very nice theoretical analysis and approximation can be found in Murray, 2002, 2003), the mathematical work developed in an adequate software environment and even in a spreadsheet will enable us to gather relevant information based on a modern, useful and interesting use of mathematics. Considering both the continuous (3) and the discrete systems presented by (1) and (2), a question that arises is that of recognizing when the number of infected individuals begins to decrease in a society—possibly leading to the disappearance of the disease. In other words, when is it that I(n) or I(t) decrease, or I(n + 1) < I(n)? The same kind of question might be asked of S, the set of susceptible, or of R, the set of resistant or temporarily resistant individuals, who after a certain period of time, return to the compartment of Susceptible: when do the populations is each compartment increase or decrease? To answer this question, we can rewrite the right-hand side of the second equation of system (2), in order to have a better working posture, as follows:In.βSI.Sn-γIR. We can then see that, since I(n) is a number of individuals and therefore greater than zero, which means that the population of infected individuals increases, then this product is positive wheneverβSI.Sn-γIR>0,orwhen Sn>γIRβSI. On the other hand, the number of infected persons in a disease decreases, i.e. I(n + 1) < I(n), when we have Sn<γIRβSI and this indicates that, mathematically, a disease begins to decrease only when (in a colloquial sense) you run out of susceptible individuals to whom the disease may be transmitted. The very same conclusion with the same inequation holds for the model that uses, continuously, system (3). Of course many other possibilities can and should be considered in analysing the dynamics of the studied disease. Nevertheless, in all its simplicity, this model is still used to identify a possible turning point in the spread of a disease. But since this model is for short-term diseases, one can always recover the dynamics of the R compartment simply by usingR=N-S+I, whether the model is discrete or continuous, allowing one to use curves in a plane to describe the interaction of the two compartments S and I. Using this ‘phase diagram’, we can observe the evolution of the behaviour of both populations simultaneously whether for discrete or continuous models (Fig. 3).Fig. 3 The phase plane for the SIR model: simultaneous behaviour of susceptible and infected individuals (developed in Wolfram Mathematica®) With the construction of phase planes like this one presented in Fig. 3, it can be remarked that, for whatever initial condition one begins with, the values of S and I tend to the same stationary point and, therefore, so does R. As we can see in this illustration, the model chosen for the simulation—with a temporary resistance—indicates the endemic situation, in which, after some time, the three compartments achieve a stationary point at which the population as a whole coexists with a permanent proportion of individuals in the susceptible, infected and temporarily resistant categories: the disease does not go away, rather, it is permanent in the population for as long as the parameters effectively describe the disease's behaviour. Another possibility is to present in the same graph the number of individuals in each one of the three compartments as a function (susceptible, infected and temporarily removed, as in Fig. 4) of the considered time steps.Fig. 4 Evolutionary dynamic of the SIR model With this graph, the observation made for Fig. 3 still holds, since S, I and R tend to a stationary situation in which the disease co-exists with the whole population at a constant level. In the graph presented in Fig. 4 we can verify that the three considered compartments do, in fact, reach an equilibrium which indicates an endemic situation. Now this model, simple as it may be, does indicate what happens if the value of βSI decreases, meaning that transmissibility is smaller as is the case, for example, if the use of masks, through some public sanitary policy, becomes mandatory, or at least, usual. Students can see this, when, using a simple programme, the results are obtained when for two different values of βSI, different equilibrium stages are assumed, either with a continuous system or with a discrete one. This same model can be used—although not in COVID-19 situations—with a simpler mathematical expression, for the study of other diseases, such as measles, mumps, and chickenpox, for example. The model may (with no loss of its generality, as seen above) consider exclusively susceptible and infected individuals. This happens for many reasons, two of which (the main ones) are as follows: this type of model, which can be used to understand rapid epidemic surges that do not last for a long time in society—so that the increase of the latter need not be considered and, secondly, the case in which, since resistance is permanent, the return from the R compartment to the S one does not exist. Therefore, the model can be used with only two compartments, S and I. We will then have, for the discrete model, also called an SI model,4 Sn+1-Sn=-βSI.Sn.In,In+1-In=βSI.Sn.In, where n represents the time-step. System (1) has now become a nonlinear one with only two equations, a form of system (4) which is easier to ‘see’ or with which it can be easier to ‘read’ the world. To use the model, students in high school (and those who wish to become mathematics teachers) can use what has been learnt for sequences (usually one-dimensional), to accompany the behaviour of a disease in society, thereby developing mathematical concepts before actually formalizing them. In fact, this process may be described as intuitively learning mathematics. On the other hand, for undergraduate students in teacher education, the introduction of this kind of system of discrete difference equations as well as differential ones presents mathematics not as an objective in itself, but rather as an important and useful tool with which to understand life, effectively showing with a present and urgent situation, the educational value of working out these models (i.e., systems of both difference and differential equations). For the authors, situations and mathematical approaches such as those described here permit students at many levels to understand that the boundaries between themes such as mathematics and epidemiology are not crisply defined, but rather fuzzy—as always happens in real life situations. Besides this, what else can be accomplished with the presentation and use of models like these (and the equivalent programmes)? Well, an important aspect is that different situations, derived from this initial one, can be proposed. One challenge is presented in the next section. The inclusion of vaccination What can be accomplished with the presentation of this model? One challenge would be that of considering a vaccination programme that modifies the model presented in (2), with the inclusion of a new compartment, which includes the vaccinated population with a temporary immunity, as seems to be the case for all COVID-19 vaccines. This model requires another compartment, besides that of Susceptible, Infected and Temporarily Resistant individuals—that of Vaccinated individuals. An illustrative chart for this new proposition, is presented in Fig. 5.Fig. 5 Flow chart for the SIRV model This figure has the purpose of describing the modification of the previous models with the introduction of a new compartment, that of vaccinated individuals. This picture shows that besides the previous movements, there is also a movement from the compartment of susceptible individuals to that of vaccinated ones. And from the latter back to the former, due to the end of the validity of the immunization given by the vaccine. This chart indicates the passage from the compartment of susceptible individuals to infected ones, as well as the return from temporary resistance and, besides the return from temporary resistance R back to Susceptibles, also the passage from Susceptibles to Vaccinated and back. Borba (2021) suggested that we use the COVID-19 pandemic as a mathematical tool, for example using the application of exponential functions to explain the spread of the coronavirus. However, we propose using the pandemic to obtain a function that describes the behaviour of vaccination. This illustrates the fact that, in real-life situations, we use what we can to study and understand phenomena. And, sometimes, we create mathematical tools in order to study and understand these facts in a better way. In Brazil, the National Vaccination Campaign against COVID-19 started on January 18, 2021, and four vaccines are currently available in Brazil. Ever since the first vaccines appeared, they have had a different vaccination schedule and variations in effectiveness (Prüβ, 2021). The graph presented in Fig. 6 provides us with information about the overall immunization behaviour as a function of time for Brazil.Fig. 6 Number of people in Brazil during 2021 according to the number of vaccination doses (Our World in Data, 2021) As the vaccination data are dynamic and there was a step in the number of people with full immunization as of May 25, we considered the data from that day onwards for adjusting an exponential curve to the data, using the least squares method. The question arises: how can students work with this type of data? The main focus is on the analysis of the dynamics of a vaccination policy. A relationship between time and the number of vaccinated persons could permit simulations to be made, and different vaccination policies to be studied for such a functional relationship to be found; a good strategy could be to obtain the best simple linear regression model, observing that the ‘best’ model is subjective to the chosen tools, as is the case with many so-called ‘objective’ mathematical results. The given data for the considered time-period can be described by a curve with exponential characteristics. This adjustment demands technological resources and this necessity for technology increases with the complexity of the chosen curves for adjustment using least square methods. The chosen curve relating a dependent variable y (the number of vaccinated individuals) to a single independent variable x (time) is given by the following:yx=a.eb.x, With this choice comes an implicit warning about the validity of mathematical modelling with regard to the range of variation of the variables, both dependent and independent: these adjustments serve the modelling process in so far as the modelling assumptions remain valid. The graph of the comparison of the number of people who received the second dose of vaccines and the obtained curve (by the least square method) is shown in Fig. 7.Fig. 7 Number of people with full immunization and the adjusted curve (Our World in Data, 2021) Adapting the original model to include vaccination Although the possibility of vaccinated individuals contracting the SARS-CoV-2 virus exists in relevant quantities, the simplified model adopted here will suppose that vaccinated persons, after a period during which any individual remains immune, become susceptible again. As in the previous example, we assume that the total population does not change significantly, so that S(t) + I(t) + V(t) + R(t) = N, a constant value. This demands the creation of the above-mentioned compartment for persons that are neither susceptible, nor infected nor temporarily resistant: the compartment of vaccinated individuals. When knowing the values of S(0), I(0), R(0) and V(0), a discrete mathematical model for obtaining the successive values of S(n), I(n), R(n), V(n) can be developed. It is similar to that of system (3), with the presence of the vaccinated individuals as described previously. And besides using the same parameters representing the very same movements between compartments, system (5) includes υSV and υVS indicating the movement of susceptible individuals to the compartment of vaccinated ones and, after the period of immunization, returning to the susceptible state.5 Sn+1-Sn=-βSI.Sn.In+δSR.Rn-ϑSV.Sn+ϑVS.V(n),In+1-In=βSI.Sn.In-γIR.In,Vn+1-Vn=ϑSV.Sn-ϑVS.Vn,Rn+1-Rn=γIR.In-δSR.Rn, where n represents the time-step. A new figure can be presented with all four compartments changing in time (with the same parameters as in system (5)) in order to be able to understand how vaccination programmes can affect the final results. For this, we can use, instead of system (5), a system of ODE's, in a continuous context:6 dS(t)dt=-βSI.St.It+δSR.Rt-ϑSV.St+ϑVS.V(t),dI(t)dt=βSI.St.It-γIR.It,dV(t)dt=ϑSV.St-ϑVS.Vt,dR(t)dt=γIR.It-δSR.Rt, for t varying in the interval [0,T]. There is a justification for the use of both kinds of modelling efforts. If, on the one hand, data are collected daily—which is to say in a discrete way and, therefore with difference equations—on the other hand, the long period of time permits the use of a continuous model and, consequently, using differential equations. A study of the phase planes of both systems (5) and (6) leads to the same kind of graph and very similar numerical values. A use of system (6) would probably not be possible in a high school environment. And a spreadsheet would only be useful for discrete systems such as (5), whereas, for simulating system (6) there is the need to approximate numerically the solutions of ordinary differential equations. In spite of the possibility of using a spreadsheet for a discrete system, the results presented here were obtained using Wolfram Mathematica®, in which numerical methods for ordinary differential equations were used in the approximations. Although the general aspect is qualitatively similar to that of the SIR model as seen in previous figures, in Fig. 8 it can be seen that the endemic stationary situation happens with a smaller population of infected individuals. In other words, the disease tends to an endemic situation (called a ‘steady state’) in which the number of individuals in each of the compartments remains the same.Fig. 8 The phase plane for SIRV model, considering only S and I (developed in Wolfram Mathematica®) This graph (Fig. 9) illustrates the positive effect of an adequate vaccination policy, showing a much smaller population level for infected individuals than the steady-state values in the graphs shown in Figs. 3 and 4, where the model does not include vaccination, in spite of the fact that the utilized vaccine does not guarantee permanent immunity for the disease.Fig. 9 Evolutionary dynamic of the SIRV model Modelling at this level, and using technologies, be it to approximate the behaviour of systems of difference equations or of systems of ordinary differential equations, will permit modelling techniques and attitudes to be tested, to be evaluated and to have results criticized both mathematically and socially. In the next section, a further modification is discussed, always in the same spirit of trying, testing, checking… and learning. The purpose of this modification is to show how a useful model can be extended by including, or eliminating, aspects such as compartments, parameters, relationships, as is done in the next section. Modelling including confinement Another possible (and necessary) modification can very well be that of having an additional compartment of those people who can stay at home, respecting social distancing in a disciplined manner. In such a case, it must be considered that for many people, this is not possible: hospital and sanitary professionals, bus and truck as well as taxi drivers, cannot operate from a home office. These and so many other professionals must be ready to face the pandemic, depending mostly on individual protection equipment, as well as hoping that others also have similar socially responsible behaviour and attitudes. Figure 10 presents this new situation in which social distancing is respected by those who have the opportunity to do so. The chart describes a situation in which, besides compartments of susceptible, infected and removed individuals, there is another compartment of confined persons. confined persons are also subject to the possibility of catching the disease, even with a much smaller possibility and, therefore, the passage from confined to infected can exist, as can also happen that due to social or economic pressures, individuals may leave the compartment of confined persons returning to the compartment of susceptible ones. The diagram, however, takes into account that the rate with which those in susceptible move to confined is, in fact, the difference between those that go from S to C minus those that return from C to S.Fig. 10 Flow chart for the SCIR model Using a system of difference equations, with the intention of working this out quantitatively at a high school level, with adequately simple spreadsheet software, we may very well have a discrete mathematical model given by the following: For S(n), C(n), I(n), R(n) and knowing S(0), C(0), I(0) and R(0), we have7 Sn+1-Sn=-βSI.Sn.In-αSC.S(n)+δSR.Rn,Cn+1-Cn=αSC.Sn-βCI.Cn.In,In+1-In=βSI.Sn.In+βCI.Cn.In-γIR.In,Rn+1-Rn=γIR.In-δSR.Rn, where n represents the time-step. Again, as in previous settings, this can be worked out numerically with a relatively simple theoretical explanation. This is qualitatively indicated in Figs. 11 and 12. If it becomes necessary to use, instead of system (7), a system of ODE's, we will have the following system:8 dS(t)dt=-βSI.St.It-αSC.S(t)+δSR.Rt,dC(t)dt=αSC.St-βCI.Ct.It,dI(t)dt=βSI.St.It+βCI.Ct.It-γIR.It,dR(t)dt=γIR.It-δSR.Rt, for t varying in the interval [0,T].Fig. 11 The phase plane for SCIR model: simultaneous behaviour of susceptible and infected cases (developed in Wolfram Mathematica®) Fig. 12 Evolutionary dynamic of the SCIR model The results of system (8) are qualitatively indicated in Figs. 11 and 12. Figures 11 and 12 also indicate an endemic situation in which, as in previous ones, the disease remains affecting society, with all its dire consequences, even if the level of infected populations is low with respect to that of other compartments. As in both previously presented situations, an endemic situation is obtained, and COVID-19 (or any analysed disease) will remain within society—and society will have to learn to live with this presence. These previous modelling efforts, however, demonstrate that employing only vaccines or only social distancing, cannot, in general, eliminate a disease such as the one theoretically described here. The use of convenient software (or, preferably, of adequate freeware) can enable us to consider a continuous modelling effort, using an analogue system of ordinary differential equations, such as the one described in system (8). All this should possibly lead us to a new step in the modelling process, namely, that of considering a combination of social distancing with self-confinement together with a vaccination programme. Modelling including confinement and vaccination The chart in Fig. 13 illustrates a situation in which the susceptible are vaccinated but the immunity is only for a period of time, after which individuals become susceptible again. Besides that, it considers that isolated individuals may still be infected albeit in a very much smaller proportion than that of non-confined individuals.Fig. 13 Flow chart for the SCIRV model For S(n), C(n), I(n), R(n), V(n) and S(0), C(0), I(0), R(0) and V(0), we have9 Sn+1-Sn=-βSI.Sn.In-αSC.S(n)+δSR.Rn-ϑSV.Sn+ϑVS.V(n),Cn+1-Cn=αSC.Sn-βCI.Cn.In,In+1-In=βSI.Sn.In+βCI.Cn.In-γIR.In,Vn+1-Vn=ϑSV.Sn-ϑVS.Vn,Rn+1-Rn=γIR.In-δSR.Rn, where n represents the time-step. Assumptions are similar to those of the previous models in order to emphasize possible and many times necessary adaptations and modifications, in order to include different considerations in the modelling processes. We can use, instead of system (9), the following system of ODE's:10 dS(t)dt=-βSI.St.It-αSC.S(t)+δSR.Rt-ϑSV.St+ϑVS.V(t),dC(t)dt=αSC.St-βCI.Ct.It,dI(t)dt=βSI.St.It+βCI.Ct.It-γIR.It,dV(t)dt=ϑSV.St-ϑVS.Vt,dR(t)dt=γIR.It-δSR.Rt, for t varying in the interval [0,T]. In this case, there are some relevant comments. The first one is that the parameter that describes the movement of confined individuals to the compartment of infected ones is significantly smaller than the one that describes the movement from susceptible to infected compartments. Another comment is that this model could possibly be improved, considering the fact that confined persons may return to the compartment of susceptible individuals facing social pressures, as mentioned before. As in other modelling efforts with other hypotheses, this graph (Fig. 14) indicates that an endemic situation is reached in the relative quantities of individuals in the Susceptible and Infected categories.Fig. 14 The phase plane for SCIRV model: simultaneous behaviour of susceptible and infected (developed in Wolfram Mathematica®) As in the Fig. 14, this graph (Fig. 15) clearly indicates an endemic situation in which the disease remains in society—unless new measures appear, introducing the need for changes in the present models.Fig. 15 Evolutionary dynamic of the SCIRV model A comparison of models In this section, we present the different types of evolutionary behaviour of the models we have discussed previously. This is not in any way a comprehensive list of models—there are, in fact a vast number of other choices for this modelling (Aguiar, et al., 2020; Amaku, et al., 2021; Ciufolini & Paolozzi, 2020; Thomas, et al., 2020; Zeb, et al., 2020). Also, there are other types of modelling efforts using different mathematical tools and numerical procedures. It is not only the choice of different models that can adapt the mathematical work of different situations and contexts. Even choosing to use only non-linear systems of ordinary differential equations, it is necessary to consider that very important changes in the parameters may be necessary and may have surprising results. For example, the choice of having the transmissibility parameter varying in time in order to simulate successively changing virus strains, as well as varying social attitudes, will modify the rates of change in confined and vaccinated compartmental populations, thus changing the outcomes of the results. For the models cited above, a comparative consideration can and should be used. From an educational point of view, these graphs can motivate a critical evaluation of the modelling efforts, as well as their capability in simulating social scenarios. And we believe that qualitative illustrations, such as that in Fig. 16, may very well be used to assess and evaluate social and sanitary policies chosen and adopted by authorities. Figure 16 illustrates, therefore, the tendencies according to different models and possible outcomes.Fig. 16 Number of the Infected people using different models A comparison with and a subsequent adaptation to real-life data, as undertaken by Meyer et al. (2021), can be quite useful in testing values, decisions, policies. Of course, when a model is forced to agree with real-life information, there is always a risk of ‘'spoiling’' simulations, since models demand empirical as well as heuristic evaluations and corrections. As we mentioned before, another challenge is to use simplicity in altering models of this kind in order to test scenarios. For example, we could very well consider a model in which vaccinated individuals, after a period of time, can be considered as temporarily resistant to the disease and, therefore, do not return immediately to the compartment of Susceptible, passing previously by the compartment of temporarily resistant persons. The pedagogical value of these models on the pandemic COVID-19 For the authors, besides some comments in the introduction (as well as elsewhere) on the importance of pedagogical values evident in working with a sequence of different mathematical models for real-life situations, there is another aspect to be mentioned. In lieu of a meaningless learning of mathematics, working with the models described here, as well as with their viable modifications, can present learners with a mathematics of necessities, to be used as an instrument to understand the world around us—and to change it. Therefore, not only in the sense that these mentioned models can enable students, also working in groups, to use mathematics to understand, criticize and simulate different governmental actions to combat the pandemic as well as to evaluate several actions that were (or that were not, in some countries) adopted in combating COVID-19. That is to say that working with these models and procedures can be considered as Mathematics for Education and not the other way round. Using different mathematical tools, researchers world-wide have evaluated governmental attitudes in terms of the criteria for adopting public health policies (an example is that of Silva & Sagastizábal, 2021, with results for some regions in Brazil). Another important factor for learners and students is the fact that, in the majority of situations, the exhibition of apparently precise numerical results of mathematical modelling of phenomena, such as those related to the pandemic, say very little in different areas, to the media and to many professionals in public health about possible simulations and results, and mathematics is not, in any way, exact (in spite of what common sense states). Instead, the presentation of qualitative results, of possible tendencies, of illustrative graphs, albeit inexact, may very well display what the mathematical results are about, in the sense of the ‘what if’ questions, which should arise in the simulations; and those cases should be discussed by teachers with their students, in order to adjust the mathematical modelling to describe the real scenarios in a better way. There is a third pedagogical value that can justify the use of the modelling of social and natural phenomena in the learning of mathematics and where it is needed: that is, using simulations and corresponding graphs can lead to a better interpretation of data. There are almost infinite links to data on the internet. These data, in order to become information, need mathematics and, in very many cases, mathematical modelling of the problem under study. It goes without saying that the same number can very well represent a catastrophic sanitary problem in one place, while showing an inevitable improvement in other regions. In this case, both the use of derivatives, as is the case with differential equations, and the use of discrete differences, which are the basis for difference equations, can immediately identify very different tendencies and, in so doing, permit a dynamical description of the problem, in spite of some similar numerical results. For the authors, there is a fourth and important factor in the use of this type of modelling in classroom and school activities: models can and do function while situations remain the same. A very small change in the transmissibility of a virus may result in a substantial modification of the complete model. In other words, applying mathematics to real-life situations through some mathematical modelling effort is almost never the same in time, neither does it necessarily hold when space variables change: mathematical modelling consists in constantly evaluating results and adjusting procedures, models, equations, and instruments whenever necessary. This section was used to emphasize that the use of mathematical modelling of social and environmental phenomena does have a very important place in the learning of mathematics, as well as in school activities. Four possibilities were mentioned, out of many other possible aspects, as follows: the mathematics of necessity, resulting from challenges to being able to study and understand a developing and demanding world; the inexactness of mathematics allied to its strength in simulations which can be decisive in evaluating social and natural situations; the always urgent need of interpreting data to obtain faithful information; and the ever-changing real world impinging on the dynamics of mathematical modelling. Finally, there is another provocative remark: in these situations, here described and mentioned, there is generally no specific mathematical question. While students learn how to obtain the unique and exact answers to classroom problems, real life demands that they be able to ask the right questions as well. Conclusions The main objective of the work described in this paper is to present modelling techniques that students can use to learn (1) to adapt models to slight changes and prepare for the creation of new models, (2) to evaluate modelling choices of state variables, parameters and mathematical tools, (3) to recognize modelling limitations and, most certainly, to understand critically how the use of mathematical models, also employing technological and numerical strategies, can be useful for learning about a certain problem and understanding its aspects, and for simulating scenarios for social and natural decisions as regards human actions in society and nature. In the introduction, the expression ‘messy’ was used for real-life situations, an evaluation to which both authors agree. On the other hand, in a text written by Kaiser et al. (2011), (with a very appropriate title in its reference to “Authentic Modelling Problems in Mathematics”) the readers' attention is called not only to the mathematical difficulties but also to the problem of choosing an adequate hypotheses for modelling a studied situation. Attitudes, as described in the mentioned text, which “promote the whole range of [learning] modelling competencies” as well as enhancing the need for students to take action (this last phrase is an interpretation of the text), must be learnt. It is not always the complex mathematical outlook which leads to the best results, even though, in general, sophisticated mathematics as well as advanced software packages do result in better understanding and evaluations. This is especially true in an environment where mathematics is used for studying both non-mathematical as well as mathematical situations and creating knowledge for relevant social responsibilities. The modelling efforts presented and discussed above can, both in a high school and at undergraduate levels, motivate the use of mathematics, coupled with other scientific outlooks, creating an essential, albeit auxiliary tool, which is, in fact, an important support for decision-making procedures in a transdisciplinary situation. Another purpose of this text, possibly implicit in all that is stated, refers to the fact that a mathematical model as well as mathematical modelling is not necessarily ‘good’ per se. It does, in fact, promote learning, evaluation, decision-making, and critical analysis concerning mathematical processes—but in which context or in which contexts? Now this is put forth from the point of view of mathematics education, since a model can be very ‘good’ in its efficiency, regardless of ethical consequences. Some years ago, in the questions-and-answers period after a conference, Professor Nelson Maculan, who was the chancellor of the Rio de Janeiro Federal University, said that in his childhood, the photograph from the simple graduation ceremony of his junior school was taken, as was the case for all his little colleagues, with a coloured terrestrial globe beside each child. He went on to say that, for older students, the equivalent of the globe, which should be used today against negationist theories, was the mathematical model (Maculan Filho, 2018). The authors tend to agree with him, but it is also true that the attitude behind modelling in a school environment must always take into account that mathematics is an essential tool in the modelling process, and that it can (and should) become a very important tool in the sense of Mathematics for Education. Mathematical Modelling has always been used, for example, in creating and improving mass destruction artifacts, in the production of toxic material which is said to be useful in the culture of cereal grains, in calculating the increase in profits coupled to both the increase in quantities as well as the reduction of quality in so many areas of human activities. The model can be an efficient tool, but wielding it correctly needs education in a critically ethical sense, education for effective human development, or, as a UNESCO study suggests, “Learning to study, inquire and co-construct together (learning to learn), Learning to collectively mobilize, Learning to live in a common world and Learning to attend and care” (Sobe, 2021, p. 1) or, in the words of Rodrigues (2021), “Learning to know, learning to do, learning to live together and learning to be” (p. 3). This attitude identifies the perspective of mathematical modelling with what it entails for an education to improve the context for humanly relevant values, for society and nature. Villa-Ochoa and Berrío (2015) affirmed that, (in the authors’ words) besides, “[activating] other dynamics of the individual knowledge of some members of a culture”, the individual’s knowledge “[is] to be discussed, … socially organized, thus becoming a body of knowledge which is a response to its members' needs and will” (p. 249). This perspective places a challenge for the educational use of mathematical modelling that does, in fact, reach quite farther than the useful and necessary learning of mathematical concepts and processes. There are, of course, many other models and possible approaches for infected diseases in a human-to-human transmission, in vector-born infections, and in situations in which these mentioned vectors provide the contact between humans and natural hosts. Simpler models are not always the best, but they are necessary for the procedure that might eventually lead to important tools for understanding, studying, analysing, and simulating infectious diseases in many environments. Besides, simpler models can and do lead the way to the learning of mathematical concepts and uses, and they enable mathematics students, most especially those who are learning to become mathematics teachers, not only to exhibit the power and the necessity of mathematics for developing the posture of conscientious citizens, but to interact in a transdisciplinary way with other knowledge, other points of view, other backgrounds and cultures, as well as so many other ways for coping with the challenge of living in an ever-changing world. In other words, this enterprise involves understanding how mathematics is, together with other sciences, essential in reading the world, creating knowledge, evaluating it, and using it for a better society and a better environment. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 986 KB) Acknowledgements M. Lima thanks CAPES (Grantee 88887.369830/2019-00) for financial support and the authors would like to thank IMECC—Unicamp for support in the development of this work. Besides, it is necessary to acknowledge the reviewers and the editor-in-chief for comments, insights, and precious help in formulating this text. Declarations Conflict of interest The authors declare no conflict of interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Aguiar M Ortuondo EM Bidaurrazaga van-Dierdonck, J. Modelling COVID 19 in the Basque country from introduction to control measure response Scientific Reports 2020 10 17306 10.1038/s41598-020-74386-1 33057119 Amaku M Covas DT Coutinho FAB Modelling the impact of delaying vaccination against SARS-CoV-2 assuming unlimited vaccine supply Theoretical Biology and Medical Modelling 2021 18 14 10.1186/s12976-021-00143-0 34325717 Borba MC The future of mathematics education since COVID-19: Humans-with-media or humans-with-non-living-things Educational Studies in Mathematics 2021 108 385 400 10.1007/s10649-021-10043-2 34934229 Ciufolini I Paolozzi A Mathematical prediction of the time evolution of the COVID-19 pandemic in Italy by a Gauss error function and Monte Carlo simulations The European Physical Journal plus 2020 135 Article ID 355 10.1140/epjp/s13360-020-00383-y D'Ambrósio, U. (1998). Etnomatemática : Arte ou técnica de explicar ou conhecer. Ática. Edelstein-Keshet L Mathematical models in biology 2005 Society for Industrial and Applied Mathematics Freire P Educação como prática da liberdade 1967 Paz e Terra Getzels JW Problem finding: A theoretical note Cognitive Science 1979 3 2 167 172 10.1207/s15516709cog0302_4 Kaiser, G., Schwarz, B., & Buchholtz, N. (2011). Authentic modelling problems in mathematics education. In G. Kaiser, W. Blum, R. Borromeo Ferri & G. Stillman (Eds.), Trends in teaching and learning of mathematical modelling (ICTMA14) (pp. 585–594). Springer. Kermack WO McKendrick AG A contribution to the mathematical theory of epidemics, part. I Proceedings of the Royal Society b: Biological Sciences 1927 115 700 721 Kermack WO McKendrick AG Contributions to the mathematical theory of epidemics, part. II Proceedings of the Royal Society b: Biological Sciences 1932 138 55 83 Kermack WO McKendrick AG Contributions to the mathematical theory of epidemics, part III Proceedings of the Royal Society b: Biological Sciences 1933 141 94 112 Maculan Filho, N. (2018). Oral communication, Invited Conference at the Institute for Mathematics, Statistics and Scientific Computing (IMECC) at the State University of Campinas (UNICAMP), Campinas. Meyer JFCA Lima M Espitia C Longo F Laiate B Gois A Kunz C Different approaches to the modelling of CoViD-19 Trends in Computational and Applied Mathematics 2021 10.5540/tcam.2021.022.04.00515 Murray JD Mathematical biology I 2002 Springer Murray JD Mathematical biology II 2003 Springer Our World in Data. (2021). Brazil: coronavirus pandemic country profile, from https://ourworldindata.org/coronavirus/country/brazil. Accessed 30 Nov 2021. Prüβ BM Current state of the first COVID-19 vaccines Vaccines 2021 9 30 10.3390/vaccines9010030 33429880 Rodrigues ZB Education: A study based on the UNESCO report on the four pillars of knowledge Revista Científica Multidisciplinar Núcleo Do Conhecimento 2021 1 4 53 60 10.32749/nucleodoconhecimento.com.br/education/four-pillars Ross R Hudson HP An application of the theory of probabilities to the study of a priori pathometry. Part I Proceedings of the Royal Society of London 1917 A93 225 240 10.1098/rspa.1917.0015 Roth WM Leash RA Hamilton E Kaput JJ Mathematical modeling ‘in the wild’: A case of hot cognition Foundations for the future in mathematics education 2007 1 Lawrence Erllbaum Associates 77 97 Silva, P. J. S., & Sagastizábal, C. (2021). Vidas salvas no Brasil pelo isolamento social e uso de máscaras (descontinuado), from https://www.ime.unicamp.br/~pjssilva/vidas_salvas.html. Accessed 5 May 2022. Sobe, N. W. (2021). Reworking four pillars of education to sustain the commons, UNESCO Futures of Education Ideas LAB, from https://en.unesco.org/futuresofeducation/ideas-lab/sobe-reworking-four-pillars-education-sustain-commons. Accessed 10 Feb 2022 Stillman GA Brown JP Geiger V Stillman GA Blum W Biembengut MS Facilitating mathematisation in modelling by beginning modellers in secondary school Mathematical modelling in education research and practice—cultural, social and cognitive influences 2015 1 Springer 93 104 Thomas DM Sturdivant R Dhurandhar NV Debroy S Clark N A primer on COVID-19 mathematical models Obesity 2020 28 8 1375 1377 10.1002/oby.22881 32386464 Villa-Ochoa JA Berrío JB Stillman GA Blum W Biembengut MS Mathematical modelling and culture: An empirical study Mathematical modelling in education research and practice—cultural, social and cognitive influences 2015 1 Springer 93 104 Williams J Goos M Clements MA Bishop AJ Keitel C Kilpatrick J Leung FKS Modelling with mathematics and technologies Third international handbook of mathematics education 2013 1 Springer 549 569 Zeb A Alzahrani E Erturk VS Zaman G Mathematical model for coronavirus disease 2019 (COVID-19) containing isolation class BioMed Research International 2020 10.1155/2020/3452402	36475296	PMC9716171	NO-CC CODE	2022-12-03 23:20:54	no	ZDM. 2022 Dec 2;:1-14	utf-8	ZDM	2,022	10.1007/s11858-022-01447-2	oa_other
==== Front Open Econ Rev Open Economies Review 0923-7992 1573-708X Springer US New York 9696 10.1007/s11079-022-09696-0 Article The Impacts of Financial Crises on the Trilemma Configurations Aizenman Joshua [email protected] 1 Chinn Menzie [email protected] 23 http://orcid.org/0000-0002-5440-1526 Ito Hiro [email protected] 4 1 grid.42505.36 0000 0001 2156 6853 Economics and International Relations, University of Southern California and NBER, University Park, Los Angeles, CA 90089-0043 USA 2 grid.28803.31 0000 0001 0701 8607 Department of Economics, University of Wisconsin and NBER, 1180 Observatory Drive, Madison, WI 53706 USA 3 Robert M. La Follette School of Public Affairs, Madison, WI USA 4 grid.262075.4 0000 0001 1087 1481 Department of Economics, Portland State University and Research Institute of Economy, Trade and Industry (RIETI), 506 SW Mill Street, Urban Center Building, #450, Portland, OR 97201 USA 2 12 2022 139 12 10 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Over the years, policymakers have explored various combinations of varying degrees of monetary policy independence, exchange rate stability, and financial openness while recognizing that not all three policies can be achieved to the fullest extent – the “monetary trilemma” hypothesis. In recent years, holding international reserves (IR) has become an important policy instrument as a buffer or insurance against liquidity shortages. Significant and fundamental economic events such as currency crises have often changed the policy mix. In this paper, we find that countries’ policy mixes have been diverse and varied over time from the perspective of the trilemma and also IR holding. We then illustrate how the combination of the three trilemma policies and IR holding drastically changed before and after the Asian Financial Crisis (AFC). However, the Global Financial Crisis did not lead to a drastic change in the policy arrangements. We find that countries that faced large terms of trade shocks or negative economic growth during the crisis increase IR holding in the post-AFC. Countries that had negative growth during the crisis also tend to pursue more exchange rate flexibility and more open financial markets. This characteristic is true for commodity exporters, but not for manufacturing exporters. Countries with large current account deficit (i.e., “large capital borrowers”) tend to be more sensitive to economic growth at the time of the AFC. Countries that are under IMF stabilization programs or those with sovereign wealth funds tend to hold more IR. These characteristics were not found in the aftermath of the GFC. In general, countries increased their IR holdings after the GFC, but did not respond to the during-crisis economic and institutional conditions. Supplementary Information The online version contains supplementary material available at 10.1007/s11079-022-09696-0. Keywords Globalization Trilemma Financial crisis JEL Classification Nos. F31 F32 F33 F41 ==== Body pmcIntroduction Achieving stable and sustainable economic growth is a long-standing goal of economic policymakers. With respect to open macro policy, policymakers have coordinated a combination of different degrees of monetary policy independence (MI), exchange rate stability (ERS), and financial openness (FO), but not all three policies can be achieved to the fullest extent – i.e., the “monetary trilemma hypothesis” (Fig. 1).Fig. 1 The “Monetary Trilemma” Sudden and fundamental changes in the economic environment due to major economic events, such as currency crises or changes in the international monetary system, have caused policymakers to change the mix of the three trilemma policies. After the European powers, the United States, and Japan left the gold standard in the 1930s, the international monetary system after World War II shifted to the Bretton Woods system. That means in the context of the trilemma, economic major powers replaced a system with full FO and ERS and zero MI with another system with full ERS and MI and zero FO as a response to the economic turmoil in the 1930s. As financial globalization progressed in the 1980s first in the advanced economies (AE) and in the 1990s in less developed countries (LDC), the influence of financial markets has become significant in the global economy. After emerging market economies (EMEs) experienced financial crises in the late 1990s and the early 2000s, how to maintain financial stability has become an important policy objective. It has been argued that financial liberalization is a double-edged sword. While financial opening would alleviate financial repression, promote more efficient allocation of financial resources, and thereby enable higher economic development, opening financial markets for cross-border capital can also exacerbate boom-bust cycles. Financial volatility has become a key barometer in open macro management. How to ensure that financial volatility does not affect the real economy and how to maintain smooth consumption against potentially volatile income flows have become important policy concerns. In addition, financial globalization has also made EMEs sensitive and vulnerable to changes in financial conditions in the center-economies, most namely the United States. Thus, countries’ macroeconomic conditions have become more sensitive to the “global financial cycles” in capital flows, asset prices, and credit growth. In the late 1990s and early 2000s, after witnessing crisis-ridden EMEs experience severe hard currency shortages, many developing countries, especially emerging economies, began to hoard international reserves (IR) as a line of defense against financial instability. In addition, the fact that the International Monetary Fund (IMF) imposed stringent conditionalities on crisis-ridden economies seeking for bailouts has led EMEs to avoid viewing the IMF as a potential source of emergency funds and regard IR holding as self-insurance against potential financial instability. China is undoubtedly a prime example of a country that is hoarding large amounts of IR for insurance against financial instability. Thus, after the EME crises, IR have become an important policy in addition to the trilemma-based three open macro policies. There do not seem to be many developing countries where the configuration of the trilemma policy variables has drastically changed in the aftermath of the Global Financial Crisis (GFC) of 2008–09. That might be partly because the epicenter of the crisis was the U.S. and several other European countries, and because the impact of the crisis on developing economies was rather uniform, not regionally concentrated like the case of the AFC. That raises a natural question of how the post-crisis response to the GFC of countries’ trilemma policy arrangements and IR holding differ from those to the AFC. In this paper, we first illustrate the development of the international monetary system in the last five decades from the perspective of the trilemma and also IR holding. We then investigate how the combination of the three trilemma policies and IR holding changed before and after the major crises, namely, the AFC and the GFC. Lastly, we examine what kind of economic and institutional factors lead to changes in the policy configurations. We start with Sect. 2 where we review the theory of the monetary trilemma and the development of the international monetary system in the post-Bretton Woods era from the theory’s perspective. We also discuss the role of IR holding as a fourth variable of the open macro policy configurations. Using the “diamond charts,” We examine how the configuration of the four variables changed over the AFC of 1997–98 and the GFC of 2008–09. In Sect. 3, we conduct a formal empirical analysis of the development of the four policy variables at the time of financial crises. Lastly, we examine what kind of economic and institutional factors lead to changes in the policy configurations. In Sect. 4, we make concluding remarks. The Trilemma Theory and Evidence The Trilemma Hypothesis The trilemma is illustrated in Fig. 1. Each of the three sides of the triangle—representing monetary independence, exchange rate stability, and financial integration—depicts a potentially desirable goal, yet it is not possible to be simultaneously on all three sides of the triangle. For example, the top vertex, labeled “floating exchange rate,” is associated with the full extent of monetary policy autonomy and financial openness, but not exchange rate stability. History has shown that different international financial systems have attempted to achieve combinations of two out of the three policy goals, such as the Gold Standard – guaranteeing capital mobility and exchange rate stability – and the Bretton Woods system – providing monetary autonomy and exchange rate stability. The fact that economies have altered the combinations as a reaction to crises or major economic events may be taken to imply that each of the three policy options is a mixed bag of both merits and demerits for managing macroeconomic conditions. Greater monetary independence could allow policy makers to stabilize the economy through monetary policy without being subject to other economies’ macroeconomic management, thus potentially leading to stable and sustainable economic growth. However, in a world with price and wage rigidities, policy makers could also manipulate output movement (at least in the short-run), thus leading to increasing output and inflation volatility. Furthermore, monetary authorities could also abuse their autonomy to monetize fiscal debt, and therefore end up destabilizing the economy through high and volatile inflation. Exchange rate stability could bring out price stability by providing an anchor, and lower risk premium by mitigating uncertainty, thereby fostering investment and international trade. Also, at the time of an economic crisis, maintaining a pegged exchange rate could increase the credibility of policy makers and thereby contribute to stabilizing output movement (Aizenman et al. 2012). However, greater levels of exchange rate stability could also rid policy makers of a policy choice of using exchange rate as a tool to absorb external shocks.1 Hence, the rigidity caused by exchange rate stability could not only enhance output volatility, but also cause misallocation of resources and unbalanced, unsustainable growth. Financial liberalization is perhaps the most contentious and hotly debated policy among the three policy choices of the trilemma. On the one hand, more open financial markets could lead to economic growth by paving the way for more efficient resource allocation, mitigating information asymmetry, enhancing and/or supplementing domestic savings, and helping transfer of technological or managerial know-how (i.e., growth in total factor productivity). Also, economies with greater access to international capital markets should be better able to stabilize themselves through risk sharing and portfolio diversification. On the other hand, it is also true that financial liberalization has often been blamed for economic instability over the last three decades. Based on this view, financial openness could expose economies to volatile cross-border capital flows resulting in sudden stops or reversal of capital flows, thereby making economies vulnerable to boom-bust cycles (Kaminsky and Schmukler 2003). Thus, theory tells us that each one of the three trilemma policy choices can be a double-edged sword, which should explain the wide and mixed variety of empirical findings on each of the three policy choices. Furthermore, to make the matter more complicated, while there are three ways of pairing two out of the three policies (i.e., three vertices in the triangle in Fig. 1), the effect of each policy choice can differ depending on what the other policy choice it is paired with. For example, exchange rate stability can be more destabilizing when it is paired with financial openness while it can be stabilizing if paired with greater monetary autonomy. Hence, it may be worthwhile to empirically analyze the three types of policy combinations in a comprehensive and systematic manner.2 Development of Policy Combinations in the Trilemma Context Now, let us take a look at the development of trilemma policy combinations. Aizenman et al. (2013) introduced a set of metrics that measure the extent of achievement in the three policy goals. Aizenman et al.’s “trilemma indexes” measure the degree to which each of the three policy choices is implemented by economies. The indexes are updated occasionally and cover more than 180 economies for 1970 through 2020.3 The monetary independence index (MI) is based on the correlation of a country’s interest rates with the base country’s interest rate. The index for exchange rate stability (ERS) is an invert of exchange rate volatility, i.e., standard deviations of the monthly rate of depreciation, using the exchange rate between the home and base economies. The degree of financial integration is measured with the Chinn and Ito (2006, 2008) capital controls index (KAOPEN).4 Figure 2 illustrates the trajectories of the trilemma indexes for different income-country groups. For the advanced economies (AE),5 financial openness accelerated after the beginning of the 1990s while the extent of monetary independence started a declining trend. After the end of the 1990s, exchange rate stability rose significantly. All these trends seem to reflect the introduction of the euro in 1999.Fig. 2 Development of the trilemma configurations over time, (a) Advanced Economies, (b) Emerging market economies, (c) Non-Emerging Market Developing Countries Developing economies on the other hand do not present such a distinct divergence of the indexes, and their experiences differ depending on whether they are emerging or non-emerging market economies.6 For EMEs, exchange rate stability declined rapidly from the 1970s through the mid-1980s. After some retrenchment around early 1980s (in the wake of the debt crisis), financial openness started rising from 1990 onwards. For the other developing economies (non-EME LDC), exchange rate stability declined less rapidly, and financial openness trended upward more slowly. In both cases though, monetary independence remained more or less trendless. Interestingly, EMEs tend to choose a policy combination composed of intermediate levels of all three policies as the indexes suggest, which we call the “middle-ground convergence.” This pattern of results suggests that EMEs may have been trying to cling to moderate levels of both monetary independence and financial openness while maintaining higher levels of exchange rate stability. In other words, they have been leaning somewhat against the trilemma over a period that interestingly coincides with the time when some of these economies began accumulating sizable IR, potentially to buffer the trade-off arising from the trilemma. None of these observations is applicable to non-emerging developing market economies (Fig. 2). For this group of economies, exchange rate stability has been the most aggressively pursued policy throughout the period. In contrast to the experience of the EMEs, financial liberalization has not been proceeding rapidly for the non-emerging market developing economies. Comparing these indexes provides some interesting insights into how the international financial architecture has evolved over time. However, just looking at the evolution of open macro policies through the lens of the three trilemma policies may not be sufficient; it is increasingly important to shed light on the role of IR holding. Over the last two decades, while a growing number of developing countries have opted for greater flexibility in exchange rate, IR/GDP ratios increased dramatically, especially in the wake of the East Asian crises, and most evidently among EMEs. Between 1990 and 2011, global reserves increased from about USD 1 trillion to more than USD 10 trillion, and to USD 15 by 2020 (Fig. 3). Today, about three quarters of the global IR are held by developing countries, geographically concentrating in Asia (Fig. 3). The most dramatic changes occurred in China; As of 1990, China held mere 2.8% of global reserves, increasing its ratio to about 23.6% in 2020.Fig. 3 (a): IR Holding by AEs and Non-AEs (in US Billions), (b): IR Holding by Country Groups (% of World Total) Many researchers have pointed out the increasing importance of financial integration as a determinant for IR hoarding (Aizenman and Lee 2007; Cheung and Ito 2009; Delatte and Fouquau 2012; and Obstfeld et al. 2009), suggesting a link between the changing configurations of the trilemma and the level of IR. In fact, holding an adequate amount of IR may indeed allow an economy to achieve a certain target combination of the three trilemma policies. For example, a country pursuing a stable exchange rate and monetary autonomy may try to liberalize cross-border financial transactions while determined not to give up the current levels of exchange rate stability and monetary autonomy. In such a case, the monetary authorities may try to hold a sizeable amount of IR so that they can stabilize the exchange rate movement while retaining monetary autonomy. Or, an economy with open financial markets and fixed exchange rate could independently relax monetary policy, though temporarily, as long as it holds a massive amount of IR. The “diamond charts” suffice this purpose and intuitively summarize the development of trilemma policy combinations while incorporating IR holding. Figure 4 illustrates the trends for different income-based or geographical groups of countries. Each country’s configuration at a given instant is summarized by a “generalized diamond,” whose four vertices measure monetary independence, exchange rate stability, IR/GDP ratio, and financial integration. The origin has been normalized so as to represent zero monetary independence, pure float, zero international reserves, and financial autarky.Fig. 4 The trilemma and international reserves configurations over time Based on the figures, AEs and EMEs have moved towards deeper financial integration while non-emerging market developing countries have barely inched toward financial integration. While pursuing greater financial openness, AEs have lost monetary independence. EMEs, after giving up some exchange rate stability during the 1970s, have not changed their stance on the exchange rate stability at an intermediate level whereas non-emerging market developing countries seem to be remaining at, or slightly oscillating around, a relatively high level of exchange rate stability. Interestingly, EMEs stand out from other groups by achieving a relatively balanced, mid-level combination of the three macroeconomic goals along with a substantially increased amount of IR holding by the 2000s. EMEs in Latin America (LATAM) and Asia have moved somewhat toward exchange rate flexibility in the 1970s, a contrast from the group of non-EME developing countries.7 LATAM countries have rapidly increased financial openness although they retrenched financial openness in the 2010s. Asian EMEs have retained a stable level of financial openness through the sample period. One distinctive characteristic of the group of Asian EMEs is that it holds much more IR than any other group while having achieved a balanced combination of the three policy goals. Impacts of the Crises on the Four Policy Combinations – Graphical Presentation These changes in the policy configurations can be abrupt and radical, caused by major economic events such as currency crisis and changes in the international monetary system. The diamond chars of Fig. 5 illustrate the impacts of financial crises on the four policy combinations. In Fig. 5, the diamonds with the orange solid lines depict the three trilemma policy configurations and IR (as a share of GDP) shown as the ten-year averages between 1999 and 2008, i.e., post-Asian Financial Crisis (AFC) decade.8 The diamonds with the green dotted lines illustrate the four policy configurations as the ten-year averages in the pre-Asian Financial Crisis (i.e., 1987 – 1996).Fig. 5 (a): Impacts of the Asian financial crisis on the trilemma and IR configurations, (b): impacts of the global financial crisis on the trilemma and IR configurations In the aftermath of the AFC, EMEs have increased the level of financial openness significantly and these economies hold more IR compared to the pre-AFC period. In contrast, non-EME developing economies do not show much change between the pre- and the post-AFC periods. Among EMEs, Latin American economies increased financial openness considerably whereas Asian EMEs did not change the level of financial openness between the pre- and the post-crisis period. However, these economies significantly increased the level of IR holding in the post-AFC period as many studies show. They also reduced the level of monetary independence to some extent in the post-crisis period. EMEs in Eastern and Central Europe increased the levels of both financial openness and IR holding.9 Did the GFC leave any impacts on the trilemma configurations and IR holding? Figure 5 illustrates the diamond charts for the four policy variables in the decades before and after the GFC. The diamond charts for AEs, EMEs, and non-EMEs show that the trilemma configurations and IR holding have not changed in the aftermath of the GFC. However, there appear to be some geographical differences across EMEs. Asian EMEs have increased the level of monetary independence and also retained more IR whereas LATAM EMEs have reduced the level of financial openness to a small extent. Eastern and Central European EMEs have become less independent in their monetary policy making and more financially openness. Their exchange rate stability has also inched up as well. All these reflect policy changes by some Eastern and Central European EMEs to link their currencies to the euro. Analysis on the Change in the Trilemma and IR Configurations Systematic Tests on the Change in the Trilemma and IR Configurations between the Pre- and Post-crisis While the above analysis with the diamond charts helps to provide pictures on long-term changes in the configurations of the trilemma and IR policies over crises, aggregations of policy variables across the sample groups may mask the nuances of the development of policy configurations in individual economies. Given that, we test the following regression analysis:1 Δyi,k,tC=yi,k,t+5\|t+1C¯-yi,k,t-1\|t-5C¯=αkC+εi,k,tC. yi,k,t+5\|t+1C¯ represents the post-crisis 5-year average of one of the trilemma variables or IR holding (k∈K) of country i in the aftermath of either the AFC or GFC (i.e., C = AFC or GFC) whereas yi,k,t-1\|t-5C¯ represents the pre-crisis 5-year average of the variable of concern prior to the AFC or GFC.10 Hence, Δyi,k,tC refers to the change over the AFC or GFC period in one of the variables of our concern: MI, ERS, FO, and IR. The above specification is essentially the t-test on the level of variable y between the pre- and the post-crisis 5-year periods. Hence, a significantly positive α^ indicates the level of variable y is significantly higher in the 5-year period after crisis C on average for a sample group. Columns 1 through 4 of Tables 1 and 2 report the results of the estimation based on Eq. (1) for the case of the AFC and the GFC, respectively.Table 1 Changes in the four policy configurations over the Asian Financial Crisis (AFC) d_IR d_ERS d_KA d_MI d_IR d_ERS d_KA d_MI d_IR d_ERS d_KA d_MI (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Constant 0.024 0.038 0.071 -0.049 0.002 0.013 0.057 -0.067 -0.005 0.025 0.087 -0.077 (0.006)*** (0.023)* (0.020)* (0.014)* (0.020) (0.071) (0.069) (0.042) (0.021) (0.077) (0.075) (0.045)* LDC 0.030 -0.006 0.013 0.044 0.044 -0.009 -0.020 -0.024 (0.021) (0.075) (0.072) (0.044) (0.035) (0.129) (0.126) (0.075) Euro -0.034 0.371 0.036 -0.203 -0.027 0.359 0.007 -0.193 (0.029) (0.103)* (0.102) (0.061)* (0.029) (0.106)* (0.106) (0.062)* Asia 0.029 -0.047 -0.119 0.041 (0.036) (0.135) (0.132) (0.078) ECE 0.024 0.112 0.026 0.134 (0.040) (0.148) (0.145) (0.087) MENA -0.047 -0.001 -0.025 -0.028 (0.041) (0.151) (0.148) (0.092) LATAM -0.019 0.014 0.067 0.060 (0.039) (0.146) (0.143) (0.085) SSA -0.024 -0.109 0.016 0.128 (0.040) (0.147) (0.143) (0.086) N 127 135 134 124 127 135 134 124 127 135 134 124 Adj. R2 0.00 0.00 0.00 0.00 0.06 0.14 -0.01 0.18 0.15 0.18 0.01 0.24 Table 2 Changes in the four policy configurations over the Global Financial Crisis (GFC) d_IR d_ERS d_KA d_MI d_IR d_ERS d_KA d_MI d_IR d_ERS d_KA d_MI (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Constant 0.032 -0.003 -0.012 0.037 0.106 -0.070 -0.032 0.161 0.126 -0.050 -0.044 0.156 (0.010)* (0.014) (0.013) (0.018) (0.032)* (0.047) (0.044) (0.057)* (0.039)* (0.056) (0.054) (0.068) LDC -0.069 0.063 0.019 -0.125 -0.027 0.029 -0.018 -0.073 (0.033) (0.048) (0.046) (0.059) (0.059) (0.084) (0.080) (0.102) Euro -0.136 0.134 0.049 -0.183 -0.157 0.115 0.060 -0.178 (0.040)* (0.058) (0.055) (0.070) (0.047)* (0.067)* (0.064) (0.081)** Asia -0.068 -0.068 0.042 0.018 (0.068) (0.097) (0.093) (0.118) ECE -0.087 0.061 0.086 -0.104 (0.073) (0.105) (0.099) (0.127) MENA -0.014 -0.009 0.035 0.051 (0.076) (0.109) (0.104) (0.133) LATAM -0.063 0.036 0.002 -0.054 (0.073) (0.105) (0.100) (0.127) SSA -0.063 0.021 0.076 -0.095 (0.074) (0.105) (0.100) (0.128) N 132 133 136 126 130 133 136 126 130 133 136 126 Adj. R2 0.00 0.00 0.00 0.00 0.07 0.03 -0.01 0.04 0.07 0.05 -0.00 0.07 The dependent variable is the change in the -year average of a trilemma configuration of concern between the pre- and the post-GFC. “LDC” refers to “less developed economies.” “ECE,” “MENA,” “LATAM,” and “SSA” refer to “Eastern & Central Europe,” “Middle East and North Africa,” “Latin America,” and Sub-Saharan Africa,” respectively. * p < 0.1; p < 0.05; * p < 0.01 In the case of the AFC, the sample countries on average increased the extent of ERS and FO, whereas they reduced the extent of MI. The average amount of IR also went up for these countries as well. In contrast, the GFC does not involve any significant changes in the ERS or FO policies, but the sample economies increased the levels of MI and IR in the post-GFC period. Whether the AFC or the GFC, in its aftermath, economies of our concern increased the holding of IR, which is consistent with the argument that countries hold IR for the sake of self-insurance (Aizenman and Lee 2007). In the previous section, we have also seen that the combinations of the three trilemma policies and IR holding differ across income and regional groups of economies. We have seen that the trilemma configurations between AEs and developing economies differ. Although we are interested in the change in the trilemma and IR configurations, we still examine the differences among AEs, euro member countries, and non-AEs by including the dummies for non-AEs (LDC) and the euro member countries (EURO). In columns 5 through 8 of Table 1, we see that compared to the pre-AFC period, the euro member countries pursued greater ERS but gave up MI, which reflects the efforts made for the inauguration of euro that almost coincides with the AFC period. Compared to non-euro AEs, LDC’s average increase in the amount of IR holding is positive in the post-AFC period, but it is not statistically significant. To examine if there is any heterogeneity on the four policy variables based on the regions, we include regional dummies in the estimation, namely, Asia, Eastern and Central Europe (ECE), Middle East and North Africa (MENA), Latin America, and Sub-Saharan Africa (SSA).11 However, we do not observe any region-specific characteristics in the average changes in the four policy variables. What about the case of the GFC (Table 2)? On average, the sample economies increased the volume of IR holding and the level of MI in the aftermath of the GFC (columns 1 and 4). Developing economies in general increased the amount of IR by 3.7% (= 0.106 – 0.069) and the level of MI by 3.6% in the post-crisis period when we do not control for regional heterogeneity. As was the case with the AFC, we do not find any region-specific behavior in the policy configurations, and when we include the regional dummies, we lose statistical significance for the non-AE dummy. In-depth Analysis of the Change in the Trilemma and IR Configurations – SURE Approach The previous analysis reported in Tables 1 and 2 is essentially comparison of the averages of policy variables for different country groups. As can be seen in the low adjusted R2, there can be missing variables that may affect the change in the trilemma and IR configurations. We now identify econometrically the determinants of the changes in the three trilemma policy combination and IR holding. Instead of assuming that the trilemma-related policy combinations can be only attributed to geographical characteristics or income levels, we examine whether and how economic and institutional factors affect the open macro policy configurations. For the estimation, we can simply use the following estimation model for policy variable k:2 Δyi,k,tC=αkC+Xi.k.tC∗Bk+εi,k,tC. Here, we assume that a policymaker determines the combinations of the three trilemma policies and IR holding jointly, which we believe is a reasonable assumption. Aizenman et al. (2013) and Ito and Kawai (2014) empirically show that the three policy variables based on the trilemma: MI, ERS, and FO are linearly related. Furthermore, as Aizenman (2017) and Aizenman et al. (2020) argue, we may now live in a world of “quadrilemma,” where financial stability has been added to the trilemma’s original policy goals. The above arguments lead us to incorporate additional two considerations to Eq. (2). First, if we assume Eq. (2) as a set of four equations and that IR, MI, ERS, and FO are jointly determined, the error terms: εi,IR,tC,εi,MI,tCεi,ERS,tC,andεi,FO,tC will be correlated. Second, the extent to which a policy variable (k) changes its value over a crisis episode can be affected by changes in the other policy variables. That is especially the case for the three trilemma policy variables because they are linearly related. Hence, in the estimation model for the change in policy variable k over crisis C, the changes in the other three variables should also be included in the estimation.12 Therefore, the set of estimation equations will be:3 Δyi,IR,tC=αIRC+γIR,MICΔyi,MI,tC+γIR,ERSCΔyi,ERS,tC+γIR,FOCΔyi,FO,tC+XCi,tBIRC+εi,IR,tCΔyi,MI,tC=αMIC+γMI,IRCΔyi,IR,tC+γMI,ERSCΔyi,ERS,tC+γMI,FOCΔyi,FO,tC+XCi,tBMIC+εi,MI,tCΔyi,ERS,tC=αERSC+γERS,IRCΔyi,IR,tC+γERS,MICΔyi,MI,tC+γERS,FOCΔyi,FO,tC+XCi,tBERSC+εi,ERS,tCΔyi,FO,tC=αFOC+γFO,IRCΔyi,IR,tC+γFO,MICΔyi,MI,tC+γFO,ERSCΔyi,ERS,tC+XCi,tBFOC+εi,FO,tC To account for the joint determination of the policy variables and the correlated error terms across the four equations, we apply the seemingly unrelated regression (SUR) estimation method to a cross-sectional data for each of AFC and GFC. X is a vector of the common explanatory variables; B is a vector of corresponding coefficients; and covεj,εk≠0forjork={IR,ES,FO,MI}. The theoretical rationale for this estimation is that the exogenous variables in Xi,t jointly determine the change in the combinations of the four policy choices. The vector of explanatory variables Xi,t includes the following variables: terms of trade (TOT) shocks; relative per-capita income (per capita GDP in PPP as a percentage of the US level); the growth rate of GDP during the crisis of concern; the dummies for the existence of the sovereign wealth fund (SWF), IMF stabilization programs, and swap agreements. We suppose that TOT shocks would capture the extent of external shocks to which the sample countries are exposed prior to and during the crisis of concern. We measure the shocks using the standard deviations of the growth rate of TOT over five years including the crisis years. Because the level of exposure could also be affected by the level of openness of the country, we include the product between the five-year standard deviations of the growth rate of TOT and the level of trade openness (i.e., (EX + IM)/GDP) as of the crisis years. We include relative income in the estimation (as of one year prior to the crisis period) because we have seen that countries may behave differently depending on their income levels. The growth rate of real GDP is measured for the crisis years, so that it supposed to examine whether the “depth” of the crisis impacts the changes in the policy combinations. We include a dummy for swap agreements, that takes the value of one if a country has a bilateral currency swap agreement with a major central bank of either the Federal Reserve Board, the European Central Bank, and the Bank of Japan, or the People’s Bank of China (regardless of the currency of the agreement) (Aizenman et al. 2015). We stipulate that a swap agreement can relax liquidity constraint and ensure accessibility to a hard currency when there is liquidity shortage. The access to hard currencies is especially helpful when the global economic conditions are still fragile in the immediate aftermath of a financial crisis. Hence, a swap agreement provision may allow countries to hold less of IR than they would otherwise. Removing or alleviating liquidity shortage may make it easier for policymakers to implement financial liberalization, which suggests the existence of a swap agreement may lead to an increase in the extent of financial openness. The greater reliance on sovereign wealth funds (SWF) as a means to manage the public sector’s saving is another example of a possible supplement to IR hoarding. The impetus of instituting an SWF has been based on the recognition that the primary mandate of the central bank is to conduct monetary policy and ensure financial stability, not managing IR. Hence, the opportunity cost of reserves in practice may be of limited relevance for the central bank’s operations. Therefore, once the level of IR (as a share of GDP) reaches a level high enough to cover self-insurance needs, countries, usually those with high saving rates, may opt to manage their public saving in their own SWFs. That can especially be true for commodity-rich countries. Unlike the central bank authorities, the mandate of SWFs is to secure stable income for future generations; therefore, an SWF generally has a higher risk tolerance than the central bank and aims for higher-than-expected income and longer-term investments. Given these considerations, the presence of SWFs may lower IR/GDP for a given savings rate. We include a dummy variable for the existence of SWFs in the estimation.13 IMF’s stabilization programs may affect the change in the IR holding. Funds available through stabilization programs may provide an additional funding source for a crisis afflicted economy. Or, an expectation that IMF’s stabilization programs would be available to mitigate liquidity shortage may make policymakers less incentivized to hold IR. Furthermore, the IMF may require a potential fund recipient country to implement financial opening as one of the conditionalities. Also, the IMF may encourage a potential recipient to adopt flexible exchange rate. We assign a value of one for the dummy for a country under an IMF stabilization program during the crisis period.14 We continue to include the regional dummies. In addition, because the euro member countries have had a unique history of the trilemma configuration, we control for the euro membership with a dummy. Table 3 presents the results from the SUR estimations on the determinants of the changes in the four policy variables over the AFC.Table 3 SUR estimations on the determinants of the changes in the open macro policy variables over the AFC dIR dERS dFO dMI (1) (2) (3) (4) Change in ERS over AFC 0.087 0.194 -0.119 (0.025)*** (0.105)* (0.065)* Change in FO over AFC -0.094 0.213 -0.394 (0.027)*** (0.115)* (0.060)*** Change in MI over AFC -0.068 -0.340 -1.023 (0.044) (0.185)* (0.156)* Change in IR over AFC 1.565 -1.526 -0.427 (0.457)* (0.436)* (0.278) EURO -0.068 0.344 -0.309 -0.160 (0.025)* (0.101)* (0.101)* (0.061)* Asia 0.077 -0.217 -0.055 -0.032 (0.025)* (0.112)* (0.107) (0.067) Eastern & Central Europe 0.098 -0.248 0.232 0.096 (0.035)* (0.151) (0.145) (0.090) MENA 0.126 -0.304 0.293 0.083 (0.030)* (0.134) (0.128) (0.080) Latin America 0.075 -0.159 0.316 0.177 (0.027)* (0.117) (0.110)* (0.067)* Sub-Saharan Africa 0.079 -0.195 0.305 0.203 (0.029)* (0.126) (0.120) (0.072)* TOT shocks x trade openness 0.293 -0.932 0.615 0.324 (0.139) (0.595) (0.573) (0.354) Relative income 0.104 -0.352 0.220 0.094 (0.034)* (0.145) (0.141) (0.087) Real GDP growth during crisis -0.549 2.084 -2.632 -0.937 (0.226) (0.956) (0.899)* (0.571) IMF 0.013 -0.057 -0.005 -0.042 (0.014) (0.059) (0.056) (0.034) SWF 0.010 0.074 0.059 0.016 (0.020) (0.085) (0.081) (0.050) Constant -0.063 0.164 -0.058 -0.073 (0.031)** (0.135) (0.129) (0.080) N 80 Let us first focus on the three trilemma variables.15 If a country increases the extent of ERS in the aftermath of AFC, it would tend to lower MI (i.e., corr(dERS, dMI) < 0). Conversely, a greater pursuit of exchange rate flexibility would yield greater MI. A country that opens up its financial markets more in the post-AFC period would lose its MI for a given change in ERS (i.e., corr (dFO, dMI) < 0). For a given change in MI, the changes in FO and ERS are in a positive relationship (i.e., corr (dFO, dERS) > 0). The combination of the three correlations as we found is consistent with the theoretical premise that the weighted average of the three variables is a constant (Mundell 1963).16 The movement of IR holding is also related to the trilemma configurations. An increase in the extent of ERS and that in IR are positively correlated (columns 1 and 2) while changes in FO are negatively correlated with IR. These findings suggest that if a central bank with an undervalued currency pursued greater ERS, it would hold more IR through active foreign exchange interventions (i.e., by buying hard currency and selling off its domestic currency. Or, a country that aborted high level of ERS would lose its reserves due to speculative attacks. A country with more open financial markets may lose its IR holding ceteris paribus, suggesting that increasing the level of financial openness might lead to a leak of IR holding. A country that experienced greater TOT shocks during and prior to the AFC or negative GDP growth during the crisis tends to hold more IR in the post-AFC period, providing evidence for self-insurance motives of IR holding. A country with negative per capita growth also tends to pursue greater exchange rate flexibility apparently with the hope of retaining greater MI. We do not find any significant impacts of IMF stabilization programs or the possession of SWF. Appendix Table 8 in Appendix 2 reports the SUR estimation results for the subsamples of AEs and non-AE countries. Overall, we can see that the results in Table 3 are consistent with the results of the non-AE subsample than with those of the AE subsample. For the subsample of non-AE countries, the estimate on the TOT shocks is now significantly negative for the ERS estimation, indicating that when it is exposed to TOT shocks, an economy of concern would respond by pursuing greater exchange rate flexibility so that exchange rate movements could bugger the shocks. If a developing country experiences negative growth during the AFC, in its aftermath, it would tend to pursue less ERS and greater FO and MI while holding more IR. Higher MI may be for the country of concern to retain more control over monetary policy while greater financial openness for the country to benefit more from international risk sharing. To insure itself against the risk from potentially great financial instability, the country would accumulate more IR. Table 4 presents the SUR estimation results for the case of the GFC.Table 4 SUR estimations on the determinants of the changes in the open macro policy variables over the GFC dIR dERS dFO dMI (1) (2) (3) (4) Change in ERS over GFC -0.122 -0.033 -0.368 (0.071)* (0.100) (0.106)* Change in FO over GFC 0.022 -0.029 -0.278 (0.068) (0.089) (0.102)* Change in MI over GFC 0.275 -0.274 -0.231 (0.058)* (0.079)* (0.085)*** Change in IR over GFC -0.212 0.043 0.640 (0.123)* (0.132) (0.136)*** EURO -0.079 0.021 0.036 -0.042 (0.047)* (0.063) (0.067) (0.073) Asia -0.047 -0.048 0.001 -0.022 (0.052) (0.068) (0.072) (0.079) Eastern & Central Europe -0.059 0.075 0.056 -0.008 (0.051) (0.067) (0.071) (0.078) MENA -0.032 -0.031 -0.043 -0.017 (0.060) (0.079) (0.084) (0.092) Latin America -0.041 0.064 -0.073 -0.035 (0.055) (0.073) (0.077) (0.084) Sub-Saharan Africa -0.053 0.035 0.019 -0.022 (0.060) (0.080) (0.084) (0.092) TOT shocks x trade openness 0.262 0.449 -0.113 -0.057 (0.184) (0.240)* (0.257) (0.282) Relative income 0.043 0.017 0.055 0.000 (0.040) (0.053) (0.056) (0.062) Real GDP growth during crisis -0.240 0.164 0.468 0.548 (0.320) (0.422) (0.444) (0.487) IMF 0.041 -0.108 0.020 -0.093 (0.027) (0.035)* (0.038) (0.041) SWF -0.040 -0.039 0.055 0.075 (0.029) (0.038) (0.041) (0.044)* Swap -0.016 0.026 -0.045 0.010 (0.033) (0.043) (0.046) (0.050) Constant 0.039 -0.012 -0.017 0.053 (0.056) (0.074) (0.078) (0.085) N 112 Overall, the results of the SUR estimation do not appear robust, indicating that countries did not respond to the GFC by altering the mix of open macro policies. While the estimate of the TOT shock is now significantly positive, its impact on IR is no longer significant. The mixture of the four policy variables is not affected by economic growth. A country under the IMF’s stabilization program would decrease the extent of ERS and MI. We see the same correlation patterns of corr(dERS, dMI) < 0 and corr(dFO, dMI) < 0, but we do not see statistically significant correlation between the post-crisis change in ERS and that in FO (corr(dERS, dFO) = 0) unlike in the case of the AFC. However, a change in ERS and that in IR are negatively correlated in the post-GFC period, a contrast to the case of the post-AFC period. In the aftermath of the GFC, those economies that pursued more exchange rate flexibility tended to hold more IR, possibly because those economies wanted to buffer themselves by ensuring more access to hard currency. Furthermore, a rise in the level of MI, possibly by reducing the level of ERS, would lead to a rise in IR holding. In general, the trilemma and IR arrangements did not change much in the post-GFC period. In the post-AFC period, many developing countries responded to the crisis by altering the trilemma and IR configuration. Their response was heterogenous. In the case of GFC, the impact was greater AEs, where the epicenter was, but it was rather weaker and homogenous in the developing economies. We do not observe much significant alteration of the policy arrangements in the aftermath of the GFC. In sum, the AFC was more impactful than the GFC in terms of how countries changed their policy configuration in the aftermath of the crisis. We conducted the OLS estimations that correspond to the SUR estimations reported in Tables 3 and 4. The results are reported in Online Appendix. Overall, the results from the OLS estimations are less robust than those from the SUR estimations. However, the signs of the estimates are mostly consistent with those in the SUR estimations. Considering that the SUR estimation improves the level of efficiency, the results are not unexpected. Further Analysis of the Heterogeneity in Crisis Response Countries with certain economic characteristics may respond to crises differently than countries without them. Here, we investigate whether and how commodity exporters, manufacturing exporters, and large capital borrowers behave differently in terms of how they respond to the AFC and the GFC. Let us first compare commodity exporters with non-commodity exporters. At the top panel of Table 5, we divide the full sample into the subsamples of commodity exporters and non-commodity exporters and report the results from the estimation for the change in policy configurations over the post-AFC period.17 The bottom panel reports the comparative results from the estimations for policy changes in the post-GFC period.18Table 5 SUR estimation with disaggregated samples, commodity exporters vs. non-commodity exporters Over the AFC Commodity Exporters Non-Commodity Exporters dIR dERS dFO dMI dIR dERS dFO dMI (1) (2) (3) (4) (5) (6) (7) (8) Change in ERS over AFC 0.117 0.460 0.025 -0.106 -0.032 -0.292 (0.025)* (0.119)* (0.091) (0.036)* (0.149) (0.088)* Change in FO over AFC -0.182 0.693 -0.471 -0.048 -0.034 -0.391 (0.026)* (0.179)* (0.093)* (0.038) (0.157) (0.084)* Change in MI over AFC -0.116 0.078 -0.961 -0.185 -0.780 -0.991 (0.048) (0.280) (0.190)* (0.058)* (0.235)* (0.214)* Change in IR over AFC 3.746 -3.887 -1.213 -1.812 -0.780 -1.183 (0.799)* (0.547)* (0.501) (0.609)* (0.616) (0.369)* TOT shocks x trade openness 0.955 -3.712 4.134 1.654 0.007 0.103 -0.812 0.008 (0.182)* (1.220)* (0.923)* (0.712) (0.165) (0.681) (0.641) (0.412) Relative income 0.071 -0.556 0.251 0.005 0.124 0.052 0.354 0.255 (0.066) (0.367) (0.306) (0.215) (0.036)* (0.164) (0.157) (0.096)* Real GDP growth during crisis -0.831 4.603 -4.171 -1.637 -0.364 -0.683 -0.243 -0.337 (0.315)* (1.753)* (1.408)* (1.039) (0.255) (1.079) (1.053) (0.661) IMF 0.028 -0.111 0.091 -0.005 0.003 -0.117 0.139 0.024 (0.017)* (0.100) (0.081) (0.057) (0.021) (0.083) (0.081)* (0.052) SWF -0.097 0.343 -0.425 -0.182 0.076 0.300 0.146 0.157 (0.029)*** (0.181)* (0.141)*** (0.103)* (0.023)* (0.097)* (0.100) (0.062) N 38 42 Over the GFC Commodity Exporters Non-Commodity Exporters dIR dERS dFO dMI dIR dERS dFO dMI (1) (2) (3) (4) (5) (6) (7) (8) Change in ERS over GFC -0.222 -0.523 0.388 0.044 0.467 -0.158 (0.104) (0.170)* (0.177) (0.107) (0.098)* (0.146) Change in FO over GFC 0.082 -0.299 0.041 -0.360 0.680 -0.230 (0.080) (0.097)* (0.136) (0.123)* (0.143)* (0.174) Change in MI over GFC 0.294 0.217 0.040 0.276 -0.126 -0.127 (0.074)* (0.099) (0.133) (0.091)* (0.117) (0.096) Change in IR over GFC -0.359 0.232 0.846 0.067 -0.375 0.523 (0.167) (0.224) (0.213)* (0.163) (0.128)* (0.173)* TOT shocks x trade openness 0.599 0.416 0.046 -0.377 -0.135 0.660 -0.568 -0.449 (0.223)* (0.299) (0.399) (0.402) (0.332) (0.401)* (0.335)* (0.452) Relative income -0.064 -0.052 0.082 0.068 0.269 0.055 0.101 -0.122 (0.044) (0.056) (0.073) (0.075) (0.080)*** (0.106) (0.088) (0.119) Real GDP growth during crisis -0.614 0.096 0.551 0.659 1.451 -0.170 0.962 -1.033 (0.352)* (0.457) (0.602) (0.607) (0.563)** (0.719) (0.582)* (0.796) IMF 0.027 -0.084 -0.065 -0.010 0.094 -0.191 0.153 -0.025 (0.032) (0.039) (0.054) (0.055) (0.046) (0.053)* (0.045)* (0.065) SWF -0.048 0.016 0.080 0.055 -0.028 -0.125 0.089 0.055 (0.038) (0.048) (0.063) (0.064) (0.048) (0.057)** (0.048)* (0.065) Swap -0.004 -0.082 -0.174 0.090 0.007 -0.033 0.033 -0.020 (0.047) (0.059) (0.075) (0.078) (0.048) (0.059) (0.049) (0.066) N 55 57 In the post-AFC case, the statistical significance of the estimates from the estimation of commodity exporters and their signs are more consistent with those in Table 3 and more robust than the results from the estimation for the non-commodity exporters. For commodity exporters, the estimates of TOT shocks are now significant for all four policy variables. Commodity exporters exposed to TOT shocks during the AFC tend to increase the holding of IR in the post-crisis period so as to insure themselves against the TOT shocks. These exporters tend to increase the levels of FO and MI while pursuing lower levels of ERS, all of which is consistent with a mix of more flexible exchange rate policy and greater monetary autonomy. Clearly, this policy mix is intended to stabilize the crisis conditions. Commodity exporters are also sensitive to the output growth during the crisis than non-commodity exporters. If commodity exporters experienced negative growth during the AFC, in its aftermath, they tend to hold more IR and pursue more exchange rate flexibility and more FO. Commodity exporters with SWFs tend to reduce foreign exchange reserves and the extent of FO whereas they tend to increase ERS in the post-AFC period. They also tend to hold more IR in the post-crisis period when they are under IMF’s stabilization programs. In the case of the post-GFC period (the bottom of Table 5), the trilemma and IR arrangements are not so responsive to economic and structural variables. While TOT shocks and real GDP growth during the crisis continue to be significantly positive and negative factors for IR holding, respectively, as was the case with the post-AFC period, being a recipient country of IMF’s stabilization programs leads a country of concern to reduce the extent of ERS. The impact of IMF’s stabilization programs on IR holding, ERS, and FO is statistically significant and greater in magnitude for non-commodity exporters, possibly indicating that commodity exporters have easier assess to hard currency liquidity and therefore that the impact of stabilization programs is weaker for commodity exporters. In general, commodity exporters policy mitigates the volatility of the real exchange rate by increasing IR and/or through SWF in good times. In bad times, it would also provide the treasury with more resources, as has been the policy of Norway, Chile, and the like. Manufacturing exporters may have some common ground in responding to the crisis by changing their open macro policy arrangements. We regard those countries whose share of manufacturing exporters in total exports is greater than 45% as manufacturing exporters and divide the full sample into the subsamples of manufacturing exporters and non-manufacturing exporters. Table 6 reports the results from the SUR estimations for the cases of the AFC and the GFC. Table 6 SUR estimation with disaggregated samples, manufacturing exporters vs. non-manufacturing exporters Over the AFC Manufacturing Exporters Non-Manufacturing Exporters dIR dERS dFO dMI dIR dERS dFO dMI (1) (2) (3) (4) (5) (6) (7) (8) Change in ERS over AFC -0.188 0.365 -0.467 0.134 0.417 -0.002 (0.032)* (0.181) (0.071)* (0.030)* (0.124)* (0.085) Change in FO over AFC 0.013 0.272 0.038 -0.151 0.558 -0.429 (0.034) (0.135) (0.079) (0.034)* (0.166)* (0.083)* Change in MI over AFC -0.359 -1.440 0.157 -0.062 -0.005 -1.030 (0.052)* (0.220)* (0.328) (0.059) (0.274) (0.199)* Change in IR over AFC -3.209 0.306 -1.989 2.866 -2.409 -0.412 (0.546)* (0.774) (0.289)* (0.637)* (0.549)* (0.393) TOT shocks x trade openness -0.431 -0.228 -2.873 -0.620 0.336 -1.398 1.234 0.631 (0.539) (2.257) (2.582) (1.285) (0.149) (0.693) (0.610) (0.389) Relative income 0.065 0.053 0.311 0.115 0.031 -0.257 -0.004 -0.043 (0.031) (0.135) (0.154) (0.076) (0.063) (0.289) (0.251) (0.163) Real GDP growth during crisis -0.565 -2.200 1.987 -1.323 -0.779 3.471 -3.705 -1.181 (0.269)** (1.138)* (1.308) (0.633) (0.305) (1.403) (1.157)* (0.801) IMF -0.013 -0.137 0.270 -0.047 -0.005 0.042 -0.116 -0.087 (0.024) (0.096) (0.105)*** (0.055) (0.018) (0.083) (0.070)* (0.045)* SWF 0.157 0.636 -0.107 0.359 -0.018 -0.000 -0.041 -0.058 (0.028)* (0.115)* (0.164) (0.068)* (0.027) (0.128) (0.110) (0.070) N 38 42 Over the GFC Manufacturing Exporters Non-Manufacturing Exporters dIR dERS dFO dMI dIR dERS dFO dMI (1) (2) (3) (4) (5) (6) (7) (8) Change in ERS over GFC 0.057 0.276 -0.599 -0.053 -0.457 0.477 (0.091) (0.106)* (0.105)* (0.114) (0.169)* (0.198)** Change in FO over GFC -0.190 0.365 -0.121 0.288 -0.289 -0.079 (0.104)* (0.141)* (0.139) (0.085)* (0.107)* (0.162) Change in MI over GFC 0.323 -0.645 -0.099 0.304 0.224 -0.059 (0.089)* (0.113)* (0.113) (0.071)* (0.093)** (0.120) Change in IR over GFC 0.111 -0.278 0.583 -0.079 0.682 0.972 (0.177) (0.152)* (0.160)* (0.171) (0.202)* (0.228)* TOT shocks x trade openness -1.691 1.229 0.161 2.050 0.560 0.402 -0.172 -0.871 (0.544)* (0.800) (0.703) (0.762)* (0.192)* (0.247) (0.315) (0.356) Relative income 0.210 0.039 0.064 -0.077 -0.093 -0.078 0.114 0.144 (0.071)* (0.105) (0.091) (0.101) (0.049)* (0.060) (0.075) (0.088) Real GDP growth during crisis 0.842 -1.231 1.304 -0.965 -0.409 0.899 0.589 0.211 (0.484)* (0.675)* (0.574)** (0.650) (0.405) (0.482)* (0.628) (0.729) IMF 0.060 -0.131 0.061 -0.087 0.046 -0.032 -0.021 -0.066 (0.044) (0.060) (0.053) (0.059) (0.032) (0.040) (0.050) (0.058) SWF 0.008 -0.130 0.097 -0.057 -0.062 -0.016 0.067 0.131 (0.041) (0.056) (0.049)** (0.055) (0.039) (0.048) (0.060) (0.068)* Swap -0.030 -0.004 -0.021 -0.003 0.006 -0.001 -0.067 0.030 (0.039) (0.055) (0.048) (0.053) (0.050) (0.061) (0.076) (0.089) N 61 51 We can see that manufacturing exporters are not responsive to TOT shocks in the post-AFC period unlike commodity exporters. Non-manufacturing exporters are more responsive to TOT shocks. While real GDP growth during the crisis is negatively correlated with IR holding for both commodity and manufacturing exporters, the correlation between real GDP growth and ERS is negative for manufacturing exporters unlike commodity exporters. If a commodity exporter experienced negative GDP growth during the AFC crisis, it would pursue greater ERS, but that would involve a reduction in IR holding (through foreign exchange interventions).Table 7 SUR estimation with disaggregated samples, large capital borrowers vs. non- large capital borrowers Over the AFC Large Capital Borrowers Non-Large Capital Borrowers dIR dERS dFO dMI dIR dERS dFO dMI (1) (2) (3) (4) (5) (6) (7) (8) Change in ERS over AFC -0.028 0.440 0.017 0.072 0.040 -0.213 (0.038) (0.149)* (0.097) (0.034) (0.140) (0.100) Change in FO over AFC -0.058 0.401 -0.446 -0.121 0.054 -0.517 (0.036) (0.136)* (0.075)* (0.038)* (0.190) (0.098)* Change in MI over AFC -0.139 0.041 -1.195 -0.013 -0.534 -0.950 (0.058) (0.236) (0.201)* (0.055) (0.249) (0.180)* Change in IR over AFC -0.437 -0.996 -0.893 1.565 -1.937 -0.114 (0.602) (0.624) (0.375) (0.744) (0.602)* (0.479) TOT shocks x trade openness 0.223 -0.870 1.062 0.558 0.156 -0.785 0.290 0.165 (0.164) (0.648) (0.685) (0.418) (0.177) (0.823) (0.713) (0.524) Relative income 0.112 -0.324 0.541 0.299 0.066 -0.403 -0.049 -0.142 (0.046)** (0.186)* (0.192)* (0.117) (0.042) (0.193) (0.171) (0.127) Real GDP growth during crisis -0.171 3.505 -3.767 -1.401 -0.616 1.590 -1.575 -0.345 (0.311) (1.135)* (1.200)*** (0.770)* (0.303) (1.461) (1.252) (0.931) IMF 0.050 0.022 0.004 -0.009 -0.013 -0.128 -0.053 -0.039 (0.020) (0.085) (0.089) (0.054) (0.018) (0.082) (0.073) (0.054) SWF 0.124 0.629 -0.271 0.005 -0.002 0.077 0.034 0.005 (0.043)* (0.164)* (0.191) (0.119) (0.019) (0.088) (0.075) (0.056) N 43 37 Over the GFC Large Capital Borrowers Non-Large Capital Borrowers dIR dERS dFO dMI dIR dERS dFO dMI (1) (2) (3) (4) (5) (6) (7) (8) Change in ERS over GFC 0.096 0.181 -0.238 -0.388 -0.367 -0.264 (0.129) (0.139) (0.157) (0.069)* (0.153) (0.162) Change in FO over GFC 0.204 0.170 -0.252 -0.139 -0.259 -0.303 (0.124) (0.131) (0.152)* (0.066) (0.108) (0.135) Change in MI over GFC 0.565 -0.175 -0.197 0.074 -0.167 -0.272 (0.095)* (0.115) (0.119)* (0.062) (0.103) (0.122) Change in IR over GFC 0.106 0.239 0.844 -1.050 -0.535 0.315 (0.142) (0.146) (0.142)* (0.188)* (0.252) (0.267) TOT shocks x trade openness 0.355 0.064 -0.496 -0.289 0.046 0.555 0.200 0.628 (0.277) (0.294) (0.296)* (0.342) (0.251) (0.406) (0.493) (0.516) Relative income 0.060 0.001 0.163 -0.012 -0.002 -0.017 -0.052 -0.047 (0.086) (0.091) (0.091)* (0.106) (0.039) (0.064) (0.076) (0.081) Real GDP growth during crisis 0.737 -0.260 0.504 -1.050 -0.598 -0.470 0.558 1.183 (0.835) (0.881) (0.903) (1.016) (0.303)** (0.511) (0.607) (0.631)* IMF 0.107 -0.041 0.090 -0.162 -0.019 -0.112 -0.060 -0.049 (0.051) (0.055) (0.055) (0.061)* (0.028) (0.044)** (0.056) (0.059) SWF -0.074 -0.042 0.061 0.082 0.032 0.016 0.073 0.049 (0.040)* (0.043) (0.044) (0.049)* (0.041) (0.068) (0.081) (0.085) Swap -0.004 -0.040 -0.043 0.004 -0.027 0.002 -0.024 0.016 (0.054) (0.057) (0.059) (0.067) (0.034) (0.057) (0.068) (0.072) N 58 54 A manufacturing exporter under an IMF stabilization program tends to open its financial markets in the post-AFC period, but if it were not a manufacturing exporter though it is an IMF fund recipient, it would tend to reduce the extent of FO and MI. Previously, we found a commodity exporter that possesses SWFs tends to reduce the amount of IR holding in the post-AFC period. For a manufacturing exporter, the impact of possessing an SWF on IR holding, ERS, and MI is positive. In the case of the GFC, the estimations for both manufacturing and non-manufacturing exporters continue to be weak. The impact of IMF stabilization programs on ERS continues to be negative. A manufacturing exporter that experienced a deep decline in real GDP growth tends to hold lower levels of IR and FO, but to entail a higher level of ERS, which is opposite to the case of the AFC. Overall, because the volatility of the TOT of manufacturing exporters is much smaller compared to commodity exporters, counter-cyclical policies may be of lesser importance. Lastly, we divide the full sample into the subsamples of “large capital borrowers” and “non-large capital borrowers“ (Table 7). The former group includes the countries whose average current balances (as a share of GDP) is below -2% as the average over five years leading up to the crisis year. We consider such economies as those running current account deficit persistently, meaning they borrow constantly from international financial markets. These economies can be more vulnerable to shocks from the crisis. The subsample of “non-large capital borrowers” includes the remainder countries. In the case of the AFC, large capital borrowers which experienced negative GDP growth during the crisis tend to change their policy arrangements toward lower ERS, and higher FO and MI, though the growth rate does not affect the volume of IR holding. That may mean that while trying to benefit from more international risk diversification, large capital borrowers also try to retain monetary autonomy. Those large capital borrowers under IMF stabilization programs tend to hold more IR in the post-AFC period, which is also the case if they have SWFs. Economic and structural variables do not appear to impact the policy configurations for non-large capital borrowers. Concluding Remarks To deal with economic and financial turmoil, whether internally generated or externally imported, economic policymakers change their policy goals and configurations to stabilize economic and financial conditions or minimize vulnerability. In an open macro setting, policymakers face the constraint of choosing two out of three policy goals: monetary independence, exchange rate stability, and financial openness.19 In a financially globalized world that emerged three decades ago, in addition to the three policy goals, how to achieve financial stability has been also an important policy goal. In response, holding international reserves (IR) has become an important policy instrument as a buffer or insurance against liquidity shortages. Significant and fundamental economic events such as currency crises have often changed the policy mix. In this paper, using the trilemma index and the data on IR as a share of GDP, we find that countries’ policy mixes have been diverse and varied over time. In particular, among EMEs, we observe that the three dimensions of the trilemma configurations are converging towards a “middle ground” among emerging market economies with managed exchange rate flexibility, underpinned by sizable holdings of international reserves, and intermediate levels of monetary independence and financial integration. We are interested in whether and to what extent the most recent major financial crises (before the COVID-19 crisis in March 2020) have led to changes in policy mix in the aftermath of the crisis. We illustrate how the combination of the three trilemma policies and IR holding drastically changed before and after the AFC, but the GFC did not lead to a drastic change in the policy arrangements. In general, developing countries increased the holding of IR in the aftermath of the AFC. Those countries that were exposed to greater TOT shocks or that experienced negative economic growth during the crisis tended to hold more IR and pursue lower ERS in the post-crisis period. These findings represent the insurance motives of holding IR against future financial instability. The results from the SUR estimation of policy responses are much less robust in the case of the post-GFC period compared to the post-AFC period. That suggests that policymakers did not respond to the GFC by altering the policy mix of the open macro variables. Hence, in terms of whether a major economic event leads to a drastic change in the open macro policy arrangement, the AFC was more impactful than the GFC. We also compare how the post-crisis response differs among different types of economies. Our SUR estimation results show that commodity exporters exposed to TOT shocks during the AFC tend to increase the holding of IR in the post-crisis period. They tend to increase the levels of FO and MI while pursuing lower levels of ERS. This policy mix can be interpreted as an attempt to retain more exchange rate flexibility and greater monetary autonomy. Commodity exporters are also sensitive to the output growth during the crisis than non-commodity exporters. If they experienced negative economic growth during the AFC, they would hold more IR and pursue less ERS and more FO. In sum, the findings on the impacts of TOT shocks and economic growth indicate that commodity exporters would opt for a policy mix that would allow them to ensure more monetary autonomy. The regression analysis of manufacturing exporters does not show the same kind of results as that of commodity exporters. A manufacturing exporter that experienced a deep decline in real GDP growth tends to hold lower levels of IR and FO, but to entail a higher level of ERS, the latter of which is opposite to what we find with commodity exporters. Lastly, we focus on the behavior of “large capital borrowers.” In the case of the AFC, large capital borrowers which experienced negative GDP growth during the crisis tend to change their policy arrangements toward lower ERS, and higher FO and MI, though the growth rate does not affect the volume of IR holding. That may mean that while trying to benefit from more international risk diversification, large capital borrowers also try to retain monetary autonomy. Our regression results indicate that economic and structural variables matter more for policy reconfigurations of large capital borrowers than those of non-large capital borrowers. Electronic supplementary material Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 58 kb) Appendix 1: Country List for the Regression Analysis (120 economies) Albania Algeria Angola Argentina EME Armenia Australia AE Austria AE Azerbaijan Bahamas, The Bahrain Bangladesh Barbados Belarus Belgium AE Belize Bolivia Botswana EME Brazil EME Bulgaria EME Cameroon Canada AE Chile EME China EME Colombia EME Congo, Rep Costa Rica Croatia Cyprus Czech Republic EME Denmark AE Dominican Republic Ecuador EME Egypt, Arab Rep EME El Salvador Estonia Fiji Finland AE France AE Gabon Germany AE Ghana EME Greece AE Grenada Guatemala Haiti Honduras Hong Kong, China EME Hungary EME Iceland AE India EME Indonesia EME Ireland AE Israel EME Italy AE Jamaica EME Japan AE Jordan EME Kazakhstan Kenya EME Korea, Rep EME Kuwait Lao PDR Latvia Lebanon Lithuania EME Malaysia EME Malta AE Mauritius EME Mexico EME Moldova Mongolia Morocco EME Mozambique Namibia Netherlands AE New Zealand AE Nicaragua Nigeria EME Norway AE Oman Pakistan Panama Paraguay Peru EME Philippines EME Poland EME Portugal AE Qatar Romania Russian Federation EME Rwanda Saudi Arabia Seychelles Singapore EME Slovak Republic EME Slovenia EME South Africa EME Spain AE Sri Lanka Suriname Sweden AE Switzerland AE Tajikistan Tanzania Thailand EME Tunisia EME Turkey EME Ukraine United Kingdom AE Uruguay Venezuela, RB EME Vietnam EME Zambia AE refers to “advanced economies” whereas EME stands for “emerging market economies” Appendix 2: SUR Estimation with Disaggregated Samples Tables 8 and 9. Table 8 SUR estimation with disaggregated samples over the AFC Advanced Economies Developing Economies IR ERS FO MI IR ERS FO MI (1) (2) (3) (4) (5) (6) (7) (8) Change in ERS over AFC -0.033 0.575 -0.453 0.127 0.194 -0.079 (0.040) (0.138)* (0.108)* (0.029)* (0.133) (0.079) Change in FO over AFC -0.010 0.895 0.209 -0.079 0.185 -0.407 (0.050) (0.215)* (0.167) (0.030)* (0.127) (0.066)* Change in MI over AFC -0.188 -1.084 0.321 -0.039 -0.213 -1.150 (0.053)* (0.259)* (0.257) (0.052) (0.213) (0.186)* Change in IR over AFC -0.930 -0.174 -2.194 2.181 -1.423 -0.246 (1.104) (0.899) (0.617)* (0.501)* (0.540)* (0.330) EURO -0.049 0.103 -0.138 -0.095 (0.016)*** (0.088) (0.076)* (0.056)* Asia 0.023 0.176 -0.147 0.101 0.023 -0.103 -0.382 -0.257 (0.018) (0.096)* (0.074) (0.062) (0.027) (0.112) (0.105)* (0.062)* Eastern & Central Europe 0.003 0.013 -0.127 -0.118 (0.034) (0.142) (0.145) (0.085) MENA 0.036 -0.069 -0.069 -0.135 (0.027) (0.114) (0.116) (0.066) Latin America -0.010 0.066 -0.010 -0.030 (0.018) (0.073) (0.075) (0.045) TOT shocks -0.033 -0.561 1.856 0.497 0.354 -1.243 0.617 0.259 (0.728) (3.838) (3.047) (2.474) (0.149) (0.623) (0.653) (0.387) Relative income 0.113 0.190 -0.197 0.291 0.195 -0.691 0.364 0.100 (0.051) (0.285) (0.227) (0.179) (0.047)* (0.200)*** (0.218)* (0.130) Real GDP growth during crisis -0.399 2.102 -1.939 0.064 -0.489 1.896 -2.941 -1.213 (0.360) (1.921) (1.540) (1.260) (0.258)* (1.061)* (1.060)*** (0.646)* IMF 0.011 -0.038 -0.018 -0.044 (0.015) (0.061) (0.062) (0.037) SWF -0.020 -0.487 0.324 -0.256 -0.032 0.230 -0.000 0.019 (0.040) (0.196) (0.160) (0.127) (0.025) (0.100) (0.107) (0.064) Constant -0.080 -0.313 0.259 -0.294 0.001 0.000 0.264 0.148 (0.043)* (0.232) (0.181) (0.143) (0.026) (0.109) (0.107) (0.065) N 22 58 Table 9 SUR Estimation with Disaggregated Samples over the GFC Advanced Economies Developing Economies IR ERS FO MI IR ERS FO MI (1) (2) (3) (4) (5) (6) (7) (8) Change in ERS over GFC -0.971 0.876 0.217 -0.027 0.061 -0.304 (0.603) (0.108)* (0.942) (0.073) (0.105) (0.114)* Change in FO over GFC 0.110 0.931 -0.909 0.166 0.061 -0.220 (0.648) (0.115)* (0.959) (0.072)** (0.105) (0.116)* Change in MI over GFC 0.277 0.011 -0.042 0.203 -0.247 -0.179 (0.138) (0.047) (0.045) (0.064)* (0.093)*** (0.094)* Change in IR over GFC -0.103 0.011 0.594 -0.057 0.350 0.524 (0.064) (0.065) (0.297) (0.153) (0.151) (0.165)*** EURO -0.013 0.010 -0.011 -0.004 -0.038 0.193 -0.198 -0.314 (0.046) (0.015) (0.014) (0.068) (0.113) (0.162) (0.163) (0.178)* Asia -0.054 -0.039 0.037 0.087 0.134 -0.107 -0.302 -0.175 (0.073) (0.022)* (0.022)* (0.106) (0.149) (0.216) (0.215) (0.240) Eastern & Central Europe 0.086 0.025 -0.230 -0.156 (0.144) (0.209) (0.208) (0.232) MENA 0.151 -0.081 -0.346 -0.165 (0.152) (0.221) (0.218) (0.245) Latin America 0.130 0.023 -0.376 -0.182 (0.149) (0.216) (0.213)* (0.240) Sub-Saharan Africa 0.100 -0.008 -0.273 -0.167 (0.149) (0.217) (0.215) (0.240) TOT shocks -0.181 0.268 -0.176 0.956 0.347 0.368 -0.304 -0.134 (0.857) (0.244) (0.252) (1.255) (0.182)* (0.265) (0.267) (0.297) Relative income 0.216 -0.060 0.066 -0.224 0.012 -0.005 0.073 0.016 (0.179) (0.056) (0.057) (0.273) (0.041) (0.059) (0.058) (0.065) Real GDP growth during crisis 1.487 -0.413 0.615 -0.527 -0.278 0.115 0.431 0.641 (1.280) (0.410) (0.377) (1.930) (0.316) (0.459) (0.458) (0.505) IMF 0.234 0.608 -0.610 -0.331 0.035 -0.128 0.046 -0.094 (0.399) (0.063)* (0.033)* (0.597) (0.028) (0.038)* (0.040) (0.044) SWF -0.062 0.006 -0.018 -0.039 -0.034 -0.033 0.067 0.096 (0.051) (0.017) (0.016) (0.078) (0.032) (0.046) (0.046) (0.050)* Swap -0.005 0.022 -0.046 0.016 (0.032) (0.047) (0.047) (0.052) Constant -0.102 0.015 -0.011 0.182 -0.120 0.049 0.267 0.201 (0.143) (0.046) (0.045) (0.210) (0.148) (0.215) (0.214) (0.238) N 22 90 Acknowledgements We thank Joscha Beckmann and anonymous referees for useful comments. The financial support of faculty research funds of University of Southern California, the University of Wisconsin, Madison, and Portland State University is gratefully acknowledged. This paper was conceived when Ito was visiting the RIETI as a visiting fellow. The authors are grateful for helpful comments and suggestions by Discussion Paper seminar participants at RIETI. All remaining errors are ours. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research. 1 Prasad (2008) argues that exchange rate rigidities would prevent policy makers from implementing appropriate policies consistent with macroeconomic reality, implying that they would be prone to cause asset boom and bust by overheating the economy. 2 For a comprehensive analysis of all of the three policy aspects of the trilemma, refer to Obstfeld et al. (2005, 2009, 2010) and Shambaugh (2004). 3 The data are available at http://web.pdx.edu/~ito/trilemma_indexes.htm. The measure of financial openness (KAOPEN) is updated only to 2019. 4 More details on the construction of the indexes can be found in Aizenman et al. (2013) as well as in http://web.pdx.edu/~ito/trilemma_indexes.htm. 5 The advanced economies (AEs) refer to traditional Organization of Economic Cooperation and Development (OECD) member countries whose IMF numerical codes are below 186 plus Australia and New Zealand. 6 EMEs are those classified as either emerging or frontier in 1980–1997 by the International Financial Corporation, plus Hong Kong and Singapore. This group of economies is a subset of the group of less developed, or developing, countries (LDC). These groupings are not time variant. 7 “Emerging Asian Economies” include Brunei, Cambodia, China, Hong Kong, Indonesia, Korean Rep., Malaysia, Philippines, Singapore, Taiwan, Thailand, and Vietnam. “Emerging Latin America” includes Argentina, Brazil, Chile, Colombia, Ecuador, Jamaica, Mexico, Peru, Trinidad and Tobago, and Venezuela. 8 We assume that the years of 1997–98 are the crisis period. 9 For AEs, the diamond chart illustrates that these economies have lost monetary independence while further raising the level of financial openness. However, these developments rather reflect the efforts made by the euro member countries than showing the impacts of the AFC on AEs. 10 We regard the years 1997–1998 as crisis year (t) for AFC, and 2008–2009 for GFC. The pre-crisis period is the five-year period leading up to crisis period, i.e., 1992–1996 for the AFC and 2003–2007 for the GFC. The post-crisis period is 1999–2003 for the AFC and 2010–2014 for the GFC. Hence, neither yi,k,t+5\|t+1C¯ nor yi,k,t-1\|t-5C¯ includes the crisis periods (t) in its calculation. 11 The dummies for Western Europe and North America are not included, which means that the estimated coefficient of the constant term represents the average change in a policy variable of concern before and after a crisis among the Western European and North American countries. 12 For example, the estimation model for the change in the IR level controls for the changes in MI, ERS, and FO over a crisis of concern, and the estimation for the change in the MI level controls for IR, ERS, and FO, etc. 13 The data is extracted from Aizenman et al. (2015). 14 The data is extracted from Aizenman and Ito (2014). 15 From the way the estimation model is specified, we should think that the dependent variables and the independent variables of the trilemma and IR variables are not strictly in a causal relationship. 16 That is, if achievement in the three policy goals can be measured by some normalized indexes, the sum of the three indexes must be a constant. More specifically, if each of the indexes is assumed to range from 0 to 1, the sum of the three indexes must be 2 (Ito and Kawai 2014). 17 We regard countries whose commodity exports account for more than 40% of total exports as commodity exporters. 18 The OLS estimation results are reported in Tables 2(a) through (c) in Online Appendix. 19 Or, a country can choose a policy mix of intermediate levels of all three policy goals. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Aizenman J Ito H Living with the trilemma constraint: relative trilemma policy divergence, crises, and output losses for developing countries J Int Money Financ 2014 49 28 51 10.1016/j.jimonfin.2014.05.005 Aizenman J Lee J International reserves: precautionary versus mercantilist views, theory and evidence Open Econ Rev 2007 18 2 191 214 10.1007/s11079-007-9030-z Aizenman J Chinn MD Ito Hiro The ‘impossible trinity’ hypothesis in an era of global imbalances: measurement and testing Rev Int Econ 2013 21 3 447 458 10.1111/roie.12047 Aizenman J Cheung YW Ito H International reserves before and after the global crisis: is there no end to hoarding? J Int Money Financ 2015 52 April 2015 102 126 10.1016/j.jimonfin.2014.11.015 Aizenman J Chinn MD Ito H Financial spillovers and macroprudential policies Open Econ Rev 2020 31 529 563 10.1007/s11079-020-09580-9 Aizenman J, Chinn MD, Ito H (2012) The financial crisis, rethinking of the global financial architecture, and the trilemma. In: Morgan P, Kawai M (eds) Monetary and Currency Policy Issues for Asia: Implications of the Global Financial Crisis. Edward Elgar (February) Aizenman J (2017) International reserves, exchange rates, and monetary policy – from the trilemma to the quadrilemma. Prepared for the Oxford Research Encyclopedia of Economics and Finance Cheung YW Ito H Cross-sectional analysis on the determinants of international reserves accumulation Int Econ J 2009 23 4 447 481 10.1080/10168730903372208 Chinn MD, Ito H (2006) What matters for financial development? capital controls, institutions, and interactions. J Dev Econo 81(1):163–192 Chinn MD, Ito H (2008) A new measure of financial openness. J Comp Policy Anal 10(3):309–322 Delatte A-L Fouquau J What drove the massive hoarding of international reserves in emerging economies? A time-varying approach Rev Int Econ 2012 20 164 176 10.1111/j.1467-9396.2011.01015.x Ito H, Kawai M (2014) Determinants of the trilemma policy combination. ADBI Working Paper No. 456 (January). Tokyo: Asian Development Bank Institute. https://www.adb.org/sites/default/files/publication/156311/adbi-wp456.pdf Kaminsky G, Schmukler S (2003) Short-run pain, long-run gain: The effects of financial liberalization, No. 9787, NBER Working Papers, National Bureau of Economic Research Mundell RA Capital mobility and stabilization policy under fixed and flexible exchange rates Can J Econ Political Sci 1963 29 4 475 485 10.2307/139336 Obstfeld M Shambaugh JC Taylor AM The trilemma in history: tradeoffs among exchange rates, monetary policies, and capital mobility Rev Econ Stat 2005 87 August 423 438 10.1162/0034653054638300 Obstfeld M Shambaugh JC Taylor AM Financial stability, the trilemma, and international reserves Am Econ J Macroecon 2010 2 57 94 10.1257/mac.2.2.57 Obstfeld M, Shambaugh JC, Taylor AM (2009) Financial instability, reserves, and central bank swap lines in the panic of 2008. NBER Working Papers 14826. Cambridge, MA: National Bureau of Economic Research (March) Prasad ES (2008) Monetary policy independence, the currency regime, and the capital account in China. In: Goldstein M, Lardy NR (eds) Debating China’s Exchange Rate Policy. Peterson Institute for International Economics, Washington Shambaugh JC The effects of fixed exchange rates on monetary policy Quart J Econ 2004 119 1 301 352 10.1162/003355304772839605	0	PMC9716176	NO-CC CODE	2022-12-03 23:20:54	no	Open Econ Rev. 2022 Dec 2;:1-39	utf-8	null	null	null	oa_other
==== Front J Biol Phys J Biol Phys Journal of Biological Physics 0092-0606 1573-0689 Springer Netherlands Dordrecht 36459249 9617 10.1007/s10867-022-09617-9 Research Clinical effects of 2-DG drug restraining SARS-CoV-2 infection: A fractional order optimal control study Samui Piu 1 Mondal Jayanta 1 Ahmad Bashir 2 Chatterjee Amar Nath [email protected] 3 1 grid.449077.9 0000 0004 8497 1102 Department of Mathematics, Diamond Harbour Women’s University, Sarisha, West Bengal 743368 India 2 grid.412125.1 0000 0001 0619 1117 Nonlinear Analysis and Applied Mathematics (NAAM)-Research Group, Department of Mathematics, King Abdulaziz University, P.O. Box 80203, Jeddah, 21589 Saudi Arabia 3 grid.444341.2 0000 0000 9681 1852 Department of Mathematics, K. L. S. College, Nawada, Magadh University, Bodh Gaya, Bihar 805110 India 2 12 2022 12 2022 48 4 415438 14 7 2022 28 10 2022 © The Author(s), under exclusive licence to Springer Nature B.V. 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Fractional calculus is very convenient tool in modeling of an emergent infectious disease system comprising previous disease states, memory of disease patterns, profile of genetic variation etc. Significant complex behaviors of a disease system could be calibrated in a proficient manner through fractional order derivatives making the disease system more realistic than integer order model. In this study, a fractional order differential equation model is developed in micro level to gain perceptions regarding the effects of host immunological memory in dynamics of SARS-CoV-2 infection. Additionally, the possible optimal control of the infection with the help of an antiviral drug, viz. 2-DG, has been exemplified here. The fractional order optimal control would enable to employ the proper administration of the drug minimizing its systematic cost which will assist the health policy makers in generating better therapeutic measures against SARS-CoV-2 infection. Numerical simulations have advantages to visualize the dynamical effects of the immunological memory and optimal control inputs in the epidemic system. Keywords Caputo fractional derivative FOCP 2-DG drug SARS-CoV-2 Host immune response Immunological memory http://dx.doi.org/10.13039/100020026 Department of Higher Education, Government of West Bengal 52-Edn(B)/5B-15/2017 52-Edn(B)/5B-15/2017 Samui Piu Mondal Jayanta issue-copyright-statement© Springer Nature B.V. 2022 ==== Body pmcIntroduction Infectious diseases have brought devastation to humanity since ancient times. From the last quarter of the year 2019 the world is enduring a pandemic due to SARS-CoV-2 or COVID-19 contagion, which is highly contagious than any other known infectious diseases. The first case of SARS-CoV-2 contagion was reported in Wuhan city in China. According to the Worldometer data, currently about 226 countries and territories have been in the grip of the COVID-19 pandemic situation [1]. World Health Organization (WHO) reported globally around 51.5 Crore confirmed cases including about 6.25 Million deaths due to COVID-19 infection as of May 2022 [2]. To combat the COVID-19 pandemic, vaccine is one of the most convenient equipment; 60% of total population have been administered with full vaccination as of May 2022. Severe acute respiratory syndrome coronavirus 2 or SARS-CoV-2, a progeny of the family Coronaviridae, contains a single-stranded positive sensed RNA genome inducing the outbreak of COVID-19 infection. This infection is also capable to affect other species like dromedary camels, civet cats, and bats. The vertical transmission of SARS-CoV-2 contagion happens through particles released during coughing and sneezing in the form of large respiratory droplets as well as smaller aerosols. The spreading of COVID-19 is also responsible for secondary infections in the crowded environment and poorly ventilated indoor environment through close contact. The SARS-CoV-2 virus substantially targets the lower part of human respiratory tract causing flu-like illness with symptoms such as cough, fever, fatigue, headache, difficulties in breathing, loss of smell and taste, and diarrhea. SARS-CoV-2 utilizes the membrane protein, angiotensin-converting enzyme 2 (ACE2) receptor to transform the host epithelial cells into more vulnerable ones for its unhindered entry [3–5]. A few of the epithelial cells like myocardial epithelial cells, kidney tubular epithelial cells, and gastrointestinal epithelial cells obstruct the expression of ACE2 to enter in host. ACE2 is enunciated in myocardial cells, proximal tubule cells of the kidney, and bladder urothelial cells, and abundantly in enterocytes of the small intestine (particularly in the ileum). Type II alveolar epithelial cells of lungs contain the better copious expression of ACE2 and thus are taken into consideration as the major target cells of COVID-19 infection [6, 7]. To discern effective strategies in controlling the COVID-19 transmission, it is necessary to understand the intermediate functional relationships among the SARS-CoV-2 strain, host epithelial cells and host immune response. During the activation and differentiation of T cells, the adaptive immunity is set off in host coupled with innate immunity. After the successful entry of the SARS-CoV-2 in host epithelial cells, secretion of different cytokines particularly INF-γ, IL-6 and IL-10 takes place stimulating the host immune response [8]. Basically CD4+T cells and CD8+T cells impose uttermost influence to confront the SARS-CoV-2 by engendering the virus-specific antibodies along with the activation of B cells (T dependent) and neutralizing the host epithelial cells from spreading infection. The SARS-CoV-2 Interagency Group (SIG) established by the U.S. Government categorized a new variant of SARS-CoV-2 constructed through mutation, namely Omicron, as a Variant of Concern (VOC) on November 30, 2021. Researches are in progress to detect drug or other approaches applicable to COVID-19 patients with the aim to mitigate the pandemic. A short while ago the Defence Research & Development Organisation (DRDO) under the Government of India provided the license to the Drug firm, Granules India for manufacturing and marketing the drug, 2-Deoxy-D-Glucose (2-DG) in treatment of COVID-19 infection. The clinical trial of this drug exhibited the facts that the molecule assisted the COVID-19 patients to recover quickly and the drug proved to be effective in the reduction to the urgency of supplementary oxygen. Additionally, 2-Deoxy-D-Glucose (2-DG) successfully prevented the growth of SARS-CoV-2 through the termination of viral synthesis and energy production. Mathematical modeling of the underlying mechanism of the reciprocity between the SARS-CoV-2 and within-host immune system during COVID-19 infection is found to be quite beneficial. A handful of mathematical studies have been accomplished highlighting the dynamical aspects of the transmission of COVID-19 infection [9–16] at the population level. However in developing booster dose or in enhancing the effectiveness of currently available vaccines, proper insight regarding the intrahost viral dynamics and host immune response hindering the contamination of the COVID-19 infection is indispensable; although studies investigating these cellular facts have not been conducted on a large scale yet. Hernandez et al. [17] established a mathematical model examining the cellular kinetics along with T cell responses against the replication of SARS-CoV-2 in COVID-19 infection. Wang et al. [18] explored pathogenic characteristics, anti-inflammatory treatment strategies or combined antiviral drugs in mitigating SARS-CoV-2 infection. Different consequences of both humoral and adaptive immune responses in COVID-19 and possible eradication strategies have been analyzed in [19]. Paul et al. [20] proposed a four-dimensional SEIR model to verify the dynamics of COVID-19 infection in India and Brazil. In Chatterjee et al. [21, 22] conducted studies focusing the lytic and non-lytic role of immune response mutation of SARS-CoV-2 virions to control COVID-19 infection. Mondal et al. [4] constructed a mathematical model highlighting the dynamical behaviors of SARS-CoV-2 virions during COVID-19 infection and essential effects of host immune response to inhibit complicated epidemic states like backward bifurcation and reinfection. The effect of antiviral drug in controlling the COVID-19 and a variable ordered fractional network in host have been studied in [23]. In view of the interdependencies among the immune response of human host and viral kinetics of SARS-CoV-2 accompanying the influences of 2-DG to optimally control the SARS-CoV-2 infection, a mathematical study is delineated here. Sections 2 and 3 contain our proposed compartmental ODE model and its corresponding fractional order model respectively. In Sect. 4, fundamental properties of the fractional order model together with investigation for the possible equilibrium points and stability of the system around them are presented. Section 5 accounts for the optimal strategies to control the SARS-CoV-2 infection and possible eradication of the pandemic in the light of optimal drug influence. In Sect. 6, numerical simulations assist to visualize the dynamics of the epidemic system due to changes in model parameters and implementation of optimal strategies. Section 7 is devoted to discussion and conclusion about the upshots from the study. Model synthesis Taking into account the interrelationships between the target cells (epithelial cells), consequences of immune response in human host and the kinesis of SARS-CoV-2 during COVID-19, we are aspired to introduce a compartmental and target cell-limited model consisting of four populations, namely, ES representing the susceptible target cells of SARS-CoV-2 infection, EI indicating the infected epithelial cells capable of virions production, A presenting the immune response of human host (combination of innate and adaptive immunity), and V standing for the SARS-CoV-2 virions. Our four-dimensional ODE model which would be beneficial to analyze the dynamical aspects of the infection is given by:1 dESdt=Π1-(1-ε1)βESVA1+kV-μ1ES,dEIdt=(1-ε1)βESVA1+kV-μ2EI,dAdt=Π2-(1-ε1)θβESVA1+kV-μ3A,dVdt=(1-ε2)pEIV-(μ4+qEIη+V)V, complemented with biologically realistic non-negative initial conditions (assuming the initial time as t0)2 ES(t0)=ES0,EI(t0)=EI0,A(t0)=A0,V(t0)=V0. Uninfected epithelial cells, that is, the target cells of SARS-CoV-2 infection are recruited in the system (1) at the constant rate Π1. Thereafter SARS-CoV-2 infects the target cells with the rate β agitating the T cells to expand as well as differentiated and the expansion of T cells would be saturated by the function (1+kV). The uninfected epithelial cells are naturally cleared at a rate μ1 owing to apoptosis. Infected cells are capable to persuade the replication of SARS-CoV-2 at the rate p and wipe out at the rate μ2 from the system due to cytolytic effects of immune response. Cytokine triggers particularly the adaptive immunity to produce virion-specific antibodies (like IgM and IgG antibodies) at the rate Π2 by neutralization. SARS-CoV-2 utilizes the JaK/STAT pathway in diminishing the production of antibodies [4]. The SARS-CoV-2 has the innate immunity suppressing attribute which is represented in the functional form as θβESVA1+kV resulting deficiency in cytokines production, where θ denotes the pathogenicity of SARS-CoV-2. At the rate μ3, the efficiency of antibody response fades away from the human host due to functional exhaustion of T cells. The destruction of the virions by host immune response is stated through functional response qEIη+V, where q stands for the immune destruction of the virions and at the rate μ4 the virions be washed off from the system. Here k and η denote saturation and half-maximal saturation constants respectively. In the system (1), we introduce 0≤ϵi,i=1,2≤1 to represent effectiveness of 2-DG (ϵ1 denotes the efficacy of the drug in blocking transmission of COVID-19 and ϵ2 stands for the efficacy of the drug in blocking production of new virions). Note that ϵ1=ϵ2=0 represents no antiviral drug effect while ϵ1=ϵ2=1 represents 100% efficacy of the drug. In our proposed study, all of the model parameters and the state variables are non-negative (their source values are listed in tabular form in Table 1).Table 1 Values and sources of the model parameters associated with the fractional order system (3) Parameters Mean value (unit) Sources Π1 5 (cellsml-1day-1) [18] ε1 0≤ε1≤1 - β 0.0001 (ml(RNAcopies)-1day-1) [17, 18] k 0.1 assumed μ1 0.2 (day-1) estimated μ2 0.189 (day-1) [18] Π2 5 (cellsml-1day-1) estimated θ 0≤θ≤1 - μ3 0.1 (day-1) estimated ε2 0≤ε2≤1 - p 0.65 (day-1) [17, 18] μ4 0.1 (day-1) [18] q 0.4 (ml(RNAcopies)-1day-1) estimated η 0.7 [17, 18] Formulation of fractional order model Intending to interpret the natural phenomena associated to nonlocality, fractional order differential equations are used as an excellent tool in epidemiology. Specifically to deal with some special epidemic behaviors like memory and hereditary properties and to obtain sufficient accurate results from the real data of a disease outbreak, fractional order model would be more adequate than its integer order counterpart. Naik et al. [24] constructed a COVID-19 model with Caputo and Atangana-Baleanu fractional derivative operators to estimate the parameter and they have carried out the qualitative analysis. With the help of fixed point theory, Chen et al. [25] verified that Caputo-Fabrizio type fractional order COVID-19 model has a unique solution. Mahata et al. [26] proposed and studied an SEIRV epidemic model of COVID-19 with optimal control in the context of the Caputo fractional derivative. Paul et al. [27] proposed an Adam-Bashforth-Moulton predictor-corrector scheme for the SIQR model. A SEIR model of COVID-19 has been studied by Paul et al. [28] with the help of the fractional order derivatives employing Caputo operator. Mahata et al. [29] considered the Caputo derivative to study the spread of COVID-19. Caputo derivative operator has been widely used in the study of the SARS-CoV-2 infection [21, 22, 26, 30, 31]. It is one of the useful derivative operators to define more effectively memory effect dynamics that exist in real-world phenomena [30]. Thus we use the well-known and reliable Caputo derivative operator in fractional calculus to our proposed model (1) and thus the system (1) is transformed into the following form:3 DtαES=Π1-(1-ε1)βESVA1+kV-μ1ES,DtαEI=(1-ε1)βESVA1+kV-μ2EI,DtαA=Π2-(1-ε1)θβESVA1+kV-μ3A,DtαV=(1-ε2)pEIV-(μ4+qEIη+V)V, supplemented with non-negative initial conditions (2). In system (3), the left-Caputo fractional derivative of order α(0<α≤1) is denoted by Dtα and it might be noticeable that for α=1, the system (3) will shrink to the system (1). Qualitative analysis of the fractional model In this section, the fundamental characteristics of the fractional order system (3), specifically the existence, uniqueness and non-negativity of the solutions of the system (3), possible equilibrium points and the stability of the system (3) around these equilibrium points would be addressed. First we express the left-Caputo differential equations system (3) of differentiation order α together with non-negative initial conditions (2) as4 Dtαϑ(t)=ϱ(t,ϑ(t)),0<α≤1,ϑ(t0)=ϑ0,t0>0, where ϱ(t,ϑ(t))=(ϱ1,ϱ2,ϱ3,ϱ4)T, ϑ(t)=(ES,EI,A,V)T∈R+4 and ϑ0=(ES0,EI0,A0,V0)T∈R+4. We consider the region R+={(ES,EI,A,V)∈R+4:max(\|ES\|,\|EI\|,\|A\|,\|V\|)≤K′}⊂Rn, n≥1 and K′ is a finite positive real number. Next we define the function ϱ(t,ϑ(t)) such that ϱ(t,ϑ(t)):[t0,∞)×R+→Rn forms a vector field where ϱ1, ϱ2, ϱ3 and ϱ4 are designated asϱ1=Π1-(1-ε1)βESVA1+kV-μ1ES,ϱ2=(1-ε1)βESVA1+kV-μ2EI,ϱ3=Π2-(1-ε1)θβESVA1+kV-μ3A,ϱ4=(1-ε2)pEIV-(μ4+qEIη+V)V. Positiveness of the solutions In this subsection, we study the criteria for positiveness of the solution trajectories of the system (3). First we state the generalized mean value theorem [32] and a corollary. Theorem 1 Let ϱ(t),Daαϱ(t)∈C[a,b], for 0<α≤1. Then [ϱ(ϑ)=ϱ(a)+1Γ(α)(Daαϱ)(ϕ)(ϑ-a)α] holds for all ϑ∈(a,b], where a≤ϕ≤ϑ. Corollary 1 Let ϱ(x),Daαϱ(ϑ)∈C[a,b], for 0<α≤1. If the fractional derivatives Daαϱ(ϑ)≥0 for all ϑ∈[a,b], then ϱ(ϑ) is non-decreasing for each ϑ∈[a,b]. On the other hand, if Daαϱ(ϑ)≤0, then, for all ϑ∈[a,b], then, ϱ(ϑ) is non-increasing for every ϑ∈[a,b]. Now we consider a time instant t1 with t0≤t≤t1 such thatES(t)=0,for0≤t≤t1,=0,fort=t1,<t1fort=t1+. From the system (3), note that DtαES\|ES(t1)=0. Applying Corollary 1, it is observed that ES(t1)=0 contradicting our assumption. Thus we have ES(t)≥0, at any instant t. In a similar manner, it can be shown that EI(t),A(t),V(t)≥0 implying that all the solutions of the system (3) are non-negative for any time t∈[t0,∞). Uniform boundedness of the solutions Here we determine whether the solutions of the system (3) are bounded. For that, we construct the function h(t):R0,+⟶R0,+ defined by5 h(t)=ϱ1+ϱ2+ϱ3+ϱ4,∀t≥t0. Thus placing the values of ϱi, i=1,2,3,4, we obtain6 h(t)=(Π1+Π2)-((1-ε1)θβESA1+kV+qEIη+V-(1-ε2)pEI)V-(μ1ES+μ2EI+μ3A+μ4V)≤(Π1+Π2)-μ(ES+EI+A+V), where μ=min{μ1,μ2,μ3,μ4} and (1-ε1)θβESA1+kV+qEIη+V<(1-ε2)pEI. Thus we get,7 h(t)≤(Π1+Π2)-μϑ(t),∀t≥t0. Therefore using Lemma 3 presented in [33], and for t→∞, it can be concluded that all the solutions trajectories of the system (3) initiating from R+ are uniformly bounded in the region8 Ω={(ES,EI,A,V)∈R+:ES+EI+A+V≤Π1+Π2μ}. Local existence and uniqueness of solutions In this subsection, we will investigate the local existence and uniqueness of solution trajectories (ES,EI,A,V) of the system (3). The existence criteria for the solutions of fractional order system (3) would be studied using the theorem proposed in [34] which is presented below: Theorem 2 LetP=[t0-a,t0+a],B={ϑ∈R+:‖ϑ-ϑ0‖≤b},D={(t,ϑ)∈[t0,∞)×R+:t∈P,ϑ∈B}. Further we assume that the function ϱ:D⟶R+4 satisfies the following conditions: (i) ϱ(t,ϑ(t)) is Lebesgue measurable for t∈P; (ii) ϱ(t,ϑ(t)) is continuous where ϑ belongs to B; (iii) there is a real-valued function w(t)∈L2(P) such that ‖ϱ(t,ϑ(t))‖≤w(t), for almost every t∈P and for all ϑ(t) belongs to B. Then for α be chosen as α>1/2, there exists at least one solution to the initial value problem (4) in the interval [t0-h,t0+h], for some h>0 [34]. With the help of Theorem 2, we discuss the uniqueness of solutions of the system (3). Theorem 3 Suppose that the assumptions (i)–(iii) of Theorem 2 hold and there exists a real-valued function Λ(t)∈L4(P) such that9 ‖ϱ(t,ϑ)-ϱ(t,ν)‖≤Λ(t)‖ϑ-ν‖, for almost every t∈P and all ϑ,ν∈B. Then there exists a unique solution of the initial value problem (4) on the interval [t0,∞)×R+ [34]. Proof In order to prove the uniqueness of the solution trajectories of the system (3), following the method proposed by Li et al. [33], we construct the function ϱ(t):R+→R+4 as ϱ(t,ϑ)=(ϱ1(t,ϑ),ϱ2(t,ϑ),ϱ3(t,ϑ),ϱ4(t,ϑ)), where ϱ1, ϱ2, ϱ3 and ϱ1 are defined previously. Here we utilize the norm ‖ϱ(t,ϑ)‖=\|ϱ1(t,ϑ)\|+\|ϱ2(t,ϑ)\|+\|ϱ3(t,ϑ)\|+\|ϱ4(t,ϑ)\|, for ϱ(t)∈R+. Note that R+ is endowed with the proper norm ‖.‖ and [t0,∞)×R+ is a Banach space with respect to this norm [34]. Let us choose any two points ϑ(t)=(ES,EI,A,V) and ν(t)=(ES′,EI′,A′,V′) belonging to the region R+ and for these two points ϑ(t) and ν(t),‖ϱ(t,ϑ)-ϱ(t,ν)‖=\|ϱ1(t,ϑ)-ϱ1(t,ν)\|+\|ϱ2(t,ϑ)-ϱ2(t,ν)\|+\|ϱ3(t,ϑ)-ϱ3(t,ν)\|+\|ϱ4(t,ϑ)-ϱ4(t,ν)\|. Using the definitions of ϱ1,ϱ2,ϱ3, and ϱ4, we can write‖ϱ(t,ϑ)-ϱ(t,ν)‖=\|μ1(ES-ES′)+(1-ε1)β(ESVA1+kV-ES′V′A′1+kV′)\| +\|μ2(EI-EI′)+(1-ε1)β(ESVA1+kV-ES′V′A′1+kV′)\|+\|μ3(A-A′)+(1-ε1)θβ(ESVA1+kV-ES′V′A′1+kV′)\|+\|μ4(V-V′)-(1-ε2)p(EIV-EI′V′)+q(EIVη+V-EI′V′η+V′)\|≤μ1\|(ES-ES′)\|+3(1-ε1)βK′(\|ES-ES′\|+\|V-V′\|+\|A-A′\|)+μ2\|(EI-EI′)\|+3(1-ε1)βK′(\|ES-ES′\|+\|V-V′\|+\|A-A′\|)+μ3\|(A-A′)\|+3(1-ε1)θβK′(\|ES-ES′\|+\|V-V′\|+\|A-A′\|)+μ4\|(V-V′)\|+2(1-ε2)pK′(\|EI-EI′\|+\|V-V′\|)+2qK′η2(\|EI-EI′)+\|V-V′\|)=(μ1+6(1-ε1)βK′+3(1-ε1)θβK′)\|ES-ES′\|+(μ2+2(1-ε2)pK′+2qK′η2)\|EI-EI′\|+(μ3+3(1-ε1)βK′+3(1-ε1)θβK′)\|A-A′\|+(μ4+6(1-ε1)βK′+3(1-ε1)θβK′+2(1-ε2)pK′+2qK′η2)\|A-A′\|≤L′‖ϑ-ν‖, considering L' = max (μ1+6(1-ε1)βK′+3(1-ε1)θβK′, μ2+2(1-ε2)pK′+2qK′η2, μ3+3(1-ε1)βK′+3(1-ε1)θβK′, μ4+6(1-ε1)βK′+3(1-ε1)θβK′+2(1-ε2)pK′+2qK′η2. Thus ϱ(t,ϑ) satisfies Lipschitz’s condition for ϑ∈R+ [33]. Now using the Banach Contractive Mapping Principle [34], it can be concluded that a unique real-valued function Λ(t)∈L4(P) must exist such thatΛ(t)=ϑ0+L′Γ(α)∫t0t(t-s)α-1ϱ(s,Λ(s))ds. Consequently, the initial value problem (4) exhibits unique solution on the interval [t0,∞)×R+. Hence the proof is completed. Global existence of the solutions In this subsection, the global existence of the solution trajectories of the system (3) is studied using Theorem 3.1 [34]. Theorem 4 Let the assumptions (i)–(ii) of Theorem 2 and the following condition: ‖ϱ(t,ϑ)‖≤κ1+κ2‖ϑ‖ hold in global space, where κ1>0 and κ2>0 are two constants, for almost each t∈[t0,∞) and all ϑ(t)∈R+. Then there exists a function ϑ(t)∈[0,+∞) which is a solution of the initial value problem (4) [34]. Proof With respect to certain t0∈[t0,∞) and ϑ0∈R+ and using the assumptions of the above statement, it can be observed that ϱ(t,ϑ(t)) is locally bounded in the region D. Again the weak singularity of ϱ(t,ϑ(t)) indicates the existence of the solution ϑ(t) of the initial value problem (4) specified on the interval [t0,∞)×R+. Let ϑ(t) possesses a maximal existence interval [0,l′)⊂[0,+∞) such that l′<+∞. According to [34], the solution ϑ(t) can be expressed asϑ(t)=ϑ0+1Γ(α)∫t0t(t-s)α-1ϱ(s,ϑ(s))ds. Thus, employing the assumptions of the statement, the Eq. (9) is transformed into the following form:‖ϑ(t)‖≤‖ϑ0‖+κ1Γ(α+1)\|l′-t0\|α+κ2∫t0t(t-s)α-1‖ϑ(s)‖ds, for t0,t∈[0,l′) together with t0≤t. With the help of the generalized Gronwall inequality, we can find a constant z such that ‖ϑ(t)‖<z on [t0,l′) [34]. Consequently, according to Theorem 3.1 of [34], the global existence and uniqueness of solution ϑ(t) of the initial value problem (4) can be ensured on the interval [0,+∞). Hence the proof is completed. Steady states and stability of the fractional order system This subsection is concerned with the steady states executed by the epidemic system (3) and the stability of the fractional order system (3) around these steady states. Disease-free equilibrium (DFE) The infected compartments of our proposed epidemic fractional order system (3) are EI(t) and V(t). Thus solving the system (3) by letting EI(t)=0 and V(t)=0, we find that the system (3) executes the disease-free equilibrium (DFE) E0=(Π1/μ1,0,Π2/μ3,0). Next we compute the Jacobian matrix JE0 of the system (3) around the DFE E0 to study the stability of the system (3) which is given byJE0=-μ100-(1-ϵ1)βΠ1Π2μ1μ30-μ20(1-ϵ1)βΠ1Π2μ1μ300-μ3-(1-ϵ1)βθΠ1Π2μ1μ3000-μ4. With the aid of the following theorem [35], we would study the local stability of the fractional order system (3) around the DFE. Theorem 5 Let the fractional order system (3) together with the initial condition (2) be expressed as Dtαϑ(t)=Mϑ(t), with the same definition of ϑ(t) as stated previously and M∈R4×4. The system (3) is locally asymptotically stable iff all the four eigenvalues λi, i=1,2,3,4, of the matrix M satisfy the relation \|arg(λi)\|>πα2 and the system is stable iff the eigenvalues would hold the relation \|arg(λi)\|≥πα2 satisfying the critical condition \|arg(λi)\|=πα2 along with the geometric multiplicity equal to one [35]. Proof Obviously the eigenvalues of Jacobian matrix JE0 are -μ1, μ2, μ3 and -μ4 which are strictly real and negative. Hence \|arg(λi)\|=π>πα2, since 0<α<1 (i=1,2,3,4). This shows the fact that the system (3) is locally asymptotically stable around the DFE. Endemic equilibrium (EE) The epidemic system (3) possesses a unique endemic equilibrium E∗(ES∗,EI∗,A∗,V∗). The components of EE are obtained by solving equations DtαES(t)=0, DtαEI(t)=0, DtαA(t)=0, DtαV(t)=0 and are given byES∗=π1-μ2EI∗μ1,A∗=π2-θμ2EI∗μ3,V∗=ημ4-{ηp(1-ϵ2)-q}EI∗(1-ϵ2)pEI∗-μ4, whereas the value of EI∗ is derived from the cubic equation10 a1EI∗3+a2EI∗2+a3EI∗+a4=0, where,a1=θμ22β(1-ϵ1){(1-ϵ2)ηp-q},a2=μ2[(1-ϵ2)p-k{ηp(1-ϵ2)-q}]-(1-ϵ1)β{(π1θ+π2)(ηp(1-ϵ2)-q)+ηθμ22μ4},a3=(1-ϵ1)β[{ηp(1-ϵ2)-q}π1π2-(π1θ+π2)ημ2μ4+μ2μ4(kη-1)],a4=-(1-ϵ1)βηπ1π2μ4. Next we calculate the Jacobian matrix of the system (3) about the EE which is given byJE∗=-j110-j13-j14j21-j22j13j14-θj210-j33-θj140j420-j44, where,j11=μ1+(1-ϵ1)βA∗V∗1+kV∗,j21=(1-ϵ1)βA∗V∗1+kV∗,j22=μ2,j42=(1-ϵ2)pV∗-qV∗η+V∗,j13=(1-ϵ1)βES∗V∗1+kV∗,j33=(1-ϵ1)βθES∗V∗1+kV∗+μ3,j14=(1-ϵ1)βES∗A∗(1+kV∗)2,j44=qEI∗η(η+V∗)2-(1-ϵ2)pEI∗+μ4. The characteristic equation of the Jacobian matrix JE∗ corresponding to the eigenvalue λ can be written as11 λ4+ζ1λ3+ζ2λ2+ζ3λ+ζ4=0, where,ζ1=j11+j22+j33+j44,ζ2=(j11+j33)(j22+j44)+j11j33+j22j44-j14j42+j13j21,ζ3=j11j33(j22+j44)+j22j44(j11+j33)-j14j42(j11+j33)+j13j21(j22+j44)θ+j13j14j42θ+j14j21j42,ζ4=j11j33(j22j44-j14j42)+j13j21j22j42θ+j14j42θ(j21j33-j13j21). Next we present two propositions to study the local asymptotic stability of the system (3) around the EE. Proposition 1 If each eigenvalue λi,i=1,2,3,4 of the Jacobian matrix JE∗ satisfies the condition \|arg(λ)\|>λπ2. Then the epidemic system (3) is locally asymptotically stable around the EE (using Theorem 5). Next, we determine the discriminant of the characteristic Eq. (11) as:Υ(φ)=1ζ1ζ2ζ3ζ40001ζ1ζ2ζ3ζ40001ζ1ζ2ζ3ζ443ζ12ζ2ζ3000043ζ12ζ2ζ3000043ζ12ζ2ζ3000043ζ12ζ2ζ3, 12 =ζ12ζ22ζ32-4ζ12ζ23ζ4-4ζ13ζ32+18ζ13ζ2ζ3ζ-27ζ14ζ22+4ζ23ζ32+16ζ24ζ4+18ζ1ζ2ζ33-80ζ1ζ22ζ3ζ4-6ζ12ζ32ζ4+144ζ12ζ2ζ42-27ζ34+144ζ2ζ32ζ4-128ζ22ζ42-192ζ1ζ3ζ42+256ζ43. In terms of the discriminant Υ(φ), we construct the following proposition to study the local asymptotic stability of the system (3) around the endemic equilibrium point. Proposition 2 [22] (a) The epidemic system (3) is locally asymptotically stable around the EE if Υ(φ)>0, in addition to the conditions (i) ζ1>0, (ii) ζ1ζ2>ζ3, and (iii) ζ1ζ2ζ3-ζ12ζ4-ζ3>0. (b) The epidemic system (3) is locally asymptotically stable around the EE for α∈(0.5,1), if Υ(φ)<0, in addition to the conditions (i) ζ1>0, (ii) ζ2>0, (iii) ζ1ζ2>ζ3, and (iv) ζ1ζ2ζ3-ζ12ζ4-ζ3=0. (c) The epidemic system (3) is unstable around the EE for α>2/3, if Υ(φ)<0 together with the conditions (i) ζ1<0, (ii) ζ2<0, and (iii) ζ3<0. Optimal control of the fractional order model In this section, we discuss the possible diminishment of the SARS-CoV-2 infection through optimal control strategy using 2-DG. In our study, we implement the robust control method - output feedback H∞ control where discrete time frame is stipulated and is based on Pontryagin’s Minimum Principle [36]. Optimal strategies and fractional order control model In Epidemiology, optimal control technique is applied to a mathematical model attributed to biological or biomedical state of a disease system. Several researches have been conducted to study the dynamical effects of optimal control in fractional order mathematical modeling [37–44]. At first, we determine optimal control strategies based on Pontryagin’s Minimum Principle [36] and use the explicit expression of the stability conditions. Two control variables u1(t) and u2(t) are introduced in the fractional order system (3) satisfying 0≤ui(t)≤1,i=1,2. Particularly ui=0,i=1,2 indicate that no drug is implemented in the epidemic system and ui(t)=1,i=1,2 indicate maximal use of the drug 2-DG and its influence to combat SRAS-CoV-2 infection. The drug 2-DG is mainly prescribed to human host for the purpose of turning on the host immune system and to mediate in stimulating the epithelial cells proliferation in a controlled manner (since extravagant proliferation of epithelial cells may cause carcinoma). Therefore, the fractional order model (3) is converted into the following fractional order control model:13 DtαES=Π1-(1-ε1)βESVA1+kV-μ1ES+u1(t)ES1+ρ1ES,DtαEI=(1-ε1)βESVA1+kV-μ2EI,DtαA=Π2-(1-ε1)θβESVA1+kV-μ3A+u2(t)A1+ρ2A,DtαV=(1-ε2)pEIV-(μ4+qEIη+V)V, with the same initial conditions (2). The constants ρ1 and ρ2 represent the half-maximal simulations for host immune response. The control system (13) can be expressed in the following form:14 Dtαy=f(y(t),u(t),t),y(0)=y0, where y=(ES,EI,A,V)T and y0=(ES0,EI0,A0,V0)T. To minimize the infected epithelial cells load and viral load, we define the following objective functional15 J(u1(t),u2(t))=∫T0Tf[w1u12(t)+w2u22(t)+w3EI2+w4V2]dt, where w1u12(t) and w2u22(t) stand for the cost-benefit of 2-DG in inducing the epithelial cells proliferation and agitating host immune response respectively. Here w1 and w2 are weights that assist in regularizing the optimal controls; w3 and w4 keep balance in the load of the infected epithelial cells and virus population. Our aim is to find the optimal controls (u1∗(t),u2∗(t)) implemented in the system (13) to minimize the objective functional (15) over the control set U such that16 inf(u1(.),u2(.))∈UJ(u1(.),u2(.))=J(u1∗(.),u2∗(.)), where U is defined as U=U1×U2={(u1(t),u2(t)):0≤ui(t)≤1,i=1,2;uiaremeasurable;t∈[T0,Tf]}. Derivation of optimal conditions for the FOCP In this subsection, we frame the general formulation and derivation of fractional order control problem (FOCP) described through the control induced fractional order system (13) [22]. We rewrite the objective functional (15) subjected to the system (14) asJ(u)=∫T0Tff(y(t),u(t),t)dt. In view of the system (14), we define the FOCP as17 infJ(u)=∫T0Tff(y(t),u∗(t),t)dt. Accordingly, we can rewrite the FOCP in compact form as18 infu(.)∈UJ(u(.))=J(u∗(.)), subjected to the state system19 Dtαy=g(y(t),u∗(t),σ),y(0)=y0, where the co-state vector σ satisfies the following condition:20 Dtgασ=∂f∂y+σT∂g∂y,σ(tf)=0. Using the algorithm presented in [45], we conclude the optimal control u∗(t) is the solution of the equation21 ∂f∂u∗+σT∂g∂u∗=0. Note that the Eqs. (19), (20), and (21) represent the Euler-Lagrange optimal conditions for the FOCP and with the help of these conditions, we determine the existence conditions so that controls u1∗(t) and u2∗(t) would be optimal. In this context, we construct a Hamiltonian function for the FOCP as22 H=f+σiTgi,i=1,2,3,4, where f=w1u12(t)+w2u22(t)+w3EI2+w4V2, σ=(σ1,σ2,σ3,σ4)T, gi=(g1,g2,g3,g4)T. The following theorems provide the means to minimize the objective functional (15). Theorem 6 There exist a pair of optimal controls ui∗(t),i=1,2 to the FOCP (13) subjected to the state system (19). We consider a Hamiltonian for the FOCP (13) subjected to the state system (19) as23 H=w1u12(t)+w2u22(t)+w3EI2+w4V2+∑i=14σigi. Next we prove the existence of the optimal control and verify the following conditions: The state system (13) possesses bounded coefficient and hence the controls ui∗(t),i=1,2 and the corresponding state variables are non-empty. As the state system (13) is bounded, the control set U is convex and closed. Since the state system (13) is bilinear in u1(t) and u2(t), the R.H.S. of the system (13) is bounded by a linear function associated with state and control variables. The integrand w1u12(t)+w2u22(t)+w3EI2+w4V2 is convex on U. There exist δ1>0, δ2>0 and p>1 such that the integrand of the objective functional (15) satisfies w1u12(t)+w2u22(t)+w3EI2+w4V2≥δ1(\|u1\|2+\|u2\|2)p/2-δ2, where δ1 and δ2 depend on the boundedness of EI and V, since w1>0 and w2>0. Now, applying Pontryagin’s Minimum Principle [36] to the Hamiltonian (23), we prove the following theorem. Theorem 7 Associated with the optimal control pair (u1∗(t),u2∗(t)) and solution of the state system (13), there exists adjoint variables σi,i=1,2,3,4 satisfying the adjoint system of equations24 Dtgασ1=-∂H∂ES,Dtgασ2=-∂H∂EI,Dtgασ3=-∂H∂A,Dtgασ4=-∂H∂V, 25 thatis,Dtgασ1=(1-ε1)βVA1+kV(σ1-σ2+θσ3)+σ1μ1-σ1u1(t)(1+ρ1ES)2,Dtgασ2=-2w3EI+σ2μ2-σ4(1-ε2)pV-σ4qVη+V,Dtgασ3=(1-ε1)βESV1+kV(σ1-σ2+θσ3)+σ3μ3-σ3u2(t)(1+ρ2A)2,Dtgασ4=-2w4EI+(1-ε1)βESA(1+kV)2(σ1-σ2+θσ3)-σ4(1-ε2)pEI+μ4σ4+σ4ηqEI(η+V)2, with the boundary conditions σi(tg)=0, for i=1,2,3,4. Furthermore, the expressions for optimal control pair (u1∗(t),u2∗(t)) are determined through the relation (21) characterized by26 u1∗(t)=-σ1ES2w1(1+ρ1ES),u2∗(t)=-σ3A2w2(1+ρ2A). Hence, the boundedness of the optimal control pair (u1∗(t),u2∗(t)) could be defined as27 u1∗(t)=min{max{-σ1ES2w1(1+ρ1ES),0},1},u2∗(t)=min{max{-σ3A2w2(1+ρ2A),0},1}. Proof Using standard results of Pontryagin’s Minimum Principle [36], the expressions of adjoint variables and boundary conditions can be derived. By partially differentiating the Hamiltonian (23) with respect to the corresponding states, the adjoint equations system can be expressed through (25) together with the boundary conditions σi(tg)=0, for i=1,2,3,4. With the help of Pontryagin’s Minimum Principle [36], it can be observed that the unrestricted optimal controls pair (u1∗(t),u2∗(t)) must satisfy∂H∂u1∗(t)=0,∂H∂u2∗(t)=0. We observe thatH=w1u12(t)+w2u22(t)+σ1u1(t)ES1+ρ1ES+σ3u2(t)A1+ρ2A+othertermsexcludingu1(t)andu2(t), which leads to28 ∂H∂u1(t)=2w1u1+σ1ES1+ρ1ES,∂H∂u2(t)=2w2u2+σ3A1+ρ2A. Consequently, solving Eq. (28) we obtain29 u1∗(t)=-σ1ES2w1(1+ρ1ES),u2∗(t)=-σ3A2w2(1+ρ2A). Thus boundedness of the optimal controls pair assists us to derive the control functions u1∗(t) and u2∗(t) in the following form:u1∗(t)=0,-σ1ES2w1(1+ζ1ES)≤0;-σ1ES2w1(1+ζ1ES),0<-σ1ES2w1(1+ζ1ES)<1;1,-σ1ES2w1(1+ζ1ES)≥1. u2∗(t)=0,-σ3A2w2(1+ζ2A)≤0;-σ3A2w2(1+ζ2A),0<-σ3A2w2(1+ζ2A)<1;1,-σ3A2w2(1+ζ2A)≥1. The optimal controls pair u1∗(t) and u2∗(t) in the compact form can be written as30 u1∗(t)=min{max{-σ1ES2w1(1+ζ1ES),0},1},u2∗(t)=min{max{-σ3A2w2(1+ζ2A),0},1}. Therefore it is notable that the solution of the FOCP could be obtained by replacing u1(t) and u2(t) in place of the optimal controls u1∗(t) and u2∗(t) in the system (13). Accordingly, the optimality of the FOCP establishes a two-point boundary value problem with reference to a system of fractional order differential equations. Remark 1 The optimality of the controlled system (13) referred to the adjoint system (25) together with the defined initial and boundary conditions and the optimal controls pair (u1∗(t),u2∗(t)) characterized by (29) could be presented asDtαES=Π1-(1-ε1)βESVA1+kV-μ1ES+u1(t)ES1+ζ1ES,DtαEI=(1-ε1)βESVA1+kV-μ2EI,DtαA=Π2-(1-ε1)θβESVA1+kV-μ3A+u2(t)A1+ζ2A,DtαV=(1-ε2)pEIV-(μ4+qEIη+V)V,DTfασ1=(1-ε1)βVA1+kV(σ1-σ2+θσ3)+σ1μ1-σ1u1(t)(1+ζ1ES)2,DTfασ2=-2w3EI+σ2μ2-σ4(1-ε2)pV-σ4qVη+V,DTfασ3=(1-ε1)βESV1+kV(σ1-σ2+θσ3)+σ3μ3-σ3u2(t)(1+ζ2A)2,DTfασ4=-2w4EI+(1-ε1)βESA(1+kV)2(σ1-σ2+θσ3)-σ4(1-ε2)pEI+μ4σ4+σ4ηqEI(η+V)2. The above optimal system reveals that it is required to keep the human host immune system strong enough to control the SARS-CoV-2 infection, which is possible if susceptible epithelial cells proliferation would be high. Implementation of the drug 2-DG benefits to lessen the level of infected epithelial cells and the load of virions in host body.Fig. 1 Solution trajectories of the system (3) for different values of memory (α=1.0,0.90,0.80) keeping other parameter values same as listed in Table 1 Numerical findings To construct a fractional order model, the Caputo fractional derivatives are applied extensively in epidemiology to model infectious diseases taking into account the interactions between host immune response and virus particles in the past by incorporating the characteristic “memory” in the system. In this section, we intend to numerically visualize the kinetic behaviors of our proposed fractional order system (3) and control induced fractional order system (13) for memory 0<α≤1. In order to solve our proposed fractional order system (3) numerically, we follow the iterative scheme presented in [46, 47] using MATLAB by taking the baseline parameter values from Table 1. Our proposed fractional order system (3) is fitted with the real-time patient data from Germany [17, 18].Fig. 2 Solution trajectories of the system (3) varying the destruction rate q of virions via host immune response (q=0.3,0.4,0.5) with memory α=0.8 keeping other parameter values same as listed in Table 1 Numerical simulation of epidemic system (3) without control Figure 1 describes the dynamical behavior of the epidemic system (3) varying the memory effect α, where α=1 implies the alignment of the fractional order system (3) with its corresponding integer order system. It is observed that the stability of the solution trajectories exhibits periodic nature for α=1, but in case of fractional order parameter values with α=0.90 and α=0.80, the disease system converges to its endemic steady state in shorter time. Figure 2 portrays the behavioral changes in the system (3) considering different values of q, the destruction rate of SARS-CoV-2 virions through immune response (taking α=0.8). The host immune response utilizes its capability to destroy the SARS-CoV-2 virions through immunological memory. Figure 2 shows that increased immune destruction using immunological memory enables to control the COVID-19 infection. Figure 3 displays the phase portraits of the system (3) in phase spaces ES-EI-A (left panel) and EI-A-V (right panel) exhibiting the dynamical behavior of the system in presence of memory (α=0.70,0.80,0.90) and also for α=1.0. The figure is showing that infection level could be monitored through immunological memory of healthy immune system.Fig. 3 Left Panel: Phase portrait of the system (3) corresponding to the state variables ES(t),EI(t) and V(t) for different values of memory (α=1.0,0.90,0.80,0.70). Right Panel: Phase portrait of the system (3) corresponding to the state variables EI(t),A(t) and V(t) for different values of memory (α=1.0,0.90,0.80,0.70) In Figs. 4 and 5, the efficacy of 2-DG in prohibiting the transmission of the infection and replication of SARS-CoV-2 virions are observed, respectively. It is worthwhile to notice that the reduced level of COVID-19 infection proper administration of the drug might be recommended. In this scenario, appropriate policies based on fractional order optimal control would be helpful in monitoring the level of infection and mitigating the infection. Fig. 4 Solution trajectories of the system (3) varying the efficacy ε1 of the drug 2-DG in blocking transmission of COVID-19 (0≤ε1,&ε1=0) while α=0.8 keeping other parameter values as same as listed in Table 1 Fig. 5 Solution trajectories of the system (3) varying the efficacy ε2 of the drug 2-DG in blocking transmission of COVID-19 (0≤ε2,&ε2=0) while α=0.8 keeping other parameter values as same as listed in Table 1 Numerical simulation of epidemic system (13) with control This subsection is concerned with directional behavior of the system (13) when fractional order optimal controls are applied to the disease system. In this regard, we numerically solve the state system (13) as an initial value problem and the co-state system (25) as a boundary value problem. Using an efficient iterative method, we obtain Figs. 6 and 7 whereas forward iterative scheme is applied to solve the state system (13) and backward iterative scheme is applied to solve the co-state system (25) for the values of the fractional order parameter α=1,0.85,0.75. In Figs. 6 and 7, varying the weightage of the drug 2-DG, it is observed that the weightage of the drug input increases (in Fig. 7) as it agitates the host immune system in a regular manner. The healthy increase of immunity via antiviral effects of the drug 2-DG works as a stimulant in proliferation of the epithelial cells. Here we consider estimated value of the half-maximal constants as ρ1,2=0.3.Fig. 6 Behaviors of the controlled system (13) varying memory (α=1,0.85,0.75) in presence of the control input u1 (left panel) and the control input u2 (right panel) Discussion Calibration of the complex multi-scale reciprocity between host and viral particles at micro level for the newly emerged COVID-19 infection is an exigent topic in the present scenario. In this study, a four-dimensional deterministic cell-limited model has been framed delineating the interplay among the host epithelial cells, host immune response and SARS-CoV-2 virions in COVID-19 transmission process. Our proposed epidemic model is perturbed into a Caputo fractional order deterministic system in presence of immunological memory. The apprehension regarding previous status of an epidemic benefits in inhibition of transmission and control of the infection. Caputo fractional differential equations are capable of proving specified and biologically interpretable initial conditions in modeling of a disease system. Additionally, the Caputo fractional order system has advantages in flexible utilization of classical initial conditions leading the non-negativity, uniqueness and local as well as global existence of solutions of the proposed epidemic system. The system is locally asymptotic stable around both the disease-free steady state and endemic steady state executed by the system.Fig. 7 Behaviors of the controlled system (13) varying memory (α=1,0.85,0.75) in presence of the increased weightage of the control input u1 (left panel) and the control input u2 (right panel) Quantifying the role of 2-DG drug in controlling the SARS-CoV-2 infection via improvement of host immune response, fractional order optimal control problem (FOCP) has presented in the study. FOCP benefits to determine optimal dose of the drug 2-DG and its minimum systemic cost using Pontryagin’s Minimum Principle. Based on real data and some estimated data, the dynamical behaviors of the fractional order system (for both without control and controlled systems) have been studied numerically. We observed from our analytic as well as numerical findings that fractional order model generates better results than its integer order counterpart. These findings will assist the health policy makers for better administration of 2-DG in prevention and control of the SARS-CoV-2 infection on a global basis. Author contribution Conceptualization: A. N. C., J. M. Formal analysis: J. M., P. S. Investigation: J. M., P. S., A. N. C. Methodology: B. A., J. M. Project administration: B. A., A. N. C. Supervision: J. M., B. A. Writing — original draft: J. M., P. S. Writing — review and editing: B. A., A. N. C. Funding This research has been supported to the first author to pursue her Ph.D. (Swami Vivekananda Merit-cum-Means Scholarship) by the West Bengal Higher Education Department, Govt. of West Bengal, Bikash Bhavan, India with G.O. No. 52-Edn(B)/5B-15/2017 dated 07.06.2017. Declarations Ethical approval The research demonstrated here has adhered to the accepted ethical standards of a genuine research study. Informed consent The content of the article is not submitted anywhere yet. Conflict of interest The authors declare no competing interests. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. COVID Live-Coronavirus Statistics-Worldometer, https://www.worldometers.info/coronavirus/. Accessed on 18 Jan 2022 2. Organization, W.H.: WHO Coronavirus (COVID-19) Dashboard, https://covid19.who.int. Accessed on 18 Jan 2022 3. Du SQ Yuan W Mathematical modeling of interaction between innate and adaptive immune responses in COVID-19 and implications for viral pathogenesis J. Med. Virol. 2020 92 9 1615 1628 10.1002/jmv.25866 32356908 4. Mondal, J., Samui, P., Chatterjee, A.N.: Dynamical demeanour of SARS-CoV-2 virus undergoing immune response mechanism in COVID-19 pandemic. Eur. Phys. J. Spec. Top., 1–14 (2022) 5. Shang J Wan Y Luo C Ye G Geng Q Auerbach A Li F Cell entry mechanisms of SARS-CoV-2 Proc. Natl. Acad. Sci. U.S.A. 2020 117 21 11727 11734 10.1073/pnas.2003138117 32376634 6. Addeo A Friedlaender A Cancer and COVID-19: Unmasking their ties Cancer Treat. Rev. 2020 88 102041 10.1016/j.ctrv.2020.102041 32516704 7. Dariya B Nagaraju GP Understanding novel COVID-19: its impact on organ failure and risk assessment for diabetic and cancer patients Cytokine Growth Factor Rev. 2020 53 43 52 10.1016/j.cytogfr.2020.05.001 32409230 8. Nath BJ Dehingia K Mishra VN Chu YM Sarmah HK Mathematical analysis of a within-host model of SARS-CoV-2 Adv. Difference Equ. 2021 2021 1 1 11 10.1186/s13662-021-03276-1 9. Mondal J Khajanchi S Mathematical modeling and optimal intervention strategies of the COVID-19 outbreak Nonlinear Dyn. 2022 109 1 177 202 10.1007/s11071-022-07235-7 35125654 10. Husniah H Ruhanda R Supriatna AK Biswas MH SEIR Mathematical Model of Convalescent Plasma Transfusion to Reduce COVID-19 Disease Transmission Mathematics 2021 9 22 2857 10.3390/math9222857 11. Khajanchi, S., Sarkar, K., Mondal, J., Nisar, K.S., Abdelwahab, S.F.: Mathematical modeling of the COVID-19 pandemic with intervention strategies. Res. Phys. 25, 104285 (2021) 12. Khoshnaw SH Salih RH Sulaimany S Mathematical modelling for coronavirus disease (COVID-19) in predicting future behaviours and sensitivity analysis Math. Mod. Natl. Phen. 2020 15 33 10.1051/mmnp/2020020 13. Mondal J Samui P Chatterjee AN Optimal control strategies of non-pharmaceutical and pharmaceutical interventions for COVID-19 control J. Interdisc. Math. 2021 24 1 125 153 10.1080/09720502.2020.1833459 14. Okuonghae D Omame A Analysis of a mathematical model for COVID-19 population dynamics in Lagos, Nigeria Chaos, Solitons & Fractals 2020 139 110032 10.1016/j.chaos.2020.110032 32834593 15. Shahzad M Abdel-Aty AH Attia RA Khoshnaw SH Aldila D Ali M Sultan F Dynamics models for identifying the key transmission parameters of the COVID-19 disease Alex. Eng. J. 2021 60 1 757 765 10.1016/j.aej.2020.10.006 16. Thomas DM Sturdivant R Dhurandhar NV Debroy S Clark N A primer on COVID-19 mathematical models Obesity 2020 28 8 1375 1377 10.1002/oby.22881 32386464 17. Hernandez-Vargas, E.A., Velasco-Hernandez, J.X.: In-host mathematical modelling of COVID-19 in humans. Annu. Rev. Control. (2020) 18. Wang S Pan Y Wang Q Miao H Brown AN Rong L Modeling the viral dynamics of SARS-CoV-2 infection Math. Biosci. 2020 328 108438 10.1016/j.mbs.2020.108438 32771304 19. Chimal-Eguia JC Mathematical model of antiviral immune response against the COVID-19 virus Mathematics 2021 9 12 1356 10.3390/math9121356 20. Paul, S., Mahata, A., Ghosh, U., Roy, B.: Study of SEIR epidemic model and scenario analysis of COVID-19 pandemic. Ecological Genetics and Genomics 19, 100087 (2021) 21. Chatterjee AN Ahmad B A fractional-order differential equation model of COVID-19 infection of epithelial cells Chaos, Solitons & Fractals 2021 147 110952 10.1016/j.chaos.2021.110952 33967407 22. Chatterjee, A.N., Al Basir, F., Almuqrin, M.A., Mondal, J., Khan, I.: SARS-CoV-2 infection with lytic and non-lytic immune responses: A fractional order optimal control theoretical study. Res. Phys. 26, 104260 (2021) 23. Wang, B., Mondal, J., Samui, P., Chatterjee, A.N., Yusuf, A.: Effect of an antiviral drug control and its variable order fractional network in host COVID-19 kinetics. Eur. Phys. J. Spec. Top., 1–15 (2022) 24. Naik, P.A., Yavuz, M., Qureshi, S., Zu, J., Townley, S.: Modeling and analysis of COVID-19 epidemics with treatment in fractional derivatives using real data from Pakistan. Eur. Phys. J. Plus 135(10), 1–42 (2020) 25. Chen TM Rui J Wang QP Zhao ZY Cui JA Yin L A mathematical model for simulating the phase-based transmissibility of a novel coronavirus Infect. Dis. Poverty 2020 9 1 1 8 10.1186/s40249-020-00640-3 31996251 26. Mahata A Paul S Mukherjee S Das M Roy B Dynamics of Caputo Fractional Order SEIRV Epidemic Model with Optimal Control and Stability Analysis Intl. J. Appl. Comp. Math. 2022 8 1 1 25 27. Paul, S., Mahata, A., Mukherjee, S., Roy, B.: Dynamics of SIQR epidemic model with fractional order derivative. Part. Diff. Eq. Appl. Math. 5, 100216 (2022) 28. Paul S Mahata A Mukherjee S Roy B Salimi M Ahmadian A Study of Fractional Order SEIR Epidemic Model and Effect of Vaccination on the Spread of COVID-19 Intl. J. Appl. Comp. Math. 2022 8 5 1 16 29. Mahata, A., Paul, S., Mukherjee, S., Roy, B.: Stability analysis and Hopf bifurcation in fractional order SEIRV epidemic model with a time delay in infected individuals. Part. Diff. Eq. Appl. Math. 5, 100282 (2022) 30. Akindeinde SO Okyere E Adewumi AO Lebelo RS Fabelurin OO Moore SE Caputo fractional-order seirp model for covid-19 pandemic Alex. Eng. J. 2022 61 1 829 845 10.1016/j.aej.2021.04.097 31. Chatterjee AN Basir FA Ahmad B Alsaedi A A Fractional-Order Compartmental Model of Vaccination for COVID-19 with the Fear Factor Mathematics 2022 10 9 1451 10.3390/math10091451 32. Odibat ZM Shawagfeh NT Generalized Taylor’s formula Appl. Math. Comput. 2007 186 1 286 293 33. Li HL Zhang L Hu C Jiang YL Teng Z Dynamical analysis of a fractional-order predator-prey model incorporating a prey refuge J. Appl. Math. Comput. 2017 54 1 435 449 10.1007/s12190-016-1017-8 34. Lin W Global existence theory and chaos control of fractional differential equations J. Math. Anal. Appl. 2007 332 1 709 726 10.1016/j.jmaa.2006.10.040 35. Petráš, I.: Fractional-Order Nonlinear Systems: Modeling, Analysis and Simulation. Springer Science & Business Media (2011) 36. Pontryagin, L.S.: Mathematical Theory of Optimal Processes. CRC press (1987) 37. Al-Basir F Elaiw AM Kesh D Roy PK Optimal control of a fractional-order enzyme kinetic model Control. Cybern. 2015 44 4 443 461 38. Baleanu D Mohammadi H Rezapour S A fractional differential equation model for the covid-19 transmission by using the caputo-fabrizio derivative Adv. Difference Equ. 2020 2020 1 1 27 10.1186/s13662-020-02762-2 39. Cao X Datta A Al Basir F Roy PK Fractional-order model of the disease psoriasis: a control based mathematical approach J. Syst. Sci. Complexity 2016 29 6 1565 1584 10.1007/s11424-016-5198-x 40. Demirci E Ozalp N A method for solving differential equations of fractional order J. Comput. Appl. Math. 2012 236 11 2754 2762 10.1016/j.cam.2012.01.005 41. Diethelm, K.: The Analysis of Fractional Differential Equations: An application-Oriented Exposition Using Differential Operators of Caputo type. Springer Science & Business Media (2010) 42. Kilbas, A.A., Srivastava, H.M., Trujillo, J.J.: Theory and Applications of Fractional Differential Equations, vol. 204. Elsevier (2006) 43. Naik PA Zu J Owolabi KM Global dynamics of a fractional order model for the transmission of HIV epidemic with optimal control Chaos, Solitons & Fractals 2020 138 109826 10.1016/j.chaos.2020.109826 32572309 44. Roy, P.K., Datta, A., Rana, S., et al.: The Fractional-order differential equation model of psoriatic pathogenesis: A mathematical study. Afr. Diaspora J. Math. New Series 15(2), 35–46 (2013) 45. Agrawal OP A formulation and numerical scheme for fractional optimal control problems J. Vib. Control 2008 14 9–10 1291 1299 10.1177/1077546307087451 46. Garrappa R Short tutorial: Solving fractional differential equations by Matlab codes 2014 Italy Department of Mathematics University of Bari 47. Sardar T Rana S Chattopadhyay J A mathematical model of dengue transmission with memory Commun. Nonlinear Sci. Numer. Simul. 2015 22 1–3 511 525 10.1016/j.cnsns.2014.08.009	36459249	PMC9716179	NO-CC CODE	2022-12-09 23:25:59	no	J Biol Phys. 2022 Dec 2; 48(4):415-438	utf-8	J Biol Phys	2,022	10.1007/s10867-022-09617-9	oa_other
==== Front Ann Am Acad Pol Soc Sci Ann Am Acad Pol Soc Sci ANN spann The Annals of the American Academy of Political and Social Science 0002-7162 1552-3349 SAGE Publications Sage CA: Los Angeles, CA 10.1177/00027162221122682 10.1177_00027162221122682 Single-Parent Families in the U.S. during COVID-19 Economic Precarity among Single Parents in the United States during the COVID-19 Pandemic Gornick Janet C. Maldonado Laurie C. Sheely Amanda Parolin Zachary Lee Emma K. [email protected] 7 2022 7 2022 7 2022 702 1 Single-Parent Families and Public Policy: Evidence from High-Income Countries 206223 © 2022 by The American Academy of Political and Social Science 2022 American Academy of Political & Social Science This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Single-parent families have historically faced greater economic precarity relative to other family types in the United States. We investigate how and whether those disparities widened after the onset of the COVID-19 pandemic. Using data on exposure to school and childcare center closures, unemployment, poverty, food hardship, and frequent worrying among single-parent families versus two-parent families throughout 2020 and 2021, we find that the challenges that single parents faced prior to the pandemic generally magnified after the arrival of COVID-19. In April 2020, one in four single parents was unemployed, and unemployment rates recovered more slowly for single parents throughout 2021, perhaps in part due to their unequal exposure to school and childcare closures. The expansion of income transfers largely buffered against potential increases in poverty and hardship, but levels of worrying among single parents continued to worsen throughout 2021. poverty hardship COVID-19 single parents economic insecurity typesetterts1 ==== Body pmcSingle-parent families have historically faced greater economic precarity relative to other family types in the United States. Studies have demonstrated that this is in part due to the United States providing less generous income support for single parents (or families with children more broadly) compared to many other high-income countries (Maldonado and Nieuwenhuis 2015; Aerts, Marx, and Parolin, this volume). Moreover, from the early 1990s through 2019, the American welfare state increasingly targeted income transfers at working parents, leaving jobless parents with few opportunities to access cash-based social assistance. In 2019, this work-oriented welfare state, combined with high levels of employment (including among single parents), contributed to a record-low poverty rate. In 2020, however, the COVID-19 pandemic sent the U.S. unemployment rate to a record high, rapidly exposing the limitations of the country’s employment-centered welfare state. In April 2020, the same month in which unemployment climbed to 19 percent in the United States, childcare centers across the country closed, most schools turned to distance learning, and levels of poverty and hardship threatened to rise. However, the challenges of the COVID-19 pandemic were not shared equally across family types. This study documents how trends in economic and social insecurity varied for single-parent families relative to two-parent families throughout 2020 and 2021. Specifically, we use data from several different sources to investigate trends in five measures related to economic precarity: exposure to school and childcare center closures, unemployment, poverty, food hardship, and subjective well-being. These five indicators are closely intertwined, and they can offer a multidimensional perspective of the challenges facing families with children: school and childcare centers may contribute to declines in employment for parents with care responsibilities, a task that is already more challenging for single-parent families relative to two-parent families. In turn, school and care closures may contribute to higher rates of joblessness (or both employment and closures may simply be endogenous to the spread of COVID-19 and government-enforced mobility restrictions). Regardless, declining employment may affect poverty and food hardship. Meanwhile, frequent worrying, one dimension of subjective well-being, may be associated with poverty and food hardship; conversely, in the context of the global health crisis, frequent worrying may be detached from current economic conditions. Investigating each of these dimensions offers a broad account of the economic precarity of single parents throughout the pandemic. Our findings lead to three broad takeaways. First, the neighborhoods with the highest shares of single parents faced greater exposure to school and childcare closures during the 2020–2021 school year. Additionally, single parents consistently faced higher rates of unemployment, poverty, food insufficiency, and frequent worrying relative to two-parent families throughout the COVID-19 pandemic. Second, we find that the strongest reductions in poverty and food hardship for single-parent families occurred after the expansion of the Child Tax Credit (CTC) in July 2021, which provided monthly and unconditional cash support for most families with children. Specifically, the CTC payments provided monthly cash payments of $300 per child under 6, and $250 per child between 6 and 18, from July through December 2021. Thus, while single-parent families generally faced more challenges throughout the pandemic relative to other family types, the trends observed after the introduction of the CTC suggest that a sustained distribution of unconditional cash support for lower-income families could reduce economic insecurity for single parents moving forward. Third, trends during the pandemic show notable ebbs and flows across our indicators of economic precarity for single parents: poverty and food hardship gradually increased throughout 2020 before declining in 2021, while unemployment among single parents steadily declined after April 2020, but at a slower rate than for two-parent families. Despite the improved economic conditions in 2021 for single parents, however, their levels of frequent worrying continued to climb throughout 2021 (relative to 2020 levels and relative to two-parent families), suggesting that the challenges of single parents during COVID-19 cannot be reduced solely to economic concerns. Background Economic precarity of single parents in the United States Economic insecurity, according to Western et al. (2012), is “the risk of economic loss faced by workers and households as they encounter the unpredictable events of social life.” Rather than focusing only on the risk of economic loss, however, this study broadens its focus to economic precarity, which encompasses (1) realized economic loss, as observed through trends in unemployment, poverty, and food hardship; (2) factors exacerbating economic insecurity, such as challenges in managing the work-care nexus for families with children; and (3) the consequences of economic insecurity for subjective well-being. Put differently, we use the term economic precarity in a broad, multidimensional manner to encompass the correlated challenges of care work, unemployment, poverty, food hardship, and subjective well-being facing single parents during the COVID-19 pandemic. The United States has higher rates of single parenthood relative to other high-income countries (Brady, Finnigan, and Hübgen 2017). Close to one in four U.S. households with children are headed by a single parent, with 80 percent of these households headed by a woman (Nieuwenhuis and Maldonado 2018; U.S. Census Bureau 2020). Additionally, U.S. single parents are at a particularly high risk for poverty compared to single parents in other high-income Western democracies (Maldonado and Nieuwenhuis 2015). According to 2019 Census data, 36.5 percent of single-parent families headed by a woman with children under the age of 18 and 16.3 percent of single-parent families headed by a man lived in poverty (U.S. Census Bureau 2019). Nieuwenhuis and Maldonado’s (2018) “Triple Bind,” which outlines the inadequacies of resources, employment, and policies for single parents, helps to explain the particularly high risk of poverty for single parents in the United States. Single-parent families tend to have fewer resources, such as lower levels of parental education and limited childcare, income, time, and flexibility, compared to coupled-parents. Despite there being a high single-parent employment rate, gendered inequality in employment and high rates of precarious employment among single parents contributes to persistent inequalities relative to other family types. Moreover, many redistributive policies and so-called “social investment” policies are inadequate at reducing these economic inequalities, as investments into childcare and similar services often advantage multiparent families more so than single-parent families. And given that the majority of single parents are women, gender inequality also tends to exacerbate the disadvantages that single parents in the United States experience relative to two-parent families. COVID-19 and challenges of single parents The onset of the COVID-19 pandemic contributed to widespread school closures, record levels of unemployment, and the potential for high rates of hardship in the United States. We posit, however, that the experience of economic precarity was not distributed equally across family types. Given the weak welfare state and labor market institutions supporting single parents in the United States prior to March 2020, we anticipate that the onset of the pandemic, and its correlated crises, exacerbated the challenges that single parents in the United States tend to face. We expect this to be true despite the large increase in social spending that resulted from the Coronavirus Aid, Relief, and Economic Security (CARES) Act passed in March 2020. Specifically, we investigate trends in economic precarity across five domains: school and childcare center closures, unemployment, poverty, food hardship, and frequent worrying. We discuss each domain in brief. School and childcare center closures In April 2020, the first peak of the COVID-19 pandemic, forty-eight U.S. states and Washington, D.C. mandated or strongly recommended the closure of schools and childcare centers (EducationWeek 2020). Our research suggests that close to 90 percent of children were exposed to school closures and 70 percent of families with young children were exposed to childcare closures in April 2020 (Parolin and Lee 2021; Lee and Parolin 2021). These school and childcare closures varied by geography and disproportionately affected minority and disadvantaged populations. Single-parent families, we posit, may be more likely to be exposed to school and childcare closures, given that more single-parent families are concentrated in large, dense cities (where lockdowns were more severe) relative to rural areas. Even if exposure to school and childcare closures were to be neutrally distributed, however, they may carry greater relative consequences for single parents’ employment opportunities. Unemployment In April 2020, the unemployment rate in the United States climbed to 19 percent, the highest level in recorded history (since 1948) by the Current Population Survey (U.S. Bureau of Labor Statistics 2020, 2022). Many more individuals, meanwhile, exited the labor force altogether; by definition, these former workers are not currently seeking work and thus would not be counted as unemployed. The closure of schools and childcare centers, a necessary step to prevent the spread of COVID-19, likely contributed to the employment crisis for parents with children, and single parents in particular. Without education and childcare services, working parents were abruptly forced to find a new family-work balance that, for many, required a reduction in paid work hours. The number of U.S. mothers of school-age children not actively working increased by 1.4 million from January 2020 to January 2021 (Heggeness et al. 2021). Early studies suggested that the decrease in the number of employed single mothers was particularly high for those with children five years old or younger, as well as Black and Hispanic single mothers (Barroso and Kochhar 2020). The decline in employment was also strongly influenced by government-imposed lockdowns. Given that many single-parent families live in denser cities, they may face a geographic penalty, on average, relative to two-parent families when it comes to the lack of employment. Beyond geographic factors, single parents are less likely to have a college degree and, in turn, are more likely to work in the service sector, which was the hardest hit by the pandemic (Lund et al. 2020). Poverty With high unemployment rates, the risk of poverty increased for many families during the COVID-19 pandemic. However, federal policy interventions over the course of the pandemic have significantly reduced this risk. The Families First Act, which expanded Supplemental Nutrition Assistance Program (SNAP) benefits, and the CARES Act, which provided economic relief through stimulus checks and increased unemployment benefits, were both passed in March 2020. A model by the Urban Institute estimates that early pandemic-response policies, specifically the Families First and CARES Acts, kept 10.3 million additional people out of poverty despite the large increase in unemployment (Giannarelli, Wheaton, and Acs 2020). The CARES Act led to April and May 2020 poverty rates that were even lower than prepandemic rates (Parolin et al. 2022). In particular, the CARES Act lifted more than eighteen million people out of monthly poverty in April 2020. However, upon the expiration of the unemployment supplement in August 2020, the number of people lifted out of poverty by the CARES Act decreased to only four million individuals (Parolin et al. 2022). In March 2021, the American Rescue Plan Act (ARPA), implemented various antipoverty policies, including the expanded CTC. Research estimates that the ARPA lifted over twelve million people out of poverty (Parolin, Collyer, et al. 2021). While several studies have documented the estimated poverty-reduction effects of each of these policies, it remains unclear how trends in poverty for single parents have evolved throughout the pandemic. Given the higher likelihood of unemployment and exposure to school and care closures, the policies may have been particularly effective at reducing poverty among single-parent families. Food hardship Despite increased unemployment and widespread economic downturn, food hardship rates in 2020 were similar to those in 2019 (Coleman-Jensen et al. 2021). Steady food insecurity rates were observed even with continuing grocery store closures, transportation barriers, and elevated food prices, which increased 3.8 percent in 2020 (higher than the 2 percent average increase for the past 20 years) (U.S. Bureau of Labor Statistics 2021). Many families were forced to reduce dietary variety and quality, as well as increase their reliance on charitable food sources (Waxman and Gupta 2021). Policies such as SNAP and the Pandemic Electronic Benefit Transfer (P-EBT), which provided families with vouchers equal in value to school meals missed due to remote learning, and the CTC have likely helped to mitigate food insecurity in the United States. As was also the case before the pandemic, U.S. households at the highest risk for food insecurity included those experiencing unemployment, families with children, households headed by women, and racial/ethnic minority families (Coleman-Jensen et al. 2021; Morales, Morales, and Beltran 2021). Stress, anxiety, and worrying With school and childcare closures, unemployment being at an all-time high, increased risk of poverty, and other hardships, many parents have experienced frequent stress, anxiety, and worrying, as well as other indicators of negative mental health. According to data from the Census Household Pulse Survey (hereafter “Pulse”), on average, 36 percent of adults reported symptoms of anxiety or depressive disorder from April 2020 to October 2021 (Kaiser Family Foundation 2021). The share of adults experiencing these symptoms was lowest in June 2021 (29 percent) and highest in November 2021 (43 percent) (Kaiser Family Foundation 2021). In contrast, only 11 percent of adults reported symptoms of anxiety or depressive disorder from January to June 2019 (National Center for Health Statistics 2021). Research suggests that parental well-being strongly correlates with the amount of hardship that a family has experienced over the COVID-19 pandemic (Gassman-Pines, Ananat, and Fitz-Henley 2020). More specifically, mothers who have taken on additional childcare responsibilities as a result of school and childcare closures have reported increased stress and anxiety (Calarco et al. 2020). Unemployment, especially for women, during the pandemic has also correlated with an increase in symptoms of anxiety and depressive disorder (Huato and Chavez 2021). Taken in full, it is likely that single parents, and especially single mothers, have experienced elevated levels of frequent worrying over the course of the COVID-19 pandemic. These five indicators of economic precarity: school and childcare closure, unemployment, poverty, food hardship, and frequent worrying, are likely to have worsened for single parents during the COVID-19 pandemic. Current literature suggests that this population is particularly vulnerable to worsening challenges, but there is limited research on how and whether disparities in these indicators have widened during the COVID-19 pandemic. Data and Methods We now discuss how we measure and track trends in each of these domains of economic precarity for single parents throughout the COVID-19 pandemic. Measuring economic precarity We use three sources of data to investigate our five domains of economic precarity (school and care closures, unemployment, poverty, food hardship, and subjective well-being) on a monthly basis throughout 2020 and 2021. The measures we operationalize across each domain have precedent in recent academic literature; thus, we primarily summarize the measures here and refer to the online appendix for in-depth detail on methodologies underlying each measure. We emphasize, however, that this article is the first to summarize trends in economic precarity across each of these five domains and also the first to focus on trends specifically for single parents in the United States. Table 1 summarizes our five measures and their underlying data sources. Table 1 Summary of Measures of Economic Precarity and Underlying Data Sources Domain Definition Data Source, Dates Precedent(s) School and childcare center closures The share of schools (or formal childcare centers) in the family’s census tract experiencing a 50%+ decline in in-person visits relative to the same month in 2019 U.S. School Closure & Distance Learning Database; U.S. Database of Child Care Closures during COVID-19, Sept. 2020–May 2021 Parolin and Lee (2021); Lee and Parolin (2021) Unemployment Whether the adult was active in the labor market but not working in the prior week U.S. Current Population Survey, Jan. 2020–Dec. 2021 (Many) Poverty Monthly poverty rates using Supplemental Poverty Measure framework U.S. Current Population Survey, Jan. 2020–Dec. 2021 Parolin et al. (2022) Food hardship Whether the family “sometimes” or “often” did not have enough to eat in the last 7 days Census Household Pulse Survey, Apr. 2020–Dec. 2021 Schanzenbach and Pitts (2020); Parolin, Ananat, et al. (2021) Frequent worrying The respondent reports “not being able to stop or control worrying” for more than half the days or nearly every day in the prior week. Census Household Pulse Survey, Apr. 2020–Dec. 2021 Parolin, Ananat, et al. (2021) Measuring single parenthood We measure single parenthood as a parent with children (under age 18) who does not live with other adults. We recognize that this definition does not identify all nonpartnered parents. For example, a nonpartnered parent with children who lives with his or her parents (i.e., grandparents of child) would not classify as a single parent in our definition, as there are multiple adults in the household who increase the earnings potential of the resource-sharing unit. The consequence of our decision is that we capture a slightly more vulnerable group of single parents relative to adopting a broader definition. For example, the 2019 poverty rate for this group of single parents was 26.7 percent, compared to 24.5 percent if we included single parents who lived in the same resource-sharing unit as other adults. In practice, however, our results suggest that adopting either definition demonstrates similar levels and trends in our outcomes relative to two-parent families. We identify 5.4 percent of the U.S. population living in single-parent families (using the 2019 Current Population Survey). We also recognize that there exists large heterogeneity among single-parent families. Among our single parents, 83 percent are women, 56 percent are non-White (31.4 percent are Black and 19.7 percent are Hispanic), and 40 percent do not have more than a high school degree (compared to 26 percent with a college degree). We focus on single-parent families at large in our primary analyses, while acknowledging that many of our outcomes may vary notably among single-parent families. Findings We present our findings in five steps, covering single parents’ exposure to school closures and childcare closures, unemployment, poverty, food hardship, and frequent worrying, in that order. Figure 1 presents a binned scatterplot documenting how the share of single parents in a given school district or census tract (or “neighborhood”) is associated with the likelihood of school closures or childcare center closures.1 Figure 1 Binned Scatterplot Showing Association of Share of Single-Parent Households in School District (Left Panel) or Census Tract Where Childcare Center Is Located (Right Panel) with Share of Schools or Childcare Centers That Are Closed or at Greatly Reduced Capacity (September 2020–May 2021) NOTE: School closures and childcare center closures represent a 50 percent decline of in-person visits to the school or care center in the month in 2020 or 2021 relative to the same month in 2019. The left panel of Figure 1, documenting exposure to school closures, shows a nonlinear relationship between the share of children in single-parent households and exposure to school closures. In the schools with the smallest share of students in single-parent homes (the leftmost point in the left panel), the mean rate of closures was 48 percent over the time span we examined. In schools with roughly average shares of single parents, the mean was lower at around 42 percent. However, the schools with the largest share of single parents were most exposed to school closures, with a mean of 52 percent. These statistics focus on rates of exposure, but evidence also suggests that even if rates of exposure were equal for single parents and other family types, it is the former who would likely struggle more in adapting to the challenges of distance learning and school closures. Consider, for example, that distance learning often requires an adult to be at home with the student throughout the day, a task that was likely particularly challenging for single-parent families. The right panel of Figure 1 documents exposure to childcare center closures. The nonlinear curve largely mirrors the relationship for school closures, although with slightly higher levels of closures for all groups. The census tracts with the highest share of single-parent families had the highest rate of childcare center closures (or strongly reduced capacity) at roughly 58 percent between September 2020 through May 2021. Again, focusing on exposure is useful for understanding differential access to formal care centers during the pandemic, but these statistics are only part of the story: even if single parents were exposed equally to care closures, it is likely that they would face more challenges than two-parent families (or childless families) in adapting to the challenges that the closures present. Without access to childcare centers, and/or with a need to remain at home to be with a child who is in distance learning, single parents are less likely to be able to combine care responsibilities with formal employment. This leads to our next indicator of interest: unemployment. Figure 2 displays rates of unemployment from January 2020 through December 2021 for single parents (dashed gray line) and all adults (solid black line). In January 2020, prior to the onset of the pandemic, the unemployment rate among single parents was 6.3 percent, compared to 4.4 percent among two-parent families. Put differently, single parents had an unemployment rate that was 1.9 percentage points higher than adults in two-parent homes. Figure 2 Trends in Unemployment from January 2020 through December 2021 In April 2020, however, unemployment rates spiked to 18.3 percent for two-parent families and 23.6 percent for single parents. In this month, single parents had an unemployment rate that was 5.3 percentage points higher than adults in two-parent homes. Thus, not only was the unemployment rate high for single parents, but it increased (in absolute terms) more steeply than for the adult population overall. This was also true for rates of joblessness (not depicted, but which includes both the unemployed and those who are not employed but not active in the labor market): joblessness was lower among single parents in January 2020 relative to parents in two-parent homes, but climbed higher in April 2020 (35.8 percent, up from 22.1 percent in January 2020, for single parents). Throughout the rest of 2020, unemployment rates declined for all, dropping to 8.3 percent for single parents and 6.8 percent for two-parent families in December 2020. In 2021, however, unemployment rates increased again for single parents, coinciding with a new wave of COVID-19 cases. With ebbs and flows throughout the year, unemployment among single parents reached 5.7 percent in December 2021, compared to 3.7 percent for adults in two-parent homes. Put differently, the unemployment rates in December 2021 for parents in both family types had fallen back down to prepandemic levels. The spikes in unemployment threatened to increase rates of poverty and hardship, our next two indicators of interest. Importantly, the federal government also introduced a large set of income transfers in the form of stimulus checks and expanded unemployment benefits, which are known to have reduced the national U.S. poverty rate (Parolin et al. 2022). Was this also true for single parents? Figure 3 documents trends in poverty for single parents (gray lines) and the population at large (black lines) before and after accounting for COVID-19-related relief. Specifically, we present estimates using a monthly version of the Supplemental Poverty Measure (SPM), which includes all taxes and transfers and adjusts the poverty threshold according to local living costs. The SPM is the preferred measure of poverty among most academic researchers and policymakers today in the U.S. context. Figure 3 Trends in Monthly SPM Poverty before and after Accounting for COVID-19-Related Income Transfers (January 2020–December 2021) NOTE: COVID-19-related income transfers include those passed in the CARES Act (expansion to unemployment benefits and stimulus checks) and those from the ARPA (expansion to unemployment benefits and monthly CTC payments). The large declines around March largely represent the payment of once per year refundable tax credits. In January 2020, the monthly SPM poverty rate was 34.6 percent for single-parent families, but 14.2 percent for two-parent homes. In April 2020, the month in which we saw large spikes in unemployment, the monthly SPM poverty rate for single parents climbed to 42.5 percent before accounting for COVID-19-related transfers, and to 21.7 percent for two-parent homes. When accounting for stimulus checks and expanded unemployment benefits, however, the story is different: the poverty rate in April 2020 when taking these benefits into account actually declined for single parents (to 32.1 percent) relative to prepandemic levels, as well as for two-parent homes (to 11.3 percent). When the $600 per week unemployment supplement expired in July 2020, however, poverty rates began to increase for both groups. By December 2020, single-parent families had a poverty rate of 34.7 percent (nearly identical to prepandemic levels), while two-parent families had a poverty rate of 15.1 percent (2.2 percentage points higher than the prepandemic level). Throughout 2021, poverty rates were volatile primarily due to two more rounds of stimulus payments (around January and March), refundable tax credit payments (around March), and the introduction of the monthly CTC payments (July onward). The end result, however, was large reductions in poverty, particularly for single parents. In December 2021, the monthly poverty rate for single-parent families was 21.9 percent, nearly a 13-percentage-point decline from prepandemic levels. This decline can be attributed to income transfers, and the CTC payments in particular, given that the poverty rate before accounting for the COVID-19-related transfers was near its prepandemic level. For two-parent families, the poverty rate in December 2021 was 9.6 percent, 4.6 percentage points lower than the January 2020 poverty rate. Do trends in food hardship align closely with trends in monthly poverty rates? Figure 4 presents evidence of this from April 2020 (the first month for which Pulse data is available) through December 2021. In April 2020, 21 percent of single parents experienced food insufficiency, compared to 12.6 percent of adults in two-parent homes. Rates of food insufficiency increased for both groups throughout 2020, aligning with the rise in poverty rates after the withdrawal of the $600 weekly unemployment benefit supplement in July 2020, as observed in Figure 4. By December 2020, food hardship had climbed to 26.9 percent for single parents (5.9-percentage-point increase from April) and to 17.8 percent for adults in two-parent homes (5.2-percentage-point increase from April). Figure 4 Trends in Food Insufficiency for Adults in Two-Parent Families and Single Parents from April 2020 through December 2021 NOTE: Authors’ calculations from the Census Household Pulse Survey. In 2021, however, levels of food insufficiency declined, particularly for single parents. This is consistent with the decline in poverty rates in Figure 4 after the introduction of the monthly CTC payments. In September 2021, 14.7 percent of single parents experienced food insufficiency, a 6.3-percentage-point decline from April 2020, while 11 percent of adults in two-parent homes experienced food insufficiency, a 1.6-percentage-point decline from April 2020. Put simply, our findings suggest that the expanded CTC payments contributed to notable declines in both poverty and food hardship for all families with children, but especially for single-parent families. Our final indicator of economic precarity focuses on one dimension of subjective well-being: frequent worrying. We view this measure as both (1) a subjective dimension of insecurity, in contrast to the economic measures presented before; and (2) an indicator that likely captures the combined stress of each of the challenges documented before (school and care closures, unemployment, poverty, and hardship), in addition to the health threat of the pandemic more broadly. Figure 5 presents these trends. Figure 5 Trends in Feelings of Frequent Worrying from April 2020 through December 2021 NOTE: Authors’ calculations from the Census Household Pulse Survey. In April 2020, 29.2 percent of single parents reported frequent worrying, compared to 25.4 percent of adults in two-parent homes. This rate climbed to 36.4 percent for single parents in December 2020, and to 31.5 percent for adults. In the early part of 2021, however, frequent worrying declined for both groups, although there was another slow rise in the latter half of 2021. By December 2021, levels of frequent worrying reached 31.9 percent for single parents, 2.7 percentage points higher than in April 2020; and 23 percent among adults in two-parent homes, a 2.4-percentage-point decline relative to April 2020. Although poverty and food insufficiency had declined from prepandemic levels by late 2021, frequent worrying had not among single parents, emphasizing that the toll of the pandemic on subjective well-being cannot be reduced purely to economic outcomes. Single parents, more so than other adults, appear to face particularly high and increased levels of worrying despite the average increase in their economic well-being throughout 2021. Conclusion Well before the onset of the COVID-19 pandemic, single parents in the United States faced particularly high rates of economic precarity relative to other family types. This study uses data from multiple sources to investigate how single parents compared with the general population throughout the COVID-19 pandemic on five measures related to economic precarity: exposure to school and childcare center closures, unemployment, poverty, food hardship, and frequent worrying. Our findings suggest that the challenges that single parents faced prior to the pandemic only magnified after the onset of the COVID-19 pandemic, albeit with notable month-to-month variation throughout 2020 and 2021. In April 2020, one in four single parents was unemployed, perhaps in part due to their unequal exposure to school and childcare center closures, as well as the greater difficulty in balancing work and care responsibilities when formal childcare is not available, or when a young student needs to attend school virtually from home. These high rates of unemployment could have led to high rates of poverty and hardship; however, the federal government intervened and distributed large levels of income support throughout 2020 and 2021, as documented in Table 2. Table 2 Summary of Major Policy Interventions Related to Income Support during 2020 and 2021 Income Support Date(s) Distributed Maximum Benefit Level Stimulus checks #1 April 2020a $1,200 per adult and $500 per dependent Expanded unemployment benefits #1 April–July 2020 Federal supplement of $600 per week with large expansions to access and maximum duration of benefits Stimulus checks #2 December 2020a $600 per adult and $600 per dependent Stimulus checks #3 March 2021a $1,400 per adult and $1,400 per dependent SNAP emergency allotments March 2020–2022b Upgrade of SNAP benefits to maximum benefit level Expanded unemployment benefits #2 March 2021–September 2021 Federal supplement of $300 per week with large expansions to access and maximum duration of benefits Expanded CTC July 2021–December 2021 $300 per child under age 6 per month; $250 per child between ages 6 and 18 per month a. The precise month of distribution of stimulus checks depends on a tax unit’s income and whether the IRS had its direct deposit information. We list the primary month of distribution here. b. Some states withdrew from provision of SNAP emergency allotments during 2021. Most dates continued providing the emergency allotments into 2022. The large distribution of income transfers in spring 2020 largely buffered against the risks of spikes in poverty and hardship: poverty rates briefly declined for single parents relative to prepandemic levels, but steadily increased again as the $600 weekly unemployment top-up expired. In 2021, however, two new rounds of stimulus checks, another expansion of unemployment benefits, and the introduction of the monthly, nonrefundable CTC payments led to strong declines in poverty and food insufficiency among single parents, both relative to two-parent families and the prepandemic poverty rate for single parents. The strong reduction effects of the CTC provide a clear policy takeaway: monthly, unconditional cash support appears to go a long way in improving the living standards for single parents. Continuing the monthly cash payments for low-income families with children would bring the United States in line with other high-income countries with child allowances (see Aerts, Marx, and Parolin, this volume) and would help reduce the comparatively high poverty rates in the United States. Despite the economic buffers, however, single parents’ levels of frequent worrying remained high throughout the pandemic and increased to around one in three single parents in late 2021 despite the economic improvements. The mismatch between rising levels of frequent worrying and declining poverty and hardship serves as a reminder that the challenges that single parents, in particular, face during the pandemic cannot be reduced simply to economic well-being. This study is unique in combining several sources of timely data to present a multidimensional portrait of economic precarity among single parents throughout 2020 and 2021. That said, our study has limitations that future research could build on. First, due to space constraints and data limitations, we focus on single parents at large despite notable heterogeneity among single parents. Our expectation is that single parents facing more layers of disadvantage (and, often, discrimination) face even greater risks than the averages we presented in this study. In particular, single parents who have low levels of education and/or who are racial/ethnic minorities are likely to have faced even greater challenges during COVID-19 than the present study documents. Second, our study exclusively uses quantitative data, but we recognize the value-add that qualitative accounts of the challenges of single parents during COVID-19 would bring to this research. Qualitative research, for example, could more clearly elucidate the mismatch between declining levels of hardship and poverty for single parents and rising levels of frequent worrying throughout 2021. Third, and last, there is likely vast regional variation across the United States in the challenges that single parents face. This variation can be due to differential exposure to the spread of COVID-19, different subnational policy regimes, and much more. Moving forward, scholars can build on this work to help make sense of different patterns of economic insecurity among single parents in different parts of the United States. Despite these limitations, the present study makes clear that the disadvantages that single parents faced in the United States relative to other family types prior to the pandemic only worsened after the arrival of COVID-19. New income support policies, such as the expanded CTC, helped to reduced poverty and hardship among single parents, but were not sufficient to reduce levels of frequent worrying among single parents in 2021, suggesting that the challenges of single parents during COVID-19 cannot be reduced solely to economic concerns. Supplemental Material sj-docx-1-ann-10.1177_00027162221122682 – Supplemental material for Economic Precarity among Single Parents in the United States during the COVID-19 Pandemic Click here for additional data file. Supplemental material, sj-docx-1-ann-10.1177_00027162221122682 for Economic Precarity among Single Parents in the United States during the COVID-19 Pandemic by Janet C. Gornick, Laurie C. Maldonado, Amanda Sheely, Zachary Parolin and Emma K. Lee in The ANNALS of the American Academy of Political and Social Science Zachary Parolin is an Assistant Professor of social policy at Bocconi University in Milan and a senior research fellow with Columbia University’s Center on Poverty and Social Policy. His recent work on poverty, inequality, and social policy has been published in the American Sociological Review, Demography, Nature Human Behaviour, and elsewhere. Emma K. Lee is a research assistant at Columbia University’s Center on Poverty and Social Policy. Her research on school and childcare closures in the United States during the COVID-19 pandemic appeared in Nature Human Behaviour and Socius. Supplemental Material: Supplemental material for this article is available online. 1. A binned scatterplot is a version of a scatterplot that is useful when plotting relationships with a very large number of observations. Rather than displaying a hundred thousand points in the scatterplot, we present ten points, with each point representing a decile value of the share of single parents in the area (X-axis) and the mean value of closures for all units within that given decile (Y-axis). ==== Refs References Aerts Elise Marx Ive Parolin Zachary . 2022. How to muster a million helping hands: On optimal minimum income support for single parents. The ANNALS of the American Academy of Political and Social Science (this volume). Barroso Amanda Kochhar Rakesh . 2020. In the pandemic, the share of unpartnered moms at work fell more sharply than among other parents. Pew Research Center. Available from https://www.pewresearch.org/fact-tank/2020/11/24/in-the-pandemic-the-share-of-unpartnered-moms-at-work-fell-more-sharply-than-among-other-parents/. Brady David Finnigan Ryan M. Hübgen Sabine . 2017. Rethinking the risks of poverty: A framework for analyzing prevalences and penalties. American Journal of Sociology 123 (3 ): 740–86. Available from https://www.journals.uchicago.edu/doi/full/10.1086/693678. Calarco Jessica M. Anderson Elizabeth Meanwell Emily Knopf Amelia . 2020. Let’s not pretend it’s fun: How COVID-19-related school and childcare closures are damaging mothers’ well-being. Available from 10.31235/osf.io/jyvk4. Coleman-Jensen Alisha Rabbitt Matthew P. Gregory Christian A. Singh Anita . 2021. Household food security in the United States in 2020. U.S. Department of Agriculture. Available from https://www.ers.usda.gov/webdocs/publications/102076/err-298.pdf?v=1143. EducationWeek. 2020. The coronavirus spring: The historic closing of U.S. schools (a timeline). Available from https://www.edweek.org/leadership/the-coronavirus-spring-the-historic-closing-of-u-s-schools-a-timeline/2020/07. Gassman-Pines Anna Ananat Elizabeth Oltmans Fitz-Henley II John . 2020. COVID-19 and parent-child psychological well-being. Pediatrics 146 (4 ): e2020007294. Available from 10.1542/peds.2020-007294. Giannarelli Linda Wheaton Laura Acs Gregory . 2020. 2020 poverty projections: Initial US policy response to the COVID-19 pandemic’s economic effects is projected to blunt the rise in annual poverty. Urban Institute. Available from https://www.urban.org/sites/default/files/publication/102521/2020-poverty-projections.pdf. Heggeness Misty L. Fields Jason García Trejo Yazmin A. Schulzetenberg Anthony . 3 March 2021. Tracking job losses for mothers of school-age children during a health crisis. Available from https://www.census.gov/library/stories/2021/03/moms-work-and-the-pandemic.html. Huato Julio Chavez Aida . 2021. Household income, pandemic-related income loss, and the probability of anxiety and depression. Eastern Economic Journal 47 (4 ): 546–70. Available from 10.1057/s41302-021-00199-3. Kaiser Family Foundation. 2021. Adults reporting symptoms of anxiety or depressive disorder during COVID-19 pandemic. Available from https://www.kff.org/other/state-indicator/adults-reporting-symptoms-of-anxiety-or-depressive-disorder-during-covid-19-pandemic/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D. Lee Emma K. Parolin Zachary . 2021. The care burden during COVID-19: A national database of child care closures in the United States. Socius 7 :1–10. Available from https://journals.sagepub.com/doi/abs/10.1177/23780231211032028. Lund Susan Ellingrud Kweilin Hancock Bryan Manyika James . 2020. COVID-19 and jobs: Monitoring the US impact on people and places. McKinsey Global Institute. Available from https://www.mckinsey.com/~/media/McKinsey/Industries/Public%20and%20Social%20Sector/Our%20Insights/COVID%2019%20and%20jobs%20Monitoring%20the%20US%20impact%20on%20people%20and%20places/COVID-19-and-jobs-Monitoring-the-US-impact-on-people-and-places.pdf. Maldonado Laurie C. Nieuwenhuis Rense . 2015. Family policies and single parent poverty in 18 OECD countries, 1978 2008. Community, Work & Family 18 :395–415. Available from https://www.tandfonline.com/doi/full/10.1080/13668803.2015.1080661. Morales Danielle Xiaodan Morales Stephanie Alexandra Beltran Tyler Fox . 2021. Food insecurity in households with children amid the COVID-19 pandemic: Evidence from the Household Pulse Survey. Social Currents 8 (4 ): 314–25. Available from https://journals.sagepub.com/doi/abs/10.1177/23294965211011593. National Center for Health Statistics. 2021. Estimates of mental health symptomatology, by month of interview: United States, 2019. Centers for Disease Control and Prevention. Available from https://www.cdc.gov/nchs/data/nhis/mental-health-monthly-508.pdf. Nieuwenhuis Rense Maldonado Laurie C. , eds. 2018. The triple bind of single-parent families: Resources, employment and policies to improve well-being. New York, NY: Policy Press. Parolin Zachary Ananat Elizabeth Collyer Sophie M. Curran Megan Wimer Christopher . 2021. The initial effects of the expanded child tax credit on material hardship. NBER Working Paper 29285, National Bureau of Economic Research, Cambridge, MA. Available from https://www.nber.org/papers/w29285. Parolin Zachary Collyer Sophie Curran Megan A. Wimer Christopher . 2021. The potential poverty reduction effect of the American Rescue Plan. Center on Poverty and Social Policy at Columbia University. Available from https://static1.squarespace.com/static/5743308460b5e922a25a6dc7/t/604aa2465cfc4a35b8a1c236/1615503943944/Poverty-Reduction-Analysis-American-Rescue-Plan-CPSP-2021.pdf. Parolin Z. Curran M. Matsudaira J. Waldfogel J. Wimer C. 2022. Estimating monthly poverty rates in the United States. Journal of Policy Analysis and Management. Available from 10.1002/pam.22403. Parolin Zachary Lee Emma K. 2021. Large socio-economic, geographic and demographic disparities exist in exposure to school closures. Nature Human Behaviour 5 (4 ): 522–28. Available from 10.1038/s41562-021-01087-8. Schanzenbach Diane Pitts Abigail . 2020. How much has food insecurity risen? Evidence from the census Household Pulse Survey. Northwestern Institute for Policy Research. Available from https://www.ipr.northwestern.edu/documents/reports/ipr-rapid-research-reports-pulse-hh-data-10-june-2020.pdf. U.S. Bureau of Labor Statistics. 2020. 14. Household survey: What would the unemployment rate be if these misclassified workers were included among the unemployed? Available from https://www.bls.gov/covid19/employment-situation-covid19-faq-april-2020.htm#ques14. U.S. Bureau of Labor Statistics. 2021. Consumer Price Index summary. Available from https://www.bls.gov/news.release/cpi.nr0.htm. U.S. Bureau of Labor Statistics. 2022. Labor force statistics from the Current Population Survey: Unemployment rate. Available from http://data.bls.gov/timeseries/LNS14000000?years_option=all_years. U.S. Census Bureau. 2019. Figure 9: Poverty rate and percentage-point change by type of family: Families and people. Available from https://www.census.gov/content/dam/Census/library/visualizations/2021/demo/p60-273/Figure9.pdf. U.S. Census Bureau. 2020. Table FG10. Family groups: 2020. Available from https://www.census.gov/data/tables/2020/demo/families/cps-2020.html. Waxman Elaine Gupta Poonam . 2021. Stories of hardship from families with young children as the COVID-19 pandemic persists. Urban Institute. Available from http://www.urban.org/Sites/default/files/publication/104147/stories-of-hardship-from-families-with-young-children-as-the-covid-19-pandemic-persists_1.pdf. Western Bruce Bloome Deirdre Sosnaud Benjamin Tach Laura . 2012. Economic insecurity and social stratification. Annual Review of Sociology 38 (1 ): 341–59. Available from https://www.annualreviews.org/doi/abs/10.1146/annurev-soc-071811-145434.	36474772	PMC9716188	NO-CC CODE	2022-12-03 23:20:55	no	Ann Am Acad Pol Soc Sci. 2022 Jul; 702(1):206-223	utf-8	Ann Am Acad Pol Soc Sci	2,022	10.1177/00027162221122682	oa_other
==== Front Comp Immunol Microbiol Infect Dis Comp Immunol Microbiol Infect Dis Comparative Immunology, Microbiology and Infectious Diseases 0147-9571 1878-1667 Elsevier Ltd. S0147-9571(22)00180-1 10.1016/j.cimid.2022.101923 101923 Article How can imported Monkeypox break the borders? A rapid systematic Review Ghazy Ramy Mohamed a1 Hammad Elsayed b2 Hall Mohamed Ashraf c3 Mahboob Amira d4 Zeina Sally e5 Elbanna Eman H. f6 Fadl Noha g7 Abdelmoneim Shaimaa Abdelaziz h8 ElMakhzangy Rony g9 Hammad Hammad Mohamed i10 Suliman Afrah humidan j11 Atia Hayat Hasab Alkreem k12 Rao Naman l13 Abosheaishaa Hazem m⁎14 Elrewany Ehab a15 Hassaan Mahmoud A. n16 Hammouda Esraa Abdellatif o17 Hussein Mai chp18 a Tropical Health Department, High Institute of Public health, Alexandria University, Egypt b Alexandria Faculty of Medicine, Egypt c Alexandria Dental Research Center, Egyptian Ministry of Health and Population, Egypt d Occupational health and industrial medicine department, high institute of public health, Alexandria university, Egypt e Department of Clinical Research, Maamora Chest Hospital, Ministry of Health and Population, Egypt f Health Management, Planning and Policy Department, High Institute of Public health, Alexandria University, Egypt g Family Health Department, High Institute of Public health, Alexandria University, Egypt h Clinical Research Administration, Alexandria Directorate of Health Affairs, Egyptian Ministry of Health and Population, Egypt i Al-Mana General Hospital, KSA j Faculty of medicine, University of Khartoum, Sudan k Central Lab, Ministry of Higher Education and Scientific Research, Maamorah, Khartoum, Sudan l Henry M. Goldman School of Dental Medicine, Boston University, USA m Mount Sinai Queens: New York, New York, USA n Institute of Graduate Studies & Research, Alexandria University Egypt o Head of clinical research department, El-Raml pediatric hospital, Ministry of health and population, Egypt p Harvard Medical School, Boston, MA, USA ⁎ Corresponding author. 1 ORCID:0000-0001-7611-706X 2 ORCID: 0000-0002-7523-2346 3 ORCID: 0000-0001-9452-4232 4 ORCID: 0000-0002-0785-3335 5 ORCID:0000-0001-5859-8437 6 ORCID:0000-0002-0419-5103 7 ORCID:0000-0001-9807-2720 8 ORCID:0000-0002-9197-6396 9 ORCID: 0000-0002-1089-2677 10 ORCID:0000-0001-6386-1682 11 ORCID: 0000-0001-6675-7047 12 ORCID:0000-0002-8307-9601 13 ORCID: 0000-0003-2805-7255 14 ORCID: 0000-0002-5581-8702 15 ORCID:0000-0002-8700-0630 16 ORCID: 0000-0003-0521-1360 17 ORCID:0000-0003-3269-6623 18 ORCID:0000-0003-2735-1428 28 11 2022 28 11 2022 10192311 9 2022 20 11 2022 25 11 2022 © 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier has created a Monkeypox Information Center (https://www.elsevier.com/connect/monkeypox-information-center) in response to the declared public health emergency of international concern, with free information in English on the monkeypox virus. The Monkeypox Information Center is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its monkeypox related research that is available on the Monkeypox Information Center - including this research content - immediately available in publicly funded repositories, with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the Monkeypox Information Center remains active. Background Monkeypox was designated as an emerging illness in 2018 by the World Health Organization Research and Development Blueprint, necessitating expedited research, development, and public health action. In this review, we aim to shed the light on the imported cases of monkeypox in attempt to prevent the further spread of the disease. Methodology An electronic search in the relevant database (Web of Science, PubMed Medline, PubMed Central, Google scholar, and Embase) was conducted to identify eligible articles. In addition to searching the grey literature, manual searching was carried out using the reference chain approach. Results A total of 1886 articles were retrieved using the search strategy with 21 studies included in the systematic review. A total of 113 cases of imported monkeypox were confirmed worldwide. Nineteen patients mentioned a travel history from Nigeria, thirty-eight infected cases had travel destinations from Europe, fifty-four cases traveled from European countries such as; Spain, France, and the Netherlands, one case from Portugal, and another one from the UK. All reported clades of the virus were West African clade. Nine studies showed the source of infection was sexual contact, especially with male partners. Six studies mentioned the cause of infection was contact with an individual with monkeypox symptoms. Two studies considered cases due to acquired nosocomial infection. Ingestion of barbecued bushmeat was the source of infection in three studies and rodent carcasses were the source of infection in the other two studies. Conclusion The development of functioning surveillance systems and point-of-entry screening is essential for worldwide health security. This necessitates ongoing training of front-line health professionals to ensure that imported monkeypox is properly diagnosed and managed. In addition, implementing effective health communication about monkeypox prevention and control is mandatory to help individuals to make informed decisions to protect their own and their communities' health. Abbreviations WHO, World Health Organization MSM, Men Who have sex with men R0, R naught JBI, Joanna Briggs Institute NOS, Newcastle-Ottawa Scale WOS, Web of Science HIV, Human immunodeficiency virus UK, United Kingdom USA, United States of America PCR, Polymerase chain reaction POE, Points of entry Keywords Monkeypox Imported Infectious disease emerging infection Nigeria ==== Body pmc1 Background Monkeypox is a re-emerging rare zoonotic infectious disease. The monkeypox virus is related to the orthopoxviral family and poxviridae genus. This virus's natural hosts are vertebrates and arthropods. Monkeypox was first identified in 1958 when two outbreaks occurred in research-held monkeys that started exhibiting symptoms of a pox-like illness. [1] In 1970, the disease was first confirmed in humans, by a child suspected of having smallpox in the Democratic Republic of Congo. In 2003, the first outbreak of monkeypox outside of Africa was reported in the United States of America (USA). [1.], [2.] Monkeypox was designated as an emerging illness in 2018 by the World Health Organization (WHO) Research and Development Blueprint, necessitating expedited research, development, and public health action [3]. The growing global monkeypox outbreak was deemed a Public Health Emergency of International Concern on July 23, 2022, by WHO Director-General. As of 10 November 2022, a total of 79 483 confirmed cases were reported with 49 deaths reported in 110 countries. The number of new cases reported increased by 2.5% in week 44 compared with week 43. Of note, 86.3% of cases were reported in 10 countries: the USA, Brazil, Spain, France, The United Kingdom (UK), Germany, Colombia, Peru, Mexico, and Canada. The highest prevalence was reported in regions of the Americas and Europe [4]. People under the age of 40 conform to the majority of confirmed cases of monkeypox, with a median age of 31 years [5]. This group was only born after the smallpox vaccination campaign was stopped, further emphasizing the absence of cross-protective immunity [5.], [6.]. The monkeypox virus is divided into two separate genetic clades, the Central African (Congo Basin) clade, and the West African clade. The Central African clade was thought to be more contagious and to produce more severe illness. Moreover, the Central African clade’s mortality rate was approximately triple that of the West African clade [7]. Of note, The majority of cases reported outside West and Central Africa are linked to the imported cases [7]. However, recently, several monkeypox outbreaks have been reported outside those endemic countries and have even gradually caused worldwide outbreaks without known epidemiological links to West or Central Africa [8]. Monkeypox can be spread by both infected humans and animals. Nevertheless, considering the known illness characteristics of the infected patients and the obvious localized or generalized skin lesions, silent human-to-human transmission from asymptomatic patients appears unlikely. [9.], [10.] Additionally, a significant portion of these male-to-male sexual (MSM) partners and bisexual men have had monkeypox virus infections, raising the risk of sexual transmission [9]. It is interesting to note that there is a cross-protectivity of the smallpox vaccination against smallpox and monkeypox virus infection due to the high nucleotide identity (96.3%) in the central region between these two viruses [9]. R naught (R0) is known as the reproduction ratio, which is a way of defining the disease's degree of transmissibility. According to an epidemiological modeling study, the monkeypox R0 value ranges from 1.10 to 2.40 in nations with minimal contact with orthopoxvirus species [11]. In circumstances of imported human or animal cases, this value implies that a monkeypox pandemic is about to break out [5.], [12.]. A person who is infected must take particular precautions to socially isolate and quarantine themselves due to the virus's ability to spread [3]. So, several steps must be taken to stop the viral transmission to halt the monkeypox outbreak. These steps include the creation of rapid diagnostic assays with high sensitivity and specificity, active surveillance and monitoring systems, and the potential use of the smallpox vaccine for post-exposure prophylaxis of close contacts [8.], [13.]. In this review, we aim to shed the light on the imported cases of monkeypox to prevent the further spread of the disease. This would help policymakers and stakeholders to implement more stringent public health and social measures for better epidemic control. 2 Methods The current study was conducted according to the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses, [14] and the Cochrane Handbook of Systematic Review and Meta-Analysis [15]. The protocol of the study was registered in the international prospective register of systematic reviews (PROSPERO) [CRD42022351486] 2.1 Studied outcomes Primary objective:• Trace the imported cases of monkeypox (source and destination) to prevent the further spread of the disease. Secondary objectives: • Identify the mode of transmission among imported cases. • Highlight the clade monkeypox. • Describe signs and symptoms of imported cases and secondary cases. • Address the control and preventive measures adopted by different health authorities that received the imported cases. 2.2 Search strategy An electronic search in the relevant database (Web of Science (WOS), PubMed Medline, PubMed Central, Google scholar, and Embase) was conducted by two authors to identify eligible articles. The keywords were included according to the different search platforms (Supplementary File). The database search ended on August 5, 2022. In addition to searching the grey literature, manual searching was carried out using the reference chain approach. It included searching for references from eligible articles, citation tracking, and looking at related articles for eligible articles. The eligibility criteria were: (a) any original articles about imported monkeypox; (b) all types of study designs; (c) no restriction regarding the year of publication. The exclusion criteria were: (a) non-human, in-vitro studies; (b) articles in languages other than English; (c) conference papers, abstracts only, author response, books, and reviews and (d) articles with inadequate or overlapping data. All references were imported into one Endnote library to delete duplicates and were exported to an excel sheet file for screening. An initial screening of titles and abstracts of selected articles was carried out by two authors independently. Next, two independent authors were assigned to screen the full texts. Any conflicts were resolved through discussion and consensus among authors. The first author was consulted if there was any disagreement. 2.3 Case definition Cases of monkeypox were confirmed by the presence of monkeypox virus deoxyribonucleic acid (DNA) by polymerase chain reaction (PCR) testing, next-generation sequencing of a clinical specimen, or isolation of monkeypox virus in culture from a clinical specimen. Imported monkeypox was confirmed when cases reported a history of traveling, within 21 days of illness onset, to a country with confirmed cases of monkeypox or where monkeypox is endemic. Clinically, monkeypox cases had a characteristic rash (i.e., deep-seated and well-circumscribed lesions, often with central umbilication; and lesion progression through specific sequential stages; macules, papules, vesicles, pustules, and scabs). [16] 2.4 Data extraction Essential data was extracted from the eligible articles including characteristics of participants (age, gender, occupation, suspected source of infection, mode of transmission, diagnosis, infection control, management, complications, and outcome) in addition to study characteristics (i.e., authors, year of publication, country, and study design). The primary outcome was the epidemiological characteristics of imported monkeypox. 2.5 Quality assessment The quality of the articles was evaluated independently by two authors and was added to the data extraction sheet. The findings of the quality assessment of eligible articles were compared. In case of inconsistent findings, a consensus was reached through discussion and consulting the first author. The Joanna Briggs Institute (JBI) critical appraisal checklist was used for case report studies. The JBI checklist consists of 8 items with four responses (yes, no, unclear, and not applicable)[17]. The quality of case series studies was assessed by National Heart, Lung, and Blood Institute (NIH) quality assessment tool. Studies were classified into good (7–9), fair (4–6) and poor (0–3).[18] The Newcastle-Ottawa Scale (NOS) was used for observational studies. Studies assessed by NOS were categorized as good, fair and poor [19]. (Supplementary File) 3 Results 3.1 Search results A total of 1886 articles were retrieved using the search strategies. We found 880 citations in 4 databases (WOS, PubMed Central/Medline, and EMBASE), 1000 citations in Google scholar, 4 citations in grey literature, and 2 citations through manual search. We excluded 215 studies as duplicates detected by endnote and 1616 studies during screening by title and abstract. After full-text screening, thirty-four full-text articles were excluded for reasons [duplicates (5), not relevant (13), not-imported monkeypox (14), full-text were not-available (1), results not-available (1)]. Of these, 20 studies were included in the systematic review. Additionally, 1 eligible article were found through manual search. Fig. 1 Fig. 1 : Flowchart of included studies. Fig. 1 3.2 Quality assessments After assessing the quality of the studies, we found 15 studies with good quality [8.], [10.], [20.], [21.], [22.], [23.], [24.], [25.], [26.], [27.], [28.], [29.], [30.], [31], [32.]. Six studies were of fair quality [8.], [33.], [34.], [35.], [36.], [37.]. Table 1 Table 1 Studies that addressed cases of Monkeypox across the globe. Table 1Author Year [Ref] Study setting Study design Population Criteria Job (Number of imported cases) Source The suspected source of infection \ mode of transmission Signs and symptoms Secondary cases complications \ side effects Diagnosis Clade of virus Vaccination history (poxvirus-derived vaccine) Treatment Control measures Outcome Quality score Erez, 2019 [21] Medical Center, Israel (Case report) male - 38 years Desk job (1) Nigeria Contact with rodent carcasses Fever, chills, generalized rash, genital lesion, Lymphadenopathy None None Electron microscopy, PCR Test, immunofluorescence assay, tissue culture, ELISA West African - NSAIDs, Penicillin, and Doxycycline Isolated in his residence Only 1\11 HCW agreed to be vaccinated and none of the household contacts agreed Recovered 7 Yong, 2020 [20] Tan Tock Seng Hospital emergency department, Singapore (Case report) Male - 38 years Administrative job (1) Nigeria Ingestion of barbecued bushmeat Fever, chills, myalgia, rash, genital lesion, vesiculopustular rash, Lymphadenopathy None None PCR Test, Electron microscopy, sequencing West African - - Isolated in a negative-pressure room in hospital Contacts received vaccine Quarantine of close contacts at home or a government facility Contact tracing Monitoring of HCWs and use of PPE Discharged 7 Costello, 2022 [33] A hospital, USA (Case report) Male -28 years - (1) Nigeria Human contact in Nigeria Skin burning, sensation, vesicular rash, oral affection, lymphadenopathy None Viral culture, RT-PCR West African - Acyclovir as empiric treatment for disseminated varicella zoster infection Contact and airborne isolation precautions Surveillance of HCWs - 6 Atkinson, 2022 [22] Specialist infectious disease hospital, UK (Case report) Male - 40 years - (1) Nigeria unknown Fever, generalized pustular rash, genital lesions None - Orthopoxvirus-specific RT- PCR, MPXV-specific assay - - - - - 7 Vaughan, 2018 [34] Accident and Emergency department at Blackpool Teaching Hospital, UK (Case report) Males Case 1: Naval officer Case 2: not mentioned (2) Nigeria Case 1: - Case 2: Contact with a monkeypox suspected case, ingestion of bush meat Fever, rash, lymphadenopathy, oral affection, scrotal lump - None Multiple molecular assays and sequencing analysis West African - Case 1: antibiotics for staphylococcal infections Case 2: antibiotics Isolated in hospital PEP (vaccine) for contacts PrEP(vaccine) for HCWs Active surveillance for high\ intermediate risk of exposure Passive surveillance for low risk of exposure Contacts received an information sheet about MPXV Contact tracing Stable and improving 5 Hobson, 2021 [35] A hospital, UK Emergency department of COVID-19 zone (Case report) Male - female - toddler 18 months (same family) - (1) Nigeria Case 1: (index case) Unknown, Case 2: direct contact (secondary case), Case 3: direct contact (tertiary case) Vesicular lesions 1 secondary + 1 tertiary cases within the family of the index case) None PCR Test and sequencing West African None - Isolated in hospital HCWs in the HCID Unit were offered vaccination Active surveillance for close contacts Passive surveillance for contacts at lower risk of exposure Decontaminat-ion of residence COVID-19 travel quarantine limited MPXV transmission Control measures in the ED All recovered 4 Hammerschlag, 2022 [23] A primary care clinic, Australia (Case report) HIV-positive male in 30 s - on ART - MSM - history of syphilis - (1) Europe Sexual contact (MSM) Rash, painless pustules that became painful and pruritic lesions, fever, malaise, lymphadenopathy - Complications: super imposed infection\ bacterial cellulitis of the penile shaft and lower abdomen Genome sequencing, Phylogenetic analysis, Electron microscopy West African - Ceftriaxone and Doxycycline for gonorrhea and chlamydia, Cephalexin for superimposed bacterial cellulitis and Cephazolin, Analgesia Contact and airborne isolation precautions in a room with negative pressure ventilation Improved and discharged 7 Adler, 2022 [10] HICD centers, UK (Case series) 4 males - 3 females - age 30-40 + 40-50 + <2 years - all patients were diagnosed from August 2018 to September 2021 and managed in (HCID) centers in Liverpool, London, and Newcastle, coordinated via a national HCID network A HCW - the rest, not mentioned (4) Nigeria, UK Imported, nosocomial, and household transmission Fever, night sweats, lymphadenopathy, groin swelling, coryzal illness, headache, lesions 1 secondary (the toddler) + 1 tertiary case (the HCW) within the family of the index case Side effects: elevated liver enzymes due to brincidofovir, conjunctivitis, and contact dermatitis from cleaning products Complications: Low mood, emotional lability, ulcerated inguinal lesion with delayed healing, neuralgia, deep tissue abscesses, severe pain, and pruritis PCR Test for Monkeypox West African Only the HCW case (as PEP) due to contact with a secondary case (toddler) Brincidofovir (3 cases) Tecovirimat (1 case) Opiate analgesia (For neuralgia), azithromycin, ophthalmic chloramphenicol Isolation 3 siblings (contacts of infected father) were isolated (post-exposure isolation) All recovered 1 mild relapse 9 Patalon, 2022 [24] Emergency room of a hospital, Israel (Case report) in 30 s - normal BMI - MSM - case 1 HIV positive - hemorrhoids - case 2 had VTE history - both had a history of Condyloma Acuminatum and were administered the HPV vaccine - (2) Europe Sexual contact with infected male partners (MSM) Fever, muscle aches, fatigue, headache, lymphadenopathy, chills, lesions, oral commissure affection, dyschezia, anal pain, pruritis, high anxiety level, dysuria case 1: no close contacts, case 2: not mentioned None PCR Test - None Topical antibiotics, analgesics, antihistamine, and Oxycodone Case 1: isolation Case 2: not mentioned Recovered with no hospitalization 7 Mauldin, 2020 [36] Hospitals in UK, Israel, Singapore, and Nigeria (Case series) Individuals traveling from Nigeria to UK (UK1, UK2) + to Israel (ISR) + to Singapore (SING)+ 1 HCW in UK (UK3) + a case from Bayelsa State (BAY) in Nigeria - 5 males and a female not mentioned for 4 cases and 2 HCWs (4) Nigeria UK1: unknown, UK2: ingestion of bushmeat and contact with monkeypox cases, UK3: a HCW, a secondary case (contact with UK2) ISR: contact with rodent carcasses, SING: ingestion of bush meat, BAY: HCW (occupational transmission) Presence of lesions, fever prodrome UK3 (secondary case) of UK2 (index case) - Case confirmation, sequencing tests West African - - - - 6 Martínez, 2022 [25] Healthcare setting, USA (Case report) Male - 36 years - MSM - on HIV as PEP - (1) UK MSM (one of sexual contact traveled from the UK to the USA) No fever, night sweats, sore throat, enlarged tonsils, lymphadenopathy, lesions, oropharyngeal erythema, satellite lesions - Complications: bacterial superinfection\ superimposed cellulitis (on nipples) Real-time PCR, MPX confirmation by CDC - - IM penicillin G Benzathine, Chlamydia treatment, Empiric gonorrhea therapy with Doxycycline and Ceftriaxone, Amoxicillin/ Clavulanate for superimposed cellulitis Isolation (not in a hospital) Recovered 7 Rao, 2022 [27] Emergency department, USA (Case report) Early middle-aged male - (1) Nigeria Contact with people in a large social gathering in Nigeria Diarrhea, vomiting, fever, cough, fatigue, purulent rash, pustules None - Real-time PCR for orthopoxviruses, species-specific RT- PCR at CDC West African None Tecovirimat Airborne and contact isolation precautions Monitoring of contacts at low\uncertain and intermediate risk of exposure, no persons were at high risk Disinfection of contaminated surfaces in the airport, two cars used by the case, and his home Mask use during the ongoing COVID-19 pandemic limited transmission of MPXV Discharged (after severe disease) 7 Kunasekaran, 2019 [26] Royal Liverpool Hospital, UK (Case report) Case1: Nigerian male resident living at a naval base in Cornwall – Case 2: male resident – case 3: female, 40 years - all in the UK not mentioned for cases 1 and 2- case 3, a nurse (2) Nigeria Unknown for cases 1 and 2 (index cases), case 3 (nurse - secondary case), contact with case 2 "nosocomial infection\ occupational transmission" Rash, headache, swelling of lymph node, back pain, myalgia, fatigue, vesicles pustules None - - - - Not mentioned for cases 1 and 2 Case 3: not specified isolation of case1 and 2, not mentioned Case 3 was isolated at a specialist unit at an infirmary follow up of close contacts after their last contact with cases - 9 Deresinski, 2022 [28] Emergency department, USA (Case report) Male - (1) Nigeria unknown, maybe from urban areas in Nigeria Cough, fever, diarrhea, and vomiting then, a purulent skin eruption None - - West African None Tecovirimat Airborne and contact isolation precautions Monitoring of contacts Disinfection of planes between flights and patient’s homes Mask use during the ongoing COVID-19 pandemic limited transmission of MPXV Discharged 8 Patel, 2022 [29] A regional HCID, UK (Retrospective study) Median age 38 years - males - 196\197 were gay, bisexual, or MSM - 35.9% had HIV- other STIs, gonorrhea for Chlamydia trachomatis, herpes simplex virus, and Treponema pallidum. Inclusion criteria: All confirmed cases between 13 May and 1 July 2022 - (54) Western Europe: Spain, France, Belgium, Germany, and Greece + West Africa Contact with monkeypox cases, travel to Western Europe, travel to West Africa, sexual contact with males (majority of cases) Lesions, fever, rash, pruritis, myalgia, lymphadenopathy, rectal pain, penile swelling - complications: abscesses, urinary retention, superimposed bacterial lower RTI, disseminated lesions severe rectal pain and perforation, proctitis, necrotizing secondary bacterial infection (ex. Streptococcus pyogenes), paraphimosis or phimosis, solitary cutaneous lesion PCR Test for MPXV - - Antibiotics, Metronidazole and Tecovirimat (some cases), Antibiotics for superadded infections, Analgesics, Opioids, Lidocaine gel, oral laxatives, Fentanyl for severe rectal pain, Co-amoxiclav for bacterial infection, Ceftriaxone and Metronidazole for procitis and other drugs Isolated in hospital, and in containment facilities All improved 7 Mileto, 2022 [30] A clinic, Italy (Case report) Italian male living in Portugal - 33 years - HIV infection - fully vaccinated against COVID-19 - (1) Portugal Sexual contact with a casual partner in Lisbon \ Portugal Asthenia, malaise, anorexia, lesions, pharyngodynia, sneezing, fever, lymphadenopathy - None Non-human orthopoxvirus Testing, Real-time PCR for MPXV - - - Isolated in a negative-pressure room in a hospital Isolation after discharge until lesions recovered Mild disease - remained well till discharge 7 Girometti, 2022 [37] Open-access sexual health clinics, UK (Retrospective study) MSM - 4% bisexual- median age is 41 years (IQR 34 –45)- 70% were White, 15% Black or mixed race, 7% Asian, 7% other ethnicities - 48% were born in the UK, 25% had concomitant STIs, 24% had HIV Inclusion criteria: all lab-confirmed cases between May 14 and May 25, 2022 - (25) European countries: Spain, France, and Netherlands MSM 94% and 46% of individuals traveled outside UK (88% of individuals who reported location of travel, reported visits to European countries) Fatigue, lethargy, fever, skin lesion, lymphadenopathy - Complications: pain, localized bacterial cellulitis mainly on the penile site RT-PCR assay with clade-specific PCR - - Ceftriaxone and Doxycycline for bacterial cellulitis, Metronidazole, Tecovirimat (1 case) Analgesics Hospital isolation, Isolation and telephone assessment for patients not hospitalized PPE for HCWs, fit-tested FFP3 respirators No mixing in waiting rooms All clinically improved and discharged 5 Perez Duque, 2022 [31] Healthcare facilities, Portugal (Observational study) MSM - median age 33 years (range: 22–51) - Almost all cases live in Lisbon and Tagus Valley health region - more than 50% had HIV Inclusion criteria: confirmed cases with the earliest symptom onset on 29 April - (4) Europe Contact with a monkeypox confirmed case, travel abroad (ex. UK), sexual contact with men (1 case with only women) most cases had sex with multiple partners, and contact with animals (cats and pigs) Exanthema, lymphadenopathy, fever, Asthenia, headache, genital ulcers, vesicles - None RT-PCR, Sanger sequencing, viral clade identification West African One case was vaccinated - Home isolation Exclusion of work (sick leave) Hospital isolation (3 cases) Self-monitoring of contacts Contact tracing was difficult Contact precautions Hand hygiene PPE Risk communication and social mobilization to reduce transmission 3 Hospitalized cases Discharged (2 of them) No severe cases 7 Yang, 2022 [8] Taiwan Center for Disease Control, Taiwan (Case report) Male - 20 years student (1) Germany - Fever, sore throat, muscle pain, lymph node swelling in groin, rash, atypical skin lesion - - PCR Test, Phylogenetic analysis West African None - Close contacts were quarantined - 7 Jang, 2022 [39] Incheon Medical Center, Korea (Case report) Male - 34 years – bisexual - (1) Germany MSM with suspected monkeypox infection partner Penile ulcer, headache, fever, sore throat, perioral erosive lesion, rash - - RT-PCR for MPXV, gene sequencing - - - Isolated at the airport, then at Incheon Medical Center - 5 Antinori, 2022 [32] Two different hospitals in central Italy, Italy (Case report) Males - in the 30 s - MSM - history of STIs as syphilis - cases 1 and 3 had HIV and received effective ART; cases 2 and 4 were on PrEP - (4) Gran Canary island (in Spain) MSM Skin lesions, lymphadenopathy, fever, asthenia, itchy papules, myalgia - None RT-PCR for MPXV, Sanger sequencing West African Case 3, vaccinated during childhood Case 1: Ciprofloxacin, Acyclovir and Benzylpenicillin case 2: Anti-inflammatory and Antihistaminic medications cases 3 and 4: not mentioned Isolation in hospitals Droplet and contact isolation measures plus filter face piece-2 (FFP2) All were in good clinical condition and recovered spontaneously 8 3.3 Study characteristics Eighteen studies were case report/series design [8.], [10.], [20.], [21.], [22.], [23.], [24.], [25.], [26.], [27.], [28.], [30.], [32.], [33.], [34.], [36.], [38.], [39.], and three were observational studies as cross-sectional and retrospective [29.], [31], [37.]. A total of 316 cases were diagnosed with Monkeypox. Fig. 2 Three hundred and nine patients were adult males; five patients were adult females and two were toddlers. From these patients, one hundred and thirteen cases of them had a history of travel and were infected by imported transmission. Four infected males were in their thirties and human immunodeficiency virus (HIV)-positive [23.], [24.], [25.], [30.]. MSM had been reported in 8 studies [23.], [24.], [25.], [29.], [31], [32.], [37.], [40.]. Key characteristics of included studies are listed in Table1.Fig. 2 : Number of Monkeypox cases registered in different countries with travel history. Fig. 2 3.4 Source and destination of imported infection Nineteen patients mentioned the travel history from Nigeria [10.], [20.], [21.], [22.], [26.], [27.], [28.], [33.], [34.], [35.], [36.]. In two studies, destination was to Israel [21.], [36.], two to Singapore [20.], [36.], three to the USA [27.], [28.], [33.], and six to the UK [10.], [22.], [26.], [34.], [35.], [36.]. Thirty-eight infected cases had travel destinations from Europe to Australia [23], Israel [24], Portugal [31], UK [29], Korea [39], Taiwan [8], and to Italy [32]. Fifty-four cases who traveled from European countries such as; Spain, France, and the Netherlands to the UK were mentioned in one study [37]. A study discussed a case traveling from Germany [8]. A case traveled from the UK to the USA was discussed by one study [25] and another study discussed a case traveling from Portugal to Italy [30]. 3.5 The mode of transmission among imported cases Six studies mentioned the cause of imported infection was contact with an individual with monkeypox symptoms [27.], [29.], [33.], [34.], [35.], [36.]. Two studies discussed cases due to acquired nosocomial infection [10.], [26.]. Ingestion of barbecued bushmeat was the source of infection in three studies [20.], [34.], [36.] and rodent carcasses were the source of infection in the other two studies [21.], [28.]. Eight studies showed the source of infection was sexual contact, especially with male partners [23.], [24.], [25.], [30.], [31], [32.], [37.], [39.]. Fig.: 3, Fig.: 4 Fig.: 3 (a) Number of Monkeypox cases traveled to Italy from different origin 3(b) Number of Monkeypox cases traveled to UK from different origin. Fig.: 3 Fig.: 4 (a) Number of Monkeypox cases traveled to USA from different origin. 4(b) Number of Monkeypox cases traveled to Portugal from different origin. Fig.: 4 3.6 The imported clade of virus and diagnosis Thirteen studies out of the whole 21 studies reported the virus clade [8.], [10.], [20.], [21.], [23.], [27.], [28.], [31], [32.], [33.], [34.], [35.], [36.]. All reported clades of virus were West African Clades. Diagnosis had been confirmed by PCR [8.], [10.], [20.], [22.], [24.], [25.], [27.], [29.], [30.], [31], [32.], [33.], [35.], [37.], [39.], [40.], genome sequencing [20.], [34.], [35.], [36.], [39.], Sanger sequencing [31], [32.], electron microscopy [20.], [21.], and case confirmation [36]. 3.7 Clinical signs and symptoms and secondary transmission Most of the cases had a fever, genital lesion, vesiculopustular rash, headache, lymphadenopathy, night sweats, and chills. Also, gastro-intestinal symptoms such as diarrhea and vomiting were presented in some cases[41]. Five studies mentioned that the cases had oral affection [24.], [25.], [33.], [34.], [39.]. Three studies confirmed the presence of secondary transmission between cases [10.], [35.], [36.]. 3.8 Control measures with contacts Among the 21 included studies, five studies mentioned that vaccination (smallpox\ poxvirus-derived vaccine) of contacts had been offered as post-exposure prophylaxis [20.], [21.], [31], [34.], [35.]. Also, quarantine and surveillance of these contacts (for the longest incubation period = 21 days from last exposure) were done to control the spread of infection [10.], [20.], [21.], [34.], [35.]. Although two studies mentioned that there was no vaccination offered to the contact, these contacts were not at high risk and they had surveillance and close monitoring [27.], [33.]. Other measures of precautions had been taken as airborne isolation precautions in hospital [23.], [27.], [28.], [33.], wearing personal protective equipment (PPE) [20.], [27.], [28.], [31], [37.]. Home isolation and telephone assessment for patients not hospitalized had been reported in two studies [31], [37.]. Active surveillance for high and intermediate risk exposure and passive surveillance for low risk of exposure [34.], [35.]. Quarantine for forward-traced contacts was done to control the spread of infection [20.], [35.], [40.]. 4 Discussion In this review, we aimed to trace the cases of imported monkeypox reported worldwide to identify the source of infection and destination of these cases for early control infection. We included 21 studies; the quality of included studies was good except for six studies. A total of 316 cases were infected with monkeypox; three hundred and nine patients were adult males, five patients were adult females, and two toddlers. One hundred and thirteen cases were infected by imported monkeypox. The cases were confirmed by PCR, genome sequencing, Sanger sequencing, electron microscopy, and clinical signs and symptoms. Fortunately, in this review, all cases reported were of the West African clade, the less severe form. However, country preparedness should be upscaled, to early identify, diagnose and manage imported cases [42]. The advance of globalization, frequent personnel exchanges and close international trade cooperation make it possible for infectious diseases from all over the world to be imported resulting in modification of certain diseases epidemiology [38]. Mobile populations may modify the epidemiology of certain infectious diseases in the world as they can introduce new infections that in the presence of a viable vector could produce outbreaks in the host country or reintroduce previously eradicated infections [43]. Usually, health care workers, in these areas, are not familiar with such conditions. In fact, there are numerous examples of failed control programs, not the least of which is the increased rate of tuberculosis in developed countries, which is concentrated in specific populations such as immigrants and refugees [44]. Imported malaria is claimed to have led to resurgences of the disease in previously eliminated areas such as Zanzibar, as well as in previously eliminated nations such as Greece and Turkmenistan.[45.], [46.], [47.] Other examples are imported vaccine-preventable diseases that are seen as individual cases or small outbreaks among immigrants and other mobile populations. Previously, almost all cases of monkeypox in people outside of Africa were attributed to international travel to countries where the disease was common or to imported animals [48]. Recent evidence of the role of travel in increasing the risk of infection was provided by large case series conducted across 16 countries including 528 cases. Travel was the second reported risk factor (20%) after sex (28%) in acquiring monkeypox [49]. In this review, we found that about one-half of the studies reported imported cases from Nigeria. Africa and Asia were the main origins of imported malaria and other mosquito-borne diseases [50.], [51.]. First, because of the rapid development of international economic exchange, trade, and travel, the number of migrant workers from Africa increased dramatically. Second, climate and sanitary conditions in Africa and Southeast Asia are suitable for mosquitoes and other vector survival. In fact, international travel witnessed a marvelous recovery after lifting measures on international travel to contain coronavirus disease 2019. According to the latest United Nations World Tourism Organization World Tourism Barometer, international tourism saw a strong rebound in the first five months of 2022, with almost 250 million international arrivals recorded. This compares to 77 million arrivals from January to May 2021 and means that the sector has recovered almost half (46%) of pre-pandemic 2019 levels. In Africa and the Middle East, arrivals could reach about 50% to 70% of pre-pandemic levels [52]. However, they have been described as carrying significant infectious disease burdens, determined by geographic origin, ethnicity, health conditions at the departure point, and the migratory route [53.], [54.]. Many of these infections may be asymptomatic for long periods [55]. In the current work, nine studies reported association between sexual contact especially MSM and infection with monkeypox. Indeed, people of any race/ethnicity, gender, gender identity, sexual orientation, or other traits can contract monkeypox through specific behaviors. Homosexual, bisexual, and MSM account for most infections in the current outbreak.[56] So, health message and distribution tactics may need to be tailored to reach these people directly, such as through particular websites, dating applications, or media programs. Messages should be explicit and nonjudgmental, and any sexual activity should not be stigmatized. In this review, the preventive measures implemented by the countries that reported imported cases varied from case isolation, vaccination, decontamination, and active surveillance. This may urge the need for development of effective standardized control plan that should be put in place for countries to prevent further diseases spread. This plan should focus on case finding, contact tracing, laboratory investigation, isolation, immunization, and case management implemented through communicable disease surveillance systems.[57] This system has two main functions: early notification of potentially transmissible diseases, and monitoring. The value of these surveillance systems is their ability to detect an unusual number of transmissible infections (e.g., an outbreak of dengue), generate an alert, and lead to the communication of this outbreak to the public health authorities to take actions to control the main source of infection, and thus prevent new infections. Timely dissemination of surveillance results can improve the planning, implementation, and evaluation of public health practice. For an efficient surveillance system, public and private health physicians need to continually review their efficiency in detecting and treating imported monkeypox. At the same time, the personnel working at different levels of surveillance need to report data quickly and accurately to ensure rapid and effective actions against possible infectious disease outbreaks.[44] Another crucial measure of imported monkey pox is points of entry (POE) screening. The WHO advised that health promotion and risk communication materials be accessible at POE, including information on how to detect signs and symptoms associated with monkeypox, prevent its spread, and seek medical care at the destination if necessary.[57] 4.1 Strengths and limitations To the best of our knowledge, this is the first study to shed light on imported monkeypox. This study will pave the way for future studies that may help in a better understanding of disease epidemiology. In this study, different databases were searched in addition to grey literature. However, published studies are scarce, and most of the included studies were either case reports or case series that would provide weak evidence and hinders external validation of the study findings. Also, some observational studies could include cases which already could be reported by previous case studies. 5 Conclusion Identifying and treating imported monkeypox could result in a benefit both for the individual concerned and for public health. Development of functioning surveillance system of communicable diseases is now, because of globalization, a key function for worldwide health security. This system necessitates ongoing training of front-line health professionals to ensure that monkeypox is properly recognized and diagnosed and all cases are promptly reported, and that all cases are investigated to determine whether the infection was acquired locally or abroad. In addition, implementing effective health communication about monkeypox prevention and control is mandatory to help individuals to make informed decisions to protect their own and their communities' health. Author contribution RMG, the conceptualization of research idea, writing the manuscript, database search; responded to reviewers’ comments. EMH, screening of articles; MAH, screening of articles; AM, writing manuscript, data extraction; SZ, data extraction; EHE, writing manuscript, quality assessment; NF, quality assessment, writing manuscript, SAA, article screening; RE, data extraction; HMH, protocol writing and registration; AHS, manual search; HHAA, database search; NR, database search; HA, search strategy, submission; EE, search strategy; MAH, writing manuscript, EAH, grey literature search, writing manuscript; MH, contributed to the conceptualization of research idea, writing the manuscript, and critically revised the manuscript. All authors gave final approval and agreed to be accountable for all aspects of the work. All authors read and approved the manuscript. All authors declare no conflict of interest. Funding there has been no financial support for this work that could have influenced its outcome. Conflict of interest We have no disclosures. We do not have conflicts of interest associated with this publication. This manuscript is original, has not been previously published, and is not currently under consideration by another journal. Conflicts of interest We do not have conflicts of interest associated with this publication, and there has been no financial support for this work that could have influenced its outcome. This manuscript is original, has not been previously published, and is not currently under consideration by another journal. Appendix A Supplementary material Supplementary material. Supplementary material. Appendix A Supplementary data associated with this article can be found in the online version at doi:10.1016/j.cimid.2022.101923. ==== Refs References 1. Ligon, B.L. Monkeypox: a review of the history and emergence in the Western hemisphere. in Seminars in pediatric infectious diseases. 2004. Elsevier. 2. Durski K.N. Emergence of monkeypox—west and central Africa 1970–2017. Morbidity and mortality weekly report 67 10 2018 306 29543790 3. Kaler J. Monkeypox: A Comprehensive Review of Transmission, Pathogenesis, and Manifestation Cureus 14 2022 7 4. World Health Organization. 2022 Monkeypox Outbreak: Global Trends. 2022 [cited 2022 November 11]; Available from: 〈https://worldhealthorg.shinyapps.io/mpx_global/〉 5. Okyay R.A. Another Epidemic in the Shadow of Covid 19 Pandemic: A Review of Monkeypox proteins 7 2022 10 6. Petersen E. Human monkeypox: epidemiologic and clinical characteristics, diagnosis, and prevention Infectious Disease Clinics 33 4 2019 1027 1043 30981594 7. Bunge E.M. The changing epidemiology of human monkeypox—A potential threat? A systematic review PLOS Neglected Tropical Diseases 16 2 2022 e0010141 8. Yang Z.-S. The first monkeypox virus infection detected in Taiwan-the awareness and preparation International Journal of Infectious Diseases 2022 9. Kumar, S., G. Subramaniam, and K. Karuppanan, Human Monkeypox Outbreak in 2022. Journal of Medical Virology. 10. Adler H. Clinical features and management of human monkeypox: a retrospective observational study in the UK The Lancet Infectious Diseases 2022 11. Azhar H. Novel COVID-19: A Comprehensive Review of Transmission, Manifestation, and Pathogenesis Cureus 12 2020 5 12. Grant R. Nguyen L.-B.L. Breban R. Modelling human-to-human transmission of monkeypox Bulletin of the World Health Organization 98 9 2020 638 33012864 13. Al Awaidy, S.T., et al., Monkeypox Outbreak: More queries posed as cases globally soar. Sultan Qaboos University Medical Journal [SQUMJ], 2021. 14. Moher D. Corrigendum to: Preferred Reporting Items For Systematic Reviews And Meta-Analyses: The PRISMA Statement International Journal of Surgery 2010; 8: 336-341 Int J Surg 8 2010 8 15. Higgins, J.P., et al., Cochrane handbook for systematic reviews of interventions. 2019: John Wiley & Sons. 16. European Centre for Disease Prevention and Control. Epidemiological update: Monkeypox multi-country outbreak. 2022 [cited 2022 August, 30]; Available from: 〈https://www.ecdc.europa.eu/en/news-events/epidemiological-update-monkeypox-multi-country-outbreak〉. 17. Moola S. Chapter 7: Systematic reviews of etiology and risk. Joanna briggs institute reviewer's manual The Joanna Briggs Institute 2017 5 18. National Heart, L. and B. Institute, Study Quality Assessment Tools\| National Heart, Lung, and Blood Institute (NHLBI), 2022 [Accessed November 16, 2022]. Avaiable in: 〈https://www.nhlbi.nih.gov/health-topics/study-quality-assessment-tools〉. 19. Moskalewicz A. Oremus M. No clear choice between Newcastle–Ottawa Scale and Appraisal Tool for Cross-Sectional Studies to assess methodological quality in cross-sectional studies of health-related quality of life and breast cancer Journal of clinical epidemiology 120 2020 94 103 31866469 20. Yong S.E.F. Imported Monkeypox, Singapore Emerging infectious diseases 26 8 2020 1826 32338590 21. Erez N. Diagnosis of imported monkeypox, Israel, 2018 Emerging infectious diseases 25 5 2019 980 30848724 22. Atkinson B. Infection‐competent monkeypox virus contamination identified in domestic settings following an imported case of monkeypox into the UK Environmental Microbiology 2022 23. Hammerschlag Y. Monkeypox infection presenting as genital rash, Australia, May 2022 Eurosurveillance 27 22 2022 2200411 35656835 24. Patalon, T., et al., Monkeypox infection in a developed country: A Case Report. 2022. 25. Martínez J.I. Monkeypox outbreak predominantly affecting men who have sex with men, Madrid, Spain, 26 April to 16 Eurosurveillance, 2022. 27(27) June 2022 2200471 26. Kunasekaran, M., Report of Monkeypox cases in 2018 in the United Kingdom. Global Biosecurity, 2019. 1(1). 27. Rao A.K. Monkeypox in a traveler returning from Nigeria—Dallas, Texas, July 2021 Morbidity and Mortality Weekly Report 71 14 2022 509 35389974 28. Deresinski S. Case A. of Monkeypox in a Returned Traveler Infectious Disease Alert 41 2022 8 29. Patel A. Clinical features and novel presentations of human monkeypox in a central London centre during the 2022 outbreak: descriptive case series bmj 2022 378 30. Mileto D. New challenges in human monkeypox outside Africa: A review and case report from Italy Travel medicine and infectious disease 2022 102386 31 Perez Duque M. Ongoing monkeypox virus outbreak, Portugal, 29 April to 23 May 2022 Euro Surveill 27 22 2022 32. Antinori A. Epidemiological, clinical and virological characteristics of four cases of monkeypox support transmission through sexual contact, Italy, May 2022 Eurosurveillance 27 22 2022 2200421 35656836 33. Costello V. Imported Monkeypox from International Traveler, Maryland, USA, 2021 Emerging Infectious Diseases 28 5 2022 1002 35263559 34. Vaughan A. Two cases of monkeypox imported to the United Kingdom, September 2018 Eurosurveillance 23 38 2018 1800509 30255836 35. Hobson G. , Family cluster of three cases of monkeypox imported from Nigeria to the United Kingdom, May 2021 Eurosurveillance 26 32 2021 2100745 34387184 36. Mauldin M.R. Exportation of monkeypox virus from the African continent. The Journal of infectious diseases 225 8 2022 1367 1376 32880628 37. Girometti N. Demographic and clinical characteristics of confirmed human monkeypox virus cases in individuals attending a sexual health centre in London, UK: an observational analysis The Lancet Infectious Diseases 2022 38. Wang H.-B. An outbreak following importation of wild poliovirus in Xinjiang Uyghur Autonomous Region, China, 2011 BMC Infectious Diseases 15 1 2015 1 9 25567701 39. Jang Y.R. The first case of monkeypox in the Republic of Korea Journal of Korean Medical Science 37 2022 27 40. Vivancos R. Community transmission of monkeypox in the United Kingdom, April to May 2022 Eurosurveillance 27 22 2022 2200422 35656834 41. Cheema A.Y. Monkeypox: a review of clinical features, diagnosis, and treatment Cureus 14 2022 7 42. World Health Organization. Monkeypox 2022. 2022 [cited 2022 August,27]; Available from: 〈https://www.who.int/news-room/fact-sheets/detail/monkeypox〉. 43. Sudre B. Mapping environmental suitability for malaria transmission, Greece Emerging infectious diseases 19 5 2013 784 23697370 44. Cobo F. Imported infectious diseases: The impact in developed countries 2014 Elsevier 45. Danis K. Autochthonous Plasmodium vivax malaria in Greece, 2011 Eurosurveillance 16 42 2011 19993 22027375 46. Organization, W.H., Achieving elimination in Turkmenistan. 2012. 47. Program, Z.M.C., Malaria elimination in Zanzibar: a feasibility assessment. 2009, ZMCP Zanzibar, Tanzania. 48. Center for Diseases Prevention and Control, C. About Monkeypox. 2020 [cited 2022 July, 22]; Available from: 〈https://www.cdc.gov/poxvirus/monkeypox/about/index.html〉. 49. Thornhill J.P. Monkeypox virus infection in humans across 16 countries—April–June 2022 New England Journal of Medicine 2022 50. Wang Y. Epidemiology of imported infectious diseases, China, 2005–2016 Emerging infectious diseases 25 1 2019 33 51. Tian L. Epidemiologic features of the imported cases of main mosquito-borne diseases in Beijing Acta Parasitologica et Medica Entomologica Sinica 23 3 2016 137 142 52. World Tourism Organization. INTERNATIONAL TOURISM CONSOLIDATES STRONG RECOVERY AMIDST GROWING CHALLENGES 2021 [cited 2021; Available from: 〈https://www.unwto.org/taxonomy/term/347〉. 53. Lifson, A.R., et al., Prevalence of tuberculosis, hepatitis B virus, and intestinal parasitic infections among refugees to Minnesota. Public health reports, 2016. 54. Monge-Maillo B. Imported infectious diseases in mobile populations, Spain Emerging infectious diseases 15 11 2009 1745 19891861 55. O’Brien D.P. Illness in returned travelers and immigrants/refugees: the 6–year experience of two Australian infectious diseases units Journal of travel medicine 13 3 2006 145 152 16706945 56. Center for disease control and Prevention. Reducing Stigma in Monkeypox Communication and Community Engagement. 2022 October 18, 2022 [cited 2022 November, 12]; Available from: 〈https://www.cdc.gov/poxvirus/monkeypox/resources/reducing-stigma.html〉. 57. World Health Organization. Multi-country monkeypox outbreak: situation update. 2022 June 27, 2022 [cited 2022 November 12]; Available from: 〈https://www.who.int/emergencies/disease-outbreak-news/item/2022-DON396〉.	36521366	PMC9716240	NO-CC CODE	2022-12-13 23:17:22	no	Comp Immunol Microbiol Infect Dis. 2023 Jan 28; 92:101923	utf-8	Comp Immunol Microbiol Infect Dis	2,022	10.1016/j.cimid.2022.101923	oa_other
==== Front Research in Diagnostic and Interventional Imaging 2772-6525 2772-6525 Published by Elsevier Masson SAS on behalf of Société française de radiologie. S2772-6525(22)00018-7 10.1016/j.redii.2022.100018 100018 Original Article Additional value of chest CT AI-based quantification of lung involvement in predicting death and ICU admission for COVID-19 patients Galzin Eloise a Roche Laurent bxy Vlachomitrou Anna c Nempont Olivier c Carolus Heike d Schmidt-Richberg Alexander d Jin Peng e Rodrigues Pedro e Klinder Tobias d Richard Jean-Christophe fg Tazarourte Karim hi Douplat Marion hi Sigal Alain i Bouscambert-Duchamp Maude ajaa Si-Mohamed Salim Aymeric ag⁎ Gouttard Sylvain a Mansuy Adeline a Talbot François ab Pialat Jean-Baptiste ag Rouvière Olivier aac Milot Laurent aac Cotton François ag Douek Philippe ag Duclos Antoine h Rabilloud Muriel bxy Boussel Loic ag a Department of Radiology, Hospices Civils de Lyon, Lyon, France b Department of Biostatistics, Hospices Civils de Lyon, Lyon F-69003, France c Philips France, 33 rue de Verdun, CS 60 055, Suresnes Cedex 92156, France d Philips Research, Röntgenstrasse 24-26, Hamburg D-22335, Germany e Philips Medical Systems Nederland BV (Philips Healthcare), the Netherlands f Department of Critical Care Medicine, Hôpital De La Croix Rousse, Hospices Civils de Lyon, Lyon, France g Univ Lyon, INSA‐Lyon, Université Claude Bernard Lyon 1, UJM-Saint Etienne, CNRS, Inserm, CREATIS UMR 5220, Lyon U1294, France h Research on Healthcare Performance (RESHAPE), INSERM U1290, Université Claude Bernard Lyon 1, Lyon, France i Emergency department and SAMU 69, Hospices civils de Lyon, France x Université de Lyon, Lyon F-69000, France y Laboratoire de Biométrie et Biologie Evolutive, Université Lyon 1, CNRS, UMR5558, Equipe Biostatistique-Santé, Villeurbanne F-69622, France aa Université de Lyon, Virpath, CIRI, INSERM U1111, CNRS UMR5308, ENS Lyon, Université Claude Bernard Lyon 1, Lyon F-69372, France ab Department of Information Technology, Hospices Civils de Lyon, Lyon, France ac LabTAU INSERM U1032, Lyon, France aj Laboratoire de Virologie, Institut des Agents Infectieux de Lyon, Centre National de Référence des virus respiratoires France Sud, Centre de Biologie et de Pathologie Nord, Hospices Civils de Lyon, Lyon F-69317, France ⁎ Corresponding author at: Address for correspondence: CHU Cardiologique Louis Pradel; Department of Cardiothoracic Radiology; 59 Boulevard Pinel, 69500 Bron. 2 12 2022 12 2022 2 12 2022 4 100018100018 26 6 2022 15 11 2022 © 2022 Published by Elsevier Masson SAS on behalf of Société française de radiologie. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objectives We evaluated the contribution of lung lesion quantification on chest CT using a clinical Artificial Intelligence (AI) software in predicting death and intensive care units (ICU) admission for COVID-19 patients. Methods For 349 patients with positive COVID-19-PCR test that underwent a chest CT scan at admittance or during hospitalization, we applied the AI for lung and lung lesion segmentation to obtain lesion volume (LV), and LV/Total Lung Volume (TLV) ratio. ROC analysis was used to extract the best CT criterion in predicting death and ICU admission. Two prognostic models using multivariate logistic regressions were constructed to predict each outcome and were compared using AUC values. The first model (“Clinical”) was based on patients’ characteristics and clinical symptoms only. The second model (“Clinical+LV/TLV”) included also the best CT criterion. Results LV/TLV ratio demonstrated best performance for both outcomes; AUC of 67.8% (95% CI: 59.5 - 76.1) and 81.1% (95% CI: 75.7 - 86.5) respectively. Regarding death prediction, AUC values were 76.2% (95% CI: 69.9 - 82.6) and 79.9% (95%IC: 74.4 - 85.5) for the “Clinical” and the “Clinical+LV/TLV” models respectively, showing significant performance increase (+ 3.7%; p-value<0.001) when adding LV/TLV ratio. Similarly, for ICU admission prediction, AUC values were 74.9% (IC 95%: 69.2 - 80.6) and 84.8% (IC 95%: 80.4 - 89.2) respectively corresponding to significant performance increase (+ 10%: p-value<0.001). Conclusions Using a clinical AI software to quantify the COVID-19 lung involvement on chest CT, combined with clinical variables, allows better prediction of death and ICU admission. Keywords Prognosis AI COVID-19 Chest-CT ICU Abbreviations AI, Artificial Intelligence CI, Confidence Interval COVID-19, coronavirus disease ICU, Intensive Care Unit ENT, Ear, Nose, Throat PACS, Picture Archiving and Communication System PCR, Polymerase Chain Reaction TLV, Total Lung Volume LV, Lesion Volume ==== Body pmc1 Introduction The COVID-19 pandemic overwhelmed the healthcare system, and especially the intensive care units (ICU), with its magnitude and rapid spread [1], [2], [3]. Prognosis is known to be partially related to early identification of patients at risk in order to rapidly start the appropriate therapy [4], [5], [6], [7]. Chest CT scan has been playing a key role in diagnosis of COVID-19 [8], [9], [10], [11], [12], [13], [14], and was found essential in disease staging, patient prognosis, and treatment efficacy assessment [15], [16], [17], [18], [19], [20], [21]. In addition, multiple studies demonstrated the additional value of artificial intelligence (AI) algorithms in analyzing chest CT examinations in terms of positive diagnosis and lung involvement quantification. Severity scores related to disease progression [22], [23], [24], [25], [26], [27], [28] and prognostic scores based on clinical or biological variables associated to CT data [29], [30], [31], [32], [33], [34], [35], [36] have been introduced based on the outcomes of AI algorithms. However, few studies focus on the real impact of adding CT to simple clinical markers available directly at the emergency department and even fewer use a clinically available software. The latter is of major importance since it allows use of the results within the clinical workflow. In this study, we evaluate the additional value of associating the quantification of the COVID-19 related lung involvement assessed by the software on chest CT to already established clinical risk factors and patient characteristics in predicting mortality and ICU admission. 2 Materials & methods 2.1 Population and study description Based on our local radiologic information system, we retrospectively selected 349 patients hospitalized in one of the public health care facilities of the Hospices Civils Lyon (HCL), Lyon, France from 3rd March to 4th April 2020, presenting a positive COVID RT-PCR test and who underwent a chest CT examination. Pregnant women and children were excluded from the study. For all patients the following clinical characteristics were recorded at admittance: chest pain, dyspnea (rated as "none", "moderate" or "severe"), ENT symptoms or asymptomatic (absence of the three previous symptoms). The following patient characteristics and comorbidities were also recorded: sex, age, obesity (defined as a BMI >30 kg/m2), hypertension, diabetes, chronic respiratory disease. Chest CT examinations have been performed either on arrival at the emergency department or within five days after hospitalization. Data usage policy of our institution in terms of confidentiality, anonymization and security was applied and approval for retrospective analysis of the patients’ data was obtained from the ethical committee. 2.2 CT scan acquisition Various CT models were used. The scanning parameters were as follows: mean tube voltage of 120.4 ± 11.4 kVp (range: 100-140 kVp), mean slice thickness of 2.3 ± 0.8 mm (range: 0.6 - 3 mm). Scan length, mAs and field of view were adapted for each patient and clinical indication, including exploration of the abdomen when required. 92 out of the 349 examinations (26%) were obtained after intravenous administration of iodinated contrast material. All acquisitions were performed at the end of a deep inspiration. 2.3 AI based software for lung and lung lesion segmentation All 349 chest CTs were analyzed using the CT Pulmo Auto Results software in the IntelliSpace Portal 11 platform (Philips Healthcare). This is a fully automatic image analysis application able to identify consolidations and ground-glass opacities (GGO) within the lungs, supporting the management of adult patients with suspected or diagnosed COVID-19 pneumonia. An example of the segmentation with quantitative results is shown in Fig. 1 .Fig. 1 Example of the lung and lung lesion assessment as well as automatic measurements of the “CT Pulmo Auto” AI solution with the quantification of the overall lung volume, overall lesion volume, volume of the left lung, volume of the right lung, volume of ground-glass opacity (GGO) per lung, volume of consolidations per lung. Fig 1 The automatic processing is performed in three cascaded steps as illustrated in Fig. 2 . The left and right lungs are first segmented with exclusion of the main airways including trachea, stem, lobar bronchi, and the main vessels. COVID-19 related lesions are then segmented within the pre-segmented lungs with exclusion of the large bronchi and vessels. Finally, the detected lung lesions are classified voxel-wise as GGO or consolidation. Segmentations and classification steps are achieved using deep learning algorithms [37,38] trained on internal and public datasets [39,40] for which lungs, consolidation and GGO were annotated. The training dataset included both contrast and non-contrast enhanced studies in order for the algorithm to be robust to different protocols.Fig. 2 Overview on the full lesion classification pipeline. Fig 2 To evaluate the performance of the CT Pulmo Auto Results software on the 349 cases of our study, two expert radiologists with more than 20 years of experience performed in consensus a subjective evaluation on 37 randomly selected cases out of 349 patients (approximately 10% of the cohort). Quality of lung and lesion segmentation and lesion classification was rated using five and two levels score respectively described in Table 1 .Table 1 Quality scoring of lung and lung lesion segmentation and lesions classification. Table 1Quality scoring Meaning for segmentation result 5 Perfect No manual correction required. 4 Good Only minor errors that do not affect the measured anatomy and do not have to be corrected. 3 Acceptable Small errors that only slightly affect the measured anatomy. In clinical routine, these errors would not be corrected, though. 2 Bad Significant errors that need correction. 1 Unacceptable Unusable segmentation even with (more) manual correction. Lesion classification Meaning for Segmentation Result Sufficient The results of lesion classification algorithm (GGO and consolidation) are acceptable in the clinically relevant lesions Insufficient The results of lesion classification algorithm (GGO and consolidation) are unacceptable in the clinically relevant lesions 2.4 Data analysis and prognostic score To measure the additional value of the automatically extracted CT lung information in predicting death, the main outcome criterion, we performed a two-phase analysis. To select the best CT criterion, we first assessed the performance of each CT-scan related criterion in predicting death. Once the best CT criterion was identified, we evaluated the benefit on the predictive performance of incorporating this CT criterion to additional clinical factors. During the first phase, data from all individuals of the study population (n=349) were used. ROC curves [41] were built for each of the following CT-scan criteria: GGO volume/ Total Lung Volume ratio (GGOV/TLV), consolidation volume/Total Lung Volume ratio (CV/TLV), lesion volume (GGO + consolidation)/Total Lung Volume ratio (LV/TLV), GGO volume/Lesion Volume ratio (GGOV/LV). AUC values were estimated, along with the associated 95% Confidence Intervals (CI) using the Delong method [42]. The CT-scan criterion with the highest AUC value was selected and used in the second phase of the analysis. For this second phase, only individuals with complete data (i.e., with no missing information; n=346) were analyzed. In step 1, a clinical prognostic model for death prediction was developed. The following clinical factors were selected based on prior knowledge about factors associated with the severity of the COVID disease and on clinicians’ expertise: age, sex, obesity, diabetes, hypertension, chronic respiratory disease, dyspnoea, ENT symptoms, chest pain and asymptomatic. A logistic regression model including all the clinical factors listed previously was built and is referred as “Clinical” model in the next step. In step 2, the previously selected CT-scan criterion was added to the Clinical model and, as recommended by Pepe et al. [43], a likelihood ratio test was performed to assess if the addition of the CT-scan criterion significantly improved the predictive performance (significance threshold set to 0.05). This final model is referred as “Clinical+LV/TLV” model. The predictive performance of the “Clinical” model and “Clinical+LV/TLV” model was quantified using AUC estimates called ‘naïve’ AUC estimates. Then a bootstrap resampling procedure (10000 bootstrap samples, stratified on living status) was applied to obtain optimism-corrected AUC estimates, as recommended by Steyerberg [44]. Finally 95% CIs of the corrected AUC estimates were derived using a two-stage bootstrap sampling procedure [45]. The same analysis strategy was carried out for the secondary outcome of our study corresponding to the admission in an Intensive Care Unit (ICU): the first phase to identify the CT scan criterion and to perform the univariate ROC-curve analyses and all 5 steps of the second phase All analyses were performed using R software version 4.0.2 [46]. ROC analyses and logistic regressions were performed using the pROC package [47] and the mgcv package [48], respectively. 3 Results 3.1 Characteristics of the population The baseline characteristics, CT-scan criteria, and outcomes of the 349 individuals of the study population are reported in Table 2 . Median age of the patients was 71 years and 199 out of the 349 individuals (57%) were male. Ninety-three patients were admitted in ICU (26.6%) and fifty-three patients (15.2%) died in the hospital. Mean volumetric Computed Tomography Dose Index (CTDIvol) was 9.77 +/- 4.65 mGy (range: 2.1 – 29.5 mGy).Table 2 Baseline characteristics, CT-scan criteria, and outcomes. Quantitative variables were described as median (range), and qualitative variables as count (frequency in %). Volumes are measured in mL and volume ratios in percent (%).TLV: Total Lung Volume; LV: Lesions Volume; GGOV: Ground-Glass Opacity Volume; CV: Condensation Volume. Table 2 Non-missing values Description of the study population (N=349) Characteristics - Medical history Age (years) 349 71 (20-99) Male 349 199 (57%) Obesity 347 39 (11.2%) Diabetes 349 73 (20.9%) Hypertension 349 168 (48.1%) Chronic respiratory disease 349 48 (13.8%) Symptoms Dyspnea 348 None 96 (27.6%) Moderate 171 (49.1%) Severe 81 (23.3%) ORL symptom 349 38 (10.9%) Chest pain 349 31 (8.9%) Asymptomatic 349 7 (2%) CT scan criteria : LV / TLV (%) 349 10 (0-65) GGOV / TLV (%) 349 3 (0-41) CV / TLV (%) 349 5 (0-63) GGO/ LV (%) 349 37 (0-100) Outcomes Hospital Death 349 53 (15.2%) Intensive Care admission 349 93 (26.6%) 3.2 Performance of CT-scan criteria and clinical score in predicting death The AUC values of the ROC curves were reported in the upper part of Table 3 and depicted in Fig. 3 (panel A). They ranged from 56.6% (95% CI: 47.7 - 65.6) for the ratio GGO/LV to 67.8% (95% CI: 59.5 - 76.1) for the LV/TLV ratio. The LV/TLV ratio was thus selected for the second phase of the analysis for the final prognosis score.Table 3 Estimation of Area Under the Receiver Operating Characteristic curve (AUC) with 95% Confidence Interval (95% CI), by CT-scan criterion, and for the “Clinical” model and the “Clinical+LV/TLV” model (LV/TLV used as CT-scan measurement). The “Clinical” and “Clinical+LV/TLV” model were obtained from multivariate logistic regressions. Naive AUC: AUC value not corrected for the optimism bias; Corrected AUC: AUC value corrected for the optimism bias. TLV: Total Lung Volume; LV: Lesions Volume; GGOV: Ground-Glass Opacity Volume; CV: Condensation Volume. Table 3 Performance in predicting death AUC in % (95% CI) Association with ICU admission AUC in % (95% CI) CT scan criteria : LV/TLV 67.8 (59.5 - 76.1) 81.1 (75.7 - 86.5) GGOV/TLV 58.5 (48.7 - 68.3) 67.6 (61.0 - 74.3) CV/TLV 63.2 (55.0 - 71.4) 73.2 (66.8 - 79.6) GGOV/LV 56.6 (47.7 - 65.6) 55.3 (48.3 - 62.4) Results from multivariate model - Naive AUC “Clinical” 81.2 (75.7 - 86.8) 78.9 (73.5 - 84.3) “Clinical + LV/TLV” 84.5 (79.3 - 89.7) 87.7 (83.6 - 91.9) Results from multivariate model - Corrected AUC “Clinical” 76.2 (69.9 - 82.6) 74.9 (69.2 - 80.6) “Clinical+ LV/TLV” 79.9 (74.4 - 85.5) 84.8 (80.4 - 89.2) Fig. 3 Receiver Operating Characteristic ROC curve for the prediction of death at hospital. Left panel A: ROC curve by CT-scan criterion. Right panel B: ROC curve of the “Clinical” prognostic model (dashed red line) and of the “Clinical + LV/TLV” model using the ratio Lesions Volume/Total Lung Volume (solid black line). AUC were 76.2 and 79.9 respectively. TLV: Total Lung Volume; LV: Lesion Volume; GGOV: Ground-Glass Opacity Volume; CV: Condensation Volume. Fig 3 The results of the multivariate logistic regressions for the “Clinical” and “Clinical+LV/TLV” models are shown in table 4 . The main predictors of death for the clinical model were age with an Odds-ratio for a 10-year increase estimated at 2.27 (95%CI: 1.70-3.05) and dyspnea, for which the odds-ratios for moderate and severe symptoms (as compared to the absence of dyspnea) were 1.82 (95%CI: 0.75-4.44) and 2.88 (95CI%: 1.09-7.60), respectively. For the “Clinical+LV/TLV” model, the odds-ratio for a 10-year increase in age was estimated at 2.51 (95%CI:1.81-3.48) and age had a high prognostic value. However, the prognostic value of dyspnea was reduced as compared to the “Clinical” model. The odds-ratio for a 10-point increase in the LV/TLV ratio was estimated at 1.73(95% CI:1.35-2.22) and adding the LV/TLV ratio to the clinical factors significantly improved the predictive performance of the model (p-value <0.001). The ROC curves of the “Clinical” model and of the “Clinical+LV/TLV” model are provided in Fig. 3 (panel B). The optimism-corrected values of AUC (Table 3, lower part, left column) were 76.2% (95% CI: 69.9 - 82.6) and 79.9% (95% CI 74.4 - 85.5) for the “Clinical” model and “Clinical+LV/TLV “model, respectively.Table 4 Results from the multivariate logistic regressions to predict death at hospital, expressed as odds-ratio (95% Confidence Interval CI). Analyses were restricted to complete cases (n=346 individuals). “Clinical” model: multivariate model including clinical factors; “Clinical+LV/TLV” model: multivariate model including the Lesions Volume / Total Lung Volume ratio (LV/TLV) in addition to clinical factors. TLV: Total Lung Volume; LV: Lesion Volume. Table 4 “Clinical” model Odds-ratio (95% CI) “Clinical+LV/TLV” model Odds-ratio (95% CI) Intercepta 0.09(0.04-0.22) 0.04(0.02-0.12) Characteristics - Medical history Age for a 10-year increase 2.27(1.70-3.05) 2.51(1.81-3.48) Female 0.74(0.38-1.46) 0.68(0.33-1.40) Obesity 1.73(0.59-5.12) 1.58(0.48-5.15) Diabetes 1.35(0.62-2.92) 1.48(0.65-3.36) Hypertension 0.82(0.41-1.63) 0.90(0.44-1.83) Chronic respiratory disease 0.97(0.38-2.47) 1.45(0.52-3.99) Symptoms Dyspnea None 1 1 Moderate 1.82(0.75-4.44) 1.08(0.42-2.80) Severe 2.88(1.09-7.60) 1.60(0.57-4.52) ORL symptom 0.25(0.05-1.24) 0.26(0.05-1.26) Chest pain 0.68(0.14-3.38) 1.07(0.21-5.49) Asymptomatic 0.00(0.00-Inf) 0.00(0.00-Inf) CT scan criterion LV/TLV for a 10-point increase - 1.73(1.35-2.22) a Values for Intercept corresponded to the odds for male individuals aged 70, without any symptom or comorbidity, and with a ratio LV/TLV equal to 0% (“Clinical+LV/TLV” model only). Thus, the predictive performance significantly increased by + 3.7% when adding the LV/TLV ratio to the clinical factors. 3.3 Association with ICU admission The ROC curves yielded by CT-scan criteria are reported in Fig. 4 (panel A). The association between CT-scan criteria and ICU admission varied greatly: the AUC ranged from 55.3% (IC 95%: 48.3 - 62.4) for the GGO Volume/LV ratio to 81.1% (IC 95%: 75.7 - 86.5) for the LV/TLV ratio (Table 3, right column). The LV/TLV ratio was thus selected for the second phase of the analysis.Fig. 4 Receiver Operating Characteristic ROC curve to evaluate the association with ICU admission. Left panel A: ROC curve by CT-scan criterion. Right panel B: ROC curve of the “Clinical” model (dashed red line) and of the “Clinical+LV/TLV” model using the ratio the ratio Lesions Volume/Total Lung Volume (solid black line). AUC were 74.9 and 84.8 respectively. TLV: Total Lung Volume; LV: Lesions Volume; GGOV: Ground-Glass Opacity Volume; CV: Condensation Volume. Fig 4 The results for the clinical multivariate model for ICU admission are shown in Table 5 (left column). Age was globally associated with ICU admission, but the strength of this association varied across the range of ages. For example, ICU admission was similar for individuals aged 70 years old compared to 60 years old, but much less frequent when comparing ages 90 versus 80 years. The main factors associated with ICU admission were the presence of dyspnea with an odds-ratio for moderate vs. no symptom of 5.05(IC 95%: 2.34-10.90), and gender with an odds-ratio for female versus male of 0.51(IC 95%: 0.29-0.90). For the “Clinical+LV/TLV” model, the odds-ratio for a 10-point increase in the LV/TLV ratio was 2.52(1.94-3.29; p-value < 0.001).Table 5 Results from the multivariate logistic regressions to evaluate the association with admission in an Intensive Care Unit (ICU), expressed as odds-ratio (95% Confidence Interval CI). Analyses were restricted to complete cases (n=346 individuals). “Clinical” model: multivariate model including clinical factors; “Clinical+LV/TLV” model: multivariate model including the Lesions Volume/Total Lung Volume ratio (LV/TLV) in addition to clinical factors. TLV: Total Lung Volume; LV: Lesion Volume. Table 5 “Clinical” model Odds-ratio (95% CI) “Clinical+LV/TLV” model Odds-ratio (95% CI) Intercepta 0.14(0.07-0.32) 0.04(0.02-0.11) Characteristics - Medical history Age: 40y Vs. 50y 1.44(0.86-2.41) 1.36(0.76-2.44) Age: 60y Vs. 70y 1.02(0.61-1.70) 1.04(0.58-1.89) Age: 80y Vs. 90y 0.19(0.07-0.48) 0.11(0.03-0.36) Female 0.51(0.29-0.90) 0.49(0.26-0.95) Obesity 1.01(0.44-2.33) 0.91(0.35-2.39) Diabetes 1.67(0.85-3.30) 2.15(0.98-4.72) Hypertension 0.79(0.43-1.45) 0.87(0.43-1.74) Chronic respiratory disease 0.30(0.11-0.77) 0.44(0.15-1.30) Symptoms Dyspnea None 1 1 Moderate 5.05(2.34-10.90) 2.48(1.07-5.76) Severe 3.39(1.41-8.15) 1.45(0.55-3.79) ORL symptom 0.58(0.23-1.46) 0.48(0.17-1.37) Chest pain 0.79(0.28-2.25) 1.49(0.49-4.55) Asymptomatic 0.00(0.00-Inf) 0.00(0.00-Inf) CT scan criterion LV/TLV for a 10-point increase - 2.52(1.94-3.29) a Values for Intercept corresponded to the odds for male individuals aged 70, without any symptom or comorbidity, and with a measure of the ratio LV/TLV equal to 0% (“Clinical+LV/TLV” model only). The ROC curves of the “Clinical” model only and of the “Clinical+LV/TLV” model are provided in Fig. 4 (panel B). The optimism-corrected values of AUC (Table 3, lower part, right column) were 74.9% (IC 95%: 69.2 - 80.6) and 84.8% (IC 95%: 80.4 - 89.2) respectively. Thus, the predictive performance significantly increased by +10% when adding the LV/TLV ratio to the clinical factors. 3.4 Performances of the CT Pulmo Auto Results software Regarding lung segmentation, the median score value was 5 (25th-75th percentile: [4.5-5]). 36 out of 37 cases received a score of 3 or more, i.e. clinically acceptable values. Similarly, for lesion segmentation, the median score value was 4 (25th-75th percentile: [4], [5]). 35 out of 37 cases were scored larger than 3. Regarding lesions classification, 30 out of 37 cases were rated sufficient. 4 Discussion In this study, we showed that adding automatically extracted quantification information of lung involvement on chest CT data to clinical variables improves death and intensive care admission prediction. Specifically, the results of the multivariate model showed an AUC increase of +3.7% (from 76.2% to 79.9%) in predicting death by adding the CT scan lung involvement quantification to the clinical data. Similarly, for ICU admission, the AUC increased by approximately 10% from 74.9% to 84.8%. These data are consistent with other studies, including the one conducted by Lassau et al [49], demonstrating that adding CT data by using an AI tool to clinical and biological data increased the AUC by 3% as compared to the model including only clinical and biological features. In the same way, the study by Shiri et al [50] showed that using a combined model of clinical and radiomic data from the CT scan increased the AUC from 87% (clinical model) to 95% (combined model) for survival prediction. Finally, in the study by Zhang et al [51], the combination of the two models increased the AUC for estimating progression to critical disease from 84% to 91%. . These differences concerning the added value of CT features to clinical variables across the different studies may be explained by differences in population characteristics. Indeed, in alignment with previous studies, we found a significant association between clinical risk factors and a severe form of the disease with a major effect of age [6,49,52,53] as a predictor of mortality and dyspnea [30,49,53] as a predictor of admission to ICU. Contrary to several other studies, we did not find any statistical association between disease severity and the presence of obesity, diabetes [54,55], hypertension [54], or chronic respiratory disease [54]. This may be due to differences in the studied population across studies. For example, the population in our study is mainly composed of elderly and hospitalized patients, whereas the evaluation population of the Lassau et al study has a large proportion of cancer immunosuppressed patients. This may result in a modification of the contribution of CT data to clinical data, which is positive but limited in the Lassau study, whereas it is clearly significant in our study. Regarding lung lesion classification, on the ROC analysis, we highlight that the ratios GGOV/TLV and CV/TLV and the ratio GGOV/LV is less predictive than the ratio LV/TLV for mortality or ICU admission. This may be related to two different factors. Firstly, the CT attenuation values between GGO and consolidation are overlapping making even the manual classification subjective and extremely challenging. This may also explain the relatively low performance of the algorithm in lesions classification. Secondly, the evolution of the disease from GGO to consolidation is not linear. Indeed, in certain cases, the evolution of the disease is fast presenting a progression of the lesions directly from the early stage (GGO) to the late phase called fibrosis, skipping the consolidation phase [56,57]. The major strength of our study concerns the use of the CT Pulmo Auto Results software, a commercial AI tool released for clinical use, accessible via the Philips IntelliSpace Portal Platform. Having AI tools integrated into the clinical workflow is indeed of major importance, as specified in the CLAIM (Checklist for Artificial Intelligence in Medical Imaging, Mongan et al [58] item 39). This is not the case for several other studies [59], [60], [61], [62] that used AI lung segmentation tools difficult to access in daily hospital practice. Besides the prognostic value of automatic CT analysis that we reported in this study, access to a systematic quantification of lung involvement is also of importance for the overall management of COVID patients and to assess the severity and progression of the disease as stated by the Radiological Society of North America [63], the Fleischner Society [64], French Society of Radiology [65], Chinese National Health Commission [66], and The World Health Organization (WHO) [67] . Our study has some limitations. First, a selection bias could be generated as the recruitment concerns severely ill patients admitted for hospitalization. Second, the retrospective nature of this study and the short period of patient inclusion at the beginning of the pandemic with a higher proportion of scans performed due to limited availability of PCR at some sites should be considered. Finally, our study concerns the original COVID-19 variant and might not be extrapolated to other variants. In conclusion, the combination of clinical factors and the lung involvement ratio measured on regular chest CT examinations using a clinical AI software allows better prediction of death and ICU admission for COVID-19 patients than clinical variables alone. 5 Author statement Conceptualization: LB, MR, AD, Data curation: AG, AV, LB, SG, Formal Analysis: AG, AV, LB, SG, Funding acquisition: LB, SG, Investigation; AG, AV, LB, SG, Methodology: MB-D, MR, LB, Project administration: AM, SG, Resources: EG, LR, AV, ON, HC, A-R, PJ, PR, TK, JR, KT, MD, AS, MB-D, SAS-M, SG, AM, FT, JP, OR, LM, FC, PD, AD, MR, LB, Software: LB, AV, ON, HC, AD-R, PJ, PR, TK, SG, Supervision: LB, MR, Validation: EG, LR, AV, ON, HC, A-R, PJ, PR, TK, JR, KT, MD, AS, MB-D, SAS-M, SG, AM, FT, JP, OR, LM, FC, PD, AD, MR, LB, Visualization: SAS-M, LB, AV, MR, Writing – original draft: EG, LB, AV, MR, Writing – review & editing: EG, LR, AV, ON, HC, A-R, PJ, PR, TK, JR, KT, MD, AS, MB-D, SAS-M, SG, AM, FT, JP, OR, LM, FC, PD, AD, MR, LB 6 Disclosure statement N/A. Declaration of Competing Interest The authors declare the following competing interest: Anna Vlachomitrou, Olivier Nempont, Heike Carolus, Alexander Schmidt-Richberg, Peng Jin, Pedro Rodrigues are Tobias Klinder are employees of Philips Healthcare. Acknowledgments We acknowledge the “Consortium COVID HCL” for its support for this publication. We acknowledge Morgane Bouin, Cécile Rémy, Hayette Djouadi, Hind Behlouli, and Sabine Debeer for their help in data collection. ==== Refs References 1 Zhu N Zhang D Wang W A novel coronavirus from patients with pneumonia in China, 2019 N Engl J Med 382 2020 727 733 10.1056/NEJMoa2001017 31978945 2 Fisher D Heymann D Q&A: the novel coronavirus outbreak causing COVID-19 BMC Med 18 2020 57 10.1186/s12916-020-01533-w 32106852 3 Khan M Adil SF Alkhathlan HZ COVID-19: a global challenge with old history, epidemiology and progress so far Molecules 26 2020 10.3390/molecules26010039 4 Wang D Hu B Hu C Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan, China JAMA 323 2020 1061 1069 10.1001/jama.2020.1585 32031570 5 Huang C Wang Y Li X Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China Lancet 395 2020 497 506 10.1016/S0140-6736(20)30183-5 31986264 6 Yang X Yu Y Xu J Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study Lancet Respir Med 8 2020 475 481 10.1016/S2213-2600(20)30079-5 32105632 7 Phua J Weng L Ling L Intensive care management of coronavirus disease 2019 (COVID-19): challenges and recommendations Lancet Respir Med 8 2020 506 517 10.1016/S2213-2600(20)30161-2 32272080 8 Ai T Yang Z Hou H Correlation of chest CT and RT-PCR testing for coronavirus disease 2019 (COVID-19) in China: a report of 1014 cases Radiology 296 2020 E32 E40 10.1148/radiol.2020200642 32101510 9 Fang Y Zhang H Xie J Sensitivity of chest CT for COVID-19: comparison to RT-PCR Radiology 296 2020 E115 E117 10.1148/radiol.2020200432 32073353 10 Huang P Liu T Huang L Use of chest CT in combination with negative RT-PCR assay for the 2019 novel coronavirus but high clinical suspicion Radiology 295 2020 22 23 10.1148/radiol.2020200330 32049600 11 Tang Y-W Schmitz JE Persing DH Stratton CW Laboratory diagnosis of COVID-19: current issues and challenges J Clin Microbiol 58 2020 10.1128/JCM.00512-20 12 Long C Xu H Shen Q Diagnosis of the coronavirus disease (COVID-19): rRT-PCR or CT? Eur J Radiol 126 2020 108961 10.1016/j.ejrad.2020.108961 13 Xie X Zhong Z Zhao W Chest CT for typical coronavirus disease 2019 (COVID-19) pneumonia: relationship to negative RT-PCR testing Radiology 296 2020 E41 E45 10.1148/radiol.2020200343 32049601 14 Ducray V, VLACHOMITROU AS, Bouscambert-Duchamp M, et al (2020) COVID-19: Role of chest CT in the triage of patients in multiple emergency departments: experience report from a large French university hospital. 10.13140/RG.2.2.31810.40649 15 Bhandari S Rankawat G Bagarhatta M Clinico-radiological evaluation and correlation of CT chest images with progress of disease in COVID-19 patients J Assoc Physicians India 68 2020 34 42 16 Abbasi B Akhavan R Ghamari Khameneh A Evaluation of the relationship between inpatient COVID-19 mortality and chest CT severity score Am J Emerg Med 2020 10.1016/j.ajem.2020.09.056 17 Francone M Iafrate F Masci GM Chest CT score in COVID-19 patients: correlation with disease severity and short-term prognosis Eur Radiol 30 2020 6808 6817 10.1007/s00330-020-07033-y 32623505 18 Li Y Xia L Coronavirus disease 2019 (COVID-19): role of chest CT in diagnosis and management AJR Am J Roentgenol 214 2020 1280 1286 10.2214/AJR.20.22954 32130038 19 Ye Z Zhang Y Wang Y Chest CT manifestations of new coronavirus disease 2019 (COVID-19): a pictorial review Eur Radiol 30 2020 4381 4389 10.1007/s00330-020-06801-0 32193638 20 Feng Z Yu Q Yao S Early prediction of disease progression in COVID-19 pneumonia patients with chest CT and clinical characteristics Nat Commun 11 2020 4968 10.1038/s41467-020-18786-x 33009413 21 Zhao W Zhong Z Xie X Relation between chest CT findings and clinical conditions of coronavirus disease (COVID-19) pneumonia: a multicenter study AJR Am J Roentgenol 214 2020 1072 1077 10.2214/AJR.20.22976 32125873 22 Wang S, Kang B, Ma J, et al (2020) A deep learning algorithm using CT images to screen for corona virus disease (COVID-19). Infectious Diseases (except HIV/AIDS) 23 Meng L Dong D Li L A deep learning prognosis model help alert for COVID-19 patients at high-risk of death: a multi-center study IEEE J Biomed Health Inform 24 2020 3576 3584 10.1109/JBHI.2020.3034296 33108303 24 Wang S Zha Y Li W A fully automatic deep learning system for COVID-19 diagnostic and prognostic analysis Eur Respir J 56 2020 10.1183/13993003.00775-2020 25 Li L Qin L Xu Z Using artificial intelligence to detect COVID-19 and community-acquired pneumonia based on pulmonary CT: evaluation of the diagnostic accuracy Radiology 296 2020 E65 E71 10.1148/radiol.2020200905 32191588 26 Lessmann N Sánchez CI Beenen L Automated assessment of COVID-19 reporting and data system and chest CT severity scores in patients suspected of having COVID-19 using artificial intelligence Radiology 298 2021 E18 E28 10.1148/radiol.2020202439 32729810 27 Kwon YJ (Fred) Toussie D Finkelstein M Combining initial radiographs and clinical variables improves deep learning prognostication in patients with COVID-19 from the emergency department Radiol Artif Intell 3 2021 e200098 10.1148/ryai.2020200098 28 Gieraerts C Dangis A Janssen L Prognostic value and reproducibility of AI-assisted analysis of lung involvement in COVID-19 on low-dose submillisievert chest CT: sample size implications for clinical trials Radiol Cardiothorac Imaging 2 2020 e200441 10.1148/ryct.2020200441 29 Levy TJ Richardson S Coppa K Development and validation of a survival calculator for hospitalized patients with COVID-19 MedRxiv 2020 10.1101/2020.04.22.20075416 30 Liang W Liang H Ou L Development and validation of a clinical risk score to predict the occurrence of critical illness in hospitalized patients with COVID-19 JAMA Intern Med 180 2020 1081 1089 10.1001/jamainternmed.2020.2033 32396163 31 Ji D Zhang D Xu J Prediction for progression risk in patients with COVID-19 pneumonia: the CALL score Clin Infect Dis Off Publ Infect Dis Soc Am 71 2020 1393 1399 10.1093/cid/ciaa414 32 Knight SR Ho A Pius R Risk stratification of patients admitted to hospital with COVID-19 using the ISARIC WHO clinical characterization protocol: development and validation of the 4C mortality score BMJ 370 2020 m3339 10.1136/bmj.m3339 32907855 33 Wynants L Van Calster B Collins GS Prediction models for diagnosis and prognosis of COVID-19 infection: systematic review and critical appraisal BMJ 369 2020 m1328 10.1136/bmj.m1328 32265220 34 Lippi G Plebani M Laboratory abnormalities in patients with COVID-2019 infection Clin Chem Lab Med 58 2020 1131 1134 10.1515/cclm-2020-0198 32119647 35 Chassagnon G Vakalopoulou M Battistella E AI-driven quantification, staging and outcome prediction of COVID-19 pneumonia Med Image Anal 67 2021 101860 10.1016/j.media.2020.101860 36 Bartoli A Fournel J Maurin A Value and prognostic impact of a deep learning segmentation model of COVID-19 lung lesions on low-dose chest CT Res Diagn Interv Imaging 1 2022 100003 10.1016/j.redii.2022.100003 37 Brosch T Saalbach A Foveal fully convolutional nets for multi-organ segmentation Medical imaging 2018: image processing 2018 International Society for Optics and Photonics 105740U 38 Milletari F Navab N Ahmadi S-A V-net: fully convolutional neural networks for volumetric medical image segmentation 2016 Fourth International Conference on 3D Vision (3DV) 2016 565 571 39 Simpson AL, Antonelli M, Bakas S, et al (2019) A large annotated medical image dataset for the development and evaluation of segmentation algorithms. ArXiv190209063 Cs Eess 40 Armato SG McLennan G Bidaut L The Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI): a completed reference database of lung nodules on CT scans: the LIDC/IDRI thoracic CT database of lung nodules Med Phys 38 2011 915 931 10.1118/1.3528204 21452728 41 Pepe MS The Statistical Evaluation of Medical Tests for Classification and Prediction 2003 Oxford University Press Oxford; New York 42 DeLong ER DeLong DM Clarke-Pearson DL Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach Biometrics 44 1988 837 845 3203132 43 Pepe MS Kerr KF Longton G Wang Z Testing for improvement in prediction model performance Stat Med 32 2013 1467 1482 10.1002/sim.5727 23296397 44 Steyerberg EW Validation of prediction models Clinical prediction models 2019 Springer International Publishing Cham 329 344 45 Steyerberg EW Bleeker SE Moll HA Internal and external validation of predictive models: a simulation study of bias and precision in small samples J Clin Epidemiol 56 2003 441 447 10.1016/s0895-4356(03)00047-7 12812818 46 R Core Team A language and environment for statistical computing 2020 R Foundation for Statistical Computing Vienna, Austria https://www.R-project.org/ 47 Robin X Turck N Hainard A pROC: an open-source package for R and S+ to analyze and compare ROC curves BMC Bioinform 12 2011 77 10.1186/1471-2105-12-77 48 Wood SN Generalized additive models: an introduction with R 2nd ed. 2017 Chapman and Hall/CRC 49 Lassau N Ammari S Chouzenoux E Integrating deep learning CT-scan model, biological and clinical variables to predict severity of COVID-19 patients Nat Commun 12 2021 634 10.1038/s41467-020-20657-4 33504775 50 Shiri I Sorouri M Geramifar P Machine learning-based prognostic modeling using clinical data and quantitative radiomic features from chest CT images in COVID-19 patients Comput Biol Med 132 2021 104304 10.1016/j.compbiomed.2021.104304 51 Zhang K Liu X Shen J Clinically applicable AI system for accurate diagnosis, quantitative measurements, and prognosis of COVID-19 pneumonia using computed tomography Cell 181 2020 1423 1433 10.1016/j.cell.2020.04.045 e11 32416069 52 Ruan Q Yang K Wang W Clinical predictors of mortality due to COVID-19 based on an analysis of data of 150 patients from Wuhan, China Intensive Care Med 46 2020 846 848 10.1007/s00134-020-05991-x 32125452 53 Li K Wu J Wu F The clinical and chest CT features associated with severe and critical COVID-19 pneumonia Investig Radiol 55 2020 327 331 10.1097/RLI.0000000000000672 32118615 54 Wang B Li R Lu Z Huang Y Does comorbidity increase the risk of patients with COVID-19: evidence from meta-analysis Aging 12 2020 6049 6057 10.18632/aging.103000 32267833 55 Gupta R Ghosh A Singh AK Misra A Clinical considerations for patients with diabetes in times of COVID-19 epidemic Diabetes Metab Syndr 14 2020 211 212 10.1016/j.dsx.2020.03.002 32172175 56 Li M Lei P Zeng B Coronavirus disease (COVID-19): spectrum of CT findings and temporal progression of the disease Acad Radiol 27 2020 603 608 10.1016/j.acra.2020.03.003 32204987 57 Pan F Ye T Sun P Time course of lung changes at chest CT during recovery from coronavirus disease 2019 (COVID-19) Radiology 295 2020 715 721 10.1148/radiol.2020200370 32053470 58 Mongan J Moy L Kahn CE Checklist for Artificial Intelligence in Medical Imaging (CLAIM): a guide for authors and reviewers Radiol Artif Intell 2 2020 e200029 10.1148/ryai.2020200029 59 Amyar A Modzelewski R Ruan S Multi-task deep learning based CT imaging analysis for COVID-19: classification and segmentation Epidemiology 2020 60 Fan D-P Zhou T Ji G-P Inf-net: automatic COVID-19 lung infection segmentation from CT images IEEE Trans Med Imaging 39 2020 2626 2637 10.1109/TMI.2020.2996645 32730213 61 Laradji I, Rodriguez P, Mañas O, et al (2020) A weakly supervised consistency-based learning method for COVID-19 segmentation in CT images. ArXiv200702180 Cs Eess 62 Shi F Wang J Shi J Review of artificial intelligence techniques in imaging data acquisition, segmentation, and diagnosis for COVID-19 IEEE Rev Biomed Eng 14 2021 4 15 10.1109/RBME.2020.2987975 32305937 63 Simpson S Kay FU Abbara S Radiological Society of North America Expert Consensus Statement on reporting chest CT findings related to COVID-19. Endorsed by the Society of Thoracic Radiology, the American College of Radiology, and RSNA—Secondary Publication J Thorac Imaging 35 2020 219 227 10.1097/RTI.0000000000000524 32324653 64 Rubin GD Ryerson CJ Haramati LB The role of chest imaging in patient management during the COVID-19 pandemic: a Multinational Consensus Statement from the Fleischner Society Radiology 296 2020 172 180 10.1148/radiol.2020201365 32255413 65 Société d’Imagerie Thoracique (SIT) La société d’Imagerie Thoracique propose un Compte-Rendu Structuré de scanner Thoracique pour les patients suspects de COVID-19 SFR E-Bull 2020 https://ebulletin.radiologie.fr/actualites-COVID-19/societe-dimagerie-thoracique-propose-compte-rendu-structure-scanner-thoracique 2020 66 National Health Commission Diagnosis and treatment protocol of COVID-19 pneumonia (Trial Version 8) Infect Dis Inf 33 2020 289 296 10.3969/j.issn.1007-8134.2020.04.001 67 Akl EA Blažić I Yaacoub S Use of chest imaging in the diagnosis and management of COVID-19: a WHO rapid advice guide Radiology 298 2021 E63 E69 10.1148/radiol.2020203173 32729811	0	PMC9716289	NO-CC CODE	2022-12-03 23:20:56	no	2022 Dec 2; 4:100018	utf-8	null	null	null	oa_other
==== Front Am J Med Am J Med The American Journal of Medicine 0002-9343 1555-7162 Elsevier Inc. S0002-9343(22)00822-1 10.1016/j.amjmed.2022.11.002 Clinical Research Study COVID Vaccine Hesitancy and Risk of a Traffic Crash Redelmeier Donald A. MD, FRCPC, MSHSR, FACP abcde⁎ Wang Jonathan MMASc bc Thiruchelvam Deva MSc ac a Evaluative Clinical Sciences, Sunnybrook Research Institute, Toronto, Ont, Canada b Department of Medicine, University of Toronto, Ont, Canada c Institute for Clinical Evaluative Sciences (ICES), Toronto, Ont, Canada d Division of General Internal Medicine e Center for Leading Injury Prevention Practice Education & Research, Sunnybrook Health Sciences Centre, Toronto, Ont, Canada ⁎ Requests for reprints should be addressed to Donald A. Redelmeier, MD, Sunnybrook Health Sciences Centre, G-151 2075 Bayview Ave, Toronto, ON M4N 3M5, Canada. 2 12 2022 2 12 2022 14 10 2022 2 11 2022 © 2022 Elsevier Inc. All rights reserved. 2022 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Coronavirus disease (COVID) vaccine hesitancy is a reflection of psychology that might also contribute to traffic safety. We tested whether COVID vaccination was associated with the risks of a traffic crash. Methods We conducted a population-based longitudinal cohort analysis of adults and determined COVID vaccination status through linkages to individual electronic medical records. Traffic crashes requiring emergency medical care were subsequently identified by multicenter outcome ascertainment of all hospitals in the region over a 1-month follow-up interval (178 separate centers). Results A total of 11,270,763 individuals were included, of whom 16% had not received a COVID vaccine and 84% had received a COVID vaccine. The cohort accounted for 6682 traffic crashes during follow-up. Unvaccinated individuals accounted for 1682 traffic crashes (25%), equal to a 72% increased relative risk compared with those vaccinated (95% confidence interval, 63-82; P < 0.001). The increased traffic risks among unvaccinated individuals extended to diverse subgroups, was similar to the relative risk associated with sleep apnea, and was equal to a 48% increase after adjustment for age, sex, home location, socioeconomic status, and medical diagnoses (95% confidence interval, 40-57; P < 0.001). The increased risks extended across the spectrum of crash severity, appeared similar for Pfizer, Moderna, or other vaccines, and were validated in supplementary analyses of crossover cases, propensity scores, and additional controls. Conclusions These data suggest that COVID vaccine hesitancy is associated with significant increased risks of a traffic crash. An awareness of these risks might help to encourage more COVID vaccination. Keywords COVID-19 Human factors Judgment and reasoning Motor vehicle accident Traffic crash Vaccine hesitancy ==== Body pmc Clinical Significance • Coronavirus disease (COVID) vaccination uptake has stalled despite being safe, effective, and free. • COVID vaccine hesitancy is associated with increased traffic risks. • The risks in unvaccinated adults apply to differing patients and severe events. • The traffic risks are comparable with the risks with sleep apnea. • Physicians counseling patients who decline COVID vaccination could consider safety reminders to mitigate traffic risks. Alt-text: Unlabelled box Introduction Motor vehicle traffic crashes are a common cause of sudden death, brain injury, spinal damage, skeletal fractures, chronic pain, and other disabling conditions. Crash risks occur as a complication of several diseases including alcohol misuse, sleep apnea, and diabetes.1 Crashes also occur in patients with controlled hypertension, prior cancer, or no disease at all.2 The proximate causes of most crashes are human behaviors including speeding, inattention, tailgating, impairment, improper passing, disobeying a signal, failing to yield right-of-way, or other infractions.3 These behaviors might partially reflect health consciousness, safety mindedness, community spirit, or other psychological characteristics that are difficult to measure in a systematic manner.4 , 5 Coronavirus disease (COVID) vaccine hesitancy is defined by the World Health Organization as a delay in acceptance or refusal of vaccination against an important contagious disease despite supply (distribution), access (availability), and awareness (albeit with possible misinformation).6 , 7 Vaccine hesitancy or confidence is not new; for example, the original polio vaccine required multifactorial efforts, including celebrity endorsements (eg, the publicized injection for Elvis Presley in 1956).8, 9, 10 Vaccination preferences mayalso reflect past misadventures (eg, the ill-advised swine-flu vaccine mandate by Gerald Ford in 1976).11 Vaccine hesitancy in regions of wide availability, however, can be contentious due to conflicting values, fallible self-report, cognitive blind spots, or other behavioral issues.12, 13, 14, 15, 16, 17 COVID vaccination is an objective, available, important, authenticated, and timely indicator of human behavior—albeit in a domain separate from motor vehicle traffic crashes. Whether COVID vaccination is associated with increased traffic risks, however, has not been tested and might seem surprising.18 Simple immune activation against a coronavirus, for example, has no direct effect on traffic behavior or the risk of a motor vehicle crash.19 Instead, we theorized that individual adults who tend to resist public health recommendations might also neglect basic road safety guidelines.20, 21, 22, 23 The study question was “Does COVID vaccine hesitancy correlate with the risks of a serious traffic crash?” Methods Study Setting Ontario is the most populous province of Canada, with 14,789,778 residents in 2021.24 The yearly crash risk was 2% for an average adult (minor incidents included), the minimum driving age was 16 years, and novice drivers initially received beginner licenses.3 The COVID vaccine became available in winter 2020, doses were widely delivered to adults by spring 2021, and uptake had plateaued in summer 2021.25 , 26 The 4 vaccines were Pfizer-BioNTech (approved December 9, 2020), Moderna (December 23, 2020), AstraZeneca (February 26, 2021), and Johnson & Johnson (March 5, 2021).27, 28, 29 Vaccination was free to all, supported by popular community outreach, accompanied by public campaigns, and connected to a central registration system (COVAXON).30 Vaccination Status We identified individuals using encrypted identifiers from official government registries.31 We included adults age 18 years or more on July 31, 2021 to ensure that each was eligible for a regular driver's license and a COVID vaccine.32 This population-based approach was fully comprehensive, with the exception of excluding cases marked as invalid, containing faulty identifiers, or missing a birthdate.33, 34, 35 We also excluded those living elsewhere (home address), having no earlier activity (record gap), or who were not alive (death database). COVID vaccination status was based on the COVAXON database, with further details on product (manufacturer), date of first dose (earlier or later), and completeness (1 or 2 doses).36 , 37 The study was registered in advance, approved by the Sunnybrook Research Ethics Board, and conducted using Institute for Clinical Evaluative Sciences safeguards. Additional Characteristics Information on age (years), sex (binary), home location (urban, rural), and socioeconomic status (quintile) was based on demographic databases.38 , 39 Linked health care records were used to identify past diagnoses (International Classification of Diseases, Ninth Revision) and access to care (clinic contacts, emergency visits, hospital admissions) based on the preceding year.40 , 41 We directed specific attention to diseases associated with traffic risks, including alcohol misuse, sleep apnea, diabetes, depression, and dementia.42 , 43 For interest, we also checked for a past diagnosis of hypertension, cancer, and COVID infection. The available databases lacked information on driver skill, functional status, personality traits, traffic infractions, political affiliation, and self-identified ethnicity.44 Traffic Crashes We identified serious traffic crashes during the subsequent month based on emergency care throughout the region (178 individual hospitals).45 This definition reflected incidents sending a patient to an emergency department as a driver, passenger, or pedestrian (codes V00-V69).46 Additional crash characteristics included time (morning, afternoon, night), day (weekend, weekday), ambulance involvement (yes, no), and triage severity score (higher, lower).47 In each case we also determined whether the patient was admitted (yes, no) and final status (dead, alive).45 , 46 , 48, 49, 50 Due to privacy restrictions we did not link to insurance records (financial costs from vehicle damage) or police records (deaths at the scene prior to reaching hospital). Other Outcomes Our study was not a randomized trial and we selected additional outcomes to check for a difference where a difference was anticipated (positive control) and no difference where no difference was anticipated (negative control).51 Specifically, we replicated methods by focusing instead on emergency care for COVID pneumonia as an alternative outcome (positive control). The rationale was that a lack of COVID vaccination, in theory, would be associated with an increased risk of subsequent COVID infection. Similarly, we tested emergency care for uncomplicated constipation (negative control). The rationale was that uncomplicated constipation is a frequent and distinct medical disorder among diverse patients unrelated to COVID vaccination or COVID infection. Statistical Analysis The main analysis evaluated emergency visits for individuals injured in traffic crashes. The primary comparison used the chi-square test to analyze those who had not received a COVID vaccine relative to those who had received a COVID vaccine. Odds ratios were used for relative risk estimates, with no censoring for interval deaths (accounting for deaths at the scene and censoring for interval deaths yielded nearly identical results). Stratified analyses assessed differences according to individual characteristics, with special attention to a diagnosis of alcohol misuse. The analysis was then replicated for patients diagnosed with subsequent COVID pneumonia (positive control) and patients diagnosed with uncomplicated constipation (negative control). Secondary analyses explored further nuances to check the robustness of a potential association between COVID vaccination and traffic crash risks. We used multivariable logistic regression analysis to test the strength of association after accounting for baseline demographic and diagnostic predictors. Prespecified subgroup analyses were used to check for replication according to specific vaccine, recency of first dose, and completeness of vaccination. Similarly, subtype analyses were used to examine whether the association extended across the spectrum of crash severity. In addition, a sensitivity analysis was conducted to account for crossover patients who eventually received a vaccination during the 1-month follow-up interval. Two more supplementary sets of analyses were conducted in a post hoc manner after examining results from the primary analysis. The first analyses tested a propensity score approach as an alternative method to adjust for observed baseline individual differences. Individual patients were pair matched one-to-one based on age (within 5 years), sex (binary), location (binary), socioeconomic status (quintile), and propensity score of specific diagnosis (total = 8). The second analyses tested additional negative controls to validate statistics and check for a further lack of difference in unrelated outcomes. The 4 separate additional emergency outcomes were a fall, a water transportation incident, appendicitis, and conjunctivitis (Appendix, available online). Study reporting followed the Strengthening the Reporting of Observational Studies in Epidemiology guideline (STROBE checklist). Results Overview A total of 11,270,763 adults were identified. Overall, 9,425,473 (84%) had received a COVID vaccine and 1,845,290 (16%) had not received a COVID vaccine at study baseline (July 31, 2021). The 2 groups spanned a diverse range of demographics, with comparable general health care utilization (Table 1 ). The largest relative differences were that those who had not received a COVID vaccine were more likely to be younger, living in a rural area, and below the middle socioeconomic quintile. Those who had not received a vaccine also were more likely to have a diagnosis of alcohol misuse or depression and less likely to have a diagnosis of sleep apnea, diabetes, cancer, or dementia. About 4% had a past COVID diagnosis, with no major imbalance between the 2 groups.Table 1 Baseline Characteristics Table 1 COVID Vaccination Yes No Variable (n = 9,425,473) (n = 1,845,290) Demographic Age (years) 18-39 3,040,343 (32.3%) 938,310 (50.8%) 40-64 3,987,941 (42.3%) 684,712 (37.1%) ≥65 2,397,189 (25.4%) 222,268 (12.0%) Sex Male 4,505,555 (47.8%) 928,543 (50.3%) Female 4,919,918 (52.2%) 916,747 (49.7%) Home Urban 8,464,905 (89.8%) 1,619,385 (87.8%) Rural 960,568 (10.2%) 225,905 (12.2%) Socioeconomic status* Higher 3,956,080 (42.0%) 620,654 (33.6%) Middle 1,913,588 (20.3%) 366,488 (19.9%) Lower 3,555,805 (37.7%) 858,148 (46.5%) Diagnoses† Alcohol misuse‡ Yes 37,118 (0.4%) 13,522 (0.7%) Sleep apnea§ Yes 507,054 (5.4%) 80,454 (4.4%) Diabetes‖ Yes 987,422 (10.5%) 109,995 (6.0%) Depression¶ Yes 1,181,992 (12.5%) 262,915 (14.2%) Dementia⁎⁎ Yes 151,776 (1.6%) 11,522 (0.6%) Hypertension†† Yes 1,069,601 (11.3%) 123,536 (6.7%) Cancer‡‡ Yes 654,151 (6.9%) 75,226 (4.1%) COVID infection§§ Yes 390,928 (4.1%) 64,696 (3.5%) General† Clinic contacts ≥3 6,283,552 (66.7%) 1,116,778 (60.5%) Emergency visit Yes 1,891,240 (20.1%) 475,786 (25.8%) Hospital admission Yes 477,873 (5.1%) 107,175 (5.8%) ⁎ Based on home neighborhood, missing data coded as lower. † Based on previous year. ‡ Code 303. § Code 786. ‖ Code 250. ¶ Code 300. ⁎⁎ Code 290. †† Code 401. ‡‡ Codes 140 to 208. §§ Code 080. Traffic Crashes A total of 6682 individuals required emergency care for a serious traffic crash during the subsequent month of follow-up. This rate averaged over 200 individuals per day and was comparable with population norms for high-income countries. Patients who had not received a COVID vaccine accounted for 1682 crashes (25% of total crashes), equal to an absolute risk of 912 per million. Patients who had received a COVID vaccine accounted for 5000 crashes (75% of total crashes), equal to an absolute risk of 530 per million. The difference corresponded to a relative risk of 1.72 for patients who had not received the COVID vaccine (95% confidence interval, 1.63-1.82; P < 0.001). The risk of a traffic crash was proportional with time for both groups (Figure 1 ).Figure 1 Cumulative incidence plots of absolute risk of a serious traffic crash. X-axis shows days following start of follow-up. Y-axis shows cumulative incidence of events per million individuals. Blue line denotes those vaccinated against coronavirus disease (COVID) and red line denotes those not vaccinated against COVID. Counts in square brackets indicate cumulative total patients in each group with an event at corresponding time. Relative risk ratio based on logistic regression model. Results show substantial incidence of serious traffic crashes that is increased for those who are not vaccinated relative to those who are vaccinated. Figure 1 Consistency for Subgroups The association between a lack of COVID vaccination and increased traffic risks extended to important subgroups. The pattern was apparent for younger and middle-aged adults, men and women, those in urban and rural locations, and across the range of socioeconomic status (Figure 2 ). The smallest relative risk was for adults older than 65 years. The results persisted after stratifying for a diagnosis of alcohol misuse or other specific diagnosis. Stratified analyses based on total clinic contacts, emergency visits, and prior admissions also yielded findings consistent with the primary analysis (Appendix). All subgroups with at least 1000 total crashes showed a significant finding replicating the primary analysis. No subgroup showed a significant opposite association.Figure 2 Forest plot of relative risk of a serious traffic crash in different subgroups. Relative risk compares unvaccinated adults with vaccinated adults for each estimate. In each subgroup, counts show total crashes along with absolute crash risk for those vaccinated and for those not vaccinated (events per million). Circles denote relative risk estimate and horizontal lines denote 95% confidence interval. Null association shown as a relative risk of 1.00 on logarithmic axis. Summary data for total cohort at bottom. Findings show substantial counts, increased relative risk for those unvaccinated, and most subgroups overlapping main analysis. High outlier of unvaccinated patients with dementia potentially attributable to chance. Figure 2 Additional Predictors of Crash Risk The baseline risk of a traffic crash was also related to other individual characteristics (Table 2 ). In accord with past studies, the risk was greater for younger than older adults, more for men than women, and higher for those with lower socioeconomic status. Living in a rural location was not associated with a large difference in risk in either univariable or multivariable analysis. A diagnosis of alcohol misuse was a substantial risk factor, sleep apnea or depression were modest risk factors, and a past diagnosis of COVID infection was an equivocal risk factor. Adjustment for all measured individual characteristics suggested a relative risk of 1.48 for individuals who had not received a COVID vaccine (95% confidence interval, 1.40-1.57; P < 0.001).Table 2 Predictors of Traffic Crash Risk Table 2 Basic Analysis* Adjusted Analysis† Relative Risk Confidence Interval Relative Risk Confidence Interval No COVID vaccination 1.72 1.63-1.82 1.48 1.40-1.57 Younger age (<40 y) 1.50 1.43-1.58 1.40 1.33-1.48 Older age (≥65 y) 0.62 0.57-0.66 0.67 0.62-0.73 Male sex 1.48 1.41-1.56 1.50 1.43-1.57 Rural home 1.03 0.95-1.11 1.06 0.98-1.15 Higher socioeconomic status‡ 0.99 0.93-1.06 1.01 0.94-1.08 Lower socioeconomic status‡ 1.16 1.09-1.24 1.13 1.06-1.21 Alcohol misuse 3.06 2.49-3.77 2.25 1.83-2.78 Sleep apnea 1.21 1.09-1.33 1.32 1.19-1.46 Diabetes 0.76 0.70-0.83 0.98 0.90-1.08 Depression 1.56 1.46-1.66 1.53 1.44-1.63 Dementia 0.39 0.28-0.54 0.59 0.43-0.82 Hypertension 0.63 0.57-0.69 0.82 0.74-0.90 Cancer 0.78 0.70-0.87 1.01 0.90-1.13 COVID infection 1.16 1.03-1.30 1.11 0.99-1.25 ⁎ No adjustments for baseline differences. † Adjusted for other differences through regression model. ‡R eferent is middle socioeconomic status. Secondary Analyses The increased traffic crash risks among those who had not received a COVID vaccine applied across diverse analyses (Table 3 ). The increased risk extended to patients who required ambulance transport, had higher triage severity, and needed hospital admission. The increased risk was accentuated in analyses distinguishing earlier rather than later vaccine timing and distinguishing those with 2 rather than 1 dose. The risk was similar for the Pfizer, Moderna, or other vaccines. As expected, the risk of subsequent COVID pneumonia was increased for those who had not received a COVID vaccine, whereas the risk of constipation was unrelated to the COVID vaccine. Results were further validated in analyses of those eventually vaccinated during follow-up, those matched by propensity scores, and those with additional outcomes (Appendix).Table 3 Secondary Analyses Table 3 Total Events Risk with Vaccine* Risk with No Vaccine* Relative Risk† Confidence Interval Primary analysis 6682 530 912 1.72 1.63-1.82 Crash details Involvement Driver 2856 218 434 1.99 1.83-2.16 Passenger 1189 92 175 1.91 1.68-2.17 Pedestrian 2637 221 303 1.38 1.25-1.51 Time‡ Morning 1490 123 178 1.45 1.28-1.64 Afternoon 3589 292 455 1.56 1.45-1.69 Night 1603 116 278 2.41 2.17-2.67 Day Weekend 2142 172 285 1.66 1.51-1.84 Weekday 4540 359 627 1.75 1.63-1.87 Ambulance transport Yes 2657 207 381 1.84 1.69-2.00 No 4025 323 531 1.64 1.53-1.77 Triage severity§ Higher 1838 137 297 2.17 1.96-2.39 Lower 4844 394 615 1.56 1.46-1.67 Hospital admission Yes 550 42 82 1.97 1.64-2.38 No 6132 489 828 1.69 1.60-1.80 Outcome§ Alive 6674 530 909 1.72 1.62-1.81 Dead 8 0.42 2.17 5.11 1.28-20.43 Vaccine details Timing‖ Earlier¶ 3901 457 912 2.00 1.88-2.13 Later¶ 4463 609 912 1.50 1.41-1.59 Completeness‖ Two doses 5895 505 912 1.81 1.71-1.91 One dose 2469 725 912 1.26 1.16-1.37 Specific type‖ Pfizer 5190 523 912 1.74 1.64-1.85 Moderna 2718 558 912 1.63 1.51-1.77 Other⁎⁎ 2138 528 912 1.73 1.56 tp 1.92 Validation analysis Eventual vaccination 6682 534 939 1.76 1.66-1.86 Propensity matched 2899 661 911 1.38 1.28-1.49 Other outcomes†† COVID pneumonia 5358 303 1354 4.47 4.23-4.74 Constipation 2985 263 272 1.03 0.94-1.14 Fall 28,805 2598 2337 0.90 0.87-0.93 Water craft‡‡ 462 40 44 1.10 0.87-1.40 Appendicitis 1164 101 115 1.14 0.98-1.32 Conjunctivitis 1677 149 150 1.01 0.89-1.15 ⁎ Risk is crash rate per million individuals. † Calculated from logistic regression. ‡ Morning 4 AM to 11:59 AM, afternoon 12 noon to 7:59 PM, night is remainder. § Based on Canadian Triage Severity Score, higher is 1 or 2, lower is remainder. ‖D enotes control group for each sub-analysis based on first dose. ¶ Earlier is prior to May 1, 2021, later is after May 1, 2021. ⁎⁎ AstraZeneca or Johnson & Johnson. †† Supplementary details in accompanying appendix. ‡‡ Transportation incident on waterway not roadway. Discussion We studied millions of adults and found that COVID vaccine hesitancy was associated with significant increased traffic risks. The increased risks included adults with diverse characteristics who spanned the range of socioeconomic status and home locations. The increased risks extended across the spectrum of crash severity, including cases requiring ambulance transport and acute hospitalization. The magnitude of estimated risk was substantial and similar to the relative risk associated with sleep apnea, less than associated with alcohol misuse, and greater than associated with diabetes. A relative risk of this magnitude, furthermore, exceeds the safety gains from modern automobile engineering advances and also imposes risks on other road users.43 , 52 Our research agrees with past studies about psychology contributing to traffic risks.53 , 54 One of the earliest studies evaluated taxi drivers and observed a 7-times greater frequency of personality disorders among those with multiple crashes compared with those with no crashes.55 A study of young drivers identified a near doubling of crash incidents associated with an aggressive personality pattern.56 A psychometric analysis of motorcycle riders found that personal temperament was the largest predictor of crash involvement.57 The weaknesses of past studies include small sample sizes, fallible self-report, cross-sectional designs, low outcome counts, and narrow generalizability.58 , 59 We are aware of no past study testing COVID vaccination and traffic risks. A limitation of our study is that correlation does not mean causality because our data do not explore potential causes of vaccine hesitancy or risky driving.60 One possibility relates to a distrust of government or belief in freedom that contributes to both vaccination preferences and increased traffic risks.61 A different explanation might be misconceptions of everyday risks, faith in natural protection, antipathy toward regulation, chronic poverty, exposure to misinformation, insufficient resources, or other personal beliefs.62 Alternative factors could include political identity, negative past experiences, limited health literacy, or social networks that lead to misgivings around public health guidelines.63 , 64 These subjective unknowns remain topics for more research. Another limitation of our study is the lack of direct data on driving exposure in different groups. A 100% increase in driving distance, however, is unlikely to explain the magnitude of traffic risks observed in this study.65 A difference in driving distance would also not explain why the increased risks extended to pedestrians, why the increased risks were not lower in urban locations, and why the increased risks were not higher on weekends (when discretionary driving is common).66 To be sure, physical factors such as vehicle speed and distance are controlled by the driver and part of the mechanism that ultimately results in a traffic crash. These physical unknowns do not change the importance of our study for estimating prognosis. Our study has other limitations. The analysis does not correct for barriers in access to care or risk compensation that each bias results in the contrary direction.67 The analysis does not include minor crashes that do not lead to emergency care or deaths at the scene prior to reaching the hospital (Appendix).68 The data do not examine the long-term recovery, quality of life, and insurance costs for those who survive initial injuries. Many vehicle factors remain unexplored, including speed, spacing, configuration, location, weather, and distances driven. The study does not test the reliability of COVID vaccination as a proxy for COVID vaccine hesitancy. The available data do not examine long-term trends, test at-fault liability, or assess measurement error that biases results toward the null.58 These uncertainties are further opportunities for future science.10 The current findings can help address 4 common misunderstandings.69 We show the high numbers and the diverse profile of adults who are not vaccinated (Table 1), contrary to claims that COVID vaccine hesitancy is concentrated in men, in poverty, and in rural regions. We validate that vaccination is associated with large reductions in subsequent COVID pneumonia (Table 3), contrary to claims that industry-funded trials are misleading. We document that traffic crashes have continued unabated during the COVID pandemic (Figure 1), contrary to claims that social distancing would lead to fewer traffic fatalities or that one pandemic somehow might replace another. We verify that traffic crashes disproportionately involve those in poverty (Table 2), contrary to claims that traffic safety is unrelated to health disparities. Our findings have direct relevance by highlighting how injury risks have continued despite the COVID pandemic.70 Primary care physicians who wish to help patients avoid becoming traffic statistics, for example, could take the opportunity to stress standard safety reminders such as wearing a seatbelt, obeying speed limits, and never driving drunk.1 , 71 The observed risks are sufficiently large that paramedics, emergency staff, and other first responders should be aware that unvaccinated patients are overrepresented in the aftermath of a traffic crash.72 , 73 The observed risks might also justify changes to driver insurance policies in the future.74 Together, the findings suggest that unvaccinated adults need to be careful indoors with other people and outside with surrounding traffic. Appendix: COVID Vaccine Hesitancy and Risk of a Traffic Crash Table of Contents:§1. Research in Context...........................................................................2 §2. Directed Acyclic Graph.....................................................................4 §3. Description of Patient Flows..............................................................5 §4. Additional Negative Controls............................................................6 §5. Additional Propensity Score Analyses..............................................7 §6. Additional Stratified Analysis...........................................................8 §7. Accounting for Scene Deaths..........................................................10 §8. Accounting for Later Vaccinations..................................................11 §1 Research in Context Evidence prior to this study: We searched MEDLINE, PsychInfo, Scopus, and Google Scholar on December 31, 2021 with no language or date restrictions. The search terms for MEDLINE were (“vaccines” OR “immunization”) AND (“traffic accidents” OR “automobile driving”). The search terms for other databases were adapted as appropriate (details on request). Only 3 surveys examined the association of vaccination with traffic crash risks. One survey (n = 104,594) correlated previous influenza vaccinations with driving safety and detected a significant inverse association (individuals who had not received an influenza vaccination were 15% more likely to report risky driving). Two other survey studies (n = 348 and n = 654) assessed general attitudes toward public health and also found clustering of risks (individuals who reported risk-taking tendencies were 39% less likely to be coronavirus disease (COVID) vaccinated and 41% less likely to follow COVID public health instructions). No studies used validated longitudinal analysis to compare objective vaccination status with actual traffic crash risks. Added value of this study: This is the first population-based longitudinal cohort study to examine an adult's COVID vaccination status and subsequent traffic crash risk. The analysis of over 10 million adults found the risk of a serious traffic crash was significantly higher for adults who had not received a COVID vaccine compared with adults who had received a COVID vaccine. The increased traffic risk associated with COVID vaccine hesitancy persisted in relevant subgroups stratifying for age, sex, home location, socioeconomic status, medical diagnoses, and access to care. The relative risk was similar to the relative risk associated with sleep apnea, less than the risk associated with alcohol misuse, and greater than the risk associated with diabetes. The increased risk was primarily explained by events when driving at night. The increased risk extended across differing degrees of crash severity, was more prominent in analyses based on 2 doses rather than 1 dose, and similar for the Pfizer, Moderna, or other COVID vaccines. Implications of all available evidence: COVID vaccine hesitancy is associated with an increased risk of a traffic crash. A direct effect from immunization is unlikely; instead, diverse psychological factors contribute to vaccine willingness and driving safety (eg, both entail inconveniences advocated by authorities to protect the community). Traffic crashes have continued during the COVID pandemic, implying that physicians have a responsibility to counsel at-risk patients in primary care. In addition, COVID vaccine status might be considered for regions that prioritize road safety, such as those that mandate physicians to warn risky drivers and report to vehicle licensing agencies. Prehospital care providers need to also be aware that unvaccinated adults are overrepresented in the aftermath of a traffic crash, thereby justifying maintaining adherence to COVID precautions at the frenzied crash scene. In addition, the clustering of risks imposed on others might indirectly promote new strategies to promote COVID vaccination. §2 Directed Acyclic Graph Footnote: Directerd Acyclic Graph of possible causal pathways relevant to vaccine hesitancy and traffic risks. The diagram displays measured factors (white), unmeasured ancestors of vaccine hesitancy (green), unmeasured ancestors of traffic risks (blue), and unmeasured ancestors to both vaccine hesitancy and traffic risks (pink). Causal pathways denoted as closed (black lines) or open (magenta lines). Specific causal pathways based on literature review, direct clinical experience (Canada's largest trauma center), and expert consultation (International Traffic Medicine Association). Unlabelled image §3 Description of Patient Flows OHIP = Ontario Health Insurance Plan. Unlabelled image §4 Additional Negative Controls ICD 10 Codes Total Events Positive Control COVID pneumonia U07 5358 Negative Control Constipation K95 2985 Falls W00 to W19 28,805 Appendicitis K35 to K38 1164 Conjunctivitis H10 to H13 1677 Water transportation V90 to V94 462 COVID = coronavirus disease; ICD = International Classification of Diseases. §5 Additional Propensity Score Analyses: General and Stringent The purpose of the first propensity score analysis was to retain a large sample size when matching an unvaccinated individual 1-to-1 with a vaccinated individual and accounting for baseline demographic characteristics and individual diseases.Unlabelled image Analysis of General Matched Cohort Pairs Unvaccinated Control YES CRASH NO CRASH Vaccinated Individual YES CRASH 3 1216 NO CRASH 1677 1,841,591 Total individuals = 3,688,974; total pairs = 1,844,487; total crashes = 2899; odds ratio = 1.38; 95% confidence interval, 1.28-1.44; P-value < 0.001. The purpose of the second propensity score analysis was to be stringent when matching an unvaccinated individual 1-to-1 with a vaccinated individual and excluding cases where any person had a medical diagnosis.Unlabelled image Analysis of Stringent Matched Cohort Pairs Unvaccinated Control YES CRASH NO CRASH Vaccinated individual YES CRASH X 42X NO CRASH 68X 584,41X “X” denotes single digit suppression for privacy regulations; total individuals = 1,171,044; total pairs = 585,522; total crashes = 1111; odds ratio = 1.63; 95% confidence interval, 1.45-1.85; P-value < 0.001. §6 Additional Stratified Analysis Total Events Risk with Vaccine* Risk with No Vaccine* Relative Risk† Confidence Interval Primary analysis 6682 530 912 1.72 1.63-1.82 Health care‡ Clinic contacts ≥3 4620 562 975 1.74 1.62-1.86 Clinic contacts ≤2 2062 468 814 1.74 1.58-1.92 Emergency visit yes 2298 834 1515 1.82 1.67-1.99 Emergency visit no 4384 454 702 1.55 1.44-1.66 Hospital admit yes 363 582 793 1.36 1.07-1.74 Hospital admit no 6319 528 919 1.74 1.65-1.84 ⁎ Risk is crash rate per million individuals. † Calculated from logistic regression. ‡ Based on previous year. §7 Accounting for Scene Deaths The study examined serious traffic crashes based on emergency care throughout the region and thereby did not include deaths at the scene. In turn, we considered extreme assumptions to examine how results might change based on these missing deaths. Specifically, traffic statistics for this setting (602 total deaths in Ontario, 2018) suggested that 50 total deaths might have occurred in our study during follow-up (602/12). Taking into account the 8 deaths that were included, therefore, we estimated potentially 42 total deaths at the scene (50−8). Making an extreme assumption and assigning all these deaths to the vaccinated group yielded minimal changes to final results. In particular, the observed event count increased from a total of 5000 crashes to 5042 crashes, equivalent to an absolute risk of 535 per million (rather than 530 per million). This observed absolute risk was still substantially lower than the observed risk of 912 per million in the unvaccinated group. These results suggested that extreme assumptions about the deaths at the scene make minimal difference to final estimates of relative risk. §8 Accounting for Later Vaccinations The study examined vaccination status based on records on July 31, 2021 and did not include possible later vaccination that might have eventually occurred. In turn, we retrieved information on these subsequent vaccinations and considered extreme assumptions to examine how results might change based on the crossover cases. Specifically, we found 219,740 individuals who were eventually vaccinated from the cohort of 1,8450,290 who had been classified as unvaccinated. These individuals accounted for 155 total traffic crashes during follow-up. Making an extreme assumption and assigning all individuals to the vaccinated group yielded minimal changes to final results. In particular, the observed event count increased from a total of 5000 crashes to 5155 crashes, equivalent to an absolute risk of 534 per million (rather than 530 per million). This observed absolute risk was still substantially lower than the recalculated risk of 939 per million in the unvaccinated group. These results suggested that extreme assumptions about possible eventual vaccination during follow-up make minimal difference to final estimates of relative risk. Acknowledgments We thank Melany Gaetani, Fizza Manzoor, Sheharyar Raza, Eldar Shafir, Richard Thaler, Robert Tibshirani, Chris Yarnell, the Stanford Department of Biomedical Data Science, and the Princeton University Center for Behavioral Science & Public Policy for helpful suggestions on specific points. Funding: This project was supported by a Canada Research Chair in Medical Decision Sciences, the Canadian Institutes of Health Research, the Graduate Diploma in Health Research at the University of Toronto, and the National Sciences & Engineering Research Council of Canada. The views expressed are those of the authors and do not necessarily reflect the Ontario Ministry of Health. Conflicts of Interest: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, and approval of the manuscript. All authors have no financial or personal relationships or affiliations that could influence the decisions and work on this manuscript. Authorship: The lead author (DAR) had full access to all the data in the study, takes responsibility for the integrity of the data, and is accountable for the accuracy of the analysis. Other contributions include: Conceptualization (DAR, JW, DT), data curation (DAR, DT), formal analysis (DAR, JW, DT), funding acquisition (DAR, JW), investigation (DAR, JW, DT), methodology (DAR, JW, DT), project administration (DAR, JW), resources (DAR, JW), software (nil), supervision (DAR), validation (DAR, JW, DT), visualization (DAR, JW, DT), original draft (DAR), and revisions (DAR, JW, DT). The protocol was approved by the Sunnybrook Research Ethics board and conducted using privacy safeguards at the Institute for Clinical Evaluative Sciences. Parts of this material are based on data compiled by CIHI; however, the analyses, conclusions, and statements expressed are those of the authors and not necessarily those of CIHI. Study participants contributed in important ways to this research yet it was not feasible to directly involve individuals in study design or conduct. Members of the public provided feedback on study results and earlier presentations of this material. Data Availability: The study dataset is held securely in coded form at the Institute for Clinical Evaluative Sciences (ICES). While legal data sharing agreements between ICES and data providers (eg, health care organizations and government) prohibit ICES from making the dataset publicly available, access may be granted to those who meet criteria for confidential access, available at www.ices.on.ca/DAS (email [email protected]). The full dataset creation plan and analytic code are available from the authors upon request, understanding that the computer programs might rely upon coding templates or macros that are unique to ICES. ==== Refs References 1 Redelmeier DA Yarnell CJ Thiruchelvam D Tibshirani RJ Physicians' warnings for unfit drivers and the risk of trauma from road crashes N Engl J Med 367 13 2012 1228 1236 23013074 2 National Highway Traffic Safety Administration Traffic Safety Facts 2019 2021 US Department of Transportation Washington, DC Available at: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813141 Accessed June 9, 2022 3 Ontario Ministry of Transportation Ontario Road Safety Annual Report 2018 2021 Government of Ontario Toronto Available at: https://files.ontario.ca/mto-2/mto-orsar-2018-en-2021-11-18.pdf Accessed June 9, 2022 4 Adavikottu A Velaga NR Analysis of factors influencing aggressive driver behavior and crash involvement Traffic Inj Prev 22 suppl 1 2021 S21 S26 34491872 5 Redelmeier DA Najeeb U Etchells EE Understanding patient personality in medical care: five-factor model J Gen Intern Med 36 7 2021 2111 2114 33506393 6 MacDonald NE SAGE Working Group on Vaccine Hesitancy Vaccine hesitancy: definition, scope and determinants Vaccine 33 34 2015 4161 4164 25896383 7 Dror AA Eisenbach N Taiber S Vaccine hesitancy: the next challenge in the fight against COVID-19 Eur J Epidemiol 35 8 2020 775 779 32785815 8 Dudley MZ Privor-Dumm L Dubé È MacDonald NE Words matter: vaccine hesitancy, vaccine demand, vaccine confidence, herd immunity and mandatory vaccination Vaccine 38 4 2020 709 711 31836257 9 Hershfield H, Brody I. How Elvis got Americans to accept the polio vaccine. Scientific American. January 18, 2021. Available at: https://www.scientificamerican.com/article/how-elvis-got-americans-to-accept-the-polio-vaccine/. Accessed June 9, 2022. 10 Sehgal NKR Impact of Vax-a-Million Lottery on COVID-19 vaccination rates in Ohio Am J Med 134 11 2021 1424 1426 34339676 11 Fineberg HV Swine flu of 1976: lessons from the past. An interview with Dr Harvey V Fineberg Bull World Health Organ 87 6 2009 414 415 19565118 12 Nisbett RE Wilson TD Telling more than we can know: verbal reports on mental processes Psychol Rev 84 3 1977 231 259 13 Hall L Johansson P Strandberg T Lifting the veil of morality: choice blindness and attitude reversals on a self-transforming survey PLoS One 7 9 2012 e45457 23029020 14 Krumpal I Determinants of social desirability bias in sensitive surveys: a literature review Qual Quant 47 2013 2025 2047 15 Lambooij MS Harmsen IA Veldwijk J Consistency between stated and revealed preferences: a discrete choice experiment and a behavioural experiment on vaccination behaviour compared BMC Med Res Methodol 15 2015 19 25887890 16 Arena R Pronk NP Laddu D Mapping one million COVID-19 deaths and unhealthy lifestyle behaviors in the United States: recognizing the syndemic pattern and taking action Am J Med 135 11 2022 1288 1295 35820461 17 Murphy J Vallières F Bentall RP Psychological characteristics associated with COVID-19 vaccine hesitancy and resistance in Ireland and the United Kingdom Nat Commun 12 1 2021 29 33397962 18 Redelmeier DA Shafir E Pitfalls of judgment during the COVID-19 pandemic Lancet Public Health 5 6 2020 e306 e308 32334647 19 Forchette L Sebastian W Liu T A comprehensive review of COVID-19 virology, vaccines, variants, and therapeutics Curr Med Sci 41 6 2021 1037 1051 34241776 20 Norris FH Matthews BA Riad JK Characterological, situational, and behavioral risk factors for motor vehicle accidents: a prospective examination Accid Anal Prev 32 4 2000 505 515 10868753 21 Shrank WH Patrick AR Brookhart MA Healthy user and related biases in observational studies of preventive interventions: a primer for physicians J Gen Intern Med 26 5 2011 546 550 21203857 22 Rosenbaum L Escaping Catch-22 – overcoming Covid vaccine hesitancy N Engl J Med 384 14 2021 1367 1371 33577150 23 Fendrich SJ Balachandran M Patel MS Association between behavioral phenotypes and sustained use of smartphones and wearable devices to remotely monitor physical activity Sci Rep 11 1 2021 21501 34728746 24 Government of Ontario. Ontario demographic quarterly: highlights of first quarter. 2021. Available at: https://www.ontario.ca/page/ontario-demographic-quarterly-highlights-first-quarter. Accessed June 9, 2022. 25 Ipsos. Global attitudes on a COVID-19 vaccine. World Economic Forum, 2021. Available at: https://www.ipsos.com/en-ca/global-attitudes-covid-19-vaccine-january-2021. Accessed June 9, 2022. 26 Tang X Gelband H Nagelkerke N COVID-19 vaccination intention during early vaccine rollout in Canada: a nationwide online survey Lancet Reg Health Am 2 2021 100055 27 Government of Canada. Canada's COVID-19 vaccine supply and donation strategy. 2021. Available at: https://www.canada.ca/en/public-health/services/diseases/coronavirus-disease-covid-19/vaccines/supply-donation.html. Accessed June 9, 2022. 28 Government of Ontario. COVID-19 Vaccine Distribution Plan. 2021. Available at: https://files.ontario.ca/moh-covid-19-vaccine-distribution-plan-en-2021-02-19.pdf. Accessed June 9, 2022. 29 Government of Canada. Drug and vaccine authorizations for COVID-19: List of authorized drugs, vaccines and expanded indications. 2022. Available at: https://www.canada.ca/en/health-canada/services/drugs-health-products/covid19-industry/drugs-vaccines-treatments/authorization/list-drugs.html. Accessed June 9, 2022. 30 ICES. ICES COVID-19 Dashboard. 2021. Available at: https://www.ices.on.ca/DAS/AHRQ/COVID-19-Dashboard. Accessed June 9, 2022. 31 Iron K Zagorski BM Sykora K Manuel DG Living and Dying in Ontario: An Opportunity for Improved Health Information 2008 Institute for Clinical Evaluative Sciences Toronto ICES Investigative Report 32 Public Health Ontario. Surveillance report: COVID-19 vaccine uptake in Ontario: December 14, 2020 to May 23, 2022. [Page 5]. Available at: https://www.publichealthontario.ca/-/media/documents/ncov/epi/covid-19-vaccine-uptake-ontario-epi-summary.pdf?la=en. Accessed June 9, 2022. 33 Williams JI Young W A summary of studies on the quality of health care administrative databases in Canada eds Goel V Williams JI Anderson GM Blackstien-Hirsch P Fooks C Naylor CD Patterns of Health Care in Ontario: The ICES Practice Atlas 1996 Canadian Medical Association Ottawa 339 345 2nd ed 34 eds Tu JV Penfold SP McColgan P Laupacis A Access to Health Services in Ontario: ICES Atlas 2006 Institute for Clinical Evaluative Sciences Toronto 35 Juurlink D Preyra C Croxford R Canadian Institute for Health Information Discharge Abstract Database: a validation study 2006 Institute for Clinical Evaluative Sciences Toronto 36 Chung H He S Nasreen S Effectiveness of BNT162b2 and mRNA-1273 covid-19 vaccines against symptomatic SARS-CoV-2 infection and severe covid-19 outcomes in Ontario, Canada: test negative design study BMJ 374 2021 n1943 34417165 37 Leung G Verma A Epidemiological study of COVID-19 fatalities and vaccine uptake: insight from a public health database in Ontario, Canada Cureus. 13 7 2021 e16160 34367770 38 Yarnell CJ Thiruchelvam D Redelmeier DA Risks of serious injury with testosterone treatment Am J Med 134 1 2021 84 94.e6 32926849 39 Redelmeier DA Ng K Thiruchelvam D Shafir E Association of socioeconomic status with medical assistance in dying: a case-control analysis BMJ Open 11 5 2021 e043547 40 Government of Ontario. Medical services – claims history database. Ontario Ministry of Health and Long-Term Care, IntelliHEALTH ONTARIO. Available at: https://intellihealth.moh.gov.on.ca/. Accessed June 9, 2022. 41 Macpherson AK Schull M Manuel D Cernat G Redelmeier DA Laupacis A Injuries in Ontario: ICES Atlas 2005 Institute for Clinical Evaluative Sciences Toronto 42 World Health Organization International Statistical Classification of Diseases and Related Health Problems 2010 World Health Organization Geneva, Switzerland 43 Redelmeier DA Tien HC Medical interventions to reduce motor vehicle collisions CMAJ 186 2 2014 118 124 24324015 44 Sehgal AR Lifetime risk of death from firearm injuries, drug overdoses, and motor vehicle accidents in the United States Am J Med 133 10 2020 1162 1167.e1 32387317 45 Schull MJ Hatcher CM Guttmann A Development of a Consensus on Evidence-Based Quality of Care Indicators for Canadian Emergency Departments. ICES Investigative Report 2010 Institute for Clinical Evaluative Sciences Toronto 46 Canadian Institute for Health Information (CIHI) CIHI Data Quality Study of Emergency Department Visits for 2004-2005: volume II of IV: Main Study Findings 2008 CIHI Ottawa 47 Bullard MJ Musgrave E Warren D Revisions to the Canadian Emergency Department Triage and Acuity Scale (CTAS) guidelines 2016 CJEM 19 S2 2017 S18 S27 28756800 48 Scales DC Guan J Martin CM Redelmeier DA Administrative data accurately identified intensive care unit admissions in Ontario J Clin Epidemiol 59 8 2006 802 807 16828673 49 Wodchis WP Bushmeneva K Nikitovic M McKillop I Guidelines on Person-Level Costing Using Administrative Databases in Ontario 2013 Health System Performance Research Network Toronto, Canada Working Paper Series. Volume 1 50 Wodchis WP Austin PC Henry DA A 3-year study of high-cost users of health care CMAJ 188 3 2016 182 188 26755672 51 Prasad V Jena AB Prespecified falsification end points: can they validate true observational associations? JAMA 309 3 2013 241 242 23321761 52 Evans L Traffic Safety 2004 Science Serving Society Bloomfield Hills, Mich 53 Nabi H Rachid Salmi L Lafont S Chiron M Zins M Lagarde E Attitudes associated with behavioral predictors of serious road traffic crashes: results from the GAZEL cohort Inj Prev 13 1 2007 26 31 17296685 54 Nunn J Erdogan M Green RS The prevalence of alcohol-related trauma recidivism: a systematic review Injury 47 3 2016 551 558 26830122 55 Tillmann WA Hobbs GE The accident-prone automobile driver; a study of the psychiatric and social background Am J Psychiatry 106 5 1949 321 331 18143862 56 Gulliver P Begg D Personality factors as predictors of persistent risky driving behavior and crash involvement among young adults Inj Prev 13 6 2007 376 381 18056312 57 Cheng ASK Ng TCK Lee HC A comparison of the hazard perception ability of accident-involved and accident-free motorcycle riders Accid Anal Prev 43 4 2011 1464 1471 21545879 58 Redelmeier DA Yarnell CJ Lethal misconceptions: interpretation and bias in studies of traffic deaths J Clin Epidemiol 65 5 2012 467 473 22212528 59 Brown TG Ouimet MC Eldeb M Personality,executive control, and neurobiological characteristics associated with different forms of risky driving PLoS One 11 2 2016 e0150227 60 Bullock J Lane JE Shults FL What causes COVID-19 vaccine hesitancy? Ignorance and the lack of bliss in the United Kingdom Humanit Soc Sci Commun 9 1 2022 1 7 61 Kirzinger A, Sparks G, Brodie M. KFF COVID-19 vaccine monitor: in their own words, six months later. Kaiser Family Foundation, July 13, 2021. Available at: https://www.kff.org/coronavirus-covid-19/poll-finding/kff-covid-19-vaccine-monitor-in-their-own-words-six-months-later/. Accessed June 9, 2022. 62 Rosenbaum L No cure without care – soothing science skepticism N Engl J Med 384 15 2021 1462 1465 33606366 63 Johnson NF Velásquez N Restrepo NJ The online competition between pro- and anti-vaccination views Nature 582 7811 2020 230 233 32499650 64 Angus Reid Institute. Dwindling group of unvaccinated cite ‘personal freedom’ and ‘health concerns’ as main reasons for dodging the jab. November 3, 2021. Available at: https://angusreid.org/canada-unvaccinated-freedom-reasons/. Accessed June 9, 2022. 65 Redelmeier DA The fallacy of interpreting deaths and driving distances Med Decis Making 34 8 2014 940 943 24713693 66 Redelmeier DA, Zipursky JS. Pedestrian deaths during the COVID-19 pandemic [online ahead of print]. Am J Lifestyle Med. Published online November 26, 2021. Available at: https://journals.sagepub.com/doi/10.1177/15598276211058378. Accessed November 15, 2022. 67 Trogen B Caplan A Risk compensation and COVID-19 vaccines Ann Intern Med 174 6 2021 858 859 33646837 68 Rojas Castro MY Orriols L Basha Sakr D A web-based prospective cohort study of home, leisure, school and sports injuries in France: a descriptive analysis Inj Epidemiol 8 1 2021 50 34348790 69 Hitti FL Weissman D Debunking mRNA Vaccine Misconceptions-An Overview for Medical Professionals Am J Med 134 6 2021 703 704 33737059 70 National Highway Traffic Safety Administration. Early estimate of motor vehicle traffic fatalities in 2020. Washington DC: United States Department of Transportation. Available at: https://www.nhtsa.gov/press-releases/2020-fatality-data-show-increased-traffic-fatalities-during-pandemic. Accessed June 9, 2022. 71 Redelmeier DA McLellan BA Modern medicine is neglecting road traffic crashes PLoS Med 10 6 2013 e1001463 72 Bledsoe BE Sweeney RJ Berkeley RP Cole KT Forred WJ Johnson LD EMS provider compliance with infection control recommendations is suboptimal Prehosp Emerg Care 18 2 2014 290 294 24401023 73 Prezant DJ Zeig-Owens R Schwartz T Medical leave associated with COVID-19 among emergency medical system responders and firefighters in New York City JAMA Netw Open 3 7 2020 e2016094 74 Weisburd S Identifying moral hazard in car insurance contracts Rev Econ Stat 97 2 2015 301 313	36470796	PMC9716428	NO-CC CODE	2022-12-08 23:16:27	no	Am J Med. 2022 Dec 2; doi: 10.1016/j.amjmed.2022.11.002	utf-8	Am J Med	2,022	10.1016/j.amjmed.2022.11.002	oa_other
==== Front Aust Crit Care Aust Crit Care Australian Critical Care 1036-7314 1036-7314 Australian College of Critical Care Nurses Ltd. Published by Elsevier Ltd. S1036-7314(22)00240-5 10.1016/j.aucc.2022.11.008 Research Paper Fallen angels and forgotten heroes: A descriptive qualitative study exploring the impact of the angel and hero narrative on critical care nurses Stokes-Parish Jessica RN PhD a∗ Barrett David RN PhD b Elliott Rosalind RN PhD cd Massey Deb RN PhD e Rolls Kaye RN DNurs f Credland Nicki RN MMedSci b a Faculty of Health Science and Medicine, Bond University, Gold Coast, Queensland 4229, Australia b Faculty of Health Sciences, University of Hull, Hull, HU6 7RX, United Kingdom c Malcolm Fisher Department of Intensive Care, Royal North Shore Hospital and Centre for Nursing and Midwifery Research, Northern Sydney Local Health District, St Leonards NSW 2065 Australia d Faculty of Health, University of Technology, Ultimo NSW 2007 Australia e Faculty of Health, Southern Cross University, Gold Coast Queensland 4225 Australia f Faculty of Science, Medicine and Health, University of Wollongong, NSW 2522 Australia ∗ Corresponding author. 2 12 2022 2 12 2022 12 6 2022 2 11 2022 12 11 2022 © 2022 Australian College of Critical Care Nurses Ltd. Published by Elsevier Ltd. 2022 Australian College of Critical Care Nurses Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background During the COVID-19 pandemic, the use of the labels ‘heroes’ and ‘angels’ to describe nurses (and especially critical care nurses) became prevalent. While often well intentioned, the use of these labels may not be the most positive image of nurses and the nursing profession. Critical care nurses have not previously been given the opportunity to provide their perceptions of the angel/hero narrative and the impact this may have on their practice and working environments. Objectives The objectives of this study were to explore the perspectives of critical care nurses and discover their perceptions about the angel/hero narrative and its impact on their clinical practice, safe working environments, and professional development during the COVID-19 pandemic. Methods A semistructured qualitative virtual interview study was conducted with critical care nurses from the United Kingdom, Australia, and North America. Digital audio data were transcribed verbatim. Thematic analysis of the transcribed data was performed. The COREQ guidelines were used to report the study. Findings Twenty-three critical care nurses located in the United Kingdom, Australia, and North America participated. Four themes were synthesised: history repeating, gender stereotypes, political pawns, and forgotten heroes. Conclusions Critical care nurses did not perceive the hero and angel labels positively. Participants were concerned about unrealistic expectations, potential safety workplace risks, and poor remuneration related to these narratives. Participants perceived that context and intention were important in the interpretation of these narratives; they spoke with pride about their work and called for improved representations of their role, recognition, and work conditions. Keywords Critical care nurses Covid-19 Hero narrative Nursing Qualitative research Healthcare workers Intensive care nursing Emergency nursing Thematic analysis ==== Body pmc1 Introduction The COVID-19 pandemic has raised the profile and visibility of the nursing profession among the general public and mainstream media. For example, there was a threefold increase in media news items about nurses during March–April 2020 (COVID-19 was declared a pandemic by the World Health Organization in March 2020) compared to the previous year.1 This increase has been particularly evident for the critical care specialty because of the key role critical care nurses played in response to the COVID-19 pandemic.2 One notable characteristic of the increased profile of nurses was the tendency for the collective or individual labels ‘angels’ or ‘heroes’ to be used.2 , 3 These labels are undoubtedly bestowed on nurses with good intentions, in an attempt to acknowledge the courage, care, and commitment that underpin the profession.4 , 5 However, there is increasing evidence that nurses are concerned about these stereotypical labels, the wider narrative, and the underlying message they convey, particularly in relation to the potential impact on the provision of safe work environments.4 , 6 A Danish ethnographic study highlighted nurses' aversion to the angel or hero label, citing the increased personal risk associated with the promotion of a voluntary sacrificial choice which placed the burden on the individual, as opposed to the system.3 Additionally, Canadian researchers identified the emotional burden of nursing during the pandemic and its link to nursing identity. This related in part to a perceived urgency to provide care ‘at all costs’—for example, not donning personal protective equipment (PPE) prior to helping a deteriorating patient.7 Personal risk and a lack of adequate PPE, skilled staff shortages, and low nurse-to-patient ratios (particularly in critical care) have been highlighted as ongoing concerns during the pandemic. Arguably, the media portrayal of nurses and other healthcare workers as angels or heroes exacerbates these potential harms by normalising risk,8 embedding public expectation of ‘sacrifice’,8 and inhibiting discussion of what limits there should be on their duty of care during a pandemic.9 Beyond the impact of the media narrative on risks faced by nurses, there is also a concern regarding the wider perception of nurses as skilled practitioners. Bennett et al highlighted the superficiality of the media discourse, suggesting that the narrative presents nurses as a “… homogenous, selfless, and unquestioning group” (p2754) and fails to take account of the role and characteristics of those who work within the profession.1 This superficial labelling of nurses as heroes and angels arguably detracts from the skills required to undertake the role, moving the perception of the profession from one of skilled practitionership to one of tireless servantship.10 The discourse analysis of Mohammed et al went further to argue that the narrative is a deliberately deployed political tool, which seeks to offer ‘hero status’ as a reward in itself, offered in lieu of tangible improvements in nurses' pay or working conditions.8 Much of the literature related to the angels and heroes narrative takes the form of editorials, commentaries, or discourse analysis. Therefore, we sought to investigate, through the voices of critical care nurses themselves, how this narrative has contributed to nursing identity, the profession, and critical care nursing practice. This important phenomenon has yet to be thoroughly investigated, and its impact on nursing and nurses is poorly understood. In this study, we sought to explore the perspectives of critical care nurses and their feelings about the angel and hero narratives impact on the nursing profession throughout the COVID-19 pandemic. The findings from the study will offer new and important insights into the impact of this narrative on the professional identify of critical care nurses and will provide researchers with potential opportunities to explore this phenomenon more widely. The findings may also provide useful information for educators and faculty in relation to curriculum development that clearly outlines the effect of these labels on nursing and nurses. 2 Methods 2.1 Design and setting We conducted a qualitative descriptive study involving thematic analysis of semistructured interviews. Individual, semistructured virtual interviews were conducted with critical care nurses. Participants were interviewed at a place and time of their choosing, with a single interviewer. Interviews were conducted virtually and conducted by the researchers over a 12-week time scale (October to December 2021). The COREQ guidelines were used to report the study.11 2.2 Participant recruitment and selection A purposive sample of critical care nurses who were exposed to the angel/hero narrative either directly (e.g., hearing patients or relatives using the terms) or indirectly (e.g., watching or reading the narrative in the media) were recruited. All researchers were nurses and therefore ‘insiders’ or part of the social group under examination.12 Recruitment and data collection continued until data saturation was achieved. We defined data saturation in line with Fush and Nest (2015), that is, no new patterns or categories are synthesised from the data.13 An invitation to participate in the study was circulated via social media platforms, through email lists for critical care nursing organisations (e.g., British Association of Critical Care Nurses) and through the researchers’ personal networks. Potential participants who expressed an interest were emailed a formal invitation, a participant information sheet, and a consent form. Potential participants were given a week during which the principal investigator can be contacted to arrange an interview time. 2.3 Data collection Semistructured interviews with critical care nurses were guided by a brief interview schedule (Table 1 ) incorporating four publicly accessible images depicting the hero and angel narrative (see Fig. 1 ) and were undertaken by most members of the research team. Participants were shown images depicting the hero and angel narrative by sharing the researcher’s computer screen and asked about their perceptions of the images. Images were selected by the research team who agreed they were representative of international public sentiment in the middle of 2020. A UK National Health Service (NHS) image was used from their campaign to “send a million hearts to our NHS heroes”,14 while the ‘Thank You to Nurses: Our Front-Line Heroes’ image was a blog from March 2020 to the CipherHealth employees.15 Table 1 Interview questions Table 1What are your views are on the following images and statements? (see Fig. 1 – Images for Study) What do you understand by the terms “angels and heroes”? How do these terms make you feel? How do these terms influence the perception of the nurse/nursing? What elements of the images do you identify with? Fig. 1 Pictures used to prompt discussion during interviews. Fig. 1 The interviews were conducted online on a platform of the participants choosing (i.e., Zoom, Teams, Skype) and audio recorded after the researcher had verbally reconfirmed informed consent. Recordings were transcribed verbatim by a commissioned commercial third party with no links to the research team. The researchers documented their reflections and impressions after each interview in field notes to enable the researchers to consider the data contemporaneously. The interviewer performed preliminary checking and clarification of some comments with the participant at the conclusion of the interview. 2.4 Data analysis Thematic analysis was guided by the methods of Braun and Clarke16 with primary data analysis undertaken independently by two researchers (J.S.-P. and D.M.) after all interviews were completed. Two primary researchers conducted line-by-line coding and analysis, using paper and post-it notes. The first step, ‘immersion’, involved developing familiarity with the interview data by listening repeatedly to the audiotapes and reading the transcription and field notes. The researchers' field notes containing reflections and impressions after each interview contributed to the thematic analysis. Generation of initial codes included identification of raw data which could be categorised in a meaningful way. Moving from creating codes to synthesising broader themes involved examining commonalities with the findings of published studies, then searching for and collating broader themes that captured the essence of participants' responses and patterns of responses. All researchers developed an in-depth understanding of the data through a combination of conducting or listening to interviews and reading transcripts and field notes. The process was iterative, with outcomes from different stages of analysis presented to the whole research team for discussion and revision over several rounds. 2.5 Validity and reliability/rigour We drew on the work of Braun and Clarke (2006) to ensure trustworthiness and rigour in our data collection and analysis.17 Data were digitally recorded and professionally transcribed to ensure accuracy and to enable review by the full team, all of whom have experience in qualitative work. Coding and theme development was conducted and checked by at least two team members as indicated above. Disagreements were resolved through discussion reexamination of the. Regular team meetings were held to discuss and review analytic progress, and all team members were involved in decisions to finalise the themes. The research team was comprised of experienced critical care nurses who now work primarily in the tertiary education sector with three researchers still active clinically (J.-S.P., D.M., and R.E.). Five out of six researchers were female (J.S.-P., D.M., R.E., K.R., and N.C.), four held PhDs (J.S.-P., D.M., D.B., and R.E.), one holds a doctorate (K.R.), and one holds a master’s degree (N.C.). While researchers drew on their experience and background to inform their interpretation of the data, they deliberately and consciously avoided making assumptions and challenged each other during the data analysis stage to ensure a shared understanding of the data and themes. None of the researchers had a working relationship with any of the participants. 2.6 Ethical considerations The study was conducted in accordance with the Good Clinical Practice (GCP) guidelines and the National Statement on the Ethical Conduct of Human Research.18 Local ethical approval was provided by the university in which the lead researchers worked (application number: 15783). All participants provided informed written consent prior to the interview and were advised that they could decline further participation any time and without reason. No participants declined to participate. This information was again provided verbally by the researcher prior to starting the interview. All identifiable data were anonymised. The audio recordings, transcripts, and field notes were saved securely and only accessible by the research team and audio transcriber. Recordings were deleted immediately after the transcripts were cleaned and data analysis was complete. 3 Findings 3.1 Number of participants and duration of audio recordings Interviews were completed with 23 of the 28 critical care nurses who initially agreed to take part, with five not responding to follow-up emails. Participants were critical care nurses located in the United Kingdom (n = 17), Australia (n = 5), and North America (n = 1). The mean duration of digital audio recordings was 19 minutes (range: 11–56 minutes). Four themes were developed: history repeating, gender stereotypes, political pawns, and forgotten heroes. A conceptual representation of how themes are connected is shown in Fig. 2 .Fig. 2 Conceptual representative of the themes. Fig. 2 3.1.1 History repeating Reflecting on the angels and heroes narrative, participants identified that the term ‘angel’ was strongly associated with historical representations of nursing and reflected the religious origins of nursing (e.g., the linkage between nuns and nursing):“I think that the idea that nursing was a vocation that, you know, single women gave up their lives in the same way that they gave up their, you know, they became nuns in that same kind of idea … you’re so self-sacrificing to become a nurse” (Participant 1) Participants rejected the ‘angel’ label, stating that nurses were often the direct opposite of angels. One nurse noted “I don’t think angels would have the dark humour … the dark ways of, of seeing the world. I don’t think that’s necessarily very angelic” (Participant 12). Participants also expressed discomfort with the purity associated with being an angel, “we are a profession … able to stand up for ourselves … not just fall back into this handmaiden, angel, hero, you know, that kind of from Florence Nightingale time” (Participant 5). By being described as angels, participants felt they were viewed as passive, submissive, and disempowered, when in reality, the challenges of critical care require intelligence, critical thinking, and problem-solving skills. Participants recognised that critical care nurses must be able to reflect on practice and respond to changes in patient condition in a proactive, responsive, and autonomous way. In the participants' narratives, knowledge, skill, and professionalism were clearly identified. Again, participants rejected notions of the historical construction of nurses as angels in their professional role and everyday life choices. Several participants highlighted that the sense of immortality and ‘calling’ associated with the angel label created unsafe work practices for critical care nurses—“we can be trodden upon … because it’s a calling” (Participant 19). They felt it led to unrealistic expectations, unsafe nurse-to-patient ratios, and inadequate access to PPE and other important resources such as education. One participant stated:“[they]… make it sound like it’s, like a Mother Teresa type profession, so to me it’s trying to make it a more emotional thing and like an emotional calling or something like that rather than people actually having a paid professional job” (Participant 9). Another voiced: “It really has a significant impact in that, my concern is that policymakers will assume that no matter what comes along, we will just step up to that … without any reward, without any recognition” (Participant 32). When asked “how would you like to see nursing portrayed”, interviewees expressed a desire to see the diversity of nursing portrayed, both in role and cultural background. They also highlighted the need to be recognised for their critical thinking, education, and skills as a nurse. 3.1.2 Gender stereotypes Participants also identified that the angel label was gendered and promoted the passive submissive construction of the nurse as to “caregiver, mother, surrogate, make everything better … we’ll take anything” (Participant 3). The gendering of critical care nurses was also evident through the identification of feminine roles assigned to critical care nursing as ‘women’s work’. The term ‘dirty work’ was used by participants to reflect how they felt their role was tainted (e.g., through contact with bodily fluids), acknowledging that this was an element of their role that nobody else was willing to do. Participants also highlighted that the gendered stereotypes occurred as a result of history. For example, one participant noted, “traditional gender roles … how that informs even if it’s subconsciously” (Participant 3). Participants also noted the predominance of women used to portray nurses—“if you ask the public to physically describe a nurse, nine out of ten times I will guarantee they’d describe a woman” (Participant 5). Participants described the expectation of nurses to be smiling all the time as a uniquely feminine role. One participant observed that men make up most of the senior nursing positions (Participant 5), and another posed, “what if they had a male nurse with angel wings? would that create an impact of us getting more money?” (Participant 8). Other participants raised media representations as a contributor to the perception of gendered stereotypes in critical care nursing. For example, some participants cited the 1970s/1980s British television series ‘Angels’. Others interpreted the portrayal in media as a form of sexism, portraying nurses in Halloween costumes as a “sexy nurse kind of thing …” (Participant 16). This tension was also evident between the genders of the interviewees—the male respondents responded more positively to the hero label but found it hard to relate to being an angel “the sentiment is right … but … as a male nurse, like identifying as an angel is … it’s not easy” (Participant 27) (which the authors note as ironic, given that biblical angels are portrayed as male!19). Participant 15 noted that he was frequently called ‘doctor’, despite being a nurse. Another participant stated that it was easy to relate to the heroes as he had ‘served in the forces’. Participants also explained that there was a difference in their perception of the angel label, considering that context and intention influenced their perception of the use of language. Participant 22 described an experience where a patient regained consciousness and said, “Are you an angel? Am I in heaven?‘ … if it’s a family member or a patient then that’s [ok]” (Participant 22). 3.1.3 Political pawns Participants described a sense of dismissal and betrayal from employers and politicians—“they were saying oh we’re heroes, giving free coffee … when you know, we were being worked to the bone” (Participant 7). While other workers could work from home, nurses were required to attend and work extra hours in challenging environments and oppressive PPE. Participants felt the focus on heroes and angels served to benefit politicians by allowing them to appear as if they were doing something, without actually doing anything. There was an overwhelming sense of tokenism whereby one day nurses were publicly acknowledged with clapping from doorsteps and balconies and the next day were greeted by protests as they arrived for work. One participant reflected that perhaps “the narrative fits for people who can’t imagine the dirty work … so it’s … butterflies and rainbows” (Participant 3). Participants felt this political tokenism had a negative impact on addressing their fundamental needs such as levels of pay and safe working conditions. One participant said, “Our Emergency Department is so unsafe, we are so busy, we are so bed-blocked and it’s so unsafe and it’s well communicated and just not addressed” (Participant 23). 3.1.4 Forgotten heroes While there was strong dislike and rejection of the label ‘angel’, participants were conflicted about their portrayal as a hero. On the one hand, they did ‘suit up with armour’ (PPE) in the uncertainty of the early days of the COVID-19 pandemic, when they were respected by their community for doing jobs that no one else would do, risking their lives by turning up to work in the presence of an unknown pathogen, for example,“it’s just kind of this Hollywood patriarchal fantasy that, that everyone just goes oh yeah, yeah, they’re heroes, they’re angels … we’ve got this awful pandemic but someone’s there to help us and it’s more palatable maybe” (Participant 3) Conversely, critical care nurses quickly felt the respect dwindle to such a point that nurses were villainised. Nurses were abused by families they cared for and disrespected by the community who saw nurses as part of the ‘global agenda’ or who actively thwarted public health efforts. Participants described receiving verbal aggression; one participant said, “I’ve seen comments and all sorts of things...[they] called us all sorts of things” (Participant 8). Some reported that their friends had accused them of killing patients, and others recounted experiences of the public ridiculing them when wearing masks “what are you wearing a mask for, you going to rob the place?” (Participant 19). Beyond this, critical care nurses felt that the angel and hero labels lead to the public having unrealistic expectations. For example, “angels and heroes don’t get sick” (Participant 9) and nurses being obliged to provide care in any circumstances. Over time, participants expressed a sense of being forgotten after all the fanfare of “stunt throwing of public companies … and then all of a sudden it stops because the pandemic’s over … you can’t now book a Thomson holiday and get an NHS discount because, oh we don’t care about you anymore” (Participant 15). Fig. 2 is the conceptual framework showing how the four themes interlink to represent how critical care nurses perceived the angels–heroes narrative during the COVID-19 pandemic. The angel imagery portrays the historical and religious origins of nursing with the outcome of reinforcing nursing as a calling rather than a profession with a delineated scope of practice and scientific basis. This willingness to sacrifice supports the gendered stereotype of nursing as feminine work, which is especially borne out when the public describes a nurse. During the initial months of the pandemic, politicians lauded nurses as heroes for their dedication to their jobs, using them as pawns to deflect from system failures but as time progressed have failed to provide any tangible benefits, such as a safe working environments or adequate compensation. Similarly, public sentiment towards nurses was generally positive during the first part of 2020; however, as crisis fatigue set in, this sentiment changed to one in which nurses were publicly vilified. While participants could understand there were contexts where the labels might be appropriate, there was a shared sense of dismissal, disregard, and diminishing of the identity of and contributions by critical care nurses throughout the pandemic. Despite the seriousness of the potential impact of these themes, the critical care nurses expressed a strong sense of pride when highlighting their expertise and professionalism. For example, “As highly educated professionals who, you know, work collaboratively but also independently from the multidisciplinary team” (Participant 23). 4 Discussion Critical care nurses were interviewed about their unique experiences of the angel and hero narrative during the COVID-19 pandemic. Our analysis of the interviews revealed significant and important concerns associated with the continued and ongoing use of the labels, particularly regarding potential impact on safety in work environments, clinical practice, and professional identify. These concerns have been echoed by other researchers.3 , 4 , 6 Our work is important because it reveals that critical care nurses mostly did not agree with these narratives and view them as detrimental to the societal image of nursing as a profession. In a clinical environment that has been challenged by the pandemic in relation to staffing skill mix, recruitment and retention, issues and increasing burnout,20 it is essential that critical care nurses’ voices are heard and acted on; otherwise, patient safety may be compromised. The narrative of angels and heroes represents critical care nurses as invincible, self-sacrificing, knowingly and willingly working in risk,6 , 21 and undertaking dirty, invisible work.22 It is hardly surprising, therefore, that the participants in our study rejected these historical, gendered, and outdated narratives, as they have done in other studies.3 , 6 , 8 , 23 Critical care nurses strive to deliver high-quality care to patients, but participants suggested that they struggled with feeling insufficiently protected (lack of PPE), suboptimal work conditions, and other obligations outside their jobs. The angel and hero narratives undermine these valid concerns because they normalise unsafe work practices and instill a culture of ‘showing up at all costs’.9 Interestingly, this history is repeated, as nursing experienced similar narratives in the 2003 SARS pandemic. In a media analysis of the portrayal of nurses during this time, researchers found that nurses were portrayed as heroic and self-sacrificing, fighting an invisible enemy and “willing to put themselves in danger and ultimately sacrifice their lives for the sake of others” (p214).24 In a digital world, this effect has no doubt been amplified during the recent pandemic as the use of social media to spread ideas has become ubiquitous.8 The nurses we interviewed felt it was important that their professional capabilities, skills, and knowledge were recognised by the public, by the organisations they worked for, and by politicians. However, the interviews illustrated that angel and/or hero narrative was perceived to be an inaccurate portrayal widely adopted by politicians so that safe staffing ratios could be ignored and dangerous working conditions accepted.3 , 4 , 6 The use of nurses as political pawns was an important finding of our analysis. Likewise, Mohammed et al.8 concluded that the nurse-as-hero discourse was positioned as a reward for nurses and as a convenient means to ignore the lack of support that nurses receive from their places of work, managers, and governments. We acknowledge that nurses have been responsible for extraordinary feats in their response to the pandemic and while they appreciated the early recognition and respect associated with this work, they still preferred tangible and measurable rewards, for example, better working conditions and improved remuneration. At the time of writing, these tangible rewards were still outstanding.25 Importantly, healthcare organisations are now faced with significant recruitment and retention challenges as experienced clinical nurses have increasingly expressed a desire to leave the profession.26 The participants acknowledged that the angel or hero narrative was not always intended by the public and media to be derogatory, quite the opposite in fact. Arguably, the angel or hero label should be used cautiously and with sensitivity so as not to facilitate increased risk of burnout. There has been increasing acknowledgement of the negative impact of the COVID-19 pandemic on critical care nurses’ mental health and wellbeing.27 , 28 Many of the factors contributing to worsening mental health in critical care nurses are related to poor communication, unrealistic role expectations, and unsafe work environments.27 , 28 It is therefore essential that critical care clinical working environments are safe and that nurses are rewarded appropriately for their role. Without this, highly qualified and skilled clinical nurses from the critical care environment may leave.29 , 30 This loss of skilled knowledgeable nurses is likely to adversely affect patient safety and outcomes, as demonstrated by previous work.31 , 32 4.1 Strengths and limitations To ensure the trustworthiness of study outcomes, the research team wove the principles of credibility, dependability, confirmability, reflexivity, and transferability into the study design. Credibility in interpretation was promoted by all researchers immersing themselves in the data. Member checking was undertaken to address credibility and dependability. Multiple quotes are used to illustrate themes and provide confirmability; discussions about the themes ensured consistency, robustness, and transparency. This study sought to explore the perspectives of a group of critical care nurses almost 2 years into the COVID-19 pandemic, a global health emergency that has challenged the way health care is delivered and how critical care nurses practice. It is therefore likely that the labels and narratives used to describe critical care nurses' roles will change and that critical care nurses’ responses to these labels will evolve as COVID-19 moves into the endemic phase. Countries experienced the pandemic differently, and these varying experiences may also have influenced how participants felt about the angel and hero narrative; however, we were unable to explore differences between hospital setting type (i.e., private, public). The transferability of findings may be limited given that most participants were geographically located in two countries: Australia and the United Kingdom. However, the findings do offer important insights into the perceptions of critical care nurses regarding the widespread use of the hero or angel narrative. 5 Conclusion Critical care nurses perceived the angel and hero narratives as damaging to workplace conditions, remuneration, and professional expectations. They felt these narratives perpetuated stereotypes of nurses as being self-sacrificing and ‘superhuman’. However, nurses perceived that context and intention were important in their interpretation of the meaning of these narratives, with important distinctions between patients and politicians. They spoke with pride about their expertise and professionalism and called for improved representations of their role, recognition, and work conditions. Funding This work was supported by a grant awarded by the British Association of Critical Care Nurses. CRediT authorship contribution statement Nicki Credland: conceptualisation, methodology, data acquisition, funding acquisition, investigation, validation, synthesis, formal analysis, writing- original draft preparation, writing-reviewing and editing, visualisation; Jessica Stokes-Parish: conceptualisation, methodology, data acquisition, investigation, validation, synthesis, formal analysis, writing-original draft, reviewing and editing, visualisation; David Barrett: conceptualisation, methodology, data acquisition, funding acquisition, investigation, validation, formal analysis, writing-reviewing and editing, visualisation; Rosalind Elliott: conceptualisation, methodology, data acquisition, writing-reviewing and editing, visualisation; Deb Massey: conceptualisation, methodology, data acquisition, synthesis, formal analysis, writing- original draft preparation, reviewing and editing, visualisation Kaye Rolls: conceptualisation, methodology, data acquisition, writing-reviewing and editing, visualisation Conflict of interest Dr Elliott is an associate editor of Australian Critical Care. Dr Rolls and Associate Professor Deb Massey are editorial board members of Australian Critical Care. This manuscript was handled independently during the review process overseen by the Editor-in-Chief. Acknowledgements We wish to thank all the participants for taking the time to be interviewed.support ==== Refs References 1 Bennett C.L. James A.H. Kelly D. Beyond tropes: towards a new image of nursing in the wake of COVID-19 J Clin Nurs 29 15–16 2020 2753 2755 32445608 2 Bagnasco A. Catania G. Gallagher A. Morley G. Media representations of nurses in the pandemic: just doing our job? Nurs Ethics 27 4 2020 901 905 32495717 3 Halberg N. Jensen P.S. Larsen T.S. We are not heroes—the flipside of the hero narrative amidst the COVID19-pandemic: a Danish hospital ethnography J Adv Nurs 77 5 2021 2429 2436 33616210 4 Stokes-Parish J. Elliott R. Rolls K. Massey D. Angels and heroes: the unintended consequence of the hero narrative J Nurs Scholarsh : an official publication of Sigma Theta Tau International Honor Society of Nursing 52 5 Sep 2020 462 466 5 Stokes-Parish J. Massey D. Rolls K. Elliott R. The angels and heroes of health care: justified and appropriate, or harmful and destructive? J Hosp Med 17 10 2022 847 849 35938210 6 McAllister M. Lee Brien D. Dean S. The problem with the superhero narrative during COVID-19 Contemp Nurse 56 3 Apr - Jun 2020 199 203 32969777 7 Lapum J. Nguyen M. Fredericks S. Lai S. McShane J. Goodbye ... Through a glass door": emotional experiences of working in COVID-19 acute care hospital environments Can J Nurs Res. Mar 53 1 2021 5 15 8 Mohammed S. Peter E. Killackey T. Maciver J. The "nurse as hero" discourse in the COVID-19 pandemic: a poststructural discourse analysis Int J Nurs Stud. May 117 2021 103887 9 Cox C.L. ‘Healthcare Heroes’: problems with media focus on heroism from healthcare workers during the COVID-19 pandemic J Med Ethics 46 8 2020 510 513 32546658 10 Chipps J. Jarvis M.A. Brysiewicz P. Heroes and angels: ED nurses' ongoing fight for meaningful recognition as professionals on the frontline of the pandemic Int Emerg Nurs. Nov 59 2021 101080 11 Tong A. Sainsbury P. Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups Int J Qual Health Care 19 6 2007 349 357 17872937 12 Bonner A. Tolhurst G. Insider-outsider perspectives of participant observation Nurse Res 9 4 2002 7 19 13 Fusch P. Ness L. Are we there yet? Data saturation in qualitative research Qual Rep 20 9 2015 1408 1416 14 Service NH. Thanks a million NHS. Available at: https://thanksamillionnhs.co.uk/. 15 Health C. A thank you letter to nurses our frontline heroes. Available at: https://cipherhealth.com/blog/a-thank-you-letter-to-nurses-our-front-line-heroes/. 16 Braun V. Clarke V. Conceptual and design thinking for thematic analysis 2021 Qualitative psychology Washington, DC) 17 Braun V. Clarke V. Using thematic analysis in psychology Qual Res Psychol 3 2 2006/01/01 2006 77 101 18 Council NHaMR The national statement on the ethical Conduct in Human research [1999 & 2007, updated 2018] 2018 19 Keathley JH. Angelology: the doctrine of angels. Available at: https://bible.org/article/angelology-doctrine-angels. 20 Endacott R. Pearce S. Rae P. Richardson A. Bench S. Pattison N. How COVID-19 has affected staffing models in intensive care: a qualitative study examining alternative staffing models (SEISMIC) J Adv Nurs. Apr 78 4 2022 1075 1088 21 Girvin J. Jackson D. Hutchinson M. Contemporary public perceptions of nursing: a systematic review and narrative synthesis of the international research evidence J Nurs Manag. Nov 24 8 2016 994 1006 22 Simpson R. Simpson A. Embodying” dirty work: a review of the literature Sociology Compass 12 6 2018 e12581 23 Weaver R. Salamonson Y. Koch J. Jackson D. Nursing on television: student perceptions of television's role in public image, recruitment and education J Adv Nurs. Dec 69 12 2013 2635 2643 24 Hall L.M. Angus J. Peter E. O'Brien-Pallas L. Wynn F. Donner G. Media portrayal of nurses' perspectives and concerns in the SARS crisis in Toronto J Nurs Scholarsh 35 3 2003 211 216 14562487 25 Nowroozi I. Healthcare workers threaten more industrial action after turning down 3pc pay rise offer 2022 26 Xu G. Zeng X. Wu X. Global prevalence of turnover intention among intensive care nurses: a meta-analysis Nurs Crit Care Jul 14 2021 27 Elliott R. Crowe L. Abbenbroek B. Grattan S. Hammond N.E. Critical care health professionals' self-reported needs for wellbeing during the COVID-19 pandemic: a thematic analysis of survey responses Aust Crit Care 35 1 2022/01/01/2022 40 45 28 Hammond N.E. Crowe L. Abbenbroek B. Elliott R Tian D Donaldson L Impact of the coronavirus disease 2019 pandemic on critical care healthcare workers’ depression, anxiety, and stress levels Aust Crit Care. Mar 34 2 2021 146 154 29 Khan N. Jackson D. Stayt L. Walthall H. Factors influencing nurses' intentions to leave adult critical care settings Nurs Crit Care 24 1 2019 24 32 29635820 30 Chen Y.A.-O. Guo Y.A.-O. Chin W.S. Cheng N.Y. Ho J.J. Shiao J.S. Patient-nurse ratio is related to nurses' intention to leave their job through mediating factors of burnout and job dissatisfaction 2019 LID - 10.3390/ijerph16234801 [doi] LID - 4801(1660-4601 (Electronic)) 31 Duffield C.M. Roche M.A. Homer C. Buchan J. Dimitrelis S. A comparative review of nurse turnover rates and costs across countries J Adv Nurs 70 12 2014 2703 2712 25052582 32 Dawson AJ, Stasa H, Roche MA, Homer CSE, Duffield C. Nursing churn and turnover in Australian hospitals: nurses perceptions and suggestions for supportive strategies.	36470775	PMC9716433	NO-CC CODE	2022-12-03 23:20:57	no	Aust Crit Care. 2022 Dec 2; doi: 10.1016/j.aucc.2022.11.008	utf-8	Aust Crit Care	2,022	10.1016/j.aucc.2022.11.008	oa_other
==== Front Urolithiasis Urolithiasis Urolithiasis 2194-7228 2194-7236 Springer Berlin Heidelberg Berlin/Heidelberg 36459218 1382 10.1007/s00240-022-01382-7 Research Validation of the Italian version of wisconsin stone quality of life (WISQOL): a prospective Italian multicenter study Mazzon Giorgio [email protected] 1 Serafin Emanuele 2 Ferretti Stefania 3 Claps Francesco 4 Zhong Wen 5 Fiori Cristian 6 Celentano Giuseppe 7 Guarino Giulio Gaetano 3 Zamengo Davide 6 Piasentin Andrea 4 Creta Massimiliano 7 Longo Nicola 7 Dordoni Roberta 3 Pavan Nicola 4 Brancelli Claudio 2 Cerruto Maria Angela 2 Antonelli Alessandro 2 Celia Antonio 1 1 grid.416724.2 0000 0004 1759 6760 Department of Urology, San Bassiano Hospital, Bassano del Grappa, Vicenza Italy 2 Department of Urology, University of Verona, Azienda Ospedaliera Universitaria Integrata Verona, Verona, Italy 3 grid.411482.a Department of Urology, University Hospital of Parma, Parma, Italy 4 grid.5133.4 0000 0001 1941 4308 Urology Clinic, Department of Medical, Surgical and Health Science, University of Trieste, Trieste, Italy 5 grid.470124.4 Department of Urology, First Affiliated Hospital of Guangzhou Medical University, Guangzhou, China 6 grid.415081.9 0000 0004 0493 6869 Department of Urology, San Luigi Gonzaga Hospital, University of Turin, Orbassano, Turin Italy 7 grid.4691.a 0000 0001 0790 385X Department of Neurosciences, Sciences and Odontostomatology, Urology Unit, University of Naples “Federico II”, Naples, Italy 2 12 2022 2023 51 1 73 11 2022 5 11 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Urolithiasis is a worldwide spread condition that affects patients’ Health-Related Quality of Life (HRQOL), which measurement is an important tool for routine clinical and research practice. Disease-specific HRQOL measures demonstrated to perform better in assessing the effects of specific conditions. A disease-specific questionnaire for kidney stones, the WISQOL, has been validated in different languages, but an Italian version is still missing. Our aim is to produce and validate the Italian version of WISQOL (IT-WISQOL). Patients undergoing any elective treatment for upper urinary tract stones were enrolled. A multi-step process with forward- and back-translation was used to translate WISQOL into Italian. Patients were evaluated within 15 days pre-operatively and then at 30-, 90 days post-operatively and administered both IT-WISQOL and SF-36v2. Post-operative data such as 30 days postoperative complications, late stone-related events, successful status, and stone complexity were collected. Cronbach’s α was used to evaluate the internal consistency of IT-WISQOL, while Spearman’s rho was used for item and inter-domain correlations and IT-WISQOL with SF-36v2 correlation. We found excellent internal consistency across all domains (α ≥ 0.88), particularly when the total score is considered (α = 0.960). Test–retest reliability showed excellent results for the total questionnaire (Pearson correlation value: 0.85). The Inter-domain association ranged from 0.497 to 0.786. Convergent validity was confirmed by a good correlation with subdomains of the SF-36v2 measures. IT-WISQOL is a reliable tool to measure HRQOL in stone patients. It shows analog characteristics if compared to English WISQOL. Supplementary Information The online version contains supplementary material available at 10.1007/s00240-022-01382-7. Keywords Health-related quality of life WISQOL SF-36v2 Kidney stones Urolithiasis Questionnaire issue-copyright-statement© Springer-Verlag GmbH Germany, part of Springer Nature 2023 ==== Body pmcIntroduction Urolithiasis is a worldwide spread condition with a prevalence ranging from 1 to 20% [1] and it is expected to represent a rising global challenge [2]. Additionally, the high risk of recurrence may determine the need to repeat periodically investigations and/or surgical treatments, and may as well expose patients to bothersome symptoms, including pain, infections, haematuria, etc. [3]. These events may affect patients’ Quality of Life (QOL) under many aspects, from social to working life. In fact, kidney stones have been shown to be associated with stress, anxiety, and depression even between acute episodes [4, 5]. As expected, Health-related QOL (HRQOL) of stone formers was found to be reduced compared to healthy adults [6]. HRQOL measurements are important tools for routine clinical and research practice, as they provide a better understanding of the specific disease’s impact, and they allow to standardize elements that are difficult to balance. In recent years, they have gained growing attention and influenced treatment strategies. In terms of urolithiasis, urologists rely on surrogates of surgical adequacy, like the absence of residual fragments or complication rates, to determine the success of different treatments. Patients’ perspectives are frequently different and consider also different elements including the invasiveness of the procedure, the peri-operative discomfort, the length and complexity of postoperative recovery. All these elements have to be measured and balanced before treatment planning and should be discussed during patients’ counseling [7]. Additionally, the evaluation and comparison of HRQOL may play an important role in data comparison and research purposes. Different generic questionnaires have been created to evaluate the effect of surgeries on patients’ general HRQOL [8]. However, they might not be sufficient to gain a deep understanding of the impact of urinary stone disease. Furthermore, it is demonstrated that disease-specific HRQOL measures perform better than their generic equivalents in assessing the effects of specific conditions [9]. To fill this gap, a disease-specific questionnaire for kidney stones, the “wisconsin stone quality of life questionnaire” (WISQOL), was recently developed and validated by Penniston et al. [7, 10]. The WISQOL has already been translated and validated in different languages, i.e., Spanish, Turkish, Korean, French,German, Chinese, Japanese and Slovak [11–18], but an Italian version has not been developed and tested yet. Aim of our study is to produce and validate the Italian version of WISQOL (IT-WISQOL). We will also test its reliability using for comparison the 36-Item Short Form Health Survey (SF-36v2), a validated questionnaire to evaluate HRQOL. Materials and methods Study design Six different Italian urological centers sited in different regions of the peninsula were involved in this prospective observational study. The initial study period was 2019–2020. Due to the advent of the COVID-19 outbreak and related difficulties in recruiting patients, we extended it to September 2021. Patients undergoing any elective treatment for upper urinary tract stones with curative intent were enrolled in the study. Surgical treatments included Extracorporeal Shock Wave Lithotripsy (SWL), ureterolithotripsy (URS), Retrograde Intra-Renal Surgery (RIRS) or Percutaneous Nephrolithotomy (PCNL). All cases have been carried out by surgeons beyond their learning curve. Patients treated in an emergency setting were excluded from the study. To be eligible, patients had to be older than 18 years, native Italian speakers, with a computed tomography (CT) scan performed pre-operatively within 3 months and willing and able to give informed consent. Exclusion criteria included any of the following: contraindications to surgical treatment; no preoperative CT scan available within 3 months before surgery; pregnant patients; American Society of Anesthesiologists (ASA) [19] score ≥ 3 or any other ongoing medical conditions causing pain or deterioration of general health. Questionnaires in use The WISQOL questionnaire is composed of 28 items and it measures disease-specific impact identifying four different domains: social impact, emotional impact, disease impact, and impact on vitality [10]. Each item of the questionnaire is scored on a Likert scale ranging from 1 to 5, with the maximum score being 140 and, thus, a high score correlates to high HRQOL [7]. The WISQOL only refers to events or symptoms occurring in the previous four weeks. SF-36v2 is a multipurpose, short-form health survey, widely adopted because of its shortness and comprehensiveness, and differently from WISQOL it is a generic measure. It is built on three levels: (1) 36 items, each of those is used in scoring only one scale; (2) eight scales that can be grouped by four according to the physical and mental health variance they have in common to form (3) two summary measures, called “Physical Health” and “Mental Health” [20]. Questionnaire preparation The translation process followed the multi-step process recommended by Hutchinson [21]. The original version of WISQOL was reviewed and translated into Italian by two native Italian-speaking urologists. Both surgeons were specialized in urolithiasis management and fluent in English. A meeting was held to review both versions and consensus was achieved on an initial version of the questionnaire. Subsequently, the questionnaire was administered to a different urologist not familiar with WISQOL to carry out the back-translation in English. This back-translated version and the original English version were reviewed together for comparison. Few further modifications were introduced to render it easier to read and understand. A final consensus was reached after two rounds of meetings. Five patients were recruited and administered the questionnaire to test its readability and comprehension. All participants were able to understand and fill in the questionnaire correctly, so no further modifications were included. Patients were evaluated within 15 days pre-operatively and then at 30-, 90-days post-operatively and administered both IT-WISQOL and the SF-36v2. 30 days postoperative complications were collected and graded using Clavien-Dindo classification System [22]. In the case of PCNL, the modified version by de la Rosette et al. was utilized [23]. Late stone-related events were also recorded. Successful status, defined as no residual fragment ≥ 2 mm, was assessed either by performing non-contrast-enhanced CT or by an ultrasound scan of the urinary tract plus kidney-ureter-bladder (KUB) radiogram within 3 months after treatment. The stone complexity was calculated using Guy’s and S.T.O.N.E. nephrolithometry scores [24, 25]. Sample size population Previous studies validating different language versions of the questionnaire documented a Spearman’s rank correlation between WISQOL and SF-36v2 varying between 0.5 and 0.7 [10, 12–14]. Expecting similar results, considering a 95% Fisher confidence interval, a sample size population of about 150–200 patients was calculated to obtain adequate results. Statistical analysis All the data were analyzed with the IBM Statistical Package for the Social Sciences Statistics version 25.0 (IBM SPSS Statistics; Armonk, NY, USA). Continuous variables are presented as means (SD) in cases of normal distribution and compared using the independent Student’s t test, while in cases of skewed distributions they are presented as median and interquartile range (IQR) and compared using the Mann–Whitney U test. Categorical variables are presented as numbers with percentages and compared using chi-square or Fisher’s exact tests. Cronbach’s α was used to evaluate internal consistency of IT-WISQOL between Centers. Test–retest reliability was evaluated by comparing the 30-day-postoperative questionnaire and the 3-month-postoperative questionnaire. Spearman test was used for item and inter-domain correlations and for IT-WISQOL Total Score. As Penniston et al. [10] did, IT-WISQOL was divided into four domains describing social impact, emotional impact, disease impact and impact on vitality, and matched with the relative domains of SF-36v2, namely “Role Physical”, “Mental Health”, “Body Pain”, “Vitality”. Correlation of total Scores of IT-WISQOL and SF-36v2 was assessed to determine convergent validity using Spearman rank correlation. Correlations between clinical variables and results from IT-WISQOL and SF-36v2 were analyzed to assess the effect of these on results. Cronbach’s α and correlation coefficients results were interpreted as in the original WISQOL development study. Statistical significance will be considered for two-tailed P values of < 0.05. Results A total of 226 patients were enrolled in the study. Patients’ characteristics and preoperative outcomes are reported in Supplementary Table 1. Among them 141 (62.4%) were male and 85 (37.6%) were female. Most of them (68.1%) had a high school degree or higher. A minority of patients were stented (24.8%) or had a nephrostomy placed (6.2%) before treatment. Stone’s cumulative diameter was in median (IQR) 12.5 (8.2–20.0) mm. Median (IQR) Guy’s score was 2 (1–2), median (IQR) S.T.O.N.E. score was 7 (5–8). Perioperative data are reported in Supplementary Table 2. Majority of patients (49.5%) underwent retrograde treatments (URS or RIRS), 28.3% underwent SWL and 22.12% PCNL. Most patients (68.6%) were discharged with indwelling ureteral stent or nephrostomy tube, that were removed after a median of 21(1–45) days. 22 patients (9.7%) experienced minor complications (Clavien–Dindo ≤ 2) during in-hospital stay, mostly haematuria. In the following month, 23 (9.3%) patients experienced complications and stent-related pain was the most common complication reported. 32 (14.2) experienced a stone-related event within 3 months from discharge. Overall SFR at 3 months was 86.7%. Questionnaires The average preoperative IT-WISQOL score of 226 patients was 96.16 ± 23.86. It raised to 103.66 ± 19.87 at 30 days postoperative and 109.07 ± 19.56 at 3 months postoperative. Fig. 1a, b, c describe the correlation between IT-WISQOL and SF-36v2 preoperative, 30- and 90 days separately. The total score of IT-WISQOL and SF-36v2 had significant correlation in all three preoperative (r = 0.836, p < 0.001), 30 days postoperative (r = 0.827, p < 0.001) and 90 days postoperative (r = 0.817, p < 0.001) questionnaires.Fig. 1 Correlation of WISQOL and SF-36v2 a Preoperative; b 30 days postoperative; c 3 months postoperative. Min–Max of IT-WISQOL score is 28–140 respectively The internal consistency of each domain was evaluated by Cronbach’s α of the preoperative questionnaire. The α values were all > 0.8, representing excellent internal consistency. Test–retest reliability was evaluated by comparing the 30-day-postoperative questionnaire and the 3-month-postoperative questionnaire. The value of Pearson correlation of each domain was > 0.6 and no significant differences were observed (p < 0.01) (Table 1).Table 1 Internal Consistency and Test–retest reliability Internal consistency Test Retest Pearson correlation value (p < 0.01) Mean ± SD Mean ± SD D1-social impact 0.899 32.66 ± 7.55 34.91 ± 5.68 0.66 D2-emotional impact 0.888 27.58 ± 6.33 29.14 ± 5.69 0.67 D3-stone-related impact 0.903 31.43 ± 7.35 33.49 ± 6.57 0.70 D4-vitality impact 0.929 11.99 ± 3.44 12.79 ± 2.92 0.65 Total score 0.960 83.45 ± 45.69 84.94 ± 49.41 0.85 D1 domain 1-social impact (WISQOL items: 3a, 3b, 3c, 3d, 3e, 6a, 6b, 6c), D2 domain 2-emotional impact (WISQOL items: 4c, 7a, 7b, 7c, 7d, 7e, 7f), D3 domain 3-disease impact (WISQOL items: 2a, 2b, 2c, 2d, 5a, 5b, 5c, 5d), D4 domain 4-impact on vitality (WISQOL items: 1a, 1b, 1c) The value of the inter-domain association ranged from 0.497 to 0.786, showing a moderate difference and correlation among domains. Different domains of all three questionnaires demonstrated a significant correlation to the total score (Table 2).Table 2 Spearman rank correlations: inter-domain association and convergent validity Domains Preoperative 30 days postoperative 3 months postoperative Inter-domain association D1-D2 0.721 0.771 0.733 D1-D3 0.614 0.723 0.699 D1-D4 0.681 0.633 0.650 D2-D3 0.593 0.640 0.573 D2-D4 0.636 0.606 0.497 D3-D4 0.786 0.709 0.704 D1-TOTAL 0.847 0.728 0.736 D2-TOTAL 0.810 0.626 0.620 D3-TOTAL 0.849 0.719 0.726 D4-TOTAL 0.850 0.695 0.712 Convergent Validity* D1-Role physical 0.668 0.570 0.531 D2-Mental health 0.605 0.425 0.347 D3-Body pain 0.624 0.646 0.559 D4-Vitality 0.709 0.639 0.604 D1 domain 1-social impact (WISQOL items: 3a, 3b, 3c, 3d, 3e, 6a, 6b, 6c), D2 domain 2-emotional impact (WISQOL items: 4c, 7a, 7b, 7c, 7d, 7e, 7f), D3 domain 3-disease impact (WISQOL items: 2a, 2b, 2c, 2d, 5a, 5b, 5c, 5d), D4 domain 4-impact on vitality (WISQOL items: 1a, 1b, 1c) *Correlations between corresponding domains in the WISQOL and SF-36v2 questionnaires All domains of IT-WISQOL significantly correlate with the corresponding domains of SF-36v2, though the value of Spearman rank test turned weaker at two postoperative questionnaires (Table 2). Discussion Urolithiasis is a common condition with increasing incidence worldwide [2]. Patients’ HRQOL was found to be decreased, particularly with respect to their physical and mental states [5, 6, 26–28]. Penniston et al. [7, 10] developed a stone-specific questionnaire, the WISQOL, with robust psychometric properties, which has been already translated and validated in different languages [11–15]. Worldwide, we are observing an increasing interest in publishing high-quality papers on urolithiasis. In fact, Abedi et al. [29], during the decade 2010–2020 if compared with the previous decade, reported an increased volume of publications corresponding to 133, 103.5 and 70.4% for ureteroscopies, PCNLs and SWL respectively. This fact highlights the need for tools to objectively report, analyze and compare HRQOL parameters. Therefore, in our opinion, the IT-WISQOL will provide important support for this purpose. This study demonstrates that IT-WISQOL is a reliable assessment tool for the evaluation of symptoms in patients suffering from urolithiasis. IT-WISQOL showed satisfactory validation results, with excellent internal consistency across all domains (α ≥ 0.88), particularly when the total score is considered (α = 0.960). Similar results were also found with the original WISQOL [10] and the French version [13], both with a total score α = 0.970. The test–retest reliability showed good values for each domain and excellent results for the total questionnaire (Pearson correlation value for total score: 0.85) (Table 1). Comparison with other studies is not easy as test–retest reliability has been measured with different tools, like Spearman rank correlation or Cronbach’s α in the German and Turkish validation studies respectively [12, 14]. Inter-domain reliability analysis showed analogous results to other validation studies [10, 12, 13]. Consistently with previous works [10, 12, 13] we found a strong correlation between the vitality domain (D4) and the emotional domain (D2) at 3-month follow-up (Spearman’s rho correlation coefficient = 0.497), but not at preoperative and 30-days postoperative questionnaires. The reason for these findings remains to be clarified. According to Bhojani et al. [13], this weaker correlation suggests that patients with lower HRQOL in the context of nephrolithiasis also tend to experience higher levels of stress (Table 2). Convergent validity of IT-WISQOL was confirmed by a good correlation with subdomains of the SF-36v2 measures (Table 2). Figure 1 shows how the total score of IT-WISQOL and SF-36v2 were significantly correlated at any of the follow-up times the questionnaire has been administered. This makes IT-WISQOL a dedicated reliable tool and comparable with the Italian SF-36v2 questionnaire. This is the first study aiming to evaluate the reliability and adequacy of the Italian version of WISQOL. The questionnaire demonstrated to be reliable if compared to other versions. We included 6 Centers located in different parts of Italy with different dialects and in all of them the questionnaire resulted to be adequately usable. Our study has some limitations. First of all, the number of patients involved is limited. Studies using IT-WISQOL in larger populations are desirable. Additionally, most of the patients had a high educational level, thus further testing on patients with low education may further improve the readability of the questionnaire. Patients have been recruited in Academic centers, thus they might have more complex stone disease than the general population, affecting our results. Furthermore, the study has been conducted during two COVID-19 outbreaks. This limited our capacity in recruiting patients and a certain amount of distress on patients could be expected, potentially affecting final results. Future perspectives for this field of study may be the evaluation of the differences of IT-WISQOL scores among sexes or patients of different age as well as to investigate whether HRQOL differs in patients with different preoperative (history of prior stones, stone burden), operative (type of surgery received) or postoperative (presence of residual fragments, stent) characteristics. Conclusion The IT-WISQOL has demonstrated to be a reliable tool to measure HRQOL in stone patients. It also showed analog characteristics if compared to English WISQOL. In our opinion it represents an important tool for treatment counseling, monitoring patients’ conditions and also research purposes. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 17 KB) Supplementary file2 (DOCX 15 KB) Author contributions GM: protocol and project development, manuscript writing, data analysis and interpretation ES: manuscript writing, data collection, data management, data analysis and interpretation WZ: statistical analysis SF, FC, CF, GC, GGG, DZ, AP, MC, NL, RD, NP, CB, MAC, AA, AC: data collection, supervision and critical revision All authors read and approved the final version of the manuscript. Funding The authors did not receive financial support. Data availability The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Conflict of interest The authors declare that they have no conflict of interest. Ethical approval The study was approved by the ethics committee of each participating center and performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments. Informed consent Informed consent was obtained from all individual participants included in the study. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Türk C, Neisius A, Petřík A et al (2021). EAU Guidelines on Urolithiasis. In: EAU Guidelines, edn. presented at the EAU Annual Congress Milan 2021, European Association of Urology Guidelines Office: Arnhem, the Netherlands, (ISBN 978-94-92671-13-4) 2. Sorokin I Mamoulakis C Miyazawa K (2017) Epidemiology of stone disease across the world World J Urol 2017 359 35 1301 1320 10.1007/S00345-017-2008-6 3. Khan SR Pearle MS Robertson WG Kidney stones Nat Rev Dis Prim 2016 21 2 1 23 10.1038/nrdp.2016.8 4. Diniz DHMP Schor N Blay SL stressful life events and painful recurrent colic of renal lithiasis J Urol 2006 176 2483 2487 10.1016/J.JURO.2006.07.156 17085135 5. Angell J Bryant M Tu H Association of depression and urolithiasis Urology 2012 79 518 525 10.1016/J.UROLOGY.2011.10.007 22136748 6. Kartha G Calle JC Marchini GS Monga M Impact of stone disease: chronic kidney disease and quality of life Urol Clin North Am 2013 40 135 147 10.1016/J.UCL.2012.09.004 23177641 7. Penniston KL Nakada SY Development of an instrument to assess the health related quality of life of kidney stone formers J Urol 2013 189 921 930 10.1016/J.JURO.2012.08.247 23017521 8. Urbach DR Measuring quality of life after surgery Surg Innov 2005 12 161 165 10.1177/155335060501200216 16034507 9. Wiebe S Guyatt G Weaver B Comparative responsiveness of generic and specific quality-of-life instruments J Clin Epidemiol 2003 56 52 60 10.1016/S0895-4356(02)00537-1 12589870 10. Penniston KL Antonelli JA Viprakasit DP Validation and reliability of the wisconsin stone quality of life questionnaire J Urol 2017 197 1280 1288 10.1016/J.JURO.2016.11.097 27889419 11. Basulto-Martínez M Olvera-Posada D Velueta-Martínez IA (2020) Quality of life in patients with kidney stones: translation and validation of the Spanish wisconsin stone quality of life questionnaire Urolithiasis 2020 485 48 419 424 10.1007/S00240-020-01192-9 12. Atalay HA Ülker V Canat L Validation of the Turkish version of the wisconsin stone-quality of life questionnaire Turkish J Urol 2019 45 118 123 10.5152/TUD.2018.35305 13. Bhojani N Moussaoui G Nguyen DD Validation of the French version of the wisconsin quality of life (WISQOL) questionnaire for patients with nephrolithiasis Can Urol Assoc J 2020 15 227 231 10.5489/cuaj.6552 14. Gottstein M Pratsinis M Güsewell S The German linguistic validation of the wisconsin stone quality of life questionnaire (WisQoL World J Urol 2020 396 39 2163 2168 10.1007/S00345-020-03405-7 15. Yoon YE Cho SY Translation and linguistic validation of the korean version of the wisconsin stone quality of life questionnaire Int Neurourol J 2020 24 77 83 10.5213/INJ.1836238.119 32252189 16. Zhong W, Xu J, Mazzon G et al (2022) Translation and validation of the Chinese version of wisconsin stone quality of life questionnaire in patients with kidney stones. Minerva Urol Nephrol (online ahead of print). 10.23736/S2724-6051.22.04905-9 17. Okada T Hamamoto S Taguchi K Validation of the Japanese version of the wisconsin stone quality of life questionnaire: results from SMART study Group J Endourol 2021 10.1089/END.2021.0292 34162226 18. Svihra J Sopilko I Svihrova V Is health-related quality of life of patients after single-use flexible ureteroscopy superior to extracorporeal shock wave lithotripsy? a randomised prospective study Urolithiasis 2021 49 73 79 10.1007/S00240-020-01224-4 33164114 19. Dripps RD Lamont A Eckenhoff JE The role of anesthesia in surgical mortality JAMA 1961 178 261 266 10.1001/JAMA.1961.03040420001001 13887881 20. Ware JEJ SF-36 Health survey update Spine 2000 25 3130 3139 10.1097/00007632-200012150-00008 11124729 21. Hutchinson A, Bentzen N, König-Zahn C (1997) Cross-cultural health outcome assessment: a user’s guide. ERGHO (European Research Group on Health Outcomes), (ISBN 9789036707480) 22. Dindo D Demartines N Clavien P-A Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey Ann Surg 2004 240 205 213 10.1097/01.sla.0000133083.54934.ae 15273542 23. De La Rosette JJMCH Opondo D Daels FPJ Categorisation of complications and validation of the clavien score for percutaneous nephrolithotomy Eur Urol 2012 62 246 255 10.1016/J.EURURO.2012.03.055 22487016 24. Thomas K Smith NC Hegarty N Glass JM The guy’s stone score—grading the complexity of percutaneous nephrolithotomy procedures Urology 2011 78 277 281 10.1016/J.UROLOGY.2010.12.026 21333334 25. Okhunov Z Friedlander JI George AK S.T.O.N.E. nephrolithometry: novel surgical classification system for kidney calculi Urology 2013 81 1154 1160 10.1016/J.UROLOGY.2012.10.083 23540858 26. Raja A, Hekmati Z, Joshi HB (2016) How do urinary Calculi influence health-related quality of life and patient treatment preference: a systematic review. https://home.liebertpub.com/end 30:727–743. 10.1089/END.2016.0110 27. Bryant M Angell J Tu H Health related quality of life for stone formers J Urol 2012 188 2 436 440 10.1016/J.JURO.2012.04.015 22704108 28. Miyaoka R, Ortiz-Alvarado O, Kriedberg C et al (2012) Correlation between stress and kidney stone disease. https://home.liebertpub.com/end 26:551–555. 10.1089/END.2010.0536 29. Abedi AR Razzaghi M Montazeri S Allameh F The trends of urolithiasis therapeutic interventions over the last 20 years: a bibliographic study J Lasers Med Sci 2021 12 1 6 10.34172/jlms.2021.14	36459218	PMC9716497	NO-CC CODE	2022-12-03 23:21:04	no	Urolithiasis. 2023 Dec 2; 51(1):7	utf-8	Urolithiasis	2,022	10.1007/s00240-022-01382-7	oa_other
==== Front Curr Microbiol Curr Microbiol Current Microbiology 0343-8651 1432-0991 Springer US New York 36459252 3112 10.1007/s00284-022-03112-z Review Article Comprehensive Investigations Relationship Between Viral Infections and Multiple Sclerosis Pathogenesis Sedighi Somayeh 1 Gholizadeh Omid 23 Yasamineh Saman 3 Akbarzadeh Sama 4 Amini Parya 3 Favakehi Parnia 5 Afkhami Hamed 6 Firouzi-Amandi Akram 7 Pahlevan Daryoush 8 Eslami Majid 9 Yousefi Bahman 10 Poortahmasebi Vahdat [email protected] 2 http://orcid.org/0000-0003-3352-7880 Dadashpour Mehdi [email protected] 1011 1 Department of Immunology, Faculty of Medicine, Medical Science of Mashhad, Mashhad, Iran 2 grid.412888.f 0000 0001 2174 8913 Department of Bacteriology and Virology, Faculty of Medical Sciences, Tabriz University of Medical Sciences, Tabriz, Iran 3 grid.411705.6 0000 0001 0166 0922 Research Center for Clinical Virology, Tehran University of Medical Sciences, Tehran, Iran 4 grid.412831.d 0000 0001 1172 3536 Department of Animal Biology, Faculty of Natural Science, University of Tabriz, Tabriz, Iran 5 grid.411757.1 0000 0004 1755 5416 Department of Microbiology, Falavargan Branch, Islamic Azad University, Isfahan, Iran 6 Department of Bacteriology, Faculty of Medicine, Medical Science of Shahed, Tehran, Iran 7 grid.412888.f 0000 0001 2174 8913 Department of Immunology, Faculty of Medicine, Tabriz University of Medical Sciences, Tabriz, Iran 8 grid.486769.2 0000 0004 0384 8779 Determinants of Health Research Center, Semnan University of Medical Sciences, Semnan, Iran 9 grid.486769.2 0000 0004 0384 8779 Department of Bacteriology and Virology, Semnan University of Medical Sciences, Semnan, Iran 10 grid.486769.2 0000 0004 0384 8779 Cancer Research Center, Semnan University of Medical Sciences, Semnan, Iran 11 grid.486769.2 0000 0004 0384 8779 Department of Medical Biotechnology, Faculty of Medicine, Semnan University of Medical Sciences, Semnan, Iran 2 12 2022 2023 80 1 158 9 2022 5 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Multiple sclerosis (MS) is a chronic autoimmune disease that affects the central nervous system (CNS). Compared to other types of self-limiting myelin disorders, MS compartmentalizes and maintains chronic inflammation in the CNS. Even though the exact cause of MS is unclear, it is assumed that genetic and environmental factors play an important role in susceptibility to this disease. The progression of MS is triggered by certain environmental factors, such as viral infections. The most important viruses that affect MS are Epstein-Barr virus (EBV), human herpes virus 6 (HHV-6), human endogenous retrovirus (HERV), cytomegalovirus (CMV), and varicella zoster virus (VZV). These viruses all have latent stages that allow them to escape immune detection and reactivate after exposure to various stimuli. Furthermore, their tropism for CNS and immune system cells explains their possible deleterious function in neuroinflammation. In this study, the effect of viral infections on MS disease focuses on the details of viruses that can change the risk of the disease. Paying attention to the most recent articles on the role of SARS-CoV-2 in MS disease, laboratory indicators show the interaction of the immune system with the virus. Also, strategies to prevent viruses that play a role in triggering MS are discussed, such as EBV, which is one of the most important. issue-copyright-statement© Springer Science+Business Media, LLC, part of Springer Nature 2023 ==== Body pmcIntroduction Multiple sclerosis (MS) is a Central nervous system (CNS) autoimmune disease [1, 2] in which nerve cells are demyelinated, causing inflammation and damage in the CNS [3–7]. Although the etiology of MS remains unclear [2, 8–10], colossal advancement has been accomplished in distinctive hazard factors connected to MS [11]. Both genetic and environmental variables are discovered in the epidemiologic analysis with a vital role in the progression of MS [11]. The role of infectious and viral agents is still controversial in MS, but there is increasing evidence that some viruses play a role in disease development [2, 12]. The most outstanding proof comes from identifying viral nucleic acids or antigens and antiviral antibody responses in patients with MS [12]. Viral infections can influence MS in different ways and combinations. These pathways include molecular mimicry, direct toxicity, bystander activation, dual T-cell receptors, and epitope spreading [12, 13]. Direct toxicity: It is based on direct damage to the cell without the intermediation of inflammation or autoimmunity. It has been observed in some of the MS plaques that the dystrophy of the oligodendrocyte cells and the precursors (without the mediation of IgG and complement) has happened, which causes a defect in the function of the blood–brain barrier (BBB) and the continuation of the disease [12]. Molecular mimicry: During this phenomenon, cross-reaction occurs, wherein myelin compounds are presented by class II major histocompatibility complex (MHC) on APCs to autoreactive CD4 + T-cell lymphocytes due to the similarity of myelin to some viral structures [12, 14]. Dual T-cell receptors: It is thought that some T lymphocytes express different TCRs with various functions, such as recognizing viral antigens or myelin antigens, and these lymphocytes activate both types of antigen responses [12]. Bystander activation: viral infections cause extensive inflammation. With this inflammation, the surrounding cells are damaged and cause the discovery and presentation of autoantigens presented by Antigen-presenting cells (APC). As a result, T and B lymphocytes became reactive [12]. Epitope expansion: damage to myelogenous cells causes myelin fragmentation in the inflammatory environment, which leads to additional epitopes with self-perpetuating destruction of myelin [12, 14]. Viruses can cause brain damage by directly infecting the CNS or through the inflammatory response that follows. Epstein-Barr virus (EBV), human herpes virus 6 (HHV-6), human endogenous retrovirus (HERV), cytomegalovirus (CMV), SARS-CoV-2 (COVID-19), and varicella zoster virus (VZV) can all enter the CNS, cause acute cellular damage and dysfunction, and remain quiescent or latent in infected cells for long periods [4, 15, 16]. They can stimulate the activation of lymphocytes and the production of pro-inflammatory cytokines that lead to neuron destruction and disrupt cellular activity in other indirect ways. These inflammatory pathways may be triggered by sensitization of brain neurons, which may result from genetic changes in MS. Although these processes related to viral roles in classical neurological disorders remain hypothetical and are still being studied. The association between infection or viral components and the onset or recurrence of symptoms in MS has long been recognized [17]. The persistent latent infections that have concealed, silent or latent phases escape from detection by the immune system and revive when exposed to multiple triggers. Their tendency for CNS and immune system cells explains their potential destructive property in neuroinflammation and neurodegeneration of CNS [18]. Also, other viruses that have been connected to the onset of MS but are not addressed in the article include Herpes simplex virus-1 (HSV-1), HSV-2, and John Cunningham Virus (JCV). For HSV-1, the seroprevalence of IgG against the virus has been found to be expanded in pediatric MS/Clinically isolated syndrome (CIS) but not adult MS compared with controls [19]. For HSV-2 the seroprevalence has been found significantly increased in MS compared with controls [19]. JCV is a non-enveloped double-stranded DNA (dsDNA) virus, which may cause progressive multifocal leukoencephalopathy (PML) characterised by infection of oligodendrocytes and astrocytes in the CNS [20–22]. In this review, we address the existing evidence linking particular viruses to MS development or aggravation. Finally, controlled clinical studies utilizing preventative or therapeutic techniques that precisely interfere with any virus in MS disease are the only way to confirm that agent’s participation in the disease effectively. Viral Triggers in MS Epstein–Barr Virus (EBV) EBV, a member of the human Herpesviridae family, has a dsDNA genome of 120 kb encoding approximately 85 proteins, and a large number of non-coding RNAs (ncRNAs) [23, 24]. A history of chronic infectious mononucleosis caused by EBV increases the risk of MS by approximately 40-fold [25]. This is one of the most critical factors in developing MS [1, 9, 26, 27]. EBV is activated, replicates, and latently exists in B lymphocytes during the lifetime of an infected individual [1, 9, 28]. The Immunological Mechanism of EBV in MS The EBV virus affects epithelial and lymphoid cells in the oropharyngeal Waldeyer’s loop. However, the virus continuously transforms B cells that express CD21, EBV’s primary cellular receptor [25, 28]. To produce the phenotypic (IgD-CD27 + and IgD + CD27 +) memory cells, where the virus causes permanent infection and alters the development of resting B cells like germinal center responses [9, 25, 28], the EBV and cellular genomes experience variable degrees of epigenetic alteration throughout the B cell transformation process, including silencing of severalviral genes required to create a sustained latent infection [1, 25, 28, 29]. Immune control is essential for this crucial process because when a virus released from each ruptured plasma cell infects an epithelial cell, virus production is increased in the tonsils, and the epithelial cell generates enough virions to infect 10,000 native B cells, implying that the harm is gradual. This cycle can have a considerable negative impact on the infected blast population [9, 28, 29]. Deficiency in cytotoxic CD8 + T-cell removal of EBV-infected B cells may increase the risk of developing MS by allowing the buildup of EBV-infected autoreactive B cells in the CNS [9, 28, 30]. There is a widespread lack of CD8 + T-cells in MS, namely the CD62L effector memory (EM/EMRA) fraction, which performs CNS immunosurveillance and defends against viral infection [28, 31]. At all MS levels, CD8 + T- cell lymphocytes do not respond to lytic phase EBV antigens, which indicates an impaired control of EBV reactivation. Contrarily, CD8 + T-cells that target EBV latent antigens are more abundant but less active, meaning an exhausted response to the more significant number of latently infected cells due to diminished CD8 + T-cell regulation of EBV reactivation [9, 28, 30, 32]. There is evidence that the course of MS is associated with cellular exhaustion of EBV-specific CD4 + and CD8 + T lymphocytes [30, 33]. However, other factors, such as a decline in EBV reactivation related to age, may also contribute to these outcomes [9, 28]. A study showed that lysis-specific CD8 + T-cells declined in peripheral blood mononuclear cells (PBMCs) at all stages of multiple sclerosis, besides clinical attacks, as well as CD8 + EMRA T-cells and CD8 + EM, the frequency of lytic-specific CD8 + T-cells was also decreased. EM/EMRA T-cells were reduced compared to ordinary cases. MS patients were significantly less likely to have lytic-specific T-cells in their CD8 + and T-cell phenotypes than normal individuals [28, 34]. The number of EBV-lytic CD8 + T-cells was reduced, and the number of latent-specific CD8 + T-cells increased in MS [28, 30, 34]. The results revealed that the numbers of latent-specific cells in the CD8 + T-cells, CD8 + EM T-cells, CD8 + CM T-cells, and CD8 + EM/EMRA T-cells substantially increased in patients with MS compared to healthy individuals with EBV-positive blood [28, 34]. A possible mechanism for CD8 + T-cell memory loss in MS is the decrease in IFN-I production because the number of EBV-specific CD4 + T-cells correlates severely with the number of EBV-specific CD8 + T-cells in MS as opposed to EBV-infected people with no previous records of MS. This could be because IFN-α or IFN-β is critical in developing CD8 + memory T-cells [28, 35]. Memory B cells can trigger the auto proliferation of Th1 cell-homing CD4 + T-cells in MS, detecting autoantigens in both B cells and MS lesions [36]. According to reports, these cells go to the brain and interact with RAS guanyl-releasing protein 2 (RASGPR2) and HLA-DR to cause inflammation [37]. CNS autoreactive T-cells may get activated in lymphoid tissue. They could, by contact with EBV-infected B cells, move into the CNS, where they receive co-stimulatory and survival cues from EBV-infected B cells. Increased B cell-mediated antigen presentation to autoreactive T-cells and prevention of their apoptosis may contribute to local inflammation persistence, attracting more inflammatory cells and causing antigen-directed harm and bystander injury to the CNS [37]. (Fig. 1).Fig. 1 The immunological mechanism involved in EBV can cause MS. 1. EBV infects naive B cells, so these cells proliferate in the germinal centers. Then the receptor of B cells or BCR and proteins of EBV are latent in the self-reactive memory cells. B: blood circulation: 2. Memory cells infected with EBV leave the lymph node and enter the bloodstream. C: Brain: 3. Memory cells that are EBV-infected will enter the brain cells (neurons) and stay there. 4. In neurons, autoreactive T-cells that have already entered the brain cause damage to neurons in two ways. 5. Infected memory cells signal B7 stimulus to CD28 receptors on the surface of active autoreactive T-cells. 6. These co-stimulus signals activate T-cells by producing interferon-gamma, interferon-beta, and IL-2. 7. Autoreactive T-cells detect apoptotic myelin fragments by microglia (brain macrophages) through MCH, eventually causing further apoptosis of the myelin sheath around neurons. 8. Autoantibodies produced by B cells infected with EBV sit on the myelin sheath, releasing myelin and oligodendrocyte fragments. 9. Released myelin fragments by MHC memory cells that are EBV-infected are given to the autoreactive T-cells, and this degradation process by the T-cells continues MS is associated with high levels of EBV-induced G protein-coupled receptor 2 (EBI2) expression, which mediates CNS autoimmunity, lymphocyte migration, and MS lesions [38]. The EBI2 receptors play a critical role in myelin formation and mediate the halt of demyelination induced by lysophosphatidylcholine-induced demyelination (LPC, lysolecithin). The oxysterol-EBI2 direction is involved in immunoregulatory responses, and the unique expression of this receptor is involved in the antigen-specific B-dependent antibody responses, and T-dependent antibody responses [39, 40]. EBV survival is elevated by EB nuclear antigen-2 (EBNA-2) modulation of gene expression, which leads to a higher incidence of lymphoma and autoimmune diseases, such as MS [26]. Therefore, a recent study has suggested that mutations in EBNA-2 might influence host responses to EBV and MS sensitivity [26, 30, 41]. As revealed here, EBNA-2 connects to five of six genes associated with MS with allele imbalance [42], and suppressing EBNA-2 changes the expression of five of these genes. There is still significant uncertainty concerning how EBNA-2 affects MS susceptibility by promoting the expression of (mutated) Single-nucleotide polymorphisms (SNPs) in lymphoblastoid cell lines, thus affecting EBV’s ability to evade immune responses and exposure to MS. Studies should be conducted to determine whether inhibiting EBNA-2 could offer an excellent therapeutic way of treating MS [26, 30, 43]. It has been suggested that astrocyte EBV infections could activate human endogenous retrovirus-W (HERVW)/MS-associated retrovirus (MSRV)/syncytin1 in humans. Laboratory Diagnostic Markers of EBV in MS Although almost all MS patients are seropositive, whereas EBV seropositivity is present in approximately 95% of the world’s population [9]. A record of infectious mononucleosis significantly accelerates the risk of multiple sclerosis. In addition, EBV-specific oligoclonal bands have been identified in cerebrospinal fluid (CSF) in a subgroup of patients with MS [12, 44–46]. Recent studies have shown that the level of serum components differs in MS patients compared to healthy people. Antibodies that cross-react with myelin essential protein (MBP) can recognize a unique epitope of EBNA-1 411–426 [12, 28]. This has led to the hypothesis that EBV can reintroduce “forbidden” memory B-lymphocytes to CNS epitopes. EBV memory B cells may lose EBV recognition DNA during replication, which explains why EBV is not always found in MS lesions [9]. Still, authentication of the “forbidden” epitope would be preserved, potentially triggering molecular mimicry. In addition, a “two-hit concept” can explain the relationship between MS and EBV infection. In the initial condition, EBV alters the penetrance of the BBB, which allows activated immune cells to enter the CNS, triggering a cascade of events that leads to inflammation of the CNS [9, 12]. Increased antibodies against the EBV latency-associated nuclear antigen 1 (EBNA-1) can be noted before the onset of neurological symptoms [25, 28]. It is proposed that the raised titer of anti-EBNA-1 IgG reflects an elevated amount of latent EBV antigen load due to faulty control of EBV by T-cells [9, 28]. Moreover, EBV infection evidence in brain-infiltrating B cells in brain lesions of patients with MS has been noted [47, 48]. A novel study about the transcriptome of B cells received from CSF, and MS patients’ peripheral blood was bare of human viral transcripts, which included EBV [9, 25, 47]. It has also been suggested that EBV boosts the survival of oligoclonal antibodies-producing autoreactive B cells in the CSF of patients suffering from MS, and boosts the survival and infiltration of these infected cells toward the CNS, leading to the pathogenesis of MS [25]. As a study showed, the anti-EBNA-1 IgG and anti-VCA IgG titers were elevated in MS patients. The negative correlation between the anti-EBNA-1 IgG titer and the LCL-specific CD8 + T-cell frequency in MS validates their former finding, which noted an inverse correlation between this titer and the LCL-specific T-cell frequency [28, 49]. The immune response to the infection of EBV in people with MS (PwMS) is different than in healthy people. For instance, levels of EBNA-1 IgG are higher in both children and adults suffering from MS [50]. There is a positive relation between anti-EBNA1, but not anti-VCA titers, but anti-EBNA-1 titers of IgG, and blood EBV DNA load in both patients with MS and healthy individuals, as well as noted for patients with Hodgkin lymphoma and healthy relatives [28, 51]. Medication Control of EBV Effective control of EBV infections has been suggested as a remedy for preventing and curing autoimmune diseases. In patients with MS, controlling infection of EBV can be done by depleting B cells, enhancing immunity, antiviral drugs, and ameliorating immune surveillance [9]. Specific agents Mechanisms of action (MOA), inactivation through viral kinase, narrowed the study duration, single-agent versus a combination of antiviral drugs (e.g., efficacy against HIV). Readers should remember that one of the top five options for the treatment for MS remains a potent antiviral protein, human IFN-beta [52]. IFN-beta, a significant contributor to MS, inhibits the ability of EBV, CMV, and other viruses to replicate, affects T-cell proliferation in response to EBNA-1, and decreases the compartment of B cell memory through a subset of cellular pathogens. [9]. The hypothesis that IFN-beta and other disease-modifying treatments (DMTs) have overlapping antiviral and anti-inflammatory mechanisms supports the concurrent testing of MS medications with various modes of action (MOAs) in the future [53]. Currently, there is not any vaccine that protects against infection of EBV. One viable approach to developing such a vaccine is targeting gp350 or other viral proteins [54]. New progressions in immunization by vaccines, focusing on DNA or RNA vaccines, which are capable of furnishing sequences coding multiple proteins, could facilitate advanced immunization investigations about MS. The idea of producing a prophylactic vaccine for averting acute infection of EBV as a strategy for preventing MS progression remains intriguing. Recent exploration of other herpesvirus vaccines has supplied encouraging results that support the view that designing a vaccine that could stop disease instead of infection might be feasible [55, 56]. The importance of B cells in the pathology of MS was highlighted by the potent efficacy of anti-CD20 B cell abatement treatments like ofatumumab, rituximab, and ocrelizumab. A recent study using adaptive delivery of autologous EBV-specific T-cells that demonstrated encouraging clinical results supports an EBV + B cell pathogenic role in the pathogenesis of MS [7, 9, 25, 57]. HHV-6 Since 1993, a relationship between HHV-6 and MS has been hypothesized, and numerous studies have been conducted. HHV-6 is a neurotrophic virus linked to various nervous disorders, including MS and neuromyelitis opticus. Numerous clinical studies have found a link between MS and HHV-6 infection [18, 58, 59]. HHV-6 A and HHV-6 B [12, 58] are two HHV-6 species with 95% homology [12, 58, 60]. HHV-6A and HHV-6B are large enveloped (Env) beta herpesviruses with a dsDNA genome [59]. Several studies have shown that the incidence of HHV-6A is higher than HHV-6B in samples from MS patients [58, 61]. Activation of the Immune System by HHV-6 in MS HHV-6 can activate the immune system to generate a “fertile field” for the proliferation of autoreactive T-cells induced by other environmental stimuli. Evidence also suggests that HHV-6 has a role in nervous system disorders by infecting microglia cells in the CNS and causing them to produce a pro-inflammatory response [12, 58]. Adult oligodendrocytes, astrocytes, and microglia cells exhibit the complement system receptor CD46, which is used by HHV-6. These characteristics make them ideal candidates for modulating MS pathogenic pathways. HHV-6A is thought to use this receptor more frequently [4]. HHV-6B’s primary receptor is the CD134 protein on the surface of T-cells [62]. HHV-6A is responsible for the development of demyelination and can cause latent infection in oligodendrocytes, which are thought to be a target of autoimmune responses in MS pathogenesis [58]. Compared to control tissue, HHV-6 DNA and protein were found in MS plaques, particularly oligodendrocytes [4]. In MS, HHV-6A activates latent EBV in B cells. Latent EBV-infected B cells are imported into the CNS after infection with neurotrophic HHV-6A [63]. In MS patients, this virus can operate as an originator or potentiator of inflammatory plaques. CD8 + T lymphocytes responses to HHV-6-infected CNS cells can cause tissue injury and the release of sequestered antigens, which activate self-reactive lymphocytes and boost autoreactive immune reactions. Improvement of the activation of the complementary systems can occur by the CD46 used by HHV-6A as a cellular receptor [4]. Both kinds of HHV-6 may cause Th2 migration in T-helper cell balance by blocking IL-12 production by DCs and macrophages. Other studies have found that HHV-6 infection increases the production of inflammatory cytokines like IL1, TNF, and IFN in PBMCs [58]. It increases the production of IL-18 and IFN-γ receptors in T lymphocytes while decreasing the expression of IL-10 and IL-14, changing the balance of T-helper cells towards the Th2 phenotype. HHV-6A has also been demonstrated to aggravate disease progression and induce the production of IL-15 in NK cells [58]. Infected astrocytes had a reduced ability to ingest glutamate, which was linked to lower expression of the glial glutamate transporter EAAT-2 [64]. Over-activation of AMPA and kainate receptors can cause cytotoxic death of oligodendrocytes and oligodendroglial death when dysregulation of glutamate levels [4]. Laboratory Diagnostic Markers of HHV-6 in MS Innumerable clinical studies have shown an association between infection of HHV-6 and MS. For instance, serum samples showing HHV-6 DNA levels representing active infection are significantly increased in patients suffering from MS compared to healthy individuals or patients with different neurological-related diseases [58]. In addition, in the CSF and PBMCs of patients with MS, HHV-6 DNA has been identified at a higher level [12, 58]. Another study showed that in brain autopsy samples from patients with MS compared to healthy individuals, genes exclusively expressed in HHV-6 were elevated [48, 58]. These investigations found an extremely high proportion of patients with MS, who were seropositive for anti-HHV-6 IgG, which was remarkably higher than that of controls [48]. A novel multiplex serological assay was used to measure IgG reactivity against the immediate-early protein 1 of HHV-6A (IE1A) and HHV-6B (IE1B) in cohorts of MS. Anti-IE1A IgG responses were positively related to MS, and there was an interaction between IE1A and EBV antibody responses on the risk of MS [61]. Human Endogenous Retrovirus MS has been linked to the existence and activation of three HERVs, including HERV H, HERV K, and HERV W, which were inserted into the human genome many years ago [12, 65]. Gag (matrix and retroviral nucleus), Pol (reverse transcriptase), Pro (integrase), and Env are the four viral proteins encoded by a HERV’s genome, which is identical to those of exogenous retroviruses’ genomes [66]. Most HERVs are found in heterochromatin and are silenced through epigenetic processes such as germinal centers methylation, histone changes, and RNA silencing. HERV expression has been linked to some physiological functions thus far [8]. The presence of HERV-W, HERV-K, and HERV-H in the embryonic brain suggests that HERV may play an essential function in brain evolution and the progression of brain injury [8, 67]. In addition, high-level expression of the HERV family such as HERV-W and HERV-K has been indicated in many diseases, such as cancers and autoimmune diseases [8]. A link between HERVs and MS has been established [66] as HERV-K18 Env expression is higher in MS patients [8]. The pathogenic coat of the HERV-W Env protein originally termed “MSRV- Env,” has been detected in the brain, serum, perivascular infiltration, and infiltrating macrophages of MS diseases [8, 66, 68], Proposing a role of HERVW/MSRV as a biomarker for the course and treatment outcome of MS [68]. Microglia activated by recombinant HERV-W Env protein can cause myelinated axon injury, implying that pHERV-W Env may function in MS neuron destructions [8, 68]. Although it is unknown whether HERV-W is to blame for the onset of multiple sclerosis, evidence suggests that it may affect the immune system. HERV-W Env has been linked to inflammatory processes and is related to active demyelination locations, and is primarily expressed in macrophages and microglia [8, 68]. Activation of the Immune System by HERV in MS In MS, some HERV products are overexpressed, which triggers an innate immune response by stimulating IFN-I and III responses. The severity of MS appears to be associated with the p40 subunits of IL-12 and IL-6 [8]. The expression of human inducible nitric oxide synthase (hiNOS) and the promoter activity of hiNOS can be increased by HERV-W Env. Nitric oxide (NO) encompasses a dual function in the CNS. By some mechanisms, NO can demyelinate or destroy oligodendrocytes, disrupting the structure of the BBB and increasing its permeability, boosting neuronal apoptosis or necrosis, and effects on damaging the axons [8, 68]. Dendritic cells’ ability to develop Th1-like effector T lymphocytes and their functional or phenotypic maturation can be induced by HERV-W Env. Expressing the HERV-W Env epitope by active B cells and monocytes in MS patients can demonstrate cross-reactivity toward myelin proteins via molecular mimicry events [8, 68]. The MSRV Env’s potential immunopathogenic and pro-inflammatory effects appear to be associated with the activation of TLR4 and its co-receptor CD14, which is expressed on endothelial and monocyte cells and lead to the formation of inflammatory cytokines, IL-6, IL1β, and TNF-α [69]. The expression induction of ICAM-1 happens by TLR4 activation on endothelial cells, which recruits T-cells from the bloodstream toward the CNS. The Env protein interacts with regions of the T-cell receptor-independent antigen-binding site. It can activate many clones regardless of the antigen dedicated after T-cells enter the CNS. The Env protein, a superantigen, can be a mediator in a cycle and leads to out-of-control autoreactive cell expansion and major secreting of pro-inflammatory cytokines in the CNS with its pro-inflammatory effects on microglia [66]. In addition, in MS patients, the Env protein can be co-localized with normal-appearing white matter oligodendrocyte progenitor cells (OPCs). The production of inflammatory cytokines and inducible nitric oxide synthase (iNOS), which in turn affects the expression of myelin proteins and also causes groups of the nitrotyrosine (superoxide) to form, is determined by the activation of TLR4 expressed in OPC. These adverse effects of the HERV-W env protein on OPC may obstruct myelin repair, result in remyelination abnormalities, and development of MS [70]. Laboratory Diagnostic Markers of HERV in MS The perceived association between HERV and autoimmune disease hangs primarily on detecting retroviral antigens at the disease site or the presence of the examination case’s serum antiretroviral antibodies [8]. In an investigation, expressing several gag genes of HERV-K was significantly higher in PBMCs and brain cells from patients suffering from MS. In addition to HERV-K and HERV-W, HERV-H Env and gag increased expression in PBMCs and serum of patients with MS [8]. Elevated expression of HERV-H Env protein was found in monocytes and B cells of patients with active MS compared with inactive MS patients or healthy controls [8]. EBV can activate HERV-W in infectious mononucleosis patients and healthy individuals with an elevated titer of EBNA-1 [8]. Medication Control of HERV These findings indicate that HERV activation may come up with the progression of MS triggering the demyelination process. HERV may be activated by several simulations, consisting of viruses such as VZV, HSV-1, EBV, and HHV-6 [8, 12]. It has been reported that treatment may also affect the expression of HERV. Rituximab can play a role in downregulating the expression of HERV by depleting B cells that co-express proteins of EBV and HERV [8, 68]. Several findings have shown that a considerable decrease may happen in viremia in interferon-beta, natalizumab, and fingolimod-treated patients [68]. Stimulation of MS by Cytomegalovirus CMV is another human herpes virus implicated in several autoimmune diseases. In most cases, immune-competent individuals infected with CMV have few or no symptoms [1]. In experimental autoimmune encephalomyelitis (EAE), the proportion of peripheral CD4 + CD28 null T-cells is correlated with the severity of the illness [6]. This suggests that peripherally enlarged CD4 + CD28 null T-cells can affect the CNS by migrating to the CNS. In summary, it has been demonstrated that CMV can increase the proliferation of CD4 + CD28 null T-cells, which in turn encourages the escalation of autoimmune-mediated inflammation, demyelination, and activation of disease-specific CD4 + T-cells [6]. CD28 is seen on the surface of naive T-cells, but the loss of CD28 expression can be caused by repeated antigen stimulations. Within chronic activation of the immune system in a subgroup of healthy control subjects and MS patients, CD4 + CD28 null memory T-cells can occur [6]. Memory T-cells specific to CMV may accumulate significantly (on average 10% of total T memory cell compartments) due to CMV persistent quiddity. Due to this high percentage of CMV-specific T cells, immunological surveillance may become noticeably weaker, impairing normal immunity [71]. A linear relationship with the severity of disease in MS has not yet been defined. However, indirect evidence, such as the ability to penetrate target tissues and cytotoxic activity on oligodendrocytes, suggests this hypothesis [6]. Laboratory Diagnostic Markers of CMV in MS A study exposed that the mean value of anti-cytomegalovirus IgG antibody in MS patients’ blood was not only increased but also statistically significant. The mean value of the anti-cytomegalovirus IgM antibody in the MS patients’ blood increased but was not statistically meaningful [72]. They found that antibody levels to CMV were higher in MS patients (98%) compared to controls (52%) and were statistically significant [72]. Our results also demonstrated the role of CMV in the elevation of autoimmune symptoms in MS patients. The acceleration in IgG and IgM antibody titers against CMV in the MS patients’ sera was statistically significant [72], suggesting a role that this virus may affect autoimmune diseases [72]. The enrichment of CMV-specific antibodies in MS is the most crucial indicator of the disease promoting state. Within those patients with MS where antibodies had been found, this became related to a reduced relapse time, a boom within the wide variety of relapses, and more advantageous mind atrophy [6]. When comparing the serosensitivity of two important antigens, VZV-IgG and CMV-IgG, between controls and patients with MS, we noted that the control group was remarkably more likely to be less favorable for both antibodies compared with the patients suffering from MS [11]. Stimulation of MS by VZV VZV has been shown to be the most common component of multi-specific humoral responses in the spinal cavity of patients who suffer from MS, which helps diagnose MS. Additionally, a higher risk of MS diagnosis can be associated with VZV infection [12]. After primary infection, VZV can remain dormant in the sensory ganglia. The virus can be reactivated during immunosuppression [48]. The mechanisms by which CMV or VZV may affect the risk of MS are unclear but may arise from an immune response to proteins of the virus or alter the local cytokine milieu by a nonspecific third-party immune response [11]. Laboratory Diagnostic Markers of VZV in MS In an investigation, anti-VZV IgG seropositivity in patients with MS was slightly elevated than in controls, demonstrating VZV DNA’s short-term presence in mononuclear cells during relapse. A higher risk of developing MS a year after reactivation of VZV was found in the Taiwanese population than in the control group [48]. Some studies have shown that VZV load increases during relapse and decreases during remission of the disease [11]. SARS-CoV-2 Role in MS SARS-CoV-2 is the seventh human coronavirus known as a positive sense non-segmented RNA virus [73–75]. Detecting coronaviruses in the CNS of patients with Alzheimer’s disease (AD), Parkinson’s disease (PD), and MS is prominent [14, 76–78]. A possible explanation is that infection occurs as spikes in viral glycoproteins bind to the angiotensin-converting enzyme 2 (ACE2) receptor, which is widespread in the brain [76, 79]. PARP9 (poly ADP-ribose polymerase 9) and PARP14 (poly ADP-ribose polymerase 14) play vital roles in eukaryotic physiology and are actively involved in developing COVID-19. Under normal physiological conditions, the function of the PARP family proteins is mainly unclear. However, both proteins play essential roles in IFN-mediated host antiviral advocacy and DNA repair [80–82]. Notably, PARP9 and PARP14 have two opposing positions in IFN-γ-induced macrophage activation, where PARP9 promotes the IFN-γ response and PARP14 inhibits it by preventing STAT1 phosphorylation. On the Open Targets Platform, PARP14 and PARP9 have shown powerful associations with a wide range of human diseases spanning multiple organ systems, many of which have an autoimmune component in their etiology (e.g., MS) [80]. In contrast to PARP14 and PARP9 proteins, lymphocytes can express two TCRs that promiscuously interact with two or more molecular mimics, thereby increasing the potential for self-oriented immunopathology and further epitope diffusion [80, 83]. MS is an interesting disease for several reasons. First, the disease itself has an immunological nature. And then, disease-modifying therapy (DMT) clinical management may alter immune function and increase susceptibility to COVID-19 [76]. People with MS treated with DMT usually have a higher risk of infection, with rituximab having the most significant severe infection incidence [84]. However, COVID-19 infection severity was not associated with the presence of DMT, which was related to a lower risk of hospitalization in univariate analysis [85, 86]. Additionally, neurologists worldwide face the daunting task of stratifying the viral infection risk, especially in MS patients receiving immunosuppressive or immunomodulatory therapies. Although it has been documented that people with MS may, in theory, have a higher risk of infection than the general population, it is still being discussed whether patients with MS are at a higher risk of contracting COVID-19 from infection with SARS-CoV-2 [84]. In particular, some thoughts are required to examine the most recent data showing a lack of relationship with DMT exposure and a significant association between the Extended Disability Status Scale (EDSS) and the severity of COVID-19 and age [85]. The highest degree of heterogeneity in severe COVID-19 outcomes has been linked to EDSS, according to reports [14, 76]. The statement that morbidity and death in COVID-19 may be caused by an overlapping immunological response generated by the virus and the subject’s immune condition can be a sensible explanation. Both adaptive and innate immune responses are critical to preventing infection by the virus. Viral infections can be prevented by innate immunity with inhibition of natural killer (NK) cells and IFN-I as resistance is achieved within an adaptive immune response to immunity induced by T lymphocytes and antibodies, mainly CD8 + T-cells [14, 87, 88]. Therefore, the infection of COVID-19 triggering further amplification of immunity pathways in patients with pre-existing immunocompromised paths like MS can be speculated. Second, COVID-19 infection and age. Age has been reputed to be related to the highest variation in severe COVID-19 outcomes [85]. Third, DMT and the disease of COVID-19. Are these drugs detrimental or protective? In theory, DMT limits the immune response, allowing more replication of the virus and potentially more severe infections. On the other hand, with the aid of limiting the exaggerated immune reaction and cytokine storm due to illness by SARS-CoV-2, those drugs can also have a few defensive and valuable results in opposition to this new virus. Additionally, most of the DMTs don’t specifically interact with the innate immune system, and few of them have serious, long-lasting effects on CD8 + T-cells, which limits protection against COVID-19 [89]. The long-lasting or acute effects of COVID-19 on disease phases in patients with MS should be the subject of future research. This can also be a resilience time; This catastrophic pandemic may be an extraordinary chance for databases to ease collaboration and investigation in the MS field [90]. Although there is an association between neurodegeneration and neuroinflammation in the MS brain, there is currently insufficient evidence that SARS-CoV-2 may have a possible role in these patients’ future neurodegeneration [76]. Conclusions The significance of viral infections in patients suffering from MS is yet unknown, and the potential for many viruses to be involved in the pathogenesis of MS has to be considered. Furthermore, because of MS heterogeneity, the interaction between viruses and other illnesses, as well as environmental and genetic variables, may vary significantly. The presence of viral components in lesions of MS or an antiviral immune response in patients suffering from MS with clinical relapses strongly suggests that viruses are involved in disease progression, potentially as stimuli or co-factors. The majority of licensed MS disease-modifying medications currently have an indirect or direct effect on memory B cells and also memory T-cells, whose cooperation is expected to be seriously involved in disease pathogenesis. Clinical trials utilizing a B cell-depleting antibody have recently backed up the B cell role in MS. Ocrelizumab is a game-changing therapy for MS, and its ability to reduce disease activity and CNS damage as long as it preferentially targets B cells is consistent with the pathophysiologic role(s) of EBV-infected B cells in MS patients. Cell-based treatments that target specific groups of B cells, such as EBV-infected ones, might be a novel way to do this, with a suggestion for treating relapsing and also progressive types of MS disease, as well as perhaps hindering the disease from developing. The frequency and specificity of these viruses for humans, however, make studying plausible pathways challenging. As a result, the development of novel MS models in infected animals is exceptionally encouraging, and it enables an essential tool for defining the viral involvement in MS. Acknowledgements The authors would like to thank the Department of Medical Biotechnology, Semnan University of Medical Sciences for supporting this project. Author Contributions SS, OG, SY, SA, PA, PF: Writing—original draft, HA, AF-A, DP, ME, BY: Writing—review & editing. VP and MD: Conceptualization, Supervision, Writing—review & editing. All authors participated in the manuscript in the critical review process of the manuscript and approved the final version. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or nonprofit sectors. Data Availability Not applicable. Declarations Conflict of interest The authors declare that they have no competing interests. Ethical Approval Not applicable. Consent to Participate Not applicable. Consent for Publication Not applicable. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Hussein HM Rahal EA The role of viral infections in the development of autoimmune diseases 2019 NY Taylor & Francis 394 412 2. Olsson T Barcellos LF Alfredsson L Interactions between genetic, lifestyle and environmental risk factors for multiple sclerosis Nat Rev Neurol 2017 13 1 25 36 10.1038/nrneurol.2016.187 27934854 3. Leitzen E Comparison of reported spinal cord lesions in progressive multiple sclerosis with theiler’s murine encephalomyelitis virus induced demyelinating disease Int J Mol Sci 2019 20 4 989 10.3390/ijms20040989 30823515 4. Mechelli R Viruses and neuroinflammation in multiple sclerosis Neuroimmunology and Neuroinflammation 2021 8 269 283 5. Houen G Trier NH Frederiksen JL Epstein-Barr virus and multiple sclerosis Front Immunol 2020 2020 3315 6. Vanheusden M Cytomegalovirus infection exacerbates autoimmune mediated neuroinflammation Sci Rep 2017 7 1 1 11 10.1038/s41598-017-00645-3 28127051 7. Cheng Y Skinner DD Lane TE Innate immune responses and Viral-Induced neurologic disease J Clin Med 2018 8 1 3 10.3390/jcm8010003 30577473 8. Latifi T Zebardast A Marashi SM The role of human endogenous retroviruses (HERVs) in multiple sclerosis and the plausible interplay between HERVs, Epstein-Barr virus infection, and vitamin D Mult Scler Relat Disorders 2021 57 103318 103318 10.1016/j.msard.2021.103318 9. Bar-Or A Epstein-Barr virus in multiple sclerosis: theory and emerging immunotherapies Trends Mol Med 2020 26 3 296 310 10.1016/j.molmed.2019.11.003 31862243 10. Marrodan M The role of infections in multiple sclerosis Mult Scler J 2019 25 7 891 901 10.1177/1352458518823940 11. Karampoor S Cytomegalovirus and varicella zoster virus seropositivity of Iranian patients with multiple sclerosis: a population-based study J Neuroimmunol 2017 309 4 6 10.1016/j.jneuroim.2017.04.004 28601285 12. Donati D Viral infections and multiple sclerosis Drug Discov Today Dis Model 2020 32 27 33 10.1016/j.ddmod.2020.02.003 13. Gholizadeh O Recent advances in treatment Crimean-Congo hemorrhagic fever virus: a concise overview Microb Pathog 2022 8 105657 10.1016/j.micpath.2022.105657 14. Freire-de-Lima L Autoimmune disorders & COVID-19 Medicines 2021 8 10 55 55 10.3390/medicines8100055 34677484 15. Keikha M HCV genotypes and their determinative role in hepatitis C treatment Virusdisease 2020 31 3 235 240 10.1007/s13337-020-00592-0 32904762 16. Hashemi B Emerging importance of nanotechnology-based approaches to control the COVID-19 pandemic; focus on nanomedicine iterance in diagnosis and treatment of COVID-19 patients Journal of Drug Delivery Science and Technology 2021 67 102967 10.1016/j.jddst.2021.102967 34777586 17. Eslami M Dietary pattern, colonic microbiota and immunometabolism interaction: new frontiers for diabetes mellitus and related disorders Diabet Med 2021 38 2 e14415 10.1111/dme.14415 33025587 18. Leibovitch EC Jacobson S Viruses in chronic progressive neurologic disease Mult Scler J 2018 24 1 48 52 10.1177/1352458517737392 19. Xu L Positive association of herpes simplex virus-IgG with multiple sclerosis: A systematic review and meta-analysis Mult Scler Relat Disorders 2021 47 102633 10.1016/j.msard.2020.102633 20. Del Valle L White MK Khalili K Potential mechanisms of the human polyomavirus JC in neural oncogenesis J Neuropathol Exp Neurol 2008 67 8 729 740 10.1097/NEN.0b013e318180e631 18648329 21. Meneguzzi G Minichromosome from BK virus as a template for transcription in vitro Proc Natl Acad Sci 1978 75 3 1126 1130 10.1073/pnas.75.3.1126 26051 22. Saribaş AS JC virus-induced progressive multifocal leukoencephalopathy Futur Virol 2010 5 3 313 323 10.2217/fvl.10.12 23. Jouanguy E Human inborn errors of immunity to herpes viruses Curr Opin Immunol 2020 62 106 122 10.1016/j.coi.2020.01.004 32014647 24. Majerciak V A genome-wide Epstein-Barr virus polyadenylation map and its antisense RNA to EBNA J Virol 2019 93 2 e01593 e1618 10.1128/JVI.01593-18 30355690 25. Soldan SS Epigenetic plasticity enables CNS-trafficking of EBV-infected B lymphocytes PLoS Pathog 2021 17 6 e1009618 10.1371/journal.ppat.1009618 34106998 26. Keane JT The interaction of Epstein-Barr virus encoded transcription factor EBNA2 with multiple sclerosis risk loci is dependent on the risk genotype EBioMedicine 2021 71 103572 103572 10.1016/j.ebiom.2021.103572 34488019 27. Endriz J Ho PP Steinman L Time correlation between mononucleosis and initial symptoms of MS Neurol-Neuroimmunol Neuroinflam 2017 4 3 e308 10.1212/NXI.0000000000000308 28. Pender MP Defective T-cell control of Epstein-Barr virus infection in multiple sclerosis Clin Trans Immunol 2017 6 1 e126 10.1038/cti.2016.87 29. Hammerschmidt W The epigenetic life cycle of Epstein-Barr virus Epstein Barr Virus 2015 1 103 117 10.1007/978-3-319-22822-8_6 30. Cencioni MT Programmed death 1 is highly expressed on CD 8+ CD 57+ T cells in patients with stable multiple sclerosis and inhibits their cytotoxic response to Epstein-Barr virus Immunology 2017 152 4 660 676 10.1111/imm.12808 28767147 31. Latour S Fischer A Signaling pathways involved in the T-cell-mediated immunity against Epstein-Barr virus: Lessons from genetic diseases Immunol Rev 2019 291 174 189 10.1111/imr.12791 31402499 32. Pender MP Decreased T cell reactivity to Epstein-Barr virus infected lymphoblastoid cell lines in multiple sclerosis J Neurol Neurosurg Psychiatry 2009 80 5 498 505 10.1136/jnnp.2008.161018 19015225 33. Serafini B Epstein-Barr virus-specific CD8 T cells selectively infiltrate the brain in multiple sclerosis and interact locally with virus-infected cells: clue for a virus-driven immunopathological mechanism J Virol 2019 93 24 e00980 e1019 10.1128/JVI.00980-19 31578295 34. Albanese M Epstein-Barr virus microRNAs reduce immune surveillance by virus-specific CD8+ T cells Proc Natl Acad Sci 2016 113 42 E6467 E6475 10.1073/pnas.1605884113 27698133 35. Jangra S Epstein-Barr virus and innate immunity: Friends or foes? Microorganisms 2019 7 6 183 10.3390/microorganisms7060183 31238570 36. Jelcic I Memory B cells activate brain-homing, autoreactive CD4+ T cells in multiple sclerosis Cell 2018 175 1 85 100 10.1016/j.cell.2018.08.011 30173916 37. Winter S Loss of RASGRP 1 in humans impairs T-cell expansion leading to Epstein-Barr virus susceptibility EMBO Mol Med 2018 10 2 188 199 10.15252/emmm.201708292 29282224 38. Wanke F EBI2 is highly expressed in multiple sclerosis lesions and promotes early CNS migration of encephalitogenic CD4 T cells Cell Rep 2017 18 5 1270 1284 10.1016/j.celrep.2017.01.020 28147280 39. Rutkowska A EBI2 regulates pro-inflammatory signalling and cytokine release in astrocytes Neuropharmacology 2018 133 121 128 10.1016/j.neuropharm.2018.01.029 29374507 40. Clottu AS EBI2 expression and function: robust in memory lymphocytes and increased by natalizumab in multiple sclerosis Cell Rep 2017 18 1 213 224 10.1016/j.celrep.2016.12.006 28052250 41. Liu X Human virus transcriptional regulators Cell 2020 182 24 37 10.1016/j.cell.2020.06.023 32649876 42. Afrasiabi A Evidence from genome wide association studies implicates reduced control of Epstein-Barr virus infection in multiple sclerosis susceptibility Genome medicine 2019 11 1 1 13 10.1186/s13073-019-0640-z 30609936 43. Glaser LV EBF1 binds to EBNA2 and promotes the assembly of EBNA2 chromatin complexes in B cells PLoS Pathog 2017 13 10 e1006664 10.1371/journal.ppat.1006664 28968461 44. Simonsen CS The diagnostic value of IgG index versus oligoclonal bands in cerebrospinal fluid of patients with multiple sclerosis Mult Scler J-Exp Trans Clin 2020 6 1 2055217319901291 2055217319901291 45. Makhani N Oligoclonal bands increase the specificity of MRI criteria to predict multiple sclerosis in children with radiologically isolated syndrome Mult Scler J-Exp Trans Clin 2019 5 1 2055217319836664 46. Arrambide G The value of oligoclonal bands in the multiple sclerosis diagnostic criteria Brain 2018 141 4 1075 1084 10.1093/brain/awy006 29462277 47. Hassani A Epstein-Barr virus is present in the brain of most cases of multiple sclerosis and may engage more than just B cells PLoS ONE 2018 13 2 e0192109 10.1371/journal.pone.0192109 29394264 48. Czarnowska A Herpesviridae seropositivity in patients with multiple sclerosis: first polish study Eur Neurol 2018 80 5–6 229 235 10.1159/000496402 30661064 49. Moreno MA Molecular signature of Epstein-Barr virus infection in MS brain lesions Neurol-Neuroimmunol Neuroinflamm 2018 5 4 e466 10.1212/NXI.0000000000000466 29892607 50. Maple PA A different response to cytomegalovirus (CMV) and Epstein-Barr virus (EBV) infection in UK people with multiple sclerosis (PwMS) compared to controls J Infect 2020 80 3 320 325 10.1016/j.jinf.2019.10.017 31678173 51. Veroni C Transcriptional profile and Epstein-Barr virus infection status of laser-cut immune infiltrates from the brain of patients with progressive multiple sclerosis J Neuroinflammation 2018 15 1 1 19 10.1186/s12974-017-1049-5 29301548 52. Jakimovski D Interferon β for multiple sclerosis Cold Spring Harb Perspect Med 2018 8 11 a032003 10.1101/cshperspect.a032003 29311124 53. Zivadinov R Teriflunomide’s effect on humoral response to Epstein-Barr virus and development of cortical gray matter pathology in multiple sclerosis Mult Scler Relat Disorders 2019 36 101388 10.1016/j.msard.2019.101388 54. Ogembo JG A chimeric EBV gp350/220-based VLP replicates the virion B-cell attachment mechanism and elicits long-lasting neutralizing antibodies in mice J Transl Med 2015 13 1 1 15 10.1186/s12967-015-0415-2 25591711 55. Smith C Khanna R The development of prophylactic and therapeutic EBV vaccines Epstein Barr Virus 2015 2 455 473 10.1007/978-3-319-22834-1_16 56. Eslami M Mycobacterium avium paratuberculosis and Mycobacterium avium complex and related subspecies as causative agents of zoonotic and occupational diseases J Cell Physiol 2019 234 8 12415 12421 10.1002/jcp.28076 30673126 57. Mulero P Midaglia L Montalban X Ocrelizumab: a new milestone in multiple sclerosis therapy Ther Adv Neurol Disord 2018 11 1756286418773025 10.1177/1756286418773025 29774057 58. Keyvani H The role of human herpesvirus-6 and inflammatory markers in the pathogenesis of multiple sclerosis J Neuroimmunol 2020 346 577313 10.1016/j.jneuroim.2020.577313 32673896 59. Dunn N Kharlamova N Fogdell-Hahn A The role of herpesvirus 6A and 6B in multiple sclerosis and epilepsy Scand J Immunol 2020 92 6 e12984 10.1111/sji.12984 33037649 60. Greninger AL Comparative genomic, transcriptomic, and proteomic reannotation of human herpesvirus 6 BMC Genomics 2018 19 1 1 17 10.1186/s12864-018-4604-2 29291715 61. Engdahl E Increased serological response against human herpesvirus 6A is associated with risk for multiple sclerosis Front Immunol 2019 10.3389/fimmu.2019.02715 32038605 62. Tang H CD134 is a cellular receptor specific for human herpesvirus-6B entry Proc Natl Acad Sci 2013 110 22 9096 9099 10.1073/pnas.1305187110 23674671 63. Fierz W Multiple sclerosis: an example of pathogenic viral interaction? Virology journal 2017 14 1 1 5 10.1186/s12985-017-0719-3 28081705 64. Fotheringham J Human herpesvirus 6 (HHV-6) induces dysregulation of glutamate uptake and transporter expression in astrocytes J Neuroimmune Pharmacol 2008 3 2 105 116 10.1007/s11481-007-9084-0 18247129 65. Morandi E Human endogenous retroviruses and multiple sclerosis: causation, association, or after-effect? Mult Scler J 2017 23 8 1050 1055 10.1177/1352458517704711 66. Küry P Human endogenous retroviruses in neurological diseases Trends Mol Med 2018 24 4 379 394 10.1016/j.molmed.2018.02.007 29551251 67. Wang X Huang J Zhu F Human endogenous retroviral envelope protein syncytin-1 and inflammatory abnormalities in neuropsychological diseases Front Psych 2018 9 422 10.3389/fpsyt.2018.00422 68. Dolei A The aliens inside us: HERV-W endogenous retroviruses and multiple sclerosis Mult Scler J 2018 24 1 42 47 10.1177/1352458517737370 69. Duperray A Inflammatory response of endothelial cells to a human endogenous retrovirus associated with multiple sclerosis is mediated by TLR4 Int Immunol 2015 27 11 545 553 10.1093/intimm/dxv025 25957268 70. Madeira A MSRV envelope protein is a potent, endogenous and pathogenic agonist of human toll-like receptor 4: relevance of GNbAC1 in multiple sclerosis treatment J Neuroimmunol 2016 291 29 38 10.1016/j.jneuroim.2015.12.006 26857492 71. Hanley PJ Bollard CM Controlling cytomegalovirus: helping the immune system take the lead Viruses 2014 6 6 2242 2258 10.3390/v6062242 24872114 72. Salim MA Determining the IgM and IgG antibody titer against CMV and helicobacter pylori in the serum of multiple sclerosis patients comparing to the control group in Hamadan Hum Antibodies 2017 26 1 23 28 10.3233/HAB-170317 28505965 73. Lippi A SARS-CoV-2: at the crossroad between aging and neurodegeneration Mov Disord 2020 35 5 716 10.1002/mds.28084 32291797 74. Yasamineh S A state-of-the-art review on the recent advances of niosomes as a targeted drug delivery system Int J Pharm 2022 624 121878 10.1016/j.ijpharm.2022.121878 35636629 75. Yousefi B A global treatments for coronaviruses including COVID-19 J Cell Physiol 2020 235 12 9133 9142 10.1002/jcp.29785 32394467 76. Ferini-Strambi L Salsone M COVID-19 and neurological disorders: are neurodegenerative or neuroimmunological diseases more vulnerable? J Neurol 2021 268 2 409 419 10.1007/s00415-020-10070-8 32696341 77. Matías-Guiu J Should we expect neurological symptoms in the SARS-CoV-2 epidemic? Neurología (English Edition) 2020 35 3 170 175 10.1016/j.nrleng.2020.03.002 78. Yasamineh S Spotlight on therapeutic efficiency of mesenchymal stem cells in viral infections with a focus on COVID-19 Stem Cell Res Ther 2022 13 1 257 10.1186/s13287-022-02944-7 35715852 79. Rodriguez-Perez AI Angiotensin type 2 receptors: Role in aging and neuroinflammation in the substantia nigra Brain Behav Immun 2020 87 256 271 10.1016/j.bbi.2019.12.011 31863823 80. Kovačić D Jotanović J Laković J The possible role of molecular mimicry in SARS-CoV-2-mediated autoimmunity: an immunobiochemical basis J Med Sci 2021 90 3 e560 e560 10.20883/medical.e560 81. Yousefi B Eslami M Genetic and structure of novel coronavirus COVID-19 and molecular mechanisms in the pathogenicity of coronaviruses Rev Med Microb 2022 33 1 e180 e188 10.1097/MRM.0000000000000265 82. Yousefi B Potential therapeutic effect of oxygen-ozone in controlling of COVID-19 disease Med Gas Res 2022 12 2 33 34677149 83. Dardalhon V Role of Th1 and Th17 cells in organ-specific autoimmunity J Autoimmun 2008 31 3 252 256 10.1016/j.jaut.2008.04.017 18502610 84. Willis M Robertson N Multiple sclerosis and the risk of infection: considerations in the threat of the novel coronavirus, COVID-19/SARS-CoV-2 J Neurol 2020 267 5 1567 1569 10.1007/s00415-020-09822-3 32303837 85. Louapre C Clinical characteristics and outcomes in patients with coronavirus disease 2019 and multiple sclerosis JAMA Neurol 2020 77 9 1079 1088 10.1001/jamaneurol.2020.2581 32589189 86. Yasamineh S An overview on nanoparticle-based strategies to fight viral infections with a focus on COVID-19 Journal of Nanobiotechnology 2022 20 1 440 10.1186/s12951-022-01625-0 36209089 87. Berger JR Brandstadter R Bar-Or A COVID-19 and MS disease-modifying therapies Neurol-Neuroimmunol Neuroinflamm 2020 7 4 e761 10.1212/NXI.0000000000000761 32414755 88. Fard NG COVID-19 coinfection in patients with active tuberculosis: first case-report in Iran 2021 Authorea Preprints 89. Baker D The underpinning biology relating to multiple sclerosis disease modifying treatments during the COVID-19 pandemic Mult Scler Relat Disorders 2020 43 102174 10.1016/j.msard.2020.102174 90. Racke MK Newsome SD Multiple sclerosis and COVID-19: a great opportunity for databases promoting research and collaboration J Neuroimmunol 2020 345 577283 10.1016/j.jneuroim.2020.577283 32518058	36459252	PMC9716500	NO-CC CODE	2022-12-03 23:21:04	no	Curr Microbiol. 2023 Dec 2; 80(1):15	utf-8	Curr Microbiol	2,022	10.1007/s00284-022-03112-z	oa_other
==== Front Asia-Pac J Reg Sci Asia-Pacific Journal of Regional Science 2509-7946 2509-7954 Springer Nature Singapore Singapore 269 10.1007/s41685-022-00269-0 Article Does financial inclusion control corruption in upper-middle and lower-middle income countries? Barik Rajesh [email protected] 1 http://orcid.org/0000-0001-9353-6872 Lenka Sanjaya Kumar [email protected] 2 1 grid.464901.e 0000 0004 1772 1939 Department of Economics, ICFAI Business School, Hyderabad, Telangana 501203 India 2 grid.440672.3 0000 0004 1761 0390 Department of Economics, School of Social Sciences, CHRIST (Deemed to be University), Bannerghatta Road Campus, Bengaluru, Karnataka 560076 India 2 12 2022 124 4 6 2022 15 11 2022 © The Japan Section of the Regional Science Association International 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Presence of corruption in a system is always a path breaker for transparent distribution of public services in the economy. Therefore, controlling corruption is a high priority for progress of a country’s growth. The main objective of this study was to empirically examine the impacts of financial inclusion on control of corruption in selected upper-middle and lower-middle income countries. Using cross-country annual data from 2004 to 2018, the study applied fixed effect, random effect, panel corrected standard errors, feasible general least square and 2SLS (two-stage least-squares regression) models to evaluate the impacts of financial inclusion on control of corruption across all samples from upper-middle and lower-middle income countries. The results from the upper-middle income (UMI) countries demonstrated that a basic level of financial inclusion has no impact on the control of corruption, whereas higher intensification of financial inclusion beyond the basic level positively impacts it. Similarly, the findings from lower-middle-income (LMI) countries indicated that financial inclusion up to a certain threshold level helps to control corruption, whereas financial inclusion above the threshold level negatively impacts the control of corruption. These empirical findings suggest that in the overall sample, financial inclusion plays an important role to control corruption. Keywords Financial inclusion Corruption Upper-middle-income Lower-middle-income Cross-country ==== Body pmcIntroduction Financial inclusion is defined as the easy and affordable access of financial services by the poor and marginalized people (Asian Development Bank 2000; Beck and Demirgüç-Kunt 2008; Rangarajan 2008; Reserve Bank of India 2014). In recent years, the magnitude of financial inclusion has enhanced significantly in both developed and developing countries (Mader and Duvendack 2019). It has been observed that over the years, the accessibility and usability of formal financial products and services among the poor and marginalized sections of people are increasing (Hussain et al. 2021). Like high-income countries, adults from middle-income and low-income countries are also frequently using different modes of formal financial services in their day-to-day lifecycle. The recent adaptability of digital payment technologies across all sections of people has drastically augmented the overall financial transactions. Specifically, the rapid use digital technologies during the COVID-19 pandemic are well noticed around the globe (see Sahay et al. 2020; Sornaganesh et al. 2020; Ozili 2020a, b). Hence, understanding the rising trend of digital technologies and the high accessibility of financial services among the mass sections of people, it is always a matter of inquisitiveness to demystify the fact how has the wide range of accessibility of basic finance impact corruption practices in the countries. Corruption is always defined as the misuse of public power for the individual benefits. Hence, the corruptive behavior of the public authority has significant impact on the other socio-economic outcomes. Control of corruption has always been a key factor for the reduction of poverty, inequality (Gupta, et al. 2002), and unemployment (Adjor and Kebalo 2018; Abé Ndjié et al. 2019), and the overall economic development. Hence, comprehending the significance of control of corruption and observing the high accessibility of financial services among all sections of people, a debate has been erupted among the world academicians and policy makers for bringing a clear-cut evidence on the inter-relationship between financial inclusion and corruption. Some economists (like Raghuram Rajan) have argued in favor of financial inclusion and control of corruption. While delivering 20th Lalit Doshi memorial lecture in Mumbai (India), prof. Rajan emphasized that the promotion of financial inclusion and digital finance [like internet/mobile banking, direct benefit transfer (DBT), linking unique biometric identification with a bank account, etc.] can break the corrupt practices in public services. Conversely, a counter argument has been put forward that the high degree of financial inclusion with less financial literacy/awareness, no supervision, and inadequate regulation can enable people to access huge credit and allow them to invest their obtained credit in different unauthorized or illegal activities like insider trading, money laundering, gambling (Ajide 2020), which ultimately enhance the process of corruption in the country. Understanding the theoretical arguments from both sides of financial inclusion on corruption, this study has endeavored to empirically examine the impact of financial inclusion in controlling corruption between the selected upper-middle- and lower-middle-income countries in the world. While, looking the World Bank classification of countries on the basis of country’s income, it can be understood that unlike low- and higher-income countries, upper-middle- and lower-middle-income countries have less income gap1 in between them. Hence, it would always be a noteworthy attempt to study how the process of financial inclusion is responding differently to two almost similar kinds of income groups’ countries. This study contributes to the existing literatures of financial inclusion in the following ways. To the best of authors’ knowledge, Ajide (2020) is the only empirical study which is depicting the role of financial inclusion in controlling corruption in 13 African countries. Hence, this is the first study which is empirically examining the same relationship in a cross-country level and trying to understand how the process of financial inclusion behaving differently in a cross-country level. Second, unlike Ajide (2020), this study uses principal component analysis (PCA) to construct financial inclusion indexes for both the upper-middle- and lower-middle-income countries. Third, this study examines the non-monotonic relationship between financial inclusion and control of corruption for both the groups of countries. Fourth, identifying the nature of the considered sample countries, it can be said that these countries are having different stages of financial inclusion and its inter-relationship with corruption. Hence, along with the overall sample, this study also endeavored to examine their inter-relationship for lower-middle and upper-middle income countries separately. Fifth, this study applies various panel data models to strengthen the regression results. Finally, this study also make an attempt to provide some valuable policy suggestions to both the groups of countries about the role of financial inclusion in control of corruption. The rest of the paper is organized in the following manner. Section 2 of this paper describes the theoretical nexus between fiancé and corruption. Section 3 provides a brief review of literature. Section 4 gives a detailed explanation of the data and econometric techniques used in this study. Similarly, Sect. 5 presents the empirical findings and intensely discusses the results. Lastly, Sect. 6 delivers a concluding remark and suggests some valuable policy recommendations for the policy makers and governments. Theoretical background The theoretical foundation of corruption and finance is well described in the ‘sand the wheels’ hypothesis. This hypothesis depicts that the presence of corruption can adversely affect the financial sector development (see Cooray and Schneider 2018; Ajide 2020). The hypothesis further highlights that because of inadequate supervisory policies, large-scale regulation, insider trading, and the absence of transparency, corruption practices are enhanced, which further hampers the financial sector’s development (see Khemani and Meyerman 1998; Song et al. 2021; Weill 2011a, b; Park 2012). That means in the context of financial inclusion, it can be said that because of the problem of adverse selection and moral hazard (Barth et al. 2004; Cooray and Schneider 2018), the financial resources are diverted that leads to inefficient allocation of money. Again due to lack of transparency and supervision in the system, misallocation of financial resources raises the volume of non-performing assets which further damages the financial stability of the country. Additionally, this study can also borrow its theoretical foundation from the “rational choice theory”, which depicts that the practice of corruption happens because of the existence of asymmetric information between the agent and the principal (Svensson 2005; Kolstad and Søreide 2009; Dupuy and Neset 2018). According to the ‘rational choice theory’, the agent (i.e., government or the public authority) hold the monopoly power over the given information which allows the agent to hide information and to use that information for personal gains. With context to financial inclusion, the agents (i.e., the bank staffs) hold all the banking information and misguide the customers. Generally, in developing and under developed countries, people hold asymmetric information related to banking services because of their low financial literacy (Hussain et al. 2021). It gives an opportunity to the bank agent to hide the key information by misguiding the customers and taking bribe for providing banking services. Review of literature Much of the earlier research on corruption has focused on the nexus between corruption and economic growth. While studying the inter-relationship between corruption and growth, studies have found mixed results. Some of the studies (likeMauro and Driscoll 1997; Mo 2001; Ahmad et al. 2012; Thach et al. 2017) have found that corruption has a negative impact on economic growth. Whereas some other kind of literatures says that corruption has positive role on growth (Leff 1964; Méon and Weill 2010; Méndez and Sepúlveda 2006; Chakravorty, 2019). Similarly, while narrating the story of corruption and financial sector mismanagement, studies like Kane (1993) have claimed that there were many ways of corruption presented in the Japanese banking system which further led to a big banking crisis that happened in 1990s. Correspondingly, authors like MacFarlane (2001) have put the argument that corruption in the banking system, fraud actives, and loan-sharking by some bankers lead to the bank crisis in Japan. Concerning the Asian financial crisis, researchers like Khemani and Meyerman (1998) have argued that excess corruption, nepotism, and cronyism are the factors behind this crisis. Similar to this kind of results, some other studies (see Cooray and Schneider 2018; Song et al. 2021) have also found that corruption has a significantly negative impact on financial sector development. Similarly, other studies like Weill (2011a, b) and Park (2012) have argued that the presence of corruption in the banking sector promotes bad loans in the system. In another kind of literature (see Ahlin and Pang 2008), financial development and corruption are used for substitutability, and the author's found that corruption raises the demand for liquidity which brings improvement in the financial system and oppositely financial underdevelopment brings more corruption in the system. Diverging from these results, other studies by Thornton (2010) and Altunbaş and Thornton (2012) have argued that the development in the financial sector can reduce the corruption level. A recent study by Sharma and Paramati (2020) conducted an empirical study on 140 sample countries to know the impact of financial sector development on the control of corruption. Their empirical findings depicted that the development of the financial sector plays a significant role to control the growth of corruption across the full sample countries, low- and lower-middle-income countries, and upper-middle and high-income countries as well. Through-out our literature journey, we observed that there are some studies which have depicted the nexus of corruption with economic growth and development. Some other studies have narrated how the presence of corruption in the banking system leads to bad loan and bank crisis. Similarly, while finding the nexus between financial development and corruption, we observed that there are a handful of studies existing on the relationship between financial development and control of corruption and vice-versa. Similar to financial development, we also found that some studies are narrating the inter-relationship between corruption and financial inclusion. The studies depicting the impact of corruption on financial inclusion have found mixed results. For example, one study was conducted by Abu et al. (2015) across the West African states and the study showed that lower level of corruption is increasing the saving rate in the West African states. Similarly, Eldomiaty et al. (2020) conducted another study across the world economics to examine how control of corruption is impacting the process of financial inclusion. In that study, the authors used control of corruption as one of the prominent indicators of good governance (along with other five indicators of good governance), and the study found that control of corruption has a positive significant effect on saving at financial institution, debit and credit card provision (Sha'ban et al. 2020; Ali et al. 2022). Conversely, one recent study by Malik et al. (2022) shows that governance quality (control of corruption is considered as one indicator) is negatively impacting financial inclusion in Asian counties. From our literature review process, we realized that there are handful of studies depicting the impact of corruption on financial inclusion. However, while looking the literatures from reverse side (i.e., impact of financial inclusion on control of corruption), we found only one study (Ajide 2020) has been conducted in 13 African countries for the period of 2005–2016, to know the impact of financial inclusion on corruption control in African countries. The finding of the paper explains that financial inclusion intensifies the control of corruption. This study further discusses that there is a threshold level up to which financial inclusion will assist to control corruption and after the surpass of the threshold level, financial inclusion will have a negative impact on the control of corruption because of weak institutional factors present in African countries. Hence, from our literature journey, we found that Ajide (2020) is the only study which is showing the impact of financial inclusion on control of corruption in African countries. Looking at the dearth of cross-country evidence in the context of financial inclusion and control of corruption, this current study is an endeavor to find out the impact of financial inclusion on control of corruption considering some selected countries from both the upper-middle- and lower-middle-income groups. The findings of this study will surely help both the upper- and lower-middle-income countries to redesign their financial inclusion policies, so that it would provide more positive results in controlling corruption. Data and econometrics techniques used Variable specification and data sources This section briefly explains the number of variables used in this paper and its various data sources. To empirically examine the impact of financial inclusion on corruption, this study has collected data from 31 countries from both the upper- and lower-middle-income regions covering the time period of 2004–2018. Depending open the availability of data for all the countries and all the selected variables, this study has chosen 31 countries for its analysis. Out of these total 31 countries, this study has taken 16 countries from the upper-middle-income (UMI) groups and the rest of 15 countries are chosen from lower-middle-income (LMI) categories (see Table 5 in Appendix). For a division of countries on an income basis, this study has relied open the World Bank income-based classification of countries for the 2021 fiscal year. Corruption Country-wise corruption perception index2 (CPI) score has been collected for 31 countries from transparency international. The trend of corruption in both the upper- and lower-middle-income countries has been presented in Figs. 3 and 4 in the appendix section of this paper. For both the income groups, periodical trend (i.e., 2004, 2014, and 2018) of corruption has been presented in Figs. 3 and 4 to know the status of corruption in these countries over the years (See Figs. 3 and 4 in Appendix). Financial inclusion index Measuring complete financial inclusion is always a challenging task because it cannot be measured in a single dimension. Financial inclusion includes multiple indicators from diverse financial services. In the financial inclusion literature, different indicators have been used by different researchers as per their data availability and suitability. In this present study, we have considered six different financial inclusion indicators covering three major dimensions of financial inclusion (i.e., demographic dimension, geographic dimension, and usage dimension). In the demographic dimension, we have considered the number of bank branches and number of ATMs per 100,000 adult population. Similarly, in the geographic section, we have chosen the number of commercial bank branches and the number of ATMs per 1000 km2. Likewise, for measuring usage of financial inclusion, we have taken outstanding deposits and credit as a percentage of GDP. All these indicators are used by Lenka and Barik (2018) in their financial inclusion index calculation. All the indicators of financial inclusion are collected from the financial access survey of the International Monetary Fund (IMF). Control variables For control variables, we have used remittance receive as a percentage of county’s GDP, GDP per capita, human capital, percentage of unemployment, and the rate of inflation. Some of the control variables like inflation, remittance, and unemployment has been used in earlier studies on financial inclusion and corruption (see Ajide 2020). Data for all the control variables have been collected from World Bank indicators (Table 1).Table 1 Description of variables Variables Explanation Data sources Dependent variable Measure the perception of corruption in the public sector Transparency International Corruption perception index (CPI) Independent variable (a) Number of bank branches per 100,000 adult population (b) Number of ATMs per 100, 000 adult population (c) Number of bank branches per 1000 km2 (d) Number of ATMs per 1000 km2 (e/f) Outstanding deposit and credit as a percentage of GDP Financial Access Survey of International Monetary Fund Financial inclusion index (FI) Control variables (a) Received remittance % of GDP (b) Gross domestic product per capita (c) School enrolment ratio (d) Percentage of unemployment (e) Rate of inflation World Bank Indicators (a) REM (remittance) (b) GDPPC (GDP per capita) (c) HUM (human capital) (d) UNEMP (unemployment) (e) INF (inflation) Source: authors estimation Measuring financial inclusion index in both upper- and lower-middle-income countries Measuring a holistic and unbiased composite financial inclusion index is a challenging assignment for the researchers. Meanwhile, previous studies (Sarma 2008; Arora 2010; Gupte et al. 2012; Chakravarty and Pal 2013) have used different methods [like distance-based approach adopted by UNDP to compute HDI, analytical and hierarchical process (AHP) and axiomatic approach] to compute the index of financial inclusion. Each method has its own merits and demerits for computation of the index. For construction of multidimensional index, weights play a major role in the overall composite indicator. Most of the studies have used AHP method for calculation of weights of the variables in the composite index construction. However, the problem with AHP is that there is no prior information available about the weight of a particular variable (Lenka and Barik 2018). So, it may be messy for a researcher to assign weights for each variable looking the previous literature and expert opinions. Hence, AHP may not be a worthy method to find out the weight of factor included in the multidimensional index. In addition, looking at the volatility nature of financial access variables, AHP and distance-based approach may not solve unbiased index construction. To overcome these deficiencies, the present study relies on the statistical procedure for the construction of weights of the factors i.e., the principal component analysis (PCA) method. PCA methods always requires the input variable to have a similar scale of measurement, i.e., variables are commonly standardized to zero mean and unit variance (Baxter 1995). This technique is used when the input variables are in different units of measurements. However, Jolliffe (1986) pointed out that if the variables are in same units, standardization amounts to be an arbitrary choice before construction of multidimensional index. Though this study used various variables with different measurement units, the authors used following standardization process of variables before constructing index through PCA method.Standarization=X-MSD where X = actual/original value of the variables, M = mean value of the series, and SD = standard deviation of the series. The study uses six different indicators [automated teller machines (ATMs)] per 1000 km2, automated teller machines (ATMs) per 100,000 adults, branches of commercial banks per 1000 km2, branches of commercial banks per 100,000 adults, outstanding deposits with commercial banks (% of GDP), and outstanding loans with commercial banks (% of GDP) of financial inclusion for constructing a single index. For measuring the sample adequacy, the authors used Kaisor–Meyer–Olkin (KMO) and Bartlett’s Test of Sphericity test before running to PCA estimation in each of the separate countries. The average score of KMO for FII variables in different countries lies between 0.68 and 0.74 and significant Bartlett test value also (see Table 8 in the appendix section). In the PCA method, first we calculate the factor scores (weights) through their eigenvalues. Then we calculate the factor score (weights) of each variable and multiply it with the respective original variable. Finally, we add them together to get the single value of the composite index for ith state for a particular time period t. Hence, for constructing a single index of financial inclusion, the formula is expressed as:1 FIIit=Wi1X1+Wi2X2+Wi3X3+……+WipXp Here, FIIit is the financial inclusion index; Wi is the weight of the factor coefficient, X is the respective original value of the component, and p is the number of variables used. Here, the financial inclusion index for all the countries are calculated by adding together the entire factor scores (weights) and their respective original values. FIIit is the financial inclusion index of ith state for the time period t and W2, W2, ………. W6 are the weights of different factor scores. Finally, the financial inclusion index for all the 31 countries (16 countries for UMI and 15 countries for LMI) from both the income groups have been calculated (see Figs. 1 and 2 in Appendix). For both the income groups, periodical trends (i.e., 2004, 2014, and 2018) of financial inclusion have been presented in the figure to know the growth of financial inclusion in these countries over the years. Empirical models The prime objective of this paper is to empirically examine the impact of financial inclusion on control of corruption from the selected upper-middle- and lower-middle-income countries. The following econometric model is specified to materialize our above cited objectives as follows:2 CRPit=α0+β1FIIit+β2REMit+β3GDPPCit+β4HUMit+β5UEMPit+β6INFit+μit Here, in the above equation, CRPit represents the control of corruption and is used as dependent variable. FII indicates a financial inclusion index and is used in the main independent variable. Along with FII, this study also uses some control variables such as received remittance as a % of GDP (REM), per capita gross domestic product (GDPPC), human capital (HUM), rate of unemployment (UEMP), and rate of inflation (INF) and the µit refers to the error term. The subscript (i,t) denotes the cross sectional and time dimensions of the panel. The selected control variables have been well supported with the previous literature. As defined in the Eq. (2), the study uses REM, PGDPC, HUM, UEMP, and INF as its control variables. Previous literatures (like Ajide 2020 and Sharma and Paramati 2020) have used the same variables in their studies. The control variables like REM and GDPPC have important connection with the control of corruption. The inflow of high remittance to the economies and the rise in per capita income can encourage the people to use modern digital technologies for their financial transaction, which may help to reduce corruption in the economies. Correspondingly, the huge inflow of remittance to countries can also help the migrants to have greater political voice and that may bring more transparency in the public administration (Ajide and Olayiwola 2020). Likewise, HUM can also reduce corruption by promoting basic literacy in general and financial literacy in particular among the citizens. The study also chooses UEMP as one of the controls because the presence of high unemployment in the economies may lead to enhance the corruptive practices among the youth. Similarly, the rate of inflation (INF) may increase the corruption behaviors among the citizens by losing their income and encouraging them to be involved in corrupt practices (see Akca et al. 2012). Furthermore, the study used nonlinear term of financial inclusion (FII2) to ensure an evaluation of threshold level of financial inclusion (see Ozili 2020a, b; Nizam et al. 2020) in both selected UMI and LMI countries. Because of wide variation in the socio-economic characteristics, financial inclusion is having non-monotonic nature in these two categories of countries. In a simple way, it can be said that there are huge socio-economic and institutional differences existing in between these two categories of nations. These differences can lead to the unequal accessibility of financial services among the peoples of these two groups of countries. Hence, to capture the variability in financial accessibility, the study has taken threshold level of financial inclusion (FII2) as an independent variable along with the baseline financial inclusion index (FII), especially for the UMI and LMI countries. Therefore, the econometric model can be described in the following manner.3 CRPit=α0+β1FIIit+β2FIIit2+β3REMit+β4GDPPCit+β5HUMit+β6UEMPit+β7INFit+μit Estimating techniques The study uses both fixed effect and random effect models to measure the impact of financial inclusion on the control of corruption. Moreover, the random effect model is chosen for final interpretation of the result based on the Hausman test. Though panel data are mainly based on the two dimensions, i.e., time and cross-sectional dimensions; there might be the issue of autocorrelation and heteroscedasticity in the dataset. To take care of these issues, the study employs panel corrected standard errors (PCSEs) and feasible general least square method (FGLS) for robustness of the results. At the end, both PCSEs and FGLS are not sufficient to solve the issue of endogeneity or any potential problem of variables omission. To overcome these problems, the study relies on the instrument variable techniques (i.e., two-stage least-square-2SLS) and re-estimate the results. Empirical findings and discussion The motto of this study is to examine the impact of financial inclusion on controlling corruption among the 31 selected upper-middle- and lower-middle-income countries for the period of 2004–2018. To address the above question, the paper estimates three different regressions. First, the study examined the impact of financial inclusion on control of corruption taking the whole 31 countries. This study further segregated the data into two different samples, i.e., upper-middle-income (UMI) countries and lower-middle-income (LMI) countries and examined the same objective differently among these two categories of nations. The main purpose of doing separate regression for two separate categories of nations is to understand if any regional characteristics have impacted the outcomes. Because two categories of nations (i.e., UMI and LMI) have different social, economic, and institutional setup in their respective regions. Moreover, while segregating the whole dataset into two different categories (based on the World Bank classification), we have selected 16 countries for UMI countries and the rest 15 countries are used for LMI countries. Results of financial inclusion and control of corruption: for full sample countries This overall sample includes data from both the UMI and LMI countries. The findings from the overall sample countries depict that financial inclusion have a positive and significant effect on the control of corruption in the overall sample countries (see Table 2). The positive impact of financial inclusion on corruption control indicates that with the rise in the degree of financial inclusion in these selected countries, the basic financial literacy increases among the people, which further restricts the citizens to engage in any form of corrupt financial behavior. On the other side, with rising financial consciousness, it becomes difficult for financial fraudsters to cheat the people or to illegally engage them in any corrupt practices.Table 2 Impact of financial inclusion on corruption for full sample countries from 2004 to 2018 Variables FE Model-1 RE Model-2 PCSEs Model-3 FGLS Model-4 IV-2SLS Model-5 FI − 0.0013 (0.0020) 0.0004 (0.0018) 0.0038* (0.0009) 0.0038* (0.0014) 0.0562** (0.0282) REM − 0.0205* (0.0096) − 0.0182* (0.0091) − 0.0154* (0.0032) − 0.0154 (0.0065) − 0.0439 (0.0199) GDPPC 3.0305* (0.2990) 2.5211* (0.2616) 0.8537* (0.0718) 0.8537*** (0.1428) 2.1764 (1.6503) HUM 1.6774* (0.7515) 1.8307* (0.7422) 1.6411*** (0.5289) 1.6411* (0.9111) 8.5296 (4.1033) UEMP − 0.0021 (0.0054) − 0.0004 (0.0054) − 0.0376* (0.0047) − 0.0376* (0.0055) − 0.0718* (0.0212) INF − 0.0138* (0.0035) − 0.0145* (0.0035) − 0.0160* (0.0049) − 0.0160* (0.0060) − 0.0118 (0.0189) C − 11.813* (1.8512) − 10.2043* (1.7454) − 3.5518*** (1.2657) − 3.5518* (1.9065) − 8.1793* (4.4744) Observation 465 465 465 465 465 R-square 0.3664 0.3641 0.3341 0.4357 F-statistics 41.25 Prob. (F-statistics) 0.0000 Wald χ2 236.82 818.91 233.34 63.36 Prob > χ2 0.0000 0.0000 0.0000 0.0000 Hausman test − 7.39 Log likelihood − 467.5802 Durbin(score) χ2(1) 13.2216 * Wu-Hausman F(1,457) 13.3744* No. of countries 31 31 31 31 31 Source: authors estimation Dependent variable: corruption Standard errors in parentheses. p < 0.10, p < 0.05, p < 0.01 Furthermore, the overall increment of financial inclusion helps to reduce poverty (see Lal 2018; Inoue 2019; Zhang and Posso 2019; Churchill and Marisetty 2020; Ahamed and Gutiérrez-Romero 2020) and enhances the basic standard of living among the national citizens. The improvement in basic standard of living allows individuals to use digital technology in their day-to-day financial transactions. The frequent use of digital payment technologies helps to check the corrupt financial transaction and brings more transparency within the system. Hence, the overall increment of financial inclusion has a positive and significant effect on the control of corruption among the overall sample countries. The result of the control variables depicts that remittance inflow (REM), unemployment rate (UEMP), and inflation (INF) have negatively impacted the process of corruption control in the overall sample countries. The huge inflow of remittance can have negative impact on the control of corruption because the huge inflow of remittance to the economies can divert the public goods provision and because of higher income, migrants can obtain more favor from the public authorities by providing them with more bribes (see Berdiev et al. 2013). Similarly, the presence of unemployment in the economies can also promote the corruptive behavior among the national citizens. This is because, the persistence of high unemployment can provide an opportunity to the youth to give bribe to the public authority to obtain a job in the public sectors. Likewise, the presence of price instability in the economies due to high inflation rate can hinder the process of corruption control in the economies. However, other control variables like GDP per capita (GDPPC) and human capital (HUM) have positively impacted the control of corruption. On one side, the increment of individual income allows the individual to access basic amenities and can encourage the individual to less involve in corrupt practices. Similarly, with rise in income level, the individuals can opt to use modern mode of digital payment in their day-to-day financial transaction, which ultimately decay the corruptive behavior within the system. Likewise, the improvement in human capital by enhancing basic literacy level can augment the social consciousness of the negative consequence of corruption, which may enforce the individuals to less engage in corruptive practices. Results of financial inclusion and control of corruption: for upper-middle-income (UMI) countries The impacts of financial inclusion on control of corruption among the UMI countries are presented in the following table (Table 3). Here, in Table 3, we have used two levels of financial inclusion. One is the baseline level and another one is the threshold level. The reasons for leveling the financial inclusion data are to verify the impact of financial inclusion on control of corruption at the primary level and with the extensive level of financial inclusion. Here, the table has shown the impact of both baseline and threshold level financial inclusion on control of corruption. The baseline result in Table 3 depicts that financial inclusion has a negative and significant impact on the control of corruption across the models presented in columns 1–5. However, the threshold result of financial inclusion demonstrates that financial inclusion has a positive and significant effect on the control of corruption in UMI countries. That means, in UMI countries, financial inclusion has no impact on the control of corruption up to a certain threshold level, while a further increase in financial inclusion beyond the threshold level helps significantly to control corruption in these selected countries. These results can further describe as that in the UMI countries, majority of citizens can access basic financial services at an easy and affordable cost and this basic accessibility of financial services does not have any impact on the control of corruption. Whereas, after the surpass of a threshold level, financial inclusion has a positive impact on the control of corruption. This is happening because the upsurge of financial inclusion beyond the threshold level helps to spur economic growth in the counties (see Sharma 2016; Lenka and Sharma 2017; Sethi and Acharya 2018), which ultimately assists to increases the government expenditure on the provision of public goods. Because of the provision of more public goods from the government, the citizens would not be involved in any corrupt practices to access their basic amenities. Correspondingly, the rise of financial inclusion above the threshold level would allow the individuals to access more credit from the formal financial system and that obtained credit will be invested in the market by the individuals, which will bring more income to the individuals. The escalation of an individual’s income through financial inclusion would allow the government to have more tax collections based on the citizen’s income (Mitchell and Scott 2019). Again, the collection of more tax revenue would support the government to spend more on the provision of public goods, which would further restrict the citizens to engage in any corrupt practices. Moreover, the rise in government revenue would also spur digital innovation through the government expenditure on research and development (R&D). In the UMI countries, with a strong bureaucratic system, the use of digital technologies would help to bring more transparency in the organizations and reduce the corruptive behavior among the citizens. Therefore, after a threshold level, financial inclusion has a very positive and significant impact on the control of corruption in UMI countries.Table 3 Impact of financial inclusion on corruption for upper-middle-income countries from 2004 to 2018 Variables FE Model-1 RE Model-2 PCSEs Model-3 FGLS Model-4 IV-2SLS Model-5 FI − 0.01647* (0.0057) − 0.0158* (0.0055) − 0.0319* (0.0087) − 0.0319* (0.0053) − 0.0733* (0.0204) FI2 0.0001* (0.0003) 0.0001* (0.0003) .0003* (0.0007) 0.0003* (0.0004) 0.0007* (0.0001) REM − 0.0973* (0.0342) − 0.0606* (0.0280) − 0.0450* (0.0067) − 0.0450* (0.0150) − 0.0591* (0.0188) GDPPC 1.8454* (0.4929) 1.6337*** (0.4509) 0.4213* (0.2242) 0.4213 (0.3302) 0.3364 (0.5921) HUM 1.4706 (1.2389) 1.5198 (1.2571) 0.9240 (0.8563) 0.9240 (1.4086) 0.2451 (1.6884) UEMP − 0.0093 (0.0071) − 0.0014 (0.0068) − 0.0556* (0.0079) − 0.0556* (0.0060) − 0.0680* (0.0095) INF − 0.0365* (0.0076) − 0.0374* (0.0078) − 0.0468* (0.0113) − 0.0468*** (0.0126) − 0.0167 (0.0225) C − 6.8750* (3.5477) − 6.2562 (3.4452) 0.4949 (2.1931) 0.4950 (3.4360) 6.5620 (4.9530) Observation 240 240 240 240 240 R-square 0.3857 0.3788 0.4317 0.498 F-statistics 19.46 Prob. (F-statistics) 0.0000 Wald χ2 116.67 302.50 182.31 107.27 Prob > χ2 0.0000 0.0000 0.0000 0.0000 Hausman test − 7.05 Log likelihood − 228.1258 Durbin(score) χ2(1) 40.8487* Wu-Hausman F (1,457) 23.5881* No. of countries 16 16 16 16 16 Source: authors estimation Dependent variable: corruption Standard errors in parentheses. p < 0.10, p < 0.05, p < 0.01 With regard to the control variables used for UMI countries, we found that remittance has a negative and significant impact on the control of corruption. The inflow of more remittance to the economies can hinder the process of corruption control. Similarly, the findings illustrate that the degree of economic development which is proxied by per capita GDP (GDPPC) has a positive and significant impact on the control of corruption (An and Kweon 2017). This result depicts that the process of economic development helps to control corruption in the UMI countries. With the rise in GDP per capita, the citizens opt to use more modern digital technologies (like mobile/internet banking, automated teller machine, credit cards, etc.) in their financial transactions which bring more transparency in their financial transactions and reduce corruption by removing the middlemen participation. Additionally, with more economic development, countries endeavor to strengthen their public institutions and bureaucratic structure, which support to convey more transparency and accountability within the system. A similar kind of study was conducted by Sharma and Paramati (2020) showing the impact of financial development on controlling corruption in 140 countries across the globe. The results of that study show that per capita income positively impacts the control of corruption across selected countries. As expected, human capital (HUM) which is proxied by primary school enrollment ratio, has a positive and significant impact on the control of corruption. Likewise, the impact of unemployment on control of corruption describes that unemployment (UEMP) has a negative and significant impact on the control of corruption. That means the presence of high unemployment in the economies can hamper the drive of corruption control. This is because the incidence of high unemployment in the countries can provoke the youths to engage in corrupt practices to obtain a job. Similarly, this can also encourage the fake employment agencies and companies to involve in corrupt practices by taking money and providing false employment assurance to the youths. However, it is found that the rate of inflation (INF) in the countries has a negative and significant impact on the control of corruption in the UMI countries. That means the rise in the inflation rate creates price instability in the economies, which negatively influences the control of corruption in the studied countries (Akca et al. 2012). Results of financial inclusion and control of corruption: for lower-middle-income (LMI) countries Here, the present section demonstrates the results of financial inclusion and control of corruption in lower-middle-income (LMI) countries. Like the previous section (i.e., UMI countries), here also we have depicted the threshold level impact of financial inclusion on control of corruption along with its baseline results. However, unlike the UMI countries, here we have found a contrast result for LMI countries. The baseline result of financial inclusion indicates that financial inclusion has a positive effect on the control of corruption across the five models (see Table 4). Whereas the threshold result depicts that after the surpass of a threshold level, financial inclusion has a negative impact on the control of corruption in the LMI countries. That means in LMI countries, up to a certain threshold level, financial inclusion significantly helps to control corruption. However, a further enhancement of financial inclusion beyond that threshold level hinders the control of corruption in these countries. These findings further indicate that the accessibility of basic formal financial services plays a pioneering role to reduce corruption in the LMI countries. Whereas, the greater intensification of financial accessibility among the LMI countries’ people can have a downside impact on the corruption reduction process. This result clearly illustrates that beyond the threshold level, more intensification of financial inclusion can produce financial risk for the LMI countries people. This is quite possible because, in LMI countries with the poor institutional quality and weak organizational structure, it would be fairly difficult to supervise and regulate the financial accessibility service. It can further be argued that the more intensification of financial inclusion allows low-income individuals to access financial services (including formal credit) without proper scrutiny and risk measure. This kind of financial accessibility sometimes encourages criminals, hackers, and fraudsters to access formal loans by giving money to the bank staffs or by providing fake documents. Similarly, greater accessibility of bank credits can provoke individuals to engage themselves in other financial crimes like insider trading, money laundering, gambling, etc. Because people from LMI countries normally have less financial literacy and low investment skills, which would further mislead them in their financial investment. Additionally, the greater intensification of financial inclusion would enhance the use of digital transaction technologies in the LMI countries. The frequent use of digital transaction technologies without proper digital training and institutional supervision can also promote cyber hacking and financial crimes in the LMI countries.Table 4 Impact of financial inclusion on corruption for lower-middle-income countries from 2004 to 2018 Variables FE Model-1 RE Model-2 PCSEs Model-3 FGLS Model-4 IV-2SLS Model-5 FI 0.0095 (0.0111) 0.0081 (0.0106) 0.0072 (0.0070) 0.0072 (0.0105) 0.0475 (0.0263) FI2 − 0.0001 (0.0001) − 0.0009 (0.0001) − 0.0002* (0.0009) − 0.0002 (0.0001) − 0.0008 (0.0004) REM − 0.0282* (0.0096) − 0.0203 (0.0090) − 0.0088 (0.0041) − 0.0088 (0.0070) − 0.0080 (0.0075) GDPPC 3.0927* (0.4629) 2.8805* (0.4216) 2.1870* (0.2338) 2.1870* (0.2794) 2.0085* (0.3098) HUM 1.4237 (0.9920) 2.0223 (0.9584) 2.7160* (0.6292) 2.7160* (0.9846) 3.0348* (1.0618) UEMP − 0.0152 (0.0101) − 0.0120 (0.0098) − 0.0046 (0.0099) − 0.0046 (0.0106) − 0.0036 (0.0121) INF − 0.0070 (0.0037) − 0.0067 (0.0037) − 0.0015 (0.0057) − 0.0015 (0.0059) − 0.0070 (0.0071) C − 11.3743* (2.1241) − 11.7328* (2.0377) − 10.2532* (1.6760) − 10.2531* (2.1430) − 0.8784* (2.3316) Observation 225 225 225 225 225 R-square 0.4861 0.4838 0.4011 0.3218 F-statistics 27.43 Prob. (F-statistics) 0.0000 Wald χ2 184.61 375.61 150.72 125.45 Prob > χ2 0.0000 0.0000 0.0000 0.0000 Hausman test 8.66 Log likelihood − 178.5791 Durbin(score) χ2(1) 4.81295 Wu-Hausman F(1,457) 2.34978 No. of countries 15 15 15 15 15 Source: authors estimation Dependent variable: corruption Standard errors in parentheses. p < 0.10, p < 0.05, **p < 0.01 Similarly, the LMI counties generally have a very low presence of bank branches in rural and remote areas; hence, it is quite difficult for the rural people to access banking services at an easy and affordable charge. This incidence provokes the middlemen to provide banking services (like proving bank credits, ATM and other digital technology services) by taking some amount of money from the customers. Even sometimes, the bank staffs may take some amount of bribe for extending the banking services. As a result of all these reasons, financial inclusion after a threshold level has a negative effect on the control of corruption in the selected LMI countries. The finding of this study is relatively associated with the finding of Ajide (2020), where the author conducted a cross-country study considering 13 African countries. The finding of that paper depicts that after the surpass of a threshold level, financial inclusion has a negative impact on the control of corruption in the selected African countries (see Ajide 2020). Except for the main variable (i.e., FI), the results of control variables for LMI are quite similar to the control variables findings of UMI countries. Like the UMI countries, we found that remittance (REM), unemployment (UNMP), and inflation rate (INF) have a negative impact on the control of corruption. Whereas the other two variables, i.e., per capita GDP (GDPPC) and human capital (HUM) have a positive and significant effect on the control of corruption in LMI countries. The analysis of the control variables for LMI countries follows the same arguments as UMI countries. Concluding remarks The practice of corruption severely affects the social, economic, and political behavior of the nation. Most specifically, corruption affects the process of economic development, issues of unemployment, social and political stability, and more importantly the everyday lives of the general people. Understanding the gravity of the issue, policy makers and academicians across the globe have taken their interest to study the various aspects of corruption (Goel and Nelson 2010; Corrado and Rossetti 2018). Correspondingly, the beginning of the twenty-first century has also witnessed rapid progress in the process of financial inclusion among the developing and lower developed countries. The rise in basic financial inclusion has also witnessed the corresponding rise in digital payment technologies across the globe. Hence, given this background, this study is an endeavor to empirically examine the role of financial inclusion in the process of corruption control. Furthermore, to understand the dynamic relationship between these variables, we further segregated our analysis into two categories of nations, i.e., upper-middle-income (UMI) and lower-middle-income (LMI) countries. The empirical finding of this study illustrates that, in the overall sample counties, financial inclusion is playing a pioneering role in the control of corruption. Furthermore, while empirically examining the same objective by considering the baseline and threshold level of financial inclusion among the UMI and LMI countries, our findings show very interesting relationship between these two variables. The findings from UMI countries demonstrate that at the basic level, financial inclusion does not affect the control of corruption. While the more intensification of the financial inclusion process significantly plays a positive role to control corruption in the selected UMI countries. Likewise, the results from LMI countries suggest pure contrast answers. The results show that the basic level of financial inclusion is having a significant role to play in the process of corruption control. Whereas, the surpass of financial inclusion beyond the threshold level is having a downside impact on the control of corruption in the selected LMI countries. Concerning to the control variables, the result depicts that across the three categories of nation, the sign of all the control variables remains the same. The control variables like remittance (REM), unemployment (UNMP), and inflation rate (INF) are having a negative impact on the control of corruption, whereas other two control variables such as per capita GDP (GDPPC) and human capital (HUM) are having a positive and significant effect on the control of corruption across the three categories of nations. Policy implications Based on our empirical findings, this study takes an opportunity to propose some valuable policy suggestions to the policy makers. The findings show that the overall sample of financial inclusion is positively impacting to the control of corruption. Similarly, in the case of UMI countries, an extension of financial inclusion beyond the basic financial accessibility significantly impacts the control of corruption. Based on this result, it can be suggested that the UMI countries should give more priority to the intensification of financial inclusion process as it is more beneficial for controlling corruption. However, in the case of LMI counties, the empirical findings suggest that, up to a certain threshold level, financial inclusion can play a pivotal role to control corruption, whereas the extension of financial inclusion beyond the threshold level can increase the level of corruption. From this result, it can be said that in case of LMI countries, because of less financial literacy/awareness, low financial supervision, and weak institutional structure, more intensification of financial inclusion after a certain level can enhance the process of corruption. This kind of result does not discourage more intensification of financial inclusion in LMI countries. Rather, it suggests, along with the enhancement of financial inclusion, LMI countries should also give equal importance to financial literacy/awareness, stringent financial supervision policy, and strong institutional structure to reduce the level of corruption in their countries. Appendix See appendix Tables 5, 6, 7, 8.Table 5 List of countries considered for this study Sl. No. WB classification Country Sl. no. WB classification Country 1 UMIC Brazil 17 LMIC Moldova 2 UMIC Bulgaria 18 LMIC Egypt 3 UMIC China 19 LMIC India 4 UMIC Costa Rica 20 LMIC Kenya 5 UMIC Ecuador 21 LMIC Morocco 6 UMIC Guatemala 22 LMIC Pakistan 7 UMIC Kazakhstan 23 LMIC Philippines 8 UMIC Malaysia 24 LMIC Tunisia 9 UMIC Mexico 25 LMIC Ukraine 10 UMIC Peru 26 LMIC Vietnam 11 UMIC Russia 27 LMIC Zambia 12 UMIC Serbia 28 LMIC Zimbabwe 13 UMIC South Africa 29 LMIC Bangladesh 14 UMIC Indonesia 30 LMIC Ghana 15 UMIC Thailand 31 LMIC Sri Lanka 16 UMIC Turkey Source: authors estimation Table 6 Descriptive statistics for all 31 countries Mean Standard deviation Minimum Maximum CPI 3.373118 0.8114137 1.5 5.9 FI 43.6712 27.37524 3.716863 100 REM 4.650598 5.23452 0.0932936 34.499 LGDPPC 3.915394 0.2964055 3.168833 4.450265 LHUM 2.018941 0.0357658 1.887118 2.128788 UEMP 7.332232 5.836214 − 10.4974 30.91959 INF 6.419217 5.359475 − 2.4095 48.69986 Source: authors estimation Table 7 Correlation matrix for all 31 countries CPI FI REM LGDPPC LHUM UEMP INF CPI 1.0000 FI 0.3047 1.0000 REM − 0.2897 0.1185 1.0000 LGDPPC 0.4841 0.5739 − 0.4078 1.0000 LHUM 0.1124 0.1652 − 0.1209 0.0129 1.0000 UEMP − 0.3248 − 0.0718 − 0.1246 0.1562 0.0827 1.0000 INF − 0.1795 − 0.1505 0.0521 − 0.1361 − 0.2183 0.0329 1.0000 Source: authors estimation Table 8 KMO and Bartlett’s test Kaiser–Meyer–Olkin measure of sampling adequacy 0.682 Bartlett’s test of sphericity Approx. χ2 234.293 df 15 Sig. 0.000 See appendix Figs. 1, 2, 3, 4.Fig. 1 Status of financial inclusion in UMI countries Fig. 2 Status of financial inclusion in LMI countries Fig. 3 Status of corruption in UMI countries Fig. 4 Status of corruption in LMI countries Data availability All the data used in this study is available in the public domain. The Individual sources of data availability is mentioned in the data description Table 1. 1 As per the 2021 classification of countries, countries’ GNI from 1046 to 4096 are classified as lower-middle-income (LMI) countries, whereas GNI from 4096 to 12,695 are classified as upper-middle-income (UMI) countries. 2 The CPI has been defined as the ‘misuse of public power for private benefit’ (Hamilton and Hammer, 2018). The CPI score refers to the perceived level of corruption in the public sector. Every individual country is assigned a corruption score which varies from 0 to 10. The score ‘0’ refers to the extreme level of corruption whereas the score ‘10’ refers no presence of corruption. Each individual country score is developed by aggregating and averaging normalized scores of ‘corruption related data’ collected from various international data sources. Though some methodological limitations have been observed in the calculation of CPI (Álvarez-Díaz et al. 2018), it has been widely used among the researchers to measure the degree of corruption in a given country. Researcher like Ajide (2020) has used the same data for analyzing the impact of financial inclusion on corruption in selected African countries. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abé Ndjié A Atangana Ondoa H Ngoa Tabi H Governance and youth unemployment in Africa Labor Hist 2019 60 6 869 882 10.1080/0023656X.2019.1645320 Abu N Abd Karim MZ Aziz MIA Low savings rates in the Economic Community of West African States (ECOWAS): the role of corruption J Econ Coop Dev 2015 36 2 3 ADB (2000) Finance for the poor: microfinance development strategy. Rural Asian study: Beyond the Green Revolution 52 Adjor DM Kebalo L Does corruption matter for unemployment in SADC countries? Rev Econ Bus Stud 2018 11 2 65 92 10.1515/rebs-2018-0074 Ahamed M, Gutiérrez-Romero R (2020) COVID-19 response needs to broaden financial inclusion to curb the rise in poverty. World Dev. Preprint at arXiv:2006.10706 Ahlin C Pang J Are financial development and corruption control substitutes in promoting growth? J Dev Econ 2008 86 2 414 433 10.1016/j.jdeveco.2007.07.002 Ahmad E Ullah MA Arfeen MI Does corruption affect economic growth? Lat Am J Econ 2012 49 2 277 305 10.7764/LAJE.49.2.277 Ajide FM Can financial inclusion reduce the presence of corruption? Evidence from selected countries in Africa Int J Soc Econ 2020 47 11 1 8 10.1108/IJSE-03-2020-0145 Ajide FM Olayiwola JA Remittances and corruption in Nigeria J Econ Dev 2020 23 1 1 15 Akca H Ata AY Karaca C Inflation and corruption relationship: evidence from panel data in developed and developing countries Int J Econ Financ Issues 2012 2 3 281 295 Ali M Nazir MI Hashmi SH Ullah W Financial inclusion, institutional quality and financial development: empirical evidence from OIC countries Singapore Econ Rev 2022 67 01 161 188 10.1142/S0217590820420084 Altunbaş Y Thornton J Does financial development reduce corruption? Econ Lett 2012 114 2 221 223 10.1016/j.econlet.2011.08.020 Álvarez-Díaz M, Saisana M, Montalto V, Moura CT (2018) Corruption Perceptions Index 2017 Statistical Assessment. European Commission Joint Research Centre Technical Report. An W Kweon Y Do higher government wages induce less corruption? Cross-country panel evidence J Policy Model 2017 39 5 809 826 10.1016/j.jpolmod.2017.03.001 Arora RU (2010) Measuring financial access. Griffith University, Discussion Paper Economics, No-7 Barth JR Caprio G Jr Levine R Bank regulation and supervision: what works best? J Financ Intermed 2004 13 2 205 248 10.1016/j.jfi.2003.06.002 Baxter MJ Standardization and transformation in principal component analysis, with applications to archaeometry Appl Stat 1995 44 4 513 10.2307/2986142 Beck T Demirgüç-Kunt A Access to Finance: an unfinished Agenda World Bank Econ Rev 2008 22 3 383 396 10.1093/wber/lhn021 Berdiev AN Kim Y Chang CP Remittances and corruption Econ Lett 2013 118 1 182 185 10.1016/j.econlet.2012.10.008 Chakravarty SR Pal R Financial inclusion in India: an axiomatic approach J Policy Model 2013 35 5 813 837 10.1016/j.jpolmod.2012.12.007 Chakravorty NN (2019) How does corruption affect economic growth? An econometric analysis. J Leadersh Account Ethics 16(4) Churchill SA Marisetty VB Financial inclusion and poverty: a tale of forty-five thousand households Appl Econ 2020 52 16 1777 1788 10.1080/00036846.2019.1678732 Cooray A Schneider F Does corruption throw sand into or grease the wheels of financial sector development? Public Choice 2018 177 1–2 111 133 10.1007/s11127-018-0592-7 Corrado G Rossetti F Public corruption: a study across regions in Italy J Policy Model 2018 40 6 1126 1139 10.1016/j.jpolmod.2018.01.001 Dupuy K, Neset S (2018) The cognitive psychology of corruption. Micro-level explanations for unethical behavior. U4 Issue, 2. https://www.u4.no/publications/the-cognitive-psychology-of-corruption.pdf Eldomiaty T Hammam R El Bakry R Institutional determinants of financial inclusion: evidence from world economies Int J Dev Issues 2020 19 2 217 228 10.1108/IJDI-08-2019-0147 Goel RK Nelson MA Causes of corruption: history, geography and government J Policy Model 2010 32 4 433 447 10.1016/j.jpolmod.2010.05.004 Gupta S Davoodi H Alonso-Terme R Does corruption affect income inequality and poverty? Econ Gov 2002 3 1 23 45 10.1007/s101010100039 Gupte R Venkataramani B Gupta D Computation of financial inclusion index for India Proc Soc Behav Sci 2012 37 133 149 10.1016/j.sbspro.2012.03.281 Hamilton A, Hammer C (2018) Can we measure the power of the grabbing hand? Policy Research Working Paper 8299 Hussain M Yahya F Waqas M Does strong governance stimulate the effect of economic freedom and financial literacy on financial inclusion? A Cross-Country Evidence Future Bus J 2021 7 1 1 10 10.1186/s43093-021-00090-9 Inoue T Financial inclusion and poverty reduction in India J Financ Econ Policy 2019 11 1 1 14 10.1108/JFEP-01-2018-0012 Jolliffe IT Principal component analysis 1986 New York Springer-Verlag Kane E What lessons should Japan learn from the US deposit-insurance mess? J Jpn Int Econ 1993 7 4 329 355 10.1006/jjie.1993.1021 Khemani R, Meyerman G (1998) East Asia’s economic crisis and competition policy. Global Compet Rev (August/September), 16–18 Kolstad I Søreide T Corruption in natural resource management: Implications for policy makers Resour Policy 2009 34 4 214 226 10.1016/j.resourpol.2009.05.001 Lal T Impact of financial inclusion on poverty alleviation through cooperative banks Int J Soc Econ 2018 45 5 1 30 10.1108/IJSE-05-2017-0194 Leff NH Economic development through bureaucratic corruption Am Behav Sci 1964 8 3 8 14 10.1177/000276426400800303 Lenka SK Barik R Has expansion of mobile phone and internet use spurred financial inclusion in the SAARC countries? Financ Innov 2018 4 1 1 19 10.1186/s40854-018-0089-x Lenka SK Sharma R Does financial inclusion spur economic growth in India? J Dev Areas 2017 51 3 215 228 10.1353/jda.2017.0069 MacFarlane J Corruption and the financial sector: the strategic impact J Financ Crime 2001 9 1 8 22 10.1108/eb026002 Mader P, Duvendack M (2019) Impact of financial inclusion in low-and middle-income countries: a systematic review of reviews 34(3):1–36 Malik AH bin Md Isa AH bin Jais M Rehman AU Khan MA Financial stability of Asian Nations: governance quality and financial inclusion Borsa Istanbul Rev 2022 22 2 377 387 10.1016/j.bir.2021.05.005 Mauro P, Driscoll DD (1997) Why worry about corruption? vol 6, pp 1–19. International Monetary Fund, Washington, DC Méndez F Sepúlveda F Corruption, growth and political regimes: cross country evidence Eur J Polit Econ 2006 22 1 82 98 10.1016/j.ejpoleco.2005.04.005 Méon PG Weill L Is corruption an efficient grease? World Dev 2010 38 3 244 259 10.1016/j.worlddev.2009.06.004 Mitchell K, Scott RH (2019) Public revenue, financial inclusion and value-added tax in Argentina. In: Pesos or Plastic? Palgrave Pivot, Cham, pp 33–58 Mo PH Corruption and economic growth J Comp Econ 2001 29 1 66 79 10.1006/jcec.2000.1703 Nizam R Karim ZA Rahman AA Sarmidi T Financial inclusiveness and economic growth: new evidence using a threshold regression analysis Economic Research-Ekonomska Istraživanja 2020 33 1 1465 1484 10.1080/1331677X.2020.1748508 Ozili PK (2020a) Financial inclusion research around the world: a review. In: Forum for social economics, Routledge, pp 1–23 Ozili PK (2020b) Financial inclusion and fintech during COVID-19 crisis: policy solutions. Company Lawyer J 8 Park J Corruption, soundness of the banking sector, and economic growth: a cross-country study J Int Money Financ 2012 31 5 907 929 10.1016/j.jimonfin.2011.07.007 Rangarajan (2008) The final report of the committee of “Financial Inclusion” under the chairmanship of Dr. C. Rangarajan Reserve Bank of India (2014) Financial Intermediation for All—Economic Growth with Equity. Speech delivered by Dr. Deepali Pant Joshi at the Financial Inclusion Conclave of Dun & Bradstreet, Mumbai on August 26, 2014 Sahay MR, von Allmen MUE, Lahreche MA, Khera P, Ogawa MS, Bazarbash M, Beaton MK (2020) The promise of fintech: financial inclusion in the post COVID-19 era. Int Monetary Fund Sarma M (2008) Index of financial inclusion. ICRIER Working paper no. 215 Sethi D Acharya D Financial inclusion and economic growth linkage: some cross country evidence J Financ Econ Policy 2018 10 3 1 18 10.1108/JFEP-11-2016-0073 Sha'ban M Girardone C Sarkisyan A Cross-country variation in financial inclusion: a global perspective Eur J Finance 2020 26 4–5 319 340 10.1080/1351847X.2019.1686709 Sharma D Nexus between financial inclusion and economic growth J Financ Econ Policy 2016 8 1 1 40 10.1108/JFEP-01-2015-0004 Sharma C Paramati SR Does financial development reduce the level of corruption? Evidence from a global sample of 140 countries Int J Financ Econ 2020 25 3 1 17 Song C Chang C Gong Q Economic growth, corruption, and financial development: Global evidence Econ Model 2021 94 822 830 10.1016/j.econmod.2020.02.022 Sornaganesh V Ganesh S Sathish MT Impact of COVID-19 outbreak in digital payments Int J Innov Res Multidiscip Field 2020 6 8 159 164 Svensson J Eight questions about corruption J Econ Perspect 2005 19 3 19 42 10.1257/089533005774357860 Thach NN Duong MB Oanh TTK Effects of corruption on economic growth-empirical study of Asia countries Imp J Interdiscip Res 2017 7 791 804 Thornton J (2010) Does Financial Development Reduce Corruption? Bangor Business School Research Paper No. 09/003. SSRN: https://ssrn.com/abstract=1564445 or 10.2139/ssrn.1564445 Weill L How corruption affects bank lending in Russia Econ Syst 2011 35 2 230 243 10.1016/j.ecosys.2010.05.005 Weill L Does corruption hamper bank lending? Macro and micro evidence Empir Econ 2011 41 1 25 42 10.1007/s00181-010-0393-4 Zhang Q Posso A Thinking inside the box: a closer look at financial inclusion and household income J Dev Stud 2019 55 7 1616 1631 10.1080/00220388.2017.1380798	0	PMC9716501	NO-CC CODE	2022-12-03 23:20:57	no	Asia-Pac J Reg Sci. 2022 Dec 2;:1-24	utf-8	null	null	null	oa_other
==== Front J Low Temp Phys J Low Temp Phys Journal of Low Temperature Physics 0022-2291 1573-7357 Springer US New York 2920 10.1007/s10909-022-02920-8 Article Development of Superconducting On-chip Fourier Transform Spectrometers http://orcid.org/0000-0002-3351-3078 Basu Thakur Ritoban [email protected] 12 Steiger A. 34 Shu S. 12 Faramarzi F. 5 Klimovich N. 12 Day P. K. 2 Shirokoff E. 34 Mauskopf P. D. 5 Barry P. S. 6 1 grid.20861.3d 0000000107068890 Department of Physics, California Institute of Technology, Pasadena, 91125 CA USA 2 grid.211367.0 0000 0004 0637 6500 Jet Propulsion Laboratory (NASA), 4800 Oak Grove Dr, Pasadena, 91109 CA USA 3 grid.170205.1 0000 0004 1936 7822 Kavli Institute for Cosmological Physics, 5640 S. Ellis Ave, Chicago, IL 60637 USA 4 grid.170205.1 0000 0004 1936 7822 Department of Astronomy & Astrophysics, University of Chicago, 5640 S. Ellis Ave., Chicago, IL 60637 USA 5 grid.215654.1 0000 0001 2151 2636 Department of Physics, School of Earth & Space Exploration, Arizona State University, Tempe, AZ 85281 USA 6 grid.187073.a 0000 0001 1939 4845 Argonne National Laboratory, HEP Division, 9700 South Cass Avenue, Argonne, IL 60439 USA 2 12 2022 110 31 10 2021 1 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Superconducting On-chip Fourier Transform Spectrometers (SOFTS) are broadband, ultra-compact and electronic interferometers. SOFTS will enable kilo-pixel spectro-imaging focal planes, enhancing sub-millimeter astrophysics and cosmology. Particular applications include cluster astrophysics, cosmic microwave background (CMB) science, and line intensity mapping. This article details the development, design and bench-marking of radio frequency (RF) on-chip architecture of SOFTS for Ka and W-bands. Keywords Nonlinear Kinetic Inductance Spectrometer CMB Line Intensity Mapping http://dx.doi.org/10.13039/100000104 National Aeronautics and Space Administration NNH18ZDA001N-APRA NNH18ZDA001N-APRA http://dx.doi.org/10.13039/100000104 Basu Thakur Ritoban Shirokoff E. Mauskopf P. D. ==== Body pmcIntroduction In thin film superconductors like NbTiN and NbN, increasing supercurrent I modifies the density of states, increasing kinetic inductance [1]. This, in turn, alters the phase velocity, ultimately enabling current-controlled delays in a transmission line geometry. For a microstrip transmission line (length ℓ, width w, inductance-per-square L□ and impedance Z0) with inductance and capacitance per unit length of L and C, respectively, the current-controlled delay is given by Eq. 1. The characteristic currents I∗,I∗′ are determined by the material properties and device geometry.1 Δτ(I)=ℓL(I)C-L(I=0)C≈L□ℓZ0w[1+(I/I∗)2+(I/I∗′)4]-1 We employ this core idea to realize Superconducting On-chip Fourier Transform Spectrometers (SOFTS [2]). A broadband input is split in two parallel transmission lines where a relative phase delay is introduced with current biasing. And recombined signals form interferograms like a classical FTS. In this paper, we present a thorough superconducting circuit design, numerical characterization, device fabrication and calibration plans expanding on our previous work in 1–10 GHz and 25–40 GHz ranges [2, 3]. Mach-Zhender Architecture SOFTS is designed as a 4-port Mach-Zhender interferometer, Fig. 1; Michelson and other architectures are also doable [4]. The two inputs are a broadband antenna with band-defining filter observing the sky and a bolometric load as a precision calibrator. These are combined with phasing via a hybrid coupler (HC). Superconducting Transmission Lines (STLs) connect to HC via diplexers. The diplexers DC bias the STLs. The STLs are effectively optical arms of a FTS. With DC current biasing one arm (potential audio-band AC biasing to be explored for multiplexing advantage), relative phase delay is added such that following the final HC, two detectors see the interfered power analogous to the symmetric and antisymmetric ports of a classical FTS. We ultimately Fourier transform the measured interferograms, where achievable frequency resolution is δν=1/max(Δτ(I)).Fig. 1 a System level diagram for 4-port SOFTS b Delay with current derived from measurements [3] Fig. 2 4-port SOFTS devices, colored dashed boxes denote sub-systems: Left, Ka-band chip, hybrid coupler (purple), transmission line (blue), diplexer (green) Right, W-band, with mask overlaid on photo: radial probe (red), hybrid coupler (purple), transmission line (blue) and bias tees (green). See Table 1 for details Table 1 Parameters for two current SOFTS architectures Band Chip-dimensions [mm] Material Tc [K] Nonlinearity (ΔLmax/L) δν [GHz] νmax[THz] Ka 1.15×6.08 NbTiN 14 20% 1.43 [2] ≲1 W 45 × 15 NbN 13 18 % 0.1 ≲1 Fig. 3 a Superconducting Hybrid Coupler (HC) structure on-chip. We vary the λ/4 length-scale to operate in requisite bands. Simulated S-parameters for b Ka-band HC and c W-band HC Fig. 4 On-chip diplexers realized with capacitative stubs and inductive lines: a Ka-band b W-band c Ka-band diplexer simulated scattering parameters d W-band diplexer simulated scattering parameters We are testing two SOFTS architectures in parallel with Ka and W-band devices, see Table 1 and Fig. 2. The variations in superconducting material, geometry, and optical coupling are intentional. The W-band chip has a larger footprint than the Ka-band device because (i) for a 4-port split block, we needed the probes in a row which takes up more space (ii) to lower δν (frequency resolution) we use a long STL for the W-band device. The Ka-band chip in contrast is optimized for lower resolution and high compactness. We will take the best aspects of each design and develop a unified SOFTS architecture for particular science cases. For CMB-science we need δν∼O(1) GHz, and for line-intensity mapping δν∼O(0.1) GHz. SOFTS resolution is tunable, i.e., we can always increase δν by lowering the bias current. Here we quote the smallest δν which depends on device geometry and material. Figures 3 and 4 show design and simulation of on-chip HCs and diplexers used in our devices. RF Analysis RF Cascade Simulation Fig. 5 SOFTS cascade-network, ‘a’s /‘b’s denote incident/ reflected voltages, super-scripts and sub-scripts denotes individual subsystems and port indices. Diplexers suppressed for simplicity N-port subsystem S-matrices are linked as shown in Fig. 1 and cascade simulations are run with MathWorks’ RF Blockset software.1 We input single tones on port 1 of first hybrid-coupler (HC0), monitoring the output power (ports 2 and 3 of HC3) and reflected power in the second input port (port 4 of HC0), Fig. 5. All powers are monitored as dissipation across 50 Ω.In the ideal case, i.e., for negligible reflections and loss from the transmission lines, and low cross-talk between output ports of each hybrid-coupler, we can model the power measured in the symmetric and anti-symmetric ports (ports 2 and 3 of HC3). For unit voltage input on the antenna port, this is given by Eq. 2, where Δτ=τ2-τ1.2 P2(3)=150ΩS21(3)(ν)S21(0)(ν)+e-i2πνΔτ(I)S24(3)(ν)S31(0)(ν)2P3(3)=150ΩS31(3)(ν)S21(0)(ν)+e-i2πνΔτ(I)S34(3)(ν)S31(0)(ν)2⇒P2/3(3)≈1V250Ω·21±cos(2πνΔτ(I)) For the 90-deg hybrid-coupler, S34≈S21≈i and S31≈S24≈1, and Eq. 2 reduces to the standard FTS results of cosine modulations as shown above. RF cascade simulations allow us to comprehensively model frequency dependencies, multi-path effects from reflections and cross-talk, from which we can expect mild anharmonicities. The ability to comprehend these anharmonicities is indeed a major advantage for SOFTS, i.e., we can understand the spectrometer performance as a pure circuit model, as compared to optical FTS where multipath effects are challenging to accurately model and correct for. Here we are using simulated S-parameters; measured ASOFTS, e.g., our previous publication [2], can also be used. From Eq. 1 and prior work [2, 3] we expect up to 2 ns of delay. For every single-tone input, we scan over this range to produce interferograms and their FFTs generate a transfer function (ASOFTS), i.e., observed frequencies for single-tone inputs. Figure 6 shows relevant figures for Ka-band studies and W-band simulations are done identically. Digital signal processing with the transfer function enables accurate spectral recovery.Fig. 6 RF simulations of SOFTS in Ka-band: a Interferograms (zoomed in) for a 30 GHz input tone b ASOFTS (transfer function matrix) for port-2 of the output hybrid coupler Error Correction Error correction implies accounting for device non-ideality generated anharmonicities so that we have accurate spectral recovery. Each input tone is a unit vector in frequency space, named U→k, where only the kth element is 1, e.g., νmin= 1,0,0,...T, νmin+δν= 0,1,0,...T. Each single tone input generates multi-tone output given by V→k=ASOFTS·U→k, see Fig. 6. Suppose that Btrue is the true multichroic sky-signal and Bobs is the SOFTS spectrum that is readout. Since U→k is essentially a delta function in frequency space, we pursue inversion following least-squares method, Eq. 3.3 ASOFTSTASOFTS-1ASOFTSTV→k=δν,νk⇒Btrue=ASOFTSTASOFTS-1ASOFTST·Bobs ASOFTSTASOFTS-1ASOFTST is analogous to a Green’s function for SOFTS devices. We demonstrate spectral reconstruction with fractional errors \|Bobs-Btrue\|/Btrue≲10-11 by considering the CMB spectrum as measured by SOFTS, Fig. 7. This is a major achievement over classical optical FTSs where reconstruction of multi-path and alignment issues are far less accurate [5]. Our error is fundamentally set by circuit non-idealities across the chip, caused by impedance mismatches originating from practical limits of fabrication. Here we demonstrate spectral recovery without noise. Robust recovery with noise is also possible in this framework [6], and has been demonstrated for photon and detector noise [7].Fig. 7 Ka-band and W-band fractional errors following complete RF cascade simulations Optical Coupling and Device Hardware Ka-band SOFTS in essence is an inverted microstrip architecture where 35 nm thick and 250 nm wide NbTiN is the workhorse superconductor, Fig. 2. The STL fabrication is a stepper driven process, and all fabrication steps are identical to our published work on the measurement of Ka-band phase-delay [3]. The minimum resolved frequency for this device is expected to be 1.43 GHz, based on prior measurements [3]. The STL design is mostly band-independent, and the maximum frequency is limited by the ∼1.2 THz gap for NbTiN [8]. It will be slightly reduced during operation due to the change in density of states from the applied bias current. For operation near Ic the fractional reduction in the gap Δ(Γ)/Δ0≈0.9 [1], still allowing for maximum frequencies ∼ 1 THz. We have fabricated a printed circuit board (PCB) for mounting the SOFTS chip, and an OFHC copper housing to encase the chip and PCB for laboratory testing, Fig. 8a. We simulated the PCB over the Ka-band and measured <-20 dB cross-talk and <-10 dB reflections. Ultimately for antenna coupled SOFTS the PCB is unnecessary. The actual SOFTS chip is ∼ 6 mm × 1 mm. The PCB is necessary to make RF and DC wire bonding connections to the device and connect to the device housing ports. W-band SOFTS housing with waveguide coupling has been designed and fabricated and coupling is done using a radial probe [9]. The STLs were formed from a 40 nm thick deposited niobium nitride film etched using a reactive ion etching (RIE) process. Other circuit elements, including the probes and the hybrids, were formed from a 150 nm thick niobium (Nb) film using a liftoff process. Silicon nitride was used as the dielectric layer (500 nm) and it was deposited on top of the circuit using a Plasma-Enhanced Chemical Vapor Deposition (PECVD) method. The “skyplane” was then deposited on top of the dielectric. Lastly, the silicon below the probes is etched away for improved probe coupling. Chip-housing consists of three parts, with the top parts being the split-block waveguide as shown in Fig. 8b–d, and a chip holder, which consists of the waveguide backshorts and acts as a heat sink. We have two extrude cuts on the chip holder for DC-biasing circuit boards.Fig. 8 a The Ka-band SOFTS chip sits in the PCB cut-out, and is wirebonded to the leads. PCB routes the signals to DC and RF connectors at the edge of the copper module. b Housing assembly for W-band SOFTS chip includes four waveguide flange fittings for each port and through holes for mounting. c A split block is used to transition from the waveguides to the W-band on-chip probes. d The chip holder design includes backshorts for the probes and trenches for the DC bias PCBs Configuration and Application Examples While specific science requirements will define SOFTS-based kilopixel array architectures [7], we consider SOFTS devices covering 90–270 GHz and compare2 with SPT-3G, a state-of-the-art CMB telescope [10]. Frequency resolution determines STL length, the largest element in a SOFTS chip. For δν≈4 GHz, a sufficient resolution for CMB science cases, SOFTS will occupy ≈2 mm2 as scaled from the Ka-band device. A 90–270 GHz antenna is ≈10 mm2 and dominates focal plane area occupation. Therefore we can commensurately fit ≈2100 SOFTS spaxels compared to 2690 in SPT-3G. Each SPT-3G pixel has 6 bolometers (3 coarse spectral bands × 2 polarizations) and SOFTS will have 4 bolometers per spaxel (2 sum/difference ports × 2 polarizations); 33% reduction in readout burden. Instead of 3 coarse bands in SPT-3G, SOFTS allows (270-90)GHz/4GHz=45 spectral channels. Reducing DC bias lowers channel counts (commensurately increases sensitivity), should dynamic optimization be needed. On a focal plane similar to state-of-the-art, we will therefore have 1/3rd fewer detectors and ×15 more spectral channels, all with similar pixel counts. Thus SOFTS enables kilopixel spectro-imaging in the sub-millimeter. Biasing O(103) devices is non-trivial, though there is precedent of time-division multiplexing [11]. We are exploring AC biasing (∼10 kHz) which can enable frequency domain multiplexing [10]. High accuracy CMB spectral distortions (CMB-SD) is an emerging field that needs new technologies [12]. SOFTS can be introduced between the antenna and detector, e.g,. Fig. 1, whilst maintaining the general architecture of kilopixel arrays as discussed above. Such multiple simultaneous spectroscopic “eyes-on-the-sky” enables measuring CMB-SD. The SPT-3G comparative design above, which is not optimized for CMB-SD yields sensitivities of ∼10 Jy/sr, approaching other optimized mission concepts [12]. Similar sensitivity will allow SOFTS designed for THz operation3 to perform line intensity mapping (LIM) studies. Observational techniques between these fields overlap significantly and as discussed in the literature [13, 14]. While CMB Spectral distortion will probe very early universe physics, LIM will probe reionization physics and structure formation. Conclusion We have outlined detailed circuit modeling of Superconducting On-chip Fourier Transform Spectrometers (SOFTS) and discussed device design and hardware progress for Ka and W-band SOFTS, including the fabrication layout of SOFTS chips and their necessary optical coupling technologies. These bands were chosen based on the commercial availability of VNA and parts. However, we can rescale the hybrid and diplexer elements and retune the transmission line impedance, therefore SOFTS design is largely frequency band independent. Ultimately we intend for antenna and detector coupled SOFTS [7]. Furthermore we elucidated comprehensive RF cascade simulations of our complete devices, and demonstrated that such on-chip circuits have fractional errors in spectral recovery at levels of ≲10-11. Although due to COVID-19 device fabrication and testing has been delayed, our imminent work will involve measurements of both Ka and W-band devices. Acknowledgements We thank Robert Webber (Caltech) for pointing out noise spectral construction is possible with our framework [6]. We thank Rick LeDuc at JPL-MDL for device fabrication support. The W-band housing is being made using micro-mill machining by Matt Underhill at ASU. Undergraduate students C. Bell, E. Linden and E. Rapaport assisted with hardware assembly and participated through JPL/Caltech SURF. This research is supported by the NASA award NNH18ZDA001N-APRA, and by the University of Chicago College Summer Research Scholarship. 1 https://www.mathworks.com/help/simrf/. 2 This is not a comparison of scientific capabilities or sensitivities. We intended to outline SOFTS focal plane architecture in the context of the current kilopixel arrays. 3 New materials such as MgB2 are under R &D to explore SOFTS operations in the THz, as the superconductors discussed here may not be optimal. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Anthore A Pothier H Esteve D Density of states in a superconductor carrying a supercurrent Phys. Rev. Lett. 2003 90 12 127001 10.1103/PhysRevLett.90.127001 12688893 2. BasuThakur R Klimovich N Day PK Shirokoff E Mauskopf PD Faramarzi F Barry PS Superconducting on-chip fourier transform spectrometer J. Low Temp. Phys. 2020 200 5 342 352 10.1007/s10909-020-02490-7 3. Shu S Klimovich N Eom BH Beyer AD Basu Thakur R Leduc HG Day PK Nonlinearity and wide-band parametric amplification in a nbtin microstrip transmission line Phys. Rev. Res. 2021 3 023184 10.1103/PhysRevResearch.3.023184 4. A. Kogut, D. Fixsen, D. Chuss, J. Dotson, E. Dwek, M. Halpern, G. Hinshaw, S. Meyer, S. Moseley, M. Seiffert, D. Spergel, E. Wollack, The primordial inflation explorer (pixie): A nulling polarimeter for cosmic microwave background observations. 5. Liu M Pan Z Thakur RB Meyer SS Simulation and calibration of a compact millimeter-wavelength Fourier transform spectrometer Appl. Opt. 2020 69 25 7726 7733 10.1364/AO.397312 6. Halko N Martinsson PG Tropp JA Finding structure with randomness: Probabilistic algorithms for constructing approximate matrix decompositions SIAM Rev. 2011 53 2 217 288 10.1137/090771806 7. A. Steiger, The Design, Characterization, and Pixel Architecture of Superconducting On-chip Fourier Transform Spectrometers, Undergraduate Thesis, The University of Chicago, (2022). [Online]. Available: https://knowledge.uchicago.edu/record/3726 8. N. Klimovich, Traveling Wave Parametric Amplifiers and Other Nonlinear Kinetic Inductance Devices, Ph.D., California Institute of Technology, (2022). [Online]. Available: https://resolver.caltech.edu/CaltechTHESIS:05202022-191221394 9. Faramarzi F Day P Glasby J Sypkens S Colangelo M Chamberlin R Mirhosseini M Schmidt K Berggren KK Mauskopf P Initial design of a w-band superconducting kinetic inductance qubit IEEE Trans. Appl. Superconduct. 2021 31 5 1 5 10.1109/TASC.2021.3065304 10. Anderson AJ Ade PAR Ahmed Z Austermann JE Avva JS Barry PS Thakur RB Bender AN Benson BA Bleem LE Byrum K Carlstrom JE Carter FW Cecil T Chang CL Cho HM Cliche JF Crawford TM Cukierman A Denison EV de Haan T Ding J Dobbs MA Dutcher D Everett W Foster A Gannon RN Gilbert A Groh JC Halverson NW Harke-Hosemann AH Harrington NL Henning JW Hilton GC Holder GP Holzapfel WL Huang N Irwin KD Jeong OB Jonas M Khaire T Knox L Kofman AM Korman M Kubik D Kuhlmann S Kuklev N Kuo CL Lee AT Leitch EM Lowitz AE Meyer SS Michalik D Montgomery J Nadolski A Natoli T Nguyen H Noble GI Novosad V Padin S Pan Z Pearson J Posada CM Rahlin A Reichardt CL Ruhl JE Saunders LJ Sayre JT Shirley I Shirokoff E Smecher G Sobrin JA Stark AA Story KT Suzuki A Tang QY Thompson KL Tucker C Vale LR Vanderlinde K Vieira JD Wang G Whitehorn N Yefremenko V Yoon KW Young MR Spt-3g: A multichroic receiver for the south pole telescope J. Low Temp. Phys. 2018 193 5 1057 1065 10.1007/s10909-018-2007-z 11. L. Moncelsi, P. A. R. Ade, A. Z, M. Amiri, D. Barkats, R. B. Thakur, C. A. Bischoff, J. J. Bock, V. Buza, J. R. Cheshire, J. Connors, J. Cornelison, M. Crumrine, A. J. Cukierman, E. V. Denison, M. Dierickx, L. Duband, M. Eiben, S. Fatigoni, J. P. Filippini, N. Goeckner-Wald, D. Goldfinger, J. A. Grayson, P. Grimes, G. Hall, M. Halpern, S. A. Harrison, S. Henderson, S. R. Hildebrandt, G. C. Hilton, J. Hubmayr, H. Hui, K. D. Irwin, J. H. Kang, K. S. Karkare, S. Kefeli, J. M. Kovac, C. L. Kuo, K. Lau, E. M. Leitch, K. G. Megerian, L. Minutolo, Y. Nakato, T. Namikawa, H. T. Nguyen, R. O’brient, S. Palladino, N. Precup, T. Prouve, C. Pryke, B. Racine, C. D. Reintsema, A. Schillaci, B. L. Schmitt, A. Soliman, T. S. Germaine, B. Steinbach, R. V. Sudiwala, K. L. Thompson, C. Tucker, A. D. Turner, C. Umiltà, A. G. Vieregg, A. Wandui, A. C. Weber, D. V. Wiebe, J. Willmert, W. L. K. Wu, E. Yang, K. W. Yoon, E. Young, C. Yu, L. Zeng, C. Zhang, S. Zhang, Receiver development for BICEP array, a next-generation CMB polarimeter at the south pole, in Millimeter, Submillimeter, and Far-Infrared Detectors and Instrumentation for Astronomy X, eds. by J. Zmuidzinas, J.-R. Gao, vol. 11453, International Society for Optics and Photonics. SPIE, 2020, pp. 189 – 206. [Online]. Available: 10.1117/12.2561995 12. J. Chluba, M. H. Abitbol, N. Aghanim, Y. Ali-Haimoud, M. Alvarez, K. Basu, B. Bolliet, C. Burigana, P. de Bernardis, J. Delabrouille, E. Dimastrogiovanni, F. Finelli, D. Fixsen, L. Hart, C. Hernandez-Monteagudo, J. C. Hill, A. Kogut, K. Kohri, J. Lesgourgues, B. Maffei, J. Mather, S. Mukherjee, S. P. Patil, A. Ravenni, M. Remazeilles, A. Rotti, J. A. Rubino-Martin, J. Silk, R. A. Sunyaev, E. R. Switzer, New horizons in cosmology with spectral distortions of the cosmic microwave background, Exp. Astron. 51, 1515–1554 (2021). 10.1007/s10686-021-09729-5 13. E. Kovetz, P. C. Breysse, A. Lidz, J. Bock, C. M. Bradford, T.-C. Chang, S. Foreman, H. Padmanabhan, A. Pullen, D. Riechers, M. B. Silva, E. Switzer, Astrophysics and cosmology with line-intensity mapping, Bull. AAS, 51(3) (2019). https://baas.aas.org/pub/2020n3i101 14. P. Serra, O. Doré, G. Lagache, Dissecting the high-z interstellar medium through intensity mapping cross-correlations, Astrophys. J. 833(2), 153, (2016). https://iopscience.iop.org/article/10.3847/1538-4357/833/2/153	36474505	PMC9716503	NO-CC CODE	2022-12-03 23:20:57	no	J Low Temp Phys. 2022 Dec 2;:1-10	utf-8	J Low Temp Phys	2,022	10.1007/s10909-022-02920-8	oa_other
==== Front J Int Bus Stud J Int Bus Stud Journal of International Business Studies 0047-2506 1478-6990 Palgrave Macmillan UK London 564 10.1057/s41267-022-00564-0 Editorial Towards integrating country- and firm-level perspectives on intellectual property rights Cui Victor [email protected] 1Victor Cui is Conrad Research Excellence Chair and Associate Professor (Entrepreneurship, Innovation, and Global Strategy) at the Conrad School of Entrepreneurship and Business, University of Waterloo, Canada. His current research focuses on business strategy, technological innovation, and IPR protection. His research has been published in leading journals such as Journal of International Business Studies, Strategic Management Journal, and Research Policy. Narula Rajneesh [email protected] 2Rajneesh Narula is the John H. Dunning Chair of International Business Regulation at the Henley Business School, University of Reading, UK. His research and consulting have focused on the role of firms in development, inequality, innovation and industrial policy, R&D alliances, outsourcing and the informal economy . He is also Director of the Dunning Africa Centre. Minbaeva Dana [email protected] 34Dana Minbaeva is a Professor of Human Resource Management at King’s College of London, UK. She also has a part-time appointment at Copenhagen Business School, Denmark. Her research area is strategic international HRM. Professor Minbaeva received several national and international awards for research achievements, including the prestigious JIBS Decade Award 2013. Vertinsky Ilan [email protected] 5Ilan Vertinsky is Vinod Sood Professor of International Business Studies, Strategy and Business Economics in the Sauder School of Business at UBC. His current research foci are: (1) the relationships between interfirm collaboration, competition, IPR litigation, and innovation; (2) the US–China technology innovation rivalry; and (3) climate change resilience. 1 grid.46078.3d 0000 0000 8644 1405 Conrad School of Entrepreneurship and Business, University of Waterloo, Waterloo, Canada 2 grid.9435.b 0000 0004 0457 9566 Henley Business School, University of Reading, Reading, UK 3 grid.13097.3c 0000 0001 2322 6764 King’s Business School, King’s College London, London, UK 4 grid.4655.2 0000 0004 0417 0154 Copenhagen Business School, Frederiksberg, Denmark 5 grid.17091.3e 0000 0001 2288 9830 Sauder School of Business, The University of British Columbia, Vancouver, Canada 2 12 2022 2022 53 9 18801894 29 4 2022 24 8 2022 25 8 2022 © Academy of International Business 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Intellectual property rights (IPRs) are of critical importance in international business. The implications for firm strategy and for policymakers are rarely aligned because the optimal level of IPR protection can be quite different from the country- and the firm-level perspectives. There is considerable heterogeneity in firm strategies, the spatial distribution of their innovation activities, and their IPR portfolios. There is still greater variation between countries, their IPR legislation and enforcement efforts, as well as their industrial and development policies. For firms, sustaining firm-specific advantages (FSAs) depends on their ability to create and extract rent from their knowledge assets, and this involves deliberate interfirm cooperation, careful location choices, and talent recruitment and retention. At the country level, the attractiveness of countries for MNEs is shaped by the provision of country-specific advantages such as IPR protection and its effective enforcement, but the kinds of IPR regimes that are optimal to attract inward investment can be disadvantageous for building domestic firm capacity, and vice-versa. Although firm IPR strategies and IPR regimes are clearly interlinked, the literature integrating across these two levels has been underdeveloped, and we propose a framework to guide future research. Résumé Les droits de propriété intellectuelle (Intellectual property rights - IPRs) sont d'une importance capitale dans les affaires internationales. Les implications pour la stratégie des entreprises et pour les décideurs politiques sont rarement alignées, car le niveau optimal de protection des IPRs peut être très différent selon que l'on se place au niveau du pays ou de l'entreprise. Il existe une hétérogénéité considérable dans les stratégies des entreprises, la distribution spatiale de leurs activités d'innovation et leurs portefeuilles de IPR. Les variations sont encore plus grandes entre les pays, leurs législations en matière de IPR, leurs efforts d'application, ainsi que leurs politiques industrielles et de développement. Pour les entreprises, le maintien des FSAs dépend de leur capacité à créer et à extraire une rente de leurs actifs intellectuels, ce qui implique une coopération interentreprises délibérée, des choix de localisation judicieux, ainsi que le recrutement et la rétention des talents. Au niveau du pays, l'attractivité des pays pour les entreprises multinationales est déterminée par la fourniture de leurs avantages spécifiques tels que la protection des IPRs et leur application efficace. Néanmoins, les types de régimes de IPR qui sont optimaux pour attirer les investissements étrangers peuvent être désavantageux pour la construction des capacités des entreprises domestiques, et vice-versa. Bien que les stratégies des entreprises en matière de IPR et les régimes de IPR soient clairement liés, la littérature intégrant ces deux niveaux a été sous-développée ; aussi, nous proposons un cadre pour guider de futures recherches. Resumen Los derechos de propiedad intelectual (DPI) tienen una importancia fundamental en los negocios internacionales. Las implicaciones para la estrategia de las empresas y para los diseñadores de políticas no suelen estar alineados, ya que el nivel óptimo de protección de los DPI puede ser muy diferente desde el punto de vista del país y de la empresa. Existe una considerable heterogeneidad en las estrategias de las empresas, la distribución espacial de sus actividades de innovación y sus portafolios de DPI. Todavía hay gran variación entre los países, su legislación en materia de legislación y hacer respetar los DPI, así como sus políticas industriales y de desarrollo. Para las empresas, el mantenimiento de las ventajas especificas empresariales (FSAs por sus iniciales en inglés) depende de su capacidad para crear y extraer rentas de sus activos de conocimiento, y esto implica una cooperación deliberada entre empresas, una cuidadosa elección de la ubicación y la contratación y retención del talento. A nivel de país, el atractivo de los países para las empresas multinacionales depende de las ventajas específicas del país, como la protección de los derechos de propiedad intelectual y su efectividad para hacer respetar los DPI, pero los tipos de regímenes de derechos de propiedad intelectual que son óptimos para atraer la inversión extranjera entrante pueden ser desventajosos para el desarrollo de la capacidad de las empresas nacionales, y viceversa. Aunque las estrategias de DPI de las empresas y los regímenes de DPI están claramente interrelacionadas, la bibliografía que integra estos dos niveles está poco desarrollada, y proponemos un marco para orientar la investigación futura. Resumo Direitos de propriedade intelectual (IPRs) são de importância crítica em negócios internacionais. As implicações para a estratégia da empresa e para formuladores de políticas raramente estão alinhadas porque o nível ótimo de proteção a IPR pode ser bem diferente das perspectivas no nível do país e da empresa. Há considerável heterogeneidade nas estratégias de empresas, na distribuição espacial de suas atividades de inovação e em seus portfólios de IPR. Há variação ainda maior entre países, sua legislação de IPR e esforços de observância (da lei), bem como suas políticas industriais e de desenvolvimento. Para empresas, sustentar FSAs depende de sua capacidade de criar e extrair renda de seus ativos de conhecimento, e isso envolve cooperação deliberada entre empresas, cuidadosas escolhas de localização e recrutamento e retenção de talentos. No nível nacional, a atratividade de países para MNEs é moldada pela oferta de vantagens específicas do país, como proteção a IPR e sua efetiva observância, mas os tipos de regimes de IPR que são ideais para atrair investimentos podem ser desvantajosos para a formação de. capacidade doméstica de firmas e vice-versa. Embora estratégias de IPR de empresas e regimes de IPR sejam claramente interligados, a literatura que integra esses dois níveis tem sido pouco desenvolvida e propomos um modelo para orientar pesquisas futuras. 知识产权 (IPR) 在国际商务中至关重要。对公司战略和政策制定者的影响很少是一致的，因为IPR保护的最佳水平可能与国家和公司层面的观点颇为不同。公司战略、其创新活动空间分布以及其IPR组合存在相当大的异质性。各国之间、其IPR立法和执法力度以及其产业和发展政策之间仍然存在较大变异。对于公司而言，维持公司特有的优势(FSA) 取决于他们创造和从知识资产中提取租金的能力，这涉及到公司间的深思熟虑的合作、谨慎的地点选择以及人才的招聘和保留。在国家层面，各国对跨国公司（MNE）的吸引力取决于国家特有优势的提供，例如IPR保护及其有效执法，但最能吸引外来投资的IPR制度可能不利于用于建设国内公司的能力，反之亦然。尽管公司的IPR战略和IPR制度显然相互关联，整合这两个层面的文献还欠开发，我们因而提出了一个指导未来研究的框架。 Keywords intellectual property rights policies cooperation mobility location FSAs innovation issue-copyright-statement© Academy of International Business 2022 ==== Body pmcINTRODUCTION Intellectual property rights (IPRs) are one of the most critical building blocks of modern society. According to the World Trade Organization (WTO), they are ‘the rights given to persons over the creations of their minds. They usually give the creator an exclusive right over the use of his/her creation for a certain period of time’.1 Where individual creators are employed by an organization, these rights may belong (in part or in whole) to the organization with which these individuals are affiliated. IPRs are commonly divided into two areas: first, copyright and rights related to copyright, and second, industrial property, including distinctive signs, patents, industrial design, and trade secrets. It is this second category that has garnered the most interest for IB, strategy, and policymaking. The assignation and protection of IPRs plays a fundamental role in shaping both the political economy and competitiveness of countries, and in the success and failure of firms. However, there is considerable conflict and variation in the extent to which intellectual property is protected, partly because the interests of the nation state and firms (whether multinational or otherwise) are rarely aligned. Even collocated firms in the same industry may have diverging attitudes to IPRs, and different countries often have heterogenous policy objectives, each viewing the implementation and enforcement of IPRs as a significant strategic tool to be wielded in disparate ways. It is to be emphasized that MNEs are not simply passive users of IPR regimes, they are actively engaged in shaping and influencing these regimes, as active actors within the milieu, to align to their own interests. The conflict during the Covid-19 pandemic over the equitable distribution of vaccines illustrates this well (Guimón & Narula, 2020). The advanced economies (and home to the major pharmaceutical MNEs) were unwilling to waive IPRs on the new vaccines, while much of the developing world sought to either partially or fully waive IP enforcement of Covid-19 technologies, to allow universal access to IP and more equitably located production capacity. Indeed, despite considerable negotiation, as of mid-2022, no consensus has been achieved, either among or between the MNEs, the home countries of these MNEs, or among the developing countries themselves, despite the active engagement of a variety of supranational institutions, NGOs, and interest groups (Jecker, 2022; Jecker & Atuire, 2021). It is self-evident, therefore, that IPRs are a critical field of study in international business, strategy, and economics, and that establishing and enforcing IPRs form a central aspect of the strategies and policies of MNEs and governments. Despite the ubiquitous nature of IPRs, and their implications at the firm, subnational, national, and supranational levels, much of the academic work in international business and related disciplines has tended to be single-leveled, in terms of the antecedents/determinants and the outcomes of IPRs. For example, macro-level research has primarily paid attention to policy determinants of IPR protection, such as political systems, formal and informal institutions, and the international rule of law (e.g., Brander, Cui, & Vertinsky, 2017; Ginarte & Park, 1997; Maskus, 2014; Mertha, 2007; Saggi, 2002; Peng, Ahlstrom, Carraher, & Shi, 2017). Likewise, many micro-level studies have focused on various firm-level strategies to protect intellectual property from misappropriation (Hamel, Doz, & Prahalad, 1989; Inkpen, Minbaeva, & Tsang, 2019; Narula & Santangelo, 2009; Shaver & Flyer, 2000). Only a few studies have investigated IPR protection across levels. They have mainly focused on the influence of differing country-level IPR regimes on the strategies of firms, their effect on the optimal alliance governance structures, how internal knowledge linkages can be most effective, and the effectiveness of the modularization of innovation (see, e.g., Alcácer & Zhao, 2012; Baldwin and Henkel, 2015; Oxley, 1999; Shi, Sun, Pinkham, & Peng, 2014; Zhao, 2006). Firm-level research has considered the macro aspects of IPR as exogenous, and vice versa, and rarely capture the complexity of macro- and micro-level interactions adequately. The original call for papers for this special issue sought to build a more comprehensive framework that specifies the underlying mechanisms that integrates and connects multiple levels. For example, the “optimal” level of IPR protection that balances between attracting foreign direct investment (FDI) and developing indigenous innovative capacities differs between countries at different stages of technological development (Criscuolo & Narula, 2008). Discrepancies exist between the formalization and implementation of IPR institutions in home- and host-countries. MNEs actively seek to influence country policies, by engaging in various levels of regulatory capture. Such macro-level characteristics have challenged assumptions underlying micro-level IPR studies, which have presumed a homogenous IPR environment. We invited papers that explicitly addressed and developed multilevel explanations. We encouraged submissions that theoretically accommodated a nested, complex and adaptive system view on IPR in MNEs. In the rest of the paper, we review studies on IPRs as a key source of firm-specific advantages (FSAs), based on the IB, strategy, and organizational behavior literatures. We focus on three interconnected aspects in creating and augmenting FSAs: interfirm cooperation, location choices, and talent mobility. We then turn our attention to IPRs as a critical component of location advantages, based on the IB and economics literatures. In particular, we highlight the development challenge of IPR policies that strongly protect FSAs. Such policies can be disadvantageous for developing countries interested in building domestic firm competitiveness, while being useful to attract inward MNE investment. Adapting and incorporating these findings and theories of single-level IPR research from various disciplines, we then develop a conceptual framework for IPR research across firm- and country-levels. We identify gaps in single-level research and suggest some plausible research avenues for multilevel analysis. OVERVIEW OF THE ARTICLES IN THIS SPECIAL ISSUE What is imminently clear from this special issue is that these two perspectives - the firm-level and the country-level – are inextricably linked. The papers in this special issue illustrate this well. The paper by Genin, Tan and Song (2022) provides nuanced insights into host-country IPR regimes, state intervention, and effectiveness of joint ventures (JVs) in curbing opportunism. Specifically, they focus on the powerful role that governments can play in the governance of an R&D collaboration in developing countries with an active policy to leverage MNEs in building up domestic capacity. Their focus is China’s objective of developing a high-speed train industry, a sector where there is considerable state ownership. The Chinese government had hoped that setting up a JV with leading foreign firms in the sector would permit the rapid internalization of foreign advanced technologies by domestic actors. The Chinese state has used collaboration as a useful mechanism to leapfrog in the past, providing strong support and resources to back domestic actors’ efforts. However, in this case, there was considerable distrust by the foreign partners who were concerned about misappropriation, leading them to refrain from sharing tacit knowledge, and thus significantly affecting the success of the collaboration. Genin et al. (2022) highlight that the government as a partner can inadvertently become a liability, and that government support to boost domestic innovation in host countries with weak IPR regimes can negatively affect international cooperation if the state acts opportunistically within the collaboration. Yan, Li and Zhang (2022) shed light on how MNEs match internal IPR protection mechanisms and host-country IPR regimes in their location choices. Specifically, they examine how the global pharmaceutical industry uses the small-world structure of internal social and knowledge networks to create IPR protection mechanisms, as a substitute for inadequate external IPR regimes in host countries. They argue that the small-world structure of internal collaborative networks can result in social complexity in generating new technologies, which increases the difficulty for other firms to interpret and imitate technologies by means of inventor recruitment. In contrast, the small-world aspect of internal knowledge networks increases the interrelatedness among knowledge sets, making it easier for competitors to misappropriate a foreign firm’s core knowledge asset by getting access to only fractions of its knowledge basis. These findings showcase important adaptive strategies when MNEs face weak IPR regimes. The paper by Bruno, Crescenzi, Estrin and Petralia (2022) highlights the role that location choices and institutional distance play in the important relationship between MNEs’ inventive capabilities and innovation performance. They find that this relationship is contingent on the IPR regime distance between the host and home countries. Overall, innovation performance decreases with IPR regime distance, because the costs of coordination and administration of R&D subsidiaries across different regimes increase with IPR distance. Such a negative effect can be asymmetric, depending on the relative strength of IPR protection between the host and home countries. The negative effect is stronger when MNEs locate their R&D activities in host countries with less strict IPR protection than their home countries. This is because the cost of setting up internal defense mechanisms to protect MNE knowledge from being misappropriated in a weak IPR environment outweighs the potential benefits from institutional arbitrage in such a host environment. In contrast, when MNEs establish R&D subsidiaries in countries with stronger IPR protection, the negative effect is reduced. This is because the MNEs rely less on internal mechanisms, than on the host-country legal system to protect its core knowledge, and because such an environment offers more opportunities for knowledge acquisition and learning. Overall, from an IB perspective, that there is an interaction between FSAs and location advantages is unsurprising, given what we know from the various strands of IB theory (Asmussen, Chi, & Narula, 2022; Forsgren & Holm, 2021; Narula, Asmussen, Chi, & Kundu, 2019). However, one of the objectives of this special issue was to demonstrate that, by explicitly bringing together firm-level and country-level research, we can deepen our understanding about how the creation, acquisition, and protection of IPRs affects international business. In the rest of this introductory paper, we aim to explicate the benefits of intentionally introducing multilevel theorizing around the antecedents/determinants and outcomes of IPR, instead of discussing those for the purposes of post hoc contextualization. Specifically, we focus on the multilevel impact of IPR regimes and MNEs’ IPR strategies on key IB decisions and outcomes, by integrating firm- and country-level IPR perspectives. While the scope of multilevel analysis can range across supernational, national, subnational, industry, firm, and individual levels, our objective in this paper is not to exhaust all multilevel possibilities, but to showcase a few plausible approaches. We explore opportunities to build an IB-specific IPR-related research agenda, based on both multilevel analysis and an interdisciplinary approach. FIRM-LEVEL IPR STRATEGIES: OPTIMIZING MNE FSA PORTFOLIOS Intellectual property lies at the heart of the competitiveness of firms, and the absence or presence of the associated rights plays a critical role in defining the capacity of MNEs to invest in asset creation and augmentation, both through formal and informal R&D. Where MNEs are unable to ensure exclusive use of their IP so as to (at least) recover the costs of R&D and the associated risks of such investment, they lose a key incentive to innovate. Indeed, a technological asset becomes a FSA only if the firm can impede the appropriation of its IP by its competitors. It is therefore in the interests of the MNE to reduce unintended leakages of firm-specific IP. Firms may choose to protect their IPRs through a number of means. Although much of the literature emphasizes patents, copyrights, and brands, by far the simplest and most common way is secrecy (for a discussion, see Arundel, 2001; Arundel & Kabla, 1998; Cohen, Goto, Nagata, Nelson, & Walsh, 2002). In the absence of legally enforceable IPR protection, MNEs can mitigate the risk of IPR violation through secrecy, and capture value from their R&D, by resorting to internal organizations that enable them to efficiently transfer, integrate, and build on technologies developed in other IPR regimes (Zhao, 2006). A growing share of FSA creation and augmentation takes place through formal and informal collaboration, either with suppliers and customers, or with competitors, in a variety of quasi-internal governance arrangements. Indeed, many firms manage their internal and external activities as indivisible aspects of an open innovation portfolio (e.g., Narula, 2001; Van de Vrande, Vanhaverbeke, & Gassmann, 2010). Such portfolios rely on intentional knowledge exchanges, and are on balance mutually beneficial, providing both cost-economizing and strategic benefits (Narula & Martinez-Noya, 2015). Nonetheless, optimal levels of innovation depend on both deliberate and unintentional knowledge flows. While all firms seek to maximize the return on their own FSAs, they are also dependent on knowledge that resides outside the boundaries of their own organization to develop new FSAs. Thus, strongly enforced and protected IPRs can act as a retardant to firm-level innovation. As we will discuss later, the optimal IPR regime for firms interested in FSA creation is one that permits informal and formal knowledge exchanges, and not one that necessarily offers strong, excessively broad or long property rights. Indeed, overly strong IPR regimes can deter innovation by firms in that location (Acemoglu & Akcigit, 2012; Dasgupta & Stiglitz, 1980). The early IB literature was focused on exploitation of FSAs, and the original thesis underlying most of the early work was that the raison d’etre of the MNEs derived from choosing between internalizing the use of its FSAs within its own organization, versus licensing or selling its IP-related assets to other firms (see Narula et al., 2019 for a review). The literature of the last few decades has come to recognize that MNEs also go abroad to seek and augment their FSAs, and, in order to do so, have developed a variety of hybrid governance mechanisms (beyond arms-length sales, licensing, or exchanges) to engage in interfirm cooperation because of the need for cutting-edge competencies in a large number of fields, and the associated cost of maintaining a high level of competence in multiple technological areas (for a review, see Martinez-Noya & Narula, 2018). Furthermore, building up FSAs depends on choosing the ‘right’ location, not (only) in the sense of location advantages (which we will discuss at length in our discussion of country-level issues) but in the sense of choosing the appropriate location relative to other locations with similar location advantages, known as determining location choice. Finally, FSAs ultimately depend on individuals as the key vector in knowledge creation, and how knowledge exchanges between MNEs are shaped by inventor mobility and talent recruitment. We discuss these three aspects next. Interfirm Cooperation Interfirm R&D alliances have become a key mechanism through which firms cooperate to create and protect knowledge (Martinez-Noya & Narula, 2018). While firms cooperate to create value, they often simultaneously appropriate proprietary knowledge from each other (Hamel, 1991). Alliances create unique challenges for jointly creating IP, due to the risks of opportunism (see Hoffmann, Lavie, Reuer, & Shipilov, 2018 for a review). Opportunism in interfirm cooperation depends on a variety of factors, such as the interplay of private and common benefits of partners (Khanna, Gulati, & Nohria, 1998), bilateral and multilateral rivalry in alliances (Lavie, 2007), asymmetrical power and learning capacity of partners (Wang, Wang, Jiang, Yang, & Cui, 2016; Yang, Zheng, & Zaheer, 2015), and cooperation structure (Cui, Yang, & Vertinsky, 2018; Gulati, 1995a, b; Polidoro, Ahuja, & Mitchell, 2011). R&D alliances typically demand intensive knowledge sharing, frequent interpersonal interactions, and long-term commitment to projects, which normally would involve high risk of misappropriation (Gulati, 1995a). However, recent research suggests that the level of opportunism may not be necessarily high in R&D alliances (Cui et al., 2018). In fact, it is highest when the cooperation portfolio between partners is balanced between exploratory (e.g., R&D alliances to create new knowledge) and exploitative alliances. In contrast, when the cooperation portfolio is dominated by exploratory alliances, partners are more likely to adopt a long-term outlook that cultivates relationship building, which minimizes the level of opportunism (Cui et al., 2018). To prevent opportunism, firms tend to show a preference for selecting familiar partners, because trust lowers transaction costs and increases information sharing (Dyer & Chu, 2003; Gulati, 1995a; Li, Eden, Hitt, & Ireland, 2008; Zaheer, McEvily, & Perrone, 1998). Hoetker (2005) showed that, as technological uncertainty increases, prior relationships take on greater positive significance relative to the importance of potential learning opportunities from unfamiliar partners. To curb opportunism, equity agreements such as JVs were the conventional governance form employed in early R&D collaboration to better align partners’ interests, by setting up a separate jointly-owned legal entity (e.g., Gulati, 1995a; Ryu, McCann, & Reuer, 2018; Hagedoorn & Narula, 1996). It is noteworthy, however, that the need for equity-based agreements varied (and continues to vary) considerably across sectors, reflecting the extent of technological uncertainty, the IPR environment of the firms, and the rapidity of technological change in the sector. For instance, the adoption of JVs has been declining in fast-moving sectors where firms do not have the luxury to engage in long-term equity partnerships (Hagedoorn, 2002; Martinez-Noya & Narula, 2018). In fast-moving industries and sectors, there is a strong preference for non-equity agreements, although there is some variation when dealing with MNEs from home countries with weak IPR protection. When operating in locations offering low protection of IPRs, MNEs tend to fragment the operations entrusted to foreign units, assigning activities with a less strategic content as a way to reduce misappropriation problems (Belderbos, Park, & Carree, 2021; Gooris & Peeters, 2016). Overall, however, there has been a shift towards use of contractual and non-equity agreements, especially as firms increasingly adopt open innovation strategies (Kranenburg, Hagedoorn, & Lorenz-Orlean, 2014; Santamaria, Nieto, & Barge-Gil, 2010). The development of a systemic and explicit set of the firm strategies needed for an optimal open innovation strategy requires understanding how firms orchestrate multiple agreements. Careful contract design of each agreement is critical, given the complex overlapping nature of such agreements and the various IPR regimes involved because of the cross-border nature of the activities they cover (Bogers, Zobel, Afuah, Almirall, Brunswicker, Dahlander, Frederiksen, Gawer, Gruber, Haefliger, & Hagedoorn, 2017; Contractor & Reuer, 2014). Balancing between intentional and unintended knowledge transfers within the confines of a formal collaborative agreement is an area of considerable interest in IB, strategy, and innovation studies, and indeed it is well acknowledged that firms may utilize alliances as a mechanism to control knowledge leakages, rather than (or as well) as a mechanism to promote them (Narula & Santangelo, 2009). Researchers have found that the structure of cooperative agreements makes a difference in the effectiveness in managing knowledge transfers and curbing opportunism in interfirm cooperation (Hagedoorn, 2002; Martinez-Noya & Garcia-Canal, 2018; Narula & Martinez-Noya, 2015). Location Choices In addition to accessing ‘traditional’ location-specific advantages, collocation of key suppliers, competitors, and customers matters in enhancing the FSAs of firms (Anand, McDermott, Mudambi, & Narula, 2021; Castellani, 2018; Castellani & Lavoratori, 2020; Papanastassiou, Pearce, & Zanfei, 2020). Knowledge transfers are highly sensitive to geographic proximity, and this is especially so where the knowledge being exchanged is tacit in nature (Cantwell & Santangelo, 1999). Firms seek to locate their R&D activities in spatial proximity to competitors, because knowledge spillovers tend to be geographically localized (Jaffe, Trajtenberg, & Henderson, 1993). Prior studies have found that technology-intensive firms tend to locate in regions with high R&D intensity, and that industry followers actively seek to locate close to the industry leaders (Chung & Alcácer, 2002). As knowledge leakage is a two-way process, scholars have suggested that an optimal location should balance between the gains from inward knowledge flows and the costs of outward flows to competitors (Alcácer & Chung, 2007; Shaver & Flyer, 2000). Knowledge leakage from, to, and between firms in the same location is often unintentional and unavoidable, due to highly embedded interfirm relationships and frequent employee mobility. Inkpen et al. (2019) also suggest that knowledge leakages can be beneficial when there is a reciprocal exchange. Whether the net flow is negative or positive depends on a number of conditions. The optimal level of exchange is a function of the mix of intellectual property in the knowledge portfolio of the firms concerned, their absorptive capacity in specific technologies, and the extent to which the product or service can be fine-sliced without compromising the integrity of the final product or service. Indeed, lead firms in any given sector are keen to avoid collocation with follower firms to minimize unintended knowledge flows, and engage in interfirm cooperation as a defensive means to minimize leakages when collocation is unavoidable (Narula & Santangelo, 2009). Conversely, follower firms actively seek collocation with leader firms. Given that the production of most products and services involves multiple technologies, and that it is exceedingly rare for a single firm to be a technological leader in all the constituent technological fields, collocation and knowledge flows are a practical reality. The decision of with whom to partner can be connected with the location of the alliance partner, and, while firms should, in principle, seek partners with the most appropriate FSA portfolio, researchers have consistently found a preference for geographically proximate partners, because it offers the advantage of facilitating control when misappropriation hazards are high (Reuer & Lahiri, 2014). Collocation of a firm’s partners and rivals can also introduce potential indirect paths of knowledge leakage to rivals, and, as such, further increase the risk that the firm’s knowledge may be misappropriated (Ryu et al., 2018). Inventor Mobility and Talent Recruitment The very basic inputs for innovation are individual ‘inventors’ (e.g., scientists and engineers), who work individually or in teams in a formal setting within an R&D facility of a firm, or innovate ‘informally’ as part of other functions. Building up FSAs is to an extent dependent on the mobility of inventors (Andersson, Brewster, Minbaeva, Narula, & Wood, 2019). The literature on inventor mobility forms a rich body of firm-level research related to IPRs in MNEs. Researchers have argued that inventor mobility is a major source of knowledge flows between firms and locations (Agarwal, Echambadi, Franco, & Sarkar, 2004; Almeida & Kogut, 1999; Mawdsley & Somaya, 2016; Schaefer, 2020; Singh, 2007). In broad brush strokes, this line of literature can be categorized into two substreams. The first substream investigates the antecedents to inventor mobility at various levels, e.g., individual, job, firm, and job market. For example, at the individual level, inventor mobility is influenced by such attributes as demographics, personalities, motivation, and portable assets (e.g., abilities, skills, and social capital). At the job level, inventor mobility could be influenced by instrumental communication, job security, conflict, task complexity, etc. At the firm level, inventor mobility is affected by such factors as centralization, culture, reputation, support, and compensation. At the job market level, inventor mobility is determined by the availability of alternatives (for a thorough review, see Mawdsley & Somaya, 2016; Rubenstein, Eberly, Lee, & Mitchell, 2018). The second substream of this literature focuses on the impact of inventor mobility on knowledge transfer, i.e., the “learning-by-hiring” effect (see Mawdsley & Somaya, 2016, for a review). Hiring firms enjoy access to new technological knowledge because newly hired inventors retain their prior social and knowledge networks, resulting in post-movement knowledge transfer. Scholars have found that the amount of knowledge acquired from the recruited talents is influenced by characteristics of the hiring firms, such as knowledge diversification (Slavova, Fosfuri, & De Castro, 2016), positioning of the recruits (Song, Almeida, & Wu, 2003), and the tenure of incumbent scientists (Slavova et al., 2016), as well as the relationship between the current and previous employers (Almeida & Kogut, 1999). Researchers have also studied a number of mechanisms to curb knowledge outflows via inventor turnover, such as compensation, litigation, and non-compete covenants (Agarwal, Ganco, & Ziedonis, 2009; Ganco, Ziedonis, & Agarwal, 2015; Wang, He, & Mahoney, 2009; Younge, Tong, & Fleming, 2015). As employee mobility is an important reason for localized knowledge spillover (Almeida & Kogut, 1999), by curbing inventor turnover, firms can control, to some extent, knowledge outflows to geographically proximate firms by careful use of alliances, and the associated use of complex contracts and non-disclosure agreements. However, the restrictive covenants in such agreements may limit the possibilities of knowledge leakage for a certain time after the inventors leave a job, firms can never completely eliminate leakages. COUNTRY-LEVEL IPR REGIMES: LOCATION ADVANTAGES AND DUALITY FOR POLICY The attractiveness of countries for MNEs is shaped by their provision of location-specific advantages. Legislation and policies relating to IPRs are critical sources of location advantages (Narula & Santangelo, 2012). IPR legislation and enforcement provide the formal and informal institutions to govern (and bolster) domestic economic activity. They also send an important ‘signal’ to foreign investors, because they indicate the openness of a market and the nature of its institutions, even for investors that are not necessarily affected by IPR policies (Javorcik, 2004; Lall, 1997). IPRs matter more for MNEs in ‘IPR sensitive’ sectors, such as in knowledge-intensive industries, and less for more mature sectors and resource-intensive ones. Similarly, IPR matter more for R&D-related investments, but less so for market- and resource-seeking investments (Narula, 2022; Nunnenkamp & Spatz, 2004). The Duality of IPRs for Policy: Conflicting Interests The role of the nation state is crucial in IPR, as it is responsible for mandating the legal basis for IPR, as well as being responsible for enforcing the rights of individuals, firms, and other actors. Governments seek ideally to balance possible conflicts of influence from both domestic interest groups (e.g., political parties, industry associations, worker unions, and large firms) and foreign stakeholders (e.g., geopolitical allies, bilateral and multilateral agreement partners, and MNEs) (Athreye, Piscitello, & Shadlen, 2020). IPR policy and its enforcement form an invaluable tool to shape the competitiveness of firms within its borders, and the state has considerable leeway in utilizing IPR law to strengthen or weaken the FSAs of specific groups of actors. It can do so by legislating or omitting to legislate legally binding IPRs (i.e., software patents are not similarly protected in all countries) or by only selectively enforcing IPRs (either for strategic reasons, or where there is an absence of effective regulatory capacity). Both mechanisms can be used by countries with active industrial and FDI policies as a means to strengthen the FSAs of domestic firms relative to those of foreign MNEs operating in its domain. In some circumstances, where the domestic sector is weak, the capacity to enforce IPRs may serve as a tool to attract inward MNE activity. However, not all development policy regimes rely on MNE investment, and there is considerable heterogeneity in MNE-assisted development regimes. The ‘optimal’ level of protection that balances between attracting MNE investment and developing domestic firms’ innovation capacities differs between countries at different stages of technological development, and within sectors over time (Narula & Dunning, 2010; Narula & Pineli, 2019). While stricter IPR enforcement tends to be more common among advanced economies with strong innovation capabilities, strong IPR can hinder the development of technological capacity and FDI inflows for developing countries seeking to catch up (Chen & Puttitanun, 2005; Kim, Lee, Park, & Choo, 2012; May & Sell, 2006; Wu, Ma, & Zhuo, 2017). Not surprisingly, while nation states establish IPR legislation as part of international IPR protection agreements (e.g., TRIPS), they differ significantly with respect to the actual implementation and enforcement of these laws (Papageorgiadis & McDonald, 2022). States may selectively comply with their obligations to protect foreign IPRs to maximize strategic goals for technological leapfrogging (Brander et al, 2017). IPRs may be waived or strengthened in a specific sector, or only for a specific group of firms, in an effort to build up national champions to create and strengthen a cluster of national competitiveness. States may intentionally weaken the competitiveness of foreign-owned firms, by making their technological assets easier to imitate for domestic actors, as part of an industrial or development policy. Alternatively, policies may strengthen the IPRs of domestic firms by giving them IP protection of unusual length and breadth. While some scholars argue that IPR protection eventually becomes stronger as the domestic technological capacity grows (Acemoglu, Aghion, & Zilibotti, 2006; Kalaycı & Pamukcu, 2014; Narula, 2022; Sweet & Eterovic, 2019; Sweet & Maggio, 2015), other scholars suggest that, without strong external coercion, poor or selective IPR enforcement will continue in developing countries (Brander et al., 2017). Developing countries may also encounter resource constraints to enforcing international IPR regulations. Few developing countries have the legal manpower to properly interpret and judge patent applications by MNEs, or to determine infringements. Where foreign patents infringe upon unpatented intellectual property, such as local traditional medicines or extracts from plants, local actors cannot afford to protect and enforce these IPRs (Giuliani, Jacqueminet, & Nieri, 2022). Even wealthier developing countries with greater resources and significant numbers of skilled bureaucrats, such as China or India, find it challenging (Papageorgiadis & McDonald, 2022). Resource constraints in the public sector, and in particular the absence of skilled and expert IPR legal experts to evaluate sophisticated IP in most developing countries, will continue to impede enforcement. In general, for governments, MNE-related IPR issues are rarely a priority compared to addressing key economic problems, such as endemic unemployment and poverty (Narula, & van der Straaten, 2021). International institutions that affect IPR protection are constantly evolving as the international business contexts change over time. As various forms of digital trade (e.g., e-commerce, digitalization of physical products, and cloud computing) play an increasingly critical role in international trade, it has become critical to effectively govern digital trade globally (Athreye et al., 2020). Crucial conditions for the functioning of the digital economy – such as maintaining free flow of data, maintaining security within digital ecosystems, and protection of source code – inevitably have consequences for IPR violation and protection. There has been a growing divide within the WTO between the advanced countries (led by the US and Europe) and the developing countries (led by India, South Africa, and Brazil) on issues associated with the right to establish IPRs for biologics. Other considerable disagreements proliferate around such issues as data localization, filtering and privacy, strategic limits to IPRs for development purposes, and the use of geographical indications (Azmeh, Foster, & Echavarri, 2020). LINKING COUNTRY-AND FIRM-LEVEL PERSPECTIVE: A FRAMEWORK FOR FUTURE RESEARCH The leitmotif of this paper is that prior research investigating IPR at country and firm levels have not yet been fully integrated. From the firm perspective, their IP strategies are not developed and implemented in a vacuum, but are necessarily adapted to the context provided by the IPR regimes in the home and various host countries. Firms, both individually and as part of various formally and informally organized interest groups, therefore strive to influence country-level (and occasionally, subnational) IPR policy formation and implementation. From the perspective of policymakers, the establishment, implementation, and reformation of IPR regimes are influenced by the interests of both domestic firms and MNEs, with respect to their IPR strategies and practices. In addition, IPR policies of nation states are rarely static, and ideally evolve with the competitiveness and economic structure of the country, in tandem with the development path and strategic objectives of other policies that address the competitiveness of the location and its locally embedded economic actors. As a variety of commentators have noted, IPRs have been a key aspect of development policy, since at least the first industrial revolution (Chang, 2002; Malerba & Lee, 2021; Narula, 2022). A research framework, depicted in Figure 1, integrates these two levels of IPR research. In this framework, we take an IB perspective, placing MNEs at the heart of the analysis. We therefore also take an MNE-centric view of the interactions between firm-level IPR strategies (interfirm cooperation, location choices, inventor mobility) and country-level IPR institutions (home, host, and home–host distance). We suggest that macro- and micro-level factors interact with each other through two processes: a top–down process, i.e., country IPR institutions cascading down to the firm level, influencing MNE decisions and performance, and a bottom–up effect, i.e., firm-level IPR practices driving country-level IPR regime evolution. Next, we describe the research agenda contained in this framework with a few examples to showcase plausible research areas to be fully explored.Figure 1 A research framework for studying IPRs in IB. The Top–Down Process IPR regimes and interfirm cooperation A key area that deserves further attention is the interaction between country IPR regimes (host and home) and interfirm cooperation. Despite ample research on interfirm cooperation, we have not fully understood how IPR regimes might influence core issues in interfirm cooperation, such as MNE alliance formation, alliance portfolio management, governance structure, and the interplay between cooperation and competition. One area of opportunity is to examine the effect of host-country IPR regimes on international cooperation. For instance, weak IPR regimes in host countries could influence local firms and MNEs from strong IPR regimes asymmetrically, with respect to private versus common interests dynamics (Arslan, 2018; Khanna et al., 1998), bilateral and multilateral rivalry (Lavie, 2007), and the transition from cooperative to competitive relationships (Cui et al., 2018). Earlier work on cooperation and collaboration focused on equity JVs, which are no longer the predominant form of governance structure in international cooperation. In a rapidly evolving global milieu that is increasingly dominated by global value chains (GVCs), we now need to ask which governance structures best limit opportunism, and to what extent are specific structures relevant in host countries with poor IPR enforcement and weak institutions? Anecdotal evidence at the micro-level would suggest that equity agreements matter more where weak institutions prevail, although the global rise of GVCs and the associated tendency towards quasi-internalization across industries and regions does not suggest that there is a great variation in the use of non-equity cooperative structures (Asmussen, Chi, & Narula, 2022). However, this remains a conjecture, in need of empirical confirmation. Another research avenue is to investigate the effect of home-country IPR regimes on interfirm cooperation, rather than the effects of host-country regimes. Earlier studies noted that home-country IPR regimes imprint on MNEs (Meyer & Rowan, 1977; Meyer & Zucker, 1989; Rosenzweig & Singh, 1991; Zucker, 1987). Although at least one study noted that this might influence how MNEs behave in international alliances (Oxley, 1999), this has remained an unexplored theme. Indeed, evidence on the location of R&D activities by MNEs suggests that more knowledge-intensive R&D continues to be biased towards the home country (and region), and, by extension, towards the corporate HQ’s strategic objectives (Castellani, 2018). Alliances remain a key means to overcome the threats to the MNE’s intellectual property when engaging with partners from weak IPR regimes. International cooperation may also involve state-owned (or -controlled) enterprises, which in some cases are a deliberate and direct tool of industrial policy. In such cases, the interaction between IPR policies and firm-level cooperation takes on a more immediate and deliberate character, and brings to the forefront a variety of political economy-related issues (Andreoni & Chang, 2019; Genin, Tan & Song, 2022; Malerba & Lee, 2021; Mazzucato, 2018). IPR regimes and location choices There is still much to be discovered about how IPR regimes (home and host countries) and firm-level IPR protection and enforcement (of both MNEs and local firms) may interactively affect MNE location-related strategies. It is apparent that the IPR regulation matters more as a location advantage for certain kinds of industries than others. Despite some clear indications that more mature, low-tech industries do not consider IPR regimes as an important issue when making location choices, current research does not make a strong differentiation between legislation and effective enforcement of IPRs. Indeed, neither is dichotomous in nature, because there is considerable nuance in the kinds of legislation enacted, and in the extent to which IPRs are enforced, and this can vary by sector, and even between firms. This means that simply classifying countries as ‘strong’ or ‘weak’ in IPR protection can be an oversimplification, and future research needs to explore IPR enforcement and legislation separately, both as continua. This matters when considering the use of GVCs and the fine-slicing of activities and their spatial distribution. Although there is considerable discussion on institutional distance in the recent IB literature (e.g., Kostova, Beugelsdijk, Scott, Kunst, Chua, & van Essen, 2020), we know less about IPR-associated institutional distance and its influence on MNE location choices. In other words, distance may matter differently for IPR-related institutions, as the paper by Bruno et al. (2022) illustrates. How does IPR-related distance and IPR enforcement in practice influence transaction costs, knowledge flows, and, consequently, MNE location choices? How are location strategies affected by IPRs in ‘new’ technological domains where IPRs are still unclear (Chen, Li, Wei, & Yang, 2022). And how can institutional changes in different countries influence location-specific advantages, and consequently MNE geographic diversification of their subsidiaries? IPR regimes and inventor mobility The third area for further development is the effect of IPR regimes on the antecedents and consequences of inventor mobility in an international context. A multilevel perspective that contextualizes talent management within heterogeneous IPR regimes could challenge existing assumptions and open new research avenues for the literatures on talent recruitment and ‘learning-by-hiring’. One promising research avenue is to investigate the effect of home- or host-country IPR regimes on talent recruitment. For example, in host countries with weak IPR enforcement, is litigation still an effective measure to prevent inventors from moving to competitors (Agarwal et al., 2009; Ganco et al., 2015)? How do home-country IPR regimes influence the reputation of MNEs in host countries (Tung, 2007; Zaheer, 1995; Zaheer & Mosakowski, 1997), which consequently influences talent recruitment? Another promising research direction is to contextualize the outcome of ‘learning-by-hiring’. The literature on ‘learning-by-hiring’ has focused mainly on strong IPR regimes, overlooking the influence of various IPR environments in which former and current employers of the recruits are embedded. For example, one may study whether institutional distance between host- and home-country IPR regimes may influence how recruits adjust, cooperate, and contribute to MNEs’ innovation, which might be different from what they do in a homogeneous IPR environment. The Bottom–Up Process The bottom–up route to studying IPR protection has been largely unexplored in the IB-related IPR literature, although there has been an active engagement with the influences of MNEs on institutional evolution (e.g., Brandl, Darendeli, & Mudambi, 2019; Cantwell, Dunning, & Lundan, 2010; Hillman & Hitt, 1999). This bottom–up perspective can be applied to study how MNE IPR strategies can influence the implementation, enforcement, and evolution of IPR institutions in home countries. For example, as MNEs from weak IPR regimes adapt to stronger IPR environments, they gain first-hand experience, which may consequently influence IPR regime changes in their home countries. Alternatively, it is interesting to see how home-country IPR regimes of MNEs influence the IPR systems of host countries, and the degree to which certain MNEs are able to utilize regulatory capture to optimize their IPR portfolio. Similarly, it is worth examining how MNEs can serve as advocates for certain types of IPR protection, in their interactions with local governments, firms, and inventors, and how such activities influence IPR institutions in host countries, in a negative or positive way. CONCLUSION The objective of this special issue is to encourage a multilevel view in studying IPRs in IB. Intellectual property plays a critical role in conferring MNEs with FSAs, and this has received ample attention in the IB literature. So too have the location advantages associated with the protection of IPRs, advantages that determine country-level attractiveness as destinations for MNE investment. Integrating these single-level research findings, this paper has proposed a multilevel IPR research agenda that takes an interdisciplinary approach. The papers in this special issue have revealed some previously unexplored facets of IPR protection in IB. All of them have taken a top–down approach, focusing on the effect of host-country IPR regimes or institutional distance. As illustrated in Figure 1, abundant opportunities are left unexplored, which are embedded in the interactions between firm-level strategies to optimize FSAs and country-level challenges over location advantages, and how such interactions influence both firm-level decisions and performance, and country-level regimes and policies. Opportunities for multilevel IPR research also exist beyond the firm and country levels. An area that has been greatly neglected is the individual level, and, unfortunately, this special issue has been unable to rectify this shortcoming. Neither has this special issue seen IPR studies across other levels (e.g., supernational, subnational, and industry), or indeed from political economy, law, and sociology. Finally, we note that there is considerable merit in pursuing bottom–up research in IP-related research, as the recent work on microfoundations has emphasized (Felin, Foss, & Ployhart, 2015). Note https://www.wto.org/english/tratop_e/trips_e/intel1_e.htm. ACKNOWLEDGEMENTS We would like to thank Davide Castellani for providing a friendly review. Accepted by Alain Verbeke, Area Editor, 25 August 2022. This article was single-blind reviewed and has been with the authors for one revision. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Acemoglu, D., Aghion, P., & Zilibotti, F. 2006. Distance to frontier, selection, and economic growth. Journal of the European Economic Association, 4(1): 37–74. Acemoglu D Akcigit U Intellectual property rights policy, competition and innovation Journal of the European Economic Association 2012 10 1 1 42 10.1111/j.1542-4774.2011.01053.x Agarwal R Echambadi R Franco AM Sarkar MB Knowledge transfer through inheritance: Spin-out generation, development, and survival Academy of Management Journal 2004 47 4 501 522 10.2307/20159599 Agarwal, R., Ganco, M., & Ziedonis, R. H. 2009. Reputations for toughness in patent enforcement: Implications for knowledge spillovers via inventor mobility. Strategic Management Journal, 30(13): 1349–1374. Alcácer J Chung W Location strategies and knowledge spillovers Management Science 2007 53 5 760 776 10.1287/mnsc.1060.0637 Alcácer J Zhao M Local R&D strategies and multilocation firms: The role of internal linkages Management Science 2012 58 4 734 753 10.1287/mnsc.1110.1451 Almeida P Kogut B Localization of knowledge and the mobility of engineers in regional networks Management Science 1999 45 7 905 917 10.1287/mnsc.45.7.905 Anand J McDermott G Mudambi R Narula R Innovation in and from emerging economies: New insights and lessons for international business research Journal of International Business Studies 2021 52 4 545 559 10.1057/s41267-021-00426-1 Andersson UR Brewster CJ Minbaeva DB Narula R Wood GT The IB/IHRM interface: Exploring the potential of intersectional theorizing Journal of World Business 2019 54 5 100998 10.1016/j.jwb.2019.100998 Andreoni A Chang HJ The political economy of industrial policy: Structural interdependencies, policy alignment and conflict management Structural Change and Economic Dynamics 2019 48 136 150 10.1016/j.strueco.2018.10.007 Arundel A The relative effectiveness of patents and secrecy for appropriation Research Policy 2001 30 4 611 624 10.1016/S0048-7333(00)00100-1 Arundel A Kabla I What percentage of innovations are patented? Empirical estimates for European firms Research Policy 1998 27 2 127 141 10.1016/S0048-7333(98)00033-X Arslan, B. 2018. The interplay of competitive and cooperative behavior and differential benefits in alliances. Strategic Management Journal, 39(12): 3222–3246. Asmussen, C., Chi, T., & Narula, R. 2022. Quasi-internalization, recombination advantages, and global value chains: Clarifying the role of ownership and control. Journal of International Business Studies, 53: 1747–1765. Athreye S Piscitello L Shadlen KC Twenty-five years since TRIPS: Patent policy and international business Journal of International Business Policy 2020 3 4 315 328 10.1057/s42214-020-00079-1 Azmeh, S., Foster, C., & Echavarri, J. 2020. The international trade regime and the quest for free digital trade. International Studies Review, 22(3): 671–692. Belderbos R Park J Carree M Do R&D investments in weak IPR countries destroy market value? The role of internal linkages Strategic Management Journal 2021 42 8 1401 1431 10.1002/smj.3268 Bogers M Zobel AK Afuah A Almirall E Brunswicker S Dahlander L Frederiksen L Gawer A Gruber M Haefliger S Hagedoorn J The open innovation research landscape: Established perspectives and emerging themes across different levels of analysis Industry and Innovation 2017 24 1 8 40 10.1080/13662716.2016.1240068 Brander JA Cui V Vertinsky I China and intellectual property rights: A challenge to the rule of law Journal of International Business Studies 2017 48 7 908 921 10.1057/s41267-017-0087-7 Baldwin, C. Y., & Henkel, J. 2015. Modularity and intellectual property protection. Strategic Management Journal, 36(11): 1637–1655. Brandl, K., Darendeli, I., & Mudambi, R. 2019. Foreign actors and intellectual property protection regulations in developing countries. Journal of International Business Studies, 50(5): 826–846. Bruno, R. L., Crescenzi, R., Estrin, S., & Petralia, S. 2022. Multinationals, innovation, and institutional context: IPR protection and distance effects. Journal of International Business Studies, 53(9). 10.1057/s41267-021-00452-z. Cantwell J Santangelo GD The frontier of international technology networks: Sourcing abroad the most highly tacit capabilities Information Economics and Policy 1999 11 1 101 123 10.1016/S0167-6245(99)00005-0 Cantwell John Dunning John H Lundan Sarianna M An evolutionary approach to understanding international business activity: The co-evolution of MNEs and the institutional environment Journal of International Business Studies 2010 41 4 567 586 10.1057/jibs.2009.95 Castellani D Cook G Johns J McDonald F Beaverstock J Pandit N The changing geography of innovation and the multinational enterprise The Routledge companion to the geography of international business 2018 London Routledge Castellani D Lavoratori K The lab and the plant: Offshore R&D and co-location with production activities Journal of International Business Studies 2020 51 1 121 137 10.1057/s41267-019-00255-3 Chang HJ Kicking away the ladder: Development strategy in historical perspective 2002 Anthem Press Chen Y Puttitanun T Intellectual property rights and innovation in developing countries Journal of Development Economics 2005 78 2 474 493 10.1016/j.jdeveco.2004.11.005 Chen L Li S Wei J Yang Y Externalization in the platform economy: Social platforms and institutions Journal of International Business Studies 2022 53 1 12 10.1057/s41267-022-00506-w Chung W Alcácer J Knowledge seeking and location choice of foreign direct investment in the United States Management Science 2002 48 12 1534 1554 10.1287/mnsc.48.12.1534.440 Cohen WM Goto A Nagata A Nelson RR Walsh JP R&D spillovers, patents and the incentives to innovate in Japan and the United States Research Policy 2002 31 8–9 1349 1367 10.1016/S0048-7333(02)00068-9 Contractor FJ Reuer JJ Structuring and governing alliances: New directions for research Global Strategy Journal 2014 4 4 241 256 10.1002/gsj.1082 Criscuolo P Narula R A novel approach to national technological accumulation and absorptive capacity: Aggregating Cohen and Levinthal The European Journal of Development Research 2008 20 1 56 73 10.1080/09578810701853181 Cui V Yang H Vertinsky I Attacking your partners: Strategic alliances and competition between partners in product markets Strategic Management Journal 2018 39 12 3116 3139 10.1002/smj.2746 Dasgupta P Stiglitz J Industrial structure and the nature of innovative activity The Economic Journal 1980 90 358 266 293 10.2307/2231788 Dyer JH Chu W The role of trustworthiness in reducing transaction costs and improving performance: Empirical evidence from the United States, Japan, and Korea Organization Science 2003 14 1 57 68 10.1287/orsc.14.1.57.12806 Felin T Foss NJ Ployhart RE The microfoundations movement in strategy and organization theory Academy of Management Annals 2015 9 1 575 632 10.5465/19416520.2015.1007651 Forsgren M Holm U Controlling without owning–owning without controlling: A critical note on two extensions of internalization theory Journal of International Business Studies 2021 52 1 13 Ganco, M., Ziedonis, R. H., & Agarwal, R. 2015. More stars stay, but the brightest ones still leave: Job hopping in the shadow of patent enforcement. Strategic Management Journal, 36(5): 659–685. Genin, A. L., Tan, J., & Song, J. 2022. Relational assets or liabilities? Competition, collaboration, and firm intellectual property breakthrough in the Chinese high-speed train sector. Journal of International Business Studies, 53(9). 10.1057/s41267-021-00482-7. Ginarte, J. C., & Park, W. G. 1997. Determinants of patent rights: A cross-national study. Research Policy, 26(3): 283–301. Giuliani E Jacqueminet A Nieri F Castellani D Perri A Scalera VG Zanfei A Bringing light to dark spots: The case of cross-border bioprospecting Cross-border innovation in a changing world: Players, places, and policies 2022 Oxford Oxford University Press Gooderham PN Elter F Pedersen T Sandvik AM The digital challenge for multinational mobile network operators. More marginalization or rejuvenation? Journal of International Management 2022 28 4 100946 10.1016/j.intman.2022.100946 Gooris J Peeters C Fragmenting global business processes: A protection for proprietary information Journal of International Business Studies 2016 47 5 535 562 10.1057/jibs.2015.44 Griffith MK Steinberg RH Zysman J From great power politics to a strategic vacuum: Origins and consequences of the TPP and TTIP Business and Politics 2017 19 4 573 592 10.1017/bap.2017.16 Guimón J Narula R Ending the COVID-19 pandemic requires more international collaboration Research-Technology Management 2020 63 5 38 41 10.1080/08956308.2020.1790239 Gulati R Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances Academy of Management Journal 1995 38 1 85 112 10.2307/256729 Gulati R Social structure and alliance formation patterns: A longitudinal analysis Administrative Science Quarterly 1995 40 619 652 10.2307/2393756 Hagedoorn J Inter-firm R&D partnerships: An overview of major trends and patterns since 1960 Research Policy 2002 31 4 477 492 10.1016/S0048-7333(01)00120-2 Hagedoorn J Narula R Choosing organizational modes of strategic technology partnering: International and sectoral differences Journal of International Business Studies 1996 27 2 265 284 10.1057/palgrave.jibs.8490135 Hamel G Competition for competence and interpartner learning within international strategic alliances Strategic Management Journal 1991 12 S1 83 103 10.1002/smj.4250120908 Hamel, G., Doz, Y. L., & Prahalad, C. K. 1989. Collaborate with your competitors and win. Harvard Business Review, 67(1): 133–139. Heiman B Nickerson JA Empirical evidence regarding the tension between knowledge sharing and knowledge expropriation in collaborations Management Decision Economics 2004 25 6–7 401 420 10.1002/mde.1198 Hillman Amy J. Hitt Michael A. Corporate Political Strategy Formulation: A Model of Approach, Participation, and Strategy Decisions Academy of Management Review 1999 24 4 825 842 10.5465/amr.1999.2553256 Hoetker G How much you know versus how well I know you: Selecting a supplier for a technically innovative component Strategic Management Journal 2005 26 1 75 96 10.1002/smj.453 Hoffmann W Lavie D Reuer JJ Shipilov A The interplay of competition and cooperation Strategic Management Journal 2018 39 12 3033 3052 10.1002/smj.2965 Inkpen A Minbaeva D Tsang EW Unintentional, unavoidable, and beneficial knowledge leakage from the multinational enterprise Journal of International Business Studies 2019 50 2 250 260 10.1057/s41267-018-0164-6 Jaffe AB Trajtenberg M Henderson R Geographic localization of knowledge spillovers as evidenced by patent citations The Quarterly Journal of Economics 1993 108 3 577 598 10.2307/2118401 Javorcik BS The composition of foreign direct investment and protection of intellectual property rights: Evidence from transition economies European Economic Review 2004 48 1 39 62 10.1016/S0014-2921(02)00257-X Jecker NS Achieving global vaccine equity: The case for an international pandemic treaty The Yale Journal of Biology and Medicine 2022 95 2 271 35782474 Jecker NS Atuire CA What’s yours is ours: Waiving intellectual property protections for COVID-19 vaccines Journal of Medical Ethics 2021 47 9 595 598 10.1136/medethics-2021-107555 34233956 Kalaycı, E., & Pamukçu, T. 2014. Assessing the drivers of R&D activities of firms in developing countries: evidence from Turkey. The European Journal of Development Research, 26(5): 853–869. Khanna T Gulati R Nohria N The dynamics of learning alliances: Competition, cooperation, and relative scope Strategic Management Journal 1998 19 3 193 210 10.1002/(SICI)1097-0266(199803)19:3<193::AID-SMJ949>3.0.CO;2-C Kim YK Lee K Park WG Choo K Appropriate intellectual property protection and economic growth in countries at different levels of development Research Policy 2012 41 2 358 375 10.1016/j.respol.2011.09.003 Kostova T Beugelsdijk S Scott WR Kunst VE Chua CH van Essen M The construct of institutional distance through the lens of different institutional perspectives: Review, analysis, and recommendations Journal of International Business Studies 2020 51 4 467 497 10.1057/s41267-019-00294-w Kranenburg H Hagedoorn J Lorenz-Orlean S Distance costs and the degree of inter-partner involvement in international relational-based technology alliances Global Strategy Journal 2014 4 4 280 291 10.1002/gsj.1085 Lall S Dunning JH Hamdani KA Investment, technology and international competitiveness The new globalism and developing countries 1997 New York United Nations University Press 232 259 Lavie D Alliance portfolios and firm performance: A study of value creation and appropriation in the US software industry Strategic Management Journal 2007 28 12 1187 1212 10.1002/smj.637 Li D Eden L Hitt M Ireland RD Friends, acquaintances, or strangers? Partner selection in R&D alliances Academy of Management 2008 51 2 315 334 10.5465/amj.2008.31767271 Malerba F Lee K An evolutionary perspective on economic catch-up by latecomers Industrial and Corporate Change 2021 30 4 986 1010 10.1093/icc/dtab008 Mariotti S Piscitello L Elia S Spatial agglomeration of multinational enterprises: The role of information externalities and knowledge spillovers Journal of Economic Geography 2010 10 4 519 538 10.1093/jeg/lbq011 Martinez-Noya A Garcia-Canal E Location, shared R&D suppliers and the performance of R&D outsourcing agreements Industry and Innovation. 2018 25 3 308 332 10.1080/13662716.2017.1329085 Martinez-Noya A Narula R What more can we learn from R&D alliances? A review and research agenda BRQ Business Research Quarterly 2018 21 3 195 221 10.1016/j.brq.2018.04.001 Maskus, K. E. 2014. The new globalisation of intellectual property rights: What's new this time? Australian Economic History Review, 54(3): 262–284. Meyer, J. W., & Rowan, B. 1977. Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83(2): 340–363. Meyer, M. W., & Zucker, L. G. 1989. Permanently failing organizations. SAGE Publications. Mawdsley JK Somaya D Employee mobility and organizational outcomes: An integrative conceptual framework and research agenda Journal of Management 2016 42 1 85 113 10.1177/0149206315616459 May C Sell SK Intellectual property rights: A critical history 2006 Boulder Lynne Rienner Mazzucato M Mission-oriented innovation policies: Challenges and opportunities Industrial and Corporate Change 2018 27 5 803 815 10.1093/icc/dty034 Mertha, A. C. 2007. Enforcement of Intellectual Property Rights. Oxford: University of Oxford. The Foundation for Law, Justice and Society in collaboration with The Centre for Socio-Legal Studies, Rule of Law in China Publications 1. Nandkumar A Srikanth K Right person in the right place: How the host country IPR influences the distribution of inventors in offshore R&D projects of multinational enterprises Strategic Management Journal 2016 37 8 1715 1733 10.1002/smj.2418 Narula R Choosing between internal and non-internal R&D activities: Some technological and economic factors Technology Analysis & Strategic Management 2001 13 3 365 387 10.1080/09537320120088183 Narula R Castellani D Perri A Scalera VG Zanfei A IPRs, cross-border (collaborative) innovation and development challenges: A comment Cross-border innovation in a changing world: Players, places, and policies 2022 Oxford Oxford University Press Narula R Dunning JH Multinational enterprises, development and globalization: Some clarifications and a research agenda Oxford Development Studies 2010 38 3 263 287 10.1080/13600818.2010.505684 Narula R Hagedoorn J Innovating through strategic alliances: Moving towards international partnerships and contractual agreements Technovation 1999 19 5 283 294 10.1016/S0166-4972(98)00127-8 Narula R Martinez-Noya A Archibugi D Filippetti A International R&D alliances by firms: Origins and development The handbook of global science, technology, and innovation 2015 Hoboken Wiley Narula R Pineli A Improving the developmental impact of multinational enterprises: Policy and research challenges Journal of Industrial and Business Economics 2019 46 1 1 24 10.1007/s40812-018-0104-2 Narula R Santangelo GD Location, collocation and R&D alliances in the European ICT industry Research Policy 2009 38 2 393 403 10.1016/j.respol.2008.11.005 Narula R Santangelo GD Location and collocation advantages in international innovation Multinational Business Review 2012 20 1 6 25 10.1108/15253831211217161 Narula R Van der Straaten K A comment on the multifaceted relationship between multinational enterprises and within-country inequality Critical Perspectives on International Business 2021 17 1 33 52 10.1108/cpoib-10-2019-0080 Narula R Asmussen CG Chi T Kundu SK Applying and advancing internalization theory: The multinational enterprise in the twenty-first century Journal of International Business Studies 2019 50 9 1231 1252 10.1057/s41267-019-00260-6 Noon P De Vita G Appleyard L What do we know about the impact of intellectual property rights on the foreign direct investment location (country) choice? A review and research agenda Journal of Economic Surveys 2019 33 2 665 688 10.1111/joes.12292 Nunnenkamp P Spatz J Intellectual property rights and foreign direct investment: A disaggregated analysis Review of World Economics 2004 140 3 393 414 10.1007/BF02665982 Oxley JE Institutional environment and the mechanisms of governance: The impact of intellectual property protection on the structure of inter-firm alliances Journal of Economic Behavior & Organization 1999 38 3 283 309 10.1016/S0167-2681(99)00011-6 Papageorgiadis N McDonald F Castellani D Perri A Scalera VG Zanfei A Cross-border innovation and knowledge protection policies Cross-border innovation in a changing world: Players, places, and policies 2022 Oxford Oxford University Press Papanastassiou M Pearce R Zanfei A Changing perspectives on the internationalization of R&D and innovation by multinational enterprises: A review of the literature Journal of International Business Studies 2020 51 4 623 664 10.1057/s41267-019-00258-0 Peng, M. W., Ahlstrom, D., Carraher, S. M., & Shi, W. S. 2017. An institution-based view of global IPR history. Journal of International Business Studies, 48(7): 893–907. Polidoro F Jr Ahuja G Mitchell W When the social structure overshadows competitive incentives: The effects of network embeddedness on joint venture dissolution Academy of Management Journal 2011 54 1 203 223 10.5465/amj.2011.59215088 Reuer J Lahiri N Searching for alliance partners: Effects of geographic distance on the formation of R&D collaborations Organization Science 2014 25 1 283 298 10.1287/orsc.1120.0805 Rubenstein AL Eberly MB Lee TW Mitchell TR Surveying the forest: A meta-analysis, moderator investigation, and future-oriented discussion of the antecedents of voluntary employee turnover Personnel Psychology 2018 71 1 23 65 10.1111/peps.12226 Rosenzweig, P. M., & Singh, J. V. 1991. Organizational environments and the multinational enterprise. Academy of Management Review, 16(2): 340–361. Ryu W McCann BT Reuer JJ Geographic co-location of partners and rivals: Implications for the design of R&D alliances Academy of Management Journal 2018 61 3 945 965 10.5465/amj.2016.0416 Saggi, K. 2002. Trade, foreign direct investment, and international technology transfer: A survey. The World Bank Research Observer, 17(2): 191–235. Santamaria L Nieto MJ Barge-Gil A The relevance of different open innovation strategies for R&D performers Cuadernos De Economía y Dirección De La Empresa 2010 13 45 93 114 10.1016/S1138-5758(10)70025-6 Schaefer K Catching up by hiring: The case of Huawei Journal of International Business Studies 2020 51 9 1500 1515 10.1057/s41267-019-00299-5 Shaver J Flyer F Agglomeration economies, firm heterogeneity, and foreign direct investment in the United States Strategic Management Journal 2000 21 12 1175 1193 10.1002/1097-0266(200012)21:12<1175::AID-SMJ139>3.0.CO;2-Q Singh J Asymmetry of knowledge spillovers between MNCs and host country firms Journal of International Business Studies 2007 38 5 764 786 10.1057/palgrave.jibs.8400289 Shi, W. S., Sun, S. L., Pinkham, B. C., & Peng, M. W. 2014. Domestic alliance network to attract foreign partners: Evidence from international joint ventures in China. Journal of International Business Studies, 45(3): 338–362. Slavova K Fosfuri A De Castro O Learning by hiring: The effects of scientists’ inbound mobility on research performance in academia Organization Science 2016 27 1 72 89 10.1287/orsc.2015.1026 Song J Almeida P Wu G Learning–by–hiring: When is mobility more likely to facilitate interfirm knowledge transfer? Management Science 2003 49 4 351 365 10.1287/mnsc.49.4.351.14429 Sweet CM Maggio DSE Do stronger intellectual property rights increase innovation? World Development 2015 66 665 677 10.1016/j.worlddev.2014.08.025 Sweet, C., & Eterovic, D. 2019. Do patent rights matter? 40 years of innovation, complexity and productivity. World Development, 115: 78–93. Tung, R. L. 2007. The human resource challenge to outward foreign direct investment aspirations from emerging economies: The case of China. The International Journal of Human Resource Management, 18(5): 868–889. Van De Vrande V Vanhaverbeke W Gassmann O Broadening the scope of open innovation: Past research, current state and future directions International Journal of Technology Management 2010 52 3/4 221 235 10.1504/IJTM.2010.035974 Wang Y Wang N Jiang L Yang Z Cui V Managing relationships with power advantage buyers: The role of supplier initiated bonding tactics in long-term buyer–supplier collaborations Journal of Business Research 2016 69 12 5587 5596 10.1016/j.jbusres.2016.03.066 Wang, H. C., He, J., & Mahoney, J. T. 2009. Firm‐specific knowledge resources and competitive advantage: the roles of economic‐and relationship‐based employee governance mechanisms. Strategic Management Journal, 30(12): 1265–1285. Wu J Ma Z Zhuo S Enhancing national innovative capacity: The impact of high-tech international trade and inward foreign direct investment International Business Review 2017 26 3 502 514 10.1016/j.ibusrev.2016.11.001 Yang H Zheng Y Zaheer A Asymmetric learning capabilities and stock market returns Academy of Management Journal 2015 58 2 356 374 10.5465/amj.2012.0675 Yan, Y., Li, J., & Zhang, J. 2022. Protecting intellectual property in foreign subsidiaries: An internal network defense perspective. Journal of International Business Studies, 53(9). 10.1057/s41267-021-00430-5. Younge, K. A., Tong, T. W., & Fleming, L. 2015. How anticipated employee mobility affects acquisition likelihood: Evidence from a natural experiment. Strategic Management Journal, 36: 686–708. Zaheer, S. 1995. Overcoming the liability of foreignness. Academy of Management Journal, 38(2): 341–363. Zaheer, S., & Mosakowski, E. 1997. The dynamics of the liability of foreignness: A global study of survival in financial services. Strategic Management Journal, 18(6): 439–463. Zaheer A McEvily B Perrone V Does trust matter? Exploring the effects of interorganizational and interpersonal trust on performance Organization Science 1998 9 2 123 141 10.1287/orsc.9.2.141 Zhao M Conducting R&D in countries with weak intellectual property rights protection Management Science 2006 52 8 1185 1199 10.1287/mnsc.1060.0516 Zucker, L. G. 1987. Institutional theories of organization. Annual Review of Sociology, 13(1): 443–464.	36474709	PMC9716505	NO-CC CODE	2022-12-03 23:20:57	no	J Int Bus Stud. 2022 Dec 2; 53(9):1880-1894	utf-8	J Int Bus Stud	2,022	10.1057/s41267-022-00564-0	oa_other
==== Front Soft comput Soft comput Soft Computing 1432-7643 1433-7479 Springer Berlin Heidelberg Berlin/Heidelberg 7700 10.1007/s00500-022-07700-w Application of Soft Computing Automatic approach for mask detection: effective for COVID-19 Banik Debajyoty [email protected] 1 Rawat Saksham [email protected] 1 Thakur Aayush [email protected] 1 http://orcid.org/0000-0002-2439-1507 Parwekar Pritee [email protected] 2 Satapathy Suresh Chandra [email protected] 1 1 grid.412122.6 0000 0004 1808 2016 School of Computer Engineering, Kalinga Institute of Industrial Technology, Deemed to be University, Odisha, India 2 grid.412742.6 0000 0004 0635 5080 SRMIST: SRM Institute of Science and Technology, Delhi-NCR Campus, Ghaziabad, India 2 12 2022 111 18 11 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The outbreak of coronavirus disease 2019 (COVID-19) occurred at the end of 2019, and it has continued to be a source of misery for millions of people and companies well into 2020. There is a surge of concern among all persons, especially those who wish to resume in-person activities, as the globe recovers from the epidemic and intends to return to a level of normalcy. Wearing a face mask greatly decreases the likelihood of viral transmission and gives a sense of security, according to studies. However, manually tracking the execution of this regulation is not possible. The key to this is technology. We present a deep learning-based system that can detect instances of improper use of face masks. A dual-stage convolutional neural network architecture is used in our system to recognize masked and unmasked faces. This will aid in the tracking of safety breaches, the promotion of face mask use, and the maintenance of a safe working environment. In this paper, we propose a variant of a multi-face detection model which has the potential to target and identify a group of people whether they are wearing masks or not. Keywords Boundary-layer meteorology CNN (Convolutional neural network) COVID-19 Grad CAM MobileNetV2 ==== Body pmcIntroduction COVID-19 is a virulent disease that has unfolded across the world. The pandemic ailment has ended in an important worldwide fitness difficulty that has profoundly impacted humanity and the manner we see the truth and our everyday lives. The unfolding of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), some other very contagious respiration ailment, commenced in Wuhan in December 2019. Before COVID-19 was declared a global pandemic, 7711 human beings were infected and 170 deaths were disclosed in China. Coronavirus has been given the designation COVID-19 in line with the World Health Organization (COVID contamination 2019) (W. H. Organization et al. 2020). COVID-19 has inflamed greater than 13,039,853 humans and brought about greater than 571,659 fatalities in more than 200 countries across the world, in line with a World Health Organization (WHO) report (beginning July 12, 2020), ensuing in a fatality price of round 37, as compared to a demise price of below 1 percent from flu. Individual to man or woman transmission of the radical COVID generating COVID contamination 2019 (COVID-19) has been reported, but it seems that transmission of the radical COVID inflicting COVID contamination 2019 (COVID-19) also can be from an asymptomatic transporter without coronavirus symptoms. There is, but if any, clinically accepted antiviral drug or antibodies that have been proven to be powerful in opposition to COVID-19. It has hastily elevated over the planet, inflicting big well-being, economic, environmental, and social issues for the complete human population. Individuals have to put on facial veils to keep away from the hazard of contamination transmission, and a social hole of a minimum of 2 m (Rota et al. 2003) has to be maintained among human beings to save their character from character unfold of sickness, consistent with WHO. Furthermore, numerous public provider establishments require customers to apply their offerings simplest if they put on veils and cling to secure social segregation. As a result, face veil identity and secure social separation checking have ended up an essential PC vision (Memish et al. 2013) project on the way to help the worldwide society. This look illustrates a technique for stopping the transmission of infection via way of means of constantly looking if people are adhering to secure social practices which include casting off their face coverings and carrying them openly. Deep learning techniques (Liu et al. 2018) have recently demonstrated a significant significance in object detection. Facial detection research necessitates expression recognition, face tracking, and position estimation, according to (Khan et al. 2019; Licheng et al. 2019). The objective is to recognize the face in a single photograph. Face detection is a tough task since faces fluctuate in size, shape, color, and other characteristics throughout time. The World Health Organization (WHO) reports proposed that the two primary courses of transmission of the COVID-19 infection are respiratory beads and actual contact. In this investigation, clinical covers were characterized as careful or technique covers that are level or creased (some are molded like cups); they are appended to the head with lashes. They are tried for adjusted high filtration, satisfactory breath ability, and alternatively, liquid infiltration obstruction. The examination investigated a bunch of video transfers/pictures to distinguish individuals who are consistent with the public authority rule of wearing clinical covers. This could assist the public authority with making a suitable move against individuals who are resistant. In the current situation, everybody has been feeling down and discouraged about the condition of the world; a huge number of individuals are biting the dust every day, and for a considerable lot of us, there is next to no (regardless) we can do. To help in any little manner conceivable, we chose to apply PC vision and profound figuring out how to tackle a genuine issue: Best-case situation—we can utilize our undertaking to help other people. As software engineers, designers, and PC vision/profound learning specialists, we let our abilities become the interruption and our sanctuary. To create this dataset, we had the brilliant idea ofCapturing faces in their natural state. Then writing a custom computer vision Python software to detect face masks on them, resulting in a fictitious (but still useful) dataset. Once you practice face landmarks to the problem, this approach is absolutely plenty less difficult than it sounds. We can use facial landmarks to routinely deduce the location of face systems such as:Eyes Eyebrows Nose Mouth Jawline To prepare a custom face cover locator, breaking our venture into two unmistakable stages was required, each with its own separate sub-steps.Preparing: Here, stacking our face veil discovery dataset from plate, preparing a model on this dataset, and afterward serializing the face cover locator to circle was the focus. Individual entries are indicated with a black dot, a so-called bullet. The text in the entries may be of any length. Sending: Once the face veil identifier is prepared, the accompanying advance of stacking the cover finder, performing face recognition, and afterward characterizing each face as with veil or without veil can be executed. To prevent the spread of infection during COVID-19 outbreak, almost everyone wears a veil. As a result, traditional facial recognition innovation, such as network access control, face access control, facial participation, facial security checks at railway stations, and so on, is virtually always inadequate. Consequently, It is vital to improve the current face recognition technology’s recognition performance on hidden faces. The majority of today’s advanced face recognition systems rely on a huge number of face samples and are based on deep comprehension. In a real-time scenario, their algorithm (Teboulbi et al. 2021) tracks persons wearing or not wearing masks and provides social separation by producing an alarm if there is a violation in the scene or in public locations. This can be combined with current embedded camera infrastructure to allow these analytics, which can be used in a variety of verticals as well as in offices and airport terminals/gates. In this task, we will go over our two-stage COVID-19 face cover identifier as well as our PC vision/profound learning pipeline. After that, we will run a check on the dataset we will be utilizing to develop our own face cover indicator. We will then show how to utilize Keras and TensorFlow to execute a Python script on our dataset to produce a face cover identifier. This Python code will be used to generate a face cover identification and a survey of the findings. We will continue to run two more Python programs to detect face covers while video transfers take place now that the COVID-19 face cover detector is ready. We will wrap off this piece by taking a peek at the results of our face veil finder. The remainder of the paper goes like this: Section II describes the methodology, followed by Section III that describes a CNN certainty by iteratively blocking parts. MobileNet V2 description is provided in Section IV. Section V describes Grad-Cam. Overview of model description and formulation is shown in Section VI followed by evaluation metrics in Section VII. Limitations of the face detection system are provided in Section VIII. Section IX describes how to overcome limits on facial recognition tools. Section X describes the conclusion followed by Section XI future work. Methodology/Approach To utilize facial imprints to develop an informational collection of facial covers, we started with a picture of an individual who does not wear a facial veil. Following this, face recognition was applied to figure the area of the jumping enclosing the image. When we knew where in the picture the face is, we could extricate the face region of interest (ROI), and from that point, we applied facial milestones, permitting us to restrict the mouth, face, and eyes. To apply covers, we required a picture of a veil (with a straightforward and top-notch picture) and afterward added the cover to the identified face and afterward resized and turned in like manner to put it over the face. This cycle is rehashed for all information of images.Training: This progression included preparing for the picture of appearances with veil and without cover individually with a fitting algorithm.Deployment: Once the models were prepared, we proceeded onward to the stacking cover identifier and perform face identification, at that point for characterization of each face.At the point when an image had been moved, the request happened normally. It was then possible to apply some interpretation ability strategies for neural association understanding. The UI presented two of the going with methods: Grad CAM envisioned how parts of the info picture influence a CNN yield by investigating the enactment maps, and Occlusion Sensitivity imagined how parts of the information picture affect, which is shown in Fig. 1.Fig. 1 Classification algorithm from PyTorch Fig. 2 MobileNetV2 classifier training process using PyTorch Fig. 3 MobileNetV2 architecture A CNN certainty by iteratively blocking parts Picture classification algorithm from PyTorch Convolutional neural network (CNN) has several well-designed and pre-built networks, such as AlexNet, ResNet, Inception, LeNet, MobileNet, and so on. Because of its lightweight and effective adaptable organized model, I chose the MobileNetV2 for our circumstance. We utilize two related demonstrating techniques to manage and test the viability of face mask usage by segments of the general population in reducing SARS-CoV-2 transmission and, as a result, in lowering the appropriate age number, Re (the ordinary number of new cases achieved by a single overpowering individual at a given point in the scourge). The basic model employs a fanning cycle to examine the reduction in transmission caused by the use of face masks, as well as the achievable adequacy of two control variables in lowering Re for the microbe. The degree to which the general population employs face masks (basically, the likelihood that a person would wear a cloak on any given day) and the appropriateness of the cover in decreasing transmission are the control factors (which relate to an extent of covers that connect from unpleasant penetrable covers (Mohamed et al. 2010, Piccardi 2004) to fronts of clinical standard). The goal of this model is to see if there are any obvious limit ranges in which the two control variables may lower Re to the point where they can be relied on to halt or stop the spread of the epidemic. We mimic the aftereffects of those who wear face masks regularly, or shortly after they begin to experience adverse effects. We modify the fundamental SIR definition and include free-living SARS-CoV-2 particles delivered by internal breath from globules in the oculum and by continuous contact with facial apertures from fomite inoculum stored on surfaces (§2b(i)). The model is intended to examine the possible effects of wearing a face mask during times of lockdown, which are then dispersed once the lockdown is lifted. Because of the flexible nature of this displaying framework, a distinction may be seen between the capacity of face masks to decrease transmission from sullied persons (where sign verbalization is used) and the security provided by face masks on weak individuals. The final point might be beneficial, as the face mask decreases oculum internal breathing. It might also be negative; for example, if there is a constant manual variation in the face mask, the likelihood of transmission increases. We will most likely provide a varied, yet almost clear displaying framework to test hypotheses regarding face mask use in conjunction with other pandemic tactics, as well as scaling from one lead to people outcomes. SARS-CoV-2 is a new illness to humanity; thus, the conclusions should be made in this context, considering the gaps in our understanding regarding certain limits. Customary object detection: A traditional item reputation version can apprehend the trouble of recognizing many veiled and unmasked faces in photographs. The majority of the time, object vicinity involves finding and categorizing matters in photographs (if there ought to be a prevalence of various items). Although traditional algorithms depend considerably on feature engineering, Haar cascade (Rafael et al. 2012) and HOG (Dalal and Triggs 2005) have proven to be beneficial for such situations. In the age of deep learning, it is miles viable to layout neural networks that keep away from those computations and do now no longer require any similarly feature engineering. Multistage detectors: The detection cycle is split into numerous ranges in a multi-level identifier. A two-level indicator, which includes RCNN, measures and shows a listing of areas of hobby primarily based totally on precise pursuit. After that, the eCNN spotlight vectors are freely deleted from every locale. Several regional proposal networks primarily based totally on algorithms, which include fast RCNN (Girshick, 2015) and faster RCNN (Ren et al. 2015), have carried out extra accuracy and desired consequences than maximum single-level detectors. Single-stage detectors: A one-level indicator conducts recognition in an unmarried step, ostensibly over an intensive examination of ability regions. These computations leave out the place proposition level utilized in multi-level detectors, making them quicker in general, however, on the value of a few accuracy losses. One of the maximum famous unmarried-level calculations, You Only Look Once (YOLO), changed into released in 2015 and attained close to non-stop performance. The single-shot detector is a tool that detects an unmarried shot. SSD is some other famous item identity technique that produces exact results. RetinaNet and feature pyramid networks are one of the best indicators and use critical misfortune. Adding more stages of learned transformations, specifically a module for feedforward connections in deconvolution and a new output module, enables this new approach and provides a potential path forward for further detection (Figs. 2 and 3). MobileNetV2 It uses depthwise separable convolution as an efficient building component, based on ideas from MobileNetV1 (W. H. Organization et al. 2020). V2, on the other hand, adds two additional aspects to the architecture:Linear bottlenecks between the layers and Shortcut connections between the bottlenecks. The basic structure is shown below. The overall point of the objective’s face has a major influence on the acknowledgment score. When a face is associated with recognition programmers, several points are frequently employed (profile, frontal, and 45 degree are normal). Anything less than a frontal perspective has an impact on the calculation’s capacity to generate a face format. The greater the score of any future matches, the more plain the picture (both enlisted and test picture) and the higher it’s objective. Loads of each layer in the model are predefined depending on the ImageNet dataset. The loads show the cushioning, steps, part size, input channels, and yield channels (Table 1). Table 1 In view of ImageNet dataset MobileNetV2 beats MobileNetV1 and ShuffleNet (1.5) with equivalent model size and computational expense. And furthermore it will perform well for the more modest dataset Model Params Multiply-Adds mAP MobileCPU MobileNetV1 + SSDLite 5.1M 1.3B 22.2% 270ms MobileNetV2 + SSDLite 4.3M 0.8B 22.1% 200ms Step 1: data visualization In the first phase, we imagined the total number of images in our collection is divided into two classes. There are 690 photographs in the ‘yes’ class and 686 pictures in the ‘no’ class, as can be seen. Step 2: data augmentation After that, we expanded our dataset to include a larger number of images for our preparation. We rotated and flipped every single photograph in our dataset throughout this process of information development. Following information expansion, we now have a total of 2751 images, with 1380 images in the ‘yes’ category and 1371 images in the ‘no’ category as shown in Fig. 4.Fig. 4 Data distribution Step 3: Splitting the data We divided our data into two sets: the preparation set, which contains the images on which the CNN model will be trained, and the test set, which contains the images on which our model will be tested. In this case, we will choose split size =0.8, which means that 80 percent of the absolute photographs will go to the preparation set, while the remaining 20 percent will go to the test set. Following separating, we discovered that the optimal level of images had been distributed to both the preparation set and the test set, as previously mentioned. Step 4: Building the model Following that, we built our sequential CNN model using several layers such as Conv2D, Max Pooling2D, Flatten, Dropout, and Dense. In the final dense layer, we use the softmax capability to generate a vector that represents the probability of each of the two classes. Because there are just two classes, we used the ‘adam’ streamlining agent and ‘paired cross-entropy’ as our unfortunate job. Furthermore, the MobileNetV2 may be used to improve accuracy. Step 5: Pre-training the CNN model Following the construction of our model, we created the ‘train generator’ and ‘approval generator’ in order to adapt them to our model in the next step. We discovered that the preparation set has 2200 images and the test set contains 551 images. Step 6: Training the CNN model This is the first step, in which we fit our photographs from the preparation and test sets to the sequential model we built using the Keras library. I have made a model for 30 different ages (cycles). Nonetheless, we can plan for a larger number of ages to obtain more precision in the event of over-fitting. Our model has a precision of 96.19 percent with the preparation set and a precision of 98.86 percent with the test set after 30 years. This indicates that it is well-prepared and not over-fitted. We added a shortcut connection that takes an input from the first CNN layer and feds the output to the final layer of CNN because it reduces the information loss problem (Fig. 5).Fig. 5 CNN model accuracy Step 7: Labeling the information We identify two probabilities for our outcomes when we finish creating the model. ‘0’ denotes ‘no veil’ and ‘1’ denotes ‘mask.’ I am also using RGB values to define the coloring of the limit square shape. ‘RED’ for ‘without-veil’ and ‘GREEN’ for ‘with-mask.’ Step 8: Importing the face detection program Following that, we want to use it to detect whether we are wearing a face veil via our PC’s camera. To do so, we must first implement facial recognition. For this, I am using Haar feature-based cascade classifiers to identify the face’s highlights. OpenCV designed this course classifier to recognize the frontal face by preparing a huge number of images. The .xml file for this purpose should be downloaded and used to recognize the face. I have saved the record to my GitHub repository. Step 9: Detecting the faces with and without masks In the closing advance, we applied the OpenCV library to run an endless circle to make use of our net digital digicam wherein we identified faces using the cascade classifier. The code webcam = cv2.VideoCapture(zero) manner using webcam. The version will assume the threat of each one of the classes ([without-cover, with-mask]). In mild of which chance is higher, the mark may be picked and proven round our countenances. Moreover, we will download the DroidCam software for each Mobile and PC to make use of our portable digital digicam and alternate the motivation from zero to at least one in webcam= cv2.VideoCapture(1). Grad-Cam Several previous works (Inbaraj and Jeng 2021) have claimed that if more number of CNN layers are applied to an input data, then it captures higher-level visual constructs. Furthermore, CNN layers contain mainly spatial information that is lost in fully connected layers. For this reason, the final convolutional layers will provide the best compromise between high-level semantics and detailed spatial information. These layers’ neurons search the image for semantic class-specific information. Grad-CAM assigns importance values to each neuron for a specific decision of interest using gradient information flowing into the last convolutional layer of the CNN.Table 2 Comparison of performance with other deep learning models Accuracy CNN (Alok et al. 2020) VGG-19 (Jian et al. 2020) Efficient-Yolov3 (Xueping et al. 2022) Proposed model OA 98.90 97.62 98.18 99.94 KA 99 96.31 95.59 99.97 AA 98 94.07 98.1 99.95 Model description and formulation Figure 2 shows the model structure. In this section, we mainly explained each and every symbol present in Equations (1–7) (Richard et al. 2020). Face mask wearers and non-face mask wearers are two distinct populations, both with persons who fall into the following categories.: helpless (S); uncovered, for example inactively tainted (E); asymptomatically irresistible IA; apparently irresistible IS; and taken out (R). The eliminated class incorporates people who recuperated from contamination and the individuals who kicked the bucket. Coming into touch with inoculum generated by persons infected with SARS-CoV-2 can taint powerless people. We distinguish between inoculum generation by irresistible individuals, which provides free-living inoculum, and inoculum take-up and illness in helpless people. The inoculum can be obtained by inhaling transitory bead (D) forms that can be seen all over or by coming into touch with a decaying repository of inoculum stored by infected persons in the environment as fomites (F), which can survive for up to 72 h on certain surfaces. Rapid evidence of bead inoculum coexists with a more gradual rot of fomite (Fig. 2). As a result, there are two sets of transmission rates: A and S for inoculum production by asymptomatic and suggestive individuals, respectively, and D and F for inoculum take-up and illness of helpless people from bead and fomite inoculum, respectively. A handful of these limits are influenced by wearing a face mask (cf. mi in Fig. 2). Face masks reduce the amount of bead inoculum that escapes irresistible persons (Richard et al. 2020) by trapping a larger number of drops behind the veil mA,ms<1. Face masks also reduce the amount of bead inoculum breathed in by capturing a larger number of beads in the air, lowering the take-up transmission rate βD by (mD). (Fig. 2). At first, we assume that coverings have no effect on the risk of inoculum accessing inoculum from surfaces βF with mF = 1. In any event, the model considers how wearing a veil might increase the risk of fomite illness contamination mF>1, for example, by exposing the face to more consecutive touch while changing the cover. We note that critical PPE, for example, a full face-hood, could act to lessen the danger of fomite disease mF<1. Furthermore, sterilization interventions such as hand-washing may be proven by reducing the life expectancy of fomite inoculum (βF), and additional cleaning of surfaces or the use of faster self-sanitizing surfaces can be demonstrated by reducing the life expectancy of fomite inoculum (τF). We will focus on the effects of face masks and lockdown times in this section. The model is designed and settled as a fundamental deterministic differential condition model, which is summarized below for completion. The model may be quickly rebuilt in a stochastic framework with progress probabilities, as shown. It is also simple to divide the target country into metapopulations with varying contact rates, such as between urban communities and country zones or across age-groups in the population, and to geographically segment the population with limited inoculum pools. We use the model to look at substantial levels of how wearing a face mask complements a significant control method that involves the lockdown of a portion of the population. Accepting that lockdown reduces transmission rates (βi,i=A,S,D,F) by a predetermined amount, q, we may simulate this. It reduces the amount of inoculum generated by irresistible persons in open zones, which reduces the amount of inoculum available in the D and F pools, as well as the number of time susceptibles, spend in touch with that inoculum. Lockdown aims to reduce generally attractive spread rates by a factor of q2 in the model along these lines.1 dS/dt=-βFmFF+βDmDDS 2 dE/dt=-βFmFF+βDmDDS-ETE 3 dIA/dt=ETE-IATA 4 dIs/dt=IATA-ISTS 5 dR/dt=ISTS 6 dD/dt=-βAmAIA+βSmSISS-DTD 7 dF/dt=DTD-FTF In Table 2, we compared the accuracy obtained from the state-of-the-art models with our proposed model. It was discovered from this table that our proposed model has a higher accuracy (Figs. 6 and 7).Fig. 6 Evolution of the accuracy during training of the MobileNetV2 classifier on our dataset with small faces Fig. 7 Evolution of the accuracy during training for the BiT classifier, again on our small-faces dataset. The model still learns under intense data augmentation Evaluation metrics We can evaluate our machine learning algorithm using a variety of metrics. The confusion matrix is used to assess how well a model performs on test results. True positive, true negative, false positive, and false negative are the four groups in the confusion matrix (Margherita et al. 2020). False positive indicates the presence of a model-predicted entity that is not true. The term ‘false negative’ refers to the entity’s nonexistence. To be more accurate, it may be assumed that the prediction of the entity’s nonexistence was incorrect. True positive defines the entity’s actual presence when it is confirmed and identified. To be more specific, the model here attempts to detect the existence of something that is actually present and can be proven right. True negative identifies the entity’s nonexistence, which must be shown and indicated. To be more specific, the model seeks to classify the absence of something that is not present and must be proven incorrect. In this case, we select overall accuracy (OA), average accuracy (AA), and kappa accuracy (KA). Overall accuracy tells us how many of the references were correctly mapped out of all of them. The overall accuracy is typically expressed as a percentage, with 100 percent accuracy representing a perfect classification in which all reference sites were correctly classified. The kappa coefficient assesses the degree of agreement between classification and truth values. Equations 8, 9, and 10 for calculating the kappa accuracy (KA), overall accuracy (OA), and average accuracy (AA) are given as follows (Margherita et al. 2020):8 KA=ACCtotal-ACCrandom1-Accrandom 9 OA=K∗ACCavg+(2-K)2 10 AA=2∗OA-1K+1 Limitations of face detection system Poor image quality limits facial recognition’s effectiveness: The quality of an image has an impact on how effectively facial recognition algorithms perform. The visual quality of scanning video is poor when compared to that of a digital camera. Even high-definition video is typically 720p (progressive scan) at best. These numbers are around 2MP and 0.9MP, respectively, but a low-cost digital camera may attain 15MP. It is easy to see the difference. Small image sizes develop more difficult facial recognition: When a face recognition algorithm detects a face in an image or a still from a video clip, the relative size of the face compared to the overall picture size influences how effectively the face will be detected. Because of the limited picture size and the fact that the target is far away from the camera, only 100 to 200 pixels of the identified face are visible on one side. Furthermore, scanning a picture for varied face sizes is a processor-intensive activity. Most algorithms enable us to choose a face-size range to assist eliminating false positives in detection and speed up picture processing. Different face angles can throw off the reliability of facial recognition The recognition ranking is heavily influenced by the relative angle of the target’s face. Several angles are generally employed when a face is used in a recognition program (profile, frontal, and 45 degree are common). Something other than a frontal perspective has an influence on the algorithm’s capacity to generate a prototype for the face. The greater the score of any resultant matches, the more exact and better the resolution of the picture (both enrolled and probing image. Data processing and storage might limit technology for facial recognition Although the high-definition video has a lower resolution than digital camera footage, it still takes up a lot of disk space. Because processing every frame of film is a huge job, only a small percentage (10 percent to 25 percent) of it is actually processed through a recognition device. To minimize total processing time, agencies may employ computer clusters. Adding computers, on the other hand, necessitates huge data transfer via a network, which might be constrained by input–output restrictions, further slowing processing performance. How to overcome limits on facial recognition tools As technology advances, more high-quality cameras will become available. PC companies will be able to transmit more data, and processors will be able to operate faster. Face recognition algorithms will be better prepared to identify faces from a photograph and remember them in a database of selected persons. The fundamental mechanisms that underpin the current computations, such as darkening parts of the face with shades and veils or altering one’s hairdo, will be able to function properly. Changing how photographs are captured is a quick way to overcome a significant number of these barriers. When using checkpoints, for example, individuals are expected to organize themselves and channel via a single point. Cameras would be able to focus on each subject with more precision, resulting in indisputably more valuable frontal, higher-goal test images. Regardless, widespread use necessitates a greater number of cameras. Biometric applications that are progressing are promising. They include face recognition, as well as signals, articulations, stride, and vascular instances, such as iris, retina, palm print, ear print, voice acknowledgment, and scent marks, as well as iris, retina, palm print, ear print, voice acknowledgment, and fragrance marks. A combination of modalities is unparalleled in terms of improving a framework’s capacity to produce outcomes with more confidence. Related efforts focus on increasing the ability to acquire data from a distance where the target is aloof and often unaware. Without a doubt, security issues surround this breakthrough and its application. Finding a balance between public security and people’s protection rights will be a hot topic in the next years, especially as technology progresses. Conclusion This paper presents an inventive method to enhance the recognition of articles on face for our situation face cover wherein we play out a quick one shot output of veil. It outflanks or is at standard with different papers with a comparative plan in any event when our model is tried with lower-quality live recordings. The model was carefully tested with probable false-positive prospects that resulted in shambles of shirts folded over faces, handkerchiefs over the lips, and so on, and our model stood out as being more effective. Instead of a basic image classifier, the preparation included a dedicated two-class object identification. The problem with this method is that a face mask, via way of means of definition, hides a part of the face. Because the face mask detector cannot locate the face if sufficient of it’s far concealed, the face mask detector will now no longer be used. To get around this, we created a two-elegance item detector with mask elegance and without mask elegance. The version changed into progressed in methods via way of means of combining an item detector and a specialized mask class. For starters, the item detector changed to be capable of hitting upon folks carrying masks that the face detector could not able to hit upon as a result of the masks overlaying an excessive amount of the face. Future work Our model is made with a limited source of data, as the performance of neural networks increases with the data it is trained on, so we will try to incorporate more data and make it more robust and fault-proof. Furthermore, many causes of mistakes, such as brightness, posture, or partial image capture, can affect this detection. We will keep working to increase the technology’s accuracy. We are also working in to expand this project to make sure if a person is wearing a mask correctly or not. Often wearing masks below the nose is considered useless. We can even extend our project as a mode of surveillance, by using the face mask detection in street camera video. This will help in following the social distancing rules as given by the government which can be deployed in public areas like offices, railway stations, and airports. Funding No funding from any source. Data availability Inquiries about data availability should be directed to the authors. Declarations Conflict of interest There is no conflict of interest. Ethical approval This article does not contain any studies with human participants performed by any of the authors. Informed consent No such consent is required in studies. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Dalal N, Triggs B (2005) Histograms of oriented gradients for human detection. 2005 IEEE computer society conference on computer vision and pattern recognition (CVPR’05), vol. 1, pp 886–893, 10.1109/CVPR.2005.177 Grandini M, Bagli E, Visani G (2020) Metrics for multi-class classification: an overview. arXiv preprint arXiv:2008.05756 Inbaraj XA, Jeng J-H (2021) Mask-GradCAM: object identification and localization of visual presentation for deep convolutional network. 2021 6th international conference on inventive computation technologies (ICICT), pp 1171–1178, 10.1109/ICICT50816.2021.9358569 Jiao L Zhang F Liu F Yang S Li L Feng Z Qu R A survey of deep learning-based object detection IEEE Access 2019 7 128837 128868 10.1109/ACCESS.2019.2939201 Khan A, Sohail A, Zahoora U, Qureshi AS (2019) A survey of the recent architectures of deep convolutional neural networks. Artif Intell, Rev Liu L Deep learning for generic object detection: a survey Int J Comput Vis 2018 128 2 261 318 10.1007/s11263-019-01247-4 Memish ZA Zumla AI Al-Hakeem RF Al-Rabeeah AA Stephens GM Middle east respiratory syndrome coronavirus infections family cluster New Engl J Med 2013 368 26 2487 2494 10.1056/NEJMoa1303729 23718156 Mohamed SS, Tahir NM, Adnan R (2010) Foundation displaying and foundation deduction proficiency for entity recognition. 2010 sixth International Colloquium on Signal Processing and its Applications, Mallaca City, pp 1–6, 10.1109/CSPA.2010.5545291 Negi A, Chauhan P, Kumar K, Rajput RS (2020) Face mask detection classifier and model pruning with keras-surgeon. In: 2020 5th IEEE international conference on recent advances and innovations in engineering (ICRAIE), IEEE. pp 1–6 Organization WH et al. (2020) Coronavirus disease 2019 (covid-19): senerio report, 96 Padilla R Costa Filho CFF Costa MGF Evaluation of haar cascade classifiers designed for face detection World Acad Sci Eng Technol 2012 64 362 365 Piccardi M (2004) Foundation deduction strategies: an analysis. 2004 IEEE international conference on systems, man and cybernetics (IEEE Cat. No.04CH37583), The Hague, vol.4, pp 3099–3104, 10.1109/IC-SMC.2004.1400815 Ren S, He K, Girshick R, Sun J (2015) Faster R-CNN: towards real-time detection of objects with networks of area proposal. In: Proc Adv Neural Inf Process Syst, pp 91–99 Rota PA Oberste MS Monroe SS Nix WA Campagnoli R Icenogle JP Penaranda S Bankamp B Maher K Chen M-H Novel coronavirus characterisation linked with extreme acute respiratory syndrome Science 2003 300 5624 1394 1399 10.1126/science.1085952 12730500 Stutt ROJH, Retkute R, Bradley M, Gilligan CA, Colvin J (2020) A modelling framework to assess the likely effectiveness of facemasks in combination with ‘lock-down’ in managing the COVID-19 pandemic. Proceedings of the Royal Society A Su X Gao M Ren J Li Y Dong M Liu X Face mask detection and classification via deep transfer learning Multimed Tools Appl 2022 81 3 4475 4494 10.1007/s11042-021-11772-5 34903950 Teboulbi S Messaoud S Hajjaji MA Mtibaa A Real-time implementation of AI-based face mask detection and social distancing measuring system for COVID-19 prevention Sci Program 2021 2021 8340779 10.1155/2021/8340779 Xiao J Wang J Cao S Li B Application of a novel and improved VGG-19 network in the detection of workers wearing masks J Phys Conf Series 2020 1518 1 012041 10.1088/1742-6596/1518/1/012041	36475038	PMC9716506	NO-CC CODE	2022-12-03 23:20:57	no	Soft comput. 2022 Dec 2;:1-11	utf-8	Soft comput	2,022	10.1007/s00500-022-07700-w	oa_other
==== Front Chin J Acad Radiol Chin J Acad Radiol Chinese Journal of Academic Radiology 2520-8985 2520-8993 Springer Nature Singapore Singapore 109 10.1007/s42058-022-00109-2 Original Article The chest X-ray radiologic severity index as a determinant of the severity of COVID-19 pneumonia: study based on the duration of treatment and inpatient rooms Satoto Bambang [email protected] 1 Handoyo Thomas [email protected] 2 Sari Maya Nuriya Widya [email protected] 3 Santoso Antonius Gunawan [email protected] 4 http://orcid.org/0000-0001-6783-5895 Prasetyo Nicolaus Erlangga [email protected] 5 1 grid.412032.6 0000 0001 0744 0787 Thoracic Radiology, Faculty of Medicine, Universitas Diponegoro–Doctor Kariadi General Hospital, 16th, Dr. Soetomo Street, Semarang, 50249 Indonesia 2 grid.412032.6 0000 0001 0744 0787 Pulmonology, Faculty of Medicine, Universitas Diponegoro–Doctor Kariadi General Hospital, 16th, Dr. Soetomo Street, Semarang, 50249 Indonesia 3 grid.412032.6 0000 0001 0744 0787 Abdominal Radiology, Faculty of Medicine, Universitas Diponegoro–Doctor Kariadi General Hospital, 16th, Dr. Soetomo Street, Semarang, 50249 Indonesia 4 grid.412032.6 0000 0001 0744 0787 Interventional Radiology, Faculty of Medicine, Universitas Diponegoro–Doctor Kariadi General Hospital, 16th, Dr. Soetomo Street, Semarang, 50249 Indonesia 5 grid.412032.6 0000 0001 0744 0787 Radiology, Faculty of Medicine, Universitas Diponegoro–Doctor Kariadi General Hospital, 16th, Dr. Soetomo Street, Semarang, 50249 Indonesia 2 12 2022 18 16 10 2022 8 11 2022 20 11 2022 © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Introduction The chest X-ray examination is an imaging modality that is widely used in screening for COVID-19 pneumonia. The problems with treating COVID-19 pneumonia patients are the high incidence and severity of the disease and the limitations of treatment room facilities. The chest X-ray Radiologic Severity Index is expected to assist clinicians in obtaining the relationship between the extent of lesions on X-ray and the duration of treatment and hospitalization for COVID-19 pneumonia patients. Results This study used an observational method using a retrospective approach. The research subjects were COVID-19 pneumonia patients from March 2020 to April 2021 who were hospitalized at Doctor Kariadi General Hospital Semarang. A total of 105 subjects confirmed positive RT-PCR and received serial X-ray examination services during treatment. The calculation of the RSI value was carried out on all X-ray chest X-rays and then statistically analyzed using the paired T test and Mann–Whitney methods. There was no significant relationship between the value of RSI1 and the duration of hospitalization with p = 0.566, as well as the value of RSI2 with the duration of hospitalization with p = 0.715. There is a significant relationship between the values of RSI1 and RSI2 with the use of the intensive care unit with p < 0.000, respectively. There was a significant relationship between the values of RSI1 and RSI2 with the use of ventilators in treatment, with p < 0.000. Furthermore RSI1 and RSI2 have a good result as predictor of intensive care and ventilator usage. Conclusion The chest X-ray RSI has no significant relationship with the duration of hospitalization. The value of the chest X-ray RSI has a significant relationship with the use of intensive care rooms and the use of ventilators in treatment. The increase in the RSI value can describe the severity of the disease so that it plays a role in planning the treatment room. Keywords COVID-19 pneumonia Chest X-ray Radiologic severity index Intensive care Ventilator Inpatient ==== Body pmcIntroduction COVID-19 is a disease that emerged around December 2019 with pneumonia symptoms that developed regionally around the city of Wuhan, China. This disease caused by SARS-CoV-2, spread widely and became a global pandemic on March 12, 2020, after it was found quite fast in 114 countries around the world. Indonesia reported the development of confirmed cases of COVID-19 from almost all provinces. With a high incidence rate and a very broad spectrum of symptoms, COVID-19 pneumonia provides a high burden, especially with the availability of health care facilities. Patients with severe symptoms require treatment in the intensive care unit as soon as possible and may even require the use of a ventilator [1–4]. Imaging has an important role in COVID-19 pneumonia, from screening to assisting clinicians in making a diagnosis before RT-PCR results are available. COVID-19 pneumonia is a new emerging disease that is not fully known yet. Several previous studies conducted by Yoon SH, Lee KH, Kim JY, Lee YK, Ko H, Kim KH, et al. and Borghesi A, stated that COVID-19 is an interstitial pneumonia with typical findings of bilateral multifocal ground-glass opacity or consolidation dominantly in the basal peripheral lung zone. Imaging also plays a role in providing an overview of the severity and progression of the disease so that it can help clinicians plan treatment. A Radiologic Severity Index assessment was used to assess the grade of the lesion based on the extension of the lesion and the density of the lesion on a chest X-ray [5–7]. This study was conducted to assess lesions on X-rays and their relationship to the duration of treatment and inpatient room in patients with COVID-19 pneumonia, so it is hoped that it can assist clinicians in planning patient care by considering the duration of treatment and inpatient room. Material and methods This study used an observational method with a retrospective approach. This research was conducted after obtaining a research permit and an ethical certificate from the Health Research Ethics Commission of Doctor Kariadi General Hospital Semarang. The minimum sample size for this study was determined based on the proportion test for 2 groups in pairs for a retrospective study. According to the proportion of COVID-19 in Indonesia of 24.5% (in September 2020), the formula np = ((Z + Zb)2 × proportion)/significant error2, with a significance level of 0.05, CI 95%, power of 80%, and significant error margin 20%, a minimum sample size of 48.3 samples is obtained. The inclusion criteria in this study were COVID-19 hospitalized patients who received serial chest X-ray examination services at least 2 times in 1 hospitalization period. Many other disease and infections can resemble the chest X-ray findings of COVID-19 pneumonia, therefore this study used several exclusion criteria. The exclusion criteria is the presence of comorbid diseases or a history of previous chest X-ray images that could resemble COVID-19 pneumonia or could obscure the assessment of pulmonary lesions such as lung tuberculosis, primary lung tumor and lung metastases lesion, lung edema, interstitial lung disease, pleural effusion and pleural or thoracic wall tumor. Of the more than 5000 patients with positive PCR results from March 2020 to April 2021, approximately 300 adult patients, 18 years or older, received inpatient services at the Doctor Kariadi General Hospital. Of the 300 hospitalized patients, only less than 200 patients met the exclusion criteria. Moreover, many samples did not have a complete data set in medical record. Therefore, this study used 105 subjects who met the inclusion and exclusion criteria. Secondary data on COVID-19 patients was obtained from the COVID team and electronic medical records. This study used a posteroanterior or anteroposterior projection thoracic X-Ray modality. The chest x-ray examination was carried out using the GE Proteus XR/a Type RAD-14 ionizing radiation source at the radiology installation and the Philips Mobile Practix 360 Type IAE X22 onsite ionizing radiation source. X-rays of the entire study subject were assessed using the RSI method. All chest X-rays on the research subjects were re-read and RSI was calculated. The chest images were read randomly by the researchers without looking at the time of examination or the order of the examination dates in chronological order, but still paying attention to the patient's name and registration number to avoid identity errors. Interpretation and calculations differences of each researcher are resolved through deliberation and decided by the main researcher. The RSI scoring system uses two main variables: the extension of the lesion and the volumetric density of the lesion. In the RSI assessment, the spread of the lesion was assessed based on the location of the lesion in the lung fields; the right and left lung fields were divided into three parts, respectively; the upper zone (up to the carina), the middle zone (below the carina to the upper border of the inferior pulmonary vein), and the lower zone (below the upper border of the inferior pulmonary vein). An inferior pulmonary vein can be identified by a linear image of the right hilum pointing to the periphery and crossing the pulmonary artery [8]. The lesion pattern was divided into 3 value categories; 1 for normal lungs, 2 for ground-glass opacity images, and 3 for consolidation images. Ground-glass opacity was defined as an increase in the density of the lung parenchyma but not as thick as consolidation so that the boundaries of the broncho-alveolar tract could be clearly seen [8]. If there were areas with both ground-glass opacity and consolidation lesion, then it was calculated as consolidation which gave highest score. The volumetric area is divided into 5 broad categories; 0%, 1 for an area of 1–24%, 2 for an area of 25–49%, 3 for an area of 50–74%, and 4 for an area of 75–100%. The RSI was obtained by adding up the multiplication of the value of the lesion pattern and its volumetric area in each zone, with a total value between 0 and 72. The independent variables of this study were the factors that formed the values of RSI. The dependent variables in this study were the RSI value on chest X-ray, the duration of hospitalization, and the inpatient room type [5–8, 13]. The data that has been collected is presented in tabulated form and analyzed using the SPSS 2.5 for Windows program. A receiver operating characteristic (ROC) curve was used to determine the cut value with high sensitivity and specificity. Furthermore, an analysis of the sensitivity and specificity was carried out, which was used to measure the accuracy of the diagnostic test. The cross tabulation of the RSI cut-off was statistically analyzed with a significant relationship analysis result if p < 0.05. Results Clinical characteristics of research subjects research The subjects were taken from the patient's medical record data. The number of research subjects was 105 patients, consisting of 58 men (55%) and 47 women (45%). The mean age of the subjects was 51.3 with the youngest age being 23 years and the oldest being 77 years old, and the median age was 54 years. The mean number of days of treatment was 13.5 days, with a standard deviation of 8.1. The shortest number of days of treatment was 1 day, and the number of days of longest treatment was 43 days, with a median value of 12 days. During the hospitalization period, 44 subjects were treated in the intensive room and the other 61 subjects were treated in the non-intensive room and. Of the 44 patients treated in the intensive care unit, 39 patients were used ventilator. 41 patients were declared dead at the end of the treatment period, and 64 other patients were declared cured and could be hospitalized. Patients receive serial X-ray examination with an average of 3.2 photos, with the minimum number is 2 photos and the maximal number is 9 photos per subject in 1 hospitalization period. No medical history or imaging finding of lung tuberculosis, primary lung tumor and lung metastases lesion, lung edema, interstitial lung disease, pleural effusion and pleural or thoracic wall tumor was include in this research. X-ray examination results with RSI The RSI calculation results are displayed in tabular form and analyzed to obtain a relationship with the duration of treatment and the type of treatment room. In this study, the correlation between RSI values and length of hospitalization and the type of inpatient room was tested based on the RSI value in the first and second chest X-ray of the subject. The first and second chest X-ray were chosen because all objects were given at least 2 chest X-ray services and the RSI results on both photos had normal data distribution on all research subjects. The third chest X-ray and so on were not analyzed because not all patients were provided with the third and so on. The chest X-ray RSI assessment can be seen in Figs. 1, 2, 3 and 4. The average RSI value in the initial chest X-ray (hereinafter referred to as RSI1) is 25.4 (standard deviation 17.7), with the smallest value of 0 and the largest value being 66, and the median 23. The average RSI value in the second chest X-ray (hereinafter referred to as RSI2) is 31.8 (standard deviation 18.6), with the smallest photo value of 0 and the largest value of 72. A normality test was carried out on the variable length of stay, RSI1 and RSI2 using the Kolmogorov–Smirnov method and obtained a normal distribution of data, so it was continued with a correlation test using the Spearman's rho method.Fig. 1 Serial chest X-ray Radiologic Severity Index in COVID-19 pneumonia patient. A RSI values on initial chest x-rays at 2020, March 16th and B RSI values on second chest x-rays at 2020, March 21st. There is an increase in RSI value at 5-days interval [12] Fig. 2 Serial chest X-ray RSI in COVID-19 pneumonia patient with shortest hospitalization duration. A RSI values on initial chest x-rays was 14 and B RSI values on second chest x-rays at the same day was 50 Fig. 3 Serial chest X-ray RSI in COVID-19 pneumonia patient with longest hospitalization duration. A RSI values on initial chest x-rays was 14 and B RSI values on second chest x-rays at 3 days interval was 51 Fig. 4 Serial chest x-ray RSI in COVID-19 pneumonia patient that using ventilator during hospitalization. A RSI values on initial chest x-rays was 60 and B RSI values on second chest x-rays at the same day was 66. There is an increase in RSI value at 5-days interval Table 1 shows the correlation between length of stay and RSI1 and RSI2 values based on the Spearman method. A significant relationship was indicated by a p value < 0.005, while in the correlation test the duration of hospitalization with RSI1 was 0.566 and with RSI2 was 0.715. From the correlation test, it was found that there was no significant relationship between the values of RSI1 and RSI2 with the duration of hospitalization. Furthermore, the relationship between RSI1 and RSI2 values was tested with the type of treatment room, namely the intensive care unit (ICU) and non-intensive unit.Table 1 Correlation of RSI value to duration of hospitalization with Spearman RSI1* RSI2** Length of Stay Correlation coefficient 0.057 0.036 Sig. (2-tailed) 0.566 0.715 Sum 105 105 Initial Chest X-ray RSI Serial Chest X-ray RSI The mean value of RSI1 in ICU patients is 36.3 (SD 17.5), with the smallest value being 0 and the largest value being 66, whereas the average value of RSI1 in non-ICU patients is 17.4 (SD. 13.1), with the smallest value being 0 and the largest value being 51. The RSI2 value in patients treated in the ICU was 45.3 (SD 14.5) with the smallest value of 6 and the largest value of 72, while in non-ICU patients the average value of RSI2 was 22.1 (SD 14.9) with the smallest value of 0 and the largest value of 57. Table 2 shows the relationship test of the RSI1 value with the type of inpatient room. After the normality test, the relationship between the RSI1 value and the treatment room was tested. From the results of statistical tests using the Mann–Whitney method, there was a significant relationship between the value of RSI1 and ICU and non-ICU treatment rooms with p = 0.000 (significant p < 0.0; CI 95%). Table 3 shows the relationship test of the RSI2 value with the type of inpatient room. From the results of statistical tests using the unpaired T test method, it was found that there was a significant relationship between the values of RSI2 with ICU and non-ICU treatment rooms with p = 0.000 (meaning p < 0.05; CI 95%).Table 2 Relevance of RSI1 value to inpatient room type Room type Sum Mean RSI1 Std. dev Mann–Whitney U Z Asymp.Sig. (2-tailed) ICU 44 36.39 17.2 544.0 − 5.186 0.000 Non ICU 61 17.48 13.1 Mann–Whitney method Table 3 Relevance of RSI2 value to inpatient room type Room type Sum Mean RSI2 Std. dev T df Sig. (2-tailed) ICU 44 45.30 14.5 7.897 103 0.000 Non ICU 61 22.18 14.9 Non paired T test method Furthermore, the relationship between RSI 1 and RSI 2 values with the use of a ventilator was tested. Of 105 subjects, 39 subjects were treated using a ventilator and 66 subjects were treated without using a ventilator. The mean RSI1 value in the subjects treated with the ventilator was 36.1 (SD 18.2). This value is higher than the mean RSI1 value in subjects treated without a ventilator, which is 19.0 (SD 14.0). Likewise, the mean RSI2 value in subjects treated with a ventilator was 45.2 (SD 15.4), which was also higher than in subjects treated without a ventilator, with a mean of 23.9 (SD 15.7). Table 4 shows the relationship between the RSI1 value and the use of a ventilator. Statistical tests were carried out using the Mann–Whitney method and there was a significant relationship between the increase in the RSI1 value and the use of a ventilator in treatment with p = 0.000 (significant p < 0.05; CI 95%). Table 5 shows the relationship between RSI2 values and the use of ventilators. Statistical tests were carried out using an unpaired T test and there was a significant relationship between the increase in the RSI2 value and the use of a ventilator in treatment with p = 0.000 (significant p < 0.05; CI 95%).Table 4 Relevance of RSI1 to ventilator usage Ventilator Sum Mean RSI1 Std. dev Mann–Whitney U Z Asymp.Sig. (2-tailed) Yes 39 36.18 2.91 605.5 − 4.523 0.000 No 66 19.03 1.73 Mann–Whitney method Table 5 Relevance of RSI2 to ventilator usage Ventilator Sum Mean RSI2 Std. dev T df Sig. (2-tailed) Yes 39 45.21 2.47 6.720 103 0.000 No 66 23.98 1.93 *Non paired T test method Based on the data distribution of RSI1 values, hypothesis testing was carried out using Chi-square test with a cut-off value of 22.5 to obtain a sensitivity of 72% and specificity of 65% to intensive care. The cut-off value of 22.5 has a significant relationship to intensive care with p = 0.000 (significant p < 0.05; CI 95%). The risk estimation analysis resulted in a PR value of 5.079. Hypothesis testing was also carried out on the cut-off value of 22.5 to the use of ventilator to obtain sensitivity of 71% and specificity of 62% with p = 0.002 (significant p < 0.05; CI 95%). The risk estimation analysis resulted in a PR value of 4.175. Cross tabulation of RSI1 to intensive care and ventilator using can be seen in Tables 6 and 7.Table 6 Cross tabulation RSI1 and RSI2 categories to inpatient room type RSI1 category Inpatient room p PR (95% CI) ICU Non ICU ≥ 22.5 (n = 53) 32 (60.4%) 21 (39.6%) 0.000 5.079 (2.175–11.860) < 22.5 (n = 52) 12 (23.1%) 40 (76.9%) RSI2 category Outcome p PR (95% CI) Died Alive ≥ 29.5 (n = 54) 37 (68.5%) 17 (31.5%) 0.000 13.681 (5.120–26.553) < 29.5 (n = 51) 7 (13.7%) 44 (86.3%) PR Prevalence ratio Table 7 Cross tabulation RSI1 and RSI2 categories to ventilator usage RSI1 category Inpatient room p PR (95% CI) ICU Non ICU ≥ 22.5 (n = 53) 28 (52.8%) 25 (47.2%) 0.002 4.175 (1.773–9.832) < 22.5 (n = 52) 11 (21.2%) 41 (78.8%) RSI2 category Outcome p PR (95% CI) Died Alive ≥ 29.5 (n = 54) 32 (59.3%) 22 (40.7%) 0.000 9.143 (3.484–23.992) < 29.5 (n = 51) 7 (13.7%) 44 (86.3%) PR Prevalence Ratio Based on RSI2 data distribution, hypothesis testing was carried out using a cut-off value of 29.5 to obtain sensitivity of 84% and specificity of 72% to intensive care. The cut-off value of 29.5 has a significant relationship to intensive care with p = 0.000 (significant p < 0.05; CI 95%). The risk estimation analysis resulted in a PR value of 13.681. Hypothesis testing was also carried out on the cut-off value of 29.5 to the use of ventilator to obtain sensitivity of 82% and specificity of 66% with p = 0.000 (significant p < 0.05; CI 95%). The risk estimation analysis resulted in a PR 9.143. Cross tabulation of RSI2 to intensive care and ventilator using can be seen in Tables 6 and 7. Discussion The results of this study indicate a significant relationship between the RSI value and the selection of ICU or non-ICU inpatient rooms and the use of ventilators in treatment. The results of the RSI1 and RSI2 diagnostic tests are in line with previous studies conducted by Sheshadri, Shah, Godoy, et al., which found that the X-scoring system using the RSI method was associated with increased mortality. Although the RSI in this study was carried out in patients with parainfluenza viral pneumonia, a significant relationship was also found in this study conducted in patients with SARS-CoV-2 pneumonia. In accordance with research conducted by Borghesi, Zigliani, Golemi et al. and research by Lotfi et al., the results of RSI values showed a significant correlation to mortality and severity of COVID-19. The cut-off value of RSI1 ≥ 22.5 gives a PR of 5.079 to intensive care room and PR of 4.175 to ventilator using, which means that subjects with a RSI1 value more than equal to 22.5 have a 5.079 times greater probability treated in intensive care room and a 4.175 times greater probability of using ventilator during treatment than subjects with RSI1 value < 22.5. The cut-off value of RSI2 ≥ 29.5 gives a PR of 13.681 to intensive care room and PR of 9.143 to ventilator using. These results indicate that subjects with RSI2 ≥ 29.5 have 13.681 times greater probability treated in intensive care room and a 9.143 times greater probability of using ventilator during treatment than subjects with RSI2 value < 29.5. This study show that increased area and density of lesions in the lung parenchyma generate an increase in the RSI value, is associated with an increase in the patient's severity of illness. The greater the severity of the disease, the greater the probability for treatment in the intensive room and the greater the need for the use of a ventilator [6, 9–13]. The results of the study did not show a significant relationship between the RSI value and the duration of treatment. This happens because the spectrum of clinical symptoms of patients with COVID pneumonia is broad, so that severe symptoms do not always correspond to a long duration of treatment. Some subjects are admitted to the hospital with mild symptoms, while another patients are admitted with severe symptoms. In addition, some patients had high RSI values with a short duration of treatment because they were declared dead at the beginning of treatment. This study showed that the increase in severity based on the increase in lesion area and lesion density was not directly proportional to the length of treatment duration. The limitations of this study include not being able to assess the trend of increasing the RSI value for the length of treatment and also not being able to show a stratification of the degree of disease, either mild, moderate, or severe. This limitation was caused, among other things, because not all research subjects received a uniform chest X-ray examination in the number and interval of examinations so that the distribution of the data was uneven. A larger number of research subjects is needed to get a better analysis studies. This study has not included the relationship of patient predisposing and comorbid factors to severity of the disease and also has not shown the effect of medication administration on changes in the lesions on the thorax during the treatment period. Conclusions The RSI value did not have a significant relationship with the duration of care for patients with COVID-19 pneumonia. RSI values have a significant relationship with intensive care rooms and ventilator use in patients with COVID-19 pneumonia. The chest X-ray RSI value gave good results as a predictor of intensive care and ventilator usage of COVID-19 pneumonia patients. The RSI value can be used by clinicians as a method of assessing the severity of COVID-19 pneumonia and planning treatment. Acknowledgements The authors thank to Radiology Departement Doctor Kariadi General Hospital for being a source of the research data set. Author contributions BS interpreted and assessed the RSI of all the chest X-ray, statistical calculations, and major contributor in writing the manuscript. TH clinically assessed all research subjects, collected basic references of COVID pathologic related diseases. MNWS re-interpreted and re-assessed all the chest X-ray RSI value, collected basic references of COVID imaging technique, minor contributor in writing manuscript. AGS re-interpreted and re-assessed all the chest X-ray RSI value, editing the manuscript format. NEC collected the data sets and editing the manuscript format. All authors read and approved the final manuscript. Funding No funding was obtained for this study. Data availability The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Conflict of interest The authors declare that they have no competing interests. Ethical approval The study is approved by Doctor Kariadi General Hospital Ethical and Scientific Committee based on research permit letter and ethical exemption certificate. The description of ethical exemption Number 600/EC/KEPK-RSDK/2020 was sign by M. Sofyan Harahap, M.D, Anesthesiologist Consultant as the Chairperson of Ethical and Scientific Committee of Doctor Kariadi General Hospital Semarang. Based on medical record and PACS database in Doctor Kariadi General Hospital Semarang, no patient informed consent was obtained written nor verbally. Consent for publication The data sets of this study was based on medical record and PACS database in Doctor Kariadi General Hospital Semarang, no patient informed consent for publication was obtained written nor verbally. All the images and data sets were entirely unidentifiable and there were no details on individuals reported within the manuscript. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Yang R Li X Liu H Zhen Y Zhang X Xiong Q Chest CT severity score: an imaging tool for assesing severe COVID-19 Radiol Soc North Am. 2020 282 2 e200047 2. Kemenkes RI. Pedoman Pencegahan dan Pengendalian Coronavirus Disease (COVID-19). Germas. 2020;0–115. 3. Burhan E, Isbaniah F, Susanto AD, Aditama TY. Pneumonia COVID-19 Diagnonsis & Penatalaksanaan di Indonesia. Perhimpunan Dokter Paru Indonesia. Jakarta: Perhimpunan Dokter Paru Indonesia; 2020. 4. Gugus Tugas Percepatan Penanganan COVID-19. Pedoman Penanganan Cepat Medis dan Kesehatan Masyarakat COVID-19 di Indonesia [Internet]. Setiawan AH, Rachmayanti S, Kiasatina T, All E, editors. Jakarta, Indonesia; 2020. 2020 p. Available from: www.covid19.go.id 5. Yoon SH Lee KH Kim JY Lee YK Ko H Kim KH Chest radiographic and CT findings of the 2019 novel coronavirus disease (Covid-19): analysis of nine patients treated in Korea Korean J Radiol 2020 21 4 498 504 10.3348/kjr.2020.0132 6. Sheshadri A Shah DP Godoy M Erasmus JJ Song J Li L Progression of the Radiologic Severity Index predicts mortality in patients with parainfluenza virus-associated lower respiratory infections PLoS ONE 2018 13 5 1 18 10.1371/journal.pone.0197418 7. ACR. ACR Recommendations for the use of Chest Radiography and Computed Tomography (CT) for Suspected COVID-19 Infection [Internet]. American College of Radiology. 2020. pp. 18–9. Available from: https://www.acr.org/Advocacy-and-Economics/ACR-Position-Statements/Recommendations-for-Chest-Radiography-and-CT-for-Suspected-COVID19-Infection 8. Lange S Walsh G Examination technique and normal findings Radiology of chest disease 2007 3 New York Thieme 1 49 9. Eldaboosy SAM Halima KM Shaarawy AT Kanany HM Elgamal EM El-Gendi A-A Comparison between CURB-65, PSI, and SIPF scores as predictors of ICU admission and mortality in community-acquired pneumonia Egypt J Crit Care Med [Internet]. 2015 3 2–3 37 44 10.1016/j.ejccm.2015.10.001 10. Lotfi M. Introduction of a Radiologic Severity Index for the 2019 Novel Corona Virus (COVID-19 ). 2019;1–19. 11. Borghesi A COVID-19 outbreak in Italy: Experimental chest x-ray scoring system for quantifying and monitoring disease progression Radiol Med 2020 10.1007/s11547-020-01200-3 32852750 12. Borghesi A Maroldi R Radiographic severity index in COVID-19 pneumonia : relationship to age and sex in 783 Italian patients Radiol Med 2020 10.1007/s11547-020-01202-1 32852750 13. Prasetyo NE Satoto B Handoyo T The relevance of chest X - ray radiologic severity index and CURB - 65 score with the death event in hospitalized patient with COVID - 19 pneumonia Egypt J Radiol Nucl Med [Internet]. 2022 53 191 10.1186/s43055-022-00877-y	36474596	PMC9716507	NO-CC CODE	2022-12-03 23:20:57	no	Chin J Acad Radiol. 2022 Dec 2;:1-8	utf-8	Chin J Acad Radiol	2,022	10.1007/s42058-022-00109-2	oa_other
==== Front AMS Rev AMS Review 1869-814X 1869-8182 Springer US New York 240 10.1007/s13162-022-00240-4 Theory/Conceptual Robots in retail: Rolling out the Whiz Rindfleisch Aric [email protected] 1 Fukawa Nobuyuki [email protected] 2 Onzo Naoto [email protected] 3 1 grid.35403.31 0000 0004 1936 9991 University of Illinois, Champaign, IL USA 2 grid.260128.f 0000 0000 9364 6281 Missouri University of Science & Technology, Rolla, MO USA 3 grid.5290.e 0000 0004 1936 9975 Waseda University, Tokyo, Japan 2 12 2022 17 21 9 2022 30 9 2022 © Academy of Marketing Science 2022, Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Robots are increasingly being employed in retail settings to accomplish a wide variety of tasks. In the years ahead, it is expected that most retailers will employ robots in some capacity and that these robots will alter the role of employees and change the nature of customer experience. However, since this revolution is in its early stages, scholarship in this domain is largely forward looking in nature and focused on the future rather than the present. Our research seeks to enrich and extend this literature by examining a recent robot deployment (i.e., the Whiz) across a large Japanese retail chain (i.e., Daiei). Specifically, we report an interview with executives from both Daiei as well as Softbank Robotics (the manufacturer of the Whiz). This interview touches upon a number of interesting topics including, how this robot is currently being deployed, how employees and customers are responding to this robot, and how it impacts store operations and retailer performance. We then reflect upon this interview to offer a set of future research directions. Our article is also accompanied by a commentary by Guha and Grewal (2022, in this issue) that offers additional insights into robots in retail. ==== Body pmcIntroduction The Robot Revolution is upon us. A growing number of retailers around the world are using robotic technology to supplement and/or enhance human employees (Grewal et al., 2021). For example, Saks Fifth Avenue is currently employing warehouse robots (by GXO Logistics) to help fulfill online orders (Loten, 2021), while Sam’s Club is using robots (by Brain Corp) to clean its store floors (Repko, 2021). According to a recent survey, approximately half of all retailers plan to employ in-store robots in some capacity by 2023 (Brain Corp., 2021). More broadly, the worldwide retail robot market is expected to grow from $7 billion in 2020 to over $55 billion by 2028 (Coherent Market Insights, 2021). This escalating interest in robot workers has been spurred by a variety of drivers, including advances in robotic technology, human labor shortages, and the need for social distancing in the wake of COVID-19 (Huang & Rust, 2021; Shankar et al., 2021; Subero-Navarro et al., 2022). The rise of retail robots has also attracted attention from a growing number of marketing scholars (e.g., Grewal et al., 2021; Huang & Rust, 2018, 2021; Noble et al., 2022). Since the Robot Revolution has just begun, most research in this domain is forward-looking in nature and focused on identifying future research directions. For example, Huang and Rust (2018) predict that in the near future, many service-related tasks (such as cleaning and maintenance) will be performed by robots rather than humans. Likewise, Grewal et al. (2021) suggest that research is needed to understand how in-store robots affect shopping behavior as well as how retailers could leverage robots to enhance profitability. In this article, we provide some preliminary insights into these types of issues via an interview with some early movers in this domain (i.e., Daiei and SoftBank Robotics/Iris Robotics). This interview is accompanied by a commentary by Abhijit Guha and Dhruv Grewal (Guha & Grewal, 2022, in this issue), who offer additional insights about retailing’s coming Robot Revolution. Background Our interview focuses upon the recent establishment of an in-store robot called the Whiz. This AI-enabled vacuum cleaning robot was initially developed by SoftBank Robotics, one of the largest robotics firms in Japan (www.softbankrobotics.com/jp/product/whiz). To date, over 20,000 of these robots have been shipped to customers across multiple countries, such as the US, Japan, and Singapore (Iris Ohyama, 2022a; SoftBank Robotics, 2020). The Whiz can operate autonomously and be trained to remember multiple cleaning routes in a given store. In addition, its built-in sensors allow the Whiz to avoid both shoppers and obstacles (e.g., products, shelving). The Whiz can operate continuously for several hours before returning to its charging base (Iris Ohyama, 2022a). In early 2021, SoftBank Robotics formed a partnership with Iris Ohyama (a Japanese home appliance manufacturer) to create a new company called Iris Robotics, which currently handles sales and services of the Whiz in Japan. This version of the Whiz is displayed in Exhibit 1. To date, this robot has been adopted by over 2,500 Japanese firms (Iris Ohyama, 2022b). One of its primary customers is Daiei, a national chain of supermarkets with over 200 stores across Japan (Daiei, 2022). Daiei introduced the Whiz into 79 of its stores in July 2021.1 At present, this represents the largest robot deployment of any retailer within Japan. This deployment has attracted considerable attention, as evidenced by Daiei’s recent selection as the recipient of the 2021 AI Retail Award (Retail AI Research Society, 2021). In order to learn more about the Whiz and how it is being employed by Daiei, we conducted an interview with executives from both SoftBank Robotics/Iris Robotics (Mr. Jun Ikeda) and Daiei (Ms. Maki Totani). Mr. Ikeda is a Director of Business Development for SoftBank Robotics/Iris Robotics and Ms. Totani is the Kanto District Director of Sales Planning for Daiei. We conducted a concurrent interview (via Zoom) with both of these executives on June 9, 2022. Our questions focused on understanding the motives behind Daiei’s introduction of the Whiz, how it currently operates, and how the introduction of this retailing robot has been received by both employees and customers. In addition, we solicited thoughts from its manufacturer, SoftBank Robotics and obtained predictions about the future of robots in retailing. This interview was conducted in Japanese and lasted approximately one hour. With permission of the participants, this interview was recorded and the contents of this recording were transcribed into Japanese and then translated into English by a professional translation firm. Exhibit 1 (Whiz, Iris Edition) Interview Question Can you tell us a bit about the robot? How did you begin using the robot? What is the story behind it? Mr. Ikeda (SoftBank Robotics) We originally introduced the Whiz (at Daiei) to enhance store hygiene and provide a new customer experience. We first placed it in 18 stores and eventually expanded it to 79 stores. Daiei views this initiative as a way to improve cleanliness and reduce costs. Also, we are using this robot for sales promotion. So far, we have used the robot to promote wine, cakes, and sushi rolls. The levels of both store cleanliness and product sales have improved. A comparison of stores with and without the robot shows that sales are higher at stores using the robots. We have also administered a customer satisfaction survey. The results of the survey indicate that customers and employees are happy with the robot so far. Ms. Totani (Daiei) The Whiz was introduced due to increased interest in hygiene management of our facilities following COVID-19. Due to limited staff, our stores needed help with cleaning and sterilization. Also, there is a limit to what can be achieved by humans. This is where the robot became useful. It proved to be an indispensable ally and attracted curiosity. We first demonstrated the Whiz in a store during business hours, which became the initial trigger. In addition to being an effective cleaning system, we believe that the unique appearance of the Whiz has a futuristic vibe. We also found that the robot received a great deal of positive feedback from our customers, particularly children and employees. Since its introduction as a cleaning robot, we have been considering its potential for a broader range of versatile applications. We introduced the Whiz to 79 stores in the Kanto (Tokyo area) and the Kinki (Osaka area) regions. Question How long does the robot operate per day? Ms. Totani (Daiei) Three hours, on average. Mr. Ikeda (SoftBank Robotics): It takes roughly one hour per run (average three runs per day). On average, there are three hours of cleaning, with each store deciding on a time convenient for them to operate, such as 10:00am, 3:00pm, and 7:00pm. Some stores have a shorter period of one hour, whereas others have a longer period of up to five or six hours. It can remember up to 600 routes. Question Can you tell us how you are using this robot and what types of outcomes that you have found thus far? Could you describe any challenges in introducing and operating the robot? Ms. Totani (Daiei) We have received excellent evaluations in terms of the robot’s cleaning precision. Mr. Ikeda (SoftBank Robotics) We have a device that lets us see the relative cleanliness of the floor. On average, using the Whiz leads to a 50% improvement (based on a test that measures the cleanliness of a floor surface). In particular, the robot decreases the number of invisible particles on the store floor. Ms. Totani (Daiei) In the beginning, its primary function was cleaning and sterilization. My department was asked if we could use the robot for sales promotions as well. However, we were unsure if it was possible but are currently working on using the Whiz for improving sales promotion. We are extremely focused on this as a department. In addition to improving sales, the robot also boosts cleanliness and prevents crime. We are looking at the satisfaction level of customers visiting the stores and the buzz generated by having robots like this in our stores. Our employees are also pleased to see customers watching the robots with interest. We are working to ensure that this initiative produces a positive outcome. Thus, we are checking what kind of posts about the robot appear on social networking sites, and after seeing some customer postings, our stores started decorating their cleaning robots. It is difficult to say whether sales promotions are more effective while the robot is in operation. It is challenging to ascertain a causal relationship between the two. It is very complex and there are a lot of things to consider. We experimented several times with sales promotions. One of the first trials was a promotion of Japanese-style confectioneries. We displayed these confectionaries on the sales floor, as well as on the robots via POPs. The POPs included promotional materials and audio recordings. Since the number of units sold vary across stores and time, it was difficult to find a cause-and-effect relation between the robot and product sales. We also experimented with catalog advertising. We loaded the robot with an advertisement that said, “We are now accepting orders for Beaujolais Nouveau wine.” We then compared the number of wine reservations between stores that deployed robots and stores that did not deploy robots. Reservations seemed to increase in the stores that used the robot compared with those that did not. We then asked manufacturers to help subsidize this type of robot-based sales promotion. We first looked at manufacturers that could not engage in in-store promotion (food sampling) due to Covid-19, thinking they may have some left-over budget. We negotiated a subsidy in return for letting them use our robot to promote and advertise their products and conducted field experiments in four stores. These experiments worked very well. It is a trial-and-error process that is still in the early stages with many twists and turns. Question What are the incentives for manufacturers to participate in these trials? Ms. Totani (Daiei) Well, a manufacturer’s goal is to sell products. The expense is justified if the store can sell more products because of the promotion. Right now, COVID-19 has made it impossible to do tasting samples. However, manufacturers still want to advertise their products, so they are now starting to look for other methods. Question How have employees and customers reacted to this robot and how has it affected your firm’s marketing efforts? Ms. Totani (Daiei) The store staff felt the robots were like characters. Our company’s mascot is “Mokkun.” So we tried to apply that to the Whiz by giving it a wide face with eyes, a nose, and a mouth. Also, many stores have started decorating their robots in various ways. For example, one store has created an “Ebinya” robot based on a character from Ebina City (a city in Kanagawa-prefecture in Kanto region). They obtained permission from the city to use that character and decorated their local Whiz to make it look like the character “Ebinya” is trundling around cleaning. Several other stores have also begun to take these kinds of steps. Then, there are customers. We see people post about the robot on Twitter and so on. In particular, customers made postings of our robots displaying different promotions. Mr. Ikeda (SoftBank Robotics) Social networking sites are already showing actual customer reactions. We receive direct feedback on these sites, with comments such as “This is an amazing age to be alive.” There are many photos and as far as I can see, the posts are almost always positive. We have seen a great number of posts about how the robots are “advanced” and “amazing.” As for whether our employees are satisfied with the Whiz, a recent survey demonstrated that 97% are satisfied and that 95% would like to continue using it in the future. Question How have other firms in your industry responded to your introduction of robots? Mr. Ikeda (SoftBank Robotics) Daiei was the first company to create an environment in which robots operate in their supermarkets during the daytime on such a large scale. So other companies began to wonder if it was safe to have a robot running around the store during the day. Everyone seems to have a preconception that it would be dangerous and that customers would have to try to avoid collisions with the robots. Surprisingly, these robots avoid people easily, and vice versa. So, competing retailers went to the Daiei store to see what was happening. They were like, “Oh, it’s okay.” You can see that people and robots can really coexist. So they said, “Well, let’s try this at our stores.” Over the past year, the number of companies that have adopted this robot has increased tenfold. Question What are your future plans for the Whiz in particular or robots in general? Could you describe new marketing efforts you are planning in relation to the Whiz? Ms. Totani (Daiei) I think our company needs robots. This is a major direction for us as a company. We are still in the early stages of using the Whiz to monetize manufacturer-sponsored promotions. So, we are trying to get this part of the business on a favorable trajectory. Currently, we are only doing this in a couple of regions; It should be done on a company-wide basis and we would like to involve as many stores as we can. Our hope is to utilize the robot to generate revenue and improve sales. However, there are some stores where we cannot use the robot, such as stores with narrow aisles or stores with uneven floors or lots of steps. It would be really great if the robots could traverse steps. The existing model struggles to detect even slight differences in surface level. I believe that this is something that needs to be addressed. We are currently working on in-house standards for promotional materials and are including various other companies in the discussion. We have received some feedback and results from trials, such as “customers are not looking at this material,” or “this material is not being looked at in this location.” During this process, we found that it is better to display promotions at the entrance. Our company has an app, and until now, we have displayed notices in the stores explaining how customers can find information about special offers via the official app. The results revealed that it is definitely better to display this information at the entrance. We would like to use the robot to give visitors information when they enter the store. I think using the robot to display signage is the most effective method. We had a store where the Whiz rolled around carrying a product. We also received information that other companies are doing something similar. However, in reality, the Whiz is somewhat limited in what it can carry and transport. It can only carry around 10 lightweight products. Right now, we are displaying signage and running a 15-second commercial on a loop. Also, if you go to a different sales floor where another Whiz is running, the same content will be played again. It is easy for customers to understand but may not be engaging for customers who already have an interest in the product. Instead, the robot could automatically play a video when it senses someone nearby. There are also a variety of things that we want to try to promote such as special offers on certain days or offers for different types of members. However, arranging the content has proven extremely difficult. Mr. Ikeda (SoftBank Robotics) We could also stop and start the content at a certain time or use “beacon” technology and make use of triggers that run the content when the beacon is touched. It may also be possible to link customers’ smartphones and show different kinds of commercials to those who have installed the Daiei app. There are still some areas that are not technically stable, so we have not yet ventured into these areas yet. Question What do you think will be the future role of robots in retail settings? Will they be replacement for humans? Ms. Totani (Daiei) This is a difficult question. I do not see robots as being a replacement for humans in any way. Certainly not. Nowadays, robots are becoming increasingly automated and self-reliant. However, when I think about how to make the customer experience quicker and more comfortable, how to make the checkout process smoother, and how to make the shopping experience more pleasant, I think people are superior to robots in this regard. Let me also give an example of a situation in which robots may be preferred to people. When retailers place orders many things need to be considered. Customer demand varies depending on a number of factors such as the weather and the day of the week. Location is another factor. Until now, these activities have been conducted based on human intuition and accumulated experience. The results haven’t been particularly great. The difference in the skill level of each person’s ability has made a big difference in the number of products sold. In order to address this concern, we are currently implementing an AI system for automatic ordering. Question Is there anything else that you would like to tell us about these robots? Ms. Totani (Daiei) We will only utilize these robots if they are mutually beneficial and if the advantages outweigh the disadvantages. I think this is why we have not deployed the robot for promotion beyond the Kinki and Kanto regions. A few risks still remain. At stores, we have been employing the Whiz for product promotion through playing manufacturer’s commercials. Daiei has been approaching various beverage producers that are keen on this project. There is definitely huge potential and we have been negotiating with manufacturers regarding ways to deploy signage for sales promotions and in conjunction with product placement. We believe many manufacturers will see the benefits of this type of robot-based promotion. Ideally, both our company and our manufacturers are satisfied. Products that don’t sell could start to accumulate, so with new products coming in each week it could be a challenge to make sure everything works. We need to plan for the long term and look a little further into the future. One of the things I think we need to do is to ensure that people will see the robot’s benefits. As a new initiative it is difficult for the robot to be accepted unless it produces benefits and reduces risks. Even within the same company, it could be difficult to get people to understand the robot. Future research directions Our hope is that this interview provides a useful portrayal of how robots, such as the Whiz, are currently being deployed in retail settings. As illustrated via this interview, retailing’s Robot Revolution is in its early days. However, the pace of technological improvement in this domain is quite rapid, and even at the current stage of development, retailers can choose from an array of different robots that vary in terms of size, shape and functionality (Huang & Rust, 2021; Repko, 2021). Thus, the Whiz is just one example of a wider robot population. In fact, SoftBank Robotics also markets other robots, such as the Pepper (a humanoid robot designed for conversation and entertainment) (https://www.softbankrobotics.com/corp/robots/). Thus, our inquiry represents only a partial look at a considerably broader (and rapidly evolving) domain and may not be representative of how firms, employees, and/or customers might interact and respond to other types of robots. This subtle but important distinction leads to a number of interesting avenues for future research. First of all, in contrast to humanoid robots such as the Pepper, the Whiz bears little resemblance to a human (or any other type of living being). In fact, the Whiz looks essentially like a Roomba on steroids. This decidedly non-human-like appearance may be one reason why retailers such as Daiei try to give this robot a more familiar appearance such as decorating it with a face. Prior research suggests that humans are anthropomorphic in nature and may respond more favorably (up to a point) to robots that have a more human-like appearance (Blut et al., 2021; Letheren et al., 2021). For example, Letheren et al. (2021) show that consumers prefer domestic service robots that appear more humanlike in nature. On the other hand, robots that appear too human-like could represent a threat and may be negatively perceived due to speciesism (Schmitt, 2020). Thus, the utility and value of human-like robot such as the Pepper vs. a non-human-like robot such as the Whiz is an intriguing question. In addition to not looking like a human, the Whiz is also not capable of directly interacting with humans. In fact, from the Whiz’s perspective, humans are simply obstacles that prevent it from performing its duties. Despite its disinterest in (and lack of interaction with) the human species, both employees and customers appear to display considerable fondness towards the Whiz. An examination of social media postings reveal that many shoppers endearingly regard the Whiz as a mascot or a pet. For example, some postings refer to the Whiz as Mokkun (i.e., Daiei’s official mascot).2 Thus, even though humans are not directly interacting with the Whiz, they seem quite intrigued by its presence. This observation leads to some interesting questions. For example, how does the ability (or inability) to interact with a robot shape employee and/or customer perceptions of these devices? Likewise, can a robot alter customer behavior (e.g., time and/or money spent in a store) even if it doesn’t interact with a customer? Prior research on the mere presence effect suggests that the simple presence of another individual may enhance arousal and increase drive (Guerin, 1986). Thus, might it be possible that the mere presence of a robot such as the Whiz has a similar effect on in-store shoppers? Finally, our interview focused on how humans react to robots. This focus makes sense since robots such as the Whiz are largely designed to serve humans. However, as technology in this domain improves, robots are likely to gain enhanced agency and greater autonomy (Repko, 2021; Roggeveen & Sethuraman, 2020). Thus, future research may wish to examine how robots react to humans. This issue will likely take on increased importance as robots expand their role in retail settings. While most humans are likely to treat robots in a respectful manner, some (especially children) may engage in robot abuse (Keijsers et al., 2022; Yamada et al., 2020). To prepare for our interview, we observed several hours of in-store video footage of the Whiz in operation. While this footage didn’t uncover any evidence of robot abuse, it revealed that some adult shoppers appear to intentionally stand in front of the Whiz in order to impede its progress and groups of children sometimes spontaneously gather and surround the robot in a gang-like manner. In effect, some shoppers appear to bully the Whiz. In response, the robot remains still and waits for these shoppers to move on (which they usually do). At present, this type of behavior is unlikely to cause much harm (other than waylaying the robot from its task). However, if robot abuse increases, it is possible that robots may be programmed (or learn) to adopt a more active response, such as asking a bully to move away or reporting him/her to store management. The degree to which these types of robot responses occur and the effect they may have upon store operations, customer experience, and robot welfare are intriguing questions for the future. Exhibit 2 (Whiz deployed at Daiei) Conclusion The rapid development of new technologies such as AI and robotics has captured the attention of both marketing scholars and practitioners. As noted by Fournier and Mick (1999), new technologies typically entail a number of paradoxes and often force humans to adopt new practices, which may be both good and bad. Thus, it is not surprising that the coming Robot Revolution is attracting anticipation as well as trepidation. While it is difficult to predict the future, our early look at Daiei’s deployment of the Whiz suggests that the future of robots in retail may be more satisfying than terrifying. For example, our interview reveals that the presence of the Whiz engenders positive affect from both employees and customers and that it supplements, rather than replaces, humans. In addition, initial indications suggest that the use of the Whiz as a promotional device (as shown in Exhibit 2) is generating positive returns for Daiei and presents an unexpected use case for SoftBank Robotics/Iris Robotics. Indeed, the improvisation of the Whiz from its intended use as a cleaning device to its new use as a promotional tool is surprising and suggests that robots may be used by retailers in ways that both practitioners and scholars may not predict. This uncertainty will likely make the development of appropriate theories and concepts particularly challenging. Thus, we encourage future scholars to follow our lead by observing and conversing with robot manufacturers, users, employees, customers and perhaps even the robots themselves. In particular, inductive-based approaches such as grounded theory (Goulding, 2005) and theories-in-use (Zeithaml et al., 2020) may be particularly valuable. We also encourage scholars and practitioners interested in this domain to read the prescient commentary to our interview offered by Abhijit Guha and Dhruv Grewal (Guha & Grewal, 2022, in this issue), which directly follows our article. In closing, we hope that this look at rolling out the Whiz provides some early and valuable insights into retail’s coming Robot Revolution. 1 The following video shows how Whiz operates in one of Daiei’s stores (https://www.youtube.com/watch?v=XfHYOEceERE). 2 For details of this mascot, see: https://www.daiei.co.jp/mokkun/. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Blut M Wang C Wünderlich NV Brock C Understanding anthropomorphism in service provision: A meta-analysis of physical robots, chatbots, and other AI Journal of the Academy of Marketing Science 2021 49 635 658 10.1007/s11747-020-00762-y Brain Corp (2021). Retailers fast-tracking robotic technologies in the wake of COVID-19. PR Newswire, April 6, https://www.prnewswire.com/news-releases/survey-retailers-fast-tracking-robotic-automation-technologies-in-the-wake-of-covid-19-301262459.html Coherent Market Insights (2021). Retail robots market analysis, Coherent Market Insights Corporate Homepage, July, https://www.coherentmarketinsights.com/market-insight/retail-robots-market-4535 Daiei (2022). Daiei company description, Daiei Corporate Homepage, https://www.daiei.co.jp/corporate/company/about_us.html Fournier S Mick DG Rediscovering satisfaction Journal of Marketing 1999 63 4 5 23 10.1177/002224299906300403 Goulding C Grounded theory, ethnography and phenomenology: A comparative analysis of three qualitative strategies for marketing research European journal of Marketing 2005 39 3/4 294 308 10.1108/03090560510581782 Guha, A., & Grewal, D. (2022). How robots will affect the future of retailing. AMS Review, 12 (3–4), 10.1007/s13162-022-00241-3 Grewal D Gauri DK Roggeveen AL Sethuraman R Strategizing retailing in the new technology era Journal of Retailing 2021 97 1 6 12 10.1016/j.jretai.2021.02.004 Guerin B Mere presence effects in humans: A review Journal of Experimental Social Psychology 1986 22 1 38 77 10.1016/0022-1031(86)90040-5 Huang MH Rust RT Artificial intelligence in service Journal of Service Research 2018 21 2 155 172 10.1177/1094670517752459 Huang MH Rust RT Engaged to a robot? The role of AI in service Journal of Service Research 2021 24 1 30 41 10.1177/1094670520902266 Iris Ohyama (2022a). Whiz – Iris Edition, Iris Ohyama Corporate Homepage, https://www.irisohyama.co.jp/b2b/robotics/products/whiz-i/ Iris Ohyama (2022b). Whiz – Iris Edition is now available at Taiwan, Iris Ohyama Corporate Homepage, March 22, https://www.irisohyama.co.jp/news/2022/?date=0322 Keijsers M Bartneck C Eyssel F Pay them no mind: The influence of implicit and explicit robot mind perception on the right to be protected International Journal of Social Robotics 2022 14 2 499 514 10.1007/s12369-021-00799-1 Letheren K Jetten J Roberts J Donovan J Robots should be seen and not heard… sometimes: Anthropomorphism and AI service robot interactions Psychology & Marketing 2021 38 12 2393 2406 10.1002/mar.21575 Loten, A. (2021). Saks e-commerce business pairs robots with workers,Wall Street Journal, October 26, https://www.wsj.com/articles/saks-e-commerce-business-pairs-robots-with-workers-11635292990 Noble SM Mende M Grewal D Parasuraman A The fifth industrial revolution: How harmonious human–machine collaboration is triggering a retail and service [r]evolution Journal of Retailing 2022 98 2 199 208 10.1016/j.jretai.2022.04.003 Repko, M. (2021). The robot work force isn’t coming - It’s already here, CNBC, December 27, https://www.cnbc.com/2021/12/27/the-robot-work-force-isnt-coming-its-already-here-.html Retail AI Research Society (2021). Retail AI award 2021, Retail AI Research Society Homepage, https://retail-ai.or.jp/award2021 Roggeveen AL Sethuraman R Customer-interfacing retail technologies in 2020 & Beyond: An integrative framework and research directions Journal of Retailing 2020 96 3 299 309 10.1016/j.jretai.2020.08.001 Schmitt B Speciesism: Sn obstacle to AI and robot adoption Marketing Letters 2020 31 1 3 6 10.1007/s11002-019-09499-3 Shankar V Kalyanam K Setia P Golmohammadi A Tirunillai S Douglass T Hennessey J Bull JS Waddoups R How technology is changing retail Journal of Retailing 2021 97 1 13 27 10.1016/j.jretai.2020.10.006 SoftBank Robotics. (2020). Worldwide sales of SoftBank Robotics Group’s AI-enabled vacuum cleaning robot ‘Whiz’ surpasses 10,000 mark,SoftBank Robotics Corporate Homepage, https://www.softbankrobotics.com/corp/pressroom/20200630a/ Subero-Navarro Á Pelegrín-Borondo J Reinares-Lara E Olarte-Pascual C Proposal for modeling social robot acceptance by retail customers: CAN model + technophobia Journal of Retailing and Consumer Services 2022 64 1 9 10.1016/j.jretconser.2021.102813 Yamada, S., Kanda, T., & Tomita, K. (2020). An escalating model of children’s robot abuse. 15th ACM/IEEE International Conference on Human-Robot Interaction (HRI), 191–199. Zeithaml VA Jaworski BJ Kohli AK Tuli KR Ulaga W Zaltman G A theories-in-use approach to building marketing theory Journal of Marketing 2020 84 1 32 51 10.1177/0022242919888477	0	PMC9716508	NO-CC CODE	2022-12-03 23:20:57	no	AMS Rev. 2022 Dec 2;:1-7	utf-8	null	null	null	oa_other
==== Front Soc Netw Anal Min Soc Netw Anal Min Social Network Analysis and Mining 1869-5450 1869-5469 Springer Vienna Vienna 1006 10.1007/s13278-022-01006-3 Original Article Delivery structure of nationalism message on Twitter in the context of Indonesian netizens Sari Dewi Kartika [email protected] 12 Kumorotomo Wahyudi 3 Kurnia Novi 4 1 grid.8570.a 0000 0001 2152 4506 Doctor in Communication Science, Universitas Gadjah Mada, Yogyakarta, Indonesia 2 grid.444224.0 0000 0001 0742 4402 Department of Communication Science, Universitas Kristen Satya Wacana, Salatiga, Indonesia 3 grid.8570.a 0000 0001 2152 4506 Department of Public Policy and Management, Universitas Gadjah Mada, Yogyakarta, Indonesia 4 grid.8570.a 0000 0001 2152 4506 Department of Communication Science, Universitas Gadjah Mada, Yogyakarta, Indonesia 2 12 2022 2022 12 1 17322 10 2022 17 11 2022 19 11 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Austria, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Nationalism has emerged to be discussed in this modern era due to the emergence of a globalized society. Its delivery structure of nationalism message is significant to be investigated due to its impact on the self-determination of a nation so that nationalism spreads among civil society. In the Indonesian context, discussions of Indonesia’s Independence Day arise as a response to this self-determination. Its discussion came up in a face-to-face conversation, as well as on Twitter through the #HUTRI76 network. As a microblogging platform, Twitter was able to facilitate discussion of Indonesian in articulating their Independence Day. The concept of nationalism was utilized to sharpen the analysis. Hence, the social network analysis method has been applied to explore various metrics in the #HUTRI76 Twitter network. The results showed some influencers who varied in their profession, emerged and played a role in sparking discourse on Indonesia’s Independence Day. The discussion of Indonesian Independence Day on Twitter was a reproduction message of the government. Moreover, users engaged the varied media to cross-posting their messages as regards Indonesia’s Independence Day, which means Twitter acts as a hub through hyperlinks embedded in tweets. Keywords Indonesia Independence day Nationalism Netizens Social network analysis Universitas Kristen Satya Wacanaissue-copyright-statement© Springer-Verlag GmbH Austria, part of Springer Nature 2022 ==== Body pmcIntroduction Even though the golden era of nationalism ideology is predicted to end at the end of the twentieth century, unexpectedly, this ideology has returned to attention and emerged as a massive movement in many countries. Events such as the Russia-Ukraine war (Breuilly and Halikiopoulou 2022; Knott 2022; Kuzio 2022), the shooting of Shinzo Abe in Japan (Duffy 2022; Magcamit 2019; Wingfield-Hayes 2022), Brexit in Britain (Brown 2017; Gusterson 2017; Mcewen 2022; Roth 2018; Stephens 2019), the election of Donald Trump in the USA (Bieber 2018; Giroux 2017; Restad 2020; Schertzer and Woods 2020), and the election of nationalist leaders in many parts of the world have aroused civil society movements in these countries to express their nationalism openly to the public. Specifically, in Indonesia, the election of Joko Widodo, a figure who is considered to represent the nationalist orientations (Madu, 2017) and nationalist economist ideology (Gede Wahyu Wicaksana 2019), as president for two terms (2014–2024) and the victory of PDIP, which is a nationalist party, for two terms (2014–2024) in the legislative elections, shows the beginning of the new era of nationalism rising (Ramdhani and Anggraini 2021). In recent times, the existence of nationalism ideology is constantly increasing and being felt in many countries. Even Bieber said that nationalism acts like air, it is omnipresent and its behavior is abstruse (Bieber 2018). Various moments are considered to trigger the emergence of the nationalist movement (Merriman 2020; Paasi 2016; Visoka 2020; Z. Wang 2021; Żuk and Żuk 2022). One of the important moments to be discussed in Indonesia around nationalism is Indonesia’s Independence Day. Independence Day can be seen as an assertion of a nation-state or as a sustaining of collective consciousness, reminiscent of memories that created nation-building (Akuupa and Kornes 2015), and constitute collective memory (Kook 2006), or pay tribute to the beginnings of independence (Nyyssönen 2009). As Independence Day is important and deserves scholarly consideration, it was set in a longer historical time frame and broader geographical perspective. Indeed, Independence Day has more variation in practice than is generally recognized (Cannadine 2008), revealing cultural analysis (Żuk and Żuk 2022), or as a cosmopolitan moment (Webber 2005). Indonesia declared its Independence Day on August 17th, 1945 by its founding father, Soekarno and Muhammad Hatta. Since then, Indonesia’s independence has been celebrated dynamically, following the social, political, and economic developments that surround it (Adams 1997; Agus et al. 2020; Hatley 1982; Tanasaldy and Palmer 2019). Television, newspapers, and radio were the mass media that made news of the celebration of Indonesian independence at that time (Hatley 2012; Hellman 2007; Wild 2007). Anderson (1983, 2006), a political scientist and historian, noticed that newspapers had become print capitalism, which was employed by elite groups to instill a sense of nationalism in society. Nowadays, in the era of digitalization, nationalism changed its shape (Budnitskiy 2018; Mihelj and Jiménez-Martínez 2021), in which social media is considered to be one of the contributing factors (Mihelj and Jiménez-Martínez 2021). Despite the era of globalization being considered to threaten the existence of this ideology, globalization through social media has spread the spirit of nationalism from one country to another as well. With 68.9 percent of social media users at the beginning of 2022 (Kemp 2022), Indonesian nationalism in the current digital era continues to experience dynamics through social media. Social media is closely related to relationships and activities of daily life. Social media increasingly makes bonds of offline contexts constantly online, and, due to the ambient nature of this technology, awareness of the opinions, interests, and activities of social ties has spread (Hall 2016; Pulido et al. 2018). It is very important to know the fashion of the nationalism movement that occurs on social media. For this purpose, analytical data from appropriate social media platforms are required. The most popular social media used for discussing a topic is Twitter. This platform has a wide audience reach (Lipschultz et al. 2022) and can specifically grasp those who are interested in a particular topic through hashtags (#) (Rauschnabel et al. 2019). This platform allows direct two-way communication so that feedback on a topic can be attained. In addition, Twitter accounts are free so they do not restrict users from joining conversations related to certain topics (Morgan-Lopez et al. 2017). Because of its characteristics that may provide current information, the social media platform Twitter was chosen (Hasan et al. 2018; Stromer-Galley 2019). Twitter is conveniently accessible and is capable of connecting users through quick and frequent comments and posts as well. Users can employ likes, retweets, and views to maximize their publicity (Awan 2017). Twitter, particularly, has been developed to support interactive multitasking. The multimedia information that is included in tweets is referred to as the interactive component (including hyperlinks, photographs, and videos). These tweets are not inherently consumed passively by the recipients. Instead, people may actively engage with this information or they can blur the lines and start creating their content if they leave comments on the original content or reply to the original tweeter (Murthy 2013; Steinert-Threlkeld 2018; Tong and Zuo 2021). Numerous research on the use of Twitter as a communication channel has been used in the Indonesian context (Santoso 2021; Sari et al. 2021), but significant research related to the use of Twitter on Independence Day, particularly in Indonesia, has not been conducted. Therefore, this study aims to investigate the delivery structure of nationalism messages on Twitter in the context of Indonesian netizens. The comprehensive picture of how Twitter is used will extend to the analysis of this research. Basic graph theory Fuhse (2022, 2009) stated that humans live in a connected environment through dyadic relationships, symbols, schemes, scripts, and communication which creates social networks, which basically consist of two structures: vertex (associated with node) and edge (referred to as tie or connection) (Yang et al. 2020). Thus, this research focuses on two-term on basic graph theories, namely vertices and edges. Investigating social network analysis, in which vertex is a social entity and edge is a relation embedded in them, will guide us in understanding the existence of virtual communities (Jan 2019). The dynamics of social relations (Bidart et al. 2020), social movements (Diani 2014), and cultural networks (DiMaggio 2014) are several studies that show the importance of studying social entities and their relations within the framework of social network analysis. Vertices, the first structure of the network, are also known as vertex, nodes, items, agents, actors, or entities. It serves as many things: people, social structures (including work groups, organizations, institutions, teams, or even countries), keyword tags, videos, web pages, or represents physical events, locations, or events as well. Vertices often correspond to posts or authors on blogs and friends on social networking sites (Hansen et al. 2011a, b). These vertices data have attributes that describe each node. Attribute data, however, can generally describe a user’s demographic characteristics (such as gender, race, and age), a system used by users (such as messages posted, number of logins, and edits made), or other characteristics such as location or income (Hansen et al., 2011a, b). The second structure of the network is the edges. Edges are associated with links, connections, ties, and relationships. It connects two vertices, which represent different types of relationships like proximity, kinship, collaboration, trade partnership, friendship, literature citation, investment, hyperlink, transaction, or any shared attribute. In short, an edge is any form of relationship or connection between two entities (Hansen et al. 2020a). Moreover, edge analysis can be classified into two types: directed and undirected. A directed edge (also known as an asymmetrical edge) has a clear origin and destination, which is represented on the graph as a line with arrows pointing from the source point to the receiving point and can be reciprocated or not. Undirected edges (also known as symmetric or reciprocal edges), on the other hand, exist only between two people or things. There is no clear origin or destination in this reciprocal relationship. A line connecting two vertices without arrows represents an undirected edge in a graph (Hansen et al. 2020b). As the use of digital media has increased and connected, Wellman (Wellman 2001) argued that networks cannot be studied in isolation because they can connect people, organizations, and knowledge (Castillo-De Mesa and Gómez-Jacinto 2020; Clark et al. 2017; Hendrickson et al. 2011; White et al. 2022). Visualized networks could help us understand the social world that generates benefits by providing many insights, which would not be possible if we used other approaches (Borgatti et al. 2018; Hansen et al. 2011a, b; Iglesias and Moreno 2020). According to Smith et al. (2009), previously unknown detailed data on social processes can now be understood through network visualization. Social researchers will also be supported when network visualization can be used to explore areas that have not previously been studied in their studies (D. L. Hansen et al. 2020a; Moody and Light 2020). Researchers, professionals, and governments employ network visualization for a variety of purposes. Network visualization can be used to identify influencers (Sanawi et al. 2017) and describe a network of media agendas and explain the narratives that form on Twitter (Guo et al. 2017). In a crisis, it can be important communication to facilitate awareness and assist the decision-making process (Drosio and Stanek 2016; Karakosta et al. 2021; Onorati and Díaz 2015). Additionally, network visualization can be used to map relationships based on the user’s profession, topics discussed, and their relationship to the mass media (Abdelsadek et al. 2018; Ausserhofer and Maireder 2013; Williamson and Ruming 2016). The effectiveness of risk communication on social media can be mapped with network visualization so that it is feasible for public health officials and agencies in planning, monitoring, or evaluating public discussions on Twitter (Pascual-Ferrá et al. 2022). Furthermore, it could describe the domain of news media and are a valuable resource for newsrooms seeking to gain insight into news events emerging from a stream of social media posts as well (Ahmed and Lugovic 2019). Network analysis metrics Social network measurement was focused on the number of simple connections. The measurement is more sophisticated over time as it develops and incorporates various variables such as network density, centrality, vertex and edge count, and bridge level. These metrics measurements integrate the complex network variables. The first mentioned metrics evaluate the characteristics of the entire network (network-level), while the rest evaluate the behavior of the individual node (vertex-level). The density of the network captures the connectivity between the nodes by calculating the percentage of observed connections from the maximum possible if one node is connected to others. The centrality is the measure of the importance level of a node in the network based on some objective criteria in which some central nodes are located at the edge or periphery while others are in the middle of the network. It can be calculated in many ways, such as degree centrality, closeness centrality, betweenness centrality, and eigenvector centrality. In this work, the centrality calculation is restricted using degree and betweenness. The vertex count is the number of nodes in the network while the edge count is the number of connections between the nodes. However, the fifth metric, a bridge, occurs when the third node bridges the connection (“friend of a friend”). It is acting as an intermediary or a connector. When the node is missing, a gap or structural hole is formed in the network. Thus, the “bridge” presence is important to fill in this hole (Golbeck 2015; Hansen et al. 2020b). Degree centrality describes the social connections (edges) quantity owned by a node in the network. This parameter quantifies the significance and effect of the individual user in the established network environment. Its value is equal to the number of social connections owned by a user. If a node has x number of connections, thus its degree of centrality is also x. For example, a node with an 8-degree centrality value shows eight connections it has. The more central a node, thus its degree of centrality value is higher, and vice versa. Zero-degree centrality of a node indicates its independency behavior. As the first vertex-level analysis, degree centrality can be divided into two terms, i.e., in-degree and out-degree centrality. The in-degree centrality counts the connection number directed to the node (pointed-in), while the out-degree counts the connection number directed from the node (pointed-out) as illustrated in Fig. 1.Fig. 1 Illustration of in-degree and out-degree centrality of a central node around the adjacent nodes Conceptualizing nationalism Discussions on nationalism have been continuing for a long time, even though they were not initially conducted in a systematic procedure. Before the second half of the twentieth century, articles on nationalism began to be published in political science journals, which were usually case studies, written by historians, sociologists, or psychologists (Mylonas and Tudor, 2021). Later, over 1983, three significant studies on nationalism were released simultaneously. Benedict Anderson published Imagined Communities (Anderson 1983), The Invention of Tradition by Hobsbawm and Ranger (1983), and Nations and Nationalism by Ernest Gellner (1983, 2009). These scholars viewed the nation and nationalism as contemporary phenomena evolved of urbanization, industrialization, print capitalism, and anti-colonial struggle (Mylonas and Tudor 2021). Ernest Gellner (2006), a leading scholar on nationalism, describes nationalism simply as mainly a political theory that argues that the political and national units should be coherent. It is also consistent with both natural and constructed methods of national identity formation because it defines nationalism along with its impact rather than its cause. Nonetheless, Gellner (2006) principal focus in Nations and Nationalism is on the causative relationship and connective development between industrialization and nationalism. Gellner puts nationalism as based on power, education, and shared culture, and as such. It is dynamic like its constituent elements and not a “permanent aspect of the human condition” (Hajjaj 2020). Anderson (1983) would subsequently identify the production and diffusion of language, particularly written language, as a precondition for a nationalistic sensibility, in many aspects adding to Gellner's concept. While Gellner asserts that cultural homogeneity is a necessity for industrial society and hence the production of nationalism, Anderson (1983) emphasizes the organic emergence of nationalism through the print media. Eric Hobsbawm and Terence Ranger, on the other hand, introduce the phrase “invented tradition” as how many long-held traditions have been reinvented in the recent past. Additionally, Hobsbawm goes on to characterize manufactured traditions as a reaction to new conditions that attempt to portray at least some aspects of social life throughout the modern world as permanent and invariant. Hobsbawm then distinguishes between “tradition,” “custom,” and “convention.” Contrary to tradition, the custom is not regarded to be immutable and must remain adaptable. It is a well-established procedure that can be simply modified if required. The convention is merely routine in the absence of any ceremonial or symbolic function. Both convention and custom are practices that make life easier, but none has an intellectual foundation like tradition (Singenberger 2021). Traditionally, nationalism has been divided into ethnic and civic variants. As Kohn, a historian and politician scientist posited that civic nationalism is founded on citizenship and individuals’ capacity to join the nation. However, ethnic nationalism is rooted in the notion of shared ancestry and hence is less inclusive (Kohn 1944). While this idea was previously valuable, it yet assists, even though on a minor scale (Bieber 2018). In advancement, in its development, Kohn’s distinction of nationalism has been criticized. Civic/ethnic distinction influences present-moment politics. First, the dichotomy presupposes that ethnic conflicts are indeed encountered in the East, encouraging us to dismiss the expansion of racial and ethnic tensions inside purported civic Western democracies. Western democracies profess to be more peaceful and inclusive than they are by flying the civic flag, developing a self-image that allows them to exonerate themselves, leaving them unable to cope with internal issues (Tamir 2019). Second, the civic/ethnic distinction promotes the notion that in civic countries, culture, religion, language, ethnicity, and race perform minor roles and may indeed be dismissed. This presumption compels us to reconsider the relationship between politics and identity. Members of minority groups pushed the majority to admit that identity influences not only who we are but also what we get during the initial stage of this debate. In light of the emergence of identity politics, identity concerns could not be ignored, resulting in the age of multiculturalism (Tamir 2019). In addition, perceiving nationalism through Kohn’s perspectives can also be observed from the modes throughout which nationalism operates. Nationalism, as per Benedict Anderson, is a mode of representation. The nation refers to the imagined community made possible by the forces of representation liberated by printing press technology (Squibb 2016). Meanwhile, Ernest Gellner considers nationalism as a mode of reproduction in a different sense, arguing that it is required for industrial production (Squibb 2016). Entering the digital age, nationalism nowadays interacts in complex ways with technological information and communication networks, frequently resulting in unexpected outcomes (Schneider 2022). Thus, scholars need to pay attention to the role information and communication technologies play in a discussion of nationalism. As Schneider argued, ‘digital nationalism’ is an emergent characteristic of communication networks and a result of technical affordances, economic incentives, and policy decisions (Schneider 2018). Furthermore, critical issues that emerge in studying digital nationalism are the investigation that requires conceptual and empirical work if we need a better understanding of how interactions between humans and technology produce and shape nationalism in advanced information societies. Therefore, the purpose of this research is to provide empirical work on these concerns. Conceptualizing netizens Performing research on nationalism in its digital era will investigate netizens as well. As MacKinnon argued, netizens are defined as those who digitally ‘inhabit’ and rely on digital platforms, networks, and services (MacKinnon 2012). They have the option of being passive ‘users,’ accepting their lot as given by governments and corporations which claim to know what is best for them. Alternatively, they can choose to express themselves as “netizens” and work actively to ensure that the Internet grows following their rights and interests (MacKinnon 2012). Extensively, there are three vectors of power in cyberspace: government, enterprises, and what can be called “the netizenry,” “digital civil society,” or “citizen commons.” The netizenry’s power must be enhanced to the point where it can successfully challenge the power of government and corporations (MacKinnon 2012). The importance of the existence of netizens has occasionally received prominent recognition (Hauben 2019). Netizens’ demand for popular science (L. Wang and Zhong 2019), netizens’ participation to support the government in times of crisis (J. Guo et al., 2018), netizens’ digital activism in Zimbabwe (Nyoka and Tembo 2022), and netizens engagement patterns (Hasfi et al. 2021; Seman et al. 2019) are several studies that have been conducted to examine netizens’ influence in each of the nations researched. Conceptualizing hashtag Over the last ten years, the popularity of social networks has also been a success for the keyword. Since the introduction of the “hashtag” by Twitter and Instagram in 2007 and 2010, respectively, a method of categorizing remarks and documents that were previously exclusive to highly specialized professional groups has marked prevalent media use (Bernard 2019). Today, every Twitter feed and Instagram post contribute to the community indexing or “keywording” of the world, which can be performed as a creative act by every user of these social networks, unconstrained by predefined rules or hierarchically tiered access procedures. Documents could only be linked to one another in the initial periods of the “World Wide Web,” as is widely recognized. In many aspects, the shift from “link” to “hashtag” as a defining networking notion represented a significant transition in the digital structuring of statements. First, it indicated that every Internet user could initiate links independently and without requiring any programming skills, paving the way for the Internet’s more even “social” and fostering a participatory era. Second, it signified that for the first time, a mechanism for establishing connections was supplied with its typographic element. The prefixed symbol “#”—referred to as a “hash” in British English or a “pound sign” in American English—converts words into networked keywords. Thus, the hashtag and the letters immediately following it have two purposes: They are both a component of visible tweets or Instagram posts, as well as a trigger for the invisible mechanism that connects them (Bernard 2019). As Rocca argued, hashtags were a cultural product. Additionally, if we consider hashtags to be cultural products, altering their meaning throughout human action, we should ask ourselves what we can learn from them (Rocca 2020). When analyzing data retrieved from social media and indexed by hashtags, we must acknowledge the significance of the hashtag’s social construction and the meanings that exist underneath it. Whether we apply them for audience analysis (Fisher and Mehozay 2019; Gallagher et al. 2019; Maireder et al. 2015) or cataclysmic events (Lovari and Bowen 2020; Mirbabaie et al. 2021; Yan and Pedraza-Martinez 2019), we must consider that they are formed with a certain sense attribution, but that it alters as a result of contact, developing the various selves that hashtag represents. Methods As part of content analysis, a quantitative approach was used. Data were collected from Twitter using NodeXL Professional for Academics (NodeXL Pro). NodeXL was chosen because it offers a wide range of options for optimizing Twitter network analysis. As it runs on Microsoft Excel, many types of statistics can be used. Furthermore, NodeXL provides a friendly way to browse the Twitter network with the automated functions offered (Yu and Muñoz-Justicia 2020). The time selected was August 18th, 2021. This time was chosen because of the high level of discussion on Indonesia’s Independence Day. The data were taken one day after Indonesia’s Independence Day with the intention it can capture all conversations on Independence Day. To collect data from Twitter related to Indonesia’s Independence Day, the #HUTRI76 keyword was used. This work on Twitter has focused on conversations coordinated by hashtags. The hashtag is a keyword or short abbreviation, beginning with the ‘#’ symbol and indicates a special discursive marker (Bruns and Stieglitz 2013; Small 2011). Moreover, we use hashtags because of their effective mechanism to capture all Twitter features (including tweets, replies, or retweets) (Bruns 2011). On Indonesian Independence Day, #HUTRI76 emerged in Indonesia’s Twitter trending topics. We chose Independence Day because it is a symbol of nationalism (Paasi 2016), a form of an exhibition of nationalism (Agastia 2017), as well as a symbolic and ritualistic form of nationalism (Hyttinen and Näre 2017). Afterward, the data were collected, and then, they were analyzed and visualized. The analysis was carried out in three parts. First, analysis of the network structure to find influential users by evaluating the value of several parameters, viz. in-degree, out-degree, betweenness centrality, and eigenvector centrality. The second analysis was conducted to examine the message structure on the #HUTRI76 network by detecting the top twenty words and words pair on the #HUTRI76 network. The third analysis was conducted to figure out the media used by users in delivering the messages of Indonesia’s Independence Day. As these three analyses were executed, the delivery structure of nationalism messages through Twitter in the Indonesian context emerged. Result and discussion The network structure of influential users on #HUTRI76 Analysis was set up by admitting the metric of the #HUTRI76 network. The network metric formed among Indonesian digital citizens on the issue of Indonesia’s Independence Day is presented in Table 1. In this network, 14,686 accounts/users joined the conversation and built 20,163 connections with the possibility of forming a connection of only 0.0088%. Moreover, each account needs 6.85 of average path length to reach another account of the network to share information, which reveals the efficiency of information transfer in the network.Table 1 Network metric of #HUTRI76 network on Twitter Parameters: Nodes Edges Density Average path length Count: 14,686 20,163 8.8 × 10–5 6.85 Incoming engagement users The significance of users on the Twitter network can be determined based on the centrality of incoming and outgoing. In addition, engagements between users and their content on Twitter can be captured through in- and out-degree centrality metrics. Actual attention given to content as well as users’ actions taken to disseminate information is indicated through these things (D. L. Hansen et al. 2020c). More attention to tweets among the community of users who participate in conversations dealing with a particular topic is gained by users with a high degree of centrality. Thus, centrality at the user level captures engagement in a community. High centrality scores exist for users who are considered the center of the conversation because other users have replied to, mentioned, or retweeted their posts. Thus, in-degree is an indication of the information flow initiated by the user. In addition, users can have high login rates on one topic network, but low login rates on another network. On the other hand, the out-degree centrality metric is measured by capturing the level of user-initiated engagement with community members. Figure 2 shows the in-degree and out-degree distribution related to the issue of Indonesia’s Independence Day events on the #HUTRI76 Twitter network, while the percentage of three categories of the degree’s values is summarized in Table 2. As revealed in Fig. 2a and Table 2, the majority of users (73%) are independent users (unconnected to one another) as indicated by zero in-degree value. It shows the independency of this Twitter user (lack of influence from another user). As Jendoubi et al. (2018) point out, independent users are generally not swayed and are independent in their choices and decisions. Independent users can draw and drive another user’s point of view. Meanwhile, the minor population (0.7%) are the interesting users. Only the minor population gained more attention to their tweets among the community of users who participate in the conversation about Indonesia’s Independence Day by mentioning, replying to, or retweeting their posts. As shown in Table 2, a Twitter user is eligible to be independent when his perspective does not depend on other people’s ideas. Thus, the user’s behavior is independent of other user behaviors. Detection of independent users is crucial since some of them can become influencers. It is similar to Räbiger and Spiliopoulou (2015) study which underlines the importance of distinguishing influencers from non-influencers on Twitter. Those who are not influencers can be directly targeted because they can be persuaded. However, we propose that independent users can be considered as influencers although not all independent users are influencers. This finding conforms to the study of Bokunewicz and Shulman (2017) which finds that popularity is not the only indication of influence in a network, or refers to the study of Räbiger and Spiliopoulou (2015) which states that writing influential tweets is not a prerequisite for becoming an influential user.Fig. 2 a In-degree, and b Out-degree centrality value of #HUTRI76 Twitter network Table 2 Frequency of some category of the degree centrality value Value of degree centrality Frequency relative to the node population In-degree Out-degree 0 73.0% 6% 1–30 26.3% 94% > 30 0.70% 0.2% Outcoming engagement users As revealed in Fig. 2b and Table 2, the majority of users are active users (initiate the connection to other users and actively post tweets) as indicated by the nonzero out-degree value (Hansen et al. 2020b). The out-degree centrality value shows the user’s attention to the topic of Indonesia’s Independence Day by mentioning or replying to a tweet (Hansen et al. 2020a). Out-degree centrality, then, reflects the level of user-initiated engagement with community members within the #HUTRI76 network. Out-degree centrality graphs the outreach of a user to the community. Users who tweet a lot about a topic are represented with a high degree of centrality value. They intend to grasp the user’s attention by replying or mentioning them. A high out-degree centrality of the #HUTRI76 network, which has a value of more than 30, was found to be 0.2%. The level of engagement a user initiates with members of the community, thus, is captured by this user. This indicates that they tweet a lot about a topic by mentioning or replying, aiming to reach users’ attention. These users generate a high level of engagement with other users for content about Indonesian Independence. The user has control over other users in terms of the dissemination of information on Indonesian Independence. This is similar to the study of Aleskerov which stated that important user has a significant role in various fields (Aleskerov et al. 2021). Moreover, 6% of the user population has zero value of out-degree which means most of the users in this Twitter network have an impact on other users. This is reinforced by Carolina, Guido, Renaud, and Fabio’s research which stated that users with the highest out-degree were identified as the most popular users, while users with the highest out-degree posts were the most “viral” (Becatti et al. 2019). It was revealed that challenges with high out-degree centrality can create many other challenges (Himelboim et al. 2017; Jami Pour et al. 2022). Furthermore, we can identify that these users pay more attention to the actual information as well as the actions taken to spread information about Indonesia’s Independence Day. This character is unique and authentic to the character of Indonesian netizens. Indonesia has a high level of participation in Internet use. According to other research, the level of community participation appears in the form of participation in various voluntary activities (Irandoost et al. 2022; Niebuur et al. 2018). Predominantly, the research shows a positive relationship between social media use and participation (Boulianne 2015). On the other hand, Arwati and Latif’s article shows that Indonesian public participation in e-government implementation remains low, as a result of public knowledge of government data and information, which has not been adequately accounted for in terms of its assistance, especially to anti-corruption efforts. (Arwati and Latif 2019). Most users have a low value of both degree centrality which lies in the ranges of 0–10, thus we only show the distribution in that interval. It shows that most users in the #HUTRI76 Twitter network have scattered users. In other words, various accounts that tweet mentioning other accounts (out-degree) in the #HUTRI76 network, as well as accounts mentioned by others (in-degree), are in the high category. However, several accounts (in which ID stands for account Identity) have in-degree and out-degree scores with a higher mean than other accounts in the network. These four accounts are ID 3176, ID 970, ID 224, and ID 261. Cross-engagement users To observe the significant users that impact outside, as well as receive much popularity (double-central user), the in-degree and out-degree of each user are paired and presented in Fig. 3. Only four users, i.e., ID 3176, ID 970, ID 224, and ID 261, fulfilled the criteria of the double-central user. They not only have control and power over others but also have popularity among the adjacent users in the #HUTRI76 network on Twitter.Fig. 3 In- and out-degree centrality cross-tabulation value of #HUTRI76 Twitter network ID 3176 mentions ID 3173 to respond to the tweet ‘Mensen temen lo yang layak juara 1 lomba menahan rindu disaat PPKM (Mention your friend who deserves 1st place in the contest to holding back during PPKM)’. User ID 3176 has an in-degree score of 37 and an out-degree value of 19. The user ID 3176 betweenness centrality value is 402748,972886. Surprisingly, ID 3176 appears on two parameters, i.e., cross-engagement metric and betweenness centrality metric. This makes ID 3176 act as an influential user as well as a bridge between users in the conversation of Indonesia’s Independence Day. Bridges between users The number of connections in a network is used to determine a vertex’s degree of centrality. Because they link users who would otherwise be unconnected or insufficiently linked, users (i.e., vertex) can also play a significant role in a network. On the other hand, betweenness centrality quantifies the extent to which a vertex serves as a connecting point in a network. The degree to which the user is located on the shortest path between another user in the network is measured by betweenness centrality. A user’s betweenness centrality increases as more people rely on them to connect them with others. Thus, actors positioned in structural gaps in a network profit strategically by exercising power, getting access to innovative information, and brokering resources. Because of their structural position and associated benefits, actors who fill structural gaps are considered engaging relationship partners. These actors are referred to as brokers or bridges because they fill a brokerage position. These actors connect otherwise less connected actors through non-redundant, frequently weak ties. Actors who bridge social gaps or act as social mediators place high importance on both betweenness and in-degree. Figure 4 shows the top ten of users with the highest betweenness centrality value on the #HUTRI76 network. The user can act as a bridge in the flow of information if he/she has a high betweenness centrality value. All information is directed to be able to go through the ten users. Users can also act to reduce the flow of information and act as gatekeepers as well. Thus, the ten users in the #HUTRI76 network play an important role in the flow of information in the entire network.Fig. 4 Betweenness centrality value of #HUTRI76 Twitter network The previously mentioned ten users with the highest score of betweenness centrality consist of various community groups. The first group is regional leaders (ID 2582 and ID 1148), football club (ID 13), content creator (ID 29), study center (ID 11), ID 20, ID 11,186, freelance MV animator (ID 1603), K-Pop fans (ID 1503), and Vtuber agency (ID 6740). The diversity of user groups based on betweenness centrality reveals that micro-blogs, particularly Twitter, can create the nature of online participation. It means the audience is not only passively reading content on social media. Audiences even contribute by creating content (Faizal Kasmani et al. 2014), as well as it has the potential for the development of democracy (Schreiner 2018). In social network analysis, betweenness centrality has other implications. From a macroscopic perspective, the bridging position indicated by high betweenness centrality reflects strength. This is because it allows them to exercise control over, for example, deciding whether or not to share information with users on a network (Burt 2009). In online social networks, high betweenness centrality corresponds to relationships between the closest friends when observed from the microscopic perspective of ego networks (i.e., in light of first-degree connections). This is due to the connection reflecting the social capital invested in the relationship when distant social circles (such as family and university) are linked, which frequently happens as a result of ego recognition (Stolz and Schlereth 2021). Based on Fig. 4, it was found that 81% of the accounts had a zero-centrality value. From a macroscopic perspective, it explains the weak strength of the user interface in the network. On the one hand, users tend to decide not to share information with other users, while on the other hand, users also tend to have no control over other users in the #HUTRI76 network. In contrast, 19% of accounts have a nonzero betweenness centrality value. From a macroscopic perspective, this group of users has control over the information they have. They tend to provide information to other users on the network. Even from a microscopic perspective, this group of users has strong interpersonal bonds with other users. Thus, on the #HUTRI76 network, the emergence of various circles of society on Twitter informs several things. First, all community groups show expressions to celebrate Indonesian Independence Day. Both groups of people admitted their national identity by posting tweets related to Indonesia’s Independence Day. The third is the excitement of Twitter users on Indonesia’s Independence Day. Regional heads, politicians, football clubs, content creators, etc., participated in this event. To expand the analysis of users who have influence, the values of betweenness centrality and eigenvector centrality are paired. The eigenvector represents the magnitude of the influence or importance of nodes in a network. Figure 5 displays four users with high eigenvector values and high betweenness centrality values. They are ID 9267, ID 10,851, ID 1110, and ID 3176. These four accounts have a strong influence as a bridge in disseminating information on the network, as well as having a strong influence on other users in the #HUTRI76 network.Fig. 5 Cross-tabulation Betweenness centrality and Eigenvector Centrality #HUTRI76 Twitter network The message structure on the #HUTRI76 Figure 6 denotes the 20 most used words and word pairs on the #HUTRI76 network. Figure 6a reveals the word ‘#hutri76’ ranks first with 19,657 uses of the word, while the word ‘selamat’ is the twentieth most frequently used word on the #HUTRI76 network with 1,826 uses. Among the top twenty words, ‘hutri76,’ ‘Indonesia,’’17-an,’ ‘dirgahayu,’ ‘kemerdekaan,’ ‘republik,’ and ‘kita,’ ‘merdeka,’ and ‘selamat’ were related to discussions of nationalism. The word ‘Indonesia’ refers to the name of the country, the word ‘independence’ refers to freedom from colonialism, and the word ‘republik’ refers to the form of the Indonesian state, which is in line with Blondel (2005) and Pettit (2002) intention that republican government prioritizes the voice of the people.Fig. 6 Top word and top word pairs in the #HUTRI76 Twitter network ‘#hutri76’ is an abbreviation of ‘hari ulang tahun Republik Indonesia’ ke-76’ (the 76th anniversary of the Republic of Indonesia-in English), which refers to the public’s memory of Indonesia’s Independence Day from the Dutch which reached 76 years of age. As Malinova (2021) stated, memory is essential for each identity. Hence, we need to review memory politics, which refers to the public activities of several social individuals and organizations. It aimed at promoting certain interpretations of a collective history and establishing an appropriate sociocultural infrastructure of remembering, school curriculum, and, in some cases, special laws (Malinova 2021). We propose that regarding Kohn’s perspective, collective interpretations of the past and shared representations thus far emerge in Indonesia within the framework of ‘civic nationalism’ (Kohn 1944). The word ‘kita’ (us-in English) refers to the bond of Indonesian national identity. As this term was used by Smith (2010) which refers to the deep bond within a country that serves to strengthen the sense of belonging together, although, on the other hand, the word ‘kita’ has implications for ethnocentrism. Similarly, Knott argued that nationalism is related to the sense of belonging. Furthermore, the nation is a particularly powerful kind of belonging that produces a sense of oneness to those regarded as members of the nation and territories, as well as a feeling of distinction to those considered as outsiders, i.e., non-members. Insiders regard belonging to the nation as a personal, deep sense of belonging which creates an emotional (or even ontological) commitment as a continuous endeavor of belonging to people and places (Knott 2017). Meanwhile, in Kohn’s idea, this sense of belonging is a form of ethnic nationalism (Kohn 1944). The use of the local Indonesian word ‘dirgahayu’ is the most frequently used word as well. ‘dirgahayu’ in Kamus Besar Bahasa Indonesia means long life. This word is usually addressed to a country or organization that is celebrating its Independence Day. In the Indonesian context, the word ‘dirgahayu’ is not only implemented on Independence Day but is often used on the anniversary of the Indonesian army as well. Thus, this word is closely related to the language used by the Indonesian government at the moment of the anniversary of the state or state organizations. This choice of words also represents nationalist language ideologies (Vessey 2021). The use of the English words ‘happy’ and ‘independence’ indicates the discourse on Indonesia’s Independence Day aimed to be known at a global level. As a community uses language to communicate, they employ language that allows them to interact with others. Indeed, English is the language of the global community (Crystal 2003; Northrup 2013; Salomone 2022). Figure 6b is a graph pointing to the top 20 most frequently used word pairs on the #HUTRI76 network. The highest number of word pairs is ‘#hutri76’ with the word ‘#17an’ which is used 3436 times, while the number of word pairs in the twentieth most frequently used is the word ‘agustus’ with the word ‘2021’ with total usage of 749 times. Among these word pairs, there are two pairs of words used by the Government of Indonesia which are used as slogans on Indonesia’s Independence Day. The two pairs of slogans are the word ‘Indonesia’ with the word ‘tangguh’ and ‘Indonesia’ with the word ‘tumbuh.’ ‘Indonesia tangguh, Indonesia tumbuh’ is a phrase on Indonesia’s Independence Day in 2021. This phrase was made by the Government of Indonesia to commemorate Indonesia’s Independence Day amid the COVID-19 pandemic. Based on the findings of the use of these words and word pairs, we argue that the discourse of nationalism on the #HUTRI76 network is dominated by the use of discourse from the Government of Indonesia. Thus, the process of reproduction discourse occurs in this network. The state, in this case, yet dominates the discourse of nationalism on Twitter. The media structure on #HUTRI76: twitter as a hub As stated by Murthy (2018), who first structured the idea of the study on Twitter, the interactive part of other platforms may turn out on Twitter. On the #HUTRI76 network, various hyperlinks, videos, and even social media on Facebook are embedded in tweets. The top ten hyperlinks on the #HUTRI76 network emerged. The first most shared medium was hyperlinked, of which six of them are hyperlinks from YouTube. Meanwhile, the other medium interpolated in tweets is one Facebook hyperlink, two Twitter hyperlinks, and one Yahoo online news hyperlink. The variety of mediums inserted in these tweets indicates a platform-sensitive framework (Burgess et al. 2018). Platform-sensitive emphasizes the specificity of the platform as a socio-technological environment that draws different users together and regulates the relationship between users of different platforms to provide things for different types of users. The specificity of this platform can also provide an overview of the users who are connected through various possible actions. Media richness then could be emerged on discussions related to Indonesia’s Independence Day on the #HUTRI76 network over Twitter as a hub. The formed modes in the nationalism message delivery related to Indonesia’s Independence Day event in the #HUTRI76 Twitter network are summarized in Fig. 7. Interrelationships between media structure, message structure, and user structure are found in this model. As has been shown, the delivery of messages of nationalism, with local and global narratives, related to #HUTRI76 which is formed in the Twitter network, is built through interactions between users. In these interactions, special word forms are used to represent certain messages involving various types of online media platforms that are linked to the Twitter network.Fig. 7 Delivery structure of the nationalism message related to Indonesia’s Independence Day events in the #HUTRI76 Twitter network Conclusion The delivery structure of nationalism message on Twitter in the context of Indonesian netizens has been investigated by employing social network analysis. Through the #HUTRI76 Twitter network, an overview of influential people, topics discussed, and content embedded in tweets on Indonesia’s Independence Day emerged. The presence of significant users was detected in the network as they have an impact on other users on the network, as well as receive much popularity (double-central users). Moreover, users which act as a bridge on the network arise and reveal the diversity of user groups in the #HUTRI76 network. This bridge has an important role to connect users on the network in discussing Independence Day. Without their existence, no conversation developed. Furthermore, the developing conversations were referring to local and global discourse. At the local level, specific terms of words reflecting Indonesia’s national identity such as ‘kemerdekaan,’ ‘kita,’ and ‘dirgahayu’ were utilized to discuss Independence Day. Even though these terms serve to strengthen the sense of national belonging, these words have implications for the formation of ethnocentrism as well. The words in English, i.e., ‘happy’ and ‘independence’ are intended to make the discourse of Indonesian Independence known globally. At this moment, especially in the discussion of #hutri76, the terms are used in forming civic nationalism—referring to individuals’ capacity to join the conversation on Twitter, as well as ethnic nationalism—attributing to the notion of shared ancestry and inclusiveness. These two forms of nationalism are possible because currently, these two concepts cannot be separated from one another. Meanwhile, there are a variety of mediums embedded in tweets discussing Indonesia’s Independence Day. The top ten hyperlinks are found and revealed on the network. Thus, Twitter could act as a hub for other social media to do cross-posting content. Due to its flexibility and openness to links and updates from other networks, Twitter is believed to be a cross-posting platform echoing the discourse on Indonesian Independence Day through the #HUTRI76 Twitter network. Author contributions This paper was conceived and designed by DKS, WK, and NK. This paper was written by DKS. Data collection and data analysis were performed by DKS. The published manuscript has been reviewed and approved by all authors. Funding This project was funded by Universitas Kristen Satya Wacana. Declarations Competing interests The authors declare no competing interests. Conflict of interest The authors declare no competing interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abdelsadek Y Chelghoum K Herrmann F Kacem I Otjacques B Community extraction and visualization in social networks applied to Twitter Inf Sci 2018 424 204 223 10.1016/J.INS.2017.09.022 Adams KM Nationalizing the local and localizing the nation ceremonials, monumental displays and national memory-making in upland Sulawesi Indones Mus Anthropol 1997 21 1 113 130 10.1525/MUA.1997.21.1.113 Agastia ID (2017) Celebrating independence, recounting common struggle. The Jakarta Post. https://www.thejakartapost.com/academia/2017/08/16/celebrating-independence-recounting-common-struggle.html Agus C Cahyanti PAB Widodo B Yulia Y Rochmiyati S Cultural-based education of Tamansiswa as a locomotive of Indonesian education system World Sustain Ser 2020 10.1007/978-3-030-15604-6_29/COVER/ Ahmed W Lugovic S Social media analytics: analysis and visualisation of news diffusion using NodeXL Online Inf Rev 2019 43 1 149 160 10.1108/OIR-03-2018-0093/FULL/XML Akuupa MU Kornes G From ‘One Namibia, one Nation’ towards ‘unity in diversity’? Shifting representations of culture and nationhood in Namibian Independence Day celebrations, 1990–2010 Anthropol S Afr 2015 36 1–2 34 46 10.1080/23323256.2013.11500041 Aleskerov FT (Faud T ogly), Shvydun S, Mescheryakova N (2021) New centrality measures in networks : how to take into account the parameters of the nodes and group influence of nodes to nodes. Chapman and Hall/CRC. 10.1201/9781003203421 Anderson BRO (1983) Imagined communities: reflections on the origin and spread of nationalism. Verso Anderson BRO (2006) Imagined communities : reflection on the origin and spread of nationalism (revised ed). Verso Arwati D Latif DV Factors inhibiting public participation in corruption prevention through E-government application in Indonesia Glob Bus Manag Res Int J 2019 11 1 81 86 Ausserhofer J Maireder A National politics on twitter Inf Commun Soc 2013 16 3 291 314 10.1080/1369118X.2012.756050 Awan I (2017) Cyber-extremism: isis and the power of social media. Society 54(2):138–149. 10.1007/S12115-017-0114-0/TABLES/5 Becatti C Caldarelli G Lambiotte R Saracco F Extracting significant signal of news consumption from social networks: the case of Twitter in Italian political elections Palgrave Commun 2019 10.1057/s41599-019-0300-3 Bernard A Theory of the hashtag 2019 Polity Press Bidart C, Degenne A, Grossetti M (2020) Living in networks: the dynamics of social relations. Cambridge University Press Bieber F Is nationalism on the rise? Assess Glob Trends Ethnopolitics 2018 17 5 519 540 10.1080/17449057.2018.1532633 Blondel J The Presidential Republic 2005 Palgrave Macmillan Bokunewicz JF Shulman J Influencer identification in Twitter networks of destination marketing organizations J Hosp Tour Technol 2017 8 2 205 219 10.1108/JHTT-09-2016-0057/FULL/PDF Borgatti SP, Evertte MG, Johnson JC (2018) Analyzing social networks. SAGE Publications Boulianne S Social media use and participation: a meta-analysis of current research Inf Commun Soc 2015 18 5 524 538 10.1080/1369118X.2015.1008542 Breuilly J Halikiopoulou D N&N Themed section: reflections on nationalism and the Russian invasion of Ukraine: Introduction Nations Natl 2022 10.1111/NANA.12874 Brown H Post-Brexit Britain: thinking about ‘English Nationalism’ as a factor in the EU referendum Int Polit Rev 2017 5 1 12 10.1057/s41312-017-0023-7 Bruns A How long is a tweet? Mapping dynamic conversation networks on twitter using Gawk and Gephi Inf Commun Soc 2011 15 9 1323 1351 10.1080/1369118X.2011.635214 Bruns A Stieglitz S Towards more systematic Twitter analysis: metrics for tweeting activities Int J Soc Res Methodol 2013 16 2 91 108 10.1080/13645579.2012.756095 Budnitskiy S (2018) Digital nationalisms: identity, strategic communication, and global internet governance [Carleton University]. https://curve.carleton.ca/system/files/etd/c88cf12b-8b77-4a50-9b91-ea458c26f227/etd_pdf/404c7c902d139b006831792437ce9b1f/budnitskiy-digitalnationalismsidentitystrategiccommunication.pdf Burgess J, Marwick AE, Poell T (Eds) (2018). The SAGE handbook of social media. Sage Publications Ltd Burt RS The social structure of competition 2009 Harvard University Press Cannadine D Introduction: Independence Day ceremonials in historical perspective 1 Round Table Commonw J Int Aff 2008 97 398 649 665 10.1080/00358530802327829 Castillo-De Mesa J Gómez-Jacinto L Connectedness, engagement, and learning through social work communities on LinkedIn Psychosoc Interv 2020 29 2 103 112 10.5093/PI2020A4 Clark JL Algoe SB Green MC Social network sites and well-being: the role of social connection Curr Dir Psychol Sci 2017 27 1 32 37 10.1177/0963721417730833 Crystal D English as a global language 2003 Cambridge University Press Diani M (2014) Social movements and collective action. In: The SAGE handbook of social network analysis, SAGE Publications Lt DiMaggio P (2014) Cultural networks. In: The SAGE handbook of social network analysis, pp 286–300. SAGE Publications Ltd. 10.4135/9781446294413.N20 Drosio S Stanek S The Big Data concept as a contributor of added value to crisis decision support systems J Decis Syst 2016 25 s1 228 239 10.1080/12460125.2016.1187404 Duffy M (2022) Shinzo abe’s murder and Japan’s History of political assassination. E-international relations. https://www.e-ir.info/2022/07/08/opinion-shinzo-abes-murder-and-japans-history-of-political-assassination/ Faizal Kasmani M Sabran R Ramle N Can Twitter be an effective platform for political discourse in Malaysia? A study of #PRU13 Procedia Soc Behav Sci 2014 155 348 355 10.1016/j.sbspro.2014.10.304 Fisher E Mehozay Y How algorithms see their audience: media epistemes and the changing conception of the individual Media Cult Soc 2019 41 8 1176 1191 10.1177/0163443719831598 Fuhse JA The meaning structure of social networks Sociol Theory 2009 27 1 51 73 10.1111/J.1467-9558.2009.00338.X Fuhse J Social networks of meaning and communication 2022 Oxford University Press Gallagher JR Chen Y Wagner K Wang X Zeng J Kong AL Peering into the internet abyss: using Big Data audience analysis to understand online comments Tech Commun Q 2019 29 2 155 173 10.1080/10572252.2019.1634766 Gellner E Nations and nationalism 1983 Cornell University Press Gellner E Nations and nationalism 2006 Blackwell Gellner E (2009) Nations and nationalism (second edi). Cornell University Press. https://www.amazon.com/Nations-Nationalism-New-Perspectives-Past/dp/0801475007/ref=sr_1_1?crid=BK412RBEX9LG&dchild=1&keywords=ernest+gellner&qid=1614871404&s=books&sprefix=ernest+gellner%2Cstripbooks-intl-ship%2C535&sr=1-1 Giroux HA White nationalism, armed culture and state violence in the age of Donald Trump Philos Soc Crit 2017 43 9 887 910 10.1177/0191453717702800 Golbeck J (2015) Analyzing networks. In: Introduction to social media investigation, pp 221–235. 10.1016/B978-0-12-801656-5.00021-4 Guo L Mays K Wang J Whose story wins on Twitter? Journal Stud 2017 20 4 563 584 10.1080/1461670X.2017.1399813 Guo J Zhang C Wu Y Li H Liu Y Examining the determinants and outcomes of netizens’ participation behaviors on government social media profiles Aslib J Inf Manag 2018 70 4 306 325 10.1108/AJIM-07-2017-0157 Gusterson H From Brexit to Trump: anthropology and the rise of nationalist populism Am Ethnol 2017 44 2 209 214 10.1111/amet.12469 Hajjaj B Nationalism and national identity formation in Bangladesh: a colonial legacy behind the clash of language and religion Asian J Comp Polit 2020 7 3 435 448 10.1177/2057891120938145 Hall JA When is social media use social interaction? Defining mediated social interaction New Media Soc 2016 20 1 162 179 10.1177/1461444816660782 Hansen DL Shneiderman B Smith MA Himelboim I Analyzing social media networks with NODEXL insights from a connected world 2020 Morgan Kaufmann Hansen DL, Shneiderman B, Smith MA, Himelboim I (2020b). Social network analysis: measuring, mapping, and modeling collections of connections. In: Analyzing social media networks with NodeXL: insights from a connected world. Morgan Kaufmann. 10.1016/B978-0-12-817756-3.00003-0 Hansen DL, Shneiderman B, Smith MA (2011a) Social network analysis: measuring, mapping, and modeling collections of connections. In: Analyzing social media networks with NodeXL, pp 31–50. 10.1016/B978-0-12-382229-1.00003-5 Hansen D, Shneiderman B, Smith MA (2011b) Analyzing social media networks with NodeXL: insights from a connected world. Elsevier Inc Hansen DL, Shneiderman B, Smith MA, Himelboim I (2020c) Twitter: information flows, influencers, and organic communities. In: Analyzing social media networks with NodeXL: insights from a connected world, pp 161–178. 10.1016/B978-0-12-817756-3.00011-X Hasan M Orgun MA Schwitter R A survey on real-time event detection from the Twitter data stream J Inf Sci 2018 44 4 443 463 10.1177/0165551517698564/ASSET/IMAGES/LARGE/10.1177_0165551517698564-FIG1.JPEG Hasfi N Fisher MR Sahide MAK Overlooking the victims: Civic engagement on Twitter during Indonesia’s 2019 fire and haze disaster Int J Disaster Risk Reduct 2021 60 102271 10.1016/J.IJDRR.2021.102271 Hatley B National ritual, neighborhood performance: celebrating Tujuhbelasan Indonesia 1982 34 55 10.2307/3350948 Hatley B Performing identity and community in Indonesia in modern times ACCESS Crit Perspec Commun Cult Policy Stud 2012 31 2 27 38 Hauben M Researching the “net” a talk on the evolution of usenet news and the significance of the global computer network Amat Comput 2019 32 2 1 30 Hellman J The use of ‘cultures’ in official representations of Indonesia: the fiftieth anniversary of independence Indones Malay World 2007 26 74 1 12 10.1080/13639819808729907 Hendrickson B Rosen D Aune RK An analysis of friendship networks, social connectedness, homesickness, and satisfaction levels of international students Int J Intercult Relat 2011 35 3 281 295 10.1016/J.IJINTREL.2010.08.001 Himelboim I Smith MA Rainie L Shneiderman B Espina C Classifying Twitter topic-networks using social network analysis Soc Media Soc 2017 10.1177/2056305117691545 Hobsbawm E, Ranger T (eds) (1983) The invention of tradition. Cambridge University Press Hyttinen A Näre L Symbolic and ritual enactments of nationalism: a visual study of Jobbik’s gatherings during Hungarian national day commemorations Vis Stud 2017 32 3 236 250 10.1080/1472586X.2017.1358104 Iglesias CA, Moreno A (2020) Sentiment analysis for social media. MDPI AG Irandoost SF Sedighi S Hoseini AS Ahmadi A Safari H Ebadi Fard Azar F Yoosefi Lebni J Activities and challenges of volunteers in confrontation with COVID-19: a qualitative study in Iran Int J Disaster Risk Reduct 2022 82 103314 10.1016/J.IJDRR.2022.103314 36168596 Jami Pour M Hosseinzadeh M Mansouri NS Challenges of customer experience management in social commerce: an application of social network analysis Internet Res 2022 32 1 241 272 10.1108/INTR-01-2021-0076/FULL/PDF Jan SK Investigating virtual communities of practice with social network analysis: guidelines from a systematic review of research Int J Web Based Commun 2019 15 1 25 43 10.1504/IJWBC.2019.098697 Jendoubi S, Chebbah M, Martin A, Evidential AM (2018) Independence maximization on Twitter network. In: International conference on belief functions. https://hal.archives-ouvertes.fr/hal-01879620 Karakosta C Mylona Z Karásek J Papapostolou A Geiseler E Tackling covid-19 crisis through energy efficiency investments: decision support tools for economic recovery Energy Strat Rev 2021 38 100764 10.1016/J.ESR.2021.100764 Kemp S (2022) Digital 2022: Indonesia. Datareportal. https://datareportal.com/reports/digital-2022-indonesia Knott E Nationalism and belonging: introduction Nations Natl 2017 23 2 220 226 10.1111/nana.12297 Knott E Existential nationalism: Russia’s war against Ukraine Nations Natl 2022 10.1111/NANA.12878 Kohn H The idea of nationalism 1944 Macmillan Kook R Changing representations of national identity and political legitimacy: Independence Day celebrations in Israel, 1952–1998 National Identities 2006 7 2 151 171 10.1080/14608940500144096 Kuzio T Imperial nationalism as the driver behind Russia’s invasion of Ukraine Nations Natl 2022 10.1111/NANA.12875 Lipschultz JH, Freberg KJ, Luttrell R (2022) The Emerald handbook of computer-mediated communication and social media. Emerald Publishing Limited Lovari A Bowen SA Social media in disaster communication: a case study of strategies, barriers, and ethical implications J Public Aff 2020 20 1 e1967 10.1002/PA.1967 MacKinnon R The netizen Development 2012 55 201 204 10.1057/dev.2012.5 Madu L Indonesia’s foreign policy under president Jokowi: more domestic and nationalist orientations Int J Sci Res Sci Technol 2017 2 3 189 197 Magcamit MI The fault in Japan’s stars: Shinzo Abe, North Korea, and the quest for a new Japanese constitution Int Polit 2019 57 4 606 633 10.1057/S41311-019-00186-8 Maireder A Weeks BE Gil de Zúñiga H Schlögl S Big Data and political social networks: introducing audience diversity and communication connector bridging measures in social network theory Soc Sci Comput Rev 2015 35 1 126 141 10.1177/0894439315617262 Malinova O Politics of memory and nationalism Natl Pap 2021 49 6 997 1007 10.1017/NPS.2020.87 Mcewen N Irreconcilable sovereignties? Brexit and Scottish self-government Territ Polit Gov 2022 10 5 733 749 10.1080/21622671.2022.2044898 Merriman P National movements Environ Plan C Polit Space 2020 38 4 585 586 10.1177/2399654420912445C Mihelj S Jiménez-Martínez C Digital nationalism: understanding the role of digital media in the rise of ‘new’ nationalism Nations Natl 2021 27 2 331 346 10.1111/NANA.12685 Mirbabaie M Ehnis C Stieglitz S Bunker D Rose T Digital nudging in social media disaster communication Inf Syst Front 2021 23 5 1097 1113 10.1007/S10796-020-10062-Z/TABLES/7 Moody J Light R The oxford handbook of social networks 2020 Oxford University Press Morgan-Lopez AA Kim AE Chew RF Ruddle P Predicting age groups of Twitter users based on language and metadata features PLoS ONE 2017 10.1371/journal.pone.0183537 Murthy D Twitter: social communication in the Twitter age 2018 Polity Press Murthy D (2013) Twitter: social communication in the Twitter age. Polity Press Mylonas H Tudor M Nationalism: what we know and what we still need to know Annu Rev Polit Sci 2021 24 109 132 10.1146/annurev-polisci-041719 Niebuur J Van Lente L Liefbroer AC Steverink N Smidt N Determinants of participation in voluntary work: a systematic review and meta-analysis of longitudinal cohort studies BMC Public Health 2018 18 1 1 30 10.1186/S12889-018-6077-2/TABLES/9 Northrup D (2013) How english became the global language. In: How english became the global language, pp 1–205. 10.1057/9781137303073 Nyoka P Tembo M Dimensions of democracy and digital political activism on Hopewell Chin’ono and Jacob Ngarivhume Twitter accounts towards the July 31st demonstrations in Zimbabwe Cogent Soc Sci 2022 10.1080/23311886.2021.2024350 Nyyssönen H (2009) The politics of calendar: Independence Day in the Republic of Finland. In: National Days, pp 136–150. 10.1057/9780230251175_10 Onorati T, Díaz P (2015) Semantic visualization of Twitter usage in emergency and crisis situations. In: Lecture notes in business information processing, vol 233, pp 3–14. 10.1007/978-3-319-24399-3_1/COVER Paasi A Dancing on the graves: Independence, hot/banal nationalism and the mobilization of memory Polit Geogr 2016 54 21 31 10.1016/J.POLGEO.2015.07.005 Pascual-Ferrá P Alperstein N Barnett DJ Social network analysis of COVID-19 public discourse on Twitter: implications for risk communication Disaster Med Public Health Prep 2022 16 2 561 569 10.1017/DMP.2020.347 32907685 Pettit P (2002) Republicanism: a theory of freedom and government. Oxford University Press Pulido CM Redondo-Sama G Sordé-Martí T Flecha R Social impact in social media: a new method to evaluate the social impact of research PLoS ONE 2018 10.1371/journal.pone.0203117 Räbiger S Spiliopoulou M A framework for validating the merit of properties that predict the influence of a twitter user Expert Syst Appl 2015 42 5 2824 2834 10.1016/J.ESWA.2014.11.006 Ramdhani N Anggraini R PDIP NTB political communication in building loyalty of cadres and constituents in elections J Mantik 2021 5 3 2081 2100 Rauschnabel PA Sheldon P Herzfeldt E What motivates users to hashtag on social media? Psychol Mark 2019 36 5 473 488 10.1002/MAR.21191 Restad HE What makes America great? Donald Trump, national identity, and US foreign policy Glob Aff 2020 6 1 21 36 10.1080/23340460.2020.1734955 Rocca GL Possible selves of a hashtag: moving from the theory of speech acts to cultural objects to interpret hashtags Int J Sociol Anthropol 2020 12 1 1 9 10.5897/IJSA2019.0833 Roth S Introduction: contemporary counter-movements in the Age of Brexit and Trump Sociol Res Online 2018 23 2 496 506 10.1177/1360780418768828 Salomone RC (2022) The rise of English: global politics and the power of language. Oxford University Press Sanawi JB Samani MC Taibi M #Vaccination: identifying influencers in the vaccination discussion on Twitter through social network visualisation Mus Chairil Samani Int J Bus Soc 2017 18 718 726 Santoso DH New media and nationalism in indonesia: An analysis of discursive nationalism in online news and social media after the 2019 indonesian presidential election J Komun Malays J Commun 2021 37 2 289 304 10.17576/JKMJC-2021-3702-18 Sari DK, Ahmad J, Hergianasari P, Harnita PC, Wibowo NA (2021) Quantitative study of the cyber-nationalism spreading on Twitter with hashtag Indonesia and Malaysia using social network analysis. Media Watch J 12(1):161–171. 10.15655/mw/2021/v12i1/205465 Schertzer R Woods E #Nationalism: the ethno-nationalist populism of Donald Trump’s Twitter communication Ethn Racial Stud 2020 44 7 1154 1173 10.1080/01419870.2020.1713390 Schneider F Emergent nationalism in China’s sociotechnical networks: how technological affordance and complexity amplify digital nationalism Nations Natl 2022 28 1 267 285 10.1111/NANA.12779 Schneider F (2018) Nationalism and its digital modes. In: China’s digital nationalism, pp 25–56. Oxford University Press. 10.1093/oso/9780190876791.003.0002 Schreiner T (2018) Information, opinion, or rumor? The role of Twitter during the post-electoral crisis in Côte d’Ivoire: Soc Media Soc 10.1177/2056305118765736 Seman RAA, Laidey NM, Ali RS (2019) Netizens’ political engagement in Malaysia: impact of anti fake news act 2018. J Pengaj Media Malays 21(1):77–87. 10.22452/JPMM.VOL21NO1.6 Singenberger NP (2021) Review: the invention of tradition. Documenting the periphery. https://dlf.uzh.ch/sites/skandinavien-postkolonial/review-the-invention-of-tradition/ Small TA What the hashtag? Inf Commun Soc 2011 14 6 872 895 10.1080/1369118X.2011.554572 Smith MA, Shneiderman B, Milic-Frayling N, Rodrigues EM, Barash V, Dunne C, Capone T, Perer A, Gleave E (2009) Analyzing (social media) networks with NodeXL. https://www.cs.umd.edu/~ben/papers/Smith2009Analyzing.pdf Smith AD (2010) Nationalism: theory, ideology, history. Polity Squibb S (2016) This machine builds fascists: nationalism as mode of distribution. E-Flux J. https://www.e-flux.com/journal/78/82918/this-machine-builds-fascistsnationalism-as-mode-of-distribution/ Steinert-Threlkeld ZC Twitter as Data 2018 Cambridge University Press Stephens AC Feeling “Brexit”: nationalism and the affective politics of movement GeoHumanit 2019 5 2 405 423 10.1080/2373566X.2019.1620623 Stolz S Schlereth C Predicting tie strength with ego network structures J Interact Mark 2021 54 40 52 10.1016/J.INTMAR.2020.10.001 Stromer-Galley J (2019) Vulgar eloquence in the digital age. In: Power shift? Political leadership and social media, pp 34–48. Routledge Tamir Y Not so civic: is there a difference between ethnic and civic nationalism? Annu Rev Polit Sci 2019 2 419 434 10.1146/annurev-polisci-022018 Tanasaldy T Palmer C Discrimination, sport and nation building among Indonesian Chinese in the 1950s Indones Malay World 2019 47 137 47 65 10.1080/13639811.2019.1559564 Tong J, Zuo L (2021) The Brexit referendum on Twitter. Emerald Publishing Limited Vessey R Nationalist language ideologies in tweets about the 2019 Canadian general election Discourse Context Media 2021 39 100447 10.1016/J.DCM.2020.100447 Visoka G Everyday peace capture: nationalism and the dynamics of peace after violent conflict Nations Natl 2020 26 2 431 446 10.1111/NANA.12591 Wang Z From crisis to nationalism? The conditioned effects of the COVID-19 crisis on neo-nationalism in Europe Chin Polit Sci Rev 2021 6 20 39 10.1007/s41111-020-00169-8 Wang L Zhong Q Research on the structures and features of netizens’ demand for popular science: a search data perspective Cult Sci 2019 2 2 129 150 10.1177/209660831900200205 Webber J Independence day as a cosmopolitan moment: teaching international relations Int Stud Perspect 2005 6 3 374 392 10.1111/J.1528-3577.2005.00214.X Wellman B (2001) Little boxes, glocalization, and networked individualism. In: Digital cities II: computational and sociological approaches. Kyoto workshop on digital cities. Springer, Berlin, Heidelberg. 10.1007/3-540-45636-8_2 White KB Resmondo ZN Jennings JC Creel LM Kelly Pryor BN A social network analysis of interorganisational collaboration: efforts to improve social connectedness Health Soc Care Community 2022 10.1111/HSC.14044 Wicaksana GWI Economic nationalism for political legitimacy in Indonesia J Int Relat Dev 2019 24 1 27 50 10.1057/S41268-019-00182-8 Wild C The radio midwife: Some thoughts on the role of broadcasting during the Indonesian struggle for independence Indones Circ Sch Orient Afr Stud Newsl 2007 19 55 34 42 Williamson W Ruming K Using social network analysis to visualize the social-media networks of community groups: two case studies from Sydney J Urban Technol 2016 23 3 69 89 10.1080/10630732.2016.1197490 Wingfield-Hayes R (2022) Shinzo Abe death: Shock killing that could change Japan forever. BBC News. https://www.bbc.com/news/world-asia-62074223 Yan L Pedraza-Martinez AJ Social media for disaster management: operational value of the social conversation Prod Oper Manag 2019 28 10 2514 2532 10.1111/POMS.13064 Yang S, Keller FB, Zheng L (2020) Basics of social network analysis. In: Social network analysis: methods and examples, pp 2–25. SAGE Publications, Inc. 10.4135/9781071802847.N1 Yu J Muñoz-Justicia J Free and low-cost Twitter research software tools for social science Soc Sci Comput Rev 2020 40 1 124 149 10.1177/0894439320904318 Żuk P Żuk P The Independence Day as a nationalist ritual: framework of the March of Independence in Poland Ethnography 2022 23 1 14 37 10.1177/14661381211073406	36475091	PMC9716509	NO-CC CODE	2022-12-03 23:20:58	no	Soc Netw Anal Min. 2022 Dec 2; 12(1):173	utf-8	Soc Netw Anal Min	2,022	10.1007/s13278-022-01006-3	oa_other
==== Front Nat Rev Microbiol Nat Rev Microbiol Nature Reviews. Microbiology 1740-1526 1740-1534 Nature Publishing Group UK London 822 10.1038/s41579-022-00822-w Review Article SARS-CoV-2 viral load and shedding kinetics http://orcid.org/0000-0002-5789-8988 Puhach Olha 1 http://orcid.org/0000-0003-0601-3550 Meyer Benjamin 2 http://orcid.org/0000-0002-4850-7172 Eckerle Isabella [email protected] 134 1 grid.8591.5 0000 0001 2322 4988 Department of Medicine, Faculty of Medicine, University of Geneva, Geneva, Switzerland 2 grid.8591.5 0000 0001 2322 4988 Centre for Vaccinology, Department of Pathology and Immunology, University of Geneva, Geneva, Switzerland 3 grid.150338.c 0000 0001 0721 9812 Geneva Centre for Emerging Viral Diseases, Geneva University Hospitals, Geneva, Switzerland 4 grid.150338.c 0000 0001 0721 9812 Division of Infectious Diseases, Geneva University Hospitals, Geneva, Switzerland 2 12 2022 115 21 10 2022 © Springer Nature Limited 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. SARS-CoV-2 viral load and detection of infectious virus in the respiratory tract are the two key parameters for estimating infectiousness. As shedding of infectious virus is required for onward transmission, understanding shedding characteristics is relevant for public health interventions. Viral shedding is influenced by biological characteristics of the virus, host factors and pre-existing immunity (previous infection or vaccination) of the infected individual. Although the process of human-to-human transmission is multifactorial, viral load substantially contributed to human-to-human transmission, with higher viral load posing a greater risk for onward transmission. Emerging SARS-CoV-2 variants of concern have further complicated the picture of virus shedding. As underlying immunity in the population through previous infection, vaccination or a combination of both has rapidly increased on a global scale after almost 3 years of the pandemic, viral shedding patterns have become more distinct from those of ancestral SARS-CoV-2. Understanding the factors and mechanisms that influence infectious virus shedding and the period during which individuals infected with SARS-CoV-2 are contagious is crucial to guide public health measures and limit transmission. Furthermore, diagnostic tools to demonstrate the presence of infectious virus from routine diagnostic specimens are needed. A better understanding of the transmission of SARS-CoV-2 is essential to inform public health measures. In this Review, Puhach, Meyer and Eckerle explore insights into what influences SARS-CoV-2 shedding, how this drives transmission and the tools available to measure this and determine infectiousness. Subject terms SARS-CoV-2 Viral infection Clinical microbiology ==== Body pmcIntroduction At the end of 2019, a novel coronavirus emerged, later termed severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which is the causative agent of coronavirus disease 2019 (COVID-19). SARS-CoV-2 primarily targets multiciliated cells in the upper respiratory tract (URT), but was also reported to infect cells outside the URT1. It can spread to the lower respiratory tract (LRT), where it infects alveoli, leading to reduced gas exchange, inflammation and pulmonary pathologies that are typical of COVID-19 (ref.2). Individuals who are infected shed the virus through the URT, with emission of infectious virus leading to secondary transmission and thus further spread of the virus. Because of their nonspecific clinical presentation, precise diagnostic tools are needed to identify SARS-CoV-2 infections. Specific real-time PCR (RT-PCR) assays were quickly available after the emergence of the virus, later followed by antigen-detecting (rapid) diagnostic tests (Ag-RDTs) and serological assays. Although detection of viral RNA in respiratory specimens by RT-PCR is highly sensitive and specific, it does not distinguish between replication-competent virus and residual RNA. In the absence of a diagnostic test, infectiousness is often established using one of two proxies: the presence of viral RNA above a defined cycle threshold (Ct) value, or a positive Ag-RDT. RT-PCR is a useful tool for initial diagnosis, whereas Ag-RDTs can serve as an indicator for ending the isolation period. This is because viral RNA (which would be picked up by RT-PCR) remains detectable in the absence of infectious virus, whereas positivity of Ag-RDTs better correlates with the presence of infectious virus. Aside from the respiratory tract, SARS-CoV-2 RNA has been detected in peripheral blood, stool, urine and ocular secretions3–7. Virus isolation from non-respiratory specimens was unsuccessful in most studies4,8,9, with very few reported cases of infectious virus presence in non-respiratory specimens10–13. Furthermore, viral loads from respiratory tract samples were found to be much higher than from other materials, the latter often with RNA viral loads that are incompatible with the presence of infectious virus. Such specimens are not considered relevant for transmission and therefore, we concentrate on SARS-CoV-2 virus shedding only through the respiratory tract. Here, we elucidate the relationship between SARS-CoV-2 viral load and infectious virus presence, the biological and host factors that determine infectious virus shedding, measurement of infectious virus and the role diagnostics can have as a proxy for infectious virus shedding. Measuring SARS-CoV-2 viral load The gold standard for laboratory diagnosis of a respiratory tract infection is demonstration of viral RNA with a virus-specific (semi-)quantitative RT-PCR from material collected from the respiratory tract. The most commonly used materials are swab specimens from the nasopharynx or oropharynx, but swabs of the nasal cavity, saliva or gargled liquid solution have also been suggested as alternative materials, with the advantage of being a less uncomfortable procedure for the participant. Viral load as determined by RT-PCR is either expressed as the number of viral RNA copies per millilitre of viral transport medium or per swab, or by the arbitrary test-specific Ct value. By contrast, infectiousness is determined by qualitative or quantitative assessment of infectious virus in a clinical specimen by replication of virus in cell culture. The limitations to measuring viral shedding are described in Box 1. In this Review, we refer to viral particles that can cause infection as infectious virus, and to viral RNA levels (which are widely used as surrogates for infectious virus) as viral load. Box 1 Limitations to measuring viral load Specimen selection site The anatomical site chosen to collect the swab specimen for detection of SARS-CoV-2 might influence viral load detection. Higher RNA viral load was reported from nasopharyngeal than oropharyngeal swabs28,124,181. As a result, nasopharyngeal samples show the highest diagnostic accuracy compared with other upper respiratory tract samples182. Similarly, higher virus isolation success was reported from nasopharyngeal swabs than from saliva, nasal or sublingual swabs124. However, another study found higher RNA viral loads in the throat and sputum than from nasal swabs30. Two studies indicate that virus can be detected earlier in the throat29 or saliva33, but reaches significantly higher levels and remains detectable for longer in the nose29,33. A meta-analysis, which evaluated different clinical sampling methods using nasopharyngeal swab as a reference, demonstrated that pooled nasal and throat swabs showed the best diagnostic performance183. Notably, this analysis revealed higher heterogeneity of results in studies using nasal or saliva specimens than using pooled nasal and throat swabs183. The effect of the swabbing method (self-administered or performed by trained person) on measured viral loads cannot be overlooked: the sensitivity of antigen-detecting (rapid) diagnostic tests achieved by health-care professionals was higher than for self-tests57,184. Impact of individual infection kinetics To date, there is a limited number of studies that describe the longitudinal dynamics of SARS-CoV-2 shedding29,33. Most of the studies used only a single time point to collect respiratory swabs from individuals who were infected for measurement of viral load. As a result, different times from symptom onset can be a confounding factor when comparing viral load between different patients, which might also explain the variation of available data on viral load. Influence of epidemic period RNA viral loads across the specimens collected at single time points were found to indicate the trajectory of the epidemic, as a high proportion of individuals who were recently infected with low cycle threshold values correlates with a higher reproduction number, indicative of a growing epidemic185. Similarly, the rise and fall of RNA viral load correlated with the number of COVID-19 cases and hospital admissions across the population as it was identified using SARS-CoV-2 RNA measurement in wastewater samples186. Moreover, variation in estimates of the mean incubation period was shorter before the epidemic peak in China than after the peak187. Therefore, sampling at single time points can be biased by the epidemic period and might reflect more epidemiological dynamics than individual shedding kinetics. Influence of SARS-CoV-2 variant of concern Available data on SARS-CoV-2 variants of concern demonstrated that, although the overall pattern of viral load dynamics is conserved between the variants, infection with different SARS-CoV-2 variants of concern led to highly distinct infectious virus amounts and RNA viral loads25,86,87,90,92,94 and variations in SARS-CoV-2 incubation period113. Therefore, extrapolation of our understanding from shedding of current or earlier SARS-CoV-2 variants to newly emerged variants may be of only limited value. Detection of infectious virus The gold standard for determining the presence of infectious (that is, replication competent) virus in respiratory specimens is the recovery of virus in cell culture, a procedure that is commonly termed virus isolation (Fig. 1).Fig. 1 Methods to measure infectious virus and RNA viral load. Swab specimens from the nasopharynx or oropharynx are used for detection of SARS-CoV-2 viral loads. Detection of viral nucleic acids (RNA) is performed by quantitative real-time PCR (qRT-PCR). Viral RNA is extracted from lysed virus, reverse transcribed and amplified by qPCR using primers specific for one or more target regions in the viral genome. The amplification cycle at which samples cross the threshold (cycle threshold) defines the amount of viral RNA. RNA viral load can be expressed as the number of viral RNA copies per millilitre, or by the arbitrary test-specific cycle threshold value. Lateral flow assays detect the presence of specific viral proteins in the lysed viral particles. SARS-CoV-2 nucleocapsid is used in most antigen-detecting (rapid) diagnostic tests. The presence of infectious (replication-competent) virus in respiratory specimens can only be determined by the recovery of virus in cell culture by isolation or by quantification of infectious virus titres using 50% tissue culture infectious dose (TCID50), focus-forming assays or plaque-forming assays. Virus isolation is performed by applying infectious medium on the monolayer of cells; isolation success is determined by the presence of a cytopathic effect approximately 3–5 days post-infection. White colour indicates the presence of a cytopathic effect in cells. For quantification of infectious virus titres, serial dilutions of respiratory samples are performed and used for inoculation on the monolayer of cells. In TCID50, 3–5 days post-infection, viral-induced cytopathic effect is classically defined using microscopy. In focus-forming assays, cells are fixed 1 day post-infection and immunostaining with virus-specific antibodies is performed to detect groups of infected cells (foci). The foci, indicating the presence of infectious virus, are displayed in blue. In plaque-forming assays, plates are fixed 2–3 days post-infection and stained with crystal violet; wells with individual plaques are used to determine viral titres. The plaques, indicating the presence of infectious virus, are displayed in white. In the case of SARS-CoV-2, various cell lines and primary cells can be used for virus isolation, including those that express angiotensin-converting enzyme 2 (ACE2; the receptor required for virus entry) or transmembrane protease 2 (TMPRSS2; which is also important for virus entry)14. A cell line derived from African green monkey kidney cells, Vero E6, is commonly used for virus isolation, propagation and titration15. Other human cell lines that have been successfully used for SARS-CoV-2 isolation are a colorectal adenocarcinoma cell line (Caco-2), a lung adenocarcinoma cell line (Calu-3), a lung adenocarcinoma cell line ectopically overexpressing ACE2 (A549) and a human hepatocellular carcinoma cell line (Huh7)16,17. The presence of infectious virus in the cell culture is qualitatively assessed using light microscopy, which can be used to identify cells undergoing the cytopathic effects (and death) caused by SARS-CoV-2 infection, consisting of syncytium formation, cell rounding, detachment and degeneration17. Infection is usually confirmed by a second method, either by a specific RT-PCR for viral RNA from the supernatant of infected cells, indicating virus replication by an increase of viral load over time in comparison to the baseline sample, or by immunostaining for viral proteins15,18. This qualitative measurement of virus presence cannot, however, quantify the infectious virions in the inoculated specimens, although samples with lower viral load commonly show delayed development of a cytopathic effect19. Instead, methods such as plaque assays, focus-forming assays or 50% tissue culture infectious dose (TCID50) can be used to quantify infectious virus in a patient sample. The above methodologies are reliable tools to detect infectious virus in clinical specimens of individuals who are infected with SARS-CoV-2, although there are limitations. Detection of viable virus particles is highly influenced by the quality of the sample, and infectious viral particles can quickly lose their infectiousness in unsuitable storage conditions. To preserve infectious virus in specimens, swab samples from patients infected with SARS-CoV-2 should be immediately submerged in a viral transport medium suitable for cell culture and stored at −80 °C as early as possible after collection. Prolonged exposure to higher temperatures or repeated freeze–thaw cycles can drastically influence the quality of the sample, leading to potentially complete loss of infectious viral particles. Therefore, many factors can influence the reproducibility of the results between different laboratories. Furthermore, cell lines used for isolation can show a high variability between laboratories even when they are presumably the same. Consumables used during cell culture, such as culture medium or additives such as fetal bovine serum and antibiotics, could potentially also impact virus isolation success. In human primary airway epithelial cells, which mimic the primary site of entry in the human respiratory tract, the probability of isolating infectious virus was reduced compared with that of Vero E6 cells, indicating that infectious virus determined using Vero E6 cells might be overestimated for assessing transmission risks in vivo20. Importantly, all cell culture work with SARS-CoV-2 is done under biosafety level 3 conditions, so only specially trained personnel in laboratories with advanced infrastructure can perform these experiments. Thus, detection of viable virus through virus isolation is not suitable for diagnostics and is restricted to research only. Detection of RNA viral load Techniques for detecting viral RNA by RT-PCR were quickly established at the beginning of the pandemic21,22 (Fig. 1). The high specificity and sensitivity of RT-PCR make it the gold standard for diagnosing SARS-CoV-2 infections. Quantitative RT-PCR assays provide a Ct value, which is inversely correlated with the concentration of the target viral RNA in the clinical sample (that is, the higher the value, the lower the target RNA in the sample). By using an external standard with a defined number of RNA copies, Ct values can be transformed into absolute viral RNA copy numbers or international units per millilitre of viral transport medium or per total swab. Although RT-PCR cannot directly determine infectiousness owing to its inability to differentiate between replication-competent (infectious) virus and residual (non-infectious) viral RNA, a correlation between RNA viral load and the presence of infectious virus has been sought. Several studies have attempted to correlate the quantity of viral RNA with infectiousness by isolating virus across a range of Ct values. Indeed, there was a stepwise decrease in the probability of virus isolation with increasing Ct values in samples collected during the first 8 days post-onset of symptoms (dpos)18,23. However, other studies have found that the correlation between infectious virus and RNA viral load was low and that viral load (or Ct values as a proxy) is only a weak predictor of infectious virus presence in the first 5 dpos4,20,24,25. Furthermore, when taking a certain Ct value or RNA copy number as a threshold, it is not possible to determine whether the RNA viral load is increasing or already decreasing; therefore, a low viral load could be measured at the end of infection or in the early (pre-)symptomatic phase before reaching peak viral load. In a routine diagnostic context, analytical sensitivity and limits of detection may vary between the tests and laboratories where they are applied. An analytical performance comparison between different RT-PCR assays showed variation between the measured Ct values and the detection rate26. Therefore, application of RNA standards and calculation of RNA genome copy number based on a standard curve can improve comparability between laboratories and assays. To facilitate easier calibration and control of nucleic acid amplification techniques, an international standard with assigned potency in the form of an inactivated SARS-CoV-2 isolate was introduced by the World Health Organization (WHO)27. As with the detection of infectious virus, several other parameters can influence whether viral load can be detected. The site of specimen collection can impact the findings on viral load; although some studies report higher RNA viral load in nasal or nasopharyngeal swabs28,29, others show higher RNA viral load in throat samples30. Moreover, the transport media used for the sample, storage condition and quality of the sample may further influence the detection of viral RNA and their usefulness and limitations when extrapolating to potential infectiousness. Although new variants have impacted some gene targets, in most instances, they did not have a major effect on molecular diagnostics, owing to the use of dual-target assays (in which at least two viral genes are detected simultaneously)31. Antigen-detecting rapid diagnostic tests Most lateral flow tests are designed to detect SARS-CoV-2 nucleocapsid protein, as a proxy for infectious virus, in nasal or nasopharyngeal swabs32–36 (Fig. 1). Indeed, most studies on Ag-RDT detection show good concordance with RT-PCR positivity when Ct values are below 25–30, a viral load compatible with the presence of infectious virus, whereas higher Ct values give less reliable results34,37–41. Early time points during infection often give negative results with Ag-RDT in individuals who have tested positive by PCR29,42. On average, the first positive Ag-RDT results are obtained about 1–2 days later than positive PCR results37, whereas the highest sensitivity in patients was shown during the first 7 dpos in the studies with ancestral SARS-CoV-2 (refs.42–44). Antigen tests show highest sensitivity for specimens containing infectious virus and with Ct values below 25 (refs.45–49), and their positivity highly correlates with the presence of infectious virus34,45,47,50. By contrast, Ag-RDTs are less sensitive to low RNA viral loads (which have higher Ct values)51. Several studies have demonstrated a strong correlation between Ag-RDT positivity and the period in which infectious virus can be detected, indicating that Ag-RDTs can add an additional safety layer for deciding when to end isolation29,39. However, some inconsistencies between studies and tests have been noted. For instance, there have been reports (across a range of studies and Ag-RDTs) of failure to detect viral antigens in specimens with a low Ct value and/or containing infectious virus (beyond the early acute phase)46,50. Moreover, there are seldom reports of Ag-RDTs remaining positive after more than 10 dpos42,46,50. As most studies failed to isolate infectious virus after more than 10 dpos, it remains unclear whether Ag-RDT positivity beyond 10 dpos correlates with infectious virus shedding. One study showed that antigen tests predict infectiousness more accurately at 1–5 dpos, than at 6–11 dpos52. Notably, there was a good correlation between Ag-RDT positivity and infectious virus isolation within the first 11 dpos52. Conflicting results were found for sensitivity and specificity of Ag-RDTs for detection of SARS-CoV-2 variants, with large variations between manufacturers, the type of setting in which the Ag-RDTs were used (self-tests versus tests collected by a health-care professional) and the type of sample used for detection (nasal versus oral)53–57. With increasing hybrid immunity and the presence of mucosal antibodies, Ag-RDTs may further lose sensitivity58. Viral load and shedding dynamics Viral loads are used as a proxy to characterize infectious viral shedding. The exact time for which individuals remain infectious is laborious to estimate and is likely to vary between patients. Viral factors, such as viral variant, and host factors, such as patient age and sex and immune status, influence shedding dynamics. Viral load as a key determinant of viral shedding After the emergence of SARS-CoV-2 in late 2019, the first details on viral load and infectious virus shedding were measured in a cluster of infections that occurred in January 2020 in Germany, assessing nine immunocompetent individuals with a mild course of disease4. Peak RNA viral loads were reached in the early symptomatic period at 5 dpos, a finding that was confirmed by other studies reporting peak viral loads at the time of symptom onset or even shortly before4,7,28,59. RNA viral loads gradually declined over the course of the disease in the nasopharyngeal and throat swabs, reaching low or undetectable levels 2 weeks after symptom onset4,23,59,60 (Fig. 2). Declining RNA viral load is associated with resolution of clinical symptoms and gradual increase in antibody titres, for both binding and neutralizing antibodies18,23. However, ongoing detection of viral RNA has been described for prolonged periods up to 28 dpos in otherwise healthy individuals61, and some studies have reported low-level detection of RNA by RT-PCR even for months62. Participants who continue to shed viral RNA for more than 4 weeks after initial detection by RT-PCR represent a minority of non-severe cases, estimated to be around 3%63, 14%64 or less than 20%65.Fig. 2 Kinetics of RNA viral loads and infectious virus for ancestral SARS-CoV-2 in patients with mild-to-moderate disease. According to different studies, the incubation period for ancestral SARS-CoV-2 was estimated to lie between 4.6 and 6.4 days. On average, symptoms continue to persist for 10 days. RNA can already be detected before the onset of symptoms; RNA levels peak around the onset of symptoms and then gradually decline. Median clearance for RNA viral load is 16 days post-onset of symptoms. Infectious virus titres are highest around symptom onset, and infectious virus can be isolated up to 8 or 10 days post-onset of symptoms. RNA can be detected for prolonged periods by real-time PCR, when infectious virus is no longer detectable, whereas virus detection by antigen-detecting (rapid) diagnostic tests (Ag-RDTs) was shown to be a better correlate for infectiousness. Gradients reflect variability between individuals (lighter shade towards the end of infection shows that viral loads continue to be detected in some but not all individuals). The grey dashed line marks the initial infection, the blue dashed lines mark the PCR-positive period and the red dashed lines mark Ag-RDT positivity. Details of the underlying studies used to generate Fig. 2 can be found in Supplementary Table 1. Infectious virus shedding of the ancestral SARS-CoV-2 strain, as determined by virus isolation in cell culture, was reported to correlate with high RNA viral load in the early acute phase after symptom onset23. Importantly, daily longitudinal sampling of respiratory specimens from individuals with mild disease or asymptomatic infection revealed that infectious virus can already be detected before the onset of symptoms33. Successful infectious virus isolation was reported within the first 8–10 dpos, but culture probability after this time period rapidly declined4,7,23,29,66,67. Studies that assessed infectious virus quantitatively found that infectious virus titres declined over the first 10 dpos25,29. In addition, a reduced chance of virus isolation coincided with the time of seroconversion in hospitalized patients and, as a result, infectious virus could no longer be isolated from seroconverted patients with detectable antibody titres18,68,69. Although similar seroconversion studies performed on mildly symptomatic patients are missing, the number of immunologically naive individuals is declining and this broadly existing underlying immunity makes such an assessment more complex. Most studies on infectious virus shedding in the acute symptomatic period were on immunocompetent patients that had mild-to-moderate disease, representing the majority of COVID-19 cases in the community. Therefore, the assessment of the presence of infectious virus in the URT from those studies was used to define the duration of the period of infectiousness and contributed to best public health practices for isolation and quarantine62,70. Although the pattern of infection is broadly similar in patients with mild and severe disease, key differences do exist. The first week of illness is comparable in terms of RNA viral load between patients with mild and severe disease. However, patients with severe disease have elevated RNA viral loads in the second week of illness, and RNA was detected for prolonged periods71. Moreover, infectious virus was recovered from hospitalized patients for prolonged periods of up to 32 dpos18,72,73; however, the median time from symptom onset to viral clearance in culture was similar to that of patients with mild or moderate disease18,73. Severe COVID-19 is also characterized by high and persistent RNA viral load in the LRT, whereas non-severe cases have similar viral loads in the URT and LRT74. Prolonged detection of viral RNA was also reported in immunocompromised patients; for example, 224 days after the beginning of the infection, virus was still detected in a man infected with HIV, including the detection of subgenomic RNA (sgRNA) indicating active viral replication75. Also, infectious virus was recovered up to 61 dpos in nasopharyngeal swabs collected from immunocompromised patients76, and low RNA viral loads were still detected at 60 dpos in another study77. Infectious virus was isolated from bronchoalveolar fluids from patients receiving chimeric antigen receptor (CAR) T cell therapy up to 28 days after admission to an intensive care unit78. A case report on an immunocompromised patient showed isolation of infectious virus up to 78 dpos79. The reports of infectious virus isolation from severely ill or immunocompromised patients are limited (owing to the low number of patients), so it is difficult to define the proportion of cases with prolonged shedding. The characteristics of viral shedding of other respiratory viruses are outlined in Box 2. Box 2 Shedding of respiratory viruses The dynamics of viral shedding differs between respiratory viruses, which influences their transmission and has an effect on diagnostics and measures applied to contain the outbreaks. SARS-CoV The epidemic of severe acute respiratory syndrome coronavirus (SARS-CoV) started in November 2002 in the Guangdong province of China and rapidly spread outside China. The virus was airborne and could also be spread via droplets of saliva, but is only moderately transmissible among humans188. Only low viral loads were detected in the early symptomatic period, generally peaking in the upper respiratory tract (URT) around 10–14 days post-onset of symptoms (dpos)189,190, and then dropping to low levels at 3–4 weeks post-infection191. In patients infected with SARS-CoV, viral RNA was detectable for a maximum of 8 weeks in samples collected from the URT191 and for 52 days in sputum samples192, whereas infectious virus was isolated up to 28 dpos from stool and respiratory specimens and up to 36 dpos from urine samples191,193. SARS-CoV replicated less efficiently at low temperatures; thus, virus replication was more efficient in the lower respiratory tract (LRT) than in the URT194. Notably, asymptomatic or pre-symptomatic viral shedding and transmission were not recorded for SARS-CoV190,195; the peaks of transmission occurred around 2 and 10 dpos195. As a result, outbreaks were successfully contained through isolation of symptomatic patients infected with SARS-CoV, which reduced onward transmission196. MERS-CoV Middle East respiratory coronavirus (MERS-CoV) was isolated from a patient with pneumonia in Saudi Arabia in 2012 and was shown to be the causative agent of a cluster of severe respiratory tract infections in the Middle East197. The disease caused by MERS-CoV is characterized by a wide range of clinical severities and by predominantly respiratory symptoms, such as acute viral pneumonia, with a high case fatality ratio198. The virus is capable of airborne transmission and has low transmissibility among humans, with a maximum estimated reproduction number below 1 (ref.198). Higher RNA viral loads were detected in the LRT than in the URT. Estimated mean shedding duration is 15.3 days in the URT and 16.3 days in the LRT62. Prolonged PCR positivity and higher RNA viral loads in the URT and LRT were associated with increased disease severity62,199. Viral RNA was also detected in the urine, stool and serum200. One study reported detection of viral RNA in the blood for 34 days and showed that presence of viral RNA in the blood is associated with higher mortality201; however, another study failed to isolate virus from PCR-positive serum samples200. Influenza virus In symptomatic patients, RNA viral loads start to be detectable by real-time PCR 2 days before the onset of symptoms and peak at 1 dpos202. Human challenge trials with influenza A viruses show that viral loads already sharply increase at 1 day post-inoculation, reach a peak at 2 days post-inoculation and become undetectable at 8 days post-inoculation. The mean duration of viral shedding for influenza viruses is 4.8 days, and the maximum duration is between 6 and 7 days203,204. Kinetics of infectious viral titres were similar to the viral load trends detected by real-time PCR for different strains of influenza205. Lower RNA viral loads and shorter infectious viral shedding were noted in asymptomatic patients202. Human respiratory syncytial virus This virus is the most frequent causative agent of LRT infections, leading to morbidity and mortality particularly in young children and older adults206. The virus is transmitted by contact with nasal secretions or large aerosols. Viral loads and symptoms increased simultaneously, reaching a peak at 5.4 days207. In human challenge trials, respiratory syncytial virus titres were detectable for an average of 4.6 days. Viral RNA could be still detected up to 9 dpos, whereas infectious virus titres could be detected from 1 to 8 dpos in adults208 and up to 9 dpos in children209. Viral shedding of SARS-CoV-2 variants Viral evolution of SARS-CoV-2 over time has led to the emergence of numerous variants. Combined with increasing population immunity due to vaccination or natural infection, this has led to a need to reassess our knowledge of viral shedding patterns. The WHO designated variants as variants of concern (VOCs) if they were associated with one or more of the following: elevated transmissibility or a detrimental change in COVID-19 epidemiology; increased virulence or a change in clinical disease presentation; or decreased effectiveness of public health measures or available diagnostics, vaccines or therapeutics80. To date, five VOCs are recognized: Alpha, Beta, Gamma, Delta and Omicron. In contrast to ancestral SARS-CoV-2, VOCs display some differences in evasion from immunity, viral loads, shedding period or even incubation period, resulting in drastically different levels of transmission81–85 (Fig. 3).Fig. 3 Infectious viral load and symptom onset in SARS-CoV-2 Delta and Omicron BA.1 variants of concern. Overall patterns of shedding dynamics are conserved between SARS-CoV-2 variants. In comparison to ancestral SARS-CoV-2, Delta and Omicron BA.1 have shorter incubation periods, estimated as approximately 3.7–4 days for Delta and approximately 3–3.4 days for Omicron BA.1. Higher infectious viral loads were detected in patients infected with Delta than in patients infected with Omicron BA.1 or ancestral SARS-CoV-2. Only a limited number of studies have determined when virus shedding for Delta and Omicron BA.1 ends, so this time point is not well defined. Owing to the low number of studies comparing the end of the infectious period between different SARS-CoV-2 variants of concern, the end point of infectivity is not well defined (shown as a colour gradient). Details of the underlying studies used to generate Fig. 3 can be found in Supplementary Table 2. All VOCs have shown changes in viral load compared with ancestral SARS-CoV-2. One study reported that infection with Alpha leads to approximately tenfold higher RNA viral load and an increased probability of cell culture isolation compared with the ancestral virus86. However, another study did not find a substantial difference in the infectious virus titre between Alpha and ancestral SARS-CoV-2 (ref.33). Delta reportedly led to an even higher increase in RNA viral load: one study reported a 1,000× increase relative to the ancestral virus87, and other studies reported 1.7× (ref.88) or 6.2× higher89 viral load than Alpha. Furthermore, Delta demonstrated elevated probability of cell culture isolation90 and higher infectious virus titres than Alpha91. Although Omicron was shown to be highly transmissible, lower RNA viral loads92, lower cell culture isolation probability93 and lower infectious virus titres25 were observed in patients infected with Omicron BA.1 than in those infected with Delta. Even within the Omicron clade, there are differences between sub-lineages, with infection with Omicron BA.2 leading to higher levels of RNA viral loads and longer time to viral clearance than with Omicron BA.1 (refs.94–96). Similarly, VOCs have shown differences in the duration of viral shedding. Analysis of Ct values in respiratory specimens found that Delta showed longer persistence of viral RNA than ancestral SARS-CoV-2 (ref.97). Another study demonstrated that there was not significant difference in the mean duration of viral RNA presence in Delta and Omicron BA.1 infections92. The duration of infectious virus shedding appears to be similar to that observed with ancestral SARS-CoV-2, with culturable virus obtained at 5 dpos85 and no replication-competent virus isolated beyond 10 dpos in patients infected with Delta and Omicron BA.1 (refs.84,98). It is important to note that pre-existing immunity to SARS-CoV-2, either from infection or vaccination, might influence the duration of infectious virus shedding (alongside immune status and disease severity, as discussed above), which may have driven some of these differences during the course of the pandemic. Influence of age and sex on viral shedding There is some evidence that age-associated and sex-associated differences in innate and adaptive immunity, as well as higher ACE2 expression in adults than in children, result in an increased risk for severe disease in older male patients99–101. Moreover, a few studies have found that age and sex influence viral loads and shedding dynamics. In cases of infection with ancestral SARS-CoV-2, resolution of RNA shedding was faster in participants <18 years of age and slower in participants >50 years of age61. According to one study, viral RNA can be detected for longer times in male patients infected with ancestral SARS-CoV-2 (ref.102), and RNA viral loads were elevated in male patients infected with either Alpha or Delta variants compared with female patients88. However, a possible association of viral load dynamics with age or sex is highly debated, as other studies demonstrated that they have no influence on infectious virus25 or RNA viral loads59. Early studies in ancestral SARS-CoV-2 did not find a difference in virus isolation success103 or RNA viral loads between children and adults104–106, but sample sizes were small. Slightly lower RNA viral loads and a more rapid clearance of viral RNA was observed in children than in adults when analysing much larger cohorts, whereas the patterns of shedding curves over time were similar between children and adults107. Furthermore, large-scale analysis of viral loads across different age groups showed no differences of distribution of RNA viral load between children and adults108 or only slightly lower viral loads (<0.5 log10 units) in children <5 years of age86. Symptoms as a correlate for shedding One of the key epidemiological parameters for SARS-CoV-2 transmission is the incubation period, defined as the time from exposure or infection to the onset of symptoms. Studies on ancestral SARS-CoV-2 have estimated that the incubation period on average is between 4.6 and 6.4 days59,109–111 (Fig. 2). A human challenge trial with ancestral SARS-CoV-2 demonstrated that symptoms start to appear 2–4 days after inoculation, and RNA viral loads reach their peak 4–5 days after inoculation29. Thus, artificial inoculation of the virus confirmed the timing of peak viral loads observed in naturally infected individuals, whereas onset of symptoms was faster in the human challenge cases. In contrast to natural infection, in artificial inoculation, virus-containing drops with high viral load are directly applied in the nose and therefore reach the nasal epithelium more quickly, which might lead to the more rapid appearance of symptoms. For Delta, the estimated incubation period was between 3.7 and 4 days81–83,97, whereas infection with Omicron BA.1 was characterized by an even shorter incubation period of 3–3.4 days83,112,113 (Fig. 3). However, as the time point of infection is rarely known outside of human challenge trials, dpos is most commonly used when analysing viral load and infectious virus. Considering that high viral loads can be detected in the URT of infected individuals regardless of their clinical manifestations, the presence of symptoms is an unreliable indicator of infectiousness. Notably, individuals infected with SARS-CoV-2 can be infectious before the onset of symptoms59, and it was estimated that about half of secondary transmissions take place in the pre-symptomatic phase59,114. Moreover, according to population surveys, asymptomatic cases represent around 40% of all SARS-CoV-2 infections with ancestral SARS-CoV-2 (refs.115–117), and tracing of close contacts of confirmed cases of SARS-CoV-2 found that up to 23% of infections were asymptomatic118. There are conflicting findings regarding viral shedding differences in symptomatic and asymptomatic patients. Comparison of viral loads between symptomatic and asymptomatic patients remains challenging, as the time of exposure cannot be clearly identified in asymptomatic individuals, and dpos cannot be used when comparing viral loads with symptomatic individuals. Furthermore, individuals who do not show clinical symptoms at the time of testing can represent either true asymptomatic individuals or pre-symptomatic individuals who will develop symptoms later. Thus, only well-controlled studies with a follow-up of assessed individuals can make a clear distinction between pre-symptomatic and asymptomatic individuals. A study on ancestral SARS-CoV-2, which followed COVID-19 confirmed cases hospitalized for isolation and recorded symptoms daily, found similar initial Ct values between asymptomatic and symptomatic individuals119. Similarly, no significant difference in RNA viral loads between symptomatic and asymptomatic patients was found in other studies in which patients were followed longitudinally and the presence of symptoms was either monitored by health-care professionals120 or was self-reported115. By contrast, other studies, in which symptoms were also recorded by clinicians, reported lower RNA viral loads in asymptomatic participants121,122. In addition, one study found a faster clearance of viral RNA in asymptomatic than in symptomatic individuals123, and another recorded a longer median duration of viral RNA shedding among asymptomatic patients119. There are limited data regarding the presence of infectious virus in asymptomatic patients. One study showed lower virus isolation success from asymptomatic patients124, but only a small number of patients were included. Therefore, more studies evaluating infectious virus in asymptomatic patients would help to elucidate the differences in their infectivity compared with symptomatic patients. SARS-CoV-2 transmission Viral loads have a key role in the SARS-CoV-2 transmission. As previously discussed, host (role of vaccination or previous infection) and viral factors (SARS-CoV-2 variants) greatly influence viral load dynamics and therefore further influence viral transmission. Influence of viral load on transmission SARS-CoV-2 can be transmitted via larger droplets and aerosols produced when breathing, speaking, sneezing or coughing and to a lesser extend also by contaminated surfaces. As an infection can only be induced by infectious viral particles and not by remnant RNA or protein alone, the presence of infectious SARS-CoV-2 is required for secondary transmission. Although transmission is a multifactorial process that is also influenced, for example, by environmental and behavioural factors (such as humidity, air quality, exposure time or closeness of contact), the viral load of SARS-CoV-2 in the URT is considered to be a proxy for transmission risk. An epidemiological study that included viral load analysis found that viral load of an index case strongly correlates with onward transmission, with higher viral loads for ancestral SARS-CoV-2 presenting a greater secondary attack rate risk125. In this study, viral load was identified as the main driver of transmission, with a more pronounced effect in household settings than in non-household settings (hospitals and nursing homes, among others). Transmission probability peaks around symptom onset, when infectious virus titres are estimated to be the highest during the course of infection. As viral load decreases with time, the probability of transmission also gradually declines in cases of infection with ancestral SARS-CoV-2 (ref.126). On this note, a study of health-care workers infected with ancestral virus documented no transmission from index cases later than 6 dpos, which is in line with findings showing reduced virus isolation success towards the end of week 1 of symptomatic disease127. However, there are limitations when using viral load of an index case as a proxy for transmission. To date, the infectious dose of SARS-CoV-2 required to lead to a secondary transmission is not yet known, and the association between presence of infectious virus in the respiratory tract and infectiousness of the same individuals is poorly understood. In the only available human challenge trial that was conducted with ancestral SARS-CoV-2, an initial infectious dose of 10 TCID50 did not lead to an infection in 16 of 36 participants29. Other factors, such as symptoms, type of contact, protective measures, vaccination status and other host factors may have an additionally strong effect on transmission128–133. Viral load can markedly vary between individuals (as a result of individual susceptibility and of immunity from previous infections or vaccination), which leads to differences in their propensity to transmit the virus. Indeed, differences have been observed in the duration of infectious virus detection and in nasal and oral viral loads for both ancestral SARS-CoV-2 and Alpha33. Inter-individual variability was suggested to have a role in the observed heterogeneity of viral load dynamics, as some early immune signatures were significantly associated with higher oropharyngeal RNA viral loads in patients134. Therefore, observed heterogeneity between individuals has an important role in ongoing viral transmission33. Such differences can lead to heterogeneity in virus transmission. Modelling with ancestral SARS-CoV-2 and Alpha estimated that individuals who are highly infectious, known as superspreaders, shed 57-fold more virus over the course of infection than those with lowest infectiousness33. By contrast, most patients with COVID-19 do not infect other individuals as they expel few to no viral particles from their airways135. Indeed, only a minority (about 8%) of patients positive for SARS-CoV-2 infected with ancestral SARS-CoV-2 or Alpha have significantly higher infectious virus titres than the rest of the population (as shown in a study measuring virus isolation probability in a large cohort of patients)86. Moreover, only 15%114 to 19%136 of individuals that were infected led to 80% of secondary transmissions of ancestral SARS-CoV-2. Similar trends were confirmed for Omicron BA.1 and BA.2, for which only 9%137 to 20%138 of the infectious contacts were responsible for 80% of all transmissions. Superspreading events are therefore characterized by infectious individuals having close contact with a high number of susceptible individuals and by a higher probability of transmission per contact. Aside from biological factors influencing these events, sociobehavioural and environmental factors contribute to the likelihood of superspreading (for example, large indoor gatherings with poor ventilation and no other infection prevention measures). Moreover, particular locations can represent a higher risk of transmission (for example, many superspreading events take place in crowded indoor settings, such as cruise ships, family gatherings, parties, elderly care centres and hospitals)139. The role of pre-existing immunity on viral shedding and transmission All currently licensed SARS-CoV-2 vaccines are administered intramuscularly, leading to a rise in serum antibodies and protection from severe disease and death due to COVID-19, but not to long-term protection from infection140–142. The levels of circulating antibodies generated following vaccination decline over time, but can be elevated by a booster dose143,144. Furthermore, currently available vaccines were developed against the ancestral SARS-CoV-2 strain using the spike protein of the first sequenced virus, and the degree of protection from severe disease against other genetic variants was shown to vary145. Moreover, vaccination leads to limited induction of neutralizing antibodies on mucosal surfaces, which may have a role in mitigating virus replication and prevention of more pronounced disease146,147. For instance, secretory component antibodies, which are specific to mucosal surfaces, were detected in the saliva in 58% of participants 2 weeks post-vaccination with mRNA vaccines in one study, but the levels were significantly lower than in convalescent participants, and their neutralizing capacity significantly decayed 6 months post-vaccination148. A study on a small group of individuals uninfected or infected with Delta demonstrated that mucosal antibody responses induced by vaccination were low or undetectable, but breakthrough infections led to substantial increases of antibody titres in saliva149. However, the role of pre-existing mucosal immunity on infectious virus shedding and the possible correlation between the mucosal antibodies and viral loads in humans has not been elucidated. As a result of waning antibodies and the emergence of VOCs with immune-evading properties, breakthrough infections have been increasingly reported among vaccinated individuals, mainly since the emergence of the Delta and Omicron VOCs. It has been debated whether vaccination with current SARS-CoV-2 vaccines impacts viral load (and therefore shedding) in breakthrough infections. The effect of vaccination on viral load and shedding is therefore of interest as it would mean that vaccination not only protects the vaccinee but can also help to mitigate virus spread by reducing infectious virus titres or shortening infectious shedding periods, thus having an impact beyond protection of the individual. Overall, vaccination has been found to lead to reduced viral load (Fig. 4), although this decreases with time. Vaccination with ChAdOx1 vaccine (the Oxford–AstraZeneca vaccine) or BNT162b2 (the Pfizer/BioNTech vaccine) leads to lower RNA viral loads in individuals infected with Alpha, but the effect was weaker for breakthrough infections with Delta150,151. Immunization with BNT162b2 led to reduced RNA viral loads in Delta breakthrough infections, although this effect declined 2 months after vaccination and ultimately faded 6 months after vaccination152. Immunization with ChAdOx1 vaccine also led to a reduction of RNA viral load in breakthrough infections with Alpha VOC153. Faster clearance of RNA viral loads was detected in the group of vaccinated patients who mostly received mRNA vaccines154,155, and lower probability of isolation of infectious virus from patients vaccinated with mRNA or adenoviral vector vaccines was observed156,157. Even though not all studies could demonstrate a reduction of RNA viral loads in Delta breakthrough infections150,154, infectious virus titres were reported to be lower in individuals vaccinated with mRNA or adenoviral vector vaccines despite similar levels of viral RNA25,93,157. Vaccination was also found to influence infectious virus isolation. Viable virus in cell culture was detected for significantly longer median time periods in unvaccinated patients infected with Delta than in vaccinated patients infected with Delta155,158. However, no significant differences in RNA viral loads were found between unvaccinated, fully vaccinated or boosted patients infected with Omicron BA.1 or BA.2 (refs.93,159), whereas infectious virus titres, measured quantitively at 5 dpos, were lower in Omicron BA.1 breakthrough infections only after a booster dose25. Other studies showed that vaccination status did not influence infectious virus isolation success93 or the time from initial positive PCR assay to culture conversion in patients infected with Omicron BA.1 (ref.85). These studies indicate that triple vaccination reduces infectious viral load but not the time period during which infectious virus can be isolated from Omicron breakthrough infections.Fig. 4 Influence of vaccination on viral load. Similar RNA viral loads were detected in vaccinated and unvaccinated patients infected with the Delta variant of concern during the first 5 days post-onset of symptoms. However, faster clearance of viral RNA was shown in vaccinated patients. Infectious viral loads (IVLs) were significantly lower in vaccinated individuals and declined faster than in unvaccinated individuals infected with Delta. Dynamics of viral loads in vaccinated individuals may vary widely in case of infection with another variant. Details of the underlying studies used to generate Fig. 4 can be found in Supplementary Table 3. There are limited data on the effect of previous infection on viral shedding. A study performed on ancestral SARS-CoV-2 demonstrated lower RNA viral loads among seropositive individuals than among seronegative individuals160. Although higher levels of reinfection with Omicron BA.1 were demonstrated among unvaccinated patients previously infected with other SARS-CoV-2 variants161, there are no relevant data on the effect of previous infections on viral load dynamics. Together, these findings suggest that vaccinated individuals are less infectious than unvaccinated individuals, although the duration of this effect has not been studied systematically. Nevertheless, there are some conflicting data on the effect of vaccination on onward transmission. An epidemiological study performed in the UK found that, despite RNA viral load declining faster among fully vaccinated than unvaccinated patients infected with Delta, the peak RNA viral loads were similar, and the secondary attack rate among household contacts exposed to fully vaccinated or unvaccinated index cases did not differ151. By contrast, data from Israel showed that less Delta transmission took place in households with vaccinated participants than with unvaccinated participants130. Another study from the UK showed that both BNT162b2 and ChAdOx1 vaccines led to the reduction of onward transmission from vaccinated index patients, although a stronger reduction was detected for Alpha than for Delta129, probably owing to the higher viral loads in the case of infection with Delta, as shown previously88,89,129. Finally, another study found that vaccination was associated with reduced onward transmission of Delta breakthrough infection due to shorter duration of viable virus shedding158. Overall, even though the currently used vaccines are still based on the ancestral virus spike protein and elicit mainly a systemic rather than a mucosal immune response, some effect on viral load, infectious virus shedding and transmission has been observed129,130,162. Furthermore, with increasing rates of breakthrough infections in the Omicron waves since the end of 2021, many individuals display hybrid immunity consisting of vaccination combined with one or more natural infections before or after vaccination163,164. It is thought that such hybrid immunity may provide better control of virus replication in the mucosa149,163,165. With the constant emergence of novel variants that can evade existing immunity, our understanding of the effect of vaccination on viral shedding should be constantly updated166. Better understanding of the role of mucosal immunity, and potentially vaccines that elicit local rather than systemic immune responses, are needed to aim for viral load reduction as a means to control SARS-CoV-2 circulation167–169. Influence of SARS-CoV-2 VOCs on transmission There are several possible underlying causes of increased transmissibility of newly emerging variants, which allow VOCs to quickly outcompete previously circulating strains, including increased viral loads, a lower infectious dose required to establish infection and prolonged period of infectiousness170. Furthermore, the immune-evading properties of new variants lead to higher susceptibility of infection for vaccinated and previously infected individuals and result in higher transmissibility, as was observed with Omicron166,171. The rapid emergence of SARS-CoV-2 variants with altered biological properties has shown that knowledge on viral loads, viral kinetics and infectious virus shedding is variant specific, and each emerging variant requires a reassessment. Although understanding of mutational profiles and associated phenotypes of SARS-CoV-2 variants has improved, reasons for enhanced transmissibility are manifold and not all understood yet. To date, shedding characteristics and transmission properties cannot be easily predicted based on sequences. Unlike immune-evasion mechanisms, shedding dynamics, such as kinetics of infectious virus titres or incubation periods of the SARS-CoV-2 variants, cannot be predicted from specific mutation patterns. With a still highly dynamic situation in terms of viral evolution of SARS-CoV-2, understanding viral kinetics and their effect on transmission remains of high public health interest. SARS-CoV-2 diagnostics in public health Our ability to define the presence of infectious virus is key to guiding public health measures, as it will enable the isolation of infectious individuals to limit secondary transmission. Unfortunately, no point-of-care diagnostic test currently exists to determine infectious SARS-CoV-2 in a patient sample172, and virus culture as described above is not suited for diagnostic purposes. Thus, a range of approaches have been suggested to find a proxy for infectiousness to guide isolation periods. One example is the detection of sgRNA transcripts, which are generated during virus replication, and specifically the synthesis of negative-strand RNA. Although sgRNAs are transcribed in infected cells, they are not packaged in the virions and can therefore serve as an indicator of active replication and thus of infectious virus. Specific RT-PCR assays were developed to detect sgRNAs in addition to the diagnostic detection of genomic SARS-CoV-2 RNA, but such assays have not made their way into routine diagnostic use owing to their lower sensitivity than conventional RT-PCR assays. Some studies found that detection of sgRNA correlates with detection of infectious virus4,173,174, and that sgRNA was rarely detectable 8 dpos67. However, sgRNA was detected in diagnostic samples up to 17 days after initial detection of infection175 or in culture-negative samples176, probably owing to the stability and nuclease resistance of double-membrane vesicles containing sgRNAs. Thus, although the absence of sgRNA would indicate absence of viral replication, the presence of sgRNA does not necessarily indicate infectiousness19. Ct values have also been used as a proxy for infectiousness, as described above. However, as already discussed, low-quality specimens resulting from technical mistakes during the collection process can falsely indicate an absence of infectious virus. Furthermore, owing to the quick increase of RNA viral load at the beginning of the infection, a low viral load, especially in the absence of symptoms or in the early symptomatic period, does not preclude that an individual will not soon enter the infectious period with the highest transmission risk. At such a period, viral loads reach their peak levels, causing the majority of transmission events59,126. Even though Ag-RDTs are less sensitive than RT-PCR, they are less expensive, can be performed outside of laboratory settings and give faster results, and so are useful tools to guide isolation and limit transmission177. RT-PCR tests have a limit of detection of 102–103 genome copies per millilitre, whereas Ag-RDTs have a limit of detection corresponding to 104–106 genome copies per millilitre177–180. Infectious individuals typically have RNA viral loads of >106 genome copies per millilitre, which corresponds largely with a Ct of 25 in most RT-PCR assays4, indicating that Ag-RDT is a good proxy for infectiousness177. However, the obvious limitations of Ag-RDT, such as lower sensitivity of infectious virus detection towards the end of infection47,52, should not be neglected. Ag-RDTs have also shown variation in their sensitivity and specificity for detection of SARS-CoV-2 VOCs53,54, which is a challenge as new variants emerge. Overall, all of the currently available diagnostic methods have certain limitations for detection of infectious virus. However, even if these tests serve only as imperfect tools when used as proxies for infectiousness, their implementation as part of a public health strategy is not intended to prevent every single infection, but rather to reduce the number of infectious people in the community and thus to decrease the number of secondary transmissions. Conclusions Entering the third year of the pandemic, much knowledge on SARS-CoV-2 viral loads, infectious virus shedding and windows of infectiousness has been gained, although emerging SARS-CoV-2 variants and an increasing population immunity add more complexity to the situation. Although much progress has been made during the pandemic in the field of diagnostics, to date, no diagnostic tests exist that reliably determine the presence of infectious virus. Continuing evaluation of viral-shedding characteristics under these changing circumstances and understanding the biological properties of novel SARS-CoV-2 variants when it comes to viral shedding remain of importance to guide public health practices. Supplementary information Supplementary Information Supplementary information The online version contains supplementary material available at 10.1038/s41579-022-00822-w. Acknowledgements The authors thank E. Boehm for help with literature search and proofreading. The work was funded by the COVID-19 National Research Program (grant number 198412) of the Swiss National Science Foundation. Author contributions O.P. and I.E. wrote the manuscript. B.M. created the figures. All authors contributed to the discussion of the content, and reviewed and edited the manuscript before submission. Peer review Peer review information Nature Reviews Microbiology thanks Nancy Leung, Quanyi Wang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Competing interests The authors declare no competing interests. Glossary Chimeric antigen receptor (CAR) T cell therapy A way to treat cancer by using T cells expressing genetically engineered receptors to target cancer cells. Cycle threshold (Ct) value The number of amplifications required for a target gene to cross the threshold determined by real-time PCR. Arbitrary test-specific Ct values inversely correlate with viral load. Focus-forming assays Assays that count the number of ‘foci’, defined as a cluster of adjacent cells expressing viral antigen stained by a specific antibody. Immunostaining A method for the detection of specific proteins in individual cells or tissues using antibodies. In the case of SARS-CoV-2, anti-nucleocapsid antibodies are used to detect virus in infected cells. Index case The infected individual who is triggering an outbreak or a cluster by transmitting an infectious agent to others. There might be multiple index cases in an outbreak or epidemiological study. Plaque assays Assays that quantify the number of infectious virions by counting plaques in a cell monolayer that correspond to single infectious particles. Secondary attack rate The probability that an infection spreads from an index case to susceptible people in a specific setting (usually, a household or close contacts). The term is used to evaluate the risk of onward transmission of pathogen within a population. Seroconversion The development of specific antibodies in the serum as a consequence of immunization by natural infection or vaccination. sgRNA Subgenomic RNA fragments that occur during viral replication. TCID50 A measurement of the presence of cytopathic effects in cells upon infection with serial dilutions of virus specimens, which indicates the dose needed to induce a cytopathic effect in 50% of the inoculated wells. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Puelles VG Multiorgan and renal tropism of SARS-CoV-2 N. Engl. J. Med. 2020 383 590 592 10.1056/NEJMc2011400 32402155 2. Lamers MM Haagmans BL SARS-CoV-2 pathogenesis Nat. Rev. Microbiol. 2022 20 270 284 10.1038/s41579-022-00713-0 35354968 3. Peng L SARS-CoV-2 can be detected in urine, blood, anal swabs, and oropharyngeal swabs specimens J. Med. Virol. 2020 92 1676 1680 10.1002/jmv.25936 32330305 4. Wölfel R Virological assessment of hospitalized patients with COVID-2019 Nature 2020 581 465 469 10.1038/s41586-020-2196-x 32235945 5. Zhang W Molecular and serological investigation of 2019-nCoV infected patients: implication of multiple shedding routes Emerg. Microbes Infect. 2020 9 386 389 10.1080/22221751.2020.1729071 32065057 6. Pérez-Bartolomé F Sánchez-Quirós J Ocular manifestations of SARS-CoV-2: literature review Arch. Soc. Esp. Oftalmol. 2021 96 32 40 10.1016/j.oftal.2020.07.020 32873480 7. Vetter P Daily viral kinetics and innate and adaptive immune response assessment in COVID-19: a case series mSphere 2020 10.1128/mSphere.00827-20 33177214 8. Jeong HW Viable SARS-CoV-2 in various specimens from COVID-19 patients Clin. Microbiol. Infect. 2020 26 1520 1524 10.1016/j.cmi.2020.07.020 32711057 9. Cerrada-Romero C Excretion and viability of SARS-CoV-2 in feces and its association with the clinical outcome of COVID-19 Sci. Rep. 2022 12 7397 10.1038/s41598-022-11439-7 35513481 10. Dergham J Delerce J Isolation of viable SARS-CoV-2 virus from feces of an immunocompromised patient suggesting a possible fecal mode of transmission J. Clin. Med. 2021 10.3390/jcm10122696 34207314 11. Xiao F Infectious SARS-CoV-2 in feces of patient with severe COVID-19 Emerg. Infect. Dis. 2020 26 1920 1922 10.3201/eid2608.200681 32421494 12. Sun J Isolation of infectious SARS-CoV-2 from urine of a COVID-19 patient Emerg. Microbes Infect. 2020 9 991 993 10.1080/22221751.2020.1760144 32342724 13. Colavita F SARS-CoV-2 isolation from ocular secretions of a patient with COVID-19 in Italy with prolonged viral RNA detection Ann. Intern. Med. 2020 173 242 243 10.7326/M20-1176 32302380 14. Matsuyama S Enhanced isolation of SARS-CoV-2 by TMPRSS2-expressing cells Proc. Natl Acad. Sci. USA 2020 117 7001 7003 10.1073/pnas.2002589117 32165541 15. Case JB Bailey AL Kim AS Chen RE Diamond MS Growth, detection, quantification, and inactivation of SARS-CoV-2 Virology 2020 548 39 48 10.1016/j.virol.2020.05.015 32838945 16. Baggen J Vanstreels E Jansen S Daelemans D Cellular host factors for SARS-CoV-2 infection Nat. Microbiol. 2021 6 1219 1232 10.1038/s41564-021-00958-0 34471255 17. Chu H Comparative tropism, replication kinetics, and cell damage profiling of SARS-CoV-2 and SARS-CoV with implications for clinical manifestations, transmissibility, and laboratory studies of COVID-19: an observational study Lancet Microbe 2020 1 e14 e23 10.1016/S2666-5247(20)30004-5 32835326 18. van Kampen JJA Duration and key determinants of infectious virus shedding in hospitalized patients with coronavirus disease-2019 (COVID-19) Nat. Commun. 2021 12 267 10.1038/s41467-020-20568-4 33431879 19. Bruce EA Predicting infectivity: comparing four PCR-based assays to detect culturable SARS-CoV-2 in clinical samples EMBO Mol. Med. 2022 14 e15290 10.15252/emmm.202115290 34862752 20. Essaidi-Laziosi M Perez Rodriguez FJ Estimating clinical SARS-CoV-2 infectiousness in Vero E6 and primary airway epithelial cells Lancet Microbe 2021 2 e571 10.1016/S2666-5247(21)00216-0 34485956 21. Liu R Positive rate of RT-PCR detection of SARS-CoV-2 infection in 4880 cases from one hospital in Wuhan, China, from Jan to Feb 2020 Clin. Chim. Acta 2020 505 172 175 10.1016/j.cca.2020.03.009 32156607 22. Corman VM Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR Eur. Surveill. 2020 10.2807/1560-7917.ES.2020.25.3.2000045 23. Bullard J Predicting infectious severe acute respiratory syndrome coronavirus 2 from diagnostic samples Clin. Infect. Dis. 2020 71 2663 2666 10.1093/cid/ciaa638 32442256 24. Jefferson T Spencer EA Brassey J Heneghan C Viral cultures for coronavirus disease 2019 infectivity assessment: a systematic review Clin. Infect. Dis. 2021 73 e3884 e3899 10.1093/cid/ciaa1764 33270107 25. Puhach O Infectious viral load in unvaccinated and vaccinated individuals infected with ancestral, Delta or Omicron SARS-CoV-2 Nat. Med. 2022 28 1491 1500 10.1038/s41591-022-01816-0 35395151 26. van Kasteren PB Comparison of seven commercial RT-PCR diagnostic kits for COVID-19 J. Clin. Virol. 2020 128 104412 10.1016/j.jcv.2020.104412 32416600 27. Bentley, E. et al. Collaborative study for the establishment of a WHO international standard for SARS-CoV-2 RNA (WHO, 2020). 28. Zou L SARS-CoV-2 viral load in upper respiratory specimens of infected patients N. Engl. J. Med. 2020 382 1177 1179 10.1056/NEJMc2001737 32074444 29. Killingley B Safety, tolerability and viral kinetics during SARS-CoV-2 human challenge in young adults Nat. Med. 2022 28 1031 1041 10.1038/s41591-022-01780-9 35361992 30. Yu F Quantitative detection and viral load analysis of SARS-CoV-2 in infected patients Clin. Infect. Dis. 2020 71 793 798 10.1093/cid/ciaa345 32221523 31. European Centre for Disease Prevention and Control & World Health Organization Regional Office for Europe. Methods for the detection and identification of SARS-CoV-2 variants: second update, August 2022 (WHO, 2022). 32. Ke R Longitudinal analysis of SARS-CoV-2 vaccine breakthrough infections reveals limited infectious virus shedding and restricted tissue distribution Open Forum Infect. Dis. 2022 9 ofac192 10.1093/ofid/ofac192 35791353 33. Ke R Daily longitudinal sampling of SARS-CoV-2 infection reveals substantial heterogeneity in infectiousness Nat. Microbiol. 2022 7 640 652 10.1038/s41564-022-01105-z 35484231 34. Pekosz A Antigen-based testing but not real-time polymerase chain reaction correlates with severe acute respiratory syndrome coronavirus 2 viral culture Clin. Infect. Dis. 2021 73 e2861 e2866 10.1093/cid/ciaa1706 33479756 35. Monel B Release of infectious virus and cytokines in nasopharyngeal swabs from individuals infected with non-Alpha or Alpha SARS-CoV-2 variants: an observational retrospective study EBioMedicine 2021 73 103637 10.1016/j.ebiom.2021.103637 34678613 36. Kirby JE SARS-CoV-2 antigen tests predict infectivity based on viral culture: comparison of antigen, PCR viral load, and viral culture testing on a large sample cohort Clin. Microbiol. Infect. 2022 10.1016/j.cmi.2022.07.010 35863629 37. Pickering S Comparative performance of SARS-CoV-2 lateral flow antigen tests and association with detection of infectious virus in clinical specimens: a single-centre laboratory evaluation study Lancet Microbe 2021 2 e461 e471 10.1016/S2666-5247(21)00143-9 34226893 38. Tariq M Viable severe acute respiratory syndrome coronavirus 2 isolates exhibit higher correlation with rapid antigen assays than subgenomic RNA or genomic RNA Front. Microbiol. 2021 12 718497 10.3389/fmicb.2021.718497 34867844 39. Chu VT Comparison of home antigen testing with RT-PCR and viral culture during the course of SARS-CoV-2 infection JAMA Intern. Med. 2022 182 701 709 10.1001/jamainternmed.2022.1827 35486394 40. Albert E Field evaluation of a rapid antigen test (Panbio™ COVID-19 Ag rapid test device) for COVID-19 diagnosis in primary healthcare centres Clin. Microbiol. Infect. 2021 27 472.e7 472.e10 10.1016/j.cmi.2020.11.004 33189872 41. Ford L Epidemiologic characteristics associated with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) antigen-based test results, real-time reverse transcription polymerase chain reaction (rRT-PCR) cycle threshold values, subgenomic RNA, and viral culture results from university testing Clin. Infect. Dis. 2021 73 e1348 e1355 10.1093/cid/ciab303 33846714 42. Berger A Diagnostic accuracy of two commercial SARS-CoV-2 antigen-detecting rapid tests at the point of care in community-based testing centers PLoS ONE 2021 16 e0248921 10.1371/journal.pone.0248921 33788882 43. Brümmer LE Accuracy of novel antigen rapid diagnostics for SARS-CoV-2: a living systematic review and meta-analysis PLoS Med. 2021 18 e1003735 10.1371/journal.pmed.1003735 34383750 44. Ngo Nsoga MT Diagnostic accuracy of Panbio rapid antigen tests on oropharyngeal swabs for detection of SARS-CoV-2 PLoS ONE 2021 16 e0253321 10.1371/journal.pone.0253321 34166410 45. Korenkov M Evaluation of a rapid antigen test to detect SARS-CoV-2 infection and identify potentially infectious individuals J. Clin. Microbiol. 2021 59 e0089621 10.1128/JCM.00896-21 34213977 46. Yamayoshi S Comparison of rapid antigen tests for COVID-19 Viruses 2020 12 1420 10.3390/v12121420 33322035 47. McKay SL Performance evaluation of serial SARS-CoV-2 rapid antigen testing during a nursing home outbreak Ann. Intern. Med. 2021 174 945 951 10.7326/M21-0422 33900791 48. Nordgren J SARS-CoV-2 rapid antigen test: high sensitivity to detect infectious virus J. Clin. Virol. 2021 140 104846 10.1016/j.jcv.2021.104846 33971580 49. Fernandez-Montero A Argemi J Validation of a rapid antigen test as a screening tool for SARS-CoV-2 infection in asymptomatic populations. Sensitivity, specificity and predictive values EClinicalMedicine 2021 37 100954 10.1016/j.eclinm.2021.100954 34127960 50. Currie DW Relationship of SARS-CoV-2 antigen and reverse transcription PCR positivity for viral cultures Emerg. Infect. Dis. 2022 28 717 720 10.3201/eid2803.211747 35202532 51. Corman VM Comparison of seven commercial SARS-CoV-2 rapid point-of-care antigen tests: a single-centre laboratory evaluation study Lancet Microbe 2021 2 e311 e319 10.1016/S2666-5247(21)00056-2 33846704 52. Lopera TJ Alzate-Ángel JC The usefulness of antigen testing in predicting contagiousness in COVID-19 Microbiol. Spectr. 2022 10 e0196221 10.1128/spectrum.01962-21 35348350 53. Osterman A Impaired detection of Omicron by SARS-CoV-2 rapid antigen tests Med. Microbiol. Immunol. 2022 211 105 117 10.1007/s00430-022-00730-z 35187580 54. Galliez RM Evaluation of the Panbio COVID-19 antigen rapid diagnostic test in subjects infected with Omicron using different specimens Microbiol. Spectr. 2022 10 e0125022 10.1128/spectrum.01250-22 35652635 55. Raïch-Regué D Performance of SARS-CoV-2 antigen-detecting rapid diagnostic tests for Omicron and other variants of concern Front. Microbiol. 2022 13 810576 10.3389/fmicb.2022.810576 35620108 56. van Ogtrop ML van de Laar TJW Eggink D Vanhommerig JW van der Reijden WA Comparison of the performance of the Panbio COVID-19 antigen test in SARS-CoV-2 B.1.1.7 (Alpha) variants versus non-B.1.1.7 variants Microbiol. Spectr. 2021 9 e0088421 10.1128/Spectrum.00884-21 34817226 57. Lindner AK Head-to-head comparison of SARS-CoV-2 antigen-detecting rapid test with self-collected nasal swab versus professional-collected nasopharyngeal swab Eur. Respir. J. 2021 57 2003961 10.1183/13993003.03961-2020 33303544 58. Meiners, L. & Horn, J. SARS-CoV-2 rapid antigen test sensitivity and viral load in freshly symptomatic hospital employees, December 2020 to February 2022. Preprint at 10.2139/ssrn.4099425 (2022). 59. He X Temporal dynamics in viral shedding and transmissibility of COVID-19 Nat. Med. 2020 26 672 675 10.1038/s41591-020-0869-5 32296168 60. Néant N Modeling SARS-CoV-2 viral kinetics and association with mortality in hospitalized patients from the French COVID cohort Proc. Natl Acad. Sci. USA 2021 118 e2017962118 10.1073/pnas.2017962118 33536313 61. Owusu D Persistent SARS-CoV-2 RNA shedding without evidence of infectiousness: a cohort study of individuals with COVID-19 J. Infect. Dis. 2021 224 1362 1371 10.1093/infdis/jiab107 33649773 62. Cevik M SARS-CoV-2, SARS-CoV, and MERS-CoV viral load dynamics, duration of viral shedding, and infectiousness: a systematic review and meta-analysis Lancet Microbe 2021 2 e13 e22 10.1016/S2666-5247(20)30172-5 33521734 63. Chen X Associations of clinical characteristics and treatment regimens with the duration of viral RNA shedding in patients with COVID-19 Int. J. Infect. Dis. 2020 98 252 260 10.1016/j.ijid.2020.06.091 32619760 64. Kim SM Hwang YJ Prolonged SARS-CoV-2 detection and reversed RT-PCR results in mild or asymptomatic patients Infect. Dis. 2021 53 31 37 10.1080/23744235.2020.1820076 65. Talmy T Tsur A Duration of SARS-CoV-2 detection in Israel Defense Forces soldiers with mild COVID-19 J. Med. Virol. 2021 93 608 610 10.1002/jmv.26374 33196120 66. Singanayagam, A. et al. Duration of infectiousness and correlation with RT-PCR cycle threshold values in cases of COVID-19, England, January to May 2020. Euro Surveill. 10.2807/1560-7917.ES.2020.25.32.2001483 (2020). 67. Perera R SARS-CoV-2 virus culture and subgenomic RNA for respiratory specimens from patients with mild coronavirus disease Emerg. Infect. Dis. 2020 26 2701 2704 10.3201/eid2611.203219 32749957 68. Killerby ME Shedding of culturable virus, seroconversion, and 6-month follow-up antibody responses in the first 14 confirmed cases of coronavirus disease 2019 in the United States J. Infect. Dis. 2021 224 771 776 10.1093/infdis/jiab125 33693830 69. Glans H Shedding of infectious SARS-CoV-2 by hospitalized COVID-19 patients in relation to serum antibody responses BMC Infect. Dis. 2021 21 494 10.1186/s12879-021-06202-8 34044758 70. Badu K SARS-CoV-2 viral shedding and transmission dynamics: implications of WHO COVID-19 discharge guidelines Front. Med. 2021 8 648660 10.3389/fmed.2021.648660 71. Munker D Dynamics of SARS-CoV-2 shedding in the respiratory tract depends on the severity of disease in COVID-19 patients Eur. Respir. J. 2021 58, 2002724 10.1183/13993003.02724-2020 33602859 72. Folgueira MD Luczkowiak J Prolonged SARS-CoV-2 cell culture replication in respiratory samples from patients with severe COVID-19 Clin. Microbiol. Infect. 2021 27 886 891 10.1016/j.cmi.2021.02.014 33631334 73. Kim MC Duration of culturable SARS-CoV-2 in hospitalized patients with Covid-19 N. Engl. J. Med. 2021 384 671 673 10.1056/NEJMc2027040 33503337 74. Chen PZ SARS-CoV-2 shedding dynamics across the respiratory tract, sex, and disease severity for adult and pediatric COVID-19 eLife 2021 10.7554/eLife 34939931 75. Cunha MDP Atypical prolonged viral shedding with intra-host SARS-CoV-2 evolution in a mildly affected symptomatic patient Front. Med. 2021 8 760170 10.3389/fmed.2021.760170 76. Aydillo T Shedding of viable SARS-CoV-2 after immunosuppressive therapy for cancer N. Engl. J. Med. 2020 383 2586 2588 10.1056/NEJMc2031670 33259154 77. Caillard S Benotmane I Gautier Vargas G Perrin P Fafi-Kremer S SARS-CoV-2 viral dynamics in immunocompromised patients Am. J. Transpl. 2021 21 1667 1669 10.1111/ajt.16353 78. Roedl K Viral dynamics of SARS-CoV-2 in critically ill allogeneic hematopoietic stem cell transplant recipients and immunocompetent patients with COVID-19 Am. J. Respir. Crit. Care Med. 2021 203 242 245 10.1164/rccm.202009-3386LE 33253054 79. Leung WF COVID-19 in an immunocompromised host: persistent shedding of viable SARS-CoV-2 and emergence of multiple mutations: a case report Int. J. Infect. Dis. 2022 114 178 182 10.1016/j.ijid.2021.10.045 34757008 80. European Centre for Disease Prevention and Control. Rapid risk assessment: assessing SARS-CoV-2 circulation, variants of concern, non-pharmaceutical interventions and vaccine rollout in the EU/EEA, 15th update (European Centre for Disease Prevention and Control, 2021). 81. Grant R Impact of SARS-CoV-2 Delta variant on incubation, transmission settings and vaccine effectiveness: results from a nationwide case–control study in France Lancet Reg. Health Eur. 2022 13 100278 10.1016/j.lanepe.2021.100278 34849500 82. Ogata T Tanaka H Shorter incubation period among unvaccinated Delta variant coronavirus disease 2019 patients in Japan Int. J. Environ. Res. Public Health 2022 10.3390/ijerph19031127 36231587 83. Backer JA Shorter serial intervals in SARS-CoV-2 cases with Omicron BA.1 variant compared with Delta variant, the Netherlands, 13 to 26 December 2021 Eur. Surveill. 2022 10.2807/1560-7917.ES.2022.27.6.2200042 84. Takahashi K Duration of infectious virus shedding by SARS-CoV-2 Omicron variant-infected vaccinees Emerg. Infect. Dis. 2022 28 998 1001 10.3201/eid2805.220197 35290176 85. Boucau J Duration of shedding of culturable virus in SARS-CoV-2 Omicron (BA.1) infection N. Engl. J. Med. 2022 387 275 277 10.1056/NEJMc2202092 35767428 86. Jones TC Estimating infectiousness throughout SARS-CoV-2 infection course Science 2021 10.1126/science.abi5273 34855491 87. Li B Viral infection and transmission in a large, well-traced outbreak caused by the SARS-CoV-2 Delta variant Nat. Commun. 2022 13 460 10.1038/s41467-022-28089-y 35075154 88. Bolze A SARS-CoV-2 variant Delta rapidly displaced variant Alpha in the United States and led to higher viral loads Cell Rep. Med. 2022 3 100564 10.1016/j.xcrm.2022.100564 35474739 89. Earnest R Comparative transmissibility of SARS-CoV-2 variants Delta and Alpha in New England, USA Cell Rep. Med. 2022 3 100583 10.1016/j.xcrm.2022.100583 35480627 90. Luo CH Infection with the SARS-CoV-2 Delta variant is associated with higher recovery of infectious virus compared to the Alpha variant in both unvaccinated and vaccinated individuals Clin. Infect. Dis. 2021 75 e715 e725 10.1093/cid/ciab986 91. Despres HW Measuring infectious SARS-CoV-2 in clinical samples reveals a higher viral titer: RNA ratio for Delta and Epsilon vs. Alpha variants Proc. Natl Acad. Sci. USA 2022 10.1073/pnas.2116518119 35058348 92. Hay JA Quantifying the impact of immune history and variant on SARS-CoV-2 viral kinetics and infection rebound: a retrospective cohort study eLife 2022 11 e81849 10.7554/eLife.81849 36383192 93. Fall A The displacement of the SARS-CoV-2 variant Delta with Omicron: an investigation of hospital admissions and upper respiratory viral loads EbioMedicine 2022 10.1016/j.ebiom.2022.104008 35460989 94. Lentini A Pereira A Monitoring of the SARS-CoV-2 Omicron BA.1/BA.2 lineage transition in the Swedish population reveals increased viral RNA levels in BA.2 cases Med 2022 3 636 664 10.1016/j.medj.2022.07.007 35981541 95. Qassim SH Effects of BA.1/BA.2 subvariant, vaccination, and prior infection on infectiousness of SARS-CoV-2 Omicron infections J. Travel Med. 2022 29 taac068 10.1093/jtm/taac068 35639932 96. Marking U Correlates of protection, viral load trajectories and symptoms in BA.1, BA.1.1 and BA.2 breakthrough infections in triple vaccinated healthcare workers medRxiv 2022 10.1101/2022.04.02.22273333 97. Wang Y Transmission, viral kinetics and clinical characteristics of the emergent SARS-CoV-2 Delta VOC in Guangzhou, China EClinicalMedicine 2021 40 101129 10.1016/j.eclinm.2021.101129 34541481 98. Siedner MJ Duration of viral shedding and culture positivity with postvaccination SARS-CoV-2 Delta variant infections JCI Insight 2022 10.1172/jci.insight.155483 36214224 99. Pierce CA Immune responses to SARS-CoV-2 infection in hospitalized pediatric and adult patients Sci. Transl Med. 2020 12 eabd5487 10.1126/scitranslmed.abd5487 32958614 100. Takahashi T Sex differences in immune responses that underlie COVID-19 disease outcomes Nature 2020 588 315 320 10.1038/s41586-020-2700-3 32846427 101. Bunyavanich S Do A Vicencio A Nasal gene expression of angiotensin-converting enzyme 2 in children and adults JAMA 2020 323 2427 2429 10.1001/jama.2020.8707 32432657 102. Zheng S Viral load dynamics and disease severity in patients infected with SARS-CoV-2 in Zhejiang province, China, January–March 2020: retrospective cohort study BMJ 2020 369 m1443 10.1136/bmj.m1443 32317267 103. L’Huillier AG Torriani G Culture-competent SARS-CoV-2 in nasopharynx of symptomatic neonates, children, and adolescents Emerg. Infect. Dis. 2020 26 2494 2497 10.3201/eid2610.202403 32603290 104. Baggio S Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) viral load in the upper respiratory tract of children and adults with early acute coronavirus disease 2019 (COVID-19) Clin. Infect. Dis. 2021 73 148 150 10.1093/cid/ciaa1157 32761228 105. Chung E Comparison of symptoms and RNA levels in children and adults with SARS-CoV-2 infection in the community setting JAMA Pediatr. 2021 175 e212025 10.1001/jamapediatrics.2021.2025 34115094 106. Han MS Clinical characteristics and viral RNA detection in children with coronavirus disease 2019 in the Republic of Korea JAMA Pediatr. 2021 175 73 80 10.1001/jamapediatrics.2020.3988 32857112 107. Bellon M Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) viral load kinetics in symptomatic children, adolescents, and adults Clin. Infect. Dis. 2021 73 e1384 e1386 10.1093/cid/ciab396 33949655 108. Madera S Nasopharyngeal SARS-CoV-2 viral loads in young children do not differ significantly from those in older children and adults Sci. Rep. 2021 11 3044 10.1038/s41598-021-81934-w 33542262 109. Lauer SA The incubation period of coronavirus disease 2019 (COVID-19) from publicly reported confirmed cases: estimation and application Ann. Intern. Med. 2020 172 577 582 10.7326/M20-0504 32150748 110. Elias C Sekri A Leblanc P Cucherat M Vanhems P The incubation period of COVID-19: a meta-analysis Int. J. Infect. Dis. 2021 104 708 710 10.1016/j.ijid.2021.01.069 33548553 111. Dhouib W The incubation period during the pandemic of COVID-19: a systematic review and meta-analysis Syst. Rev. 2021 10 101 10.1186/s13643-021-01648-y 33832511 112. Brandal LT Outbreak caused by the SARS-CoV-2 Omicron variant in Norway, November to December 2021 Eur. Surveill. 2021 26 2101147 10.2807/1560-7917.ES.2021.26.50.2101147 113. Wu Y Incubation period of COVID-19 caused by unique SARS-CoV-2 strains: a systematic review and meta-analysis JAMA Netw. Open 2022 5 e2228008 10.1001/jamanetworkopen.2022.28008 35994285 114. Sun K Transmission heterogeneities, kinetics, and controllability of SARS-CoV-2 Science 2021 371 eabe2424 10.1126/science.abe2424 33234698 115. Lavezzo E Suppression of a SARS-CoV-2 outbreak in the Italian municipality of Vo’ Nature 2020 584 425 429 10.1038/s41586-020-2488-1 32604404 116. Gudbjartsson DF Spread of SARS-CoV-2 in the Icelandic population N. Engl. J. Med. 2020 382 2302 2315 10.1056/NEJMoa2006100 32289214 117. Glenet M Asymptomatic COVID-19 adult outpatients identified as significant viable SARS-CoV-2 shedders Sci. Rep. 2021 11 20615 10.1038/s41598-021-00142-8 34663858 118. Wang Y Characterization of an asymptomatic cohort of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infected individuals outside of Wuhan, China Clin. Infect. Dis. 2020 71 2132 2138 10.1093/cid/ciaa629 32442265 119. Long QX Clinical and immunological assessment of asymptomatic SARS-CoV-2 infections Nat. Med. 2020 26 1200 1204 10.1038/s41591-020-0965-6 32555424 120. Lee S Clinical course and molecular viral shedding among asymptomatic and symptomatic patients with SARS-CoV-2 infection in a community treatment center in the Republic of Korea JAMA Intern. Med. 2020 180 1447 1452 10.1001/jamainternmed.2020.3862 32780793 121. Zhou R Viral dynamics in asymptomatic patients with COVID-19 Int. J. Infect. Dis. 2020 96 288 290 10.1016/j.ijid.2020.05.030 32437933 122. Hall SM Comparison of anterior nares CT values in asymptomatic and symptomatic individuals diagnosed with SARS-CoV-2 in a university screening program PLoS ONE 2022 17 e0270694 10.1371/journal.pone.0270694 35830378 123. Kissler SM Viral dynamics of acute SARS-CoV-2 infection and applications to diagnostic and public health strategies PLoS Biol. 2021 19 e3001333 10.1371/journal.pbio.3001333 34252080 124. Tallmadge RL Viral RNA load and infectivity of SARS-CoV-2 in paired respiratory and oral specimens from symptomatic, asymptomatic, or postsymptomatic individuals Microbiol. Spectr. 2022 10 e0226421 10.1128/spectrum.02264-21 35575498 125. Marks M Transmission of COVID-19 in 282 clusters in Catalonia, Spain: a cohort study Lancet Infect. Dis. 2021 21 629 636 10.1016/S1473-3099(20)30985-3 33545090 126. Marc A Quantifying the relationship between SARS-CoV-2 viral load and infectiousness eLife 2021 10.7554/eLife.69302 34569939 127. Cheng HY Contact tracing assessment of COVID-19 transmission dynamics in Taiwan and risk at different exposure periods before and after symptom onset JAMA Intern. Med. 2020 180 1156 1163 10.1001/jamainternmed.2020.2020 32356867 128. Mugglestone MA Presymptomatic, asymptomatic and post-symptomatic transmission of SARS-CoV-2: joint British Infection Association (BIA), Healthcare Infection Society (HIS), Infection Prevention Society (IPS) and Royal College of Pathologists (RCPath) guidance BMC Infect. Dis. 2022 22 453 10.1186/s12879-022-07440-0 35549902 129. Eyre DW Effect of Covid-19 vaccination on transmission of Alpha and Delta variants N. Engl. J. Med. 2022 386 744 756 10.1056/NEJMoa2116597 34986294 130. Prunas O Vaccination with BNT162b2 reduces transmission of SARS-CoV-2 to household contacts in Israel Science 2022 375 1151 1154 10.1126/science.abl4292 35084937 131. Lyngse FP Household transmission of SARS-CoV-2 Omicron variant of concern subvariants BA.1 and BA.2 in Denmark Nat. Commun. 2022 13 5760 10.1038/s41467-022-33498-0 36180438 132. Leung NHL Transmissibility and transmission of respiratory viruses Nat. Rev. Microbiol. 2021 19 528 545 10.1038/s41579-021-00535-6 33753932 133. Alihsan B The efficacy of facemasks in the prevention of COVID-19: a systematic review medRxiv 2022 10.1101/2022.07.28.22278153 134. Hu Z Early immune markers of clinical, virological, and immunological outcomes in patients with COVID-19: a multi-omics study Elife 2022 11 e77943 10.7554/eLife.77943 36239699 135. Chen PZ Heterogeneity in transmissibility and shedding SARS-CoV-2 via droplets and aerosols eLife 2021 10.7554/eLife.65774 34939931 136. Adam DC Clustering and superspreading potential of SARS-CoV-2 infections in Hong Kong Nat. Med. 2020 26 1714 1719 10.1038/s41591-020-1092-0 32943787 137. Guo Z Superspreading potential of infection seeded by the SARS-CoV-2 Omicron BA.1 variant in South Korea J. Infect. 2022 85 e77 e79 10.1016/j.jinf.2022.05.041 35659549 138. Guo Z Superspreading potential of COVID-19 outbreak seeded by Omicron variants of SARS-CoV-2 in Hong Kong J. Travel Med. 2022 29 taac049 10.1093/jtm/taac049 35435992 139. Althouse BM Superspreading events in the transmission dynamics of SARS-CoV-2: opportunities for interventions and control PLoS Biol. 2020 18 e3000897 10.1371/journal.pbio.3000897 33180773 140. Baden LR Efficacy and safety of the mRNA-1273 SARS-CoV-2 vaccine N. Engl. J. Med. 2021 384 403 416 10.1056/NEJMoa2035389 33378609 141. Lopez Bernal J Effectiveness of Covid-19 vaccines against the B.1.617.2 (Delta) variant N. Engl. J. Med. 2021 385 585 594 10.1056/NEJMoa2108891 34289274 142. Feikin DR Duration of effectiveness of vaccines against SARS-CoV-2 infection and COVID-19 disease: results of a systematic review and meta-regression Lancet 2022 399 924 944 10.1016/S0140-6736(22)00152-0 35202601 143. Belik M Comparative analysis of COVID-19 vaccine responses and third booster dose-induced neutralizing antibodies against Delta and Omicron variants Nat. Commun. 2022 13 2476 10.1038/s41467-022-30162-5 35513437 144. Pegu A Durability of mRNA-1273 vaccine-induced antibodies against SARS-CoV-2 variants Science 2021 373 1372 1377 10.1126/science.abj4176 34385356 145. Lipsitch M Krammer F Regev-Yochay G Lustig Y Balicer RD SARS-CoV-2 breakthrough infections in vaccinated individuals: measurement, causes and impact Nat. Rev. Immunol. 2022 22 57 65 10.1038/s41577-021-00662-4 34876702 146. Mostaghimi D Valdez CN Larson HT Kalinich CC Iwasaki A Prevention of host-to-host transmission by SARS-CoV-2 vaccines Lancet Infect. Dis. 2021 10.1016/S1473-3099(21)00472-2 34534512 147. Russell MW Moldoveanu Z Ogra PL Mestecky J Mucosal immunity in COVID-19: a neglected but critical aspect of SARS-CoV-2 infection Front. Immunol. 2020 11 611337 10.3389/fimmu.2020.611337 33329607 148. Sheikh-Mohamed S Systemic and mucosal IgA responses are variably induced in response to SARS-CoV-2 mRNA vaccination and are associated with protection against subsequent infection Mucosal Immunol. 2022 15 799 808 10.1038/s41385-022-00511-0 35468942 149. Collier AY Characterization of immune responses in fully vaccinated individuals after breakthrough infection with the SARS-CoV-2 Delta variant Sci. Transl Med. 2022 14 eabn6150 10.1126/scitranslmed.abn6150 35258323 150. Pouwels KB Effect of Delta variant on viral burden and vaccine effectiveness against new SARS-CoV-2 infections in the UK Nat. Med. 2021 27 2127 2135 10.1038/s41591-021-01548-7 34650248 151. Singanayagam A Community transmission and viral load kinetics of the SARS-CoV-2 Delta (B.1.617.2) variant in vaccinated and unvaccinated individuals in the UK: a prospective, longitudinal, cohort study Lancet Infect. Dis. 2021 22 183 195 10.1016/S1473-3099(21)00648-4 34756186 152. Levine-Tiefenbrun M Viral loads of Delta-variant SARS-CoV-2 breakthrough infections after vaccination and booster with BNT162b2 Nat. Med. 2021 27 2108 2110 10.1038/s41591-021-01575-4 34728830 153. Emary KRW Efficacy of ChAdOx1 nCoV-19 (AZD1222) vaccine against SARS-CoV-2 variant of concern 202012/01 (B.1.1.7): an exploratory analysis of a randomised controlled trial Lancet 2021 397 1351 1362 10.1016/S0140-6736(21)00628-0 33798499 154. Chia PY Virological and serological kinetics of SARS-CoV-2 Delta variant vaccine breakthrough infections: a multicentre cohort study Clin. Microbiol. Infect. 2021 28 612.e1 612.e7 10.1016/j.cmi.2021.11.010 34826623 155. Garcia-Knight M Infectious viral shedding of SARS-CoV-2 Delta following vaccination: a longitudinal cohort study PLoS Pathog. 2022 18 e1010802 10.1371/journal.ppat.1010802 36095030 156. Peña-Hernández MA Comparison of infectious SARS-CoV-2 from the nasopharynx of vaccinated and unvaccinated individuals medRxiv 2022 10.1101/2021.12.28.21268460 35982667 157. Shamier MC Virological characteristics of SARS-CoV-2 vaccine breakthrough infections in health care workers medRxiv 2021 10.1101/2021.08.20.21262158 158. Jung J Transmission and infectious SARS-CoV-2 shedding kinetics in vaccinated and unvaccinated individuals JAMA Netw. Open 2022 5 e2213606 10.1001/jamanetworkopen.2022.13606 35608859 159. Hirotsu Y Similar viral loads in Omicron infections regardless of vaccination status medRxiv 2022 10.1101/2022.04.19.22274005 160. Letizia AG SARS-CoV-2 seropositivity and subsequent infection risk in healthy young adults: a prospective cohort study Lancet Respir. Med. 2021 9 712 720 10.1016/S2213-2600(21)00158-2 33865504 161. Pulliam JRC Increased risk of SARS-CoV-2 reinfection associated with emergence of Omicron in South Africa Science 2022 376 eabn4947 10.1126/science.abn4947 35289632 162. Harris RJ Effect of vaccination on household transmission of SARS-CoV-2 in England N. Engl. J. Med. 2021 385 759 760 10.1056/NEJMc2107717 34161702 163. Bates TA Vaccination before or after SARS-CoV-2 infection leads to robust humoral response and antibodies that effectively neutralize variants Sci. Immunol. 2022 7 eabn8014 10.1126/sciimmunol.abn8014 35076258 164. Wratil PR Three exposures to the spike protein of SARS-CoV-2 by either infection or vaccination elicit superior neutralizing immunity to all variants of concern Nat. Med. 2022 28 496 503 10.1038/s41591-022-01715-4 35090165 165. Malato J Risk of BA.5 infection among persons exposed to previous SARS-CoV-2 variants N. Engl. J. Med. 2022 387 953 954 10.1056/NEJMc2209479 36044619 166. Cao Y BA.2.12.1, BA.4 and BA.5 escape antibodies elicited by Omicron infection Nature 2022 608 593 602 10.1038/s41586-022-04980-y 35714668 167. Hassan AO A single-dose intranasal ChAd vaccine protects upper and lower respiratory tracts against SARS-CoV-2 Cell 2020 183 169 184 10.1016/j.cell.2020.08.026 32931734 168. Lapuente D Protective mucosal immunity against SARS-CoV-2 after heterologous systemic prime-mucosal boost immunization Nat. Commun. 2021 12 6871 10.1038/s41467-021-27063-4 34836955 169. Afkhami S Respiratory mucosal delivery of next-generation COVID-19 vaccine provides robust protection against both ancestral and variant strains of SARS-CoV-2 Cell 2022 185 896 915 10.1016/j.cell.2022.02.005 35180381 170. Grubaugh ND Hodcroft EB Public health actions to control new SARS-CoV-2 variants Cell 2021 184 1127 1132 10.1016/j.cell.2021.01.044 33581746 171. Carreño JM Activity of convalescent and vaccine serum against SARS-CoV-2 Omicron Nature 2022 602 682 688 10.1038/s41586-022-04399-5 35016197 172. Ashcroft P Lehtinen S Bonhoeffer S Test-trace-isolate-quarantine (TTIQ) intervention strategies after symptomatic COVID-19 case identification PLoS ONE 2022 17 e0263597 10.1371/journal.pone.0263597 35148359 173. Santos Bravo M Viral culture confirmed SARS-CoV-2 subgenomic RNA value as a good surrogate marker of infectivity J. Clin. Microbiol. 2022 60 e0160921 10.1128/JCM.01609-21 34669457 174. Kim JY Diagnostic usefulness of subgenomic RNA detection of viable SARS-CoV-2 in patients with COVID-19 Clin. Microbiol. Infect. 2022 28 101 106 10.1016/j.cmi.2021.08.009 34400343 175. Alexandersen S Chamings A SARS-CoV-2 genomic and subgenomic RNAs in diagnostic samples are not an indicator of active replication Nat. Commun. 2020 11 6059 10.1038/s41467-020-19883-7 33247099 176. Bonenfant G Surveillance and correlation of SARS-CoV-2 viral RNA, antigen, virus isolation, and self-reported symptoms in a longitudinal study with daily sampling Clin. Infect. Dis. 2022 10.1093/cid/ciac282 35442437 177. Peeling RW Heymann DL Teo YY Garcia PJ Diagnostics for COVID-19: moving from pandemic response to control Lancet 2022 399 757 768 10.1016/S0140-6736(21)02346-1 34942102 178. Pilarowski G Performance characteristics of a rapid severe acute respiratory syndrome coronavirus 2 antigen detection assay at a public plaza testing site in San Francisco J. Infect. Dis. 2021 223 1139 1144 10.1093/infdis/jiaa802 33394052 179. Leber W Comparing the diagnostic accuracy of point-of-care lateral flow antigen testing for SARS-CoV-2 with RT-PCR in primary care (REAP-2) EClinicalMedicine 2021 38 101011 10.1016/j.eclinm.2021.101011 34278276 180. Kohmer N The comparative clinical performance of four SARS-CoV-2 rapid antigen tests and their correlation to infectivity in vitro J. Clin. Med. 2021 10.3390/jcm10020328 34945047 181. Wang W Detection of SARS-CoV-2 in different types of clinical specimens JAMA 2020 323 1843 1844 32159775 182. Lee, R. A., Herigon, J. C., Benedetti, A., Pollock, N. R. & Denkinger, C. M. Performance of saliva, oropharyngeal swabs, and nasal swabs for SARS-CoV-2 molecular detection: a systematic review and meta-analysis. J. Clin. Microbiol. 59, e02881-20 (2021). 183. Tsang NNY Diagnostic performance of different sampling approaches for SARS-CoV-2 RT-PCR testing: a systematic review and meta-analysis Lancet Infect. Dis. 2021 21 1233 1245 10.1016/S1473-3099(21)00146-8 33857405 184. Peto T COVID-19: rapid antigen detection for SARS-CoV-2 by lateral flow assay: a national systematic evaluation of sensitivity and specificity for mass-testing EClinicalMedicine 2021 36 100924 10.1016/j.eclinm.2021.100924 34101770 185. Hay JA Estimating epidemiologic dynamics from cross-sectional viral load distributions Science 2021 373, eabh0635 10.1126/science.abh0635 34083451 186. Peccia J Measurement of SARS-CoV-2 RNA in wastewater tracks community infection dynamics Nat. Biotechnol. 2020 38 1164 1167 10.1038/s41587-020-0684-z 32948856 187. Xin H The incubation period distribution of coronavirus disease 2019: a systematic review and meta-analysis Clin. Infect. Dis. 2021 73 2344 2352 10.1093/cid/ciab501 34117868 188. Yu IT Evidence of airborne transmission of the severe acute respiratory syndrome virus N. Engl. J. Med. 2004 350 1731 1739 10.1056/NEJMoa032867 15102999 189. Cheng PK Viral shedding patterns of coronavirus in patients with probable severe acute respiratory syndrome Lancet 2004 363 1699 1700 10.1016/S0140-6736(04)16255-7 15158632 190. Peiris JS Clinical progression and viral load in a community outbreak of coronavirus-associated SARS pneumonia: a prospective study Lancet 2003 361 1767 1772 10.1016/S0140-6736(03)13412-5 12781535 191. Chan PKS Laboratory diagnosis of SARS Emerg. Infect. Dis. 2004 10 825 831 10.3201/eid1005.030682 15200815 192. Liu W Long-term SARS coronavirus excretion from patient cohort, China Emerg. Infect. Dis. 2004 10 1841 1843 10.3201/eid1010.040297 15504274 193. Xu D Persistent shedding of viable SARS-CoV in urine and stool of SARS patients during the convalescent phase Eur. J. Clin. Microbiol. Infect. Dis. 2005 24 165 171 10.1007/s10096-005-1299-5 15789222 194. V’Kovski P Disparate temperature-dependent virus–host dynamics for SARS-CoV-2 and SARS-CoV in the human respiratory epithelium PLoS Biol. 2021 19 e3001158 10.1371/journal.pbio.3001158 33780434 195. Pitzer VE Leung GM Lipsitch M Estimating variability in the transmission of severe acute respiratory syndrome to household contacts in Hong Kong, China Am. J. Epidemiol. 2007 166 355 363 10.1093/aje/kwm082 17493952 196. Riley S Transmission dynamics of the etiological agent of SARS in Hong Kong: impact of public health interventions Science 2003 300 1961 1966 10.1126/science.1086478 12766206 197. Zaki AM van Boheemen S Bestebroer TM Osterhaus AD Fouchier RA Isolation of a novel coronavirus from a man with pneumonia in Saudi Arabia N. Engl. J. Med. 2012 367 1814 1820 10.1056/NEJMoa1211721 23075143 198. Breban R Riou J Fontanet A Interhuman transmissibility of Middle East respiratory syndrome coronavirus: estimation of pandemic risk Lancet 2013 382 694 699 10.1016/S0140-6736(13)61492-0 23831141 199. Oh MD Viral load kinetics of MERS coronavirus infection N. Engl. J. Med. 2016 375 1303 1305 10.1056/NEJMc1511695 27682053 200. Corman VM Viral shedding and antibody response in 37 patients with Middle East respiratory syndrome coronavirus Infection Clin. Infect. Dis. 2016 62 477 483 26565003 201. Min C-K Comparative and kinetic analysis of viral shedding and immunological responses in MERS patients representing a broad spectrum of disease severity Sci. Rep. 2016 6 25359 10.1038/srep25359 27146253 202. Ip DK Viral shedding and transmission potential of asymptomatic and paucisymptomatic influenza virus infections in the community Clin. Infect. Dis. 2017 64 736 742 28011603 203. Carrat F Time lines of infection and disease in human influenza: a review of volunteer challenge studies Am. J. Epidemiol. 2008 167 775 785 10.1093/aje/kwm375 18230677 204. Pawelek KA Modeling within-host dynamics of influenza virus infection including immune responses PLoS Comput. Biol. 2012 8 e1002588 10.1371/journal.pcbi.1002588 22761567 205. Ip DKM The dynamic relationship between clinical symptomatology and viral shedding in naturally acquired seasonal and pandemic influenza virus infections Clin. Infect. Dis. 2016 62 431 437 26518469 206. Nair H Global burden of acute lower respiratory infections due to respiratory syncytial virus in young children: a systematic review and meta-analysis Lancet 2010 375 1545 1555 10.1016/S0140-6736(10)60206-1 20399493 207. Bagga B Comparing influenza and RSV viral and disease dynamics in experimentally infected adults predicts clinical effectiveness of RSV antivirals Antivir. Ther. 2013 18 785 791 10.3851/IMP2629 23714753 208. DeVincenzo JP Viral load drives disease in humans experimentally infected with respiratory syncytial virus Am. J. Respir. Crit. Care Med. 2010 182 1305 1314 10.1164/rccm.201002-0221OC 20622030 209. Kutter JS Small quantities of respiratory syncytial virus RNA only in large droplets around infants hospitalized with acute respiratory infections Antimicrob. Resist. Infect. Control 2021 10 100 10.1186/s13756-021-00968-x 34193302	36460930	PMC9716513	NO-CC CODE	2022-12-03 23:20:58	no	Nat Rev Microbiol. 2022 Dec 2;:1-15	utf-8	Nat Rev Microbiol	2,022	10.1038/s41579-022-00822-w	oa_other
==== Front Immunogenetics Immunogenetics Immunogenetics 0093-7711 1432-1211 Springer Berlin Heidelberg Berlin/Heidelberg 36459183 1282 10.1007/s00251-022-01282-5 Original Article Bioinformatics analysis of structural protein to approach a vaccine candidate against Vibrio cholerae infection Oladipo Elijah Kolawole [email protected] 12 Akindiya Olawumi Elizabeth 18 Oluwasanya Glory Jesudara 1 Akanbi Gideon Mayowa 13 Olufemi Seun Elijah 14 Adediran Daniel Adewole 14 Bamigboye Favour Oluwadara 1 Aremu Rasidat Oyindamola 1 Kolapo Kehinde Temitope 1 Oluwasegun Jerry Ayobami 15 Awobiyi Hezekiah Oluwajoba 1 Jimah Esther Moradeyo 1 Irewolede Boluwatife Ayobami 1 Folakanmi Elizabeth Oluwatoyin 13 Olubodun Odunola Abimbola 15 Akintibubo Samuel Adebowale 13 Odunlami Foluso Daniel 15 Ojo Taiwo Ooreoluwa 14 Akinro Omodamola Paulina 13 Hezikiah Oluwaseun Samuel 15 Olayinka Adenike Titilayo 17 Abiala Grace Asegunloluwa 15 Idowu Akindele Felix 14 Ogunniran James Akinwunmi 17 Ikuomola Mary Omotoyinbo 15 Adegoke Hadijat Motunrayo 16 Idowu Usman Abiodun 13 Olaniyan Oluwaseyi Paul 9 Bamigboye Olutoyin Omolara 10 Akinde Sunday Babatunde 11 Babalola Musa Oladayo 12 1 Genomics Unit, Helix Biogen Institute, Ogbomoso, Oyo State, Nigeria 2 grid.472242.5 0000 0004 4649 0041 Department of Microbiology, Laboratory of Molecular Biology, Bioinformatics and Immunology, Adeleke University, Osun State, P.M.B 250, Ede, Nigeria 3 grid.411270.1 0000 0000 9777 3851 Department of Pure and Applied Biology, Microbiology Unit, Ladoke Akintola University of Technology, Ogbomoso, Oyo State, Nigeria 4 grid.411270.1 0000 0000 9777 3851 Department of Biochemistry, Ladoke Akintola University of Technology, Ogbomoso, Oyo State, Nigeria 5 grid.411270.1 0000 0000 9777 3851 Department of Physiology, Ladoke Akintola University of Technology, Ogbomoso, Oyo State, Nigeria 6 grid.411270.1 0000 0000 9777 3851 Department of Pure and Applied Chemistry, Laboratory of Computational and Biophysical Chemistry, Ladoke Akintola University of Technology, Ogbomoso, Oyo State, Nigeria 7 grid.411270.1 0000 0000 9777 3851 Department of Medical Microbiology and Parasitology, Ladoke Akintola University of Technology, Ogbomoso, Oyo State, Nigeria 8 Department of Biology, Olusegun Agagu University of Science and Technology, Okiti-Pupa, Ondo State, Nigeria 9 grid.412422.3 0000 0001 2045 3216 Department of Biochemistry, Faculty of Basic and Applied Sciences, Osun State University, P.M.B. 4494, Oke-BaaleOsogbo, Nigeria 10 grid.472242.5 0000 0004 4649 0041 Department of Microbiology, Faculty of Science, Adeleke University, Osun State, P.M.B 250, Ede, Nigeria 11 grid.412422.3 0000 0001 2045 3216 Department of Microbiology, Faculty of Basic and Applied Sciences, Osun State University, P.M.B. 4494, Oke-BaaleOsogbo, Nigeria 12 grid.411782.9 0000 0004 1803 1817 Department of Biochemistry, College of Medicine, University of Lagos, Lagos, Nigeria 2 12 2022 116 25 8 2022 23 10 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The bacteria Vibrio cholerae causes cholera, an acute diarrheal infection that can lead to dehydration and even death. Over 100,000 people die each year as a result of epidemic diseases; vaccination has emerged as a successful strategy for combating cholera. This study uses bioinformatics tools to create a multi-epitope vaccine against cholera infection using five structural polyproteins from the V. cholerae (CTB, TCPA, TCPF, OMPU, and OMPW). The antigenic retrieved protein sequence were analyzed using BCPred and IEDB bioinformatics tools to predict B cell and T cell epitopes, respectively, which were then linked with flexible linkers together with an adjuvant to boost it immunogenicity. The construct has a theoretical PI of 6.09, a molecular weight of 53.85 kDa, and an estimated half-life for mammalian reticulocytes in vitro of 4.4 h. These results demonstrate the construct’s longevity. The vaccine design was docked against the human toll-like receptor (TLR) to evaluate compatibility and effectiveness; also other additional post-vaccination assessments were carried out on the designed vaccine. Through in silico cloning, its expression was determined. The results show that it has a CAI value of 0.1 and GC contents of 58.97% which established the adequate expression and downstream processing of the vaccine construct, and our research demonstrated that the multi-epitope subunit vaccine exhibits antigenic characteristics. Additionally, we carried out an in silico immunological simulation to examine the immune reaction to an injection. Our results strongly suggest that the vaccine candidate on further validation would induce immune response against the V. cholerae infection. Keywords Cholera Vibrio cholerae Immune-informatics Multi-epitope Peptide vaccine ==== Body pmcIntroduction Cholera, which can cause very serious acute watery diarrhea and is also a symptom of social inequality and a lack of development, continues to pose a threat to public health which has cause emergency and brought to notice of the scientists. Dunkin (2021) and Nezafat et al. (2016) assert that cholera is a highly contagious disease that is spread by ingesting water or food that has been contaminated with the Vibrio cholerae bacterium. Cholera outbreaks have been documented during the previous 60 years in a number of African and Asian countries in 2021 and 2022. According to reports, there are sizable outbreaks still going on in Afghanistan, Bangladesh, the Democratic Republic of the Congo, Ethiopia, and Nigeria. Since the latest update on February 16, 2022, about 30,629 suspected cholera cases, including 39 fatalities, have been recorded globally. According to Ali et al. (2015), cholera cases range from 1.3 to 4.0 million each year, and the disease kills 21,000 to 143,000 people worldwide. However, 5–10% of people who have severe cholera cases—which are 50% deadly if untreated—are affected. There have been over 287 suspected cholera cases reported in Nigeria, with 12 deaths. As of February 27, 2022, a total of 701 probable cases have been recorded from 12 states and Federal Capital Territory (FCTs), with 19 fatalities (CFR 2.7%). Five states, Taraba (242 cases), Cross River (111 cases), Borno (91 cases), Bayelsa (76 cases), and Adamawa, account for 82% of all cases. The age group most likely to be affected by the suspected cases are children under the age of 5 (WHO 2022). Even if the prevalence of cholera is not well known, it is nevertheless a serious tropical illness that is worth studying (CDC 2020; NORD 2021). According to Somboonwit et al. (2017), not all strains of V. cholerae are considered to be the cause of cholera. Out of the more than 200 serology groups of V. cholerae, only the O1 and O139 groups have resulted in cholera epidemics (Waldor and Mekalanos 1994). Vaccination is a way to control and prevent cholera epidemics in the underdeveloped countries (Helen and Shoubai 2020) . Some of the known virulence factors of Vibrio cholerae include V. cholera toxin (CTX), toxin-coregulated pili (TCP), lipopolysaccharide (LPS), and outer membrane proteins (Omps), which are important candidates for the cholera vaccine (Nezafat et al. 2016). Additionally, studies on their adverse consequences have been conducted (María et al. 2017). Also, additional research demonstrates that the multi-epitope vaccination has antigenic components that can elicit an immunological reaction (Validi et al. 2018; Khan et al. 2019; ul Qamar et al. 2021; Jyotisha and Qureshi 2022). This approach hereby eliminates some evaluations that have suggested that additional antigens may likely cause inflammation which are necessary for the immunogenicity of innovative vaccines in addition to identifying every antigen that may be investigated using conventional methods (Coscolla et al. 2015; Parvizpour et al. 2020). In order to promote immunity and elicit both cellular and humoral immune responses to combat the cholera sickness, this study developed a novel multi-epitope peptide vaccine encompassing the cytotoxic T lymphocyte (CTL), helper T lymphocyte (HTL), and linear B lymphocyte (LBL) epitopes employing immune-informatics techniques in the development process. Using an in silico simulation, it was demonstrated that the constructed vaccine is capable of inducing immunological upon injection. The vaccine’s effectiveness and safety were assessed using physicochemical, molecular docking, and thermodynamic stability profiling approaches. Our study lays the groundwork for the experimental creation of a potent cholera vaccine. Materials and methods Retrieval of Vibrio cholerae polyprotein In this research work, the best potential candidate protein sequences of Vibrio cholerae, O1 and O139 strain, were retrieved from NCBI Protein Database (https://www.ncbi.nlm.nih.gov/) in FASTA format for constructing multi-epitope vaccine by the use of immune-informatics approach (Dong et al. 2020). The protein sequences used were partial sequences (Shahab et al. 2022). Antigenicity and allergenicity prediction Antigenicity prediction was carried on the retrieved sequences using VaxiJen 2.0 webserver (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html (Abraham Peele et al. 2021). The server carries out the prediction of antigenic and non-antigenic amino acid sequences based on physicochemical properties only at a default threshold of 0.4 (Doytchinova and Flower 2007; Zaharieva et al. 2017; Habib 2020). All the sequences that passed the antigenicity test were further subjected to AlgPred webserver (https://webs.iiitd.edu.in/raghava/algpred/submission.html), an online tool to determine the allergenicity. The server uses six different approaches for the allergenicity prediction with 85% accuracy at 0.4 thresholds (Saha and Raghava 2006). Cytotoxic T lymphocyte (CTL) epitope prediction Cytotoxic T lymphocyte’s prediction of V. cholerae antigenic proteins were carried out using an online webserver- NetCTL (http://www.cbs.dtu.dk/services/NetCTL/) (Banerjee et al. 2020). The threshold was set to 0.75 to carry out the CTL epitopes prediction (Stranzl et al. 2010; Larsen et al. 2007). Helper T lymphocyte (HTL) epitope prediction Helper T lymphocyte’s prediction was carried out on the antigenic protein sequences using IEDB tool (http://www.iebd.org/immunogenicity/) according to the method of Vita et al. (2019). The human leukocyte antigen class II (HLA-DR) supertype alleles (DRB103:01, DRB301:01, DRB302:01, DRB302:02, DRB501:01) were selected, and three different methods were used in the prediction of the HLA-II epitopes; they are Consensus (smm/nn/sturniolo), Consensus (comb.lib./smm/nn), and NetMHCIIpan. The peptide affinity for each of the receptors is based on the IC50 score given to each and every predicted epitope. The same server was used for immunogenicity prediction as well, using major histocompatibility complex (MHC) class I immunogenicity (Vita et al. 2015). Prediction of B cell epitopes The B cell epitopes were predicted with an online tool server, BCpred (http://ailab-projects1.ist.psu.edu:8080/bcpred/SimpleServlet), and all parameters were left at default. The peptides, scores, and start position were shown by the predicted B cells (Chen et al. 2007). Toxicity prediction The predicted CTL, HTL, and B cell epitopes were submitted to ToxinPred, an online web server (http://crdd.osdd.net/raghava/toxinpred/) for toxicity prediction. The server classifies the epitopes as toxic and non-toxic based on their physicochemical properties (Gupta et al. 2013). Construction of multi-epitope vaccine sequence The predicted CTL, HTL, and B cell epitopes, sorted out through the discussed immune-informatics approach were used to design the vaccine construct. The adjuvant used was APPHALS, then EAAK linkers were used to link the adjuvant with the CTL epitope. The CTL epitope was linked with the 8 HTL epitopes using GPGPG linkers, while the 14 B cell epitopes were linked to the HTL epitopes with GPGPG linkers (Nagamune 2001). Population coverage prediction The epitopes selected for the Vibrio cholerae vaccine design must be predicted for their ability to binds to HLA molecules across different human population and the likelihood of it inducing long-lasting immune response in this population. IEDB population coverage tool was used in the prediction of the epitopes with high percentage of been presented in the context of HLA molecules to induce immune response in population (Bui et al. 2006). The endemic regions are selected for the population coverage. Allergenicity prediction of the vaccine Allergenicity prediction of the constructed vaccine was done using AlgPred (http://www.imtech.res.in/raghava/algpred/). The server used six different approaches for the prediction with 85% accuracy at 0.4 thresholds (Saha and Raghava 2006). Antigenicity prediction of the vaccine VaxiJen (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) was used for antigenicity prediction of the vaccine construct with bacteria selected as a model organism at a threshold of 0.4. The prediction of antigenic and non-antigenic amino acid sequences was carried out based on physicochemical properties only (Doytchinova and Flower 2007). Physicochemical property prediction ProtParam web tool (http://web.expasy.org/protparam/) was used for the examination of the physicochemical properties of the constructed sequence. Evaluation of the constructed amino acid sequence, the vaccine’s amino acid composition, in vivo/in vitro half life, theoretical pI, GRAVY (grand average of hydropathy) instability index, and molecular weight were all predicted (Gasteiger et al. 2005). Secondary structure prediction for the vaccines Secondary structure prediction of the constructed vaccine sequence was done using PSIPRED tool (http://bioinf.cs.ucl.ac.uk/psipred/), an online web-based server that performs secondary structure prediction given accurate result (Jones 1999). Tertiary structure prediction Tertiary structure prediction of the constructed sequence for the vaccine was carried out using I-TASSER (https://zhanggroup.org/I-TASSER/). The server generated 5 models, and the model with the highest C-score was selected (Yang and Zhang 2015) as predicted and validated by Zhou et al. (2022) to also model small protein structure. Tertiary structure refinement The final vaccine 3D structure was refined using Galaxy Refine tools (http://galaxy.seoklab.org/refine) (Adero et al. 2021). For the repacking of the amino acid sequence and side chain reconstruction, the server uses CASP10 refinement method. Also, the same method is employed for molecular dynamic simulation to relax the 3D structure of the query protein. Molecular docking of vaccine ClusPro 2.0 (https://cluspro.org), an online bioinformatics sever, was utilized for the docking of the vaccine with human TLR-4 accessed through PD ID of 4G8A from Pubchem (Ezediuno et al. 2021; Sanami et al. 2021). Twenty-six models were generated in respect to low energy, and large size was selected as it indicated a good interaction between the receptor and ligand. In silico cloning and codon optimization of vaccine protein The codon optimization was significant for the expression of vaccine structure in a host E. coli (strain K12) as the usage of a codon is comparatively different in E. Coli than the native host. Rho-independent transcription, restriction cleavage sites, and prokaryotic ribosome-binding sites were eluded by considering three extra options. Java codon adaptation tools provide the output in terms of CAI (codon adaptation index) and GC content in order to confirm the high level of protein expression. For cloning, the final vaccine construct in E. coli pET-28a ( +) vector modification N- and C-terminal with Xhol and Notl restriction sites were performed, respectively. Finally, for expression, the prepared optimized sequence, along with the restriction sites were incorporated into the pET-28a ( +) vector utilizing the SnapGene tool (https://www.snapgene.com) (Grote et al. 2005). Immune simulation To understand the molecular dynamics of the human immune system in response to designed vaccine, the relationship between the human immune system and vaccine construct was predicted using C-ImmSim 10.1 (https://kraken.ian.rm.cnr.it/C-IMMSIM/index.php?page=1), a server that uses agent-based modeling (Rapin et al. 2010). PSSM method was used to assess the production of cytokines and other substances like interferon and antibodies. Also, the response for T helper cell 1 and T helper cell 2 (Th1 and Th2) was predicted with the parameters left at default to measure the diversity or Simps Index by the server. Results Retrieval of antigenic proteins, antigenicity, and allergenicity prediction A hundred (100) protein sequences of Vibrio cholerae were retrieved from the NCBI database (https://www.ncbi.nlm.nih.gov/) out of which only 46 passed the antigenicity and allergenicity analysis. Five structural polyproteins toxin-coregulated pili (TCPA and TCPF), outer membrane protein (OMPW and OMPU), and cholera toxin B subunit (CTB) of Vibrio cholerae with an antigenic probability score of greater than 0.8 were considered further for vaccine construction. Cytotoxic T lymphocyte epitopes prediction From all the 46 protein sequences that passed antigenicity and allergenicity analysis, a total of 290 epitopes were generated, out of which 195 were toxin-coregulated pili and others were cholera toxin subunit. However, based on MHC class 1 binding affinity, toxicity, allergenicity, and immunogenicity, only one CTL epitope as shown in Table 1 was selected; this is due to the fact that other epitopes did not pass the allergenicity test.Table 1 Non-toxic and non-allergen CTL epitope with its combined score and immunogenicity score Protein name Epitopes Sequence Combined score Immunogenicity score TcpA Partial AAL09686.1 LTDFETTQA 0.9916 0.00388 T lymphocyte epitopes prediction Five reference sets of human alleles which includes HLA-DRB103:01, HLA-DRB301:01, HLA-DRB302:01, HLA-DRB302:02, and HLA-DRB501:01 were used for the prediction. A total of 230 HTL epitopes were subjected to the same predictions as CTL epitopes, in which only 8—208–222 (HLA-DRB501:01), 93–87 (HLA-DRB301:01), 248–262 (HLA-DRB302:01), 88–102 (HLA-DRB103:01), 84–98 (HLA-DRB103:01), 83–97 (HLA-DRB103:01), 130–144 (HLA-DRB103:01), and 85–99 (HLA-DRB103:01)—epitopes passed the predictions, and these were used for the designing of the vaccine (Table 2).Table 2 List of final non-toxic and non-allergen HTL epitopes with their prediction methods and percentile rank, respectively Proteins Allele Position Peptide sequence Method Percentile rank CTB HLA- DRB103:01 85–99 AIERMKDTLRIA YLT Consensus (smm/nn/sturniolo) 24.00 CTB HLA- DRB103:01 83–97 KKAIERMKDTLR IAY Consensus (smm/nn/sturniolo) 20.00 CTB HLA- DRB103:01 84–98 KAIERMKDTLRI AYL Consensus (smm/nn/sturniolo) 19.00 CTB HLA- DRB103:01 88–102 RMKDTLRIAYLT EAK Consensus (smm/nn/sturniolo) 17.00 TCPA HLA-DRB103:01 130–144 KTLITSVGDMFP YVL Consensus (smm/nn/sturniolo) 9.60 TCPF HLA- DRB301:01 73–87 PTSQDMFYDAYP STE Consensus (comb.lib./smm/nn) 6.50 TCPF HLA- DRB302:01 248–262 DVQFKVLVGVP HAET NetMHCIIpan 6.30 TCPF HLA- DRB501:01) 208–222 YPHIKVYEGTLS RLK Consensus (smm/nn/sturniolo) 3.10 Prediction of B cell epitope Out of the five cholera group of proteins, only 14 TcpA protein passed the allergenicity and toxicity test. These 14 linear B cell epitopes were used for the construction of multi-epitope vaccine using BCpred web server (http://ailab-projects1.ist.psu.edu:8080/bcpred/SimpleServlet) with the defaulted parameters. The predicted B cell epitopes shows (Table 3) the peptides, scores, and start position.Table 3 List of final non-toxic and non-allergen B cell epitopes with their start positions and scores, respectively Accession no Sequence Start position Score TcpA (Vibrio cholerae) DLNDFETNAANAAAGTGIIK 146 1 TcpA (Vibrio cholerae) GDLTDFETTPGAADTGIGVI 143 0.999 TcpA (Vibrio cholerae) AAAGTGTGIIKSIAPTSVNL 180 0.999 TcpA (Vibrio cholerae) QTYRSLGNYPTTADANAAAA 74 0.996 TcpA (Vibrio cholerae) TQTYRGLGQYPATADGTAAA 46 0.995 TcpA (Vibrio cholerae) DLGDFETGPANAVTGKGIIK 145 0.989 TcpA (Vibrio cholerae) QTYRSLGNYPATADATAAAA 48 0.987 TcpA (Vibrio cholerae) KKFVKEEHDKKTGQEGMTLL 10 0.984 TcpA (Vibrio cholerae) QTYRSLGNYPTTADAAAAAA 49 0.983 TcpA (Vibrio cholerae) LGKISPDEAKNPFTGTDMNI 76 0.978 TcpA (Vibrio cholerae) LIQTYRGLGNYPETTDDTAA 47 0.974 TcpA (Vibrio cholerae) LGKISPDEAKNPFTGADMNI 76 0.97 TcpA (Vibrio cholerae) VQVSMTQTYRALGNYPATAN 52 0.935 TcpA (Vibrio cholerae) LVTSVGDMFPYINVQEKAAV 122 0.91 Construction of multi-epitope peptide vaccine The final vaccine construct was designed by the combination of adjuvant and epitopes through the use of some linkers. 1 CTL, 8 HTL, and 14 B cells were made use in the designing of the subunit vaccine construct. With the use of EAAAK linker, the adjuvant was fused with the CTL epitope at the N-terminal; 8 HTL and 14 B cells were fused through the use of GPGPG linkers Figs. 1, 2 and 3. Population coverage The population coverage of the CTL epitopes across the endemic regions in Africa using the references HLA-A, HLA-B, and HLA-C alleles and that of HTL epitopes across the same regions using the references, HLA-DP, HLA-DQ, and HLA-DP selected during the MHC binding predictions are depicted in Table 4, while Fig. 4 is showing the percentage population coverage of the combined MHC molecules.Table 4 The population coverage of cholera vaccine epitopes Population/area Class I coverage Class II coverage MHC class combined Burkina Faso 56.43% 12.76% 67.63% South Africa 57.63% 23.61% 71.46% Kenya 71.96% 23.03% 71.96% Morocco 78.0% 64.38% 92.16% Senegal 69.36% 6.49% 71.35% Sudan 67.55% 43.6% 81.7% Tunisia 75.43% 51.08% 87.98% Uganda 74.81% 20.81% 74.81% Zambia 81.65% 12.76% 81.65% Zimbabwe 69.26% 22.21% 76.09% Average 70.21 30.89 60.41 Fig. 1 Systematic flow diagram for cholera vaccine construct. It is a flow chart of all the key steps involved in creating potential T cell, B cell, and multi-epitope vaccines against cholera. The details of the various steps are described in this section Fig. 2 Construction process of the vaccine Fig. 3 Epitope population coverage of the designed vaccine in endemic region of Africa Fig. 4 The peptide structure of the vaccine Prediction, validation and refinement of 3D structure I-TASSER was used for the construction of the 3D structure. Five models were generated, out of which the best was selected based on the C-scores, the model with high C-score was selected because C-score of a higher value signifies a model with a higher confidence and vice versa. For refinement of the vaccine construct, Galaxy Refinement server was used in which five refined model were given but the best of them all was chosen based on RMSD, Mol Probity, High GDT-HA score, poor rotamers, and Rama favored. Model 5 3D structure is provided below. The model structure was chosen based on its structure information such as GDT-HA score of 0.9437, RMSD score of 0.450, Mol Probity calculated as 2.259, low clash score of 15.7, poor rotamers calculated as 0.5, and the Rama favored estimated to be 89.5 which is the highest and it is far better than other models of the vaccine, as shown in Table 5. Therefore, model 5 was selected as a refined vaccine structure out of the five refined models for further analysis. The 3D model of the vaccine selected was then subjected to structure assessment stage, and these includes ProCheck for protein model showing that 76.1% of amino acid are plotted in the most favorable region, 18.8% in allowed regions, and 2.7% in disallowed region. Also, ProSA-web and ERRANT estimated and verified the whole quality of the basic 3D model. The overall quality calculated by ERRANT was 64.02% and ProSA-web displayed a score of − 7.88 for the 3D model of the vaccine construct, which authenticate the precision of the vaccine. Result form ProSA-web and RAMPAGE are given in Fig. 5.Table 5 Galaxy Refined results for vaccine construct. Model was selected based on its low Mol Probity and high Rama favored score Model GDT-HA RMSD Mol Probity Clash score Poor rotamers Rama favored Initial 1.0000 0.000 2.552 22.7 1.1 83.6 MODEL 1 0.9455 0.442 2.281 15.8 0.0 88.8 MODEL 2 0.9441 0.449 2.239 14.2 0.0 88.8 MODEL 3 0.9432 0.447 2.227 13.9 0.3 89.0 MODEL 4 0.9423 0.452 2.258 15.3 0.5 89.2 MODEL 5 0.9437 0.450 2.259 15.7 0.5 89.5 Fig. 5 3D structure refinement, evaluation, and prediction. a The use of rampage Ramachandran plot for quality assessment of the vaccine model; the black square denotes torsion angle distribution relative to the core (red) and allowed (yellow) regions. Residues are plotted in generously allowed (pale yellow) and disallowed (white) regions. b In ProsaWeb, the prosa plot score of the vaccine can be identified as a black dot in the plot of all PDB structures’ quality scores of the same size. c ProsaWeb showing sequence position Antigenicity and allergenicity assessment of the vaccine construct Vaxijen and AlgPred servers were used for the prediction of antigenicity and allergenicity of the vaccine sequences, respectively, at a threshold of 0.4 for bacteria; the antigenicity server confirms that the candidate vaccine is antigenic with a score of 0.7844 and with the nearest protein in UniProtKB accession number Q86UU0, defined the vaccine as non-allergen on AlgPred. Physiochemical properties of vaccines The physiochemical properties of the designed vaccine sequences were predicted via ProtParam. Theoretical isoelectric point (theoretical PI) was predicted to be 6.09, molecular weight as 53.85 kDa, the estimated half-life for in vitro for mammalian reticulocytes is 4.4 h, while for in vivo in yeast and Escherichia coli is > 20 and > 10, respectively. The vaccine sequence aliphatic index was computed to be 60.32, which indicate the thermostability of the vaccine, and was classified as stable as the instability index (II) is computed to be 13.99 and the hydrophilic nature of the vaccine sequence was shown as the negative grand average of hydropathicity (GRAVY) was − 0.435. Secondary structure prediction In the secondary structure prediction of the vaccine construct which was carried out by PSIPRED, alpha-helix (Hh) was computed as 18.06%, extended strand (Ee) as 18.81%, and random coil (Cc) as 58.29%. Graphical representation of the secondary structure elements is shown in Fig. 6.Fig. 6 Graphical representation of the secondary structure of the vaccine construct Molecular docking of constructed vaccine with human-toll like receptor-4 TLR-4 having PDB ID of 4G8A was selected as the receptor of choice for docking the vaccine which was carried out through a bioinformatics web server known as ClusPro 2.0 server. Twenty-six models were generated, but the best of it all in respect to low energy and large size was selected as it indicates a good interaction between the receptor and ligand. Graphically, docked complex intermolecular conformation and chemical interactions are presented below (Fig. 7).Fig. 7 Molecular docking of the vaccine with toll-like receptor-4. a Ribbon-like top-ranked model of docking assay. b Cartoon-like top-ranked model of docking assay Immune simulation of the multi-epitope vaccine Through in silico immune simulation, the injection of vaccine doses, the human immune system response was observed. The antibody titer was very high after the injection, and this triggered the immune response. IgM + IgG titer was observed at 700,000 antigens count per ml. IgM was around 400,000 xx/ml and IgG1 was reported as 250,000 xx/ml. Interleukin and cytokine responses were also observed,while IFN-gamma and IL-2 titer were increased significantly. All these indicate the dependability and high immune-triggering response upon injection. It also shows that there is high cellular immune response and memory cells’ development to pathogen recognition at re-occurrence. The population of the T cells, dendritic cells, phagocytic macrophages, and phagocytic natural killer cells were reported to be > 1150 cell/mm3, 225 cell/mm3, 225 cell/mm3, 374 cell/mm3, respectively. Therefore, all these results depicted in Fig. 8 denote that our vaccine candidate can trigger immune response effectively Fig. 9.Fig. 8 In silico immune simulation by the vaccine followed by injection. The immune response to the vaccine is demonstrated by antibodies titer. a Interferon and interleukins. b Natural killer cells. c Macrophages. d Cytotoxic T cell. e Dendritic cells. f Antigen and immunoglobulin Fig. 9 Codon adaptation using E. coli K12 strain Codon adaptation and insilico cloning Vaccine sequence design was modified using JCAT in reference to the E. coli K12 strain. The CAI value obtained was 0.1, and the GC content was 58.97%, indicating a greater level of vaccine design expression in the E. coli K12 strain. According to reports, GC levels between 35 and 70% have greater gene expression. The designed vaccine was cloned into the Pet-28a ( +) plasmid using the restriction cloning tool SnapGene. To do this, suitable restriction enzymes Xhol and Notl were selected, and their restriction sites were inserted at both ends of the modified nucleotide sequence (Habib 2020). The restriction sites for Xhol and Notl were added to the N- and C-termini of the optimized vaccine design and cloned in the pET-28a ( +) vector as shown in Fig. 10.Fig. 10 Cloned vaccine using Pet-28a( +) vector Discussion The use of vaccines has greatly improved the world’s health over a long period of time (Oli et al. 2020). In the past, developing vaccinations was thought to be the most practical method of preventing infectious diseases (Greenwood 2014). Herd immunity greatly reduces disease load, death, and disability (Helen and Shoubai 2020). Personalized vaccination is required for infectious organisms with complicated life cycles and antigenic diversity, and newly developing and re-emerging infectious diseases (ERID) bring additional challenges in vaccine development (Poland et al. 2016; Oli et al. 2020). With immunological findings and understanding of computational methods for epitope predictions, a new pattern of vaccine design has been revealed (Adhikari et al. 2018). Immune-informatics approaches are quick and efficient methods for discovering viable candidates for vaccine design using a pathogen’s proteome (Khan et al. 2019). According to Oladipo et al. (2022), epitope-based vaccine design is advantageous, profitable, extremely stable, non-toxic, and easy to engineer in comparison to conventional vaccines, which may raise several issues in patients who are compromised in immune system which also agrees with Wang et al. (2022) on the use of nucleic based vaccine. In silico immune-informatics methods can be used to accurately predict B and T cell epitopes using a plethora of bioinformatics tools, according to Farhadi et al. (2015). This can have an impact on the formulation of epitope vaccines. Utilizing highly immunogenic and precise B cell and T cell epitopes that are extracted from the pathogen’s proteome, these multi-epitope-based vaccination peptides are made (Sajjad et al. 2020). Therefore, this method can not only identify all antigens that can be investigated using conventional approaches, but it can also identify additional antigens that are crucial to the immunogenicity of novel vaccines (Flower et al. 2010). Consequently, cholera caused 2.8 million illnesses and 91,000 fatalities in more than 50 endemic nations. However, according to the most recent data, there were 69 different countries’ total of 2.9 million illnesses and 95,000 fatalities (Ali et al. 2015). It is highly desirable to develop an efficient and reliable cholera epitope vaccination. The current study’s objectives were to generate a cholera vaccine to lower worldwide mortality, and the findings will be validated by prior research in the field using the same approach with Banerjee et al. (2020). The majority of cholera vaccines now on the market or in development are inactivated (like Dukoral1, mORCAX1, and SancholTM) and live attenuated (like CVD103-HgR and Peru-15 or CholeraGarde1) (Holmgren 2021). Nezafat et al. (2016) proposed that using immunostimulatory adjuvant (CTB) in epitope vaccine development, with the selection promiscuous epitopes from various V. cholerae protective antigens (OmpW, OmpU, TcpA, and TcpF) that bind to different HLA-II supertype alleles, and linking the epitopes with suitable linkers (EAAAK and GPGPG) to each other are important strategies to create CTL, HTL, and B cell epitopes were predicted using a variety of bioinformatics methods. The closest UniProt number, Q86UU0, which is designated as a non-allergen, antigenicity, and immunogenicity of synthetic peptides, was used to evaluate the allergenicity using the AlgPred service (Sharma et al. 2021; Adam 2021; Ghandadi 2022). The immunization remained stable, nevertheless, as evidenced by the vaccine construct’s molecular weight of 53.85 Da and predicted PI of 6.09. The vaccine construct’s molecular weight of 53.85 Da and Pi of 6.09 show that it was stable. Cholera epitopes of various protective antigens (Ompu, Ompw, TcpA, and TcpF) that bind to different HLA-II supertype alleles with suitable linkers (EAAAK, GPGPG) are integrated into immune-stimulating extra vesicular CTB combined with CTB (Nezafat et al. 2016; Wieczorek et al. 2017). To generate more precise epitopes that are limited to particular HLA-II supertype alleles, three servers (IEBD, SVMHC, and PSIpred) were employed, each with a different technique (Nielsen et al. 2010, 2020). B cell epitope is also necessary for the immunization to boost antibody response. The antigens ompU, ompw, TcpA, and tcpF were used to represent the frequently occurring area between mixed epitopes and B cell epitopes. Because of its dual role in the formation of the epitope vaccination, the later region of the antigen was chosen. First, it limits the emergence of junctional epitopes (neoepitopes), and second, it improves immune processing and displays HLA-II binding epitopes (Livingstone et al. 2002). This area was joined using short amino acid linkers, particularly the GPGPG linker. To prevent contact with other vaccine segments, helical linkers were introduced to the N- and C-terminals, which improved the separation. CTB is a non-toxic subunit that acts as a mucosal immunostimulatory adjuvant, stimulating both mucosal and systemic immune responses (Stratmann 2015; Lavelle and Ward 2022). The population coverage of the designed cholera vaccine show that the designed cholera vaccine can induce immune response in an average of about over 60% of African population that cholera has been found to be endemic. I-TASSER was used to simulate the initial 3D structure of the protein vaccine which has been reported to also modelled 3D of small protein sequences as validated by (Zhou et al. 2022). The model has to be modified, according to the results of the ProSA-web plot analysis of the preliminary version (Yang and Zhang 2015). After loop refinement and energy reduction of the core model, a high-quality 3-D standard necessary for the identification of conformational B cell epitopes and docking techniques was attained. The CTB region of the vaccine and TLR2 were successfully docked by Cluspro (Kozakov et al. 2017). Due to CTB’s strong sequence and structural similarity to heat-labile enterotoxin B (LTB) and TLR2’s functional characteristics, an adequate docking model may be achieved even though there is no direct evidence of interaction between CTB and TLR2. The immunological and physicochemical characteristics of the protein vaccine were assessed (Dey et al. 2014). The vaccine is antigenic and also non-allergenic, according to our findings. Additionally, the vaccine demonstrated the appropriate level of solubility when it was overexpressed. The protein vaccines has high aliphatic index and high thermal stability. Finally, by modifying the DNA codon for production in the E. coli host and adding limit sites to gene flanks, the epitope vaccine was cloned in silico suggesting it a good candidate vaccine for Vibrio cholerae infection, and therefore, there is need for the use of wet laboratory techniques for further validation of the in silico prediction. Conclusion In our study, we used an immune-informatics approach to design a potential multi-epitope vaccine using six different groups of protein from vibrio cholerae. This cholera vaccine has the potential to provide prophylactic benefits. However, this in silico work requires experiment validation for confirmation, which will be quickly followed by in vitro and in vivo research to ensure the immunogenicity, wholesomeness, and safety of the potential vaccine. Acknowledgements The authors appreciate the staffs and management of Helix Biogen Institute for providing technical assistance during the cause of the work. Author contribution EKO: Conceptualization, experimental design. OEA, GMA, FOB, ROA, HOA, KTK, JAO, GAA: Data retrieval, analysis, and wrote the first draft of the manuscript. SEO, TOO, OPA, OSH, ATO: Data analysis and result interpretation. EMJ, BAI, AFA, MOI, HMA, UAI: Result interpretation. EOF, OAO, DAA, SAA, FDO, OPO, OOB, SBA, MOB: Manuscript review and editing; all authors participated in the review of the final edition of the manuscript. Data availability statement All data analyzed in this study are publicly available data at the National Centre for Biotechnology Information GenBank repositories (https://www.ncbi.nlm.nih.gov) and which could be made available upon request. Declarations Competing of interest The authors declare no competing interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abraham Peele K Srihansa T Krupanidhi S Ayyagari VS Venkateswarulu TC Design of multi-epitope vaccine candidate against SARS-CoV-2: a in-silico study J Biomol Struct Dyn 2021 39 10 3793 3801 10.1080/07391102.2020.1770127 32419646 Adam KM Immunoinformatics approach for multi-epitope vaccine design against structural proteins and ORF1a polyprotein of severe acute respiratory syndrome coronavirus-2 (SARS-CoV- 2) Tropical Diseases, Travel Medicine and Vaccines 2021 7 1 1 13 10.1186/s40794-021-00147-1 33436085 Adero W Nyanjom S Mwirigi MK Yattinder B In-silico epitope prediction of small ruminant morbillivirus proteins as potential vaccine candidates East African Agricultural and Forestry Journal 2021 85 1–4 14 14 Adhikari UK, Tayebi M, Rahman MM (2018) Immunoinformatics approach for epitope-based peptide vaccine design and active site prediction against polyprotein of emerging oropouche virus. J Immunol Res 2018 Ali M Nelson AR Lopez AL Sack D PLoS Negl Trop Dis 2015 9 6 e0003832 10.1371/journal.pntd.0003832 26043000 Banerjee S, Majumder K, Gutierrez GJ, Gupta D, Mittal B (2020) Immuno-informatics approach for multi-epitope vaccine designing against SARS-CoV-2. BioRxiv Bui HH Sidney J Dinh K Southwood S Newman MJ Sette A Predicting population coverage of T-cell epitope-based diagnostics and vaccines BMC Bioinformatics 2006 7 153 10.1186/1471-2105-7-153 16545123 CDC (2020) Cholera: Vibrio cholerae infection. Centers for Disease Control and Prevention. Cole Supe´rieure de Biotechnologie de Strasbourg, Boulevard Se´bastien Brandt, 67400 Illkirch, France. Biologicals (2001) 29:209–213. 10.1006/biol.2001.0308, http://www.idealibrary.com Chen J Liu H Yang J Chou KC Prediction of linear B-cell epitopes using amino acid pair antigenicity scale Amino Acids 2007 33 3 423 428 10.1007/s00726-006-0485-9 17252308 Coscolla M, Copin R, Sutherland J, Gehre F, de Jong B, Owolabi O, Gagneux SM (2015) Tuberculosis T cell epitope analysis reveals paucity of antigenic variation and identifies rare variable TB antigens. Cell Host Microbe 18(5):538–548 Dey AK Malyala P Singh M Physicochemical and functional characterization of vaccine antigens and adjuvants Expert Rev Vaccines 2014 13 5 671 685 10.1586/14760584.2014.907528 24702271 Dong R Chu Z Yu F Zha Y Contriving multi-epitope subunit of vaccine for COVID-19: immunoinformatics approaches Front Immunol 2020 11 1784 10.3389/fimmu.2020.01784 32849643 Doytchinova IA Flower DR VaxiJen: a server for prediction of protective antigens, tumour antigens and subunit vaccines BMC Bioinformatics 2007 8 1 1 7 10.1186/1471-2105-8-4 17199892 Dunkin MA (2021) Cholera. WebMD. https://www.webmd.com/a-to-z-guides/cholera-faq Ezediuno LO Onile OS Oladipo EK Majolagbe ON Jimah EM Senbadejo TY Designing multi-epitope subunit vaccine for ocular trachoma infection using Chlamydia trachomatis polymorphic membrane proteins G Informatics in Medicine Unlocked 2021 26 100764 10.1016/j.imu.2021.100764 Farhadi T Nezafat N Ghasemi Y Karimi Z Hemmati S Erfani N Designing of complex multi-epitope peptide vaccine based on omps of Klebsiella pneumoniae: an in-silico approach Int J Pept Res Ther 2015 21 3 325 341 10.1007/s10989-015-9461-0 Flower DR, Macdonald IK, Ramakrishnan K, Davies MN, Doytchinova IA (2010) Computer aided selection of candidate vaccine antigens. Immunome Research 6(Suppl 2):S1. 10.1186/1745-7580-6-S2-S1 Gasteiger E, Hoogland C, Gattiker A, Duvaud S, Wilkins MR, Appel RD, Bairoch A (2005) In John M. Walker (ed): The Proteomics Protocols Handbook, Humana Press pp.571–607 Full text - Copyright Humana Press Ghandadi M An Immunoinformatic strategy to develop new Mycobacterium tuberculosis multi-epitope vaccine Int J Pept Res Ther 2022 28 3 1 14 10.1007/s10989-022-10406-0 Greenwood B The contribution of vaccination to global health: past, present and future Philosophical Transactions of the Royal Society B: Biological Sciences 2014 369 1645 20130433 10.1098/rstb.2013.0433 Grote A, Hiller K, Scheer M, Münch R, Nörtemann B, Hempel DC, Jahn D (2005) JCat: a novel tool to adapt codon usage of a target gene to its potential expression host. Nucleic Acids Res 33(Web Server issue):W526–W531. 10.1093/nar/gki376 Gupta S, Kapoor P, Chaudhary K, Gautam A, Kumar R, Open Source Drug Discovery Consortium and Raghava GP (2013) In silico approach for predicting toxicity of peptides and proteins. PLoS One 8(9):e73957. 10.1371/journal.pone.0073957 Habib PT (2020) Learning from COVID-19 Pandemic: In Silico Vaccine and Cloning Design Against Nipah Virus by Studying and Analyzing the Whole Nipah Virus Proteome Helen HM, Shoubai C (2020) Advances in Vaccines. Current Application of Pharmaceutical Biotechnology 171:155–188. https://www.cdc.gov/cholera/infection-sources.html Holmgren J An update on cholera immunity and current and future cholera vaccines Tropical Medicine and Infectious Disease 2021 6 2 64 10.3390/tropicalmed6020064 33925118 Jones DT Protein secondary structure prediction based on position-specific scoring matrices J Mol Biol 1999 292 2 195 202 10.1006/jmbi.1999.3091 10493868 Jyotisha SS, Qureshi IA (2022) Multi-epitope vaccine against SARS-CoV-2 applying immunoinformatics and molecular dynamics simulation approaches. J Biomol Struct Dyn 40(7):2917–2933 Khan M Khan S Ali A Akbar H Sayaf AM Khan A Wei DQ Immunoinformatics approaches to explore Helicobacter Pylori proteome (Virulence Factors) to design B and T cell multi-epitope subunit vaccine Sci Rep 2019 9 1 1 13 10.1038/s41598-019-49354-z 30626917 Kozakov D Hall DR Xia B Porter KA Padhorny D Yueh C Vajda S The ClusPro web server for protein–protein docking Nat Protoc 2017 12 2 255 278 10.1038/nprot.2016.169 28079879 Larsen MV Lundegaard C Lamberth K Buus S Lund O Nielsen M Large-scale validation of methods for cytotoxic T-lymphocyte epitope prediction BMC Bioinformatics 2007 8 1 1 12 10.1186/1471-2105-8-424 17199892 Lavelle EC Ward RW Mucosal vaccines—fortifying the frontiers Nat Rev Immunol 2022 22 4 236 250 10.1038/s41577-021-00583-2 34312520 Livingston B Crimi C Newman M Higashimoto Y Appella E Sidney J Sette A A rational strategy to design multiepitope immunogens based on multiple Th lymphocyte epitopes J Immunol 2002 168 11 5499 5506 10.4049/jimmunol.168.11.5499 12023344 María RR Arturo CJ Alicia JA Paulina MG Gerardo AO The impact of bioinformatics on vaccine design and development Vaccines 2017 2 3 6 29263864 Nagamune T Design of the linkers which effectively separate domains of a bifunctional fusion protein Protein Eng Des Sel 2001 14 8 529 532 10.1093/protein/14.8.529 Nezafat N Karimi Z Eslami M Mohkam M Zandian S Ghasemi Y Designing an efficient multi-epitope peptide vaccine against Vibrio cholerae via combined immunoinformatics and protein interaction-based approaches Comput Biol Chem 2016 62 82 95 10.1016/j.compbiolchem.2016.04.006 27107181 Nielsen M, Lund O, Buus S, Lundegaard C (2010) MHC class II epitope predictive algorithms. Immunology 130(3):319–28. 10.1111/j.1365-2567.2010.03268.x. Epub 2010 Apr 12. PMID: 20408898; PMCID: PMC2913211 Nielsen M, Andreatta M, Peters B, Buus S (2020) Immunoinformatics: predicting peptide–MHC binding NORD (2021) Rare disease https://rarediseases.org/rare-diseases/cholera/ Oladipo EK, Jimah EM, Irewolede BA, Folakanmi EO, Olubodun OA, Adediran DA, Akintibubo SA, Odunlami FD, Olufemi SE, Ojo TO, Akinro OP, Hezekiah OS, Olayinka AT, Abiala GA, Idowu AF, Ogunniran JA, Ikuomola MO, Adegoke HM, Idowu UA, Akindiya OE, Adelusi TI (2022) Immunoinformatics design of multi-epitope peptide for the diagnosis of Schistosoma haematobium infection. J Biomol Struct Dyn 1–8. Advance online publication. 10.1080/07391102.2022.2111358 Oli AN Obialor WO Ifeanyichukwu MO Odimegwu DC Okoyeh JN Emechebe GO Ibeanu GC Immunoinformatics and vaccine development: an overview ImmunoTargets and Therapy 2020 9 13 10.2147/ITT.S241064 32161726 Parvizpour S Pourseif MM Razmara J Rafi MA Omidi Y Epitope-based vaccine design: a comprehensive overview of bioinformatics approaches Drug Discov Today 2020 25 6 1034 1042 10.1016/j.drudis.2020.03.006 32205198 Poland GA Whitaker JA Poland CM Ovsyannikova IG Kennedy RB Vaccinology in the third millennium: scientific and social challenges Curr Opin Virol 2016 17 116 125 10.1016/j.coviro.2016.03.003 27039875 Rapin N Lund O Bernaschi M Castiglione F Computational immunology meets bioinformatics: the use of prediction tools for molecular binding in the simulation of the immune system PLoS ONE 2010 5 4 e9862 10.1371/journal.pone.0009862 20419125 Saha S Raghava GP Prediction of continuous B-cell epitopes in an antigen using recurrent neural network Proteins 2006 65 1 40 48 10.1002/prot.21078 16894596 Sajjad R Ahmad S Azam SS In silico screening of antigenic B-cell derived T-cell epitopes and designing of a multi-epitope peptide vaccine for Acinetobacter nosocomialis J Mol Graph Model 2020 94 107477 10.1016/j.jmgm.2019.107477 31654980 Sanami S Azadegan-Dehkordi F Rafieian-Kopaei M Salehi M Ghasemi-Dehnoo M Mahooti M Bagheri N Design of a multi-epitope vaccine against cervical cancer using immunoinformatics approaches Sci Rep 2021 11 1 1 15 10.1038/s41598-021-91997-4 33414495 Shahab M Hayat C Sikandar R Zheng G Akter S In silico designing of a multi-epitope vaccine against Burkholderia pseudomallei: reverse vaccinology and immunoinformatics J Genet Eng Biotechnol 2022 20 1 1 12 10.1186/s43141-022-00379-4 34978647 Sharma N, Patiyal S, Dhall A, Pande A, Arora C, Raghava GPS (2021) AlgPred 2.0: an improved method for predicting allergenic proteins and mapping of IgE epitopes. Brief Bioinform 22(4):bbaa294. 10.1093/bib/bbaa294. PMID: 33201237 Somboonwit C Menezes LJ Holt DA Sinnott JT Shapshak P Current views and challenges on clinical cholera Bioinformation 2017 13 12 405 10.6026/97320630013405 29379258 Stratmann T (2015) Cholera Toxin Subunit B as Adjuvant--An Accelerator in Protective Immunity and a Break in Autoimmunity. Vaccines (Basel) 3(3):579–96. 10.3390/vaccines3030579. PMID: 26350596; PMCID: PMC4586468 Stranzl T Larsen MV Lundegaard C Nielsen M NetCTLpan: pan-specific MHC class I pathway epitope predictions Immunogenetics 2010 62 6 357 368 10.1007/s00251-010-0441-4 20379710 ul Qamar MT, Ahmad S, Fatima I, Ahmad F, Shahid F, Naz A, Chen LL (2021) Designing multi-epitope vaccine against Staphylococcus aureus by employing subtractive proteomics, reverse vaccinology and immuno-informatics approaches. Comput Biol Med 132:104389 Validi M Karkhah A Prajapati VK Nouri HR Immuno-informatics based approaches to design a novel multi epitope-based vaccine for immune response reinforcement against Leptospirosis Mol Immunol 2018 104 128 138 10.1016/j.molimm.2018.11.005 30448609 Vita Randi Mahajan Swapnil Overton James A Dhanda Sandeep Kumar Martini Sheridan Cantrell Jason R Wheeler Daniel K Sette Alessandro Peters Bjoern The Immune Epitope Database (IEDB): 2018 update Nucleic Acids Research 2019 47 D1 D339 D343 10.1093/nar/gky1006 30357391 Vita R, Overton JA, Greenbaum JA et al (2015) The immune epitope database (IEDB) 3.0. Nucleic 449 acids research. 43:D405-D412 Waldor MK Mekalanos JJ ToxR regulates virulence gene expression in non-O1 strains of Vibrio cholerae that cause epidemic cholera Infect Immun 1994 62 1 72 78 10.1128/iai.62.1.72-78.1994 7903285 Wang L Geng J Chen L Guo X Wang T Fang Y Belingon B Jiao Wu Li M Zhan Y Shang W Wan Y Feng X Xianghui Li Hu Wang Improved transfer efficiency of supercharged 36 + GFP protein mediate nucleic acid delivery Drug Deliv 2022 29 1 386 398 10.1080/10717544.2022.2030430 35075948 Wieczorek M Abualrous ET Sticht J Álvaro-Benito M Stolzenberg S Noé F Freund C Major histocompatibility complex (MHC) class I and MHC class II proteins: conformational plasticity in antigen presentation Front Immunol 2017 8 292 10.3389/fimmu.2017.00292 28367149 World Health Organization (2022) Disease Outbreak News; Cholera – Malawi. Available at: https://www.who.int/emergencies/disease-outbreak-news/item/2022-DON419 Yang J Zhang Y Protein structure and function prediction using I-TASSER Curr Protoc Bioinformatics 2015 52 1 5 8 10.1002/0471250953.bi0508s52 Zaharieva N Dimitrov I Flower D Doytchinova I Immunogenicity prediction by VaxiJen: a ten-year overview J Proteomics Bioinformatics 2017 10 298 310 Zhou Xiaogen Zheng Wei Li Yang Pearce Robin Zhang Chengxin Bell Eric W. Zhang Guijun Zhang Yang I-TASSER-MTD: a deep-learning-based platform for multi-domain protein structure and function prediction Nature Protocols 2022 17 10 2326 2353 10.1038/s41596-022-00728-0 35931779	36459183	PMC9716527	NO-CC CODE	2022-12-03 23:21:04	no	Immunogenetics. 2022 Dec 2;:1-16	utf-8	Immunogenetics	2,022	10.1007/s00251-022-01282-5	oa_other
==== Front Curr Microbiol Curr Microbiol Current Microbiology 0343-8651 1432-0991 Springer US New York 36459213 3103 10.1007/s00284-022-03103-0 Review Article Insight into the Pathogenic Mechanism of Mycoplasma pneumoniae Hu Jie 1 Ye Youyuan 1 Chen Xinxin 1 Xiong Lu 1 Xie Weimin [email protected] 1 http://orcid.org/0000-0002-6723-8801 Liu Peng [email protected] 12 1 grid.412017.1 0000 0001 0266 8918 Institute of Pathogenic Biology, Basic Medical School, Hengyang Medical School, University of South China, Hengyang Central Hospital, Hengyang, 421001 Hunan China 2 grid.412017.1 0000 0001 0266 8918 Present Address: Hunan Provincial Key Laboratory for Special Pathogens Prevention and Control, University of South China, Hengyang, 421001 Hunan China 2 12 2022 2023 80 1 1428 2 2022 28 10 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Mycoplasma pneumoniae, an obligate parasitic pathogen without cell wall, can cause severe upper and lower respiratory tract symptoms. It is the pathogen of human bronchitis and walking pneumonia, and named community-acquired pneumonia. In addition to severe respiratory symptoms, there are clinical extrapulmonary manifestations in the skin, brain, kidney, musculoskeletal, digestive system, and even blood system after M. pneumoniae infection. Hereby, we comprehensively summarized and reviewed the intrapulmonary and extrapulmonary pathogenesis of M. pneumoniae infection. The pathogenesis of related respiratory symptoms caused by M. pneumoniae is mainly adhesion damage, direct damage including nutrient predation, invasion and toxin, cytokine induced inflammation damage and immune evasion effect. The pathogenesis of extrapulmonary manifestations includes direct damage mediated by invasion and inflammatory factors, indirect damage caused by host immune response, and vascular occlusion. The intrapulmonary and extrapulmonary pathogenic mechanisms of M. pneumoniae infection are independent and interrelated, and have certain commonalities. In fact, the pathogenic mechanisms of M. pneumoniae are complicated, and the specific content is still not completely clear, further researches are necessary for determining the detailed pathogenesis of M. pneumoniae. This review can provide certain guidance for the effective prevention and treatment of M. pneumoniae infection. http://dx.doi.org/10.13039/501100004735 Natural Science Foundation of Hunan Province 2019JJ50493 Liu Peng Research Foundation of Education Bureau of Hunan Province190SJY092 Liu Peng Research Foundation of University of South China190XQD015 Liu Peng Hunan Provincial College Students’ innovation and Entrepreneurship Training ProgramS202110555121 Hu Jie issue-copyright-statement© Springer Science+Business Media, LLC, part of Springer Nature 2023 ==== Body pmcIntroduction Mycoplasmas are the smallest free-living, wall-less and self-replicating prokaryotes having an extremely small genome size of 580–2200 kb [1, 2]. More than 200 species of mycoplasma have been identified in humans, animals, plants and arthropods, but only a few of them have been proven to cause human diseases. The main pathogenic mycoplasmas include Mycoplasma pneumoniae, M. genitalium, M. fermentation, M. hominis, M. penetrans, M. pirum and Ureaplasma urealyticum, which are identified to be responsible for humans and animal diseases in the respiratory tract and urogenital system. Among pathogenic Mycoplasma, M. pneumoniae is the most predominant and intensely studied species. Mycoplasma pneumoniae is one of the main pathogens that cause chronic human respiratory tract disease and pneumonia, especially children and adolescents are most susceptible. Mycoplasma pneumoniae infection is generally self-limiting and mild. However, it may develop into a severe or life-threatening disease in some patients. This atypical pathogen was identified to be responsible for up to 40% community-acquired pneumoniae (CAP) among children over five years of age, and lower respiratory tract infections are considered a common cause of morbidity and mortality in children [1, 3]. Mycoplasma pneumoniae infection are also thought to be associated with chronic lung disease and bronchial asthma [4]. In addition to causing severe lower respiratory tract disease and milder upper respiratory tract symptoms, M. pneumoniae also produces other extrapulmonary diseases and post-infection events. Extrapulmonary complications occur in the skin, kidneys, stomach, intestines, heart, musculoskeletal, brain and blood system, etc., resulting in some unusual clinical symptoms. Central nervous system manifestations are the most common extrapulmonary complications of M. pneumoniae infection, at times be life-threatening. Many studies have addressed that due to the atypical symptoms, no obvious clinical and imaging features of M. pneumoniae infection, the early stage of infection is easily underestimated [5]. The main laboratory diagnosis methods are rapid culture based on throat swabs, PCR and serological test, as well as other laboratory diagnostic methods such as rapid antigen test. Extrapulmonary manifestations often occur in the absence of pneumonia, and both involve independent pathological mechanisms [6]. The extrapulmonary manifestations caused by M. pneumoniae can be explained by three possible mechanisms: direct damage caused by invasion or locally induced inflammatory cytokines, Immune-mediated indirect damage, and vascular occlusion caused by vasculitis or thrombosis. The intrapulmonary infection mechanism includes adhesion, nutrient depletion, invasion, toxin, immune and inflammatory damage [7]. It should be noted that these mechanisms are not considered mutually exclusive, but can act in the patient's body at the same time. Based on the severe infection of M. pneumoniae, we summarized the intrapulmonary and extrapulmonary pathogenesis of M. pneumoniae, and this review will provide a certain reference for the pathogenesis research and treatment strategy of M. pneumoniae infection. Intrapulmonary Infection Mechanism The pathogenesis of M. pneumoniae is complicated. At the initial stage, M. pneumoniae adheres to the host bronchial epithelium through the terminal structure, induces intracellular metabolism and ultrastructural alters in the infected cells. At the same time, M. pneumoniae invades host cells, depletes nutrients and releases CARDS toxin, hydrogen peroxide and superoxide radicals, leading to direct damage. Combined with M. pneumoniae Hape enzymes, lipids, lipoproteins, glycolipids and other components to induce cytokines production, the occurrence of inflammation ultimately causes indirect damage. Furthermore, M. pneumoniae evades the host immune system through its immune evasion mechanism, which may help surviving in the body for a long time and thus cause more serious clinical manifestations (Fig. 1).Fig. 1 Pathogenic mechanisms of M. pneumoniae intrapulmonary infection. a M. pneumoniae adhesion causes cell damage. Adhesins binds to sialoglycoproteins and sulfated glycolipids receptors on host cell surface to obtain nutrition for M. pneumoniae, which induces intracellular metabolism and ultrastructural changes in the infected cells. Additionally, EF-Tu can bind to a range of host molecules (such as fibronectin), increase mycoplasmas attach to tracheal epithelial cells. b M. pneumoniae releases CARD toxin, H2O2 and superoxide radicals into host cell to produce host cytotoxicity. c Inflammation-inducing factors (membrane lipid, lipoprotein, HapE enzyme, nuclease, oxidase GlpO, capsular materials) activate host inflammatory pathways to produce inflammatory damage. d M. pneumoniae produces a nuclease (encoded by MPN491) and an antioxidant enzyme (encoded by MPN668) to degrade NETs and peroxides, respectively. Homologous DNA recombination leads to antigenic variation. Furthermore, IbpM and EF-Tu enable M. pneumoniae evade the host's immune response Adhesion Related Proteins Adhesion is the primary factor and prerequisite for the pathogenicity of M. pneumoniae, this ability depends on a special polarized terminal attachment organelle, some pathogenic factors, such as toxic effects, are based on the adhesion step. Mycoplasma pneumoniae interacts with the host respiratory epithelium by attaching to the surface of the bronchial ciliary epithelium, which induces intracellular metabolism and ultrastructural alters in the infected cells, rearranges the cytoskeleton, and causes nutrition depletion of host cells [5]. Changes in intracellular metabolism such as parallel decrease of the uptake rate of host cells orotic acid and amino acid, and the evident inhibition of ribonucleic acid and protein synthesis. The ultrastructural changes such as an accompanying deterioration in the integrity of the airway lumenal surface membranes and subsequent loss of the epithelial cell cytosol. In addition, the disorder of host cells carbohydrate metabolism, amino acid uptake and protein synthesis, eventually promotes the transmission of pathogens in cells, leading to cilia stagnation, cell death, and coordinating other factors to produce human respiratory symptoms[5]. Mycoplasma pneumoniae tightly binds to the host epithelial cells through its unique attachment organelle, which are considered to mediate cell division, cytoadherence, and cell motility at host cell surface [8]. The main receptors for M. pneumoniae to recognize are sialylated and sulfated oligosaccharide receptors [9] (Fig. 1a). The nature and density of the host receptors can profoundly affect the adhesion and sliding of M. pneumoniae, which in turn affects the pathogenic mechanism and infection outcome [10]. The attachment organelle at a cell polar is a membrane protrusion composed of some nap-like surface structures and an internal core (Fig. 2). It realizes a complex and multi-factor adhesion process through the interaction between the internal network-like cytoskeletal system and the surface adhesion protein [11]. The nap-like surface structure is mainly composed of P1 adhesin, P30, P40 and P90. The internal core structure can be divided into three parts, including a terminal button, paired plates, and a bowl (wheel) complex from the front end, and that it is essential for the formation of an attachment organelle. The main proteins include high-molecular-weight proteins (HMW1, HMW2, HMW3), proteins P65, P200, CpsG, mpn387, Lon protease, P41 and P24, etc. [12] (Fig. 2).Fig. 2 Component proteins of the internal structure and the surface adhesion complex of attachment organelle. The nap-like surface structure composed of the main adhesins (P1 and P30) and accessory proteins (P40 and P90) surrounding the cell membrane. The internal structure is made up of terminal button (HMW2, HMW3, P65), paired plates (HMW1, HMW2, CpsG, HMW3), and a bowl complex (Lon, P24, TopJ, P200, P41, MPN387, HMW2). HMW1, HMW2, HMW3 refer to three high molecular weight (HMW) proteins P1 Membrane protein P1 is considered as the major cellular adhesin, which is surface localized and trypsin sensitive [13]. When M. pneumoniae contacts the target cell, the P1 precursor proteins, which are scattered in the cell membranes, rapidly traffics to the terminal organelle, and the leader peptide on the amino terminal is hydrolyzed to become a mature P1 protein that binds to the host receptor [7]. It should be noted that P1 adhesin can mediate adhesion only if its correct positioned on the terminal organelle. Researches have confirmed that P1 adhesin not only participates in the binding between M. pneumoniae and host receptors to play an adhesion role, but also in gliding on the surface of host cells. Studies have also shown that P1 adhesin plays an important role in the mast cells cytokine response induced by M. pneumoniae, the mast cells are activated to cause inflammation damage by the direct contact between M. pneumoniae and the sialylated residues on the surface of mast cells. The antibodies against the highly immunogenic carboxyl terminus of P1 produced by the humoral immunity are considered to reduce M. pneumoniae adhering to non-biological and host cells [13]. P30 Studies have found that there is a degree of sequence homology between some specific domains of the P30 protein and the P1 protein, both representing the dominant proteins responsible for adherence. P30 adhesin located at the tip of attachment organelle, plays a crucial role in conveying signals from the cell interior to the exterior to activate key steps in cytoadherence and motility, such as the arrangement of P1 adhesin complex and the binding between P1 and host receptors [12]. Romero-Arroyo et al. have proved that P30 is instrumental to cell development, the loss of P30 could lead to abnormal mycoplasma morphology, including oval or leafy cells with poorly defined apical structures, while transformation of P30 mutants and wild-type P30 alleles can restore their normal morphology. P116 P116 protein was verified to be surface exposed and considered as a crucial cell adhesin because the anti-P116 antibody has been shown to prevent attachment of M. pneumoniae to the HEp-2 cells independently of P1. P116 has also been identified as an important immunogenic antigen of M. pneumoniae [5]. The overall level of P116-C-terminal protein has been used for serological diagnosis of M. pneumoniae, and Tabassum et. al have shown that an N-terminal 27 kDa fragment of P116 protein also held a promise for serodiagnosis of M. pneumoniae infection. Other Accessory Proteins The protein P65 have a close spatial and functional relationship with P30. The analysis of the P65 and P30 fluorescent fusion proteins expressed by the growing mycoplasma culture showed that they were situated at the developing terminal organelles almost concurrently. P65 might interact with the internal structural domain of P30 to achieve a close combination between the terminal button and the front side of the membrane [12]. P40 and P90 are adhesins produced by the cleavage of mpn142, which form a transmembrane adhesion complex with protein P1 [13]. The presence of accessory proteins is essential for the formation of functional attachment organelle. Vizarraga has reported that the binding site for sialic acid was found in P40/P90 and not in P1, genetic variability of the N-terminal domain surfaces of P1 and P40/P90 results in clinical symptoms variability, these founds provide new strategies in vaccine development against M. pneumoniae infections [14]. Studies have shown a function for HMW (1, 2, 3) proteins in the architecture and stability of attachment organelles, also HMW (1, 2, 3) proteins are related to adherence and gliding, participating in the correct positioning of adhesins and the maintenance of cell morphology. Protein P200 was speculated as an accessory structural component in cytoadherence, since it shared several unusual features with the proteins HMW1 and HMW3. However, P200 was considered more essential for motility rather than adherence, and associated with biofilms and cells maturation [15]. P41/P24 are involved in anchoring the terminal organelles on the cell body, both of them play a significant role in the assembly and development of M. pneumoniae attachment organelles. Direct Damage Direct damage refers to damage to host cells caused by M. pneumoniae, instead of inflammatory or immune-mediated injury caused by M. pneumoniae infection. The direct damage includes nutrition depletion, intracellular localization, toxin, oxidative damage and induction of apoptosis. Nutrition Depletion Mycoplasma pneumoniae depends on host cells to supply necessary nutrients for their survival and development, since their small genome and limited biosynthetic capabilities. The cell membrane of M. pneumoniae can closely contact with the host cell membrane, which promotes the exchange of compounds essential for their growth and proliferation [1]. In addition, it is thought to be able to absorb nutrients such as glucose, cholesterol and amino acids by inserting microtubules into host cells (Fig. 1b). Intracellular Localization The recently elucidated genomic structure of M. pneumoniae strongly suggests that this organism might have undergone a unique and reductive genetic evolution, such as other intracellular bacteria, which indicates this pathogen may be a highly specialized parasitic bacteria in respiratory tissue cells, and provides preliminary evidence for the intracellular invasion of M. pneumoniae. In fact, some experimental data and studies have concluded that M. pneumoniae may have intracellular permeability and the ability to penetrate the host cell membrane for nutrients acquisition [6] (Fig. 1b). CARDS Toxin Community-Acquired Respiratory Distress Syndrome Toxin (CARDS) is a protein shared significant sequence homologies with the S1 subunit of pertussis toxin, which cause clinical symptoms like pertussis. CARDS, encoded by M. pneumoniae mpn372, is a unique ADP ribosylating and vacuolating toxin, and the maintenance of these activities requires disulfide bond [16]. Recent study has shown that CARDS toxin binds to SP-A receptors of host target cells (Fig. 1b) and is internalized rapidly in a dose and time-dependent manner by clathrin-mediated pathway. After internalization, CARDS toxin is transported in a retrograde manner from endosome through the golgi complex into the endoplasmic reticulum. Moreover, retrograde transport facilitates toxin clipping and is required to induce vacuole formation [17]. An acidic environment in host cell intracellular vesicle is considered essential for clipping, trafficking and translocation of M. pneumoniae CARDS toxin. Regulating the acidic environment in host cells may open new possibilities to protect host target cells against M. pneumoniae CARDS toxin-induced vacuolation [17]. Oxidative Damage After adhering to the host cell, M. pneumoniae inserts the microtubule into the host cell and releases hydrogen peroxide and superoxide radicals. These substances and the endogenous toxic oxygen molecules produced by the host cell cause respiratory tract epithelial cells oxidative stress. Moreover, M. pneumoniae lacks superoxide dismutase and catalase, and the superoxide free radicals produced by M. pneumoniae can also inhibit the activity of catalase in the host cell, both leading to reduced decomposition of peroxides and making host cells more sensitive to the toxic effects of oxygen molecules (Fig. 1b). The study results of Yamamoto suggested that M. pneumoniae might develop a mechanism to regulate infected cell detachment by producing hydrogen peroxide, which may contribute to sustaining the bacterial infection [18]. As a wall-less bacteria of the genus Mycoplasma, glycerol derived from phospholipids of animal or human hosts is the major source of carbon and energy [19]. L-α-glycerophosphate oxidase (GlpO), a surface-exposed enzyme involved in the metabolism of glycerol, is responsible for the production of hydrogen peroxide during glycerol metabolism and important for the pathogenesis of M. pneumoniae (Fig. 1c) [20]. However, Melanie et al. have demonstrated that the GlpO as a candidate vaccine antigen is unlikely to induce a protective immune response [20]. In addition, HPrK, a key regulator of carbon metabolism in many Gram-positive bacteria, is also one of the nine known regulatory proteins encoded by the M. pneumoniae genome, the HPrK can be activated by glycerol and induced oxidative stress by the production of peroxides. Oxidative Damage ultimately leads to respiratory epithelial cells alters, such as cilia loss, vacuolar degeneration, reduced oxygen consumption, decreased glucose utilization, amino acid uptake and macromolecular synthesis. Inducing Apoptosis Lung macrophages play an important role in controlling M. pneumoniae infection. They can recognize M. pneumoniae through TLR2, activate MyD88-NF-κB signaling pathway and phagocytose M. pneumoniae. MyD88 is the major adaptor molecule for signaling downstream of TLR, MyD88 signaling is essential for macrophage response to M. pneumoniae in the lung. At the same time, the activation of the NF-κB pathway can cause severe inflammation, induce the apoptosis of monocytes, macrophages and lymphocytes, and ultimately reduce immune function (Fig. 1b) [21]. Inflammation Injury The cellular components including metabolites and toxin released by M. pneumoniae could act as pro-inflammatory molecules to stimulate the inflammatory response. The release of M. pneumoniae-induced cytokines such as gamma interferon (IFN-γ), tumor necrosis factor alpha (TNF-α), and interleukins (including interleukins-1β[IL-β], IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-18) are thought to be related to asthma exacerbation. The changes of cytokines may be one of the pathogenic mechanisms of M. pneumoniae infection. HapE The HapE is an enzyme and potential virulence factor of M. pneumoniae, which can produce H2S to damage blood cells [4] (Fig. 1c). Further, H2S produced by HapE has been demonstrated to induce phagocytes to secrete pro-inflammatory factors. The up-regulated expression of various inflammatory mediators and cytokines can activate and aggravate inflammatory reactions, leading to tissue damage [22]. In addition, this enzyme also mediates inflammatory reactions via adenosine triphosphate (ATP)-sensitive K+ (KATP) channels [5] (Fig. 1c). HapE can degrade cysteine within the KATP channel complex to form H2S, enhanced production of H2S alters cellular excitability via modulation of ion channel function and exacerbates inflammation. Studies have shown that mutants with defective HapE genes cannot be isolated, hence suggesting that HapE gene is essential for the development of M. pneumoniae. Lipid and Lipoproteins Lipid is considered to bind to TLR4 as potential TLR4 ligands, stimulating macrophage autophagy, and then promoting the activation of NLRP3 inflammasomes and NF-κB pathway by inducing ROS production, ultimately leading to the secretion of pro-inflammatory cytokines [23] (Fig. 1c). Lipoproteins are thought to be recognized by TLR2 and TLR6/TLR1 through extracellular leucine repeat sequence region, which induces cytokines production and the expression of mediators in immune cells (Fig. 1c). A recent study by Mara et al. has reported that after inoculating BALB/c mice with M. pneumonia, lipid-associated membrane proteins (LAMPs) induce lung lesions consistent with exacerbated disease following challenge, while removing the lipid moieties from LAMP before vaccination could eliminate symptoms. The overall results can be concluded that the lipid moieties of the lipoproteins are the causative factors of M. pneumoniae vaccine-enhancing disease (VED) [24]. CARDS CARDS activates NLRP3-related inflammasomes to regulate the activation of caspase-1, promotes the maturation and release of IL-1β and IL-18 in the natural immune defense, which results in inflammatory cell death and stress pathological conditions (Fig. 1c). At the same time, CARDS toxin can increase the expression of pro-inflammatory cytokines IL-6 and TNF-α in a dose and activity-dependent manner [11, 25]. Gang et al. has found that CARDS toxin was positively correlated with TNF-α level in refractory M. pneumoniae pneumonia (RMPP) cases. Therefore, CARDS provides a good diagnostic biomarker for differentiating children with RMPP and non-RMPP(NRMPP) [26]. Nuclease Studies have demonstrated the nuclease of parasitic Mollicutes bacteria may contribute to host pathology by nuclease catalyzing reactions which help M. pneumoniae plunder nucleic acids from host nucleic acid precursors or DNA for survival in the host. The lipoprotein, encoded by M. pneumoniae mpn133, participates in the uptake of free glycerol and acts as a calcium-dependent nuclease to degrade DNA and RNA which causes programmed death of host cells, local inflammatory cell infiltration and tissue damage (Fig. 1c) [4, 5]. Others Glycolipid and capsule are considered as potential virulence factors, but the pathogenic mechanisms are unclear and needs to be further explored [5]. Histone deacetylase 5 (HDAC5) is involved in the regulation of inflammation, which may promote M. pneumoniae-induced inflammatory response in macrophages through NF-κB activation [27]. Immune Evasion IbpM Mpn400 protein is a surface protein that binds strongly to various immunoglobulins (IgM, IgA and IgG) produced by the host [5]. Therefore, mpn400 protein was named as immunoglobulin binding protein of Mycoplasma (IbpM). Study has demonstrated that M. pneumoniae strains lacking IbpM are slightly impaired in terms of cytotoxicity and thus IbpM is considered to be a virulence factor [28] (Fig. 1d). Intracellular Invasion Mycoplasma pneumoniae is thought to invade cells and tissues of the body, parasitize inside the cell to escape the phagocytosis of immune cells and the effect of antibiotics, and finally survive in the body for a long time. However, the mechanisms by which M. pneumoniae evade host defense while intercellular remain unknown, and the pathways related to intracellular survival remain to be elucidated. Intracellular invasion may be responsible for M. pneumoniae evasion, long-term incubation in the host body and establishment of chronic infection (Fig. 1d). Antigen Polymorphism and Variation In spite of limited size, the genomes of M. pneumoniae consist of a significant portion of repeated elements, which are dispersed throughout the genome and constitute approximately 8% of the genome. It has been demonstrated that these genes are sufficient to generate antigenic variation by homologous recombination between specific repetitive genomic elements [29] (Fig. 1d). Studies have implicated that the repetitive sequences serve as a reservoir for antigenic variation generation of P1 adhesin genes in M. pneumoniae, and the production of recombinant P1 adhesin is essential for adherence and motility of M. pneumoniae [30]. Due to mutation and rearrangement of M. pneumoniae surface antigens, deficiency of protective antibodies in the host after M. pneumoniae infection is considered responsible for repeated M. pneumoniae infection. The Mechanism of Oxide Degradation The production of reactive oxygen species (ROS) is a part of the host cell non-specific immune defense against invading pathogens [31]. ROS is produced by NADPH oxidase in the host. On the one hand, ROS is directly antimicrobial, which targets on nucleic acids, carbohydrates, lipids and proteins in M. pneumoniae cells, and causing considerable damage to these biological macromolecules. On the other hand, ROS is also an essential signal for innate immune signaling, which activates the innate immune system to fight against pathogens. Therefore, M. pneumoniae needs to evolve a mechanism to deal with this oxidative challenge. Recent studies have found that the protein encoded by mpn668 is a protective antioxidant enzyme in M. pneumoniae (Fig. 1d), which may degrade hydroperoxide and limit the oxidative damage exerted by the host [32]. Nuclease After infected with M. pneumoniae, neutrophils can rapidly accumulate at the infected area by chemokines chemotaxis, become highly phagocytic and eventually undergo morphological changes, which lead to Neutrophil extracellular traps (NET) and a variety of bactericidal substances release, ultimately effectively eliminate pathogenic microorganisms. M. pneumoniae produces some extracellular nucleases capable of degrading NETs. The magnesium-dependent nuclease encoded by M. pneumoniae mpn491 is a major extracellular nuclease which improves the survival rate of M. pneumoniae and helps the pathogen to escape the host immune response by degrading NETs [33], resulting in further damage to the host (Fig. 1d). Elongation Factor Thermo Unstable (EF-Tu) Factor H is a negative regulator of the complement system in host, which helps avoiding unexpected complement activation [34]. C3 convertase cleaves complement component C3 into C3b, which is the main effector molecule of the complement system that can furtherly activate the complement system. Yanfei Yu et al. have revealed that M. hyopneumoniae could bind factor H via EF-Tu, which contributes to decreased C3 deposition on the M. hyopneumoniae surface, and ultimately blocks further complement activation. Meanwhile, many mycoplasmas, including M. pneumoniae, could hijack factor H via EF-Tu and then simulate host molecular to escape from complement attack [34] (Fig. 1d). In addition to helping mycoplasmas escape from complement killing, EF-Tu also strengthens adherence between mycoplasmas and tracheal epithelial cells (Fig. 1a). Immunity Disorder M. pneumoniae infection can cause innate immunity and adaptive immunity disorder in host. Stelmach et.al have shown that there were no increases of IgG, IgM, and IgA immunoglobulins in M. pneumoniae infected patients during a 1-year observation, which indicated immunity damage caused by M. pneumoniae infection. The levels of C3 and C4 increased significantly in the acute stage of infection, while immune suppression caused by M. pneumoniae infection could lead C3 and C4 to a normal or lower level in the later stage. In addition, M. pneumoniae infection induces pro-inflammatory cytokines and chemokines released in respiratory tract and activates variety of immune cells, causing T cells overactive exhausted and access to the verge of apoptotic progress [35]. Furthermore, adhesins and metabolites of M. pneumoniae can cause immune damage to respiratory epithelial cells lymphocytes, resulting in decreased activity and accelerated apoptosis of lymphocytes. Mainly, the decrease of CD4+ function leads to the imbalance of immune function in patients infected by M. pneumoniae, ultimately causing antigen presentation disturbance, B cells maturation disorder, and the relative reduction of antibody production. The damage of respiratory system is further aggravated by disorders of humoral immunity, cellular immunity and innate immune system caused by M. pneumoniae. Extrapulmonary Infection Mechanism In addition to typical respiratory symptoms, M. pneumoniae can also cause some extrapulmonary complications. Importantly, extrapulmonary manifestations due to M. pneumoniae infection sometimes occur in the absence of pneumonia and even respiratory symptoms [36]. There is a myriad of extrapulmonary manifestations of M. pneumoniae infection that can potentially involve all systems and organs [2, 37]. The concomitant occurrence of mycoplasmaemia was obtained in the mycoplasmal central nervous system involvement, which proved M. pneumoniae could transfer to distant organs by blood transmission to cause disease. The extrapulmonary pathogenic mechanisms of M. pneumoniae can be divided into three parts: direct damage, indirect damage and vascular occlusion (Fig. 3). Early-onset extrapulmonary complications may be related to direct damage caused by the blood transmission of M. pneumoniae, while late-onset disease may be associated with indirect damage caused by autoimmunity, vascular damage and drug reaction.Fig. 3 Pathogenic mechanisms of M. pneumoniae extrapulmonary infection. a Direct invasion and inflammatory damage induced by cytokines lead to direct damage to host cells. b M. pneumoniae antigens mimic host cell components or cause changes in the structure of host cell membrane antigens to stimulate host autoimmunity. Immune complex deposition is considered responsible for M. pneumoniae extrapulmonary infection. Self-reactive IgEs promote the occurrence of allergic reactions and cause certain damage to tissue cells. c M. pneumoniae locally induces cytokines and chemokines to affect the vascular wall, causing vasculitis and thrombotic vascular occlusion through medical mediators (such as complement and fibrin D-dimer) Direct Damage M. pneumoniae existed in blood, pericardial fluid, synovial fluid and skin lesions by PCR and culture testing. Therefore, there is the possibility of direct invasion and damage in the extrapulmonary pathogenesis caused by M. pneumoniae. Some patients with an immature or damaged immune barrier on the respiratory tract surface are not sufficient to develop into pneumonia after M. pneumoniae infection, and the pathogen can be passively transferred into the circulation through weak gaps between injured lungs epithelial cells [38]. Since erythrocytes carry sialoglycoproteins, M. pneumoniae is of hemadsorption and has the ability to absorb to erythrocytes. Therefore, M. pneumoniae is considered to cause systemic infection after invading into the blood system (Fig. 4). The occurrence of Mycoplasma bacteremia constitutes a direct extrapulmonary manifestation.Fig. 4 M. pneumoniae can be passively transferred into the circulation through weak gaps between injured epithelial cells in some patients with an immature or damaged immune barrier on the respiratory tract surface and transfer to other organs by adhering to erythrocytes to cause extrapulmonary infection There are two possible forms of direct injury, firstly, the direct M. pneumoniae invasion outside the respiratory tract (Fig. 3a). Since early-onset hepatitis sometimes develops in the absence of pneumoniae, it may be associated with a direct-type extrapulmonary manifestation [39]. Mycoplasma pneumoniae is speculated to directly colonize and infect liver epithelial cells, but this situation has not been proven. Secondly, inflammatory damage induced by M. pneumoniae (Fig. 3a). In tissues riched in cytokine-producing cells, the membrane lipoprotein of M. pneumoniae can induce local cytokines production which leads to inflammatory damage in tissues and organs [25]. According to reports, IL-17 is an important immune mediator in the systemic immune response, which may be related to the disease severity and extrapulmonary pathogenesis [40]. Indirect Damage It has been hypothesized that recognition of M. pneumoniae by innate immune cells and consequent activation of the cells may be considered as main candidates to induce some serious M. pneumoniae complications [41]. Fink et al. using indirect immunofluorescence and PCR detection for serum IgM, IgA, IgG and cerebrospinal fluid (CSF) samples of patients with sudden neurological diseases. They found that the damage to the nervous system did not seem to be caused by direct invasion of M. pneumoniae, and most likely an immune response to infection. Immune-mediated mechanisms have been mainly implicated in Mycoplasma pneumoniae-related extrapulmonary diseases (MpEPDs) [6]. Autoimmunity Caused by Molecular Mimicry Mycoplasma pneumoniae antigen could mimic host cell components and cause shifts in the structure of host cell membrane antigens to activate auto-immune responses which forms immune complexes with corresponding organs to activate complements, producing neutrophil chemotaxis factors and C3a, C5a, C3b. A large number of white blood cells infiltrate the diseased site, release the hydrolase in the lysosome, cause destructive injuries and disease in multiple organs (Fig. 3b). For example, the P1 and P30 proteins on the attachment organelles of M. pneumoniae show high levels of homology to troponin, cytoskeletal proteins, keratin and fibrinogen of the host. Antibodies in response to M. pneumoniae infections target various host tissues and form immune complexes to cause damage in various tissues and organs such as liver, kidney, brain, smooth muscle and lungs [5]. Conclusively, Autoimmunity plays an important role in extrapulmonary complications caused by M. pneumoniae. Immune Complex Deposition In cases of M. pneumoniae infection complicated by acute nephritis and renal failure, there are reports that the genome of M. pneumoniae and immune complexes containing M. pneumoniae antigens have been detected in the glomerulus, it may be associated with excessive immune complexes deposition and complement activation in the tissues (Fig. 3b). Circulating immune complexes are responsible for the pathological mechanism of glomerulonephritis and IgA nephropathy related to M. pneumoniae infection. Non-Specific Antibody M. pneumoniae can activate B lymphocytes and produce non-specific polyclonal antibodies that are not directly against M. pneumoniae. The experiment of Sauteur et al. has shown that the level of serum antibodies against M. pneumoniae proteins and glycolipids arise in M. pneumoniae-infected children and mice. The equal recovery of serum antibodies level from M. pneumoniae infection in Btk-deficient (a mice species developed M. pneumoniae-specific IgG responses to M. pneumoniae proteins but not to M. pneumoniae glycolipids) and wild-type mice suggests that pulmonary M. pneumoniae clearance is mainly mediated by IgG reactive with M. pneumoniae proteins, and M. pneumoniae glycolipid-specific IgG or IgM is not essential [42]. Cold agglutinins, a kind of IgM antibodies, are produced in 50% patients infected by M. pneumoniae and may persist for several weeks. Cold agglutinins can be used to confirm clinical suspicions of primary atypical pneumonia caused by M. pneumoniae. One theory is that cold agglutinins are the result of cross-reactive autoantibodies developed against the M. pneumoniae glycolipid antigen and I antigen of human erythrocytes during acute M. pneumoniae infection. This non-specific antibody can cause auto-immune hemolytic anemia (cold agglutinin disease), which is the most famous indirect extrapulmonary manifestation caused by M. pneumoniae infection. Atopy and Elevated IgE Levels The body with atopy refers to an inherited tendency can produce IgE antibodies in response to small amounts of common environmental proteins [40]. Patel et al. observed 162 hospitalized children and found a significant increase in the total serum IgE level of children infected by MpEPDs, and it was significantly higher than that of children with only classic Mycoplasma pneumoniae-related respiratory illnesses, which indicates the existence of atopy in MpEPDs children [43]. The incidence of atopy in patients with extrapulmonary manifestations is higher than that of patients without extrapulmonary manifestations, so atopy may be associated with MpEPDs [40, 44]. Increased IgE levels are considered to be a sign of immune disorders. Self-reactive IgEs aggravate immune-mediated diseases and manifestations such as allergic reactions (Fig. 3b). For example, after infected by M. pneumoniae, the P1 protein could induce the production of P1-specific IgE in patients allergic to M. pneumoniae, and ultimately resulting in allergic symptoms and tissue damage [45]. Some patients who are prone to produce IgE, may be predisposed to develop extra-respiratory diseases associated with M. pneumoniae acute infections [46]. Vascular Occlusion Extrapulmonary manifestations are not only directly related to the infection process and auto-immune, but also vascular complications. Thrombosis can appear in a vessel of any part of the body, pulmonary vessels are the most commonly involved sites, and accordingly chest pain was the most common symptom, followed by neurological symptoms and abdominal pain [47]. Jacobs et al. has reported a new case of pediatric priapism and proved that this symptom may be an extremely rare but reasonable type of vascular occlusion produced by M. pneumoniae infection. The occurrence of extrapulmonary manifestations caused by vascular occlusion involves direct and indirect mechanisms. The direct type is that M. pneumoniae can be hematogenously transferred to distant organs, locally induced cytokines and chemokines (including TNF-α and IL-8) to affect the vascular wall, and eventually lead to local vasculitis or thrombosis without systemic hypercoagulability (Fig. 3c). The indirect type is systemic hypercoagulability through the activation of chemical mediators including complement and fibrin D-dimer, which may result in thrombotic vessel occlusion (Fig. 3c). Liu Jingwei has reported that some of the factors causing thrombosis are transient, and some are due to hereditary thrombophilia in patients with thrombosis caused by M. pneumoniae infection [48]. On the other side, some substances and phenomena have been suggested to increase the risk of thrombosis but transient, including cold agglutinin, vascular malformations, sickle cell trait, positive for anticardiolipin antibodies, β2-glycoprotein antibodies, lupus anticoagulant antibodies and anti-prothrombin antibodies. (Fig. 3c) [48, 49]. Others The M. pneumoniae superantigen may be one of the factors of extrapulmonary manifestations. Superantigen produced by several bacteria could stimulate the production of a large number of T lymphocytes and lipid-related membrane proteins, leading to an uncontrolled immune response like the damage of Kawasaki disease [50]. Conclusion Over the past few years, the understanding of the mechanisms by which M. pneumoniae causes intrapulmonary and extrapulmonary manifestations has gradually increased. The occurrence of related disease is the result of multiple pathogenic factors. Whether respiratory symptoms or extrapulmonary manifestations, although both are independent of each other, there are still similarities such as direct invasion, inflammatory damage and immune-mediated damage. The development of drugs targeting for the common pathogenic mechanism of intrapulmonary and extrapulmonary infection may support the M. pneumoniae infection prevention. For extrapulmonary complications caused by M. pneumoniae, more researches are required to construct a comprehensive treatment strategy including microbiology (antibiotics), hematology (anticoagulants) and immunology (immunomodulators). In addition, exploring alternative treatment options for macrolide antibiotics may improve the clinical symptoms of macrolide treatment failure. Vaccine development is the first choice for any infected disease control strategy. A safe vaccine that can provide protective immunity is essential to reduce M. pneumoniae infection. Coinfection with M. pneumoniae occurred in patients infected with other common respiratory pathogens. Clinicians managing patients with COVID‐19 infection should be mindful of coinfections with M. pneumoniae, which may exacerbate clinical symptoms during this COVID‐19 outbreak. To explore the synergistic mechanism between M. pneumoniae and other pulmonary pathogens may provide some strategies for prevention and treatment of M. pneumoniae co-infection. Abbreviations CARDS Community-acquired respiratory distress syndrome toxin GlpO L-α-glycerophosphate oxidase KATP channels ATP-sensitive K+ channels ROS Reactive oxygen species LAMPS Lipid-associated membrane proteins HDAC 5 Histone deacetylase 5 IbpM Immunoglobulin binding protein of Mycoplasma NET Neutrophil extracellular traps EF-Tu Elongation factor thermo unstable MpEPDs Mycoplasma pneumoniae-related extrapulmonary diseases Author Contributions JH drafted the original manuscript. YY modified the manuscript. XC and LX prepared a part of graphic and text materials. WX supervised the writing. PL conceived the idea. All authors have read and approved the manuscript. Funding This work was supported by the Natural Science Foundation of Hunan Province, China [2019JJ50493]; Research Foundation of Education Bureau of Hunan Province, China [190SJY092]; Research Foundation of University of South China [190XQD015]; and Hunan Provincial College Students’ innovation and Entrepreneurship Training Program [S202110555121]. Declarations Conflict of interest The authors have no conflict of interest to declare. Consent to Participate Not applicable. Consent to Publication Not applicable. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Kumar S Mycoplasma pneumoniae: a significant but underrated pathogen in paediatric community-acquired lower respiratory tract infections Indian J Med Res 2018 147 1 23 31 10.4103/ijmr.IJMR_1582_16 29749357 2. Bajantri B Venkatram S Diaz-Fuentes G Mycoplasma pneumoniae: a Potentially Severe Infection J Clin Med Res 2018 10 7 535 544 10.14740/jocmr3421w 29904437 3. Kutty PK Jain S Taylor TH Bramley AM Diaz MH Ampofo K Arnold SR Williams DJ Edwards KM McCullers JA Pavia AT Winchell JM Schrag SJ Hicks LA Mycoplasma pneumoniae among children hospitalized with community-acquired pneumonia Clin Infect Dis 2019 68 1 5 12 29788037 4. Waites KB Xiao L Liu Y Balish MF Atkinson TP Mycoplasma pneumoniae from the respiratory tract and beyond Clin Microbiol Rev 2017 30 3 747 809 10.1128/CMR.00114-16 28539503 5. Jiang Z Li S Zhu C Zhou R Leung P Mycoplasma pneumoniae Infections: pathogenesis and vaccine development Pathogens 2021 10 2 119 10.3390/pathogens10020119 33503845 6. Poddighe D Extra-pulmonary diseases related to Mycoplasma pneumoniae in children: recent insights into the pathogenesis Curr Opin Rheumatol 2018 30 4 380 387 10.1097/BOR.0000000000000494 29432224 7. He L Ye T Liu Y Zeng Wu Insights into the pathogenesis of Mycoplasma pneumoniae (Review) Mol Med Rep 2016 14 5 4030 4036 10.3892/mmr.2016.5765 27667580 8. Nakane D Murata K Kenri T Shibayama K Nishizaka T Molecular ruler of the attachment organelle in Mycoplasma pneumoniae PLoS Pathog 2021 17 6 e1009621 10.1371/journal.ppat.1009621 34111235 9. Williams CR Chen L Sheppard ES Chopra P Locklin J Boons GJ Krause DC Distinct Mycoplasma pneumoniae interactions with sulfated and sialylated receptors Infect Immun 2020 10.1128/IAI.00392-20 32839185 10. Williams CR Chen L Driver AD Arnold EA Sheppard ES Locklin J Krause DC Sialylated receptor setting influences Mycoplasma pneumoniae attachment and gliding motility Mol Microbiol 2018 109 6 735 744 10.1111/mmi.13997 29885004 11. Chaudhry R Ghosh A Chandolia A Pathogenesis of Mycoplasma pneumoniae: an update Indian J Med Microbiol 2016 34 1 7 16 10.4103/0255-0857.174112 26776112 12. Miyata M Hamaguchi T Integrated information and prospects for gliding mechanism of the pathogenic bacterium Mycoplasma pneumoniae Front Microbiol 2016 7 960 10.3389/fmicb.2016.00960 27446003 13. Widjaja M Berry IJ Jarocki VM Padula MP Dumke R Djordjevic SP Cell surface processing of the P1 adhesin of Mycoplasma pneumoniae identifies novel domains that bind host molecules Sci Rep 2020 10 1 6384 10.1038/s41598-020-63136-y 32286369 14. Vizarraga D Kawamoto A Matsumoto U Illanes R Pérez-Luque R Martín J Mazzolini R Bierge P Pich OQ Espasa M Sanfeliu I Esperalba J Fernández-Huerta M Scheffer MP Pinyol J Frangakis AS Lluch-Senar M Mori S Shibayama K Kenri T Kato T Namba K Fita I Miyata M Aparicio D Immunodominant proteins P1 and P40/P90 from human pathogen Mycoplasma pneumoniae Nat Commun 2020 11 1 5188 10.1038/s41467-020-18777-y 33057023 15. Feng M Schaff AC Cuadra Aruguete SA Riggs HE Distelhorst SL Balish MF Development of Mycoplasma pneumoniae biofilms in vitro and the limited role of motility Int J Med Microbiol 2018 308 3 324 334 10.1016/j.ijmm.2018.01.007 29426802 16. Balasubramanian S Pandranki L Maupin S Ramasamy K Taylor AB Hart PJ Baseman JB Kannan TR Disulfide bond of Mycoplasma pneumoniae community-acquired respiratory distress syndrome toxin is essential to maintain the ADP-ribosylating and vacuolating activities Cell Microbiol 2019 21 8 e13032 10.1111/cmi.13032 30977272 17. Ramasamy K Balasubramanian S Kirkpatrick A Szabo D Pandranki L Baseman JB Kannan TR Mycoplasma pneumoniae CARDS toxin exploits host cell endosomal acidic pH and vacuolar ATPase proton pump to execute its biological activities Sci Rep 2021 11 1 11571 10.1038/s41598-021-90948-3 34078958 18. Yamamoto T Kida Y Kuwano K Mycoplasma pneumoniae protects infected epithelial cells from hydrogen peroxide-induced cell detachment Cell Microbiol 2019 21 6 e13015 10.1111/cmi.13015 30702185 19. Blötz C Stülke J Glycerol metabolism and its implication in virulence in Mycoplasma FEMS Microbiol Rev 2017 41 5 640 652 10.1093/femsre/fux033 28961963 20. Schumacher M Nicholson P Stoffel MH Chandran S D'Mello A Ma L Vashee S Jores J Labroussaa F Evidence for the cytoplasmic localization of the L-α-glycerophosphate oxidase in members of the "Mycoplasma mycoides Cluster" Front Microbiol 2019 10 1344 10.3389/fmicb.2019.01344 31275271 21. Segovia JA Chang TH Winter VT Coalson JJ Cagle MP Pandranki L Bose S Baseman JB Kannan TR NLRP3 Is a critical regulator of inflammation and innate immune cell response during Mycoplasma pneumoniae infection Infect Immun 2018 86 1 e00548 10.1128/IAI.00548-17 29061706 22. Li S Xue G Zhao H Feng Y Yan C Cui J Sun H 2019 The Mycoplasma pneumoniae HapE alters the cytokine profile and growth of human bronchial epithelial cells Biosci Rep 10.1042/BSR20182201 23. Luo H He J Qin L Chen Y Chen L Li R Zeng Y Zhu C You X Wu Y Mycoplasma pneumoniae lipids license TLR-4 for activation of NLRP3 inflammasome and autophagy to evoke a proinflammatory response Clin Exp Immunol 2021 203 1 66 79 10.1111/cei.13510 32894580 24. Mara AB Gavitt TD Tulman ER Geary SJ Szczepanek SM Lipid moieties of Mycoplasma pneumoniae lipoproteins are the causative factor of vaccine-enhanced disease npj Vaccines 2020 5 1 31 10.1038/s41541-020-0181-x 32284882 25. Mitsuo N Classification of extrapulmonary manifestations due to Mycoplasma pneumoniae infection on the basis of possible pathogenesis Fron Microbiol 2016 7 23 26. Li G Fan L Wang Y Huang L Wang M Zhu C Hao C Ji W Liang H Yan Y Chen Z High co-expression of TNF-α and CARDS toxin is a good predictor for refractory Mycoplasma pneumoniae pneumonia Mol Med 2019 25 1 38 10.1186/s10020-019-0105-2 31399022 27. Zhao Y Ma G Yang X HDAC5 promotes Mycoplasma pneumoniae-induced inflammation in macrophages through NF-κB activation Life Sci 2019 221 13 19 10.1016/j.lfs.2019.02.004 30738045 28. Blötz C Singh N Dumke R Stülke J Characterization of an immunoglobulin binding protein (IbpM) from Mycoplasma pneumoniae Front Microbiol 2020 11 685 10.3389/fmicb.2020.00685 32373096 29. Hakim MS Annisa L Jariah ROA Vink C The mechanisms underlying antigenic variation and maintenance of genomic integrity in Mycoplasma pneumoniae and Mycoplasma genitalium Arch Microbiol 2021 203 2 413 429 10.1007/s00203-020-02041-4 32970220 30. Kenri T Kawakita Y Kudo H Matsumoto U Mori S Furukawa Y Tahara YO Shibayama K Hayashi Y Arai M Miyata M Production and characterization of recombinant P1 adhesin essential for adhesion, gliding, and antigenic variation in the human pathogenic bacterium Mycoplasma pneumoniae Biochem Biophys Res Commun 2019 508 4 1050 1055 10.1016/j.bbrc.2018.11.132 30551878 31. Lee D Lal NK Lin ZD Ma S Liu J Castro B Toruño T Dinesh-Kumar SP Coaker G Regulation of reactive oxygen species during plant immunity through phosphorylation and ubiquitination of RBOHD Nat Commun 2020 11 1 1838 10.1038/s41467-020-15601-5 32296066 32. Chen LS Li C You XX Lin YW Wu YM The mpn 668 gene of Mycoplasma pneumoniae encodes a novel organic hydroperoxide resistance protein Intern J Med Microbiol 2018 308 7 776 783 10.1016/j.ijmm.2018.04.006 33. Yamamoto T Kida Y Sakamoto Y Kuwano K Mpn491, a secreted nuclease of Mycoplasma pneumoniae, plays a critical role in evading killing by neutrophil extracellular traps Cell Microbiol 2016 19 3 e12666 10.1111/cmi.12666 34. Yu Y Wang J Han R Wang L Xiong Q Mycoplasma hyopneumoniae evades complement activation by binding to factor H via elongation factor thermo unstable (EF-Tu) Virulence 2020 11 1 1059 1074 10.1080/21505594.2020.1806664 32815770 35. Chen X Liu F Zheng B Kang X Wang X Mou W Zhang H Jiao A Zhao S Gui J Exhausted and apoptotic BALF T cells in proinflammatory airway milieu at acute phase of severe mycoplasma pneumoniae pneumonia in children Front Immunol 2021 12 760488 10.3389/fimmu.2021.760488 35111152 36. Saraya T Mycoplasma pneumoniae infection: Basics J Gen Fam Med 2017 18 3 118 125 10.1002/jgf2.15 29264006 37. Mirijello A La Marca A D'Errico MM Curci S Vendemiale G Grandone E De Cosmo S Venous thromboembolism during mycoplasma pneumoniae infection: case report and review of the literature Eur Rev Med Pharmacol Sci 2020 24 19 10061 10068 33090413 38. Choi SY Choi YJ Choi JH Choi KD Isolated optic neuritis associated with Mycoplasma pneumoniae infection: report of two cases and literature review Neurol Sci 2017 38 7 1 5 10.1007/s10072-017-2922-9 39. Song WJ Kang B Pyung LH Joongbum C Jeong LH Ho CY Pediatric Mycoplasma pneumoniae infection presenting with acute cholestatic hepatitis and other extrapulmonary manifestations in the absence of pneumonia Pediatr Gastroenterol Hepatol Nutr 2017 20 2 124 129 10.5223/pghn.2017.20.2.124 28730137 40. Wang Z Sun J Liu Y Wang Y Impact of atopy on the severity and extrapulmonary manifestations of childhood Mycoplasma pneumoniae pneumonia J Clin Lab Anal 2019 33 5 e22887 10.1002/jcla.22887 30924557 41. Naghib M Hatam-Jahromi M Niktab M Ahmadi R Kariminik A Mycoplasma pneumoniae and toll-like receptors: a mutual avenue Allergol Immunopathol 2018 46 5 508 513 10.1016/j.aller.2017.09.021 42. Meyer Sauteur PM de Bruijn A Graça C Tio-Gillen AP Estevão SC Hoogenboezem T Hendriks RW Berger C Jacobs BC van Rossum AMC Huizinga R Unger WWJ Antibodies to protein but not glycolipid structures are important for host Defense against Mycoplasma pneumoniae Infect Immun 2019 10.1128/IAI.00663-18 30396892 43. Poddighe D Comi EV Brambilla I Licari A Marseglia GL Increased total serum immunoglobulin E in children developing Mycoplasma pneumoniae-related extra-pulmonary diseases Iran J Allergy Asthma Immunol 2018 17 5 490 496 30518191 44. Poddighe D Marseglia GL Is there any relationship between extra-pulmonary manifestations of Mycoplasma pneumoniae infection and atopy/respiratory allergy in children? Pediatric Rep 2016 8 1 6395 6395 10.4081/pr.2016.6395 45. Ye Q Mao JH Shu Q Shang SQ Mycoplasma pneumoniae induces allergy by producing P1-specific immunoglobulin E Ann Allergy Asthma Immunol 2018 121 1 90 97 10.1016/j.anai.2018.03.014 29555351 46. Poddighe D Marseglia GL Is there any relationship between extra-pulmonary Manifestations of Mycoplasma pneumoniae infection and atopy/respiratory allergy in children? Pediatr Rep 2016 8 1 6395 10.4081/pr.2016.6395 27114818 47. Liu J He R Wu R Wang B Xu H Zhang Y Li H Zhao S Mycoplasma pneumoniae pneumonia associated thrombosis at Beijing Children's hospital BMC Infect Dis 2020 20 1 51 10.1186/s12879-020-4774-9 31948402 48. Liu J Li Y Thrombosis associated with mycoplasma pneumoniae infection (Review) Exp Ther Med 2021 22 3 967 10.3892/etm.2021.10399 34335909 49. Pachet A Dumestre-Perard C Moine M Marlu R Rubio A Bost-Bru C Splenic infarction associated with transient anti-prothrombin antibodies is a rare manifestation of acute Mycoplasma pneumoniae infection Arch Pediatr 2019 26 8 483 486 10.1016/j.arcped.2019.10.001 31740285 50. Dietz SM van Stijn D Burgner D Levin M Kuipers IM Hutten BA Kuijpers TW Dissecting Kawasaki disease: a state-of-the-art review Eur J Pediatr 2017 176 8 995 1009 10.1007/s00431-017-2937-5 28656474	36459213	PMC9716528	NO-CC CODE	2022-12-03 23:21:04	no	Curr Microbiol. 2023 Dec 2; 80(1):14	utf-8	Curr Microbiol	2,022	10.1007/s00284-022-03103-0	oa_other
==== Front Eur J Pediatr Eur J Pediatr European Journal of Pediatrics 0340-6199 1432-1076 Springer Berlin Heidelberg Berlin/Heidelberg 36459226 4687 10.1007/s00431-022-04687-2 Study Protocol Pediatric emergency care admissions for somatic symptom disorders during the COVID-19 pandemic Turco R. [email protected] [email protected] 1 Russo M. 1 Lenta S. 1 Apicella A. 1 Gagliardo T. 1 Savoia F. 2 Corona A. M. 1 De Fazio F. 1 Bernardo P. 3 Tipo V. 1 1 grid.415247.1 0000 0004 1756 8081 Pediatric Emergency Unit, Santobono-Pausilipon Children’s Hospital, Naples, Italy 2 grid.415247.1 0000 0004 1756 8081 Childhood Cancer Registry of Campania, Santobono-Pausilipon Children’s Hospital, 80129 Naples, Italy 3 grid.415247.1 0000 0004 1756 8081 Department of Neurosciences, Pediatric Psychiatry and Neurology, Santobono-Pausilipon Children’s Hospital, 80120 Naples, Italy Communicated by Peter de Winter 2 12 2022 18 17 5 2022 25 10 2022 31 10 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. During the COVID-19 pandemic, children and adolescents with psychiatric disorders experienced an exacerbation of their symptoms with more access to the emergency department (ED). However, little is known about the experience of somatic symptom disorders (SSDs) during the COVID-19 pandemic in children. Therefore, we aimed to compare the rates of pediatric ED admissions for SSDs before and during the COVID-19 pandemic and to understand whether the relative risk of ED admissions for SSDs changed between the two periods. We retrospectively enrolled all children between 4 and 14 years admitted for SSDs in the pediatric ED of Santobono-Pausilipon Hospital, Naples, Italy, from March 11th, 2020, to March 11th, 2021 (pandemic period), and in the same time period of the previous year (pre-pandemic period). We identified 205/95,743 (0,21%) children with SSDs presenting in ED in the pre-pandemic year and 160/40,165 (0,39%) in the pandemic year (p < 0.05). Considering the accesses for age, we observed a relative decrease of the accesses for SSDs over 12 years old (IRR 0,59; CI 0,39–0,88), while we found no differences under 12 years old (IRR 0,87; CI 0,68–1,10). Conclusion: In this study, we found that despite the massive decrease in pediatric admissions due to the COVID-19 pandemic, somatic symptom disorders’ admissions to the pediatric ED increased, suggesting an impact of the pandemic also on pediatric psychiatric disorders. What is Known: • During the COVID -19 pandemic, children and adolescents with a psychiatric disorder experienced exacerbation of their symptoms with more accesses in Emergency Department. What is New: • We found that despite the massive decrease of the pediatric admissions due to the COVID-19 pandemic, somatic symptom disorders admissions in healthy children to the pediatric Emergency Department increased ,suggesting an impact of the pandemic also on the pediatric psychiatric disorders Keywords Somatic symptom disorders COVID-19 Emergency department Children ==== Body pmcIntroduction Coronavirus-19 (COVID-19) infection, caused by SARS-CoV-2, was declared a pandemic by the World Health Organization on March 11th, 2020 [1]. COVID-19 radically changed medical practices and the provision of health services in all specialties [2, 3]. Since the beginning of the pandemic, a clear correlation between age and disease severity and mortality has been identified [4, 5]. A systematic review of 1065 children infected with SARS-CoV-2 demonstrated uniformly mild phenotypes of disease, presenting mostly with self-limiting respiratory symptoms [6]. Nevertheless, the impact of COVID-19 on children and their families has been high not only because of the disease itself but also because of all the containment measures to reduce virus spread. Children and adolescents experienced a drastic routine disruption due to the loose of social interaction, the schools’ closure, and the interruption of recreational activities. The impact of these measures on the routine children’s lifestyle is unexpected [7]. Wenter et al. showed that the COVID-19 pandemic spread among children’s negative effects on their mental health [8]. In particular, a recent survey of 23 different hospital emergency departments in 10 countries demonstrated an increased proportion of children and adolescents with self-harm presentation [9]. Satcher and Kenned observed that anxiety and increased fear are the most common symptoms reported among people living under lockdown [10]. Emergency departments (ED) are often the first point of care for children experiencing mental health emergencies, particularly when other services are inaccessible or unavailable [11]. During the pandemic, children and adolescents with psychiatric disorders experienced an exacerbation of their symptoms with more access to ED [12]. However, little is known about the experience of somatic symptom disorders during the COVID-19 pandemic in children. Therefore, we aimed to investigate the impact of the COVID-19 pandemic on somatic symptom disorders in children and to understand whether the relative risk of ED admissions for somatic symptom disorders changed between pandemic and pre-pandemic periods. Moreover, we aimed to understand if children with somatic symptom disorders had a similar ED presentation pattern between the two periods. Methods This retrospective study was conducted at the Santobono-Pausilipon Children’s Hospital of Naples, Italy. The inclusion criteria were as follows.aged from 4 to 14 years old, admission to the pediatric ED; and diagnosis of “somatic symptom and related disorders” according to the DSM V [13], including somatic symptom disorders, illness anxiety disorder, conversion disorder (functional neurological symptom disorder), psychological factors affecting other medical conditions, factitious disorder, other specified somatic symptom and related disorder, and unspecified somatic symptom and related disorder. The onset symptoms were classified into three large groups:pain symptoms (head, abdomen, back, joints, extremities, chest, rectum, during menstruation, during sexual intercourse, or during urination); gastrointestinal symptoms (nausea, bloating, vomiting other than during pregnancy, diarrhea, or intolerance of several different foods); pseudo-neurological symptoms (conversion symptoms such as impaired coordination or balance, paralysis, or localized weakness, difficulty swallowing or lump in throat, aphonic, urinary retention, hallucinations, loss of touch or pain sensation, double vision, blindness, deafness, seizures; dissociative symptoms such as amnesia; or loss of consciousness other than fainting). Data included demographic characteristics, gender, age, and previous ED admissions. Moreover, data concerning the psychiatric evaluation were collected as well as the need for psychiatric follow-up. The data were anonymously recorded and analyzed. We excluded children with known psychiatric disorders or with chronic diseases as well as children with acute diseases that required hospitalization or those with multiple accesses in a single month. The study group was divided into the following two sub-groups:pre-pandemic group: children admitted to the Pediatric ED from March 11th, 2019, to March 11th, 2020; and pandemic group: children admitted to the pediatric ED from March 11th, 2020, to March 11th, 2021. The overall pediatric emergency department admissions were calculated in both pandemic and pre-pandemic periods. Statistical analysis ED visits are expressed in absolute frequencies and percentage change between pandemic and pre-pandemic periods. Incidence rate ratios (IRR) were calculated using weekly ED access counts, modeled with a Poisson regression model. A COVID-19-related covariate assumed the value of 0 in the pre-pandemic year and the value of 1 in the same period of the pandemic year. Poisson regression model was also carried out using as offset all ED accesses per week in order to assess the change in pediatric somatic disorders adjusting by total ED accesses. All p-values were from 2-sided tests, and results were deemed statistically significant at p < 0.05. Data analysis was conducted using STATA version 13.1 (STATA Corp). Results In this retrospective study, we identified a total of 95,744 admissions to our pediatric ED in the pre-pandemic year with respect to a total of 40,165 admissions in the pandemic year (Fig. 1). Concerning the admission for somatic symptom disorders, we collected 205 children in the pre-pandemic group (0.21%) and 160 (0.39%) in the pandemic group. In particular, evaluating the admission pattern during the year, we observed a relative decrease in the access for somatic symptom disorders in the first quarter of the pandemic year (IRR 0.78; CI 0,64–0,96) with respect to the pre-pandemic year, while there were no differences for the rest of the study periods (Table 1; Fig. 2). Nevertheless, when we analyzed the ratio between the number of children admitted to the ED for somatic symptom disorders of both groups related to the total number of admission in ED in pandemic and pre-pandemic years, we found an incident rate ratio (IRR) of 1.86 (CI 1.51–2.29) in the pandemic group with respect to the pre-pandemic group (p < 0.05). The median age was 10.2 (4.2–13.11) years in the pre-pandemic group and 9.9 (4.10–13.9) years in the pandemic group (p = 0.29). In the pre-pandemic group, 54% of children were female with respect to the 45% in the pandemic group (p = 0.66). Considering the accesses for age, we observed a relative decrease in the admissions for somatic symptom disorders over 12 years old between the pandemic and pre-pandemic groups (IRR 0.59, CI 0.39–0.88), while no differences were found for the other age groups (IRR 0,87; CI 0.68–1,10). Related to gender, females had a lower number of accesses for somatic symptom disorders to the ED in the pandemic group with respect to the pre-pandemic group (IRR 0.66, CI 0.49–0.88; p < 0.05).Fig. 1 Emergency department admissions in pre- and pandemic year Table 1 Incidence rates for emergency department admissions for somatic symptoms in pre-pandemic (11th March 2019–10th March 2020) and pandemic period (11th March 2020–11th March 2021) 1 Mar–30 Jun 1 Jul–30 Oct 1 Nov–28 Feb Total Overall Pre-pandemic 63 84 58 205 Pandemic 42 72 46 160 IRR (95% CI) 0,67 (0,45–0,99) 0,86 (063–1,18) 0,79 (0,54–1,17) 0,78 (0,64–0,96) Age < 12 years Pre-pandemic 41 60 41 142 Pandemic 34 50 39 123 IRR (95% CI) 0,83 (0,53–1,31) 0,83 (0,57–1,21) 0,95 (0,61–1,47) 0,87 (0,68–1,10) Age 12–14 years Pre-pandemic 22 24 17 63 Pandemic 8 22 7 37 IRR (95% CI) 0,36 (0,16–0,82) 0,92 (0,51–1,63) 0,41 (0,17–0,99) 0,59 (0,39–0,88) Female Pre-pandemic 33 45 33 111 Pandemic 19 35 19 73 IRR (95% CI) 0,58 (0,33–1,01) 0,78 (0,50–1,21) 0,58 (0,33–1,01) 0,66 (0,49–0,88) Male Pre-pandemic 30 39 25 94 Pandemic 23 37 27 87 IRR (95% CI) 0,77 (0,44–1,31) 0,95 (0,61–1,49) 1,08 (0,63–1,86) 0,93 (0,69–1,24) Fig. 2 Emergency department admissions among children with somatic disorders in the pandemic year compared to the pre-pandemic year When we adjusted the frequencies of somatic symptom disorders for the total number of access, we found a significant difference between pandemic and pre-pandemic groups for chest pain (IRR: 4.17, p < 0.05), breathing difficulties (IRR: 3.46; p < 0.05), anxiety (IRR: 2.12, p < 0.05), general discomfort (IRR:1.60; p < 0.05), anorexia (IRR: 3.34; p < 0.05), dysphagia (IRR: 2.68; p < 0.5), and tachycardia (IRR 2,19; p < 0.05). No differences between the two studied groups were found in constipation, abdominal pain, paraesthesia, dizziness, fainting/pre-syncope, asthenia, headache, nausea, and vomiting (Table 2). When classifying the somatic symptom disorders in the 3 subtypes (pain, gastrointestinal, and pseudo-neurological symptoms) and adjusting the data for the total number of accesses at the ED, we found that gastrointestinal symptom’s presentation was more frequent in the pandemic group with respect to the pre-pandemic group (IRR 3.28) (Table 3). We observed that 14/205 (6.8%) children with somatic symptom disorders had repeated admissions compared to 13/160 (8%) in the pandemic period (p < 0.2).Table 2 Type of somatic symptoms complained during the pandemic and pre-pandemic period Pre-pandemic year Post-pandemic year IRR (IC95%) P value IRR based on the total P value based on the total Total accesses 95,743 40,165 Number of accesses for somatic symptom disorders 205 160 0,78 (0,64–0,96) < 0.05 1,86 (1,51–2,29) < 0.05 Dizziness 9 3 0,33 (0,09–1,23) 0,10 0,79 (0,22–2,94) 0,73 Abdominal pain 30 15 0,50 (0,27–0,93) < 0,05 1,19 (0,64–2,22) 0,78 Headache 11 8 0,73 (0,29–1,81) 0,50 1,73 (0,70–4,31) 0,24 Nausea 2 2 1,00 (0,14–7,12) 1 2,38 (0,34–16,92) 0,39 General discomfort 67 45 0,67 (0,46–0,98) < 0,05 1,60 (1,10–2,34) < 0,05 Breathing difficulties 31 45 1,46 (0,92–2,30) 0,11 3,46 (2,19–5,47) < 0,05 Chest pain 8 14 1,75 (0,74–4,18) 0,20 4,17 (1,75–9,94) < 0,05 Table 3 Differences of presentation of somatic symptoms divided into three subgroups according to DSM V into pre-pandemic and post-pandemic years, respectively Pre-pandemic year Pandemic year Variation (%) Results adjusted for total pediatric ED accesses Pain symptoms group 49 37 − 24 IRR 1.80 (1.17–2.76), P < 0.01 Gastrointestinal symptoms group 16 22 38 IRR 3.28 (1.72–6.24), P < 0.01 Pseudo-neurological group 170 130 − 24 IRR 1.82 (1.45–2.29), P < 0.01 A total of 15.6% of patients required neuropsychiatric consultation in ED in the pandemic group with respect to 9.7% in the pre-pandemic group (IRR: 2.98, p < 0.05). Discussion In this retrospective study, we demonstrated a relative increase in the admissions rate for somatic symptom disorders in the pandemic year with respect to the previous year. We observed these findings despite the massive decrease in pediatric admissions due to the COVID-19 pandemic in agreement with the data reported by other Italian centers [14], especially in the first 8 weeks of the COVID-19-induced social lockdown [15]. As well-renowned COVID-19 pandemic forced a reorganization of the ED, imposing a filter to accesses, which has reduced all admissions to the emergency rooms other than for COVID-19 infection. Moreover, this decreasing in ED admissions could be explained by the fear of possible infection by the COVID-19 virus, which was new at the time, which led families to avoid the clinic as much as possible at the beginning of the pandemic. Hartnett et al. have already reported that the mean weekly number of ED visits for children < 14 years old reduced by approximately 70% during March 29–April 25, 2020, with respect to the corresponding period in 2019, for asthma, otitis, sprain, and strain-related injuries, while it increased of 69% for psychosocial factors [16]. Moreover, Leeb et al. found that compared with 2019, the proportion of mental-health-related visits in ED for children aged 5–11 years old and 12–17 years old increased by approximately 24% and 31%, respectively [17]. In particular, analyzing the admission pattern for somatic symptom disorders during the pandemic and pre-pandemic year, we found that the admission rates were significantly lower in the first quarter of the pandemic period with respect to the corresponding period of the pre-pandemic year and similar in the rest of the year. Our results are consistent with other Italian data [15, 18, 19], where during the first two months of COVID-19 lockdown, no significant changes were found in hospitalization rate or in the prevalence distribution of the primary reason for the psychiatric ED visit [15], while a significant increase was recorded in the following months [18, 19]. This trend was partly due to a return to social life after months of isolation and, on the other hand, to the onset of neuropsychological issues that led parents conducting their children to pediatric ED. Although children demonstrated to have milder clinical manifestations when infected by SARS-COV2 with respect to adults [20], they certainly experienced considerable discomfort. As a matter of fact, during the pandemic year, a spread of psychological problems in children with an increased prevalence of symptoms like anxiety, fearness, and breathing difficulties have been reported [21–25]. Singh et al., in a recent review on the impact of COVID-19 on the mental health of children, showed high levels of stress, insomnia, poor appetite, and inattentiveness [26]. Xie et al., moreover, found that 22.6% and 18.9% of children and pre-adolescents in Hubei reported symptoms of depression and anxiety independently of demographic characteristics [27]. Children presented to our ED complained, in particular, chest pain, breathing difficulties, anxiety, insomnia, fearness, anorexia, dysphagia, and tachycardia with a significant difference with respect to the pre-pandemic year, while we did not find any differences for other somatic disorders. It is well known that somatic symptom disorders are very common in the general pediatric population with major presentation symptoms such as abdominal pain, headache, and seizures [28]. Considering the somatic symptom disorders’ subtypes, we found a major presentation of gastrointestinal symptoms during the COVID-19 pandemic with respect to the other pattern presentations. This finding is in contrast with Solmi et al., who reported a higher prevalence of pain and pseudo-neurological subtypes. Further studies, however, are needed to understand if the general changing in routine, habits, and lifestyle due to the COVID-19 pandemic led to an overspread of the gastrointestinal somatic symptom disorders’ pattern. The reason for this significant increase is still not completely understood. At the beginning of the pandemic, Sama et al. [29] conducted a systematic review of 3166 published articles on previous quarantine for SARS (11 studies), Ebola (five), the 2009 and 2010 H1N1 influenza pandemic (three), Middle East respiratory syndrome (two), and equine influenza (one) showing that lockdown may have negative psychological effect causing anxiety, anger, sleep disorder, depression, and post-traumatic stress disorder (PTSD) in the general population. Previously, Loades et al. [30] also described the negative effects of quarantine during the outbreak of H1N1 in 2009, showing a four times higher PSTD score among the children who lived under lockdown than those who did not. However, the level of psychological impacts of lockdown is related to various factors such as lifestyle, society, and culture [31, 32]. Certainly, children had to change their daily routines, suffering a lack of social interactions with the closure of schools and kindergartens. They also had to challenge new fears, potentially with a lot of unanswered questions like the possibility of losing or being separated from their parents. Changing behavior during the pandemic as the reduction in outdoor physical activity, diets’ modifications, and increasing in time spending in front of screens may have contributed to the spreading of somatic disorders in children. Moreover, the number of children exposed to direct or indirect domestic violence and abuses seems to be increased [33–35]. In most studies, girls reported higher levels of worry, concern, and fear regarding COVID-19 [27, 35–37] compared with boys. Two different studies observed that girls exhibited equivalent changes in depressive symptoms related to the COVID-19 pandemic compared with boys [19, 39]. Differently, in our population, we did not find any difference with regard to gender, but evaluating the rate of somatic disorders admissions in ED for gender, we found that males had more admissions with respect to females under 12 years of age. This data was in line with Uccella et al. who demonstrated, in a recent survey on the behavioral changes related to COVID-19, that adolescents seemed less affected by behavioral problems than younger children [19]. The elevated number of admissions for somatic symptom disorders led to an increase in neuropsychiatric consultations with the need to improve the mental health assistance in our ED. We acknowledge that our results have to be interpreted with caution, taking into account our study limitations. First, our data are restricted to ED visits and, as such, may not be generalizable to the overall population outside and do not fully capture the incidence of somatic symptom disorders among children who did not present to a tertiary pediatric ED. Second, being a retrospective study, it is possible that some somatic disorders may have been lost. Third, in our study, we did not investigate the impact of COVID-19 infections on the children included in the analysis, which may have specifically contributed to some of the reported symptoms. In conclusion, in our population, we observed an increased rate of somatic symptom disorders admissions in ED during the pandemic with respect to the corresponding previous year, highlighting the great impact of the COVID-19 infection also on pediatric behavior. The consequences of these phenomena on the pediatric population are unknown and undetectable, and further studies will be needed to monitor them and the correlation to the onset of new psychiatric disorders in the involved children. Author contribution Rossella Turco, Marina Russo, and Vincenzo Tipo contributed to the study’s conception and design. Material preparation, data collection, and analysis were performed by Alberto M Corona, Selvaggia Lenta, Andrea Apicella, Thaililjia Gagliardo, and Floriana De Fazio. The first draft of the manuscript was written by Rossella Turco, Marina Russo, Pia Bernardo, and Fabio Savoia, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Declarations Ethics approval and consent to participate This is an observational study. The Santobono-Pausilipon Research Ethics Committee has confirmed that no ethical approval is required. Consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. R. Turco and M. Russo contributed equally to this work. ==== Refs References 1. World Health Organization (2020)WHO Director-General’s opening remarks at the media briefing on COVID-19 2020 [cited 2020 Jul 16]. Available from: https://www.who.int/dg/speeches/detail/who-director-general-s-opening remarks-at-the-media-briefing-on-covid-19--11-march-2020. 2. Lee CCM, Thampi S, Lewin B, Lim TJD, Rippin B, Wong WH, Agrawal RV (2020) Battling COVID-19: critical care and peri-operative healthcare resource management strategies in a tertiary academic medical centre in Singapore. Anaesthesia 75:861–871. 10.1111/anae.15074 3. Wu Z, McGoogan JM (2020) Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention. JAMA 7;323(13):1239–1242. 10.1001/jama.2020.2648 4. Pathak EB Salemi JL Sobers N Menard J Hambleton IR COVID-19 in children in the United States: intensive care admissions, estimated total infected, and projected numbers of severe pediatric cases in 2020 J Public Health Manag Pract 2020 26 4 325 333 10.1097/PHH.0000000000001190 32282440 5. Castagnoli R Votto M Licari A Brambilla I Bruno R Perlini S Rovida F Baldanti F Marseglia GL Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in children and adolescents: a systematic review JAMA Pediatr 2020 20201 174 882 889 10.1001/jamapediatrics.2020.1467 6. Parmet WE, Sinha MS (2020) Covid-19 - The law and limits of quarantine. N Engl J Med 9;382(15):e28 10.1056/NEJMp2004211. 7. Wenter A Schickl M Sevecke K Juen B Exenberger S Children’s Mental health during the first two years of the COVID-19 pandemic: burden, risk factors and posttraumatic growth – a mixed-methods parents’ perspective Front Psychol 2022 2 13 901205 10.3389/fpsyg.2022.901205 8. Ougrin D Wong BH Vaezinejad M Pandemic-related emergency psychiatric presentations for self-harm of children and adolescents in 10 countries (PREP-kids): a retrospective international cohort study Eur Child Adolesc Psychiatry 2022 31 1 13 10.1007/s00787-021-01741-6 33677628 9. Satcher D, Kenned PJ (2020) Failure to address coronavirus mental health issues will prolong impact. The Hill. hlk 10. Dolan MA Fein JA Committee on Pediatric Emergency Medicine. Pediatric and adolescent mental health emergencies in the emergency medical services system Pediatrics 2011 127 e1356 e1366 10.1542/peds.2011-0522 21518712 11. Palacio-Ortiz JD, Londoño-Herrera JP, Nanclares-Márquez A, Robledo-Rengifo P, Quintero-Cadavid CP (2020) Psychiatric disorders in children and adolescents during the COVID-19 pandemic. Rev Colomb Psiquiatr (Engl Ed)49:279–288. 10.1016/j.rcp.2020.05.006 12. American Psychiatric Association (2013) Diagnostic and statistical manual of mental disorders (5th ed.). 10.1176/appi.books.9780890425596 13. Raffaldi I, Castagno E, Fumi I et al (2021) ICOPER Study Group. Pediatric admissions to emergency departments of North-Western Italy during COVID-19 pandemic: a retrospective observational study. Lancet Reg Health Eur5:10008. 10.1016/j.lanepe.2021.1000811 14. Davico C, Marcotulli D, Lux C et al (2021) Impact of the COVID-19 pandemic on child and adolescent psychiatric emergencies. J Clin Psychiatry 6; 82:20m13467. 10.4088/JCP.20m13467 15. Hartnett KP, Kite-Powell A, DeVies J, Coletta MA, Boehmer TK, Adjemian J, Gundlapalli AV (2020) National Syndromic Surveillance Program Community of Practice. Impact of the COVID-19 pandemic on emergency department visits - United States, January 1, 2019-May 30, 2020. MMWR Morb Mortal Wkly Rep. 2020 Jun 12;69(23):699–704. 10.15585/mmwr.mm6923e1 16. Vicari S Pontillo M Developmental psychopathology in the COVID-19 period. COVID-19 pandemic impact on children and adolescents’ mental health Psychiatr Danub 2021 33 33 35 34862887 17. Uccella S De Grandis E De Carli F Impact of the COVID-19 outbreak on the behavior of families in Italy: a focus on children and adolescents Front Public Health 2021 5 9 608358 10.3389/fpubh.2021.608358 18. RavalliS, Musumeci G (2020) Coronavirus outbreak in Italy: physiological benefits of home-based exercise during J Funct Morphol Kinesiol 7;5:31. 10.3390/jfmk5020031 19. Samji H Wu J Ladak A Vossen C Stewart E Dove N Long D Snell G Review: mental health impacts of the COVID-19 pandemic on children and youth – a systematic review Child Adolesc Ment Health 2022 27 173 189 10.1111/camh.12501 34455683 20. Breaux R Dvorsky MR Marsh NP Green CD Cash AR Shroff DM Buchen N Langberg JM Becker SP Prospective impact of COVID-19 on mental health functioning in adolescents with and without ADHD: protective role of emotion regulation abilities J Child Psychol Psychiatry 2021 62 1132 1139 10.1111/jcpp.13382 33543486 21. Zhou J Yuan X Qi H Liu R Li Y Huang H Chen X Wang G Prevalence of depression and its correlative factors among female adolescents in China during the coronavirus disease 2019 outbreak Global Health 2020 28 16 69 10.1186/s12992-020-00601-3 22. Zhou S-J Wang L-L Yang R Yang X-J Zhang L-G Guo Z-C Chen J-X Sleep problems among Chinese adolescents and young adults during the coronavirus-2019 pandemic Sleep Med 2020 74 39 47 10.1016/j.sleep.2020.06.001 32836185 23. Zhou S-J Zhang L-G Wang L-L Guo Z-C Wang J-Q Chen J-C Chen J-X Prevalence and sociodemographic correlates of psychological health problems in Chinese adolescents during the outbreak of COVID-19 Eur Child Adolesc Psychiatry 2020 29 749 758 10.1007/s00787-020-01541-4 32363492 24. Singh S, Roy D, Sinha K, Parveen S, Sharma G, Joshi G (2020) Impact of COVID-19 and lockdown on mental health of children and adolescents: a narrative review with recommendations. Psychiatry Res 293:113429. 10.1016/j.psychres.2020.113429 25. Xie X Xue Q Zhou Y Zhu K Liu Q Zhang J Mental health status among children in home confinement during the coronavirus disease 2019 outbreak in Hubei Province, China JAMA Pediatr 2020 174 898 900 10.1001/jamapediatrics.2020.1619 32329784 26. Solmi M Dragioti E Croatto G Risk and protective factors for personality disorders: an umbrella review of published meta-analyses of case-control and cohort studies Front Psychiatry 2021 12 679379 10.3389/fpsyt.2021.679379 34552513 27. Brooks SK, Webster RK, Smith LE, Woodland L, Wessely S, Greenberg N, Rubin GJ (2020) The psychological impact of quarantine and how to reduce it: rapid review of the evidence. Lancet 14; 395 :912–920. 10.1016/S0140-6736(20)30460-8 28. Sprang G Silman M Posttraumatic stress disorder in parents and youth after health-related disasters Disaster Med Public Health Prep 2013 7 1 105 110 10.1017/dmp.2013.22 24618142 29. Sama BK Kaur P Thind PS Verma MK Kaur M Singh DD Implications of COVID-19-induced nationwide lockdown on children’s behaviour in Punjab India Child Care Health Dev 2021 47 1 128 135 10.1111/cch.12816 33047842 30. Loades ME Chatburn E Higson-Sweeney N Higson-Sweeney N Reynolds S Shafran R Rapid systematic review: the impact of social isolation and loneliness on the mental health of children and adolescents in the context of COVID-19 J Am Acad Child Adolesc Psychiatry 2020 59 1218 39.e3 10.1016/j.jaac.2020.05.009 32504808 31. Fore HH A wake-up call: COVID-19 and its impact on children’s health and wellbeing Lancet Glob Health 2020 8 e861 e862 10.1016/S2214-109X(20)30238-2 32405458 32. Sacco MA Caputo F Ricci P Sicilia F De Aloe L Bonetta CF The impact of the COVID-19 pandemic on domestic violence: the dark side of home isolation during quarantine Med Leg J 2020 88 71 73 10.1177/0025817220930553 32501175 33. Buzzi C, Tucci M, Ciprandi R, Brambilla I, Caimmi S, Ciprandi G, Marseglia GL (2020) The psycho-social effects of COVID-19 on Italian adolescents’ attitudes and behaviors. Ital J Pediatr 24;46(1):69. 10.1186/s13052-020-00833-4 34. Magson NR Freeman JYA Rapee RM Richardson CE Oar EL Fardouly J Risk and protective factors for prospective changes in adolescent mental health during the COVID-19 pandemic J Youth Adolesc 2021 50 44 57 10.1007/s10964-020-01332-9 33108542 35. Matovu JKB Kabwama SN Ssekamatte T Ssenkusu J Wanyenze RK COVID-19 Awareness, adoption of COVID-19 preventive measures, and effects of COVID-19 lockdown among adolescent boys and young men in Kampala Uganda J Community Health 2021 46 4 842 853 10.1007/s10900-021-00961-w 33481156 36. Xiang M, Zhang Z, Kuwahara K (2020) Impact of COVID-19 pandemic on children and adolescents’ lifestyle behavior larger than expected. Prog Cardiovasc Dis 63 :531–532. 10.1016/j.pcad.2020.04.013 37. Ren H He X Bian X Shang X Liu J The protective roles of exercise and maintenance of daily living routines for Chinese adolescents during the COVID-19 quarantine period J Adolesc Health 2021 68 35 42 10.1016/j.jadohealth.2020.09.026 33121902	36459226	PMC9716529	NO-CC CODE	2022-12-03 23:21:04	no	Eur J Pediatr. 2022 Dec 2;:1-8	utf-8	Eur J Pediatr	2,022	10.1007/s00431-022-04687-2	oa_other
==== Front Soft comput Soft comput Soft Computing 1432-7643 1433-7479 Springer Berlin Heidelberg Berlin/Heidelberg 7682 10.1007/s00500-022-07682-9 Application of Soft Computing Ziwi: indoor and outdoor planning network—framework to collection, modeling and network structure based on computational optimization and measurements http://orcid.org/0000-0002-1018-9033 Rocha Lidia [email protected] 1 Ferreira Sidnir [email protected] 2 Vivaldini Kelen C. Teixeira [email protected] 1 Araújo Jasmine [email protected] 2 Batalha Iury [email protected] 2 1 grid.411247.5 0000 0001 2163 588X Department of Computing, Federal University of São Carlos, São Carlos, Brazil 2 grid.271300.7 0000 0001 2171 5249 Electrical Engineering Pós-Graduate Department, Federal University of Pará, Belém, Brazil 2 12 2022 121 18 11 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Society is increasingly connected, utilizing more data that demands greater capacity and better channel quality. Furthermore, wireless networks are being inserted into the population's daily lives. Therefore, solutions capable of transferring a high volume of data are increasingly needed. In this context, we present a framework that aims to network planning through data collection, modeling, and routers optimization in the environment. Ziwi framework can simulate wireless networks in indoor and outdoor environments with the main classical propagation models, obtain and calculate metrics and performance parameters. It is possible to measure data by cell phone and send it to the website quickly. Furthermore, it can model the data and compare with different propagation models. Also, optimize them using a genetic algorithm or permutation, choosing whether or not to consider sockets to turn on the routers and how many routers are needed to place in the environment. In addition, have a virtual reality environment aiming at greater interactivity with the data. We analyzed framework results comparing with Close-In propagation model, free space model, and statically using the root mean square error metric. Measurements were made in a real environment using the Ziwi mobile application, inserting data captured on Ziwi website to validate the framework. Keywords Wireless network Mobile networks Simulation Propagation Software ==== Body pmcIntroduction Over time the population is more connected, through the internet, with objects, and with each other (Skouby and Lynggaard 2014). In the current scenario, due to COVID-19, video conferences and video streams have increased exponentially since remote work, classes, and meetings are being held virtually, demanding more from internet services. NOKIA’s traffic information in the COVID-19 pandemic period showed an increase of 30–50% concerning the peak of “normal” traffic over the weekend (Labovitz 2020). Video conference applications are the main activity on weekdays and streaming on the weekend. Conference applications like Zoom have seen more than 700% growth in the US since February 1st, 2020. Furthermore, internet services can assist in healthcare by connecting hospitals, ambulances, and homes reliably to make the health service more resource-efficient and more effective in managing Pandemic and normal operations (Saeed et al. 2020). Moreover, services such as remote monitoring and patient diagnosis enabled for artificial intelligence allow anyone with a medical condition to report a central health system without having to be physically present (Shakir 2020). Besides that, applications based on artificial intelligence and wireless chipsets/sensors are intended to diagnose whether a person is infected or not (Maghdid et al. 2020). Due to the rise in the number of online services, such as streaming and video calls, the necessity of mobile and Wireless Local Area Networks (WLANs) with a good connection has increased (Sharma and Wang 2019). Therefore, there is a growing demand for coverage at any time and place without interruption, leading to a need for mobile networks and Wi-Fi to deal with greater capacity and offer the quality required by customers. Operators are implementing new external cells, and routers are experiencing increasing speed (Rappaport et al. 2013). Due to this fact, the demand for an adequate tool for network planning arose to assist in predicting and optimizing signal coverage (Rodrigues et al. 2011). Rodrigues et al. (2011) shows the growth of services that need the internet and require increasingly complex data. Since 2011, devices such as smartphones, urban and industrial sensing devices have been used by 9 billion people (Huang et al. 2020). The significant increase in data traffic puts huge pressure on network management. Hence, it is noticed the importance of carrying out good network planning, aiming at future network generations. In this context, this paper aims to develop an easy and practical framework named Ziwi for engineers and academics in the field. Providing an intuitive interface for planning mobile and Wi-Fi networks. Ziwi has a graphical web interface that allows communication with a mobile application capable of measuring electromagnetic signals in indoor and outdoor environments. In addition, allows the use of virtual reality software, aiming to increase the interactivity of the user with the signal of the environment and assist in its analysis. The developed Ziwi framework helps manage the high data consumption required by 5th Generation of Mobile Communications (5G), smart cities, and by the increase in data traffic due to the COVID-19 Pandemic (Ijaz et al. 2016). In addition to increasing user productivity from an interconnected software that both brings agility for research and increasing user interactivity with the electromagnetic signal through virtual reality. Ziwi was inspired by the Ranplan (Charbonneau 2005) and IBWave (Qin et al. 2011) simulators. Ranplan performs indoor, outdoor, and hybrid networks planning and IBWave can perform indoor networks planning. Both aim to improve network coverage in 5G scenarios. However, both software are paid and focused on large projects, such as smart cities and smart hospitals that provide real-time treatment. Situations in which significant investments are available. In this context, the main difference between Ziwi and these softwares is that it is free software that can be used in different scenarios. Besides, Ziwi has a mobile application to measure data in indoor and outdoor scenarios. Furthermore, the website of the Ziwi can optimize the planning, and the framework has a virtual reality system to encourage and facilitate the learning of new students in the area. This paper is organized as follows: Sect. 2 showed related works and their differences from the current work. Section 3 shows all technologies applied. Section 4 describes the framework development. Section 5 describes the experimental environment configuration, followed by experimental results. Finally, in Sect. 6 we provide our conclusion. Related works This section provides an overview of the main related works to this research. Kar et al. (2016) developed an indoor simulator for the Motley Keenan propagation model. However, the software is limited to only Motley Keenan models and outdated internet standards such as IEEE 802.11a, IEEE 802.11b, and IEEE 802.11g. Besides, since it does not have a graphical interface, it makes the simulation process more complex. It is not possible to optimize the router position in the environment. The application differential is to provide signal strength, path loss, Signal-to-Noise Ratio (SNR), Signal-to-Noise-plus-Interference Ratio (SINR), and channel capacity metrics. These data also were inserted in Ziwi. When observing McEl-Roy et al. (2007) work, we decided to add the intensity of the electric field in the Ziwi, so the user can analyze the data and compare with the legal limits of electromagnetic field exposure. Furtado et al. (2016) created a simulator for the WLAN network very close to the IBWaves simulator, which provides simulation only for indoor environments and runs on desktop computers. Nevertheless, much data is consumed in outdoor environments through mobile networks. Ziwi has a module that allows both indoor and outdoor environments simulations for planning mobile networks, provided by an application for smartphones that also runs in web browsers. Najnudel (2004) presents the main propagation models for indoor environments, such as Close-In (CI), Motley Keenan, Floating Interception, and ITU-R P.1238-8. Maccartney et al. (2015) propose Floating Interception model, a propagation model that tends to adapt better to the measured path loss. In this model, two parameters are modeled according to the measurements: alpha (the anchor point) and beta (the slope of the curve). Lastly, Castro et al. (2010) present the main propagation models for outdoor environments: SUI, COST 231, and ECC 33. All models cited above are semi-empirical, i.e., they need measured data in the environment to carry out the simulation. However, due to the rapid development of technology, it is possible to conduct environmental measurements using applications compatible with mobile devices. Bhatt et al. (2016) present a literature review of the main applications developed with this focus. However, most of them are capable of recording only the signal strength but not generating data logs. Therefore, we developed a measurement application for the Ziwi framework to capture the mentioned metrics and save them in log files. None of the works presented above perform router optimization. However, some commercial software does this, such as “Raplan” and “IBWave”, which can plan networks for the future generation automatically, optimizing the time of the network planner. Thus, we realize that it would be necessary to implement an optimization technique to plan indoor environments to have the largest network coverage using the shortest possible time. Teixeira et al. (2017), uses neural networks to predict the signal, but this technique requires training sets, which may not be a viable solution for different situations. On the other hand, Kelly (2016) uses Particle Swarm Optimization (PSO) to find the optimal solution to the problem. Nevertheless, Genetic Algorithms achieve a more accurate solution due to genetic parameters intensifying and diversifying the solution population (Garcia-Martinez et al. 2018). Consequently, we applied genetic algorithms as an optimization technique, which also seeks the optimal solution. Despite spending a little more time than the PSO technique, it tends to have a more accurate result. Several researchers have used genetic algorithms to optimize routers’ positions. For example, Teguh et al. (2014) presented an optimization technique for routers based on energy restriction using a genetic algorithm. The optimization uses routers’ positions to minimize total communication distance, aiming to maximize sensor network lifetime. Results found that it is possible to reduce both the number of disconnected sensors and total communication distance. In (Gao et al. 2013), is proposed a multiobjective genetic algorithm to optimize network coverage for mobile cellular users. First, is assessed the status of cell coverage using measurement data reported by mobile stations. With this, is generated a system to optimize network coverage for users. The results showed that network coverage performance could be improved after optimization with the genetic algorithm. Optimization with a genetic algorithm is not limited to routers. It can also be applied to antennas. Chou et al. (2016) implemented a genetic algorithm in the antenna structure measurement system to optimize them, radiating the directional beam in the desired direction. The actual outputs of the phase shifters are fashionable, so they can be used to optimize the pattern in real-time, without the need to measure the phase changes of all phase shifters. For situations where high computational performance devices are not available, semi-empirical propagation models are more appropriate. However, this approach requires measured environment data as the basis demonstrated by the authors in Blackard et al. (1993), where measurements took place in indoor environments, using both omnidirectional and directional antennas. With that, it was possible to analyze the characteristics and sources of noise in the channels. The results are shown in peak amplitude probability distributions, pulse duration distributions, and time distributions between arrivals. Measurements must also be made in outdoor environments. Liechty (2007), performed measurements in an outdoor environment on a 2.4 GHz IEEE 802.11g Wi-Fi network to validate the Seidel-Rappaport propagation model adapted by the authors. The measurements showed that predictive planning for network coverage is possible without the need for overly complicated modeling techniques, such as ray tracing, and the use of semi-empirical techniques is valid for this. Many surveys use the G-Net Track Pro mobile application to perform measurements (Yudha et al. 2016), (Alias et al. 2016). However, this application is paid. There is a free version, but it is possible to generate only Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), and SNR values. Besides, the application only performs measurements in outdoor environments. Bhatt et al. (2016), showed some applications capable of performing measurements in environments, but most are no longer available or have an intuitive user interface. Thus, to provide these measurements, we developed an application that communicates with the framework. We present in this section the literature review, detailing the differentials of the proposed simulator. The main points were to use indoor and outdoor propagation models on a web platform. Furthermore, all models implemented in the framework are semi-empirical. Therefore, we developed a mobile application capable of performing measurements for the simulations. A Ziwi differential is that it simulates the heat map of the environment in virtual reality, facilitating its analysis and helping learning in the area of signal propagation, making it more interactive. According to Wickens (1992), virtual reality applications can increase long-term user learning retention and improve performance in a specific area. Thus, it can be used to explain electromagnetic and propagation signals in classrooms. Ziwi’s tools In this section, we demonstrate the tools' concepts and how they are used in the framework. These concepts are wireless networks, propagation models, and optimization techniques. Wireless network At the end of the last century, WLAN communication technologies became popular in the market, with the arrival of the IEEE 802.11 standard. Today, this well established technology composes data infrastructure of communication solutions for the most diverse environments due to its relatively low cost and ease implementation (Ahmavaara et al. 2003). It is possible to use several IEEE 802.11 standards in the framework, such as 802.11n, 802.11ac, 802.11ad, 802.11ax, and 802.11ay, changing frequency and bandwidth. The differences between the standards are shown in Tables 1 and 2.Table 1 Main properties for different IEEE 802.11 standards available in the simulator IEEE 802.11n IEEE 802.11ac Maximum data rate 72.2 Mbps (20 MHz, 400 ns) 96.3 Mbps (20 MHz, 400 ns) 150 Mbps (40 MHz, 400 ns) 200 Mbps (40 MHz, 400 ns) 433.3 Mbps (80 MHz, 400 ns) 866.7 Mbps (160 MHz, 400 ns) Frequency 2.4, 5 GHz 5 GHz PHY protocol OFDM OFDM Bandwidth 20, 40 MHz 20, 40, 80, 160 MHz Modulation 64 QAM 256 QAM MIMO 4 × 4 8 × 8 (UL) 4 × 4 (DL, MU-MIMO) Table 2 Main properties for different 802.11 standards, in development, present in the simulator IEEE 802.11ad IEEE 802.11ax IEEE 802.11ay Maximum data rate 7 Gbps 10.53 Gbps 20 Gbps Frequency 60 GHz 2.4, 5 GHz 60 GHz PHY protocol DMG MIMO-OFDM OFDM Bandwidth 2160 MHz * 8000 MHz Modulation OFDM MIMO-OFDM OFDM MIMO N/D * 4 People are increasingly dependent on mobility and connectivity. Through mobile devices, users have information from around the world (Backholm et al. 2013). However, these devices may not always be close to an access point, such as Wi-Fi, requiring other technologies to connect to the Internet. Solutions capable of meeting this demand are the 2nd Generation of Mobile Communications (2G), 3rd Generation of Mobile Communications (3G), 4th Generation of Mobile Communications (4G), and more recently 5th Generation of Mobile Communications (5G). Nowadays, 4G is the main technology used for outdoor environments, which operates in Brazil's 700, 1800, and 2500 MHz bands. The latest advancement in LTE, called LTE-A PRO, allows carrier aggregation with up to 32 carriers. Each carrier with 100 MHz bandwidth offering a maximum aggregate bandwidth of 640 MHz folded without any additional spectrum or base stations. Increasing capacity allows multiple transmit and receive signals simultaneously and makes it possible to carry more bits per symbol, improving throughput and better use of the spectrum. Besides, the battery lifetime is approximately ten times longer and has a closer alignment with the 5G to improve security of future networks (Dahlman et al. 2016). We intend to add the 5G in future updates as it promises to carry up to 1000 times more traffic than the current LTE infrastructure (Beyranvand et al. 2016; Ji et al. 2018). The 5G networks should consume up to 90% less energy than current 4G networks. Connection times between mobile devices must be less than five milliseconds (ms). The number of devices connected per area must be 50 to 100 times greater than the current one. Propagation models Knowing the communication channel is essential when the objective is to carry out a good coverage plan. One way to do this is through propagation models also known as path loss models. Propagation models consist in one or more mathematical expressions capable of predicting the attenuation suffered by the electromagnetic signal during its propagation in the communication channel (Furtado et al. 2016). There is a wide variety of propagation models, most obtained empirically. Empirical propagation models are based on several measurements and observations in real propagation environments. The equation that represents an empirical model is created to best fit the measured path loss data. For an empirical model to efficiently represent propagation losses in a given environment, it must have its parameters derived from the location characteristics, linked to the system frequency operation and effective antenna heights used for signal transmission and reception (Saunders and Aragon-Zavala 2007). In an outdoor environment, there are large distances between the transmitter and the receiver. In general, the signal propagation is somewhat predictable. Therefore, if we have information about the topography and the constructions in the terrain, we can efficiently determine the coverage area (Abbas et al. 2012). For the outdoor network design, we implemented the SUI, Cost-231, and ECC- 33 path loss models. The SUI model is an empirical model for radio channel characterization frequencies below 11 GHz, considering an initial distance of 100 m (Lima 2017). Cost-231 considers differences in height between the transmitter (Tx) and receiver antennas (Rx). The location to perform coverage analysis is more recommended for urban environments, using 1 km as the initial distance (Singh 2012). The ECC-33 propagation model can predict signals in frequencies bands up to 3.5 GHz, with the peculiarity of considering both Tx and Rx antenna gains and their heights, as variables in channel modeling (Abhayawardhana et al. 2005; Ayyappan and Dananjayan 2008). Thus, this model can predict path loss in urban environments, but it is not recommended for rural environments. In indoor environments, the coverage area has a much smaller coverage radius combined with greater complexity and significant variability of materials, so predicting electromagnetic signal propagation is considerably more complex than in outdoor environments (Sarkar et al. 2003). Therefore, for indoor modeling, was implemented Motley Keenan and ITU-R P.1238-1 path loss models. Motley Keenan is a complete model for signal prediction in indoor environments with obstacles since it considers both floors and walls (Solahuddin and Mardeni 2011). The ITU-R P.1238-1 model predict signals in a frequency range between 900 MHz and 100 GHz in indoor environments. This model considers reflection and diffraction on fixed objects, transmission through walls, floors, and other fixed obstacles, energy confinement in corridors, people, and objects in motion in the environment (Series 2012). However, both models consider floors and the simulator does not yet have a multi-floor simulation mode. Some path loss models can be used for both indoor and outdoor environments, they are based on the measured data, and due to not considering walls, they can adapt to both environments. The models implemented are Close-In and Floating Intercept. The Close-In model is a generic model to analyze propagation signals describing the loss on a large scale in a given environment. This model provides a preliminary prediction for path loss for future 5G systems (Sun et al. 2016, 2015; Maccartney et al. 2015). The Floating Intercept model requires two parameters and is not physically based on the transmitted power of the measurements (MacCartney et al. 2013; Samimi et al. 2015). It has the variables alpha and beta, alpha being the point of interception in dB, and beta is the slope of the line, also with a random variable Gaussian with zero mean (in dB) (Maccartney et al. 2015). For each model mentioned above, we use metrics to classify the environment better. These metrics are:Path loss measure how much the signal density is decreased with distance; Signal strength the amount of signal arriving at the mobile device; SNR the relationship between the amplitude of the desired analog or digital data signal and the noise amplitude in a transmission channel at a specific point in time; SINR measure the quality of wireless connections; Electrical field strength the harmfulness of the electromagnetic field in the environment; Channel capacity the quality of the internet. Also, the RSRQ measurements and the Root Mean Square Error (RMSE) of the models concerning the measured data are available. The RSRQ is more reliable to visualize the quality of the signal since it uses the number of resources for that. Moreover, the RMSE is important to know how close the predicted model is to real data. Optimization techniques The metaheuristic optimization techniques are those whose problem-solving incorporates intelligent processes that tend to optimize better the solution obtained (Colin 2007). An optimal problem solution is not always the target of these methods (Arcanjo 2014). Therefore, these methods usually find the best possible solutions to problems, not exact solutions. A metaheuristic optimization technique called a genetic algorithm (GA) was used to achieve a more precise solution by using genetic parameters to intensify and diversify the population (Deb et al. 2002). The GA has several steps to be completed, specifically: selection, crossing, crossover, and mutation. The selection methods can be used both in choosing which individuals will be parents and choosing the best adapted to pass on to the next generation (Mirjalili 2019). Each individual is an abstract representation of a solution to the problem. Natural selection is a criterion for choosing the best solutions, eliminating the bad ones, and crossing and mutation techniques to obtain new solutions (Mirjalili 2019). In this work, we use a genetic algorithm to determine the optimal position for the installation of routers in order to obtain greater signal coverage. Each gene represents X and Y points of the position of the router, having as fitness the quality of the coverage of the signal in the environment. The quality of the coverage represents how good the signal strength is in the environment, and is calculated by Ziwi with the data of measurement, that is, latitude, longitude, and signal strength. The signal strength shows the quality of the coverage, and the user can define which will be the best for his project. Therefore, the best chromosome will be the one with the most significant coverage of the environment according to the signal strength defined by the user. This chromosome in GA uses path loss as an evaluation method. Each iteration performed the selection by the tournament method, the crossover by binary mask, and a mutation at a rate of 0.1, considering elitism. Each population has a size of 100 chromosomes. The user defines the number of generations for the simulation, the more generations, the more accurate it will be, also increasing the algorithm execution time. In the developed simulator also is possible to optimize from the permutation, a technique used when defining the possible points to install the access point. In this way, are exchanged the sockets coordinates, and is calculated the coverage of each generated option, showing only the best positions coverage in the environment. Developed framework This section describes the framework development and functioning. Besides, it explains the use case of each software. Finally, the virtual reality application can do a simulation with high interactivity and make it easier to perform analyses in the environment. The developed application can be found at https://play.google.com/store/apps/details?id=com.lidiaxp.ziwi and the virtual reality application can be downloaded from https://play.google.com/store/apps/details?id=lidiaxpziwivr. The website can be accessed at http://ziwi.herokuapp.com/, which also provides a link to mobile application, virtual reality, and instructions for using each software. Mobile application The application has an indoor and outdoor mode, being different in frequency values and how distance is recorded. To begin radio channel modeling, it is necessary to perform measurements in the environment. So, mobile applications can make these measurements, being compatible with Ziwi websites or virtual reality applications. Besides, its interface optimization makes the network planning process faster and more accurate. For example, it can capture up to 20 points per second and record localization dynamically and automatically. In indoor mode, the user can choose between preset frequencies for Wi-Fi currently, 2400, 5200 MHz, or 60 GHz. It is possible to pause the recording between each point to save them in a single file. Indoor measurements based on a Cartesian plane, so it is necessary to insert the X and Y axes at each point recorded. Beyond showing signal strength and electric field strength values in real-time and plotting the history of the last 30 s on a graph, it is possible to export the measurement log to both e-mail and the internal memory of the device. In outdoor mode, the user chooses between predefined frequencies for LTE, being 700, 1800, or 2500 MHz. The recordings can be made just by walking around the environment, capturing the latitude and longitude of the mobile device. The signal strength values, electric field, and RSRQ are shown in real-time and in a graph with the last 30 s of measurements. It is possible to export both to email and the internal memory of the device. Both interfaces are shown in Fig. 1.Fig. 1 Mobile application We adopted a top-down approach in the tool development, which consists of starting from an overview, finalization, and summarization of the functionalities of the system and then detailing its subcomponents and most basic functionalities. Figure 2 shows the use case diagram for the mobile application, capable of performing signal measurements, which served as a basis for detailing the structure of the tool.Fig. 2 Use case diagram of mobile application There is only one stakeholder in the software in question, the user interested in performing the signal measurement, with two primary functions: indoor and outdoor measurement, subdivided into mandatory and optional functions. For example, record data when entered in indoor measurement mode as a mandatory function and send the data by email as an optional function. Indoor mode on website The indoor mode is divided into five sections: simulate models (that generate heat maps), compare models, evaluate the model, optimize the simulation, and create scenarios, respectively, as described below. First section: indoor mode In the first section, data on the maximum room width and length (m), the X and Y axis of the routers (m), and the received power at initial point d0 (dBm) are required. There is other information that is also necessary. However, it comes predefined and can be changed, namely: minimum measurement distance (m), transmission power (dBm), frequency (MHz), transmitting and receiving antenna gain (dBi), bandwidth (MHz), and the propagation model to simulate, being able to choose between Close-In, Floating Intercept, Motley Keenan and ITU-R P.1238-1. Whether user not defines the environment or using the Floating Intercept model is also necessary to insert the indoor measurement file from the Ziwi application. In addition to inserting the wall file generated by the scenery creation, available on the website itself, this field is only necessary if using the Motley Keenan model since the model considers the walls of the environment. The heat map of the propagation model chosen in the inserted environment will be generated from this data. Second section: indoor mode The second section can compare the main propagation models with the measured data and perform the main calculations: the value of path loss exponent and the standard deviation to the measured data. It is necessary to enter the point values X and Y of the router (m), the environment where the measurements took place, and the indoor measurement file, generated by the Ziwi mobile application. Some information is predefined by the application, but can be changed according to the needs of the user, such as: frequency (MHz); the first point distance measured (m); transmission power (dBm), and the receiving and transmitting antenna gain (dBi). The simulator uses this data to adequately calculate the radiated power. Figure 3. shows the comparison mode.Fig. 3 Comparison mode—indoor mode When comparing models, the comparison graph among the propagation models and the measured data. Third section: indoor mode This section is generated automatically after comparing the data in the second section above. The third section presents the dispersion graph of path loss exponent and a table containing the path loss exponent value and standard deviation for each model, according to the comparison made in the Sect. 4.2.2. Fourth section: indoor mode In indoor mode, the model optimization, and it needs the same information used in the first section. The section aims to help indoor network planning, making the process faster to define the best location for routers. Initially, to do the optimization, the user chooses the propagation model. Then, he will be informed of the best location to install one or more routers. Finally, it can consider sockets locations that routers will be installed. There are two ways to use this section, considering or not the location of sockets. First, the algorithm performs the permutation between all points X and Y of the sockets if they are considered. Second, it calculates all possible combinations of sockets to inform the user which will be the best combination ultimately. If not considering the sockets, the best location for the routers will be simulated using the genetic algorithm. If the user wishes to consider the sockets, he must select the “with sockets” option at the beginning. Then, inform the X and Y axes of the sockets and the number of routers to insert into the environment. Otherwise, must select “no sockets” and inform the number of GA iterations and quality threshold in dB. These predefined values can be configured according to the needs of the user. The more iterations, the closer to the optimal value the algorithm will be. However, it will take a longer time to get the results. A heat map of the last route will be generated with the best result when performing the simulation. Below the title and best positions for the X and Y axis will be shown. It is possible to download the heat map of the environment. If the user wishes not to consider the sockets, the result will be a fitness graph of the genetic algorithm and heat maps with the best location of the router. In addition, heat maps will be generated informing the signal strength, path loss, SNR, SINR, electric field strength, and capacity. It is possible to download all heat maps in a compressed file. Figure 4. Presents the optimization mode.Fig. 4 Optimization mode—indoor mode Fifth section: indoor mode For the scenario construction, the environment width and total length in meters, both values must be informed and separated by the letter x, without space, e.g., 5 × 7, for an environment with 5 m of width and 7 m of length. In the central part of the page, a square appears with the exact dimensions of the room. In this square, the user designs his environment, and the software automatically identifies whether the user wants to insert a vertical or horizontal line according to the clicks given. Then the start and end values of the line must be entered and the distance to the edge in meters. For example, when building a wall, lines are created in the square, where the thickest black line represents the concrete wall, the blackest thin line for the wall with the window, and the red line is the window. On the left side of the page, there are three buttons and a text box. The text box is the interactive history of the environment that the user is building. For example, when creating walls on the canvas, the information will be added dynamically in the text box, informing the orientation of the wall, its coordinates, and material. It is also possible to add or delete lines by typing or removing information from the text box. Deleting the wall coordinates from the text box will remove the lines on the canvas. However, it may be a line placed a long time ago, so the user no longer knows which line contains the wall information. Therefore, it is possible to select a line and display it on the canvas. In this way, the line corresponding to that information will turn green, and the user can decide if he wants to change it. Finally, is used the “Download” button to download the information in a text file, used for modeling with the Motley Keenan model. The scenario construction mode is shown in Fig. 5.Fig. 5 Scenario construction mode—indoor model Outdoor mode on website The propagation models available in the outdoor website mode are SUI, Cost 231, ECC 33, Close-In, and Floating Intercept. The main function of this mode is the comparison among the models and the measured data and calculations of path loss exponent value and standard deviation. Aiming to perform the simulation is necessary to enter the values of:Frequency (MHz); Latitude and longitude of the transmitting antenna; Transmitting antenna height (m); Receiving antenna height (m); Transmitting antenna power (dBm); The shortest distance that occurred the measurement (m); Transmitting antenna gain (dBi); Receiving antenna gain (dBi); The file with the outdoor measurement values. The mobile version of Ziwi can export the file with outdoor measurements with the data from the transmitting and receiving antennas used to calculate the adequate radiated power. For website development, a top-down approach, aiming to present the functionalities of the system briefly and then detail its subcomponents and its most basic functionalities. Figure 6 shows the use case diagram for the website, where are carried out data modeling, comparison, and optimization, which served as a basis for detailing the structure of the tool.Fig. 6 Use case diagram of website—outdoor mode There is only one stakeholder in the software, the user interested in performing signal analysis or optimizing it, and that is the user shown in the diagram. The website has five main functions shown in the image, which have sub-functions that can be mandatory or optional. The mandatory ones include relationship, as is the case when comparing indoor or outdoor models. Consequently, the calculation of propagation models parameters will be made. The options demonstrate the extended relationship, such as downloading the files after each operation. Virtual reality application Finally, we developed a virtual reality application to visualize data in 3D to make observation more interactive. For example, they can simulate environments and visualize their heat map to the user, informing the X and Y axis of the router and path loss of the point the user is examining. The software also makes it possible to walk around the scene to make the observations more closely. It is necessary to inform these data to simulate an environment:Width and length of the environment to be simulated (m); X and Y axis of the routers (m); Received power in the shortest distance to the router (dBm); The shortest distance to the router (m); The transmission power of the router (dBm); Transmitting and receiving antenna gain (dBi); Router frequency (MHz); Technology bandwidth (MHz); Walls (only mandatory if using Motley Keenan model); Path loss exponent (which can be calculated on the website); Propagation model (between Close-In, Motley Keenan, and ITU-R P.1238-8). The user will see the screen where the virtual reality environment will be when entering this data (Fig. 7). It is presented on a heat map on the floor with path loss measures according to the chosen propagation model and environment. Upper left corner shows the value of each color and the user can move around the scenario using the controller and look around with their head. The screen center has a circle that is moved along with the head of the user, when positioned over the path loss, the precise value in the location and the position of X and Y axes are in the upper right corner.Fig. 7 Visualization of virtual reality application As with previous software, the virtual reality application also adopted a top-down approach in the tool development, aiming to present a summary version of the functionalities of the system and then detail its subcomponents and its most basic functionalities. Figure 8 shows the use case diagram for the VR application able to view the path loss in 3D, interactively, which served as a basis for detailing the tool structure.Fig. 8 Use case diagram of virtual reality application The software has only one stakeholder. He is the user interested in analyzing the signal propagation more interactively, and that is the user shown in the diagram. Its main function is indoor modeling, being divided only into functions that must be performed for the mode, represented by include in the diagram. For example, take the heat map generation in virtual reality and visualize the X and Y coordinates and the path loss in the scenario. Case study In this section, two use cases of the Ziwi framework will be demonstrated. Initially, in an indoor environment, measurement campaigns were carried out with the mobile application shown in Sect. 4.1. The scenario used for measurement is described in Sect. 5.1. Then, with the captured data, it was possible to use the tools on the website to generate the results shown in Sects. 5.2 and 5.3. Also, Sect. 5.4 shows the same scenario being simulated in virtual reality. Finally, measurement campaigns were carried out in an outdoor environment with the mobile application described in Sect. 4.1. The scenario used for the measurements is described in Sect. 5.5. Thus, the captured data were used to compare the outdoor models, the results being shown in Sect. 5.6. In this way, it was possible to validate the framework of the Ziwi and its accuracy for wireless network applications. Indoor scenario The indoor measurements were made in a classroom at the Federal University of Pará (UFPA). The cell phone was vertical during the entire measurement, with a cross-polarization concerning the router, approximately 1 m from the floor. The room is 8 × 6 m. The router was installed in one corner of the room, being half a meter from each wall. Therefore, there is also one meter from the floor. The trajectory performed in the room is shown in Fig. 9. The measurements were made in 3 radials every 1 m of distance, totaling 22 points. The cell phone remained stationary at each point for approximately 10 s, and with the application, it is possible to record 20 points per second. Therefore, approximately 4400 points were measured across the room.Fig. 9 Measurements performed in the indoor setting Measurements were performed using a Samsung Galaxy S9. Moreover, the router used for the measurements was a D-Link IEEE 802.11n standard, at the frequency of 2400 MHz, having an antenna gain of approximately five dBi, and transmission power of 15 dBm. Indoor models In the first section of the website (as shown in Sect. 4.2.1), it is possible to simulate an indoor environment on a heat map according to several metrics, such as signal strength, path loss, SNR, SINR, channel capacity, and electric field strength. In addition to being possible to simulate the environment using the Close-In model, Motley Keenan, ITU-R P.1238-1 and Floating Intercept. For this simulation, an IEEE 802.11n router was used at 2400 MHz, a transmission antenna gain of 5 dBi, and a receiver antenna gain of 1 dBi. The router was positioned at a half-meter distance from each wall, one meter high from the ground. The router transmission power is 15 dBm, and the received power at point d0 was − 43 dBm. Figure 10 shows the heat map of the path loss in the environment. The signal tends to attenuate as the distance from the router increases. The simulator also generates the heat map of the environment considering the signal strength, SNR, SINR, channel capacity, and electric field intensity, aiming to optimize the work of the user by generating simulations for the most diverse purposes.Fig. 10 Heat map of path loss In the second section, the indoor mode of the website (as shown in Sect. 4.2.2), it is possible to compare all the propagation models available for indoor mode and the measured data. Needing to be inserted the frequency value, transmission power, received power at point d0, antennas gains, routers position, the environment where the measurements took place, and the file with measured data. Figure 11 shows a comparison between the models.Fig. 11 Comparison between indoor models The parameters used for Close-In and Floating Intercept models were calculated from the data entered in the system. For the ITU-R P.1238-1 and Motley Keenan models, the attenuation coefficient used was tabulated according to the environment specified by the user, in the case of this work being a furnished environment. According to the Friis model, the signal loss for 1 m of distance at the 2400 MHz frequency is 40 dB, as shown in the graph for the Close-In, ITU-R P.1238-1, and Motley Keenan models. In the Floating Intercept model, the initial loss was approximately 50 dB, closer to the measured data as the model is based only on measurements of the environment, adjusting to them. As the measurements were made with cross-polarization, the path loss of the environment tends to be greater than with the Friis model. From this information, the RMSE value of each model is also generated concerning the measured data and auxiliary values (attenuation coefficient and the alpha and beta values used in the Floating Intercept model). The values are being shown in Table 3.Table 3 Comparison metrics of indoor models Close-in Motley Keenan ITU-R P.1238-1 Floating intercept Auxiliary n = 2.58 n = 3 n = 2.8 Alpha = 49.86 Beta = 1.45 RMSE 5.79 5.81 5.72 3.57 The attenuation coefficient and beta values tend to be close, and the Free Space Path Loss (FSPL) and alpha values representing the curve slope and the initial loss, respectively. The main difference is that the alpha and beta values are calculated based exclusively on measurements made in the environment. The RMSE value shows how close each model is to the measured data. For example, it can be seen that the Floating Intercept model was the one that came closest to the measured data as the model depends on two physical parameters to adjust, the alpha and the beta. In contrast, the other models depend only on the attenuation coefficient. The other models, on the other hand, obtained a slight difference between the RMSE. Indoor optimization Through the website, it is possible to optimize indoor environments. The “No socket” mode is the one that uses a genetic algorithm to define the best points to install the routers. The optimization by genetic algorithm aims at closed environments that are still under construction, that is, when it is possible to install sockets in any position. In this way, GA searches for the entire environment where there will be the best points to insert a router, aiming at the best quality of signal in the environment. The data required to perform this simulation are the same as in the first section, adding the number of interactions with the genetic algorithm and the quality threshold in dB. Of course, with more interactions, the result will be better. However, it will also take a longer time. Furthermore, the quality threshold is the acceptable received power for the project being optimized. Figure 12 presents the genetic algorithm fitness curve showing the best individual fitness and the average population that tends to converge in a minimal error over the generations. However, it is also possible to observe elitism in the algorithm, since fitness has continuously decreased over the generations.Fig. 12 Fitness throughout generations Figure 13 shows the heat map before and after optimization. With the position of the router optimization, there was an improvement of 2 dB in the environment. Also, on the path loss map, it is possible to view the routers’ location implemented in the environment and coverage percentage, taking into account the acceptable signal strength defined by the user. For the scenario used in this work, we obtained 89.81% of coverage, with the best location for the router being the point [4.05, 3.01]. There is also the “With socket” mode that performs the optimization with the aid of permutation. This mode is ideal for environments that are ready and need to optimize the signal, as one of the inputs of this method is the location of the environment sockets and the number of routers that will be installed. Thus, the permutation between the coordinates of the sockets is made, generating the X and Y axes of input to simulate which will have the least path loss.Fig. 13 Heat map of loss between original and optimized positions The necessary data are the same as the simulation in the first section, adding the X and Y axis of the sockets and the router number that is intended to be installed. The simulation is performed between all the coordinates. The simulator exports only the one with the best result (Fig. 14), showing the position of the router in the title text, represented by red dots in the image. The sockets for this simulation were located on the axes [0, 1], [0, 3], [0, 5], [2, 4], [4, 3] and [6, 2].Fig. 14 Optimization considering sockets Analyzing Fig. 10 with Fig. 14, it is possible to see from the path loss range in the environment that the loss has decreased considerably. The maximum loss improved from 60 to 56 dB, and the minimum loss was improved from 48 to 42 dB. Thus, this function aims to optimize propagation signals in environments that are already ready and want to improve signal quality or add a new router. Virtual reality A virtual reality simulation was performed using Ziwi with the same environment and settings as previous simulations. Figure 15 shows how the scenario looks as soon as the user enters the environment. For example, Fig. 15a is looking down, and Fig. 15b is looking forward. Figure 15c shows the floor with path loss when the user turns his head to look at the router and walks to the end of the environment to have another view of the scenario. With this, the user can see that the path loss on the router is almost null and increases with distance. As the user will be in virtual reality, he will be able to assess how much the loss increases with distance and the exact loss value at each point.Fig. 15 Virtual reality simulation It is possible to see that in the upper left part, there is a bar that shows the minimum and maximum path loss and the equivalent path loss in the environment. In the upper right part are the X and Y position in meters and the path loss in dB for the point observed by the user. The point is defined for the area where the white circle in the center of the screen is pointing. Observing the upper right area of the image shows the values changes according to the location the user is looking. The importance of a virtual reality application for the area is both to increase interactivity with the environment as well helping to analyze more complex environments. Nevertheless, it is also essential for the educational area when it will be able to show students in a more interactive way how signal propagation works. Outdoor scenario The outdoor measurements were taken at the Science and Technology Park (PCT) at UFPA, which presents a wooded environment with some constructions along the way. The street just in front of a building was chosen, being approximately 200 m long, totaling 8676 measured points since approximately 20 points are recorded per second. The cell phone was vertical during the entire measurement, with a cross-polarization concerning the transmitting antenna, approximately 1 m from the ground. The measurements were made using a Samsung Galaxy S9 smartphone connected to an LTE network, a transmission antenna approximately 450 m from the park. The transmission antenna has a frequency of 1870 MHz, a height of 50 m from the ground. It has a transmission power of 60 W, and the antenna gain is 16.71 dB. Outdoor models For the outdoor website mode, a comparison was made between the models and the measured data. The available models are SUI, Cost 231, ECC-33, Close-In, and Floating Intercept. For this simulation, an LTE network was used, whose antenna distance is approximately 450 m from the environment of the measurement. The antenna has a frequency of 1870 MHz, 50 m high, and has a transmission power of 47.78 dB and an antenna gain of 16.71 dB. It is also necessary to enter the measured data obtained through the measurement of a mobile device. Therefore, the gain of the receiving antenna is 1 dB and defines the environment of a big city since measurements were made in a wooded environment. The models were not made specifically for that environment. Figure 16 compares the models and the measured data. Notice that the graph shows only the Floating Intercept model curve and the Close-In model just below this curve, showing a difference of 0.0032 dB between the RMSE of each model.Fig. 16 Comparison between outdoor models The SUI, COST 231, and ECC 33 models were not developed to be applied in wooded environments. Also, these models do not have any physical parameters based on measurements of the environment. The parameters of the models are adjustable for the predefined environments for each one, so the path loss for these models tends to present a more significant error concerning the measured data. The Close-In and Floating Intercept models are based on physical parameters obtained from environmental measurements, making it easier to adapt to a wooded environment. According to the Friis model, the path loss for 450 m and 1870 MHz is 150 dB. In the measurement, a path loss of 153 dB was obtained in both models for this distance. There was a minor variation because the path loss calculated by the Friis model is an initial distance of 100 m in an outdoor environment, so there may be a slight variation in the path loss compared to that calculated by the Friis model due to the curve slope. From this information, the RMSE value of each model is also generated concerning the measured data and auxiliary values, such as the value of the attenuation coefficient used in the Close-In model and the alpha and beta values used in the Floating Intercept model. The values are being shown in Table 4.Table 4 Comparison metrics of outdoor models SUI COST 321 ECC 33 Close-in Floating intercept Auxiliary – – – – Alpha = 163.97 Beta = 2.74 RMSE 21.09 14.1 4.89 2.3994 2.3962 The Floating Intercept model was the one that came closest to the measured data. As the environment was wooded and several constructions around, physical parameters, alpha and beta of the model, made it fit better to the environment than the other propagation models. Conclusion This work aimed to develop a framework for measuring, modeling, and planning environments, indoor or outdoor. For this, three software were developed with communication between them for data transfer aiming at the efficiency of the work of a telecommunications engineer, who can do his job faster, without losing quality. In addition to having a virtual reality version aiming to make the signal analysis mode more interactive, it is more attractive for new professionals in the area. The framework can assist in network planning, making signal modeling, measurement, and optimization faster. Furthermore, assist in the academic and professional area by having a more straightforward and more intuitive interface and analyzing the spread of the environment in virtual reality, increasing the interactivity and understanding of students, researchers and engineers with the content. Acknowledgements This study was funded by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES – Brazil) with Grant Number 88882.426579/2019-01. Author contribution LR: Conceptualization, Methodology, Software, Formal analysis, Investigation, Data Curation, Writing – Original Draft, Visualization. SF: Validation, Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualization. KV: Writing - Review & Editing. JA Conceptualization, Writing – Review & Editing, Supervision. IB: Conceptualization, Methodology, Validation, Data Curation, Writing – Review & Editing, Supervision. Funding The authors have not disclosed any funding. Data availability Enquiries about data availability should be directed to the authors. Declarations Conflict of interest All authors declare that they have no conflict of interest. Human or animal rights This article does not contain any studies with human participants or animals performed by any of the authors. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abbas Z, Nasreddine J, Riihijarvi J, Mahonen P (2012) Long-term indoor propagation models for radio resource management. In: 2012 IEEE international symposium on a world of wireless, mobile and multimedia networks (WoWMoM), IEEE, pp 1–9 Abhayawardhana V, Wassell I, Crosby D, Sellars M, Brown M (2005) Com- parison of empirical propagation path loss models for fixed wireless access systems. In: 2005 IEEE 61st vehicular technology conference, IEEE, vol 1, pp 73–77 Alias AH, Ali DM, bin Sahrani MN (2016) Performance measurement of lTE along light rapid transit (lRT) railway track of Kelana Jaya line. In: 2016 7th IEEE control and system graduate research colloquium (ICSGRC), IEEE, pp 67–72 Ahmavaara K Haverinen H Pichna R Interworking architecture between 3GPP and WLAN systems IEEE Commun Mag 2003 41 11 74 81 10.1109/MCOM.2003.1244926 Arcanjo D (2014) Metodologia multiestágio para restabelecimento de sistemas elétricos de distribuição utilizando algoritmos bio-inspirados. Master thesis, UFJV, 2014 Ayyappan K Dananjayan P Propagation model for highway in mobile communication system Ubiquitous Comput Commun J 2008 3 4 61 66 Backholm A, Salorinne S, Ylinen H, Groeber M, Vuornos L, Ikonen R, Ahonen J, Everitt A, McLeod A, Salmi P et al (2013) Database synchronization via mobile network. US Patent 8,620,858 Beyranvand H Levesque M Maier M Salehi JA Verikoukis C Tipper D Toward 5G: Fiwi enhanced lte-a hetnets with reliable low-latency fiber backhaul sharing and wifi offloading IEEE/ACM Trans Netw 2016 25 2 690 707 10.1109/TNET.2016.2599780 Bhatt CR Redmayne M Abramson MJ Benke G Instruments to assess and measure personal and environmental radiofrequency-electromagnetic field exposures Australas Phys Eng Sci Med 2016 39 1 29 42 10.1007/s13246-015-0412-z 26684750 Blackard KL Rappaport TS Bostian CW Measurements and models of radio frequency impulsive noise for indoor wireless communications IEEE J Sel Areas Commun 1993 11 7 991 1001 10.1109/49.233212 Castro BSL et al (2010) Modelo de propagação para redes sem fio fixas na banda de 5, 8 GHz em cidades típicas da região amazônica. Master thesis. UFPA, 2010 Charbonneau H (2005) System and method for designing a communications network. US Patent App. 10/744,018 Chou HT, Cheng DY, Chen NW (2016) The pattern calibration of phased array antennas via the implementation of genetic algorithm with measurement system. In: 2016 USNC-URSI radio science meeting, IEEE, pp 83–84 Colin EC (2007) Pesquisa Operacional: 170 aplicações em estratégia, finanças, logística, produção, marketing e vendas. Livros Técnicos e Científicos Dahlman E Parkvall S Skold J 4G, LTE-advanced pro and the road to 5G 2016 Academic Press Deb K Pratap A Agarwal S Meyarivan T A fast and elitist multi-objective genetic algorithm: Nsga-ii IEEE Trans Evol Comput 2002 6 2 182 197 10.1109/4235.996017 Furtado PPF et al (2016) Ferramenta em Java para simulação de cobertura de redes wlan em ambientes indoor Gao M, Huang L, Cui X, Cai H, Gao Z (2013) Intelligent coverage optimization with multi-objective genetic algorithm in cellular system. In: 8th international conference on Computer Science & Education. IEEE, pp 859–863 Garcia-Martinez C, Rodriguez FJ, Lozano M (2018) Genetic algorithms, Springer International Publishing, Cham, pp 431–464. 10.1007/978-3-319-07124-4_28 Huang M Liu A Xiong NN Wang T Vasilakos AV An effective service-oriented networking management architecture for 5G-enabled internet of things Comput Netw 2020 173 107208 10.1016/j.comnet.2020.107208 Ijaz A Zhang L Grau M Mohamed A Vural S Quddus AU Imran MA Foh CH Tafazolli R Enabling massive IoT in 5G and beyond systems: phy radio frame design considerations IEEE Access 2016 4 3322 3339 10.1109/ACCESS.2016.2584178 Ji H Park S Yeo J Kim Y Lee J Shim B Ultra-reliable and low-latency communications in 5G downlink: physical layer aspects IEEE Wirel Commun 2018 25 3 124 130 10.1109/MWC.2018.1700294 Kar K, Datta S, Pal M, Ghatak R (2016) Motley Keenan model of in-building coverage analysis of IEEE 802.11 n WLAN signal in electronics and communication engineering department of national institute of technology Durgapur. In: 2016 international conference on microelectronics, computing and communications (MicroCom), IEEE, pp 1–6 Kelly MG (2016) The automatic placement of multiple indoor antennas using particle swarm optimization. PhD thesis, Loughborough University Labovitz C (2020) Network traffic insights time covid19. https://www.nokia.com/blog/network-traffic-insights-time-covid-19-march-23-29-update/, online, Accessed 19 June 2020 Liechty LC (2007) Path loss measurements and model analysis of a 2.4 GHz wireless network in an outdoor environment. PhD thesis, Georgia Institute of Technology Lima GD (2017) Desenvolvimento de um aplicativo baseado em fdtd para o estudo da propagação de ondas eletromagnéticas Maccartney GR Rappaport TS Sun S Deng S Indoor office wideband millimeter-wave propagation measurements and channel models at 28 and 73 GHz for ultra-dense 5G wireless networks IEEE Access 2015 3 2388 2424 10.1109/ACCESS.2015.2486778 MacCartney GR, Zhang J, Nie S, Rappaport TS (2013) Path loss models for 5G millimeter wave propagation channels in urban microcells. In: 2013 IEEE global communications conference (GLOBECOM), IEEE, pp 3948–3953 Maghdid HS, Ghafoor KZ, Sadiq AS, Curran K, Rabie K (2020) A novel AI-enabled framework to diagnose coronavirus covid 19 using smartphone embedded sensors: design study. arXiv preprint arXiv:200307434 McElroy JA Egan KM Titus-Ernstoff L Anderson HA Trentham-Dietz A Hampton JM Newcomb PA Occupational exposure to electromagnetic field and breast cancer risk in a large, population-based, case-control study in the United States J Occup Environ Med 2007 49 3 266 274 10.1097/JOM.0b013e318032259b 17351512 Mirjalili S (2019) Evolutionary algorithms and neural networks. In: Studies in computational intelligence, vol 780. Springer, Berlin/Heidelberg, Germany Najnudel M (2004) Estudo de propagação em ambientes fechados para o planejamento de wlans. Rio de Janeiro 136 Qin C Liu D Lai Z Zhang J Analysis and solution for WLAN wireless interference based on Ranplan iBuildNet Telecommun Technol 2011 9 127 131 Rappaport TS Sun S Mayzus R Zhao H Azar Y Wang K Wong GN Schulz JK Samimi M Gutierrez F Millimeter wave mobile communications for 5G cellular: it will work! IEEE Access 2013 1 335 349 10.1109/ACCESS.2013.2260813 Rodrigues JDC et al (2011) Planejamento de redes de comunicação sem fio para ambiente indoor considerando os efeitos da polarização das antenas: abordagem baseada em medições Saeed N, Bader A, Al-Naffouri TY, Alouini MS (2020) When wireless communication faces covid-19: combating the pandemic and saving the economy. arXiv preprint arXiv:200506637 Samimi MK Rappaport TS MacCartney GR Probabilistic omnidirectional path loss models for millimeter-wave outdoor communications IEEE Wirel Commun Lett 2015 4 4 357 360 10.1109/LWC.2015.2417559 Sarkar TK Ji Z Kim K Medouri A Salazar-Palma M A survey of various propagation models for mobile communication IEEE Antennas Propag Mag 2003 45 3 51 82 10.1109/MAP.2003.1232163 Saunders SR Aragon-Zavala A Antennas and propagation for wireless communication systems 2007 John Wiley and Sons Series P (2012) Propagation data and prediction methods for the planning of indoor radiocommunication systems and radio local area networks in the frequency range 900 MHz to 100 GHz. Recommendation ITU-R pp 1238–7 Shakir MZ, Ramzan N (2020) Ai for emerging verticals: robotic-human computing, sensing and networking. The IET, UK Sharma SK Wang X Toward massive machine type communications in ultra-dense cellular IoT networks: current issues and machine learning-assisted solutions IEEE Commun Surv Tutor 2019 22 1 426 471 10.1109/COMST.2019.2916177 Singh Y Comparison of okumura, hata and cost-231 models on the basis of path loss and signal strength Int J Comput Appl 2012 59 11 37 82 Skouby KE, Lynggaard P (2014) Smart home and smart city solutions enabled by 5G, IOT, AAI and cot services. In: 2014 international conference on contemporary computing and informatics (IC3I), IEEE, pp 874–878 Solahuddin Y, Mardeni R (2011) Indoor empirical path loss prediction model for 2.4 GHz 802.11 n network. In: 2011 IEEE international conference on control system, computing and engineering, IEEE, pp 12–17 Sun S, Thomas TA, Rappaport TS, Nguyen H, Kovacs IZ, Rodriguez I (2015) Path loss, shadow fading, and line-of-sight probability models for 5G urban macro-cellular scenarios. In: 2015 IEEE Globecom workshops (GC Wkshps), IEEE, pp 1–7 Sun S, Rappaport TS, Rangan S, Thomas TA, Ghosh A, Kovacs IZ, Rodriguez I, Koymen O, Partyka A, Jarvelainen J (2016) Propagation path loss models for 5G urban micro-and macro-cellular scenarios. In: 2016 IEEE 83rd vehicular technology conference (VTC Spring), IEEE, pp 1–6 Teguh R, Murakami R, Igarashi H (2014) Optimization of router deployment for sensor networks using genetic algorithms. In: International conference on artificial intelligence and soft computing, Springer, pp 468–479 Teixeira RBM, de Melo LFT, de Mesquita VA, Leandro dSP, de Souza IT (2017) Predição do sinal em uma rede local sem fio através de redes neurais artificiais Wickens CD (1992) Virtual reality and education. In: [Proceedings] 1992 IEEE international conference on systems, man, and cybernetics, IEEE, pp 842– 847 Yudha DMM Sudiarta PK Indra EN Analisis parameter jaringan hsdpa kondisi indoor dengan tems investigation dan G-Nettrack Pro E-J Spektrum 2016 3 1 40 46	36475039	PMC9716530	NO-CC CODE	2022-12-03 23:20:58	no	Soft comput. 2022 Dec 2;:1-21	utf-8	Soft comput	2,022	10.1007/s00500-022-07682-9	oa_other
==== Front AMS Rev AMS Review 1869-814X 1869-8182 Springer US New York 241 10.1007/s13162-022-00241-3 Commentary How robots will affect the future of retailing http://orcid.org/0000-0002-9605-2993 Guha Abhijit [email protected] 1 Grewal Dhruv [email protected] 23 1 grid.254567.7 0000 0000 9075 106X Department of Marketing, Darla Moore School of Business, University of South Carolina, Columbia, SC 29208 USA 2 grid.423152.3 0000 0001 0686 270X Toyota Chair in E-Commerce and Electronic Business and Professor of Marketing, Department of Marketing, Babson College, Babson Park, MA 02457 USA 3 grid.7340.0 0000 0001 2162 1699 Fractional, University of Bath, Bath, England 2 12 2022 18 16 9 2022 30 9 2022 © Academy of Marketing Science 2022, Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. To predict how robots will affect the future of retailing, the authors begin this article with a brief review of recent deployments of robots in the retail and service sectors, along with relevant research pertaining to their uses. Next, they focus on two crucial dimensions, reflecting whether the robots (1) are customer facing or not and (2) intend to augment or substitute for retail associates. These two dimensions suggest four strategic quadrants relating to future robot deployments; some quadrants are attractive in the near term, and others attractive in the years beyond. In closing, the authors also suggest an agenda for continued research. Keywords Robots Artificial Intelligence Retailing ==== Body pmcRetailers have started to deploy robots in stores (Forgan, 2020). In some cases, robots may help retailers gain efficiency by streamlining non–customer-facing functions (e.g., cleaning, inventory checking, price changes) and likely replacing the need for employees to do these tasks, while in other cases robots may directly perform customer-facing tasks (e.g., interacting with customers, providing promotional or product information), or assist sales associates with their customer-facing tasks. Our commentary builds on the work of Rindfleisch et al. (2022, in this issue), and presents both a new conceptual framework and a forward-looking research agenda. Rindfleisch et al. (2022, in this issue) present the results of their interviews with a senior manager at Softbank (which manufacturers robots, including Whiz), and a senior manager at Daiei (Japanese supermarket chain, which uses Whiz). They determined that initially Daiei deployed Whiz mainly to execute cleaning functions, which was particularly valuable during the COVID-19 pandemic, when staffing was limited but cleanliness demands were especially acute. This deployment attracted substantial consumer interest, leading to various photos of Whiz appearing on social media sites. Over time, the retailer deployed Whiz to share sales promotions, and employees added personalized decorations to the robot in their individual stores to reflect local landmarks. Yet Whiz’s success also has a backstory, in that Softbank had initially introduced a robot called Pepper, designed to interact directly with customers and address their customer service queries (Softbank, 2022b). But due to Pepper’s limited functionality, Softbank shifted its focus to the cleaning robot Whiz (Nussey, 2021), which might suggest that while robots may be able to execute non-customer-facing tasks, they are ill-equipped to engage in customer-facing applications. Beyond Pepper and Whiz, there have been many instances of retailers deploying robots. Considering first non-customer-facing domains, the Tally robot was deployed in Meijer stores, where it searches constantly for missing, mispriced, or misplaced products, and also assists with inventory updates. With its assistance, retailers are two to fifteen times more likely to identify out-of-stock items, than are retailers that do not rely on Tally (Ferguson, 2021). The Whiz robot cleans floors; some Hilton hotels earned nearly 2 × higher guest cleanliness scores after introducing it (Softbank, 2022a). Other robots deployed by retailers include (i) K5, which can patrol shopping malls and parking lots (Robinson, 2017), (ii) Millie, used in Woolworths stores (in Australia) to perform safety tasks, such as clean ups of spills, (iii) Amazon’s Kiva robots, located in distribution centers, that can perform sophisticated operations and keep track of millions of items (Forgan, 2020; Mims, 2018), and (iv) Walmart’s Bossa Nova robots, which track inventory (Nassauer & Cutter, 2019). Yet some robot deployments also entail customer-facing domains, and it is these robots that are more widely reported in the popular press. LoweBot (at Lowe’s) helps customers find items in stores, after they request its assistance via either a voice command or by typing on its touchscreen (Morgan, 2020). Robot servers, such as Peanut, deliver food; using infrared technology, it can navigate its path, relying on lidar to detect any obstacles (Simon, 2021). Various versions of robot baristas include Café X, Rocky, Monty Café, and Rozum Café (e.g., Hochman, 2018). Despite excitement about such encounters, expectations about the robots’ capabilities still need to be tempered (Davenport et al., 2020); their limited functionalities mean they cannot (fully) function sans human support. For example, Peanut assists human servers by bringing orders to the table, but the servers still take orders and interact with customers (Simon, 2021). Table 1 lists exemplars of robots deployed by retailers.Table 1 Exemplars of robot deployment in retail Name Domain (relatively speaking) Intent (relatively speaking) Photo Café X Customer-facing Augment From: https://thespoon.tech/cafe-x-shuts-down-its-three-downtown-san-francisco-locations/ retrieved on 09 Sep 2022 K5 Non–customer-facing Augment From: https://robots.ieee.org/robots/k5/ retrieved on 09 Sep 2022 LoweBot Customer-facing Replace From: https://www.lowesinnovationlabs.com/projects/lowebot retrieved on 09 Sep 2022 Peanut Customer-facing Augment From: https://www.wired.com/story/peanut-the-waiter-robot-is-proof-that-your-job-is-safe/ retrieved on 09 Sep 2022 Pepper (suspended) Customer-facing Replace From: https://www.softbankrobotics.com/emea/en/pepper retrieved on 09 Sep 2022 Tally Non–customer-facing Augment From: https://www.woodtv.com/news/kent-county/meet-tally-the-robot-helping-eliminate-empty-shelves-at-meijer/ retrieved on 09 Sep 2022 Whiz Non–customer-facing Replace From: https://www.businessinsider.com/softbank-robot-vacuum-whiz-self-driving-tech-photos-2019-11 retrieved on 09 Sep 2022 In this paper, we first provide an overview of pertinent AI and Robotics literature, predominantly from the marketing domain. Next, we develop and present a four-quadrant framework to understand the evolution and deployment of robotics in retail and service settings. Finally, we present an agenda for future research, to help stimulate and guide future examinations in this rapidly evolving field. Literature review We provide a summary of relevant prior research in Table 2. The summary pertains to either work in robotics or to work in artificial intelligence (hereafter, AI) that can inform how robots with AI capabilities will likely function, and influence customer, employee, and retailer/service provider performance. For a more comprehensive summary of robotic work, interested readers are referred to Mende et al. (2019).Table 2 Literature review Cite Technology focus Substantive focus Key points Castelo et al. (2018) Robots Consumers, generally (Very) humanlike robots may lead to discomfort Davenport et al. (2020) AI Marketing, broadly Important to temper expectations about AI (and AI-empowered robots); benefits are likely to be overestimated in the short-term More effective for AI to augment human capabilities Grewal et al. (2020a) In-store technology Retail/ services Role of social presence and convenience Grewal et al. (2020b) Cyborgs Retail/ services Perceived warmth and perceived competence mediate the impact on evaluations Guha et al. (2022) AI Consumers, generally Perceived artificiality and perceived intelligence mediate the impact on evaluations Guha et al. (2021) AI Retail Applications of AI likely to be (initially) in non–customer-facing domains More effective for AI to augment human capability Benefits of AI likely to be overestimated in the short-term Mende et al. (2019) Robots Retail/ services In some cases, deploying robots may elicit perceptions of eeriness, with downstream effects Noble et al. (2022) Technology, broadly Retail/ services Initial prioritization is non–customer-facing domains Rindfleisch et al. (2022, in this issue) Robots Retail Whiz enhances customer evaluations and also is perceived positively by employees Shankar (2018) AI Retail AI affects retailing through demand-side and supply-side pathways Van Doorn and Holthoewer (2020) Robots Retail Robots, without human assistance, may be useful for sales of embarrassing products Van Doorn et al. (2017) Robots Retail/ services Perceived warmth and perceived competence mediate the impact on evaluations Artificial intelligence and robots In their theoretical consideration of how AI might alter retailing, Guha et al. (2021) offer three main predictions and recommendations. First, early applications of AI should address non–customer-facing tasks because such deployments offer substantial value to retailers. In contrast, customer-facing interactions are relatively harder to control and more variable. Due to the technological capacities and limitations of today’s AI, such interactions increase the risk of service failures. Second, retailers and service providers should use AI to augment their employees’ capacities, rather than replacing them (Guha et al., 2021). In essence, retailers should look at balancing human employees’ and AI input, by positioning AI input as a complement to human input. With this approach, the retailers can overcome the limitations of current AI technology by leveraging human resources to intervene or smooth over potential service failures. Third, retailers need to develop realistic expectations. As suggested by Davenport et al. (2020), the effects of robots are likely to be evolutionary, not revolutionary, despite some exaggerated claims in the popular press. That is, we currently have access to artificial narrow intelligence (ANI), not artificial general intelligence (AGI) (Guha et al., 2021). Although ANI performs better than humans in domains with structured data and predictable outcomes, it is less effective in novel domains. In contrast, AGI performs better in novel domains and on complex, idiosyncratic tasks (Huang & Rust, 2018), but it is not a near-term reality (Davenport et al., 2020). According to AI researchers, the odds of achieving AGI by 2050 are 50–50 at best. (Guha et al., 2021). Thus, the benefits of robots may well be overestimated for the near term (Davenport et al., 2020). It is likely easier for retailers to start using robots (i) for non–customer-facing applications, such as cleaning, checking inventory and prices, and (ii) in ways that could augment the human capabilities of their existing workforce. Recent research on robots Recent papers have highlighted some important points as regards the use and deployment of robots in retail and service settings. Noble et al. (2022) argue that, in retail and service settings, robots and humans should collaborate closely, rather than allowing robots to replace humans. They also identify an initial prioritization on “robots running warehouses or performing constant inventory assessments” (Noble et al., 2022, p. 202), reflecting the notion that robots should initially be deployed in non–customer-facing functions. In contrast though, studies of how robots might influence satisfaction tend to focus on customer-facing domains. Both Van Doorn et al. (2017) and Grewal et al. (2020b) posit that perceived warmth and perceived competence mediate robots’ impacts on customer satisfaction. Similarly, in a study of AI-powered voice assistants (Guha et al., 2022), the findings imply that perceived artificiality and perceived intelligence mediate the impact of these devices on customers’ continued usage intentions. Deploying robots may also have negative outcomes. Mende et al. (2019) posit that robots can elicit perceptions of eeriness, with downstream effects on consumption, including inducing defensive consumption (e.g., eating more comfort food). As a broader point, Castelo et al. (2018) argue that very humanlike robots may prompt discomfort, in line with uncanny valley theory; this effect pertains to the sense of discomfort or eeriness experienced when robots are too humanoid (Grewal et al., 2020b). Such concerns seem more relevant in customer-facing domains, so again, the recommendation that retailers should focus on initially deploying AI (and robots) in non–customer-facing domains appears appropriate (Guha et al., 2021). Proposed framework How should retailers think about deploying robots? Building on contributions from Guha et al (2021) and Davenport et al. (2020), we propose a four-quadrant framework (Fig. 1) for guiding their deployment decisions. In this framework, we specify two key influences: whether the usage domain involves non– or customer-facing applications, and if retailers’ intent is to augment or replace human resources.Fig. 1 Deployment of robots in retail Usage domain Guha et al. (2021) (see also Shankar, 2018) outline two pathways by which AI is likely to affect retailing: demand-side (e.g., in-store customer experience management) or supply-side (e.g., inventory optimization). But we argue that the usage domain goes beyond such a dichotomous split, and instead reflects a continuum that varies in the extent to which robots face customers. At one end of the continuum, robots like Tally, which exclusively track inventory, are fully non–customer-facing; Whiz is somewhat customer-facing (Rindfleisch et al. 2022, in this issue); and LoweBot is substantially customer-facing, because its primary functions relate to interacting with customers. Customer-facing interactions are relatively complex and variable. Because AI currently achieves only ANI (not AGI), there is high risk of service failures, and considering the direct interaction with customers, such service failures may have dire consequences. Usage intent Both Davenport et al. (2020) and Guha et al. (2021) suggest using AI, in its ANI form, to augment rather than replace human capability, reflecting ‘human plus machines’, rather than ‘humans versus machines’. In practice though, we note that some retailers choose to deploy robots with the intent that such robots augment human capability. For example, the Café X robot has the intent of augmenting human capability, with the robot focusing on coffee preparation, and with human associates focusing on order taking, order advice, and troubleshooting if there is service failure. In contrast, other retailers choose to deploy robots with the intention of largely replacing human capability. For example, Lowebot (and to an extent Pepper) looked to replace human capability, especially for relatively simple customer service tasks. Four-quadrant framework Non–customer-facing applications, augmenting human capability This quadrant (Backstage Robot Assistant) relates to the proverbial low hanging fruit, involving immediate benefits at relatively low costs. For example, Tally checks prices and alerts employees to fix any incorrect price tags, and the K5 security robot alerts security personnel to any disturbances in parking lots. The backstage, non–customer-facing domain is relatively easier to operate in, and the goals set for the robot are relatively modest (they still require human assistance). Thus, these robots should be relatively effective in the limited tasks they are designed to perform, and retailers should embrace such assistance and deployments. Non–customer-facing applications, substituting for humans In this quadrant (Backstage Robot Worker), robots take on a more autonomous role in non-customer-facing domains. Although the non–customer-facing domain remains relatively easy to operate in, the usage intent (i.e., substituting for humans, and operating fairly autonomously) is ambitious, such that it is unclear whether current ANI robot technology can achieve this goal. Whiz operates reasonably autonomously, but even for the seemingly straightforward task of cleaning floors, it sometimes needs human support (e.g., if it gets stuck under furniture, or if it encounters stairs). To the extent possible, retailers should pursue deployments in this quadrant, but they also should be cautious and prioritize simpler tasks that can be executed by existing ANI technology. Customer-facing applications, augmenting human capability In this quadrant (Robot Coworker), retailers deploy robots to augment human capabilities in customer-facing domains, notwithstanding the challenges in customer-facing domains. For example, the Café X barista robot prepares various coffee brews, whereas human baristas engage customers, perform cleaning tasks, and refill coffee beans and other supplies. To the extent possible, retailers should pursue deployments in this quadrant, but we again call for caution, to ensure the robots are assigned tasks they can actually execute, given current ANI technology. Further, retailers should provide sufficient human support, such that a human employee can intervene to address any service failures. Customer-facing applications, substituting for humans In this final quadrant (involving Robot Associates), we examine whether retailers should deploy robots to operate somewhat autonomously, in customer-facing domains. We highlight two prominent, linked risks: (i) existing ANI technology struggles with operating autonomously in customer-facing domains, and (ii) if the robot aims to replace human employees, any service failures are difficult to recover from. LoweBot can respond to simple requests from shoppers (e.g., “Where can I find 1″ bolts?”), but if the requests are complex or phrased in unconventional ways, it might fail to understand, and thereby induce customer frustration. Thus, autonomous robot deployments demands great caution. The risk of service failure is high, due to both the limitations of current ANI technology and the lack of available human support, and in these direct interactions, such failures would have immediate and costly impacts on the firm-customer relationship. Some initial evidence indicates that customer-facing robots without human assistance may be useful in sales interactions involving embarrassing products (Van Doorn & Holthoewer, 2020). Customers arguably may perceive that robots are less likely to judge them, so when they have a need for sensitive, personal products, they may be more comfortable interacting with a robot. But, generally speaking, as the experience of Softbank with its Pepper robot indicates, it remains risky to implement potentially unreliable robot technology that can induce service failures, sans human support to address such failures. Consistent with this point, Softbank has suspended the Pepper program, while increasing its focus on Whiz. Research agenda The robot revolution in retail is just beginning, opening a wide variety of areas for research. Below, we present three areas for future research focus, relating to the interaction of robots and (i) customers, (ii) retail employees, and (iii) retailers; in addition, we also look at public interest issues. Robots and customers We briefly discuss four factors relevant to how customers interact with and react to robots. First, understanding how robots influence customers’ evaluations of the retail setting is critical. Rindfleisch et al. (2022, in this issue) posit that customers react positively to Whiz, even though Whiz may not interact with them directly. Any future proposed frameworks must be flexible enough to encompass customers’ evaluations relating to robot deployments across both customer-facing and non–customer-facing roles. Second, several concerns arise when robots interact with customers. Interactions with a robot may allow the retailer to capture substantial, personal information, e.g., through sensors, which raises ethics and privacy concerns. Should such information even be collected? How may it be used? Furthermore, what are the distinct responsibilities of retailers, customers, and policy makers regarding how such private information is protected, after its collection? Third, customers might react to and evaluate robots differently, depending on their features. For example, robot evaluations could be contingent on perceived artificiality and perceived intelligence (Guha et al., 2022), and certain features of the robots themselves (e.g., those that evoke anthropomorphism) might have stronger or weaker relative effects on perceived artificiality versus perceived intelligence. Consumers’ anthropomorphism could enhance their evaluations, but at very high levels, it could lead to negative impacts (Castelo et al., 2018; Mende et al., 2019). Fourth, customers might not respond positively (or neutrally) to robots, as demonstrated by real-world cases where customers and passers-by have caused intentional harm to robots (Wilson, 2017). How should retailers think about such cases, and what protections–if any–should they put in place to limit such damage to the robots they deploy? Should robots have some elements of self-defense capability? Robots and employees There is limited research into how retail employees perceive–and react to–robots. We note that Rindfleisch et al. (2022, in this issue) find that employees perceive positively, and react positively, to Whiz. What drives employees’ perceptions of and attitudes toward robots? What downstream outcomes, such as retail store evaluations, stem from such perceptions and attitudes? Can positive employee attitudes towards robots, both non-customer-facing robots and customer-facing robots, influence downstream outcomes, like retail store evaluations, and stock market evaluations? Robots and retailers Regarding the retailer (at an organizational level), we suggest four areas for research. First, Guha et al. (2021) investigate which factors drive retailers’ adoption of AI; similarly, it may be worthwhile to examine which factors drive retailers’ adoption of robots. We have identified some likely influences (see Fig. 1), but other factors also could be important. Second, research should specify which types of retailers can benefit most from using robots, as well as whether robot adoption paths or deployment strategies might be contingent on the retailers’ type. Third, robots can take many (physical) forms, such as humanoid forms (e.g., LoweBots), functional forms (e.g., K5, Peanut), or relatively immobile designs (e.g., Café X). Research can determine which factors define the optimal physical form for a robot, contingent on retailers’ type. Fourth, retailers are deploying a host of in-store technologies (Grewal et al., 2020a), so it will be important to study how robots can co-exist and enhance the deployment of those other technologies. Robots and the public interest Retailers may use robots in variety of ways. In this paper, we have described how retailers may deploy robots in the store, in both non-customer-facing domains and customer-facing domains. In addition, retailers may deploy robots outside the store, e.g., Domino’s plans to use an autonomous vehicle to deliver pizza to customers (Benveniste, 2021). All of this raises a variety of public policy questions (i) how will robot deployment impact other retail stakeholders e.g., retail employees, delivery companies like Door Dash and Uber Eats, (ii) what public infrastructure may be needed if robots are deployed outside the store e.g., would robots be able to use bicycle lanes (robots cannot co-exist with cars on main roads, as cars are both heavier and move faster), (iii) would the public be accepting of robots, both in-store and outside-of-the-store, noting that robots may be coming close to customers and thence capturing personal customer information as they move around in-store or outside the store while executing their tasks, and (iv) noting that customers sometimes exhibit violent tendencies towards robots (Wilson, 2017), how acceptable would it be, if the robots were allowed some self-defense mechanism (even a relatively non-violent self-defense mechanism, e.g., a high-pitched alarm) towards customers who may wish it harm? Conclusion To reflect on how robots are likely to inform the future of retailing, we propose a framework that gives retailers suggestions for how they should deploy robots. We also offer a three-part research agenda, related to the implications of robots for customers, employees, and retailers. Accordingly, we hope this article provides insights for retailers, researchers, and public policy experts. Although deploying robots promises substantial benefits for retailers, we end on two notes of caution. First, already pressing ethics and privacy concerns are likely to intensify even further, as privacy laws and regulations take full effect. Second, it is important for retailers to develop realistic expectations. According, to previous theorizing, AI (and robots) can provide evolutionary benefits in the immediate term but provide revolutionary benefits in the long-run. We echo this point and anticipate that the benefits of robots might be overestimated in the short term, whereas the cautionary tale of various robots (like Pepper) highlights the need for realism. Still, the long-run potential benefits of robots offer great promise. Declarations Conflict of Interest The authors declare that they have no conflict of interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Benveniste, A. (2021). Domino’s is launching a pizza delivery robot car. Retrieved September 11, 2022, from https://www.cnn.com/2021/04/12/tech/dominos-pizza-delivery-robot/index.html Castelo, N., Schmitt, B., & Sarvary, M. (2018). Human or robot? The uncanny valley in consumer robots. In: Advances in Consumer Research, Volume 46, eds. Andrew Gershoff, Robert Kozinets, and Tiffany White, Duluth, MN: Association for Consumer Research, 183–187. Davenport T Guha A Grewal D Bressgott T How artificial intelligence will change the future of marketing Journal of the Academy of Marketing Science 2020 48 1 24 42 10.1007/s11747-019-00696-0 Ferguson, C. (2021). Meet Tally, the robot helping eliminate empty shelves at Meijer. Retrieved September 3, 2022, from https://www.woodtv.com/news/kent-county/meet-tally-the-robot-helping-eliminate-empty-shelves-at-meijer/#:~:text=The%20autonomous%20robot%20now%20roams,scan%20RFID%20tags%20on%20products Forgan, B. (2020). What robots can do for retail. Harvard Business Review, 01 Oct. Grewal D Kroschke M Mende M Roggeveen AL Scott ML Frontline cyborgs at your service: How human enhancement technologies affect customer experiences in retail, sales, and service settings Journal of Interactive Marketing 2020 51 9 25 10.1016/j.intmar.2020.03.001 Grewal D Noble SD Roggeveen AL Nordfält J The future of in-store technology Journal of the Academy of Marketing Science 2020 48 1 96 113 10.1007/s11747-019-00697-z Guha, A., Bressgott, T., Grewal, D., Mahr, D., Wetzels, M., & Schweiger, E. (2022). How artificiality and intelligence affect voice assistant evaluations. Journal of the Academy of Marketing Science. Forthcoming. Guha A Grewal D Kopalle PK Haenlein M Schneider MJ Jung H Hawkins G How artificial intelligence will affect the future of retailing Journal of Retailing 2021 97 1 28 41 10.1016/j.jretai.2021.01.005 Hochman, D. (2018). This $25,000 robotic arm wants to put your Starbucks barista out of business. Retrieved September 8, 2022, from https://www.cnbc.com/2018/05/08/this-25000-robot-wants-to-put-your-starbucks-barista-out-of-business.html Huang M-H Rust RT Artificial Intelligence in Service Journal of Service Research 2018 21 2 155 172 10.1177/1094670517752459 Mende M Scott ML van Doorn J Grewal D Shanks I Service robots rising: How humanoid robots influence service experiences and elicit compensatory consumer responses Journal of Marketing Research 2019 56 4 535 556 10.1177/0022243718822827 Mims, C. (2018). How robots and drones will change retail forever. The Wall Street Journal, 15 Oct. Morgan, B. (2020). The 3 best in-store robots and why they work. Retrieved September 3, 2022, from https://www.forbes.com/sites/blakemorgan/2020/05/13/the-3-best-in-store-robots-and-why-they-work/?sh=3f3297f137b2 Noble SM Mende M Grewal D Parasuraman A The Fifth Industrial Revolution: How harmonious human–machine collaboration is triggering a retail and service [r]evolution Journal of Retailing 2022 98 199 208 10.1016/j.jretai.2022.04.003 Nassauer, S., & Cutter , C. (2019). Walmart is rolling out the robots. The Wall Street Journal, 9 Apr. Nussey, S. (2021). Softbank shrinks robotics business, stops Pepper production. Retrieved September 3, 2022, from https://www.reuters.com/technology/exclusive-softbank-shrinks-robotics-business-stops-pepper-production-sources-2021-06-28/ Rindfleisch, A., Fukawa, N., & Onzo, N. (2022). Robots in retail: Rolling out the Whiz. AMS Review. 10.1007/s13162-022-00240-4 Robinson, M. (2017). The tech giants of Silicon Valley are starting to rely on crime-fighting robots for security. Retrieved September 8, 2022, from https://finance.yahoo.com/news/tech-giants-silicon-valley-starting-150800457.html Shankar V How artificial intelligence (AI) is reshaping retailing Journal of Retailing 2018 94 4 vi xi 10.1016/S0022-4359(18)30076-9 Simon, M. (2021). Peanut the waiter robot is proof that your job is safe. Retrieved September 3, 2022, from https://www.wired.com/story/peanut-the-waiter-robot-is-proof-that-your-job-is-safe/ Softbank. (2022a). Hilton Hotel case study. Retrieved September 3, 2022, from https://us.softbankrobotics.com/case-study/hilton-garden-inn Softbank. (2022b). For better business just add Pepper. Retrieved September 8, 2022, from http://us.softbankrobotics.com/pepper Van Doorn, J. & Holthoewer, J. (2020). This robot doesn’t judge me – service robots and the choice of embarrassing products. In Advances in Consumer Research Volume 48, eds. Jennifer Argo, Tina M. Lowrey, and Hope Jensen Schau, Duluth, MN: Association for Consumer Research, 1052–1057. Van Doorn J Mende M Noble SM Hulland J Ostrom AL Grewal D Petersen JA Domo arigato Mr. Roboto: Emergence of automated social presence in organizational frontlines and customers’ service experiences Journal of Service Research 2017 20 1 43 58 10.1177/1094670516679272 Wilson, T. (2017). Drunk man knocked down 300-pound robot in Mountain View. Retrieved September 3, 2022, from https://abc7news.com/police-say-drunk-man-knocked-down-security-robot-in-mountain-view-k-5-knightscope-robots-named-k5/1915713/	0	PMC9716531	NO-CC CODE	2022-12-03 23:20:58	no	AMS Rev. 2022 Dec 2;:1-8	utf-8	null	null	null	oa_other
==== Front J Real Estate Finan Econ The Journal of Real Estate Finance and Economics 0895-5638 1573-045X Springer US New York 9933 10.1007/s11146-022-09933-w Article Local Housing Market Sentiments and Returns: Evidence from China Shen Shulin [email protected] 1 Zhao Yiyi [email protected] 1 http://orcid.org/0000-0002-2022-2263 Pang Jindong [email protected] 2 1 grid.33199.31 0000 0004 0368 7223 School of Economics, Huazhong University of Science and Technology, 1037 Luoyu Road, Wuhan, 430074 Hubei China 2 grid.49470.3e 0000 0001 2331 6153 Economics and Management School, Wuhan University, Hubei 430072 Wuhan, China 2 12 2022 135 22 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This paper examines the impacts of local housing sentiments on the housing price dynamics of China. With a massive second-hand transaction dataset, we construct monthly local housing sentiment indices for 18 major cities in China from January 2016 to October 2020. We create three sentiment proxies representing the local housing market liquidity and speculative behaviors from the transaction dataset and then use partial least squares (PLS) to extract a recursive look-ahead-bias-free local housing sentiment index for each city considered. The local housing sentiments are shown to have robust predictive powers for future housing returns with a salient short-run underreaction and long-run overreaction pattern. Further analysis shows that local housing sentiment impacts are asymmetric, and housing returns in cities with relatively inelastic housing supply are more sensitive to local housing sentiments. We also document a significant feedback effect between housing returns and market sentiments, indicating the existence of a pricing-sentiment spiral which could potentially enhance the ongoing market fever of Chinese housing markets. The main estimation results are robust to alternative sentiment extraction methods and alternative sentiment proxies, and consistent for the sample period before COVID-19. Keywords Housing market Sentiment Market liquidity Speculation PLS JEL Classification P25 G41 R30 R31 http://dx.doi.org/10.13039/501100001809 National Natural Science Foundation of China 71903145 71903060 Shen Shulin Pang Jindong ==== Body pmcIntroduction In the past several decades, housing prices have skyrocketed in major Chinese cities. The annual growth rate of real Chinese national housing prices could be as high as 16.7% from 2006 to 2010 according to Wu et al. (2014), and the annual growth rate of housing prices in first-tier cities of China was as high as 13% over 2003–2013 according to Fang et al. (2016). While some scholars (e.g., Chow & Niu, 2015; Tan et al., 2020; Wang & Zhang, 2014; Wu et al., 2016) have documented how these housing price appreciations are driven by fundamental factors of demand and supply, still a large part of these price appreciations remained unexplained. Advances in behavioral finance have offered one potential explanation for this housing market fever in China, i.e., investor sentiment (or market sentiment). Originating from “animal spirits” by Pigou (1927) and Keynes (1936), Shiller (2000) proposes that the “irrational exuberance” of investors renders them to rely on many psychological factors for asset valuations. These early studies have laid the foundations for research on investor sentiment in the stock and many other financial markets (e.g., Baker & Wurgler, 2006, 2007). In behavioral finance models, investor sentiment, independent of market fundamentals, is believed to play a role in price determination and could induce some investors to possess systematic biases in their beliefs. Many studies have shown how market sentiments greatly affect stock returns (Baker & Wurgler, 2006, 2007; Da et al., 2015; Huang et al., 2015). In viewing the important role of market sentiment in real estate markets, there is a growing literature examining the market sentiment on housing returns and dynamics, with the majority of the study focusing on commercial real estate markets (including REIT markets) in developed markets. Examples of these studies include Clayton and MacKinnon (2002, 2000), Gallimore and Gray (2002), Clayton et al. (2009), Lin et al. (2009), Ling et al. (2014), Jin et al. (2014), Das et al. (2015), Letdin et al. (2021) and so on. All these studies on commercial real estate markets and REIT markets highlight the significant role of market sentiments in the valuations of real estate properties. However, empirical examinations of market sentiment on Chinese residential housing markets are limited. More formally, housing market sentiment could be defined as a misguided belief about housing price appreciations, which cannot be justified by the current economic information set available to housing market participants, as in Ling et al. (2015). As emphasized in Hui and Wang (2014) and Hui et al. (2017), the private (versus commercial) housing sector stands out as an ideal “victim” of investor sentiment for several reasons. First, market participants in private housing markets are mostly individuals and households with limited knowledge and information and hence more susceptible to sentiment. Second, the illiquidity of housing properties leads to high segmentation of the market and asymmetry of information. Among all housing sectors, the private housing market has the highest liquidity and makes it resemble the stock market. Lastly, short-selling restrictions render mispricing in the private housing market hard to eliminate. As the real estate market is the mainstay of this emerging economy and housing composes a large part of household asset holdings, examining the impacts of investor sentiments on the private housing market is imperative, as emphasized in Case and Shiller (2003), Shiller (2010) and Case et al. (2012). In this paper, we try to fill this gap by applying sentiment analysis techniques to housing markets in China, intending to examine how housing sentiments affect local housing markets in this emerging economy. To achieve this goal, we first construct city-specific housing sentiment indices for 18 major Chinese cities and then conduct a series of empirical analyses to examine the sentiment impacts on local housing returns. More specifically, we construct three sentiment proxies representing the local housing market liquidity and speculative behaviors from a massive second-hand transaction dataset and then use the recursive look-ahead-bias-free implementation of the partial least squares (PLS) method to extract a local housing sentiment index from these three proxies for each city considered. A panel regression of housing return forecasting is then examined with local housing sentiments as an explanatory variable. To investigate the persistence of sentiment impacts, we also estimate impulse responses of cumulative housing returns to sentiment shocks over different return horizons. For sentiment construction, the first two proxies are the median time on the market of sold housing units and the turnover rate of listed housing units in a given month, serving as two housing market liquidity measures. These two housing market liquidity measures resemble the turnover rate of the stock market in Baker and Wurgler (2006). For the stock market, Baker and Stein (2004), Baker and Wurgler (2006), and Baker and Wurgler (2007) suggest that turnover, or more generally liquidity can serve as a sentiment index. For the real estate market, Clayton et al. (2008) and Ling et al. (2015) point out that increased liquidity is the amplifying channel of a pricing-sentiment spiral in the housing market. The third sentiment proxy is constructed as a small-house return premium over large houses. This small-house return premium measures the speculative investment in Chinese housing markets, as speculative investors in China tend to over-invest in smaller housing units because of less capital requirement and higher liquidity, and hence smaller housing units appreciate faster in cities with more speculative investors. With these three local housing sentiment proxies, we first orthogonalize each of these proxies against a set of fundamental variables and then implement the recursive look-ahead-bias-free estimation of the PLS method by Kelly and Pruitt (2013, 2015) and Huang et al. (2015) to construct the local housing market sentiment for each city in our sample. Compared with traditional dimension reduction methods such as principal component analysis (PCA), PLS filters out irrelevant common components and summarizes the most useful information in these proxies to construct a single complex sentiment index with the largest covariance with the targeted series of predictions. To remove the potential look-ahead bias of the in-sample estimation of PLS, we recursively conduct the look-ahead-bias-free implementation of PLS with an expanding window scheme to obtain a look-ahead-bias-free estimate of the housing sentiment sequence. The panel housing return forecasting regressions show that local housing sentiment indices have robust predictive powers for future housing market returns, while the impulse response estimations indicate a salient pattern of short-run underreaction and long-run overreaction. Further analysis shows that local housing sentiment impacts are asymmetric in that local housing sentiments only have significant impacts on housing returns when the market sentiment is below the sample average. In addition, the sentiment effects are much stronger and more significant for cities with a relatively inelastic housing supply. Housing sentiments exhibit strong inertia and are positively correlated with past short-term cumulative housing returns, implying that housing sentiments have a backward-looking property and there exists a significant feedback effect between housing returns and sentiments. The documented sentiment impacts are comparable to what has been documented for developed housing markets. As shown in the results section, a one-standard-deviation increase in local housing sentiments will cause the annual housing return to increase by about 1.68% in China, while this effect is approximately 5.58% for the U.S. housing market in Soo (2018) and approximately 3.73% in Bork et al. (2020).1 These results indicate that the housing markets in China are also susceptible to irrational sentiments, with arbitrage and speculation prevalent in this emerging market. To prove that the recursive look-ahead-bias-free implementation of the PLS method does not lead to “false detection of predictability”, we conduct a placebo test in which three randomly-generated sentiment proxies are drawn from the standard normal distribution and then summarized into a “placebo” sentiment index by the recursive look-ahead-bias-free procedure. We find no significant predictive power of this “placebo” sentiment on future housing returns. The placebo test documents that the significant predictability of our extracted sentiment index is not generated mechanically by the recursive look-ahead-bias-free implementation of the PLS method. Our main regression results are also robust to alternative sentiment construction methods, such as the scaled PCA method by Huang et al. (2022) and the traditional PCA method, and are consistent for the sample period before COVID-19. Also, we find consistently significant predictability power of an alternative sentiment index based on the number of newly listed properties (instead of the turnover rate as a proxy) on future housing returns. Housing market sentiment is particularly difficult to measure compared with other financial markets for several reasons. First, as stated in Soo (2018), typical sentiment proxies for the stock market, such as closed-end fund discounts, mutual fund flows, and dividend premiums, do not have straightforward counterparts for the housing market. Second, housing markets are highly segregated geographically. The housing market conditions for different cities vary substantially even within a given region during the past housing cycle (Ferreira & Gyourko, 2012). Third, transactions in the housing market usually take several months or even years to complete. Hence, the impact of sentiments on housing market outcomes may take a long time to unfold. In this paper, we choose to follow the studies on market-based (indirect) stock market sentiments to construct housing market sentiments (Baker & Wurgler, 2006; Da et al., 2015; Huang et al., 2015). The two housing market liquidity measures of our sentiment indices mimic the turnover rate of the stock market. The speculation measure of local housing markets could serve as an alternative to first-day returns on IPOs, which measures investor enthusiasm for the stock market. More importantly, by the richness of the transaction dataset, we can construct a sentiment index for each city considered, capable of reflecting heterogeneous local housing market conditions. And the impulse response analysis of cumulative housing returns to sentiment shocks over different return horizons enables us to examine how the sentiment impacts unfold with time. Compared with the textual approach in Soo (2018), which develops housing sentiments for 34 cities across the United States by quantifying the qualitative tone of local housing newspaper articles, our market-based proxies may be a delayed reflection of market sentiments. However, these transaction-based proxies provide a good complementary to the textual sentiments when representative local housing newspapers are not available. Compared with the survey method to quantify market sentiments, such as those in Bork et al. (2020) and Ling et al. (2015) which use household survey responses to questions about buying conditions for houses in the U.S., our transaction-based sentiment index is an indirect measure of market sentiment.2 But given the fact that surveys regarding Chinese customers are limited, our transaction-based sentiment index is a good alternative to those survey-based measures. Moreover, even though internet usage is common nowadays, an aggregate search index, such as the mortgage default risk based on the Google search index of Chauvet et al. (2016) for the U.S. market, cannot provide localized sentiment measures given the heterogeneous properties of local housing markets. For Chinese real estate markets, Zheng et al. (2016) use Google search counts to construct city-specific confidence indices for 35 Chinese cities from 2005 to 2013. The confidence index constructed by Zheng et al. (2016) is calculated as the ratio of the count of positive entries to the total count of both positive and negative entries regarding the targeted real estate market. For the sample period considered in this paper (2016 to 2020) and the sample period of Zheng et al. (2016), Google search has been blocked and hence Google search entries may not work as a good indicator of confidence index or investor sentiments. The sentiment indices constructed in this paper are based on market transaction outcomes, which are the results of demand and supply factors,3 and could serve as a complement to these methods listed above, especially the city-specific confidence index by Zheng et al. (2016). The contribution of this paper lies in three parts. First, we contribute to a growing literature evaluating the effect of market sentiments on housing markets by analyzing a representative sample of cities in China. Among existing studies on housing sentiments regarding developing markets (Hui & Wang, 2014; Hui et al., 2017, 2018; Lam & Hui, 2018; Soo, 2018; Zhou, 2018), most of them construct a sentiment index for a specific city and examine the relationship between sentiments and housing returns of the city such as Shanghai or Hong Kong, which lack a representative dataset. Aided by massive transaction data, we construct city-specific housing sentiment indices for 18 major cities in China, which are more representative of Chinese housing markets. This panel data structure of our sentiment indices provides more reliable and comprehensive empirical findings on the relationship between local housing sentiments and returns. This paper confirms that investor sentiment is an important factor in Chinese housing markets, suggesting arbitrage and speculation are prominent in this emerging market. Second, we provide more evidence on the role of local housing supply conditions in analyzing housing market sentiment impacts. Existing studies have shown that supply-side conditions greatly affect housing prices and constraints on housing supply can explain large differences in housing price dynamics in different regions (Glaeser & Ward, 2009; Glaeser et al., 2006, 2008; Green et al., 2005; Saiz, 2010). The role of housing supply may be more significant in Chinese housing markets as indicated by Liu (2014) and Wang et al. (2012). However, supply-side conditions have been overlooked in the empirical examinations of housing sentiment impacts on local housing returns (except Zheng et al., 2016). In this paper, we take advantage of supply elasticity estimates from Liu et al. (2019) to study whether local supply conditions alter market sentiment impacts. Our result shows that housing markets in cities with relatively inelastic housing supply are more sensitive to local housing sentiments than places with elastic housing supply. For cities with a much more elastic housing supply, the impacts of market sentiments on returns are much smaller or even negligible. The finding that local housing supply elasticities play an important role in the sentiment effects highlights the necessity of considering local market conditions in the analysis of sentiment analysis. Third, we also contribute to the literature on China’s high housing price puzzle. There exist many theoretical as well as empirical works documenting and explaining why the housing price level or growth rate is so high in China (Chen & Wen, 2017; Fang et al., 2016; Glaeser et al., 2017; Wei et al., 2012; Wu et al., 2016). Many of these works debate the existence of a bubble in Chinese housing markets by analyzing the fundamentals underlying this emerging market. The findings in this paper provide some evidence on this issue from the aspect of the nonfundamental market sentiments. Our local housing sentiment indices are shown to impact local housing returns significantly in addition to many fundamental controls, indicating nonfundamental sentiments also add up to the housing market fever of China. The short-run underreaction and long-run overreaction pattern of housing sentiments imply that there exist irrational investors in the Chinese housing market, exaggerating the turmoil of local housing markets in the short run. We also document that the expectations of Chinese real estate investors are backward-looking and there exists a significant feedback effect between housing returns and market sentiments. This pricing-sentiment spiral could enhance the ongoing market fever of Chinese housing markets and potentially extend the length and magnitude of housing market cycles in China. In brief, our empirical analysis enriches the literature on the role of sentiments in the housing market and provides better insights into the housing market dynamics, which may help policymakers to stabilize and improve the functioning of the housing markets. The rest of this paper is organized as follows. In "Data and Housing Price Index" section, we introduce our transaction data and construct second-hand housing price indices. Descriptions of sentiment proxies and the construction of city-level housing sentiment indices are discussed in "Housing Sentiment Index" section. "Local Housing Sentiments and Returns" section studies the relationship between housing market sentiments and returns. "Robustness Checks" section provides some robustness checks on pre-COVID-19 estimation, alternative sentiment construction methods, alternative sentiment proxies, and the placebo test. "Conclusions" section concludes. Data and Housing Price Index Transaction Data In this paper, we use a massive second-hand housing transaction dataset to construct the housing price index and the sentiment index.4 The private (residential) transaction records are retrieved from Lianjia.com, one of the most popular second-hand housing trading platforms in China. As a real estate brokerage firm, Lianjia helps housing sellers to list their properties online and provides brokerage services for potential sellers and buyers. Once a transaction (sale) has been concluded, Lianjia records the sale and posts its information in the Transacted-Units section on its website. Each transaction record includes detailed information on the characteristics of the sold housing unit: listing date, listing price, transaction date, transaction price, and project name (xiao qu), as well as unit attributes, such as floor area, layout, decoration, orientation, etc. We collect all the transaction records from this Transacted-Units section on Lianjia.com for each city in our sample. In total, we obtain 1,799,272 transaction records from January 2015 to October 2020 covering 18 major cities in China. Table 1 lists all the cities and their corresponding sample periods. The earliest date of the transaction data is January 2015 for Shanghai and Tianjin. In the following, we first use the transaction data to construct the Hedonic Housing Price Index (HPI) and then derive local housing sentiment using the recursive look-ahead-bias-free implementation of the PLS method for each city. Since 2016, a large-scale purchase restriction policy and a series of tightening policies were introduced to fight against speculative housing investment and rapid housing price growth.5 Hence, our analysis is under a general tightening policy environment.Table 1 Transaction data and sentiment sample periods Num City Transaction Data Look-ahead-bias-free Sentiment 1 Shanghai 2015/01 ~ 2020/10 2016/01 ~ 2020/10 2 Beijing 2016/07 ~ 2020/10 2017/07 ~ 2020/10 3 Nanjing 2015/10 ~ 2020/10 2016/10 ~ 2020/10 4 Xiamen 2015/11 ~ 2020/10 2016/11 ~ 2020/10 5 Hefei 2017/01 ~ 2020/10 2018/01 ~ 2020/10 6 Dalian 2015/08 ~ 2020/10 2016/08 ~ 2020/10 7 Tianjin 2015/01 ~ 2020/10 2016/01 ~ 2020/10 8 Guangzhou 2016/02 ~ 2020/09 2017/02 ~ 2020/09 9 Chengdu 2015/05 ~ 2020/10 2016/05 ~ 2020/10 10 Hangzhou 2016/03 ~ 2020/10 2017/03 ~ 2020/10 11 Wuhan 2016/02 ~ 2020/10 2017/02 ~ 2020/10 12 Shenyang 2017/01 ~ 2020/09 2018/01 ~ 2020/09 13 Jinan 2015/12 ~ 2020/10 2016/12 ~ 2020/10 14 Shenzhen 2015/09 ~ 2020/10 2016/09 ~ 2020/10 15 Suzhou 2015/08 ~ 2020/10 2016/08 ~ 2020/10 16 Chongqing 2017/01 ~ 2020/07 2018/01 ~ 2020/07 17 Changsha 2016/03 ~ 2020/10 2017/03 ~ 2020/10 18 Qingdao 2016/11 ~ 2020/10 2017/11 ~ 2020/10 Sample periods are chosen by their availability. As discussed below, we use the first twelve months' transaction data to get the first sentiment estimate. Hence, the sample periods for the look-ahead-bias-free sentiment start one year later than their transaction data Housing Price Index Based on a Hedonic Model The first step of our empirical study is to construct a monthly quality-adjusted housing price index (HPI) for each city in our sample. We use the hedonic regression to build the HPI. The Hedonic model is introduced into the housing markets by Rosen (1974). According to the theory, housing prices contain a part that is driven by some attributes of the house itself (area, layout, orientation, and so on). Following this idea, if these properties were decomposed from the price and then the remaining part is purely driven by supply and demand factors. Accordingly, our HPI construction is based on the following equation:1 lnPilt=α+∑t=2TβtDt+γXilt+πl+εilt, where Pilt represents the trading price per square meter for housing unit i in the project (xiao qu) l on transaction date t. Xilt denotes a vector of unit attributes, including floor area, elevator, number of bedrooms, floor level, orientation, and decoration type (fancy, simple, or other). We also control for the project-fixed effect πl. Note that a project in Chinese cities is a very small geographic unit, similar to a Census block in the U.S. The project-fixed effects control for neighborhood amenities, including the school district, crime rate, traffic service, distance to CBD, and so on. The error term is denoted by εilt. A series of month dummies, Dt, are included in the regression to capture the time variation in the housing prices clear of quality changes. The value of HPI for month t is then calculated as expβt×100 and the value is set as 100 for the origin month. We apply the hedonic regression for each city in our sample and obtain 18 city-specific monthly HPI series from January 2015 to October 2020. It is worth noting that the final HPI series is an unbalanced panel with the earliest date in January 2015 for Shanghai and Tianjin. Housing Sentiment Index In this section, we use second-hand transaction data to construct our housing sentiment indices. We first introduce our sentiment proxies and then illustrate the recursive look-ahead-bias-free procedure of the PLS method from which the housing sentiment index is constructed. Finally, we display some correlation analysis between our sentiment indices and several confidence indices from official sources. Sentiment Proxies In the construction of sentiment proxies, we majorly follow the approach of Zhou (2018). We construct three sentiment proxies based on our trade-by-trade data. The first proxy is MedianIntv, the natural logarithm of the median holding period of sellers, i.e., the median number of days on the market of sold housing units in a given month. Intuitively, a lower level of MedianIntv would indicate a relatively higher market sentiment. The second proxy is Turnover, the ratio of housing areas being sold to the total housing areas for sale in a given month. To construct this proxy, for each month, we divide the total floor areas transacted in this month by the floor areas of listed properties including transacted houses and available-for-sale houses (which have been listed but not yet sold until the current month). Note that this Turnover ratio is a quasi-turnover ratio of the housing market. The real turnover rate of the housing market is the ratio of trading volume to the total housing stock as in Ling et al. (2014). Usually, the real turnover rate of the housing market is small, since most housing units are not for sale. By definition, this Turnover proxy measures the liquidity of listed properties in the housing market and usually, it is much larger than the real turnover ratio of the housing market. Essentially, constructed in the same spirit as measures in Clayton et al. (2008), Ling et al. (2014), and Ling et al. (2015), MedianIntv and Turnover measure the liquidity of real estate markets. Unlike MedianIntv, Turnover should be positively correlated with the housing market liquidity since a liquid housing market should be associated with a shorter time on the market and a high turnover rate. As the correlation matrix in Table 9 of Appendix A shows, Turnover is significantly negatively correlated with MedianIntv. Meanwhile, Turnover exhibits a positive correlation with the contemporaneous housing return Rt while MedianIntv exhibits a negative correlation with Rt. These correlation coefficients provide some support for the positive (negative) correlation between Turnover (MedianIntv) and market liquidity. The third proxy is SMB, the small-house return premium over large houses. Each year we calculate the quintile breakpoints of the transacted houses’ floor areas and then divide houses into five groups according to the latest breakpoints. Small (big) houses are those in the first (fifth) quintile. Then we use the Hedonic method in "Housing Price Index Based on a Hedonic Model" section to construct a housing price index for small houses and big houses, respectively. In the end, SMB is equal to the small-house return minus the big-house return (in percentage). Zhou (2018) states that over-optimistic investors tend to buy big houses and push up prices for these large houses. Hence, the proxy SMB, which measures the return premium of smaller houses, is expected to be negatively correlated with sentiments according to Zhou (2018). However, in contrast to this argument, speculative investors in the Chinese housing market tend to over-invest in smaller houses for several reasons. First of all, the financial pressure of speculating in small houses is relatively low because of lower total costs and hence less capital requirement. Secondly, the risk in investing in small houses is much lower as a flipper of small houses could more easily find a buyer in the market because of the higher liquidity and the dominant demand for small houses. Thirdly, the potential return on investing in small houses is higher since the price appreciation tends to be higher for smaller houses. For the Singapore housing market, Fu and Qian (2014) also find that short-term speculators typically target smaller units as they require less capital and are easier to sell. Furthermore, some online news reports also reveal that small housing units, the primary target of speculators, are always sold out even during periods of a market downturn.6 Some speculators were personally interviewed and admitted that “they invest in small houses because of the high return-on-investment ratio”.7 In short, from the point of view of speculating behavior, the prices for smaller housing units appreciate faster in cities with more speculative investors, and hence SMB should be positively correlated with housing sentiments. We will examine the correlation between SMB and housing sentiment in the next section. There are two more proxies used by Zhou (2018), namely, the newly-opened housing construction area, and the housing sector investment to construct the housing sentiment for Shanghai. In this paper, we choose to construct the housing sentiment using MedianIntv, Turnover, and SMB only for three reasons. Firstly, since we are constructing sentiments for 18 Chinese cities, data for the other two proxies are not available for some cities in our sample. Secondly, this paper tries to construct a market outcome-based sentiment. Newly-opened housing construction and housing investment reflect sentiments of the supply side, which has been reflected in market outcomes to some extent. So, omitting these two proxies in the construction of the sentiment will not result in big information loss. Finally, housing investment and construction usually take several years to accomplish. So, sentiments from the supply side may not be synchronous with these based on market outcomes. Hence, omitting the housing investment and construction variables and relying on the other three transaction-based proxies will leave us with a more synchronous sentiment measure. At the end of this section, we display some summary statistics for these three proxies. To save space, Table 2 only shows the average values of these three proxies for Beijing, Shanghai, Guangzhou, and Shenzhen, respectively. For example, the average median days on the market of second-hand properties in Beijing is 23.54 days (exp(3.159)) in 2016, while this value increases to 70.39 days (exp(4.25)) in 2020. Also, the turnover rate of listed second-hand properties has decreased from 46 to 29% during the same period for Beijing. The average small-house return premium is about 0.24% for Beijing in 2016, decreasing to a negative value of -0.17% in 2020.Table 2 Housing sentiment proxies from 2016 to 2020 City Year MedianIntv Turnover SMB Beijing 2016 3.159 0.462 0.235 2017 3.511 0.306 -0.283 2018 3.809 0.252 0.239 2019 4.157 0.212 0.211 2020 4.254 0.296 -0.172 Shanghai 2016 6.108 0.102 1.320 2017 6.679 0.117 -0.178 2018 4.451 0.171 -0.501 2019 4.136 0.220 0.074 2020 4.055 0.313 0.055 Guangzhou 2016 3.836 0.195 0.698 2017 3.776 0.317 0.402 2018 3.867 0.193 0.561 2019 4.332 0.142 -0.323 2020 4.530 0.273 -0.941 Shenzhen 2016 3.591 0.270 0.460 2017 4.006 0.235 0.072 2018 4.077 0.190 0.254 2019 4.274 0.195 0.302 2020 3.940 0.353 0.146 Values are averaged by year for each city presented. MedianIntv is the natural logarithm of the median number of days on the market. Turnover equals the total area of transacted houses divided by the total area of houses available for sale in a given month, and SMB equals the small-house return minus the big-house return (in percentage) Housing Sentiment Index With these above three proxies, we then construct sentiment indices for the 18 cities in our sample. First, all these raw proxies are standardized with zero means and unit standard deviations. Second, to eliminate the impact of the business cycle, we regress the standardized proxies on some macroeconomic variables and obtain the residuals. Following Zhou (2018), we choose four macroeconomic variables including Purchasing Managers’ Index (PMI), the growth rate of consumer price index (CPI), the growth rate of M2 (M2G), and the difference between the yield of AA-grade corporate bonds and AAA-grade corporate bonds (Default). The residuals from these regressions contain sentiment information that is orthogonal to business cycle factors. Third, to smooth out jumps, we impose a three-month moving average to the residuals, as in Huang et al. (2015). The smoothed residuals are called “clean” proxies. Fourth, considering that the clean proxies may have a lag in reflecting the sentiment of market entities, we need to choose between the current value and the lagged value of each proxy, as in Baker and Wurgler (2006). In detail, we first perform principal component analysis on MedianIntvt, MedianIntvt-1, Turnovert, Turnovert-1, SMBt, and SMBt-1, to extract its first principal component. Then the correlation between the first component and each of the six variables is calculated to determine whether to choose the current value or the lagged value. Finally, the value which is more correlated with the first component is chosen as the final sentiment proxy. Fifth, we conduct the recursive look-ahead-bias-free implementation of the PLS approach of Kelly and Pruitt (2013, 2015) and Huang et al. (2015) to construct a look-ahead-bias-free housing sentiment index for each city in our sample. Specifically, at month t, for each sentiment proxy proxyi(MedianIntv, Turnover, or SMB), we run the following first-stage regression:2 proxyi,s-1=πi,0+πi,1Returns+ui,s-1,s≤t, in which proxyi,s-1 denotes the lagged sentiment proxy (proxy i at time s-1) and Returns denotes housing return (in percentage) at time s. To obtain the loadings for month t, we only use data up to month t in the above regression. The latest return (Returnt) used on the right-hand side is from time t-1 to t. The latest sentiment proxies used on the left-hand side is proxyi,t-1. Thus, the first-stage coefficient estimates π^i1 are in the time t information set since they use monthly returns {Return2,⋯,Returnt} and monthly proxies {proxyi,1,⋯,proxyi,t-1}. In the second-stage regression, for month t, we run the cross-sectional regression as follows:3 proxyi,t=α+StPLSπ^i1+vi,t,i=1,2,⋯,N. In the above regression, the independent variable is the loadings that we obtain in Eq. (2). Then the slope estimate StPLS obtained in the above regression is the look-ahead-bias-free sentiment estimate for month t. For sentiment estimate St+1PLS at month t + 1, we re-estimate the above first-stage regressions by expanding the first-stage sample to {s: s ≤ t + 1} and re-estimate the second-stage regression for time-period t + 1. By recursively implementing the above procedure with an expanding estimation window, we obtain a recursive look-ahead-bias-free estimate of the housing sentiment sequence {StPLS,t=t0, …., T}. Note that the first available sentiment estimate comes on the date t0, which is usually later than the starting period of the proxy and housing return sample. This sample loss is because we need to use an initial training data set {s: s ≤t0} to estimate the first-stage regression for the month t0. In the empirical implementation, we use the first twelve months as the initial training sample and hence our final sentiment sample starts one year later than the transaction data as exhibited in Table 1. We display the housing sentiment index (in red) and the hedonic HPI returns (in blue) for selected cities in Fig. 1. As Fig. 1 shows, the trend of housing sentiments is similar to that of housing returns, especially in Beijing, Guangzhou, and Shenzhen. This co-movement pattern indicates a positive sentiment-return relationship. We also calculate the contemporaneous correlation between the sentiment index (StPLS) and the housing return Rt as shown in Table 9 in Appendix A. The correlation coefficient (0.327) is similar to that of Zhou (2018) with a 0.26 correlation between her sentiment indices and returns for the Shanghai housing market.Fig. 1 Housing sentiments and returns. Notes. The blue line represents hedonic HPI returns (in percentage) and the red line represents housing sentiments Correlation with Proxies and Confidence Indices To confirm the reliability of our sentiment indices, we first calculate the correlations between StPLS and these three sentiment proxies. As Table 9 in Appendix A shows, the housing sentiment StPLS is positively correlated with Turnover and negatively correlated with MedianIntv. These results are consistent with the findings on the positive correlation between market liquidity and sentiments in the literature. To further investigate the correlation between local sentiment and proxies, we calculate their correlation for each city and display the results in Table 3. Results in Table 3 verify the positive correlation between sentiment and Turnover and the negative correlation between sentiment and MedianIntv.Table 3 Correlation of local housing sentiments with three sentiment proxies Num City MedianIntv Turnover SMB 1 Shanghai -0.366 0.035 0.023 2 Beijing -0.554 0.206 -0.159 3 Nanjing -0.400 0.270 0.003 4 Xiamen -0.379 -0.030 -0.080 5 Hefei 0.197 0.488 0.002 6 Dalian -0.284 0.219 0.103 7 Tianjin -0.558 0.113 0.117 8 Guangzhou 0.433 -0.471 0.010 9 Chengdu -0.684 0.481 -0.058 10 Hangzhou -0.790 0.486 0.075 11 Wuhan -0.651 0.187 0.003 12 Shenyang -0.438 -0.549 -0.071 13 Jinan -0.776 0.321 0.129 14 Shenzhen -0.287 0.412 0.157 15 Suzhou -0.795 0.454 0.098 16 Chongqing 0.475 -0.230 -0.035 17 Changsha -0.635 0.280 0.098 18 Qingdao -0.590 0.246 -0.061 Values are the correlation coefficients between the local housing sentiment index and three sentiment proxies for each city. MedianIntv is the logarithm of the median days on the market. Turnover equals the total area of transacted houses divided by the total area of houses available for sale in a given month, and SMB equals the small-house return minus the big-house return (in percentage) Interestingly, distinct from Zhou (2018), we find that 12 out of 18 cities show a positive correlation between sentiment and SMB. As argued by Zhou (2018), SMB should be negatively correlated with housing sentiment. However, this is not the case for Chinese housing markets. As mentioned above, speculative investors in Chinese housing markets tend to over-invest in smaller houses, and hence prices for smaller housing units appreciate faster in cities with more speculative investors. So, the proxy SMB reflecting housing investors’ speculative behavior should be positively correlated with housing sentiments. Our empirical results in Table 3 verify the prevalence of speculative investors in Chinese housing markets. Cities such as Shanghai, Nanjing, Hefei, Dalian, Tianjin, Guangzhou, Hangzhou, Wuhan, Jinan, Shenzhen, Suzhou, and Changsha, all show significantly positive correlations between local housing sentiments and the small-house return premium. These findings indicate that speculative behaviors are prevalent in Chinese housing markets, and over-optimistic investors are not buying bigger houses but instead over-investing in smaller housing units. We also compare our sentiment index with some official indices for Beijing and Shanghai, the two largest cities in China. Since housing buyers include both consumers and investors (Han, 2013; Miller & Pandher, 2008), we consider consumer confidence indices and investor confidence indices. As Table 10 of Appendix A shows, our housing sentiment index of Shanghai is significantly positively correlated with the Housing Boom Index of China (HBCN) but negatively correlated with the Investor Confidence Index for Economic Policy (ICCNP) and the Investor Confidence Index of China (ICCN). The sentiment index of Beijing exhibits positive but insignificant correlations with the Consumer Confidence Index of Beijing (CCBJ) and the Consumer Expectation Index of Beijing (CEBJ). While these results provide some support to the reliability of our second-hand transaction-based housing market sentiment index, it also reveals that housing sentiment may exhibit distinct characteristics relating to macroeconomic indices. Local Housing Sentiments and Returns Predicting Returns with Local Housing Sentiments In this section, we examine the relationship between local housing sentiments and housing market returns. The basic regression equation is set as follows:4 Ri,t+1=α+βSitPLS+γSpringt+ηAutumnt+θMacrost+δFundamentalsit+πi+λt+εit, where Ri,t+1 denotes the housing market return of city i at time t+1, calculated by the log return (in percentage) of HPI constructed in "Housing Price Index Based on a Hedonic Model" section. SitPLS is the housing sentiment of city i at time t. Springt and Autumnt are seasonality dummy variables, standing for the cold and hot seasons for Chinese housing markets, respectively.8Macrost refers to some macroeconomic control variables which only vary across time but not across cities, including Purchasing Managers Index (PMIt), the growth rate of M2 (M2Gt), and the yield spread between AA-grade corporate bonds and AAA-grade corporate bonds (Defaultt). We also control for some city-specific economic fundamentals (Fundamentalsit), including the growth rate of the consumer price index (CPIit), the growth rate of the population (populationit), the unemployment rate (unemploymentit), the growth rate of the average income of on-the-job employees (incomeit), the growth rate of the average monthly per-square-meter rents (rentsit), the house price to rent ratio (PRRit).9 City-fixed effects (πi) are included in the regression, and time-fixed effects, λt, are also controlled for in the last specification in the result table. All these control variables are obtained from the Wind dataset. Finally, εit denotes the regression residual. Table 4 presents the results from Eq. (4) with standard errors clustered at the city level while Table 11 in Appendix A provides the detailed results of this regression. From Column (1) to Column (6), control variables are added step by step. In the first column, we only control for seasonality dummies (Autumnt and Springt) besides the sentiment variable and the city-fixed effects. In the second column, we further control for time-varying macroeconomic variables. City-specific fundamental controls are added in Column (3) and lagged macroeconomic and fundamental variables are further included in Column (4) of Table 4. In Column (5), we also include the first lag of housing returns in consideration of the persistence of housing returns. In the last column of Table 4, time-fixed effects are controlled for and hence these time-varying macroeconomic variables and seasonality dummies are omitted.Table 4 Predicting future returns by local housing sentiments (1) (2) (3) (4) (5) (6) Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 SitPLS 0.451 0.447 0.522* 0.296 0.113* 0.137 (0.176) (0.177) (0.174) (0.120) (0.060) (0.062) Autumnt -0.991* -1.025* -0.922* -0.638* -0.619* (0.197) (0.202) (0.186) (0.176) (0.152) Springt 0.240 0.212 0.280 0.289 -0.191 (0.155) (0.143) (0.185) (0.136) (0.153) Ri,t 0.606* 0.598*** (0.067) (0.068) Macrost No Yes Yes Yes Yes No Fundamentalsit No No Yes Yes Yes Yes laggedControlit No No No Yes Yes Yes CityFE Yes Yes Yes Yes Yes Yes TimeFE No No No No No Yes Num.ofcities 18 18 18 18 18 18 R2 0.114 0.129 0.174 0.352 0.494 0.519 Obs. 799 799 786 780 780 780 Controlit contains both of Macrost and Fundamentalsit. Macrost refers to macroeconomic variables including PMIt (Purchasing Managers Index), M2Gt (the growth rate of M2), and Defaultt (the yield spread between AA-grade corporate bonds and AAA-grade corporate bonds). Fundamentalsit contains CPIit (the growth rate of the consumer price index), populationit (the growth of population), unemploymentit (the unemployment rate), incomeit (the growth rate of the average income of on-the-job employees), rentsit (the growth rate of the average rents per square meter per month), and PRRit (the house price to rent ratio). City-fixed effects are controlled for in each column. Time-fixed effects are controlled for in Column (6). In the last three columns, the first lagged values of Macrost and Fundamentalsit are also included. , , and indicate significance at the 10%, 5%, and 1%, respectively. City-clustered standard errors are reported in parentheses As shown in Table 11 in Appendix A, we can see that housing returns exhibit a salient seasonality pattern (with lower returns in Autumn) and decrease with the default risk (Defaultt). Other macroeconomic variables’ impacts are not persistent across different specifications. For fundamental controls, the unemployment rate and the inflation rate display significant negative impacts on housing returns according to the last three columns. Other fundamental controls’ impacts are not significant or persistent across different specifications. The most important finding from Tables 4 and Table 11 is that the local housing sentiment has a significant (at the 5% significance level) positive impact on future housing returns. This significant impact is consistent across different specifications, although with different magnitudes. Our conclusions are mainly based on Columns (6) with time-fixed effects of Tables 4 and Table 11. Column (6) shows that housing sentiments can strongly predict future housing returns even when we control for time-fixed effects. In detail, a one-standard-deviation increase in housing sentiments is positively associated with a future monthly return appreciation of approximately 0.14%. This result indicates that the annual housing return will increase by about 1.68% (0.14% × 12) in China given a one-standard-deviation positive shock to housing sentiments. Soo (2018) also finds a positive predicting power of news media sentiments on quarterly housing returns for the U.S. housing markets. It is shown that for every one percent increase in four quarters of accumulated lagged sentiments, the future quarterly price appreciation is approximately 0.93%. With a standard deviation of the news sentiments of value 1.5, annual housing appreciation will be 5.58% (1.5 × 0.93% × 4) to a positive one standard deviation of the news sentiment shock. The housing sentiment impacts on Chinese markets estimated in this paper are comparable (but smaller) to that inferred from Soo (2018). More recently, Bork et al. (2020) use household survey responses from the University of Michigan consumer surveys to construct a housing sentiment index for the U.S. housing market. The housing sentiment constructed by Bork et al. (2020) is associated with a standard deviation of 0.08 and is estimated to impact the quarterly housing price growth rate with a coefficient of 11.67. So, one standard deviation increase in housing sentiments will lead to a 3.73 (11.67 × 0.08 × 4) percentage points increase in the annual housing appreciation. Our estimate of housing sentiment impacts on the Chinese housing markets is more comparable to the estimate from Bork et al. (2020). We offer one explanation for this disparity as well as the similarity among these sentiment impacts. Soo (2018) draws the conclusion based on city-specific sentiment indices from media tone in local newspaper articles for 34 U.S. cities from 2000 to 2013, while Bork et al. (2020) construct an aggregate housing sentiment index for the U.S. housing market based on the University of Michigan consumer surveys for a much longer period of 1975 to 2017. In contrast, the housing sentiment indices constructed in this paper are based on market transaction outcomes for 18 major Chinese cities covering the period of 2016 to 2020. Within this sample period, Chinese housing markets are under strict regulation policies, such as purchasing restrictions on multi-house owners and so on. However, even given these strict regulations, the estimated annual housing sentiment impact is still as high as 1.68%, indicating that Chinese housing markets may be also highly susceptible to irrational sentiment compared with the U.S. market. As Baker and Wurgler (2006) suggested, assets with limited arbitrage were prone to be influenced by sentiment more easily. This comparable sentiment impact on housing returns in China under a stringent regulatory environment also indicates that there may be many arbitrage and speculation behaviors in the Chinese housing market. Underreaction and Overreaction Further, it is meaningful to explore whether the transaction-based sentiment captures fundamental information or just market sentiments. According to behavioral asset-pricing theories, if the sentiment captures investors’ behavior bias, the cumulative response curve of asset price should exhibit short-run underreaction and long-run reversal. Liu et al. (2019) show that the effect of sentiments on stock returns behaves the pattern of short-run underreaction and long-run reversal in China’s stock market. In this section, we want to explore whether the same pattern will appear in the housing market. Following Liu et al. (2019), we adopt the local projection method of Jordà (2005) to estimate cumulative impulse responses of housing returns to local housing sentiment indices. As shown by Jordà (2005), this local projection method is more robust to misspecifications than conventional vector-autoregression (VAR) for estimating impulse responses. Specifically, we consider the following multiple-horizon predictive regressions:5 logPib-logPit=αb+βbSitPLS+δbControlit+πi+λt+εit, where b ranges from month t+1 to month t+12, extending the return horizon from 1 month to 1 year. Pit represents the second-hand HPI constructed in "Housing Price Index Based on a Hedonic Model" section. logPib-logPit measures the cumulative logarithmic return (in percentage) for city i from t to t+b, denoted by Ri,t,t+b, for b=1,⋯,12. The cumulative impulse response of housing prices to SitPLS is captured by βb at different horizons, which can be separately estimated by ordinary least squares. As in Column (6) of Table 4, we control for city-fixed effects (πi), time-fixed effects (λt), and some fundamental variables including CPIit, populationit, unemploymentit, incomeit, rentsit, PRRit, and their lagged terms in the estimation (variable definitions see "Predicting Returns with Local Housing Sentiments" section).10 Table 5 reports the estimated housing sentiment coefficients at different return horizons. We can find that the cumulative impulse response gradually increases, peaks at around 5 months, and reverses to a lower and statistically insignificant level at longer horizons. Figure 2 depicts the coefficients of housing sentiment βb from Table 5 with the time horizon. The salient pattern of short-run underreaction and long-run overreaction is similar to what Liu et al. (2019) document for the Chinese stock market. The significant short-run impacts and insignificant long-run impacts indicate that the sentiment indeed leads to mispricing in China’s housing markets in the short run. And the lack of permanent effects of housing sentiments on cumulative returns in the long run indicates that there is no fundamental information contained in the housing sentiments. This result verifies the existence of irrational traders in the Chinese housing market by standard behavioral asset-pricing models as in Barberis et al. (1998); De Long et al. (1990); Daniel et al. (1998); Hong and Stein (1999).Table 5 Multi-Horizon Return Regressions Return Horizon Ri,t,t+b 1 Month 2 Months 3 Months 4 Months 5 Months 6 Months 7 Months 8 Months 9 Months 10 Months 11 Months 12 Months SitPLS 0.137 0.311* 0.581* 0.766* 0.844* 0.845 0.717 0.519 0.350 0.320 0.353 0.280 (0.062) (0.169) (0.316) (0.409) (0.484) (0.525) (0.613) (0.654) (0.667) (0.656) (0.635) (0.643) R2 0.519 0.542 0.528 0.527 0.506 0.516 0.516 0.500 0.480 0.490 0.481 0.449 Obs. 780 762 744 724 703 681 656 631 608 589 570 550 This table reports the estimated sentiment coefficient βb in logPib-logPit=αb+βbSitPLS+δbControlit+πi+λt+εi,t, with b varying from 1 to 12 months. Pit denotes second-hand housing price index. Control variables include lagged monthly returns,CPIit, populationit, unemploymentit, incomeit, rentsit, PRRit, and their lagged values. City-fixed effect πi and time-fixed effect λt are controlled for in regression. Finally, , , and indicate significance at the 10%, 5%, and 1% levels, respectively. City-clustered standard errors are reported in parentheses Fig. 2 Sentiments and future multi-horizon returns. Notes. This figure depicts the coefficient of sentiment βb for horizon b ranged from 1 to 12 months in regression (5). The dots on the curve indicates the coefficient βb is significant and no dots means insignificant Asymmetric Effects of Sentiments In this section, we decompose our sentiment SitPLS into a positive part and a negative part to investigate potential asymmetric housing sentiment impacts. We define Sitpos=maxSitPLS,0 as the positive sentiment, and Sitneg=minSitPLS,0 as the negative sentiment.11 Then the Sit in Eq. (5) is replaced by Sitpos and Sitneg while other control variables remain the same. The empirical results are displayed in Table 6. Table 6 shows that Sitneg has significantly positive impacts on cumulative returns at short horizons (b = 1, 3, 6) while Sitpos has no significant coefficients for all horizons. These results indicate that local housing sentiments have an asymmetric impact on housing returns, with only negative sentiments having significant positive effects on future housing returns.Table 6 Predicting returns by negative and positive sentiments (1) (2) (3) (4) (5) Return Horizon b = 1 b = 3 b = 6 b = 9 b = 12 Sitneg 0.265* 1.045* 1.746 1.174 0.939 (0.084) (0.322) (0.748) (1.176) (1.201) Sitpos 0.043 0.241 0.197 -0.204 -0.124 (0.124) (0.541) (0.731) (0.661) (0.702) Controlit Yes Yes Yes Yes Yes CityFE Yes Yes Yes Yes Yes TimeFE Yes Yes Yes Yes Yes Num.ofcities 18 18 18 18 18 R2 0.520 0.530 0.519 0.483 0.450 Obs. 780 744 681 608 550 This table reports the results of Ri,t,t+b=αb+βbnegSitneg+βbposSitpos+δbControlit+πi+λt+εit. Sitpos=maxSitPLS,0, denoting the positive sentiment. Sitneg=minSitPLS,0, denoting the negative sentiment. Controlit comprises CPIit, populationit, unemploymentit, incomeit, rentsit, PRRit, and their lagged values. City-fixed effects and time-fixed effects are also controlled for. Finally, , , and indicate significance at the 10%, 5%, and 1% levels, respectively. City-clustered standard errors are reported in parentheses Our findings are distinct from Zhou (2018) who shows that high positive sentiments are followed by low housing returns. On one hand, our conclusions are based on panel data analysis for 18 cities other than Shanghai only. The broad set of cities makes our findings more representative of the Chinese housing markets. On the other hand, this different pattern may be related to our sample period. Zhou (2018) constructed a trade-based sentiment for Shanghai from January 2010 to May 2015, in which the housing price experienced several rounds of upswing. In contrast, our sample covers the period of January 2016 to October 2020, during which a series of tightening policies were introduced to curb the rapid growth of housing prices. The tightening policy trend and the global COVID-19 epidemic may result in a generally low level of housing sentiment relative to the period before 2016. Sentiments and Housing Supply Elasticities Given housing supply elasticity is one of the key determinants of housing prices (Glaeser et al., 2006; Liu, 2014), we will investigate potential heterogeneous sentiment effects across cities with different housing supply elasticities in this section. The supply elasticity estimates are obtained from Liu et al. (2019), which construct housing supply elasticities for 282 cities in China by combining natural geographical constraints, cultivated land protection constraints, and floor area ratio regulations of local housing markets. These elasticity measures show housing supply in most Chinese cities is overall inelastic and exhibits cross-sectional heterogeneity. Based on these elasticity estimates, we divide cities into two groups, one with a relatively elastic housing supply and the other with a relatively inelastic housing supply. More specifically, we define a dummy variable, ElastHighi, taking the value of 1 if the elasticity of city i is greater than the 75 percentile of elasticities of 282 cities given by Liu et al. (2019), and the value of zero otherwise. In our sample, Shanghai, Xiamen, and Hefei belong to the supply elastic group (ElastHighi=1). Note that Shanghai has a relatively high supply elasticity, which could be related to its looser land use regulations and flat terrain in the area (Liu et al., 2019). Then an interactive term of SitPLS and ElastHighi is included in Eq. (5). The results are given in Table 7. For b = 1, the coefficient on the interactive term is significantly negative, which means that the local housing sentiment has a stronger impact on returns for cities with relatively inelastic housing supply than cities with a more elastic housing supply. Also, the heterogeneous sentiment effects are persistent even for longer horizons (b = 3, 6, 9). This significant heterogeneous sentiment impacts on housing returns of cities with varying supply elasticities are consistent with findings in Zheng et al. (2016), which show that the confidence index based on Google search also has a larger impact on housing appreciations of cities with relatively inelastic housing supply.Table 7 Interactive effects of housing supply elasticities and sentiments (1) (2) (3) (4) (5) (6) (7) (8) Return Horizon b = 1 b = 1 b = 3 b = 3 b = 6 b = 6 b = 9 b = 9 SitPLS 0.137 0.214** 0.581* 0.851** 0.845 1.205* 0.350 0.880 (0.062) (0.077) (0.316) (0.344) (0.525) (0.583) (0.667) (0.708) SitPLS×ElastHighi -0.361 -1.253* -1.718 -2.441* (0.156) (0.403) (0.673) (0.779) Controlit Yes Yes Yes Yes Yes Yes Yes Yes laggedControlit Yes Yes Yes Yes Yes Yes Yes Yes CityFE Yes Yes Yes Yes Yes Yes Yes Yes TimeFE Yes Yes Yes Yes Yes Yes Yes Yes Num.ofcities 18 18 18 18 18 18 18 18 R2 0.519 0.524 0.528 0.537 0.516 0.523 0.480 0.496 Obs. 780 780 744 744 681 681 608 608 ElastHighi equals 1 if the housing supply elasticity of city i is greater than the 75 percentile of 282 cities estimates in China given by Liu et al. (2019), otherwise equals 0. Controlit includes CPIit, populationit, unemploymentit, incomeit, rentsit, and PRRit. We also include the first lags of all the control covariates (denoted by laggedControlit). City-fixed effects and time-fixed effects are controlled for in all columns. Finally, , , and indicate significance at the 10%, 5%, and 1% levels, respectively. City-clustered standard errors are reported in parentheses The finding that local housing supply elasticities play an important role in the sentiment effects highlights the necessity of considering local market conditions in the analysis of sentiment analysis. For cities with a much more elastic housing supply, the impacts of market sentiments on returns are much smaller or even negligible. This may be because that cities with elastic housing supply could adjust housing supply easily when expecting a warming-up market sentiment. However, the supply elasticity is not time-invariant. According to Liu et al. (2019), natural geographical constraints, cultivated land protection constraints, and floor area ratio regulations are the three main factors in determining local supply elasticities. While the natural physical landscape is hard to change and land protection constraints are not allowed to loosen, local official governments could loosen regulations on floor area ratios in land development to alleviate the impacts of housing sentiments. Predicting Sentiments with Returns So far, we have shown that the transaction-based local housing sentiment has a positive effect on future housing returns, exhibits the pattern of short-run underreaction and long-run overreaction, and has an asymmetry impact on housing returns. In addition, the market sentiment effects are much stronger and more significant for cities with relatively inelastic housing supply. Intuitively, sentiment may be affected by past housing returns. Soo (2018) shows that past price appreciations predict higher media sentiment for the U.S. housing markets. In this section, we investigate the determinants of housing sentiments by running the regression as follows:6 Si,t+1PLS=α+βRit+γControlit+δSitPLS+μi+λt+εit, where Si,t+1PLS denotes the sentiment of city i in month t+1, Rit denotes the housing return of city i in month t. We control for SitPLS,rentsit,PRRit,CPIit, populationit, unemploymentit, incomeit (and their first lags), city-fixed effects as well as time-fixed effects in Eq. (6). Table 8 reports the estimation results. Columns (1)-(4) show that the housing return Rit is positively correlated with future sentiment Si,t+1PLS, consistent with Soo (2018). Higher past housing returns will foster higher market sentiments. Results of Column (4) indicate a significant positive serial correlation in housing sentiments. These results indicate that the expectations of Chinese real estate investors are backward-looking and there exists a significant feedback effect between housing returns and market sentiments. This feedback effect could result in a pricing-sentiment spiral, as defined by Ling et al. (2015), in which the dynamic interplay between sentiment and house price appreciations can create a self-reinforcing spiral. The pricing-sentiment spiral could potentially extend the length and magnitude of housing market cycles in China and hence largely exaggerates the ongoing housing market fever in China.Table 8 Predicting sentiments with past returns (1) (2) (3) (4) (5) (6) (7) Return Horizon b = 1 b = 1 b = 1 b = 1 b = 3 b = 6 b = 9 Ri,t-b,t 0.195* 0.185* 0.155 0.041* 0.009 0.005 -0.000 (0.046) (0.048) (0.063) (0.023) (0.008) (0.005) (0.003) rentsit -0.001 0.016 -0.010 0.002 0.001 -0.004 (0.009) (0.037) (0.017) (0.020) (0.022) (0.020) PRRit 0.085 0.698 -0.095 0.199 0.183 0.048 (0.123) (0.858) (0.334) (0.380) (0.441) (0.408) SitPLS 0.825* 0.824* 0.821* 0.833* (0.034) (0.039) (0.044) (0.045) Controlit No Yes Yes Yes Yes Yes Yes laggedControlit No No Yes Yes Yes Yes Yes CityFE Yes Yes Yes Yes Yes Yes Yes TimeFE Yes Yes Yes Yes Yes Yes Yes Num.ofcities 18 18 18 18 18 18 18 R2 0.163 0.191 0.220 0.751 0.749 0.738 0.729 Obs. 813 797 793 780 776 759 705 Dependent variable is Si,t+1PLS, the sentiment of city i in month t+1; Ri,t-b,t denotes the cumulative returns (in percentage) of the past 1 month (b=1), 3 months (b=3), 6 months (b=6), and 9 months (b=9), respectively. Controlit include PRRit, rentsit CPIit, populationit, unemploymentit, and incomeit. We also include the first lags of all the control covariates (denoted by laggedControlit). City-fixed effects and time-fixed effects are also controlled for in the regression. Standard errors clustered at the city level are in parentheses. Finally, , , and indicate significance at the 10%, 5%, and 1% levels, respectively Moreover, we find negative but insignificant effects of rent growth rates (rentsit) and price-rent ratios (PRRit) on local housing sentiments as shown in Table 8. To further explore how persistently the past accumulated returns can push up local sentiments, we replace Rit with Ri,t-b,t in Eq. (6). The last three columns of Table 8 report the estimates for 3-month, 6-month, and 9-month return horizons, respectively. The results show that only the past 1-month return significantly predicts higher market sentiments, while past cumulative returns over longer horizons (i.e., 3-month, 6-month, and 9-month) seem to exert no significant impacts on housing sentiments. The most recent past returns (1-month) appear to exert the greatest impact on housing sentiment, similar to findings in Soo (2018). Robustness Checks In this section, we conduct several robustness checks on our main regression results. Firstly, to remove the potential shock of COVID-19 on the Chinese real estate market, we exclude the sample period after November 2019 and re-estimate our main regression. Secondly, we use alternative approaches (scaled PCA and PCA) to construct the sentiment index and re-estimate the main regression. Thirdly, we substitute an alternative proxy of sentiment, the new listings of second-hand properties, for turnover and then re-construct our sentiment index and re-estimate the main regression. Finally, to confirm the reliability of the recursive look-ahead-bias-free implementation of the PLS approach, we conduct a placebo test following Kelly and Pruitt (2013). Pre-Covid-19 Sub-sample Estimation COVID-19 broke out in November 2019 and was considered to have caused a huge negative shock to the Chinese economy and especially to the real estate market. To remove the potential effect of COVID-19, we discard the data after November 2019 and re-estimate our main empirical regression. Using the sample from January 2016 to October 2019, Panel A of Table 12 in Appendix B shows that a one-standard-deviation increase in local housing sentiment is associated with a future monthly return increase of approximately 0.14%, which is almost the same as that from the full sample estimation in "Predicting Returns with Local Housing Sentiments" section. Alternative Approaches of Sentiment Construction: scaled PCA and PCA To further prove the robustness of the predictability of local housing sentiment on future returns, we consider two alternative sentiment construction methods, i.e., the scaled PCA by Huang et al. (2022) and the traditional PCA method. For both alternative methods, we first follow the first four steps as described in "Housing Sentiment Index" section to get the final sentiment proxies (MedianIntv, Turnover, and SMB). And then, for the PCA method, we extract the first principal component of these three proxies to derive the PCA housing sentiment SitPCA. For the scaled PCA method, after we get the final sentiment proxies following the first four steps, for month t, we run the following regression: Returns=δi,t+γi,tproxyi,s-1+ui,s,s≤t, (7). by using information only up to month t. We then apply PCA to scaled proxies (γ^1,tMedianIntv,γ^2,tTurnover, γ^3,tSMB) to extract the first component as the housing sentiment StsPCA.12 By recursively estimating the above predictive regression (7) and applying PCA to scaled proxies with an expanding window scheme, we obtain a recursive look-ahead-bias-free scaled PCA estimate of the housing sentiment sequence {StsPCA,t=t0, …., T}. Like the recursive implementation of PLS, we use the first twelve months as the initial training sample and the constructed scaled PCA housing sentiment sequences are associated with one-year shorter sample periods. As the correlation matrix in Table 9 of Appendix A shows, the new alternative scaled PCA sentiment index SitsPCA is significantly positively correlated with the original sentiment SitPLS (with a correlation coefficient of 0.491). The PCA sentiment SitPCA is also significantly positively correlated with SitPLS, but with a much smaller correlation coefficient of 0.233. We re-estimate our main regression by replacing the sentiment estimates with SitPCA and SitsPCA with results summarized in Panel B (scaled PCA) and Panel C (PCA) of Table 12 in Appendix B, respectively. As results show, housing sentiment indices derived from both alternative methods still exert a significant impact on future housing returns, although with smaller and less significant impacts than that from the PLS method. To briefly sum up, the predictability of the local housing sentiment we document is robust to the choice of the sentiment index construction method. However, our results do suggest that the PLS method seems to do a better job at constructing a predictive sentiment index than scaled PCA and PCA in our research setting. Alternative Proxy: New listings Lowry (2003) suggests that the number of initial public offerings (IPOs) reflects stock market sentiment. Similarly, in the housing market, a larger number of new listings could indicate a higher market sentiment. In this section, we replace the raw proxy Turnoverit with NewListingsit, the number of new listings at month t in city i.13 Then we apply these proxy cleaning procedures to MedianIntv, NewListings, and SMB, and conduct the recursive look-ahead-bias-free implementation of PLS to construct a look-ahead-bias-free sentiment index SitPLS′ for each city in our sample. As the correlation matrix in Table 9 of Appendix A shows, NewListings is significantly positively correlated with Turnover (with a correlation coefficient of 0.118) but negatively correlated with MedianIntv (with a coefficient of -0.122). And the new alternative sentiment index SitPLS′ is significantly positively correlated with the original SitPLS (with a correlation coefficient of 0.466). We display the main regression results using this alternative sentiment index SitPLS′ in Panel D of Table 12 in Appendix B. The coefficient on SitPLS′ (0.178 and significant at 5% level) is close to 0.137 from the main estimation in Column (6) of Table 4. This result suggests that our findings on the significant predictability of housing sentiment on housing returns are robust with the alternative proxy NewListings. Placebo Test Following Kelly and Pruitt (2013), we conduct a placebo test to confirm that our estimation procedure does not generate sentiment predictability mechanically. Intuitively, we can test our method by using simulated random sentiment known to have no true return forecasting ability. In detail, we conduct the placebo test as follows. Firstly, we generate three random series from the standard normal distribution (with zero mean and unit variance) as “fake” raw proxies for each city. Secondly, we then conduct the five-steps approach as described in "Housing Sentiment Index" section to these randomly generated proxy series to construct a “placebo” look-ahead-bias-free PLS sentiment index SitPlacebo. Finally, we examine the predictability of this “placebo” sentiment on future returns. If we find no significant predictive power of this “placebo” sentiment, we could provide evidence that the look-ahead-bias-free implementation of PLS does not yield “false predictability” mechanically. As the results in Table 13 of Appendix C show, this “placebo” sentiment exhibits no consistent predictability on future housing returns.14 This test confirms that the recursive look-ahead-bias-free implementation of PLS does not mechanically yield a sentiment with significant predictability. Conclusions This paper investigates the relationship between local housing sentiments and returns in Chinese housing markets. We construct monthly city-level sentiment indices for 18 Chinese cities from January 2016 to October 2020 by using a massive second-hand transaction dataset through a recursive look-ahead-bias-free implementation of the PLS method. These local housing sentiment indices are based on two housing market liquidity proxies and a small-house return premium measure. Empirically, we find that local housing sentiments can significantly predict future housing market returns. The sentiment impact is comparable with estimates for the U.S. housing market in the literature, implying that the Chinese housing markets are also susceptible to irrational sentiments even under a stringent regulatory environment. Furthermore, a salient short-run underreaction and long-run overreaction pattern of sentiment effects is documented. Further analysis shows that local housing sentiment impacts are asymmetric, and housing returns in cities with relatively inelastic housing supply are more sensitive to local market sentiments. Last but not the least, we show that the expectations of Chinese real estate investors are backward-looking and there exists a significant feedback effect between housing returns and market sentiments. Our major findings are robust to alternative sentiment construction methods and alternative sentiment proxy choice, and consistent for the sub-sample before COVID-19. Our empirical analysis enriches the literature on the role of sentiments in the housing market and provides better insights into the housing market dynamics. This paper’s findings can provide references for policymakers to stabilize and improve the functioning of the housing market. Appendix A Tables Table 9 Correlation of sentiments, proxies, and returns StPLS StSPCA StPCA SitPLS′ MedianIntvt Turnovert SMBt NewListingst Rt StPLS 1.000 StSPCA 0.491* 1.000 StPCA 0.233* 0.234* 1.000 StPLS′ 0.466* 0.420* 0.372* 1.000 MedianIntvt -0.212* -0.297* -0.066 -0.134* 1.000 Turnovert 0.293* 0.167* 0.115* 0.163* -0.575* 1.000 SMBt 0.001 0.022 0.064 0.030 -0.018 -0.032 1.000 NewListingst -0.037 0.003 -0.033 -0.039 -0.122* 0.118* -0.015 1.000 Rt 0.327* 0.151* 0.211* 0.365* -0.233* 0.458* 0.013 0.065 1.000 StPLS stands for the sentiment index from the PLS method, while StsPCA stands for the sentiment index from the scaled PCA method, StPCA stands for the sentiment index from the PCA method, and SitPLS′ stands for the sentiment index based on the alternative proxy NewListings. Significance levels of 10%, 5%, 1% are marked by , , and , respectively 9, Table 10 Correlation of sentiment index with confidence indices Panel A: Beijing CCBJ CEBJ HBCN ICCNF ICCNP ICCN CCCN CECN Correlation 0.460 0.472 -0.042 0.205 0.006 0.058 -0.262 -0.252 P-Value 0.114 0.104 0.799 0.447 0.983 0.758 0.103 0.117 Panel B: Shanghai CCSH CESH HBCN ICCNF ICCNP ICCN CCCN CECN Correlation 0.145 0.036 0.344 -0.126 -0.432 -0.561* -0.105 -0.115 P-Value 0.636 0.908 0.017 0.557 0.035 0.000 0.480 0.436 HBCN is the Housing Boom Index of China; ICCNF is the Investor Confidence Index for Domestic Economic Fundamentals; ICCNP is the Investor Confidence Index for Economic Policy; ICCN is the Investor Confidence Index of China; CCCN is the Consumer Confidence Index of China; CECN is the Consumer Expectations Index of China. In Panel A, CCBJ is the Consumer Confidence Index of Beijing; CEBJ is the Consumer Expectation Index of Beijing; In Panel B, CCSH is the Consumer Confidence Index of Shanghai; CESH is the Consumer Expectation Index of Shanghai. All these data are sourced from the National Bureau of Statistics and Wind. Significance level of 10%, 5%, 1% are marked by , , and , respectively 10, Table 11 Predicting future returns by local housing sentiments VARIABLES (1) (2) (3) (4) (5) (6) Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 SitPLS 0.451 0.447 0.522* 0.296** 0.113* 0.137 (0.176) (0.177) (0.174) (0.120) (0.060) (0.062) Autumnt -0.991* -1.025* -0.922* -0.638* -0.619* (0.197) (0.202) (0.186) (0.176) (0.152) Springt 0.240 0.212 0.280 0.289** -0.191 (0.155) (0.143) (0.185) (0.136) (0.153) Defaultt -1.363* -1.756 -3.401* -2.502* (0.754) (0.757) (1.001) (0.822) PMIt 0.028 0.038 -0.013 -0.046 (0.038) (0.036) (0.033) (0.033) M2Gt 0.095 0.114 0.328* 0.178 (0.074) (0.068) (0.098) (0.103) CPIit 0.029 -0.085 -0.232 -0.364 (0.103) (0.093) (0.105) (0.152) Populationit -0.216 0.098 0.159 0.103 (0.205) (0.573) (0.331) (0.374) Unemploymentit 0.649 -3.950* -3.420 -2.367 (0.778) (2.024) (1.264) (1.077) Incomeit 0.875 3.139 2.383 1.660 (0.602) (2.735) (1.823) (1.696) Rentsit 0.025 0.318* 0.011 0.014 (0.019) (0.059) (0.031) (0.035) PRRit -0.715 7.555*** -0.414 -0.294 (0.275) (1.359) (0.876) (0.989) Defaultt-1 2.185 1.602* (1.257) (0.897) PMIt-1 0.059* 0.028 (0.030) (0.030) M2Gt-1 0.329* 0.169 (0.078) (0.060) CPIi,t-1 0.025 -0.003 -0.093 (0.080) (0.074) (0.128) Populationi,t-1 -0.255 -0.286 -0.232 (0.574) (0.348) (0.408) Unemploymenti,t-1 4.785* 3.697** 2.568* (2.581) (1.501) (1.299) Incomei,t-1 -2.777 -2.321 -1.673 (2.752) (1.868) (1.728) Rentsi,t-1 0.010 -0.004 -0.000 (0.013) (0.015) (0.012) PRRi,t-1 -8.350* -0.192 -0.367 (1.376) (0.810) (0.919) Ri,t 0.606* 0.598* (0.067) (0.068) Constant 0.779* 0.048 -0.142 -1.318 2.960 1.465 (0.061) (1.599) (2.272) (1.975) (1.780) (1.417) CityFE Yes Yes Yes Yes Yes Yes TimeFE No No No No No Yes Num.ofcities 18 18 18 18 18 18 R2 0.114 0.129 0.174 0.352 0.494 0.519 Obs. 799 799 786 780 780 780 Controlled covariates include seasonality dummies (Autumnt and Springt), macroeconomic covariates (Macrost), and fundamental controls (Fundamentalsit).Macrost contains PMIt (Purchasing Managers Index), M2Gt (the growth rate of M2), and Defaultt (the yield spread between AA-grade corporate bonds and AAA-grade corporate bonds). Fundamentalsit contains CPIit (the growth rate of the consumer price index), populationit (the growth of population), unemploymentit (the unemployment rate), incomeit (the growth rate of the average income of on-the-job employees), rentsit (the growth rate of the average rents per square meter per month), and PRRit (the house price to rent ratio). City-fixed effects are controlled for in each column. Time-fixed effects are controlled for in Column (6). In the last three columns, we also include the first lagged values of Macrost and Fundamentalsit. , , and indicate significance at the 10%, 5%, and 1% levels, respectively. City-clustered standard errors are reported in parentheses 11 Appendix B Table Table 12 Robustness Checks: Predicting future returns by local housing sentiments (1) (2) (3) (4) (5) (6) Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Panel A. Pre-COVID-19 estimation (for January 2016-October 2019) SitPLS 0.519 0.483 0.576* 0.339** 0.117* 0.142 (0.205) (0.211) (0.195) (0.123) (0.062) (0.062) Ri,t 0.581* 0.590*** (0.072) (0.066) Macrost No Yes Yes Yes Yes No Fundamentalsit No No Yes Yes Yes Yes laggedControlit No No No Yes Yes Yes CityFE Yes Yes Yes Yes Yes Yes TimeFE No No No No No Yes R2 0.128 0.173 0.263 0.440 0.556 0.584 Obs. 606 606 594 589 589 589 Panel B. Alternative method: Scaled PCA SitsPCA 0.176* 0.182* 0.257 0.168 0.072 0.092* (0.089) (0.090) (0.113) (0.079) (0.049) (0.045) Ri,t 0.614* 0.606* (0.062) (0.062) Macrost No Yes Yes Yes Yes No Fundamentalsit No No Yes Yes Yes Yes laggedControlit No No No Yes Yes Yes CityFE Yes Yes Yes Yes Yes Yes TimeFE No No No No No Yes R2 0.073 0.091 0.130 0.347 0.497 0.523 Obs. 804 804 790 784 784 784 Panel C. Alternative method: PCA SitPCA 0.313* 0.308* 0.428 0.254 0.103* 0.100* (0.160) (0.159) (0.168) (0.106) (0.052) (0.052) Ri,t 0.590* 0.582* (0.068) (0.068) Macrost No Yes Yes Yes Yes No Fundamentalsit No No Yes Yes Yes Yes laggedControlit No No No Yes Yes Yes CityFE Yes Yes Yes Yes Yes Yes TimeFE No No No No No Yes R2 0.072 0.077 0.175 0.361 0.494 0.517 Obs. 964 964 919 907 907 907 Panel D. Alternative proxy:MedianIntv, NewListings, and SMB SitPLS′ 0.535* 0.534* 0.582* 0.373* 0.161 0.178 (0.144) (0.146) (0.152) (0.101) (0.075) (0.073) Ri,t 0.590* 0.580* Macrost No Yes Yes Yes Yes No Fundamentalsit No No Yes Yes Yes Yes laggedControlit No No No Yes Yes Yes CityFE Yes Yes Yes Yes Yes Yes TimeFE No No No No No Yes R2 0.138 0.154 0.195 0.366 0.496 0.522 Obs. 800 800 788 782 782 782 SitsPCA stands for the sentiment from scaled PCA. SitPCA is the sentiment from PCA. SitPLS′ is the PLS sentiment based on MedianIntv, NewListings, and SMB. Controlit contains both of Macrost and Fundamentalsit. Macrost contains PMIt, M2Gt, and Defaultt. Fundamentalsit contains CPIit, populationit, unemploymentit, incomeit, rentsit, and PRRit. , , and ** indicate significance at the 10%, 5%, and 1% levels, respectively. City-clustered standard errors are reported in parentheses 12 Appendix C Table Table 13 Placebo Test: Predicting future returns by a “placebo” sentiment (1) (2) (3) (4) (5) (6) Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 Ri,t+1 SitPlacebo 0.168 0.157 0.188* 0.118 -0.020 -0.014 (0.101) (0.101) (0.095) (0.070) (0.054) (0.050) Autumnt -1.004* -1.083* -0.953* -0.641* -0.627*** -0.474 (0.176) (0.187) (0.174) (0.173) (0.154) (0.491) Springt 0.278* 0.201 0.287 0.256* -0.226 -0.527 (0.154) (0.143) (0.171) (0.126) (0.149) (0.537) Ri,t 0.627* 0.620* (0.066) (0.065) Macrost No Yes Yes Yes Yes No Fundamentalsit No No Yes Yes Yes Yes laggedControlit No No No Yes Yes Yes CityFE Yes Yes Yes Yes Yes Yes TimeFE No No No No No Yes Num.ofcities 18 18 18 18 18 18 R2 0.067 0.083 0.113 0.339 0.495 0.519 Obs. 798 798 785 779 779 779 SitPlacebo stands for a “placebo” sentiment based on three randomly generated sentiment proxies using the recursive look-ahead-bias-free implementation of PLS. Controlit contains both Fundamentalsit and Macrost. Macrost refers to macroeconomic variables including PMIt, M2Gt, and Defaultt. Fundamentalsit contains CPIit, populationit, unemploymentit, incomeit, rentsit, and PRRit. , , and ** indicate significance at the 10%, 5%, and 1%, respectively. City-clustered standard errors are reported in parentheses 13 Acknowledgements We thank Kun Duan at Huazhong University of Science and Technology, Zehua Zhang at Hunan University, and seminar participants at the 2021 Jinan-SMU-ABFER Virtual Conference on Urban and Regional Economics for helpful comments. We appreciate Naqun Huang and Yanmin Yang at Nanjing Audit University for the technical support of the data. We appreciate housing supply elasticity measures provided by Xiuyan Liu at Dongnan University. This study is under the support of the National Natural Science Foundation of China (Grant No. 71903145 and 71903060). All errors are our own. Declarations Conflicts of Interest All authors declare no in this manuscript. 1 For detailed analysis, see "Predicting Returns with Local Housing Sentiments" section. 2 Other survey-based sentiment proxies, such as Anastasiou et al. (2021) which use European Commission’s consumer surveys to construct sentiment indicators representing inflationary expectations and precautionary saving incentives, and Jin et al. (2014) which use the U.S. Conference Board Consumer Sentiment Index (a measurement of consumer perception on market conditions), are more of an overall confidence index towards the entire economy than a direct measure of housing sentiment. 3 We do not consider proxies such as housing construction or housing investment in the construction of sentiment index, in that housing investment and construction, which are more of a supply side measure, usually takes several years to accomplish and hence may contain un-synchronous information with housing transaction outcomes. 4 The reason for not using new house transaction data is that housing prices for new houses are highly regulated by local governments during our sample period. In contrast, second-hand housing transactions are not subject to price ceilings set by local governments and thus are more representative of local housing market conditions. 5 In March 2016, the Fourth Session of the 12th National People's Congress proposed to stabilize housing prices in first—and second-tier cities and destock in third—and fourth-tier cities. On September 30th, 2016, Beijing Municipal Commission of Housing and Urban–Rural Development and other departments issued the official document "Several measures to promote the steady and healthy development of the real estate market in cities", (http://zjw.beijing.gov.cn/bjjs/gcjs/tzgg36/391380/index.shtml) which raised the down payment ratio of the second house to 50% and started a new round of purchase and loan restriction policy nationwide. 6 An agent of a real estate firm in Yantai (a Chinese city) suggests that “while the big houses stay lag in sale, the smaller houses are sold well, even in the periods of a housing market downturn” according to the report available at http://sd.dzwww.com/sdgd/sdgdxw/201112/t20111214_6817566.htm. 7 Please find the news report at http://news.sohu.com/20170405/n486498292.shtml. 8 Springt equals 1 if month t is in March, April, or May, while Autumnt equals 1 if month t is in September, October, or November. 9 PRRit is calculated by dividing the hedonic HPI by rentsit. Note that populationit, unemploymentit, incomeit are yearly data, and we convert them into monthly series by linear interpolation. 10 As in Column (6) of Table 4, seasonality dummies and macroeconomic variables are omitted when we control for time-fixed effects. 11 Note the sample mean of housing sentiment series are set to zero. So, the negative (positive) sentiment stands for the part of sentiment below (above) the sample average in the sample period considered. 12 The PCA procedure of the scaled PCA analysis is also conducted recursively with an expanding window scheme. In other words, for month t, the PCA analysis is applied to (γ^1,tMedianIntv,γ^2,tTurnover, γ^3,tSMB) with proxy sequences ending at month t. 13 We thank an anonymous referee for the suggestion on the NewListings proxy. 14 Except for Column (3). In Column (3) of Table 13 of Appendix C, the “placebo” sentiment is significant at the 10% level. However, this significance is not consistent as it becomes insignificant in the preferred regression of Column (6). Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Anastasiou, D., Kapopoulos, P., Zekente, K.-M. (2021). Sentimental shocks and house prices. The Journal of Real Estate Finance and Economics, 1–29. Baker M Stein JC Market liquidity as a sentiment indicator Journal of Financial Markets 2004 7 3 271 299 10.1016/j.finmar.2003.11.005 Baker M Wurgler J Investor sentiment and the cross-section of stock returns The Journal of Finance 2006 61 4 1645 1680 10.1111/j.1540-6261.2006.00885.x Baker M Wurgler J Investor sentiment in the stock market Journal of Economic Perspectives 2007 21 2 129 152 10.1257/jep.21.2.129 Barberis N Shleifer A Vishny R A model of investor sentiment Journal of Financial Economics 1998 49 3 307 343 10.1016/S0304-405X(98)00027-0 Bork L Møller SV Pedersen TQ A new index of housing sentiment Management Science 2020 66 4 1563 1583 10.1287/mnsc.2018.3258 Case KE Shiller RJ Is there a bubble in the housing market? Brookings Papers on Economic Activity 2003 2003 2 299 362 10.1353/eca.2004.0004 Case, K. E., Shiller, R. J., Thompson, A. (2012). What have they been thinking? Home buyer behavior in hot and cold markets. National Bureau of Economic Research. Chauvet M Gabriel S Lutz C Mortgage default risk: New evidence from internet search queries Journal of Urban Economics 2016 96 91 111 10.1016/j.jue.2016.08.004 Chen K Wen Y The great housing boom of china American Economic Journal: Macroeconomics 2017 9 2 73 114 Chow GC Niu L Housing prices in urban china as determined by demand and supply Pacific Economic Review 2015 20 1 1 16 10.1111/1468-0106.12080 Clayton J Ling DC Naranjo A Commercial real estate valuation: Fundamentals versus investor sentiment The Journal of Real Estate Finance and Economics 2009 38 1 5 37 10.1007/s11146-008-9130-6 Clayton, J., MacKinnon, G. (2000). Explaining the discount to nav in reit pricing: Noise or information? Available at SSRN 258268. Clayton, J., MacKinnon, G. (2002). Departures from nav in reit pricing: The private real estate cycle, the value of liquidity and investor sentiment. Real Estate Research Institute, Working Paper. Clayton J MacKinnon G Peng L Time variation of liquidity in the private real estate market: An empirical investigation Journal of Real Estate Research 2008 30 2 125 160 10.1080/10835547.2008.12091217 Da Z Engelberg J Gao PJ The sum of all fears investor sentiment and asset prices Review of Financial Studies 2015 28 1 1 32 10.1093/rfs/hhu072 Daniel, K., Hirshleifer, D., Subrahmanyam, A. (1998). Investor psychology and security market under-and overreactions. The Journal of Finance, 53(6), 1839–1885. Das PK Freybote J Marcato G An investigation into sentiment-induced institutional trading behavior and asset pricing in the reit market The Journal of Real Estate Finance and Economics 2015 51 2 160 189 10.1007/s11146-014-9490-z De Long JB Shleifer A Summers LH Waldmann RJ Noise trader risk in financial markets Journal of Political Economy 1990 98 4 703 738 10.1086/261703 Fang H Gu Q Xiong W Zhou L-A Demystifying the Chinese housing boom NBER Macroeconomics Annual 2016 30 1 105 166 10.1086/685953 Ferreira F Gyourko J Heterogeneity in neighborhood-level price growth in the united states, 1993–2009 American Economic Review 2012 102 3 134 140 10.1257/aer.102.3.134 Fu YM Qian WL Speculators and price overreaction in the housing market Real Estate Economics 2014 42 4 977 1007 10.1111/1540-6229.12071 Gallimore P Gray A The role of investor sentiment in property investment decisions Journal of Property Research 2002 19 2 111 120 10.1080/09599910110110671 Glaeser E Huang W Ma Y Shleifer A A real estate boom with Chinese characteristics Journal of Economic Perspectives 2017 31 1 93 116 10.1257/jep.31.1.93 Glaeser EL Gyourko J Saiz A Housing supply and housing bubbles Journal of Urban Economics 2008 64 2 198 217 10.1016/j.jue.2008.07.007 Glaeser EL Gyourko J Saks RE Urban growth and housing supply Journal of Economic Geography 2006 6 1 71 89 10.1093/jeg/lbi003 Glaeser EL Ward BA The causes and consequences of land use regulation: Evidence from greater Boston Journal of Urban Economics 2009 65 3 265 278 10.1016/j.jue.2008.06.003 Green RK Malpezzi S Mayo SK Metropolitan-specific estimates of the price elasticity of supply of housing, and their sources American Economic Review 2005 95 2 334 339 10.1257/000282805774670077 Han L Understanding the puzzling risk-return relationship for housing The Review of Financial Studies 2013 26 4 877 928 10.1093/rfs/hhs181 Hong H Stein JC A unified theory of underreaction, momentum trading, and overreaction in asset markets The Journal of Finance 1999 54 6 2143 2184 10.1111/0022-1082.00184 Huang DS Jiang FW Li KP Tong GS Zhou GF Scaled PCA: A new approach to dimension reduction Management Science 2022 68 3 1678 1695 10.1287/mnsc.2021.4020 Huang DS Jiang FW Tu J Zhou GF Investor sentiment aligned: A powerful predictor of stock returns Review of Financial Studies 2015 28 3 791 837 10.1093/rfs/hhu080 Hui EC Dong Z Jia S How do housing price and sentiment affect consumption distribution in china? Habitat International 2018 77 99 109 10.1016/j.habitatint.2018.01.001 Hui ECM Dong Z Jia S Lam CHL How does sentiment affect returns of urban housing? Habitat International 2017 64 71 84 10.1016/j.habitatint.2017.04.013 Hui ECM Wang ZY Market sentiment in private housing market Habitat International 2014 44 375 385 10.1016/j.habitatint.2014.08.001 Jin C Soydemir G Tidwell A The us housing market and the pricing of risk: Fundamental analysis and market sentiment Journal of Real Estate Research 2014 36 2 187 220 10.1080/10835547.2014.12091390 Jordà Ò Estimation and inference of impulse responses by local projections American Economic Review 2005 95 1 161 182 10.1257/0002828053828518 Kelly B Pruitt S Market expectations in the cross-section of present values The Journal of Finance 2013 68 5 1721 1756 10.1111/jofi.12060 Kelly B Pruitt S The three-pass regression filter: A new approach to forecasting using many predictors Journal of Econometrics 2015 186 2 294 316 10.1016/j.jeconom.2015.02.011 Keynes JM The supply of gold The Economic Journal 1936 46 183 412 418 10.2307/2224879 Lam CH-L Hui EC-M How does investor sentiment predict the future real estate returns of residential property in hong kong? Habitat International 2018 75 1 11 10.1016/j.habitatint.2018.02.009 Letdin, M., Sirmans, S., Sirmans, G. S. (2022). Betting against the sentiment in reit nav premiums. The Journal of Real Estate Finance and Economics, 64(4), 590–614. Lin CY Rahman H Yung K Investor sentiment and reit returns The Journal of Real Estate Finance and Economics 2009 39 4 450 471 10.1007/s11146-008-9115-5 Ling DC Naranjo A Scheick B Investor sentiment, limits to arbitrage and private market returns Real Estate Economics 2014 42 3 531 577 10.1111/1540-6229.12037 Ling DC Ooi JT Le TT Explaining house price dynamics: Isolating the role of nonfundamentals Journal of Money, Credit and Banking 2015 47 S1 87 125 10.1111/jmcb.12194 Liu X Housing supply elasticity, determinant factors and housing price performance of urban china Finance & Trade Economics 2014 35 4 125 137 Liu X Du C Li S Natural geographical constraints, land use regulations and china’s housing supply elasticity The Journal of Economic Research 2019 54 4 99 105 Lowry M Why does IPO volume fluctuate so much? Journal of Financial Economics 2003 67 1 3 40 10.1016/S0304-405X(02)00230-1 Miller N Pandher G Idiosyncratic volatility and the housing market Journal of Housing Research 2008 17 1 13 32 10.1080/10835547.2008.12091984 Pigou AC The laws of diminishing and increasing cost The Economic Journal 1927 37 146 188 197 10.2307/2224353 Rosen S Hedonic prices and implicit markets: Product differentiation in pure competition Journal of Political Economy 1974 82 1 34 55 10.1086/260169 Saiz A The geographic determinants of housing supply The Quarterly Journal of Economics 2010 125 3 1253 1296 10.1162/qjec.2010.125.3.1253 Shiller RC Irrational exuberance Philosophy and Public Policy Quarterly 2000 20 1 18 23 Shiller RJ Animal spirits 2010 Princeton University Press Princeton Soo CK Quantifying sentiment with news media across local housing markets The Review of Financial Studies 2018 31 10 3689 3719 10.1093/rfs/hhy036 Tan, Y., Wang, Z., Zhang, Q. H. (2020). Land-use regulation and the intensive margin of housing supply. Journal of Urban Economics, 115, 103199. Wang S Chan SH Xu B The estimation and determinants of the price elasticity of housing supply: Evidence from china Journal of Real Estate Research 2012 34 3 311 344 10.1080/10835547.2012.12091336 Wang Z Zhang Q Fundamental factors in the housing markets of china Journal of Housing Economics 2014 25 53 61 10.1016/j.jhe.2014.04.001 Wei, S.-J., Zhang, X., Liu, Y. (2012). Status competition and housing prices. National Bureau of Economic Research. Wu J Deng YH Liu HY House price index construction in the nascent housing market: The case of china Journal of Real Estate Finance and Economics 2014 48 3 522 545 10.1007/s11146-013-9416-1 Wu J Gyourko J Deng Y Evaluating the risk of Chinese housing markets: What we know and what we need to know China Economic Review 2016 39 91 114 10.1016/j.chieco.2016.03.008 Zheng S Sun W Kahn ME Investor confidence as a determinant of china's urban housing market dynamics Real Estate Economics 2016 44 4 814 845 10.1111/1540-6229.12119 Zhou ZY Housing market sentiment and intervention effectiveness: Evidence from china Emerging Markets Review 2018 35 91 110 10.1016/j.ememar.2017.12.005	0	PMC9716533	NO-CC CODE	2022-12-03 23:20:58	no	J Real Estate Finan Econ. 2022 Dec 2;:1-35	utf-8	null	null	null	oa_other
==== Front J Solgel Sci Technol J Solgel Sci Technol Journal of Sol-Gel Science and Technology 0928-0707 1573-4846 Springer US New York 5988 10.1007/s10971-022-05988-6 Original Paper: Industrial and technological applications of sol–gel and hybrid materials Industrial applications of sol–gel derived coatings http://orcid.org/0000-0001-8997-7579 Lu Songwei [email protected] 1 Shao Jiping 2 Wu Fanghui 2 1 grid.427187.f 0000 0001 0491 2267 Coatings Innovations Center, PPG Industries, Inc., 4325 Rosanna Drive, Allison Park, PA 15101 USA 2 grid.427187.f 0000 0001 0491 2267 Monroeville Business and Technology Center, PPG Industries, Inc., 440 College Park Drive, Monroeville, PA 15146 USA 2 12 2022 111 26 8 2022 2 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Sol–gel derived coatings have many practical applications in different industries. In this paper, HI-GARD® hard coat, multi-layer antireflective coatings, and an anti-glare coating with organic particles are described. Optical and mechanical performances of these coatings are discussed in addition to adhesion properties. The HI-GARD® hard coat was dip or spin coated from a sol by hydrolyzing alkoxysilanes with water in an acidic condition. The hard coat acts as a protective coating for optical lenses with excellent optical properties with a Bayer ratio of 4.8 and an adhesion of 5B. The multi-layer antireflective coatings were prepared by incorporating titanium oxide sol into the HI-GARD® hard coat solution to obtain different layers by spin-coating with tunable refractive index. These two-layer or three-layer antireflective coatings increase transmittance by at least 3% compared to an uncoated glass substrate. Anti-glare coatings were spray-coated at room temperature on glass substrates by embedding cationic or anionic polystyrene particles in an acid-hydrolyzed silane sol. The anti-glare coating with organic particles can provide a significant glare reduction with a haze value of up to 13% for display surface without sparkling. In addition to these transparent coatings, non-transparent sol–gel derived coatings such as a sol–gel non-stick coating for cookware and bakeware, and two zinc-silicate protective coatings hydrolyzed from a silane with addition of zinc dusts for corrosion protection are also discussed briefly. The schematic structures of the hard coat on CR-39® substrate, three-layer AR coatings on glass substrate, and organic particle embedded anti-glare coatings on glass substrate, and their respective transmittance or reflective curves Highlights HI-GARD® hard coat with excellent scratch resistance, optical properties, and adhesion. Multi-layer anti-reflective coatings with at least 3% decrease of single side reflectance. Anti-glare coatings with organic particles coated at room temperature to reduce glare. Keywords Hard coat Antireflective coatings Multi-layer coatings Anti-glare coatings Non-stick coatings Protective coatings ==== Body pmcIntroduction Sol–gel-derived transparent functional coatings have many industrial applications in recent years due to their superior optical [1–5], mechanical [6, 7], electrical [8], and functional properties [4, 9]. Kajioka et al. achieved low sparkling in anti-glare spray coatings by controlling the size of polymerized species in silica sols with the transmittance haze as high as 12.38% [1]. However, no mechanical durability is reported, which is a requirement for anti-glare coatings on touch screen displays. By adjusting the spray parameters, Huang et al. were able to achieve a sol–gel anti-glare coating with a high haze of 57.0% and a gloss of only 9.2 gloss unit [2]. Wang et al. used a double-layer film to combine the anti-reflective performance with self-cleaning performance using anti-reflective hydrophobic coatings for PV modules [4]. The water contact angle of the surface layer is 150° in the super-hydrophobic state, and the transmittance increases by 5% compared to uncoated glass. Although it is claimed that the coatings have excellent durability, the coating showed scratches with 4H pencil hardness due to its weak surface nanostructures. Zhang et al. prepared ladder-like, cage-like, and/or partial cage-like structures of cycloaliphatic epoxy-functionalized oligosiloxanes (CEOS) by a sol–gel method and incorporated them in a coating followed by a thermal curing [6]. The coating is transparent with a transmittance of 94.9% at 500 nm, flexible, and high pencil harness at 9H on PET substrate. However, in order to get 9H pencil hardness, a thermal curing process of 25 h is necessary which is too long for most production processes. Wu et al. investigated a sol–gel hard coat on polycarbonate substrate concluding that the main factors towards improved pencil scratch resistance are layer thickness, elastic modulus, fracture toughness and intrinsic hardness of the coating material [7]. Pencil hardness is increased from grade 2B to 5H by adjusting these parameters. A good review of sol–gel coatings in the energy sector was recently given by Kaliyanna et al. [10]. Minami reported advanced sol–gel coatings for practical applications including protective coating on metal sheets using methyltrialkoxysilane-derived films, micropatterning on glass substrates, water-repellent coatings for windshields, colored coatings on glass bottles for easy recycling, superhydrophobic and superhydrophilic coatings on glass substrates, and antireflective coatings on glass lenses [11]. Due to the recent work-from-home environment during the COVID-19 pandemic, the needs for smartphones, tablets, and touch screen laptop computers have skyrocketed, as have the needs for transparent functional coatings on these devices. These transparent functional coatings include hard coat, easy-to-clean coatings, anti-glare coatings, antireflective coatings, anti-fingerprint coatings, and anti-static coatings. Besides the optical and functional properties required by these coatings on touch screens, mechanical performance is another requirement that must be met for this application. For example, 9H pencil-hardness hard coating has been applied on plastic substrates for flexible displays [12]. In addition to meet these performance specifications, ease of coating material manufacturing, coating application, and curing are also key requirements for successful practical industrial applications. Furthermore, the use of plastic eyewear substrates has opened a wide door for sol–gel coating application on optical lenses by dip-coating and spin-coating processes. Sol–gel derived hard coat and antireflective coating are common optical coatings in addition to sputtering coatings under vacuum. On the other hand, sol–gel non-transparent coatings have applications in the areas where optical transparency is not a critical factor, e.g., corrosion protection and protective coatings. PPG Industries, Inc. has developed easy-to-clean (EC) coatings for easy-to-clean and silky touch features, along with scratch resistance to protect the uncoated glass substrates; anti-glare (AG) coatings to reduce surface glare to provide better readability under direct light or outdoor without sparkling; antireflective (AR) coatings to reduce surface reflection; and anti-fingerprint (AFP) coatings to reduce fingerprint visibility and better fingerprint cleanability [13]. In this paper, a hard coat on optical lenses, a multi-layer AR coating stack for glass substrates, and an anti-glare coating with organic particles will be described in detail. The non-transparent sol–gel coatings for non-stick coating application, and corrosion protection will also be described briefly. Experimental HI-GARD® hard coat A typical proprietary hard coat solution preparation is described as following: a mixture of alkoxysilanes was added to a suitable vessel such as a 1 L glass flask, and a catalytic amount of an acidic catalyst was added under stirring. Within 10 min, the exotherm generated from the hydrolysis of the silanes caused an increase in the temperature of the reaction mixture from ~25 to >50 °C. The mixture was stirred for an additional hour while cooling to about 30 °C. A 50/50 weight ratio of alcohol and ether solvents was added to the mixture and stirred. Then an additive, such as polyether-modified polysiloxane copolymer in solvent was added at a low level, and the resulting mixture was stirred for at least 30 min. The resulting coating solution was filtered through a nominal 0.45 µm capsule filter and stored at 4 °C until use. The hard coat can be easily coated on substrates using standard dip and spin coating processes. A typical application by dip coating is described as following: uncoated optical lenses made of CR-39® monomer or other plastic materials, available from PPG Industries, Inc. were cleaned with different concentrations of caustic cleaning solutions twice under ultrasonics for 5 min for each cleaning. Then the lenses were rinsed with deionized water for three times with each rinse lasting for 5 min before drying at 70–75 °C; the dried lenses were cooled at room temperature for 5 min before the hard coat was applied by a dip-coater. The withdrawal rate used was ~15 cm/min to achieve a cured coating thickness of 3–5 µm. Afterwards, the lenses coated with the hard coat were dried and cured in an air circulating oven for 3 h at 120 °C. Similarly, optical lenses can be coated via spin coating with a spinning speed of 110 rpm, resulting in a coating thickness of ~2–3 µm. The coating can then be cured at 120 °C for 3 h. Multi-layer antireflective coatings Preparation of titanium oxide sol [14] A sol–gel titanium oxide sol was prepared in a 1 L plastic bottle by slowly adding 440 g of a 6.4% nitric acid aqueous solution to a mixture of 200 g of titanium iso-propoxide (97%, sigma Aldrich), 100 g of DOWANOL™ PM (Propylene glycol mono methyl ether, ≥99.5%, Dow Chemical Company), and 100 g of DOWANOL™ PMA (Propylene glycol methyl ether acetate, ≥99.5%, Dow Chemical Company) under vigorous stirring. White precipitates formed upon adding the acidic solution, and after stirring for 24 h the hydrolyzed solution peptized and became transparent without precipitation or gelation. The titanium oxide sol was stored in a freezer at −20 °C for future use. Preparation of low refractive index sol–gel solution for a coating with n = 1.44 ± 0.01 The low refractive index sol–gel solution for a coating with n = 1.44 ± 0.01 was prepared using tetraethyl orthosilicate (98% purity, Sigma-Aldrich Corporation), a silane functional acrylic polymer [15] with modification, colloidal silica MT-ST (30 wt% colloidal silica mono-dispersed in methanol, Nissan Chemical), methyltrimthoxysilane (Evonik), 2-propanol, HNO3, deionized water, and BYK-306 (12.5 wt% polyether-modified polydimethylsiloxane in xylene and monophenyl glycol, BYK USA Inc.). The details of the preparation can be found in Example 5 of referenced patent application [15]. Preparation of sol–gel solutions for sol–gel derived coatings with tunable refractive index In order to produce multi-layer AR coatings, there should be at least two to three layers with different refractive indices in the final films. It is critical to have appropriate stable sol–gel solutions available. One way to obtain sol–gel films with different refractive indices is to combine one sol–gel solution for high refractive index film with another sol–gel solution for low refractive index film. Figure 1 shows the tunable refractive index of the final films after blending the titanium oxide sol in Section 2.2.1. with a HI-GARD® hard coat sol from PPG Industries, Inc. as mentioned above in Section 2.1. The hard coat sol itself can produce a film with a refractive index of 1.51. Various refractive indices ranging from 1.63 to 1.91 can then be obtained after properly blending by controlling the weight ratio of titanium oxide sol to hard coat sol, thus giving the wider options to design multi-layer AR coatings based on the tunable refractive indices.Fig. 1 Tunable refractive index of the final films after blending hard coat sol and titanium oxide sol. The x axis is the weight ratio of titanium oxide sol to the hard coat sol Two-layer antireflective (AR) coating stack by spin coating [14] A soda lime glass substrate was cleaned with isopropyl alcohol and pre-treated with nitrogen plasma treatment for 15 min using an ATTO™ plasma treater (Diener Electronics, Germany). The first layer consisting of a coating from the solution for a refractive index of 1.68 ± 0.02 was coated using a Cee™ 200X spin-coater (Cost Effective Equipment, LLC.) on the substrate, followed by curing at 150 °C for 1 h. The film thickness was adjusted by using different spin speeds (500–2000 rpm), and optimal thickness was in the range of 100–140 nm. The second layer consisting of a coating from the solution for a refractive index of 1.44 ± 0.01 was spin coated on top of first layer after nitrogen plasma treatment for 15 min. The thickness of the second layer was in the range of 70 to 90 nm. Three-layer antireflective (AR) coating stack by spin coating [14] A soda lime glass substrate was cleaned by isopropyl alcohol, and treated with nitrogen plasma treatment for 15 min using an ATTO™ plasma treater (Diener Electronics, Germany). A coating from the solution for a refractive index of 1.78 ± 0.02 was spin coated on the glass substrate as the first layer. The coating was then dried at 80 °C for 20 min. A second coating from the solution for a refractive index of 1.97 ± 0.03 was spin coated as the second layer followed by drying at 80 °C for 20 min. Thereafter, a coating from the solution for refractive index 1.44 ± 0.01 was coated as the third layer (top layer). The final coating stack was then cured at 150 °C for 1 h. The average film thickness for the first layer, the second layer and the third layer was 50, 83, and 81 nm, respectively. Anti-glare coatings with particles [16] Anti-glare coatings solution can be made by incorporating a dispersion of cationic or anionic polystyrene particles in water into the sol–gel matrix. A detailed procedure for preparing a dispersion of cationic or anionic polystyrene particles in water can be found in a granted patent [16]. 30.0 g of tetraethyl orthosilicate (TEOS) from Sigma-Aldrich Corporation, 17.5 g of deionized water, and 17.5 g of denatured ethyl alcohol were mixed together in a flask with a magnetic stirring bar. 1.8 g of 4.68 wt% aqueous nitric acid solution was added to the above mixture to hydrolyze the TEOS. After 1 h stirring, an additional 29.2 g of ethyl alcohol and 4.0 g of cationic or anionic polystyrene particles in water were added to the hydrolyzed sol. The sol was then keep stirring for 10 min. A glass substrate was treated using a low-pressure plasma system from Diener Electronics, Germany. The above coating solutions were then sprayed on to the glass surface at room temperature using a Spraymation™ and a Binks™ 95 automatic HVLP™ spray gun with a traverse speed of 600 in./min followed by a curing at 150 °C for 1 h. Coating characterizations The optical characteristics such as color and % haze of hard coat were measured with a Hunter UltraScan™ PRO. The optical transmittance and reflectance spectra and color of AR coatings were measured using a Lambda™ 9 optical spectrophotometer from 350 to 800 nm. The single side reflection was measured by eliminating the backside reflection by placing a 3M™ black tape on the backside of the sample. The gloss value of the anti-glare-coating coated glass sample was measured using a BYK Gardner™ Micro-Tri-Gloss 20/60/85° gloss meter (BYK-Gardner GmbH, Germany) at 60°. A matte surface of a black paint with a gloss value of <0.5 gloss unit was used under the glass samples to ensure accurate reading. Transmittance, L, a, b, haze, and reflectance of anti-glare coatings were measured using a Color i7 Benchtop Spectrophotometer (X-Rite, Inc., Grand Rapids MI). The abrasion resistance of hard coat was evaluated via Bayer testing by measuring the haze of both coated and uncoated lenses using a Hunter UltraScan™ PRO after 600 oscillation cycles with 500 g of alundum grits. The Bayer abrasion ratio was calculated using the following formula (Eq. (1)), with the “reference lens” being an uncoated CR-39® as Bayer standard lens.1 BayerRatioTestLens=Hazefinal−HazeinitialAvereageCR−39Refs.Hazefinal−HazeinitialTestLens The scratch resistance of hard coat was measured using a steel wool test. Optical lenses were subjected to rubbing by selected steel wool under a specified weight, e.g., 1 lb, and a specified number of cycles, e.g., 200 cycles. The haze can be measured before and after rubbing but giving a qualitative rating from poor to excellent is common since the scratches could be localized. Pencil hardness of anti-glare coatings was measured using an HA-3363 Gardco™ Pencil Scratch Hardness Kit (Paul N. Gardner Company, Inc., Pompano Beach, FL) with a 500 g load and a Derwent graphic pencil. The Erichsen™ hardness of a hybrid non-stick coating was evaluated using an Erichsen™ hardness test pencil model 318S. A dry reciprocating abrasion test was conducted for non-stick coatings to simulate the effect of scraping by spatulas and other cooking utensils. A 2-in. abrasive pad (3M Scotch-Brite 07447) was mounted on a 3 kg armature, which was cycled for 1000. The abrasive pad was replaced with a fresh pad every 1000 cycles, and the test continued until 10% of the abraded area has been exposed to bare metal. The adhesion of hard coat on substrates can be evaluated with the ASTM D3359 Tape Test or Crosshatch Adhesion Test. First the coated substrate is scored with a crosshatch tool twice to form a grid. Then a Scotch™ tape is applied to the grid and quickly removed. For coatings on optical lens, adhesion is evaluated for cured samples at room temperature, and also for the samples submerged in boiling water for 15 min to 3 h. For anti-glare coatings, the crosshatch test was conducted only at room temperature for cured coatings. Optical profilometry analysis and surface roughness Ra were carried out using a Veeco™ WYKO NT3300 Optical Profilometer. SEM cross-section and top morphology images were taken from a LEO 1530 Field Emission SEM w/Noran™ EDX System. Results and discussion HI-GARD® hard coat performances Optical performance The hard coat has excellent optical performances in terms of neutral color, high transmittance, and no haze. Figure 2 shows the transmittance spectra of the hard coat coated CR-39® and uncoated CR-39® substrate. Their spectra are almost identical. The integrated transmittance from 450 nm to 800 nm is about 92% for both coated and uncoated CR-39® substrates. The refractive index of the hard coat of 1.51 matches with that of 1.50 for CR-39® very well. One can also see from Fig. 2 that the CR-39® substrate has significant low transmittance below 380 nm, indicating an excellent UV protection for human eyes. After the coating process, the color data is almost identical to the substrate with slightly higher b, which is also confirmed from the yellowness index (YI) with a slightly higher value of 0.76 vs. that of 0.41 for an uncoated CR-39® substrate (Table 1). It is acceptable for vision application with yellowness index as ≤1.0. There is no haze for the coated sample, compared to an industrial acceptable standard of less than 0.5% of haze value for optical eyewear applications.Fig. 2 Transmittance spectra of the hard coat coated CR-39® (solid line) vs uncoated CR-39® substrate (dotted line) Table 1 Optical performance data of the hard coat coated CR-39® vs uncoated CR-39® substrate Sample L* a* b* YI Haze X Y Z CR-39® substrate 96.92 −0.08 0.26 0.41 0.12 87.41 92.25 98.56 Hard coat coated CR-39® 96.89 −0.15 0.47 0.76 0.03 87.31 92.18 98.21 The optical performance data include L, a, and b* from CIBLAB color space, X, Y, Z from CIE 1931 color space, yellow index (YI), and haze Scratch resistance The hard coat has not only excellent optical performance, but also excellent mechanical performance to protect the eyewear substrates (Fig. 3). The scratch resistance of the hard coat can be measured with the Bayer test and steel wool test as described in the experimental section. The Bayer test results indicate that the Δ% Haze is 24.0% for the uncoated CR-39® reference lens before and after the Bayer test (Fig. 3a left). The Δ% Haze is only 5.0% for CR-39® lens coated with the hard coat before and after the Bayer test (Fig. 3a right). According to Eq. (1), the Bayer ratio is calculated as 4.8. The steel wool testing results show enhanced scratch resistance when the plastic lens material is protected with the hard coat, with several scratches on the uncoated CR-39® reference lens but no scratches on the hard coat coated CR-39® lens (Fig. 3b). The pencil hardness of the hard coat is 3H. The scratch resistance and pencil hardness are very similar to hard coatings with similar thickness on CR-39® lens prepared based on the nanocomposite of Al2O3–ZrO2 nanoparticles in a sol–gel matrix using 3-glycidoxypropyltrimethoxysilane (GPTMS) and TEOS by Chantarachindawong et al. [5]. However, it seems that the addition of Al2O3–ZrO2 nanoparticles does not show significant benefits for enhancing scratch resistance, e.g., pencil hardness. Only the cracked coating example with 6 μm thickness showed a pencil hardness of >5H [5]. Wu et al. reported that pencil hardness shows continuous improvement to three higher grades with increasing silica content up to 30.5 vol%, but remains stable till 36 vol% silica content [7]. In the current work, the HI-GARD® hard coat has excellent mechanical performance without the addition of alumina, silica, or zirconia nanoparticles.Fig. 3 Mechanical performance of the hard coat a left: uncoated CR-39® reference lens with Δ% Haze = 24.0% before and after Bayer test; and right: CR-39® lens coated with the hard coat with Δ% Haze = 5.0% before and after Bayer test; b left: uncoated CR-39® reference lens after scratch testing with steel wool; and right: CR-39® lens coated with the hard coat after scratching testing with steel wool Hard coat adhesion The hard coat also has excellent adhesion on CR-39® substrates (Fig. 4). The adhesion on CR-39® substrates is 5B for samples before and after further boiling water immersion for up to 1 to 3 h.Fig. 4 CR-39® lens coated with the hard coat after cross-hatch adhesion testing with Scotch™ 600 tape showing no coating peeling off after the coated lens was boiled in hot water for 30 min Optical performances of two-layer AR coatings The single side reflection spectra of two samples of two-layer AR coatings are shown in Fig. 5 in comparison to uncoated glass. The reflection curves have a lowest value around 600 nm. This position of the lowest value can be shifted by adjusting the thickness of each layer. These two samples of two-layer AR coatings have a very slight difference in the reflection curve. Table 2 lists the optical data including color of these two samples in comparison to the uncoated glass. The integrated transmittance of the AR coatings increased by around 3% from the uncoated glass, which is corresponding to the decrease of front reflection by a similar value. The transmittance color of the AR coatings is slightly yellowish giving the positive b* value. In contrast, the reflective color of the AR coatings is slightly blueish giving the significant negative b* value.Fig. 5 Single side reflection spectra of two-layer AR coating samples (solid line and dotted line) in comparison with an uncoated glass (dashed line) Table 2 Optical performance data of two-layer AR coating samples in comparison with an uncoated glass T Front R Sample name T (%) Single side R (%) Front R (%) Back R (%) L* a* b* L* a* b* Uncoated glass 91.83 4.24 8.12 8.12 96.75 −0.03 0.14 34.26 −0.09 −0.59 Two-layer AR 1 94.85 0.95 4.97 5.00 97.93 0.03 1.53 27.00 −0.08 −9.18 Two-layer AR 2 94.84 0.98 5.01 4.99 97.92 −0.07 1.36 27.09 0.38 −8.69 Optical performances of three-layer AR coatings The single side reflection spectra of two samples of three-layer AR coatings are shown in Fig. 6 in comparison to the uncoated glass. Comparing to two-layer AR coatings, the spectra show two valleys, with one located in the 370–380 nm range while the second one is located in the 570–590 nm range. The curve forms a W-shape for three-layer AR coatings (Fig. 6), while it is more a V-shaped for two-layer AR coatings (Fig. 5). The W shape can be changed and the valley locations can be shifted by adjusting the thickness of each layer. Table 3 lists the optical performance data of these three-layer AR coatings. It can be seen that these three-layer AR coatings have slightly higher transmittance, and lower reflectance than the two-layer AR coatings. The three-layer AR coatings may be further optimized to reduce the single side reflection to less than 0.5% or even 0.3%. For these two particular three-layer AR coating samples, their transmittance color is neutral due to b* value of less than 1, while their reflective color is slightly blue due to slightly negative b* value (Table 3).Fig. 6 Single side reflection spectra of three-layer AR coating samples (solid line and dotted line) in comparison with an uncoated glass (dashed line) Table 3 Optical performance data of three-layer AR coating samples in comparison with an uncoated glass T Front R Sample name T (%) Single side R (%) Front R (%) Back R (%) L* a* b* L* a* b* Uncoated glass 91.88 4.23 8.11 8.12 96.77 −0.05 0.15 34.23 −0.06 −0.60 Three-layer AR 1 95.17 0.60 4.71 4.73 98.07 −0.05 0.72 26.09 0.39 −5.40 Three-layer AR 2 95.20 0.59 4.73 4.72 98.10 −0.22 0.53 26.08 0.80 −3.72 Antiglare coatings performances Traditionally, the anti-glare display surface can be formed from an anti-glare film from 3M™ for plastic substrates, or etched glass surface in the case of glass substrates. An anti-glare film may have weak scratch resistance and low pencil hardness on plastic substrates. The etched glass has significant environmental issues using hydrofluoric acid or similar acid as etchant. The by-products during the glass etching process include hexafluorosilicic acid (H2SiF6) which can release hydrogen fluoride when evaporated, posing further environmental, health, and safety concerns. On the other hand, anti-glare coatings using the sol–gel process do not generate harmful hydrogen fluoride gas. Figure 7 shows the SEM surface morphologies of three commercial anti-glare surfaces (a) anti-glare film, (b) anti-glare surface by etching process, and (c) anti-glare coatings from PPG Industries, Inc. [13] While the surface morphologies of these samples are all different with (a) consisting of particles, (b) bowl-shaped dents, and (c) random hills and valleys, the principle of the anti-glare surface is the same as using roughed surface features to scatter the light, resulting in significant glare reduction.Fig. 7 SEM images of a commercially available anti-glare film (a), an anti-glare glass substrate by etching process (b), and an anti-glare coating by sol–gel spray process (c) The commercially available anti-glare coatings from PPG Industries, Inc. usually require a pre-heated glass substrate [13]. In order to further reduce energy consumption and operational cost for pre-heating process, anti-glare coatings with embedded polystyrene particles have been developed without the pre-heating process. The cationic or anionic polystyrene particles were first prepared and their dispersion was characterized by high resolution TEM. Typical dispersed particles are shown in Fig. 8(a) and (b). The mean particle size is 290–300 nm with the particle size ranging from 150 nm to 500 nm (Fig. 8c). When these particles are incorporated in the anti-glare coatings on a Kapton® tape as a substrate, the particles are visible from the cross-sectional HRTEM image (Fig. 9). The surface structure from the anti-glare coatings using particles can be seen in optical profilometry images (Fig. 10). The transmittance curve shown in Fig. 11 indicated that the low haze sample (e.g., haze 6.3%) did not change much of the transmittance curve of the anti-glare coatings, while the high haze sample (e.g., haze 12.2%) has a decrease of about 5% transmittance compared to the uncoated gorilla glass. The haze of 12.2% is comparable with the reported haze of sample S4 from Kajioka et al. [1]; however, the sample in this work has a higher 60° gloss value. The discrepancy could be caused by different structures in the anti-glare coatings, i.e., polystyrene particles vs polymerized species. The haze value is much lower than the reported highest haze of 57.0% [2].Fig. 8 Typical dispersed cationic or anionic polystyrene particles a low magnification, bar 1 μm; b high magnification, bar 100 nm, c Particle size distribution of typical sample with mean particle size of 292 nm and size ranging from 150 to 500 nm Fig. 9 High resolution TEM cross-sectional image of anti-glare coatings on a Kapton® tape Fig. 10 Optical profilometry images (2.5× magnification) of anti-glare coatings for a 73 gloss unit anti-glare coatings with Ra = 77 nm; b 61 gloss unit anti-glare coatings with Ra = 106 nm; and c 46 gloss unit anti-glare coatings with Ra = 127 nm Fig. 11 Transmittance of anti-glare coatings in comparison with the uncoated Gorilla glass substrate Table 4 lists the optical property, adhesion, and pencil hardness of anti-glare coatings incorporating polystyrene particles. The relationship between haze and 60° gloss value is shown in Fig. 12. When haze decreases, the 60° gloss value increases. Usually, a 70 ± 10 gloss unit will give rise to a 50 ± 10% glare reduction. Depending on the desired extent of the gloss reduction, various 60° gloss values can be obtained by adjusting the spray parameters while still keeping adhesion at 5B and pencil hardness at ≥8H. Due to the roughness of the anti-glare surface, there is a possibility of sparkling from the display with high degree of roughness. Sparkling is a result of the prism effect from the interaction of light from a regular display pixel matrix and anti-glare irregular microstructured surfaces. Displays with higher pixel per inch (PPI) have a higher possibility of sparkling. A higher haze and low gloss anti-glare coating will result in higher possibility of sparkling. Figure 13 illustrates the dependence of sparkling on the surface roughness Ra and PPI, with non-sparkling occurring when Ra is less than 55 nm depending on PPI. This result is contradictory to the conclusion from Kajioka et al.’s work, which indicated that larger polymerized species led to lower gloss, higher haze, and lower sparkle [1]. Sparkling effect wasn’t discussed for the highest haze of 57.0% from a sol–gel coating [2].Table 4 Optical properties, adhesion, and pencil hardness of anti-glare coatings incorporating polystyrene particles Sample Gloss (GU) Haze (%) T% at 550 nm a* b* L* Cross hatch adhesion Pencil hardness 1 75.7 6.41 91.46 −0.22 0.11 96.56 5B 9H 2 70.0 6.95 91.15 −0.20 0.08 96.45 5B 9H 3 64.7 10.90 89.85 −0.23 −0.03 95.89 5B 9H 4 64.0 11.63 89.75 −0.22 0.02 95.86 5B 9H Fig. 12 The relationship of haze vs 60° gloss of the anti-glare coatings Fig. 13 The dependence of sparkling phenomenon on surface roughness Ra for different display resolutions PPI values Other industrial applications of sol–gel derived coatings Non-stick coatings for cookware and bakeware A sol–gel-derived non-stick coating for cookware and bakeware has been commercialized using a two-part coating composition [17]. The sol–gel coating composition is based on one or more organoalkoxysilanes such as methyltrimethoxysilane, an acid catalyst for hydrolysis reaction, water, organic polymer particles such as polyphenylene sulfide (PPS), and fillers. The coated panel is then cured above 200 °C, such as at 327 °C for 30 min. When organic polymers are added to the sol–gel precursor mixtures, significant improvements in hardness and abrasion properties can be obtained and preserved up to temperatures close to the melting point, glass-transition temperature, or heat deflection temperature of the chosen polymer. The Fusion™ waterborne sol–gel coatings available from PPG Industries, Inc. is a hybrid organic inorganic coating with good release, stain resistance, and high gloss. An example of a hybrid coating composition is listed in Table 5 in comparison to a conventional coating composition. The key difference is the addition of polyphenylene sulfide organic particles in the hybrid sol–gel coating composition, which significantly increase the mechanical performances, such as a doubling of the Erichsen™ hardness compared to the conventional composition, and more than a tripling of the performance in the dry reciprocating abrasion test (Table 6).Table 5 Compositional summary of a conventional sol–gel coating and a hybrid sol–gel coating composition for non-stick coatings for cookware and bakeware [18] Component Conventional coating composition (wt%) Hybrid Coating Composition (wt%) Methyltrimethoxysilane 31.2 33 Silicone fluid 1.16 1.2 Polyphenylene sulfide 0 6.7 Colloidal silica (30%) 43.2 45.2 Black pigments 2.17 2.92 Acid catalyst 0.66 0.7 Pigment dispersant 0.5 1.15 Defoamer 0.58 0.52 Silicon carbide 1 0.98 Dipropylene glycol methyl 6.6 6.4 Other fillers 12.93 1.23 Total 100 100 Table 6 Mechanical performance comparison of conventional sol–gel coating vs. hybrid sol–gel coatings for non-stick coating applications Erichsen™ hardness (Newtons) Dry reciprocating abrasion test (cycles/mil) Conventional sol–gel coating 8 25,000 Hybrid sol–gel coating 18–20 87,000 Zinc rich sol–gel primer and Aquaweld™ 100 waterborne shop primer A zinc-rich sol–gel primer is developed for protective and marine coatings, which provides galvanic protection to steel in corrosive environments. The zinc rich primer is based on the hydrolysis of a silane and addition of zinc dusts to form a zinc–silicate matrix for the coatings. After solvent evaporation, the zinc silicate primer is cured by absorption of ambient moisture over 16–24 h. The corrosion protection mechanism is that the steel structure acts as the cathode, while the zinc dust in the primer is the sacrificial metal anode in this electrochemical system. When exposed to an electrolyte such as sea water, the zinc in the primer will corrode preferentially, thus protecting the steel galvanically. In a well-designed coating, the zinc salts that are formed when the zinc corrodes will eventually fill the voids in the zinc primer, and the protection mechanism slowly shifts from galvanic protection to barrier protection. In addition to corrosion resistance, the zinc-silicate coating also has excellent solvent resistance and heat resistance. A different two-part zinc-silicate waterborne primer called Aquaweld™ 100 has also been developed as a shop primer for pre-construction and pre-fabrication which provides temporary anti-corrosion performance for steel substrates [18]. The two-part sprayable shop primer is an acid stabilized sol–gel coating with addition of zinc dusts and other pigments, forming a moisture curing zinc-silicate network. In general, the shop primer has pot life of about 4 h for application. This red-brown shop primer has good workability through automated spray lines. It has excellent corrosion resistance, cutting properties, and weldability. Summary PPG Industries Inc. has developed and/or commercialized several sol–gel derived coatings for different applications. The HI-GARD® hard coat has excellent optical and mechanical properties for ophthalmic applications with excellent adhesion on PPG Industries, Inc.’s CR-39®, TRIVEX®, and TRIBRID® substrates and other commercial lens substrates. The hard coat sol can also be blended with a high index titanium oxide sol to form thin films with various refractive indices. Multi-layer antireflective coatings, such as two-layer or three-layer antireflective coatings can be spin-coated from coating solutions with desired refractive index on glass substrates. These antireflective coatings will have at least 3% increase in transmittance from 350 nm to 800 nm. The anti-glare coatings with cationic or anionic polystyrene particles scatter light to result in a glare reduction without sparkling by controlling the surface roughness Ra. The anti-glare coatings have a crosshatch adhesion of 5B and a pencil hardness of ≥8H. A sol–gel derived non-stick coating has significant improvements in hardness and abrasion properties. Two sol–gel derived zinc-silicate coatings are used as primers on steel substrates to prevent corrosion. Acknowledgements The authors would like to acknowledge experimental results from David C. Martin, Zilu Li, and Xiangling Xu, TEM, SEM, and particle size analyses by PPG Industries, Inc. analytical team, and valuable discussions with Kees Van Vliet, James E. McCarthy, Matteo Sperindio, Vincent Pagnotti, Matthew Luchansky, and Elizabeth Horner, all from PPG Industries, Inc. The authors also would like to thank Michael Buchanan for optical spectrum measurement for antireflective coatings, and Nathaniel Hazelton for optical profilometry measurement for anti-glare coatings, both from Vitro Architectural Glass. Attributions: The PPG Industries, Inc. Logo, CR-39 and Hi-Gard, Trivex and Tribrid are registered trademarks of PPG Industries Ohio, Inc. Fusion is a trademark of PPG Industries Australia Pty Ltd. Aquaweld is a trademark of PPG Coatings Nederland B.V. All company names and third-party marks appearing in this article are property of their respective owners. © 2022 PPG Industries, Inc. All Rights Reserved. Author contributions All authors contributed to part of the experimental and results of the manuscript. The first draft of the manuscript was written by SL and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Compliance with ethical standards Conflict of interest The authors declare no competing interests. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Kajioka T Kanai T Ikegami K Kozuka H Achievement of low sparkle in anti-glare spray coatings by controlling the size of polymerized species in silica sols J Sol–Gel Sci Technol 2021 100 244 251 10.1007/s10971-021-05664-1 2. Huang YH Chen LC Chou HM Optimization of process parameters for anti-glare spray coating by pressure-feed type automatic air spray gun using response surface methodology Materials 2019 12 751 10.3390/ma12050751 30841570 3. Abe K Sanada Y Morimoto T Anti-reflective coatings for CRTs by sol–gel process J Sol–Gel Sci Technol 2003 26 709 713 10.1023/A:1020737902758 4. Wang P Yan X Zeng J Luo C Wang C Anti-reflective superhydrophobic coatings with excellent durable and self-cleaning properties for solar cells Appl Surf Sci 2022 602 154408 10.1016/j.apsusc.2022.154408 5. Chantarachindawong R Osotchan T Chindaudom P Srikhirin T Hard coatings for CR-39 based on Al2O3–ZrO2 3-glycidoxypropyltrimethoxysilane (GPTMS) and tetraethoxysilane (TEOS) nanocomposites J Sol–Gel Sci Technol 2016 79 190 200 10.1007/s10971-016-4006-3 6. Zhang K Huang S Wang J Liu G Transparent organic/silica nanocomposite coating that is flexible, omniphobic, and harder than a 9H pencil Chem Eng J 2020 396 125211 10.1016/j.cej.2020.125211 7. Wu LYL Chwa E Chen Z Zeng XT A study towards improving mechanical properties of sol–gel coatings for polycarbonate Thin Solid Films 2008 516 1056 1062 10.1016/j.tsf.2007.06.149 8. Aegerter MA Reich A Ganz D Gasparro G Piitz J Krajewski T Comparative study of SnO2:Sb transparent conducting films produced by various coating and heat treatment techniques J Non-Cryst Solids 1997 218 123 128 10.1016/S0022-3093(97)00134-8 9. Esfahani MB Eshaghi A Bakhshi SR Transparent hydrophobic, self-cleaning, anti-icing and anti-dust nano-structured silica based thin film on cover glass solar cell J Non-Cryst Solids 2022 583 121479 10.1016/j.jnoncrysol.2022.121479 10. Kaliyannan GV Palanisamy SV Priyanka EB Thangavel S Sivaraj S Rathanasamy R Investigation on sol–gel based coatings application in energy sector—a review Mater Today: Proc 2021 45 1138 1143 11. Minami T Advanced sol–gel coatings for practical applications J Sol–Gel Sci Technol 2013 65 4 11 10.1007/s10971-011-2572-y 12. Choi GM Jin J Shin D Kim YH Ko JH Im HG Jang J Jang D Bae BS Flexible hard coating: glass-like wear resistant, yet plastic-like compliant, transparent protective coatings for foldable displays Adv Mater 2017 29 1700205 10.1002/adma.201700205 13. Lu S Shao J Martin DC Li Z Schwendeman I Commercialization of sol–gel based transparent functional coatings J Sol–Gel Sci Technol 2018 87 105 112 10.1007/s10971-018-4694-y 14. Lu S, Shao J, Li Z (2019) Multi-layer anti-reflective coated articles. US 2019/0033491 A1 15. Basil JD, Hunia RM, McGrady LB (2013) Polysiloxane coating with hybrid copolymer. US 8,507,631 B2 16. Lu S, Vanier N, Xu X, Martin DC, Olson KG, Schwendeman I (2020) Curable film-forming Sol–gel compositions and anti-glare coated articles formed from them. US 10,723,890 B2 17. Sperindio M, Lake SC, Piras R, Lindstrom MJ (2021) Sol–gel compositions with improved hardness and impact resistance. US 11,193,021 B2 18. Plehiers M, Van Loon S (2013) Binder Composition. US 2013/0289190 A1	36475095	PMC9716534	NO-CC CODE	2022-12-03 23:20:58	no	J Solgel Sci Technol. 2022 Dec 2;:1-11	utf-8	J Solgel Sci Technol	2,022	10.1007/s10971-022-05988-6	oa_other
==== Front Empir Econ Empir Econ Empirical Economics 0377-7332 1435-8921 Springer Berlin Heidelberg Berlin/Heidelberg 2335 10.1007/s00181-022-02335-0 Article Observed and expected interest rate pass-through under remarkably high market rates http://orcid.org/0000-0001-7359-7539 Divino Jose Angelo [email protected] 1 http://orcid.org/0000-0002-3710-2348 Haraguchi Carlos 2 1 grid.411952.a 0000 0001 1882 0945 Catholic University of Brasilia, QS 07 Lote 01 EPCT, Office K-245, Brasilia, DF 71966-700 Brazil 2 grid.466507.6 0000 0001 2163 7594 Central Bank of Brazil, Brasilia, DF Brazil 2 12 2022 144 12 7 2022 16 11 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This paper investigates the pass-through from observed and expected policy interest rates to the remarkably high lending rates in the Brazilian economy, accounting for financial-institution specific characteristics, borrower types, asymmetric adjustment and persistence in loan rates. We use a unique and non-public dataset with expected variables identified by professional forecasters and apply a fixed-effects approach to alternative specifications as robustness checks. Financial institutions correctly forecast the next target level of the policy rate and anticipate adjustments in their loan rates. There is evidence of over-proportional and positively asymmetric pass-through to loans with higher interest rate margins, implying a positive correlation between degrees of pass-through and spreads across persistent lending rates. These findings contribute to explain why loan interest rates are so high in the Brazilian economy. Keywords Interest rate pass-through Asymmetry Lending rates Monetary policy JEL Classification E43 E44 E52 http://dx.doi.org/10.13039/501100003593 Conselho Nacional de Desenvolvimento Científico e Tecnológico 302632-2019-0 Divino Jose Angelo http://dx.doi.org/10.13039/501100002322 Coordenação de Aperfeiçoamento de Pessoal de Nível Superior 88887.201766/2018-00 760/2018 Haraguchi Carlos http://dx.doi.org/10.13039/501100005668 Fundação de Apoio à Pesquisa do Distrito Federal 00193-00001157/2022-10 Divino Jose Angelo ==== Body pmcIntroduction The degrees of pass-through from the policy interest rate to lending and deposit rates unveil the transmission channel of the monetary policy to the financial sector of the economy. In an ideal scenario, changes in the policy (or base) interest rate should be completely transmitted to the market rates in a full and symmetric pass-through environment, characterizing the efficiency of the monetary policy to affect the market rates, and so the real side of the economy through the credit channel. However, in practice, this might not be the case, as the degree of interest rate pass-through might be either smaller or bigger than one, featuring an incomplete or over-proportional pass-through, respectively. In addition, the pass-through might be asymmetric, meaning that increases or decreases in the policy rate are conveyed in different proportions to the market rates. As a result, the monetary policy might not affect the market interest rates, domestic credit and economic activity as desired. The issue is of special concern in the Brazilian economy, which historically has very high loan interest rates in both nominal and real terms. Despite the lowest level of 2% per year achieved in 2020 as a monetary policy response to stimulate the economy during the COVID-19 pandemic, market rates have remained at very high levels and not followed the downward bias in the policy interest rate. This pattern has yielded very high interest rate spreads for the banking sector. The explanations include specific features of the financial market, such as high probability of default by borrowers, market power by banks, concentration in the financial sector and poor institutional quality. These assertions, however, sound misleading as long as they only focus on the interest rate margins (or spreads) and overlook elements from the interest rate pass-through. We argue that consistent estimates of the degree of pass-through from the observed and expected policy interest rates to different lending rates by financial institutions and borrower types might contribute to fulfil this gap. Specifically, departing from high interest rates margins, an over-proportional pass-through coupled with asymmetric behavior by financial institutions that overreact to increases and under-react to decreases in the policy interest rate, both observed and expected, might sustain the remarkably high loan interest rates in the Brazilian economy. The financial institutions might even anticipate asymmetric adjustments in their lending rates by correctly forecasting future changed in the policy rate. Thus, a complete investigation would require to consider the pass-through form both observed and expected policy interest rate. The objective of this paper is to investigate the interest rate pass-through from the observed and expected policy rates to the remarkably high lending rates in the Brazilian economy. We estimate observed and expected degrees of pass-through by accounting for financial institution specific characteristics, asymmetric behavior and partial adjustment due to persistence in the lending rates. We use a unique and non-public dataset of loan interest rates, observed Over-Selic rate and expected Over-Selic rate identified by professional forecasters (financial institutions) covering the period from January 2012 to April 2019 on a weekly basis, available from the Central Bank of Brazil. The sample is disaggregated by interest rates, financial institutions and loan operations for households and non-financial corporations. In addition to the static panel data estimation, we also allow for partial adjustment in the lending interest rates in a dynamic panel data environment. We apply a fixed effects approach to panels of financial institutions and non-earmarked lending interest rates disaggregated by households and non-financial corporations. The policy interest rate is the Over-Selic rate, which is the monetary policy instrument in the inflation targeting regime adopted by the Central Bank of Brazil since June of 1999. We also use the expected Over-Selic rate identified by professional forecasters to assess whether financial institutions anticipate future changes in the policy rate when setting their loan interest rates. By doing so, they might avoid unexpected losses due to unanticipated changed in the policy rate.1 This unique and non-public dataset with identified expectations reduces loss of information caused by aggregation of expectations by the mean or median, for instance, making the results more reliable. Empirical studies have found asymmetric responses of lending rates (Castro and Mello 2012) and deposit rates (Chong 2010; Hannan and Berger 1991) to downward versus upward movements in policy interest rates. Liu et al. (2008) provided evidences of asymmetric responses in both rates, while Neumark and Sharpe (1992) only for deposit rates of banks in concentrated markets. These findings suggest that rigidity in the pass-through is bigger when there is stimulus for downward movements in lending rates or for upward changes in deposit rates. Market power might affect the banks’ responses to changes in the policy rate, although the effects are unclear in the pass-through literature (Kopecky and Hoose 2012). Hannan and Berger (1991) argued that banks in concentrated markets exhibit higher rigidity in deposit rates, and Holton and d’Acri (2018) found similar results for lending rates. However, while bank concentration is one of the most common indicators of market power, measures of competition are considered more relevant to assess banks’ behavior (e.g., Ornelas et al. 2020; Berger et al. 2004; Cottarelli and Kourelis 1994). Cottarelli and Kourelis (1994) claimed that lack of competition increases stickiness of lending rates and simulations of a DSGE model by Hristov et al. (2014) yielded similar results under weaker competition. Holton and d’Acri (2018) are also in accordance, since large banks (proxy for banks with bigger market power) showed a lower long-run pass-through, especially for small loans (proxy for small and medium sized enterprises). On the other hand, Coelho et al. (2010) suggested that larger Brazilian banks had stronger reactions to the monetary policy than the smaller ones. Conflicting evidences also prevail when assessing the ownership control and capital origin of the banks. Cottarelli and Kourelis (1994) considered a heterogeneous panel of 31 countries from all over the world and found that lending rates appear to be stickier in publicly owned banking systems, and privatizing would substantially increase flexibility of lending rates.2 Using Brazilian data from May 2006 to March 2010, Pereira and Maia-Filho (2013) also obtained lower pass-through for public-owned government banks (GCBs) before the financial crisis, but found no evidence that private banks and GCBs adjusted their lending rates differently afterward. This behavior before the financial crisis contrasts with Coelho et al. (2010), who uncovered similar responses for both types of Brazilian banks in the period of June 2000 to December 2006. Arena et al. (2007) argued that deposit and lending rates of foreign banks are less sensitive to changes in monetary conditions during periods of financial crisis, but Coelho et al. (2010) found that both types of banks displayed similar responses for lending rates. The combined effects of high-risk balance sheets and distress in the banking sector to a sluggish pass-through were highlighted by the financial crisis (Altavilla et al. 2020; Holton and d’Acri 2018; Von Borstel et al. 2016; Hristov et al. 2014). Such environment changed the interest rate setting strategy, making loan spreads higher in banks that incurred larger losses or shortfall in capital and liquidity buffers (Gambacorta and Mistrulli 2014; Santos 2011). Slowing down in the speed of pass-through is also associated with longer term of loans or deposits (Liu et al. 2008), repeated discount rate as a signaling device (Cottarelli and Kourelis 1994), and absence of lending relationship (Gambacorta and Mistrulli 2014). These latter elements, however, were not included in our specifications because would require some arbitrary adjustments to synchronize the data frequency, since a larger dataset is not readily available.3 We found convincing evidence of full pass-through from both the observed and the expected policy interest rates to the majority of lending rate types. For the overall sample, sub-samples by households and non-financial corporations and some specific lending types, the estimates indicated an over-proportional pass-through, meaning that banks increase loan interest rates more than proportional to any raise in the Over-Selic interest rate, either observed or expected. The banks’ behavior is asymmetric, as downward adjustments in the lending rates are always smaller than the upward ones. The degrees of pass-through are strongly correlated with the interest rate margins, meaning that higher spreads are coupled with larger and positively asymmetric pass-through coefficients. Banks anticipate future changes in the policy rate when setting interest rates on loans, as the pass-through estimates are similar for both observed and expected rates.4 These findings are robust to the inclusion of other control variables and partial adjustment in a dynamic panel data environment, which additionally revealed high persistence in some loan rates. Taken together, they contribute to explain why loan interest rates are kept so high in the Brazilian economy, regardless of downward movements in the policy rate during the period. Any increase in the policy interest rate, either observed or expected, leads to increases at least as proportional as in highly persistent lending rates, while any stimulus to decrease in these rates is refrained by the financial institutions. Other complementary findings also contribute to disentangle the role of the pass-through to keep the remarkably high lending rates in the Brazilian economy. First, it is important to control for the heterogeneity in the lending rates by both loan and borrower types because interest rate margins, credit risk and other specific characteristics are quite different among them. Second, financial institutions correctly forecast the next level of the policy rate and use this information to anticipate asymmetric adjustments in their lending rates, as the estimated degrees of pass-through from either the observed or the expected Over-Selic rate are very similar. Finally, there is a strong and positive correlation between the degree of pass-through and the interest rate margins across borrower types and policy rates, meaning that loan types with higher interest rate margins are coupled with larger degrees of pass-through and lower stickiness in lending rates. Schmeling et al. (2022) showed that deviations from the conventional Taylor rule are correlated with expectation errors in the US economy. They found asymmetric short rate predictability because financial institutions correctly anticipate the direction of the changes, but most surprises were in rate cuts, not in unexpected rate hikes. Furthermore, the magnitude of the decline is more often underestimated than the size of the increase. Cieslak (2018) obtained similar results, with negative forecasting errors during and after recessions suggesting underestimation of monetary policy easing. These findings are in line with the Brazilian case, as the financial institutions predict the interest rate rule followed by the Central Bank of Brazil and correctly forecast the next target level for short horizons according to Divino and Haraguchi (2022). However, they might underestimate monetary easing and overestimate monetary tightening as captured by the asymmetric interest rate pass-through for the lending rates.5 The paper is organized as follows. The next section discusses the dataset and provides a summary of descriptive statistics and illustrates the several interest rate types. Section 3 outlines the empirical strategy. Section 4 reports and analyses the major findings. Section 5 describes and applies robustness tests. Resorting to the theoretical literature, it also discusses and explains the major empirical findings. Finally, Sect. 6 is dedicated to the concluding remarks. Data The dataset comprises interest rates from new credit operations (lending rates), Over-Selic interest rate6 and expectations identified by professional forecasters (financial institutions) of the next Over-Selic target level. The sample covers the period from January 5th, 2012 to April 4th, 2019 on a weekly basis. The original dataset of loan operations contains the five-business-days weighted moving average of interest rates by financial institutions and loan types.7 To synchronize with the dates of the Monetary Policy Committee (Copom) meetings and capture Selic changes, we considered only observations beginning on Thursdays or the next business day in case the Thursday was a holiday. This procedure resulted in up to 378 weekly observations per financial institution and loan type, as illustrated in Figs. 1 and 2.8Fig. 1 Observed Over-Selic rate and household lending rates. Notes: The figure reports the observed Over-Selic rate (black line) and households lending rates by financial institutions and loan types (colored dots). Each color corresponds to a financial institution. CC stands for credit card. All types are fixed interest rates. (Color figure online) Fig. 2 Observed Over-Selic rate and non-financial corporations lending rates. Notes: The figure reports the observed Over-Selic rate (black line) and non-financial corporations lending rates by financial institutions and loan types (colored dots). Each color corresponds to a financial institution. ACC and CC stands for advances on exchange contracts and credit card, respectively. The types in the first row are floating interest rates, except ACC which is a foreign-currency indexed interest rate. The remaining types are fixed interest rates. (Color figure online) Within this period, there were 58 Copom meetings, with an average interval of 46 days between two consecutive meetings (ranging from 35 to 63 days). Meetings always begin on Tuesday and end on Wednesday, when the Selic target is decided and publicly released. The target rate is effective from the next business day after the meeting until a new decision is made in the next meeting. Selic expectations always refer to the next Over-Selic target level. These expectations are collected daily through the “Focus Survey” carried out by Central Bank of Brazil across financial institutions and a median expectation is weekly released to the public.9 Selic expectations are also available by financial institutions, but with one-year delay in the release and each institution anonymously identified by a non-public code, as illustrated in Fig. 3. For this research, the Central Bank of Brazil has kindly provided a list of the confidential codes that matches lending rates and Selic expectations by financial institutions.10 As a result, we were able to build an accurate dataset of lending interest rates and Selic expectations both identified by financial institutions. This unique dataset reduces loss of information that would be caused by using aggregate median expectations as usually done by other studies.11 It also allows us to estimate the pass-through from the identified Selic expectations to the loan interest rates and infer whether future changes in the Selic rate are correctly anticipated by the financial institutions and transmitted to their lending interest rates.Fig. 3 Observed and expected Over-Selic rates. Notes: The figure presents the observed Over-Selic rate (black line) and expected Over-Selic rate (green bubbles). The size of the bubble represents the number of financial institutions that reported the same expected value in a given week. (Color figure online) The financial institutions are identified by the National Register of Legal Entity (CNPJ), a public enterprise tax identification number of the financial institution that granted the loan. On the other hand, the Selic expectations are associated to a code other than the CNPJ that identifies the financial institution responsible for entering the forecasts in the Focus Survey. There is a unique and confidential list from the Central Bank of Brazil matching CNPJ and Selic expectation codes by financial institution. However, CNPJ from lending rates and codes from Selic expectations hardly match one another without further information. In some cases, several financial institutions are part of the same conglomerate, where each one has its own area of experts responsible for forecasting the next target level of the Over-Selic. In many cases, the area in charge of making the forecasts has a different CNPJ than the area that grants loans to individuals and firms. The information binding these distinct CNPJ is the conglomerate. Therefore, we replaced all financial institutions’ specific CNPJ by their respective conglomerate’s CNPJ. In case there is no corresponding conglomerate, we considered the financial institution as a conglomerate with only one subsidiary. By doing this manipulation in the original dataset, we were able to faithfully match CNPJ and Selic expectation codes by financial institutions. The original dataset contains Selic expectations for all dates in which financial institutions entered their initial forecast or revision of forecast in the Focus survey. Selic expectations are not restricted to the next Copom meeting and so might refer to any future meeting. In order to standardize the dataset and match the lending rates frequency, we selected the last expectations in effect on Thursdays to transform the data frequency in weekly figures. We also filtered observations to keep only expectations for the target level to be decided in the next Copom meeting. As pointed out by Coelho et al. (2010), banks costs of funds increase immediately in response to a raise in the basic interest rate, especially for short-maturity loans. The expected increase in costs that leads to raise in lending rates is better approximated by the expectations of the policy rate for the next Copom meeting than for future meetings. Expectations with horizons greater than 45 days were not considered because forecasts are more reliable as the Copom meeting approaches. Loan operations are classified by size and capital origin of the financial institution, type of borrowers and interest rate modality. Segment S1, as defined by the Central Bank of Brazil, is composed of systemically important banks whose characteristics are a size equals to or bigger than 10% of the Brazil GDP or a relevant international activity, regardless its size. Regarding the proprietorship, a financial institution might be either private- or public-owned and the capital origin might be either domestic or foreign. There are two types of borrowers, represented by households (HH) and non-financial corporations (NFC). Loans for NFC are categorized in 12 types, while for HH in 11 modalities. All HH types and the majority of NFC types have fixed interest rates (Fixed). For NFC, three loan types have floating interest rates (Float) and one has foreign-currency-indexed (FCI) interest rate. This later type is used as a placebo in the empirical evidence, given that there should be no pass-through from the domestic rates to the FCI rate. In order to avoid estimation biases, we trimmed outliers above the 97th percentile of each type. Descriptive statistics for the whole sample are reported in Table 1. Table 2 describes the distribution of loan operations and financial institutions by borrower and lender types.Table 1 Descriptive statistics Type Observations Mean SS Minimum 25% Median 75% Maximum Households CC financing 3611 137.3 42.7 15.1 103.2 145.9 166.3 226.6 CC revolving 4035 384.2 154.8 53.9 253.8 399.5 495.5 662.3 Discount-checks 1259 51.5 11.0 26.9 42.3 51.4 60.6 70.8 Other goods financing 3468 49.5 24.5 2.1 29.6 44.3 66.4 118.6 Overdraft 3732 201.9 106.0 12.7 101.4 207.6 292.5 422.3 Payroll-deducted-private 4914 36.2 8.8 0.0 29.9 35.7 41.2 56.8 Payroll-deducted-public 4630 25.4 3.2 11.6 23.0 25.4 27.8 32.8 Payroll-deducted-retirees 5184 27.4 2.5 15.9 26.1 27.6 28.9 32.3 Personal credit 4992 84.7 57.4 0.0 51.6 70.9 93.2 293.4 Vehicle financing 5190 22.0 4.4 9.8 19.3 22.4 25.3 30.2 Vehicle leasing 1408 17.7 4.1 7.5 14.7 17.2 20.3 29.8 Non-financial corporations ACC (FCI) 5995 4.2 1.7 0.0 2.9 4.0 5.4 8.8 Discount-CC bills 2095 31.1 12.0 6.6 20.6 32.6 40.3 54.8 Discount-checks 2834 34.6 7.8 15.8 28.6 34.9 40.6 48.6 Discount-trade bills 4542 26.3 10.1 0.0 18.9 26.4 33.8 49.6 Guaranteed overdraft 3689 51.7 32.3 9.6 31.2 39.5 62.9 192.2 Guaranteed overdraft (Float) 5542 22.4 4.8 7.2 19.2 22.0 25.1 36.3 Overdraft 3581 196.6 101.6 42.7 92.1 185.7 281.4 370.9 Vendor 2905 16.6 3.6 3.2 14.0 16.2 18.9 27.2 Working capital ∼365 4859 24.8 9.5 0.0 18.0 22.4 29.9 53.4 Working capital ∼365 (Float) 5151 17.8 4.5 3.7 14.5 17.5 20.7 30.7 Working capital 365∼ 4386 23.6 8.6 0.0 17.2 21.9 28.4 50.9 Working capital 365∼ (Float) 4550 16.5 3.8 1.7 13.8 16.2 19.0 27.6 Selic Selic rate 378 10.1 2.8 6.4 7.2 10.2 12.9 14.2 Selic expectation 14,390 10.0 2.8 6.0 7.2 10.0 12.8 15.2 Interest rates are non-weighted and in percent values. CC and ACC stand for credit card and advances on exchange contracts; FCI designates foreign-currency-indexed interest rate Table 2 Number of observations and financial institutions by borrower and lender types All financial institutions S1 financial institutions Total Public Private Foreign Total Public Private Foreign Number of observations Total 92,552 21,279 48,528 22,745 52,002 13,893 28,335 9774 Households 42,423 9865 23,981 8577 27,978 6595 15,950 5433 Non-financial corporations 50,129 11,414 24,547 14,168 24,024 7298 12,385 4341 Number of financial institutions Total 57 4 34 19 30 3 20 7 Households 49 4 33 12 30 3 20 7 Non-financial corporations 32 3 17 12 11 2 7 2 S1 stands for systemically important financial institutions This dataset contains more accurate information and covers an updated period when compared to other studies (Pereira and Maia-Filho 2013; Castro and Mello 2012; Coelho et al. 2010). According to the Central Bank of Brazil, in the new database of credit operations, the data coverage was extended and the operations were reclassified to meet needs for households and corporate financing.12 Another distinguish feature is that Selic expectations are uniquely identified by financial institutions, unlike earlier information on aggregate expectations by the mean or median across financial institutions. A potential limitation, however, is that data with weekly figures of interest rates by financial institutions are only available after the year of 2012. Nonetheless, all types of loan interest rates are freely negotiated between financial institutions and borrowers, meaning that they are market rates. Empirical strategy We are interested in estimating the degree of pass-through from the observed and the expected policy rates to the loan interest rates and testing whether the estimates are over-proportional or asymmetric. To do so, we propose a panel-based approach to measure how changes in the observed and the expected Over-Selic rates might currently affect the lending interest rates. In case there is evidence of over-proportional and positively asymmetric pass-through, this might be used to account for the remarkably high lending rates in the Brazilian economy. The fixed effects approach controls for unobserved individual heterogeneity, which is a relevant feature among financial institutions and loan types in the full sample. Panels are unbalanced because financial institutions are not obligated to report Selic expectations to the Focus survey of the Central Bank of Brazil and we trimmed outliers above the 97th percentile for each loan type.13 The next sections report the empirical models and discuss the major results. Baseline specification In order to have a comprehensive view of the interest rate pass-through, we use not only aggregate data, but also sub-samples by lending rate types and type of borrowers. This is rather relevant due to the heterogeneity in interest rate types, as illustrated in Table 1 and Figs. 1 and 2. The overall sample comprises all types except credit card revolving and advances on exchange contracts (ACC). There is a structural break in the former14 and the funding of the latter comes from the foreign market, whose interest rate is not affected by the domestic monetary policy.15 Sub-samples by household (HH) and non-financial corporation (NFC) loans also do not include these types. In addition to the overall sample and two sub-samples, we also estimate panels for each one of the 23 lending rate types across all financial institutions. Considering the fact that we estimate the pass-through for both observed and expected Over-Selic rates, there are 52 panels in total in the empirical analysis. The baseline model is:1 LendingRatem,i,t=α+βBaseRatei,t+Ctδ+εm,i,t where LendingRatem,i,t is the lending rate of type m and financial institution i during time t, BaseRatei,t is the explanatory variable (either observed or expected Over-Selic rate), Ct is a row vector of control variables, and εm,i,t is the compound error term. Let’s define A≡InflationteEMBIt, where Inflationte is the 12-months-ahead expected inflation rate and EMBIt is the EMBI+ Brazil index, used as a proxy for risk perception.16 We have Ct=A, except for two sub-samples. First, Ct=AD(CC)tBaseRatei,t×D(CC)t for Credit card revolving, where D(CC)t is a dummy variable for the structural change in the rules of this loan type. D(CC)t accounts for the change in level while BaseRatei,t×D(CC)t for the change in slope or in the pass-through coefficient. Second, Ct=ALibort for ACC, where Libort is the US dollar Libor rate. Since ACC funding comes from the foreign market, we consider the US dollar Libor rate as a proxy for the foreign funding cost. We assume the one-way error component model for the compound disturbance:2 εm,i,t=μm,i+γt+νm,i,t where μm,i is the unobservable type–financial institution specific effect, γt is the unobservable time fixed effect, and νm,i,t is the aggregate time varying disturbance. Coefficients α and β are scalars while δ is a column vector. The explanatory variable BaseRatei,t is either the Over-Selic rate (Selict) or the identified expectations of the Over-Selic rate (Expeci,t). Sub-index i is ineffective for the observed Over-Selic rate because it varies over time but not across financial institutions. The expected Over-Selic rate, however, is identified by financial institutions (professional forecasters) and so varies in both dimensions, i and t. The coefficient of primary interest is β. We should have β>1 for over-proportional interest rate pass-through. In case β=0, there is no pass-through, while 0<β<1 and β=1 means incomplete and full pass-through, respectively.17 We assume that μm,i is the loan-type and financial-institution fixed effects. Hausman’s and other specification tests might be used to check the alternative specifications of fixed-effects, random effects and pooled sample. We found evidence in favor of the consistent generalized least squares (GLS) estimator for the aggregate samples and 17 lending rate types. Nevertheless, instead of using different specifications, we choose to apply the fixed-effects estimator for the overall sample and all sub-samples. We prefer to lose efficiency, but get consistent estimators under eventual correlation between explanatory variables and the unobserved time-invariant component of the error term, μm,i.18 The constraint ∑m,iμm,i=0 is applied to compute the overall intercept, α, meaning that it makes the weighted average of fixed effects null. This condition equalizes the averages of the observed and fitted values, leaving the remaining fixed effects as deviations from the estimated lending rates. Additionally, under this constraint, the fixed-effects estimator, although less efficient, becomes adequate for estimating the random-effects model as well. The intercept, α, represents a constant average bank margin—or mark up, or interest rate spread—over the reference rate (e.g., Gregor et al. 2021; Banerjee et al. 2013). It is an average margin independent from the monetary policy upon the risk-free interest rate, the Over-Selic rate. It will also be computed as an expected average margin upon the expected Over-Selic rate identified by financial institutions. We apply a robust variance-covariance matrix given by the Huber/White/sandwich estimator for within-groups, which is heteroskedasticity and serial correlation consistent according to Arellano (1987). Standard errors are clustered by loan types and financial institutions in the aggregate samples and by financial institutions in the sub-samples. Asymmetric pass-through In order to test for asymmetric responses of the loan interest rates to changes in the Over-Selic rate or expected Over-Selic rate, we estimate the following model:3 LendingRatem,i,t=α+βBaseRatei,t+θ-BaseRatei,t×D(ΔBaseRate<0)i,t+θ+BaseRatei,t×D(ΔBaseRate>0)i,t+γ-D(ΔBaseRate<0)i,t+γ+D(ΔBaseRate>0)i,t+Ctδ+εm,i,t where D(ΔBaseRate<0)i,t and D(ΔBaseRate>0)i,t are dummy variables that assume values equal to 1 in the following cases (and zero otherwise): D(ΔSelic<0)t=1 for negative variation in the Selic rate, D(ΔExpec<0)i,t=1 for negative variation in the expected Selic rate, D(ΔSelic>0)t=1 for positive variation in the Selic rate, D(ΔExpec>0)i,t=1 for positive variation in the expected Selic rate. The compound error term, εm,i,t, follows the same specification described in Eq. (2). We are interested in θ- and θ+, which capture the differentials in the pass-through coefficient due to decreases and increases in the Selic rate or the expected Selic rate, respectively. Differentials in the level of the loan interest rates are measured by γ- and γ+, and are included in the model to avoid bias in the estimated asymmetry coefficients. We cannot reject the hypothesis of positively asymmetric pass-through when either θ+>0, θ-<0, or θ+>0 and θ-<0 simultaneously. In case θ+<0 and θ->0, either simultaneously or independently, then there is evidence of negatively asymmetric pass-through. Results Baseline interest rate pass-through We first estimate the baseline model for the overall sample and the HH and NFC sub-samples, whose results are reported in Table 3. Confidence intervals for the coefficients of Selict and Expeci,t indicate the existence of over-proportional pass-through in all panels, with similar responses in the HH and NFC loan interest rates. The confidence intervals also suggest that the pass-through from the observed and expected policy interest rates to the loan rates are analogous in all samples. A remarkable difference, however, is the interest rate margins, α, which are clearly higher for HH loans.Table 3 Interest rate pass-through Type Pass-through Interest rate margin Selic (β) (α) Overall (1) 1.77∗∗∗ 55.1∗∗∗ OBS (1.36, 2.18) (47.9, 62.3) Overall (2) 1.80∗∗∗ 57.0∗∗∗ EXP (1.37, 2.23) (49.7, 64.3) Households (3) 1.78∗∗∗ 74.3∗∗∗ OBS (1.07, 2.50) (63.1, 85.5) Households (4) 1.79∗∗∗ 76.2∗∗∗ EXP (1.04, 2.54) (64.9, 87.5) Non-financial corporations (5) 1.76∗∗∗ 38.1∗∗∗ OBS (1.33, 2.20) (29.1, 47.1) Non-financial corporations (6) 1.82∗∗∗ 40.0∗∗∗ EXP (1.37, 2.27) (30.8, 49.3) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. 95% confidence interval in parentheses. Estimated with fixed effects. All regressions are controlled by expected inflation and EMBI. OBS indicates that the explanatory variable in the regression is Selict while EXP indicates that the explanatory variable is Expecit However, these apparently strong results should be interpreted with caution because of the wide heterogeneity in interest rate loan types in the overall sample as well in the HH and NFC sub-samples, as illustrated earlier. The disaggregation of the overall sample by HH and NFC sub-samples did not affect the degree of pass-through as the heterogeneity in the loan types is still high within each borrower category. Table 4 increases the disaggregation and reports estimates by interest rate types. For HH interest rate types, there is no pass-through from both observed and expected Over-Selic only for the Credit card financing rate (panels 1 and 2). Two types—Credit card revolving rate (3 and 4) and Personal credit rate (17 and 18)—revealed significant β for the Selic rate, but not for the expected Selic rate. For all remaining interest rate types, however, there is evidence of pass-through at the 95% confidence level from both observed and expected Selic rates.Table 4 Interest rate pass-through by loan types Households Non-financial corporations Type Pass-through Interest rate Selic Type Pass-through Interest rate Selic (β) margin(α) (β) margin(α) CC financing (1) 2.44 151.8∗∗∗ OBS ACC (1) 0.00 3.4∗∗∗ OBS (-0.53, 5.40) (115.5, 188.2) (-0.05, 0.06) (2.3, 4.4) CC financing (2) 2.41 154.0∗∗∗ EXP ACC (2) -0.02 3.6∗∗∗ EXP (-0.69, 5.50) (119.9, 188.2) (-0.07, 0.03) (2.5, 4.7) CC revolving (3) 17.07∗∗∗ 361.6∗∗∗ OBS Discount-CC bills (3) 2.96∗∗∗ 8.3 OBS (8.03, 26.11) (236.1, 487.0) (2.23, 3.69) (-8.4, 25.0) CC revolving (4) 7.14∗ 499.1∗∗∗ EXP Discount-CC bills (4) 3.07∗∗∗ 11.6∗ EXP (-0.15, 14.42) (369.9, 628.3) (2.21, 3.92) (-4.2, 27.3) Discount-checks (5) 1.31∗∗∗ 42.9∗∗∗ OBS Discount-checks (5) 1.38∗∗∗ 31.4∗∗∗ OBS (0.63, 1.98) (35.4, 50.4) (1.22, 1.54) (25.0, 37.8) Discount-checks (6) 1.44∗∗∗ 44.5∗∗∗ EXP Discount-checks (6) 1.40∗∗∗ 32.8∗∗∗ EXP (0.65, 2.24) (37.9, 51.1) (1.22, 1.58) (26.5, 39.1) Other goods financing (7) 1.82∗∗∗ 51.4∗∗∗ OBS Discount-trade bills (7) 1.66∗∗∗ 11.2∗∗∗ OBS (0.72, 2.91) (36.0, 66.8) (1.17, 2.15) (4.5, 18.0) Other goods financing (8) 1.59∗∗∗ 52.4∗∗∗ EXP Discount-trade bills (8) 1.69∗∗∗ 13.0∗∗∗ EXP (0.68, 2.50) (38.0, 66.8) (1.16, 2.21) (6.5, 19.5) Overdraft (9) 6.34∗∗∗ 312.9∗∗∗ OBS Guaranteed overdraft (9) 2.34∗∗ 45.4∗∗∗ OBS (3.18, 9.49) (276.3, 349.4) (0.49, 4.19) (33.6, 57.3) Overdraft (10) 6.81∗∗∗ 321.3∗∗∗ EXP Guaranteed overdraft (10) 2.28∗∗ 47.6∗∗∗ EXP (3.55, 10.08) (283.2, 359.4) (0.41, 4.14) (36.4, 58.8) Payroll-deducted (11) 0.85∗∗∗ 34.8∗∗∗ OBS Guaranteed overdraft (11) 1.00∗∗∗ 12.8∗∗∗ OBS - Private (0.54, 1.16) (32.0, 37.5) (Float) (0.81, 1.19) (9.8, 15.8) Payroll-deducted (12) 0.86∗∗∗ 35.7∗∗∗ EXP Guaranteed overdraft (12) 1.03∗∗∗ 14.0∗∗∗ EXP - Private (0.52, 1.19) (32.9, 38.6) (Float) (0.84, 1.22) (11.1, 16.9) Payroll-deducted (13) 0.65∗∗∗ 22.1∗∗∗ OBS Overdraft (13) 7.13∗∗∗ 295.1∗∗∗ OBS - Public (0.47, 0.82) (19.9, 24.4) (4.24, 10.02) (223.0, 367.1) Payroll-deducted (14) 0.64∗∗∗ 22.8∗∗∗ EXP Overdraft (14) 7.34∗∗∗ 303.0∗∗∗ EXP - Public (0.45, 0.83) (20.6, 24.9) (4.30, 10.38) (229.2, 376.9) Payroll-deducted (15) 0.59∗∗∗ 23.4∗∗∗ OBS Vendor (15) 0.82∗∗∗ 10.7∗∗∗ OBS - Retirees (0.48, 0.70) (22.0, 24.8) (0.62, 1.02) (7.4, 13.9) Payroll-deducted (16) 0.59∗∗∗ 24.1∗∗∗ EXP Vendor (16) 0.83∗∗∗ 11.5∗∗∗ EXP - Retirees (0.48, 0.71) (22.7, 25.4) (0.61, 1.05) (8.5, 14.5) Personal credit (17) 2.53∗∗ 75.4∗∗∗ OBS Working capital (17) 1.21∗∗∗ 17.0∗∗∗ OBS (0.09, 4.98) (36.8, 113.9) ∼365 (0.84, 1.58) (9.3, 24.7) Personal credit (18) 2.43∗ 77.8∗∗∗ EXP Working capital (18) 1.26∗∗∗ 18.4∗∗∗ EXP (-0.25, 5.11) (41.7, 113.9) ∼365 (0.87, 1.64) (10.9, 25.8) Vehicle financing (19) 0.66∗∗∗ 17.7∗∗∗ OBS Working capital (19) 0.90∗∗∗ 6.2∗∗∗ OBS (0.49, 0.83) (15.7, 19.7) ∼365 (Float) (0.77, 1.03) (4.5, 8.0) Vehicle financing (20) 0.69∗∗∗ 18.4∗∗∗ EXP Working capital (20) 0.96∗∗∗ 7.4∗∗∗ EXP (0.51, 0.87) (16.4, 20.4) ∼365 (Float) (0.82, 1.11) (5.7, 9.1) Vehicle leasing (21) 0.61∗∗∗ 13.8∗∗∗ OBS Working capital (21) 1.12∗∗∗ 10.6∗∗∗ OBS (0.29, 0.93) (7.6, 19.9) 365∼ (0.77, 1.47) (4.6, 16.6) Vehicle leasing (22) 0.65∗∗∗ 14.5∗∗∗ EXP Working capital (22) 1.20∗∗∗ 12.0∗∗∗ EXP (0.28, 1.02) (8.7, 20.3) 365∼ (0.81, 1.59) (6.3, 17.7) Working capital (23) 0.74∗∗∗ 5.1∗∗∗ OBS 365∼ (Float) (0.52, 0.95) (3.8, 6.4) Working capital (24) 0.78∗∗∗ 6.0∗∗∗ EXP 365∼ (Float) (0.56, 1.01) (4.6, 7.5) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. 95% confidence interval in parentheses. Estimated with fixed effects. CC is for credit cards and ACC is for advances on exchange contracts. All regressions are controlled by expected inflation and EMBI. CC revolving is also controlled by the structural change in the rules of this loan type, and ACC is also controlled by the Libor rate. OBS indicates that the explanatory variable in the regression is Selict, while EXP indicates that the explanatory variable is Expecit The estimated confidence intervals indicate incomplete pass-through for three types—Payroll-deducted loans to public sector employees (13 and 14), Payroll-deducted loans to retirees (15 and 16), and Vehicle financing (19 and 20)—and full pass-through for four other types—Discount of checks (5 and 6), Other goods financing (7 and 8), Payroll-deducted loans to private sector employees (11 and 12), and Personal credit (17). One type—Vehicle leasing (21 and 22)—shows incomplete pass-through from the Selic rate, but full pass-through from the expected Selic rate. An interesting result is that, for Credit card revolving (3) and Overdraft (9 and 10), there is evidence of over-proportional pass-through, similarly to the estimates for the overall, HH and NFC samples reported in Table 3. Credit card revolving and Overdraft are the most expensive credit lines and have the highest margins in the sample, suggesting that the over-proportional pass-through was not found merely by chance. Estimated pass-through from the expected Selic rate (even-numbered panels), in general, confirms the findings from the observed Selic rate (odd-numbered panels), and the degrees of pass-through are very similar when changing between them for a given loan type. The only exception is Credit card revolving rate (3 and 4), where the pass-through for the expected Selic rate was not statistically significant at the 5% level. One possible explanation is a potential structural break resulting from the legal change in the Credit card revolving rules. This legal change was announced some months before the effective implementation, allowing for the financial institutions and borrowers to adjust behaviors in advance. Results for NFC are even more homogeneous. Estimated pass-through coefficients are statistically significant for all types, except for Advances on exchange contracts (panels 1 and 2) as expected because it was used as a placebo.19 There is over-proportional pass-through for the following types: Discount of credit card bills (3 and 4), Discount of checks (5 and 6), Discount of trade bills (7 and 8), and Overdraft (13 and 14). Not a coincidence, the highest interest rate margin is coupled with the highest degree of over-proportional pass-through for the Overdraft type under both observed and expected Selic rate. For the other NFC types, the pass-through is complete for both observed and expected Selic rates. The only exceptions are Working capital over 365 days and floating rate (23 and 24), which showed incomplete pass-through under the observed Selic rate. Similarly to the HH results, the NFC types with higher loan interest rates also revealed less rigidity and over-proportional pass-through. The top five most expensive types, considering the average interest rates, also presented the highest pass-through coefficients. Among them, only for Guaranteed overdraft fixed rates (9 and 10) there is evidence of full, but not over-proportional, pass-through. Similarly to the HH types, the estimated degrees of pass-through are very similar for both observed and expected Selic rates, indicating that financial institutions successfully forecasts the next target level of the Over-Selic rate and adjust in advance their lending interest rates. The interest rate margins, α, are positive and well dispersed across the loan types. It is not statistically significant only for Discount of credit card bills of NFC. There is a striking pattern of positive correlation between the margins and the degrees of pass-through, as reported in Fig. 4. The correlations are very strong, irrespective of the borrower category (HH or NFC) or Selic rate (observed or expected). The positive slopes of the fitted regressions illustrate that types with the highest margins also present over-proportional degrees of pass-through. While the margins in Fig. 4 might be correlated with the risk levels by loan and borrower types, the degree of pass-through is bigger for loans with higher interest rate margins. There are other factors that might affect margins, such as operating, administrative and taxing costs, but banks claim that the credit risk is a key component of the interest rate spread.20Fig. 4 Interest rate margins and degrees of pass-through for both observed and expected Selic rates. Notes: The figure reports the margins (α) and the degrees of pass-through (β) by type of borrower (HH and NFC) and policy rate (observed and expected Selic). Filled dots are for statistically significant β, while open dots are for non-statistically significant β. Vertical bars represent the 95% confidence interval for β. Shaded areas are the 95% confidence interval for predictions by a linear model with significant β’s. Modalities with over-proportional pass-through are highlighted with their type of loans It is worth highlighting that the heterogeneity in lending rates shall be taken into account when assessing the pass-through from the observed or expected policy rates. Loan types with higher rates and margins appear to show lower stickiness and over-proportional degrees of pass-through. The prevalence of full and over-proportional pass-through differs from previous findings by Holton and d’Acri (2018) and Hristov et al. (2014), but is in line with Coelho et al. (2010), who accounted for the concentration in the Brazilian banking system. Asymmetric interest rate pass-through We estimate Eq. (3) to evaluate asymmetry in the interest rate pass-through, and the results are reported in Table 5. The estimates of θ- and θ+ measure the asymmetric effects of changes in the observed or expected Selic rates on the degree of pass-through for distinct lending rates. They are not statistically significant for the HH sub-sample, but θ- is negative and statistically significant for the overall sample and NFC sub-sample, meaning lower pass-through under decreases in the policy rates. On the contrary, none of the estimates for θ+ is statistically significant. HH and NFC sub-samples have different findings, as there are significant asymmetric effects only for the latter. To account for the heterogeneity, we disaggregate the sub-samples by loan types.Table 5 Asymmetric interest rate pass-through Type Pass-through Asymmetry Asymmetry Selic (β) (θ-) (θ+) Overall (1) 1.80∗∗∗ -0.24∗∗ -0.19 OBS (1.39, 2.22) (-0.44, -0.03) (-0.50, 0.12) Overall (2) 1.84∗∗∗ -0.30∗∗ -0.13 EXP (1.41, 2.28) (-0.54, -0.06) (-0.33, 0.07) Households (3) 1.82∗∗∗ -0.22 -0.22 OBS (1.09, 2.54) (-0.56, 0.13) (-0.87, 0.43) Households (4) 1.83∗∗∗ -0.21 -0.12 EXP (1.08, 2.59) (-0.65, 0.23) (-0.51, 0.27) Non-financial corporations (5) 1.80∗∗∗ -0.28∗∗ -0.13 OBS (1.35, 2.24) (-0.51, -0.04) (-0.37, 0.11) Non-financial corporations (6) 1.86∗∗∗ -0.38∗∗∗ -0.11 EXP (1.39, 2.33) (-0.62, -0.15) (-0.29, 0.08) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. 95% confidence interval in parentheses. Estimated with fixed effects. All regressions are controlled by expected inflation and EMBI. OBS indicates that the explanatory variable in the regression is Selict while EXP indicates that the explanatory variable is Expecit For the HH sub-sample, Table 6 reveals that four types—Overdraft (panels 9 and 10), Payroll-deducted loans to public sector employees (14), Payroll-deducted loans to retirees (15 and 16), and Vehicle financing (20)—show statistically significant asymmetry for either observed or expected Selic rates. For the majority of these types, decreases in policy rate are coupled with smaller degree of pass-through when compared to increases in this rate. Only Payroll-deducted loans to retirees revealed an opposite behavior. The coefficient θ+ is negative for credit card financing, but β is not statistically significant for this type.Table 6 Asymmetric interest rate pass-through by loan types Households Non-financial corporations Type Pass-through Asymmetry Asymmetry Selic Type Pass-through Asymmetry Asymmetry Selic (β) (θ-) (θ+) (β) (θ-) (θ+) CC financing (1) 2.43 0.24 -9.12∗∗∗ OBS ACC (1) 0.01 -0.06∗∗ -0.09∗∗ OBS (-0.54, 5.40) (-0.88, 1.37) (-13.51, -4.73) (-0.04, 0.07) (-0.11, -0.00) (-0.17, -0.02) CC financing (2) 2.40 0.03 -1.32∗∗∗ EXP ACC (2) -0.02 -0.03 -0.04∗ EXP (-0.72, 5.53) (-1.55, 1.61) (-2.22, -0.43) (-0.07, 0.03) (-0.08, 0.02) (-0.08, 0.00) CC revolving (3) 18.20∗∗∗ 0.20 -11.39∗ OBS Discount-CC bills (3) 2.98∗∗∗ 0.00 -0.58∗∗ OBS (8.82, 27.58) (-3.80, 4.19) (-24.89, 2.11) (2.25, 3.71) (-0.50, 0.50) (-1.02, -0.13) CC revolving (4) 8.08∗∗ 0.16 -2.32 EXP Discount-CC bills (4) 3.06∗∗∗ -0.05 -0.44∗∗∗ EXP (0.38, 15.77) (-3.50, 3.83) (-5.94, 1.30) (2.23, 3.90) (-0.72, 0.63) (-0.70, -0.17) Discount-checks (5) 1.28∗∗∗ 0.09∗ 0.44∗ OBS Discount-checks (5) 1.38∗∗∗ -0.07 -0.01 OBS (0.59, 1.97) (-0.01, 0.19) (-0.07, 0.96) (1.23, 1.54) (-0.19, 0.05) (-0.33, 0.31) Discount-checks (6) 1.39∗∗∗ 0.14∗ 0.16 EXP Discount-checks (6) 1.41∗∗∗ -0.19∗∗ 0.04 EXP (0.57, 2.21) (-0.02, 0.31) (-0.42, 0.75) (1.23, 1.59) (-0.38, -0.01) (-0.10, 0.18) Other goods financing (7) 1.87∗∗∗ -0.66 -0.58 OBS Discount-trade bills (7) 1.69∗∗∗ -0.31∗∗ -0.44∗∗ OBS (0.74, 3.00) (-1.54, 0.23) (-1.63, 0.48) (1.21, 2.18) (-0.57, -0.05) (-0.86, -0.02) Other goods financing (8) 1.67∗∗∗ -1.46 -0.36 EXP Discount-trade bills (8) 1.70∗∗∗ -0.19∗∗ -0.10 EXP (0.73, 2.61) (-3.33, 0.42) (-1.05, 0.34) (1.17, 2.22) (-0.38, -0.00) (-0.29, 0.08) Overdraft (9) 6.22∗∗∗ -0.96 5.28∗∗∗ OBS Guaranteed overdraft (9) 2.52∗∗ -1.64∗ -1.11 OBS (2.91, 9.53) (-3.14, 1.22) (2.69, 7.86) (0.59, 4.46) (-3.60, 0.31) (-2.72, 0.50) Overdraft (10) 7.03∗∗∗ -2.18∗ 2.61∗∗ EXP Guaranteed overdraft (10) 2.33∗∗ -0.43 -0.55 EXP (3.74, 10.31) (-4.51, 0.16) (0.42, 4.80) (0.41, 4.25) (-1.58, 0.72) (-1.87, 0.76) Payroll-deducted (11) 0.86∗∗∗ -0.06 -0.14 OBS Guaranteed overdraft (11) 1.00∗∗∗ -0.05 -0.08 OBS - Private (0.56, 1.17) (-0.27, 0.15) (-0.50, 0.22) (Float) (0.81, 1.20) (-0.19, 0.10) (-0.33, 0.16) Payroll-deducted (12) 0.87∗∗∗ -0.07 -0.08 EXP Guaranteed overdraft (12) 1.03∗∗∗ -0.08 -0.07 EXP - Private (0.52, 1.21) (-0.26, 0.12) (-0.25, 0.08) (Float) (0.84, 1.23) (-0.23, 0.07) (-0.22, 0.08) Payroll-deducted (13) 0.65∗∗∗ -0.08∗ -0.01 OBS Overdraft (13) 7.32∗∗∗ -2.13∗∗ 1.01 OBS - Public (0.47, 0.83) (-0.16, 0.00) (-0.19, 0.16) (4.35, 10.29) (-3.80, -0.47) (-1.63, 3.65) Payroll-deducted (14) 0.64∗∗∗ -0.13∗∗∗ 0.03 EXP Overdraft (14) 7.68∗∗∗ -3.52∗∗∗ 0.31 EXP - Public (0.44, 0.84) (-0.23, -0.04) (-0.05, 0.12) (4.63, 10.74) (-5.09, -1.95) (-1.74, 2.35) Payroll-deducted (15) 0.60∗∗∗ 0.00 -0.19∗∗∗ OBS Vendor (15) 0.83∗∗∗ -0.06 -0.07∗ OBS - Retirees (0.49, 0.70) (-0.03, 0.04) (-0.30, -0.08) (0.64, 1.02) (-0.25, 0.13) (-0.13, 0.00) Payroll-deducted (16) 0.59∗∗∗ -0.01 -0.09∗∗ EXP Vendor (16) 0.83∗∗∗ -0.03 -0.14∗∗∗ EXP - Retirees (0.47, 0.71) (-0.07, 0.05) (-0.17, -0.01) (0.61, 1.05) (-0.14, 0.08) (-0.23, -0.05) Personal credit (17) 2.69∗∗ -0.53 -1.90∗ OBS Working capital (17) 1.20∗∗∗ 0.10 0.14 OBS (0.30, 5.08) (-1.74, 0.69) (-3.96, 0.16) ∼365 (0.82, 1.57) (-0.20, 0.40) (-0.25, 0.53) Personal credit (18) 2.41∗ 0.84 -0.46 EXP Working capital (18) 1.26∗∗∗ -0.14 0.05 EXP (-0.28, 5.09) (-0.22, 1.90) (-1.19, 0.28) ∼365 (0.85, 1.66) (-0.42, 0.14) (-0.15, 0.25) Vehicle financing (19) 0.65∗∗∗ -0.00 0.15∗ OBS Working capital (19) 0.91∗∗∗ -0.13∗∗∗ 0.06 OBS (0.48, 0.83) (-0.09, 0.08) (-0.02, 0.33) ∼365 (Float) (0.77, 1.04) (-0.21, -0.05) (−0.02, 0.14) Vehicle financing (20) 0.70∗∗∗ -0.11∗∗∗ 0.01 EXP Working capital (20) 0.97∗∗∗ -0.05 0.01 EXP (0.51, 0.88) (-0.19, -0.03) (-0.08, 0.10) ∼365 (Float) (0.82, 1.11) (-0.17, 0.08) (-0.09, 0.11) Vehicle leasing (21) 0.61∗∗∗ -0.17 0.18∗ OBS Working capital (21) 1.11∗∗∗ -0.03 0.16 OBS (0.28, 0.93) (-0.40, 0.07) (-0.02, 0.38) 365∼ (0.77, 1.46) (-0.21, 0.16) (-0.05, 0.37) Vehicle leasing (22) 0.64∗∗∗ 0.04 0.16 EXP Working capital (22) 1.21∗∗∗ -0.16∗ 0.09 EXP (0.27, 1.00) (-0.22, 0.30) (-0.08, 0.39) 365∼ (0.82, 1.60) (-0.33, 0.00) (-0.02, 0.19) Working capital (23) 0.73∗∗∗ 0.06 -0.06 OBS 365∼ (Float) (0.51, 0.95) (-0.08, 0.19) (-0.20, 0.08) Working capital (24) 0.78∗∗∗ -0.03 -0.04 EXP 365∼ (Float) (0.55, 1.01) (-0.11, 0.06) (-0.13, 0.06) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. 95% confidence interval in parentheses. Estimated with fixed effects. CC is for credit cards and ACC is for advances on exchange contracts. All regressions are controlled by expected inflation and EMBI. CC revolving is also controlled by the structural change in the rules of this loan type, and ACC is also controlled by the Libor rate. OBS indicates that the explanatory variable in the regression is Selict, while EXP indicates that the explanatory variable is Expecit There is statistically significant asymmetry for five NFC types, represented by Discount of checks (panel 6), Overdraft (13 and 14), Working capital up to 365 days and floating rate (19), Discount of credit card bills (3 and 4), and Vendor (16), as reported in Table 6. For the first three, the asymmetry is positive, while for the last two it is negative. The results are inconclusive for Discount of trade bills (7 and 8), since both θ- and θ+ are negative and statistically significant in the observed Selic rate regression. In summary, out of the 23 loan types, 9 revealed asymmetric interest rate pass-through for the observed, expected or both Selic rates. Among them, there is evidence of positive asymmetry for six types. The negative estimates for θ- or positive for θ+ imply smaller degrees of pass-through for decreases and higher for increases in the observed or expected Selic rates, respectively. These findings are in line with the argument that higher rigidity occurs for movements in interest rates that might decrease the banks’ profitability (e.g., Castro and Mello 2012; Chong 2010; Liu et al. 2008; Neumark and Sharpe 1992; Hannan and Berger 1991). Despite the asymmetry in some loan types, in general, the pass-through coefficients and their confidence intervals have not significantly changed relatively to the baseline estimates, confirming the previous findings. Alternative specifications and robustness checks We examine whether the degrees of pass-through for the loan interest rates reported in Sect. 4.1 are robust to alternative model specifications. First, we control for size, ownership type and capital origin of the financial institutions. Then, we allow for persistence in the loan interest rates and consider the effects of partial adjustment in a dynamic panel data environment. We also included unobserved time-specific fixed effects to account for the possibility that unobserved macroeconomic conditions would influence the evolution of both monetary policy and lending interest rates. However, they were not statistically significant and excluded from the final estimates. Financial-institutions specific characteristics In the previously estimated models, we accounted for macroeconomic conditions (expected inflation, sovereign risk, foreign interest rate) and a loan-specific dummy variable to control for a structural change in credit card revolving rules. However, as discussed in Sect. 1, specific characteristics of the financial institutions might potentially affect the interest rate pass-through. Size, ownership type (private or public) and capital origin (domestic or foreign) of the financial institution are some of the specific characteristics explicitly accounted for in the estimation of the following model:4 LendingRatem,i,t=α+βBaseRatei,t+σBaseRatei,t×D(non-S1)i+ψBaseRatei,t×D(Public)i+ϕBaseRatei,t×D(Foreign)i+Ctδ+εm,i,t, where the dummies D(non-S1)i, D(Public)i, and D(Foreign)i are equal to one for non-systemically important institutions, public-owned government institutions and foreign-controlled private institutions, respectively, and equal to zero otherwise.21 The term εm,i,t follows the one-way error component model described by Eq. (2). Since financial-institution-specific effects, such as those captured by D(non-S1)i, D(Public)i, and D(Foreign)i, are already accounted for in the fixed-effects component, μm,i, the inclusion of level dummies has no role in the estimation. However, their interactions with the observed and expected Selic rates measure disproportional effects from different types of financial institutions in the degree of pass-through. The estimates of β are now for systemically important (S1), private and domestic financial institutions, while the coefficients σ, ψ, and ϕ captures the differentials in the degree of pass-through for non-systemically important, public-owned, and foreign-controlled financial institutions, respectively. The results for the complete sample and sub-samples by HH and NFC lending rates are reported in Table 7. None of the interaction coefficients between the dummy variables and either the observed or expected Selic rates was statistically significant at the 5% significance level. Therefore, the previous findings were not driven by the financial-institutions specific characteristics in the overall sample and two sub-samples.Table 7 Pass-through controlling for size and ownership of the financial institutions Type Pass-through Size Ownership Origin Selic (β) (σ) (ψ) (ϕ) Overall (1) 1.79∗∗∗ -0.06 0.03 -0.02 OBS (1.12, 2.46) (-0.68, 0.56) (-0.79, 0.86) (-0.74, 0.69) Overall (2) 1.83∗∗∗ 0.04 -0.20 0.01 EXP (1.16, 2.50) (-0.55, 0.63) (-0.98, 0.59) (-0.66, 0.67) Households (3) 1.49∗∗∗ 0.43 0.14 0.55 OBS (0.40, 2.58) (-0.81, 1.68) (-1.34, 1.62) (-0.93, 2.03) Households (4) 1.55∗∗∗ 0.47 -0.13 0.57 EXP (0.45, 2.66) (-0.70, 1.65) (-1.56, 1.30) (-0.80, 1.94) Non-financial corporations (5) 2.19∗∗∗ -0.54∗ -0.17 -0.41 OBS (1.46, 2.92) (-1.15, 0.08) (-0.86, 0.53) (-1.11, 0.28) Non-financial corporations (6) 2.22∗∗∗ -0.41 -0.35 -0.39 EXP (1.51, 2.93) (-0.98, 0.16) (-0.98, 0.27) (-1.02, 0.23) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. 95% confidence interval in parentheses. Estimated with fixed effects. All regressions are controlled by expected inflation and EMBI. OBS indicates that the explanatory variable in the regression is Selict, while EXP indicates that the explanatory variable is Expecit Taking into account the heterogeneity in the loan operations, the results for the HH lending rates are reported in Table 8. In general, the previous findings by HH types are also robust to the inclusion of the new dummy variables. The non-systemically important financial institutions yield a significant differential in the degree of pass-through only for Discount of checks (panels 5 and 6), Overdraft (9) and Payroll-deducted loans to retirees (15 and 16), but with no specific pattern among these types and similar effects for both observed and expected Selic rates. For Other goods financing (7 and 8), β was not statistically significant, meaning that the non-S1 institutions might have driven the estimated pass-through in the baseline specification. On the other hand, the public-owned government banks, whenever statistically significant, yielded positive differentials for the estimated degrees of pass-through, except for Credit card revolving (3 and 4) where it was negative. This was the case for Discount of checks (5 and 6), Payroll-deducted loans to retirees (15 and 16) and Vehicles leasing (21 and 22). Finally, foreign-controlled financial institutions, except for Discount of checks (panels 5 and 6) and Vehicles leasing (21 and 22), yielded positive differentials for the pass-through whenever statistically significant. This also happened with Overdraft (9), Payroll-deducted loans to public sector employees (13 and 14), and Payroll-deducted loans to retirees (15 and 16). Interesting to notice that these differentials are very similar for either the observed or expected Selic rates in the regressions, confirming that financial institutions correctly anticipated the next target level of the policy interest rate regardless of their specific characteristics.Table 8 Pass-through for HH loans controlling for size and ownership of the financial institutions Type Pass-through Size Ownership Origin Selic (β) (σ) (ψ) (ϕ) CC financing (1) 1.80 4.84 -5.59 3.30 OBS (-1.81, 5.40) (-1.13, 10.81) (-12.50, 1.31) (-0.70, 7.29) CC financing (2) 1.85 4.80 -6.04∗ 3.01 EXP (-1.70, 5.41) (-0.94, 10.54) (-12.71, 0.64) (-1.03, 7.06) CC revolving (3) 24.28∗∗∗ -7.99 -16.47∗∗∗ -11.85 OBS (13.59, 34.97) (-19.15, 3.17) (-28.90, -4.05) (-32.68, 8.98) CC revolving (4) 15.06∗∗∗ -7.57 -17.00∗∗∗ -10.89 EXP (5.98, 24.14) (-18.37, 3.24) (-29.27, -4.73) (-31.25, 9.46) Discount-checks (5) 1.37∗∗∗ -2.10∗∗∗ 0.89∗∗∗ -0.22∗∗∗ OBS (1.23, 1.52) (-2.26, -1.94) (0.87, 0.91) (-0.25, -0.18) Discount-checks (6) 1.54∗∗∗ -2.11∗∗∗ 0.88∗∗∗ -0.30∗∗∗ EXP (1.40, 1.68) (-2.22, -2.00) (0.86, 0.90) (-0.35, -0.26) Other goods financing (7) 1.20 2.18∗∗∗ 0.03 0.15 OBS (-0.25, 2.66) (1.33, 3.03) (-1.28, 1.33) (-1.61, 1.92) Other goods financing (8) 1.12 2.33∗∗∗ -0.46 0.09 EXP (-0.28, 2.52) (1.49, 3.18) (-1.92, 1.00) (-1.71, 1.89) Overdraft (9) 7.38∗∗∗ -4.67∗∗ 1.74 4.23∗∗ OBS (3.49, 11.28) (-8.96, -0.37) (-2.69, 6.17) (0.13, 8.33) Overdraft (10) 7.91∗∗∗ -3.91∗ 0.81 3.30∗ EXP (3.85, 11.98) (-8.05, 0.23) (-3.49, 5.11) (-0.32, 6.92) Payroll-deducted (11) 0.80∗∗∗ -0.06 0.56∗ -0.46 OBS - Private (0.43, 1.17) (-0.50, 0.39) (-0.06, 1.18) (-1.15, 0.23) Payroll-deducted (12) 0.78∗∗∗ 0.02 0.53∗ -0.44 EXP - Private (0.43, 1.13) (-0.41, 0.45) (-0.05, 1.10) (-1.12, 0.23) Payroll-deducted (13) 0.43∗∗∗ 0.12 0.34 0.37∗∗∗ OBS - Public (0.15, 0.70) (-0.19, 0.43) (-0.10, 0.79) (0.16, 0.59) Payroll-deducted (14) 0.43∗∗∗ 0.11 0.32 0.37∗∗∗ EXP - Public (0.16, 0.69) (-0.21, 0.43) (-0.15, 0.80) (0.16, 0.58) Payroll-deducted (15) 0.41∗∗∗ 0.12∗∗ 0.41∗∗∗ 0.24∗∗∗ OBS - Retirees (0.30, 0.52) (0.01, 0.23) (0.30, 0.53) (0.11, 0.37) Payroll-deducted (16) 0.42∗∗∗ 0.12∗∗ 0.40∗∗∗ 0.25∗∗∗ EXP - Retirees (0.30, 0.53) (0.01, 0.22) (0.27, 0.54) (0.13, 0.36) Personal credit (17) 3.40 -1.64 0.39 -1.77 OBS (-1.43, 8.23) (-5.91, 2.63) (-3.88, 4.66) (-5.65, 2.10) Personal credit (18) 3.24 -1.53 0.26 -1.43 EXP (-1.91, 8.39) (-6.05, 2.99) (-4.14, 4.67) (-5.52, 2.65) Vehicle financing (19) 0.68∗∗∗ 0.05 0.06 -0.14 OBS (0.40, 0.95) (-0.21, 0.30) (-0.28, 0.39) (-0.58, 0.30) Vehicle financing (20) 0.71∗∗∗ 0.04 0.02 -0.11 EXP (0.42, 0.99) (-0.22, 0.30) (-0.33, 0.38) (-0.54, 0.31) Vehicle leasing (21) 0.85∗∗∗ 0.03 0.43∗∗∗ -0.66∗∗∗ OBS (0.50, 1.19) (-0.25, 0.32) (0.23, 0.62) (-0.99, -0.32) Vehicle leasing (22) 0.90∗∗∗ 0.01 0.39∗∗∗ -0.69∗∗∗ EXP (0.53, 1.27) (-0.28, 0.31) (0.26, 0.52) (-1.05, -0.33) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. 95% confidence interval in parentheses. Estimated with fixed effects. All regressions are controlled by expected inflation and EMBI. CC revolving is also controlled by the structural change in the rules of this loan type. OBS indicates that the explanatory variable in the regression is Selict while EXP indicates that the explanatory variable is Expecit For the NFC loan types, the results reported in Table 9 are even more stronger, in the sense that the baseline results were basically kept unchanged. The new estimates confirmed that all loan types, except Advances on exchange contracts (1 and 2), experienced a full or over-proportional pass-through in all alternative specifications. Advances on exchange contracts is the placebo and should not have any pass-through, as expected. For systemically important, private and domestic financial institutions, the over-proportional pass-through was confirmed for Discount of credit card bills (3 and 4), Discount of checks (5 and 6), Discount of trade bills (7 and 8), and Overdraft (13 and 14). For these institutions, full pass-through held in place for all remaining loan types. Overall, these findings are basically the same for either the observed or expected Over-Selic rates.Table 9 Pass-through for NFC loans controlling for size and ownership of the financial institutions Type Pass-through Size Ownership Origin Selic (β) (σ) (ψ) (ϕ) ACC (1) 0.01 -0.02 -0.05 0.05 OBS (-0.07, 0.09) (-0.12, 0.07) (-0.18, 0.08) (-0.06, 0.17) ACC (2) -0.01 -0.03 -0.06 0.06 EXP (-0.10, 0.07) (-0.12, 0.06) (-0.18, 0.07) (-0.05, 0.16) Discount-CC bills (3) 3.76∗∗∗ -2.08∗∗∗ -0.44 -1.70∗∗∗ OBS (2.83, 4.68) (-3.28, -0.89) (-1.47, 0.59) (-2.72, -0.68) Discount-CC bills (4) 3.92∗∗∗ -2.19∗∗∗ -0.59 -1.88∗∗∗ EXP (2.81, 5.04) (-3.52, -0.86) (-1.75, 0.57) (-3.04, -0.72) Discount-checks (5) 1.60∗∗∗ -0.15 -0.27∗ -0.52∗∗∗ OBS (1.48, 1.72) (-0.40, 0.09) (-0.57, 0.03) (-0.63, -0.42) Discount-checks (6) 1.64∗∗∗ -0.15 -0.32∗ -0.50∗∗∗ EXP (1.48, 1.79) (-0.45, 0.15) (-0.71, 0.06) (-0.66, -0.34) Discount-trade bills (7) 2.34∗∗∗ -1.28∗∗∗ 0.28 0.09 OBS (1.77, 2.90) (-1.76, -0.80) (-0.20, 0.76) (-0.44, 0.61) Discount-trade bills (8) 2.38∗∗∗ -1.28∗∗∗ 0.14 0.12 EXP (1.80, 2.96) (-1.77, -0.79) (-0.32, 0.61) (-0.43, 0.66) Guaranteed overdraft (9) 2.04∗∗ 0.70 -0.74 0.75 OBS (0.25, 3.84) (-1.81, 3.21) (-3.03, 1.56) (-1.33, 2.83) Guaranteed overdraft (10) 2.07∗∗ 0.58 -0.71 0.59 EXP (0.23, 3.91) (-1.80, 2.95) (-2.85, 1.43) (-1.38, 2.56) Guaranteed overdraft (11) 1.04∗∗∗ 0.10 -0.09 -0.20 OBS (Float) (0.66, 1.42) (-0.40, 0.61) (-0.59, 0.42) (-0.69, 0.29) Guaranteed overdraft (12) 1.07∗∗∗ 0.13 -0.12 -0.20 EXP (Float) (0.70, 1.43) (-0.35, 0.60) (-0.62, 0.38) (-0.66, 0.26) Overdraft (13) 9.25∗∗∗ -0.11 -3.84∗ -3.73 OBS (5.24, 13.27) (-4.08, 3.87) (-7.92, 0.23) (-8.25, 0.80) Overdraft (14) 9.03∗∗∗ 1.04 -4.45∗∗ -3.15 EXP (4.84, 13.22) (-2.94, 5.02) (-8.40, -0.49) (-7.77, 1.46) Vendor (15) 0.95∗∗∗ -0.14 -0.10 -0.18 OBS (0.73, 1.17) (-0.39, 0.10) (-0.48, 0.27) (-0.46, 0.10) Vendor (16) 0.96∗∗∗ -0.17 -0.12 -0.14 EXP (0.71, 1.21) (-0.44, 0.11) (-0.54, 0.29) (-0.45, 0.17) Working capital (17) 0.91∗∗ 0.37 0.74 -0.21 OBS ∼365 (0.15, 1.66) (-0.35, 1.09) (-0.46, 1.94) (-0.78, 0.36) Working capital (18) 0.98∗∗ 0.38 0.60 -0.22 EXP ∼365 (0.21, 1.74) (-0.36, 1.13) (-0.69, 1.89) (-0.79, 0.35) Working capital (19) 1.00∗∗∗ -0.13 0.04 -0.07 OBS ∼365 (Float) (0.79, 1.20) (-0.33, 0.08) (-0.20, 0.28) (-0.28, 0.14) Working capital (20) 1.08∗∗∗ -0.14 -0.00 -0.07 EXP ∼365 (Float) (0.86, 1.29) (-0.34, 0.07) (-0.25, 0.24) (-0.28, 0.14) Working capital (21) 1.16∗∗∗ -0.40 0.76∗ -0.22 OBS 365∼ (0.95, 1.37) (-0.89, 0.08) (-0.13, 1.64) (-0.65, 0.22) Working capital (22) 1.25∗∗∗ -0.37 0.68 -0.24 EXP 365∼ (0.98, 1.52) (-0.89, 0.15) (-0.27, 1.63) (-0.69, 0.21) Working capital (23) 0.79∗∗∗ -0.12 0.13 -0.10 OBS 365∼ (Float) (0.53, 1.06) (-0.31, 0.07) (-0.10, 0.36) (-0.27, 0.06) Working capital (24) 0.85∗∗∗ -0.12 0.09 -0.10 EXP 365∼ (Float) (0.58, 1.12) (-0.29, 0.05) (-0.10, 0.29) (-0.26, 0.06) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. 95% confidence interval in parentheses. Estimated with fixed effects. All regressions are controlled by expected inflation and EMBI. ACC is also controlled by the Libor rate. OBS indicates that the explanatory variable in the regression is Selict, while EXP indicates that the explanatory variable is Expecit The differential for non-systemically important financial institutions is statistically significant only for Discount of credit card bills (3 and 4) and Discount of trade bills (7 and 8). The public-owned government banks differential is not significant for any of the NFC types, except Overdraft (14) in the expected Selic regression. The foreign-controlled financial institutions yielded significant differentials only for Discount of credit card bills (3 and 4) and Discount of checks (5 and 6). In all these cases, the pass-through differentials are basically the same for both the observed and expected Selic rates. Notice that, in addition of being negative, all statistically significant differentials are for types with over-proportional pass-through. Despite some statistically significant pass-through differentials, the major findings remained unchanged. However, the negative differentials for NFC loan types indicate that over-proportional pass-through from the baseline estimates might have been driven by S1, private and domestic financial institutions. The first two characteristics are related to market power, market concentration and political interference, which might help to explain the high degrees pass-through according to the discussion in Sect. 5.3. Persistence in the lending rates In order to investigate how potential inertia in the lending rates might affect the interest rate pass-through, we estimate the following dynamic panel-data specification:5 LendingRatem,i,t=ρLendingRatem,i,t-1+(1-ρ)[α+βBaseRatei,t+Ctδ]+εm,i,t where 0<ρ<1 measures the persistence in the lending rates. The other variables and parameters follow the previous definitions. In this set up, (1-ρ)β measures the short-run pass-through while β accounts for the long-run interest rate pass-through. Thus, in the estimation of Eq. (5), we have to identify β in order to compare it with the previous static estimates. In the case of dynamic panels, it is well known that the fixed-effects estimator is inconsistent for typical applications in microeconomic data where there are few time periods and a large number of individuals (here, financial institutions). The estimator bias is caused by correlation between the lagged dependent variable and the unobserved specific heterogeneity. However, the current dataset does not fit this profile because it has a large number of time periods and relatively fewer individuals. Then, correlation induced by the Within transformation vanishes and the fixed-effects estimator becomes consistent according to Bond (2002). Additionally, the Least Squares Dummy Variable estimator generally has the lowest residual mean square error (RMSE) when compared to alternative methods usually applied to dynamic panels, as pointed out by Judson and Owen (1999).22 Table 10 reports the results for alternative models with persistence in lending rates. All aggregate samples revealed full pass-through with estimated coefficients slightly lower than the ones found in the static models.Table 10 Inertia in lending rates and the interest rate pass-through Type Persistence Pass-through Selic (ρ) (β) Overall (1) 0.90∗∗∗ 1.54∗∗∗ OBS (0.85, 0.95) (0.78, 2.30) Overall (2) 0.90∗∗∗ 1.61∗∗∗ EXP (0.85, 0.95) (0.86, 2.36) Households (3) 0.90∗∗∗ 1.64∗∗∗ OBS (0.85, 0.96) (0.73, 2.56) Households (4) 0.90∗∗∗ 1.76∗∗∗ EXP (0.85, 0.96) (0.84, 2.68) Non-financial corporations (5) 0.89∗∗∗ 1.50∗∗ OBS (0.79, 0.98) (0.16, 2.83) Non-financial corporations (6) 0.89∗∗∗ 1.53∗∗ EXP (0.79, 0.98) (0.18, 2.88) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. ρ measures the persistence in the lending rates and β corresponds to the identified long-run interest rate pass-through coefficient according to Eq. (5). 95% confidence interval in parentheses. Estimated with fixed effects. All regressions are controlled by expected inflation and EMBI. OBS indicates that the explanatory variable in the regression is Selict while EXP indicates that the explanatory variable is Expecit Table 11 reports the estimates for the HH and NFC loan types. Basically, most of the baseline results were kept unchanged. The full pass-through is still present for the modalities that presented this result in the static models. Statistical significance of β for Discount of checks (panels 5 and 6), Vehicle financing (19 and 20) and Vehicle leasing (21 and 22) were a bit lower when compared to the estimates from Table 4. The evidence of over-proportional pass-through for Overdraft (9 and 10) was maintained. Payroll-deducted loans to public employees (13 and 14) and Payroll-deducted loans to retirees (15 and 16) revealed incomplete pass-through as in the static models.Table 11 Inertia in lending rates and the interest rate pass-through by loan types Households Non-financial corporations Type Persistence Pass-through Selic Type Persistence Pass-through Selic (ρ) (β) (ρ) (β) CC financing (1) 0.85∗∗∗ 2.03 OBS ACC (1) 0.38∗∗∗ 0.01 OBS (0.78, 0.92) (-1.01, 5.07) (0.29, 0.47) (-0.05, 0.07) CC financing (2) 0.85∗∗∗ 2.15 EXP ACC (2) 0.38∗∗∗ -0.02 EXP (0.78, 0.92) (-1.09, 5.39) (0.29, 0.47) (-0.07, 0.04) CC revolving (3) 0.69∗∗∗ 20.07∗∗ OBS Discount-CC bills (3) 0.89∗∗∗ 3.05∗∗ OBS (0.38, 1.00) (4.44, 35.69) (0.80, 0.98) (0.69, 5.41) CC revolving (4) 0.70∗∗∗ 9.54∗∗ EXP Discount-CC bills (4) 0.89∗∗∗ 3.25∗∗∗ EXP (0.39, 1.01) (0.23, 18.86) (0.82, 0.97) (1.09, 5.41) Discount-checks (5) 0.73∗∗∗ 1.31∗∗ OBS Discount-checks (5) 0.90∗∗∗ 1.32 OBS (0.52, 0.94) (0.24, 2.39) (0.78, 1.02) (-0.26, 2.90) Discount-checks (6) 0.73∗∗∗ 1.46∗∗ EXP Discount-checks (6) 0.90∗∗∗ 1.40∗ EXP (0.52, 0.93) (0.33, 2.58) (0.79, 1.02) (-0.15, 2.95) Other goods financing (7) 0.86∗∗∗ 1.88∗∗∗ OBS Discount-trade bills (7) 0.78∗∗∗ 1.62∗∗∗ OBS (0.78, 0.94) (0.55, 3.22) (0.66, 0.90) (0.83, 2.42) Other goods financing (8) 0.86∗∗∗ 1.76∗∗∗ EXP Discount-trade bills (8) 0.78∗∗∗ 1.72∗∗∗ EXP (0.78, 0.94) (0.43, 3.10) (0.66, 0.90) (0.83, 2.61) Overdraft (9) 0.92∗∗∗ 7.38∗∗∗ OBS Guaranteed overdraft (9) 0.36∗∗∗ 2.20∗∗∗ OBS (0.86, 0.98) (2.89, 11.87) (0.17, 0.55) (0.73, 3.68) Overdraft (10) 0.92∗∗∗ 7.87∗∗∗ EXP Guaranteed overdraft (10) 0.36∗∗∗ 2.13∗∗∗ EXP (0.86, 0.98) (3.09, 12.65) (0.17, 0.55) (0.66, 3.61) Payroll-deducted (11) 0.91∗∗∗ 0.93∗∗∗ OBS Guaranteed overdraft (11) 0.53∗∗∗ 0.96∗∗∗ OBS - Private (0.87, 0.96) (0.43, 1.43) (Float) (0.35, 0.71) (0.54, 1.38) Payroll-deducted (12) 0.91∗∗∗ 0.99∗∗∗ EXP Guaranteed overdraft (12) 0.54∗∗∗ 1.00∗∗∗ EXP - Private (0.87, 0.95) (0.51, 1.46) (Float) (0.36, 0.72) (0.57, 1.43) Payroll-deducted (13) 0.91∗∗∗ 0.63∗∗∗ OBS Overdraft (13) 0.91∗∗∗ 7.15∗ OBS - Public (0.86, 0.96) (0.36, 0.91) (0.81, 1.01) (-1.36, 15.66) Payroll-deducted (14) 0.91∗∗∗ 0.64∗∗∗ EXP Overdraft (14) 0.91∗∗∗ 7.30∗ EXP - Public (0.87, 0.96) (0.40, 0.88) (0.81, 1.01) (-0.97, 15.57) Payroll-deducted (15) 0.93∗∗∗ 0.47∗∗∗ OBS Vendor (15) 0.61∗∗∗ 0.80∗∗∗ OBS - Retirees (0.90, 0.96) (0.18, 0.76) (0.39, 0.83) (0.27, 1.33) Payroll-deducted (16) 0.94∗∗∗ 0.50∗∗∗ EXP Vendor (16) 0.62∗∗∗ 0.81∗∗∗ EXP - Retirees (0.91, 0.96) (0.21, 0.79) (0.42, 0.83) (0.30, 1.33) Personal credit (17) 0.69∗∗∗ 1.98∗∗ OBS Working capital (17) 0.31∗∗ 1.20∗∗∗ OBS (0.61, 0.76) (0.20, 3.77) ∼365 (0.03, 0.59) (0.64, 1.76) Personal credit (18) 0.69∗∗∗ 1.93∗ EXP Working capital (18) 0.31∗∗ 1.23∗∗∗ EXP (0.61, 0.76) (-0.02, 3.88) ∼365 (0.03, 0.59) (0.63, 1.83) Vehicle financing (19) 0.89∗∗∗ 0.52∗∗ OBS Working capital (19) 0.27∗∗∗ 0.90∗∗∗ OBS (0.79, 0.98) (0.05, 0.99) ∼365 (Float) (0.16, 0.39) (0.69, 1.10) Vehicle financing (20) 0.89∗∗∗ 0.57∗∗ EXP Working capital (20) 0.27∗∗∗ 0.96∗∗∗ EXP (0.79, 0.98) (0.06, 1.08) ∼365 (Float) (0.16, 0.38) (0.75, 1.17) Vehicle leasing (21) 0.43∗∗∗ 0.61∗∗ OBS Working capital (21) 0.63∗∗∗ 1.08∗∗∗ OBS (0.21, 0.65) (0.01, 1.21) 365∼ (0.51, 0.75) (0.66, 1.50) Vehicle leasing (22) 0.43∗∗∗ 0.65∗ EXP Working capital (22) 0.63∗∗∗ 1.16∗∗∗ EXP (0.21, 0.66) (-0.03, 1.32) 365∼ (0.51, 0.75) (0.67, 1.64) Working capital (23) 0.34∗∗∗ 0.78∗∗∗ OBS 365∼ (Float) (0.23, 0.46) (0.56, 1.00) Working capital (24) 0.35∗∗∗ 0.83∗∗∗ EXP 365∼ (Float) (0.23, 0.47) (0.60, 1.05) ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. ρ measures the persistence in the lending rates and β corresponds to the identified long-run interest rate pass-through coefficient according to Eq. (5). 95% confidence interval in parentheses. Estimated with fixed effects. All regressions are controlled by expected inflation and EMBI. CC revolving is also controlled by the structural change in the rules of this loan type, and ACC is also controlled by the Libor rate. OBS indicates that the explanatory variable in the regression is Selict, while EXP indicates that the explanatory variable is Expecit For the NFC loans, most of the previous static findings were also held in the dynamic panel data environment, as reported in Table 11. The degree of pass-through is not statistically significant for Advances on exchange contracts (1 and 2), Discount of checks (5 and 6) and Overdraft (13 and 14). For all other loan types, the estimated values and significance levels of β were very close to the ones from the static models. However, in the dynamic environment, there is over-proportional pass-through only for Discount of credit card bill (4) when regressed against the expected Selic rate. Lending rates are highly persistent for most types, as indicated by the estimates of ρ. All R2 coefficients are much higher than in the static models.23 This was expected since inertia is an important component of the lending rates, increasing the explanatory power of the regressions. Overall, the major results are robust to the alternative dynamic panel data specification, despite the high persistence in most of the interest rate loan types. This finding, coupled with high interest rate margins, full (or over-proportional) and positively asymmetric pass-through contribute to explain the historically high levels of loan interest rates in the Brazilian economy. In the next section, we lay out some explanations that might help to understand the financial institutions’ behavior. Discussion The findings of full (or over-proportional) and positively asymmetric pass-through coupled with high interest rate margins and highly persistent lending rates might be assessed by complementary explanations from the literature. One is the traditional structure-conduct-performance hypothesis arguing that market power creates an environment that affects the banks’ behavior and performance in unfavorable ways from a social perspective (Berger et al. 2004). When borrowers are subjected to collusive price arrangements, banks may react differently to upward and downward movements in the policy rate. Notwithstanding this hypothesis is extensively used in studies of bank spreads, concentration, and other competition measures, it is also prominent in the interest rate pass-through literature. Collusive behavior can occur due to the costs that borrowers incur in switching loans from a bank to another. Switching costs are one source of market power which affects bank competition. While these costs induce bank competition to enlarge customer base by capturing new clients with lower lending rates, the spreads raise to the borrowers once they are locked in (Carletti 2008). There are evidences of significant switching costs in Brazilian private banks, meaning that the longer is the duration of the relationship with the borrower, the higher is the spread (Ornelas et al. 2020). In case of collusive price arrangements, expected costs of breakdown should lead to a slowdown in pass-through (Cottarelli and Kourelis 1994; Hannan and Berger 1991), unless the interest rate change results in higher gains. Thus, lending rates would be less likely to respond to a decrease than to an increase in the policy rate, or in the expected policy rate. This asymmetric behavior by banks was successfully identified in our previous findings. While switching costs directly affects borrowers, adjustment costs are charged on lenders. However, borrowers’ behavior against changes in lending interest rates might affect the pass-through and persistence of these rates. Adjustment costs are associated to more sluggishness of the pass-through as the market become less competitive because banks are more capable of smoothing their loan adjustments over time (Kopecky and Hoose 2012). Hannan and Berger (1991) claim that negative customer’s reaction (here, borrower’s reaction) to unstable prices, coupled with a more negative reaction to unfavorable price changes (increases in the lending rate), imply a higher rigidity in pass-through. On the other hand, in the presence of fixed adjustment costs, the lending rates will be adjusted only if these costs are lower than the costs of keeping them unchanged (Banerjee et al. 2013; Cottarelli and Kourelis 1994). This claim is reinforced by Hofmann and Mizen (2004), who found that nonlinearities in the adjustment of retail rates to changes in base rates have arisen from menu cost models. In our sample, where changes were relatively frequent and interest rate margins high, extra surplus by increasing lending rates could have overcome adjustment costs. The relevance of these costs depends on the demand elasticity for bank loans (Cottarelli and Kourelis 1994), but this issue is beyond the scope this study. If the gains surpass the costs of adjusting the lending rate, then banks might have incentive to a full or even over-proportional interest rate pass-through. As gains are supposed to be greater after rising lending rates, this would lead to distinct strategies of upward and downward movements in response to changes in the policy rate (or in the expected policy rate). Adjustment costs and extra surplus might also play a central role when there is a perception that changes in money market rates (or in policy rates) would be temporary (Cottarelli and Kourelis 1994). At the beginning of our sample, throughout 2012, there was a fall in the Over-Selic rate perceived by the financial sector as inconsistent with the inflation targeting regime in place. This is the case because the Central Bank of Brazil implemented a monetary policy easing starting on August 2011 and reduced the Selic rate in the following nine Copom meetings, stopping the drop only on November of 2012. Meanwhile, the expected inflation for 2012 was above the target and increasing for 2013, which would have required the Central Bank to increase instead of reducing the policy interest rate.24 In the following years, the policy rate climbed once again, and expectations by financial institutions indicated that the policy rate could have reached higher levels in 2016, when it peaked in our sample (Fig. 3). There might have been some lack of confidence in the monetary authority during this period, and banks preferred not to fully pass-through movements in the Over-Selic rate to lending rates fearing sudden changes in the monetary policy conduction. This behavior might explain the high persistence in lending rates and asymmetric movements in cases where extra surplus were higher than the costs of changing lending rates. An alternative hypothesis to the market power is the efficient structure hypothesis (e.g., Berger et al. 2004; Berger and Hannan 1989). It posits that differences in firm-specific efficiencies within markets create unequal market shares and high levels of concentration (Berger and Hannan 1989). Thus, concentration would be endogenous and, as well as performance, stem from high market share of firms that are efficient. We argue that, under this view, banks would also be efficient in adjusting lending rates after changes in the policy rate or, at least, would incur in lower adjustment costs. It might be added that over-proportional pass-through was stronger in the loan types with the highest interest rate margins. Presumably, these loan operations should have a wider interval to adjust their interest rates. Therefore, specific elements of market power, bank concentration, lack of competition and bank efficiency should be put together to adequately assess the over-proportional and positively asymmetric pass-through to highly persistent lending rates. These striking pass-through features contribute to explain why loan interest rate are so high in Brazil. Which market imperfection will prevail to account for the banks’ behavior, however, is an empirical issue that shall be tested against the data and is left for further research. Conclusion This paper investigated the interest rate pass-through from the observed and expected policy rates to the remarkably high lending interest rates in the Brazilian economy, accounting for financial institutions specific characteristics, asymmetric adjustment and persistence in the loan rates. We used a unique and non-public dataset with identified Over-Selic expectations by financial institutions, which reduces loss of information that would be caused by aggregation of expectations by the mean or median. The sample covers the period from January 2012 to April 2019, on weekly basis, with variability by loan types, financial institutions and time. In addition to the standard static specification, we also accounted for partial adjustment of the lending rates in response to changes in both observed and expected Over-Selic rates in a dynamic panel data environment. The results provided robust evidence of full (and over-proportional) pass-through from the observed and the expected policy rates to the lending interest rates. For some loan operations, we found an asymmetric behavior by the financial institutions, as captured by smaller degrees of pass-through for decreases than for increases in the observed or expected policy rates. For the overall sample, sub-samples by households and non-financial corporations and specific loan types, there is evidence of over-proportional pass-through, meaning that increases in lending rates overpass any raise in the policy interest rate, either observed or expected. Loan types with the highest interest rate margins also revealed over-proportional degrees of pass-through. In general, the higher the interest rate margin, the bigger the degree of pass-through from both observed and expected policy rate. These findings are robust to the inclusion of other control variables, such as specific characteristics by size, ownership type and capital origin, as well as to a dynamic panel data specification. In fact, the loan interest rates are highly persistent and the long run pass-through closely resembles the short run estimates from the static models. When addressing the pass-through, one should account for heterogeneity in the loan types, as the interest rate margins, degrees of pass-through and asymmetry might vary considerably among them. A common feature, however, is that financial institutions anticipate adjustments in their lending rates by correctly forecasting the next target level of the policy interest rate. This price-setting strategy, coupled with persistently high margins, full (or over-proportional) and positively asymmetric pass-through contribute to explain the remarkably high loan interest rates in the Brazilian economy. The economic reasoning behind the financial institutions’ behavior, however, demands complementary explanations from the specialized literature. Elements of market power, market concentration, lack of competition and other frictions should be theoretically addressed and empirically tested in an integrated environment. Expectation formation, on its turn, might be affected by forward guidance of the monetary policy, as the Central Bank communication might affect the economy even in the absence of changes in the short-term policy rate. The Brazilian case is worth investigating to check whether the effects of forward guidance are as strong as the conventional monetary policy.25 These suggestions, however, are left for further research. Appendix A: Description of the loan types Table 12 Description of the loans modalities Type Description Credit card financing Installment loans financed by the card issuer with incidence of interest. These operations are linked to financed purchases or to refinanced credit card balances. This type includes also cash withdrawals that generate scheduled installment payments Credit card revolving credit Financing of the outstanding credit card balance (remaining after payment due date) or cash withdrawals that generate one payment due at next credit card bill Other goods financing Financing of goods, except vehicles, for consumption of households contractors Overdraft Revolving credit line related to checking accounts, in which limited funds are made available for customers to use discretionarily and for short periods, through withdrawals, checks, payments or bank transfers. In such transactions, the outstanding debt balance must be promptly amortized whenever there is any deposit to the checking account. This type includes situations where the negative balance exceeds the authorized overdraft limit Payroll-deducted personal loans—to private sector employees Credit for non-government employees, in which part of their salaries or wages is withheld by the employer in order to pay the loan installments to the lending institutions Payroll-deducted personal loans—to public sector employees Credit to government employees (federal, state or local; active or inactive) in which part of their wage or retirement income is withheld by the public entities in order to pay the loan installments to the lending institutions Payroll-deducted personal loans—to retirees and pensioners Loans to retirees or pensioners of the National Institute of Social Security (INSS), in which part of their monthly stipends is withheld by INSS in order to pay the loan installments to the lending institutions Personal credit Credit to individuals not bound to any specific destination and without withholding wages for the payment of loan installments (i.e., no payroll-deducted) Vehicle financing To consumption of households contractors. The contract must contain a lien clause, with the financed good constituting the guarantee. Funding for vehicles intended for commercial stocks are not classified in this type of credit Vehicle leasing Finance lease operations, where the lessor grants the lessee the use of the object of the lease (vehicles), with a purchase option at the end of the contract Advance on exchange contracts (ACC) Partial or total advance of funds linked to export contracts, in order to finance the production of export goods. This type includes operations of advances on delivered exchange contracts (ACE) Discount of credit card bills Advance of funds to non-financial corporations based on future cash flows linked to receivables from credit card bills Discount of checks Advance of funds to non-financial corporations based on future cash flows linked to checks Discount of trade bills Advance of funds to non-financial corporations based on future cash flows linked to trade bills or other receivables, except checks and credit card bills Guaranteed overdraft accounts Revolving credit related to bank accounts of non-financial corporations, in which limited funds are made available for customers to use, whether by running the checking account or by formally requesting to the financial institution, which may eventually seek binding guarantees from receivables, or other collaterals. This type includes situations where the negative balance exceeds the authorized overdraft limit Vendor Sales financing transaction where the borrowing company (seller) to finance their sales and to get immediately paid by the financial institution. The buyer commits itself to the payment schedule which will settle the transaction with the financial institution. In general, the financial institution will hold the receivables of the selling company, which undertakes the risk of the operation Working capital up to 365 days Short-term credit to finance operating activities of non-financial corporations, related to a specific contract that establishes deadlines, fees and guarantees. Its maturity may not exceed 365 days Working capital over 365 days Medium and long term credit to finance the operating activities of non-financial corporations, related to a specific contract that establishes deadlines, fees and guarantees. Its maturity should be above 365 days Source: Central Bank of Brazil Funding This work was funded by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and Fundação de Apoio a Pesquisa do Distrito Federal (FAP-DF). J. A. Divino has received financial support from CNPq and CAPES (Grant Numbers 302632-2019-0 and 760/2018, respectively) and Carlos Haraguchi from CAPES (Grant Number 88887.201766/2018-00). Declarations Conflict of interest Author J. A. Divino declares that he has no conflict of interest. Author Carlos Haraguchi declares that he has no conflict of interest. Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors. 1 Banerjee et al. (2013) used aggregate data for the four major Euro area economies and argued that banks anticipate short-term market rates when setting interest rates on loans and deposits, and even more so when they will have to refinance the loans that they make in the future. We found a similar result by using a loan-specific dataset with expected policy interest rate identified by professional forecasters. 2 From the Latin American region, the panel included Colombia, Jamaica, Mexico and Venezuela, but not Brazil. 3 See Gregor et al. (2021) for a comprehensive review of the pass-through literature. 4 According to Schmeling et al. (2022), Cieslak (2018) and Divino and Haraguchi (2022), the financial institutions forecast ability of the policy rate depends on the knowledge of the interest rate rule followed by the Central Bank. We highlight that similar results for observed and expected interest rate pass-through indicate anticipated adjustments in the lending rates, but with room for some misalignment due to the asymmetric pass-through. 5 An alternative approach claims that the financial institutions forecasting strategy of the policy rate might rely on a pro-conservative monetary policy convention in Brazil. See Bresser-Pereira et al. (2020) for details. 6 The Over-Selic rate is the daily average of the overnight rates of interbank loans backed by federal securities, carried out in the Special System for Settlement and Custody (the Selic System). 7 It is available from the Open Data Portal https://opendata.bcb.gov.br/, from where we also extracted the observed and expected Over-Selic rates. Data on the Monetary Policy Committee meetings and financial institutions were obtained from Central Bank of Brazil website https://www.bcb.gov.br/en. 8 See Appendix A for a detailed description of the loan types. 9 Focus Survey monitors the market expectations for several economic indicators, including Selic target level and inflation rate. 10 The confidential financial institutions codes list was kindly provided by Department of Statistics (DSTAT) of the Central Bank of Brazil only for the purposes of this work. 11 Estimates using median Selic expectations were significantly different from those with Selic expectations identified by financial institutions, especially in models with disaggregated loan operations. The higher the disaggregation in the sub-samples, the bigger the difference in the estimated pass-through coefficients between the median Selic expectations and the identified expectations by financial institutions. These results are available from the authors upon request. 12 See BCB’s methodological notes in https://www.bcb.gov.br/content/statistics/methodologicalnotes_docs/financialsystemloans/notaempri.pdf and https://www.bcb.gov.br/content/statistics/methodologicalnotes_docs/financialsystemloans/notaempr201502i.pdf. 13 As a robustness check, we also used winsorized data by setting the top 3% to the 97th percentile. The results were similar and are available from the authors upon request. 14 National Monetary Council Resolution 4549 of 2017 (http://www.bcb.gov.br/pre/normativos/busca/downloadNormativo.asp?arquivo=/Lists/Normativos/Attachments/50330/Res_4549_v1_O.pdf) states that the outstanding balance in the credit card invoice, once not completely paid at the due date, may be financed by revolving credit only until the next invoice. This measure led consumers to settle down the debt in full, to pay it in instalments, or to seek more advantageous credit sources for financing the debt. The new rule has become effective in April 3, 2017. 15 Advances on exchange contracts is a credit type directed at foreign trade, mainly to advance funds to exporters before payment by importers. Financial institutions that offer this type of credit line obtain funds from abroad and charge interest rates indexed to credit costs in the international markets. As stated earlier, it is included as a placebo in the analysis by loan rate type because no pass-through should be observed from the domestic interest rates. 16 We do not control for credit risk because this information is confidential and not released by loan type and financial institution. The Central Bank of Brazil computes loan ratings and borrower ratings for every new loan in the credit registry system (SCR). However, the SCR is strictly confidential and subject to specific rules and special authorization to be accessed. We only had access to monthly default rates for some loan types that did not match our weekly-basis sample. While controlling for credit risk of loan operations is relevant to explain interest rate margins (or spread), this might also be the case in the estimation of the degree of pass-through. However, the correlation between the credit risk by loan type and the Over-Selic rate (observed and expected) might not be strong enough to bias the pass-through estimates, an issue that deserves further investigation depending on data availability. 17 Kopecky and Hoose (2012) developed a dynamic adjustment cost model with imperfect competition where bank retail deposit and loan rates depend on own lagged values and on lagged, current, and expected future values of the security rate, but without providing further empirical evidence. The problem with applying this framework is that the observed Over-Selic rate varies only over time and is highly correlated with the expected rate, which changes over time and by financial institutions. This prevented us from including both observed and expected Over-Selic rates in a unique panel-data pass-through regression. The results were meaningless and are available from the authors upon request. 18 In a robustness check, we applied the random effects specifications to all regressions and there was no significant change in the results, which are available from the authors upon request. 19 As explained earlier, funding for this type comes from abroad and is not related to the domestic interest rates. 20 The Central Bank of Brazil Banking Report 2018 brings a decomposition of the average cost of outstanding loans in which delinquency—losses arising from non-payment of debts or interest and discounts granted—represented 23% of the total cost and 37% of the spread in the last three years. The report is available at https://www.bcb.gov.br/content/publications/bankingreport/BAR_2018.pdf. 21 The Central Bank of Brazil established the S1 segmentation for proportional implementation of prudential regulation to prevent any “domino effect” in the financial system. It is composed of financial institutions with the largest market shares in addition to other features, as explained in Sect. 2. The S1 institutions (Banco do Brasil, Bradesco, BTG Pactual, Caixa Econômica Federal, Itau, and Santander) accounted for 80.45% market share in outstanding credit for households and 58.24% share in outstanding credit for non-financial corporations in a universe of 172 authorized institutions, according to the Central Bank of Brazil Banking Report from 2018 (available at https://www.bcb.gov.br/content/publications/bankingreport/BAR_2018.pdf). 22 We also applied the traditional Arellano and Bond (1991) estimator, but the coefficient of the lagged dependent variable did not lie within the bounds defined by the OLS and Within estimators, indicating that these estimates are not reliable according to Bond (2002) and Roodman (2009). Another practical issue is that a large number of time periods adds too many instrumental variables to the IV matrix and generates a dimensionality problem that requires some sort of arbitrary truncation. By using a fixed-effects estimator, we also avoid this issue. 23 This is especially evident for the aggregate samples in Table 10, Credit card financing, Other goods financing, Personal credit, and Guaranteed overdraft (fixed rate) in Table 11. For the HH types, the estimates of ρ ranged from 0.69 to 0.94, except for Vehicle leasing, where it was 0.43. NFC rates showed lower estimated values of ρ, ranging from 0.27 to 0.91. R2 coefficients are available from the authors upon request. 24 Expected inflation ranged from 4.85% to 5.71% for 2012 and from 5.00% to 5.60% for 2013, while the inflation target was 4.5% for both years, according to Focus Survey (available at https://www3.bcb.gov.br/expectativas2/#/consultas) and the inflation targeting track record (available at https://www.bcb.gov.br/en/monetarypolicy/historicalpath). 25 See Ferreira (2022) for a recent empirical evidence for the US economy. We are grateful to Central Bank of Brazil, and in particular Fernando Rocha, Luciana Roppa, and Monica Une from Department of Statistics (DSTAT), and Cassio Silva from Information Technology Department (Deinf), for providing crucial data used in this work. We would like also to thank comments from Osvaldo Candido, Thiago Silva, Andre Minella, Joao Mello, Jose Renato Ornelas, Anderson Okinokabu, Sergio Leao, Thiago Trafane and from participants in the 2022 Econometric Society European Meeting (EEA-ESEM 2022), 42nd Meeting of the Brazilian Econometric Society (SBE) and of the Central Bank of Brazil Research Network Workshop. C. Haraguchi thanks CAPES Foundation and J. A. Divino thanks CNPq for financial support. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-Brasil (CAPES)—Finance Code 001. The authors acknowledge the Fundação de Apoio a Pesquisa do Distrito Federal (FAP-DF) for the financial support. The views expressed in the paper are those of the authors and do not necessarily reflect those of the Central Bank of Brazil. All remaining errors are the authors’ sole responsibility. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Altavilla C Canova F Ciccarelli M Mending the broken link: heterogeneous bank lending rates and monetary policy pass-through J Monet Econ 2020 110 81 98 10.1016/j.jmoneco.2019.01.001 Arellano M Computing robust standard errors for within-groups estimators Oxford Bull Econ Stat 1987 49 431 434 10.1111/j.1468-0084.1987.mp49004006.x Arellano M Bond S Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations Rev Econ Stud 1991 58 277 297 10.2307/2297968 Arena M, Reinhart C, Vázques F (2007) The lending channel in emerging economies: are foreign banks different?. Working Paper 48 International Monetary Fund. https://www.elibrary.imf.org/view/journals/001/2007/048/article-A001-en.xml Banerjee A Bystrov V Mizen P How do anticipated changes to short-term market rates influence banks’ retail interest rates? Evidence from the four major euro area economies J Money Credit Bank 2013 45 1375 1414 10.1111/jmcb.12056 Berger AN Demirgüç-Kunt A Levine R Haubrich JG Bank concentration and competition: an evolution in the making J Money Credit Bank 2004 36 433 451 10.1353/mcb.2004.0040 Berger AN, Hannan TH (1989) The price-concentration relationship in banking. Rev Econ Stat, pp 291–299. 10.2307/1926975 Bond SR Dynamic panel data models: a guide to micro data methods and practice Port Econ J 2002 1 141 162 10.1007/s10258-002-0009-9 Bresser-Pereira LC de Paula LF Bruno M Financialization, coalition of interests and interest rate in Brazil Revue de la régulation 2020 10.4000/regulation.16636 Carletti E (2008) Handbook of financial intermediation and banking. Handbooks in Finance chapter Competition and regulation in banking, pp 449–482. Elsevier Castro PHR, Mello JMP (2012) Is the bank interest rate pass-through of selic rate movements asymmetric? Braz Rev Econom, 32, 03–30. 10.12660/bre.v32n12012.2967 Chong BS Interest rate deregulation: monetary policy efficacy and rate rigidity J Bank Finance 2010 34 1299 1307 10.1016/j.jbankfin.2009.11.026 Cieslak A Short-rate expectations and unexpected returns in treasury bonds Rev Financ Stud 2018 31 3265 3306 10.1093/rfs/hhy051 Coelho CA, de Mello JMP, Garcia MGP (2010) Identifying the bank lending channel in brazil through data frequency. Economía, 10:47–79. https://www.jstor.org/stable/25800046 Cottarelli C, Kourelis A (1994) Financial structure, bank lending rates, and the transmission mechanism of monetary policy. Working Paper 39 International Monetary Fund. 10.2307/3867521 Divino JA, Haraguchi C (2022) (Mis)alignment between observed and expected monetary policy. Unpublished Paper, Catholic University of Brasilia Ferreira LN Forward guidance matters: disentangling monetary policy shocks J Macroecon 2022 73 103423 10.1016/j.jmacro.2022.103423 Gambacorta L Mistrulli PE Bank heterogeneity and interest rate setting: what lessons have we learned since Lehman Brothers? J Money Credit Bank 2014 46 753 778 10.1111/jmcb.12124 Gregor J Melecký A Melecký M Interest rate pass-through: a meta-analysis of the literature J Econ Surveys 2021 35 141 191 10.1111/joes.12393 Hannan TH Berger AN The rigidity of prices: evidence from the banking industry Am Econ Rev 1991 81 938 Hofmann B Mizen P Interest rate pass-through and monetary transmission: evidence from individual financial institutions’ retail rates Economica 2004 71 99 123 10.1111/j.0013-0427.2004.00359.x Holton S d’Acri CR Interest rate pass-through since the euro area crisis J Bank Finance 2018 96 277 291 10.1016/j.jbankfin.2018.08.012 Hristov N Hülsewig O Wollmershäuser T The interest rate pass-through in the euro area during the global financial crisis J Bank Finance 2014 48 104 119 10.1016/j.jbankfin.2014.08.004 Judson RA Owen AL Estimating dynamic panel data models: a guide for macroeconomists Econ Lett 1999 65 9 15 10.1016/s0165-1765(99)00130-5 Kopecky KJ Hoose DDV Imperfect competition in bank retail markets, deposit and loan rate dynamics, and incomplete pass through J Money Credit Bank 2012 44 1185 1205 10.1111/j.1538-4616.2012.00527.x Liu M-H Margaritis D Tourani-Rad A Monetary policy transparency and pass-through of retail interest rates J Bank Finance 2008 32 501 511 10.1016/j.jbankfin.2007.06.012 Neumark D Sharpe SA Market structure and the nature of price rigidity: evidence from the market for consumer deposits Q J Econ 1992 107 657 680 10.2307/2118485 Ornelas JRH, Silva MSd, Van Doornik BFN (2020) Informational switching costs, bank competition and the cost of finance. Working Paper Series 1104 Inter-American Development Bank. 10.18235/0002508 Pereira CM Maia-Filho LF Brazilian retail banking and the 2008 financial crisis: were the government-controlled banks that important? J Bank Finance 2013 37 2210 2215 10.1016/j.jbankfin.2012.03.009 Roodman D How to do xtabond2: an introduction to difference and system GMM in Stata Stand Genomic Sci 2009 9 86 136 10.1177/1536867x0900900106 Santos JAC Bank corporate loan pricing following the subprime crisis Rev Financ Stud 2011 24 1916 1943 10.1093/rfs/hhq115 Schmeling M Schrimpf A Steffensen SA Monetary policy expectation errors J Financ Econ 2022 146 841 858 10.1016/j.jfineco.2022.09.005 Von Borstel J Eickmeier S Krippner L The interest rate pass-through in the euro area during the sovereign debt crisis J Int Money Financ 2016 68 386 402 10.1016/j.jimonfin.2016.02.014	36474602	PMC9716535	NO-CC CODE	2022-12-03 23:20:58	no	Empir Econ. 2022 Dec 2;:1-44	utf-8	Empir Econ	2,022	10.1007/s00181-022-02335-0	oa_other
==== Front Acta Polit Acta Politica 0001-6810 1741-1416 Palgrave Macmillan UK London 273 10.1057/s41269-022-00273-4 Original Article Participatory budgeting and the perception of collective empowerment: institutional design and limited political interference http://orcid.org/0000-0002-6627-5598 Gherghina Sergiu [email protected] 1 http://orcid.org/0000-0001-6476-2563 Tap Paul 2 Soare Sorina 3 1 grid.8756.c 0000 0001 2193 314X Department of Politics, University of Glasgow, Glasgow, UK 2 grid.7399.4 0000 0004 1937 1397 Department of International Studies and Contemporary History, Babes-Bolyai University, Cluj-Napoca, Romania 3 grid.8404.8 0000 0004 1757 2304 Department of Political Science, University of Florence, Florence, Italy 2 12 2022 118 15 11 2022 © Springer Nature Limited 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Participatory budgeting gains momentum around the world, and increasing evidence provides mixed results about its effects. Under these circumstances, it is unclear if citizens consider it a source of empowerment and an avenue for effective decision-making in the life of their local community. We know very little about how participants in participatory budgeting perceive the collective empowerment. This article seeks to identify the factors that shape these perceptions about the empowerment potential of participatory budgeting. It focuses on the critical case of Cluj-Napoca and uses 25 semi-structured interviews conducted in October–November 2020 with three categories of participants. Our findings indicate that participants acknowledge the potential for collective empowerment and praise the limited political involvement but identify design issues and resource allocation as weakening the empowerment potential. Keywords Participatory budgeting Citizens Empowerment Political involvement Romania ==== Body pmcIntroduction The theory and practice of participatory budgeting (PB) indicates concrete avenues to improve the legitimacy, transparency, accountability, and government effectiveness at local level (Sintomer et al. 2008; Wampler 2012; Brun-Martos and Lapsley 2017; Wampler et al. 2018). PB is regularly portrayed as a means to enhance democratic quality and an opportunity for mobilizing a ‘countervailing power’ able to neutralize the power-advantages of political actors (Fung and Wright 2003a). It engages ordinary citizens in specific co-governance arrangements related to the allocation of budgets for local projects and provides more responsive and equitable budgets that meet community needs (Wampler 2012). PB allows experienced government and administrative representatives to share political power with common citizens with the overarching goal to contribute to raising the level of knowledge of the citizens and, more in general, their civic awareness and education (Smith 2009; Talpin 2011, 2012). The international organizations support PB as a model of good governance and a “citizenship school” educating, and engaging citizens in politics between elections (Shah 2007). Several scholars consider PB as a core element among the remedies to the crisis of representative democracy and a potential successful strategy for collective empowerment (Smith 2009; Geissel and Newton 2012; Baiocchi and Ganuza 2014). However, there is also research about the unintended consequences and risks linked to the capture of the process by politicians or interest groups that reproduce social, economic, and political hierarchies (Fung and Wright 2003b; Cabannes 2004; Shah 2007; Marian 2018; Wampler et al. 2018; Williams and Waisanen 2020). This mixed evidence makes us wonder if citizens who engage in PB consider this process as a source of empowerment by learning about their rights and by expressing their views to shape policies. Or, on the contrary, the process creates the illusion of empowerment, being hijacked by political actors. In this sense, it can provide a compromised reality marked by a wide space of maneuver for “politics as usual” that transforms PB in an end in a itself (Baiocchi and Ganuza 2014) and steers attention away from the need of further political innovation and institutional improvement (Röcke 2014; Peck and Theodore 2015). We define empowerment as the possibility to tie action to discussion. That is, the process through which citizens’ proposals are transformed into policies through the coupling with government actions (Fung and Wright 2003b). It is about the extent to which public deliberation, demands, and preferences—through voting on PB projects—are translated into local reforms. PB is mainly about collective empowerment although individual empowerment is also possible for specific segments in society (Hajdarowicz 2018). Earlier research points in the direction of an empowerment dimension for the PB especially linked with small discretionary budgets (Baiocchi and Ganuza 2014). While this dimension is investigated as a general process embedded in a broader context (Alves and Allegretti 2012; Pin 2017), very little is known so far about how participants to PB perceive the collective empowerment. There is limited research about what makes participants to the PB consider this process as an empowering avenue that provides citizens an effective voice in the decision-making of their local community. This perception is important for at least three reasons: (1) it can provide information about what goes well and what can be improved in PB, which reflects the continuity and success of the process (Smith 2009; Åström and Grönlund 2012); (2) it serves as a point of departure to understand citizens’ engagement in community issues; and (3) it complements the objective assessment of empowerment available in the literature. The focus on perception is in line with earlier research about the importance of feelings and emotions for deliberative practices (Steiner 2012). This article aims to address this gap in the literature and seeks to identify the factors that shape the perception of participants to PB about its empowerment potential. We focus on the critical case of Cluj-Napoca where the local political authority initiated and maintained PB for several years. The mayor, in office for many years, is a strong supporter of PB and started a pilot project at district level in 2013, which then continued at city level between 2017 and 2021 with a break in 2020 due to the COVID-19 restrictions. The city-level PB uses a digital platform on which projects are uploaded and votes are cast, with thousands of citizens getting involved over the years. Our qualitative analysis uses 25 semi-structured interviews conducted with three categories of participants actively involved in the 2019 PB: ordinary participants, applicants, and consultants. The findings of our inductive thematic analysis indicate that the context and the institutional design shape the perception about the empowerment potential of PB. On the one hand, the citizens assess positively the possibility for collective empowerment through PB and appreciate the limited political involvement in the process. On the other hand, the participants to the PB claim that the empowerment is limited due to the absence of real deliberation, the use of online platforms that excludes several segments of society and the small-scale funding. Overall, most participants have a nuanced and complex perception of the PB acknowledging both the advantages and problems of the process. Beyond their discourse, their participatory behavior on a continuous basis reflects that they value the PB process despite its downsides. The following section reviews the literature about PB and empowerment with emphasis on potential drivers and obstacles. Next, we provide details about case selection, data, and methodology. The third section briefly describes the PB process in Cluj-Napoca, while the fourth section presents the results of our inductive thematic analysis for interviews. The discussion and conclusions summarize the key findings and reflect on the implications of this analysis for the broader field of study. PB, empowerment, and citizens’ perceptions Three decades after the initiation of PB in Porto Alegre (Brazil), the procedure through which citizens can decide how to spend a share of the local budget has gained momentum (Dias et al. 2019). It emerged as a participatory innovation that intertwines state and civil society with the goal to increase citizens’ propensity to pay taxes and to effectively redistribute municipal funds (De Sousa Santos 1998; Novy and Leubolt 2005). Since then, variations of PB were embraced by thousands of local communities around the world and the allocated budgets have gradually grown (Cabannes and Lipietz 2018). It is a process of community decision-making in which a share of the budget is directly allocated by residents. PB provides a voice to those who contribute to the budget and who are affected by how money are spent (Cabannes 2004). The PB can become “citizenship schools” where ordinary people develop better social and cognitive skills with notable “spillover effects” in terms of further civic society activities (Nylen 2002; Fung and Wright 2003b; Shah 2007). Citizens can voice their view and shape the agenda, increase their abilities to understand the policy agenda, assess government performance, and keep the local authorities accountable. The PB experiences help transforming the local political culture, provide a space for sustainable forms of nonelite grass-roots activism, and foster the civic consciousness (Nylen 2002). In the context of participatory processes, empowerment can be broadly understood as the ability to make or influence a decision that would otherwise not take place. The usual categorization of empowerment is the dichotomy between individual and collective. Individual empowerment refers to the capacity of each person to make a change. Collective empowerment brings individuals together in an effort through which they can achieve greater influence than on their own. Rowlands (1997) adds a third dimension of empowerment: the relational empowerment in which the emphasis lies on the ability to influence the relationships and decisions that are taken through these. Some communities are more oriented toward empowerment than others; they have internal or external features that impede or foster empowerment (Maton 2008). One possibility to achieve collective empowerment is the mobilization of individuals through a common purpose. For example, female community leaders seek to persuade ordinary citizens to engage in urban participatory democracy to create a change. These leaders make an attempt to increase both inclusiveness and support for a common cause (Hajdarowicz 2018). Four specific dimensions can be used to assess whether PB projects can reach collective empowerment: the primacy of participatory forums in the decision-making process, the scope of the budget (how much is directly allocated by citizens), the degree of power allocated to PB (whether authorities retain discretion in implementing the projects), and the self-regulation of PB (if rules are established by participants) (Baiocchi and Ganuza 2014, p. 49). Pin (2017) adds a fifth criterion which consists of the permanency of the PB forums, which refers to the institutionalized feature of PB and to its stability in community, based on support received from citizens. This contrasts with authorities’ complete degree of agency over the establishment and existence of the PB process. Much of the literature on empowerment focuses on the supply side, embedding it in the broader political and economic context in which PB takes place. The demand-side perspective remains somehow peripheral and limited attention is paid to the perception of empowerment. The latter motivates public participation (Kang 2014): when people feel that their efforts can make a difference, they are more likely to engage in various processes. With reference to PB, Zolotov et al. (2018) assess the effect of four dimensions of the psychological empowerment (competence, impact, meaning, and self-determination) on the intention to use online PB. While competence and meaning have a positive effect on the willingness to continue using PB, habit is a strong predictor. People who used PB are more inclined to continue using it, which is consistent with earlier findings about how people who actively used deliberative democracy are much more likely to support (Gherghina and Geissel 2020). There is empirical evidence that documents the existence of a relationship between the use of PB and perceived legitimacy of local governments. The inhabitants of the council districts that use PB have greater feelings of access to the local government and believe they understand better the spending of public money. All these leading to a positive perception of the government officials and to higher subjective / perceived legitimacy (Swaner 2017). Participants to PB express a variety of feelings and perceptions about the process. A comparative analysis of e-PBs in Brazil indicates the existence of both positive feelings such as the belief in political effectiveness and negative feelings that include frustration or indignation. The perceptions oscillate between approval for what is done through PB to perceptions of poor representation or low political effectiveness (Barros and Sampaio 2016). Over time, the participants appeared to lose trust in the process and in local level representatives. In other instances, the engagement in deliberation during the PB leads to a reflection of people’s opinions in the budget or policy decision-making. The offline is more effective compared to the online participation because of higher levels of representativeness and deliberation (Lim and Oh 2016). This literature review indicates the importance of institutional design in shaping the perceptions of participants about the PB. Although much research focuses on the perceptions about the authorities that implement the PB decisions and about who gets involved, they can be used as a point of departure for this study. The experience of participants with the PB and the existence of deliberation in the PB are two main drivers for perceptions. This study isolates the variable of experience in using the PB because it focuses exclusively on people who have participatory experience. In doing so, it tries to reveal what other issues—related to institutional design or beyond—can shape participants’ perceptions about the PB’s collective empowerment potential. Due to the limited research on the topic, we use an inductive approach in which the reasons are derived from interviews. Data and method To investigate this perception, we use a case study approach (George and Bennett 2005) that focuses on Cluj-Napoca as the first city in Romania to use PB. The city is a critical case for three reasons. First, there is continuity in terms of local administration: the mayor, who is a supporter of PB, is in office uninterrupted since 2012. Before that, he was elected as a mayor of the city in 2004 and re-elected in 2008, but his second term in office was ended after roughly half a year because he served as country prime minister between 2008 and 2012. Second, the experience of PB in Cluj-Napoca achieved an emblematic status in Romanian politics. The party to which the mayor belongs used it during election campaign as a benchmark of innovation and good democratic governance. Third, the PB meets partially the minimum characteristics of the process (Sintomer et al. 2008, p. 168). It includes the component of proposal design and submission, information about ideas, final vote, and implementation of projects at the city—as opposed to neighborhood—level, and it is organized regularly. However, it misses the component of debate and deliberation, and the accountability / monitoring of implementation (Stortone and de Cindio 2016) is only passive. As such, this PB includes some design particularities that could influence the perceptions and the interviews will show whether the participants identify these elements as drivers for their opinion on the collective empowerment potential. We conducted 25 semi-structured interviews in November 2020 with citizens who took part in PB in different roles. The interviewees were (1) ordinary participants by casting a vote on the projects that raised their interest, (2) project applicants who submitted projects in the PB competition or (3) consultants, i.e., they provide support to those who did not have access to the Internet or did not know how to use the online platform. We used a snowballing technique to select the participants and the selection sought to maximize variation on several key variables such as gender, area of residence, and profession (Appendix 1). We interviewed eight women and 17 men, with age between 20 and 47 years, living in different neighborhoods of the city. Their professions ranged from private corporation employees or public clerks to school and university teachers. We also tried to get variation in terms of education but that was not possible since all the respondents we contacted had university degrees. This is not surprising since the city is the home to the largest university in the country and less than 0.5% of the residents engage with the PB. Quite likely, there is self-selection bias in terms of education and the complexity of their answers regarding the empowerment component is related to the fact that many are middle-class and educated participants. We stopped at 25 interviews because that is where we reached the saturation point. The interview guide includes eight main questions. The first two questions were related to the overall satisfaction toward the development of Cluj-Napoca in the last 10 years and the benefits of participating budgeting. The next two questions were more specific and asked about how and why the interviewees involved in the PB and questions five and six asked the interviewees to describe the PB process and to explain if they believe that participating budgeting is a real tool for citizen empowerment (they were asked to elaborate their answers also). The last two questions were about the PB potential to reduce the involvement of the political factor in the decision-making and about directions for future PB improvement. We used ad-hoc follow-up questions when more details were required from respondents. The interviews were made by telephone, with an average length of 15 min, and recorded with interviewees’ consent. We use inductive thematic analysis that relies on themes identified in the answers provided by respondents (Nowell et al. 2017). In the absence of pre-established themes / determinants of perceptions that could be identified in the literature, the inductive thematic approach provides the appropriate avenue to interpret meaning from the answers. We read all the answers and sought to assign sentences to sub-themes and then larger themes (Table 1). The latter are preferable since they make the interpretation of results straightforward and substantial. The process of collecting, grouping, and analyzing the data was divided into five phases: 1) Identify the respondents and schedule the interviews; 2) Conducting the interviews; 3) Interview transcription; 4) Independent reading (of the authors) of the interviews and outlined the initial themes; and 5) Comparison of the themes, stylization, final version, and interpretation. Based on the interview, we could formulate two broad themes: the inappropriate format of the PB and the limited funds allocated to it, each of these with three sub-themes (Table 1).Table 1 An overview of the themes derived from the interviews Themes of PB Sub-Themes Inappropriate format Lack of communication between participants and organizers Online format that limits the involvement of elders The absence of debates aiming and improving projects Limited funds Little control over the public funds Limited influence of the political factor in decision-making Small funds limit the advancement of large-scale projects PB in Cluj-Napoca Cluj-Napoca is the fourth most populous in Romania (slightly over 300,000 people) and uses PB at city level on a continuous basis since 2015. PB was adopted for the first time in 2013 as a pilot project for the largest neighborhood in Cluj-Napoca that includes roughly one third of the city’s population. The main idea of the pilot project was to allow the inhabitants to discuss ideas and propose projects related to the improvement of the quality of their lives in the neighborhood. The City Hall organized discussion sessions, which were also joined by the mayor, in specific locations where people exchanged ideas between them, public clerks and officials. The PB expanded to the level of the entire city and continued in 2015 and 2016 as a project dedicated to informal groups of individuals aged between 14 and 35. The participants proposed projects aimed at changing the Cluj community overall. The winning projects were financed from the budget of the City Hall (Cluj-Napoca 2017) and implemented the subsequent years. In 2017, the City Hall created an online platform dedicated exclusively to the project. Cluj-Napoca became the first city in Romania to ground PB as a stand-alone project opened for its inhabitants of at least 18 years.1 It provides the participants with the opportunity to shape and observe the implementation of projects concerning their communities, to transfer their ideas into projects, to signal the authorities the main concerns and shortcomings of their neighborhoods, to find solutions for their problems, and to take part in the process of setting priorities in spending local money. Until 2022, the Cluj-Napoca PB funded 15 projects annually with a maximum value of 150,000 € for each project. In 2019, the city had a budget of 326 million € out of which the PB received 2.25 million €, which is roughly 0.70% of the total city budget. In 2020, the PB was allocated the same amount of 2.25 million €, roughly 0.65% of the total municipal budget that which was 344 million € (Gherghina and Tap 2021). The 2020 PB was not organized due to the COVID-19 pandemics, but the process was continued in 2021. In 2022, the number of funded projects was reduced to five, but the maximum value increased to 400,000 € for each project. All the accepted projects are implemented the next calendar year and the stage of project development can be accessed on the PB’s platform.2 All those who live, work, or study in the city can submit projects or cast a vote on existing proposals. They must access the PB platform, create an account, and proceed with the application or vote. All submitted projects are analyzed by the technical departments of the City Hall, and if the project is eligible, (i.e., it can be implemented by the authorities, does not exceed the maximum budget), it will be placed in the competition. Since 2017 the City Hall established an office where its employees organize the PB. This office collects the project proposals, coordinates the technical check, communicates with the applicants, organizes the voting, communicates the results, and maintains the online platform. Between 2017 and 2021, the time frame covered by our analysis, the project competition included six domains of submission: (1) alleys, sidewalks, and pedestrian areas; (2) mobility, accessibility, and traffic safety; (3) green spaces and playgrounds; (4) arrangement of public spaces (urban furniture, public lightning); (5) educational and cultural infrastructure; and (6) digital city. The projects that passed the technical check became eligible for the voting process which was divided in two stages: the vote according to the domains and the final one. In the first stage, 30 projects were chosen from all domains and every citizen has the right to vote six proposals (i.e., one belonging to each domain). The top three projects according to the voting share went directly to the second stage and the rest were selected based on the number of votes they gathered. In the second stage, every citizen could vote for one project (irrespective of its domain) and a total of 15 projects are selected. The project with most votes in each domain will be automatically selected and the rest of the winning projects will be decided according to their voting share. The projects that gather the highest number of votes will be implemented by the City Hall in the following calendar year; usually the PB is over in November. Those who wish to vote or submit projects but do not have access to the Internet or do not know how to use the platform can benefit from the help of City Hall’s employees in specific locations in Cluj-Napoca during the entire process.3 The number of projects submitted to PB decreased from 383 in 2017 to 103 in 2021, while the number of eligible projects decreased from 126 in 2017 to 20 in 2021.4 Out of the 20 eligible projects in 2021, 15 got funded and would be implemented, three of them with as few as 29, 45, or 50 votes. The decreasing trend in the number of submitted proposals and citizens’ participation could have several possible explanations such as demotivating competitiveness, understanding that projects must meet certain standards to be accepted for voting, loss of interest in the PB process, the migration of those who engaged before, or different priorities during the COVID-19 pandemic. How citizens perceive empowerment Overall, the 25 interviews emphasized that PB in Cluj-Napoca is perceived as a form of citizen empowerment in decision-making for two reasons (Fig. 1). First, the openness of City Hall for PB creates opportunities for citizens to engage directly in the decision-making. Second, there is no political interference in the PB process. At the same time, the participants signaled that the collective empowerment of PB in Cluj-Napoca is limited by three elements. First, the absence of discussions, debates, and communication with the organizers and project applicants complicates the process of voting, writing, or improving projects before the competition. Second, the online format of PB discourages specific age groups (i.e., elders) from participating. Since senior citizens are usually active in political participation, a format that is friendly to this age category is likely to increase the feeling of empowerment and inclusiveness. Third, the funds allocated to PB are too limited. The sum is tiny compared to the overall budget of the city and thus PB cannot make a real influence on the community decision-making; the participants cannot propose expensive projects due to financial constraints. The following sub-sections cover each of these issues.Fig. 1 The perception of citizens’ empowerment according to PB participants General opportunity and no political interference Many respondents perceive the Cluj-Napoca PB as a possibility to empower citizens and provide them a voice in the decision-making process at local level. A large share of interviewees (20 out of 25) stated that PB brings plenty of benefits to the overall development of the city, neighborhoods and stressed that the voices of mere citizens are heard by the authorities via PB. Some of them emphasized that PB “allows the accumulation or gathering the ideas from the citizens which might not otherwise have reached the table of the local administration (…) allows the administration to see where citizens say that there is a need to invest public money” (I9) or associated the project with a “channel for collecting people’s demands, so prioritizing investments according to public opinion because the citizen sees best in the vicinity of home or workplace (…) which are the problems and if they are sufficiently addressed” (I13). PB was described as “one of the best collaborative tools that Cluj-Napoca provides and it is (…) an example of good practices for several cities in the country. It is a unique and interesting way (…) to consult citizens regarding the projects in their neighborhood that directly affect them” (I14). It could stimulate “(…) the involvement of citizens in direct decision-making (…) crates an involved civic community that find local solutions to their problems, much closer than the general opinion of the Local Council and the City Hall” (I6). PB was described as a “democratic exercise” (I12), “a form of democratic consolidation” (I11), “a proof of democracy” (I4) or a “right of direct democracy” (I9). All respondents outlined that PB is directly related to the right of citizens to decide how public money should be invested. Some responses highlighted that “we are the main contributors to the local budget (…) we should have a very big word to say not only once every four years when we vote for the Mayor or the Local Council but in the decisions taken by the City Hall in these four years. First of all, decisions that directly affect us” (I6) or “(…) man is the most important, democracy and the whole public administration revolve around him and somehow you give him power through this issue of participatory budgeting” (I16). PB was not related only to a right but with a “privilege (…) there are not so many cities that do this project so it seems to me that it is a very useful tool (…) this public consultation and the whole process because you are practically included from start to finish” (I14) as well as “an obligation to get involved in the decision-making process in the localities in which we live (…) this would be a key factor for me in consolidating Romania’s democratic tradition after the post-December Revolution” (I11). All these are potential outcomes when the citizens realize how much power they have and get more involved in these processes (I12). One interviewee stated that “no one gets involved and there is no advantage in participatory budgeting (…) very few people get involved and very few people think that they have power (…) without people you cannot do anything” (I2). However, the interviewee’s dissatisfaction was related to the little number of individuals who get involved in PB and not with the program itself. The interviewees described PB as a tool to increase citizen empowerment in decision-making and as an avenue for greater social cohesion and civic engagement. For instance, one respondent emphasized that he made lobby for a project that was advanced via PB. The interviewee said that he had no linkage with its initiator, but the project aimed at issues concerning his neighborhood. The interviewee found very useful the ideas of the project and promoted it around the neighborhood and on social media (I16). Similarly, other interviewees emphasized that promoted PB projects on social media with the aim of mobilizing new voters (I1) and others stressed that were contacted by friends or other individuals to vote for specific projects (I17). Several interviewees said that PB stimulates collective decision-making because “as long as hundreds of votes are collected (…) maybe there were over 1000 (…) means that it is a collective decision” (I16) or said that “I see a collective decision because any citizen in Cluj-Napoca can vote for projects that seem more suitable for the city” (I19) and underlined the idea that “through the votes on the PB website it is a collective decision because everyone has a say in choosing the projects” (I21). PB goes beyond the individual empowerment and enforce the collective decision-making because the votes mirror the desire of the collectivity and increase social cohesion and civic engagement of individuals. Several answers explicitly underlined that PB reduces the political influence in decision-making, and the process itself is not influenced by political interference. Apart from three interviewees who said that politicians could use the implemented projects via participatory budgeting as electoral gains (I1, I6, I11),5 the other interviewees explicitly stated that the political actors do not get involved in the process. The mayor and his team created the institution of PB but do not intervene politically during its functioning. There is illustrative evidence in this sense. There is no biased promotion of some projects on the PB platform and citizens have access to the same type of information about all submitted projects. All the 15 projects that get voted through the PB are implemented by the City Hall and the information is publicly available online. No interviewee mentioned that some communities’ projects were favored over other communities. On the contrary, many mentioned that the projects in their neighborhood made it on the list of voted projects and that they knew that even the applicants were surprised by this. All the interviewees stated that they did not feel any kind of political influence and pressure when participating in participatory budgeting and referred to the process as free from political interference. Some of the respondents explicitly appreciate this as a positive feature since the independence of PB increases the confidence that changes can be made, and thus increases the feeling of empowerment. Poor communication and no possibility for deliberation Despite the positive perceptions toward PB, interviewees stressed certain shortcomings of the project and suggested avenues for improvement. For instance, some outlined that “(…) in the future the second point that should be improved would be a diversification (…) of the types of projects (…). At the moment, the projects are divided into categories, but they could go further for each one, somehow (…) more personalized for each part of the city” (I7) or “(…) given that the process has been going on for several years now, some new domains could be introduced, for example, the social domain that is now missing or sports field” (I14). Similarly, some interviewees noted that PB “should be more marketed, made more public (…) posted in the city (…) the City Hall can make posters to put them in the city and they could also promote it on social networks” (I8) or that the project needs “more visibility (…) I would (…) recommend putting up banners around the city as we see in election campaigns (…) banners with ‘Participate in participatory budgeting’ (…) to make the process much more attractive or (…) and use online paid ads” (I4). Also, one interviewee noted that it takes quite a long time until the projects are adopted and implemented, and faster development of the projects could be an advantage for streamlining PB (I21). Inappropriate format Despite the transparency, accessibility, inclusiveness, and openness of PB interviewees emphasized that lack of public or online debates represents one of the major problems of the program, it was stressed thatThere are no meetings, deliberative forums where the proposed projects can be debated (…) many projects are very succinct, it is very difficult to figure out what the project proposes and I think you need to give citizens the opportunity to debate those programs via online forums or face to face meetings (I12). Similarly, other interviewees emphasized that “some debates should be held, either in real or virtual way (…) meetings on Zoom, Skype or other possibilities because (…) apart from the project that appears on the site, participants do not know a lot of information about it, only if they know the initiator and ask for additional information” (I19) or “I would like to have a mix between the online and face-to-face meetings (…) for discussing topics of participatory budgeting (…) I say this from the perspective of reaching as many people as possible” (I16). Also, some interviewees highlighted thatApart from submitting projects and consulting after a project is submitted, I do not know to exist public debates (…) where these projects are raised from time to time (…) I do not know to be open discussions or chats or messages directly with the organizers (…) probably if you contact the Facebook page or the platform by e-mail, those interested will probably receive answers, but so far it has not been the case to send an e-mail to request any clarification or something like that (I14). Therefore, interviewees stated that PB lacks centralized forums of discussions even though interactions between participants could improve the quality of the projects. In this sense, it was emphasized that “any dialogue can bring improvements and bring projects to a much better form, which will better meet the needs and form synergies with other needs (…) would also stimulate people’s desire to propose projects” (I13) or “I think there should be a much broader debate, especially in the final project categories (…) or a top three in each project category and an online debate or poll for each project (…) a SWOT analysis with what the project brings positive to the whole community” (I6). Also, interviewees signaled the need for organizing debates related not only to the proposed projects and how these can be improved but for helping people to understand what PB is, how to write a project or what kind of projects could be implemented easily, are necessary for the overall development of the city and could be improved before the initiators advance them. In this sense, interviewees noted that PB needs better communication between organizers and citizens (I1, I4, I7, I8, I9, I11, I13, I21). Another line of argumentation emphasizes that the predominant online format limits the involvement of elders in the program. Some interviewees highlighted that older generations could face difficulties when it comes to accessing the online platform, use it or even writing a project. Also, they are not aware of the possibilities in which projects targeting their interests are available on the platform and how they can vote for them (I2, I11, I16, I17, I20). Even though there were City Hall’s employees who “offered support by creating an account, by helping the demanders to find the project they wanted (…) there were older people who did not have the opportunity to vote, to make an e-mail address, an account and in this way I helped them” (I21), the lack of skills in using the online platform explains why PB is somehow problematic for this social group. These shortcomings transform PB in a limited form of empowerment of citizens in decision-making because they have no opportunities to debate the projects, improve their quality, and find real solutions for the pressing problems of their communities. They can only vote for the projects that are uploaded on the platform without the possibility of discussing how these can be improved and how can be transformed in valuable investments for the city. Also, elders are underrepresented in PB and they cannot benefit properly from this form of empowerment because the program itself is based mainly on online procedures and not all the elders are literate in the Internet using or eager to go to specific points in the city where they could benefit from the help of PB organizers. Limited funding PB reduces political influence in decision-making “because not politicians decide here but citizens (…) the collective that vote for certain projects decides, nobody intervenes (…) let us say Mayor, Deputies, Ministers” (I18) and “I say that practically the strongest voice is the voice of the citizens and it is practically the citizens who have the strongest word to say in this project, so I say that it is a rather big reduction of political involvement” (I19). Other interviewees stated that “I agree with this type of process because it gives a sign to the political parties that they are not really in control of the budget and they cannot do exactly what they want (…) they must be kept account of what citizens say (…)” even that the voted project or idea is contrary to their interests (I3) or emphasized that political influence is diminished “(…) quite a lot (…) considering that people (…) talk freely and present their ideas, come up with proposals and promote them (…) here I think it is the gain that there is no political involvement, people can simply decide their priorities” (I16). Many interviewees stressed that their empowerment is limited by the funds that are allocated for the process. They referred to the limited funding provided to PB, which is reflected also by objective statistics. One interviewee argued that the City Hall should allocate 10% of the city’s budget for PB. When the funds of the program increase, citizens will gain much more power in decision-making because the authorities “can no longer influence the allocation of funds on clientelism or for electoral or sociological purposes where they think they have a more favorable electorate” (I6). Other interviewees used examples from Western societies to underline that “A project like participatory budgeting should allocate much more resources and much more funds (…) the cities in Latin America, in Brazil (…) allocate 20% or even 30% of local government funds in support of projects started or proposed by citizens” (I11) or “I think it should be a much more important part of the local budget (…) authorities should reflect on and borrow from the practices of other cities where are much more substantial budget allocations (…) I mean the countries of Western Europe” (I12). Similarly, some interviewees outlined that due to the low funds allocated for PB, their empowerment and influence in decision-making is limited because they can decide only how a little amount of public money are invested. Consequently, they stressed that political influence in decision-making is not significantly reduced because projects of a maximum of 150,000 € cannot produce major changes inside the society and affect the overall budget of the city. In this sense, the interviewees stated that the budget allocated for PB should be raised to improve their empowerment and significantly reduce the political influence in decision-making (I3, I4, I8, I13, I14, I17). Moreover, interviewees highlighted that the allocated funds must be raised according to the interest of the citizens. They said that many good projects are not implemented due to the fact that their expenses surpass the allocated funds, and this represents a limit for citizens who should have good ideas, but they cannot propose them due to the financial limits (I4, I14, I17). Discussion and conclusions This article analyzed the factors that shape the perception of participants to PB in Cluj-Napoca about its collective empowerment potential. The results indicate the existence of a complex picture in which participants acknowledge the potential for collective empowerment but also identify several obstacles. The political interference is not among these obstacles. On the contrary, the limited involvement of political actors is a characteristic that enhances the collective empowerment of the PB. The mayor and his political party are behind the project, they support it and organize it on a regular basis, but this is where the involvement of politics in PB appears to stop according to our interviewees. The limited presence of political actors in the PB process is broadly appreciated by respondents and perceived as an important avenue that allows an effective engagement of citizens in the decision-making process at community level. The existing obstacles can be addressed through several reforms of the PB process, which are relevant for policy makers. One important change would be the introduction of a deliberative component, which provides the possibility to discuss, debate, and decide about the most appropriate projects. Many participants complained about the absence of such possibilities and the absence of deliberation brings the entire process much closer to e-voting on projects rather than to an actual PB. A deliberative component is crucial for PB processes around the world and earlier analyses—coming from academics, observers or even organizers—indicate the benefits of open discussions for the quality of approved projects. A second change could include physical meetings with citizens that could complement the online presentation of projects. Finally, an important reform refers to a larger budget allocated to projects, to increase the impact on community development along citizens’ priorities. All these changes are likely to increase the perception of empowerment, the visibility of the process, and the number of participants. Overall, this study illustrates a complex web of perceptions and shows how several elements can lead to different perceptions. People pick on the PB design issues and consider these to weaken the empowerment potential, which can be a valuable lesson for further implementation of the PB in this city and in other similar settings. The limited decision-making power, reflected mainly in design flaws, is in line with earlier results from West and Southern European countries where it led to a general disappointment on the side of citizens with the PB process (Talpin 2011). In addition to decision-making, our results indicate that the objectives and resources are crucial elements that could enhance citizens’ approach toward PB as it happens in other European cases (Lehtonen 2022). In essence, we illustrate that the ideas linked to citizens’ participation in the PB are mainly institutional in nature, with little emphasis on the individual or policy dimensions (Röcke 2014). All these indicate that the Cluj-Napoca PB has characteristics that are comparable with PB processes conducted elsewhere, and thus the implications of our study reach beyond the single case analyzed here. At theoretical level, the analysis illustrates how citizens use a combination of personal and community-based features to assess the PB’s empowerment potential. Such features can form the basis of a framework for analysis to be used in further research. At empirical level, our analysis identifies several themes that allow a better understanding of how the participants see PB. Its functionality and obstacles to a larger impact may be defining factors to explain further processes such as the desire to become an applicant in the process or the willingness to continue spending time resources with the project. There is a convergence of opinions between three different categories of participants—ordinary citizens, applicants, and consultants—which increase the robustness of observations. At policy level, and essential for the goal of this article, the findings illustrate that the limited political interference in the PB process has a positive effect on the citizens’ perception regarding their empowerment. The Romanian political parties can use this piece of information to decide on future implementation of PB in the same city or in other cities around the country. These findings can be generalized to other settings where PB takes place under similar circumstances. Our analysis yields several findings that open the door to future research. One immediate avenue could be a closer investigation of the causes leading to the formation of perceptions about empowerment. Our paper focuses on attitudes and experiences, but we do not touch upon the ways in which these are shaped. Further studies could explore this causal mechanism to better understand citizens’ expectations when engaging with PB. It could investigate, for example, to what extent participants' satisfaction after participating altered their initial expectations. Alternatively, a comparison with other political contexts would be useful both in the East European region and more broadly in Europe. This would allow understanding whether the context could influence the perceptions of empowerment. As such, the analysis can be replicated to compare results and reach broader implications for the field of study. Another avenue for research could include a more quantitative approach in which more PB participants are recruited, with different socio-demographic characteristics, to identify whether and how the perceptions of empowerment are shaped by who the participants are. Appendix 1: List of interviews Interviewee Age Gender Occupation PB role Interview length (min.) I1 30 Female IT consultant Applicant 18 I2 47 Male Restaurant manager Participant 21 I3 20 Male Student Participant 19 I4 20 Male Student Participant 20 I5 20 Male Student Participant 15 I6 34 Male Project manager Applicant 17 I7 24 Male Student Participant 15 I8 37 Male Programmer Participant 12 I9 33 Male IT worker Participant 10 I10 40 Female School teacher Participant 12 I11 23 Male Student Participant 19 I12 40 Female University lecturer Participant 23 I13 42 Female Policy councillor Participant 18 I14 24 Male Marketing specialist Participant 12 I15 38 Male Inspector Participant 15 I16 34 Male Clerk (local authority) Participant 15 I17 35 Male IT technical analyst Participant 15 I18 32 Female Economist Participant 9 I19 23 Male Student Applicant 14 I20 27 Female Contractor employee Participant 10 I21 47 Female Clerk (local authority) Consultant 10 I22 31 Male Teaching assistant Participant 15 I23 33 Female Quality assurance inspector Consultant 19 I24 33 Male University lecturer Participant 10 I25 28 Male Clerk (central authority) Participant 10 Declarations Conflict of interests All authors declare that they have no conflicts of interest. Funding The work of Sorina Soare was supported by Babes-Bolyai University Cluj through an International Advanced Fellowship (project number CNFIS-FDI-2022-0179). 1 In practice, since 2017 the process is an e-PB, but for the sake of consistency, we use the PB label throughout this article. 2 The platform available at https://bugetareparticipativa.ro/, last accessed 18 August 2022. 3 Data available at https://primariaclujnapoca.ro/informatii-publice/comunicate/bugetare-participativa-cluj-napoca-2019/, last accessed 11 November 2020. 4 Data for the 2017–2019 PB in Cluj are available in Gherghina and Tap (2021). 5 The three interviewees said that certain projects (e.g., the one concerning public transport for students) that were implemented via participatory budgeting were claimed as successes of the administration and the officials did not say that the projects were framed by citizens and implemented through participatory budgeting. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Alves ML Allegretti G (In)stability, a key element to understand participatory budgeting: Discussing Portuguese cases Journal of Deliberative Democracy 2012 8 2 1 19 10.16997/jdd.147 Åström J Grönlund Å Coleman S Shane PM Online consultations in local government: What works, when, and why? Connecting democracy: Online consultation and the flow of political communication 2012 Cambridge MIT Press 75 96 Baiocchi G Ganuza E Participatory budgeting as if emancipation mattered Politics & Society 2014 42 1 29 50 10.1177/0032329213512978 Barros SAR Sampaio RC Do citizens trust electronic participatory budgeting? Public expression in online forums as an evaluation method in Belo Horizonte; do citizens trust electronic participatory budgeting? Public expression in online forums as an evaluation method in Belo Ho Policy and Internet 2016 8 3 292 312 10.1002/poi3.125 Brun-Martos MI Lapsley I Democracy, governmentality and transparency: Participatory budgeting in action Public Management Review 2017 19 7 1006 1021 10.1080/14719037.2016.1243814 Cabannes Y Participatory budgeting: A significant contribution to participatory democracy Environment and Urbanization 2004 16 1 27 46 10.1177/095624780401600104 Cabannes Y Lipietz B Revisiting the democratic promise of participatory budgeting in light of competing political, good governance and technocratic logics Environment and Urbanization 2018 30 1 67 84 10.1177/0956247817746279 Cluj-Napoca, P. 2017. Bugetare participativă la Cluj-Napoca (Participatory budgeting in Cluj-Napoca). Dias, N., S. Enríquez, and S. Júlio, eds. 2019. Participatory budgeting world atlas. Cascais: Municipality of Cascais. Fung A Wright EO Fung A Wright EO Countervailing power in empowered participatory governance Deepening democracy: Institutional innovations in empowered participatory governance 2003 London and New York Verso 259 290 Fung A Wright EO Deepening democracy: Institutional innovations in empowered participatory governance 2003 London and New York Verso Geissel B Newton K Evaluating democratic innovations: Curing the democratic malaise? 2012 London Routledge George AL Bennett A Case Studies and Theory Development in the Social Sciences 2005 Cambridge, MA MIT Press Gherghina S Geissel B Support for direct and deliberative models of democracy in the UK: Understanding the difference Political Research Exchange 2020 2 1 1809474 10.1080/2474736X.2020.1809474 Gherghina S Tap P Ecology projects and participatory budgeting: Enhancing citizens support Sustainability 2021 13 19 10561 10.3390/su131910561 Hajdarowicz I ‘Does participation empower? The example of women involved in participatory budgeting in Medellin Journal of Urban Affairs 2018 44 22 37 10.1080/07352166.2018.1431048 Kang M Understanding public engagement: Conceptualizing and measuring its influence on supportive behavioral intentions Journal of Public Relations Research 2014 26 5 399 416 10.1080/1062726X.2014.956107 Lehtonen P Policy on the move: The enabling settings of participation in participatory budgeting Policy Studies 2022 43 5 1036 1054 10.1080/01442872.2021.1895981 Lim S Oh Y Online versus offline participation: Has the democratic potential of the internet been realized? Analysis of a participatory budgeting system in Korea Public Performance & Management Review 2016 39 3 676 700 10.1080/15309576.2016.1146553 Marian, C. 2018. The Social Democrat Party and the use of political marketing in the 2016 elections in Romania. Sfera Politicii 26 (3–4): 70–82. Maton KI Empowering community settings: Agents of individual development, community betterment, and positive social change American Journal of Community Psychology 2008 41 1–2 4 21 10.1007/s10464-007-9148-6 18175215 Novy A Leubolt B Participatory Budgeting in Porto Alegre: Social Innovation and the Dialectical Relationship of State and Civil Society Urban Studies 2005 42 11 2023 2036 10.1080/00420980500279828 Nowell LS Thematic analysis: Striving to meet the trustworthiness criteria International Journal of Qualitative Methods 2017 16 1 1 16 10.1177/1609406917733847 Nylen WR Testing the empowerment thesis: The participatory budget in Belo Horizonte and Betim, Brazil Comparative Politics 2002 34 2 127 145 10.2307/4146934 Peck J Theodore N Fast policy: Experimental statecraft at the thresholds of neoliberalism 2015 Minneapolis University of Minnesota Press Pin L Does participatory budgeting lead to local empowerment? The case of Chicago, IL Alternate Routes 2017 28 1 114 140 Röcke A Framing citizen participation: Participatory budgeting in France, Germany and the United Kingdom 2014 Basingstoke Palgrave Macmillan Rowlands J Questioning empowerment. Working with women in Honduras 1997 Oxford Oxfam GB Shah A Participatory budgeting 2007 Washington DC World Bank Sintomer Y Herzberg C Röcke A Participatory budgeting in Europe: Potentials and challenges International Journal of Urban and Regional Research 2008 32 1 164 178 10.1111/j.1468-2427.2008.00777.x Smith G Democratic innovation: Designing institutions for citizen participation 2009 Cambridge Cambridge University Press De Sousa Santos B Participatory budgeting in Porto Alegre: Toward a redistributive democracy Politics & SocietySociety 1998 26 4 461 510 10.1177/0032329298026004003 Steiner J The foundations of deliberative democracy empirical research and normative implications 2012 Cambridge Cambridge University Press Stortone, S., and F. de Cindio. 2016. Styles of online participation in the Monza participatory budgeting. In: 2016 Conference for E-democracy and open government (CeDEM), pp. 17–24. Swaner R Trust matters: Enhancing government legitimacy through participatory budgeting New Political Science 2017 39 1 95 108 10.1080/07393148.2017.1278856 Talpin J Schools of Democracy: How ordinary citizens (sometimes) become competent in participatory budgeting institutions 2011 Colchester ECPR Press Talpin J Geissel B Newton K When democratic innovations let the people decide. An evaluation of co-governance experiments Evaluating democratic innovations: Curing the democratic Malaise? 2012 London Routledge 184 206 Wampler B Participatory budgeting: Core principles and key impact Journal of Public Deliberation 2012 8 2 1 13 Wampler, B., S. McNulty, and M. Touchton. 2018. Participatory budgeting: Spreading across the globe, transparency and accountability initiative. https://www.transparency-initiative.org/wp-content/uploads/2018/03/spreading-pb-across-the-globe_jan-2018.pdf. Williams D Waisanen D Real money, real power? The challenges with participatory budgeting in New York City 2020 Basingstoke Palgrave Macmillan Zolotov, M. N., T. Oliveira, and S. Casteleyn. 2018. Continued intention to use online participatory budgeting: The effect of empowerment and habit. In: Proceedings of the 11th international conference on theory and practice of electronic governance, pp. 209–216.	0	PMC9716538	NO-CC CODE	2022-12-03 23:20:58	no	Acta Polit. 2022 Dec 2;:1-18	utf-8	null	null	null	oa_other
==== Front Psychol Stud (Mysore) Psychol Stud (Mysore) Psychological Studies 0033-2968 0974-9861 Springer India New Delhi 686 10.1007/s12646-022-00686-3 Research in Progress Mental Fitness: Psychological Warfare from Battlefield to Playground http://orcid.org/0000-0002-4780-5100 Idaya Rani C. [email protected] Subbu Lakshmi M. [email protected] Department of EFL, SRM IST, Kattankulathur, Tamilnadu India 2 12 2022 17 23 2 2022 26 3 2022 © The Author(s) under exclusive licence to National Academy of Psychology (NAOP) India 2022, Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. India and its people are commonly known for their unique culture and tradition. Cricket and mythology are much interwoven into the lives of people as both are almost inseparable part of their life and culture. Although the two fields are completely different from each other, there is a deep-rooted connection between them when it comes to their popularity in India. Most of the people of India have spent their childhood either by listening to the stories of mythology or watching cricket, because the two interesting activities that consciously impact the mind and gain the attention so easily. The psychological aspects of both game and stories literally leave a strong impact in human mind. Hence, this paper attempts to integrate the psychological aspects of the game and mythology by analyzing the existing mental health problems of Indian cricketers with reference to the mythological stories of Indian heroes. It further aims to provide the proposed model for mental fitness named SPORTS as a guide to mental training for contemporary cricketers to manage their emotions and control their mind for optimal performance. Keywords Mental fitness Warriors Indian mythology Mental training Self-awareness Cricket ==== Body pmcIntroduction Sport is a physical game involving players or athletes in competitive physical activities and exercises. Generally, a multitude of people get involved into sporting activities for some notable reasons. These people can be categorized into two sets. The first set of people utilizes the sporting activities for relaxation, fun, pleasure, enjoyment, and most importantly to reduce stress. The second set of people is named as elite athletes who engage themselves in sport with utmost dedication and commitment along with intensive hours of training and practice. The motive behind these two sets of people completely contradicts, but, something binds them together is the psychological aspects of the game. Mia Hamm, former American soccer player, quoted that “the most important attribute a player must have is mental toughness” (Hamm, n.d.). Though sport seems to have interconnected with physical attributes, the result that leads to success in any sport is determined by the players’ power of focus, concentration and commitment. Thus, sport is highly related to mental fitness than physical fitness (Afremow, 2014). This paper focuses on the importance of mental fitness, considering Indian mythological warriors and the contemporary cricketers. Indian mythology is a part and parcel of our Indian culture. It enriches the modern society by educating the morals and ethics through its symbolic characters, narrations and historical events. Mythological warriors are an integral part of human existence. They are considered as ideals, which help to construct the lives of human being in the right direction through their morals and ethics. The great Indian epics, such as Ramayana and Mahabharata, have gifted numerous legends to Indian mythology. India’s diversity of culture, people, and tribes are reflected in the wide variety of sporting disciplines in the country. There are various games in India, ranging from tribal games to mainstream sports such as cricket, hockey, tennis, badminton, kabaddi, archery, basketball, volleyball, cycling, shooting, squash, weightlifting, gymnastics, and football. Currently, cricket is considered to be the most popular game in India above all other sports. Hence, it gains more number of audience compared to any other sports. Cricket players have a huge fan base in India and, obviously, the expectations of the audience are high on Indian cricketers. Taking up cricket as a profession and fulfilling the expectation of millions of Indian fans are not an easy task. The cricketers need to focus more on both their physical and mental fitness. This paper analyzes the mental health problems faced by cricketers on and off the field. Further, it attempts to provide the proposed model for mental fitness with reference to Indian mythological heroes through their life lessons. Mental Fitness Mental fitness can be defined as being aware of one’s own thoughts, emotions, and feelings that influence a human behavior. It is a process of cultivating the ability to handle difficult situations, building resilience, maintaining a state of well-being, and developing mental strength to stay calm and composed without being affected by emotional turbulence (Mack & Casstevens, 2002). The raising concern over mental health-related problems, in the field of sports, has sparked many widespread debates and has drawn attention from the sports fraternity all around the globe (Purcell et al., 2019). Being fit and maintaining physical fitness is a basic requirement for any player in sports (Wood, 2010). The players have to clear the Yo-Yo test which is a benchmark for physical fitness Indian cricketers making into international cricket and is also considered to be the main reason for massive improvement of physical fitness level among players (SportsAdda, 2021). Although both psychological and physical components equally determine the players’ optimal performance in sports, there is no test being taken to check whether the player is mentally fit to play or not. This raises an issue that “Is mental health the most neglected areas in sports?” After the pandemic, the status of mental health is a great concern for people worldwide. According to the World Health Organization (WHO), the COVID-19 pandemic has disrupted or, in some cases, halted critical mental health services in 93% of countries, while the demand for improving mental health is increasing worldwide (Kovacevic, 2021). For any sports player, factors such as quarantine, less practice sessions, and bio-bubble create more stress that greatly impact their performance on the field (Gentile et al., 2021). After India’s poor T20 World cup performance in 2021, Indian head coach and few players have expressed and assured that the mental health problem is existing in cricket. “The team has been in a bubble for the last 6 months. I am mentally drained but these players are mentally and physically drained. We would have ideally liked a bigger gap between the IPL and the T20 World Cup,” Ravi Shastri, former Indian head coach told the T20 World Cup broadcasters Star Sports (CricketNext, 2021). Physical fitness is the most important factor for any athlete who gets into a game but the mental fitness determines their success and sustainability in sports. Attempting to address the current mental health problem and to provide solution, this paper aims to propose a model for mental fitness guide with an acronym SPORTS that consists of six components, such as Self-awareness, Preparation, Optimism, Resilience, Team spirit, and Self-motivation. It can be helpful for their overall development to become a better performer and also a better individual as well. Self-Awareness Self-awareness is the very first component of mental fitness. Cricket is a game being played in different formats, such as test, one day match, and T20. Strategies, game plan, and skills are playing a prominent role in cricket, but, winning the game often depends on mental strength of players (Barker & Slater, 2015). Sometimes, one bad over from a bowler, few misfielding, missed catches, and certain decisions can change the course of the game. More focus and concentration are needed throughout the game for every player who gets into the field in order to lead the team toward victory. Hence, mindfulness is the key factor for players to win the match. It is also one of the main aspects of self-awareness. “Knowing yourself is the beginning of all wisdom” quotes Aristotle, Greek philosopher. Self-awareness is a process of understanding oneself, their strength, weakness, abilities, emotions, and potential (Morin, 2011). It also gives them the space to interrogate their own failures, flaws, disabilities, and defects. So, players can utilize their potential to overcome their weaknesses which leads to better performance. When players encounter a do-or-die situation in cricket, their quickness and agility determine that whether a player is good enough in handling the difficult situation with ease or struggling to cope up with that situation. Therefore, self-awareness acts as a core component to build one’s personality and helps them to become stronger individual. Further, the player also gains the confidence to face any problems and the consequences in their professional life (Colman, 2008). To understand the power of self-awareness, the Indian mythological hero, Arjuna’s life can be taken for this study. Arjuna is known for his courage, valor, and determination. The magical hero has extended boundaries of Yudhistira’s empire through his mighty skills by defeating great kings. Arjuna is the only student who fulfilled the expectation of his guru Drona in Gurukul by answering “I see the eye of the bird,” (Bhagavatam katha, n.d.). For this reason, he is remembered for the focused attention and the power of concentration. Even though Arjuna stood as the Warrior of ages, he finds himself in a dilemma in the battlefield of Kurukshetra to fight against his own kith and kin. He forgot his duty and responsibility to fight for good against evil. His mental instability and confusions pushed him to decide to go back to jungle instead of participating in the war. Krishna, as a mentor and charioteer, tries to balance the mind of Arjuna from negative thoughts and emotions about the consequences of future and asked him to focus on the present to do his duty. This is a good example for mindfulness. He has clarified his confusions, removed all his doubts, and prepared him to participate for war. Having listened to Krishna, Arjuna decides to participate in the war with more focus and confidence (Modh, 2015). The same way, in cricket, a mentor or a sports psychologist needs to be hired full time like other coaches in order to take care of the psychological, emotional, and social well-being of players. A player can possess skill set, physical fitness, have better strategies, and game plan, but the presence of mind, mental strength leads to success. This would prevent a player from distractions, sharpen their focus, and boost their confidence. It helps them to be more aware of their thoughts and the process of that particular game. Undoubtedly, the most successful player in cricket is M.S. Dhoni. He often expresses his mantra for success in his interviews; he insisted that the process is more important rather than the result (Indian Express, 2019). Self-awareness in cricket is more of engaging themselves in the present and involving the process of the game. No player should focus on winning or losing the game, instead they should be very attentive on the process of giving their best. Thinking about the result will make the player emotive. The fear and nervousness that is connected with the result often prevent a player from giving his best. Thus, self-awareness is a technique of giving a chance to connect with their self and gifts the ability to have a control over his own mind (Frank & Ronald, 1980). Preparation Preparation, being a second component of mental fitness, is an essential process in any sports. It engages the players in different activities in order to prepare them for national and international competitions. It includes gym workout plans, practice game, training sessions, enhancing skills, learning strategies, and aerobic exercises. The most important aspect for any form of sporting activities requires mental strength, but the amount of time is being spent for mental preparation is comparatively lesser than physical fitness. Cricketers likely utilize some of the well-known mental conditioning techniques such as meditation and yoga for mental preparation (Rao, 2020). Every young player has to realize the importance of mental preparation (Biddle, 1985). To enhance cricketers’ mental strength, the role of mentoring is crucial and a proper guidance is much needed. In a game-like cricket, the overall team spirit is essential than individual strength and abilities. To adapt players’ mindset to work as a team to have a better cooperation with captain and team members is quite important. So the cricketers as an individual should be trained to strengthen their mental abilities to have a positive atmosphere among themselves. The main motive of preparation is to transform the players’ mindset as a stronger individual (Crook, 1980). Hanuman, the son of wind god Vayu in the Hindu epic Ramayana, is believed to be an incarnation of Lord Shiva. He is considered as a role model to understand the nature of human mind and the art of controlling it. He is an embodiment of total self-control, discipline, concentration and ever steadfast in helping his lord Rama to defeat the demon king Ravana and to rescue Sita. Truly, Hanuman is a symbolic of the perfect mind and embodies the highest potential one can achieve. His image is enshrined in gymnasiums all over India and wrestlers worship him before commencing their practice. Valmiki, in this epic, makes no secret of his admiration for Hanuman. He brilliantly depicts Hanuman’s state of mind before he undertakes to leap across the sea in search of Sita. The mental readiness and enthusiasm motivates his fellow mates before they venture into any mission. When they face any challenges, Hanuman focus on the task completely without any diversion. This is an excellent illustration of elevating oneself joining hands with others (Csnarasimhan’s Weblog, 2010). This mental state is more important before undertaking any important mission which requires extraordinary mental stability and physical strength (Vanamali, 2010). The preparation of cricketers must include the integrated method of both mental skills along with physical strength. Another interesting method for mental preparation should be visualization (Predoiu et al., 2020). Brown (n.d.) emphasizes that “You must see your goals clearly and specifically before you can set out for them. Hold them in your mind until they become second nature.” Visualizing your goal or success is a guided imagery which is needed for the athlete as a mental rehearsal. Current team Indian captain Rohit Sharma has expressed “Nothing is easy in cricket. Maybe when you watch it on TV, it looks easy. But it is not. You have to use your brain and time the ball” (Lokapally, 2017). This clearly shows that the role of mind in cricket. Visualization is key factor in training the mind to execute the plans on the field. The power of visualizing the success will have an impact on mind which enhances a player’s performance. By training the mind, a player can actually perform the skill more effectively and efficiently as well. Hence, being mentally prepared will help the players to balance their success and failures. This readiness can be achieved by cricket players through more psychological training sessions (Pybus, n.d.). Optimism The third component of mental fitness is Optimism. It is an attitude that has to be developed in all individuals to attain greater heights in life. It is a basic need for the professional cricketers who represent the country and has been considered as role model for many people at the very young age. Optimism ingrains strength, courage, powerful energy for players to face any challenge on and off the field quite easily. To create a positive atmosphere, positive talks have to be encouraged among the players to boost their confidence level. It changes the perception of every player and helps them to balance emotions to handle both the success and failure in the same way. Indian cricketer Virat Kohli emphasized that his mantra lies in optimism. He quotes “It’s very basic, simple. One-day cricket, T20 cricket, anything, you have to be positive in your mindset” (Cricket Country, 2017). For every player who involved in team sport-like cricket having a positive mindset is quiet difficult, because though they expect success they may get failure which is unexpected. In cricket, the negative result may not affect the players but the people who believe in them. Generating a sense of positive attitude and giving positive reinforcement to oneself will have a greater impact in the game. Though a player is physically strong and makes better strategies, without positive approach, he/she would not have succeeded. A player may possess talent and great skills, but if he carries negative thoughts and fear in his mind which obviously lead to many failures (Donnelly, 2017). The life of Karna has taken for this study as an instance to prove how focusing on the negative aspects in life lead to failures in spite of having greater physical strength and skills. The great archer Karna has possessed extraordinary skills but spent all his life in a dilemma due to the humiliations of being born as a sutaputra (low caste) and the external negative impulses. Karna throughout his life seeks for recognition and focuses more on negative aspects of his life. Being a kind-hearted and righteous person, the tragic hero has failed to take right decisions at the crucial times due to mental stress. Duryodhana has utilized Karna and recognized as a King of Anga. Karna values friendship and he is ready to do anything for the sake of Duryodhana. This blind bond misguided him which led to his defeat and finally ended his life. Saguni and Duryodhana misguided and Karna used his skills for their evil intentions which he is not known. It proves that the psychological injuries is quite harmful than physical injuries. In his deathbed, he came to know about the evil intentions of Duryodhana and realized his mistakes at the end. Throughout Karna’s life, he focuses on the negative aspects of his own life. As a consequence, he is easily being trapped by Saguni and lost his precious life (Ande, 2021). When a person is into trap of negativity and unable to control his own destructive thoughts, it simply means that he/she literally gives the remote of their own life for taking over the control to somebody else. Hence, they tend to lose their confidence and uniqueness that leads to failures in life. Success and failures are inseparable part of cricket or in any games. To accept failures, by learning from the mistakes and to handle success, staying humble by understanding the truth is called sportsmanship. “The people who succeed aren’t the ones who avoid failures; they’re the ones who learn how to respond to failures with optimism” (Donnelly, 2017, p. 92). If a player is strong enough to handle failures, he/she will focus on his performance, then success not on the success, gradually, that leads to his/her success. Cricketers should have the same tendency to look both success and failures as one because both are temporary. For any cricketer, in particular young players, optimism is a key factor in attaining such mindset. It would help them to fight against emotional turbulence that arises within them during crucial hours of play. It induces energy to develop the fighting spirit and finally sustains them in the professional cricket in the long run. Resilience Resilience is the fourth component of mental fitness. In cricket, there are various factors that affect a player’s career; those are continuous failures, loss of form, injuries, mental health issues, disciplinary issues, etc. Sometimes, these factors end their professional life and make them even more depressed. Resilience helps the cricketers getting back to rhythm after struggles. It is an arduous task for anyone. Obviously it takes time, but it depends on individual’s inner strength which helps them to overcome such hardships and develop their confidence level (Edgley, 2021). Looking for encouragement and support can be helpful, in contrary it may lead the victim to fall for dependency. Cricketers are the heroes of the game. A hero’s essential nature is to keep hope alive and never give up easily against any sorts of hurdles. It sets a hero apart from the other normal people. Abhimanyu, Arjuna’s son and the great mythological hero from Mahabharata, is remembered for his physical and mental strength (Daily Herald, 2016). Being a young warrior, he valiantly defended his army and defeated many warriors from Kaurava army. Encountering the Chakravyuha of Kurukshetra, Abhimanyu is completely aware that he cannot defeat the mighty Kauravas alone. He is high on his confidence and the fearless warrior never ready to give up. This confidence is an important key for resilience. Abhimanyu is the resilient person who fought till his last breath with his full potential. He has taught us an important lesson that confidence and resilience are more powerful than the strength of the opponent (Deepak, 2021). Resilience is one of the popular concepts in positive psychology (Ackerman, 2021). The art of resilience would help cricketers to deal any high-pressure games easily. It instills confidence that players can bounce back into the game at any moment to change the course of the game. In cricket, winning or losing the game hardly matters, but the real challenge is continuing the fighting spirit till the last ball of the game. “Have failed more times than I have succeeded, but I never gave up, and will never give up, till my last breath, and that’s what cricket has taught me” quotes Yuvraj Singh, Former Indian Cricketer, during his final farewell speech (Hindu, 2019). He is the man who hit six runs in the consecutive balls over the ropes and inscribed his name in the golden book of cricket. He is a fighter against cancer and an epitome of perseverance. His struggle to build a success story is reflected in the above quote (Hindu, 2019). In cricket, every ball and even one run can decide the result of the match. During such high-intensity games, resilience is an essential factor that determines the success and every cricketer should be more focused on the field. It gives an opportunity for players to come back with full spirit at any moment of the game. Team Spirit The fifth component of mental fitness is Team spirit. Players should be trained to work as a team and develop their mindset to play accordingly. Cricket is a team game, and each player of the team should have a commitment and responsibility when they represent the country as a team. In a team game, winning the match should not depend on the individual’s performance or records. The team should never rely on any individual player instead every player in the team has to contribute for success. Hence, team building is an essential aspect of team game. Even if an individual player scores hundred or takes many wickets, if there is no contribution from the other team members, they cannot enjoy the success. And also if there is no coordination, the individual efforts go in vain (Heermann, 1997). The concept of team spirit has played an important role in the stories of Indian Mythology. The battle between Pandavas and Kauravas at Kurukshetra in Mahabharata teaches the spirit of team work. Kauravas were assisted in the battlefield by Bhisma, the grand uncle of Pandavas and Kauravas and the statesman of Kuru Kingdom, Guru Dronacharya and his son Ashwathama, the Kauravas’ brother-in-law Jayadratha, the Brahmin Kripa, Kritavarma, King Shalya, Shakuni, and many more. Among these, some are bound by their loyalty toward either Hastinapura or Dhritarashtra, and others are reluctant to fight against Pandavas. Compared to Pandavas, Kauravas have a powerful army which consists of many great warriors. However, Pandavas succeed over Kauravas because of their team spirit, commitment, and coordination which were missing in Kauravas’ army. Arjuna as a great leader leads the army in the war front; the rest of the warriors are obliged to his orders and executed the plans. Though the Kauravas have enough resources, power, and strength, there have failed miserably as a team since they lack team spirit. Pandavas, with minimal resources though equally strength, rely on mental strength and believe in coordination and team spirit which earn them the success. Rahul Dravid former captain, and current head coach of India, often emphasized the importance of mental health in cricket during his interviews. He believed that team work plays a vital role in determining the winning or losing. He quoted that “You don’t win or lose the games because of the 11 you select. You win or lose with that those 11 do on the field” (Dravid, n.d.). All players are from different parts of the country representing specific region or state. The team captain has many responsibilities start from analyzing the players’ strength and weakness, motivating them whenever necessary, and creating a good rapport with the team members. Hence, team spirit and coordination play in determining the success of all ages. Self-Motivation The final and sixth component of mental fitness is self-motivation. Every player need to have self-motivation in order to encourage himself to move forward. Self-motivation is an internal drive propelling everyone to set goals, and keep moving forward to achieve them. To develop self-motivation, self-talk is the very first step (Mead, n.d.). Practicing positive self-talk energizes the cricketers, improvises the mental strength, gives the courage to solve life puzzles, and also guides them to take important decisions (Afremow, 2014). In times of trials and tribulations, the great warriors of Indian mythology can be taken as role models that everyone looks up to. The Indian mythological inspiring stories of Krishna, Arjuna, Rama, Karna, Bheema, Bhisma, Abhimanyu, and so on have set the path for everyone who faces the hurdles of life. The struggles and hardships that these warriors faced are no way inferior to the existing challenges in sports. These heroes are excelled in specific sport to participate in war. Arjuna, Karna, and Rama excelled in archery, Bhisma in martial arts, Bheema and Duryodhan in mace, etc. The positive and negative aspects of these warriors can be the motivating factors and lessons for the contemporary cricketers. This will motivate them to evolve as a great player and also to their overall development as a better human. Former Indian player and bowling coach, Venkatesh Prasad, shared a verse from Bhagavad Gita through twitter. He expressed that “The Bhagavad Gita is a phenomenal book of learning and applying in life and I have been very inspired by the Bhagavad Gita. Wanted to share a verse: “Chapter 3, Verse 35.” He said in his video that “is far better to perform one’s natural prescribed duty, though tinged with faults, than to perform another’s prescribed duty, though perfectly. In fact, it is preferable to die in the discharge of one’s duty, than to follow the path of another, which is fraught with danger.” He further expressed how the application of great lessons from mythological texts would change his perspectives and act as a guidance for self-motivation. A real player is the one who motivates himself without seeking appreciation from others. In Cricket, the players who represent the country at very young age can consider the teachings from Bhagavad Gita as great lessons and mythological heroes as an inspiration to motivate them. For young cricketers, self-motivation is a key factor in bringing out one’s true potential, preventing them from distractions and making them to survive in the cricket for longer. Conclusion This paper limelights the importance of mental fitness for cricketers through the mythological stories. The life of the mythological heroes serves as a psychic escort for contemporary cricketers. This analysis highlights the need for mental fitness along with physical fitness for professional cricketers to make them mentally strong individuals. This paper proposes a model for mental fitness named SPORTS that consists of six components and its functions in detail. These components of mental fitness which is discussed above would help players mastering their own mind for their overall development physically, mentally, emotionally, and spiritually. Lao Tzu, Chinese Philosopher, quotes “Knowing others is intelligence; knowing yourself is true wisdom. Mastering others is strength, mastering yourself is true power.” Being in limelight, the elite cricketers have to face the challenges on and off the field. As mastering own mind is true power, this proposed model guides the players to focus on the aspects of self-awareness, mental preparation, increasing positivity, confidence, team spirit, and self-motivation. As an individual, this would definitely pave the way to understand themselves to overcome any challenges on and off the field. Hence, the players can utilize these major aspects for their own welfare for personal growth. The stories of the mythological warriors act as a motivating factor to impact the minds of the contemporary cricketers. It would help them to connect themselves with those great warriors to boost their confidence level. By incorporating these aspects of mental fitness along with physical fitness into the training session will result in better output from the players. Finally, the purpose of this paper is to highlight the need of mental fitness in cricket. Hence, playing cricket will not only benefit the players to become stronger individuals physically, but to make them stronger emotionally, mentally, and spiritually. It guides them to understand the essence of their unique individuality and utilize it for their own betterment to represent their country with responsibility and also make them eligible enough for being a role model for young generation. Acknowledgements Not applicable. Authors' Contributions Both authors contributed to the study conception and design. Material preparation was done by IR and analyzed by SL. The first draft of the manuscript was written by IR and edited by SL. Both authors read and approved the final manuscript. Funding No funding was received to assist with the preparation of this manuscript. Data Availability Statement Not applicable. Declarations Conflict of interest Not applicable. Ethics Approval Not applicable. Consent to Participate Not applicable. Consent for Publication Not applicable. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Ackerman, E. (2021, December 12). What is resilience and why is it important to bounce back? Retrieved from https://positivepsychology.com/what-is-resilience/ Afremow J The champion’s mind: How great athletes think, train, and thrive 2014 Rodale Books Ande N Suryaputra Karna: Chasing my identity and dreams 2021 White Falcon Publishing Barker J Slater M It’s not just cricket The Psychologist 2015 28 552 557 Bhagavatam katha. (n.d.). Mahabrarata story: I see only the round black eye of the bird! Retrieved from http://www.bhagavatam-katha.com/mahabrarata-story-i-see-only-the-round-black-eye-of-the-bird/ Biddle SJH Mental preparation, mental practice and strength tasks: A need for clarification Journal of Sports Sciences 1985 3 1 67 74 10.1080/02640418508729733 4094020 Brown, L. (n.d.). AZQuotes.com. Retrieved January 24, 2022, from https://www.azquotes.com/quote/1310022 Colman W On being, knowing and having a self Journal of Analytical Psychology 2008 53 3 351 366 10.1111/j.1468-5922.2008.00731.x 18494672 Cricket Country. (2017, August 23). Virat Kohli’s success mantra lies in ‘optimism’ and ‘positivity.’ Retrieved from https://www.cricketcountry.com/news/virat-kohlis-success-mantra-lies-in-optimism-and-positivity-635613 CricketNext. (2021, November 08). Indian players are mentally and physically drained: Ravi Shastri on bubble fatigue. Retrieved from https://www.news18.com/cricketnext/news/t20-world-cup-indian-players-are-mentally-and-physically-drained-ravi-shastri-on-bubble-fatigue-4419302.html Crook JH The evolution of human consciousness 1980 Clarendon Press Csnarasimhan’s Weblog. (2010, January 10). Hanuman as a motivating leader. Stories from Ramayana. Retrieved from https://csnarasimhan.wordpress.com/2010/01/10/hanuman-as-a-motivating-leader/ Daily Herald. (2016, February 18). Resilience: Learning from mythological heroes. Retrieved from https://www.heraldextra.com/special-section/2016/feb/18/resilience-learning-from-mythological-heroes/ Deepak MR Abhimanyu: The warrior Prince 2021 Bloomsbury Donnelly D Relentless optimism: How a commitment to positive thinking changes everything 2017 Shamrock New Media Inc Dravid, R. (n.d.). AZQuotes.com. Retrieved from https://www.azquotes.com/quote/865299 Edgley R The art of resilience: Strategies for an unbreakable mind and body 2021 HarperCollins Frank LS Ronald ES Techniques for improving self-awareness of youth sports coaches Journal of Physical Education and Recreation 1980 51 2 46 52 10.1080/00971170.1980.10624089 Gentile A Trivic T Bianco A Lakicevic N Figlioli F Roklicer R Eliseev S Tabakov S Maksimovic N Drid P Living in the “bubble”: Athletes’ psychological profile during the Sambo world championship Frontiers in Psychology 2021 12 657652 10.3389/fpsyg.2021.657652 34122241 Hamm M. (n.d.). AZQuotes.com. Retrieved January 24, 2022, from https://www.azquotes.com/quote/537411 Heermann B Building team spirit: Activities for inspiring and energizing teams 1997 McGraw Hill Higher Education Kovacevic, R. (2021, February 11). Mental health: Lessons learned in 2020 for 2021 and forward. Retrieved from https://blogs.worldbank.org/health/mental-health-lessons-learned-2020-2021-and-forward Lokapally, V. (2017, December 13). Nothing's easy in cricket', SPORTSTAR. https://sportstar.thehindu.com/cricket/india-vs-srilanka-2017/rohit-sharma-nothings-easy-in-cricket/article21609488.ece Mack G Casstevens D Mind gym: An athlete’s guide to inner excellence 2002 McGraw Hill Mead, E. (n.d.). What is positive self-talk? Retrieved from https://positivepsychology.com/positive-self-talk/ Modh S Discover the Arjuna in you 2015 Jaico Publishing House Morin A Self-awareness part 1: Definition, measures, effects, functions, and antecedents Social and Personality Psychology Compass 2011 5 10 807 823 10.1111/j.1751-9004.2011.00387.x Predoiu R Predoiu A Mitrache G Firănescu M Cosma G Dinuţă G Bucuroiu RA Visualisation techniques in sport - the mental road map for success Discobolul – Physical Education, Sport and Kinetotherapy Journal 2020 59 3 245 256 10.35189/dpeskj.2020.59.3.4 Purcell R Gwyther K Rice SM Mental health in elite athletes: Increased awareness requires an early intervention framework to respond to athlete needs Sports Medicine – Open 2019 5 46 10.1186/s40798-019-0220-1 31781988 Pybus, R. (n.d.). Cricket mental training for optimal performance. Cricketlab. Retrieved from http://www.cricketlab.co/cricket-mental-training.html Rao MR Efficacy of yoga in facilitating mindfulness among asymptomatic male cricket players Indian Journal of Science and Technology 2020 13 22 2182 2188 10.17485/IJST/v13i22.614 SportsAdda. (2021, April 1). Yo-Yo test in cricket: The ultimate fitness benchmark for players. Retrieved from https://www.sportsadda.com/cricket/features/what-yo-yo-test-fitness-cricket The Hindu. (2019). Yuvraj Singh’s final farewell speech. Retrieved from https://www.thehindu.com/news/resources/yuvraj-singhs-final-farewell-speech/article27762557.ece The Indian Express. (2019, August 7). The process is more important than the result: M S Dhoni. Retrieved from https://indianexpress.com/article/lifestyle/life-positive/ms-dhoni-process-is-more-important-than-the-result-m-s-dhoni-inspiring-video-good-morning-5882798/ Vanamali D Hanuman: The devotion and power of the monkey god 2010 Inner Traditions Wood, R. J. (2010). Complete guide to fitness testing. Topendsports.com. Retrieved 7 March 2016, from https://www.topendsports.com/testing/	36474579	PMC9716539	NO-CC CODE	2022-12-03 23:20:58	no	Psychol Stud (Mysore). 2022 Dec 2;:1-7	utf-8	Psychol Stud (Mysore)	2,022	10.1007/s12646-022-00686-3	oa_other
==== Front Eur Phys J Spec Top Eur Phys J Spec Top The European Physical Journal. Special Topics 1951-6355 1951-6401 Springer Berlin Heidelberg Berlin/Heidelberg 36475056 724 10.1140/epjs/s11734-022-00724-1 Editorial Dynamics of the COVID-19 pandemic: nonlinear approaches on the modelling, prediction and control Banerjee Santo [email protected] grid.4800.c 0000 0004 1937 0343 Department of Mathematical Sciences, Politecnico di Torino, Turin, Italy 2 12 2022 2022 231 18-20 32753280 © The Author(s), under exclusive licence to EDP Sciences, Springer-Verlag GmbH Germany, part of Springer Nature 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This special issue contains 35 regular articles on the analysis and dynamics of COVID-19 with several applications. Some analyses are on the construction of mathematical models representing the dynamics of COVID-19, and some are on the estimations and predictions of the disease, a few with possible applications. The various contributions report important, timely, and promising results, such as the effects of several waves, deep learning-based COVID-19 classifications, and multivariate time series with applications. issue-copyright-statement© EDP Sciences, Springer-Verlag GmbH Germany, part of Springer Nature 2022 ==== Body pmcIntroduction Over 200 countries have been affected by the COVID-19 pandemic, resulting in the death of millions of humans. Numerous researchers from different interdisciplinary fields around the world are involved in studies to examine the behavior and nature of the virus and also to develop a potential vaccine against the virus. Several analyses are also focused on “long COVID,” and the possible post-COVID conditions and implications for human health and society. Nonlinear tools have been effectively used for the predictions, analysis, and understanding of the dynamics of the pandemic. This special issue is a compilation of original research articles that address the dynamics and applications of COVID-19 through nonlinear dynamics. The articles are organized in five sections, comprising mathematical modeling and epidemics [1–7], the dynamics of several waves and transmission [8–17], neural network and deep learning related to COVID-19 [18–24], predictions and estimations related to COVID-19 [25–30], and detailed analysis on the pandemic and its applications [31–35]. Epidemics and mathematical modeling This section presents seven research articles on the construction of discrete and continuous mathematical models representing the dynamics of COVID-19. The dynamics of the spread of an infectious disease together with the economic challenges faced by the country are examined through a combined epidemic–economic model proposed by Karim et al. [1]. Nonlinear analysis of the proposed model is performed along with global stability analysis. They investigate the impact of complete vaccination on reduction in the rate of infections. An epidemic model with the effect of noise is investigated by He and Mukherjee [2]. They study the dynamical behavior and complexity of the stochastic model, and verify the sensitivity of the system with varying noise strength and effect of changes in the initial condition on the noise-induced system. The results can be effective for understanding the uncertainty of a noise-induced epidemic model. A mathematical model of unemployment as a consequence of the COVID-19 pandemic in middle-income countries is developed by Chinnadurai et al. [3], using population dynamics. The results show that the model is effective in controlling unemployment in all kinds of populations. A discrete time pandemic model is proposed by Ghosh et al. [4] to predict disease outbreak, employing real-world data, and suitable control measures are presented. They claim that in the future, the dynamics of COVID-19 can demonstrate an oscillating behavior. A compartment model of COVID-19 with inclusion of the vaccinated human populations is proposed by Rana et al. [5]. The vaccine reproduction number is obtained, and the study of equilibrium and stability is carried out. The model parameters are estimated by fitting the model with real-time COVID-19 data employing a nonlinear fitting technique. A discrete susceptible-infected mathematical model is proposed by Bashkirtseva and Ryashko [6]. They used a stochastic sensitivity technique to investigate various noise-induced dynamics related to the system. A fractional-order dynamical system is proposed by Khoojine et al. [7] for understanding and predicting the spread of COVID-19 in Thailand. The accuracy of the model in representing the real-time data of Thailand is investigated via sensitivity analysis of the model with respect to fractional order. Dynamics of the waves and COVID-19 transmission Based on COVID-19 transmission, this section presents 10 regular articles on the analysis of the second wave, transmission of airborne viral aerosol droplets, the effect of a contaminated environment, and direct transmission between humans, etc. The number of infections of COVID-19 in India during the second wave is estimated by Gopal et al. [8] with the help of a four-compartment model comprising susceptible, exposed, infected, and recovered populations. They report that factors such as the individual efforts of people and the support they render to governmental initiatives including the implementation of curfews and vaccination strategies are vital in controlling the pandemic from both a present and future perspective. Based on a nonlinear dynamical model representing the dynamics of the pandemic, Natiq and Saha [9] present the random behavior and unpredictable nature of the COVID-19 outbreak among the prey species and humans infected by the species by means of direct and indirect contact. A simplistic reaction–diffusion model demonstrating the contraction of the virus from contaminated aerosols in a closed room with ventilation is proposed by Turkyilmazoglu [10]. The spatiotemporal dynamics of the aerosol concentration with infectious coronavirus is explored in the article. The necessary safety measures and precautions to be adopted against the risk of infection in the private and public sectors can be framed based on the results they present. The outbreak of COVID-19 infections in terms of amplitude equations is investigated by Frank and Smucker [11]. The dynamics of the infected population are studied at different epidemic waves; an analytical method is employed to derive the eigenvectors and their respective amplitudes for low-dimensional models, and computational methods are implemented for high-dimensional models. The authors introduce the idea of stages of epidemics, and system behavior is discussed based on the nature of the eigenvectors. In particular, the first wave of the COVID-19 epidemic in the state of New York and in Pakistan is explicitly investigated. A nonlinear mathematical model is proposed by Sarkar et al. [12] to analyze the transmission dynamics of the COVID-19 pandemic in India. The data on daily cases of infection in India are used for model fitting and parameter estimation. The proportional impact of environmental contamination on the increase in infection due to COVID-19 is established, but with disinfection of the environment by sanitation, no drastic increase in the rate of infections is seen. The spread of the coronavirus under the mobility effect in certain regions is investigated by Amoedo et al. [13], and the study focuses specifically on Spain. The way to control inconsistency in the spread of COVID-19 is studied using a regression model. A new method is developed to optimize the usage of data from Google Trends. The results suggest that diversity among the regions is vital in understanding pandemic containment strategies. The impact of cytokine cells in building the immune response against COVID-19 infections is presented as a mathematical model by Rana et al. [14]. An optimal control problem with immunomodulatory therapy is discussed, employing a linear feedback method to improve healthy cells in humans. The results highlight the importance of immunomodulatory therapy in controlling cytokines for restoring the immune system of ill patients and helping them to reach a healthy state. A COVID-19 transmission model taking into account the direct transmission of the virus between two individuals of which one is infected, together with virus transmission through air, is proposed by Pal and Ghosh [15]. Real-time data of two districts are used to estimate the parameter values employing the nonlinear least-squares technique. The study suggests that increasing vaccination among the susceptible population and treatment of the infected population play a crucial role in containing the disease in the two districts considered. In their work, Ghosh et al. [16] consider an epidemic model of disease spread considering the movement of people and indirect virus transmission by means of a contaminated environment and puff clouds. The numerical calculation of the probability of infection among individuals and survival distances are performed, and values are presented. The influence and impact of the indirect transmission of infection are presented via simulations with respect to several parameters. Based on statistical COVID-19 data, interpretation of multiple waves is studied in the sense of local and global scenarios. A six-compartment mathematical model including the vaccinated, home-isolated, and hospitalized populations together with susceptible, infected, and recovered populations is developed by Devi et al. [17]. The vaccination campaign in controlling the disease spread is given importance by including the impact of speed and effectiveness of the campaigns. The model constructed is validated with real-time data, and the graphical results are presented. The impact of vaccination is studied with data, and the results suggest that isolation and hospitalization are highly necessary until direct transmission among individuals is reduced to a certain level after complete vaccination of every nation. Deep learning and cellular automata related to COVID-19 Chowdhury et al. [18] investigate the spatial and temporal behavior of the disease with the help of a cellular automata (CA) model, and discuss the spread of the disease based on the model. Neighborhood criteria are proposed for measuring the social confinement at the time of the spread of the disease. The model is fitted to the real-time statistical data for different waves in India, and predictions are presented based on infections and social interaction by varying the values of different parameters. In their study, Appasami et al. [19] propose a deep learning-based convolutional neural network (CNN) model with an automatic extraction method for detection of COVID-19 from chest X-ray images collected from different data sources. The chest X-ray images of patients are used for analysis by means of data augmentation, which guides medical professionals in the diagnosis of COVID-19 under their heavy workload situation. Accuracy of 93% is achieved by the choice of best optimizer for classification of COVID-19 data. In their study, Kumar and Alphonse [20] propose the classification of COVID-19 and other human respiratory sound diseases by a lightweight convolutional neural network with a modified mel-frequency cepstral coefficient (M-MFCC) using different depths and kernel sizes. A comparison between the abnormalities of COVID-19 and other respiratory sound diseases is carried out. Different contextual information supporting the classification of data is obtained as the application of different receptive fields and depths. The authors demonstrated the effectiveness of the classification technique employed with the existing technique and applicability of the technique under different scenarios. The effects of the coronavirus in countries such as India and the USA that exhibit great variation from day to day in cases of infection are investigated by Bhardwaj and Bangia [21]. The study validates the performance measure via mean absolute standard error (MASE), mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), and coefficient of determination (R2). The goodness of fit for data values of India and the USA are obtained using wavelet neuronal network fuzzified inferences’ layered multivariate adaptive regression spline (WNNFIL–MARS). The authors also suggest that WNNFIL–MARS provides a better fit than the other prediction models. A new regularized deep convolutional neural network (RDCNN) architecture capable of accepting the sound data and their features is proposed by Kumar and Alphonse [22], and they evaluated the sound data for COVID-19. The proposed model presents a better learning curve in comparison with the other models when trained with COVID-19 sound datasets. Their experimental results suggest that 2–3% greater accuracy in classification for 3 × 3 kernel size is exhibited by RDCNN (without max pooling) than by RDCNN with max pooling. The authors also suggest that the proposed technique is better suited and achieves better results for respiratory diseases. To determine the effect of lockdown due to COVID-19 on public health, Adak et al. [23] collected data before and after lockdown from 200 random individuals from a municipal region of the state of West Bengal, primarily focusing on six parameters, namely fasting blood sugar, systolic blood pressure, insomnia, diastolic blood pressure, respiratory distress, and cholesterol. A model is formulated using the adaptive neuro-fuzzy inference system (ANFIS) approach with the use of real-time data. The study results suggest that the impact of lockdown on healthy individuals is negligible, while individuals with poor health incur significant effects on health. A knowledge-based approach for the implementation of safety measures to combat the spread of COVID-19 is introduced by Geetha et al. [24]. The analysis is performed based on five different dimensions, including correlating the detected and confirmed COVID-19 cases, speed of detection, maintaining social distance, wearing of masks, and correlating the imported and inbound cases. The actions are based on the level of security in the considered region. The proposed algorithm will serve as a guide to the government for the implementation of several measures, including maintaining sufficient distance among individuals, wearing safety masks, and other policies. This proves to be an effective way to prevent future danger and to implement precautionary policies. Also, the proposed approach will improve the governance of communities. COVID-19-related estimations and predictions This section presents six regular articles based on prediction analysis related to COVID-19 data. The article by Biswas [25] aims at real-time prediction of confirmed infections in India and the USA. Neural network autoregression (NNAR)- and autoregressive moving average (ARIMA)-based models are considered for daily data on COVID-19 infections. Their results provide evidence proving that the performance of the hybrid models is better than that of single models. They also present results on advanced hybrid models implementing a wavelet-based approach for better accuracy in assessments (MAE and RMSE). The present situation in India and the USA is illustrated by obtaining a reproduction number that is time-dependent. The aim of the study by Saha et al. [26] is improving the adherence to medication based on active reminders. A detailed study is performed to measure the impact of several factors, including the perception of patients regarding side effects, beliefs, and physician instruction, on medication adherence. These factors play a vital role in improving the rate of adherence. The authors collected real-time data from Sikkim, a state in India. An equation is framed based on the responses obtained from the individuals, and the outcome of the study suggests that there is a significant effect of the importance given by the patients to the physicians and their beliefs. A mathematical compartment model calibrating the effect of the host immune system on COVID-19 infections is proposed by Mondal et al. [27]. They present a mathematical analysis to study the occurrence of the transcritical bifurcation and the analysis of the stability of the steady states. The study also investigates the effect of external factors on the virus reproduction rate, and the formation of backward bifurcation is analyzed. The authors also illustrate the role of the immune system and immunopathology in inhibiting complex epidemic states, and suitable support for the analytical findings is provided by means of simulations. One of the important aspects of analysis based on multivariate time series is adopted by James and Menzies [28] to understand the state-wise progression of the rate of infection to rate of death in the USA as a function of time. They propose a nonlinear framework to investigate the time-varying offset. The study is performed employing four different approaches, and the results are found to behave in an “up-down-up” pattern and also help in predicting the workload of the healthcare systems and assist in the allocation of sufficient resources to the states in need. Markovian-type models for estimating the spread of the COVID-19 virus in the future considering the exponentially distributed time duration of the population in each compartment of the model are proposed by Basnarkov et al. [29]. The study reveals that the starting time of the infection among individuals coincides with the same time as that of symptoms, while the incubation period is not exponentially distributed. The distribution of the incubation period and the infectivity time period are used to estimate the basic reproduction number (R0) for COVID-19. The importance of employing the Markovian approach in prediction is presented, and the need for cautious implementation of the method is also suggested. The end time of the pandemic together with the final size, maximum number of populations infected and the time taken to attain the peak infection, and the removal rate of the infected population in China due to COVID-19 are estimated by Pei and Hu [30] with a nonautonomous susceptible–infected–removed (SIR) model with time delay. Though the model appears to be simple and requires very few data, the predictive results are more accurate and effective in reflecting the real-time situation. The results suggest that governmental policymakers can employ several techniques to control the infection rate and improve the economic and social status of the country by increased production and resumption of work. COVID-19-related analysis and applications This section presents five regular articles based on the analysis and applications of the novel coronavirus. The article by He et al. [31] implements the theory of multifractals to illustrate the impact of infections at different age groups by classifying X-ray images. The study is performed at different levels of noise using filtered and normal X-ray images for better understanding of lung infections. Denoised and noisy X-ray images are compared for calculating the peak signal-to-noise ratio values and mean absolute error with a median filter approach. Age-based analysis of oxygen levels is performed by constructing three-dimensional (3D) visualization using fractal dimension values. The study also reveals increasing complexity of lung infections among older people in comparison with younger people based on their X-ray images. The rate of increase in infection and death due to COVID-19, introducing the orthonormal basis, is investigated by Chen [32]. The study based on an orthonormal basis reveals several unknown features of the pandemic via Fourier coefficients. The values of the coefficients for a considered sample of countries are ranked as ranking vectors. The countries are then categorized by spectral clustering based on the Manhattan metric. The spectral analysis proves superior to classical analysis techniques in describing the internal structures of the time series. Further, they calculate the approximated numbers of infected individuals and deaths. The primary goal of the article by Malla and Alphonse [33] is to create awareness among people by educating them in differentiating real information from fake news that is being spread. They investigate several social networking websites including Instagram, Facebook, and Twitter that contain fake data on the spread of COVID-19. Various deep learning techniques are employed to test the fake data from tweets that are categorized. The technique proposed in this article outperforms the most effective deep learning models CT-BERT (COVID-Twitter Bidirectional Encoder Representations from Transformers) and RoBERTa (Robustly Optimized BERT Pre-training Approach) employed for fake COVID-19 data analysis with the help of the multiplicative fusion technique. The performance of the proposed model is superior due its ability to overcome the disadvantages of those techniques, and the results are obtained at an accuracy rate of 98.88% and F1 score of 98.93%. In the work by Samadder and Ghosh [34], a study is carried out considering the stock indices of some influential markets of the countries that experienced major impacts from COVID-19. Six major impacted countries with considerable influence are selected. The authors conclude that the economic growth of those markets will recover when the impact of the COVID-19 virus begins to fade. Thangaraj and Easwaramoorthy [35] present a different characterization of the disease and its severity by comparing denoised and noisy computed tomography (CT) scan images collected from COVID-19-infected patients, implementing the theory of multifractals. Multifractal measures are examined, and the CT scan images are classified and investigated by means of filtered and edge detection methods. The authors employ the Robert, Prewitt, and Sobel edge detection algorithms for the converted images to compute various qualitative measures. The comparison results reveal that the Sobel method provides better classification of the CT scan images of COVID-19-infected patients than the other algorithms due to the greater image complexity exhibited by the Sobel method. The study is supported with some statistical analysis of variance (ANOVA) tests, and simulations such as box plots are provided and the experimental images are explored statistically. Conclusion The outbreak of COVID-19 changed the human perception of day-to-day life and tested the bounds of medical technology in protecting the welfare of humans. Several approaches and safety measures have been implemented to minimize the countless lives that are being affected. However, public health and educational breaches are evidenced in most countries in which not all citizens have the same opportunities to deal with the pandemic. Therefore, this has led to pervasive consequences, including mental health problems because of the disruption of everyday life routines. This special issue is a collection of 35 research articles providing insight into the spread of the coronavirus and control measures against the COVID-19 pandemic. Acknowledgements The editor of this special issue would like to thank the authors for their valued contributions, and the referees for their dedicated efforts in reviewing the articles. We believe that the selected papers gathered here will enrich readers’ knowledge and will help scientists and researchers to further develop the theory of fractal analysis and related applications. Lastly, we wish to express our sincere gratitude to all members of EPJ ST for hosting this special issue. Data availability No data was associated with the manuscript. ==== Refs References 1. Karim F Chauhan S Dhar J Analysing an epidemic–economic model in the presence of novel corona virus infection: capital stabilization, media effect, and the role of vaccine Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00539-0 2. He S Mukherjee S Exploration of stochastic dynamics and complexity in an epidemic system Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00658-8 3. Chinnadurai K Athithan S Mathematical modelling of unemployment as the effect of COVID-19 pandemic in middle-income countries Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00620-8 4. Ghosh D Santra PK Mahapatra GS Elsonbaty A Elsadany AA A discrete-time epidemic model for the analysis of transmission of COVID19 based upon data of epidemiological parameters Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00537-2 5. Rana PS Sharma N The modeling and analysis of the COVID-19 pandemic with vaccination and treatment control: a case study of Maharashtra, Delhi, Uttarakhand, Sikkim, and Russia in the light of pharmaceutical and non-pharmaceutical approaches Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00534-5 6. Bashkirtseva I Ryashko L Analysis of stochastic dynamics in a multistable logistic-type epidemiological model Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00618-2 7. Khoojine AS Mahsuli M Shadabfar M Hosseini VR Kordestani H A proposed fractional dynamic system and Monte Carlo-based back analysis for simulating the spreading profile of COVID-19 Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00538-1 8. Gopal R Chandrasekar VK Lakshmanan M Analysis of the second wave of COVID-19 in India based on SEIR model Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00426-8 9. Natiq H Saha A In search of COVID-19 transmission through an infected prey Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00429-5 10. Turkyilmazoglu M Indoor transmission of airborne viral aerosol with a simplistic reaction-diffusion model Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00614-6 11. Frank TD Smucker J Characterizing stages of COVID-19 epidemics: a nonlinear physics perspective based on amplitude equations Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00530-9 12. Sarkar K Mondal J Khajanchi S How do the contaminated environment influence the transmission dynamics of COVID-19 pandemic? Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00648-w 13. Amoedo JM Atrio-Lema Y del Carmen Sánchez-Carreira M Neira I The heterogeneous regional effect of mobility on Coronavirus spread Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00533-6 14. Rana P Chauhan S Mubayi A Burden of cytokines storm on prognosis of SARS-CoV-2 infection through immune response: dynamic analysis and optimal control with immunomodulatory therapy Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00435-7 15. Pal S Ghosh I A mechanistic model for airborne and direct human-to-human transmission of COVID-19: effect of mitigation strategies and immigration of infectious persons Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00433-9 16. Ghosh S Chakraborty A How surface and fomite infection affect contagion dynamics: a study with self-propelled particles Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00431-x 17. Devi MB Devi A Gupta PK Tripathi D Response of vaccination on community transmission of COVID-19: a dynamical approach Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00652-0 18. Chowdhury S Roychowdhury S Chaudhuri I Cellular automata in the light of COVID-19 Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00619-1 19. Appasami G Nickolas S A deep learning-based COVID-19 classification from chest X-ray image: case study Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00647-x 20. Kumar LK Alphonse PJA COVID-19 disease diagnosis with light-weight CNN using modified MFCC and enhanced GFCC from human respiratory sounds Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00432-w 21. Bhardwaj R Bangia A Hybridized wavelet neuronal learning-based modelling to predict novel COVID-19 effects in India and USA Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00531-8 22. Kumar LK Alphonse PJA COVID-19: respiratory disease diagnosis with regularized deep convolutional neural network using human respiratory sounds Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00649-9 23. Adak S Majumder R Majee S Jana S Kar TK An ANFIS model-based approach to investigate the effect of lockdown due to COVID-19 on public health Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00621-7 24. Geetha S Narayanamoorthy S Manirathinam T Ahmadian A Bajuri MY Kang D Knowledge-based normative safety measure approach: systematic assessment of capabilities to conquer COVID-19 Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00617-3 25. Biswas S Forecasting and comparative analysis of Covid-19 cases in India and US Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00536-3 26. Saha SK Adhikary A Jha A Enhancement in medication adherence amidst COVID-19 using active reminders Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00532-7 27. Mondal J Samui P Chatterjee AN Dynamical demeanour of SARS-CoV-2 virus undergoing immune response mechanism in COVID-19 pandemic Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00437-5 28. James N Menzies M Estimating a continuously varying offset between multivariate time series with application to COVID-19 in the United States Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00430-y 29. Basnarkov L Tomovski I Avram F Estimation of the basic reproduction number of COVID-19 from the incubation period distribution Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00650-2 30. Pei L Hu Y Long-term prediction of the sporadic COVID-19 epidemics induced by δ-virus in China based on a novel non-autonomous delayed SIR model Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00622-6 31. He S Thangaraj C Easwaramoorthy D Muhiuddin G Multifractal analysis on age-based discrimination in X-ray images for sensing the severity of COVID-19 disease Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00615-5 32. Chen R-M Analysing deaths and confirmed cases of COVID-19 pandemic by analytical approaches Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00535-4 33. Malla SJ Alphonse PJA Fake or real news about COVID-19? Pretrained transformer model to detect potential misleading news Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00436-6 34. Samadder S Ghosh K The changing economic relationship between some of the major COVID-19 impacted countries with prominent wealth: a comparative study from the view point of stock markets Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00616-4 35. Thangaraj C Easwaramoorthy D Generalized fractal dimensions based comparison analysis of edge detection methods in CT images for estimating the infection of COVID-19 disease Eur. Phys. J. Spec. Top. 2022 10.1140/epjs/s11734-022-00651-1	36475056	PMC9716540	NO-CC CODE	2022-12-16 23:17:51	no	Eur Phys J Spec Top. 2022 Dec 2; 231(18-20):3275-3280	utf-8	Eur Phys J Spec Top	2,022	10.1140/epjs/s11734-022-00724-1	oa_other
==== Front J Econ Race Policy J Econ Race Policy Journal of Economics, Race, and Policy 2520-8411 2520-842X Springer International Publishing Cham 109 10.1007/s41996-022-00109-5 Original Article Racial and Ethnic Disparities in Housing Instability During the COVID-19 Pandemic: the Role of Assets and Income Shocks http://orcid.org/0000-0003-4042-7984 Chun Yung [email protected] 12 Roll Stephen 12 Miller Selina 2 Lee Hedwig 13 Larimore Savannah 3 Grinstein-Weiss Michal 12 1 grid.4367.6 0000 0001 2355 7002 Social Policy Institute, Washington University in St. Louis, St. Louis, USA 2 grid.4367.6 0000 0001 2355 7002 Brown School of Social Work, Washington University in St. Louis, St. Louis, USA 3 grid.4367.6 0000 0001 2355 7002 Department of Sociology, Washington University in St. Louis, St. Louis, USA 2 12 2022 119 16 5 2022 3 11 2022 12 11 2022 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Stable and adequate housing is critical to sound public health responses in the midst of a pandemic. This study explores the disproportionate impact of the COVID-19 pandemic on housing-related hardships across racial/ethnic groups in the USA as well as the extent to which these disparities are mediated by households’ broader economic circumstances, which we operationalized in terms of prepandemic liquid assets and pandemic-related income losses. Using a longitudinal national survey with more than 23,000 responses, we found that Black and Hispanic respondents were more vulnerable to housing-related hardships during the pandemic than white respondents. These impacts were particularly pronounced in low- and moderate-income households. We found that liquid assets acted as a strong mediator of the housing hardship disparities between white and Black/Hispanic households. Our findings imply that housing became less stable for minority groups as a result of the pandemic, particularly those households with limited liquid assets. Such housing-related disparities demonstrate the need for policies and practices that target support to economically marginalized groups and families of color in particular. Keywords COVID-19 Race Ethnicity Housing Liquid assets Employment ==== Body pmcIntroduction The COVID-19 pandemic has had unprecedented economic effects in the USA. In April 2020 alone, an estimated 20.5 million Americans lost their jobs, increasing the unemployment rate to 14.7% (Bureau of Labor Statistics, 2020). Just as COVID-19 mortality and hospitalization rates have disproportionately burdened racial and ethnic minorities (Shah et al. 2020; Townsend et al. 2020), so too have the economic effects of the pandemic. Early data show racial/ethnic disparities in unemployment during the pandemic, with Hispanic workers suffering especially high job losses (Fairlie et al. 2020; Karpman et al. 2020). Compounding these disproportionate employment impacts, racial and ethnic minorities tended to hold much lower levels of emergency savings prior to the pandemic. A recent study of nearly 1 million bank accounts found that in the years prior to the pandemic white1 account owners held roughly 2 and 3 times as much in liquid savings as Hispanic and Black account owners, respectively (Farrell et al. 2020). Thus, even as the economic and health burdens of the pandemic fell disproportionately on Black and Hispanic families, these groups were also in a worse position to withstand them financially. As economic burdens and housing hardship frequently go hand in hand, housing impacts may also have fallen unequally across racial and ethnic lines. This study aims to explore and explain the evolution of disparities in housing-related hardships across racial/ethnic groups over the course of the pandemic. Though researchers have examined the relationships between both race/ethnicity and socioeconomic status on housing instability (Desmond 2012; Heflin 2017; Medina et al. 2020; Niedt and Martin 2013; Pilkauskas et al. 2012), our work builds on this research by (a) examining how racial/ethnic disparities in housing instability manifest and evolve in the context of a broad-based and acute economic shock like the COVID-19 pandemic and (b) by conducting an explanatory analysis of the extent to which housing-related disparities are driven by disparities in economic circumstances. Leveraging a novel, national longitudinal survey conducted throughout the first year of the pandemic, we found that between May and August, 2020, non-Hispanic Black (hereafter Black) and Hispanic respondents disproportionately experienced eviction, mortgage/rent delinquency, and utility bill payment delays compared to non-Hispanic white respondents (hereafter white). These disproportionate impacts were particularly pronounced among lower-income respondents within these minority groups. In the fall of 2020, as white respondents began to increasingly experience housing instability, the gap between racial/ethnic groups narrowed. As the US economy entered the recovery stage of the pandemic due to the development and distribution of COVID-19 vaccines in early 2021, housing instability levels lowered across all racial/ethnic groups. Building upon empirical evidence from previous research linking racial and ethnic disparities in liquid assets, income, and employment to disparities in housing hardships, we also explore whether disparities in liquid assets and employment shocks explain the impact of the pandemic on Black and Hispanic populations. We found that prepandemic liquid asset amounts mediated the disparities in housing-related hardships between white and Black/Hispanic respondents. However, we find limited evidence that employment shocks during the pandemic explain the disproportionate impacts across racial/ethnic groups. Thus far, the media as well as the academia have reported racial and ethnic disparities in housing stability before and during the pandemic. However, to our knowledge, this study is the first attempt to examine a connection between racial and ethnic disparities in housing instability and financial/employment attributes during an exogenous shock. Focusing on the nationwide (and worldwide) COVID-19 pandemic, our findings add that the connection is not a local problem, but a nationwide issue of the housing market. The remainder of this paper is structured as follows. The first section reviews the literature on disproportionate housing-related hardships across racial/ethnic groups and presents our research questions. The second and third sections describe our data sources and empirical strategy, respectively. The fourth section includes a detailed examination of our results. We conclude with a discussion of the implications for scholars and practitioners. Theoretical Expectations Disproportionate Housing Hardships Across Racial/Ethnic Groups Large disparities exist in the experience of housing hardship across racial and ethnic groups. Black and Hispanic households are more likely to experience housing hardships, such as eviction (Desmond, 2012; Greenberg et al. 2016; Medina et al. 2020) and delays in mortgage, rent, and utility bill payments (Heflin, 2017) compared to white households even after controlling for education and household resources. Medina et al. (2020) used a spatial data analysis model to demonstrate that evictions were clustered in minority-dominant neighborhoods and that residents in these neighborhoods were 66% more likely to be evicted than residents of other neighborhoods. Based on the Survey of Income and Program Participation (SIPP) data, Heflin (2017) found that both Black and Hispanic respondents were more likely to fall behind on rent or mortgage payments than white respondents. External financial shocks generally increase housing hardship, especially for households that are already financially strapped. Financially distressed homeowners are more likely to experience foreclosure than financially nondistressed homeowners (Niedt & Martin, 2013; Pilkauskas et al. 2012). For example, Niedt and Martin (2013) found that those who reported their finances had recently worsened were approximately 1.5 times more likely to experience foreclosure than those in a comparison group, and more than half of those who had experienced foreclosure had also lost a job in the prior 2 years. At the macrolevel, Pilkauskas et al. (2012) found that a 1% increase in unemployment rate was associated with 13% and 16% increases in the probability of a rent/mortgage/utility bill payment delay and having utilities cut off, respectively. Recent evidence from the global financial crisis in the late 2000s suggests that Black and Hispanic households are also disproportionately vulnerable to external shocks. In an analysis of national SIPP data from 2009 to 2011, Zhang and Lerman (2019) found that in the years immediately following the Great Recession Black households were 16.5%p and Hispanic households were 9.5%p more likely to be behind on housing, utility, or other bills than white households. Black-dominant neighborhoods experienced steep property value declines during that economic crisis and relatively slow recovery compared to white-dominant neighborhoods (Raymond et al. 2016). Recent data identified similar patterns of racial and ethnic hardship during the COVID-19 pandemic. Based on a nationally representative sample, Lopez et al. (2020) found disproportionate impacts of the pandemic on minority groups in the first months with respect to employment, rainy day funds, and monthly bill payments. Choi and Pang (2020) used Census Pulse data to estimate delinquency rates across racial and ethnic groups and found, as of July 2020, Black and Hispanic homeowners were more than twice as likely to experience mortgage delinquency than white homeowners. Media reports also indicated that minority groups were more at risk for utility shutoffs during the pandemic (Duster, 2020; Kowalski, 2020; Tomich et al. 2020). Linking Racial and Ethnic Disparities in Liquid Assets, Income, and Employment to Disparities in Housing Hardships Racial and ethnic disparities in housing hardship reflect racial and ethnic disparities in other areas, including access to liquid assets2 and stable, high-quality jobs. Due to the legacies of codified violence and discrimination—including slavery, Jim Crow laws, and more recently redlining, racial steering, and racially biased mortgage and hiring practices—Blacks and Hispanics often live in racially and ethnically segregated neighborhoods with poor housing stock and a lack of access to quality education and job opportunities. As a result, members of these minority groups have been systematically limited from building wealth and transferring money and other assets across generations (Pattillo, 2013; Rich et al. 1993; Rothstein, 2017; Sharkey, 2013). Bayer et al. (2016) suggested that households with lower levels of savings and wealth may face an increased risk of mortgage delinquency and foreclosure during economic shocks, compounding their financial hardships. Likewise, Ren (2020) found that much of the widening in the Black–white homeownership gap during the foreclosure crisis could be explained by accounting for racial differences in liquid wealth. In addition to having less wealth, Black and Hispanic households are overrepresented in low-wage, less secure, and precarious jobs (Grodsky & Pager, 2001; Huffman & Cohen, 2004; McCall, 2001; Pager & Shepherd, 2008), leaving these populations continually vulnerable to economic instability. Bayer et al. (2016) found that Black and Hispanic homeowners were disproportionately exposed to surging unemployment rates, which made them more vulnerable to foreclosure. This finding is consistent with other researchers’ conclusions, who found that Black employees are frequently the “first fired” during economic downturns (Brown & Pagán, 1998; Couch & Fairlie, 2010; Freeman et al. 1973). Taken together, this research indicates that a large-scale economic shock like the COVID-19 pandemic can have far-reaching economic consequences for racial and ethnic minority households that can lead to further disparities in housing hardships within these groups. Theoretical Framework We assumed the current COVID-19 pandemic was an exogenous financial shock that has led to massive housing-related hardships among US households. Building upon the evidence of previous empirical research, we posited four hypotheses regarding the pandemic’s disproportionate impacts on housing-related hardships across racial and ethnic groups (Fig. 1): [Hypothesis 1] The disproportionate housing-related hardship experiences across racial/ethnic groups vary over the course of the pandemic. [Hypothesis 2] The disproportionate impacts of the pandemic across racial/ethnic groups are stronger among lower-income respondents. [Hypothesis 3] Liquid assets mediate the disproportionate impacts of the pandemic across racial/ethnic groups. [Hypothesis 4] Job and income losses mediate the disproportionate impacts of the pandemic across racial/ethnic groups.Fig. 1 Theoretical frameworks Data Data for this study come from the longitudinal Socioeconomic Impacts of COVID-19 Survey (SEICS), administered by the Social Policy Institute at Washington University in St. Louis (Roll et al. 2021). The five-wave longitudinal survey was distributed by a large online-panel provider at quarterly intervals between April 2020 and June 2021. Figure 2 illustrates the administration periods for each survey wave, along with reference information on COVID infections, vaccines, and key social and political events that occurred during the administration period (e.g., stimulus check disbursal).3 More than 5000 respondents from all 50 US states and Washington, D.C. completed each wave of the survey. The survey sample was developed using quota sampling techniques to ensure that the sample represented US demographic characteristics with respect to age, gender, race/ethnicity, and income.4, 5 The survey response rate was 9.6%, with 71,800 adults entering the survey. Of these respondents, 46,842 were excluded because they either failed to meet quota requirements to ensure national representativeness on the established sampling criteria, or failed quality checks embedded in the survey. After these exclusions, 24,958 completed surveys composed the sample. Additional checks on the characteristics of this sample revealed that they approximated the US population in terms of the state of residence, homeownership, and other key demographic and financial criteria. For the purposes of this study, respondents who did not provide a response to any item used in this analysis were excluded using listwise deletion. The final analytical sample comprised 22,939 white, Black, Asian/other (non-Hispanic), and Hispanic respondents.Fig. 2 Administration periods for each survey wave Methods Measures Varying definitions of housing hardship exist. Some researchers use the term to focus on a family’s lack of their own place to live (Neckerman et al. 2016) or issues with the quality of the physical dwelling (e.g., pests, leaks, broken windows, overcrowding; Eamon & Wu, 2011), whereas some use the term to denote problems in making housing-related payments (Heflin, 2017), and others use it to refer to a combination of these problems (Caswell & Zuckerman, 2018; Long, 2003). Often, conceptualizations of housing hardship that focus on housing-related payments, such as missing a rent/mortgage payment or late/skipped payment of a utility bill, are examined as one of the areas within the broader concept of material hardship (Despard et al. 2018; Gjertson, 2016; Heflin, 2016; McKernan et al. 2009). In this paper, we measure housing-related hardships (e.g., eviction and foreclosure, mortgage and rent delinquency, and utility bill payments) during the pandemic using the following survey questions:[Eviction/foreclosure] In the past 3 months, was anyone in your household forced to move by a landlord or bank when you did not want to? [Rent/mortgage delinquency] In the past 3 months, have you or someone in your household not paid the full amount of the rent or mortgage because you could not afford it? [Utility payment delay] In the past 3 months, have you or someone in your household skipped paying a bill or paid a bill late due to not having enough money? To examine the relationship between race/ethnicity, income, and housing hardships, the survey asked respondents to indicate if they identified as white/Caucasian, Black/African American, Asian, Native American/Pacific Islander, or some other race. Respondents could select multiple options. The survey also asked whether a respondent considered themself Hispanic or Latino/a/x. Of the two survey questions, the one regarding Hispanic origin was dominant over the race question—those who consider themselves Hispanic or Latino/a/x were coded as Hispanic or Latino/a/x regardless of their racial identity. To measure income, the survey asked respondents to report their total pretax household income from all sources in 2019. This question allowed us to identify households’ income prior to any income fluctuations caused by the COVID-19 pandemic. Because the cost of living varies across geography and family size, we constructed our measure of income as a function of households’ total income in 2019, household size, and the US Department of Housing and Urban Development’s (2020) measure of area median income (AMI) at the county level. Therefore, income indicates the proportion of AMI adjusted for household size. For ease of reporting, we measured the marginal effects at 30%, 50%, 80%, 120%, and 170% of AMI to represent extremely low-income, very low-income, low-income, moderate-income, middle-income, and high-income thresholds, respectively.6 To construct the liquid asset amount indicator, we used self-reported asset measures from the survey. Specifically, we defined liquid assets as the sum of assets held in checking accounts (or money market accounts), savings accounts, and cash (or pre-paid cards); our liquid asset measure is, therefore, the sum of assets held in these forms. We asked respondents to report the value of their liquid assets currently and the value of these assets 3 months ago. We used the retrospective liquid asset measure (e.g., liquid assets 3 months prior) to construct our liquid asset variable. To address extreme outliers, we winsorized asset amounts at the upper 99th percentile. To construct our measure of employment shocks during the pandemic, we used three survey questions: (a) “Have you lost a job or lost income as a result of the COVID-19 pandemic?”; (b) “Has your spouse lost a job or lost income as a result of the COVID-19 pandemic?”; and (c) “Has anyone else in your household lost a job or lost income as a result of the COVID-19 pandemic?”. If a respondent answered “yes” to any of these questions, they were considered to have experienced a household-level employment shock. In addition to the measures of race/ethnicity, income, liquid asset amount, and employment shocks during the pandemic, our empirical models accounted for housing status (whether respondents own their home with or without a mortgage or pay rent) and demographic characteristics (gender, age, marital status, educational attainment, and the number of dependents). We also include two policy variables as controls—receipt of the Economic Impact Payments offered through the CARES Act and the presence of state-level eviction moratoria.7, 8 Empirical Model Design Disproportionate Housing Hardships During the Pandemic The housing-related hardship variables, including eviction/foreclosure risk, rent/mortgage delinquency, utility bill payment delay, and any of the three hardship experiences, are binary. Thus, we employed a set of logistic regression models as follows:1 lnPrYi=1\|X1-PrYi=1\|X=β0+β1xirace+β2xiinc+β3xirace∗xiinc+Xiγ+λDi+ηWi where the probability of a given housing hardship for an individual i,Pr(Yi=1), was a function of race/ethnicity,xirace, income,xiinc, and the interaction of race/ethnicity and income indicators as well as a set of covariates including demographic characteristics (gender, age, marital status, number of dependents), socioeconomic attributes (educational attainment, homeownership), and the receipt of the first stimulus check. To account for the geographic heterogeneity of the economic impacts of the pandemic, each empirical model also considered geographic (division, Di) and time (wave,Wi) fixed effects, as well as robust standard errors. For simplicity, we report the predicted housing hardships of each combination of race/ethnicity and income.9 Mediation Effects of Liquid Assets and Employment Shocks Building upon evidence from previous empirical studies, we assumed that the pandemic’s disproportionate impacts on housing hardships across racial/ethnic groups are at least partly associated with varying liquid assets of these groups. To measure the mediation impacts of liquid assets and employment shocks, we employ Buis’ (2010) model to estimate direct and indirect effects in a logit model. Using the model, we decompose the total effects of racial/ethnic attributes on housing hardships into direct (i.e., race/ethnicity to housing hardships) and indirect (i.e., race/ethnicity to liquid asset amount/employment shocks to housing hardships) effects as follows10:A. Mediation effect of liquid asset amount Oddsblack,asset\|blackOddswhite,asset\|white↓total=Oddswhite,asset\|blackOddswhite,asset\|white↓indirect×Oddsblack,asset\|blackOddswhite,asset\|black↓direct Oddshisp,asset\|hispOddswhite,asset\|white↓total=Oddswhite,asset\|hispOddswhite,asset\|white↓indirect×Oddshisp,asset\|hispOddswhite,asset\|hisp↓direct Oddshisp,asset\|asianOddswhite,asset\|white↓total=Oddswhite,asset\|asianOddswhite,asset\|white↓indirect×Oddshisp,asset\|asianOddswhite,asset\|asian↓direct B. Mediation effect of employment shocks Oddsblack,unemp\|blackOddswhite,unemp\|white↓total=Oddswhite,unemp\|blackOddswhite,unemp\|white↓indirect×Oddsblack,unemp\|blackOddswhite,unemp\|black↓direct Oddshisp,unemp\|hispOddswhite,unemp\|white↓total=Oddswhite,unemp\|hispOddswhite,unemp\|white↓indirect×Oddshisp,unemp\|hispOddswhite,unemp\|hisp↓direct Oddshisp,unemp\|asianOddswhite,unemp\|white↓total=Oddswhite,unemp\|asianOddswhite,unemp\|white↓indirect×Oddshisp,unemp\|asianOddswhite,unemp\|asian↓direct The indirect effect estimates the relative odds of predicted housing hardship risk of a given minority group over the counterfactual housing hardship risk of that minority group if it had the asset (or employment shock) distribution of white respondents. For example, in the first equation in model A, the denominator, Oddswhite,asset\|white, is the odds of having experienced a housing hardship for white respondents. The numerator, Oddswhite,asset\|black, is the counterfactual odds of a housing hardship experience for white respondents if they had the same distribution of assets as the group. The relative odds ratio, Oddsblack,asset\|blackOddswhite,asset\|white, represents the indirect effect of assets on the disparity in housing hardship among Black respondents compared to white respondents; if the odds ratio is greater than 1, the asset amounts positively mediate the association between race/ethnicity and housing hardship. To compute standard errors for the decomposed effects, we used a bootstrapping procedure with 999 iterations. In addition to the decomposed effects, we also estimated the size of the indirect effect relative to the total effect. All the mediation models in this study also controlled for all the covariates in the logistic models above, as well as the AMI and family size adjusted annual household income in 2019. To enhance the external validity of the analysis,11 we weighted our analytic sample with respect to age, gender, race and ethnicity, marital status, number of dependents, educational attainment, income, and geography (division), based on the Census Bureau’s American Community Survey (ACS) 2018 Public Use Microdata Sample (PUMS). The data analysis in this study was conducted using Stata (Version 16; StataCorp, 2019), and we used a threshold of p < 0.05 to assess statistical significance. Empirical Findings Descriptive Analysis Table 1 presents summary statistics on model variables for the entire sample as well as by racial/ethnic group. Overall, these findings indicate that the pandemic worsened housing problems in the USA. In August 2020, during the nationwide lockdown, 6.3% of respondents were forced to move by a bank or a landlord, 11.1% were having difficulty keeping up with their mortgage or rent payments, and 14.4% skipped paying a utility bill or paid a bill late in the prior 3 months. These measures of housing instability peaked in the fall of 2020, when 7.3%, 11.8%, and 15.2% of the respondents experienced eviction/foreclosure risk, rent/mortgage payment delay, and utility bill payment delay, respectively. Though these aspects of housing instability decreased in 2021, the levels for each were still higher than before the pandemic.12Table 1 Summary statistics of housing hardship experiences in the past 3 months, over time, across race/ethnicity All By race/ethnicity White Black Hispanic Asian/other (1) (2) (3) (4) (5) Any housing hardships Wave 1 14.8% 13.2% 20.5% 17.9% 11.8% Wave 2 19.2% 15.6% 29.5% 26.9% 14.2% Wave 3 20.7% 21.6% 21.2% 23.6% 8.2% Wave 4 16.4% 14.5% 22.1% 22.0% 11.1% Wave 5 15.0% 12.2% 20.8% 23.9% 11.4% Eviction/foreclosure Wave 1 3.1% 3.2% 1.8% 4.2% 1.5% Wave 2 6.3% 5.2% 8.7% 10.3% 2.8% Wave 3 7.3% 7.9% 6.1% 8.7% 1.5% Wave 4 4.7% 4.3% 5.3% 6.5% 3.3% Wave 5 4.2% 3.5% 5.4% 7.5% 2.1% Rent/mortgage delinquency Wave 1 7.3% 6.6% 9.1% 9.4% 5.9% Wave 2 11.1% 8.6% 16.6% 17.3% 7.9% Wave 3 11.8% 12.0% 12.8% 14.6% 3.3% Wave 4 8.7% 8.0% 10.1% 11.7% 5.6% Wave 5 9.1% 6.9% 12.4% 16.4% 6.3% Utility bill payment delay Wave 1 11.5% 10.1% 16.7% 13.9% 8.9% Wave 2 14.4% 12.1% 21.3% 19.7% 10.7% Wave 3 15.2% 16.2% 15.2% 16.6% 5.6% Wave 4 12.4% 11.4% 16.3% 15.4% 8.2% Wave 5 12.0% 9.8% 15.8% 20.1% 8.4% Liquid assets Wave 1 $24,862.2 $27,596.4 $14,371.0 $19,631.5 $31,005.9 Wave 2 $24,282.3 $27,841.4 $12,120.3 $17,863.4 $29,865.7 Wave 3 $24,597.1 $25,995.0 $15,312.6 $22,022.2 $32,570.0 Wave 4 $25,856.2 $29,124.3 $13,096.1 $20,784.5 $30,697.1 Wave 5 $27,118.7 $30,613.2 $15,047.4 $21,170.1 $29,468.0 Job/income shock Wave 1 28.6% 29.9% 22.6% 29.7% 26.0% Wave 2 30.3% 28.4% 30.2% 38.2% 29.1% Wave 3 29.7% 29.5% 25.1% 35.3% 26.6% Wave 4 27.1% 24.2% 29.6% 36.5% 24.9% Wave 5 24.5% 20.9% 23.5% 35.8% 29.2% Housing-related hardships during the pandemic varied somewhat depending upon racial/ethnic identity and income. Table 1 indicates that families in minority groups were more vulnerable to housing-related hardships than white families during the earlier stages of the pandemic. In August 2020, Black respondents were 1.7 times as likely to be forced to move, 1.9 times as likely to be delinquent on housing payments, and 1.8 times as likely to be delinquent on utility bill payments compared with white respondents. In the same period, Hispanic respondents were 2.0 times as likely to have had an eviction/foreclosure risk, 2.0 times as likely to have missed a housing payment, and 1.6 times as likely to have missed paying a utility bill. In contrast, the Asian/other group experienced fewer housing hardships than white respondents throughout the pandemic. As with housing-related hardships, more Black and Hispanic respondents reported experiencing financial instability during the pandemic than those in the white and Asian/other groups. Black and Hispanic groups held much smaller amounts of liquid assets than white respondents. In particular, Black respondents reported the lowest liquid assets; the average liquid asset amount of Black respondents was almost half that of white respondents throughout the pandemic. Also, Hispanic families exhibited higher levels of job/income shocks than the other three groups. Throughout the pandemic, more than a third of the Hispanic respondents reported a loss of income/job due to COVID-19. Explanatory Analysis Disproportionate Pandemic Impacts on Household Hardship over Time To address potential bias due to heterogeneity in the cohort, we employed a set of logistic regression models to control for demographic characteristics and geography at the Census division level. Figure 3 shows that housing instability measures varied over the course of the pandemic. In the early stages of the pandemic, white respondents exhibited the lowest levels of housing instability compared to respondents of color. In the first 3 months of the pandemic (Wave 1, March to May 2020), white respondents were significantly less likely to experience any housing-related hardships than Black respondents (p < 0.05).Fig. 3 Housing instability measures In the next survey wave (wave 2, June to August 2020), the housing hardship gap between white and Black/Hispanic respondents widened and remained significant; 15.3% of white respondents experienced housing-related hardship between June and August 2020, 22.6% of Black and 21.9% of Hispanic respondents experienced housing instability during the same period (p < 0.01). After June and August of 2020, housing inequality declined. Some of this decline was due to an increase in housing stability for Black and Hispanic households; however, white respondents increasingly experienced housing hardships, and that increase was larger than the relative decrease among Black and Hispanic households. From wave 2 to wave 3 (November to December 2020), the proportion of white respondents who experienced housing hardships increased by 4.0%p. In contrast, housing hardship decreased in Black and Asian/other respondents by 2.3 and 2.8%p, respectively. During the same period, the housing instability of Hispanic respondents slightly increased by 0.6%p. Differences in housing hardships widened again in wave 4 (February to March 2021) as the proportion of white respondents with housing instability rapidly decreased by 4.3%p. However, the housing instability gap narrowed in the final survey wave (wave 5, May 2021), as the proportions of Black and Hispanic respondents with housing hardships decreased by 4.9%p each. In sum, we found that Black and Hispanic respondents were hit faster, harder, and longer than white respondents during the pandemic. White respondents experienced lower rates of housing hardship than Black and Hispanic respondents, their housing hardships increased later in the pandemic, and they recovered more quickly after their hardship rates peaked. Notably, Asians exhibited a steady decline in hardships throughout the study period after their peak in the first wave of the study. Disproportionate Pandemic Impacts on Household Hardship across Income Cohorts Next, we explore how housing inequality varied across income cohorts. Here, we focus on the wave 2 survey administered in August 2020 when housing inequality peaked due to the nationwide lockdown. Figure 4 displays significant disparities between white and Black/Hispanic families in low- and moderate-income cohorts. Families with 30 to 120% AMI exhibited significant disparities in housing hardship between the two racial/ethnic groups. Compared to white respondents, Black and Hispanic respondents at 80% AMI were 1.5 and 1.6 times more likely to have experienced housing hardship, respectively (p < 0.01). Specifically, Black respondents were 2.3 times more likely to be forced to move (4.4% vs. 10.2%, p < 0.01), 1.9 times more likely to fall behind on housing payments (8.6% vs. 16.6%, p < 0.01), and 1.2 times more likely to miss utility bill payments (12.8% vs. 15.7%, not significant). Similarly, Hispanic respondents were 1.8 times more likely to be forced to move (4.4% vs. 8.1%, p < 0.01), 1.9 more likely to miss a rent/mortgage payment (8.6% vs. 16.3%, p < 0.001), and 1.3 more likely to miss a utility bill payment (12.8% vs. 15.7%, not significant). In extremely high- and low-income groups, the risks of these three housing hardship indicators were not significantly different at the 0.05 level across the racial/ethnic groups after controlling for covariates. There was no significant disparity in housing instability between white and Asian/other groups overall and across income cohorts.Fig. 4 Significant disparities between white and Black/Hispanic families in low- and moderate-income cohorts It is worth noting that we observed significant positive associations between housing instability and Economic Impact Payments receipt at the individual level (odds ratio = 1.607; p < 0.001). Though this positive association is contrary to our expectations, it might be due to potential selection bias in who qualified to receive these payments (Roll and Grinstein-Weiss 2020).13 The presence of a state-level eviction moratorium was not significantly associated with housing hardship experiences. Mediation Effects of Liquid Assets and Employment Shocks Race and ethnicity alone do not determine housing hardship. Rather, we hypothesized that preexisting disparities in liquid asset amounts prior to the pandemic and employment shocks during the pandemic could be two key pathways in the relationship between race/ethnicity and housing hardship risks. Panel A in Table 2 shows the estimated indirect mediation effects of liquid asset amount on the association between race/ethnicity and housing hardships. Notably, the indirect effects of race/ethnicity through the liquid asset pathway were highly significant in the models comparing white and Black respondents. The indirect effects via liquid assets explain 32%, 24.3%, and 34.9% of the estimated disproportionate pandemic impacts on eviction risk, mortgage/rent delinquency, and utility bill payment delay, respectively (p < 0.001). On the other hand, the indirect effects are somewhat smaller and less significant when comparing white and Hispanic respondents. The indirect effects represent 24.0% (p < 0.05.), 17.0% (p < 0.01), and 28.3% (p < 0.001) of the total effects on eviction risk, mortgage/rent delinquency, and utility bill payment delinquency, respectively.Table 2 Mediation effects of liquid asset amounts and COVID-19-related job/income loss, wave 2 Black to white Hispanic to white Asian/other to white Eviction Rent/mortgage Utility Eviction Rent/mortgage Utility Eviction Rent/mortgage Utility (1) (2) (3) (4) (5) (6) (7) (8) (9) Panel A: liquid asset amount Total effectodds ratio 2.346* 2.252* 1.725* 2.125* 2.121 1.541* 0.856 1.265 1.222 (0.390) (0.320) (0.287) (0.470) (0.550) (0.181) (0.283) (0.384) (0.235) Indirect effectodds ratio 1.314* 1.218* 1.210* 1.198* 1.136* 1.130* 0.946 0.963 0.965 (0.085) (0.056) (0.033) (0.054) (0.033) (0.026) (0.041) (0.032) (0.028) Direct effectodds ratio 1.786* 1.849* 1.426* 1.774* 1.867* 1.363 0.905 1.313 1.266 (0.281) (0.290) (0.232) (0.411) (0.481) (0.138) (0.300) (0.373) (0.241) Indirect effect/total effect 32.0%* 24.3%* 34.9%* 24.0%* 17.0%* 28.3%* 36.0% − 16.1% − 17.9% Panel B: job/income loss Total effectodds ratio 2.194* 2.104* 1.575 2.169 2.068 1.466 0.811 1.194 1.154 (0.373) (0.342) (0.259) (0.625) (0.554) (0.181) (0.260) (0.392) (0.241) Indirect effectodds ratio 1.040 1.031 1.029 1.218* 1.171* 1.161* 1.016 1.012 1.012 (0.034) (0.034) (0.033) (0.052) (0.040) (0.045) (0.044) (0.033) (0.034) Direct effectodds ratio 2.109* 2.040* 1.530** 1.780* 1.765* 1.262* 0.798 1.180 1.140 (0.368) (0.335) (0.242) (0.509) (0.482) (0.149) (0.258) (0.397) (0.234) Indirect effect/total effect 5.0% 4.2% 6.4% 25.5%* 21.8% 39.1%** − 7.5% 6.9% 8.0% Gender, marital status, number of dependents, educational attainment, homeownership, pre-pandemic annual income (2019) and division and survey wave fixed effects are controlled. Exponentiated coefficients for total/indirect/direct effects. Bootstrap standard errors in parentheses p < 0.05, p < 0.01, p < 0.001 Panel B in Table 2 presents the estimated indirect mediation effects of job and/or income losses during the pandemic. The indirect effects of race/ethnicity through the job/income loss pathway were not significant in the models comparing white and Black respondents. Across the three housing hardships, the indirect effect was less than 10%. However, the mediation effects of job and income loss on white and Hispanic disparities for the three measured housing hardships were highly and positively significant in eviction risk (25.5%, p < 0.05) and utility bill payment delay (39.1%, p < 0.01). Discussion The results of our study indicate that the pandemic disproportionately affected the housing stability of minority groups. Although the entire US population faced increased housing risk, Black and Hispanic populations bore these risks disproportionately, especially during the early stages of the pandemic. Though these groups were more vulnerable to the pandemic’s impacts on housing hardships, the temporal dynamics of housing instability varied. White respondents experienced a slight lag when the housing-related shock of the pandemic hit, but they then rapidly recovered. Those in the Asian/other group were immediately affected by the pandemic shock but recovered quickly. Black respondents were immediately affected by the pandemic shock and recovered very slowly. Hispanic respondents’ exhibited both a slow response to the pandemic shock and a slow recovery. These observed inequalities across racial/ethnic groups were particularly prevalent for those with low or moderate incomes. This finding might be due to various associations between income and housing instability across racial/ethnic groups. For white families in our sample, housing instability during the pandemic was inversely correlated with their income before the pandemic. The probability of experiencing housing-related hardship dropped by more than half for respondents with income measures between 30 and 120% AMI. On the other hand, housing instability levels were relatively stable across income levels for Black and Hispanic families. The risk of housing hardship decreased by 30% in households with income between 30 and 120% AMI. In other words, income was a strong predictor of housing instability for white people but not for people of color. Our findings do not mean that the pandemic created disparities in housing instability among minority groups. As our literature review clearly demonstrates, racial disparities in the housing market existed well before the pandemic. Instead, the racial and income gaps between wave 1 and wave 2 of our study indicate that the pandemic exacerbated these disparities. Our mediation models suggest that the mechanisms contributing to housing instability during the pandemic varied between Black and Hispanic groups. Disparities in prepandemic liquid assets explain the relatively high housing risks among Black and Hispanic families. The partial mediation effect of prepandemic liquid assets on the disproportionate housing hardships faced by Black and Hispanic families implies that the current disparities are to some extent a function of preexisting economic inequities. Over decades, wage disparities, homeownership disparities, unequal access to affordable financial products and services, and myriad other factors have left Black families less able to build up the type of emergency savings buffers that are the lynchpin of economic security. The disparate exposure to housing hardships during large-scale economic crises like the COVID-19 pandemic is just one result of this intergenerational economic inequality. In addition to preexisting disparities, COVID-19-related employment/income shocks explain why Hispanic families were more likely to experience some housing-related hardships during the pandemic. Recent evidence shows that Black and Hispanic workers were more vulnerable to employment shocks in the early stages of the pandemic as they are disproportionately concentrated in so-called essential industries that were the first and hardest hit by the economic downturn (Klein and Shiro 2020; Williams 2020). Our results corroborate these findings and show that Hispanic families experienced the highest level of job/income loss among the three racial/ethnic groups (white—26.7%, Black—26.3%, Hispanic—35.0%, Asian/other—27.1%). Though we do not examine them in this study, historical and contemporary forms of racial/ethnic discrimination in the labor market, as well as discrimination by immigration status, may also contribute to the housing hardship disparities we observed among Hispanic households. In particular, Hispanic families may have been in vulnerable housing situations prior to the pandemic, which could be a function of discriminatory mortgage lending practices, characteristics specific to the communities where Hispanic families live, or other individual or societal factors. Lack of access to government services may be another reason for their disproportionate housing hardship experiences during the pandemic. Poor access to social services in the Hispanic population has been attributed to the immigration status of the population, a lack of adequate information due to language barriers, and discrimination at both institutional and individual levels (Einstein and Glick 2017). We also found that those in the Asian/other group had unique experiences of housing stability during the pandemic. Except for the very early months of the pandemic, the Asian/other group displayed more housing stability than other minority groups and exhibited a pattern similar to that of the white group. Although they exhibited the highest level of housing instability immediately after the outbreak of the pandemic, that level steadily declined. This uniqueness of the Asian/other population, as well as varying mediation effects of unemployment on the Black and the Hispanic groups, calls for more sophisticated approaches in exploring housing inequality in minority groups. Last, we observed that racial and ethnic disparities in housing instability also varied by hardship type (i.e., eviction/foreclosure, rent/mortgage delinquency, and utility payment delay). In the summer of 2020, the disparity in utility payment delay across the four groups was not statistically significant. However, Black and Hispanic respondents exhibited significantly higher risks of eviction/foreclosure and rent/mortgage delinquency than white respondents during the same time frame. This difference might be simply related to families’ priorities once hardship hit a family. More plausibly, these disparities could represent the systemic discriminatory nature of the US housing market and its related policies. Once hardship hits families, housing stability is one of the first things that families of color lose due to this discrimination (Rothstein 2017). Limitations Although our study offers novel contributions to the field, it is not without limitations. First, this study was limited by the absence of a prepandemic study wave. As our panel survey was conducted after the COVID-19 pandemic began, we were limited in asking retrospective questions about prepandemic experiences.14 In this regard, the results in wave 1 might partly represent the prepandemic housing disparities as well as at the beginning of the pandemic. However, the purpose of this study was not to examine the extent to which housing hardships occurred because of the pandemic but rather to explore how disparities evolved over the course of the pandemic and how these disparities were mediated by households’ financial endowment and financial shocks due to the pandemic. A second limitation of the study stems from relying on online surveys. The fact that our survey was conducted online introduces potential sampling bias as we did not include those without access to a stable Internet connection in the study sample. Survey responses may also reflect some degree of measurement error; for example, our financial variables may be subject to errors stemming from an inability (or unwillingness) to accurately report income or assets. We minimize this error in several ways: (a) assuring respondents that their answers were confidential; (b) eliciting a commitment from respondents that they would provide their best answers to survey questions and eliminating anyone from the survey who did not answer this question affirmatively; (c) paying an additional fee to our survey panel provider to engage in a data cleaning procedure that identified and eliminated respondents with suspicious response patterns and other indicators of low-quality responses (e.g., improbable completion times); (d) asking respondents about the value of their checking accounts, savings accounts, and cash on hand separately to help them calculate the value of these accounts more easily (rather than having to aggregate them mentally); and (e) excluding suspicious asset amount reports and winsorizing the asset amount at the 99th percentile to minimize the impact of extreme asset amounts. A final limitation concerns the level of detail captured in the survey. Our measure of employment shock due to the pandemic only captures whether households lost any job or income as a result of the pandemic, rather than capturing how many jobs or the percent of income they lost, and so on. As such, some measures used in this study are relatively coarse and may not capture the full details of households’ economic experiences during the pandemic. Conclusion In addition to highlighting potentially long-lasting implications for inequality in the housing market during the COVID-19 pandemic, the findings from this study make unique contributions to the current literature. First, this study is the first attempt to integrate the racial and ethnic disparities in housing instability and financial/employment attributes during an exogenous shock like the pandemic. Although previous empirical studies explored individual sets of associations between the same variables we examined—exogenous shocks and housing hardships (e.g., Niedt and Martin 2013; Pilkauskas et al. 2012), race/ethnicity and housing hardships (e.g., Heflin 2017; Medina et al. 2020), or race/ethnicity and wealth (e.g., Bayer et al. 2016)—they have not considered all of these components together due to the lack of comprehensive data. Our comprehensive survey allowed us to explore the mediation effects of liquid assets and income shocks on racial disparities in housing issues at the family level. In particular, the observed mediation effect of liquid assets calls for proactive and fundamental remedies beyond the Economic Impact Payments offered through the CARES Act and eviction moratorium policies. Going forward, identifying and addressing the causes of liquid asset gaps across racial and ethnic groups will be essential to helping these families better withstand future economic shocks. Secondly, our empirical study of the COVID-19 pandemic implies that disparities in housing instability in response to shocks can occur nationwide. Previous external shocks to the housing market, such as Hurricane Katrina in 2005 and the mortgage crisis of the late 2000s, tended to be concentrated in certain geographic areas. Thus, much of the research on prior shocks focused on specific local housing markets and had external validity limitations. The COVID-19 pandemic differs from previous shocks in that it was global, and its onset was almost simultaneous regardless of geography. By using a nationally representative survey, our study offers a broader understanding of housing market dynamics and racial issues. Our findings indicate that large minority groups in the USA are not only exposed to all the hardships that accompany housing instability, but also have likely faced the disproportionately high risks of COVID-19 infection that accompany the inability to effectively shelter in place. Understanding the particular needs of these groups and taking positive steps to address both the disparate burdens placed on them during the pandemic and the prepandemic inequities that led to these disparities, will be essential to forming effective pandemic responses both now and in the future. Though always important, stable and adequate housing is even more critical in the midst of a pandemic to maintain public health. Stay-at-home orders have been a core component of the public health response to COVID-19 in the USA. Without housing, individuals and families cannot shelter in place to prevent the spread of disease (Ellen et al. 2020). An increase in residential evictions increases the demand for services at homeless shelters, which may become overcrowded, thereby facilitating viral spread. Housing hardship may also cause families to double up, increasing overcrowding in residential units and making all residents more vulnerable to infection. Early research supports these theories; the end of eviction moratoria and the corresponding increase in evictions were associated with further spread of COVID-19 (Jowers et al. 2021; Pan et al. 2020). In addition, housing hardship, even without culminating in eviction, may operate as a form of chronic stress and weaken immune system responses (Jelleyman and Spencer 2008; Ross and Squires 2011). Understanding and combating housing hardship among vulnerable populations is therefore essential to a sound public health response. Appendix 1 Table 3 Table 3 Summary statistics of explanatory variables in use All By race/ethnicity White Black Hispanic Asian/other (1) (2) (3) (4) (5) Race/ethnicity White 61.4% 100% - - - Black 12.5% - 100% - Hispanic 16.8% - - 100% - Asian/other 9.3% - - - 100% Incomea Very low income, AMI = [0, 50) 25.4% 22.7% 34.5% 30.6% 21.1% Low income, AMI = [50, 80) 18.7% 18.6% 20.5% 18.8% 17.0% Moderate income, AMI = [80, 120) 20.5% 20.2% 20.1% 21.7% 21.1% Middle income, AMI = [120, 170) 16.0% 16.9% 12.7% 14.8% 17.5% High income, AMI = [170,.) 19.3% 21.6% 12.2% 14.1% 23.3% Gender Female 50.3% 47.4% 59.5% 54.1% 50.6% Age 18–25 9.8% 11.1% 9.9% 8.2% 4.2% 25–34 18.9% 20.9% 14.6% 18.0% 12.9% 35–44 16.5% 14.0% 18.9% 23.0% 18.3% 45–54 17.9% 17.3% 19.4% 18.4% 19.5% 55 + 36.8% 36.7% 37.2% 32.4% 45.2% Marital status Married 53.1% 55.4% 34.6% 55.6% 58.3% Single, never married 32.9% 31.5% 44.4% 31.1% 29.7% Single, separated, divorced, widowed 14.0% 13.1% 21.0% 13.3% 12.0% Dependents No dependents 72.5% 73.6% 73.6% 65.3% 77.0% 1 13.4% 12.4% 15.5% 16.5% 11.8% 2 10.6% 10.8% 7.8% 12.7% 9.2% 3 3.5% 3.3% 3.1% 5.5% 2.0% Educational attainment High school/GED or lower 13.9% 13.7% 17.1% 16.2% 6.8% Some college/certificate/associate’s degree 29.7% 28.4% 36.3% 34.4% 20.3% Bachelor’s degree 30.5% 30.6% 26.4% 29.6% 36.8% Graduate or professional degree 26.0% 27.3% 20.2% 19.8% 36.1% Homeownership Own home, with mortgage 37.7% 37.2% 36.1% 39.5% 40.4% Own home, without mortgage 30.7% 33.4% 21.1% 25.7% 34.8% Rent home 26.5% 24.3% 37.0% 29.5% 21.7% Neither rent nor own home 5.1% 5.2% 5.8% 5.3% 3.1% First stimulus check receipt Received the first stimulus check 66.5% 68.0% 66.4% 65.9% 57.7% Eviction moratorium (state level) Enacted/recommended 71.4% 70.8% 61.5% 76.7% 78.9% Reference groups are underlined aArea median income (AMI) was estimated in 2019 at the country level; in the regression analysis, we treated the adjusted income variable as continuous (a household’s proportion of AMI adjusting for household size) bIn the regression analysis, liquid asset amounts are winsorized at upper 99th percentile Appendix 2 Full logit model results Table 4 and Table 5 Table 4 Full logit model results, over time Any hardships Eviction Rent/mortgage Utility (1) (2) (3) (4) Black 1.864* 1.342 1.509+ 1.715 (0.344) (0.525) (0.356) (0.332) Hispanic 1.376+ 2.302 1.576* 1.227 (0.242) (0.691) (0.346) (0.235) Asian/other 1.895* 0.692 1.967+ 1.934* (0.512) (0.335) (0.759) (0.613) Wave 2 1.129 1.752** 1.327* 1.093 (0.120) (0.331) (0.183) (0.125) Wave 3 1.578* 2.203* 1.665* 1.428 (0.164) (0.415) (0.229) (0.160) Wave 4 1.106 1.324 1.217 1.103 (0.122) (0.268) (0.178) (0.132) Wave 5 0.994 1.253 1.165 1.008 (0.119) (0.273) (0.185) (0.130) Wave 2 × Black 0.951 2.138+ 1.401 0.721 (0.224) (0.960) (0.404) (0.178) Wave 2 × Hispanic 1.225 0.868 1.195 1.063 (0.271) (0.305) (0.321) (0.257) Wave 2 × Asian/other 0.638 1.847 0.706 0.541 (0.213) (1.211) (0.325) (0.207) Wave 3 × Black 0.577* 1.216 0.992 0.515 (0.137) (0.574) (0.296) (0.131) Wave 3 × Hispanic 0.917 0.818 0.9 0.775 (0.205) (0.312) (0.251) (0.192) Wave 3 × Asian/other 0.357 1.378 0.209 0.302 (0.126) (0.951) (0.107) (0.130) Wave 4 × Black 0.857 1.649 0.962 0.763 (0.212) (0.842) (0.310) (0.206) Wave 4 × Hispanic 1.234 0.947 0.793 1.129 (0.292) (0.385) (0.235) (0.294) Wave 4 × Asian/other 0.377** 1.282 0.355* 0.336 (0.130) (0.780) (0.167) (0.134) Wave 5 × Black 0.645+ 0.91 0.958 0.595+ (0.164) (0.441) (0.306) (0.160) Wave 5 × Hispanic 0.937 0.543 1.034 1.095 (0.226) (0.209) (0.303) (0.285) Wave 5 × Asian/other 0.369 0.771 0.425+ 0.315** (0.126) (0.512) (0.199) (0.125) Gender: female 0.897* 0.500* 0.762* 0.972 (0.048) (0.047) (0.052) (0.057) Age = [25,35) 0.702* 0.520* 0.822+ 0.802* (0.061) (0.073) (0.090) (0.075) Age = [35,45) 0.461* 0.336* 0.614* 0.561* (0.043) (0.052) (0.070) (0.056) Age = [45,55) 0.367* 0.138* 0.413* 0.486* (0.036) (0.027) (0.053) (0.051) Age = 55 or more 0.128* 0.026* 0.153* 0.159* (0.015) (0.009) (0.025) (0.021) Marital status: single, never married 0.871+ 0.81 0.893 0.945 (0.065) (0.108) (0.081) (0.077) Marital status: separated/divorced/widowed 1.200* 0.707+ 1.127 1.186+ (0.103) (0.146) (0.130) (0.112) Education: some college/certificate/associate's degree 1.117+ 1.114 0.955 1.112 (0.072) (0.137) (0.080) (0.076) Education: bachelor’s degree 0.789** 1.322* 0.825+ 0.704* (0.060) (0.184) (0.082) (0.059) Education: graduate or professional degree 0.967 2.271* 1.013 0.882 (0.084) (0.358) (0.116) (0.085) # kid[s]: 1 1.889* 2.394* 2.016* 1.770* (0.139) (0.287) (0.176) (0.144) # kid[s]: 2 1.921* 2.777* 1.817* 1.816* (0.157) (0.369) (0.176) (0.161) # kid[s]: 3 + 2.042* 2.303* 2.111* 2.089* (0.234) (0.485) (0.293) (0.253) Income, AMI and family size adjusted 0.992* 0.994* 0.994* 0.992* (0.001) (0.001) (0.001) (0.001) Own home free and clear 0.927 1.218+ 0.739 0.849+ (0.069) (0.136) (0.071) (0.073) Pay rent 1.239 0.479*** 0.824* 1.288* (0.085) (0.063) (0.070) (0.097) Neither own home nor pay rent 0.523* 0.293* 0.315* 0.627* (0.066) (0.067) (0.056) (0.084) Public benefits 1.202 1.603*** 1.210* 1.319* (0.070) (0.169) (0.090) (0.085) Eviction moratorium: enacted 1.026 1.095 0.984 1.029 (0.075) (0.142) (0.092) (0.082) Constant 0.503* 0.099* 0.250* 0.318*** (0.102) (0.034) (0.066) (0.072) Observations 22,939 22,939 22,939 22,939 Pseudo R2 0.172 0.22 0.136 0.161 AIC 8.52E + 08 3.11E + 08 5.94E + 08 7.42E + 08 BIC 8.52E + 08 3.11E + 08 5.94E + 08 7.42E + 08 Exponentiated coefficients; standard errors in parentheses. Std. Err. adjusted for 9 clusters in division (division FE omitted) + p < 0.10, p < 0.05, p < 0.01, p < 0.001 Table 5 Full logit model results, across income cohorts (wave 2) Any hardships Eviction Rent/mortgage Utility (1) (2) (3) (4) Black 1.344 2.510 1.503 1.063 (0.296) (0.875) (0.402) (0.248) Hispanic 1.403 2.854** 1.917* 1.021 (0.324) (1.024) (0.518) (0.253) Asian/other 1.020 1.397 1.123 0.607 (0.420) (1.554) (0.688) (0.294) Income, AMI, and family size adjusted 0.990*** 0.995* 0.991* 0.988* (0.002) (0.003) (0.002) (0.002) Black × income, AMI, and family size adjusted 1.004 1.002 1.006+ 1.002 (0.003) (0.004) (0.003) (0.003) Hispanic × income, AMI, and family size adjusted 1.004 0.996 1.002 1.004 (0.002) (0.004) (0.003) (0.003) Asian/other × income, AMI, and family size adjusted 1.003 1.001 1.004 1.006 (0.004) (0.008) (0.005) (0.005) Gender: female 0.783* 0.508* 0.553* 0.926 (0.088) (0.101) (0.077) (0.115) Age = [25,35) 0.621 0.420* 0.777 0.708+ (0.105) (0.107) (0.155) (0.129) Age = [35,45) 0.384* 0.224* 0.422* 0.484* (0.070) (0.063) (0.092) (0.093) Age = [45,55) 0.408* 0.127* 0.436 0.529 (0.082) (0.060) (0.116) (0.112) Age = 55 or more 0.106* 0.014* 0.098* 0.139* (0.025) (0.008) (0.029) (0.035) Marital status: single, never married 0.999 0.821 0.954 0.952 (0.146) (0.191) (0.170) (0.147) Marital status: separated/divorced/widowed 1.195 0.544 1.207 1.179 (0.215) (0.209) (0.285) (0.227) Education: some college/certificate/associate’s degree 1.290+ 1.165 0.889 1.510** (0.168) (0.285) (0.145) (0.210) Education: bachelor’s degree 0.869 1.275 0.720 0.915 (0.138) (0.386) (0.146) (0.157) Education: graduate or professional degree 1.085 2.131* 0.955 1.074 (0.194) (0.738) (0.218) (0.211) # kid[s]: 1 1.939* 2.939* 2.316* 1.559 (0.271) (0.652) (0.390) (0.239) # kid[s]: 2 1.894* 2.938* 2.275* 1.676 (0.343) (0.789) (0.477) (0.316) # kid[s]: 3 + 1.854* 4.055* 2.218 1.615+ (0.476) (1.578) (0.635) (0.438) Own home free and clear 0.754+ 1.215 0.733+ 0.607 (0.111) (0.251) (0.128) (0.102) Pay rent 0.994 0.367* 0.643* 1.092 (0.139) (0.089) (0.111) (0.163) Neither own home nor pay rent 0.354* 0.225 0.197* 0.429 (0.095) (0.103) (0.078) (0.123) RECODE of ben_gov_cares_ever (Has your HH received the second relief payment) = 1 1.607*** 1.396 1.417* 1.891*** (0.201) (0.322) (0.217) (0.269) Eviction modratorium: enacted 0.869 0.928 0.793 0.893 (0.127) (0.230) (0.142) (0.145) Constant 0.793 0.313* 0.531 0.377* (0.314) (0.167) (0.259) (0.161) Observations 4805 4805 4805 4805 Pseudo R2 0.192 0.257 0.184 0.179 AIC 1.82E + 08 7.56E + 07 1.32E + 08 1.56E + 08 BIC 1.82E + 08 7.56E + 07 1.32E + 08 1.56E + 08 Exponentiated coefficients; standard errors in parentheses. Std. Err. adjusted for 9 clusters in division (division FE omitted) + p < 0.10, p < 0.05, p < 0.01, **p < 0.001 Data Availability The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request. Declarations Conflict of Interest The authors declare no competing interests. 1 We capitalize “Black” and “Hispanic” in a racial, ethnic or cultural sense, conveying an essential and shared sense of history, identity and community among people who identify as Black, including those in the African diaspora and within Africa. The lowercase black indicates a color not a person. On the other hand, we do not capitalize “white” as they generally do not share the same history and culture or the experience of being discriminated against because of skin color (see Dumas 2016; Kolchin 2002). 2 In this study, we define liquid assets as cash in hand or assets that can easily be converted into cash in a short amount of time, such as assets in checking and savings accounts. 3 Although it was not originally intended to do so, most survey waves corresponded to important social event(s) during the pandemic. The wave 1 survey was administered approximately 3 months after the pandemic outbreak when the first stimulus checks from the government (Economic Impact Payments) had just been distributed. The wave 2 survey was implemented when the Black Lives Matter movement peaked. The Wave 3 survey was distributed a day after the US presidential election (and Pfizer announced its vaccine candidate a week after survey administration began). By wave 4, the COVID-19 infection cases had sharply decreased due to the COVID-19 vaccines and the infection level remained low in wave 5. 4 Research has demonstrated that online, nonprobability samples using Qualtrics panels generate samples that closely approximate those of the General Social Survey, which uses a probability sampling design, which is the gold standard in survey administration (Zack et al. 2019). 5 Although the Washington University in St. Louis institutional review board (IRB) established that this study was not human subject research, researchers still obtained informed consent from participants prior to administering the survey. 6 The use of this relative income measure has several advantages over the use of the absolute dollar amount. First, it allows us to account for the cost of living in an area. $100 K in San Francisco County, for instance, is not equal to $100 K in McDowell County, West Virginia. The use of this measure also allows us to create categories that correspond to policy-relevant income groups, in this case following HUD’s income classification system to determine Fair Market Rents and Sect. 8 qualifications. However, we ran another set of analyses using absolute dollar amounts (and a control for median income at the county level) instead of relative dollar amounts and confirm that the results are consistent and robust. 7 States handled eviction restrictions in a wide variety of different ways, with a wide variety of start and end dates. We code states as 1 if they put any restrictions on residential evictions or foreclosures over and above what was included in the CARES Act for any length of time between March 1 (start of the pandemic) and August 10 (start of the survey) in 2020, and as 0 if they did not. 8 Summary statistics of the covariates in our empirical models are available in Appendix 1. 9 Full logistic regression model results are available in Appendix 2. 10 To estimate the mediation effect of liquid asset amount (at 3 months prior to the survey) and employment shocks (in 3 months prior to the survey), we used the housing hardship experiences within 3 months prior to the survey instead of the hardship experiences during the pandemic. 11 Though our analytic sample is nationally representative, each racial/ethnic group is not well balanced (for instance, more than 70% of the Black respondents were female). To make each racial/ethnic group as well as each income cohort be representative, therefore, we employed a weighting scheme. 12 These figures are much higher than in 2019 when according to CoreLogic’s report (2019), 0.4% of homeowners were foreclosed upon. The mortgage or rent delinquency rate in our survey is also notable, as this is much higher than the 4.5% delinquency rate in 2019 (CoreLogic, 2019). 13 Appendix 2 presents the full regression output. In addition to the main findings of this paper, these models show that, younger respondents, male respondents, and those with dependents were more likely to experience the measured housing hardships. Interestingly, these models show that renters were less likely to experience eviction than those who own a home with a mortgage, and that those with graduate/professional degrees were more likely to experience eviction than those with a high school degree or less. However, these associations are a function of controlling for other variables such as income, which is correlated with homeownership status, education, and our outcomes of interest. When examining these relationships in isolation, renters are more likely, and graduate degree holders less likely, to experience eviction. 14 Each survey is retrospective, asking for each hardship experience in the 3 months prior to the survey. For those interviewed at the earliest time (April 27), 3 months earlier started January 27; for those interviewed at the latest time, 3 months earlier started on February 12. Because the pandemic did not start having a large effect on the economy until mid-March, the data capture at least 1 month prepandemic. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Bayer P Ferreira F Ross SL The vulnerability of minority homeowners in the housing boom and bust Am Econ J Econ Pol 2016 8 1 1 27 10.1257/pol.20140074 Brown CJ Pagán JA Changes in employment status across demographic groups during the 1990–1991 recession Appl Econ 1998 30 12 1571 1583 10.1080/000368498324652 Buis ML Direct and indirect effects in a logit model Stata Journal 2010 10 1 11 29 10.1177/1536867x1001000104 22468140 Bureau of Labor Statistics, U.S. Department of Labor, The Economics Daily, Unemployment rate rises to record high 14.7 percent in April 2020 at https://www.bls.gov/opub/ted/2020/unemployment-rate-rises-to-record-high-14-point-7-percent-in-april-2020.htm (visited November 29, 2022). Caswell KJ, Zuckerman S. Food insecurity, housing hardship, and medical care utilization. Urban Institute. 2018 Choi JH, Pang D. Six facts you should know about current mortgage forbearances. Urban Institute. 2020. Couch KA Fairlie R Last hired, first fired? Black-white unemployment and the business cycle Demography 2010 47 1 227 247 10.1353/dem.0.0086 20355692 Desmond M Eviction and the reproduction of urban poverty Am J Sociol 2012 118 1 88 133 10.1086/666082 Despard MR Grinstein-Weiss M Guo S Taylor S Russell B Financial shocks, liquid assets, and material hardship in low- and moderate-income households: Differences by race J Econ, Race, Policy 2018 1 4 205 216 10.1007/s41996-018-0011-y Dumas MJ. Against the dark: Antiblackness in education policy and discourse. Theory Pract. 2016;55(1):11–9. Duster C. Utility shutoffs threaten a fresh crisis for low-income and Black families as Covid surges again. CNN Politics. 2020. https://www.cnn.com/2020/07/12/politics/utility-shutoffs-coronavirus/index.html. Accessed 16 May 2022 Eamon MK Wu CF Effects of unemployment and underemployment on material hardship in single-mother families Child Youth Serv Rev 2011 33 2 233 241 10.1016/j.childyouth.2010.09.006 Einstein KL, Glick DM. Does race affect access to government services? An experiment exploring street-level bureaucrats and access to public housing. Am J Political Sci. 2017;61(1):100–16. Ellen I, Graves E, O’Regan K, Schuetz J. Strategies for increasing affordable housing amid the COVID-19 economic crisis. Brookings Institution, Washington, DC. 2020. https://www.brookings.edu/research/strategies-for-increasing-affordable-housing-amid-the-covid-19-economic-crisis/ Fairlie RW, Couch KA, Xu H, Fairlie Kenneth Couch Huanan Xu RW, Robert Fairlie MW, Couch KA, Xu H, Fairlie RW, and Xu Judd H. The impacts of COVID-19 on minority unemployment: First evidence from April 2020 CPS microdata. In NBER Work Paper Ser. 2020. 10.2139/ssrn.3604814 Farrell D, Greig F, Wheat C, Liebeskind M, Ganong P, Noel P, Jones D. Racial gaps in financial outcomes: Big data evidence. SSRN Electron J. 2020. 10.2139/ssrn.3582557. Freeman RB Gordon RA Bell D Hall RE Changes market in the for labor black Brook Pap Econ Act 1973 1973 1 67 131 10.2307/2534085 Gjertson L Emergency saving and household hardship J Fam Econ Issues 2016 37 1 1 17 10.1007/s10834-014-9434-z Greenberg D Gershenson C Desmond M Discrimination in evictions: empirical evidence and legal challenges Harv CR-CLL Rev 2016 51 115 Grodsky E Pager D The structure of disadvantage: Individual and occupational determinants of the black-white wage gap Am Sociol Rev 2001 66 4 542 567 10.2307/3088922 Heflin C Family instability and material hardship: Results from the 2008 Survey of Income and Program Participation J Fam Econ Issues 2016 37 3 359 372 10.1007/s10834-016-9503-6 Heflin C The role of social positioning in observed patterns of material hardship: New evidence from the 2008 Survey of Income and Program Participation Soc Probl 2017 64 4 513 531 10.1093/socpro/spw041 Huffman ML Cohen PN Racial wage inequality: Job segregation and devaluation across U.S. labor markets Ame J Sociol 2004 109 4 902 936 10.1086/378928 Jelleyman T Spencer N Residential mobility in childhood and health outcomes: a systematic review J Epidemiol Community Health 2008 62 7 584 592 10.1136/jech.2007.060103 18559440 Jowers K, Timmins C, Bhavsar N, Hu Q, and Marshall J. Housing precarity & the covid-19 pandemic: Impacts of utility disconnection and eviction moratoria on infections and deaths across us counties. Nat Bureau Econ Res. 2021. 10.3386/w28394 Karpman M, Zuckerman S, Gonzalez D, and Kenney GM. The COVID-19 pandemic is straining families’ abilities to afford basic needs: Low-income and hispanic families the hardest hit. 2020. Washington, DC: Urban Institute. Klein A Shiro AG The COVID-19 recession hit Latino workers hard Here’s what we need to do 2020 Brookings Institute Kolchin P Whiteness studies: The new history of race in America J Am His 2002 89 1 154 173 10.2307/2700788 Kowalski KM. Racial disparities persist in electric service. Is ‘willful blindness’ to blame? Energy News Net. 2020. https://energynews.us/2020/07/01/racial-disparities-persist-in-electric-service-is-willful-blindness-to-blame/ Long SK. Hardship among the uninsured: Choosing among food, housing, and health insurance. Urban Inst Ser B. 2003;B–54. https://webarchive.urban.org/UploadedPDF/310775_B-54.pdf Lopez MH, Rainie L, and Budiman A. Financial and health impacts of COVID-19 vary widely by race and ethnicity. Pew Research Center. 2020. https://www.pewresearch.org/fact-tank/2020/05/05/financial-and-health-impacts-of-covid-19-vary-widely-by-race-and-ethnicity/ McCall L Sources of racial wage inequality in metropolitan labor markets: Racial, ethnic, and gender differences Am Sociol Rev 2001 66 4 520 541 10.2307/3088921 McKernan S, Ratcliffe C, and Vinopal K. Do assets help families cope with adverse events? In Washington, DC: Urban Institute. 2009. Medina RM, Byrne K, Brewer S, and Nicolosi EA. Housing inequalities: Eviction patterns in Salt Lake County, Utah. Cities. 2020;104. 10.1016/j.cities.2020.102804 Neckerman KM Garfinkel I Teitler JO Waldfogel J Wimer C Beyond Income Poverty : Measuring Disadvantage in Terms of Material Hardship and Health Acad Pediatr 2016 16 3 S52 S59 10.1016/j.acap.2016.01.015 27044702 Niedt C Martin IW Who are the foreclosed? A statistical portrait of america in crisis Hous Policy Deb 2013 23 1 159 176 10.1080/10511482.2012.702119 Pager D Shepherd H The sociology of discrimination: Racial discrimination in employment, housing, credit, and consumer markets Ann Rev Sociol 2008 34 181 209 10.1146/annurev.soc.33.040406.131740 20689680 Pan WK Tyrovolas S Iago G-V Dasgupta RR Daniel F Zaitchik B Lantos PM Woods CW Covid-19: Effectiveness of non-pharmaceutical interventions in the United States before phased removal of social distancing protections varies by region MedRxiv 2020 69 458 Pattillo M. Black picket fences, second edition: Privilege and peril among the Black middle class. University of Chicago Press. 2013. Pilkauskas NV Currie JM Garfinkel I The great recession, public transfers, and material hardship Social Serv Rev 2012 86 3 401 427 10.1086/667993 Raymond E Wang K Immergluck D Race and uneven recovery: neighborhood home value trajectories in Atlanta before and after the housing crisis Hous Stud 2016 31 3 324 339 10.1080/02673037.2015.1080821 Ren C A Framework for Explaining Black – White Inequality in Homeownership Sustainability In CSD Working Papers Demogr 2020 57 1297 10.1007/s13524-020-00894-4 Rich WC Massey DS Denton NA American apartheid: Segregation and the making of the underclass Political Sci Quar 1993 108 3 574 10.2307/2151725 Roll S, Bufe S, Chun Y, and Grinstein-Weiss M. The socioeconomic impacts of COVID-19 study: Survey methodology report. In Social Policy Institute Research. Social Policy Institute, Washington University in St. Louis. 2021. 10.7936/r4cj-5041. Accessed 16 May 2022 Roll S, and Grinstein-Weiss M.Did CARES Act benefits reach vulnerable Americans? Evidence from a national survey. Brookings. 2020. https://www.brookings.edu/research/did-cares-act-benefits-reach-vulnerable-americans-evidence-from-a-national-survey/ Ross LM Squires GD The personal costs of subprime lending and the foreclosure crisis: a matter of trust, insecurity, and institutional deception Soc Sci Q 2011 92 1 140 163 10.1111/j.1540-6237.2011.00761.x Rothstein R The color of law: A forgotten history of how our government segregated America 2017 Liveright Publishing Shah M Sachdeva M Dodiuk-Gad RP COVID-19 and racial disparities J Am Acad Dermatol 2020 83 1 e35 10.1016/j.jaad.2020.04.046 32305444 Sharkey P. Stuck in place: Urban neighborhoods and the end of progress toward racial equality. University of Chicago Press. 2013. StataCorp Stata Statistical Software: Release 16 (Version 16) 2019 StataCorp LLC Tomich J, Klump E, Swartz KE, and Skibell A. Pandemic, unpaid bills test utilities’ social justice vows. E&E News. 2020. https://www.eenews.net/articles/pandemic-unpaid-bills-test-utilities-social-justice-vows/ Townsend MJ Kyle TK Stanford FC Outcomes of COVID-19: disparities in obesity and by ethnicity/race Int J Obes 2020 1–3 1807 10.1038/s41366-020-0635-2 Williams J Laid off more, hired less: Black workers in the COVID-19 recession 2020 RAND Zack ES Kennedy J Long JS Can nonprobability samples be used for social science research? A cautionary tale Surv Res Meth 2019 13 2 215 227 Zhang S Lerman RI Does homeownership protect individuals from economic hardship during housing busts? Hous Policy Debate 2019 29 4 522 541 10.1080/10511482.2018.1532447	36474600	PMC9716543	NO-CC CODE	2022-12-03 23:20:58	no	J Econ Race Policy. 2022 Dec 2;:1-19	utf-8	J Econ Race Policy	2,022	10.1007/s41996-022-00109-5	oa_other
==== Front Pharmacoeconomics Pharmacoeconomics Pharmacoeconomics 1170-7690 1179-2027 Springer International Publishing Cham 1225 10.1007/s40273-022-01225-8 Editorial Best Practices: A Collection of Systematic Critical Reviews of Modeling Approaches in Specific Disease Areas http://orcid.org/0000-0001-5470-3754 Caro J. Jaime [email protected] [email protected] [email protected] 12345 1 grid.14709.3b 0000 0004 1936 8649 McGill University, Montreal, Canada 2 grid.13063.37 0000 0001 0789 5319 London School of Economics, London, UK 3 grid.4280.e 0000 0001 2180 6431 National University of Singapore, Singapore, Singapore 4 Evidera, Boston, MA USA 5 39 Bypass Rd, Lincoln, MA 01773 USA 2 12 2022 13 10 11 2022 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. ==== Body pmcThe use of models to forecast disease outcomes and the effects of interventions seems to be exploding, perhaps prompted by the prominent use of models during the coronavirus disease 2019 (COVID-19) pandemic. A cursory literature search for ‘modeling’ and ‘cost-effectiveness’, and limiting this to health care journals, reveals a steady rise from less than 40 papers per year at the turn of the century to nearly 300 in 2020 and 2021. In 2022, the number is approaching 500, with 20% of those related to COVID-19. With the growing number of models published in any given disease area, it is important to ensure that the models have been validated and that they undergo systematic critical review. To foment this activity, Pharmacoeconomics is launching a series of papers reporting on critical review of modeling approaches in specific disease areas. In this short introduction to the series, I lay out what the review papers should cover. Comprehensiveness Each review should strive to cover all models in a disease area that seek to inform decisions about the use of health technologies, regardless of the language or year of publication. The focus of the review should be on papers that report on the methods used to conceptualize and implement the model, no matter what specific technologies are assessed or what analyses are reported. Supplementary online materials, and any additional documentation posted on websites should also be sought and reviewed. How papers were identified should be described following applicable guidelines, such as the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [1]. If the model itself is available, it would be ideal if the review authors examined it, but it is recognized that this may go too far. Publications reporting results with only cursory descriptions of the model should be used to trace back to the paper that provides the required details. If none is found, it would be most helpful if an attempt is made to contact the authors to request the details. When this also fails to turn up sufficient information, the model should still be listed, noting that it could not be critically reviewed and the reason for this. Focus on Methods Although the quality of the reporting will undoubtedly affect the ability to critically review the model, the review should be about the methods not about how they are reported. If some methodological aspect is left out or is unclear, this should be noted, but there is no need to score each paper on its reporting quality. Results of particular analyses are not of interest, except where they reflect on methodological choices. If so, the review should address what the results indicate about those choices. Similarly, the specific values used as inputs (e.g., the discount rate) are not relevant unless they form part of the model structure (e.g., the cycle length in a Markov model). The review should be structured using a methodological best practice guidance (e.g., Caro et al. [2]). Generally, it should cover how the model was conceptualized and its intended uses; the type of framework selected; details specific to that type of framework; sources for the structural assumptions; how uncertainty was considered; and how transparent the model is and how it was validated. Most of these details can be tabulated for each model, with the text reserved for the reviewers’ assessment of their adequacy. Conceptualization In this section, the review should consider what the objectives were for the model; who is the intended audience(s); what problems it was meant to address and its intended uses. Is it meant for a single or multiple application (‘whole disease model’ [3])? The scope of the model (e.g., is it limited to a particular severity of disease); what perspective(s) are enabled for analyses; details of the disease types covered; which populations are targeted; which interventions are modeled and how feasible it is to add new ones; what outcomes are considered and how they are measured; and what formal process was followed in designing the model should be covered. The reviewers should not just list the details but rather critically assess these and note any gaps. The type of model selected, and its justification, should be reported. A simple classification can be used: is the model deterministic or stochastic? If it is stochastic, does it consider what happens periodically or does it contemplate the time until each event occurs? Decision trees, partitioned survival, cohort state-transition (‘Markov’) and static SEIR models are deterministic frameworks. Stochastic structures can be individual-level Markov (‘microsimulation’), discrete-event simulation (usually unconstrained), and agent-based simulation. Other types (e.g., dynamic transmission models, systems dynamics, general equilibrium) are much less common in our field. Whatever type was selected, the review should also assess whether it is adequate for the stated purpose(s) of the model. Structure The review should provide details of the structure according to the model type. For example, for a cohort Markov model, it should be specified if it is a chain (i.e., constant transition probabilities); what states were defined, together with which transitions are allowed and whether these adequately represent the problem; what sources were used to derive transition probabilities and how projections were made; how heterogeneity in determinants of the transition probabilities was handled (e.g., were there separate states for males and females or was a proportion used in a single state); what cycle length was chosen, whether it was short enough to adequately reflect the frequency of transitions; if it can vary over time and was there a half-cycle correction; which outcomes are accrued and how utilities are applied. In stochastic models, the review should describe the possible trajectories; what patient profiles are considered, on what basis they were defined and whether the same individuals are modeled for each intervention to reduce nuisance variability; the handling of continuous disease parameters (e.g., are they regularly updated); for time-to-event models what event time distributions are used, how they were determined, how times are drawn and redrawn; how stochastic uncertainty is handled and the availability of stability analyses. Other model types may require description of additional or alternative aspects. While the choice of input values is not generally relevant for the review, some inputs may affect the structure. For example, how a determinant of patient trajectories is handled, especially if it is projected into the future, would be germane. For all aspects, appraisal of the structural choices must be made. Uncertainty The results of specific uncertainty analyses are not of much interest—they are entirely dependent on the purpose of that particular study. Instead, the review should address the types of uncertainty analyses the model enables. Of greatest importance is structural uncertainty that arises from the assumptions made and methodological choices. Does the model facilitate analysis with different structural assumptions? For example, is a structure approach that facilitates scenarios mentioned? If time-to-event distributions are used (e.g., for mortality), can the user select a different distribution? Are alternative structures available with toggles for easy activation? Has structural uncertainty been parameterized to facilitate testing? Also of importance is parameter uncertainty. Most models today allow for one-way deterministic analyses across a range of input values and also for probabilistic analyses that draw input value sets according to distributions that describe their uncertainty. The review should address the extent to which the model enables these analyses. Can the user readily modify how the uncertainty is characterized? Were any inputs derived via calibration and can their uncertainty be examined? Are the inputs that control the analysis (e.g., the number of replications to be run) easily changed? Validation The first step in validation is to appraise the face validity of the model concept, its structure, and the evidence used in its design. Modelers should have independent experts evaluate face validity and document what questions were raised, or, at a minimum, how they were resolved. It is also helpful if the model has been submitted for review by an agency or other external organization. Errors in implementation of models are common and thus it is important to subject a model to formal, rigorous verification that it is correctly specified and works as intended. This verification should be documented and its results should be available to anyone interested in that model. Assessing the extent to which a model’s forecasts are accurate is perhaps the most important aspect to document in a review. As the reviewer cannot be expected to independently verify this, it behooves the modelers to compare their model’s forecasts to actual observations. Ideally, these external data were not used in the building of the model and thus an independent validation was performed. However, at a minimum, modelers should have compared the model predictions with what was obtained in the studies used as sources for the model. These dependent validations are not as strong but are more practical to do. For all three types of validation, the review should note whether the modelers have indicated that it was done and the process that was followed. Published validation checklists can be leveraged in this regard [4, 5]. For face validity, they should note who was involved and their degree of independence from the project and funders. The review should also list known limitations of the model specified by the authors and the extent to which external validation was performed. While this is rarely included in a paper reporting on a model, the availability of report(s) detailing what was done, and any resulting modifications, should be noted. Transparency All models should be accompanied by non-technical documentation that describes the model in terms that any reader can follow. The review should note whether this documentation is available freely, perhaps in the model itself, or via supplementary materials or posted on a website. Models should also have full technical documentation that provides all the details that would enable an interested person with the proper skills to rebuild the model. The review should address whether this documentation exists and whether it is available freely or under some non-disclosure agreement. Apart from documentation, the review should note whether the model is available for others to review and use, and whether openly or under licensing. The software used to implement the model should also be given. Funding sources and other potential conflicts of interest should also be listed. Conclusion Modeling review papers submitted to Pharmacoeconomics will undergo peer review guided by the criteria set out here. We encourage researchers to contribute to this Series—those having already completed such a review some time ago are welcome to update it, ensuring that it meets the criteria, and submit it for consideration. We hope that this Series will not only be a useful resource for researchers in the area but will also provide guidance on best practice for future modeling efforts. Hopefully, it will also help elevate the standards these models must meet. Funding J. Jaime Caro is an employee of Evidera, a part of Thermo Fisher Scientific that receives funding from health sciences companies for health economic work. Data availability Data availabiltiy statement was not required for Editorial submission, per on-line website. ==== Refs References 1. Page MJ Moher D Bossuyt PM Boutron I Hoffmann TC Mulrow CD PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews BMJ 2021 372 160 10.1136/bmj.n160 2. Caro JJ Briggs AH Siebert U Kuntz KM Modeling good research practices—overview: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force -1 Value Health. 2012 15 798 803 10.1016/j.jval.2012.06.012 3. Tappenden P Chilcott J Brennan A Squires H Stevenson M Whole disease modeling to inform resource allocation decisions in cancer: a methodological framework Value Health. 2012 15 1127 1136 10.1016/j.jval.2012.07.008 23244816 4. Vemer P Corro Ramos I van Voorn GAK AdViSHE: a validation-assessment tool of health-economic models for decision makers and model users Pharmacoeconomics 2016 34 349 361 10.1007/s40273-015-0327-2 26660529 5. Büyükkaramikli NC Rutten-van Mölken MPMH Severens JL TECH-VER: a verification checklist to reduce errors in models and improve their credibility Pharmacoeconomics 2019 37 1391 1408 10.1007/s40273-019-00844-y 31705406	36459348	PMC9716546	NO-CC CODE	2022-12-03 23:20:58	no	Pharmacoeconomics. 2022 Dec 2;:1-3	utf-8	Pharmacoeconomics	2,022	10.1007/s40273-022-01225-8	oa_other
==== Front J Cult Econ Journal of Cultural Economics 0885-2545 1573-6997 Springer US New York 9465 10.1007/s10824-022-09465-4 Original Article How does urban violence impact choices of cultural participation? The case of the Maré favela complex in Rio de Janeiro http://orcid.org/0000-0002-7824-4828 Iachan Luisa [email protected] 1 Moreau François [email protected] 1 http://orcid.org/0000-0001-7432-5404 Heritage Paul [email protected] 2 http://orcid.org/0000-0002-0715-1018 Valiati Leandro [email protected] 3 Silva Eliana Sousa [email protected] 4 1 grid.462844.8 0000 0001 2308 1657 Centre de recherche en économie et gestion (CEPN) & LabEx ICCA, Université Sorbonne Paris Nord, Avenue Jean-Baptiste Clément, Villetaneuse, France 2 grid.4868.2 0000 0001 2171 1133 School of English and Drama, Queen Mary University of London, Mile End Rd, Bethnal Green, London, UK 3 grid.5379.8 0000000121662407 School of Arts, Languages and Culture, University of Manchester, Oxford Road, Manchester, UK 4 grid.503479.c Redes da Maré, Rua Sargento Silva Nunes, Rio de Janeiro, Brazil 2 12 2022 133 28 1 2022 16 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The impact of urban violence on society has been the subject of several studies, but the consequences of fear for habits of cultural consumption are missing in cultural economics research. This article investigates whether the fear of urban violence explains individuals’ choice between different options of cultural participation with a particular focus on the activities of watching movies and listening to music. Based on individual data from a survey conducted in 2019 with 1211 residents from a conglomeration of sixteen favelas (slums) located in the Maré neighbourhood in Rio de Janeiro (Brazil), this study employs Simultaneous Bivariate Ordered Probit Models to verify the association between individuals’ fear of violence and their choice of consuming culture in private or public spaces. Controlling for socioeconomic, demographic, and territorial variables, the findings indicate that consuming culture in private spaces is a substitute for public spaces when individuals are more afraid of violence. The results presented in this work provide evidence for the design and implementation of policies targeting territories impacted by high levels of violence. Keywords Cultural participation Urban violence Brazilian favelas Probit model JEL Classification D91 R22 Z11 Economic and Social Research Council (UK)ES/S000720/1 ESRC-AHRC GCRF Mental Health 2017 Heritage Paul Arts council of England (UK)ES/S000720/1 ESRC-AHRC GCRF Mental Health 2017 Heritage Paul ==== Body pmcIntroduction Urban violence has been the object of several economics-based studies. Empirical evidence revealed that violence generally imposes costs on society and, at the individual level, the fear of becoming a victim often affects behaviours and habits (Greenbaum & Tita, 2004; Skogan & Maxfield, 1981; Warr, 2000). This paper looks at the way this fear affects the means through which people participate in cultural activities. Many different approaches have dealt with the indirect consequences of urban violence. These consequences range from losses in terms of gross domestic product and costs imposed on the labour market, to the influence over individuals’ decisions of where to live and what to consume (Carboni & Detotto, 2016; Cullen & Levitt, 1999; Greenbaum & Tita, 2004; Mejía & Restrepo, 2016). However, none of these studies investigated the impact of the fear of victimization on decisions of where to participate in activities that can be taken either in public or in private spaces. This question is particularly relevant in a global context where new technologies progressively offer digital consumption alternatives that compete with traditional modalities (presential, physical or face-to-face) in numerous sectors. The cultural sector was especially affected by this in recent years. Access to cultural and creative content is increasingly made via digital means, through platforms such as YouTube, Spotify, Netflix, and others (Waldfogel, 2017). At the same time, part of the importance of culture and creativity in terms of socioeconomic development seems to be related to public spaces. There is an acknowledgement of the role of culture and creativity for socioeconomic development, economic growth, quality of life and urban regeneration (Florida, 2002; Galloway, 2006; Schlesinger, 2016). Since the 90s, several cultural-based initiatives at the local level were implemented to support and stimulate economic and social development. Those are committed to appealing to the creative class and providing an intensive cultural supply (Florida, 2002). This is also part of the creative economy paradigm, which associates creativity with human capital and long-term productivity and competitiveness (Bakhshi et al., 2015). This paper focuses on the decision between participating in cultural activities in public or private spaces in territories where armed conflicts are frequent and the fear of violence restricts mobility. If the fear of being a victim constitutes an obstacle to participation in cultural life, urban areas affected by violence may lag behind in terms of socioeconomic development and in any possibility of improving quality of life through a creative economy. It is widely known that urban violence is unevenly distributed across territories and that some segments of the population commonly face greater risk than others (Winton, 2004). Briceño-León and Zubillaga (2002) reported, for instance, that the homicide rate in Rio de Janeiro is on average 4.5 times higher in the lower-income districts than in the city’s middle-class and tourist areas. Most of the favelas1 are particularly affected by violence due to the presence of gangs linked to illegal drug trade or of militias that exploit basic services through the use or threat of violence, and precariously monopolize functions that the State is supposed to regulate. At the same time, public security agents generally do not recognize the right of this population to live in safety and use practices that intimidate and contribute to creating insecurity and fear in daily life. The current logic of the State in most of these spaces is warmongering, characterized by specific actions of police operations marked by intense armed conflict. These conflicts often employ heavy weaponry, such as grenades and modern military machine guns, and regularly lead to deaths, including of those not directly involved in the drug trade. The imminence and unpredictability of armed confrontations, whether between rival armed groups or between those groups and the police forces, have massive adverse impacts on the daily lives of favela residents, particularly on their freedom to move. In Rio de Janeiro, 1.3 million people (around 22% of the population) live in favelas according to the Brazilian Institute of Geography and Statistics (IBGE, 2010). This article studies the case of the sixteen favelas of the Maré district, which is home to approximately 10% of the population living in favelas in Rio de Janeiro. Based on individual data from a survey conducted in 2019 with 1211 residents, it associates the fear of victimization with choices between different ways in which people can consume culture. More specifically, between listening to music and watching movies in public spaces as compared to private spaces. The findings show a statistically significant and negative effect of the fear of being hit by a stray bullet on cultural practices that require individuals to go out (as compared to modalities that take place in private spaces). Results also show that fear represents a stronger determinant of individuals’ behaviour than the actual occurrence of armed conflict events. The paper is organized as follows. Section 2 reviews the literature in two parts: first, looking at the impacts of violence on individuals and society, then examining the findings that describe the typical determinants of cultural participation. Section 3 defines the hypothesis of this study and the theoretical background that supports them. Section 4 describes the context of the empirical study, the data and variables used for the analysis and the estimation strategy. Sections 5 and 6 provide the results and discuss them respectively. Section 7 concludes the article. Literature review The impact of violence on individuals and society The consequences of violence and the fear of crime were the subject of several empirical studies in a variety of research fields, including economics, psychology, criminology, and urban studies. The effect of territorial violence on individuals and the society have been observed in different aspects, but for the purposes of this article this review focuses on the economic and behavioural impacts. In this sense, multiple results were observed. These range from negative impacts on macroeconomics outcomes, such as the gross domestic product (e.g., Carboni & Detotto, 2016) to impacts on the individual level, such as health deterioration (e.g., Ross & Mirowsky, 2001). Most of the analyses discuss and provide evidence of what can be called the indirect costs of violence, which in turn affect the quality of life of individuals and society as a whole (Hale, 1996). Among those studies, a few are on the effects of the fear of crime on individuals’ behaviours and habits, including where they choose to live (e.g., Cullen & Levitt, 1999), their use of public transportation (Patterson, 1985), handgun ownership (DeFronzo, 1979) and conspicuous consumption (Mejía & Restrepo, 2016). No study was found however on the association between the fear of violence and individuals’ avoidance to go out for certain activities that can alternatively be done without leaving their residence. The literature suggests that this kind of avoidance behaviour must inevitably provoke negative feedback effects on violence rates (Hale, 1996), decrease levels of social interaction (Garofalo, 1981) and negatively impact economic activities (Warr, 2000), with all of these bringing losses in terms of well-being. From a development economics and human capital accumulation perspective, Brown and Velásquez (2017) showed that young adults exposed to increased local violence attained significantly fewer years of education and were less likely to complete compulsory schooling. Furthermore, the effect of armed conflict on the accumulation of schooling in Tajikistan were studied by Shemyakina (2011). The author’s findings indicate that exposure to violent conflict had a large and statistically significant negative effect on girls’ enrolment and completion of compulsory studies, but no significative effect on the education of boys. Similarly, evidence from the 1994 Rwandan Genocide revealed its strong negative impact on schooling, with exposed children presenting a drop in education attainment of 18.3% relative to the average (Akresh et al., 2008). In Brazil, Monteiro and Rocha (2017) showed that the gunfights between drug gangs in Rio de Janeiro decreased students’ scores at school, and that the supply side of education played as an important mechanism, since schools in violent areas in Brazil experienced larger teacher absenteeism, less stability in administration and were more likely to temporarily close. Other indirect economic costs of urban violence include the effects on businesses and the labour market. Greenbaum and Tita (2004) found a negative impact of violent crime on the creation of new business establishments and the growth of employment in existing businesses. Looking at the businesses’ choice in terms of location, Rosenthal and Ross (2010) showed that retailers are more likely to locate in safer areas as compared to wholesalers in the same industry. Regarding impacts on the labour market, Hamermesh (1999) found that homicide rates deter working in the evening and at night, shifting it to the daytime. Besides, multiple studies have shown that the Mexican Drug War adversely impacted employment and earning outcomes (BenYishay & Pearlman, 2013; Dell, 2015; Robles et al., 2013; Velásquez, 2020). The housing market is also deeply affected by criminality and violence in urban areas, especially in most crime ridden areas (Lynch & Rasmussen, 2001). For instance, Linden and Rockoff (2008) and Pope (2008) found significative negative effects on property values when a registered sex offender moved into a neighbourhood. This happens as a result of a lower demand for living in areas perceived as dangerous (Hartnagel, 1979). Studies on the effect of crime on an individual’s residence location confirm this trend. Cullen and Levitt (1999) found that high crime rates cause people to move out of cities and Gould Ellen and O'Regan (2008) indicated that cities are better able to retain residents after crime rates are reduced. The choice of where to live is not the only behavioural aspect impacted by urban violence that was the subject of research. Mejía and Restrepo (2016) showed that property crime reduces the consumption of visible goods, not only because some of these may be stolen, but also because they reveal information about an individual’s wealth. Besides, scholars refer to other multiple individual reactions to the fear of crime, such as: adopting more protective behaviour and ensuring personal safety, like owning guns (DeFronzo, 1979; Krahn & Kennedy, 1985), avoiding activities perceived as dangerous, like walking down some shared-use routes (Ravenscroft et al., 2002), and using less public transportation (Patterson, 1985). Moreover, Hale (1996) argues that the fear of crime is expected to make individuals stay at home more. No study was found, however, on the impact of violence and fear on cultural participation. The determinants of cultural participation There is a vast body of literature on the socioeconomic determinants of cultural participation, from the studies of Baumol and Bowen (1966) and Bourdieu and Passeron (1964) to the present days. These studies provide empirical evidence on how differences between social strata (such as income, type of employment and education) and demographic groups (like age and gender) are associated with differences in terms of cultural participation. Among all socioeconomic determinants, schooling (measured by years of study in the formal education system) is seen as the main factor associated with the intensity of cultural participation (Seaman, 2006). Most empirical studies are carried out in developed countries, but the few studies conducted in the developing world find similar results: education and income are positively associated with participation in most cultural activities (Courty & Zhang, 2018). In areas considered to be of low income in developed countries such as the United States, United Kingdom and Ireland, findings indicate low cultural participation in general for all inhabitants (Moore, 1998). In Brazil, Diniz and Machado (2011) and De Almeida et al. (2020) analysed data from the Family Budget Survey (POF) and found that spending on culture is strongly determined by education and income. Other personal characteristics are also indicated by the empirical literature as determinants of cultural participation. Particularly, the type of content accessed and of activity practiced can be influenced by socioeconomic and demographic factors. For example, social class, location of residence, age and race seem to be specially associated with styles of musical content accessed (Mellander et al., 2018). Besides, gender is a strong determinant for participation in visual arts, where women are generally more participative than men (Bennett & Silva, 2006). Finally, the level of participation in cultural activities in childhood is also distinguished as a strong determinant for cultural participation in adult life (Orend, 1988). Concerning the relationship between violence and culture, this was studied mainly in the opposite direction: how arts, culture and the creative sector make territories safer, reducing and preventing urban violence. The report of the European Working Group on Culture and Development “In from the Margins: A contribution to the debate on Culture and Development in Europe” (1997), identifies the reduction of crime as an indirect social impact of arts and culture. According to the document, this impact results from the ability of culture to enrich the social environment with public amenities, to induce educational effects, to stimulate creativity, among others. Several studies verified this association empirically (Azevedo, 2016; Matarasso, 1997; Tubadji et al., 2015). However, no study was found on the other direction of this association, that is, on how violence may shape cultural participation. Theoretical background and hypothesis This study investigates whether urban violence explains individuals’ choice between participating in cultural activities in private or public spaces. A large body of evidence in economic psychology, criminology and urban studies suggests that fear of crime shapes individuals’ behaviour and choices (Becker & Rubinstein, 2011; Garofalo, 1981; Greenbaum & Tita, 2004; Liska et al., 1988). Particularly, behavioural economists acknowledge that individuals’ emotions cannot be ignored by any theory of choice (Kahneman, 2002). Among different types of individual behavioural reactions to crime (as a consequence of fear), avoidance is defined as “actions taken to decrease exposure to crime by removing oneself from or increasing the distance from situations in which the risk of criminal victimization is believed to be high.” (DuBow et al., 1979, pp. 16). For instance, the fear of being a victim makes people alternate the routes they take when commuting, the form of transportation they use and the number of times they choose to leave their residence (see DuBow et al., 1979; Warr, 1994). According to survey data, spatial avoidance is the most common reaction to fear of urban violence in the United States (Warr, 1994). Such responses of avoidance must undeniably affect economic, leisure and social activities (Warr, 2000). An implication of this for the cultural and creative sector is that fear is expected to make people reduce attendance to these activities in public spaces. Hence, this study formulates the following hypothesis: H1 Individuals with more fear will choose to participate more in culture in private spaces as compared to public spaces. From the perspective of the economic geography field, the literature on objective and subjective geography of opportunities provides a theoretical framework to understand the way urban violence limits behaviours in the territory (Galster & Killen, 1995). This theory links individuals’ process of decision making to their geographical context. In short, subjective geography is seen as a limitation of the opportunities that are in fact available for individuals (objective geography). In the particular case of this study, investment on cultural equipment is essential to improve available opportunities for cultural participation (objective geography), but the fear of violence (subjective geography) might limit these opportunities. From an economic development perspective, this is also in line with Amartya Sen’s capability approach (Sen, 1999). More specifically, the fear of violence can be seen as a factor that limits individuals’ capabilities (or their freedom to act and make choices). In other words, the fear caused by urban violence is expected to restrict the real opportunities and the freedom of choice that people have. The empirical approach Context of the empirical study To contextualise the empirical examination, it is important to provide a brief characterization of the area under study (The Maré Favela Complex). The territory of Maré is composed of sixteen favelas occupying an area of 5.79 km2. The last official data accounted for a population of 139,073 inhabitants in 2013 (Redes da Maré, 2019). This makes it the ninth most populous neighbourhood in Rio de Janeiro, and more populous than 96% of Brazilian municipalities. At the same time, according to IBGE, Maré has the fourth lowest human development index and the fifth lowest income per capita among 126 neighbourhoods in Rio de Janeiro (IBGE, 2010). This is a reflection of a public sector that historically neglects favelas. Also, this implies a shortage of many basic services and public infrastructures. In what concerns the cultural and creative field, the neglect of the state produces a scenario in which cultural initiatives usually come from the third sector or informal community mobilizations. For instance, the Centre of Arts of Maré,2 founded by the non-governmental organization Redes da Maré, or an itinerant cinema, which is organized by a community resident.3 According to the Building Barricades mapping of cultural spaces (2019), there is no formal cinema in the whole territory of Maré and live music events mainly occur in bars and public squares. Besides, problems related to urban mobility, such as poor public transport and traffic jams, reduce the possibilities of cultural outings in the vicinity and in other parts of the city of Rio de Janeiro. This is an issue that affects all the dwellers of Maré in the same way. Indeed, in the metropolitan region of Rio de Janeiro, the average time people spend commuting per route is of 67 min, which is the highest in Brazil and one of the highest in the world. The sixteen favelas of Maré demarcate three different areas in the territory, as distinguished by Table 1. The demarcation of the territory by different areas emerged with the development, history of occupation, migratory movements and urban policies of the several favelas that compose it.4Table 1 Maré’s favelas encompassed by each geographical stratification Geographical strata Favelas Area 1 Nova Holanda, Parque Maré, Parque Rubens Vaz e Parque União Area 2 Baixa do Sapateiro, Conjunto Bento Ribeiro Dantas, Conjunto Esperança Conjunto Pinheiros, Morro do Timbau, Nova Maré, Salsa e Merengue, Vila do João, Vila dos Pinheiros Area 3 Marcílio Dias, Parque Roquete Pinto, Praia de Ramos There is a correspondence of these areas with the occupation and regulation by different armed groups.5 Since the public sector traditionally neglects favelas, these criminal groups ended up finding an aperture to control the territory (Silva et al., 2008). The actuation of these armed groups occurs through the governance of a series of illegal and irregular economic activities, such as drug trafficking, security services, public transportation, taxation of the sale of gas cylinders, distribution of TV signals and Internet connection, among others. All of this is sustained by an armed base and the frequent use of violence (Silva et al, 2008). This setting ends up producing a scenario marked by conflicts between different factions of the drug trafficking and between the state police and the armed gangs, which are central sources of violence and fear. People who live in areas affected by armed conflicts or near them are deeply impacted. Particularly, freedom of movement is considerably restricted during armed conflicts, given the danger of being hit by stray bullets. Some facts and figures can help positioning the violence in Maré in light of other regions of Brazil and the world. In the year 2019 alone, there were 49 deaths by firearms in Maré, 34 as a result of police action and 15 as a result of the action of armed groups, resulting in 1 death every 7 days and a firearm homicide rate of 35 per 100,000 inhabitants. The firearm homicide rate in the state of Rio de Janeiro is also high: a total number of 2321 deaths resulted in a firearm homicide rate of 34.5 per 100,000 inhabitants. These firearm homicide rates in Maré and the state of Rio de Janeiro are comparable to those of the most violent countries of the world, such as El Salvador (36.8), Venezuela (33.3), Guatemala (26.1) and Colombia (26.4) and to the most violent states in the United States, such as Alaska (24.4) and Mississippi (24.2). In Brazil, the firearm homicide rate was of 21.9 per 100,000 inhabitants in 2019, showing the differences in exposure to violence within the country.6 These data indicate that although the problem of violence is particularly serious in Maré, it is not an exclusivity of the territory, and other contexts in the world might be similarly affected by it. Data and variables description The data used on this work were extracted from the Building Barricades survey7 carried out in the period between September 2019 and January 2020 with 1211 adults residing in Maré. The in-person, door-to-door survey8 was organised and mediated by Redes da Maré, a local non-governmental organization. The surveyed population was made up of adults aged 18 years old or over, residing in households of the sixteen favelas in Maré.9 The estimated number of people aging 18 years old or over in Maré was 101,549 for the year of 2019. The sample size of 1211 adults was chosen as sufficient to reduce the margin of error and significantly represent the total population. Appendix 1 includes the tables containing the estimated population of Maré by age groups and gender in 2019, as well as the sample distribution by age groups and gender. The Building Barricades survey included sixty questions and lasted around fifty minutes. The respondents were asked about the frequency of their participation in a variety of cultural practices, their perceptions and experiences of violence, as well as their personal and socioeconomic characteristics, such as age, gender, education, employment, and the favela of residence. The variables used in this article are described in Table 2.Table 2 Description of the variables Dependent variables Gap movies cinema—Internet Categorical ordered variable ranging from − 4 to 4 Gap movies cinema—by other means Categorical ordered variable ranging from − 4 to 4 Gap music live—Internet Categorical ordered variable ranging from − 4 to 4 Gap music live—by other means Categorical ordered variable ranging from − 4 to 4 Independent variables Fear of being hit by stray bullet No fear = 0, rarely = 1, sometimes = 2, many times = 3, always = 4 Age Numerical discrete variable No education or pré-school Dummy (if this was the last level of education completed = 1, otherwise = 0) Elementary or middle education Dummy (if this was the last level of education completed = 1, otherwise = 0) High school Dummy (if this was the last level of education completed = 1, otherwise = 0) University, specialization or Master’s degree Dummy (if this was the last level of education completed = 1, otherwise = 0) Female Dummy (female = 1, male = 0) Household income Numerical discrete variable Childhood incentive Dummy (yes = 1, no = 0) Unemployment Dummy (unemployed = 1, employed = 0) Internet quality No Internet = 0, terrible = 1, bad = 2, regular = 3, good = 4, excellent = 5 Favela Baixa do Sapateiro Dummy (if residing in this favela = 1, otherwise = 0) Favela Conjunto Bento Ribeiro Dantas Dummy (if residing in this favela = 1, otherwise = 0) Favela Conjunto Esperança Dummy (if residing in this favela = 1, otherwise = 0) Favela Conjunto Pinheiros Dummy (if residing in this favela = 1, otherwise = 0) Favela Marcilio Dias Dummy (if residing in this favela = 1, otherwise = 0) Favela Morro do Timbau Dummy (if residing in this favela = 1, otherwise = 0) Favela Nova Holanda Dummy (if residing in this favela = 1, otherwise = 0) Favela Nova Maré Dummy (if residing in this favela = 1, otherwise = 0) Favela Parque Maré Dummy (if residing in this favela = 1, otherwise = 0) Favela Parque Roquete Pinto Dummy (if residing in this favela = 1, otherwise = 0) Favela Parque Rubens Vaz Dummy (if residing in this favela = 1, otherwise = 0) Favela Parque União Dummy (if residing in this favela = 1, otherwise = 0) Favela Praia de Ramos Dummy (if residing in this favela = 1, otherwise = 0) Favela Salsa e Merengue Dummy (if residing in this favela = 1, otherwise = 0) Favela Vila do João Dummy (if residing in this favela = 1, otherwise = 0) Favela Vila do Pinheiros Dummy (if residing in this favela = 1, otherwise = 0) For the purpose of this study, two variables of cultural practice were selected: watching movies and listening to music. The choice of these practices is justified by the fact that both are well developed and widely popularized in their domestic modalities (through online platforms or by other means, such as CDs and DVDs), which enables the comparison between participation in private and public spaces. Activities in public spaces typically involve leaving the home, while practices in private spaces do not require leaving the home. More specifically, watching movies in the cinema and listening to live music are cultural activities that take place in public spaces. On the other hand, practices that take place in private spaces are listening to music through online platforms or by other means (such as CDs and radio) and watching movies through online platforms and by other means (such as DVDs and TV). Participation in cultural activities were assessed in the survey using five levels of frequency, ranging from daily participation to not participating at all. Since the interest of this article is to assess the individuals’ choice between participating in cultural activities in private or public spaces, four new variables were created to be used as dependent variables: (i) the individual gap between the frequency of watching movies in the cinema and watching movies through online platforms; (ii) the individual gap between the frequency of watching movies in the cinema and watching movies by other means (DVDs, TV, etc.); (iii) the individual gap between the frequency of listening to live music and listening to music via online platforms; (iv) the individual gap between the frequency of listening to live music and listening to music by other means (CDs, radio, etc.). Each of the four gaps are measured by the difference between the individuals’ frequency of consuming movies and music in public and private spaces. Therefore, these variables assess individuals’ propensity to consume culture in public spaces as compared to private spaces. Relying on the gaps’ measurement, instead of considering only the frequency of cultural participation forms in public spaces, allows this study to control for personal preferences for culture. In other words, it certifies that it is not other factors, such as lack of motivation or interest, that determine the individual’s decision not to attend the form of the activity that takes place in public spaces. Furthermore, the separation of online and offline activities in private spaces allows to assess differences when considering different types of cultural participation at home. For the calculation of the gap variables, the five levels of frequency of the variables of cultural participation were transformed into categorical ordered values ranging between 0 and 4 (where never = 0, less than once per month = 1, at least once per month = 2, at least once a week = 3, daily or almost daily = 4). Since the gaps are calculated by the difference between two variables of cultural participation, the gaps variables can assume nine levels of frequency, and, when transformed into categorical ordered values, they range between − 4 and 4. To account for the explanatory variable of violence, the individuals' subjective assessment of danger was measured by five degrees of fear of being hit by a stray bullet, ranging from never to always. The categorical frequencies of fear were thus recoded into a numerical ordinal scale as follows: no fear at all = 0, rarely = 1, sometimes = 2, many times = 3, always = 4. The choice of the fear of being hit by a stray bullet as the indicator representing the individuals’ perception of violence is because of its relevance in Rio de Janeiro, where previous studies have shown that this is the biggest fear of people living in the city (Rio Como Vamos, 2011). Not only this subjective measurement of violence was included in the model. An objective measurement of violence was also subsequently considered. This variable is measured by the number of violence events occurred in the favela where the respondent resides during the 30 days previous to the survey interrogation, including police operations and armed conflicts by criminal groups. When included in the model, it was divided by the population size of the corresponding favela. The distribution graphics of the dependent variables and the violence variables are represented in Fig. 1.10Fig. 1 Histogram with the distribution of the dependent and the violence variables Besides, the model includes as control variables socioeconomic and demographic characteristics, such as age, gender, education, household income, and employment. There is a consensus in the literature that socioeconomic and demographic variables are strong determinants for both fear and cultural participation. Generally, studies report that older people, women, people with low income, the unemployed and people with less formal education feel more vulnerable to crime and consequently show more fear (for a review of the literature, see Hale, 1996). Also, low income, unemployed and less educated people are associated with less cultural participation, and gender and age determine specific patterns and preferences of participation in the cultural life (see, for instance, Falk & Katz-Gerro, 2016 and Seaman, 2006). In the model, age and household income11 are included as numerical discrete variables. The education level is transformed into four dummy variables (for the categories ‘no education’ or pré-school; elementary or middle education; high school; and university, specialization or Master’s studies), which assume the value 1 if the person achieved a specific level, and zero otherwise. Gender and unemployment are also dummy variables: gender assumes the value 1 if female and 0 if male; and unemployment is equal to 1 when the respondent does not have a job, and 0 otherwise. Finally, the model also includes as control variables the Internet quality, the incentives to carry out cultural activities in the childhood, and the favela of residence among the sixteen in Maré. These variables are included to avoid endogeneity problems. Particularly in the case of the favela of residence, intrinsic territorial characteristics, such as the proximity to public equipment and better services infrastructure, might impact both the cultural participation behaviour and the perception of violence. In the model, the Internet quality is an ordinal variable that can assume values ranging between 0 and 4 (no Internet = 0, terrible = 1, bad = 2, regular = 3, good = 4, excellent = 5); the childhood incentive variable is a dummy that assumes the value 1 when the person had incentive to carry out cultural activities as a child, and 0 in the contrary case; and the favelas are incorporated as sixteen dummies, assuming the value 1 if the individual resides in the corresponding favela, and 0 otherwise. Table 3 summarizes the (weighted) descriptive statistics of the variables included in the model.12 Observations with missing values were dropped from the analysis. This amounted to less than 5% of the observations for each of the variables.Table 3 Descriptive statistics of the variables Obs Mean SD Min Max Dependent variables Live music 1209 0.745 1.216 0 4 Music on the Internet 1209 2.289 1.869 0 4 Music by other means 1209 1.935 1.887 0 4 Movies in the cinema 1209 0.463 0.883 0 4 Movies by other means 1209 1.643 1.751 0 4 Movies on the Internet 1206 1.511 1.74 0 4 Gap movies cinema—Internet 1208 − 1.541 2.025 − 4 4 Gap movies cinema—other means 1208 − 1.19 2.183 − 4 4 Gap music live—Internet 1206 − 1.046 1.674 − 4 4 Gap music live—other means 1208 − 1.182 1.863 − 4 4 Independent variables Fear of stray bullet 1201 2.53 1.603 0 4 Violence events 1211 5.635 3.274 1.634 23.595 Age 1211 43.86 16.942 18 91 No education or pré-school 1211 0.079 0.27 0 1 Elementary or middle education 1211 0.458 0.498 0 1 High school 1211 0.403 0.491 0 1 University, specialization or Master’s degree 1211 0.057 0.232 0 1 Female 1211 0.614 0.487 0 1 Household income 1112 1801.259 1091.485 0 5000 Childhood incentive 1200 0.308 0.462 0 1 Unemployment 1210 0.441 0.497 0 1 Internet quality 1205 2.417 1.591 0 5 Favela Baixa do Sapateiro 1211 0.038 0.191 0 1 Favela Conjunto Bento Ribeiro Dantas 1211 0.016 0.124 0 1 Favela Conjunto Esperança 1211 0.019 0.137 0 1 Favela Conjunto Pinheiros 1211 0.035 0.183 0 1 Favela Marcilio Dias 1211 0.107 0.31 0 1 Favela Morro do Timbau 1211 0.042 0.201 0 1 Favela Nova Holanda 1211 0.088 0.283 0 1 Favela Nova Maré 1211 0.012 0.107 0 1 Favela Parque Maré 1211 0.073 0.26 0 1 Favela Parque Roquete Pinto 1211 0.162 0.368 0 1 Favela Parque Rubens Vaz 1211 0.044 0.205 0 1 Favela Parque União 1211 0.131 0.338 0 1 Favela Praia de Ramos 1211 0.065 0.247 0 1 Favela Salsa e Merengue 1211 0.016 0.124 0 1 Favela Vila do João 1211 0.079 0.27 0 1 Favela Vila do Pinheiros 1211 0.074 0.262 0 1 The estimation strategy The econometric model aims at analysing the effect of violence on the individuals’ choice between cultural participation in public or private spaces. The econometric strategy chosen is the simultaneous bivariate ordered probit model. In the case of this study, the ordered probit model is more suitable than conventional regression procedures (e.g., ordinary least squares) because the dependent variables are discrete ordered values, ranging between limited lower and upper boundaries (Greene, 2002). The variables are discrete ordered values because they were constructed based on the frequency of practice of the two cultural activities. The choice of a simultaneous bivariate econometric model is to analyse the joint determination of two dependent variables with correlated disturbances (Greene, 2002). In this case, it was necessary to analyse simultaneously the determination of the choice of the cultural outgoing activity of reference (cinema or live concerts) over doing the activity both through the Internet and by other means. The model is built around latent variables Y∗, which are unobserved, but can be associated to observed discrete dependent variables Y (Wooldridge, 2012). In this case, the latent variables are: (i) the gap between watching movies in the cinema and through the Internet; (ii) the gap between watching movies in the cinema and by other means; (iii) the gap between listening to live music and through the Internet; and (iv) the gap between listening to live music and by other means. The observed dependent variables Y are available in discrete form, with nine possible values, ranging from − 4 (no participation in public spaces = 0, and daily or almost daily participation in private spaces = 4) to 4 (daily or almost daily participation in public spaces = 4, and no participation in private spaces = 0). The values that the observed variable can assume result from different combinations of participation frequencies in public and private spaces. For instance, when the observable variable is equal to 3, it can result of two different combinations: daily participation in public spaces and less than once per month participation in private spaces (Yi=4-1), or at least once a week in public spaces and no participation in private spaces (Yi=3-0). The correspondence between the latent variable and the observed dependent variable is expressed as follows:Yi-4if-∞<Yi∗≤y1-3ify1<Yi∗≤y2-2ify2<Yi∗≤y3-1ify3<Yi∗≤y40ify4<Yi∗≤y51ify5<Yi∗≤y62ify6<Yi∗≤y73ify7<Yi∗≤y84ify8<Yi∗≤+∞ The dependent variable is thus a function of the latent variable Y∗, which is assumed to depend linearly on a set of personal, socioeconomic and demographic variables X, and on the continuous explanatory variable of fear Z. Thus, the dependent variable is determined as such:1 Yi(Yi∗)=∑n=1NβnXni+αZi+εi∴εi∼N0,1, where i is the individual, β is the vector of coefficients to be estimated for the set of personal, socioeconomic and demographic variables, α is the coefficient to be estimated for the explanatory variable fear of violence, and ε is the error term, which is assumed to be normally and identically distributed with a mean of zero and variance normalized to one. When applying the econometric model, the sample was reduced to consider only the individuals who do at least one of the three possible modalities of participating in each of the practices (in public spaces, in private spaces through Internet or in private spaces by other means). Thus, when analysing watching movies, the sample was reduced to 761 individuals, considering only those who watch movies at least “less than once per month” either in the cinema, or through Internet, or by other means. When analysing listening to music, the sample was reduced to 953 individuals, who listen to music at least “less than once per month” either live, or through Internet, or by other means. In other words, individuals who do not consume movies or music at all (in public or private spaces) are dropped from the regression. In this way, the study can deal with two problems. First, it avoids a zero-inflated regression model, which happens because of a large number of zero-valued observations (Heilbron, 1994). In addition, it prevents people who have no interest in the cultural practice (whether listening to music or watching movies) from being considered equal to those who participate with the same frequency in the modalities that take place in public and private spaces, given that both cases would result in a zero-valued observation. In other words, by eliminating the non-participants in any modality, the analysis focuses only on individuals who show interest for the corresponding cultural practice.13 Results Descriptive statistics Among the weighted sample, 13.3% of the people declared not to be afraid of being hit by a stray bullet, while 9.2% rarely fear it, 14.8% fear it sometimes, 12.4% fear it many times, and 50.2% fear it every day. This represents a total of 86.7% fearing it at least partially, which shows the relevance of the fear of being hit by a stray bullet among the residents of Maré. Although relatively more widespread in Maré, it is worth noting that the fear of being hit by a stray bullet is not a particularity of this territory. Contrarily, it affects the whole state of Rio de Janeiro, where 80% of the people are afraid of being hit by a stray bullet, against 66% of the people living in São Paulo (Forum Brasileiro de Segurança Pública, 2022). Concerning the cultural practices of interest for this study, attendance in public spaces is significantly smaller than participation in private spaces for both movies and music. Besides, among those participating in cultural practices in private spaces, the majority do it on a daily base (or almost daily). While 71.3% of the interrogated declared not having gone to the cinema in the three months preceding the interview, the non-participants corresponded only to 46.5% and 47.2% of the people for watching movies in the Internet and by other means respectively.14 Attending the cinema was done by 7.2% of the respondents less than once per month, by 17.7% at least once per month, by 2.9% at least once a week, and by 0.9% daily or almost. Watching movies on the Internet was done by 1.9% of the people less than once per month, by 6.8% at least once per month, by 18.3% at least once a week, and by 26.4% daily or almost. Watching movies by other means was done by 2.2% less than once per month, by 6.5% at least once per month, by 18.9% at least once a week, and by 25.1% daily or almost. Analogously, 65.4% of the respondents did not attend live music concerts, far above the 30.3% and 49.4% who did not listen to music on the Internet or by other means respectively.15 Listening to live music was done by 4.6% of the people less than once per month, by 14.2% at least once per month, by 12.9% at least once a week, and by 2.9% daily or almost. Among those listening to music on the Internet and by other means, the majority do it daily or almost. This corresponded, respectively, to 54.6% and 37.3% of the interviewed. Listening to music on the Internet was also done by 0.6% less than once per month, by 1.9% at least once per month, by 12.6% at least once a week. Listening to music by other means was also done by 2.0% less than once per month, by 2.4% at least once per month, and by 9.0% at least once a week. Moreover, it is possible to notice that practices in public spaces and on the Internet are more common among individuals who are younger and reached higher educational levels. On the other hand, listening to music and watching movies by other means are more homogeneously practiced across the population groups. The descriptive statistics of the average frequency (ranging between 0 and 4) of cultural participation in each modality of watching movies and listening to music, by educational level, age range and household income, are displayed in Table 4.Table 4 Average frequency (between 0 and 4) of cultural participation in each modality of watching movies and listening to music, by educational level, age range and household income Live music Music Internet Music other means Cinema Movies Internet Movies other means Educational level No education/pre-school 0.769 1.003 1.914 0.229 0.392 1.028 Elementary/middle education 0.603 1.898 1.922 0.192 1.108 1.673 High school 1.002 3.31 1.723 0.817 2.444 1.862 University/specialization/Master 1.237 3.598 1.824 1.244 2.414 1.71 Age range 18–29 y.o 1.076 3.572 1.4 0.948 2.619 2.019 30–49 y.o 0.898 2.815 1.883 0.481 1.795 1.752 50–65 y.o 0.453 1.38 2.357 0.236 0.846 1.502 66 y.o. or more 0.359 .225 1.993 0.067 0.223 0.875 Household income 0–1500 0.67 2.28 1.81 0.39 1.30 1.64 1500–2000 0.71 2.72 1.91 0.55 1.77 1.76 > 2500 1.02 2.69 1.77 0.64 2.01 1.74 Econometric estimation This study aims at analysing the hypothesis that individuals’ cultural choices are affected by their fear of violence in the territory, which in this case is represented by the fear of being hit by a stray bullet. For that, two simultaneous bivariate ordered probit models were estimated, measuring if fear affects the frequency with which individuals chose to participate in cultural activities that take place in public spaces (cinema and live music) over those that can take place in private spaces (watching movies and listening to music on the Internet and by other means). People with more fear of violence are expected to choose more often cultural participation in private over public spaces (H1). Therefore, the coefficient of the variable that captures the individuals’ fear is expected to be negative. The joint regression of the dependent variables that represent the gaps between movies (music) in the cinema (live) and on the Internet, and movies (music) in the cinema (live) and by other means is proved to be necessary. The LR tests of independence of equations (see Tables 5, 6) indeed show that we can reject the null hypothesis of independence between the two equations. This corroborates the choice of the simultaneous bivariate model for the analysis of the joint determination of the dependent variables, which present correlated disturbances (Greene, 2002).Table 5 Regression results for the gap between listening to music in public and private spaces Simultaneous bivariate ordered probit for listening to music Gap music live—Internet Gap music live—by other means Age 0.0177* (0.00292) − 0.0188* (0.00292) No education/pre-school Ref Ref Elementary/middle education − 0.0717 (0.152) − 0.0608 (0.159) High school − 0.191 (0.167) − 0.0627 (0.172) University/specialization/Master − 0.279 (0.215) − 0.126 (0.215) Female − 0.0543 (0.0756) − 0.00177 (0.0747) Household income − 0.000000521 (0.0000353) 0.0000786 (0.0000346) Fear of being hit by stray bullet − 0.0509 (0.0247) − 0.0661*** (0.0244) Childhood incentive 0.00482 (0.0812) 0.152* (0.0792) Unemployed − 0.0309 (0.0775) − 0.0525 (0.0765) Internet quality − 0.283* (0.0286) 0.0839* (0.0279) Controlled for 16 favelas Yes Observations 953 LR test of indep. eqns Chi2(1) = 95.17 Prob > chi2 = 0.0000 Athrho and cuts were not reported in the tables in order to summarize the results *p < 0.01, p < 0.05, p < 0.1 Table 6 Regression results for the gaps between watching movies in public and private spaces Simultaneous bivariate ordered probit for watching movies Gap movies cinema—Internet Gap movies cinema—by other means Age 0.00776* (0.00322) − 0.00912*** (0.00325) No education/pre-school Ref Ref Elementary/middle education − 0.310 (0.201) 0.139* (0.0819) High school − 0.310 (0.213) 0.0000811** (0.0000382) University/specialization/Master 0.183 (0.247) − 0.218 (0.207) Female 0.138* (0.0816) 0.124 (0.218) Household income − 0.0000779** (0.0000382) 0.425* (0.251) Fear of being hit by stray bullet − 0.0576** (0.0270) − 0.0509* (0.0271) Childhood incentive − 0.0286 (0.0841) − 0.0433 (0.0843) Unemployed 0.0230 (0.0837) − 0.171 (0.0846) Internet quality − 0.227* (0.0321) 0.115* (0.0320) Controlled for 16 favelas Yes Observations 761 LR test of indep. eqns.: chi2 = 49.79 Prob > chi2 = 0.0000 Athrho and cuts were not reported in the tables in order to summarize the results p < 0.01, p < 0.05, p < 0.1; standard errors in parenthesis Let us consider first the estimates for the traditional socioeconomic and demographic variables. Age is significant for determining all the four dependent variables. While it positively affects the individuals’ choice of participating in cultural activities in public spaces as opposed to the Internet (significant at 1% level for music and at 5% for movies), it negatively affects participation in public spaces as opposed to other means (significant at 1% level for both music and movies). In other words, all other variables being constant, the older a person is, more she is expected to participate in culture by other means (CDs, radio, DVDs, TV, etc.) as compared to public spaces (cinema and live music), and more she is expected to participate in culture taking place in public spaces as compared to doing it on the Internet. Likewise, Internet access and quality is significant (at 1% level) for determining all dependent variables. Not surprisingly, it is negatively associated with the choice between cultural participation in public spaces and on the Internet, and positively associated with the choice between participation in public spaces and by other means. Also, household income is significant for determining most of the dependent variables (except for the gap between live music and music on the Internet). While it is positively associated with the gap between the frequency of cultural participation in public spaces and by other means (at 5% level for music and 10% level for movies), it is negatively associated with the choice of watching movies in the cinema over doing it on the Internet. Furthermore, while education is distinguished by the research literature as an important characteristic associated with greater levels of cultural participation (see, for instance, Seaman, 2006), the results indicate mostly that it is not significant for impacting individuals’ choice between modalities of participation. The only exception is that the probability of watching movies in the cinema as compared to other means increases when the individual accomplishes elementary studies or middle education (significant at 10% level) and high school (significant at 5% level). Similarly, unemployment is only significant (at 5% level) for determining the same choice. In other words, unemployed individuals are less likely to choose cinema attendance over watching movies by other means. In its turn, gender is only significant for determining the choice between going to the cinema and watching movies on the Internet, which probability of being greater increases if the respondent is a woman (significant at 10% level). Finally, the incentive to carry out cultural activities in the childhood is only significant for determining the choice between listening to live music and doing it by other means (significant at 10% level). Both simultaneous bivariate ordered probit models, for movies and music, corroborate the hypothesis 1 (H1) that the fear of stray bullet negatively impacts the choice of attending cultural events in public or private spaces.16 The results are significant at 5% level when comparing cinema and live music to watching movies and listening to music through the Internet respectively, at 10% level when comparing cinema to watching movies by other means, and at 1% level when comparing live music to listening to music by other means. Table 5 describes the econometrics results for the gap between listening to music in public spaces and in private spaces, and Table 6 the ones for the gaps between watching movies in public spaces and in private spaces. Table 7 reports the marginal effects of changes from the minimum to the maximum level of fear on the dependent variables representing the gaps between cultural participation in public and private spaces. For instance, let us consider the case of the variable Gap music live—Internet. Each of the four steps of the variation from fear of stray bullet = 0 to fear of stray bullet = 4 increases the probability of being “intensive in private spaces” (category − 4 of the gap variable) by 2.7 percentage points. This means that all other variables being constant, when the fear of violence goes from the minimum to the maximum (from 0 to 4), the probability to be a music listener intensive in private spaces increases by 4 × 2.7 = 10.8 percentage points. Likewise, when the fear of violence goes from the minimum to the maximum, the probability to be intensive in private spaces increases by 9.8 percentage points for the Gap music live—by other means, by 10.0 percentage points for the Gap movies cinema—Internet, and by 6.3 percentage points for the Gap movies cinema—by other means. Therefore, it is noteworthy that an increase in the level of fear has mainly the effect of rising cultural participation in private as opposed to public spaces.Table 7 Marginal effects for the variable that captures the fear of violence Gap music live—Internet Gap music live—by other means Gap movies cinema—Internet Gap movies cinema—by other means − 4 0.0272* 0.0245 0.0250*** 0.0157* − 3 0.0001 0.0012* 0.0075*** 0.0027* − 2 − 0.0035 − 0.0014 − 0.0056** − 0.0026* − 1 − 0.0083* − 0.0033 − 0.0089*** − 0.0031* 0 − 0.0031 − 0.0019 − 0.0023 − 0.0007 1 − 0.0021 − 0.0026 − 0.0039 − 0.0028* 2 − 0.0032 − 0.0059 − 0.0068** − 0.0068* 3 − 0.0042 − 0.0082 − 0.0029 − 0.0018 4 − 0.0030 − 0.0024** − 0.0021* − 0.0007 *p < 0.01, p < 0.05, p < 0.1 The fact that the results are similar for movies and music indicates the robustness of the analysis. It implies that the association between fear and individuals’ choice among different modalities of cultural participation is not simply a problem of supply. That is, it could be expected to find lower supply of spaces for cultural participation in parts of the territory that are more affected by violence. However, the Building Barricades mapping of cultural spaces (2019) indicates that there is no formal cinema in the whole territory of Maré and the equipment for listening to live music are more or less uniformly distributed across all the 16 favelas.17 Hence, the ability to go out and watch movies and watch live shows is the same across different areas and different crime rates in the territory of Maré. To test for endogeneity in the model, the instrumental variable method is used (Newhouse & McClellan, 1998). In this case, endogeneity could occur because of omitted variables correlated with both fear and the error term, or because of simultaneity, meaning that the dependent variables could also affect the explanatory variable of fear. Indeed, there are evidence of how social behaviour and cultural life affect individuals’ fear (De Donder et al., 2005; Liska et al., 1988). An instrumental variable (IV) must satisfy both exclusion and inclusion restrictions. The exclusion restriction stipulates that the IV should not directly impact the dependent variable, whereas the inclusion restriction means that the IV should be correlated with the possible endogenous variable. This robustness check is done by introducing as the IV an indicator of the fear that anyone close to the respondent would be hit by a stray bullet (0–4 scale). The fear for others is not expected to affect directly the dependent variables (satisfying the exclusion restriction), while it is correlated with the individuals' fear for themselves (satisfying the inclusion restriction). A simple ordered probit model with the instrument as an independent variable and the potentially endogenous variable as the dependent variable should show that both variables are highly correlated. The conduction of four simultaneous bivariate ordered probit models for the joint determination between each of the dependent variables and the variable of fear should not allow the rejection of the null hypothesis of independence between the equations. The results (presented in the Appendix 2) indicate that, for all dependent variables, the observed LR test of independence of equations indicates that the null hypothesis of independence between the two variables cannot be rejected. Therefore, the models do not seem to present any endogeneity problem and the regressions presented in Tables 5 and 6 are relevant. Finally, to test differences between subjective violence (measured by fear) and objective violence, data with the actual number of violence events in the favelas of Maré were extracted from a database provided by Redes da Maré (2019) and incorporated into the econometric model. The objective violence indicator is a value attributed to each of the sixteen favelas and associated with the respondents according to the favela where they reside (among the sixteen in Maré). It is measured by the number of violence events occurred in the favela where the respondent resides during the 30 days previous to the survey interrogation, including police operations and armed conflicts by criminal groups, divided by the population size of the corresponding favela. The regression tables including the variable of objective violence are presented in the Appendix 3. The results indicate that fear (subjective violence) affect people's behaviour and choices more than the actual occurrence of violent events (objective violence). Discussion This paper provides evidence on the way violence restricts individuals’ behaviour. It documents a statistically significant and negative effect of the fear of being hit by a stray bullet on cultural participation (watching movies and listening to music) in public spaces (as compared to the modalities in private spaces). Results also show that subjective violence (fear) is a stronger determinant of individuals’ behaviour than objective violence (its actual occurrence). This is in line with other studies: while the frequency of violent events might affect individuals’ perception of danger, the literature of behavioural economics shows that individuals’ fear and responses to risk are not proportional to the probability of victimization, and individuals’ decisions and behaviours are more likely to be based on their fear than on the actual probability of occurrence of the feared situation (Becker & Rubinstein, 2011).18 The findings of this study are important because of their territorial and economic implications, as well as the implications to the development of the cultural sector. They also fit into a broader cultural economics approach to the understanding of the determinants of inequalities in cultural participation and contribute to the formulation of evidence-based cultural policies. There are various implications linked to the effect of violence on decisions between undertaking activities in private and public spaces. At the territorial level, it represents a limit for the potential of culture and creativity to generate positive effects in terms of socioeconomic development and quality of life. Arts and culture are typically associated with safer territories, since it enriches the social environment with public amenities, induces educational effects, stimulates creativity, among others (Azevedo, 2016; Matarasso, 1997; Tubadji et al., 2015). Hence, if violence prevents cultural outgoings, it is likely that in less safe territories there will be a feedback cycle where violence diminishes cultural participation and this, in turn, leads to an even less safe environment. Furthermore, restricting cultural participation in public spaces might also reduce the socialization and well-being benefits that are typical of cultural outgoings for the inhabitants of the territory. At the economy level, the avoidance of cultural consumption in public spaces might provoke direct and indirect economic losses. Direct losses come from reduced ticket sales, while indirect losses result from less sales of goods and services that are normally consumed along the way or in a complementary way to the fruition of the cultural activity in public spaces (for instance, popcorn in cinemas or drinks in bars where live music is played). At the policy level, this paper provides evidence to inform policymaking in a context where there is usually a lack of official data and empirical research that can inform public policies. Based on the results, the intertwined nature of cultural policies with other public policies should be emphasized. While localized investment in the cultural supply side could benefit urban regeneration in favelas, it should be designed in coordination with comprehensive security and social policies. Supply side investments on cultural services and infrastructure in socioeconomically disadvantaged territories such as favelas will only be effective if coordinated with other policies to guarantee the conditions for individuals to effectively access it. Policymakers of all sectors should pay particular attention to the way that security policies are designed and implemented. Those should not be solely concentrated on reducing objective violence, whereas ignoring subjective violence. For instance, there is a risk that their focus on confronting armed groups and drug trafficking ends up strengthening or reproducing the feeling of fear in favelas. This would have a negative impact on cultural outgoings and the overall cultural sector. Another policy implication is related to the fact that the Internet could be a particularly relevant tool in contexts in which violence restricts mobility and consequently the fruition of activities that require leaving home. People can substitute activities that demand going out for cultural activities on the Internet or other more traditional means. In the precise case of cultural participation, there is a global trend of shifting from offline to online (Waldfogel, 2017). Since the Coronavirus pandemic outbreak and the social distancing that it imposed, this trend accentuated even more and digital access to culture has become more critical than ever (Radermecker, 2021). However, this study shows that different individual characteristics explain choices of substitution by each of the alternative means. In the context of Maré, the Internet does not seem to be a perfect substitute for other means yet. More specifically, the data show that the youngest, with greater educational achievements and earning higher income are the ones most frequently accessing culture online, while access by other means is almost homogenous across personal characteristics (see Table 4). Indeed, the digital divide might represent a strong obstacle for the development of online activities as alternatives (Van Dijk, 2006). Opportunities for access to and use of the Internet do not seem to be equally distributed across different demographic and socioeconomic strata of the population (see, for instance, Ateca-Amestoy & Castiglione, 2016). On top of that, the distribution of Internet access in Brazilian favelas is economically controlled by armed groups. It should be a mission of the public sector to intertwine security, technological and cultural polices to develop and guarantee the access and effective use of Internet in these territories. Conclusion This study provides evidence that the fear of being a victim (more than the actual occurrence of armed conflicts) negatively affects the frequency of cultural outgoings as compared to the modalities of cultural participation that take place in private spaces. Although several studies have examined the negative externalities of urban violence, this study contributes to filling a gap in the examination of how this affects cultural participation. The impact that violence has on cultural participation behaviour may represent an obstacle for the improvement of the quality of life in territorially disadvantaged areas through the stimulus of a creative environment, as well as a restriction for the overall development of the cultural sector in the given context. This is an issue that might similarly affect a variety of other territories in Latin America and the rest of the world. Hence, it is key to incorporate territorial aspects when analysing inequalities in cultural participation and formulating cultural policies to address them. To be effective, these policies should account for territorial specificities and be developed in harmony with other public policies, such as security and Internet dissemination policies. This study opens avenues for future research. First, to inform cultural policies, more research is needed on the association between territorial aspects and cultural participation. Prospective studies can examine the influence of other territorial features on choices of cultural participation, such as the influence of the availability and quality of public transportation or of climate conditions. Second, future studies can analyse the dimension of the negative impact of violence on cultural participation, including its consequences for the territory and the economy. This includes an assessment of the difference between cultural consumption in public and private spaces in terms of how it impacts individuals’ well-being, mental health and safety in the territory; and an assessment of the direct and indirect economic losses resulting from a reduction in cultural outgoings when people fear violence. Finally, the ability of the Internet to mitigate negative externalities linked to territorial aspects should be assessed. Given the restrictions that fear causes to mobility, Internet could be a compensation tool for certain activities. However, Internet access inequalities and the digital divide are still a worldwide reality, which is expected to impact especially socioeconomically disadvantaged territories such as favelas. Indeed, the case of Maré suggests that new technologies are not yet perfect substitutes for traditional modalities of consumption in private spaces. Future research can investigate differences in the determinants of digital cultural participation when compared to other means of cultural access in private spaces. This study is not without limitation. The econometric strategy and robustness check allowed the assessment of the causal association between an increase in the fear of violence and a rise in the probability of being more intensive in consumption in private spaces, disentangling the effects of other personal characteristics that correlate with cultural engagement. However, it is based on cross-sectional individual differences rather than temporal variation. An ideal investigation would, instead, rely on a database that collects information of the same individuals along the time, so as to enable the assessment of how variations in fear (and in the actual occurrence of violence events) provoke variations in the type of cultural participation. This would minimize measurement errors and omitted variable biases. Another limit of the analysis is that there could be inverse causality. Instead of preventing people from consuming culture in public spaces, violence could for instance prevent artists from performing. Also, having cultural equipment could be the reason why an area would offer a greater sense of security. This reasoning would imply that safer favelas would be closer to movie theaters and music venues, and less safe areas would be more distant to them. However, the Building Barricades mapping of cultural spaces (2019) indicates that cultural supply is uniform across all the favelas. There is no formal cinema in the whole territory and there are bars where people can listen to live music across all the favelas. Moreover, the econometric analysis controls for fixed effects associated to each favela, thus accounting for any supply discrepancies. Hence, this does not seem to be a pertinent problem. Also, it is important to emphasize that going out to watch movies and listen to live shows and consumption in private spaces are not perfect substitutes, and individuals are likely to have an underlying preference for one over the other. There are several reasons for people to prefer consuming culture in private or in public spaces. They may prefer to consume it at home to avoid traffic, congestion, and just in general save time. They may prefer public spaces to go out and socialize. The intention of the paper is not to capture all these reasons. It is to verify the importance of violence. The omitted variables don’t interfere the effect of the perception of violence. Therefore, the result is still valid to identify the association between violence and lower cultural consumption in public spaces. Despite its limitations, this study provides theoretical, empirical and methodological contributions to the literature, in particular to the field of cultural economics. From a methodological standpoint, this work offers an alternative for dealing with the problem of measuring what people are not doing. Indeed, a major difficulty to deal with the behavioural effects of fear in empirical sciences is the difficulty involved in measuring what people are not doing (Warr, 2000). By relying on the gaps between the frequencies of cultural consumption in public and private spaces, the empirical method proposed controls for personal preferences and avoids putting under the same umbrella those who are not attending cultural activities in public spaces because of fear and those who are not doing it because of lack of interest. From an empirical and theoretical point of view, although the impact of urban violence for individuals and the society as a whole have been studied in other fields of research, there’s a lack of analysis on the effects of spatial avoidance on cultural outgoings. This paper not only investigates this phenomena, but it does it in a type of territory that is usually overlooked by cultural economics studies. While inequalities in cultural participation have been a primary research focus in the field of cultural economics, most of the existing studies have focused on socioeconomic and demographic individual characteristics to explain these differences, including factors such as age, gender, education, income, occupation, race, and the household structure. The location where individuals live is usually only superficially considered, typically using aggregate measures indicating for instance if it is an urban or rural area. This study brings a new perspective to understand inequalities in cultural participation. By shifting the focus from individual preferences to the influence of territorial specificities, it helps understanding other factors that affect cultural consumption and interfere in the development of the cultural sector. It shows that incorporating territorial aspects into the discussion is key to better inform policies that aim at addressing inequalities in cultural participation. Appendix 1 See Tables Table 8 Estimated population older than 18 years old in Maré by gender and age groups (1st July 2019) Age group Total Male Female Total 101.549 49.435 52.114 18–29 y.o 30.603 15.186 15.417 30–49 y.o 45.911 22.911 23.000 50–65 y.o 17.750 8.418 9.332 More than 66 y.o 7.285 2.920 4.365 8 and Table 9 Sample description by age group, gender and geographical stratification Geographical strata Total Age group and gender 18–29 y.o 30–49 y.o 50–65 y.o More than 66 y.o Male Female Male Female Male Female Male Female Total 1.211 113 168 190 303 116 166 49 106 Area 1 406 32 66 68 101 36 53 14 36 area 2 400 38 50 63 107 34 63 18 27 Area 3 405 43 52 59 95 46 50 17 43 9. Appendix 2 See Tables Table 10 Simultaneous bivariate ordered probit for the variable of fear and the gap between listening to live music and listening to music by other means Simultaneous bivariate ordered probit Gap music live—by other means Fear of stray bullet Age − 0.0175** (0.00340) 0.00726* (0.00400) No education/pre-school Ref Ref Elementary/middle education 0.0163 (0.178) 0.240 (0.200) High school 0.0354 (0.197) 0.340 (0.227) University/specialization/Master − 0.0746 (0.249) 0.204 (0.291) Female − 0.0294 (0.0884) 0.151 (0.106) Household income 0.000101 (0.0000415) 0.00000719 (0.0000503) Fear of being hit by stray bullet − 0.0746 (0.0350) Fear for others 1.012*** (0.0467) Childhood incentive 0.173* (0.0941) 0.0399 (0.117) Unemployed − 0.0441 (0.0909) − 0.121 (0.108) Internet quality 0.0418 (0.0319) 0.0844 (0.0390) Controlled for 16 favelas Yes Observations 721 LR test of indep. eqns Chi2(1) = 0.09 Prob > chi2 = 0.7649 *p < 0.01, p < 0.05, p < 0.1; standard errors in parenthesis 10, Table 11 Simultaneous bivariate ordered probit for the variable of fear and the gap between listening to live music and listening to music on the Internet Simultaneous bivariate ordered probit Gap music live—Internet Fear of stray bullet Age 0.0118 (0.00355) 0.0000106 (0.00414) No education/pre-school Ref Ref Elementary/middle education − 0.274 (0.243) 0.239 (0.283) High school − 0.323 (0.251) 0.301 (0.293) University/specialization/Master − 0.346 (0.285) 0.326 (0.331) Female − 0.105 (0.0857) 0.147 (0.0994) Household income − 0.00000674 (0.0000401) − 0.00000604 (0.0000473) Fear of being hit by stray bullet − 0.0930* (0.0353) Fear for others 1.009* (0.0450) Childhood incentive 0.136 (0.0885) − 0.00629 (0.104) Unemployed − 0.0909 (0.0902) − 0.143 (0.104) Internet quality − 0.271* (0.0368) 0.0600 (0.0433) Controlled for 16 favelas Yes Observations 748 LR test of indep. eqns Chi2(1) = 0.70 Prob > chi2 = 0.4014 *p < 0.01; standard errors in parenthesis 11, Table 12 Simultaneous bivariate ordered probit for the variable of fear and the gap between watching movies in the cinema and by other means Simultaneous bivariate ordered probit Gap movies cinema—by other means Fear of stray bullet Age − 0.00833 (0.00359) 0.00391 (0.00440) No education/pre-school Ref Ref Elementary/middle education − 0.269 (0.218) 0.000242 (0.255) High school 0.152 (0.232) 0.0369 (0.276) University/specialization/Master 0.550 (0.268) 0.0564 (0.321) Female 0.189 (0.0908) 0.139 (0.109) Household income 0.0000737* (0.0000423) − 0.0000234 (0.0000517) Fear of being hit by stray bullet − 0.0322 (0.0378) Fear for others 1.039*** (0.0503) Childhood incentive − 0.0518 (0.0939) − 0.104 (0.115) Unemployed − 0.174* (0.0936) − 0.0928 (0.112) Internet quality 0.0856 (0.0345) 0.0891 (0.0417) Controlled for 16 favelas Yes Observations 641 LR test of indep. eqns Chi2(1) = 0.93 Prob > chi2 = 0.3336 *p < 0.01, p < 0.05, p < 0.1; standard errors in parenthesis 12 and Table 13 Simultaneous bivariate ordered probit for the variable of fear and the gap between watching movies in the cinema and on the Internet Simultaneous bivariate ordered probit Gap movies cinema—Internet Fear of stray bullet Age − 0.000685 (0.00413) 0.00630 (0.00502) No education/pre-school Ref Ref Elementary/middle education − 1.023 (0.302) − 0.287 (0.384) High school − 0.747 (0.303) − 0.268 (0.387) University/specialization/Master − 0.160 (0.332) − 0.196 (0.420) Female 0.213 (0.0964) 0.272 (0.116) Household income − 0.0000164 (0.0000441) − 0.00000485 (0.0000544) Fear of being hit by stray bullet − 0.0975 (0.0398) Fear for others 1.036* (0.0531) Childhood incentive 0.0636 (0.0978) − 0.0813 (0.120) Unemployed − 0.0163 (0.100) − 0.103 (0.122) Internet quality − 0.164*** (0.0469) 0.0979* (0.0564) Controlled for 16 favelas Yes Observations 556 LR test of indep. eqns Chi2(1) = 0.20 Prob > chi2 = 0.6554 *p < 0.01, p < 0.05, p < 0.1; standard errors in parenthesis 13. Appendix 3 See Tables Table 14 Regression results (with objective violence) for the gap between listening to music in public and private spaces Simultaneous bivariate ordered probit for listening to music Gap music live—Internet Gap music live—by other means Age 0.0178 (0.00293) − 0.0186* (0.00293) No education/pre-school Ref Ref Elementary/middle education − 0.0676 (0.152) − 0.0588 (0.159) High school − 0.198 (0.167) − 0.0595 (0.172) University/specialization/Master − 0.284 (0.215) − 0.117 (0.216) Female − 0.0662 (0.0762) 0.00630 (0.0753) Household income − 0.00000458 (0.0000357) 0.0000735** (0.0000350) Fear of being hit by stray bullet − 0.0455* (0.0248) − 0.0667*** (0.0245) Violence events/inhabitants − 0.201* (0.111) − 0.0308 (0.109) Childhood incentive − 0.00734 (0.0816) 0.155* (0.0795) Unemployed − 0.0428 (0.0777) − 0.0523 (0.0767) Internet quality − 0.277* (0.0288) 0.0826* (0.0281) Controlled for 16 favelas Yes Observations 945 LR test of indep. eqns Chi2(1) = 95.50 Prob > chi2 = 0.0000 *p < 0.01, p < 0.05, p < 0.1 14 and Table 15 Regression results (with objective violence) for the gaps between watching movies in public and private spaces Simultaneous bivariate ordered probit for watching movies Gap movies cinema—Internet Gap movies cinema—by other means Age 0.00805* (0.00323) − 0.00891*** (0.00326) No education/pre-school Ref Ref Elementary/middle education − 0.312 (0.201) − 0.225 (0.207) High school − 0.307 (0.213) 0.119 (0.218) University/specialization/Master 0.184 (0.247) 0.419* (0.251) Female 0.150* (0.0821) 0.148* (0.0824) Household income − 0.0000767 (0.0000386) 0.0000814 (0.0000386) Fear of being hit by stray bullet − 0.0630** (0.0272) − 0.0528* (0.0273) Violence events/inhabitants 0.126 (0.121) − 0.111 (0.121) Childhood incentive − 0.0248 (0.0844) − 0.0335 (0.0846) Unemployed 0.0195 (0.0839) − 0.172 (0.0847) Internet quality − 0.225* (0.0322) 0.115* (0.0321) Controlled for 16 favelas Yes Observations 756 LR test of indep. eqns.: chi2 = 49.91 Prob > chi2 = 0.0000 p < 0.01, p < 0.05, p < 0.1; standard errors in parenthesis 15. Acknowledgements This work is part of the multidisciplinary research project “Building Barricades”, carried out by People's Palace Projects, Redes da Maré, Queen Mary University of London, School of Social Service at UFRJ, Institute of Psychiatry at UFRJ and NECCULT, with financial support from the Economic and Social Research Council and Arts and Humanities Research Council of the United Kingdom through the Global Challenges Fund and Arts Council of England (ES/S000720/1 ESRC-AHRC GCRF Mental Health 2017). We thank Eduardo Ribeiro, Dalcio Marinho and Mauricio de Vasconcellos for their support with data cleaning and organization. We also thank Bruna Cataldo for her research assistance and Marcelo Santos Cruz, Miriam Krenzinger, Luna Arouca, Maíra Gabriel, Brenno Erick and Renata Peppl for their support and insightful comments along the duration of the “Building Barricades” project. We are also greatful to Viviane Linares da Silva, Isabele Sales dos Anjos, Giselle Moraes de Souza, Elza Sousa Silva, Jordana Farias do Espírito Santo, Maria Daiane de Araújo Alves and Maykon da Silva for their support with survey administration. Funding This study was funded by the Economic and Social Research Council and Arts and Humanities Research Council of the United Kingdom through the Global Challenges Fund and Arts Council of England (ES/S000720/1 ESRC-AHRC GCRF Mental Health 2017). Declarations Conflict of interest The authors declare that they have no conflict of interest. 1 Favela is used in the remainder of the paper to refer to Rio de Janeiro’s slums. 2 The Centre of Arts of Maré does not exhibit movies or live concerts. 3 The itinerant cinema is organized once per month in varied locations of the 16 slums of Maré. 4 These three areas are not homogeneous: they count with different housing characteristics, social dynamics, as well as access roads and public facilities. While the firsts favelas originated at the beginning of the XXth century, when the Pereira Passos reform removed the poorest population living in the city centre and in the south zone of Rio de Janeiro towards Maré, other favelas emerged more recently. For instance, during the dictatorship period, or by the removal of populations living in territories at risk of landslides and floods. 5 The introduction of drug trafficking in the 1990s divided the favela complex into areas controlled by rival armed factions (the Militia, the Comando Vermelho and the Terceiro Comando). 6 Data about Maré is from the Boletim Direito a Segurança Pública na Maré (Redes da Maré, 2019). Data about Brazil and the Other countries are from the World Population Review. 7 This survey was carried out in the context of the research project 'Building the Barricades', supported by the Economic and Social Research Council (ESRC) and the Arts and Humanities Research Council (AHRC) of the UK (ES/S000720/1 ESRC-AHRC GCRF Mental Health 2017). 8 There are no concerns that a door-to-door survey would lead to a biased sample. The sample of people at home is a random sample representative of the whole population. The door-to-door survey was conducted every day, and at different times of the day. When the selected person was not at home, the interviewer would redo the visit. The sampling process employed in the survey is stratified and selected in two stages. In the first stage, households were selected by inverse sampling (Haldane, 1945; Vasconcellos et al., 2005; Vasconcellos et al., 2013). In the second stage, an adult resident was selected with equal probability among the adult residents of the selected household. Inverse sampling is a sequential sampling procedure that aims to mitigate the non-responses observed in classic household surveys. In this process, the interviewer receives the list of addresses and visits them sequentially until reaching the number predicted or exhausting the area (stratum or unit sample) of research. For that, the research population was stratified into three geographic strata, composed of clusters of favelas in Maré, delimited according to their location, housing characteristics and social dynamics of the favelas, in addition to access to roads and common public facilities. 9 The linear trend method (Madeira & Simões, 1972) was used to estimate the population of Maré for July 1st, 2019. This was based on the population estimates by Federation Unit and simple age produced by the Brazilian Institute of Geography and Statistics (IBGE) for the years of 2010 (2010 Demographic Census, from IBGE), 2013 (year of the Maré Census) and 2019 (year of the Building Barricades survey). 10 In accordance with what is done subsequently, the non-consumers are removed from the histogram of the dependent variables. That is, only individuals who listened to music by at least one form of music consumption were considered in the graphs represented in the second line of Fig. 1, and only individuals who watched movies by at least one form of movies consumption were considered in the graphs represented in the third line of graphs of Fig. 1 11 A second analysis separated the sample by income level, doing two regressions: one for the 50% lower revenue and the other for the 50% higher revenue. 12 Due to the sampling design, a weight must be employed for the statistical analysis, so as to adjust the sampling in accordance to the gender and age of the whole population. This is calculated by multiplying the inverse of the probability of selection of an adult for a calibration factor (Silva, 2004). 13 We have checked that excluding the non-participants does not bias the results. In a configuration where the non-participants are added to the sample with a gap equal to zero, the results remain unchanged. 14 At the same time, a survey on cultural practices in 12 capital cities of different regions in Brazil found that only 32% of the people in Rio de Janeiro did not go to the cinema in the previous 12 months (and 54% of the least educated people). Despite differences between the two surveys in the methodology and the period of time analysed, it signals a significative difference between cultural outgoings in Maré and in the city of Rio de Janeiro as a whole (Cultura nas Capitais, 2017). The percentage of people in the southeast of Brazil who declared not having watched movies online in the previous 3 months is only 27%, way below the 46.5% in Maré (TIC Domincílios, 2016). 15 At the same time, a survey on cultural practices in 12 capital cities of different regions in Brazil found that only 51% of the people in Rio de Janeiro did not attend live concerts in the previous 12 months (and 69% of the least educated people). Despite differences between the two surveys in the methodology and the period of time analysed, it signals a significative difference between cultural outgoings in Maré and in the city of Rio de Janeiro as a whole (Cultura nas Capitais, 2017). The percentage of people in the southeast of Brazil who declared not having listened to music online in the previous 3 months is only 27%, below the 30.3% in Maré (TIC Domincílios, 2016). 16 The two regressions were repeated separating the sample by income level to check if there are any differences for the 50% lower income and the 50% higher income. The results remained the same for movies and music regardless of the income level. Furthermore, the results were also consistent when doing linear regression instead of the ordered probit model and when grouping together the ‘Internet’ and ‘other means’ variables into a single variable representing participation at home (in private spaces). All these regression tables can be provided by the author upon request. 17 There are bars and squares in all parts of Maré where the majority of the respondents listen to live music. Also, there is the itinerant cinema, which is informally organized once per month in different locations of the 16 slums of Maré. 18 Researchers have focused on bounded rationality and availability heuristics to explain these divergences between individuals’ judgments and the probability of occurrence of an event (Tversky and Kahneman, 1982). Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Akresh, R., & de Walque, D. (2008). Armed conflict and schooling: evidence from the 1994 Rwandan Genocide. The World Bank Policy Research Working Paper No. 4606. Ateca-Amestoy, V., & Castiglione, C. (2016). The consumption of cultural goods through the internet. How is it affected by the digital divide? Association for Cultural Economics International Working Paper No. AWP-04–2016. Azevedo, M. (2016). The evaluation of the social impacts of culture: culture, arts and development (Doctoral dissertation, Université Panthéon-Sorbonne-Paris I). Bakhshi H Cunningham S Mateos-Garcia J Jones C Lorenzen M Sapsed J Public policy for the creative industries The Oxford handbook of creative industries 2015 Oxford University Press 465 485 Baumol WJ Bowen WG Performing arts: The economic dilemma: a study of problems common to theater 1966 Opera, Music and Dance Becker, G. S., & Rubinstein, Y. (2011). Fear and the response to terrorism: an economic analysis. Centre for Economic Performance Discussion Paper No 1079. London School of Economics and Political Sciences. Bennett T Silva EB Introduction Cultural capital and Inequality: Policy issues and contexts Cultural Trends 2006 15 2–3 87 106 10.1080/09548960600712777 BenYishay, A., & Pearlman, S.F. (2013). Homicide and Work: The Impact of Mexico's Drug War on Labor Market Participation. Latin American Economics eJournal. Bourdieu P Passeron JC Les héritiers: les étudiants et la culture 1964 Minuit Briceño-León R Zubillaga V Violence and globalization in Latin America Current Sociology 2002 50 1 19 37 10.1177/0011392102050001003 Brown R Velásquez A The effect of violent crime on the human capital accumulation of young adults Journal of Development Economics 2017 127 1 12 10.1016/j.jdeveco.2017.02.004 Carboni OA Detotto C The economic consequences of crime in Italy Journal of Economic Studies 2016 43 1 122 140 10.1108/JES-07-2014-0121 Courty P Zhang F Cultural participation in major Chinese cities Journal of Cultural Economics 2018 42 4 543 592 10.1007/s10824-018-9319-3 Cullen JB Levitt SD Crime, urban flight, and the consequences for cities Review of Economics and Statistics. 1999 81 2 159 169 10.1162/003465399558030 De Almeida CCR Lima JPR Gatto MFF Expenditure on cultural events: Preferences or opportunities? An analysis of Brazilian consumer data Journal of Cultural Economics 2020 44 451 480 10.1007/s10824-019-09370-3 DeFronzo J Fear of crime and handgun ownership Criminology 1979 17 3 331 340 10.1111/j.1745-9125.1979.tb01299.x Dell M Trafficking networks and the Mexican Drug War American Economic Review 2015 105 6 1738 1779 10.1257/aer.20121637 Diniz SC Machado AF Analysis of the consumption of artistic-cultural goods and services in Brazil Journal of Cultural Economics 2011 35 1 1 18 10.1007/s10824-010-9129-8 De Donder L Verté D Messelis E Fear of crime and elderly people: Key factors that determine fear of crime among elderly people in West Flanders Ageing International 2005 30 4 363 376 10.1007/s12126-005-1021-z DuBow., F, McCabe, E., & Kaplan, G. (1979). Reactions to Crime: A Critical Review of the Literature. National Institute of Law Enforcement and Criminal Justice, U.S. Government Printing Office, Washington, D.C. Falk M Katz-Gerro T Cultural participation in Europe: Can we identify common determinants? Journal of Cultural Economics 2016 40 2 127 162 10.1007/s10824-015-9242-9 Florida R The rise of the creative class: And how it’s transforming work, leisure, community, and everyday life 2002 Basic Books Galloway S Cultural participation and individual quality of life: A review of research findings Applied Research in Quality of Life 2006 1 3–4 323 342 Galster GC Killen SP The geography of metropolitan opportunity: A reconnaissance and conceptual framework Housing Policy Debate 1995 6 1 7 43 10.1080/10511482.1995.9521180 Garofalo J The Fear of Crime: Causes and Consequences The Journal of Criminal Law and Criminology 1981 72 2 839 857 10.2307/1143018 Gould Ellen I O'Regan K Reversal of fortunes? Lower-income urban neighbourhoods in the US in the 1990s Urban Studies 2008 45 4 845 869 10.1177/0042098007088471 Greenbaum RT Tita GE The impact of violence surges on neighbourhood business activity Urban Studies 2004 41 13 2495 2514 10.1080/0042098042000294538 Greene WH Econometric analysis 2002 7 Prentice Hall Haldane JBS On a method of estimating frequencies Biometrika 1945 33 222 225 10.1093/biomet/33.3.222 21006837 Hale C Fear of crime: A review of the literature International Review of Victimology 1996 4 2 79 150 10.1177/026975809600400201 Hamermesh DS Crime and the Timing of Work Journal of Urban Economics 1999 45 2 311 330 10.1006/juec.1998.2106 Hartnagel TF The perception and fear of crime: Implications for neighborhood cohesion, social activity, and community affect Social Forces 1979 58 1 176 193 10.2307/2577792 Heilbron DC Zero-altered and other regression models for count data with added zeros Biometrical Journal 1994 36 5 531 547 10.1002/bimj.4710360505 IBGE (2010). Demographic census. Kahneman D Maps of bounded rationality: A perspective on intuitive judgment and choice Nobel Prize Lecture 2002 8 351 401 Krahn H Kennedy LW Producing personal safety: The effects of crime rates, police force size, and fear of crime Criminology 1985 23 4 697 710 10.1111/j.1745-9125.1985.tb00370.x Linden L Rockoff J Estimates of the impact of crime risk on property values from Megan’s law The American Economic Review 2008 98 3 1103 1127 10.1257/aer.98.3.1103 Liska AE Sanchirico A Reed MD Fear of crime and constrained behavior specifying and estimating a reciprocal effects model Social Forces 1988 66 3 827 837 10.2307/2579577 Lynch AK Rasmussen DW Measuring the impact of crime on house prices Applied Economics 2001 33 15 1981 1989 10.1080/00036840110021735 Madeira JL Simões CCDS Estimativas preliminares da população urbana e rural segundo as unidades da federação, de 1960/1980 por uma nova metodologia Revista Brasileira De Estatística 1972 33 129 3 11 Matarasso F Use or ornament The Social Impact of Participation in the Arts 1997 4 2 34 41 Mejía D Restrepo P Crime and conspicuous consumption Journal of Public Economics 2016 135 1 14 10.1016/j.jpubeco.2015.06.004 Mellander C Florida R Rentfrow PJ Potter J The geography of music preferences Journal of Cultural Economics 2018 42 4 593 618 10.1007/s10824-018-9320-x Monteiro J Rocha R Drug battles and school achievement: Evidence from Rio de Janeiro's favelas Review of Economics and Statistics 2017 99 2 213 228 10.1162/REST_a_00628 Moore J Poverty and access to the arts: Inequalities in arts attendance Cultural Trends 1998 8 31 53 73 10.1080/09548969809365050 Newhouse JP McClellan M Econometrics in outcomes research: The use of instrumental variables Annual Review of Public Health 1998 19 1 17 34 10.1146/annurev.publhealth.19.1.17 Orend, R. J. (1988). Socialization and Participation in the Arts. Research Division Report 21. Patterson AH Fear of crime and other barriers to use of public transportation by the elderly Journal of Architectural and Planning Research 1985 2 277 288 Pope JC Fear of crime and housing prices: Household reactions to sex offender registries Journal of Urban Economics 2008 64 3 601 614 10.1016/j.jue.2008.07.001 Radermecker ASV Art and culture in the COVID-19 era: For a consumer-oriented approach SN Business & Economics 2021 1 1 1 14 10.1007/s43546-020-00003-y Ravenscroft N Uzzell D Leach R Danger ahead? The impact of fear of crime on people's recreational use of nonmotorised shared-use routes Environment and Planning C: Government and Policy 2002 20 5 741 756 10.1068/c24m Redes da Maré (2019). Censo Populacional da Maré (2013). Rio de Janeiro: Redes da Maré. Available in: https://www.redesdamare.org.br/media/downloads/arquivos/CensoMare_WEB_04MAI.pdf Rio Como Vamos (2011). Pesquisa de percepção 2011. Available in: https://www.yumpu.com/pt/document/read/13367283/pesquisa-de-percepcao-rio-como-vamos-2011 Robles, G., Calderon, G., & Magaloni B. (2013). The Economic Consequences of Violence and Crime in Mexico, Working Paper for the Costs of Crime and Violence in Latin American and the Caribbean: Methodological Innovations and New Dimensions, IDB 2013. Rosenthal SS Ross A Violent crime, entrepreneurship, and cities Journal of Urban Economics 2010 67 1 135 149 10.1016/j.jue.2009.09.001 Ross CE Mirowsky J Neighborhood disadvantage, disorder, and health Journal of Health and Social Behavior 2001 42 3 258 276 10.2307/3090214 11668773 Schlesinger P The creative economy: Invention of a global orthodoxy Les Enjeux De L'information Et De La Communication 2016 17 2 187 205 10.3917/enic.021.0187 Seaman BA Empirical studies of demand for the performing arts Handbook of the Economics of Art and Culture 2006 1 415 472 10.1016/S1574-0676(06)01014-3 Sen A Development as Freedom 1999 Oxford University Press Shemyakina O The effect of armed conflict on accumulation of schooling: Results from Tajikistan Journal of Development Economics 2011 95 2 186 200 10.1016/j.jdeveco.2010.05.002 Silva, J. S., Fernandes, F. L., & Braga, R. W. (2008). Grupos criminosos armados com domínio de território: reflexões sobre a territorialidade do crime na Região Metropolitana do Rio de Janeiro, 16–24. In: Segurança, Tráfico e Milícias no Rio de Janeiro. Justiça Global. Silva, P. D. N. (2004). Calibration estimation: when and why, how much and how. Textos para discussão da Diretoria de Pesquisa, nº 15. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística. Skogan, W. G., & Maxfield, M. G. (1981). Coping with Crime and Neighborhood Reaction. Beverly Hills. The European Task Force on Culture and Development (1997). In from The Margins: A contribution to the debate on Culture and Development in Europe. Council of Europe Publishing. Tubadji A Osoba BJ Nijkamp P Culture-based development in the USA: Culture as a factor for economic welfare and social well-being at a county level Journal of Cultural Economics 2015 39 3 277 303 10.1007/s10824-014-9232-3 Tversky A Kahneman D Kahneman D Slovic SP Slovic P Tversky A Judgment under uncertainty: Heuristics and biases Judgment under uncertainty: Heuristics and biases 1982 Cambridge University Press Van Dijk JAGM Digital divide research, achievements and shortcomings Poetics 2006 34 4–5 221 235 10.1016/j.poetic.2006.05.004 Vasconcellos MTLD Silva PLDN Szwarcwald CL Sampling design for the world health survey in Brazil Cadernos De Saúde Pública 2005 21 89 99 10.1590/S0102-311X2005000700010 16463000 Vasconcellos MTL Silva PLN Anjos LA Sample design for the Nutrition, Physical Activity and Health Survey (PNAFS), Niterói, Rio de Janeiro Brazil. Estadística 2013 65 184 47 61 Velásquez A The Economic Burden of Crime Evidence from Mexico Journal of Human Resources 2020 55 4 1287 1318 10.3368/jhr.55.4.0716-8072R2 Waldfogel J How digitization has created a golden age of music, movies, books, and television Journal of Economic Perspectives 2017 31 3 195 214 10.1257/jep.31.3.195 Warr M Reiss AJ Jr Roth JA Public perceptions and reactions to violent offending and victimization Consequences and control, 4: Understanding and preventing violence 1994 National Academy Press Warr M Fear of crime in the United States: Avenues for research and policy Criminal Justice 2000 4 4 451 489 Winton A Urban violence: A guide to the literature Environment and Urbanization 2004 16 2 165 184 10.1177/095624780401600208 Wooldridge, J. M. (2012). Introductory econometrics: A modern approach. Fifth edition.	0	PMC9716548	NO-CC CODE	2022-12-03 23:20:58	no	J Cult Econ. 2022 Dec 2;:1-33	utf-8	null	null	null	oa_other
==== Front Lancet Reg Health Am Lancet Reg Health Am Lancet Regional Health. Americas 2667-193X Elsevier Ltd S2667-193X(22)00220-4 10.1016/j.lana.2022.100403 100403 Articles Estimated SARS-CoV-2 antibody seroprevalence trends and relationship to reported case prevalence from a repeated, cross-sectional study in the 50 states and the District of Columbia, United States—October 25, 2020–February 26, 2022 Wiegand Ryan E. a∗ Deng Yangyang b Deng Xiaoyi b Lee Adam b Meyer William A. III c Letovsky Stanley d Charles Myrna D. a Gundlapalli Adi V. a MacNeil Adam a Hall Aron J. a Thornburg Natalie J. a Jones Jefferson a Iachan Ronaldo b Clarke Kristie E.N. a a COVID-19 Response, Centers for Disease Control and Prevention, Atlanta, GA, USA b ICF Inc., Fairfax, VA, USA c Quest Diagnostics, Secaucus, NJ, USA d Laboratory Corporation of America, Burlington, NC, USA ∗ Corresponding author. Centers for Disease Control and Prevention, 1600 Clifton Road NE, MS A06, Atlanta, 30329, GA, USA. 2 12 2022 2 2023 2 12 2022 18 100403100403 19 7 2022 5 11 2022 14 11 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Sero-surveillance of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can reveal trends and differences in subgroups and capture undetected or unreported infections that are not included in case-based surveillance systems. Methods Cross-sectional, convenience samples of remnant sera from clinical laboratories from 51 U.S. jurisdictions were assayed for infection-induced SARS-CoV-2 antibodies biweekly from October 25, 2020, to July 11, 2021, and monthly from September 6, 2021, to February 26, 2022. Test results were analyzed for trends in infection-induced, nucleocapsid-protein seroprevalence using mixed effects models that adjusted for demographic variables and assay type. Findings Analyses of 1,469,792 serum specimens revealed U.S. infection-induced SARS-CoV-2 seroprevalence increased from 8.0% (95% confidence interval (CI): 7.9%–8.1%) in November 2020 to 58.2% (CI: 57.4%–58.9%) in February 2022. The U.S. ratio of the change in estimated seroprevalence to the change in reported case prevalence was 2.8 (CI: 2.8–2.9) during winter 2020–2021, 2.3 (CI: 2.0–2.5) during summer 2021, and 3.1 (CI: 3.0–3.3) during winter 2021–2022. Change in seroprevalence to change in case prevalence ratios ranged from 2.6 (CI: 2.3–2.8) to 3.5 (CI: 3.3–3.7) by region in winter 2021–2022. Interpretation Ratios of the change in seroprevalence to the change in case prevalence suggest a high proportion of infections were not detected by case-based surveillance during periods of increased transmission. The largest increases in the seroprevalence to case prevalence ratios coincided with the spread of the B.1.1.529 (Omicron) variant and with increased accessibility of home testing. Ratios varied by region and season with the highest ratios in the midwestern and southern United States during winter 2021–2022. Our results demonstrate that reported case counts did not fully capture differing underlying infection rates and demonstrate the value of sero-surveillance in understanding the full burden of infection. Levels of infection-induced antibody seroprevalence, particularly spikes during periods of increased transmission, are important to contextualize vaccine effectiveness data as the susceptibility to infection of the U.S. population changes. Funding This work was supported by the 10.13039/100000030 Centers for Disease Control and Prevention , Atlanta, Georgia. Keywords SARS-CoV-2 Seroprevalence United States COVID-19 ==== Body pmc Research in context Evidence before this study Analyses of trends in COVID-19 reported cases, deaths, emergency room visits, hospitalizations, and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections have been important to understand the progression of the pandemic. Sero-surveillance of SARS-CoV-2 antibody prevalence can provide an additional and unique measure of disease burden by assess evidence of past undetected or unreported infections. An initial SARS-CoV-2 seroprevalence study utilizing remnant serum from people seeking routine medical care was conducted in 10 US. geographic areas with known community transmission. The study estimated the cumulative number of infections to be 6 to 24 times higher than the number from reported, laboratory-confirmed COVID-19 case reports. Another nationwide study from blood donors found infection-induced seroprevalence reached approximately 20% by May 2021. Added value of this study The results of this study increase the knowledge base of SARS-CoV-2 infection and demonstrate the utility of SARS-CoV-2 sero-surveillance by describing trends in seropositivity and the ratios of changes in seroprevalence to changes in reported case prevalence, overall and among subpopulations, from October 25, 2020, through February 26, 2022, in the United States. Seroprevalence can be used to explore testing gaps, evaluate disease transmission, and identify population subgroups that are at higher risk of infection. Ratios of changes in serologically defined estimated infection rates to changes in reported case rates can add important context to the interpretation of case counts in the population and vaccine effectiveness data. Implications of all the available evidence Ratios of the change in seroprevalence to the change in reported case prevalence suggest that, during periods of increased transmission in the United States, a greater proportion of infections go unreported. The increased availability and use of self-administered viral antigen tests may lead to decreased reporting of cases. As self-testing becomes more commonplace, the U.S. CDC's updated community levels, which incorporate case counts as well as hospitalizations to measure COVID-19 impact, may become more variable; in that event, sero-surveillance will become more valuable. Sero-surveillance can more accurately characterize the infection burden, especially during periods of high transmission, and contextualize reported case counts. In addition, these sero-surveillance results highlight the utility of incorporating infection-induced immunity into vaccine effectiveness estimates, since infection-induced immunity also provides some protection against subsequent infection. Finally, these analyses demonstrate the higher infection burdens faced by certain subgroups, especially children and people in the Midwestern and southern United States. Introduction Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was first detected in the United States in late January 20201 with initial waves of coronavirus disease 2019 (COVID-19) cases in the spring and summer 2020. A third wave started in the fall 2020 and case counts reached an apex in mid-January 2021, then started declining due to the population-level effectiveness of the vaccine rollout2 and continued a downward trajectory until reaching a nadir in early July 2021. Afterwards, reported case counts have risen sharply due to the B.1.617.2 (Delta)3 variant beginning in late June 2021 and B.1.1.529 (Omicron)4 variant beginning in December 2021. COVID-19 has imposed a tremendous burden of disease in the United States with approximately 97.2 million cases and 1.1 million deaths reported through October 28, 2022.5 Monitoring the COVID-19 burden over time in the United States has often utilized reported case and death counts,2 , 5, 6, 7 emergency department visits,2 , 7 or hospital admissions.2 , 7 Conversely, seroprevalence has often been measured at a single time point8, 9, 10 and before the popularity of at-home testing.11 Sero-surveillance studies with either cohort or repeated, cross-sectional designs have been limited in their ability to assess the national burden of COVID-19 infection because they were administered in subnational geographical areas12 or sampled specific patient populations.13, 14, 15, 16 Modeling or simulation-based approaches have attempted to fill these gaps.17, 18, 19 To better understand nationwide temporal trends of SARS-CoV-2 seroprevalence and contextualize case-based surveillance data, we analyzed data from an all-ages, national, repeated, cross-sectional SARS-CoV-2 seroprevalence study.20 Remnant sera from commercial laboratory specimens from patients who had blood drawn for routine screening or clinical care were collected regularly from the 50 United States and the District of Columbia (D.C.) and tested for SARS-CoV-2 antibodies using commercially available U.S. Food and Drug Administration Emergency Use Authorized test kits. The objectives of the study were to a) examine antibody seroprevalence trends overall and in subgroups by age, sex, and urbanicity, and b) compare the change in serologically estimated infection prevalence to changes in prevalence derived from reported cases at different stages of the pandemic and by geographic region, including during case surges due to the omicron variant. Methods Study design Remnants of serum specimens submitted to clinical laboratories for routine, clinical screening, or diagnostic testing from 51 U.S. jurisdictions (50 U.S. states and D.C.) between October 25, 2020, and February 26, 2022, were examined. Until July 11, 2021, specimens were tested for SARS-CoV-2 antibodies at biweekly time periods; after a break of 56 days, specimens were tested at monthly time periods beginning September 6, 2021. If SARS-CoV-2 antibody testing was requested by the ordering clinician on the same day as the specimen was identified in the convenience sample, the specimen was excluded to reduce selection bias. Three laboratories (A, B, and C) collected specimens from five, 23, and 23 jurisdictions, respectively. All laboratories serve or can serve patients in all 50 U.S. states, D.C., and Puerto Rico. Laboratory A serves approximately 259,000 of physicians in the U.S., laboratory B serves approximately 50% of physicians and hospitals in the U.S., and laboratory C serves 64% of hospitals and 400,000 physicians in the U.S. In each jurisdiction, participating laboratories selected a convenience sample of 1300 remnant sera specimens during each biweekly testing period, divided equally among four age groups (0–17, 18–49, 50–64, and ≥65 years); for the monthly samples, 1750 samples were included per jurisdiction per month divided among five age groups (0–11, 12–17, 18–49, 50–64, and ≥65 years). Laboratories were unable to provide specimens for the following time periods and states: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. As a result, national seroprevalence estimates excluded these states in these time periods. Data collected included information on patient age, sex, state, specimen collection date, ZIP code of residence, and ordering provider ZIP code, but not race, ethnicity, or vaccination status because these were not provided by the clinical laboratories. Ethics This activity was reviewed by the U.S. Centers for Disease Control and Prevention (CDC) and was conducted consistent with applicable federal law and CDC policy.e Informed consent was waived as data were de-identified and Health Insurance Portability and Accountability Act (HIPAA)-compliant. Supporting data Case count data from CDC's COVID Data Tracker5 were used to explore trends in case counts reported by jurisdictions over time, and to assess correlations with the trends in seroprevalence. Urbanicity is defined as metro or non-metro based on the U.S. Department of Agriculture's Rural-Urban Continuum Codes (metro 1–3, non-metro 4–9).21 The social vulnerability index (SVI)22 is a county-level measure of potential negative effects on communities caused by external stresses on human health that has been associated with higher SARS-CoV-2 seroprevalence.23 In modeling, we included SVI as a continuous variable but, for demographic information (Supplementary Table S2), was divided into terciles at the 33.3rd and 66.7th percentiles and defined as low, medium, and high. Laboratory methods Laboratories performed specimen processing and transportation according to their established procedures. Laboratory A tested all specimens at a central facility, laboratory B performed testing at 12 facilities, and laboratory C performed testing at 23 sites. Specimens were tested by either an assay detecting antibodies against the nucleocapsid (N) protein (the Abbott ARCHITECT SARS-CoV-2 IgG immunoassay or the Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay) or an assay detecting antibodies against the spike (S) protein (the Ortho-Clinical Diagnostics VITROS SARS-CoV-2 IgG immunoassay). All specimens tested with the VITROS platform after December 18, 2020 (n = 17,026 samples in 11 jurisdictions) were excluded from these analyses since the VITROS anti-S platform targets only the S-protein and could conflate antibodies from infection with antibodies from vaccination. All specimens from Puerto Rico were also excluded because they were tested with the VITROS platform and we were unable to establish an anti-N seroprevalence estimate. Beginning in September 2021, all specimens were tested using the Roche Elecsys assay which targets the N-protein; anti-N antibodies are produced by the body in response to infection, but not after vaccination with vaccines approved or authorized in the United States. All three assays were granted Emergency Use Authorization by the U.S. Food and Drug Administration and were used according to manufacturer instructions. Statistical analysis A generalized linear mixed effects model24 assuming a binomial distribution with a logit link function was used to associate the serologic test result (positive or negative) with multiple covariates. The testing round, age category, sex, metro or non-metro designation, SVI, biweekly period, census region, assay type, and an interaction between round and assay type were included as fixed effects. State and county were included as random effects to account for correlation among specimens collected in the same geographic area. After models were fit, seroprevalence estimates were generated from linear combinations of the regression coefficients. Seroprevalence estimates were standardized by adjusting to the parameters of the Roche Elecsys N-target assay, and weighting the age, sex, region, SVI, and metro or non-metro distributions to the U.S. population. The Roche Elecsys assay was chosen because, among the three assays used in this study, it has been shown to yield the most stable antibody responses over multiple months.25 Models were fit in R (The Comprehensive R Archive Network, version 4.0.3) with the lme4 package.26 Where appropriate, 95% confidence intervals are shown. For a small number of specimens with missing data (Supplementary Table S1), imputation was performed for missing age, sex, U.S. County Federal Information Processing Standard (FIPS) code, and metro status data. A probabilistic method was used which imputes the missing data for each variable from the distribution of non-missing data within each jurisdiction. Records with missing SVI were excluded from the analysis. The ratio of the change in seroprevalence to the change in reported case prevalence were calculated by finding the quotient of the difference between model-estimated seroprevalence at the beginning and end of a given period and the difference in population-based cumulative case counts prevalence during the same period. Seroprevalence was taken directly from the model-based point and confidence limit estimates, while the change in cumulative case prevalence was calculated by dividing the number of reported cases by the estimated 2019 population.27 To synchronize case prevalence data with the seroprevalence time periods, we used the median date within each time interval and then lagged the reported case count by 14 days to account for the time required to develop antibodies, such that, cumulatively, the reported case prevalence is estimated for a date that is 14 days earlier than the date of the seroprevalence. Further details on the statistical methods can be found in the Supplementary Material. This ratio is henceforth referred to as the change ratio. Role of the funding source The CDC was involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication. Results A total of 1,469,792 remnant serum specimens were collected between October 25, 2020, and February 26, 2022 (Table 1 ). Women accounted for 58.9% (n = 603,571) of specimens with a range of 58.1% to 59.8% across surveys. Specimens from children aged 0–17 years made up the smallest percentage among age groups (15.0%, n = 220,272, range: 11.5%–24.6%) while people aged 18–49 years made up the largest percentage (30.9%, n = 454,756, range 24.5%–33.1%). The oldest two age groups each contributed slightly over a quarter of specimens (50–64 years: 26.6%, n = 390,491, 21.1%−29.2%; 65+ years: 27.5%, n = 404,273, 26.4%−28.8%). Prior to December 18, 2020, 32.8% (n = 74,613) of specimens were tested with the Abbott ARCHITECT assay, 9.8% (n = 22,365) with the Ortho-Clinical Diagnostics VITROS assay, and 57.4% (n = 130,684) with the Roche Elecsys assay. After December 18, 2020, 29.4% (n = 365,005) of specimens were tested with the Abbott ARCHITECT assay and 70.6% (n = 877,125) with the Roche Elecsys assay. Specimens were collected from people divided evenly across SVI categories, and most came from metro areas (Supplementary Table S2).Table 1 Demographic information and assay used for participants tested for SARS-CoV-2 antibodies for each cross-sectional survey period, 50 United States and District of Columbia, October 25, 2020–February 26, 2022. Date range Total population or samples Sex Age category Assay Male 0–17 18–49 50–64 ≥65 Abbott Architect Ortho VITROS Roche Elecsys ACS population total 322,903,030 158,984,190 (49.2) 73,553,240 (22.8) 137,062,784 (42.4) 6,3048,425 (19.5) 49,238,581 (15.3) – – – Overall 1,469,792 603,571 (41.1) 220,272 (15.0) 454,756 (30.9) 390,491 (26.6) 404,273 (27.5) 439,618 (29.9) 22,365 (1.5) 1,007,809 (68.6) Oct 25–Nov 15, 2020 63,111 25,595 (40.6) 8360 (13.2) 19,688 (31.2) 18,064 (28.6) 16,999 (26.9) 21,117 (33.5) 7364 (11.7) 34,630 (54.9) Nov 9–29, 2020 61,995 25,357 (40.9) 7285 (11.8) 20,547 (33.1) 17,813 (28.7) 16,350 (26.4) 17,854 (28.8) 4431 (7.1) 39,710 (64.1) Nov 23–Dec 12, 2020 57,859 24,226 (41.9) 7444 (12.9) 18,111 (31.3) 16,286 (28.1) 16,018 (27.7) 20,315 (35.1) 6008 (10.4) 31,536 (54.5) Dec 8–27, 2020 56,742 23,503 (41.4) 6519 (11.5) 18,452 (32.5) 16,564 (29.2) 15,207 (26.8) 18,753 (33.0) 4562 (8.0) 33,427 (58.9) Dec 22, 2020–Jan 10, 2021 50,969 21,119 (41.4) 6227 (12.2) 16,660 (32.7) 14,361 (28.2) 13,721 (26.9) 19,902 (39.0) 0 (0.0) 31,067 (61.0) Jan 4–24, 2021 55,529 22,930 (41.3) 6729 (12.1) 17,042 (30.7) 15,979 (28.8) 15,779 (28.4) 21,315 (38.4) 0 (0.0) 34,214 (61.6) Jan 19–Feb 7, 2021 57,819 24,215 (41.9) 7736 (13.4) 17,969 (31.1) 16,361 (28.3) 15,753 (27.2) 23,802 (41.2) 0 (0.0) 34,017 (58.8) Feb 1–21, 2021 59,009 24,623 (41.7) 7631 (12.9) 18,605 (31.5) 16,517 (28.0) 16,256 (27.5) 27,293 (46.3) 0 (0.0) 31,716 (53.7) Feb 15-Mar 7, 2021 59,008 24,281 (41.1) 8620 (14.6) 17,993 (30.5) 16,654 (28.2) 15,741 (26.7) 27,471 (46.6) 0 (0.0) 31,537 (53.4) Mar 1–21, 2021 61,779 25,210 (40.8) 8605 (13.9) 19,025 (30.8) 17,110 (27.7) 17,039 (27.6) 27,211 (44.0) 0 (0.0) 34,568 (56.0) Mar 15–Apr 4, 2021 60,280 24,515 (40.7) 8295 (13.8) 19,131 (31.7) 16,541 (27.4) 16,313 (27.1) 26,998 (44.8) 0 (0.0) 33,282 (55.2) Mar 29–Apr 18, 2021 60,359 24,570 (40.7) 7726 (12.8) 19,073 (31.6) 17,176 (28.5) 16,384 (27.1) 26,841 (44.5) 0 (0.0) 33,518 (55.5) Apr 12–May 2, 2021 58,767 23,816 (40.5) 7867 (13.4) 18,418 (31.3) 16,409 (27.9) 16,073 (27.4) 26,896 (45.8) 0 (0.0) 31,871 (54.2) Apr 26–May 16, 2021 60,640 24,813 (40.9) 8009 (13.2) 19,138 (31.6) 16,907 (27.9) 16,586 (27.4) 26,936 (44.4) 0 (0.0) 33,704 (55.6) May 10–30, 2021 58,501 23,805 (40.7) 7566 (12.9) 18,292 (31.3) 16,618 (28.4) 16,025 (27.4) 26,677 (45.6) 0 (0.0) 31,824 (54.4) May 24–Jun 13, 2021 59,841 24,276 (40.6) 6940 (11.6) 19,220 (32.1) 17,202 (28.7) 16,479 (27.5) 26,727 (44.7) 0 (0.0) 33,114 (55.3) Jun 7–27, 2021 62,584 25,392 (40.6) 7540 (12.0) 20,352 (32.5) 17,295 (27.6) 17,397 (27.8) 27,196 (43.5) 0 (0.0) 35,388 (56.5) Jun 21–Jul 11, 2021 58,591 23,970 (40.9) 7370 (12.6) 19,035 (32.5) 16,155 (27.6) 16,031 (27.4) 26,314 (44.9) 0 (0.0) 32,277 (55.1) Sep 6–Oct 3, 2021 63,199 26,464 (41.9) 12,728 (20.1) 19,200 (30.4) 13,942 (22.1) 17,329 (27.4) 0 (0.0) 0 (0.0) 63,199 (100) Oct 4–31, 2021 72,096 29,961 (41.6) 16,414 (22.8) 20,314 (28.2) 15,241 (21.1) 20,127 (27.9) 0 (0.0) 0 (0.0) 72,096 (100) Nov 1–28, 2021 79,649 32,860 (41.3) 16,027 (20.1) 22,538 (28.3) 18,201 (22.9) 22,883 (28.7) 0 (0.0) 0 (0.0) 79,649 (100) Nov 29–Dec 26, 2021 75,865 30,900 (40.7) 14,070 (18.5) 23,233 (30.6) 17,444 (23.0) 21,118 (27.8) 0 (0.0) 0 (0.0) 75,865 (100) Dec 27, 2021–Jan 29, 2022 70,091 28,887 (41.2) 13,376 (19.1) 21,580 (30.8) 15,585 (22.2) 19,550 (27.9) 0 (0.0) 0 (0.0) 70,091 (100) Jan 27–Feb 26, 2022 45,509 18,283 (40.2) 11,188 (24.6) 11,140 (24.5) 10,066 (22.1) 13,115 (28.8) 0 (0.0) 0 (0.0) 45,509 (100) Estimated SARS-CoV-2 infection-induced seroprevalence in October 25–November 15, 2020, was 8.0% (95% confidence interval (CI): 7.9%–8.1%) (Fig. 1 ). Over the next three months, seroprevalence increased by at least 2 percentage points between each biweekly period until January 19-February 7, 2021, when seroprevalence was estimated at 20.6% (CI: 20.4%–20.9%). From then until June 28–July 11, 2021, seroprevalence increased very slowly. National seroprevalence was approximately 25% in June–July 2021. After a pause in data collection from July 12, 2021, to September 5, 2021, seroprevalence increased to 29.3% (CI: 29.0%–29.7%) on September 6–October 3, 2021. Infection-induced seroprevalence again increased slowly during the fall and early winter 2021 and then experienced two one-month increases of over 9 percentage points to 43.7% (CI: 43.3%–44.2%) in December 27, 2021–January 29, 2022, and an additional increase of 15 percentage points to 58.2% (CI: 57.4%–58.9%) in January 27–February 26, 2022.Fig. 1 Estimated SARS-CoV-2 antibody seroprevalence in the United States, October 25, 2020–February 26, 2022. Data collected as part of a national, repeated, cross-sectional study of convenience samples of specimens from patients who sought routine screening or clinical care. Footnotes: Data were collected from the 50 United States and the District of Columbia and tested for SARS-CoV-2 antibodies using the commercially available COVID-19 test kits specified in the methods section. Regression models were used to estimate associations between specimen positivity and covariates and then were used to create seroprevalence estimates as if all specimens were tested with Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay. Laboratories were unable to provide specimens for the following time periods and states and were excluded from analyses: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. Infection-induced seroprevalence was associated with age; the youngest age group, aged 0–17 years, had the highest seroprevalence, which increased from 10.4% to 75.7% over the study period. The next highest seroprevalence was noted in those aged 18–49 years (increased from 9.2% to 64.5%) followed by people aged 50–64 years (increased from 7.0% to 49.6%) and people aged 65 years or older (increased from 4.1% to 32.7%) (Fig. 2 ). Infection-induced seroprevalence estimates had overlapping confidence intervals for males and females (Supplementary Fig. S1). Metro areas had consistently lower seroprevalence compared to non-metro areas, though the difference was 2.2 percentage points or less in every time interval (Supplementary Fig. S2). The Midwestern and southern U.S. regions had higher seroprevalence than the northeastern and western (Supplementary Fig. S3); the Midwest had the largest increase, from 9.8% to 62.6%, during the study period.Fig. 2 Estimated SARS-CoV-2 antibody seroprevalence in the United States by age category, October 25, 2020–February 26, 2022. Data collected as part of a national, repeated, cross-sectional study of convenience samples from specimens of patients who sought routine screening or clinical care. Footnotes: Data were collected from the 50 United States and the District of Columbia and tested for SARS-CoV-2 antibodies using the commercially available COVID-19 test kits specified in the methods section. Regression models were used to estimate associations between specimen positivity and covariates and then were used to create seroprevalence estimates as if all specimens were tested with Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay. Laboratories were unable to provide specimens for the following time periods and states and were excluded from analyses: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. Change ratios, ratios of the change in seroprevalence to the change in reported case prevalence, followed a convex pattern over time (Fig. 3 ). From November 4, 2020, to January 27, 2021, the ratio was 2.8 (CI: 2.8–2.9) and then decreased, reaching a low point from April 21 to July 1, 2021 (1.1, CI: 0.6–1.7). These ratios increased to 2.3 (CI: 2.0–2.5) from July 1 to September 20, 2021, held steady from September 20 to December 8, 2021 (2.2, CI: 2.0–2.5), and increased again from December 8, 2021, to February 26, 2022 (3.1, CI: 3.0–3.3).Fig. 3 Estimated change in seroprevalence to change in reported case prevalence ratios for SARS-CoV-2 in the United States, October 25, 2020–February 11, 2022. Data collected as part of a national, repeated, cross-sectional study of convenience samples of specimens of patients who sought routine screening or clinical care. Footnotes: Bars represent 95% confidence intervals. Data were collected from the 50 United States and the District of Columbia and tested for SARS-CoV-2 antibodies using the commercially available COVID-19 test kits specified in the methods section. Regression models were used to estimate associations between specimen positivity and covariates. The associations were used to create seroprevalence estimates as if all specimens were tested with Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay and then compared to case counts from CDC's COVID Data Tracker. Laboratories were unable to provide specimens for the following time periods and states and were excluded from analyses: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. Trends of change ratios differed by U.S. Census region (Fig. 4 ). The South region had the highest ratios during the winter months (3.2, CI: 3.1–3.3; and 3.5, CI:3.3−3.7) but a ratio of approximately 1.5 during all other time periods. In contrast, the Northeast region had a ratio of approximately 1.0 from January 27 to July 1, 2021, and greater than 2.5 in all other time periods measured. Change ratios varied by U.S. Census division, e.g., some divisions did not have overlapping confidence intervals (Supplementary Fig. S4). Compared to non-metro locations, metro locations had a slightly higher ratio in two time periods but otherwise the two groups had overlapping confidence intervals (Supplementary Fig. S5).Fig. 4 Estimated change in seroprevalence to change in reported case prevalence ratios by census region for SARS-CoV-2 in the United States, October 25, 2020–February 11, 2022. Data collected as part of a national, repeated, cross-sectional study of convenience samples of specimens of patients who sought routine screening or clinical care. Footnotes: Bars represent 95% confidence intervals. Data were collected from the 50 United States and the District of Columbia and tested for SARS-CoV-2 antibodies using the commercially available COVID-19 test kits specified in the methods section. Regression models were used to estimate associations between specimen positivity and covariates. The associations were used to create seroprevalence estimates as if all specimens were tested with Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay and then compared to case counts from CDC's COVID Data Tracker. Laboratories were unable to provide specimens for the following time periods and states and were excluded from analyses: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. Discussion Sero-surveillance data are an important source of data on the burden of infection, particularly during periods of increased transmission and changing testing practices. First, the change ratios, ratios estimating the change in seroprevalence compared to the change in reported case prevalence, can be used as a multiplier to enhance the understanding of the infection burden represented by officially reported case rates. Second, sudden increases in infection rates are essential to consider in vaccine efficacy studies since infection-induced immunity can also provide some protection from infection. The increasing infection-induced seroprevalence rates shown in this study demonstrate that accurate estimates of vaccine effectiveness will need to incorporate previous infection, including in pediatric populations.28 Sudden increases in infection rates impact the disease susceptibility of the unvaccinated population and are vital in the interpretation of vaccine effectiveness data. The change ratios were highest during periods of high transmission, specifically during winter case surges. While increased seroprevalence compared to case prevalence during surges are concerning, they demonstrate the improvement in access to testing compared to spring 2020, when early infection to reported case ratios ranged from 6 to 24 infections per case in metropolitan areas29 , 30 due to a shortage of diagnostic tests. The increased availability and uptake of home testing11 may be an important factor driving winter 2021–2022 increases in seroprevalence compared to case prevalence. Changes in seroprevalence may continue to increase relative to changes in reported case prevalence as home testing becomes more common and fewer jurisdictions engage in active case finding through case interviews and contact tracing. This illustrates the importance of continued sero-surveillance for understanding the true infection burden represented by officially reported case counts and other metrics; interpretation of the significance of reported case counts may require revision in the event of continued increases in this ratio. During times of lower transmission, smaller changes in seroprevalence between time periods suggest a greater proportion of infections were included in reported case counts, though with greater uncertainty. At the nadir of the ratio of the change in seroprevalence compared to the change in case prevalence in April–July 2021 (1.1; CI: 0.6–1.7), the confidence interval was widest because small changes in both seroprevalence and reported case rates resulted in a high coefficient of variation. These data from the current analysis are reinforced by findings from 2020, when jurisdictional infection to reported case ratios declined (range: 1.0–12.5) as transmission decreased during summer 2020 and testing became more widely accessible,20 though the variability in jurisdiction-level change ratios also underscore the importance of sub-national surveys to more accurately depict seroprevalence. Serosurveys can highlight population subgroups that are at higher risk of infection and help target interventions to those subgroups. For example, children had higher seroprevalence31 and have had higher infection to case ratios32 despite suggestions that seroprevalence in children might be underestimated compared with adults.33 Additionally, fewer infections per case were reported during periods of high transmission, especially in certain regions. These seroprevalence results possess some differences from prior seroprevalence estimates, which may be because convenience sampling limits the generalizability of the sample pool. Approximately 14.3% (range, 11.6%–18.5%) of the U.S. population were estimated to have been infected by mid-November 2020,34 which is slightly higher than our estimate of 10.7% from November 9–29, 2020. Our lower estimate may be due to the possibility of underestimation of seroprevalence from the available specimens obtained from clinical laboratories. People engaged in routine medical care and clinical screening likely have greater access to and utilization of healthcare resources, while people with less access are more likely to belong to racially- and ethnically-minoritised groups, disproportionately affected by chronic conditions and COVID-19.35 The lack of race and ethnicity data is especially limiting since disparities have been noted in other seroprevalence surveys.30 , 36, 37, 38 The combined bias is likely to underestimate seroprevalence and the ratio of change in seroprevalence to change in reported case prevalence, though a comparison in a diverse location revealed a higher seroprevalence estimate.39 Persons with specimens collected under routine screening for high-risk conditions may be more likely to be vaccinated than those with high-risk conditions unable to utilize healthcare resources.40 This could be a result of health insurance status and workplace sick leave policies.41 Compared to estimates from a longitudinal study of blood donors from November 2020 through May 2021,14 seroprevalence estimates in this study were consistently 1.0–4.5% higher. This could be explained by the exclusion of the pediatric age group from the blood donor estimate, higher vaccination rates in blood donors,14 or the underrepresentation in blood donor pools42 of people from racially- and ethnically-minoritised groups or other groups disproportionately affected by the COVID-19 pandemic.43 The lack of probabilistic sampling, which has been highlighted as a potential source of bias in serosurveys,44 was one of multiple limitations in this investigation. Several other limitations also may have led to an underestimation of seroprevalence, including the exclusion of specimens from people specifically seeking SARS-CoV-2 antibody testing, the inability of these sero-surveillance methods to detect reinfection (particularly during the Omicron phase45 , 46), and the potential that some fully vaccinated people who are subsequently infected may develop levels of N-antibody that fall below the assay's limit of detection.47 In addition, the likelihood of a given assay to detect antibody post-infection varies by the time since infection and the type of antibody binding.48 , 49 Although the Roche Elecsys N-target assay is less affected by antibody waning compared to other assays, antibody concentrations wane and specimens could fall below the limit of detection.25 While we were unable to control or adjust for antibody waning directly, we mitigated these effects by adjusting results to a single assay, and by performing all testing with a single assay type (Roche Elecsys) since September 2021. Nevertheless, we did not directly adjust our estimates for errors in SARS-CoV-2 antibody measurement from the assays used in this study. Qualitative testing cannot quantify the level of SARS-CoV-2 antibody in the blood,50 and current FDA emergency use authorized SARS-CoV-2 antibody assays are not validated to measure a specific level of immunity or protection from SARS-CoV-2 infection.51 Thus, a SARS-CoV-2 N-target-based seroprevalence estimate does not necessarily indicate the percentage of the population susceptible or immune to SARS-CoV-2 infection or reinfection. Also, while we hoped to be able to calculate an infection-to-reported-case ratio in this manuscript, our ratios of the change in seroprevalence to the change in reported case prevalence are only an approximation. Finally, our power of 70% for detecting a 2% increase in seroprevalence may mean our study was underpowered. Nevertheless, these results provide information to more fully understand the burden of SARS-CoV-2 infection in the United States from late 2020 to early 2022, and to more accurately interpret case reporting to aid public health decision-making.52 The U.S. CDC utilizes case surveillance, hospital admissions, and staffed inpatient beds to evaluate the community-level impact of COVID-19 illness.53 These analyses highlight the importance of a multi-faceted approach to surveillance since reported case rates must be interpreted in the context of variable ratios of the change in seroprevalence to the change in reported case prevalence by geographic area and across time. Conclusions Sero-surveillance data suggest that reported case counts did not completely capture the SARS-CoV-2 infection burden in the U.S. between late 2020 and early 2022, especially during periods of high transmission. Some subgroups, such as children and people living in the South and Midwest regions, experienced a higher infection burden compared to that suggested by case-based surveillance. Sero-surveillance data can aid in the appropriate interpretation of vaccine effectiveness data, demonstrate the increasing importance of accounting for previous infection, provide a more complete picture of COVID-19 impact for community-level decision-making, and identify subgroups at higher risk for infection. With the potential for increased use of at home, viral-based testing, national sero-surveillance will become pivotal in efforts to estimate disease burden and appropriately interpret case rates over time and by geographic region. Contributors Dr. Wiegand and Dr. Clarke had full access to the data presented in the study and take responsibility for data integrity and accuracy of analysis. Study concept and design: Wiegand, Y. Deng, X. Deng, Lee, Jones, Iachan, Clarke. Acquisition, analysis, or interpretation of data: Wiegand, Y. Deng, X. Deng, Lee, Meyer, Letovsky, Jones, Iachan, Clarke. Drafting of the manuscript: Wiegand, Y. Deng, X. Deng, Lee, Jones, Iachan, Clarke. Critical revision of the manuscript for important intellectual content: all authors. Obtained funding: Gundlapalli, Charles, MacNeil, Hall, Clarke. Administrative, technical, or material support: Meyer, Letovsky, Charles, Gundlapalli, MacNeil, Hall, Jones, Clarke. Study supervision: Meyer, Letovsky, Charles, Jones, Iachan, Clarke Disclaimer: The findings and conclusions in the article are those of the authors and do not necessarily represent the views of the U.S. Centers for Disease Control and Prevention or the corporate employers of the authors. Data sharing statement A public use dataset containing a data dictionary and seroprevalence estimates for all rounds of data collection is available at https://data.cdc.gov/Laboratory-Surveillance/Nationwide-Commercial-Laboratory-Seroprevalence-Su/d2tw-32xv. Individual-level data will not be made available. Declaration of interests BioReference Laboratories, Inc., ICF Inc., Laboratory Corporation of America Holdings, and Quest Diagnostics, Inc. were awarded federal contracts from the U.S. Centers for Disease Control and Prevention (CDC) for the execution of this project. No other disclosures were reported. Appendix A Supplementary data Supplementary Material S1 Acknowledgments We thank the following members of CDC for administrative and technical support: Anna Bratcher, PhD, Elizabeth Cole-Greenblatt, JD, Elise Nycz, MHS, Lauren Peel, JD. We thank Tonja Kyle, MS, from ICF Inc. for administrative and technical support. We thank Quest Diagnostics, BioReference Laboratories, and Laboratory Corporation of America for testing specimens. From Quest Diagnostics: Sara Ansari, PhD, Scott Deschenes, Brooke Ethington, MS, MBA, Sara Peters, Caterina Powell, BS, Dianna Tate, Brian Young, AA. From BioReference Laboratories: James Weisberger, MD. From Laboratory Corporation of America: Dorothy Adcock, MD, Kelly Chun PhD, Marla Williams. These individuals were not compensated directly by CDC for their participation in this specific study. Funding: This work was supported by the 10.13039/100000030 CDC , Atlanta, Georgia. e See e.g., 45 C.F.R. part 46; 21 C.F.R. part 56; 42 U.S.C. §241(d), 5 U.S.C. §552a, 44 U.S.C. §3501 et seq. Appendix A Supplementary data related to this article can be found at https://doi.org/10.1016/j.lana.2022.100403. ==== Refs References 1 Holshue M.L. DeBolt C. Lindquist S. First case of 2019 novel coronavirus in the United States N Engl J Med 382 10 2020 929 936 32004427 2 McNamara L.A. Wiegand R.E. Burke R.M. Estimating the early impact of the US COVID-19 vaccination programme on COVID-19 cases, emergency department visits, hospital admissions, and deaths among adults aged 65 years and older: an ecological analysis of national surveillance data Lancet 399 10320 2021 152 160 34741818 3 del Rio C. Malani P.N. Omer S.B. Confronting the delta variant of SARS-CoV-2, summer 2021 JAMA 326 11 2021 1001 1002 34406361 4 CDC COVID-19 Response Team SARS-CoV-2 B.1.1.529 (omicron) variant - United States, December 1-8, 2021 MMWR Morb Mortal Wkly Rep 70 50 2021 1731 1734 34914670 5 Centers for Disease Control and Prevention CDC COVID data tracker https://covid.cdc.gov/covid-data-tracker/#datatracker-home 2022 6 Neelon B. Mutiso F. Mueller N.T. Pearce J.L. Benjamin-Neelon S.E. Spatial and temporal trends in social vulnerability and COVID-19 incidence and death rates in the United States PLoS One 16 3 2021 e0248702 7 Siegel D.A. Reses H.E. Cool A.J. Trends in COVID-19 cases, emergency department visits, and hospital admissions among children and adolescents aged 0-17 Years - United States, August 2020-August 2021 MMWR Mortal Wkly Rep 70 36 2021 1249 1254 8 Lopez L. Nguyen T. Weber G. Seroprevalence of anti-SARS-CoV-2 IgG antibodies in the staff of a public school system in the midwestern United States PLoS One 16 6 2021 e0243676 9 Rogawski McQuade E.T. Guertin K.A. Becker L. Assessment of seroprevalence of SARS-CoV-2 and risk factors associated with COVID-19 infection among outpatients in Virginia JAMA Netw Open 4 2 2021 e2035234 33555331 10 Anand S. Montez-Rath M. Han J. Estimated SARS-CoV-2 seroprevalence in US patients receiving dialysis 1 year after the beginning of the COVID-19 pandemic JAMA Netw Open 4 7 2021 e2116572 34251441 11 Rader B. Gertz A. Iuliano A.D. Use of at-home COVID-19 tests — United States, August 23, 2021–March 12, 2022 MMWR Mortal Wkly Rep 71 13 2022 10.15585/mmwr.mm7113e1 12 Malecki K. Nikodemova M. Schultz A. Population changes in seroprevalence among a statewide sample in the United States medRxiv 2021 2020.12.18.20248479 13 Fisher B.T. Sharova A. Boge C.L.K. Evolution of severe acute respiratory coronavirus virus 2 (SARS-CoV-2) seroprevalence among employees of a US academic children's hospital during coronavirus disease 2019 (COVID-19) pandemic Infect Control Hosp Epidemiol 2021 1 9 14 Jones J.M. Stone M. Sulaeman H. Estimated US infection- and vaccine-induced SARS-CoV-2 seroprevalence based on blood donations, July 2020-May 2021 JAMA 326 14 2021 1400 1409 34473201 15 Walker A.G. Sibbel S. Wade C. SARS-CoV-2 antibody seroprevalence among maintenance dialysis patients in the United States Kidney Med 3 2 2021 216 222.e1 33564746 16 Zyskind I. Rosenberg A.Z. Zimmerman J. SARS-CoV-2 seroprevalence and symptom onset in culturally linked orthodox jewish communities across multiple regions in the United States JAMA Netw Open 4 3 2021 e212816 33688968 17 Chiu W.A. Ndeffo-Mbah M.L. Using test positivity and reported case rates to estimate state-level COVID-19 prevalence and seroprevalence in the United States PLoS Comput Biol 17 9 2021 e1009374 18 Noh J. Danuser G. Estimation of the fraction of COVID-19 infected people in U.S. states and countries worldwide PLoS One 16 2 2021 e0246772 19 Reif J. Heun-Johnson H. Tysinger B. Lakdawalla D. Measuring the COVID-19 mortality burden in the United States Ann Intern Med 174 12 2021 1700 1709 34543588 20 Bajema K.L. Wiegand R.E. Cuffe K. Estimated SARS-CoV-2 seroprevalence in the US as of september 2020 JAMA Intern Med 181 4 2021 450 460 33231628 21 U.S. Department of Agriculture Economic Research Service Rural-urban Continuum codes https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/ 2013 22 Flanagan B.E. Gregory E.W. Hallisey E.J. Heitgerd J.L. Lewis B. A social vulnerability index for disaster management J Homel Secur Emerg Manag 8 1 2011 23 Li Z. Hallisey E. Lewis B. Social vulnerability and rurality associated with higher severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection–induced seroprevalence: a nationwide blood donor study—United States, July 2020–June 2021 Clin Infect Dis 75 1 2022 e133 e143 35137014 24 Breslow N.E. Clayton D.G. Approximate inference in generalized linear mixed models J Am Stat Assoc 88 421 1993 9 25 25 Takahashi S. Peluso M.J. Hakim J. SARS-CoV-2 serology across scales: a framework for unbiased seroprevalence estimation incorporating antibody kinetics and epidemic recency medRxiv 2021 2021.09.09.21263139 26 Bates D. Mächler M. Bolker B. Walker S. Fitting linear mixed-effects models using lme4 J Stat Softw 67 1 2015 1 48 27 United States Census Bureau County population totals: 2010-2019 https://www.census.gov/data/datasets/time-series/demo/popest/2010s-counties-total.html 2021 28 Hall V. Foulkes S. Insalata F. Protection against SARS-CoV-2 after covid-19 vaccination and previous infection N Engl J Med 386 13 2022 1207 1220 35172051 29 Havers F.P. Reed C. Lim T. Seroprevalence of antibodies to SARS-CoV-2 in 10 sites in the United States, March 23-May 12, 2020 JAMA Intern Med 180 12 2020 1576 1586 30 Routledge I. Epstein A. Takahashi S. Citywide serosurveillance of the initial SARS-CoV-2 outbreak in San Francisco using electronic health records Nat Commun 12 1 2021 3566 34117227 31 Clarke K.E.N. Jones J.M. Deng Y. Seroprevalence of infection-induced SARS-CoV-2 antibodies — United States, september 2021–February 2022 MMWR Morb Mortal Wkly Rep 71 17 2022 606 608 35482574 32 Couture A. Lyons B.C. Mehrotra M.L. SARS-CoV-2 seroprevalence and reported COVID-19 cases in U.S. Children, August 2020—May 2021 Open Forum Infect Dis 9 3 2022 ofac044 35198651 33 Toh Z.Q. Anderson J. Mazarakis N. Comparison of seroconversion in children and adults with mild COVID-19 JAMA Netw Open 5 3 2022 e221313 35262717 34 Angulo F.J. Finelli L. Swerdlow D.L. Estimation of US SARS-CoV-2 infections, symptomatic infections, hospitalizations, and deaths using seroprevalence surveys JAMA Netw Open 4 1 2021 e2033706 33399860 35 Tai D.B.G. Shah A. Doubeni C.A. Sia I.G. Wieland M.L. The disproportionate impact of COVID-19 on racial and ethnic minorities in the United States Clin Infect Dis 72 4 2021 703 706 32562416 36 Anand S. Montez-Rath M. Han J. Prevalence of SARS-CoV-2 antibodies in a large nationwide sample of patients on dialysis in the USA: a cross-sectional study Lancet 396 10259 2020 1335 1344 32987007 37 Feehan A. Velasco C. Fort D. Racial and workplace disparities in seroprevalence of SARS-CoV-2, Baton Rouge, Louisiana, USA Emerg Infect Dis 27 1 2021 314 33171096 38 Parrott J.C. Maleki A.N. Vassor V.E. Prevalence of SARS-CoV-2 antibodies in New York City adults, June–October 2020: a population-based survey J Infect Dis 224 2 2021 188 195 34086923 39 Bajema K.L. Dahlgren F.S. Lim T.W. Comparison of estimated SARS-CoV-2 seroprevalence through commercial laboratory residual sera testing and a community survey Clin Infect Dis 73 9 2020 e3120 e3123 40 Galanis P. Vraka I. Siskou O. Konstantakopoulou O. Katsiroumpa A. Kaitelidou D. Predictors of COVID-19 vaccination uptake and reasons for decline of vaccination: a systematic review medRxiv 2021 2021.07.28.21261261 41 Peipins L.A. Soman A. Berkowitz Z. White M.C. The lack of paid sick leave as a barrier to cancer screening and medical care-seeking: results from the National Health Interview Survey BMC Publ Health 12 1 2012 520 42 Shaz B.H. James A.B. Hillyer K.L. Schreiber G.B. Hillyer C.D. Demographic patterns of blood donors and donations in a large metropolitan area J Natl Med Assoc 103 4 2011 351 357 21805814 43 Van Dyke M.E. Mendoza M.C.B. Li W. Racial and ethnic disparities in COVID-19 incidence by age, sex, and period among Persons aged <25 Years - 16 U.S. Jurisdictions, January 1-december 31, 2020 MMWR Mortal Wkly Rep 70 11 2021 382 388 44 Shook-Sa B.E. Boyce R.M. Aiello A.E. Estimation without representation: early severe acute respiratory syndrome coronavirus 2 seroprevalence studies and the path forward J Infect Dis 222 7 2020 1086 1089 32750135 45 Pulliam J.R.C. van Schalkwyk C. Govender N. Increased risk of SARS-CoV-2 reinfection associated with emergence of the Omicron in South Africa Science 376 6593 2022 eabn4947 35289632 46 Altarawneh H.N. Chemaitelly H. Hasan M.R. Protection against the omicron variant from previous SARS-CoV-2 infection N Engl J Med 386 13 2022 1288 1290 35139269 47 Whitaker H.J. Gower C. Otter A.D. Nucleocapsid antibody positivity as a marker of past SARS-CoV-2 infection in population serosurveillance studies: impact of variant, vaccination, and choice of assay cut-off medRxiv 2021 2021.10.25.21264964 48 Alfego D. Sullivan A. Poirier B. Williams J. Adcock D. Letovsky S. A population-based analysis of the longevity of SARS-CoV-2 antibody seropositivity in the United States EClinicalMedicine 36 2021 100902 49 Di Germanio C. Simmons G. Kelly K. SARS-CoV-2 antibody persistence in COVID-19 convalescent plasma donors: dependency on assay format and applicability to serosurveillance Transfusion 2021 50 Gundlapalli A.V. Salerno R.M. Brooks J.T. SARS-CoV-2 serologic assay needs for the next phase of the US COVID-19 pandemic response Open Forum Infect Dis 8 1 2021 ofaa555 33442555 51 U.S. Food & Drug Administration Antibody testing is not currently recommended to assess immunity after COVID-19 vaccination: FDA safety communication https://www.fda.gov/medical-devices/safety-communications/antibody-testing-not-currently-recommended-assess-immunity-after-covid-19-vaccination-fda-safety 5/19/2021 2021 52 Murhekar M.V. Clapham H. COVID-19 serosurveys for public health decision making Lancet Global Health 9 5 2021 e559 e560 33705691 53 Centers for Disease Control and Prevention COVID-19 community levels March 17 2022 2022 https://www.cdc.gov/coronavirus/2019-ncov/science/community-levels.html	36479424	PMC9716971	NO-CC CODE	2022-12-03 23:21:02	no	Lancet Reg Health Am. 2023 Feb 2; 18:100403	utf-8	Lancet Reg Health Am	2,022	10.1016/j.lana.2022.100403	oa_other
==== Front Lancet Reg Health Am Lancet Reg Health Am Lancet Regional Health. Americas 2667-193X Elsevier Ltd S2667-193X(22)00220-4 10.1016/j.lana.2022.100403 100403 Articles Estimated SARS-CoV-2 antibody seroprevalence trends and relationship to reported case prevalence from a repeated, cross-sectional study in the 50 states and the District of Columbia, United States—October 25, 2020–February 26, 2022 Wiegand Ryan E. a∗ Deng Yangyang b Deng Xiaoyi b Lee Adam b Meyer William A. III c Letovsky Stanley d Charles Myrna D. a Gundlapalli Adi V. a MacNeil Adam a Hall Aron J. a Thornburg Natalie J. a Jones Jefferson a Iachan Ronaldo b Clarke Kristie E.N. a a COVID-19 Response, Centers for Disease Control and Prevention, Atlanta, GA, USA b ICF Inc., Fairfax, VA, USA c Quest Diagnostics, Secaucus, NJ, USA d Laboratory Corporation of America, Burlington, NC, USA ∗ Corresponding author. Centers for Disease Control and Prevention, 1600 Clifton Road NE, MS A06, Atlanta, 30329, GA, USA. 2 12 2022 2 2023 2 12 2022 18 100403100403 19 7 2022 5 11 2022 14 11 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Sero-surveillance of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can reveal trends and differences in subgroups and capture undetected or unreported infections that are not included in case-based surveillance systems. Methods Cross-sectional, convenience samples of remnant sera from clinical laboratories from 51 U.S. jurisdictions were assayed for infection-induced SARS-CoV-2 antibodies biweekly from October 25, 2020, to July 11, 2021, and monthly from September 6, 2021, to February 26, 2022. Test results were analyzed for trends in infection-induced, nucleocapsid-protein seroprevalence using mixed effects models that adjusted for demographic variables and assay type. Findings Analyses of 1,469,792 serum specimens revealed U.S. infection-induced SARS-CoV-2 seroprevalence increased from 8.0% (95% confidence interval (CI): 7.9%–8.1%) in November 2020 to 58.2% (CI: 57.4%–58.9%) in February 2022. The U.S. ratio of the change in estimated seroprevalence to the change in reported case prevalence was 2.8 (CI: 2.8–2.9) during winter 2020–2021, 2.3 (CI: 2.0–2.5) during summer 2021, and 3.1 (CI: 3.0–3.3) during winter 2021–2022. Change in seroprevalence to change in case prevalence ratios ranged from 2.6 (CI: 2.3–2.8) to 3.5 (CI: 3.3–3.7) by region in winter 2021–2022. Interpretation Ratios of the change in seroprevalence to the change in case prevalence suggest a high proportion of infections were not detected by case-based surveillance during periods of increased transmission. The largest increases in the seroprevalence to case prevalence ratios coincided with the spread of the B.1.1.529 (Omicron) variant and with increased accessibility of home testing. Ratios varied by region and season with the highest ratios in the midwestern and southern United States during winter 2021–2022. Our results demonstrate that reported case counts did not fully capture differing underlying infection rates and demonstrate the value of sero-surveillance in understanding the full burden of infection. Levels of infection-induced antibody seroprevalence, particularly spikes during periods of increased transmission, are important to contextualize vaccine effectiveness data as the susceptibility to infection of the U.S. population changes. Funding This work was supported by the 10.13039/100000030 Centers for Disease Control and Prevention , Atlanta, Georgia. Keywords SARS-CoV-2 Seroprevalence United States COVID-19 ==== Body pmc Research in context Evidence before this study Analyses of trends in COVID-19 reported cases, deaths, emergency room visits, hospitalizations, and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections have been important to understand the progression of the pandemic. Sero-surveillance of SARS-CoV-2 antibody prevalence can provide an additional and unique measure of disease burden by assess evidence of past undetected or unreported infections. An initial SARS-CoV-2 seroprevalence study utilizing remnant serum from people seeking routine medical care was conducted in 10 US. geographic areas with known community transmission. The study estimated the cumulative number of infections to be 6 to 24 times higher than the number from reported, laboratory-confirmed COVID-19 case reports. Another nationwide study from blood donors found infection-induced seroprevalence reached approximately 20% by May 2021. Added value of this study The results of this study increase the knowledge base of SARS-CoV-2 infection and demonstrate the utility of SARS-CoV-2 sero-surveillance by describing trends in seropositivity and the ratios of changes in seroprevalence to changes in reported case prevalence, overall and among subpopulations, from October 25, 2020, through February 26, 2022, in the United States. Seroprevalence can be used to explore testing gaps, evaluate disease transmission, and identify population subgroups that are at higher risk of infection. Ratios of changes in serologically defined estimated infection rates to changes in reported case rates can add important context to the interpretation of case counts in the population and vaccine effectiveness data. Implications of all the available evidence Ratios of the change in seroprevalence to the change in reported case prevalence suggest that, during periods of increased transmission in the United States, a greater proportion of infections go unreported. The increased availability and use of self-administered viral antigen tests may lead to decreased reporting of cases. As self-testing becomes more commonplace, the U.S. CDC's updated community levels, which incorporate case counts as well as hospitalizations to measure COVID-19 impact, may become more variable; in that event, sero-surveillance will become more valuable. Sero-surveillance can more accurately characterize the infection burden, especially during periods of high transmission, and contextualize reported case counts. In addition, these sero-surveillance results highlight the utility of incorporating infection-induced immunity into vaccine effectiveness estimates, since infection-induced immunity also provides some protection against subsequent infection. Finally, these analyses demonstrate the higher infection burdens faced by certain subgroups, especially children and people in the Midwestern and southern United States. Introduction Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was first detected in the United States in late January 20201 with initial waves of coronavirus disease 2019 (COVID-19) cases in the spring and summer 2020. A third wave started in the fall 2020 and case counts reached an apex in mid-January 2021, then started declining due to the population-level effectiveness of the vaccine rollout2 and continued a downward trajectory until reaching a nadir in early July 2021. Afterwards, reported case counts have risen sharply due to the B.1.617.2 (Delta)3 variant beginning in late June 2021 and B.1.1.529 (Omicron)4 variant beginning in December 2021. COVID-19 has imposed a tremendous burden of disease in the United States with approximately 97.2 million cases and 1.1 million deaths reported through October 28, 2022.5 Monitoring the COVID-19 burden over time in the United States has often utilized reported case and death counts,2 , 5, 6, 7 emergency department visits,2 , 7 or hospital admissions.2 , 7 Conversely, seroprevalence has often been measured at a single time point8, 9, 10 and before the popularity of at-home testing.11 Sero-surveillance studies with either cohort or repeated, cross-sectional designs have been limited in their ability to assess the national burden of COVID-19 infection because they were administered in subnational geographical areas12 or sampled specific patient populations.13, 14, 15, 16 Modeling or simulation-based approaches have attempted to fill these gaps.17, 18, 19 To better understand nationwide temporal trends of SARS-CoV-2 seroprevalence and contextualize case-based surveillance data, we analyzed data from an all-ages, national, repeated, cross-sectional SARS-CoV-2 seroprevalence study.20 Remnant sera from commercial laboratory specimens from patients who had blood drawn for routine screening or clinical care were collected regularly from the 50 United States and the District of Columbia (D.C.) and tested for SARS-CoV-2 antibodies using commercially available U.S. Food and Drug Administration Emergency Use Authorized test kits. The objectives of the study were to a) examine antibody seroprevalence trends overall and in subgroups by age, sex, and urbanicity, and b) compare the change in serologically estimated infection prevalence to changes in prevalence derived from reported cases at different stages of the pandemic and by geographic region, including during case surges due to the omicron variant. Methods Study design Remnants of serum specimens submitted to clinical laboratories for routine, clinical screening, or diagnostic testing from 51 U.S. jurisdictions (50 U.S. states and D.C.) between October 25, 2020, and February 26, 2022, were examined. Until July 11, 2021, specimens were tested for SARS-CoV-2 antibodies at biweekly time periods; after a break of 56 days, specimens were tested at monthly time periods beginning September 6, 2021. If SARS-CoV-2 antibody testing was requested by the ordering clinician on the same day as the specimen was identified in the convenience sample, the specimen was excluded to reduce selection bias. Three laboratories (A, B, and C) collected specimens from five, 23, and 23 jurisdictions, respectively. All laboratories serve or can serve patients in all 50 U.S. states, D.C., and Puerto Rico. Laboratory A serves approximately 259,000 of physicians in the U.S., laboratory B serves approximately 50% of physicians and hospitals in the U.S., and laboratory C serves 64% of hospitals and 400,000 physicians in the U.S. In each jurisdiction, participating laboratories selected a convenience sample of 1300 remnant sera specimens during each biweekly testing period, divided equally among four age groups (0–17, 18–49, 50–64, and ≥65 years); for the monthly samples, 1750 samples were included per jurisdiction per month divided among five age groups (0–11, 12–17, 18–49, 50–64, and ≥65 years). Laboratories were unable to provide specimens for the following time periods and states: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. As a result, national seroprevalence estimates excluded these states in these time periods. Data collected included information on patient age, sex, state, specimen collection date, ZIP code of residence, and ordering provider ZIP code, but not race, ethnicity, or vaccination status because these were not provided by the clinical laboratories. Ethics This activity was reviewed by the U.S. Centers for Disease Control and Prevention (CDC) and was conducted consistent with applicable federal law and CDC policy.e Informed consent was waived as data were de-identified and Health Insurance Portability and Accountability Act (HIPAA)-compliant. Supporting data Case count data from CDC's COVID Data Tracker5 were used to explore trends in case counts reported by jurisdictions over time, and to assess correlations with the trends in seroprevalence. Urbanicity is defined as metro or non-metro based on the U.S. Department of Agriculture's Rural-Urban Continuum Codes (metro 1–3, non-metro 4–9).21 The social vulnerability index (SVI)22 is a county-level measure of potential negative effects on communities caused by external stresses on human health that has been associated with higher SARS-CoV-2 seroprevalence.23 In modeling, we included SVI as a continuous variable but, for demographic information (Supplementary Table S2), was divided into terciles at the 33.3rd and 66.7th percentiles and defined as low, medium, and high. Laboratory methods Laboratories performed specimen processing and transportation according to their established procedures. Laboratory A tested all specimens at a central facility, laboratory B performed testing at 12 facilities, and laboratory C performed testing at 23 sites. Specimens were tested by either an assay detecting antibodies against the nucleocapsid (N) protein (the Abbott ARCHITECT SARS-CoV-2 IgG immunoassay or the Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay) or an assay detecting antibodies against the spike (S) protein (the Ortho-Clinical Diagnostics VITROS SARS-CoV-2 IgG immunoassay). All specimens tested with the VITROS platform after December 18, 2020 (n = 17,026 samples in 11 jurisdictions) were excluded from these analyses since the VITROS anti-S platform targets only the S-protein and could conflate antibodies from infection with antibodies from vaccination. All specimens from Puerto Rico were also excluded because they were tested with the VITROS platform and we were unable to establish an anti-N seroprevalence estimate. Beginning in September 2021, all specimens were tested using the Roche Elecsys assay which targets the N-protein; anti-N antibodies are produced by the body in response to infection, but not after vaccination with vaccines approved or authorized in the United States. All three assays were granted Emergency Use Authorization by the U.S. Food and Drug Administration and were used according to manufacturer instructions. Statistical analysis A generalized linear mixed effects model24 assuming a binomial distribution with a logit link function was used to associate the serologic test result (positive or negative) with multiple covariates. The testing round, age category, sex, metro or non-metro designation, SVI, biweekly period, census region, assay type, and an interaction between round and assay type were included as fixed effects. State and county were included as random effects to account for correlation among specimens collected in the same geographic area. After models were fit, seroprevalence estimates were generated from linear combinations of the regression coefficients. Seroprevalence estimates were standardized by adjusting to the parameters of the Roche Elecsys N-target assay, and weighting the age, sex, region, SVI, and metro or non-metro distributions to the U.S. population. The Roche Elecsys assay was chosen because, among the three assays used in this study, it has been shown to yield the most stable antibody responses over multiple months.25 Models were fit in R (The Comprehensive R Archive Network, version 4.0.3) with the lme4 package.26 Where appropriate, 95% confidence intervals are shown. For a small number of specimens with missing data (Supplementary Table S1), imputation was performed for missing age, sex, U.S. County Federal Information Processing Standard (FIPS) code, and metro status data. A probabilistic method was used which imputes the missing data for each variable from the distribution of non-missing data within each jurisdiction. Records with missing SVI were excluded from the analysis. The ratio of the change in seroprevalence to the change in reported case prevalence were calculated by finding the quotient of the difference between model-estimated seroprevalence at the beginning and end of a given period and the difference in population-based cumulative case counts prevalence during the same period. Seroprevalence was taken directly from the model-based point and confidence limit estimates, while the change in cumulative case prevalence was calculated by dividing the number of reported cases by the estimated 2019 population.27 To synchronize case prevalence data with the seroprevalence time periods, we used the median date within each time interval and then lagged the reported case count by 14 days to account for the time required to develop antibodies, such that, cumulatively, the reported case prevalence is estimated for a date that is 14 days earlier than the date of the seroprevalence. Further details on the statistical methods can be found in the Supplementary Material. This ratio is henceforth referred to as the change ratio. Role of the funding source The CDC was involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication. Results A total of 1,469,792 remnant serum specimens were collected between October 25, 2020, and February 26, 2022 (Table 1 ). Women accounted for 58.9% (n = 603,571) of specimens with a range of 58.1% to 59.8% across surveys. Specimens from children aged 0–17 years made up the smallest percentage among age groups (15.0%, n = 220,272, range: 11.5%–24.6%) while people aged 18–49 years made up the largest percentage (30.9%, n = 454,756, range 24.5%–33.1%). The oldest two age groups each contributed slightly over a quarter of specimens (50–64 years: 26.6%, n = 390,491, 21.1%−29.2%; 65+ years: 27.5%, n = 404,273, 26.4%−28.8%). Prior to December 18, 2020, 32.8% (n = 74,613) of specimens were tested with the Abbott ARCHITECT assay, 9.8% (n = 22,365) with the Ortho-Clinical Diagnostics VITROS assay, and 57.4% (n = 130,684) with the Roche Elecsys assay. After December 18, 2020, 29.4% (n = 365,005) of specimens were tested with the Abbott ARCHITECT assay and 70.6% (n = 877,125) with the Roche Elecsys assay. Specimens were collected from people divided evenly across SVI categories, and most came from metro areas (Supplementary Table S2).Table 1 Demographic information and assay used for participants tested for SARS-CoV-2 antibodies for each cross-sectional survey period, 50 United States and District of Columbia, October 25, 2020–February 26, 2022. Date range Total population or samples Sex Age category Assay Male 0–17 18–49 50–64 ≥65 Abbott Architect Ortho VITROS Roche Elecsys ACS population total 322,903,030 158,984,190 (49.2) 73,553,240 (22.8) 137,062,784 (42.4) 6,3048,425 (19.5) 49,238,581 (15.3) – – – Overall 1,469,792 603,571 (41.1) 220,272 (15.0) 454,756 (30.9) 390,491 (26.6) 404,273 (27.5) 439,618 (29.9) 22,365 (1.5) 1,007,809 (68.6) Oct 25–Nov 15, 2020 63,111 25,595 (40.6) 8360 (13.2) 19,688 (31.2) 18,064 (28.6) 16,999 (26.9) 21,117 (33.5) 7364 (11.7) 34,630 (54.9) Nov 9–29, 2020 61,995 25,357 (40.9) 7285 (11.8) 20,547 (33.1) 17,813 (28.7) 16,350 (26.4) 17,854 (28.8) 4431 (7.1) 39,710 (64.1) Nov 23–Dec 12, 2020 57,859 24,226 (41.9) 7444 (12.9) 18,111 (31.3) 16,286 (28.1) 16,018 (27.7) 20,315 (35.1) 6008 (10.4) 31,536 (54.5) Dec 8–27, 2020 56,742 23,503 (41.4) 6519 (11.5) 18,452 (32.5) 16,564 (29.2) 15,207 (26.8) 18,753 (33.0) 4562 (8.0) 33,427 (58.9) Dec 22, 2020–Jan 10, 2021 50,969 21,119 (41.4) 6227 (12.2) 16,660 (32.7) 14,361 (28.2) 13,721 (26.9) 19,902 (39.0) 0 (0.0) 31,067 (61.0) Jan 4–24, 2021 55,529 22,930 (41.3) 6729 (12.1) 17,042 (30.7) 15,979 (28.8) 15,779 (28.4) 21,315 (38.4) 0 (0.0) 34,214 (61.6) Jan 19–Feb 7, 2021 57,819 24,215 (41.9) 7736 (13.4) 17,969 (31.1) 16,361 (28.3) 15,753 (27.2) 23,802 (41.2) 0 (0.0) 34,017 (58.8) Feb 1–21, 2021 59,009 24,623 (41.7) 7631 (12.9) 18,605 (31.5) 16,517 (28.0) 16,256 (27.5) 27,293 (46.3) 0 (0.0) 31,716 (53.7) Feb 15-Mar 7, 2021 59,008 24,281 (41.1) 8620 (14.6) 17,993 (30.5) 16,654 (28.2) 15,741 (26.7) 27,471 (46.6) 0 (0.0) 31,537 (53.4) Mar 1–21, 2021 61,779 25,210 (40.8) 8605 (13.9) 19,025 (30.8) 17,110 (27.7) 17,039 (27.6) 27,211 (44.0) 0 (0.0) 34,568 (56.0) Mar 15–Apr 4, 2021 60,280 24,515 (40.7) 8295 (13.8) 19,131 (31.7) 16,541 (27.4) 16,313 (27.1) 26,998 (44.8) 0 (0.0) 33,282 (55.2) Mar 29–Apr 18, 2021 60,359 24,570 (40.7) 7726 (12.8) 19,073 (31.6) 17,176 (28.5) 16,384 (27.1) 26,841 (44.5) 0 (0.0) 33,518 (55.5) Apr 12–May 2, 2021 58,767 23,816 (40.5) 7867 (13.4) 18,418 (31.3) 16,409 (27.9) 16,073 (27.4) 26,896 (45.8) 0 (0.0) 31,871 (54.2) Apr 26–May 16, 2021 60,640 24,813 (40.9) 8009 (13.2) 19,138 (31.6) 16,907 (27.9) 16,586 (27.4) 26,936 (44.4) 0 (0.0) 33,704 (55.6) May 10–30, 2021 58,501 23,805 (40.7) 7566 (12.9) 18,292 (31.3) 16,618 (28.4) 16,025 (27.4) 26,677 (45.6) 0 (0.0) 31,824 (54.4) May 24–Jun 13, 2021 59,841 24,276 (40.6) 6940 (11.6) 19,220 (32.1) 17,202 (28.7) 16,479 (27.5) 26,727 (44.7) 0 (0.0) 33,114 (55.3) Jun 7–27, 2021 62,584 25,392 (40.6) 7540 (12.0) 20,352 (32.5) 17,295 (27.6) 17,397 (27.8) 27,196 (43.5) 0 (0.0) 35,388 (56.5) Jun 21–Jul 11, 2021 58,591 23,970 (40.9) 7370 (12.6) 19,035 (32.5) 16,155 (27.6) 16,031 (27.4) 26,314 (44.9) 0 (0.0) 32,277 (55.1) Sep 6–Oct 3, 2021 63,199 26,464 (41.9) 12,728 (20.1) 19,200 (30.4) 13,942 (22.1) 17,329 (27.4) 0 (0.0) 0 (0.0) 63,199 (100) Oct 4–31, 2021 72,096 29,961 (41.6) 16,414 (22.8) 20,314 (28.2) 15,241 (21.1) 20,127 (27.9) 0 (0.0) 0 (0.0) 72,096 (100) Nov 1–28, 2021 79,649 32,860 (41.3) 16,027 (20.1) 22,538 (28.3) 18,201 (22.9) 22,883 (28.7) 0 (0.0) 0 (0.0) 79,649 (100) Nov 29–Dec 26, 2021 75,865 30,900 (40.7) 14,070 (18.5) 23,233 (30.6) 17,444 (23.0) 21,118 (27.8) 0 (0.0) 0 (0.0) 75,865 (100) Dec 27, 2021–Jan 29, 2022 70,091 28,887 (41.2) 13,376 (19.1) 21,580 (30.8) 15,585 (22.2) 19,550 (27.9) 0 (0.0) 0 (0.0) 70,091 (100) Jan 27–Feb 26, 2022 45,509 18,283 (40.2) 11,188 (24.6) 11,140 (24.5) 10,066 (22.1) 13,115 (28.8) 0 (0.0) 0 (0.0) 45,509 (100) Estimated SARS-CoV-2 infection-induced seroprevalence in October 25–November 15, 2020, was 8.0% (95% confidence interval (CI): 7.9%–8.1%) (Fig. 1 ). Over the next three months, seroprevalence increased by at least 2 percentage points between each biweekly period until January 19-February 7, 2021, when seroprevalence was estimated at 20.6% (CI: 20.4%–20.9%). From then until June 28–July 11, 2021, seroprevalence increased very slowly. National seroprevalence was approximately 25% in June–July 2021. After a pause in data collection from July 12, 2021, to September 5, 2021, seroprevalence increased to 29.3% (CI: 29.0%–29.7%) on September 6–October 3, 2021. Infection-induced seroprevalence again increased slowly during the fall and early winter 2021 and then experienced two one-month increases of over 9 percentage points to 43.7% (CI: 43.3%–44.2%) in December 27, 2021–January 29, 2022, and an additional increase of 15 percentage points to 58.2% (CI: 57.4%–58.9%) in January 27–February 26, 2022.Fig. 1 Estimated SARS-CoV-2 antibody seroprevalence in the United States, October 25, 2020–February 26, 2022. Data collected as part of a national, repeated, cross-sectional study of convenience samples of specimens from patients who sought routine screening or clinical care. Footnotes: Data were collected from the 50 United States and the District of Columbia and tested for SARS-CoV-2 antibodies using the commercially available COVID-19 test kits specified in the methods section. Regression models were used to estimate associations between specimen positivity and covariates and then were used to create seroprevalence estimates as if all specimens were tested with Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay. Laboratories were unable to provide specimens for the following time periods and states and were excluded from analyses: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. Infection-induced seroprevalence was associated with age; the youngest age group, aged 0–17 years, had the highest seroprevalence, which increased from 10.4% to 75.7% over the study period. The next highest seroprevalence was noted in those aged 18–49 years (increased from 9.2% to 64.5%) followed by people aged 50–64 years (increased from 7.0% to 49.6%) and people aged 65 years or older (increased from 4.1% to 32.7%) (Fig. 2 ). Infection-induced seroprevalence estimates had overlapping confidence intervals for males and females (Supplementary Fig. S1). Metro areas had consistently lower seroprevalence compared to non-metro areas, though the difference was 2.2 percentage points or less in every time interval (Supplementary Fig. S2). The Midwestern and southern U.S. regions had higher seroprevalence than the northeastern and western (Supplementary Fig. S3); the Midwest had the largest increase, from 9.8% to 62.6%, during the study period.Fig. 2 Estimated SARS-CoV-2 antibody seroprevalence in the United States by age category, October 25, 2020–February 26, 2022. Data collected as part of a national, repeated, cross-sectional study of convenience samples from specimens of patients who sought routine screening or clinical care. Footnotes: Data were collected from the 50 United States and the District of Columbia and tested for SARS-CoV-2 antibodies using the commercially available COVID-19 test kits specified in the methods section. Regression models were used to estimate associations between specimen positivity and covariates and then were used to create seroprevalence estimates as if all specimens were tested with Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay. Laboratories were unable to provide specimens for the following time periods and states and were excluded from analyses: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. Change ratios, ratios of the change in seroprevalence to the change in reported case prevalence, followed a convex pattern over time (Fig. 3 ). From November 4, 2020, to January 27, 2021, the ratio was 2.8 (CI: 2.8–2.9) and then decreased, reaching a low point from April 21 to July 1, 2021 (1.1, CI: 0.6–1.7). These ratios increased to 2.3 (CI: 2.0–2.5) from July 1 to September 20, 2021, held steady from September 20 to December 8, 2021 (2.2, CI: 2.0–2.5), and increased again from December 8, 2021, to February 26, 2022 (3.1, CI: 3.0–3.3).Fig. 3 Estimated change in seroprevalence to change in reported case prevalence ratios for SARS-CoV-2 in the United States, October 25, 2020–February 11, 2022. Data collected as part of a national, repeated, cross-sectional study of convenience samples of specimens of patients who sought routine screening or clinical care. Footnotes: Bars represent 95% confidence intervals. Data were collected from the 50 United States and the District of Columbia and tested for SARS-CoV-2 antibodies using the commercially available COVID-19 test kits specified in the methods section. Regression models were used to estimate associations between specimen positivity and covariates. The associations were used to create seroprevalence estimates as if all specimens were tested with Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay and then compared to case counts from CDC's COVID Data Tracker. Laboratories were unable to provide specimens for the following time periods and states and were excluded from analyses: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. Trends of change ratios differed by U.S. Census region (Fig. 4 ). The South region had the highest ratios during the winter months (3.2, CI: 3.1–3.3; and 3.5, CI:3.3−3.7) but a ratio of approximately 1.5 during all other time periods. In contrast, the Northeast region had a ratio of approximately 1.0 from January 27 to July 1, 2021, and greater than 2.5 in all other time periods measured. Change ratios varied by U.S. Census division, e.g., some divisions did not have overlapping confidence intervals (Supplementary Fig. S4). Compared to non-metro locations, metro locations had a slightly higher ratio in two time periods but otherwise the two groups had overlapping confidence intervals (Supplementary Fig. S5).Fig. 4 Estimated change in seroprevalence to change in reported case prevalence ratios by census region for SARS-CoV-2 in the United States, October 25, 2020–February 11, 2022. Data collected as part of a national, repeated, cross-sectional study of convenience samples of specimens of patients who sought routine screening or clinical care. Footnotes: Bars represent 95% confidence intervals. Data were collected from the 50 United States and the District of Columbia and tested for SARS-CoV-2 antibodies using the commercially available COVID-19 test kits specified in the methods section. Regression models were used to estimate associations between specimen positivity and covariates. The associations were used to create seroprevalence estimates as if all specimens were tested with Roche Elecsys Anti-SARS-CoV-2 pan-immunoglobulin immunoassay and then compared to case counts from CDC's COVID Data Tracker. Laboratories were unable to provide specimens for the following time periods and states and were excluded from analyses: September 6–October 3, 2021, from Indiana, Maryland, New Jersey, and Virginia; November 1–November 28, 2021, from North Dakota; and December 27, 2021–January 29, 2022, from Nevada. Discussion Sero-surveillance data are an important source of data on the burden of infection, particularly during periods of increased transmission and changing testing practices. First, the change ratios, ratios estimating the change in seroprevalence compared to the change in reported case prevalence, can be used as a multiplier to enhance the understanding of the infection burden represented by officially reported case rates. Second, sudden increases in infection rates are essential to consider in vaccine efficacy studies since infection-induced immunity can also provide some protection from infection. The increasing infection-induced seroprevalence rates shown in this study demonstrate that accurate estimates of vaccine effectiveness will need to incorporate previous infection, including in pediatric populations.28 Sudden increases in infection rates impact the disease susceptibility of the unvaccinated population and are vital in the interpretation of vaccine effectiveness data. The change ratios were highest during periods of high transmission, specifically during winter case surges. While increased seroprevalence compared to case prevalence during surges are concerning, they demonstrate the improvement in access to testing compared to spring 2020, when early infection to reported case ratios ranged from 6 to 24 infections per case in metropolitan areas29 , 30 due to a shortage of diagnostic tests. The increased availability and uptake of home testing11 may be an important factor driving winter 2021–2022 increases in seroprevalence compared to case prevalence. Changes in seroprevalence may continue to increase relative to changes in reported case prevalence as home testing becomes more common and fewer jurisdictions engage in active case finding through case interviews and contact tracing. This illustrates the importance of continued sero-surveillance for understanding the true infection burden represented by officially reported case counts and other metrics; interpretation of the significance of reported case counts may require revision in the event of continued increases in this ratio. During times of lower transmission, smaller changes in seroprevalence between time periods suggest a greater proportion of infections were included in reported case counts, though with greater uncertainty. At the nadir of the ratio of the change in seroprevalence compared to the change in case prevalence in April–July 2021 (1.1; CI: 0.6–1.7), the confidence interval was widest because small changes in both seroprevalence and reported case rates resulted in a high coefficient of variation. These data from the current analysis are reinforced by findings from 2020, when jurisdictional infection to reported case ratios declined (range: 1.0–12.5) as transmission decreased during summer 2020 and testing became more widely accessible,20 though the variability in jurisdiction-level change ratios also underscore the importance of sub-national surveys to more accurately depict seroprevalence. Serosurveys can highlight population subgroups that are at higher risk of infection and help target interventions to those subgroups. For example, children had higher seroprevalence31 and have had higher infection to case ratios32 despite suggestions that seroprevalence in children might be underestimated compared with adults.33 Additionally, fewer infections per case were reported during periods of high transmission, especially in certain regions. These seroprevalence results possess some differences from prior seroprevalence estimates, which may be because convenience sampling limits the generalizability of the sample pool. Approximately 14.3% (range, 11.6%–18.5%) of the U.S. population were estimated to have been infected by mid-November 2020,34 which is slightly higher than our estimate of 10.7% from November 9–29, 2020. Our lower estimate may be due to the possibility of underestimation of seroprevalence from the available specimens obtained from clinical laboratories. People engaged in routine medical care and clinical screening likely have greater access to and utilization of healthcare resources, while people with less access are more likely to belong to racially- and ethnically-minoritised groups, disproportionately affected by chronic conditions and COVID-19.35 The lack of race and ethnicity data is especially limiting since disparities have been noted in other seroprevalence surveys.30 , 36, 37, 38 The combined bias is likely to underestimate seroprevalence and the ratio of change in seroprevalence to change in reported case prevalence, though a comparison in a diverse location revealed a higher seroprevalence estimate.39 Persons with specimens collected under routine screening for high-risk conditions may be more likely to be vaccinated than those with high-risk conditions unable to utilize healthcare resources.40 This could be a result of health insurance status and workplace sick leave policies.41 Compared to estimates from a longitudinal study of blood donors from November 2020 through May 2021,14 seroprevalence estimates in this study were consistently 1.0–4.5% higher. This could be explained by the exclusion of the pediatric age group from the blood donor estimate, higher vaccination rates in blood donors,14 or the underrepresentation in blood donor pools42 of people from racially- and ethnically-minoritised groups or other groups disproportionately affected by the COVID-19 pandemic.43 The lack of probabilistic sampling, which has been highlighted as a potential source of bias in serosurveys,44 was one of multiple limitations in this investigation. Several other limitations also may have led to an underestimation of seroprevalence, including the exclusion of specimens from people specifically seeking SARS-CoV-2 antibody testing, the inability of these sero-surveillance methods to detect reinfection (particularly during the Omicron phase45 , 46), and the potential that some fully vaccinated people who are subsequently infected may develop levels of N-antibody that fall below the assay's limit of detection.47 In addition, the likelihood of a given assay to detect antibody post-infection varies by the time since infection and the type of antibody binding.48 , 49 Although the Roche Elecsys N-target assay is less affected by antibody waning compared to other assays, antibody concentrations wane and specimens could fall below the limit of detection.25 While we were unable to control or adjust for antibody waning directly, we mitigated these effects by adjusting results to a single assay, and by performing all testing with a single assay type (Roche Elecsys) since September 2021. Nevertheless, we did not directly adjust our estimates for errors in SARS-CoV-2 antibody measurement from the assays used in this study. Qualitative testing cannot quantify the level of SARS-CoV-2 antibody in the blood,50 and current FDA emergency use authorized SARS-CoV-2 antibody assays are not validated to measure a specific level of immunity or protection from SARS-CoV-2 infection.51 Thus, a SARS-CoV-2 N-target-based seroprevalence estimate does not necessarily indicate the percentage of the population susceptible or immune to SARS-CoV-2 infection or reinfection. Also, while we hoped to be able to calculate an infection-to-reported-case ratio in this manuscript, our ratios of the change in seroprevalence to the change in reported case prevalence are only an approximation. Finally, our power of 70% for detecting a 2% increase in seroprevalence may mean our study was underpowered. Nevertheless, these results provide information to more fully understand the burden of SARS-CoV-2 infection in the United States from late 2020 to early 2022, and to more accurately interpret case reporting to aid public health decision-making.52 The U.S. CDC utilizes case surveillance, hospital admissions, and staffed inpatient beds to evaluate the community-level impact of COVID-19 illness.53 These analyses highlight the importance of a multi-faceted approach to surveillance since reported case rates must be interpreted in the context of variable ratios of the change in seroprevalence to the change in reported case prevalence by geographic area and across time. Conclusions Sero-surveillance data suggest that reported case counts did not completely capture the SARS-CoV-2 infection burden in the U.S. between late 2020 and early 2022, especially during periods of high transmission. Some subgroups, such as children and people living in the South and Midwest regions, experienced a higher infection burden compared to that suggested by case-based surveillance. Sero-surveillance data can aid in the appropriate interpretation of vaccine effectiveness data, demonstrate the increasing importance of accounting for previous infection, provide a more complete picture of COVID-19 impact for community-level decision-making, and identify subgroups at higher risk for infection. With the potential for increased use of at home, viral-based testing, national sero-surveillance will become pivotal in efforts to estimate disease burden and appropriately interpret case rates over time and by geographic region. Contributors Dr. Wiegand and Dr. Clarke had full access to the data presented in the study and take responsibility for data integrity and accuracy of analysis. Study concept and design: Wiegand, Y. Deng, X. Deng, Lee, Jones, Iachan, Clarke. Acquisition, analysis, or interpretation of data: Wiegand, Y. Deng, X. Deng, Lee, Meyer, Letovsky, Jones, Iachan, Clarke. Drafting of the manuscript: Wiegand, Y. Deng, X. Deng, Lee, Jones, Iachan, Clarke. Critical revision of the manuscript for important intellectual content: all authors. Obtained funding: Gundlapalli, Charles, MacNeil, Hall, Clarke. Administrative, technical, or material support: Meyer, Letovsky, Charles, Gundlapalli, MacNeil, Hall, Jones, Clarke. Study supervision: Meyer, Letovsky, Charles, Jones, Iachan, Clarke Disclaimer: The findings and conclusions in the article are those of the authors and do not necessarily represent the views of the U.S. Centers for Disease Control and Prevention or the corporate employers of the authors. Data sharing statement A public use dataset containing a data dictionary and seroprevalence estimates for all rounds of data collection is available at https://data.cdc.gov/Laboratory-Surveillance/Nationwide-Commercial-Laboratory-Seroprevalence-Su/d2tw-32xv. Individual-level data will not be made available. Declaration of interests BioReference Laboratories, Inc., ICF Inc., Laboratory Corporation of America Holdings, and Quest Diagnostics, Inc. were awarded federal contracts from the U.S. Centers for Disease Control and Prevention (CDC) for the execution of this project. No other disclosures were reported. Appendix A Supplementary data Supplementary Material S1 Acknowledgments We thank the following members of CDC for administrative and technical support: Anna Bratcher, PhD, Elizabeth Cole-Greenblatt, JD, Elise Nycz, MHS, Lauren Peel, JD. We thank Tonja Kyle, MS, from ICF Inc. for administrative and technical support. We thank Quest Diagnostics, BioReference Laboratories, and Laboratory Corporation of America for testing specimens. From Quest Diagnostics: Sara Ansari, PhD, Scott Deschenes, Brooke Ethington, MS, MBA, Sara Peters, Caterina Powell, BS, Dianna Tate, Brian Young, AA. From BioReference Laboratories: James Weisberger, MD. From Laboratory Corporation of America: Dorothy Adcock, MD, Kelly Chun PhD, Marla Williams. These individuals were not compensated directly by CDC for their participation in this specific study. Funding: This work was supported by the 10.13039/100000030 CDC , Atlanta, Georgia. e See e.g., 45 C.F.R. part 46; 21 C.F.R. part 56; 42 U.S.C. §241(d), 5 U.S.C. §552a, 44 U.S.C. §3501 et seq. Appendix A Supplementary data related to this article can be found at https://doi.org/10.1016/j.lana.2022.100403. ==== Refs References 1 Holshue M.L. DeBolt C. Lindquist S. First case of 2019 novel coronavirus in the United States N Engl J Med 382 10 2020 929 936 32004427 2 McNamara L.A. Wiegand R.E. Burke R.M. Estimating the early impact of the US COVID-19 vaccination programme on COVID-19 cases, emergency department visits, hospital admissions, and deaths among adults aged 65 years and older: an ecological analysis of national surveillance data Lancet 399 10320 2021 152 160 34741818 3 del Rio C. Malani P.N. Omer S.B. Confronting the delta variant of SARS-CoV-2, summer 2021 JAMA 326 11 2021 1001 1002 34406361 4 CDC COVID-19 Response Team SARS-CoV-2 B.1.1.529 (omicron) variant - United States, December 1-8, 2021 MMWR Morb Mortal Wkly Rep 70 50 2021 1731 1734 34914670 5 Centers for Disease Control and Prevention CDC COVID data tracker https://covid.cdc.gov/covid-data-tracker/#datatracker-home 2022 6 Neelon B. Mutiso F. Mueller N.T. Pearce J.L. Benjamin-Neelon S.E. Spatial and temporal trends in social vulnerability and COVID-19 incidence and death rates in the United States PLoS One 16 3 2021 e0248702 7 Siegel D.A. Reses H.E. Cool A.J. Trends in COVID-19 cases, emergency department visits, and hospital admissions among children and adolescents aged 0-17 Years - United States, August 2020-August 2021 MMWR Mortal Wkly Rep 70 36 2021 1249 1254 8 Lopez L. Nguyen T. Weber G. Seroprevalence of anti-SARS-CoV-2 IgG antibodies in the staff of a public school system in the midwestern United States PLoS One 16 6 2021 e0243676 9 Rogawski McQuade E.T. Guertin K.A. Becker L. Assessment of seroprevalence of SARS-CoV-2 and risk factors associated with COVID-19 infection among outpatients in Virginia JAMA Netw Open 4 2 2021 e2035234 33555331 10 Anand S. Montez-Rath M. Han J. Estimated SARS-CoV-2 seroprevalence in US patients receiving dialysis 1 year after the beginning of the COVID-19 pandemic JAMA Netw Open 4 7 2021 e2116572 34251441 11 Rader B. Gertz A. Iuliano A.D. Use of at-home COVID-19 tests — United States, August 23, 2021–March 12, 2022 MMWR Mortal Wkly Rep 71 13 2022 10.15585/mmwr.mm7113e1 12 Malecki K. Nikodemova M. Schultz A. Population changes in seroprevalence among a statewide sample in the United States medRxiv 2021 2020.12.18.20248479 13 Fisher B.T. Sharova A. Boge C.L.K. Evolution of severe acute respiratory coronavirus virus 2 (SARS-CoV-2) seroprevalence among employees of a US academic children's hospital during coronavirus disease 2019 (COVID-19) pandemic Infect Control Hosp Epidemiol 2021 1 9 14 Jones J.M. Stone M. Sulaeman H. Estimated US infection- and vaccine-induced SARS-CoV-2 seroprevalence based on blood donations, July 2020-May 2021 JAMA 326 14 2021 1400 1409 34473201 15 Walker A.G. Sibbel S. Wade C. SARS-CoV-2 antibody seroprevalence among maintenance dialysis patients in the United States Kidney Med 3 2 2021 216 222.e1 33564746 16 Zyskind I. Rosenberg A.Z. Zimmerman J. SARS-CoV-2 seroprevalence and symptom onset in culturally linked orthodox jewish communities across multiple regions in the United States JAMA Netw Open 4 3 2021 e212816 33688968 17 Chiu W.A. Ndeffo-Mbah M.L. Using test positivity and reported case rates to estimate state-level COVID-19 prevalence and seroprevalence in the United States PLoS Comput Biol 17 9 2021 e1009374 18 Noh J. Danuser G. Estimation of the fraction of COVID-19 infected people in U.S. states and countries worldwide PLoS One 16 2 2021 e0246772 19 Reif J. Heun-Johnson H. Tysinger B. Lakdawalla D. Measuring the COVID-19 mortality burden in the United States Ann Intern Med 174 12 2021 1700 1709 34543588 20 Bajema K.L. Wiegand R.E. Cuffe K. Estimated SARS-CoV-2 seroprevalence in the US as of september 2020 JAMA Intern Med 181 4 2021 450 460 33231628 21 U.S. Department of Agriculture Economic Research Service Rural-urban Continuum codes https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/ 2013 22 Flanagan B.E. Gregory E.W. Hallisey E.J. Heitgerd J.L. Lewis B. A social vulnerability index for disaster management J Homel Secur Emerg Manag 8 1 2011 23 Li Z. Hallisey E. Lewis B. Social vulnerability and rurality associated with higher severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection–induced seroprevalence: a nationwide blood donor study—United States, July 2020–June 2021 Clin Infect Dis 75 1 2022 e133 e143 35137014 24 Breslow N.E. Clayton D.G. Approximate inference in generalized linear mixed models J Am Stat Assoc 88 421 1993 9 25 25 Takahashi S. Peluso M.J. Hakim J. SARS-CoV-2 serology across scales: a framework for unbiased seroprevalence estimation incorporating antibody kinetics and epidemic recency medRxiv 2021 2021.09.09.21263139 26 Bates D. Mächler M. Bolker B. Walker S. Fitting linear mixed-effects models using lme4 J Stat Softw 67 1 2015 1 48 27 United States Census Bureau County population totals: 2010-2019 https://www.census.gov/data/datasets/time-series/demo/popest/2010s-counties-total.html 2021 28 Hall V. Foulkes S. Insalata F. Protection against SARS-CoV-2 after covid-19 vaccination and previous infection N Engl J Med 386 13 2022 1207 1220 35172051 29 Havers F.P. Reed C. Lim T. Seroprevalence of antibodies to SARS-CoV-2 in 10 sites in the United States, March 23-May 12, 2020 JAMA Intern Med 180 12 2020 1576 1586 30 Routledge I. Epstein A. Takahashi S. Citywide serosurveillance of the initial SARS-CoV-2 outbreak in San Francisco using electronic health records Nat Commun 12 1 2021 3566 34117227 31 Clarke K.E.N. Jones J.M. Deng Y. Seroprevalence of infection-induced SARS-CoV-2 antibodies — United States, september 2021–February 2022 MMWR Morb Mortal Wkly Rep 71 17 2022 606 608 35482574 32 Couture A. Lyons B.C. Mehrotra M.L. SARS-CoV-2 seroprevalence and reported COVID-19 cases in U.S. Children, August 2020—May 2021 Open Forum Infect Dis 9 3 2022 ofac044 35198651 33 Toh Z.Q. Anderson J. Mazarakis N. Comparison of seroconversion in children and adults with mild COVID-19 JAMA Netw Open 5 3 2022 e221313 35262717 34 Angulo F.J. Finelli L. Swerdlow D.L. Estimation of US SARS-CoV-2 infections, symptomatic infections, hospitalizations, and deaths using seroprevalence surveys JAMA Netw Open 4 1 2021 e2033706 33399860 35 Tai D.B.G. Shah A. Doubeni C.A. Sia I.G. Wieland M.L. The disproportionate impact of COVID-19 on racial and ethnic minorities in the United States Clin Infect Dis 72 4 2021 703 706 32562416 36 Anand S. Montez-Rath M. Han J. Prevalence of SARS-CoV-2 antibodies in a large nationwide sample of patients on dialysis in the USA: a cross-sectional study Lancet 396 10259 2020 1335 1344 32987007 37 Feehan A. Velasco C. Fort D. Racial and workplace disparities in seroprevalence of SARS-CoV-2, Baton Rouge, Louisiana, USA Emerg Infect Dis 27 1 2021 314 33171096 38 Parrott J.C. Maleki A.N. Vassor V.E. Prevalence of SARS-CoV-2 antibodies in New York City adults, June–October 2020: a population-based survey J Infect Dis 224 2 2021 188 195 34086923 39 Bajema K.L. Dahlgren F.S. Lim T.W. Comparison of estimated SARS-CoV-2 seroprevalence through commercial laboratory residual sera testing and a community survey Clin Infect Dis 73 9 2020 e3120 e3123 40 Galanis P. Vraka I. Siskou O. Konstantakopoulou O. Katsiroumpa A. Kaitelidou D. Predictors of COVID-19 vaccination uptake and reasons for decline of vaccination: a systematic review medRxiv 2021 2021.07.28.21261261 41 Peipins L.A. Soman A. Berkowitz Z. White M.C. The lack of paid sick leave as a barrier to cancer screening and medical care-seeking: results from the National Health Interview Survey BMC Publ Health 12 1 2012 520 42 Shaz B.H. James A.B. Hillyer K.L. Schreiber G.B. Hillyer C.D. Demographic patterns of blood donors and donations in a large metropolitan area J Natl Med Assoc 103 4 2011 351 357 21805814 43 Van Dyke M.E. Mendoza M.C.B. Li W. Racial and ethnic disparities in COVID-19 incidence by age, sex, and period among Persons aged <25 Years - 16 U.S. Jurisdictions, January 1-december 31, 2020 MMWR Mortal Wkly Rep 70 11 2021 382 388 44 Shook-Sa B.E. Boyce R.M. Aiello A.E. Estimation without representation: early severe acute respiratory syndrome coronavirus 2 seroprevalence studies and the path forward J Infect Dis 222 7 2020 1086 1089 32750135 45 Pulliam J.R.C. van Schalkwyk C. Govender N. Increased risk of SARS-CoV-2 reinfection associated with emergence of the Omicron in South Africa Science 376 6593 2022 eabn4947 35289632 46 Altarawneh H.N. Chemaitelly H. Hasan M.R. Protection against the omicron variant from previous SARS-CoV-2 infection N Engl J Med 386 13 2022 1288 1290 35139269 47 Whitaker H.J. Gower C. Otter A.D. Nucleocapsid antibody positivity as a marker of past SARS-CoV-2 infection in population serosurveillance studies: impact of variant, vaccination, and choice of assay cut-off medRxiv 2021 2021.10.25.21264964 48 Alfego D. Sullivan A. Poirier B. Williams J. Adcock D. Letovsky S. A population-based analysis of the longevity of SARS-CoV-2 antibody seropositivity in the United States EClinicalMedicine 36 2021 100902 49 Di Germanio C. Simmons G. Kelly K. SARS-CoV-2 antibody persistence in COVID-19 convalescent plasma donors: dependency on assay format and applicability to serosurveillance Transfusion 2021 50 Gundlapalli A.V. Salerno R.M. Brooks J.T. SARS-CoV-2 serologic assay needs for the next phase of the US COVID-19 pandemic response Open Forum Infect Dis 8 1 2021 ofaa555 33442555 51 U.S. Food & Drug Administration Antibody testing is not currently recommended to assess immunity after COVID-19 vaccination: FDA safety communication https://www.fda.gov/medical-devices/safety-communications/antibody-testing-not-currently-recommended-assess-immunity-after-covid-19-vaccination-fda-safety 5/19/2021 2021 52 Murhekar M.V. Clapham H. COVID-19 serosurveys for public health decision making Lancet Global Health 9 5 2021 e559 e560 33705691 53 Centers for Disease Control and Prevention COVID-19 community levels March 17 2022 2022 https://www.cdc.gov/coronavirus/2019-ncov/science/community-levels.html	36459225	PMC9717553	NO-CC CODE	2022-12-06 23:23:25	no	Nervenarzt. 2022 Dec 2; 93(12):1277-1296	latin-1	Nervenarzt	2,022	10.1007/s00115-022-01414-y	oa_other
==== Front Eur J Pediatr Eur J Pediatr European Journal of Pediatrics 0340-6199 1432-1076 Springer Berlin Heidelberg Berlin/Heidelberg 36459227 4738 10.1007/s00431-022-04738-8 Research Ultrasonographic assessment of diaphragmatic function in preterm infants on non-invasive neurally adjusted ventilatory assist (NIV-NAVA) compared to nasal intermittent positive-pressure ventilation (NIPPV): a prospective observational study Elkhouli Mohamed 12 Tamir-Hostovsky Liran 123 Ibrahim Jenna 2 Nasef Nehad 4 Mohamed Adel [email protected] 12 1 grid.17063.33 0000 0001 2157 2938 Department of Pediatrics, University of Toronto, Toronto, ON Canada 2 grid.416166.2 0000 0004 0473 9881 Department of Pediatrics, Mount Sinai Hospital, 600 University Avenue, Toronto, ON M5G 1X5 Canada 3 grid.12136.37 0000 0004 1937 0546 Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel 4 grid.10251.37 0000000103426662 Department of Pediatrics, Mansoura University, Mansoura, Egypt Communicated by Daniele De Luca 2 12 2022 19 6 10 2022 19 11 2022 27 11 2022 © Crown 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. NIV-NAVA mode for respiratory support in preterm infants is not well-studied. This study aimed to describe the diaphragmatic function, diaphragmatic excursion (DE), and thickness fraction (DTF), in preterm infants < 30 weeks’ gestation supported by NIV-NAVA compared to NIPPV using bedside ultrasonography. In this consecutive prospective study, DE, diaphragmatic thickness at end of expiration (DTexp), end of inspiration (DTins), and DTF were assessed using bedside ultrasound. Lung aeration evaluation using lung ultrasound score (LUS) was performed for the two groups. Diaphragmatic measurements and LUS were compared for the 2 groups (NIV-NAVA group versus NIPPV group). Statistical analyses were conducted using the SPSS software version 22. Out of 70 infants evaluated, 40 were enrolled. Twenty infants were on NIV-NAVA and 20 infants on NIPPV with a mean [SD] study age of 25.7 [0.9] weeks and 25.1 [1.4] weeks respectively (p = 0.15). Baseline characteristics and respiratory parameters at the time of the scan showed no significant difference between groups. DE was significantly higher in NIV-NAVA with a mean SD of 4.7 (1.5) mm versus 3.5 (0.9) mm in NIPPV, p = 0.007. Additionally, the mean (SD) of DTF for the NIV-NAVA group was 81.6 (30) % vs 78.2 (27) % for the NIPPV group [p = 0.71]. Both groups showed relatively high LUS but no significant difference between groups [12.8 (2.6) vs 12.6 (2.6), p = 0.8]. Conclusion: Preterm infants managed with NIV-NAVA showed significantly higher DE compared to those managed on NIPPV. This study raises the hypothesis that NIV-NAVA could potentially improve diaphragmatic function due to its synchronization with patients’ own breathing. Longitudinal studies to assess diaphragmatic function over time are needed. Trial registry: Clinicaltrials.gov (NCT05079412). Date of registration September 30, 2021.What is Known: • NIV-NAVA utilizes diaphragmatic electrical activity to provide synchronized breathing support. • Evidence for the effect of NIV-NAVA on diaphragmatic thickness fraction (DTF) and excursion (DE) is limited. What is New: • Ultrasonographic assessment of diaphragmatic function (DTF and DE) is feasible. • In preterm infants, DE was significantly higher in infants supported with NIV-NAVA compared to those supported with NIPPV. Keywords Diaphragm thickness Diaphragm excursion Ultrasound Preterm infant Non-invasive neurally adjusted ventilatory assist ==== Body pmcIntroduction Over the past few decades, avoidance of invasive mechanical ventilation (IMV) and promoting of the use of non-invasive ventilation (NIV) in infants born prematurely has been accepted as a standard of care to reduce lung injury and subsequent development of bronchopulmonary dysplasia (BPD) [1]. Although nasal continuous positive airway pressure (nCPAP) is the most frequently used NIV mode for infants born before 32 weeks gestation, NIPPV was found to decrease the incidence of reintubation within 2–7 days of life compared to nCPAP [2]. Furthermore, synchronized NIPPV is considered the most effective NIV in preventing extubation failure in preterm neonates with respiratory distress syndrome [3]. The advantages of NIPPV over nCPAP include the ability to deliver higher mean airway pressure (MAP) and carbon dioxide (CO2) clearance [4, 5]. Recently, neurally adjusted ventilatory assist (NAVA) for invasive and non-invasive ventilation (NIV-NAVA) has emerged as a new respiratory support mode for preterm infants with respiratory insufficiency. Typically, NAVA mode (invasive and non-invasive) uses the electrical activity (Edi) of the diaphragm to trigger, set the amount of pressure, and cycle off the ventilator which in turn reduces asynchrony during NIV [6, 7]. In comparison to NIPPV, several studies have reported that NIV-NAVA is associated with a higher success rate of preventing reintubation [8–10], alongside fewer episodes of bradycardia and apnea of prematurity per day [11]. However, data regarding the effect of NIV-NAVA compared to NIPPV on diaphragmatic function and dimensions in preterm infants remains unknown. Lung ultrasound (LU) has been increasingly used in neonates as a non-invasive and radiation-free imaging modality to assess lung aeration and diaphragm function. Moreover, there is a growing interest among researchers in using ultrasound to monitor the evolution of diaphragmatic contractility and dimensions during IMV, for clinical and research purposes [12–15]. Several parameters have been investigated to evaluate diaphragmatic functions, such as diaphragmatic excursion (DE) and diaphragmatic thickness and its fraction (DTF). While DE is known as the distance in which diaphragm can move during the respiratory cycle, DTF is a ratio between diaphragm thickness during inspiration and expiration [16, 17]. Although data about interpreting those parameters in preterm infants are still emerging, high DE was found to be a good indicator for prediction of extubation success in preterm infants < 32 weeks [18]. Likewise, DTF was found to be greater in preterm infants with BPD indicating increased diaphragmatic function to compensate for or the underlying parenchymal disease in this population [19]. Despite NIV-NAVA utilizes diaphragmatic electrical activity to provide synchronized breathing support, evidence for the effect of NAVA on DTF and DE is scarce. We hypothesized that infants on NIV-NAVA will have better diaphragmatic function (DE and DTF) compared to infants on NIPPV. Thus, in this study, our primary objective was to characterize the DE, diaphragmatic thickness at end of expiration (DTexp) and end of inspiration (DTins), and DTF in preterm infants (< 30 weeks’ gestation) who were managed on NIV-NAVA compared to those managed on NIPPV using bedside ultrasonography. The secondary objective was to evaluate lung aeration using lung ultrasound score (LUS) in infants on NIV-NAVA compared to those on NIPPV. Methods Study design We conducted a consecutive prospective study between March 2020 and November 2021 on eligible infants who were admitted to the neonatal intensive care unit (NICU) at Mount Sinai Hospital, Toronto, Canada. Local research ethics approval was obtained (Mount Sinai Hospital REB (19–0324-E), and consecutive patients were enrolled after informed written consent was obtained from parents or guardians. This study was registered with the US Library of Medicine clinical trials registry, www.clinicaltrials.gov (NCT: 05,079,412). Registration has been delayed because of concomitant COVID pandemic announcement that led to withhold all interventional studies include this study for almost a year. Reporting of this study followed the Strengthening Reporting of Observational Studies in Epidemiology (STROBE) Statement [20]. Study participants Consecutive infants born at < 30 weeks’ gestation who were receiving NIV-NAVA or NIPPV at 2–4 weeks after birth and for at least 24 h at the time of recruitment were included. Infants who had congenital or chromosomal abnormalities, neuromuscular disease, known lung malformation or diaphragm dysfunction, or whose parents declined to consent were excluded. Respiratory management for preterm infants < 30 weeks gestational age Respiratory management of preterm neonates < 30 weeks’ gestation at our unit was by unit guidelines developed as part of “Bronchopulmonary Dysplasia Prevention bundle.” The primary mode of non-invasive ventilatory support for infants born at < 24 weeks’ gestation post-extubation is nasal intermittent positive pressure ventilation (NIPPV). Infants born at ≥ 24 weeks’ gestation and spontaneously breathing are initiated on nCPAP (6–10 cmH2O). If infants experience bradycardia and/or apnea while on nCPAP, they will be transitioned to NIPPV. FiO2 is titrated to maintain SpO2 within the target range (91–95%). Given the limited number of NAVA ventilators available in our unit during the study period, NIV-NAVA was used as a rescue NIV respiratory mode for cases failing NIPPV (FiO2 > 0.4–0.5 to maintain oxygen saturation target, and pCO2 > 65 mmHg with a pH < 7.20) to prevent reintubation. Another indication for the use of NIV-NAVA was the presence of significant abdominal distension that can potentially lead to compromised ventilation. Study procedure All eligible infants were consecutively approached for the parent’s consent. We evaluated diaphragmatic thickness (DT) and DE for all enrolled infants using bedside ultrasound (US). We used a high-resolution L20–5 MHz linear probe for the measurement of DT and a C10–3 MHz curvilinear probe for assessing DE (Z. One—Mindray, Inc.). All US assessments were undertaken using the standard technique while infants were in the supine position. Scans were performed either before or 1 h after feeding to avoid any concerns of a full stomach on diaphragmatic assessment. Infants were on continuous cardiorespiratory monitoring for apnea, bradycardia (heart rate < 80 beats per minute), and desaturation (oxygen saturation < 85%) events during the scan. Sonographic assessment of the diaphragm was completed by personnel trained in point-of-care neonatal US with a minimum of 6 months of experience in lung and diaphragm ultrasounds. To minimize errors in measurements and improve our measurement accuracy, we standardized the technique for obtaining diaphragmatic views and how to measure diaphragm thickness and excursion among all investigators prior to study enrollment. Using a recently published standardized technique [12, 19, 21–25], the right hemidiaphragm was assessed with the infant in a relaxed state facilitated by modified swaddling and occasionally pacifier administered by a second person. For DE measurements, the probe was placed subcostal between the anterior axillary and midclavicular line with a curvilinear probe using a standardized protocol for image acquisition. DE was measured as the difference in the position of the outer line of the diaphragm in M-mode at peak inspiration and expiration (Fig. 1a). DT was measured at the zone of apposition where thickening and shortening of the diaphragm could be appropriately evaluated. DT was obtained by placing the linear transducer at the 9th or 10th intercostal space near the mid-axillary line while the transducer was perpendicular to the chest wall [23, 24]. By B-mode, the diaphragm was identified as the hypo-echoic area (muscular layer) bordered by two echogenic lines of the diaphragmatic pleura (upper line) and peritoneum (lower line). Using M-mode tracing, the end of expiratory and inspiratory DT was calculated by determining the maximum perpendicular distance between the pleural and peritoneal layers, measuring only the distance of the hypoechoic area (Fig. 1b). To assess the efficiency of the diaphragm as a pressure generator, the DTF was determined [25]. DTF was calculated as the change in DTexp DTins using the following formula:Fig. 1 a, b Measurement of diaphragmatic excursion and thickness. a Measurement of diaphragmatic excursion using M-mode from a study patient, marked by the length of the orange arrows pointing up. b Measurement of diaphragmatic thickness at the end of inspiration and expiration using M-mode from a study patient, marked by the vertical distance (orange bidirectional arrows) between the 2 layers of diaphragm DTF=[inspiratorythickness-expiratorythickness]expiratorythickness×100 For each diaphragm parameter (DTexp, DTins, DTF, and DE), the average from at least three respiratory cycles was reported to reduce the risk of over or underestimating the diaphragmatic measurements [26]. Standard neonatal LU views, three chest areas on each side (upper anterior, lower anterior and lateral) were obtained, and LUS using grading score (ranges from 0–18 points) were determined as previously described [27, 28]. At the end of the study, all scans for the two groups were anonymized. All scans collected during the study were kept confidential and securely locked in computerized files saved on hospital drive which is password protected. Assessment of diaphragm function (DE and DTF) was performed by one of the investigators who was blind to study groups. Similarly, LUS was evaluated by another study investigator who was unaware of the study groups or results of diaphragmatic measurements of the study patients. To assess the inter-observer correlation, 25% of the anonymized study scans (n = 10 cases of DT and 10 cases of DE) were randomly assigned and revaluated. Measurements were compared, and the intra-class correlation coefficient (ICC) was calculated. Outcomes and clinical data The primary outcome was the sonographic evaluation of DE, DTexp, DTins, and DTF for infants born at < 30 weeks’ gestation after receiving NIV-NAVA or NIPPV for ≥ 24 h. Secondary outcome was LUS for the two groups. Demographic and respiratory data at the time of the scan was collected. Typically, respiratory parameters are documented by respiratory therapist or bedside nurse every hour in the patient electronic medical record. The average from three consecutive measured PIP and MAP prior to LUS scan was reported to minimize risk of bias. Sample size A convenience sample was planned over one year (the study period) since no previous data on DE or DTF in preterm infants on NIV-NAVA was available. Due to COVID-19 pandemic, study recruitment was extended until November 2021. Statistical analysis Statistical analysis was performed using commercially available software (SPSS for Windows Inc. Version 22. Chicago, Illinois). The Kolmogorov–Smirnov test was performed to examine the distribution of data. Independent Student t test was used to compare continuous parametric variables to determine the differences between groups; Mann–Whitney U test was used for continuous non-parametric variables; chi-square test (χ2) or Fisher exact test was used for categorical variables when appropriate. Clinical data and ultrasonography measurements were presented as mean and standard deviation or median and interquartile range for continuous variables or frequency and percentages for categorical variables. The ICC was determined for variables with continuous measurements (mixed factorial design). We calculated the ICC using 30% of the images, which were evaluated by another individual blinded to the groups. p values less than 0.05 were considered significant. Results We evaluated 70 infants born at < 30 weeks’ gestation in a primary screen, of whom 46 cases met eligibility and consented to participate. Out of the 46 consented patients, 40 infants completed the study protocol (20 infants in each group). Six infants (4 cases in the NIPPV group and 2 infants from NIV-NAVA group) were not scanned due to timing issues and were excluded (Fig. 2). Baseline characteristics and respiratory status prior to the LU scanning of the two groups were summarized in Tables 1 and 2. There was no significant difference between the NIV-NAVA group compared to the NIPPV group regarding birth gestation (GA) or postmenstrual age (PMA) at the time of the scan. DE was significantly higher in NIV-NAVA [mean SD 4.7 (1.5)] versus [3.5 (0.9)] in NIPPV, p = 0.007 (Table 3). DTF was higher in NIV-NAVA compared to NIPPV group [mean (SD) 81.6 (30) vs 78.2 (27), p = 0.71). There was no significant difference regarding DTexp and DTins between groups. Both groups showed relatively high LUS but no significant difference [mean (SD) 12.8 (2.6) versus 12.6 (2.6) p = 0.8] (Fig. 3). Duration of IMV and incidence of BPD did not show significant difference between NIV-NAVA and NIPPV group [mean (SD) 5.35 (6.2) days versus 3.35 (4.4) days p = 0.24] and [80% versus 70% p = 0.85], respectively.Fig. 2 Study flow diagram Table 1 Baseline characteristics of the study population NIPPV group (n = 20) NIV-NAVA group (n = 20) p value Gestational age at birth (week), median (IQR) 25.9 (25.4–26.3) 24.6 (24–26.3) 0.24 Birth weight (gram), mean (SD) 756 (657–895) 700 (592- 795) 0.37 Male sex, n (%) 13 (65%) 12 (60%) 0.74 Mode of delivery (caesarian), n (%) 11 (55%) 12 (60%) 0.75 APGAR score at 5 min, median (IQR) 8 (5–9) 7 (4–8.5) 0.35 Antenatal steroid, n (%) 17 (85%) 19 (95%) 0.29 Chorioamnionitis, n (%) 10 (50%) 13 (65%) 0.34 Surfactant n (%) 20 (100%) 19 (95%) 0.31 SNAP-II scores, median (IQR) 14 (5–19.7) 14 (5–14) 0.67 Sepsis prior to US scan, n (%) 3 (15%) 2 (10%) 0.63 HsPDA, n (%) 10 (50%) 9 (45%) 0.75 Duration of IMV before the scan (days), median (IQR) 2 (0–4.7) 3.5 (0–11) 0.43 BPD, n (%) 16 (80%) 14 (70%) 0.85 Corrected age (weeks) at US scan, median (IQR) 28.6 (27.1–29.3) 27.7 (16.5–29.9) 0.88 Weight (grams) at US scan, median (IQR) 880 (800–1045) 910 (767–1015) 0.78 Data expressed as median (IQR) or as number (%) IQR interquartile range, NIPPV non-synchronized non-invasive positive pressive ventilation, NIV-NAVA non-invasive neurally adjusted ventilatory assist, SNAPE II Score for Neonatal Acute Physiology-Perinatal Extension Table 2 Respiratory status prior to the US scan NIPPV group (n = 20) NIV-NAVA group (n = 20) p value Postnatal systemic steroid, n (%) 0 (0%) 2 (10%) 0.11 Respiratory parameters at the scan PIP median (IQR) 16 (15.3–18) 15.5 (13–18) 0.41 PEEP median (IQR) 8 (7–10) 9 (7–10) 0.15 MAP median (IQR) 10.5 (10–12) 12 (8.8–15.3) 0.23 FiO2 median (IQR) 28.5 (22.3–33.8) 27 (23–35.5) 0.82 NAVA level median (IQR) NA 1.7 (1.2–2) NA Edi peak median (IQR) NA 5.3 (4.2–7.7) NA Edi min median (IQR) NA 2.5 (1.03–3.9) NA RR (breath/min) median (IQR) 37 (28.8–50) 45 (38.3–57.5) 0.13 pH median (IQR) 7.35 (7.25–7.43) 7.25 (7.23–7.33) 0.26 pCO2 median (IQR) 58 (40.3–67.5) 53 (50–67.5) 0.71 RSS (resp severity score) median (IQR) 2.7 (2.2–4.2) 3.3 (2.1–5.6) 0.49 Duration of NIV-NAVA prior to the scan (Days) median (IQR) NA 1 (1–2.8) NA Duration of NIPPV prior to the scan (Days) median (IQR) 10 (5.3–11) 9.5 (5.3–15) 0.74 Total duration of NIV before the scan (Days) median (IQR) 12.5 (10–17) 12 (8–26.3) 0.86 Data expressed as median (IQR) or as number (%) IQR interquartile range, NIPPV non-synchronized non-invasive positive pressive ventilation, NIV-NAVA non-invasive neurally adjusted ventilatory assist, RR respiratory rate, RSS respiratory severity score Table 3 Diaphragmatic measurements and lung ultrasound score NIPPV group (n = 20) NIV-NAVA group (n = 20) p value Diaphragmatic excursion (mm), mean (SD) 3.5 ± 0.9 4.7 ± 1.5 0.007 Inspiratory diaphragm thickness (mm), mean (SD) 1.18 ± 0.4 1.18 ± 0.3 0.99 Expiratory diaphragm thickness (mm), mean (SD) 0.69 ± 0.3 0.67 ± 0.2 0.78 Diaphragmatic thickness fraction (mm), mean (SD) 78.2 ± 27 81.6 ± 30 0.71 Lung ultrasound score, median (IQR) 12.6 ± 2.6 12.8 ± 2.6 0.81 Data expressed as mean (SD; standard deviation) or as median (IQR) Fig. 3 Lung ultrasound patterns and scoring: Lung ultrasound findings were categorized into 4 patterns 1 A-lines pattern (score = 0) indicates presence of A-lines artifact, pleural line sliding, and less than 3 B-lines; 2 Nonconfluent B-lines pattern (interstitial syndrome, score = 1), indicates more than 3 B-lines, and pleural line sliding; 3 Confluent B-lines pattern (white lung pattern, score = 2) indicates confluent B-lines, and pleural line sliding; 4 consolidation pattern (subpleural consolidation > 5 mm, score = 3) with absent A-lines, confluent B lines We found a strong inter-observer agreement for the measurement of DT and DE (ICC 0.86 [95% confidence interval (CI) 0.84–0.87] and 0.93 [95% CI 0.91–0.94] respectively). The ultrasonographic exams were generally well tolerated, and none of the scans were aborted because of bradycardia or desaturation during the ultrasonographic exams. Discussion Our study is the first to report that DE was significantly higher in preterm infants on NIV-NAVA compared to DE in infants supported by NIPPV. The significantly higher DE in preterm infants supported with NIV-NAVA can be explained by the fact that this mode of ventilation has a better patient-ventilator synchronization which will eventually augment diaphragmatic contractility. Several studies have suggested that better DE indicates improved diaphragmatic function and can potentially predict successful weaning from mechanical ventilation [29, 30]. Our results aligned with Hadda et al., who randomized adult patients to either invasive NAVA or conventional ventilators. Authors of this study found that invasive NAVA was associated with a higher DE, in comparison to patients supported by conventional ventilators [31]. Alam et al. [32] reported that successful extubation was significantly correlated with DE with an area under the curve (AUC) of 0.83 (p < 0.001) and sensitivity 77.8% and specificity 84.6%. These results were consistent with several other studies that showed promising findings for using NIV-NAVA to facilitate extubation in preterm infants [10, 33, 34]. During our study period, NIV-NAVA was used as rescue mode for infants who failed NIPPV. Currently, attending physicians in our unit are recommending the use of NIV-NAVA as the primary NIV mode post extubation to improve rate of successful extubation. Interestingly, NAVA has been useful in preterm neonate with postsurgical unilateral diaphragmatic paralysis and cases of congenital diaphragmatic hernia with encouraging results [35–37]. Furthermore, Rehan et al. found that a higher level of PEEP in preterm infants supported by nCPAP causes significant reduction in DE suggesting impaired diaphragmatic functions with high PEEP [38]. While the PEEP used in NIV-NAVA and NIPPV in our study were relatively high in both groups, infants on NIV-NAVA had less reduced DE compared to DE in infants receiving similar PEEP on NIPPV. Additionally, our study did not show a significant difference between the 2 groups regarding the diaphragmatic thickness (DTins and DTexp) or DTF. There is no available data regarding the effect of NIV-NAVA on diaphragmatic dimensions in preterm infants to compare with our findings. However, in a recent study conducted in our unit by Yeung et al., authors reported higher DTF in preterm infants who were diagnosed with BPD [19]. Despite up to 80% of our study population had been diagnosed with BPD at 36 weeks PMA, our results did not show similar results as Yeung et al. This could be explained by the fact that our study was done during the earlier neonatal period before BPD established. Another potential reason of not finding difference in DTF between groups was due to the small sample size of our study. In contrast, our results were aligned with Hadda et al., who evaluated the DTF in adults’ patients admitted to ICU with acute respiratory failure, to evaluate the effect of invasive NAVA on the diaphragmatic function. In this study, investigators found no significant difference in DT and DTF between invasive NAVA and conventional ventilator [31]. Furthermore, we evaluated lung aeration using LUS in both groups. Although we found no significant difference between NIV-NAVA and NIPPV regarding the LUS, we noted that LUS were significantly high in both groups when compared with our cut-off score (> 10 points) for early prediction of BPD in similar preterm population [27]. The high LUS in both groups is likely because our study population were born at a mean GA of 25 weeks’ gestation with very immature lung. Several studies that evaluated lung aeration using LUS in the first 2 weeks of postnatal age in preterm infants found similar high LUS that accurately predict the diagnosis of BPD at 36 weeks PMA [39–41]. We acknowledge our study limitations. First, we had a small sample size that could be attributed to interrupted/low recruitment rate due to the COVID-19 pandemic. Second, the study design was based on consecutive recruitment of all eligible patients but lacked randomization. Thirdly, we did not do a serial ultrasonographic assessment to evaluate the changes of diaphragmatic dimensions and functions over time while infants were supported by these two types of NIV. Another limitation is that infants in the NIV-NAVA group were scanned after short duration post transitioning from NIPPV and the risk of “carryover effect” cannot be ruled out completely. Finally, diaphragm ultrasound is operator dependent; therefore, some variations in the measurements are not uncommon. However, our study results have shown high interobserver reliability which validate, to some degree, the study findings. Conclusion In infants born at < 30 weeks’ gestation, NIV-NAVA was associated with significantly higher DE compared to NIPPV reflecting improvement in the diaphragmatic functions. There were no significant differences regarding other measurement such as DTexp, DTins, DTF, and LUS. Further studies, with a larger sample size and serial assessment of the diaphragm are needed to draw a firm conclusion. Acknowledgements We would also like to thank the families, nurses, and respiratory therapists in Mount Sinai Hospital who helped us complete our study. Guarantor AM, who takes responsibility for the content of the manuscript, including the data and analysis (Original Research). Authors' contributions ME and LT has equal contribution to the study design, data collection, and manuscript writing. AM had full access to all of the data and takes responsibility for the content of this manuscript, including study design, data, and data analysis. The study design was conducted by LT, ME, JI, NN, and AM; data collection was performed by JI, ME, and AM. Data analysis was performed by NN and AM. The manuscript was prepared by ME and LT then edited by JI, NN, and AM. All authors of this study approved the final draft of the manuscript. Data availability All data generated or analysed during this study are included in this published article (and its supplementary information files). Declarations Ethics approval Local research ethics approval was obtained (Mount Sinai Hospital REB (19–0324-E), Toronto, ON, Canada. Consent to participate Written informed consent was obtained from parents or guardians. Consent for publication All the authors have seen the final version of the manuscript and gave their full consent for the publication. Conflicts of interest The authors declare no competing interests. Abbreviations DE Diaphragm excursion DTexp Diaphragm thickness at end of expiration DTF Diaphragm thickness fraction DTins Diaphragm thickness at end of inspiration ICC Intraclass correlation coefficient IMV Invasive mechanical ventilation LU Lung ultrasound LUS Lung ultrasound score NAVA Neurally adjusted ventilatory assist nCPAP Nasal continuous positive airway pressure NICU Neonatal intensive care unit NIPPV Nasal intermittent positive pressure ventilation NIV Non-invasive ventilation RCT Randomized control trial SD Standard deviation Abstract publication/presentation Portions of this paper were presented at the Pediatric Academic Society (PAS) meeting in Denver, USA, in May 2022 a as poster presentation as well as at the 4th Neonatal Research Day- Toronto, Canada, on April 2022 as an oral presentation. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Stein H Beck J Dunn M Non-invasive ventilation with neurally adjusted ventilatory assist in newborns Semin Fetal Neonatal Med 2016 21 154 161 10.1016/j.siny.2016.01.006 26899957 2. Lemyre B Davis PG De Paoli AG Kirpalani H Nasal intermittent positive pressure ventilation (NIPPV) versus nasal continuous positive airway pressure (NCPAP) for preterm neonates after extubation Cochrane Database Syst Rev 2017 2 Cd003212 28146296 3. Ramaswamy VV More K Roehr CC Bandiya P Nangia S Efficacy of noninvasive respiratory support modes for primary respiratory support in preterm neonates with respiratory distress syndrome: systematic review and network meta-analysis Pediatr Pulmonol 2020 55 2940 2963 10.1002/ppul.25011 32762014 4. Mukerji A Belik J Neonatal nasal intermittent positive pressure ventilation efficacy and lung pressure transmission J Perinatol 2015 35 716 719 10.1038/jp.2015.61 26043417 5. Latremouille S Bhuller M Shalish W Sant'Anna G Cardiorespiratory effects of NIV-NAVA, NIPPV, and NCPAP shortly after extubation in extremely preterm infants: A randomized crossover trial Pediatr Pulmonol 2021 56 3273 3282 10.1002/ppul.25607 34379891 6. Firestone KS Beck J Stein H Neurally adjusted ventilatory assist for noninvasive support in neonates Clin Perinatol 2016 43 707 724 10.1016/j.clp.2016.07.007 27837754 7. Treussart C Decobert F Tauzin M Bourgoin L Danan C Dassieu G Carteaux G Mekontso-Dessap A Louis B Durrmeyer X Patient-ventilator synchrony in extremely premature neonates during non-invasive neurally adjusted ventilatory assist or synchronized intermittent positive airway pressure: a randomized crossover pilot trial Neonatology 2022 119 386 393 10.1159/000524327 35504256 8. Makker K Cortez J Jha K Shah S Nandula P Lowrie D Smotherman C Gautam S Hudak ML Comparison of extubation success using noninvasive positive pressure ventilation (NIPPV) versus noninvasive neurally adjusted ventilatory assist (NI-NAVA) J Perinatol 2020 40 1202 1210 10.1038/s41372-019-0578-4 31911641 9. Yonehara K Ogawa R Kamei Y Oda A Kokubo M Hiroma T Nakamura T Non-invasive neurally adjusted ventilatory assist versus nasal intermittent positive-pressure ventilation in preterm infants born before 30 weeks' gestation Pediatr Int 2018 60 957 961 10.1111/ped.13680 30133079 10. Shin SH Shin SH Kim SH Song IG Jung YH Kim EK Kim HS Noninvasive neurally adjusted ventilation in postextubation stabilization of preterm infants: a randomized controlled study J Pediatr 2022 247 53 59.e51 10.1016/j.jpeds.2022.04.025 35460702 11. Tabacaru CR Moores RR Jr Khoury J Rozycki HJ NAVA-synchronized compared to nonsynchronized noninvasive ventilation for apnea, bradycardia, and desaturation events in VLBW infants Pediatr Pulmonol 2019 54 1742 1746 10.1002/ppul.24464 31373180 12. Zhao XQ Tang ZJ Xia B Clinical application of ultrasound in paediatric diaphragmatic dysfunction Pediatr Dimens 2021 6 1 6 10.15761/PD.1000214 13. Kharasch SJ Dumas H O'Brien J Shokoohi H Al Saud AA Liteplo A Schleifer J Kharasch V Detecting ventilator-induced diaphragmatic dysfunction using point-of-care ultrasound in children with long-term mechanical ventilation J Ultrasound Med 2021 40 845 852 10.1002/jum.15465 32881067 14. Wait JL Johnson RL Patterns of shortening and thickening of the human diaphragm J Appl Physiol 1997 83 1123 1132 10.1152/jappl.1997.83.4.1123 9338420 15. Boon AJ Harper CJ Ghahfarokhi LS Strommen JA Watson JC Sorenson EJ Two-dimensional ultrasound imaging of the diaphragm: quantitative values in normal subjects Muscle Nerve 2013 47 884 889 10.1002/mus.23702 23625789 16. Ferrari G De Filippi G Elia F Panero F Volpicelli G Aprà F Diaphragm ultrasound as a new index of discontinuation from mechanical ventilation Crit Ultrasound J 2014 6 8 10.1186/2036-7902-6-8 24949192 17. Kim WY Suh HJ Hong SB Koh Y Lim CM Diaphragm dysfunction assessed by ultrasonography: influence on weaning from mechanical ventilation Crit Care Med 2011 39 2627 2630 10.1097/CCM.0b013e3182266408 21705883 18. Bahgat E El-Halaby H Abdelrahman A Nasef N Abdel-Hady H Sonographic evaluation of diaphragmatic thickness and excursion as a predictor for successful extubation in mechanically ventilated preterm infants Eur J Pediatr 2021 180 899 908 10.1007/s00431-020-03805-2 32986125 19. Yeung T, Mohsen N, Ghanem M, Ibrahim J, Shah J, Kajal D, Shah PS, Mohamed A (2022) Diaphragmatic thickness and excursion in preterm infants with bronchopulmonary dysplasia compared with term or near term infants: a prospective observational study. Chest 20. von Elm E Altman DG Egger M Pocock SJ Gotzsche PC Vandenbroucke JP Initiative S The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies Lancet 2007 370 1453 1457 10.1016/S0140-6736(07)61602-X 18064739 21. El-Halaby H Abdel-Hady H Alsawah G Abdelrahman A El-Tahan H Sonographic evaluation of diaphragmatic excursion and thickness in healthy infants and children J Ultrasound Med 2016 35 167 175 10.7863/ultra.15.01082 26679203 22. El-Mogy M El-Halaby H Attia G Abdel-Hady H Comparative study of the effects of continuous positive airway pressure and nasal high-flow therapy on diaphragmatic dimensions in preterm infants Am J Perinatol 2018 35 448 454 10.1055/s-0037-1608682 29132179 23. Grosu HB Lee YI Lee J Eden E Eikermann M Rose KM Diaphragm muscle thinning in patients who are mechanically ventilated Chest 2012 142 1455 1460 10.1378/chest.11-1638 23364680 24. Cohn D Benditt JO Eveloff S McCool FD Diaphragm thickening during inspiration J Appl Physiol 1997 83 291 296 10.1152/jappl.1997.83.1.291 9216975 25. Goligher EC Laghi F Detsky ME Farias P Murray A Brace D Brochard LJ Bolz SS Rubenfeld GD Kavanagh BP Ferguson ND Measuring diaphragm thickness with ultrasound in mechanically ventilated patients: feasibility, reproducibility and validity Intensive Care Med 2015 41 642 649 10.1007/s00134-015-3687-3 25693448 26. Umbrello M Formenti P Longhi D Galimberti A Piva I Pezzi A Mistraletti G Marini JJ Iapichino G Diaphragm ultrasound as indicator of respiratory effort in critically ill patients undergoing assisted mechanical ventilation: a pilot clinical study Crit Care 2015 19 161 10.1186/s13054-015-0894-9 25886857 27. Mohamed A Mohsen N Diambomba Y Lashin A Louis D Elsayed Y Shah PS Lung ultrasound for prediction of bronchopulmonary dysplasia in extreme preterm neonates: a prospective diagnostic cohort study J Pediatr 2021 238 187 192.e182 10.1016/j.jpeds.2021.06.079 34237347 28. Brat R Yousef N Klifa R Reynaud S Shankar Aguilera S De Luca D Lung ultrasonography score to evaluate oxygenation and surfactant need in neonates treated with continuous positive airway pressure JAMA Pediatr 2015 169 e151797 10.1001/jamapediatrics.2015.1797 26237465 29. Hayat A Khan A Khalil A Asghar A Diaphragmatic excursion: does it predict successful weaning from mechanical ventilation? J Coll Physicians Surg Pak 2017 27 743 746 29185398 30. Zambon M Greco M Bocchino S Cabrini L Beccaria PF Zangrillo A Assessment of diaphragmatic dysfunction in the critically ill patient with ultrasound: a systematic review Intensive Care Med 2017 43 29 38 10.1007/s00134-016-4524-z 27620292 31. Hadda V Pahuja S Mittal S Madan K Khan MA Mohan A Guleria R Effects of neurally adjusted ventilation assist (NAVA) and conventional modes of mechanical ventilation on diaphragm functions: a randomized controlled trial Heart Lung 2022 53 36 41 10.1016/j.hrtlng.2022.01.016 35124338 32. Alam MJ Roy S Iktidar MA Padma FK Nipun KI Chowdhury S Nath RK Rashid HO Diaphragm ultrasound as a better predictor of successful extubation from mechanical ventilation than rapid shallow breathing index Acute Crit Care 2022 37 94 100 10.4266/acc.2021.01354 35081706 33. Lee BK, Shin SH, Jung YH, Kim E-K, Kim H-S (2019) Comparison of NIV-NAVA and NCPAP in facilitating extubation for very preterm infants. BMC Pediatr 298 34. Yagui AC Gonçalves PA Murakami SH Santos AZ Zacharias RSB Rebello CM Is noninvasive neurally adjusted ventilatory assistance (NIV-NAVA) an alternative to NCPAP in preventing extubation failure in preterm infants? J Matern Fetal Neonatal Med 2021 34 3756 3760 10.1080/14767058.2019.1697669 31762348 35. Soreze Y Motte E Dell'Orto V Yousef N De Luca D Use of neurally adjusted ventilator assist in postsurgical hemidiaphragmatic paralysis Arch Dis Child Fetal Neonatal Ed 2018 103 F86 f87 10.1136/archdischild-2017-313570 29092910 36. Gentili A Masciopinto F Mondardini MC Ansaloni S Reggiani ML Baroncini S Neurally adjusted ventilatory assist in weaning of neonates affected by congenital diaphragmatic hernia J Matern Fetal Neonatal Med 2013 26 598 602 10.3109/14767058.2012.745502 23126605 37. Roosens S Derriks F Cools F Case report: Non-invasive neurally adjusted ventilatory assist in a newborn with unilateral diaphragmatic paralysis Pediatr Pulmonol 2016 51 E37 e39 10.1002/ppul.23447 27093203 38. Rehan VK Laipraset J Wallach M McCool FD Effects of CPAP on diaphragm dimensions in the neonate Pediatr Res 1999 45 317 317 39. Alonso-Ojembarrena A Lubián-López SP Lung ultrasound score as early predictor of bronchopulmonary dysplasia in very low birth weight infants Pediatr Pulmonol 2019 54 1404 1409 10.1002/ppul.24410 31216121 40. Loi B Vigo G Baraldi E Raimondi F Carnielli VP Mosca F De Luca D Lung ultrasound to monitor extremely preterm infants and predict bronchopulmonary dysplasia. A multicenter longitudinal cohort study Am J Respir Crit Care Med 2021 203 1398 1409 10.1164/rccm.202008-3131OC 33352083 41. Oulego-Erroz I Alonso-Quintela P Terroba-Seara S Jiménez-González A Rodríguez-Blanco S Early assessment of lung aeration using an ultrasound score as a biomarker of developing bronchopulmonary dysplasia: a prospective observational study J Perinatol 2021 41 62 68 10.1038/s41372-020-0724-z 32665687	36459227	PMC9717554	NO-CC CODE	2022-12-06 23:23:25	no	Eur J Pediatr. 2022 Dec 2;:1-9	utf-8	Eur J Pediatr	2,022	10.1007/s00431-022-04738-8	oa_other
==== Front Can J Public Health Can J Public Health Canadian Journal of Public Health = Revue Canadienne de Santé Publique 0008-4263 1920-7476 Springer International Publishing Cham 36459365 714 10.17269/s41997-022-00714-9 Commentary Defining the “local public health unit” for public health systems and services research in Canada http://orcid.org/0000-0001-8569-4395 Plante Charles [email protected] 12 Sandhu Navdeep 3 Bandara Thilina 234 Fuller Daniel 2 Neudorf Cordell 23 1 grid.25152.31 0000 0001 2154 235X Saskatchewan Health Authority, Research Department, St. Andrew’s College, 1121 College Drive, Saskatoon, Saskatchewan S7N 0W3 Canada 2 grid.25152.31 0000 0001 2154 235X Department of Community Health and Epidemiology, University of Saskatchewan, Health Science Building, 107 Wiggins Road, Saskatoon, Saskatchewan S7N 5E5 Canada 3 Urban Public Health Network, 655 Whiteswan Drive, Saskatoon, Saskatchewan S7K 8A2 Canada 4 grid.25152.31 0000 0001 2154 235X School of Public Health, University of Saskatchewan, 104 Clinic Place, S7N 2Z4 Saskatoon, SK Canada 2 12 2022 15 1 4 2022 19 10 2022 © The Author(s) under exclusive license to The Canadian Public Health Association 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. ==== Body pmcIntroduction Public health systems and services research (PHSSR) refers to the systematic study of “the impact of the organization, staffing, financing, and management of public health systems on access to, delivery, cost, quality and outcomes of population-based services and interventions” (Acheson, 1988, p. 284). This field is underdeveloped in Canada, due in large part to a lack of infrastructure for the routine collection of nationally comparable evidence on public health system and service operation across the country. By way of contrast, the field of PHSSR is far better developed in the United States where, since 1989, there have been important efforts to develop and administer regular surveys that provide a national profile of local public health departments (National Association of County and City Health Officials, 2017). One of the primary reasons that the routine collection of nationally comparable data on local public health activities has not emerged in Canada is that public health in Canada is highly decentralized and orchestrated very differently from one region of the country to the next. In effect, public health practice in Canada consists of 14 different and largely siloed public health systems, one for each province, territory, and the Public Health Agency of Canada operating at the federal level, and each one is characterized by different levels of local regionalization and organization. Advancing PHSSR requires us to be able to identify apples-to-apples comparisons between similar local public health units from one jurisdiction to the next. However, at present, there is no established approach for identifying comparable local public health units across jurisdictions due to their diverse and frequently changing organizational structures. We submit that the elementary unit of PHSSR in Canada should be what we elect to call the “local public health unit” (LPHU), which is analogous to the “local health departments” that have emerged as the elementary units of PHSSR in the USA (NACCHO, 2020) and similar arrangements which exist in more or less centralized forms throughout Europe (Jakubowski et al., 2018). In this commentary, we propose a definition for LPHU which can be applied uniformly throughout Canada to identify common units and support interregional comparative research on public health systems and services. Although not our focus, we anticipate that it could also be applied to diverse regions internationally, providing additional opportunities for comparison and learning. Proposed definition An ideal definition for LPHU for research should be specific enough that it can identify a meaningful unit of analysis and generic enough that it can be applied consistently across diverse contexts. With this in mind, we propose that for Canadian research an LPHU should be defined as: The lowest unit of independent (or delegated) responsibility for a defined population, having direct responsibility for the administration of public health programs and services, and led (or co-led) by a qualified Medical Health Officer/Medical Officer of Health. According to this definition, an LPHU is a health unit that meets three overlapping criteria. In the following sections, we discuss in turn each of these three criteria and our reasoning for selecting them. A health unit that fails to meet any one of the three criteria should not be considered an LPHU for research purposes. 1. “The lowest unit of independent (or delegated) responsibility for a defined population” We can think of a “public health unit” as an institution that is responsible for the administration of public health programs and services and led (or co-led) by a qualified Medical Health Officer/Medical Officer of Health (MHO/MOH). A public health unit is considered local when it is the “lowest unit of independent (or delegated) responsibility for a defined population.” Historically, it was more common in Canada for the responsibility of LPHU to be independent of wider health or public health authorities (Rutty & Sullivan, 2010); however, in recent years, amalgamations have resulted in LPHU becoming embedded within wider health systems in most provinces. For example, while the responsibility of Toronto Public Health derives from the City of Toronto and it does not answer to any wider health authority, the responsibility of the Edmonton Zone in Alberta follows most of the country and is delegated by the provincial health authority, Alberta Health Services. If there is no unit of independent (or delegated) responsibility below the provincial or territorial level, then a local health unit may cover an entire province or territory. This is the case, for example, in Prince Edward Island. Additionally, “a defined population” often pertains to a population within a defined geography. In our definition, they do not have to be. For example, the defined local population of the First Nations Health Authority (FNHA) in British Columbia is delineated by First Nations status. In this instance, its responsibility covers a defined population for which there is no lower level of delegated responsibility. 2. “Having direct responsibility for the administration of public health programs and services” An LPHU has direct responsibility for administering health programs and services which fall within recognized categories of “core” or “essential” public health functions. In recent years, representative public health bodies across Canada have made efforts to define the scope of the field, including by defining “core” and “essential” public health functions (CPHA, 2017; Government of British Columbia, 2017; Government of Ontario, 2018). In Canada, six categories of essential public health functions are generally most widely recognized, having been identified and distinguished in Learning from SARS: Renewal of Public Health in Canada—also referred to as “The Naylor Report” (Naylor et al., 2003). These categories are (1) health protection, (2) health surveillance, (3) disease and injury prevention, (4) population health assessment, (5) health promotion, and, sometimes, (6) disaster response. A unit can be considered an LPHU as long as it has direct responsibility to administer programs and services providing any one of these core functions. In some regions of the country, some of these functions are distributed separately between local and regional units, which we describe elsewhere (Plante et al., 2022) Additionally, to say that the unit has direct responsibility for the administration of public health programs and services means that they have the ability to decide on program and service priorities and budget accordingly for their defined population. In recent years, we have begun to see an erosion of the role of LPHU in some regions, such that their administrative and budgetary responsibilities are being stripped from them and they are increasingly only being called upon to act as implementers or “technical consultants” (Cassola et al., 2022). Oftentimes, even these relatively modest advisory roles are not being clearly defined. 3. “Led (or co-led) by a qualified Medical Health Officer/Medical Officer of Health” LPHU are not the only units that may have local responsibility for a defined population and administer public health programs and services. Crucially, the last characteristic that has to be met for a local unit to qualify as an LPHU is that it has to be led by a “qualified Medical Health Officer/Medical Officer of Health.” A qualified MHO/MOH is typically defined at the provincial or territorial level by a combination of provincial legislation (e.g. a provincial public health act, but oftentimes regulations under the act) and provincial medical regulatory or licensing authorities. Most commonly, a qualified MHO/MOH is defined as a medical doctor with an FRCPC or equivalent specialization in Public Health and Preventive Medicine. In less common instances, a qualified MHO/MOH could be a medical doctor with another public health specialization (such as a Master’s in Public Health). These qualified public health professionals may lead their LPHU by themselves or as part of formalized collaborative partnerships with others, oftentimes non-medical administrative leadership. Sometimes organizations that are not LPHU administer public health services and programs to local populations. These organizations could be other health service providers or they could be non-public sector community organizations that have stepped in to fill a need. For example, community health centres (CHC) work to provide a more integrated approach to primary care, which means that they also often provide some services that fall within public health’s core functions, such as health promotion. This also means that they will routinely employ public health professionals as part of their teams, but this alone does not make them LPHU. Conclusion Public health systems and services research in Canada needs an agreed-upon unit of analysis in order to facilitate comparisons between diverse settings and ensure the advancement of the field. The purpose of this commentary is not to supplant existing public health naming conventions or unit arrangements. Rather, it is to provide a common language so that diverse existing arrangements can be related and compared and so that their experiences can more effectively inform one another in a research context. This includes also being able to identify common units within the same jurisdiction over time in such a way that transcends system reorganization. In this commentary, we propose that this unit should be the local public health unit (LPHU) and provide a generic definition. Table 1 provides a complete list of all the LPHU in each province and territory in 2021 based on this definition. This list was compiled from official websites and documentation and through consultation with local MHO/MOH. Table 1 Non-First Nation regional and local public health units (LPHU) by province and territories in accordance with our proposed definition Province/territory Regional Public Health Unit Local Public Health Unit Alberta Alberta Health Services North Zone Edmonton Zone Central Zone Calgary Zone South Zone British Columbia Vancouver Coastal Health Authority Richmond Vancouver North Shore/Coast Garibaldi Vancouver Island Health South Vancouver Island Central Vancouver Island North Vancouver Island Fraser Health Fraser East Fraser North Fraser South Interior Health East Kootenay Kootenay Boundary Okanagan Thompson Cariboo Shuswap Northern Health Northwest Northern Interior Northeast Manitoba N/A Winnipeg Regional Health Authority Interlake-Eastern Regional Health Authority Southern Health Prairie Mountain Health Northern Health Region New Brunswick N/A North Region East Region South Region Central Region Newfoundland and Labrador Central Health Eastern Health Labrador-Grenfell Health Western Health Northwest Territories Health and Social Services Nova Scotia N/A Western Zone Northern zone Eastern Zone Central Zone Nunavut N/A Department of Health Ontario N/A Chatham-Kent Health Unit Lambton Public Health Windsor-Essex County Health Unit Middlesex-London Health Unit Grey Bruce Health Unit Southwestern Public Health Huron Perth Public Health Region of Waterloo, Public Health Brant County Health Unit Hamilton Public Health Services Haldimand-Norfolk Health Unit Niagara Region Public Health Wellington-Dufferin-Guelph Public Health Peel Public Health Halton Region Health Department Toronto Public Health York Region Public Health Peterborough Public Health Haliburton, Kawartha, Pine Ridge District Health Unit Durham Region Health Department Hastings Prince Edward Public Health Leeds, Grenville and Lanark District Health Unit Kingston, Frontenac and Lennox and Addington Public Health Eastern Ontario Health Unit Ottawa Public Health Renfrew County and District Health Unit Simcoe Muskoka District Health Unit Timiskaming Health Unit North Bay Parry Sound District Health Unit Algoma Public Health Public Health Sudbury and Districts Porcupine Health Unit Northwestern Health Unit Thunder Bay District Health Unit Prince Edward Island N/A Health Prince Edward Island Quebec N/A Bas-Saint Laurent Saguenay-Lac Saint Jean Capitale-Nationale Mauricie-et-Centre-du-Québec Estrie Montréal Outaouais Abitibi-Témiscamingue Côte-Nord Nord-du-Québec Gaspésie – Îles-de-la-Madeleine Chaudière-Appalaches Laval Lanaudière Laurentides Montérégie Centre-du-Québec Saskatchewan Saskatchewan Health Authority North Urban (Regina) Urban (Saskatoon) Rural Yukon Territory N/A Department of Health and Social Services We have discussed at greater length some of the limitations and challenges posed by our definition (Plante et al., 2022). A unit of public health professionals that is not empowered to act on public health should not be considered a public health unit for research purposes. Likewise, a unit that is empowered to act on public health but is not headed by a qualified practitioner should also not be considered one. These conditions could be a barrier to the adoption of our definition in some regions and settings but we believe that the clarity it brings offsets this risk. For instance, it allows for comparison between LPHU and different forms of non-LPHU and their impacts on health. There still remain important differences between LPHU as we define them. However, rather than rendering these units incomparable, these differences represent variations that can be used to generate new hypotheses and advance our understanding of LPHU and their impacts. For example, in some regions of the country, LPHU are embedded within ministries of health. Working with our common unit allows us to see how these units are similar to those in regions that decouple their local public health operations from government while also leading us to consider the impacts of this difference. Acknowledgements A previous version of this paper was presented at the Annual Meeting of the Canadian Public Policy Network in February 2021. The authors thank Lori Baugh Littlejohns for her helpful comments and suggestions. Author contributions CP led the program of research and manuscript preparation, and was involved in all aspects of the project. NS provided research support throughout. TB and DF helped with manuscript preparation. CN was involved in planning and supervised the work. All authors provided critical feedback and helped shape the research, analysis, and manuscript. Funding This project was funded in part by the Urban Public Health Network, the Canadian Institutes of Health Research, and the Canadian Partnership Against Cancer. Data availability Not applicable. Code availability Not applicable. Declarations Ethics approval Not applicable. Consent to participate Not applicable. Consent for publication Not applicable. Conflict of interest The authors declare no competing interests. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Cassola A Fafard P Nagi R Hoffman SJ Tensions and opportunities in the roles of senior public health officials in Canada: A qualitative study Health Policy 2022 126 10 988 95 10.1016/j.healthpol.2022.07.009 36002358 Committee of Inquiry into the Future Development of the Public Health Function Public Health in England: Report of the Committee of Inquiry into the Future Development of the Public Health Function 1988 London The Stationery Office CPHA Public health: A conceptual framework 2017 Canadian Public Health Association Government of British Columbia. (2017). Promote, protect, prevent: Our health begins here: BC’s guiding framework for public health. Government of British Columbia. Government of Ontario. (2018). Protecting and promoting the health of Ontarians: Ontario Public Health Standards - Requirements for programs, services, and accountability. Government of Ontario. Jakubowski E Kluge H Rechel B Rechel B Jakubowski E McKee M Nolte E Chapter 3: Organization of public health services Organization and financing of public health services in Europe (pp. 17–58) 2018 World Health Organization NACCHO 2019 National Profile of Local Health Departments 2020 National Association of County and City Health Officials National Association of County and City Health Officials 2016 National Profile of Local Health Departments 2017 National Association of County and City Health Officials Naylor D Basrur S Bergeron MG Brunham RC Butler-Jones D Dafoe G Learning from SARS: Renewal of public health in Canada 2003 National Advisory Committee on SARS and Public Health Plante, C., Sandhu, N., Bandara, T., & Neudorf, C. (2022). Toward a definition of ‘local public health unit’ for public health systems and services research in Canada. Urban Public Health Network. Rutty C Sullivan SC This is public health: A Canadian history Public Health 2010 4 10	36459365	PMC9717555	NO-CC CODE	2022-12-06 23:23:25	no	Can J Public Health. 2022 Dec 2;:1-5	utf-8	Can J Public Health	2,022	10.17269/s41997-022-00714-9	oa_other
==== Front Innov. Infrastruct. Solut. Innovative Infrastructure Solutions 2364-4176 2364-4184 Springer International Publishing Cham 1010 10.1007/s41062-022-01010-8 State-of-the-art Paper A state-of-the-art review on sustainable low-cost housing and application of textile reinforced concrete http://orcid.org/0000-0002-9393-7809 Immanuel Sophia [email protected] Baskar K. [email protected] grid.419653.c 0000 0004 0635 4862 Department of Civil Engineering, National Institute of Technology, Tiruchirappalli, India 2 12 2022 2023 8 1 3922 8 2022 24 11 2022 © Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This article reviews the various housing schemes implemented in India post-independence. This year marking the 75 years of independence, the demand for affordable housing for economically weaker sections and low-income group are at stake in the country. In achieving “Housing for all” in the rapidly urbanizing Indian context, Textile reinforced concrete (TRC) technology promises to be environmentally and economically “affordable” consuming significantly fewer quantities of building materials with high embodied energy. Textile reinforced concrete is the most durable, lightweight, and highly ductile structural component with its non-corrosive textiles replacing the conventional steel reinforcement used, and it can be the most feasible solution to the problem of implementing affordable housing in urban India. TRC has been an excellent solution for the retrofitting and strengthening of existing infrastructures paving its progressive research direction of using this as a load-bearing structural component. With a substantially lower carbon footprint than conventional RCC structures, TRC offers better potential for sustainability. Furthermore, this article proposes a futuristic direction to enhance the research on the application of TRC as a structural component for prefabricated low-cost housing. Keywords Affordable housing Sustainable Low-cost materials EWS TRC PMAY-U issue-copyright-statement© Springer Nature Switzerland AG 2023 ==== Body pmcIntroduction Urban sprawl and migration from rural to cities have resulted in severe housing scarcity in metropolitan India, especially among the economically weaker sections of the society [1]. The “Summit on Sustainable Development” in New York in September 2015 with 193 United Nations member nations gathered with a solid motive to eradicate poverty by 2030 in developing countries and to build a sustainable future considering the 17 sustainability global goals. By 2030, one of the most important goals was to ensure everyone would have access to adequate, secured, reasonably priced housing with all the basic amenities and to rehabilitate the slums. As per the RBI report, by 2025, the Indians residing in cities are estimated to exceed 543 million. By 2020, urban regions will have contributed to more than 70% of India's Gross Domestic Product (GDP). In terms of Foreign Direct Investment (FDI) inflows, construction is the fourth largest industry. “Housing for All” a government initiative in India, is anticipated to attract 1.3 trillion US dollars in investments in the housing industry by 2025 [1]. In India, an approximated deficit of around 18 million housing units were identified, of which 99% belong to the economically weaker sections of society [2]. Hence, the policymakers in India foresaw this worldwide goal and set a lofty deadline of 2022 for achieving it in the country. The Indian government, along with the Reserve Bank of India (RBI), has taken several measures to help low-income families with various schemes and initiatives so that poverty in India is wholly eradicated. “Affordable housing” is one such initiative of the government in which these housing units make them more accessible to the country's poorest citizens. Housing regulations, government incentive schemes, and an interest subsidy scheme under the Pradhan Mantri Awas Yojana (PMAY) were all positive steps in this direction [3]. This sector has enormous demand but very minimal supply [4]. The United Nations financed the research project on Mainstreaming Sustainable Social Housing in India Project (MaS-SHIP), which is quite like the PMAY-U, it aimed to examine the social implications on the development of housing units in urban India. Hence, this research on affordable housing is even more critical since it has the potential to address the problems, meet the tremendous demand in this low-cost housing sector, and bridge the supply–demand gap. The building phase was at complete halt during the covid-19 epidemic, resulting in a significant delay in constructing houses for EWS communities. Following covid, the cost of conventionally used construction materials has reached an all-time hike, necessitating the need to adopt a low-cost alternative material that functions well in all weather conditions and can be prefabricated and deployed on-site as soon as possible. Most of the Housing models projected have steel reinforcement which is costly and exhibit corrosion well before the targeted service life of the structures. In alternative to this, a thin and light weight structural elements using Non-Corrosive Textile reinforced composite panels (TRCP) [5] which eliminates thicker cover requirement can be a promising solution for affordable housing. TRC allows the design of very thin-structured concrete elements with high strength in compression and tension [6]. This article proposes the most effective solution incorporating TRC made with innovative sustainable high performance fine-grained cementitious matrix and textiles embedded in the matrix as shown in Fig. 1, which are not only cost effective but also would perform well in any kind of exposure. Techniques for large scale production and deploying its effective application in various places in our country. The key objective of low-cost housing construction approaches is to lower the cost of construction by employing good alternate materials and effective techniques and methods, which is what this paper outlines. This article also discusses the necessity for affordable housing, obstacles to its creation, and a feasible low-cost housing construction technique using Textile Reinforced Concrete.Fig. 1 Textile reinforced concrete casting process and hardened component [7] Housing schemes in India: an overview Government of India has significantly introduced various housing schemes in India following Independence which are outlined in this section. It reviews the design of these initiatives to see if the strategy defining such design has satisfactorily answered housing poverty in the cities [8]. The housing policy of India must change in response to the extent the urbanization has evolved. The failure of several housing initiatives by the Government of India that have been tried over the last eight decades has been attributed to the lack of an integrated approach in addressing the housing poverty and the reason for the still existing slums. Construction of Low-Cost housing units or the facilitation of such construction through financial aid has been the dominating technique for addressing housing poverty [9, 10]. Such actions do not really accomplish anything to stop the spread of slums. Additionally, the tactics being used do not fit the nature of the urban poverty the nation is experiencing. Under PMAY-U, 1.3 crores of new Low-cost Housing for Economically Weaker section (EWS) were sanctioned to be completed by March 2022 but currently only 61.77 Lakhs houses were only possible to be constructed within the stipulated time. Using a technology that not only is sustainable but cost effective with minimum CO2 footprint is what is feasible for the remaining housings to be constructed for EWS. In Fig. 2, Various Housing schemes implemented in India post-independence is briefly highlighted using a flowchart.Fig. 2 Various housing schemes from 1950 to 2022 [9, 10] Need for low-cost housing in India There has been a significant imbalance in the supply–demand of urban housing in India, according to the Technical Group (TG-12) Report on Estimation of Urban Housing Shortage [11]. Housing units for the EWS sectors of society are perpetually scarce in all developing countries worldwide. It is challenging to determine the precise extent of the housing shortage in metropolitan cities due to unchecked growth, fast migration, and a lack of reliable statistics. For 66.30 million urban families, a technical team established by the Ministry of Housing and Poverty Alleviation (MHPUA) estimated a deficit of 24.71 million housing units at the finish of the 10th five-year plan. According to the group's estimates, most of the deficit was in the EWS category, where the gap was estimated as 88 percent; LIG accounted for 11 percent; and MIG/HIG groupings, where the anticipated shortfall was only 0.04 million housing units. The group estimated that the total housing demand in urban centers during the 11th Five Year Plan (2007–12), including backlog, will range from 26.53 million housing units for 75.01 million families. It is stated by several researchers that 30 million homes would be needed by the year 2020 to satisfy the national aim of providing affordable shelter for all, if the current trend of increasing the backlog of housing continues. Technical Committee also critically examined the housing shortage in each category separately and found that the EWS category had the highest housing shortage, with 99.9% of all EWS households experiencing a shortage, followed by LIG with 10.5 percent and MIG/HIG with only 0.2 percent of households experiencing a shortage. It is observed that 96% of India's entire housing deficit was attributable to the Economically weaker sections (EWS) and low-income group (LIG) as stated in Table 1 from MHUP annual report 2012.Table 1 Shortage of Housing in Urban India as reported by the Ministry of Housing and Urban Poverty (MHUP) report [11] Types of housing units Shortage (in 2012) People residing in substandard housing (Katcha Houses) 0.99 People living in obsolescent houses 2.27 People residing in overcrowded localities 14.99 Poverty-stricken homeless people 0.53 Total Housing requirement 18.78 I. Economically weaker Sections(EWS) 10.55 (56%) II. Low income group (LIG) 7.41 (40%) III.Medium and high-income group (MIG + HIG) 0.82(4%) The McKinsey Report [12] predicts that by 2030, 68 cities in India will have a population of one million or higher, up from 42 now, with 40% of the country's people residing in urban regions as depicted in Fig. 3. It predicts that from 19 million housing units in 2012, the demand for affordable housing will go up to 38 million units in 2030.Fig. 3 Urbanization rate in India from 1990 to 2030 [12] The housing unit requirement for EWS with a max carpet area of 30 sq.m is given in Table 2 is possible only with low-cost building materials which will be durable with low maintenance costs, Eco-friendly with low Carbon footprint, and Economical.Table 2 Under credit-linked subsidy scheme (CLSS) Category of beneficiaries Income (per annum) Interest subsidy (%) Loan amount (₹ in lakhs) Carpet area (sq.m.) Economically weaker section (EWS) Up to ₹ 3 lakhs 6.5 6 30 Low income group (LIG) ₹3–6 lakhs 6.5 6 60 Middle income group (MIG –I) ₹6–12lakhs 4 9 120 Middle income group (MIG−II) ₹ 12–18 lakhs 3 12 150 The number of houses finished under PMAY-U has increased by three times since April 2017, demonstrating the central government's relentless effort for affordable Housing. The current status of the PMAY-U scheme is presented in Table 3.Table 3 Current status of PMAY-U “Housing for all by 2022”. Source: Compiled from PMAY(U)-HFA'22 progress report, MoHUA (GoI), 2022 (1 crore = 100 lakhs; 1 lakh crore = 0.1 million) Project Proposal considered 21,566 projects Project costs ₹ 8.31 Lakhs Cr Central assistance approved ₹ 2.03 Lakhs Cr Central assistance released ₹ 1.20 Lakhs Cr Houses sanctioned ₹ 122.69 Lakhs Houses grounded ₹ 102.59 Lakhs Houses completed ₹ 61.77 Lakhs corros As of August 1, 2022, *Included incomplete houses of earlier NURM (National Urban Renewal Mission) Within the framework of this programs, the government intends to build 2.03 lakh crore housing units. According to the latest data given by the MoHUA in August 1 2022, approximately 21,566 Projects have been approved for the building of 122.69 Lakhs homes under this initiative, of which 102.59lakhs homes are already under construction and 61.77 Lakhs homes have been completed [13]. PMAY-U's current report also mentions that 16 Lakhs new houses will be constructed using new technology. Sustainable materials for low-cost housing Utilizing materials that are readily accessible locally helps reduce the negative impact on the environment of the infrastructure being built in the locality. The suitability of sustainable construction materials and building structures can be determined by their adherence to variability in climate conditions and natural catastrophic threats. Advancements in technologies and information sharing with local populations are sought in every situation. Local people can be made aware of these materials and may be trained and educated on low-cost housing construction to create more sustainable building materials [14]. The Building Materials and Technology Promotion Council (BMTPC), Ministry of Housing and Urban Poverty Alleviation (MHUPA), Government of India (GoI), has been in charge of studying, assessing, validating, promoting, and endorsing innovative construction technologies for acceptance to be implemented for low-cost mass housing schemes across the nation. BMTPC has recommended the following technologies, including [15, 16]. There are numerous technologies that can be used to provide affordable housing. It is challenging to recommend a particular technology for use throughout the entire country. Because of this, it is crucial to conduct research in each area to determine the best technology to utilize given the context, location, climate, local requirements, available resources, tenant preferences, cultural considerations, time, cost, sustainability, and other factors. To select the optimal technology and accomplish sustainable development goals, it is crucial to conduct a trade-off analysis between several factors. The demand for housing will necessitate the construction of millions of houses worldwide. Technology and sustainability were not discussed in further detail in the housing efforts. These details may significantly affect an aim of massive accumulation. To reduce embodied energy, each project should be approached as a unique circumstance and the appropriate technology should be selected based on its unique requirements. Embodied energy has a substantial impact on the overall life cycle energy of naturally ventilated and partially air-conditioned buildings. Additionally, the element of thermal comfort, which is directly related to energy, is really what keeps the tenants in any type of housing. Thus, thermal comfort does have a big impact on how well housing plans work. Passive design strategies, for example, may in some cases be effective in attaining thermal comfort and lowering operational energy in tropical regions. In some cases, however, passive design is unable to offer the required comfort for independent life. Therefore, research should also point in that direction. more cost-effective cooling and heating options that are also more energy efficient. As a result, many considerations should be considered when choosing a technology, including whether prioritizing the criteria will help with decision-making in conflict situations. Sustainability in building can only be achieved if these needs are considered at the planning stage. In Fig. 4, bamboo dominates a variety of uses when evaluating environmentally acceptable building materials. In spite of its relatively low fire resistance, bamboo has demonstrated its effectiveness as a material for reinforcing. In particular the north-eastern regions of India, bamboo continues to prove to be a cost-effective alternative in composite forms and with the use of various protective coatings. Mud blocks and coir-based solutions are becoming more effective in South India. The coir industry is currently booming in the twenty-first century, with a diverse range of products being developed, from coir concrete pavements to coir polymer composite boards and panels for doors and walls. These materials are not long-lasting and eventually deteriorate.Fig. 4 Various eco-friendly, energy efficient, cost-effective composite products for Low-Cost Housing (BMTPC) [14, 15, 17–25] Figure 5 depicts the modular housing technologies that the BMTPC [14] recommends. Utilizing waste materials, researchers have developed bricks such Calcium Silicate Bricks, Lime Fly Ash Bricks, and Aerocon Blocks for Walls that are more lightweight and efficient than regular bricks. The conventional in situ construction method, which requires months to complete, has been superseded with precast construction thanks to technological advancements. Precast technologies allow for the rapid construction of housing units [19–22]. Complex geometrical designs are achievable with precast and pre-engineered technologies thanks to the use of optimized materials, which is not possible with traditional methods. Both quantity and quality are fully considered utilizing hollow core, waffle, and pre-stressed torn precast slabs has enhanced strength and durability while reducing material consumption, cost, and geometrical requirements. In comparison with the cost of reinforcement used in above structures and sustainability as the major concern, the use of textiles is quite economical with high performance.Fig. 5 Modular housing technologies recommended & approved by BMTPC [8, 15, 19, 21–27] Out of the eighteen technologies mentioned most of them are lightweight, but the thickness of the walls or slabs is very similar to the conventional RC structures. Most of these techniques use steel reinforcement subjected to corrosion during its lifetime. Hence, the use of technologies that survives to a greater extent is the need for affordable Housing in India. Housing units with durable modern technology, earthquake resistant, stable in waterlogged areas during monsoon and flood, withstand high temperatures during summer, and has good freeze and thaw resistance during peak winter is the most critical solution for the construction industry in India. Researchers are working on still more advanced technology to design the structural components based on crack-resistant design. Bridging the crack in the structure under various loadings utilizing fibers and textiles instead of steel will help prevent catastrophic failure of structural systems, promising the life safety of the people residing. Performance criteria for affordable housing technologies in India In Table 4, the building technologies currently adopted in India for Affordable housing in various regions is compared based on its physical properties, structural performance and considering sustainability and durability aspect. It is observed that some of the low-cost technologies like Aerocon Panels [28] and EPS Panels [29] which are widely used perform low when durability is concern, cracks are developed within a decade of the structure built. The failure mechanism of EPS panels are through shear failure, even though cost is really low but in long run they do not perform well. Low-cost housing does not mean to compromise on the quality of construction.Table 4 Properties of building technologies used currently in India [6, 7, 37–45] SLNo Properties Bamboo Concrete Blocks Ferrocement Panels Aerocon Panels Fiber Cement Composites Fly Ash Bricks Compressed Mud Blocks Rice husk bricks EPS Panels Cellular Foam Concrete Panel TRC Sandwich Panels 1 Structural Good in Shear, High flexibility than steel and lower modulus of elasticity High Strength and be tailored as per mortar mix Lightweight High strength Low density High crack resistance of mortar Lightweight, Dry wall connections Easy to handle and place Lightweight, High strength to weight ratio, Corrosion resistance, Crack resistance Eco-friendly, Saves energy, High strength, Local mud used, Economic and energy efficient Pozzolanic, Economical, Corrosion resistance increases Lightweight Low shear resistance, Cracks very early Lightweight Foam concrete Lightweight, High strength to weight ratio, High performance, High Ductility, Crack Resistant design, Corrosion free, Minimum Cover requirement (thin structural element) 2 Thermal Good resistance Excellent Excellent Excellent Excellent Excellent Excellent Moderate Good Excellent Excellent 3 Water Resistant Moderate Good Excellent Excellent Good low Low Low Excellent Moderate Excellent 4 Buildability Moderate Good Good Good Moderate Good Good Moderate Good Excellent Excellent (Any complex geometries) 5 Durability Low Moderate Good Good (but not crack resistant) Good Low Low Very low Good Excellent Longer Service life Crack resistant Bricks made of Flyash [30–32], compressed mud blocks [33–35] were extensively used for rural housing in various districts in India. Reuse of waste materials inspired researchers to develop new building materials out of wastes. Ferrocement technology was quite popular in 20th century for its quicker construction and less requirement of shuttering for construction of panels, water tanks, STP tanks, etc. Textile reinforced concrete incorporated ferrocement concrete concepts by replacing the wire meshes with textile layers [36]. TRC is now utilized extensively for nonstructural components like exterior paneling and facades, and research into its application in sandwich panels is still underway. Numerous prototypes have lately been created, but due to the difficult design processes, their practical use is restricted. Additionally, there was a lack of assurance that such a material could be used as a structural element due to the absence of production and testing standards. The modeling of intricate components as TRC shells, bridges, and decks will be made simpler if the existing problems are solved, guaranteeing that this material finds more uses. TRC as a feasible solution for low-cost housing Adoption of cutting-edge and creative building technologies is primarily required to shorten construction time, minimize construction costs, and guarantee construction quality. A new era in affordable housing would be ushered in by the Government of India's Technology Mission under PMAY to source the most cutting-edge and cost-effective global technology. Sustainable construction techniques are increasingly being used in the modern day as a prime driving criterion for advancement in the building and construction industry [46]. The spectrum of sustainability features provided by TRC [47, 48] is extensive. It has the potential to produce components using a substantially lesser material as well as a longer service life in comparison to the traditional concrete which is still being used in our country [49]. TRC can also be used to upgrade the mechanical performance of the already existing old, deteriorating structures in order to increase their ability to tolerate higher static and dynamic loads [50]. For more than a century, steel has been employed as reinforcement in concrete. However, steel reinforcement is vulnerable to corrosion, which reduces the effective cross-sectional area of the rebar and, in severe exposure conditions, causes the concrete to spall and collapse structurally. It was demonstrated that substituting textile reinforcement for steel reinforcement can boost the longevity and dependability of buildings [51]. High-performance fine-grained mortar and strong textile materials make up Textile Reinforced Concrete (TRC). The textiles are multi-yarns with a high tensile strength that are often constructed of polymer, carbon, basalt, or alkali resistant (AR) glass. The maximum aggregate size that may be utilized in TRC depending on the dimensions of the structural components and yarn distance. Typically, TRC aggregate has a thickness of less than 2 mm. The positives of TRC are its high tensile strength, pseudo-ductile behavior, and corrosion and acid attack resistance. The TRC composite is also suitable for the fabrication of lightweight structural components as well as for rehabilitating and restoring aging structural elements due to its superior mechanical properties and durability features [50, 52]. Properties of the fiber mesh as reinforcements Synthetic fibers such as carbon, alkali-resistant glass, E-glass, aramid, polypropylene, and natural fibers like basalt and sisal as shown in Fig. 6 are adopted to develop lightweight TRC structural components. These fibers have high mechanical characteristics compared to metallic fibers like steel which are described in Table 5. Textiles of these fibers are available in woven, non-woven or in knitted forms as shown in Fig. 7 [49]. Several coatings, including epoxy resin coated, styrene butadiene rubber (SBR) coated, and alkali resistant coating, which preserves these textiles from deteriorating within the alkaline fine grained cementitious matrix.Fig. 6 Various fibers used as reinforcement in the TRC [52, 53] Table 5 Properties of various fibers [5, 54, 55] Fiber Tensile strength (MPa) Modulus of elasticity (GPa) Ultimate strain (%) Density (g/cc) AR glass 2500 70 3.6 2.78 Carbon 3500–6000 230–600 1.5–2.0 1.60–1.95 Aramid 3000 60–130 2.1–4.0 1.4 Polypropylene 140–690 3–5 25 0.9–0.95 Basalt 3000–4800 79.3–93.1 3.1 2.7 Sisal 600–700 38 2–3 1.33 Steel 1200 200 3–4 7.85 Fig. 7 Configuration of various textile used as reinforcement in the TRC [54–58] Considering the facts that have been reviewed and displayed in Fig. 8, the normalized values highlights the merits and demerits of each reinforcing textile. Carbon textiles have lower ultimate strain in contrast to all textiles but have higher tensile strength and elastic modulus. The steel reinforcement bar has the second-highest modulus of elasticity, the third-highest strain, and the lowest tensile strength (at yield). Compared to steel's yield strength characteristics, alkali-resistant glass textiles have better tensile strength and ultimate strain but a significantly lower E-modulus. Steel reinforcement despite having an E-modulus close to 1 basalt fiber yarn tensile strength which is 5 times higher than steel and a significantly higher ultimate strain than the yield strain of steel [60].Fig. 8 Comparison of normalized mechanical properties of all Textiles [59, 60] Precast application of TRC The positives of TRC technology include highly polished surfaces, reduced product thickness owing to lightweight structures [38], the potential for free-form architecture, and enhanced mechanical and durability properties. It has a high strength-to-weight ratio, hence its ability to manufacture into complex shapes and geometries can be tailored for its mechanical properties according to specific needs. Properties of textile include its non-corrosive, high fracture toughness [61], low thermal conductivity, and high fatigue performance [62–64]. TRC is used in the production of precast structural components and in the rehabilitation and repair of reinforced structures. Prefabricated TRC components utilize less concrete, have lower manufacture, transportation, and construction costs, and have less wastage [52]. Hence, they are employed for both architectural and structural components. For the creation of facades, employ fabrics like carbon or AR glass [62]. In India, lot many of architectural structures have incorporated the use of TRC facades which are crack-resistant and innovative designs with aesthetically pleasing forms. In the fabrication of prototype ventilated facade components [65], using a mix of continuous AR-glass threads and textile meshes embedded in self-compacting high-performance fiber-reinforced micro concrete Precast pedestrian bridges utilized TRC. The first TRC Bridge, with an 8.6 m span, was constructed in Oschatz, Germany. For the same, four layers of fine concrete with AR glass yarns embedded as reinforcement within the matrix were used. Then a 17 m long TRC Bridge in Kempten, Germany built to allow vehicles to cross over with maximum weight of 3.5 tons. Later, in 2010, the longest TRC Bridge with 97 m span length was constructed using AR glass textiles in Albstadt, Germany (Figs. 9 and 10).Fig. 9 TRC Bridge ‘Rottachsteg’ in Kempten, Germany [60] Fig. 10 First segmental TRC Bridge in Oschatz, Germany [66] Free-form structural elements are becoming the most remarkable trends in modern architecture which could be possible only using TRC. Pavilion Building at RWTH Aachen University paved the way for TRC to be utilized as Structural Component [47, 60] (Figs. 11, 12, 13, 14 and 15).Fig. 11 Shell elements for TRC Pavilion [60] Fig. 12 TRC shell elements with six cast elements [60] Fig. 13 TRC hypar-shells at RWTH Aachen [60] Fig. 14 Roof of the bicycle stand made of laminated TRC barrel vault shells at RWTH Aachen [60] Fig. 15 TRC facades with AR glass Textiles at RWTH Aachen University [65] Modular precast TRC sandwich wall panels have been used to construct low-cost housing in Germany and many other countries like Japan, China, and Italy. This technology applies to many regions of the world and is particularly attractive in places where steel and timber are limited [50]. Cast-in-place construction using Textiles as reinforcement with fully integrated molds is attractive for developing energy-efficient, economical, and sustainable infrastructure in India [49, 67, 68]. Conclusion and way forward Prime Minister Awas Yojana (PMAY), which promises to provide “Housing for All” within a constrained timeframe by 2022, has presented several difficulties to engineers, decision-makers, and administrators everywhere. Due to these difficulties, several cutting-edge technologies have emerged in the building sector to accomplish mass housing plans and schemes like PMAY. The critical criteria that drive the development of innovative technologies for low-cost housing are speedy construction, economical cost, ease of construction, quality of construction, and sustainability as a significant concern. Even in current conventional construction practices, structural and functional requirements plays a critical factor for all currently existing and developing technologies; each one is unique due to its conditions to adapt to various geographic, climatic, and other exposure situations. However, major factors that determine which specific form of technology will be used to achieve low-cost mass housing programs were suggested by various researchers. Through this effort, it has also become clear that current technologies are more focused on individual building components than the entire building system in one go. The main area of this research work is to identify the problem of slow implementation of urban housing in India and propose a feasible solution to it. It aims to help the Government of India PMAY-U scheme to complete its unaccomplished tasks of “Housing for all” at the earliest by incorporating alternative, affordable, low-cost construction materials with higher durability and climatic resistance. This article reviews many modern building technologies adopted in various states in India with locally available raw materials. The study report attempts to offer a feasible and sustainable solution to the housing requirements issue based on its findings, it is evident that Precast lightweight textile reinforced concrete panels will be the best promising choice. Textile-reinforced composites have drawn a lot of interest as strong, durable, high-performance, lightweight materials for various cutting-edge applications, including aerospace, construction, etc. Textile composites will be used more often in a wider range of applications because of recent advancements in fiber and matrix systems, interface modification techniques, and nanotechnology. In future, textile materials are anticipated to play a significant role in reinforcing sophisticated composite materials In view of all the above applications, for low-cost modular Housing for EWS communities in India, Textile Reinforced Concrete (TRC) is the most effective and feasible solution. With the review of various literatures, it is been observed that very limited research is carried out to adopt TRC as a structural component for infrastructure development. Hence, our current research focuses on developing structural components of TRC and proposing its use for rural housing in India. Author contributions SI, Ph.D. scholar at NIT, Trichy, has written the original draft. Prof. KB has reviewed and supervised the draft. Funding Sophia Immanuel reports financial support for the Ph.D. fellowship (MHRD) is provided by the National Institute of Technology Tiruchirappalli for carrying out this research. Declarations Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Ethical approval This article does not contain any studies involving human participants directly performed by any authors. The data mentioned are from the Government of India Portal used for these studies. This is purely a review paper. Hence it reviews various technologies implemented in India and proposes a new solution to the problem. Informed consent None. ==== Refs References 1. Gandhi S Economics of affordable housing in Indian cities: the case of Mumbai Environ Urban ASIA 2012 10.1177/097542531200300112 2. Sarkar A Dhavalikar U Agrawal V Morris S Examination of affordable housing policies in India Business and Management Horizons 2016 10.5296/bmh.v4i1.9508 3. Sengupta U Affordable housing development in India: a real deal for low-income people? Int Dev Plan Rev 2013 10.3828/idpr.2013.17 4. Agarwal A, Jain V, Karamchandani A (2013) Lead sponsors: Michael & Susan Dell Foundation and Mahindra Lifespaces Sponsors: Department for international development and Ambuja cement co-sponsors: Muthoot Housing Finance and Dewan Housing Finance corporation limited state of the low-income housing market encouraging progress & opportunity to realize dreams of millions 5. Portal NW (2015) Usability of textile reinforced concrete: structural performance, durability and sustainability 6. Djamai ZI Bahrar M Salvatore F Textile reinforced concrete multiscale mechanical modelling: application to TRC sandwich panels Finite Elem Anal Des 2017 10.1016/j.finel.2017.07.003 7. Zhang Z Provis JL Reid A Wang H Geopolymer foam concrete: an emerging material for sustainable construction Constr Build Mater 2014 56 113 127 10.1016/j.conbuildmat.2014.01.081 8. Gavali HR Bras A Faria P Ralegaonkar RV Development of sustainable alkali-activated bricks using industrial wastes Constr Build Mater 2019 215 180 191 10.1016/j.conbuildmat.2019.04.152 9. D’Souza R (2019) Housing poverty in urban India: the failures of past and current strategies and the need for a new blueprint 10. Ahmad S (2015) Housing poverty and inequality in urban India. In: Economic studies in inequality, social exclusion and well-being 11. Housing MOF, Alleviation UP, Organisation NB (2012) Government of India ministry of housing and urban poverty alleviation national buildings organisation www.mhupa.gov.in. Group (New York) 12. Sankhe S, Vittal I, Dobbs R, et al (2010) India’s urban awakening: building inclusive cities, sustaining economic growth. McKinsey Quarterly 13. Kundu A (2022) Housing for all: from rhetoric to a reality check on slum statistics for operationalization of flagship missions in India. In: Accessible housing for South Asia 14. Bredenoord J Sustainable building materials for low-cost housing and the challenges facing their technological developments: examples and lessons regarding bamboo, earth-block technologies, building blocks of recycled materials, and improved concrete panels J Archit Eng Technol 2017 10.4172/2168-9717.1000187 15. BMTPC and Mohua (2018) Compendium of prospective emerging technologies for mass housing. Building Materials & Technology Promotion Council 16. Keswani K The contribution of building centres to low-cost housing in India Build Res Inf 1997 10.1080/096132197370606 17. Roshan R Sivapriya SV Understanding sustainable strategies in low-cost housing Int J Hous Mark Anal 2020 10.1108/IJHMA-05-2019-0050 18. Mastrucci A Rao ND Bridging India’s housing gap: lowering costs and CO2 emissions Build Res Inf 2019 10.1080/09613218.2018.1483634 19. Torres-Sanchez C Huang C Steel G Optimization of assembly instructions for a low-cost housing solution Inf Des J 2016 10.1075/idj.22.1.04tor 20. Kulshreshtha Y Mota NJA Jagadish KS The potential and current status of earthen material for low-cost housing in rural India Constr Build Mater 2020 10.1016/j.conbuildmat.2020.118615 21. Sinha S Mishra S Kumar P Saurabh S Sustainable low-cost housing using cost effective construction technology “Rat trap bond masonry” and “Filler roof slab” in Bihar Int J Eng Res Technol 2020 10.37624/ijert/13.7.2020.1780-1785 22. Kumar S A perspective study on fly ash-lime-gypsum bricks and hollow blocks for low cost housing development Constr Build Mater 2002 10.1016/S0950-0618(02)00034-X 23. Jasvi AH sustainable use of low cost building materials in the rural India Int J Res Eng Technol 2015 10.15623/ijret.2015.0425077 24. Singh MR, Naskar SD (2020) Smart construction: case of ‘3-S’ Prefab technology for sustainable mass housing. In: Lecture notes in civil engineering 25. Raj PV Teja PS Siddhartha KS Rama JSK Housing with low-cost materials and techniques for a sustainable construction in India-a review Mater Today Proc 2020 43 1850 1855 26. Garg AA, Yalawar A, Kamath A, Vinay J (2014) Effect of varying cement proportions on properties of compressed stabilized earth blocks (CSEB) - a sustainable low-cost housing material. In: International conference on sustainable civil infrastructure, ASCE India Section, October 17–18, 2014, Hyderabad, India 27. Khadka B Rammed earth, as a sustainable and structurally safe green building: a housing solution in the era of global warming and climate change Asian J Civ Eng 2020 10.1007/s42107-019-00202-5 28. Tarafdar SS (2016) Facade: sustainable facade materials lead to green buildings. Athena Information Solutions Pvt Ltd 29. Mamat R, Rahim JA, Hamzah SH (2015) Behaviour of unreinforced expanded polystyrene lightweight concrete (EPS-LWC) wall panel enhanced with steel fibre. J Eng Sci Technol 10 30. Lavanya B Kuriya PD Suganesh S Properties of geopolymer bricks made with flyash and GGBS IOP Conf Ser Mater Sci Eng 2020 872 012141 10.1088/1757-899X/872/1/012141 31. Senthil K, Rupali S, Pratap A, et al (2020) Behaviour of concrete brick and flyash brick on infilled frame under cyclic loading. Ecol Environ Conserv 26: 32. Sisodia H, Sharma P (2017) A review on utilization aspects of natural and artificial fiber in flyash brick. Int J Mech Eng Technol 8: 33. Vinod BR Shobha R Vikas C Compressed stabilized mud block using alccofine and sodium meta silicate Eng Sci Int J 2021 10.30726/esij/v7.i4.2020.74020 34. Subramanian GKM Balasubramanian M Jeya Kumar AA A review on the mechanical properties of natural fiber reinforced compressed earth blocks J Nat Fibers 2021 19 7687 7701 10.1080/15440478.2021.1958405 35. Vinayak M, Nitish M, Siddarth Gowda HM, et al (2021) Studies on properties of compressed mud blocks. In: Lecture notes in civil engineering 36. Naaman AE (2010) Textile reinforced cement composites: competitive status and research directions. In: International RILEM conference on material science 37. Chowdhury S Roy S Prospects of low cost housing in India Geomaterials 2013 10.4236/gm.2013.32008 38. Hegger J, Will N, Chudoba R, Scholzen A (2015) Classification of approaches to dimensioning and assessment of lightweight Trc structures. In: FERRO-11–11th international symposium on ferrocement and 3rd ICTRC-international conference on textile reinforced concrete 39. Colombo IG, Colombo M, di Prisco M, et al (2018) TRC sandwich panel for energy retrofitting exposed to environmental thermal actions. In: American Concrete Institute, ACI Special Publication 40. D'Orazio M Stipa P Sabbatini S Maracchini G Experimental investigation on the durability of a novel lightweight prefabricated reinforced-EPS based construction system Constr Build Mater 2020 252 119134 10.1016/j.conbuildmat.2020.119134 41. Wayal A Kumthekar MB Aerocon blocks and panels: sustainable and cost effective material for speedy construction SSRN Electron J 2019 10.2139/ssrn.3366741 42. Jaiganesh K Dinesh S An investigation of using aerocon block and M-sand in constructing low cost housing IOSR J Mech Civ Eng 2017 10.9790/1684-1403017075 43. Chithambaram SJ Kumar S Flexural behaviour of bamboo based ferrocement slab panels with flyash Constr Build Mater 2017 134 641 648 10.1016/j.conbuildmat.2016.12.205 44. Mali PR Datta D Experimental evaluation of bamboo reinforced concrete slab panels Constr Build Mater 2018 188 1092 1100 10.1016/j.conbuildmat.2018.08.162 45. Sreadha AR Pany C Review on fabrication of bamboo composite materials reinforced concrete J Sci Tech 2020 5 3 258 279 46. RILEM TC 232-TDT (2016) Test methods and design of textile reinforced concrete. Mater Struct 49: 47. Scholzen A Chudoba R Hegger J Thin-walled shell structures made of textile-reinforced concrete: part I: structural design and construction Struct Concr 2015 16 1 106 114 10.1002/suco.201300071 48. Döbrich O Gereke T Cherif C Modeling the mechanical properties of textile-reinforced composites with a near micro-scale approach Compos Struct 2016 135 1 7 10.1016/j.compstruct.2015.09.010 49. Alexander AE Shashikala AP Sustainability of construction with textile reinforced concrete-a state of the art IOP Conf Ser Mater Sci Eng 2020 936 012006 10.1088/1757-899X/936/1/012006 50. Mesticou Z Bui L Junes A Larbi AS Experimental investigation of tensile fatigue behaviour of textile-reinforced concrete (TRC): effect of fatigue load and strain rate Compos Struct 2017 160 1136 1146 10.1016/j.compstruct.2016.11.009 51. Karaduman NS, Karaduman Y, Ozdemir H, Ozdemir G (2017) Textile reinforced structural composites for advanced applications. In: Textiles for advanced applications 52. Butler M, Lieboldt M, & Mechtcherine V (2010). Application of textile-reinforced concrete (TRC) for structural strengthening and in prefabrication. In: Proceedings of the international conference on advanced concrete materials (ACM), Stellenbosch, South Africa (pp. 125–134) 53. Stolyarov O Quadflieg T Gries T Effects of fabric structures on the tensile properties of warp-knitted fabrics used as concrete reinforcements Text Res J 2015 85 18 1934 1945 10.1177/0040517515578334 54. Guo S Ren J Yang T Influences of surface treatment on the mechanical performances of carbon and basalt textiles-reinforced concretes under harsh environments Compos B Eng 2022 246 110195 10.1016/J.COMPOSITESB.2022.110195 55. Garcia D Larrinaga P San-Jose JT Garmendia L Ummantelung des Betons mit Basalttextilmatten bewehrtem Mörtel/confinement of concrete with basalt textile reinforced mortar Restorat Build Monum 2010 16 4–5 341 352 10.1515/rbm-2010-6392 56. Du Y Zhang X Zhou F Zhu D Zhang M Pan W Flexural behavior of basalt textile-reinforced concrete Constr Build Mater 2018 183 7 21 10.1016/j.conbuildmat.2018.06.165 57. Shamseldein A ELgabbas F Elshafie H Tensile behavior of basalt textile-reinforced mortar (BTRM) Ain Shams Eng J 2022 10.1016/j.asej.2021.05.003 58. Alan Strauss Rambo D Umbinger de Oliveira C Pícolo Salvador R Sisal textile reinforced concrete: improving tensile strength and bonding through peeling and nano-silica treatment Constr Build Mater 2021 10.1016/j.conbuildmat.2021.124300 59. Zhu D Liu S Yao Y Effects of short fiber and pre-tension on the tensile behavior of basalt textile reinforced concrete Cement Concr Compos 2019 10.1016/j.cemconcomp.2018.11.015 60. Scheerer S Chudoba R Garibaldi MP Curbach M Shells made of textile reinforced concrete -applications in Germany J Int Assoc Shell Spatial Struct 2017 10.20898/j.iass.2017.191.846 61. Udhayaraman R Subramanian H Mulay SS Venkatachalam S Multi-scale approach-based studies on the damage-healing and fracture behavior of plain woven textile composite Mech Adv Mater Struct 2022 10.1080/15376494.2020.1809037 62. Engberts, E. (2006). Large-size facade Elements of textile reinforced concrete. In ICTRC'2006–1st International RILEM conference on textile reinforced concrete (pp. 309–318). RILEM Publications SARL 63. Junes A Larbi AS An indirect non-linear approach for the analysis of sandwich panels with TRC facings Constr Build Mater 2016 112 406 415 10.1016/j.conbuildmat.2016.02.190 64. Colombo IG, Colombo M, di Prisco M, Salvalai G, & Sesana MM (2018). TRC sandwich panel for energy retrofitting exposed to environmental thermal actions. In: 2nd international workshop on durability and sustainability of concrete structures, DSCS 2018 (pp. 1–10). American Concrete Institute, ACI Special Publication 65. Hegger J, Zell M, & Horstmann M (2008). Textile reinforced concrete–realization in applications. In: proceedings: international fib symposium tailor made concrete structures: new solutions for our society (pp. 357–362) 66. Karaduman NS, Karaduman Y, Ozdemir H, & Ozdemir G (2017). Textile reinforced structural composites for advanced applications. In: Textiles for advanced applications, 87 67. Peled A, Mobasher B, & Bentur A (2017). Innovative applications of textile reinforced concrete (TRC) for sustainability and efficiency. In: Textile reinforced concrete (pp. 451–464) 68. Laiblová L Pešta J Kumar A Hájek P Fiala C Vlach T Kočí V Environmental impact of textile reinforced concrete facades compared to conventional solutions—LCA case study Materials 2019 12 19 3194 10.3390/ma12193194 31569478	0	PMC9717557	NO-CC CODE	2022-12-06 23:23:25	no	Innov. Infrastruct. Solut.. 2023 Dec 2; 8(1):39	utf-8	null	null	null	oa_other
==== Front Water Air Soil Pollut Water Air Soil Pollut Water, Air, and Soil Pollution 0049-6979 1573-2932 Springer International Publishing Cham 5984 10.1007/s11270-022-05984-0 Article Municipal Wastewater Treatment uses Vertical Flow Followed by Horizontal Flow in a Two-Stage Hybrid-Constructed Wetland Planted with Calibanus hookeri and Canna indica (Cannaceae) Singh Krishna Kumar [email protected] Vaishya Rakesh Chandra grid.419983.e 0000 0001 2190 9158 Department of Civil Engineering, Motilal Nehru National Institute of Technology Allahabad, Prayagraj, 211004 India 1 12 2022 2022 233 12 51013 8 2022 22 11 2022 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The utilization of hybrid-constructed wetland systems has recently expanded due to more rigorous municipal wastewater discharge and also complex wastewaters treated in hybrid-constructed wetlands (HCWs). A lab-scale two-stage experimental setup of vertical flow followed by horizontal flow hybrid-constructed wetland (VFHCW-HFHCW) configuration was built. First-stage vertical flow hybrid-constructed wetland reactor with the surface area was 1963.49 cm2 and second-stage horizontal flow hybrid-constructed wetland reactor with the surface area was 2025 cm2. The HCW unit was planted with two type plants: Calibanus hookeri and Canna indica (Cannaceae). Influent Municipal wastewater flow rate 112.32 l/day, hydraulic loading rate (HLR) 0.55 m/day, and hydraulic retention time of 1 day. The efficiency was evaluated in municipal wastewater quality improvement and physico-chemical analysis in our laboratory. The removal rate after the second-stage horizontal flow of BOD3 at 27 °C, COD, TSS, TP, NH3-N, and NO3-N reached 92.75%, 89.90%, 85.45%, 88.83%, 99.09%, and 96.05%, respectively. The results shown after both stage hybrid-constructed wetland VFHCW-HFHCW, treated effluent of Municipal wastewater produced high-quality effluent which may be reused in gardening, agriculture, and flushing in toilet purpose according to Bureau of Indian Standards (BIS) code for practices. However, in the future, hybrid-constructed wetlands could be standards design criteria developing and enhancing the performance standards and economic meets both to make more popular technology of the hybrid-constructed wetland (HCW). Supplementary Information The online version contains supplementary material available at 10.1007/s11270-022-05984-0. Keywords Hybrid-constructed wetlands Vertical flow Horizontal flow Municipal wastewater treatment Reuse issue-copyright-statement© Springer Nature Switzerland AG 2022 ==== Body pmcIntroduction All over the world, many countries face freshwater problems, and even India is not exempted. As per the CPCB (Central Pollution Control Board) study, India’s present scenario is approximately 72,368 MLD Municipal wastewater generation all over countries, and approximately 40,527 MLD untreated Municipal wastewater generation is discharged directly to surface water bodies (CPCB, 2021). It has led to deteriorating aquatic life of surface water quality, threatened human health, environment, and major water-borne diseases. India faces a problem hues gap exists between treated Municipal wastewater and untreated Municipal wastewater because of lack of funding and management create this large gap (CPCB, 2019 & ENVIS n.d.). In view of capital cost limitation and minimizing large gap can use the less capital cost and less maintenance technique for the treatment of Municipal wastewater. Various advanced municipal wastewater treatment technologies are the solution in this context. Some conventional treatment technology such as stabilization pond, anaerobic filter, green filter, septic tank, sand filtration, and activated sludge process are still in use. Nowadays, a new low capital cost, less maintenance cost, and eco-friendly constructed wetland (CWs) is becoming popular for the municipal wastewater treatment (Abidi et al., 2009; Vymazal & Masa, 2003; Molle et al., 2008; Kouki et al., 2009; Brix0 & Arias, 2005; Zhang et al., 2014). Constructed wetlands (CWs) are the new artificially engineered system that remove the pollutants load from various types of wastewater (Masi et al., 2002; Brix & Arias, 2005; Vymazal, 2005; Kouki et al., 2009; Dotro et al., 2015). It has been found to effectively remove pollutants like organic and inorganic contamination, nutrients, and pathogens as well as transmitted virus in wastewater bodies and pathogens (Lesage, 2006 & 2007; Keffala & Ghrabi, 2005; Tanner et al., 2012; Haiming et al., 2013; Dotro et al., 2015). Nowadays, mainly horizontal flow or vertical flow hybrid-constructed wetlands (HFHCW or VFHCW) are more used individually for the treatment of municipal and industrial wastewaters (Karathanasis et al., 2003; García et al. 2005; Wang et al., 2012; Abou-Elela & Hellal, 2012). Horizontal flow or vertical flow hybrid-constructed wetlands (HFHCW or VFHCW) have been successfully used in industries like fertilizer, textile, dairy, tannery, food and beverages, and pulp & paper etc. to remove varieties of pollutants (Abou-Elela & Hellal, 2012; Zurita et al., 2009; Yaseen & Scholz, 2019; Avila, 2020). The removal mechanism may include ion exchange, soil adsorption, and uptake by plants, chemical precipitation, and anaerobic or aerobic microbial growth or decomposed activity (Karathanasis et al., 2003; Keffala & Ghrabi, 2005; Klomjek & Nitisoravut, 2005; Molle et al., 2008; Lesage, 2006; Zhang et al., 2009). Constructed wetland (CW) treatment performance can be increased using anaerobic and aerobic processes (Lesage, 2006; Lesage et al., 2007). The study focuses on the development of one lab scale at two-stage vertical flow followed by horizontal flow hybrid-constructed wetland (VFHCW-HFHCW). In this configuration, the first-stage VFHCW is circular shape filled with integrated gravel, sand, and soil for better organic and inorganic nutrient, phosphate uptake, and oxidation of ammonia under aerobic condition; the second-stage HFHCW is square shaped filled with integrated gravel, sand, cold drink plastic bottle chips, and soil for effective de nitrification for better removal of nitrogen, BOD, COD, and TP respectively. The configured two-stage VFHCW-HFHCW were planted with two different kinds of plant species, such as Calibanus Hookeries and Canna Indica red color (Cannaceae). These wetlands were used to treat actual municipal wastewater which collected from the outside of Motilal Nehru National Institute of Technology, Allahabad Prayagraj, India, campus. The developed hybrid-constructed wetland system’s performance was evaluated for 4 months in order to evaluate actual municipal wastewater treatment performance. By introducing new techniques, this study addresses the few key concerns with the technologies currently in use for treating municipal wastewater: (i) reducing the cost of an external mechanical system-based aeration with algal-based passive aeration for pollutant oxidation, algal uptake of pollutants, and developing energy-efficient processes; (ii) replacing various costly advanced STPs techniques with cheap and eco-friendly hybrid integrated layer of constructed wetland; (iii) instead of independent and segregated treatment approaches, a synergistic approach to fully treating organic and nutrient pollutants in a single system is preferred; and (iv) zero sludge generation from wastewater in this treatment system. Also at present, there is lack of proper guidelines in our country India for the design criteria, flow regulation, and hydraulic retention time (HRTs) for the constructed wetland process. If these guidelines are in place then in future more and more ULBs and small town areas can select the most suitable hybrid-constructed wetlands configuration with eco-friendly technology for the treatment of sewage and grey wastewater. Materials and Methods Experimental Setup A lab-scale two-stage experimental setup of vertical flow followed by horizontal flow hybrid-constructed wetland (VFHCW-HFHCW) configuration consisted of one circular and one square polycarbonate compact transparent 6-mm-thickness Sheet (PCTS). PCTS has high-impact strength, high-temperature resistance, and ultraviolet (UV) protection. The circular box has an inside diameter of 50 cm and depth of 55 cm, and the square box has inside dimensions length, width, and depth 45 cm each. In circular vertical flow and square horizontal flow, hybrid-constructed wetland reactor layered with the gavel of size 16–20 mm, 10 mm, and 4.75 mm; sand of grade 2.36 mm, 1.18 mm, 600 µm, and 300 µm; and with soil from the bottom to top in Fig. 1 (CPCB, 2019). Before filling the gravel in both stages of the wetland reactor, gravel is properly washed by de-ionized water. In second-stage wetland placed a layered of cold drink plastic bottles chips between 300 µm sand and soil. Flow path from vertical circular to the horizontal square stage to be maintained by gravimetric flow and from vertical circular stage to horizontal square stage connected with polyvinyl chloride pipes (PVC) of diameter 14 mm. The dimensions and operating conditions of hybrid-constructed wetland reactors are given in Table 1.Fig. 1 Lab-scale two-stage experimental setup of hybrid-constructed wetland Table 1 Dimensions and operating conditions of hybrid-constructed wetland reactors Stage Type Diameter/length × width of flow (cm) Depth (cm) Height of flow (cm) Surface area (cm2) Temperature (0C) 1st Circular VFCHCW 50 55 15 1963.49 15–42 2nd Square HFSHCW 45 × 45 45 15 2025 15–42 Effective Volume of Hybrid-Constructed Wetland Bed and Plantation The total volume of the first-stage circular VFHCW reactor was 107.99 L and the second-stage square HFHCW reactor was 91.125 L. In the hybrid-constructed wetland, the actual volume of wastewater was filled in the wetland reactor determined by taking a known volume of wastewater and filling the hybrid-constructed wetland bed and knowing the quantity of wastewater filled until the flows first drop through the outlet pipe. The adequate volume in the first-stage circular VFHCW reactor was 28.45 L and the second-stage square HFHCW reactor was 28.37 L. The two different vegetation plant species used, such as Calibanus hookeri and Canna indica red color (Cannaceae), were available in the Motilal Nehru National Institute of Technology Allahabad (MNNIT), Prayagraj India Campus. The vegetation plant can absorb a high level of pollutant and organic load from municipal wastewater. The vegetation plants roots were washed with the deionized water 2–3 times. Both reactors were planted with 4 Calibanus hookeri and 2 Canna indica red color (Cannaceae), with heights of Calibanus hookeri in between 20 and 35 cm and Canna indica red color (Cannaceae) in between 30 and 45 cm respectively (Brix & Arias, 2005; Tanveer Saeed et al., 2021). Sampling and Physico-chemical Analysis Municipal wastewater was collected from Prayagraj city, India at latitude 25° 29′ 40.8372″ N, and longitude 81° 51′ 53.2044″ E every week and stored in a tank. After that, municipal wastewater was pumped through Watson Marlow peristaltic pump at the rate of 25–35 rpm 8 h daily in these reactors. Inlet raw wastewater and treated wastewater samples were collected daily from stage 1 outlet and stage 2 outlet of the hybrid wetland reactor in Fig. 1. The collected samples were analyzed on regular basis for almost 4 months in the first phase from February 2021 to May 2021. Due to second-wave COVID-19, Institute was closed officially during June 2021 to August 2021, that’s why the experimental analysis stopped. Therefore, during the rainy season, the analysis was not shown here in the research. The samples were examined for inlet raw wastewater and treated wastewater through both stages. The physico-chemical parameters analysis are pH, total suspended solids (TSS), biochemical oxygen demand (BOD3 at 27 °C), chemical oxygen demand (COD), total phosphate (TP), nitrate nitrogen (NO3−-N), and ammonia nitrogen (NH4-N). The pH value was measured using HI 2210 pH meter. TSS value was calculated using Matrix eco solution-111 Hot air oven followed by Wensar electronic balance, BOD3 at 27 °C was measured using MKSI BOD incubator, and COD was measured using HACH company closed reflux COD meter. While TP and NO3−-N were measured by using LAB INDIA analytical UV/VIS double beam spectrophotometer, and NH4-N was measured by using universal Kjeldahl digestion and distillation apparatus. All the physico-chemical parameters analysis procedure was followed by standard methods for examining water and wastewater, 23rd edition (APHA 2017). All the experimental data analysis was carried using Microsoft Excel 2013 version and Origin Pro 2021b version. The removal efficiency of the pollutant in percentage as calculated by following Eq. (1).1 %R=ci-ceci where Ci initial concentration in mg/l and Ce effluent concentration in mg/l. Results and Discussion Influent Municipal Wastewater Characterization In this research, the minimum, maximum, and average characterization of influent municipal wastewater to the hybrid-constructed wetland are shown in the Table 2. The characterization results indicated that the influent concentration value of BOD3 at 27 °C, COD, TSS, TP, NO3−-N, and NH4-N varied during the study period. The pH value and organic loading rate also varied during the study period. The influent concentration value of heavy metals such as fluoride, iron, and chromium was measured during the study period but did not exceed 0.01 mg/l. The analysis was conducted after achieving a stable removal rate in the second week of February 2021.Table 2 Characteristics of influent municipal wastewater of four month operation S.N Parameters unit Minimum value Maximum value Average value + St. dv 1 pH – 7.45 8.3 7.89 ± 0.35 2 BOD3 at 27 °C mg/l 49.8 119.2 103.75 ± 29.75 3 COD mg/l 224 800 546.47 ± 235.71 4 TSS mg/l 140 597 275.75 ± 191.62 5 TP mg/l 5.38 30.72 15.138 ± 10.44 6 NO3−-N mg/l 3.694 7.573 4.478 ± 3.094 7 NH4-N mg/l 2.52 17.96 13.474 ± 3.776 pH Analysis The observed pH values in the two-stage hybrid-constructed wetland in which VFHCW followed HFHCW are shown in Fig. 2 and data were given in the SI Table 1. The average pH of influent municipal wastewater was at about 7.89 ± 0.35, which decreased slightly in the effluent of first-stage circular VFHCW to 7.6 and also slightly in the effluent of second-stage square HFHCW to 7.59. The pH of integrated sand media was 7.54 for both stages of HCW. The formation of volatile fatty acids as a result of anaerobic breakdown of complex organics present in wastewater by microbes resulted in the lower average pH of VFHCW and HFHCW. The accumulation of protons produced during organic matter oxidation also contributed to the pH decrease. It may happen to start the degradation process of municipal wastewater due to increment in alkalinity.Fig. 2 pH value of influent municipal wastewater and effluent of VFHCW and HFHCW BOD3 at 27 °C and COD Removal Study The aerobic microbial degradation and sedimentation processes are thought to be responsible for BOD and COD removal in vegetated submerged wetlands. Microbial growth on media surface removes soluble organic compounds, which are then attached to plant roots and rhizomes. Figures 3 and show the BOD3 and COD concentration in two-stage VFHCW followed HFHCW effluents Fig. 4. The concentration value of BOD and COD of influent, effluent of VFHCW, and effluent of HFHCW was shown in SI Table 2 and SI Table 3. In terms of BOD3 and COD, the results reveal that lab-scale setup units are very effective at removing contaminants. However, the average surface organic loading rate in the two-stage hybrid wetland was 26.35 g BOD3/m2/day and 144.11 g COD/m2/day respectively while in the second stage, square HFHCW removal efficiency was reached 92.75% and 89.90% with an average treated effluent concentration 3.82 mg/l and 45.62 mg/l for BOD3 and COD respectively. The BOD3 and COD removal efficiency was found to be stable beginning with the fourth and third weeks of operation of the first-stage VFHCW, and beginning with the end of the second and first weeks of operation of the second-stage HFHCW. At this period, plant growth in the first-stage VFHCW was 35–56 cm for Canna Indica red color and 22–32 cm for Calibanus hookeries, and 15–26 cm for Canna Indica red color and 6–14 cm for Calibanus hookeries in the second-stage HFHCW. This effective removal depends on the amalgamation of physical and microbial activity. Because in a hybrid-constructed wetland, the physical phenomenon mechanism allows filtering the water through low porosity of constructed wetland. The solid organic traps through the bed for the long hydraulic retention time, so that these organic traps in the presence of sunlight and through soil and vegetation plant allow to biodegradation of the organic matter (Thalla et al., 2019). These removal rates are high because of the retention of organic and inorganic solid materials on the topsoil bed and rapid decomposition in aerobic conditions (Yaseen & Scholz, 2019; Karathanasis et al., 2003). Some organic solid with the wastewater through the low porosity of gravel settles down to the bed of the hybrid-constructed wetland, and it decomposes in anaerobic conditions and through the roots of the vegetable plants, in the both aerobic and anaerobic conditions removal of organic matter and reduction by bacteria and microbes taken place (Karathanasis et al., 2003; Zhang et al., 2014). From the other studies of wetlands, our two-stage hybrid-constructed wetland shows better BOD and COD removal efficiency.Fig. 3 BOD3 at 27 °C concentrations in influent municipal wastewater and effluent of VFHCW and HFHCW Fig. 4 COD concentrations in influent municipal wastewater and effluent of VFHCW and HFHCW Table 3 Vegetation plant growth measured in two-stage hybrid-constructed wetland during the operational period Month Week Canna indica height (in cm) Calibanus hookeries height (in cm) VFHCW HFHCW VFHCW HFHCW Februry 2021 1–2 10–18 8–15 4–12 2–6 3–4 18–35 15–26 12–22 6–14 March 2021 1–2 35–56 26–45 22–32 14–22 3–4 56–83 45–68 32–48 22–40 April 2021 1–2 83–98 68–89 48–70 40–65 3–4 98–115 89–102 70–88 65–82 May 2021 1–2 115–132 102–125 88–104 82–98 3–4 132–158 125–149 104–112 98–107 Total Suspended Solid Removal Study Removing total suspended solid (TSS) is a significant physical phenomenon for treating municipal wastewater, in contrast to the results obtained in the reduction of BOD and COD for two-stage lab-scale VFHCW and HFHCW. The concentrations of total suspended solids (TSS) in the two-stage hybrid-constructed wetland circular VFHCW followed by square HFHCW effluent and influent of the Municipal wastewater are shown in SI Table 4 and graph was shown in Fig. 5. The results were observed that the average removal rate in percentage in the first-stage circular VFHCW was reached to 75.09% with an average treated effluent quality of suspended solid reached to 96.47 mg/l while in the second stage, square HFHCW was reached to 85.45% with an average treated effluent quality of suspended solid 37.32 mg/l. These effective removal of suspended solids filtered through the surface flow and roots of the Calibanus hookeries and Canna indica red color plants in the hybrid-constructed wetland. Physical processes like sedimentation and filtration, which are used in accordance with aerobic and anaerobic microbial degradation inside the substrate, are the main methods used to remove TSS (Spangler et al., 2019; Avila, 2020). For the lab-scale plant, the integrated graded gravel, sand, and soil substrate along with the vegetation improved treatment efficiency (Abou-Elela & Hellal, 2012; Lesage et al., 2007). The high TSS removal rates found in this study are comparable to those found in previous research. These results were significant because of physical phenomenon for removal of solid and small particles settling plant’s stems and roots play a significant role.Fig. 5 TSS concentrations in influent municipal wastewater and effluent of VFHCW and HFHCW Removal Study of Total Phosphorous The physico-chemical process associated with phosphorous removal mechanism in a hybrid-constructed wetland occurred by precipitation with metals, adsorption from the substrate, and vegetable plant roots taken for the growth (Lesage et al., 2007; Molle et al., 2008). Phosphorus that is soluble will flow with the flow in subsurface flow wetlands, but phosphorus that is linked to particulate matter will be caught and removed by filtration and interception systems built into the wetland bed (Spangler et al., 2019; Avila, 2020). In the lab, the phosphorous is measured as PO4-P, and concentration results were shown in SI Table 5 and graph shown in Fig. 6 shows increment in the phosphorus removal rate indicates the biological activity as the substrate and algal uptake for the growth. However, decreased phosphorus removal rate indicates that adsorption takes over the sites with an increase in time. The total phosphorous (TP) concentration in a two-stage hybrid-constructed wetland circular VFHCW followed by square HFHCW effluent, and influent of the Municipal wastewater is shown in Fig. 6. Moreover, the results were observed that the average removal rate in percentage in the first-stage circular VFHCW was reached to 80.84% with an average treated effluent quality of suspended solid reached to 2.52 mg/l while in the second stage, square HFHCW was reached to 88.83% with an average treated effluent quality of suspended solid 1.56 mg/l. Thus for the phosphorous removal, the role of vegetation plants and oxygen in a two-stage hybrid-constructed wetland circular VFHCW followed by square HFHCW is most appropriate Fig. 7 and Fig. 8. Fig. 6 TP concentrations in influent municipal wastewater and effluent of VFHCW and HFHCW Fig. 7 NH4-N concentrations in influent municipal wastewater and effluent of VFCHCW and HFSHCW Removal study of Ammonia Nitrogen (NH3-N) and Nitrate Nitrogen (NO3.−-N) The sewage wastewater contains one of the significant pollutants, nitrogen, which can cause the toxicity effect to the surviving aquatic organism. In sewage wastewater, there exists inorganic and organic forms. Nitrogen in the inorganic forms is nitrate (NO3−), nitrite (NO2−), ammonium (NH4), and in the gaseous form of nitrous oxide (N2O), nitrogen gas (N2), and free ammonia (Masi et al., 2002; Karathanasis et. al., 2003; Effendi et al., 2020). Although in the organic form of nitrogen are urea, peptide in amino acid forms. In the hybrid-constructed wetland, nitrogen removal was done by transforming biological processes such as nitrification, denitrification ammonification, reduction of nitrate, assimilation of biomass matter, and uptake by plant roots (Karathanasis et al., 2003; Wang et al., 2012; Haiming et al., 2013). The transformation of nitrogen is shown in the schematic diagram Fig. 9.Fig. 8 NO3-N concentrations in influent municipal wastewater and effluent of VFHCW and HFHCW Fig. 9 Nitrogen transformation in hybrid-constructed wetland The ammonia nitrogen and nitrate nitrogen concentrations are variation of effluent and influent of the sewage wastewater in the two-stage hybrid-constructed wetland circular VFHCW followed by square HFHCW shown in SI Table 6 and SI Table 7 respectively and graph was shown in the Fig. 7 and Fig. 8. The results were observed in the two-stage hybrid wetland with effective removal of inorganic nitrogen in terms of ammonia nitrogen and nitrate nitrogen. The average removal rate of ammonia nitrogen through Kjeldahl nitrogen and nitrate nitrogen through absorption process in the percentage in first-stage circular VFCHCW reached to 94.64% and 88.63% with an average treated effluent value of 0.64 mg/l and 0.613 mg/l respectively shown in Fig. 7 and Fig. 8. In contrast, square HFHCW removal rate was reached to 99.09% and 96.05% in the second stage with an average treated effluent value of 0.105 mg/l and 0.21 mg/l respectively shown in Fig. 7 and Fig. 8. The removal performance results were better when treated from both-stage hybrid wetlands. Performance studies were done on first-stage circular VFHCW and second-stage square HFHCW planted with two different types of vegetable plants species such as Calibanus hookeries and Canna indica red color. Vegetation in Hybrid-Constructed Wetland Table 3 represents the results of measurements taken during the operational phase of the laboratory-scale hybrid wetland model with two stages of vegetation growth. With that, VFHCW showed significantly higher plant growth compared to HFHCW. It was discovered that the plant’s roots extended all the way down into the substrates. Conclusions The wetland treatment process is generally complicated to comprehend due to the complex physical, chemical, and biological processes involved, as well as variations in real-time wastewater. The major conclusions reached during the study of treatment technology is up-and-coming, and the treated parameters not only pH, BOD3 at 27 °C, COD, and TSS but also removal of TP, NH3-N, and NO3-N for nitrification and de nitrification. The removal rates after the second-stage HFHCW of BOD3 at 27 °C, COD, TSS, TP, NH3-N, and NO3-N reached 92.75%, 89.90%, 85.45%, 88.83%, 99.09%, and 96.05%, respectively. When compared to the VFHCW, the HFHCW had a higher mass removal efficiency. The interaction of vegetation, integrated gravel, sand strata, and substrate influences the entire treatment process. Also, Canna Indica and Calibanus Hookeries, two locally available plants used in the lab scale of two-stage wetland model, demonstrated rapid growth and survival in the treatment wetland bed. BOD, COD, phosphates, and NH3-N removal efficiency were promising and stable during the treatment process, so it can be used on a small scale and low population areas. Constructed wetland process is simplified geographical design and vegetation classification studies should be conducted to provide appropriate guidance for upcoming CWs. Therefore, a hybrid-constructed wetland can be seen as a greener option for the traditional tertiary treatment of domestic wastewater, allowing for its reuse. The low-priced constructed wetland technology can aid in the relief of the current wastewater management problems in developing countries, provided the minimal maintenance requirements, the ease of operation, and the decent removal performance of contaminants. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 39 KB) Abbreviations HCW Hybrid-constructed wetland VFHCW Vertical flow hybrid-constructed wetland HFHCW Horizontal flow hybrid-constructed wetland HLR Hydraulic loading rate OLR Organic loading rate BOD3 Biochemical oxygen demand COD Chemical oxygen demand TSS Total suspended solid TP Total phosphate NH3-N Ammonia nitrogen NO3—N Nitrate nitrogen PCTS Polycarbonate compact transparent sheet VF Vertical flow HF Horizontal flow Author Contribution Krishna Kumar Singh contributed to the study setup installation, methodology, data collection, data analysis, and original manuscript preparation. Rakesh Chandra Vaishya Supervision setup installation and manuscript preparation, resource provided. Data Availability All data generated and analyzed during this study are included in this published article. Declarations Ethical Approval Not applicable. Conflict of Interest The author declare no competing interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abidi S Kallali H Jedidi N Bouzaiane O Hassen A Comparative pilot study of the performances of two constructed wetland wastewater treatment hybrid systems Desalination 2009 246 370 377 10.1016/j.desal.2008.03.061 Abou-Elela SI Hellal MS Municipal wastewater treatment using vertical flow constructed wetlands planted with Canna, Phragmites and Cyperus Ecological Engineering 2012 47 209 213 10.1016/j.ecoleng.2012.06.044 American Public Health Association (APHA) Standard methods for the examination of water and wastewater 2017 23 American Public Health Association/United Book Press Avila FG Treatment of municipal wastewater by vertical subsurface flow constructed wetland: Data collection on removal efficiency using Phragmites Australis and Cyperus Papyrus Data in Brief 2020 30 105584 10.1016/j.dib.2020.105584 32420420 Brix H Arias CA The use of vertical flow constructed wetlands for onsite treatment of domestic wastewater: New Danish guidelines Ecological Engineering 2005 25 491 500 10.1016/j.ecoleng.2005.07.009 CPCB (2019). Constructed Wetland as an Alternative Technology for Sewage Management of India. CPCB (2021). National Inventory Treatment Plant. https://cpcb.nic.in/status-of-stps/. Accessed Sept 2021. Dotro, G., Fort, R.P., Barak, J., Jones, M., Vale, P., Jefferson, B. (2015). Long-term performance of constructed wetlands with chemical dosing for phosphorus removal. In: The Role of Natural and Constructed Wetland in Nutrient Cycling and Retention on the Land Scape. Springer, Cham, 273–292. 10.1007/978-3-319-08177-919 Effendi H Widyatmoko BA Pratiwi Utomo NTM Ammonia and orthophosphate removal of tilapia cultivation wastewater with Vetiveria zizanioides J. King Saud Univ. Sci. 2020 32 207 212 10.1016/j.jksus.2018.04.018 ENVIS Centre of Hygiene, Sani., Sewage Treat. Sys. & Tech. (n.d.). National status of wastewater gen. and treat. http://www.sulabhenvis.nic.in/Database/STST_wastewater_2090.aspx. Accessed Sept 2021. García J Aguirre P Barrag’an J Mujeriego R Matamoros V Bayona JM Effect of key design parameters on the efficiency of horizontal subsurface flow constructed wetlands Ecological Engineering 2005 25 405 418 10.1016/j.ecoleng.2005.06.010 Haiming Wu Zhang J Wei R Liang S Li C Xie H Nitrogen transformations and balance in constructed wetlands for slightly polluted river water treatment using different macrophytes Environmental Science and Pollution Research 2013 20 443 451 10.1007/s11356-012-0996-8 22707115 Karathanasis AD Potter CL Coyne MS Vegetation effects on fecal bacteria, BOD, and suspended solid removal in constructed wetland treating domestic wastewater Ecological Engineering 2003 20 157 169 10.1016/S0925-8574(03)00011-9 Keffala C Ghrabi A Nitrogen and bacterial removal in constructed wetlands treating domestic wastewater Desalination 2005 185 383 389 10.1016/j.desal.2005.04.045 Klomjek P Nitisoravut S Constructed treatment wetland: A study of eight plant species under saline conditions Chemosphere 2005 58 585 593 10.1016/j.chemosphere.2004.08.073 15620752 Kouki S Mhiri F Saidi N Belaid S Hassen A Performances of a constructed wetland treating domestic wastewaters during a macrophytes life cycle Desalination 2009 246 452 467 10.1016/j.desal.2008.03.067 Lesage E. (2006). Behavior of heavy metals in constructed treatment wetlands. Ph.D. thesis, Faculty of Bios. Engineering, Ghent University, Belgium. https://lib.ugent.be/catalog/rug01:001041909. Accessed Sept 2021. Lesage E Rousseau D Meers E Van de Moortel A Du Laing G Verloo MG Accumulation of metals in the sediments and reed biomass of a combined constructed wetland treating domestic wastewater Water, Air and Soil Pollution 2007 183 253 264 10.1007/s11270-007-9374-4 Masi, F., Conte, G., Martinuzzi, N., & Pucci, B. (2002). Winery high organic content wastewaters treated by constructed wetlands in Mediterranean climate. In: Proceedings of the 8th International Conference on Wetland Systems for Water Pollution Control. University of Dar-es-Salaam, Tanzania and IWA. 274–282. Molle P Prost BS Lienard A Potential for total nitrogen removal by combining vertical flow and horizontal flow constructed wetlands: A full scale experimental study Eco. Eng. 2008 34 23 29 10.1016/j.ecoleng.2008.05.016 Spangler JT Sample DJ Fox LJ White SA Data on floating treatment wetland aided nutrient removal from agricultural runoff using two wetland species Data in Brief 2019 22 756 761 10.1016/j.dib.2018.12.037 30671519 Tanner CC Sukias JPS Headley TR Yates CR Stott R Constructed wetlands and denitrifying bioreactors for on-site and decentralized wastewater treatment: Comparison of five alternative configurations Ecological Engineering 2012 42 112 123 10.1016/j.ecoleng.2012.01.022 Tanveer Saeed Md Miah J Khan T Intensified constructed wetlands for the treatment of municipal wastewater: Experimental investigation and kinetic modelling Environ Sci. Pollution Research 2021 28 30908 30928 10.1007/s11356-021-12700-8 Thalla AK Devatha CP Anagh K Sony E Performance evaluation of horizontal and vertical flow constructed wetlands as tertiary treatment option for secondary effluents Applied Water Science 2019 9 147 10.1007/s13201-019-1014-9 Vymazal J Horizontal sub-surface flow and hybrid constructed wetlands systems for wastewater treatment Ecological Engineering 2005 25 478 490 10.1016/j.ecoleng.2005.07.010 Vymazal J Masa M Horizontal sub-surface flow constructed wetland with pulsing waterlevel Water Science and Technology 2003 48 5 143 148 10.2166/wst.2003.0304 Wang R Baldy V Perissol C Korboulewsky N Influence of plants on microbial activity in a vertical-down flow wetland system treating waste activated sludge with high organic matter concentrations Journal of Environmental Management 2012 95 158 164 10.1016/j.jenvman.2011.03.021 Yaseen DA Scholz M Textile dye wastewater characteristics and constituents of synthetic effluents: A critical review 2019 Berlin Heidelberg Springer Zhang D Gersberg RM Keat TS Review constructed wetlands in China Ecological Engineering 2009 35 1367 1378 10.1016/j.ecoleng.2009.07.007 Zhang J Haiming Wu Zhen Hu Liang S Fan J Examination of oxygen release from plants in constructed wetlands in different stages of wetland plant life cycle Environmental Science and Pollution Research 2014 21 9709 9716 10.1007/s11356-014-2905-9 24777322 Zurita F De Anda J Belmont MA Treatment of domestic wastewater and production of commercial flowers in vertical and horizontal subsurface-flow constructed wetlands Ecological Engineering 2009 35 861 869 10.1016/j.ecoleng.2008.12.026	36504546	PMC9717559	NO-CC CODE	2022-12-06 23:23:25	no	Water Air Soil Pollut. 2022 Dec 1; 233(12):510	utf-8	Water Air Soil Pollut	2,022	10.1007/s11270-022-05984-0	oa_other
==== Front Clin Exp Med Clin Exp Med Clinical and Experimental Medicine 1591-8890 1591-9528 Springer International Publishing Cham 36459278 959 10.1007/s10238-022-00959-1 Research The impact of the secondary infections in ICU patients affected by COVID-19 during three different phases of the SARS-CoV-2 pandemic Murgia Federica 1 Fiamma Maura 2 Serra Silvia 2 Marras Giulia 2 Argiolas Raul 2 Mattana Chiara 2 Mattu Maria Grazia 2 Garau Maria Cristina 2 Doneddu Sonia 2 Olla Sabrina 2 Cocco Eleonora 34 Lorefice Lorena 4 Muntoni Sandro 1 Paffi Peppino 5 Porru Stefano 5 Abis Marta 5 Bellizzi Saverio 6 Pani Antonello 7 Angioi Andrea 7 Simbula Gabriella 8 Mussap Michele 9 Massidda Orietta 10 Carta Franco 2 Atzori Luigi [email protected] 11 1 grid.7763.5 0000 0004 1755 3242 Department of Biomedical Sciences, Clinical Metabolomics Unit, University of Cagliari, Cagliari, Italy 2 Clinical-Chemical Analysis and Microbiology Laboratory, “San Francesco” Hospital, Nuoro, Italy 3 grid.7763.5 0000 0004 1755 3242 Department of Medical Science and Public Health, University of Cagliari, Cagliari, Italy 4 Multiple Sclerosis Center, Binaghi Hospital, ATS Sardegna, ASSL Cagliari, Cagliari, Italy 5 Intensive Care Unit, “San Francesco” Hospital, Nuoro, Italy 6 Amman, Jordan 7 grid.7763.5 0000 0004 1755 3242 Nephrology, Dialysis and Transplantation, University of Cagliari, Cagliari, Italy 8 grid.7763.5 0000 0004 1755 3242 Department of Biomedical Sciences, Oncology and Molecular Pathology Unit, University of Cagliari, Cagliari, Italy 9 grid.7763.5 0000 0004 1755 3242 Department of Surgical Sciences, University of Cagliari, Cagliari, Italy 10 grid.11696.39 0000 0004 1937 0351 Department of Cellular, Computational and Integrative Biology, Center of Medical Sciences (CISMed), University of Trento, Trento, Italy 11 grid.7763.5 0000 0004 1755 3242 Department of Biomedical Sciences, Clinical Metabolomics Unit, University of Cagliari, Strada Interna Policlinico Universitario, 09042 Monserrato, CA Italy 2 12 2022 113 30 9 2022 18 11 2022 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Microbial secondary infections can contribute to an increase in the risk of mortality in COVID-19 patients, particularly in case of severe diseases. In this study, we collected and evaluated the clinical, laboratory and microbiological data of COVID-19 critical ill patients requiring intensive care (ICU) to evaluate the significance and the prognostic value of these parameters. One hundred seventy-eight ICU patients with severe COVID-19, hospitalized at the S. Francesco Hospital of Nuoro (Italy) in the period from March 2020 to May 2021, were enrolled in this study. Clinical data and microbiological results were collected. Blood chemistry parameters, relative to three different time points, were analyzed through multivariate and univariate statistical approaches. Seventy-four percent of the ICU COVID-19 patients had a negative outcome, while 26% had a favorable prognosis. A correlation between the laboratory parameters and days of hospitalization of the patients was observed with significant differences between the two groups. Moreover, Staphylococcus aureus, Enterococcus faecalis, Candida spp, Pseudomonas aeruginosa and Klebsiella pneumoniae were the most frequently isolated microorganisms from all clinical specimens. Secondary infections play an important role in the clinical outcome. The analysis of the blood chemistry tests was found useful in monitoring the progression of COVID-19. Supplementary Information The online version contains supplementary material available at 10.1007/s10238-022-00959-1. Keywords SARS-COV-2 infection COVID-19 Blood chemistry parameters Microbiological data Secondary infections Clinical outcome ==== Body pmcIntroduction SARS-CoV-2 infection is the cause of COVID-19, a respiratory disease that can range from asymptomatic to symptomatic disease with severe pneumonia, inflammation and death [1, 2]. The clinical outcome and prognosis depend on various factors [3], including bacterial/viral/fungal co-infections or superinfections (secondary infections) which can increase the risk of mortality in these patients [4–7] The importance of secondary infections during the COVID-19 pandemic has been somewhat investigated [8, 9] as they were reported to be associated with a decreased survival, playing a crucial role especially in those patients with a severe illness admitted to the ICU [10]. Several reports indicated an increased number of COVID-19 patients with secondary infections during hospitalization [11, 12]. SARS-CoV-2, like other respiratory viruses, weakens the host immunity, facilitating such infections [13], especially during hospitalization. While the specific source and nature of these infections have not been thoroughly investigated [5], some evidence suggests that nosocomial, often multidrug-resistant (MDR), pathogens, are likely responsible [10, 14]. The dramatic aspect of this scenario is that during this pandemic up to 50% of patients who died from COVID-19 had secondary infections [9, 15], likely related to the fact that critically ill hospitalized patients needed invasive mechanical ventilation for a prolong period and this rendered them susceptible to hospital-acquired infections [16]. Inflammatory chemical biomarkers and proper microbiological surveillance may be fundamental for the management of COVID-19 patients because they allow the early detection of markers which correlate with the presence of such infections and a negative outcome. Laboratory data are thus fundamental to support clinicians in monitoring the hospitalized patients and follow the clinical evolution of COVID-19 [17]. In this study, we collected the clinical, laboratory and microbiological data of hospitalized COVID-19 patients, admitted to the ICU from March 2020 to May 2021, to evaluate the significance and the prognostic value of these parameters [18]. Methods COVID-19 patients, positive for SARS-CoV-2 infection (diagnosed through the extraction of RNA and reverse transcription Real-Time PCR) hospitalized in the ICU of the S. Francesco Hospital of Nuoro (Sardinia, Italy) in the period from March 2020 to May 2021 were enrolled in this study (n = 178). The three different phases of the pandemics were classified as follows: (i) March–May 2020; (ii) September–December 2020; and (ii) January–May 2021. The study was conducted in accordance with the current revision of the Helsinki Declaration. By analyzing the medical records of 156 patients (22 records were not available), it was possible to collect information regarding the symptoms at the onset of the infection, the presence of risk factors and co-morbidities and the microbiological data. Moreover, blood chemistry parameters relative to three different time points were collected: (T0), sampling at the moment of the admission of the patient at the emergency room, the infectious diseases ward, or directly at the ICU; (T1), sampling between the admission day and the clinical outcome; and (T2), the last sampling before the transfer of the patient from the ICU to others hospital wards or the patient' demise. The blood chemistry parameters considered were: leucocytes, neutrophils, lymphocytes, platelets, hemoglobin, lactate dehydrogenase (LDH), C reactive protein (CRP), procalcitonin, D-dimer, fibrinogen, ferritin, creatinine, aspartate aminotransferase (AST), alanine aminotransferase (ALT), sodium and potassium. Details of the selected biomarkers and laboratory instruments used are reported in Table 1.Table 1 Panel of the laboratory parameters evaluated in the study Blood chemistry parameters Panels Parameters Instrument Blood count Hemoglobin ADVIA 2120 SIEMENS Leukocytes Platelet Coagulation D-dimer CS 1500 SIEMENS Fibrinogen Kidney/liver function electrolytes AST DIMENSION VISTA 1500 ALT Creatinine Sodium Potassium Inflammation and sepsis indices Ferritin LDH CRP PCT ADVIA CENTAUR XP SIEMENS Microbiological data related to blood, bronchoaspirate, urine and intravascular device specimens (central venous catheter, CVC) were collected every day through diagnostic microbiological methods. Each isolate was carefully reviewed by the members of the microbiology team to evaluate its clinical significance. For our analysis, we considered the positivity of the microbiological samples after 7 days of ICU hospitalization to exclude possible bias due to the presence of previous infections. To determine the antimicrobial susceptibility profile of the clinical isolates, Matrix-assisted laser desorption ionization-time of flight (MALDI-TOF), biotyper (Bruker Daltonics), MicroScan (Beckman Coulter. Inc., Brea, USA), MIC/Combo and NMDR Panel were used. The breakpoint panels use concentrations equivalent to the breakpoint of the Clinical and Laboratory Standards Institute (CLSI) following the European Committee on Antimicrobial Susceptibility Testing (EUCAST) interpretive criteria. Isolated considered not of clinical relevance, and thus not warranting a specific therapy, were classified as commensal and/or non-significant. Laboratory data were organized in matrices for statistical analysis. Two different statistical approaches were applied: first, the blood chemistry parameters collected at different time points were processed through multivariate analysis, using the SIMCA-P software (ver. 16.0, Sartorius Stedim Biotech, Umea, Sweden) [19]. Variables were UV scaled, and then the Principal Component Analysis (PCA) was applied to explore the sample distributions without classification and identify potential strong outliers through the application of the Hotelling’s T2 test. Subsequently, to study a possible linear relationship between a matrix Y (dependent variable, time) and a matrix X (predictor variables, e.g., blood chemistry parameters), the Partial Least Square (PLS) model was carried out [20]. The variance and the predictive ability (R2X, R2Y, Q2) were evaluated to establish the model’s suitability. In addition, a permutation test (n = 400) was performed to validate the models [21]. The most significant variables were extracted from the PLS model’s loading plot. Variables’ Influence on Projection (VIP) value was also evaluated for the selection of the discriminant variable. This analysis allowed to observe, simultaneously, changing of the variables during the hospitalization of the patients based on the progression of the COVID-19 infection. Univariate statistical analysis was then performed. In particular, the concentrations of the discriminant parameters were tested through the non-parametric Wilcoxon test. This test is appropriate to investigate differences in paired samples (GraphPad Prism software, version 7.01, Inc., San Diego, CA, USA). Only differences between variables with p value < 0.05 were considered significant. For these variables, we also evaluated the trend during the three different phases of the pandemic waves specified above. Results A total of 178 patients were enrolled in this study. About 74% (131) of these patients died during ICU hospitalization (Deceased patients) while about 26% (47) of them became SARS-CoV-2 negative during the ICU hospitalization and were transferred to other hospital wards (Transferred Patients). Demographic data of the enrolled patients are reported in Table 2.Table 2 Demographic data of the enrolled patients. Data are presented considering firstly the whole cohort of patients and then considering the different phases of the Covid-19 pandemic Database Covid-19 Patients Age mean ± SD (Range) F/M (%) Deceased (%) Transferred (%) Hospitalization days (Mean ± SD) All 178 68.1 ± 11.3 (22 − 88) 31/69 74 26 24 ± 15 Deceased 132 69.7 ± 9.5* (37 − 86) 31/69 17 ± 11 Transferred 46 63.2 ± 14.3 (22–88) 30/70 31 ± 20 1 Phase (March–May 2020) 31 69 Deceased 4 81 ± 1.7* (80 − 83) 67/33 11 ± 2 Transferred 9 57 ± 15.7 (22 − 78) 22/78 34 ± 24 2 Phase (September–December 2020) 82 18 Deceased 80 69 ± 9.4* (80 − 83) 25/75 17 ± 11 Transferred 18 63 ± 14.8 (33 − 88) 38/62 26 ± 18 3 Phase (January–May 2021) 72 28 Deceased 48 70 ± 9.6* (43 − 84) 39/61 18 ± 10 Transferred 19 66 ± 11.5 (43 − 82) 36/64 33 ± 21 * = p < 0.05 For the deceased patients, the mean time from the onset of symptoms to the hospitalization was 5.6 days, and 26% of these patients were admitted directly to the ICU. For the transferred patients, the mean time from the onset of the symptoms to the hospitalization was equal to 6.1 days and 22% of these patients were admitted directly to the ICU. A summary of the onset symptoms, co-morbidities and lifestyle information of the patients is shown in Table 3.Table 3 Summary of the onset symptoms and co-morbidities of the patients with COVID-19 classified based on the clinical outcome Onset of symptoms Deceased (%) Transferred (%) Dyspnea 73 60 Fever 72 80 Cough 45 53 Asthenia 31 33 Diarrhea 11 16 Myalgia 9.5 16 Hyposmia/Hypogeusia 5.5 3 Headache 5 7 Pharyngodynia 5 7 Sputum 4 0 Nausea/Vomiting 5 7 Co-morbidities Arterial hypertension* 57 33 Ischemic heart disease 9.5 3 Heart failure 4 – Atrial fibrillation 13 10 Aneurysm of the aorta 3 7 Congenital heart disease – 3.4 Valvulopathies 2.4 – Altered coagulation 2.4 3.4 Brain stroke 3 – Dementia 7 – Parkinson disease 1.5 – Depression 7 7 Dyslipidemia 21 17 Diabetes 22 23 Obesity 41 50 Bronchial asthma 7.5 7 Chronic obstructive pulmonary disease 14 10 Chronic renal failure 5 Liver diseases 2 – Colon diverticulosis 3 – Inflammatory bowel diseases – 3 Gastroesophageal reflux disease 11 – Rheumatoid arthritis 4.5 14 Systemic lupus erythematosus – 3 Hypothyroidism 11 14 Arthrosis 3 – Osteoporosis 5 3 Benign prostatic hyperplasia 11 77 Cancer 11 3 Lifestyle BMI 29.6 31 Smokers 15.9 6.9 Habitual alcohol users 2.3 3.4 * = p < 0.05 Analysis of the laboratory and microbiological results were based on the different outcomes of the infection to highlight the predictive role of the data concerning the progression and prognosis of the disease. Analysis of the blood chemistry parameters The data matrix of the deceased patients was firstly analyzed with the PCA model to identify outliers that could affect the validity of the analysis. No outliers were identified with this analysis, so all the samples were considered for the subsequent step. Samples were analyzed through a PLS approach, using as Y-variable the three different time points (T0, T1 and T2). The correlation model showed a time-dependent distribution of the samples based on the days of hospitalization (Fig. 1A, statistical parameters were R2X = 0.297, R2Y = 0.464, Q2 = 0.420, p < 0.0001). The model was then validated by using a permutation test (Fig. 1S-A,). This statistical approach allows correlating the number of days of hospitalization with numerical variables, such as the laboratory parameters, with the aim to quickly identify which of them changes its concentration in line with the clinical evolution of the patients.Fig. 1 A PLS model built considering the blood chemistry parameters of the deceased patients. This statistical approach allows to correlate the number of the days of hospitalization with numerical variables, such as the laboratory parameters, with the aim to quickly identify which of them change its concentration in line with the clinical evolution of the patients. White circles represent samples at the moment of the hospital admission (T0); gray circles represent samples belonging to the same patients but collected at an intermediate time of hospitalization; black circles represent the samples collected from the same patients before death. B The trend of the discriminant parameters resulted from the multivariate statistical analysis which changed their concentration significantly during the hospitalization of the deceased patients. Wilcoxon test was employed. = p value < 0.01, * = p value < 0.001, ** = p value < 0.0001 Variables (laboratory parameters) that showed a VIP value > 1 were considered discriminant and underwent the Wilcoxon test for paired data. The results are shown in Fig. 1B. Significant changes of some parameters during hospitalization were observed, such as an increase in leukocytes, neutrophils, PCT, D-dimer, sodium and potassium and a decrease in hemoglobin and lymphocyte levels. Moreover, the trend of the laboratory parameters was also evaluated during the three different phases of the SARS-CoV-2 pandemic (Fig. 2S).Fig. 2 A PLS model built considering the blood chemistry parameters of the transferred patients. This statistical approach allows to correlate the number of the days of hospitalization with numerical variables, such as the laboratory parameters, with the aim to quickly identify which of them change its concentration in line with the clinical evolution of the patients. White circles represent samples at the moment of admission to the hospital (T0); gray circles represent samples belonging to the same patients but collected at an intermediate time of hospitalization; blue circles represent the samples collected from the same patients before the transfer from the ICU to other hospital wards. B The trend of the discriminant parameters resulted from the multivariate statistical analysis which significantly changed their concentration during the hospitalization of the transferred patients. Wilcoxon test was employed. = p value < 0.01, * = p value < 0.001, ** = p value < 0.0001 Considering the matrix of the transferred patients, the analysis was conducted following the same workflow applied to the deceased patients. Also in this case, no outliers were identified with this analysis, so all the samples were considered for the subsequent PLS approach, using as Y-variable the three different time points (T0, T1 and T2). The model showed a time-dependent distribution of the samples based on the days of hospitalization (Fig. 2A, statistical parameters were R2X = 0.352, R2Y = 0.49, Q2 = 0.36, p < 0.0001). Also in this case, the distribution of the samples reflected the change in the laboratory parameters used as variables, during the hospitalization. The model was then validated by using a permutation test (Fig. 1S-B). Variables that showed a VIP value > 1 were considered discriminant and underwent the Wilcoxon test for paired data. The results are shown in Fig. 2B. Changes in some parameters throughout hospitalization were observed with a significant increase in lymphocytes, sodium, platelets, and a significant decrease in neutrophils, hemoglobin, CRP, LDH, fibrinogen, creatinine and AST. The trend of the laboratory parameters was also evaluated during the three different phases of the SARS-CoV-2 pandemic (Fig. 3S).Fig. 3 PLS-DA model of the T0 samples of the patients which had different outcomes (gray circles are transferred patients while light blue circles represent deceased patients) To investigate a possible predictive role of the laboratory parameters in terms of prognosis/outcome of the SARS-CoV-2 infection, a supervised model (PLS-DA) of the T0 samples of the transferred and deceased patients was performed. No separation was observed between the two classes of samples and the model was not significant (Fig. 3). This result was also evidenced when we compared each laboratory parameter of the two classes at T0 time. No significant comparison emerged. However, we also performed the comparisons between deceased and transferred patients considering the laboratory parameters at the three different time points. Interestingly, leukocytes, neutrophils, lymphocytes, LDH, CRP, D-dimer, creatinine, sodium and potassium showed significant (p < 0.05) differences in their concentrations starting from the T1 sampling as demonstrated in Fig. 4.Fig. 4 Comparisons of the concentrations of the laboratory parameters considering the three different time points of the deceased and transferred patients (red and blue bars, respectively). Mann–Whitney U test was used (* = p < 0.05, = p value < 0.01, * = p value < 0.001, **** = p value < 0.0001) Microbiological analysis Of the enrolled patients, after 7 days of hospitalization, 44% showed the presence of a secondary infection during the hospitalization (and about 80% of these patients had a negative outcome (Fig. 5A). About 38% of the patients showed a microbiological negative result at 7 days, and for about 18% it was not possible to retrieve the information. The microbiological analysis evaluated several specimens: blood, bronchoaspirate, urine and CVC. The summary of the positive samples is reported in Fig. 5B. The analysis of the isolated microbiological species in the different specimens was performed based on the different outcomes of the SARS-CoV-2 infection (Fig. 6). The panel of the antibiotics used is summarized in Fig. 4S. All the patients were exposed to at least 1 dose of an antibiotic.Fig. 5 A Percentage of COVID-19 patients with secondary infections after 7 days of hospitalization and mortality relative to the positive patients. B Summary of the positive microbiological specimens in the deceased groups and transferred patients (black and white bars, respectively) Fig. 6 A Microbiological results of the different specimens relative to the deceased patients. Blood cultures were positive mainly for CoNS (16%) and S. aureus (12%). Bronchoaspirates were positive mainly for Candida spp. (24%), P. aeruginosa (13%), Klebsiella pneumoniae (8%). Urine cultures were positive for E. faecalis (31%), Candida spp. (18%) and Klebsiella pneumoniae (18%); Finally, CVC cultures, were positive for CoNS (37%), while Candida spp. were found in 27% of the cases. B Microbiological results of the different specimens relative to the transferred patients. Blood cultures were positive for CoNS (28%), E. faecalis, E. aerogenes, and K. pneumoniae (12%). Bronchoaspirates were mainly positive for Candida spp. (23%), P. aeruginosa (18%), CoNS and Escherichia Coli (12%); Urine cultures were positive for Candida spp (26%). Finally, in CVC cultures, CoNS were the prevailing bacteria (32%), while P. aeruginosa and K. pneumoniae were found in 14% Discussion By observing the cohort of patients, it emerged that males were more affected by COVID-19 compared to females, regardless of the clinical outcome and the pandemic phase. These data confirm previous studies, in which a higher percentage of males with COVID-19 were reported [22]. Different causes, for example the expression of ACE2 receptor, have been considered as an indicator of gender susceptibility [23]. Based on the demographic data, the mean age of the non-surviving patients was significantly higher than the surviving ones, as expected. If only deceased patients are considered, it is interesting to note that the age of the patients in the first phase was higher, probably because SARS-CoV-2 infection hit the oldest and more susceptible individuals in the population. Other risk factors were also reported, and the percentage of smokers was twice as high in deceased patients compared to the transferred patients, suggesting cigarette smoke as a negative predicting factor of the COVID-19 prognosis [24]. No differences resulted from BMI and alcohol consumption analysis. Several co-morbidities, including hypertension, heart failure, atrial fibrillation, asthma, chronic obstructive pulmonary disease, and chronic renal failure were more frequent in deceased patients [25]. The deceased patients appeared more susceptible to contracting secondary infections and had a reduced ability to compensate for the alterations induced by them [26]. Regarding the onset of symptoms, our data show that only dyspnea, hyposmia/hypogeusia and diarrhea have a higher incidence among deceased patients. Blood chemistry parameters Covid-19, and possibly associated secondary infections, caused various laboratory alterations, with a different trend between the two groups analyzed. In deceased patients, there was a statistically significant increase in the indices of inflammation, creatinine, sodium and potassium, and a significant reduction in hemoglobin and lymphocytes, probably due to the organism's failure to counteract infections and consequent organ damage [27]. The same laboratory parameters in surviving patients showed a significant increase in lymphocytes, platelets, sodium, and potassium plus a reduction in the indices of inflammation, hemoglobin and markers of liver damage [28]. These changes are consistent with a probable resolution of the infections and, therefore, improve the clinical outcome. In this study, the changes in our patients' laboratory test result over time and the differences between the three pandemic waves were also analyzed. Regarding the group of deceased patients, all the considered parameters were stable in the first and third waves. In contrast, in the second wave, leukocytes, procalcitonin, C reactive protein, D-dimer, creatinine, LDH, AST and ALT increased considerably, particularly from T1 to death, while lymphocytes were reduced. These differences over time are probably linked to the fact that in September–December 2020, the patients admitted to the Nuoro ICU were more affected by other bacterial/fungal infections which may have contributed to the poor prognosis of these patients. Regarding the transferred patients, the trend over time of the laboratory results in the three waves was rather stable in the three periods analyzed with minimal variations. In particular, a reduction in the inflammation indices was observed in the first wave, probably because, in our region, this was the phase of the pandemic with fewer COVID-19 cases and fewer deaths, likely related to the geographical position and the severe restrictive measures adopted by the government (national lockdown). Based on the comparison of the laboratory parameters between the deceased and transferred patients at the admission time (T0), we didn’t find any significant result, and this evidenced the lack of a predictive role of them in terms of prognosis and outcome of the COVID-19 disease. However, several parameters changed significantly their concentrations between the two classes of patients starting from the T1 sampling (leukocytes, neutrophils, lymphocytes, LDH, CRP, D-dimer, creatinine, sodium and potassium), suggesting the presence of secondary infections during hospitalization, which may have started to act as reinforcing factors to promote worsening of the disease in vulnerable patients. Microbiological data Another crucial aspect of this study was to focus on secondary infections, which comprise the group of all patients with a positive culture (urine, bronchoaspirate, blood culture, and CVC) during hospitalization in the ICU. From our analysis, it emerges that among the 156 patients, approximately 44% presented positive cultures, and of these, 80% died. By analyzing the data from the literature, the percentage of COVID-19 patients with secondary infections is highly variable, ranging between 14 and 100%, depending on the different inclusion criteria used [8, 15, 29]. Differences in the population, specimen source, and pathogens of interest are likely responsible for the wide variations reported [30]. We observed a high percentage of positive samples, and it is probably because, in our study, we considered all the positive cultures occurring after 7 days of hospitalization, including blood and urine cultures, bronchoaspirate and CVC cultures. Another critical aspect to consider is the logistical conditions in which the healthcare personnel had to work, especially during the second wave, as there was a rapid and sudden increase in COVID-19 cases and ICU admissions. The higher incidence of catheter-related bloodstream infection could be explained by the heavy pressure put on the ICU by the COVID-19 outbreak; in fact, during the peak, the ICU capacity of the Nuoro hospital had to be increased by 300% to accommodate all patients requiring critical care. Interestingly, from the microbiological analysis, it emerged that deceased patients showed a higher percentage of positive samples than the transferred patients considering blood cultures and bronchoaspirate. On the other hand, a higher percentage of positive samples in urine cultures and CVC cultures was observed in the group of transferred patients suggesting that it is correlated with the longer period of hospitalization. Indeed, the average ICU stay for the transferred patients was longer than for deceased patients (31 vs 17 days, respectively), explaining the greater occurrence of nosocomial infections in the urine and CVC cultures of the first group of patients. Our attention focused on the microorganisms likely responsible for the secondary infections in the two groups of patients. Staphylococcus aureus, Klebsiella pneumonia, Candida spp, Enterococcus faecalis, Pseudomonas aeruginosa and above all CoNS were mainly detected in both groups. These findings are in line with other published studies [15]. During the hospitalization of our patients, Klebsiella pneumoniae isolates developed acquired resistance mechanisms, such as ESBL (Extended-Spectrum β-Lactamase) and carbapenem-resistance, determining a difficult challenge for the choice of the antibiotic treatment, especially during the second wave. Our results are overall consistent with what is already reported in the literature. Microorganisms mainly involved in the COVID-19 secondary infections were Mycoplasma spp., Haemophilus influenzae, P. aeruginosa and S. aureus, with a discrete variability in the species most represented and widespread in the various territories and hospitals [12, 31]. Based on our results, it seems clear that secondary infections may have played a critical role in the negative outcome of the patients affected by COVID-19. Secondary infections may have acted as mutually reinforcing factors to promote the progression to severe and fatal disease. Indeed, severe COVID-19 caused multiple damages [32], which may favor rapid bacterial growth; on the other hand, bacterial virulence factors can alter the immune responses, resulting in a rebound of SARS-CoV-2 infection and disease progression, leading to higher mortality in severe and critical patients [33, 34]. Finally, the analysis of our data revealed that the most used antibiotics were Azithromycin (60%), Piperacillin-Tazobactam (40%), and Vancomycin (20%). Such a high percentage of Azithromycin could be attributed to the fact that, in many cases, it was already administered to the patients before hospitalization or in the infectious disease divisions to patients with symptomatic COVID-19. However, despite Azithromycin has been one of the few drugs used in COVID-19 patients during the start of the pandemic, there is currently no evidence to justify the widespread use of this antibiotic [35]. As for Piperacillin-Tazobactam, it has been used in many patients as a broad-spectrum empirical therapy, pending the microbiological and antibiogram results. Finally, Vancomycin was also used in a high percentage of patients to treat infections caused by Gram-positive bacteria resistant to other classes of antibiotics. Conclusion The clinical presentation, the presence of co-morbidities, and the laboratory, as well as microbiology data, can be valuable tools for identifying the most severe COVID-19 patients facing a poor prognosis. ICU represented a critical area for the management of patients with severe COVID-19. Between the first and the other two pandemic phase, considered in this study, there has not been an implementation of the resources necessary to properly assist these patients, such as the number of devices and health workers. This may have been an indirect cause for an increased spread of infectious agents within the hospital, particularly in the ICU. Moreover, the higher frequency of isolation of pathogens may have been caused by the fact that the most severe and critically ill COVID-19 cases in ICU underwent invasive procedures, which per se increase the chance of hospital-acquired infections. For these reasons, it is essential to monitor the patients with serial microbiological examinations to undertake optimal targeted treatments to avoid the development of antimicrobial resistance and the spread of MDR microorganisms, which represents a complex global problem that cannot be ignored even in the case of superimposing emergencies as that of the Sars-Cov2 pandemic. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 1041 KB) Acknowledgements We are grateful to the healthcare operators and the patients included in this study. Author contributions Conceptualization FM, MF, FC, LA, Data curation FM, OM, MM, FC, AP, AA, Formal analysis FM, SP, MA, SO, RA, SS, GM, CM, MGM, SD, Investigation FM, MA, SP, SO, Supervision MCG, LA, Visualization EC, LL, SB, SM, Writing-original draft FM, MF, FC, OM, Writing-review and editing LA, MM, OM, EC, FM. All authors read and approved the final manuscript. Funding Not applicable. Declarations Conflict of interests The authors declare no competing interests. Ethical approval This study was approved by the institutional review board of the San Francesco Hospital (428/2022/CE). The study was performed in accordance with the ethical standards of the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective nature of the study. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Federica Murgia Maura Fiamma Franco Carta and Luigi Atzori have contributed equally. ==== Refs References 1. Wang C Horby PW Hayden FG A novel coronavirus outbreak of global health concern Lancet 2020 395 470 473 10.1016/S0140-6736(20)30185-9 31986257 2. Pollard CA Morran MP Nestor-Kalinoski AL The COVID-19 pandemic: a global health crisis Physiol Genomics 2020 52 549 557 10.1152/physiolgenomics.00089.2020 32991251 3. Parohan M Yaghoubi S Seraji A Risk factors for mortality in patients with Coronavirus disease 2019 (COVID-19) infection: a systematic review and meta-analysis of observational studies Aging Male 2020 23 1416 1424 10.1080/13685538.2020.1774748 32508193 4. McArdle AJ Turkova A Cunnington AJ When do co-infections matter? Curr Opin Infect Dis 2018 31 209 215 10.1097/QCO.0000000000000447 29698255 5. Mirzaei R Goodarzi P Asadi M Bacterial co-infections with SARS-CoV-2 IUBMB Life 2020 72 2097 2111 10.1002/iub.2356 32770825 6. Jia L Xie J Zhao J Mechanisms of severe mortality-associated bacterial co-infections following influenza virus infection Front Cell Infect Microbiol 2017 7 338 10.3389/fcimb.2017.00338 28824877 7. Quah J Jiang B Tan PC Impact of microbial Aetiology on mortality in severe community-acquired pneumonia BMC Infect Dis 2018 18 451 10.1186/s12879-018-3366-4 30180811 8. Huang C Wang Y Li X Clinical features of patients infected with 2019 novel coronavirus in Wuhan China Lancet 2020 395 497 506 10.1016/S0140-6736(20)30183-5 31986264 9. Zhou F Yu T Du R Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study Lancet 2020 395 1054 1062 10.1016/S0140-6736(20)30566-3 32171076 10. Zhang G Hu C Luo L Clinical features and short-term outcomes of 221 patients with COVID-19 in Wuhan China J Clin Virol 2020 127 104364 10.1016/j.jcv.2020.104364 32311650 11. Hendaus MA Jomha FA Covid-19 induced superimposed bacterial infection J Biomol Struct Dyn 2021 39 4185 4191 10.1080/07391102.2020.1772110 32448041 12. Rawson TM Wilson RC Holmes A Understanding the role of bacterial and fungal infection in COVID-19 Clin Microbiol Infect 2021 27 9 11 10.1016/j.cmi.2020.09.025 32979569 13. Ritchie AI Singanayagam A Immunosuppression for hyper inflammation in COVID-19: a double-edged sword? Lancet 2020 395 1111 10.1016/S0140-6736(20)30691-7 32220278 14. Blasco ML Buesa J Colomina J Co-detection of respiratory pathogens in patients hospitalized with Coronavirus viral disease-2019 pneumonia J Med Virol 2020 92 1799 1801 10.1002/jmv.25922 32320082 15. Chen N Zhou M Dong X Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study Lancet 2020 395 507 513 10.1016/S0140-6736(20)30211-7 32007143 16. Cox MJ Loman N Bogaert D Co-infections: potentially lethal and unexplored in COVID-19 Lancet Microbe 2020 1 1 e11 10.1016/S2666-5247(20)30009-4 32835323 17. Costa VO Nicolini EM da Costa BMA Evaluation of the Risk of Clinical Deterioration among Inpatients with COVID-19 Adv Virol 2021 2021 6689669 10.1155/2021/6689669 34257657 18. Murgia F Fiamma M Serra S The impact of secondary infections in COVID-19 critically ill patients J Infect 2022 84 6 e116 e117 10.1016/j.jinf.2022.03.017 35314266 19. Eriksson L, Byrne T, Johansson E, et al. Multi- and megavariate data analysis basic principles and applications. Umetrics Academy; 2013. 509. 20. Wold S Sjöström M Eriksson L PLS-regression: a basic tool of chemometrics Chemom Intell Lab Syst 2001 58 109 130 10.1016/S0169-7439(01)00155-1 21. Lindgren F Hansen B Karcher W Model validation by permutation tests: applications to variable selection J Chemom 1996 10 521 532 10.1002/(SICI)1099-128X(199609)10:5/6<521::AID-CEM448>3.0.CO;2-J 22. Wang P Lu J-A Jin Y Statistical and network analysis of 1212 COVID-19 patients in Henan China Int J Infect Dis 2020 95 391 398 10.1016/j.ijid.2020.04.051 32339715 23. Zhao Y Zhao Z Wang Y Single-Cell RNA expression profiling of ACE2, the receptor of SARS-CoV-2 Am J Respir Crit Care Med 2020 202 756 759 10.1164/rccm.202001-0179LE 32663409 24. Kashyap VK Dhasmana A Massey A Smoking and COVID-19: adding fuel to the flame Int J Mol Sci 2020 21 E6581 10.3390/ijms21186581 25. Ejaz H Alsrhani A Zafar A COVID-19 and comorbidities: deleterious impact on infected patients J Infect Public Health 2020 13 1833 1839 10.1016/j.jiph.2020.07.014 32788073 26. Parohan M Yaghoubi S Seraji A Risk factors for mortality in patients with Coronavirus disease 2019 (COVID-19) infection: a systematic review and meta-analysis of observational studies Aging Male 2020 23 1416 1424 10.1080/13685538.2020.1774748 32508193 27. Huang W Berube J McNamara M Lymphocyte subset counts in COVID-19 patients: a meta-analysis Cytometry A 2020 97 772 776 10.1002/cyto.a.24172 32542842 28. Kermali M Khalsa RK Pillai K The role of biomarkers in diagnosis of COVID-19 - a systematic review Life Sci 2020 254 117788 10.1016/j.lfs.2020.117788 32475810 29. Garcia-Vidal C Sanjuan G Moreno-García E Incidence of co-infections and superinfections in hospitalized patients with COVID-19: a retrospective cohort study Clin Microbiol Infect 2021 27 83 88 10.1016/j.cmi.2020.07.041 32745596 30. Kubin CJ McConville TH Dietz D Characterization of bacterial and fungal infections in hospitalized patients with coronavirus disease 2019 and factors associated with health care-associated infections Open Forum Infect Dis. 2021 8 6 ofab201 10.1093/ofid/ofab201 34099978 31. Huang ACC Huang CG Yang CT Concomitant infection with COVID-19 and Mycoplasma pneumoniae Biomed J. 2020 43 5 458 61 10.1016/j.bj.2020.07.002 32891590 32. Ghoneim HE Thomas PG McCullers JA Depletion of alveolar macrophages during influenza infection facilitates bacterial superinfections J Immunol 2013 191 3 1250 1259 10.4049/jimmunol.1300014 23804714 33. Zhang H Zhang Y Wu J Risks and features of secondary infections in severe and critical ill COVID-19 patients Emerg Microbes Infect 2020 9 1 1958 1964 10.1080/22221751.2020.1812437 32815458 34. Zangrillo A Beretta L Scandroglio AM Characteristics, treatment, outcomes and cause of death of invasively ventilated patients with COVID-19 ARDS in Milan Italy Crit Care Resusc 2020 22 3 200 211 32900326 35. Echeverría-Esnal D Martin-Ontiyuelo C Navarrete-Rouco ME Azithromycin in the treatment of COVID-19: a review Expert Rev Anti Infect Ther 2021 19 2 147 163 10.1080/14787210.2020.1813024 32853038	36459278	PMC9717567	NO-CC CODE	2022-12-06 23:23:25	no	Clin Exp Med. 2022 Dec 2;:1-13	utf-8	Clin Exp Med	2,022	10.1007/s10238-022-00959-1	oa_other
==== Front Nat Rev Psychol Nat Rev Psychol Nature Reviews Psychology 2731-0574 Nature Publishing Group US New York 136 10.1038/s44159-022-00136-x Review Article Benevolent and hostile sexism in a shifting global context http://orcid.org/0000-0002-6973-7233 Barreto Manuela [email protected] http://orcid.org/0000-0002-8010-6870 Doyle David Matthew grid.8391.3 0000 0004 1936 8024 Department of Psychology, University of Exeter, Exeter, UK 2 12 2022 114 1 11 2022 © Crown 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The theory of and research on ambivalent sexism — which encompasses both attitudes that are overtly negative (hostile sexism) and those that seem subjectively positive but are actually harmful (benevolent sexism) — have made substantial contributions to understanding how sexism operates and the consequences it has for women. It is now clear that sexism takes different forms, some of which can be disguised as protection and flattery. However, all forms of sexism have negative effects on how women are perceived and treated by others as well as on women themselves. Some of these findings have implications for understanding other social inequalities, such as ableism, ageism, racism and classism. In this Review, we summarize what is known about the predictors of ambivalent sexism and its effects. Although we focus on women, we also consider some effects on men, in particular those that indirectly influence women. Throughout the Review we point to societal shifts that are likely to influence how sexism is manifested, experienced and understood. We conclude by discussing the broader implications of these changes and specifying areas of enquiry that need to be addressed to continue making progress in understanding the mechanisms that underlie social inequalities. Sexism encompasses attitudes that are both overtly negative and those that seem subjectively positive but are actually harmful. In this Review, Barreto and Doyle describe the predictors of ambivalent sexism and its effects on women, and consider societal shifts that might influence how sexism is manifested, experienced and understood. Subject terms Human behaviour Psychology Society ==== Body pmcIntroduction Addressing the substantial gender inequalities that exist across a range of life domains1 requires an understanding of the effects of sexism. According to ambivalent sexism theory2, which was developed to account for the relationship between (cisgender and heterosexual) men and women, sexism includes a hostile component (overtly negative attitudes about men and women) and a benevolent component (attitudes towards men and women that seem subjectively positive but are actually harmful). These components differ in tone but are positively correlated and work together to perpetuate gender inequalities2. Research suggests that children3,4, young people5,6 and adult men and women around the world7 endorse ambivalent sexism (that is, agree with items that measure both benevolent sexism, such as “women should be protected by men,” and hostile sexism, such as “women seek to gain power by getting control over men.”). Indeed, according to one study, half of the British population holds these attitudes8. Ambivalent sexism is therefore a critical factor in shaping girls’ and women’s lives in a variety of social contexts. Although there has been substantial progress in this area of research9, theoretical insights are often assumed to hold across time, cultures and social groups. Consequently, theoretical advances do not account for societal shifts in gender relationships over time, or consider the socio-political and cultural contexts in which they operate. For example, binary views of gender are more widely challenged than before10 (at least in some places), which influences ideas about what it means to be a man or a woman, as well as what relationships between individuals of different gender groups should look like. In addition, legal and policy developments change the background against which relationships between men and women play out. For example, the number of countries offering paid paternity leave has increased, and so has its uptake11, which has led to greater labour participation of both mothers and fathers12. Although the negative effects of the COVID-19 pandemic on workload and household work burdens disproportionately influenced women13,14, changes in women’s participation in the workforce provided a normative climate against which couples could evaluate, and be evaluated by others as a function of, their decisions in this area. In addition, because divorce and same-sex parenting and single parenting are increasing15,16, men and women now often have both traditionally male and traditionally female roles within families. More broadly, the spread of neoliberalism as a prevailing socio-political ideology has influenced ideas of equality and how best to achieve progress (for example, by changing individuals rather than social structures)17. For women, this shift has been associated with greater agency in terms of workplace involvement and contribution to the global marketplace18, but often without adequate policy and structural support (such as adequate parental leave or strong employment non-discrimination laws). Instead, women are simultaneously tasked with traditional gender chores, such as childcare and housekeeping, while also being told to ‘lean into’ their careers when they inevitably experience obstacles not faced by heterosexual men. Neoliberalism both empowers women to strive for, and blames women for failing to achieve, outcomes that are often beyond their individual control, masking subtler and more blatant ways in which sexism shapes and constricts lives. Although the full extent of the consequences of this global shift is not straightforward, these changes might influence how sexism is expressed and experienced. Researchers have begun to recognize such societal shifts in ideas about gender and romantic relationships beyond heterosexual couples10, but research in this area is still scarce. In addition, the geographical contexts of research on ambivalent sexism have diversified19, but the majority of research is still carried out in a restricted number of countries (including New Zealand, Spain, Turkey, the UK and the USA), so comparative work and reflections on cultural specificities are still largely missing. In this Review, we take stock of the current understanding of ambivalent sexism to facilitate further research that addresses relevant societal shifts and their global contexts. First, we describe benevolent and hostile sexism and their predictors. We then review what is currently known about how both benevolent sexism and hostile sexism influence how women are perceived and treated (by both men and women). Next, we discuss how these types of sexism influence how women feel and behave, as well as romantic relationships between men and women. Although the applicability of findings to present socio-political contexts will be flagged throughout the paper, the final section more thoroughly considers shifts in global context and how these open up avenues for future research. We focus on research published within the past five years, but key older studies are also mentioned where they exemplify core theoretical aspects. We also focus primarily on sexism towards women. Ambivalent attitudes towards men also encompass hostile and benevolent components20, but these attitudes are less well understood. Importantly, they are strongly related to ambivalent sexism towards women and have been proposed to serve the same function of supporting male dominance over women21. Some examples of effects of ambivalent sexism on men are mentioned, especially where their effects on women are most direct. Two forms of sexism Prejudice is traditionally conceptualized as a negative attitude that explains and shapes antagonistic relationships between dominant and subordinate groups22. Sexism is a form of prejudice that specifically subordinates women to men. Although sexism can take very clearly negative (and even violent) forms, attitudes towards women are not necessarily negative in obvious ways; in fact, people often describe women more positively than they describe men — the ‘women are wonderful’ effect23. However, positive descriptions of women tend to be restricted to traits related to warmth (women are sociable and nice), whereas men are more positively described in domains such as agency and competence that determine status and power in society (men are bright and capable)23,24. In addition, relationships between men and women are not necessarily characterized by antagonism; instead, they often involve the coexistence of male dominance with cooperation and even intimacy. Ambivalent sexism theory2,25 was developed to account for these specific circumstances and proposes that sexism combines antipathy (hostile sexism) with subjective benevolence (benevolent sexism) towards women, which together maintain men’s dominance over women. Hostile sexism is similar to the traditional conceptualization of prejudice as antipathy: it is negative in tone and disparages women who challenge traditional gender roles and ideologies (for example, professionally successful women). It communicates a view of gender relationships as competitive, with women wanting to dominate men and threatening men’s higher status in society. By contrast, benevolent sexism has a more positive tone: it idealizes and flatters women who embody traditional ideals (such as stay-at-home mothers), and portrays women as morally pure and uniquely caring, but also as weak and unable to take care of themselves. Benevolent sexism portrays gender relationships as cooperative and complementary, with men in charge of protection and security and women dedicated to nurture and reproduction. Both hostile and benevolent sexism encompass three components, which are assessed using the Ambivalent Sexism Inventory25: paternalism, gender differentiation and heterosexual intimacy (Table 1). Paternalism refers to men’s superiority over women, either aggressively (in hostile sexism) or protectively (in benevolent sexism). Gender differentiation draws a line between men and women, distributing roles associated with power to men (in hostile sexism) and nurturing roles to women (in benevolent sexism). Heterosexual intimacy accommodates heterosexual men’s dependency on women for sexual satisfaction; hostile sexism aims to restrict women’s use of sex to manipulate men and benevolent sexism idealizes women as necessary to complete men.Table 1 Components of hostile and benevolent sexism and example items from the ambivalent sexism inventory25 Overarching component Hostile sexism Benevolent sexism Component Example item Component Example item Paternalism Dominative: defending men’s power over women “Women seek to gain power by getting control over men.” Protective: restricting women’s access to resources and freedoms in order to protect them “Women should be cherished and protected by men.” Gender differentiation Competitive: portraying qualities necessary for high status positions as unique to men “Women exaggerate problems they have at work.” Complementary: ascribing positive traits to women in domains that are inconsequential for status and power “Women, compared to men, have a superior moral sensibility.” Heterosexuality Hostile: controlling women’s sexuality and fearing its use to manipulate men “Once a woman gets a man to commit to her, she usually tries to put him on a tight leash.” Intimate: idealizing women as romantic partners “Every man ought to have a woman whom he adores.” Although hostile and benevolent sexism are opposite in tone, they both draw on gender stereotypes and therefore tend to be positively associated25 across nations7. The more hostile sexism there is in a given society, the more individuals in that society also endorse benevolent sexism7. Correspondingly, women who report more daily experiences with hostile sexism also report more daily experiences with benevolent sexism26. However, because hostile and benevolent sexism express gender stereotypes in distinct ways, there are important differences in how these two forms of sexism are perceived: hostile sexism is regarded as more objectionable than benevolent sexism27, in part because it is perceived as more sexist28. Benevolent sexism is perceived as harmless29 and even romantic30, and this makes men who endorse benevolent sexism seem likeable19,28,31. Hostile sexism is less frequently endorsed7 and expressed, and indeed women report more lifetime experiences with benevolent than hostile sexism26. However, in part because of the warmth it transmits, benevolent sexism can make hostile sexism seem more acceptable when expressed by the same person32. Benevolent sexism is also seen as less objectionable than hostile sexism because it offers women benefits. For example, because benevolent sexism offers protection to women33, men who express benevolent sexism are seen as caring34. In addition, women who endorse benevolent sexism see the social system as fair35 and consequently report greater life satisfaction36. In sum, both benevolent and hostile sexism express the belief that women are and should be submissive to men. However, benevolent sexism is considered more acceptable, and at times even flattering. This positive perception is a key property of ambivalent sexism that contributes to the perpetuation and pervasiveness of gender inequalities. Predictors of ambivalent sexism Understanding how sexism operates requires consideration of why people might endorse sexist views. Whereas some factors predict endorsement of both benevolent and hostile sexism, others appear to uniquely predict one type of sexism (Fig. 1).Fig. 1 Predictors of benevolent and hostile sexism. Unique and shared predictors of benevolent and hostile sexism. Results can differ across studies; the predictors represented here reflect the most consistent associations reported in the literature. Demographic factors Existing comparative evidence using the Ambivalent Sexism Inventory relies upon non-representative samples25, so it is not possible to establish precisely how benevolent and hostile sexism vary across countries. However, the evidence suggests that hostile sexism is strongest in countries characterized by lower gender equality and less wealth, health and education, as measured by United Nations indicators37. These findings suggest that sexism is not only detrimental to women’s own advancement, but might also be detrimental to society as a whole, reducing overall educational achievement and impairing social prosperity. Because benevolent and hostile sexism serve to justify and perpetuate male privilege, it is not surprising that men endorse benevolent and hostile sexism to a greater extent than women25 across nations7, with gender differences typically being larger for hostile than benevolent sexism. Research comparing sexism scores between cisgender (those who identify with the gender they were assigned at birth), transgender (those who identify with a gender different from that assigned at birth) and gender-diverse individuals (those who identity as non-binary, genderfluid or genderqueer, for example) has produced mixed results. One study revealed higher hostile sexism scores among cisgender men, and lower benevolent sexism scores among cisgender women and gender-diverse individuals assigned female at birth, than other gender groups38. However, another study revealed higher scores on both components among transgender than cisgender individuals39. These discrepancies highlight the need for more research in this area. Regarding age, men’s hostile sexism and women’s hostile and benevolent sexism are higher in adolescence and young adulthood, lower in middle adulthood, and again higher in older age. By contrast, men’s benevolent sexism increases with age6,40. This finding is argued to reflect age-normative changes in the importance of goals related to power, identity and relationships that underlie ambivalent sexism, such as the fact that middle-aged individuals have greater relational and role stability as well as greater independence than young and older adults. It remains to be seen whether these age and gender patterns hold across time and cultures with different views on power, identity and relationships. Studies are beginning to show the importance of taking race into account when attempting to understand the drivers of sexism (Box 1). One study showed that Black American women endorse benevolent sexism to a greater extent than white American women41. Crucially, benevolent and hostile sexism are not significantly correlated among Black American participants42, and there is also no gender difference in the endorsement of these two types of sexism among these participants41,42. However, the benevolent sexism subscale of the Ambivalent Sexism Inventory has poor measurement properties for Latinx and African American participants, suggesting that it is not appropriate for assessing this construct in all racial or ethnic groups42. This measurement issue also highlights the need to expand understanding to other cultural contexts and intersections between multiple identities. Even though ambivalent sexism is endorsed across sexual orientations43,44, individuals who are, or desire to be, in heterosexual romantic relationships report stronger benevolent and hostile sexist attitudes than sexual-minority respondents43–45. However, as mentioned above, existing measurement tools are not appropriate for comparing heterosexual and sexual-minority samples, creating doubt about how these differences in scores should be interpreted46. Box 1 How benevolent sexism justifies racist attitudes Research on ambivalent sexism can contribute to understanding how sexism can exacerbate race inequalities. For example, stereotypes of white American women tend to infantilize them as dependent and helpless (in line with benevolent sexism), whereas stereotypes of Black American women portray them as hypersexualized (in line with hostile sexism)173 and Asian American women are seen as both hypersexual and submissive182. Indeed, women from racial and ethnic minorities often experience a combination of racism and sexism183. This racialized sexual harassment can, in turn, justify their sexual exploitation, increase the extent to which they are blamed for sexual violence184, and reduce the extent to which they are willing to complain about sexual harassment for fear of not being believed or of attracting attention to themselves as targets of sexual attention183. Moreover, American students associate whiteness with femininity and blackness with masculinity, which can justify harsher treatment of Black men and women more generally. In addition, stereotypes of strength and aggressiveness contribute to why police officers intervene less frequently in domestic abuse incidents involving Black American women184. Muslim women are essentialized by particular understandings of Islam that render them responsible for the family’s honour by being modest and chaste. Yet, at the same time, these views of Muslim women are at the core of Western Islamophobic perceptions of Muslim people as oppressive and inferior, with the use of the hijab seen to symbolize women’s forceful subjugation185. This simultaneously racist and (benevolently) sexist discourse is well encapsulated by the term ‘hijabophobia’186, which reflects the view of a “submissive and voiceless Muslim woman who needs to be saved from her barbaric and misogynistic religion.”185 Situational factors The more an individual’s circumstances reflect traditional gender roles, the higher their benevolent sexism scores. For example, having more children predicts stronger endorsement of benevolent sexism two years later — and not the other way around47. That is, people might endorse benevolent sexism to justify the traditional gender roles they have adopted in their life, rather than adopting these roles because they endorse benevolent sexism. If this is the case, then changes in gender roles — for example, through increases in same-sex parenting, or men’s increased participation in childcare — might lead to reductions in endorsement of benevolent sexism. Ideological factors Religiosity is another form of traditionalism that drives sexism. Both forms of sexism, but benevolent sexism in particular, have been positively associated with religiosity across affiliations such as Christianity and Islam48–52. Simple reminders of religion can be sufficient to increase endorsement of benevolent sexism53. Some have argued (but not yet demonstrated) that reductions in religiosity worldwide coincide with scientific and technological advances that increase fertility and reduce child mortality. These advances thereby reduce the need to control women’s reproduction and sphere of activity, which was historically facilitated by religious norms54. Thus, one prediction is that declines in religiosity might translate into a reduction in sexist attitudes. Ideological variables related to political conservatism also predict sexism. In fact, political conservatism has been found to explain more variance in ambivalent sexism than gender8. Moreover, in both men and women, hostile sexism is predicted most strongly and consistently by social dominance orientation (a view of the world in which groups of people compete for dominance and superiority), whereas benevolent sexism is most strongly and consistently predicted by right-wing authoritarianism (which stems from perceptions of the world as a dangerous place and reflects a desire for security)55,56. These findings support the idea that hostile sexism is primarily driven by the idea that men’s dominance over women is both appropriate and desirable, a belief that can be shared by men and women. By contrast, benevolent sexism is driven by a need for security (implied in right-wing authoritarianism). These findings lead to the prediction that political rhetoric associated with the rise in right-wing populism and world events that promote the idea that the world is an unsafe place (such as the COVID-19 pandemic) might increase endorsement of these forms of sexism. Further evidence that benevolent sexism is driven by a need for security is that women’s endorsement of benevolent (but not hostile) sexism increases when they believe that men have more hostile attitudes towards women7. Women also endorse benevolent sexism to a greater extent when their fear of crime is enhanced57. This finding leads to the prediction that actions that highlight women’s vulnerability to sexual violence (for example the #MeToo movement) might ironically increase women’s feelings of insecurity and their endorsement of benevolent sexism in an attempt to secure protection. Similarly, particularly high exposure to discrimination among Black American women (which raises the need for safety) might explain why they endorse benevolent sexism to a greater extent than white American women41, but this has not been directly tested. Furthermore, men and women who are more afraid of disease and contagion endorse benevolent sexism to a greater extent, presumably because the restrictions benevolent sexism imposes on women’s behaviour can protect against disease58. This finding is particularly interesting in light of the COVID-19 pandemic — fear of disease during the pandemic might have led to increases in benevolent sexism. Finally, men’s benevolent sexism increases when they feel anxious about their sense of manhood59 or their romantic relationship60. Interestingly, men who do not have such security needs (men with a tendency to avoid attachment) report low benevolent and high hostile sexism60. In sum, a range of factors increase benevolent and hostile sexism, some of which are unique to each form of sexism. Importantly, changes within a given society in these various predictors (for example, general decreases in religiosity, or temporary fluctuations in insecurity, particularly for women) might have implications for the manifestation of ambivalent sexism. The direct links between these societal changes and endorsement of ambivalent sexism requires further evidence. Effects of ambivalent sexism It is important to understand the different ways in which sexism can be expressed because they can have different consequences. In this section we summarize and compare the effects of benevolent and hostile sexism. Although the review is not exhaustive, it includes those effects that are most crucial for understanding the impact of ambivalent sexism across a range of domains (Table 2).Table 2 Summary of key effects of benevolent and hostile sexism Domain Associations with hostile sexism Associations with benevolent sexism Gender roles Negative attitudes towards men and women who behave non-traditionally61 Positive attitudes towards men and women who behave traditionally25 Self-views Body dissatisfaction79 Stereotypical self-descriptions75 and body dissatisfaction79 Affect and physiology Increased stress response87 and anger92 Delayed stress recovery87 and anxiety90 Violence towards women Belief that victims of sexual assault actually want sex97 Belief that victims of sexual assault have behaved inappropriately99 Careers Fewer hiring recommendations for women62 and less support for female managers106 Stereotypical career choices116, reduced self-efficacy120, and more dependency-oriented support for women at work111, leading women to be perceived as incompetent112 Healthcare Less support for women’s (but not men’s) pain management125 Discouraging women from accessing medical treatment126; restrictive attitudes towards pregnant women127 Legal decisions — More lenient criminal sentencing for women than men134 Gender roles Hostile and benevolent sexism contribute to maintaining the status quo by regulating how women (and men) behave. Hostile sexism is correlated with negative stereotypes or disparaging views about women who challenge the status quo by behaving non-traditionally, such as career women61, women in stereotypical male employment positions (such as managers)62 or feminists61. By contrast, benevolent sexism is associated with positive stereotypes about or support for women who reinforce gender inequalities by behaving in line with traditional gender roles, such as housewives7,25 or women who do not confront sexism63. In addition, hostile sexism punishes women who deviate from traditional gender roles and benevolent sexism encourages women to abide by them in exchange for protection and financial security. For example, women’s endorsement of hostile sexism is associated with the derogation of women who breastfeed in public64 and women who are highly sexually active65; men’s endorsement of benevolent sexism is associated with favourable views of women who breastfeed their children in private66, and predicts unfavourable attitudes towards women who engage in pre-marital sex67. Men who endorse benevolent sexism often engage in protective behaviours towards women (the ‘white knight’ effect)33, and the idea that women need protection is often used as an argument in favour of restricting transgender women’s access to the bathroom of their affirmed gender68. Men do not necessarily benefit from these restrictive attitudes. Indeed, both men and women who do not conform to the rigid gender role prescriptions that underlie ambivalent sexism — such as LGB individuals69,70, men who perform stereotypically feminine behaviours (such as styling someone’s hair)71, men who express gender-egalitarian beliefs72 and transgender individuals73,74 — are the target of negative attitudes, particularly by those high in hostile sexism70. This lack of conformity is perceived to threaten the gender hierarchy in which men dominate, so it is not surprising that these negative attitudes tend to be stronger among men than women73,74. These rigid notions of gender contribute to regulating men’s behaviour, and directly or indirectly influence women’s social standing. It is unclear whether these gender role prescriptions (and their effects on how men and women are perceived) are retained as men and women are seen to successfully take on more counter-stereotypical roles, such as women being successful at work or men successfully parenting. Self-views Sexism influences how women feel and think about themselves and their bodies. Benevolent sexism is particularly problematic in this regard because its flattering and less obviously sexist tone discourages women from rejecting the stereotypes it makes salient. Consequently, women exposed to benevolent (but not hostile) sexism describe themselves more in line with gender stereotypes and remember more gender-stereotypical information about themselves75,76. Beauty ideals are important for the subjugation of women because they often reduce women to sex objects, draw attention away from their competence, and undermine their self-confidence, thereby facilitating men’s dominance. Both benevolent and hostile sexism are associated with the endorsement of beauty ideals (such as thin bodies)77, self-objectification78 and body dissatisfaction79. These, in turn, make women vulnerable to psychological ill health, for example, by decreasing adherence to physical medical exams and exacerbating eating disorders80. Interestingly, benevolent sexism has been associated with both thin77,81 and large79 body ideals, the former presumably because they render women fragile and dependent, and the latter presumably because large bodies signal fertility. In addition, benevolent sexism has also been associated with women’s increased use of cosmetics, which can improve satisfaction with appearance81 and reverse the relationship between benevolent sexism and body image79,82. In sum, both forms of sexism lead to attitudes that seek to control, and draw attention to, women’s appearance, but the effects of benevolent sexism are slightly more complex. Sexism also influences men’s views of themselves and their bodies. Although sexism can enhance the value of being a man, such narrow notions of masculinity can lead those who do not (always) fit this notion to experience low self-esteem and body dissatisfaction83. The role of ambivalent sexism in beauty ideals for transgender and gender-diverse people has not been directly researched and is an important focus for future research. Affect and physiology Automatic responses to both types of sexism are evident in changes in physiology and affect, which might place women at increased risk of physiological ‘wear and tear’, including cardiovascular disease, over the life course84. Cardiovascular disease is the leading cause of mortality among women around the world, but it remains under-recognized, underdiagnosed and undertreated85. Thus, research into the specific contributions of ambivalent sexism to this condition is critical to health equity. For example, being a target of either benevolent86,87 or hostile87 sexism leads to cardiovascular signatures indicative of threat. However, being a target of hostile sexism leads to a greater initial spike in cardiovascular reactivity, whereas benevolent sexism leads to a lower initial spike but slower recovery to baseline87 (Fig. 2). These findings might be consistent with evidence that exposure to benevolent sexism increases activation of the dorsolateral prefrontal cortex, a brain region involved in cognitive control and thought suppression, suggesting that women ruminate about benevolent sexism for some time after experiencing it88. Sexism can also be a substantial physical stressor for men when they feel their adherence to strict notions of masculinity is questioned89.Fig. 2 Cardiovascular responses to ambivalent sexism. Women’s cardiovascular systems are relatively more reactive to experiences of hostile sexism, but relatively slower to recover and return to baseline after experiences of benevolent sexism. This illustration is based on data from ref.87, where cardiovascular responses were measured by systolic and diastolic blood pressure, heart rate, cardiac output, pre-ejection period and total peripheral resistance. Consistent with the portrayal of sexism as a stressor, it can elicit anxiety in men29. For women, experiences with both benevolent and hostile sexism are associated with increased self-reported anxiety90,91 and anger28,92. However, these associations are relatively stronger for hostile than benevolent sexism28,91, perhaps because women do not always identify benevolent sexism as overtly (or uniquely) negative. Men and women tend to overestimate and underestimate how women’s affect will be influenced by exposure to hostile and benevolent sexism, respectively, potentially because they have only a naive understanding of the difference between them29. Furthermore, some evidence suggests that the affective impact of benevolent sexism varies depending on the specific component of benevolent sexism experienced; specifically, one study showed that experiences with protective paternalism are associated with more self-doubt, lower self-esteem and poorer psychological wellbeing, whereas experiencing complementary gender differentiation was associated with less self-doubt, more self-esteem and better wellbeing93. Future work must continue to disentangle the overlapping and unique affective and physiological sequelae of exposure to various forms of ambivalent sexism among women. Violence towards women Restrictive gender role prescriptions can encourage men who feel their masculinity is threatened to behave in ways that they believe demonstrate their manhood, such as displaying aggression94. Only hostile sexism has been shown to predict men’s self-reported likelihood to sexually harass women95 and tolerance of sexual harassment96. However, both hostile and benevolent sexism predict men’s inclination to commit acquaintance rape and blame victims of sexual assault97,98. For hostile sexism, this is because it is associated with the idea that women actually want and control sex even when they claim not to. For benevolent sexism, this is restricted to cases of acquaintance rape and attributed to the idea that women who enter a relationship with a man invite sexual attention97. Because of these perceptions of victims’ culpability, those high (versus low) in benevolent sexism recommend more lenient sentences for perpetrators of acquaintance rape99. In addition, because those high in hostile sexism believe that victims actually want sex, hostile sexism predicts less support for measures that reduce male violence towards women and more support for measures that encourage women to avoid male violence100; benevolent sexism is positively associated with support for both types of measure owing to its focus on women’s protection100. However, the protection against violence offered by benevolent sexism does not necessarily extend to Black women. In situations where police shoot suspects of armed robberies, benevolent sexism leads to perceptions of white (versus Black) female suspects as more feminine, which in turn leads to more blame on the officer than the suspect when the suspect is white, but not when she is Black101. This underlines the need for more research into the intersection of race and gender to examine the limits of ambivalent sexism theory, or expand it to diverse racial groups. Hostile sexism is also linked to sexual aggression towards women by increasing objectification102,103 and denying women uniquely human emotions104. Benevolent sexism has no such effect. In fact, one study showed that, for both men and women, benevolent sexism increases the association of women with positive and uniquely human emotions104. The fact that benevolent sexism can promote this positive image of women might be another reason why women feel flattered by it, despite the fact that it can nevertheless be associated with negative outcomes, including gender violence. Careers Sexism influences how women are perceived and treated in the work domain. For example, hostile sexism is associated with the idea that gender income inequality is legitimate because it arises from women’s choice of work arrangements that are associated with lower salaries105. In addition, hostile sexism leads to fewer recommendations to hire women as managers62 and predicts negative attitudes towards women managers106. Once at work, female employees are often treated in benevolently sexist ways by receiving ample praise but little concrete recognition for their work, such as career-enhancing opportunities107, promotions or salary raises108,109. Benevolent sexism is associated with lower competency standards for female (versus male) employees, resulting in positive evaluations of women when they are compared to other women, but not when they are compared to men (to whom they are deemed inferior)110. Benevolent sexism also results in more dependency-oriented (versus autonomy-oriented) help offered to female employees111, which leads others to perceive women as less competent112, irrespective of whether or not they have requested the help offered113. Merely observing a female job candidate being treated in a benevolently sexist manner leads observers to infer that she is less competent or hireable114. Finally, benevolent sexism has been related to more support for employment equity policies, but only for stereotypically feminine, not masculine, positions115. Taken together, this evidence suggests that benevolent sexism encourages behaviours towards female employees that seem positive, but in fact undermine women’s careers. Thus benevolent sexism might partially explain why women remain under-represented in higher-status and more-powerful roles. It remains to be examined whether these relationships become weaker when and where sexist individuals are a minority in the workplace and their attitudes towards female employees have less power. In terms of career choices, benevolent sexism directs boys to stereotypically male domains, such as business- and maths-related fields, and girls to stereotypically female domains, such as the arts116,117. These choices are often influenced by mothers’ benevolent sexist attitudes118. In addition to shaping career choices, benevolent sexism can impair how women actually perform at work, especially if the task is stereotypically masculine119, by decreasing self-efficacy120 and increasing thought intrusions88,121. At the same time, benevolent sexism restricts women’s access to career-enhancing support122. Women high (versus low) in benevolent sexism are more likely to accept patronizing behaviour from men, which they might perceive as supportive, but which can perpetuate their dependence on men and undermine their career prospects34,111. Irrespective of their benevolent sexist attitudes, women might refuse support when they believe that accepting such support would confirm the sexist belief that they are dependent upon men123. Together, these findings show that although hostile sexism has more immediate and negative emotional effects than benevolent sexism, both negatively influence women’s views of themselves, and benevolent sexism in particular shapes women’s career choices and performance. The fact that benevolent sexism is often not identified as problematic means that an important deterrent of women’s careers frequently remains unaddressed. However, research on this topic might need updating, particularly because some effects of benevolent sexism rely on its subtlety and perceived flattery, which might wane when and where its sexist nature is more visible. Healthcare Both men’s and women’s healthcare is compromised by sexist views of women as emotional and men as brave124. However, only support for addressing women’s (but not men’s) pain is negatively related to benevolent and hostile sexism125. Moreover, patronizing attitudes characteristic of benevolent sexism are associated with discouraging women to undergo mammography to avoid the anxiety it might provoke, despite evidence suggesting that mammography reduces women’s anxiety about having breast cancer126. In addition, the idealization of women as mothers (which is fundamental to benevolent sexism) leads to controlling attitudes about pregnant women’s choices127 and opposition to both elective and traumatic abortion128. Men’s and women’s benevolent sexism is associated with negative attitudes towards women who have an abortion, even if it is medically motivated129. In fact, although sexist attitudes can coerce women towards abortion when families seek to restrict the birth of female children130, sexism can also limit access to abortion. For example, benevolently sexist language has been identified in policy-making discussions to justify restricting women’s access to abortion services131. Ironically, rather than protecting women’s health, research in the USA has shown that state-level abortion bans are tied to increased total maternal mortality132. Consistent with benevolent sexism, those who object to abortion often claim that they wish to protect women from the negative emotions it might elicit (such as shame, grief and regret) and portray women as incapable of making good decisions133. Such arguments might take on greater importance as abortion becomes legal in more places because they provide an additional (but informal) hurdle women might need to overcome to access this care130. Of course, benevolently sexist arguments can also be used to ensure that abortion does not become legal, as in the USA, where the Supreme Court overturned previously established abortion rights in 2022. Legal decisions Finally, court decisions and criminal sentencing often reflect benevolent sexism, in this case often benefiting women134. For example, judges tend to sentence female defendants to less time in prison than male defendants for the same crime, which can be attributed to benevolent sexist ideas that women are weaker than men. Similarly, judges are more likely to allow a divorced mother to relocate with her children away from the father than when exactly the same case is presented by a father, which in turn can be attributed to the benevolently sexist belief that mothers are inherently more essential to children than fathers. The legalization of gay marriage in some countries, and associated shifts in the prevalence and visibility of same-sex parenting, might make men’s ability to provide appropriate parenting more evident and bring about change in this type of decision-making. Clearly, although these effects of benevolent sexism might bring some benefits to women, they contribute towards portraying women as weak and restricting them to the domestic sphere. Effects on heterosexual relationships The desire to sustain the historical norm of heterosexual relationships between cisgender men and women to raise children was originally proposed as one of the driving forces behind ambivalent sexism25,135. Accordingly, an impressive body of research now addresses how ambivalent sexism plays out within heterosexual relationships between cisgender men and women136. Some women (and men) might be romantically persuaded by the chivalry inherent in benevolent sexism137. Benevolent sexism might play a seductive part in heterosexual women’s initial attraction to men because it promises adoration and willingness to invest by potential male partners138. Indeed, women rate benevolently sexist male strangers as more likeable and sexually attractive than hostilely sexist, or even non-sexist, male strangers31. This is especially true for women higher in need of security in romantic relationships (for instance, women higher in attachment anxiety)139. Women’s benevolent sexism is also associated with preferences for male romantic partners who possess traits more consistent with traditional gender roles, such as the ability to provide status and/or resources140,141. By contrast, men’s hostile sexism is associated with preferences for female romantic partners who possess traits more consistent with traditional gender roles, such as attractiveness or vitality140. Among men and women, both benevolent and hostile sexism are associated with greater endorsement of double standards in heterosexual dating (such as the idea that men, not women, should ask for the first date and pay for the date)142. Thus, both hostile and, particularly, benevolent sexism influence heterosexual cisgender women to pursue more traditional heteronormative partners and potential relationships. Once in established heterosexual intimate relationships, both men’s and women’s benevolent and hostile sexism can shape the ways in which romantic partners interact and how their relationships function over time. For example, benevolent sexism promotes traditional task divisions for women143 and men in heterosexual couples144. By ostensibly providing women with a sphere of influence (within rather than outside the home), ambivalent sexism tempts women to become complicit in their own subjugation. For example, hostile sexism among mothers is associated with maternal gatekeeping (behaviours that limit or exclude fathers from childcare), which leads to women performing a greater share of childcare tasks and spending more hours on these tasks than men145. Furthermore, benevolent sexism among women (but not men) is related to intentions to provide dependency-oriented help to male romantic partners when completing stereotypically feminine domestic tasks (such as doing laundry), allowing men to avoid this type of labour in the long run146. Thus, ambivalent sexism perpetuates broader social inequalities around gender by steering women away from education and careers in favour of a primary caregiving role in relationships and family life147. Benevolent sexism can also influence sexual functioning within relationships by focusing the couple on men’s sexual needs and women’s sexual duties148,149. In heterosexual relationships, women’s hostile and benevolent sexism is associated with greater and lesser frequency of faking orgasms, respectively (potentially indicating that women higher in benevolent sexism place less value upon their own sexual pleasure)150. Furthermore, exposure to benevolent sexism reduces condom use during sex, partially owing to women’s motivation to have sex to please a male partner rather than for their own pleasure148. Such behaviours can increase the risk of sexually transmitted infections as well as pregnancy, which can have detrimental health effects and further limit women’s educational and career attainment. Perhaps owing to differences in social acceptability, benevolent versus hostile sexism from male romantic partners is more prevalent in public versus private contexts, respectively151. However, women higher in benevolent sexism are more likely to accept paternalistic restrictions on their behaviours outside of the home (for example, declining a ‘risky’ educational or career opportunity) at their romantic partner’s behest (particularly when the partner offers a justification that is ostensibly about protecting the woman)34. Importantly, women’s endorsement of benevolent sexism is strongly influenced by perceptions of their male partners’ benevolent sexism93,152. Thus, being involved in a relationship with a man who holds benevolently sexist attitudes and ideals might tempt women to view benevolent sexism as a manifestation of love and protection rather than sexism and subjugation. Ambivalent sexism probably leads to a deterioration of relationship quality in heterosexual couples. However, processes by which this might happen can differ for hostile and benevolent sexism and longitudinal research is currently lacking. In general, current evidence suggests that men’s hostile sexism decreases relationship satisfaction for men and women. Indeed, men’s hostile sexism leads to insecurities about women’s independence153 and increases conflict154 and aggression155,156 in heterosexual relationships, which can lead some women to perceive these behaviours as normative and acceptable in intimate relationships157. Women’s benevolent sexism can increase their partner’s relationship satisfaction158, but is associated with shorter relationship length154. The more women endorse the romanticized relationship ideals linked to benevolent sexism, the more dissatisfied they are with their relationship when the couple faces conflict159,160. However, women with attachment insecurities can benefit from perceiving that their partner endorses benevolent sexism when there are low levels of conflict because this reassures the women of their partners’ commitment to the relationship161. Hostile sexism is also associated with negative attitudes towards non-traditional family planning, such as surrogacy162. However, there is little research on how ambivalent sexism influences minority sexual relationships163,164. Furthermore, the Ambivalent Sexism Inventory produces different means and item loadings across heterosexual individuals and sexual-minority individuals46. Thus, this inventory might not reflect how sexism is experienced by sexual-minority individuals and should not be used to compare groups on the basis of sexual orientation. Future research on the effects of ambivalent sexism on romantic relationships should investigate how these processes might function among individuals of diverse sexual and gender identities. Summary and future directions Theoretical and empirical knowledge about ambivalent sexism has improved our understanding of gender inequalities by shedding light on how women are subordinated through the tandem operation of hostile and benevolent sexism. Hostile sexism has more obvious effects, but benevolent sexism is equally damaging and more insidious, largely because it wears a cloak of flattery and protection. Over the almost three decades of research in this area, there has been little effort to consider the changing and global context in which sexism operates. Future research will need to examine whether these societal shifts have been accompanied by changes in how these forms of sexism are expressed, perceived and experienced (Table 3). For example, it is likely that an increased understanding of how sexism operates has produced reductions in both types of sexism, at least in some places, with benevolent sexism potentially showing a slower decline owing to its positive tone40. Ambivalent sexism theory was developed to account for the specific characteristics of gender relationships as they were understood at the time. However, men’s and women’s roles have changed, even if not everywhere165. For example, more women in the USA occupy high-status positions in employment, or are family breadwinners, in the 2020s than in the 1990s166. These changes in gender roles can have contradictory effects. For example, they might serve to showcase women’s perceived competence in the work domain and men’s perceived suitability as carers, and increase cooperation between men and women, which could reduce sexism167,168. However, more egalitarian gender roles might ironically increase gender competition and dominative paternalism to keep women in place and protect the gender hierarchy. The direction of these changes might be influenced by factors such as individuals’ baseline levels of sexism, ultimately leading to a more polarized society (though perhaps with a smaller minority of sexist individuals).Table 3 Potential ramifications of the shifting global context for ambivalent sexism Contextual shift Ramifications for hostile sexism Ramifications for benevolent sexism Greater awareness of how sexism operates might… Render hostile sexism increasingly less acceptable Render benevolent sexism increasingly less acceptable, but with a slower decline than for hostile sexism More egalitarian gender roles might… Increase gender competition Increase dominative paternalism to keep women in their place Increase the need for cooperation between men and women, reducing hostile sexism overall Reduce paternalism and gender complementarity Increased visibility of same-sex couples might… Increase targeting of non-traditional families (such as lesbian mothers) to protect the status quo Increase efforts to reward traditional families to protect the status quo Encourage progressively more inclusive views of what traditional families consist of, reducing benevolent sexism Increased visibility of gender diversity might… Increase targeting of gender-nonconforming individuals to protect the gender hierarchy Increase efforts to reward gender conformity to protect the gender hierarchy Encourage progressively more inclusive views of what gender (and therefore also gender conformity) consists of The increased awareness and acceptability of non-traditional notions of gender, such as transgender and non-binary gender identities or expressions169, or of non-traditional families, such as those with same-sex parents, might also influence gender-related processes. Those who endorse hostile sexism might attempt to protect the gender hierarchy by targeting gender-nonconforming individuals and non-traditional families (such as lesbian mothers) and rewarding women who abide by gender norms. However, it is also possible that these non-traditional gender identities and families could contribute to further changes in societal understanding of gender and gender norms. Future research should examine how perceptions and experiences of sexual-minority, transgender and nonbinary individuals might be influenced by the restrictive views of gender communicated and supported by hostile and benevolent sexism, and how these, in turn, might change with increased exposure to gender-nonconforming individuals. Although research on ambivalent sexism has shed light on how attitudes towards other groups operate (Box 2), more research is needed to understand the intersection between gender and other characteristics, such as age, disability or sexual orientation. For example, little is known about how ambivalent sexism influences wellbeing and relationship functioning in same-sex couples. There is evidence that sexism contributes to intimate partner violence163, attitudes towards same-sex parenting164 and objectification170 by sexual-minority individuals and within minority sexual relationships. However, these studies used a measure that is now known not to adequately capture sexism in these populations46. Indeed, the appropriateness of existing measures of sexism beyond populations that are cisgender, heterosexual, mostly white and living in specific cultures has as yet to be confirmed171,172. For example, efforts to validate the ambivalent sexism inventory across cultures have revealed that it might need adjustment to capture sexism in those cultures42. Future research needs to examine the appropriateness of measures for a range of populations and, if necessary, develop new tools to enable comparative research and better serve these groups. Despite growing evidence that the intersectionality between gender and race shapes women’s experiences of ambivalent sexism41,42,101,173 the majority of research in this area has either not specified the racial composition of the samples or has described them as predominantly white. The findings of this research raise questions about the generalizability of ambivalent sexism theory. More research is needed to clarify whether the theory is less applicable to women of diverse racial groups, whether it can be adjusted and expanded to increase its generalizability, and what measures might be needed to capture sexism across racial or ethnic groups. More generally, research examining predictors and consequences of ambivalent sexism tend to be restricted to a few cultural contexts, which cannot be regarded as a proxy for the rest of the world. The vast majority of this research fails to acknowledge the cultural context where it is carried out and so does not always reflect on how these contexts influence the processes uncovered. Although men tend to have more power than women in most societies, the precise cultural and historical context in which gender relationships are lived cannot be ignored. Indeed, there is some evidence that predictors and consequences of sexism can vary across societies as culturally similar as the UK and the USA174. At the same time, some of the research reviewed here reported similar phenomena across different cultural settings. Ultimately, what is needed is more comparative research to shed further light on the cultural contexts of sexism. Rapid developments in societal norms and attitudes towards sex, gender and sexuality across many countries in the past few decades175,176 reflect a global context that is shifting in response to a more intensely interconnected era. These changes are rarely welcomed by everyone and in some cases they are also not permanent. Research needs to more directly examine the effects of these changes, their trajectories across time, and how they influence and are influenced by changes in gender roles and gender-based equality. Socio-political features of this context, such as dominant neoliberal ideology, are likely to influence the ways in which sexism is manifested and entrenched177,178. It is therefore important to understand the effects of ambivalent sexism and its components40,179–181 as manifestations and consequences of sexism morph in response to this shifting global context. Box 2 Implications for other contexts Sexist beliefs can be recruited to justify prejudice and discrimination towards sexual- and gender -minority individuals. By restricting men and women to particular roles and behaviours, sexism discourages other options187. It is therefore not surprising that both hostile and benevolent sexism have been associated with negative attitudes towards homosexuality188–190 and parenting by same-sex couples45 in countries such as England, Turkey and the USA. In addition, the view that (specific types of) cisgender women are vulnerable and in need of protection from and by men has been core to arguments against gender-neutral public bathrooms68. These sexist views simultaneously restrict and subjugate women, sexual minority individuals and transgender people, protecting the status quo and heteronormative male dominance. Insights regarding ambivalent sexism have inspired research into other types of prejudice, which also have ambivalent characteristics that often combine hostility and benevolence, typically involving paternalism. For example, classism135,191, ageism192 and ableism193 are all rooted in stereotypes of warmth and incompetence in groups characterized by low status and cooperative relationships with more powerful groups194. Just as with sexism, ambivalence leads to polarizing attitudes towards members of these groups, with some offered benevolence (for example, those with visible disabilities, who are perceived as dependent) and others being targets of hostility (such as those with less apparent disabilities, who are perceived as wanting unwarranted special privileges)193. This conceptualization has further implications: for example, it emphasizes the importance of clarifying where the line might lie between paternalism and legitimate help. Indeed, although some individuals might legitimately require a level of care or assistance (for example, individuals with memory impairments), where help goes beyond the required level it can easily become paternalistic, which, by discouraging agency, can accelerate the individual’s deterioration. Acknowledgements The authors thank all colleagues whose work has contributed to the state of the art in this field, whether or not space allowed for it to be explicitly mentioned in the article. Author contributions Both authors contributed to all aspects of the article. Peer review Peer review information Nature Reviews Psychology thanks Theresa Vescio and the other, anonymous, reviewers for their contribution to the peer review of this work. Competing interests The authors declare no competing interests. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. United Nations. Gender equality. Goal 5: Achieve gender equality and empower all women and girls. Sustainable Development Goals https://www.un.org/sustainabledevelopment/gender-equality/ (2015). 2. Glick P Fiske ST An ambivalent alliance: hostile and benevolent sexism as complementary justifications for gender inequality Am. Psychol. 2001 56 109 118 10.1037/0003-066X.56.2.109 11279804 3. Hammond MD Cimpian A “Wonderful but weak”: children’s ambivalent attitudes toward women Sex Roles 2021 84 76 90 10.1007/s11199-020-01150-0 4. Gutierrez BC Halim MLD Martinez MA Arredondo M The heroes and the helpless: the development of benevolent sexism in children Sex Roles 2020 82 558 569 10.1007/s11199-019-01074-4 5. de Lemus S Moya M Glick P When contact correlates with prejudice: adolescents’ romantic relationship experience predicts greater benevolent sexism in boys and hostile sexism in girls Sex Roles 2010 63 214 225 10.1007/s11199-010-9786-2 6. Ferragut M Blanca MJ Ortiz-Tallo M Bendayan R Sexist attitudes and beliefs during adolescence: a longitudinal study of gender differences Eur. J. Dev. Psychol. 2017 14 32 43 10.1080/17405629.2016.1144508 7. Glick P Beyond prejudice as simple antipathy: hostile and benevolent sexism across cultures J. Pers. Soc. Psychol. 2000 79 763 775 10.1037/0022-3514.79.5.763 11079240 8. de Geus R Ralph-Morrow E Shorrocks R Understanding ambivalent sexism and its relationship with electoral choice in Britain Br. J. Polit. Sci. 2022 10.1017/S0007123421000612 9. Connor, R. A., Glick, P. & Fiske, S. T. in The Cambridge Handbook of the Psychology of Prejudice (eds Sibley, C. G. & Barlow, F. K.) 295–320 (Cambridge Univ. Press, 2017). 10. Hyde JS Bigler RS Joel D Tate CC van Anders SM The future of sex and gender in psychology: five challenges to the gender binary Am. Psychol. 2019 74 171 193 10.1037/amp0000307 30024214 11. van der Gaag, N., Heilman, B., Gupta, T., Nembhard, C. & Barker, G. State of the world’s fathers: unlocking the power of men’s care. ProMundo https://promundoglobal.org/wp-content/uploads/2019/06/BLS19063_PRO_SOWF_REPORT_015.pdf (2019). 12. ONS. Families and the labour market UK: 2019. Census 2021 https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/articles/familiesandthelabourmarketengland/2019 (2019). 13. Alon, T., Doepke, M., Olmstead-Rumsey, J. & Tertilt, M. This time it’s different: the role of women’s employment in a pandemic recession. National Bureau of Economic Research http://www.nber.org/papers/w27660.pdf (2020). 14. Giurge LM Whillans AV Yemiscigil A A multicountry perspective on gender differences in time use during COVID-19 Proc. Natl Acad. Sci. USA 2021 118 e2018494118 10.1073/pnas.2018494118 33798094 15. Golombok, S. Modern Families: Parents And Children In New Family Forms (Cambridge Univ. Press, 2015). 16. The Williams Institute. How many same-sex couples in the US are raising children? UCLA https://williamsinstitute.law.ucla.edu/wp-content/uploads/Same-Sex-Parents-Jul-2018.pdf (2018). 17. Goudarzi S Badaan V Knowles ED Neoliberalism and the ideological construction of equity beliefs Perspect. Psychol. Sci. 2022 17 1431 1451 10.1177/17456916211053311 35536556 18. Government Equalities Office. Gender equality monitor: tracking progress on gender equality. Gov.uk https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/814080/GEO_GEEE_Strategy_Gender_Equality_Monitor_tagged.pdf (2019). 19. Chisango T Javangwe G Are people better at recognizing ambivalent sexism on the basis of the non-standard profiles than the standard ASI ones? Sex Roles 2012 67 69 82 10.1007/s11199-012-0146-2 20. Glick P Fiske ST The ambivalence toward men inventory: differentiating hostile and benevolent beliefs about men Psychol. Women Q. 1999 23 519 536 10.1111/j.1471-6402.1999.tb00379.x 21. Glick P Bad but bold: ambivalent attitudes toward men predict gender inequality in 16 nations J. Pers. Soc. Psychol. 2004 86 713 728 10.1037/0022-3514.86.5.713 15161396 22. Allport, G. W. The Nature Of Prejudice (Addison-Wesley, 1954). 23. Eagly AH Mladinic A Are people prejudiced against women? Some answers from research on attitudes, gender stereotypes, and judgments of competence Eur. Rev. Soc. Psychol. 1994 5 1 35 10.1080/14792779543000002 24. Fiske ST Cuddy AJC Glick P Universal dimensions of social cognition: warmth and competence Trends Cogn. Sci. 2007 11 77 83 10.1016/j.tics.2006.11.005 17188552 25. Glick P Fiske ST The ambivalent sexism inventory: differentiating hostile and benevolent sexism J. Pers. Soc. Psychol. 1996 70 491 512 10.1037/0022-3514.70.3.491 26. Salomon K The experiences with ambivalent sexism inventory (EASI) Basic Appl. Soc. Psychol. 2020 42 235 253 10.1080/01973533.2020.1747467 27. Becker JC Wright SC Yet another dark side of chivalry: benevolent sexism undermines and hostile sexism motivates collective action for social change J. Pers. Soc. Psychol. 2011 101 62 77 10.1037/a0022615 21355657 28. Barreto M Ellemers N The burden of benevolent sexism: how it contributes to the maintenance of gender inequalities Eur. J. Soc. Psychol. 2005 35 633 642 10.1002/ejsp.270 29. Bosson JK Pinel EC Vandello JA The emotional impact of ambivalent sexism: forecasts versus real experiences Sex Roles 2010 62 520 531 10.1007/s11199-009-9664-y 30. Rudman LA Heppen JB Implicit romantic fantasies and women’s interest in personal power: a glass slipper effect? Pers. Soc. Psychol. Bull. 2003 29 1357 1370 10.1177/0146167203256906 15189575 31. Bohner G Ahlborn K Steiner R How sexy are sexist men? Women’s perception of male response profiles in the ambivalent sexism inventory Sex Roles 2010 62 568 582 10.1007/s11199-009-9665-x 32. Hopkins-Doyle A Sutton RM Douglas KM Calogero RM Flattering to deceive: why people misunderstand benevolent sexism J. Pers. Soc. Psychol. 2019 116 167 192 10.1037/pspa0000135 30359066 33. Leone RM Schipani-McLaughlin AM Haikalis M Parrott DJ The “white knight” effect: benevolent sexism accounts for bystander intervention in party situations among high status men Psychol. Men. Masc. 2020 21 704 709 10.1037/men0000314 34764820 34. Moya M Glick P Expósito F de Lemus S Hart J It’s for your own good: benevolent sexism and women’s reactions to protectively justified restrictions Pers. Soc. Psychol. Bull. 2007 33 1421 1434 10.1177/0146167207304790 17933737 35. Jost JT Kay AC Exposure to benevolent sexism and complementary gender stereotypes: consequences for specific and diffuse forms of system justification J. Pers. Soc. Psychol. 2005 88 498 509 10.1037/0022-3514.88.3.498 15740442 36. Connelly K Heesacker M Why is benevolent sexism appealing?: Associations with system justification and life satisfaction Psychol. Women Q. 2012 36 432 443 10.1177/0361684312456369 37. Brandt MJ Sexism and gender inequality across 57 societies Psychol. Sci. 2011 22 1413 1418 10.1177/0956797611420445 21987694 38. Schiralli JE Peragine DE Chasteen AL Einstein G Explicit and implicit gender-related stereotyping in transgender, gender expansive, and cisgender adults Arch. Sex. Behav. 2022 51 2065 2076 10.1007/s10508-022-02339-y 35467169 39. Khorashad BS Roshan GM Talaei A Arezoomandan S Sadr M Views of individuals with gender dysphoria and disorders of sex development on sexism: an Iranian study Int. J. Transgend. 2019 20 459 470 10.1080/15532739.2018.1445573 32999630 40. Hammond MD Milojev P Huang Y Sibley CG Benevolent sexism and hostile sexism across the ages Soc. Psychol. Personal. Sci. 2018 9 863 874 10.1177/1948550617727588 41. Davis TM Settles IH Jones MK Standpoints and situatedness: examining the perception of benevolent sexism in Black and white undergraduate women and men Psychol. Women Q. 2022 46 8 26 10.1177/03616843211043108 42. Hayes E-R Swim JK African, Asian, Latina/o, and European Americans’ responses to popular measures of sexist beliefs: some cautionary notes Psychol. Women Q. 2013 37 155 166 10.1177/0361684313480044 43. Cowie LJ Greaves LM Sibley CG Sexuality and sexism: differences in ambivalent sexism across gender and sexual identity Personal. Individ. Differ. 2019 148 85 89 10.1016/j.paid.2019.05.023 44. López-Sáez MÁ García-Dauder D Montero I Correlate attitudes toward LGBT and sexism in Spanish psychology students Front. Psychol. 2020 11 2063 10.3389/fpsyg.2020.02063 32973622 45. Pistella J Tanzilli A Ioverno S Lingiardi V Baiocco R Sexism and attitudes toward same-sex parenting in a sample of heterosexuals and sexual minorities: the mediation effect of sexual stigma Sex. Res. Soc. Policy 2018 15 139 150 10.1007/s13178-017-0284-y 46. Cross EJ Muise A Hammond MD Do scales measuring sexist attitudes have equivalent meaning for sexual minorities and majorities? Sex Roles 2021 85 707 720 10.1007/s11199-021-01248-z 47. Deak CK Hammond MD Sibley CG Bulbulia J Individuals’ number of children is associated with benevolent sexism PLoS One 2021 16 e0252194 10.1371/journal.pone.0252194 34043695 48. Glick P Lameiras M Castro YR Education and Catholic religiosity as predictors of hostile and benevolent sexism towards women and men Sex Roles 2002 47 433 441 10.1023/A:1021696209949 49. Mikołajczak M Pietrzak J Ambivalent sexism and religion: connected through values Sex Roles 2014 70 387 399 10.1007/s11199-014-0379-3 24910494 50. Burn SM Busso J Ambivalent sexism, scriptural literalism, and religiosity Psychol. Women Q. 2005 29 412 418 10.1111/j.1471-6402.2005.00241.x 51. Maitner AT Henry PJ Ambivalent sexism in the United Arab Emirates: quantifying gender attitudes in a rapidly modernizing society Group Process Intergroup Relat. 2018 21 831 843 10.1177/1368430217740433 52. Taşdemir N Sakallı-Uğurlu N The relationships between ambivalent sexism and religiosity among Turkish university students Sex Roles 2010 62 420 426 10.1007/s11199-009-9693-6 53. Haggard MC Kaelen R Saroglou V Klein O Rowatt WC Religion’s role in the illusion of gender equality: supraliminal and subliminal religious priming increases benevolent sexism Psychol. Relig. Spiritual. 2019 11 392 398 10.1037/rel0000196 54. Inglehart, R. Religion’s Sudden Decline: What’s Causing It, And What Comes Next? (Oxford Univ. Press, 2021). 55. Hellmer K Stenson JT Jylhä KM What’s (not) underpinning ambivalent sexism?: revisiting the roles of ideology, religiosity, personality, demographics, and men’s facial hair in explaining hostile and benevolent sexism Personal. Individ. Differ. 2018 122 29 37 10.1016/j.paid.2017.10.001 56. Sibley CG Wilson MS Duckitt J Antecedents of men’s hostile and benevolent sexism: the dual roles of social dominance orientation and right-wing authoritarianism Pers. Soc. Psychol. Bull. 2007 33 160 172 10.1177/0146167206294745 17259578 57. Phelan JE Sanchez DT Broccoli TL The danger in sexism: the links among fear of crime, benevolent sexism, and well-being Sex Roles 2010 62 35 47 10.1007/s11199-009-9711-8 58. Fitzgerald HN McDonald R Thomas R Shook NJ Disease avoidance: a predictor of sexist attitudes toward females Curr. Psychol. 2021 10.1007/s12144-020-01343-6 59. Valved T Kosakowska-Berezecka N Besta T Martiny SE Gender belief systems through the lens of culture — differences in precarious manhood beliefs and reactions to masculinity threat in Poland and Norway Psychol. Men. Masc. 2021 22 265 276 10.1037/men0000331 60. Fisher MI Hammond MD Personal ties and prejudice: a meta-analysis of romantic attachment and ambivalent sexism Pers. Soc. Psychol. Bull. 2019 45 1084 1098 10.1177/0146167218804551 30413137 61. Glick P Diebold J Bailey-Werner B Zhu L The two faces of Adam: ambivalent sexism and polarized attitudes toward women Pers. Soc. Psychol. Bull. 1997 23 1323 1334 10.1177/01461672972312009 62. Masser B Abrams D Reinforcing the glass ceiling: the consequences of hostile sexism for female managerial candidates Sex Roles 2004 51 609 615 10.1007/s11199-004-5470-8 63. Kahn KB van Breen JA Barreto M Kaiser CR When is women’s benevolent sexism associated with support for other women’s agentic responses to gender‐based threat? Br. J. Soc. Psychol. 2021 60 786 807 10.1111/bjso.12441 33455018 64. Huang Y Sibley CG Osborne D Breast is best, but where? Hostile sexism underlies men’s opposition to breastfeeding in public J. Soc. Issues 2020 76 219 238 10.1111/josi.12379 65. Zaikman Y Marks MJ Ambivalent sexism and the sexual double standard Sex Roles 2014 71 333 344 10.1007/s11199-014-0417-1 66. Forbes GB Adams-Curtis LE Hamm NR White KB Perceptions of the woman who breastfeeds: the role of erotophobia, sexism, and attitudinal variables Sex Roles 2003 49 379 388 10.1023/A:1025116305434 67. Sakalh‐Uğurlu N Glick P Ambivalent sexism and attitudes toward women who engage in premarital sex in Turkey J. Sex. Res. 2003 40 296 302 10.1080/00224490309552194 14533024 68. Blumell LE Huemmer J Sternadori M Protecting the ladies: benevolent sexism, heteronormativity, and partisanship in online discussions of gender-neutral bathrooms Mass Commun. Soc. 2019 22 365 388 10.1080/15205436.2018.1547833 69. Carnaghi A Maass A Fasoli F Enhancing masculinity by slandering homosexuals: the role of homophobic epithets in heterosexual gender identity Pers. Soc. Psychol. Bull. 2011 37 1655 1665 10.1177/0146167211424167 21975948 70. Rollè L Sechi C Santoniccolo F Trombetta T Brustia P The relationship between sexism, affective states, and attitudes toward homosexuality in a sample of heterosexual Italian people Sex. Res. Soc. Policy 2022 19 194 206 10.1007/s13178-021-00534-5 71. Bosson JK Prewitt-Freilino JL Taylor JN Role rigidity: a problem of identity misclassification? J. Pers. Soc. Psychol. 2005 89 552 565 10.1037/0022-3514.89.4.552 16287418 72. Rudman LA Mescher K Moss-Racusin CA Reactions to gender egalitarian men: perceived feminization due to stigma-by-association Group Process Intergroup Relat. 2013 16 572 599 10.1177/1368430212461160 73. Konopka K Rajchert J Dominiak-Kochanek M Roszak J The role of masculinity threat in homonegativity and transphobia J. Homosex. 2021 68 802 829 10.1080/00918369.2019.1661728 33650942 74. Uluboy Z Husnu S Turkish speaking young adults attitudes toward transgender individuals: transphobia, homophobia and gender ideology J. Homosex. 2022 69 101 119 10.1080/00918369.2020.1813510 32875957 75. Barreto M Ellemers N Piebinga L Moya M How nice of us and how dumb of me: the effect of exposure to benevolent sexism on women’s task and relational self-descriptions Sex Roles 2010 62 532 544 10.1007/s11199-009-9699-0 76. Dumont M Sarlet M Dardenne B Be too kind to a woman, she’ll feel incompetent: benevolent sexism shifts self-construal and autobiographical memories toward incompetence Sex Roles 2010 62 545 553 10.1007/s11199-008-9582-4 77. Forbes GB Collinsworth LL Jobe RL Braun KD Wise LM Sexism, hostility toward women, and endorsement of beauty ideals and practices: are beauty ideals associated with oppressive beliefs? Sex Roles 2007 56 265 273 10.1007/s11199-006-9161-5 78. Calogero RM Jost JT Self-subjugation among women: exposure to sexist ideology, self-objectification, and the protective function of the need to avoid closure J. Pers. Soc. Psychol. 2011 100 211 228 10.1037/a0021864 21186932 79. Forbes GB Body dissatisfaction in college women and their mothers: cohort effects, developmental effects, and the influences of body size, sexism, and the thin body ideal Sex Roles 2005 53 281 298 10.1007/s11199-005-5686-2 80. Forbes GB Body dissatisfaction and disordered eating in three cultures: Argentina, Brazil, and the US Sex Roles 2012 66 677 694 10.1007/s11199-011-0105-3 81. Franzoi SL Is female body esteem shaped by benevolent sexism? Sex Roles 2001 44 177 188 10.1023/A:1010903003521 82. Oswald DL Franzoi SL Frost KA Experiencing sexism and young women’s body esteem J. Soc. Clin. Psychol. 2012 31 1112 1137 10.1521/jscp.2012.31.10.1112 83. Atkins, D. Looking Queer (Routledge, 2012). 84. Molix L Sex differences in cardiovascular health: does sexism influence women’s health? Am. J. Med. Sci. 2014 348 153 155 10.1097/MAJ.0000000000000300 25054736 85. Vogel B The Lancet Women and Cardiovascular Disease Commission: reducing the global burden by 2030 Lancet 2021 397 2385 2438 10.1016/S0140-6736(21)00684-X 34010613 86. Lamarche VM Seery MD Kondrak CL Saltsman TL Streamer L Clever girl: benevolent sexism and cardiovascular threat Biol. Psychol. 2020 149 107781 10.1016/j.biopsycho.2019.107781 31618662 87. Salomon K Burgess KD Bosson JK Flash fire and slow burn: women’s cardiovascular reactivity and recovery following hostile and benevolent sexism J. Exp. Psychol. Gen. 2015 144 469 479 10.1037/xge0000061 25844626 88. Dardenne B Benevolent sexism alters executive brain responses NeuroReport 2013 24 572 577 10.1097/WNR.0b013e3283625b5b 23660680 89. Himmelstein MS Kramer BL Springer KW Stress in strong convictions: precarious manhood beliefs moderate cortisol reactivity to masculinity threats Psychol. Men Masc. 2019 20 491 502 10.1037/men0000187 90. Pacilli MG Spaccatini F Giovannelli I Centrone D Roccato M System justification moderates the relation between hostile (but not benevolent) sexism in the workplace and state anxiety: an experimental study J. Soc. Psychol. 2019 159 474 481 10.1080/00224545.2018.1503993 30142299 91. Spaccatini F Roccato M The palliative function of sexism: individual sexism buffers the relationship between exposure to workplace sexism and psychological distress Sex. Cult. 2021 25 767 785 10.1007/s12119-020-09793-7 92. Lemonaki E Manstead ASR Maio GR Hostile sexism (de)motivates women’s social competition intentions: the contradictory role of emotions Br. J. Soc. Psychol. 2015 54 483 499 10.1111/bjso.12100 25564748 93. Oswald DL Baalbaki M Kirkman M Experiences with benevolent sexism: scale development and associations with women’s well-being Sex Roles 2019 80 362 380 10.1007/s11199-018-0933-5 94. Bosson JK Vandello JA Precarious manhood and its links to action and aggression Curr. Dir. Psychol. Sci. 2011 20 82 86 10.1177/0963721411402669 95. Begany JJ Milburn MA Psychological predictors of sexual harassment: authoritarianism, hostile sexism, and rape myths Psychol. Men Masc. 2002 3 119 126 10.1037/1524-9220.3.2.119 96. Russell BL Trigg KY Tolerance of sexual harassment: an examination of gender differences, ambivalent sexism, social dominance, and gender roles Sex Roles 2004 50 565 573 10.1023/B:SERS.0000023075.32252.fd 97. Abrams D Viki GT Masser B Bohner G Perceptions of stranger and acquaintance rape: the role of benevolent and hostile sexism in victim blame and rape proclivity J. Pers. Soc. Psychol. 2003 84 111 125 10.1037/0022-3514.84.1.111 12518974 98. Chapleau KM Oswald DL Russell BL How ambivalent sexism toward women and men support rape myth acceptance Sex Roles 2007 57 131 136 10.1007/s11199-007-9196-2 99. Viki GT Abrams D Masser B Evaluating stranger and acquaintance rape: the role of benevolent sexism in perpetrator blame and recommended sentence length Law Hum. Behav. 2004 28 295 303 10.1023/B:LAHU.0000029140.72880.69 15264448 100. Brownhalls J Make it safe at night or teach women to fight? Sexism predicts views on men’s and women’s responsibility to reduce men’s violence toward women Sex Roles 2021 84 183 195 10.1007/s11199-020-01159-5 101. Brown-Iannuzzi JL Cooley E Cipolli W Mehta S Race, ambivalent sexism, and perceptions of situations when police shoot black women Soc. Psychol. Personal. Sci. 2022 13 127 138 10.1177/1948550620987659 102. Rudman LA Mescher K Of animals and objects: men’s implicit dehumanization of women and likelihood of sexual aggression Pers. Soc. Psychol. Bull. 2012 38 734 746 10.1177/0146167212436401 22374225 103. Cikara M Eberhardt JL Fiske ST From agents to objects: sexist attitudes and neural responses to sexualized targets J. Cogn. Neurosci. 2011 23 540 551 10.1162/jocn.2010.21497 20350187 104. Viki GT Abrams D Infra-humanization: ambivalent sexism and the attribution of primary and secondary emotions to women J. Exp. Soc. Psychol. 2003 39 492 499 10.1016/S0022-1031(03)00031-3 105. Connor RA Fiske ST Not minding the gap: how hostile sexism encourages choice explanations for the gender income gap Psychol. Women Q. 2019 43 22 36 10.1177/0361684318815468 106. Sakalli-Ugurlu N Beydogan B Turkish college students’ attitudes toward women managers: the effects of patriarchy, sexism, and gender differences J. Psychol. 2002 136 647 656 10.1080/00223980209604825 12523452 107. King EB Benevolent sexism at work: gender differences in the distribution of challenging developmental experiences J. Manag. 2012 38 1835 1866 108. Biernat M Tocci MJ Williams JC The language of performance evaluations: gender-based shifts in content and consistency of judgment Soc. Psychol. Personal. Sci. 2012 3 186 192 10.1177/1948550611415693 109. Vescio TK Gervais SJ Snyder M Hoover A Power and the creation of patronizing environments: the stereotype-based behaviors of the powerful and their effects on female performance in masculine domains J. Pers. Soc. Psychol. 2005 88 658 672 10.1037/0022-3514.88.4.658 15796666 110. Cassidy BS Krendl AC A crisis of competence: benevolent sexism affects evaluations of women’s competence Sex Roles 2019 81 505 520 10.1007/s11199-019-1011-3 111. Shnabel N Bar-Anan Y Kende A Bareket O Lazar Y Help to perpetuate traditional gender roles: benevolent sexism increases engagement in dependency-oriented cross-gender helping J. Pers. Soc. Psychol. 2016 110 55 75 10.1037/pspi0000037 26461798 112. Reilly ED Rackley KR Awad GH Perceptions of male and female STEM aptitude: the moderating effect of benevolent and hostile sexism J. Career Dev. 2017 44 159 173 10.1177/0894845316641514 113. Becker JC Glick P Ilic M Bohner G Damned if she does, damned if she doesn't: consequences of accepting versus confronting patronizing help for the female target and male actor Eur. J. Social Psychol. 2011 41 761 773 10.1002/ejsp.823 114. Good JJ Rudman LA When female applicants meet sexist interviewers: the costs of being a target of benevolent sexism Sex Roles 2010 62 481 493 10.1007/s11199-009-9685-6 115. Hideg I Ferris DL The compassionate sexist? How benevolent sexism promotes and undermines gender equality in the workplace J. Pers. Soc. Psychol. 2016 111 706 727 10.1037/pspi0000072 27414234 116. Farkas T Leaper C Chivalry’s double-edged sword: how girls’ and boys’ paternalistic attitudes relate to their possible family and work selves Sex Roles 2016 74 220 230 10.1007/s11199-015-0556-z 117. Fernández ML Castro YR Otero MC Foltz ML Lorenzo MG Sexism, vocational goals, and motivation as predictors of men’s and women’s career choice Sex Roles 2006 55 267 272 10.1007/s11199-006-9079-y 118. Montañés P Intergenerational transmission of benevolent sexism from mothers to daughters and its relation to daughters’ academic performance and goals Sex Roles 2012 66 468 478 10.1007/s11199-011-0116-0 119. Gervais SJ Vescio TK The effect of patronizing behavior and control on men and women’s performance in stereotypically masculine domains Sex Roles 2012 66 479 491 10.1007/s11199-011-0115-1 120. Jones K Negative consequence of benevolent sexism on efficacy and performance Gend. Manag. Int. J. 2014 29 171 189 10.1108/GM-07-2013-0086 121. Dardenne B Dumont M Bollier T Insidious dangers of benevolent sexism: consequences for women’s performance J. Pers. Soc. Psychol. 2007 93 764 779 10.1037/0022-3514.93.5.764 17983299 122. Hideg I Shen W Why still so few? A theoretical model of the role of benevolent sexism and career support in the continued underrepresentation of women in leadership positions J. Leadersh. Organ. Stud. 2019 26 287 303 10.1177/1548051819849006 123. Wakefield JRH Hopkins N Greenwood RM Thanks, but no thanks: women’s avoidance of help-seeking in the context of a dependency-related stereotype Psychol. Women Q. 2012 36 423 431 10.1177/0361684312457659 124. Samulowitz A Gremyr I Eriksson E Hensing G “Brave men” and “emotional women”: a theory-guided literature review on gender bias in health care and gendered norms towards patients with chronic pain Pain. Res. Manag. 2018 2018 1 14 10.1155/2018/6358624 125. Prego-Jimenez S The impact of sexism and gender stereotypes on the legitimization of women’s low back pain Pain Manag. Nurs. 2022 10.1016/j.pmn.2022.03.008 35428592 126. Lee J Hardesty LA Kunzler NM Rosenkrantz AB Direct interactive public education by breast radiologists about screening mammography: impact on anxiety and empowerment J. Am. Coll. Radiol. 2016 13 12 20 10.1016/j.jacr.2015.07.018 26482812 127. Sutton RM Douglas KM McClellan LM Benevolent sexism, perceived health risks, and the inclination to restrict pregnant women’s freedoms Sex Roles 2011 65 596 605 10.1007/s11199-010-9869-0 128. Huang Y Davies PG Sibley CG Osborne D Benevolent sexism, attitudes toward motherhood, and reproductive rights: a multi-study longitudinal examination of abortion attitudes Pers. Soc. Psychol. Bull. 2016 42 970 984 10.1177/0146167216649607 27226340 129. Osborne D Davies PG When benevolence backfires: benevolent sexists’ opposition to elective and traumatic abortion J. Appl. Soc. Psychol. 2012 42 291 307 10.1111/j.1559-1816.2011.00890.x 130. Alotey P Ravindran S Sathivelu V Trends in abortion policies in low- and middle-income countries Annu. Rev. Public Health 2021 42 505 518 10.1146/annurev-publhealth-082619-102442 33138701 131. Greubel AD Benevolent sexism in the targeted regulation of abortion providers (TRAP): a case study of Texas House Bill 2 Sex Roles 2021 85 1 11 10.1007/s11199-021-01244-3 132. Vilda D State abortion policies and maternal death in the United States, 2015–2018 Am. J. Public Health 2021 111 1696 1704 10.2105/AJPH.2021.306396 34410825 133. Duerksen KN Lawson KL “Not brain-washed, but heart-washed”: a qualitative analysis of benevolent sexism in the anti-choice stance Int. J. Behav. Med. 2017 24 864 870 10.1007/s12529-017-9633-8 28124193 134. Rachlinski JJ Wistrich AJ Benevolent sexism in judges San. Diego Law Rev. 2021 58 101 142 135. Jackman, M. R. The Velvet Glove: Paternalism And Conflict In Gender, Class, And Race Relations (Univ. California Press, 1994). This is a classic analysis of paternalism in three different intergroup contexts, which inspired ambivalent sexism theory and continues to inspire understandings of the complexity of prejudice. 136. Hammond MD Overall NC Dynamics within intimate relationships and the causes, consequences, and functions of sexist attitudes Curr. Dir. Psychol. Sci. 2017 26 120 125 10.1177/0963721416686213 137. Lamont E Negotiating courtship: reconciling egalitarian ideals with traditional gender norms Gend. Soc. 2014 28 189 211 10.1177/0891243213503899 138. Gul P Kupfer TR Benevolent sexism and mate preferences: why do women prefer benevolent men despite recognizing that they can be undermining? Pers. Soc. Psychol. Bull. 2019 45 146 161 10.1177/0146167218781000 29957149 139. Cross EJ Overall NC Women’s attraction to benevolent sexism: needing relationship security predicts greater attraction to men who endorse benevolent sexism Eur. J. Soc. Psychol. 2018 48 336 347 10.1002/ejsp.2334 140. Lee TL Fiske ST Glick P Chen Z Ambivalent sexism in close relationships: (hostile) power and (benevolent) romance shape relationship ideals Sex Roles 2010 62 583 601 10.1007/s11199-010-9770-x 23914004 141. Travaglia LK Overall NC Sibley CG Benevolent and hostile sexism and preferences for romantic partners Personal. Individ. Differ. 2009 47 599 604 10.1016/j.paid.2009.05.015 142. Paynter A Leaper C Heterosexual dating double standards in undergraduate women and men Sex Roles 2016 75 393 406 10.1007/s11199-016-0628-8 143. Silván-Ferrero MDP Bustillos López A Benevolent sexism toward men and women: justification of the traditional system and conventional gender roles in Spain Sex Roles 2007 57 607 614 10.1007/s11199-007-9271-8 144. Chen Z Fiske ST Lee TL Ambivalent sexism and power-related gender-role ideology in marriage Sex Roles 2009 60 765 778 10.1007/s11199-009-9585-9 24058258 145. Gaunt R Pinho M Do sexist mothers change more diapers? Ambivalent sexism, maternal gatekeeping, and the division of childcare Sex Roles 2018 79 176 189 10.1007/s11199-017-0864-6 146. Bareket O Shnabel N Kende A Knab N Bar-Anan Y Need some help, honey? Dependency-oriented helping relations between women and men in the domestic sphere J. Pers. Soc. Psychol. 2021 120 1175 1203 10.1037/pspi0000292 32584100 147. Overall NC Hammond MD How intimate relationships contribute to gender inequality: sexist attitudes encourage women to trade off career success for relationship security Policy Insights Behav. Brain Sci. 2018 5 40 48 10.1177/2372732217745096 148. Fitz CC Zucker AN Everyday exposure to benevolent sexism and condom use among college women Women Health 2015 55 245 262 10.1080/03630242.2014.996721 25794197 149. Harris EA Hornsey MJ Barlow FK On the link between benevolent sexism and orgasm frequency in heterosexual women Arch. Sex. Behav. 2016 45 1923 1931 10.1007/s10508-016-0780-9 27305907 150. Harris EA Hornsey MJ Larsen HF Barlow FK Beliefs about gender predict faking orgasm in heterosexual women Arch. Sex. Behav. 2019 48 2419 2433 10.1007/s10508-019-01510-2 31309430 151. Chisango T Mayekiso T Thomae M The social nature of benevolent sexism and the antisocial nature of hostile sexism: is benevolent sexism more likely to manifest in public contexts and hostile sexism in private contexts?: antisocial and social sexism Int. J. Psychol. 2015 50 363 371 10.1002/ijop.12106 25303091 152. Hammond MD Overall NC Cross EJ Internalizing sexism within close relationships: perceptions of intimate partners’ benevolent sexism promote women’s endorsement of benevolent sexism J. Pers. Soc. Psychol. 2016 110 214 238 10.1037/pspi0000043 26785062 153. Hammond MD Overall NC Men’s hostile sexism and biased perceptions of partners’ support: underestimating dependability rather than overestimating challenges to dominance Pers. Soc. Psychol. Bull. 2020 46 1491 1506 10.1177/0146167220907475 32153238 154. Leaper C Gutierrez BC Farkas T Ambivalent sexism and reported relationship qualities in emerging adult heterosexual dating couples Emerg. Adulthood 2020 10 776 787 10.1177/2167696820934687 155. Cross EJ Overall NC Low RST McNulty JK An interdependence account of sexism and power: men’s hostile sexism, biased perceptions of low power, and relationship aggression J. Pers. Soc. Psychol. 2019 117 338 363 10.1037/pspi0000167 30475007 156. Overall NC Chang VT Cross EJ Low RST Henderson AME Sexist attitudes predict family-based aggression during a COVID-19 lockdown J. Fam. Psychol. 2021 35 1043 1052 10.1037/fam0000834 33734757 157. Alvarez C Lameiras-Fernandez M Holliday CN Sabri B Campbell J Latina and Caribbean immigrant women’s experiences with intimate partner violence: a story of ambivalent sexism J. Interpers. Violence 2021 36 3831 3854 10.1177/0886260518777006 29860910 158. Hammond MD Overall NC Benevolent sexism and support of romantic partner’s goals: undermining women’s competence while fulfilling men’s intimacy needs Pers. Soc. Psychol. Bull. 2015 41 1180 1194 10.1177/0146167215593492 26160333 159. Hammond MD Overall NC When relationships do not live up to benevolent ideals: women’s benevolent sexism and sensitivity to relationship problems Eur. J. Soc. Psychol. 2013 43 212 223 10.1002/ejsp.1939 160. Hammond MD Overall NC Endorsing benevolent sexism magnifies willingness to dissolve relationships when facing partner-ideal discrepancies Pers. Relatsh. 2014 21 272 287 10.1111/pere.12031 161. Cross EJ Overall NC Hammond MD Perceiving partners to endorse benevolent sexism attenuates highly anxious women’s negative reactions to conflict Pers. Soc. Psychol. Bull. 2016 42 923 940 10.1177/0146167216647933 27287752 162. Hashim PH Michniewicz K Richmond K Attitudes toward transnational surrogacy, ambivalent sexism, and views on financial allocation Womens Reprod. Health 2022 10.1080/23293691.2022.2062923 163. Li D Zheng L Intimate partner violence and controlling behavior among male same-sex relationships in China: relationship with ambivalent sexism J. Interpers. Violence 2021 36 208 230 10.1177/0886260517724835 29294885 164. Zhao R Zheng Y Child development in same-sex families: beliefs of Chinese lesbians, gays, and heterosexuals Sex. Res. Soc. Policy 2021 18 588 597 10.1007/s13178-020-00483-5 165. Ortiz-Ospira, E. & Tzvetkova, S. Working women: key facts and trends in female labor force participation. Our World In Data https://ourworldindata.org/female-labor-force-participation-key-facts (2017). 166. Lippa RA Preston K Penner J Women’s representation in 60 occupations from 1972 to 2010: more women in high-status jobs, few women in things-oriented jobs PLoS One 2014 9 e95960 10.1371/journal.pone.0095960 24788710 167. Eagly AH Nater C Miller DI Kaufmann M Sczesny S Gender stereotypes have changed: a cross-temporal meta-analysis of U.S. public opinion polls from 1946 to 2018 Am. Psychol. 2020 75 301 315 10.1037/amp0000494 31318237 168. Donnelly K Attitudes toward women’s work and family roles in the United States, 1976–2013 Psychol. Women Q. 2016 40 41 54 10.1177/0361684315590774 169. Hegarty, P., Ansara, Y. G. & Barker, M.-J. in Gender, Sex, And Sexualities: Psychological Perspectives 53–76 (Oxford Univ. Press, 2018). 170. Xiao L Wang F Examining the links between beauty ideals internalization, the objectification of women, and ambivalent sexism among Chinese women: the effects of sexual orientation Arch. Sex. Behav. 2021 50 553 562 10.1007/s10508-020-01718-7 32394113 171. Aktan T Yalçındağ B Çelişik Duygulu Cinsiyetçilik Ölçeğinin (ÇDCÖ) Yeniden Gözden Geçirilmesi: Korumacı Cinsiyetçiliğin Yapı Geçerliliği ve ÇDCÖ’nün Ölçüm Değişmezliği Psikol. Çalışmaları Stud. Psychol. 2022 10.26650/SP2022-820401 172. Trut V Sinovčić P Milavić B Initial validation of the ambivalent sexism inventory in a military setting Soc. Sci. 2022 11 176 10.3390/socsci11040176 173. McMahon JM Kahn KB Benevolent racism? The impact of target race on ambivalent sexism Group Process Intergroup Relat. 2016 19 169 183 10.1177/1368430215583153 174. Blumell LE Rodriguez NS Ambivalent sexism and gay men in the US and UK Sex. Cult. 2020 24 209 229 10.1007/s12119-019-09635-1 175. Charlesworth TES Banaji MR Patterns of implicit and explicit stereotype. III: Long-term change in gender stereotypes Soc. Psychol. Personal. Sci. 2022 13 14 26 10.1177/1948550620988425 176. Kite, M. E., Togans, L. J. & Schultz, T. J. in Cross-Cultural Psychology (ed. Keith, K. D.) 427–448 (Wiley, 2019). 177. Azevedo F Jost JT Rothmund T Sterling J Neoliberal ideology and the justification of inequality in capitalist societies: why social and economic dimensions of ideology are intertwined: neoliberal ideology and justification J. Soc. Issues 2019 75 49 88 10.1111/josi.12310 178. Girerd L Bonnot V Neoliberalism: an ideological barrier to feminist identification and collective action Soc. Justice Res. 2020 33 81 109 10.1007/s11211-020-00347-8 179. Huang Y Osborne D Sibley CG The gradual move toward gender equality: a 7-year latent growth model of ambivalent sexism Soc. Psychol. Personal. Sci. 2019 10 335 344 10.1177/1948550617752472 180. Gomes A Gonçalves G Sousa C Santos J Giger J-C Are we getting less sexist? A ten-year gap comparison analysis of sexism in a Portuguese sample Psychol. Rep. 2022 125 2160 2177 10.1177/00332941211011073 33878957 181. Jones CW Mitchell JS Martin JD Ambivalent sexism? Shifting patterns of gender bias in five Arab countries Int. Stud. Q. 2021 65 277 293 10.1093/isq/sqab007 182. Mukkamala S Suyemoto KL Racialized sexism/sexualized racism: a multimethod study of intersectional experiences of discrimination for Asian American women Asian Am. J. Psychol. 2018 9 32 46 10.1037/aap0000104 183. Buchanan NT Ormerod AJ Racialized sexual harassment in the lives of African American women Women Ther. 2002 25 107 124 10.1300/J015v25n03_08 184. Esqueda CW Harrison LA The influence of gender role stereotypes, the woman’s race, and level of provocation and resistance on domestic violence culpability attributions Sex Roles 2005 53 821 834 10.1007/11199s-005-8295-1 185. Keddie A Disrupting (gendered) Islamophobia: the practice of feminist ijtihad to support the agency of young Muslim women J. Gend. Stud. 2018 27 522 533 10.1080/09589236.2016.1243047 186. Hamzeh, M. Pedagogies Of De-veiling: Muslim Girls And The Hijab Discourse (Information Age Publishing, 2012). 187. Gaunt R Breadwinning moms, caregiving dads: double standard in social judgments of gender norm violators J. Fam. Issues 2013 34 3 24 10.1177/0192513X12438686 188. Bills MA Hayes BE The association between adherence to sexist beliefs and traditional family norms, religion, and attitudes toward sexual minorities J. Homosex. 2022 69 499 524 189. Davies M Correlates of negative attitudes toward gay men: sexism, male role norms, and male sexuality J. Sex. Res. 2004 41 259 266 10.1080/00224490409552233 15497054 190. Sakalli N The relationship between sexism and attitudes toward homosexuality in a sample of Turkish college students J. Homosex. 2002 42 53 64 10.1300/J082v42n03_04 12066992 191. Jordan JA Lawler JR Bosson JK Ambivalent classism: the importance of assessing hostile and benevolent ideologies about poor people Basic Appl. Soc. Psychol. 2021 43 46 67 10.1080/01973533.2020.1828084 192. Cary LA Chasteen AL Remedios J The ambivalent ageism scale: developing and validating a scale to measure benevolent and hostile ageism Gerontologist 2016 57 e27 e36 193. Nario‐Redmond MR Kemerling AA Silverman A Hostile, benevolent, and ambivalent ableism: contemporary manifestations J. Soc. Issues 2019 75 726 756 10.1111/josi.12337 194. Cuddy AJC Stereotype content model across cultures: towards universal similarities and some differences Br. J. Soc. Psychol. 2009 48 1 33 10.1348/014466608X314935 19178758	36504692	PMC9717569	NO-CC CODE	2022-12-06 23:23:25	no	Nat Rev Psychol. 2022 Dec 2;:1-14	utf-8	Nat Rev Psychol	2,022	10.1038/s44159-022-00136-x	oa_other
==== Front Arch Sex Behav Arch Sex Behav Archives of Sexual Behavior 0004-0002 1573-2800 Springer US New York 36459349 2486 10.1007/s10508-022-02486-2 Original Paper Frequency and Combination of Sequential Sexual Acts That May Lead to Sexually Transmitted Infections at Different Anatomic Sites Within the Same Person http://orcid.org/0000-0002-9743-0829 Khosropour Christine M. [email protected] 1 Coomes David M. 1 Barbee Lindley A. 23 1 grid.34477.33 0000000122986657 Department of Epidemiology, University of Washington, 325 Ninth Avenue, Box 359777, Seattle, WA 98104 USA 2 grid.34477.33 0000000122986657 Department of Medicine, University of Washington, Seattle, WA USA 3 grid.238801.0 0000 0001 0435 8972 HIV/STD Program, Public Health – Seattle & King County, Seattle, WA USA 2 12 2022 19 20 6 2022 19 10 2022 16 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Modeling studies suggest that transmission of gonorrhea and chlamydia to multiple anatomic sites within the same person is necessary to reproduce observed high rates of extragenital gonorrhea/chlamydia. Limited empiric behavioral data support this idea. In this cross-sectional study, we enrolled individuals assigned male at birth who reported sex with men (MSM) and denied receptive anal sex (RAS) in the past 2 years. Participants enrolled in-person at the Sexual Health Clinic in Seattle, Washington (December 2019–September 2021) or online (July 2021–September 2021), and completed a sexual history questionnaire that asked about specific sexual acts and sequence of those acts during their last sexual encounter. We enrolled 210 MSM during the 16-month recruiting period. The median number of sex acts reported at last sexual encounter was 4 (interquartile range 3–5). The most commonly reported acts at last sex were: kissing (83%), receiving oral sex (82%), and insertive anal sex (65%). There was substantial variability in the sequence of acts reported; no unique sequence of sex acts was reported by more than 12% of the population. Ninety percent of participants reported sequences of behaviors that could lead to gonorrhea or chlamydia transmission within the same person (respondent or partner); the most common of these combinations was kissing followed by receiving oral sex (64% reporting). Engaging in multiple sex acts within a single sexual encounter is common and may lead to gonorrhea/chlamydia transmission within the same person. This complicates empiric measurements of transmission probabilities needed to estimate population-level transmission. Supplementary Information The online version contains supplementary material available at 10.1007/s10508-022-02486-2. Keywords Oral sex Anal sex Men who have sex with men Infectious diseases transmission Sexually transmitted infections http://dx.doi.org/10.13039/100000060 National Institute of Allergy and Infectious Diseases R21AI142369 Khosropour Christine M. ==== Body pmcIntroduction The pharynx and rectum are common anatomic sites of infection for Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (GC) (Centers for Disease Control & Prevention, 2019; Workowski et al., 2021). Accordingly, there is increasing interest in understanding the potential role of these extragenital sites of infection in sustained population-level CT and GC transmission, particularly in an era with record-high CT and GC rates (Centers for Disease Control & Prevention, 2019). However, transmission of CT and GC is difficult to measure, as individuals may engage in multiple sex acts during a single sexual encounter, which could result in transmission to multiple anatomic sites within the same person and to the other partner(s). Thus, quantifying how often extragenital infections lead to transmission requires a better understanding of the frequency and sequence of behaviors that may lead to CT/GC acquisition and transmission. To date, most studies and surveys of sexual behavior have examined sexual behaviors as single sexual acts during a specific time period (e.g., participant has engaged in receptive oral sex in the past 2 months). These data are helpful to understand the prevalence of behaviors that may lead to STI acquisition and transmission. However, insofar as most individuals do not engage in just one sexual act during a given sexual encounter, these data do not allow us to fully understand CT and GC transmission that may occur between individuals or within the same person. Recently, two mathematical models of men who have sex with men (MSM) that included “sequential sexual practices” (i.e., one sex act following another) found that transmission between multiple anatomic sites within a single sexual encounter are necessary to explain the high rates of CT and GC at more than one anatomic site (Xu et al., 2020, 2021). These models are critically important tools to help us understand the behavioral drivers of transmission and to inform how STI interventions may impact disease transmission. However, current models are lacking in empiric data on the frequency of different combinations of sex acts and on the most common sequence of behaviors. The goals of the present study were to describe the combination and sequence of sex acts during sexual encounters in order to inform studies of GC and CT transmission. Method Participants This is a cross-sectional analysis which is a subset of a larger parent study (“Bottoms Up”) that was designed to examine which behaviors other than receptive anal sex are associated with acquisition of rectal CT/GC. We began recruitment for Bottom’s Up in December 2019; for this analysis, we used data from individuals who were enrolled through September 2021. Participants were eligible to enroll if they were at least 16 years old, were male sex at birth, reported sex with a man in the past 12 months, and did not report receptive anal sex in the past 2 years. These eligibility criteria were developed to be able to answer the parent study’s primary research question. Participants were recruited for the study through two mechanisms, in-person and online. In-person recruitment occurred at the Public Health—Seattle & King County Sexual Health Clinic in Seattle, Washington. We started in-person recruitment in December 2019, stopped recruiting in mid-March 2020 due to COVID-19 restrictions, and began recruiting again in October 2020 through September 2021. Study staff in the clinic approached patients about participating if they appeared to be eligible based on their responses to the clinic’s routine clinical intake form, which queries patients on demographics and sexual history, among other topics. Study staff confirmed eligibility with the patient prior to informed consent. We started online recruitment in July 2021 and continued through September 2021. We recruited individuals from a geospatial social networking app and the social media platforms Facebook and Instagram, as well as third-party apps and website with which Facebook partners. We placed image-based study advertisements and text-based pop-up advertisements on these apps and sites, and we geo-targeted the advertisements to residents of King, Pierce, and Snohomish Counties (the general catchment area of the Sexual Health Clinic) who reported being at least 18 years old. Upon clicking on the survey advertisement, individuals were taken to a survey landing page which described the purpose of the study. Individuals who chose to proceed were asked to complete a brief eligibility screener to ascertain their age, birth sex, residence, gender of sex partners in the past year, and history of receptive anal sex in the past 2 years. Eligible individuals were taken to an electronic consent page. Measures and Procedure There were two components to the study: completion of a survey (described below) and self-collection of a rectal specimen for CT/GC testing. The CT/GC test results are not relevant to the present analysis and are not discussed further. Participants were paid $25 for participating in the study. Study procedures were reviewed and approved by the University of Washington Institutional Review Board (IRB #00007226). Enrolled participants were asked to complete a 10–20-min electronic survey (programmed into REDCap [Harris et al., 2009, 2019]) that queried participants about their sexual behavior history. Prior to implementing the survey, we conducted cognitive interviews with six patients from the Sexual Health Clinic to refine the survey. The goal of the interviews was to ensure the questions would be understood by the study population and were being answered as intended. We refined the survey questions after each interview in an iterative process. The definitions used in the survey are summarized in Table 1.Table 1 Descriptions of sexual behaviors provided to participants within the study survey Behavior Explanation/description in survey Deep kissing This is also called “French kissing” and is when you touch tongues with another person while kissing, and there is saliva exchanged Insertive oral sex (i.e., received oral sex) This is when someone puts their mouth/tongue on your penis (i.e., gives you a blow job) Receptive oral sex (i.e., gave oral sex) This is when you put your mouth/tongue on a man’s penis (i.e., give a man a blow job) Insertive anal sex This is also called topping and is when you put your penis in your partner’s anus (butt) Rimming someone Rimming is when you put your mouth and/or tongue on someone’s anus Being rimmed Being rimmed is when someone puts their mouths and/or tongue on your anus Non-penetrative anal play (perianal play) This is when your partner touches your anus with his penis without full penetration or when your partner’s penis enters your anus a little bit but does not completely penetrate Fingering Being fingered in the anus is when someone puts their fingers on or in your anus. This is different than fisting, where someone puts their entire hand in your anus Fisting Being fisted is when someone puts their hand in your anus Watersports This is when someone urinates (pees) in your butt Felching This is when you suck semen out of someone’s anus or vagina after ejaculation The present analysis focuses on one of the questions in the survey, which queried participants about the acts they engaged in the last time they had sex. The specific question was: “The last time you had sex, which of the following activities did you do? Check all that apply.” The definitions to the behaviors had been previously provided in the survey (summarized in Table 1), but were re-iterated as shortened definitions within the response options. The next question presented a matrix with the behaviors that participants indicated they had engaged in the last time they had sex, and asked participants to indicate the order in which they engaged in the activities. These questions are included as Supplementary material. Statistical Analyses All analyses are descriptive. We report characteristics of the study population (Table 2) and the number and distribution of all sexual acts reported at the last sexual encounter (Table 3). We report the most common combinations of acts, stratified by number of acts during the last sexual encounter (Table 4). For example, for individuals who reported engaging in three acts at the time of their last sexual encounter, we list all three acts reported. In Table 4, we also report the most commons sequence of behaviors reported (far right column of Table 4). For this analysis of sequences, we only included unique behaviors in each sequence; thus, if an individual reported three acts (kissing, insertive oral sex, and insertive anal sex) but reported them in the sequence of kissing followed by insertive oral sex, followed by kissing, followed by insertive anal sex, we did not include the second act of kissing in the sequence and instead considered this to be kissing followed by insertive oral sex followed by insertive anal sex. For Tables 3 and 4, we included all behaviors, regardless of whether or not they could lead to transmission within the same person. In Fig. 1, we report the most common two-act sequences that could lead to GC or CT transmission within the same person (respondent or their partner), regardless of the total number of sex acts reported, in order to directly inform mathematical models of transmission. This analysis addresses the fact that some individuals may report the same sex act multiple times during a single sexual encounter, and allows individuals to be “counted” in the analysis multiple times. For example, if an individual reported kissing, followed by insertive oral sex, followed by kissing, followed by rimming, we considered that individual to have engaged in kissing followed insertive oral sex, insertive oral sex followed by rimming, and kissing followed by rimming.Table 2 Characteristics of study participants, N = 210 Characteristic N (%) Age 16–24 22 (10.5) 25–29 42 (20.0) 30–34 51 (24.3) 35–44 47 (22.4) > 44 48 (22.9) Race/ethnicity American Indian/Alaska native 9 (4.3) Asian 26 (12.4) Black or African-American 39 (18.6) Native Hawaiian, other Pacific islander 3 (1.4) White 120 (57.1) Unknown 13 (6.2) Latinx ethnicity 48 (23.1) Gender Female 1 (0.5) Male 202 (96.7) Non-binary/genderqueer 6 (2.9) Highest level of education High school 66 (31.6) Associate’s degree 22 (10.5) Bachelor’s degree 85 (40.7) Graduate or professional school 36 (17.2) Number of sex partners in past 12 months, median (interquartile range) 4 (2–9) Number of sexual acts reported at last sexual encounter 1 18 (8.5) 2 22 (10.5) 3 55 (26.2) 4 42 (20.0) 5 42 (20.0) 6 19 (9.5) 7 10 (4.8) 8–9 2 (1.0) Table 3 Distribution of sexual behaviors at last sexual encounter, by number of sex acts reported at last encounter, N = 210* Total N = 210 2 acts N = 22 3 acts N = 55 4 acts N = 42 5 acts N = 42 > 5 acts N = 31 Behavior N (%) N (%) N (%) N (%) N (%) N (%) Kissing 175 (83) 12 (55) 50 (91) 40 (95) 41 (98) 30 (97) Insertive oral sex 173 (82) 13 (59) 48 (87) 35 (83) 40 (95) 30 (95) Receptive oral sex 133 (63) 5 (23) 30 (55) 26 (62) 38 (90) 30 (97) Insertive anal sex 137 (65) 11 (50) 27 (49) 31 (74) 35 (83) 30 (97) Perform rimming 82 (39) 2 (9) 3 (5) 17 (40) 32 (76) 28 (90) Received rimming 38 (18) 0 (0) 3 (5) 2 (5) 8 (19) 25 (81) Fingering 25 (12) 0 (0) 3 (5) 5 (12) 5 (12) 12 (39) Fisting 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Perianal play 30 (14) 0 (0) 1 (2) 9 (21) 9 (21) 12 (39) Watersports 1 (0.5) 0 (0) 0 (0) 0 (0) 0 (0) 1 (3) Felching 3 (1.4) 0 (0) 1 (2) 1 (2) 0 (0) 1 (3) *18 individuals reported only one sex act during the last sexual encounter. They are included in the total column but not reported in a separate column Table 4 Frequency of the combination and sequence of sexual acts reported at last sexual encounter, N = 210a Combination and sequence of behaviors Prevalence of combination among the total population (N = 210) Prevalence of combination, among those reporting that number of acts Prevalence of combination in the sequence specified, among those reporting that combinationd N (%) n/N (%)b n/N (%)c Two acts (n = 22) IOS–IAS 8 (3.8%) 8/22 (36.3%) 7/8 (88%) K–ROS 5 (2.4%) 5/22 (22.7%) 3/5 (60%) K–IOS 4 (1.9%) 4/22 (18.2%) 4/4 (100%) Three acts (n = 55) K–IOS–ROS 24 (11.4%) 24/55 (43.6%) 19/24 (79%) K–IOS–IAS 18 (8.6%) 18/55 (32.7%) 14/18 (78%) K–ROS–IAS 3 (1.4%) 3/55 (5.5%) 2/3 (67%) Four acts (n = 42) K–ROS–IOS–IAS 10 (4.8%) 10/42 (23.8%) 6/10 (60%) K–IOS–R–IAS 10 (4.8%) 10/42 (23.8%) 7/10 (70%) K–ROS–R–IAS 3 (1.4%) 3/42 (7.1%) 3/3 (100%) K–IOS–ROS–PP 3 (1.4%) 3/42 (7.1%) 2/3 (67%) Five acts (n = 42) K–IOS–ROS–R–IAS 23 (10.9%) 23/42 (54.8%) 17/23 (74%) K–IOS–ROS–BR—IAS 4 (1.9%) 4/42 (9.5%) 3/4 (75%) Six acts (n = 19) K–IOS–ROS–R–BR–IAS 13 (6.2%) 13/19 (68.4%) 8/13 (62%) K–IOS–ROS–R–F–IAS 3 (1.4%) 3/19 (15.7%) 1/3 (33%) K kissing, IOS insertive oral sex (received oral sex), ROS receptive oral sex (gave oral sex), IAS insertive anal sex, R rimming someone, BR being rimmed, PP perianal play aThe combinations listed represent those where at least 5% of individuals in each category report that combination (categories being “two acts”, “three acts”, etc.) bn = number who reported that combination of sex acts; N = number who reported that number of acts cn = number who reported that sequence in the order specified; N = number who reported that combination of sex acts dOf the 84 individuals who report sequences in the order specified (the sum of the numerators for those reporting at least 3 acts), 14 (17%) reported a behavior that was repeated in the sequence after first identification. For 10 of the 14, this behavior was kissing; for 4, it was kissing plus another behavior Fig. 1 Prevalence of sequences of sex acts that could lead to GC/CT transmission within the same person (respondent or partner). Percentages calculated as the number of respondents who reported that sequence of sex acts out of the total population of 210. Panel A: Kissing followed by insertive oral sex; Panel B: Kissing followed by receptive oral sex; Panel C: Insertive oral sex followed by being rimmed; Panel D: Receptive oral sex followed by performing rimming; Panel E: Insertive oral sex followed by insertive anal sex; Panel F: Kissing following by being rimmed; Panel G: Kissing followed by performing rimming; Panel H: Performing rimming followed by insertive anal sex Results During the 16-month recruiting period, we enrolled a total of 210 participants, including 177 participants in-person and 33 participants online. About 30% of participants were less than 30 years old, nearly 20% self-reported their race as Black or African-American, and 97% identified as male gender (Table 2). Participants reported a median 4 sex partners in the past year. The vast majority of participants (> 90%) reported more than one sex act during their last sexual encounter; the median number of sex acts reported was 4 (Table 2). The distribution of sexual behaviors at last sexual encounter is displayed in Table 3. Most participants reported kissing (83%), insertive oral sex (82%), receptive oral sex (63%), or insertive anal sex (65%) during their last sexual encounter. Rimming (receipt and giving) were the next most frequently reported behaviors, and were more often reported by individuals who engaged in at least three sex acts. Watersports and felching were not frequently reported, and fisting was not reported by any respondents. There was substantial variability in the combination of behaviors reported (Table 4), with no single combination being reported by more than 12% of the study population. The two most common combinations of behaviors reported were: (1) kissing, insertive oral sex, and receptive oral sex, which was reported by 24 (11.4%) of 210 participants, of whom 79% reported those behaviors in that sequence; and (2) kissing, insertive oral sex, receptive oral sex, performing rimming, and insertive anal sex, which was reported by 23 (10.9%) of 210 participants, of whom 74% reported those behaviors in that sequence. As described in the Method, we only included unique behaviors in each sequence (i.e., if someone reported a behavior twice, we only included it in the sequence the first time it was reported). Of the 84 individuals who reported sequences of acts in the order specified in Table 4 (sum of the numerators in the far-right column of Table 4 for those reporting at least 3 acts), 17% (n = 14) of individuals reported a behavior that was repeated in the sequence after first identification; for 10 of these 14, the behavior was kissing. For four, it was kissing plus another behavior. The order of the placement of the repeated behavior for these 14 individuals was unique to each person. Overall, 189 (90%) of participants reported a sequence of sex acts at their last sexual encounter that could lead to GC or CT transmission within the same person (respondent or partner) (Fig. 1). The most commonly reported sequence of these behaviors was kissing followed by insertive oral sex (Fig. 1, Panel A), which was reported by 135 (64%) of 210 respondents (theoretical CT/GC transmission route: respondent’s pharynx to partner’s pharynx to respondent’s urethra). Kissing followed by receptive oral sex (Fig. 1, Panel B; theoretical transmission route: partner’s pharynx to respondent’s pharynx to partner’s urethra) and insertive oral sex followed by insertive anal sex (Fig. 1, Panel E; theoretical transmission route: partner’s pharynx to respondent’s urethra to partner’s anus) were also commonly reported sequences, reported by 57% and 49% of the total population, respectively. Discussion In this cross-sectional study, we found that over 50% of participants reported at least four sex acts during their last sexual encounter, with the most commonly reported acts being kissing, insertive oral sex, receptive oral sex, and insertive anal sex. There was considerable variability in the combination of behaviors being reported; no single combination was reported by more than 12% of the population. Sequences of behaviors that could lead to GC or CT transmission within the same person were reported by 90% of study participants. Our findings lend support to recent modeling studies that have suggested that sequential sexual practices are necessary to replicate observed multi-site infections among MSM. There have been a number of large studies—including population-based studies—that have measured the prevalence of sexual behaviors among MSM in the USA (Centers for Disease Control and Prevention, 2021; Sanchez et al., 2016; Wiatrek et al., 2021). However, most studies have asked about engagement in behaviors during a period of time, and not about combinations of behaviors or sequences of acts within a single sexual encounter. To our knowledge, the largest US study that has examined sex acts within a sexual encounter was a 2011 internet-based study by Rosenberger et al. (2011) that enrolled nearly 25,000 respondents. In that study, the majority of respondents reported between 5 and 9 sex acts at the last sexual encounter, and most men reported kissing (75%), insertive oral sex (73%), and receptive oral sex (75%). Likewise, we noted that the majority of participants reported at least four behaviors, and that kissing, insertive oral sex, and receptive oral sex were also the most commonly reported behaviors at 83%, 80%, and 63%, respectively. Notably, our study population—by design—excluded individuals who reported receptive anal sex in the past 2 years, whereas the Rosenberger et al. study did not. Despite these studies enrolling populations with different sexual behavior histories, the prevalence of reported behaviors was remarkably consistent across the two studies. Our study builds upon the findings of Rosenberger et al. by additionally providing information on the sequence of reported behaviors, which is more informative for measuring transmission than the combination alone. The results of our study highlight the complexities of empirically measuring per-act CT and GC transmission probabilities, which are essential parameters to be able to understand disease transmission (Spicknall et al., 2019). We observed substantial variability in the combination of sex acts reported, as well as variability in the sequence in which acts were reported. This variability in combination and sequences creates a large number of potential transmission pathways at the anatomic sites of the pharynx, rectum, and urethra between partners and within a single person. Notably, our analysis of sequences of behaviors was a simplified one, in that we only considered reported acts one time per sequence. For example, if someone reported kissing, receptive oral sex, kissing, and insertive oral sex, we did not include the second kissing act in the sequence. We note that 17% of individuals (n = 14) who reported behaviors in the sequences outlined in Table 4 did report the same act multiple times in the sequence; each of these 14 individuals had unique sequences of behaviors. Our need to simplify this analysis highlights the complexities in measuring potential GC and CT transmission routes. However, we did attempt to address this limitation of our analysis by also examining the most commonly reported two-act sequences that could lead to GC or CT transmission within the same person, allowing for multiple two-act sequences of behaviors to be counted for a single respondent. We found that 90% of individuals reported at least one sequence of behaviors that could lead to GC or CT transmission within the same person. The commonness of these sequences may help explain the relatively high prevalence of pharyngeal and rectal GC and CT observed in many sexual health clinic settings and highlights the need for and importance of extragenital GC and CT screening. Our findings can be used to inform parameter estimation for mathematical models designed to estimate disease transmission. There have been a number of recent STI models that incorporate anatomic-specific STIs (Fairley et al., 2019; Jenness et al., 2017; Spicknall et al., 2019; Xiridou et al., 2013; Zhang et al., 2017), but until recently none had included sequential practices. In novel models of multi-site infection of GC and CT, Xu and colleagues found that the inclusion of sequential sexual practices is necessary to replicate observed GC and CT site-specific prevalences (Xu et al., 2020, 2021). These new models demonstrated the importance of including sequential sexual practices to accurately model disease transmission; however, the prevalence of sequential behaviors used in the models were estimates. Our findings fill this gap by providing empiric data to inform these behavioral parameters. Given the complexities in directly measuring GC and CT transmission, coupled with recent calls for more data to understand the benefits and harms of screening among MSM (US Preventive Services Task Force et al., 2021)—including the role of screening in the development of antimicrobial resistance (Dijck et al., 2020; Kenyon, 2020; Kenyon et al., 2020a, 2020b)—developing robust mathematical models may be our best method to understand how interventions (e.g., screening, vaccines) may alter disease transmission, sequelae, and antimicrobial resistance(Craig et al., 2015; Gray et al., 2009; Spicknall et al., 2019). Our analysis also highlights the dearth of empiric data on combinations of sexual acts and the sequence of these acts among heterosexual individuals. The same sequences of behaviors we describe in this analysis are also relevant for individuals who are female sex at birth, a population at risk of adverse reproductive health outcomes. For example, a female partner who has pharyngeal GC and performs oral sex on their male partner could transmit GC from their pharynx to their partner’s urethra; and if the partners subsequently engage in vaginal sex, the male partner could then transmit GC from their urethra to their partner’s vagina. To date, mathematical models of heterosexual GC and CT transmission have not included sequential sexual behaviors (Althaus et al., 2010, 2012; Lewis & White, 2018). Doing so may more accurately predict the role of extragenital screening on GC and CT transmission and health outcomes within heterosexual populations. Our study is strengthened by recruitment from in-person and online venues, and by our use of a survey tool that was developed after extensive testing and interviewing with the study population. This study is also subject to important limitations. The participants we enrolled were individuals who attended our Sexual Health Clinic or clicked on an online advertisement and had not had receptive anal sex in the past 2 years. Thus, their reported behaviors may not be generalizable to populations who do report anal sex or to other MSM populations more broadly. However, the consistency of our findings with the largest study to date on this topic is reassuring that the behaviors of our study population may be at least somewhat representative. Second, we made analytic decisions to simplify the reporting of the prevalences of behaviors. For example, participants could indicate that they engaged in behavior more than one time during the course of the encounter, but for the purposes of this analysis we only reported the behavior in the sequence the first time it was reported. This affected 17% of reported sequences. We also split up sequences into two-act sequences (Fig. 1), acknowledging that this does not allow one to fully appreciate all anatomic sites that could be exposed during a single act. However, it does follow how modeling studies have parameterized sequential sex acts (Xu et al., 2020, 2021). Third, we included a set list of behaviors from which participants could choose, and this was not an exhaustive list of all behaviors. For example, we did not include mutual masturbation or ask about use of saliva as a lubricant (Cornelisse et al., 2018; Fairley et al., 2019). Instead, we focused on behaviors that have the potential to transmit GC or CT directly to the pharynx, urethra, or rectum. Fourth, we focused only on participants’ last sexual encounter in an attempt to diminish recall bias. It is possible that the most recent sexual encounter is not representative of an individual’s typical sexual behaviors, particularly for a population attending a sexual health clinic where their most recent encounter may have prompted the visit to the clinic. This may also overrepresent behaviors with participants’ regular partners, but we did not specifically ask participants any details about their last sexual partner. Fifth, although we asked for data from the last sexual encounter, the results are still subject to recall bias, insofar as participants were unable to accurately recall their most recent sexual encounter. Finally, the two questions included in this analysis came at the end of a larger behavioral survey and it is possible that participants may have experienced survey fatigue which could have led to inaccuracies in reporting. In summary, we found that the variability in the combination and sequence of sexual behaviors during sexual encounters complicates empiric measurement of per-act transmission probabilities of GC and CT. Gaining a better understanding of GC and CT transmission from all anatomic sites is essential in order to fully estimate population-level GC and CT transmission, and the prevalence estimates from our study can be used in mathematical models aimed to understanding disease transmission. We hope our study inspires more specific and detailed collection of sexual behavior data to fully capture sequential events. However, even well-designed prospective studies may not be able to directly estimate per-act transmission probabilities, as multiple behaviors could contribute to GC or CT acquisition at a single anatomic site. In the future, controlled human infection models (Hobbs & Duncan, 2019) could be a useful tool to be able to measure disease transmission in a controlled environment. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 38 kb) Acknowledgements We would like to thank our study participants and the Public Health—Seattle & King County Sexual Health Clinic for donating study space. We thank Seila Vorn, Farchung Saechao, and Angela LeClair for recruiting and coordinating with study participants and thank Tanya Avoundjian for creating the figure. Author Contributions CK and LB led the study conception and design and data collection of the in-person study. DMC led the design and data collection of the online arm of the study. Data analyses were performed by CK. The first draft of the manuscript was written by CK, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding This work was supported by the National Institutes of Health (NIH) [Grant R21AI142369 to C.M.K. and L.A.B.] Data Availability A complete de-identified dataset and code will be made available to interested parties by request, at the completion of the study (anticipated end date May 2023). Declarations Conflict of interest C.M.K. and L.A.B. report research support from Hologic, Inc. L.A.B. has received research support from SpeeDX outside of the submitted work and has received research support and consulting fees from Nabriva, unrelated to the submitted work. D.M.C. has no relevant financial or non-financial interests to disclose. Consent to Participate Informed consent was obtained from all individuals included in the study. Ethical Standard Approval for this study was obtained from the ethics committee (Institutional Review Board) at the University of Washington. The procedures used in the study adhere to the tenets of the Declaration of Helsinki. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Althaus CL Heijne JC Low N Towards more robust estimates of the transmissibility of Chlamydia trachomatis Sexually Transmitted Diseases 2012 39 5 402 404 10.1097/OLQ.0b013e318248a550 22504608 Althaus CL Heijne JCM Roellin A Low N Transmission dynamics of Chlamydia trachomatis affect the impact of screening programmes Epidemics 2010 2 3 123 131 10.1016/j.epidem.2010.04.002 21352783 Centers for Disease Control & Prevention. (2019). Sexually Transmitted Disease Surveillance 2018. US Department of Health and Human Services. https://www.cdc.gov/std/stats18/STDSurveillance2018-full-report.pdf Centers for Disease Control & Prevention. (2021). National HIV Behavioral Surveillance (NHBS). https://www.cdc.gov/hiv/statistics/systems/nhbs/index.html Cornelisse VJ Fairley CK Read TRH Lee D Walker S Hocking JS Chen MY Bradshaw CS Chow EPF Associations between anorectal chlamydia and oroanal sex or saliva use as a lubricant for anal sex: A cross-sectional survey Sexually Transmitted Diseases 2018 45 8 506 510 10.1097/OLQ.0000000000000800 29465648 Craig AP Gray RT Edwards JL Apicella MA Jennings MP Wilson DP Seib KL The potential impact of vaccination on the prevalence of gonorrhea Vaccine 2015 33 36 4520 4525 10.1016/j.vaccine.2015.07.015 26192351 Fairley CK Cornelisse VJ Hocking JS Chow EPF Models of gonorrhoea transmission from the mouth and saliva The Lancet Infectious Diseases 2019 19 10 e360 e366 10.1016/S1473-3099(19)30304-4 31324517 Gray RT Beagley KW Timms P Wilson DP Modeling the impact of potential vaccines on epidemics of sexually transmitted Chlamydia trachomatis infection Journal of Infectious Diseases 2009 199 11 1680 1688 10.1086/598983 19432550 Harris PA Taylor R Minor BL Elliott V Fernandez M O’Neal L McLeod L Delacqua G Delacqua F Kirby J Duda SN Consortium RE The REDCap consortium: Building an international community of software platform partners Journal of Biomedical Informatics 2019 95 103208 10.1016/j.jbi.2019.103208 31078660 Harris PA Taylor R Thielke R Payne J Gonzalez N Conde JG Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support Journal of Biomedical Informatics 2009 42 2 377 381 10.1016/j.jbi.2008.08.010 18929686 Hobbs MM Duncan JA Experimental human infection with Neisseria gonorrhoeae Methods in Molecular Biology 2019 1997 431 452 10.1007/978-1-4939-9496-0_25 31119638 Jenness SM Weiss KM Goodreau SM Gift T Chesson H Hoover KW Smith DK Liu AY Sullivan PS Rosenberg ES Incidence of gonorrhea and chlamydia following human immunodeficiency virus preexposure prophylaxis among men who have sex with men: A modeling study Clinical Infectious Diseases 2017 65 5 712 718 10.1093/cid/cix439 28505240 Kenyon C Does intense sexually transmitted infection screening cause or prevent antimicrobial resistance in sexually transmitted infections? it depends on one's underlying epistemology A Viewpoint. Sexually Transmitted Diseases 2020 47 8 506 510 10.1097/OLQ.0000000000001199 32520879 Kenyon C Baetselier I Wouters K Screening for STIs in PrEP cohorts results in high levels of antimicrobial consumption International Journal of STD and AIDS 2020 31 12 1215 1218 10.1177/0956462420957519 32903136 Kenyon C Buyze J Spiteri G Cole MJ Unemo M Population-level antimicrobial consumption is associated with decreased antimicrobial susceptibility in Neisseria gonorrhoeae in 24 European countries: An ecological analysis Journal of Infectious Diseases 2020 221 7 1107 1116 30957153 Lewis J White PJ Changes in chlamydia prevalence and duration of infection estimated from testing and diagnosis rates in England: A model-based analysis using surveillance data, 2000–15 Lancet Public Health 2018 3 6 e271 e278 10.1016/S2468-2667(18)30071-9 29776798 Rosenberger JG Reece M Schick V Herbenick D Novak DS Van Der Pol B Fortenberry JD Sexual behaviors and situational characteristics of most recent male-partnered sexual event among gay and bisexually identified men in the United States Journal of Sexual Medicine 2011 8 11 3040 3050 10.1111/j.1743-6109.2011.02438.x 21883941 Sanchez T Zlotorzynska M Sineath C Kahle E Sullivan P The annual American men's internet survey of behaviors of men who have sex with men in the United States: 2014 key indicators report JMIR Public Health and Surveillance 2016 2 1 e23 10.2196/publichealth.5476 27244770 Spicknall IH Mayer KH Aral SO Romero-Severson EO Assessing uncertainty in an anatomical site-specific gonorrhea transmission model of men who have sex with men Sexually Transmitted Diseases 2019 46 5 321 328 10.1097/OLQ.0000000000000953 30516722 US Preventive Services Task ForceDavidson KW Barry MJ Mangione CM Cabana M Caughey AB Davis EM Donahue KE Doubeni CA Krist AH Kubik M Li L Ogedegbe G Pbert L Silverstein M Simon MA Stevermer J Tseng CW Wong JB Screening for chlamydia and gonorrhea: US Preventive Services Task Force Recommendation Statement Journal of the American Medical Association 2021 326 10 949 956 10.1001/jama.2021.14081 34519796 Van Dijck C Laumen J Zlotorzynska M Manoharan-Basil SS Kenyon C Association between STI screening intensity in men who have sex with men and gonococcal susceptibility in 21 States in the USA: An ecological study Sexually Transmitted Infections. 2020 96 7 537 540 10.1136/sextrans-2019-054313 32066589 Wiatrek S Zlotorzynska M Rai R Sullivan P Sanchez T The annual American Men’s Internet Survey of Behaviors of Men Who Have Sex with Men in the United States: Key Indicators Report 2018 JMIR Public Health and Surveillance 2021 7 3 e21812 10.2196/21812 33496669 Workowski KA Bachmann LH Chan PA Johnston CM Muzny CA Park I Reno H Zenilman JM Bolan GA Sexually transmitted infections treatment guidelines, 2021 MMWR–Recommendations and Reports 2021 70 4 1 187 10.15585/mmwr.rr7004a1 Xiridou M Vriend HJ Lugner AK Wallinga J Fennema JS Prins JM Geerlings SE Rijnders BJ Prins M de Vries HJ Postma MJ van Veen MG Schim van der Loeff MF van der Sande MA Modelling the impact of chlamydia screening on the transmission of HIV among men who have sex with men BMC Infectious Diseases 2013 13 436 10.1186/1471-2334-13-436 24047261 Xu X Chow EPF Ong JJ Hoebe CJPA Zou Z Hocking JS Fairley CK Zhang L Chlamydia trachomatis transmission between the oropharynx, urethra and anorectum in men who have sex with men: A mathematical model BMC Medicine 2020 18 1 326 10.1136/sextrans-2020-054565 33198750 Xu X Chow EPF Ong JJ Hoebe CJPA Williamson D Shen M Kong FYS Hocking JS Fairley CK Zhang L Modelling the contribution that different sexual practices involving the oropharynx and saliva have on Neisseria gonorrhoeae infections at multiple anatomical sites in men who have sex with men Sexually Transmitted Infections 2021 97 3 183 189 10.1136/sextrans-2020-054565 33208511 Zhang L Regan DG Chow EPF Gambhir M Cornelisse V Grulich A Ong J Lewis DA Hocking JS Fairley CK Neisseria gonorrhoeae transmission among men who have sex with men: An anatomical site-specific mathematical model evaluating the potential preventive impact of mouthwash Sexually Transmitted Diseases 2017 44 10 586 592 10.1097/OLQ.0000000000000661 28876289	36459349	PMC9717570	NO-CC CODE	2022-12-06 23:23:26	no	Arch Sex Behav. 2022 Dec 2;:1-9	utf-8	Arch Sex Behav	2,022	10.1007/s10508-022-02486-2	oa_other
==== Front Z Rheumatol Z Rheumatol Zeitschrift Fur Rheumatologie 0340-1855 1435-1250 Springer Medizin Heidelberg 36459172 1295 10.1007/s00393-022-01295-1 Originalien Effect of cervical stabilization exercises on cervical position error in patients with axial spondyloarthritis: a randomized controlled pilot study Wirkung von Übungen zur Halsstabilisierung auf zervikale Positionsfehler der Halswirbelsäule bei Patienten mit axialer Spondyloarthritis: randomisierte kontrollierte PilotstudieOz Hande Ece 1 Duran Gozde 2 http://orcid.org/0000-0002-2852-8910 Bayraktar Deniz [email protected] 3 Kara Mete 1 Solmaz Dilek 1 Akar Servet 1 1 grid.411795.f 0000 0004 0454 9420 Division of Rheumatology, Department of Internal Medicine, Faculty of Medicine, Izmir Katip Celebi University, Izmir, Turkey 2 grid.21200.31 0000 0001 2183 9022 Institute of Health Sciences, Dokuz Eylul University, Izmir, Turkey 3 grid.411795.f 0000 0004 0454 9420 Department of Physiotherapy and Rehabilitation, Faculty of Health Sciences, Izmir Katip Celebi University, Izmir, Turkey Redaktion Ulf Müller-Ladner, Bad Nauheim Uwe Lange, Bad Nauheim 2 12 2022 17 31 10 2022 © The Author(s), under exclusive licence to Springer Medizin Verlag GmbH, ein Teil von Springer Nature 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Objective To investigate the effect of cervical stabilization exercises on cervical position error in patients with axial spondyloarthritis (axSpA). Materials and methods Thirty-nine patients with axSpA were randomly allocated to two groups as exercise group (n = 20, 11 males) and control group (n = 19, 12 males). The exercise group performed a progressive home-based cervical stabilization exercise program, while the control group did not receive any exercise intervention. To control exercise adherence and progression, text messages and video instructions were delivered via a freeware and cross-platform messaging service on a weekly basis. All patients were evaluated regarding physical characteristics, disease activity (Bath Ankylosing Spondylitis Disease Activity Index), functional status (Bath Ankylosing Spondylitis Functional Index), and spinal mobility (Bath Ankylosing Spondylitis Metrology Index). Cervical position error was evaluated in flexion, extension, rotation, and lateral flexion directions. All evaluations were performed at baseline and after 6 weeks. Results Baseline physical and disease-related characteristics were similar between the groups (p > 0.05). After 6 weeks, significant improvements were observed in cervical position error in all directions in the exercise group (p < 0.05), whereas no improvements were detected in the control group (p > 0.05). Conclusion A 6-week home-based cervical stabilization exercise program seems to be beneficial for improving impaired cervical proprioception in patients with axSpA. Ziel Ziel der Arbeit war es, die Wirkung von Übungen zur Stabilisierung der Halswirbelsäule auf deren zervikaler Positionsfehler bei Patienten mit axialer Spondyloarthritis (axSpA) zu untersuchen. Material und Methoden Dazu wurden 39 Patienten mit axSpA randomisiert in 2 Gruppen eingeteilt, eine Übungsgruppe (n = 20, 11 m.) und eine Kontrollgruppe (n = 19, 12 m.). Die Übungsgruppe führte ein stufenweise aufbauendes Heimprogramm zur Halswirbelsäulenstabilisierung durch, während die Kontrollgruppe keine Übungsintervention erhielt. Um die Einhaltung der Übungen und ihren Fortschritt zu kontrollieren, wurden wöchentlich Textnachrichten und Videoanleitungen über einen freien und plattformübergreifenden Messagerdienst gesendet. Sämtliche Patienten wurden hinsichtlich körperlicher Merkmale, Krankheitsaktivität (Bath Ankylosing Spondylitis Disease Activity Index), funktionellem Status (Bath Ankylosing Spondylitis Functional Index), und Wirbelsäulenbeweglichkeit (Bath Ankylosing Spondylitis Metrology Index) untersucht. Zervikale Positionsfehler der Halswirbelsäule wurden in Flexion, Extension, Rotation und Lateralflexion beurteilt. Alle Untersuchungen wurden zu Anfang der Studie und nach 6 Wochen durchgeführt. Ergebnisse Die körperlichen und krankheitsbezogenen Merkmale zu Anfang waren zwischen den beiden Gruppen ähnlich (p > 0,05). Nach 6 Wochen wurden in der Übungsgruppe signifikante Verbesserungen der zervikalen Positionsfehler der Halswirbelsäule in allen Richtungen festgestellt (p < 0,05), während in der Kontrollgruppe keine Verbesserungen zu beobachten waren (p > 0,05). Schlussfolgerung Ein 6‑wöchiges häusliches Übungsprogramm zur Stabilisierung der Halswirbelsäule scheint vorteilhaft zu sein, um die beeinträchtigte zervikale Propriozeption bei Patienten mit axSpA zu verbessern. Keywords Spondyloarthritis Exercise Proprioception Neck Functional status Schlüsselwörter Spondyloarthritis Bewegungsübungen Propriozeption Hals Funktioneller Status ==== Body pmcIntroduction Axial spondyloarthritis (axSpA) is a chronic inflammatory disease which is classified into two subtypes according to radiographic findings as radiographic axSpA (ankylosing spondylitis) and non-radiographic axSpA [1]. Spinal involvement is a dominant feature of the disease, and the sacroiliac joints along with lumbar and cervical regions are frequently involved. It is well known that functional status, muscle functions, quality of life, and emotional status may be diminished in patients with axSpA [2–4]. Recently, our group showed deterioration in the sense of cervical joint position in patients with axSpA [5]. Cervical joint position accuracy is important to maintain optimal head and neck position which is crucial for most activities of daily living. A sense of joint position along with a sense of movement and sense of force are the components of proprioception, which represents the awareness of the body in the space [6]. Sense of joint position is provided by receptors in tendons, ligaments, joint capsules, and joint surfaces [7]. In inflammatory arthritis, chronic inflammation might damage joint surfaces which contribute to the sense of joint position. On the other hand, proprioceptive acuity is also affected by muscle activity [8]. Knoop et al. reported that muscle weakness along with damage in the mechanoreceptors contributes to a decreased sense of position in patients with arthritis [9]. Thus, improving muscle function may help to enhance proprioceptive function. Jull et al. showed that conventional proprioceptive and craniocervical flexion exercises improved cervical joint position sense in patients with neck pain [10]. Later, they reported that craniocervical exercises improve proprioception by activating the deep flexor muscles and restoring the coordination between deep and superficial neck muscles [11]. Similarly, cervical stabilization exercises primarily aim to activate deep cervical muscles. Thus, utilizing these exercises may help to improve proprioception sense. Moreover, Lee et al. showed that performing cervical stabilization exercises leads to improvement in the accuracy of joint position sense in healthy participants [12]. However, to the best of our knowledge, there are no previous studies investigating the effects of cervical stabilization exercises on proprioception in patients with axSpA. Therefore, in the present pilot study, we aimed to investigate the effects of cervical stabilization exercises on cervical joint position error in patients with axSpA. Methods This was a prospective study performed between September and December 2019 in Izmir Katip Celebi University, Ataturk Research and Training Hospital. The study was approved by the Izmir Katip Celebi University Ethical Committee with approval number of 325/2019. The study was registered in ClinicalTrials.gov (NCT04483648). Patients Fifty patients with axSpA according to the Assessment of Spondyloarthritis International Society (ASAS) criteria were invited to join the study [1]. Seven patients declined to participate. Patients who had neck trauma history, complete spinal ankylosis, orthopedic disability, cervical spinal surgery, and/or vestibular system problems were excluded. The flowchart of the study is given in Fig. 1. All patients gave informed consent prior to participation in the study.Fig. 1 Flowchart of the study Procedures Every participant completed a structured form containing demographic data (age, sex, height, weight), disease activity (Bath Ankylosing Spondylitis Disease Activity Index) [13], and functional status (Bath Ankylosing Spondylitis Functional Index) [14]. Spinal mobility (Bath Ankylosing Spondylitis Metrology Index) [15] was assessed by a rheumatologist and the cervical joint position error (JPE) was evaluated by a physiotherapist. Patients were then randomly allocated to two groups as exercise and control using opaque envelopes. The patients in the exercise group performed a 6-week progressive home-based cervical stabilization exercise program along with their regular treatment, while the patients in the control group maintained their regular treatment only. Cervical JPE was reassessed after 6 weeks. Outcome measures Cervical joint position error The neutral head position method, which shows high reliability and accuracy, was used to evaluate cervical JPE [16]. The patients were seated upright with a straight head position on a standard chair with a back support at 100 cm from the measurement wall. A piece of 1‑mm-squared graphical drawing paper (width 100 cm, height 100 cm) was placed on the wall. Then a plastic helmet with a laser pointer inserted at the top was placed on the head of the patient [5]. The method was also recently employed by our group to determine the status of cervical JPE in patients with axSpA [5]. A soft opaque fabric eye mask was placed over their eyes, and they were asked to memorize their neutral head position. The patients were then instructed to move their head into flexion, extension, right and left lateral flexion, and right and left rotation. Initially, a reference point was determined for each direction and each movement was repeated three times. Ten seconds of rest were provided between attempts for the same direction and 1 min rest was allowed between different directions. The absolute JPE was calculated using the formula angle = tan − 1 (error distance/90 cm) for each trial, and the average of three measurements was used in the analysis for each direction for each participant. All JPE assessments were performed by the same investigator. Interventions Cervical stabilization exercises The patients in the cervical stabilization exercise group performed progressive home-based cervical stabilization exercises three times a week for 6 weeks. The unilateral/bilateral upper and lower extremity movements with/without an external loads were performed in four different positions: lying in supine, lying in prone, in quadrupedal, and in standing. Chin-tuck position was maintained during each movement. The exercise program was progressed by increasing isometric contraction times of the chin-tuck exercises and the repetitions of the movements (Table 1; Fig. 2). Each week, a short video message including exercises along with the instructions was delivered to patients via a freeware and cross-platform messaging service (WhatsApp messenger; Meta, Menlo Park, CA, USA) in order to control exercise adherence and to inform about exercise progression. An exercise session lasted approximately 20 min.Table 1 Progression of the cervical stabilization exercises Week Movements Chin-tuck time Repeats (for each position) 1–2 Maintaining chin-tuck position in supine, prone, quadrupedal, and standing positions 10 s 10 times 3 Maintaining chin-tuck position plus unilateral and bilateral upper and reciprocal lower extremity movements in supine, prone, quadrupedal, and standing positions 5 s 5 times 4 Maintaining chin-tuck position plus unilateral and bilateral upper extremity with an external load of 500 ml full water bottle and reciprocal lower extremity movements in supine, prone, quadrupedal, and standing positions 5 s 12 times 5 Maintaining chin-tuck position plus unilateral and bilateral upper extremity with an external load of 500 ml full water bottle and reciprocal lower extremity movements in supine, prone, quadrupedal, and standing positions 10 s 12 times 6 Maintaining chin-tuck position plus unilateral and bilateral upper extremity with an external load of 500 ml full water bottle and reciprocal lower extremity movements in supine, prone, quadrupedal, and standing positions 15 s 12 times Fig. 2 Cervical stabilization exercise samples (red arrows to this direction) Statistical analysis Statistical Package for the Social Sciences software (SPSS) version 18.0 (IBM®, Armonk, NY, USA) was used for statistical analyses. The normality of the data was investigated using the Shapiro–Wilk test and histograms. Non-parametric analyses were used due to the heterogeneity of the data. Thus, continuous variables were expressed as median (interquartile range) and categorical variables were presented as percentages. Mann–Whitney U test and chi-square tests were used to compare the groups. Wilcoxon signed-rank test was used to compare in-group changes. A p-value of < 0.05 was accepted as the statistical significance level. Results In total, 39 patients with axSpA completed the study (exercise group: n = 20, 11 males; control group: n = 19, 12 males). Fifteen (75%) and 16 patients (84%) were using anti-tumor necrosis factor (TNF) agents in the exercise group and in the control group, respectively. No changes in patients’ medication were performed during the study. Thirteen (65%) and 14 patients (74%) were radiographic axSpA in the exercise group and in the control group (p = 0.557), respectively. Ten patients had cervical syndesmophytes in the exercise group, and 5 patients has syndesmophytes in the control group (p = 0.049). The demographic, physical, and disease-related characteristics were comparable between groups (Table 2). No patients reported adverse effects related to the exercise program.Table 2 Physical and disease-related characteristics Exercise group (n = 20) Control group (n = 19) P-value Median (IQR 25/75) Median (IQR 25/75) Physical characteristics Age (years) 40.5 (36.0/52.5) 44.0 (39.0/49.5) 0.496a Height (cm) 168.5 (160.0/176.5) 175.0 (168.5/178.5) 0.214a Weight (kg) 78.5 (67.0/91.5) 84.0 (70.0/90.0) 0.687a Body mass index (kg/m2) 27.5 (24.5/30.2) 26.8 (23.6/29.3) 0.569a Male sex (n, %) 11 (55%) 12 (63%) 0.605b Disease-related characteristics Time since onset of symptoms (years) 14.5 (9.0/20) 11 (7.0/14.0) 0.403a Time since diagnosis (years) 9.0 (4.0/11.5) 6.0 (3.0/7.0) 0.113a BASDAI (score) 2.0 (1.0/3.3) 1.8 (1.3/2.5) 0.687a BASMI tragus–wall distance (score) 2.0 (1.0/3.2) 2.0 (1.0/3.2) 0.945a BASMI cervical rotation (score) 2.0 (1.6/3.1) 2.0 (1.4/3.2) 0.708a BASMI total (score) 2.9 (1.7/4.1) 2.3 (1.8/3.1) 0.127a BASFI (score) 1.8 (0.6/2.9) 1.2 (1.0/2.2) 0.496a Cervical proprioception error Flexion (o) 4.9 (2.2/7.0) 6.3 (3.5/7.3) 0.444a Extension (o) 4.5 (3.3/6.4) 5.5 (4.5/7.3) 0.247a Right Rotation (o) 5.2 (3.0/8.9) 6.4 (4.3/9.0) 0.513a Left Rotation (o) 4.3 (2.5/5.0) 5.4 (3.5/7.9) 0.084a Right Lateral Flexion (o) 4.9 (3.3/6.8) 5.9 (3.6/8.4) 0.531a Left Lateral Flexion (o) 4.3 (2.1/6.7) 3.8 (2.4/5.6) 0.857a IQR 25/75 Interquartile range 25/75, BASDAI Bath Ankylosing Spondylitis Disease Activity Index, BASMI Bath Ankylosing Spondylitis Metrology Index, BASFI Bath Ankylosing Spondylitis Functional Index aMann–Whitney U test bChi-square test Cervical JPE was similar between groups for all directions at baseline (p > 0.05, Table 3). Cervical JPE was significantly improved after 6 weeks in the exercise group for all directions (p < 0.05, Table 3), while no improvements were observed in the control group (p > 0.05, Table 3). Although cervical stabilization exercises resulted higher net changes in JPE measurements in comparison to control patients, those differences did not reach statistical significance (Table 4, p > 0.05).Table 3 Changes in cervical joint position errors in the groups Before After p-valuea Median (IQR 25/75) Median (IQR 25/75) Exercise group (n = 20) Flexion (o) 4.9 (2.2/7.0) 2.8 (1.7/3.8) 0.033* Extension (o) 4.5 (3.3/6.4) 3.1 (1.8/4.8) 0.040* Right rotation (o) 5.2 (3.0/8.9) 3.7 (1.9/4.7) 0.006* Left rotation (o) 4.3 (2.5/5.0) 2.8 (2.2/3.3) 0.017* Right lateral flexion (o) 4.9 (3.3/6.8) 2.3 (1.8/3.7) 0.009* Left lateral flexion (o) 4.3 (2.1/6.7) 2.0 (1.5/3.4) 0.010* Control group (n = 19) Flexion (o) 6.3 (3.5/7.3) 5.2 (3.8/7.0) 0.856 Extension (o) 5.5 (4.5/7.3) 4.1 (3.3/8.2) 0.809 Right rotation (o) 6.4 (4.3/9.0) 5.5 (3.0/8.5) 0.472 Left rotation (o) 5.4 (3.5/7.9) 5.0 (3.5/7.2) 0.778 Right lateral flexion (o) 5.9 (3.6/8.4) 4.3 (2.7/7.7) 0.717 Left lateral flexion (o) 3.8 (2.4/5.6) 4.9 (2.9/5.7) 0.904 IQR 25/75 Interquartile range 25/75 aWilcoxon signed-rank test *p < 0.05 Table 4 Comparison of the net changes in the cervical joint position errors between groups Direction Exercise group Control group p-valuea (n = 20) (n = 19) Median (IQR 25/75) Median (IQR 25/75) ∆ Flexion (o) −1.5 (−4.8/0.8) 0.3 (−1.6/2.5) 0.079 ∆ Extension (o) −1.9 (−3.5/0.9) 0.3 (−2.4/2.7) 0.177 ∆ Right rotation (o) −2.5 (−5.2/−0.6) −0.4 (−2.3/1.0) 0.081 ∆ Left rotation (o) −2.0 (−3.3/−0.7) 0.7 (−3.7/2.7) 0.133 ∆ Right lateral flexion (o) −1.0 (−2.6/−0.1) 0.2 (−1.9/1.4) 0.164 ∆ Left lateral flexion (o) −1.9 (−4.9/−0.3) −0.2 (−2.7/2.8) 0.109 IQR 25/75 Interquartile range 25/75, ∆ difference aMann–Whitney U test Discussion The results of the present pilot study showed for the first time that performing a home-based cervical stabilization exercise program three times a week is beneficial for improving cervical position error compared to usual care in patients with axSpA. The cervical spine is a commonly affected region in axSpA. Our group recently showed that patients with axSpA have higher cervical joint position error compared to healthy controls [5]. The cervical region contains a high density of mechanoreceptors [17], where most are located in the deep cervical muscles and their attachments. Thus, a static chin-tuck exercise was mainly used as the cervical stabilization exercise in various body positions in the present study and combined with dynamic extremity movements. Sustaining chin-tuck exercises activates deep cervical muscles and allows patients to be aware of their cervical positions at a certain time which might lead to improvements in proprioception. In addition, cervical stabilization exercises might lead to improvements in muscle functions such as strength and endurance, which may also improve the cervical joint position error. The positive effects of exercise on cervical proprioception have been reported previously [10, 18]. However, there is no established standardized exercise protocol for proprioceptive rehabilitation of the cervical region. Previous studies used eye/head coordination exercises, gaze stability exercises, head relocation exercises, eye-follow exercises, and craniocervical flexion training. Craniocervical training as described by Jull et al. targets the deep flexor muscles that control cervical lordosis flattening during exercises [10]. The chin-tuck exercises in the present study also flatten the cervical lordosis, similar to exercises in the study of Jull et al. [10]. It seems that positioning the neck in a flattened cervical position may lead to improvements in proprioception sense by activating deep cervical flexor muscles. Repeated contractions during cervical stabilization exercises may also enhance the spindle function, which contributes to cervical proprioception. In addition, Lee et al. showed improvements in the cervical position error following cervical stabilization exercises in healthy subjects [12]. Improvements in the cervical position error were observed in all directions following the cervical stabilization exercise program in the present study. On the other hand, the control group, who received their usual care, did not show any significant improvements in the study period. However, when the changes were compared between groups, even though the cervical stabilization exercise group obtained better results, no statistical significance was detected between groups. This may be related to the relatively small sample size of the study, or the 6‑week exercise period may not be adequate to establish a possible difference. The effect of the exercises on cervical proprioception was investigated for the first time in the present study. The exercise program was delivered via a messaging service (WhatsApp messenger; Meta, Menlo Park, CA, USA) throughout the study. Patients received their exercises as short video instructions on a weekly basis, and they performed their exercises at home. In addition, they were regularly controlled via the same messaging platform weekly. Obtaining promising results from a such home exercise program is especially important for current conditions with the world facing the COVID-19 pandemic. In this regard, performing regular exercise programs with minimum physical contact may be a good example of telerehabilitation for these kinds of emergencies. Previous studies have ensured the correct position during cervical exercises using a biofeedback unit (Stabilizer Pressure Biofeedback Unit; Chattanooga Group Inc., Hixson, TN, USA). However, as we used cervical stabilization exercises as a home-exercise program, we could not benefit from this tool. Proprioception was evaluated in terms of one aspect only (position sense) in the present study; however, the effect of a stabilization exercise program on other parameters of proprioception such as kinesthesia may be different. Also, the effects of cervical stabilization exercises on other parameters such as balance, physical functions, and activities of daily living yet to be discovered and should be addressed in future studies. In addition, different cervical stabilization exercise programs with various loadings, durations, or times should be investigated to determine the optimal program. The lack of longer follow-up is another limitation of the present study. Moreover, the small sample size of the present pilot study may not allow us to draw firm conclusions. Future studies with more participants are needed to confirm the results. According to the results of the present study, a 6-week home-based cervical stabilization exercise program is beneficial for improving cervical proprioception in patients with axSpA. Patients should be encouraged to exercise on a regular basis and home-based exercise programs with regular follow-ups may be used in cases where physical contact should be avoided. Declarations Conflict of interest H.E. Oz, G. Duran, D. Bayraktar, M. Kara, D. Solmaz, and S. Akar declare that they have no competing interests. All procedures performed in studies involving human participants or on human tissue were in accordance with the ethical standards of the institutional and/or national research committee and with the 1975 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study. Additional informed consent was obtained from all individual participants from whom identifying information is included in this article. Clinical trial registration: ClinicalTrials.gov (NCT04483648). Scan QR code & read article online ==== Refs References 1. Rudwaleit M van der Heijde D Landewé R Listing J Akkoc N Brandt J The development of Assessment of SpondyloArthritis international Society classification criteria for axial spondyloarthritis (part II): validation and final selection Ann Rheum Dis 2009 68 777 783 10.1136/ard.2009.108233 19297344 2. Strand V Singh JA Patient burden of axial spondyloarthritis J Clin Rheumatol 2017 23 383 391 10.1097/RHU.0000000000000589 28937474 3. Garrido-Cumbrera M Delgado-Domínguez CJ Gálvez-Ruiz D Mur CB Navarro-Compán V Atlas Working Group The effect of axial spondyloarthritis on mental health: results from the atlas J Rheumatol 2019 46 1284 1289 10.3899/jrheum.180868 30770507 4. Valido A Crespo CL Pimentel-Santos FM Muscle evaluation in axial spondyloarthritis-the evidence for sarcopenia Front Med 2019 6 219 10.3389/fmed.2019.00219 5. Ozen T Tonga E Polat MG Bayraktar D Akar S Cervical proprioception accuracy is impaired in patients with axial spondyloarthritis Musculoskelet Sci Pract 2021 51 102304 10.1016/j.msksp.2020.102304 33227676 6. Röijezon U Clark NC Treleaven J Proprioception in musculoskeletal rehabilitation. Part 1: Basic science and principles of assessment and clinical interventions Man Ther 2015 20 368 377 10.1016/j.math.2015.01.008 25703454 7. Hogervorst T Brand RA Mechanoreceptors in joint function J Bone Joint Surg Am 1998 80 1365 1378 10.2106/00004623-199809000-00018 9759824 8. Proske U Wise AK Gregory JE The role of muscle receptors in the detection of movements Prog Neurobiol 2000 60 85 96 10.1016/s0301-0082(99)00022-2. 10622377 9. Knoop J Steultjens MP van der Leeden M van der Esch M Thorstensson CA Roorda LD Proprioception in knee osteoarthritis: a narrative review Osteoarthritis Cartilage 2011 19 381 388 10.1016/j.joca.2011.01.003 21251988 10. Jull G Falla D Treleaven J Hodges P Vicenzino B Retraining cervical joint position sense: the effect of two exercise regimes J Orthop Res 2007 25 404 412 10.1002/jor.20220 17143898 11. Jull GA Falla D Vicenzino B Hodges PW The effect of therapeutic exercise on activation of the deep cervical flexor muscles in people with chronic neck pain Man Ther 2009 14 696 701 10.1016/j.math.2009.05.004 19632880 12. Lee MY Kim SG Lee HY The effect of cervical stabilization exercise on active joint position sense: a randomized controlled trial J Back Musculoskelet Rehabil 2016 29 85 88 10.3233/BMR-150601 26406219 13. Akkoc Y Karatepe AG Akar S Kirazli Y Akkoc N A Turkish version of the bath ankylosing spondylitis disease activity index: reliability and validity Rheumatol Int 2005 25 4 280 284 10.1007/s00296-003-0432-y 14730386 14. Ozer HT Sarpel T Gulek B Alparslan ZN Erken E The Turkish version of the bath ankylosing spondylitis functional index: reliability and validity Clin Rheumatol 2005 24 2 123 128 10.1007/s10067-004-0984-6 15340866 15. Jenkinson TR Mallorie PA Whitelock HC Kennedy LG Garrett SL Calin A Defining spinal mobility in ankylosing spondylitis (AS). The Bath AS Metrology Index J Rheumatol 1994 21 9 1694 1698 7799351 16. Revel M Andre-Deshays C Minguet M Cervicocephalic kinesthetic sensibility in patients with cervical pain Arch Phys Med Rehabil 1991 72 288 291 2009044 17. Boyd-Clark LC Briggs CA Galea MP Muscle spindle distribution, morphology, and density in longus colli and multifidus muscles of the cervical spine Spine 1976 27 694 701 10.1097/00007632-200204010-00005 18. Humphreys BK Irgens PM The effect of a rehabilitation exercise program on head repositioning accuracy and reported levels of pain in chronic neck pain subjects J Whiplash Rel Disord 2002 1 99 112	36459172	PMC9717571	NO-CC CODE	2022-12-06 23:23:26	no	Z Rheumatol. 2022 Dec 2;:1-7	utf-8	Z Rheumatol	2,022	10.1007/s00393-022-01295-1	oa_other
==== Front Soc Netw Anal Min Soc Netw Anal Min Social Network Analysis and Mining 1869-5450 1869-5469 Springer Vienna Vienna 1003 10.1007/s13278-022-01003-6 Original Article Media moments: how media events and business incentives drive twitter engagement within the small business community Trifiro Briana [email protected] 1 Clarke Michael [email protected] 1 Huang Sunny [email protected] 2 Mills Brittney [email protected] 3 Ye Yijun [email protected] 1 Zhang Siming [email protected] 1 Zhou Maoxin [email protected] 1 Su Chris Chao [email protected] 1 1 grid.189504.1 0000 0004 1936 7558 Boston University, Boston, USA 2 grid.35403.31 0000 0004 1936 9991 University of Illinois at Urbana-Champaign, Urbana, USA 3 grid.256023.0 000000008755302X Fordham University, Bronx, USA 2 12 2022 2022 12 1 17414 10 2022 17 11 2022 19 11 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Austria, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Twitter is one of the most popular social networking platforms today with nearly 238 million active daily users. While the platform is used by a myriad of individuals for various purposes, businesses both large and small have begun to adopt Twitter into their business strategy to better connect with consumers. Considering the growing emphasis on social media engagement in the business sector, the present study examines some of the fastest-growing American small businesses from the perspective of media events theory. According to media events theory, certain large-scale events will attract excess viewership and attention from the public, both on traditional and digital platforms. We examine how small businesses leveraged media events of 2020, including COVID-19 and the 2020 US presidential election, so as to increase engagement and foster the growth of their businesses via Twitter. Using 35000 tweets based on media event-related hashtags collected throughout 2020, we investigated Twitter engagement among 100 of the fastest-growing small businesses in the USA. Through the use of network analysis metrics, we illustrate that businesses that tweeted about media events often received greater levels of user engagement and exerted greater influence over their respective networks. Keywords Network analysis Media events Engagement Social networking issue-copyright-statement© Springer-Verlag GmbH Austria, part of Springer Nature 2022 ==== Body pmcIntroduction Twitter, known for its short 280‐character tweets, has become a fertile environment for trending hashtags, sensational headlines, and fledgling social movements (Bennett and Segerberg 2011; Rosenbaum and Bouvier 2020). Presently, Twitter is one of the largest social media networks in the world, with 237.8 million active daily users as of late 2022 (Reimann 2022). Among the leading social media platforms, Twitter has garnered considerable scholarly attention due to its popularity, size, and accessibility for researchers. Although much of the existing literature focuses on how individual users leverage the platform (Williams et al 2013), there is a notable gap in both theory and practice regarding how businesses can leverage the platform to maximize engagement, foster brand awareness, and build a consumer base. Furthermore, studies within the existing literature pertaining to how businesses utilize Twitter focus primarily on larger, international brands, such as Disney or McDonald’s, as opposed to small businesses (Curran et al. 2011; Cripps et al. 2020). Relatively little is known about small businesses’ patterns and daily practices using Twitter and social media in general. We contend that the Twitter accounts run by small businesses can be distinguished through two perspectives. Firstly, these accounts share content in an effort to promote their business, advertise their brands, or sell their products—functions we categorize as business-centric activities. Second, these accounts share content when engaging with media events, such as championing social causes, “live tweeting” events as they happen, or otherwise commenting on major current events. The present study endeavors to understand the daily behavior of small businesses on Twitter and examines how media events can affect these online practices, thereby determining the interaction networks of small businesses on Twitter. Through this, we utilize novel research methodologies to elucidate how small businesses utilize social media to leverage their online presence and better connect with potential customers. Given the goals of the present study, we root our work within Dayan and Katz’s (1992) theory of media events. The framework offered by media events theory demonstrates a clear understanding of the benefits of fostering social media engagement of users on social media platforms such as Twitter (Smith and Gallicano 2015). While the existing literature illustrates how large, transnational conglomerates and mega-corporations have successfully utilized media events to foster their online presence (Lin et al. 2014), small businesses have the same opportunity to use these events to garner attention and brand awareness. With this in mind, the present study compares the nuances between individual, business-centric and event-driven social media engagement by assessing how and why users interact with each other on which topics, and how these interactions result in overall engagement with a brand, whereas the existing literature primarily conceptualizes user engagement as an individual’s personal, intrinsic connection with a platform (Li et al. 2013), and this study differs as we will utilize engagement as a metric to quantify how media events impact and how Twitter users engage with online content issued by small businesses. Furthermore, considering the existing literature surrounding the notion of media storms (Boydstun et al. 2014), it stands to reason those periods of intense media coverage can transcend into instances marked by high levels of social media engagement. From the Arab Spring revolutions to the #BlackLivesMatter protests that captivated the attention of millions worldwide—individuals are increasingly turning to social media to share their thoughts about these significant moments. Considering the tenets offered in the existing literature about media events theory in addition to that of media storms, the present study seeks to address the ways in which small businesses can capitalize on massive media events to leverage their social media presence. Through this, we seek to better understand how small businesses utilize Twitter to make recommendations to aid these industries. Finally, through the use of social network analysis (SNA), the present study seeks to utilize a novel methodological approach to understanding how small businesses interact—and leverage their presence—on social media. Literature review Small businesses on twitter As noted in the existing research (Bulearca and Bulearca 2010; Humphreys and Wilken 2015), Twitter can be a valuable tool for businesses, especially those that would find ways to learn how consumers perceive their products. As a great venue for businesses to interact and communicate with their customers, business owners can use Twitter to gauge how the public feels about their products, business practices, and more. Further, businesses can utilize Twitter to adjust their internal strategies based on their goals and by using the feedback and information provided by the platform’s users. By establishing contact with potential target markets, the business can gain insight into their needs, wants, and behaviors (Culnan et al. 2010). Twitter provides businesses with opportunities to reach out to consumers and understand their thoughts, illustrating the platform’s usefulness for small businesses. Despite the demonstrated advantages that Twitter provides small businesses, there remains a noticeable gap within the literature regarding how these organizations optimize their presence on the platform. Much of the existing literature focuses exclusively on how large, transnational conglomerates and mega-corporations capitalize on media events to foster their online presence (Lin et al. 2014). However, with nearly 31 million small businesses accounting for nearly 60 million jobs as of 2019 in the USA (U.S. Small Business Administration 2019), we argue that small businesses represent an under-studied case study in the relationship between media events and social media engagement. The present study aims to understand how these small businesses leverage event-driven activities to optimize their presence on Twitter. While there is a myriad of conceptualizations used to define what constitutes a small business in contemporary corporate America, we offer the following parameters through which we identified the small businesses that comprise our sample. For the present study, we are particularly interested in organizations that are US-based, privately held, for-profit, and independent (meaning they are not subsidiaries or divisions of another parent company), with a minimum revenue of $100,000 (PRNewswire 2021). Given the general lack of knowledge about small business activities on Twitter (Lin et al. 2014), we offer the following set of research questions in an effort to understand the characteristics of the interaction networks of the fastest-growing American small businesses:RQ1a: Do American small businesses form collaboration networks through their interactions on Twitter? RQ1b: What are the characteristics of the Twitter interaction network among American small businesses? Small businesses on twitter and media events theory To date, ample research has been dedicated to understanding how major media moments compel and demand audience attention (Al Nashmi 2017; Lin et al. 2014). Perhaps one of the most seminal theoretical frameworks in this area is media events theory, which has been applied to a myriad of contexts in an effort to illustrate how media moments captivate the attention of audiences and spur engagement. Originally offered in the 1990s, the seminal framework sought to address how ceremonial media spectacles interrupt people’s daily routines with preplanned and scripted media content. However, since its original conceptualization, the theory has been applied to analyze the impact of significant, headline-worthy events on the behavior of large enterprises and global business entities. Furthermore, while Dayan and Katz (1992) focused on the influence of media events on television coverage, the framework has since been expanded to include new and digital media, including Twitter (Jungherr 2014; Katz and Liebes 2007). The societal shift to digital media has afforded audiences greater opportunities to engage with global and local media events. To this end, media events highlighted on Twitter have been found to draw excess attention compared to other events, such as the Black Lives Matter protests in Ferguson, Missouri (Jackson and Foucault Welles 2016). Existing research shows that these events have been leveraged by large businesses to generate media exposure (Lin et al. 2014; Jungherr 2014). As defined by Dayan and Katz (1992), a media event is a phenomenon unique to the communication industry, as it attracts the public’s attention away from their daily routine by emphasizing something of social relevance (Jackson and Foucault Welles 2016; Jungherr 2014; Sumiala et al. 2016). The traditional conceptualization of a “media event” required eight elements in order to be characterized under their typology: (1) be broadcast live by television, (2) constitute an interruption of everyday life and everyday broadcasting, (3) be preplanned and scripted, and (4) be viewed by a large audience. There should also be (5) a normative expectation that viewing was obligatory and (6) a reverent, awe-filled narration, and the event had to be (7) integrative of society and (8) primarily conciliatory (Sonnevend 2017). Thus, the original authors presented three basic characterizations that media events could be categorized as contests (for example, the World Cup, the Olympic Games, or presidential debates), conquests (such as the moon landing or the Pope’s visit to the USA), and coronations (such as President Kennedy’s funeral, the coronation of Elizabeth II, and the various royal weddings that have taken place over the last few years) (Sonnevend 2017). To summarize, these events were characteristically mesmerizing and spectacular, capturing audiences’ attention across the globe. Considering this, a core concept of media events theory focuses on the nature of the community setting and the inherent communal element of the events themselves. As demonstrated throughout the existing literature, media events invite participation from both spectators and community members. As postulated in the existing literature, they draw interested viewers or community members when media events occur, sparking conversations among individuals and fostering a collective conversation. One avenue where this can mainly be observed is on social media, as networking platforms offer a space for individuals to disseminate information through social interactions with other individuals, groups, and organizations (Botha and Mills 2012). Further, online platforms offer several affordances that help to foster these interactions. For example, the use of hashtags helps catalog conversations and interactions in an archival fashion, conveniently locating similar content in one space. This ultimately assists users in decreasing their time spent searching for information. In summary, the communal aspect of media events has provided users and businesses on Twitter with shared public forums that encourage the free flow of information and connections. These collaborative platforms are spaces where businesses should theoretically circulate their content freely and connect with potential consumers. This presents a valuable opportunity for small businesses looking to grow their presence on social media and foster brand loyalty among consumers. As demonstrated throughout the existing literature, the theoretical foundation provided by media events theory demonstrates that media events have proven to be helpful in generating content exposure for large businesses on Twitter in the same fashion that has been successful for traditional television-based media events (Sumiala et al. 2016). It stands to reason that, with their behavior mimicking that of large businesses in other ways on Twitter, small businesses would benefit from media events-related practices in the same way that larger businesses do (Lin et al. 2014; Jungherr 2014). In today’s economy, small businesses often need to utilize social media to connect with potential consumers, enhance their brand awareness, and captivate the public’s attention in today’s saturated media market. Initially, this attention was captivated through traditional formats such as broadcast television; however, throughout the last decade, social media has become a vital tool in attracting audience attention (Al Nashmi 2017; Lin et al. 2014). Dayan and Katz (1992) demonstrated that media events attract exceptional levels of attention to shared spaces, such as a collaborative internet platform like Twitter. If a small business participates in these shared spaces surrounding a media event, it can leverage its attention to bring greater awareness to its brand (Dayan and Katz 1992; Lin et al. 2014). With the goal of increased awareness and the function of media events to increase participation and thus awareness, the features afforded by Twitter provide small businesses with the ability to connect with users, who may then serve as potential customers. At present, much of the existing literature focuses on how large corporations utilize social media as a portion of their marketing strategy (Lin et al. 2014), effectively ignoring how small businesses can also benefit from these new digital platforms. Based on the existing literature on the impact of media events on small businesses (Roig et al. 2017; Sumiala et al. 2016), the present study seeks to analyze the role of media events in how small businesses position themselves on Twitter. Like traditional media formats such as network television (Dayan and Katz 1992), the existing literature indicates that media events have generated audience attention on social media platforms such as Twitter (Jungherr 2014; Lin et al. 2014). Furthermore, research indicates that small businesses stand to benefit from media events, as these events are often coupled with increased audience participation, resulting in greater levels of engagement (Roig et al. 2017). With a substantial online presence surrounding a media event (Jackson and Foucault Welles 2016), there is ample opportunity for a small business to engage and draw attention from users to their brand from events that the brand is engaging, if not their product, platform, or service. Suppose small businesses are not using the opportunity of media events to their advantage. In that case, it seems clear that they are missing out on an opportunity for growth that comes at little to no cost to take advantage of (Cripps et al. 2020; Curran et al. 2011). Posts issued by small businesses concerning media events can draw additional attention compared to regular (non-event) social media posts due to the “must-see” nature of media events. Considering the significant media events of the last decade—Arab Spring, #MeToo, #BlackLivesMatter, Occupy Wall Street—millions of individuals worldwide continuously turned to social media for updates and connected with others. Social media, and these media events, help connect individuals throughout the world and create a dialog that would be impossible without digital media. Furthermore, individuals often mimic observed behaviors to conform to social platforms, such as participating in online conversations surrounding these major media events (Wang et al. 2012). One such example could be observed in June 2020, when countless Instagram users took to the platform to post images of black squares in protest of police brutality. This online protest occurred as a result of the murder of George Floyd and the ensuing Black Lives Matter protests in 2020 (Heilweil 2020). This combination of eyeball-drawing factors increases the exposure of small business’ media posts when made concerning media events. In summary, we posit that small businesses, like individual Twitter users, can also take advantage of these events and leverage their brands by connecting with potential consumers. Media storms Another critical element that merits discussion here is that of media storms. In traditional media environments, audience attention oscillates in reaction to newsworthy events that occur on a daily basis. As articulated in their seminal article, Boydstun and colleagues (2014) refer to these oscillations as media storms. These storms refer to sudden surges in news coverage surrounding a particular event that ultimately leads to a sustained period of great audience attention (Boydstun et al. 2014). To date, media storms have been studied in a myriad of different contexts and are often referred to as various other idioms, such as media “hypes” or “waves” (Boydstun et al. 2014) or “news flashpoints” (Waisbord and Russell 2020). Given the existing literature pertaining to how audience attention coalesces around major events, we contend that there is major theoretical overlap to be found in the contexts of media storms and Dayan and Katz’s (1992) original framework of media events theory. To date, scholars have focused almost exclusively on how attention cascades and coalesces in line with political events (Walgrave et al. 2017). For example, analyzing data capturing US media attention and congressional hearings spanning from 1996 to 2006, Walgrave et al. (2017) found that the presence of media storms often conditions the effects of media attention on congressional attention. These sudden, large bursts of media attention often direct the public agenda as they captivate large swaths of the audience. As a result, Walgrave et al. (2017) argue that a one-story increase in media attention can have a more significant effect on congressional attention as media storms often surpass a crucial threshold for catching the attention of policymakers—demonstrating the vast impact that media storms can have on public salience and issue importance. These findings are also reinforced through Boydstun et al. (2014) work, which not only shows that media storms can impact audiences’ awareness of specific issues and events, but also their perceptions of the world around them. Similarly, Waisbord and Russell (2020) describe their concept of news flashpoints, which refer to bursts of news attention that are unique to the networked news environment, in which various forces vie to influence public discourse. These flashpoints are brief and sudden periods where interest in specific topics rises and falls rapidly across multiple news outlets. Theoretical contribution and research hypotheses Describing how audience attention shifts and coalesces in reaction to major media events, the present study seeks to address the theoretical linkages that exist within the seminal frameworks offered by Dayan and Katz’s (1992) seminal theory of media events and the newer, burgeoning work on media storms (Boydstun et al. 2014; Waisbord and Russell 2020). One key difference that merits mention between these two frameworks is that media storms refer to a surge of attention from traditional media sources in much of the existing literature. In contrast, the present study seeks to expand upon this existing work and address how social media engagement and attention changes in reaction to these events. We posit that as major events occur—such as presidential elections, national protest movements or global events like the Olympics—a media storm is generated, thus garnering the attention of the public and motivating Twitter users to participate in the broader conversation. However, while existing work focuses on how individuals engage with these media events, we contend that they also offer unique opportunities for businesses and organizations to engage in a meaningful way and optimize their social media presence. Rather than sudden attention from traditional media sources, we argue that media events spur increased public attention, resulting in increased levels of social media engagement. Here is where we see the potential for small businesses, as these organizations should capitalize on these events and participate in these conversations the same way individuals can. This provides a free, effective way to increase the visibility of small businesses that may otherwise struggle to optimize their social media presence and reach potential consumers. Considering the structure of media storms, we argue that small businesses should engage in storm-like behavior in the same way traditional media formats do in order to connect with potential consumers and further leverage their social media presence. Considering the existing literature, we contend that a potential theoretical extension of media events theory and media storms is the potential distinction between individual-centered and event-driven activities. This discrepancy reflects how individuals, such as small businesses in our specific case study, react to unprecedented events while maintaining their daily activities. The media events theory suggests that individuals have no agency in maintaining their everyday routines as the entire network produces narratives relative to a single topic or event. However, small business owners simultaneously balance their engagement with these events with the facilitation of their business goals—such as sharing information about new products and upcoming promotions. This observation enables us to examine both individual-centric activities and engagement that maintain these organizations’ business-natured daily routines and event-driven activities and engagement that are interrupted by media events or media storms. With this goal in mind, we offer the following hypotheses to understand the relationship between media events and Twitter engagement: H1 For small businesses on Twitter, business-centric engagement will be positively related to event-driven engagement. H2 Small businesses with higher levels of business-centric engagement on Twitter are more likely to be in central positions on the interaction network formed through tweet activity. H3 Small businesses with higher levels of event-driven engagement on Twitter are more likely to be in central positions on the interaction network formed through tweet activity. H4 The relationship between business-centric engagement and the degree of network centrality among small businesses on Twitter can be enhanced if small businesses also show higher levels of event-driven engagement. H5 The hashtag networks of small businesses’ business-related tweets formed through (a) user-generated content, (b) sharing content/retweets, and (c) replies are positively correlated with the hashtag networks of small businesses’ media events-related tweets formed through the same kinds of tweeting activities. Method Data collection The present study utilizes a content analysis to ascertain how small businesses capitalize upon media events to build brand awareness. Thus, the unit of analysis for the present study is an individual tweet issued by a small business. To collect data, we utilized a simple random sample approach to select the businesses that would comprise our dataset. We randomly selected 100 small businesses from a list of the 5000 fastest-growing US-based small businesses as published by INC Magazine. To reiterate, the organizations on this list must be US-based, privately held, for-profit, and independent (meaning they are not subsidiaries or divisions of another parent company), with a minimum revenue of $100,000 (PRNewswire 2021). To merit inclusion in our sample, the business must also (1) have a Twitter account and (2) have tweeted on said account during 2020. The data compiled using INC.com included various attributes, such as industry, year founded, and company size. To obtain the most comprehensive perspective of how small businesses interact with media events on Twitter, we collected all tweets issued by the 100 small businesses from January 1, 2020, to December 31, 2020. The data collection yielded a total corpus of 35,177 tweets. In order to achieve a subsample for data analysis, 3704 of the tweets were randomly selected to code to surpass the 10% threshold for representativeness. Coding procedure The coding process was completed by two researchers. Each tweet was hand-coded according to a codebook identifying the media events as concerned in this study. Coders were tasked with identifying the types of tweets issued by each organization in the sample and categorizing whether they facilitated business-oriented content or media event-related content. We were specifically interested in four major global media events that we believe generated considerable media attention, resulting in what the existing literature describes as a media storm (Boydstun et al. 2014). These events were the COVID-19 pandemic, the 2020 US presidential election, the Black Lives Matter protests, and the 2021 (previously 2020) Olympics. Tweets that did not mention a media event but instead focused on some aspect of business strategy (e.g., branding, advertising, and promotions) were coded as such. Three rounds of intercoder reliability were conducted, yielding a final percent agreement of 99.5 and Krippendorff’s alpha of 0.99. Measurements Twitter affords three separate measurements to gauge interaction with individual tweets. Each tweet can be liked and will display a total number of times it has been liked. Tweets can also be replied to, allowing for a direct response to an individual post. Lastly, there is the option to retweet, wherein a user will repost another user’s tweet, furthering its reach. These three features are built into the platform and allow for a quick and easy measure of engagement. Additionally, with these three features requiring user input, they provide direct, definitive proof of engagement. More specifically, our study analyzes two specific forms of social media engagement. The first, business-centric engagement, is measured by the number of tweets posted about any business, such as promotions, sales, black Friday events, new product launches, and more. For example, if Twitter account A, which represents a small business, posts three tweets or retweets about any business affairs, the business-centric engagement is three. The second form of engagement we analyzed is event-driven engagement. Like business-centric engagement on Twitter, the event-driven engagement is measured by the number of tweets posted about any media events, including Black Life Matters (BLM), COVID-19, Olympics in Tokyo, and the 2020 US Presidential Election. Each mention or tweet about a media event counts as an instance of event-driven engagement, which is added together to form the total event-driven engagement of each small business. Network analysis As discussed in the existing literature, analyzing the structure of a social network often lends great insight into the interactions and relationships that connect users while simultaneously revealing patterns of behavior (Kumar Behara et al. 2019). An activity-based interaction network emerges from all the tweets made by the selected small businesses and their associated Twitter users, including quotes, retweets, replies, and mentions. This network consisted of 10902 nodes—representing various Twitter accounts—and 11390 edges—representing interactions among these accounts. Based on the interaction networks on Twitter, this study adopted the in-degree centrality concept from social network analysis to measure the relative positions of small businesses on Twitter’s interaction network. Degree centrality metrics count the number of interactions or connections a Twitter user has with other Twitter users. Given that Twitter networks are directed (e.g., account A may mention account B, but account B may not mention account A), this study uses in-degree centrality to measure the number of connections or interactions others have initiated with a given Twitter account. For instance, if small business A was mentioned five times by other Twitter users in the above-formed interaction network, A’s in-degree centrality metric would be five. The higher the in-degree centrality, the more central the Twitter user account is (Kumar Behara et al. 2019). To better understand how small businesses on Twitter engage with (1) individual-centric business topics and (2) media events-related topics, this study also computes the hashtag networks based on the number of tweets each pair of hashtags appears. If two hashtags #BLM and #COVID appeared in ten tweets simultaneously, an edge with a weight of ten was created between the two hashtags. A total of 780 hashtags were retrieved from all tweets in our sample. In addition, we created three different hashtag networks using the same strategy based on the type of tweeting activity among Twitter accounts, namely (a) self-posting user-generated content, (b) sharing content/retweeting, and (c) replies. Depending on how all small businesses and Twitter users interact, the level of interaction varies for each type of hashtag network. Results An overview of small businesses on twitter We first ran descriptive statistics to gain a comprehensive overview regarding the industries of the small businesses in our sample. The accounts in the sample date back to 2008, with seven accounts, created that year. The youngest of the accounts were created in 2020, with six accounts created within the last year. The businesses span twelve different industries, the most prevalent being technology and computers (n = 37). The least represented industries were automotive, education, and home and garden, each of which only had a single business. Thirty-four businesses were classified as “other,” with industries such as marketing and recruiting making up some of those in that category. Seventy-six accounts were national small businesses. These businesses served clients across the USA. Of the remaining twenty-four, nine were international businesses, and fifteen were categorized as local businesses in only a few selected cities or small regions throughout the USA. Businesses were also categorized by having a traditional industry model, where they have physical business locations or products. Each business was coded as either a traditional industry, an online/digital business, or both. Of one hundred businesses in the sample, fifty-four were categorized as “traditional industry,” thirty as “online/digital businesses,” and the final sixteen were “both.” Lastly, each Twitter account was categorized as either a verified account or not. At the time of data collection, Twitter verification was assigned by the platform to indicate the authenticity of an account. In order to be verified, an account must have represented or somehow been associated with a prominently recognized individual or brand that has received significant media attention (Twitter n.d.). The Twitter verification process has since changed following Elon Musk’s acquisition of Twitter in late 2022 (Bowman and Dillon 2022). Of the accounts in the study’s sample, only five were verified by Twitter. Social network of small businesses on twitter Broadly, RQ1 sought to analyze the network of small businesses on Twitter, specifically focusing on how the network was structured. A network analysis was also conducted to visualize the interactions among the small businesses in the sample (Fig. 1) to address RQ1a and RQ1b. In the network presented in Fig. 1, each node represents an individual Twitter account, including the small businesses within our sample and those with whom they have interacted on Twitter. Although sparsely connected, the network indicates that American small businesses have interacted among themselves and formed collaboration networks on Twitter (RQ1a). Specifically, the total number of nodes was 10,902, representing 10,902 accounts, with 11,390 corresponding edges. Within this context, an edge represents an interaction between Twitter user accounts. A descriptive network analysis was conducted to examine the characteristics of the Twitter interaction network among American small businesses (RQ1b). The average degree was 1.045, with an average weighted degree of 5.838. The modularity (Resolution = 1) was 0.84, representing 164 communities. Fig 1 highlights the three largest communities. Highlighted in purple, we can observe accounts and interactions that focus on media events and business affairs (30.29%). Green represents interactions focused solely on business affairs (20.78%). Finally, orange represented interactions that were exclusive to media events (15.48%).Fig. 1 Network visualization based on the small businesses and their tweeting activities In the 3704 tweets analyzed, 115 mentioned one of the four media events specified in the codebook (3.1%). We offered a set of hypotheses to analyze the relationship between media events and Twitter engagement with these small businesses. H1 asserted that business-centric engagement would be positively related to event-driven engagement. A Pearson product-moment correlation coefficient was computed to address H1 to assess the relationship between business-centric engagement and event-driven engagement. There was a weak, positive correlation between the two variables, r = 0.15 (p < 0.01), supporting H1. In addition, we conducted a series of regression analyses to examine the effects of two types of social media engagement on users’ in-degree centrality within small businesses’ Twitter networks (H2–H4). Table 1 presents the standardized regression coefficients using social media engagement to predict the network centrality of the small businesses within our sample. The first regression model uses two types of social media engagement of small businesses (business-centric and event-driven) as independent variables. The dependent variable is in-degree centrality (node-level centrality measuring how central a small business is located) in the activity-based interaction network. Both independent variables show positive correlations to small businesses’ in-degree centrality, meaning that the more involved the small business is with business-centric activities and media event-related activities on Twitter, the more centrally it is located and ranked. Accordingly, event-driven social media engagement (β = 0.38, p < 0.001) will lead to greater increases in small businesses’ centrality in networks as opposed to business-centric engagement (β = 0.21, p < 0.001). The two independent variables manage to explain 18% variance in small businesses’ network centrality, lending evidence to H2, H3, and H4.Table 1 Predicting the network centrality of small businesses on twitter IV: Social media engagement DV: In-degree centrality (β) Main effects Business-centric engagement 0.21* Event-driven engagement 0.38* Adjusted R square 0.18*** Moderation effect Business-centric engagement * Event-driven Engagement 0.10 Total Adjusted R square 0.19* Standardized regression coefficients were reported *p < 0.001, p < 0.01, p < 0.05 As well as the independent variables from the first regression model, the second regression model adds an additional independent variable: the interaction term created based on the two types of social media engagement (two independent variables serving as main effects). Results show that the interaction effect is significant, which indicates that event-driven social media engagement has a stronger impact on network centrality when the level of business-centric engagement is high (see Table 1). It stands to reason that the increase in network centrality score due to business-centric social media engagement will be amplified among small businesses with a high score for event-driven social media engagement, lending support for H2. Hashtag networks from twitter discussions A hashtag network of 780 hashtags was created to address H5. To further analyze this hypothesis, we also conducted a Multiple Regression Quadratic Assignment Procedure (MRQAP), the results of which are available in Table 2. MRQAP extends the quadratic assignment procedure (QAP) with the double semi-partialing permutation method (Dekker et al. 2007). This form of statistical analysis can assess the unique effect of one independent matrix on the dependent matrix by partialling out the effects of other predictor variables. Thus, this technique randomizes the residuals from the regression on each predictor, or fixed effect, to obtain the p value.Table 2 Predicting hashtag networks formed through event-related tweets #Hashtag networks IV (Business-related tweets) B β Adjusted R2 Event-related tweets Model 1—Self-post Self-post 0.001 0.01 0.08 Reply 0.03 0.07 Retweet 0.04 0.07 Model 2—Reply Self-post 0.002 0.17 0.19* Reply 0.12 0.01 Retweet 0.46* 1.25* Model 3—Retweet Self-post 0.003* 0.26* 0.39* Reply − 0.08 − 0.01 Retweet 0.52* 1.28* MRQAP Multiple regression quadratic assignment procedure p < 0.05, p < 0.01, **p < 0.001 Each hashtag network consisted of 780 rows and 780 columns. As shown in model 1, none of the three types of business-related hashtag networks (i.e., self-post, reply, and retweet) was significantly correlated with the event-related hashtag network of self-post content, suggesting that small businesses’ own opinions about media events on Twitter were not affected by their business incentives. In model 2, the event-related hashtag network of replies was significantly and positively predicted by the business-related hashtag network of self-post tweets (β = 0.17, p < 0.01) and the hashtag network of retweets (β = 1.25, p < 0.001). Similarly, the event-related hashtag network of retweets shared the same patterns as the hashtag network of replies (see model 3 in Table 2), significantly affected by the business-related hashtag network of self-post tweets (β = 0.26, p < 0.01) and the hashtag network of retweets (β = 1.28, p < 0.001). Both model 2 and model 3 suggest robust statistical explanatory power in predicting the dependent variables (Adjusted R-square = 0.19 and 0.39, respectively, p < 0.001). Our results indicate that H5a is not supported, whereas both H5b and H5c are. Discussion For the last decade, the role of social media has evolved considerably, often serving as a conduit for communication among individuals who may never otherwise have had the opportunity to connect. As a result of this information and communication technology proliferation, social media companies are increasingly establishing relationships with small businesses (Humphreys and Wilken 2015). Namely, Twitter has grown to be a particularly important channel for businesses due to its open nature and non-reciprocal networked platform (Culnan, McHugh and Zubillaga 2010), enabling organizations to expand their reach and potential customer base. Considering the existing literature of media storms/news flashpoints, heightened attention surrounding popular media spectacles has been linked to several offline outcomes (Boydstun et al. 2014; Waisbord and Russell 2020; Walgrave et al. 2017). Our research attempts to provide these organizations with an outline for success in increasing their brand awareness and subsequent user engagement. With this goal in mind, the present study applied a modern approach to Dayan and Katz’s (1992) media events theory to understand how current events coincide with social media engagement. While descriptive in nature, our goal was to analyze whether the framework of media events theory could be expanded to analyze social media content issued by small business organizations. Furthermore, the present study leverages a novel methodological approach to operationalize and visualize the engagement patterns of small businesses online. Our findings offer several implications regarding the interplay between business-related Twitter activity and event-driven activity. As discussed, several of our hypotheses were supported—indicating rigorous support to suggest that the engagement that a Twitter account receives is related to its prominence and influence within a network. Specifically, we observed that business-centric engagement, or engagement driven by daily business operations, was positively related to event-driven engagement. Similarly, our results indicate that event-driven social media engagement leads to greater increases in small businesses’ centrality in networks as opposed to business‐centric engagement. The more involved the small business is with business‐centric activities and media event-related activities on Twitter, the more central it is located and ranked within its network. Furthermore, with a specific focus on the hashtag networks spurred by these businesses, we found that these networks were correlated with both the sharing of content/retweets and replies, indicating the proliferation of the networks that these businesses were a part of. Considering the existing literature about media events (Jackson and Foucault Welles 2016; Jungherr 2014; Sumiala et al. 2016) and the storms that they generate (Boydstun et al. 2014), we argue that our findings contribute to this existing body of work by contending that engagement that is spurred by a media event contributes to an organization’s greater prominence within its network. As a result of our findings, we offer several practical, as well as theoretical, implications. Namely, we suggest that small businesses harness the opportunities provided by more significant media events. By joining the broader conversation sparked by these events, small businesses can expand their networks and leverage their position within their networks. From a theoretical vantage point, we posit that our findings offer a contemporary extension of the seminal framework as Dayan and Katz (1992) offered. While contemporary media has evolved considerably since the theory’s first conceptualization in the early 1990s, our work illustrates that media events still serve as critical moments that captivate the public’s attention. Furthermore, our study offers one of the first—to our knowledge—to utilize an extensive data content analysis to observe the impacts of a media event. Similarly, our study extends these theoretical frameworks by utilizing online user engagement as a primary dependent variable. However, the present study is not without limitations. As we only sought to measure the effect of global media events, particularly those that were deemed relevant for the 2020 calendar year, it is possible that there were local and regional media events that were also relevant to small businesses and their engagement. Due to the nature of our study, these are relationships that this study was not posed to examine. Additionally, this study focused on some of the fastest-growing businesses in the USA, many of whom already had an established presence on social media. Media events may benefit small businesses that are newcomers to social media differently from businesses with an established brand presence. Lastly, the present study only sought to look at businesses in the USA, leaving open the question of what small businesses worldwide have to gain from media event engagement. This opens potential avenues for future research, as scholars would likely benefit from a more comprehensive understanding of how the average small business performs online daily. Furthermore, future researchers could benefit from a greater understanding of the more finite uses and gratifications that motivate small business owners to engage with their online networks. In summary, the use of media events to foster social media engagement appears to be a tangible opportunity for small businesses to take advantage of. However, with only three percent of our sample discussing these events, it appears that small businesses may not pay enough attention to media events in the current environment. Furthermore, we found that most tweets issued within our sample were related to business strategy, such as advertising or developing a company’s branding. To this end, most of the content within our sample was related to promoting a specific product or service offered by the small business. While daily business operations are a vital facet of an organization’s social media strategy, our results indicate that event-driven social media engagement has a more substantial impact on a business’s network centrality when the level of business-centric engagement is high. Based solely on our preliminary research, we argue that small businesses would benefit from greater participation in these significant social and media events and can ultimately leverage them to increase their visibility. As previously articulated, if small businesses are not using the opportunities afforded by media events to their advantage, it seems clear that they are missing out on the ability to grow and expand their network that comes at little to no cost to take advantage of (Cripps et al. 2020; Curran et al. 2011). In conclusion, our findings offer several implications that can be of particular use to small business owners, social media managers, and marketing professionals. Author’s contribution A. wrote the main manuscript text. C.D. aided with question creation. B.H. conducted network analysis. E.F.G. assisted with data collection. Declarations Conflict of interest The authors declare no conflict of interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Al Nashmi E From selfies to media events Digit J 2017 6 1 98 117 10.1080/21670811.2017.1306787 Bennett WL Segerberg A Digital media and the personalization of collective action: social technology and the organization of protests against the global economic crisis Inf Commun Soc 2011 14 6 770 799 10.1080/1369118X.2011.579141 Botha E Mills AJ Close Scheinbaum A Managing the new media tools for brand management in social media Online consumer behavior: theory and research in social media advertising and E-tail 2012 New York Routledge Bowman E and Dillon RM (2022) Twitter begins advertising a paid verification plan for $8 per month. Accessed from https://www.npr.org/2022/11/05/1134561542/twitter-blue-check-paid-verification-elon-musk Boydstun AE Hardy A Walgrave S Two faces of media attention: media storm versus non-storm coverage Polit Commun 2014 31 4 509 531 10.1080/10584609.2013.875967 Bulearca M Bulearca S Twitter: a viable marketing tool for SMEs? Glob Bus Manag Res Int J 2010 2 296 Cripps H, Singh A, Mejtoft T, Salo J (2020) The use of twitter for innovation in business markets. Marketing Intelligence & Planning Culnan MJ, McHugh PJ, Zubillaga JI (2010) How large US companies can use twitter and other social media to gain business value. MIS Q Exec (9)4: 243–259. Accessed from https://aisel.aisnet.org/misqe/vol9/iss4/6 Curran K O’Hara K O’Brien S The role of Twitter in the world of business Int J Bus Data Commun Netw (IJBDCN) 2011 7 3 1 15 10.4018/jbdcn.2011070101 Dayan D Katz E Media events 1992 US Harvard University Press Dekker D Krackhardt D Snijders TA Sensitivity of MRQAP tests to collinearity and autocorrelation conditions Psychometrika 2007 72 4 563 581 10.1007/s11336-007-9016-1 20084106 Heilweil R (2020) Why people are posting black squares on instagram. Accessed from https://www.vox.com/recode/2020/6/2/21278051/instagram-blackout-tuesday-black-lives-matter Humphreys L Wilken R Social media, small businesses, and the control of information Inf Commun Soc 2015 18 3 295 309 10.1080/1369118X.2014.989249 Jackson SJ Foucault Welles B #Ferguson is everywhere: initiators in emerging counterpublic networks Inf Commun Soc 2016 19 3 397 418 10.1080/1369118X.2015.1106571 Jungherr A The logic of political coverage on twitter: temporal dynamics and content J Commun 2014 64 2 239 259 10.1111/jcom.12087 Katz E Liebes T ‘No more peace!’: how disaster, terror and war have upstaged media events Int J Commun 2007 1 1 10 Kumar Behera R Kumar Rath S Misra S Damaševičius R Maskeliūnas R Distributed centrality analysis of social network data using mapreduce Algorithms 2019 12 8 161 10.3390/a12080161 Li T Berens G de Maertelaere M Corporate twitter channels: the impact of engagement and informedness on corporate reputation Int J Electron Commer 2013 18 2 97 126 10.2753/JEC1086-4415180204 Lin Y-R Keegan B Margolin D Lazer D Rising tides or rising stars?: dynamics of shared attention on twitter during media events Plos One 2014 9 5 E94093 10.1371/journal.pone.0094093 24854030 PRNewswire (2021) Inc. Magazine Reveals Annual List of America's Fastest-Growing Private Companies--the Inc. 5000. Accessed from https://www.prnewswire.com/news-releases/inc-magazine-reveals-annual-list-of-americas-fastest-growing-private-companiesthe-inc-5000-301358141.html Reimann N (2022) Twitter Hemorrhaging Its Most Active Users, Report Says. https://www.forbes.com/sites/nicholasreimann/2022/10/25/twitter-hemorrhaging-its-most-active-users-report-says/?sh=e34ed5e6e5e1 Roig EM Fuentes EM Ramon ND User-generated social media events in tourism MDPI 2017 9 12 2250 Rosenbaum JE Bouvier G Twitter, social movements and the logic of connective action: activism in the 21st century–an introduction J Audience Recept Stud 2020 17 1 120 125 Smith BG Gallicano TD Terms of engagement: analyzing public engagement with organizations through social media Comput Hum Behav 2015 5 82 90 10.1016/j.chb.2015.05.060 Sonnevend J Media events Communication 2017 10.1093/obo/9780199756841-0189 Sumiala J Tikka M Huhtamaki J Valaskivi K #Jesuischarlie: towards a multi-method study of hybrid media events Cogitatio 2016 4 4 97 108 10.17645/mac.v414.593 Twitter (n.d.) About verified accounts. Accessed from https://help.twitter.com/en/managing-your-account/about-twitter-verified-accounts U.S. Small Business Administration. (2019). 2019 Small Business Profile. Accessed from https://cdn.advocacy.sba.gov/wp-content/uploads/2019/04/23142719/2019-Small-Business-Profiles-US.pdf Waisbord S Russell A News flashpoints: networked journalism and waves of coverage of social problems J Mass Commun Q 2020 97 2 376 392 Walgrave S Boydstun AE Vliegenthart R Hardy A The nonlinear effect of information on political attention: media storms and US congressional hearings Polit Commun 2017 34 4 548 570 10.1080/10584609.2017.1289288 Wang X Yu C Wei Y Social media peer communication and impacts on purchase intentions: a consumer socialization framework J Interact Mark 2012 26 198 208 10.1016/j.intmar.2011.11.004 Williams SA Terras MM Warwick C What do people study when they study twitter? classifying twitter related academic papers J Doc 2013 69 3 384 410 10.1108/JD-03-2012-0027	36505398	PMC9717572	NO-CC CODE	2022-12-06 23:23:26	no	Soc Netw Anal Min. 2022 Dec 2; 12(1):174	utf-8	Soc Netw Anal Min	2,022	10.1007/s13278-022-01003-6	oa_other
==== Front ACS Biomater Sci Eng ACS Biomater Sci Eng ab abseba ACS Biomaterials Science & Engineering 2373-9878 American Chemical Society 36445062 10.1021/acsbiomaterials.2c01094 Article SARS-CoV-2 Spike Protein-Activated Dendritic Cell-Derived Extracellular Vesicles Induce Antiviral Immunity in Mice Barnwal Anjali †‡# Basu Brohmomoy §# Tripathi Aarti § Soni Naina § Mishra Debasish § Banerjee Arup § Kumar Rajesh ∥ Vrati Sudhanshu § https://orcid.org/0000-0003-0202-4770 Bhattacharyya Jayanta †‡ † Centre for Biomedical Engineering, Indian Institute of Technology Delhi, New Delhi 110016, India ‡ Department of Biomedical Engineering, All India Institute of Medical Science, New Delhi 110029, India § Laboratory of Virology, Regional Centre for Biotechnology, Faridabad 121001, Haryana, India ∥ Translational Health Science & Technology Institute, Faridabad 121001, Haryana, India * Email: [email protected]. Tel.: +911292848801. * Email: [email protected]. Tel.: +911126591337. 29 11 2022 12 12 2022 8 12 53385348 15 09 2022 17 11 2022 © 2022 American Chemical Society 2022 American Chemical Society This article is made available via the PMC Open Access Subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The onset and spread of the SARS-CoV-2 virus have created an unprecedented universal crisis. Although vaccines have been developed against the parental SARS-CoV-2, outbreaks of the disease still occur through the appearance of different variants, suggesting a continuous need for improved and effective therapeutic strategies. Therefore, we developed a novel nanovesicle presenting Spike protein on the surface of the dendritic cell-derived extracellular vesicles (DEVs) for use as a potential vaccine platform against SARS-CoV-2. DEVs express peptide/MHC-I (pMHC-I) complexes, CCR-7, on their surface. The immunogenicity and efficacy of the Spike-activated DEVs were tested in mice and compared with free Spike protein. A 1/10 Spike equivalent dose of DEVs showed a superior potency in inducing anti-Spike IgG titers in blood of mice when compared to dendritic cells or free Spike protein treatment. Moreover, DEV-induced sera effectively reduced viral infection by 55–60% within 15 days of booster dose administration. Furthermore, a 1/10 Spike equivalent dose of DEV-treated mice was found to be equally effective in inducing CD19+CD38+ T-cells in the spleen and lymph node; CD8 cells in the bone marrow, spleen, and lymph node; and CD4+CD25+ T-cells in the spleen and lymph node after 90 days of treatment. Thus, our results support the immunogenic nature of DEVs, demonstrating that a low dose of DEVs induces antibodies to inhibit SARS-CoV-2 infection in vitro, therefore warranting further investigations. SARS-CoV-2 DC-derived extracellular vesicle vaccine neutralization humoral response Department of Biotechnology , Ministry of Science and Technology 10.13039/501100001407 BT/PR15984/MED/31/325/2015 Indian Institute of Technology Delhi 10.13039/501100007488 MI02233G Science and Engineering Research Board 10.13039/501100001843 EMR/2017/001490 Indian Council of Medical Research 10.13039/501100001411 2020-6132/SCR-BMS Department of Biotechnology , Ministry of Science and Technology 10.13039/501100001407 BT/PR40401/COT/142/17/2020 document-id-old-9ab2c01094 document-id-new-14ab2c01094 ccc-price This article is made available via the ACS COVID-19 subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. ==== Body pmcIntroduction COVID-19 is an outbreak of the novel coronavirus disease caused by the new coronavirus 2019-nCoV. The virus has been named SARS-CoV-2 by the International Committee on Taxonomy of Viruses (ICTV). The potential of these viruses to spread rapidly to become a pandemic worldwide seems to be a severe public health risk drawing significant global attention. Scientists worldwide have been understanding the disease and its epidemiology to find possible therapeutic agents and develop vaccines. More than 90 vaccines are being developed against the virus across the world by academic, governmental, industrial, and nongovernmental organizations.1 These research teams have tried and developed various approaches and platforms, such as virus vaccines, viral vectors, nucleic acid, or protein-based vaccines.2−4 Most vaccines target the Spike protein (S) since SARS-CoV-2 uses the receptor-binding domain (RBD) of the S protein to enter the host cells and lead to infection. However, free nucleic acid and a protein-based vaccine are associated with disadvantages such as poor stability and limited humoral and cell-mediated responses, which can be solved by incorporating some delivery vehicle that can add to the therapeutic value.3−5 Another strategy to overcome these disadvantages is to use a dendritic cell (DC)-based vaccine. DCs link innate and adaptive immunity by recognizing, ingesting, processing, and presenting antigenic peptides on MHC molecules. Despite being central antigen-presenting cells having the capacity to induce a strong immune response, the usage of the DC vaccine is associated with many drawbacks, such as their storage over a long period without losing efficacy.6,7 Several studies demonstrated that DCs undergo apoptosis shortly after interacting with the antigen-specific T-cells and get cleared from the lymph nodes within 2–3 days.8−11 As a result, the half-life of DCs in the lymph node is not enough to provide sustained antigen-specific T-cell activation.8−11 To overcome the problems associated with DCs, DC-derived extracellular vesicles (DEVs) can be used. Compared to in vivo allogeneic administration of cells and artificial nanoparticles, extracellular vesicles are less immunogenic and, thus, can be used to treat various diseases.12−14 Hence, DEVs have been acclaimed as a solution to many of the disadvantages of DC-based vaccines. DEVs are nanosized extracellular vesicles released by DCs. Unlike DCs, DEVs can be stored for at least six months at −80 °C. Activation of DCs by any specific antigen releases extracellular vesicles, which express MHC-I, MHC-II, and all the essential costimulatory molecules, important for the activation of T-cells.6,7 DEVs are known to activate naïve DCs in blood circulation, which further stimulates T-cell activation.6,7 Previously, DEVs have also been reported to activate specific T-lymphocytes directly by presenting the antigen on the surface as an MHC–antigen complex and enhancing the immunogenicity of the antigen.6,7 Recently, the DEV-based vaccine demonstrated protection against Toxoplasma gondii infection in syngeneic and allogeneic mice.15 As DEVs are merely vesicular structures, they are resistant to activated CTL, supporting the observation that DEVs could prolong Ag presentation compared to DCs in a murine model.6 Furthermore, it also increases the plasma half-life of the free protein.16 It is also reported that extracellular vesicles isolated from mRNA-vaccinated individuals are immunogenic in mice,17 suggesting the important role of Spike protein-containing extracellular vesicles in inducing an immune response. A recent work has demonstrated that engineered extracellular vesicles induced a strong SARS-CoV-2 N-specific CD8+ T immunity to protect the mice from lethal infection.18 However, none of those studies have used dendritic cells as tools to generate engineered extracellular vesicles. In this study, we have developed a cell-free vaccine using the extracellular vesicles derived from Spike-activated DCs (DEVs). A 1/10 Spike protein equivalent dose of DEVs was administered in mice, and the in vivo efficacy was compared with that of DCs and Spike protein. DEVs depicted a superior potency in inducing anti-Spike IgG titers in blood of mice, even at a lower concentration of 1/10 Spike protein equivalent dose, at early time points compared to DCs or free Spike protein. The efficacy of DEV-induced antibodies was further tested in vitro against the live SARS-CoV-2 virus. It also induced the differentiation of various subsets of T-cells, such as CD4+ T-cells, CD8+ T-cells, and T-follicular helper cells (Tfh) in secondary lymphoid organs. Our study shows that delivery of Spike protein as a DEV-based vaccine enhances the immunogenicity of free protein and induces humoral immunity by producing neutralizing antibodies. Results Synthesis and Characterization of DEVs from Bone Marrow-Derived Dendritic Cell (BMDC) Culture The DEVs isolated from the culture media of BMDCs were purified by differential centrifugation followed by ultrafiltration (Figure 1A,B). DEVs were characterized for the vesicle morphology by SEM and cryo-TEM, which revealed a spherical shape of DEVs with different sizes ranging from 50 to 200 nm (Figure 1C,D). NTA was performed for an accurate size distribution analysis of DEVs, and the data showed that the average diameter of DEVs was 134 ± 47.4 nm, including a subpopulation of various sizes. Figure 1E shows three major peaks at 96, 155, and 213 nm with a total concentration of 1.54 × 1010 particles/mL (Figure 1E). Figure 1 Characterization of Spike protein-activated DEVs isolated from BMDC culture. (A) Schematic representation of the culture of BMDC and isolation of DEVs. (B) BMDCs at day 3, day 5, and day 8 show the change of the morphology after the activation with increased dendrite formation and adherence. DEVs were isolated from Spike protein-activated DCs and analyzed by SEM (C) and cryo-TEM (D). (E) The concentration and size distribution of DEVs were analyzed by nanotracking analysis. (F) DC and DEV lysates were used in Western blot analysis for the expression of CD63 and CD86. (G) Expression of MHC-I and CCR-7 on the surface of DEVs was analyzed by Western blotting. Next, protein markers were analyzed by Western blotting of the DEV lysate. The presence of tetraspanin protein CD63 and Alix, representative markers of extracellular vesicles, confirmed that the vesicles purified were extracellular vesicles (Figure 1F). CD86 is a costimulatory molecule necessary for T-cell activation and survival; hence, the presence of CD86 on DEVs was checked by Western blotting. As shown in Figure 1F, both DCs and DEVs showed an abundance of CD86. The abundance of MHC molecules and the lymph node targeting moiety CCR-7 was also analyzed by Western blotting. As shown in Figure 1G, DEVs were rich in both MHC-I and CCR-7. The stability of DEVs was assessed by performing TEM and Western blotting of 4–5-months-old DEVs stored at −80 °C. As shown in Figure S1A, the shape and size of the old DEVs were identical to those isolated freshly from the DC culture media. The expression of MHC-I and CCR-7 was determined by Western blotting, and stored DEVs showed similar expression of both the markers to the freshly isolated DEVs (Figure S1B). Next, to determine whether DEVs have any cytotoxic effect, we treated RAW 264.7, a murine macrophage cell line, with various concentrations of DEVs (0.01–40 μg/mL), and the viability of the cells was analyzed by the MTT assay. As shown in Figure S2A, DEVs did not affect the viability of the cells at any concentration, indicating that they did not impart any cytotoxicity on the macrophages. DEV Treatment Induces a Humoral Immune Response To assess the humoral immune response, blood was collected from the tails of mice at 14, 30, 60, and 90 days after injection. The binding affinity of the DEV-induced antibodies against in-house developed Spike protein and RBD was evaluated by ELISA. Recombinant Spike protein or RBD protein was coated on 96-well ELISA plates. Serum from each group was then added to the Spike-coated wells, RBD-coated wells, and wells having the negative control (nonfatty milk). Normalization of the absorbance values consisted of dividing the absolute absorbance by the average value of absorbance readings of nonfatty milk (NFM) used in each set of experiments. All the treatment groups were observed to produce specific antibodies (Figure 2) against Spike protein as early as 15 days following the first dose. Further, the antibody levels increased multiple-fold following booster vaccination (Figure 2) and reached slightly higher levels 90 days after first injection (Figure 2A). All the mice receiving treatment produced specific antibodies against Spike protein, while the absorbance signal remained at the basal level in the control group (i.e., injected with PBS only). In addition, Figure 2A,B demonstrates that sera from DEV-treated mice (1:100 and 1:300 dilutions) exhibited significantly higher IgG levels against whole Spike proteins across the time points compared to the mice treated with DCs and Spike protein (Figure 2A). Moreover, the RBD-specific antibody was also detectable in the serum of all treatment groups (1:100 dilution) at day 15, maintained till day 30, then increased slightly at day 60, and elevated several-fold at day 90 post-treatment (Figure 2C). However, elevated levels of RBD-specific antibodies in all treatment groups at 1:100 dilutions were only observed at 90 days post-treatment (Figure 2C). The increased IgG level against both Spike protein and RBD indicated that DEVs have the potential to provide a stronger humoral response against SARS-CoV-2 infection. To analyze the overall toxicity of DEVs in mice, body weights of mice in every group were checked throughout the experiment. As shown in Figure S2B, the body weights of mice gradually increased with time in all the groups, depicting that none of the treatments imparted any toxic effect. Figure 2 Binding affinity of DEV-induced antibodies in mouse sera. An enzyme-linked immunosorbent assay (ELISA) was performed with mouse serum samples in 1:100 and 1:300 dilutions. Corrected OD450 values were normalized with the negative control. The Y axis represents log 10 values of the normalized data. (A,B) ELISA with recombinant SARS-CoV-2 Spike protein-coated plates with serum dilutions of 1:100 and 1:300, (C,D) ELISA with SARS-CoV-2 binding domain recombinant protein-coated plates with serum dilutions of 1:100 and 1:300. The comparison and statistics were done by unpaired Mann–Whitney t-tests with a 95% confidence level; 0.12 (ns), 0.033 (), and 0.002 (). The error bars represent ± SD. Levels of antibodies produced by DEVs and DCs at all time points were nonsignificant compared to the levels of the Spike antibody. Each bar graph represents data from five mouse serum runs in duplicate. DEV-Treated Mouse Sera Inhibit SARS-CoV-2 Infection Sera from treated mice were used to assess the neutralization of wild-type SARS-CoV-2 USA/WA1/2020. For the virus inhibition assay, the SARS-CoV-2 virus was incubated with heat-inactivated sera obtained from all the groups followed by incubation with Vero E6 cells. Twenty-four hours post-infection, cells were fixed and stained with a monoclonal mouse antibody against SARS-CoV-2 nucleocapsids. Total and SARS-CoV-2 nucleocapsid protein-positive cells (in 16 image fields per well) were counted and analyzed by multiple wavelength cell sorting algorithms to obtain the percentage of infected cells. The representative images of SARS-CoV-2-infected Vero E6 cells after treatment with sera from PBS-, Spike-, DC-, or DEV-treated mice are shown in Figure S3. Figure 3A demonstrates that SARS-CoV-2 replication was dramatically reduced up to 45% with the sera of mice treated (1:50 dilution) with a 1/10 Spike equivalent dose of DEVs at day 14 (p < 0.03 vs DC treatment). In contrast, sera from DC- or Spike-treated mice were able to reduce only ∼20 and 35% of SARS-CoV-2 replication, respectively. Even with a dilution of 1:100, DEV-treated sera were still able to reduce viral infection with better efficacy at early time points than Spike- or DC-treated mouse sera (Figure 3B). The results confirmed that DEV treatment could induce a superior antibody response against Spike protein among all the treated groups. The reduction was further observed with DEV-treated mouse sera after 15 days of the second dose of administration (p < 0.03 and p < 0.002 vs DC and Spike treatment, respectively, n = 5 in each group). However, as the days progressed, the ability of DEV-induced antibodies to reduce viral infection was reduced by 10–15% and became comparable with the reduction observed with sera of mice treated with DCs or Spike (Figure 3A,B). Figure 3 Inhibition of SARS-CoV-2 infection by DEV-induced antibodies in mouse sera. Microneutralization of SARS-CoV-2 with the mouse serum was performed in Vero E6 cells. The data were compared with the percentage inhibition with reference to the 1× PBS-treated mice (negative control). The mouse serum was added (A) in 1:50 dilutions and (B) in 1:100 dilutions. The comparison and statistics were done by one-way ANOVA with Dunnett’s multiple comparison analysis at 95% CI (P < 0.002; P < 0.03, n = 5 in each group). The error bars represent ± SD. Each bar graph represents data from five mouse serum runs in triplicate. DEVs Induce IL-2, IFN-γ, and IL-6 Levels in Mouse Serum In response to stress-inducing viral infection, host cells secrete different cytokines by reprogramming cell metabolism. The different cytokines, including those involved in adaptive immunity (e.g., IL-2), proinflammatory cytokines, and interleukins (ILs) (e.g., interferon IFN-γ and IL-6), comprise a vital role in defensive response.19 In this study, the levels of IL-2, IFN-γ, and IL-6 were measured at days 7 and 15 after treatment administration. As mentioned in Figure 4A, the serum IL-2 level was significantly increased in mice treated with DEVs compared to those treated with DCs and Spike protein (DEVs vs Spike at day 7, p = 0.0082; DEVs vs Spike or DCs at day 15, p ≤ 0.0005, n = 3). Figure 4 Serum cytokine levels after vaccination. After vaccination, blood was collected at days 7 and 14 from all the groups and cytokine levels in sera; IL-2 (A), IFN-γ (B), and IL-6 (C) were analyzed by ELISA. The results were presented as the mean ± 95% CI, n = 3. P values for the DEV-treated group relative to Spike protein are **P < 0.0001, P < 0.01, and P < 0.05 or for DC groups are ###P < 0.001 and ##P < 0.01. Next, we determined the serum IFN-γ level as it is highly involved in adaptive immune response and inflammatory processes. It helps in the activation of immune cells, including T-cells and macrophages, and enhances their antigen presentation. It also has a crucial role in antimicrobial immunity.19 The results demonstrated that the serum IFN-γ level was significantly elevated at day 7 and further increased at day 15 in mice treated with DEVs compared to mice treated with PBS (Figure 4B) (DEVs vs PBS for both days 7 and 15, p < 0.0001, n = 3). The serum IFN-γ level on day 7 of mice treated with a 1/10 Spike protein equivalent dose of DEVs was comparable with that of mice treated with DCs and Spike protein; however, it was significantly increased at day 15 as compared to Spike-treated mice (Figure 4B) (DEVs vs Spike at day 15, p ≤ 0.05, n = 3). Several studies indicated that IL-6 plays an important role in Tfh cell differentiation, and the production of IL-6 by different T-cell types after SARS-CoV-2 infection is beneficial for resolving the viral infection.20,21 As shown in Figure 4C, at day 7, IL-6 was not elevated in the serum of mice for all groups. However, on day 15, the IL-6 level was significantly elevated in the serum obtained from mice treated with DEVs (Figure 4C) (DEVs vs DCs or Spike at day 15, p ≤ 0.005, n = 3). Most interestingly, at day 15, IL-6 levels in the serum of mice treated with DCs and Spike protein did not elevate from day 7 and are comparable to the mice that received PBS only (Figure 4C). DEV Treatment Elevates Different Subsets of T and B Cells Next, we evaluated the effect of DEVs on the immune cell population in the spleen, lymph node (LN), and bone marrow (BM). In clinics, CD4+ and CD8+ T-cell counts in patients indicate the severity of COVID-19 infection and predict their clinical outcomes.22 On day 90, all mice were sacrificed, and cells were harvested from the spleen, LN, and BM to analyze the population of T-cells and B cells. A significant increase in CD4+ (∼25% vs 16% in the control) T-cell population was observed in the spleen of mice treated with a 1/10 Spike equivalent dose of DEVs, DC, or free Spike protein as compared to control (PBS)-treated mice (Figure 5A right panel and Figure S4). However, in the LN, there was no significant change in the CD4+ T-cell population. Additionally, the CD8+ T-cell population was significantly elevated in the bone marrow (35% in DEVs vs 26% in the control), spleen (36% in DEVs vs 16% in the control), and LN (49% in DEVs vs 27% in the control) of the DEV-treated mice when compared to PBS-treated mice (Figure 5B and Figure S4). Compared to Spike, DC-treated mice had increased levels of CD8+ cells in the bone marrow, whereas DEV-treated mice had a higher number of cells in the spleen compared to Spike-treated mice. Figure 5 Induction of different T-cell subsets and memory B cells in the secondary lymphoid organ. Immune cell abundance in the bone marrow, spleen, and lymph node of the experimental groups: Gr-1 control, PBS-treated (n = 3); Gr-2 Spike-pulsed dendritic cells (DCs) (n = 5); Gr-3 DEVs (n = 5); and Gr-4 Spike protein (n = 5) on day 90 post-treatment. Leukocytes enriched from the bone marrow, lymph node, and spleen were subjected to flow cytometric analysis. The bar graph indicates the quantification of (A) CD4+ cells, (B) CD8+ cells, (C) CD4+CXCR5+ cells, and (D) CD4+CD25+ cells in the spleen and lymph node in the left and right panel, respectively. (E) Memory B cells were assessed by analyzing the CD19+CD38+ population in the spleen and lymph node in the left and right panel, respectively. Each dot in the bar graph represents cells from an individual mouse. Data are represented as the mean ± SD (p < 0.01; **p < 0.001). CD4+CD25+ T-cells, naturally occurring T-regulatory cells (Tregs), are crucial for maintaining immune homeostasis and preventing excessive inflammation.23 However, a dominant Tregs population might lead to suppression of immune cell activation; hence, a balance is required for an effective immune response against any disease.24 To this end, the population of CD4+CD25+ cells was quantified, and Figure S5 demonstrates that the percentage of CD4+CD25+ cells was in a range of 2–12% of total CD4+ T-cells in both the spleen and LN in all the groups, which are similar to the reported population in healthy mice or humans in homeostasis conditions.25 As compared to Spike, DEV-treated mice had an increased number of CD4+CD25+ cells, indicating a better potential of DEVs in inducing Treg population. Tfh cells, characterized by the expression of CXCR5, are critical for antibody affinity maturation and production of memory B cells and long-lived plasma cells.26 In this study, we also found that the population of CD4+CXCR5+ T-cells was increased compared to the control in the spleen and LN of mice treated with a 1/10 Spike protein equivalent dose of DEVs and is comparable to the mice treated with DCs and Spike protein (Figure 5D). Next, the number of CD19+CD38+ B cells was determined, as, upon recognition of the antigen, murine B cells get activated and differentiate into CD19+CD38+ memory cells. It was found that the DEV treatment was equally effective in inducing CD19+CD38+ cells in both the spleen and lymph node (Figure 5 and Figure S4). Figure 5E demonstrates that the population of CD19+CD38+ B cells was significantly increased in the spleen (58% vs 45% in the control) and LN (35% vs 20% in the control) of mice treated with DEVs when compared to the PBS-treated group. Our data also revealed that a 1/10 Spike equivalent dose of DEVs could induce memory B cells similar to DCs and Spike protein (Figure 5E). Representative dot plots of the acquired cells are shown in Figure S4, and the mean fluorescence intensities of all the cell populations are represented in Tables S1–S9. Thus, these data demonstrate that DEV treatment has the potential to induce various subsets of immune cells in mice. Discussion Here, we aimed at developing Spike-activated DEVs presenting MHC-peptide complexes on the surface that elicit T-cell and memory B cell responses against SARS-CoV-2. The results depicted that MHC class I, CD86, and CCR-7 markers are expressed on the surface of DEVs. DEVs also induce Spike protein-specific antibodies in the injected mice, indicating that they carry peptides of the Spike proteins. Several nanotechnology-based vaccines, including extracellular vesicles, have recently shown antiviral or neutralizing antibody responses against SARS-CoV-2 spike protein.18,27−29 However, none of those studies have used dendritic cells as tools to engineer extracellular vesicles, which is conceptually novel and different from the existing literature. DEVs have many advantages over conventional vesicles: apart from generating a strong neutralizing response, DEVs in the study contain processed antigens present on MHCs, which can directly activate T-cells and other immune cells. Furthermore, the CCR-7 receptor present on DEVs helps in the trafficking of cargos toward secondary lymphoid organs. The process can additionally boost the activation of T-cells in lymph nodes. Further, in vivo assays demonstrated that subcutaneous administration of both Spike and DEVs in mice appeared safe and well-tolerated. Sera from DEV-injected mice enhanced the Spike protein and the RBD-specific antibody titer. A previous study reported that infectivity of the patient plasma in the in vitro neutralization assays of SARS-CoV-2 provides an important indication of the effectiveness of humoral response against specific viral variants.30 Our study also indicated that DEVs were able to induce the Spike protein and RBD-specific antibodies in the sera of mice. It also reduced the infection of SARS-CoV-2 viruses to the Vero E6 cells. The data suggest that DEV-treated mouse sera can neutralize the virus. In the case of respiratory syncytial virus (RSV), engineered extracellular vesicles were found to be immunogenic and safe.31 Extracellular vesicles derived from DCs treated with IFN-α have a protective effect against DENV infection in other cells, confirming potential tools to counteract viral infection.32 Previous reports demonstrated the durability of the immunization by analyzing the population of various immune cells even beyond 90 days.33−36 In this study, we also wanted to check the durability of the immune response and assess whether this vaccination can provide long-term immunity against SARS-CoV-2. Interestingly, an increase in the CD8+ and CD4+ T-cell populations was observed after 90 days of the first dose of DEV administration. Again, the CD4+CD25+ Tregs population is a critical factor as they control inflammation and also help in maintaining immune homeostasis.23 Our results depicted that the CD4+CD25+ T-cell population was in the range of 2–12% in all the groups, which is similar to the Tregs population in healthy humans or mice (6–12%).24 When the virus enters the body, naïve antigen-specific CD4+ T-cells generally differentiate into two subsets of effector CD4 T-cells: Th1 cells and Tfh cells.37 Th1 cells regulate and control viral infections by producing different cytokines, such as IFN-γ and TNF-a.38 However, in secondary lymphoid organs, antigen-specific conventional DCs or monocyte-derived DCs initiate the differentiation of naïve CD4+ T-cells into Tfh cells by the upregulation of CXCR5.39 Tfh cells have a vital role in inducing germinal center reactions and helping in antibody secretion by promoting cognate B cell differentiation.20 Interestingly, a significant increase in the CD4+CXCR5+ T-cell count was observed in mice treated with DEVs, indicating the differentiation of T-cells into Tfh cells upon vaccination. The mainstay for vaccine efficacy is the generation of memory B cells, which provides protective immunity upon pathogen encounter and generates plasma cells capable of producing Spike-specific antibodies.40 Herein, we found that the CD19+CD38+ B cell population was enhanced in the spleens and lymph nodes of mice treated with DEVs. A similar pattern was also observed in mice treated with DCs or Spike protein, suggesting that DEVs are immunogenic and can induce memory B cells. However, the efficacy of these memory B cells and CD8 T-cells needs further studies in the virus challenge model. IL-2 plays an important role in T-cell proliferation and the generation of effector and memory T-cells.41 It is involved in adaptive immunity and increases the proliferation and activation of T, B, and NK cells.42 Similarly, IFN-γ, which is secreted by a wide variety of lymphocytic cells, including CD4+ and CD8+ T-cells, B cells, and NK cells, is involved in many innate and adaptive immune processes. It also stimulates macrophage activation and antigen presentation and is significantly involved in immunity against bacteria and viruses.43 Our study demonstrated that serum IL-2 and IFN-γ levels significantly increased in mice treated with DEVs. Again, IL-6 upregulates the STAT1/3-Bcl-6 signal and plays a pivotal role in Tfh cell differentiation. Moreover, studies have related the dysregulation of IL-6 to the COVID-19 progression and associated disease severity, respiratory failure, and mortality in patients.44,45 Interestingly, optimum production of IL-6 and other proinflammatory markers by various cell types after SARS-CoV-2 infection provided a positive indication in resolving the viral infection.21 It is important to note that although DEVs induce antibodies at early time points, however, the antibody level becomes equivalent at later stages, as observed with DCs and Spike. A similar observation was also correlated with the ability to neutralize the virus in an in vitro model. A recent study showed that targeting antigens to the surface of EVs improves the in vivo immunogenicity of human and nonhuman adenoviral vaccines in mice.46 Notably, the dose of DEVs was equivalent to 1/10 of Spike protein, indicating that DEVs are more potent in inducing antibody generation. However, further study is warranted to compare the efficacy of DEVs with commercially available vaccines in terms of antibody production, stability, and capacity to neutralize the virus. In conclusion, DEVs can express SARS-CoV-2-specific peptides on the surface and induce a humoral response by producing specific antibodies, which induces virus neutralization in vitro. Moreover, DEVs also induce an adaptive immune response by secreting serum IL-2, IFN-γ, and IL-6 and inducing the differentiation of various T-cell subsets, including CD8+, CD4+ T-cells, and Tfh cells (CD4+CXCR5+). Noticeably, it elevated the generation of memory B cells (CD19+CD38+) in the secondary lymphoid organ, which might impart long-lasting protection from viral infection. Hence, a 1/10 Spike protein equivalent dose of DEVs was able to induce a comparable level of humoral and cellular immunity, therefore warranting further investigations. Materials and Methods Mice Experiments were performed in C57BL/6 mice. Mice were housed in a 12:12 light:dark cycle with constant temperature and humidity and ad libitum access to food and water. All procedures followed the guidelines of the Committee for Control and Supervision of Experiments on Animals, Ministry of Environment and Forestry, Government of India, and were approved by the Regional Centre for Biotechnology ethics committee approval no. RCB/IAEC/2019/060. Culture of Bone Marrow-Derived Dendritic Cells and Isolation of DEVs Femur and tibia bones from 8-weeks-old C57BL/6 mice were isolated, and the bone marrow was flushed out into cold DMEM (cat. no. A007S, Himedia). The bone marrow was crushed with the back of a needle and passed through a 70 μm strainer to make a single-cell suspension. The cells were plated at a density of 2 × 105 cells/mL of DMEM with 10% FBS and 20 ng/mL GM-CSF and 10 ng/mL IL-4 (cat. nos. 576306 and 574306, respectively) in a Petri dish and cultured for seven days at 37 °C in a humidified CO2 incubator. On days 3 and 5, the media was replaced with fresh DMEM supplemented with 10 ng/mL GM-CSF and 10 ng/mL IL-4. On day 7, the immature DCs were activated with recombinant Spike protein and a cytokine cocktail having LPS (cat. no. L2654, Sigma Aldrich), IFN-γ, IL-6, and IL-12 (cat. nos. 575306, 575704, and 577004, respectively, from BioLegend, USA) for 24 h. On day 8, fresh extracellular vesicle-depleted DMEM was added to the BMDCs. After 16 h, the media was collected and centrifuged at 300g for 10 min followed by 3000g for 30 min to remove debris and dead cells. The microparticles were removed by centrifuging the supernatant at 10,000g for 30 min. The extracellular vesicles from the supernatant were purified using an Amicon cell with a 500 kDa filter membrane (cat. no. PBVK06210, Millipore), concentrated in PBS, and stored at −80 °C for further use. For the activation, 5 μg of Spike protein was added to 106 DCs. DEVs (200 μg) were isolated from 106 DCs. We finally injected 20 μg of DEVs/mouse, which was equivalent to 1/10 of DCs (or 1/10 of 5 μg Spike protein used). Nanotracking Analysis and Scanning Electron Microscopy Purified DEVs were analyzed by nanotracking analysis (NTA) using a nanosight model LM14 (United Kingdom). DEVs were diluted in PBS, and a few 90 s videos were recorded using camera levels 13 and 14. The detection threshold was optimized for the sample, and the data were analyzed using NTA software 2.3 with screen gain at 10 to track all possible particles with minimal background. The shape and sizes of the DEVs were confirmed by a scanning electron microscope (SEM). Briefly, DEVs were diluted 100 times with PBS and drop-casted on a glass substrate followed by fixing with 2% glutaraldehyde in PBS for 10 min. Postfixation, the sample was washed with an increasing concentration of ethanol and dried at room temperature for 1 h. Finally, the DEVs were analyzed by SEM (Zeiss EBO-50, Germany) after gold–palladium sputtering. Cryo-Transmission Electron Microscopy DEVs in PBS (2 μL) were drop-casted on a carbon grid and immediately plunge-frozen in liquid nitrogen. The grid was then placed into an FEI Tecnai F20 G2 FEG TEM at 200 kV using a cryo-TEM sample holder (Gatan model 626). Images were analyzed on a 4K × 4K CCD camera (FEI Eagle) using TIA software. Transmission Electron Microscopy A DEV suspension was placed on Formvar/carbon-coated copper grids and allowed to adsorb for 5 min in a dry environment. The grid was then washed in RNAase-free water, stained with 1.5% phosphotungstic acid for 1 min, and placed in a clean environment for drying. The grid was observed at the 80k threshold under electron microscopy (Tecnai, FEI, Hillsboro, OR, USA). Western Blot Analysis DCs or DEVs were lysed by incubating them in RIPA buffer for 20 min followed by centrifugation at 12,000 rpm for 15 min. The protein concentration in the supernatant was quantified by a BCA assay. DC and DEV proteins (10 μg) were separated using SDS polyacrylamide gel electrophoresis and transferred onto a polyvinylidene fluoride (PVDF) membrane. The membranes were blocked with 3% bovine serum albumin in tris-buffered saline with tween 20 (TBST) followed by incubation with anti CD63, Alix, CD86, MHC-I, and CCR-7 antibodies (cat. nos. 143902, 634502, 105002, 110307, and 120107; BioLegend, USA). Binding was then detected with HRP-conjugated secondary antibodies. Blots were developed using enhanced chemiluminescence (ECL) (cat. no. RPN2209 Cytiva, United States) reagents and visualized by a G:Box Chemi XRQ imaging system (Syngene USA, Inc.). Administration of DEVs in C57BL/6 Mice The immunogenicity and efficacy of the DEVs were tested in mice and compared against Spike protein. Twenty C57BL/6 mice (3–4-weeks-old) were randomly divided into four groups having five mice in each group: Gr-1 control, PBS-treated; Gr-2 Spike-pulsed dendritic cells (DCs); Gr-3 DEVs; and Gr-4 Spike protein. Corresponding groups got two injections subcutaneously at day 0 and day 15 with filtered PBS, DCs (0.5 × 106 cells/mouse), 20 μg of exosomal protein equivalent DEVs/mouse, and Spike protein (5 μg/mice). Sera post-immunization were collected on days 14, 30, 60, and 90. Determination of Binding Affinity to Spike and RBD ELISA Nunc MaxiSorp plates (Thermo Scientific) were coated with recombinant soluble Spike protein, nonfatty milk, and the receptor-binding domain (RBD) of SARS-CoV-2 (the schematic is presented in Figure S6). The Spike protein and the RBD were purified by expressing the codon-optimized sequence of the respective proteins cloned in pCDNA3.1. The proteins were expressed and purified in the mammalian Expi 293 F system as per the literature.47,48 The 96-well plates were coated with Spike or RBD protein at a concentration of 2 μg per well at 4 °C overnight. The next day, the wells were blocked with 5% (w/v) nonfatty milk in 1× PBS (phosphate buffer) at 37 °C for 1.5 h. The wells were washed three times with 1× PBS before adding the serum samples diluted in 5% (w/v) nonfatty milk in 1× PBS (1:100 and 1:300 dilutions). The dilutions were added to the wells in technical duplicates and incubated for 1 h at room temperature. Each sample was added to the Spike-coated wells, RBD-coated wells, and wells having the negative control (nonfatty milk). After rinsing six times with 0.05% PBST, the wells were incubated with HRP-conjugated goat antimouse IgG (Invitrogen G-21040; 1:10,000 diluted in PBST) at room temperature for 1 h. A TMB substrate (3,3′,5,5′-tetramethylbenzidine) (Life Technologies, 002023) was added and incubated for 20 min after washing six times with 0.05% PBST. The plate was read for A570 and A450 using an ELISA plate reader (xMark microplate spectrophotometer, BIO-RAD), and the final values were plotted after wavelength correction. Cells Vero E6 cells were cultured in complete modified Eagle medium (MEM; Himedia AL047S) containing 10% fetal bovine serum (FBS; Gibco, 10270-106) supplemented with 100 U/mL penicillin and 100 mg/mL streptomycin (Himedia, A001A-100 mL). Vero E6 cells, obtained from NCCS Pune, were independently validated in our laboratories and routinely tested for mycoplasma contamination. SARS-CoV-2 Virus Strain Culture and Propagation The SARS-CoV-2 virus USA/WA1/2020 was obtained from the World Reference Center for Emerging Viruses and Arboviruses (WRCEVA, UTMB), The University of Texas Medical Branch, and used as a wild-type reference.49 Viruses were grown in Vero E6 cells for 3 days; the supernatant was clarified by centrifugation at 3800g for 15 min, and aliquots were frozen at −80 °C for long-term use. Expanded viral stocks were titrated on Vero E6 cells before using in microneutralization assays. A concentration of 3 × 106/mL of the viral titer was used for the assay. SARS-CoV-2 Microneutralization Assay Sera from treated mice were used to assess the neutralization of wild-type SARS-CoV-2. All procedures were performed in the BSL3 facility at the NCR-Biotech Science Cluster, following standard safety guidelines. On the day before infection, Vero E6 cells were seeded in 96-well high-binding tissue culture plates (Corning cat. no. 3904) at a density of 10,000 cells/well in complete modified Eagle medium (MEM; cat. no. AL047S). The serum samples were heat-inactivated at 56 °C for 15 min. The samples were diluted in MEM at 1:50 and 1:100 dilutions in technical duplicates. The SARS-CoV-2 virus USA/WA1/2020 isolate was added at 0.05 MOI per well and incubated at 37 °C for 1 h in a biosafety level 3 lab for neutralization. The negative control (serum samples from 1× PBS-immunized mice) and positive control (MonoRab BS-R2B2 and 4G6 from Genscript) neutralization antibodies against SARS-CoV-2 were also included in the 96-well plates for the neutralization assay. Postneutralization, the combination of the diluted serum sample and the virus was added to Vero E6 cells, which were grown overnight in 96-well plates (Corning clear bottom cat. no. 3904) with 10% FBS and complete MEM and incubated for 1 h in a biosafety level 3 lab. After an hour, the serum–virus mixture was aspirated, and complete MEM with 2% FBS was added to the cells and incubated for 24 h before fixing the cells with 4% paraformaldehyde. The fixed cells were washed three times with 1× PBS, permeabilized with 0.3% Tween 20, and blocked with 2.5% BSA in 1× PBS. Next, the cells were incubated with a monoclonal mouse antibody against SARS-CoV-2 nucleocapsids (Sino Biologicals 40143-05) followed by a secondary antibody (goat antimouse Alexa 568 conjugated A11031 Invitrogen). Before imaging, Hoechst 33342 (cat. no. 62249, Thermo Scientific) was diluted in 1× PBS and added to the cells. The cells were imaged in a micro confocal high-content imaging system: ImageXpress, by Molecular Devices in DAPI and Texas Red channel laser settings and a 10× Plan Apo NA 0.45 objective with a correction collar. The data were analyzed by the multiwavelength cell scoring algorithm in fast mode analysis. The percent positive cells for the Texas Red signal (against the DAPI count) were obtained and normalized with the control. Isolation and Characterization of Immune Cells from Tissues for Flow Cytometry Analysis Mice were euthanized on day 90 post-DEV administration by isofluorane inhalation. The spleen, lymph nodes, femur, and tibia were quickly removed in RPMI media supplemented with 10% FBS. The single-cell suspension was prepared by grinding and filtering the spleen and lymph nodes through a 70 μm diameter nylon mesh (BD Bioscience). The femur and tibia were flushed with RPMI media supplemented with 10% FBS and filtered through a 70 μm diameter nylon mesh to collect bone marrow-derived cells. Single cells were treated with ACK lysis buffer (BioLegend, Biolegend Way, San Diego, CA) for red blood cell lysis as per manufacturer protocol. To identify T and B cells, single cells from the spleen, lymph node, and bone marrow were first preincubated with the antimouse CD16/32 antibody TruStain FcX as per manufacturer protocol (BioLegend, San Diego, CA). The procedure was performed for 15 min at 4 °C to block Fc receptors and then simultaneously stained with fluorochrome-conjugated antibodies (FITC-CD45, [at a dilution of 1:100; Miltenyi Biotec, Gladbach, Germany], PE-CD38 [at a dilution of 1:50; BioLegend, San Diego, CA, USA], and APC-CD19 [at a dilution of 1:50; BioLegend, San Diego, CA, USA]. For T-cell immune profiling, all the antibodies were procured from BioLegend (FITC-CD3 [at a dilution of 1.5:50], APC-CD4 [at a dilution of 1:50], Brilliant Violet 421-CXCR5 [at a dilution of 1:50], FITC-CD25 [at a dilution of 1.5:50], and APC-CD8 [at dilution 1:25]). Cells were then rinsed with FACS buffer and run on a BD FACSVerse (BD Biosciences, San Jose, CA). Data were analyzed using BD FACS Suite v1.0.6 (BD Biosciences, San Jose, CA) and FlowJo v10 (FlowJo LLC). Statistical Analysis The results show the mean value ± 95% CI or ± SD. Statistical analysis was carried out in GraphPad Prism 6, using unpaired t-tests with one-way ANOVA and one-way ANOVA with Dunnett’s multiple comparison to calculate P values. Tukey’s multiple comparison tests analyzed the significance of the difference between treatments. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsbiomaterials.2c01094.TEM analysis of DEVs stored in PBS at −80 °C (Figure S1A); Western blot data indicating the presence of surface markers (CCR7 and MHC-I) in old and fresh DEVs (Figure S1B); cytotoxicity effect of DEVs on RAW 264.7 macrophage cells (Figure S2A); body weight of mice throughout the experiment (Figure S2B); representative confocal images of the microneutralization assay (Figure S3); representative dot plots showing cell populations (Figure S4); percentage of Tregs in the spleen and lymph node (Figure S5); schematic workflow of ELISA (Figure S6); mean fluorescence intensities of all the cell populations (Tables S1–S9) (PDF) Supplementary Material ab2c01094_si_001.pdf Author Contributions # A.B. and B.B. equally contributed. Author Contributions J.B., A.B., and S.V. conceived the study and designed the experiments. A.B. performed primary DC culture, DEV isolation, and biophysical characterization, while J.B. provided guidance. B.B. and D.M. were involved in all BSL3 work related to cell culture. B.B. was involved in SARS-CoV-2 virus propagation, titration, and microneutralization assays. R.K. provided critical reagent Spike and RBD used for the ELISA assay, and B.B. and R.K. were involved in the antibody ELISA assay. N.S. and A.T. performed mouse experiments and immune cell isolation from different tissue sources and performed FACS analysis. J.B., A.B., and S.V. wrote the manuscript, and all authors contributed to editing the document. J.B. and S.V. secured funding for the study. All authors have read and agreed to the published version of the manuscript. The research work was supported through the project (grant nos. BT/PR15984/MED/31/325/2015 and BT/PR40401/COT/142/17/2020) sanctioned to A.B. and S.V., respectively, funded by the Department of Biotechnology, Government of India. This work was supported by the Indian Institute of Technology Delhi (no. MI02233G) to J.B. On a student’s fellowship, B.B. was supported by CSIR, India. N.S. was supported by ICMR, India (fellowship award no. 2020-6132/SCR-BMS). A.T. was supported by the SERB, DST, India project, awarded to A.B. (grant no. #EMR/2017/001490). The authors declare no competing financial interest. Acknowledgments A.B. acknowledges IIT Delhi for providing her doctoral fellowship. ==== Refs References Kim Y. C. ; Dema B. ; Reyes-Sandoval A. COVID-19 vaccines: breaking record times to first-in-human trials. NPJ Vaccines 2020, 5 , 34 10.1038/s41541-020-0188-3.32377399 Huang W. C. ; Zhou S. ; He X. ; Chiem K. ; Mabrouk M. T. ; Nissly R. H. ; Bird I. M. ; Strauss M. ; Sambhara S. ; Ortega J. ; Wohlfert E. A. ; Martinez-Sobrido L. ; Kuchipudi S. V. ; Davidson B. A. ; Lovell J. F. SARS-CoV-2 RBD Neutralizing Antibody Induction is Enhanced by Particulate Vaccination. Adv. Mater. 2020, 32 , e2005637 10.1002/adma.202005637.33111375 Mabrouk M. T. ; Huang W. C. ; Martinez-Sobrido L. ; Lovell J. F. Advanced Materials for SARS-CoV-2 Vaccines. Adv. Mater. 2022, 2107781 10.1002/adma.202107781.34894000 Zhou Q. ; Gu H. ; Sun S. ; Zhang Y. ; Hou Y. ; Li C. ; Zhao Y. ; Ma P. ; Lv L. ; Aji S. ; Sun S. ; Wang X. ; Zhan L. Large-Sized Graphene Oxide Nanosheets Increase DC-T-Cell Synaptic Contact and the Efficacy of DC Vaccines against SARS-CoV-2. Adv. Mater. 2021, 33 , 2102528 10.1002/adma.202102528.34396603 Gale E. C. ; Powell A. E. ; Roth G. A. ; Meany E. L. ; Yan J. ; Ou B. S. ; Grosskopf A. K. ; Adamska J. ; Picece V. C. T. M. ; D’Aquino A. I. ; Pulendran B. ; Kim P. S. ; Appel E. A. Hydrogel-based slow release of a receptor-binding domain subunit vaccine elicits neutralizing antibody responses against SARS-CoV-2. Adv. Mater. 2021, 2104362 10.1002/adma.202104362.34651342 Pitt J. M. ; André F. ; Amigorena S. ; Soria J.-C. ; Eggermont A. ; Kroemer G. ; Zitvogel L. Dendritic cell-derived exosomes for cancer therapy. J. Clin. Invest. 2016, 126 , 1224–1232. 10.1172/JCI81137.27035813 Andre F. ; Escudier B. ; Angevin E. ; Tursz T. ; Zitvogel L. Exosomes for cancer immunotherapy. Ann. Oncol. 2004, 15 , iv141-4 10.1093/annonc/mdh918.15477298 Kamath A. T. ; Henri S. ; Battye F. ; Tough D. F. ; Shortman K. Developmental kinetics and lifespan of dendritic cells in mouse lymphoid organs. Blood 2002, 100 , 1734–1741. 10.1182/blood.V100.5.1734.h81702001734_1734_1741.12176895 Ingulli E. ; Mondino A. ; Khoruts A. ; Jenkins M. K. In vivo detection of dendritic cell antigen presentation to CD4(+) T cells. J. Exp. Med. 1997, 185 , 2133–2141. 10.1084/jem.185.12.2133.9182685 Loyer V. ; Fontaine P. ; Pion S. ; Hétu F. ; Roy D.-C. ; Perreault C. The in vivo fate of APCs displaying minor H antigen and/or MHC differences is regulated by CTLs specific for immunodominant class I-associated epitopes. J. Immunol. 1999, 163 , 6462–6467. 10.4049/jimmunol.163.12.6462.10586037 Hermans I. F. ; Ritchie D. S. ; Yang J. ; Roberts J. M. ; Ronchese F. CD8+ T cell-dependent elimination of dendritic cells in vivo limits the induction of antitumor immunity. J. Immunol. 2000, 164 , 3095–3101. 10.4049/jimmunol.164.6.3095.10706699 Dinh P.-U. C. ; Paudel D. ; Brochu H. ; Popowski K. D. ; Gracieux M. C. ; Cores J. ; Huang K. ; Hensley M. T. ; Harrell E. ; Vandergriff A. C. ; George A. K. ; Barrio R. T. ; Hu S. ; Allen T. A. ; Blackburn K. ; Caranasos T. G. ; Peng X. ; Schnabel L. V. ; Adler K. B. ; Lobo L. J. ; Goshe M. B. ; Cheng K. Inhalation of lung spheroid cell secretome and exosomes promotes lung repair in pulmonary fibrosis. Nat. Commun. 2020, 11 , 1064 10.1038/s41467-020-14344-7.32111836 Ha D. ; Yang N. ; Nadithe V. Exosomes as therapeutic drug carriers and delivery vehicles across biological membranes: current perspectives and future challenges. Acta Pharm. Sin. B 2016, 6 , 287–296. 10.1016/j.apsb.2016.02.001.27471669 Ansari M. A. ; Thiruvengadam M. ; Venkidasamy B. ; Alomary M. N. ; Salawi A. ; Chung I. M. ; Shariati M. A. ; Rebezov M. Exosome-based nanomedicine for cancer treatment by targeting inflammatory pathways: Current status and future perspectives. Semin. Cancer Biol. 2022, 86 , 678–696. 10.1016/j.semcancer.2022.04.005.35452820 Beauvillain C. ; Ruiz S. ; Guiton R. ; Bout D. ; Dimier-Poisson I. A vaccine based on exosomes secreted by a dendritic cell line confers protection against T. gondii infection in syngeneic and allogeneic mice. Microbes Infect. 2007, 9 , 1614–1622. 10.1016/j.micinf.2007.07.002.17905628 Luketic L. ; Delanghe J. ; Sobol P. T. ; Yang P. ; Frotten E. ; Mossman K. L. ; Gauldie J. ; Bramson J. ; Wan Y. Antigen presentation by exosomes released from peptide-pulsed dendritic cells is not suppressed by the presence of active CTL. J. Immunol. 2007, 179 , 5024–5032. 10.4049/jimmunol.179.8.5024.17911587 Bansal S. ; Perincheri S. ; Fleming T. ; Poulson C. ; Tiffany B. ; Bremner R. M. ; Mohanakumar T. Cutting Edge: Circulating Exosomes with COVID Spike Protein Are Induced by BNT162b2 (Pfizer-BioNTech) Vaccination prior to Development of Antibodies: A Novel Mechanism for Immune Activation by mRNA Vaccines. J. Immunol. 2021, 207 , 2405–2410. 10.4049/jimmunol.2100637.34654691 Ferrantelli F. ; Chiozzini C. ; Manfredi F. ; Leone P. ; Spada M. ; Di Virgilio A. ; Giovannelli A. ; Sanchez M. ; Cara A. ; Michelini Z. ; Federico M. Strong SARS-CoV-2 N-Specific CD8(+) T Immunity Induced by Engineered Extracellular Vesicles Associates with Protection from Lethal Infection in Mice. Viruses 2022, 14 , 329 10.3390/v14020329.35215922 Vabret N. ; Britton G. J. ; Gruber C. ; Hegde S. ; Kim J. ; Kuksin M. ; Levantovsky R. ; Malle L. ; Moreira A. ; Park M. D. ; Pia L. ; Risson E. ; Saffern M. ; Salomé B. ; Selvan M. E. ; Spindler M. P. ; Tan J. ; van der Heide V. ; Gregory J. K. ; Alexandropoulos K. ; Bhardwaj N. ; Brown B. D. ; Greenbaum B. ; Gümüş Z. H. ; Homann D. ; Horowitz A. ; Kamphorst A. O. ; Curotto de Lafaille M. A. ; Mehandru S. ; Merad M. ; Samstein R. M. ; Immunology of COVID-19: Current State of the Science. Immunity 2020, 52 , 910–941. 10.1016/j.immuni.2020.05.002.32505227 Cui D. ; Tang Y. ; Jiang Q. ; Jiang D. ; Zhang Y. ; Lv Y. ; Xu D. ; Wu J. ; Xie J. ; Wen C. ; Lu L. Follicular Helper T Cells in the Immunopathogenesis of SARS-CoV-2 Infection. Front. Immunol. 2021, 12 , 731100 10.3389/fimmu.2021.731100.34603308 Asri N. ; Nazemalhosseini Mojarad E. ; Mirjalali H. ; Mohebbi S. R. ; Baghaei K. ; Rostami-Nejad M. ; Yadegar A. ; Rezaei-Tavirani M. ; Asadzadeh Aghdaei H. ; Rostami K. ; Masotti A. Toward finding the difference between untreated celiac disease and COVID-19 infected patients in terms of CD4, CD25 (IL-2 Rα), FOXP3 and IL-6 expressions as genes affecting immune homeostasis. BMC Gastroenterol. 2021, 21 , 462 10.1186/s12876-021-02056-1.34895167 Liu Z. ; Long W. ; Tu M. ; Chen S. ; Huang Y. ; Wang S. ; Zhou W. ; Chen D. ; Zhou L. ; Wang M. ; Wu M. ; Huang Q. ; Xu H. ; Zeng W. ; Guo L. Lymphocyte subset (CD4+, CD8+) counts reflect the severity of infection and predict the clinical outcomes in patients with COVID-19. J. Infect. 2020, 81 , 318–356. 10.1016/j.jinf.2020.03.054. Sakaguchi S. ; Mikami N. ; Wing J. B. ; Tanaka A. ; Ichiyama K. ; Ohkura N. Regulatory T Cells and Human Disease. Annu. Rev. Immunol. 2020, 38 , 541–566. 10.1146/annurev-immunol-042718-041717.32017635 de Wolf A. C. M. T. ; van Aalst S. ; Ludwig I. S. ; Bodinham C. L. ; Lewis D. J. ; van der Zee R. ; van Eden W. ; Broere F. Regulatory T cell frequencies and phenotypes following anti-viral vaccination. PLoS One 2017, 12 , e0179942 10.1371/journal.pone.0179942.28658271 Wei S. ; Kryczek I. ; Zou W. Regulatory T-cell compartmentalization and trafficking. Blood 2006, 108 , 426–431. 10.1182/blood-2006-01-0177.16537800 Crotty S. Follicular helper CD4 T cells (TFH). Annu. Rev. Immunol. 2011, 29 , 621–663. 10.1146/annurev-immunol-031210-101400.21314428 Jiang L. ; Driedonks T. A. P. ; Jong W. S. P. ; Dhakal S. ; Bart van den Berg van Saparoea H. ; Sitaras I. ; Zhou R. ; Caputo C. ; Littlefield K. ; Lowman M. ; Chen M. ; Lima G. ; Gololobova O. ; Smith B. ; Mahairaki V. ; Riley Richardson M. ; Mulka K. R. ; Lane A. P. ; Klein S. L. ; Pekosz A. ; Brayton C. ; Mankowski J. L. ; Luirink J. ; Villano J. S. ; Witwer K. W. A bacterial extracellular vesicle-based intranasal vaccine against SARS-CoV-2 protects against disease and elicits neutralizing antibodies to wild-type and Delta variants. J. Extracell. Vesicles 2022, 11 , e12192 10.1002/jev2.12192.35289114 Scott T. A. ; Supramaniam A. ; Idris A. ; Cardoso A. A. ; Shrivastava S. ; Kelly G. ; Grepo N. A. ; Soemardy C. ; Ray R. M. ; McMillan N. A. J. ; Morris K. V. Engineered extracellular vesicles directed to the spike protein inhibit SARS-CoV-2. Mol. Ther.--Methods Clin. Dev. 2022, 24 , 355–366. 10.1016/j.omtm.2022.01.015.35127966 Staufer O. ; Gupta K. ; Hernandez Bücher J. E. ; Kohler F. ; Sigl C. ; Singh G. ; Vasileiou K. ; Yagüe Relimpio A. ; Macher M. ; Fabritz S. ; Dietz H. ; Cavalcanti Adam E. A. ; Schaffitzel C. ; Ruggieri A. ; Platzman I. ; Berger I. ; Spatz J. P. Synthetic virions reveal fatty acid-coupled adaptive immunogenicity of SARS-CoV-2 spike glycoprotein. Nat. Commun. 2022, 13 , 868 10.1038/s41467-022-28446-x.35165285 Khoury D. S. ; Wheatley A. K. ; Ramuta M. D. ; Reynaldi A. ; Cromer D. ; Subbarao K. ; O’Connor D. H. ; Kent S. J. ; Davenport M. P. Measuring immunity to SARS-CoV-2 infection: comparing assays and animal models. Nat. Rev. Immunol. 2020, 20 , 727–738. 10.1038/s41577-020-00471-1.33139888 Hong S. ; Ruan S. ; Greenberg Z. ; He M. ; McGill J. L. Development of surface engineered antigenic exosomes as vaccines for respiratory syncytial virus. Sci. Rep. 2021, 11 , 21358 10.1038/s41598-021-00765-x.34725399 Martins S. T. ; Kuczera D. ; Lötvall J. ; Bordignon J. ; Alves L. R. Characterization of Dendritic Cell-Derived Extracellular Vesicles During Dengue Virus Infection. Front. Microbiol. 2018, 9 , 1792 10.3389/fmicb.2018.01792.30131785 Nchinda G. ; Amadu D. ; Trumpfheller C. ; Mizenina O. ; Überla K. ; Steinman R. M. Dendritic cell targeted HIV gag protein vaccine provides help to a DNA vaccine including mobilization of protective CD8+ T cells. Proc. Natl. Acad. Sci. U. S. A. 2010, 107 , 4281–4286. 10.1073/pnas.1000621107.20160099 Gasper D. J. ; Neldner B. ; Plisch E. H. ; Rustom H. ; Carrow E. ; Imai H. ; Kawaoka Y. ; Suresh M. Effective Respiratory CD8 T-Cell Immunity to Influenza Virus Induced by Intranasal Carbomer-Lecithin-Adjuvanted Non-replicating Vaccines. PLoS Pathog. 2016, 12 , e1006064 10.1371/journal.ppat.1006064.27997610 Uddback I. E. M. ; Pedersen L. M. I. ; Pedersen S. R. ; Steffensen M. A. ; Holst P. J. ; Thomsen A. R. ; Christensen J. P. Combined local and systemic immunization is essential for durable T-cell mediated heterosubtypic immunity against influenza A virus. Sci. Rep. 2016, 6 , 20137 10.1038/srep20137.26831578 Zorgi N. E. ; Galisteo A. J. Jr. ; Sato M. N. ; do Nascimento N. ; de Andrade H. F. Jr. Immunity in the spleen and blood of mice immunized with irradiated Toxoplasma gondii tachyzoites. Med. Microbiol. Immunol. 2016, 205 , 297–314. 10.1007/s00430-015-0447-5.26732075 Hale J. S. ; Youngblood B. ; Latner D. R. ; Mohammed A. U. R. ; Ye L. ; Akondy R. S. ; Wu T. ; Iyer S. S. ; Ahmed R. Distinct memory CD4+ T cells with commitment to T follicular helper- and T helper 1-cell lineages are generated after acute viral infection. Immunity 2013, 38 , 805–817. 10.1016/j.immuni.2013.02.020.23583644 Swain S. L. ; McKinstry K. K. ; Strutt T. M. Expanding roles for CD4(+) T cells in immunity to viruses. Nat. Rev. Immunol. 2012, 12 , 136–148. 10.1038/nri3152.22266691 Crotty S. T Follicular Helper Cell Biology: A Decade of Discovery and Diseases. Immunity 2019, 50 , 1132–1148. 10.1016/j.immuni.2019.04.011.31117010 Ciabattini A. ; Pastore G. ; Fiorino F. ; Polvere J. ; Lucchesi S. ; Pettini E. ; Auddino S. ; Rancan I. ; Durante M. ; Miscia M. ; Rossetti B. ; Fabbiani M. ; Montagnani F. ; Medaglini D. Evidence of SARS-CoV-2-Specific Memory B Cells Six Months After Vaccination With the BNT162b2 mRNA Vaccine. Front. Immunol. 2021, 12 , 740708 10.3389/fimmu.2021.740708.34650563 Rosenberg S. A. IL-2: the first effective immunotherapy for human cancer. J. Immunol. 2014, 192 , 5451–5458. 10.4049/jimmunol.1490019.24907378 Turner M. D. ; Nedjai B. ; Hurst T. ; Pennington D. J. Cytokines and chemokines: At the crossroads of cell signalling and inflammatory disease. Biochim. Biophys. Acta 2014, 1843 , 2563–2582. 10.1016/j.bbamcr.2014.05.014.24892271 Costela-Ruiz V. J. ; Illescas-Montes R. ; Puerta-Puerta J. M. ; Ruiz C. ; Melguizo-Rodriguez L. SARS-CoV-2 infection: The role of cytokines in COVID-19 disease. Cytokine Growth Factor Rev. 2020, 54 , 62–75. 10.1016/j.cytogfr.2020.06.001.32513566 Nasonov E. ; Samsonov M. The role of Interleukin 6 inhibitors in therapy of severe COVID-19. Biomed. Pharmacother. 2020, 131 , 110698 10.1016/j.biopha.2020.110698.32920514 Grifoni E. ; Valoriani A. ; Cei F. ; Lamanna R. ; Gelli A. M. G. ; Ciambotti B. ; Vannucchi V. ; Moroni F. ; Pelagatti L. ; Tarquini R. ; Landini G. ; Vanni S. ; Masotti L. Interleukin-6 as prognosticator in patients with COVID-19. J. Infect. 2020, 81 , 452–482. 10.1016/j.jinf.2020.06.008. Bliss C. M. ; Parsons A. J. ; Nachbagauer R. ; Hamilton J. R. ; Cappuccini F. ; Ulaszewska M. ; Webber J. P. ; Clayton A. ; Hill A. V. S. ; Coughlan L. Targeting Antigen to the Surface of EVs Improves the In Vivo Immunogenicity of Human and Non-human Adenoviral Vaccines in Mice. Mol. Ther.--Methods Clin. Dev. 2020, 16 , 108–125. 10.1016/j.omtm.2019.12.003.31934599 Parray H. A. ; Chiranjivi A. K. ; Asthana S. ; Yadav N. ; Shrivastava T. ; Mani S. ; Sharma C. ; Vishwakarma P. ; Das S. ; Pindari K. ; Sinha S. ; Samal S. ; Ahmed S. ; Kumar R. Identification of an anti-SARS-CoV-2 receptor-binding domain-directed human monoclonal antibody from a naïve semisynthetic library. J. Biol. Chem. 2020, 295 , 12814–12821. 10.1074/jbc.AC120.014918.32727845 Murugavelu P. ; Perween R. ; Shrivastava T. ; Singh V. ; Ahmad Parray H. ; Singh S. ; Chiranjivi A. K. ; Thiruvengadam R. ; Singh S. ; Yadav N. ; Jakhar K. ; Sonar S. ; Mani S. ; Bhattacharyya S. ; Sharma C. ; Vishwakarma P. ; Khatri R. ; Kumar Panchal A. ; Das S. ; Ahmed S. ; Samal S. ; Kshetrapal P. ; Bhatnagar S. ; Luthra K. ; Kumar R. Non-neutralizing SARS CoV-2 directed polyclonal antibodies demonstrate cross-reactivity with the HA glycans of influenza virus. Int. Immunopharmacol. 2021, 99 , 108020 10.1016/j.intimp.2021.108020.34426117 Harcourt J. ; Tamin A. ; Lu X. ; Kamili S. ; Sakthivel S. K. ; Murray J. ; Queen K. ; Tao Y. ; Paden C. R. ; Zhang J. ; Li Y. ; Uehara A. ; Wang H. ; Goldsmith C. ; Bullock H. A. ; Wang L. ; Whitaker B. ; Lynch B. ; Gautam R. ; Schindewolf C. ; Lokugamage K. G. ; Scharton D. ; Plante J. A. ; Mirchandani D. ; Widen S. G. ; Narayanan K. ; Makino S. ; Ksiazek T. G. ; Plante K. S. ; Weaver S. C. ; Lindstrom S. ; Tong S. ; Menachery V. D. ; Thornburg N. J. Isolation and characterization of SARS-CoV-2 from the first US COVID-19 patient. bioRxiv 2020, 10.1101/2020.03.02.972935.	36445062	PMC9717688	NO-CC CODE	2022-12-03 23:21:05	no	ACS Biomater Sci Eng. 2022 Dec 12; 8(12):5338-5348	utf-8	ACS Biomater Sci Eng	2,022	10.1021/acsbiomaterials.2c01094	oa_other
==== Front Br J Anaesth Br J Anaesth BJA: British Journal of Anaesthesia 0007-0912 1471-6771 British Journal of Anaesthesia. Published by Elsevier Ltd. S0007-0912(22)00624-9 10.1016/j.bja.2022.10.035 Clinical Investigation Ventilatory ratio, dead space, and venous admixture in acute respiratory distress syndrome Maj Roberta 12† Palermo Paola 12† Gattarello Simone 12 Brusatori Serena 1 D’Albo Rosanna 1 Zinnato Carmelo 1 Velati Mara 12 Romitti Federica 1 Busana Mattia 1 Wieditz Johannes 1 Herrmann Peter 1 Moerer Onnen 1 Quintel Micheal 1 Meissner Konrad 1 Sanderson Barnaby 3 Chiumello Davide 4 Marini John J. 5 Camporota Luigi 3 Gattinoni Luciano 1∗ 1 Department of Anaesthesiology, Medical University of Göttingen, University Medical Centre Göttingen, Göttingen, Germany 2 Department of Anaesthesia and Intensive Care, IRCCS San Raffaele Scientific Institute, Milan, Italy 3 Department of Adult Critical Care, Guy’s and St Thomas’ NHS Foundation Trust, Health Centre for Human and Applied Physiological Sciences, London, UK 4 Department of Anaesthesiology and Intensive Care, ASST Santi Paolo e Carlo Hospital, University of Milan, Milan, Italy 5 Department of Pulmonary and Critical Care Medicine, Regions Hospital, St. Paul, MN, USA ∗ Corresponding author. † Contributed equally to the study. 2 12 2022 2 12 2022 16 8 2022 13 10 2022 © 2022 British Journal of Anaesthesia. Published by Elsevier Ltd. All rights reserved. 2022 British Journal of Anaesthesia Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Ventilatory ratio (VR) has been proposed as an alternative approach to estimate physiological dead space. However, the absolute value of VR, at constant dead space, might be affected by venous admixture and CO2 volume expired per minute (VCO2). Methods This was a retrospective, observational study of mechanically ventilated patients with acute respiratory distress syndrome (ARDS) in the UK and Italy. Venous admixture was either directly measured or estimated using the surrogate measure PaO2/FiO2 ratio. VCO2 was estimated through the resting energy expenditure derived from the Harris–Benedict formula. Results A total of 641 mechanically ventilated patients with mild (n=65), moderate (n=363), or severe (n=213) ARDS were studied. Venous admixture was measured (n=153 patients) or estimated using the PaO2/FiO2 ratio (n=448). The VR increased exponentially as a function of the dead space, and the absolute values of this relationship were a function of VCO2. At a physiological dead space of 0.6, VR was 1.1, 1.4, and 1.7 in patients with VCO2 equal to 200, 250, and 300, respectively. VR was independently associated with mortality (odds ratio [OR]=2.5; 95% confidence interval [CI], 1.8–3.5), but was not associated when adjusted for VD/VTphys, VCO2, PaO2/FiO2 (ORadj=1.2; 95% CI, 0.7–2.1). These three variables remained independent predictors of ICU mortality (VD/VTphys [ORadj=17.9; 95% CI, 1.8–185; P<0.05]; VCO2 [ORadj=0.99; 95% CI, 0.99–1.00; P<0.001]; and PaO2/FiO2 (ORadj=0.99; 95% CI, 0.99–1.00; P<0.001]). Conclusions VR is a useful aggregate variable associated with outcome, but variables not associated with ventilation (VCO2 and venous admixture) strongly contribute to the high values of VR seen in patients with severe illness. Keywords ARDS dead space mechanical ventilation venous admixture ventilatory ratio Handling editor: Gareth Ackland ==== Body pmc Editor's key points • The use of ventilatory ratio to quantify physiological dead space is potentially limited by venous admixture and CO2 volume expired per minute. • The authors examined data from patients mechanically ventilated with ARDS. • Ventilatory ratio increased exponentially as a function of the dead space. • Ventilatory ratio was associated with mortality, but non-ventilatory variables were the chief contributors to high ventilatory ratio values associated with severe illness. The physiological dead space (VD/VTphys) reflects the severity of lung injury1 and is a powerful prognostic factor in acute respiratory distress syndrome (ARDS).2, 3, 4 Its use, however, is uncommon, as it requires measurement of mixed expired CO2 and the simultaneous arterial blood sample to determine Paco 2. The ventilatory ratio (VR) has recently emerged as an alternative measure of ventilatory efficiency. VR correlates strongly with VD/VTphys,5 does not require measurement of mixed expired CO2, and can be easily calculated from a few routinely collected variables.6 In addition, the unitless VR is easy to interpret, as it is normalised to a predefined ‘standard’ and quantifies the degree of impaired CO2 elimination in relation to an expected reference value. However, VR may be affected by factors such as venous admixture (Qva/Q) and CO2 volume expired per minute (VCO2), which can alter the absolute value of VR despite an unchanged dead space ventilation. The potential effects of these two factors on VR, in particular Qva/Q, have been described but not quantified.7 Specifically, there are no clinical data that establish the relative importance of measured Qva/Q on VR, nor the relative importance of VCO2 on VR when the VD/VTphys is adjusted for the degree of Qva/Q. These considerations are particularly important in patients with more severe disease, in whom the assumption that virtually all of the variations in VR are attributable to an increased VD/VTphys 8 may be confounded by the effect of larger venous admixture. We compared VR and VD/VTphys in a large cohort of ventilated patients with ARDS, aiming to: (1) define the effect of Qva/Q and VCO2 on VR; (2) examine the relationship between mortality and VR corrected for physiological confounders; (3) provide theoretical models to explain the variations in VR which may occur for the same VD/VTphys. Methods Study design This was a multicentre, retrospective, observational study including 641 patients with ARDS (448 patients admitted to Guy's & St Thomas' NHS Foundation Trust, London, UK, from March 2020 to March 2021; and 193 admitted to San Paolo Hospital in Milan, Italy, from 2003 to 2018). All patients present in the databases were included into the analysis, except for seven patients who underwent extra-corporeal support. The study was approved by the institutional review board of each hospital, and written informed consent was obtained according to the national regulations of the participating institutions (see Supplementary material for details). All patients met ARDS criteria, according to the Berlin definition.9 Variables The following variables were collected contemporaneously at the time of lowest PaO 2/FiO2 ratio during the first 24 h of mechanical ventilation: minute ventilation (VE), tidal volume (VT), ventilatory frequency, and arterial PCO2 (PaCO2). VCO2 was estimated in all 641 patients using the Harris–Benedict formula10 and mixed-expired P co 2 (PECO2) was computed using the estimated VCO2/VE ratio. In 129 patients, both VCO2 and PECO2 were directly measured with capnometry. A subgroup of patients with normal Paco 2 (4.5–6 kPa) and VD/VTphys <0.35 was used to calculate the theoretical reference VE needed to estimate the VR. Modelling of ventilatory ratio To understand the relationships between VD/VTphys, VR, and venous admixture, we created a model as a function of their independent determinants: VCO2, minute ventilation (VE), venous admixture (Qva/Q), cardiac output (Qt), and arterial CO2 tension (Paco 2). The model derivation described in the supplement, shows that the VD/VTphys utilising arterial P co 2 as a surrogate of alveolar P co 2 depends on the alveolar/total ventilation ratio (Supplementary material, equation [6]) and the Qva/Q (Supplementary material, equation [8]). In addition, the VR depends both on VD/VTphys and VCO2 (Supplementary material, equation [15]). Definition of dead space Physiological dead space (VD/VTphys) was defined as the dead space calculated using the Bohr–Enghoff formula, which assumes a Qva/Q of zero (i.e. arterial P co 2 is equal to the alveolar P co 2). Corrected dead space (VD/VTcorr) was defined as VD/VTphys corrected for the Qva/Q using the Kuwabara equation11 and its modification using CO2 content in the blood,12 rather than the CO2 pressure. As the dead space fraction obtained with both methods were similar, we have used the classical Kuwabara equation for simplicity. Quantitative computed tomography Quantitative chest CT was performed as previously described using a dedicated software (Maluna).13 , 14 We estimated lung weight, gas volume, and the amount of well-aerated, poorly aerated, and non-aerated tissues. Statistical analysis All continuous data are presented as means (standard deviation [sd]) with comparisons between two means performed using with Student's t-test, and with analysis of variance (anova) between more values. Categorical data were presented as counts and percentages, with comparisons between categories made using χ2 tests. Linear regression was used to test associations among variables. The association between VR and ICU mortality was examined through univariable and multivariable logistic regression models. To assess the association of VR with mortality when adjusted for covariates which could have a contribution to the VR, we performed a multivariable logistic regression including VCO2, PaO 2/FiO2, and VD/VTphys. To make an additional comparison of the ORs, we standardised all covariates by subtracting the mean and dividing by their sd. After standardisation, the ORs refer to a unit change in sd of each covariate – therefore giving all covariates numerically similar scales. Model coefficients are reported for standardised and non-standardised data. The aim of this multivariable analysis was not to find a model which included all the factors potentially associated to outcome (i.e. age, mechanical power), but to explore the effects of the physiological variables which contribute to the VR and its association to outcome once adjusted by these physiological confounders. Two-tailed P-values <0.05 were considered statistically significant. All analyses were performed with R for Statistical Computing 4.0 (R Foundation for Statistical Computing, Vienna, Austria). Results Patient characteristics Table 1 shows the characteristics of the study cohort (n=641) at baseline, with ARDS severity categorised according to the Berlin definition.9 COVID-19 was the aetiology in 70% of the population (see Supplementary material for details). Patients with more severe ARDS had higher BMI, VE, Paco 2, and VR, and driving pressure and mechanical power; and lower P end-tidal co 2 (P et co 2) to Paco 2 ratio. ICU mortality increased with severity from 18% in mild, to 31% in moderate, and 46% in severe ARDS.Table 1 Demographic data of the whole study cohort. All data are presented as means (standard deviation). Analysis of variance (anova) was used to analyse mean values. ∗Data presented as absolute and relative frequencies (%). Table 1 Overall (n=641) Mild (n=65; 10.1%) Moderate (n=363; 56.6%) Severe (n=213; 33.2%) P-value Age (yr) 58 (17–88) 55 (18–84) 59 (17–88) 58 (19–86) 0.138 Female (%)∗ 195 (30.5) 24 (37) 108 (39) 63 (30) 0.485 Height (cm) 171 (8.9) 172 (8.7) 171 (9) 171 (8.9) 0.541 Actual weight (kg) 83 (20.8) 78 (19.4) 82 (19.5) 88 (22.9) <0.001 BMI (kg m−2) 28.4 (6.7) 26.2 (5.5) 27.8 (5.9) 30.1 (7.9) <0.001 PaO2/FiO2 (kPa) 17.2 (6.9) 32 (4.4) 18.4 (3.5) 10.5 (1.8) <0.001 Minute ventilation (L min−1) 8.7 (2.3) 8.3 (2.1) 8.5 (2.2) 9.0 (2.6) 0.007 Respiratory rate (bpm) 18 (4.3) 17 (4.6) 18 (4.2) 19 (4.3) <0.001 Tidal volume kgPBW−1 (ml kg−1) 7.4 (1.6) 7.7 (1.6) 7.3 (1.5) 7.4 (1.7) 0.349 VCO2 (ml min−1) 116.9 (34) 114.2 (33.2) 114.1 (32.1) 122.4 (36.8) 0.012 Paco2 (kPa) 6.1 (1.2) 5.6 (0.9) 6 (1.2) 6.3 (1.3) <0.001 Physiological dead space 0.55 (0.15) 0.69 (0.13) 0.72 (0.12) 0.72 (0.13) 0.225 Ventilatory ratio 1.51 (0.50) 1.32 (0.37) 1.47 (0.46) 1.63 (0.57) <0.001 Pend-tidalco2/Paco2 0.79 (0.16) 0.88 (0.14) 0.81 (0.14) 0.73 (0.16) <0.001 Mechanical power 17 (7) 15 (7) 16 (6) 18 (8) 0.002 Driving pressure (mm H2O) 14 (4) 13 (4) 14 (4) 15 (4) <0.001 ICU mortality (%)∗ 224 (35%) 12 (18%) 114 (31%) 98 (46%) <0.001 Calculation of ventilatory ratio The reference value used for VR computation was 5.3 kPa 0.1 L min−1 kg−1, derived from patients during anaesthesia,15 where 0.1 L min−1 kg−1 is assumed as the normal VE. To assess whether a similar reference value can be applied to critically ill patients, we selected among our 641 ICU patients 26 patients with ‘normal’ Paco 2 and VD/VTphys. The characteristics of this ARDS reference cohort are reported in Supplementary Table E1. The measured VE averaged 0.1 L min−1 kgPBW −1 and Paco 2 averaged 5.3 kPa (0.4) – which were similar to those found in patients undergoing anaesthesia5 , 15 and led to identical VR. Quantitative chest CT scan and contemporaneous arterial and central venous blood gas samples were available in 153 patients (Table 2 ), allowing for the calculation of Qva/Q and determination of its effects on the VD/VTphys and VR. Qva/Q increased from 0.31 (0.15) in mild ARDS, and to 0.39 (0.11) and 0.61 (0.12), respectively, for moderate and severe ARDS, likely reflecting the increase in the non-aerated tissue fraction. The VD/VTphys in this cohort also increased with disease severity, increasing from 0.54 (0.13) in mild disease, to 0.57 (0.12) in moderate disease, and 0.65 (0.11) in severe disease (P<0.001). However, the physiological dead space corrected (VD/VTcorr) for Qva/Q was similar in all patients across severity categories.Table 2 Demographic data of the subgroup with chest computed tomography and paired central venous and arterial gases. All data are presented as means (standard deviation). Analysis of variance (anova) was used to analyse mean values. ∗Data presented as absolute and relative frequencies. Table 2 Overall (n=153) Mild (n=27) Moderate (n=97) Severe (n=29) P-value Age (yr) 59 (18–88) 54 (18–84) 60 (21–88) 61 (19–86) 0.226 Female (%)∗ 49 (33%) 11 (41%) 30 (31%) 8 (28%) 0.533 Height (cm) 170 (9.5) 171 (9.5) 170 (9.9) 171 (7.9) 0.785 Actual weight (kg) 77 (21.3) 75 (18.1) 75 (20.8) 85 (29.3) 0.056 BMI (kg m−2) 26.3 (6.4) 25.7 (5.6) 25.8 (5.8) 28.8 (9.1) 0.059 PaO2/FiO2 (kPa) 20 (7.6) 32.4 (4.1) 19.5 (3.7) 10 (2.1) <0.001 Venous admixture 0.42 (0.15) 0.31 (0.15) 0.39 (0.11) 0.61 (0.12) <0.001 Minute ventilation (L min−1) 8.7 (2.1) 8.4 (1.6) 8.6 (2.1) 9.7 (2.3) 0.032 Ventilatory frequency (bpm) 17 (4.5) 16 (5.1) 16 (4.3) 19 (4.8) 0.062 Tidal volume/kgPBW (ml kg−1) 8.3 (1.4) 8.6 (1.4) 8.6 (1.4) 8.1 (1.3) 0.485 VCO2 (ml min−1) 180 (39.8) 180 (40.1) 176 (36.7) 191 (47.5) 0.148 Paco2 (kPa) 45.2 (7.7) 42.1 (5.4) 44.4 (7.6) 50.6 (7.7) <0.001 Physiological dead space 0.58 (0.12) 0.54 (0.13) 0.57 (0.12) 0.65 (0.11) <0.001 True dead space 0.53 (0.13) 0.51 (0.15) 0.53 (0.13) 0.53 (0.27) 0.199 Ventilatory ratio 1.54 (0.45) 1.38 (0.39) 1.49 (0.45) 1.86 (0.39) <0.001 Mechanical power 19 (7) 20 (8) 18 (7) 21 (8) 0.002 Driving pressure 14 (4) 14 (3) 14 (4) 15 (4) <0.001 Lung tissue mass (g) 1480 (504) 1234 (186.1) 1429 (481) 1894 (565.2) <0.001 Lung gas volume (ml) 1141 (635) 1192 (489) 1173 (657) 977 (673) 0.319 Non-aerated tissue fraction (%) 42.6 (15.5) 36.2 (13) 41.5 (15.1) 52.7 (14.8) <0.001 Poorly aerated tissue fraction (%) 31.1 (11.3) 30.6 (10.8) 31.5 (11.8) 30.1 (10.1) 0.830 Normally aerated tissue fraction (%) 25.9 (13.3) 32.9 (11.5) 26.7 (12.9) 16.8 (11.3) <0.001 ICU mortality (%)∗ 63 (41%) 7 (26%) 34 (36%) 22 (75%) <0.001 Relationship between dead space, VCO2, and ventilatory ratio The theoretical relationship between VR and VD/VTphys demonstrates that VR increases asymptotically with VD/VTphys, and its value depends on VCO2 (Fig 1 a). VRs <1 (0.86 [0.11], n=71), between 1 and 2 (1.44 [0.27], n=484) and >2 (2.41 [0.52], n=86) were respectively associated (Fig 1b) with VD/VTphys of 0.32 (0.11), 0.55 (0.1), and 0.73 (0.07). This exponential relationship was shifted on the vertical axis (VR) depending on VCO2. Patients with VCO2 values exceeding the median (208 [29] ml min−1) had a higher VR than patients with VCO2 below the median for the same calculated VD/VTphys. In both the theoretical (Fig 1a) and the actual cohort (Fig 1b), remarkably different VRs were associated with the same VD/VTphys, depending on the VCO2.Fig 1 (a) Theoretical model: ventilatory ratio as a function of the physiological dead space at VCO2 equal to 186 ml min−1 (median of the clinical cohort – blue line), 228 ml min−1 (50% above the median – red line), and 142 ml min−1 (50% below the median). (b) Clinical cohort (n=641): ventilatory ratio as a function of physiological dead space. Patients with VCO2 higher than median (186 ml min−1) are represented by red points (average VCO2 equal to 208 (29) ml min−1); patients with VCO2 below the median are represented by green points (average VCO2 equal to 164 (17) ml min−1). Fig 1 Effects of venous admixture on physiological dead space and ventilatory ratio Venous admixture and physiological dead space Figure 2 shows the relationship between VD/VTcorr and VD/VTphys in our theoretical model (panel a) and in the subgroup of 153 patients in whom the computation of the VD/VTcorr for Qva/Q was possible (panel b). As shown, the relationship between VD/VTcorr and VD/VTphys was linear but shifted to the right with higher Qva/Q. Figure 2 shows that VD/VTphys, as measured in clinical practice, corresponds to VD/VTcorr only if Qva/Q is zero, that is the alveolar P co 2 equals the arterial P co 2. With increasing Qva/Q (Fig 2a), VD/VTcorr was remarkably lower than VD/VTphys. The difference between physiological and VD/VTcorr as a function of Qva/Q is reported in Supplementary Figure E1, panel A.Fig 2 (a) Theoretical model: the true physiological dead space as a function of physiological dead space at different venous admixture. The blue line is the identity line (venous admixture equal to zero), green line denotes venous admixture equals to 0.31, whereas the red line has venous admixture equal to 0.48. (b) Corrected dead space as a function of the physiological dead space in the clinical cohort of patients in which venous admixture was available. The venous admixture levels were the average above the median (0.31 [0.07]) and below the median (0.48 [0.13]). Fig 2 Venous admixture and ventilatory ratio In Fig 3 we report the differences between VR and the VR corrected for Qva/Q in mild, moderate, and severe ARDS. The difference between VR and VR corrected for Qva/Q becomes progressively larger with greater disease severity at different levels of VD/VTphys (Supplementary Figure E1, panel B).Fig 3 Effect of the venous admixture, measured as difference between ventilatory ratio and ventilatory ratio corrected for venous admixture, in different classes of ARDS severity. The greater is the venous admixture, the higher is its effect on the difference between measured and corrected ventilatory ratios. ARDS, acute respiratory distress syndrome; VR, ventilatory ratio. Fig 3 Associations of dead space and ventilatory ratio with mortality In the entire cohort, VD/VTphys, PaO 2/FiO2, and VR were independently associated with mortality. The OR for mortality of VR and VD/VTphys were respectively 2.5 (95% CI, 1.8–3.5) and 7.04 (95% CI, 1.9–27.7). The area under the receiver operating characteristic (ROC) curve was 0.64 (95% CI, 0.59–0.68) for VR and 0.66 (95% CI, 0.62–0.71) for VD/VTphys. When the effect of VR on mortality was adjusted – in a multivariable model – for variables proven to affect VR in the physiological modelling (i.e. VD/VTphys, VCO2, PaO 2/FiO2), VR was no longer independently associated with mortality, ORadj=1.2 (95% CI, 0.7–2.1). On the contrary, VD/VTphys (ORadj=17.9; 95% CI, 1.8–185; P<0.05); VCO2 (ORadj=0.99; 95% CI, 0.99–1.00; P<0.001); and PaO 2/FiO2 (ORadj=0.99; 95% CI, 0.99–1.00; P<0.001) remained independent predictors of ICU mortality. To further investigate the relative association of each covariate on mortality, we used a standardised model including the same variables so that the resulting adjusted ORs refer to a unit change in sd – regardless of the real units, therefore giving all covariates similar numerical scale. Using this model, the standardised adjusted OR (ORst-adj) for mortality were 1.09 (95% CI, 0.8–1.5), 1.5 (95% CI, 1.1–2.1; P<0.05); 0.71 (95% CI, 0.58–0.87; P<0.001), 0.67 (95% CI, 0.55–0.81; P<0.001) for VR, VD/VTphys, VCO2, and PaO 2/FiO2, respectively, all independently associated with mortality. These results indicate that variations in VCO2 and PaO 2/FiO2 have a similar and important independent association with mortality and affect the prognostic prediction of VR. Discussion The main results of this study are: (1) the effect of Qva/Q on absolute VR becomes larger with increasing VD/VTphys; (2) the effect of VCO2 is also of major significance, particularly when VR is corrected for Qva/Q; (3) VR is a useful aggregate variable associated with outcome; however, it does not only reflect VD/VTphys but also important contributions from VCO2 (Fig 1) and Qva/Q, reflected by PaO 2/FiO2 (Supplementary Figure E2). These data suggest that VCO2 and Qva/Q contribute to the high values of VR seen in the most severe categories of patients. The CO2-related variables are strongly related with structural lung changes in ARDS1 and in COVID-19 pneumonia.16 In the Bohr's formulation,17 VD/VTphys was measured as the difference between alveolar and mixed expired CO2 normalised to the alveolar P co 2 (Supplementary material, equation [1]). The alveolar P co 2, according to Riley and colleagues,18 is the pressure present continuously and uniformly in functioning alveoli, assuming that the quantity of CO2 exchanged from blood to alveoli occurs in equal proportion to the VCO2 measured in the expired air (Supplementary material, equation [2]). The dead space fraction computed in this model depends on the ratio between alveolar ventilation and minute ventilation, regardless of oxygenation status or VCO2. The measurement of alveolar P co 2, however, is complex and not easily performed in clinical practice; therefore, VD/VTphys is estimated using the Enghoff modification, where alveolar P co 2 is assumed equal to arterial P co 2.19 The dead space computed in this way was defined as ‘physiological’, as during health, the alveolar P co 2 and arterial P co 2 differ only by 0.1–0.4 kPa. In ARDS, however, the arterial P co 2 may substantially exceed the alveolar P co 2, because of the effect of Qva/Q. Indeed, the difference between arterial and alveolar P co 2 increases with Qva/Q and VCO2, whereas it decreases with cardiac output (Supplementary Figure E3). Therefore, the substitution of alveolar with the arterial P co 2 overestimates the true dead space. To correct for the Qva/Q effect, Kuwabara and Duncalf11 proposed an equation based on the mass conservation principle:CaCO2=CcCO2∗(1−QvaQ)+CvcCO2∗QvaQ where CaCO2, CcCO2, and CvcCO2 are the CO2 contents in arterial, pulmonary (ventilated) capillary, and mixed venous blood, respectively. Kuwabara and Duncalf11 assumed that tensions and contents are in equilibrium and vary proportionately, and therefore the formula to correct dead space for shunt uses gas tensions instead of their contents. Although this assumption is not strictly accurate, using CO2 contents or tensions provided similar results (see supplement). Therefore, despite its limitations, the Kuwabara equation is the best available option to correct the dead space. The impact of Qva/Q on VD/VTphys may be relevant at Qva/Q>0.2–0.3 (Supplementary Fig. E1, panel A). Indeed, the ‘physiological’ dead space in pathological conditions represents the entirety of the gas exchange dysfunction, as it is influenced both by wasted ventilation (dead space ventilation) and wasted perfusion (Qva/Q). VR has been proposed by Sinha and colleagues5 as an estimate of ventilatory efficiency. A theoretical analysis8 indicated that VCO2 and VD/VTphys are both determinants of VR. VR uses as a reference the product of ‘standard’ VE and the ‘standard’ Paco 2. The standard VE was derived, more than five decades ago, from normal subjects undergoing anaesthesia.15 Interestingly, we found similar values (0.1 kgPBW −1) in our subgroup of ARDS patients. VR values in the literature range from <1 in the anaesthetised cohorts (indicating the effects of normal VD/VTphys and Qva/Q and reduced VCO2) to >5 in ICU patients. The largest values of VR are unlikely to reflect the magnitude of dead space ventilation alone, and it is therefore unclear whether the higher absolute value of VR observed in severe ARDS reflects a worse dead space or the greater contribution of the Qva/Q. Our multivariable logistic regression indicates that VR alone is a useful aggregate variable associated with outcome with odds ratios similar to other studies.7 Because of the relationship between VR and PaO 2/FiO2 ratio, particularly in severe disease, VR should be interpreted accordingly and not considered a bedside index to estimate purely dead space. The physiological dead space and VR have a near-exponential relationship whose level depends on VCO2. Indeed, we found higher VR in patients with higher VCO2 (Fig 1). Sinha and colleagues7 found weak and non-significant association between VCO2 and VR. They attributed this to the smaller and short-lived variation in VCO2 compared with the larger variations of VD/VTphys. However, we found that the effects of VCO2 are more marked when VR is corrected for Qva/Q. The recognition that venous admixture (Qva/Q) and VCO2 can change the absolute value of VR despite an unchanged dead space ventilation has several potential implications: (1) Changes in VR may not be attributed to a change in VD/VT if there are associated variations in oxygenation or VCO2. This may affect the interpretation of the effect of therapeutic manoeuvres such as prone position, PEEP selection, or pulmonary vasodilators on the change in VD/VT. In these examples, changes in VR may be determined by a variable combination of reduction in Qva/Q and VD/VT – but not necessarily exclusively in VD/VT. (2) In patients with more severe disease, the variations in VR may be confounded by the effect of larger Qva/Q. In this case, interventions that affect Qva/Q may disproportionally affect VR and affect the assumption of the underlying pathophysiological mechanisms. (3) Prediction models using VR as a proxy of VD/VT can inflate the range and its prognostic effect. (4) Although VCO2 disparities may appear a minor problem in general cohort, the VR dependency on this variable makes its use misleading in cases of abnormal VCO2 or during extracorporeal support, where a substantial portion of CO2 may be cleared by the membrane lung. In that setting, VD/VTphys fully reflects the lung status, whereas VR may appear normal or even low. The major limitation of this work, beyond its retrospective design, is that VCO2 was estimated rather than measured. The computation relies on the Harris–Benedict equation, which estimates the ‘standard’ VCO2 production based on age, height, and weight (Supplementary material, equation [19]). In ICU patients, we may expect remarkable discrepancies between the actual and the predicted VCO2. Yet, in the 176 patients in which VCO2 was measured, the relationship with the computed VCO2 was acceptable and the bias between measured and computed VCO2 was –22 (48) ml, despite the large variability of their absolute values (Supplementary Fig. E4). Any inaccuracy of VCO2 estimation should affect the VD/VT (Supplementary material, Equation [2]), whereas it would not affect the calculation of VR. The measured and estimated VD/VTphys values computed in 176 individuals, however, were similar (0.65 [0.13] and 0.59 [0.12], respectively). In conclusion, our data suggest that: (1) if one aims to strictly determine the true dead space in ARDS, the VD/VTphys must be corrected for Qva/Q; (2) the VD/VTphys is an estimate of the overall gas exchange (oxygenation and CO2 clearance) and, as such, a powerful clinical tool to assess the severity of the lung impairment; (3) ventilatory ratio alone is a useful variable associated with outcome; however, ventilatory ratio, as VD/VTphys, does not only reflect ventilated regions alone but also the important contributions of VCO2 and Qva/Q. Authors' contributions Study concept and design: LG, LC, JM, MQ, KM Acquisition, analysis, or interpretation of data: RM, PP, LG, LC, OM, MB, SG, CZ, RD, MV, FR First drafting of manuscript-writing committee: RM; PP, LG, FR, LC, BS, JM, MQ Critical revision for important intellectual content and final approval of manuscript: LC, JM, MQ, RM, LC, JM, MQ, JW, DC Statistical analysis: BS, JW, RM Paper review and modifications: all authors Administrative, technical or material support: KM, OM, PH Declaration of Interest LG reports a consultancy for General Electric and SIDAM. He also receives lecture fees from Estor and Mindray. Funding Sartorius AG (Göttingen, Germany) for an unrestricted grant for lung injury-related research to the Department of Anesthesiology of Göttingen University Medical Center; Klaus-Tschira Stiftung gGmbH, Heidelberg (development of the software for CT analysis). Appendix A Supplementary data The following is the Supplementary data to this article:Multimedia component 1 Multimedia component 1 Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.bja.2022.10.035. ==== Refs References 1 Gattinoni L. Bombino M. Pelosi P. Lung structure and function in different stages of severe adult respiratory distress syndrome JAMA 271 1994 1772 1779 8196122 2 Nuckton T.J. Pittet J.F. Pulmonary dead-space fraction as a risk factor for death in the acute respiratory distress syndrome N Engl J Med 346 2002 1281 1286 11973365 3 Cressoni M. Cadringher P. Chiurazzi C. Lung inhomogeneity in patients with acute respiratory distress syndrome Am J Respir Crit Care Med 189 2014 149 158 24261322 4 Kallet R.H. Zhuo H. Liu K.D. Calfee C.S. Matthay M.A. On behalf of the National Heart Lung and Blood Institute ARDS Network Investigators. The association between physiologic dead-space fraction and mortality in subjects with ARDS enrolled in a prospective multi-center clinical trial Respir Care 59 2014 1611 1618 24381187 5 Sinha P. Fauvel N.J. Singh S. Soni N. Ventilatory ratio: a simple bedside measure of ventilation Br J Anaesth 102 2009 692 697 19346233 6 Morales-Quinteros L. Schultz M.J. Bringué J. MARS Consortium Estimated dead space fraction and the ventilatory ratio are associated with mortality in early ARDS Ann Intensive Care 9 2019 128 31754866 7 Sinha P. Calfee C.S. Beitler J.R. Physiologic analysis and clinical performance of the ventilatory ratio in acute respiratory distress syndrome Am J Respir Crit Care Med 199 2019 333 341 30211618 8 Sinha P. Fauvel N.J. Singh P. Soni N. Analysis of ventilatory ratio as a novel method to monitor ventilatory adequacy at the bedside Crit Care 17 2013 R34 23445563 9 Ranieri V.M. Rubenfeld G.D. Thompson B.T. ARDS Definition Task Force Acute respiratory distress syndrome: the Berlin Definition JAMA 307 2012 2526 2533 22797452 10 Harris A. Benedict F. A biometric study of human basal metabolism Proc Natl Acad Sci U S A 4 1918 370 373 16576330 11 Kuwabara S. Duncalf D. Effect of anatomic shunt on physiologic deadspace-to-tidal volume ratio—a new equation Anesthesiology 31 1969 575 577 5348825 12 Douglas A.R. Jones N.L. Reed J.W. Calculation of whole blood CO2 content J Appl Physiol 65 1988 473 477 3136136 13 Herrmann P. Nguyen X. Luecke T. Quintel M. MALUNA 1.03 a software tool for the analysis of computed tomographic slice images of the lung Jamal R Jaschinski H Published Online 2002 389 395 14 Herrmann P. Busana M. Cressoni M. Using artificial intelligence for automatic segmentation of CT lung images in acute respiratory distress syndrome Front Physiol 12 2021 676118 15 Nunn J.F. Ventilation nomograms during anaesthesia Anaesthesia 15 1960 65 14427919 16 Rossi S. Palumbo M.M. Sverzellati N. Mechanisms of oxygenation responses to proning and recruitment in COVID-19 pneumonia Intensive Care Med 48 2022 56 66 34825929 17 Bohr C. Ueber die Lungenathmung1 Skand Arch Für Physiol 2 1891 236 268 18 Riley R.L. Lilienthal J.L. Proemmel D.D. Franke R.E. On the determination of the physiologically effective pressures of oxygen and carbon dioxide in alveolar air Am J Physiol 147 1946 191 198 21000736 19 Enghoff H. Volumen inefficax. Bermekungen zur Frage des shadlichen Raumes Upsala Lakareforen Forh 44 1938 191 218	36470747	PMC9718027	NO-CC CODE	2022-12-06 23:23:27	no	Br J Anaesth. 2022 Dec 2; doi: 10.1016/j.bja.2022.10.035	utf-8	Br J Anaesth	2,022	10.1016/j.bja.2022.10.035	oa_other
==== Front Anal Chem Anal Chem ac ancham Analytical Chemistry 0003-2700 1520-6882 American Chemical Society 36448939 10.1021/acs.analchem.2c03055 Article Chemiluminescent Nanogels as Intensive and Stable Signal Probes for Fast Immunoassay of SARS-CoV-2 Nucleocapsid Protein Qu Fajin Shu Jiangnan * Wang Shanshan Haghighatbin Mohammad A. https://orcid.org/0000-0003-4769-9464 Cui Hua * CAS Key Laboratory of Soft Matter Chemistry, Collaborative Innovation Center of Chemistry for Energy Materials, Department of Chemistry, University of Science and Technology of China, Hefei, Anhui230026, P. R. China * E-mail: [email protected]. Fax: +86-551-63600730. * E-mail: [email protected]. Fax: +86-551-63600730. 30 11 2022 acs.analchem.2c0305515 07 2022 16 11 2022 © 2022 American Chemical Society 2022 American Chemical Society This article is made available via the PMC Open Access Subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. It is highly desired to exploit good nanomaterials as nanocarriers for immobilizing chemiluminescence (CL) reagents, catalysts and antibodies to develop signal probes with intensive and stable CL properties for immunoassays. In this work, N-(4-aminobutyl)-N-ethylisoluminol (ABEI) and Co2+ bifunctionalized polymethylacrylic acid nanogels (PMAANGs-ABEI/Co2+) were synthesized via a facile strategy by utilizing carboxyl group-rich PMAANGs as nanocarriers to immobilize ABEI and Co2+. The obtained PMAANGs-ABEI/Co2+ showed extraordinary CL performance. The CL intensity is 2 orders of magnitude higher than that of previously reported ABEI and Cu2+–cysteine complex bifunctionalized gold nanoparticles with high CL efficiency. This was attributed to the excellent catalytic ability of Co2+ and polymethylacrylic acid nanogels, as well as the improved CL catalytic efficiency from a decreased spatial distance between ABEI and the catalyst. The as-prepared nanogels also possess abundant surface reaction sites and good CL stability. On this basis, a sandwich immunoassay for the nucleocapsid protein of SARS-CoV-2 (N protein) was developed by using magnetic bead connected primary antibody as a capture probe and PMAANGs-ABEI/Co2+ connected secondary antibody as a signal probe. The linear range of the proposed method for N protein detection was 3.16–316 ng/mL, and its detection limit was 2.19 ng/mL (S/N = 3). Moreover, the developed immunoassay was performed with a short incubation time of 5 min, which greatly reduced the detection time for N protein. By using corresponding antibodies, the developed strategy might be applied to detect other biomarkers. Clinical Research Hospital, Chinese Academy of Sciences NA YD2060002008 document-id-old-9ac2c03055 document-id-new-14ac2c03055 ccc-price This article is made available via the ACS COVID-19 subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. ==== Body pmcChemiluminescent functionalized nanomaterials (CFNMs) have been widely used as analytical interfaces or probes to develop bioassays with low detection limit and wide linear dynamic range, due to their advantages of high chemiluminescent activity, good biocompatibility and easy assembly. To date, various CFNMs have been exploited, such as chemiluminescent functionalized carbon-based nanomaterials,1 metal nanoparticles,2 silica nanoparticles,3 magnetic beads,4 and metal–organic frameworks.5,6 The synthesis methods are mainly divided into two categories: (a) Chemiluminescent molecules are loaded through noncovalent bonds, such as Au–N bonds, hydrogen bonds, and π–π stacking.4,5,7,8 However, most of the CFNMs synthesized by these methods have disadvantages of reagent shedding and leakage, which reduces the chemiluminescence (CL) stability of CFNMs and affects the reproducibility of bioassays. (b) Chemiluminescent molecules are loaded through covalent bonds.1,6 Although these CFNMs are assembled by stable covalent bonds and could avoid reagent shedding and leakage, most of the reactive sites on their surface are used to connect with CL molecules and catalysts to obtain high CL intensity, which hinders their further direct connection with the recognition elements when constructing bioassays. Therefore, it is highly desirable to develop new CFNMs with stable properties and abundant surface reactive sites for the development of new bioassays with high stability and sensitivity. Nanogels are commonly identified as hydrogel nanoparticles with tunable size ranges of 1–1000 nm formed by physically or chemically cross-linked polymeric chains.9 Owing to the advantages of high tunability, high porosity, abundant surface groups, good biocompatibility, and biodegradability, nanogels are used to load various molecules for constructing functionalized nanomaterials with good stability and abundant binding sites. They have been widely applied in the fields of drug delivery, catalysis, bioimaging and so on.10 In particular, great attention has been given to its application in bioassays. To date, a series of bioassays based on nanogels have been developed, including electrochemistry,11 fluorescence,12 Raman spectroscopy,13 colorimetry, and so on.14 For example, Liu and co-workers designed a self-assembled nanogel of thiolated silver nanoclusters and chitosan with fluorescence, which was further applied to rapid and convenient bacterial detection.12 Achadu and co-workers used pH-responsive nanogels to encapsulate MoO3-QD nanotags as a SERS-based platform for the immunoassays of hepatitis E virus or norovirus.13 However, nanogels with CL for bioassays have rarely been reported, which has the advantages of simple instrumentation and high sensitivity owing to no light source and a lack of distractions from the excitation light source. Therefore, it is important to design chemiluminescent functionalized nanogels for developing bioassays with excellent analytical performance. Polymethylacrylic acid nanogels (PMAANGs) are nanogels synthesized by the free radical copolymerization of methacrylic acid and cross-linkers (such as divinylbenzene or N,N′-methylenebis(acrylamide), etc.).15−17 It can be synthesized in a simple and economical way. Moreover, it has extremely rich carboxyl groups both inside and on its surface, which is suitable for connecting CL reagents and catalysts for the synthesis of CFNMs with high stability, high CL activity, and abundant surface reaction sites. Coronavirus disease 2019 (COVID-19) is a viral disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2).18 To date, the spread of the virus has not been controlled, although huge efforts are being made to counter the virus. Developing sensitive, accurate and fast detection methods for rapid identification of infected individuals is benefited to stop the spread of the outbreak. At present, the most universal method for COVID-19 diagnosis is real-time RT-PCR. However, RT-PCR requires skilled technicians, time-consuming amplification steps, and expensive reagents.19 Therefore, convenient, fast and economical testing methods are urgently needed as complementary tools to RT-PCR, such as antibody or antigen testing.18 However, antibodies in serum samples increased after 7 days of infection with the virus, which makes the antibody test unsuitable for early diagnosis of COVID-19.4,20 The nucleocapsid protein of SARS-CoV-2 (N protein) plays a significant role in the replication of the virus. It is largely expressed during SARS-CoV-2 replication, which can be detected in serum samples even just 1 day after infection. Therefore, N protein is a suitable target for the early diagnosis of COVID-19.21 In this work, we synthesized CL reagent N-(4-aminobutyl)-N-ethylisoluminol (ABEI) and catalyst Co2+ bifunctionalized polymethylacrylic acid nanogels (PMAANGs-ABEI/Co2+) with outstanding CL performance. The resulting materials were characterized by fluorescence spectroscopy, transmission electron microscopy (TEM), UV–vis absorption spectroscopy, inductively coupled plasma atomic emission spectroscopy (ICP-AES), dynamic light scattering (DLS) measurements, X-ray photoelectron spectroscopy (XPS), and Fourier transform infrared spectroscopy (FT-IR). The as-prepared PMAANGs-ABEI/Co2+ showed extraordinary CL efficiency and good stability. A possible CL mechanism is proposed by investigating the effects of radical scavengers. In addition, a new CL signal probe was designed based on PMAANGs-ABEI/Co2+ and used to develop a sandwich CL immunoassay to realize the specific detection of the N protein. Experimental Section Chemicals and Materials 2,2-Azobis(isobutyronitrile) (AIBN, 99%) was obtained from Meryer (Shanghai, China). ABEI (≥97.0%) powders were purchased from TCI (Japan). Methacrylic acid (MAA, ≥98.0%), glycine (≥99.5%), acetonitrile, sodium hydroxide (NaOH), hydrogen peroxide (H2O2, 30.0%), and cobalt chloride hexahydrate (CoCl2·6H2O) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Bovine serum albumin (BSA) was obtained from Solarbio (Beijing, China). A 10 mM PBS buffer (pH 7.4) was purchased from Sangon Biotechnology Co., Ltd. (Shanghai, China). Carboxyl magnetic beads (MBs-COOH, 1.0 μm) were purchased from Hangzhou Watson Biotechnology Co., Ltd. (Wuxi, China). Divinylbenzene (DVB, 80%), 1-(3-(dimethylamino)propyl)-3-ethylcarbodiimide hydrochloride (EDC, 98.0%), and N-hydroxysuccinimide (NHS, 98%) were purchased from Aladdin Reagent (Shanghai, China). SARS-CoV-2 antigens, including N protein and the receptor-binding domain (RBD) of spike 1 protein, were provided by Jin’s Group (University of Science and Technology of China). Primary antibody (Ab1) and secondary antibody (Ab2) of N protein were obtained from Guangdong Feipeng Biological Co., Ltd. (Guangdong, China). A milli-Q system (Milli-pore, France) was utilized to produce ultrapure water, which was used throughout the experiment. Preparation of PMAANGs-ABEI/Co2+ PMAANGs were synthesized according to the method reported in the literature.22 PMAANGs-ABEI/Co2+ was synthesized with a convenient and advisable strategy. First, ABEI was covalently bonded to PMAANGs through amide bonds. In a typical synthesis, a total of 8 mL of aqueous solution containing 80 mg of EDC and 80 mg of NHS was mixed with 2 mL of 20 mg/mL PMAANGs solution, followed by adding 0.5 mL of 4 mM ABEI (pH 12.0) solution. After reacting for an hour at room temperature, the product (PMAANGs-ABEI) was separated by centrifugation, washed with deionized water twice to remove excess activator and unreacted ABEI, and then redispersed in 4 mL of deionized water. Subsequently, Co2+ was loaded on PMAANGs-ABEI. Briefly, the obtained suspension was mixed with 0.4 mL of 1 mg/mL CoCl2·6H2O solution. After reacting for 15 min at room temperature, unreacted Co2+ was removed via centrifugation at 6000 rpm for 3 min. Finally, the precipitate was dispersed with deionized water to obtain the PMAANGs-ABEI/Co2+ sample for further use. Characterization TEM images were taken with a transmission electron microscope (H7700, Hitachi, Japan). FT-IR spectra were recorded on a FT-IR spectrophotometer (EQUINX55, Broker, Germany). The XPS data was obtained using XPS (ESCALAB 250Xi, VG Scientific, U.K.) with Al Kα radiation (hν = 1486.6 eV) as the X-ray excitation source. The concentration of ABEI in PMAANGs-ABEI/Co2+ was measured by UV–vis spectrophotometer (Agilent 8453, Agilent Technologies, U.S.A.). The concentration of Co2+ in PMAANGs-ABEI/Co2+ was obtained by the ICP-AES measurements carried out on a plasma atomic emission spectrometer (Optima 7300 DV, PerkinElmer, U.S.A.). Hydrodynamic diameter of PMAANGs-ABEI/Co2+ was measured using a DLS spectrometer (SYNC, U.S.A.). Fluorescence emission spectra were obtained using a fluorescence spectrophotometer (F-7000, Hitachi, Japan) equipped with a 150 W xenon lamp as the light source. CL spectra were obtained by a spectral detector assembled by ourselves, including a spectrometer (SpectraPro HRS-300, Teledyne Princeton Instruments, U.S.A.) and an EMCCD (Newton, Andor Technology, U.K.). CL Immunoassay for N Protein Detection Signal probes PMAANGs-ABEI/Co2+-Ab2/BSA for CL immunoassay were prepared as follows. A total of 50 μL of 10 mg/mL prepared PMAANGs-ABEI/Co2+ was dispersed in 1.0 mL of aqueous solution containing 4 mg EDC and 4 mg NHS. After shaking at room temperature for 20 min, the activated PMAANGs-ABEI/Co2+ were washed three times by centrifugation and redispersed in 0.8 mL of PBS buffer. Subsequently, the above suspension was mixed with 50 μL of 1 mg/mL Ab2 to react for 2 h under shaking, followed by blocking with 50 μL of 10 mg/mL BSA and 50 μL of 1 M glycine for another 0.5 h. Finally, the obtained PMAANGs-ABEI/Co2+-Ab2/BSA was washed by centrifugation and redispersed in 1 mL of PBS buffer containing 0.2 mg/mL BSA. Capture probes MBs-Ab1/BSA for the CL immunoassay were assembled as follows. First, 100 μL of 10 mg/mL MBs-COOH was washed three times with deionized water by magnetic separation. Then, the purified MBs-COOH were dispersed in 1.0 mL of solution containing 4 mg EDC and 4 mg NHS. After shaking at room temperature for 20 min, the activated MBs-COOH were washed three times with PBS buffer and redispersed in 0.8 mL of PBS buffer. Subsequently, the above suspension was mixed with 50 μL of 1 mg/mL Ab1 to react for 2 h under shaking, followed by blocking with 50 μL of 10 mg/mL BSA for another 0.5 h. Finally, the obtained MBs-Ab1/BSA were purified and redispersed in 1 mL of PBS buffer containing 0.2 mg/mL BSA. N protein was detected by sandwich immunoassay as follows. First, 150 μL of N protein with different concentrations (1.0 × 10–8.5 to 1.0 × 10–6.5 g/mL) was added into 100 μL of PBS buffer containing 0.5 mg/mL MBs-Ab1/BSA and 0.125 mg/mL PMAANGs-ABEI/Co2+-Ab2/BSA. After shaking at 37 °C for 5 min, the immunocomplex was purified by magnetic separation and redispersed in 500 μL of PBS buffer. Subsequently, the CL emission of the immunocomplex was measured with a Centro LB960 microplate luminometer (Berthold, Germany). In a typical CL measurement, 80 μL of the immunocomplex suspension was added into the wells of a 96-well plate. Then, 80 μL of 50 mM H2O2 in 0.1 M NaOH solution was injected into the well to initiate the CL reaction, and the CL kinetic curve was recorded at the same time. Results and Discussion Synthesis and Characterization of PMAANGs-ABEI/Co2+ The synthetic route for PMAANGs-ABEI/Co2+ is schematically illustrated in Scheme 1. First, when the solvent acetonitrile was heated to boiling, the free-radical copolymerization of monomer (MAA) and cross-linker (DVB) was initiated by the free-radical initiator (AIBN).15 Since acetonitrile was a poor solvent for the polymer, the generated macromolecules began to precipitate and grow into particles by continuing to capture monomers and oligomers. Thus, PMAANGs were obtained as the color of the solution changed from transparent to milky white. Subsequently, the carboxyl group in PMAANGs was activated by EDC and NHS to form a NHS-ester intermediate, which could be substituted by the primary amino group of ABEI via the nucleophilic reaction to obtain PMAANGs-ABEI.23 Finally, by virtue of the coordination reaction, CL catalyst Co2+ combined with the carboxyl groups which were still abundantly present in PMAANGs-ABEI to obtain PMAANGs-ABEI/Co2+. Scheme 1 Schematic Illustration for Assembly of PMAANGs-ABEI/Co2+ The successful assembly of PMAANGs-ABEI/Co2+ was verified by various characterization methods. First, the morphology of PMAANGs-ABEI/Co2+ was investigated by TEM. As shown in Figure 1A, all of the nanogels were uniform spheres with an average size of 580 ± 16 nm. In addition, the nanogels in aqueous solution were also characterized by DLS. As shown in Figure 1B, the average hydrodynamic diameter of PMAANGs-ABEI/Co2+ measured by DLS was 674 nm, which was obviously larger than the average size obtained by TEM. This was attributed to the swelling of the nanogels in aqueous solutions.24 Furthermore, the polydispersity index (PDI) of PMAANGs-ABEI/Co2+ was 0.016, also suggesting that the nanogels had good dispersity. The results indicated that the obtained nanogels have uniform morphology and good monodispersity. Figure 1 (A) TEM image of PMAANGs-ABEI/Co2+. Inset: size distribution histogram of PMAANGs-ABEI/Co2+. (B) Hydrodynamic diameter distributions of PMAANGs-ABEI/Co2+ obtained by DLS. (C) FT-IR spectra of PMAANGs, PMAANGs-ABEI and PMAANGs-ABEI/Co2+. (D) Survey XPS of PMAANGs, PMAANGs-ABEI and PMAANGs-ABEI/Co2+. (E) Deconvolution of N 1s spectra of PMAANGs, PMAANGs-ABEI and PMAANGs-ABEI/Co2+. (F) Fluorescence emission spectra of ABEI, PMAANGs and PMAANGs-ABEI/Co2+ excited at 270 nm. Then, the ingredients and molecular interactions of the nanogels were investigated by FT-IR. The black curve in Figure 1C shows the FT-IR spectrum of PMAANGs. Specifically, the C≡N stretching vibration at 2238 cm–1 came from initiator AIBN.25 C=C stretching vibration of phenyl at 1510, 1580, and 1600 cm–1 belonged to cross-linker DVB.26 O–H stretching vibration of carboxyl at 3450 cm–1, as well as C–H shear type vibration of −CH3 at 1388 cm–1 were assigned to monomer MAA.26,27 These characteristic peaks indicated the successful synthesis of PMAANGs via copolymerization of MAA and DVB. As shown by the red curve in Figure 1C, when PMAANGs were modified with ABEI, one of the primary changes was that the characteristic peaks of C–N stretching vibrations at 1019 and 1068 cm–1 became stronger, indicating successful assembly of ABEI.28 Another obvious change was that a new peak arose at 1800 cm–1, which might be assigned to the C=O antisymmetric stretching vibration of succinimide in NHS-ester, the activated product of carboxyl. Meanwhile, the C=O symmetric stretching vibration of succinimide could also be observed at 1750 cm–1, indicating that excessive NHS-ester did not fully react with ABEI, and some of them were kept.29 After loading Co2+, the O–H stretching vibration of the carboxyl at 3450 cm–1 could still be observed in the FT-IR spectrum of PMAANGs-ABEI/Co2+, indicating that there were still unreacted carboxyl groups in PMAANGs-ABEI/Co2+, which could be further used to connect antibodies for constructing immunoassay probes. As shown in Figure 1D,E, the survey XPS revealed changes in element contents of the nanogels during assembly. The intensity of the N 1s peak at around 400 eV in Figure 1D observably increased after integration of ABEI into PMAANGs. To learn more about ingredient changes of the nanogels during assembly, further investigations about the N 1s peak were performed. As shown in Figure 1E, the N 1s peak of PMAANGs-ABEI and PMAANGs-ABEI/Co2+ could be curved-fitted into three peaks at 399.8, 401.2, and 402.3 eV.4,30 Specifically, the peak at 399.8 eV, which was also present in the XPS spectrum of PMAANGs, was assignable to C≡N of AIBN and the tertiary amine of ABEI. Another peak at 402.3 eV could be assigned to the NHS-ester, admitting the presence of the intermediate NHS-ester in PMAANGs-ABEI/Co2+.30 The last peak at 401.2 eV was attributed to the amide bond, indicating that ABEI was integrated into PMAANGs via amide bonds.4 Then, the presence of ABEI in PMAANGs-ABEI/Co2+ was further confirmed by the fluorescence emission spectra. As shown in Figure 1F, compared with a single peak in the fluorescence emission spectra of ABEI (440 nm) and PMAANGs (320 nm), two peaks located in the same range could be simultaneously observed in that of PMAANGs-ABEI/Co2+, indicating the successful loading of ABEI.2 Furthermore, the concentration of ABEI was further measured by UV–vis absorption spectrometry. It was calculated that there was 0.157 mM ABEI in 10 mg/mL PMAANGs-ABEI/Co2+. Finally, the concentration of Co2+ was analyzed by ICP-AES and the result showed that there was 11.9 μg/mL Co2+ in 10 mg/mL PMAANGs-ABEI/Co2+. CL Behavior of PMAANGs-ABEI/Co2+ First, the CL kinetic behavior of the nanogels was investigated using a microplate photometer. As shown in Figure 2A, PMAANGs-ABEI/Co2+ exhibited an intense CL emission (red curve) when reacting with H2O2. However, PMAANGs did not show any CL activity (cyan curve), and PMAANGs-ABEI exhibited a weak CL emission (blue curve), indicating that the CL activity of the obtained nanogels was related to ABEI. Furthermore, the CL spectrum of PMAANGs-ABEI/Co2+ and H2O2 systems shows an emission centered at 450 nm, which matches the CL spectrum of ABEI–H2O2, as shown in Figure S1.2 The above results indicated that the luminophore in the reaction of PMAANGs-ABEI/Co2+ with H2O2 was the excited-state oxidation product of ABEI (ABEI-ox). Figure 2 (A) CL kinetic curves of PMAANGs (cyan curve), PMAANGs-ABEI (blue curve), and PMAANGs-ABEI/Co2+ (red curve) with H2O2. The inset of A is a magnification of the CL kinetic curves of PMAANGs (cyan curve) and PMAANGs-ABEI (blue curve) with H2O2. (B) CL intensity of ABEI (a), mixture of PMAANGs and ABEI (b), PMAANGs-ABEI (c), mixture of ABEI and Co2+ (d), mixture of PMAANGs, ABEI and Co2+ (e), and PMAANGs-ABEI/Co2+ (f) with H2O2. The inset of C is a magnification of CL intensity of ABEI (a), mixture of PMAANGs and ABEI (b), PMAANGs-ABEI (c) with H2O2. (C) CL stability of PMAANGs-ABEI/Co2+ within 30 days. In order to further illustrate the outstanding CL performance of PMAANGs-ABEI/Co2+, a series of controlled experiments were performed. As shown in Figure 2B, the CL intensities of pure ABEI (a), the mixture of PMAANGs and ABEI (b) and the mixture of ABEI and Co2+ (d) increased in turn, and the CL intensity of the mixture of ABEI and Co2+ was much higher than that of the others. The results indicate that both Co2+ and PMAANGs exhibited catalytic activity for ABEI–H2O2 CL system and the catalytic activity of Co2+ was much better than that of PMAANGs. According to the molecular structure of PMAANGs, the catalytic activity of PMAANGs was likely attributed to the large number of carboxyl groups from its monomers MAA.31 Furthermore, the CL intensities of PMAANGs-ABEI (c) and PMAANGs-ABEI/Co2+ (f) were higher than those of the mixture of PMAANGs and ABEI (b), and the mixture of PMAANGs, ABEI, and Co2+ (e), respectively. The results indicated that the CL efficiency was enhanced after the assembly of ABEI and Co2+ on PMAANGs, due to the decrease in the spatial distance between catalysts and CL molecules, as reported in the literature.32 In addition, as shown in Figure S2, the CL kinetic curves of PMAANGs-ABEI and PMAANGs-ABEI/Co2+ were slow glow type, which was obviously different from that of other controlled samples with sharp CL curves. The slow glow type CL emissions of PMAANGs-ABEI and PMAANGs-ABEI/Co2+ occurred because the diffusion of H2O2 was limited by the three-dimensional network structure of the nanogels, resulting in ABEI molecules could not directly react with H2O2 at the moment of injection.33,34 Consequently, based on the above results, the novel CL performance of PMAANGs-ABEI/Co2+ could be attributed to the combination of the following three merits: (i) The superior catalytic effect of Co2+ and PMAANGs in the ABEI–H2O2 system. (ii) The catalytic efficiency was improved by decreasing the spatial distance between the ABEI and catalyst. (iii) The restriction on the fast diffusion of H2O2 because of the three-dimensional network structure of the nanogels. To further understand the intense CL emission property of PMAANGs-ABEI/Co2+, we compared its CL intensity with previously reported ABEI and Cu2+–cysteine complex bifunctionalized gold nanoparticles (Cu2+-Cys/ABEI-AuNPs) at the same synthesis concentration of ABEI.35 As shown in Figure S3, under the respective optimal reaction conditions, the CL intensity of PMAANGs-ABEI/Co2+ was approximately 100 times higher than that of Cu2+-Cys/ABEI-AuNPs. The result indicated that the as-prepared PMAANGs-ABEI/Co2+ nanogel exhibited superior CL performance. Furthermore, the CL stability of PMAANGs-ABEI/Co2+ was investigated by CL measurements within 30 days, as shown in Figure 2C. The RSD of CL intensity of PMAANGs-ABEI/Co2+ (interday stability, n = 30) was 3.26%, indicating that PMAANGs-ABEI/Co2+ had outstanding CL stability, which laid a solid foundation for its applications in bioassays. Optimization of the Experimental Conditions According to previous researches, the CL reaction of ABEI–H2O2 system could be catalyzed by various metal ions.32 In order to explore whether Co2+ was the best metal ion catalyst for CL reaction of PMAANGs-ABEI with H2O2, other metal ions, including Fe3+, Cr3+, Cd2+, Sn2+, Pb2+, Ni2+, Mn2+, Ce3+, and Cu2+, were used to synthesize PMAANGs-ABEI/Mn+ and further studied their CL behaviors. As shown in Figure S4A, PMAANGs-ABEI/Co2+ exhibited much higher CL activity than PMAANGs-ABEI/Mn+, which was consistent with the fact that Co2+ was the best metal ion catalyst for the ABEI–H2O2 CL system. Therefore, Co2+ was selected for the optimal metal ion catalyst for the following studies. Furthermore, the experimental conditions were optimized to achieve the best CL performance of PMAANGs-ABEI/Co2+. First, the effect of the Co2+ content on CL intensity was investigated by adjusting the mass ratios of CoCl2·6H2O and the nanogels. As shown in Figure S4B, the CL intensity of PMAANGs-ABEI/Co2+ increased as the increase of the mass ratio of CoCl2·6H2O and nanogels from 1:1000 to 1:100, and reached a plateau after the mass ratio reached 1:100. Thus, a mass ratio of CoCl2·6H2O and nanogels of 1:100 was selected for further experiments. In addition, previous research studies demonstrated that the conditions of the coreaction reagents were important factors for CL intensity. Therefore, the effects of the concentration and pH of H2O2 on the CL intensity of this system were investigated. As shown in Figure S4C, the CL intensity increased as the increase of the concentration of H2O2 and reached a plateau after reaching 50 mM. Thus, 50 mM of H2O2 was selected for further experiments. The CL intensity increased slowly as the pH of H2O2 increased from 10.0 to 12.0 in Figure S4D. However, it was significantly enhanced as the pH changed from 12.0 to 13.0. There were two main reasons accounting for this phenomenon. On one hand, it was well-known that the CL reaction of ABEI–H2O2 system was carried out in alkaline solution because higher pH could accelerate the decomposition of H2O2 and the deprotonation of ABEI, which in turns enhanced CL emission.8,36 On the other hand, the CL kinetic curves of PMAANGs-ABEI/Co2+ with H2O2 at pH 12.0 and 13.0 were analyzed as shown in Figure S5. It was found that the CL kinetic curve at pH 12.0 was obviously slower than that at pH 13.0. The reason might be due to that stronger alkaline solution could promote the swelling of PMAANGs-ABEI/Co2+, thereby increased the diffusion rate of H2O2 in the three-dimensional network structure and accelerated the CL reaction.37 As a result, the above reasons contribute to significant increase of the CL intensity with the change of pH from 12.0 to 13.0. Finally, the optimal conditions are as follows: the concentration of H2O2 was 50 mM, and the pH of the H2O2 solution was 13.0. CL Mechanism of PMAANGs-ABEI/Co2+ Early studies have shown that the reaction mechanism of the ABEI–H2O2 system was a radical mediated process, specifically referring to OH•, O2•–, and ABEI•–.2,38 Thus, the effects of thiourea (OH• radical scavengers) and SOD (O2•– radical scavengers) on the CL intensity of the PMAANGs-ABEI/Co2+–H2O2 system were investigated. As shown in Figure S6, the CL intensity of the PMAANGs-ABEI/Co2+–H2O2 system decreased obviously with increasing concentrations of thiourea or SOD, demonstrating that OH• and O2•– are both involved in the CL reaction between PMAANGs-ABEI/Co2+ and H2O2. Moreover, according to the above experimental results, both Co2+ and PMAANGs exhibited catalytic activity for the CL reaction of ABEI with H2O2. Thus, a possible CL mechanism of the PMAANGs-ABEI/Co2+–H2O2 system is proposed as follows: (i) Co2+ in PMAANGs-ABEI/Co2+ catalyzes the decomposition of H2O2 to obtain highly reactive OH•, which further reacts with O2 to form O2•–.38,39 (ii) −COO– combines with O2•– to form −CO4•2–.40 (iii) OH•, O2•–, and −CO4•2– react with ABEI– to form ABEI radicals (ABEI•–). (iv) ABEI•– is oxidized by O2•– to yield ABEI-ox, which then returns to the ground state, resulting in strong CL emission. Sandwich CL Immunoassay for N Protein Considering the outstanding CL property, good stability, and abundant carboxyl groups on the surface, PMAANGs-ABEI/Co2+ are used to construct CL signal probes PMAANGs-ABEI/Co2+-Ab2/BSA for detecting N protein, as shown in Scheme 2A. First, after activated by EDC and NHS, PMAANGs-ABEI/Co2+ could directly conjugate with Ab2 via amide bonds, obtaining PMAANGs-ABEI/Co2+-Ab2. Subsequently, nonspecific binding sites were blocked by BSA and glycine, obtaining PMAANGs-ABEI/Co2+-Ab2/BSA. Meanwhile, to improve the washing and separation efficiency during the immunoassay process, MBs-COOH were chosen as carriers of Ab1 to construct capture probes MBs-Ab1/BSA for immunoassay, as shown in Scheme 2B. Specifically, Ab1 was integrated into MBs-COOH via amide bonds, obtaining MBs-Ab1. Subsequently, nonspecific binding sites of MBs-Ab1 were blocked by BSA, and the capture probe MBs-Ab1/BSA was obtained. A schematic representation of the detection principle of the CL immunoassay is shown in Scheme 2C. First, N protein, MBs-Ab1/BSA and PMAANGs-ABEI/Co2+-Ab2/BSA were directly mixed and incubated at 37 °C to form an immunocomplex. After removing the unbound PMAANGs-ABEI/Co2+-Ab2/BSA by magnetic separation, the CL intensity of the immunocomplex with H2O2 was obtained by a microplate luminometer for detecting N protein. Scheme 2 Schematic Illustration for the Preparation of Signal Probe PMAANGs-ABEI/Co2+-Ab2/BSA (A), Capture Probe MBs-Ab1/BSA (B), and Sandwich Immunoassay (C) Incubation time was a crucial factor influencing the immunoassay performance. Thus, the incubation time for the immunoreaction was investigated from 5 to 40 min with different N protein concentrations. As shown in Figure S7A, at each concentration of N protein, the relative CL intensity I – I0 do not show a significant increase with prolonged incubation time from 5 to 40 min, indicating that long incubation periods could not effectively improve sensitivity in this system, which might be attributed to the fast interaction kinetics of N protein with the antibodies. Furthermore, due to nonspecific adsorption between the signal probe and the capture probe, the CL intensity of the control in the absence of N protein obviously increased with prolonged incubation time, as shown in Figure S7B. Therefore, considering that prolonging the incubation time could not effectively improve the sensitivity and meanwhile increased nonspecific absorption, 5 min of incubation time was chosen for the following experiments. Under the above conditions, the analytical performance of the proposed sandwich CL immunoassay for N protein was investigated. As shown in Figure 3A, the logarithm of CL intensity was positively correlated with the logarithm of N protein concentration. A good linear relationship was obtained in N protein concentration range of 3.16–316 ng/mL. The linear regressive equation with a correlation coefficient of 0.996 was log(I) = 0.7680 × log(C) + 10.36 (I: maximum CL intensity; C: the concentration of N protein). The detection limit for detecting N protein was calculated to be 2.19 ng/mL (S/N = 3). A comparison of the chemiluminescent nanogels-based immunoassay with previously developed sandwich immunoassay has been shown in Table S1. Although our method is not as sensitive as most reported methods, it can be conducted with a short incubation time of 5 min and can give detection results within 8 min, which greatly reduces the detection time for N protein. Figure 3 (A) Linear relationship between logarithm of maximum CL intensity and the logarithm of N protein concentration. (B) Selectivity of the CL immunoassay to N protein (100 ng/mL) by comparison with blank, IgG (1.0 mg/mL), IgM (1.0 mg/mL), RBD (1.0 mg/mL), and their mixture. Afterward, the selectivity of the proposed immunoassay method was investigated by using IgG, IgM, and RBD as interferents instead of N protein. As shown in Figure 3B, when the concentration of each interferent was 10 times higher than that of N protein, only two samples that contained N protein showed strong CL signals. Therefore, the as-prepared immunoassay had excellent selectivity for detecting N protein of SARS-CoV-2. Detection of N Protein in Human Serum Samples The ability of the proposed CL immunoassay to resist complex substrates was further investigated by detecting N protein in human serum. Healthy human serum samples were collected from the Second Hospital of Anhui Medical University and stored at −78 °C. All serum samples were diluted 100 times with 0.01 M PBS (pH 7.4) before detecting. Different concentrations (25.0, 50.0, 100.0 ng/mL) of N protein were added to the diluted samples. The measurement results in Table 1 showed that the immunoassay was of good performance with recoveries in the range of 91.2–107.3%. Therefore, the developed CL immunoassay possessed the potential to be applied in clinical samples. Table 1 Quantitative Determination of N Protein in Human Serum (n = 3) serum sample initial (ng/mL) added (ng/mL) found (ng/mL) recovery (%) 1 not detected 25.0 23.5 ± 2.6 94.1 2 not detected 50.0 45.5 ± 4.1 91.2 3 not detected 100.0 107.3 ± 5.2 107.3 Conclusion Carboxyl group-rich PMAANGs were used as a nanocarrier for the effective immobilization of ABEI and Co2+ to synthesize ABEI and Co2+ bifunctionalized nanogels PMAANGs-ABEI/Co2+ for the first time. The obtained PMAANGs-ABEI/Co2+ were homogeneous spherical particles with an average size of 580 ± 16 nm. They exhibited extraordinary CL emission, and their CL intensity was approximately 100 times higher than that of Cu2+-Cys/ABEI-AuNPs with excellent CL performance. This was attributed to the significant catalytic ability of Co2+ and PMAANGs, as well as the improved CL catalytic efficiency from a decreased spatial distance between ABEI and catalysis. Based on the outstanding CL property of PMAANGs-ABEI/Co2+, a sandwich CL immunoassay was constructed for the specific detection of N protein in the concentration range of 3.16–316 ng/mL, with a detection limit of 2.19 ng/mL. Moreover, the developed immunoassay was performed with a short incubation time of 5 min, which greatly reduced the detection time for N protein. This work shows that nanogels are ideal nanocarriers for designing novel CFNMs with high stability and CL efficiency. The proposed fast immunoassay strategy can also be extended to the detection of other proteins using corresponding antibodies, which is of great significance for the rapid diagnosis of diseases. Furthermore, it is foreseeable that chemiluminescent functionalized nanogels are of great potential for in vivo imaging applications due to their advantages of high tunability, abundant surface groups, and good biocompatibility. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.2c03055.Synthesis of PMAANGs; CL spectra of ABEI and PMAANGs-ABEI/Co2+; CL kinetic curves of nanogels and controlled samples; CL intensity of PMAANGs-ABEI/Co2+ compared with Cu2+-Cys/ABEI-AuNPs; effect of different metal ions on the CL intensity; effect of Co2+ concentration on the CL intensity; effect of pH values and H2O2 concentration on the CL intensity; effect of thiourea and SOD on the CL intensity of PMAANGs-ABEI/Co2+; CL kinetic curves of PMAANGs-ABEI/Co2+ with H2O2 at pH 12 and 13; optimization of incubation time for immunoreaction; comparison of the chemiluminescent nanogels-based immunoassay with previously reported sandwich immunoassay for detecting N protein (PDF) Supplementary Material ac2c03055_si_001.pdf Author Contributions The manuscript was written through contributions of all authors. The authors declare no competing financial interest. Acknowledgments This research was supported by a COVID-19 special task grant by the Chinese Academy of Sciences Clinical Research Hospital (Hefei) with Grant No. YD2060002008. ==== Refs References Gao Y. ; Huang Y. ; Chen J. ; Liu Y. ; Xu Y. ; Ning X. A Novel Luminescent ″Nanochip″ as a Tandem Catalytic System for Chemiluminescent Detection of Sweat Glucose. Anal. Chem. 2021, 93 , 10593–10600. 10.1021/acs.analchem.1c01798.34291923 Zhao K. ; Shen W. ; Cui H. Highly chemiluminescent silver nanoclusters with a dual catalytic center. J. Mater. Chem. C 2018, 6 , 6549–6555. 10.1039/C8TC01966E. Fan W. ; Li L. ; Yuan J. ; Ma X. ; Jia J. ; Zhang X. Aggregation-Induced Emission Effect within Peroxyoxalate-Loaded Mesoporous Silica Nanoparticles for Efficient Harvest of Chemiluminescence Energy in Aqueous Solutions. Anal. Chem. 2021, 93 , 17043–17050. 10.1021/acs.analchem.1c03798.34907772 Wang S. ; Shu J. ; Lyu A. ; Huang X. ; Zeng W. ; Jin T. ; Cui H. Label-Free Immunoassay for Sensitive and Rapid Detection of the SARS-CoV-2 Antigen Based on Functionalized Magnetic Nanobeads with Chemiluminescence and Immunoactivity. Anal. Chem. 2021, 93 , 14238–14246. 10.1021/acs.analchem.1c03208.34636246 Li Z. ; Xi Y. ; Zhao A. ; Jiang J. ; Li B. ; Yang X. ; He J. ; Li F. Cobalt-imidazole metal-organic framework loaded with luminol for paper-based chemiluminescence detection of catechol with use of a smartphone. Anal. Bioanal. Chem. 2021, 413 , 3541–3550. 10.1007/s00216-021-03305-8.33782733 Wang C. ; Zhang N. ; Li Y. ; Yang L. ; Wei D. ; Yan T. ; Ju H. ; Du B. ; Wei Q. Cobalt-based metal-organic frameworks as co-reaction accelerator for enhancing electrochemiluminescence behavior of N-(aminobutyl)-N-(ethylisoluminol) and ultrasensitive immunosensing of amyloid-β protein. Sens. Actuators, B 2019, 291 , 319–328. 10.1016/j.snb.2019.04.097. Wu Y. ; Wang J. ; Cui H. Chemiluminescent magnetic nanoparticles with good catalytic activity and rapid separation capability and sensitive sensing for H2O2. Anal. Bioanal. Chem. 2022, 414 , 367–375. 10.1007/s00216-021-03597-w.34363088 Li Q. ; Liu X. ; Zhuang M. ; Wang X. ; Cui H. Cobalt(ii)-8-hydroxyquinoline-5-sulfonic acid complex/N-(4-aminobutyl)-N-ethylisoluminol/reduced graphene hybrids as nanocatalytic reaction platforms for chemiluminescence. RSC Adv. 2017, 7 , 37261–37267. 10.1039/C7RA06327J. Basak S. The Age of Multistimuli-responsive Nanogels: The Finest Evolved Nano Delivery System in Biomedical Sciences. Biotechnol. Bioprocess Eng. 2020, 25 , 655–669. 10.1007/s12257-020-0152-0. Pinelli F. ; Saadati M. ; Zare E. N. ; Makvandi P. ; Masi M. ; Sacchetti A. ; Rossi F. A perspective on the applications of functionalized nanogels: promises and challenges. Int. Mater. Rev. 2022, 1–25. 10.1080/09506608.2022.2026864. Li W. ; Shu D. ; Han H. ; Ma Z. An amperometric immunoprobe based on multifunctional nanogel for sensitive detection of tumor marker. Sens. Actuators, B 2018, 273 , 1451–1455. 10.1016/j.snb.2018.07.067. Liu J. ; Liu L. ; Li S. ; Kang Q. ; Zhang R. ; Zhu Z. Self-assembled nanogels of luminescent thiolated silver nanoclusters and chitosan as bactericidal agent and bacterial sensor. Mater. Sci. Eng., C 2021, 118 , 111520 10.1016/j.msec.2020.111520. Achadu O. J. ; Abe F. ; Li T. C. ; Khoris I. M. ; Lee D. ; Lee J. ; Suzuki T. ; Park E. Y. Molybdenum Trioxide Quantum Dot-Encapsulated Nanogels for Virus Detection by Surface-Enhanced Raman Scattering on a 2D Substrate. ACS Appl. Mater. Interfaces 2021, 13 , 27836–27844. 10.1021/acsami.1c04793.34105944 Jayeoye T. J. ; Rujiralai T. Green, in situ fabrication of silver/poly(3-aminophenyl boronic acid)/sodium alginate nanogel and hydrogen peroxide sensing capacity. Carbohydr. Polym. 2020, 246 , 116657 10.1016/j.carbpol.2020.116657.32747289 Ghalehkhondabi V. ; Fazlali A. ; Soleymani M. Folic acid-conjugated pH-responsive poly(methacrylic acid) nanospheres for targeted delivery of anticancer drugs to breast cancer cells. J. Mol. Liq. 2022, 348 , 118028 10.1016/j.molliq.2021.118028. Jiang G. ; Wang N. ; Jia L. ; Che H. ; Wang L. ; Yang J. ; Xu H. ; Wu C. ; Liu M. Multi-functional DNA-conjugated nanohydrogels for aptamer-directed breast cancer cell targeting. New J. Chem. 2021, 45 , 20410–20418. 10.1039/D1NJ04152E. Jin S. ; Li D. ; Yang P. ; Guo J. ; Lu J. Q. ; Wang C. Redox/pH stimuli-responsive biodegradable PEGylated P(MAA/BACy) nanohydrogels for controlled releasing of anticancer drugs. Colloids Surfaces A 2015, 484 , 47–55. 10.1016/j.colsurfa.2015.07.041. Mutantu P. N. ; Ngwe Tun M. M. ; Nabeshima T. ; Yu F. ; Mukadi P. K. ; Tanaka T. ; Tashiro M. ; Fujita A. ; Kanie N. ; Oshiro R. ; et al. Development and Evaluation of Quantitative Immunoglobulin G Enzyme-Linked Immunosorbent Assay for the Diagnosis of Coronavirus Disease 2019 Using Truncated Recombinant Nucleocapsid Protein as Assay Antigen. Int. J. Environ. Res. Public Health 2021, 18 , 9630 10.3390/ijerph18189630.34574555 Lyu A. ; Jin T. ; Wang S. ; Huang X. ; Zeng W. ; Yang R. ; Cui H. Automatic label-free immunoassay with high sensitivity for rapid detection of SARS-CoV-2 nucleocapsid protein based on chemiluminescent magnetic beads. Sens. Actuators B Chem. 2021, 349 , 130739 10.1016/j.snb.2021.130739.34611381 Che X. Y. ; Hao W. ; Wang Y. D. ; Di B. ; Yin K. ; Xu Y. C. ; Feng C. S. ; Wan Z. Y. ; Cheng V. C. C. ; Yuen K. Y. Nucleocapsid protein as early diagnostic marker for SARS. Emerging Infect. Dis. 2004, 10 , 1947–1949. 10.3201/eid1011.040516. Dutta N. K. ; Mazumdar K. ; Gordy J. T. The Nucleocapsid Protein of SARS-CoV-2: a Target for Vaccine Development. J. Virol. 2020, 94 , e00647-20 10.1128/JVI.00647-20.32546606 Jin S. ; Pan Y. ; Wang C. Reflux Precipitation Polymerization: A New Technology for Preparation of Monodisperse Polymer Nanohydrogels. Acta Chim. Sinica 2013, 71 , 1500–1504. 10.6023/A13070776. Wu H. ; Jin H. ; Wang C. ; Zhang Z. ; Ruan H. ; Sun L. ; Yang C. ; Li Y. ; Qin W. ; Wang C. Synergistic Cisplatin/Doxorubicin Combination Chemotherapy for Multidrug-Resistant Cancer via Polymeric Nanogels Targeting Delivery. ACS Appl. Mater. Interfaces 2017, 9 , 9426–9436. 10.1021/acsami.6b16844.28247750 Gelber C. ; Margel S. Synthesis and Characterization of Free and Grafted N-Halamine Nanoparticles for Decomposition of Organic Dyes in an Aqueous Continuous Phase. ACS Omega 2020, 5 , 4004–4013. 10.1021/acsomega.9b03529.32149227 Dalapati R. ; Nandi S. ; Biswas S. Post-synthetic modification of a metal-organic framework with a chemodosimeter for the rapid detection of lethal cyanide via dual emission. Dalton Trans. 2020, 49 , 8684–8692. 10.1039/D0DT00837K.32598422 Luo J. ; Hu W. ; Suo Z. ; Wang Y. ; Zhang Y. Co-pyrolysis of spent radioactive ion exchange resin and manganese dioxide: Decrease the decomposition temperatures of functional groups. J. Hazard. Mater. 2021, 418 , 126275 10.1016/j.jhazmat.2021.126275.34102364 Weerapoprasit C. ; Prachayawarakorn J. Characterization and properties of biodegradable thermoplastic grafted starch films by different contents of methacrylic acid. Int. J. Biol. Macromol. 2019, 123 , 657–663. 10.1016/j.ijbiomac.2018.11.083.30445086 Nasrollahzadeh M. ; Motahharifar N. ; Sajjadi M. ; Naserimanesh A. ; Shokouhimehr M. Functionalization of chitosan by grafting Cu(II)-5-amino-1H-tetrazole complex as a magnetically recyclable catalyst for C-N coupling reaction. Inorg. Chem. Commun. 2022, 136 , 109135 10.1016/j.inoche.2021.109135. Lin S. ; Yu H. ; Li M. Formation of six-membered cyclic anhydrides by thermally induced intramolecular ester condensation in Eudragit E film. Polymer 1999, 40 , 3589–3593. 10.1016/S0032-3861(98)00488-1. Zhao J. ; Gao F. ; Pujari S. P. ; Zuilhof H. ; Teplyakov A. V. Universal Calibration of Computationally Predicted N 1s Binding Energies for Interpretation of XPS Experimental Measurements. Langmuir 2017, 33 , 10792–10799. 10.1021/acs.langmuir.7b02301.28921989 Yang R. ; Li F. ; Zhang W. ; Shen W. ; Yang D. ; Bian Z. ; Cui H. Chemiluminescence Immunoassays for Simultaneous Detection of Three Heart Disease Biomarkers Using Magnetic Carbon Composites and Three-Dimensional Microfluidic Paper-Based Device. Anal. Chem. 2019, 91 , 13006–13013. 10.1021/acs.analchem.9b03066.31525898 Liu Y. ; Shen W. ; Cui H. Metal ion-induced chemiluminescence recovery for highly intensive chemiluminescence bifunctionalized polydopamine nanospheres. J. Mater. Chem. C 2019, 7 , 14588–14593. 10.1039/C9TC05223B. Liu Y. ; Shen W. ; Li Q. ; Shu J. ; Gao L. ; Ma M. ; Wang W. ; Cui H. Firefly-mimicking intensive and long-lasting chemiluminescence hydrogels. Nat. Commun. 2017, 8 , 1003 10.1038/s41467-017-01101-6.29042537 Song Q. ; Yan X. ; Cui H. ; Ma M. Efficient Cascade Resonance Energy Transfer in Dynamic Nanoassembly for Intensive and Long-Lasting Multicolor Chemiluminescence. ACS Nano 2020, 14 , 3696–3702. 10.1021/acsnano.0c00847.32150394 Li F. ; Liu Y. ; Zhuang M. ; Zhang H. ; Liu X. ; Cui H. Biothiols as chelators for preparation of N-(aminobutyl)-N-(ethylisoluminol)/Cu2+ complexes bifunctionalized gold nanoparticles and sensitive sensing of pyrophosphate ion. ACS Appl. Mater. Interfaces 2014, 6 , 18104–18111. 10.1021/am504985w.25275558 Yue L. ; Liu Y. T. Mechanistic Insight into pH-Dependent Luminol Chemiluminescence in Aqueous Solution. J. Phys. Chem. B 2020, 124 , 7682–7693. 10.1021/acs.jpcb.0c06301.32790377 Heckel J. ; Loescher S. ; Mathers R. T. ; Walther A. Chemically Fueled Volume Phase Transition of Polyacid Microgels. Angew. Chem., Int. Ed. 2021, 60 , 7117–7125. 10.1002/anie.202014417. Burdo T. G. ; Seitz W. R. Mechanism of Cobalt Catalysis of Luminol Chemiluminescence. Anal. Chem. 1975, 47 , 1639–1643. 10.1021/ac60359a019. Lin J. M. ; Shan X. Q. ; Hanaoka S. ; Yamada M. Luminol chemiluminescence in unbuffered solutions with a cobalt(II)-ethanolamine complex immobilized on resin as catalyst and its application to analysis. Anal. Chem. 2001, 73 , 5043–5051. 10.1021/ac010573+.11721898 Gao L. ; Ju L. ; Cui H. Determination of the binding constant of specific interactions and binding target concentration simultaneously using a general chemiluminescence method. RSC Adv. 2016, 6 , 5305 10.1039/C5RA24928G.	36448939	PMC9718083	NO-CC CODE	2022-12-05 23:15:01	no	Anal Chem. 2022 Nov 30;:acs.analchem.2c03055	utf-8	Anal Chem	2,022	10.1021/acs.analchem.2c03055	oa_other
==== Front ACS Bio Med Chem Au ACS Bio Med Chem Au bg abmcb8 ACS Bio & Med Chem Au 2694-2437 American Chemical Society 10.1021/acsbiomedchemau.2c00046 Article Substrate Specificity and Kinetics of RNA Hydrolysis by SARS-CoV-2 NSP10/14 Exonuclease https://orcid.org/0000-0003-1413-2082 Dangerfield Tyler L. https://orcid.org/0000-0002-6575-2823 Johnson Kenneth A. * Institute for Cellular and Molecular Biology, Department of Molecular Biosciences, University of Texas, 2500 Speedway, Austin, Texas78712, United States * Email: [email protected]. 16 11 2022 acsbiomedchemau.2c0004613 07 2022 09 09 2022 09 09 2022 © 2022 The Authors. Published by American Chemical Society 2022 The Authors This article is made available via the PMC Open Access Subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), the virus that causes COVID-19, continues to evolve resistance to vaccines and existing antiviral therapies at an alarming rate, increasing the need for new direct-acting antiviral drugs. Despite significant advances in our fundamental understanding of the kinetics and mechanism of viral RNA replication, there are still open questions regarding how the proofreading exonuclease (NSP10/NSP14 complex) contributes to replication fidelity and resistance to nucleoside analogs. Through single turnover kinetic analysis, we show that the preferred substrate for the exonuclease is double-stranded RNA without any mismatches. Double-stranded RNA containing a 3′-terminal remdesivir was hydrolyzed at a rate similar to a correctly base-paired cognate nucleotide. Surprisingly, single-stranded RNA or duplex RNA containing a 3′-terminal mismatch was hydrolyzed at rates 125- and 45-fold slower, respectively, compared to the correctly base-paired double-stranded RNA. These results define the substrate specificity and rate of removal of remdesivir for the exonuclease and outline rigorous kinetic assays that could help in finding next-generation exonuclease inhibitors or nucleoside analogs that are able to evade excision. These results also raise important questions about the role of the polymerase/exonuclease complex in proofreading during viral replication. Addressing these questions through rigorous kinetic analysis will facilitate the search for desperately needed antiviral drugs to combat COVID-19. kinetics proofreading exonuclease SARS-CoV-2 polymerase NSP14 remdesivir National Institute of Allergy and Infectious Diseases 10.13039/100000060 NIH R01AI163336 document-id-old-9bg2c00046 document-id-new-14bg2c00046 ccc-price ==== Body pmcIntroduction COVID-19, caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), presents a major global health threat that is unlikely to dissipate anytime soon. Scientists around the world have published enzymatic mechanisms and structures of viral proteins at an unprecedented rate. In particular, the kinetics and mechanism of fast RNA replication (300 nt/s) by the RNA-dependent-RNA polymerase (RdRp) and its inhibition by the nucleoside analog remdesivir have been established through a combination of kinetic and structural analyses.1,2 The bifunctional NSP10/NSP14 exonuclease/methyltransferase complex has been implicated in genome proofreading3,4 and RNA capping,5 and structural biologists have provided atomic structures of the exonuclease complex in isolation6−8 as well as bound to the RdRp complex (NSP12/NSP7/NSP8) and the NSP13 helicase.9 However, important details regarding kinetics and specificity of the proofreading exonuclease are lacking. Other groups have reported expression, purification, and preliminary activity assays for the SARS-CoV-2 exonuclease complex.3,6−8,10,11 From these studies, it is clear that the exonuclease activity of NSP14 (1) requires either Mg2+ or Mn2+ for catalysis4,6,7,10 (2) strongly prefers RNA bases over DNA bases,3,8 and (3) is greatly stimulated by the addition of the noncatalytic NSP10.11 However, there are contradictions in the literature as to whether a remdesivir incorporated into the primer strand of the RNA can be efficiently excised by the exonuclease complex7,8,12 and whether the enzyme preferentially hydrolyzes single-stranded RNA over base-paired RNA or RNA with terminal mismatches.3,4,7,8,10 To date, the ambiguity present in currently available studies on the exonuclease arises largely because initial experiments to characterize the exonuclease reaction have been performed under steady-state conditions with a single long time point, anywhere from 20 to 45 min, which does not provide a valid measure of substrate specificity.13 Fixed time point assays famously underestimate enzyme discrimination because the amounts of product formed after long incubation times do not reflect the different rates of reaction for various substrates. For example, while the preferred substrate may complete the reaction in a second, incubation for 30 min gives the alternative substrate nearly 2000-fold more time to complete the reaction. In the past, this error in experimental design has led to extraordinary misinterpretations, such as the conclusion that hydrogen bonds are not needed for efficient DNA replication, an error that was corrected by subsequent single turnover kinetic analysis.14 Moreover, steady-state assays using enzymes that operate on DNA are often rate-limited by the slow release of DNA from the enzyme, so they fail to measure the rates of catalysis and thereby greatly underestimate specificity.13 Single turnover kinetic experiments directly measure intrinsic rates of DNA cleavage or polymerization and can identify rate-limiting and specificity-determining steps in the reaction pathway that are masked by steady-state methods.1,13,15 Here, we present our initial work to characterize the substrate specificity and kinetics of the exonuclease hydrolysis reaction catalyzed by the NSP10/14 complex using transient kinetic methods. We performed experiments on three physiologically relevant substrates as well as an RNA substrate containing a remdesivir monophosphate incorporated into the primer strand. Our findings have implications for the design of experiments to screen viral exonuclease inhibitors and for the biological role of NSP10/14 beyond the proposed genome proofreading function. Results NSP10/NSP14 Exonuclease Hydrolyzes Mismatched and Single-Stranded RNA Much More Slowly Than Correctly Base-Paired RNA We first measured rates of excision of the 3′-terminal nucleotide using 5′-[6-FAM] labeled RNA substrates using oligonucleotides based on those we previously used to measure kinetics of polymerization by the viral RdRp complex.1 The FAM label on the 5′-end of the primer strand was deemed to be outside of the oligonucleotide binding domain so that it would not interfere with binding at the active site, as shown in studies on DNA polymerases.15,16 As shown in Figure 1, we chose three physiologically relevant substrates to measure the rates of hydrolysis: single-stranded RNA (FAM-LS2), correctly base-paired primer/template double-stranded RNA (FAM-LS2/LS1), and primer/template double-stranded RNA containing a terminal A:A mismatch (FAM-LS2/LS1.2.2). Double-stranded RNA containing either correctly base-paired or mispaired 3′-terminal primer strands represents RNA replication intermediates that would be continuously formed by the RdRp during genome replication and possibly excised by the exonuclease. For well-characterized proofreading exonucleases, single-stranded nucleic acids are the preferred substrates.17,18 When primer extension at the polymerase active site stalls, the primer strand of duplex DNA can melt away from the complementary template strand and enter the exonuclease active site where the 3′-terminal base is rapidly excised.19 Figure 1 Duplex RNA without mismatches is the preferred substrate for the SARS-CoV-2 NSP10/NSP14 exonuclease complex. A solution of 1 μM NSP10/NSP14 was mixed with 100 nM FAM-ssRNA (FAM-LS2, blue), FAM-primer/template double-strand RNA (dsRNA) with an A:A mismatch (FAM-LS2/LS1.2.2, green), or FAM-primer/template dsRNA (FAM-LS2/LS1, red) to start the reaction. Data are shown on a logarithmic time scale, fit to single-exponential functions, with observed rates of 0.016, 0.044, and 2 s–1 for ssRNA, A:A mismatch dsRNA, and primer/template dsRNA, respectively, as summarized in Table 1. Note that the template strands LS1 and LS1.2.2 differ by three nucleotides in the single-strand region of the template strand, which is unlikely to have any nearest-neighbor effects, which are due to duplex stability and structure.21 In our experiments, a solution of 1 μM NSP10/NSP14 complex was mixed with 100 nM FAM-labeled RNA in the presence of 5 mM Mg2+. The reaction was then stopped at various time points by quenching with EDTA, and products were resolved and quantified by capillary electrophoresis.20 The experiments were designed with a concentration of enzyme in large excess over the substrate so that complete substrate conversion to product could occur in a single turnover. In Figure 1, we show the loss of the full-length starting material over time for each RNA substrate, displayed on a logarithmic time scale and fit to a single-exponential function. Surprisingly, hydrolysis was fastest for the double-stranded RNA substrate without any mismatches (2 s–1), while hydrolysis of the double-stranded substrate with a mismatch and single-stranded RNA was 45- and 125-fold slower, respectively. Since the primer/template RNA without mismatches was the preferred substrate, we performed further experiments to characterize the kinetics of this reaction. To determine whether the rate of RNA binding limited the observed rate of hydrolysis of the RNA substrate, we performed a rapid quench experiment with the dsRNA substrate using two protocols (Figure 2). In the first experiment, duplex RNA was mixed with the NSP10/14 exonuclease complex to start the reaction, as described above. Here, the enzyme must first bind to the RNA before hydrolyzing it, so the observed rate could be limited by the rate of RNA binding to the enzyme. In a separate experiment, we preincubated the RNA and the NSP10/14 complex in the presence of EDTA for 30 min before adding excess Mg2+ to start the reaction. In this experiment, the RNA equilibrates with the exonuclease complex during the preincubation step and catalysis is initiated upon adding the metal ions. For the experiment where RNA is mixed with the enzyme to start the reaction, the observed rate was 2 s–1, whereas the observed rate after adding Mg2+ to a preformed enzyme–RNA complex was approximately 5.4 s–1 (determined from the fit to the loss of the 20 nt starting material). These results suggest that the intrinsic rate of chemistry (cleavage) is at least 5.4 s–1 and the slower rate observed upon mixing enzyme with RNA may be limited by the rate of RNA binding. We observed a linear increase in the observed rate of RNA cleavage with increasing enzyme concentration (Figure S1) with an approximate apparent second-order rate constant of ∼2.2 μM–1 s–1, supporting a rate-limiting RNA binding step when the reaction was initiated by mixing RNA with 1 μM enzyme. Figure 2 Processive exonuclease kinetics observed with double-stranded RNA. (A) Excision upon mixing the enzyme and RNA to start the reaction. A solution of 1 μM NSP10/NSP14 complex was mixed with 100 nM FAM-LS2/LS1 RNA to start the reaction, monitored using a quench-flow instrument. Both syringes contained 5 mM Mg2+. Data for the loss of the 20 nt primer fit a single exponential with an observed rate of 2 ± 0.1 s–1. Data for the formation and decay of the 19 nt product best fit a double-exponential function with both rates at approximately 2 s–1, given in Table 1. (B) Excision upon the Mg2+ addition to E-RNA complex. A solution of 1 μM NSP10/NSP14, 2.5 mM EDTA, and 100 nM FAM-LS2/LS1 RNA was mixed with 7.5 mM Mg2+ to start the reaction. Samples were quenched with EDTA, and products were resolved by capillary electrophoresis. Fitting the fast phase of the loss of 20 nt starting material gave an observed rate of 6.5 ± 0.9 s–1. A double-exponential fit of the data for the formation and decay of the 19 nt intermediate is summarized in Table 1. Note that on the time scale of the experiment in panel (B) a fraction of the starting material (∼10%) fails to react. We could not account for this fraction of slow-reacting RNA by any simple model based on RNA–enzyme equilibration, so we have focused on the faster reaction phase accounting for 90% of the reaction. We monitored the loss of the 20 nt starting material, the rise and fall of the 19 nt product, and the subsequent formation of products less than or equal to 18 nt in length (Figure 2) since peaks for the 17 and 18 nt product were not well resolved (as shown in Figures S2 and S3). Although the errors are somewhat larger for the double-exponential fit to define the rise and fall of the 19 nt intermediate, the comparable rates of formation and decay (Table 1), and the fact that the decay of the 19 nt intermediate goes to completion suggest that the enzyme does not dissociate after hydrolyzing one base but rather removes at least 2 nt in a processive manner. Similar experiments were performed with the ssRNA substrate and A:A mismatch dsRNA substrate, as shown in Figures S4–S7. For the ssRNA substrate (Figures S4 and S5), the observed rate of cleavage was comparable with or without preincubation to form the E–ssRNA complex. These results could suggest either weak binding of the RNA to the enzyme at the concentrations tested or a slow rate of chemistry. Structures of the complex showing double-stranded RNA stably bound at the exonuclease active site suggest that ssRNA may bind weakly.7 Data for the mismatched RNA substrate were biphasic upon the Mg2+ addition to the ES complex, with approximately half of the amplitude corresponding to the fast phase at a rate greater than 1 s–1 and the other half at an observed rate of 0.06 s–1 (Figure S7). These results suggest that for a mismatch, the intrinsic rate of cleavage may be relatively fast; however, the initial binding may be weak and limit the enzyme’s ability to process these substrates. Table 1 Observed Ratesa figure substrate/phase observed rate (s–1) Figure 1 ssRNA (FAM-LS2) 0.016 ± 0.0005 dsRNA—0 mismatches (FAM-LS2/LS1) 2.0 ± 0.05 dsRNA—A:A mismatch (FAM-LS2/LS1.2.2) 0.046 ± 0.0033 Figure 2 RNA initiated reaction (A)—19 nt product phase 1 2.3 ± 0.2 phase 2 2.3 ± 0.2 Mg2+ initiated reaction (B)—19 nt product phase 1 6.5 ± 0.9 phase 2 5.4 ± 0.6 Figure 3 RMP RNA fast phase 2.4 ± 0.4 slow phase 0.42 ± 0.083 a Here, we summarize the results of rate measurements. For the data in Figure 2, we show the rate measurements for the formation and decay of the 19 nt intermediate derived from a double-exponential fit. Fitting the fast phase for the loss of the starting 20 nt RNA gave an observed rate of 6.5 ± 0.9 s–1. In fitting these data, we constrained the fast phase for the loss of 20 nt and formation of 19 nt intermediate to follow the same observed rate. Remdesivir Is Hydrolyzed at a Rate Comparable to Correctly Base-Paired RNA Remdesivir is a nucleoside analog drug that, in its triphosphate form (RTP), is rapidly incorporated by the SARS-CoV-2 RdRp.1 Remdesivir was approved by the FDA after it was shown to be effective in treating COVID-19, provided the drug is given early in the course of infection.22 Others have shown that a remdesivir monophosphate incorporated into a primer strand can be hydrolyzed8 contrary to earlier predictions that this nucleoside analog would resist exonuclease hydrolysis due to the 1′ cyano group.12 The rates of this hydrolysis remain to be quantitatively measured to estimate the effect of the 1′ cyano group. We therefore directly measured this rate of hydrolysis using our pre-steady-state exonuclease assay. To address whether remdesivir monophosphate (RMP) incorporated into an RNA primer strand is efficiently hydrolyzed, we used the purified SARS-CoV-2 RdRp complex to enzymatically synthesize an RNA substrate containing a terminal RMP (see the Materials and Methods section). We achieved a 92% efficiency in synthesizing the RNA primer containing RMP, as quantified by capillary electrophoresis. We then heat-denatured the RdRp complex and reannealed the primer/template duplex RNA. We performed the excision reaction on the resulting RNA in the quench flow by mixing with 1 μM NSP10/NSP14 complex and analyzed the time course of RNA cleavage by capillary electrophoresis. The reaction is processive, as shown in Figure S8. A fit to the time dependence of RMP removal is shown in Figure 3. The data best fit a double-exponential function (for comparison, a single-exponential fit is shown in Figure S9) with similar amplitudes for the fast and slow phases with observed rates of 2.4 and 0.42 s–1, respectively. The reason for two phases of the reaction is still unknown but could represent two different binding modes of the terminal RNA to the enzyme or different conformations of the remdesivir in the RNA. Nonetheless, these results show that at least half of the terminal remdesivir is hydrolyzed at a rate comparable to correctly base-paired RNA, showing that it is easily removed after incorporation into the genome. Nonetheless, it is likely that remdesivir could escape removal by the proofreading exonuclease since the RNA primer is extended more rapidly than RMP is excised and polymerization stalls only after the incorporation of three additional nucleotides on top of RMP.1,2 Figure 3 Excision of incorporated remdesivir monophosphate. (A) Scheme for the enzymatic synthesis of RMP containing substrate. SARS-CoV-2 RdRp complex and remdesivir triphosphate were added to enzymatically incorporate RMP into the primer strand (see the Materials and Methods section). The RdRp complex was heat-denatured, and then the RNA was reannealed before measuring excision by NSP10/NSP14 in the quench flow. (B) Structure of remdesivir monophosphate. Minor modifications relative to ATP in the ring and the addition of a 1′ cyano group that causes delayed chain termination by the RdRp.1,2 (C) Time course of remdesivir excision from the RNA. Data are shown fit to a double-exponential function with rates of 2.4 and 0.42 s–1 for the fast and slow phases, respectively. Amplitudes for the two phases are approximately equal. Discussion Steady-state experiments commonly performed on the NSP10/NSP14 exonuclease are difficult to interpret since the exonuclease reaction is over in a fraction of a second, while steady-state time scales close to 1 h are typically employed. Single turnover experiments have the advantage of directly measuring the relative rates of excision on biologically relevant time scales with enzyme in excess to monitor reactions occurring at the active site of the enzyme. These methods provide quantitative results that can be directly interpreted and include estimates of the concentration of active enzyme. The ambiguity of the steady-state results and the lack of a standard for acceptable enzyme activity in the literature provided the motivation for the direct experiments outlined in this paper to unambiguously address these questions. We began by choosing three RNA substrates on which to measure excision. Most proofreading exonucleases function on at least partially single-stranded nucleic acids23 arising from the melting of duplex DNA or RNA at the polymerase active site to transfer the 3′-end of the primer into the exonuclease active site.17,19,24−26 Our data show that, unlike other proofreading exonucleases, neither single-stranded RNA nor double-stranded RNA containing a mismatch is a preferred substrate, exhibiting rates of hydrolysis 45- to 125-fold slower than rates of excision on correctly base-paired RNA. This observation may imply that the RNA duplex must dissociate completely from the polymerase active site and both strands of the RNA must rebind at the exonuclease active site without tethering the template strand in the polymerase active site, as previously suggested.9 While the kinetics for the NSP10/NSP14 complex in isolation is clear, the major limitation of this study and others is that the experiments fail to consider that the exonuclease presumably functions in a complex containing the helicase, RdRp, and potentially other viral nonstructural proteins, which may change the kinetics of RNA hydrolysis and substrate specificity. For example, a recent preprint suggests that the kinetics of exonuclease hydrolysis changes in the presence of NSP16.27 Previous studies have clearly shown a robust change in activity upon the association of NSP10 with NSP14, so a further change in activity when complexed with other viral proteins is conceivable. High-quality single turnover kinetic analysis of the exonuclease complex in conjunction with the RdRp and other viral nonstructural proteins are required to address these questions about whether the substrate specificity of the NSP10/14 complex changes in conjunction with other binding partners. Moreover, the critical role of the proofreading exonuclease will rely on the relative rates of primer extension versus excision from a single multienzyme complex containing both activities, as shown for DNA polymerases.17,19 We also quantified rates of excision of remdesivir monophosphate incorporated into the primer strand of a duplex RNA and found that this analog was hydrolyzed at a rate similar to that for correctly base-paired RNA. While remdesivir is incorporated with higher efficiency than ATP by the viral RdRp,1 the high excision efficiency we observed here is slower than the rate of extension. Thus, remdesivir may be protected from the proofreading exonuclease by being rapidly buried by subsequent incorporation of normal nucleotides. By stalling polymerization only after being buried by three nucleotides, remdesivir would be protected from immediate removal by the exonuclease.1,2 Nonetheless, this protection is not absolute, and processive exonuclease activity could ultimately remove RMP. Further studies are currently underway to measure the kinetics of excision of molnupiravir, another promising nucleoside analog.28 While to the best of our knowledge no one has designed a nucleoside analog that evades excision by the exonuclease, it is possible that certain modifications could inhibit exonuclease hydrolysis. If such modifications were also efficiently incorporated by the polymerase, such a molecule would presumably be an attractive candidate for antiviral studies in vivo. Alternatively, exonuclease inhibitors, which have been successfully discovered in vitro,29 could be used in combination with nucleoside analog therapies to make both inhibitors more potent. Single turnover experiments using the methods outlined in this paper and our previous studies could prevent many of the pitfalls of steady-state experiments on these enzymes and provide direct, quantitative answers on the potential effectiveness of new inhibitors. Besides proofreading, the exonuclease activity of the NSP10/14 complex may have other roles in the life cycle of the virus. One study proposed a role in the translational shutdown of the infected host cell,30 although the mechanism of this remains unknown. When exonuclease activity is knocked out in cell culture experiments, SARS-CoV-2 curiously fails to replicate rather than simply accumulating more mutations31 as would be expected if the exonuclease activity only functions in proofreading. Interestingly, for reasons that are still unknown, this is not the case for the original SARS-CoV.32 One role of exonuclease activity that has been proposed is in viral recombination,31 which is consistent with our data where the enzyme would efficiently hydrolyze stretches of correctly base-paired RNA on different viral genomic templates allowing for efficient recombination. Efficient recombination is a hallmark of coronaviruses and could contribute to the large number of variants that have evolved relatively quickly in the COVID-19 pandemic. The necessity of the exonuclease complex for virus propagation makes this complex an attractive target for the design of next-generation exonuclease inhibitors that may boost the efficacy of currently available antiviral drugs. Studies outlined here provide a boilerplate for the analysis and evaluation of other nucleotide analogs designed to inhibit SARS-CoV-2 RNA replication. Materials and Methods Enzymes, Oligonucleotides, and Reagents Expression and Purification of NSP10/14 NSP10 and NSP14 were expressed and purified as previously described.29 Briefly, the coexpression of NSP10/14 with N-terminal Strep tag II and N-terminal 8xHis tag, respectively, in BL21(DE3) E. coli was induced with 0.5 mM IPTG at 16 °C for 20 h. Cells were harvested, lysed, then purified on a low-resolution Ni-NTA column (Qiagen), dialyzed, and purified on a high-resolution Ni-NTA column (HiTrap IMAC HP, GE). The complex was then further purified by size exclusion chromatography (Superdex 200, GE), flash-frozen, and stored at −80 °C. Expression and Purification of NSP12-His/NSP7L8 His-tagged NSP12 and NSP7L8 were coexpressed and purified as previously described.1 Briefly, the coexpression of 8xC-His-tagged NSP12 and NSP7L8 in BL21 E. coli/pG-Tf2 was induced with 0.5 mM IPTG, 10 ng/mL tetracycline, and 50 μg/mL nalidixic acid at 16 °C for 16 h. Harvested cells were lysed, clarified by centrifugation, and purified by Ni-NTA chromatography (His-Trap FF, Cytiva). Fractions containing the NSP12-His/NSP7L8 complex were flash-frozen in liquid nitrogen and stored at −80 °C. Oligonucleotides and Reagents RNA oligonucleotides were synthesized by Integrated DNA Technologies with RNase-free HPLC purification. The concentration of purified oligonucleotides was determined by absorbance at 260 nm using the extinction coefficients given in Table 2. Oligo stocks were stored in annealing buffer (10 mM tris–HCl pH 7, 50 mM NaCl, 0.1 mM EDTA) at −20 °C. Double-stranded RNA substrates were prepared by mixing the primer and template strands at a 1:1 molar ratio in annealing buffer, heating to 90 °C for 3 min, and then cooling slowly to room temperature over the course of 1 h. Remdesivir triphosphate (GS-443902, RTP) was kindly provided by Joy Feng and Brian Schultz of Gilead Sciences. The concentration of RTP was determined by absorbance at 245 nm using an extinction coefficient of 24,100 M–1 cm–1.1 All buffer components and other chemicals were purchased from either Sigma-Aldrich or Thermo Fisher Scientific. Table 2 Oligonucleotides Used in This Study oligo name sequence 5′–3′ extinction coefficient at 260 nm(M–1 cm–1) FAM-LS2 [6-FAM]—GUCAUUCUCCUAAGAAGCUA 222,360 LS1 CUAUCCCCAUGUGAUUUUAAUAGCUUCUUAGGAGAAUGAC 403,100 LS1.2-U CUAUCCCCAUGUGAUACGAUUAGCUUCUUAGGAGAAUGAC 401,600 LS1.2.2 CUAUCCCCAUGUGAUACGAAAAGCUUCUUAGGAGAAUGAC 404,600 Cy3–28mer DNA [Cy3]-CCGTGAGTTGGTTGGACGGCTGCGAGGC 266,800 Enzymatic Synthesis of Remdesivir-Containing RNA Substrate An RNA substrate containing a terminal remdesivir monophosphate was synthesized enzymatically using the following protocol. A solution of 2.5 μM NSP12-His/NSP7L8,1 5 μM NSP8, 200 nM FAM-LS2/LS1.2-U RNA, and 20 μM RTP was incubated for 90 min at room temperature to form approximately 92% product RNA containing a terminal remdesivir monophosphate. The reaction was heated to 95 °C for 5 min, cooled slowly to room temperature over the course of 1 h, and then centrifuged at 20,000g in a tabletop microcentrifuge for 20 min to pellet the precipitated protein. The supernatant was then used in kinetics experiments. Kinetics Experiments All kinetics experiments were performed at 37 °C in SARS-CoV-2 reaction buffer (40 mM tris–HCl pH 7, 50 mM NaCl, 5 mM MgCl2, 1 mM DTT).1,2 In some experiments, Mg2+ was only present in one syringe but at 2× the final concentration to measure the kinetics of hydrolysis of RNA already bound to the enzyme. Rapid quench experiments were performed using a KinTek RQF-3 (KinTek Corp) with reaction buffer in the drive syringes and 0.6 M EDTA in the quench syringe for a final concentration of 0.2 M after quenching. A circulating water bath was used for temperature control. Kinetics time points were analyzed by capillary electrophoresis on an ABI 3130xl Genetic Analyzer with a 36 cm array and a nanoPOP-6 polymer (Molecular Cloning Laboratories) at 65 °C. Samples were prepared for analysis by mixing 1 μL of sample with 10 μL of HiDi formamide (Thermo Fisher) containing a 28 nt Cy3-labeled DNA oligo internal standard for sizing. Samples were injected for 6–12 s, depending on the experiment, at 3.6 kV. The peak area was determined with GeneMapper software, and sizing and quantification were performed with a program written in house.2 Concentrations of reaction components given in the text are final concentrations after mixing unless otherwise noted. Experiments were repeated at least once to ensure reproducibility. Electropherograms from which figures in the main paper are derived are given in the Supporting Information. Data Fitting and Analysis Data fitting and analysis were performed using KinTek Explorer simulation and data fitting software v11 (www.kintekexplorer.com).33,34 This software was also used in preparing figures for kinetic data. Conventional data fitting was performed in the software using built-in functions. The equation for a single exponential is y = A0 + A1(1 – exp(−b1t)), where A0 is the y-value at time zero, A1 is the amplitude, b1 is the decay rate, and t is time. The equation for a double exponential is y = A0 + A1(1 – exp(−b1t)) + A2(1 – exp(−b2t)), where A0 is the y-value at time zero, A1 and A2 are the amplitudes of the first and second phases, b1 and b2 are the decay rates of the first and second phases, respectively, and t is time. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsbiomedchemau.2c00046.Data showing the enzyme concentration dependence of cleavage; electropherograms showing raw data for experiments described in the main text; global data fitting defining the kinetics of processive exonuclease reactions; and alternative (single-exponential) fit of remdesivir excision data (PDF) Supplementary Material bg2c00046_si_001.pdf Author Contributions CRediT: Tyler L. Dangerfield conceptualization, formal analysis, investigation (lead), methodology (lead), visualization (lead), writing-original draft (lead), writing-review & editing (equal); Kenneth A Johnson conceptualization (equal), formal analysis (equal), funding acquisition (lead), project administration (lead), software (lead), supervision (lead), writing-review & editing (equal). This work was supported by a grant from the National Institute of Allergy and Infectious Diseases (NIAID) to K.A.J. (NIH R01AI163336). The authors declare the following competing financial interest(s): K.A.J. is president of KinTek Corporation, which provided the RQF-3 rapid quench flow instrument and KinTek Explorer software used in this study. Acknowledgments The authors would like to thank Brian Schultz and Joy Feng at Gilead Sciences for kindly providing the remdesivir triphosphate used in this study and Jerome Deval and Cheng Liu at Aligos therapeutics for providing the purified NSP10/NSP14 used in this study, expressed and purified as described.29 ==== Refs References Dangerfield T. L. ; Huang N. Z. ; Johnson K. A. Remdesivir Is Effective in Combating COVID-19 because It Is a Better Substrate than ATP for the Viral RNA-Dependent RNA Polymerase. iScience 2020, 23 , 101849 10.1016/j.isci.2020.101849.33283177 Bravo J. P. K. ; Dangerfield T. L. ; Taylor D. W. ; Johnson K. A. Remdesivir is a delayed translocation inhibitor of SARS-CoV-2 replication. Mol. Cell 2021, 81 , 1548–1552.e4. 10.1016/j.molcel.2021.01.035.33631104 Baddock H. T. ; Brolih S. ; Yosaatmadja Y. ; Ratnaweera M. ; Bielinski M. ; Swift ; Lonnie P. ; Cruz-Migoni A. ; Fan H. ; Keown J. R. ; Walker A. P. ; et al. Characterization of the SARS-CoV-2 ExoN (nsp14ExoN–nsp10) complex: implications for its role in viral genome stability and inhibitor identification. Nucleic Acids Res. 2022, 50 , 1484–1500. 10.1093/nar/gkab1303.35037045 Frazier M. N. ; Riccio A. A. ; Wilson I. M. ; Copeland W. C. ; Stanley R. E. Recent insights into the structure and function of coronavirus ribonucleases. FEBS Open Bio 2022, 12 , 1567–1583. 10.1002/2211-5463.13414. Bouvet M. ; Debarnot C. ; Imbert I. ; Selisko B. ; Snijder E. J. ; Canard B. ; Decroly E. In Vitro Reconstitution of SARS-Coronavirus mRNA Cap Methylation. PLoS Pathog. 2010, 6 , e1000863 10.1371/journal.ppat.1000863.20421945 Lin S. ; Chen H. ; Chen Z. ; Yang F. ; Ye F. ; Zheng Y. ; Yang J. ; Lin X. ; Sun H. ; Wang L. ; et al. Crystal structure of SARS-CoV-2 nsp10 bound to nsp14-ExoN domain reveals an exoribonuclease with both structural and functional integrity. Nucleic Acids Res. 2021, 49 , 5382–5392. 10.1093/nar/gkab320.33956156 Liu C. ; Shi W. ; Becker S. T. ; Schatz D. G. ; Liu B. ; Yang Y. Structural basis of mismatch recognition by a SARS-CoV-2 proofreading enzyme. Science 2021, 373 , 1142–1146. 10.1126/science.abi9310.34315827 Moeller N. H. ; Shi K. ; Demir Ö. ; Belica C. ; Banerjee S. ; Yin L. ; Durfee C. ; Amaro R. E. ; Aihara H. Structure and dynamics of SARS-CoV-2 proofreading exoribonuclease ExoN. Proc. Natl. Acad. Sci. U.S.A. 2022, 119 , e2106379119 10.1073/pnas.2106379119.35165203 Yan L. ; Yang Y. ; Li M. ; Zhang Y. ; Zheng L. ; Ge J. ; Huang Y. C. ; Liu Z. ; Wang T. ; Gao S. ; et al. Coupling of N7-methyltransferase and 3′-5′ exoribonuclease with SARS-CoV-2 polymerase reveals mechanisms for capping and proofreading. Cell 2021, 184 , 3474–3485.e11. 10.1016/j.cell.2021.05.033.34143953 Ma Z. ; Pourfarjam Y. ; Kim I. K. Reconstitution and functional characterization of SARS-CoV-2 proofreading complex. Protein Expression Purif. 2021, 185 , 105894 10.1016/j.pep.2021.105894. Riccio A. A. ; Sullivan E. D. ; Copeland W. C. Activation of the SARS-CoV-2 NSP14 3′-5′ exoribonuclease by NSP10 and response to antiviral inhibitors. J. Biol. Chem. 2022, 298 , 101518 10.1016/j.jbc.2021.101518.34942146 Shannon A. ; Le N. T. ; Selisko B. ; Eydoux C. ; Alvarez K. ; Guillemot J. C. ; Decroly E. ; Peersen O. ; Ferron F. ; Canard B. Remdesivir and SARS-CoV-2: Structural requirements at both nsp12 RdRp and nsp14 Exonuclease active-sites. Antiviral Res. 2020, 178 , 104793 10.1016/j.antiviral.2020.104793.32283108 Johnson K. A. Kinetic Analysis for the New Enzymology; KinTek Corporation, 2019. Lee H. R. ; Helquist S. A. ; Kool E. T. ; Johnson K. A. Importance of hydrogen bonding for efficiency and specificity of the human mitochondrial DNA polymerase. J. Biol. Chem. 2008, 283 , 14402–14410. 10.1074/jbc.M705007200.17650502 Dangerfield T. L. ; Johnson K. A. Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid. J. Biol. Chem. 2021, 296 , 100143 10.1074/jbc.RA120.016617.33273013 Hanes J. W. ; Johnson K. A. Analysis of single nucleotide incorporation reactions by capillary electrophoresis. Anal. Biochem. 2005, 340 , 35–40. 10.1016/j.ab.2005.02.013.15802127 Donlin M. J. ; Patel S. S. ; Johnson K. A. Kinetic partitioning between the exonuclease and polymerase sites in DNA error correction. Biochemistry 1991, 30 , 538–546. 10.1021/bi00216a031.1988042 Reha-Krantz L. J. ; Marquez L. A. ; Elisseeva E. ; Baker R. P. ; Bloom L. B. ; Dunford H. B. ; Goodman M. F. The Proofreading Pathway of Bacteriophage T4 DNA Polymerase. J. Biol. Chem. 1998, 273 , 22969–22976. 10.1074/jbc.273.36.22969.9722519 Dangerfield T. L. ; Kirmizialtin S. ; Johnson K. A. Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase. J. Biol. Chem. 2022, 298 , 101627 10.1016/j.jbc.2022.101627.35074426 Dangerfield T. L. ; Huang N. Z. ; Johnson K. A. High throughput quantification of short nucleic acid samples by capillary electrophoresis with automated data processing. Anal. Biochem. 2021, 629 , 114239 10.1016/j.ab.2021.114239.33979658 Banerjee D. ; Tateishi-Karimata H. ; Ohyama T. ; Ghosh S. ; Endoh T. ; Takahashi S. ; Sugimoto N. Improved nearest-neighbor parameters for the stability of RNA/DNA hybrids under a physiological condition. Nucleic Acids Res. 2020, 48 , 12042–12054. 10.1093/nar/gkaa572.32663294 Russo P. ; Tacconelli E. ; Olimpieri P. P. ; Celant S. ; Colatrella A. ; Tomassini L. ; Palù G. Mortality in SARS-CoV-2 Hospitalized Patients Treated with Remdesivir: A Nationwide, Registry-Based Study in Italy. Viruses 2022, 14 , 1197 10.3390/v14061197.35746668 Freemont P. S. ; Friedman J. M. ; Beese L. S. ; Sanderson M. R. ; Steitz T. A. Cocrystal Structure of an Editing Complex of Klenow Fragment with DNA. Proc. Natl. Acad. Sci. U.S.A. 1988, 85 , 8924–8928. 10.1073/pnas.85.23.8924.3194400 Dodd T. ; Botto M. ; Paul F. ; Fernandez-Leiro R. ; Lamers Meindert H. ; Ivanov I. Polymerization and editing modes of a high-fidelity DNA polymerase are linked by a well-defined path. Nat. Commun. 2020, 11 , 5379 10.1038/s41467-020-19165-2.33097731 Lamichhane R. ; Berezhna S. Y. ; Gill J. P. ; Van der Schans E. ; Millar D. P. Dynamics of Site Switching in DNA Polymerase. J. Am. Chem. Soc. 2013, 135 , 4735–4742. 10.1021/ja311641b.23409810 Lieberman K. R. ; Dahl J. M. ; Wang H. Kinetic Mechanism at the Branchpoint between the DNA Synthesis and Editing Pathways in Individual DNA Polymerase Complexes. J. Am. Chem. Soc. 2014, 136 , 7117–7131. 10.1021/ja5026408.24761828 Matsuda A. ; Plewka J. ; Chykunova Y. ; Jones A. N. ; Pachota M. ; Rawski M. ; Mourão A. ; Karim A. ; Kresik L. ; Lis K. ; Despite the odds: formation of the SARS-CoV-2 methylation complex. Submitted 2022-03-01. bioRxiv https://www.biorxiv.org/content/10.1101/2022.01.25.477673v2 (accessed 2022-07-01). Kabinger F. ; Stiller C. ; Schmitzova J. ; Dienemann C. ; Kokic G. ; Hillen H. S. ; Hobartner C. ; Cramer P. Mechanism of molnupiravir-induced SARS-CoV-2 mutagenesis. Nat. Struct. Mol. Biol. 2021, 28 , 740–746. 10.1038/s41594-021-00651-0.34381216 Scholle M. D. ; Liu C. ; Deval J. ; Gurard-Levin Z. A. Label-Free Screening of SARS-CoV-2 NSP14 Exonuclease Activity Using SAMDI Mass Spectrometry. SLAS Discovery 2021, 26 , 766–774. 10.1177/24725552211008854.33870746 Hsu J. C.-C. ; Laurent-Rolle M. ; Pawlak J. B. ; Wilen C. B. ; Cresswell P. Translational shutdown and evasion of the innate immune response by SARS-CoV-2 NSP14 protein. Proc. Natl. Acad. Sci. U.S.A. 2021, 118 , e2101161118 10.1073/pnas.2101161118.34045361 Ogando N. S. ; Zevenhoven-Dobbe J. C. ; van der Meer Y. ; Bredenbeek P. J. ; Posthuma C. C. ; Snijder E. J. The Enzymatic Activity of the nsp14 Exoribonuclease Is Critical for Replication of MERS-CoV and SARS-CoV-2. J. Virol. 2020, 94 , e01246-01220 10.1128/JVI.01246-20. Ogando N. S. ; Ferron F. ; Decroly E. ; Canard B. ; Posthuma C. C. ; Snijder E. J. The Curious Case of the Nidovirus Exoribonuclease: Its Role in RNA Synthesis and Replication Fidelity. Front. Microbiol. 2019, 10 , 1813 10.3389/fmicb.2019.01813.31440227 Johnson K. A. ; Simpson Z. B. ; Blom T. Global kinetic explorer: a new computer program for dynamic simulation and fitting of kinetic data. Anal. Biochem. 2009, 387 , 20–29. 10.1016/j.ab.2008.12.024.19154726 Johnson K. A. Fitting enzyme kinetic data with KinTek global kinetic explorer. Methods Enzymol. 2009, 467 , 601–626. 10.1016/S0076-6879(09)67023-3.19897109	0	PMC9718090	NO-CC CODE	2022-12-05 23:15:02	no	ACS Bio Med Chem Au. 2022 Nov 16;:acsbiomedchemau.2c00046	utf-8	ACS Bio Med Chem Au	2,022	10.1021/acsbiomedchemau.2c00046	oa_other
==== Front Anal Chem Anal Chem ac ancham Analytical Chemistry 0003-2700 1520-6882 American Chemical Society 36399654 10.1021/acs.analchem.2c03813 Article Accurate Diagnosis of COVID-19 from Self-Collectable Biospecimens Using Synthetic Apolipoprotein H Peptide-Coated Nanoparticle Assay Kang Junwon †‡ Jang Haewook † Kim Tae Hyun §∥ Cho Untack ⊥# Bang Hyeeun ⊥ Jang Jisung ⊥ Lee Wooseok ∥ Joo Hyelyn † Noh Jinsung §∥ Lee Gi Yoon ∥ Shin Dong Hoon ¶ Kang Chang Kyung ¶ Choe Pyoeng Gyun ¶ Kim Nam Joong ¶ Oh Myoung-don ¶ Song Manki ∇ https://orcid.org/0000-0003-3514-1738 Kwon Sunghoon †§∥⊥†† Veas Francisco ○⧫△ Park Wan Beom ¶ † Interdisciplinary Program in Bioengineering, Seoul National University, Seoul08826, Korea ‡ Integrated Major in Innovative Medical Science, Seoul National University, Seoul03080, Korea § Bio-MAX Institute, Seoul National University, Seoul08826, Korea ∥ Department of Electrical and Computer Engineering, Seoul National University, Seoul08826, Korea ⊥ QuantaMatrix Inc., Seoul08506, Korea # Interdisciplinary Program in Cancer Biology, Seoul National University College of Medicine, Seoul03080, Korea ¶ Department of Internal Medicine, Seoul National University College of Medicine, Seoul03080, Korea ∇ International Vaccine Institute, Seoul08826, Korea ○ Copernicus Integrated Solutions for Biosafety Risks (CISBR), Mauguio34130, France ⧫ ApoH-Technologies, 94 Allée des Fauvettes, La Grande Motte34280, France †† Biomedical Research Institute, Seoul National University Hospital, Seoul03080, Korea △ UMR5151/French Research Institute for Development (IRD), University of Montpellier (UM), Montpellier34093, France Email: [email protected]. * Email: [email protected]. * Email: [email protected]. 18 11 2022 acs.analchem.2c0381331 08 2022 02 11 2022 © 2022 American Chemical Society 2022 American Chemical Society This article is made available via the PMC Open Access Subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. A high-throughput, accurate screening is crucial for the prevention and control of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Current methods, which involve sampling from the nasopharyngeal (NP) area by medical staffs, constitute a fundamental bottleneck in expanding the testing capacity. To meet the scales required for population-level surveillance, self-collectable specimens can be used; however, its low viral load has hindered its clinical adoption. Here, we describe a magnetic nanoparticle functionalized with synthetic apolipoprotein H (ApoH) peptides to capture, concentrate, and purify viruses. The ApoH assay demonstrates a viral enrichment efficiency of >90% for both SARS-CoV-2 and its variants, leading to an order of magnitude improvement in analytical sensitivity. For validation, we apply the assay to a total of 84 clinical specimens including nasal, oral, and mouth gargles obtained from COVID-19 patients. As a result, a 100% positivity rate is achieved from the patient-collected nasal and gargle samples, which exceeds that of the traditional NP swab method. The simple 12 min pre-enrichment assay enabling the use of self-collectable samples will be a practical solution to overcome the overwhelming diagnostic capacity. Furthermore, the methodology can easily be built on various clinical protocols, allowing its broad applicability to various disease diagnoses. Seoul National University 10.13039/501100002551 0320200330 Quantamatrix Inc. NA NA Bio-MAX Institute, Seoul National University NA NA Ministry of Science and ICT, South Korea 10.13039/501100014188 NA National Research Foundation of Korea 10.13039/501100003725 NRF-2020R1A3B3079653 National Research Foundation of Korea 10.13039/501100003725 NRF-2020R1A2C1007242 National Research Foundation of Korea 10.13039/501100003725 2020M3H1A1073304 document-id-old-9ac2c03813 document-id-new-14ac2c03813 ccc-price This article is made available via the ACS COVID-19 subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. ==== Body pmcIntroduction Despite the varying levels of public health efforts being made to prevent the rapid spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), a considerable number of new infected cases and deaths continue to arise globally.1 Since the World Health Organization declared coronavirus disease 2019 (COVID-19) outbreak as a pandemic, various forms of vaccines and therapeutic reagents have been actively developed and clinically validated for emergency use approval.2 However, the emergence and rapid succession of new variants along with growing evidences showing that some mutated types often evade immune responses produced by prior vaccinations or infections raise concerns existentially.3 As the social and economic burdens extend, demands and attempts to gradually reopen business and schools regardless of the recurrent outbreak have become inevitable. To effectively contain the local viral transmission, while returning to daily life, large-scale diagnostic tests to identify, trace, and isolate infected individuals are mandated.4,5 The current standards to detect SARS-CoV-2 involve testing nasopharyngeal (NP) swab samples using quantitative reverse transcription-polymerase chain reaction (RT-qPCR).6 Approaches using isothermal amplification or extraction-free methods have been proposed, with emphasis on simplifying the procedure and reducing the turnaround time.7−9 However, the delay in obtaining test results mainly occurs not only in the time required for a single assay. More importantly, the overall capacity to sample and perform multiplexed reactions appears to be a great hurdle for regular base population-level screening.10,11 Due to the quality of specimens, which highly depends on the proper swabbing technique and safety issues, NP swab sampling requires the assistance of trained professionals.12 Consequently, this leads to an increase in the wait time and gatherings at each testing location which also escalates the risk of exposure to SARS-CoV-2. On the other hand, the absence of medical staffs being reassigned to areas to support COVID-19 severely disrupts the general healthcare services. Diagnostic tests utilizing self-collectable samples such as nasal swab, oral swab, saliva, or mouth gargle are being evaluated as potential alternatives to resolve the accessibility and throughput constraints, negating the needs for human-to-human interactions, and improve patient compliance.10,13−19 In spite of these advantages, the viral load associated with disease progression and severity is typically low,14−16 which constitutes the possibility of generating false-negative results. Recently, some studies have shown that self-collectable specimens often have similar or higher viral material compared to the NP swabs.10,17,18 However, it has been found that the total viral load can vary dynamically, not only between sampling locations, but also according to the time of collection after infection.19−21 Along with these studies, several attempts have been made to improve the analytical sensitivity or testing capacity, for instance, utilizing next-generation sequencing;11 however, the additional time and cost required per operation have hindered their widespread clinical adoption for SARS-CoV-2 diagnosis. To expedite the use of self-collectable samples to our current COVID-19 screening practice as an NP substitute, continuous efforts to enhance the assay sensitivity accompanied by the exploration of the correlation of viral loads between sampling sites are necessary. In this study, we present a synthetic apolipoprotein H (ApoH) peptide-coated magnetic nanoparticle that can efficiently bind to live viruses. Considering that biological specimens are eluted and stored in 2–4 mL of viral transport media prior to RT-qPCR and the maximum reaction volume of a typical PCR machine used in clinical laboratories is around <100 μL per slot, in practice, only 2.5–5% of the total samples is subjected for testing. We use the ApoH nanoparticle assay to capture, purify, and concentrate viral particles and demonstrate the detection of SARS-CoV-2 with a 10-fold increase in analytical sensitivity. The synthetic ApoH peptide-coated nanoparticle differs from other immunomagnetic based strategies such as antibody-coated nanobeads as it can bind to and cover a wide range of viral species including its variants. Next, we measure the viral loads from the nasal and oral swabs and mouth gargles collected by the patients themselves along with NP and oropharyngeal (OP, throat) swabs taken by healthcare workers simultaneously. We discover that the total viral load obtained from the self-collectable specimens is, on average, significantly lower than that of the NP swab samples which have, so far, questioned its clinical adoption. However, by applying the assay, we realize a 100% positivity rate of detecting SARS-CoV-2 from both patient-collected nasal swab and mouth gargles which were unable to achieve even through conventional NP swab-based testing methods, regardless of the large variation of the sample collection quality. Our work finally provides a pragmatic solution of implementing self-collection of biospecimens into current COVID-19 testing standards by overcoming the false-negative concerns. Moreover, with the unique binding characteristics of the synthetic ApoH peptide to a broad spectrum of viruses, ultimately, the assay will resolve the capacity limit of diagnosis for COVID-19 as well as for the forthcoming pandemic and infectious diseases. Experimental Section Clinical Specimen Collection Clinical samples were obtained from hospitalized patients, who were confirmed positive for SARS-CoV-2, after obtaining informed consent for research testing that complied with the ethical regulations under an Institutional Review Board (IRB)-approved protocol (2009-098-1157) at the Seoul National University Hospital (SNUH). Written informed consent was obtained from all patients. All specimens including NP, OP, nasal, and oral swab samples and mouth gargles were collected at the same time and stored at 4 °C before processing. NP and OP swab samples were collected by clinicians, whereas the nasal and oral swabs and mouth gargles were obtained by the patients themselves. NP swabs were taken from the posterior nasopharynx via the nostril, whereas the OP swabs were taken from the pharyngeal tonsil and posterior pharynx, excluding the tongue. The swabs were slowly rotated in place, removed, and inserted in a transfer tube containing 2 mL of 0.9% PBS (10010-023, Gibco). For the self-collectable nasal and oral swab samples, all participants were first asked to clean their hands and cough three to five times. The patients were then asked to gently insert the swab either into the nostril (nasal) or the floor of the mouth (oral), including the outer gums and inside the cheeks, and slowly rotate 10 times. Then, similar to the NP and OP specimens, these swabs were inserted into a transfer tube. Finally, in the case of mouth gargles, the patients were instructed to hold 10 mL of saline solution in their mouth, gargle the solution for 5 s and repeat by tilting their head back, and expel the content into an empty 50 mL sterile tube. SARS-CoV-2 Enrichment Using ApoH Nanoparticles ApoH nanoparticles and a custom capture buffer (75 μL used for 1 mL of sample) were added to the sampled specimen and gently mixed by inverting the tube three times. The primary role of the capture buffer was to ensure that the solution was kept at neutral pH. Considering the isoelectric point of SARS-CoV-2 and its variants,22 these viruses bear negative charge at pH = 7. As the functional region of the ApoH peptide presents a positive charge on its surface, a tight control of the solution’s pH to neutrality was important. The samples were placed in a thermomixer (KTM-100C, KBT), set at 4 °C with a rotation speed of 800 rpm, for 5 min to enable an efficient binding of ApoH nanoparticles to the virions. Next, the sample tubes were placed in a magnetic rack (DynaMag Magnet, Invitrogen) for 5 min for particle enrichment. After removing the supernatant, the magnet was removed, and the remaining content consisting of ApoH-bound viral particles was used for the following RNA extraction and RT-qPCR. In the case of mouth gargles, the samples were treated with 3% NAC solution (850 μL used for 1 mL of sample, A7250, Sigma) and incubated for 30 min at 37 °C prior to the virus enrichment procedure mentioned above. RNA Extraction and RT-qPCR Following virus enrichment, RNA was extracted using the QIAamp Viral RNA Mini kit (Qiagen) according to the manufacturer’s instruction. In brief, the ApoH particle-bound viruses were resuspended using 560 μL of lysis buffer and incubated for 10 min at 37 °C. Next, the solution was placed in a magnetic rack for 5 min. The supernatant containing the viral nucleic acids, separated from the ApoH nanoparticles, was collected and purified. Finally, the extracted RNA was eluted and stored at −70 °C before analysis. RT-qPCR was performed using the QPLEX COVID-19 test (QMCOVID02, QuantaMatrix Inc.). A mixture containing 10 μL of 2× One-step Premix, 5 μL of Oligo Mix, and 5 μL of RNA sample was prepared and used for complementary DNA (cDNA) synthesis using reverse transcription and amplification of the target genes in the same reaction tube. The primer and probe sets used in the kit were designed to target the RdRp and E genes of SARS-CoV-2. Of note, the primers were applied to 22 different types of other respiratory viruses and bacteria to test the specificity of the assay (Table S1). After five cycles of pre-amplification, the product was subjected to 35 cycles of PCR for measurement. The positivity of infection was defined when both RdRp and E genes were detected; inconclusive when either gene target was detected; and negative when both were undetectable.23 Results Design and Synthesis of ApoH-Coated Nanoparticles The ApoH protein, also known as beta-2 glycoprotein 1, is a plasma glycoprotein which is predominantly expressed in the liver. Although, the functions of ApoH are not yet fully understood, it appears to be involved in blood coagulation pathways and immune responses.24 An interesting property of ApoH is that it interacts with hydrophobic or negatively charged substances with high affinity, such as anionic phospholipids, heparin, activated platelets, and apoptotic cells.25−28 Also, we and others have previously shown that ApoH binding frequently occurs with infectious agents like bacteria or viruses, making the protein promising for use in various diagnostic approaches.29,30 Inspired by these aspects, a synthetic ApoH peptide was designed and engineered by mimicking the corresponding functional regions of the ApoH protein. In this form, the size and the molecular complexity of the virus-binding domain were minimized, allowing the ApoH peptide to be densely packed to functionalized surfaces, resulting in a 27.29% improvement in the virus capture efficiency (Figure S1). Moreover, the protein-by-protein variation was controlled, allowing reliability and repeatability in its production. The synthesized peptide was then attached and functionalized to magnetic nanoparticles of a mean diameter of 200 nm. The ApoH nanoparticles were used for SARS-CoV-2 enrichment following a simple protocol which consists of ApoH particle binding and magnetic separation (Figure 1). During the virus-binding process, a capture buffer was added to the solution to support the viral capture and cell lysis to extract the virus from the infected cells for enrichment. Figure 1 Schematic of SARS-CoV-2 enrichment procedure using ApoH nanoparticles. (a) Virus sampling locations available for COVID-19 testing. Samples that can be self-collected are outlined in red. (b) Molecular structure of the ApoH protein (left) and illustration of a synthetic ApoH peptide-coated nanoparticle bound to a SARS-CoV-2 virion (right). (c) Strategy to enhance the virus detection efficiency. The total amount of viruses in the sampled specimen can be concentrated using ApoH nanoparticles, overcoming the traditional RT-qPCR volume limit, which results in the increase of analytical sensitivity and detection rate. (d) SARS-CoV-2 enrichment workflow. The entire procedure consists of ApoH nanoparticle binding, magnetic separation, and removal of the supernatant to concentrate and collect the viruses. The total time required for the enrichment assay is less than 12 min, and the entire viral detection procedure is completed within 2 h. SARS-CoV-2 Enrichment Using ApoH-Coated Nanoparticles To optimize the viral enrichment procedure, mixtures of different volume ratios of solutions containing ApoH nanoparticles (1.9 × 1010 particles/μL) to samples spiked with 125 plaque-forming units (PFUs) of SARS-CoV-2 virions in diethylpyrocarbonate (DEPC)-treated water were prepared. After collecting the viruses following ApoH nanoparticle binding (Figure S2) and magnetic separation, each sample was subjected to RT-qPCR to estimate the recovery rate. The primers used in the reaction were designed to specifically target the RNA-dependent RNA polymerase (RdRp) and envelope (E) genes (Table S1 and Figure S3). As a result, a particle concentration above 4.75 × 1010 particles/mL was determined to be deemed optimal for efficient virus retrieval (Figure 2a). Figure 2 Characterization of SARS-CoV-2 enrichment performance of ApoH nanoparticles. (a) Optimization of the ApoH nanoparticle concentration for SARS-CoV-2 capture. (b) Virus capture efficiency as a function of processing volume. A constant amount of 125 PFU viruses are spiked into different volumes of solution. (c) Improvements in the CT value after applying the ApoH assay to varying amounts of sample volume. Samples are spiked with a constant concentration of 125 PFU/mL viruses. ΔCT is calculated by subtracting the CT values from each enrichment test to the CT value obtained from a 140 μL sample, which represents the conventional RT-qPCR procedure. Comparison between the RT-qPCR detection limit, with and without using the ApoH nanoparticles, for (d) RdRp and (e) E genes. (f) Variant independent binding of ApoH nanoparticles. The average ΔCT of RdRp gene for each alpha, beta, and gamma variant after ApoH enrichment is presented in a line plot (red) above. To quantitatively assess the volume-dependent virus capture efficiency, a fixed number of 125 PFU virions was again dispensed into varying amount of sample volume ranging from 1 to 10 mL (Figure 2b). Using the optimized concentration of ApoH nanoparticles for capture, the CT values of each sample were measured before and after the enrichment process. The total number of RNA copies was derived from CT values by converting a single infectious unit (PFU) to 1 × 104 copies of viral RNA to calculate the capture efficiency.31 In total, an average of 90.32 ± 2.90% and 93.20 ± 4.12% capture efficiency was achieved for RdRp and E gene, respectively, revealing the successful recovery of virus, regardless of the processing volume. Similar results were also derived through a different quantification method based on the plaque assay, showing a total capture efficiency of 96.0 ± 4.45% (Figure S4). Next, to demonstrate the viral concentration using the ApoH assay for improved COVID-19 detection accuracy, a solution containing 125 PFU/mL of SARS-CoV-2 was divided into aliquots of 1, 2, 4, and 10 mL and tested (Figure 2c). Compared to the CT cycles obtained from a 140 μL suspension which refers to the maximum volume that a conventional PCR machine can accommodate, a substantial reduction in CT value was observed with the increasing sample volume used. The increase in ΔCT verifies that the loss of viral RNA due to limited reaction volume during RT-qPCR can be significantly reduced through the viral concentration. Finally, the limit of detection (LoD) of the assay was evaluated (Figure 2d,e). Based on the 33 cycles defined as the PCR cutoff, a 10-fold increase in analytical sensitivity was achieved. In addition, the ApoH assay was performed against several SARS-CoV-2 mutations, including the alpha, beta, and gamma variants (Figures 2f and S5). Each type of viruses was spiked in 2 mL of DEPC-treated water at a concentration of approximately 12.5 PFU/mL and processed. Similar to the original COVID-19 virus, the ApoH nanoparticles efficiently bound to all variants, exhibiting an average CT reduction of 3.04 ± 0.14 and 3.12 ± 0.27 for RdRp and E gene, respectively. The variant-independent binding behavior illustrates the broad applicability and efficacy of the ApoH assay for the sensitive diagnosis of viral infections. Before applying the assay directly to clinical specimens, the viral enrichment procedure was validated using nasal swab, oral swab, and mouth gargles obtained from healthy individuals. Here, mouth gargles were chosen over saliva due to the ease of collection, patient comfort, and to neglect the additional preprocessing steps required to digest the residual substances such as sputum (Figure S6). Considering that the majority of viruses from COVID-19 patients reside within the cells infected through continuous replication and propagation (assumed to be around 105 virions per cell),32 first, the cell lysis efficiency of the capture buffer was inspected (Figure S7a,b). The capture buffer of 150 μL was added to 2 mL of phosphate-buffered saline (PBS) containing 2 × 106 of Vero cells and slightly vortexed. A portion of cells was sequentially sampled after 5, 10, 30, and 60 min to observe the amount of cell membrane rupture using a live/dead viability assay. An average of >99% of cells was completely permeable within 5 min, allowing the viruses to be extracted and exposed in suspension for ApoH nanoparticle binding. Next, each nasal and oral swab was eluted in 2 mL of PBS, and mouth gargles were collected by rinsing out 10 mL of saline solution in an empty 15 mL sterile tube. PBS was selected as a sample collection and transport medium based on the 4 day virus preservation test at 4 °C in which negligible amount of RNA degradation was observed (Figure S8). To mimic the presence of viral infection in SARS-CoV-2 patients, 125 PFUs of live viruses were spiked into these samples for the test. Here, the amount of ApoH nanoparticles was slightly adjusted to a concentration of 1.9 × 1011 particles/mL to ensure sufficient binding to viruses which may be affected by impurities such as epithelial cells. As expected, a significant decrease in CT values was noticed in all sample types after viral concentration (Figure S7c–e). Clinical Validation of the ApoH Nanoparticle Assay Following the ApoH nanoparticle characterization, the assay was applied to a total of 84 clinical specimens obtained from 19 hospitalized patients who were confirmed positive to SARS-CoV-2 infection. Based on the patient consent available, NP and OP swabs sampled by medical staff members as well as nasal swab, oral swab, or mouth gargles collected by the patient themselves were subjected to the test. Similar to the approach above, the swab samples were eluted in 2 mL of PBS, and mouth gargles were collected after rinsing out 10 mL of saline solution in a 15 mL tube. However, this time, 140 μL of the sampled specimens was separated and directly ran through RT-qPCR, following the current diagnostic procedure used for COVID-19 detection, and later, the remaining volume was used for ApoH capture. All samples were processed within 6 h upon collection. To investigate the correlation between the sampling sites and determine the potential candidate among the self-collected specimens that could be used to diagnose SARS-CoV-2 most effectively, the viral loads from the 140 μL aliquots were analyzed in pairs (Figure 3a–c). Within the patient cohort studied, the average viral loads found in the self-collected nasal swab, oral swab, and mouth gargles were substantially low compared to the matched NP samples (Figure 3d–g). Surprisingly, even in OP swabs, which are often used in some COVID-19 screening kits, the number of viral materials was only comparable to the self-collected nasal swabs. Considering the relation to the NP specimen, the highest correlation was found in the nasal swab samples. As a result, a higher number of false-negatives were observed in all samples compared to the NP specimen. However, in the case of mouth gargles, given that only 1.4% of the sample volume was analyzed, the total amount of virus contained in the 10 mL collection was estimated to be comparable to the NP swab samples. Figure 3 Correlation of viral loads between sampling sites. Matched-paired viral loads of NP, OP, nasal, and oral swabs and mouth gargles measured using (a) RdRp and (b) E genes. Bold line indicates the median viral load obtained from each sampling location. (c) Heatmap of correlation coefficient between sample types. Viral RNA copy number from (d) OP (n = 18), (e) nasal (n = 19), (f) oral swab (n = 19), and (g) mouth gargles (n = 9) compared to NP swab samples. Each data point reflects samples that are collected from the same patient at the same time. Regression lines for RdRp and E genes are expressed in blue and orange dotted lines, respectively. ρ represents Spearman’s rank correlation coefficient. Finally, the ApoH nanoparticles were applied to the remaining swab samples and mouth gargles of 1.84 and 9.86 mL, respectively, to evaluate the accuracy and enhancement of detecting COVID-19 compared to the traditional RT-qPCR method measured from the 140 μL aliquots (Figures 4a and S9). The CT values from all samples decreased significantly, demonstrating the benefit of ApoH nanoparticles for improved analytical sensitivity. The largest difference in CT values was shown in mouth gargles, particularly due to the increased sampling volume of 10 mL, compared to the swab specimens. Interestingly, the virus concentration using the ApoH assay was more efficiently analyzed in samples with lower viral loads (Figure S10). The noticeable difference in CT numbers also led to changes in positivity rate for COVID-19 detection (Figure 4b–f). The incidence of false-negative cases decreased dramatically from all sampling locations. Specifically, for the NP swab, nasal swab, and gargle samples, 5.3, 26.3, and 33.3% of the false-negative cases, respectively, all turned positive after applying the ApoH assay. Figure 4 Diagnosis of COVID-19 from hospitalized patients using the ApoH assay. (a) Comparison between matched-paired CT values of the RdRp gene measured from NP, OP, nasal, and oral swabs, and mouth gargles, with (red) and without (blue) applying the ApoH assay. Bold line indicates the median CT values obtained from each sampling location. Change in CT values and COVID-19 positivity rate from (b) NP (n = 19), (c) OP (n = 18), (d) nasal (n = 19), and (e) oral (n = 19) swabs and (f) mouth gargle (n = 9), with and without using the ApoH assay. Paired samples are connected in lines. The gray area represents the PCR cutoff of CT = 33. Green, orange, and yellow areas in the pie chart indicate the proportion of positive, negative, and inconclusive test results, respectively. (p-values: < 0.01; * < 0.001, the Wilcoxon signed-rank test). Discussion Despite various strategies that are being developed to improve the throughput for COVID-19 diagnosis, the sampling procedure which involves NP swabbing constitutes a huge barrier of scaling up the overall testing capacity. We explore the possibility of utilizing self-collectable specimens that could easily be acquired noninvasively from home, to avoid the needs of timely in-person visits to designated examination areas, mitigating the fundamental hurdles of population-level screening. The synthetic ApoH nanoparticle we describe enables the sensitive capture and concentration of live viruses from diluted biological specimens which significantly increase the number of virions subjected to RT-qPCR analysis. The improvements made through the simple 12 min protocol allow a 10-fold increase in analytical sensitivity compared to the regular RT-qPCR process performed without prior enrichment steps, permitting the accurate diagnosis of SARS-CoV-2 from the nasal and oral swab and gargle samples which typically suffer from insufficient viral loads. Surprisingly, by testing the nasal swab or mouth gargles collected by COVID-19 patients themselves, the ApoH assay shows a 100% detection rate which outperforms the use of conventional NP swab-based testing methods. Various types of self-collectable biospecimens have been exploited for their use in SARS-CoV-2 diagnosis. However, due to the lack of standardized protocols, differing groups of patient cohorts, and samples being tested between studies, the optimal sampling site that holds the largest, stable amount of viral material has remained unclear.16,33 From our matched-paired analysis, swabs taken from the nasal cavity revealed the highest correlation in viral load to the NP area, at the same time, demonstrating the lowest false-negative rate of 26.3%. However, the fact that the detection rate cannot be reached up to the level of NP specimen, even from the nasal swab sample itself, questions the adoption of using self-collectable specimens for clinical purposes. On the other hand, using the ApoH assay, the amount of viral material has significantly been increased, particularly in mouth gargles which instantly became comparable to the NP swabs. The effect of virus concentration has most apparently been observed in gargles due to the large collection volume. Consequently, together with nasal swabs, the false-negative rate from using mouth gargles decreased dramatically down to 0%, presenting potentials as a NP swab substitute. Although the approach in our study focuses on utilizing self-collectable biospecimens, the ApoH assay can also be selectively applied to and benefit the most widely used NP-based testing methods in situations where the test accuracy is critical. Recent studies have revealed that the false-negative rate produced by NP swab samples can range from 15–30%, which is yet considered high regarding the disease infectivity.34,35 This percentage further increases during the presymptomatic phase in which the viral material is relatively insufficient but highly contagious.36 Also, unlike previously emerged coronaviruses, epidemiological studies have indicated that the infectious period for SARS-CoV-2 frequently begins prior to the symptom onset, approximately 2 days after the initial exposure, making the disease highly transmissible.37 Thus, failures to identify such pauci-symptomatic or even asymptomatic subjects during the early spread can lead to an explosive transmission of SARS-CoV-2 through super-spreaders.36,38 Furthermore, the occurrence of false-negative results can raise particular concerns in places such as hospitals or intermediate care facilities, where a group of high-risk individuals are present. Studies have shown that vulnerable patients with older age, comorbidities, impaired immune function, or pregnancy are found to contain low viral loads due to their prolonged nuclear acid conversion (viral shedding).39,40 Therefore, sensitivity improvements made by implementing the ApoH assay to the standard NP-based surveillance tests can contribute to mitigate the local surge of COVID-19 and prevent the collapse of the overall healthcare systems. With efforts to reduce the cost and number of RT-qPCR for population-level screening, several pooling strategies have been proposed and developed to scale up the COVID-19 testing capacity.11,41 However, as each pool contains multiple samples in a single pot, it seems inevitable to dilute the viral loads contained in individual specimens. This inherently restricts the available number of samples being included in a single pool. We propose that the ApoH assay can address the trade-off between the analytical sensitivity and throughput caused by the PCR volume limit. Taking into account that the current NP swab testing methods utilize a specimen volume of around 20–100 μL per reaction, the successful demonstration of concentrating viruses from a 10 mL mouth gargle implies that a total number of samples of more than 2 orders of magnitude can be pooled together and processed without sacrificing the sensitivity. This opens up opportunities to realize the pooled-test method for accurate screening and massive surveillance tests. Conclusions As the disease evolves and spread continues, emphasis is being placed on establishing new strategies or examination guidelines to overcome the capacity limit of current COVID-19 screening practices. In this regard, the capability to diagnose SARS-CoV-2 using self-collectable specimens and to perform multiplexed sample reaction without affecting the test accuracy using the ApoH assay can be a practical solution for a scalable surveillance option. To accelerate the deployment of our technology to clinical fields, further validation over extended patient cohorts must be followed. Nevertheless, from the results featuring higher enrichment efficiency at lower viral loads, we anticipate that the ApoH assay will also be able to accommodate groups of presymptomatic patients that were considered difficult to be diagnosed using conventional methods. Finally, beyond SARS-CoV-2 and its variants, the high affinity of ApoH nanoparticles against a variety of viruses and pathogens, allows us to easily expand our approach to multiple diseases including future species with pandemic potential and also integrate with other clinical protocols. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.2c03813.Description of cell/virus culture, plaque assay, ApoH nanoparticle fabrication, and virus extraction methods; comparison between native ApoH protein and synthetic ApoH peptide; characterization of SARS-CoV-2 primers; optimization of sample collection, preservation, preparation condition, and virus capture protocol; and ApoH assay efficiency on samples with low and high viral loads (PDF) Supplementary Material ac2c03813_si_001.pdf Author Contributions J.K., H.J., and T.H.K. contributed equally to this work. J.K., H.J., T.H.K., F.V., W.B.P., and S.K. conceived and designed the study. J.K., H.J., T.H.K, U.C., and F.V. optimized the protocol for virus capture and viral RNA extraction. U.C., H.B., and J.J. designed the primers for the RT-qPCR assay. W.L., H.J., and M.S. carried out the cell culture for virus cultivation and plaque assay. W.L., H.J., and G.Y.L. performed the live and dead assay. J.K. and J.N. conducted the statistical analysis. D.H.S, C.K.K, P.G.C, and N.J.K. collected and prepared the patient samples. W.B.P. and M.O. provided advice on the clinical data for validation. J.K., H.J., T.H.K., F.V., W.B.P., and S.K. analyzed the data and co-wrote the manuscript. All authors discussed the results and commented on the manuscript. The authors declare the following competing financial interest(s): The work was done in collaboration between QuantaMatrix Inc. and ApoH-Technolgoies. F.V. is the co-founder and CSO of ApoH-Technologies. Acknowledgments This research was supported by the Ministry of Science and ICT (MSIT, Republic of Korea), the National Research Foundation of Korea (NRF-2020R1A3B3079653, NRF-2020R1A2C1007242), the Bio & Medical Technology Development Program of the NRF of Korea (MSIT, 2020M3H1A1073304), BK21 FOUR program of the Education and Research Program for Future ICT Pioneers (Seoul National University in 2021), Seoul National University Bio-MAX Institute (K-BIO KIURI Center, 2020M3H1A1073304), QuantaMatrix Inc., and the Seoul National University Hospital Research Fund (grant no. 0320200330). The pathogen resource (NCCP43326) for this study was provided by the National Culture Collection for Pathogens (Korea National Institute of Health). F.V. acknowledges the ApoH-Technologies’ team as well as the European Commission for the European Projects: Horizon 2020 EDCTP “PANDORA-ID-NET” (grant no. RIA2016E-1609) and Horizon Europe “EPIC-Crown-2” (grant no. 101046084). ==== Refs References Adam D. The pandemic’s true death toll: millions more than official counts. Nature 2022, 601 , 312–315. 10.1038/d41586-022-00104-8.35042997 Baedeker M. ; Ringel M. S. ; Schulze U. 2021 FDA approvals: value driven by COVID-19 vaccines. Nat. Rev. Drug Discov. 2022, 21 , 90 10.1038/d41573-022-00015-3.35039671 Harvey W. T. ; Carabelli A. M. ; Jackson B. ; Gupta R. K. ; Thomson E. C. ; Harrison E. M. ; Ludden C. ; Reeve R. ; Rambaut A. ; Peacock C. ; Robertson U. ; Peacock S. J. ; Robertson D. L. SARS-CoV-2 variants, spike mutations and immune escape. Nat. Rev. Microbiol. 2021, 19 , 409–424. 10.1038/s41579-021-00573-0.34075212 Hu B. ; Guo H. ; Zhou P. ; Shi Z.-L. Characteristics of SARS-CoV-2 and COVID-19. Nat. Rev. Microbiol. 2021, 19 , 141–154. 10.1038/s41579-020-00459-7.33024307 Mercer T. R. ; Salit M. Testing at scale during the COVID-19 pandemic. Nat. Rev. Genet. 2021, 22 , 415–426. 10.1038/s41576-021-00360-w.33948037 Kevadiya B. D. ; Machhi J. ; Herskovitz J. ; Oleynikov M. D. ; Blomberg W. R. ; Bajwa N. ; Soni D. ; Das S. ; Hasan M. ; Patel M. ; Senan A. M. ; Gorantla S. ; McMillan J. ; Edagwa B. ; Eisenberg R. ; Gurumurthy C. B. ; Reid S. P. M. ; Punyadeera C. ; Chang L. ; Gendelman H. E. Diagnostics for SARS-CoV-2 infections. Nat. Mater. 2021, 20 , 593–605. 10.1038/s41563-020-00906-z.33589798 Amaral C. ; Antunes W. ; Moe E. ; Duarte A. G. ; Lima L. M. P. ; Santos C. ; Gomes I. L. ; Afonso G. S. ; Vieira R. ; Teles H. S. S. ; Reis M. S. ; da Silva M. A. R. ; Henriques A. M. ; Fevereiro M. ; Ventura M. R. ; Serrano M. ; Pimentel C. A molecular test based on RT-LAMP for rapid, sensitive and inexpensive colorimetric detection of SARS-CoV-2 in clinical samples. Sci. Rep. 2021, 11 , 16430 10.1038/s41598-021-95799-6.34385527 Ludwig K. U. ; Schmithausen R. M. ; Li D. ; Jacobs M. L. ; Hollstein R. ; Blumenstock K. ; Liebing J. ; Słabicki M. ; Ben-Shmuel A. ; Israeli O. ; Weiss S. ; Ebert T. S. ; Paran N. ; Rüdiger W. ; Wilbring G. ; Feldman D. ; Lippke B. ; Ishorst N. ; Hochfeld L. M. ; Beins E. C. ; Kaltheuner I. H. ; Schmitz M. ; Wöhler A. ; Döhla M. ; Sib E. ; Jentzsch M. ; Moench J. D. ; Borrajo J. ; Strecker J. ; Reinhardt B. ; Cleary M. ; Geyer M. ; Hölzel R. ; Macrae M. M. ; Nöthen P. ; Hoffmann M. ; Exner A. ; Regev F. ; Zhang J. L. ; Schmid-Burgk J. L. LAMP-Seq enables sensitive, multiplexed COVID-19 diagnostics using molecular barcoding. Nat. Biotechnol. 2021, 39 , 1556–1562. 10.1038/s41587-021-00966-9.34188222 Smyrlaki I. ; Ekman M. ; Lentini A. ; Rufino de Sousa N. R. ; Papanicolaou N. ; Vondracek M. ; Aarum J. ; Safari H. ; Muradrasoli S. ; Rothfuchs A. G. ; Albert J. ; Högberg B. ; Reinius B. Massive and rapid COVID-19 testing is feasible by extraction-free SARS-CoV-2 RT-PCR. Nat. Commun. 2020, 11 , 4812 10.1038/s41467-020-18611-5.32968075 Wyllie A. L. ; Fournier J. ; Casanovas-Massana A. ; Campbell M. ; Tokuyama M. ; Vijayakumar P. ; Warren J. L. ; Geng B. ; Muenker M. C. ; Moore A. J. ; Vogels C. B. F. ; Petrone M. E. ; Ott I. M. ; Lu P. ; Venkataraman A. ; Lu-Culligan A. ; Klein J. ; Earnest R. ; Simonov M. ; Datta R. ; Handoko R. ; Naushad N. ; Sewanan L. R. ; Valdez J. ; White E. B. ; Lapidus S. ; Kalinich C. C. ; Jiang X. ; Kim D. J. ; Kudo E. ; Linehan M. ; Mao T. ; Moriyama M. ; Oh J. E. ; Park A. ; Silva J. ; Song E. ; Takahashi T. ; Taura M. ; Weizman O.-E. ; Wong P. ; Yang Y. ; Bermejo S. ; Odio C. D. ; Omer S. B. ; Cruz C. S. D. ; Farhadian S. ; Martinello R. A. ; Iwasaki A. ; Grubaugh N. D. ; Ko A. I. Saliva or nasopharyngeal swab specimens for detection of SARS-CoV-2. N. Engl. J. Med. 2020, 383 , 1283 10.1056/nejmc2016359.32857487 Bloom J. S. ; Sathe L. ; Munugala C. ; Jones E. M. ; Gasperini M. ; Lubock N. B. ; Yarza F. ; Thompson E. M. ; Kovary K. M. ; Park J. ; Marquette D. ; Kay S. ; Lucas M. ; Love T. ; Sina Booeshaghi A. S. ; Brandenberg O. F. ; Guo L. ; Boocock J. ; Hochman M. ; Simpkins S. W. ; Lin I. ; LaPierre N. ; Hong D. ; Zhang Y. ; Oland G. ; Choe B. J. ; Chandrasekaran S. ; Hilt E. E. ; Butte M. J. ; Damoiseaux R. ; Kravit C. ; Cooper A. R. ; Yin Y. ; Pachter L. ; Garner O. B. ; Flint J. ; Eskin E. ; Luo C. ; Kosuri S. ; Kruglyak L. ; Arboleda V. A. Massively scaled-up testing for SARS-CoV-2 RNA via next-generation sequencing of pooled and barcoded nasal and saliva samples. Nat. Biomed. Eng. 2021, 5 , 657–665. 10.1038/s41551-021-00754-5.34211145 Marty F. M. ; Chen K. ; Verrill K. A. How to obtain a nasopharyngeal swab specimen. N. Engl. J. Med. 2020, 382 , e76 10.1056/nejmvcm2010260.32302471 Goldfarb D. M. ; Tilley P. ; Al-Rawahi G. N. ; Srigley J. A. ; Ford G. ; Pedersen H. ; Pabbi A. ; Hannam-Clark S. ; Charles M. ; Dittrick M. ; Gadkar V. J. ; Pernica J. M. ; Hoang L. M. N. Self-collected saline gargle samples as an alternative to health care worker-collected nasopharyngeal swabs for COVID-19 diagnosis in outpatients. J. Clin. Microbiol. 2021, 59 , e02427-20 10.1128/JCM.02427-20.33514627 Alemany A. ; Millat-Martinez P. ; Ouchi D. ; Corbacho-Monné M. ; Bordoy A. E. ; Esteban C. ; Hernández Á. ; Casañ C. ; Gonzalez V. ; Costes G. ; Capdevila-Jáuregui M. ; Torrano-Soler P. ; José A. S. ; Ara J. ; Prat N. ; Clotet B. ; Bassat Q. ; Gimenez M. ; Blanco I. ; Baro B. ; Mitjà O. Self-collected mid-nasal swabs and saliva specimens, compared with nasopharyngeal swabs, for SARS-CoV-2 detection in mild COVID-19 patients. J. Infect. 2021, 83 , 709 10.1016/j.jinf.2021.09.012. Wang W. ; Xu Y. ; Gao R. ; Lu R. ; Han K. ; Wu G. ; Tan W. Detection of SARS-CoV-2 in different types of clinical specimens. JAMA 2020, 323 , 1843–1844. 10.1001/jama.2020.3786.32159775 Lee R. A. ; Herigon J. C. ; Benedetti A. ; Pollock N. R. ; Denkinger C. M. Performance of saliva, oropharyngeal swabs, and nasal swabs for SARS-CoV-2 molecular detection: a systematic review and meta-analysis. J. Clin. Microbiol. 2021, 59 , e02881-e20 10.1128/jcm.02881-20.33504593 Kojima N. ; Turner F. ; Slepnev V. ; Bacelar A. ; Deming L. ; Kodeboyina S. ; Klausner J. D. Self-Collected Oral Fluid and Nasal Swab Specimens Demonstrate Comparable Sensitivity to Clinician-collected nasopharyngeal swab specimens for the detection of SARS-CoV-2. Clin. Infect. Dis. 2021, 73 , e3106 10.1093/cid/ciaa1589.33075138 Gertler M. ; Krause E. ; van Loon W. ; Krug N. ; Kausch F. ; Rohardt C. ; Rössig H. ; Michel J. ; Nitsche A. ; Mall M. A. ; Nikolai O. ; Hommes F. ; Burock S. ; Lindner A. K. ; Mockenhaupt F. P. ; Pison U. ; Seybold J. Self-collected oral, nasal and saliva samples yield sensitivity comparable to professionally collected oro-nasopharyngeal swabs in SARS-CoV-2 diagnosis among symptomatic outpatients. Int. J. Infect. Dis. 2021, 110 , 261–266. 10.1016/j.ijid.2021.07.047.34302962 Wölfel R. ; Corman V. M. ; Guggemos W. ; Seilmaier M. ; Zange S. ; Müller M. A. ; Niemeyer D. ; Jones T. C. ; Vollmar P. ; Rothe C. ; Hoelscher M. ; Bleicker T. ; Brünink S. ; Schneider J. ; Ehmann R. ; Zwirglmaier K. ; Drosten C. ; Wendtner C. Virological assessment of hospitalized patients with COVID-2019. Nature 2020, 581 , 465–469. 10.1038/s41586-020-2196-x.32235945 Savela E. S. ; Winnett A. V. ; Romano A. E. ; Porter M. K. ; Shelby N. ; Akana R. ; Ji J. ; Cooper M. M. ; Schlenker N. W. ; Reyes J. A. ; Carter A. M. ; Barlow J. T. ; Tognazzini C. ; Feaster M. ; Goh Y.-Y. ; Ismagilov R. F. Quantitative SARS-CoV-2 viral-load curves in paired saliva samples and nasal swabs inform appropriate respiratory sampling site and analytical test sensitivity required for earliest viral detection. J. Clin. Microbiol. 2022, 60 , e0178521 10.1128/JCM.01785-21.34911366 Zou L. ; Ruan F. ; Huang M. ; Liang L. ; Huang H. ; Hong Z. ; Yu J. ; Kang M. ; Song Y. ; Xia J. ; Guo Q. ; Song T. ; He J. ; Yen H.-L. ; Peiris M. ; Wu J. SARS-CoV-2 viral load in upper respiratory specimens of infected patients. N. Engl. J. Med. 2020, 382 , 1177–1179. 10.1056/nejmc2001737.32074444 Areo O. ; Joshi P. U. ; Obrenovich M. ; Tayahi M. ; Heldt C. L. Single-particle characterization of SARS-CoV-2 isoelectric point and comparison to variants of interest. Microorganisms 2021, 9 , 1606 10.3390/microorganisms9081606.34442686 Stohr J. J. J. M. ; Wennekes M. ; van der Ent M. ; Diederen B. M. W. ; Kluytmans-van den Bergh M. F. Q. K. ; Bergmans A. M. C. ; Kluytmans J. A. J. W. ; Pas S. D. Clinical performance and sample freeze-thaw stability of the cobas 6800 SARS-CoV-2 assay for the detection of SARS-CoV-2 in oro-/nasopharyngeal swabs and lower respiratory specimens. J. Clin. Virol. 2020, 133 , 104686 10.1016/j.jcv.2020.104686.33221622 Ninivaggi M. ; Kelchtermans H. ; Lindhout T. ; de Laat B. de. Conformation of beta2glycoprotein I and its effect on coagulation. Thromb. Res. 2012, 130 , S33–S36. 10.1016/j.thromres.2012.08.269.22925530 Wurm H. β2-glycoprotein-I (apolipoprotein H) interactions with phospholipid vesicles. Int. J. Biochem. 1984, 16 , 511–515. 10.1016/0020-711x(84)90168-x.6427035 Nakaya Y. ; Schaefer E. J. ; Brewer H. B. Activation of human post heparin lipoprotein lipase by apolipoprotein H (β2-glycoprotein I). Biochem. Biophys. Res. Commun. 1980, 95 , 1168–1172. 10.1016/0006-291x(80)91595-8.7417307 Nimpf J. ; Bevers E. M. ; Bomans P. H. H. ; Till U. ; Wurm H. ; Kostner G. M. ; Zwaal R. F. A. Prothrombinase activity of human platelets is inhibited by β2-glycoprotein-I. Biochim. Biophys. Acta, Gen. Subj. 1986, 884 , 142–149. 10.1016/0304-4165(86)90237-0. Niessen H. W. ; Lagrand W. K. ; Rensink H. J. ; Meijer C. J. ; Aarden L. ; Hack C. E. ; VisserApolipoprotein C. H. Apolipoprotein H, a new mediator in the inflammatory changes ensuing in jeopardised human myocardium. J. Clin. Pathol. 2000, 53 , 863 10.1136/jcp.53.11.863.11127271 Vutukuru M. R. ; Sharma D. K. ; Ragavendar M. ; Schmolke S. ; Huang Y. ; Gumbrecht W. ; Mitra N. A rapid, highly sensitive and culture-free detection of pathogens from blood by positive enrichment. J. Microbiol. Methods 2016, 131 , 105–109. 10.1016/j.mimet.2016.10.008.27765617 Stefas I. ; Tigrett S. ; Dubois G. ; Kaiser M. ; Lucarz E. ; Gobby D. ; Bray D. ; Ellerbrok H. ; Zarski J. P. ; Veas F. Interactions between Hepatitis C virus and the human apolipoprotein H acute phase protein: a tool for a sensitive detection of the virus. PLoS One 2015, 10 , e0140900 10.1371/journal.pone.0140900.26502286 Ganguli A. ; Mostafa A. ; Berger J. ; Aydin M. Y. ; Sun F. ; Ramirez S. A. S. ; Valera E. ; Cunningham B. T. ; King W. P. ; Bashir R. Rapid isothermal amplification and portable detection system for SARS-CoV-2. Proc. Natl. Acad. Sci. U.S.A. 2020, 117 , 22727–22735. 10.1073/pnas.2014739117.32868442 Sender R. ; Bar-On Y. M. ; Gleizer S. ; Bernshtein B. ; Flamholz A. ; Phillips R. ; Milo R. The total number and mass of SARS-CoV-2 virions. Proc. Natl. Acad. Sci. U.S.A. 2021, 118 , e2024815118 10.1073/pnas.2024815118.34083352 Kerimov D. ; Tamminen P. ; Viskari H. ; Lehtimäki L. ; Aittoniemi J. Sampling site for SARS-CoV-2 RT-PCR—An intrapatient four-site comparison from Tampere, Finland. PLoS One 2021, 16 , e0260184 10.1371/journal.pone.0260184.34784386 Watson J. ; Whiting P. F. ; Brush J. E. Interpreting a COVID-19 test result. BMJ 2020, 369 , m1808 10.1136/bmj.m1808.32398230 Alcoba-Florez J. ; Gil-Campesino H. ; Artola D. G.-M. ; González-Montelongo R. ; Valenzuela-Fernández A. ; Ciuffreda L. ; Flores C. Sensitivity of different RT-qPCR solutions for SARS-CoV-2 detection. Int. J. Infect. Dis. 2020, 99 , 190–192. 10.1016/j.ijid.2020.07.058.32745627 Jarvis K. F. ; Kelley J. B. Temporal dynamics of viral load and false negative rate influence the levels of testing necessary to combat COVID-19 spread. Sci. Rep. 2021, 11 , 9221 10.1038/s41598-021-88498-9.33911131 He X. ; Lau E. H. Y. ; Wu P. ; Deng X. ; Wang J. ; Hao X. ; Lau Y. C. ; Wong J. Y. ; Guan Y. ; Tan X. ; Mo X. ; Chen Y. ; Liao B. ; Chen W. ; Hu F. ; Zhang Q. ; Zhong M. ; Wu Y. ; Zhao L. ; Zhang F. ; Cowling B. J. ; Li F. ; Leung G. M. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nat. Med. 2020, 26 , 672–675. 10.1038/s41591-020-0869-5.32296168 Nicolelis M. A. L. ; Raimundo R. L. G. ; Peixoto P. S. ; Andreazzi C. S. The impact of super-spreader cities, highways, and intensive care availability in the early stages of the COVID-19 epidemic in Brazil. Sci. Rep. 2021, 11 , 13001 10.1038/s41598-021-92263-3.34155241 Sahin O. ; Yildirmak T. ; Karacalar S. ; Aydın E. ; Ciftci M. A. ; Bagci H. ; Yildirim S. ; Emeklioglu C. ; Balci B. G. ; Genc S. ; Cingillioglu B. ; Mihmanli V. ; Khalil A. ; Kalafat E. Short-term outcomes of pregnant women with convalescent COVID-19 and factors associated with false-negative polymerase chain reaction test: A prospective cohort study. Int. J. Clin. Pract. 2021, 75 , e14670 10.1111/ijcp.14670.34342119 Xiao A. T. ; Tong Y. X. ; Zhang S. False-negative of RT-PCR and prolonged nucleic acid conversion in COVID-19: Rather than recurrence. J. Med. Virol. 2020, 92 , 1755 10.1002/jmv.25855.32270882 Cleary B. ; Hay J. A. ; Blumenstiel B. ; Harden M. ; Cipicchio M. ; Bezney J. ; Simonton B. ; Hong D. ; Senghore M. ; Sesay A. K. ; Gabriel S. ; Regev A. ; Mina M. J. Using viral load and epidemic dynamics to optimize pooled testing in resource-constrained settings. Sci. Transl. Med. 2021, 13 , eabf1568 10.1126/scitranslmed.abf1568.33619080	36399654	PMC9718094	NO-CC CODE	2022-12-05 23:15:02	no	Anal Chem. 2022 Nov 18;:acs.analchem.2c03813	utf-8	Anal Chem	2,022	10.1021/acs.analchem.2c03813	oa_other
==== Front Mol Pharm Mol Pharm mp mpohbp Molecular Pharmaceutics 1543-8384 1543-8392 American Chemical Society 36448927 10.1021/acs.molpharmaceut.2c00448 Communication Telmisartan Nanosuspension for Inhaled Therapy of COVID-19 Lung Disease and Other Respiratory Infections Chen Daiqin †‡ Yun Xin § Lee Daiheon †‡ DiCostanzo Joseph R. ∥ Donini Oreola ⊥ Shikuma Cecilia M. # Thompson Karen ∇ Lehrer Axel T. ◆ Shimoda Larissa § https://orcid.org/0000-0002-2437-0001 Suk Jung Soo †‡¶ † Center for Nanomedicine, Wilmer Eye Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland21231, United States ‡ Department of Ophthalmology, Wilmer Eye Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland21231, United States § Division of Pulmonary and Critical Care Medicine, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland21205, United States ∥ Kaiser Permanente, Honolulu, Hawai‘i96814, United States ⊥ Soligenix, Inc., Princeton, New Jersey08540, United States # Department of Medicine, John A. Burns School of Medicine, University of Hawai‘i at Ma̅noa, Honolulu, Hawai‘i96813, United States ∇ Department of Pathology, John A. Burns School of Medicine, University of Hawai‘i at Ma̅noa, Honolulu, Hawai‘i96813, United States ◆ Department of Tropical Medicine, Medical Microbiology & Pharmacology, John A. Burns School of Medicine, University of Hawai‘i at Ma̅noa, Honolulu, Hawai‘i96813, United States ¶ Department of Chemical & Biomolecular Engineering, Johns Hopkins University, Baltimore, Maryland21218, United States Email: [email protected]. Phone: +1-410-614-4526. 30 11 2022 acs.molpharmaceut.2c0044803 06 2022 17 11 2022 16 11 2022 © 2022 American Chemical Society 2022 American Chemical Society This article is made available via the PMC Open Access Subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Vaccine hesitancy and the occurrence of elusive variants necessitate further treatment options for coronavirus disease 2019 (COVID-19). Accumulated evidence indicates that clinically used hypertensive drugs, angiotensin receptor blockers (ARBs), may benefit patients by mitigating disease severity and/or viral propagation. However, current clinical formulations administered orally pose systemic safety concerns and likely require a very high dose to achieve the desired therapeutic window in the lung. To address these limitations, we have developed a nanosuspension formulation of an ARB, entirely based on clinically approved materials, for inhaled treatment of COVID-19. We confirmed in vitro that our formulation exhibits physiological stability, inherent drug activity, and inhibitory effect against SARV-CoV-2 replication. Our formulation also demonstrates excellent lung pharmacokinetics and acceptable tolerability in rodents and/or nonhuman primates following direct administration into the lung. Thus, we are currently pursuing clinical development of our formulation for its uses in patients with COVID-19 or other respiratory infections. renin-angiotensin system nanosuspension acute respiratory distress syndrome (ARDS) respiratory infection inhalational therapy U.S. Department of Health and Human Services 10.13039/100000050 P30EY001765 Cystic Fibrosis Foundation 10.13039/100000897 SUK18I0 U.S. Department of Health and Human Services 10.13039/100000050 R01HL137716 U.S. Department of Health and Human Services 10.13039/100000050 R01HL073859 document-id-old-9mp2c00448 document-id-new-14mp2c00448 ccc-price This article is made available via the ACS COVID-19 subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. ==== Body pmcIntroduction The outbreak of the coronavirus disease 2019 (COVID-19) pandemic has spurred global efforts to contain the spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent for the disease. Although vaccination is the pre-eminent public health strategy, vaccine hesitancy and the occurrence of increasingly more elusive variants have reinforced the need for effective treatment strategies, especially those that prevent the need for hospitalization. Treatment targeted specifically to prevent the acute respiratory distress syndrome (ARDS) caused by COVID-19 offers both short- and long-term utility in the event of escape variants of SARS-CoV-2 and/or other respiratory pandemics. Accumulated evidence suggests that clinically used hypertensive drugs, angiotensin receptor blockers (ARBs), may mitigate deleterious lung pathology, ARDS, in patients with COVID-19 presumably via modulation of the renin-angiotensin system (RAS) perturbed by the disease.1−5 Specifically, ARBs can shift the pro-inflammatory angiotensin II (ANG II)-dominant pathological state of ARDS toward an anti-inflammatory angiotensin converting enzyme 2 (ACE2)-dominant state.6 Further, potential inhibitory roles of ARBs on replication of SARS-CoV-2 have been recently suggested and experimentally validated in vitro.7−9 Thus, ARBs could exert meaningful therapeutic intervention in COVID-19 lungs in a multimodal manner. However, universal use of oral ARB formulations poses safety concerns due to the established systemic adverse effects, particularly for those with normal blood pressure or hypotension.10−13 Additionally, unlike the original use for systemic pressure-reducing effects, a very high oral dose is likely needed to achieve desirable therapeutic concentrations in the lung tissue to yield meaningful clinical outcomes. Inhalable formulations may elegantly enhance dose flexibility as well as reduce overall dose and/or dosing frequency, therefore minimizing undesired systemic exposure of drug payloads.14,45 To this end, we have developed and extensively characterized an inhalable nanosuspension formulation of a clinically used ARB, telmisartan, which can be administered directly into the lung via nebulization. Of note, nebulized nanoformulations, such as drug nanosuspensions, can effectively reach the deep lung when appropriate aerodynamic properties are achieved.15,16 Materials and Methods Materials Telmisartan was purchased from Biosynth International, Inc. (San Diego, CA) and stored at 4 °C before use. Free drug telmisartan solution (FD-TEL, Semintra) was purchased from Boehringer Ingelheim Vetmedica, Inc. (St. Joseph, MO). Polysorbate 80 (Tween 80) was purchased from VWR Life Science (Rouses Point, NY). Water and methanol for HPLC analysis, dithiothreitol, fura-2-acetoxymethyl, and Krebs solution were purchased from Fisher Scientific (Hampton, NH). UltraPure DNase/Rnase-Free distilled water, HEPES-buffered saline (HBS), collagenase, papain, and bovine serum albumin (BSA) were purchased from Invitrogen (Carlsbad, CA). Minimum Essential Medium (MEM, 1×), fetal bovine serum (FBS), and penicillin/streptomycin (P/S) were purchased from Thermo Fisher Scientific – Gibco (Carlsbad, CA), and 0.25% Trypsin was purchased from Corning (Corning, NY). ANG II was purchased from Sigma-Aldrich (St. Louis, MO). Smooth muscle cell medium was purchased from ScienCell (Carlsbad, CA), and SmGM-2 SingleQuots were purchased from Lonza (Basel, Switzerland). Formulation and Physicochemical Characterization of Inhalable Polysorbate 80-Coated Telmisartan Nanosuspensions (INH-TEL) 12 mg of polysorbate 80 was dissolved in 3 mL of ultrapure water and then 1, 3, 4, or 5 mg of telmisartan was added. The resulting suspension was sonicated with a Branson 2800 ultrasonic bath (Branson Ultrasonic Corp., Danbury, CT) for 15 min at room temperature (R.T.), and a spoonful of stainless-steel beads (0.5 mm, Next Advance, Troy, NY) was added before loading on to a Tissue Lyser LT (Qiagen, Hilden, Germany) and milled for up to 10 h with the oscillation speed of 50 s–1. Some batches were subjected to the second pulverization step where the obtained suspension was homogenized with a probe sonicator (Vibra-Cell VCX500, Sonics & Materials, Newtown, CT) for 1 h (pulse: 1 s on and 1 s off, amplitude: 40%). Subsequently, the mixture was centrifuged at 900g for 10 min to remove large aggregates. This first centrifuge step also removes metal pieces released from the beads during the milling process, if any. The supernatant was centrifuged at 17,000g for 10 min to remove the polysorbate 80 micelles, and the pellet was resuspended in 1 mL of ultrapure water. Finally, the suspension was freeze-dried and stored as a white powder (i.e., INH-TEL) at R.T. for future uses. The hydrodynamic diameter/polydispersity index (PDI) and ζ-potential were measured by dynamic light scattering and laser Doppler anemometry, respectively, using a Nanosizer ZS90 (Malvern Instruments, Southborough, MA). Transmission electron microscopy (TEM, Hitachi H7600, Tokyo, Japan) was performed to determine the morphology of INH-TEL. The drug content in INH-TEL was determined with HPLC (LC-2030C 3D, Shimadzu, Columbia, MD) equipped with Luna Omega 5 μm Polar C18 column (Phenomenex, Torrance, CA) at a flow rate of 1 mL/min and an injection volume of 25 μL. The fluorescence detector of HPLC was set to the excitation and emission wavelengths at 300 and 360 nm, respectively, for the detection of telmisartan. To determine the relative amounts of telmisartan and polysorbate 80 in INH-TEL, the final product was lyophilized and measured for the total dry mass. The amount of polysorbate 80 was then calculated by subtracting the mass of telmisartan, determined by HPLC, from the mass of the lyophilized powder. Physiological Stability of INH-TEL in Bronchoalveolar Lavage Fluid (BALF) To assess the physiological stability of INH-TEL, BALF was collected from C57BL/6 mice. After euthanizing mice, a small hole was cut in the trachea, and an 18 G metal tube adaptor was inserted into the trachea through the hole. Three successive volumes of 1 mL of DPBS were instilled and gently aspirated to be pooled. The pooled solution was centrifuged at 1,500g for 10 min at 4 °C, and the supernatant was collected and lyophilized for future use. The lyophilized BALF and INH-TEL were rehydrated with DPBS to prepare a mixed solution containing 0.06 mg/mL BALF solid content and 0.6 mg/mL INH-TEL (based on the telmisartan concentration). The INH-TEL in BALF solution was incubated at R.T. and the hydrodynamic diameters and PDI values were measured over time using a Nanosizer ZS90. Drug Release Kinetics of INH-TEL The lyophilized INH-TEL were rehydrated in DPBS to a telmisartan concentration of 5 μg/mL and 1 mL of INH-TEL resuspension was loaded into a dialysis bag (MWCO = 3 kDa). The dialysis bag was then immersed in 20 mL of DPBS supplemented with 0.05% polysorbate 80 (to generate an artificial sink condition) and placed in a shaking incubator (Incu-Shaker, Newport Beach, CA) at 150 rpm and 37 °C. 1 mL of solution was taken out for HPLC analysis to determine the telmisartan content, and the same volume of fresh DPBS was replenished at each time point. In Vitro Drug Activity of INH-TEL Primary pulmonary arterial smooth muscle cells (PASMCs) were isolated from rats as previously described.17,18 Briefly, the heart and lung were excised and transferred to a Petri dish containing HBS. The intrapulmonary arteries were isolated, and the tissue was digested for 20 min in reduced Ca2+ HBS containing collagenase (type I, 1750 U/mL), papain (9.5 U/mL), BSA (2 mg/mL), and dithiothreitol (1 mM) at 37 °C. Individual smooth muscle cells were isolated from the digested tissue via trituration and plated on glass coverslips. Cells were cultured in smooth muscle cell medium containing SmGM-2 SingleQuots supplemented with 1% P/S for 24–48 h and placed into low-serum media (smooth muscle cell medium containing 0.5% FBS and 1% P/S) 16–24 h before beginning experiments. Intracellular calcium concentration ([Ca2+]i) was measured as previously described.17,18 Briefly, PASMCs were loaded with 5 μM fura-2-acetoxymethyl for 1 h at 37 °C before being placed in a temperature-controlled (37 °C) laminar flow chamber in a live cell Ca2+ imaging system (Intracellular Imaging Inc., Cincinnati, OH). Cells were perfused with warmed modified Krebs solution bubbled with 16% O2 & 5% CO2 gas at 37 °C. At the beginning of each experiment, cells were perfused for 15 min to allow for establishment of a stable baseline. [Ca2+]i was calculated from the F340/F380 ratio using a calibration curve created with known Ca2+ concentrations. Cells were pretreated with DPBS only, FD-TEL, or INH-TEL at a final telmisartan concentration of 50 μM for 30 min prior to exposure to ANG II (10–7 M). In Vitro Antiviral Efficacy of INH-TEL against SARS-CoV-2 The assessment of in vitro antiviral efficacy was conducted by the University of Pennsylvania High-throughput Screening Core.19,20 Calu-3 cells (ATCC, HTB-55) grown in MEM supplemented with 1% nonessential amino acids, 1% P/S, 2 mM l-glutamine, and 10% FBS were plated on a 384-well assay plate (Corning, Corning, NY). On the next day, 1 μL of INH-TEL was resuspended in an aqueous buffer and added as an 8-point dose–response assay with 3-fold dilutions between test concentrations in triplicate, starting at a final concentration of 33.3 μg/mL. In parallel, cells were treated with FD-TEL at a concentration range for comparison, and the negative controls (0.2% DMSO, n = 32) were included on each assay plate. Calu3 cells were pretreated with INH-TEL for 2 h prior to infection. In BSL3 containment, SARS-CoV-2 (isolate USA WA1/2020) diluted in serum-free growth medium was added to plates to achieve a multiplicity of infection (MOI) = 0.5. Cells were incubated continuously with INH-TEL and SARS-CoV2 for 48 h. Cells were fixed with 4% formaldehyde for 15 min at R.T., washed 3 times with PBS, blocked with 2% BSA (w/v) in PBS supplemented with 0.1% Triton-X-100 (PBST) and incubated with primary antibody against dsRNA (J2) diluted in PBST overnight at 4 °C. Cells were washed 3 times with PBST and incubated with a secondary antibody (anti-mouse Alexa Fluor 488) and Hoechst 33342 for 1 h at R.T. Cells were washed 3 times with PBST and imaged on an automated microscope (ImageXpress Micro, Molecular Devices, San Jose, CA) at 10×, four sites per well. The total number of cells and the number of infected (dsRNA+) cells were measured using the cell scoring module (MetaXpress 5.3.3, Molecular Devices), and the percentage of infected cells (dsRNA+ cells/cell number) per well was calculated. SARS-CoV-2 infection at each drug concentration was normalized to aggregated DMSO plate control wells and expressed as percentage-of-control (POC; % infection sample/avg % infection DMSO control). A nonlinear regression curve fit analysis with Prism 8 (GraphPad Software, La Jolla, CA) of POC infection and cell viability versus the log10 transformed concentration values were used to determine the IC50/IC90 values for infection and CC50 values for cell viability. Selectivity index (SI) was calculated as a ratio of drug’s CC50 and IC50 values (SI = CC50/IC50). Pharmacokinetics Study with Mice Lyophilized INH-TEL was reconstituted in ultrapure water to a telmisartan concentration of 0.05 mg/mL and intratracheally administrated into the lungs of C57BL/6 mice at a telmisartan dose of 0.1 mg/kg. Mice were euthanized at predetermined time points postadministration. Plasma and lungs were harvested, and drug content was analyzed by HPLC as described earlier. Pharmacokinetics and Tolerability Studies with Nonhuman Primates (NHPs) Dosing of NHPs and subsequent sample collection were conducted at BIOQUAL, Inc. (Rockville, MD). The lyophilized INH-TEL was reconstituted in 0.45% saline to a telmisartan concentration of 0.833 mg/mL and intratracheally administrated into the lungs of two cynomolgus macaques (CM1 and CM2) at a dosage of 2.5 mg/animal using a laryngo-tracheal mucosal atomization device (MADgic, Teleflex, Morrisville, NC). In parallel, the other macaque (CM3) received daily oral administration of FD-TEL solution at a telmisartan dose of 1 mg/kg. The blood samples and lung tissues were collected at predetermined time points listed in Table S1. Drug contents in the plasma and lungs were measured by HPLC as described earlier. The blood samples were subjected to complete blood count and blood chemistry assays. The livers, kidneys, and lungs were formalin-fixed, paraffin embedded, and sectioned for hematoxylin and eosin (H&E) staining for histopathologic analysis by a board-certified pathologist (Auther K. T.). Statistical Analysis Statistical analyses were performed with Student’s t-test and one-way analysis of variance (ANOVA), followed by the Tukey’s multiple comparisons tests, for two-group and multiple (i.e., more than two)-group comparisons, respectively, using Prism (v.8.4.2, GraphPad Software). Result and Discussion Physicochemical Properties of INH-TEL We used a modified top-down method21,22 to formulate an inhalable telmisartan formulation (INH-TEL) composed of TEL drug nanosuspension core stably coated by polysorbate 80, one of the rare polymer-based surfactants approved by the FDA for respiratory use.23 We conducted physicochemical characterization of INH-TEL freshly prepared by an optimized two-step pulverization method (i.e., 10 h of milling, followed by 1 h of probe sonication) where the hydrodynamic diameters and ζ-potentials were measured to be 322 ± 15 nm (polydispersity index or PDI = 0.24 ± 0.03) and −2.9 ± 0.5 mV, respectively (Figure 1A and Table 1). The hydrodynamic diameters of INH-TEL fall in the size range of 100–400 nm reported for polysorbate 80 stabilized nanosuspension formulations prepared with other water-insoluble drugs.24−27 We found that omitting or shortening either the milling or the probe sonication step resulted in marked increases in the hydrodynamic diameters and the PDI values and/or dramatic reduction in the final yield (Table S2). On the other hand, physicochemical properties and yield were comparable when nanosuspensions were prepared by the optimized pulverization method while varying the telmisartan-to-polysorbate 80 ratios at a fixed polysorbate 80 concentration of 4 mg/mL (Table S3). The ratio between telmisartan and polysorbate 80 in INT-TEL was determined to be roughly 4:1 (Table 1). Figure 1 Physicochemical properties and in vitro antivirus activity of INH-TEL. (A) Hydrodynamic diameters of freshly prepared (black) and lyophilized-rehydrated (red) INH-TEL measured by DLS. (B) Representative transmission electron micrographs of freshly prepared (left) and lyophilized-rehydrated (right) INH-TEL. Scale bar = 100 nm. (C) Colloidal stability of INH-TEL in a physiological lung environment determined by the changes of particle hydrodynamic diameters in mouse BALF at 37 °C over time (n = 3 independent experiments). (D) Cumulative in vitro release of the drug payloads (i.e., telmisartan) from INH-TEL in DPBS supplemented with 0.05% polysorbate 80 over time (n = 3 independent experiments). (E) In vitro drug activity (i.e., inhibition of intracellular calcium spike induced by ANG II) of INH-TEL in comparison to FD-TEL. (F) In vitro inhibitory effect on SARS-CoV-2 replication (red) and cytotoxicity (black) of INH-TEL in Calu-3 cells (n = 3 independent experiments). , p < 0.05 (one-way ANOVA). Table 1 Physicochemical Properties of Freshly Prepared and Lyophilized–Rehydrated INH-TEL INH-TEL hydrodynamic diameter (nm) PDI ζ-potential (mV) drug loading density (%) fresh 322 ± 15 0.24 ± 0.03 –2.9 ± 0.5 82 ± 3 lyophilized-rehydrated 351 ± 9 0.25 ± 0.04 –4.0 ± 1.0 We then found that lyoprotectant-free lyophilization and subsequent reconstitution (i.e., rehydration) did not yield particle aggregates and resulted in only moderate changes in hydrodynamic diameters (351 ± 0.9 nm; PDI = 0.25 ± 0.04) and ζ-potentials (−4.0 ± 1.0 mV) (Figure 1A and Table 1). Likewise, TEM revealed that both freshly prepared and lyophilized-rehydrated INH-TEL possessed a rod-shaped morphology with similar geometric sizes (Figure 1B). The findings here underscore that INH-TEL could be stored long-term in a powder form prior to reconstitution in an aqueous vehicle solution for inhaled administration. INH-TEL also demonstrated excellent colloidal stability in a physiologically relevant lung environment, mouse BALF, for at least up to 2 h, as evidenced by negligible changes in hydrodynamic diameters and PDI (Figure 1C). In parallel, we conducted an in vitro drug release study using Dulbecco’s phosphate-buffered saline (DPBS) supplemented with 0.05% polysorbate 80 as an artificial sink condition where nearly 90% of telmisartan was released within the first 5 h (Figure 1D). The rapid drug release may be beneficial for managing acute pathological conditions that require prompt drug action, such as ARDS triggered by respiratory pathogens. In Vitro Pharmacological Activities of INH-TEL We next tested whether the drug release from INH-TEL preserved its inherent drug activity by assessing the ability to prevent ANG II-mediated elevation of intracellular calcium ion concentration ([Ca2+]i) in lung smooth muscle cells. ANG II binding to its cell surface receptor (i.e., ANG II type 1 receptor) activates the voltage-gated Ca2+ channels to elevate [Ca2+]i, which is effectively inhibited by ARBs.17,18,29 We found that the ANG II-mediated transient [Ca2+]i spike was equally and entirely abrogated when cells were treated with dose-matched free drug telmisartan (FD-TEL) or INH-TEL (Figure 1E), suggesting that the drug activity of INH-TEL was fully retained. To test our hypothesis that INH-TEL might provide antiviral efficacy, we then assessed the ability of INH-TEL to deter SARS-CoV-2 replication in vitro.19,20 We found that the viral replication was inhibited by INH-TEL in a dose-dependent manner (up to the telmisartan concentration of 33.3 μg/mL) in Calu-3 cells without incurring significant cytotoxicity (Figure 1F). Of note, Calu-3 has been confirmed for expression of the cell surface portal for SARS-CoV-2 (i.e., ACE2) and susceptibility to the viral infection accordingly.31 Our observation agrees with recent reports demonstrating inhibitory effects of various ARBs, including telmisartan, against SARS-CoV-2 replication in Vero-E6 or Caco-2 cells.7−9 In relevance to these findings, it has been shown that intracellular calcium is essential for viral assembly and budding of SARS-CoV,32 the virus responsible for outbreak of severe acute respiratory syndrome in 2003, and that ARB reduces viral spread by preventing release of several enveloped viruses from infected cells.33−35 More recently, the potential role of ARBs in blocking the main protease of SARS-CoV-2 essential for viral replication and transcription has been suggested,8 but further investigation is warranted to fully unravel the mechanism(s) of inhibition. Although ARBs were initially speculated to upregulate ACE2 to promote viral infection and disease severity, recent independent studies refuted such a hypothesis.36 We note that while FD-TEL exhibited modest dose-dependent antiviral effect (Figure S1), telmisartan was markedly more effective when combined with the polysorbate 80 as a nanosuspension (i.e., INH-TEL; Figure 1F). As the antiviral effect of the two ingredients of INH-TEL appears to be synergistic, direct pulmonary administration is likely the most practical dosing method to dispatch both components simultaneously to the apical lung surface. Importantly, the highest viral load in SARS-CoV-2, as well as numerous other respiratory viral infections, occurs in the apical lung. The exact mechanism of the enhanced in vitro antiviral effect enabled by the addition of polysorbate 80 is yet to be unraveled. However, we speculate that it may be linked to an antiviral effect of its hydrolysis product, oleic acid, which has been previously shown to inactivate several enveloped virus.37−40 Pharmacokinetics of INH-TEL Following Intratracheal Administration into the Lungs of Mice and Cynomolgus Macaques We went on to test our hypothesis that direct administration of INH-TEL into the lung would provide high telmisartan concentrations in the lung. To test this, we intratracheally administered INH-TEL into the lungs of C57BL/6 mice at a telmisartan dose of 0.1 mg/kg and compared the drug content in the lung and the plasma at different time points. Of note, INH-TEL was reconstituted to include <0.02% w/w polysorbate 80 to meet the FDA cutoff for respiratory use.23 We found that telmisartan content was about an order of magnitude greater in the lung compared to that in the plasma at 1 and 12 h postadministration (Figure 2A). To complement this mouse study, we recently conducted a small pharmacokinetic study using NHPs (i.e., cynomolgus macaques) in which we compared locally administered INH-TEL versus oral FD-TEL. Specifically, two macaques were intratracheally treated with INH-TEL at a telmisartan dose of 2.5 mg per animal (0.81–0.87 mg/kg), and lung tissues were harvested at 0.5 or 8-h postadministration for the assessment of drug content in the lung. As a clinically relevant control, one macaque received daily oral FD-TEL for 7 days at a telmisartan dose of 1 mg/kg and lung tissue was harvested 2 h after the final dose. The 2 h time point was selected based on a previous human study demonstrating the median time to maximum plasma concentration of systemically administered telmisartan to be 0.5–2 h.41 Macaques that received intratracheal INH-TEL, regardless of the time of lung harvest, exhibited an at least 10-fold and up to 40-fold greater drug content in the lung compared to the macaque that received 7 daily oral FD-TEL (Figure 2B). In parallel, we monitored plasma pharmacokinetics of these animals. The plasma drug content of the animals that received intratracheal INH-TEL was transiently elevated but quickly reduced to the level on par with or lower than the steady-state plasma drug content observed with the animal under the daily oral FD-TEL regimen (Figure 2C). In agreement with our mouse study (Figure 2A), telmisartan content was markedly and significantly greater in the lung compared to the plasma at both 0.5 and 8 h postadministration of the intratracheal INH-TEL (Figure 2D). In contrast, the drug content was significantly greater in the plasma than in the lung 2 h after the final (i.e., seventh) oral FD-TEL administration (Figure 2D). Figure 2 Pharmacokinetics of INH-TEL following intratracheal administration into the lungs of wild-type C57BL/6 mice and cynomolgus macaques. (A) Lung and plasma concentrations of telmisartan 1 and 12 h after a single intratracheal administration of INH-TEL at a telmisartan dose of 0.1 mg/kg into the lungs of wild-type C57BL/6 mice (n = 5 mice per group). (B–D) Two macaques, CM1 (3.08 kg) and CM2 (2.86 kg), received a single intratracheal (IT) administration of INH-TEL at a fixed telmisartan dose of 2.5 mg (0.81–0.87 mg/kg, calculated based on the body weights) and one macaque (CM3; 3.08 kg) received daily oral gavage (OG) administration of FD-TEL at a telmisartan dose of 1 mg/kg for 7 days. (B) Telmisartan content in the lung tissues from CM1 and CM2 harvested at 0.5 and 8 h postadministration of INH-TEL into the lung, respectively, and from CM3 harvested 2 h after the last (i.e., seventh) daily oral administration of FD-TEL. (C) Plasma pharmacokinetics of telmisartan in CM1 and CM2 after receiving a single intratracheal administration of INH-TEL and CM3 after receiving the sixth daily oral administration of FD-TEL. Plasma pharmacokinetics were monitored until CM1 and CM2 were euthanized to harvest lung tissues and up to 12 h for CM3. (D) Relative telmisartan content in the lung tissue versus the plasma harvested from CM1 and CM2 at 0.5- and 8-h postadministration of INH-TEL into the lung and from CM3 at 2 h postadministration of the last (i.e., seventh) oral FD-TEL dose. , p < 0.05; *, p < 0.005; and **, p < 0.001 (one-way ANOVA). We note that the NHP study was conducted as a comparison in extremes where we sought to evaluate the tolerability of a very high dose of locally, lung-administered telmisartan in NHPs. Employing a 10-fold safety factor42 recommended by the FDA, we chose the intratracheal telmisartan dose at 10 times the efficacious dose predicted from our mouse study (0.1 mg/kg). Based on the reference body and lung weights of different species,43 the intratracheal telmisartan dose of 2.5 mg corresponds to a human equivalent dose of 125 mg. On the other hand, the oral dose was given at a lower level in NHPs in order to allow us to obtain a well-tolerated steady-state dose. Specifically, the 1 mg/kg oral NHP dose converts to the human equivalent of 19.4 mg for a reference human body weight of 60 kg,42 which is 25–50% of the standard oral daily dose of 40–80 mg given to patients with hypertension. Clearly, using a ∼ 6.5-fold greater local dose resulted in up to a 40-fold greater drug amount delivered to the lung (Figure 2B). Assuming lung concentrations scale linearly with oral dose, human equivalent oral doses of ∼780 mg/day would be required to achieve the same level of lung exposure in these NHPs. Histopathological Analysis of Lungs of Macaques That Received Intratracheal INH-TEL We next assessed tolerability of intratracheally administered INH-TEL in NHPs. Specifically, lung tissues harvested from three macaques at the respective times of pulmonary drug content analysis (Figure 2C) were subjected to paraffin section and hematoxylin and eosin staining. Sections were taken from the peripheral and central cranial, peripheral and central caudal, and peripheral and central midportion of both right and left lungs of each of the three animals. The lung slides were then scored in a blinded manner for edema, composite inflammation, increased bronchus-associated lymphoid tissue (BALT), reactive epithelial changes, alveolar collapse, and interstitial fibrosis by a board-certified pathologist (author K.T.). The evaluation revealed no significant histopathologic differences between INH-TEL intratracheally administered at an extremely highly local dose and oral FD-TEL given at a well-tolerated dose with acceptable tolerability in the lung tissues (Figure 3). We note that mild alveolar collapse observed shortly after the intratracheal administration of INH-TEL was quickly resolved (Figure 3). We also conducted blood biochemistry analysis at the times of lung harvest and respective baselines (i.e., prior to the administration). We found that most of the biochemical readouts were comparable before and after the treatments (Table 2), underscoring that our formulation did not exert significant systemic toxicity. The high drug level of lung exposure achieved in NHPs resulted in circulating plasma levels that are commensurate with well-tolerated systemic telmisartan levels after oral dosing, ensuring systemic safety; of note, human dosing with 120 mg tablets has been reported to yield a Cmax of 1.635 μg/mL.44 Notably, the human experience with telmisartan also indicates that increased dosage is not associated with greater blood pressure reductions, but rather with longer duration of effect.44 Thus, the transient systemic exposure of telmisartan observed after the intratracheal administration of INH-TEL (Figure 2C) is not expected to yield adverse events. Further, the lung exposure shown to be well tolerated in NHPs is well beyond any level expected to be needed in humans. Figure 3 Histopathological analysis of lung tissues from macaques after receiving either intratracheal INH-TEL or oral FD-TEL. Lung tissues were harvested at different time points after a single intratracheal administration of INH-TEL at a fixed telmisartan dose of 2.5 mg (0.81–0.87 mg/kg) or after the seventh oral gavage administration of FD-TEL at a telmisartan dose of 1 mg/kg. Table 2 Blood Biochemistry Panel of Macaques That Received Either Intratracheal INH-TEL or Oral FD-TEL macaque ID CM1 CM2 CM3 administration IT IT OG treatment INH-TEL (2.5 mg) INH-TEL (2.5 mg) FD-TEL (1 mg/kg) time B.L.a 0.5 hc B.L.a 8 hc B.L.a dose 6 B.L.b dose 7 B.L.b 2 hc ALP (U/L) 137 128 135 147 164 150 158 155 ALT (U/L) 64 61 15 51 86 131 177 176 AST (U/L) 44 47 23 235 85 86 163 222 albumin (g/dL) 3.8 3.5 3.7 3.8 3.5 3.4 3.6 3.5 total protein (g/dL) 6.8 6.3 6.7 7 6.9 6.2 6.4 6.2 globulin (g/dL) 3 2.8 3 3.2 3.4 2.8 2.8 2.7 total bilirubin (mg/dL) 0.2 0.2 0.1 0.3 0.1 0.2 0.2 0.2 bilirubin-conjugated (mg/dL) 0 0 0 0.1 0 0.1 0.1 0 BUN (mg/dL) 31 27 19 20 26 22 26 27 creatinine (mg/dL) 0.7 0.6 0.6 0.5 0.7 0.6 0.8 0.8 cholesterol (mg/dL) 133 127 125 137 125 106 103 104 glucose (mg/dL) 79 100 87 100 97 77 85 59 calcium (mg/dL) 9 8.8 9.1 8.7 9.6 8.9 9.7 9.3 phosphorus (mg/dL) 3.9 3.7 3.7 5.6 5 3.1 3 3.2 chloride(mmol/L) 104 106 109 113 105 105 103 108 potassium(mmol/L) 3.3 3.8 3.5 3.7 3 3.8 4.1 3.4 sodium(mmol/L) 146 146 148 149 146 147 145 148 ALB/GLOB ratio 1.3 1.3 1.2 1.2 1 1.2 1.3 1.3 BUN/Creatinine ratio 44.3 45 31.7 40 37.1 36.7 32.5 33.8 bilirubin-unconjugated (mg/dL) 0.2 0.2 0.1 0.2 0.1 0.1 0.1 0.2 NA/K ratio 44 38 42 40 49 39 35 44 hemolysis index N + N +++ N + N N lipemia index N N N N N N N N a Baseline levels prior to the intratracheal administration of INH-TEL or the first oral administration of FD-TEL. b Baseline levels prior to sixth or seventh daily oral dose of FD-TEL. c Levels at different times postadministration of intratracheal INH-TEL or of the last (i.e., seventh) oral administration of FD-TEL. Conclusion In summary, we have developed a surface-stabilized nanosuspension formulation of telmisartan suitable for long-term storage and shipping in a powder form and experimentally confirmed its physiological stability, unperturbed drug activity, and inhibitory potential against SARS-CoV-2 infection. Further, our formulation demonstrates excellent lung pharmacokinetics and acceptable local and systemic tolerability as revealed by our NHP studies. To this end, we are currently continuing our effort toward the clinical development of our formulation to be nebulized for treating patients with COVID-19 or ARDS associated with other respiratory infections. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.molpharmaceut.2c00448.Schedule of dosing and sample collection for cynomolgus macaques (PDF) Supplementary Material mp2c00448_si_001.pdf Author Contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. The authors declare no competing financial interest. Acknowledgments This work was supported by the National Institute of Health (R01HL137716, R01HL073859, and P30EY001765) and the Cystic Fibrosis Foundation (SUK18I0). The NHP study was supported by a personal donation of J.R.D. Johns Hopkins University utilized the nonclinical and preclinical services program offered by the National Institute of Allergy and Infectious Diseases. The authors thank Drs. Sara Cherry and David C. Schultz and the University of Pennsylvania High-throughput Screening Core for supporting the in vitro anti-SARS-CoV-2 studies in Calu-3 cells. We would like to thank BIOQUAL, Inc. for conduct of the NHP study. Specifically, thanks to Deborah Weiss for veterinary care; John Misamore, Holly Thomasson, and Nick Solomotis for NHP facility management and coordination; Jake Yalley-Ogunro for sample processing and shipments, and Hanne Andersen, the BIOQUAL, Inc. PI. ==== Refs References Zeinalian M. ; Salari-Jazi A. ; Jannesari A. ; Khanahmad H. A potential protective role of losartan against coronavirus-induced lung damage. Infect Control Hosp Epidemiol 2020, 41 (6 ), 752–753. 10.1017/ice.2020.80.32248854 Rothlin R. P. ; Vetulli H. M. ; Duarte M. ; Pelorosso F. G. Telmisartan as tentative angiotensin receptor blocker therapeutic for COVID-19. Drug Dev Res. 2020, 81 (7 ), 768–770. 10.1002/ddr.21679.32356926 Duarte M. ; Pelorosso F. ; Nicolosi L. N. ; Salgado M. V. ; Vetulli H. ; Aquieri A. ; Azzato F. ; Castro M. ; Coyle J. ; Davolos I. Telmisartan for treatment of Covid-19 patients: An open multicenter randomized clinical trial. eClinicalMedicine 2021, 37 , 100962 10.1016/j.eclinm.2021.100962.34189447 Rothlin R. P. ; Duarte M. ; Pelorosso F. G. ; Nicolosi L. ; Salgado M. V. ; Vetulli H. M. ; Spitzer E. Angiotensin Receptor Blockers for COVID-19: Pathophysiological and Pharmacological Considerations About Ongoing and Future Prospective Clinical Trials. Front. Pharmacol. 2021, 12 , 603736 10.3389/fphar.2021.603736.33854432 Bengtson C. D. ; Montgomery R. N. ; Nazir U. ; Satterwhite L. ; Kim M. D. ; Bahr N. C. ; Castro M. ; Baumlin N. ; Salathe M. An Open Label Trial to Assess Safety of Losartan for Treating Worsening Respiratory Illness in COVID-19. Front Med. 2021, 8 , 630209 10.3389/fmed.2021.630209. Zaheer J. ; Kim H. ; Kim J. S. Correlation of ACE2 with RAS components after Losartan treatment in light of COVID-19. Sci. Rep 2021, 11 (1 ), 24397 10.1038/s41598-021-03921-5.34937861 Nejat R. ; Sadr A. S. ; Freitas B. ; Crabttree J. ; Pegan S. D. ; Tripp R. A. ; Najafi D. Losartan Inhibits SARS-CoV-2 Replication in Vitro. J. Pharm. Pharm. Sci. 2021, 24 , 390–399. 10.18433/jpps31931.34319871 Alnajjar R. ; Mostafa A. ; Kandeil A. ; Al-Karmalawy A. A. Molecular docking, molecular dynamics, and in vitro studies reveal the potential of angiotensin II receptor blockers to inhibit the COVID-19 main protease. Heliyon 2020, 6 (12 ), e05641 10.1016/j.heliyon.2020.e05641.33294721 Reus P. ; Schneider A. K. ; Ulshofer T. ; Henke M. ; Bojkova D. ; Cinatl J. ; Ciesek S. ; Geisslinger G. ; Laux V. ; Grattinger M. ; Gribbon P. ; Schiffmann S. Characterization of ACE Inhibitors and AT1R Antagonists with Regard to Their Effect on ACE2 Expression and Infection with SARS-CoV-2 Using a Caco-2 Cell Model. Life 2021, 11 (8 ), 810–822. 10.3390/life11080810.34440554 Pitt B. Adverse events in HEAAL: when to hold and when to fold. European journal of heart failure 2012, 14 (12 ), 1319–1321. 10.1093/eurjhf/hfs167.23159580 Mann J. F. ; Schmieder R. E. ; Dyal L. ; McQueen M. J. ; Schumacher H. ; Pogue J. ; Wang X. ; Probstfield J. L. ; Avezum A. ; Cardona-Munoz E. ; Dagenais G. R. ; Diaz R. ; Fodor G. ; Maillon J. M. ; Ryden L. ; Yu C. M. ; Teo K. K. ; Yusuf S. Effect of telmisartan on renal outcomes: a randomized trial. Ann. Int. Med. 2009, 151 (1 ), 1–10. 10.7326/0003-4819-151-1-200907070-00122.19451556 Kiernan M. S. ; Wentworth D. ; Francis G. ; Martinez F. A. ; Dickstein K. ; Komajda M. ; Zannad F. ; Neaton J. D. ; Konstam M. A. Predicting adverse events during angiotensin receptor blocker treatment in heart failure: results from the HEAAL trial. European journal of heart failure 2012, 14 (12 ), 1401–1409. 10.1093/eurjhf/hfs145.23187648 Yusuf S. ; Healey J. S. ; Pogue J. ; Chrolavicius S. ; Flather M. ; Hart R. G. ; Hohnloser S. H. ; Joyner C. D. ; Pfeffer M. A. ; Connolly S. J. Irbesartan in patients with atrial fibrillation. N Engl. J. Med. 2011, 364 (10 ), 928–938. 10.1056/NEJMoa1008816.21388310 Chen D. ; Liu J. ; Wu J. ; Suk J. S. Enhancing nanoparticle penetration through airway mucus to improve drug delivery efficacy in the lung. Expert Opin Drug Deliv 2021, 18 (5 ), 595–606. 10.1080/17425247.2021.1854222.33218265 Andrade da Silva L. H. ; Vieira J. B. ; Cabral M. R. ; Antunes M. A. ; Lee D. ; Cruz F. F. ; Hanes J. ; Rocco P. R. M. ; Morales M. M. ; Suk J. S. Development of nintedanib nanosuspension for inhaled treatment of experimental silicosis. Bioeng. Transl. Med. 2022, e10401 10.1002/btm2.10401. Dabbagh A. ; Abu Kasim N. H. ; Yeong C. H. ; Wong T. W. ; Abdul Rahman N. Critical Parameters for Particle-Based Pulmonary Delivery of Chemotherapeutics. J. Aerosol Med. Pulm Drug Deliv 2018, 31 (3 ), 139–154. 10.1089/jamp.2017.1382.29022837 Mangal S. ; Gao W. ; Li T. ; Zhou Q. T. Pulmonary delivery of nanoparticle chemotherapy for the treatment of lung cancers: challenges and opportunities. Acta Pharmacol Sin 2017, 38 (6 ), 782–797. 10.1038/aps.2017.34.28504252 Shimoda L. A. ; Luke T. ; Sylvester J. T. ; Shih H.-W. ; Jain A. ; Swenson E. R. Inhibition of hypoxia-induced calcium responses in pulmonary arterial smooth muscle by acetazolamide is independent of carbonic anhydrase inhibition. Am. J. Physiol.: Lung Cell. Mol. Physiol. 2007, 292 (4 ), L1002–L1012. 10.1152/ajplung.00161.2006.17209136 Shimoda L. A. ; Suresh K. ; Undem C. ; Jiang H. ; Yun X. ; Sylvester J. T. ; Swenson E. R. Acetazolamide prevents hypoxia-induced reactive oxygen species generation and calcium release in pulmonary arterial smooth muscle. Pulm. Circ. 2021, 11 (4 ), 1–12. 10.1177/20458940211049948. Dittmar M. ; Lee J. S. ; Whig K. ; Segrist E. ; Li M. ; Kamalia B. ; Castellana L. ; Ayyanathan K. ; Cardenas-Diaz F. L. ; Morrisey E. E. ; Truitt R. ; Yang W. ; Jurado K. ; Samby K. ; Ramage H. ; Schultz D. C. ; Cherry S. Drug repurposing screens reveal cell-type-specific entry pathways and FDA-approved drugs active against SARS-Cov-2. Cell Rep 2021, 35 (1 ), 108959 10.1016/j.celrep.2021.108959.33811811 Schultz D. C. ; Johnson R. M. ; Ayyanathan K. ; Miller J. ; Whig K. ; Kamalia B. ; Dittmar M. ; Weston S. ; Hammond H. L. ; Dillen C. Pyrimidine biosynthesis inhibitors synergize with nucleoside analogs to block SARS-CoV-2 infection. bioRxiv (Microbiology), 2021.06.24.449811, June 24, 2021. DOI: 10.1101/2021.06.24.449811. Van Eerdenbrugh B. ; Van den Mooter G. ; Augustijns P. Top-down production of drug nanocrystals: nanosuspension stabilization, miniaturization and transformation into solid products. Int. J. Pharm. 2008, 364 (1 ), 64–75. 10.1016/j.ijpharm.2008.07.023.18721869 Fu X. ; Cai J. ; Zhang X. ; Li W. D. ; Ge H. ; Hu Y. Top-down fabrication of shape-controlled, monodisperse nanoparticles for biomedical applications. Adv. Drug Deliv Rev. 2018, 132 , 169–187. 10.1016/j.addr.2018.07.006.30009884 Metz J. K. ; Scharnowske L. ; Hans F. ; Schnur S. ; Knoth K. ; Zimmer H. ; Limberger M. ; Gross H. ; Lehr C. M. ; Hittinger M. Safety assessment of excipients (SAFE) for orally inhaled drug products. ALTEX 2020, 37 (2 ), 275–286. 10.14573/altex.1910231.32052853 de Cassia Zaghi Compri J. ; Andres Felli V. M. ; Lourenco F. R. ; Takatsuka T. ; Fotaki N. ; Lobenberg R. ; Bou-Chacra N. A. ; Barros de Araujo G. L. Highly Water-Soluble Orotic Acid Nanocrystals Produced by High-Energy Milling. J. Pharm. Sci. 2019, 108 (5 ), 1848–1856. 10.1016/j.xphs.2018.12.015.30599168 Corrias F. ; Schlich M. ; Sinico C. ; Pireddu R. ; Valenti D. ; Fadda A. M. ; Marceddu S. ; Lai F. Nile red nanosuspensions as investigative model to study the follicular targeting of drug nanocrystals. Int. J. Pharmaceut 2017, 524 (1–2 ), 1–8. 10.1016/j.ijpharm.2017.03.042. Starkloff W. J. ; Bucala V. ; Palma S. D. ; Gonzalez Vidal N. L. Design and in vitro characterization of ivermectin nanocrystals liquid formulation based on a top-down approach. Pharm. Dev. Technol. 2017, 22 (6 ), 809–817. 10.1080/10837450.2016.1200078.27346432 Zhang J. J. ; Lv H. X. ; Jiang K. ; Gao Y. Enhanced bioavailability after oral and pulmonary administration of baicalein nanocrystal. Int. J. Pharmaceut 2011, 420 (1 ), 180–188. 10.1016/j.ijpharm.2011.08.023. Abadir P. M. ; Walston J. D. ; Carey R. M. ; Siragy H. M. Angiotensin II Type-2 receptors modulate inflammation through signal transducer and activator of transcription proteins 3 phosphorylation and TNFα production. J. Interferon Cytokine Res. 2011, 31 (6 ), 471–474. 10.1089/jir.2010.0043.21288138 Saccon E. ; Chen X. ; Mikaeloff F. ; Rodriguez J. E. ; Szekely L. ; Vinhas B. S. ; Krishnan S. ; Byrareddy S. N. ; Frisan T. ; Vegvari A. Cell-type-resolved quantitative proteomics map of interferon response against SARS-CoV-2. iScience 2021, 24 (5 ), 102420 10.1016/j.isci.2021.102420.33898942 Lu W. ; Xu K. ; Sun B. , SARS Accessory Proteins ORF3a and 9b and Their Functional Analysis. In Molecular Biology of the SARS-Coronavirus; Springer Berlin Heidelberg, 2009; pp 167. Choi J. A. ; Kim J. E. ; Ju H. H. ; Lee J. ; Jee D. ; Park C. K. ; Paik S. Y. The effects of losartan on cytomegalovirus infection in human trabecular meshwork cells. PLoS One 2019, 14 (6 ), e0221693 10.1371/journal.pone.0221693.31437233 Gardner P. L. ; Mbuy G. N. ; Knabb M. T. Effects of the angiotensin II receptor antagonist losartan on herpes simplex virus-type 2 infection of cultured Vero and cardiac neonatal myocytes. Life Sci. 1994, 55 (4 ), 283–289. 10.1016/0024-3205(94)00730-6.8028446 Bai D. ; Fang L. ; Xia S. ; Ke W. ; Wang J. ; Wu X. ; Fang P. ; Xiao S. Porcine deltacoronavirus (PDCoV) modulates calcium influx to favor viral replication. Virology 2020, 539 , 38–48. 10.1016/j.virol.2019.10.011.31670218 Jackson C. B. ; Farzan M. ; Chen B. ; Choe H. Mechanisms of SARS-CoV-2 entry into cells. Nat. Rev. Mol. Cell Biol. 2022, 23 (1 ), 3–20. 10.1038/s41580-021-00418-x.34611326 Kohn A. ; Gitelman J. ; Inbar M. Unsaturated Free Fatty-Acids Inactivate Animal Enveloped Viruses. Arch. Virol. 1980, 66 (4 ), 301–307. 10.1007/BF01320626.7447706 Chen D. Y. ; Luo W. ; Hoffman J. ; Huang L. H. ; Sandefur S. ; Hall T. ; Murphy M. ; O’Donnell S. Insights into virus inactivation by polysorbate 80 in the absence of solvent. Biotechnol. Prog. 2020, 36 (3 ), 2953 10.1002/btpr.2953. Bayard F. ; Calande T. ; Pritchard D. I. ; Thielemans W. ; Young S. ; Paine S. ; Bosquillon C. Esterase activity in the Human Bronchial Epithelial Calu-3 Cell Line. J. Pharm. Pharmacol. 2010, 62 (10 ), 1316–1317. Brebner J. ; Kalow W. Soluble Esterases of Human Lung. Can. J. Biochem. 1970, 48 (9 ), 970–978. 10.1139/o70-151.5475473 Stangier J. ; Su C. A. ; Roth W. Pharmacokinetics of orally and intravenously administered telmisartan in healthy young and elderly volunteers and in hypertensive patients. J. Int. Med. Res. 2000, 28 (4 ), 149–67. 10.1177/147323000002800401.11014323 Nair A. B. ; Jacob S. A simple practice guide for dose conversion between animals and human. J. Basic Clin Pharm. 2016, 7 (2 ), 27–31. 10.4103/0976-0105.177703.27057123 Tepper J. S. ; Kuehl P. J. ; Cracknell S. ; Nikula K. J. ; Pei L. Q. ; Blanchard J. D. Symposium Summary: ″Breathe In, Breathe Out, Its Easy: What You Need to Know About Developing Inhaled Drugs″. Int. J. Toxicol 2016, 35 (4 ), 376–392. 10.1177/1091581815624080.26857693 Sharpe M. ; Jarvis B. ; Goa K. L. Telmisartan: a review of its use in hypertension. Drugs 2001, 61 (10 ), 1501–29. 10.2165/00003495-200161100-00009.11558835	36448927	PMC9718101	NO-CC CODE	2022-12-05 23:15:03	no	Mol Pharm. 2022 Nov 30;:acs.molpharmaceut.2c00448	utf-8	Mol Pharm	2,022	10.1021/acs.molpharmaceut.2c00448	oa_other
==== Front ACS Appl Mater Interfaces ACS Appl Mater Interfaces am aamick ACS Applied Materials & Interfaces 1944-8244 1944-8252 American Chemical Society 36448483 10.1021/acsami.2c16446 Research Article Highly Adsorptive Au-TiO2 Nanocomposites for the SERS Face Mask Allow the Machine-Learning-Based Quantitative Assay of SARS-CoV-2 in Artificial Breath Aerosols Hwang Charles S. H. †∥⊥ Lee Sangyeon †⊥ Lee Sejin †∥ Kim Hanjin † https://orcid.org/0000-0002-5387-6458 Kang Taejoon ‡§ Lee Doheon † https://orcid.org/0000-0003-4799-7816 Jeong Ki-Hun †∥ † Department of Bio and Brain Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon34141, Korea ‡ Bionanotechnology Research Center, Korea Research Institute of Bioscience and Biotechnology (KRIBB), 125 Gwahak-ro, Yuseong-gu, Daejeon34141, Korea § School of Pharmacy, Sungkyunkwan University, Suwon16419, Korea ∥ KAIST Institute for Health Science and Technology (KIHST), Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon34141, Korea * Email: [email protected]. * Email: [email protected]. 30 11 2022 acsami.2c1644613 09 2022 21 11 2022 © 2022 The Authors. Published by American Chemical Society 2022 The Authors This article is made available via the PMC Open Access Subset for unrestricted RESEARCH re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Human respiratory aerosols contain diverse potential biomarkers for early disease diagnosis. Here, we report the direct and label-free detection of SARS-CoV-2 in respiratory aerosols using a highly adsorptive Au-TiO2 nanocomposite SERS face mask and an ablation-assisted autoencoder. The Au-TiO2 SERS face mask continuously preconcentrates and efficiently captures the oronasal aerosols, which substantially enhances the SERS signal intensities by 47% compared to simple Au nanoislands. The ultrasensitive Au-TiO2 nanocomposites also demonstrate the successful detection of SARS-CoV-2 spike proteins in artificial respiratory aerosols at a 100 pM concentration level. The deep learning-based autoencoder, followed by the partial ablation of nondiscriminant SERS features of spike proteins, allows a quantitative assay of the 101–104 pfu/mL SARS-CoV-2 lysates (comparable to 19–29 PCR cyclic threshold from COVID-19 patients) in aerosols with an accuracy of over 98%. The Au-TiO2 SERS face mask provides a platform for breath biopsy for the detection of various biomarkers in respiratory aerosols. SARS-CoV-2 surface-enhanced Raman spectroscopy breath biopsy machine-learning plasmonics nanocomposite Korea Medical Device Development Fund 10.13039/100019266 KMDF_PR_20200901_0074 Ministry of Science and ICT, South Korea 10.13039/501100014188 NRF-2012M3A9C4048758 National Research Foundation of Korea 10.13039/501100003725 2021R1A2B5B03002428 National Research Foundation of Korea 10.13039/501100003725 2021M3E5E3080379 Korea Research Institute of Bioscience and Biotechnology 10.13039/501100003715 1711134081 Electronics and Telecommunications Research Institute 10.13039/501100003696 21YR2500 document-id-old-9am2c16446 document-id-new-14am2c16446 ccc-price ==== Body pmcIntroduction Respiratory airborne particulates from human breath contain vital information on the current health status of an individual.1,2 Different forms of such particulates, i.e., aerosols and volatile organic compounds (VOCs), contain potential biomarkers for a wide spectrum of diseases including viruses, asthma, cancers, and neurodegenerative disorders.3−8 In particular, respiratory aerosols from human breath have attracted significant interest with the recent global outbreak of the 2019 coronavirus disease (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2).9−11 With the first report in the late December of 2019 in Wuhan, China, there are over 500 million confirmed cases globally as of April 2022.12 Both the World Health Organization and the United States Centers for Disease Control and Prevention have confirmed that COVID-19 is principally transmitted through respiratory aerosols.12,13 Combined with the high proportion of asymptomatic patients reaching up to 40% of the total confirmed patients, the reproduction number of COVID-19 has been reported to be 4–20 times greater than that of seasonal flu.14 However, the current gold standard for the diagnosis of such large-scale COVID-19 patients is still limited to nucleic acid detection by amplification,15,16 where the examination process takes up to several hours under the supervision of highly qualified medical personnel. COVID-19 face masks have recently been demonstrated that diagnose the collected viruses on a cellulose matrix;17,18 yet, these also require time-consuming post-extraction processes for a polymerase chain reaction. In contrast, a comprehensive and direct examination of breath aerosols has a high potential for the rapid diagnosis of COVID-19 but has been relatively unexplored for such diagnostic applications. Recently, surface-enhanced Raman spectroscopy (SERS) has demonstrated the detection and analysis of aerosols for diverse air monitoring applications by utilizing the electromagnetic “hotspots” of plasmonic nanostructures.4,19−22 Unlike conventional aerosol detection methods such as physical impactor and evaporative light scattering detection, SERS serves as a cost-effective alternative for rapid, sensitive, and quantitative analysis of chemical fingerprints within the aerosols. However, plasmonic SERS substrates often show some technical limitations in detecting respiratory breath aerosols due to a low Weber number of an aerosol, i.e., the ratio of the aerosol’s momentum to the surface tension.23 In other words, the fast emission velocity of such aerosols (2–10 m/s)24 substantially deters effective adsorption onto a substrate. While conventional SERS substrates with assorted nanostructures25−29 and nanoparticles30,31 display highly packed geometries for enhanced electromagnetic hotspots, they often result in low surface energy that is inefficient for aerosol adsorption. As a result, hotspot-rich SERS substrates with high surface energies are still in need of efficient capture and facile preconcentration of respiratory aerosols. Results and Discussion Here, we report a SERS face mask for the label-free detection of the aerosolized SARS-COV-2 virus using gold-titanium dioxide (Au-TiO2) nanocomposites. The face mask features a highly adsorptive Au-TiO2 nanocomposite SERS chip on the inner cloth of a filtering facepiece respirator (Figure 1a). Human breath aerosols are oronasally released during respiratory action such as breathing, coughing, sneezing, or speaking and are continuously preconcentrated onto the SERS chip inside the respirator. Au-TiO2 nanoislands on a quartz substrate exhibit high aerosol adsorption and dense electromagnetic hotspots, allowing the rapid, facile, and quantitative SERS detection of the SARS-CoV-2 virus (Figure 1b). An autoencoder neural network is further employed for the accurate classification of the SARS-CoV-2 virus at various concentrations. The aerodynamic behavior of aerosols impacting a surface mainly depends on the surface energy (Figure 1c). For instance, respiratory aerosols with an average velocity of 2–10 m/s24 simply rebound upon impact with conventional Au nanoislands due to low surface energy.23 In contrast, an ultrathin film of TiO2 nanoclustered in the Volmer–Weber mode on the Au nanoislands significantly increases the surface energy (contact angle, θ = 20–30°), allowing the efficient adsorption of the respiratory aerosols. The Au-TiO2 SERS chip with an active sensing area of 28 mm2 was diced and attached to a commercial KF94 respirator with a round medical plaster bandage (Figure 1d). Figure 1 Schematic illustration of SARS-CoV-2 detection from respiratory breath aerosols using the Au-TiO2 SERS chip on a face mask. (a) Oronasal aerosols emitted upon respiratory actions are easily and directly collected onto the SERS chip. (b) Machine-learned SERS detection of SARS-CoV-2 aerosols. The measured SERS signals are quantitatively classified using an ablation-assisted autoencoder. (c) Aerodynamic behavior of aerosols impacting on a solid substrate with different surface energies. Au-TiO2 nanoislands with high surface energies efficiently adsorb the low-volume and high-velocity respiratory aerosols. (d) Optical image of the SERS face mask as an application example. The Au-TiO2 SERS chip was attached to the inner cloth of a KF94 mask with a medical plaster. The enhanced surface energy of the Au-TiO2 nanocomposites is crucial for the highly sensitive SERS detection of respiratory aerosols. The nanofabrication of Au-TiO2 nanocomposites exploits a facile two-step process including repeated thermal dewetting of Au thin films and thermal evaporation of TiO2 (Figure 2a). The hotspot-rich Au nanoislands were fabricated at repeated dewetting of 6 + 6 nm thick Au thin films, showing a 21% increase in average size, 72% decrease in the interparticle distance, and 5-fold E-field enhancement (Figure. S1), compared to single dewetting of a 12 nm thick Au thin film. See the Methods section for the repeated thermal dewetting of Au nanoislands, geometric characterization of Au nanoislands, and FDTD calculation of E-field enhancements. A 2 nm thick TiO2 film was evaporated in the Volmer–Weber mode onto the as-fabricated Au nanoislands to form Au-TiO2 nanocomposites. The geometry and the composition of Au-TiO2 nanocomposites were characterized using a field-emission transmission electron microscope (300 keV FE-TEM, Tecnai G2 F30 S-TWIN, FEI) equipped with energy-dispersive X-ray (EDX) (Figures 2b and S2). The ultrathin TiO2 film forms nanoclusters on Au nanoislands, while TiO2 of larger thickness leads to uniform thin-film formation.32,33 The contact angle of a water droplet was then measured to compare the surface energy of Au-TiO2 substrates with different TiO2 thicknesses ranging from 0, 2, 4, 6, 8, and 10 nm (Figure 2c). While simple Au nanoislands without TiO2 show a contact angle of 60°, even a slight addition of 2 nm thick TiO2 significantly reduces the contact angle to 29°. The measured SERS signals of nebulized 10–7 M rhodamine 6 g (R6G) also show that the Au-TiO2 substrate efficiently captures aerosols from the increased surface energy. The E-field intensities of the Au-TiO2 nanocomposites were calculated using the FDTD method and compared with the observed SERS peak intensity of nebulized R6G aerosols at 1360 cm–1 (Figures 2d and S3). See the Methods section for the FDTD calculation of E-field enhancements. The relative E-field intensity ratio (E4/EAu4) rapidly decays to 0.67, 0.49, 0.43, 0.39, and 0.36 for increasing TiO2 thickness of 2, 4, 6, 8, and 10 nm because the TiO2 thin films behave as a dielectric gap spacer.34 However, the SERS intensity ratios at 1360 cm–1 are observed as 1.47, 0.81, 0.74, 0.58, and 0.51 with respect to that of simple Au. This phenomenon is explained by the highly adsorptive Au-TiO2 substrate efficiently capturing more aerosols that compensates for the SERS intensity decay. In particular, the 2 nm TiO2 thin film evaporated as nanoclusters on Au allows direct contact of the target analytes in the aerosol to the plasmonic hotspots, exhibiting substantially enhanced SERS intensity. The preconcentration of aerosols on the face mask was further investigated by measuring the SERS signals of 10–7 M R6G for increasing nebulizing times (Figure 2e). The low-volume aerosols dry quickly, allowing for the continuous accumulation of target analytes within the aerosols. The SERS intensity at 1360 cm–1 linearly increases as the preconcentration time of the aerosols changes from 5, 10, 15, and 20 s. Note that the collection of aerosols nebulized during five seconds is equivalent to a four-hour preconcentration of aerosols from respiration.35 Figure 2 Highly adsorptive Au-TiO2 nanocomposites substrate for aerosol preconcentration and enhanced SERS signals. (a) Nanofabrication process employs repeated solid-state dewetting of Au thin films for high-density EM hotspot generation. A thin layer of TiO2 is then thermally evaporated to obtain Au-TiO2 nanocomposites. (b) FE-TEM images of the fabricated Au-TiO2 nanoislands for 2 nm (left) and 10 nm (right) thin films of TiO2. (c) Contact angles for various thicknesses of TiO2 on Au nanoislands. The 2 nm thick TiO2 thin film substantially increases the surface energy. (d) Comparison of measured SERS intensity of R6G aerosols and calculated E-field intensities for different thicknesses of TiO2. The Au-TiO2 substrate with 2 nm TiO2 exhibits 47% increased SERS intensity of R6G aerosols at 1360 cm–1 due to highly adsorptive surfaces efficiently adsorbing nebulized R6G aerosols. (e) Preconcentration of 100 nM R6G aerosols using a nebulizer. The SERS intensity of R6G at 1360 cm–1 linearly increases with respect to the preconcentration times. Characteristic SERS peaks of SARS-CoV-2 spike proteins in artificial respiratory aerosols were characterized using the Au-TiO2 substrates. The SARS-CoV-2 is surrounded by protruding spike proteins, which serve as a primary biomarker for immunoassays in the receptor recognition and membrane fusion process.36 Several characteristic SERS peaks in 1 μM SARS-CoV-2 spike proteins are observed including CH2 rocking of phenylalanine at 651, 772, and 996 cm–1, CH2 rocking of tryptophan at 852 cm–1, NH3 rocking of histidine at 1177 cm–1, CH2 wagging of L-arginine at 1267 cm–1, C-N stretching and amide III band at 1372 cm–1, and CH2 deformation and NH bending of tryptophan and phenylalanine at 1523 and 1587 cm–1 (Figure 3a).37 Note that the SERS signals of SARS-CoV-2 spike proteins show high selectivity to other viral proteins.38,39 Also see Figure S4 for SERS signals of the spike proteins with different concentrations ranging from 1 μM to 100 pM. Next, the SARS-CoV-2 spike proteins with concentrations ranging from 100 nM to 100 pM were mixed in an artificial respiratory solution and nebulized onto the SERS substrate for the emulation of the human oronasal emission. The chemical constituents,40 concentrations, and observed SERS peaks of the artificial respiratory aerosols (ARA) are respectively summarized in Figures 3b and S5. The ARA samples were prepared by adding spike proteins of different concentrations to the stock artificial respiratory solution. See Methods for the emulation of oronasal aerosols and artificial respiratory solutions. Strong SERS bands are observed in ARA at the 1400–1600 cm–1 range, resulting from various chemicals such as potassium citrate, lactic acid urea, etc. As a result, the SERS signals of the spike protein in ARA (Figure 3c) are partially different from that of Figure 3a, particularly at the 1400–1600 cm–1 range, due to signal interference and different adsorption kinetics in complex mixtures caused by ARA.41 The characteristic SERS peaks of the SARS-CoV-2 spike proteins in Figure 3a were then quantitatively compared by calculating the SERS peak intensity ratios depending on the spike protein concentration, with respect to ARA as the reference (Figure 3d). While the signal intensity ratio of nondiscriminant peaks (1523 and 1587 cm–1) remains relatively constant, all other characteristic peaks are highly discriminant and increase for a higher concentration of the SARS-CoV-2 spike protein in ARA. Figure 3 SERS measurements of SARS-CoV-2 spike proteins in artificial respiratory aerosols. (a) SERS signals of SARS-CoV-2 spike protein aerosols adsorbed on Au-TiO2 nanocomposite substrates. The characteristic Raman peaks of the spike proteins are clearly observable at 651, 772, 853, 927, 996, 1177, 1267, 1372, 1523, and 1587 cm–1. (b) Composition of artificial respiratory aerosols (ARA) and measured SERS peaks. Raman bands of the spike proteins at 1523 and 1587 cm–1 overlap with those of artificial respiratory aerosols. (c) SERS signals at various concentrations of the spike proteins in ARA. (d) SERS signal intensity ratios at various characteristic Raman bands of the spike proteins were calculated using ARA as the reference. Some Raman bands, i.e., 1523 and 1587 cm–1 Raman bands, are shadowed by the reference signals, while most others provide quantitative fingerprints of SARS-CoV-2 spike proteins in ARA. An ablation-assisted autoencoder with a logistic regression model was further employed for a quantitative assay of SARS-Cov-2 lysate in aerosols. The autoencoder algorithm allows characteristic feature extraction by reducing the dimensions of input signals into low-dimensional features in the latent space.42,43 The autoencoder-based dimensionality reduction model was trained in a supervised manner to minimize the total loss function including latent loss, interclass, intraclass, and distances from centroids44 (Figure 4a). See the Methods section for training the autoencoder model. The initial SERS signals of SARS-CoV-2 lysate aerosols in artificial respiratory solutions were acquired for different concentrations ranging from 0, 101, 102, 103, and 104 pfu/mL at five different positions on the Au-TiO2 substrates (Figures 4a and S6). The SERS signals mapped onto the two-dimensional (2D) latent space via the autoencoder reveal that the individual classes are well-clustered and diagonally aligned depending on the lysate concentration (Figure 4b). This allows the logistic regression model to be used as a prediction model for the highly accurate quantitative assay of target molecules in the complex solution, i.e., artificial respiratory aerosols, that supersede the conventional chemometrics such as principal component analysis. The ablation of the nondiscriminant Raman features of spike proteins in ARA further substantially increases the classification accuracy of the prediction model (Figure 4c). The two nondiscriminant SERS features of the spike proteins in ARA, i.e., 1532 and 1587 cm–1 bands, were ablated based on Figure 3. Data sets with the random ablation of the same length of vectors were also generated for comparison. Each of the data sets with nondiscriminatory ablation, random ablation, and without ablation was then trained using the autoencoder-based dimensionality reduction model, followed by logistic regression models. The receiver operating characteristic (ROC) curve explicitly demonstrates that the ablation of the 1532 and 1587 cm–1 SERS peaks provides 7.6% higher classification accuracy compared to nonablated SERS signals. The confusion matrix of the ablation-assisted autoencoder model indicates that the SARS-CoV-2 lysates adsorbed on the Au-TiO2 nanocomposite SERS substrate exhibit over 98% classification accuracy for concentrations ranging from 101–104 pfu/mL (Figure 4d). Note that typical COVID-19 patients demonstrate an average cyclic threshold value of 27 during PCR testing,45 which corresponds to 102–103 pfu/mL concentration of the SARS-CoV-2 virus.46 As a result, the highly accurate assay of the SARS-CoV-2 lysates down to 101 pfu/mL infers the successful diagnosis of COVID-19 from respiratory aerosols. Figure 4 Deep learning-based quantitative SERS assay for label-free SARS-CoV-2 lysates in ARA. (a) Ablation-assisted autoencoding. The nasopharyngeal SARS-CoV-2 lysates were nebulized with ARA for initial SERS measurements, followed by the ablation of the low-importance SERS features of spike proteins in ARA. Then, an autoencoder was used to quantitatively classify the SERS signals of the viral lysates (101–104 pfu/mL) in ARA. The 2D latent space scores were optimized with hyperparameters including latent loss (L0), intraclass (L1), interclass (L2), and distances from centroids (L3). (b) Two-dimensional latent scores of SERS signals of SARS-CoV-2 lysates with different concentrations. The classified clusters show a strong diagonal alignment with SARS-CoV-2 concentrations. (c) Area under the receiver operating characteristic curve (ROC) for the ablation-assisted autoencoder algorithm. (d) Confusion matrix of the autoencoder algorithm exhibiting the average prediction accuracy of over 98% from the label-free SERS detection of SARS-CoV-2 lysates. Conclusions To conclude, this work has successfully demonstrated a highly adsorptive Au-TiO2 nanocomposite-based SERS face mask for the SERS detection of SARS-CoV-2 in artificial breath aerosols. Nanoclustered TiO2 on Au nanoislands efficiently adsorbs respiratory aerosols and shows 47% enhanced SERS signals compared to conventional Au nanoislands. The SERS chip inside a face mask further exhibits preconcentration of low-volume respiratory aerosols. The autoencoder prediction model with a modified loss function demonstrates the quantitative assay of SARS-CoV-2 lysates with 98% accuracy by ablating nondiscriminant SERS peaks of spike proteins. This Au-TiO2 SERS face mask provides a rapid, robust, and facile screening method for the pre-emptive diagnosis of COVID-19 and a biosensing platform for breath biopsy, detecting various disease biomarkers in respiratory aerosols. Methods Repeated Thermal Dewetting of Au Nanoislands First, a 6 nm thick Au thin film was thermally evaporated onto a 4-inch quartz wafer at a constant rate of 0.5 Å/s. The Au thin film was then thermally dewetted inside a box furnace (Lindberg/Blue M, Moldatherm Box Furnace) for one hour at 700 °C to form large-area Au nanoislands with high adhesion to glass. The target temperature was steadily increased from the room temperature with ramp-up and ramp-down rates of 20 and 5 °C/min, respectively. The above procedure was repeated once more to form closely packed and enlarged Au nanoislands for highly uniform and substantially increased SERS signals.47 Geometric Characterization of Au Nanoislands The SEM images were directly converted into binary images. The radius and interparticle distances of the Au nanoislands after single and repeated dewetting were calculated with ImageJ software, assuming that Au nanoislands are periodic arrays with uniform sizes. FDTD Calculation of E-Field Intensities The three-dimensional finite-difference time-domain (FDTD) simulation was performed to calculate the E-field enhancement of the nanoislands by directly importing the two-dimensional SEM images. The Au nanoislands were assumed to have a thickness of 58 nm on the dielectric surface (n = 1.4) with a uniform thin-film TiO2 layer on top of the nanoislands with respect to the target thickness. The TiO2 layer was disabled for the E-field comparison between single and repeated thermal dewetting. The E-field intensity was monitored at a 633 nm wavelength region positioned at 20 nm above the substrate. The average E-field intensity for the region of interest was utilized for the calculation of the relative E-field intensity (Enanocomposite4/EAu4). (Release: 2019a r6, Version 8.21.1933) Emulation of Oronasal Aerosols and Artificial Respiratory Solutions The oronasal emission of human respiratory aerosols was emulated using a commercial nebulizer (Teledyne CETAC Technologies) (Figure 2d). The carrier gas at a 200 sccm flow rate was controlled using a mass flow controller (MKS Instruments), while the Au-TiO2 nanocomposite SERS substrate was positioned 25 cm away from the nebulizer’s end in a fume hood. The average size of the nebulized aerosol is <10 μm. The nebulizing time was 10 seconds. The artificial respiratory solution was prepared to emulate respiratory aerosols from human breath. SARS-CoV-2 spike proteins with respective concentrations were diluted in a stock solution with final concentrations of 27.3 μM NaCl, 4.1 μM NH4NO3, 4.6 μM KH2PO4, 1.6 μM KCl, 1.0 μM C6H5K3O7, 0.1 μM C5H4N4O3, 3.3 μM CH4N2O, and 1.2 μM C3H6O3 in deionized water. SARS-CoV-2 Lysate Preparation The SARS-CoV-2 (BetaCoV/Korea/KCDC03/2020) was provided by the National Culture Collection for pathogens, which is operated by the Korea National Institute of Health. The virus was carefully cultured in a biosafety level 3 laboratory at the Korea Research Institute of Bioscience and Biotechnology (KRIBB) and heated with the lysis buffer.48 The stock SARS-CoV-2 lysates were then diluted to concentrations ranging from 101–104 pfu/mL and mixed in artificial respiratory solution prior to nebulizing. The control sample in Figure 4 refers to an artificial respiratory solution mixed with the culture media and lysis buffer without any viral lysates. SERS Signal Measurement of Nebulized R6G Aerosols The 10–7 M R6G was nebulized for 10 s, and the SERS signals were measured using a benchtop spectrometer equipped with a CCD camera and 50× objective lens (MicroSpec 2300i, Princeton Instruments) under excitation of 5 mW 633 nm HeNe laser. The acquisition time was 1 s. Training the Autoencoder Model First, five SERS spectra measured at different positions on the SERS substrate were obtained for different concentrations of aerosolized SARS-CoV-2 lysates in an artificial respiratory solution (control, 101, 102, 103, 104 pfu/mL). The measured SERS spectra with a vector length of 1096, containing the signal intensities for the Raman shift range of 600–1580 cm–1, were used for the quantitative analysis of SARS-CoV-2 lysates. The number of measured SERS data was augmented to improve the performance and the robustness of the prediction model and to prevent overfitting by adding random noise and applying the synthetic minority oversampling technique (SMOTE) algorithm provided by the Python package.49 Twenty synthetic data with random noise were generated from five measured SERS data sets, followed by generating 80 additional synthetic data from the SMOTE algorithm with a shrinkage value of 1.8 for each concentration class of SARS-CoV-2 lysates. The amplitude of noise for a wavenumber was randomly determined between ±1/5 of the original SERS value. One hundred vectors were acquired for each SARS-CoV-2 lysate concentration. The total 500 vectors, resulting from five different concentrations, were finally utilized for the autoencoder training. The model consists of a symmetric encoder and decoder with 1069 input nodes and two fully-connected layers, each consisting of 256 and 36 nodes and two latent nodes. The batch size and epoch were manually set as 32 and 40, respectively, with a 1e–4 learning rate and 1e–4 weight decay. Other hyperparameters for the linearizing autoencoder, such as weights of loss, are heuristically determined through the experiment. An entirely different data set of 500 test data was generated for validation and evaluation after the training—475 synthetic vectors were generated using the aforementioned protocol of generating synthetic training data and combined with 25 measured SERS data. Fifty data points on both sides of the nondiscriminant SERS bands of the SARS-CoV-2 spike were set to zero during the training of the ablation-assisted autoencoder. The same number of wavenumbers at random points was set to zero as a control set. The concentrations of the SARS-CoV-2 lysate data were classified by the logistic regression of the two-dimensional latent features from the autoencoder. The ROC curve is calculated during the classification and macro-averaged for every label. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.2c16446.SERS EF comparison between single and repeatedly dewetted Au nanoislands (Figure S1); EDX elemental analysis of Au-TiO2 nanocomposites (Figure S2); SERS intensity of nebulized R6G aerosols on Au-TiO2 nanocomposites (Figure S3); SERS spectrum of SARS-CoV-2 spike proteins for various concentrations (Figure S4); SERS spectra of artificial respiratory aerosols (Figure S5); and SERS spectra of nebulized SARS-CoV-2 lysates in artificial respiratory aerosols for concentrations ranging from 0 to 104 pfu/mL (Figure S6). The data that support the findings of this study are available upon reasonable request (PDF) Supplementary Material am2c16446_si_001.pdf Author Contributions ⊥ C.S.H.H. and S.L. contributed equally to this work. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. This work was supported by the Electronics and Telecommunications Research Institute, the Korea Research Institute of Bioscience & Biotechnology, the Korea Medical Device Development, and the National Research Foundation of Korea. The authors declare no competing financial interest. Acknowledgments This work was supported by ETRI (Electronics and Telecommunications Research Institute) internal funds (21YR2500), the Korea Medical Device Development fund grant funded by the Korean government (KMDF_PR_20200901_0074), the National Research Foundation of Korea (NRF) grant funded by the Korean government (2021R1A2B5B03002428, 2021M3E5E3080379), the Bio-Synergy Research Project (NRF-2012M3A9C4048758) of the Ministry of Science and ICT, and the KRIBB Research Initiative Program (1711134081). ==== Refs References Pereira J. ; Porto-Figueira P. ; Cavaco C. ; Taunk K. ; Rapole S. ; Dhakne R. ; Nagarajaram H. ; Camara J. S. Breath Analysis as a Potential and Non-Invasive Frontier in Disease Diagnosis: An Overview. Metabolites 2015, 5 , 3–55. 10.3390/metabo5010003.25584743 Chan L. W. ; Anahtar M. N. ; Ong T. H. ; Hern K. E. ; Kunz R. R. ; Bhatia S. N. Engineering Synthetic Breath Biomarkers for Respiratory Disease. Nat. Nanotechnol. 2020, 15 , 792–800. 10.1038/s41565-020-0723-4.32690884 Kharitonov S. A. ; Barnes P. J. Biomarkers of Some Pulmonary Diseases in Exhaled Breath. Biomarkers 2002, 7 , 1–32. 10.1080/13547500110104233.12101782 Chen Y. ; Zhang Y. ; Pan F. ; Liu J. ; Wang K. ; Zhang C. ; Cheng S. ; Lu L. ; Zhang W. ; Zhang Z. ; Zhi X. ; Zhang Q. ; Alfranca G. ; de la Fuente J. M. ; Chen D. ; Cui D. Breath Analysis Based on Surface-Enhanced Raman Scattering Sensors Distinguishes Early and Advanced Gastric Cancer Patients from Healthy Persons. ACS Nano 2016, 10 , 8169–8179. 10.1021/acsnano.6b01441.27409521 Pleil J. D. Breath Biomarkers in Toxicology. Arch. Toxicol. 2016, 90 , 2669–2682. 10.1007/s00204-016-1817-5.27585664 Amal H. ; Ding L. ; Liu B. B. ; Tisch U. ; Xu Z. Q. ; Shi D. Y. ; Zhao Y. ; Chen J. ; Sun R. X. ; Liu H. ; Ye S. L. ; Tang Z. Y. ; Haick H. The Scent Fingerprint of Hepatocarcinoma: In-Vitro Metastasis Prediction with Volatile Organic Compounds (VOCs). Int. J. Nanomed. 2012, 7 , 4135–4146. 10.2147/IJN.S32680. Shirasu M. ; Touhara K. The Scent of Disease: Volatile Organic Compounds of The Human Body Related to Disease and Disorder. J. Biochem. 2011, 150 , 257–266. 10.1093/jb/mvr090.21771869 Mazzatenta A. ; Pokorski M. ; Sartucci F. ; Domenici L. ; Di Giulio C. Volatile Organic Compounds (Vocs) Fingerprint of Alzheimer’s Disease. Respir. Physiol. Neurobiol. 2015, 209 , 81–84. 10.1016/j.resp.2014.10.001.25308706 van Doremalen N. ; Bushmaker T. ; Morris D. H. ; Holbrook M. G. ; Gamble A. ; Williamson B. N. ; Tamin A. ; Harcourt J. L. ; Thornburg N. J. ; Gerber S. I. ; Lloyd-Smith J. O. ; de Wit E. ; Munster V. J. Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1. N. Engl. J. Med. 2020, 382 , 1564–1567. 10.1056/NEJMc2004973.32182409 Tang S. ; Mao Y. ; Jones R. M. ; Tan Q. ; Ji J. S. ; Li N. ; Shen J. ; Lv Y. ; Pan L. ; Ding P. ; Wang X. ; Wang Y. ; MacIntyre C. R. ; Shi X. Aerosol Transmission of SARS-Cov-2? Evidence, Prevention and Control. Environ. Int. 2020, 144 , 106039 10.1016/j.envint.2020.106039.32822927 Lamote K. ; Janssens E. ; Schillebeeckx E. ; Lapperre T. S. ; De Winter B. Y. ; van Meerbeeck J. P. The Scent of COVID-19: Viral (Semi-)Volatiles As Fast Diagnostic Biomarkers?. J. Breath Res. 2020, 14 , 042001 10.1088/1752-7163/aba105.32599571 WHO. https://covid19.who.int/. CDC https://www.cdc.gov/. Ito K. ; Piantham C. ; Nishiura H. Relative instantaneous reproduction number of Omicron SARS-CoV-2 variant with respect to the Delta variant in Denmark. J. Med. Virol. 2022, 94 , 2265–2268. 10.1002/jmv.27560.34967453 Corman V. M. ; Landt O. ; Kaiser M. ; Molenkamp R. ; Meijer A. ; Chu D. K. ; Bleicker T. ; Brunink S. ; Schneider J. ; Schmidt M. L. ; Mulders D. G. ; Haagmans B. L. ; van der Veer B. ; van den Brink S. ; Wijsman L. ; Goderski G. ; Romette J. L. ; Ellis J. ; Zambon M. ; Peiris M. ; Goossens H. ; Reusken C. ; Koopmans M. P. ; Drosten C. Detection of 2019 Novel Coronavirus (2019-nCoV) by Real-time RT-PCR. Eurosurveillance 2020, 25 , 2000045 10.2807/1560-7917.ES.2020.25.3.2000045.31992387 Tahamtan A. ; Ardebili A. Real-time RT-PCR in COVID-19 Detection: Issues Affecting the Results. Expert Rev. Mol. Diagn. 2020, 20 , 453–454. 10.1080/14737159.2020.1757437.32297805 Kanaujia R. ; Biswal M. ; Angrup A. ; Ray P. Inhale, Then Exhale: Start Afresh to Diagnose Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-Cov-2) by Non-Invasive Face-Mask Sampling Technique. Clin. Microbiol. Infect. 2020, 26 , 1701–1702. 10.1016/j.cmi.2020.06.034.32629023 Daniels J. ; Wadekar S. ; DeCubellis K. ; Jackson G. W. ; Chiu A. S. ; Pagneux Q. ; Saada H. ; Engelmann I. ; Ogiez J. ; Loze-Warot D. ; et al. A Mask-based Diagnostic Platform for Point-Of-Care Screening of Covid-19. Biosens. Bioelectron. 2021, 192 , 113486 10.1016/j.bios.2021.113486.34260968 Dong X. ; Ohnoutek L. ; Yang Y. ; Feng Y. ; Wang T. ; Tahir M. A. ; Valev V. K. ; Zhang L. Cu/Ag Sphere Segment Void Array as Efficient Surface Enhanced Raman Spectroscopy Substrate for Detecting Individual Atmospheric Aerosol. Anal. Chem. 2019, 91 , 13647–13657. 10.1021/acs.analchem.9b02840.31580648 Craig R. L. ; Bondy A. L. ; Ault A. P. Surface Enhanced Raman Spectroscopy Enables Observations of Previously Undetectable Secondary Organic Aerosol Components at the Individual Particle Level. Anal. Chem. 2015, 87 , 7510–7514. 10.1021/acs.analchem.5b01507.26176648 Sivaprakasam V. ; Hart M. B. ; Eversole J. D. Surface Enhanced Raman Spectroscopy of Individual Suspended Aerosol Particles. J. Phys. Chem. C 2017, 121 , 22326–22334. 10.1021/acs.jpcc.7b05310. Phan-Quang G. C. ; Yang N. ; Lee H. K. ; Sim H. Y. F. ; Koh C. S. L. ; Kao Y. C. ; Wong Z. C. ; Tan E. K. M. ; Miao Y. E. ; Fan W. ; Liu T. ; Phang I. Y. ; Ling X. Y. Tracking Airborne Molecules from Afar: Three-Dimensional Metal-Organic Framework-Surface-Enhanced Raman Scattering Platform for Stand-Off and Real-Time Atmospheric Monitoring. ACS Nano 2019, 13 , 12090–12099. 10.1021/acsnano.9b06486.31518107 Caviezel D. ; Narayanan C. ; Lakehal D. Adherence and Bouncing of Liquid Droplets Impacting on Dry Surfaces. Microfluid. Nanofluid. 2008, 5 , 469–478. 10.1007/s10404-007-0248-2. Mhetre M. R. ; Abhyankar H. K. Human Exhaled Air Energy Harvesting with Specific Reference to PVDF Film. Eng. Sci. Technol. 2017, 20 , 332–339. 10.1016/j.jestch.2016.06.012. Lee W. S. ; Ahn J. ; Jung S. ; Lee J. ; Kang T. ; Jeong J. Biomimetic Nanopillar-based Biosensor for Label-Free Detection of Influenza A Virus. BioChip J. 2021, 15 , 260–267. 10.1007/s13206-021-00027-y.34122741 Potejanasak P. ; Duangchan S. Gold Nanoisland Agglomeration upon The Substrate Assisted Chemical Etching Based on Thermal Annealing Process. Crystals 2020, 10 , 533 10.3390/cryst10060533. Telecka A. ; Li T. ; Ndoni S. ; Taboryski R. Nanotextured Si Surfaces Derived from Block-Copolymer Self-Assembly with Superhydrophobic, Superhydrophilic, or Superamphiphobic Properties. RSC Adv. 2018, 8 , 4204–4213. 10.1039/C8RA00414E. Kang M. ; Kim J. J. ; Oh Y. J. ; Park S. G. ; Jeong K. H. A Deformable Nanoplasmonic Membrane Reveals Universal Correlations Between Plasmon Resonance and Surface Enhanced Raman Scattering. Adv. Mater. 2014, 26 , 4510–4514. 10.1002/adma.201305950.24668875 Oh Y. J. ; Jeong K. H. Glass Nanopillar Arrays with Nanogap-Rich Silver Nanoislands for Highly Intense Surface Enhanced Raman Scattering. Adv. Mater. 2012, 24 , 2234–2237. 10.1002/adma.201104696.22454295 Meng X. ; Dyer J. ; Huo Y. ; Jiang C. Greater SERS Activity of Ligand-Stabilized Gold Nanostars with Sharp Branches. Langmuir 2020, 36 , 3558–3564. 10.1021/acs.langmuir.0c00079.32176502 Gambinossi F. ; Mylon S. E. ; Ferri J. K. Aggregation Kinetics and Colloidal Stability of Functionalized Nanoparticles. Adv. Colloid Interface Sci. 2015, 222 , 332–349. 10.1016/j.cis.2014.07.015.25150615 El-Atab N. ; Chowdhury F. ; Ulusoy T. G. ; Ghobadi A. ; Nazirzadeh A. ; Okyay A. K. ; Nayfeh A. Similar to 3-nm ZnO Nanoislands Deposition and Application in Charge Trapping Memory Grown by Single ALD Step. Sci. Rep. 2016, 6 , 38712.27991492 Vogt K. W. ; Kohl P. A. ; Carter W. B. ; Bell R. A. ; Bottomley L. A. Characterization of Thin Titanium-Oxide Adhesion Layers on Gold - Resistivity, Morphology, and Composition. Surf. Sci. 1994, 301 , 203–213. 10.1016/0039-6028(94)91300-5. Lee S. J. ; Guan Z. ; Xu H. ; Moskovits M. Surface-enhanced Raman Spectroscopy and Nanogeometry: The Plasmonic Origin of SERS. J. Phys. Chem. C 2007, 111 , 17985–17988. 10.1021/jp077422g. Bake B. ; Larsson P. ; Ljungkvist G. ; Ljungstrom E. ; Olin A. C. Exhaled Particles and Small Airways. Respir. Res. 2019, 20 , 8 10.1186/s12931-019-0970-9.30634967 Liu D. ; Wu F. ; Cen Y. ; Ye L. ; Shi X. Y. ; Huang Y. L. ; Fang S. S. ; Ma L. Comparative Research on Nucleocapsid and Spike Glycoprotein as The Rapid Immunodetection Targets of COVID-19 and Establishment of Immunoassay Strips. Mol. Immunol. 2021, 131 , 6–12. 10.1016/j.molimm.2021.01.005.33450670 Daoudi K. ; Ramachandran K. ; Alawadhi H. ; Boukherroub R. ; Dogheche E. ; El Khakani M. A. ; Gaidi M. Ultra-sensitive and Fast Optical Detection of The Spike Protein of The SARS-Cov-2 Using Agnps/Sinws Nanohybrid Based Sensors. Surf. Interfaces 2021, 27 , 101454 10.1016/j.surfin.2021.101454.34957346 Zhang M. ; Li X. ; Pan J. ; Zhang Y. ; Zhang L. ; Wang C. ; Yan X. ; Liu X. ; Lu G. Ultrasensitive Detection of SARS-Cov-2 Spike Protein in Untreated Saliva Using SERS-Based Biosensor. Biosens. Bioelectron. 2021, 190 , 113421 10.1016/j.bios.2021.113421.34134070 Pezzotti G. ; Boschetto F. ; Ohgitani E. ; Fujita Y. ; Shin-Ya M. ; Adachi T. ; Yamamoto T. ; Kanamura N. ; Marin E. ; Zhu W. ; Nishimura I. ; Mazda O. Raman Molecular Fingerprints of SARS-CoV-2 British Variant and the Concept of Raman Barcode. Adv. Sci. 2022, 9 , 2103287 10.1002/advs.202103287. Gal J. Y. ; Fovet Y. ; Adib-Yadzi M. About a Synthetic Saliva for In Vitro Studies. Talanta 2001, 53 , 1103–1115. 10.1016/S0039-9140(00)00618-4.18968202 Zou S. ; Hou M. ; Li J. ; Ma L. ; Zhang Z. Semi-Quantitative Analysis of Multiple Chemical Mixtures in Solution at Trace Level by Surface-Enhanced Raman Scattering. Sci. Rep. 2017, 7 , 6186 10.1038/s41598-017-06543-y.28733621 Ciloglu F. U. ; Caliskan A. ; Saridag A. M. ; Kilic I. H. ; Tokmakci M. ; Kahraman M. ; Aydin O. Drug-Resistant Staphylococcus Aureus Bacteria Detection by Combining Surface-Enhanced Raman Spectroscopy (SERS) and Deep Learning Techniques. Sci. Rep. 2021, 11 , 18444 10.1038/s41598-021-97882-4.34531449 Thrift W. J. ; Ronaghi S. ; Samad M. ; Wei H. ; Nguyen D. G. ; Cabuslay A. S. ; Groome C. E. ; Santiago P. J. ; Baldi P. ; Hochbaum A. I. ; Ragan R. Deep Learning Analysis of Vibrational Spectra of Bacterial Lysate For Rapid Antimicrobial Susceptibility Testing. ACS Nano 2020, 14 , 15336–15348. 10.1021/acsnano.0c05693.33095005 Lee S. ; Kim H. ; Lee D. Linearization Autoencoder: an Autoencoder-based Regression Model with Latent Space Linearization bioRxiv 2022. Azzi L. ; Carcano G. ; Gianfagna F. ; Grossi P. ; Dalla Gasperina D. ; Genoni A. ; Fasano M. ; Sessa F. ; Tettamanti L. ; Carinci F. ; Maurino V. ; Rossi A. ; Tagliabue A. ; Baj A. Saliva is a Reliable Tool to Detect SARS-Cov-2. J. Infect. 2020, 81 , E45–E50. 10.1016/j.jinf.2020.04.005.32298676 Chung Y. S. ; Lee N. J. ; Woo S. H. ; Kim J. M. ; Kim H. M. ; Jo H. J. ; Park Y. E. ; Han M. G. Validation of Real-time RT-PCR For Detection of SARS-Cov-2 in The Early Stages of The COVID-19 Outbreak in The Republic of Korea. Sci. Rep. 2021, 11 , 14817 10.1038/s41598-021-94196-3.34285290 Kang M. ; Park S. G. ; Jeong K. H. Repeated Solid-state Dewetting of Thin Gold Films for Nanogap-rich Plasmonic Nanoislands. Sci. Rep. 2015, 5 , 14790 10.1038/srep14790.26469768 Cha H. ; Kim H. ; Joung Y. ; Kang H. ; Moon J. ; Jang H. ; Park S. ; Kwon H. J. ; Lee I. C. ; Kim S. ; Yong D. ; Yoon S. W. ; Park S. G. ; Guk K. ; Lim E. K. ; Park H. G. ; Choo J. ; Jung J. ; Kang T. Surface-enhanced Raman Scattering-based Immunoassay for Severe Acute Respiratory Syndrome Coronavirus 2. Biosens. Bioelectron. 2022, 202 , 114008 10.1016/j.bios.2022.114008.35086030 Lemaitre G. ; Nogueira F. ; Aridas C. K. Imbalanced-learn: A Python Toolbox to Tackle the Curse of Imbalanced Datasets in Machine Learning arXiv 2017, p 18. 10.48550/arXiv.1609.06570.	36448483	PMC9718102	NO-CC CODE	2022-12-05 23:15:03	no	ACS Appl Mater Interfaces. 2022 Dec 14; 14(49):54550-54557	utf-8	ACS Appl Mater Interfaces	2,022	10.1021/acsami.2c16446	oa_other
==== Front Radiology Radiology Radiology Radiology 0033-8419 1527-1315 Radiological Society of North America 36445225 222087 10.1148/radiol.222087 Original Research Thoracic Imaging Detection of Post-COVID-19 Lung Abnormalities: Photon-counting CT versus Same-day Energy-integrating Detector CT https://orcid.org/0000-0001-5607-8657 Prayer Florian MD PhD 1 https://orcid.org/0000-0001-6407-9561 Kienast Patric MD 1 https://orcid.org/0000-0002-7320-0279 Strassl Andreas BSc Msc 1 Moser Philipp. T. MD PhD 1 Bernitzky Dominik MD 2 https://orcid.org/0000-0002-6924-0075 Milacek Christopher MD 2 Gyöngyösi Mariann MD 3 https://orcid.org/0000-0001-8087-4303 Kifjak Daria MD 1 4 https://orcid.org/0000-0002-8303-6492 Röhrich Sebastian MD PhD 1 https://orcid.org/0000-0003-4388-7580 Beer Lucian MD PhD 1 Watzenboeck Martin L. MD 1 https://orcid.org/0000-0001-7254-5606 Milos Ruxandra I. MD 1 https://orcid.org/0000-0002-6468-5577 Wassipaul Christian Mag. MD 1 Gompelmann Daniela MD 2 https://orcid.org/0000-0001-7097-2057 Herold Christian J. MD 1 https://orcid.org/0000-0002-6119-6364 Prosch Helmut MD 1 https://orcid.org/0000-0002-4827-2457 Heidinger Benedikt H. MD 1 [email protected] Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Vienna, Austria Department of Medicine II, Division of Pulmonology, Medical University of Vienna, Vienna, Austria Department of Medicine II, Division of Cardiology, Medical University of Vienna, Vienna, Austria Department of Radiology, UMass Memorial Medical Center and University of Massachusetts Chan Medical School, Worcester, MA, USA Correspondence to: Benedikt Heidinger, MD Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna Währinger Gürtel 18-20, 1090 Vienna, Austria [email protected] 29 11 2022 29 11 2022 222087© 2022 by the Radiological Society of North America, Inc. 2022 This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background Photon-counting detector (PCD) CT allows ultra-high-resolution lung imaging and may shed light on morphologic correlates of persistent symptoms after COVID-19. Purpose To compare PCD CT with energy-integrating detector (EID) CT for noninvasive assessment of post-COVID-19 lung abnormalities. Materials and Methods For this prospective study, adult participants with one or more COVID-19-related persisting symptoms (resting or exertional dyspnea, cough, and fatigue) underwent same-day EID and PCD CT scans between April 2022 and June 2022. EID CT 1.0mm images and, subsequently, 1.0mm, 0.4mm, and 0.2mm PCD CT images were reviewed for the presence of lung abnormalities. Subjective and objective EID and PCD CT image quality was evaluated using a 5-point Likert scale (-2 to 2) and lung signal-to-noise ratios (SNR). Results Twenty participants (mean age, 54 years ±16 [SD], 10 men) were included. EID CT showed post-COVID-19 lung abnormalities in 15 of 20 (75%) participants with a median involvement of 10% of lung volume [IQR 0-45%], and 3.5 lobes [IQR 0-5]. Ground-glass opacities (GGO) and linear bands (both 10 of 20 participants, 50%) were the most frequent findings on EID CT. PCD CT revealed additional lung abnormalities in 10 of 20 (50%) participants, most commonly bronchiolectasis (10 of 20, 50%). Subjective image quality was improved for 1.0mm PCD vs. 1.0mm EID CT images (1 [IQR 1-2], P<.001) and 0.4mm vs. 1.0mm PCD CT images (1 [IQR 1-1], P<.001), but not for 0.4mm vs. 0.2mm PCD CT images (0 [IQR 0-0.5], P=.26). PCD CT delivered higher lung SNR vs. EID CT 1.0mm images (mean difference 0.53 ± 0.96, P=.03), but lower SNRs for 0.4mm vs. 1.0mm images, and 0.2mm vs. 0.4mm images, respectively (−1.52 ± 0.68, P<.001 and -1.15 ± 0.43, P<.001). Conclusion Photon-counting detector CT outperformed energy-integrating detector CT with regard to visualization of subtle post-COVID-19 lung abnormalities and image quality. ==== Body pmcSummary Photon-counting detector CT detected subtle post-COVID-19 lung abnormalities indicative of irreversible pulmonary fibrosis and provided superior image quality compared with conventional energy-integrating detector CT. Key Results ■ In a prospective study of 20 adults with COVID-19-related persisting symptoms undergoing same-day photon-counting detector (PCD) CT and energy-integrating detector (EID) CT, PCD CT revealed additional lung abnormalities in 10 of 20 participants (50%), most frequently bronchiolectasis (10 of 20 [50%]). ■ PCD CT provided higher subjective and objective image quality at 1.0mm section thickness than EID CT (median 5-point Likert scale: 1 [IQR 1 to 2], P<.001); mean lung signal-to-noise ratio: 0.53±0.96 [SD], P=.03). Introduction Following the acute phase of COVID-19, the persistence of symptoms such as chronic cough and exertional dyspnea (i.e., feeling short of breath during exercise) for longer than four weeks has previously been described as ‘post-acute COVID-19 syndrome’ (1) or ‘long COVID’ (2). To assess for lung sequelae as a potential cause for persistent respiratory symptoms, patients with post-COVID-19 symptoms are recommended to undergo CT of the chest (3, 4). Indeed, six months to one year following moderate to severe COVID-19, between 24 and 72% of patients show lung sequelae on CT, most frequently ground-glass opacities (GGO) and subpleural reticulations (5-8). However, subtle lung abnormalities may be below the spatial resolution of conventional energy-integrating detector (EID) CT. These abnormalities include fine reticulations, faint GGO, or bronchiolectasis, indicative of incipient pulmonary fibrosis. Thus, current state-of-the-art EID CT may not recognize early stages of pulmonary fibrosis following COVID-19, hindering timely treatment allocation. Technological advances have led to the introduction and approval of photon counting detector CT (PCD-CT) in 2021 (9). In contrast to EID, PCD uses continuous semiconductor materials, eliminating the need for metal septae between detector elements and enabling individual photon energy measurement (10). Thereby, PCD CT allows ultra-high resolution image acquisition with a section thickness of 0.2 mm compared with the current standard of 1.0 mm in high resolution EID CT (11). While PCD CT may allow detection of subtle lung abnormalities and provide insights into the morphologic correlates of persistent respiratory symptoms in participants after COVID-19, there is a lack of evidence regarding its application in this population. Therefore, the aim of this study was to compare PCD CT with EID CT for noninvasive assessment of post-COVID-19 lung abnormalities in symptomatic participants after COVID-19 using same-day chest CT scans. Materials and Methods This prospective study was approved by the institutional review board of the Medical University of Vienna (No 2065/2017). Written informed consent was obtained from all study participants prior to study inclusion. The Department of Biomedical Imaging and Image-guided Therapy at the Medical University of Vienna has an institutional research grant by Siemens Healthineers. The authors had full control of the data and information submitted for publication. Participants Consecutive participants with post-COVID-19 symptoms referred to the Department of Radiology of a single tertiary care university hospital between April 2022 and June 2022 for CT of the chest due to persistent symptoms were screened for inclusion and exclusion criteria. Inclusion criteria consisted of the following: a) clinically-indicated chest CT to evaluate or monitor post-COVID-19 lung sequelae; b) prior history of COVID-19 including at least one positive polymerase chain reaction (PCR) test for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); c) at least 28 days between initial positive SARS-CoV-2 PCR test and CT; d) age of 18 years or older; e) one or more of the following COVID-19-related persisting symptoms: resting or exertional dyspnea, cough, and fatigue. Exclusion criteria consisted of a) clinical CT indication requiring administration of contrast (e.g. to exclude pulmonary thromboembolism); b) pregnancy; c) current symptoms of acute infection; d) prior history of interstitial lung disease, lung malignancy, or lung surgery; e) inability to give informed consent. A sample size of n=20 was defined for this experimental study including same-day EID CT and PCD CT scans. Current symptoms, symptoms during COVID-19 (up to 28 days following the initial positive severe acute respiratory syndrome coronavirus 2 [SARS-Cov2] PCR test), need for hospitalization, intensive care, and intubation were recorded. CT A clinically indicated conventional chest EID CT scan was acquired using a third-generation multi-detector dual-source CT system (Somatom Drive, Siemens Healthineers). On the same day, an additional PCD CT scan in ultra-high-resolution scan mode was acquired for research purposes using a dual-source PCD CT system approved for clinical use (Naeotom Alpha, Siemens Healthineers). All CT scans were performed with the participant in supine body position with elevated arms, in a craniocaudal scan direction, in full inspiration breath-hold, and without administration of intravenous contrast agent. Detailed CT image acquisition and reconstruction parameters are provided in Table E1. CT dose index (CTDIvol) and dose length product (DLP) for each CT scan, as well as the time between EID CT and PCD CT scans were recorded for each participant. Image analysis CT images were independently assessed by a radiologist with nine years of experience in lung imaging (reader 1, B.H.H.) and a radiologist-in-training with five years of experience (reader 2, F.P.), who were blinded to imaging reports and clinical data. All images were assessed using a diagnostic quality PACS display (BARCO NV) and readers were allowed to change window settings. EID CT images were reviewed for the presence as well as overall extent (lobes affected) of the following post-COVID-19 lung abnormalities described by Solomon et al. (12) and defined according to the Fleischner Society's glossary of terms for thoracic imaging (13): bronchial wall thickening, bronchiectasis, bronchiolectasis, consolidation, GGO, honeycombing, linear bands, mosaic attenuation, pleural thickening, reticulation, and volume loss. Subsequently, PCD CT images with 1.0mm, 0,4mm and 0.2mm section thicknesses were reviewed for the presence of additional lung abnormalities, defined as abnormalities in lung areas where EID CT showed no or different (e.g. reticulations in PCD CT images classified as GGO based on EID CT images) lung abnormalities. To investigate the detectability of subtle lung abnormalities by PCD CT and considering the limited practicality of blinded comparison between the obviously much larger number of particularly PCD CT images compared with EID CT images, a stepwise approach was taken: First, 1.0mm PCD CT images were compared with 1.0mm EID CT images. Second, 0.4mm PCD CT images were compared with 1.0mm PCD CT images. Third, 0.2mm PCD CT images were compared with 0.4mm PCD CT images. Following the same principle, subjective image quality differences were assessed according to a five-point Likert scale (14) (-2 definitely worse with likely effect on detectability of lung abnormalities, -1 definitely worse with unclear effect on detectability of lung abnormalities, 0 similar without decrement or benefit, 1 definitely better with unclear effect on detectability of lung abnormalities, 2 definitely better with likely effect on detectability of lung abnormalities). CT readings were reported for reader 1, and absolute inter-observer agreement between reader 1 and reader 2 was calculated. Finally, to evaluate objective image quality, circular regions of interest (ROIs) with a diameter of 20 to 30 mm were placed in normal-appearing lung parenchyma avoiding large bronchi and vessels at the same location on EID CT (1.0 mm) and PCD CT (1.0 mm, 0.4 mm, 0.2 mm) images (15). Signal-to-noise ratio (SNR) was then calculated by dividing the mean HU density of each ROI by its standard deviation. This was performed three times for each possible combination of CT scanner and section thickness to obtain average lung SNRs. Statistical analysis Statistical analyses were performed by B.H.H. using Stata (v14.2; StataCorp LP). Continuous variables (age, body mass index, CTDIvol, DLP, extent of lung abnormalities, height, lung signal-to-noise ratio, time COVID-19 to CT scan, weight) were tested for normal distribution using the Shapiro-Wilk tests. Normally distributed variables were expressed as mean ± standard deviation, non-normally distributed as median along with their IQR. Categorical variables (affected lobes, gender, hospitalization, lung abnormalities, need for intensive care, need for intubation, symptoms) were expressed as absolute numbers and their percentages. Subjective image quality differences between PCD CT images and respective reference images (1.0mm PCD CT vs. 1.0mm EID CT images, 0.4mm vs. 1.0mm PCD CT images, and 0.2mm vs. 0.4mm PCD CT images) were assessed for statistical significance using Wilcoxon signed-rank tests. ANOVA for repeated measurements was performed to test for statistically significant differences in lung signal-to-noise ratio among the unique scanner and reconstruction combinations with post-hoc pairwise comparisons corrected for multiple testing using Bonferroni method. Absolute and relative (percentage) interobserver agreement with regard to the presence of lung abnormalities on EID CT, and additional lung abnormalities on PCD CT was calculated. The differences between the two readers were evaluated using McNemar tests or Wilcoxon signed-rank test. A two-sided P-value <.05 was considered statistically significant. Results Participant Characteristics Twenty participants with post-COVID-19 symptoms (mean age, 54 years ±16 [SD], 10 men) formed the study sample after 47 participants were excluded (see Fig 1). Key characteristics of the study sample are given in Table 1. The median time interval between COVID-19 (day of first positive SARS-CoV-2 PCR test) and same-day EID CT and PCD CT scans was 101 days [IQR 89 to 219 days]. The median time between same-day EID CT and PCD CT scans was 26 min [IQR 19 to 40 min]. Figure 1: Flow chart of included and excluded individuals with persistent symptoms after COVID-19; PCR polymerase chain reaction; SARS-CoV2 severe acute respiratory syndrome coronavirus 2. Flow chart of included and excluded individuals with persistent symptoms after COVID-19; PCR polymerase chain reaction; SARS-CoV2 severe acute respiratory syndrome coronavirus 2. Table 1: Characteristics of Study Participants CT On EID CT, lung abnormalities were found in 15/20 (75%) participants with a median extent of 10% [IQR 0 to 45%] lung volume and 3.5 [IQR 0 to 5] affected lobes. The most frequently observed lung abnormalities were GGO and linear bands (both 10/20, 50%). PCD CT revealed additional lung abnormalities in 10/20 (50%) participants with post-COVID-19 symptoms compared with EID CT. Most additional findings were identified in 0.4mm PCD CT images compared with 1.0mm PCD CT images (17 abnormalities in 9 participants), followed by 1.0mm PCD CT images compared with 1.0mm EID CT images (5 abnormalities in 4 participants), and 0.2mm PCD CT images compared with 0.4mm PCD CT images (1 abnormality in 1 participant). Table 2 provides an overview of participants with lung abnormalities on EID and PCD CT. Table E2 semi-quantitatively describes the extent of lung abnormalities detected by PCD CT. The most frequently observed additional lung abnormalities detected with PCD CT were bronchiolectasis (10/20 participants, 50%, Fig 2) and reticulations (7/20 participants, 35%). In six participants, fine reticulations were visible on PCD CT in areas classified as GGO on EID CT (Fig 3). No additional abnormalities were found in the five participants that did not exhibit lung abnormalities on EID CT. Table 3 details additional lung abnormalities detected with PCD CT at different section thicknesses. Table 2: Participants with Post-COVID-19 Lung Abnormalities in Energy-integrating Detector CT and Photon-counting Detector CT Figure 2: Ultra-high-resolution photon-counting CT reveals bronchiolectasis: Axial CT lung images without contrast agent of a 70-year-old woman with persistent fatigue 401 days after COVID-19: 1.0mm image obtained with energy-integrating detector (EID) CT (A), and 1.0mm (B), 0.4mm (C) and 0.2mm (D) images obtained with photon-counting detector CT at the same level. Bronchiolectasis (white arrow) was not detected by EID CT but was found by PCD-CT. Ground-glass opacity detected in EID CT images (black arrows in A) was found to contain reticulations in PCD CT images (black arrows in B-D). Ultra-high-resolution photon-counting CT reveals bronchiolectasis: Axial CT lung images without contrast agent of a 70-year-old woman with persistent fatigue 401 days after COVID-19: 1.0mm image obtained with energy-integrating detector (EID) CT (A), and 1.0mm (B), 0.4mm (C) and 0.2mm (D) images obtained with photon-counting detector CT at the same level. Bronchiolectasis (white arrow) was not detected by EID CT but was found by PCD-CT. Ground-glass opacity detected in EID CT images (black arrows in A) was found to contain reticulations in PCD CT images (black arrows in B-D). Figure 3: Ultra-high-resolution photon-counting CT reveals reticulations in lung areas classified as ground-glass opacity by high-resolution energy-integrating CT: Axial CT lung images without contrast agent of a 55-year-old man with persistent exertional dyspnea and chronic fatigue 399 days after COVID-19: 1.0mm image obtained with energy-integrating detector (EID) CT (A), and 1mm (B), 0.4mm (C) and 0.2mm (D) images obtained with photon-counting detector CT at the same level. Ground-glass opacities detected by EID CT (black arrows in A) were found to contain reticulations in PCD CT (black arrows in B-D). Ultra-high-resolution photon-counting CT reveals reticulations in lung areas classified as ground-glass opacity by high-resolution energy-integrating CT: Axial CT lung images without contrast agent of a 55-year-old man with persistent exertional dyspnea and chronic fatigue 399 days after COVID-19: 1.0mm image obtained with energy-integrating detector (EID) CT (A), and 1mm (B), 0.4mm (C) and 0.2mm (D) images obtained with photon-counting detector CT at the same level. Ground-glass opacities detected by EID CT (black arrows in A) were found to contain reticulations in PCD CT (black arrows in B-D). Table 3: Additional Post-COVID-19 Lung Abnormalities Detected by Photon-counting CT Overall and for Different Section Thicknesses Mean CTDIvol and DLP of PCD CT (6.3 ± 2.0mGy; 199.4 ± 60.7mGycm) were higher than of EID CT (5.2 ± 1.8 mGy; 181.9 ± 62.8 mGycm) (P<.001; P=.009). Image quality Subjective image quality improved in 1.0 mm PCD CT compared with 1.0 mm EID CT images (1 [IQR 1 to 2], P<.001), and in 0.4 mm PCD CT compared with 1.0 mm PCD CT images (1 [IQR 1 to 1], P<.001) (Fig 4). No evidence of a difference was observed between 0.4 mm and 0.2 mm PCD CT images (0 [IQR 0 to 0.5], P=.26). Figure 4: Comparison of image quality between energy-integrating detector and photon-counting detector CT images: Axial CT lung images without contrast agent of a 66-year-old man with persistent exertional dyspnea, chronic fatigue and anosmia 94 days after COVID-19: Subjective image quality was rated as ‘definitely better with likely effect on detectability of lung abnormalities’ for 1.0mm photon-counting detector (PCD) CT (B) compared with 1.0mm energy-integrating detector (EID) CT (A) images, ‘definitely better with unclear effect on detectability of lung abnormalities’ for 0.4mm (C) compared with 1.0mm (B) PCD CT images, and ‘similar without decrement or benefit’ for 0.2mm (D) compared with 0.4mm (C) PCD CT images. Comparison of image quality between energy-integrating detector and photon-counting detector CT images: Axial CT lung images without contrast agent of a 66-year-old man with persistent exertional dyspnea, chronic fatigue and anosmia 94 days after COVID-19: Subjective image quality was rated as ‘definitely better with likely effect on detectability of lung abnormalities’ for 1.0mm photon-counting detector (PCD) CT (B) compared with 1.0mm energy-integrating detector (EID) CT (A) images, ‘definitely better with unclear effect on detectability of lung abnormalities’ for 0.4mm (C) compared with 1.0mm (B) PCD CT images, and ‘similar without decrement or benefit’ for 0.2mm (D) compared with 0.4mm (C) PCD CT images. Objective image quality, measured as signal-to-noise ratio (SNR), was higher on 1.0 mm PCD CT images than on 1.0 mm EID CT images (0.53 ± 0.96, P=.03). On PCD-CT, SNR was reduced between images with 1.0 mm and 0.4 mm section thicknesses, and between images with 0.4 mm and 0.2 mm section thicknesses (-1.52 ± 0.68 and -1.15 ± 0.43, both P<.001). Table 4 summarizes subjective and objective image quality parameters for EID CT and PCD CT lung images. Table 4: Subjective and Objective Image Quality Differences between Energy-integrating Detector and Photon-counting Detector CT Interobserver agreement Interobserver agreement with regard to the presence of each lung abnormality (e.g. reticulation, GGO) on EID CT, and additional lung abnormalities on PCD CT ranged between 18/20 (90%) and 20/20 (100%) with the observed differences not being statistically significant (P=.50 to >.99; see Table E3 for details). Interobserver agreement with regard to subjective comparison of image quality was 19/20 (95%) for 1.0mm EID CT vs. 1.0mm PCD-CT, 20/20 (100%) for 0.4mm vs. 1.0mm PCD-CT, and 18/20 (90%) for 0.2mm vs. 0.4mm PCD-CT. The observed differences were not statistically significant (P=.16 to >.99). Discussion Ultra-high-resolution photon-counting detector (PCD) CT lung imaging may improve the detection of subtle lung abnormalities in patients with persisting symptoms after COVID-19, but evidence is lacking regarding its application in this population. In this prospective PCD CT study of post-COVID-19 lung changes, twenty symptomatic participants received same-day conventional energy-integrating detector (EID CT) scans and PCD CT scans. PCD CT revealed additional lung abnormalities compared with EID CT in 10/20 (50%) participants, most frequently bronchiolectasis. Non-specific ground-glass opacities (GGO) on EID CT images revealed to include fine reticulations, a possible sign of incipient pulmonary fibrosis, on PCD CT scans in 6/20 (30%) participants. In addition, using standard 1.0mm sections (16), PCD CT delivered improved subjective and objective image quality compared with EID CT (median on a 5-point Likert scale: 1 [IQR 1-2], P<.001; mean signal-to-noise (SNR) difference 0.53 ± 0.96, P=.03). Thereby, PCD CT was shown to enable detection and characterization of subtle post-COVID-19 lung abnormalities, which may help shed light on the morphologic correlates with respiratory symptoms in participants with post-COVID-19 symptoms in the future. While acute phase COVID-19 lung disease patterns have been well characterized (17), there is increasing evidence of morphologic lung sequelae on CT in a substantial number of patients following COVID-19 pneumonia (5-8, 12, 18-24). GGO, subpleural bands, and reticulations were the most commonly observed pulmonary sequelae one year after severe COVID-19 pneumonia in previous literature (21, 22, 24), and in our current study sample. Recently, air-trapping in expiratory chest CT scans has been described as another frequent finding of participants with post-COVID-19 symptoms (25, 26), indicating the presence of small airways disease. However, subtle lung abnormalities in participants with post-COVID-19 symptoms may escape detection with conventional EID CT due to limitations in spatial resolution. Indeed, in our study, PCD CT detected additional subtle lung abnormalities in participants with post-COVID-19 symptoms, which were not perceivable on EID CT. The most common additional finding revealed by PCD CT was bronchiolectasis, which was observed in half of all participants. While the pathophysiology of persistent lung abnormalities following COVID-19 pneumonia is still poorly understood, the presence of bronchiolectasis is a crucial finding, as it may serve as an early indicator of pulmonary fibrosis. In addition, PCD CT allowed visualization of fine reticular opacities in lung areas previously characterized as GGO with EID CT in 6/20 (35%) participants. This finding corroborates observations by Inoue et al., who reported several instances of GGO on EID CT that represented reticulations in same-day PCD CT of patients with suspected interstitial lung disease (27). GGO represents non-specific lung changes that may be caused by a wide spectrum of reversible or irreversible causes including interstitial abnormalities, inflammation, infection, edema, hemorrhage, and malignant disease (28). Thus, the unmasking of fine reticular opacities within areas of GGO by PCD CT may help identify patients at risk of developing irreversible fibrosis. Our study confirms existing reports on the improved image quality delivered by chest PCD CT compared with EID CT (15, 29). PCD CT images reconstructed following the recommendations for CT imaging of interstitial lung abnormalities (16, 30), using an edge-enhancing kernel and 1.0mm sections were subjectively perceived of higher quality than EID CT images in all (20/20, 100%) participants. However, for PCD CT images, there seemed to be a trade-off between thinner sections and subjective image quality: While PCD CT images with 0.4mm section thickness were rated as ‘definitely better’ compared with 1.0mm in all (20/20, 100%) participants, images with 0.2mm section thickness were considered of higher quality to 0.4mm sections in 5/20 (25%) participants but of lower quality in 2/20 (10%) participants. Objective signal-to-noise ratio (SNR) analysis confirmed the visual impression as PCD CT delivered significantly higher lung SNR compared with EID CT for 1.0mm sections, but as expected, SNR was reduced for 0.4mm and 0.2mm sections. Optimal section thickness for PCD CT lung imaging is thus far unclear. Previous studies assessing image quality of human lung PCD CT used section thicknesses of 1.0mm (15, 29), 0.6mm (27, 31), and 0.25mm (in a single human volunteer) (32), but, to our knowledge, no prior study used 0.4mm and 0.2mm section thickness for human lung imaging as shown here. Our findings suggest that 0.4mm sections represent a favorable compromise between ultra-high-resolution and perceived (and quantitative) image quality, permitting confident detection of subtle lung abnormalities. Noteworthy, instead of applying the 120×0.2mm collimation of the UHR scan mode, a 144×0.4mm collimation could deliver comparable PCD CT images while reducing scan time and increasing dose efficiency. Our study had limitations that may affect the generalizability of the results. First, as in a previous study, same-day EID CT and PCD CT scans facilitated comparison of lung abnormalities but limited the sample size (27). Second, a newly available clinical PCD CT scanner was used with a fixed tube voltage of 120 kVp for image acquisition, which explains the slightly higher amounts of radiation exposure compared with EID CT. Since the introduction of PCD CT, software updates have enabled automatic tube voltage selection, allowing reduction of radiation exposure for lung imaging beneath EID CT levels (33). Third, to enable detection of subtle lung abnormalities, the sharpest available edge-enhancing CT reconstruction kernels were used, which varied slightly between EID CT (BI57) and PCD CT (BI64) scanners. Fourth, due to the lack of histopathological correlation, false positive findings by PCD CT or missed lung abnormalities in all EID and PCD CT images cannot be ruled out. Fifth, the sequential rating of EID and PCD CT images may have biased subjective image quality rating, and may have failed to reflect abnormalities visible in EID but not in PCD CT images. Sixth, while PCD CT could detect subtle lung abnormalities indicative of irreversible lung damage, optimal treatment of persistent post-COVID-19 symptoms is still a matter of intense research (34). In conclusion, photon-counting detector (PCD) CT revealed subtle lung abnormalities in symptomatic participants with persistent post-COVID-19 symptoms that were not detectable on energy-integrating detector (EID) CT but may be indicative of irreversible fibrosis. In addition, PCD CT image quality was perceived higher compared with conventional EID CT, improving diagnostic confidence. Further research is required to confirm the potentially leading role of PCD CT lung imaging in screening and monitoring of post-COVID-19 lung sequelae. Appendix E1 Table E1: Image acquisition and reconstruction parameters of energy-integrating detector and photon-counting detector CT Table E2: Semi-quantitative rating of additional post-COVID-19 lung abnormalities detected by photon-counting detector CT Table E3: Interobserver agreement regarding the presence of lung abnormalities on energy-integrating detector CT and the presence of specific additional lung abnormalities on photon-counting detector CT This work originated at the Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Währinger Gürtel 18-20, 1090 Vienna, Austria. The institutional Department of Biomedical Imaging and Image-guided Therapy Medical University of Vienna has an institutional research grant form Siemens Healthineers. Data generated or analyzed during the study are available from the corresponding author by request. List of Abbreviations: EID energy-integrating detector GGO ground-glass opacity PCD photon-counting detector PCR Polymerase chain reaction ==== Refs References 1. Nalbandian A , Sehgal K , Gupta A , Madhavan MV , McGroder C , Stevens JS , et al . Post-acute COVID-19 syndrome . Nat Med . 2021 ; 27 ( 4 ): 601 – 15 . Epub 2021/03/24. doi: 10.1038/s41591-021-01283-z . PubMed PMID: 33753937; PubMed Central PMCID: PMCPMC8893149 . 33753937 2. Sudre CH , Murray B , Varsavsky T , Graham MS , Penfold RS , Bowyer RC , et al . Attributes and predictors of long COVID . Nat Med . 2021 ; 27 ( 4 ): 626 – 31 . Epub 2021/03/12. doi: 10.1038/s41591-021-01292-y . PubMed PMID: 33692530; PubMed Central PMCID: PMCPMC7611399 . 33692530 3. Martini K , Larici AR , Revel MP , Ghaye B , Sverzellati N , Parkar AP , et al . COVID-19 pneumonia imaging follow-up: when and how? A proposition from ESTI and ESR . Eur Radiol . 2022 ; 32 ( 4 ): 2639 – 49 . Epub 2021/10/30. doi: 10.1007/s00330-021-08317-7 . PubMed PMID: 34713328; PubMed Central PMCID: PMCPMC8553396 . 34713328 4. Yelin D , Moschopoulos CD , Margalit I , Gkrania-Klotsas E , Landi F , Stahl JP , et al . ESCMID rapid guidelines for assessment and management of long COVID . Clin Microbiol Infect . 2022 ; 28 ( 7 ): 955 – 72 . Epub 2022/02/20. doi: 10.1016/j.cmi.2022.02.018 . PubMed PMID: 35182760; PubMed Central PMCID: PMCPMC8849856 . 35182760 5. Caruso D , Guido G , Zerunian M , Polidori T , Lucertini E , Pucciarelli F , et al . Post-Acute Sequelae of COVID-19 Pneumonia: Six-month Chest CT Follow-up . Radiology . 2021 ; 301 ( 2 ): E396 – E405 . Epub 2021/07/28. doi: 10.1148/radiol.2021210834 . PubMed PMID: 34313468; PubMed Central PMCID: PMCPMC8335814 . 34313468 6. Chen Y , Ding C , Yu L , Guo W , Feng X , Yu L , et al . One-year follow-up of chest CT findings in patients after SARS-CoV-2 infection . BMC Med . 2021 ; 19 ( 1 ): 191 . Epub 2021/08/10. doi: 10.1186/s12916-021-02056-8 . PubMed PMID: 34365975; PubMed Central PMCID: PMCPMC8349604 . 34365975 7. Huang C , Huang L , Wang Y , Li X , Ren L , Gu X , et al . 6-month consequences of COVID-19 in patients discharged from hospital: a cohort study . Lancet . 2021 ; 397 ( 10270 ): 220 – 32 . Epub 2021/01/12. doi: 10.1016/S0140-6736(20)32656-8 . PubMed PMID: 33428867; PubMed Central PMCID: PMCPMC7833295 . 33428867 8. Wu X , Liu X , Zhou Y , Yu H , Li R , Zhan Q , et al . 3-month, 6-month, 9-month, and 12-month respiratory outcomes in patients following COVID-19-related hospitalisation: a prospective study . Lancet Respir Med . 2021 ; 9 ( 7 ): 747 – 54 . Epub 2021/05/09. doi: 10.1016/S2213-2600(21)00174-0 . PubMed PMID: 33964245; PubMed Central PMCID: PMCPMC8099316 . 33964245 9. Rajendran K , Petersilka M , Henning A , Shanblatt ER , Schmidt B , Flohr TG , et al . First Clinical Photon-counting Detector CT System: Technical Evaluation . Radiology . 2022 ; 303 ( 1 ): 130 – 8 . Epub 2021/12/15. doi: 10.1148/radiol.212579 . PubMed PMID: 34904876; PubMed Central PMCID: PMCPMC8940675 . 34904876 10. Leng S , Bruesewitz M , Tao S , Rajendran K , Halaweish AF , Campeau NG , et al . Photon-counting Detector CT: System Design and Clinical Applications of an Emerging Technology . Radiographics . 2019 ; 39 ( 3 ): 729 – 43 . Epub 2019/05/07. doi: 10.1148/rg.2019180115 . PubMed PMID: 31059394; PubMed Central PMCID: PMCPMC6542627 . 31059394 11. Ferda J , Vendis T , Flohr T , Schmidt B , Henning A , Ulzheimer S , et al . Computed tomography with a full FOV photon-counting detector in a clinical setting, the first experience . Eur J Radiol . 2021 ; 137 : 109614 . Epub 2021/03/04. doi: 10.1016/j.ejrad.2021.109614 . PubMed PMID: 33657475 . 33657475 12. Solomon JJ , Heyman B , Ko JP , Condos R , Lynch DA . CT of Post-Acute Lung Complications of COVID-19 . Radiology . 2021 ; 301 ( 2 ): E383 – E95 . Epub 2021/08/11. doi: 10.1148/radiol.2021211396 . PubMed PMID: 34374591; PubMed Central PMCID: PMCPMC8369881 . 34374591 13. Hansell DM , Bankier AA , MacMahon H , McLoud TC , Muller NL , Remy J . Fleischner Society: glossary of terms for thoracic imaging . Radiology . 2008 ; 246 ( 3 ): 697 – 722 . Epub 2008/01/16. doi: 10.1148/radiol.2462070712 . PubMed PMID: 18195376 . 18195376 14. Bartlett DJ , Koo CW , Bartholmai BJ , Rajendran K , Weaver JM , Halaweish AF , et al . High-Resolution Chest Computed Tomography Imaging of the Lungs: Impact of 1024 Matrix Reconstruction and Photon-Counting Detector Computed Tomography . Invest Radiol . 2019 ; 54 ( 3 ): 129 – 37 . Epub 2018/11/22. doi: 10.1097/RLI.0000000000000524 . PubMed PMID: 30461437; PubMed Central PMCID: PMCPMC6363870 . 30461437 15. Woeltjen MM , Niehoff JH , Michael AE , Horstmeier S , Moenninghoff C , Borggrefe J , et al . Low-Dose High-Resolution Photon-Counting CT of the Lung: Radiation Dose and Image Quality in the Clinical Routine . Diagnostics (Basel) . 2022 ; 12 ( 6 ). Epub 2022/06/25. doi: 10.3390/diagnostics12061441 . PubMed PMID: 35741251; PubMed Central PMCID: PMCPMC9221815 . 16. Hatabu H , Hunninghake GM , Richeldi L , Brown KK , Wells AU , Remy-Jardin M , et al . Interstitial lung abnormalities detected incidentally on CT: a Position Paper from the Fleischner Society . Lancet Respir Med . 2020 ; 8 ( 7 ): 726 – 37 . Epub 2020/07/11. doi: 10.1016/S2213-2600(20)30168-5 . PubMed PMID: 32649920; PubMed Central PMCID: PMCPMC7970441 . 32649920 17. Kanne JP , Bai H , Bernheim A , Chung M , Haramati LB , Kallmes DF , et al . COVID-19 Imaging: What We Know Now and What Remains Unknown . Radiology . 2021 ; 299 ( 3 ): E262 – E79 . Epub 2021/02/10. doi: 10.1148/radiol.2021204522 . PubMed PMID: 33560192; PubMed Central PMCID: PMCPMC7879709 . 33560192 18. Luger AK , Sonnweber T , Gruber L , Schwabl C , Cima K , Tymoszuk P , et al . Chest CT of Lung Injury 1 Year after COVID-19 Pneumonia: The CovILD Study . Radiology . 2022 ; 304 ( 2 ): 462 – 70 . Epub 2022/03/30. doi: 10.1148/radiol.211670 . PubMed PMID: 35348379; PubMed Central PMCID: PMCPMC8988857 . 35348379 19. Vijayakumar B , Tonkin J , Devaraj A , Philip KEJ , Orton CM , Desai SR , et al . CT Lung Abnormalities after COVID-19 at 3 Months and 1 Year after Hospital Discharge . Radiology . 2022 ; 303 ( 2 ): 444 – 54 . Epub 2021/10/06. doi: 10.1148/radiol.2021211746 . PubMed PMID: 34609195; PubMed Central PMCID: PMCPMC8515207 . 34609195 20. Han X , Fan Y , Alwalid O , Zhang X , Jia X , Zheng Y , et al . Fibrotic Interstitial Lung Abnormalities at 1-year Follow-up CT after Severe COVID-19 . Radiology . 2021 ; 301 ( 3 ): E438 – E40 . Epub 2021/07/28. doi: 10.1148/radiol.2021210972 . PubMed PMID: 34313470; PubMed Central PMCID: PMCPMC8335811 . 34313470 21. Lorent N , Vande Weygaerde Y , Claeys E , Guler Caamano Fajardo I , De Vos N , De Wever W , et al . Prospective longitudinal evaluation of hospitalised COVID-19 survivors 3 and 12 months after discharge . ERJ Open Res . 2022 ; 8 ( 2 ). Epub 2022/04/14. doi: 10.1183/23120541.00004-2022 . PubMed PMID: 35415186; PubMed Central PMCID: PMCPMC8994962 . 22. Pan F , Yang L , Liang B , Ye T , Li L , Li L , et al . Chest CT Patterns from Diagnosis to 1 Year of Follow-up in Patients with COVID-19 . Radiology . 2022 ; 302 ( 3 ): 709 – 19 . Epub 2021/10/06. doi: 10.1148/radiol.2021211199 . PubMed PMID: 34609153; PubMed Central PMCID: PMCPMC8515211 . 34609153 23. Zhan Y , Zhu Y , Wang S , Jia S , Gao Y , Lu Y , et al . SARS-CoV-2 immunity and functional recovery of COVID-19 patients 1-year after infection . Signal Transduct Target Ther . 2021 ; 6 ( 1 ): 368 . Epub 2021/10/15. doi: 10.1038/s41392-021-00777-z . PubMed PMID: 34645784; PubMed Central PMCID: PMCPMC8512652 . 34645784 24. Zhao Y , Yang C , An X , Xiong Y , Shang Y , He J , et al . Follow-up study on COVID-19 survivors one year after discharge from hospital . Int J Infect Dis . 2021 ; 112 : 173 – 82 . Epub 2021/09/15. doi: 10.1016/j.ijid.2021.09.017 . PubMed PMID: 34520845; PubMed Central PMCID: PMCPMC8434916 . 34520845 25. Cho JL , Villacreses R , Nagpal P , Guo J , Pezzulo AA , Thurman AL , et al . Quantitative Chest CT Assessment of Small Airways Disease in Post-Acute SARS-CoV-2 Infection . Radiology . 2022 ; 304 ( 1 ): 185 – 92 . Epub 2022/03/16. doi: 10.1148/radiol.212170 . PubMed PMID: 35289657; PubMed Central PMCID: PMCPMC9270680 . 35289657 26. Huang R , Zhu J , Zhou J , Shang Y , Lin X , Gong S , et al . Inspiratory and Expiratory Chest High-resolution CT: Small-airway Disease Evaluation in Patients with COVID-19 . Curr Med Imaging . 2021 ; 17 ( 11 ): 1299 – 307 . Epub 2021/01/14. doi: 10.2174/1573405617999210112194621 . PubMed PMID: 33438547 . 33438547 27. Inoue A , Johnson TF , White D , Cox CW , Hartman TE , Thorne JE , et al . Estimating the Clinical Impact of Photon-Counting-Detector CT in Diagnosing Usual Interstitial Pneumonia . Invest Radiol . 2022 . Epub 2022/06/16. doi: 10.1097/RLI.0000000000000888 . PubMed PMID: 35703439 . 28. Remy-Jardin M , Remy J , Giraud F , Wattinne L , Gosselin B . Computed tomography assessment of ground-glass opacity: semiology and significance . J Thorac Imaging . 1993 ; 8 ( 4 ): 249 – 64 . Epub 1993/01/01. doi: 10.1097/00005382-199323000-00001 . PubMed PMID: 8246323 . 8246323 29. Graafen D , Emrich T , Halfmann MC , Mildenberger P , Duber C , Yang Y , et al . Dose Reduction and Image Quality in Photon-counting Detector High-resolution Computed Tomography of the Chest: Routine Clinical Data . J Thorac Imaging . 2022 . Epub 2022/06/15. doi: 10.1097/RTI.0000000000000661 . PubMed PMID: 35699680 . 30. Hata A , Schiebler ML , Lynch DA , Hatabu H . Interstitial Lung Abnormalities: State of the Art . Radiology . 2021 ; 301 ( 1 ): 19 – 34 . Epub 2021/08/11. doi: 10.1148/radiol.2021204367 . PubMed PMID: 34374589; PubMed Central PMCID: PMCPMC8487219 . 34374589 31. Si-Mohamed SA , Greffier J , Miailhes J , Boccalini S , Rodesch PA , Vuillod A , et al . Comparison of image quality between spectral photon-counting CT and dual-layer CT for the evaluation of lung nodules: a phantom study . Eur Radiol . 2022 ; 32 ( 1 ): 524 – 32 . Epub 2021/06/30. doi: 10.1007/s00330-021-08103-5 . PubMed PMID: 34185147 . 34185147 32. Si-Mohamed S , Boccalini S , Rodesch PA , Dessouky R , Lahoud E , Broussaud T , et al . Feasibility of lung imaging with a large field-of-view spectral photon-counting CT system . Diagn Interv Imaging . 2021 ; 102 ( 5 ): 305 – 12 . Epub 2021/02/22. doi: 10.1016/j.diii.2021.01.001 . PubMed PMID: 33610503 . 33610503 33. Jungblut L , Euler A , von Spiczak J , Sartoretti T , Mergen V , Englmaier V , et al . Potential of Photon-Counting Detector CT for Radiation Dose Reduction for the Assessment of Interstitial Lung Disease in Patients With Systemic Sclerosis . Invest Radiol . 2022 . Epub 2022/06/01. doi: 10.1097/RLI.0000000000000895 . PubMed PMID: 35640003 . 34. Ruggiero V , Aquino RP , Del Gaudio P , Campiglia P , Russo P . Post-COVID Syndrome: The Research Progress in the Treatment of Pulmonary sequelae after COVID-19 Infection . Pharmaceutics . 2022 ; 14 ( 6 ). Epub 2022/06/25. doi: 10.3390/pharmaceutics14061135 . PubMed PMID: 35745708; PubMed Central PMCID: PMCPMC9229559 .	36445225	PMC9718279	NO-CC CODE	2022-12-05 23:15:06	no	Radiology. 2022 Nov 29;:222087	utf-8	Radiology	2,022	10.1148/radiol.222087	oa_other
==== Front J Public Health Manag Pract J Public Health Manag Pract JPUMP Journal of Public Health Management and Practice 1078-4659 1550-5022 Wolters Kluwer Health, Inc. 36448764 10.1097/PHH.0000000000001680 jpump2901p108 3 News From NACCHO Interruptions in Routine Blood Pressure Screening Services Among Local Health Departments During the COVID-19 Pandemic Cunningham Margaret C. MPH, RN [email protected] Royster Jordan MSc [email protected] McCall Timothy C. PhD [email protected] National Association of County and City Health Officials, Washington, District of Columbia (Ms Cunningham, Mr Royster, and Dr McCall) and Clinical Research and Leadership, The George Washington University School of Medicine and Health Sciences, Washington, District of Columbia (Dr McCall). Correspondence: Margaret C. Cunningham, MPH, RN, National Association of County and City Health Officials, 1201 Eye St, Washington, DC 20005 ([email protected]). 1 2023 16 11 2022 16 11 2022 29 1 Data Informs Health Equity 108111 © 2023 Wolters Kluwer Health, Inc. All rights reserved. 2023 This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. ==== Body pmcHypertension is a leading risk factor for cardiovascular disease and a significant contributor to preventable morbidity, mortality, and health care delivery system costs in the United States. According to the National Center for Health Statistics (NCHS), hypertension was a primary or contributing cause to more than 670 000 deaths nationwide in 2020.1 While the overall prevalence of hypertension among US adults is estimated at 45.4%, significant disparities exist among geographic regions, racial and ethnic groups, and socioeconomic status levels.2,3 Serious sequelae of uncontrolled hypertension, including heart disease and stroke, are also inequitably distributed nationwide.4 Like many other chronic conditions, hypertension-associated morbidity and mortality can be mitigated at multiple levels of prevention. Chronic disease prevention is widely recognized as a foundational public health service informing the activities of local health departments (LHDs).5 The role of LHDs in chronic disease prevention frequently includes access to secondary prevention via screenings. According to nationally representative survey data collected by the National Association of County and City Health Officials (NACCHO), blood pressure (BP) screening is a common LHD function, endorsed by 54% of LHDs nationwide in 2016 and 59% in 2019.6 For most LHDs, routine services—including BP screening and other chronic disease prevention work—were interrupted during the initial public health emergency response to the COVID-19 pandemic. Reasons for routine service interruptions included facility closures, local stay-at-home orders, and the reassignment of LHD staff from their usual duties to support the pandemic response workflow. These changes to LHD service provision and staffing have occurred against the backdrop of an evolving and increasingly inequitable chronic disease risk environment. Since the spring of 2020, lifestyle disruptions have contributed to changes in health behaviors for many US individuals and communities. Individual-level risk factors associated with altered physical activity, diet, and substance use patterns are well documented7 and have often been compounded by economic hardships, caregiving duties, grief, and other stressors. Health care system access and utilization patterns have changed as well, including delayed preventive care, interrupted medication access, loss of insurance coverage, and other barriers to clinical care have been widespread.8 Recognizing that the burden of these and other risk factors will affect certain populations more severely than others, some LHDs launched or prioritized initiatives targeted to people with chronic conditions during service disruptions. Methods NACCHO's Forces of Change (FoC) survey is a periodic survey fielded to a sample of LHDs across the United States to assess changes in LHD capacity driven by trends in public health. The 2020 FoC survey focused on impacts to LHD infrastructure in the context of the COVID-19 pandemic. The survey was administered electronically through Qualtrics9 and was distributed to a population of 2392 LHDs. All LHDs in the study population received a common core set of questions from November 2020 to March 2021 (n = 583; response rate = 24%). In addition to the core questionnaire sent to the population of LHDs, a stratified random sample of 905 LHDs was invited to complete a module questionnaire on topics such as pandemic preparedness, recovery, and equity, with strata defined by the size of the population served by the LHD (n = 237; response rate = 26%). Statistics were computed using poststratification weights to adjust for oversampling and nonresponses by jurisdiction size proportionate to their distribution in the United States. This analysis provides national estimates for all LHDs in the United States by using weighted survey results based on the size of population served. Results A majority of LHDs responding to the 2020 FoC survey reported a reduction in 1 or more routine public health services during the COVID-19 emergency response.10 Of the LHDs that reported having provided BP screening services at any time during calendar year 2019, more than two-thirds (67.2%) indicated that these services were reduced during the pandemic response, and only 1.4% reported that they had expanded screening services. In contrast, in NACCHO's 2019 National Profile of Local Health Departments survey, only 7.3% LHDs had reported reductions in BP screening between calendar years 2017 and 2018, while 13.8% had expanded these services.6 Reductions in BP screening services were reported more frequently by LHDs serving large populations (populations >500 000; 85.9%) than by those serving small (<50 000) and medium (50 000-500 000) populations (67.2% and 63.9%, respectively). State-governed LHDs (36.2%) were less likely to report BP screening service reductions than LHDs with shared or local governance structure (89.6% and 75.4%, respectively) (Table). TABLE BP Screening Service Changes, Reassignment of Chronic Disease Prevention Staff, and LHDs That Prioritized Chronic Care Initiatives as Reported in Forces of Change 2020 LHDs That Provided BP Screening Services at Any Time in Calendar Year 2019 (n = 202) LHDs That Reduced BP Screening Services During the COVID-19 Pandemic Responsea (n = 116) LHDs That Reassigned Chronic Disease Prevention During Pandemic Responseb (n = 513) LHDs Prioritizing Initiatives for People With Chronic Conditions During COVID-19 Service Disruptionsc (n = 225) All LHDs 57.6% 67.2% 43.1% 19.9% Population size <50 000 67.4% 67.2% 32.7% 20.1% 50 000-499 999 41.1% 63.9% 55.5% 19.3% 500 000+ 47.2% 85.9% 81.9% 22.1% Type of governance State 49.1% 36.2% 37.6% 23.2% Local 59.8% 75.4% 43.8% 18.0% Shared 72.6% 89.6% 55.9% 36.4% Census region Northeast 46.9% 71.4% 25.8% 4.2% Midwest 67.7% 76.1% 49.1% 20.9% South 55.2% 54.6% 40.6% 23.3% West 43.3% 71.8% 65.1% 33.9% Abbreviations: BP blood pressure; LHD, local health department. aLHDs selecting response option “Reduced services” to question: “Between the start of our COVID-19 response and today, my LHD....” This response option was available only to LHDs reporting that they had provided BP screening at any time in calendar year 2019. bLHDs selecting response option “Chronic disease” to question: “From which of the following program areas have staff been reassigned from their regular duties to perform duties in support of your LHD's COVID-19 response?” cLHDs selecting response option “Clinical care for people with chronic conditions during service disruptions” to question: “Please indicate if your LHD has prioritized or developed targeted initiatives to address these issues at any time between the start of your COVID-19 response and today.” A significant factor in the reduction in services may have been the reassignment of staff. In 2020, 4 in 5 (82%) LHDs reassigned staff from normal duties to support COVID-19 response activities. More than 2 in 5 LHDs (43%) reported reassigning chronic disease staff to perform duties in support to the COVID-19 response. This differed by population size served (Table): 81.9% of LHDs serving large populations reassigned chronic disease staff, compared with 32.7% of LHDs serving small populations. In response to a question about targeted initiatives to address population health issues exacerbated by the pandemic, only 1 in 5 LHDs (20%) endorsed “clinical care for people with chronic conditions during service disruptions” as a priority. Commitment to targeted clinical care initiatives did not vary greatly by LHDs' population size, type of governance, or Census region (Table). Discussion A majority of LHDs reduced BP screening services during the COVID-19 pandemic response. Higher rates of pandemic-associated service reductions were observed among LHDs serving large populations and LHDs with state or shared governance structure. Most LHDs did not name people with chronic conditions as a priority issue to address via targeted initiatives during the 2020-2021 pandemic response. This may have been due to competing public health priorities, limited funding, or lack of staff capacity or because other stakeholders in the community were perceived as filling this gap. BP screening is only one component of LHD chronic disease prevention programming, which may also include primary prevention activities (eg, health promotion and community education activities, advocacy for healthier built environments) and facilitating access to medical care for community members living with chronic conditions. Given the inequitable impact that the COVID-19 pandemic has made on cardiovascular health risks and outcomes in the United States, an equity-focused approach is recommended for LHDs looking to continue, relaunch, or start new BP screening programs as part of their chronic disease prevention strategies. LHDs should utilize existing community-level data to assess the distribution of risk factors, protective factors, and pertinent outcomes. Geospatial data analysis tools such as the Centers for Disease Control and Prevention's PLACES project11 and the Interactive Atlas of Heart Disease and Stroke12 can help LHDs as they plan where, when, and how to deliver screening services. These data should be supplemented with both quantitative and qualitative data from key stakeholders, including community members, medical providers, local government, and community-based organizations. LHDs should also recognize the syndemic nature of behavioral health issues and chronic disease, forming interdisciplinary partnerships and working as an integrated health promotion and care delivery system. LHDs are uniquely positioned to address the chronic disease health debt introduced by the COVID-19 pandemic due to their knowledge of local contexts, relationships with key stakeholders, and capacity to provide or facilitate access to low-barrier preventive care to the community. Limitations The present analysis is not without limitations. Survey data were self-reported and were not verified independently by NACCHO. In addition, NACCHO's FoC is a cross-sectional survey reflecting a point in time; LHDs may have revised their priorities and established programming since completing this survey. Finally, nonresponse bias may impact the results presented here, due to a low response rate during the COVID-19 public health emergency response. Directions for future work As LHDs enact pandemic recovery plans, many will resume or initiate BP screening and other chronic disease screening programs. Whether service provision will return to pre-2020 frequencies will depend on local priorities, as well as LHD funding levels and workforce capacity. Analysts of future LHD surveys should examine the associations of service provision with workforce and budget indicators, with community-level indicators of the social determinants, and with indicators of relevant population health outcomes. The authors declare no conflicts of interest. ==== Refs References 1. Centers for Disease Control and Prevention, National Center for Health Statistics. About Multiple Cause of Death, 1999-2020. CDC WONDER. Atlanta, GA: Centers for Disease Control and Prevention; 2022. https://wonder.cdc.gov/mcd-icd10.html. Accessed September 23, 2022. 2. Ostchega Y Fryar CD Nwankwo T Nguyen DT . Hypertension Prevalence Among Adults Aged 18 and Over: United States, 2017-2018. Hyattsville, MD: National Center for Health Statistics; 2020. NCHS Data Brief No. 364. https://www.cdc.gov/nchs/products/databriefs/db364.htm. Accessed September 26, 2022. 3. Commodore-Mensah Y Turkson-Ocran RA Foti K Cooper LA Himmelfarb CD . Associations between social determinants and hypertension, stage 2 hypertension, and controlled blood pressure among men and women in the United States. Am J Hypertens. 2021;34 :707–717.33428705 4. Wadhera RK Figueroa JF Rodriguez F Racial and ethnic disparities in heart and cerebrovascular disease deaths during the COVID-19 pandemic in the United States. Circulation. 2022;143 (24 ):2346–2354. 5. The Public Health National Center for Innovations. Foundational Public Health Services [issue brief]. https://phnci.org/uploads/resource-files/FPHS-Factsheet-2022.pdf. Published 2022. Accessed September 28, 2022. 6. NACCHO. 2019 National Profile of Local Health Departments. https://www.naccho.org/uploads/downloadable-resources/Programs/Public-Health-Infrastructure/NACCHO_2019_Profile_final.pdf. Published 2020. Accessed September 28, 2022. 7. Hacker KA Briss PA Richardson L Wright J Petersen R . COVID-19 and chronic disease: the impact now and in the future. Prev Chronic Dis. 2021;18 :E62.34138696 8. Ruth L Alongi J Robitscher J . Confronting the health debt: the impact of COVID-19 on chronic disease prevention and management. Health Affairs Blog. Posted 2021. https://www.healthaffairs.org/do/10.1377/forefront.20210914.220940. Accessed September 28, 2022. 9. Qualtrics [Computer software]. Version May 2022. Provo, UT: Qualtrics; 2005. 10. NACCHO. Forces of Change 2020: the COVID-19 edition. https://www.naccho.org/uploads/downloadable-resources/2020-Forces-of-Change-The-COVID-19-Edition.pdf. Published 2022. Accessed September 28, 2022. 11. Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Division of Population Health. PLACES: local data for better health. https://www.cdc.gov/places/index.html. Published 2022. Accessed October 3, 2022. 12. Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Division for Heart Disease and Stroke Prevention. Interactive Atlas of Heart Disease and Stroke. https://www.cdc.gov/dhdsp/maps/atlas/index.htm. Published 2020. Accessed October 3, 2022.	36448764	PMC9718291	NO-CC CODE	2022-12-07 23:21:56	no	J Public Health Manag Pract. 2023 Jan 16; 29(1):108-111	utf-8	J Public Health Manag Pract	2,022	10.1097/PHH.0000000000001680	oa_other
==== Front 9432230 8745 Cancer Gene Ther Cancer Gene Ther Cancer gene therapy 0929-1903 1476-5500 32632271 10.1038/s41417-020-0189-4 nihpa1606631 Article Oncolytic Virus Promotes Tumor-Reactive Infiltrating Lymphocytes for Adoptive Cell Therapy Feist Mathilde 12 Zhu Zhi 1 Dai Enyong 13 Ma Congrong 1 Liu Zuqiang 12 Giehl Esther 12 Ravindranathan Roshni 1 Kowalsky Stacy J. 1 Obermajer Natasa 1 Kammula Udai S. 1 Lee Andrew J. H. 1 Lotze Michael T. 1 Guo Zong Sheng http://orcid.org/0000-0002-4624-9907 1* Bartlett David L. http://orcid.org/0000-0003-3099-2613 14* 1 Departments of Surgery, University of Pittsburgh School of Medicine, and UPMC Hillman Cancer Center, Pittsburgh, PA, USA 2 Department of Surgery, CCM/CVK, Charité - Universitaetsmedizin Berlin, Berlin, Germany 3 Department of Oncology and Hematology, The Third Hospital of Jilin University, Changchun, Jilin, China 4 Current Address: Allegheny Health Network - Cancer Institute‎, Pittsburgh, PA 15212 * Corresponding authors: Dr. David L Bartlett, MD, [email protected], Dr. Zong Sheng Guo, PhD, [email protected] 26 6 2020 2 2021 07 7 2020 02 12 2022 28 1-2 98111 http://www.nature.com/authors/editorial_policies/license.html#terms Users may view, print, copy, and download text and data-mine the content in such documents, for the purposes of academic research, subject always to the full Conditions of use:http://www.nature.com/authors/editorial_policies/license.html#terms Adoptive cell therapy (ACT) using tumor-specific tumor-infiltrating lymphocytes (TILs) has demonstrated success in patients where tumor-antigen specific TILs can be harvested from the tumor, expanded, and re-infused in combination with a preparatory regimen and IL-2. One major issue for non-immunogenic tumors has been that the isolated TILs lack tumor specificity and thus possess limited in vivo therapeutic function. An oncolytic virus (OV) mediates an immunogenic cell death for cancer cells, leading to elicitation and dramatic enhancement of tumor-specific TILs. We hypothesized that the tumor-specific TILs elicited and promoted by an OV would be a great source for ACT for solid cancer. In this study, we show that a local injection of oncolytic poxvirus in MC38 tumor with low immunogenicity in C57BL/6 mice, led to elicitation and accumulation of tumor-specific TILs in the tumor tissue. Our analyses indicated that IL-2-armed OV-elicited TILs contain lower quantities of exhausted PD-1hiTim-3+ CD8+ T cells and regulatory T cells. The isolated TILs from IL-2-expressing OV-treated tumor tissue retained high tumor-specificity after expansion ex vivo. These TILs resulted in significant tumor regression and improved survival after adoptive transfer in mice with established MC38 tumor. Our study showcases the feasibility of using an OV to induce tumor-reactive TILs that can be expanded for ACT. oncolytic virus vaccinia virus IL-2 antitumor adaptive immunity tumor-specific T cells expansion ex vivo adoptive T cell therapy ==== Body pmcINTRODUCTION Immunotherapy for cancer has become a reality in the last few years, with the approval by the FDA of a therapeutic cancer vaccine for advanced prostate cancer (Provenge), an oncolytic virus T-VEC for advanced melanoma, two types of CAR T cells for leukemia, and multiple antibody-based drugs of immune checkpoint blockade for a variety of cancers. Immunotherapy is rapidly evolving and treating cancer by activating the patient’s immune system presents an attractive therapeutic strategy [1]. One promising strategy is adoptive cell therapy (ACT) using the patient’s own TILs – a form of personalized cancer immunotherapy. This approach consists of excising a tumor, then expanding the lymphocytes from the tumor ex vivo. The patient is then treated with a non-myeloablative preparative chemotherapy regimen followed by TIL infusion and systemic IL-2 [2]. This approach has been successful in patients with cutaneous melanoma where the tumor mutational load (TMB) is high and tumor reactive lymphocytes are abundant in the tumor. Objective response rates (ORR) were observed ranging from 49% to 72%, with long term durable and potentially curative complete response rates of up to 25% [3], ORR was 42% with an attenuated IL-2 dose in another study [4]. However, most human solid tumors are non-immunogenic and tumor-reactive TIL cannot be isolated or expanded for ACT. Recently, however, ACT of autologous TILs, isolated and separated for tumor-reactive TILs (which made up a small percentage of total TILs), was shown to induce objective tumor regression in patients with metastatic uveal melanoma, a low mutational burden, non-immunogenic tumor [5]. In this trial, there was a strong association between the anti-tumor reactivity of the infused TILs and objective clinical response, suggesting that mechanisms to improve recovery of tumor reactive TILs could lead to more effective therapy for other solid tumors. There are at least two major hurdles in successful ACT for cancers other than melanoma. First, the source of TILs is a major issue as the majority of human solid cancers are considered poorly immunogenic and very few TILs exist in the tumor tissues [6]. Even if it is possible to isolate TILs in non-immunogenic cancers such as uveal melanoma, most TILs isolated were not tumor-reactive [5]. Second, other than TILs from melanoma patients, there have not been many successful studies of isolation and expansion of tumor-specific TILs. The standard operating procedures for isolation and expansion of tumor-specific TILs ex vivo from other types of solid cancers have not been established. This was the case even for relatively high immunogenic renal cell carcinomas until recently [7]. Local immunotherapy, including oncolytic virotherapy, may modulate the tumor immune contexture and convert tumors from immunologically “cold” to “hot” [8–10]. We hypothesized that cytokine armed oncolytic VV could create a local pro-inflammatory environment in otherwise poorly immunogenic tumors leading to tumor reactive T-cells that could be harvested for ACT of cancer. Oncolytic viruses (OVs) selectively infect and/or replicate in cancer cells in vivo leaving normal cells unharmed. The immunogenic cell death induced by VV exposes a natural repertoire of tumor-associated antigens in conjunction with danger signals: damage associated molecular pattern molecules (DAMPs) and virus-derived pathogen-associated molecular pattern molecules (PAMPs), and inflammatory cytokines to reverse the tumor induced immunosuppression and elicit anti-tumor cellular immunity [11, 12]. OVs have been shown in both pre-clinical studies [13–16], and clinical trials with T-VEC and Pexa-Vec [9, 17, 18] to induce potent adaptive antitumor immunity contributing to the overall efficacy of the therapy. To further improve the antitumor immunity, investigators have engineered various oncolytic VV to express tumor antigens, T-cell co-stimulatory molecules and inflammatory chemokines and cytokines, and they have been studied for their efficacy and safety in preclinical tumor models [19–22]. In the current study, we utilized relatively low immunogenic MC38 murine colon tumor, which displays low levels of inflammatory infiltrate and tumor-reactive TILs, as a model for study. We present for the first time that oncolytic VVs can induce local tumor-specific TILs, and these TILs can be harvested, expanded ex vivo, and utilized for ACT to achieve a therapeutic response in a tumor model with low-immunogenicity. We envision that this approach using OV to induce tumor reactive TILs in low- or non-immunogenic tumors could be translated to cancer patients, and such ex vivo expanded TILs could be used for ACT for solid tumors with the potential for long term cure. MATERIALS AND METHODS Mice and cell lines. Female C57BL/6J (B6) mice, age 5–6 weeks old, were obtained from The Jackson Laboratory (Bar Harbor, ME) and housed in specific pathogen-free conditions in the University of Pittsburgh Animal Facility. All animal studies were approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh. Murine colon cancer cell line B16 was originally obtained from American Type Culture Collection (Manassas, VA). Mouse colon cancer cell line MC38-luc was described previously [23]. All cell lines were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, and 1 penicillin/streptomycin (Invitrogen, Carlsbad, CA) in 37°C, 5% CO2 incubator. All cell lines were tested for mycoplasma once every three months or so, to ensure they were free of the contamination. Viruses. WR strain-derived recombinant oncolytic VVs, vvDD, vvDD-CCL5 and vvDD-CXC11, [24–26], vvDD-IL15-Rα [27] and vvDD-IL2 expressing a membrane-bound IL-2 [28], have been constructed and studied. These VVs were amplified in HeLa cells and then purified as previously described [23, 26]. Viral titers were determined by plaque assays in CV-1 cells. For intratumoral injection of the viruses, a dose of 1.0e8 pfu per mouse was used unless indicated otherwise. Rodent Tumor models and virus treatments. For subcutaneous (s. c.) tumor model, B6 mice were subcutaneously inoculated with 5.0 × 105 MC38 colon cancer cells. When the tumors reached the size ~ 5 × 5 mm, vvDD, vvDD-IL2, vvDD-CCL5, vvDD-IL15Rα, vvDD-CXC11, or PBS was intratumorally injected at 1.0e8 pfu/tumor. In some experiments human IL-2 has been injected intratumorally at 1.0 × 106 IU per tumor (Prometheus, San Diego, CA). The primary tumor size was measured using an electric caliper in two perpendicular diameters followed by measurement every other day. Ten days post virus treatment, tumor tissues or spleens were harvest and then single cell suspension was made for T cell separation and further analysis. For peritoneal (i.p.) tumor models, mice were intraperitoneally inoculated with 5.0 × 105 MC38-luc cancer cells and randomized according to tumor size based on live animal IVIS imaging 7 days post tumor cell injection using a Xenogen IVIS 200 Optical In Vivo Imaging System (Caliber Life Sciences, Hoptikon, MA). Generation of tumor-reactive TILs for adoptive cell transfer. B6 mice were s.c. inoculated with 5.0 × 105 MC38 cancer cells at the right hind frank. When the tumor area reached 5 × 5 mm, vvDD-IL2 (1.0e8 pfu per tumor) was intratumorally injected. 10 days later tumors were collected, cut into pieces and incubated at 37°C in digestion buffer (Miltenyi Biotec, San Diego, CA) before being mashed over a 100 μM tissue strainer. Lysis of red blood cells was performed using ACK Lysing buffer (Thermo Fisher Scientific, Waltham, MA). For Leukocyte separation, Percoll gradient centrifuge procedure adapted from the protocol by Liu et al. [29] was used. Leucocytes were collected at the interface between 40% and 80% discontinuous Percoll gradient followed by magnetic separation (CD90.2 beads, Miltenyi Biotec). T cells have been cultured in 24-well plates in a concentration with 1.0 × 106 cells per well in RPMI complete media with 30 IU/ml IL-2 (MiltenyiBiotec) and 5 ng/ml IL-7 (BioLegend, San Diego, CA) over 4 days. As control T cells, spleens from non-tumor bearing untreated mouse have been harvested and proceeded to single cell suspension followed by magnetic separation (CD90.2 beads). The control T cells have been cultured under the same conditions as the virus induced T cells. Prior to adoptive cell transfer, T cells have been analyzed for tumor specificity in a co-culture assay. T cells (2.0 × 104 per well, 96-well plate) were either left unstimulated (medium) or challenged with γ-irradiated MC38 tumor cells (2.0 × 104 per well) or irrelevant target cells as γ-irradiated B16 tumor cells (2.0 × 104 per well) or naïve splenocytes (2.0 × 104 per well) from non-tumor-bearing B6 mice in duplicate for 24 h. The plate setup has been used for IFN-γ ELISPOT assay or flow cytometry analysis as described below. Tumor model, adoptive cell transfer and cytokine administration. B6 mice were intraperitoneally inoculated with 5.0 × 105 MC38-luc cancer cells and divided into required groups according to tumor growth condition based on live animal IVIS imaging 7 days post tumor cell injection. Grouped mice received 5.0 × 106 TILs that were isolated from vvDD-IL2-treated tumors and then expanded ex vivo, naïve T cells from control mice or just PBS. All treated mice also received 5 Gy of sublethal irradiation according to clinical protocols prior to cell transfer and exogenous cytokine support of IL-2 (100,000 IU/mouse, i.p.) for 3 days post transfer every 12 h (Prometheus). Flow cytometry. Collected tumor tissues were minced and incubated in RPMI 1640 medium containing 2% FBS, 1mg/mL collagenase IV (Sigma: #C5138), 0.1mg hyaluronidase (Sigma: #H6254), and 200U DNase I (Sigma: #D5025) at 37°C for 1 hours to make single cells. In vitro virus-infected cells or single cells from tumor tissues were blocked with α-CD16/32 Ab (clone 93, eBioscience: #14-0161-85; 1:1000) and then cells were stained with 100 μL Zombie Aqua Fixable Viability Kit cell dye (BioLegend, San Diego, CA, USA) at a dilution of 1:1000 and left in room temperature for 15 min in the dark. Cells were stained in 100 μL total stain volume (50 μL BV stain buffer, 50 μl 2% FBS) with antibody at a dilution of 1:200 for 30 min on ice in the dark. The sources of antibodies are listed in Supplementary Table 1. The intracellular staining kit for Foxp3 and IFN-γ was purchased from BioLegend. Cells were then washed once with FACS buffer and fixed in 1% paraformaldehyde (EK Industries, Joliet, IL, USA) before being stored at 4°C overnight and acquired the next day on a BD LSR Fortessa II analyzer (BD Biosciences, San Jose, CA, USA). Data were analyzed using flowJo cytometer software. The enzyme-linked immunospot (ELISpot) assay. Collected tumor tissues were cut into pieces and incubated at 37°C in digestion buffer (Miltenyi Biotec, San Diego, CA) before being mashed over a 100 μM tissue strainer. Lysis of red blood cell was performed using ACK Lysing buffer (Thermo Fisher Scientific, Waltham, MA) and single cell suspension was proceeded by straining cell suspension over 40 μM filter. Isolation of CD8+ TILs was performed using negative α-mouse CD8 microbeads isolation protocol (Miltenyi Biotec, San Diego, CA). Ninety-six well plates (MAHAS4510, Millipore, Burlington, MA) were coated with anti-mouse IFN-γ mAb 15mg/ml (clone AN18, Mabtech Inc., Cincinnati, OH). T cells (2.0 × 104 per well) were either left unstimulated (medium) or challenged with γ-irradiated MC38 tumor cells (2.0 × 104 per well in 96-well plate) or irrelevant target cells as γ-irradiated B16 tumor cells (2.0 × 104 per well) or naïve splenocytes (2.0 × 104 per well) from non-tumor-bearing B6 mouse in duplicate for 24 h. After appropriate washes, biotylated secondary antibody (clone R4–6A2-biotin, Mabtech, Inc) was added and incubated at room temperature for 2 h. Spot development followed using Vectastain Elite ABC and AEC peroxidase substrate kit (Vector Laboratories, Inc. Burlingame, CA). Number of IFN-γ spots were analyzed by ImmunoSpot™ (Cellular Technology, Ltd., Shaker Heights, OH). To determine MC38-reactive responses the average value of spots from control wells were substracted from the number of MC38 challenged wells. Immunofluorescence staining. Resected tumors were fixed for 2 h in 2% Paraformaldehyde and incubated in 30% sucrose overnight. Sections were cut (5 μm) and stained with combined primary antibodies CD3 Alexa 488 (100212, BioLegend), CD4 Alexa 594 (100446, BioLegend) and CD8 Alexa 647 (100727, BioLegend) and nuclei were labeled with Hoechst dye (bis benzimide, Sigma B-2283–1 mg/100 ml in dH20). Images were acquired digitally from 9 fields under each condition. Density of positive cells was evaluated by automated image analysis using Nikon Elements (Nikon Instruments Inc, Melville, NY). Percentage of CD3+ T cells, CD3+CD4+, CD3+CD8+ T cells per area has been calculated by number of cells positive for the antibody versus the total number of cells. Long-term survival of mice. The health and survival of treated mice was closely monitored. All mice bearing subcutaneous or peritoneal tumors were monitored via caliper measurements for changes in tumor size or abdominal girth. Mice were dead naturally due to the disease or sacrificed when their subcutaneous tumor size exceeded 20 mm in diameter or when abdominal girth exceeded 1.5× the original measurement. Statistics. For the majority of experiments, the numbers of mice per group was equal to, or bigger than 5. We have chosen these sample sizes to ensure adequate statistical power either by pilot studies or previous studies we had conducted. For tumor-bearing mice, right before the treatment, they were pooled and then randomly distributed into different groups. For most of the animal experiments, it is not blinded. In rare occasions, if the value of one individual is considered to be an outlier (the value of 2 times of s.d outside of the mean), we might have excluded the value from the presented data. Statistical analyses were performed using unpaired Student’s t test for two group comparison. For multiple group comparison One-way ANOVA were used where p value is adjusted for multiple test by Dunnett method (GraphPad Prism version 5). Animal survival is presented using Kaplan-Meier survival curves and compared by using log rank test (GraphPad Prism version 5). Value of p < 0.05 is considered to be statistically significant, and all p values were two sided. In the figures, the standard symbols were used: * p < 0.05; p < 0.01; * p < 0.001; ** p < 0.0001; and NS: not significant. RESULTS Oncolytic VVs elicit and attract high numbers of tumor-reactive TILs We engineered several vvDD-based oncolytic VVs expressing different cytokines/chemokines (vvDD-CCL5, vvDD-CXC11, vvDD-IL15Rα, vvDD-IL2) to enhance the anti-tumor immune response and to attract T cells into the tumor tissue. We first explored a variety of time points after injection and found that 10 days was the optimal time to harvest tumor reactive T-cells from the tumor microenvironment (data not shown). We studied TILs in implanted subcutaneous MC38 tumors 10 days after intratumoral treatment with either PBS, vvDD, vvDD-CCL5, vvDD-CXC11, vvDD-IL15Rα or vvDD-IL2 (Fig. 1A). CD8+ TILs from these tumors were isolated and analyzed using an IFN-γ ELISpot assay against γ-irradiated MC38 tumor cells. From control tumors injected with PBS, there were no tumor-reactive T cells found in the tumor. In contrast, each of the virus-treated tumors demonstrated a significant increase in tumor-reactive CD8+ TILs over control (p < 0.05). vvDD-IL2-treated tumors had high frequencies of tumor-reactive CD8+ TILs compared to control (Fig. 1A). Also, vvDD-IL2 yielded one of the most promising therapeutic results in a separate study with MC38 colon tumor models [28]. Thus, we have picked vvDD-IL2 for the rest of our study here. To investigate whether the OV-induced CD8+ TILs were specific for the injected tumor, the experiment was repeated with the murine tumor B16 and naïve splenocytes as additional control targets. Subcutaneous MC38 tumors were injected with vvDD-IL2 or PBS, and 10 days later the TILs were harvested. CD8+ T cells from treated tumors were again analyzed using an IFN-γ ELISPOT assay against γ-irradiated MC38 tumor cells, B16 tumor cells, or naïve splenocytes from a non-tumor-bearing B6 mouse (Fig. 1B). vvDD-IL2-treated tumors again demonstrated a significant increase in MC38 tumor reactive CD8+ T cells. The vvDD-IL2-induced CD8+ TILs were highly reactive and specific against MC38 tumor cells compared to the irrelevant target cells. In summary, vvDD-IL2 injection into MC38 tumors induced high levels of MC38 specific TILs. We repeated the experiment again with vvDD and IL-2 protein injection as additional controls, using an effective dose based on a prior study [30], and examined both systemic and intratumoral T-cells for tumor reactivity. MC38 s.c. tumor-bearing mice were treated intratumorally with PBS, IL-2, vvDD or vvDD-IL2. Ten days later, both splenocytes and the CD8+ fraction of TILs were isolated and analyzed by IFN-γ ELISPOT assays. Both viruses induced greater frequency of tumor reactive T cells in the spleen compared to controls as shown by IFN-γ release (Fig. 1C). Further, each of the viruses induced a greater frequency of tumor-reactive TILs compared with PBS, and the vvDD-IL-2 was superior to vvDD or IL-2 alone in promoting recovery of antitumor CD8+ lymphocytes from the tumor (Fig. 1D). While vvDD-IL2 in the above studies induced a higher percentage of tumor-reactive CD8+ TILs amongst recovered CD8+ cells, we also hypothesized that the absolute numbers of CD8+ T cells were increased after viral infection. We determined the quantity of CD4+ and CD8+ T cells in the tumor at day 10 after treatment, using two methods of analyses. First, tumors were analyzed by immunofluorescence staining for total CD3+ infiltrating cells, CD3+CD8+ and CD3+CD4+ T cells (Fig. 2). vvDD-IL2 promoted significant increases of total CD3+ and CD3+CD8+ cell infiltration in the tumor tissue compared to PBS treatment. Oncolytic vvDD-IL2 treatment presented a trend towards increased CD3+CD4+ TILs (% CD3+CD4+ T cells: vvDD-IL2 vs PBS: p = 0.0588). Second, T cells were purified from tumor tissues and quantified as per gram of tumor tissue, and similar patterns were observed (Suppl. Fig. 1). Cumulatively, these results indicate that viral treatment, especially vvDD-IL2, elicits potent adaptive antitumor immunity leading to an increased number of CD3+CD8+ T cells in the tumor tissues, as well as an increased percentage of tumor-reactive CD3+CD8+ T cells in the tumor. OV-induced TILs profiling indicated a strong pro-antitumor immunity. We then analyzed the status of T cells in the tumor tissues by flow cytometry (Fig. 3). To start with, we looked at the splenocytes and set them as control for comparison (Fig. 3A). In the spleen, CTLA-4+CD8+ T cells were slightly reduced in both groups treated with OVs. Tim-3+CD8+ T cells were not changed, while PD-1hiTim-3+ CD8+ T cells were slightly reduced. The PD-1hiTim-3+ CD8+ T cells represent the highly exhausted cell population [31]. Interestingly, Treg (Foxp3+CD4+ T cells) cells were reduced in the mice treated with vvDD-IL2, but not in the group treated with vvDD. Then we analyzed the same set of markers in TILs (Fig. 3B). We observed increased CTLA-4+ or TIM-3+ CD8+ T cells in the virus-treated groups. Surprisingly, when we examined PD-1hiTim-3+ CD8+ T cells, considered to be highly exhausted cells, we saw a very exciting pattern of changes. The group treated with vvDD exhibited a reduction of this exhausted cells, and this reduction was much greater in the mice treated with vvDD-IL2 (p < 0.01). When we analyzed Treg cells, the same patterns of reduction were observed. During T cell activation, there is a dynamic transition from Tn to Tcm to Tem, and finally effector T cells. We therefore examined the CD8+ T cell subsets (Fig. 3C). First, there was a large percentage of Tn (naïve T cells) in the PBS group, while Tn was greatly reduced in the two groups treated with OVs. In the Tcm subset, there was an increase in the groups treated with OVs, and finally, there was a large increase of the Tem, from ~17% in PBS group to ~43% in the groups treated with OVs. When we analyzed the populations of 4–1BB+CD8+ T cells, there were clearly an enhancement in the OV-treated groups (p < 0.001, compared to PBS group). These results showed convincingly that OV activated and promoted differentiation of CD8+ T cells toward effector T cells. We have also conducted preliminary studies on the specificity of TILs recognizing certain antigens. We pulsed naïve splenocytes with H-2Kb-restricted epitope peptides, p15E604–611 (retrovirally-encoded tumor antigen) [32], B8R20–27 (dominant vaccinia virus antigen) [33], and β-galactosidase-derived epitope (DAPIYTNV), then incubated with TILs isolated from MC38-tumor-bearing mouse treated with vvDD-IL2 for ~12 h. The flow cytometry analysis for IFN-γ+ CD8+ T cells suggested that fractions of TILs were able to recognize p15E and B8R epitopes, respectively (Supplementary Figure 2). OV-induced TILs can be isolated from tumor tissues and expanded ex vivo As our eventual aim was to utilize these tumor-reactive TILs for therapeutic ACT, we investigated the ability to expand these TILs while retaining their tumor-specificity. First, we established protocols for TIL isolation with CD90.2 magnetic beads, and then for ex vivo culture and expansion. For induction of tumor-specific TILs, MC38 s. c. tumor-bearing mice were treated as before with intratumoral PBS or vvDD-IL2. Ten days later tumors were harvested and TILs from each tumor were cultured in the presence of IL-2 and IL-7 in RPMI complete media for 3 days. To test if the cultured TILs retained their tumor specificity, these T cells were tested for their tumor recognition using a co-culture assay including irradiated MC38 tumor cells and irrelevant target cells -γ-irradiated B16 tumor cells or naïve splenocytes from non-tumor-bearing B6 mice. After 24 h these T cells were analyzed for tumor specificity using IFN-γ ELISpot or 4–1BB expression by flow cytometry. According to the IFN-γ ELISpot assay, T cells expanded from vvDD-IL2 or PBS-treated tumors demonstrated increased MC-38 specific IFN-γ secretion compared to irrelevant target cells (B16 or naïve splenocytes) (Fig. 4A; Suppl. Fig. 3). To distinguish between CD8+ and CD4+ tumor-specific T cell response, flow cytometry results of 4–1BB expressing CD8+ and CD4+ T cells from each expanded TIL culture are summarized in Fig. 4 (B, C) and Suppl. Fig. 3. The expanded cultures from the vvDD-IL2-induced TILs demonstrated a higher percentage of tumor-specific CD8+ and CD4+ T cells when compared to PBS treatment [% CD8+ 4–1BB+ T cells: vvDD-IL2: 16.4 ± 11; PBS: 1.96 ± 1.6. % CD4+ 4–1BB+ T cells: vvDD-IL2: 6.3 ± 2.4; PBS: 2.4 ± 2.0]. ACT of OV-induced TILs led to therapeutic efficacy in a low-immunogenic colon tumor model We then explored the ability of the vvDD-IL2-induced TILs to be utilized for ACT for cancer treatment. The overall schema is shown in Fig. 5A. We harvested the TILs from MC38 colon cancer-bearing mice treated with vvDD-IL2 as described in Methods. As a control, T cells were isolated from splenocytes from naïve mice. These TILs and control T cells were expanded ex vivo for 4 days as described above and then portions of the cultures were characterized. Expanded TILs from vvDD-IL2 injected tumors had a higher specific reactivity against MC38 compared to naïve splenocytes expanded in an identical manner (Fig. 5, B & C). No IFN-γ was released when cultured control T cells were tested against MC38 cancer cells. We also tested the activation status of the vvDD-IL2-induced TILs using 4–1BB staining and found that 6% of the CD8+ T cells and 5% of CD4+ T cells were 4–1BB positive after co-culture with irradiated MC38 tumor cells, as analyzed by flow cytometry (Fig. 5, D & E; Suppl. Fig. 4). Very low levels of 4–1BB were found when these TILs were co-cultured with B16 tumor cells, naïve splenocytes or growth medium. These results indicated that CD4+ and CD8+ TILs specifically recognized irradiated MC38 cancer cells after expansion. We next examined the therapeutic efficacy of the ex vivo expanded TILs via ACT in a peritoneal MC38 colon cancer model (Fig. 6). Seven days after inoculation of 5.0e5 MC38-luc cells i.p. into C57BL/6J mice, the tumor growth was monitored by bioluminescence imaging using the Optical In Vivo Imaging System, and mice were randomly divided into groups with an equivalent range of tumor burden. Prior to ACT, tumor-bearing mice received 5 Gy of sublethal irradiation to mimic lymphodepletion similar to clinical protocols. The mice were injected i.p. with vvDD-IL2-induced and ex vivo expanded TILs, ex vivo expanded control T cells, or PBS saline. All treated mice received exogenous cytokine support with recombinant IL-2. We then followed the therapeutic response by live animal bioluminescence imaging to monitor the kinetics of tumor growth over time (Fig. 6A, B). It is clear that by day 17 post ACT, the group treated with ACT with T cells from vvDD-IL2 mice and radiation showed the best results (p < 0.05, compared to any other group). Mice treated with ACT from vvDD-IL2-induced TILs showed the least amounts of tumor burden and survived the longest when compared to PBS or 5Gy/IL-2 controls (p < 0.001) (Fig. 6C). If we compare the median survival of mice in different groups, the ACT with T cells from VV-IL2 group gave the very best result, with median survival time at 61 days (Fig. 6D; p < 0.01, compared to any other group). DISCUSSION ACT using autologous TILs represents a personalized cancer immunotherapy strategy, targeting shared and unique tumor antigens expressed by a patient’s tumor [2, 34]. ACT for cancer using patients’ own TILs has been a successful systemic treatment leading to the cure of a number of advanced melanoma patients, and a few other tumor types. The suppressive TME is addressed with a preparatory non-myeloablative chemotherapy regimen (cytoxan and fludarabine) then ex vivo expanded TILs are delivered intravenously followed by IL-2 to maintain T cell activity. A recent study in uveal melanoma demonstrated a strong association between the anti-tumor reactivity of the infused TILs and objective clinical response. Thus, isolation and delivery of TILs with greater anti-tumor reactivity may result in improved clinical responses. Discovering new methods to improve the recovery of tumor-reactive TILs could have a significant impact on the field of cancer immunotherapy and ultimately patient outcomes. Immune cell infiltration in the tumor impacts tumor progression and patient survival, and a strong lymphocyte infiltration has been reported to be associated with an antitumor response and improved clinical outcome in a variety of types of cancers, including colorectal cancer [35–39]. The majority of solid tumors, especially those of low tumor mutational burden (TMB), however, lack an inflammatory infiltrate. Cancers develop immune escape mechanisms to avoid detection by effector immune cells. This includes cell surface expression of immune system checkpoint ligands such as PD-L1 [40, 41]; secretion of soluble immunosuppressive factors such as transforming growth factor beta, vascular endothelial growth factor, interleukin-10, galectin-1, indoleamine 2,3-dehydrogenase [42–44]; down-regulation of major histocompatibility complex (MHC) class I expression; and overexpression of receptors such as C-X-C chemokine receptor type 4, basic fibroblast growth factor and epidermal growth factor [45, 46]. The immunosuppressive tumor microenvironment includes immunosuppressive macrophages, myeloid-derived suppressor cells and regulatory T cells interfering with an efficient anti-tumor T cell response. Inhibitory checkpoint molecules such as CTLA-4, PD-1, TIM-3, LAG3, are upregulated in chronically stimulated T cells, promoting T cell anergy. OVs overcome the immunosuppressive TME as they turn cold tumor ‘hot’ [8–10]. They infect and kill tumor cells in vivo and induce an inflammatory response that clears the virus and some non-infected tumor cells, despite the immunosuppressive TME. The virus is ultimately immunologically cleared from the tumor and tumor reactive T cells are generated. This local cytotoxic and anti-tumor immunologic activity has limited systemic activity. Successful systemic delivery of OVs with adequate tumor infection throughout the body for a meaningful effect on metastatic tumors, remains a challenge. One way to take advantage of local anti-tumor immunity for systemic therapy is to harvest the infected tumor tissues, isolate and expand the tumor-reactive T cells ex vivo, and adoptively transfer them back into the patient. In a clinical study done by Ribas et al., intratumoral injection of talimogene laherparepvec in melanoma patients resulted in an increase in infiltrating T cells independent of the T cell baseline [9]. In our study, we have demonstrated for the first time the feasibility of using these TILs for therapy. When the subsets of T cells in the TIL populations were analyzed, the quality of TILs was much improved in the mice treated with vvDD. They contained much higher percentages of Tcm and Tem cells. Moreover, vvDD-IL2 improved the quality even further, as it contained fewer exhausted PD-1hiTim-3+ CD8+ T cells and Treg cells. This would effectively expand the utility of both TILs and OV. This immunotherapy would be personalized for each patient’s tumor neoantigens, and in contrast to CAR-T cells would be polyclonal in terms of antigen recognition and immune cell type and unlikely to have reactivity against normal cells (off target toxicity). Our laboratories have extensive experience with both OVs and T cell-mediated immunotherapy [47, 48]. The anti-tumor immune response induced by oncolytic VV has been studied extensively. Previous depletion studies in animal models indicated the anti-tumor response and long-term survival are attributed to CD8+ T cells, and to a smaller extent CD4+ T cells, NK cells and neutrophils [8, 26, 49]. In the current study, we compared the parental oncolytic VV (vvDD) with those expressing various cytokines and chemokines (IL-2, CCL5, CXCL11 and IL-15) for their capacity to elicit and attract tumor-reactive TILs. Under the experimental conditions, we found that vvDD-IL2 performed the best, leading to the highest levels of tumor-reactive CD8+ TILs. The virus induced a MC38-specific T cell response and the effect was not reproducible by injection of the cytokine without the virus. The cytokine IL-2 was approved to treat metastatic melanoma and renal cancer in the 1990s. Since then different application strategies and combinations have been tested to overcome toxicity and enable a specific anti-tumor immune response [50, 51]. Although the transfer of virus-specific peripheral blood T cells to treat cancer patients have been done [52, 53], our current study has demonstrated, for the first time, a new therapeutic method-ACT using virally induced TILs isolated from solid tumors. The data from immunostaining confirmed that vvDD-IL2 attracts high numbers of total CD3+CD8+ T cells in the tumor with an increased percentage of specific tumor reactivity. These TILs isolated from tumors treated with vvDD-IL2 were able to be expanded in culture and maintain their activity and specificity. We observed high levels of IFN-γ release and 4–1BB expression from CD8+ and CD4+ TILs after 4 days of in vitro expansion when co-cultured with the same tumor cells. Adoptive transfer of these TILs in a model of peritoneal metastases reduced tumor burden and achieved better survival compared to the control groups. Thus, we have demonstrated the principle of therapeutic efficacy for OV-elicited tumor-reactive TILs in a murine tumor model. Virally induced tumor-reactive TILs have not previously been tested in ACT for therapeutic effectiveness. In the literature, most studies on TILs have been performed in patients by culturing small pieces of the tumor and expanding the outgrowing T cells [2]. It is well known that isolation of murine TILs is challenging and there exist only a few reports documenting the culturing of murine TILs [54, 55]. Most ACT studies in the murine system have been conducted using T cells derived from splenocytes and universally activated with super-antigens (such as anti-CD3 and CD28 antibodies). A number of investigators have explored the combination of an OV with ACT in cancer models, using the virus to improve trafficking of adoptively transferred cells [56–58], but not as a “pre-TIL” approach to improve the recovery of tumor-reactive T cells. Murine TILs have been isolated with magnetic beads or FACS sorting from tumor tissue-derived single cell suspensions [29, 54, 55]. We utilized magnetic bead sorting (CD90.2 antibody) for isolation, and murine TILs were able to be expanded ex vivo only when clearly separated from the tumor tissue. When assayed for their functionalities after ex vivo short-term expansion they retained their tumor specificity. In our studies MC38 tumor specific 4–1BB expression was enhanced in about 5–20 % of the CD8+ T cells and about 5–6 % of the CD4+ T cells. We are fully aware that the murine system for ex vivo TIL expansion has limitations and that the i.p. tumor serves only as a model to prove the therapeutic principle. In the future, we will explore the intravenous delivery of TILs to solid tumor models as the homing potential of such ex vivo expanded TILs needs to be analyzed as well. We would like to explore ways to induce, isolate, and expand more optimized TILs for ACT using OVs. For example, it is known that stem cell-like T cells may offer more therapeutic potential [59, 60]. Our comparison of cytokines and chemokines have not been exhaustive, and other agents may have similar or more potent activity as a “pre-TIL” approach. Once these types of questions are addressed, it will be time to translate this improved approach and technology from bench to bedside. In addition, one key issue in ACT is the trafficking of TILs intravenously delivered back to the tumor tissues where they exert their key functions. The study on trafficking of TILs back to tumor tissue is out of scope of this study as many other studies have been done to find the right answers [61]. In summary, we have demonstrated that IL-2 armed OV promotes T cell infiltration and enhances the population of tumor-reactive TILs in a murine colon cancer model. Our report presents a new therapeutic strategy to promote the generation and infiltration of tumor-reactive TILs in lowly or poorly immunogenic tumors, and the expansion of such TILs for ACT. The new strategy may be translated into a clinical application and may allow adoptive T cell therapy for an expanded group of cancer patients. Supplementary Material 1 2 3 4 Acknowledgments We thank the University Imaging Core Facility for technical assistance in immunostaining and imaging, the Hillman Cancer Center Biostatistics facility for consultation, and Prometheus (San Diego, CA) for human recombinant IL-2 used in the study. Funding This work was supported in part by David C. Koch Regional Cancer Therapy Center. MF (DFG research fellowship 1655/1-1) and EG were supported by fellowships from German Research Foundation. ED was supported by fellowships from the China Scholarship Council, China. This project has used the University of Pittsburgh shared facilities (Animal Facility, Genomics Research Core, Flow Cytometry) that are supported in part by the NIH grant award P30CA047904. Figure 1. Oncolytic VVs elicit tumor-specific antitumor CD8+ T cell response in the tumor tissue. Ten days after viral treatments, subcutaneous MC38 tumors and/or splenocytes were collected and single cell suspensions were made followed by magnetic separation. Then isolated CD8+ T cells or splenocytes were tested by IFN-γ ELISPOT assay. (A). CD8+ T cells isolated from tumors (n = 4 – 5 mice/group) were either left unstimulated or challenged with γ-irradiated MC38 tumor cells for 24 h. Results were shown as individual data points (number of spots in each well) and bars (means ± standard deviation) of IFN-γ+ CD8+ T cells from each mouse evaluated in triplicate. Data from one experiment representing 2 independent experiments are shown. (B). Tumor-specificity of the OV-induced CD8+ TILs (n = 7 mice/group). Data are from one experiment representative of 3 independent experiments. For multiple group comparison, One-way ANOVA was used. p < 0.001; **p < 0.0001. (C). MC38 tumor cell reactivity of splenocytes by IFN-γ ELISPOT assay. Splenocytes from treated mice were either left unstimulated or challenged with γ-irradiated MC38 tumor cells for 24 h. (D). MC38 tumor cell reactivity of isolated CD8+ TILs by IFN-γ ELISPOT assay. One experiment representative of 2 independent experiments is shown (n = 4–5 mice/group). Student’s t-test was used to analyze the statistical significance for data presented in panels A, C and D. p < 0.05; p < 0.01; p < 0.001; *p < 0.0001. Figure 2. vvDD-IL2 treatments increase the number of TILs. The tumor cell inoculation and tumor harvest were as described previously. Tumors were collected, fixed and stained for Hoechst (blue), CD3 (green), CD4 (red) and CD8 (white). A. Representative immunofluorescence image of one sample from each group (n = 5). For the representative panel, a single representative field was cropped from the original nine field acquisition. Scale bar 25 μm. B. Summary of the percentage of CD3+ T cells and CD3+CD4+ and CD3+CD8+ T cells per area. Student’s t-test was used to analyze the statistical significance. p < 0.001. Figure 3. Immune status in the TME and spleen post virus treatment. B6 mice were inoculated subcutaneously with 5.0e5 MC38 tumor cells. When tumors reached 5 × 5 mm (about day 9) they were treated intratumorally with PBS, vvDD, vvDD-IL-2 at a dose of 1.0e8 PFU. The mice were sacrificed 10 days post viral treatment and primary tumors and splenocytes were collected and analyzed by flow cytometry to determine the activation and immunosuppressive markers in (A) splenocytes, (B) TILs and (C) subsets of CD8+ T cells in the TILs. Tn for naïve T cells; Tcm for central memory T cells, and Tem for effector memory T cells. The percentages are over total CD8+ T cells, or over total CD4+ T cells for Treg cells. p < 0.05, p < 0.01, * p < 0.001; **** p < 0.0001. Figure 4. OV-induced TILs can be cultured and expanded ex vivo and retain their tumor specificity. (A). Tumor specificity of the expanded TILs. TILs from each individual mouse have been cultured for 4 days and then tested by IFN-γ ELISPOT assay. (B, C). Analysis of 4–1BB upregulation on (B) CD4+ and (C) CD8+ T cells by flow cytometry. As previously described, T cells were either left unstimulated (medium) or challenged with γ-irradiated MC38 tumor cells or γ-irradiated B16 tumor cells or naïve splenocytes from non-tumor-bearing B6 mouse in duplicate. After 24 h the cells have been stained for flow cytometry analysis for CD3, CD4, CD8, 4–1BB. Results are shown as individual data points (percentage of CD8+4–1BB+ T cells and CD4+4–1BB+ T cells) and bars (means ± standard deviation) of T cells from each mouse evaluated in duplicate. Data are presented as summary from 2 out of 5 independent experiments (n = 3–4 mice/group). For multiple group comparison One-way ANOVA was used. p < 0.05; p < 0.01; p < 0.001; *p < 0.0001. Figure 5. Analysis of the vvDD-IL2 induced TILs after ex vivo expansion. Prior to ACT, samples (T cells from naïve spleens: n = 4; TILs: n = 6) of the ex vivo cultured and expanded TILs were tested for tumor reactivity using IFN-γ ELISpot assay and co-culture assays for 4–1BB expression analyzed by flow cytometry as described previously. (A). Schema of experimental procedure. (B, C). Shown are (B) the ELISpot plate and (C) calculated IFN-γ+ spots per 1.0e6 T cells. (D, E). The cultured and expanded TILs were analyzed for cell surface 4–1BB expression of (D) CD4+ and (E) CD8+ T cells by flow cytometry. For multiple group comparison One-way ANOVA was used. p < 0.05; p < 0.01; p < 0.001; **p < 0.0001. Figure 6. ACT of oncolytic virus-induced TILs led to significant therapeutic efficacy in mice bearing peritoneal carcinomatosis of MC38 colon cancer. B6 mice were intraperitoneally inoculated with 5 × 105 MC38-luc cancer cells followed in 7 days, these mice were imaged for tumor growth and randomly grouped (n = 7 mice/group). Prior to ACT, mice in the treated groups received 5 Gy of sublethal irradiation. Grouped mice were intraperitoneally injected with PBS, naïve T cells, or vvDD-IL2 induced and ex vivo expanded TILs. All treated mice received exogenous IL-2 (1.0e5 IU/mouse, i.p. injection once every ~12 h for 3 days). One experiment representative of two independent experiments is shown (A, B). (A). Tumor growth has been monitored by live animal imaging 9 days and 17 days post treatment. (B). Radiance data quantified for two time points at days 9 and 17 post ACT. Student’s t-test was used to analyze the statistical significance. The variance was similar between the groups in this experiment. (C). The long-term survival of tumor-bearing mice was monitored by Kaplan-Meier analysis. These data represent one of the two independent experiments. p < 0.05; p < 0.01; *p < 0.001. (D). Summary of median survival in the different treatment groups. This experiment was performed 2 times. Conflicts of Interests: MF, ZL, ZSG and BLB filed a patent partly based on this work. Other authors declare that they have no conflict of interest. ==== Refs References 1. Yang Y : Cancer immunotherapy: harnessing the immune system to battle cancer. J Clin Invest 2015; 125 :3335–3337.26325031 2. Rosenberg SA , Restifo NP : Adoptive cell transfer as personalized immunotherapy for human cancer. Science 2015; 348 :62–68.25838374 3. Rosenberg SA , Yang JC , Sherry RM , Kammula US , Hughes MS , Phan GQ : Durable complete responses in heavily pretreated patients with metastatic melanoma using T-cell transfer immunotherapy. Clin Cancer Res 2011; 17 :4550–4557.21498393 4. Andersen R , Donia M , Ellebaek E , Borch TH , Kongsted P , Iversen TZ : Long-Lasting Complete Responses in Patients with Metastatic Melanoma after Adoptive Cell Therapy with Tumor-Infiltrating Lymphocytes and an Attenuated IL2 Regimen. Clin Cancer Res 2016; 22 :3734–3745.27006492 5. Chandran SS , Somerville RPT , Yang JC , Sherry RM , Klebanoff CA , Goff SL : Treatment of metastatic uveal melanoma with adoptive transfer of tumour-infiltrating lymphocytes: a single-centre, two-stage, single-arm, phase 2 study. Lancet Oncol 2017; 18 :792–802.28395880 6. Fridman WH , Zitvogel L , Sautes-Fridman C , Kroemer G : The immune contexture in cancer prognosis and treatment. Nat Rev Clin Oncol 2017; 14 :717–734.28741618 7. Baldan V , Griffiths R , Hawkins RE , Gilham DE : Efficient and reproducible generation of tumour-infiltrating lymphocytes for renal cell carcinoma. Br J Cancer 2015; 112 :1510–1518.25867267 8. Liu Z , Ravindranathan R , Kalinski P , Guo ZS , Bartlett DL : Rational combination of oncolytic vaccinia virus and PD-L1 blockade works synergistically to enhance therapeutic efficacy. Nat Commun 2017; 8 :14754.28345650 9. Ribas A , Dummer R , Puzanov I , VanderWalde A , Andtbacka RHI , Michielin O : Oncolytic virotherapy promotes intratumoral T cell infiltration and improves anti-PD-1 immunotherapy. Cell 2017; 170 :1109–1119 e1110.28886381 10. Gujar S , Pol JG , Kroemer G : Heating it up: Oncolytic viruses make tumors ‘hot’ and suitable for checkpoint blockade immunotherapies. Oncoimmunology 2018; 7 :e1442169.30221036 11. Bartlett DL , Liu Z , Sathaiah M , Ravindranathan R , Guo Z , He Y : Oncolytic viruses as therapeutic cancer vaccines. Mol Cancer 2013; 12 :103.24020520 12. Guo ZS , Liu Z , Bartlett DL : Oncolytic immunotherapy: dying the right way is a key to eliciting potent antitumor immunity. Front Oncol 2014; 4 :74.24782985 13. Todo T , Rabkin SD , Sundaresan P , Wu A , Meehan KR , Herscowitz HB : Systemic antitumor immunity in experimental brain tumor therapy using a multimutated, replication-competent herpes simplex virus. Hum Gene Ther 1999; 10 :2741–2755.10584921 14. Prestwich RJ , Errington F , Ilett EJ , Morgan RS , Scott KJ , Kottke T : Tumor infection by oncolytic reovirus primes adaptive antitumor immunity. Clin Cancer Res 2008; 14 :7358–7366.19010851 15. Gauvrit A , Brandler S , Sapede-Peroz C , Boisgerault N , Tangy F , Gregoire M : Measles virus induces oncolysis of mesothelioma cells and allows dendritic cells to cross-prime tumor-specific CD8 response. Cancer Res 2008; 68 :4882–4892.18559536 16. Li X , Wang P , Li H , Du X , Liu M , Huang Q : The efficacy of oncolytic adenovirus is mediated by T-cell responses against virus and tumor in Syrian hamster model. Clin Cancer Res 2017; 23 :239–249.27435398 17. Kaufman HL , Kim DW , DeRaffele G , Mitcham J , Coffin RS , Kim-Schulze S : Local and distant immunity induced by intralesional vaccination with an oncolytic herpes virus encoding GM-CSF in patients with stage IIIc and IV melanoma. Ann Surg Oncol 2010; 17 :718–730.19915919 18. Heo J , Reid T , Ruo L , Breitbach CJ , Rose S , Bloomston M : Randomized dose-finding clinical trial of oncolytic immunotherapeutic vaccinia JX-594 in liver cancer. Nat Med 2013; 19 :329–336.23396206 19. Li H , Zeng Z , Fu X , Zhang X : Coadministration of a herpes simplex virus-2 based oncolytic virus and cyclophosphamide produces a synergistic antitumor effect and enhances tumor-specific immune responses. Cancer Res 2007; 67 :7850–7855.17699791 20. Chard LS , Maniati E , Wang P , Zhang Z , Gao D , Wang J : A vaccinia virus armed with interleukin-10 is a promising therapeutic agent for treatment of murine pancreatic cancer. Clin Cancer Res 2015; 21 :405–416.25416195 21. de Graaf JF , de Vor L , Fouchier RAM , van den Hoogen BG : Armed oncolytic viruses: A kick-start for anti-tumor immunity. Cytokine Growth Factor Rev 2018; 41 :28–39.29576283 22. Ylosmaki E , Cerullo V : Design and application of oncolytic viruses for cancer immunotherapy. Curr Opin Biotechnol 2019; 65 :25–36.31874424 23. Guo ZS , Parimi V , O’Malley ME , Thirunavukarasu P , Sathaiah M , Austin F : The combination of immunosuppression and carrier cells significantly enhances the efficacy of oncolytic poxvirus in the pre-immunized host. Gene Ther 2010; 17 :1465–1475.20703311 24. McCart JA , Ward JM , Lee J , Hu Y , Alexander HR , Libutti SK : Systemic cancer therapy with a tumor-selective vaccinia virus mutant lacking thymidine kinase and vaccinia growth factor genes. Cancer Res 2001; 61 :8751–8757.11751395 25. Li J , O’Malley M , Urban J , Sampath P , Guo ZS , Kalinski P : Chemokine expression from oncolytic vaccinia virus enhances vaccine therapies of cancer. Mol Ther 2011; 19 :650–657.21266959 26. Liu Z , Ravindranathan R , Li J , Kalinski P , Guo ZS , Bartlett DL : CXCL11-armed oncolytic poxvirus elicits potent antitumor immunity and shows enhanced therapeutic efficacy. Oncoimmunology 2016; 5 :e1091554.27141352 27. Kowalsky SJ , Liu Z , Feist M , Berkey SE , Ma C , Ravindranathan R : Superagonist IL-15-Armed Oncolytic Virus Elicits Potent Antitumor Immunity and Therapy That Are Enhanced with PD-1 Blockade. Mol Ther 2018; 26 :2476–2486.30064894 28. Liu Z , Ge Y , Wang H , Ma C , Feist M , Ju S : Modifying the cancer-immune set point using vaccinia virus expressing re-designed interleukin-2. Nat Commun 2018; 9 :4682.30410056 29. Liu Y , Chen K , Wang C , Gong W , Yoshimura T : Isolation of mouse tumor-infiltrating leukocytes by percoll gradient centrifugation Bio-Protocol 2013; 3 :e892. 30. Jacobs JJ , Sparendam D , Den Otter W : Local interleukin 2 therapy is most effective against cancer when injected intratumourally. Cancer Immunol Immunother 2005; 54 :647–654.15685449 31. Ma J , Zheng B , Goswami S , Meng L , Zhang D , Cao C : PD1(Hi) CD8(+) T cells correlate with exhausted signature and poor clinical outcome in hepatocellular carcinoma. J Immunother Cancer 2019; 7 :331.31783783 32. Zeh HJ 3rd , Perry-Lalley D , Dudley ME , Rosenberg SA , Yang JC : High avidity CTLs for two self-antigens demonstrate superior in vitro and in vivo antitumor efficacy. J Immunol 1999; 162 :989–994.9916724 33. Oseroff C , Kos F , Bui HH , Peters B , Pasquetto V , Glenn J : HLA class I-restricted responses to vaccinia recognize a broad array of proteins mainly involved in virulence and viral gene regulation. Proc Natl Acad Sci U S A 2005; 102 :13980–13985.16172378 34. Crompton JG , Klemen N , Kammula US : Metastasectomy for tumor-infiltrating lymphocytes: an emerging operative indication in Surgical Oncology. Ann Surg Oncol 2018; 25 :565–572.29188500 35. Pages F , Berger A , Camus M , Sanchez-Cabo F , Costes A , Molidor R : Effector memory T cells, early metastasis, and survival in colorectal cancer. N Engl J Med 2005; 353 :2654–2666.16371631 36. Galon J , Costes A , Sanchez-Cabo F , Kirilovsky A , Mlecnik B , Lagorce-Pages C : Type, density, and location of immune cells within human colorectal tumors predict clinical outcome. Science 2006; 313 :1960–1964.17008531 37. Gao Q , Qiu SJ , Fan J , Zhou J , Wang XY , Xiao YS : Intratumoral balance of regulatory and cytotoxic T cells is associated with prognosis of hepatocellular carcinoma after resection. J Clin Oncol 2007; 25 :2586–2593.17577038 38. Goc J , Germain C , Vo-Bourgais TK , Lupo A , Klein C , Knockaert S : Dendritic cells in tumor-associated tertiary lymphoid structures signal a Th1 cytotoxic immune contexture and license the positive prognostic value of infiltrating CD8+ T cells. Cancer Res 2014; 74 :705–715.24366885 39. Turksma AW , Coupe VM , Shamier MC , Lam KL , de Weger VA , Belien JA : Extent and location of tumor-infiltrating lymphocytes in microsatellite-stable colon cancer predict outcome to adjuvant active specific immunotherapy. Clin Cancer Res 2016; 22 :346–356.26420855 40. Nomi T , Sho M , Akahori T , Hamada K , Kubo A , Kanehiro H : Clinical significance and therapeutic potential of the programmed death-1 ligand/programmed death-1 pathway in human pancreatic cancer. Clin Cancer Res 2007; 13 :2151–2157.17404099 41. Gao Q , Wang XY , Qiu SJ , Yamato I , Sho M , Nakajima Y : Overexpression of PD-L1 significantly associates with tumor aggressiveness and postoperative recurrence in human hepatocellular carcinoma. Clin Cancer Res 2009; 15 :971–979.19188168 42. Martínez-Bosch N , Fernández-Barrena MG , Moreno M , Ortiz-Zapater E , Munné-Collado J , Iglesias M : Galectin-1 drives pancreatic carcinogenesis through stroma remodeling and Hedgehog signaling activation. Cancer Res 2014; 74 :3512–3524.24812270 43. Kobayashi N , Kubota K , Kato S , Watanabe S , Shimamura T , Kirikoshi H : FOXP3+ regulatory T cells and tumoral indoleamine 2,3-dioxygenase expression predicts the carcinogenesis of intraductal papillary mucinous neoplasms of the pancreas. Pancreatology 2010; 10 :631–640.21051918 44. Brandacher G , Perathoner A , Ladurner R , Schneeberger S , Obrist P , Winkler C : Prognostic value of indoleamine 2,3-dioxygenase expression in colorectal cancer: effect on tumor-infiltrating T cells. Clin Cancer Res 2006; 12 :1144–1151.16489067 45. Feig C , Jones JO , Kraman M , Wells RJ , Deonarine A , Chan DS : Targeting CXCL12 from FAP-expressing carcinoma-associated fibroblasts synergizes with anti-PD-L1 immunotherapy in pancreatic cancer. Proc Natl Acad Sci U S A 2013; 110 :20212–20217.24277834 46. Chatterjee S , Behnam Azad B , Nimmagadda S : The intricate role of CXCR4 in cancer. Adv Cancer Res 2014; 124 :31–82.25287686 47. Guo ZS , Lu B , Guo Z , Giehl E , Feist M , Dai E : Vaccinia virus-mediated cancer immunotherapy: cancer vaccines and oncolytics. J Immunother Cancer 2019; 7 :6.30626434 48. Dafni U , Michielin O , Lluesma SM , Tsourti Z , Polydoropoulou V , Karlis D : Efficacy of Adoptive Therapy with Tumor-infiltrating Lymphocytes and Recombinant Interleukin-2 in Advanced Cutaneous Melanoma: A Systematic Review and Meta-analysis. Ann Oncol 2019; 30 :1902–1913.31566658 49. John LB , Howland LJ , Flynn JK , West AC , Devaud C , Duong CM : Oncolytic virus and anti-4–1BB combination therapy elicits strong anti-tumor immunity against established cancer. Cancer Res 2012; 72 :1651–1660.22315352 50. Jiang T , Zhou C , Ren S : Role of IL-2 in cancer immunotherapy. Oncoimmunology 2016; 5 :e1163462.27471638 51. Pol JG , Caudana P , Paillet J , Piaggio E , Kroemer G : Effects of interleukin-2 in immunostimulation and immunosuppression. J Exp Med 2020; 217 :e20191247.31611250 52. Berger C , Turtle CJ , Jensen MC , Riddell SR : Adoptive transfer of virus-specific and tumor-specific T cell immunity. Curr Opin Immunol 2009; 21 :224–232.19304470 53. Houghtelin A , Bollard CM : Virus-Specific T Cells for the Immunocompromised Patient. Front Immunol 2017; 8 :1272.29075259 54. Kodumudi KN , Siegel J , Weber AM , Scott E , Sarnaik AA , Pilon-Thomas S : Immune checkpoint blockade to improve tumor infiltrating lymphocytes for adoptive cell therapy. PLoS One 2016; 11 :e0153053.27050669 55. Fernandez-Poma SM , Salas-Benito D , Lozano T , Casares N , Riezu-Boj JI , Mancheno U : Expansion of tumor-infiltrating CD8(+) T cells expressing PD-1 improves the efficacy of adoptive T-cell therapy. Cancer Res 2017; 77 :3672–3684.28522749 56. Fu X , Rivera A , Tao L , Zhang X : An HSV-2 based oncolytic virus can function as an attractant to guide migration of adoptively transferred T cells to tumor sites. Oncotarget 2015; 6 :902–914.25460506 57. Ajina A , Maher J : Prospects for combined use of oncolytic viruses and CAR T-cells. J Immunother Cancer 2017; 5 :90.29157300 58. Klebanoff CA , Gattinoni L , Palmer DC , Muranski P , Ji Y , Hinrichs CS : Determinants of successful CD8+ T-cell adoptive immunotherapy for large established tumors in mice. Clin Cancer Res 2011; 17 :5343–5352.21737507 59. Lugli E , Dominguez MH , Gattinoni L , Chattopadhyay PK , Bolton DL , Song K : Superior T memory stem cell persistence supports long-lived T cell memory. J Clin Invest 2013; 123 :594–599.23281401 60. Klebanoff CA , Scott CD , Leonardi AJ , Yamamoto TN , Cruz AC , Ouyang C : Memory T cell-driven differentiation of naive cells impairs adoptive immunotherapy. J Clin Invest 2016; 126 :318–334.26657860 61. Melero I , Rouzaut A , Motz GT , Coukos G : T-cell and NK-cell infiltration into solid tumors: a key limiting factor for efficacious cancer immunotherapy. Cancer Discov 2014; 4 :522–526.24795012	32632271	PMC9718357	NO-CC CODE	2022-12-05 23:15:07	no	Cancer Gene Ther. 2021 Feb 7; 28(1-2):98-111	utf-8	Cancer Gene Ther	2,020	10.1038/s41417-020-0189-4	oa_other
==== Front J Biomol Struct Dyn J Biomol Struct Dyn Journal of Biomolecular Structure & Dynamics 0739-1102 1538-0254 Taylor & Francis 36331082 10.1080/07391102.2022.2141887 2141887 Version of Record Research Article Research Article Design of a novel multiple epitope-based vaccine: an immunoinformatics approach to combat monkeypox C. Hayat et al. Journal of Biomolecular Structure and Dynamics Hayat Chandni a Shahab Muhammad b Khan Salman Ali d Liang Chaoqun b Duan Xiuyuan b Khan Haleema c Zheng Guojun b https://orcid.org/0000-0002-8530-8711 Ul-Haq Zaheer de a Department of Biochemistry, Computational Medicinal Chemistry Laboratory, UCSS, Abdul Wali Khan University, Mardan, Pakistan b State Key Laboratories Of Chemical Resources Engineering, Beijing University of Chemical Technology, Beijing, China c Department of Chemistry, UCSS, Abdul Wali Khan University, Mardan, Pakistan d Dr. Panjwani Center for Molecular Medicine and Drug Research, International Center for Chemical and Biological Sciences, University of Karachi, Karachi, Pakistan e Third World Center for Science and Technology, H.E.J. Research Institute of Chemistry, International Center for Chemical and Biological Sciences, University of Karachi, Karachi, Pakistan CONTACT Zaheer Ul-Haq [email protected] Dr. Panjwani Center for Molecular Medicine and Drug Research, International Center for Chemical and Biological Sciences, University of Karachi, Karachi, Pakistan Guojun Zheng [email protected] State Key Laboratories Of Chemical Resources Engineering, Beijing University of Chemical Technology, Beijing, China 4 11 2022 2022 112 1 8 2022 24 10 2022 KnowledgeWorks Global Ltd.4 11 2022 published online ahead of issue4 11 2022 © 2022 Informa UK Limited, trading as Taylor & Francis Group 2022 Informa UK Limited, trading as Taylor & Francis Group This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the monkeypox public health emergency or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Abstract Monkeypox virus is an infectious agent that causes fever, Pneumonitis encephalitis, rash, lymphadenopathy and bacterial infection. The current outbreak of monkeypox has reawakened the global health concern. In the current situation of increasing viral infection, no vaccine or drug is available for monkeypox. Thus, there is an urgent need for viable vaccine development to prevent viral transmission by boosting human immunity. Herein, using immunoinformatics approaches, a multi-epitope vaccine was constructed for the Monkeypox virus. In this connection, B-Cell and T-cell epitopes were identified and joined with the help of adjutants and linkers. The vaccine construct was selected based on promising vaccine candidates and immunogenic potential. Further epitopes were selected based on antigenicity score, non-allergenicity and good immunological properties. Molecular docking reveals strong interactions between TLR-9 and the predicted vaccine construct. Finally, molecular dynamics simulations were performed to evaluate the stability and compactness of the constructed vaccine. The MD simulation results demonstrated the significant stability of the polypeptide vaccine construct. The predicted vaccine represented good stability, expression, immunostimulatory capabilities and significant solubility. Design vaccine was verified as efficient in different computer-based immune response investigations. Additionally, the constructed vaccine also represents a good population coverage in computer base analysis. Communicated by Ramaswamy H. Sarma Keywords Monkeypox epitope vaccine B-cell T-cell immunoinformatics This research was funded by the National Key R&D Program of China No. 2021YFC2102900 and Beijing Natural Science Foundation No. L212001. ==== Body pmcIntroduction Human monkeypox (MPX) is a rare viral infection caused by smallpox-like orthopoxvirus. In April 2022, hundreds of MPXV-positive cases were reported from more than 30 countries. At the end of July 2022, World Health Organization (WHO) declared the recent monkeypox (MPX) outbreak a global health emergency. MPXV is a close family member of the variola virus, the causative agent of smallpox, which killed 300 million people worldwide in the twentieth century (Kenneth, 1993). MPXV has a 200-kilobyte double-stranded DNA genome (Shchelkunov et al., 2001). Although MPXV is not a natural host for the virus, it was named when the virus was discovered in a laboratory for the first time in 1958. Monkeypox is a zoonotic disease, and the infection is frequently severe in children (Magnus et al., 2009). MPXV is transmitted mostly by contact with infected individuals and animals, along with infected material. The diseases can be prevented by avoiding close contact with infected individuals and animals, and also contaminated products. Monkeypox is a neglected disease prevalent in Central and Western Africa. However, it has recently gained international attention as a result of more than 100 confirmed and suspected cases (by 21 May 2022), affecting more than ten countries especially Australia, North America and Europe. It is predicted that the number of instances would increase worldwide (Potter et al., 2007). Bunge et al. analyze the trend of suspected and confirmed cases of MPXV over the last 50 years. The number of cases has increased over time: only 47 cases were reported during 1970–1979, 356 cases in 1980–1989, 520 cases in 1990–1999, 10,166 cases in 2000–2009 and over 19 thousand cases in 2010–2019 (Bunge et al., 2022). In 2020 and 2021, the WHO Bulletin reported about 6200 and 9400 confirmed and suspected cases, respectively (Impouma et al., 2018). Smallpox immunizations protect against monkeypox, and TPOXX, a smallpox medication, is promising and effective against monkeypox. Two US FDA-approved vaccines that can protect against monkeypox are JYNNEOS and ACAM2000. Unfortunately, there is still a need to develop a safe and effective multi-epitope vaccine against monkeypox. Resistance to antiviral medications is prevalent, therefore novel treatments that target viral sites are critical for treating monkeypox (Fine et al., 1988). To prevent uncontrollable spread, more intensive surveillance, and research on MPXV biology, history, transmission mode, host interactions, drug targets and vaccine development are extremely crucial. Governments, academics, industries and healthcare systems work together to achieve this goal. The conventional approach to vaccine development is risky and time-consuming. However, the immunoinformatics-based multi-epitope vaccine approach is an attractive alternative. Different vaccines are developed by using immunoinformatics approaches against human infections caused by H. pylori (Naz et al., 2015), Ebola virus (Bazhan et al., 2019), Marburg virus (Hasan et al., 2019), Mokola Rabies virus (Aa et al., 2017) and Crimean-Congo hemorrhagic virus (Nosrati et al., 2019). In this study, we designed an effective, safe and thermodynamically stable epitope-based vaccine for monkeypox, to trigger innate and adaptive immune responses. Here we employed an immunoinformatics approach to design non-allergic, and top-most antigenic epitopes for vaccine development, and the workflow is presented in Figure 1. The findings provide the way for the construction of the Monkeypox vaccine; however further experimental validation is needed to reduce the Monkeypox disease. Figure 1. Workflow and tools used in this study. Methodology Sequence retrieval The reference sequence of the monkeypox (accession no: AAQ09810.1) was first retrieved from NCBI (https://www.ncbi.nlm.nih.gov/protein/ AAQ09810.1), and then subjected to protein-protein blast (Blastp) against the non-redundant database and the resulting top 6 sequences (including the reference sequence) were retrieved as FASTA format. Multiple sequence alignment was conducted by the MUSCLE v3.6 program. Phylogenetic analysis of all 6 sequences was performed using Mega X. Different immunoinformatics tools and software were used for the development of the multi-epitope vaccine against monkeypox. First, the sequence was submitted to online tools Vaxijen 2.0 (Doytchinova & Flower, 2007a) and AllerTop (Dimitrov et al., 2014) to check antigenicity and allergenicity. Further, the Soluprot (https://loschmidt.chemi.muni.cz/soluprot/) (Hon et al., 2021) tools were employed to check the soluble protein expression in Escherichia coli. Physical and chemical characteristics were predicted via the online server Expasy (https://web.expasy.org/protparam/) (Wilkins et al., 2005). Similarly, the secondary structure was predicted via the online server, Psipred (http://bioinf.cs.ucl.ac.uk/psipred) (Jones, 1999). Prediction of B-cell and T-cell epitopes Selecting effective B-cell epitopes requires surface accessibility. Thus, the Emini surface accessibility tool was used to evaluate surface accessibility. IEDB Resource (http://tool.iedb.org/main/) (Nielsen et al., 2003) was used to predict the B-cell and T-cell epitopes and the binding scores were calculated for T-cell epitopes with MHC-I and MHC-II. The threshold value was kept at 0.5 and IC50 scores were assigned by the online server. A threshold score of more than 0.5 was taken as a good epitope candidate. IC50 and binding affinity are inversely proportional to each other. Mean if IC50 value is small, epitope binding affinity to MHC-II will be high. IC50 values <10 nM < 100 nM <1000 nM means high, intermediate and low binding affinity with MHC-II. Protein construct allergenicity prediction To check the antigenic and allergic behavior of B-cells and T-cells epitopes, online servers VaxiJen v2.0 (Doytchinova & Flower, 2007b) and AllerTop (Dimitrov et al., 2014) were used. The epitopes with a higher value than the reference and non-allergic were chosen for further research. Population coverage Using the online IEDB Analysis Resource, MHC-I and MHC-II epitopes were used to predict population coverage (tools.iedb.org/population/). A vaccine construct was designed using B and T cell epitopes with high binding affinity, non-allergenicity and antigenicity. Based on earlier works (Bhattacharya et al., 2020; Shams et al., 2020), AAY and GPGPG, linkers were used to connect all MHC-I and MHC-II epitopes, respectively. Further, the EAAAK linker was used to bind the adjuvant to the N-terminal of the vaccine construct (Arai et al., 2001). Secondary and tertiary structure prediction and validation PSIPRED 3.3 online server (http://bioinf.cs.ucl.ac.uk/psipred/) and TrRosetta web (Du et al., 2021) server were used to predict the secondary and tertiary structures of the vaccine construct. Galaxy Refine (https://galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE) was used to validate the vaccine. For further verification, ProSA-Web (Wiederstein & Sippl, 2007) Ramachandran Plot and PROCHECK (https://saves.mbi.ucla.edu/) (Laskowski et al., 2006) servers were used. Molecular docking of vaccine construct The three-dimensional structure of TLR-9 (PDB ID: 3WPF) was downloaded from PDB. The structure was subjected to geometry correction and missing residues were added. All the water molecules were removed. The MOE software was used to correct the protonation states. Then the structure was saved in TLR-9.pdb format. The vaccine construct was docked with the human Toll-like receptor 9 using Cluspro (https://cluspro.bu.edu) (Vajda et al., 2017), the most extensively used docking server, which is based on six energy functions. Cluspro predicted the 10 models in about three hours, based on densely packed low-energy clusters for each docking parameter. This procedure illustrates how to use various choices such as creating extra restrictions files, selecting various energy parameters and analyzing the outcomes. PDBsum (https://www.ebi.ac.uk/thornton-srv/databases/pdbsum/Generate.html) was used to evaluate the vaccine and TLR-9 interactions. Codon optimization and MD simulation The JCat tool was used to optimize codons and reverse the vaccination sequence (Grote et al., 2005). The JCat program is also used to guarantee that the vaccination sequence is expressed in a vector with a high level of expression. Three extra parameters were chosen in this tool, including Rho-independent transcription termination, restriction enzyme cleavage sites and bacterial ribosome binding sites. JCat also determines the CAI score and GC content of the vaccine sequence. MD simulation iMODS, an open-source modeling server for molecular dynamics simulation and image processing provides a user-friendly interface for internal normal mode analysis (NMA) of deformability, stability, and mobility of the target protein. Users can perform NMA or simulate possible trajectories between two conformations and view the results in 3D, even for several biomolecular complexes. Herein, iMODS was utilized to evaluate the dynamic stability of the TLR-9-vaccine complex. Results Sequence retrieval, phylogenetic analysis and sequence prioritization Protein information of the top 6 protein sequences, obtained by BlastP against the nr-database is depicted in Table 1, along with their properties including antigenicity, allergenicity and toxicity. MUSCLE v3.6 was used for multiple sequence alignment. A phylogenetic tree showing the phylogenetic relatedness among the sequences was constructed using the MEGA X program by neighbor-joining method with a bootstrap replication of 1000, shown in Figure 2. Among the 6 sequences, the protein sequence with accession number NP_536428.1 was found to be the most potent antigenic protein with a VaxiJen score of 0.7703 and selected to further design a multi-epitope based vaccine for monkeypox. Figure 2. Phylogenetic relationship among the studied protein (AAQ09810.1), reference protein and other proteins obtained from non-redundant database using BlastP. The evolutionary distances were computed using the Poisson correction method (Magnus et al., 2009) and are in the units of the number of amino acid substitutions per site. Table 1. Protein information of the 6 protein sequences obtained from NCBI by BlastP search. Accession No Protein name Sequence VaxiJen score Antigenicity NP_536428.1 Monkeypox virus Zaire-96-I-16 MKQYIVLACMCLVAAAMPTSLQQSSSSCTEEENKHHMGIDVIIKVTKQDQTPTNDKICQSVTEVTETED DEVSEEVVKGDPTTYYTIVGAGLNMNFGFTKCPKISSISESSDGNTVNTRLSSVSPGQGKDSPAI TREEALAMIKDCEMSIDIRCSEEEKDSDIKTHPVLGSNISHKKVSYKDIIGSTIVDTKCVKNLEFSV RIGDMCEESSELEVKDGFKYVDGSASEGATDDTSLIDSTKLKACV 0.7703 ANTIGEN AAQ09810 secreted chemokine binding protein MKQYIVLACMCLVAAAMPTSLQQSSSSCTEEENKHHMGIDVIIKVTKQDQTPTNDKICQSVTEVTETEDD EVSEDDEVSEEVVKGDPTTYYTIVGAGLNMNFGFTKCPKISSISESSDGNTVNTRLSSVSPGQGK DSPAITREEALAMIKDCEMSIDIRCSEEEKDSDIKTHPVLGSNISHKKVSYKDIIGSTIVDTKCVKN LEFSVRIGDMCEESSELEVKDGFKYVDGSASEGATDDTSLIDSTKLKACV 0.7574 ANTIGEN AAM76335.1 secreted chemokine binding protein MKQYIVLACMCLVAAAMPTSLQQSSSSCTEEENKHHMGIDVIIKVTKQDQTPTNDKICQSVTEVTETEDD EVSEEVVKGDPTTYYNIVGAGLNMNFGFTKCPKISSISESSDGNTVNTRLSSVSPGQGKDSPAITR EEALAMIKDCEMSIDIRCSEEEKDSDIKTHPVLGSNISHKKVSYKDIIGSTIVDTKCVKNLEFSVRIGD MCEESSELEVKDGFKYVDGSASEGATDDTSLIDSTKLKACVKDGDTWYYLEASGAMKASQWFK VSDKWYYVNGLGALAVNTTVDGYKVNANGEWV 0.7003 ANTIGEN AAM76334.1 secreted chemokine binding protein [Monkeypox virus] [Streptococcus pneumoniae] MKQYIVLACMCLVAAAMPTSLQQSSSSCTEEENKHHMGIDVIIKVTKQDQTPTNDKICQSVTEVTETEDDEV SEDDEVSEEVVKGDPTTYYNIVGAGLNMNFGFTKCPKISSISESSDGNTVNTRLSSVSPGQGKDSP AITREEALAMIKDCEMSIDIRCSEEEKDSDIKTHPVLGSNISHKKVSYKDIIGSTIVDTKCVKNLEFSV RIGDMCEESSELEVKDGFKYVDGSASEGATDDTSLIDSTKLKACV 0.7563 ANTIGEN URF91554.1 hypothetical protein MPXV-SI-2022V52144_00002 [Monkeypox virus] MKQYIVLACMCLVAAAMPTSLQQSSSSCTEEENKHHMGIDVIIKVTKQDQTPTNDKICQSVTEVTETEDDEV SEEVVKGDPTTYYTIVGAGLNMNFGFTKCPKILSISESSDGNTVNTRLSSVSPGQGKDSPAITREEAL AMIKDCEMSIDIRCSEEEKDSDIKTHPVLGSNISHKKVSYKDIIGSTIVDTKCVKNLEFSVRIGDMCEE SSELEVKDGFKYVDGSASEGATDDTSLIDSTKLKACV 0.7634 ANTIGEN USS79525.1 chemokine binding protein [Monkeypox virus] MKQYIVLACMCLVAAAMPTSLXXXXSSCTEEENKHHMGIDVIIKVTKQDQTPTNDKICQSVTEVTETEDDEV SEEVVKGDPTTYYTIVGAGLNMNFGFTKCPKILSISESSDGNTVNTRLSSVSPGQGKDSPAITREEALA MIKDCEMSIDIRCSEEEKDSDIKTHPVLGSNISHKKVSYKDIIGSTIVDTKCVKNLEFSVRIGDMCEESSE LEVKDGFKYVDGSASEGATDDTSLIDSTKLKACV 0.7521 ANTIGEN Physiochemical properties The sequence of monkeypox (NP_536428.1) was submitted to Protparam (https://web.expasy.org/protparam/), an online tool to calculate the physical and chemical properties of the vaccine. The results reveal that the vaccine construct has a total number of 246 amino acid residues, and a molecular weight of 33547.32 kDa, and the theoretical isoelectric point (PI) value of 7.60. The total number of negatively charged residues (Asp + Glu) was 36 and the total number of positively charged was 37 which indicates that the protein is positively charged, as is the case with isoelectric points over 7.0. An instability index (II) of 58.15 determined by Protparam classified our protein to be stable. The aliphatic index was 98.99, indicating that it is thermo-stable over a wide temperature range. At 0.076, the grand average of hydropathcity (GRAVY) was computed using the chemical formula C1449H2412N394O450S31. Secondary structure analysis The monkeypox has 11.4% β-strands, 48.9% α-helices and 39.7% coil structures, according to the secondary structure prediction using PSIPRED and 3D structure prediction using trRosetta (https://yanglab.nankai.edu.cn/trRosetta/) (Figure 3) (Zheng et al., 2021). TMHMM, an online tool, was used to predict transmembrane topology. Surface exposed residues were detected at positions 150–160, 165–175, 185–190 and 195–200, whereas residues from 10–20, 50–80, 85–105 and 115–120 were found inside the transmembrane region. The core area of the chemokine-binding protein of monkeypox was determined to have residues from positions 200–206, 210–225 and 228–235. Figure 3. Graphical representation of secondary structure prediction of target protein. H; helix, E; strands and C; coils. B-cell epitope prediction The FASTA sequence of the monkeypox was subjected to the IEDB online server with default parameters to identify potential B-cell epitopes. Residues with a higher value than the given threshold of 0.5 were designated as B-cell epitopes and are graphically presented in Figure 4. The IEDB server predicted a total number of 6 B-cell epitopes. All six epitopes were deposited to online tools VaxiJen 2.0 and AllerTop to calculate their antigenicity and allergenicity. Among the six epitopes, 3 epitopes were finalized based on good antigenic scores and non-allergic behavior which is presented in Table 2 along with their amino acid sequence, length and position. The antigenicity analysis showed that the minimum antigenicity value was 0.5181 and the maximum value was 0.847. However, the average value of 0.62 was observed. Figure 4. Graphical representation of B-cell epitope (0.5 shows the threshold value of the yellow color epitopic portion while the green color represents the non-epitopic portion). Table 2. IEDB analysis resource predicts a list of Bepipred linear epitopes. Start End Peptide Length 46 81 TKQDQTPTNDKICQSVTEVTETEDDEVSEEVVKGDP 36 156 195 EEEKDSDIKTHPVLGSNISHKKVSYKDIIGSTIVDTKCVK 40 206 242 MCEESSELEVKDGFKYVDGSASEGATDDTSLIDSTKL 37 T-cell epitope prediction MHC-1 epitopes Stabilized matrix method was utilized in order to prioritize the selected epitopes using the IC50 threshold of 100 nM. IC50 value less than 100, representing strong and higher binding affinity of epitopes that bind with MHC-1. A lesser IC50 suggests a greater affinity for MHC-I molecules. To maximize affinity for MHC-1 alleles, the total number of epitopes was designed to be fewer than 100. Based on MHC-1 allele interactions and IC50 value, 89 epitopes were chosen. A total of 10 epitopes were selected showing good antigenicity, allergenicity and non-toxicity. The antigenic scores of toxic and allergic epitopes were ruled out. The MHC-1 epitopes were finalized that bind to alleles HLA-A02:06, HLA-B15:01, HLA-A68:02 and HLA-A03:01. NISHKKVSYK has an antigenic score of 1.3013 (Table 3). Table 3. IEDB analysis resource predicts a list of Bepipred linear epitopes. Start End Peptide antigenic score Length 5 14 MGIDVIIKV 0.5798 10 8 17 ACMCLVAAAM 0.9166 10 45 53 TVNTRLSSV 1.1279 9 6 14 VLACMCLVA 0.7627 9 6 15 VLACMCLVAA 0.8605 10 28 37 VLGSNISHKK 0.9404 10 4 13 YIVLACMCLV 1.1068 10 5 13 IVLACMCLV 1.1304 9 5 10 IVLACMCLVA 1.0097 10 2 10 KQYIVLACM 0.7219 9 MHC-II epitopes MHC-II alleles interacted with 550 conserved predicted epitopes with IC50 less than 60 nM. 30 epitopes were selected among 550 that interacted with five MHC-II alleles. 13 epitopes were selected for further study based on their allergenicity, toxicity and antigenicity. The epitopes MSIDIRCSE, LGSNISHKK, IIKVTKQDQ, ISHKKVSYK and LGSNISHKK were considered the top binder, with alleles HLA-DRB103:01, HLA-DRB501:01, HLA-DRB107:01, HLA-DRB501:01 HLA-DRB401:01 and HLA-DRB103:01 (Table 4). Table 4. Antigenicity prediction using the Kolaskar and Tongaonkar technique. Start End Peptide antigenic score Length 5 13 MSIDIRCSE 2.4214 9 22 30 LGSNISHKK 1.0759 9 40 49 IIKVTKQDQ 1.2531 9 30 38 ISHKKVSYK 1.3341 9 26 34 LGSNISHKK 1.0759 9 15 23 ISHKKVSYK 1.3341 9 39 47 IIKVTKQDQ 1.2531 9 3 11 MSIDIRCSE 1.1304 9 8 16 LGSNISHKK 1.0759 9 Construction of vaccine Vaccine ensembles were performed by joining a total number of 3 B-cell epitopes and 19 T-cell epitopes (10 MHC-I and 9 MHC-II). 50S ribosomal protein is used as an adjuvant for the construction of vaccines. Through different linkers, adjuvant were combined with B-cell epitope to make a specific immune response. At the C-terminus of the vaccine sequence, a 6× His tag was inserted in order to facilitate the protein identification and purification processes. The final vaccine construct was deposited to the online tools VaxiJen2.0 and AllerTop to evaluate the antigenicity and allergenicity. Prediction of antigenicity and allergenicity found that the construct was a probable antigen (VaxiJen score 0.8780) and non-allergen. The generated vaccine construct is shown below. >vaccine protein EAAAKTKQDQTPTNDKICQSVTEVTETEDDEVSEEVVKGDPEEEKDSDIKTHPVLGSNISHKKVSYKDIIGSTIVDTKCVKMCEESSELEVKDGFKYVDGSASEGATDDTSLIDSTKLCPGPGMGIDVIIKVACMCLVAAAMTVNTRLSSVVLACMCLVAVLACMCLVAAVLGSNISHKKYIVLACMCLVIVLACMCLVIVLACMCLVAKQYIVLACMAAYMSIDIRCSELGSNISHKKIIKVTKQDQISHKKVSYKLGSNISHKKISHKKVSYKIIKVTKQDQMSIDIRCSELGSNISHKKHHHHHH Population coverage The MH Class-I allele was found to be present in 84.06% of the world’s population, followed by East Asia (74.06%), South Asia (78.07%), Hong Kong (66.57%), Europe (75.1%), North-East Asia (74.39%), Southwest Asia (69.03%), South America (80.72%), Central America (75.32%) and South Africa (80.56%) as depicted in Figure 5. The lowest population was found in Hong Kong. South America and South Africa have the highest percentage of MH Class-I and MH Class-II alleles in the population followed by East Asia. Three epitopes in MH Class-I (TVNTRLSSV, YIVLACMCLV and IVLACMCLVA) are crucial for the majority of interactions. Five Epitope MH Class-II alleles (LGSNISHKK, MSIDIRCSE, MSIDIRCSE, ISHKKVSYK and LGSNISHKK); describe a considerable coverage in contrast to the whole world population. For IVLACMCLVA, the proportion of concentrated population coverage in the world was anticipated at 70.57%. The population coverage data for the abundant binders to MH Class-I and MH Class-II alleles reveal 88.06 and 75.06% coverage, respectively. Figure 5. Bar-graph represents the population coverage of final 10 epitopes in different regions of the world. 3D structure prediction and validation The PROSA 3D server was used to predict the 3D structure of the multi-epitope vaccine sequence, as a result, ten structures were predicted for a given query sequence. Model five was selected for further investigation (Figure 6). Using the ERRAT, ProSA-web and PROCHECK servers, the structure was validated and any potential errors in the projected tertiary structure were corrected. The ERRAT server projected the overall quality of the vaccine 3D structure, with an estimated quality score of 95.0%. The Z-score was calculated to determine whether the input structure was within the range of natural proteins of similar size. The computed Z-score for the input structure was −6.67, indicating that it was outside the usual range for natural proteins of the same size, as shown in Figure 7. For Ramachandran analysis, the PROCHECK server shows the total number of 308 amino acid residues, 96.9% of the residues in the most favored areas, 2.8% in extra permitted regions, 0.3% in generously allowed regions, and 0.0% in the forbidden region, with 0.0% residues in disallowed regions. Figure 6. Graphical and 3D view of vaccine construction. 50 s ribosomal protein, EAAAL Linker, CPGPG Linker and AAY Linker are shown. Figure 7. Validation of the final vaccine 3D models. PROSA 3D structure validation showing corresponding Z-score of −6.67. Ramachandran plot shows most favored (96.99%), allowed (2.8%), generously allowed (0.3%) and disallowed regions (0.0%), respectively. Molecular docking analysis To analyze the interaction and forecast the final 3D complex, TLR-9 and the vaccine construct were docked using the Clustpro docking server. A total of ten models were obtained. Using the Pymol program, all ten docking models were visually examined and analyzed. Among the 10 models, the first model showed a good docking result with a total of 11 H-bond interactions with a binding score of −1202.4. For a graphical illustration of the residual interaction between the vaccine construct and TLR-9, the PDBsum online database was used. To evaluate the hydrogen bonding between the vaccine construct and TLR-9 complex, a graphical picture was generated (Figure 8). 11 hydrogen bond contacts between TLR-9 and the vaccine construct were observed. Analysis of the vaccine-TLR complex revealed that hydrogen bonds were formed between HIS203-ASN145, THR202-ASN145, GLU223-ASN145, HIS203-SER149, HIS506-LYS179, HIS531-LYS179, ARG482-LYS180, ASN372-CYS188, THR395-CYS195, THR395-CYS195 and ARG337-VAL199 at a distance of 3.32, 2.86, 2.88, 3.27, 2.83, 2.80, 2.53, 3.03, 3.05, 3.05 and 2.81 Å, respectively. Figure 8. TLR-9 (PDB ID: 3WPF)-vaccine docked complex. The TLR-9 is shown in magentas, while the yellow color represents the multi-epitope subunit vaccine. Left panel graphically represents the hydrogen bonds interaction between TLR-9 and vaccine complex. Molecular dynamic simulation MD simulation was performed by using iModS. iModS analyze structure by adjusting complex force field with various time interval. In the heat map, high co-related area and low RMSD indicated good interactions of individual residues. Figure 9 represents a more detailed explanation of IModS results. Figure 9A represents the NMA mobility in protein structure, Figure 9B represents the deformability and Figure 9C represents the B-factor. Figure 9D shows Eigen value, Figure 9E showed variance and Figure 9F shows the complex elastic network and covariance. Figure 9. MD simulation of multi-epitope vaccine complexed with the TLR-9. Immune simulations The vaccine’s potential to induce a robust immunological response if delivered globally was evaluated using the C-ImmSim program. After the primary reaction, both secondary and 3D immune response was high. Different antibodies IgG and IgM were detected. Correspondingly, IFN-γ and IL-2 were also observed (Figure 10). Figure 10. Vaccine immune simulation through C-ImmSim server. (A) The production of antibodies. (B) The population of B-cell. (C) Cytokines production and (D) T cell-population. Codon optimization The Jcat server was used to identify reverse translation and codon optimization in E. coli in order to locate expression in the vaccine. The vaccine sequence consisted of nucleotides, whereas the CAI was 0.53 and the GC content was 61.25, indicating a high level of expression. Two main restriction sites, BanI and Eco241 were added. The restriction sites and vaccines were cloned with Snap gene software. The vaccines along with another site in the cloning vector are depicted in Figure 11. Figure 11. The cloning of the final vaccine where the black color shows the vector and the red color shows the vaccine insert. Discussion As the world recovers from the pandemic of COVID-19, a new infectious disease, monkeypox widespread and has clusters across Europe, America and Australia. There is currently no antiviral treatment that has been licensed by the Food and Drug Administration (FDA) of the United States particularly against MPXV. Alternative preventative measures include smallpox vaccinations such as JYNNEOS and ACAM2000. Despite these preventive measures, the widespread transmission of monkeypox necessitates the development of novel therapeutics against MPXV. The implementation of advanced bioinformatics tools is more beneficial than conventional methods (Oli et al., 2020). Bio-computational research, especially reverse vaccinology, is an attractive alternative to developing an epitopic vaccination for monkeypox in the current situation of rapidly spreading disease. Therefore, using immunoinformatic approaches, our research aims to design a vaccine against monkeypox. Using a similar method, Bazhan et al. designed a T-cell multi-epitope vaccine model that was highly immunogenic in mice against the Ebola virus (Bazhan et al., 2019). The FASTA sequence of monkeypox was obtained from the NCBI database. The antigenicity, non-toxicity and non-allergenic behavior of the proteins were investigated. MHC-I and MHC-II epitopes on B-cells and T-cells were predicted. By using linkers CPGPG and AAY, selected epitopes joined with each other. The effectiveness of the predicted vaccine has demonstrated that our design vaccine is both non-allergic and has a high antigenic score (0.8780), along with a good solubility expression inside E. coli (0.5075). Constructed vaccine further proceeded for physicochemical properties analysis, where molecular weight was 33547.32, and GRAVY was 0.076, which represents our vaccine is hydrophobic. Similar in silico methodologies were also used by Foroutan et al. against Toxoplasma gondii to analyse the antigenicity and physiochemical properties of their model vaccine, in addition to laboratory validation (Foroutan, 2020). It was demonstrated that this strategy for the design of vaccines was successful in eliciting an immunological response in mice. The vaccine’s stability was demonstrated by its low instability score of 58.15 and its high aliphatic index of 98.99. For secondary structure, PSIPRED V3.3 online tool was used, which was found to be 11.4% β-strands, 48.9% α-helices and 39.7% coil structures while the 3D structure of constructed vaccine was modeled by online server trRosetta. 3D model validation was performed by using ERRAT and PROCHECK. Ramachandran plot represents the suitable stereochemical statics, essential for the 3D structure. Results verified that most of the residues were inside the allowed regions. The Z score predicted by ProSa-Web was −6.67, although the score was outside the usual range for natural proteins of the same size; however with Z-scores of −5.26 and −9.5 kcal/mol, Droppa-Almeida et al. and Rekik et al. concluded that the predicted 3D structures were accurate and of high quality (Droppa-Almeida et al., 2018; Rekik et al., 2015). Taken together the stereochemical quality of the predicted model is acceptable to be used further. Previous studies have suggested the importance of toll-like receptors in immune response stimulation (Iwasaki & Medzhitov, 2004). Thus, the vaccine construct was subjected to molecular docking studies with TLR-9. By Cluspro server docking was performed and the results reveal that the vaccine construct exhibits strong interactions with the binding site residues of TLR-9 which indicated the ability of the vaccine to elicit immunogenic responses. Finally, MD simulation was used to validate the stability of the vaccine construct and TLR-9 docked complex. The results indicated the stable molecular interactions between the vaccination and immunological receptor, assuring the molecular stability of the multi-epitope vaccine complex in a cellular environment. Immune simulation results predicted cellular immune responses that were similar to those observed in nature. High levels of T-cytotoxic, memory cells and Ig production were observed, along with an increase in IFN-γ and IL-2 levels. Jcat software is used to predict the best protein expression in E. coli K12 strain for codon optimization to enhance transcription and translation efficacy. Similar to the results of previous immunoinformatics vaccine designed studies, the designed vaccines may offer protection against MPXV (Ismail et al., 2020; Tahir Ul Qamar et al., 2020). This study suggested an alternate vaccine method based on the multi-epitope assembly of MPXV genome protein components to deal with antigenic complexity. The predicted vaccine is believed to be immunogenic based on immunoinformatics techniques and it could contribute to eradicating the disease. However, in vitro immunological assays are needed to validate the potency of the vaccine. Disclosure statement The authors declare that they have no conflicts of interest. Author’s contributions C.H. & M.S.: Contribute Equally, Conceptualization, Methodology, Visualization, Data Curation. C.L. & X.D.: Investigation, S.A.K.: Writing - review & editing, Data Curation. H.K.: Methodology write-up. G.Z. & Z.U.-H.: Supervision, Project administration, Writing - review & editing, Software, Funding acquisition. Data availability statement NCBI database in FASTA format with ID (AAQ09810.1). ==== Refs References Aa, M., O, H., Kaa, E., A, H., & Ma, H. (2017). Epitope-based peptide vaccine design against Mokola rabies virus glycoprotein G utilizing in silico approaches. Immunome Research, 13 (3 ), 2. 10.4172/1745-7580.1000144 Arai, R., Ueda, H., Kitayama, A., Kamiya, N., & Nagamune, T. (2001). Design of the linkers which effectively separate domains of a bifunctional fusion protein. Protein Engineering, 14 (8 ), 529–532. 10.1093/protein/14.8.529 11579220 Bazhan, S. I., Antonets, D. V., Karpenko, L. I., Oreshkova, S. F., Kaplina, O. N., Starostina, E. V., Dudko, S. G., Fedotova, S. A., & Ilyichev, A. A. (2019). In silico designed ebola virus T-cell multi-epitope DNA vaccine constructions are immunogenic in mice. Vaccines, 7 (2 ), 34. 10.3390/vaccines7020034 30934980 Bhattacharya, M., Sharma, A. R., Patra, P., Ghosh, P., Sharma, G., Patra, B. C., Lee, S.-S., & Chakraborty, C. (2020). Development of epitope‐based peptide vaccine against novel coronavirus 2019 (SARS‐COV‐2): Immunoinformatics approach. Journal of Medical Virology, 92 (6 ), 618–631. 10.1002/jmv.25736 32108359 Bunge, E. M., Hoet, B., Chen, L., Lienert, F., Weidenthaler, H., Baer, L. R., & Steffen, R. (2022). The changing epidemiology of human monkeypox—A potential threat? A systematic review. PLoS Neglected Tropical Diseases, 16 (2 ), e0010141. 10.1371/journal.pntd.0010141 35148313 Dimitrov, I., Bangov, I., Flower, D. R., & Doytchinova, I. (2014). AllerTOP v. 2—a server for in silico prediction of allergens. Journal of Molecular Modeling, 20 (6 ), 2278–2276. 10.1007/s00894-014-2278-5 24878803 Doytchinova, I. A., & Flower, D. R. (2007a). Identifying candidate subunit vaccines using an alignment-independent method based on principal amino acid properties. Vaccine, 25 (5 ), 856–866. 10.1016/j.vaccine.2006.09.032 17045707 Doytchinova, I. A., & Flower, D. R. (2007b). VaxiJen: A server for prediction of protective antigens, tumour antigens and subunit vaccines. BMC Bioinformatics, 8 (1 ), 1–7. 10.1186/1471-2105-8-4 17199892 Droppa-Almeida, D., Franceschi, E., & Padilha, F. F. (2018). Immune-informatic analysis and design of peptide vaccine from multi-epitopes against Corynebacterium pseudotuberculosis. Bioinformatics and Biology Insights, 12 , 1177932218755337.29780242 Du, Z., Su, H., Wang, W., Ye, L., Wei, H., Peng, Z., Anishchenko, I., Baker, D., & Yang, J. (2021). The trRosetta server for fast and accurate protein structure prediction. Nature Protocols, 16 (12 ), 5634–5651. 10.1038/s41596-021-00628-9 34759384 Fine, P. E., Jezek, Z., Grab, B., & Dixon, H. (1988). The transmission potential of monkeypox virus in human populations. International Journal of Epidemiology, 17 (3 ), 643–650. 10.1093/ije/17.3.643 2850277 Foroutan, M. (2020). Vaccination with a novel multi-epitope ROP8 DNA vaccine against acute Toxoplasma gondii infection induces strong B and T cell responses in mice. Comparative Immunology, Microbiology and Infectious Diseases, 65 , 101413. Grote, A., Hiller, K., Scheer, M., Münch, R., Nörtemann, B., Hempel, D. C., & Jahn, D. (2005). JCat: A novel tool to adapt codon usage of a target gene to its potential expression host. Nucleic Acids Research, 33 (Web Server issue ), W526–W531. 10.1093/nar/gki376 15980527 Hasan, M., Azim, K. F., Begum, A., Khan, N. A., Shammi, T. S., Imran, A. S., Chowdhury, I. M., & Urme, S. R. A. (2019). Vaccinomics strategy for developing a unique multi-epitope monovalent vaccine against Marburg marburgvirus. Infection, Genetics and Evolution, 70 , 140–157. 10.1016/j.meegid.2019.03.003 Hon, J., Marusiak, M., Martinek, T., Kunka, A., Zendulka, J., Bednar, D., & Damborsky, J. (2021). SoluProt: Prediction of soluble protein expression in Escherichia coli. Bioinformatics, 37 (1 ), 23–28. 10.1093/bioinformatics/btaa1102 33416864 Impouma, B., Archer, B. N., Lukoya, O. C., Hamblion, E. L., & Fall, I. S. (2018). World Health Organization Regional Office for Africa weekly bulletin on outbreaks and other emergencies. Emerging Infectious Diseases, 24 (7 ), 1394–1395. 10.3201/eid2407.180573 Ismail, S., Ahmad, S., & Azam, S. S. (2020). Vaccinomics to design a novel single chimeric subunit vaccine for broad-spectrum immunological applications targeting nosocomial Enterobacteriaceae pathogens. European Journal of Pharmaceutical Sciences, 146 , 105258.32035109 Iwasaki, A., & Medzhitov, R. (2004). Toll-like receptor control of the adaptive immune responses. Nature Immunology, 5 (10 ), 987–995. 10.1038/ni1112 15454922 Jones, D. T. (1999). Protein secondary structure prediction based on position-specific scoring matrices. Journal of Molecular Biology, 292 (2 ), 195–202. 10.1006/jmbi.1999.3091 10493868 Kenneth, F. K. (1993). The Cambridge world history of human disease. Cambridge University Press. Laskowski, R., MacArthur, M., & Thornton, J. (2006). PROCHECK: Validation of protein-structure coordinates. Magnus, P. v., Andersen, E. K., Petersen, K. B., & Birch-Andersen, A. (2009). A pox‐like disease in cynomolgus monkeys. Acta Pathologica Microbiologica Scandinavica, 46 (2 ), 156–176. 10.1111/j.1699-0463.1959.tb00328.x Naz, A., Awan, F. M., Obaid, A., Muhammad, S. A., Paracha, R. Z., Ahmad, J., & Ali, A. (2015). Identification of putative vaccine candidates against Helicobacter pylori exploiting exoproteome and secretome: A reverse vaccinology based approach. Infection, Genetics and Evolution, 32 , 280–291. 10.1016/j.meegid.2015.03.027 Nielsen, M., Lundegaard, C., Worning, P., Lauemøller, S. L., Lamberth, K., Buus, S., Brunak, S., & Lund, O. (2003). Reliable prediction of T‐cell epitopes using neural networks with novel sequence representations. Protein Science, 12 (5 ), 1007–1017. 10.1110/ps.0239403 12717023 Nosrati, M., Behbahani, M., & Mohabatkar, H. (2019). Towards the first multi-epitope recombinant vaccine against Crimean-Congo hemorrhagic fever virus: A computer-aided vaccine design approach. Journal of Biomedical Informatics, 93 , 103160.30928513 Oli, A. N., Obialor, W. O., Ifeanyichukwu, M. O., Odimegwu, D. C., Okoyeh, J. N., Emechebe, G. O., Adejumo, S. A., & Ibeanu, G. C. (2020). Immunoinformatics and vaccine development: an overview. ImmunoTargets and Therapy, 9 , 13–30.32161726 Potter, M. A., Sweeney, P., Iuliano, A. D., & Allswede, M. P. (2007). Performance indicators for response to selected infectious disease outbreaks: A review of the published record. Journal of Public Health Management and Practice : JPHMP, 13 (5 ), 510–518. 10.1097/01.PHH.0000285205.40964.28 17762697 Rekik, I., Chaabene, Z., Grubb, C. D., Drira, N., Cheour, F., & Elleuch, A. (2015). In silico characterization and Molecular modeling of double-strand break repair protein MRE11 from Phoenix dactylifera v deglet nour. Theoretical Biology and Medical Modelling, 12 (1 ), 1–14. 10.1186/s12976-015-0013-2 25566687 Shams, N., Shakarami Gandabeh, Z., Nazifi, N., Forouharmehr, A., Jaydari, A., & Rashidian, E. (2020). Computational design of different epitope-based vaccines against Salmonella typhi. International Journal of Peptide Research and Therapeutics, 26 (3 ), 1527–1539. 10.1007/s10989-019-09959-4 Shchelkunov, S. N., Totmenin, A. V., Babkin, I. V., Safronov, P. F., Ryazankina, O. I., Petrov, N. A., Gutorov, V. V., Uvarova, E. A., Mikheev, M. V., Sisler, J. R., Esposito, J. J., Jahrling, P. B., Moss, B., & Sandakhchiev, L. S. (2001). Human monkeypox and smallpox viruses: Genomic comparison. FEBS Letters, 509 (1 ), 66–70. 10.1016/S0014-5793(01)03144-1 11734207 Tahir Ul Qamar, M., Shokat, Z., Muneer, I., Ashfaq, U. A., Javed, H., Anwar, F., Bari, A., Zahid, B., & Saari, N. (2020). Multiepitope-based subunit vaccine design and evaluation against respiratory syncytial virus using reverse vaccinology approach. Vaccines, 8 (2 ), 288. 10.3390/vaccines8020288 32521680 Vajda, S., Yueh, C., Beglov, D., Bohnuud, T., Mottarella, S. E., Xia, B., Hall, D. R., & Kozakov, D. (2017). New additions to the C lus P ro server motivated by CAPRI. Proteins, 85 (3 ), 435–444. 10.1002/prot.25219 27936493 Wiederstein, M., & Sippl, M. J. (2007). ProSA-web: Interactive web service for the recognition of errors in three-dimensional structures of proteins. Nucleic Acids Research, 35 (Web Server issue ), W407–W410. 10.1093/nar/gkm290 17517781 Wilkins, M. R., Gasteiger, E., Bairoch, A., Sanchez, J. C., Williams, K. L., Appel, R. D., & Hochstrasser, D. F. (1999). Protein identification and analysis tools in the ExPASy server. Methods in Molecular Biology (Clifton, N.J.), 112, 531–552. 10.1385/1-59259-584-7:531 Zheng, W., Zhang, C., Li, Y., Pearce, R., Bell, E. W., & Zhang, Y. (2021). Folding non-homologous proteins by coupling deep-learning contact maps with I-TASSER assembly simulations. Cell Reports Methods, 1 (3 ), 100014. 10.1016/j.crmeth.2021.100014 34355210	36331082	PMC9718376	NO-CC CODE	2022-12-07 23:18:55	no	J Biomol Struct Dyn.;:11-12	utf-8	J Biomol Struct Dyn	2,022	10.1080/07391102.2022.2141887	oa_other
==== Front Resour Policy Resour Policy Resources Policy 0301-4207 1873-7641 Elsevier Ltd. S0301-4207(22)00564-5 10.1016/j.resourpol.2022.103121 103121 Article The role of foreign trade and technology innovation on economic recovery in China: The mediating role of natural resources development Hasan Mohammad Maruf ab Du Fang ac∗ a School of International Studies, Sichuan University, Chengdu, Sichuan, 610065, PR China b School of Economics, Sichuan University, Chengdu, Sichuan, 610065, PR China c School of Public Administration, Sichuan University, Chengdu, 610065, Sichuan, China ∗ Corresponding author. School of International Studies, Sichuan University, Chengdu, Sichuan, 610065, PR China. 2 12 2022 1 2023 2 12 2022 80 103121103121 23 7 2022 4 11 2022 6 11 2022 © 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The coronavirus disease (COVID-19) has caused the most recent global economic collapse, which severely impacted worldwide economic operations. Natural resource volatility significantly affects global economic recovery. Therefore, the study aims to determine the significance of natural resource volatility, foreign trade, technological innovation, urbanisation, and investment in energy resources on the economic recovery of Chinese provinces. A total of 30 provinces in China were examined between 1995 and 2020. The generalised method of movement (GMM) technique was used to demonstrate that investments in energy sector resources, international commerce, and technological innovation are inconsistent than gross domestic product (GDP) and natural resources. The results demonstrated how trade blocs limit the effect of natural resources on regional economic development in the central provinces. Specifically, 33.4% of energy was saved while 35.2% of emissions were reduced. Although abundant natural resources significantly influence economic growth, studies discovered a negative impact on urbanisation. Nonetheless, the positive effects of trade openness outweighed those of economic recovery. The study proposed stabilising the fluctuating costs of natural resources, encouraging green financing, and increasing investment in energy resources. The findings also provided a novel strategy for achieving high economic growth and recovery. Keywords Natural resources volatility Technology innovation Ecological footprint Foreign trade Economic recovery ==== Body pmc1 Introduction The world has experienced exceptional economic development over the last 20 years, which accounts for 59% of global inhabitants and 40% of global GDP with approximately US$ 4.5 trillion in international currency holdings. Essentially, developing economies are the primary participants in the global economy (Rasoulinezhad, 2020). The total GDP of these transitioning economies escalated from US$ 4121 billion (in constant US$ 2010) to US$ 24,490 billion with an average growth rate of approximately 5.4% from 1983 to 2017 (Yoshino et al., 2021). Economic development enables countries to build essential infrastructure, combat poverty, and improve people's quality of life. Nevertheless, the growth process has disadvantages, specifically when nations choose artificial luxury over the ecosystem health (Saboori et al., 2017). Developing nations have compromised their natural resource reserves to achieve rapid economic expansion, hence resulting in severe environmental instability, such as ecological deterioration, vast solid and commercial waste, and soil, water, and air quality issues (Rasoulinezhad and Saboori, 2018; Ahmad et al., 2019). The current study examined how urbanisation, human capital, and natural resources affected the environmental footprint and economic development in China from 1995 to 2020 The total resources rent as a percentage of GDP is a measure of natural resources (Taghizadeh-Hesary et al., 2021), including gas, oil, coal, minerals, and forest. The relationship between natural resources and environmental deterioration is a controversial issue. Economic expansion, urbanisation, and industrialisation trends raise the requirement for the exploitation and use of domestic resources, which impacts environmental sustainability (Yin et al., 2021). Meanwhile, a surplus of natural resources might discourage the use of fossil fuels by reducing imports. In Yousaf and Ali (2020) and Ahmad et al. (2019), the main contributors to natural habitat degradation and water, soil, and air pollution are human activities, such as mining, deforestation, and chainsaw usage. Studies on natural resource contribution to environmental sustainability presented similar conclusions. For example, Walther et al. (2019a) stated that using natural resources increases ecological footprint, while other studies revealed the contrary. Hence, more research is required to advance towards a sustainable environment considering the lack of studies on the relationship between natural resources and a broad ecological measure (environmental footprint). The study subject is of critical importance for China as recent economic growth has caused a surge in the exploitation of natural resources to meet the rising needs in manufacturing and other industries. In 2016, approximately 32% of resource rent in China came from coal followed by minerals (31%), oil (20%), forests (8%), and other natural resources (7%) (Ullah et al., 2021; Walther et al., 2019b). Apart from being the world's most significant coal producer, China consumes more fossil fuels than any nation. Coal production comprised approximately 62.8% of the overall national energy consumption in 2014 (Ahmad et al., 2019). China is among the world-leading economy and mining but every level of the mining process endangers the environment (Iqbal et al., 2021). According to Yarovaya et al. (2020), coal mining and urbanisation are major contributors to the deteriorating environment in China. Walther et al. (2019a) disclosed the vital role of mining towards water pollution, which harms human health and the ecosystem in China. Meanwhile, recent research discovered that the rapid consumption of natural resources and growing energy demand results from rapid economic expansion in China (Yasmeen et al., 2022). Reliance on foreign energy sources has thus increased from 1.4% in 1980 to 19.4% in 2014 (Iqbal et al., 2021). Energy security and the availability of natural resources are essential for national sustainable development (Umar et al., 2021a). The primary causes of energy scarcity are overemphasising economic expansion and irresponsible exploitation of natural resources. China demonstrated a 3.75% ecological deficit in 2015 for an environmental footprint (3.62 gha per capita), which is higher than the national biocapacity of 1.03 gha per person (Zhao et al., 2022). The environmental gap indicated that the demand for Chinese environmental products and services exceeds what the country ecosystem can provide. Due to persistent ecological deficit, declining biodiversity, deteriorating natural resources, depletion of ecological reserves, and environmental degradation due to a lack of available resources. Despite the environmental regulations in China, achieving sustainable use of natural resources remain complicated due to issues with administration, execution, and limited public engagement. Thus, the current topic is critical for China. The country also experiences a widespread bias towards prioritising economic values over ecological ones (Umar et al., 2021b). Total natural resource rents are considered synonymous with natural resource availability. Existing literature often neglects situations concerning differential rent amounts. Therefore, legislators must assess whether natural resource rents burden or benefit economic progress. The previous study is limited in scope. Studies usually consider industrialised countries when comparing the natural resource curse (NRC) with the financial resource curse (FRC). Nonetheless, natural resources hold a greater and significant role in the well-being of emerging economies. Limited studies have provided an in-depth analysis of the resource curse theory in emerging countries (Umar et al., 2021a; Dogan et al., 2020). Thus, evaluating the FRC of emerging markets is critical. The study significantly contributes to the literature. Two single-equation models were used to examine the effects of natural resource abundance, foreign trade, technology innovation, investment of energy resources, and urbanisation on economic recovery. The GMM method was used to examine 30 provinces in China from 1995 to 2020. This study is also the first to examine the relationship between financial success and the volatility of natural resources. Past research was used to provide empirical support. The study also contributes to the body of literature to address the lack of research investigating how investments in energy resources affect the volatility of natural resources. Therefore, the study provides evidence of the hypothesised relationship. China is developing technologically and plays a significant role in world commerce. Nonetheless, no research has addressed how these two factors affect the volatility of natural resources. The study also analysed the national resource tax contribution to economic growth in China. China is one of the top importers of natural resources for energy (including oil), which inspired the current investigation on the country. Price fluctuations in natural resources could significantly impact the industrial economy in China. Moreover, the COVID-19 outbreak has greatly disrupted most industrial and economic activities, which might substantially impact national financial performance. China is also a rising nation concentrating on its steady economic progress. As the first nation to experience the pandemic, the examination of the Chinese economy is crucial. The current study is also one of the first to assess the volatility of natural resource commodity prices, economic performance, and new policy insights. Few studies have discussed the topic despite the significance of investments in green finance and energy resources for financial performance. Therefore, the current study would attract academic and governor interest in evaluating these aspects for policy considerations. Thus, the study expanded the body of knowledge by analysing the aforementioned relationship and offering valuable and novel policy tools and approaches to mitigate the adverse effects on financial and economic indicators. The remainder of the paper is organised as follows: Section 2 contains the literature review, while Section 3 discusses the data and methodology. Results and discussion are explained in Section 4, while Section 5 presents the study conclusions and policy implications. 2 Literature review Empirical studies on the relationship between economic expansion and environmental degradation address a wide range of topics, such as resource depletion, pollution, and global warming. Climate change remains the modern world's biggest and most pressing concern that can harm people's health and the economy (Tu and Xue, 2019). Consequently, academics, intellectuals, and politicians have recently emphasised ecological deterioration and natural resources. Past studies highlighted ecological footprint (EF) to measure ecological degradation. Symitsi and Chalvatzis (2019) investigated the link between EF, natural resources, urbanisation, and human capital in China from 1972 to 2017. Observably, increasing human capital reduces environmental degradation, while the rent from natural resources, economic development, and urbanisation contribute to the EF level. Human capital and urbanisation interaction term also contribute to improving the environment. Shi et al. (2020) discovered that natural resources and economic development in China positively impacted the national EF from 1968 to 2015, thus suggesting that an over-reliance on natural resources negatively impacts environmental quality. Shahzad et al. (2021) examined China nations from 1993 to 2018 and noted the significance of urbanisation, renewable energy usage, and natural resources in determining the EF. Conclusively, the rent from natural resources, urbanisation, and renewable energy detrimentally impacted the EF of an area or country. The inverted U-shaped relationship between economic expansion and EF also confirmed the findings. A study analysed the effect of natural resources, foreign direct investment (FDI), and human capital on the EF in the United States of America (USA) between 1980 and 2014. Resultantly, the EF of a region is mitigated by its stock of human capital, FDI, and natural resources. The current study identified a clear link between the availability of natural resources and EF. Bouri et al. (2021) revealed that resource rent lowers emissions in China. Increasing economic activity and coal usage contribute to environmental degradation. Conversely, Omane-Adjepong and Alagidede (2019) examined the relationship between emissions, natural resources, wealth, and renewable energy in China using the SGMM technique and discovered a negligible influence of natural resources in the China panel. The EKC between wealth and emissions was negative, while clean energy negatively affected emissions. Some scholars investigated the link between natural assets and ecological imprints. According to research by, China benefits from energy consumption (EC) mitigation due to its natural resources. A similar correlation was revealed between EC and human capital but energy and money negatively impact EC. The connection between natural resources and EC in China was further highlighted in Pesaran et al. (2001). Furthermore, EC is reduced by using renewable energy and urbanisation, while EKC exists between income and EC in China nations. Human capital is a holistic term that considers the source of the wealth boost (whether innate or acquired skills). Health and knowledge are the two most effective forms of human capital in improving human output (Okorie and Lin, 2020). Human capital affects the generation of renewable energy by accepting new information and contributing work (Connor et al., 2015). Therefore, efficient administration of information and technology-intensive assets is especially crucial for renewable energy firms. Human capital is vital to organisational long-term success and expansion. Human capital may also synergistically affect energy usage with a negligible but meaningful socioeconomic influence. The dual benefits of human capital include economic expansion and beneficial externalities, such as better health and a cleaner atmosphere (Narayan, 2005). Endogenous growth models regard human capital as an alternative to technical advancement in manufacturing, which is a significant factor in the economic expansion (Kumar and Anandarao, 2019). Nonetheless, limited studies have emphasised the significance of human capital in achieving sustainable development, specifically regarding natural ecosystem pollution. Klein et al. (2018) questioned the direct link between financial progress and GDP expansion. Although other studies failed to identify a short-term causal relationship, a long-term connection was displayed between financial depth and economic expansion. Goodell and Goutte (2021) discovered a causal relationship between financial development and economic growth in most countries except China. The literature review outlined the need for a more in-depth discussion on whether a robust financial industry is necessary for sustained economic expansion. The mixed findings are due to the lower influence of financial variables on GDP growth. Furthermore, knowledge on the mechanisms where financial development changes cause agent behaviour shifts and other economic elements that contribute to expansion remains lacking. According to Karamti and Belhassine (2022), the direction of causality is sensitive to the choice of variables, specifically to the potential omission of a third significant factor that impacts economic growth and financial development. Moreover, the methodology used to investigate the link between financial development and economic growth varies based on the country and over time. The current study investigated whether the relationship between financial intermediaries and GDP expansion is more nuanced than a standard bivariate model, which varies considerably based on country. Specifically, the significance of the oil industry on the Chinese economy was investigated using a three-variable model involving multiple connections among the financial sector, the oil sector, and the non-oil sector. 3 Methodology 3.1 Data A total of 30 Chinese provinces were chosen as study samples but Tibet, Hong Kong, Macao, and Taiwan were excluded following insufficient data. The sample data were extracted from the Provincial Statistical Yearbook, China Energy Statistical Yearbook, and China Environmental Statistical Yearbook. The federal government requires local governments to publish pertinent energy intensity metrics since 2005. Hence, the study used data from 1995 to 2020. The non-ratio variables were processed logarithmically to address the heteroscedasticity issue. Table 1 displays the descriptive statistics for each variable.Table 1 Description of variables and sources. Table 1Variable name Data source Description of variables Dependent Natural Resources Rent (TRNT) Provincial Statistical Yearbook REC per capita Gross Domestic Product (GDP) Provincial Statistical Yearbook US dollar per 1000 Independent GDP US dollar per 1000 Foreign Trade (FT) Provincial Statistical Yearbook The percentage of exports & imports over GDP measures trade openness Investment in Energy Resources (IER) China Energy Statistical Yearbook Index of fuel production, power generation, energy infrastructure, transport Technology Innovation (TI) Provincial Statistical Yearbook Measure as the percentage of GDP Urbanisation (URB) China Environmental Statistical Yearbook URB index 3.2 Theoretical framework Company profitability is the phenomenon under investigation. In place of economic indicators, such as Tobin's Q or share prices, accountants use IER (return on assets) and GDP (return on equity), which calculate profits as a percentage of average equity. Managers and shareholders may use market-based measurements to their advantage (Berner et al., 2022). According to Frondel et al. (2022), accounting-based indicators may provide a more accurate portrayal of organizational information. Studies of Chinese clean energy businesses have frequently used URB and TI (Salisu et al., 2022). In China, digital currency is assessed using two different approaches. First, the digital banking index is built by analyzing the frequency of digital finance-related terms in the Yahoo news database using a text processing approach (Aweke and Navrud, 2022). The strategy involves two aspects: No objective criteria for selecting keywords, hence the ranking can only be built at the national level (Selmi et al., 2022). The second is (Azancot et al., 2022). Or the Philippine Normal university Financial Technology Accessibility Index. Information from 1995 to 2020 may be accessed from municipal and regional government levels. Digital financial research in China has extensively used the index since its inception (Silvestroni et al., 2022). The current study utilized the URB index using the data at the prefectural level. Generally, no academic research has provided a fixed standard for assessing financial hardship. The budget limit is quantified using the TRNT, GDP, and FT based on Farkas et al. (2022)4 IER employ the analysis based on the interest coverage ratio to measure stability. A higher TI or URB index for a specific company indicates a more constrained net profit. 3.3 Control variables The study parameters were divided into three broad categories: business demographics, and corporate governance structure. The natural log of total assets (RMB) determines a company size (Size). The state control marker was set to 1 when the state regulates a company and 0 when it is not. Economic governance structure factors include top share (Yasmeen et al., 2022), board size, and FT duality (Zhao et al., 2022). In order to account for the impact of the classic financial sector, Finder defines regional capital formation as the ratio of banking capital market deposits and lending to province GDP. Lastly, a group of time dummies was provided to account for year impacts as listed in Equation (1):(1) yit=αyit−1+Xit′TRNT+Xit′GDP+Xit′FI+Xit′IER+Xit′TI+Xit′URB+βZit′σ+ui+εit Logit modelling and a matching test were employed to determine how well GMM had balanced and control groups. Equation (1) presents that all six variables were statistically significant, hence demonstrating that the individuals were chosen adequately as covariates. Long run and short run coefficients are represented in Equation (1) by α and yit−1, respectively. The short-run and long-run analyses evaluated the disequilibrium and the rate at which the equilibrium level will return concerning the immediate impact of independent factors shock on the explanatory variables. The optimal lags for the explanatory and multiple regression were β and Zit−1. 4 Results and discussion Nations with high fossil fuel usage and poor economic growth tend to participate in the China Initiative (Jareño et al., 2019). Considering that most of China comprise emerging nations struggling with excessive energy usage and pollution, assessing the China Initiative ecological and energy impacts is critical. In Table 2 , the average EC is 0.046 and G is 0.099. Meanwhile, the financial innovation in China depicts considerable variation where the mean value of TI was 1.818 and the error margin was 0.643. The mean of 0.988 and the measure of the dispersion of 1.485 indicate that the degree to which resources are stretched may vary considerably (Tapia et al., 2022). The descriptive statistics are listed in Table 2. The findings suggested (Sun et al., 2022) no evidence of multivariate regression due to the low degree of relationship between the independent components.Table 2 Descriptive statistics. Table 2Variable N Mean SD Min Medium Max TRNT 927 0.048 0.047 −0.065 0.876 0.487 GDP 927 0.095 0.087 −0.098 0.098 0.876 FT 927 1.819 0.645 0.287 1.887 4.087 IER 927 0.986 1.798 −21.554 1.254 6.998 TI 927 23.177 1.454 20.287 22.760 28.045 URB 927 18.765 5.098 4.001 18.000 37.000 Note: TRNT = Natural resources rent, GDP = Economic recovery, FT = Foreign Trade, IER = Investment in energy resources, TI = Technology Innovation, URB = Urbanisation. The study determined which factors to be used for the analysis followed by measuring propensity scores using relevant variables. Meanwhile, matching was performed to reduce sample heterogeneity and improve the GMM estimator precision. Table 2 indicates that the GMM technique successfully eliminated the pre-existing discrepancies between the control and treatment groups upon completing the matching process. The overall community (Kalyuzhnova and Lee, 2020) bias among the control and treatment groups was approximately 18.6% before matching, which decreased to 4.8% after matching, hence reflecting a 73.3% decrease in bias. Hsu et al. (2021) stated that the matching criteria are satisfied if the bias is under 20% points after completing the process. Additionally, no significant difference exists between the control and therapy groups in terms of the biases of the factors used for matching, thus indicating a successful balancing test. Guo et al. (2021) examined how the propensity scores for the control and therapy groups are distributed differently, which provided more insight into the balancing test outcomes. The kernel density of the propensity scores for the control and therapy groups was relatively close after a matched sample. Therefore, the abovementioned findings validated the GMM approach efficacy as depicted in the correlation analysis in Table 3 .Table 3 Correlation analysis. Table 3Variable 1 2 3 4 5 6 TRNT 1.000 GDP 0.876* 1.000 FT −0.076 −0.098 1.000 IER −0.598* −0.387* 0.007 1.000 TI −0.008 0.187* 0.287* 0.069 1.000 URB 0.098 0.098 0.436* 0.069 0.190 1.000 Table 3 displays the findings of how the China Initiative affects energy consumption and pollution levels in the natural ecosystem. Results from conserving energy and lowering carbon emissions are reported based on González et al.'s (2021) models, which demonstrate the full scope of the policy impacts. The GMM technique was employed to build models (1)–(2) and (5)–(6) without matching the therapy and control groups. Although the direction of influence aligned with that of GMM (de la O González et al., 2020) models, the magnitude of the effect and the size of the coefficients were notably lower due to the indigeneity issue caused by the selection bias in the samples in the control pool. The situation concealed the true extent to which the China (González et al., 2020) Initiative was responsible for increased energy consumption and pollution. The statistic model in Table 4 was developed with and without the control variables to demonstrate the robustness of the findings. Meanwhile, the policy impact remained essentially the same in both cases.Table 4 Statistic model results. Table 4Empty Cell TRNT TRNT TRNT GDP GDP GDP TRNT 0.087* 0.024 0.028 0.194* 0.093 0.078* (0.087) (0.080) (0.080) (0.028) (0.028) (0.096) GDP −0.043* −0.071* −0.091* −0.018* −0.015* −0.004* (0.003) (0.006) (0.008) (0.008) (0.009) (0.007) FT 0.007 0.009 (0.007) (0.007) IER 0.007 0.004 0.087* 0.018* (0.008) (0.009) (0.008) (0.007) TI 0.007 0.007 0.007 0.008 (0.006) (0.008) (0.008) (0.009) Year fixed effect Yes Yes Yes Yes Yes Yes Province fixed effect Yes Yes Yes Yes Yes Yes Industry fixed effect Yes Yes Yes Yes Yes Yes N 933 843 887 965 8476 910 The results implied that autocratic territories with well-functioning bureaucracies and democratic regions with poor-quality bureaucracies (Ghazani and Khosravi, 2020) benefit from oil and gas money, thus stimulating economic expansion (Diebold and Yilmaz, 2009). Interestingly, the role of potential outliers revealed that the second finding is driven by a single observation: the Tyumen area. Removing Tyumen from the data set diminished the reported impact for governments with low-quality bureaucracies. The results remained when examining the marginal impacts for triple relationship terms or the four subsamples alone (Diebold and Yilmaz, 2011). Conversely, the current findings were similar even after the regressions were re-estimated, thus eliminating all areas with below-average Carnegie Centre scores of democracy and above-average Opera indices of the bureaucracy quality. Moreover, sufficient observations of resource-rich territories exist within the subgroup of non-democratic regions with efficient bureaucracy to rule out the possibility that the findings are driven only by a few outliers (Demir et al., 2020). The findings for democracies with poor administration quality proved negligible after re-estimating impacts for the entire sample and omitting Tyumen. Table 5 demonstrates that the results for non-democracies with high administration quality remained strong.Table 5 Dynamic model results and robustness test results with R&D controlled. Table 5Empty Cell TRNT TRNT GDP GDP TRNT TRNT GDP GDP TRNT 0.398* 0.409* 0.358* 0.387* (0.097) (0.082) (0.078) (0.098) GDP 0.477* 0.498* 0.487* 0.498* (0.109) (0.190) (0.080) (0.981) FT 0.087 0.018** 0.029* 0.090* 0.070 0.018 0.087 0.076 (0.009) (0.008) (0.092) (0.095) (0.090) (0.009) (0.092) (0.009) IER −0.008* −0.009* −0.004* −0.008 −0.018* * −0.019 −0.008* (0.008) (0.009) (0.007) (0.009) (0.009) (0.009) (0.009) (0.009) TI −0.009 −0.018 −0.009 −0.019 (0.007) (0.008) (0.007) (0.009) Control Variables Controlled Controlled Controlled Controlled Controlled Controlled Controlled Controlled R&D controlled NO NO NO NO Yes Yes Yes Yes Firm fixed effect Yes Yes Yes Yes Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes Yes Yes Yes Yes AR (1) −3.68 −3.98 −3.39 −4.37 −2.78 −3.50 −1.88 −2.88 [0.00] [0.008] [0.016] [0.090] [0.008] [0.087] [0.049] [0.059] AR (2) −1.58 −1.49 −1.19 −1.29 −2.08 −1.98 −0.59 −0.58 [0.182] [0.192] [0.299] [0.289] [0.048] [0.098] [0.587] [0.680] Hansen test 81.65 90.48 58.29 78.58 90.48 88.72 67.717 103.18 [0.187] [0.396] [0.487] [0.676] [0.199] [0.498] [0.198] [0.399] N 687 687 687 687 687 687 687 687 Tyumen is a unique territory in China due to having the most oil and gas reserves, which has garnered significant interest from federal, state, and federal business organisations. Second, measuring TRNT, GDP, FT, IER, and TI is a complex process as Tyumen is where most oil and gas is extracted in the independent areas (Das et al., 2020). Most literature on growth regression assessments in China Federation (Corbet et al., 2020a, Corbet et al., 2020b) has omitted Tyumen as an outlier. Nonetheless, Table 6 displays that the impact of oil and gas on economic development in non-democratic areas with high-quality administration is highly resilient and not uncommon.Table 6 Robustness test results with the alternative financial constraint index. Table 6 TRNT TRNT GDP GDP TRNT TRNT GDP GDP TRNT 0.399* 0.380* 0.376* 0.387* (0.108) (0.096) (0.095) (0.098) GDP 0.469* 0.495* 0.498* 0.485* (0.086) (0.082) (0.181) (0.800) FT 0.087 0.019 0.027* 0.076 0.078 0.098 0.098 0.090 (0.00) (0.090) (0.018) (0.098) (0.098) (0.001) (0.083) (0.018) IER −0.009* −0.009*** −0.006* −0.007 −0.009* −0.009* −0.009 −0.007 (0.009) (0.009) (0.006) (0.009) (0.009) (0.008) (0.008) (0.009) TI −0.009* −0.076* −0.009* −0.018* (0.006) (0.009) (0.008) (0.008) Control Variables Controlled Controlled Controlled Controlled Controlled Controlled Controlled Controlled RD controlled NO NO NO NO Yes Yes Yes Yes Firm fixed effect Yes Yes Yes Yes Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes Yes Yes Yes Yes AR (1) −2.46 −2.61 −2.18 −2.49 −2.06 −2.32 −1.98 −2.09 [0.087] [0.007] [0.081] [0.017] [0.049] [0.000] [0.058] [0.092] AR (2) −1.19 −1.18 −1.27 −1.28 −1.18 −0.77 −0.58 −0.48 [0.250] [0.287] [0.288] [0.206] [0.292] [0.485] [0.696] [0.675] Hansen test 62.58 76.08 75.99 90.66 94.78 90.00 94.18 103.00 [0.286] [0.497] [0.487] [0.272] [0.27] [0.589] [0.399] [0.389] N 676 677 679 677 558 557 558 558 The bound's testing method was performed in the second stage to determine the long-run connection. The critical values were compared by performing bound tests, which were evaluated using the GMM method and calculating the F-statistics for the combined importance of the lagged levels variables (Corbet et al., 2020a, Corbet et al., 2020b). The heterogeneity analysis of the firm scale was significantly larger than the aforementioned critical boundaries (see Table 7 ).Table 7 Heterogeneity analysis of the firm scale. Table 7 TRNT GDP TRNT GDP TRNT 0.399* 0.409* (0.087) (0.079) GDP 0.985* 0.489* (0.098) (0.096) FT 0.080* 0.076 0.018* 0.028* (0.009) (0.019) (0.007) (0.094) IER −0.008* −0.098 −0.009 −0.007 (0.009) (0.008) (0.009) (0.009) TI −0.009 −0.006* (0.009) (0.009) URB −0.007* −0.018 (0.009) (0.009) Control variables controlled controlled controlled controlled Firm fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes AR (1) −2.79 −2.37 −38 −2.79 [0.009] [0.019] [0.0787] [0.009] AR (2) −1.67 −1.08 −1.37 −0.98 [0.099] [0.309] [0.143] [0.391] Hansen test 97.08 90.08 80.87 68.68 [0.186] [0.447] [0.387] [0.587] N 677 679 678 679 Two-way causation exists between growing economies and polluting ecosystems. Therefore, determining the link between TRNT, GDP, FT, and IER and the effect of CO2 emission on economic development is critical. Table 8 presents the findings of the heterogeneity analysis. The F-statistic and the associated probability values were derived from the null hypothesis of no causation (Charfeddine et al., 2020). The results indicated a link between economic growth and carbon dioxide emissions. Thus, the importance of this directionality cannot be overstated.Table 8 Heterogeneity analysis of ownership. Table 8Empty Cell TRNT GDP TRNT GDP TRNT 0.381* 0.390* (0.097) (0.098) GDP 0.495* 0.492* (0.108) (0.098) FT 0.008 0.029 0.095* 0.098 (0.007) (0.019) (0.009) (0.098) IER −0.008* −0.018** −0.008* −0.018 (0.009) (0.007) (0.009) (0.008) TI −0.008* −0.016* (0.007) (0.009) URB −0.008 −0.009 (0.008) (0.010) Control variables controlled controlled controlled controlled Firm fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes AR (1) −3.79 −4.78 −3.39 −1.65 [0.009] [0.007] [0.080] [0.081] AR (2) −1.78 −1.48 −1.28 −1.18 [0.080] [0.197] [0.213] [0.281] Hansen test 70.08 90.56 70.24 78.13 [0.186] [0.391] [0.411] [0.492] N 679 679 678 678 The various dependent variables approach in Bouri et al., 2017a, Bouri et al., 2017b was applied to ensure the validity of the causality test. The results for TRNT, TI, and IER indicated Integration when using financial development indicators as dependent variables. Therefore, the economic growth in China could be a long-term driving (Bouri et al., 2017a, Bouri et al., 2017b) variable of ecological degradation. National ecological efficiency may also affect the rate of economic growth. The study thus utilized the GDP growth rate per person as the dependent variable as depicted in Table 9 .Table 9 Heterogeneity analysis of region. Table 9 TRNT TRNT GDP GDP 0.387* 0.388* (0.054) (0.080) GDP 0.489* 0.498* (0.108) (0.194) FT 0.019* 0.018 0.018 0.019 (0.011) (0.097) (0.086) (0.095) IER −0.008* −0.019* −0.029* −0.085*** (0.008) (0.007) (0.010) (0.085) TI −0.009 −0.009 (0.009) (0.007) URB 0.081⁎ 0.007 (0.022) (0.009) Control variables controlled controlled controlled controlled Firm fixed effect Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes AR (1) −2.78 −2.68 −2.38 −2.39 [0.007] [0.008] [0.019] [0.080] AR (2) −1.67 −0.87 −1.17 −1.07 [0.081] [0.386] [0.299] [0.278] Hansen test 74.38 19.98 60.09 65.37 [0.186] [0.097] [0.495] [0.687] N 678 687 678 677 Table 3 depicts the first four model outputs from 1953 to 2006. The harmful factor are statistically essential, hence proving that better environmental performance is associated with rising levels of financial growth in China. Notably, TRNT, GDP, and TI displayed a magnitude of −0.447. Therefore, environmental pollution increases by 0.447% for every 1% growth in economic development, which is consistent with Bouri et al. (2021). A value of e of 0.323 is noteworthy. In the long term, every 1% increase in energy usage per person will increase CO2 emissions by 0.323%. The y-coefficient was 1.216, which was significantly higher than random. Therefore, a 1% increase in a real per person's wealth increases CO2 emissions by 3.157%. Energy usage predictions in the long-run pollution equation are optimistic and statistically crucial in all circumstances. Canh et al. (2019) and other studies verified that energy use has a favourable impact. The findings on energy consumption and CO2 emissions in China are also consistent with Bouri et al. (2020). Carbon emissions in China are mainly driven by wealth and energy usage. Hence, the findings align with the current research. Ten models were generated using data from 1978 to 2006 regarding the three measures of monetary growth. Specifically, the study noted a statistically significant and harmful factor for the case of private, therefore suggesting a harmful connection between environmental pollution and economic development. Correspondingly, Beneki et al. (2019) mentioned that private companies would avoid polluting sectors if they gain easier access to capital. Table 9 reveals a negative long-term relationship between human assets and energy use. The Chinese government has emphasised certain issues, such as resource and environmental preservation and clean and efficient energy use. The economy has shifted the focus from rapid expansion to quality growth. The value of human capital has grown and the quality of human capital has advanced significantly over time. The results were consistent with Baur and Lucey (2010) where human capital results in decreased energy use. Moreover, Baker et al. (2020) noted that increasing levels of human capital benefit energy use. The findings disclosed a strong positive causal link between energy use and economic expansion. Specifically, a rise of 1% in economic growth accounts for a 1.419% increase in energy usage. Nevertheless, energy efficiency demonstrates how capitalisation negatively impacts energy consumption, which can be mitigated with improved practices. Suppose everything else holds constant (Akram, 2020), a 1% increase in capital results in a 0.2641% reduction in energy use. Importing power increased overall energy use significantly. Rising energy imports were mainly responsible for the 0.0604% rise in energy usage. The findings revealed a statistically significant negative relationship between R&D spending and energy use. Thus, a 1% increase in R&D spending results in a 0.15% drop in energy use. Education and experience are less significant in improper energy use, as measured by the demand function. In the unclean energy model, the coefficient of human capital causes remained at 0.6907%, hence suggesting that lowering energy usage increases human capital. Using fossil fuels has a significant, harmful effect on GDP growth. A 1% rise in GDP, the factor (1.3308) predicts a 3% drop in dirty energy usage (Maciej Serda, 2013). A strong inverse correlation exists between financial resources and the use of filthy forms of energy. A 1% increase in capitalisation is associated with a 0.228% decrease in dirty energy usage, while everything else was equal (Adekoya and Oliyide, 2021). The calculations indicated that a 1% increase in imported energy results in a 0.0578% rise in filthy energy consumption. Meanwhile, destructive energy use was negatively correlated with research and development budgets. A 1% rise in spending on research and development leads to a 0.1557% drop in the use of dirty energy. The need for pollution management of energy usage has become essential, which is in line with the economic strategy in China following economic growth and rising human asset level. Increases in economic development, human (Jareño et al., 2021) capital, R&D spending, and imports provide favourable environments to lower polluting energy usage. Government legislation has facilitated reducing the use of polluting energy. 5 Conclusion The study examined the relationship between the volatility of natural resources and macroeconomic factors, such as economic performance, investment in energy resources, international commerce, and technological innovation. The volatility of natural resources is the current trend. The GMM approach was used in the present research to examine 30 Chinese provinces between 1995 and 2020. The substantial reliance on natural resource revenue in the area creates structural barriers to attaining sustainable development objectives. Several multinational corporations with little to no interest in local development have dominated the exploitation of natural resources. Although natural resources may detrimentally affect economic advancement, the impact can be mitigated by the limited ability to create productive links and the low building of a constructive network with local businesses. The considerable price fluctuation of the commodities exported from the area lessens the overall impact of natural richness on economic development. The natural resources utilized in the area are oriented towards the global market, which suggests a moderating influence on establishing commercial blocs at the intersection of the two variables. Additionally, the erratic nature of raw material pricing complicates preparing for and predicting when the advantages of natural riches will materialize. The conclusions provide several possible political repercussions. Effective rent collection from natural resources depends on strong institutions that prevent corruption by preserving the rule of law, good governance, and property rights. Monitoring mechanisms should be established to minimize trade imbalances and enhance the influence of global commerce on economic growth. Absorptive capacity can be enhanced through human capital development, R&D, institutional quality improvement, and financial growth. Stock markets within nations should be improved due to being a more effective instrument in contributing to financial development towards economic progress. These strategies enable countries to benefit from the technical gains of international commerce and trade openness. The measures could also increase human development by promoting economic growth. Therefore, economic growth should be enhanced with caution by considering the interaction between various elements. The study provided opportunities for further investigation to strengthen the body of literature:1. Many studies tend to concentrate on two or three factors. Hence, the current research contributed by econometrically analyzing the impact of the five variables on economic growth in two single-equation models. 2. A wider understanding of economic development was investigated by utilizing a human development index in addition to conventional economic growth as a proxy for economic development. 3. Using the trade balance as a proportion of GDP to represent international commerce as a proxy was another contribution as the trade share metric was used. Given the emphasis on the economic growth proxy, research should examine a wider view of economic development. Trade balance could be used as a proxy for global trade as a proportion of GDP to provide different outcomes. Research should include other factors, such as poverty and unemployment rates to assess the impact on human development in light of the small coefficients in the human development model. Additional data could enable future studies to include low-income nations and provide updated information. Credit author statement Mohammad Maruf Hasan: Conceptualization, Methodology, Software, Data Creation, and Writing-original draft preparation. Fang Du (F. Du): Conceptualization, Methodology, Software, and Data Creation, Revision of manuscript Data availability Data will be made available on request. Acknowledgement This study supported by the Institute of South Asian Studies at Sichuan University, a Key Research Centre for Humanities and Social Sciences at Universities. ==== Refs References Adekoya O.B. Oliyide J.A. How COVID-19 drives connectedness among commodity and financial markets: evidence from TVP-VAR and causality-in-quantiles techniques Resour. Pol. 70 2021 10.1016/J.RESOURPOL.2020.101898 Ahmad M. Beddu S. binti Itam Z. Alanimi F.B.I. State of the art compendium of macro and micro energies Adv. Sci. Technol. Res. J. 13 1 2019 88 109 10.12913/22998624/103425 Akram Q.F. Oil price drivers, geopolitical uncertainty and oil exporters' currencies Energy Econ. 89 2020 10.1016/J.ENECO.2020.104801 Aweke A.T. Navrud S. Valuing energy poverty costs: household welfare loss from electricity blackouts in developing countries Energy Econ. 109 2022 10.1016/j.eneco.2022.105943 Azancot L. Bobadilla L.F. Centeno M.A. Odriozola J.A. Catalytic reforming of model biomass-derived producer gas Fuel 320 2022 10.1016/j.fuel.2022.123843 Baker S.R. Bloom N. Davis S.J. Kost K. Sammon M. Viratyosin T. The unprecedented stock market reaction to COVID-19 Rev. Asset Pricing Stud. 10 4 2020 742 758 Baur D.G. Lucey B.M. Is gold a hedge or a safe haven? An analysis of stocks, bonds and gold Financ. Rev. 45 2 2010 217 229 10.1111/J.1540-6288.2010.00244.X Beneki C. Koulis A. Kyriazis N.A. Papadamou S. Investigating volatility transmission and hedging properties between Bitcoin and Ethereum Res. Int. Bus. Finance 48 2019 219 227 10.1016/j.ribaf.2019.01.001 Berner A. Lange S. Silbersdorff A. Firm-level energy rebound effects and relative efficiency in the German manufacturing sector Energy Econ. 109 2022 10.1016/j.eneco.2022.105903 Bouri E. Gupta R. Tiwari A.K. Roubaud D. Does Bitcoin hedge global uncertainty? Evidence from wavelet-based quantile-in-quantile regressions Finance Res. Lett. 23 2017 87 95 10.1016/j.frl.2017.02.009 Bouri E. Molnár P. Azzi G. Roubaud D. Hagfors L.I. On the hedge and safe haven properties of Bitcoin: is it really more than a diversifier? Finance Res. Lett. 20 2017 192 198 10.1016/j.frl.2016.09.025 Bouri E. Shahzad S.J.H. Roubaud D. Cryptocurrencies as hedges and safe-havens for US equity sectors Q. Rev. Econ. Finance 75 2020 294 307 10.1016/j.qref.2019.05.001/ Bouri E. Cepni O. Gabauer D. Gupta R. Return connectedness across asset classes around the COVID-19 outbreak Int. Rev. Financ. Anal. 73 2021 10.1016/J.IRFA.2020.101646 Canh N.P. Binh N.Q. Thanh S.D. Cryptocurrencies and investment diversification: empirical evidence from seven largest cryptocurrencies Theor. Econ. Lett. 9 3 2019 431 452 10.4236/TEL.2019.93031 Charfeddine L. Benlagha N. Maouchi Y. Investigating the dynamic relationship between cryptocurrencies and conventional assets: implications for financial investors Econ. Modell. 85 2020 198 217 10.1016/j.econmod.2019.05.016 Corbet S. Hou Y. Greg Hu Y. Larkin C. Oxley L. Any port in a storm: cryptocurrency safe-havens during the COVID-19 pandemic Econ. Lett. 194 2020 10.1016/J.ECONLET.2020.109377 Corbet S. Larkin C. Lucey B. The contagion effects of the COVID-19 pandemic: evidence from gold and cryptocurrencies Finance Res. Lett. 2020 10.1016/j.frl.2020.101554 Das D. Le Roux C.L. Jana R.K. Dutta A. Does Bitcoin hedge crude oil implied volatility and structural shocks? A comparison with gold, commodity and the US Dollar Finance Res. Lett. 36 2020 10.1016/J.FRL.2019.101335 de la O González M. Jareño F. Skinner F.S. Nonlinear autoregressive distributed lag approach: an application on the connectedness between bitcoin returns and the other ten most relevant cryptocurrency returns Mathematics 8 5 2020 10.3390/MATH8050810 Demir E. Simonyan S. García-Gómez C.D. Lau C.K.M. The asymmetric effect of bitcoin on altcoins: evidence from the nonlinear autoregressive distributed lag (NARDL) model Finance Res. Lett. 2020 10.1016/J.FRL.2020.101754 Diebold F.X. Yilmaz K. Measuring financial asset return and volatility spillovers, with application to global equity markets Econ. J. 2009 10.1111/j.1468-0297.2008.02208.x Diebold F.X. Yilmaz K. Measuring financial asset return and volatility spillovers, with application to global equity markets SSRN Electron. J. 2011 10.2139/SSRN.956918 Dogan E. Altinoz B. Tzeremes P. The analysis of ‘Financial Resource Curse’hypothesis for developed countries: evidence from asymmetric effects with quantile regression Resour. Pol. 68 2020 101773 10.1016/j.resourpol.2020.101773 Farkas A. Degiuli N. Martić I. Barbarić M. Guzović Z. The impact of biofilm on marine current turbine performance Renew. Energy 190 2022 584 595 10.1016/j.renene.2022.03.134 Frondel M. Kaeding M. Sommer S. Market premia for renewables in Germany: the effect on electricity prices Energy Econ. 109 2022 105874 10.1016/j.eneco.2022.105874 Ghazani M.M. Khosravi R. Multifractal detrended cross-correlation analysis on benchmark cryptocurrencies and crude oil prices Phys. Stat. Mech. Appl. 2020 560 10.1016/j.physa.2020.125172 González M. de la O. Jareño F. Skinner F.S. Portfolio effects of cryptocurrencies during the COVID-19 crisis A New World Post COVID-19 2020 10.30687/978-88-6969-442-4/011 González M. de la O. Jareño F. Skinner F.S. Asymmetric interdependencies between large capital cryptocurrency and Gold returns during the COVID-19 pandemic crisis Int. Rev. Financ. Anal. 76 2021 10.1016/j.irfa.2021.101773 Goodell J.W. Goutte S. Diversifying equity with cryptocurrencies during COVID-19 Int. Rev. Financ. Anal. 76 2021 10.1016/J.IRFA.2021.101781 Guo X. Lu F. Wei Y. Capture the contagion network of bitcoin - evidence from pre and mid COVID-19 Res. Int. Bus. Finance 58 2021 10.1016/J.RIBAF.2021.101484 Hsu S.H. Sheu C. Yoon J. Risk spillovers between cryptocurrencies and traditional currencies and gold under different global economic conditions N. Am. J. Econ. Finance 57 2021 10.1016/J.NAJEF.2021.101443 Iqbal N. Fareed Z. Wan G. Shahzad F. Asymmetric nexus between COVID-19 outbreak in the world and cryptocurrency market Int. Rev. Financ. Anal. 73 2021 10.1016/J.IRFA.2020.101613 Jareño F. Tolentino M. González M. de la O. Oliver A. Impact of changes in the level, slope and curvature of interest rates on U.S. sector returns: an asymmetric nonlinear cointegration approach Economic Research-Ekonomska Istrazivanja 32 1 2019 1275 1297 10.1080/1331677X.2019.1632726 Jareño F. González M. de la O. López R. Ramos A.R. Cryptocurrencies and oil price shocks: a NARDL analysis in the COVID-19 pandemic Resour. Pol. 74 2021 10.1016/j.resourpol.2021.102281 Kalyuzhnova Y. Lee J. Will COVID-19 Change Oil Markets Forever? A New World Post COVID-19 2020 10.30687/978-88-6969-442-4/012 Karamti C. Belhassine O. COVID-19 pandemic waves and global financial markets: evidence from wavelet coherence analysis Finance Res. Lett. 45 2022 10.1016/J.FRL.2021.102136 Klein T. Pham Thu H. Walther T. Bitcoin is not the New Gold - a comparison of volatility, correlation, and portfolio performance Int. Rev. Financ. Anal. 59 2018 105 116 10.1016/j.irfa.2018.07.010 Kumar A.S. Anandarao S. Volatility spillover in crypto-currency markets: some evidences from GARCH and wavelet analysis Phys. Stat. Mech. Appl. 524 2019 448 458 10.1016/j.physa.2019.04.154 Narayan P.K. The saving and investment nexus for China: evidence from cointegration tests Appl. Econ. 37 17 2005 1979 1990 10.1080/00036840500278103 Okorie D.I. Lin B. Crude oil price and cryptocurrencies: evidence of volatility connectedness and hedging strategy Energy Econ. 87 2020 10.1016/j.eneco.2020.104703 Omane-Adjepong M. Alagidede I.P. Multiresolution analysis and spillovers of major cryptocurrency markets Res. Int. Bus. Finance 49 2019 191 206 10.1016/j.ribaf.2019.03.003 O'Connor F.A. Lucey B.M. Batten J.A. Baur D.G. The financial economics of gold - a survey Int. Rev. Financ. Anal. 41 2015 186 205 10.1016/j.irfa.2015.07.005 Pesaran M.H. Shin Y. Smith R.J. Bounds testing approaches to the analysis of level relationships J. Appl. Econom. 2001 10.1002/jae.616 Rasoulinezhad E. Environmental impact assessment analysis in the kahak's wind farm J. Environ. Assess. Pol. Manag. 22 1n02 2020 2250006 10.1142/S1464333222500065 Rasoulinezhad E. Saboori B. Panel estimation for renewable and non-renewable energy consumption, economic growth, CO2 emissions, the composite trade intensity, and financial openness of the commonwealth of independent states Environ. Sci. Pollut. Control Ser. 25 18 2018 17354 17370 10.1007/s11356-018-1827-3 Saboori B. Rasoulinezhad E. Sung J. The nexus of oil consumption, CO2 emissions and economic growth in China, Japan and South Korea Environ. Sci. Pollut. Control Ser. 24 8 2017 7436 7455 10.1007/s11356-017-8428-4 Salisu A.A. Olaniran A. Tchankam J.P. Oil tail risk and the tail risk of the US Dollar exchange rates Energy Econ. 109 2022 10.1016/j.eneco.2022.105960 Selmi R. Hammoudeh S. Kasmaoui K. Sousa R.M. Errami Y. The dual shocks of the COVID-19 and the oil price collapse: a spark or a setback for the circular economy? Energy Econ. 109 2022 105913 10.1016/J.ENECO.2022.105913 Serda Maciej Synteza i aktywność biologiczna nowych analogów tiosemikarbazonowych chelatorów żelaza Uniwersytet Śląski 2013 343 354 10.2/JQUERY.MIN.JS Shahzad A. Hassan S. Yusuff Aremu A. Examine factors affecting success of enterprise resource planning system Adoption on organization performance among medium-sized enterprises sector Int. J. Bus. Syst. Res. 15 2 2021 1 10.1504/IJBSR.2021.10026434 Shi Y. Tiwari A.K. Gozgor G. Lu Z. Correlations among cryptocurrencies: evidence from multivariate factor stochastic volatility model Res. Int. Bus. Finance 53 2020 10.1016/j.ribaf.2020.101231 Silvestroni L. Capiani C. Sciti D. Sani E. Coloring zirconium oxide for novel energy saving industrial applications Renew. Energy 190 2022 223 231 10.1016/j.renene.2022.03.083 Sun C. Khan A. Liu Y. Lei N. An analysis of the impact of fiscal and monetary policy fluctuations on the disaggregated level renewable energy generation in the G7 countries Renew. Energy 189 2022 1154 1165 10.1016/j.renene.2022.03.027 Symitsi E. Chalvatzis K.J. The economic value of Bitcoin: a portfolio analysis of currencies, gold, oil and stocks Res. Int. Bus. Finance 48 2019 97 110 10.1016/j.ribaf.2018.12.001 Taghizadeh-Hesary F. Rasoulinezhad E. Shahbaz M. Vinh Vo X. How energy transition and power consumption are related in Asian economies with different income levels? Energy 237 2021 121595 10.1016/j.energy.2021.121595 Tapia M. Heinemann D. Ballari D. Zondervan E. Spatio-temporal characterization of long-term solar resource using spatial functional data analysis: understanding the variability and complementarity of global horizontal irradiance in Ecuador Renew. Energy 189 2022 1176 1193 10.1016/j.renene.2022.03.049 Tu Z. Xue C. Effect of bifurcation on the interaction between bitcoin and litecoin Finance Res. Lett. 31 2019 382 385 10.1016/J.FRL.2018.12.010 Ullah S. Khan U. Rahman K.U. Ullah A. Problems and benefits of the China-Pakistan economic corridor (cpec) for local people in Pakistan: a critical review Asian Perspect. 45 4 2021 861 876 10.1353/apr.2021.0036 Umar Z. Jareño F. Escribano A. Static and dynamic connectedness between oil price shocks and Spanish equities: a sector analysis Eur. J. Finance 27 9 2021 880 896 10.1080/1351847X.2020.1854809 Umar Z. Jareño F. Escribano A. Agricultural commodity markets and oil prices: an analysis of the dynamic return and volatility connectedness Resour. Pol. 73 2021 10.1016/j.resourpol.2021.102147 Walther F. Koerver R. Fuchs T. Ohno S. Sann J. Rohnke M. Zeier W.G. Janek J. Characterization of battery material interfaces by time-of-flight secondary ion mass spectrometry ECS Meeting Abstracts MA2019–04 2 2019 10.1149/MA2019-04/2/101 101–101 Walther T. Klein T. Bouri E. Exogenous drivers of Bitcoin and Cryptocurrency volatility - a mixed data sampling approach to forecasting J. Int. Financ. Mark. Inst. Money 63 2019 10.1016/J.INTFIN.2019.101133 Yarovaya L. Brzeszczynski J. Goodell J.W. Lucey B.M. Lau C.K. Rethinking financial contagion: information transmission mechanism during the COVID-19 pandemic SSRN Electron. J. 2020 10.2139/SSRN.3602973 Yasmeen R. Tao R. Jie W. Padda I.U.H. Shah W.U.H. The repercussions of business cycles on renewable & non-renewable energy consumption structure: evidence from OECD countries Renew. Energy 190 2022 572 583 10.1016/J.RENENE.2022.03.138 Yin L. Nie J. Han L. Understanding cryptocurrency volatility: the role of oil market shocks Int. Rev. Econ. Finance 72 2021 233 253 10.1016/J.IREF.2020.11.013 Yoshino N. Rasoulinezhad E. Taghizadeh-Hesary F. Economic impacts of carbon tax in a general equilibrium framework: empirical study of Japan J. Environ. Assess. Pol. Manag. 23 1n02 2021 2250014 10.1142/S1464333222500144 Yousaf I. Ali S. The COVID-19 outbreak and high frequency information transmission between major cryptocurrencies: evidence from the VAR-DCC-GARCH approach Borsa Istanbul Rev. 20 2020 10.1016/J.BIR.2020.10.003 S1–S10 Zhao Z. Feng S. Zhao Y. Wang Z. Ma J. Xu L. Yang J. Shen B. Investigation on the fuel quality and hydrophobicity of upgraded rice husk derived from various inert and oxidative torrefaction conditions Renew. Energy 189 2022 1234 1248 10.1016/j.renene.2022.03.087	36507092	PMC9718377	NO-CC CODE	2022-12-06 23:23:38	no	Resour Policy. 2023 Jan 2; 80:103121	utf-8	Resour Policy	2,022	10.1016/j.resourpol.2022.103121	oa_other
==== Front Drug Saf Drug Saf Drug Safety 0114-5916 1179-1942 Springer International Publishing Cham 1256 10.1007/s40264-022-01256-2 Original Research Article The Food and Drug Administration’s (FDA’s) Drug Safety Surveillance During the COVID-19 Pandemic http://orcid.org/0000-0002-9468-5324 Diak Ida-Lina [email protected] 1 Swank Kimberley 1 McCartan Kate 1 Beganovic Maya 1 Kidd James 1 Gada Neha 1 Kapoor Rachna 1 http://orcid.org/0000-0002-7372-6051 Wolf Lisa 1 Kangas Laura 1 Wyeth Jo 1 Salvatore Toni 1 Fanari Melina 1 LeBoeuf Andrew A. 2 Mishra Poonam 2 http://orcid.org/0000-0003-2306-0914 Blum Michael D. 1 http://orcid.org/0000-0003-4874-5864 Dal Pan Gerald 1 1 grid.417587.8 0000 0001 2243 3366 Office of Surveillance and Epidemiology, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, MD USA 2 grid.417587.8 0000 0001 2243 3366 Office of New Drugs, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, MD USA 2 12 2022 111 6 11 2022 © This is a U.S. Government work and not under copyright protection in the US; foreign copyright protection may apply 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Introduction On 4 February, 2020, the Secretary of the Department of Health and Human Services declared a public health emergency related to coronavirus disease 2019 (COVID-19), and on 27 March, 2020 declared circumstances existed to justify the authorization of the emergency use of drug and biological products (hereafter, “drugs”) for COVID-19. At the outset of the pandemic with uncertainty relating to the virus, many drugs were being used to treat or prevent COVID-19, resulting in the US Food and Drug Administration’s (FDA’s) need to initiate heightened surveillance across these drugs. Objective We aimed to describe the FDA’s approach to monitoring the safety of drugs to treat or prevent COVID-19 across multiple data sources and the subsequent actions taken by the FDA to protect public health. Methods The FDA conducted surveillance of adverse event and medication error data using the FDA Adverse Event Reporting System, biomedical literature, FDA-American College of Medical Toxicology COVID-19 Toxicology Investigators Consortium Pharmacovigilance Project Sub-registry, and the American Association of Poison Control Centers National Poison Data System. Results From 4 February, 2020, through 31 January, 2022, we identified 22,944 unique adverse event cases worldwide and 1052 unique medication error cases domestically with drugs to treat or prevent COVID-19. These were from the FDA Adverse Event Reporting System (22,219), biomedical literature (1107), FDA-American College of Medical Toxicology COVID-19 Toxicology Investigator’s Consortium Sub-registry (638), and the National Poison Data System (32), resulting in the detection of several important safety issues. Conclusions Safety surveillance using near real-time data was critical during the COVID-19 pandemic because the FDA monitored an unprecedented number of drugs to treat or prevent COVID-19. Additionally, the pandemic prompted the FDA to accelerate innovation, forging new collaborations and leveraging data sources to conduct safety surveillance to respond to the pandemic. ==== Body pmcKey Points The emergence of a variety of drugs to treat or prevent coronavirus disease 2019 highlighted the need for the US Food and Drug Administration to move beyond routine pharmacovigilance practices to rapidly identify emerging safety concerns. Establishing collaborations and leveraging new data sources including near real-time data were essential in the Food and Drug Administration’s response to the pandemic. The Food and Drug Administration evaluated over 20,000 adverse event cases from multiple data sources with drugs used to treat or prevent coronavirus disease 2019 during the first 2 years of the pandemic, resulting in the identification of safety issues and subsequent actions to protect the health of Americans during the pandemic. Introduction In December 2019, a novel coronavirus (severe acute respiratory syndrome coronavirus 2 [SARS-CoV-2]) was first detected in Wuhan City, Hubei Province, China. On 4 February, 2020, the Secretary of the Department of Health and Human Services declared a national public health emergency related to coronavirus disease 2019 (COVID-19) [1]. On 27 March, 2020, pursuant to section 564 of the Federal Food, Drug, and Cosmetic Act, the Secretary of the Department of Health and Human Services declared that circumstances existed justifying the authorization of the emergency use of drug and biological products during the COVID-19 pandemic [2]. This declaration allowed the FDA to authorize unapproved drug or biological products or unapproved uses of approved drug or biological products for emergency use during the COVID-19 public health emergency upon an Agency determination for a particular product that the criteria for issuance of an Emergency Use Authorization (EUA) are met. In addition to using drugs authorized under an EUA, some members of the public and medical community turned to other drugs to treat or prevent COVID-19 without any available data to support their efficacy. Therefore, we quickly identified the need to adapt our existing surveillance practices into a system that could accomplish heightened surveillance based on the urgency related to the rapidly evolving nature of the virus, the illness it caused, and the therapies used to treat or prevent it. In 2009, during the H1N1 pandemic, our surveillance was limited to adverse event (AE) reports in the FDA Adverse Event Reporting System (FAERS) [3]. During the COVID-19 pandemic, the FAERS remains the FDA’s principal repository for drug safety information. Over the course of the pandemic, we established collaborations and leveraged new sources that enabled us to conduct safety surveillance using near real-time data across all drug and biological products (hereafter, “drugs”) regulated by the FDA’s Center for Drug Evaluation and Research that were used to treat or prevent COVID-19. The FDA was able to rapidly identify new safety issues, revise Fact Sheets (FSs) for drugs authorized under EUAs (“EUA drugs”), and issue public communications. Herein, we provide a descriptive analysis of the safety data from four sources: FAERS, biomedical literature, FDA-American College of Medical Toxicology (ACMT) COVID-19 Toxicology Investigator’s Consortium [FACT] Pharmacovigilance Project Sub-registry, and American Association of Poison Control Centers (AAPCC) National Poison Data System (NPDS). Our use of administrative claims data and electronic health records has been previously described elsewhere [4]. Methods This analysis covers a 2-year period starting with the Department of Health and Human Services Secretary’s declaration of the public health emergency on 4 February, 2020 through 31 January, 2022. To conduct heightened safety surveillance of drugs used during the pandemic, the FDA assembled a team of safety experts from the Center for Drug Evaluation and Research to monitor AEs and medication errors (MEs) across all available sources reporting data related to any drug used to treat or prevent COVID-19. These drugs included both new therapies and off-label uses of drugs with proven efficacy for their approved indications. Because of differences among the data sources, including the coding of AE terms and the method of case retrieval or submission, we used a customized approach to search for and identify relevant cases of interest from each source. We defined a “relevant” case as a case that reported an AE or death with a drug used for the treatment or prevention of COVID-19. We systematically collected and evaluated all the data received and excluded duplicate cases using a stepwise approach to identify duplicate cases by looking for exact matches of various data elements, such as age, sex, country, drug product, and description of the AE, and comparing them with cases already documented. Regardless of source, the overall assessment of the cases focused on demographic information, reported AEs and MEs, concomitant medications, time to onset of events, clinical outcome, reporter source, and dechallenge and rechallenge information. We evaluated each case to identify emerging safety issues. Once we identified a safety signal through our surveillance, we performed a stepwise approach to building a case series for a pharmacovigilance evaluation. This stepwise approach consisted of creating and applying a case definition and assessing the causal association between the reported AE and product using the World Health Organization-Uppsala Monitoring Center Causality Assessment System [5]. FAERS Established AE reporting requirements for applicants, manufacturers, packers, and distributors are detailed under 21CFR314.80 [6], 21CFR600.80 [7], 21CFR314.98 [8], and 21CFR310.305 [9]. For EUA drugs, under the conditions of authorization, the FDA requires that EUA sponsors and healthcare professionals (HCPs) report all serious AEs and MEs. The FDA determined that more stringent reporting requirements for EUA drugs are necessary to protect the public health. MedWatch is the FDA’s medical product safety reporting program for HCPs, patients, and consumers. Cases submitted to MedWatch are incorporated into the FAERS, a large database of spontaneously reported AEs used to support the FDA’s post-marketing surveillance program for drugs [10]. The FAERS data are useful for identifying unknown, rare, and serious AEs reported with the use of FDA-approved or authorized drugs, which is especially important throughout the pandemic. Because FAERS is an established AE database, it is our primary data source for conducting surveillance, especially early in the pandemic when there were many drugs being used to treat or prevent COVID-19. We adapted our strategies throughout the pandemic, from broadly searching for any case mentioning the novel coronavirus or COVID-19-related terms (e.g., coronavirus infection, coronavirus test positive, SARS-CoV-2), to searching more narrowly for cases involving drugs authorized or approved to treat or prevent COVID-19. We reviewed all cases identified from the search strategies and tracked all the relevant cases. As the pandemic evolved, we began utilizing additional data sources. Biomedical Literature We screened recurring weekly search queries of PubMed and Embase and reviewed article titles and abstracts to identify cases describing an AE occurring after administration of any drug to treat or prevent COVID-19. FACT Pharmacovigilance Project Sub-registry [11] In 2010, the ACMT established the Toxicology Investigators Consortium as a multicenter toxico-surveillance and research network of medical toxicologists and has continued expanding the capabilities of the registry to serve as a database for public health surveillance and emerging concerns within the field of toxicology, including AEs [12, 13]. In response to the COVID-19 pandemic, the FDA contracted with ACMT’s Toxicology Investigators Consortium to establish the FACT Pharmacovigilance Project Sub-registry to leverage the existing capabilities of the Toxicology Investigators Consortium registry as a surveillance tool to identify emerging safety issues associated with drugs to treat or prevent COVID-19 [14]. The FACT Sub-registry includes medical toxicology site investigators and research assistants from 17 medical centers across the USA. FACT enables direct outreach by the FDA to site investigators when additional information on reported cases is needed. In conjunction with the FDA, ACMT and site investigators developed enhanced data collection forms for some AEs of special interest to the FDA in an effort to obtain additional clinical details that could be used in FDA evaluations of safety concerns. The FDA received, and continues to receive, real-time e-mail alerts as new and follow-up cases are entered into the Sub-registry providing an opportunity to review cases in near real-time. All cases identified from the real-time e-mail alerts were reviewed for relevance and tracked. NPDS [15] The NPDS, a database managed by AAPCC and derived from a nationwide network of US Poison Control Centers (PCCs), captures near real-time data from calls from individuals, HCPs, and others regarding exposures to prescription drugs, over-the-counter medications, and unapproved drugs. Healthcare professionals at PCCs who respond to calls about exposures have specialized training in toxicology needed to assess, triage to the most appropriate level of care, provide recommendations, and follow up on toxic emergencies. Within the NPDS, calls for exposures may result in documentation of an event, provision of information, or advice regarding medical management; PCC staff managing these cases undergo training in the effort to standardize documentation across centers. Documentation of cases includes details on the drug(s), patient characteristics, route of exposure, reasons for exposure, level of care received (e.g., admitted to critical care unit vs treated and released), medical outcomes (e.g., death, no effect), and other more curated variables, such as “relatedness” of the reported exposure to the outcomes of interest. Reasons for use are categorized into groups by the AAPCC and include such categories as “intentional” and “unintentional,” the former encompassing the subgroups of intentional misuse, abuse, suspected suicide, or unknown intent. The PCC case data do not represent the national-level incidence of exposures or cases of misuse/abuse related to any substance. These data only capture events if the exposure resulted in a call to a PCC. The PCC data rely on information voluntarily shared by patients and healthcare personnel, and most substance classification is based on history alone and does not involve any laboratory confirmation. Exposures may be unconfirmed ingestions (i.e., the product may not have been ingested at all by the patient). Drug exposures resulting in an unattended or out-of-hospital death are unlikely to generate a call to a PCC, and therefore, fatal poisonings are expected to be substantially under-documented in PCC case data. Throughout the pandemic, we used the NPDS to identify exposure cases reporting an AE across a variety of prescription (e.g., hydroxychloroquine, ivermectin [16]), over-the-counter (e.g., hand sanitizers [17, 18], hydrogen peroxide), and unapproved products (e.g., Miracle Mineral Solution [19], oleander extract) used in an effort to treat or prevent COVID-19. Daily surveillance queries were generated to identify exposure calls related to these products. All cases identified from the queries were reviewed for relevance and tracked. The NPDS served as a valuable data source in the surveillance efforts for many of these products, especially hand sanitizers, including those contaminated with methanol [20]. Results AEs Through our surveillance, we monitored the safety of COVID-19 drugs and evaluated potential AEs of interest across the four data sources. From 4 February, 2020, through 31 January, 2022, we identified 22,944 unique cases for inclusion in this analysis reporting an AE with a drug to treat or prevent COVID-19. Of these, we identified 21,167 in the FAERS database, 1107 in the biomedical literature, 638 in the FACT Sub-registry, and 32 in the NPDS database. A majority of these cases were reported from US sources. Variation in reporting between other countries and the USA is likely multifactorial and may include reasons such as authorization/approval status of drugs in other countries, the standard of care for treating COVID-19, and the number of literature publications varying from each country. In addition, in the USA, there are different regulatory requirements for domestic and foreign case reports. Adverse events that are both serious and unexpected (i.e., not in the drug product label), whether domestic or foreign, from all sources must be submitted to the FDA within 15 days of initial receipt of the information by the applicant. For AEs that are serious and expected or non-serious, only domestic reports must be submitted to the FDA by the applicant. Additionally, the NPDS database and the FACT Sub-registry are US-based data sources; therefore, foreign cases from these sources are not anticipated. Table 1 summarizes the descriptive characteristics of the patients described in the 22,944 cases.Table 1 Patient characteristics and drug product data from 4 February, 2020 through 31 January, 2022 Case demographics All casesa (n = 22,944) HCQ or CQ (n = 3649) RDV (n = 6747) BAM (n = 3377) BCT (n = 685) CASI and IMDE (n = 3464) BAM and ETE (n = 1497) SOT (n = 409) TCZ (n = 2913) TIX and CIL (n = 21) NIRM and RTV (n = 148) MOV (n = 50) Date of authorization – 28 March, 2020b 1 May, 2020c 9 November, 2020 19 November, 2020d 21 November, 2020 9 February, 2021 26 May, 2021 24 June , 2021 8 December, 2021 22 December, 2021 23 December, 2021 Age, median (IQR) n = 20,594 62 (48–73) n = 3340 60 (48–71) n = 6331 60 (50–74) n = 3006 68 (59–77) n = 627 61 (51–71) n = 3356 55 (40–68) n = 1391 52 (37.5–66) n = 336 61 (45–73) n = 2295 62 (50–71) n = 18 55 (44–63) n = 108 62 (47–72) n = 47 67 (20–79) Sex n = 21,529 n = 3429 n = 6628 n = 3279 n = 645 n = 3414 n = 1440 n = 361 n = 2353 n = 20 n = 128 n = 50 Male, n (%) 12,205 (56) 2299 (67) 4,131 (62) 1,714 (52) 427 (66) 1469 (43) 541 (38) 166 (46) 1,681 (71) 13 (65) 45 (35) 21 (42) Female, n (%) 9319 (43) 1130 (33) 2495 (37) 1565 (48) 218 (34) 1943 (57) 899 (62) 194 (54) 672 (29) 7 (35) 83 (65) 29 (58) Transgender, n (%) 4 (<1) 0 2 (<1) 0 0 2 (<1) 0 0 0 0 0 0 Intersex, n (%) 1 (<1) 0 0 0 0 0 0 1 (<1) 0 0 0 0 Country USA 16,036 1112 5412 3330 586 3450 1370 284 1058 20 147 32 Other countries 6908 2537 1335 47 99 14 127 125 1855 1 1 18 Reported deaths,e n (%) 5061 (22) 1217 (33) 1932 (29) 201 (6) 186 (27) 148 (4) 64 (4) 28 (7) 1386 (48) 1 (5) 2 (1) 7 (14) BAM bamlanivimab, BCT baricitinib, CASI casirivimab, CIL cilgavimab, COVID-19 coronavirus disease 2019, CQ chloroquine, ECMO extracorporeal membrane oxygenation, ETE etesevimab, EUA emergency use authorization, FDA Food and Drug Administration, HCQ hydroxychloroquine, IMDE imdevimab, IQR interquartile range, MOV molnupiravir, NIRM nirmatrelvir, RDV remdesivir, RTV ritonavir, SOT sotrovimab, TCZ tocilizumab, TIX tixagevimab aThis total includes all drugs, EUA and non-EUA, identified during surveillance. Other drugs used to treat or prevent COVID-19 that were included in this total and reported in 50 or more cases are: azithromycin, lopinavir/ritonavir, convalescent plasma, oseltamivir, favipiravir, sarilumab, interferon (i.e., alpha, alfa-2b, beta-1a, 2b, 1b), anakinra, nitric oxide, ivermectin, umifenovir, ribavirin, ritonavir, canakinumab, darunavir/cobicistat, darunavir, and tofacitinib. More than one treatment may have been reported in each case; therefore, the total number for “all cases” is lower than the sum across all products bThe FDA revoked the HCQ/CQ EUA on 15 June, 2020 cOn 22 October, 2020, the FDA approved remdesivir for use in adults and pediatric patients (12 years of age and older and weighing at least 40 kg) for the treatment of COVID-19 requiring hospitalization. On April 25, 2022, the FDA approved remdesivir for the remaining populations (pediatric patients aged 28 days and older and weighing at least 3 kg) removing the need for the EUA. Corresponding revisions to the EUA with respect to its authorized uses and populations were similarly made during this time frame dOn 10 May, 2022, the FDA approved baricitinib for the treatment of COVID-19 in hospitalized adults requiring supplemental oxygen, non-invasive or invasive mechanical ventilation, or ECMO. This is the first immunosuppressant to be approved for this condition. The FDA also revised the EUA to authorize baricitinib for emergency use to treat COVID-19 in hospitalized pediatric patients aged 2 to less than 18 years eIncludes mortality from all causes and does not imply a causal relationship to the drug Figure 1 illustrates the trend in cases reporting an AE with a drug to treat or prevent COVID-19 from 4 February, 2020 through 31 January, 2022. There were several upward trends in AE reporting during this period that appeared to correlate with the COVID-19 pandemic trajectory [21]. Early in the pandemic, prior to the issuance of any EUAs, certain drugs were used off-label to combat COVID-19, including antivirals and antimicrobials, such as lopinavir/ritonavir, azithromycin, ceftriaxone, and interferon. In addition to the overall reporting trends during this period, there were several trends observed with EUA drugs. As illustrated in Fig. 1, peaks and valleys in AE reporting for EUA drugs related to the issuance of new EUAs, shifts in variants, or revocations of EUAs. For example, the FDA issued an EUA for bamlanivimab on 9 November, 2020, which was followed by an increase in AE reporting for this product shortly thereafter in December 2020 and January 2021, followed by a sharp decline in April 2021 after revocation of the EUA of bamlanivimab because of a lack of susceptibility [22]. In addition, there are mandatory reporting requirements of all serious AEs and MEs for EUA drug products, which could be one explanation for an increase in AE reporting trends seen in Fig. 1. The trends depicted in the graph are unlikely related to changes in search strategies or the addition of new data sources, as these changes occurred early in the surveillance process prior to granting EUAs for COVID-19 products. As the pandemic evolved, changes in therapy to treat or prevent COVID-19 are reflected in the AE reporting trends.Fig. 1 Cases from the FDA Adverse Event Reporting System, biomedical literature, FDA-American College of Medical Toxicology COVID-19 Toxicology Investigators Consortium Pharmacovigilance Sub-registry, and the National Poison Data System reporting an adverse event with a drug to treat or prevent COVID-19 from 4 February, 2020 through 31 January, 2022. *All products, EUA and non-EUA, identified during surveillance. Monthly COVID-19 cases were calculated from the CDC COVID Data Tracker: https://covid.cdc.gov/covid-data-tracker/#trends_dailycases. BAM bamlanivimab, BCT baricitinib, CASI casirivimab, CDC Centers for Disease Control and Prevention, CIL cilgavimab, COVID coronavirus disease, CQ chloroquine, ETE etesevimab, EUA Emergency Use Authorization, HCQ hydroxychloroquine, IMDE imdevimab, MOV molnupiravir, NDA New Drug Application, NIRM nirmatrelvir, RDV remdesivir, RTV ritonavir, SOT sotrovimab, TCZ tocilizumab, TIX tixagevimab, US United States Actions Resulting from Surveillance Activities Here forward, we focus on data collected through our heightened surveillance efforts to support actions related to certain COVID-19 EUA drugs. These actions included FDA public communications and safety-related updates to EUA FSs. Reporting frequencies, summarized in Tables 2, 3 and 4, were used to identify potential AEs of interest, which the FDA further investigated and assessed the need for regulatory action. An overview of these actions and the evidence supporting them follows. Hydroxychloroquine and Chloroquine On 28 March, 2020, the FDA issued an EUA for the antimalarial drugs, hydroxychloroquine and chloroquine, for adults and adolescents weighing ≥ 50 kg hospitalized with COVID-19 for whom a clinical trial was not available, or participation was not feasible. The FDA revoked the EUA on 15 June, 2020, after emerging scientific data, including a randomized trial, failed to show evidence of a benefit [23], prompting the National Institutes of Health to recommend against using these drugs to treat or prevent COVID-19 [24]. Cardiotoxicity, particularly when associated with the cardiac conduction system, is a known risk of hydroxychloroquine and chloroquine. Notably, prolongation of the QT interval was the most common AE reported with hydroxychloroquine and chloroquine (Table 2). Related life-threatening arrhythmias, torsades de pointes (26 cases), and ventricular fibrillation (16 cases), were also reported. An FDA evaluation of cases, which included four cases of torsades de pointes, prompted the FDA to issue a Drug Safety Communication on 24 April, 2020 (during the EUA period), reminding the public of the risk of arrhythmia and cautioning against hydroxychloroquine and chloroquine use for COVID-19 outside of the hospital setting or a clinical trial [25]. The FDA further evaluated cardiotoxicity in patients receiving these drugs for COVID-19 in a safety assessment during the EUA period, and concluded that the current hydroxychloroquine and chloroquine labeling, which describes cardiotoxicity in the Warnings and Precautions section, adequately conveyed these risks for approved indications, and further recommended an update to the chloroquine labeling to better describe cardiac electrophysiology study findings. The non-cardiac AEs were more likely related to COVID-19 infection than the use of hydroxychloroquine and chloroquine. For example, a recent study described the use of the FDA’s Sentinel System to measure the 90-day risk of arterial thromboembolism and venous thromboembolism in patients hospitalized with COVID-19 before or during COVID-19 vaccine availability versus patients hospitalized with influenza. The authors concluded that there was a higher risk of venous thromboembolism in hospitalized patients with COVID-19 compared with hospitalized patients with influenza [26].Table 2 Most frequently reported adverse events with a reporting frequency ≥ 2% with hydroxychloroquine or chloroquine (n = 3649) Adverse eventsa Reporting frequencyb, n (%) Electrocardiogram QT prolongedc 424 (11.6) Respiratory disordersd 368 (10.1) Liver impairmente 353 (9.7) Renal impairmentf 229 (6.3) Gastrointestinal disordersg 153 (4.2) Cardiac arrythmiah 138 (3.8) Cardiac arrest and cardio-respiratory arrest 94 (2.6) Pulmonary embolism 78 (2.1) Data sources include the FDA’s Adverse Event Reporting System, literature, National Poison Data System, and FDA-American College of Medical Toxicology COVID-19 Toxicology Investigator’s Consortium Sub-registry aA single case could describe more than one adverse event bReporting frequency is the number of reported cases of a particular adverse event for a product used to treat or prevent coronavirus disease 2019 divided by the total number of reported cases for that product × 100 cIncludes long QT syndrome dIncludes respiratory failure, acute respiratory distress syndrome, hypoxia, dyspnea, and acute respiratory failure eIncludes hepatitis, increased alanine aminotransferase, increased aspartate aminotransferase, hepatic cytolysis, increased transaminases, liver injury, and hypertransaminasemia fIncludes acute kidney injury, increased serum creatinine, and renal failure gIncludes diarrhea, nausea, and vomiting hIncludes atrial fibrillation, ventricular tachycardia, bradycardia, and tachycardia Remdesivir Remdesivir, a SARS-CoV-2 nucleotide analog RNA polymerase inhibitor, became the first novel antiviral product issued an EUA on 1 May, 2020, and on 22 October, 2020 became the first approved treatment for COVID-19 in the USA. Hypersensitivity reactions, including infusion-related reactions (IRRs), were recognized as an important safety issue for remdesivir at the time of EUA issuance and were included in the Warnings and Precautions section of the FS for HCPs. Post-authorization AE cases received through the FAERS for remdesivir identified signs and symptoms of hypersensitivity reactions that were not included in the initial FS, including tachycardia, bradycardia, dyspnea, wheezing, angioedema, rash, and anaphylaxis. On 15 June, 2020, shortly after identification of these additional events, the FDA updated the FS to inform the medical community and public of these emerging safety issues. Hepatotoxicity was another important safety issue for remdesivir, with liver impairment being the most frequently reported AE (Table 3). Transaminase elevations were observed across the remdesivir development program. Upon EUA issuance, the FS for HCPs included the risk of transaminase elevations in the Warnings and Precautions section with advice to discontinue or not initiate remdesivir in patients whose alanine aminotransferase (ALT) was greater than five times the upper limit of normal or who had an ALT elevation with signs or symptoms of liver inflammation. Although additional cases of transaminase elevations were identified post-authorization, all cases reporting hepatic failure had information suggesting alternative explanations or did not have sufficient information to attribute this AE to remdesivir. Labeling upon FDA approval retained transaminase elevations in the Warnings and Precautions with risk mitigation strategies including discontinuing remdesivir if ALT levels exceeded ten times the upper limit of normal or if the ALT elevation was accompanied by signs or symptoms of liver inflammation.Table 3 Most frequently reported adverse events with a reporting frequency ≥ 2% with remdesivir (n = 6746) Adverse eventsa Reporting frequencyb, N (%) Liver impairmentc 1750 (26) Renal impairmentd 1016 (15) Respiratory disorderse 748 (11) Bradycardiaf 698 (10) Hypotensiong 222 (3) Cardiac arrest 163 (2) Infusion site extravasation 130 (2) Data sources include the FDAs Adverse Event Reporting System, literature, and FACT Sub-registry aA single case could describe more than one adverse event bReporting frequency is the number of reported cases of a particular adverse event for a product used to treat or prevent coronavirus disease 2019 divided by the total number of reported cases for that product × 100 cIncludes increased alanine aminotransferase, increased aspartate aminotransferase, increased hepatic enzyme, increased transaminases, increased liver function tests, increased blood alkaline phosphatase, increased blood bilirubin, acute hepatic failure, liver injury, drug-induced liver injury dIncludes increased blood creatinine, acute kidney injury, decreased glomerular filtration, renal failure eIncludes dyspnea, hypoxia, decreased oxygen saturation, respiratory distress, respiratory failure, acute respiratory failure fIncludes decreased heart rate gIncludes decreased blood pressure Renal impairment was also a frequently reported AE for remdesivir. Review of the cases did not identify a causal relationship between remdesivir and renal impairment; therefore, no FS changes were warranted. Neutralizing Monoclonal Antibodies The FDA issued EUAs for six SARS-CoV-2 neutralizing IgG1 monoclonal antibody (mAb) therapies to treat or prevent COVID-19, including bamlanivimab, bamlanivimab/etesevimab, casirivimab/imdevimab, sotrovimab, tixagevimab/cilgavimab, and bebtelovimab, with the first, bamlanivimab, authorized on 9 November, 2020. Through heightened surveillance, the FDA identified AEs of interest, resulting in several updates to the mAb FS for HCPs. Based on clinical trial data for mAbs administered via an infusion, the FSs for HCPs included a warning for hypersensitivity reactions including anaphylaxis and IRRs, which included bronchospasm, hypotension, angioedema, rash including urticaria, myalgia, and dizziness. Most IRRs and hypersensitivity reactions in the clinical trials were reported as mild to moderate in severity. To mitigate the potential risks of severe IRRs, the FDA requires the administration of bamlanivimab, bamlanivimab/etesevimab, casirivimab/imdevimab, and sotrovimab, in settings in which HCPs have immediate access to medications to treat a severe reaction, such as anaphylaxis, and the ability to activate the emergency medical system as necessary. Patients should be monitored during infusion/injection(s) and observed for at least 1 hour after the infusion is complete. A review of the post-authorization data identified difficulty breathing, reduced oxygen saturation, fatigue, arrhythmia, altered mental status, hypertension, and diaphoresis as the most frequently reported AEs occurring within 2 hours of mAb administration (i.e., IRRs) that were not included in the FSs at the time of initial authorization. In February 2021, the FDA updated the FS for HCP for the authorized mAbs at the time based on post-authorization cases identified in the FAERS database and FACT Sub-registry. The updates included the addition of new signs and symptoms under hypersensitivity including IRRs in the Warnings and Precautions section of the FSs. Additionally, the FDA included a new warning communicating the potential for clinical worsening after mAb administration based on cases identified of rapid respiratory decompensation during or shortly after mAb administration, some of which required hospitalization. In June 2021, the FDA updated the Warnings and Precautions section of the mAb FS for HCPs to include hypersensitivity reactions occurring more than 24 hours after infusion and vasovagal reactions (i.e., pre-syncope and syncope) based on post-authorization cases identified in the FAERS database and FACT Sub-registry describing pre-syncopal and syncopal episodes occurring during or shortly after administration of mAbs that were identified in the context of IRRs and hypersensitivity reactions, many of which required treatment in the emergency department or hospital admission. Additionally, the FDA identified cases of non-immediate (>2 hours after infusion) hypersensitivity reactions, which included cutaneous reactions and angioedema. The most frequently reported AEs (Table 4) with a reporting frequency of ≥5% with mAbs, including IRRs, are all labeled AEs in the mAb FS for HCPs, except for cough.Table 4 Most frequently reported adverse events with a reporting frequency ≥5% with monoclonal antibodies (n = 8737) Adverse eventsa Reporting frequencyb, n (%) Infusion-related reactionsc 3923 (45) Respiratory disordersd 2537 (29) Nausea/vomiting 966 (11) Fever 912 (10) Chest pain/discomfort 918 (10) Chills 697 (8) Dizziness 651 (7) Rash/erythema 581 (7) Cough 546 (6) Flushing 507 (6) Hypotension 538 (6) Headache 446 (5) Tachycardia 446 (5) Data sources include the FDA Adverse Event Reporting System, literature, and FACT Sub-registry aA single case could describe more than one adverse event bReporting frequency is the number of reported cases of a particular adverse event for a product used to treat or prevent coronavirus disease 2019 divided by the total number of reported cases for that product × 100 cInfusion-related reactions was defined as those adverse events that occurred within 2 hours of monoclonal antibody administration dIncludes dyspnea, hypoxia, oxygen saturation decreased, respiratory distress, respiratory failure, and respiratory depression MEs From 4 February, 2020, through 31 January, 2022, we identified 1052 unique US ME cases submitted to the FAERS. Healthcare professionals reported 90% of the cases. Approximately 20% (227/1,052) of the cases reported a serious AE; however, based on individual case reviews, the serious AEs were more likely related to drug therapy or underlying medical conditions (including COVID-19) than a ME. As shown in Table 5, the number of reported ME cases varied by product, from one case (molnupiravir) to 484 cases (casirivimab/imdevimab). The number of cases was influenced by the length of time the product was authorized for use, individual definitions of an ME, utilization patterns, and product characteristics (e.g., route of administration, packaging, and procedures for use).Table 5 EUA drug product characteristics and number of medication error cases received by the US FDA between 4 February, 2020 through 31 January, 2022 (n = 1052) Producta Case count Route of administration Dosage forms Packaging Bamlanivimab or etesevimab 175 IV Injection, 700 mg/20 mL (bamlanivimab) Injection, 700 mg/20 mL (etesevimab) Single-dose vials in individual cartons Olumiant (baricitinib) 17 Oral Tablets, 1 mg, 2 mg, and 4 mg Bottles REGEN-COV (casirivimab/imdevimab or casirivimab or imdevimab) 484 IV or SC Injection, 600 mg of casirivimab and 600 mg of imdevimab per 10 mL (co-formulated) Injection, 300 mg/2.5 mL, 1332 mg/11.1 mL (casirivimab) Injection, 300 mg/2.5 mL, 1332 mg/11.1 mL (imdevimab) Single-dose vials (co-formulated) in individual cartons Single-dose vials in individual cartons Single-dose vials, co-packaged in one carton Dose packs providing 1200 mg casirivimab and 1200 mg imdevimab (containing 2, 5, or 8 single-dose vials in individual cartons per dose pack) Evusheld (cilgavimab; tixagevimab) 4 IM Injection, 150 mg/1.5 mL (cilgavimab) Injection, 150 mg/1.5 mL (tixagevimab) Single-dose vials, co-packaged in one carton Plaquenil (hydroxychloroquine) or Aralen (chloroquine) 3 Oral 200 mg (hydroxychloroquine sulfate) 500 mg (chloroquine phosphate) Bottles Veklury (remdesivir) 246 IV Injection, 100 mg/20 mLb For injection, 100 mg of lyophilized powder for reconstitution Single-dose vials in individual cartons Sotrovimab 96 IV Injection, 500 mg/8 mL Single-dose vial in individual cartons Actemra (tocilizumab) 12 IV Injection, 80 mg/4 mL, 200 mg/10 mL, or 400 mg/20 mL Single-dose vials in individual cartonsc Lagevrio (molnupiravir) 1 Oral Capsules, 200 mg Bottles Paxlovid (nirmatrelvir; ritonavir) 17 Oral Nirmatrelvir tablets, 150 mg Ritonavir tablets, 100 mg Blister packs of tablets, co-packaged COVID-19 coronavirus disease 2019, EUA Emergency Use Authorization, FDA Food and Drug Administration, IM intramuscular, IV intravenous, SC subcutaneous aLimited to EUA drugs authorized to treat or prevent COVID-19. See Table 1 for the date of first EUA issuance bBoth remdesivir injection and remdesivir for injection formulations were authorized for emergency use in the management of COVID-19; both formulations are now FDA approved cTocilizumab is also approved in the USA as an autoinjector (Actemra Actpen); however, only the single-dose vial was authorized for emergency use in the management of COVID-19 The most frequently reported MEs across all products were improper dose, product prepared incorrectly, product dispensing error (e.g., wrong strength or formulation dispensed), transcription error (e.g., incorrect order entry), wrong drug product administered (e.g., casirivimab administered instead of the intended COVID-19 vaccine), and wrong route of administration. Depending on the product and type of ME, multiple contributing factors for the errors were reported, including inconsistent or unfamiliar formatting and content on container labels (e.g., use of investigational drug nomenclature [27], lack of barcodes for product verification), multiple packaging presentations for the same product, limitations of electronic health systems, miscommunication, and changes to dosage regimens. Fast-paced stressful work environments with staffing and supply shortages and deviations from usual workflow practices also contributed to the errors. As part of the EUA process, the FDA reviewed proposed FS language and packaging to minimize potential MEs. Post-authorization, the FDA monitored ME cases and subsequently updated FSs and requested manufacturers revise container labels, carton labeling, and packaging to help mitigate reported errors. Discussion Although many drugs have been used throughout the pandemic to treat or prevent COVID-19, this analysis primarily focused on the FDA’s heightened surveillance efforts that were integral to identifying emerging safety issues across many EUA drugs leading to rapid actions to alert the public of these risks. The FDA was uniquely positioned as an agency because of our access to both internal and external AE and ME data sources providing robust safety data. This was possible through existing and new partnerships and leveraging and building onto our infrastructure through data sources providing near real-time data. Additionally, to improve transparency and expand data access to the public, the FDA launched the COVID-19 EUA FAERS Public Dashboard [28]. The COVID-19 pandemic provided the opportunity to learn key lessons that contributed to the success of our surveillance efforts. During the COVID-19 pandemic, the FDA requires stringent AE/ME reporting for FDA EUA drugs and has had access to valuable data through multiple sources, which allowed us to rapidly assess and identify trends/signals to take prompt action. One example of a new collaboration between the FDA and ACMT resulted in the FACT Sub-registry, which was developed to capture AEs associated with COVID-19 drugs in near real-time. At the outset, it was unknown what type of AEs would be captured given the setting of data collection, but it proved to be a very useful source that complemented other data sources. We were also able to contribute to developing targeted data collection forms and, as a result, were better positioned to assess these cases given the additional details on AEs of interest obtained. The FDA staff also collaborated with the AAPCC to develop recurring queries in the NPDS to capture potential toxic exposures related to hydroxychloroquine and chloroquine early in the pandemic. Because of these collaborations, the near real-time data sources diversified our sources of AE case reports, thereby enhancing our surveillance efforts during the pandemic. Post-authorization surveillance of AEs and MEs for EUA drugs was critical to inform benefit-risk determinations to support continuation of an EUA. When safety concerns were identified, the FDA took actions, including FS revisions and communications. Data obtained through the FDA’s surveillance for hydroxychloroquine and chloroquine resulted in the release of a Drug Safety Communication alerting the public of the risk of arrhythmias and cautioning against its use for COVID-19 outside of the hospital setting or clinical trials. The FDA also used data identified with remdesivir to rapidly update the FS and during the New Drug Application review to ensure the product labeling on approval adequately described the known safety profile. Additionally, we continue to identify emerging safety issues through ongoing surveillance. Most recently, the FDA identified hypersensitivity reactions related to nirmatrelvir/ritonavir and molnupiravir shortly after distribution began that resulted in updates to the FSs for each product within weeks of identification, which further highlights the importance of the near real-time access to data on AEs and MEs. The FDA’s safety assessments provide a valuable window into the safety of the drugs to treat or prevent COVID-19. Conclusions The emergence of a variety of drugs to treat or prevent COVID-19 highlighted the need for the FDA to move beyond routine pharmacovigilance practices to rapidly identify emerging safety concerns related to these drugs and communicate these risks to the public. Establishing collaborations and leveraging new data sources were essential in our response to the pandemic. Declarations Funding No sources of funding were received for the preparation of this article. Conflicts of Interest/Competing Interests None of the authors has any conflicts of interest to disclose. Ethics Approval Ethics approval was not considered necessary for this study because FDA surveillance activity is deemed necessary for the protection of public health. Consent to Participate Not applicable. Consent for Publication Not applicable. Availability of Data and Material The FAERS data are available via the FAERS Public Dashboard and as Quarterly Data files. Additionally, individual case reports can be requested via a Freedom of Information Act request to the FDA. Additional details can be found here: https://www.fda.gov/drugs/surveillance/questions-and-answers-fdas-adverse-event-reporting-system-faers. The FDA-AAPCC NPDS data were obtained under contract with ACMT and AAPCC, respectively. Code Availability Not applicable. Authors Contributions Conceptualization: ILD, NG, RK, KS, KM, MB, JK, LW, LK, JW, MDB, GDP; data analysis: KS, KM, MB, JK, TS, MF, JW; writing: ILD, NG, KS, KM, MB, JK, TS, JW; review and editing: ILD, KS, KM, MB, JK, NG, RK, LW, ALB, PM, MDB, GDP; approval of the manuscript: all authors. ==== Refs References 1. Determination of Public Health Emergency. https://www.federalregister.gov/documents/2020/02/07/2020-02496/determination-of-public-health-emergency. Accessed 25 Mar 2022. 2. Notice of Emergency Use Authorization Declaration. https://www.federalregister.gov/documents/2020/04/01/2020-06905/emergency-use-authorization-declaration. Accessed 25 Mar 2022. 3. Sorbello A Jones SC Carter W Emergency use authorization for intravenous peramivir: evaluation of safety in the treatment of hospitalized patients infected with 2009 H1N1 influenza A virus Clin Infect Dis 2012 55 1 1 7 10.1093/cid/cis351 22491501 4. Cocoros NM Fuller CC Adimadhyam S A COVID-19-ready public health surveillance system: the Food and Drug Administration’s Sentinel System Pharmacoepidemiol Drug Saf 2021 30 7 827 837 10.1002/pds.5240 33797815 5. WHO-UMC system for standardization case causality assessment. http://www.who-umc.org/. Accessed 14 Oct 2022. 6. CFR - Code of Federal Regulations Title 21 Sec. 314.80 Postmarketing reporting of adverse drug experiences. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=314.80. Accessed 22 May 2022. 7. CFR - Code of Federal Regulations Title 21 Sec. 600.80 Postmarketing reporting of adverse experiences. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=600.80. Accessed 22 May 2022. 8. CFR - Code of Federal Regulations Title 21 Sec. 314.98 Records and reports concerning adverse drug experiences on marketed prescription drugs for human use without approved new drug applications: abbreviated applications. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=314.98. Accessed 22 May 2022. 9. CFR - Code of Federal Regulations Title 21 Sec. 310.305 Records and reports concerning adverse drug experiences on marketed prescription drugs for human use without approved new drug applications. Available from: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=310.305. Accessed 22 May 2022]. 10. Questions and answers on FDA’s Adverse Event Reporting System (FAERS). https://www.fda.gov/drugs/surveillance/questions-and-answers-fdas-adverse-event-reporting-system-faers. Accessed 17 Nov 2022. 11. FDA ACMT COVID-19 ToxIC (FACT) Pharmacovigilance Project Sub-Registry description. https://www.toxicregistry.org/FACT.html. Accessed 17 Nov 2022. 12. Wax PM Kleinschmidt KC Brent J The Toxicology Investigators Consortium (ToxIC) Registry J Med Toxicol 2011 7 4 259 265 10.1007/s13181-011-0177-z 21956161 13. Spyres MB Aldy K Farrugia LA The toxicology investigators consortium 2020 annual report J Med Toxicol 2021 17 4 333 362 10.1007/s13181-021-00854-3 34535889 14. Aldy K Wax P Brent J on behalf of the ToxIC FACT Study Group Rapid development of the FDA ACMT COVID-19 ToxIC (FACT) pharmacovigilance pilot project to monitor adverse events reported in association with COVID-19 therapeutics Clin Toxicol (Phila) 2021 59 11 1037 1093 15. American Association of Poison Control Centers National Poison Data System Information. https://www.aapcc.org/national-poison-data-system. Accessed 22 Mar 2022. 16. FDA consumer update: why you should not use ivermectin to treat or prevent COVID-19. https://www.fda.gov/consumers/consumer-updates/why-you-should-not-use-ivermectin-treat-or-prevent-covid-19. Accessed 18 Mar 2022. 17. FDA updates on hand sanitizers consumers should not use. https://www.fda.gov/drugs/drug-safety-and-availability/fda-updates-hand-sanitizers-consumers-should-not-use. Accessed 18 Mar 2022. 18. McCulley L Cheng C Mentari E Alcohol-based hand sanitizer exposures and effects on young children in the U.S. during the COVID-19 pandemic Clin Toxicol (Phila) 2021 59 4 355 356 10.1080/15563650.2020.1811298 32852232 19. Lardieri A Cheng C Jones SC McCulley L Harmful effects of chlorine dioxide exposure Clin Toxicol (Phila) 2021 59 5 448 449 10.1080/15563650.2020.1818767 32960110 20. Coronavirus (COVID-19) update: FDA takes action to warn, protect consumers from dangerous alcohol-based hand sanitizers containing methanol. https://www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-takes-action-warn-protect-consumers-dangerous-alcohol-based-hand. Accessed 18 Mar 2022. 21. Centers for Disease Control and Prevention. COVID data tracker. https://covid.cdc.gov/covid-data-tracker/#trends_dailytrendscases. Accessed 31 Mar 2022. 22. US FDA. FDA revocation letter of the Emergency Use Authorization for bamlanivimab. https://www.fda.gov/media/147629/download. Accessed 3 Apr 2022. 23. US FDA. FDA revocation letter of the Emergency Use Authorization of oral formulations of chloroquine phosphate and hydroxychloroquine sulfate. https://www.fda.gov/media/138945/download. Accessed 3 Apr 2022. 24. COVID-19 Treatment Guidelines Panel. Coronavirus disease 2019 (COVID-19) treatment guidelines. National Institutes of Health. https://www.covid19treatmentguidelines.nih.gov/. Accessed 30 Mar 2022. 25. US FDA. FDA cautions against use of hydroxychloroquine or chloroquine for COVID-19 outside of the hospital setting or a clinical trial due to risk of heart rhythm problems. https://www.fda.gov/drugs/drug-safety-and-availability/fda-cautions-against-use-hydroxychloroquine-or-chloroquine-covid-19-outside-hospital-setting-or. Accessed 30 Mar 2022. 26. Lo Re V Dutcher SK Connolly JG Association of COVID-19 vs influenza with risk of arterial and venous thrombotic events among hospitalized patients JAMA 2022 328 7 637 651 10.1001/jama.2022.13072 35972486 27. Institute for Safe Medication Practices. COVID-19 related medication errors. 14 May 2020. ISMP Med Saf Alert Acute Care. 25(9):1–3. https://www.ismp.org/acute-care/special-edition-medication-safety-alert-may-14-2020/covid-19. Accessed 6 May 2022. 28. FDA Adverse Event Reporting System Public Dashboard. https://www.fda.gov/drugs/questions-and-answers-fdas-adverse-event-reporting-system-faers/fda-adverse-event-reporting-system-faers-public-dashboard. Accessed 10 Apr 2022.	36460854	PMC9718450	NO-CC CODE	2022-12-06 23:23:38	no	Drug Saf. 2022 Dec 2;:1-11	utf-8	Drug Saf	2,022	10.1007/s40264-022-01256-2	oa_other
==== Front Netw Model Anal Health Inform Bioinform Netw Model Anal Health Inform Bioinform Network Modeling and Analysis in Health Informatics and Bioinformatics 2192-6662 2192-6670 Springer Vienna Vienna 397 10.1007/s13721-022-00397-9 Original Article Structural analysis of SARS-CoV-2 Spike protein variants through graph embedding http://orcid.org/0000-0001-5542-2997 Guzzi Pietro Hiram [email protected] Lomoio Ugo Puccio Barbara Veltri Pierangelo grid.411489.1 0000 0001 2168 2547 Department of Surgical and Medical Sciences, University of Catanzaro, Catanzaro, Italy 2 12 2022 2023 12 1 312 7 2022 21 10 2022 16 11 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Austria, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Since December 2019, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has affected almost all countries. The unprecedented spreading of this virus has led to the insurgence of many variants that impact protein sequence and structure that need continuous monitoring and analysis of the sequences to understand the genetic evolution and to prevent possible dangerous outcomes. Some variants causing the modification of the structure of the proteins, such as the Spike protein S, need to be monitored. Protein contact networks (PCNs) have been recently proposed as a modelling framework for protein structures. In such a framework, the protein structure is represented as an unweighted graph whose nodes are the central atoms of the backbones (C-α), and edges connect two atoms falling in the spatial distance between 4 and 7 Å. PCN may also be a data-rich representation since we may add to each node/atom biological and topological information. Such formalism enables the possibility of using algorithms from graph theory to analyze the graph. In particular, we refer to graph embedding methods enabling the analysis of such graphs with deep learning methods. In this work, we explore the possibility of embedding PCN using Graph Neural Networks and then analyze in the embedded space each residue to distinguish mutated residues from non-mutated ones. In particular, we analyzed the structure of the Spike protein of the coronavirus. First, we obtained the PCNs of the Spike protein for the wild-type, α, β, and δ variants. Then we used the GraphSage embedding algorithm to obtain an unsupervised embedding. Then we analyzed the point of mutation in the embedded space. Results show the characteristics of the mutation point in the embedding space. issue-copyright-statement© Springer-Verlag GmbH Austria, part of Springer Nature 2023 ==== Body pmcIntroduction Proteins play a prominent role in many biological processes. The molecular structure determines the function of each protein. Structural data about each protein are usually determined from experiments (e.g. X-ray crystallography or NMR; Petrey and Honig 2005). More recently, a set of computational prediction methods (e.g., Jumper and Pritzel 2021; Kukic et al. 2014; Palopoli et al. 2009; Gu et al. 2022) predicting protein structure has been introduced. Protein structures are finally stored in publicly available databases such as the Protein Databank (PDB) (Bittrich et al. 2022). Such data are also useful to unravel many biologically relevant problems, such as the structure-to-function relationship and the interaction among proteins (Eswar et al. 2003). The so-far introduced analysis requires the use of appropriate computational models to represent structures and enable the development of novel algorithms.Fig. 1 Figure shows an example of a PCN. On the right part of the figure, the spatial structure of the 7sbk protein is depicted. After the structure analysis, a graph, as represented on the right, is obtained Protein contact networks (PCNs) emerged as a relevant paradigm for the analysis of protein molecular structures (Di Paola et al. 2013). PCN are networks whose nodes represent the C-α atoms of the backbone of proteins, while edges represent a relative spatial distance among 4 and 8 Å. Figure 1 depicts an example of a protein structure and fragment of the resulting graph. Topological descriptors of PCNs, such as centrality measures, are used to discover protein properties, even at the sub-molecular level. Protein modularity, for instance, is specifically suited to identify modular domains in a structure, whose mutual interactions are responsible for allosteric regulation, i.e., the protein structure adaptation to environmental cues (Khan and Ghosh 2015; Das et al. 2021). Existing approaches use spectral clustering to identify network modules that correspond to allosteric regions (Tasdighian et al. 2014). Despite the relevance, these approaches are based only on network structure. Thus, they cannot gather biological and biochemical information broadly available for both nodes and edges. Moreover, classical clustering approaches are inherently transductive since they need to re-analyze all the networks in the presence of modification in both nodes and edges. Many approaches have recently demonstrated that graph structures may be efficiently mapped into latent spaces and then analyzed using deep learning and data-mining methods. Such processes, known as embedding methods, map graph nodes into a so-called embedding space, and the transformation preserves node-similarity (Hamilton et al. 2017b). There exist different node-embedding methods. A large class of methods were based only on the analysis of the topology of the input graph [such as node2vec (Grover and Leskovec 2016), deep walk (Perozzi et al. 2014)]. These methods had two main drawbacks: (i) the impossibility of including information related to nodes; (ii) the need to recalculate the whole embedding in case of graph changes. Subsequent methods overcame these limitations by leveraging the computational power of an ad-hoc-developed neural network. GraphSage (Hamilton et al. 2017b) is a general framework that can leverage node features (e.g. attributes associated with each node) to generate node embeddings. It is based on an inductive process, so it can generate node embedding for unseen nodes without requiring the analysis of the whole graph again. It is based on learning a function that generates embeddings by the analysis of the neighbourhood of each node aggregating the features. It has been used for node classification and clustering tasks with good performances. Here we integrate the previous methods into a single framework of analysis. Our framework is based on the integration of existing software modules for the whole process of the analysis: Creation of the Protein Contact Network Embedding and analysis of the network Visualization of the results We apply the framework by presenting a case study of the analysis of the structure of the Spike protein of SARS-CoV-2 (Guzzi et al. 2020; Ortuso et al. 2021). SARS-CoV-2, which caused the recent pandemic, presents many sequence mutations that impact its protein structure. Among the others, mutations of Spike protein are particularly of interest since they impact the transmission of the virus. To the best of our knowledge, existing work does not face the analysis of such mutation on the embedding space. The aims of this work are:Providing a mechanicist framework for the analysis of the structure of PCN in general; applying the framework to study structures of the variants of the Spike protein; contributing to elucidating the differences among variants of the SARS-CoV-2 proteins. Therefore we first consider protein structures of the wild-type (i.e. unmutated) Spike protein and the structures of main variants: alpha, beta, delta, and omicron (Eskandarzade et al. 2022; Gordon et al. 2020; Kumar Das et al. 2021; Ortuso et al. 2022). Then we obtain the PCN representation for each structure. Finally, we map each structure into the embedding space using GraphSage and analyze the differences between mutated and unmutated residues. We train an unsupervised learning model, aiming to distinct in the topological space points of mutation from other points. Results evidence that PCN of variants are globally different, while more investigation needs to be performed to characterise these points better. The paper is structured as follows: Sect. 2 introduces the Protein Interaction Network formalism. Section 3 discusses briefly the state-of-the-art approaches for node embedding. Then, Sect. 4 presents the proposed framework, its architecture and main modules. Finally, Sect. 5 concludes the paper. Protein interaction networks A protein structure can be represented as a complex three-dimensional object, formally defined by its atoms’ coordinates in 3D space. Despite the large availability of protein molecular structures data, there are yet many problems regarding the relationship between protein structures and their functions. For this reason, it is necessary to define simple descriptors that can describe protein structures with few numerical variables. Structure and function are based on the complex network of inter-residue interactions, where residues are identified by aminoacids sequences (Di Paola et al. 2013). Therefore, the residues interactions are used to define protein interaction networks. Protein interaction networks are thus used to study protein functions. The most simple choice to define networks is to represent the protein structure using α-carbon location. The spatial position of Cα is still reminiscent of the protein backbone, allowing us to highlight the three-dimensional structure’s most important characteristics. Starting from the Cα spatial distribution, a distance matrix d is evaluated where each di,j represents the Euclidean distance in the 3D space between the i-th and j-th residues, defined as1 di,j=((xi-xj)2)+((yi-yj)2)+(zi-zj)2) where (xi,yi,zi) and (xj,yj,zj) respectively are the Cartesian coordinates of residue i and j. Matrix d is used to define a Protein Contact Network concept, which is an alternative and different representation of using graph-based models to represent protein structures. A graph is the most natural structure to represent proteins, where nodes (or vertices) are the protein residues and links (or edges) between the i-th and the j-th nodes (residues) represent residue contacts. In the graph representation, there exists a link between two residues i and j if the distance between two residues (i.e., di,j) is higher than 4 and lower than 8 Å. The lower end excludes all covalent bonds, which are not sensitive to environment change (so to protein functionality). In contrast, the upper end removes weaker non-covalent bonds (so not significant for protein functionality). At this point, it is possible to build up adjacency matrix A, whose generic element is defined as:2 Aij=1if4Å≤dij≤8Å0otherwise. A graph’s adjacency matrix is unique regarding the ordering nodes. In the case of proteins, in which the order of nodes (residues) corresponds to the residues sequence (primary structure) it can be said that its corresponding network is unique: this establishes a one-to-one correspondence between protein and its network. Graph embeddings Graph embedding approaches represented an answer to the primary challenge within machine learning: finding a way to represent or encode graph structure so that the machine learning model can easily exploit it. These approaches, that are usually referred to as graph representation learning or graph-embedding, automatically learn to encode graph structure into low-dimensional embeddings, using techniques based on deep learning and nonlinear dimensionality reduction (Hamilton et al. 2017a; Agapito et al. 2019; Guzzi and Zitnik 2022). The main purpose of graph embedding methods is to encode nodes in a latent vector space which means packaging the properties of each node into a vector with a smaller size. The embeddings learned can also support graph analysis much faster and more accurately compared to the direct execution of such tasks in the domains of complex high-dimensional graphs. Considering, for instance, node embedding, the aim of the mapping is to associate each node (and the associated features) to a low-dimensional vector. These vector spaces correspond to a notion of similarity by preserving a graph’s inherent properties and structure, i.e., similar nodes in the original graph space will be closer to each other in latent vector space. The generated embeddings reflect a network’s updated features that carry the non-linear graph information. Node embedding techniques generate low-dimensional vectors by solving an optimization problem that follows an unsupervised learning schema. Based on the approach to generate embeddings, node embedding methods can be categorized into three major categories: (1) Matrix-Factorization, (2) Random walks, and (3) Graph Neural Networks. Embedding methods that fall under the matrix factorization and random-walk category are known as shallow embedding methods. They are hard-engineered, transductive and often fail to capture node attribute information. In contrast, the graph neural network-based embedding methods are known as deep graph encoders as they produce better representations by specifically involving in deep, multilayered approach for learning or training mechanisms. In particular, the Random Walks-based approaches to learn node embeddings described over a walk with a successive number of random steps in a network. The Random Walks incorporate local and higher-order topological neighbourhood information of a network. The key idea is to derive the similarity between two nodes based on the co-occurrence of nodes in the respective random walks by observing that two similar nodes have a greater chance of having similar random walks. Different methods based on random walks have been developed depending on the strategy used to calculate similarity (e.g. the way to simulate random walks). DeepWalk (Perozzi et al. 2014) introduces unsupervised feature learning on graph data by incorporating truncated random walks to learn latent representations. The method exploits structural regularities and processes random walks equivalent to sentences in neural language modelling. Node2Vec (Grover and Leskovec 2016), which is a modified version of Deepwalk, samples the sequence of random walk based on DFS (depth-first-search) and BFS (Breadth-FirstSearch) strategies. LINE (Tang et al. 2015) can embed networks of large sizes and arbitrary types: undirected, directed, and weighted. The model carefully designs an objective function that optimizes and preserves both the local and the global structural information of graphs by testing the performance on word analogy, text classification, and node classification. Struct2vec (Ribeiro et al. 2017) model learning latent representations for classification task. The representations are generated via a biased random walk to produce node sequences. All the proposed approaches have two main limitations: (i) they do not take into account data related to nodes (i.e. node features); (ii) they are inherently transductive, so they need to recalculate the whole embedding in case of any graph modification (e.g. node/edge insertion or removal). A set of different approaches have been introduced to overcome these limitation. The first method presented in literature is GraphSAGE (Hamilton et al. 2017a), which uses Graph Convolutional Networks to learn the mappings. GraphSAGE is based on an inductive framework that leverages node feature information (e.g., text attributes), so it can gather such information during the embeddings and efficiently generate node embeddings for previously unseen data without analysing the whole graph. For each node i of the graph, GraphSage can learn the embedding by analyzing the neighbours’ information, e.g. all the nodes at a distance K=2. After the determination of the neighborhood for each node, it use two functions for generating the embedding, aggregation, and concatenation. Aggregation functions accept a neighbourhood as input and combine each neighbour’s embedding with weights to create a neighbourhood embedding. The aggregation function use weights that are shared among all the node of the graph, and such weights are either learned or fixed. The proposed approach Framework architecture The proposed framework is based on some main modules, as represented in Fig. 2. Users may insert the input protein data, encoded as a PDB file through a Graphical User Interface. The GUI is responsible for invoking the building of the Protein Contact Network at first. The PCN-Creation module is realized by wrapping the PCN-Miner libraries (Guzzi et al. 2022a, b; Gu et al. 2022). After creating the network, the user may leverage embedding and subsequent mining libraries to analyse the embeddings. Both libraries are included in our framework by wrapping the Stellar Graph library (Data61 2018). The StellarGraph library offers state-of-the-art algorithms for machine learning on graphs. It provides algorithms of representation learning for nodes and edges; Classification of nodes and edges, and link prediction. StellarGraph is built on TensorFlow 2 and its Keras high-level API, as well as Pandas and NumPy. The framework is implemented in the Python Programming Language.Fig. 2 Architecture of the proposed framework The first step is the creation of a Protein Contact Network from structural data. The PCN-Creation module is responsible for this task. It reads a Protein Data Bank File, and it can build a PCN. For each atom of the C-α backbone the module adds a node into the PCN. Then, all the pairwise distances are calculated. Finally, for each distance among two atoms i, j that fall into the range 4–8 Å, the module adds an edge among nodes i, and j into the network. This step adds to each node the information about the centrality values of the node itself. The PCN-enrichment module is responsible for this task. Currently, we calculate closeness, eigenvector, and betweenness centrality (Guzzi and Milenković 2018). These values are a set of node features we add to each node through the networkX library. Finally, we learn the representation of each PCN through Graphsage in an unsupervised mode. GraphSAGE learns node embeddings in this modality by solving a classification task: nodes are subdivided into positive and negative groups. Positive nodes are generated from the analysis of simulated random walks (i.e. nodes that frequently co-occur in random walks), while negative nodes are randomly selected pairs. The binary classifier predicts the likelihood of co-occurrence in a random walk performed on the graph. The classification task is used to learn an inductive mapping from attributes of nodes and their neighbours to node embedding. We downloaded three protein structures, specified with a code (PDB code), from the Protein Data Bank (PDB https://www.rcsb.org/), an archive of 3D structure data: 7FET (variant alpha), 7SBK (variant Delta), and 7WK2 (variant Omicron). Coordinates of the Carbon -α atoms were used to get PCNs. Starting from 3D structure, we obtained the corresponding PCN for each structure using PCN-Miner. This tool (Zitnik et al. 2018) allows to import the structure in .pdb format, to extract structural information and to build the corrispective PCN. Protein network nodes are built to represent single residues. Links between nodes (residues) are defined if the distance between corresponding residues lies between 4 and 8 Å. This threshold, that PCN-miner allows to set, is chosen to map connections only for relevant non-covalent intra-molecular interactions. Then we calculated for each node following centrality measures: Degree, Eigenvector, Closeness and Betweenness. We then apply GraphSAGE with the standard parameters and we these points using t-sne transformation. Figures 3, 4, and 5, depict the embedding of the alpha, delta, and omicron variant. Each point on the figure represent a residue. Red dots evidences points in which a mutation has been occurred.Fig. 3 t-sne visualisation of node embeddings obtained by GraphSAGE for Spike protein of the alpha variant. Red dots represent site of mutation (colour figure online) Fig. 4 t-sne visualisation of node embeddings obtained by GraphSAGE for Spike protein of the Delta variant. Red dots represent site of mutation (colour figure online) Fig. 5 t-sne visualisation of node embeddings obtained by GraphSAGE for Spike protein of the Omicron variant. Red dots represent site of mutation (colour figure online) The analysis of each map reveals that there is a substantial difference in the PCN of variants. Conversely, there are no appreciable differences in the topological parameters of the mutated variants with respect to those that are preserved. Hence, more deep investigations need to be performed. Conclusion Protein contact networks (PCNs) are a modelling framework for protein structures. In such a framework, the protein structure is represented as an unweighted graph. Graph embedding methods enable to map of nodes into a latent space, including a set of information to each node, and then to analyze such graph with deep learning methods. In this work, we proposed a framework to perform such analysis and to study the mutation of the S protein of SARS-CoV-2. Acknowledgements Authors thank Luisa di Paola and Alessandro Giuliani for fruitful discussion and collaboration during the preparation of this paper Author Contributions For research articles with several authors, a short paragraph specifying their individual contributions must be provided. The following statements should be used “Conceptualization, PHG. and PV.; methodology, PHG.; software, BP and UL.; data curation, UL; writing—original draft preparation, PHG, PV, and BP.; writing—review and editing, PHG and PV. Funding acquisition, PV. All authors have read and agreed to the published version of the manuscript.” Data availability Data are available upon request. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Agapito G Guzzi PH Cannataro M Parallel extraction of association rules from genomics data Appl Math Comput 2019 350 434 446 Bittrich S Rose Y Segura J Lowe R Westbrook JD Duarte JM Burley SK RCSB Protein Data Bank: improved annotation, search and visualization of membrane protein structures archived in the PDB Bioinformatics 2022 38 5 1452 1454 10.1093/bioinformatics/btab813 Das JK Roy S Guzzi PH Analyzing host-viral interactome of SARS-CoV-2 for identifying vulnerable host proteins during COVID-19 pathogenesis Infect Genet Evol 2021 93 104921 10.1016/j.meegid.2021.104921 34004362 Data61 C (2018) Stellargraph machine learning library. https://github.com/stellargraph/stellargraph. Accessed Sept 2022 Di Paola L De Ruvo M Paci P Santoni D Giuliani A Protein contact networks: an emerging paradigm in chemistry Chem Rev 2013 113 3 1598 1613 10.1021/cr3002356 23186336 Eskandarzade N, Ghorbani A, Samarfard S, Diaz J, Guzzi PH, Fariborzi N, Tahmasebi A, Izadpanah K (2022) Network for network concept offers new insights into host-SARS-CoV-2 protein interactions and potential novel targets for developing antiviral drugs. Comput Biol Med 105575 Eswar N John B Mirkovic N Fiser A Ilyin VA Pieper U Stuart AC Marti-Renom MA Madhusudhan MS Yerkovich B Tools for comparative protein structure modeling and analysis Nucleic Acids Res 2003 31 13 3375 3380 10.1093/nar/gkg543 12824331 Gordon DE Jang GM Bouhaddou M Xu J Obernier K White KM O’Meara MJ Rezelj VV Guo JZ Swaney DL A SARS-CoV-2 protein interaction map reveals targets for drug repurposing Nature 2020 583 7816 459 468 10.1038/s41586-020-2286-9 32353859 Grover A, Leskovec J (2016) node2vec: Scalable feature learning for networks. In: Proceedings of the 22nd ACM SIGKDD international conference on Knowledge discovery and data mining, pp 855–864 Gu S Jiang M Guzzi PH Milenković T Modeling multi-scale data via a network of networks Bioinformatics 2022 38 9 2544 2553 10.1093/bioinformatics/btac133 35238343 Guzzi PH Milenković T Survey of local and global biological network alignment: the need to reconcile the two sides of the same coin Brief Bioinform 2018 19 3 472 481 28062413 Guzzi PH Zitnik M Editorial deep learning and graph embeddings for network biology IEEE/ACM Trans Comput Biol Bioinform 2022 19 2 653 654 Guzzi PH Mercatelli D Ceraolo C Giorgi FM Master regulator analysis of the SARS-CoV-2/human interactome J Clin Med 2020 9 4 982 10.3390/jcm9040982 32244779 Guzzi PH, Di Paola L, Giuliani A, Veltri P (2022a) Design and development of PCN-miner: a tool for the analysis of protein contact networks. arXiv preprint. arXiv:2201.05434 Guzzi PH Di Paola L Giuliani A Veltri P PCN-miner: an open-source extensible tool for the analysis of protein contact networks Bioinformatics 2022 38 17 4235 4237 10.1093/bioinformatics/btac450 35799364 Hamilton WL, Ying R, Leskovec J (2017a) Representation learning on graphs: methods and applications. arXiv preprint. arXiv:1709.05584 Hamilton W, Ying Z, Leskovec J (2017b) Inductive representation learning on large graphs. Adv Neural Inf Process Syst 30 Jumper JE Pritzel A Highly accurate protein structure prediction with AlphaFold Nature 2021 596 583 589 10.1038/s41586-021-03819-2 34265844 Khan T Ghosh I Modularity in protein structures: study on all-alpha proteins J Biomol Struct Dyn 2015 33 12 2667 2681 10.1080/07391102.2014.1003969 25669306 Kukic P Mirabello C Tradigo G Walsh I Veltri P Pollastri G Toward an accurate prediction of inter-residue distances in proteins using 2D recursive neural networks BMC Bioinform 2014 15 1 1 15 10.1186/1471-2105-15-6 Kumar Das J Tradigo G Veltri P Guzzi PH Roy S Data science in unveiling COVID-19 pathogenesis and diagnosis: evolutionary origin to drug repurposing Brief Bioinform 2021 22 2 855 872 10.1093/bib/bbaa420 33592108 Ortuso F, Mercatelli D, Guzzi PH, Giorgi FM (2021) Structural genetics of circulating variants affecting the SARS-CoV-2 spike/human ace2 complex. J Biomol Struct Dyn 1–11 Ortuso F Mercatelli D Guzzi PH Giorgi FM Structural genetics of circulating variants affecting the SARS-CoV-2 spike/human ACE2 complex J Biomol Struct Dyn 2022 40 14 6545 6555 10.1080/07391102.2021.1886175 33583326 Palopoli L Rombo SE Terracina G Tradigo G Veltri P Improving protein secondary structure predictions by prediction fusion Inf Fusion 2009 10 3 217 232 10.1016/j.inffus.2008.11.004 Perozzi B, Al-Rfou R, Skiena S (2014) Deepwalk: online learning of social representations. In: Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining, pp 701–710 Petrey D Honig B Protein structure prediction: inroads to biology Mol Cell 2005 20 6 811 819 10.1016/j.molcel.2005.12.005 16364908 Ribeiro LF, Saverese PH, Figueiredo DR (2017) Struc2vec: learning node representations from structural identity. In: Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining. KDD ’17. ACM, New York, pp 385–394. 10.1145/3097983.3098061 Tang J, Qu M, Wang M, Zhang M, Yan J, Mei Q (2015) LINE. In: Proceedings of the 24th international conference on world wide web. International world wide web conferences steering committee. 10.1145/2736277.2741093 Tasdighian S Di Paola L De Ruvo M Paci P Santoni D Palumbo P Mei G Di Venere A Giuliani A Modules identification in protein structures: the topological and geometrical solutions J Chem Inf Model 2014 54 1 159 168 10.1021/ci400218v 24289204 Zitnik M Agrawal M Leskovec J Modeling polypharmacy side effects with graph convolutional networks Bioinformatics 2018 34 13 i457 i466 10.1093/bioinformatics/bty294 29949996	36506261	PMC9718452	NO-CC CODE	2022-12-06 23:23:38	no	Netw Model Anal Health Inform Bioinform. 2023 Dec 2; 12(1):3	utf-8	Netw Model Anal Health Inform Bioinform	2,022	10.1007/s13721-022-00397-9	oa_other
==== Front Drugs Drugs Drugs 0012-6667 1179-1950 Springer International Publishing Cham 36459382 1815 10.1007/s40265-022-01815-y Acknowledgement to Referees Acknowledgement to Referees 2 12 2022 13 © Springer Nature Switzerland AG 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. ==== Body pmcDear Reader Welcome to the final issue of Drugs for 2022. This is a time to reflect on this year’s achievements, and to acknowledge and thank everyone who has contributed to ensure the continued high quality of the journal. I particularly want to acknowledge the continuing impact of the COVID-19 pandemic on all healthcare professionals. The quality of content published in Drugs has been reflected in the most recent Impact Factor of 11.431 and CiteScore™ of 16.3, placing the journal 13th in the Pharmacy and Pharmacology, and 4th in the Pharmacology (Medicine) categories, respectively. Further, Drugs has continued to publish content in a timely manner, with a median time from submission to first decision of 8 days. I would like to begin by thanking the authors of articles published in Drugs over the course of 2022. The skill and dedication of these experts is critical to the continued success of the journal. The quality of published articles is also testament to the diligence of the peer reviewers, and I would like to acknowledge the following individuals who acted as reviewers for Drugs in the last 12 months: Ghanashyam Acharya, USA Sascha Alles, USA Rustam Al-Shahi Salman, UK Talha M. Al-Shawaf, UK Danieli Andrade, Brazil Panagiota Anyfanti, Greece Kazuhiro Araki, Japan Carmen Ardanuy, Spain Wilbert S. Aronow, USA Rodrigo Azevedo Loiola, France Zsuzsa Bagoly, Hungary Monica Baiula, Italy Dipika Bansal, India Kalyani Barve, India Viviana Bazan, Italy Meral Beksac, Turkey Joel Belmin, France Daniela Berg, Germany Deborah Blumenthal, Israel Josip A. Borovac, Croatia Dave Boucher, UK David Boyer, USA Johan Burisch, Denmark Christopher F. Carpenter, USA Louise Carton, France Jose V. Castell, Spain Maddalena Centanni, Sweden Hong Y. Choi, Canada Junlan Chuan, China Marcin Chwistek, USA Patricia Cristofaro, USA Alessandro Dalpiaz, Italy Lucio D'Anna, UK Larry Danziger, USA Rajan Darjee, Australia Amaury Daste, France Adem Dawed, UK Stefano de Biase, Italy Theo M. de Reijke, The Netherlands Mette S. Deleuran, Denmark Costantino Di Carlo, Italy Hans-Christoph Diener, Germany M.A. Dimopoulos, Greece Cécile Droz-Perroteau, France Urban Emmenegger, Canada Narendranth Epperla, USA Martin Faehling, Germany Sonia Fargue, USA Bruno Fattizzo, Italy Steven R. Feldman, USA Javier Fernandez-Ruiz, Spain Simone Ferrero, Italy Charles Flexner, USA Ronan Flippot, France Ana Fraguas-Sanchez, Spain Joseph H. Friedman, USA Brian M. Frier, UK Walter Fries, Italy Thorsten Fuereder, Austria Reinhard Gabathuler, UK Om P. Ganda, USA Pedro J. Garcia Ruiz, Spain Moshe Gatt, Israel Daniel Gaudet, Canada Eleni Gavriilaki, Greece Franco Gemignani, Italy Nils E. Gilhus, Norway Natividad Gomez-Roman, UK Anders Gottsäter, Sweden Morag Griffin, UK John Gubatan, USA Mario Habek, Croatia Hans-Peter Hartung, Germany Roxanne Hastie, Australia H. Hayashi, Japan Yujiro Hayashi, USA Qiyu He, China Thatcher Heumann, USA Martin Hoenigl, USA Christian Holscher, China Jorge Ibanez, USA Luís S. Inês, Portugal Giovanni Iolascon, Italy Rohit Jain, USA Meghna Jani, UK Emmanuele A. Jannini, Italy Matti S. Jauhiainen, Finland Reed Johnson, USA Páll Karlsson, Denmark Bernd Kasper, Germany Carol A. Kauffman, USA Chul Kim, USA Laura Kimberly, USA Denise Kreutzwiser, Canada Christine C. Krieger, USA Yoshiaki Kubota, Japan Mikael Landén, Sweden Simona Lattanzi, Italy Matthew Lee, UK Young Ran Lee, USA Antoine Lemaire, France Katarzyna Leszczynska, Poland Russel E. Lewis, Italy James H. Lewis, USA Li Li, China Evangelos N. Liberopoulos, Greece David Liew, Australia Chia-Chi Lin, Taiwan, Republic of China Giuseppe Losurdo, Italy Jörn Lötsch, Germany Giancarlo Lucchetti, Brazil M. Isabel Lucena, Spain Iain C. Macdougall, UK Giuseppe Maltese, UK Giorgia Mandrile, Italy Christophe Mariat, France Eric Marietta, USA Paolo Martelletti, Italy Rami Massie, Canada Stephanie Mathieson, Australia Hisako Matsumoto, Japan Benjamin L. Maughan, USA Kelly E. McCann, USA James Millard, UK Alice Mims, USA Peter G.M. Mol, The Netherlands Filippo F. Montemurro, Italy Jarrod J. Mousa, USA Eli Muchtar, USA Debabrata Mukherjee, USA Dirk Müller-Wieland, Germany Misako Nagasaka, USA Takehiro Nakai, Japan Beatrice Nardone, USA Walaa A. Negm, Egypt Sarah Nevitt, UK Stephen J. Nicholls, Australia Lindsay E. Nicolle, Canada Alicia Okines, UK Carlos J. Orihuela, USA Pedro Orihuela, Chile Tobias Overbeck, Germany Giovanni Palladini, Italy Hardev Pandha, UK Ioannis Parodis, Sweden Ana V. Pejčić, Serbia Charles A. Peloquin, USA Helen Petsky, Australia JoAnn V. Pinkerton, USA Bertram Pitt, USA Mathias Pletz, Germany Werner Poewe, Austria Antonio A.E. Pontiroli, Italy Rashmi Prasad, Sweden Juergen Prattes, Austria Emanuel Raschi, Italy Olivier Rascol, France Hennie Raterman, The Netherlands Gerard Reach, France Giuseppe Remuzzi, Italy Keith A. Rodvold, USA Jorge A. Roman-Blas, Spain Stefano Romano, UK Danial Roshandel, Australia Sherin J. Rouhani, USA Antoine Rousseau, France Piero Ruggenenti, Italy Alessandro Russo, Italy Helio S. Sader, USA Else Charlotte Sandset, Norway Naomi Schlesinger, USA Elaine Shum, USA Michael A. Singer, USA Tania C. Sorrell, Australia Pierre P.J. Souquet, France Stacey Stein, USA Thomas E. Stinchcombe, USA Carmen Suay-Corredera, Spain Jin-Yu Sun, China Raghav Sundar, Singapore Michael Szarek, USA Sasha Taleban, USA Astrid J. Terkelsen, Denmark Sebastian Thams, Sweden Florence Thibaut, France René Thieme, Germany Martin J. Thomas, UK Vicente E. Torres, USA Antonella Tosti, USA Anne-Priscille Trouvin, France Elena Tsourdi, Germany Kazuo Tsubota, Japan Christopher P. Twine, UK K. Utsugisawa, Japan Bijay Vaidya, UK Jennifer Vaughn, USA Viviane C. Veiga, Brazil Jeffery M. Venner, Canada Michael Verret, Canada Liza Villaruz, USA Maj Vinberg, Denmark Ioannis Voutsadakis, Canada Sarah Wall, USA Jing Wang, China Zhengqiang Wang, USA Michael H. Weisman, USA Ilene C. Weitz, USA Ethel Weld, USA Fletcher A. White, USA Cynthia Willey, USA John Windsor, New Zealand Jay B. Wish, USA Angel Wong, UK Clifford J. Woolf, USA Emily Wright, Australia Jing Xue, China Yoon-Sim Yap, Singapore Chi-Hsiao Yeh, Taiwan, Republic of China Chun-Nan Yeh, Taiwan, Republic of China Moussa B.H. Youdim, Israel Rafael Zaragoza, Spain I am also extremely grateful to the members of the journal’s Honorary Editorial Board, who have acted as peer reviewers and authors, and have provided guidance on journal content. This year, we said to farewell some long-standing members, Professors Paul Abrams, Luisa de Angelis, Christine Katlama, Wasaburo Koizumi, Hartmut Lode, Francesco Montorsi and Geoffrey Playford, and thank them for their outstanding contributions to the journal. We were also very fortunate to welcome several new members to the board this year, and thank them for their willingness to contribute to the success of the journal: Sabina Antoniu, Vera Bril, Christoph Correll, Lucia del Vecchio, Roy Fleischmann, Peter Goadsby, Johnny Mahlangu, Bradley McGregor, Ole Haagen Nielsen, Michael Nurmohamed, Kimberly Scarsi, Einar Sigurdsson, Elizabeth Smyth, Yu Sunakawa and Konrad Talbot. The editorial program for 2023 is well under way, and we are looking forward to continuing to bring you many high-quality and authoritative articles in the field of clinical drug treatment over the coming year. On behalf of all the Section Editors and staff, thank you so much for your continued support. Best wishes Dene C. Peters Editor, Drugs	0	PMC9718453	NO-CC CODE	2022-12-11 23:16:16	no	Drugs. 2022 Dec 2; 82(18):1691-1693	utf-8	Drugs	2,022	10.1007/s40265-022-01815-y	oa_other
==== Front Curr Psychol Curr Psychol Current Psychology (New Brunswick, N.j.) 1046-1310 1936-4733 Springer US New York 4081 10.1007/s12144-022-04081-z Article Self-rated mental health in the transition to adulthood predicts depressive symptoms in midlife http://orcid.org/0000-0003-2371-1449 Galambos Nancy L. [email protected] 1 http://orcid.org/0000-0001-9440-4839 Johnson Matthew D. [email protected] 2 http://orcid.org/0000-0003-0445-1573 Krahn Harvey J. [email protected] 3 1 grid.17089.37 0000 0001 2190 316X Department of Psychology, University of Alberta, P217 Biological Sciences Building, Edmonton, AB T6G 2E9 Canada 2 grid.17089.37 0000 0001 2190 316X Department of Human Ecology, University of Alberta, Edmonton, AB Canada 3 grid.17089.37 0000 0001 2190 316X Department of Sociology, University of Alberta, Edmonton, AB Canada 2 12 2022 112 24 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Self-rated mental health (SRMH), a single item asking individuals to evaluate their mental or emotional health, is included in some surveys as an indicator of risk for mental disorders and to monitor population health, yet little longitudinal research examines how well it predicts future outcomes. Following a life course perspective, the current longitudinal study of 502 Canadian high school seniors tracked into midlife examined to what extent SRMH at ages 20, 25, and 32 years predicted depressive symptoms at ages 43 and 50. Hierarchical linear regressions showed that lower SRMH at age 25 and at 32 years was a significant predictor of higher levels of depressive symptoms at ages 43 and 50, even when controlling for sex, participant education, marital/cohabitation status, self-rated physical health, and baseline depressive symptoms. The results provide evidence that SRMH assessed during the transition to adulthood may be useful as a broad and powerful measure of risk for mental health problems decades into the future. Keywords Self-rated mental health Depressive symptoms Life course Transition to adulthood Midlife http://dx.doi.org/10.13039/501100000155 Social Sciences and Humanities Research Council of Canada 435-2014-0076 Krahn Harvey J. ==== Body pmcIntroduction Mental health at any point in the lifespan is an important concern for individuals, families, communities, and public health professionals and policymakers. Self-rated mental health (SRMH), a one-item measure asking individuals to evaluate their mental or emotional health on a scale from “poor” to “excellent”, is included in some national surveys as an indicator of general risk for mental disorders and to monitor population health (e.g., Canadian Community Health Survey, Capaldi et al., 2021; Israel National Health Study, Levinson & Kaplan, 2014; U.S. Medical Expenditure Panel Survey, Nguyen et al., 2015). Although this item has been increasingly used since the mid-1990s (Ahmad et al., 2014), it is an understudied measure of well-being (McAlpine et al., 2018). We know that poorer SRMH is concurrently associated with depression, psychological distress, and other measures of mental health problems (Ahmad et al., 2014; Fleishman & Zuvekas, 2007; Hoff et al., 1997; Mawani & Gilmour, 2010), but little is understood about how well it foretells future mental health. Linking earlier SRMH with later mental health issues could inform healthcare research and policy and also provide insight into the predictive validity of this one-item measure. As such, there is a critical need for longitudinal studies (Ahmad et al., 2014; McAlpine et al., 2018). The objective of the current longitudinal study is to assess the predictive validity of SRMH by examining to what extent SRMH in the transition to adulthood (age 20 and 25) and early adulthood (age 32) predicts age 43 and age 50 depressive symptoms, an important indicator of midlife mental health. The usefulness of SRMH for anticipating future mental health problems rests on the underlying construct it measures, a topic of ongoing discussion (Hoff et al., 1997; Levinson & Kaplan, 2014; Magwene et al., 2017). Research shows that while poor SRMH is correlated significantly with psychiatric diagnoses, it does not serve as a proxy for specific mental illnesses. SRMH may capture generalized psychological unease or distress, and it could pick up on subclinical symptoms that have yet to solidify into a particular disturbance such as depression or anxiety (Ahmad et al., 2014; Casu & Gremigni, 2019; Fleishman & Zuvekas, 2007; Hoff et al., 1997). Some researchers have argued in favor of SRMH as a measure of subjective well-being or mental health more than of mental illness (Levinson & Kaplan, 2014; Lombardo et al., 2018). Furthermore, SRMH appears to be an indicator of broad perceptions of one’s mental health, shaped by factors additional to any symptoms (e.g., physical health and limitations, life circumstances). These perceptions are believed to play an important role in help seeking, treatment compliance, and recovery from mental health conditions (Mawani & Gilmour, 2010; McAlpine et al., 2018; Nguyen et al., 2015; Zuvekas & Fleishman, 2008). Our examination of the longer-term association between SRMH in the 20s and 30s and depressive symptoms in midlife will add to knowledge about SRMH as an indicator of developing mental health problems. A life course perspective on mental health Our study is guided by a life course perspective on mental health. A core tenet is that human development is a lifelong process in which experiences of individuals in their familial, social, economic, and historical contexts are antecedents that set the stage for later experiences and well-being (Crosnoe & Elder, 2004; Shanahan, 2000). As such, understanding linkages between indicators of mental health across life stages is an important objective requiring longitudinal research over decades. A related assumption is that transitional periods, in which individuals may experience multiple role changes and challenges, provide opportunities for and threats to well-being, highlighting, for example, the transition to adulthood (which takes place through the 20s) as an important period of some vulnerability for the emergence of mental health problems (George, 1999). A life course approach considers risk (e.g., low socioeconomic status) and protective factors (e.g., social support) in describing pathways to mental health in adulthood. Poorer mental health is likely when there is an accumulation over time of risks along with few resources or protections that can mitigate the risks and promote mental health. Finally, a life course perspective assumes human agency, that is, that individuals are capable of actively making decisions about their futures and show resilience, although there are significant social structural constraints (e.g., gender bias, racism, poverty) that can place obstacles in the way and lead to psychopathology (Crosnoe & Elder, 2004; George, 1999; Mechanic & McAlpine, 2011; Shanahan, 2000). Mental health in the transition to adulthood to early adulthood to midlife Preceded by competencies gained and adversities experienced in childhood and adolescence, the transition to adulthood is a key period during which young people may access and acquire further resources that set them on a path toward a healthy and stable adulthood or they may flounder due to actions, cumulative vulnerabilities, or new circumstances that negatively affect their futures (Krahn et al., 2015; Masten & Tellegen, 2012). Developmental advances such as improved perspective taking and regulation of emotions are necessary for adapting to the demands of adulthood, which include developmental tasks such as leaving home, completing one’s education, finding a romantic partner, securing employment, establishing financial independence, and becoming a parent (Krahn et al., 2015; Schulenberg et al., 2004). The demands of this period may exceed available resources for meeting them. Indeed, a U.S. longitudinal study of 22- to 77-year-olds followed for 20 years revealed that daily stress exposure and stress reactivity were highest among individuals in their 20s, which declined as they aged into their 30s and 40s (Almeida et al., 2022). Anxiety, depression, and substance use disorders often emerge in the transition to adulthood, with troubling prevalence rates (George, 1999; Tanner et al., 2007). A 10-year longitudinal study of a Norwegian sample, for example, found that the 12-month prevalence of any mental disorder on the Munich Composite International Diagnostic Interview (M-CIDI; Wittchen & Pfister, 1997) was 32.2% of women and 19.8% of men among participants in their 20s. Anxiety disorders were most common in both sexes, followed by depressive disorders for women and alcohol use disorders for men (Gustavson et al., 2018). Although longitudinal research following people from their 20s into early adulthood (their 30s) is scarce, the Almeida et al. (2022) study suggests that early adulthood may be relatively less stressful. In support, Gustavson et al. (2018) found that the 12-month prevalence of any mental disorder declined by the time women (25.3%) and men (12.9%) were in their 30s. Furthermore, according to scores on the Structured Diagnostic Interview for Psychopathologic Somatic Syndromes (SPIKE; Angst & Dobler-Mikola, 1985), the prospective Zurich Cohort Study documented decreased 12-month prevalence rates for major depression, minor depression, and depressive symptoms for participants in their 30s compared to their high points in the 20s (Merikangas et al., 2003). While depression may decrease on average into early adulthood, there is also stability in mental health problems from the 20s into the 30s. A diagnosis of anxiety or depressive disorder in the 20s, for example, increases by manyfold the risk of a repeat diagnosis in the 30s (Gustavson et al., 2018). Nevertheless, accomplishments and competencies gained during the transition to adulthood (e.g., fulfilling educational and employment expectations; getting married) may offset some of its risks, leading to positive mental health in early adulthood and beyond (Howard et al., 2010; Johnson et al., 2017; Masten et al., 2010; Mossakowski, 2011). Midlife, which is generally considered to encompass ages 40 to 60 (Lachman et al., 2015), follows early adulthood and poses both opportunities for and risks to well-being. In midlife many people may experience the positive social and emotional impacts of their childrearing years and work productivity. At the same time, challenges such as launching children, caring for aging parents, maintaining a career, planning for retirement, and coping with emerging physical health issues can be turning points for worsening mental health (Infurna et al., 2020). Although there is little support for a generalized “midlife crisis” (Galambos et al., 2020), the longitudinal Dunedin Birth Cohort Study showed that by age 45 nearly everyone will have experienced at least one mental disorder (Caspi et al., 2020), meeting criteria of the Diagnostic and Statistical Manual of Mental Disorders (e.g., DSM-5; American Psychiatric Association, 2013). Major depressive episodes (MDEs) are especially common, with a lifetime prevalence in the U.S. National Comorbidity Survey Replication of about one in five in the 18–49-year-old age group (Kessler et al., 2010), as assessed by the CIDI (Kessler & Ustün, 2004). Furthermore, Infurna et al. (2020) concluded that middle-aged adults, particularly in some segments of the population (e.g., lower SES, women) have elevated rates of depression and distress compared to younger and older adults. As such, it is important to identify markers of distress earlier in life that may foreshadow depressive symptoms in midlife. SRMH in the transition to adulthood or early adulthood might signal the potential for future depression. Correlates of self-rated mental health SRMH is commonly assessed with only one item, but this item may be a potent subjective indicator that provides insight into self-awareness and feelings about one’s well-being (McAlpine et al., 2018). A scoping review of the literature found that poorer SRMH was associated with psychological distress and serious mental health conditions such as major depression, based on a variety of interview and questionnaire instruments (Ahmad et al., 2014). Mawani and Gilmour (2010), for example, reported that a national sample of Canadians (age 15 and above) who experienced an MDE in the past month, identified by scores on the CIDI, had 26 times the odds of rating their mental health as poor/fair compared to those with no MDE in the past month. Similarly, participants with high levels of psychological distress on the K-6 (Kessler et al., 2002) were 36 times more likely to report poor/fair SRMH compared to those with few symptoms of psychological distress (Mawani & Gilmour, 2010). In a sample of Canadian adolescents in British Columbia, Sawatzky et al. (2010) documented a significant association between SRMH and depressive symptoms, assessed with 12 items from the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977). In a sample of older (ages 60 to 98) Korean Americans residing in New York City, SRMH was a significant predictor of three different measures of depressive symptoms (CESD-10, Andresen et al., 1994; PHQ-9, Kroenke et al., 2001; Geriatric Depression Scale–Short Form, Sheikh & Yesavage, 1986), after controlling for variables such as education, gender, and marital status (Jang et al., 2012). Poorer SRMH is also related to lower life satisfaction, sense of community, and belongingness (Ahmad et al., 2014; Lombardo et al., 2018; Palis et al., 2018; Sawatzky et al., 2010), higher experience of family violence (Farhat et al., 2022), and subjectively poorer physical health and the presence of chronic physical conditions (Fleishman & Zuvekas, 2007; McAlpine et al., 2018; Sawatzky et al., 2010). SRMH in Canada was lower during the COVID-19 pandemic than before (Capaldi et al., 2021), and individuals who reported lower SRMH were also more likely to use health and mental health services (Ahmad et al., 2014; McAlpine et al., 2018; Nguyen et al., 2015). As a holistic assessment of one’s mental health, SRMH may be useful for screening and potentially identifying those in need of treatment (Casu & Gremigni, 2019; Mawani & Gilmour, 2010; McAlpine et al., 2018). A paucity of longitudinal research on SRMH limits knowledge about how well it forecasts future well-being, information that is essential for understanding mental health across the life cycle as well as for evaluating the predictive value of this single item. In one exception, a prospective study of a community sample of adults in New Haven, Connecticut showed that participants with poor SRMH were significantly more likely than those with better SRMH to experience an MDE in the next 12 months (assessed with the DIS; Robins et al., 1981), controlling for confounders such as age, gender, and depression at baseline (Hoff et al., 1997). In a U.S. sample of adults with depression (based on the PHQ-2; Kroenke et al., 2003) or severe psychological distress on the K6 at baseline, participants who rated their SRMH as good/very good/excellent had 30% lower odds of experiencing depression or serious psychological distress one year later, controlling for sociodemographic characteristics, mental health care utilization, and initial symptoms (McAlpine et al., 2018). In U.S. students tracked across their first year of university, higher SRMH at baseline predicted lower loneliness in fall and spring semesters on a brief version of the UCLA Loneliness Scale (Roberts et al., 1993), less anxiety in fall and spring (Generalized Anxiety Disorders 7-item Scale; Spitzer et al., 2006), and higher self-reported GPA in the fall semester; SRMH was not related to depressive symptoms (CESD-10; Andresen et al., 1994) in either semester (Jones & Schreier, 2021). In a prospective cohort study of French participants (ages 23 to 93 years), average SRMH was higher before the COVID-19 pandemic than during lockdown, with women, young to middle-aged adults (23 to 49 years), and the elderly (age 70 and over) more negatively affected (Ramiz et al., 2021). Finally, in a U.S. sample of caregivers for the elderly there were steeper declines in SRMH over two years among those who initially reported lower SRMH (Haug et al., 1999). These few longitudinal studies suggest that single-item SRMH might be sensitive to emerging mental health problems, but we do not know whether SRMH predicts well-being more than two years into the future. Given cross-sectional results supporting a link between SRMH, on the one hand, and major depression and depressive symptoms, on the other (e.g., Mawani & Gilmour, 2010; Sawatzky et al., 2010), along with some longitudinal evidence that SRMH predicts future episodes of major depression and psychological distress (Hoff et al., 1997; McAlpine et al., 2018), it makes sense to ask to what extent SRMH in the transition to adulthood and early adulthood predicts depressive symptoms in midlife. SRMH may provide a window into distress that does not meet formal criteria for psychopathology but that may be a harbinger of future depressive symptoms. The current study Influenced by a life course perspective, we examine to what extent single-item SRMH measured in the same individuals at two points during the critical transition to adulthood (i.e., ages 20 and 25) and once in early adulthood (i.e., age 32) anticipates depressive symptoms up to 30 years later in middle adulthood (i.e., ages 43 and 50). The data are drawn from a sample of high school students living in a city in western Canada who were surveyed repeatedly over three decades. First, we assess the relative strength of SRMH scores in the transition to adulthood and early adulthood for predicting midlife depressive symptoms without controlling for possible confounding variables. Second, we examine the impact of SRMH, controlling for participants’ sex, whether participants attained a university degree by age 43 or 50, marital status and self-rated physical health (SRPH) at ages 43 or 50, and depressive symptoms at baseline. Confounding variables were selected because they could be associated with SMRH and/or depressive symptoms and were common controls in previous SRMH studies (e.g., Ahmad et al., 2014; Hoff et al., 1997). Sex of participant is controlled because women are more likely than men to experience major depression (e.g., Hoff et al., 1997; Kessler et al., 2010), and a few studies have found poorer SRMH among women (Assari & Lankarani, 2017; Cohen & Patten, 2005). We control for whether participants held a university degree, an indicator of socioeconomic status (SES), as higher education and SES are related to better SRMH (Assari & Lankarani, 2017; Jang et al., 2012; Su et al., 2021; Zuvekas & Fleishman, 2008). Married people and those with partners may have better SRMH than non-partnered individuals, suggesting the importance of marital status (Chiu et al., 2017; Hoff et al., 1997; McAlpine et al., 2018). SRPH is controlled due to significant positive correlations of SRMH with SRPH (Levinson & Kaplan, 2014; Sawatzky et al., 2010). Finally, Hoff et al. (1997) showed that in the New Haven Catchment Study, the best predictor of a depressive episode in the following year was major depression at baseline. Thus, baseline (age 18) depressive symptomatology is controlled to determine whether SRMH indicators in the transition to adulthood are robust enough to explain variability in midlife depressive symptoms once adolescent depression is also in the model. Controlling for possible confounding variables provides a conservative test of the SRMH-depressive symptoms connection across decades. Method Participants In Spring 1985 (baseline; Wave 1), 983 Grade 12 students (age 18) were surveyed in six public high schools selected to represent middle- and lower-income neighborhoods in Edmonton, Alberta as part of the Edmonton Transitions Study (ETS). Questionnaires completed in class included indicators of family background, educational and work experiences, and well-being. Over 90% (n = 894) consented to be contacted to participate in the future. Follow-up questionnaires were mailed to consenters in 1986 (age 19; n = 665 returned), and again in 1987 (age 20; n = 547), 1989 (age 22; n = 504), and 1992 (age 25; n = 404) but only to those who participated in each previous wave. In 1999, follow-up telephone surveys targeted all consenting Wave 1 participants (age 32; n = 509), as did telephone and web-based surveys in 2010 (age 43; n = 405), and 2017 (age 50; n = 404). Of the original sample (n = 983), 20% participated only at Wave 1, 18% completed all eight waves, and 42% completed between four and seven waves. This study received research ethics approval prior to each data collection; the most recent approval (June 26, 2017) was granted by the University of Alberta Research Ethics Board 2 (application title: Transitions to Midlife: 32 Year Follow-up of the Class of 1985; protocol number: Pro00074272). Given our focus on midlife depressive symptoms and the fact that the Center for Epidemiologic Studies 10-item Depression Scale (CESD-10; Andresen et al., 1994) was administered only at age 43 and age 50, the study sample was filtered to include 502 individuals who had CESD-10 scores available at age 43 and/or age 50: n = 303 with the CESD-10 at age 43 and 50; n = 99 only at age 43; and n = 100 with the CESD-10 only at age 50. Table 1 presents demographic characteristics at baseline for the original full sample as well as the reduced sample of 502 participants used for analyses in the current study. These figures show participants were predominately White, born in Canada, and living in households where no parent held a university degree. Census data (McVey & Kalbach, 1995) showed that the original full sample was representative of western Canadian urban youth born in 1967 on parents’ education, immigration status, and ethnicity. By midlife (age 43 or 50), 40% of the participants had attained a university (e.g., bachelor’s) degree or higher. Three-fourths were married or cohabiting at age 43 (76%) and age 50 (77%), and the majority were parents (age 43: 78%; age 50: 82%). A large majority were employed at age 43 and 50 (92% and 91%, respectively) with about 2% unemployed (looking for work) at each age.Table 1 Demographic Characteristics of Original Sample and Study Sample Variable Original Sample N = 983 Study Sample N = 502 Sex Male 53% 50% Female 47% 50% High School Lower income 54% 48% Middle income 46% 52% Parent Education No university degree 74% 72% One or both parents with university degree 26% 28% Immigration Status Born in Canada 80% 86% Born outside Canada 20% 14% Ethnicity White 85% 89% Asian 11% 7% Mixed ethnicity/other non-White 2% 3% Indigenous 1% < 1% Black 1% < 1% Study sample consists of participants with CESD-10 scores at age 43 and/or age 50. Comparisons of participants included in the current study sample (n = 502) with excluded participants (n = 481) were conducted on the key variables used in our regression analyses. A four-item measure of depressive symptoms at age 18 was not significantly different for continuing participants versus those who attrited. SRMH was not assessed until age 20, so a baseline comparison could not be made, but included and excluded participants did not differ on SRMH at ages 20, 25, or 32 years, nor did they differ significantly on SRPH at those ages. With respect to control variables, women and individuals with at least one university-educated parent were more likely to participate in midlife (age 43 and/or 50). Measures Depressive symptoms At ages 43 and 50, the mean of 10 items (Andresen et al., 1994) from the 20-item Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977) assessed “how often in the past few months” respondents experienced symptoms of depression such as feeling “depressed” or “lonely.” Items were evaluated on a scale from 1 (never) to 5 (almost always). Internal consistency was high at age 43 (α = 0.87) and age 50 (α = 0.87). The CESD-10 has been widely used and validated in diverse samples around the world as a screening instrument for probable clinical depression (e.g., Baron et al., 2017). Because we modified the timeframe of symptoms (over a few months rather than a week), the response scale, and the scoring (calculating a mean rather than total) to suit our survey format, we interpret the scale score as a continuous measure with higher scores indicating more frequent relatively recent depressive symptoms. Self-rated mental health (SRMH) Beginning in 1987 when participants were age 20, we asked how healthy they felt “mentally” in the past few months on a scale from 1 (very unhealthy) to 5 (very healthy). This item was adapted from the 1983 Edmonton Area Survey (Population Research Laboratory, 2015), which predated the increased use of single-item SRMH measures in surveys in the 1990s and 2000s. As such, our response scale differs from more recently used SRMH items, which are typically rated on a scale ranging from “poor” to “excellent” (Ahmad et al., 2014). SRMH at ages 20, 25, and 32 are used in the current study to predict midlife CESD-10 scores. Control variables Sex was coded as 0 (female) or 1 (male). Participant education, based on highest level of education reported at either age 43 or 50, was dichotomized, with 1 = university degree or higher. Marital status was coded at age 43 and age 50 as a binary variable, with 1 = married or cohabiting. Self-rated physical health (SRPH) was assessed at age 43 and 50 by asking how healthy participants felt “physically” in the past few months on a scale from 1 (very unhealthy) to 5 (very healthy), parallel to our SRMH item and similar to SRPH as measured in other studies (Jones & Schreier, 2021; Levinson & Kaplan, 2014; Sawatzky et al., 2010). At age 18, the mean of four items from the 20-item CES-D assessed baseline depressive symptoms (“depressed,” “lonely,” “talked less than usual,” and “people were unfriendly”) in the past few months on a scale from 1 (never) to 5 (almost always). As shown in previous research with ETS participants, internal consistency of the four-item scale was acceptable (α = 0.68; Galambos et al., 2006). In the current sample within time, this four-item scale correlated significantly (p < 0.001) with the full CESD-10 at age 43 (r = 0.83) and age 50 (r = 0.87). Analytic plan For each outcome (CESD-10 at age 43 and at age 50) two regression models were estimated. In the first model, SRMH items at three earlier ages were included as predictors: CES-D Age 43 or Age 50 = a + B1(SRMH Age 20) + B2(SRMH Age 25) + B3(SRMH Age 32), where a = constant. In the second model, control variables were added to the set of SRMH items: CES-D Age 43 or 50 = a + B1(SRMH Age 20) + B2(SRMH Age 25) + B3(SRMH Age 32) + B4(Sex) + B5 (Education) + B6(Concurrent Marital Status, Age 43 or 50) + B7(Concurrent SRPH, Age 43 or 50) + B8(Depressive symptoms Age 18). Missing data and multiple imputation Missing data ranged from 0% to 39.64% (see Table 2), and most missing values were due to individuals not participating at a given wave rather than to item non-response. To handle missing data, and in accordance with recommended procedures, multiple imputation (MI) with auxiliary variables maximized the use of all observed data for those with CESD-10 scores at age 43 and/or 50. MI takes many copies of a dataset and imputes (or assigns scores for) the missing values in each copy using a model based on observed data. Then the substantive analyses (e.g., the regressions described above) are conducted on each of the imputed datasets, with the parameters combined to generate the final estimates. MI is generally considered superior to mean substitution or complete cases analysis as it is more likely to lead to non-biased inferences (Graham, 2009; Lee et al., 2016: Schlomer et al., 2010).Table 2 Descriptive Statistics for Study Variables Variable M SD N % Missing Data % Missing Due to Item Non Response CESD-10 Age 43 2.24 0.61 402 19.92 0.40 CESD-10 Age 50 2.33 0.63 403 19.72 0.20 SRMH Age 20 3.91 0.95 374 25.50 0.00 SRMH Age 25 3.74 0.98 303 39.64 0.80 SRMH Age 32 4.06 0.91 371 26.10 0.00 Sexa 0.50 0.50 502 00.00 0.00 Educationb 0.40 0.49 502 00.00 0.00 Marriedc Age 43 0.76 0.43 404 19.52 0.00 Marriedc Age 50 0.77 0.42 404 19.52 0.00 SRPH Age 43 3.70 1.01 389 22.51 3.00 SRPH Age 50 3.60 0.99 402 19.92 0.40 Depress Age 18 2.74 0.65 493 01.79 1.79 CESD-10 = Center for Epidemiologic Studies Depression Scale (10 item; possible range: 1 = never to 5 = almost always). SRMH = self-rated mental health (1 = very unhealthy to 5 = very healthy). SRPH = self-rated physical health (1 = very unhealthy to 5 = very healthy). Depress = depressive symptoms (possible range: 1 = never to 5 = almost always). a1 = male. b1 = university degree or higher at age 43 or 50. c1 = married or cohabiting. Our imputation model included 36 auxiliary variables selected because they were known or likely to be associated with SRMH, CESD-10 scores, the control variables, or missingness (e.g., parent education, immigration status, happiness at baseline). The inclusion of a large number of carefully chosen auxiliary variables is an optimal strategy for reducing bias and increasing power (Graham, 2009; Schlomer et al., 2010). Thirty imputed datasets were generated with 100 iterations, and regression models were run for all 30, with the estimates pooled. Significance was set at 0.05 (two-tailed). SPSS 28.0.1 was used for all analyses. MI assumes that data are missing at random. The lack of significant differences between continuers and attriters on key study variables (baseline depressive symptoms, SRMH and SRPH scores), along with the inclusion of possible predictors of missingness in the imputation and analytic (e.g., sex) models, support the missing at random criterion (Graham, 2009; Lee et al., 2016). Results Table 2 presents descriptive statistics for each measure. On average, and consistent with a non-clinical sample, participants did not report high levels of depressive symptoms on the CESD-10 at ages 43 or 50, although depressive symptoms at age 18 were higher, consistent with earlier research on this sample (Johnson et al., 2021). SRMH and SRPH mean scores indicated that participants felt quite healthy mentally and physically at all ages, but there was substantial variability across participants. Table 3 presents correlations among the measures. SRMH at ages 20, 25, and 32 showed significant small to moderate intercorrelations. Higher SRMH at ages 20, 25, and 32 were weakly to moderately associated with lower depressive symptoms in midlife (i.e., CESD-10 at ages 43 and 50). Among control variables, sex was related to CESD-10 only at age 50, with men less likely to report depressive symptoms. A university education was related to a lower CESD-10 score only at age 43. Marriage or cohabitation in midlife was associated with higher SRMH at some earlier points and with lower CESD-10 scores in midlife, while better SRPH was consistently associated with better SRMH and lower CESD-10 scores. Finally, higher depressive symptoms at age 18 was related to lower SRMH at all ages and to higher midlife CESD-10 scores.Table 3 Bivariate Correlations Among SRMH, Control Variables, and CESD-10 by Age SRMH CESD-10 Age 20 25 32 43 50 Variable CESD-10 Age 43 .53* SRMH Age 20 .22* .18* -.19* -.17* SRMH Age 25 .32* -.34* -.31* SRMH Age 32 -.36* -.36* Sexa .09 .08 .08 -.07 -.13* Educationb .09 -.03 .01 -.15* -.08 Marriedc Age 43 .01 .05 .18* -.23* -.13* Marriedc Age 50 .12 .16* .13* -.28* -.22* SRPH Age 43 .14* .21* .27* -.49* -.28* SRPH Age 50 .15* .16* .29* -.33* -.34* Depress Age 18 -.29* -.24* -.15* .26* .32* N = 502. CESD-10 = Center for Epidemiologic Studies Depression Scale (10 item). SRMH = self-rated mental health. SRPH = self-rated physical health. Depress = depressive symptoms. a1 = male. b1 = university degree at age 43 or 50. c1 = married or cohabiting. p < .05. Table 4 presents the results of the regression analyses. Considering Model 1, which explained 19% of the variance in CESD-10 at age 43, higher SRMH at ages 25 and 32 predicted lower CESD-10 scores. Model 2 explained 39% of the variance in age 43 CESD-10 symptoms, with age 25 and 32 SRMH again attaining significance. A university education, being married or cohabiting, reporting better physical health, and showing lower depressive symptoms at age 18 predicted lower CESD-10 scores at age 43.Table 4 Regressions Predicting Age 43 and Age 50 CESD-10 from SRMH at Earlier Ages and Control Variables Variable Age 43 CESD-10 Age 50 CESD-10 B 95% CI for B SE β t B 95% CI for B SE β t LL UL B LL UL B Model 1 Constant 3.72 3.35 4.08 .19 19.96* 3.78* 3.43 4.13 .18 20.96* SRMH Age 20 -.06 -.13 .01 .04 -.09 -1.60 -.05 -.11 .02 .03 -.07 -1.43 SRMH Age 25 -.14* -.22 -.06 .04 -.23* -3.54* -.13* -.20 -.05 .04 -.20* -3.19* SRMH Age 32 -.18* -.25 -.10 .04 -.26 -4.80* -.20* -.27 -.12 .04 -.28* -5.09* Model R2a .19* .18* Model 2 Constant 3.82* 3.35 4.30 .24 15.82* 3.30* 2.78 3.83 .27 12.39* SRMH Age 20 -.01 -.08 .05 .03 -.02 -.47 .01 -.05 .07 .03 .01 .27 SRMH Age 25 -.10* -.18 -.03 .04 -.17* -2.91* -.09* -.16 -.01 .04 -.14* -2.28* SRMH Age 32 -.10* -.17 -.03 .04 -.15* -2.82* -.14* -.22 -.07 .04 -.21* -3.68* Sexb -.01 -.11 .08 .05 -.01 -.30 -.03 -.14 .08 .06 -.02 -.51 Educationc -.14* -.24 -.04 .05 -.11* -2.66* -.05 -.16 .06 .06 -.04 -.91 Marriedd Age 43 -.22* -.34 -.10 .06 -.15* -3.52* Marriedd Age 50 -.20* -.32 -.07 .06 -.13* -3.06* SRPH Age 43 -.23* -.28 -.17 .03 -.38* -8.06* SRPH Age 50 -.13* -.20 -.07 .03 -.21* -4.13* Depress Age 18 .12* .04 .21 .04 .13* 2.92* .21* .11 .30 .05 .21* 4.38* Model R2a .39* .29* N = 502. CESD-10 = Center for Epidemiologic Studies Depression Scale (10 item). SRMH = self-rated mental health. SRPH = self-rated physical health. Depress = depressive symptoms. LL = lower limit; UL = upper limit. aMean R2 across 30 imputed datasets. b1 = male. c1 = university degree at age 43 or 50. d1 = married or cohabiting. *p < .05. With age 50 CESD-10 scores as the criterion, higher SRMH at ages 25 and 32 predicted lower depressive symptoms at age 50 in both models. Model 1 explained 18% of the variance and Model 2 explained 29% of the variance in age 50 CESD-10. Being married or cohabiting, reporting better SRPH, and lower depressive symptoms scores at age 18 predicted lower CESD-10 scores at age 50. Discussion The results of this study showed that lower SRMH, assessed with a single item, forecast depressive symptoms up to 25 years into the future. The predictive strength of single-item SRMH assessed in the middle of the transition to adulthood (age 25) and in early adulthood (age 32) is remarkable, particularly considering that the import of SRMH held when controlling for sex, depressive symptoms at age 18, educational level, and concurrently measured marital status and SRPH in midlife. The results are interpretable from a life course perspective, which highlights the likelihood of connections between early and later mental health experiences. Self-assessment that one’s mental health may not be in good shape constitutes a risk particularly if it becomes a continuing part of one’s narrative in the 20s and early 30s. We concur with McAlpine et al. (2018) in their assessment that SRMH may be a powerful construct. Whereas SRMH explained a significant 18 to 19 percent of the variance in depressive symptoms in midlife, the control variables explained an additional 10 to 20 percent. Having a university education, being married or cohabiting, reporting better physical health, and reporting fewer depressive symptoms in high school appeared to protect against midlife depressive symptoms. Furthermore, bivariate correlations showed that better SRMH at ages 25 and 32 was linked with having a marital/romantic partner at age 50, and SRMH at ages 20, 25, and 32 were positively correlated with SRPH at ages 43 and 50 (see Table 3). Consistent with a life course perspective, SRMH in the transition to adulthood and early adulthood may have cascading effects on other circumstances in young people’s lives, which ultimately carry through to midlife – and likely beyond. The fact that SRMH early in the transition to adulthood (i.e., age 20) did not predict midlife depressive symptoms is of interest. The early 20s is a period of instability in living, financial, romantic, educational, and employment situations (Krahn et al., 2015; Masten et al., 2010), and possibly, of improving mental health on average as young people meet developmental challenges and begin to settle into their roles (Galambos et al., 2006; Howard et al., 2010). It could be that SRMH is too unstable when lives are in flux to be a reliable predictor of mental health problems decades later. Indeed, correlations (Table 3) showed there was somewhat stronger stability in SRMH from age 25 to 32 (r = 0.32) than from age 20 to age 25 (r = 0.22). And SRMH at age 25 and age 32 was more strongly correlated with midlife depressive symptoms (rs between -0.31 and -0.36) than was SRMH at age 20 (rs of -0.19 and -0.17 for age 43 and age 50 depressive symptoms, respectively). This does not mean that SRMH in the early 20s is irrelevant. It may be an important harbinger of difficulties in the shorter term for individuals early in the transition to adulthood (see e.g., Jones & Schreier, 2021), which could spill over into well-being in their later 20s. Questions about the meaning of SRMH, and what it measures, remain. The factors that shape individuals’ constructions and interpretations of their overall mental health may be many, and they may also differ from person to person. Given that specific disorders like major depression or generalized anxiety disorder can wax and wane over time (Caspi et al., 2021), SRMH might provide a broad but important glimpse into psychological health, useful for tracking the well-being of individuals through life’s stages as they encounter hills and valleys on the way. As many people with mental health issues do not seek professional help or wait years until they do (Wang et al., 2005), SRMH could be an early marker of developing problems, constituting a good screening tool for use by clinicians (Casu & Gremigni, 2019), in addition to its utility for monitoring population health. Recently, Allen et al. (2022) called for more research on the reliability and validity of single-item measures, arguing that such items have a poor and probably undeserved reputation (see also Fisher et al., 2016). Our longitudinal results suggest there is merit to using single-item SRMH in research and possibly clinical applications. It could be that, like the better known and well-studied self-rated global health item that is robust enough to predict eventual morbidity and mortality (Benyamini et al., 2011; Idler & Benyamini, 1997), SRMH will earn its place among short, easy-to-administer, and meaningful tools. Limitations Although this prospective study has considerable strength, including multiple waves at which SRMH was measured and its representativeness of youth in western cities in Canada at baseline, there are limitations. First, because SRMH was first measured in the ETS prior to its more common usage in survey research, our response scale ranged from very unhealthy to very healthy rather than from poor to excellent. It is unclear whether this difference is consequential, but the substantive results (i.e., its positive associations with the CESD-10) suggest our item may perform similarly to others in more recent studies. Second, to better fit into the format of our survey, the timeframe and response scale of the CESD-10 was modified. We believe our continuous measure provides important insight into the recent frequency of depressive symptoms in a non-clinical sample but we were unable to conduct analyses to determine whether earlier SRMH predicted a probable diagnosis of major depression in midlife. Third, as is typical with a three-decade longitudinal study, attrition resulted in missing data. To ensure the validity of our results, we followed current best practices for managing missing data, one of which is to conduct MI with many auxiliary variables. This strategy uses all observed data and produces less biased estimates than complete cases analysis or mean substitution (Graham, 2009; Lee et al., 2016). Finally, because the sample is largely White the results may not be generalizable to other racialized groups. Research shows, for example, that SRMH may be more strongly related to mental health conditions and health service utilization in White than in Black and Hispanic samples (McAlpine et al., 2018; Zuvekas & Fleishman, 2008). Conclusion In the last 25 or 30 years, single-item SRMH has appeared increasingly in population health surveys as a measure of well-being, psychological distress, and mental health problems and disorders. Although it is a broad measure not interchangeable with psychiatric diagnoses, the literature shows that it does appear to tie into symptoms that may reflect serious mental health conditions as well as possibly emerging mental health problems. Most research on SRMH has been cross-sectional rather than longitudinal, and the few prospective studies that are published have not followed participants for more than one or two years. By taking a life course perspective on SRMH and examining whether SRMH in the transition to adulthood and early adulthood foretells depressive symptoms in midlife, the current study provides a unique view into the value of SRMH as an indicator of developing mental health problems. The finding that lower SRMH predicted higher depressive symptoms up to 25 years later, after controlling for life events such as marriage and achieving a university education, is remarkable and represents a unique contribution to the literature. SRMH is an indicator deserving of further empirical attention. Acknowledgements Data were collected by the Population Research Laboratory, University of Alberta. Funding This research was funded by grants from the Social Sciences and Humanities Research Council of Canada, Alberta Advanced Education, and the University of Alberta. Declarations Competing Interests The authors have no relevant financial or non-financial interests to disclose. Ethics Approval Ethics approval was obtained for all waves of this study from the University of Alberta (most recent approval was granted June 2017; Pro00074272; “Transitions to midlife: 32 year follow-up of the class of 1985”). Informed consent was obtained for all individuals participating at each wave. Data availability The data used in the analyses for the current study are available from the corresponding author on reasonable request. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Ahmad F Jhajj AK Stewart DE Burghardt M Bierman AS Single item measures of self-rated mental health: A scoping review BMC Health Services Research 2014 14 398 10.1186/1472-6963-14-398 25231576 Allen MS Iliescu D Greiff S Single item measures in psychological science: A call to action European Journal of Psychological Assessment 2022 38 1 1 5 10.1027/1015-5759/a000699 Almeida, D. M, Rush, J., Mogle, J. Piazza, J. R., Cerino, E., & Charles, S. T. (2022). Longitudinal change in daily stress across 20 years of adulthood: Results from the National Study of Daily Experiences. Developmental Psychology. Advance online publication. 10.1037/dev0001469 American Psychiatric Association Diagnostic and statistical manual of mental disorders 2013 5 American Psychiatric Association Andresen EM Malmgren JA Carter WB Patrick DL Screening for depression in well older adults: Evaluation of a short form of the CES-D (Center for Epidemiologic Studies Depression Scale) American Journal of Preventive Medicine 1994 10 2 77 84 10.1016/s0749-3797(18)30622-6 8037935 Angst J Dobler-Mikola, The Zurich Study: A prospective epidemiological study of depressive, neurotic and psychosomatic syndromes. IV: Recurrent and non-recurrent brief depression European Archives of Psychiatry and Neurological Sciences 1985 234 6 408 416 10.1007/BF00386060 3875489 Assari, S., & Lankarani, M. M. (2017). Demographic and socioeconomic determinants of physical and mental self-rated health across 10 ethnic groups in the United States. International Journal of Epidemiologic Research, 3(12), 185–193. 10.15171/ijer.2017.02 Baron EC Davies T Lund C Validation of the 10-item Centre for Epidemiological Studies Depression Scale (CES-D-10) in Zulu, Xhosa and Afrikaans populations in South Africa BMC Psychiatry 2017 17 6 10.1186/s12888-016-1178-x 28068955 Benyamini Y Blumstein T Murad H Lerner-Geva L Changes over time from baseline poor self-rated health: For whom does poor self-rated health not predict mortality? Psychology & Health 2011 26 11 1446 1462 10.1080/08870446.2011.559231 22011289 Capaldi, C. A., Liu, L., & Dopko, R. L. (2021). Positive mental health and perceived change in mental health among adults in Canada during the second wave of the COVID-19 pandemic. Health Promotion and Chronic Disease Prevention in Canada, 41(11). 10.24095/hpcdp.41.11.05 Caspi A Houts RM Ambler A Danese A Elliott ML Hariri A Harrington H Hogan S Poulton R Ramrakha S Rasmussen LJH Reuben A Richmond-Rakerd L Sugden K Wertz J Williams BS Moffitt TE Longitudinal assessment of mental health disorders and comorbidities across 4 decades among participants in the Dunedin Birth Cohort Study JAMA Network Open 2020 3 4 e203221 10.1001/jamanetworkopen.2020.3221 32315069 Casu G Gremigni P Is a single-item measure of self-rated mental health useful from a clinimetric perspective? Psychotherapy and Psychosomatics 2019 88 3 177 178 10.1159/000497373 30909284 Chiu M Rahman F Kurdyak P Cairney J Jembere N Vigod S Self-rated health and mental health of lone fathers compared with lone mothers and partnered fathers: A population-based cross-sectional study Journal of Epidemiology and Community Health 2017 71 5 417 423 10.1136/jech-2016-208005 27923873 Cohen JS Patten S Well-being in residency training: A survey examining resident physician satisfaction both within and outside of residency training and mental health in Alberta BMC Medical Education 2005 5 21 10.1186/1472-6920-5-21 15972100 Crosnoe R Elder GH Jr From childhood to the later years: Pathways of human development Research on Aging 2004 26 6 623 654 10.1177/0164027504268491 Farhat SMM Berbiche D Vasiliadis H Family violence associated with self-rated health, satisfaction with life, and health service use in primary care older adults Journal of Interpersonal Violence 2022 37 5–6 2487 2509 10.1177/0886260520938513 32646316 Fisher GG Mathews RA Gibbons AM Developing and investigating the use of single-item measures in organizational research Journal of Occupational Health Psychology 2016 21 1 3 23 10.1037/a0039139 25894198 Fleishman JA Zuvekas SH Global self-rated mental health: Associations with other mental health measures and with role function Medical Care 2007 45 7 602 609 10.1097/MLR.0b013e31803bb4b0 17571008 Galambos NL Barker ET Krahn HJ Depression, self-esteem, and anger in emerging adulthood: Seven-year trajectories Developmental Psychology 2006 42 2 350 365 10.1037/0012-1649.42.2.350 16569173 Galambos NL Krahn HJ Johnson MD Lachman ME The U shape of happiness across the life course: Expanding the discussion Perspectives on Psychological Science 2020 15 4 898 912 10.1177/1745691620902428 32375015 George, L. K. (1999). Life-course perspectives on mental health. In C. S. Aneshensel & J. C. Phelan (Eds.), Handbook of the sociology of mental health (pp. 565–583). Kluwer Academic/Plenum Publishers. 10.1007/0-387-36223-1_27 Graham JW Missing data analysis: Making it work in the real world Annual Review of Psychology 2009 60 549 576 10.1146/annurev.psych.58.110405.085530 18652544 Gustavson K Knudsen AK Nesvåg R Knudsen GP Vollset SE Reichborn-Kjennerud T Prevalence and stability of mental disorders among young adults: Findings from a longitudinal study BMC Psychiatry 2018 18 1 10.1186/s12888-018-1647-5 29304757 Haug MR Ford AB Stange KC Noelker LS Gaines AD Effect of giving care on caregivers’ health Research on Aging 1999 21 4 515 538 10.1177/0164027599214001 Hoff R Bruce M Kasl S Jacobs S Subjective ratings of emotional health as a risk factor for major depression in a community sample British Journal of Psychiatry 1997 170 2 167 172 10.1192/bjp.170.2.167 Howard AL Galambos NL Krahn HJ Paths to success in young adulthood from mental health and life transitions in emerging adulthood International Journal of Behavioral Development 2010 34 6 538 546 10.1177/0165025410365803 Idler EL Benyamini Y Self-rated health and mortality: A review of twenty-seven community studies Journal of Health and Social Behavior 1997 38 1 21 37 10.2307/2955359 9097506 Infurna FJ Gerstorf D Lachman ME Midlife in the 2020s: Opportunities and challenges American Psychologist 2020 75 4 470 485 10.1037/amp0000591 32378943 Jang Y Park NS Kim G Kwag KH Roh S Chiriboga DA The association between self-rated mental health and symptoms of depression in Korean American older adults Aging & Mental Health 2012 16 4 481 485 10.1080/13607863.2011.628981 22126263 Johnson MD Galambos NL Krahn HJ Family context, life transitions, and subjective well-being from age 18 to 50 years Developmental Psychology 2021 57 11 1968 1980 10.1037/dev0001243 34914457 Johnson MD Krahn HJ Galambos NL Better late than early: Marital timing and subjective well-being in midlife Journal of Family Psychology 2017 31 5 635 641 10.1037/fam0000297 28114770 Jones, E. J., & Schreier, H. M. C. (2021). Self-rated mental and physical health are prospectively associated with psychosocial and academic adjustment to college. Journal of American College Health. Advance online publication. 10.1080/07448481.2021.1904956. Kessler RC Andrews G Colpe LJ Hiripi E Mroczek DK Normand S-LT Walters EE Zaslavsky AM Short screening scales to monitor population prevalences and trends in non-specific psychological distress Psychological Medicine 2002 32 6 959 976 10.1017/s0033291702006074 12214795 Kessler RC Birnbaum H Bromet E Hwang I Sampson N Shahly V Age differences in major depression: Results from the National Comorbidity Survey Replication (NCS-R) Psychological Medicine 2010 40 2 225 237 10.1017/S0033291709990213 19531277 Kessler RC Ustün TB The World Mental Health (WMH) Survey Initiative version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI) International Journal of Methods in Psychiatric Research 2004 13 2 93 121 10.1002/mpr.168 15297906 Krahn HJ Howard AL Galambos NL Exploring or floundering? The meaning of employment and educational fluctuations in emerging adulthood Youth and Society 2015 47 2 245 266 10.1177/0044118X12459061 Kroenke K Spitzer RL Williams JB The PHQ-9: Validity of a brief depression severity measure Journal of General Internal Medicine 2001 16 9 606 613 10.1046/j.1525-1497.2001.016009606.x 11556941 Kroenke K Spitzer RL Williams JBW The Patient Health Questionnaire-2: Validity of a two-item depression screener Medical Care 2003 41 11 1284 1292 10.1097/01.MLR.0000093487.78664.3C 14583691 Lachman ME Teshale S Agrigoroaei S Midlife as a pivotal period in the life course: Balancing growth and decline at the crossroads of youth and old age International Journal of Behavioral Development 2015 39 1 20 31 10.1177/0165025414533223 25580043 Lee KJ Roberts G Doyle LW Anderson PJ Carlin JB Multiple imputation for missing data in a detailed case study involving imputation of missing outcome data International Journal of Social Research Methodology 2016 19 5 575 591 10.1080/13645579.2015.1126486 Levinson D Kaplan G What does self rated mental health represent? Journal of Public Health Research 2014 3 287 122 127 10.4081/jphr.2014.287 Lombardo P Jones W Wang L Shen X Goldner EM The fundamental association between mental health and life satisfaction: Results from successive waves of a Canadian national survey BMC Public Health 2018 18 342 10.1186/s12889-018-5235-x 29530010 Magwene E Quiñones AR Marshall GL Makaroun LK Thielke S Older adults rate their mental health better than their general health Journal of Public Health Research 2017 6 2 967 10.4081/jphr.2017.967 29071258 Masten AS Desjardins CD McCormick CM Kuo SI-C Long JD The significance of childhood competence and problems for adult success in work: A developmental cascade analysis Development and Psychopathology 2010 22 3 679 694 10.1017/S0954579410000362 20576187 Masten AS Tellegen A Resilience in developmental psychopathology: Contributions of the Project Competence Longitudinal Study Development and Psychopathology 2012 24 2 345 361 10.1017/S095457941200003X 22559118 Mawani FN Gilmour H Validation of self-rated mental health Health Reports 2010 21 3 61 75 20973435 McAlpine DD McCreedy E Alang S The meaning and predictive value of self-rated mental health among persons with a mental health problem Journal of Health and Social Behavior 2018 59 2 200 214 10.1177/0022146518755485 29406825 McVey W Jr Kalbach W Canadian population 1995 Nelson Canada Mechanic D., & McAlpine D.D. (2011). Mental health and aging: A life-course perspective. In R. Settersten & J. Angel (Eds.), Handbook of sociology of aging (pp. 477–493). Springer. 10.1007/978-1-4419-7374-0_30 Merikangas KR Zhang H Avenevoli S Acharyya S Neuenschwander M Angst J Longitudinal trajectories of depression and anxiety in a prospective community study Archives of General Psychiatry 2003 60 10 993 1000 10.1001/archpsyc.60.9.993 14557144 Mossakowski KN Unfulfilled expectations and symptoms of depression among young adults Social Science & Medicine 2011 73 5 729 7363 10.1016/j.socscimed.2011.06.021 21798639 Nguyen MT Chan WY Keeler C The association between self-rated mental health status and total health care expenditure: A cross-sectional analysis of a nationally representative sample Medicine 2015 94 35 e1410 10.1097/MD.0000000000001410 26334899 Palis H Marchand K Oviedo-Joekes E The relationship between sense of community belonging and self-rated mental health among Canadians with mental or substance use disorders Journal of Mental Health 2018 29 2 168 175 10.1080/09638237.2018.1437602 29431529 Population Research Laboratory. (2015). Edmonton Area Survey, 1983 [Code book]. University of Alberta Libraries Dataverse, V3. 10.7939/DVN/10582 Radloff LS The CES-D scale: A self-report depression scale for research in the general population Applied Psychological Measurement 1977 1 3 385 401 10.1177/014662167700100306 Ramiz L Contrand B Castro MYR Dupuy M Lu L Sztal-Kutas C Lagarde E A longitudinal study of mental health before and during COVID-19 lockdown in the French population Globalization and Health 2021 17 29 10.1186/s12992-021-00682-8 33752717 Roberts RE Lewinsohn PM Seeley JR A brief measure of loneliness suitable for use with adolescents Psychological Reports 1993 72 3_suppl 1379 1391 10.2466/pr0.1993.72.3c.1379 8337350 Robins LN Helzer JE Croughan JL Ratcliff KS National Institute of Mental Health Diagnostic Interview Schedule: Its history, characteristics, and validity Archives of General Psychiatry 1981 38 4 381 389 10.1001/archpsyc.1981.01780290015001 6260053 Sawatzky R Ratner PA Johnson JL Kopec JA Zumbo BD Self-reported physical and mental health status and quality of life in adolescents: A latent variable mediation model Health and Quality of Life Outcomes 2010 8 17 10.1186/1477-7525-8-17 20128913 Schlomer GL Bauman S Card NA Best practices for missing data management in counseling psychology Journal of Counseling Psychology 2010 57 1 1 10 10.1037/a0018082 21133556 Schulenberg JE Sameroff AJ Cicchetti D The transition to adulthood as a critical juncture in the course of psychopathology and mental health Development and Psychopathology 2004 16 4 799 806 10.1017/S0954579404040015 15704815 Shanahan MJ Pathways to adulthood in changing societies: Variability and mechanisms in life course perspective Annual Review of Sociology 2000 26 667 692 10.1146/annurev.soc.26.1.667 Sheikh JI Yesavage JA Geriatric Depression Scale (GDS): Recent evidence and development of a shorter version Clinical Gerontologist: The Journal of Aging and Mental Health 1986 5 1–2 165 173 10.1300/J018v05n01_09 Spitzer RL Kroenke K Williams JBW Löwe B A brief measure for assessing generalized anxiety disorder: The GAD-7 Archives of Internal Medicine 2006 166 10 1092 1097 10.1001/archinte.166.10.1092 16717171 Su Y D'Arcy C Caron J Meng X Increased income over time predicts better self-perceived mental health only at a population level but not for individual changes: An analysis of a longitudinal cohort using cross-lagged models Journal of Affective Disorders 2021 292 487 495 10.1016/j.jad.2021.05.118 34146900 Tanner JL Reinherz HZ Beardslee WR Fitzmaurice GM Leis JA Berger SR Change in prevalence of psychiatric disorders from ages 21 to 30 in a community sample Journal of Nervous and Mental Disease 2007 195 4 298 306 10.1097/01.nmd.0000261952.13887.6e 17435479 Wang PS Berglund P Olfson M Pincus HA Wells KB Kessler RC Failure and delay in initial treatment contact after first onset of mental disorders in the National Comorbidity Survey Replication Archives of General Psychiatry 2005 62 6 603 613 10.1001/archpsyc.62.6.603 15939838 Wittchen HU Pfister H DIA-X interviews (M-CIDI): Manual fur screening - Verfahren und interview 1997 Swets & Zeitlinger Zuvekas SH Fleishman JA Self-rated mental health and racial/ethnic disparities in mental health service use Medical Care 2008 46 9 915 923 10.1097/MLR.0b013e31817919e5 18725845	36504487	PMC9718454	NO-CC CODE	2022-12-06 23:23:39	no	Curr Psychol. 2022 Dec 2;:1-12	utf-8	Curr Psychol	2,022	10.1007/s12144-022-04081-z	oa_other
==== Front Public Organiz Rev Public Organization Review 1566-7170 1573-7098 Springer US New York 692 10.1007/s11115-022-00692-z Article Vaccine Strategy During the Covid-19 Pandemic: A Community Engaged Research Supporting a Policy Oriented Towards Nonprofit Organizations and Volunteers http://orcid.org/0000-0001-6256-6905 Plaisance Guillaume [email protected] 12 1 grid.412041.2 0000 0001 2106 639X Research Institute in Management Science, Bordeaux University, IAE Bordeaux - IRGO, 35 avenue Abadie, Bordeaux, France 2 IAE Bordeaux – IRGO, 35 Avenue Abadie, 33000 Bordeaux, France 2 12 2022 117 25 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. In the face of the Covid-19 crisis, vaccination was the medical tool and nonprofit organizations have tried to reduce its social impact. Nevertheless, they are mostly constituted of elderly volunteers, who chose to suspend their commitment. In France, within community-engaged research, a proposition from practitioners was to adapt the health strategy by including volunteers in the vaccine strategy. A survey dedicated to these topics and testing the proposition in January-February 2021 obtained 1,862 responses from volunteers. It confirms that the pandemic has disrupted volunteering and that the vaccination of volunteers would allow NPOs to reduce the lack of human resources. Keywords Vaccination Nonprofits Volunteer France Health Policy ==== Body pmcIntroduction The rapid emergence of effective vaccines has been a hope for coping with the Covid-19 pandemic (Belle & Cantarelli, 2021). However, vaccines are useful only in cases of broad public acceptance and high vaccination coverage. In other words, public authorities faced the challenge of converting the vaccine acceptability into actual injections. Scholars have studied the determinants explaining the willingness to be vaccinated against Covid-19 (Lazarus et al., 2021; Yin et al., 2022). However, to date, no survey analyzed the role of membership or volunteering in non-profit organizations (NPOs). NPOs are only mentioned as key influencers (e.g. Tsheten et al., 2022). These organizations indeed fulfil great responsibilities in building public trust. For instance, Belle and Cantarelli (2021) have shown that social nudges are particularly useful in promoting vaccination, especially in the case of Covid-19. Social norms facilitate vaccination, and not only within an organization. NPOs, as trusted organizations, would also have an advocacy role to play. Moreover, some NPOs do much more than playing a social advocacy role : many are also on the front line in addressing the health, social and economic crises caused by the pandemic. The NPOs’ response to the crisis heavily relies on volunteers. This is why they are studied by scholars, for instance within the French Red Cross (Heyerdahl et al., 2021). In response to the health and social crises, Recherches & Solidarités, a network of French experts and academics, has set up an experiment to facilitate the vaccination of NPOs’ volunteers. The purpose also was to show the importance of their priority vaccination. In February 2021, in France, only healthcare professionals, employees working for the elderly or handicapped people, firefighters, people with co-morbidities and those over 75 could be vaccinated. In April 2021, the minimum age has been reduced to 55 (Ministère des Solidarités et de la Santé, 2021). The opening to all took place in May 2021. By early August 2021, 63.2% of the total population had received at least one dose of vaccine. At the beginning of 2022, when the fifth wave hit France because of the Omicron variant, 79.6% of the population had received a first dose and 77.7% had a complete vaccination schedule (three doses of vaccines or Covid-19 infections). However, France remains among the most suspicious countries, alongside Russia and Poland. Hardly more than one citizen in two was likely to be vaccinated (Lazarus et al., 2021). In this, French authorities had to put in place strong incentives such as a health/vaccine pass, a document that attests that a person meets the criteria for a full vaccination schedule and which was required for access to leisure and food services. Recherches & Solidarités’s proposal was based on the identification of an additional problem. The lack of priority given to volunteers has indeed had an effect on the life of NPOs. By the end of April 2021, 61% of NPOs had lost contact with a majority of their volunteers and in 43% of NPOs, volunteers stopped their activity for fear of the virus (Le Mouvement Associatif et al., 2021). The approach proposed here can be described as a community-engaged research and the objective of this article is to verify the relevance of volunteers as a special vaccination group. Three research questions are thus associated to these objectives: Research question 1 (RQ1): To what extent has the pandemic weakened volunteering? Research question 2 (RQ2): How can volunteer status be linked to vaccination willingness? Research question 3 (RQ3): To what extent has the community-engaged research conducted by Recherches & Solidarités served the NPOs? The article therefore follows an original structure. First, a review of the literature highlights the importance of the issues raised by Recherches & Solidarités. Then, the details of the community-engaged research approach and the survey method are presented. Finally, the results are used to discuss the three research questions. Background in Literature Recherches & Solidarités wanted to bring their idea to the public authorities. However, going beyond the context of a single experiment seems appropriate. The literature review below offers some initial insights into this subject. NPOs and Covid-19: Between Threatening Their Viability and Strengthening Their Social role NPOs include organizations of different nature and from various sectors. Within NPOs, grassroots volunteer organizations (GVOs) occupy a major place. They are characterized by the importance of their members and volunteers. Their activity is essentially based on people who give their time and money in order to act in favor of the beneficiaries. In this, the dependence on volunteers has become a common feature (Nesbit et al., 2018). The survival of NPOs and GVOs in particular is therefore at risk when volunteers retire. In a period of crisis, NPOs have several roles to play. First of all, their initial mission has a strong social impact. For instance, they can maintain and strengthen their links within communities (Misener et al., 2020). In times of social distancing and restrictive measures, these connections are crucial, within the organization between people but also externally with stakeholders. Even a temporary suspension of volunteering is therefore an urgent problem. In addition, another mission of public utility is assigned to NPOs. They are trustworthy organizations, compared to a context of mistrust towards political powers and, sometimes, pharmaceutical laboratories. In this, NPOs could promote vaccination within their community by becoming soft prescribers. They would complement the recommendations of health authorities and healthcare professionals (French et al., 2020). The Vaccine Strategy Against Covid-19: Prioritization and Population Acceptance In the absence of an obligation, optimal vaccination coverage can only be achieved if a majority wishes to be vaccinated. Studies show that the vaccine acceptability is quite good but depends on traditional factors such as age, respondent’s health and the role of caregivers (e.g. Reiter et al., 2020). A vaccination strategy must therefore convince the population and manage the vaccines shortage. While Western countries have greatly benefited from the first produced doses, other countries are still receiving reduced quantities. The prioritization issue is therefore crucial. Health authorities had to decide on ethical paradoxes, particularly between competing values (O’Flynn, 2021) such as equal access to vaccines and the need to prioritize people. People with co-morbidities (Ribas et al., 2021), the elderly (Jeyanathan et al., 2020) and health professionals (Chirico et al., 2020) are favored by scholars. Beyond this apparent consensus, scholars are questioning the relevance of a prioritization solely based on health concerns. The place of the youngest is discussed (Giubilini et al., 2020), because they are gathered in their schools for example. Essential workers with high added value for the society also include very different categories (e.g. food industries, public transport, security). They are at various vaccination stages according to the studies reviewed (Hassan-Smith et al., 2020; Russell & Greenwood, 2020). The weak consideration of essential workers in favor of age criteria is criticized by scholars (Persad et al., 2020). To extend this criticism, NPO volunteers can also be seen as essential workers. However, these people are not identified in the target groups to be vaccinated, including in researches proposing a full vaccination strategy (J. Yang et al., 2021). To show the value of this group, the present study should therefore determine if the pandemic has weakened volunteering and if being a volunteer can be related to vaccination. This was the proposal from Recherches & Solidarités. A Specific Approach: A community-engaged Research Leading to a Broader Survey Community-engaged research is a “research in any field that partners university scholarly resources with those in the public and private sectors to enrich knowledge, address and help solve critical societal issues, and contribute to the public good” (Stanton, 2008, p. 20). This research approach places the community at the heart of the knowledge production. The practitioners and professionals develop the main aspects of the research. Their points of view and expertise are the research base. The research target is therefore the community (Touboulic et al., 2020). In this, the results are provided to organizations and society and then contribute to the advancement of scientific knowledge (Cunliffe & Scaratti, 2017). Wallerstein et al., (2017) provide a four-stage model for engaged research, used here to describe the context of this study. First, the partnership process is the pre-existing collaboration between nonprofit professionals and researchers within Recherches & Solidarités. It is a French NPO created in 2008. It is constituted by a network of professionals, experts and scholars, who are all specialists in solidarity and NPOs. As a partner of public authorities and NPOs federations, the organization studies the French non-profit sector. Second, the context is the pandemic: volunteers within NPOs have been withdrawing and suspending their activities to protect themselves (Bazin & Malet, 2021). This is significant because, despite their annual budget of 113 billion euros, French NPOs are above all GVOs. They rely on their volunteers in order to exist and work: only 12% of French GVOs have employees and frequently fewer than 2 (Tchernonog & Prouteau, 2019). Third, the intervention and research processes are based on a dual approach. The first part was an original proposal from Recherches & Solidarités. They wanted to show that vaccinating volunteers was logistically possible without depriving vulnerable people of doses. The idea was to adjust the French vaccination strategy (Bazin & Malet, 2021). Recherches & Solidarités proposed to identify volunteers who were ready to be quickly vaccinated. They were called by the vaccination center in case of appointment cancellations or absences. This proposal did not hinder the policy oriented towards the over 75-year-olds (at the time of the proposal in January 2021) and did not ask for more doses of vaccine. Recherches & Solidarités in January, 2021, noted that there were fewer volunteers and the vulnerable groups’ needs were increasing (Sebbag et al., 2021). At the same time, the vaccination campaign only covered people over 75 and those over 50 with health problems. The proposed initiative focused on volunteers between 50 and 74 and volunteering in health, charity and social NPOs. The proposal is based on three principles (Bazin & Malet, 2021): 1) “Avoiding imperative measures to respect vaccination implementation actors and the difficulties of their mission”. 2) “Not interfering with the process underway with people over 75 and operating at a constant number of doses”. 3) Taking advantage of the absence of people expected to attend an appointment: “replace them with volunteers identified and proposed in advance by NPOs managers and able to move quickly”. The local experiment was carried in the town of Châteaudun (Eure-et-Loir in the centre of the country) and its surroundings with the help of the Territorial Professional Health Community (Communauté Professionnelle Territoriale de Santé, CPTS Sud 28) as a vaccination center for Châteaudun. Eleven NPOs were approached, including Les Restaurants du Coeur, the Secours Populaire Français, the Society of Saint-Vincent-de-Paul, the Petits Frères des Pauvres, the French Red Cross and the Secours Catholique. These NPOs have proposed lists of volunteers to be vaccinated (especially those who can be vaccinated with a medical certificate) and the vaccination center called volunteers in cases of absences. In one month, 66 volunteers between 50 and 74 received a first injection following absences or cancellations. They represent 3% of the 2,500 people vaccinated on February 22, 2021 in Châteaudun and its surroundings. Volunteers and NPOs have expressed their relief. The first ones no longer had difficulties in getting an appointment to be vaccinated and the second ones were glad to have some volunteers back (Sebbag et al., 2021). The second part of this community-engaged research is the study underlying the present article. In order to show the generalizability of this experiment, a scientific approach was adopted. A survey was launched by Recherches & Solidarités on January 14, 2021. It was disseminated through Recherches & Solidarités’s social networks and newsletter and some French NPOs federations and umbrella organizations in the sector also relayed the survey. As a result, a majority of large French NPOs received the questionnaire and forwarded it to their volunteers. However, it is not possible to know exactly how many NPOs were reached or how many volunteers received the survey. By February 10, 2021, 1,862 volunteers had responded. Table 1 then presents the four questions asked and the sorting variables. Organizations from the health, social, sports and youth sectors are the most represented. They are both the most numerous in France and the targets of the survey conducted. Volunteers over the age of 50 are in the majority in the sample. They were precisely the target of the survey, in terms of vaccination policy. The sample obtained is in line with the work carried out by Recherches & Solidarités with regard to individual characteristics and is not randomized because the target is reached. Table 1 Presentation of the questionnaire and responses distribution Question Proposed modalities to respondents Modalities name N Q1- In view of the current health risks and your volunteer activity You stay active by going into the field. Active 922 49.92% You have preferred on your own to temporarily interrupt your activity in the field. Chosen suspension 232 12.56% The leaders of your organization have asked you to suspend your volunteer activities. Imposed suspension 183 9.91% Your organization has temporarily ceased its activities in the field. Closure 510 27.61% Q2- When you are offered the vaccine, what will you do? I will get vaccinated without hesitation. Yes 1,146 61.95% I am hesitating but I plan to find out how to make my choice / I prefer to wait a little to have a better visibility I don’t know 582 31.46% I will not be vaccinated No 122 6.59% Q3A- How do you envisage your volunteer activity between now and your vaccination? (question available if the first answer was chosen to Q2) Without a vaccine, I do not wish to have a volunteer activity Yes, suspension until vaccine injection 165 14.84% I will continue or return to my volunteer activity with special care and attention No 947 85.16% Q3B- Do you think your volunteer activity might influence your desire to be vaccinated? (question available if the second or the third answer was chosen to Q2) Yes, above all to be more serene in my volunteer activity Yes, for me 74 10.44% Yes, primarily for the protection of the members or beneficiaries of my organization. Yes, for others 180 25.39% No, my volunteer activity has little influence on my position on the vaccine / I don’t see the link No link 455 64.17% Organizational sector Advocacy 43 2.34% Charity 78 4.24% Culture 119 6.46% Economy 128 6.95% Environment 49 2.66% Health 199 10.81% Leisure 70 3.80% Social 607 32.97% Sport 186 10.10% Youth 192 10.43% Other 170 9.23% What are the main publics your organization addresses? To all publics All 880 47.85% To young people Young 235 12.78% To adults Adults 185 10.06% To the elderly Older 133 7.23% To persons with disabilities or handicaps Handicap 70 3.81% To people in great difficulty In difficulty 270 14.68% To ill persons Ill 66 3.59% Gender Female F 1,009 54.48% Male M 846 45.52% Age Under 50 years old < 50 323 17.54% Between 50 and 65 years old 50–65 449 24.38% Over 65 years old > 65 1,070 58.09% Your volunteer commitment is approximately… Less than one hour per week < 1 h 357 19.36% Between 1 and 10 h per week 1–10 h 921 49.95% More than 10 h per week > 10 h 566 30.69% Does your current volunteer activity lead you to have contact with members, beneficiaries and/or the public? No 512 27.86% Yes 1,326 72.14% The variables are all categorical and Table 1 presents the descriptive statistics. In the context of the community-engaged research, the multivariate and scientific approach was assigned to me with the aim of publishing an academic article. I have carried out multiple logistic regressions. They are all relevant with regard to the R² obtained and likelihood tests. Multinomial logistic regressions were not used because the analysis by modality was the one chosen by Recherches & Solidarités. Finally, the outcomes of the research, the fourth step of Wallerstein et al. (2017)’s model, are the 66 vaccinated volunteers, the dissemination of good practices and the production of scientific knowledge on vaccine prioritization. Findings The first and the third findings helps to answer the research question RQ1 (To what extent has the pandemic weakened volunteering?) and the final three subsections are differentiated responses to the research question RQ2 (How can volunteer status be linked to vaccination willingness?). The Impact of the Covid-19 Pandemic on Volunteering Table 2 shows that the characteristics of volunteer involvement account for almost one-third of volunteer continuation (R²=0.30). The health crisis particularly affected NPOs in the cultural, leisure, advocacy and sports sectors. Stopping sports activities posed a medium-term health problem and the lack of leisure and cultural activities also had an effect on the citizens and volunteers’ mental health. Fragile populations also saw the NPOs around them lose their volunteers, especially the ill, the handicapped and the youngest people. Volunteers over 65 were the most likely to suspend their volunteering and they have chosen this interruption. In contrast, people who volunteer more than 10 h a week remained largely active in their NPOs. Similarly, NPOs in direct contact with people were less affected by volunteer suspensions because their mission was crucial in times of health crisis. Table 2 Logistic regression of the variable Q1 on current volunteer activity Imposed suspension Chosen suspension Active B p SE B p SE B p SE Constant -4.5955 * (1.0430) -2.3473 * (0.4791) 1.1277 ** (0.3699) Other 1.3880 (1.0564) 0.2597 (0.4993) -0.7129 ^ (0.3941) Sport -0.8208 (1.2456) -1.2244 * (0.5791) -0.8829 * (0.3902) Advocacy 1.3088 (1.1912) 0.7747 (0.5940) -1.1627 * (0.4862) Social 1.4496 (1.0355) -0.1811 (0.4786) -0.0402 (0.3704) Leisure 1.1962 (1.1010) -0.2935 (0.5929) -1.0094 * (0.4591) Charity 1.3360 (1.0984) 0.3451 (0.5434) -0.5613 (0.4370) Health 1.6883 (1.0540) 0.2459 (0.5099) -0.5274 (0.3992) Youth 1.5346 (1.0637) -0.2801 (0.5322) -0.6174 (0.4053) Culture 0.9836 (1.0786) -0.6979 (0.5598) -1.2132 (0.4214) Economy 0.9944 (1.0797) 0.3052 (0.5121) -0.6008 (0.4119) Young 0.4929 ^ (0.2991) 0.2592 (0.2584) -0.0288 (0.1932) Adults 0.1708 (0.3001) 0.0756 (0.2458) -0.1738 (0.1952) Older 0.4133 (0.3143) -0.3751 (0.3372) -0.0377 (0.2266) In difficulty 0.2530 (0.2662) -0.0790 (0.2430) 0.7116 * (0.1837) Ill 1.3137 *** (0.3789) -0.2936 (0.4459) -0.6273 (0.3311) Handicap 0.7674 ^ (0.4065) 0.1695 (0.3898) -0.3224 (0.2976) M 0.1172 (0.1720) 0.1838 (0.1525) 0.1913 ^ (0.1133) 50–65 -0.3983 ^ (0.2111) -0.4343 * (0.1922) 0.5766 * (0.1349) < 50 -0.7823 (0.2697) -0.5586 * (0.2230) 0.8043 * (0.1569) 1–10 h 0.8909 * (0.2365) 0.4509 * (0.1893) -0.7235 * (0.1281) < 1 h 1.2700 * (0.2713) 0.8029 * (0.2255) -1.2991 * (0.1728) Contact - N 0.6254 * (0.1751) 0.5914 * (0.1596) -1.7772 * (0.1372) -2Log(Likelihood) 116.28 * 75.86 * 455.19 * R²(Nagelkerke) 0.1332 0.0792 0.3030 N 1,766 1,766 1,766 Notes: Standard errors in parentheses. Comparison groups for categorical variables are: environmental sector, all public, female, age > 65, commitment of more than 10 h per week, contact with the public. For each regression conducted for this variable, 1 is assigned to the modality tested, 0 is assigned to the other modalities. ^ : p < .10; * p < .05; p < .01 and * p < .001. Willingness of Volunteers to be Vaccinated Table 3 indicates that volunteer involvement may play a role in the individual vaccination decision. Volunteers in the health, economic, social and charitable sectors were most likely to be vaccinated. Moreover, men and people over 65 were more concerned about vaccination. However, the NPO’s beneficiary public did not have a major effect. The decision of volunteers in direct contact with people is still uncertain; but volunteers who have chosen to suspend their activity are the most willing to be vaccinated. The results are therefore in line with Detoc et al. (2020). Table 3 Logistic regression of the modalities of the variable Q2 on willingness to be vaccinated Yes I don’t know No B p SE B p SE B p SE Constant 0.4473 (0.3769) -0.9781 (0.3726) -2.7352 * (0.6028) Other 0.8035 * (0.3675) -0.4771 (0.3580) -0.8707 (0.5516) Sport 0.9090 * (0.3655) -0.8612 * (0.3588) -0.0970 (0.4888) Advocacy 0.1951 (0.4665) 0.2922 (0.4536) -1.7390 (1.1066) Social 1.1698 *** (0.3460) -0.8633 * (0.3357) -0.7951 ^ (0.4781) Leisure 0.7695 ^ (0.4204) -0.5752 (0.4137) -0.4001 (0.6167) Charity 1.0808 (0.4175) -0.6676 (0.4120) -1.1274 (0.7365) Health 1.2340 (0.3763) -0.8435 * (0.3683) -1.2803 * (0.5882) Youth 0.8451 * (0.3828) -0.5391 (0.3743) -0.9313 (0.5869) Culture 0.7860 * (0.3829) -0.4114 (0.3719) -0.9321 (0.5687) Economy 1.2421 ** (0.3924) -0.9300 * (0.3871) -1.0611 ^ (0.6358) Young 0.4814 * (0.1960) -0.2952 (0.2018) -0.6752 ^ (0.3973) Adults -0.0997 (0.1825) 0.2616 (0.1848) -0.4656 (0.3789) Older 0.1198 (0.2222) 0.1667 (0.2243) -1.4715 * (0.7366) In difficulty 0.1741 (0.1763) -0.1120 (0.1843) -0.3254 (0.3508) Ill 0.0113 (0.3224) -0.3094 (0.3513) 0.5436 (0.5650) Handicap -0.4598 (0.2833) 0.7106 * (0.2804) -1.0225 (0.7418) M 0.4054 * (0.1098) -0.3383 (0.1133) -0.1982 (0.2123) 50–65 -0.9008 * (0.1250) 0.7548 * (0.1292) 0.8632 * (0.2449) < 50 -1.1666 * (0.1439) 0.9236 * (0.1461) 1.0591 * (0.2593) 1–10 h -0.0807 (0.1254) 0.1238 (0.1302) 0.0283 (0.2528) < 1 h -0.3247 * (0.1613) 0.1690 (0.1661) 0.6416 * (0.2885) Contact - N -0.2836 * (0.1277) 0.2262 ^ (0.1307) 0.1856 (0.2415) Imposed suspension -0.5684 * (0.2325) 0.5375 * (0.2403) 0.2557 (0.5085) Closure -0.6088 (0.1922) 0.5403 (0.1992) 0.3763 (0.3988) Active -0.6182 * (0.1845) 0.5024 (0.1919) 0.4984 (0.3890) -2Log(Likelihood) 205.46 * 138.30 * 74.91 *** R²(Nagelkerke) 0.1495 0.1062 0.1092 N 1,766 1,766 1,766 Notes: Standard errors in parentheses. Comparison groups for categorical variables are: Environmental sector, all public, female, age > 65, commitment of more than 10 h per week, contact with the public, chosen suspension of volunteering. For each regression conducted for this variable, 1 is assigned to the modality tested, 0 is assigned to the other modalities. ^ : p < .10; * p < .05; p < .01 and * p < .001. Vaccinate Volunteers to Encourage Their Return to the Field Table 4 focuses on volunteers who wish to be vaccinated and who will not volunteer without an injection. They are primarily volunteers in contact with elderly people, over 65 and with a commitment of less than 10 h per week. These volunteers have above all chosen to interrupt their commitment. However, one out of three volunteers in France is over 65. These volunteers are the most active: NPOs are therefore losing their main human resource. Table 4 Logistic regression of the variable Q3A of volunteering suspension until injection for volunteers wishing to be vaccinated Suspension until injection B p SE Constant -1.6019 ^ (0.8810) Other 0.2027 (0.8981) Sport 0.3730 (0.8938) Advocacy 0.8660 (1.0923) Social 0.3360 (0.8653) Leisure 0.4404 (0.9452) Charity 0.7851 (0.9308) Health 0.6977 (0.9118) Youth 0.5257 (0.9061) Culture 0.0014 (0.9154) Economy 0.5048 (0.9025) Young 0.3707 (0.3367) Adults -0.0693 (0.3479) Older 1.0224 ** (0.3647) In difficulty 0.1722 (0.3910) Ill 0.3950 (0.5551) Handicap -0.1575 (0.6995) M -0.1293 (0.2078) 50–65 -0.6028 * (0.2880) < 50 -0.6284 ^ (0.3694) 1–10 h 0.5274 * (0.2694) < 1 h 0.9088 (0.3215) Contact - N 0.6213 (0.2080) Imposed suspension -1.2658 *** (0.3192) Closure -0.4718 * (0.2356) Active -4.1292 * (0.5383) -2Log(Likelihood) 260.31 * R²(Nagelkerke) 0.3813 N 1,067 Notes: Standard errors in parentheses. Comparison groups for categorical variables are the same as the previous table. ^ : p < .10; * p < .05; p < .01 and * p < .001. Promoting Vaccination Thanks to the Relay of non-profit Organizations Table 5 analyzes the effects of volunteering on attitude towards vaccination. 65% of the respondents said that their volunteering activity has little influence over the vaccine intention. In fact, this absence of link is primarily the attitude of volunteers who are not very present in the field (less than one hour a week). Within the category of volunteers seeing a link between vaccination and volunteering, volunteers in contact with young people, volunteers in the field for more than 10 h a week and men are likely to be vaccinated to protect themselves. Volunteers who are in contact with elderly or disabled people or who are not currently in the field may want to protect others by getting vaccinated. Table 5 Logistic regression of the modalities of the variable Q3B on the influence of volunteerism on vaccination attitude Yes. for me Yes. for others No link B p SE B p SE B p SE Constant -2.2767 ** (0.7095) -1.1227 * (0.5182) 0.5998 (0.4733) Other -0.0340 (0.7775) 0.7052 (0.5518) -0.6348 (0.5125) Sport -1.2856 (0.8973) 0.0701 (0.5576) 0.3350 (0.5149) Advocacy -0.0906 (0.9206) 0.8162 (0.6559) -0.7295 (0.6225) Social 0.1027 (0.7122) 0.2621 (0.5224) -0.2573 (0.4782) Leisure -0.3206 (0.9969) -0.1122 (0.6752) 0.2158 (0.6160) Charity 0.0609 (0.9324) 0.0411 (0.7073) -0.1190 (0.6326) Health 0.1988 (0.7676) -0.0185 (0.5883) -0.1867 (0.5292) Youth -1.2039 (0.8800) 0.8089 (0.5916) -0.2306 (0.5478) Culture -0.5131 (0.9052) -0.0915 (0.6041) 0.2338 (0.5547) Economy -0.3714 (0.8828) -0.5199 (0.7017) 0.4892 (0.6116) Young 1.8498 *** (0.4654) -0.3507 (0.3760) -0.5634 ^ (0.3231) Adults 1.0029 * (0.4636) 0.4643 (0.3101) -0.7802 (0.2925) Older 0.4883 (0.6089) 0.9792 (0.3748) -1.0535 (0.3697) In difficulty 0.5260 (0.4740) 0.2401 (0.3252) -0.4068 (0.3037) Ill 1.1959 ^ (0.6858) 0.1490 (0.6202) -0.6555 (0.5224) Handicap 0.6271 (0.6118) 1.0619 (0.4055) -1.2317 ** (0.4101) M 0.5801 * (0.2911) 0.2090 (0.1992) -0.4469 * (0.1862) 50–65 0.2107 (0.3246) -0.4193 ^ (0.2263) 0.3033 (0.2082) < 50 0.1980 (0.3639) -0.1301 (0.2400) 0.0581 (0.2246) 1–10 h -0.8143 * (0.3239) -0.1831 (0.2264) 0.5033 * (0.2122) < 1 h -1.5084 ** (0.4815) -0.5738 * (0.2933) 1.0388 *** (0.2748) Contact - N -0.0397 (0.3591) -0.4797 * (0.2405) 0.4354 * (0.2203) Chosen suspension 0.2875 (0.4884) -0.1647 (0.3767) 0.0113 (0.3334) Imposed suspension 0.6430 (0.4666) 0.2068 (0.3482) -0.4049 (0.3210) Closure -0.2758 (0.3997) 0.2599 (0.2431) -0.1462 (0.2284) -2Log(Likelihood) 48.43 47.24 72.10 *** R²(Nagelkerke) 0.1455 0.1003 0.1407 N 664 664 664 Notes: Standard errors in parentheses. Comparison groups for categorical variables are the same as the previous table. For each regression conducted for this variable, 1 is assigned to the modality tested, 0 is assigned to the other modalities. ^ : p < .10; * p < .05; p < .01 and * p < .001. Discussion This section lists the research questions and their final status. The discussion then moves on to the implications for the community and then for Recherches & Solidarités and researchers; as they are targets for a community-engaged research. Research Questions Statement Table 6 proposes a synthesis of the results and contributions in particular for countries building their vaccination strategy. The answers to the research questions are quite clear. On the one hand, in response to RQ1 (To what extent has the pandemic weakened volunteering?), the Covid-19 crisis had a massive effect on volunteering. On the other hand, in response to RQ2 (How can volunteer status be linked to vaccination willingness?), the link between vaccination and volunteering is not obvious. However, when conducting regression analyses, the link appears: the most committed people and those in the field are particularly concerned. If the volunteers had been among the priority groups to be vaccinated, people would have been able to return to their beneficiaries more quickly (since they had stopped their activities because of the virus). Table 6 A summary of the results and contributions Conducted analysis Result Comment Contribution for other countries Research question 1 (RQ1): To what extent has the pandemic weakened volunteering? Study of the impact of the Covid-19 pandemic on volunteering 22.5% of volunteers stopped their activities due to fear of the virus (not due to administrative closure). Vulnerable groups, as beneficiaries of these volunteers, have lost the support of these volunteers. The social impact of the forced closure of NPOs must be questioned, especially when they target vulnerable groups. 27.6% of NPOs were closed. Research question 2 (RQ2): How can volunteer status be linked to vaccination willingness? Analysis of willingness to be vaccinated within volunteers NPO volunteers were more likely to want to be vaccinated than the rest of the population. If volunteers had been prioritized in the vaccination strategy, more people could have returned to support NPOs’ beneficiaries. The inclusion of volunteers in the list of key occupations and activities is a relevant avenue to explore, depending on the national context. Volunteers who suspended their activity were more likely to be vaccinated. Analysis of the return to the field of volunteers who suspended their activity 15% of volunteers refuse to get involved without a vaccine, especially those over 65. Volunteers over 65 are the most numerous in French NPOs. Prioritization by age could be coupled with prioritization by core occupation or activity. Analysis of the link between volunteering and desire to be vaccinated 35% of volunteers see a link between the two. Volunteers who do not see a link between the two are those who are not very involved and who are not in contact with the public. The link between vaccination and volunteering exists among those most involved and in contact with the public, of which NPOs are the most dependent. In order to answer the research question RQ3 (To what extent has the community-engaged research conducted by Recherches & Solidarités served the NPOs?), the chosen method unfortunately did not lead to changes in public vaccination policy in France in 2021. Presently, the French vaccination strategy is to target hesitant people and no longer to prioritize fragile people. We therefore have a little hindsight on the present French strategy. Nevertheless, the French experience of prioritization by age can provide lessons for other countries developing their vaccination strategy. The horizon of this community-engaged research is thus broader than expected, even if it had not a direct effect on the French community, except for the 66 vaccinated volunteers. Implications for the Community (Public Authorities and NPOs) First of all, the results show a nonprofit sector in difficulty due to the suspension of the commitment of certain volunteers (Table 2). In social, health and charity NPOs, 40% of volunteers between 50 and 75 have withdrawn while needs are increasing. People who are ill or in difficulty need volunteers from NPOs in addition to the accompaniment of healthcare professionals. However, they are deprived of a part of their action. The results then show greater vaccine acceptance among volunteers in the social, health and charity sectors (Table 3). In other words, not including volunteers among essential workers can be seen as a gap opened by previous studies and health recommendations. Volunteers in priority sectors (maybe identified by public authorities) could thus be added to essential workers. This could protect NPOs’ viability, because they have few resources and often only have their volunteers to run their operations and carry out their mission with a positive social impact. In addition, beyond the health and social sectors, volunteers from the sports and cultural sectors, from community animating sectors and sectors with low dematerialization capability could also be integrated in a later phase. NPOs indeed have an important societal role to counter the negative effects of the health crisis. Second, NPOs may have a role in promoting vaccination, even if it is limited to their communities. Within volunteers in hesitation or against vaccination, a nudge from their NPOs can make them to rethink about it, even if it only concerns 36% of them. These results illustrate recent publications on the subjective nature of the individual vaccine decision. Chou and Budenz (2020) indeed invites to “activate positive emotions” to reinforce the “prosocial motivations” of people hesitating to be vaccinated. The positive message sent by NPOs promoting vaccination for their members can have a great impact. Through direct discourse or communication on their internal vaccination policy, NPOs become opinion makers for their community. Such action is a key to success of the vaccination strategy (French et al., 2020; Schoch-Spana et al., 2020). With 12.5 million volunteers (Bazin et al., 2022), French GVOs have a large community that could be sensitized. This is also the case for other countries. Implications for Recherches & Solidarités and for Scholars There is also a contribution for Recherches & Solidarités. In view of the results, their proposal can be seen as relevant. In particular, the organization was concerned about interruptions in volunteering in the social, health and charitable sectors. The proposal meets its objective: volunteers in these sectors were willing to be vaccinated. Immediate practical and professional contributions existed: (1) for the design of health initiatives with NPOs, (2) in order to justify the legitimacy and relevance of these initiatives to the public authorities, (3) to prevent wastage of vaccine doses and (4) to ensure that vaccinated volunteers were able to return to the field. The general public and the press embraced the initiative, but the paradox between centralized health decision-making and decentralized initiatives (Yang, 2020) remained strong in France and the experimentation was not generalized. Even if this initiative has not found its place in France, the present survey has demonstrated that volunteers are a relevant public in the vaccination strategy. If the authorities had retained the proposed criteria (of age and sector of activity), the vaccination policy would not have been upset. 20.3 million of French people are between 50 and 74. 4.85 million are volunteers, and more precisely 1.5 million in the health, charity and social sectors. According to the survey conducted, 61% of the target population are in the field, i.e. 915,000 volunteers. 71% would be willing to be vaccinated, or 650,000 people. The proposed initiative therefore covers less than 1% of the French population. In this, this proposal can be generalized internationally and the experimentation (Sebbag et al., 2021) could inspire other countries. The results also provide information on public governance and NPOs’ capacity building. On the one hand, the results highlight the limitations of centralized vaccine logistics and of a top-down approach. While one vaccination center implemented an accurate and cost-effective process, it has not been replicated. The public administration suffered from a lack of organizational learning and recognition of local initiatives. On the other hand, based on recent research on non-profit capacity (Nordin et al., 2022), some lessons can be drawn. First, the dependence of NPOs on volunteers largely explains the organizational failures. The pandemic has given rise to a new practice of remote volunteering that could help to overcome this shortcoming. Second, while financial aspects are often crucial, here, NPOs would have needed above all clarity on health measures. At the beginning of 2021, France was experiencing a new Covid-19 wave that finally led to a new lockdown. The uncertainty was very difficult to manage and the non-profit capacity also relies on trusting relationships with partners. Finally, NPOs have demonstrated their capacity for innovation and absorption of shocks as global as a pandemic. While there is a tendency to attempt to impose corporate practices on NPOs, it is undoubtedly a good idea to draw inspiration from what NPOs have done to survive. Conclusion This article is based on a community-engaged research conducted by a collaborative network of professionals and scholars. It sought to put NPOs and volunteers at the heart of the vaccine strategy. These people are still neglected in prioritization policies despite the inclusion of essential workers in some of the vaccination plans. The results of the statistical analyses show that the NPO sector is suffering from the health crisis and is lacking of human resources. The potential of this sector to act against the current crisis is highlighted too. This article therefore calls for adaptations and agility (Joyce, 2021) in the vaccination strategy, such as the initiative led by Recherches & Solidarités, or for a strong signal sent to volunteers in strategic sectors by including them in an earlier phase of the vaccination plan. This article has thus responded to the scholars’ demands, both in terms of analyzing the role of NPOs despite their limited resources (French et al., 2020) and in terms of surveys of specific sub-populations to understand their attitudes and beliefs towards Covid-19 (Schoch-Spana et al., 2020). It also extended the analysis and ground proposal of a non-profit actor into a possible public policy. Finally, the French vaccination policy is now open to all persons who wish to be vaccinated. In the short and medium term, the results of this article may be useful to other countries with the same problems of access to vaccine doses. In the longer term, the findings could inspire French and other countries’ considerations on vaccine policies and potential prioritization. This article has essentially theoretical limitations: its aim has been to discuss the French authorities’ vaccine strategy. It is therefore not anchored on a theoretical framework to be tested with hypotheses. Furthermore, the survey was produced by Recherches & Solidarités with no possibility of adjusting the questions or the diffusion. These limitations could be overcome by including a theoretical framework such as the terror management theory (e.g. Pyszczynski et al., 2021) and producing a survey with a mixed methodology. In addition, this limitation may be of interest to other researchers involved in community-engaged research. It seems important to reinsert the conceptual and theoretical issues earlier in the process to avoid the pitfall encountered here. Finally, the impact of including volunteers in the vaccination strategy is a new research direction. A survey taking into account the current context of a majority of vaccinated people versus a resistant minority might be useful. Finally, understanding why the French authorities never retained the initiative proposed by Recherches & Solidarités would be useful, in line with Baekkeskov (2016). Acknowledgements The author wishes to thank Jacques Malet, Cécile Bazin and Marie Duros of Recherches & Solidarités for their kind provision of the Covid-19 survey database and their support in the research process undertaken on the basis of this survey. Funding: No funding was received for conducting this study. Data Availability The data that support the findings of this study are available from Recherches & Solidarités. Declarations Informed Consent: Respondents in the data have freely chosen to answer the questionnaire online. All data is stored without any identifying information. Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Conflict of interest The author is a board member and a member of the expert committee of Recherches & Solidarités, and receives no compensation. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Baekkeskov E Same threat, different responses: experts steering politicians and stakeholders in 2009 H1n1 vaccination policy-making Public Administration 2016 94 2 299 315 10.1111/padm.12244 Bazin, C., Duros, M., Bastiani, B., Ben Ayed, A., Fauritte, J., & Malet, J. (2022). La France Associative en Mouvement—20ème édition (p. 44). Recherches & Solidarités. https://recherches-solidarites.org/wp-content/uploads/2022/10/La-France-associative-04-10-2022.pdf Bazin, C., & Malet, J. (2021). Les bénévoles à l’heure du vaccin—Focus sur les bénévoles de 50 74 ans des associations à caractère social. Recherches & Solidarités. Belle N Cantarelli P Nudging public employees through descriptive social norms in healthcare organizations Public Administration Review 2021 10.1111/puar.13353 Chirico, F., Nucera, G., & Magnavita, N. (2020). COVID-19: protecting Healthcare Workers is a priority. Infection Control & Hospital Epidemiology, 41(9), 10.1017/ice.2020.148. Article 9. Chou, W. Y. S., & Budenz, A. (2020). Considering emotion in COVID-19 Vaccine Communication: addressing vaccine hesitancy and fostering vaccine confidence. Health Communication, 35(14), 10.1080/10410236.2020.1838096. Cunliffe AL Scaratti G Embedding impact in Engaged Research: developing socially useful knowledge through Dialogical Sensemaking: embedding impact in Engaged Research British Journal of Management 2017 28 1 29 44 10.1111/1467-8551.12204 Detoc, M., Bruel, S., Frappe, P., Tardy, B., Botelho-Nevers, E., & Gagneux-Brunon, A. (2020). Intention to participate in a COVID-19 vaccine clinical trial and to get vaccinated against COVID-19 in France during the pandemic. Vaccine, 38(45), 10.1016/j.vaccine.2020.09.041. French J Deshpande S Evans W Obregon R Key guidelines in developing a pre-emptive COVID-19 vaccination Uptake Promotion Strategy International Journal of Environmental Research and Public Health 2020 17 16 16 10.3390/ijerph17165893 Giubilini, A., Savulescu, J., & Wilkinson, D. (2020). COVID-19 vaccine: vaccinate the young to protect the old? Journal of Law and the Biosciences, 7(1), 10.1093/jlb/lsaa050. Hassan-Smith, Z., Hanif, W., & Khunti, K. (2020). Who should be prioritised for COVID-19 vaccines? Lancet (London England), 396(10264), 10.1016/s0140-6736(20)32224-8. Article 10264. Heyerdahl, L. W., Vray, M., Leger, V., Le Fouler, L., Antouly, J., Troit, V., & Giles-Vernick, T. (2021). Evaluating the motivation of Red Cross Health volunteers in the COVID-19 pandemic: a mixed-methods study protocol. British Medical Journal Open, 11(1), 10.1136/bmjopen-2020-042579. Jeyanathan, M., Afkhami, S., Smaill, F., Miller, M. S., Lichty, B. D., & Xing, Z. (2020). Immunological considerations for COVID-19 vaccine strategies. Nature Reviews Immunology, 20(10), 10.1038/s41577-020-00434-6. Joyce P Public governance, agility and pandemics: a case study of the UK response to COVID-19 International Review of Administrative Sciences 2021 87 3 536 555 10.1177/0020852320983406 Lazarus, J. V., Ratzan, S. C., Palayew, A., Gostin, L. O., Larson, H. J., Rabin, K., Kimball, S., & El-Mohandes, A. (2021). A global survey of potential acceptance of a COVID-19 vaccine. Nature Medicine, 27(2), 10.1038/s41591-020-1124-9. Le Mouvement Associatif, RNMA, Recherches & Solidarités, & Ministère de l’éducation nationale et de la jeunesse - DJEPVA (2021). Covid-19: Où en sont les associations un an après ? (p. 60). https://recherches-solidarites.org/wp-content/uploads/2021/06/COVID-3-8-06-2021.pdf Ministère des Solidarités et de la Santé (2021). La stratégie vaccinale. Ministère des Solidarités et de la Santé. https://solidarites-sante.gouv.fr/grands-dossiers/vaccin-covid-19/article/la-strategie-vaccinale Misener K Morrison K Shier M Babiak K The influence of organizational social responsibility on involvement behavior in community sport membership associations Nonprofit Management and Leadership 2020 30 4 591 611 10.1002/nml.21406 Nesbit R Christensen RK Brudney JL The limits and possibilities of volunteering: a Framework for explaining the scope of Volunteer involvement in Public and Nonprofit Organizations Public Administration Review 2018 78 4 502 513 10.1111/puar.12894 Nordin N Khatibi A Azam SMF Nonprofit capacity and social performance: mapping the field and future directions Management Review Quarterly 2022 10.1007/s11301-022-00297-2 O’Flynn, J. (2021). Confronting the big challenges of our time: making a difference during and after COVID-19. Public Management Review, 23(7), 10.1080/14719037.2020.1820273. Article 7. Persad, G., Peek, M. E., & Emanuel, E. J. (2020). Fairly prioritizing groups for Access to COVID-19 vaccines. Journal Of The American Medical Association, 324(16), 10.1001/jama.2020.18513. Article 16. Pyszczynski, T., Lockett, M., Greenberg, J., & Solomon, S. (2021). Terror Management Theory and the COVID-19 pandemic. Journal of Humanistic Psychology, 61(2), 10.1177/0022167820959488. Reiter, P. L., Pennell, M. L., & Katz, M. L. (2020). Acceptability of a COVID-19 vaccine among adults in the United States: how many people would get vaccinated? Vaccine, 38(42), 10.1016/j.vaccine.2020.08.043. Ribas, A., Sengupta, R., Locke, T., Zaidi, S. K., Campbell, K. M., Carethers, J. M., Jaffee, E. M., Wherry, E. J., Soria, J. C., & D’Souza, G. (2021). Priority COVID-19 vaccination for patients with Cancer while Vaccine Supply is Limited. Cancer Discovery, 11(2), 10.1158/2159-8290.CD-20-1817. Russell, F. M., & Greenwood, B. (2020). Who should be prioritised for COVID-19 vaccination? Human Vaccines & Immunotherapeutics, 1–5. 10.1080/21645515.2020.1827882. Schoch-Spana, M., Brunson, E. K., Long, R., Ruth, A., Ravi, S. J., Trotochaud, M., Borio, L., Brewer, J., Buccina, J., Connell, N., Hall, L. L., Kass, N., Kirkland, A., Koonin, L., Larson, H., Lu, B. F., Omer, S. B., Orenstein, W. A., Poland, G. A., & White, A. (2020). The public’s role in COVID-19 vaccination: Human-centered recommendations to enhance pandemic vaccine awareness, access, and acceptance in the United States. Vaccine, S0264410X20313682. 10.1016/j.vaccine.2020.10.059 Sebbag, R., Joseph, B., Huguet, D., Bouffard, M. C., Duros, M., Bazin, C., & Malet, J. (2021). Les bénévoles à l’heure du vaccin L’exemple de Châteaudun (Eure & Loir). Recherches & Solidarités. Stanton TK New times demand new scholarship: Opportunities and challenges for civic engagement at research universities Education Citizenship and Social Justice 2008 3 1 19 42 10.1177/1746197907086716 Tchernonog, V., & Prouteau, L. (2019). Le paysage associatif français: Mesures et évolutions. Dalloz. Touboulic A McCarthy L Matthews L Re-imagining supply chain challenges through critical engaged research Journal of Supply Chain Management 2020 56 2 36 51 10.1111/jscm.12226 Tsheten T Tenzin P Clements ACA Gray DJ Ugyel L Wangdi K The COVID-19 vaccination campaign in Bhutan: strategy and enablers Infectious Diseases of Poverty 2022 11 1 6 10.1186/s40249-021-00929-x 34986883 Wallerstein, N., Duran, B., Oetzel, J., & Minkler, M. (2017). Community-based Participatory Research for Health: advancing Social and Health Equity. Jossey-Bass. Yang J Zheng W Shi H Yan X Dong K You Q Zhong G Gong H Chen Z Jit M Viboud C Ajelli M Yu H Who should be prioritized for COVID-19 vaccination in China? A descriptive study BMC Medicine 2021 19 1 45 10.1186/s12916-021-01923-8 33563270 Yang, K. (2020). Unprecedented Challenges, familiar Paradoxes: COVID -19 and governance in a New Normal State of Risks. Public Administration Review, 80(4), 10.1111/puar.13248. Article 4. Yin, F., Ji, M., Yang, Z., Wu, Z., Xia, X., Xing, T., She, Y., & Hu, Z. (2022). Exploring the determinants of global vaccination campaigns to combat COVID-19. Humanities and Social Sciences Communications, 9(1), 10.1057/s41599-022-01106-7.	0	PMC9718455	NO-CC CODE	2022-12-06 23:23:39	no	Public Organiz Rev. 2022 Dec 2;:1-17	utf-8	null	null	null	oa_other
==== Front Anal Sci Anal Sci Analytical Sciences 0910-6340 1348-2246 Springer Nature Singapore Singapore 216 10.1007/s44211-022-00216-1 Review Promising instrument-free detections of various analytes using smartphones with Spotxel® Reader Qin Ningyi 1 Liu Zirui 2 Zhao Lanbin 3 Bao Mengfan 1 Mei Xifan [email protected] 23 http://orcid.org/0000-0002-3452-0798 Li Dan [email protected] 1 1 grid.454145.5 0000 0000 9860 0426 Department of Pharmacy, Jinzhou Medical University, Jinzhou, 121000 China 2 grid.454145.5 0000 0000 9860 0426 Liaoning Provincial Key Laboratory of Medical Testing, Jinzhou Medical University, Jinzhou, 121001 China 3 grid.454145.5 0000 0000 9860 0426 The Third Affiliated Hospital, Jinzhou Medical University, Jinzhou, People’s Republic of China 2 12 2022 110 3 9 2022 8 11 2022 © The Author(s), under exclusive licence to The Japan Society for Analytical Chemistry 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. In consideration of the problems related to food safety, environmental pollution, and the spread of infected diseases nowadays, we urgently need testing methods that can be easily performed by common people. Smartphone-based detections are promising for general applications. However, some of these analytical strategies require a combination of accessories and instruments, such as portable electrochemical workstations, mini multi-mode microplate readers, and complex temperature control devices, etc., which are small but still expensive. Herein, we comprehensively introduce a free app (Spotxel® Reader) that can provide accurate data analysis for microplate or parallel-format test sensors without an instrument. By simulating the optical signal of the test samples through a smartphone, the sensing results can be obtained for free. We discuss the detection strategies involved in the reported smartphone-based analyses using Spotxel® Reader. Prospects for the development of this free app for future detection applications are presented. This review aims to popularize free analysis software, so that ordinary people may realize convenient tests. Graphical abstract Keywords Spotxel® Reader Instrument-free Wide available Smartphone University Student Innovation Project10160101 Li Dan ==== Body pmcIntroduction Many hazards may threaten human health these days. For instance, the recently emerged SARS-CoV-2 that causes COVID-19, seriously influence people’s life [1]; The food contaminants such as aflatoxin are often found in cereals and has a high risk to cause cancer [2, 3]; Some microbial such as Helicobacter pylori may be infected by sharing food and seriously damage people’s stomach [4]. These toxic, harmful, or over-expressed substances may threaten people’s health, but people cannot know their safety without detection. If the detections required professional and expensive testing methods, enough tests may not be performed, since people are tired of complicated analyses. On the other hand, the development of inexpensive and convenient methods that are preferred by the public will make an important contribution to the choices of more detections, which in turn prevent harm to attack people’s health [5]. Recently, smartphone-based detection methods have been developed, which have a great opportunity to be adopted by normal people due to their convenience [6–8]. For instance, an analyte may cause a color or fluorescence change in a sensor/probe [9–11]. These sensors can be developed by test chips, which may show a visible color change in the presence of a certain amount of the analyte [12–14]. The naked eye alone can only roughly judge color change, but cannot accurately evaluate the concentration of the analyte [15–17]. On the other hand, the smartphone can photograph this optical signal change, and then use appropriate software to transmit the specific value of the signal, and calculate the concentration of the target [18, 19]. To achieve low-cost analysis, optical signal analysis software including Color picker [20–22], ImageJ [23–25], Matlab [21, 26–28], etc., have been used for the determination of different analytes in combination with a smartphone. For instance, the Color picker can convert an image obtained from a smartphone to RGB information. This information has to be converted into an optical signal by the researchers using a defined formula. Then, the data can be further correlated with the concentration of the analyte. In addition, this method can only analyze one image at a time, so it needs further simplification and optimization for sample analysis. ImageJ supports image processing, such as logical and arithmetical operations, convolution, contrast manipulation, Fourier analysis, and edge detection. This program supports the analysis of several images simultaneously. However, to establish the connection between the analyte and the image from the smartphone, the researcher needs to establish a formula to obtain the concentration information of the analyte, which is still difficult for ordinary users. Matlab can be used to analyze sensing data, which allows numeric computing, matrix manipulations, plotting of functions and data, and implementation of algorithms with a specific written program. Thus, it is relatively complicated and requires some professional training [26, 29]. Some other smartphone-based detections can be performed at the point of care and obtain the data directly using mature programs, but expensive instruments such as Sensit smart are involved [30–32]. Spotxel® Reader is a free smartphone app that can read the color or fluorescence intensity of multiple samples in microplate wells and other samplers with plate formats (Fig. 1). It can also analyze images in rows, columns, or microarray formats. When people use Spotxel® Reader software to analyze the test objects, they no longer need to create formulas or further analysis. The premium version of this app enables the simulation of the assay’s standard curves and shows the unknown samples’ concentrations directly. This software mainly analyzes the intensity of optical signals through four channels including blue, yellow, purple, and brightness. Since its development in 2017, the Spotxel® Reader has been used by many researchers for the rapid and inexpensive detection of several analytes including environmental pollutants [33, 34], microbial [35], disease-related biomarkers [36–38], etc. The detained mechanism, detection limit, and analysis accuracy are shown in Table 1. However, there is still limited research on using this app to enable wide-available tests for common people. Herein, we introduce in detail the reported smartphone-based detection strategy using Spotxel® Reader to analyze different analytes. The related mechanisms of different strategies are systematically discussed. A prospect on the new applications of using a free app for future detections is also proposed. This review is aiming to enable common people to benefit from the free apps for cost-effective tests, which may guard people’s health against the attack of various hazards at early stages. Fig. 1 Spotxel® Reader app: bare interface (a), camera interface (b), analysis array design interface (c), and its optical signal intensity result display (d) Table 1 Simple analysis of various analytes based on Smartphones with Spotxel® Reader Probe Mechanisms Detection limit; accuracy Analytes Equipment Reader Ref. m-CDs@SiO2 The probe exhibited fluorescence changes on interaction with pyrethroids under UV irradiation 0.048 μg/L; 87.93–101.4% λ-Cyhalothrin A portable UV-integrated box with a 365 nm light source Smartphone with Spotxel® Reader, Version 1.5.5 [33] RCDs The fluorescence of RCDs-based nanomaterials was quenched by interactions with λ-Cyhalothrin 6.66 μg/L; 88–104% λ-Cyhalothrin A portable ultraviolet light box Smartphone with Spotxel® Reader, Version 1.5.5 [34] RT-LAMP Assay Primer recognition of the target genome of SARS-CoV-2 leads to a colorimetric reaction Sensitivity: 81%; Specificity: 83%; Accuracy: 75–86% SARS-CoV-2 65 °C incubator Spotxel® Reader: Germersheim, Germany, 2017 [35] DTA-functionalized Tp DTA-functionalized Tp showed chiral recognition with fluorescence change 13 ng mL−1; 86.00–118.33% Chiral amino acids None Spotxel® Reader (version: not mentioned) [36] TPTA-assembled GQDs Chiral TPTA-modified GQDs showed fluorescence change with the recognition of D‑phenylalanine 0.050 μM; 86.20–110.0% D-Phe A dark UV analyzer with 365 nm irradiation light Spotxel® Reader (version: not mentioned) [37] Humanized organ-on-a-chip model The model showed synergistic paracrine signaling from sympathetic neurons Not mentioned Neuro-breast cancer crosstalk G-Series Human Bone Metabolism Array 1000 (RayBiotech) Spotxel® software (Version 2.2.2, SICASYS Software GmbH) [38] m-CDs@SiO2 silica nanoparticles embedded with carbon dots (CDs) based on m-phenylenediaminem, RCDs red-emission carbon dots; ( +)-diacetyl-L-tartaric anhydride-functionalized 1,3,5-triformylphloroglucinol, TPTA triformylphloroglucinol-functionalized chiral ( +)-diacetyl-L-tartaric anhydride, GQDs graphene quantum dots, D-Phe D-phenylalanine Evaluation of food safety Aflatoxin B1 (AFB1) is a carcinogen that naturally exists in food, especially spoiled grains [39–41]. There are many test devices for detecting ABF1 on the market based on the observation of color or fluorescence change [42]. However, the sensitivity of most naked eye-based tests can not accurately detect AFB1 in food at relatively low concentrations, though the amount may already be toxic. Some bioreceptors such as antibodies and aptamers can recognize AFB1, but the price is mostly considerable [43–45]. Molecularly Imprinted (MIP) membranes have synthetic binding sites, which can mimic natural bioreceptors to recognize ABF1 [46]. Meanwhile, these synthesized membranes are cost-effective to meet large-scale productions. Sergeyeva et al. developed a smartphone-based fluorescent sensor using MIP membranes (Fig. 2a) [47]. Two functional monomers [2-acrylamido-2-methyl-1-propansulfonic acid (AMPSA) and acrylamide (AA)] have binding sites to interact with AFB1 selectively. Although the authors did not mention the detailed binding sites, b2 and b3 of AMPSA and AA from the selected molecules containing amide-conjugated alkene groups might recognize b1 of ABF1 based on Michael reaction (Fig. 2b). With UV irradiation and the interaction of MIP membranes with the samples, the fluorescence changes as a function of the concentration of the target. This signal can be read by a smartphone with Spotxel® Reader, which calculates the concentration of AFB1 subsequently. The detection limit of ABF1 was 20 ng mL−1. By analysis of different extracts of the aflatoxin B1-free wheat and maize flour spiked with AFB1 of 10 ng mL−1, 50 ng mL−1, 80 ng mL−1, and 100 ng mL−1, 80–94%, 83–109%, 78–114%, and 88–104% recoveries were obtained. This method facilitates a cost-effective and simple approach to food safety evaluation.Fig. 2 Evaluation of food safety using smartphone with Spotxel® Reader based on MIP sensing: a Scheme for the detection of AFB1 using an MIP-membrane-based smartphone sensor. Reproduced with permission from [47] by Elsevier. b the structure of AFB1 and several molecule candidates for recognizing ABFB1; the possible binding sites were marked based on Michael reaction Fumonisin B1 (FB1) is a carcinogenic mycotoxin that is present in food and can cause health threats to animals or humans [48–50]. Yu et al. developed a colorimetric assay for the detection of FB1 (Fig. 3) [51]. Initially, AuNP@MnO2 nanoparticles were synthesized from KMnO4 and 13 nm AuNPs (Fig. 3A). Alkaline phosphatase (ALP) labeled goat anti-mouse IgG interacted with the target through the link of FB1 monoclonal antibody (McAb) (Fig. 3B). FB1-BSA was coated on the microplate and competes with FB1 to bind the McAb [Fig. 3B(a, b)], which further captured ALP-IgG [Fig. 3B(c)]. This conjugate hydrolyzed ascorbic acid 2-phosphate (AAP) and generated ascorbic acid (AA), which reduced the MnO2 shell to Mn2+ [Fig. 3B(d, e)]. The released Mn2+ induced the aggregation of gold nanoparticles (Au NPs), resulting in a color change of the colloid that can be read by the smartphone with Spotxel® reader [Fig. 3B(f)]. The method has a detection limit of 0.15 ng/mL and a recovery rate of 86.4–110.9% for FB1 in the maize flour samples.Fig. 3 Evaluation of food safety using smartphone with Spotxel® Reader based on an antibody-modified nanosensor: A schematic of synthesis of AuNP@MnO2. B Mn2+-mediated aggregation of the colloid. a coating of FB1-BSA on the well, b Competitive reaction, c ALP-IgG interacted with McAb, d the hydrolysis of AAP by ALP, e ascorbic acid induced the generation of Mn2+ causing aggregation of AuNPs. f Spotxel® Reader analysis. Reproduced with permission from [51] by Elsevier Evaluation of food nutrition α-Lactalbumin (α-LA) is a component of the milk proteins and may possess some important functions of immunologic defense, which may induce apoptosis in lymphoid cell lines [52]. Zhang et al. employed a facile hydrothermal method and synthesized laccase mimics (named LM nanozymes) using glutathione (GSH) and copper (II) chloride as precursors [53]. LM nanozymes catalyze the oxidative coupling reaction between 4-aminoantipyrine (4-AP) and 2,4-dichlorophenol (2,4-DP) resulting in the production of a red product. This phenomenon was used for the ELISA analysis of alpha-lactalbumin (α-LA) (Fig. 4). LM nanozymes were modified by target antibodies that recognize α-LA and show color change by catalyzing the substrate in the presence of 4-AP and 2,4-DP. A smartphone with Spotxel® Reader detected the signal change and showed a detection limit of 0.056 ng/mL with high specificity for α-LA in food samples. The method was accurate as the recovery of α-LA from spiked raw milk, UHT milk, yogurt, candy, EHF, and PHF were 100.04%, 101.17%, 99.24%, 97.64%, 96.51%, and 99.48%, respectively. These strategies may enable the development of an inexpensive method for food monitoring for public use.Fig. 4 Evaluation of food nutrition using smartphone with Spotxel® Reader based on antibody-modified nanozymes: schematic illustration of the integration system of nanozyme-based immunoassay with a smartphone for the detection of α-LA. Reproduced with permission from [53] by Elsevier Detection of pesticides Pyrethroids are commercial pesticides that are toxic to insects, such as dragonflies, bees, gadflies, mayflies, and some other invertebrates [54–56]. The presence of pyrethroids in the aquatic environment may be toxic to fish and other organisms [57–59]. They also threaten people’s health indirectly. Jiang et al. developed a smartphone-based fluorescence detection method for the analysis of pyrethroids (Fig. 5). The artificial receptors inside a covalent Carbazole (CAR)-conjugated framework (CCFs) with strong fluorescence were used to recognize pyrethroids, such as λ-cyhalothrin (LC). The fluorescence changes as a function of the target under UV excitation at 365 nm. The smartphone with Spotxel® Reader detected the change and calculated the concentration of LC with a detection limit of 4.067 μg L−1. A recovery ratio of 88–103% was found for LC in food samples, which indicated the current method could accurately determine this pesticide.Fig. 5 Evaluation of pesticides using smartphone with Spotxel® Reader based on artificial receptors inside a CCFs sensor: scheme for the fabrication of the CARs-CCFs and fluorescence sensing of pyrethroids using artificial receptors bound inside a covalent organic framework by smartphone-based analysis; b the quenching constant of the CARs-CCFs and control group for LC and its analogs; c Spotxel® Reader software processing results. Reproduced with permission from [60] by Elsevier Zhu et al. synthesized red fluorescent carbon dots (RCDs) and developed molecular imprinting technology to detect pyrethroids, i.e., Lambda-cyhalothrin (LC) [33]. The –NH2 groups on the surface of the RCDs interacted with LC, allowing the detection of pyrethroids in the concentration range of 1–120 μg/L) with a detection limit of 0.89 μg/L using a Synergy H1 microplate reader. Furthermore, a portable UV light box was used to induce fluorescence of the RCDs-based sensor, the smartphone with Spotxel® Reader facilitate the sensitive detection of LC without an instrument and the detection limit was 6.66 μg/L. 90–99% recovery rates were obtained for analysis of pyrethroids from tea samples. Detection of SARS-CoV-2 Due to the continued spread of COVID-19, there is a growing need for effective testing of the disease-causing virus, i.e., SARS-CoV-2 [61–63]. Various analysis strategies such as the detection of nucleic acid [64], antibodies [65], and antigens of SARS-CoV-2 have been used to evaluate the infection of COVID-19 patients [66]. However, most methods are relatively expensive and complicated. The applications of these tests are still limited. Fabiani et al. applied magnetic beads to support the immunological chain for developing a paper-based immunoassay for analysis of SARS-CoV-2 antigen (spiked protein). The color change was observed on a wax-printed 96-well paper plate (Fig. 6) based on the catalysis of TMB to ox-TMB [67]. A smartphone with Spotxel® Reader can read the signal can calculate the concentration of SARS-CoV-2 antigen with a detection limit of 0.1 μg/mL. Compared to the gold standard PCR tests for analysis of the nasopharyngeal swab specimens samples, 100% agreement was found for 12 saliva samples. This approach brings new hope for the popularity of virus self-testing.Fig. 6 Evaluation of SARS-CoV-2 using smartphone with Spotxel® Reader based on an Immuno nanosensor: scheme of the analysis of SARS-CoV-2 antigen using smartphone with Spotxel® Reader. Reproduced with permission from [67] by Elsevier Prospect and conclusions The detection strategies based on smartphones combined with Spotxel® Reader show prospects for instrument-free analysis in the fields of food safety evaluation, environmental pollutant detection, microbial infections, and disease diagnosis. Further development of strategies associated with it is likely to bring new hope for mass detections by common people. To facilitate the development of portable, inexpensive, and simple detection methods, we propose the following outlook. At present, Spotxel® Reader mainly obtains the concentration of an analyte based on the analysis of color and fluorescence change of the reacted samples, and both of these signals are prone to have background interferences. The reported works have not studied in depth how to avoid the influence of background. Some simple devices with the necessary setup are expected to reduce the interferences and obtain more actuate results. Spotxel® Reader currently cannot detect some signals with low interferences, such as near-infrared fluorescence. In this case, the near-infrared fluorescence is expected to transfer to the visible optical signal that can be directly analyzed. For instance, combined with nano-upconversion technology, near-infrared light may induce the generation of visible light for the analysis of some targets, so that the tests can be performed. However, to maintain the sensitivity of a test after a series of signal transformations, the high energy-transfer efficiency of the system is expected to be fabricated. Finally, we also expect more free software to analyze different signals of the sensors in combination with smartphones, bringing more feasibility for ordinary people to detect analytes by themselves. Acknowledgements We acknowledge Jinzhou Medical University for its financial support. Funding This work is supported by University Student Innovation Project (grant number 10160101). Data Availability Statement Data sharing is not applicable to this review article as no new data were created in this study. Declarations Conflict of interest No potential conflict of interest was reported by the authors. Ningyi Qin and Zirui Liu have contributed equally to this work. ==== Refs References 1. Ye Q Shao W Meng H Performance and application evaluation of SARS-CoV-2 antigen assay J Med Virol 2022 94 3548 3553 10.1002/jmv.27798 35445404 2. Sharafi K Matin BK Omer AK Mansouri B Soleimani H Fattahi N Sharafi H Kiani A A worldwide systematic literature review for aflatoxin M1 in infant formula milk: human health risk assessment by Monte Carlo simulation Food Control 2022 10.1016/j.foodcont.2021.108681 3. Li B Zhang Y Ren X Ma H Wu D Wei Q No-wash point-of-care biosensing assay for rapid and sensitive detection of aflatoxin B1 Talanta 2021 235 122772 10.1016/j.talanta.2021.122772 34517631 4. Yuan C Adeloye D Luk TT Huang L He Y Xu Y Ye X Yi Q Song P Rudan I The global prevalence of and factors associated with Helicobacter pylori infection in children: a systematic review and meta-analysis Lancet Child Adolescent Health 2022 6 185 194 10.1016/S2352-4642(21)00400-4 35085494 5. Jean S Burnham CD Chapin K Garner OB Pant Pai N Turabelidze G Butler-Wu S At-home testing for infectious diseases: the laboratory where you Live Clin Chem 2021 68 19 26 10.1093/clinchem/hvab198 34969103 6. Yang T Luo Z Bewal T Li L Xu Y Mahdi Jafari S Lin X When smartphone enters food safety: a review in on-site analysis for foodborne pathogens using smartphone-assisted biosensors Food Chem 2022 394 133534 10.1016/j.foodchem.2022.133534 35752124 7. Hao Y Yang Z Dong W Liu Y Song S Hu Q Shuang S Dong C Gong X Intelligently design primary aromatic amines derived carbon dots for optical dual-mode and smartphone imaging detection of nitrite based on specific diazo coupling J Hazard Mater 2022 430 128393 10.1016/j.jhazmat.2022.128393 35149507 8. Placer L Lavilla I Pena-Pereira F Bendicho C Bromine speciation by a paper-based sensor integrated with a citric acid/cysteamine fluorescent probe and smartphone detection Sens Actuators B Chem 2022 10.1016/j.snb.2022.131499 9. Lu Z Chen M Li M Liu T Sun M Wu C Su G Yin J Wu M Zou P Lin L Wang X Huang Q Yin H Rao H Zhou X Ye J Wang Y Smartphone-integrated multi-color ratiometric fluorescence portable optical device based on deep learning for visual monitoring of Cu2+ and thiram Chem Eng J 2022 10.1016/j.cej.2022.135686 34539223 10. Çifteci A Çelik SE Apak R Gold-nanoparticle based turn–on fluorometric sensor for quantification of sulfhydryl and disulfide forms of biothiols: measurement of thiol/disulfide homeostasis Analyt Lett 2021 55 648 664 10.1080/00032719.2021.1958830 11. Jiang R Lin D Zhang Q Li L Yang L Multiplex chroma-response based fluorescent smartphone sensing platform for rapid and visual quantitative determination of antibiotic residues Sens Actuators B Chem 2022 350 130902 10.1016/j.snb.2021.130902 12. Hunt JP Zhao EL Free TJ Soltani M Warr CA Benedict AB Takahashi MK Griffitts JS Pitt WG Bundy BC Towards detection of SARS-CoV-2 RNA in human saliva: a paper-based cell-free toehold switch biosensor with a visual bioluminescent output N Biotechnol 2022 66 53 60 10.1016/j.nbt.2021.09.002 34555549 13. Yang L Cui M Zhang Y Jiang L Liu H Liu Z A colorimetric aptasensing assay with adjustable color mutation points for threshold-readout detection of carcinoembryonic antigen Sens Actuators B Chem 2022 350 130857 10.1016/j.snb.2021.130857 14. Hou J Jia P Yang K Bu T Zhao S Li L Wang L Fluorescence and colorimetric dual-mode ratiometric sensor based on Zr-tetraphenylporphyrin tetrasulfonic acid hydrate metal-organic frameworks for visual detection of copper ions ACS Appl Mater Interfaces 2022 14 13848 13857 10.1021/acsami.1c23199 35286802 15. Sun BR Zhou AG Li X Yu HZ Development and application of mobile apps for molecular sensing: a review ACS Sens 2021 6 1731 1744 10.1021/acssensors.1c00512 33955727 16. Franca AS Oliveira LS Smartphone-based detection devices 2021 Elsevier 249 268 17. Junaid HM Solangi AR Batool M Carbon dots as naked eye sensors Analyst 2021 146 2463 2474 10.1039/D0AN02399J 33725047 18. Sivakumar R Lee NY Recent progress in smartphone-based techniques for food safety and the detection of heavy metal ions in environmental water Chemosphere 2021 275 130096 10.1016/j.chemosphere.2021.130096 33677270 19. Chellasamy G Ankireddy SR Lee KN Govindaraju S Yun K Smartphone-integrated colorimetric sensor array-based reader system and fluorometric detection of dopamine in male and female geriatric plasma by bluish-green fluorescent carbon quantum dots Mater Today Bio 2021 12 100168 10.1016/j.mtbio.2021.100168 34877521 20. Wang Y Jian J Sun B Wei Y Pan D Cao J Shen Y Engineering of onsite point-of-care testing of Fe3+ with visual ratiometric fluorescent signals of copper nanoclusters-driven portable smartphone Sens Actuators B Chem 2022 370 132413 10.1016/j.snb.2022.132413 21. Tarasi S Ramazani A Morsali A Hu ML Highly sensitive colorimetric naked-eye detection of Hg(II) using a sacrificial metal-organic framework Inorg Chem 2021 60 13588 13595 10.1021/acs.inorgchem.1c01894 34435495 22. Yadav P Laddha H Agarwal M Gupta R Fun with smartphones: handy solution for quantification of debilitating fluoride ions in drinking water J Cheml Educ 2022 99 2677 2683 10.1021/acs.jchemed.2c00216 23. de Carvalho Oliveira G Machado CCS Inacio DK Silveira Petruci JFD Silva SG RGB color sensor for colorimetric determinations: evaluation and quantitative analysis of colored liquid samples Talanta 2022 241 123244 10.1016/j.talanta.2022.123244 35121545 24. Nguyen LH Oveissi F Chandrawati R Dehghani F Naficy S Naked-eye detection of ethylene using thiol-functionalized polydiacetylene-based flexible sensors ACS Sens 2020 5 1921 1928 10.1021/acssensors.0c00117 32551585 25. Pinyorospathum C Rattanarat P Chaiyo S Siangproh W Chailapakul O Colorimetric sensor for determination of phosphate ions using anti-aggregation of 2-mercaptoethanesulfonate-modified silver nanoplates and europium ions Sens Actuators B Chem 2019 290 226 232 10.1016/j.snb.2019.03.059 26. Fan K Zeng J Yang C Wang G Lian K Zhou X Deng Y Liu G Digital quantification method for sensitive point-of-care detection of salivary uric acid using smartphone-assisted muPADs ACS Sens 2022 7 2049 2057 10.1021/acssensors.2c00854 35820152 27. Caroleo F Magna G Damiano C Cavalleri M Gallo E Di Natale C Paolesse R Colour Catcher® sheet beyond the laundry: a low-cost support for realizing porphyrin-based mercury ion sensors Sens Actuators B Chem 2022 10.1016/j.snb.2022.131900 28. Adeniyi O Mashazi P Kirigami paper-based colorimetric immunosensor integrating smartphone readout for determination of humoral autoantibody immune response Microchem J 2022 178 107427 10.1016/j.microc.2022.107427 29. Heinonen R Mattila TJ Smartphone-based estimation of green cover depends on the camera used Agronomy J 2021 113 5597 5601 10.1002/agj2.20752 30. Nelis JLD Migliorelli D Muhlebach L Generelli S Stewart L Elliott CT Campbell K Highly sensitive electrochemical detection of the marine toxins okadaic acid and domoic acid with carbon black modified screen printed electrodes Talanta 2021 228 122215 10.1016/j.talanta.2021.122215 33773701 31. Li J Lillehoj PB Microfluidic magneto immunosensor for rapid, high sensitivity measurements of SARS-CoV-2 nucleocapsid protein in serum ACS Sens 2021 6 1270 1278 10.1021/acssensors.0c02561 33629833 32. Vasquez V Navas MC Jaimes JA Orozco J SARS-CoV-2 electrochemical immunosensor based on the spike-ACE2 complex Anal Chim Acta 2022 1205 339718 10.1016/j.aca.2022.339718 35414393 33. Zhu X Han L Liu H Sun B A smartphone-based ratiometric fluorescent sensing system for on-site detection of pyrethroids by using blue-green dual-emission carbon dots Food Chem 2022 379 132154 10.1016/j.foodchem.2022.132154 35078057 34. Zhu X Yuan X Han L Liu H Sun B A smartphone-integrated optosensing platform based on red-emission carbon dots for real-time detection of pyrethroids Biosens Bioelectron 2021 191 113460 10.1016/j.bios.2021.113460 34186303 35. Londono-Avendano MA Libreros G Osorio L Parra B A rapid RT-LAMP assay for SARS-CoV-2 with colorimetric detection assisted by a mobile application Diagnostics 2022 12 848 10.3390/diagnostics12040848 35453896 36. Zhao Y Yuan X Jiang W Liu H Sun B Chiroptical-responsive nanoprobe for the optosensing of chiral amino acids Microchim Acta 2022 189 184 10.1007/s00604-022-05282-w 37. Zhao Y Zhang Y Liu H Sun B A visual chiroptical system with chiral assembly graphene quantum dots for D-phenylalanine detection Anal Bioanal Chem 2022 414 4885 4896 10.1007/s00216-022-04113-4 35562570 38. Conceicao F Sousa DM Loessberg-Zahl J Vollertsen AR Neto E Soe K Paredes J Leferink A Lamghari M A metastasis-on-a-chip approach to explore the sympathetic modulation of breast cancer bone metastasis Mater Today Bio 2022 13 100219 10.1016/j.mtbio.2022.100219 35243294 39. Mao L Liu H Yao L Wen W Chen M-M Zhang X Wang S Construction of a dual-functional CuO/BiOCl heterojunction for high-efficiently photoelectrochemical biosensing and photoelectrocatalytic degradation of aflatoxin B1 Chem Eng J 2022 10.1016/j.cej.2021.132297 40. Li Y Sun L Yu H Zhao Q Sensitive competitive aptamer surface plasmon resonance sensor for aflatoxin b1 using streptavidin as a signal enhancer ACS Food Sci Technol 2022 2 936 940 10.1021/acsfoodscitech.2c00079 41. Li J Wang Q Xiong C Deng Q Zhang X Wang S Chen MM An ultrasensitive CH3NH3PbBr 3 quantum dots@SiO2-based electrochemiluminescence sensing platform using an organic electrolyte for aflatoxin B1 detection in corn oil Food Chem 2022 390 133200 10.1016/j.foodchem.2022.133200 35580516 42. Singh AK Sri S Garimella L Dhiman TK Sen S Solanki PR Graphene quantum dot-based optical sensing platform for aflatoxin B1 detection via the resonance energy transfer phenomenon ACS Appl Bio Mater 2022 5 1179 1186 10.1021/acsabm.1c01224 35179346 43. Mao X Yu B Li Z Li Z Shi G Comparison of lateral flow immunoassays based on oriented and nonoriented immobilization of antibodies for the detection of aflatoxin B1 Anal Chim Acta 2022 1221 340135 10.1016/j.aca.2022.340135 35934406 44. Charernchai S Chikae M Phan TT Wonsawat W Hirose D Takamura Y Automated paper-based femtogram sensing device for competitive enzyme-linked immunosorbent assay of aflatoxin B1 using submicroliter samples Anal Chem 2022 94 5099 5105 10.1021/acs.analchem.1c05401 35302345 45. Zhang C Wang Y Sun Y Duan J Wang M Ai S Hou J A novel immunocolorimetric probe for aflatoxin B1 based on multifunctional metal−organic frameworks Sens Actuators B Chem 2022 369 132362 10.1016/j.snb.2022.132362 46. Sergeyeva T Yarynka D Dubey L Dubey I Piletska E Linnik R Antonyuk M Ternovska T Brovko O Piletsky S El'skaya A Sensor based on molecularly imprinted polymer membranes and smartphone for detection of fusarium contamination in cereals Sensors 2020 20 4304 10.3390/s20154304 32752255 47. Sergeyeva T Yarynka D Piletska E Linnik R Zaporozhets O Brovko O Piletsky S El'skaya A Development of a smartphone-based biomimetic sensor for aflatoxin B1 detection using molecularly imprinted polymer membranes Talanta 2019 201 204 210 10.1016/j.talanta.2019.04.016 31122412 48. Wang W Zhang Q Ma F Li P Simultaneous determination of aflatoxins, fumonisin B1, T-2 and cyclopiazonic acid in agri-products by immunomagnetic solid-phase extraction coupled with UHPLC-MS/MS Food Chem 2022 378 132020 10.1016/j.foodchem.2021.132020 35033719 49. Feng J Xue Y Wang X Song Q Wang B Ren X Zhang L Liu Z Sensitive, simultaneous and quantitative detection of deoxynivalenol and fumonisin B1 in the water environment using lateral flow immunoassay integrated with smartphone Sci Total Environ 2022 834 155354 10.1016/j.scitotenv.2022.155354 35460773 50. Zhu H Yu Q Ouyang H Zhang R Li J Xian R Wang K Li X Cao C Antagonistic effect of selenium on fumonisin b1 promotes neutrophil extracellular traps formation in chicken neutrophils J Agric Food Chem 2022 70 5911 5920 10.1021/acs.jafc.2c01329 35535747 51. Yu Y Li Y Zhang Q Zha Y Lu S Yang Y Li P Zhou Y Colorimetric immunoassay via smartphone based on Mn2+—mediated aggregation of AuNPs for convenient detection of fumonisin B1 Food Control 2022 10.1016/j.foodcont.2021.108481 36091572 52. Shi R Chen W Pan F Zhao P He Y Yu R Fu R Munkh-Amgalan G Jiang Z Characterization of the binding behavior, structure and foaming properties of bovine α-lactalbumin combined with saponin by the multi-spectroscopic and silico approaches Food Hydrocolloids 2022 124 107259 10.1016/j.foodhyd.2021.107259 53. Zhang X Wu D Wu Y Li G Bioinspired nanozyme for portable immunoassay of allergenic proteins based on a smartphone Biosens Bioelectron 2021 172 112776 10.1016/j.bios.2020.112776 33157408 54. Andersen HR David A Freire C Fernandez MF D'Cruz SC Reina-Perez I Fini JB Blaha L Pyrethroids and developmental neurotoxicity—a critical review of epidemiological studies and supporting mechanistic evidence Environ Res 2022 214 113935 10.1016/j.envres.2022.113935 35870501 55. Singh S Mukherjee A Jaiswal DK de Araujo Pereira AP Prasad R Sharma M Kuhad RC Shukla AC Verma JP Advances and future prospects of pyrethroids: toxicity and microbial degradation Sci Total Environ 2022 829 154561 10.1016/j.scitotenv.2022.154561 35296421 56. Li M Lv M Liu T Du G Wang Q Lipid metabolic disorder induced by pyrethroids in nonalcoholic fatty liver disease of Xenopus laevis Environ Sci Technol 2022 54 8463 8474 10.1021/acs.est.2c00516 57. Xie W Zhao J Zhu X Chen S Yang X Pyrethroid bioaccumulation in wild fish linked to geographic distribution and feeding habit J Hazard Mater 2022 430 128470 10.1016/j.jhazmat.2022.128470 35180516 58. El Ayari T Mhadhbi L Trigui El Menif N El Cafsi M Acute toxicity and teratogenicity of carbaryl (carbamates), tebufenpyrad (pyrazoles), cypermethrin and permethrin (pyrethroids) on the European sea bass (Dicentrarchus labrax L, 1758) early life stages Environ Sci Pollut Res Int 2022 10.1007/s11356-11022-20421-11359 35501436 59. Wongmaneepratip W Gao X Yang H Effect of food processing on reduction and degradation pathway of pyrethroid pesticides in mackerel fillet (Scomberomorus commerson) Food Chem 2022 384 132523 10.1016/j.foodchem.2022.132523 35231708 60. Jiang W Zhao Y Zhang D Zhu X Liu H Sun B Efficient and robust dual modes of fluorescence sensing and smartphone readout for the detection of pyrethroids using artificial receptors bound inside a covalent organic framework Biosens Bioelectron 2021 194 113582 10.1016/j.bios.2021.113582 34461567 61. Zhang D Duran SSF Lim WYS Tan CKI Cheong WCD Suwardi A Loh XJ SARS-CoV-2 in wastewater: from detection to evaluation Mater Today Adv 2022 13 100211 10.1016/j.mtadv.2022.100211 35098102 62. Ravi N Cortade DL Ng E Wang SX Diagnostics for SARS-CoV-2 detection: a comprehensive review of the FDA-EUA COVID-19 testing landscape Biosens Bioelectron 2020 165 112454 10.1016/j.bios.2020.112454 32729549 63. Alygizakis N Markou AN Rousis NI Galani A Avgeris M Adamopoulos PG Scorilas A Lianidou ES Paraskevis D Tsiodras S Tsakris A Dimopoulos MA Thomaidis NS Analytical methodologies for the detection of SARS-CoV-2 in wastewater: protocols and future perspectives Trends Analyt Chem 2021 134 116125 10.1016/j.trac.2020.116125 33235400 64. Au WY Cheung PPH Diagnostic performances of common nucleic acid tests for SARS-CoV-2 in hospitals and clinics: a systematic review and meta-analysis The Lancet Microbe 2021 2 e704 e714 10.1016/S2666-5247(21)00214-7 34661181 65. Werbel WA Segev DL SARS-CoV-2 antibody testing for transplant recipients: a tool to personalize protection versus COVID-19 Am J Transplant 2022 22 1316 1320 10.1111/ajt.16993 35119179 66. Simon-Loriere E Schwartz O Towards SARS-CoV-2 serotypes? Nat Rev Microbiol 2022 20 187 188 10.1038/s41579-022-00708-x 35181769 67. Fabiani L Mazzaracchio V Moscone D Fillo S De Santis R Monte A Amatore D Lista F Arduini F Paper-based immunoassay based on 96-well wax-printed paper plate combined with magnetic beads and colorimetric smartphone-assisted measure for reliable detection of SARS-CoV-2 in saliva Biosens Bioelectron 2022 200 113909 10.1016/j.bios.2021.113909 34995838	36460855	PMC9718457	NO-CC CODE	2022-12-06 23:23:39	no	Anal Sci. 2022 Dec 2;:1-10	utf-8	Anal Sci	2,022	10.1007/s44211-022-00216-1	oa_other
==== Front Eur Radiol Eur Radiol European Radiology 0938-7994 1432-1084 Springer Berlin Heidelberg Berlin/Heidelberg 9265 10.1007/s00330-022-09265-6 Neuro Validation of a highly accelerated post-contrast wave-controlled aliasing in parallel imaging (CAIPI) 3D-T1 MPRAGE compared to standard 3D-T1 MPRAGE for detection of intracranial enhancing lesions on 3-T MRI Goncalves Filho Augusto Lio M. 12 Awan Komal Manzoor 12 Conklin John 12 Ngamsombat Chanon 23 Cauley Stephen F. 12 Setsompop Kawin 12 Liu Wei 4 Splitthoff Daniel N. 5 Lo Wei-Ching 6 Kirsch John E. 12 Schaefer Pamela W. 1 Rapalino Otto 1 Huang Susie Y. [email protected] 127 1 grid.38142.3c 000000041936754X Department of Radiology, Massachusetts General Hospital (MGH), Harvard Medical School, 55 Fruit Street, Boston, MA 02114 USA 2 grid.509504.d 0000 0004 0475 2664 Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA USA 3 grid.416009.a Department of Radiology, Faculty of Medicine, Siriraj Hospital, Mahidol University, Nakhon Pathom, Thailand 4 Siemens Shenzhen Magnetic Resonance Ltd., Shenzhen, China 5 grid.5406.7 000000012178835X Siemens Healthcare GmbH, Erlangen, Germany 6 Siemens Medical Solutions Inc., Boston, MA USA 7 grid.32224.35 0000 0004 0386 9924 Division of Neuroradiology, Department of Radiology, Massachusetts General Hospital, 55 Fruit St, GRB-273A, Boston, MA 02114 USA 2 12 2022 111 27 4 2022 26 9 2022 30 9 2022 © The Author(s), under exclusive licence to European Society of Radiology 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Objectives High-resolution post-contrast T1-weighted imaging is a workhorse sequence in the evaluation of neurological disorders. The T1-MPRAGE sequence has been widely adopted for the visualization of enhancing pathology in the brain. However, this three-dimensional (3D) acquisition is lengthy and prone to motion artifact, which often compromises diagnostic quality. The goal of this study was to compare a highly accelerated wave-controlled aliasing in parallel imaging (CAIPI) post-contrast 3D T1-MPRAGE sequence (Wave-T1-MPRAGE) with the standard 3D T1-MPRAGE sequence for visualizing enhancing lesions in brain imaging at 3 T. Methods This study included 80 patients undergoing contrast-enhanced brain MRI. The participants were scanned with a standard post-contrast T1-MPRAGE sequence (acceleration factor [R] = 2 using GRAPPA parallel imaging technique, acquisition time [TA] = 5 min 18 s) and a prototype post-contrast Wave-T1-MPRAGE sequence (R = 4, TA = 2 min 32 s). Two neuroradiologists performed a head-to-head evaluation of both sequences and rated the visualization of enhancement, sharpness, noise, motion artifacts, and overall diagnostic quality. A 15% noninferiority margin was used to test whether post-contrast Wave-T1-MPRAGE was noninferior to standard T1-MPRAGE. Inter-rater and intra-rater agreement were calculated. Quantitative assessment of CNR/SNR was performed. Results Wave-T1-MPRAGE was noninferior to standard T1-MPRAGE for delineating enhancing lesions with unanimous agreement in all cases between raters. Wave-T1-MPRAGE was noninferior in the perception of noise (p < 0.001), motion artifact (p < 0.001), and overall diagnostic quality (p < 0.001). Conclusion High-accelerated post-contrast Wave-T1-MPRAGE enabled a two-fold reduction in acquisition time compared to the standard sequence with comparable performance for visualization of enhancing pathology and equivalent perception of noise, motion artifacts and overall diagnostic quality without loss of clinically important information. Key Points • Post-contrast wave-controlled aliasing in parallel imaging (CAIPI) T1-MPRAGE accelerated the acquisition of three-dimensional (3D) high-resolution post-contrast images by more than two-fold. • Post-contrast Wave-T1-MPRAGE was noninferior to standard T1-MPRAGE with unanimous agreement between reviewers (100% in 80 cases) for the visualization of intracranial enhancing lesions. • Wave-T1-MPRAGE was equivalent to the standard sequence in the perception of noise in 94% (75 of 80) of cases and was preferred in 16% (13 of 80) of cases for decreased motion artifact. Supplementary Information The online version contains supplementary material available at 10.1007/s00330-022-09265-6. Keywords Magnetic resonance imaging Neuroimaging Three-dimensional imaging Acceleration Gadolinium http://dx.doi.org/10.13039/100000009 Foundation for the National Institutes of Health P41EB030006) Huang Susie Y. http://dx.doi.org/10.13039/501100011699 Siemens Healthineers ==== Body pmcIntroduction Contrast enhanced T1-weighted imaging of the brain is essential for comprehensive evaluation of a wide range of inflammatory, neoplastic, and neurovascular diseases of the central nervous system. Although several MRI techniques can detect disruption of the blood-brain barrier after contrast injection, the magnetization-prepared rapid gradient echo (MPRAGE) sequence has been used in clinical neuroimaging for decades [1]. Despite reports that post-contrast three-dimensional (3D) T1-weighted spin echo sequences such as SPACE offer better conspicuity of enhancing lesions [2] and leptomeningeal abnormalities [3], the incomplete suppression of slow-flowing blood may confound the detection of intracranial enhancing lesions, including the misdiagnosis of intraparenchymal veins as multiple sclerosis lesions [4]. As such, T1-MPRAGE remains the most commonly used 3D T1-weighted post-contrast sequence used in clinical neuroimaging, as evidenced by its inclusion in several major consensus recommendations for imaging of brain tumors and metastases [5, 6], multiple sclerosis [7], and epilepsy [8]. Due to its high isotropic resolution, 3D T1-MPRAGE provides exquisite 3D visualization of enhancing pathology for guiding biopsy, surgical resection, and treatment planning, as well as precise delivery of stereotactic radiation. However, the long scan time (~5–6 min) and motion sensitivity of the 3D encoding often degrades the image quality, thereby compromising the visualization of small enhancing lesions and decreasing the overall diagnostic yield of this sequence. Therefore, new approaches to accelerating the T1-MPRAGE sequence for evaluation of intracranial enhancing lesions would be highly beneficial and see significant impact across a wide range of protocols in neuroradiology. The wave-controlled aliasing in parallel imaging (Wave-CAIPI) acquisition and reconstruction technique is a fast imaging approach that provides up to an order of magnitude of acceleration compared to standard parallel imaging [9]. Wave-CAIPI efficiently encodes 3D k-space by synergistically combining CAIPI shifts along ky/kz with a corkscrew trajectory along the readout (kx), resulting in voxel aliasing along all three spatial dimensions. Wave-CAIPI takes full advantage of the 3D coil sensitivity information when using high-channel count array coils to provide high acceleration factors with negligible noise amplification across a variety of contrasts. Prior studies have demonstrated the benefits of Wave-CAIPI encoding in decreasing scan time and motion artifact while preserving diagnostic quality in specific clinical scenarios, including susceptibility-weighted imaging for the detection of hemorrhage [10], noncontrast T1-weighted MPRAGE for the assessment of cortical volumes [11], fluid attenuated inversion recovery imaging for detection of white matter lesions [12], and post-contrast T1-weighted SPACE for detection of enhancing pathology [13, 14]. Incorporating Wave-CAIPI acceleration into post-contrast 3D T1-MPRAGE imaging would increase the accessibility and yield of high-resolution post-contrast imaging, particularly in motion-prone patients, and merits systematic validation in a realistic clinical environment. The goal of this study was to compare a highly accelerated Wave-CAIPI post-contrast 3D T1-MPRAGE sequence (Wave-T1-MPRAGE) with the standard high-resolution 3D T1-MPRAGE sequence for routine clinical brain imaging with contrast at 3 T. We hypothesized that Wave-T1-MPRAGE would be noninferior to the standard sequence accelerated using GRAPPA (generalized autocalibrating partially parallel acquisition) parallel imaging in visualizing enhancing lesions and would provide equivalent diagnostic quality while reducing the acquisition time by more than half. Materials and methods Study participants This single-center retrospective study was approved by the institutional review board and was compliant with the Health Insurance Portability and Accountability Act. Given that the additional scan time from the Wave-T1-MPRAGE sequence was less than 3 min per scan, the institutional review board waived the need for signed informed consent. Instead, an information sheet presenting a detailed description of the research study was provided to the study participants, who could decline participation in the study prior to initiating their scan. Between February and May of 2020, a total of 96 consecutive patients undergoing contrast-enhanced brain MRI at the Massachusetts General Hospital were scanned as part of the study. Inclusion criteria were referral for contrast-enhanced brain MRI as an inpatient or outpatient and assent to participation in the study. Patients could not receive a contrast-enhanced brain MRI if they had compromised renal function (i.e., eGFR < 30 mL/min/1.73 m2) resulting in inability to receive contrast, claustrophobia, or other contraindications to MRI. Datasets were then reviewed for inclusion in the retrospective analysis of image quality. Datasets were excluded from our retrospective analysis if one of the sequences for evaluation (Standard vs Wave-T1-MPRAGE) was missing, if contrast extravasation occurred during the examination, or if the sequence parameter were not well matched. A flowchart of the screening process for inclusion in the comparative evaluation study is provided in Fig. 1. Fig. 1 Flowchart of the screening process for inclusion in the comparative evaluation study MRI protocol Whole-brain MRI scans were performed on a 3-T MRI system (MAGNETOM Prisma, Siemens Healthineers) using a commercially available 20- or 32-channel receiver coil array, which was chosen for best patient fit by the scanning MR technologist. In addition to the conventional structural sequences for each clinical indication (T1-weighted, T2-weighted, FLAIR, diffusion-weighted, and susceptibility-weighted images), the imaging protocol of all MRI studies included a standard post-contrast T1-MPRAGE sequence (acceleration factor [R] = 2 using GRAPPA parallel imaging, acquisition time [TA] = 5 min 18 s, 0.9 mm isotropic spatial resolution) and a prototype post-contrast Wave-CAIPI T1-MPRAGE sequence (R = 4, TA = 2 min 32 s, 1.0 mm isotropic spatial resolution) with comparable effective spatial resolution. The “standard” T1 MPRAGE sequence/protocol was implemented by the clinical MRI physics team on the 3-T scanners using the default parameters suggested by the vendor (e.g., TR = 2300 ms, TI = 900 ms, shortest possible TE) and are similar to those suggested by several consensus recommendations [6]. The Wave-CAIPI T1 MPRAGE protocol was created following the “standard” MPRAGE protocol to achieve comparable contrast and spatial resolution, with matched TR, TI, TE and flip angle values. Contrast-enhanced images were obtained after intravenous administration of standard dose of 0.2 mL/kg (0.1 mmol/kg) of gadoterate meglumine (Dotarem®, Guerbet) at a flow rate of approximately 2 mL/s. To minimize potential differences related to the order of acquisition, the order of the standard and Wave-T1-MPRAGE sequences was switched halfway through the study, as done in previous studies to mitigate against bias introduced by differences in timing of imaging after contrast administration. Post-contrast wave-CAIPI T1-MPRAGE sequence acquisition and reconstruction The standard T1-MPRAGE sequence used in our institution’s routine clinical protocol employs the default vendor reconstruction filter that introduces a small degree of spatial smoothing. The Wave-T1-MPRAGE sequence is currently only available as a works-in-progress prototype and as such does not yet have spatial filtering incorporated as an option. To overcome the constraints imposed by the vendor on our post-processing options, we performed a systematic evaluation of image sharpness using different effective resolutions (1 mm isotropic versus 0.9 mm isotropic) for both the standard and Wave T1-MPRAGE sequences and found that 1 mm isotropic voxel size for Wave T1-MPRAGE as compared to 0.9 mm isotropic voxel size for standard T1-MPRAGE (which incorporated spatial filtering by default) provided the most comparable effective spatial resolution. Therefore, a marginally larger isotropic voxel size was used in the Wave-T1-MPRAGE compared to the standard T1-MPRAGE acquisitions (1.0 mm vs 0.9 mm). The TR, TE, and flip angle values were matched between the Wave and standard T1-MPRAGE sequences as those are the main parameters that contribute to T1-weighted contrast. These strategies ensured comparable contrast and visual spatial resolution as evaluated by the study neuroradiologists. On-line reconstruction of the Wave-CAIPI 3D T1-MPRAGE was performed using an auto-calibrated procedure in which the true gradient trajectory is estimated during the reconstruction without the need for additional calibration scans [15]. This allowed for simultaneous estimation of the parallel imaging reconstruction and the true k-space trajectory, with a reconstruction time of approximately 60 s. Detailed sequence parameters are presented in Table 1. Table 1 Technical parameters of sequence acquisition Parameter Standard T1-MPRAGE Wave-T1-MPRAGE Matrix size 256 x 256 256 x 256 Isotropic voxel size (mm) 0.9 1.0 TR/TE/TI (msec) 2300/2.3/900 2300/3.2/900 Flip Angle (degree) 8 8 Bandwidth (Hz/px) 200 215 Acceleration factor (R) GRAPPA, R = 2 Wave-CAIPI, R = 4 Scan time (min:s) 5:18 2:32 CNR (mean) 9.40 6.24 * SNR in gray matter (mean) 59.10 41.33 * SNR in white matter (mean) 68.48 47.57 * TR repetition time, TE echo time, TI inversion time, R acceleration factor, CNR contrast-to-noise ratio, SNR signal-to-noise ratio, * = p < 0.001 Image evaluation Two neuroradiologists (C.N. and S.Y.H.; 14 and 11 years of experience, respectively) performed a randomized independent review of all contrast-enhanced images. All raters were blinded to sequence type, clinical indication and other imaging test results. The post-contrast images of the anonymized DICOM datasets were evaluated on an independent workstation. A predetermined 5-point grading scale was used to compare Wave-T1-MPRAGE with standard T1-MPRAGE in the evaluation of intracranial enhancement (Electronic Supplementary Table). The Wave-T1-MPRAGE and standard T1-MPRAGE images were evaluated in a head-to-head comparison for the detection of abnormal enhancement in the parenchyma, leptomeninges, pachymeninges (dura), and ependymal surface. The reviewers rated the images on an independent workstation and were allowed to choose the windowing levels for optimal visualization of contrast enhancement on both sequences. The window level (WL) and width (WW) settings may slightly differ for each sequence as they use fundamentally different acquisition and reconstruction techniques, thereby making it impossible to match the exact values of WW/WL. For full transparency, we have reported the window width and level settings in each panel of Figs. 2 and 3. Reviewers also evaluated the presence of artifacts related to motion, degree of background noise, and overall diagnostic quality of the image series. In addition, for a more objective assessment of the image quality, the reviewers compared the sharpness of both sequences. Measurements of signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were also performed in each image series as described below. The order of the cases and the left/right position of each sequence on the screen were randomized. All cases were rated for each feature with the 5-point grading scale, in which positive numbers favored the sequence on the right, and negative numbers favored the sequence on the left of the screen. A third neuroradiologist with over 20 years of experience (O.R.) adjudicated cases in which the reviewers differed by > 1 point in their ratings. We chose not to have the third radiologist rate all the cases in order to avoid complicating the statistical analysis, which would have been more complex with three raters. Fig. 2 Head-to-head comparison of post-contrast Wave-CAIPI T1-MPRAGE and standard T1-MPRAGE sequences. A Axial contrast-enhanced images of a 43-year-old man with a glioblastoma in the left frontal and temporal lobes. B Sagittal contrast-enhanced images of a 71-year-old man with brain metastases from non-small cell lung cancer. Window width (WW) and window level (WL) are reported alongside each image Fig. 3 Head-to-head comparison of post-contrast Wave-CAIPI T1-MPRAGE and standard T1-MPRAGE sequences. A Coronal contrast-enhanced images of a 74-year-old woman presenting with a dural enhancing lesion compatible with a meningioma along the right parietal convexity. B Axial contrast-enhanced images of a 52-year-old woman with non-small cell lung cancer showing tiny enhancing metastatic lesions in the left cerebellar hemisphere (arrow). Window width (WW) and window level (WL) are reported alongside each image Quantitative evaluation Quantitative evaluation was performed by measuring SNR and CNR values for postcontrast Wave T1 MPRAGE and standard T1 MPRAGE. Signal intensity measurements were performed in ROIs based in the left basal ganglia (gray matter) and in the left inferior frontal subcortical white matter. We sampled noise ROIs measuring 25 voxels in size in air-containing regions above the left aspect of the head for each participant. The standard deviation of the background noise was calculated in these ROIs on the standard T1 MPRAGE and Wave T1 MPRAGE images. The SNR in gray matter and white matter were calculated by dividing the mean signal intensity in the respective ROIs by the standard deviation of the background noise. The contrast to noise ratios were calculated by taking the difference in signal intensities for the gray matter and white matter ROIs and dividing by the standard deviation of the background noise. Statistical analysis We tested for noninferiority of Wave-T1-MPRAGE compared to standard T1-MPRAGE in the head-to-head analysis. A noninferiority margin (Δ) of 15% was chosen with the null hypothesis (H0) that the proportion of cases where standard T1-MPRAGE was preferred over Wave-T1-MPRAGE was > 15% [16]. We used the Z statistic to calculate the probability of the standard sequence being preferred over the Wave-T1-MPRAGE sequence in more than 15% of cases (H0 > Δ), with a significance level (α) of 0.05. The required sample size was estimated as described [17] for a single proportion (the proportion of subjects in which visualization of enhancing lesions in all compartments was preferred on standard over Wave-T1-MPRAGE), for a type I error rate (α) of 0.05, a power (1-β) of 0.90 and noninferiority margin of 15%, a minimum of 63 cases was required. Descriptive data were summarized by the calculation of percentile proportions, means and standard deviations. We also calculated the upper bound of the 95% confidence interval for the proportion of cases where standard T1-MPRAGE was preferred over Wave-T1-MPRAGE, i.e., the critical value, pcritical. A Welch’s t-test was used for the comparison of mean SNR/CNR. Interrater agreement was reported using the quadratically weighted Cohen κ to disproportionately penalize larger disagreements. The interrater agreement was interpreted according to Landis and Koch [18]. Intra-rater reliability was measured by intraclass correlation coefficients (ICC) estimates based on a single-measurement, absolute-agreement, two-way mixed-effects model. All statistical calculations were performed using R version 3.6.3. Results Of the 96 recruited study participants, 16 were excluded from the comparative evaluation for the following reasons: (a) contrast extravasation during intravenous injection (n = 1), (b) inadvertent mismatches between sequence parameters, including errors in the initial protocol set-up resulting in the acquisition of the Wave sequence with fat-suppression, which restricted the comparison with the standard sequence which did not have fat suppression (n = 10), and (c) errors in the execution of the imaging protocol resulting in the acquisition of only one type of post-contrast T1-weighted sequence during the scan (n = 5). Of note, none of the sequences needed to be repeated because there were two iterations of the same sequence (Wave and standard-T1-MPRAGE) performed after contrast administration. A total of 80 adult participants (mean age, 56 years ± 18 [standard deviation]; 42 men, 38 women) were included in the study. Demographic information for the 80 participants included in the study can be found in Table 2. Table 2 Demographic information of study participants Characteristic Study participants Age (mean ± SD, yr.) 56 ± 18 Sex (%) Men 42 (53) Women 38 (47) Clinical indication for MRI (%) Neoplasm 51 (64) Vascular 8 (10) Infection / inflammatory 2 (2) Other (headache, trauma, altered mental status) 19 (24) Order of sequence acquisition after contrast injection (%) Standard T1-MPRAGE first 54 (67) Wave-T1-MPRAGE first 26 (33) The most common clinical indication for MRI examination (Table 2) was the study of neoplastic disease in 64% (51 of 80) of study participants, followed by vascular diseases in 10% (8 of 80), and infection / inflammatory diseases in 2% (2 in 80) of participants. There were other indications for examination of the brain that together accounted for 19 of 80 participants (24%), including headache, trauma and altered mental status. In the head-to-head analysis, abnormal enhancement was detected in 47 of 80 cases (59%). Of the 47 cases that showed abnormal enhancement, 28 (60%) had parenchymal enhancement, 37 (79%) had dural enhancement, 8 (17%) had leptomeningeal enhancement, and 2 (4%) had ependymal enhancement. In 30 of 47 (64%) cases with enhancing lesions, the raters visualized pathological enhancement in more than one category. Fifty-four of 80 studies (67%) were performed with the standard post-contrast T1-MPRAGE sequence acquired before Wave-T1-MPRAGE, and 26 of 80 studies (33%) were performed with the Wave-T1-MPRAGE acquired before the standard T1-MPRAGE sequence. Wave-T1-MPRAGE was noninferior to standard T1-MPRAGE for delineating enhancement in all 47 cases that showed pathological enhancement. The head-to-head comparison showed that the degree of lesion enhancement of both sequences was indistinguishable in all instances, which enabled the effective characterization of multiple and heterogenous enhancing lesions (Fig. 2) in patients undergoing evaluation for gliomas and metastases. Similarly, the conspicuity of enhancement was also identical for dural lesions and in small metastatic lesions in the posterior fossa (Fig. 3). In Figs. 2 and 3, the Wave sequence was before the standard sequence after administration of contrast. For the other comparison criteria, interrater agreement ranged from moderate to substantial (κ = 0.59 for motion, p < 0.001; 0.64 for noise, p < 0.001; and 0.74 for the overall diagnostic quality, p < 0.001). Similarly, the ICC for intra-rater reliability ranged from good to excellent (for noise, ICC = 0.83 (S.Y.H.) and 0.80 (C.N.); for motion, ICC = 0.99 (S.Y.H.) and 1 (C.N.); and for diagnostic quality, ICC = 1 for both image reviewers). Wave-T1-MPRAGE was considered noninferior in the perception of noise (upper-bound 95% CI: pcrit = 9%; p < 0.001), motion artifact (upper-bound 95% CI: pcrit = 7%; p < 0.001), and overall diagnostic quality (upper-bound 95% CI: pcrit = 5%; p < 0.001). With respect to image noise, Wave and standard T1-MPRAGE were rated as equivalent in 75 of 80 (94%) cases, the standard sequence was preferred in 4 of 80 (5%) cases, and Wave-T1-MPRAGE was preferred for showing less background noise than the standard sequence in 1 of 80 (1%) cases. In the evaluation of motion, 64 of 80 (80%) cases were considered equivalent. Wave-T1-MPRAGE showed fewer motion artifacts in 13 of 80 (16%) cases, and the standard sequence showed fewer motion artifacts in 3 of 80 (4%) cases. For the overall diagnostic quality, Wave-T1-MPRAGE was preferred in 4 of 80 (5%) of cases, while the standard sequence was preferred in 2 of 80 (2%) of cases. The standard and Wave-T1-MPRAGE sequences were considered equivalent for the overall diagnostic quality in 74 of 80 (93%) of cases. Wave-T1-MPRAGE was also noninferior for the sharpness of images, and both sequences were equally sharp in 78 of 80 (97.5%) of cases. From the total 80 cases, in only one case (1.25%) Wave was considered sharper and in one case the standard sequence was sharper (1.25%). Additional details of the head-to-head comparison and noninferiority testing results are shown in Fig. 4. The quantitative assessment of SNR/CNR demonstrated that Wave-T1-MPRAGE shows significantly reduced mean CNR (p < 0.001), as well as decreased mean SNR in the gray- and white-matter (p < 0.001) compared to the average values in the standard sequence (Fig. 5). Fig. 4 Balloon plot showing the results of the head-to-head comparison of post-contrast standard T1 MPRAGE and post-contrast Wave-CAIPI T1 MPRAGE for visualization of abnormal intracranial enhancing lesions, sharpness, the perception of noise, presence of artifacts due to motion, and the overall diagnostic quality. A zero-score indicates equivalency, negative scores (left) favor standard T1-MPRAGE, and positive scores (right) favor Wave-T1-MPRAGE Fig. 5 Boxplot charts demonstrating the distribution of contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) in the gray matter and white matter in Wave-T1-MPRAGE versus Standard MPRAGE Out of the final 80 included participants, 72 individuals (90%) were scanned with a 20-channel coil and 8 individuals (10%) were scanned with a 32-channel coil. Nevertheless, the performance of Wave-T1-MPRAGE did not differ significantly when post-hoc subgroup analyses were performed. Additional details of the head-to-head comparison for each of the two coil set-ups are provided in the Electronic Supplementary Material. Discussion This study compared the diagnostic performance of an ultrafast post-contrast 3D Wave-T1-MPRAGE sequence to standard 3D T1-MPRAGE in the evaluation of abnormal intracranial enhancement among patients undergoing routine inpatient and outpatient contrast-enhanced brain MRI at a large volume tertiary-care hospital. The findings show that a Wave-T1-MPRAGE sequence acquired in less than half the time as the standard sequence (2:32 min versus 5:18 min) provides equivalent, robust visualization of all enhancing parenchymal, leptomeningeal and dural lesions with unanimous agreement between expert neuroradiologists. In addition to preserving sharpness (p < 0.001) and overall diagnostic quality (p < 0.001), Wave-T1-MPRAGE was noninferior to standard T1-MPRAGE in the degree of noise (p < 0.001) and motion artifact (p < 0.001). The use of high-resolution isotropic 3D T1 post-contrast MPRAGE has been established as a workhorse sequence in the systematic evaluation of brain tumors [6], metastatic disease [19], epilepsy [8], and multiple sclerosis [7]. The Wave-CAIPI approach enables higher accelerations than standard parallel imaging with minimal g-factor noise amplification and has been shown to yield comparable diagnostic quality to standard acquisitions, as demonstrated in clinical validation studies of the post-contrast Wave-T1-SPACE sequence for the evaluation of brain metastases [13], Wave susceptibility-weighted imaging for the detection of hemorrhage [10], and non-contrast Wave T1-weighted MPRAGE for the assessment of cortical volumes [11]. The fact that Wave-T1-MPRAGE was rated noninferior with respect to image noise in the current work was interesting given that a signal-to-noise ratio (SNR) reduction of 1/√2 (approximately 30%) is expected due to the higher acceleration factor of the Wave-CAIPI sequence. Despite the reductions in SNR and CNR for the Wave sequences, the qualitative analyses showed that there was no impact on diagnostic quality. This result suggests that the small reduction in SNR of the Wave-T1-MPRAGE sequence did not influence the diagnostic evaluation of the radiologist reviewers. The benefits of our current approach are that the R = 4 images were equivalent to standard MPRAGE for both the 20-ch and 32-ch coils, and were more robust to noise compared to prior studies that used higher acceleration factors with Wave-CAIPI [11]. At higher acceleration factors, the difference in SNR may become more perceptible, which could place a limit on the degree of acceleration that can be clinically achieved. The advantages of using a fast and diagnostically robust acquisition include faster patient turnaround times, which may increase the accessibility and throughput of MRI examinations [20]. The highly accelerated Wave-T1-MPRAGE sequence may particularly benefit acutely ill patients or patients who may be unable to remain still for prolonged periods. In addition, given that the high-resolution 3D T1 post-contrast sequence is usually the last to be performed in contrast-enhanced brain MRI protocols, it is the most susceptible to motion and stands to benefit greatly from faster acquisitions, which have been shown to decrease motion artifact and improving overall image quality [10]. The time savings from Wave-CAIPI sequences are projected to improve patient throughput by minimizing protocol length and improving the operational capacity of imaging services [21]. As a case in point, at our institution, the use of Wave-CAIPI accelerated sequences contributed to streamlining the MRI workflow, which was significantly impacted by the global surge of coronavirus disease 2019 (COVID-19). The changes and restrictions imposed by COVID-19 created a large backlog of imaging studies, which disproportionately affected oncology patients requiring more frequent surveillance examinations [22]. Our findings support the idea that Wave-T1-MPRAGE could replace standard T1-MPRAGE for the clinical evaluation of enhancing brain lesions in both inpatient and outpatient settings and improve patient access to valuable MRI resources. This study has several limitations. First, we sought to balance the order of acquisition to minimize potential differences in conspicuity of enhancement due to the time elapsed from contrast injection. In our cohort, a slight preponderance of studies were acquired with the post-contrast standard T1-MPRAGE (54 of 80, 67%) prior to Wave-T1-MPRAGE, even after inverting the order of sequences halfway through the study, with Wave-T1-MPRAGE acquired prior to the standard sequence in 26 of 80 (33%) cases. Although we did not achieve an even number of each acquisition order, the results showed that the conspicuity of contrast enhancement was rated as equivalent in the entire study sample. The strategy of inverting the order of sequences halfway through the study was mandated by the constraints of staffing and streamlined technologist workflow during the COVID-19 pandemic. In theory, our approach could be subject to bias due to changes in the system, e.g., scanner upgrades and/or degradation, that could affect one order more than the other. In practice, no scanner upgrades occurred during the four-month study period, and system wear and tear was minimal during this time. Alternating the acquisition order in a scan-by-scan fashion is a potential solution that may be implemented in future studies when feasible. Future studies could also focus on characterizing the time-to-peak enhancement of enhancing pathology occurring in different intracranial compartments and timing the acquisition of Wave and standard T1 MPRAGE sequences more precisely in order to ensure that an equal number of cases were acquired with peak enhancement for each sequence. While theoretically more rigorous, we believe this approach is more complex to implement and may not be practically achievable given the variability of enhancement kinetics for different types of intracranial enhancing lesions. Second, despite being blinded to the acquisition protocol, the reviewers may still have been able to identify images acquired with Wave- versus standard-T1-MPRAGE due to the slightly higher noise at the center of image on the Wave sequence owing to the higher degree of acceleration. An experienced neuroradiologist could become sensitized to the slightly higher noise level after reviewing multiple images, thereby introducing potential bias into the image evaluation. We sought to minimize this possibility by pairing closely the most important parameters that determine image quality and image contrast between acquisitions, including TR, TE, flip angle, and spatial resolution. Lastly, the selection of a proper noninferiority margin for assessing the similarity of diagnostic imaging studies is often challenging. Our choice was guided by a review of similar imaging-based noninferiority publications and consensus among the group of neuroradiologists that the new sequence could be considered noninferior if the standard sequence was preferred in fewer than 15% of cases. Since this threshold is subjective, we also reported the critical value (Pcritical), equivalent to the upper bound of a 95% confidence interval, for the proportion of cases in which the standard sequence was preferred. In conclusion, contrast-enhanced Wave-CAIPI 3D T1-MPRAGE was noninferior to the standard 3D T1-MPRAGE sequence in the visualization of enhancing lesions and overall diagnostic quality, with an equivalent degree of background noise and susceptibility to motion artifacts. The clinical application of the Wave-T1-MPRAGE enabled a two-fold reduction in the acquisition time of post-contrast images, leading to more efficient utilization of MR resources without loss of clinically important information. At our institution, in accelerating contrast-enhanced brain MRI protocols across a variety of indications, the ultrafast post-contrast Wave-T1-MPRAGE sequence has played an outsize role in improving the diagnostic evaluation and imaging workflow in an inpatient and outpatient setting, with potential for further downstream operational impact and efficiency. Supplementary information ESM 1 (PDF 209 kb) Acknowledgements This work was supported by the National Institutes of Health (grant no. P41EB030006) and a research grant from Siemens Healthineers. This work was conducted with support from Harvard Catalyst \| The Harvard Clinical and Translational Science Center (National Center for Advancing Translational Sciences, National Institutes of Health Award UL 1TR002541) and financial contributions from Harvard University and its affiliated academic healthcare centers. The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic healthcare centers, or the National Institutes of Health. Funding This study has received funding from National Institutes of Health (grant no. P41EB030006) and a research grant from Siemens Healthineers. Declarations Guarantor The scientific guarantor of this publication is Susie Y. Huang. Conflict of interest The authors of this manuscript declare relationships with the following companies: Siemens Healthineers. Statistics and biometry One of the authors has significant statistical expertise and received advice by the Harvard Catalyst Biostatistical Consulting Program. Informed consent Written informed consent was waived by the Institutional Review Board. Ethical approval Institutional Review Board approval was obtained. Methodology • retrospective • comparative study • performed at one institution Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Augusto Lio M. Goncalves Filho and Komal Manzoor Awan contributed equally to this work. ==== Refs References 1. Mugler JP Brookeman JR Three-dimensional magnetization-prepared rapid gradient-echo imaging (3D MP RAGE) Magn Reson Med 1990 15 152 157 10.1002/mrm.1910150117 2374495 2. Kato Y Higano S Tamura H Usefulness of contrast-enhanced T1-weighted sampling perfection with application-optimized contrasts by using different flip angle evolutions in detection of small brain metastasis at 3 T MR imaging: comparison with magnetization-prepared rapid acquisitio AJNR Am J Neuroradiol 2009 30 923 929 10.3174/ajnr.A1506 19213825 3. Jeevanandham B Kalyanpur T Gupta P Cherian M Comparison of post-contrast 3D-T(1)-MPRAGE, 3D-T(1)-SPACE and 3D-T(2)-FLAIR MR images in evaluation of meningeal abnormalities at 3-T MRI Br J Radiol 2017 90 20160834 10.1259/bjr.20160834 28375660 4. Danieli L Roccatagliata L Distefano D Nonlesional sources of contrast enhancement on postgadolinium “Black-Blood” 3D T1-SPACE images in patients with multiple sclerosis AJNR Am J Neuroradiol 2022 43 872 880 10.3174/ajnr.A7529 35618421 5. Kaufmann TJ Smits M Boxerman J Consensus recommendations for a standardized brain tumor imaging protocol for clinical trials in brain metastases Neuro Oncol 2020 22 757 772 10.1093/neuonc/noaa030 32048719 6. Ellingson BM Bendszus M Boxerman J Consensus recommendations for a standardized brain tumor imaging protocol in clinical trials Neuro Oncol 2015 17 1188 1198 10.1093/neuonc/nov095 26250565 7. Traboulsee A Simon JH Stone L Revised recommendations of the consortium of MS centers task force for a standardized MRI protocol and clinical guidelines for the diagnosis and follow-up of multiple sclerosis AJNR Am J Neuroradiol 2016 37 394 401 10.3174/ajnr.A4539 26564433 8. Bernasconi A Cendes F Theodore WH Recommendations for the use of structural magnetic resonance imaging in the care of patients with epilepsy: a consensus report from the International League Against Epilepsy Neuroimaging Task Force Epilepsia 2019 60 1054 1068 10.1111/epi.15612 31135062 9. Bilgic B Gagoski BA Cauley SF Wave-CAIPI for highly accelerated 3D imaging Magn Reson Med 2015 73 2152 2162 10.1002/mrm.25347 24986223 10. Conklin J Longo MGF Cauley SF Validation of highly accelerated WAVE-CAIPI SWI compared with conventional SWI and T2*-weighted gradient recalled-echo for routine clinical brain MRI at 3 T AJNR Am J Neuroradiol 2019 40 2073 2080 10.3174/ajnr.A6295 31727749 11. Longo MGF Conklin J Cauley SF Evaluation of ultrafast wave-CAIPI MPRAGE for visual grading and automated measurement of brain tissue volume AJNR Am J Neuroradiol 2020 41 1388 1396 10.3174/ajnr.A6703 32732274 12. Ngamsombat C, Filho ALMG, Longo MGF et al (2021) Evaluation of ultrafast Wave-CAIPI 3D FLAIR in the visualization and volumetric estimation of cerebral white matter lesions. AJNR Am J Neuroradiol 42:1584-1590. 10.3174/ajnr.A7191 13. Goncalves Filho ALM, Conklin J, Longo MGF et al (2020) Accelerated post-contrast wave-CAIPI T1 SPACE achieves equivalent diagnostic performance compared with standard T1 SPACE for the detection of brain metastases in clinical 3-T MRI. Front Neurol 11:587327. 10.3389/fneur.2020.587327 14. Filho ALMG, LongoMGF, Conklin J et al (2021) MRI Highly Accelerated Wave-CAIPI T1-SPACE versus Standard T1-SPACE to detect brain gadolinium-enhancing lesions at 3 T. J Neuroimaging 31:893-901. 10.1111/jon.12893 15. Cauley SF Setsompop K Bilgic B Autocalibrated wave-CAIPI reconstruction; Joint optimization of k-space trajectory and parallel imaging reconstruction Magn Reson Med 2017 78 1093 1099 10.1002/mrm.26499 27770457 16. Ahn S Park SH Lee KH How to demonstrate similarity by using noninferiority and equivalence statistical testing in radiology research Radiology 2013 267 328 338 10.1148/radiol.12120725 23610094 17. Chow SC Wang H Shao J Sample size calculations in clinical research 2003 New York CRC Press 18. Landis JR Koch GG The measurement of observer agreement for categorical data Biometrics 1977 33 159 10.2307/2529310 843571 19. Lin NU Lee EQ Aoyama H Response assessment criteria for brain metastases: proposal from the RANO group Lancet Oncol 2015 16 e270 e278 10.1016/S1470-2045(15)70057-4 26065612 20. Lang M, Cartmell S, Tabari A et al (2021) Evaluation of the Aggregated Time Savings in Adopting Fast Brain MRI Techniques for Outpatient Brain MRI. Acad Radiol S1076-6332(21)00319-6 21. Boland GW Duszak R Modality access: strategies for optimizing throughput J Am Coll Radiol 2015 12 1073 1075 10.1016/j.jacr.2015.06.012 26435120 22. Luker GD Boettcher AN Impact of COVID-19 on clinical care and research in cancer imaging: where we are now Radiol Imaging Cancer 2021 3 e210003 10.1148/rycan.2021210003 33778764	36460923	PMC9718459	NO-CC CODE	2022-12-06 23:23:39	no	Eur Radiol. 2022 Dec 2;:1-11	utf-8	Eur Radiol	2,022	10.1007/s00330-022-09265-6	oa_other
==== Front J Behav Health Serv Res J Behav Health Serv Res The Journal of Behavioral Health Services & Research 1094-3412 1556-3308 Springer US New York 9827 10.1007/s11414-022-09827-y Article A Qualitative Evaluation of an Adapted Assertive Community Treatment Program: Perspectives During COVID-19 http://orcid.org/0000-0002-3007-2553 Tran Jennifer T. PhD [email protected] 1 Kosyluk Kristin A. PhD 1 Dion Charles MA 1 Torres Katie BA 2 Jeffries Victoria MA, MSPH 3 1 grid.170693.a 0000 0001 2353 285X Department of Mental Health Law & Policy, University of South Florida, 13301 Bruce B Downs Blvd, Tampa, FL 33612 USA 2 grid.170693.a 0000 0001 2353 285X University of South Florida, 13301 Bruce B Downs Blvd, Tampa, FL 33612 USA 3 grid.10698.36 0000000122483208 Institute of Best Practices, UNC Center for Excellence in Community Mental Health, 101 Manning Drive First Floor, Chapel Hill, NC 27514 USA 2 12 2022 114 15 11 2022 © National Council for Mental Wellbeing 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The Pinellas County Empowerment Team (PCET) was an adapted assertive community treatment (ACT) program created to meet the needs of Pinellas County residents with serious behavioral health concerns and high frequency of hospitalization (medical and psychiatric) and incarceration. Recent research demonstrates that individuals participating in ACT programs can transition to lower-intensity services. To understand the needs and barriers in transitioning PCET clients to lower-intensity services and the unique experiences during the coronavirus (COVID-19) pandemic, the researchers conducted a qualitative evaluation which includes a case record review and in-depth interviews with clients of PCET and staff members. Our findings indicated several barriers to transitioning PCET clients, including a lack of sufficient behavioral health support outside the ACT program and some clients’ concerns regarding their abilities once out of the program. Keywords Assertive community treatment COVID-19 Serious mental illness Discharge Lower -intensity services Pinellas County ==== Body pmcBackground Assertive Community Treatment (ACT) Assertive community treatment (ACT),1 an evidence-based practice (EBP) in psychiatric rehabilitation, is an intensive form of community-based treatment designed for individuals diagnosed with serious and persistent mental illness (SPMI) who have a recent history of psychiatric hospitalizations or have criminal justice involvement or homelessness, and are at risk for rehospitalization.2–5 Limited research suggests that some clients experience clinical decline after ACT discharge, so ACT has often been understood to entail time-unlimited support.6 Indefinite participation in such intensive rehabilitation services may not be consistent with the principles of recovery.7,8 ACT clinicians and researchers have recognized that ACT is a resource-intensive (expensive) rehabilitation service. Due to this resource intensiveness and the time-unlimited nature of ACT services, ACT teams’ capacity is finite without transitioning participants to lower-intensity services. Lower-intensity services are defined as less intense and frequent services that may be considered traditional case management or usual care.1 ACT teams cannot accommodate all those who might benefit from ACT support. Therefore, assessing the transition readiness of current ACT clients to accommodate those with intensive needs who are currently unserved by ACT is an essential topic for exploration.7,9 It is also critical that we understand how to accomplish such transitions without leading to deterioration in functioning or undesirable outcomes for clients transitioning to lower-intensity services. A qualitative study with 15 ACT clinicians described their support for the goal of discharging clients but expressed concern over clients’ future stability and significant barriers to erupt.9 ACT has been linked to several important outcomes, including shorter hospital stays, improved quality of life, treatment adherence, and patient satisfaction.10,11 The structure of ACT is an essential component of its success which includes a ratio of 1:10 (providers: clients) and a team composed of a psychiatrist, a nurse, a substance abuse specialist, and a vocational specialist.12 The Pinellas County Empowerment Team (PCET) as an ACT-like program also included a ratio of 1:10 (providers: clients) and consisted of a team that is composed of a program director, psychiatrist, psychiatric nurse, case managers, trauma-informed counselor, an officer from the County Sheriff’s department, and support from community housing services. The importance of the structure of the PCET team is in line with prior research indicating that the most critical predictor of success and failure in ACT for clients are the aspects of the ACT team itself (e.g., providing support, consistent contact, advocacy, and the team approach to treatment).13 Pinellas County Empowerment Team (PCET) In 2016, Pinellas County, located on Florida’s Gulf Coast, developed a modified version of ACT (this version included permanent supportive housing, whereas this is not a component of the ACT EBP) to address the needs of identified residents with serious behavioral health concerns who were frequently hospitalized or incarcerated. These individuals were considered “high utilizers” with higher-level needs from various key issues across behavioral health, homelessness, criminal justice, health care, and insufficient funding dedicated to these service needs. PCET was a program within a community non-for-profit in Southeast Florida that provided a wide array of services including sexual assault services, peer support, children’s advocacy, center/child protection team services, pediatric psychiatric care, short-term case management, parenting support classes, parenting/family support counseling, school-based services, forensic case management, competency restoration, and intensive in-home therapy and case management. As a modified ACT team, PCET implemented a similar team approach as ACT which included a holistic team to address clients’ needs. PCET integrated key elements of Permanent Supportive Housing, an evidence-based practice designed to address the needs of homeless persons dealing with mental health and substance use disorders.14 The PCET team also provided law enforcement support, case management, therapy, psychiatric care, and medication management through its comprehensive team of a psychiatrist, case manager, psychiatric nurse, counselors, housing representative, and law enforcement representative. Distinguishing characteristics of the ACT model implemented by PCET included providing comprehensive services at the person’s residence, case review meetings with all team members twice per week, caseloads that do not exceed ten clients, and shared responsibility among team members in delivering specific tasks. A prior evaluation of the PCET program provided an understanding of the experiences of individuals considered “high utilizers” participating in the program (qualitative arm) as well as a cost analysis of the program (quantitative arm). The findings from the 2-year evaluation demonstrated that the PCET program addressed the needs of participants considered high utilizers, including healthcare, medication management, housing, trauma-informed counseling services, addressing substance use, and providing participants with social and vocational opportunities. Regarding cost analysis, findings from the evaluation indicated that PCET dramatically reduced the cost of cycling through acute crisis stabilization facilities and the jail system. Due to the high costs associated with intensive services, despite documented cost savings to the system, as well as the desire to allow PCET participants to utilize lower levels of support with greater independence as they are able, the current evaluation focused on understanding the unique barriers and needs for PCET clients to transition to lower-intensity services. COVID-19 Effects on Mental Health Services The coronavirus disease first infected humans in 2019 (COVID-19). It was declared a pandemic on March 11, 2020.14 This led to lockdown measures and abrupt changes in daily routine as COVID-19 spread across the USA. Due to the timing of the evaluation (from February 15 to December 15, 2020) with the COVID-19 pandemic, the evaluation of PCET also examined the experiences of individuals participating in the program and staff members during COVID-19. The uncertainty and changes during the pandemic resulted in concerns about mental health and the continuation of services for individuals with SPMI. While the Centers for Medicare & Medicaid Services (CMS) encouraged telehealth through common communication applications such as Zoom, there was little guidance or evidence-based recommendations for implementation. ACT teams, like PCET, found themselves moving forward and providing services to clients however they could manage during the pandemic.15 The Current Evaluation The evaluation aimed to understand stakeholder perspectives on creating transition plans for those ready to move out of PCET ACT-like services to lower-intensity services (defined as usual care or case management that is less frequent). Specific research questions to be answered via qualitative methods included:What were the perceived needs of stakeholders (clients in the PCET services and providers of the service) to successfully step those engaged with PCET down to lower intensity supports? What were the perceived barriers of stakeholders to successfully step those engaged with PCET down to lower intensity supports? Due to the unique timing of the evaluation of PCET beginning and lasting throughout the COVID-19 pandemic, the researchers also asked: How have services been affected during the pandemic? Through engagement with PCET, the authors of this report learned that the preferred language surrounding stepping down to lower intensity supports is “stepping up” or “graduation,” as these terms accurately capture the fact that moving to lower-intensity supports is a sign of progress. Therefore, the researchers will use “graduation to lower intensity services” or simply “graduation” to refer to clients transitioning to lower-intensity supports. Methods The PCET qualitative evaluation included three components: (a) a case record review, (b) in-depth interviews with individuals participating in PCET, and (c) in-depth interviews with PCET staff. This evaluation was submitted to the University’s Institutional Review Board and was deemed not research involving human subjects and was considered an evaluation; therefore, IRB approval and review were not required. The evaluation began on February 1, 2020, and was completed by December 15, 2020. The evaluation of PCET was conducted by a team of three researchers associated with a local University contracted by Pinellas County. The research team has no relationship with the PCET program other than as the program’s evaluators. Case Reviews The evaluation and review of case records included a review of several documents, including biopsychosocial assessments, treatment plans, public criminal records, and case progress notes for 14 participants (all current PCET participants during the time of the evaluation in 2020). Case reviews were conducted to gather demographic and background information of all current PCET clients and to provide context for client interviews. To review case records (biopsychosocials and treatment plans) during the COVID-19 quarantining protocol, a member of the PCET team shared the screen of their record management software via a HIPAA-compliant video chat with two research team members simultaneously. Over the course of 5 sessions (lasting 2–3 hours each), two researchers independently extracted data to include family history, reasons for mental health services, family substance use, education, friends/social interests, spiritual involvement, previous treatment, trauma experiences, past suicidal ideation, and past homicidal ideation. The two reviewers agreed on the data extracted in the case review by reviewing the data together after each session. Any conflicts were resolved among the two reviewers in order to reach consensus on the data extracted. A review of criminal records of PCET clients was conducted using the public search within the County Clerk of Courts and Comptroller website by two researchers, and consensus was reached via the same process used for biopsychosocial and treatment plan data. The entire case review was conducted from September 1, 2020, to December 15, 2020. Client Interviews The initial goals of interviews with clients were to understand the client’s perceived challenges and support to transition to lower-intensity services. The initial interview guide asked clients about their views on transitioning from PCET services to lower-intensity services. However, feedback from PCET staff members indicated concerns with the line of questions for clients. Concerns around anticipation that questions about transition may cause anxieties in the PCET participants, which Bromley and colleagues have also voiced.18 Therefore, PCET clients were asked about their experiences with the program and their perceived progress and needs in meeting their goals. Clients were asked three main questions: (a) How would they describe their experiences with the PCET thus far? (b) What do they see as their next steps? and (c) How have things been since the changes with COVID-19? While we could not ask clients directly about their needs and barriers to graduating from PCET, our interview protocol aligns with our research questions. Asking clients about challenges may align with barriers to transitioning to lower-intensity services. For example, if participants report challenges with housing, this may indicate a need for housing supports to be considered when graduating clients from ACT. Positive experiences expressed by PCET clients might suggest what is needed for movement toward graduation. For example, if having someone to check in on them daily was a positive experience, perhaps exploring proxies for this, like peer support groups or technology, might make sense. Two researchers conducted a total of seven in-depth interviews with clients between July 16, 2020, and July 22, 2020. The evaluation team received the name of seven clients that were eligible for interviews from PCET staff. PCET staff excluded clients currently incarcerated and those who staff determined would not be able to complete the interview due to communication barriers. Staff Interviews Qualitative in-depth interviews were conducted to solicit feedback and understand the perceived barriers and perceived the needs from PCET staff to graduate clients of PCET to lower intensity services successfully. During the interviews, PCET staff were asked questions to explore thoughts on their experiences with the PCET, effectiveness of the PCET, supports that the clients of the PCET may need to graduate, challenges for clients in graduating, and the impact of COVID-19 on clients receiving PCET services. A total of 10 in-depth interviews with individuals associated with PCET were conducted by the same two researchers who conducted client interviews. These interviews were conducted between June 24, 2020, and July 15, 2020. All PCET staff were invited to participate in an interview with the evaluation team. Participants that accepted an invitation included PCET staff and management (case managers, social workers, counselors, nurses, and program managers) and individuals working for housing centers within the County. The research team followed up with invited PCET team members up to four times across 2 months. PCET staff not interviewed included the psychiatrist, law enforcement personnel, and county workers due to a lack of response after reaching out four times. Interview Qualitative Analysis All client and staff interviews were recorded and transcribed verbatim through rev.com, an online transcription service. Two evaluation team members analyzed the transcribed interviews utilizing an applied thematic analysis approach with a codebook consisting of a priori codes (based on the research questions) and emergent codes.16 Both evaluation team members (the same researchers conducting the interviews) analyzed a single transcript using a codebook for guidance and reached inter-rater reliability of 82%. All interviews were then systematically analyzed by the same two researchers using the codebook to apply codes to each transcript. All codes were examined for relationships and organized into several themes. Results Client Demographics All information extracted from case notes, treatment plans, biopsychosocial, and criminal records is summarized in Table 1. The review of the case notes, treatment plans, and biopsychosocial provided further information regarding the histories and current treatment of PCET clients. Table 2 provides an extensive summary of the data extracted for review about PCET clients.Table 1 Summary of client demographic information N 14 PCET participants Age range 29–62 years of age Sex Male (93.3%)Female (6.7%) Sexual orientation Heterosexual (93.3%) Bisexual (6.7%) Race/ethnicity Caucasian/White (55.5%) African American/Black (45.5%) Diagnoses Schizoaffective disorder, bipolar type Schizophrenia Bipolar disorder Substance use disorder Criminal record Does not have a criminal record (15.4%) Has a criminal record (84.6%) • Felonies (battery, petit theft, possession of an illicit substance, aggravated assault) •Misdemeanors (disorderly conduct and intoxication, loitering, theft, trespassing, resisting arrest, criminal mischief, battery) • Municipal (possession of illicit substance, open container, panhandling, violation of park hours, sleeping/camping in public) Education Completed high school/GED (35.7%) Some high school (28.6%) Some college (21.4%) College degree (7.1%) Religion/spiritual involvement Identified as Christian (50%) Celebrates religious holidays (42.9%) Identified as Buddhist (7.1%) Table 2 Summary of case notes, treatment plans, and biopsychosocial Domains Codes Reason for mental health services and connection to PCET High utilization of county services Crisis unit services Psychiatric hospitalizations Jail Extensive history of homelessness Family history of mental health Immediate family member diagnoses: Bipolar disorder, schizophrenia, manic depressive disorder, and undiagnosed mental health concerns No family history of mental health concerns Trauma history History of trauma (57.1%) Denies history of trauma (42.9%) Prior suicidal ideation Yes, indicated (57.1%) No, denies past suicidal ideation (42.9%) Strengths Family involvement Spiritual support Interests/hobbies outside of home Expresses thoughts about solutions Communication skills Creative/imaginative Community support Physical health Work Empathetic School/education Hopeful about current treatment Has benefitted from past treatment Interventions Case management Therapeutic counseling Psychiatric meetings Medication management Therapeutic approach Behavior modification Motivational interviewing Psychoeducation Cognitive behavioral therapy Solution-focused therapy Case management activities Housing (connection, financing hotels/motels) Cell phone (and minutes/bill) Groceries Miscellaneous items (hygiene products, clothing items) Purchased meals Transportation (Uber, bus pass) Laundry Scheduling appointments Client Interviews The researchers identified several themes from the interviews: (a) description of struggles, (b) reported goals, (c) views on PCET, and (d) experiences during COVID-19. Client Description of Personal Struggles During the interviews, clients described a variety of personal hardships and struggles with their behavioral health. Clients reported struggling with cognitive issues such as memory, financial strains, homelessness, and the death of loved ones. Clients’ common obstacle was financial issues due to lack of employment and skills (i.e., cognitive capabilities and financial management skills). Clients reported that the financial strain led to housing, personal care, and health care issues. Previous and current personal life difficulties led clients to experience more severe behavioral health issues (i.e., substance use and mental health concerns). Client-Reported Goals Most of the clients had goals about improving their lives regarding health, finances, or independence. Many clients aimed to seek out skills for employment to improve their financial situations. They also spoke about getting their apartment to gain independence. In addition, many clients stated abstinence and staying out of jail were common goals, hoping that reaching them would improve their well-being and allow for reconnection with family. Client Views on the PCET Overall, many clients felt that PCET and its team members provided great support that has improved their lives, including assistance in finances, finding stable housing, initiation and management of medications, and providing transportation. Some PCET team members included caseworkers, nurses, and counselors, who were described as great workers who genuinely cared for their clients. One client reported that,They've [PCET] been very positive for me, very good for me, and ...I mean, they've been nothing but helpful, and I'm very grateful for what they do for me. However, some clients had issues with turnover within PCET, the development of an interpersonal relationship with different PCET members, and some issues regarding being told what to do by a team member. Generally, clients established through these interviews that PCET positively affected their lives. One client even stated, “the program saved my life.” Client Experiences During COVID-19 Due to COVID-19, clients experienced multiple adjustments to their daily life and their treatment process with PCET. Social distancing and quarantining led to major changes in how PCET team members and clients interacted. PCET team members communicated differently and provided extra support to their clients. In addition, some clients reported higher anxiety, boredom, and stress levels. Due to the adjustments that needed to be made for COVID-19. One client said that,Since the quarantine started, they’ve been helping me get rides to get my groceries because I don’t want to ride the public buses at the moment. And they’ve been really good about getting me my meds and on-the-phone support. They could come to the house, and they could stay outside. Because of the rules, they have to put a mask on, come over, and talk for a minute. And I can talk to them on the phone every day. These changes during the COVID-19 pandemic created another barrier PCET clients, and team members needed to navigate through together. Staff Interviews The researchers identified several themes from the staff interviews: (a) adjustments and challenges during COVID-19, (b) successful program strategies that the PCET utilizes, (c) recommendations for improvement of the PCET, and (d) recommendations for clients transitioning to lower intensity services (graduation). Adjustments and Challenges During COVID-19 Due to the timing of the interviews, staff were asked to describe how COVID-19 impacted individuals participating in PCET. All the staff described struggles with shifting services for clients from in-person to virtual due to restrictions of distancing and quarantine during the pandemic. Staff described struggles due to these adjustments, such as technological capabilities and connection, communication, increased mental health symptoms and substance use, housing stability, and impacts on treatment. While there were a variety of struggles, staff indicated that clients showed remarkable resilience. As one staff member put it,This has been a problem for everyone, universally. But just that not having the connection to human beings. And in addition to that, not being able to get out with our staff the way they used to and have that opportunity to do things outside of the home has really been hard for some people. And I think we’ve seen an increase in symptoms, with anxiety, with depression, maybe with some hallucinations or delusional thinking kind of all of the above. We’ve seen more so than the researchers would have in the past and an increase in substance use because they’re bored, and they’re sitting at home all day long. Successful Program Strategies All staff interviewed indicated that PCET was an effective program that has decreased the amount of jail time, number of hospitalizations, and use of other county-funded services. Several factors emerged as contributing, in the opinion of the PCET team members, to the effectiveness of the PCET, including (a) the composition of the team as holistic, (b) regular contact with clients, (c) providing a connection between the client and the community, (d) the culture of the PCET as a family, and (e) funding and the ability to provide supports for the clients. Perceived Challenges and Recommendations for Program Improvements Staff reported several challenges with supporting clients engaging with PCET, including the lack of housing available within the community, the requirements of abstinence to engage in community substance abuse/use treatment for clients, limits on the number of times clients are involved in PCET services, and funding for other supportive service programs. One staff member indicated,Many of the challenges the clients face are environmental, but many of our folks thrive when the researchers take them away from their old stomping grounds. Due to these perceived challenges, several recommendations were made regarding PCET, such as (a) having an initial psychological assessment for clients, (b) a continued connection to a housing specialist, (c) peer support, (d) substance abuse/use specialist and connection to community support programs, and (e) more capacity to do night and weekend service delivery. Recommendations for Client Graduation Staff described a variety of community barriers and needs to aid PCET clients to graduate (transition into lower-intensity services). Staff reported several barriers to transitioning clients to lower-intensity services, including the lack of available community housing, sufficient funding to support clients, concerns surrounding continuing medication and medical care, the client’s struggles with substance use, and the client’s abilities for independent living. Several staff members reported concerns about the ability of PCET clients to transition to lower-intensity services due to perceived individual clients’ needs and abilities for independence. One staff member reported that,For as long as they're living, they probably will need intense services. Some of them, I do feel that way about ... I can picture all of them sober and living like great lives. But in reality, I know that some of our clients wouldn't be able to transition towards lower intensity services, or it won't work out well for them if that happens. When asked to provide guidance on the process of aiding clients’ transition to lower intensity services, staff indicated several recommendations: (a) create or infuse services in between intensive services and outpatient, (b) the importance of establishing community engagement with the client, (c) stable housing, (d) sobriety supports, (e) linking individuals to transportation, and (f) providing a psychiatric management plan including medication and behavioral treatment. Several staff members indicated the importance of individualizing the transition to lower-intensity services for each client. Staff reported that clients vary in needs and abilities; therefore, transition services should be specific to the client. One staff member said,It’s important to make sure that they’re connected community-wise. They have something that they look forward to at least on a weekly or monthly basis, whether that’s some type of, depending on what type of community they live in some type of activity they do monthly, or like some people go to church...some type of meaningful continued activity that they can look forward to and they feel engaged in. Several staff indicated that currently, there is a lack of community services that would be able to provide sufficient support to transition clients into lower-intensity services. Staff reported that clients would continue to need intensive case management and a program such as assertive community treatment (ACT) in transition. Conclusion COVID-19 has altered community mental health treatment programs and has highlighted the barriers and needs of providing care for individuals living with SPMI participating in ACT and ACT-like programs. The researchers highlighted staff perspectives on the continued needs and barriers of clients to transition to lower-intensity services, including the lack of systemic support in mental health care outside of PCET and the concerns of stakeholders (i.e., PCET team, clients, county officials, community members) regarding clients’ abilities and preferences to transition out of ACT. A follow-up evaluation would guide any long-term effectiveness and response to transitioning clients to lower-intensity services and whether operational changes during COVID-19 were beneficial. Limitations and Future Implications These conclusions are provisional and, therefore, not generalizable because of the limitations in the data collected. Due to concerns from the PCET staff, our evaluation did not directly ask clients about their perceived needs and barriers to graduating to lower-intensity services. Furthermore, we only interviewed seven of the fourteen clients that PCET staff deemed eligible to participate based on current circumstances and cognitive abilities. PCET staff believed that asking clients questions about discharge would raise concerns and produce anxiety for clients. Therefore, our findings on the barriers and needs of graduating clients are from the perspective of staff members alone with interpretations from client perspectives. It is worth noting that Florida does not contain many ACT services, with Pinellas County only containing one ACT-like program. Given the limited exposure to ACT-like services and the recent development of the PCET program at the time of the evaluation, the concerns noted from this evaluation may change as more successful discharges are observed or with more experience by PCET staff. Therefore, more research should be done and is needed to examine clients’ perspectives on transitioning to lower-intensity services, providing implications for the development of support. Prior research has reported that ACT clinicians recommend building skills for transition, engaging natural supports, celebrating success, fostering coordination with new providers, and integrating and structuring transition in ACT routines to support graduation.2 Another recommendation is incorporating the ACT discharge readiness tool7 and the TRS17 to inform their decisions about discharge readiness or as a planning tool about whether the client may need it or is doing well already. However, it is important to note that some ACT participants may not be able to graduate to lower-intensity services feasibly. Due to the unique circumstances surrounding COVID-19 and the adjustments made to interactions and care between clients and staff members, future research should include follow-up interviews to understand client and staff experiences during COVID-19, lessons learned, and whether changes in treatment delivery methods should continue post-COVID-19. Implications for Behavioral Health To our knowledge, this is the first evaluation of an ACT-like program to assess the barriers and needs of transitioning clients to lower-intensity services with perspectives from both staff and clients. While Bromley and colleagues18 focused on barriers to transitioning clients to lower-intensity services, they did not focus on the needs of transitions. The evaluation of the PCET program was in line with prior assessments of assertive community treatment (ACT) as an integral form of treatment and support for individuals with serious and persistent mental illness.5 ACT effectively provides care for individuals with SPMI and has shown cost-effectiveness for the communities and agencies involved. In understanding the implementation of ACT in communities, Mancini and colleagues4 identified barriers and needs to successful community treatment, such as the importance of leadership, limiting bureaucracy, allocation of sufficient resources, and fidelity in successfully implementing ACT within community mental health settings. This is in line with findings from our study looking at the barriers and needs from a client and staff perspective of PCET. Reports from PCET staff members agree with the past year’s assessment data. They suggest that the PCET continued to effectively provide care for individuals with SPMI and reduce costs to the County and system. Both clients and staff members reported that the structure of PCET itself as a multidisciplinary team with low staff to client ratios was vital to clients' success. Our results showed that the perceived needs of stakeholders to graduate clients to lower-intensity services successfully require a solid connection to the community (i.e., community church, support from community organizations, family, or friends, and regular activities supported by the community). This is in line with prior research of staff interviews where staff identified a lack of social relationships (e.g., socially connected, family reconciliation, isolation) as contributing most to failure.13 In a recent report, Salzer and Baron19 report that community inclusion in aiding individuals with disabilities (such as severe psychiatric disabilities) is fundamental in improving lives. Furthermore, staff members reported a need for community services outside the PCET to meet client’s needs as they transition. Staff reported concerns about the lack of community support for PCET clients following graduation (e.g., clinical services, substance use treatment, appropriate transitional care, and housing). Clients reported struggles during times of PCET staff turnover, which may provide insight into the importance of a seamless transition to lower-intensity services for client. A suggestion may be to include a period of overlap between PCET and lower-intensity services to provide a warm handoff or transition for clients. Our results showed that the stakeholders’ perceived barriers to successfully graduating clients to lower-intensity services included individual clients’ capabilities for independence. Staff members reported concerns that there were clients that may never be able to transition to lower-intensity services. These findings align with prior research, which identified clients’ characteristics (e.g., determination, high self-esteem, risky behaviors, lack of insight) as most related to failure in transitioning to lower-intensity services.13 In another study, staff members also shared concerns for clients’ stability after transitioning to lower-intensity services, including the unpredictability of the future for clients and violating the client’s preferences of continuing with ACT.18 While ACT provides positive outcomes for many, and some participants may eventually be ready to graduate to lower-intensity services, it is not feasible to transition everyone to lower-intensity services. Clients reported their struggles with cognitive issues such as memory and financial strains, which may be considered possible barriers to transitioning to lower-intensity services. Bromley and colleagues18 reported that staff had concerns that clients would feel rejected, abandoned, or frightened when discussing discharge/transitioning out of ACT. Finnerty and colleagues2 reported clinicians’ beliefs that clients and families would not want to terminate services (due to losing relationships with the ACT team members, fear of failure, and preference for ACT). Our study reiterated these concerns in the development of the interview protocol for clients of PCET. PCET staff members reported concerns about provoking anxiety in clients when asking clients directly about needs and barriers to transitioning out of PCET. Clients consistently reported appreciating and feeling like they needed PCET, which may provide insight into possible concerns regarding transitioning to lower-intensity services from the client’s perspective. The perceived concerns of transitioning PCET clients to lower-intensity services indicate, aligned with literature on psychiatric rehabilitation, the importance of ACT-like programs to incorporate recovery-oriented programs such as skill building toward independence and function and illness and symptom management. In an evaluation of ACT services and incorporation of recovery orientation, researchers found that ACT programs with higher recovery-oriented programs had better outcomes in hospitalization days, education involvement, and employment.20 ACT and ACT-like programs, such as PCET, are well suited to incorporate recovery-oriented practices. Salyers and Tsemberis 21 suggest that hiring peer specialists as ACT team staff members can positively impact the outcomes for service recipients. This was also noted by PCET staff in recommending peer support to be included in the program's future. The COVID-19 pandemic presented unique adjustments affecting both clients and staff members of the PCET. Clients and staff members reported adjusting to quarantine and physical distancing lockdown measures. Clients and staff overall said they continued connection during the pandemic via telehealth. One staff member even reported a higher rate of client treatment participation with telehealth as previously there were transportation challenges. These adjustments during COVID-19 for the PCET are like reported adjustments from other ACT programs. Guan and colleagues said changes and adaptations during COVID-19 in maintaining essential services while limiting contagion risk, promoting health, mitigating physical and mental health impacts, and promoting staff resilience and wellness.22 PCET clients and staff reported that overall adjustments during COVID-19 went well, like the reported experiences of a program in Minnesota.23 Furthermore, lessons learned during this evaluation on the effects of COVID-19 and changes during the pandemic may provide insight into possibilities for transitioning clients to lower-intensity services. For example, the use of telehealth or telecommunications may be used for the continuation of care when graduating.15 Funding This evaluation of the Pinellas County Empowerment Team was supported by funding from Pinellas County. Declarations Conflict of Interest The authors declare no competing interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Substance Abuse and Mental Health Services Administration. Assertive Community Treatment: How to Use the Evidence-Based Practice KITs. DHHS Pub. No. SMA-08-4344, Rockville, MD: Center for Mental Health Services, Substance Abuse and Mental Health Services Administration, U.S. Department of Health and Human Services, 2008. 2. Finnerty MT Manuel JI Tochterman AZ Clinicians’ perceptions of challenge and strategies of transition from assertive community treatment to less intensive services Community Mental Health Journal 2015 51 85 95 10.1007/210597-014-9706-y 24526472 3. Huz S Thorning H White CN Time in assertive community treatment: A statewide quality improvement initiative to reduce length of participation State Mental Health Policy 2017 68 539 541 10.1176/appi.ps.201700127 4. Mancini AD Moser LL Whitley R Assertive community treatment: Facilitators and barriers to implementation in routine mental health settings Psychiatric Services 2009 60 189 195 10.1176/ps.2009.60.2.189 19176412 5. Rosen A Mueser KT Teeson M Assertive community treatment – issues from scientific and clinical literature with implications for practice Journal of Rehabilitation Research & Development 2007 44 813 826 10.1682/JRRD.2006.09.0110.Accessed18May 18075939 6. McGrew JH Bond GR Critical ingredients of assertive community treatment: Judgments of the experts The Journal of Mental Health Administration 1995 22 2 113 125 10.1007/BF02518752 10142125 7. Cuddeback GS, Morrissey JP, Meyer PS. (2006). How many assertive community treatment teams do we need? Psychiatric Services, 57(12), 1803–1806. 8. Hackman AL Stowell KR Transitioning clients from assertive community treatment to traditional mental health services Community Mental Health Journal 2009 45 1 1 5 10.1007/s10597-008-9179-y 19130222 9. Bromley E Mikesell L Whelan F Clinical factors associated with successful discharge from assertive community treatment Community Mental Health Journal 2017 53 916 921 10.1007/s10597-017-0083-1 28116636 10. Dixon L Assertive community treatment: Twenty-five years of gold Psychiatric Services 2000 51 6 759 765 10.1176/appi.ps.51.6.759 10828107 11. Spivak S Cullen BA Green C Availability of assertive community treatment in the United States: 2010 to 2016 Psychiatric Services 2019 70 10 948 951 10.1176/appi.ps.201900032 31185855 12. Cuddeback GS Morrissey JP Meyer PS How many ACT teams do the researchers need? Results from a large, urban community Psychiatric Services 2006 57 12 1803 1806 10.1176/ps.2006.57.12.1803 17158499 13. Stull LG McGrew JH Salyers MP Processes underlying treatment success and failure in assertive community treatment Journal of Mental Health 2012 21 1 49 56 10.3109/09638237.2011.629239 22142361 14. Coronavirus disease (COVID-19) Weekly Epidemiological Update and Weekly Operational Update. Geneva, Switzerland: World Health Organization, 2020. Available at https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports. Accessed 18 May 2022. 15. Couser GP Taylor-Desir M Lewis S Further Adaptations and Reflections by an Assertive Community Treatment Team to Serve Clients with Severe Mental Illness During COVID-19 Community Mental Health Journal 2020 57 7 1217 1226 10.1007/s10597-021-00860-3 16. Guest G, MacQueen KM, Namey EE. Applied Thematic Analysis. Sage Publications, Inc., Thousand Oaks. 2012. http://www.sagepub.com/books/Book233379. Accessed 4 Oct 2022. 17. Donahue SA Manuel JI Herman DB Development and use of a transition readiness scale to help manage ACT team capacity Psychiatric Services 2012 63 3 223 229 10.1176/appi.ps.201100041 22307877 18. Bromley E Mikesell L Armstrong NP “You might lose him through the cracks”: Clinicians’ views on discharge from assertive community treatment Administration and Policy in Mental Health and Mental Health Services Research 2015 42 1 99 110 10.1007/s10488-014-0547-3 24692027 19. Salzer M, Baron R. Well Together—A Blueprint for Community Inclusion: Fundamental Concepts, Theoretical Frameworks and Evidence; Wellways Australia Limited: Melbourne, Australia. 2016. 20. Kidd SA George L O’Connell M Recovery-oriented service provision and clinical outcomes in Assertive Community Treatment Psychiatric Rehabilitation Journal 2011 2011 34 194 201 10.2975/34.3.2011.194.201 21. Salyers MP Tsemberis S ACT and recovery: Integrating evidence-based practice and recovery orientation on Assertive Community Treatment teams Community Mental Health Journal 2007 43 6 619 641 10.1007/s10597-007-9088-5 17514503 22. Guan I Kirwan N Beder M Adaptations and innovations to minimize service disruptions for patients with severe mental illness during COVID-19: Perspectives and reflections from an assertive community psychiatry program Community Mental Health Journal 2021 57 1 10 17 10.1007/s10597-020-00710-8 32930903 23. Couser GP Taylor-Desir M Lewis S Further adaptations and reflections by an Assertive Community Treatment team to serve clients with severe mental illness during COVID-19 Community Mental Health Journal 2021 5 1217 1226 10.1007/s10597-021000860-3.Accessed18May	36460895	PMC9718462	NO-CC CODE	2022-12-06 23:23:39	no	J Behav Health Serv Res. 2022 Dec 2;:1-14	utf-8	J Behav Health Serv Res	2,022	10.1007/s11414-022-09827-y	oa_other
==== Front Ther Innov Regul Sci Ther Innov Regul Sci Therapeutic Innovation & Regulatory Science 2168-4790 2168-4804 Springer International Publishing Cham 36459346 483 10.1007/s43441-022-00483-0 Review Bayesian Methods in Human Drug and Biological Products Development in CDER and CBER http://orcid.org/0000-0001-5892-8141 Ionan Alexei C. [email protected] 1 Clark Jennifer 1 Travis James 1 Amatya Anup 1 Scott John 2 Smith James P. 3 Chattopadhyay Somesh 1 Salerno Mary Jo 1 Rothmann Mark 1 1 grid.417587.8 0000 0001 2243 3366 Center for Drug Evaluation and Research, Office of Translational Sciences, Office of Biostatistics, U.S. Food and Drug Administration, Silver Spring, MD USA 2 grid.417587.8 0000 0001 2243 3366 Division of Biostatistics, Center for Biologics Evaluation and Research, Office of Biostatistics and Pharmacovigilance, U.S. Food and Drug Administration, Silver Spring, MD USA 3 grid.417587.8 0000 0001 2243 3366 Center for Drug Evaluation and Research, Office of New Drugs, Office of New Drug Policy, U.S. Food and Drug Administration, Silver Spring, MD USA 2 12 2022 19 13 9 2022 22 11 2022 © This is a U.S. Government work and not under copyright protection in the US; foreign copyright protection may apply 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The Center for Drug Evaluation and Research (CDER) and the Center for Biologics Evaluation and Research (CBER) of the U.S. Food and Drug Administration (FDA) have been leaders in protecting and promoting the U.S. public health by helping to ensure that safe and effective drugs and biological products are available in the United States for those who need them. The null hypothesis significance testing approach, along with other considerations, is typically used to demonstrate the effectiveness of a drug or biological product. The Bayesian framework presents an alternative approach to demonstrate the effectiveness of a treatment. This article discusses the Bayesian framework for drug and biological product development, highlights key settings in which Bayesian approaches may be appropriate, and provides recent examples of the use of Bayesian approaches within CDER and CBER. Keywords FDA Bayesian methods Clinical trials Adaptive design ==== Body pmcIntroduction Under the Federal Food, Drug, and Cosmetic Act, a drug’s effectiveness must be established by “substantial evidence,” which is “evidence consisting of adequate and well-controlled investigations, including clinical investigations, by experts qualified by scientific training and experience to evaluate the effectiveness of the drug involved, on the basis of which it could fairly and responsibly be concluded by such experts that the drug will have the effect it purports or is represented to have under the conditions of use prescribed, recommended, or suggested in the labeling or proposed labeling thereof.” [1] FDA has also generally considered “substantial evidence” of effectiveness to be necessary to support licensure of a biological product under Section 351 of the Public Health Services Act [2]. The FDA regulatory review determines whether “substantial evidence” has been demonstrated. Additionally, because all drugs have the potential for adverse effects, FDA integrates a structured benefit-risk assessment as part of the regulatory review of marketing applications for drugs and biological products [3]. The strength of evidence in each trial contributing to meeting the substantial evidence standard is assessed by appropriate statistical methods [4]. The Bayesian approach has a wide variety of potential applications in drug and biological product development [5]. However, there is a limited number of submissions featuring Bayesian methods. The limited number of submissions featuring Bayesian methods may be, at least in part, a result of insufficient knowledge of Bayesian approaches [6]. This article discusses the use of the Bayesian framework for drug and biological product development, highlights key settings in which Bayesian approaches may be appropriate, and provides recent examples of the use of Bayesian approaches within the Center for Drug Evaluation and Research (CDER) and the Center for Biologics Evaluation and Research (CBER) at the U.S. Food and Drug Administration (FDA). Areas with Emerging Bayesian Methods Use Although the use of Bayesian methods may be well suited in some clinical contexts, the methods may be considered regardless of study type, population, or therapeutic area, as appropriate. As with any proposal for a clinical trial, whether the proposed Bayesian study design and/or analysis is fit-for-purpose will be determined on a case-by-case basis. Some areas of drug or biological product development are highlighted below. Early Drug/Biological Product Development The objective of early-phase studies includes generating preliminary safety and efficacy data to help select drug or biological product dose(s) for subsequent development. There is an opportunity to use Bayesian model-based design and analyses to meet this objective. For example, a Bayesian model could potentially be used to estimate the probability of dose-limiting toxicity based on prior toxicity information or initial assumptions. The estimates could be continuously updated as more data are collected. Bayesian modeling in early-phase studies has been studied extensively, e.g., [7]. Some of the challenges in early-phase studies using Bayesian methods include clear characterization of the assumptions, study design, and model settings. Generally, providing sufficient details to describe and justify Bayesian study design, statistical model(s), and assumptions in proposals for early-phase studies is helpful to gain alignment among stakeholders. Detailed description and justification of the optimal and efficient designs (if applicable), sample size, duration of toxicity monitoring, potential dose levels, the model, prior distribution and its informativeness, parametrizations, and success criteria are especially helpful. Simulation results with operating characteristics under various scenarios and assumptions provide valuable information to assess a proposal for an early-phase study [8]. Noninferiority Bayesian methods are not limited to superiority trials and can be used in noninferiority (NI) trials as well. The goal of an NI trial is often to show that the difference between a new investigational product and an active control is small enough to allow the active control’s known effectiveness to support the conclusion that the new investigational product is also effective. An NI study seeks to show that the amount by which the investigational product is inferior to the active control is less than some prespecified NI margin. The FDA guidance on noninferiority trials acknowledges the potential utility of a Bayesian approach in this setting [9]. As examples, Bayesian methods can potentially be used in either the design of an NI trial (e.g., to determine the NI margin) or the analysis of the data generated by the trial. Use of Bayesian approaches can be particularly well suited to determining the NI margin, which involves synthesizing data from past studies. Some examples of the use of Bayesian methods to determine the NI margin and to analyze NI trial data can be found in Rothwell et al. [10], Gamalo et al. [11], and Price and Scott [12]. Adaptive Clinical Trials Some considerations on Bayesian adaptive study designs are included in the guidance for industry, Adaptive Design Clinical Trials for Drugs and Biologics. An adaptive design allows for prospectively planned modifications to the study design based on accumulating data from participants in the clinical trial. Bayesian adaptive study designs use the Bayesian framework for study adaptations and/or posterior distributions for decision-making. Examples of Bayesian adaptive design features include the following categories per Adaptive Design Clinical Trials for Drugs and Biologics:“Use of predictive statistical modeling, possibly incorporating information external to a trial, to govern the timing and decision rules for interim analyses. Use of assumed dose–response relationships to govern dose escalation and selection. Explicit borrowing of information from external sources (e.g., previous trials, natural history studies, and registries) via informative prior distributions to improve the efficiency of a trial. Use of posterior probability distributions to form criteria for trial success.” Pediatric Clinical Trials Informative Bayesian methods can be a good fit for pediatric studies, as these methods can allow the incorporation of prior information about efficacy from adults or other source populations (for example, using information from adolescents in drawing inferences on younger children), when clinically appropriate, to facilitate a trial design that requires the enrollment of fewer children [13]. Bayesian studies are consistent with the established concept of pediatric extrapolation, which allows for efficacy to be assessed in pediatric patients with support from information gathered in other populations [14]. Extrapolation, when scientifically justified, is particularly attractive in pediatric investigations as it can allow us to reduce the size or need for studies in children who are a vulnerable population in need of extra patient protections. Bayesian methods accomplish this by creating a prior distribution based on the information leveraged from the source population which can be used in the analysis in the target pediatric population. In the past, extrapolation was employed using a decision-tree approach, through which the development program could be reduced if certain thresholds of evidence were reached [15]. Bayesian methods allow more flexibility for pediatric studies, given that the prior information can be weighted based on the level of applicability rather than an all-or-nothing approach to incorporating the information. Deciding upon applicability a priori is preferable because it separates applicability from outcomes of the new study, which may bias reviewers with regard to how previous studies are viewed. These methods were discussed in a September 2021 joint FDA and Maryland Center for Excellence in Regulatory Science and Innovation workshop, “Advancing the Development of Pediatric Therapeutics Complex Innovative Trial Design Public Workshop,” which included presentations and discussion on the use of Bayesian methods in pediatric clinical trials from FDA, the European Medicines Agency, and industry [16]. There are also ongoing international efforts within the International Council for Harmonization to gain alignment on the scope and applicable methodologies of pediatric extrapolation generally, including the use of Bayesian methods [17]. First, to do extrapolation, one needs to create the prior distribution containing the borrowed information intended for use in the analysis. FDA has seen several different proposed approaches, but they tend to follow a similar process:Determine the relevant data to be used in the construction of the prior distribution. Synthesize the different sources of information. This involves weighing the relative relevance of the different sources to the new scientific question under investigation. Determine the final overall weight of the prior based on balancing the amount of prior evidence, the uncertainty about the applicability of this evidence, and the required sample size for the study. The main challenges involved in constructing the prior distribution are synthesizing the information and determining the level of weighting. Synthesizing the information becomes complex when a range of heterogeneous data sources are used. For example, if we have a mix of adult and pediatric clinical trials, or if some of the trials used different designs than the study being planned while others used the same design, then we have different amounts of relevance. More complex statistical models may be used to reflect these relevant imbalances. In all cases, there are necessary subjective judgments to be made in building these models, requiring multidisciplinary collaboration in order to design and implement. Some examples of these subjective judgments include assessments about the degree of heterogeneity between trials and the relative relevance of the different sources of information. Schmidli et al. [18] provides an example of this process in practice, where the authors used information from both adult and pediatric studies for fingolimod to create a prior distribution for a study utilizing fingolimod as an active control in pediatric patients. Discussions related to how to weight the prior information relative to the information collected in the trial can be challenging as finding an acceptable balance in weighting between the information collected in the trial and the prior information is critical in order for the results of the subsequent analyses to be persuasive. If the prior information is weighted too heavily, then the prior information will overpower the data generated from the pediatric population, and the result may not be persuasive. If too little weight is given to the prior information, the pediatric trials may need to enroll more pediatric patients than might be necessary. Rare Diseases Similar to pediatric clinical trials, rare disease clinical trials present challenges that may also offer an opportunity for the Bayesian framework. These methods can be useful when there is a limited pool of patients, and data exist that can inform a prior distribution, such as data from earlier clinical trials. Registry data or other real-world data might also be useful for informing a prior distribution, if such use is sufficiently justified in terms of data quality, applicability, and other aspects [19, 20]. In such cases, evidence from various sources can be integrated into the assessment with uncertainty expressed in a probability scale. Interpretation from the functions of the posterior probability distribution can often be straightforward if the early study design strategy is appropriately deliberated and selected [21]. Although designing rare disease clinical trials is different from designing pediatric clinical trials and presents its own distinct set of issues, the approaches for choosing a prior distribution as discussed for pediatric clinical trials would also apply for rare disease clinical trials. What constitutes an appropriate prior distribution for a rare disease will vary depending on the experimental treatment, control treatment, study population, and the degree of similarity between prior and prospective data. It is useful to incorporate a skeptical prior, which quantifies skepticism that the treatment is beneficial or harmful. As the therapeutic landscape evolves, the relevance of prior data for the population of interest may change. Assessing data quality attributes (e.g., source, usability, timing, accuracy, completeness) and comparability attributes (e.g., demographics, disease features, phenotype, genotype, disease progress, standards of care) are helpful to ensure that analysis results using prior data will provide usable and reliable information that is fit for regulatory purposes. Subgroup Assessment A subgroup is a given subset of a clinical trial population that has a certain characteristic(s). Along with the evaluation of a treatment effect in the overall study population, assessment within subgroups of patients where the efficacy may vary is important. Traditionally, treatment effects in an individual subgroup in a pivotal trial are estimated by the observed effect within that subgroup. Although these direct estimates are easily understood, the observed treatment effects can be highly imprecise estimates of the true treatment effects. The variability of observed treatment effects across the subgroups in a trial is generally greater than the variability of the true subgroup treatment effects [22]. The observed variability of treatment effects across subgroups can be extreme when subgroups are small, given that observed treatment effects are more susceptible to random highs and lows of the sample data. Bayesian methods that weigh data in different subgroups in a way that increases the precision of the subgroup treatment effect estimates can correct these extremes by pulling the estimates toward the overall treatment effect in the total study population. Bayesian hierarchical models for subgroup analysis have been discussed extensively in the literature and featured in the CDER impact stories as an innovative statistical approach to provide the most reliable treatment outcomes information to patients and clinicians [23]. The main idea is to analyze relevant subgroups together rather than in isolation. The Bayesian method accomplishes this by linking individual subgroups through a prior distribution encompassing underlying treatment effects in each subgroup. Borrowing information from other subgroups improves the overall accuracy of the estimates across the subgroups. This property is particularly useful for the analysis of data from the smaller subgroups. Examples of Bayesian Methods Use in CDER- and CBER-Reviewed Clinical Trials Example 1: Pediatric Trials (Approval, Belimumab) [24] In 2018, FDA received a New Drug Application (NDA) supplement containing the results of a Pediatric Research Equity Act (PREA)-required postmarketing study for belimumab, a B-lymphocyte stimulator (BLyS)-specific inhibitor, in the treatment of pediatric patients with active, autoantibody-positive systemic lupus erythematosus (SLE) who were receiving standard therapy. Due to the rarity of the disease in children, an adequately powered pediatric study was not feasible but a study nonetheless ran for over 5 years and managed to enroll and randomize 93 participants. The primary endpoint was the proportion of patients who met the SLE Responder Index (SRI) Response criteria at Week 52. The results of the primary analysis for this endpoint are shown in Table 1. The primary analysis failed to demonstrate a statistically significant difference between belimumab and placebo.Table 1 Primary efficacy analysis of SRI response rate at week 52 from the pediatric study Placebo N = 40a Belimumab 10 mg/kg N = 53 Response, % (n) 44% (17) 53% (28) Observed difference – 9.2% Odds ratio (95% CI) – 1.5 (0.6, 3.5) aOne subject in placebo did not have a baseline SELENA SLEDAI assessment and, therefore, did not contribute to SRI analyses Given concerns regarding the feasibility of conducting another study in this pediatric population, the review team considered whether a Bayesian analysis could be informative, incorporating the data collected in the adult SLE studies. To conduct such an analysis, a prior distribution had to be determined. The review team considered this in a similar fashion to the three steps discussed previously in Section d. First, they identified the relevant data that could be used; specifically, there were two adult efficacy studies that compared two dose levels (1 mg/kg and 10 mg/kg) against placebo. These data were used since the clinical review team believed the disease and patient response to treatment were likely to be similar between the adult and pediatric subjects, given that the pediatric and adult diseases have similar underlying pathophysiology and management, with BLyS, the target of belimumab, being similarly relevant; systemic belimumab exposures were also similar. The data from the 1 mg/kg dose arms were not considered relevant because the pediatric study only used the higher dose. The results from the adult studies for the 10 mg/kg dose are shown in Table 2.Table 2 Primary efficacy analysis of SRI response rate at week 52 from the adult studies Adult study 1 Adult study 2 Placebo N = 275 Belimumab 10 mg/kg N = 273 Placebo N = 287 Belimumab 10 mg/kg N = 290 Response, % (n) 34% (93) 43% (118) 44% (125) 58% (167) Observed difference – 9% – 14% Odds ratio (95% CI) – 1.5 (1.1, 2.1) – 1.8 (1.3, 2.6) Next, the data from the adult studies were combined to produce a single probability distribution. The final step of the process was to use a mixture prior approach to reweight the results to ensure that the adult data did not overwhelm the pediatric data in the analysis. This approach was used to vary the amount of information borrowing between no borrowing (represented by a weight of zero of 0) and full borrowing (represented by a weight of 1) where the adult and pediatric data are essentially pooled together, with every patient counted equally. The Bayesian analysis was performed for the entire range of weights, from no borrowing to full borrowing, to allow a complete view of the spectrum of outcomes. Point estimates (posterior means) and uncertainty intervals (95% credible intervals) were computed for each weight value and are presented in Fig. 1. As more weight was placed on the adult study results, the point estimates moved toward the overall average slightly and the width of the uncertainty intervals shrank considerably.Figure 1 Posterior mean (points) and 95% credibility intervals (lines) of the odds ratio of SRI response. Source: Reproduced using Table 34 FDA Statistical Review. Typically, to try to reduce subjective biases, the amount of borrowing should be prespecified. Although it was not prespecified in this case, the clinical team did have a pre-existing belief in the similarity between adults and pediatrics based on the similarity of the disease pathogenesis and management. The Bayesian analysis found that a prior distribution weight of at least 0.55 resulted in posterior probabilities of positive treatment effects of greater than 97.5%. This 0.55 weight was found to be reasonable by the review team, and thus this analysis provided support, along with additional evidence, for the approvability of belimumab in the pediatric SLE population. Example 2: COVID-19 Vaccine In response to the COVID-19 pandemic, BioNTech Manufacturing GmBH, in partnership with Pfizer Inc., proposed a phase 1/2 trial of their SARS-CoV-2 mRNA vaccine candidate, BNT162b2. The original protocol for trial C4591001 was finalized on April 15, 2020, and was subsequently amended on July 24, 2020, to include a phase 3 trial. The trial was titled “A phase 1/2/3 Placebo-Controlled, Randomized, Observer-Blind, Dose-Finding Study to Evaluate the Safety, Tolerability, Immunogenicity, and Efficacy of SARS-CoV-2 RNA Vaccine Candidates Against COVID-19 in Healthy Individuals.” Trial C4591001 had a number of different objectives, including objectives related to dose-finding, safety, and immunogenicity. The efficacy analyses in the phase 2/3 portion of the study are most relevant for this discussion. Of note, the results described in this section are based on the data submitted as part of the COMIRNATY original Biologics License Application (BLA); the study was ongoing at that time and there have since been additional analyses. Over 44,000 subjects, aged 16 years and older, were randomized 1:1 to receive two doses of BNT162b2 or placebo at a 21-day interval. The primary efficacy endpoint was confirmed cases of COVID-19 (see protocol [25] and FDA review documents for exact definition [26]), and the primary efficacy null hypothesis was Vaccine Efficacy (VE) ≤ 30%, where VE = 100 * (1 – incidence rate ratio [IRR]). Under the assumption that the number of cases in each group follows a Poisson distribution, the number of cases in the vaccine group, s1, has a binomial distribution conditional on the total number of cases, s; s1 ~ Binomial(s, θ) where θ=r(1-VE)r1-VE+1 and r is the ratio of study time between the vaccine and placebo groups. The study was designed with a simple Bayesian efficacy analysis using a beta-binomial model with a weakly informative beta(0.700, 1) prior distribution for θ, chosen to have a prior mean of 0.4118 (corresponding to a VE of 30%). The null hypothesis would be rejected if the posterior probability that VE > 30% exceeded prespecified thresholds at interim or final analyses. The thresholds were chosen to maintain a familywise one-sided Type I error rate of 2.5%. There were also interim futility analyses planned based on the trial’s posterior predictive probability of success, calculated with a beta-binomial model. At the time of the interim analysis, conducted with a data cutoff date of November 4, 2020, there were four confirmed cases in the BNT162b2 arm and 90 confirmed cases in the placebo arm, yielding a VE of 95.5% with a 95% credible interval of (88.8%, 98.4%). The posterior probability of VE > 30% was 0.9999, which exceeded the prespecified threshold of 0.995. Updated final efficacy analyses were similarly strong [25, 26]. These efficacy findings played a critical role in BNT162b2 becoming the first COVID-19 vaccine authorized in the United States under Emergency Use Authorization (December 11, 2020) and subsequently becoming the first U.S.-licensed COVID vaccine under the trade name COMIRNATY (August 23, 2021), landmark milestones in the pandemic response [27]. Example 3: Drug Trials Snapshots (Approval, Bempedoic Acid) CDER statisticians have used Bayesian hierarchical models to provide information on treatment effect for various demographic subgroups in the FDA Drug Trials Snapshots program. For approved new molecular entities and original biological products, the Drug Trials Snapshots webpage [28] provides information on study design, results of efficacy, and safety studies, and whether there were differences in efficacy and side effects among various subgroups defined by sex, race, and age. The subgroup assessment in the Drug Trials Snapshots for bempedoic acid is an example of the use of a Bayesian method [29]. The applicant’s phase 3 clinical development program consisted of two randomized, placebo-controlled trials that enrolled adult patients with heterozygous familial hypercholesterolemia or established atherosclerotic cardiovascular disease who were on maximally tolerated statin therapy but required additional lowering of LDL-cholesterol (LDL-C). A treatment effect was assessed by percent change in LDL-C from baseline to Week 12. At the time of the application review, the difference in the LDL-C change between treatment groups within sexes, age groups, and ethnicities was estimated independently in isolation as well as by borrowing information from other subgroups using a Bayesian hierarchical model. For the Bayesian analysis of treatment effects in subgroups based on sex, before observing the trial data (i.e., a priori), it was assumed that the underlying treatment effect was not expected to be in a particular order by sex. That is, it was assumed that the probability that the treatment effect in male patients would be greater than in female patients was the same as the probability that the treatment effect in female patients would be greater than in male patients. This exchangeability assumption must hold for the resulting treatment effect estimates to be valid. Furthermore, the treatment effects in male and female subgroups were linked by imposing a prior distribution over the unknown true treatment effects within subgroups. An additional layer of prior distributions, called a hyper-prior, was imposed on the parameters of the prior distribution, so that estimated values of the parameters were driven primarily by the observed data. Mathematical details of the model are included in the FDA’s application review [29]. Estimates for subgroups of age and region were obtained similarly. The results based on observed data alone and the Bayesian analysis in the same subgroups from one of the phase 3 trials (Trial 047) are presented in Fig. 2. The 95% confidence intervals are shown for the observed sample estimates, and 95% credible intervals are shown for the estimates based on the Bayesian analysis. The credible intervals were derived from the posterior distribution of the average difference in LDL-C for the corresponding subgroups. As expected, 95% credible intervals are mostly narrower for the estimates based on the Bayesian analysis, reflecting the precision gained through borrowing information across all the subgroups. The gain in precision via Bayesian analysis is particularly notable in the subgroups with wider 95% confidence intervals.Figure 2 Trial 047 subgroup analyses. Source: FDA’s review of bempedoic acid tablets (NDA 211,616). For the bempedoic acid Drug Trials Snapshots, the results from two trials were combined to provide the overall treatment effect estimate for each subgroup [30]. This required an additional layer of hierarchy to link the study-specific treatment effects from the two trials. Example 4: Complex Innovative Trial Design Clinical Trials FDA announced the Complex Innovative Trial Design (CID) Pilot Meeting Program in the Federal Register on August 30, 2018 [31]. The goals of the CID pilot meeting program were toFacilitate the advancement and use of complex adaptive, Bayesian, and other novel clinical trial designs and Promote innovation by allowing FDA to publicly discuss the trial designs developed through the pilot meeting program as case studies, including while the drug studied in the trial has not yet been approved by FDA. The CID pilot meeting program was intended especially for innovative clinical trial designs that require simulations to estimate the operating characteristics. Several examples of CID case studies in the program have been published [12, 32]. Per Federal Register notice on October 20, 2022 [33], FDA is continuing the program as the Complex Innovative Trial Design Paired Meeting Program [32]. Discussion Fitness-for-purpose of a study design and/or analysis is determined on a case-by-case basis. This article highlighted some settings that might be well suited for Bayesian methods. The Bayesian framework has many benefits as well as challenges [5]. One of the challenges of Bayesian approaches is the absence of a single, universally accepted criterion for study success. Instead, at this time, clinical context (e.g., seriousness of disease, prevalence of disease, unmet medical need) is important in determining an appropriate criterion for study success. However, this flexibility is also a strength of the Bayesian framework because it requires a thoughtful cross-disciplinary discussion about how multiple factors influence both the prior distribution as well as the criteria for study success. FDA has numerous avenues to facilitate the appropriate use of Bayesian methods in drug and biological product development. Moreover, examples and discussion of the use of Bayesian methods have been highlighted in publications and presentations made by FDA staff e.g., [34–37], in Drug Trials Snapshots, and in several guidance for industry documents [4, 38–40]. FDA is committed to continuing to advance the appropriate use of Bayesian methods through ongoing and future efforts, including publishing a draft guidance on the Use of Bayesian Methodology in Clinical Trials of Drugs and Biologics by the end of the fiscal year 2025 as a part of the Prescription Drug User Fee Act VII commitment [41]. Acknowledgements The authors express their thanks to Sylva Collins, John Concato, Thomas Gwise, Frank Harrell, Laura Higginbotham, Lei Huang, Anna Kettermann, Paul Kluetz, Stefanie Kraus, Gregory Levin, Nikolay Nikolov, Dionne Price, Gregory Reaman, John Sharretts, Peter Stein, and Lihan Yan. Disclaimer This article reflects the views of the authors and should not be construed to represent FDA’s views or policies. Author Contributions All authors contributed to the writing and editing of this manuscript. Funding No funding was received for this manuscript. Declarations Conflict of interest The authors have no conflicts of interest relevant to this article to disclose. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Federal Food, Drug, and Cosmetic Act section 505(d) (21 USC § 355(d)). 2. Public Health Services Act section 351 (42 USC § 262). 3. Section 505(d) of the Federal Food, Drug, and Cosmetic Act (21 USC 355(d)). 4. Demonstrating substantial evidence of effectiveness for human drug and biological products. Draft Guidance for Industry. U.S. Department of Health and Human Services. Food and Drug Administration; 2019. 5. Spiegelhalter DJ, Abrams KR, Myles JP. Bayesian approaches to clinical trials and health-care evaluation. Wiley, New York; 2004. 6. Medical Outreach Subteam of the Drug Information Association Bayesian Scientific Working GroupClark J Muhlemann N Natanegara F Hartley A Wenkert D Wang F Harrell FE Jr Bray R Why are not there more Bayesian clinical trials? Perceived barriers and educational preferences among medical researchers involved in drug development Therap Innov Regul Sci 2022 10.1007/s43441-021-00357-x 7. Yuan Y, Nguyen HQ, Thall PF. Bayesian designs for phase I–II clinical trials. Chapman and Hall, London; 2017. 8. Mayer C Perevozskaya I Leonov S Simulation practices for adaptive trial designs in drug and device development Stat Biopharm Res 2019 11 4 325 335 10.1080/19466315.2018.1560359 9. Non-Inferiority Clinical Trials. Guidance for Industry. U.S. Department of Health and Human Services. Food and Drug Administration; 2016 10. Rothwell R Nikolov NP Maynard JW Levin G Noninferiority trials to evaluate drug effects in rheumatoid arthritis Arthritis Rheumatol 2020 72 8 1258 1265 10.1002/art.41257 32182406 11. Gamalo MA Tiwari RC LaVange LM Bayesian approach to the design and analysis of non-inferiority trials for anti-infective products Pharm Stat 2014 13 1 25 40 10.1002/pst.1588 23913880 12. Price D, Scott J. The U.S. Food and Drug Administration's complex innovative trial design pilot meeting program: progress to date. Clin Trials. 2021;18(6):706–710. 10.1177/17407745211050580 13. Goodman SN Sladky JT A Bayesian approach to randomized controlled trials in children utilizing information from adults: the case of Guillain-Barre Clin Trials 2005 2 4 305 310 10.1191/1740774505cn102oa 16281429 14. Applications for FDA Approval to Market a New Drug, Pediatric use information. 21 C.F.R. § 314.55. 15. General Clinical Pharmacology Considerations for Pediatric Studies for Drugs and Biological Products. Draft Guidance for Industry. U.S. Department of Health and Human Services. Food and Drug Administration; 2014. 16. Public Workshop. Food and Drug Administration and University of Maryland Center for Excellence in Regulatory Science and Innovation: Advancing the Development of Pediatric Therapeutics Complex Innovative Trial Design. September 1–2, 2021. Videocast. https://collaboration.fda.gov/p45rbov94km1/. Accessed 7 Apr 2022 17. E11A Pediatric Extrapolation. Draft Guidance Document. U.S. Department of Health and Human Services. Food and Drug Administration; 2022. 18. Schmidli H Häring DA Thomas M Cassidy A Weber S Bretz F Beyond randomized clinical trials: use of external controls Clin Pharmacol Ther 2020 107 4 806 816 10.1002/cpt.1723 31725899 19. Real-world data: assessing registries to support regulatory decision-making for drug and biological products. Draft Guidance for Industry. U.S. Department of Health and Human Services. Food and Drug Administration; 2021. https://www.fda.gov/media/154449/download 20. Real-world data: assessing electronic health records and medical claims data to support regulatory decision-making for drug and biological products. Draft Guidance for Industry. U.S. Department of Health and Human Services. Food and Drug Administration; 2021. https://www.fda.gov/media/152503/download 21. Kidwell KM, Roychoudhury S, Wendelberger B, et al. Application of Bayesian methods to accelerate rare disease drug development: scopes and hurdles. Orphanet J Rare Dis. 2022;17(1):186.10.1186/s13023-022-02342-5 22. Pennello G Rothmann M Peace KE Chen DG Menon S Bayesian subgroup analysis with hierarchical models Biopharmaceutical applied statistics symposium, volume 2 biostatistical analysis of clinical trials 2018 Springer 175 192 23. Impact Story: Using innovative statistical approaches to provide the most reliable treatment outcomes information to patients and clinicians. Food and Drug Administration. June 10, 2019. https://www.fda.gov/drugs/regulatory-science-action/impact-story-using-innovative-statistical-approaches-provide-most-reliable-treatment-outcomes. Accessed 7 Apr 2022 24. Nikolov N, Neuner R, Pottackal G, Niu J, et al. Reviews of pediatric studies conducted under BPCA and PREA from 2012—present: BLA 125370/s-064 and BLA 761043/s-007 multi-disciplinary review and evaluation Benlysta® (belimumab) for intravenous infusion in children 5 to 17 years of age with SLE. Food and Drug Administration, U.S. Department of Health and Human Services; 2019. https://www.fda.gov/media/127912/download. Accessed 11 Apr 2022 25. Polack FP Thomas SJ Kitchin N Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine N Engl J Med 2020 383 27 2603 2615 10.1056/NEJMoa2034577 33301246 26. Huang L. Comirnaty; approval history, letters, reviews and related documents: statistical review. Food and Drug Administration, U.S. Department of Health and Human Services; 2021. https://www.fda.gov/vaccines-blood-biologics/comirnaty. Accessed 11 Apr 2022 27. FDA approves first COVID-19 vaccine, approval signifies key achievement for public health. FDA News Release. August 23, 2021. https://www.fda.gov/news-events/press-announcements/fda-approves-first-covid-19-vaccine. Accessed 8 Apr 2022 28. Drug Trials Snapshots. Food and Drug Administration. April 9, 2021. https://www.fda.gov/drugs/drug-approvals-and-databases/drug-trials-snapshots. Accessed 7 Apr 2022 29. Sharretts J, Higginbotham L, Ketterman A, Liu T, et al. Drugs at FDA: FDA-approved drugs, Application Number: 211616Orig1s000 Integrated Review, Executive Summary, Interdisciplinary Assessment, Appendices. Food and Drug Administration, U.S. Department of Health and Human Services; 2019. https://www.accessdata.fda.gov/drugsatfda_docs/nda/2020/211616Orig1s000IntegratedR.pdf. Accessed 11 Apr 2022 30. Drug Trials Snapshots: NEXLETOL. March 2, 2020. https://www.fda.gov/drugs/resources-information-approved-drugs/drug-trials-snapshots-nexletol. Accessed 4 Nov 2022 31. Complex Innovative Designs Pilot Meeting Program, 83 FR 44274, federalregister.gov/d/2018-18801. 32. Complex Innovative Trial Design Meeting Program. Food and Drug Administration. 27 October 2022. https://www.fda.gov/drugs/development-resources/complex-innovative-trial-design-meeting-program. Accessed 4 Nov 2022 33. Complex Innovative Design Paired Meeting Program, 87 FR 63787, federalregister.gov/d/2022–22794 34. Irony T, Huang L. Bayesian approaches in the regulation of medical products. In: Lakshminarayanan M, Natanegara F, editors. Bayesian Applications in Pharmaceutical Development. Chapman and Hall, London; 2019. 35. Scott J. Bayes in medical product regulatory decision-making. Lecture presented at ASA Symposium on Statistical Inference; October 2017; Bethesda, MD. 36. Nie, L. Applications of Bayesian methods in regulatory decisions. Lecture presented at ASA Biopharmaceutical Section Regulatory-Industry Statistics Workshop; September 2020; Online. 37. Fernandes, L. Applications of Bayesian Methods within CDER. Lecture presented at ASA Biopharmaceutical Section Regulatory-Industry Statistics Workshop; September 2020; Online. 38. Guidance for the Use of Bayesian Statistics in Medical Device Clinical Trials. Guidance for Industry and FDA Staff. U.S. Department of Health and Human Services. Food and Drug Administration; 2010. 39. Interacting with the FDA on Complex Innovative Trial Designs for Drugs and Biological Products. Guidance for Industry. U.S. Department of Health and Human Services. Food and Drug Administration; 2020. 40. Adaptive Designs for Clinical Trials of Drugs and Biologics. Guidance for Industry. U.S. Department of Health and Human Services. Food and Drug Administration; 2019. 41. PDUFA VII: Fiscal years 2023–2027. Food and Drug Administration. August 23, 2021. https://www.fda.gov/media/151712/download. Accessed 7 Apr 2022	36459346	PMC9718464	NO-CC CODE	2022-12-06 23:23:39	no	Ther Innov Regul Sci. 2022 Dec 2;:1-9	utf-8	Ther Innov Regul Sci	2,022	10.1007/s43441-022-00483-0	oa_other
==== Front Mem Cognit Mem Cognit Memory & Cognition 0090-502X 1532-5946 Springer US New York 1374 10.3758/s13421-022-01374-3 Article The independent effects of source expertise and trustworthiness on retraction believability: The moderating role of vested interest Susmann Mark W. [email protected] 1 Wegener Duane T. 2 1 grid.261331.4 0000 0001 2285 7943 Department of Computer Science and Engineering, Ohio State University, Columbus, OH USA 2 grid.261331.4 0000 0001 2285 7943 Department of Psychology, Ohio State University, Columbus, OH USA 2 12 2022 117 13 11 2022 © The Psychonomic Society, Inc. 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Past research suggests that the trustworthiness of a source issuing a retraction of misinformation impacts retraction effectiveness, whereas source expertise does not. However, this prior research largely used expert sources who had a vested interest in issuing the retraction, which might have reduced the impact of those expert sources. We predicted that source expertise can impact a retraction’s believability independent of trustworthiness, but that this is most likely when the source does not have a vested interest in issuing a retraction. Study 1 demonstrated that retractions from an expert source are believed more and lead to less continued belief in misinformation than retractions from an inexpert source while controlling for perceptions of trustworthiness. Additionally, Study 1 demonstrated that this only occurs when the source had no vested interest in issuing the retraction. Study 2 found similar effects using a design containing manipulations of both expertise and trustworthiness. These results suggest that source expertise can impact retraction effectiveness and that vested interest is a variable that is critical to consider when determining when this will occur. Supplementary Information The online version contains supplementary material available at 10.3758/s13421-022-01374-3. Keywords Misinformation Expertise Trustworthiness Vested interest ==== Body pmcIntroduction Misinformation poses a challenge to society. For example, misinformation about the COVID-19 pandemic is prevalent and harmful, having been linked to thousands of deaths and hospitalizations (Islam et al., 2020). Although sources ranging from credible news and fact-checking organizations to individuals with little expertise on the topics did try to correct such false claims, belief in them persisted and led to tragic consequences. From the perspective of research on the continued influence effect of misinformation (CIE; Johnson & Seifert, 1994), this outcome is sadly unsurprising. The CIE refers to the robust finding that misinformation tends to have continued influences on beliefs and judgments even after it has been retracted (for reviews, see Lewandowsky et al., 2012; Seifert, 2002; Swire & Ecker, 2018). Given this, it is important to understand the factors that influence a retraction’s effectiveness. Most explanations of the CIE stem from cognitive perspectives (Lewandowsky et al., 2012; Swire & Ecker, 2018). For instance, one prominent account posits that the CIE is the result of memory processes (Lewandowsky et al., 2012; Swire & Ecker, 2018). According to this account, misinformation is automatically activated when one encounters a cue related to the misinformation. A strategic monitoring process must then retrieve the retraction in order for the misinformation to be accurately evaluated as false. This strategic monitoring process is proposed to be susceptible to failure, resulting in continued belief in misinformation due to failures to retrieve the retraction. Such memory processes are likely involved in the CIE. However, additional factors not necessarily linked to those memory processes could also contribute. For example, one set of factors relates to whether people believe that the retraction is valid. O’Rear and Radvansky (2019) demonstrated that retractions are not always believed, and people who do not believe a retraction show greater continued influence effects than those who do. If retraction believability is a key determinant of the CIE, it is important to understand what factors might make a retraction more or less believable. One factor that could have such an effect is how credible (how generally believable) the source of the retraction appears (Cooper et al., 2016; Petty & Cacioppo, 1981; Petty & Wegener, 1998), whereby more credible sources might confer greater believability to their retractions than less credible sources. Source credibility is traditionally conceptualized as a summary judgment arising from consideration of two independent components: source expertise, which refers to how knowledgeable a source is in the focal domain, and source trustworthiness, which refers to how honest a source is likely to be when providing information (Cooper et al., 2016; McGuire, 1985; Petty & Cacioppo, 1981; Petty & Wegener, 1998). Sources higher in expertise and trustworthiness are typically seen as more credible than those lower in these characteristics, and credible sources are generally more persuasive than noncredible sources (Hovland & Weiss, 1951; Pornpitakpan, 2004; Priester & Petty, 1995). Past research supports some aspects of these notions when examining the impact of source credibility on the CIE, but not others. Guillory and Geraci (2013) found that retractions of misinformation from credible sources are more effective at reducing continued belief in misinformation than retractions from noncredible sources. However, subsequent studies suggested that this effect was driven by source trustworthiness but not source expertise. That is, sources who normatively differed in trustworthiness but were equated on expertise created differences in the CIE. Yet, sources who differed on expertise but were equated on trustworthiness did not show this effect. Additional research has since reinforced these results (Ecker & Antonio, 2021; Pluviano et al., 2020). These findings are surprising given past work documenting source expertise effects on persuasion, information processing, and resulting behavioral intentions. Expert sources have been found to be more persuasive than nonexpert sources (Hovland & Weiss, 1951; Petty et al., 1981; Pornpitakpan, 2004), to validate thoughts in response to persuasive messages more than nonexpert sources (Tormala et al., 2006), and to have greater impact on subsequent product purchasing intentions than nonexpert sources (Ohanian, 1991; Yoon et al., 1998). It is therefore informative to consider why past research has found no impact of source expertise in a CIE context given the importance of source expertise in other domains. One possibility stems from the specific materials used in the past CIE research (i.e., Ecker & Antonio, 2021; Guillory & Geraci, 2013; Pluviano et al., 2020), which might have inadvertently disadvantaged any potential impact of expertise. Specifically, the expert sources used commonly had vested interests in retracting the misinformation, whereas many of the low-expertise sources did not have vested interests. A source has a vested interest when the source stands to personally benefit if others believe information the source is disseminating. In the persuasion literature, sources with a vested interest are generally less persuasive than sources without a vested interest (Eagly et al., 1978; Kelman & Hovland, 1953; Walster et al., 1966). As such, it is possible that any possible impact of perceptions of source expertise in past studies looking at the impacts of expertise in the CIE was hindered by the presence of vested interests. For example, Guillory and Geraci (2013) used a paradigm in which a politician is alleged to have taken a bribe, with that claim later being corrected. In their second study, they sought to manipulate retraction source expertise and hold trustworthiness constant by using sources normatively rated to be high/low in expertise but neutral in trustworthiness. At issue, two of the three high expertise sources used were the politician himself and his campaign manager, who both would have obvious vested interests in retracting claims he had taken a bribe. As such, the lack of expertise effects in this study might have been due to this confound between expertise and vested interest. One reason vested interest could undermine expertise effects is that vested interest might act as a cue that shifts emphasis towards perceptions of trustworthiness and away from perceptions of expertise. If this is the case, the vested interest cues present in both the high and low expertise sources in past research might have generally shifted participants’ focus onto the sources’ trustworthiness and away from expertise. This is possible given that vested interest cues might prompt individuals to consider whether the source is telling the truth (Walster et al., 1966), which might then make source trustworthiness more salient than expertise when judging the source’s overall credibility. As such, it is possible that only perceptions of trustworthiness will impact perceptions of credibility when the source has a vested interest, but that effects of expertise would emerge if no vested interest cues were present. This possibility is theoretically interesting because it suggests a possible moderator for source expertise effects, both within the CIE domain and beyond. Namely, it is possible that source expertise primarily matters when the source has no self-interest in disseminating a piece of information. As such, examining when source expertise can impact the effectiveness of a retraction would not only further understanding of the CIE, but would also inform the literature on expertise effects more generally. An additional distinction that might be important regards what the key outcome variable in most CIE studies is truly capturing and whether trustworthiness or expertise might be more likely to predict that specific construct. Much past work has considered continuing influence of misinformation in terms of whether people use misinformation as the basis to make misinformation-relevant inferences (e.g., Ecker et al., 2010, 2011a, b, 2014, 2015; Johnson & Seifert, 1994; Rich & Zaragoza, 2016; Wilkes & Leatherbarrow, 1988). These measures rely on asking participants open-ended questions that require an inference to be made and evaluating whether the inference is based on misinformation. For example, in a paradigm where misinformation that combustible materials caused a fire was presented and later corrected, if one were to infer that combustible materials caused explosions during the fire that would be considered a misinformation-based inference. However, it is unclear that this is the only form of continued influence that should be considered. Specifically, assessing how much participants continue to believe the misinformation after it is retracted also seems valuable. Though a belief measure might seem redundant with inferential measures of misinformation reliance, past research shows that these two outcomes are only moderately correlated (Susmann & Wegener, 2022a, b). Although it is difficult to make a strong a priori prediction about whether perceptions of a retraction source’s expertise or trustworthiness should have more of an effect on misinformation belief or its use to make inferences, it is possible that considerations of a source’s credibility are most closely related to belief in the misinformation. As such, effects of expertise might be most expected on this outcome whereas effects of expertise might be less likely to carry over to inferential reasoning, which might be influenced by factors beyond belief in misinformation. Therefore, the present research had several primary aims. First, this work tested whether manipulations of source expertise (Studies 1 and 2) and trustworthiness (Study 2) impact the believability of a retraction, an effect not directly tested by past CIE research examining the impacts of these source perceptions on continued misinformation belief and use (Ecker & Antonio, 2021; Guillory & Geraci, 2013; Pluviano et al., 2020). Second, this work examined whether effects of expertise on retraction believability would only emerge when the source has no vested interest. Third, this work examined whether these effects on retraction believability might mediate the effects of source characteristics on continued belief in and use of the misinformation. Study 1 Study 1 was designed to examine whether expertise of a source of a retraction only impacts that retraction’s effectiveness when the source has no vested interest. Participants received a retraction from a source who either had experience that qualified him to provide the retraction or a source who lacked this experience. This manipulation was designed to mainly impact perceptions of source expertise. The source also either had or did not have a vested interest. The primary prediction for this study was that vested interest would interact with source experience such that the experienced source’s retraction would be rated as more believable, lead to less endorsement of the misinformation being true, and lead to less use of the misinformation when participants were asked to make misinformation-relevant inferences primarily when the retraction source did not have a vested interest. Effects on misinformation endorsement and misinformation-based inferences were expected to be mediated by perceptions of retraction believability. Retractions from the experienced source would be seen as more believable and greater retraction believability would lead to greater endorsement of and reliance on the misinformation, but only when the source had no vested interest. Therefore, an overall moderated mediation pattern was expected. Method Participants Participants were 226 Mechanical Turk workers (58% women) who participated in exchange for monetary compensation. We based our decision of how many participants to recruit for this study on past research within the CIE domain. Specifically, past work has often employed samples large enough to have around 20 participants per cell of the experimental design (e.g., Ecker et al., 2011a, b, 2014, 2015). We decided to recruit a larger sample for the present study to bolster statistical power. Sensitivity analyses We conducted several sensitivity analyses to examine whether, given different levels of assumed power, our studies were likely sufficiently sensitive to detect the primary effects of interest. Prior research examining effects of source features on continued influences of misinformation have generally observed effect sizes medium to large in magnitude (Ecker & Antonio, 2021; Guillory & Geraci, 2013; Pluviano et al., 2020). We used GPower (Faul et al., 2007) to conduct several sensitivity analyses for multiple regressions. Sample sizes collected in Study 1 or 2 were included along with one of three levels of assumed power: 0.8, 0.9, or 0.95. Given Study 1’s sample size of 226, such designs would be sensitive to a Cohen’s f2 = 0.035 at an assumed power of 0.8, f2 = 0.047 at an assumed power of 0.9, and f2 = 0.058 at an assumed power of 0.95, all of which would generally be considered relatively small effect sizes (Selya et al., 2012). Likewise, using Study 2’s sample size of 353, such designs would be sensitive to a Cohen’s f2 = 0.022 at an assumed power of 0.8, f2 = 0.030 at an assumed power of 0.9, and f2 = 0.037 at an assumed power of 0.95, which would again all be considered to be relatively small effects. As such, these sensitivity analyses suggest our designs were likely sufficiently sensitive to detect our effects of interest, even if those effects were notably smaller than those observed in prior studies examining the CIE and source features. Design This study used a 2 (Source Experience: inexperienced vs. experienced) × 2 (Vested Interest: no vested interest vs. vested interest) between-subjects design. There was a minimum of 56 participants and a maximum of 58 in each cell of the design (see the Online Supplemental Material (OSM) for a full breakdown the number of participants within each cell). Measures and manipulations Source experience and vested interest manipulations The messages used were adapted from those used by Wilkes and Leatherbarrow (1988) and pertained to a warehouse fire that started at a paper company. Four versions of the message were created crossing retraction source experience with source vested interest (see the OSM for the full text of these messages). The messages were presented as a series of statements. In all messages, Statement 4 stated that a short circuit in a side room started the fire. In Statement 5, participants were told that the Fire Marshal received reports that combustible materials, such as paint and gas cylinders, had been carelessly stored in a side room prior to the start of the fire. This statement also suggested that this indicates the paper company might be liable for the damages caused by the fire. The presence of these combustible materials constituted the misinformation in this study. After several filler statements about the fire, Statement 14 contained a retraction of the misinformation. Two versions of the retraction were created: one attributed to an inexperienced source and one attributed to an experienced source. In both conditions, Statement 8 indicated that no records were kept regarding the contents of the side room before the fire, so determining whether combustible materials were stored in the side room would rely on examinations of forensic evidence. In the experienced retraction source condition, Statement 12 stated that John Anderson, the police investigator who would be examining this evidence, had been on the police force for over 20 years. Participants were also told that he had investigated the causes of hundreds of previous fires and had specialist training in advanced forensic techniques. In the inexperienced retraction source condition, participants were told that John Anderson had been on the police force for less than a year and that this was his first investigation into the cause of a fire. To manipulate vested interest, extra information was included or excluded from Statement 12 regarding Investigator Anderson. Those in the vested interest condition were told that Anderson owns a significant amount of stock in the company where the fire occurred and that, if combustible materials were stored in the side room, the company’s stock values would decrease dramatically. This information was omitted for those in the no vested interest condition. Source perception measures Perceptions of both source expertise and source trustworthiness were assessed. Perceived expertise was measured using three items asking how qualified, knowledgeable, and expert the source was about whether combustible materials had been stored in the side room. An example item is: “To what degree did you think Investigator Anderson was qualified to discuss the possible presence of combustible materials in the side room?” (1 – Not at all Qualified to 7 – Very Qualified) (α = 0.90). To assess trustworthiness, participants were asked to rate how honest (1 – Not at all Honest to 7 – Very Honest), sincere (1 – Not at all Sincere to 7 – Very Sincere), and trustworthy (1 – Not at all Trustworthy to 7 – Very Trustworthy) they found the source when he asserted that no combustible materials were present in the side room at the time of the fire (α = 0.97). Retraction believability measure Participants were asked the extent to which they believed Investigator Anderson’s assertion that there is no evidence that combustible materials were present in the side room at the time of the fire (1 – I Did Not Believe His Assertion at All to 7 – I Believed His Assertion Completely), how true they thought the assertion was (1 – I Did Not Think His Assertion Was True at All to 7 – I Thought His Assertion Was Very True), how accurate they thought it was (1 – I Did Not Think His Assertion Was Accurate at All to 7 – I Thought His Assertion Was Very Accurate), and how credible they thought it was (1 – I Did Not Find His Assertion to be Credible at All to 7 – I Found His Assertion to be Very Credible). Reliability between these items was high (α = 0.98). Misinformation endorsement measure Three items were employed to assess the extent to which participants believed the misinformation to be true. An example item is: “To what extent do you agree with the following statement: combustible materials stored in the side room contributed to the fire?” (1 – Strongly Disagree to 7 – Strongly Agree). Reliability between items was good (α = 0.98). Number of misinformation-based inferences measure Nine open-ended questions used by Wilkes and Leatherbarrow (1988) were employed to measure the extent to which participants would base inferences about the warehouse fire event on the misinformation. This measure is intended to capture how much the misinformation is continuing to guide participant’s reasoning about the event. Sample questions are: “Why do you think the fire was particularly intense?” and “What could have caused the explosions?” Responses to these questions were coded by two independent coders. The coders categorized whether each response directly or indirectly referenced the misinformation. A response that referenced the misinformation was coded as a 1, whereas a response that made no reference was coded as a 0. If a participant referenced the misinformation but also acknowledged that it is false within a response, that response was coded as a 0. Because there was a fair amount of discrepancy between the coders’ ratings (inter-rater correlation: r = 0.73), a third, independent coder examined the responses to which the original two coders had disagreed and provided a judgment to resolve the discrepancy. In cases where the response was too ambiguous to resolve whether it did or did not reference the misinformation, a code of 0.5 was given. Codes to all nine items were summed for each participant, resulting in an overall measure of how much participants continued to rely on the misinformation to inform their inferences. Procedure After providing informed consent, participants were randomly assigned to read one of the four messages, assigning them to either the high or low experience and high or low vested interest conditions. After this, participants completed the source perception measures followed by the retraction believability measure, the misinformation endorsement measure, and the misinformation-based inference measure.1 Lastly, participants were debriefed and thanked for their participation. Results All study data and analysis syntax are available here: https://osf.io/p6gry/?view_only=26fffc39bf1841d1ad90e364188db151. Source perceptions Two multiple regressions were conducted to examine how the source experience (-1 = inexperienced, 1 = experienced) and vested interest (-1 = no vested interest, 1 = vested interest) manipulations impacted perceptions of expertise and perceptions of trustworthiness respectively. Regarding expertise, there was a main effect of experience such that those who read about an experienced source perceived the source to be significantly more of an expert than those who read about an inexperienced source (see Table 1). There was no main effect of the vested interest manipulation. Unexpectedly, there was a significant interaction between the experience and vested interest manipulations such that the impact of the experience manipulation was larger when the source had no vested interest, b = 1.29, se = 0.14, t(222) = 9.55, p < 0.001, 95% confidence interval (CI) [1.027, 1.562], r = 0.54, than when the source had a vested interest, b = 0.80, se = 0.13, t(222) = 5.95, p < 0.001, 95% CI [0.535, 1.065], r = 0.37.Table 1 Study 1 results from multiple regressions predicting perceptions of retraction source expertise and trustworthiness Variable Outcome b se df t p 95% CI lower bound 95% CI upper bound r Intercept Expertise 4.77 0.10 222 49.96 <0.001 4.583 4.959 0.96 Trustworthiness 4.69 0.09 222 49.68 <0.001 4.507 4.879 0.96 Experience Manipulation (EM) Expertise 1.05 0.10 222 10.97 <0.001 0.859 1.236 0.59 Trustworthiness 0.40 0.09 222 4.25 <0.001 0.215 0.587 0.27 Vested Interest Manipulation (VM) Expertise -0.06 0.10 222 -0.62 0.533 -0.248 0.129 0.04 Trustworthiness -1.18 0.09 222 -12.45 <0.001 -1.362 -0.990 0.64 EMVM Expertise -0.25 0.10 222 -2.59 0.010 -0.435 -0.059 0.17 Trustworthiness -0.02 0.09 222 -0.23 0.821 -0.208 0.165 0.02 Regarding trustworthiness, there was a main effect of experience such that the experienced source was perceived to be significantly more trustworthy than the inexperienced source (see Table 1). There was also a main effect of the vested interest manipulation such that the source with a vested interest was perceived to be significantly less trustworthy than the source without a vested interest. The interaction between experience and vested interest was not significant. Because the source experience manipulation impacted perceptions of trustworthiness in addition to expertise, we controlled for trustworthiness and its interaction with the vested interest manipulation when examining the impact of the experience manipulation on our outcome measures by including these factors as covariates. We did this to ensure that effects of the expertise manipulation were not attributable to its impacts on trustworthiness (or any possible interactions between trustworthiness and the vested interest manipulation). Retraction believability Predicting retraction believability, the predicted interaction between the experience and vested interest manipulations emerged, b = -0.30, se = 0.07, t(220) = -4.22, p < 0.001, 95% CI [-0.439, -0.160], r = 0.27. When the source did not have a vested interest, retractions from the inexperienced source were less believable than retractions from the experienced source, b = 0.40, se = 0.10, t(220) = 3.94, p < 0.001, 95% CI [0.202, 0.607], r = 0.26. When the source had a vested interest, this effect was actually reversed, such that retractions from the inexperienced source were seen as more believable than retractions from the experienced source (see Fig. 1), b = -0.19, se = 0.10, t(220) = -1.99, p = 0.048, 95% CI [-0.387, -0.002], r = 0.13. There was no main effect of experience, b = 0.10, se = 0.07, t(220) = 1.48, p = 0.14, 95% CI [-0.035, 0.245], r = 0.10, but, unexpectedly, there was a significant main effect of vested interest such that retractions from sources without a vested interest were rated as less believable than retractions from sources with a vested interest, b = 0.19, se = 0.09, t(220) = 2.04, p = 0.042, 95% CI [0.007, 0.366], r = 0.14. Because this effect was not replicated in Study 2, we hesitate to place meaning on it.Fig. 1 Study 1 estimated marginal mean values of retraction believability as a function of source experience and vested interest. Note. Perceptions of source trustworthiness and the interaction between trustworthiness and vested interest were controlled for in this analysis. Bars represent means and error bars represent 95% confidence intervals Misinformation endorsement The predicted interaction between experience and vested interest was significant, b = 0.24, se = 0.11, t(220) = 2.19, p = 0.030, 95% CI [0.024, 0.460], r = 0.15. When the source did not have a vested interest, those who saw a retraction from the experienced source endorsed the misinformation significantly less than those who saw a retraction from an inexperienced source, b = -0.38, se = 0.16, t(220) = -2.37, p = 0.019, 95% CI [-0.694, -0.063], r = 0.16. There was no significant effect of experience when the source had a vested interest (see Fig. 2), b = 0.11, se = 0.15, t(220) = 0.69, p = 0.49, 95% CI [-0.195, 0.405], r = 0.05. There were no significant main effects of either experience, b = -0.14, se = 0.11, t(220) = -1.24, p = 0.22, 95% CI [-0.355, 0.081], r = 0.08, or vested interest, b = 0.06, se = 0.14, t(220) = 0.39, p = 0.70, 95% CI [-0.224, 0.336], r = 0.03.Fig. 2 Study 1 estimated marginal mean values of misinformation endorsement and number of misinformation-based inferences as a function of retraction source experience and vested interest. Note. Perceptions of source trustworthiness and the interaction between trustworthiness and vested interest were controlled for in this analysis. Bars represent means and error bars represent 95% confidence intervals Number of misinformation-based inferences Misinformation endorsement was significantly correlated with the number of misinformation-based inferences, r(224) = 0.53, p < 0.001, but this correlation was not strong enough to conclude that these two measures are redundant. The interaction between experience and vested interest was significant in the predicted direction, b = 0.35, se = 0.12, t(220) = 2.88, p = 0.004, 95% CI [0.111, 0.589], r = 0.19. When the source lacked a vested interest, there was a non-significant tendency for participants to make fewer misinformation-based inferences when the retraction came from an experienced versus inexperienced source, b = -0.27, se = 0.18, t(220) = -1.52, p = 0.13, 95% CI [-0.613, -0.079], r = 0.10. When the source had a vested interest, the effect actually reversed such that participants made significantly more misinformation-based inferences when the retraction came from an experienced versus inexperienced source, b = 0.43, se = 0.17, t(220) = 2.59, p = 0.010, 95% CI [0.103, 0.762], r = 0.17. There was no main effect of either experience, b = 0.08, se = 0.12, t(220) = 0.68, p = 0.50, 95% CI [-0.156, 0.322], r = 0.05, or vested interest, b = 0.16, se = 0.16, t(220) = 1.01, p = 0.32, 95% CI [-0.150, 0.464], r = 0.07. Mediation analyses To examine whether retraction believability might mediate the effects of the experience manipulation on misinformation endorsement and the number of misinformation-based inferences, two moderated mediation analyses were conducted. In the first model, the experience manipulation was included as the focal predictor, retraction believability as the mediating variable, and misinformation endorsement as the outcome variable. The vested interest manipulation was included as a moderator of the paths between the experience manipulation and retraction believability and the path between the experience manipulation and misinformation endorsement. Perceptions of trustworthiness and their interaction with vested interest were included as covariates. Hayes’ (2017) Process Macro was used to calculate the conditional indirect effects at high and low vested interest. We used 5,000 bootstrapped samples to test their significance and to determine whether these conditional indirect effects significantly differed from each other. When the source had no vested interest, there was a significant indirect effect of experience on misinformation endorsement through retraction believability, ab = -0.30, se = 0.11, 95% CI [-0.542, -0.094], such that the retraction from the experienced source was seen as more believable than retractions from the inexperienced source, a = 0.40, se = 0.10, t(220) = 3.94, p < 0.001, 95% CI [0.202, 0.607], r = 0.26, and retraction believability negatively predicted misinformation endorsement, b = -0.74, se = 0.09, t(219) = -7.91, p < 0.001, 95% CI [-0.918, -0.552], r = 0.47. When the source had a vested interest, the indirect effect was unexpectedly significant in the opposite direction, ab = 0.14, se = 0.06, 95% CI [0.037, 0.265]. This was because retractions from the inexpert source were now seen as more believable than retractions from the expert source (see Fig. 3), a = -0.19, se = 0.10, t(220) = -1.99, p = 0.048, 95% CI [-0.387, -0.002], r = 0.13. A significant index of moderated mediation indicated that these two conditional indirect effects were significantly different from each other, index = 0.44, se = 0.13, 95% CI [0.201, 0.722]. The direct effect of the Experience × Vested Interest interaction on misinformation endorsement was not significant, b = 0.02, se = 0.10, t(219) = 0.21, p = 0.83, 95% CI [-0.178, 0.222], r = 0.01.Fig. 3 Study 1 moderated mediation analysis predicting misinformation endorsement from source experience through retraction believability moderated by source vested interest. Note. Perceptions of source trustworthiness and the interaction between trustworthiness and vested interest were included as covariates within each regression within this moderated mediation analysis. Unstandardized regression coefficients appear next to the respective mediation paths, with a, b, and c’ denoting the regression coefficients for the respective paths. Conditional coefficients are reported for moderated paths The second model was constructed identically to the first except that number of misinformation-based inferences was the outcome variable. When the source did not have a vested interest, the indirect effect of experience on the number of misinformation-based inferences through retraction believability was significant, ab = -0.14, se = 0.07, 95% CI [-0.300, -0.030]. As in the first model, the retraction from the expert source was seen as more believable, and retraction believability negatively predicted the number of misinformation-based inferences, b = -0.35, se = 0.11, t(219) = -3.07, p = 0.002, 95% CI [-0.571, -0.125], r = 0.20. As in the first model, the conditional indirect effect was in the opposite direction when the source had a vested interest, ab = 0.07, se = 0.03, 95% CI [0.012, 0.144], because retractions from the inexperienced source were now seen as more believable. These two indirect effects were significantly different from one another, index = 0.21, se = 0.09, 95% CI [0.062, 0.406]. The direct effect of the Experience × Vested Interest interaction remained significant, b = 0.25, se = 0.12, t(219) = 1.98, p = 0.048, 95% CI [0.002, 0.489], r = 0.11. There was no significant effect of expertise at low vested interest, b = -0.13, se = 0.18, t(219) = -0.71, p = 0.48, 95% CI [-0.478, 0.225], r = 0.04, but when vested interest was high, those who saw a retraction from the experienced (vs. inexperienced) source made significantly more misinformation-based inferences, b = 0.36, se = 0.17, t(219) = 2.20, p = 0.029, 95% CI [0.038, 0.691], r = 0.12. Study 1 Discussion The results of Study 1 suggest retractions of misinformation are more believable when they come from an expert versus inexpert source, but only when the source has no vested interest. Additionally, retractions from expert versus inexpert sources led to less misinformation endorsement and had a similar trending effect on the number of misinformation-based inferences, but this was again only when the source had no vested interest. Perceptions of retraction believability mediated these latter effects. Importantly, these effects emerged while controlling for perceptions of trustworthiness and its interaction with vested interest, indicating that differential perceptions of trustworthiness did not account for these effects. Unexpectedly, the presence of a vested interest led the source experience manipulation to backfire on perceptions of retraction believability and the number of misinformation-based inferences. We are hesitant to place much importance in these unexpected findings, but it is possible that people would be especially disappointed in an expert source who is perceived to be lying in service of their own self-interest. Put together, these results indicate that, when people judge whether to believe a retraction or not, information about the source’s expertise might matter most when the source has no vested interest in issuing the retraction. When the source has a vested interest, their expertise seems to matter less. This could be because vested interest cues make people think more about whether the retraction is an attempt to deceive, and this might overshadow considerations of how qualified the source is to issue the retraction. These results support the idea that perceptions of source vested interest are necessary to consider when determining whether source expertise will have an impact on retraction effectiveness. A limitation of this study was that the experience manipulation influenced perceptions of trustworthiness, requiring those perceptions to be statistically controlled in our analyses. It would be beneficial to experimentally manipulate trustworthiness in addition to expertise to achieve greater control of both perceptions and more precisely examine their impacts. Study 2 Study 2 was designed to replicate the overall pattern of results observed in Study 1 using direct manipulations of source expertise and trustworthiness. As in Study 1, we predicted that retractions from the high expertise source would be seen as more believable, and subsequent belief in and use of the misinformation would be lower, than when a less expert source issued the retraction. Critically, we predicted that this would only be the case when the source did not have a vested interest in issuing the retraction. Additionally, we expected that retractions from a more trustworthy source would be seen as more believable and lead to less continued belief in and use of the misinformation than retractions from a less trustworthy source. Importantly, we predicted that this effect would remain even when the source had a vested interest in issuing the retraction. Method Participants Three-hundred and fifty-three undergraduates (54.4% men, 44.2% women, 0.6% other, 0.3% non-binary) were recruited to participate in this study in exchange for course credit. Design This study used a 2 (Source Expertise: low vs. high) × 2 (Source Trustworthiness: low vs. high) × 2 (Vested Interest: no vested interest vs. vested interest) between-subjects design. There was a minimum of 42 participants and a maximum of 46 in each cell of the design (see the OSM for a full breakdown the number of participants within each cell). Measures and manipulations Source expertise manipulation The same source experience manipulation used in Study 1 was used to manipulate source expertise in this study. Source trustworthiness manipulation Trustworthiness was manipulated by providing additional details about Investigator Anderson immediately following the source expertise information. Those in the high trustworthiness condition were told that Investigator Anderson is known to be a man of integrity who always tells the truth, even when doing so could be personally costly. Those in the low trustworthiness condition were told that he is a man of questionable integrity and that there are rumors he will sometimes lie, especially if he stands to personally benefit from doing so. Vested interest manipulation The vested interest manipulation was largely the same as that used in Study 1 with several minor wording changes (see OSM for exact wording). Source perception measures The measure of source expertise was identical to that used in Study 1 (α = 0.79). The measures used to assess trustworthiness were also largely the same, although the item asking about trustworthiness was replaced with an item asking participants the degree to which they believed Investigator Anderson was saying what he truly believed (7-point scale: 1 – Not at all to 7 – Very Much; α = 0.93). Measures of perceived vested interest were also included. These items asked to what extent the presence or absence of combustible materials in the side room could have personal implications for Investigator Anderson, if it could impact him financially, and if it has implications for him outside of his job (7-point scales: 1 – Not at all to 7 – Very Much; α = 0.85). Retraction believability measure This measure was identical to the one used in Study 1 (α = 0.94). Misinformation endorsement and inference measures The misinformation endorsement (α = 0.90) and inference measures were the same as those used in Study 1. Procedure After providing informed consent, participants were randomly assigned to read the same warehouse fire message used in Study 1 but with either the high or low expertise, trustworthiness, and vested interest information. After reading the randomly assigned message, participants responded to the source perception measures followed by the retraction believability, misinformation endorsement, and inference measures. Finally, participants were debriefed and thanked for their participation. Results Source perceptions Three multiple regressions were constructed predicting each of the three measured source perceptions from the expertise, trustworthiness, and vested interest manipulations as well as the interactions between the manipulations of expertise and vested interest, trustworthiness and vested interest, and expertise and trustworthiness. Regarding perceptions of source expertise, the source expertise manipulation significantly impacted perceptions of expertise such that the high expertise source was rated as being more expert than the low expertise source (see Table 2). The trustworthiness manipulation also significantly impacted perceptions of expertise such that the more trustworthy source was perceived to be more expert than the less trustworthy source. The vested interest manipulation did not have a significant effect.Table 2 Study 2 results from multiple regression predicting perceptions of retraction source expertise Variable b se df T p 95% CI lower bound 95% CI upper bound r Intercept 4.45 0.07 346 66.39 < 0.001 4.320 4.584 0.96 Expertise Manipulation (EM) 0.48 0.07 346 7.11 < 0.001 0.345 0.609 0.36 Trustworthiness Manipulation (EM) 0.18 0.07 346 2.64 0.009 0.045 0.309 0.14 Vested Interest Manipulation (VM) 0.05 0.07 346 0.77 0.442 -0.080 0.184 0.04 EMTM 0.05 0.07 346 0.70 0.487 -0.085 0.179 0.04 EMVM -0.10 0.07 346 -1.54 0.124 -0.235 0.028 0.08 TMVM -0.13 0.07 346 -1.89 0.059 -0.259 0.005 0.10 Regarding perceptions of source trustworthiness, the more trustworthy source was perceived to be more trustworthy than the less trustworthy source (see Table 3). The source without a vested interest was also perceived to be more trustworthy than the source with a vested interest. Interestingly, a significant interaction between trustworthiness and vested interest emerged such that the effect of the trustworthiness manipulation was stronger when the source had no vested interest, b = 0.76, se = 0.10, t(346) = 7.35, p < 0.001, 95% CI [0.555, 0.961], r = 0.37, than when the source had a vested interest, b = 0.40, se = 0.10, t(346) = 3.90, p < 0.001, 95% CI [0.198, 0.600], r = 0.21. The expertise manipulation did not have an impact.Table 3 Study 2 results from multiple regression predicting perceptions of retraction source trustworthiness Variable b se df t p 95% CI lower bound 95% CI upper bound r Intercept 3.93 0.07 346 54.14 < 0.001 3.788 4.074 0.95 Expertise Manipulation (EM) 0.08 0.07 346 1.11 0.266 -0.062 0.224 0.06 Trustworthiness Manipulation (EM) 0.58 0.07 346 7.97 < 0.001 0.436 0.721 0.39 Vested Interest Manipulation (VM) -0.43 0.07 346 -5.92 < 0.001 -0.573 -0.287 0.30 EMTM -0.02 0.07 346 -0.27 0.788 -0.162 0.123 0.01 EMVM -0.13 0.07 346 -1.75 0.081 -0.270 0.016 0.09 TMVM -0.18 0.07 346 -2.47 0.014 -0.322 -0.037 0.13 Regarding perceptions of source vested interest, the source with a vested interest was perceived to have significantly more of a vested interest in issuing a retraction than the source without a vested interest (see Table 4). The untrustworthy source was also perceived to have more of a vested interest than the trustworthy source. The expertise manipulation had no effect.Table 4 Study 2 results from multiple regression predicting perceptions of retraction source vested interest Variable b se df t p 95% CI lower bound 95% CI upper bound r Intercept 4.25 0.07 346 58.59 < 0.001 4.108 4.394 0.95 Expertise Manipulation (EM) 0.00 0.07 346 -0.02 0.984 -0.144 0.141 < 0.001 Trustworthiness Manipulation (EM) -0.21 0.07 346 -2.93 0.004 -0.355 -0.070 0.16 Vested Interest Manipulation (VM) 0.80 0.07 346 11.04 < 0.001 0.659 0.944 0.51 EMTM 0.00 0.07 346 0.004 0.997 -0.142 0.143 < 0.001 EMVM 0.10 0.07 346 1.36 0.174 -0.044 0.241 0.07 TMVM 0.14 0.07 346 1.94 0.053 -0.002 0.283 0.10 Retraction believability Another multiple regression with the same predictors was used to predict retraction believability. There were significant main effects of expertise, trustworthiness, and vested interest, such that retractions from the expert, trustworthy, or low vested interest sources were rated as more believable than retractions from the inexpert, untrustworthy, or high vested interest sources (see Table 5). Critically, the predicted Expertise × Vested Interest interaction was significant. Replicating Study 1, a significant effect of expertise emerged when vested interest was low, b = 0.34, se = 0.10, t(346) = 3.32, p = 0.001, 95% CI [0.137, 0.534], r = 0.18, but not when vested interest was high (see Fig. 4), b = -0.02, se = 0.10, t(346) = -0.16, p = 0.87, 95% CI [-0.213, 0.181], r = 0.01. There was also a significant Trustworthiness × Vested Interest interaction such that the effect of trustworthiness was stronger when the source did not have a vested interest, b = 0.62, se = 0.10, t(346) = 6.16, p < 0.001, 95% CI [0.424, 0.821], r = 0.31, than when they had a vested interest, b = 0.30, se = 0.10, t(346) = 3.02, p = 0.003, 95% CI [0.106, 0.500], r = 0.16.Table 5 Study 2 results from multiple regression predicting retraction believability Variable b se df t P 95% CI lower bound 95% CI upper bound r Intercept 3.73 0.07 346 52.43 < 0.001 3.589 3.869 0.94 Expertise Manipulation (EM) 0.16 0.07 346 2.24 0.026 0.020 0.300 0.12 Trustworthiness Manipulation (EM) 0.46 0.07 346 6.50 < 0.001 0.323 0.602 0.33 Vested Interest Manipulation (VM) -0.29 0.07 346 -4.09 < 0.001 -0.431 -0.151 0.21 EMTM 0.09 0.07 346 1.24 0.216 -0.052 0.228 0.07 EMVM -0.18 0.07 346 -2.47 0.014 -0.316 -0.036 0.13 TMVM -0.16 0.07 346 -2.25 0.025 -0.300 -0.020 0.12 Fig. 4 Study 2 estimated marginal mean values of retraction believability as a function of source expertise and vested interest. Note. The main effect of the trustworthiness manipulation and interactions between the trustworthiness and vested interest manipulations and the expertise and trustworthiness manipulations were also included in this analysis. Bars represent means and error bars represent 95% confidence intervals Misinformation endorsement A multiple regression identical to the one used to predict retraction believability was created to predict misinformation endorsement. There were significant main effects of expertise, trustworthiness, and vested interest such that participants endorsed the misinformation less following a retraction from an expert, trustworthy, or low vested interest source than one from an inexpert, untrustworthy, or high vested interest source (see Table 6). Importantly, the predicted Expertise × Vested Interest manipulation was significant such that the effect of expertise was significant when the source had no vested interest, b = -0.35, se = 0.10, t(346) = -3.54, p < 0.001, 95% CI [-0.540, -0.154], r = 0.19, but not when the source had a vested interest (see Fig. 5), b = 0.06, se = 0.10, t(346) = 0.59, p = 0.56, 95% CI [-0.134, 0.249], r = 0.03. There was also a significant Trustworthiness × Vested Interest interaction such that the effect of trustworthiness was significant when the source did not have a vested interest, b = -0.37, se = 0.10, t(346) = -3.76, p < 0.001, 95% CI [-0.562, -0.176], r = 0.20, but not when the source had a vested interest, b = -0.04, se = 0.10, t(346) = -0.45, p = 0.66, 95% CI [-0.235, 0.148], r = 0.02.Table 6 Study 2 results from multiple regression predicting misinformation endorsement Variable b se df t p 95% CI lower bound 95% CI upper bound r Intercept 4.90 0.07 346 70.97 < 0.001 4.767 5.039 0.97 Expertise Manipulation (EM) -0.14 0.07 346 -2.10 0.037 -0.281 -0.009 0.11 Trustworthiness Manipulation (EM) -0.21 0.07 346 -2.99 0.003 -0.342 -0.071 0.16 Vested Interest Manipulation (VM) 0.19 0.07 346 2.68 0.008 0.050 0.321 0.14 EMTM -0.04 0.07 346 -0.53 0.594 -0.173 0.099 0.03 EMVM 0.20 0.07 346 2.93 0.004 0.066 0.338 0.16 TMVM 0.16 0.07 346 2.36 0.019 0.027 0.299 0.13 Fig. 5 Study 2 estimated marginal mean values of misinformation endorsement and the number of misinformation-based inferences as a function of source expertise and vested interest. Note. The main effect of the trustworthiness manipulation and interactions between the trustworthiness and vested interest manipulations and the expertise and trustworthiness manipulations were also included in this analysis. Bars represent means and error bars represent 95% confidence intervals Number of misinformation-based inferences Misinformation endorsement was significantly positively correlated with inferences, r(351) = 0.33, p < 0.001, but not to a degree to which these measures could be considered redundant. An identical multiple regression as that used to predict retraction believability and misinformation endorsement was also created to predict the number of misinformation-based inferences participants made. As opposed to what was observed with misinformation endorsement, there were no significant main effects of expertise or vested interest (see Table 7). There was a significant main effect of trustworthiness, however, such that fewer misinformation-based inferences were made when the retraction came from a trustworthy versus untrustworthy source. Unexpectedly, there was also no significant Expertise × Vested Interest interaction or Trustworthiness × Vested Interest interaction.Table 7 Study 2 results from multiple regression predicting the number of misinformation-based inferences Variable b se df t p 95% CI lower bound 95% CI upper bound r Intercept 4.12 0.11 346 38.46 < 0.001 3.908 4.330 0.90 Expertise Manipulation (EM) -0.06 0.11 346 -0.54 0.589 -0.269 0.153 0.03 Trustworthiness Manipulation (EM) -0.24 0.11 346 -2.19 0.029 -0.446 -0.024 0.12 Vested Interest Manipulation (VM) 0.10 0.11 346 0.95 0.340 -0.108 0.313 0.05 EMTM 0.02 0.11 346 0.17 0.864 -0.192 0.229 0.01 EMVM -0.03 0.11 346 -0.26 0.796 -0.238 0.183 0.01 TMVM 0.06 0.11 346 0.56 0.579 -0.151 0.270 0.03 Mediation analyses As in Study 1 we conducted two moderated mediation analyses. In the first analysis, source expertise served as the focal predictor, retraction believability as the mediator, and misinformation endorsement as the outcome measure. The vested interest manipulation moderated the paths between expertise and retraction believability, and the path between expertise and misinformation endorsement. The trustworthiness manipulation and its interaction with vested interest were also included as covariates. Five thousand bootstrapped samples were again used to calculate the significance of the indirect effects. When the source had no vested interest, there was a significant negative indirect effect of expertise on misinformation endorsement through retraction believability, ab = -0.13, se = 0.05, 95% CI [-0.235, -0.051]. The retraction from the expert source was seen as more believable than the one from the inexpert source, a = 0.34, se = 0.10, t(347) = 3.34, p < 0.001, 95% CI [0.139, 0.537], r = 0.18, and retraction believability negatively predicted misinformation endorsement, b = -0.40, se = 0.05, t(346) = -8.37, p < 0.001, 95% CI [-0.491, -0.304], r = 0.41. When the source had a vested interest, the conditional indirect effect was no longer significant, ab = 0.01, se = 0.04, 95% CI [-0.074, 0.086]. This was because expertise no longer impacted retraction believability in this condition (see Fig. 6), a = -0.02, se = 0.10, t(347) = -0.16, p = 0.87, 95% CI [-0.213, 0.181], r = 0.01. These two indirect effects differed significantly from each other, index = 0.14, se = 0.06, 95% CI [0.029, 0.268]. A direct effect of the Expertise × Vested Interest interaction remained significant, b = 0.13, se = 0.06, t(346) = 2.08, p = 0.038, 95% CI [0.007, 0.257], r = 0.11, such that the effect of expertise was significant when the source had no vested interest, b = -0.21, se = 0.09, t(346) = -2.35, p = 0.019, 95% CI [-0.392, -0.035], r = 0.13, but not when the source had a vested interest, b = 0.05, se = 0.09, t(346) = 0.57, p = 0.57, 95% CI [-0.124, 0.226], r = 0.03.Fig. 6 Study 2 moderated mediation analysis predicting misinformation endorsement from source expertise through retraction believability moderated by source vested interest. Note. The trustworthiness manipulation and its interaction with vested interest were included as covariates within each regression within this moderated mediation analysis. Unstandardized regression coefficients appear next to the respective mediation paths, with a, b, and c’ denoting the regression coefficients for the respective paths. Conditional coefficients are reported for moderated paths Though there were no significant total effects of expertise on inferences, we still examined whether there might be indirect effects on inferences through retraction believability. Therefore, we ran a second mediation analysis identical to the first except that the outcome measure was the number of misinformation-based inferences. When the source had no vested interest, there was a significant indirect effect, ab = -0.14, se = 0.05, 95% CI [-0.237, -0.054]. Retractions from the expert source were seen as more believable than those from the inexpert source, and retraction believability negatively predicted the number of misinformation-based inferences, b = -0.41, se = 0.08, t(346) = -5.22, p < 0.001, 95% CI [-0.559, -0.253], r = 27. The indirect effect was no longer significant when the source had a vested interest, ab = 0.01, se = 0.04, 95% CI [-0.080, 0.087], because expertise did not impact retraction believability in this condition. These indirect effects differed significantly from each other, index = 0.14, se = 0.06, 95% CI [0.028, 0.276]. There was no significant direct effect of the Expertise × Vested Interest interaction, b = -0.10, se = 0.10, t(346) = -0.96, p = 0.34, 95% CI [-0.304, 0.105], r = 0.05. Study 2 Discussion The effect of expertise on retraction believability and misinformation endorsement emerged as predicted. Retractions from the expert source were seen as more believable and led to less continued endorsement of the misinformation than retractions from the inexpert source. Importantly, this only occurred when the retraction source had no vested interest. No significant effects of expertise emerged when predicting the number of misinformation-based inferences. As discussed earlier, this might have been because a retraction source’s characteristics are more closely linked to belief in the misinformation than its use to make inferences. If inferences are impacted by factors and other relevant information beyond misinformation belief, effects of expertise might not carry over to inferential reasoning. An additional, unexpected finding was that the effect of the trustworthiness manipulation on retraction believability was reduced, and its effect on misinformation endorsement became non-significant, when the source had a vested interest compared to when the source did not have a vested interest. One possible explanation for this could be because the trustworthiness manipulation was less impactful on perceptions of trustworthiness in the presence of vested interest information. Participants might have used this information in conjunction with the trustworthiness information to judge the source’s trustworthiness, which could have diluted the impact of the trustworthiness manipulation. As such, that manipulation could be expected to have less of an impact when vested interest information was present. Therefore, these findings likely do not suggest that trustworthiness per se is less important when a source has a vested interest, but rather that information about a source independently pertaining to their trustworthiness is less impactful when information about vested interest is also present. General discussion This research suggests that both source trustworthiness and source expertise can independently impact perceptions of retraction believability, and that those perceptions predict continued belief in the misinformation and its use to inform inferences. Additionally, source expertise only predicted retraction believability when the source had no vested interest. These findings suggest that past research was too quick to dismiss the consequentiality of a retraction source’s expertise in the CIE context (Ecker & Antonio, 2021; Guillory & Geraci, 2013; Pluviano et al., 2020; Swire & Ecker, 2018; Swire et al., 2017). Indeed, a casual assessment of the current CIE research on this topic would lead one to conclude that source expertise is meaningless in the CIE, and that sources retracting misinformation can safely ignore how expert they appear. However, the present research offers a critical clarification to research on this topic. Source expertise does matter, specifically when the retracting source has no self-interest in issuing the retraction. Because retracting sources, such as independent fact-checkers, often do not have such self-interest, the present findings suggest that it is quite important for such sources to emphasize their expertise. Additionally, this research suggests that there might be utility in using multiple outcome measures to assess continued influence of misinformation. Namely, the present research suggests that some effects might appear more consistently when examining continued belief in misinformation than when examining misinformation-based inferential reasoning. Given that both outcomes are important, it is possible that consequential effects might sometimes be missed if one were to focus on one outcome at the exclusion of the other. The present findings also illustrate the importance of considering recipients’ perceptions of a retraction’s believability. Past CIE theorizing has largely focused on processes that occur after the retraction is accepted (Johnson & Seifert, 1994; Lewandowsky et al., 2012; Swire & Ecker, 2018; Wilkes & Leatherbarrow, 1988). However, consistent with the findings of O’Rear and Radvansky (2019), the present data suggest that people do not always believe the retraction, and that belief in the retraction is an important predictor of continued belief in the misinformation and its use to inform inferences. Importantly, the present research suggests that both perceptions of expertise and trustworthiness can impact retraction believability. Therefore, future theorizing should consider the conditions under which a retraction should or should not be believed. More broadly, the present research suggests that the impacts of trustworthiness and expertise on perceptions of credibility might be more nuanced than currently theorized (Cooper et al., 2016; McGuire, 1985; Petty & Cacioppo, 1981; Petty & Wegener, 1998). Specifically, it is possible that factors such as vested interest can shift emphasis away from one perception (e.g., expertise) and place more emphasis on the other perception (e.g., trustworthiness). Future research should examine other factors besides vested interest that could play a similar role. One such factor could be perceptions of source bias (Wallace et al., 2020). Though vested interest has been linked to perceptions of source bias (Wallace et al., 2020), no research to date has examined how source bias itself might raise questions in the minds of source perceivers that differentially focus them on some source perceptions, such as trustworthiness, at the expense of other perceptions, such as expertise. A potential limitation of the present studies’ methods was that vested interest was manipulated by including or withholding information suggesting that the reaction source has a relevant vested interest. A possible issue with this approach is that participants in the vested interest condition received more information than those in the no vested interest, creating an inequality between conditions. That said, we do not believe this inequality was meaningful. It is unlikely that participants in the vested interest condition became notably more fatigued reading a couple of additional sentences than those in the no vested interest conditions, and it is unlikely that the vested interest information had other meaningful impacts on participants other than leading them to believe the source had a vested interest. Nevertheless, future research could use a no vested interest condition that contains neutral, filler information about the source in an amount equal to the vested interest information to rule out this potential issue. The present research has several notable real-world implications. Because a retracting source’s expertise, in addition to trustworthiness, can impact how effective the retraction is at reducing continued belief in misinformation, sources attempting to retract misinformation should generally try to emphasize credentials documenting their trustworthiness and expertise. However, whether one has a vested interest in issuing a retraction might also be important. In cases where a person with a vested interest must be the one issuing the retraction, they might be most successful when emphasizing their trustworthiness over their expertise. This might be particularly important in the political realm, where politicians regularly have a vested interest in correcting negative misinformation about themselves, other members of their party, or policy proposals they endorse. Additionally, it is notable that, in both studies, participants indicated post-retraction misinformation endorsement that was near or above the midpoint of the scale, even when the retracting source had high expertise and no vested interest. This suggests that the continued influence of misinformation can be persistent even in cases where retractions are expected to be most effective. As such, it is possible that individual factors, such as source credibility, are not always sufficient to satisfactorily reduce the CIE on their own, and that combining them with other factors known to reduce the CIE might be needed (Ecker et al., 2010). To conclude, these results suggest that retraction source expertise can independently impact whether one sees a retraction as believable, and that this perception predicts continued reliance on misinformation. Importantly, however, these effects might not always occur. Variables such as source vested interest can prevent one from considering a source’s expertise and instead place focus on the source’s motives to issue the retraction. Therefore, by further understanding these additional variables, future theorizing in the CIE domain will be able to more accurately predict when a retraction will be effective at dispelling belief in misinformation. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 29 KB) Author contributions M.W. Susmann designed the reported studies in collaboration with D.T. Wegener. M.W. Susmann conducted all of the data analyses in consultation with D.T. Wegener. M.W. Susmann drafted the manuscript and revised it based on feedback from D.T. Wegener. Data availability All study materials are publicly available in the Online Supplemental Materials for this article. All study data are publicly available here: https://osf.io/p6gry/?view_only=26fffc39bf1841d1ad90e364188db151. Code availability All study analysis syntax is publicly available here: https://osf.io/p6gry/?view_only=26fffc39bf1841d1ad90e364188db151. Declarations Ethics Approval All research within this article received prior ethics approval from the university Institutional Review Board. Consent to Participate All participants provided informed consent prior to taking part in the studies reported in this article. Consent to Publish Not applicable. Conflicts of Interest The authors have no conflicts of interests to disclose. 1 We also assessed whether participants recalled encountering the retraction. However, we had reservations making inferences based on this measure because it is not clear what construct this measure taps. Specifically, this measure could have assessed participants’ memory that there was a retraction, or alternatively their construal of what the retraction meant. For instance, some might not identify a retraction from an inexperienced source as a retraction if they do not think it was believable. Because of these concerns, and because this measure did not moderate any of our primary findings, we chose not to report this measure or any analyses with this measure. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Practices Statement All study materials from the reported studies are available in the OSM for this article. All study data and analysis syntax are available here: https://osf.io/p6gry/?view_only=26fffc39bf1841d1ad90e364188db151. ==== Refs References Cooper J Blackman SJ Keller K The science of attitudes 2016 Routledge Press Eagly AH Wood W Chaiken S Causal inferences about communicators and their effect on opinion change Journal of Personality and Social Psychology 1978 36 4 424 10.1037/0022-3514.36.4.424 Ecker U Antonio L Can you believe it? An investigation into the impact of retraction source credibility on the continued influence effect Memory & Cognition 2021 49 4 631 644 10.3758/s13421-020-01129-y 33452666 Ecker UK Lewandowsky S Tang DT Explicit warnings reduce but do not eliminate the continued influence of misinformation Memory & Cognition 2010 38 8 1087 1100 10.3758/MC.38.8.1087 21156872 Ecker UK Lewandowsky S Apai J Terrorists brought down the plane!—No, actually it was a technical fault: Processing corrections of emotive information The Quarterly Journal of Experimental Psychology 2011 64 2 283 310 10.1080/17470218.2010.497927 20694936 Ecker UK Lewandowsky S Swire B Chang D Correcting false information in memory: Manipulating the strength of misinformation encoding and its retraction Psychonomic Bulletin & Review 2011 18 3 570 578 10.3758/s13423-011-0065-1 21359617 Ecker UK Lewandowsky S Fenton O Martin K Do people keep believing because they want to? Preexisting attitudes and the continued influence of misinformation Memory & Cognition 2014 42 2 292 304 10.3758/s13421-013-0358-x 24005789 Ecker UK Lewandowsky S Cheung CS Maybery MT He did it! She did it! No, she did not! Multiple causal explanations and the continued influence of misinformation Journal of Memory and Language 2015 85 101 115 10.1016/j.jml.2015.09.002 Faul F Erdfelder E Lang A-G Buchner A G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences Behavior Research Methods 2007 39 2 175 191 10.3758/BF03193146 17695343 Guillory JJ Geraci L Correcting erroneous inferences in memory: The role of source credibility Journal of Applied Research in Memory and Cognition 2013 2 4 201 209 10.1016/j.jarmac.2013.10.001 Hayes AF Introduction to mediation, moderation, and conditional process analysis: A regression-based approach 2017 Guilford Press Hovland CI Weiss W The influence of source credibility on communication effectiveness Public Opinion Quarterly 1951 15 4 635 650 10.1086/266350 Islam, M. S., Sarkar, T., Khan, S. H., Kamal, A. H. M., Hasan, S. M., Kabir, A. ... Seale, H. (2020). COVID-19–related infodemic and its impact on public health: A global social media analysis. The American Journal of Tropical Medicine and Hygiene, 103(4), 1621.10.4269/ajtmh.20-0812 Johnson HM Seifert CM Sources of the continued influence effect: When misinformation in memory affects later inferences Journal of Experimental Psychology: Learning, Memory, and Cognition 1994 20 6 1420 1436 10.1037/0278-7393.20.6.1420 Kelman HC Hovland CI " Reinstatement" of the communicator in delayed measurement of opinion change The Journal of Abnormal and Social Psychology 1953 48 3 327 10.1037/h0061861 Lewandowsky S Ecker UK Seifert CM Schwarz N Cook J Misinformation and its correction: Continued influence and successful debiasing Psychological Science in the Public Interest 2012 13 3 106 131 10.1177/1529100612451018 26173286 McGuire W Lindzey G Aronson E Attitudes and attitude change Handbook of social psychology 1985 3 Random House 233 346 O’Rear, A. O., & Radvansky, G. A. (2019). Failure to accept retractions: A contribution to the continued influence effect. Memory & Cognition, 1–18.10.3758/s13421-019-00967-9 Ohanian R The impact of celebrity spokespersons' perceived image on consumers' intention to purchase Journal of Advertising Research 1991 31 1 46 54 Petty RE Cacioppo JT Attitudes and persuasion : Classic and contemporary approaches 1981 Brown Petty RE Wegener DT Gilbert D Fiske S Lindzey G Attitude change: Multiple roles for persuasion variables The handbook of social psychology 1998 4 McGraw-Hill 323 390 Petty RE Cacioppo JT Goldman R Personal involvement as a determinant of argument-based persuasion Journal of Personality and Social Psychology 1981 41 5 847 10.1037/0022-3514.41.5.847 Pluviano, S., Della Sala, S., & Watt, C. (2020). The effects of source expertise and trustworthiness on recollection: The case of vaccine misinformation. Cognitive Processing, 1–10.10.1007/s10339-020-00974-8 Pornpitakpan C The persuasiveness of source credibility: A critical review of five decades' evidence Journal of Applied Social Psychology 2004 34 2 243 281 10.1111/j.1559-1816.2004.tb02547.x Priester JR Petty RE Source attributions and persuasion: Perceived honesty as a determinant of message scrutiny Personality and Social Psychology Bulletin 1995 21 6 637 654 10.1177/0146167295216010 Rich PR Zaragoza MS The continued influence of implied and explicitly stated misinformation in news reports Journal of Experimental Psychology: Learning, Memory, and Cognition 2016 42 1 62 10.1037/xlm0000155 26147670 Seifert CM The continued influence of misinformation in memory: What makes a correction effective? The Psychology of Learning and Motivation 2002 41 265 292 10.1016/S0079-7421(02)80009-3 Selya AS Rose JS Dierker LC Hedeker D Mermelstein RJ A practical guide to calculating Cohen’s f2, a measure of local effect size, from PROC MIXED Frontiers in Psychology 2012 3 111 10.3389/fpsyg.2012.00111 22529829 Susmann, M. W., & Wegener, D. T. (2022a). How attitudes impact the continued influence effect of misinformation: The mediating role of discomfort. Personality and Social Psychology Bulletin, 01461672221077519. 10.1177/01461672221077519 Susmann MW Wegener DT The role of discomfort in the continued influence effect of misinformation Memory & Cognition 2022 50 2 435 448 10.3389/fpsyg.2012.00111 34533754 Swire B Ecker UKH Southwell BG Thorson EA Sheble L Misinformation and its correction: Cognitive mechanisms and recommendations for mass communication Misinformation and mass audiences 2018 University of Texas Press Swire B Berinsky AJ Lewandowsky S Ecker UK Processing political misinformation: Comprehending the Trump phenomenon Royal Society Open Science 2017 4 3 160802 10.1098/rsos.160802 28405366 Tormala ZL Brinol P Petty RE When credibility attacks: The reverse impact of source credibility on persuasion Journal of Experimental Social Psychology 2006 42 5 684 691 10.1016/j.jesp.2005.10.005 Wallace LE Wegener DT Petty RE When sources honestly provide their biased opinion: Bias as a distinct source perception with independent effects on credibility and persuasion Personality and Social Psychology Bulletin 2020 46 3 439 453 10.1177/0146167219858654 31282841 Walster E Aronson V Abrahams D Rottman L Importance of physical attractiveness in dating behavior Journal of Personality and Social Psychology 1966 4 5 508 10.1037/h0021188 6008393 Wilkes AL Leatherbarrow M Editing episodic memory following the identification of error The Quarterly Journal of Experimental Psychology 1988 40 2 361 387 10.1080/02724988843000168 Yoon K Kim CH Kim MS A cross-cultural comparison of the effects of source credibility on attitudes and behavioral intentions Mass Communication and Society 1998 1 3–4 153 173 10.1080/15205436.1998.9677854	36460863	PMC9718466	NO-CC CODE	2022-12-06 23:23:39	no	Mem Cognit. 2022 Dec 2;:1-17	utf-8	Mem Cognit	2,022	10.3758/s13421-022-01374-3	oa_other
==== Front Behav. Soc. Iss. Behavior and Social Issues 1064-9506 2376-6786 Springer International Publishing Cham 115 10.1007/s42822-022-00115-0 Forum on Diversity and Inclusion Understanding the Role of Cultural Values in ABA Service Delivery: Perspectives from Latino Families http://orcid.org/0000-0002-5438-6055 Castro-Hostetler Mariela [email protected] 1 Kille Ircia 2 Lopez Lizbeth Vega 1 Contreras Bethany P. 1 1 grid.266818.3 0000 0004 1936 914X Department of Psychology, University of Nevada, Reno, MS 296, 1664 North Virginia Street, Reno, NV 89557 USA 2 Helix Behavioral Services, Reno, NV USA 2 12 2022 124 31 10 2022 © Association for Behavior Analysis International 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The purpose of this study was to identify and learn about the cultural and language barriers that Latino families face when accessing applied behavior analysis (ABA) services for their children with autism spectrum disorder (ASD). We administered a survey to gather basic descriptive information regarding language and cultural barriers faced by Latino families, specifically from a sample of families living in Nevada. We then conducted follow-up interviews and focus groups to collect detailed accounts of the barriers that these families faced. We used qualitative research methods to provide a detailed analysis that captured the voices of the families who participated in this study. The information provided from the survey and focus groups provides preliminary information for practitioners to help bridge gaps and provide more effective and quality services for the Latino population. Results of the focus groups and interviews are summarized and implications for future research are discussed. Supplementary Information The online version contains supplementary material available at 10.1007/s42822-022-00115-0. Keywords Latino Cultural values Multilingual diversity ABA service delivery Autism spectrum disorder ==== Body pmcAs one of the fastest growing communities within the United States (US), it is expected that by 2045, the Latino1 population will grow up to 24.6% (Frey, 2018). As a population, it is projected that Latinos will make up 28% of the total US population by 2060 (US Census Bureau, 2018). The Latino population represents a large group of individuals originating from over 20 countries, including Central and South America, Cuba, Mexico, Puerto Rico, among others (Office of Management & Budget, 1997). In the US, the majority of Latino subgroups include those who identify as Mexican (64%), followed by Puerto Ricans (9.4%; Motel & Patten, 2012). The Latino population varies significantly by each state in the US. In particular, Nevada has one of the largest Latino populations in the US, ranking 5th nationally at 29% (US Census Bureau, 2021; Wee, 2018). Despite the large representation in numbers, Latinos continue to face barriers in access to healthcare, including behavioral services. According to the Behavior Analyst Certification Board (BACB; n.d) most certified behavior analysts work in the area of autism providing behavioral services. However, it is unclear how many of the services that are being provided by behavior analysts are being accessed by the Latino community. A prevalent barrier among Latino families is obtaining an autism spectrum disorder (ASD) diagnosis for their children in order to receive behavioral services. While the national prevalence of ASD is currently 1 in 44 children in 2018, the current estimate for Latino children is lower (Maenner et al., 2021). Language can furthermore play a critical role in this setting. For example, issues may arise if Latino families are trying to communicate their concerns in the Spanish language and professionals are unable to comprehend family concerns. It is important to stress that limited English proficiency and bilingualism are not inherently barriers in and of themselves. The barrier arises when there are differences in language proficiency between the practitioner and the family receiving services. Thus, the barrier lies within the service delivery system, as there are not enough bilingual professionals to help guide families who need support in the Spanish language (Lanesskog et al., 2015; Ruiz-Adams, 2019). Another barrier that is often faced by Latino families is a lack of knowledge and information about ASD (Zuckerman et al., 2014). Increasing Latino families’ knowledge of ASD can also guide families to find interventions derived from applied behavior analysis (ABA) that are empirically supported (National Autism Center, 2015). Unrecognized differences in the cultural practices between service providers and the recipients of those services is an additional barrier. Culture is always changing and influences families differently (Magaña, 2000). Cultural values may influence how families interpret their child’s diagnosis, decisions regarding whether or not to pursue treatment, the selection of and preference for treatment providers, methods, and goals, and the continuation of services once initiated (Buzhardt et al., 2016; Pitten, 2008; Smith et al., 2011). In addition, cultural values and practices are critical during the intervention process as they may influence adherence to treatment (Slim & Celiberti, 2021). This barrier lies in a system that is not able to readily adapt to the needs of a population and account for such differences (Ferguson & Vigil, 2019). From a behavior analytic perspective, culture can be defined as behaviors of a particular group that are maintained as a result of social reinforcement (Skinner, 1984). Culture impacts the way we use the science of behavior, such as changes in the topography of our interventions. Furthermore, individuals can use their culture to adapt and transform their environment (Lavenda & Schultz, 2010). Cultural values and beliefs are always part of the context in which behaviors occur, and thus, influence patterns of behavior. Understanding cultural values in the Latino community can be critical when providing behavioral services. For example, learning about a family’s culture can help practitioners understand how certain sets of values and beliefs fit within their broader context (Neely et al., 2020; Slim & Celiberti, 2021). In addition, by recognizing a family’s cultural values, behavior analysts can address disparities in access to services (Castro-Hostetler et al., 2021; Neely et al., 2020). Each family brings their own set of cultural values that is unique for their environment, underscoring the importance of service providers developing flexibility skills (i.e., cultural responsiveness) and learning to apply what is needed within the context for each family (Williams, 2021). While behavior analysts should avoid stereotyping cultural beliefs with families or assuming families hold certain cultural values based on their racial and ethnic identity (Neely et al., 2020), there is also risk in ignoring cultural values in service delivery. Given the diversity and different subgroups in Latino families, there is no “set list” of values that will apply to all groups. However, there are some cultural values and practices that are likely shared across Latino families (Garcia-Preto, 2005; Santiago-Rivera 2003). The literature demonstrates a tendency for more collectivist values than individualistic centered values (Dingfelder, 2005). Research suggests that some of those shared values and constructs include familismo (familism), personalismo (friendliness), and confianza (trust; Añez et al., 2008; Juckett, 2013; Magaña, 2020). Although this is not an exhaustive list, we briefly discuss these values here and explore how they can pertain to the context of the delivery of ASD services for Latino families. Cultural Values and Practices Familismo (Familism) The value of familismo refers to values and behaviors around inclusiveness, participation, and a strong bond in nuclear and extended family networks that reflect the way family ties are prioritized (Falicov, 1998; Hernández & Bámaca-Colbert, 2016; Marsiglia et al., 2013). Familismo may involve sharing critical responsibilities such as involvement in important decision making, caretaking, and emotional support (Sabogal et al., 1987). Therefore, when providing behavioral services, Latino families may request for extended family members (e.g., grandparents) to be included in meetings and parent trainings. Furthermore, Latino families may also request goals that are centered around family traditions or values. Personalismo (Interpersonal Relationships) Personalismo can be discussed as creating a personal relationship rather than an institutional relationship (Juckett, 2013; Magaña, 2020). In efforts to establish an interpersonal connection, Añez et al. (2005) recommends that throughout treatment, providers spend at least 5 min at the beginning of the session building rapport. Similarly, Jimenez-Gomez and Beaulieu (2022) discuss the importance of building rapport during the assessment process by using cultural humility and engaging in perspective taking skills with families. Ultimately, this concept will begin to establish a sense of respect from the family’s point of view and understanding of the working relationship moving forward. Confianza (Trust and Intimacy in a Relationship) The value of confianza is often built into the other discussed values held by Latinos and can be essential to clinical engagement practices (Falicov, 1998; Santiago-Rivera 2003). Confianza refers to a sense of trust and intimacy within the created interpersonal relationships (Bracero, 1998). In the context of providing ABA services, this can present itself by families coming in and requesting services from a specific practitioner because of a family or friend’s recommendation. Service providers should be aware that building trust can take time to develop and is an ongoing building experience. While the above listed barriers and cultural barriers might be held by many Latinos, it is inappropriate to assume that this is the case for every family or individual. Furthermore, there are likely differences in the way that families experience a barrier or engage with a value. Therefore, it is worthwhile to directly engage with Latino families of children with ASD to learn about the barriers they face and the values that they hold as it relates to behavior analytic treatment. The purpose of this qualitative study was to learn about the cultural and language barriers that are faced by Latino families when accessing ABA services for their children with ASD, specifically from a sample of families living in Nevada. This was a two-part study; part one was the Participant Demographic and Experience Survey, and part two was centered on the structured interviews and focus groups. The aim of the focus groups was to explore the quality of behavioral services Latino families from Nevada are receiving. Specifically, we aimed to examine the role that cultural values played in family centered planning and treatment services for Latino parents. Method Study 1: Demographic and Experience Survey Recruitment and Participants We recruited participants through local community partners, ABA-based agencies, and organizations that serve children and families with developmental disabilities. We distributed recruitment flyers directly to families and through online platforms such as social media pages (i.e., Facebook) and community pages. In addition, we distributed flyers in English and Spanish. Furthermore, we sent recruitment information via email to approximately five to ten community partners to be shared with families. Families were included to participate in the survey if they met the following criteria: (1) respondent was 18 years or older, (2) identified as Hispanic/Latino/a/x, (3) had at least one child diagnosed with any developmental disability, and (4) currently received or had received ABA services in the past. Eighteen Latino parents participated in the Demographic and Experience Survey. Eight (45%) of the participants completed the survey in English and ten (55%) completed the survey in Spanish. Aggregated participant demographics are presented in Table 1, including the parents’ cultural identity, their primary language spoken in the home, and parents’ educational level. The majority of parents (n = 16, 89%) reported Mexican being their cultural identity. Ages ranged from 25–44, with an average age of 36 years old. Parents reported a range of responses for their primary language spoken in the home; (n = 7, 39%) of parents reported speaking English and Spanish, followed by primarily Spanish (n = 6, 33%) and then English (n = 5, 28%).Table 1 Sociodemographic characteristics of online survey participants Characteristic N % Characteristic N % Gender Hispanic/Latino origin Female 16 89 Mexican 16 89 Male 2 11 Central American 1 5.5 Age Puerto Rican 1 5.5 25–34 3 17 Education level 35–44 15 83 Did not graduate 1 5.5 County of NV residence High school diploma/GED 7 39 Clark 5 28 Some college/associate 5 28 Washoe 13 72 Technical degree 1 5.5 Ethnicity Bachelor’s degree 1 5.5 Hispanic/Latino Language spoken in the home 18 100 Graduate/professional degree 2 11 English 5 28 Other 1 5.5 Spanish 6 33 Both equally 7 39 N = 18. Participants were on average 36.4 years old Survey Instrument and Data Collection Participant Demographic and Experience Survey In the Participant Demographic section, we asked participants about their gender identity, age, what primary language is spoken in the home, county of Nevada residence, ethnicity, Hispanic/Latino origin, and their highest level of education. The design of the survey incorporated both closed and open-ended questions. The second part of the questionnaire asked the participants to share their experiences in receiving ABA services and the extent to which those services were received. The survey also included questions asking what cultural values (if any) were important to them in the process of receiving services. The survey instrument used is presented in Online Resource 1. Upon completing the survey, we asked participants if they wanted to be further contacted to participate in either a focus group or interview. The focus group would be an opportunity for families to further describe their experiences in a group setting, while the interview would be a 1:1 opportunity to discuss their experiences further. The method and results of the interviews and focus groups are described below under Study 2. Coding and Data Analysis We used Qualtrics as a software to obtain frequency counts for all selection-based questions (i.e., Likert-scale questions). To analyze the responses to the open-ended questions, the research team used qualitative methods to code responses according to themes and similar responses across participants. The first author and second author reviewed each response and created a general category based on each participant’s response. For example, to the question, “Were there any barriers in receiving ABA services?” a participant responded with “Issues with insurance and cost of services.” From this response, we created a theme and categorized it under “insurance/funding” for reporting purposes. When other participants responded similarly to this question (i.e., ABA agencies not accepting their insurance plan, not having sufficient funds to pay out of pocket), we placed it under the same category. Once the questions had been reviewed, we used a frequency count for similar codes using Google Sheets. For other open-ended questions, we took a more in-depth process due to varying responses. These questions included the following: “When you started ABA services, how much did you know about ABA or about your child’s diagnosis?”; “What are some cultural values that you hold that are important for your child’s provider to know?” and “What did you find to be the most helpful/effective in receiving ABA services?” For these questions, the first and second author independently read the survey responses and created our own summary based on whether there were frequent responses to the questions or similarities across the responses. Finally, we met to discuss our own summaries and created an aggregate product to report from the findings. Results We include direct quotes from the participants throughout our results. For participants who responded in Spanish, we present the Spanish quotes first, followed by an English translation in brackets. Table 2 contains a summary of the open-ended questions collected through the Participant Demographic and Experience Survey. Seven participants (36%) indicated that they had received ABA services for more than 3 years, and 11 participants (62%) indicated receiving services for less than 3 years. Across participants, the most common reported initial barriers to begin ABA services were an extended waitlist time, issues around insurance/funding, and the COVID-19 pandemic. Other barriers that parents indicated were the lack of providers providing therapy for their child’s age and the lack of flexibility in scheduling to attend both school and ABA related services. Several participants (n = 6, 34%) reported learning about ABA services through community resources. When we asked parents what they valued around ABA services, they reported: treatment planning, individualized planning, data progress, and involvement. For example, one parent stated, “I like the planning of all his programs. They push him to do it even when he gets frustrated. He gets challenged, but I know he can do it.”Table 2 Open-ended parent responses from online survey N % N % Length of ABA services Parent values around services Less than 6 months 1 6 Treatment planning 6 33 6 months – 1 year 4 23 Individualized planning 3 17 1 year – 3 years 6 33 Data progress 3 17 More than 3 years 7 36 Involvement 5 28 Major initial barriers Other 1 5 Waitlist 5 28 Learning about ABA services Insurance/funding 3 17 Internet search 2 12 COVID-19 pandemic 2 10 Resources 6 34 None 3 17 Family/friends 3 18 Other 5 28 Service provider 2 12 Referral 3 18 Walk-In 1 6 N = 18. “Other” major initial barriers identified were the lack of providers in the area (including providers that would accept child’s age because they were older than 10) and lack of flexibility with the child’s school schedule and doing ABA at the same time Table 3 provides a summary of parents’ experiences with ABA services. Overall, the majority (n = 14, 82%) of parents reported always feeling included around treatment planning, followed by 12% (n = 2) reporting that they often felt included. Similarly, the majority (n = 9, 53%) of parents reported ABA agencies were always taking cultural considerations around treatment planning, followed by 24% (n = 4) of parents reporting cultural considerations were considered often. However, one family (6%) rated “rarely” for agencies taking cultural considerations. Finally, the majority of parents (n = 13, 77%) reported either always or often having someone available to communicate their needs in Spanish if needed. In contrast, two parents (12%) reported “rarely” or “never” having someone available to communicate with them in Spanish.Table 3 Parent experiences around ABA services Question Always Often Sometimes Rarely Never N (%) N (%) N (%) N (%) N (%) Do you feel included around treatment planning for your child? 14 (82) 2 (12) 1 (6) 0 (0) 0 (0) Do you feel that ABA agencies took cultural considerations in treatment planning for your family? 9 (53) 4 (24) 3 (19) 1 (6) 0 (0) Does your service provider speak Spanish or was there someone you could communicate your needs to? 9 (53) 4 (24) 2 (11) 1 (6) 1 (6) N = 17 For the remainder of the open-ended questions, we were unable to create repetitive themes around some of the responses due to the varying responses. However, there are implications from the collected results, which are outlined here. For example, in response to “When you started ABA services, how much did you know about ABA or about your child’s diagnosis?” – all families reported very minimal to no knowledge of what ABA services were or the effects it would bring to their family. One parent reported that they “Did not know all the benefits from ABA, and would have enrolled sooner if they had.” In addition, another parent reported that because they “did not know much about their child’s diagnosis, a lot of research had to be done on their own. In response to the next question – “What are some cultural values that you hold that are important for your child’s provider to know?” – some parents identified some of the following values: trust, respect, open communication, personal cultural practices, and family. Method Study 2: Structured Interviews and Focus Groups Recruitment, Participants, Setting, and Materials Once surveys were completed, we contacted families (with their preference of contact) that indicated an interest to participate in either an individual structured interview or in a focus group within a week to follow-up; thus, all participants for Study 2 were pooled from those who participated in Study 1. Based on the participants’ availability and language preference, we organized two interviews to take place over the phone and two focus groups online. Offering the interviews and focus groups over the phone and online allowed participants across the state to participate and ensure appropriate safety precautions due to this study taking place during the COVID-19 pandemic. Two Latino parents participated in the structured interviews. Seven Latino parents participated in two focus groups. Focus Group 1, conducted in English, included a married couple and an additional participant. Focus Group 2, conducted in Spanish, included four Latino parents as participants. The majority of parents were still currently receiving ABA services (n = 5, 56%), while the other participants had previously received services (n = 4, 44%). Aggregated participant demographics are presented in Table 4, including the parents’ cultural identity, their primary language spoken in the home, and parents’ educational level. All participants reported Mexican being their cultural identity. Parents reported a range of responses for their primary language spoken in the home; (n = 5, 62%) of parents reported speaking English and Spanish, followed by primarily English (n = 2, 25%) and then Spanish (n = 1, 13%).Table 4 Sociodemographic characteristics of structured interviews and focus group participants Characteristic N % Characteristic N % Gender Hispanic/Latino origin Female 8 89 Mexican 8 100 Male 1 11 Education Level Age Did not graduate 1 12.5 25–34 2 22 High school diploma/GED 4 50 35–44 7 78 Some college/Associate 2 24 County of NV residence Technical degree 1 12.5 Clark 2 22 Washoe 7 78 Ethnicity Hispanic/Latino 9 100 Language spoken in the home English 2 25 Spanish 1 13 Both equally 5 62 N = 9. Only some demographic information was collected from the male parent (participant from the married couple) from the English-speaking group due to only the other parent completing the participation forms Procedures All interviews and focus groups were conducted by the first author. As stated above, we conducted two separate interviews: one in English and one in Spanish. The one-on-one interviews were conducted for families who did not select to participate in a focus group and provided additional flexibility for families to discuss their experiences. The interview was structured similarly to the focus group (outlined below), including the purpose and general overview for the interview at the start. The semi structured interviews were conducted over the phone and lasted approximately 20 to 45 min. We conducted two separate focus groups: one in English and one in Spanish. We conducted all focus groups online via the Zoom platform and each lasted approximately 90 min. At the beginning of each focus group, the moderator (first author) described the purpose and general overview of the focus group. Following the introduction, we introduced a group activity using The ACT Matrix (Polk & Schoendorff, 2014). The ACT Matrix is a tool used to track short-term and long-term outcomes of an individual’s behavior and determine the behavior’s function (see Polk & Schoendorff, 2014). The purpose of running The ACT Matrix was to identify parents’ values and actions to take in order to move toward those values (i.e., spending quality time together). At the start of the activity, we discussed the difference between values and goals with a guided example. This distinction is important to demonstrate how values are never ending and goals are achievable. The example was further demonstrated with the presentation of a blank ACT Matrix. We explained each of the ACT Matrix quadrants, as well as demonstrating how the discussed example (value around health) would fit in each quadrant area. Next, in a discussion format, we asked the group: “What are things you value/are important to you?” We waited to obtain responses from the majority of participants and allowed for the group discussions to occur. We guided the group around the different areas of the matrix and moderated the group in order to allow all participants to share their experiences. During this activity, several values were identified by the group that helped with further discussion during the session. At the end, we provided parents with a copy of a blank matrix to complete independently and track their own changes for future use. At the conclusion of the focus groups, we provided participants with additional resources for support groups and parent training opportunities in the state. In addition, participants were encouraged to further contact the moderator/researchers if they had additional questions or comments that were not addressed in the group context. A separate researcher scored a treatment fidelity checklist to ensure the moderator included the same guiding questions and protocol for both focus groups. Treatment fidelity was scored at 100% across both focus groups. Coding, Data Analysis, and Interrater Reliability We used qualitative research methods to code and analyze the responses from the structured interviews and focus groups. Qualitative research is a “systematic approach to understanding qualities, or the essential nature, of a phenomenon within a particular context” (Brantlinger et al., 2005, p. 195). Qualitative methods are empirical, involve observation and experience, and enable us to address complex issues of practice including working with diverse individuals. In addition, qualitative methods can be useful during the pre-intervention stages to gather perspectives from the population and examine the acceptability of an intervention (Hitchcock et al., 2010). The outcomes of qualitative research can inform practices to provide a detailed description of a given problem. Qualitative methods can be conceptualized as a systematic descriptive assessment of social validity. In this study, we evaluated the social validity of ABA from the perspective of Latino families using three coding cycles (see, Saldaña, 2021 for more detail). We audio recorded all focus groups and interviews. We used a transcription company to produce the initial transcription, then manually compared the transcription to the audio recording to ensure it was correct. Additionally, we edited some words from the initial transcriptions that did not translate appropriately (e.g., use of slang terms, ‘Spanglish’, or acronyms). In addition, a separate researcher listened to and verified the transcription from the Spanish-speaking focus group to ensure the context was transcribed appropriately. We proceeded to the coding process using the three cycles of coding as outlined by Saldaña (2021). The overall goal of the cycles of coding was to identify themes in the participants’ discussions that related to cultural values and barriers that were faced by the participants in the context of receiving ABA services for their children with ASD. For the first cycle of coding, elemental coding methods were utilized. Elemental coding methods are “primary approaches to qualitative data analysis” (Saldaña, 2021, p. 129). The form of elemental coding that we used was in vivo coding. The in vivo method is a form of qualitative data analysis that places emphasis on the actual spoken words of the participants (Manning, 2017; Saldaña, 2021). In addition, in vivo coding can be helpful when researchers are interacting with participants from a particular culture and better understand stories or phrases used throughout (Manning, 2017; Manning & Kunkel, 2014). The first cycle involved listening to and reading the focus group transcripts to review the overall content. This initial process allowed us to become familiar with the collected data. As we listened to the audio recordings, we took notes and began to highlight important areas of the participant’s experiences. Next, as suggested by Creswell (2007), two researchers individually read and manually coded the transcripts. This first cycle allowed for researchers to get familiarized with the content and apply initial codes to the relevant areas. We utilized in vivo coding methods in order to best capture the meaning of participants’ experiences and identify open codes. The second cycle of coding involved focused coding. Focused coding searches for the most frequent or significant codes from in vivo coding to develop salient categories in the data and “requires decisions about which initial codes make the most analytic sense” (Charmaz, 2014, p. 138). The aim of focused coding is to connect statements and experiences across all of the participants’ responses. Next, as Creswell (2007) recommended, each researcher independently coded and created separate themes. Once this was complete, all researchers met to discuss the themes and categories they had created. Furthermore, they agreed/disagreed on themes and categories that would be presented and further analyzed (Creswell, 2007). The researchers discussed all areas of disagreement on themes and categories and reached a consensus on which themes and categories to use for the remainder of the coding process. Upon agreement, the themes were discussed to increase reliability and improve the validity and credibility of the data as a whole and what codes would meet the criteria. In addition, a third researcher coded the data without seeing the themes that had been finalized. This allowed for further reliability to decrease possible researcher bias and make sure no further themes had been missed (Luker, 2008). The third and final cycle of coding used was codeweaving. Codeweaving is “the actual integration of key code words and phrases into a narrative form to see how the puzzle pieces fit” (Saldaña, 2021, p. 345). Part of the final codeweaving process was to jointly review the significant overlap in themes into comprehensive categories. The list of themes was ultimately jointly reviewed by all the researchers. This collaboration helped to clarify and refine the specific themes. The number of codes were not specifically counted for reliability as “counting conveys a quantitative orientation of magnitude and frequency contrary to qualitative research” (Creswell, 2013, p. 185). Instead, reliability in the coding process was achieved through the process of ongoing discussion and consensus among the three researchers. Results The overall goal of this qualitative approach to coding the participants’ discussions was to identify themes within and across their responses that can then provide information to help practitioners move toward culturally responsive service provision for Latino families. Therefore, we present the results in terms of themes and subcategories identified within and across their responses. Similar to the above, we include direct quotes from the participants to illustrate the themes and subcategories that we identified during the coding process. As we present the results, the Spanish quotes are provided first in italics, followed by the English translation in brackets. While we identified some themes that were more frequently discussed in one focus group than the other (and vice versa), we identified consistent themes and categories across both focus groups and the structured interviews. We identified four main themes that emerged across the focus groups and structured interviews, which are outlined in Table 5. These themes included (1) family and cultural values; (2) reaction to receiving a diagnosis; (3) impact of ABA services; and (4) the future of ABA and recommendations. Each theme comprised two to three subcategories. For each theme, we describe the characteristics of the parents’ responses, followed by illustrative quotes. We provide additional illustrative quotes for each theme in Online Resources (2–6).Table 5 Themes and subcategories for all focus groups and structured interviews Theme Subcategories Family and cultural values Family Trust and friendliness in therapeutic relationships Role as a parent Reaction to receiving a diagnosis Perspectives/knowledge of receiving a diagnosis Lack of support Impact of ABA services Positive outcomes The future of ABA and recommendations Transitional services (aging out) Need for resources and further education Family and Cultural Values Family Several parents described the importance of family (familismo) across different lenses. First, parents described the importance of family as doing the best they could for their child. For instance, to ensure their children have the appropriate support system to be successful in the future, as well as to support everyone else in the household. For example, at the start of the English-speaking focus group an overarching value that a parent reported and others agreed on was for their “kids to do well in life.” Parents expressed the different ways this looked in their household. For instance, one parent described the importance of communication and flexibility that needed to happen in their household, “I’ll just work, you know, cleaning jobs, fast food jobs, whatever it is just so that the schedule is flexible for therapies-pick up drop-offs whatever it needed to be like. I was a hundred percent and I still am a hundred percent dedicated to whatever my boys need, that always comes first. So, it just took a lot of communication between my husband and I to decide that factor.” Trust and Friendliness in Therapeutic Relationships Similarly aligning with the literature, the value of confianza (trust) and personalismo (friendliness) were discussed (Magaña, 2020). With some families having their children enrolled in 20–35 hrs of ABA therapy, many are left with faith and trust that their child is getting the best treatment and care possible. One parent stated, “Sometimes they [children] cry at drop off, sometimes they don’t and then you’re wondering. I think what helps with the parent classes is being able to see how my kids interact with the therapists.” Another parent discussed the importance of trust with their service provider, “I wanted there to be honesty of what was going on with my child’s progress, so I could really know how well or not well he was doing.” Due to parents having little information about ASD when they first started ABA therapy, building therapeutic and open communication relationships with their child’s therapists was important. One parent agreed that she always felt comfortable sharing her concerns with maladaptive behaviors she was seeing in the home, “They let you know when you bring up these issues. They’re like this is normal, this is what we’re seeing here, this is what we’re going to work on or this is how we’re going to track this and that I need. You don’t feel crazy when you’re around them.” Not only does building a therapeutic relationship help families gain trust with services, but it may also help providers have better communication with families. Role as a Parent Another clear value that was present was their active and engaging role as a parent. Across the different stages of raising their children, parents were often questioning if they were doing enough or the right thing (e.g., decisions regarding treatment, engaging with other family members). One parent from the Spanish-speaking focus group stated “Yo me sentía muy mal. Yo creía que yo era la responsable de muchas cosas que le pasaban a mi hijo, porque no le enseñé a defenderse. Pero yo siento que en el momento hacemos lo que sentimos, qué es lo mejor para ellos y nuestro instinto de mamá.” [I felt really bad. I thought that I was responsible for many things that happened to my son because I did not show him how to stand up for himself. But I feel that in the moment we do what we feel is best for them and trust our motherly instinct]. See Online Resource 2 for additional illustrative quotes. Reaction to Receiving a Diagnosis Perspectives/Knowledge of Receiving a Diagnosis A commonality across many of the participants was their shared experience of when their child was first diagnosed with autism. Several parents described how they were initially ignorant of what autism was, and even in denial that their child had autism. One mother stated, ““Al principio cuando me dijeron que mi hijo tenia autismo, a ratos se le va a ir; vino la ignorancia de uno como Latino. Yo creo, que la comunidad Latina es muy ignorante en ese aspecto, deberíamos estar más educados.” [At first when I was told my son had autism, I thought it’d go away; here came the ignorance of one as a Latino came. I believe that the Latino community is very ignorant in this respect, we should be more educated [about autism]. Other parents were not prepared to receive the information that their child has autism. Families discussed that receiving an ASD diagnosis was stressful for their family and they needed time to emotionally adapt. A parent from the English-speaking focus group stated, “For me the diagnosis was a traumatic experience. I wish we were guided by some type of family counselor. At that point, I could not handle it.” Lack of Support The term “stigma” was never utilized by either the moderator or parents during the discussion; however, the context of stigma was related to a lack of support from the community/service providers and isolation from family members. As mentioned earlier, while participants identified “family” as a value, the effects of the lack of support they had received from extended family members was emphasized as well, “Yo en lo personal me alejé de mis amistades y de mi propia familia por pensar que van a juzgar, van a criticar, y no entender la condición de mi hijo.” [Personally, I distanced myself from my friends and my own family because I thought they would judge, criticize, and not understand my son’s condition]. Several participants described struggling to maintain strong family relationships and overall feeling isolated, “Outside family doesn’t understand what it is like to have our kids, they don’t know what our daily struggles or challenges are. As much as they want to be supportive, they just don’t get it.” See Online Resource 3 for additional illustrative quotes. Impact of ABA Services Positive Outcomes All parents (n = 9) reported positive impacts from receiving ABA services. Some areas of importance included teaching daily living skills, independence skills, and developmental skills. One parent from the structured interview stated, “He’s not head banging as much and he’s learning to say words. I remember crying just because he learned to say ‘eat.’ They are little milestones!” Another parent from the Spanish-speaking focus group shared a similar positive outcome for her child, “Ha pasado muchas horas en terapia [ABA], pero pienso que pues sí le ha ayudado mucho porque ahora come, come comida. Creo que a los 4 años empezó a comer chicken nuggets. A los cuatro años que empezamos me dio un beso porque no daba besos, no hacía nada. Sólo pasaba metido debajo de la cama y en el closet.” [He has been through many hours of therapy [ABA], but I think it has helped him a lot because now he eats food. I think at 4 years old he started eating chicken nuggets. At 4 years old when we started [therapy] he gave me a kiss; he did not kiss, he did not do anything. He would only hide under his bed and in the closet]. Other effects were improvements in quality of life for both parents and children due to ABA services. The activity presented at the beginning for the focus groups provided guidance for families to identify what they valued for both themselves and their families. A parent from the English-speaking focus group stated, “The behavioral team went out to different locations and I thought that was really life changing for my family because now we can go to restaurants, we can go to the grocery store, and we do a lot better with meltdowns at those different places. Because it was not possible to walk into Walmart and my child would start screaming at the top of her lungs for reasons I don’t know and so now we can…life is better.” Barriers The time from the initial ASD diagnosis to the time of receiving services varies from each family and can depend on several external factors to start services; for instance, the number of providers in the area, insurance/funding that are in network with providers, and whether providers have the capacity with staff to provide services to an incoming family. As mentioned above, while parents saw the impact of ABA services, a barrier that many families discussed was the initial wait for services to become available. One parent stated, “The wait list was very scary because although they told me I was going to get help, I sat there alone for 8 months before I got it.” Similarly, another parent agreed on the difficulties of being on a waiting list, “First, the waitlist was a big barrier. I wish there would have been something available to start with while we waited for services.” Language Barrier A significant difference and added subcategory that emerged from the focus group conducted in Spanish, was the presence of a language barrier in order to receive an ASD diagnosis and ultimately ABA services. One parent stated, “Ha sido un poco difícil en parte, porque también como latina no hablo mucho inglés.” [It has been a little difficult in part, because as a Latina, I do not speak English]. Similarly, another parent stated, “Pero si era muy difícil encontrar quien hablara español, pues yo como podía yo les decía mi hijo necesitaba servicios y así fue como que a ellos me refirieron a los servicios de ABA.” [It was very difficult to find someone who spoke Spanish. As I could, I told them my son needed services and that’s how they referred me to ABA services]. For another parent, communication was similarly a present barrier to being involved once services were attained, “No creo que me sentía muy incluida por ser parte de esas conversaciones para estar involucrada [planificación del tratamiento], sin embargo, pero sentía la necesidad de involucrarme. No hablo muy bien el inglés, pero lo entiendo bien. Así que solo miraba en casa cómo trabajaban con mi hijo y luego hacía lo mismo.” [I don't think I felt very included to be part of those conversations to be involved (regarding treatment planning), however I still felt the need to involve myself. I do not speak very good English, but I understand it well. So, I would just watch at home how they worked with my son and then did the same thing]. This barrier was less evident from the English-speaking group. See Online Resource 4 for additional illustrative quotes. The Future of ABA and Recommendations The Need for Resources and Further Education Across both focus groups and interviews, the greatest area of need reported by parents was the importance of providing learning opportunities about ASD and ABA. While some parents had heard about the positives and negatives of ABA prior to starting services, they really did not know what ABA therapy was and why it would be the best fit for their child. One parent stated, “Both therapists [from early interventions] brought up ABA early on and at first I didn’t know if it was something I was comfortable with. Yeah, it was just something that I didn’t know. Because there was so much hearsay that it’s like basically you’re putting your kids in this program to shut off all his emotions or to tell him that his stims (stimming/self-stimulatory behaviors) are not okay.” Others had expressed wanting to receive more training on the techniques and explaining the rationale behind the behavioral strategies, “I had to watch without understanding, so that can definitely be changed with more understanding and education.” In both focus groups, parents offered recommendations for providers to consider, especially with families who are newly diagnosed with ASD. One parent stated, “Necesitamos un entendimiento de lo que es ABA para nuestra comunidad y yo creo que es importante para los papás que vienen en este nuevo ciclo a los que están recientemente siendo diagnosticados, que hubiera un manual o algo en una introducción mejor.” [We need a better understanding of what ABA is for our community and I believe is important for parents who come into this new cycle of being recently diagnosed, there should be a manual or something for a better introduction]. Transitional Services (Aging out) A concern/worry that was brought up during the English-speaking focus group was the availability of services after their children turn 18 years old. Many parents were already aware that in their area there are little to no services available for adults with ASD, “There’s so much for the little babies and there’s nothing for you know, those little babies. They grow up of course and there’s no continuation. We’re parents, we’re not going to be here forever. There’s going be times when our children are here without us and there’s just such a lack of services for that point. It’s just really scary.” While many ABA organizations do not provide services after children turn 18 years old, a common process is working on individualized skills to prepare them for their next phase of life (i.e., job searching, independent living). While there is still much work to be done in this area and creating these services, further support is also needed for parents to navigate the continuation of services. See Online Resource 5 for additional illustrative quotes. Discussion The aim of this study was to identify and learn about the cultural values and beliefs held by Latino families in Nevada. In addition, we also examined barriers faced by Latino families when accessing ABA services. In Study 1, we distributed the Participant Demographic and Experience Survey to Latino families who were currently receiving ABA services or had received services in the past. The survey included questions about the families’ cultural identity, their primary language spoken in the home, and parent educational level. The second part of the questionnaire asked the parents to share their experiences in receiving ABA services and the extent to which those services were received. In Study 2, we conducted structured interviews and focus groups with some of the families who participated in Study 1. From the structured interviews and focus groups, we identified four main themes: (1) family and cultural values; (2) reaction to receiving a diagnosis; (3) impact of ABA services (4) future recommendations for the field of ABA. From these themes, we found what aspects were meaningful in receiving ABA services, as well as barriers that families faced when seeking services. An initial barrier that 28% of families reported was being on an extended waitlist for services. Similarly, a survey conducted as part of the Nevada Legislative Counsel Bureau Audit Division (2021) found that waitlist issues were the most frequently mentioned barrier. They found that 55% of providers had a 4-month waitlist for their practice and 18% had a waitlist of over a year. It is estimated that for Nevada in 2020, there were only enough providers to serve about two out of every three children who needed ABA services (Legislative Counsel Bureau Audit Division, 2021). While the number of ABA providers in Nevada has significantly increased, it is still insufficient to meet the needs of children who need services. The wait time to receive treatment is even longer for low income households in comparison to families with private insurance (Legislative Counsel Bureau Audit Division, 2021). While extended time on a waitlist is a common barrier for Latino families (Magaña, 2020; Rosales et al., 2021), ways behavior analysts can address this are limited in the literature. An interesting aspect worth noting from both studies were implications regarding language and barriers around language. First, when analyzing the data completed from the English and Spanish forms. In general, we found no differences in some of the general demographic data. However, there were meaningful differences to the following question, “What is your primary language spoken at home?” From the participants who completed the form in Spanish, 60% reported primarily speaking Spanish in the home, while no participants who completed the form in English reported Spanish being the primary language spoken in the home. These data demonstrate how much language matters. When given the option between Spanish and English, the families that primarily speak Spanish in the home, chose to complete the study in Spanish. This lends support to the assertion that given the option of having a part of services and information being available in Spanish is important to families. As discussed in the literature review, language barriers play a critical role in developing and maintaining a good therapeutic relationship between families and providers (Ferguson & Candib, 2002). A difference that emerged between the focus group conducted in Spanish in comparison to the focus group conducted in English, was the presence of a language barrier when seeking services. This can be a call for service providers to ensure their services are modified to meet the needs of the Latino community. For example, ensuring paperwork and communication is in a language that families will completely understand. For instance, having onboarding paperwork, behavioral support plans, and trainings be available in their language of preference. This feedback aligns with the subsection of the current Ethics Code for Behavior Analysts (BACB, 2020) under Sect. 2.08 Communicating about Services:2.08 Behavior analysts use understandable language in, and ensure comprehension of, all communications with clients, stakeholders, supervisees, trainees, and research participants. Across both focus groups and interviews, the greatest area of need reported by parents was the importance of providing learning opportunities about ASD and ABA services. In another of our open-ended questions in Study 1, 34% of families reported learning about ABA services through community resources. This number could be a good indicator as to where and how service providers can disseminate general information about services. As noted in the results of Study 2, several families indicated an overall lack of knowledge about ASD and how to attain information about ABA services. The findings of Latino families having limited knowledge about ASD are consistent with the literature (Chlebowski et al., 2020; Zuckerman et al., 2014). However, to our knowledge, available information for Latino families about their awareness in ABA related services is limited in comparison to Latino’s knowledge about ASD in general. Families in this study reported that even when they started ABA services, they still had several questions about ABA (i.e., the evidence support) and wished they had more knowledge about their child’s ASD diagnosis. One recommendation offered by parents in the English-speaking group was for service providers to give families more information about the time commitment needed for ABA services. One way service providers can increase parent knowledge on ABA is to provide ongoing training opportunities on specific topic areas that are applicable to their children. For example, during parent training sessions the importance of reinforcement strategies can be discussed, followed by an application of reinforcement strategies for their child. In addition, collaborations with outside community agencies would be beneficial to create an easier referral process for families or another resource for information on the topic of ASD and ABA. For instance, collaboration events can include practitioners from the community providing general information and answering questions families may have. In addition, this would also help build a trusting relationship between families and service providers from the initial start of services. Overall, families responded positively to being included in treatment planning for their child. While the majority of families reported ABA agencies always taking cultural considerations around treatment planning, some families indicated that ABA agencies sometimes or never took cultural considerations in treatment planning, respectively. This question was asked on a Likert-type scale and did not allow participants to add follow up with additional information. While “cultural considerations” were not defined for participants, it is still an indication that there are still ways behavior analysts should be mindful and open to learning about a family’s cultural values and practices. For families who did respond that culture was considered, a further follow up would aid the analyses of what exact considerations were taken from providers. For example, examining if families were given the opportunity to select goals that are culturally relevant (e.g., mealtime prayer). A strength from the interviews and focus groups were the inclusion of different experiences from families that had gone through the process of receiving ABA services. For instance, participants were able to share a range of experiences due to the large spectrum of their children’s ages and duration of experience with ABA. The participants had children ages 0–6 years old (n = 3), 6–15 years old (n = 2), and 16–18 years old (n = 3). This also allowed the focus groups to serve as a support network for families who were in their first year of services, while others were looking for support to transition services once their children turned 18 years old. The themes that were created from the results demonstrated the importance of those topics across groups, as each family offered different viewpoints. In general, the discussion of values aligned very similarly with the literature reviewed above, specifically with the values of familismo (familism) and personalismo (interpersonal relationships). Although, the presentation of these values was somewhat different. In the literature, the value of familismo is typically focused on the engagement and inclusion of extended family members during treatment planning. For our participants, the value of familismo was centered more toward the well-being of their child and the role of the nuclear family members in the household to be supportive as a unit. For instance, goals were centered for their child to gain independence and be part of the community. In addition, families discussed the changes and sacrifices that had to be made for their family in order to integrate ABA services into their lives. Our results suggest an emphasis on persistence and grit from their family values to reach those goals. This included working additional jobs to support their family and maintaining jobs that offered flexibility for their children’s schedule. On the other hand, familismo was also discussed as a value that was there for their nuclear family, but was not demonstrated from their extended family. Participants discussed how their relationships with family members had negatively shifted and the difficulties they faced from the family's lack of understanding what their child needed. Our results also expand on current knowledge about the decision-making roles Latino families, particularly the roles that Latino mothers, take during this experience. Across both focus groups, all but one participant was a mother of a child with ASD. The one participant that identified as male, was part of a dyad couple; no fathers of children with ASD participated on their own. A challenge that mothers expressed was feeling that they had to make the “right” decision of care for their child. Many of the mothers discussed the constant questioning of whether they were doing the right thing in having their kids in ABA services and the unknown if they were doing the best that they could. For one parent, her decision was not clear for her until years later. However, whether to begin services or not was only the start for some families taking the best care of action for their child with ASD. Parents often struggled with their own personal thoughts, questioning if they were doing enough for their children and related it back to questioning if they were a good enough parent. This is not a unique aspect for Latino families, but it is an important issue that needs to be addressed. While not discussed in depth in this study, the cultural construct of gender roles can be prevalent in the Latino community. This value also extended to their other children in the home. Parents often recognized that much more of their time was spent on their child with ASD, in comparison to their other children. Understanding how these cultural values impact ABA services may help service providers foster a better therapeutic relationship that can ultimately lead to better service outcomes. For example, promoting values such as personalismo (interpersonal relationships) can facilitate conversations, in which expectations for services and treatment can be discussed (Chlebowski et al., 2020; Hackethal et al., 2013). In the context of bringing the value of personalismo to families during the focus group session, the researcher initiated separate conversations and ensured confidentiality for families to feel more comfortable. While the researcher had limited rapport built with participants, service providers can ensure the value of personalismo continues to build across meetings or sessions. Furthermore, creating an interpersonal relationship and discussing treatment expectations can often be built into cultural adaptation frameworks (Domenech Rodríguez et al., 2011). Thus, cultural values are a vital component when adapting treatment services, specifically when making modifications to parent training programs. There are some limitations from the study that should be noted. First, the sample size is small, especially given the total Latino population in Nevada (29.2%; n = 918,045; US Census Bureau, 2021). As such, the data collected from this small sample is likely not representative of the total population of Latino families receiving services in the state of Nevada. The extent to which our study is representative of Latino families within the state is that participants in this study resided within two of Nevada’s 16 counties, Washoe and Clark County (72% and 28% respectively). Both of these counties are the largest metropolitan areas in Nevada. Participation from other rural counties would provide more insightful information due to rural populations being less diverse, both racially and ethnically. Future research can investigate and compare the needs of Latino families living in rural Nevada to Latino families receiving services in larger metropolitan areas. Similarly, the diversity of families that participated in the study was limited with 89% of participants identified as Mexican as their Hispanic/Latino origin. As discussed previously, there are several subgroups within the Latino population and values can vary across each group. Future research can investigate other regions of the country and compare their experiences receiving ABA services. Another limitation was the accessibility and distribution of the survey. Due to recruitment taking place during the COVID-19 pandemic, recruitment and dissemination of the study primarily took place online. While we shared recruitment flyers with ABA providers who were currently working with Latino families and posted on ASD community websites, we were still unable to yield a higher representative sample. Future studies may expand recruitment strategies by sharing at community events in Latino communities. A final limitation of the study was not including a question in the survey regarding how many years families had lived in the US. Responses to this question may provide better information as to whether these perspectives and values align with a specific generation for Latino families or if values mentioned came from across generations. The number of years living in the US may also be correlated with their knowledge about ASD and accessibility for services (Rosales et al., 2021). Future research may include adding a question on how many years they have lived in the US and/or how many generations their family has lived in the US (e.g., first generation, second generation). The results of Study 1 and Study 2 provide some implications for future research to address the needs of Latino families receiving ASD services. While research in the field is starting to explore ABA service delivery to Latino families with children with ASD (see Rosales et al., 2021), there is still limited research focused on experiences with Latino families receiving ABA services and how to culturally adapt treatments. Findings from this study can help further identify the different contexts and methods that can be adapted for a given community. The data from this study also suggest that there is still work that needs to be done from a service provider’s standpoint. For example, even when Latino families receive ABA services there are still questions regarding the purpose of treatment procedures and knowledge of other resources. Behavior analysts should continue to work on learning about different family cultures and practice cultural humility across all families that they serve. In summary, we were able to identify important cultural values for a small sample of Latino families receiving services for ASD and other related developmental disabilities. In addition, findings from this study provided an opportunity for families to share their experiences about receiving ABA services. Overall, families responded positively with the services they received; however, there is still much work to be done in order to provide high quality ABA services to Latino families with children with ASD. This study offers support to what cultural values are important to the Latino community in Nevada and can be used to culturally adapt interventions. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 42 kb) Data Availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics Approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the University of Nevada, Reno’s Institutional Review Board, protocol#1690679–2. Consent to Participate Informed consent was obtained from all individual participants included in the study. Consent for Publication Participants signed informed consent regarding publishing their data. Conflict of Interest The authors have no relevant financial or non-financial interests to disclose. 1 While the term Latinx is currently considered to be a progressive and inclusive identifier (Scharrón-del Rio & Aja, 2020), the authors elected to use the term Latino throughout this manuscript. The authors believed that the term Latino would be a better representation because when working directly with the families in the study, the term Latino was primarily used due to the familiarity and preference indicated by the families. We would like to provide some important background information given the topic of the manuscript. MCH, IK, and LVL all identify as Latinas and Spanish-English bilingual females. BPC identifies as a White female English speaker. ==== Refs References Añez, L. M., Paris Jr, M., Bedregal, L. E., Davidson, L., & Grilo, C. M. (2005). Application of cultural constructs in the care of first generation Latino clients in a community mental health setting. Journal of Psychiatric Practice, 11(4), 221–230. 7u Añez LM Silva MA Paris M Bedregal LE Engaging Latinos through the integration of cultural values and motivational interviewing principles Professional Psychology: Research and Practice 2008 39 2 153 159 10.1037/0735-7028.39.2.153 Behavior Analyst Certification Board. (n.d). BACB certificant data. Retrieved from https://www.bacb.com/BACB-certificant-data. Behavior Analyst Certification Board. (2020). Ethics code for behavior analysts. https://bacb.com/wp-content/ethics-code-for-behavior-analysts/ Bracero W Intimidades: Confianza, gender, and hierarchy in the construction of Latino-Latina therapeutic relationships Cultural Diversity and Mental Health 1998 4 4 264 10.1037/1099-9809.4.4.264 9818515 Brantlinger E Jimenez R Klingner J Pugach M Richardson V Qualitative studies in special education Exceptional Children 2005 71 2 195 207 10.1177/001440290507100205 Buzhardt J Rusinko L Heitzman-Powell L Trevino-Maack S Mcgrath A Exploratory evaluation and initial adaptation of parent training program for Hispanic families of children with autism Family Process 2016 55 1 107 122 10.1111/famp.12146 25776105 Castro-Hostetler, M., Greenwald, A. E., & Lewon, M. (2021). Increasing access and quality of behavior-analytic services for the Latinx population. Behavior and Social Issues, 1–26. 10.1007/s42822-021-00064-0 Charmaz, K. (2014). Constructing grounded theory (2nd ed.). Sage. Chlebowski C Hurwich-Reiss E Wright B Brookman-Frazee L Using stakeholder perspectives to guide systematic adaptation of an autism mental health intervention for Latinx families: A qualitative study Journal of Community Psychology 2020 48 4 1194 1214 10.1002/jcop.22296 31816103 Creswell, J. W. (2007). Qualitative inquiry & research design: Choosing among five approaches (2nd ed.). Sage Publications. Creswell, J. W. (2013). Qualitative inquiry & research design: Choosing among five approaches (3rd ed.). Sage Publications. Dingfelder, S. F. (2005). Closing the gap for Latino patients. Monitor on Psychology, 36(1), 58–61. Domenech Rodríguez MM Baumann AA Schwartz AL Cultural adaptation of an evidence based intervention: From theory to practice in a Latino/a community context American Journal of Community Psychology 2011 47 1 170 186 10.1007/s10464-010-9371-4 21116707 Falicov, C. J. (1998). Latino families in therapy: A guide to multicultural practice. Guilford Pres Ferguson A Vigil DC A comparison of the ASD experience of low-SES Hispanic and non-Hispanic White parents Autism Research 2019 12 12 1880 1890 10.1002/aur.2223 31639292 Ferguson WJ Candib LM Culture, language, and the doctor–patient relationship Modern Culture and Physician- Patient Communication 2002 34 353 361 Frey, W. (2018). The US will become ‘minority White’ in 2045, census projects. Brookings. Retrieved from https://www.brookings.edu/blog/the-avenue/2018/03/14/the-us-will-become-minority-white-in-2045-census-projects/ Garcia-Preto N McGoldrick M Giordano J Garcia-Preto N Latino Families: An overview Ethnicity and family therapy 2005 The Guilford Press 153 165 Hackethal V Spiegel S Lewis-Fernández R Kealey E Salerno A Finnerty M Towards a cultural adaptation of family psychoeducation: Findings from three Latino focus groups Community Mental Health Journal 2013 49 5 587 598 10.1007/s10597-012-9559-1 23179044 Hernández MM Bámaca-Colbert MY A behavioral process model of familism Journal of Family Theory & Review 2016 8 4 463 483 10.1111/jftr.12166 28496520 Hitchcock, J. H., Nastasi, B. K., & Summerville, M. (2010). Single-case designs and qualitative methods: Applying a mixed methods research perspective. Mid-Western Educational Researcher, 23(2). Jimenez-Gomez C Beaulieu L Cultural responsiveness in applied behavior analysis: Research and practice Journal of Applied Behavior Analysis 2022 55 337 356 10.1002/jaba.920 35266143 Juckett G Caring for Hispanic patients American Family Physician 2013 87 1 48 54 23317025 Lanesskog D Piedra LM Maldonado S Beyond bilingual and bicultural: Serving Latinos in a new-growth community Journal of Ethnic and Cultural Diversity in Social Work 2015 24 4 300 317 10.1080/15313204.2015.1027025 Lavenda RH Schultz EA Core concepts in cultural anthropology 2010 McGraw-Hill Luker K Salsa dancing into the social sciences 2008 Harvard University Press Maenner, M. J., Shaw, K. A., Bakian, A. V., Bilder, D. A., Durkin, M. S., Esler, A., ... & Cogswell, M. E. (2021). Prevalence and characteristics of autism spectrum disorder among children aged 8 years—autism and developmental disabilities monitoring network, 11 sites, United States, 2018. MMWR Surveillance Summaries, 70(11), 1. Magaña D Local voices on health care communication issues and insights on Latino cultural constructs Hispanic Journal of Behavioral Sciences 2020 42 3 300 323 10.1177/0739986320927387 Magaña SM Mental retardation research methods in Latino communities Mental Retardation 2000 38 4 303 315 10.1352/0047-6765(2000)038<0303:MRRMIL>2.0.CO;2 10981192 Manning, J. (2017). In vivo coding. In Matthes, J. (Ed.), The International Encyclopedia of Communication Research Methods. New York, NY: Wiley-Blackwell. 10.1002/9781118901731.iecrm0270 Manning J Kunkel A Researching interpersonal relationships: Qualitative methods, studies, and analysis 2014 Sage Marsiglia, F. F., Booth, J. M., Baldwin, A., & Ayers, S. (2013). Acculturation and life satisfaction among immigrant Mexican adults. Advances in Social Work, 14(1), 49–64. 10.18060/3758 Motel, S. & Patten, E. (2012). The 10 largest Hispanic origin groups: Characteristics, rankings, top counties. Pew Research Center: Hispanic trends. https://www.pewresearch.org/hispanic/2012/06/27/the-10-largest-hispanic-origin-groups-characteristics-rankings-top-counties/#:~:text=Overall%2C%20the%2010%20largest%20Hispanic,of%20the%20U.S.%20Hispanic%20population . National Autism Center. (2015). Findings and conclusions: National Standards Project, Phase 2. Randolph: National Autism Center. Neely L Gann C Castro-Villarreal F Villarreal V Preliminary findings of culturally responsive consultation with educators Behavior Analysis in Practice 2020 13 1 270 281 10.1007/s40617-019-00393-y 32231990 Nevada Legislative Counsel Bureau Audit. (2021). Performance audit: Delivery of treatment services for children with autism. Retrieved from https://www.leg.state.nv.us/audit/Full/BE2022/LA22-04%20Delivery%20of%20Treatment%20Services%20for%20Children%20With%20Autism.pdf Office of Management and Budget. (1997). Revisions to the standards for the classification of federal data on race and ethnicity. Retrieved from https://www.whitehouse.gov/wp-content/uploads/2017/11/Revisions-to-the-Standards-for-the-Classification-of-Federal-Data-on-Race-and-Ethnicity-October30-1997.pdf Pitten K How cultural values influence diagnosis, treatment and the welfare of families with an autistic child Rivier Academic Journal 2008 4 1 1 3 Polk, K. L., & Schoendorff, B. (2014). The ACT matrix: A new approach to building psychological flexibility across settings & populations. New Harbinger Publications, Inc. Rosales, R., Leon, A., Serna, R. W., Maslin, M., Arevalo, A., & Curtin, C. (2021). A first look at applied behavior analysis service delivery to Latino American families raising a child with autism spectrum disorder. Behavior Analysis in Practice, 1–10. 10.1007/s40617-021-00572-w Ruiz-Adams R Bilingual families: Using therapist-facilitated interpretation in psychotherapy Family Journal 2019 27 2 142 149 10.1177/1066480719832506 Sabogal F Marín G Otero-Sabogal R Marín BV Perez-Stable EJ Hispanic familism and acculturation: What changes and what doesn’t? Hispanic Journal of Behavioral Sciences 1987 9 4 397 412 10.1177/07399863870094003 Saldaña, J. (2021). The coding manual for qualitative researchers (4th ed.). SAGE Publications. Santiago-Rivera A Latinos values and family transitions: Practical considerations for counseling Counseling and Human Development 2003 35 6 1 12 Scharrón-del Río MR Aja AA Latinx: Inclusive language as liberation praxis Journal of Latinx Psychology 2020 8 1 7 20 10.1037/lat0000140 Slim L Celiberti D Sadavoy JA Zube ML Culturally aware practice A scientific framework for compassion and social justice: Lessons in applied behavior analysis 2021 Routledge 94 99 Smith TB Domenech-Rodríguez M Bernal G Culture Journal of Clinical Psychology 2011 67 166 175 10.1002/jclp.20757 21105069 Skinner BF The evolution of behavior Journal of the Experimental Analysis of Behavior 1984 41 2 217 221 10.1901/jeab.1984.41-217 6716037 United States Census Bureau. (2018). Hispanic population to reach 111 million by 2060. Retrieved from: https://www.census.gov/library/visualizations/2018/comm/hispanic-projected-pop.html United States Census Bureau. (2021). Retrieved from: https://www.census.gov/quickfacts/NV Wee, R. Y. (2018). States with the largest Latino and Hispanic populations. World Atlas. Retrieved from https://www.worldatlas.com/articles/us-states-with-the-largest-relative-hispanic-and-latino-populations.html Williams CM Sadavoy JA Zube ML Culturally aware practice A scientific framework for compassion and social justice: Lessons in applied behavior analysis 2021 Routledge 84 93 Zuckerman KE Sinche B Mejia A Cobian M Becker T Nicolaidis C Latino parents’ perspectives on barriers to autism diagnosis Academic Pediatrics 2014 14 3 301 308 10.1016/j.acap.2013.12.004 24767783	0	PMC9718467	NO-CC CODE	2022-12-06 23:23:39	no	Behav. Soc. Iss.. 2022 Dec 2;:1-24	utf-8	null	null	null	oa_other
==== Front Behav. Soc. Iss. Behavior and Social Issues 1064-9506 2376-6786 Springer International Publishing Cham 114 10.1007/s42822-022-00114-1 Original Paper Behavioral Interventions Contributing to Reducing Poverty and Inequities http://orcid.org/0000-0003-4796-2605 Mattaini Mark A. [email protected] 1 https://orcid.org/0000-0001-7300-1598 Roose Kathryn M. 2 https://orcid.org/0000-0003-4572-4208 Fawcett Stephen B. 3 1 grid.185648.6 0000 0001 2175 0319 Jane Addams College of Social Work, University of Illinois Chicago, PO Box 1045, Paguate, NM 87040 USA 2 Puerto Vallarta, Jalisco Mexico 3 grid.266515.3 0000 0001 2106 0692 Department of Applied Behavioral Science, University of Kansas, Lawrence, KS USA 2 12 2022 124 31 10 2022 © Association for Behavior Analysis International 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Behavioral science has a long history of engaging in efforts to understand and address socially important issues. Poverty and inequities in health and development are among the most important and complex social issues facing the world today. With its Sustainable Development Goals (SDGs), the United Nations (2015) has focused attention and guidance on addressing key global challenges, including to “end poverty” (SDG 1), “ensure good health and well-being for all” (SDG3), and “reduce inequality within and among countries” (SDG 10). In this paper, we provide a framework and illustrative examples of contributions of behavioral science to these issues. We feature illustrative behavioral interventions at the individual, relationship, community, and societal levels. We highlight the diversity of issues, intervention methods, and settings reflected in applications of behavioral science. By joining methods from behavioral science, public health, and other disciplines—and the experiential knowledge of those most affected by inequities—behavioral methods can make significant contributions to collaborative efforts to assure health and well-being for all. Keywords Poverty Inequities Behavioral interventions Public health Advocacy Policy Systems Collaborative partnerships ==== Body pmcIn 2015, the United Nations established 17 Sustainable Development Goals (SDGs) as a blueprint for achieving a better and more sustainable future for all (United Nations, 2015). In this paper, we focus on behavioral interventions that address three of the SDGs: SDG1: End poverty in all its forms everywhere; SDG3: Ensure healthy lives and promote well-being for all at all ages (linked to work on health inequities); and SDG10: Reduce inequality within and among countries. Behavioral science has a long history of engaging in efforts to understand and address socially important issues (e.g., Baer & Wolf, 1987; Baer et al., 1968; Heward et al., 2022), with particular attention to issues of poverty, health, and other inequities. In this paper, we describe how behavioral interventions are being used to address environmental conditions, behaviors, and outcomes related to problems and inequities in health and human development. From a social determinants perspective (World Health Organization, 2008), the focus is on taking action to address conditions—including social, environmental, and economic conditions—that produce unfair and unequal outcomes experienced disproportionally by some individuals and groups. The World Health Organization’s conceptual framework for social determinants calls attention to the intermediary determinants of inequities in health and development outcomes (Solar & Irwin, 2010), including (a) Differential exposures (e.g., to stress, violence) and opportunities (e.g., for employment); (b) Differential vulnerabilities (e.g., related to power, disabilities, gender) and capabilities (e.g., education, skills); and (c) Differential access (e.g., to health care) and consequences (e.g., discrimination related to race, social class). These and other social determinants can be addressed by reasonable means, including through established behavioral (and other types of) interventions. In this paper, we describe behavioral approaches grounded in the Centers for Disease Control and Prevention’s (CDC’s) Social-Ecological Model (Dahlberg & Crug, 2002), which conveys the complex interplay among influences at multiple levels—including individuals, relationships, communities, and society. The illustrative set of examples discussed in this paper, and those included in Tables 1 and 2, are not exhaustive. Rather, they offer a perspective on the diversity of possible applications, no doubt with some emphasis on those with which the authors are more familiar. The next section describes examples that address poverty, inequities, and their connections at the individual and relationship levels; and the following section offers examples at community and societal levels. Because there are reciprocal and relational connections among all levels, the two sections are best viewed as interlocked.Table 1 Illustrative behavioral interventions addressing poverty, by sector or setting engaged, and type and level of intervention Behavioral interventions addressing poverty Sector or setting engaged Type/level of intervention used Research on contingency management for substance use (e.g., Silverman et al., 2019) – Policymakers – Universities – Health Organizations – Community/civic organizations – Providing information and enhancing skills – Modifying access, barriers, and opportunities – Changing consequences Level: Individual Development and implementation of the “therapeutic workplace” which provides jobs to individuals with a history of substance use contingent on abstinence from substances (Silverman et al., 2018) – Clients – Human Service Organizations – Health Organizations – Providing information and enhancing skills – Enhancing supports and services – Modifying access, barriers, and opportunities – Changing consequences Level: Individual Emerging research on psilocybin treatment for substance use (Bogenschutz et al., 2015; Johnson et al., 2017 – Policymakers – Health Organizations – Providing information and enhancing skills – Modifying access, barriers, and opportunities Level: Individual Development of the Community Reinforcement Approach (CRA) model for treatment of substance use (Hunt & Azrin, 1973; Azrin, 1976) – Clients – Human Service Organizations – Health Organizations – Providing information and enhancing skills – Enhancing supports and services – Modifying access, barriers, and opportunities – Changing consequences Level: Individual/Relationships Development of the Community Reinforcement and Family Training (CRAFT) training family members of substance use treatment-resistant individuals to modify environments and provide reinforcement to support the sober behavior of their family member (Johnson, 1986) – Clients – Families – Human Service Organizations – Health Organizations – Providing information and enhancing skills – Enhancing supports and services – Modifying access, barriers, and opportunities – Changing consequences Level: Individual/Relationships Modifications of the CRA model to support adolescents (Dennis et al., 2004), seniors (Moyers et al., 2013) – Adolescents – Supporters (Family, Peers) – Human Service Organizations – Health Organizations – Providing information and enhancing skills – Enhancing supports and services – Modifying access, barriers, and opportunities – Changing consequences Level: Individual/Relationships Reduction of risk for poverty through a school-community initiative to reduce adolescent pregnancy (Paine-Andrews et al., 1999; Paine-Andrews et al., 2002) – Businesses – Community/Civic Organizations – Education/Schools – Health Organizations – Human Service Organizations – Media – Neighborhoods – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences Level: Community Reduction of risk for poverty through a community coalition to prevent substance abuse (Fawcett, et al., 1997). – Businesses – Community/Civic Organizations – Education/Schools – Health Organizations – Human Service Organizations –Media –Neighborhoods – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences Level: Community Community change capacity building intervention in urban neighborhood coalitions (Thompson et al., 2008). – Community/Civic Organizations – Education/Schools – Community/Civic Organizations – Neighborhoods –Providing information and enhancing skills –Enhancing services and supports –Modifying access, barriers, and opportunities –Changing consequences –Modifying policies and broader systems Level: Community Implementation and evaluation of training programs for staff of poverty self-help organizations to learn helping skills (Fawcett et al., 1976) and counseling and problem-solving skills (Whang et al., 1982), enhance leadership skills (Seekins et al., 1984), and make referrals for low-income families to access needed services (Mathews and Fawcett, 1979). – Human Service Organizations – Community/Civic Organizations – Neighborhoods –Providing information and enhancing skills –Enhancing services and supports Level: Individual/Community Implementation and evaluation of a training program for those most affected by issues to give personal testimony about their experiences in efforts to change public policies (Seekins et al., 1987a). – Policymakers – Community/Civic Organizations – Advocates – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities Level: Individual/Community Implementation and evaluation of a community concerns report method for local agenda setting in a low-income neighborhood (Schriner and Fawcett, 1988). – Policymakers – Universities – Community/Civic Organizations – Advocates – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities Level: Community Construction of cultural practices consistent with social justice (Biglan, 1995, 2015, 2016, 2020) – Policymakers – Behavioral scientists – Activists – Community members – Providing guidelines for changing cultural practices to support a just and inclusive society Level: Societal Outline of advocacy practices supporting social justice as defined by fair dispersal of benefits and burdens in society (Ardila Sánchez, Richling, et al. 2020b; Mattaini et al., 2020; Mattaini & Roose, 2021; Devlin-Foltz et al.,2012) – Policymakers – Behavioral scientists – Activists – Teachers – Community members – Presenting analytic and evaluative tools for advocacy consistent with economic justice with specific emphasis on systemic analysis and experimentation Level: Community/Societal Implementation and evaluation of the Programa Bolsa Familia, a program in Brazil that provides direct and conditional support for families experiencing poverty (Fava & Vasconcelos, 2017). – Families – Policymakers – Education/Schools – Health Organizations – Providing financial contingencies, information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences Level: Relationships/Societal Implementation of Opportunity NYC, a government sponsored program based on Programa Bolsa Familia, to provide direct and conditional support for families experiencing poverty (Holtyn et al., 2017) – Families – Policymakers – Education/Schools – Health Organizations – Attempted to use financial incentives producing minimal results Level: Relationships/Societal Promotion of intersectoral action to address chronic poverty and social exclusion (Freitas Lemos & Todorov, 2020) – Policymakers – Funders – Behavioral scientists – Activists – Community members – Describing and analyzing a study focused on a large-scale intervention across actors and sectors, with results indicating that intersectoral arrangements increased the supply and demand of apprenticeship positions and apprenticeship increased school attendance on average. Level: Societal Examination of the role of participatory community development in alleviating colonial relations and supporting independence between western and African nations (Smilak & Putnam, 2022) – Policymakers – Community members – Funders – Supporting agencies – Identifying and presenting issues of colonial relations and reciprocal and dominating/dependent relationships to funders – Recognizing collaboration as key – Expanding research in this area Level: Societal Table 2 Illustrative behavioral interventions addressing inequities in health and development, by social determinants addressed, and type and level of intervention Behavioral interventions addressing inequities Social determinant addressed Type/level of intervention used Financial incentives to increase participation in HIV treatment (Brantley et al., 2018) – Differential exposures – Differential vulnerabilities – Differential consequences – Enhancing services and supports –Modifying access, barriers, and opportunities – Changing consequences Level: Individual Financial incentives to improve medication adherence and increase vaccinations (Nowalk et al., 2010) – Differential exposures/opportunities – Differential vulnerabilities/capabilities – Differential access/consequences – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences Level: Family/Relationships Financial incentives to increase health screenings (Malotte et al., 1998) – Differential exposures/opportunities – Differential vulnerabilities/capabilities – Differential access/consequences – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences Level: Family/Relationships Analysis of potential behavioral contributions to a socially just framework for integration of immigrants to the US (Rakos & Switzer, 2021) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Analytic publication of possible enhancements to contemporary policies – Active consultations with organizations providing immigrant services Level: Community/Societal Integration of Behavioral Community Psychology in practice with homeless families (Switzer & Rakos, 2022) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Development and publication of theory-driven enhancements to contemporary policies for providing stability for homeless families – Active consultations with organizations providing homelessness services Level: Community/Societal Analysis of behavioral dynamics of police violence directed toward minority populations (Mattaini & Rehfeldt, 2020) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Analytic publication of possible enhancements to contemporary policing and legal policies and community services – Presentations to advocates of improved relations between police and communities Level: Community, Societal Analysis of alternative interventions for youth actively involved with street violence (Aspholm, 2020) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Conducting and publishing onsite research – Presentations to community and scientific groups, and advocates involved in violence prevention Level: Individual/Community Development and initiation of alternative service center for homeless youth (Holtschneider, 2021) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Establishing an alternative, and alternatively funded center on the South Side of Chicago – Providing access to multiple developmental opportunities onsite through extensive fundraising – Establishing political support – Establishing connections to needed community services Level: Individual/Community Provision of principles and successful examples of advocacy for funding from federal state and local governments (Baron & Hoeksema, 2021) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Identifying and explaining the advocacy ecosystem – Analyzing and presenting how staffers and legislators get and use research information within the financial appropriation process – Identifying actionable steps to support science in legislation Level: Societal Scientific analyses of dynamics of cultural change with focus on vulnerable populations (Biglan, 1995, 2015, 2016, 2020) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Publication of analysis and evaluations of cultural change consistent with stated values – Conference presentations on analysis and evaluations of cultural change consistent with stated values – Development of community Action Circles to reduce toxic biological and social experiences and promote prosocial behavior Level: Community/Societal Analysis and advocacy for increased emphasis on contributions of behavioral systems science to leadership for a new progressive movement integrating Israel Goldiamond’s constructional approach and M. K. Gandhi’s constructive program (Mattaini, 2015; Mattaini & Aspholm, 2016; Roose & Mattaini, 2020) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Publishing on advocacy and activist options for behavioral science – Offering collaborative conference presentations encouraging progressive advocacy options for behavioral scientists – Offering presentations to advocacy groups, and candidates for such groups Level: Community/Societal Construction of multisector matrices to develop strategic options for groups working for social justice (Seniuk et al., 2019; Mattaini, 2013). – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Integration into academic coursework – Integration into high school coursework – Presentations to professional groups Level: Community/Societal Community centered application and evaluations of the “Actively Caring for People” model (Geller, 2014). – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Extensive evaluation research in community settings – Publication of multiple books and other publications accessible to most community members – Multiple professional publications and presentations Level: Community Advocacy with congressional representatives and staff to integrate and evaluate science to policy dynamics (Crowley et al., 2018; Devlin-Foltz, et al.,2012) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Instructions for building relationships with legislators and similar officials Level: Societal Outlining strategic advocacy practices supporting social justice as defined by fair dispersal of benefits and burdens in society (Ardila Sánchez, Cihon, et al., 2020a; Mattaini et al., 2020; Mattaini & Roose, 2021; Devlin-Foltz, et al.,2012) – Differential exposures – Differential vulnerabilities and capabilities – Differential access/consequences – Sharing outline of core values, principles, and practices of scientific advocacy – Sharing outline of core advocacy repertoires – Sharing outlines of key cultural systems technologies including nested contingencies mapping, force field analysis, and use of systemic matrices Level: Community, Societal Effects of poverty clients' agenda on resource allocations by community decision makers (Fawcett, Seekins et al., 1982; Seekins & Fawcett, 1987) – Differential opportunities – Differential access/consequences – Providing information – Changing consequences – Modifying policies and broader systems Level: Community Effects of public polling data on policy consideration of utility subsidies for low-income families (Seekins, Maynard-Moody et al., 1987b) – Differential exposures – Differential access/consequences – Providing information – Changing consequences – Modifying policies and broader systems Level: Community Effects of agency-based voter registration on voter registration by low-income individuals (Fawcett, Seekins, & Silber, 1988) – Differential opportunities – Differential access – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities Level: Community Implementing a community health concerns report and microgrant program to expand initiatives for low-income residents (Paine et al., 1994, Fawcett, Seekins et al., 1982) – Differential opportunities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities Level: Community Implementing and evaluating multi-sector coalitions in low-income communities to reduce health disparities related to chronic disease (Collie-Akers et al., 2009; Schultz et al., 2009; Collie-Akers, et al., 2013) – Differential opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences – Modifying policies and broader systems Level: Community Developing and disseminating open-source, online tools for building capacity for promoting community health and development; reaching over 6 million unique users from 300 countries in past year (Community Tool Box http://ctb.ku.edu/; Holt et al., 2013) – Differential exposures and opportunities – Differential capabilities – Differential access – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities Level: Community/Societal Community mobilization to enroll low-income residents in health insurance coverage through the Affordable Care Act. (Fawcett, Sepers, et al., 2015b) – Differential exposures and opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences – Modifying policies and broader systems Level: Community Assuring health access and culturally competent health services through a Latino health coalition (Fawcett et al., 2018) – Differential exposures and opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences – Modifying policies and broader systems Level: Relationship/Community Examining the relationship between community programs and policies to prevent childhood obesity and BMI in children of different races/ethnicities (Strauss et al., 2018; Fawcett, Collie-Akers et al., 2015a) – Differential exposures and opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences – Modifying policies and broader systems Level: Community/Societal Evaluating multi-sector, multi-level responses to the Ebola outbreak in Liberia (Munodawafa et al., 2018; Hassaballa, Fawcett, Sepers, et al., 2019) and the COVID-19 pandemic in different African countries (Phori, P.M., Fawcett, S.B., Nikiema Nidjergou, et al. in press; Mwakisha, Adika et al., in press) – Differential exposures and opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences – Modifying policies and broader systems Level: Community/Societal Examining factors enabling and impeding COVID-19 response activities and incidence of cases of COVID-19 in a local public health system (Holt et al., 2021). – Differential exposures and opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences –Modifying policies and broader systems Level: Community Implementing and evaluating the effects of policy research information on adoption of child passenger safety legislation at state level (Fawcett et al., 1987; Seekins et al., 1988). – Differential exposures and opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences – Modifying policies and broader systems Level: Community/Societal Examining effects of environmental design and police enforcement on violations of handicapped parking spaces and access for people with disabilities (Suarez de Balcazar, Fawcett et al., 1988). – Differential exposures and opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences – Modifying policies and broader systems Level: Community Gathering and communicating information for policymakers on common concerns of disabled Americans [Issues and options presented to U.S. Congress, during debate on the Americans with Disabilities Act]. (Suarez de Balcazar, Bradford et al., 1988). – Differential opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities – Changing consequences – Modifying policies and broader systems Level: Community/Societal Implementing and evaluating an advocacy training program for people with disabilities to influence city policy related to disability rights (Balcazar et al., 1990). – Differential opportunities – Differential vulnerabilities and capabilities – Differential access/consequences – Providing information and enhancing skills – Enhancing services and supports – Modifying access, barriers, and opportunities Level: Individual/Community Individual and Relationship-Focused Interventions Using Behavioral Approaches Over the past 50 years, the field of behavior science has developed a body of research on behavioral interventions addressing poverty and inequities at the individual and relationship (particularly family) levels. (Many of these could potentially be scaled into community-wide interventions and larger-scale policy solutions.) The descriptions of behavioral interventions convey the diversity of issues and intermediary determinants addressed, sectors or settings engaged, and types of behavioral intervention components used. See Tables 1 and 2 for a fuller set of interventions Supporting Children and Families: Education Many behavioral interventions have been developed and implemented to improve educational outcomes for children in low-income families and neighborhoods. For instance, established in 1961, the Juniper Gardens Children’s Project (JGCP) has a sustained program of research to improve the educational experiences of children, especially of those living in low-income areas of Greater Kansas City (Kansas and Missouri). Part of the Life Span Institute of the University of Kansas, the JGCP has worked to address a number of education issues disproportionally affecting low-income families, including quality of early childhood education, school readiness, and school performance (Greenwood, 2006; Hall, Schiefelbusch et al., 2006). The Juniper Garden’s team has also played a leadership role with the Bridging the Word Gap (BWG) Research Network that aims to reduce the vocabulary gap by enriching early language experiences of young children in poverty (Greenwood, Carta et al., 2017). The BWG Research Network, a collaboration among over 100 researchers and practitioners, aims to scale up impact by disseminating information and fostering widespread adoption of education activities that promote language acquisition and school readiness. Additional important work by Rob Horner and colleagues in Portland, Oregon, in establishing the Positive Behavioral Interventions and Supports (PBIS) program has been effective in improving learning and social relations in thousands of schools across the United States (Horner & Kittelman, 2021). Other key examples include the multinational Comprehensive Application of Behavior Analysis to Schooling (CABAS) directed by R. Douglas Greer, which has produced dramatic results for children with disabilities and emotional disturbance, as well as mainstream students (Greer, 1997); and Morningside Academy in Seattle (Johnson & Street, 2014), which emphasizes Precision Teaching and the Morningside Model of Generative Instruction (MMGI, Johnson et al., 2021). Individual Interventions: Treatment Contexts A wealth of behavioral research and intervention focuses on the intersections of poverty, health, and social conditions. To illustrate, we focus on the work of the Center for Learning and Health, Department of Psychiatry and Behavioral Sciences at the Johns Hopkins University School of Medicine. Silverman and colleagues from that center suggested two levels of intervention for poverty and health: proximal, promoting healthy behaviors directly, and distal, targeting risk factors for poor health to impact health indirectly (Silverman et al., 2019). With respect to proximal interventions, Silverman and colleagues (2019) indicate that the most effective application of behavioral approaches to addressing substance abuse and associated poverty is to use direct reinforcement of abstinence, offering an incentive for providing proof of abstinence (e.g., urinalysis). Monetary rewards have been shown to be the most effective in this type of research (e.g., Benishek et al., 2014; Davis et al., 2016). Reinforcement or incentives, in the form of monetary vouchers, has been successful in increasing the probability of enrollment in treatment for opiate abuse (e.g., Holtyn et al., 2021), promoting abstinence from alcohol (e.g., Koffarnus, DeFulio et al., 2021), cocaine (e.g., Silverman et al., 1996), opiates (e.g., Robles et al., 2002), and other applications (e.g., polydrug abuse, injection substance users). Another example of an incentive-based intervention is Brantley and colleagues’ (2018) research with clients diagnosed with HIV, which found that financial incentives increased engagement in care; with 70.1% of participants engaged in HIV treatment at enrollment increasing to 98.6% during the first year of the study, and 96.3% during the second year of the study. As a result, levels of viral suppression increased from 57.7% of participants at enrollment to 82.7% after 12 months of participation in the program. Financial incentives have also been used to improve medication adherence and increase vaccinations (e.g., Nowalk et al., 2010), and promote health screenings (e.g., Malotte et al., 1998)—all behavioral interventions related to health with implications for poverty and financial stability. The “therapeutic workplace” is another example of a context for applying incentives to improve financial stability (Silverman et al., 2018; Silverman et al., 2001). In this behavioral intervention, unemployed adults with a history of substance abuse are employed and receive a salary, and in some cases are able to earn an increased salary contingent upon proof of drug abstinence and/or taking addiction treatment medications (e.g., methadone). Such cash incentives improved attendance and performance in these training programs (e.g., Silverman et al., 2018; Koffarnus, DeFulio et al., 2013; Koffarnus, Wong et al., 2013b). Following training, participants may participate in the therapeutic workplace in which they demonstrate abstinence through urine samples, or comply with abstinence-promoting medication regimens to maintain their employment and/or to receive maximum pay for their work. Family Interventions: Bolsa Família Compared to the robust history of behavioral research on treatment issues such as substance abuse, research on addressing poverty is still emerging. Some behavioral interventions have been implemented as part of large-scale anti-poverty programs targeting employment, education, and heath. The Programa Bolsa Família is a government funded cash transfer program designed to support Brazilian families by intervening directly on the causes of poverty. The program has three phases: direct income transfer for immediate relief of poverty; conditional income transfer upon meeting certain program requirements; and supplementary programs to support sustained social and economic vulnerability (Fava & Vasconcelos, 2017). In the conditional phase, families must meet specific goals relating to school attendance and not allowing children and adolescents to work, vaccine schedules, prenatal and postnatal medical care for pregnant people, and participation in services designed to strengthen family ties. Researchers have analyzed the program from a behavior-analytic perspective focusing on rule-governed behavior (Kaiser et al., 2016), and as a cultural intervention (Freitas Lemos & Todorov, 2020; Valderlon & Elias, 2019). Outcomes have been widely studied, with positive effects on school attendance (Bourguignon et al., 2002), child mortality and poverty-related infectious diseases (e.g., Nery et al., 2017; Rasella et al., 2013), maternal mortality (Rasella et al., 2021), birth weight (Rasella et al., 2013), child labor (Rawlings et al., 2005), and suicide rates (Alves et al., 2018). Programa Bolsa Família was replaced in November of 2021 by a new program, Auxílio Brasil; and the results of associated changes in eligibility and program elements are not yet known. Based on Brazil’s Bolsa Familia Program, the government-sponsored Opportunity NYC (New York City) provided financial incentives to families living in poverty for achieving goals related to education, health care, and employment (described in Silverman et al., 2019). The program was associated with reducing poverty and extreme poverty in the incentive group, increased savings for some families, and increased usage of preventive dental care (Riccio, 2010). Employment rates between groups were similar, 63% in the incentive group and 65% in the control group; participation in education and training classes was low for both groups (less than 6% of the families); both groups had low employment rates (and low wages) after the program ended (Holtyn et al., 2017). Family Interventions: Homelessness Switzer and Rakos (2022), who are active in efforts to combat family homelessness, provided a behavioral community psychology1 analysis for addressing poverty related homelessness while working within the guidelines of the federal Housing First/Rapid Re-Housing (HF/RRH) approach. Grounded in the federal 2009 Homeless Emergency Assistance and Rapid Transition to Housing Act, Housing First acknowledges that “housing itself is a precondition for resolving risks to housing stability such as mental health and substance abuse, incomplete education, and unemployment/underemployment” (Switzer & Rakos, 2022, p. 5). Switzer and Rakos have developed a comprehensive approach that reflects cultural, sociological, and ecological factors that influence the behaviors of multiple persons interacting in families and groups; they offer potential points of intervention that could improve health and development outcomes for low-income families. Community and Societal Interventions Using Behavioral Approaches Community Interventions Behavioral scientists have been interested in community-level interventions since quite early in the development of the discipline in the 1960s and 1970s (Fawcett, 1991; Watson-Thompson et al., 2020). Both the Journal of Applied Behavior Analysis, and the Behaviorists for Social Action Journal2 (now Behavior and Social Issues) were established, demonstrating strong commitment to the application of behavioral methods to address issues of social significance (Baer & Wolf, 1987; Baer et al., 1968). The subdiscipline of behavioral community psychology, which drew on other disciplines, particularly public health and human development, significantly engaged with issues of poverty and inequities. For instance, researchers in the department of applied behavioral science (and the Center for Community Health and Development) at the University of Kansas established a 50+ year commitment to this work, viewing community development as both a process and a product that required community engagement over time (Fawcett, 2021). Such community-engaged research was guided by a core set of values, including (a) Community researchers should form collaborative relationships with the participants with whom they do research; (b) Community research should provide information about the variety of behavior-environment relationships of importance to communities; and (c) Community action should occur at the level of change and timing likely to optimize beneficial outcomes (Fawcett, 1991). This perspective has proven helpful in the development and expansion of community interventions supporting health and development, in partnership with community members and organizations. Among projects in which faculty and students from this program have focused over time include those promoting healthy youth development (e.g., Watson-Thompson et al. 2008; Watson-Thompson et al., 2013), reductions in substance abuse (e.g., Anderson-Carpenter et al. 2016), and building multisectoral partnerships for population health and health equity (Fawcett et al., 2010). To build capacity for this work of promoting community health and development, the Center for Community Health and Development at the University of Kansas developed and maintains the Community Tool Box (http://ctb.ku.edu/). Begun in 1994, this open-source resource reaches six million unique users annually in 300 countries (Holt et al., 2013). Prevention science is another important approach to community intervention; it often uses behavioral interventions to reduce risk factors, and increase protective factors, especially for behaviors and outcomes related to child and youth health and development. Prevention researchers (e.g., Hawkins et al., 2002) identified a matrix of risk factors for adolescent behaviors related to poverty and inequities such as substance abuse, violence, delinquency, and dropping out of school. Such matrix thinking reflects the reality that most social issues are embedded in multiple interactive and systemic processes (Biglan, 1995; Seniuk et al., 2019). In the spirit of application, Anthony Biglan, an important behavior scientist at the Oregon Research Institute, and other prevention scientists have developed the Values to Action Coalition, which works to bring together a “community of scientists, clinicians, and concerned citizens that seeks to leverage behavioral science, effect change, and make the world a better place for everyone” (https://www.valuestoaction.org/), as well as organizing “action circles” to strengthen local communities in applying behavioral science for social change. There are many other behavioral scientists involved in reducing social problems by strengthening communities. For instance, E. Scott Geller (2014, 2017) from Virginia Tech has developed and actively researched community intervention for over 50 years in the United States and other nations. This includes applications to increase work and traffic safety, prevent substance abuse, reintegrate incarcerated felons, and strengthen community commitment of police officers; often by increasing rates of public positive reinforcement within organizational structures. In another example, the LYTE Collective in Chicago, which has developed a new form of locally grounded partnership with homeless youth, has worked to develop a culture of respectful, reciprocal relationships (consistent with Biglan’s and Geller’s approaches); in part this is due to some lead participants’ exposure to behavioral science principles during education at Jane Addams College of Social Work at the University of Illinois Chicago (Holtschneider, 2021). Societal Interventions Many of the examples included for Individual, Relationship, and Community levels could only be implemented through advocacy efforts seeking policy and systems change at multiple levels. Research on social justice and social policy is common in several behavioral science journals. For example, a scan of Perspectives on Behavior Science includes 231 references to “social policy”; Behavior and Social Issues includes 229, and the Journal of Applied Behavior Analysis, 236. For instance, Seekins and Fawcett (1986) published a paper entitled “Public Policy Making and Research Information”, and two years later the Association for Behavior Analysis (now the Association for Behavior Analysis: International—ABAI) sponsored a statement from their Task Force on Public Policy, accompanied by an article by Fawcett et al. (1988) titled “Behavior Analysis and Public Policy.” Faculty and students from a number of different university research centers have been engaged in such work since the 1980s. There are many other useful models of behavioral scientists working at the policy level. For instance, in 1995, Biglan published an inspiring and valuable book, Changing Cultural Practices: A Contextualist Framework for Intervention Research. That volume presented a general framework for cultural change, and included policy discussions related to tobacco use, childrearing practices, sexist behaviors, and environmentally harmful practices (see also Biglan, 2016, for an update). Biglan and colleagues have made continuing contributions to policy advocacy ever since, including two more heavily policy-focused books (Biglan, 2015, 2020). Mattaini and Aspholm (2016), in the paper “Contributions of Behavioral Systems Science to Leadership for a New Progressive Movement,” called for widespread social action and advocacy, including promoting and sustaining scientific activism, advanced systems science education, and protecting and caring for those harmed by unjust environments, all in collaboration with existing non-governmental organizations (NGOs). Many other behavioral scientists have engaged in advocacy and systems change efforts addressing poverty and inequities. Examples include research and action to address chronic poverty and social exclusion in Brazil (Freitas Lemos et al., 2019; Freitas Lemos & Todorov, 2020); challenging conditions related to colonialism in Africa (Smilak & Putnam, 2022); increasing supports for people immigrating to the United States (Rakos & Switzer, 2021); and working on improvements in services for unhoused persons in the United States (Holtschneider, 2021; Switzer & Rakos, 2022). Mattaini and colleagues have also developed an analysis of possible ways to reduce violence by police in the United States (Mattaini & Rehfeldt, 2020), promote youth activism as violence prevention (Aspholm & Mattaini, 2017; Roose & Mattaini, 2020), and (referencing Gandhi) establish a “constructive program” for the twenty-first century (Mattaini, 2015). Effective policy advocacy typically requires applications of strategic influence skills (Baron & Hoeksema, 2021). Such influence may be directed specifically toward the actions of decision-makers, but it is more common to focus at least in part on individuals and organizations in positions to establish or modify the actions of decision-makers. In most cases, those policy changes require engagement of those most affected as well as people and organizations that can influence decision makers. Such engagement is often required over extended periods of time to build ongoing relationships necessary to influence policy adoption and implementation. Some Challenges and Opportunities for Contribution This paper outlines ways that behavioral approaches have been used to address poverty and inequities in health and human development. The focus is on taking action to address conditions—including social, environmental, and economic conditions—that cause poverty and produce unfair and unequal outcomes in health and human development. Widespread behavior change will be needed—including among those most affected and those with the power to change programs, policies, and practices—if we are to assure the conditions for health and development for all. Behavioral science brings particular strengths to this work, especially in measurement and intervention. Behavioral science methods have been developed for the reliable and valid measurement of behavior (e.g., different health behaviors such as healthy eating and physical activity) and the products of those behaviors, including healthy weight (Fischer et al., 2022). It has also developed methods for measuring changes in the community and system (e.g., new or modified programs, policies, and practices) and their association with community-level outcomes (e.g., Fawcett, Collie-Akers, et al., 2015a; Strauss et al., 2018). Behavioral science has contributed to a wide variety of evidence-based interventions grounded in the science of behavior (as illustrated by Tables 1 and 2). There are a number of challenges to be engaged to realize the potential of behavioral science methods for helping address poverty and inequities in health and development. First, the bulk of tested behavioral interventions use relatively weak forms of behavior change (e.g., providing information and enhancing skills), rather than stronger forms (e.g., modifying policies and broader systems) that are often less feasible to implement. Second, the preponderance of behavioral interventions have been tested at lower socio-ecological levels (i.e., individuals, relationships), not at broader levels (i.e., community, society). Third, to achieve greater impact, behavioral interventions will need to be designed and implemented with greater strength and broader reach (for instance, more comprehensive forms that include modifying barriers/opportunities and policy and systems change). Fourth, efforts to change conditions that produce inequities will meet with resistance and opposition (e.g., delaying, denying, discrediting). To bring about needed systems change, behavioral scientists and practitioners must join with advocacy groups —and organized groups of those most affected—to exert necessary influence on decision makers (Fawcett, 1999). Collaborative partnerships provide a promising context for the contribution of behavioral scientists and practitioners to address poverty and inequities in health and development (Fawcett et al., 2010). They assure structures for sharing risks, resources, and responsibilities for the work needed to change conditions in multiple sectors, settings, and levels. By joining methods from behavioral science, public health, and other disciplines—and experiential knowledge of those most affected by inequities—we can work more effectively to assure conditions for health and well-being for all. Authors’ Contributions The authors agree that all authors contributed equally to this work. Data Availability Data sharing not applicable to this article as no datasets were generated or analyzed during the current study. Declarations Conflicts of Interest The authors have no known conflict of interest to disclose. 1 More detail on behavioral community psychology is discussed in the next section. 2 The journal Behavior and Social Issues was previously named Behaviorists for Social Action Journal (1978–1981) and Behavior Analysis and Social Action (1982–1990). Both predecessor journals can be found at: https://link.springer.com/journal/43038/volumes-and-issues ==== Refs References Alves FJO Machado DB Barreto ML Effect of the Brazilian cash transfer programme on suicide rates: A longitudinal analysis of the Brazilian municipalities Social Psychiatry and Psychiatric Epidemiology 2018 54 5 599 606 10.1007/s00127-018-1627-6 30456426 Anderson-Carpenter K Watson-Thompson J Chaney L Jones M Reducing binge drinking in adolescents through implementation of the strategic prevention framework American Journal of Community Psychology 2016 57 1 36 46 10.1002/ajcp.12029 27217310 Ardila Sánchez JG Cihon TM Malott ME Mattaini MA Rakos RF Rehfeldt RA Richling SM Roose KM Seniuk HA Watson-Thompson J Collective editorial: Ten guidelines for strategic social action Behavior and Social Issues 2020 29 1 15 30 10.1007/s42822-020-00038-8 Ardila Sánchez JG Richling SM Benson ML Rakos RF Cihon TM Mattaini MA Activism, advocacy, and accompaniment Behavior science perspectives on culture and community 2020 Springer 413 436 Aspholm, R. (2020). Views from the streets: The transformation of gangs and violence on Chicago's south side. Columbia University Press. 10.7312/asph18772 Aspholm RR Mattaini MA Sturmey P Youth activism as violence prevention The Wiley handbook of violence and aggression 2017 Wiley 1 12 Azrin NH Improvements in the community reinforcement approach to alcoholism Behaviour Research and Therapy 1976 14 5 339 348 10.1016/0005-7967(76)90021-8 971207 Baer DM Wolf MM Some still-current dimensions of applied behavior analysis Journal of Applied Behavior Analysis 1987 20 4 313 327 10.1901/jaba.1987.20-313 16795703 Baer DM Wolf MM Risley TR Some current dimensions of applied behavior analysis Journal of Applied Behavior Analysis 1968 1 1 91 97 10.1901/jaba.1968.1-91 16795165 Balcazar FE Seekins T Fawcett SB Hopkins BL Empowering people with physical disabilities through advocacy skills training American Journal of Community Psychology 1990 18 281 295 10.1007/BF00931305 2143052 Baron J Hoeksema MJ Science advocacy 101: Realizing the benefits, overcoming the challenges Behavior and Social Issues 2021 30 1 121 138 10.1007/s42822-021-00069-9 Benishek LA Dugosh KL Kirby KC Matejkowski J Clements NT Seymour BL Festinger DS Prize-based contingency management for the treatment of substance abusers: A metaanalysis Addiction (Abingdon, England) 2014 109 9 1426 1436 10.1111/add.12589 24750232 Biglan, A. (1995). Changing cultural practices: A contextualist framework for intervention research. Context Press. Biglan A The nurture effect: How the science of human behavior can improve our lives and our world 2015 New Harbinger Publications Biglan A The need for a more effective science of cultural practices The Behavior Analyst 2016 39 1 97 107 10.1007/s40614-016-0051-z 27606190 Biglan A Rebooting capitalism: How we can forge a society that works for everyone 2020 Values to Action Bogenschutz, M. P., Forcehimes, A. A., Pommy, J. A., Wilcox, C. E., Barbosa, P. C., & Strassman, R. J. (2015). Psilocybin-assisted treatment for alcohol dependence: a proof-of-concept study. Journal of psychopharmacology, 29(3), 289–299. 10.1177/0269881114565144 Bourguignon F Ferreira FH Leite PG Ex-ante evaluation of conditional cash transfer programs: The case of Bolsa Escola 2002 World Bank Publications Brantley AD Burgess S Bickham J Wendell D Gruber D Using financial incentives to improve rates of viral suppression and engagement in care of patients receiving HIV care at 3 health clinics in Louisiana: The health models program, 2013–2016 Public Health Reports 2018 133 2_suppl 75S 86S 10.1177/0033354918793096 30457949 Collie-Akers VL Fawcett SB Schultz JA Measuring progress of collaborative action in a community health effort Revista Panamericana de Salud Pública/Pan American Journal of Public Health. 2013 34 6 422 428 24569971 Collie-Akers VL Schultz JA Carson VL Fawcett SB Evaluating mobilization of neighborhood and faith organizations to reduce risk for health disparities within a REACH 2010 coalition in Kansas City Missouri. Health Promotion Practice 2009 10 2 118S 127S 10.1177/1524839908331271 Crowley M Scott J Fishbein D Translating prevention research for evidence-based policymaking: Results from the research-to-policy collaboration pilot Prevention Science 2018 19 2 260 270 10.1007/s11121-017-0833-x 28849362 Dahlberg, L. L., & Krug, E. G. (2002). Violence: A global public health problem. In E. Krug, L. L. Dahlberg, J. A. Mercy, A. B. Zwi, & R. Lozano (Eds.), World report on violence and health (pp. 1–21). World Health Organization. https://apps.who.int/iris/bitstream/handle/10665/42495/9241545615_eng.pdf. Accessed 12 Nov 2022. Davis DR Kurti AN Skelly JM Redner R White TJ Higgins ST A review of the literature on contingency management in the treatment of substance use disorders, 2009–2014 Preventive Medicine 2016 92 36 46 10.1016/j.ypmed.2016.08.008 27514250 Dennis, M., Godley, S. H., Diamond, G., Tims, F. M., Babor, T., Donaldson, J., ... & Funk, R. (2004). The Cannabis Youth Treatment (CYT) Study: main findings from two randomized trials. Journal of substance abuse treatment, 27(3), 197–213. 10.1016/j.jsat.2003.09.005 Devlin-Foltz D Fagen MC Reed E Medina R Neiger BL Advocacy evaluation: Challenges and emerging trends Health Promotion Practice 2012 13 5 581 586 10.1177/1524839912446482 22773623 Fava VMD Vasconcelos LA Behavior of Programa Bolsa Família beneficiaries. A behavior analytic perspective on fulfillment of education and health conditionalities Behavior and Social Issues 2017 26 1 156 171 10.5210/bsi.v26i0.7825 Fawcett, S. B. (2021). A reflection on community research and action as an evolving practice. Behavior and Social Issues., 30, 535–544. 10.1007/s42822-021-00083-x. Fawcett SB Some values guiding community research and action Journal of Applied Behavior Analysis 1991 24 621 636 10.1901/jaba.1991.24-621 16795759 Fawcett SB Rothman J Some lessons on community organization and change Reflections on community organization: Enduring themes and critical issues 1999 F. E. Peacock Publishers 314 334 Fawcett SB Collie-Akers V Schultz J Kelley M Measuring community programs and policies in the healthy communities study American Journal of Preventive Medicine 2015 49 4 636 641 10.1016/j.amepre.2015.06.027 26384934 Fawcett SB Lewis RK Paine-Andrews A Francisco VT Richter KP Williams EL Copple B Evaluating community coalitions for the prevention of substance abuse: The case of project freedom Health Education & Behavior 1997 24 6 812 828 10.1177/109019819702400614 9408793 Fawcett SB Mathews RM Fletcher RK Morrow R Stokes TF Personalized instruction in the community: Teaching helping skills to low-income neighborhood residents Journal of Personalized Instruction 1976 1 2 86 90 Fawcett, S., Schultz, J., Watson-Thompson, J., Fox, M., & Bremby, R. (2010). Building multisectoral partnerships for population health and health equity. Preventing Chronic Disease, 7(6), 1–7. http://www.cdc.gov/pcd/issues/2010/nov/10_0079.htm. Accessed 12 Nov 2022. Fawcett SB Seekins T Jason LA Policy research and child passenger safety legislation: A case study and experimental evaluation Journal of Social Issues 1987 43 133 148 10.1111/j.1540-4560.1987.tb01300.x Fawcett SB Seekins T Silber L Low-income voter registration: A small-scale evaluation of an agency-based registration strategy American Journal of Community Psychology 1988 16 5 751 758 10.1007/BF00930024 Fawcett SB Seekins T Whang PL Muiu C Suarez de Balcazar Y The concerns report method: Involving consumers in setting local improvement agendas Social Policy 1982 13 4 36 41 10260485 Fawcett SB Sepers CE Jones J Jones L McKain W Participatory evaluation of a community mobilization effort to enroll residents of Wyandotte County, Kansas residents through the affordable care act American Journal of Public Health 2015 105 S3 S433 S437 10.2105/AJPH.2014.302505 25905820 Fawcett SB Torres J Jones L Moffett M Bradford K Ramirez Mantilla M Cupertino AP Bravo de los Rios, J., & Collie-Akers, V. Assuring health access and culturally competent health services through the Latino health for all coalition Health Promotion Practice 2018 19 5 765 774 10.1177/1524839918781525 29991288 Fischer, E. F., Paz Lemus, T., Reichert, A., Alsopp, M., & Harvey, T. S. (2022). Reframing childhood obesity: Cultural insights on nutrition, weight, and food systems. Vanderbilt Cultural Contexts of Health Initiative. https://www.vanderbilt.edu/cultural-contexts-health/wp-content/uploads/sites/350/2022/07/Reframing-Childhood-Obesity-CCH-Report-Final-7.pdf. Accessed 12 Nov 2022. Freitas Lemos R Favacho CRN Favilla KC Baia FH Managing environmental policies: Lessons from traditional communities Behavior and Social Issues 2019 28 1 269 297 10.1007/s42822-019-00022-x Freitas Lemos R Todorov JC Promoting intersectoral action to address chronic poverty and social exclusion Behavior and Social Issues 2020 29 1 35 51 10.1007/s42822-020-00029-9 Geller, E. S. (2014). Actively caring for people: Cultivating a culture of compassion. Make-A-Difference, LLC. Geller ES Actively caring for people: Policing 2017 Morgan James Publishing Greenwood CR Schiefelbusch RL Schroeder SR The Juniper gardens Children’s project Doing science and doing good 2006 Brookes 327 349 Greenwood CR Carta JJ Walker D Watson-Thompson J Gilkerson J Larson AL Schnitz A Conceptualizing a public health prevention intervention for bridging the 30 million word gap Clinical Child and Family Psychology Review 2017 20 1 3 24 10.1007/s10567-017-0223-8 28150059 Greer RD The comprehensive application of behavior analysis to schooling (CABAS®) Behavior and Social Issues 1997 7 59 63 10.5210/bsi.v7i1.300 Hall RV Schiefelbusch RL Greenwood CR Hoyt RK Schiefelbusch RL Schroeder SR The Juniper gardens Children’s project Doing science and doing good 2006 Brookes 125 139 Hassaballa I Fawcett S Sepers C Reed F Schultz J Munodawafa D Phori P Chiriseri E Kouadio K Participatory monitoring and evaluation of Ebola response activities in Lofa County, Liberia: Some lessons learned International Quarterly of Community Health Education 2019 40 1 57 66 10.1177/0272684X19846742 31081453 Hawkins JD Catalano RF Arthur MW Promoting science-based prevention in communities Addictive Behaviors 2002 27 6 951 976 10.1016/s0306-4603(02)00298-8 12369478 Heward, W. L., Critchfield, T. S., Reed, D. D., Detrich, R., & Kimball, J. W. (2022). ABA from a to Z: Behavior science applied to 350 domains of socially significant behavior. Perspectives on Behavior Science, 1–33. 10.1007/s40614-022-00336-z Holt CM Fawcett SB Hassaballa-Muhammed R Partridge D Jordan S in collaboration with the local COVID-19 Unified Command Participatory monitoring and evaluation of the COVID-19 response in a local public health system Health Promotion Practice 2021 22 6 750 757 10.1177/15248399211041085 34590896 Holt CM Fawcett SB Schultz JA Jones JA Disseminating online tools for building capacity among community practitioners Journal of Prevention and Intervention in the Community 2013 41 3 201 211 10.1080/10852352.2013.788349 23751063 Holtschneider C But how homeless are you? Toward a more just and effective response to youth homelessness Behavior and Social Issues 2021 30 1 139 147 10.1007/s42822-021-00053-3 Holtyn AF Jarvis BP Silverman K Behavior analysts in the war on poverty: A review of the use of financial incentives to promote education and employment Journal of the Experimental Analysis of Behavior 2017 107 1 9 20 10.1002/jeab.233 28078664 Holtyn AF Toegel F Novak MD Leoutsakos JM Fingerhood M Silverman K Remotely delivered incentives to promote buprenorphine treatment engagement in out-of-treatment adults with opioid use disorder Drug and Alcohol Dependence 2021 225 108786 10.1016/j.drugalcdep.2021.108786 34087746 Horner RH Kittelman A Advancing the large-scale implementation of applied behavior analysis Behavior and Social Issues 2021 30 94 105 10.1007/s42822-021-00049-z Hunt GM Azrin NH A community-reinforcement approach to alcoholism Behaviour Research and Therapy 1973 11 1 91 104 10.1016/0005-7967(73)90072-7 4781962 Johnson, K., & Street, E. M. (2014). Precision teaching. The Wiley Blackwell handbook of operant and classical conditioning, 581–609. 10.1002/9781118468135.ch23 Johnson K Street EM Kieta AR Robbins JK The Morningside model of generative instruction: Bridging the gap between skills and inquiry teaching 2021 Cambridge Center for Behavioral Studies Johnson MW Garcia-Romeu A Griffiths RR Long-term follow-up of psilocybin-facilitated smoking cessation The American Journal of Drug and Alcohol Abuse 2017 43 1 55 60 10.3109/00952990.2016.1170135 27441452 Johnson, V. E. (1986). Intervention: How to help someone who doesn’t want help? Johnson Institute Books. Kaiser, F. A., Haydu, V. B., & Gallo, A. E. (2016). Uma interpretação analítico-comportamental de contingências estabelecidas pelo Programa Bolsa Família [a behavior-analytic interpretation of contingencies established by the Bolsa Família program]. Revista Brasileira de Terapia Comportamental e Cognitiva, 17(3), 70–83 Retrieved from http://www.usp.br/rbtcc/index.php/RBTCC/article/view/816/450. Accessed 12 Nov 2022. Koffarnus MN DeFulio A Sigurdsson SO Silverman K Performance pay improves engagement, progress, and satisfaction in computer-based job skills training of low-income adults Journal of Applied Behavior Analysis 2013 46 2 395 406 10.1002/jaba.51 24114155 Koffarnus MN Kablinger AS Kaplan BA Crill EM Remotely administered incentive-based treatment for alcohol use disorder with participant-funded incentives is effective but less accessible to low-income participants Experimental and Clinical Psychopharmacology 2021 29 5 555 565 10.1037/pha0000503 34110885 Koffarnus MN Wong CJ Fingerhood M Svikis DS Bigelow GE Silverman K Monetary incentives to reinforce engagement and achievement in a job-skills training program for homeless, unemployed adults Journal of Applied Behavior Analysis 2013 46 3 582 591 10.1002/jaba.60 24114221 Malotte CK Rhodes F Mais KE Tuberculosis screening and compliance with return for skin test reading among active drug users American Journal of Public Health 1998 88 5 792 796 10.2105/AJPH.88.5.792 9585747 Mathews RM Fawcett SB Community information systems: Analysis of an agency referral program Journal of Community Psychology 1979 7 4 281 289 10.1002/1520-6629 Mattaini MA Strategic nonviolent power: The science of satyagraha 2013 Athabasca University Press Mattaini, M. A. (2015). Toward a twenty-first century, science based “constructive programme.” In R. Amster, L. Finley, E. Pries, & R. McCutcheon (Eds.), Peace studies between tradition and innovation. Cambridge Scholar Publishing. Mattaini MA Aspholm R Contributions of behavioral systems science to leadership for a new progressive movement The Behavior Analyst 2016 39 1 109 121 10.1007/s40614-015-0043-4 27606185 Mattaini MA Esquierdo-Leal JL Ardila Sánchez JG Richling SM Ethridge AN Cihon TM Mattaini MA Public policy advocacy in culturo-behavior science Behavior science perspectives on culture and community 2020 Springer 385 412 Mattaini MA Rehfeldt RA Rendezvous with truth and discovery Behavior and Social Issues 2020 29 1 1 14 10.1007/s42822-020-00034-y Mattaini MA Roose KM Emerging culturo-behavior science contributions to global justice Behavior and Social Issues 2021 30 1 215 236 10.1007/s42822-021-00073-z Moyers, T. B., Andersen, K., Behrendt, S., Braun, B., Bogenschutz, M., Bühringer, G., Forcehimes, A. A., & Nielsen, A. S. (2013). Community Reinforcement Approach for Seniors (CRAS) Treatment Manual for Project Elderly, Extended Condition (MET+ CRA) Study Version. Retrieved from: https://sdu.azureedge.net/-/media/files/om_sdu/centre/ucar/materiale/cras+-+community+reinforcement+approach+seniors.pdf. Accessed 12 Nov 2022. Munodawafa D Moeti M Phori PM Fawcett SB Hassaballa I Sepers C Reed FD Schultz J Chiriseri ET Monitoring and evaluating the Ebola response effort in two Liberian communities Journal of Community Health 2018 43 2 321 327 10.1007/s10900-017-0425-5 28929318 Mwakisha, J., Adika, B., Nyawade, S., Phori, P., NiNidjergou, N., Silouakadila, C., & Fawcett, S. B. (in press). Kenya’s experience: Factors enabling and impeding the COVID-19 response. Health Promotion Practice. Nery JS Rodrigues LC Rasella D Aquino R Barreira D Torrens AW Boccia D Penna GO Penna MLF Barreto ML Pereira SM Effect of Brazil's conditional cash transfer programme on tuberculosis incidence The International Journal of Tuberculosis and Lung Disease 2017 21 7 790 796 10.5588/ijtld.16.0599 28633704 Nowalk MP Lin CJ Toback SL Rousculp MD Eby C Raymund M Zimmerman RK Improving influenza vaccination rates in the workplace: A randomized trial American Journal of Preventive Medicine 2010 38 3 237 246 10.1016/j.amepre.2009.11.011 20036102 Paine AL Francisco VT Fawcett SB Assessing community health concerns and implementing a microgrant program for self-help initiatives American Journal of Public Health 1994 84 2 316 318 8296965 Paine-Andrews A Fisher J Berkley-Patton J Fawcett SB Williams EL Lewis R Harris K Analyzing the contribution of community change to population health outcomes in an adolescent pregnancy prevention initiative Health Education & Behavior 2002 29 2 183 193 10.1177/1090198102029002004 11942713 Paine-Andrews A Harris KJ Fisher JL Lewis RK Williams EL Fawcett SB Vincent ML Effects of a replication of a school/community model for preventing adolescent pregnancy in three Kansas communities Family Planning Perspectives 1999 31 4 182 189 10.2307/2991591 10435217 Phori, P. M., Fawcett, S. B., Nikiema Nidjergou, N., Silouakadila, C., Ruaa Hassaballa-Muhammed, R., & Kakule Siku, D. (2022). Participatory monitoring and evaluation of the COVID-19 response in the Africa region. Health Promotion Practice.10.1177/15248399221095524 Rakos RF Switzer K Culturo-behavior science: Now is the time to focus on US immigration policy Behavior and Social Issues 2021 30 1 148 169 10.1007/s42822-021-00062-2 Rasella D Alves FJO Rebouças P de Jesus GS Barreto ML Campello T Paixao ES Long-term impact of a conditional cash transfer programme on maternal mortality: A nationwide analysis of Brazilian longitudinal data BMC Medicine 2021 19 1 1 9 10.1186/s12916-021-01994-7 33390155 Rasella D Aquino R Santos CA Paes-Sousa R Barreto ML Effect of a conditional cash transfer programme on childhood mortality: A nationwide analysis of Brazilian municipalities The Lancet 2013 382 9886 57 64 10.1016/S0140-6736(13)60715-1 Rawlings LB Rubio GM Evaluating the impact of conditional cash transfer programs The World Bank Research Observer 2005 20 1 29 55 10.1093/wbro/lki001 Riccio J Sharing lessons from the first conditional cash transfer program in the United States 2010 National Poverty Center Robles E Stitzer ML Strain EC Bigelow GE Silverman K Voucher-based reinforcement of opiate abstinence during methadone detoxification Drug and Alcohol Dependence 2002 65 2 179 189 10.1016/S0376-8716(01)00160-0 11772479 Roose KM Mattaini MA Cihon TM Mattaini MA Challenging violence: Toward a twenty-first century, science-based “constructive Programme” Behavior science perspectives on culture and community 2020 Springer 307 331 Schriner KF Fawcett SB A community concerns method for local agenda setting Journal of the Community Development Society 1988 19 2 108 118 10.1080/15575338809490008 Schultz J Collie-Akers V Fernandez C Fawcett S Ronan M Implementing community-based participatory research with two ethnic minority communities in Kansas City, Missouri International Journal of Migration, Health and Social Care 2009 5 1 47 57 10.1108/17479894200900006 Seekins T Fawcett SB Public policymaking and research information The Behavior Analyst 1986 9 35 45 10.1007/BF03391928 22478646 Seekins T Fawcett SB Effects of a poverty clients' agenda on resource allocations by community decision makers American Journal of Community Psychology 1987 15 3 305 320 10.1007/BF00922700 3604999 Seekins T Fawcett SB Cohen S Elder J Jason L Schnelle J Winett R Experimental evaluation of public policy: The case of state legislation for child passenger safety Journal of Applied Behavior Analysis 1988 21 3 233 243 10.1901/jaba.1988.21-233 3198544 Seekins T Fawcett SB Mathews RM Effects of self-help guides on three consumer advocacy skills: Using personal experiences to influence public policy Rehabilitation Psychology 1987 32 1 29 38 10.1037//0090-5550.32.1.29 Seekins T Mathews RM Fawcett SB Enhancing leadership skills for community self-help organizations through behavioral instruction Journal of Community Psychology 1984 12 2 155 163 10.1002/1520-6629(198404)12:2<155::AID-JCOP2290120208>3.0.CO;2-3 10266579 Seekins T Maynard-Moody S Fawcett SB Understanding the policy process: Preventing and coping with community problems Prevention in Human Services 1987 5 2 65 89 10.1300/J293v05n02_04 Seniuk HA Cihon TM Benson M Luke MM Making a footprint in environmental sustainability: A behavioral systems approach to engaging the behavioral community Perspectives on Behavior Science 2019 42 4 911 926 10.1007/s40614-019-00233-y 31976466 Silverman K Higgins ST Brooner RK Montoya ID Cone EJ Schuster CR Preston KL Sustained cocaine abstinence in methadone maintenance patients through voucher-based reinforcement therapy Archives of General Psychiatry 1996 53 5 409 415 10.1001/archpsyc.1996.01830050045007 8624184 Silverman K Holtyn AF Subramaniam S Behavior analysts in the war on poverty: Developing an operant antipoverty program Experimental and Clinical Psychopharmacology 2018 26 6 515 524 10.1037/pha0000230 30265062 Silverman K Holtyn AF Toegel F The utility of operant conditioning to address poverty and drug addiction Perspectives on Behavior Science 2019 42 3 525 546 10.1007/s40614-019-00203-4 31976448 Silverman K Svikis D Robles E Stitzer ML Bigelow GE A reinforcement-based therapeutic workplace for the treatment of drug abuse: Six-month abstinence outcomes Experimental and Clinical Psychopharmacology 2001 9 1 14 23 10.1037/1064-1297.9.1.14 11519628 Smilak, N. R., & Putnam, R. F. (2022). A critique of colonialism and modern aid in Africa: What would skinner say? Behavior and Social Issues, 1–20. 10.1007/s42822-022-00093-3 Solar, O., & Irwin, A. (2010). A conceptual framework for action on the social determinants of health. World Health Organization. Strauss WJ Nagarajal J Landgraf AJ Arteaga SS Fawcett SB Ritchie LD John LV Gregoriou M Frongillo EA Loria CM Weber SA Collie-Akers VL McIver KL Schultz J Sagatov RD Leifer ES Webb K Pate RR on behalf of the Healthy Communities Study Team The longitudinal relationship between community programmes and policies to prevent childhood obesity and BMI in children: The healthy communities study Pediatric Obesity 2018 13 82 92 10.1111/ijpo.12266 29493122 Suarez de Balcazar Y Bradford B Fawcett SB Common concerns of disabled Americans: Issues and options Social Policy 1988 19 2 29 35 10318229 Switzer, K., & Rakos, R. F. (2022). A behavioral community psychology framework for analyzing housing stability for homeless families: Modifying the rapid re-housing metacontingency. Behavior and Social Issues, 1–25. 10.1007/s42822-022-00098-y Thompson J Fawcett SB Schultz JA Differential effects of strategic planning on community change in two urban neighborhood coalitions American Journal of Community Psychology 2008 42 1 25 38 10.1007/s10464-008-9188-6 18607719 United Nations (2015). Transforming our world: The 2030 agenda for sustainable development. United Nations. http://sustainabledevelopment.un.org/. Accessed 12 Nov 2022. Valderlon Y Elias LR The Bolsa Família program and cultural design: Behavioral interpretations of cultural interventions Behavior and Social Issues 2019 28 1 114 126 10.1007/s42822-019-0003-9 Watson-Thompson J Fawcett SB Schultz JA A framework for community mobilization to promote healthy youth development American Journal of Preventive Medicine 2008 34 3 S72 S81 10.1016/j.amepre.2007.12.016 18267205 Watson-Thompson J Keene-Woods NK Schober DJ Schultz JA Enhancing the capacity of substance abuse prevention coalitions through training and technical assistance Journal of Prevention & Intervention in the Community 2013 41 3 176 187 10.1080/10852352.2013.788345 23751061 Watson-Thompson J Francisco VT Anderson-Carpenter KD A behavioral community approach to community health and development: Tools for collaborative action Behavior science perspectives on culture and community 2020 Springer 333 356 Whang PL Fletcher RK Fawcett SB Training counseling skills: An experimental analysis and social validation Journal of Applied Behavior Analysis 1982 15 3 325 334 10.1901/jaba.1982.15-325 7142059 World Health Organization, Commission on Social Determinants of Health. (2008). Closing the gap in a generation: Health equity action on the social determinants of health. Final Report of the Commission on Social Determinants of Health. . ISBN: 9789241563703 (https://www.who.int/publications/i/item/9789241563703, Accessed, September 28, 2022)	0	PMC9718469	NO-CC CODE	2022-12-06 23:23:39	no	Behav. Soc. Iss.. 2022 Dec 2;:1-24	utf-8	null	null	null	oa_other
==== Front Environ Dev Sustain Environ Dev Sustain Environment, Development and Sustainability 1387-585X 1573-2975 Springer Netherlands Dordrecht 2787 10.1007/s10668-022-02787-5 Article Suitable site selection by using full consistency method (FUCOM): a case study for maize cultivation in northwest Turkey http://orcid.org/0000-0002-3670-2114 Everest Timuçin [email protected] 1 Savaşkan Gönül Selin [email protected] 2 Or Aykut [email protected] 3 Özcan Hasan [email protected] 4 1 grid.412364.6 0000 0001 0680 7807 Lapseki Vocational School, Çanakkale Onsekiz Mart University, 17800 Çanakkale, Turkey 2 grid.412364.6 0000 0001 0680 7807 Department of Economics, Faculty of Political Science, Çanakkale Onsekiz Mart University, 17020 Çanakkale, Turkey 3 grid.412364.6 0000 0001 0680 7807 Department of Mathematics, Faculty of Science, Çanakkale Onsekiz Mart University, 17020 Çanakkale, Turkey 4 grid.412364.6 0000 0001 0680 7807 Department of Soil Science and Plant Nutrition, Faculty of Agriculture, Çanakkale Onsekiz Mart University, 17020 Çanakkale, Turkey 2 12 2022 120 6 5 2022 17 11 2022 © The Author(s), under exclusive licence to Springer Nature B.V. 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The agricultural land evaluation procedure is a valuable guide for growing plants where they are best suitable, and it has a critical role in actualizing sustainable plans for providing food security for the growing population. In agricultural land suitability analysis, different multi-criteria decision-making methods are applied. The main objective of this study is to introduce the potential usage of a new multi-criteria decision-making method the Full Consistency Method (FUCOM) in agricultural land suitability analysis. The study was carried out in the northern part of the Karamenderes plain in NW Turkey. Nine land characteristics (soil texture, soil depth, organic matter content, electrical conductivity, pH, slope, drainage, CaCO3%, and cation exchange capacity) were used for the land evaluation study. The weighting values of the land characteristics were calculated by the FUCOM. According to the results, 223 ha (6.26%) were highly suitable, 2650 ha (74.40%) were moderately suitable, 508 ha (14.26%) were marginally suitable, and 181 ha (5.08%) were not suitable for maize cultivation. The weighted values of the parameters were also tested with Analytic Hierarchy Process (AHP) and the Best-Worst Method (BWM). There is a general compatibility between the methodologies. The data obtained from these methods showed that analysis consists of a very positive relationship with each other. The comparisons of these methodologies showed that FUCOM’s prioritization order simplicity in parameter weighting and ability to reduce the processing intensity would provide a significant contribution and advantage to the land evaluation experts and planners. It is recommended that the Full Consistent Method could be reliably used in agricultural land suitability analysis. Keywords Land suitability Multi-criteria decision making Sustainable agriculture Overlay analysis ==== Body pmcIntroduction The rapid increase in the world’s population since the industrial revolution has led authorities to seek solutions to issues related to food security, the quality of life of future generations, and increasing land degradation (Beek et al., 1997). The COVID-19 pandemic began at the end of 2019 and rapidly exacerbated, and other pandemics that may occur in future have shown the importance of reaching healthy food. Restricted food mobility and production due to the pandemic have further increased the importance of sustainable use of natural resources. The reduction of agricultural lands with increasing pressures poses severe threats to producing safe and healthy food. The most important way to reach safe and healthy food can be provided with a sustainable agricultural production model. Only rational planning studies can ensure sustainability in agricultural production and improve soil health. Özkan et al. (2019) mentioned that countries need plans at various scales to meet their nutritional needs and ensure self-sufficiency in primary food products. Land evaluation studies are critical tools for the realization of these plans. Land evaluation studies ensure using land resources at a sustainable level, and it aids in safeguarding them for future generations’ needs. FAO (1976) defined the land evaluation concept as the determination of the capabilities and productivity of the land. Land evaluation is a land use planning tool, and it evaluates the expected benefits, limitations, and environmental impacts that may occur from sustainable land use (Rossiter, 1996). In agricultural land evaluation studies, complex parameters such as physical and chemical soil characteristics, land morphology, and climatic parameters affect each other. Multi-criteria decision-making (MCDM) methods are often used to evaluate these interrelated parameters (Bilgilioğlu 2021; Zhang et al. 2021). Since the second half of the 20th century, mathematical models have been used in land evaluation studies and valuable results have been obtained with the support of developing computer technologies (Odeh & McBratney, 2005). MCDM is a methodology that allows determining the best alternative among all available options in the presence of more than one criterion. Due to the wide range of data and the complexity of the criteria in land suitability studies, the use of MCDM in these studies has been suggested by different researchers (Sarkar et al., 2021; Makungwe et al., 2021) Several land suitability studies are used different MCDM methods in the literature. Most of these studies were performed using the AHP method. Various researchers have conducted different land suitability and site selection studies using the AHP method (Zolekar, 2018; Dedeoğlu & Dengiz, 2019; Everest, 2021; Özkan et al., 2020; Everest et al., 2021; Bilgilioğlu, 2021; Günal et al., 2022; Everest & Gür, 2022). There are also studies using MCDM methods other than the AHP in the literature. Mendas et al. (2021) conducted an agricultural land-use suitability study in Algeria using the ELECTRE Tri method. Mistri and Sengupta (2019) used Weighted Principal Component Analysis and the AHP method in the agricultural land suitability study. Montgomery et al. (2017) used the Logic Scoring of Preference method to classify land use capability and suitability in Colorado, USA. Jahanpoor et al., (2018) used the PROMOTHEE technique to determine land suitability for pomegranate and pistachio in Iran. Bagherzadeh and Gholizadeh (2017) used the TOPSIS method to asses land suitability for the alfalfa plant in Joveyn Plain in Iran. Everest et al. (2022) determined the land suitability for paddy using the Best-Worst Method (BWM). MCDMs, used by different disciplines and diversified with the development of new mathematical models, have been increasingly used by land evaluation experts to determine the suitability of different land-use types. Similarly, the Full Consistency Method (FUCOM) method, recently added to the literature by Pamučar et al. (2018), is discussed by different disciplines. In the literature survey, it was noticed that the FUCOM had been used in many other fields such as; site selection for textile production (Ulutaş & Karakuş 2021), landfill site selection (Badi & Kridish, 2020) determination of groundwater potential (Akbari et al. 2021), site selection for solar panel energy (Cao et al. 2019), wind farm site selection (Ecer, 2021), mapping the mineral potential (Feizi et al. 2021), selection of technologies for municipal wastewater treatment (Srivastava & Singh, 2021). However, agricultural land suitability and suitable site selection studies for different crops are not available in the literature with the use of FUCOM. To the best of our knowledge, this study will be the first agricultural land suitability study to be conducted using the FUCOM. This deficiency in the literature constituted the primary motivation of the study. The main objective of this study is to test the potential use of the FUCOM subjective weighting method for agricultural land suitability. Within the scope of the study, suitable lands for maize cultivation were evaluated by the FUCOM method. Then, the results of FUCOM were compared with other commonly used multi-criteria decision-making methods (AHP and BWM). Materials Study area The study area consists of lands north of the Karamenderes basin in NW Turkey. It is located between 39°57′36″–40°00′18″N latitudes and 26°10′12″–26°18′36″E longitudes and covers an area of 3562 ha (Fig. 1). The main geological units in the study area are the Quaternary aged alluvium deposit carried by the Karamenderes and Dümrek rivers and the Miocene aged Terrigenous clastic and calcareous units in the sloping and high terrain areas. The climate is characterized by the transition between the Marmara and Aegean regions, with cool and rainy winters and hot and dry summers. The annual average temperature is 15 °C, and the total precipitation is 625 mm (MGM, 2021). Soil resources in the study area are Typic Ustifluvents, Typic Fluvaquents, Typic Haplustepts, Typic Calciustepts, Mollic Ustifluvents, Typic Ustorthents, Inceptic Haplustalfs according to Soil Taxonomy (Everest, 2015). The most important economic activity in the region is agricultural production. In addition, the most cultivated crops are wheat, barley, alfalfa, tomato, pepper, maize, and paddy. Fig. 1 Study area Methodology Within the scope of this study, maize was chosen as the target plant due to its strategic importance. Maize is in third place after wheat and paddy in the cultivation area, but it is first in production. Maize is used for food, animal feed, and biofuel. The most maize-producing countries are the USA, China, and Brazil. Turkey ranks 24th in maize production (ZMO, 2016). In 2020, maize was cultivated in 196,982 × 103 hectares globally and 591,900 hectares in Turkey (TEPGE, 2021). Maize is grown intensively in Turkey’s Mediterranean, Southeastern Anatolia, and Aegean regions. In this study, related studies in the literature were examined to determine the soil requirements of the maize plant (Sys et al., 1993; Jimoh et al. 2016; Tashayo et al. 2020; Costantini, 2009). Only physical, chemical, and morphological characteristics were evaluated for the suitability analysis. Nine land characteristics (soil texture, depth, organic matter, EC, pH, slope, drainage, CaCO3%, and Cation Exchange Capacity (CEC)) were used in the model for land evaluation. The soil properties were taken from the 1/10,000 scale detailed soil survey and mapping report produced by Everest (2015). Spatial information on each physical and chemical characteristic was obtained from land mapping units of the detailed soil survey report. The Digital Elevation Model (DEM) was generated from 30 × 30 m spatial resolution data downloaded from NASA’s website. The slope map was also derived from the DEM map. An impact scoring was performed to score the sub-parameters. Score 4 was given for optimum conditions for the plant requirements, and score 1 was given for not meeting conditions (Table 1). Middle scores were used among optimal and limiting conditions. The flowchart of the study (Fig. 2) is also presented. Table 1 Sub-criteria for maize cultivation Surface soil texture Depth (cm) Organic matter (%) EC (dS m− 1) Slope (%) Classification Score Classification Score Classification Score Classification Score Classification Score Cmassive 1 100 < 4 < 1 1 0–4 4 0–2 4 S 2 50–100 3 1–2 2 4–6 3 2–6 3 SL, LS 3 20–50 2 2–3 3 6–8 2 6–12 2 C, SiC, SiCL, CL, Si, SiL 4 0–20 1 3 < 4 > 8 1 12 < 1 Drainage CaCO3 (%) Cation exchange capacity (CEC) cmol kg−1 pH Classification Score Classification Score Classification Score Classification Score Excess 4 < 6 4 < 25 4 6.5–7.5 4 Moderate 3 6–15 3 12–25 3 5.8–6.5, 7.5–7.8 3 Poor 2 15–25 2 6–12 2 5.5–5.8, 7.8–8.2 2 Very poor 1 25 < 1 < 6 1 < 5.5 and 8.2 < 1 Fig. 2 Flowchart of the study FUCOM subjective weighting method FUCOM is a subjective criterion weighting method introduced to the literature by Pamučar et al. (2018). This method makes it possible to obtain a solution by making significantly fewer pairwise comparisons than other criterion weighting methods. For n criteria, this approach performs n-1 comparisons (Pamučar et al., 2018; Ayçin et al., 2021). Furthermore, one of the other outstanding features of this method is that it is not complicated and can be used in the group decision-making process. The application steps of the FUCOM, which include all the formulas, are explained in three stages (Pamucar et al., 2018; Ayçin et al., 2021). Step 1. Ranking according to the significance of criteria Firstly, the decision-maker(s) rank the criteria from most important to least important (Table 2). Thus, criteria rankings are obtained according to the expected values of the weight coefficients as in the expression (1). Table 2 Ranking criteria for FUCOM Criteria no. CR1 CR2 CR3 CR4 CR5 CR6 CR7 CR8 CR9 Criteria Texture EC Drainage Depth Slope CaCO3 pH OM CEC Rank 1 2 3 4 5 6 7 8 9 1 Cj1>Cj2>...>Cjk where k represents the rank of considered criteria. In the case of criteria considered to be of equal importance by the decision-maker “=” sign can be used instead of “>”. Step 2. Determination of comparative priorities of criteria Secondly, comparing the ranking criteria is performed, and the comparative priority of the evaluation criteria ϕk/(k+1) is defined. As a result, the vector of the comparative priority is obtained, as shown in expression (2).2 ϕ=φ1/2,φ2/3,φ3/4,...,φk/(k+1) where the value φk/(k+1) represents a superiority of Cjk rank criterion over Cj(k+1) rank. In the FUCOM, the decision-maker(s) can use integers, decimals, or values of certain scales to compare criteria. This provides flexibility to decision-makers in the evaluation of criteria. Step 3. Calculation of the weights of criteria The final weight coefficients ω1, ω2,…,ωT are obtained in the last step. These values should provide the two requirements: The proportion of the weight coefficients given in expression (3) is equal to the comparative priority of the criteria determined in Step 2. 3 ωkωk+1=φk/(k+1) (2) The final values of the weight coefficients should satisfy the condition of mathematical transitivity; that is. φk/(k+1)×φ(k+1)/(k+2)=φk/(k+2). Since φk/(k+1)=ωkωk+1 and φ(k+1)/(k+2)=ωk+1ωk+2, that ωkωk+1×ωk+1ωk+2=ωkωk+2 is obtained. Therefore, another requirement that the final values of the weight coefficients of the evaluation criteria need to satisfy is obtained as shown in expression (4). 4 ωkωk+2=φk/(k+1)×φ(k+1)/(k+2) Full consistency is validated when the conditions in the expression (3) and (4) are satisfied. Namely, a deviation from full consistency (DFC) is minimum in this case. In this way, the criterion for maximum consistency is satisfied, and DFC is χ=0 for the calculated values of the weight coefficients. The linear programming model shown in expression (5) should be solved for calculating the final values of the weight coefficients of the evaluation criteria.5 Minχωjkωj(k+1)-φk/(k+1)≤χ,∀jωjkωj(k+2)-φk/(k+1)×φ(k+1)/(k+2)≤χ,∀j∑j=1nωj=1ωj≥0,∀j By solving the linear programming model (5), the final values of the evaluation criteria ω1,ω2,...,ωnT and DFCχ are calculated. FUCOM: a case study for maize cultivation In this study, FUCOM was applied to determine the relative weighted value of criteria required for maize farming. Firstly, criteria and sub-criteria were determined. In the suitability analysis, nine selected parameters were used. Secondly, pairwise comparisons were made. The comparisons are presented in Table 3. The relative significance of parameters in the same hierarchical level was considered for binary comparisons. Saaty (1980)’s 1–9 scale was used in binary comparisons. Finally, the resultant values of the weighted coefficients were calculated. Table 3 Comparisons in FUCOM Criteria according to rank Texture EC Drainage Depth Slope CaCO3 pH OM CEC Comparisons 1 2 3 3 4 5 6 6 7 By using evaluations in Table 3, all the priorities of the evaluation criteria are calculated as follows:φC1/C2=2/1=2φC2/C3=3/2=1,5φC3/C4=3/3=1φC5/C6=5/4=1,25φC6/C7=6/5=1,2φC7/C8=6/6=1φC4/C5=4/3=1,33φC8/C9=7/6=1,16 Although the ratio of the weight coefficients given in expression (3) is equal to the comparative priority of the criteria, the ratio of the weight coefficients is obtained as follows:ωC1/C2=2ωC2/C3=1,5ωC3/C4=1ωC5/C6=1,25ωC6/C7=1,2ωC7/C8=1ωC4/C5=1,33ωC8/C9=1,16 Furthermore, the final values of the weight coefficients should provide the condition of mathematical transitivity as follows:ω1ω3=ωC1/C2×ωC2/C3=2×32=3⋮ω7ω9=ωC7/C8×ωC8/C9=1×1,16=1,16 Finally, the linear programming model for determining the final values of the weight coefficients can be stated in expression (6):6 Minχω1ω2-2≤χ,ω2ω3-1,5≤χ,⋯,ω8ω9-1,16≤χωjkωj(k+2)-φk/(k+1)×φ(k+1)/(k+2)≤χ,∀j∑j=19ωj=1ωj≥0,∀j When Excel Solver solves a model (6), the final values of the weight coefficient are calculated (Table 4). Table 4 Weighted values obtained from FUCOM calculations Parameters Weighted values Texture 0.323 EC 0.162 Drainage 0.108 Depth 0.108 Slope 0.081 CaCO3 0.065 pH 0.054 OM 0.054 CEC 0.046 ArcGIS 10.3 was used for overlay analysis and mapping procedure. According to FAO (1977) criteria, the suitability was reclassified, and the final suitability map was obtained. Comparison of FUCOM results with AHP and BWM The weighted values obtained by the FUCOM were also calculated using the BWM method (Rezaei, 2015) and the AHP (Saaty, 1980). BWM methodology BWM was introduced to the literature by Rezaei (2015). BWM is a pairwise comparison-based multi-criteria decision-making method. Initially, the best and the worst criteria were determined by the decision-maker. Then, the best criterion is compared with the other criteria. In the BWM methodology, a 1–9 scale was used to determine the relative preference levels of the criteria. The BWM calculations are provided below step by step (Rezaei, 2015; Rezaei, 2016): Step 1 A set of decision criteria are built. Step 2 The best and the worst criteria are determined. Step 3 The best criterion is compared to all the other criteria by using numbers 1–9 with AB=aB1,aB2,...,aBnandaBB=1; Step 4 The worst criterion is compared to other criteria by using numbers 1–9 with AW=a1W,a2W,...,anWandaWW=1; Step 5 Optimal weighted values for each criterion are obtained. For each pairwise comparison, among the best criterion and the others (WBWj) and the worst criterion and the others (Wj/WW) should be satisfied. For this wBwj-aBj and wjwW-ajW for all j, should be minimized. In the methodology consistency ratio is checked with formulation (7) below:7 ConsistencyRatioCR=ξConsistencyindexCI Here ξ is the optimal value of the method. CI value was obtained from Rezaei (2015). The BWM comparisons are presented in Table 5. Table 5 Pairwise comparisons based on BWM Best to others Texture EC Drainage Depth Slope CaCO3 pH OM CEC Texture 1 2 3 4 4 4 5 5 5 Others to worst CEC Texture 5 EC 4 Drainage 3 Depth 3 Slope 2 CaCO3 2 pH 1 OM 1 CEC 1 AHP AHP (Saaty, 1980) is a widely used multi-criterion decision-making method. AHP is an MCDM, that can solve complex problems (Moreno-Jiménez et al., 2008). AHP provides to solve problems in a realistic facilitates. In AHP, problems are broken into different hierarchical levels (Badi et al., 2019). AHP enables evaluating qualitative and quantitative factors. With binary comparisons, priority values are obtained. With the AHP, decision-makers can prefer the best alternative. AHP methodology is described in three stages below: Stage 1: Hierarchical model is structured. Stage 2: Pairwise comparisons are realized. Stage 3: Priority values are obtained. In the first stage, the problem is separated into subsections. In this stage, the determination of criteria, sub-criteria, and alternatives are performed. In the second stage, binary comparisons are structured, and, in this way, a decision matrix is built. The binary comparisons determine the relative significance of parameters that belong to the same hierarchy. In comparison, Saaty (1980)’s 1–9 scale is used. Matrix calculation is carried out in the third stage. With the resultant information of matrix calculations, eigenvector values are obtained. These values are used for assigning weighted values to the criteria and sub-criteria. Finally, the consistency check is controlled. If the consistency is <%10 the model is accepted as validated. Otherwise, the binary comparison should be restructured (Saaty, 1980). The AHP comparisons are presented in Table 6. Table 6 Pairwise comparisons based on AHP Texture EC Drainage Depth Slope CaCO3 pH OM CEC Texture 1 2 3 4 4 5 6 6 7 EC 1/2 1 2 3 3 4 6 6 6 Drainage 1/3 1/2 1 3 3 4 5 5 5 Depth 1/4 1/3 1/3 1 2 3 4 4 4 Slope 1/4 1/3 1/3 1/2 1 2 3 3 3 CaCO3 1/5 1/4 1/4 1/3 1/2 1 3 2 2 pH 1/5 1/6 1/5 1/4 1/3 1/3 1 1 2 OM 1/6 1/6 1/5 1/4 1/3 1/2 1 1 2 CEC 1/7 1/6 1/5 1/4 1/3 1/2 1/2 1 1 λ = 9.60, n = 9, CI (consistency index) = 0.074, CR (consistency ratio) = 0.051 Results and discussions In this study, suitability analyses for maize farming were carried out. The maize plant was chosen because of its strategic importance. Maize, the feed source for humans and animals, is the most important grain in the world after wheat and paddy (Preedy & Watson, 2019; Tashayo et al., 2020). In addition, maize is used as an industrial raw material source (Ramamurthy et al., 2020). Different researchers have conducted suitability studies for maize cultivation in various parts of the world (Braimoh et al., 2004; Tashayo et al., 2020; Pilevar et al., 2020; Wanyama et al., 2019; Ramamurthy et al. 2020). Researchers also suggested that it is essential to determine suitable areas for maize cultivation and that studies should be increased. In addition, Sharma et al. (2018) stated that the literature should be supported, especially with crop-based studies. The FUCOM was used for assigning weighted values for the selected land characteristics and then suitability analyses for maize cultivation were performed. As a result of the calculations, Texture (0.323) was the most effective factor, and EC (0.162) was evaluated as the second effective factor. The lowest weighted value was assigned to the CEC (0.046) criterion (Table 4). Soil texture is an important parameter affecting the soil’s biophysical properties, soil fertility, and quality in the long term (Upadhyay & Raghubanshi, 2020). Leclerc et al. (2001) defined soil texture as the most critical component in soil fertility. In addition, Ziadi (2013) mentioned that soil texture is a very effective parameter in maize cultivation and have an essential role in the uptake of nitrogen element. For these reasons, texture has been considered this study’s most critical land characteristic. The weighted value of the EC parameter was determined as (0.162). Soil salinity is an important limiting factor that reduces the quality and productivity of crops. Although maize can tolerate moderate salinity, it is a salinity-sensitive crop in the early stages of growth (Sabagh et al., 2021). Salinity adversely affects maize crops’ vegetative and generative development, so plant growth and yield decrease (Iqbal et al., 2020). The weighted value of the drainage parameter was determined as (0.108). Saturated soil conditions in poor drainage restrict the rooting and plant growth of the maize and significantly limit its production (Nielsen, 2012; Nash et al., 2015). The weighted value of the soil depth parameter was determined as (0.108). Shallowness is an important limiting factor for land use, land management, and crop production. In shallow soils, the water holding capacity of the soil and the availability of nutrients are reduced, and crop yields decrease due to the limitation of root elongation (Peralta et al., 2021). Moreover, Sadras and Calvino (2001) reported that maize cultivation in shallow soils significantly reduced crop yield. The weighted value of the slope parameter was determined as (0.081). As the slope increases, the severity of the erosion rises, and using agricultural machinery becomes more problematic. Changere and Lal (1997) reported that sloping lands have disadvantages while flatlands are more suitable for maize growth. On the other hand, Fujisao et al. (2020), reported that plant nutrients decreased in sloping lands where maize cultivation was carried out, and as a result, this limited the yield. The weighted value of the CaCO3 parameter was determined as (0.065). El-Tilib (2005) reported that increasing CaCO3 values in the soil decreased maize’s dry matter weight and grain yield. Similarly, Elamin et al. (2005) also reported that high levels of CaCO3 adversely affect the N, P, and Mg contents in maize cultivated soils. The weighted value of the soil pH parameter was determined as (0.054). The solubility of plant nutrients is associated with soil pH. It has been reported that the ideal pH range for maize cultivation is neutral or near-neutral conditions (The et al., 2006). Tandzi et al. (2018) stated that low pH conditions reduce maize grain yield. Rahman et al. (2011) noted that high pH conditions adversely affect maize growth, especially in calcareous environments. The weighted value of the soil organic matter was determined as (0.054). Soil organic matter improves the soil’s physical, chemical, and biological processes and influences soil fertility. It has been reported that increasing soil organic matter increases maize yield (Kane et al., 2021). Low soil organic matter has also been reported to reduce maize yield (Mendez et al., 2019). The weighted value of the cation exchange capacity was determined as (0.046). Cation exchange capacity indicates soil nutrient holding capacity (Fugger, 1999; Braimoh et al., 2004). High amounts of CEC in the soil indicate that plant nutrients are sufficient, and low amounts indicate that the soil needs fertilization (Arunrat et al. 2020). Figure 3 presents the land suitability map obtained by the FUCOM. The analyses showed that 223 ha (6.26%) were highly suitable, 2650 ha (74.40%) were moderately suitable, 508 ha (14.26%) were marginally suitable, and 181 ha (5.08%) were not suitable for maize cultivation (Table 7). Highly suitable lands have optimum conditions in terms of all land characteristics, where there are no or very few limiting factors in the study area. Moderately suitable lands are the most common lands in the study area. These lands consist of coarser textured, slightly alkaline soils with medium organic matter content. Marginally suitable lands contain coarser textured, sloping, shallow, low organic matter content, moderately alkaline, moderately limy soils. Although not suitable lands have similar characteristics with marginally suitable lands, these lands also consist of sections with severe saline soils in the delta plain and very steep lands in higher altitudes. Table 7 Land suitability for maize cultivation based on FUCOM Suitability classification Class Area (ha) Area (%) Highly suitable S1 223 6.26 Moderately suitable S2 2650 74.40 Marginally suitable S3 508 14.26 Not suitable N 181 5.08 Total 3562 100 Validating the result of MCDA analysis is essential for testing the reliability of the data (Chen et al., 2010; Ghorbanzadeh et al., 2018). For accuracy assessment, interviews were conducted with the experts in Çanakkale Agriculture Directorate. The comparison was easily performed since maize production (forage crop, maize grain, and seed production) was recorded very sensitively. The data produced from this study and the land records showed a significant similarity. Also, it was seen that the farmers chose especially highly and moderately suitable lands for maize cultivation. The results were also compared with AHP and BWM methods. The statistical evaluation showed general compatibility between the three methodologies. The data obtained from FUCOM, BWM, and AHP analysis consist of very closer values. In the correlation analysis (p < .05), it was determined that there was a positive (r = .992) relationship between FUCOM and BWM and a positive relationship (r = .941) between FUCOM and AHP method (Table 8). The statistical evaluation supported that there was consistency among the methods. Similarly, Ecer (2021) reported a strong correlation between FUCOM, BWM, and AHP methods in his study. For this reason, it can be concluded that the values obtained by the study are acceptable and reliable. Table 8 Comparisons of the weighted values according to methodologies Parameters FUCOM BWM AHP Mean Std. dev. CV Texture 0.323 0.292 0.290 0.30 0.015 5.01 EC 0.162 0.173 0.209 0.18 0.020 11.07 Drainage 0.108 0.115 0.165 0.13 0.025 19.62 Depth 0.108 0.115 0.107 0.11 0.004 3.24 Slope 0.081 0.086 0.078 0.08 0.003 4.04 CaCO3 0.065 0.069 0.055 0.06 0.006 9.35 pH 0.054 0.058 0.034 0.05 0.010 21.57 OM 0.054 0.058 0.035 0.05 0.010 20.48 CEC 0.046 0.034 0.027 0.04 0.008 22.00 Std. Dev.—standard deviation, CV—coefficient of variation Different researchers have used the multi-criteria approach for decision support in land suitability studies. Zhang et al. (2015) used the AHP method to determine suitable lands for tobacco farming. They reported that one of their study’s challenges was determining the weighting values of the factors in the multi-criteria decision-making process. Flynn (2019) used AHP in his research and reported that the most critical step in the study is the determination of weights. Zabihi et al. (2015) determined suitable lands for citrus cultivation with AHP in their research. They reported that prioritization of all alternatives and criteria to each other have complex relationships. Seyedmohammadi et al. (2019) emphasized that there are a lot of studies on land suitability evaluation and they suggest more research should be conducted. Therefore, they suggest applying new interactive methodologies for land evaluation. Recently, new methodologies were used in land evaluation studies in the literature. Everest et al. (2022) and Tercan & Dengiz (2022) used the BWM in site selection for rice cultivation. Kheybari et al. (2021) used BWM for land suitability for corn cultivation. Researchers performed fewer pairwise comparisons to AHP and they obtained acceptable results. Although the AHP method is used effectively to solve many decision-making problems, it is criticized in some aspects. These can be listed as the problems experienced in pairwise comparisons, inability to deal with uncertainty and indecision situations, and being too dependent on the expert’s knowledge (Deng, 1999; Özkan et al., 2019). The BWM, on the other hand, makes fewer comparisons compared to AHP. In our study, fewer comparisons were made with the FUCOM compared to the AHP and the BWM. It is one of the most important results of this study, and it is especially useful in preventing confusion arising from pairwise comparisons in multi-criteria decision-making. These results are supported by different researchers. Puška et al. (2021) reported that the FUCOM method performed fewer binary comparisons than AHP. Biswas et al. (2021) stated that making (n − 1 number) comparisons in FUCOM decreases the inconsistency due to judgment. Akbari et al. (2021) reported that the FUCOM methodology removed the redundancy of comparisons in criteria pairs. Required pairwise comparisons for FUCOM, AHP, and BMW methods are given in Table 9. According to Table 9, it was observed that the most comparison was in the AHP (36) method, and the fewer comparison was in the FUCOM (8). These calculations support our results. The approach put forward in this study reveals that the FUCOM can be easily applied in agricultural land evaluation studies. It contributes to reducing the inconsistency of the judgments of the decision-makers. The method offers users a more flexible and comfortable mathematical model, and the results produced by the model are more optimistic. These results are in compliance with other studies. Pamučar et al. (2021) reported that FUCOM used a simple algorithm in the calculation and made fewer binary comparisons in deciding the criterion weights. Multi-criteria land suitability assessments are expressed with mathematical formulas. Each parameter evaluated with these methods is used in the mathematical model and land suitability classes are determined according to these calculated index values (Dengiz & Sarıoğlu, 2013). For this reason, the widespread use of different mathematical methods and practical options for land evaluation experts may produce more effective land evaluation studies. The currently unplanned use of the lands increases the problems of limited resources (Chaudhary et al. 2008). As a solution to this problem, planning studies should be expanded. Increasing land evaluation studies can be achieved by developing easy-to-use methods. For this reason, we recommend that the FUCOM can be used in land evaluation studies due to its flexibility, ability to reduce the processing intensity, and only assigning values to parameters by prioritizing. Fig. 3 Land suitability for maize cultivation based on FUCOM Table 9 The required number of comparisons in AHP, BWM, and the FUCOM MCDM method The number of criteria (n) and the required number of pairwise comparisons n = 2 n = 3 n = 4 n = 5 n = 6 n = 7 n = 8 n = 9 AHP (n(n − 1)/2) 1 3 6 10 15 21 28 36 BWM (2n − 3) 1 3 5 7 9 11 13 15 FUCOM (n − 1) 1 2 3 4 5 6 7 8 Bold indicates the study's numbers of pairwise comparison Conclusion Selection of suitable lands is critical for the sustainable and rational use of resources in agricultural production. The unplanned use of agricultural lands increases the importance of land suitability studies day by day. In this study, suitable lands for maize cultivation were determined using the FUCOM. This study is the first to integrate the FUCOM into agricultural land suitability studies. In this study, nine land characteristics were used and weighted coefficients were calculated with the FUCOM. The weighted values calculated with the FUCOM were compared with the AHP and the BWM, and it was revealed that the results have a strong relationship with each other. It is concluded that the FUCOM’s prioritization order provides a significant advantage and contribution to land evaluation processes. It uses a much simpler algorithm than other methods and removes the redundancy of comparisons in pairs of criteria. The FUCOM can be easily applied to land suitability studies due to these positive advantages. Since the agricultural lands have reached the uppermost limits, macro plans for large areas must have been upgraded with crop-based micro plans. Widespread applications of FUCOM and other techniques could support decision-makers, users, and policymakers in managing natural resources. Because finding the best place for a plant to grow is the most fundamental factor for sustainable and rational planning for food security. For this reason, using different multi-criteria decision-making methods in land evaluation studies and determining which methods are most suitable and easily useable for this area may be the subject of detailed studies in future. Comparing FUCOM results using fuzzy & gray methodologies in future and in-depth studies may add a different dimension to the studies. The authors are indebted to the anonymous reviewers and the editors for their valuable and constructive suggestions. Data availability All data included in this study are available upon request by contact with the corresponding author. Declarations Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Akbari M Meshram SG Krishna RS Pradhan B Shadeed S Khedher KM Sepehri M Ildoromi AR Alimerzaei F Darabi F Identification of the groundwater potential recharge zones using MCDM models: Full consistency method (FUCOM), best worst method (BWM) and analytic hierarchy process (AHP) Water Resources Management 2021 35 14 4727 4745 10.1007/s11269-021-02924-1 Arunrat N Kongsurakan P Sereenonchai S Hatano R Soil organic carbon in sandy paddy fields of Northeast Thailand: A review Agronomy 2020 10 8 1061 10.3390/agronomy10081061 Ayçin E Aşan H İş zekası uygulamaları seçimindeki kriterlerin önem ağırlıklarının FUCOM yöntemi ile belirlenmesi. Afyon Kocatepe Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi 2021 23 2 195 208 Badi I Abdulshahed A Shetwan A Eltayeb W Evaluation of solid waste treatment methods in Libya by using the analytic hierarchy process Decision Making: Applications in Management and Engineering 2019 2 2 19 35 Badi, I., & Kridish, M. (2020). Landfill site selection using a novel FUCOM-CODAS model: A case study in Libya. Scientific African, 9:00537 Bagherzadeh A Gholizadeh A Parametric-based neural networks and TOPSIS modeling in land suitability evaluation for alfalfa production using GIS Modeling Earth Systems and Environment 2017 3 1 2 10.1007/s40808-016-0263-y Beek KJ De Bie CA Driessen PM Land evaluation for sustainable land management 1997 ITC Bilgilioğlu SS Land suitability assessment for olive cultivation using GIS and multi-criteria decision-making in Mersin City, Turkey Arabian Journal Of Geosciences 2021 14 2434 10.1007/s12517-021-08768-8 Biswas S Pamucar D Kar S Sana SS A New Integrated FUCOM–CODAS framework with Fermatean fuzzy information for multi-criteria group decision-making Symmetry 2021 13 12 2430 10.3390/sym13122430 Braimoh AK Vlek PL Stein A Land evaluation for maize based on fuzzy set and interpolation Environmental Management 2004 33 2 226 238 15285400 Cao Q Esangbedo MO Bai S Esangbedo CO Grey SWARA-FUCOM weighting method for contractor selection MCDM problem: A case study of floating solar panel energy system installation Energies 2019 12 13 2481 10.3390/en12132481 Chaudhary BS Saroha GP Yadav M Human induced land use/land cover changes in northern part of Gurgaon district, Haryana, India: natural resources census concept Journal of Human Ecology 2008 23 3 243 252 10.1080/09709274.2008.11906077 Changere A Lal R Slope position and erosional effects on soil properties and corn production on a Miamian soil in central Ohio Journal of Sustainable Agriculture 1997 11 1 5 21 10.1300/J064v11n01_03 Chen Y Yu J Khan S Spatial sensitivity analysis of multi-criteria weights in GIS-based land suitability evaluation Environmental Modelling & Software 2010 25 12 1582 1591 10.1016/j.envsoft.2010.06.001 Costantini EA Manual of methods for soil and land evaluation 2009 Science 564 Dedeoğlu M Dengiz O Generating of land suitability index for wheat with hybrid system aproach using AHP and GIS Computers and Electronics in Agriculture 2019 167 105062 10.1016/j.compag.2019.105062 Deng, H. (1999). Multicriteria analysis with fuzzy pairwise comparison. In 1999 IEEE international fuzzy systems conference proceedings, vol 2, (pp. 726–731) Dengiz O Sarıoğlu FE Arazi değerlendirme çalışmalarında parametrik bir yaklaşım olan doğrusal kombinasyon tekniği Tarım Bilimleri Dergisi 2013 19 2 101 112 10.1501/Tarimbil_0000001234 Ecer F An analysis of the factors affecting wind farm site selection through FUCOM subjective weighting method Pamukkale University Journal of Engineering Sciences 2021 27 1 24 34 10.5505/pajes.2020.93271 Elamin, E. A., Tiblib, E., M. A., & Gaziri, M. E. (2005). Nitrogen fertilization and CaC03 interaction on corn (Zea. mays L.) grain and dry matter yields and their residual effects on Al-Marj Soil, Libya. Annals of Arid Zone, 44(1). El-Tilib MA Elamin EA El-Gaziri MM El-Mahi YE Combined effects on nitrogen fertilization and soil of CaCO3 contents on corn performance in Al-Marj Soil, Libya Journal of Plant Nutrition 2005 28 9 1619 1632 10.1080/01904160500203556 Everest, T. (2015). Detailed soil survey and mapping with land evaluation of Troia Historical National Park Lands. Ph.D. thesis. Canakkale Onsekiz Mart University, Graduate School, p. 203 Everest T Suitable site selection for pistachio (Pistacia vera) by using GIS and multi-criteria decision analyses (a case study in Turkey) Environment Development and Sustainability 2021 23 7686 7705 10.1007/s10668-020-00941-5 Everest T Gür E A GIS-based land evaluation model for peach cultivation by using AHP: A case study in NW Turkey Environmental Monitoring and Assessment 2022 194 241 10.1007/s10661-022-09898-6 35243554 Everest, T., Koparan, H., Sungur, A., & Özcan, H. (2021). An important tool against combat climate change: Land suitability assessment for canola (a case study: Çanakkale, NW Turkey). Environment, Development and Sustainability, pp. 1–36. Everest T Sungur A Özcan H Applying the best–worst method for land evaluation: a case study for paddy cultivation in northwest Turkey International Journal of Environmental Science and Technology 2022 19 3233 3246 10.1007/s13762-021-03373-4 FAO (1976). A framework for land evaluation. In: FAO Soils Bulletin 32, Food and Agriculture Organization, Rome, Italy. p. 72 ISBN: 9251001111 FAO. (1977). A framework for land evaluation, vol. 22. International Institute for Land Reclamation and Improvement, (p. 87). Feizi F Karbalaei-Ramezanali AA Farhadi S FUCOM-MOORA and FUCOM-MOOSRA: New MCDM-based knowledge-driven procedures for mineral potential mapping in Greenfields SN Applied Sciences 2021 3 3 1 19 10.1007/s42452-021-04342-9 Flynn KC Site suitability analysis for tef (Eragrostis tef) within the contiguous United States Computers and Electronics in Agriculture 2019 159 119 128 10.1016/j.compag.2019.02.016 Fugger, W. D. (1999). Evaluation of potential indicators for soil quality in Savanna soils in Northern Ghana. Ph.D. thesis, Georg-August University, Gottingen. Fujisao K Khanthavong P Oudthachit S Impacts of the continuous maize cultivation on soil properties in Sainyabuli province, Laos Scientific Reports 2020 10 11231 10.1038/s41598-020-67830-9 32641775 Ghorbanzadeh O Feizizadeh B Blaschke T Multi-criteria risk evaluation by integrating an analytical network process approach into GIS-based sensitivity and uncertainty analyses Geomatics Natural Hazards and Risk 2018 9 1 127 151 10.1080/19475705.2017.1413012 Günal, H., Kılıç, O. M., Ersayın, K., & Acir, N. (2022). Land suitability assessment for wheat production using analytical hierarchy process in a semi-arid region of Central Anatolia. Geocarto International, (just-accepted), pp. 1–18. Iqbal, S., Hussain, S., Qayyaum, M. A., & Ashraf, M. (2020). The response of maize physiology under salinity stress and its coping strategies. In Plant Stress Physiology. IntechOpen. Jahanpoor, N., Momtaz, H. R., & Soori, M. (2018). Application of PROMOTHEE technique to evaluate the optimal land suitability for pomegranate and pistachio crops in Miandoab Plain. Jimoh AI Yusuf YO Yau SL Soil suitability evaluation for rain-fed maize production at Gabari District Zaria Kaduna State, Nigeria Ethiopian Journal of Environmental Studies and Management 2016 9 2 137 147 10.4314/ejesm.v9i2.2 Kane DA Bradford MA Fuller E Oldfield EE Wood SA Soil organic matter protects US maize yields and lowers crop insurance payouts under drought Environmental Research Letters 2021 16 4 044018 10.1088/1748-9326/abe492 Kheybari S Javdanmehr M Rezaie FM Rezaei J Corn cultivation location selection for bioethanol production: An application of BWM and extended PROMETHEE II Energy 2021 228 120593 10.1016/j.energy.2021.120593 Leclerc ML Nolin MC Cluis D Simard RR Grouping soils of the Montreal Lowlands (Quebec) according to fertility and P sorption and desorption characteristics Canadian Journal of Soil Science 2001 81 1 71 83 10.4141/S00-021 Makungwe, M., Chabala, L. M., Van Dijk, M., Chishala, B. H., & Lark, R. M. (2021). Assessing land suitability for rainfed paddy rice production in Zambia. Geoderma Regional, 27, 00438. Mendas A Mebrek A Mekranfar Z Comparison between two multicriteria methods for assessing land suitability for agriculture: Application in the area of Mleta in western part of Algeria Environment Development and Sustainability 2021 23 9076 9089 10.1007/s10668-020-01012-5 Mendez MJ Volke-Haller VH Cortés-Flores JI Galvis-Spínola A Santiago-Cruz MDJ Soil organic matter and grain yield of rainfed maize in Luvisols of Campeche México Agricultural Sciences 2019 10 12 1602 1613 10.4236/as.2019.1012118 MGM (2021). Turkish state meteorological service. Meteorological bulletin (in Turkish). https://www.mgm.gov.tr/ Accessed 21 December 2021. Mistri, P., & Sengupta, S. (2019). Multi-criteria decision-making approaches to Agricultural Land Suitability classification of Malda District, Eastern India (pp. 1–20). Natural Resources Research. Montgomery B Dragićević S Dujmović J Griffith D Chun Y Dean D Using soft computing logic and the logic scoring of preference method for agricultural land suitability evaluation Advances in geocomputation 2017 Springer Moreno-Jiménez JM Aguarón J Escobar MT The core of consistency in AHP-Group decision making Group Decision and Negotiation 2008 17 249 265 10.1007/s10726-007-9072-z Nash P Nelson K Motavalli P Corn response to drainage and fertilizer on a poorly drained, river bottom soil Agronomy Journal 2015 107 1801 1808 10.2134/agronj14.0380 Nielsen, R. L. (2012). Corn management for extreme conditions. http://www.kingcorn.org/news/timeless/ExtremeCornMgmt.html Odeh IOA McBratney AB Pedometrics. Encyclopedia of Soils in the Environment 2005 Elsevier 166 175 Özkan B Dengiz O Turan İD Site suitability assessment and mapping for rice cultivation using multi-criteria decision analysis based on fuzzy-AHP and TOPSIS approaches under semihumid ecological condition in delta plain Paddy and Water Environment 2019 17 4 665 676 10.1007/s10333-019-00692-8 Özkan B Dengiz O Turan İD Site suitability analysis for potential agricultural land with spatial fuzzy multi-criteria decision analysis in regional scale under semi-arid terrestrial ecosystem Scientific Reports 2020 10 1 1 18 10.1038/s41598-020-79105-4 31913322 Pamučar D Ecer F Deveci M Assessment of alternative fuel vehicles for sustainable road transportation of United States using integrated fuzzy FUCOM and neutrosophic fuzzy MARCOS methodology Science of The Total Environment 2021 788 147763 10.1016/j.scitotenv.2021.147763 34029824 Pamučar D Stević Ž Sremac S A new model for determining weight coefficients of criteria in mcdm models: full consistency method (FUCOM) Symmetry 2018 10 9 393 10.3390/sym10090393 Peralta, N. R., Alesso, C. A., Costa, J. L., & Martin, N. F. (2021). Mapping soil depth in southern pampas Argentina using ancillary data and statistical learning. Soil Science Society of America Journal. Pilevar AR Matinfar HR Sohrabi A Sarmadian F Integrated fuzzy, AHP and GIS techniques for land suitability assessment in semi-arid regions for wheat and maize farming Ecological Indicators 2020 110 105887 10.1016/j.ecolind.2019.105887 Preedy, V. R., & Watson, R. R. (2019). Flour and breads and their fortification in health and disease prevention. Academic press. Puska, A., Stević, Ž., & Pamučar, D. (2021). Evaluation and selection of healthcare waste incinerators using extended sustainability criteria and multi-criteria analysis methods (pp. 1–31). Environment, Development and Sustainability. Rahman MM Soaug AA Darwish FHA Golam F Sofian-Azirun M Growth and nutrient uptake of maize plants as affected by elemental sulfur and nitrogen fertilizer in sandy calcareous soil African Journal of Biotechnology 2011 10 60 12882 12889 10.5897/AJB11.2075 Ramamurthy, V., Reddy, G. O., & Kumar, N. (2020). Assessment of land suitability for maize (Zea mays L) in semi-arid ecosystem of southern India using integrated AHP and GIS approach. Computers and Electronics in Agriculture, 179, 105806. Rezaei J Best–worst multi-criteria decision-making method Omega 2015 53 49 57 10.1016/j.omega.2014.11.009 Rezaei J Best–worst multi-criteria decision-making method:some properties and a linear model Omega 2016 64 126 130 10.1016/j.omega.2015.12.001 Rossiter DG A theoretical framework for land evaluation Geoderma 1996 72 3 165 190 10.1016/0016-7061(96)00031-6 Saaty TL The analytic hierarchy process: Planning, priority setting, resources allocation 1980 McGraw 281 Sabagh, E. L., Ayman, Ç. F., Seydoşoğlu, S., Battaglia, M. L., Javed, T., Iqbal, M. A., Mubeen, M., Ali, M., Ali, M., Bengisu, G., Konuşkan, Ö., Barutcular, C., Erman, M., Açikbaş, S., Hossain, A., Islam, M. S., Wasaya, A., Ratnasekera, D., Arif, M., Ahmad, Z., & Awad, M. (2021). Salinity stress in maize: Effects of stress and recent developments of tolerance for improvement, cereal grains—Volume 1, Aakash Kumar Goyal, IntechOpen. 10.5772/intechopen.98745. Available from: https://www.intechopen.com/chapters/77397 Sadras VO Calvino PA Quantification of grain yield response to soil depth in soybean, maize, sunflower, and wheat Agronomy Journal 2001 93 3 577 583 10.2134/agronj2001.933577x Sarkar, B., Das, P., Islam, N., Basak, A., Debnath, M., & Roy, R. (2021). Land suitability analysis for paddy crop using GIS-based Fuzzy-AHP (F-AHP) method in Koch Bihar district, West Bengal (pp. 1–26). Geocarto International. Seyedmohammadi J Sarmadian F Jafarzadeh AA McDowell RW Development of a model using matter element, AHP and GIS techniques to assess the suitability of land for agriculture Geoderma 2019 352 80 95 10.1016/j.geoderma.2019.05.046 Sharma R Kamble SS Gunasekaran A Big GIS analytics framework for agriculture supply chains: A literature review identifying the current trends and future perspectives Computers And Electronics In Agriculture 2018 155 103 120 10.1016/j.compag.2018.10.001 Srivastava RR Singh PK Reuse-focused selection of appropriate technologies for municipal wastewater treatment: A multi-criteria approach International Journal of Environmental Science and Technology 2021 19 1 18 Sys, C., Ranst, V., Debaveye, J., & Beernaert, F. (1993). Land evaluation. Part III, crop requirements (7). Agricultural Publications No. Tandzi NL Mutengwa CS Ngonkeu ELM Gracen V Breeding maize for tolerance to acidic soils: A review Agronomy 2018 8 6 84 10.3390/agronomy8060084 Tashayo B Honarbakhsh A Azma A Akbari M Combined fuzzy AHP–GIS for agricultural land suitability modeling for a watershed in southern Iran Environ Manag 2020 66 1 13 10.1007/s00267-020-01310-8 TEPGE. (2021). Tarımsal Ekonomi ve Politika Geliştirme Enstitüsü Müdürlüğü. Mısır Sektör Raporu Ankara-Turkey. Tercan, E., & Dengiz, O. (2022). Determination of rice (Oryza sativa L.) site suitability using multi-criteria decision-making based on the best-worst method. Geocarto International, (just-accepted), pp. 1–16. The C Calba H Zonkeng C Ngonkeu ELM Adetimirin VO Response of maize grain yield to changes in acid soil characteristics after soil amendment Plant and Soil 2006 284 45 57 10.1007/s11104-006-0029-9 Ulutaş A Karakuş CB Location selection for a textile manufacturing facility with GIS based on hybrid MCDM approach Industria Textila 2021 72 2 126 132 10.35530/IT.072.02.1736 Upadhyay, S., & Raghubanshi, A. S. (2020). Determinants of soil carbon dynamics in urban ecosystems. Urban Ecology (pp. 299–314). Elsevier. Wanyama, D., Mighty, M., Sim, S., & Koti, F. (2019). A spatial assessment of land suitability for maize farming in Kenya (pp. 1–18). Geocarto International. Zabihi H Ahmad A Vogeler I Said MN Golmohammadi M Golein B Nilashi M Land suitability procedure for sustainable citrus planning using the application of the analytical network process approach and GIS Computers and Electronics in Agriculture 2015 117 114 126 10.1016/j.compag.2015.07.014 Zhang J Su Y Wu J Liang H GIS based land suitability assessment for tobacco production using AHP and fuzzy set in Shandong province of China Computers and Electronics in Agriculture 2015 114 202 211 10.1016/j.compag.2015.04.004 Zhang S Liu X Wang X Gao Y Yang Q Evaluation of coffee ecological adaptability using fuzzy, AHP, and GIS in Yunnan Province, China Arabian Journal of Geosciences 2021 14 14 1 18 10.1007/s12517-021-07795-9 Ziadi N Cambouris AN Nyiraneza J Nolin MC Across a landscape, soil texture controls the optimum rate of N fertilizer for maize production Field Crops Research 2013 148 78 85 10.1016/j.fcr.2013.03.023 ZMO (2016). Ziraat Mühendisler Odası 2016 Yılı Mısır Raporu. Zolekar RB Integrative approach of RS and GIS in characterization of land suitability for agriculture: A case study of Darna catchment Arabian Journal of Geosciences 2018 11 24 1 17 10.1007/s12517-018-4148-4	36506642	PMC9718473	NO-CC CODE	2022-12-06 23:23:39	no	Environ Dev Sustain. 2022 Dec 2;:1-20	utf-8	Environ Dev Sustain	2,022	10.1007/s10668-022-02787-5	oa_other
==== Front Fract Calc Appl Anal Fract Calc Appl Anal Fractional Calculus & Applied Analysis 1311-0454 1314-2224 Springer International Publishing Cham 111 10.1007/s13540-022-00111-6 Original Paper Limitations and applications in a fractional Barbalat’s Lemma http://orcid.org/0000-0002-0556-2462 Zeraick Monteiro Noemi [email protected] 1 http://orcid.org/0000-0002-6863-3723 Rodrigues Mazorche Sandro [email protected] 2 1 grid.411198.4 0000 0001 2170 9332 Postgraduate Program in Computational Modeling, Federal University of Juiz de Fora, Str. José Lourenço Kelmer, Juiz de Fora, 36036-900 Minas Gerais Brazil 2 grid.411198.4 0000 0001 2170 9332 Department of Mathematics, Federal University of Juiz de Fora, Str. José Lourenço Kelmer, Juiz de Fora, 36036-900 Minas Gerais Brazil 2 12 2022 123 15 5 2022 14 11 2022 22 11 2022 © Diogenes Co.Ltd 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Barbalat’s Lemma is a mathematical result that can lead to the solution of many asymptotic stability problems. On the other hand, Fractional Calculus has been widely used in mathematical modeling, mainly due to its potential to make explicit the dependence of previous stages through nonlocal operators. In this work, we present a fractional Barbalat’s Lemma and its proof, as proposed in [31]. The proof is analyzed in order to show an imprecision. In fact, for orders 0<α<1, we are not able to get the supreme limit of the integrand. Then, a counterexample and a corrected version of the lemma are presented, according to [9]. The objective of this work is to draw attention to the potential and limitations of a fractional Barbalat’s Lemma, given its wide use in recent articles. In a fractional SIR model, we exhibit the constraint of the result by introducing a non-periodic relapse. So, the supreme limit could not be verified. Also in this context, we provide a general discussion of the classical Calculus’ properties that are not inherited if we change the integer orders to fractional ones. Keywords Fractional Calculus (primary) Barbalat’s Lemma persistence of properties Mathematics Subject Classification 26A33 (primary ) 34D20 65R99 92D30 Coordination for the Improvement of Higher Education Personnel - Brazilhttp://dx.doi.org/10.13039/501100007378 Universidade Federal de Juiz de Fora ==== Body pmcIntroduction Abel is remembered as the first mathematician that applied the Arbitrary-Order Calculus, more known as Fractional Calculus, to a real-world problem. In fact, Abel’s solution for the tautochrone problem, based upon mathematical analysis, leads to a fractional integral. After that, some pioneers of the applications of Fractional Calculus are recalled in [27]. However, it was only after the first International Conference on Fractional Calculus and Applications, in 1974, that the Fractional Calculus has been broadly used in several areas. One can cite applications in epidemiology [2], quantum transport [7], fibrillation [26], diffusion [24], and so on. In addition, behaviors of long memory type can also be studied via delay differential equations [3], which allows a cross between the Fractional Calculus and the theory of DDE’s, expanding the range of applications and studies. Thus, it is natural that the theories of the classical Calculus, in particular the integer-order differential equations, are revisited, in view of their possible adaptation to Fractional Calculus. This adaptation, however, is not always immediate. For instance, the Mean Value theorem is not local in Fractional Calculus, which implies that the sign of the derivative is not sufficient to indicate the monotonicity. In this context, we aim to verify the validity of a fractional Barbalat’s Lemma. In ODE’s theory, Barbalat’s Lemma is a mathematical result concerning the asymptotic properties of functions and their derivatives. When used correctly for dynamical systems, it can lead to the solution of many asymptotic stability problems, among which we emphasize compartmental epidemiological models ([4, 10]). In general terms, it deals with the convergence to zero of a sufficiently well-behaved function whose integral is bounded. An intuitive idea for the generalization of this lemma would be to consider the fractional integral of order α in the statement. In 2017, this generalization was publicized in [31]. However, in 2015, [9] had already demonstrated that the lemma is not valid for 0<α<1. Even though the same authors are cited in [31], the proved fallibility of the lemma is not considered. The Barbalat’s Lemma as proposed in [31] is used in many works from last years, some of which also cite the authors of [9]. Two examples from 2020 are [12] and [30]. Already in 2022, we found in [13] the same lemma, used to investigate the global stability of a fractional-order HBV infection. Also in 2022, the result is used for the Caputo-Fabrizio fractional derivative, in an asymptotic analysis of novel coronavirus disease via fractional-order epidemiological model [14]. Thus, given the importance of the lemma for recent biomathematical models, it is necessary to provide a dialogue between the proof of [31] and the counterexample of [9]. For this, we introduce in Sections 2 and 3 some preliminary discussion and the classic Barbalat’s Lemma. So, in Section 4, we discuss the generalization problems. Hoping to attract the attention of researchers to the care with the use of the fractional Barbalat’s Lemma, we also provide some numerical results. In Section 5, we discuss conditions for the validity of fractional Barbalat’s Lemma and, in Section 6, we end with a brief discussion about Lyapunov’s fractional theory and an example of application in the Caputo fractional SIR model with and without relapse. The Fractional Calculus and some of its uncommon behaviors Fractional Calculus is over 320 years old, but only in the last few decades has it shown substantial growth as an area of research. It is a fertile area, including allowing to deal explicitly with the “memory effect” of many phenomena, considering the dependence of previous stages on materials or processes and, in this context, optimizing the modeling of phenomena and systems. However, Fractional Calculus has some unusual behaviors and, in this sense, many classical Calculus’ theories need to be revised before they can continue to be used. In [21], we draw attention to properties that are not inherited from the integer-order Calculus, beginning with the issue of different definitions, then moving on to Mean Value theorems and topics of “persistence” of properties. As the focus of this work is to discuss different approaches about a fractional Barbalat’s Lemma and parallel questions, we begin with some basic considerations. Not only the Barbalat’s Lemma cannot be directly translated to fractional integrals, but even the sign of the derivative is no longer valid as an indication of the monotone behavior of the function, as we see in Subsection 2.2. Preliminaries The Fractional Calculus probably was born in 1695, when, as the legend says, l’Hôpital asked Leibniz about the meaning of a derivative of order 1/2. Over next centuries, important advances were made by Liouville, Riemann, Grünwald, Caputo, and many others. However, it was only after the first International Conference on Fractional Calculus and Applications, in 1974, that the number of researchers in Fractional Calculus showed great growth. Currently, congresses and symposia take place more frequently, and the reader may refer to the reference [5] for a chronology of publications in Fractional Calculus until 2019, as well as for general results. Below we consider [a, b] a finite real interval, and α a real number such that 0≤n-1<α<n, with n integer. The extension of the idea of the iterated integral leads to the following definition: Definition 1 (Riemann-Liouville integral in finite intervals) The Riemann-Liouville integral of an arbitrary order α is set to t∈[a,b] by2.1 Ia+αf(t)=1Γ(α)∫at(t-θ)α-1f(θ)dθ. After introducing the arbitrary-order integral, it is natural to search for the definition of the corresponding derivative: Definition 2 [Riemann-Liouville derivative in finite intervals] The Riemann-Liouville derivative of an arbitrary order α is set to t∈[a,b] by2.2 Da+αf(t)=Dn[Ia+n-αf(t)]=1Γ(n-α)dndtn∫at(t-θ)n-α-1f(θ)dθ, with Dn representing the integer-order derivative. We also present the definition of Caputo derivative of arbitrary order, for which, among other characteristics, the derivative of a constant is zero: Definition 3 (Caputo derivative in finite intervals) The Caputo derivative with arbitrary order α is defined for t∈[a,b] by2.3 CDa+αf(t)=Ia+n-α[Dnf(t)]=1Γ(n-α)∫at(t-θ)n-α-1dndθnf(θ)dθ. In the next section, we see that the sign of the fractional derivative does not imply monotonicity, which, in particular, is a challenge to fractional Lyapunov theories. The sign of the fractional derivative does not imply monotonicity In classical Calculus, the sign of the derivative indicates when the function is decreasing or increasing. However, in Arbitrary-Order Calculus, this feature does not have a simple correspondent: the sign of the arbitrary-order derivative is not indicative of monotonicity. This is a great challenge in the recent plenty of publications in Fractional Calculus. In fact, based on an understandable mistake about the fractional Mean Value theorem, many authors (e.g. [11, 13]) state the following result, especially in the case of the arbitrary-order Caputo derivative: Let α∈(0,1) and suppose f(t),CDa+αf(t)∈C[a,b]. It follows from the Mean Value theorem that, if CDa+αf(t)≥0 in the interval [a, b], then f is nondecreasing in [a, b]. Similarly, if CDa+αf(t)≤0 in the interval [a, b], then f is nonincreasing in that interval. The problem is that, without other hypotheses, this result is not valid. The example that we discuss is based on [6] and, in [15], we also can see that this assertion fails in the SIR model. Consider the function y(t)=t3-1.5t2+0.5t. We have y′(t)=3t2-3t+0.5. Given any α∈(0,1), the arbitrary-order derivative in the Caputo sense of y(t) is given by2.4 CD0+αy(t)=t1-αΓ(4-α)(6t2-3(3-α)t+0.5(3-α)(2-α)). Thus, CD0+αy(t) has no sign variation in the interval2.5 Tα=0,3(3-α)-3(1+α)(3-α)12. Now, we notice that2.6 ddα[3(3-α)-3(1+α)(3-α)12]=112[-3(1-α)(1+α)(3-α)-3]<0. So,Fig. 1 Integer and fractional derivatives Fig. 2 Behavior of y(t) 2.7 3(3-α)-3(1+α)(3-α)12>3(3-1)-3(1+1)(3-1)12=3-36. This implies that, for any α∈(0,1), CD0+αy(t) has no sign variation in the interval Tα=0,(3-3)/6+ϵα for some ϵα>0, as illustrated in Figure 1. However, y(t) has a local maximum at (3-3)/6, as we can see in Figure 2. So, despite the fact that we have CD0+αy(t)≥0 for all t∈Tα, the function y(t) is not monotonous in Tα. In [16], we also discuss the difficulty of defining initial conditions for IVP, since, as the arbitrary-order derivatives are nonlocal, all the past must be taken into account. Moreover, we highlight the difference in the analysis of extreme points and equilibrium, in comparison with classical Calculus, between some other aspects that deserve attention. The classic Barbalat’s Lemma Now, we focus on the classic Calculus’ result that is supposed to be extended to Fractional Calculus. But, before discussing Barbalat’s Lemma, we recall two points about asymptotic properties of functions and their derivatives. Given a function of t, the following facts are important: Remark 1 [25] The limit df/dt→0 does not imply convergence of f. Consider, for example, the function f(t)=sin(logt). While3.1 dfdt=cos(logt)t→0, the f function continues to oscillate, more and more slowly, as illustrated in Fig. 3 Remark 2 [25] Convergence of f does not imply df/dt→0. For example, the nonnegative function f(t)=e-tsin2(e2t) tends to zero, but its derivative,3.2 dfdt=4etcos(e2t)sin(e2t)-e-tsin2(e2t), is unlimited, as illustrated in Fig. 4 . Fig. 3 Remark 1: f(t)=sin(log(t)) Fig. 4 Remark 2: f(t)=e-tsin2(e2t) So, given that a function tends to a finite limit, what additional requirement can guarantee that its derivative converges to zero? Barbalat’s Lemma indicates that the derivative itself must have some smoothness. More precisely, Lemma 1 [Barbalat] If the differentiable function f(t) has a finite limit when t→∞, and if df/dt is uniformly continuous, then df/dt→0 when t→∞. The proof of the lemma can be consulted at [25]. To apply Barbalat’s Lemma to the analysis of dynamical systems, we usually use the following immediate corollary: Lemma 2 (Lemma “Lyapunov-type”) [25] If a scalar function V(x, t) satisfies the following conditions:V(x, t) is bounded from below, V˙(x,t) is negative semidefinite, V˙(x,t) is uniformly continuous in time, then V˙(x,t)→0 when t→∞. In fact, V approaches a finite limit V∞, such that V∞≤V(x(0),0) (this does not require uniform continuity). The above lemma then follows from Barbalat’s Lemma. In this work, we use the following version of the lemma, which can be obtained directly from the first one: Lemma 3 [Barbalat] If the uniformly continuous nonnegative function f(t) in [0,∞] is such that ∫0tf(τ)dτ<C, for some constant C and all t>0, then f(t)→0 when t→∞. This version is broadly used in mathematical modeling. For instance, it was used to prove the boundedness of solutions of a viral infection in a pest control model, in [10]. In 2017, was published a result for the convergence of a function based on its fractional derivative and integration, given by Lemma 4 [28] If ∫0tf(τ)dτ has a finite limit as t→+∞, and if CD0+αf(t) is bounded, where f:[0,+∞)→R, then f(t)→0 as t→+∞, where 0<α<1. However, this is not always applicable, so an intuitive idea for the generalization of Lemma 3 is to consider the fractional integral of order α in its statement. Also in 2017, this generalization was publicized in [31], as we discuss in next section. Fractional Barbalat’s Lemma: the generalization problem As mentioned in the Introduction, in Theorem 3.1 of [31] and related references it is found the following fractional Barbalat’s Lemma: Let ϕ:R→R be a uniformly continuous function on [t0,∞], and, with α∈(0,1), p and M two positive constants, It0+α∣ϕ(t)∣p≤M for all t>0. Then,limt→∞ϕ(t)=0. The case α=1 is the traditional case. Let us discuss the presented demonstration. We assume for the sake of contradiction that there exists a positive scalar ε and a sequence (tk)k∈N with tk→∞ such that ∣ϕ(tk)∣>ε. Let us assume without loss of generality that ∣tk+1-tk∣>δ0 for all k. This implies that the intervals [tk-δ0/2,tk+δ0/2] do not overlap. By the uniform continuity of ϕ, there is a δ, which we will assume without loss of generality less than δ0/2, such that4.1 ∣ϕ(t′)-ϕ(t′′)∣<ε2, for any t′,t′′ such that ∣t′-t′′∣<δ. Then, for every t in [tk-δ,tk+δ], we have4.2 ∣ϕ(t)∣≥∣ϕ(tk)∣-∣ϕ(tk)-ϕ(t)∣>ε2, whence4.3 ∣ϕ(t)∣p>ε2p. If 0<α<1, and n>k, we have -2δ/(tn-tk+δ)>-1. So, by Bernoulli’s inequality,4.4 1-2δtn-tk+δα≤1-2δαtn-tk+δ, that is,4.5 1-1-2δtn-tk+δα≥2δαtn-tk+δ. Then, for n>k, we have4.6 (tn-tk+δ)α-(tn-tk-δ)α=(tn-tk+δ)α1-1-2δtn-tk+δα≥(tn-tk+δ)α2δαtn-tk+δ≥2δα(tn-t1+δ)α-1. Hence, it follows that, for t>tn,4.7 It0+α∣ϕ(t)∣p≥1Γ(α)∫t0tn∣ϕ(τ)∣p(tn-τ)α-1dτ≥1Γ(α)∑k=1n-1∫tk-δtk+δ∣ϕ(τ)∣p(tn-τ)α-1dτ+1Γ(α)∫tn-δtn∣ϕ(τ)∣p(tn-τ)α-1dτ≥εp2pαΓ(α)∑k=1n-1[(tn-tk+δ)α-(tn-tk-δ)α]+εpδα2pαΓ(α)≥εp(n-1)δ(tn-t1+δ)α-12p-1Γ(α)+εpδα2pαΓ(α). A touchy situation occurs here, since it is considered γ=max{tk+1-tk}, for k=1,2,⋯,n-1. With this assumption, we have4.8 It0+α∣ϕ(t)∣p≥εp(n-1)δ[(n-1)γ+δ]α-12p-1Γ(α)+εpδα2pαΓ(α)=εp(n-1)αδ2p-1Γ(α)[γ+δ/(n-1)]1-α+εpδα2pαΓ(α). So, letting n→∞, we get It0+α∣ϕ(t)∣p→∞, contradiction. This implies that ϕ(t)→0. Remark 3 The proof above considers γ=max{tk+1-tk},k=1,2,⋯,n-1. Thus, γ depends on n. If we write γn to remember this dependency, what we get is4.9 It0+α∣ϕ(t)∣p≥εp(n-1)αδ2p-1Γ(α)[γn+δ/(n-1)]1-α+εpδα2pαΓ(α). When doing n→∞, we should be able to study limn→∞γn. If we have limsupn→∞γn=L for finite L, then the proof is valid. However, the proof completion can be invalid if we have limsupn→∞γn=∞. As in the case of the function exhibited in Proposition 1, we can have It0+α∣ϕ(t)∣p limited, even though ϕ(t)↛0. The idea is that the influence of the past is decreasing, with has relation with the Short Memory principle of the arbitrary-order operators [23]. In fact, for 0<α<1, a direct generalization of Barbalat’s Lemma is not true, as shown in next proposition, published in 2015: Proposition 1 [9] If 0<α<1, there exists a nonnegative uniformly continuous function f such that I0+αf(t)<M for all t, with M a positive constant, but f does not converge to zero when t goes to infinity. The proof uses the following lemma: Lemma 5 [9] If f is a bounded function that vanishes for all t>T, then I0+αf→0. Also, I0+αf will be a uniformly continuous function. Proof Proceeding with the proof of Proposition 1, let p(t) be a null function at all points except the intervals [ti,ti+δ], where it takes the value 1 (nonperiodic pulse), with (ti)i∈N an increasing divergent sequence to be specified and δ a fixed positive real to be specified. Note that, for every t and every τ<t, the function p(τ) can be written as p(τ)=∑i=1npi(τ), where n=max{i:t≥ti} and pi(t) is a function that vanishes outside the interval [ti,ti+δ], in which it takes the value 1. For every i, we have I0+αpi(t)<C1, where C1 is a positive constant. Indeed,If t<ti, then I0+αpi(t)=0; If t∈[ti,ti+δ], then 4.10 I0+αpi(t)=1Γ(α)∫tit(t-τ)α-1dτ=(t-ti)ααΓ(α)≤δααΓ(α); If t>ti+δ, then 4.11 I0+αpi(t)=1Γ(α)∫titi+δ(t-τ)α-1dτ=(t-ti)α-(t-ti-δ)ααΓ(α)≤I0+αpi(ti+δ)=δααΓ(α), since I0+αpi(t) is monotonous decreasing for t>ti+δ, because, in this case, 4.12 ddt[(t-ti)α-(t-ti-δ)α]=α[(t-ti)α-1-(t-ti-δ)α-1]<0. By the property of Lemma 5, for all i, we have I0+αpi(t)→0. So, by the definition of convergence, there is Ti such that, for every t>Ti, we have I0+αpi(t)<1/i2. Since (ti)i∈N is a divergent sequence, we can without loss of generality exclude unnecessary terms and recursively redefine the sequence in a way that ti+1>Ti. In particular, this makes the maximum step γn of Remark 3 tending to infinity, which characterizes the conflict between the two proofs. A viable choice would be to consider ti+1=Ti+1 and δ defined by the equality δαα-1Γ(α)-1=1. So, we have I0+αpi+1(ti+1+δ)=1, and4.13 I0+αpi+1(Ti+1)≤1(i+1)2≤1, which implies Ti+1≥ti+1+δ>ti+1=Ti+1. Although I0+αpi+1(t)>I0+αpi(t) for all t>ti+1+δ, the idea is that, for t>Ti+1>Ti, I0+αpi(t) is so small, and I0+αpi(t)<I0+αpi+1(t)<1/(i+1)2. This is illustrated in Figure 7. Note that the right tail of each wave is above of the right tail of previous wave, but all of them are decreasing. Hence, ti+1-ti>1 and (ti)i∈N is an increasing divergent sequence, as expected. Also,4.14 ti+1-(ti+δ)=Ti-(ti+δ)+1>0, hence the intervals [ti,ti+δ] do not overlap. In this context, for any t∈R+, there is n∈N such that tn+1≤t≤tn+2. By the linearity of the integral operator, we can write4.15 I0+αp(t)=I0+α∑i=1npi(t)+pn+1(t)=∑i=1nI0+αpi(t)+I0+αpn+1(t). By construction, it follows that4.16 I0+αp(t)≤∑i=1n1i2+I0+αpn+1(t)<2+C1. Thus, we have a limited, positive function p that does not vanish at infinity and whose fractional integral remains limited. Now let f(t) be a positive triangular function such that, for all t>0, we have f(t)≤p(t). For example, let f be null at every point, except in the intervals [ti,ti+δ/2], where it takes the values 2δ-1(t-ti) and in the intervals [ti+δ/2,ti+δ], where it takes the values 2δ-1(ti+δ-t). Then, f is uniformly continuous (more than that, it is Lipschitz continuous with constant 2δ-1), is bounded, positive, does not vanish at infinity, and is such that, if C≥2+C1, one has4.17 I0+αf(t)≤I0+αp(t)<C, for all t. This completes the proof. □ Numerical results In this subsection, we discuss Proposition 1 numerically. For that, we constructed a MATLAB code for plotting functions f and their Riemann-Liouville integrals I0+αf. We define the f function with bumps in each interval [ti,ti+δ] defined in last section, taking4.18 g(t)=exp1(t-ti)(t-ti-δ),ift∈(ti,ti+δ),0,otherwise, and normalize by convenience, writing f(t)=g(t)/max(g(t)). Note that this function is C∞ and, as its derivative is bounded, f is also Lipschitz. For the sequence (ti)i∈N, we define t1=2 and, based in Eq. (4.11), we solve recursively the inequatility4.19 xα-(x-1)α<1/i2. So, we take ti+1=δx+ti+1. This is sufficient for the proof’s purpose. Figure 5 exhibits a suitable function for α=0.5 and δ=(Γ(α+1))1/α, as proposed. In Figure 6, we can see closely the bumps’ profile.Fig. 5 A Lipschitz aperiodic function Fig. 6 Bumps’ profile In Figure 7, are illustrated the fractional integrals over time for each one of the bumps of Figure 5.Fig. 7 Each integral wave illustrate the fractional integral for each bump We note that the function does not converge to 0. However, I0+αf, the sum of the waves of Figure 7, is illustrated in Figure 8, and is limited, because each bump only can occur when the definite integral until the time t is already lower than a threshold.Fig. 8 The limited Riemann-Liouville integral of f Figures 9 and 10 exhibit the case α=1, for which the classic Barbalat’s Lemma implies that I0+αf diverges. The same occurs with α>1, as proved in Lemma 6. Of course, in this case we are not able to solve Eq. (4.19), because 1<1/i2 is an absurd. So, the example is merely illustrative.Fig. 9 A possible suitable function Fig. 10 Unlimited R-L integral, α=1 Finally, we exemplify that, if the step of the sequence (ti)i∈N is limited, so the Barbalat’s Lemma with strong limit is valid, as proposed in [31]. This is precisely the heart of the counterexample constructed for the sake to prove Proposition 1. So, Figures 11 and 12 illustrate that, if without loss of generality the step is regular, the integral diverges.Fig. 11 A function with regular bumps Fig. 12 Unlimited R-L integral, α=0.5 This detail is what [31] considers as ubiquitous in the discussed proof. But, as seen, it may be not true, and the supremum limit of the integrand may be not zero if the step size of the divergent sequence goes to infinity. The fractional Barbalat’s Lemma, with conditions From the previous discussion, it is not possible to generalize Barbalat’s Lemma to the case 0<α<1. However, the result is valid for α≥1: Lemma 6 [9] Let α≥1 and f be a nonnegative uniformly continuous function such that I0+αf(t)<M for all t, with M a positive constant. So, f converges to zero. Proof The proof is given by contradiction. We assume that f does not converge to zero. Then, there is ε>0 and an increasing divergent sequence (ti)i∈N such that f(ti)>ε. Since f is uniformly continuous, there is δ>0 such that, for all i∈N, if ∣t-ti∣<δ, then ∣f(t)-f(ti)∣≤ε/2. So, if t∈[ti,ti+δ], we have5.1 f(t)≥f(ti)-∣f(ti)-f(t)∣>ε/2. Let p(t) be a null function at every point except if t∈[ti,ti+δ], where it takes the value ε/2. By definition, for t>1,5.2 Γ(α)I0+αf=∫0t-1(t-τ)α-1f(τ)dτ+∫t-1t(t-τ)α-1f(τ)dτ. Since f is a positive function, then5.3 Γ(α)I0+αf≥∫0t-1(t-τ)α-1f(τ)dτ. Now, for 0≤τ≤t-1 and α≥1, remembering that f(t)≥p(t) for all t, we have5.4 Γ(α)I0+αf≥∫0t-1f(τ)dτ≥∫0t-1p(τ)dτ≥ntεδ2, where nt=max{i:ti+δ≤t-1}. Taking the limit when t→∞, we get5.5 Γ(α)I0+αf(t)≥limt→∞ntεδ2=∞, contradiction. Therefore, f converges to zero. Remark 4 Since I0+αf can be expressed in terms of tα-1∗f, where ∗ denotes the convolution operator, the lemma can be extended: if the convolution g∗f is uniformly bounded, where g is a nonnegative monotone increasing function (g(t)=0 for t<0) and f is a uniformly positive continuous function, then f converges to zero at infinity. We note that (5.4) is not valid for 0<α<1. Therefore, we cannot extend the proof to these values of α, which is supported by the previous section. However, one can at least assure the statement given in the next lemma: Lemma 7 [9] Let f be a nonnegative bounded function such that I0+αf(t)<M for all t, with M a positive constant. So, lim inft→∞f(t)=0. Proof Within the context, the proof is even quite simple. Since f is bounded and nonnegative, lim inft→∞f(t) exists and is nonnegative. Let us assume for the sake of contradiction that lim inft→∞f(t)=l>0. Given sufficiently small ε>0, there exists Tε>0 such that f(t)>l-ε>0 for t>Tε. Define the function g such that g takes the values of f for t≤Tε and g(t)=l-ε for t>Tε. So, given t>Tε, we have5.6 I0+αf(t)≥I0+αg(t)=1Γ(α)∫0t(t-τ)α-1g(τ)dτ=1Γ(α)∫0Tε(t-τ)α-1g(τ)dτ+(l-ε)(t-Tε)αα≥(l-ε)(t-Tε)αα→∞, contradiction. We recommend the reference [9] for an extensive study of results with stronger hypotheses and extra conditions. Application It is natural that one thinks beyond and, inspired in the Lyapunov-type Lemma 2 for the classical Barbalat’s Lemma, aims to extend other features of the Lyapunov’s theory to Fractional Calculus. The same authors of [9] recall in [8] that Lyapunov’s basic theory requires monotonicity with respect to time for the functional L(x(t), t). A fundamental tool to prove its monotonicity is the sign of its derivative. More than that, if L(t)=L(x(t),t) is monotonous decreasing and bounded, then it converges. But, as we see in Subsection 2.2, the property of monotonicity is not easily determined in Fractional Calculus. Thus, many issues remain controversial in fractional-order systems. In [8], several studies on Lyapunov’s basic theory are proposed, and one of the useful tools is Barbalat’s Lemma 7. The classic SIR model In [16], we apply the presented theory to a SIR model of arbitrary order with the Caputo derivative. One of the results obtained directly from Barbalat’s Lemma is Theorem 1. We note that the result is not as conclusive as we would like: Theorem 1 In the fractional SIR model given by6.1 CD0+αS(t)=-βS(t)I(t)N, 6.2 CD0+αI(t)=βS(t)I(t)N-γI(t), 6.3 CD0+αR(t)=γI(t), we have6.4 lim inft→∞I(t)=0. Proof The proof consists in rewriting the last equation as6.5 I0+α(γI(t))=R(t)-R(0). Since S, I and R are limited [16], it follows from Lemma 7 that6.6 lim inft→∞I(t)=0. Remark 5 We cannot use here Lemma 4, because the integer-order integral of I(t) is unlimited, even though CD0+αI(t) is bounded. Proof In fact, it is assumed that the function S(t) is continuous in any finite interval [0, T], and so, takes on a minimum value m. Then, for t∈[0,T], we have6.7 CD0+αI(t)≥-AI(t), for some A. This implies the existence of a continuous positive function q(t) s.t.6.8 CD0+αI(t)=-AI(t)+q(t). Using the Laplace Transform, we get6.9 sαL{I}-sα-1I(0)=-AL{I}+L{q}, from which6.10 L{I}=sα-1I(0)+L{q}sα+A. Applying the inverse Laplace Transform and the Mittag-Leffler function (see, e.g., [5]), we get6.11 I(t)=I(0)Eα(-Atα)+q(t)⋆(tα-1Eα,α(-Atα)), where ⋆ is the Laplace convolution. Note that tα-1Eα,α(-Atα) is the derivative of -1AEα(-Atα) [16], which is increasing for any A. Once q(t) is also positive,6.12 I(t)≥I(0)Eα(-Atα)>0, for all t∈[0,T]. The same arguments prove the non-negativity of the other compartments, what is nontrivial because of nonlocal effects. Particularly, once S(t)>0 for all t, we can state A=γ and it is possible to write6.13 ∫0TI(t)dt≥I(0)∫0TEα(-γtα)dt, for all T. As Eα(-γtα) is not integrable, then ∫0∞I(t) diverges. Figure 13 illustrates the (S, I) plane, where is possible to observe several characteristics that do not persist when we change the order of the derivative, such as the monotonicity of S compartment, the stability region of the equilibrium (S,I,R)∞, in black, and the peak point condition. In the title, μ=0 indicates the absence of vital dynamics. The problem of non-monotonicity is not solved by balancing the units of the parameters initially considered: since the orders are the same, these corrections only numerically transform them into other constants. We notice that the peak points, in blue, no longer follow the relationship CD0+αI=0, varying according to α. The external trajectory is equivalent to the traditional model and the red dot indicates (S,I)=(γN/β,0). However, as we see in [15], the equilibrium is not globally stable. It is also possible to note that, as the greater is α, faster is the model.Fig. 13 Trajectories for I(0)=1, β=1.5, γ=0.5,α=0.2:0.1:1 Fig. 14 Zoom as t→∞ for the trajectory with α=0.9. Figure 14 illustrates the behavior of the solution for α=0.9 as t→∞. We notice autointersections of the trajectory, what does not occur in classical case. This is the simplest case where monotonicity and other classic behaviors are not maintained, but even so, its possible to observe numerically that limt→∞I(t)=0. However, we reassert that we cannot use the Barbalat’s Lemma for this limit, as it only gives us the minimum limit of I(t). This is also an important thought about other works we cite: the results may be correct; however, another kind of theory may be needed to prove them. Although, in next section, we insert a relapse rate varying as the f function of Subsection 4.1, illustrating that the supremum limit really cannot be extracted from Barbalat’s Lemma. SIR model with relapse - A critical example For this discussion, we consider as basis a SIR model with relapse, based in a fractional tuberculosis model [29]. For simplicity, we disregard the birth rate, the mortalities, the vaccination and the loss of immunity, taking only the relapse rate σ. So, the model is the same that the (6.1)-(6.3), only plus the relapse rate:6.14 CD0+αS(t)=-βS(t)I(t)N, 6.15 CD0+αI(t)=βS(t)I(t)N-γI(t)+σ(t)R(t), 6.16 CD0+αR(t)=γI(t)-σ(t)R(t). Once relapses are mainly due to treatment fails, we could suppose that they are less and less frequent, because of the new treatment developments and dissemination. So, for the purpose of this example, we consider σ(t)=σf(t), where f is the function defined in Subsection 4.1. Once 0≤R(t)≤N, in analogy with [16], we have6.17 γI0+αI(t)=R(t)-R(0)+σI0+αf(t)R(t)<N+σNI0+αf(t). This implies that I0+αI(t) is limited by some constant and, from Lemma 7, we conclude that lim inft→∞I(t)=0. Hence, as can be seen in Figures 15-16, the infimum limit of I(t) is the same than that of the model (6.1)-(6.3).Fig. 15 Comparison between the I compartment of the model with and without relapse - the infimum limit is maintained, but not the supremum Fig. 16 (S,I)-plane However, the supremum limit does not follow the same behavior. The oscillation of I is shown in Figures 15-16, indicating that lim supt→∞I(t)≠0. Its worth to note that lim supt→∞I(t)≠0 for any α∈(0,1], and this does not invalidates the integer-order analysis. In fact, in the integer case, i.e., when α=1, Barbalat’s Lemma hypothesis fail, once ∫0∞f(t) is not limited (see Figure 10). By other hand, if α<1, the fractional integral I0+αf(t) is limited, even though lim supt→∞I(t)≠0. This is the key point of the fractional Barbalat’s Lemma. An extend discussion of the model (6.14)-(6.16) is still open. Conclusion Although Arbitrary-Order Calculus is almost as old as the classical Calculus, with its origins in Leibniz, l’Hôpital and Bernoulli, its greatest expansion has occurred only in recent decades. Thus, it is natural that many results need validation from other researchers, until the theory of Fractional Calculus is consolidated as that of Classical Calculus. This process can take decades, and the contribution of each researcher is important in assembling the puzzle. In this sense, this work reviews the proof of a fractional Barbalat’s Lemma published in [31], where there is an imprecision. Unconsciously, one of the main assumptions in the proof is the existence of a limit to the step of a sequence (ti)i∈N such that f(ti)>ϵ. However, this is not mandatory. So, based in [9], a counterexample and a corrected result are presented, which uses the infimum limit. Then, we discuss in short the Lyapunov’s theory for fractional systems, showing another tricky point: the sign of a fractional derivative does not imply its monotonic behavior. In fact, there are several characteristics that do not persist when we change the order of the derivative. Next, we show a brief application of Barbalat’s Lemma 7 in the SIR model with Caputo fractional derivative. It is important to note that we could not guarantee that the limit of the I compartment is zero, but only its infimum limit. Even thought in the simplest example the strong limit of I(t) seems to be valid, we also discuss an example with relapse in which it is not. In fact, inserting a relapse rate σ(t), varying as the f function of Subsection 4.1, one can see that the supreme limit cannot be extracted from Barbalat’s Lemma. We remind that a broader study of this example is still open. Indeed, relapse models are useful for tuberculosis, herpes, and other major diseases. For completeness, we recall in this last section that, in our research project, we are interested in the following question: is it possible to build fractional SIR-type models with precise mathematical and biological basis like that of the original construction did by Kermack & McKendrick? What characteristics are maintained simply exchanging orders? Does the change in the order of derivatives automatically establish consistent models, regarding the parameters’ definition, physical meaning, conservation, and units? The use of techniques to solve fractional models analytically or numerically is an interesting field in itself. However, it is important to try to verify how, where, and why the fractional derivatives interfere with the model. In this context, a fractional model can be obtained through time-since-infection dependent Mittag-Leffler based laws in the infectiousness and removal functions. Thus, in [19], we follow the footsteps of Angstmann, Henry & McGann [1], where they use the probabilistic language of Continuous Time Random Walks (CTRW). The Riemann-Liouville derivative appears throughout the construction, and following up results and applications are discussed in [17, 20, 22, 18]. Acknowledgements To the Academic Master’s program in Mathematics and to the Doctorate program in Computational Modeling - Federal University of Juiz de Fora, Brazil. Partially funded by CAPES, Coordination for the Improvement of Higher Education Personnel - Financing code 001, and Federal University of Juiz de Fora. Declarations Conflict of interest The authors declare that they have no conflict of interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Angstmann CN Henry BI McGann AV A fractional-order infectivity and recovery SIR model Fractal and Fractional 2017 1 1 11 10.3390/fractalfract1010011 2. Angstmann CN Henry BI McGann AV A fractional order recovery SIR model from a stochastic process Bulletin of Mathematical Biology 2016 78 3 468 499 10.1007/s11538-016-0151-7 26940822 3. Bilinsky LM Baer SM Slow passage through a Hopf bifurcation in excitable nerve cables: Spatial delays and spatial memory effects Bulletin of Mathematical Biology 2018 80 1 130 150 10.1007/s11538-017-0366-2 29150760 4. Bhunu CP Mushayabasa S Tchuenche JM A theoretical assessment of the effects of smoking on the transmission dynamics of tuberculosis Bulletin of Mathematical Biology 2011 73 6 1333 1357 10.1007/s11538-010-9568-6 20725798 5. Capelas de Oliveira, E.: Solved Exercises in Fractional Calculus. Springer, Switzerland (2019) 6. Diethelm K Monotonicity of functions and sign changes of their Caputo derivatives Fract. Calc. Appl. Anal. 2016 19 2 561 566 10.1515/fca-2016-0029 7. Dubbeldam JLA Tomovski Z Sandev T Space-time fractional Schrödinger equation with composite time fractional derivative Fract. Calc. Appl. Anal. 2015 18 5 1179 1200 10.1515/fca-2015-0068 8. Gallegos J Duarte-Mermoud M On the Lyapunov theory for fractional order systems Applied Mathematics and Computation 2016 277–278 149 160 9. Gallegos J Duarte-Mermoud M Manuel A Aguila-Camacho N Castro-Linares R On fractional extensions of Barbalat Lemma Systems & Control Letters 2015 84 7 12 10.1016/j.sysconle.2015.07.004 10. Ghosh S Bhattacharyya S Bhattacharya DK The role of viral infection in pest control: a mathematical study Bulletin of Mathematical Biology 2007 69 8 2649 2691 10.1007/s11538-007-9235-8 17661151 11. Huo J Zhao H Zhu L The effect of vaccines on backward bifurcation in a fractional order HIV model Nonlinear Analysis: Real World Applications 2015 26 6 289 305 10.1016/j.nonrwa.2015.05.014 12. Jahanshahi, H., Yousefpour, A., Munoz-Pacheco, J. M., Kacar, S., Pham, V-T., Alsaadi, Fawaz E.: A new fractional-order hyperchaotic memristor oscillator: Dynamic analysis, robust adaptive synchronization, and its application to voice encryption. Applied Mathematics and Computation 383(C), 125310 (2020) 13. Karaman, B.: The global stability investigation of the mathematical design of a fractional-order HBV infection. Journal of Applied Mathematics and Computing, 1–17 (2022) 14. Khan T Ahmad S Ullah R Bonyah E Ansari KJ The asymptotic analysis of novel coronavirus disease via fractional-order epidemiological model AIP Advances 2022 12 3 035349 10.1063/5.0087253 15. Mazorche SR Monteiro NZ Modelos epidemiológicos fracionários: o que se perde, o que se ganha, o que se transforma? Proceeding Series of the Brazilian Society of Computational and Applied Mathematics 2021 8 1 010448 1 16. Monteiro, N. Z.: Aplicação do Cálculo de Ordem Arbitrária à Epidemiologia. Master’s Thesis. Universidade Federal de Juiz de Fora (2021). Supervised by Mazorche, S. R 17. Monteiro, N. Z., Mazorche, S. R.: Analysis and application of a fractional SIR model constructed with Mittag-Leffler distribution. In: Proceedings of the XLII Ibero-Latin-American Congress on Computational Methods in Engineering, CILAMCE, Online (2021) 18. Monteiro, N. Z., Mazorche, S. R.: Application of a fractional SIR model built with Mittag-Leffler distribution. In: Mathematical Congress of the Americas MCA, Online (2021) 19. Monteiro NZ Mazorche SR Fractional Derivatives Applied to Epidemiology Trends in Computational and Applied Mathematics 2021 22 2 157 177 10.5540/tcam.2021.022.02.00157 20. Monteiro, N. Z., Mazorche, S. R.: Numerical study of the parameters in an arbitrary-order SIR model built with Mittag-Leffler distribution. C.Q.D. Revista Eletrônica Paulista de Matemática 22(2), 230–245 (2022) 21. Monteiro, N. Z., Mazorche, S. R.: O Cálculo Fracionário e alguns de seus comportamentos incomuns. In: I Reunião Mineira de Matemática - RMM, Online (2021) 22. Monteiro NZ Mazorche SR Some remarks on an arbitrary-order SIR model constructed with Mittag-Leffler distribution Matemática Contemporânea 2022 51 25 42 10.21711/231766362022/rmc512 23. Podlubny I Fractional Differential Equations 1998 San Diego Elsevier 24. Povstenko Y Kyrylych T Time-fractional diffusion with mass absorption under harmonic impact Fract. Calc. Appl. Anal. 2018 21 1 118 133 10.1515/fca-2018-0008 25. Slotine J-JE Li W Applied Nonlinear Control 1999 New Jersey Prentice Hall Englewood Cliffs 26. Ugarte, J. P., Machado, J. A. T., Tobón, C.: Fractional generalization of entropy improves the characterization of rotors in simulated atrial fibrillation. Applied Mathematics and Computation 425(C), 127077 (2022) 27. Valério D Machado JT Kiryakova V Some pioneers of the applications of Fractional Calculus Fract. Calc. Appl. Anal. 2014 17 2 552 578 10.2478/s13540-014-0185-1 28. Wang F Yang Y Fractional order Barbalat’s lemma and its applications in the stability of fractional order nonlinear systems Mathematical Modelling and Analysis 2017 22 4 503 513 10.3846/13926292.2017.1329755 29. Widyaningsih, P., Nugroho, A. A., Saputro, D R. S.: Susceptible infected recovered model with vaccination, immunity loss, and relapse to study tuberculosis transmission in Indonesia. In: AIP Conference Proceedings, 020121, AIP Publishing LLC (2018) 30. Yaghooti B Siahi Shadbad A Safavi K Salarieh H Adaptive synchronization of uncertain fractional-order chaotic systems using sliding mode control techniques Proceedings of the Institution of Mechanical Engineers, Journal of Systems and Control Engineering 2020 234 1 3 9 31. Zhang, R., Liu, Y.: A new Barbalat’s lemma and Lyapunov stability theorem for fractional order systems. In: 2017 29th Chinese control and decision conference (CCDC), 3676–3681, Chongqing (2017)	36506647	PMC9718479	NO-CC CODE	2022-12-06 23:23:40	no	Fract Calc Appl Anal. 2022 Dec 2;:1-23	utf-8	Fract Calc Appl Anal	2,022	10.1007/s13540-022-00111-6	oa_other
==== Front J Bus Res J Bus Res Journal of Business Research 0148-2963 0148-2963 Elsevier Inc. S0148-2963(22)00945-6 10.1016/j.jbusres.2022.113480 113480 Article Vaccine supply chain management: An intelligent system utilizing blockchain, IoT and machine learning Hu Hui ab Xu Jiajun b Liu Mengqi c⁎ Lim Ming K. d a Economic Development Research Centre, Wuhan University, China b School of Economics and Management, Wuhan University, China c Business School, Hunan University, China d Adam Smith Business School, University of Glasgow, UK ⁎ Corresponding author. 2 12 2022 2 2023 2 12 2022 156 113480113480 31 3 2022 15 11 2022 18 11 2022 © 2022 Elsevier Inc. All rights reserved. 2022 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Vaccination offers health, economic, and social benefits. However, three major issues—vaccine quality, demand forecasting, and trust among stakeholders—persist in the vaccine supply chain (VSC), leading to inefficiencies. The COVID-19 pandemic has exacerbated weaknesses in the VSC, while presenting opportunities to apply digital technologies to manage it. For the first time, this study establishes an intelligent VSC management system that provides decision support for VSC management during the COVID-19 pandemic. The system combines blockchain, internet of things (IoT), and machine learning that effectively address the three issues in the VSC. The transparency of blockchain ensures trust among stakeholders. The real-time monitoring of vaccine status by the IoT ensures vaccine quality. Machine learning predicts vaccine demand and conducts sentiment analysis on vaccine reviews to help companies improve vaccine quality. The present study also reveals the implications for the management of supply chains, businesses, and government. Keywords Vaccine supply chain Blockchain Internet of things Machine learning COVID-19 pandemic Intelligent system Abbreviations BILSTM, Bidirectional Long-Short Term Memory CNN, Convolutional Neural Network dApp, Decentralized Application DTs, Digital Technologies GRU, Gate Recurrent Unit IoT, Internet of Things IPFS, Interplanetary File System LSTM, Long-Short Term Memory RFID, Radio Frequency Identification RNN, Recurrent Neural Network VSC, Vaccine Supply Chain ==== Body pmc1 Introduction Vaccination is the most economical and effective public health intervention to control infectious diseases (Adida et al., 2013). It also contributes to ending poverty and reducing inequality and is an essential means of achieving the United Nations Sustainable Development Goals (Decouttere et al., 2021). However, global vaccination coverage remains low due to inefficiencies in the vaccine supply chain (VSC) (Chandra and Kumar, 2018, Zaffran et al., 2013). Successful immunization programs are based on sustainable VSCs and logistics systems (Chandra & Kumar, 2018). One main objective of immunization programs is to optimize the VSC so that vaccines can be delivered safely and efficiently to the recipients. The VSC encompasses personnel, systems, equipment, and activities at all stages, including production, transportation, allocation, and distribution (Duijzer et al., 2018). At each node and along each channel, potential risks affect the efficiency of the entire VSC (Finkenstadt & Handfield, 2021). In sum, three key issues that need to be addressed are vaccine quality, demand forecasting, and trust among stakeholders (De Boeck et al., 2020, Duijzer et al., 2018). Vaccine quality refers to various aspects, such as authenticity, safety, potency, and activity (Preiss et al., 2016). Counterfeit, defective, expired, and substandard vaccines are widespread around the world. Large number of counterfeit vaccine incidents, documented by the International Institute of Research Against Counterfeit Medicines, occurred primarily in Africa and Asia (Jarrett et al., 2020) (see Table 1 ). These counterfeit vaccines threaten public health and there is an urgent need for a credible anti-counterfeiting and traceability system to ensure the authenticity and safety of vaccines. Moreover, vaccines are usually temperature-sensitive and must be controlled within a certain range to maintain their potency and activity (Lin et al., 2020). However, vaccines are often exposed to inappropriate temperatures during transport (Hanson et al., 2017). Intelligent cold chain management is vital to ensure vaccine quality (Kartoglu & Milstien, 2014).Table 1 Recorded events of counterfeit vaccines. Year Vaccine Country/region where counterfeit vaccines were identified 2020 COVID-19 Russia 2020 COVID-19 Ecuador 2019 Rabies Philippines 2019 Human papillomavirus (HPV) Hong Kong 2019 Meningitis Cameroon 2019 Cholera Bangladesh 2019 Meningitis Niger 2018 Chickenpox Venezuela 2018 Hepatitis B Uganda 2017 Diphtheria-tetanus-pertussis (DTP) triple vaccine China 2017 Meningitis Nigeria 2016 Rabies and meningitis China 2016 Polio and hepatitis B Indonesia 2016 Yellow fever Angola Source: Jarrett et al. (2020). Table 2 Text classification accuracy of four deep learning models. Rating Static-CNN Trainable-CNN Static-BILSTM Trainable-BILSTM Support 1 0.7378 0.7437 0.7828 0.7878 7299 2 0.4386 0.5972 0.569 0.6239 2334 3 0.4236 0.5828 0.5403 0.6058 2205 4 0.3926 0.5769 0.5065 0.6001 1659 5 0.4341 0.5814 0.5442 0.5953 2710 6 0.3644 0.5621 0.4943 0.5798 2119 7 0.4107 0.5691 0.5178 0.5851 3091 8 0.4719 0.5795 0.5644 0.5951 6156 9 0.5394 0.613 0.6108 0.6362 9177 10 0.7565 0.7668 0.7898 0.7895 17,016 Macro avg. 0.4969 0.6173 0.592 0.6399 53,766 Weighted avg. 0.594 0.6664 0.6643 0.6905 53,766 High uncertainty of supply and demand is a characteristic of the VSC (Duijzer et al., 2018). When the information flow is transmitted from the client to the supplier, it cannot be shared effectively. This magnifies the information distortion and results in increasing fluctuations in demand information. This phenomenon is prevalent and is known as the “bullwhip effect” (Remko, 2020). Vaccine demand forecasts are based on target populations, estimated periods, and historical vaccination records. Nevertheless, the lack of reliable vaccination data and analysis methods results in the inappropriate analysis of vaccine utilization trends and inaccurate demand forecasts (Kaufmann et al., 2011). Production and procurement strategies based on inaccurate demand lead to loss of profits and additional costs (Chick et al., 2017). Trust issues among stakeholders are prevalent in the health- and safety-related supply chains (Betcheva et al., 2021, Hobbs, 2020). In the VSC, conflicting goals and fragmented decision making among suppliers, organizations, and customers lead to a lack of trust among actors (Duijzer et al., 2018). Vaccine falsifications further deepen customers’ distrust of suppliers (Jarrett et al., 2020). The public’s lack of knowledge about immunization in some developing countries leads to distrust of immunization programs (Bangura et al., 2020) and decreases global immunization coverage (Weintraub et al., 2021). The long-lasting COVID-19 pandemic has exacerbated the weaknesses of the global supply chain (Edwards et al., 2021). The lockdowns caused by COVID-19 have limited supply chain processes (Guan et al., 2020, Karmaker et al., 2021). COVID-19 has disrupted routine immunization programs, impeded vaccines shipments, and increased demand uncertainty (Bloom et al., 2021). However, COVID-19 offers opportunities for VSC’s sustainable development. On the one hand, growing public concern about health highlights the importance of vaccination (Dai and Song, 2021). On the other hand, digital technologies (DTs) have a greater appreciation for addressing supply chain management issues (Sarkis, 2020). DTs have been used in the health care ecosystem for over a decade (Secundo et al., 2021). But their market penetration is slight due to strict regulation and poor supporting payment structures (Keesara et al., 2020). The COVID-19 pandemic offers a wide range of applications for DTs. Blockchain, the internet of things (IoT), artificial intelligence, machine learning, big data analysis, geospatial technology, and virtual and augmented reality have been widely used for COVID-19 detection, monitoring, diagnosing, screening, surveillance, and tracking (Chamola et al., 2020, Lee and Trimi, 2021, Mbunge et al., 2021). Digital transformation of the health care supply chain has become inevitable (Kraus et al., 2021). In VSC management, the transparency, traceability, and immutability of blockchain can address vaccine safety issues and improve trust and coordination among stakeholders (Adarsh et al., 2021, Badhotiya et al., 2021, Cui et al., 2021, Liu et al., 2021). The literature discusses the use of the IoT in vaccine temperature detection (Monteleone et al., 2017), the application of artificial intelligence in vaccine development (Arora et al., 2021), and the operational value of using drones to transport vaccines (Haidari et al., 2016). These studies explore the potential application of DTs in the VSC, but they focus on the application scenario of a DT in a specific supply chain link (Liu et al., 2021). They neglect the management and operation of the VSC as a whole and the necessity for innovative integration among DTs. For example, while blockchain provides a platform for vaccine traceability and real-time vaccine consumption data, IoT sensors are still needed to monitor vaccine temperatures to ensure vaccine potency and data accuracy (Hasan et al., 2019, Singh et al., 2020). Machine learning is needed for data mining and analysis to generate practical applications (Ivanov & Dolgui, 2021). This study aims to build an intelligent VSC management system from a holistic view, combining blockchain, IoT, and machine learning. The system efficiently manages the VSC affected by the COVID-19 pandemic and prioritizes three key issues: vaccine quality, demand forecasting, and stakeholder trust. This is an original work, as there is no integrated VSC management system with multiple DTs in the literature. Especially in the context of the COVID-19 pandemic, our designed intelligent VSC management system is a valuable contribution because it balances the intelligent management of both the short-term VSC and long-term immunization programs. This work has important application implications, as it provides effective digital solutions to practical problems in the VSC. First, the combination of blockchain and the IoT addresses issues of vaccine quality and stakeholder trust and provides vast amounts of authentic real-time data. Second, using the Gated Recurrent Unit (GRU) machine learning model to forecast vaccine demand enables the average annual prediction error of influenza vaccine to stay within 3 %. Third, the trainable bidirectional long short-term memory (BILSTM) deep learning model provides sentiment analysis of vaccine reviews with nearly 80 % accuracy, providing strong support for enterprise credibility evaluation and consumer decision making. Finally, our proposed intelligent VSC management system is scalable and can enlighten other supply chains in terms of supply and demand management, security, and supervision. The rest of this study is arranged as follows. Section 2 discusses the impact of the COVID-19 pandemic on the VSC and DTs applications. Section 3 presents the characteristics of blockchain, the IoT, and machine learning and their applications. Section 4 builds an intelligent VSC management system based on these three DTs. Section 5 analyzes data from blockchain platforms using machine learning. The study concludes with Section 6. 2 Impact of the COVID-19 pandemic on the health care and vaccine supply chains This section discusses the risks posed by the COVID-19 pandemic to the health care supply chain and VSC, as well as the application of DTs. 2.1 COVID-19 pandemic risks to the health care and vaccine supply chains Supply chain risks can be divided into operational and disruption risks (Fahimnia et al., 2018, Xu et al., 2020). Operational risk is mainly related to common disturbances in supply chain operation, such as demand fluctuations and early delivery times. Disruption risk mainly refers to events with low frequency and high impact (Kinra et al., 2020, Sreedevi and Saranga, 2017). The COVID-19 pandemic is a special case of supply chain disruption risk due to its long duration, high uncertainty, and chain reaction spread (El Baz and Ruel, 2021, Ivanov and Dolgui, 2021). The illnesses, deaths, and trade and travel restrictions caused by the COVID-19 pandemic have led to disruptions in supply chain networks (Nagurney, 2021). The efficacy of a vaccine depends on its availability to the public. This means that the supply chain supporting vaccine production, transportation, allocation, and distribution must be resilient to quickly resume operations and continue to provide vaccines to the public after a disruption (Golan et al., 2021). Therefore, the problems in the VSC affected by COVID-19 must be properly addressed. The protracted COVID-19 pandemic has overtaxed health systems, with significant resources being devoted to responding to the surge of COVID-19 patients, thereby disrupting other preventive vaccination campaigns (Bloom et al., 2021, Zeitouny et al., 2021). The COVID-19 pandemic and associated disruptions resulted in 22.7 million children missing vaccinations, an increase of 3.7 million from 2019. In addition, movement restrictions to prevent COVID-19 infection have impeded international shipments of vaccines and exacerbated uncertainties in vaccine supply and demand. Moreover, as the COVID-19 vaccines continue to be developed, the risk of vaccine counterfeiting has become acute (Jarrett et al., 2020). Although COVID-19 vaccine research has focused on developing the vaccine and measuring its effectiveness (Georgiadis & Georgiadis, 2021), the most critical response to the outbreak is not the vaccine itself, but vaccination (Dai & Song, 2021). Therefore, no link in the COVID-19 VSC can be ignored. Alam et al. (2021) identify 15 challenges in the COVID-19 VSC, among which the most critical challenges are the limited number of vaccine manufacturing companies, lack of vaccine monitoring institutions, difficulty in monitoring and controlling vaccine temperature, difficulty in organizing and coordinating vaccine distribution, and the high cost of vaccination. The supply chain of the COVID-19 vaccine, as a specific species of vaccine, also faces the same three key issues that we summarized in the introduction section: vaccine quality, uncertainty of vaccine supply and demand, and coordination among stakeholders. 2.2 COVID-19 pandemic brings opportunities for digital technology applications Four themes recur in research related to the COVID-19 pandemic: the impact of the pandemic, supply chain resilience, supply chain sustainability, and DTs (Chowdhury et al., 2021, Sharma et al., 2020). Digitization has been the key deliverable from COVID-19 (Nandi et al., 2021). On the one hand, public health activities in response to the COVID-19 pandemic—such as mass population surveillance and screening, case identification, contact tracing, and evaluation of interventions based on mobility data and communication with the public—have facilitated the widespread adoption of DTs (Budd et al., 2020, Whitelaw et al., 2020). On the other hand, the lockdown and isolation measures in response to the COVID-19 pandemic forced people to keep their distance, pushing many behaviors (e.g., shopping, learning, working, meeting, and entertainment) to move from offline to online. New technology services have emerged to meet the needs of customers and businesses shopping and operating in the digital environment (Dwivedi et al., 2020, Pandey and Pal, 2020). All-round and multi-field digital transformation became inevitable (Kraus et al., 2021). Vargo et al. (2021) summarize the use of DTs during the COVID-19 pandemic in four dimensions: technology, population, activity, and impact. They find that 28 different forms of technology, ranging from computers to artificial intelligence, have been used by eight groups of users, primarily health care professionals. Of these, 32 activities are involved, including remote delivery of health services, analysis of data, and communication. Among the many DTs, some key disruptive technologies are highlighted, including artificial intelligence, the IoT, big data, and blockchain (Abdel-Basset et al., 2021). These technologies provide digital transformation and research and development opportunities that not only help mitigate the negative impacts of COVID-19 but also help manage supply chains affected by COVID-19 (Chamola et al., 2020). The role of DTs in responding to the COVID-19 pandemic has been extensively studied, but research related to the long-term digital management of the health care supply chain is insufficient. The application potential of blockchain technology in the health care supply chain has been explored to some extent. Ng et al. (2021) summarize the health care applications of blockchain, including artificial intelligence model development, medical supply chain encryption, and mobile health care, among others. Further, studies have shown that blockchain technology can be used in the VSC to address issues such as vaccine quality and safety, pricing and coordination, distribution and delivery, and improved trust among stakeholders (Adarsh et al., 2021, Antal et al., 2021; Badhotiya et al., 2021; Cui et al., 2021; Liu et al., 2021; Musamih et al., 2021, Yong et al., 2020). Yet, the potential of other DTs in the health care supply chain and VSC has not been explored in depth. Ting et al. (2020) propose that DTs are highly interconnected, especially among the IoT, big data analysis, artificial intelligence, and blockchain. Specifically, the proliferation of IoT devices and instruments in hospitals enables the collection of real-time data at scale, which can then be used by artificial intelligence and deep learning systems to understand health care trends, model risk associations, and predict outcomes. Blockchain technology guarantees the security and traceability of the data used for modeling and prediction. Therefore, there is a need to integrate multiple DTs jointly to solve supply chain management problems. Since blockchain is a fundamental platform for integrating multiple DTs (Ng et al., 2021), it is feasible to construct an intelligent VSC management system based on blockchain technology and complemented by other DTs. 3 Digital technologies in supply chains Traditional supply chains face many challenges, such as uncertainty, high costs, complexity, and vulnerability. To overcome these issues, supply chains must become more intelligent (Abdel-Basset et al., 2018). The COVID-19 pandemic further diminished supply chain efficiency, making it more urgent for organizations to implement data-driven technologies as part of their supply chain strategies. DTs have a significant impact on the nature and structure of supply chains, enabling internal integration of processes and, more importantly, external integration with suppliers and customers. This is achieved through improved communication and acquisition and transmission of data to enable effective decision making and improve supply chain performance. DTs have been an essential driver of effective supply chain management (Ross et al., 2010). They are vital in helping supply chains respond to changing environments and risks at all levels. In the following, we specify several important DTs and their application in supply chains. 3.1 Blockchain: Improving the security, visibility, and resilience of supply chains Blockchain is a cryptographically secure distributed ledger that has the advantages of decentralization, information immutability, and transparency. The purpose of using blockchain is to solve the problem of overreliance on third parties in electronic payments. During the COVID-19 pandemic, many blockchain-related studies have highlighted its benefits for the supply chain management. First, the decentralized nature of blockchain increases supply chain resilience by mitigating the risks associated with intermediary intervention (Min, 2019). The reduced costs of labor, operation, and maintenance reduce the cost of supply chain transactions and improve management effectiveness (Catalini and Gans, 2020, Cole et al., 2019). Second, the immutable blockchain information combined with smart contracts can be used to coordinate supply chain members and enhance trust among stakeholders (Casado-Vara et al., 2018, Saberi et al., 2019). Third, the transparency of blockchain enhances the visibility of the supply chain, thus strengthening product security (Abeyratne and Monfared, 2016, Francisco and Swanson, 2018, Korpela et al., 2017). In the wake of the COVID-19 outbreak, a large body of literature has emerged on the use of blockchain to solve health care supply chain problems. Marbouh et al. (2020) develop a blockchain-based tracking system for COVID-19 data collected from a variety of external sources. Yaqoob et al. (2021) apply blockchain to data management in health care. Omar et al. (2021) use blockchain smart contracts to automate procurement contracts in health care supply chains. In the field of VSC, blockchain has been explored for every link. Alkhoori et al. (2021) introduce a blockchain-based smart container system (CryptoCargo) for safe and efficient distribution of vaccines. Adarsh et al. (2021) propose a traceable VSC (Immunochain) based on blockchain technology for immunization programs in India. Liu et al. (2021) analyze the pricing and coordination of the VSC based on blockchain technology and find that the introduction of blockchain technology increased the total profit, consumer surplus, and social welfare of the VSC. Musamih et al. (2021) propose an Ethereum blockchain-based solution for managing data related to COVID-19 vaccine distribution and delivery. Moreover, blockchain applications in supply chains often work together with other cutting-edge technologies (Wamba & Queiroz, 2020). For example, blockchain can be combined with the IoT (De Villiers et al., 2021, Rejeb et al., 2019), big data analytics, and machine learning (Dwivedi et al., 2021). 3.2 Internet of things: Real-time monitoring and remote operation management The IoT is an intuitive and scalable technology that conveys all relevant information in real-time across the supply chain via the internet (Manavalan and Jayakrishna, 2019). With the complexity of the network and more and more stakeholders, the end-to-end visibility of the supply chain becomes more critical (Pundir et al., 2019). However, technology adoption is still slower than current needs, and many supply chains still rely on manual processes for data recording, collection, and access. This approach is both complex and lagging and fails to handle the large-scale data effectively from the COVID-19 pandemic or identify and resolve supply chain issues promptly. Applying IoT technologies to collect data and turn it into useful information provides visibility to supply chain participants. A typical example is the use of Radio Frequency Identification (RFID) technology to achieve fine-grained product tracking and tracing while reducing implementation costs (Li et al., 2017, Ustundag and Tanyas, 2009). Moreover, the IoT significantly reduces data acquisition time and enables the supply chain to respond to changes in real-time, thus increasing the efficiency of decision making (Ellis et al., 2015). For example, temperature-sensitive products can be monitored by IoT sensors to prevent product spoilage or expiration (Tajima, 2007). While saving labor costs, IoT technologies enable supply chain stakeholders to monitor the condition of products in real-time for timely decisions and coordination (Accorsi et al., 2017, Umair et al., 2021). The combination of the IoT and other DTs is expected to enhance the role of the IoT. Research has concluded that the combination of machine learning technology and the IoT is promising because IoT devices automatically generate large amounts of data, providing a good basis for machine learning. In addition, the data provided by the IoT is safe and reliable. Combined with blockchain, it further promotes automation and solves the problem of blockchain data source forgery (Zelbst et al., 2019). Conversely, blockchain manages the accuracy of IoT devices and avoids malicious behavior through authentication (Chen et al., 2020, Wang et al., 2018); and the decentralized nature of blockchain is used to enhance the privacy and security of IoT systems (Banerjee et al., 2018, Kouzinopoulos et al., 2018). 3.3 Machine learning: Making decisions for supply chain uncertainties Machine learning automatically improves computer algorithms through data or previous experience and provides more accurate decision results than humans in many areas (Ni et al., 2020). Making decisions based on uncertainty is an important problem in supply chains (Steckel et al., 2004), and machine learning solves this problem to some extent. Machine learning has performed well in production (Chen et al., 2012), demand prediction (Carbonneau et al., 2008), transportation and distribution optimization (Ćirović et al., 2014), inventory management improvement (Gumus et al., 2010), and supply chain risk prediction (Baryannis et al., 2019). In recent years, several studies have explored the application of machine learning in the health care supply chain. Chatterjee et al. (2021) explore customer satisfaction in e-commerce for health care products, using text mining and machine learning techniques. Piccialli et al. (2021) predict medical appointments using machine learning models and hybrid neural networks. The application of machine learning in supply chain management is still in the development stage (Ni et al., 2020), and its potential value needs to be further developed. However, using machine learning methods usually requires a large amount of data, and it is a significant challenge to collect the data efficiently and ensure its authenticity. Some studies suggest that this problem can be overcome by combining machine learning with other DTs, such as blockchain (Section 3.1) and the IoT (Section 3.2). 4 Intelligent vaccine supply chain management system 4.1 Overview of the system This study designs a blockchain-based intelligent VSC management system combined with the IoT and machine learning (Fig. 1 ), which can track the whole vaccination process from manufacturers to end-users and provide decision support for VSC management. Trusted participants are vetted by the government to be eligible to join the system. Participants connect to the blockchain through a decentralized application (dApp). For vaccines to be delivered and traded in the supply chain, participants need to use this dApp to upload information about vaccines and transactions. The participants initiate vaccine transfer in the system, and the system verifies the data for compliance with the criteria through predefined smart contracts. After successful verification, the participants’ records and transaction information are packaged and loaded into the system.Fig. 1 The framework of an intelligent vaccine supply chain management system based on digital technologies. Among them, the blockchain mainly stores simple information and signatures of these transactions, while detailed vaccine information is stored in the InterPlanetary File System (IPFS). The decentralized and immutable nature of blockchain and IPFS ensures the authenticity of transaction records in the VSC. The IoT devices deployed during the transportation process monitor and track the vaccines and collect real-time monitoring data into the blockchain. In case of vaccine failures, such as expiration or high temperature, all relevant participants can obtain information on the abnormal status of the vaccine through the system. If there is a quality problem after a vaccine injection, accountability can be traced through the blockchain system to find the party responsible for the problem quickly. In addition, the large amount of reliable and real-time data on the blockchain can be used by research institutions or government agencies, for example, for big data analysis, combined with machine learning techniques to contribute to the improvement of supply chain management performance. More details of the system are provided in the following subsections. 4.2 Participants in the system The key participants in the VSC include manufacturers, quality inspection agencies, distributors, vaccination centers, and patients. The government is included in the intelligent VSC management system. Participants must apply for an account (both public and private keys) with the government before they can join the vaccine supply chain, which the government releases only if the participant meets the required conditions. Vital records for vaccines with private key signatures also need to be uploaded when participants engage in transactions. The signatures are verified by the blockchain system to ensure that the records are correctly submitted by the respective institutions. These records are key to ensuring the quality of vaccines at every point in the supply chain. Among them, manufacturers provide vaccine production information; quality inspection agencies provide vaccine inspection reports; logistics companies provide real-time monitoring data on temperature, light level, and transport routes during vaccine transportation; and vaccination centers provide vaccination time and doctor information. The information provided may be falsified, so the government must regularly check the uploaded records. If there are any problems with the records, the corresponding agencies will be investigated and held accountable. Records and violations stored on and off the chain are accessible to all participants, promoting trust among participants in the VSC. 4.3 Basic technology of the system: Blockchain and smart contracts Smart contracts are one of the appealing features of blockchain technology. They are a set of commitments defined in digital form to provide a transparent, free channel for all parties, allowing trusted, traceable, and irreversible transactions without third parties. Smart contracts are run exactly as programmed, without any possibility of downtime, fraud, or third-party interference. They are automatically triggered when the specified conditions are the same as the database events. Typically, smart contracts are stored in the distributed ledger of the blockchain platform and are fully protected from deletion and tampering. By designing the right smart contracts, our system intelligently monitors the state of the system and provides secure data. Here we describe some of the key smart contracts in the system. First and foremost, once a participant initiates a vaccine transfer request, the system reviews the data submitted for compliance with the criteria through the predefined smart contracts. For vaccine manufacturers to transfer vaccines to logistics companies, the records must meet good manufacturing practices before the transfer is allowed. For the transfer of vaccines from logistics companies to vaccination centers, the records must meet the conditions of vaccine validity. If the vaccines have expired or failed in transit, by presetting smart contracts, information about vaccine failure will be disseminated throughout the VSC, preventing people from receiving ineffective vaccines. In addition, predefined smart contracts enable the system to provide intelligent oversight of the VSC. In the vaccine accident, blockchain will be used to track the entire process of vaccine circulation to investigate and pursue accountability. Smart contracts can also be used to gather additional data for big data analysis by scientific institutions. For example, smart contracts incentivize users to provide vaccine reviews by giving them coins on the blockchain, and then text sentiment analysis using machine learning technology evaluates the credibility of enterprises. Alternatively, a marketplace for buying and selling personal vaccination data can be formed through smart contracts, where vaccination users choose to sell their age, medical history, and other data. Machine learning technology can analyze the data to enable vaccine development and personalized vaccination recommendations. Thus, combining smart contracts and machine learning provides big data analysis and makes management recommendations for the VSC. 4.4 Assurance technologies for the system: IoT and IPFS Blockchain databases are trusted in untrusted environments (Tsolakis et al., 2021). The uploaded data in blockchain have the advantage of being untamable, but it is difficult to solve the problem of forgery in data collection. Automation of the IoT ensures data accuracy as it eliminates human errors and intentional fraud, which lead to wrong information. In addition, the volume of data in the blockchain is growing due to the feature of not being able to delete but only to add. Blockchain needs to be supported by new storage technologies to achieve scaling. Therefore, the DTs supporting the blockchain-based intelligent VSC management system designed in this study also include the IoT and IPFS. 4.4.1 Internet of things: Efficient access to real information Our system combines the advantages of the IoT and blockchain: the former solves the problem of how to get real information efficiently and the latter solves the problem of trust with its immutability and information transparency. RFID is a non-contact data communication technology that automatically identifies tags and input information (Angeles, 2005). RFID technology has a high identification accuracy even in harsh working environments (Mondal et al., 2019). Our system assigns a unique RFID tag to each vaccine as it is packaged by the manufacturer to reduce the labor cost of entering vaccine information. It also prevents human error in data collection and ensures the authenticity and accuracy of the data on the chain, while the blockchain further ensures that there can be no tampering with the data on the chain. The combination of blockchain and RFID provides full vaccine traceability and greatly enhances vaccine safety. In addition, IoT sensing devices are used to monitor the status of vaccines during transportation (Hasan et al., 2019). Vaccines are thermosensitive and photosensitive, and there is a risk of vaccine expiration or failure during transport, which has been exacerbated during the COVID-19 pandemic (Alam et al., 2021). Real-time monitoring of vaccine temperature during transport can prevent vaccine failure. The vaccines are placed in smart containers with IoT sensors, which automatically digitize the real-time monitored temperature and light status and send violation information to the blockchain system through smart contracts. The intelligent VSC management system greatly improves the transparency of the supply chain and enhances the credibility of information through the combination of the IoT and blockchain technology, enabling the identification and tracking of information throughout the VSC. 4.4.2 InterPlanetary File System: Avoiding blockchain information explosion IoT sensors provide real-time and continuous monitoring data for the blockchain, but the accompanying huge data storage and the increasing number of participants on the blockchain may lead to an information explosion (Wang et al., 2021). Our system uses the IPFS to solve this problem. The IPFS is a decentralized storage network with fast file storage and downloading, and the distributed storage feature ensures that the data inside are not arbitrarily tampered with or deleted (Zheng et al., 2018). A block in a traditional blockchain consists of a header and a body. The header is mainly used to verify the correctness of transactions and stores a smaller volume of data. The body stores specific transaction data and other larger data, which occupy most of the space in the blockchain. Our system keeps the body in IPFS, a design that greatly reduces the size of the blockchain and thus avoids blockchain information explosion. 4.5 Decision support technology for the system: Machine learning Blockchain and IoT enable access to real information, while the use of machine learning models allows analysis of the data and provision of useful recommendations to participants based on the results. Reliability and sharing of data are important in machine learning to improve the accuracy of the results. Blockchain motivates and collects real data and machine learning processes the data to make accurate decisions. The combination of the two provides highly accurate results (Tanwar et al., 2019). The intelligent VSC management system of this study combines blockchain and machine learning to form a system in which the two constantly promote each other. In the next section, we specify how machine learning techniques are used in this intelligent system to provide decision support for participants and thus improve the management performance of the VSC. 5 Decision support analysis of the intelligent vaccine supply chain management system In this section, we use machine learning to analyze two intelligent VSC management systems’ cases—vaccine demand forecasting and vaccine review sentiment analysis. Thereby, machine learning provides decision support to solve two important issues of VSC (vaccine demand forecasting and vaccine quality management). It is worth noting that the data used in the two cases are not collected by the proposed intelligent VSC management system. However, we discuss the advantages of the data collected by the proposed system compared to traditional data. 5.1 Vaccine demand forecasting The imbalance between vaccine supply and demand is a prominent issue in the VSC, especially in the context of the COVID-19 pandemic. Incorrect estimation of vaccine demand is also a major cause of vaccine expiration. Supply exceeding demand results in wasteful use of resources, while greater demand than supply leads to vaccine nationalism and the risk of price fraud (Eccleston-Turner and Upton, 2021, Nhamo et al., 2021). Therefore, accurate forecasting of vaccine demand is critical for vaccine production. In our system, vaccination centers provide real-time vaccine usage quantity data to the blockchain, which is vaccine demand data. Using machine learning to train the data and time-series models to predict demand and contribute to less vaccine expiration. Among the many machine learning models, RNN, LSTM, and GRU models are often used. In the following, we compare the performance of the three models for vaccine demand forecasting with a specific data set. We use the U.S. influenza vaccination volume (million doses) from 1980 to 2020 as the vaccine demand forecasting dataset, with the data from 1980 to 2011 as the training set and the data from 2012 to 2020 as the testing set. We designed RNN, LSTM and GRU models to predict the demand for influenza vaccines, respectively. The vaccination doses predict the vaccine demand for each year from the previous seven years. The trends of predicted and actual amounts for the three models are shown in Fig. 2 .Fig. 2 Vaccine demand forecasting based on three machine learning models. In the figure, all three models, especially the GRU model, have almost identical trends in predicted and actual doses of the influenza vaccine. There is no overfitting between the predicted and actual amounts, suggesting that the prediction models are scalable. Further, we compare the errors of the three machine learning models calculated using different metrics (see Appendix Table A.1). The results show that GRU is more accurate than LSTM and RNN in vaccine demand prediction. If the GRU model is applied to predict the demand for influenza vaccine and keep the annual remaining vaccine ratio within about 3 % of the predicted value, a balance between vaccine supply and demand will be effectively achieved. According to Rappuoli et al. (2019), the global vaccine market will reach $62 billion by 2027, and the COVID-19 pandemic will increase this value significantly. Therefore, using appropriate machine learning models to predict vaccine demand will help the health care sector and vaccine enterprises save billions of dollars annually. Due to data limitations, the case study only provides annual vaccine demand forecasts for the United States. The proposed intelligent system will provide timely and accurate vaccine demand data down to the city and even hospital level. A fine-grained vaccine demand forecast would offer a critical reference for vaccine distribution within a country. The blockchain and IoT in the proposed system make the data more credible, which will further improve the demand forecasting accuracy. 5.2 Sentiment analysis of vaccine reviews Providing safe and effective vaccines to consumers has always been an essential concern for VSC management, and consumer satisfaction is the most direct response to vaccine quality. The literature has conducted consumer satisfaction surveys but has rarely examined online reviews (Chatterjee et al., 2021). Sentiment analysis using machine learning is good at understanding the attitudes of online reviews (Bag et al., 2019, Fan et al., 2017). In our proposed system, reviews from vaccine consumers can first be collected using the incentives of the blockchain. The vaccine reviews in the blockchain could then be mined and analyzed using a text sentiment analysis model to help vaccine companies better understand consumer attitudes and needs. On the one hand, this supports the system in recommending to consumers high-quality vaccines that meet individual needs. On the other hand, it promotes vaccine enterprises to compete to improve vaccine quality and address consumers’ concerns. Fig. 3 illustrates the process of vaccine review sentiment analysis.Fig. 3 Process for sentiment analysis of vaccine reviews. We use the data set proposed by Grasser et al. (2018) as a case study, which is a collection of 215,063 online drug reviews obtained from the Drugs.com website. Table A.2 in the Appendix shows vaccine-related examples of scores 1 to 10 in the data set, where reviews and ratings are used as features and labels for the machine learning model. To validate the generalizability of the machine learning model on the new data set, we divide the corpus into a training set and a testing set, at a ratio of 75:25. Fig. A.1 in the Appendix shows the ratings distribution in the training and testing sets. The distribution has a polarized character, where the number of reviews with ratings of 1, 8, 9, and 10 accounts for 74 % of the overall sample, indicating that consumers’ reviews are significantly emotional. Before inputting them into the machine learning model, we preprocess the data, which includes tokenization, removing punctuation and numbers, lemmatization, padding, and truncation. Then, we compare the Convolutional Neural Network (CNN) and BILSTM models to find a suitable model and show the results of the two models for the trainable and static data sets. F1-score is the most commonly used metric to measure the accuracy of text classification. Table 2 shows the F1-scores of the four models for each classification, where support represents the number of samples per class and rating represents the score. The macro average is a simple average of the F1-scores for each classification, and the weighted average is the average obtained by assigning the sample size of each class as the weight. As can be seen, the accuracy of the four models differed little in identifying reviews with apparent emotions. However, they show significant differences in identifying reviews with intermediate attitudes, where the trainable-BILSTM model has higher accuracy than the other models for almost all classifications (except for the rating of 10). This proves that the trainable-BILSTM model is the most suitable deep learning model for sentiment analysis of vaccine reviews. The number of consumer reviews is essential for improving the efficiency and accuracy of the model (Hwang et al., 2020). One of the advantages of the proposed system is encouraging consumers’ comments with the help of blockchain tokens payment system (Yong et al., 2020). Using the proposed system to collect vaccine reviews, we can continue training based on the trained model to improve the accuracy of text sentiment analysis. Consumers’ attitudes toward vaccines, including side effects and utility after vaccination, will be captured automatically. This will provide an essential reference for consumers and manufacturers to help vaccine quality continue to improve. 6 Discussion and conclusion To manage the COVID-19-affected VSC, we combined blockchain, the IoT, and machine learning to build an intelligent VSC management system. The research has important implications for both theory and practice. 6.1 Theoretical implications In the context of the continuing spread of COVID-19, relevant studies have focused on managing the COVID-19 VSC. This particular attention is reasonable but has led to a lack of research on the supply chain management of the many routine vaccines affected by COVID-19. We bridge this gap, as the intelligent VSC management system is applicable not only to COVID-19 VSC management but also to routine VSC management affected by COVID-19. Meanwhile, this study highlights the critical role of the combined application of DTs in solving the pressing problems faced by VSC. The respective benefits of blockchain, IoT, and machine learning applications in the supply chain are relatively well established in the literature. However, these DTs are usually applied separately in supply chain systems. In the proposed system, we further combine the three DTs by complementing each other to provide more significant benefits in the VSC than they could individually. IoT ensures efficient and accurate blockchain data collection. Machine learning relies on extensive and trusted amounts of data from the IoT and blockchain as input. In turn, the suggestions derived from machine learning contribute to the reliable operation of the supply chain and the ongoing generation of useful data records. Our study indicates that the combined application of these three DTs provides significant joint benefits to the VSC. Our proposed Intelligent VSC management system contributes to the existing literature by helping to better understand how different DTs can be combined in general supply chain management. Also, it expands the use and fields of application of DTs and contributes to the advancement of research in supply chain management. 6.2 Practical implications 6.2.1 Implications for supply chain management There are three critical issues in the VSC: vaccine quality, demand forecasting, and trust among stakeholders. The recent outbreak of the COVID-19 pandemic has made these three issues even more challenging. Our proposed intelligent VSC management system solves these three major problems to a large extent. In terms of vaccine quality, blockchain's decentralization and data immutability effectively avoid vaccine counterfeiting and ensure the authenticity and safety of vaccines. The IoT-based smart containers enable real-time monitoring of all vaccine information during the circulation process, ensuring the potency and activity of the vaccines. These merits help to cope with the national and international disruption of vaccine shipments brought about by the COVID-19 pandemic. Further, text sentiment classification models in machine learning enable accurate interpretation of the vaccine review data stored in the blockchain platform, facilitating all vaccine enterprises to improve vaccine quality and address consumers’ concerns. In particular, the trainable BILSTM model provides a high accuracy rate of nearly 80 %. The continued training based on the pre-trained model in the future can continuously improve the accuracy of text sentiment analysis to help to enhance the quality of the vaccine. As for demand forecasting, the open and transparent data in blockchain make the transmission of information real and reliable, avoiding the risk of the “bullwhip effect”. Moreover, based on the data stored in the blockchain, the time-series machine learning prediction models allow for a more accurate prediction of vaccine demand. For example, when using the GRU model to predict the demand for influenza vaccine in the United States, it is possible to keep the average annual prediction error within 3 %, which helps coordinate the supply and demand of the vaccine. Vaccine demand during the Covid-19 pandemic is highly uncertain, and accurate predictions by machine learning will provide considerable cost savings to the VSC. On trust issues among stakeholders, blockchain and the IoT allow all relevant participants in the system to track the status of vaccine delivery in real-time and enable rapid traceability for accountability, greatly enhancing trust among participants. The system guarantees the safety and effectiveness of vaccines and encourages enterprises to improve the quality, which further enhances customers’ trust in suppliers. This will improve supply chain resilience and enable VSC to recover quickly after the Covid-19 pandemic. Thus, our research has practical significance, as it effectively solves actual problems in the VSC, especially during the COVID-19 pandemic. In addition, the intelligent system we constructed is scalable. It is not only applicable to the VSC, but also provides management insights for other supply chains to solve problems of supply and demand, security, and supervision. 6.2.2 Implications for business management While improving demand forecasting accuracy and motivating enterprises to enhance product quality, our system also highlights the importance of consumers’ online reviews. Current customer satisfaction surveys are often conducted through questionnaires, ignoring the value of online reviews. Enterprises must be aware that consumers’ online reviews are data that can and should be leveraged. Online reviews are massive and transparent. Consumers use them to learn about the information and quality of different enterprises’ products to make their consumption decisions. Enterprises use them to understand customers’ needs and other enterprises’ products and achieve more competitive product improvements. Machine learning is a significant processing and analyzing tool for large-scale data, and collecting more data is essential to improve the accuracy of the analysis. Therefore, enterprises should adopt business strategies to motivate consumers to make online reviews. 6.2.3 Implications for government management The credibility of the system requires the government to take on more responsibility. First, implementing the intelligent VSC management system requires the collaborative efforts of many participants and high costs. The government should lead, organize, and coordinate all parties and bear most of the investment. Second, although this blockchain-based system is transparent, some hidden frauds still need to be detected by the government. Therefore, at the beginning of the system’s operation, it is the government’s responsibility to vet the participants, and those who pass the vetting process are eligible to join the system. During the operation of the system, the government should check the records uploaded by each participant and disclose and hold them accountable for violations. This will enhance the trust among the participants in the supply chain. In addition, we highlighted another important trust issue: the lack of public knowledge about immunization leads to mistrust in the national immunization program, thus hindering immunization coverage. So popularizing immunization knowledge is an important task for the government. Similarly, the ethical and moral risks associated with the widespread use of DTs have caused concerns in the public mind that need to be alleviated through government-guided actions, which would contribute to the innovative development of DTs. 6.3 Limitations and future research We used past vaccination doses for the vaccine demand forecast. However, population size, vaccine prices, immunization awareness, and local policies are all influencing factors (Chiu et al., 2008, Dizbay and Öztürkoğlu, 2020, Sarkar et al., (2020, Oct)., Wong et al., 2020). Complex models considering various factors can be constructed in the future to predict vaccine demand more accurately. Our proposed intelligent VSC management system also has implementations and adaptation challenges. A major challenge is the costs of implementation, energy consumption, and operations. Despite the benefits of an efficient and smart VSC system, the considerable upfront cost in the integrated application of DTs may hinder the implementation of this system. The cost includes the installation and maintenance cost of the large number of IoT devices, the transaction cost of real-time blockchain monitoring, and the computational cost of fine-grained machine learning. Therefore, a valuable direction for research is to reduce the cost of applying multiple DTs. Disclosure of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. CRediT authorship contribution statement Hui Hu: Writing – original draft. Jiajun Xu: Software, Investigation. Mengqi Liu: Writing – review & editing, Project administration. Ming K. Lim: Supervision, Conceptualization. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Hui Hu is an Associate Professor in Business Economics and Management at Wuhan University. His research interests are in the area of sustainable production and consumption and sustainable supply chain. His work has been published in journals such as Emerging Markets Finance & Trade, Journal of Cleaner Production, Applied Energy, Energy, IET Renewable Power Generation, Journal of Intelligent and Fuzzy Systems and Australasian Accounting Business and Finance Journal. Jiajun Xu is a Research Fellow and Ph.D. Student at Wuhan University. His research interests are in the area of business analytics and management. Mengqi Liu is an Associate Professor in Operations Management at Hunan University. His research interests are in the area of supply chain and operational research. His work has been published in journals such as European Journal of Operational Research, International Journal of Production Research, Journal of Industrial & Management Optimization, International Transactions in Operational Research, Asia-Pacific Journal of Operational Research, Annals of Operations Research, Transportation Research Part E-Logistics and Transportation Review and Omega. Ming K. Lim is currently a Professor of Supply Chain Management and Digitalization at the Adam Smith Business School, University of Glasgow. Before joining the University of Glasgow, he was Dean of the College of Mechanical Engineering and Distinguished Professor of Manufacturing & Logistics Engineering at Chongqing University and Co-cluster Lead/Professor of Supply Chain and Operations Management at Coventry University. Professor Lim is Editor-in-Chief of the International Journal of Logistics Research and Applications and an editorial board member of a range of leading journals. He has published over 175 papers in leading journals in the fields, such as Int. J. of Operations & Production Management, European J. of Operational Research, Int. J. of Production Economics, Omega, Transportation Research, Computers & Industrial Engineering, J. of the Operational Research Society, Annals of Operations Research, Int. J. of Production Research, Production Planning and Control, Expert Systems with Applications, Resources, Conservation & Recycling, and Industrial Management & Data System. Appendix See the Tables A.1 and A.2 .Table A.1 Computational errors of the three machine learning models. RNN LSTM GRU Mean absolute error (million doses) 10.07 7.21 5.42 Root mean square error (million doses) 10.37 9.02 5.97 Mean absolute percentage error (%) 5.55 3.16 3.02 Table A.2 Vaccine-related examples of consumer reviews in the data set. ID Vaccine Review Rating 90,103 Pneumococcal 13-valent vaccine I have suffered extreme muscle stiffness and weakness in my back, shoulders, and legs since taking this vaccine. I have also developed a chronic upper respiratory infection. This vaccine should be taken off the market. 1 214,931 Human papillomavirus vaccine Given to daughter at age 13. Diagnosed with chronic idiopathic thrombocytopenic purpura at age 13.5 years. Side effects list bruising, bleeding tendency, and red pin picks on skin. Are the several side effects due to a temporary decrease in platelet function, much like aspirin or the trigger of a platelet antibody? 2 180,871 Influenza virus vaccine, inactivated There was no pain at the injection site. Both my boyfriend and I received the shot and had close to the same experience, except I have been having extreme vomiting, after 2 weeks of getting the shot. We both experienced a severe headache starting within 30 min of receiving the shot that lasted about a day and a half, followed by severe diarrhea that is still ongoing, three weeks from the time of injection with nausea. He is having lots of muscle aches still, while I'm having severe pain in one of my legs near the hip. Neither of us had problems or sickness before the shot that would indicate it was something other than the shot. 3 88,838 Mumps virus vaccine Lots of itching all over the body, sometimes severe. 4 90,112 Pneumococcal 13-valent vaccine Side effects of injection include very sore arm effectively rendered useless. Muscle aches and joint pain without relief would not allow me to sleep. Fatigue and back pain plus headache contributed to an overall miserable night and day. I do not know how long this will continue, it has been 36 h. This vaccine requires at least two days off work and should be accompanied by a good narcotic. 5 193,708 Influenza virus vaccine, live, trivalent This was very simple to use. I thought it would be great. My 3-year-old has vomiting as a side effect. My 6-year-old and I seem to have no side effects. 6 230,081 Meningococcal group B vaccine My child got a headache, fever for two days after having the Bexsero vaccine. It made me worry about if she will get meningitis disease. 7 214,942 Human papillomavirus vaccine I got the shot, the second one hurt the most. The pain in my arm only lasted maybe 2 h at the most. When I got the last shot, the nurse injected it into the joint of my shoulder, which hurt for a lot longer but that was the nurse's fault, not the Gardasil. I had no side effects. I believe getting the vaccination was a smart thing to do and was worth it. 8 180,874 Influenza virus vaccine, inactivated Barely felt the pinch. Just a little redness, bump around the injection and itchiness for a few days. Not sore at all compared to regular shots. 9 46,651 Gardasil vaccine I am 18 years old and have had all three of these shots. I don't understand why people make this shot to be such a big deal. It barely hurt when I got them. Even if it did, who cares? If the shot is going to help you out in the long run. I never had any bad side effects. I recommend it to any woman who likes to be proactive. 10 Source: Drugs.com website. See the Fig. A.1 .Fig. A.1 Rating distribution in the training and testing sets. Data availability Data will be made available on request. Acknowledgments We thank the National Natural Science Foundation of China (71974151, 71871091 and 72171079) and the Major Program of the National Social Science Foundation (20&ZD072 and 19ZDA083) for the financial support. ==== Refs References Abdel-Basset M. Chang V. Nabeeh N.A. An intelligent framework using disruptive technologies for COVID-19 analysis Technological Forecasting and Social Change 163 2021 120431 Abdel-Basset M. Manogaran G. Mohamed M. Sep). Internet of Things (IoT) and its impact on supply chain: A framework for building smart, secure and efficient systems Future Generation Computer Systems-the International Journal of Escience 86 2018 614 628 Abeyratne S.A. Monfared R.P. Blockchain ready manufacturing supply chain using distributed ledger International Journal of Research in Engineering and Technology 5 9 2016 1 10 Accorsi R. Bortolini M. Baruffaldi G. Pilati F. Ferrari E. Internet-of-things paradigm in food supply chains control and management Procedia Manufacturing 11 2017 889 895 Adarsh S. Joseph S.G. John F. Lekshmi M. Asharaf S. A Transparent and Traceable Coverage Analysis Model for Vaccine Supply-Chain Using Blockchain Technology IT Professional 23 4 2021 28 35 Adida E. Dey D. Mamani H. Operational issues and network effects in vaccine markets European Journal of Operational Research 231 2 2013 414 427 32288068 Alam S.T. Ahmed S. Ali S.M. Sarker S. Kabir G. Challenges to COVID-19 vaccine supply chain: Implications for sustainable development goals International Journal of Production Economics 239 2021 108193 Alkhoori O. Hassan A. Almansoori O. Debe M. Salah K. Jayaraman R. …Rehman M.H.U. Design and Implementation of CryptoCargo: A Blockchain-Powered Smart Shipping Container for Vaccine Distribution IEEE Access 9 2021 53786 53803 Angeles R. RFID technologies: Supply-chain applications and implementation issues Information Systems Management 22 1 2005 51 65 Antal C. Cioara T. Antal M. Anghel I. Blockchain platform for COVID-19 vaccine supply management IEEE Open Journal of the Computer Society 2 2021 164 178 Arora G. Joshi J. Mandal R.S. Shrivastava N. Virmani R. Sethi T. Artificial intelligence in surveillance, diagnosis, drug discovery and vaccine development against COVID-19 Pathogens 10 8 2021 1048 34451513 Badhotiya G.K. Sharma V.P. Prakash S. Kalluri V. Singh R. Investigation and assessment of blockchain technology adoption in the pharmaceutical supply chain Materials Today: Proceedings 46 2021 10776 10780 Bag S. Tiwari M.K. Chan F.T.S. Jan). Predicting the consumer's purchase intention of durable goods: An attribute-level analysis Journal of Business Research 94 2019 408 419 Banerjee M. Lee J. Choo K.-K.-R. A blockchain future for internet of things security: A position paper Digital Communications and Networks 4 3 2018 149 160 Bangura J.B. Xiao S. Qiu D. Ouyang F. Chen L. Barriers to childhood immunization in sub-Saharan Africa: A systematic review BMC Public Health 20 1 2020 1 15 31898494 Baryannis G. Dani S. Antoniou G. Predicting supply chain risks using machine learning: The trade-off between performance and interpretability Future Generation Computer Systems 101 2019 993 1004 Betcheva L. Erhun F. Jiang H. OM Forum—Supply Chain Thinking in Healthcare: Lessons and Outlooks Manufacturing & Service Operations Management 23 6 2021 1333 1353 Bloom D.E. Cadarette D. Ferranna M. The societal value of vaccination in the age of COVID-19 American Journal of Public Health 111 6 2021 1049 1054 33856880 Budd J. Miller B.S. Manning E.M. Lampos V. Zhuang M. Edelstein M. …Keegan N. Digital technologies in the public-health response to COVID-19 Nature Medicine 26 8 2020 1183 1192 Carbonneau R. Laframboise K. Vahidov R. Application of machine learning techniques for supply chain demand forecasting European Journal of Operational Research 184 3 2008 1140 1154 Casado-Vara R. Prieto J. De la Prieta F. Corchado J.M. How blockchain improves the supply chain: Case study alimentary supply chain Procedia Computer Science 134 2018 393 398 Catalini C. Gans J.S. Jul). Some Simple Economics of the Blockchain Communications of the ACM 63 7 2020 80 90 Chamola V. Hassija V. Gupta V. Guizani M. A comprehensive review of the COVID-19 pandemic and the role of IoT, drones, AI, blockchain, and 5G in managing its impact IEEE Access 8 2020 90225 90265 Chandra D. Kumar D. A fuzzy MICMAC analysis for improving supply chain performance of basic vaccines in developing countries Expert Review of Vaccines 17 3 2018 263 281 29115165 Chatterjee S. Goyal D. Prakash A. Sharma J. Exploring healthcare/health-product ecommerce satisfaction: A text mining and machine learning application Journal of Business Research 131 2021 815 825 Chen J. Cai T. He W. Chen L. Zhao G. Zou W. Guo L. A blockchain-driven supply chain finance application for auto retail industry Entropy 22 1 2020 95 33285870 Chen M.K. Tai T.W. Hung T.Y. Apr). Component selection system for green supply chain Expert Systems with Applications 39 5 2012 5687 5701 Chick S.E. Hasija S. Nasiry J. Information elicitation and influenza vaccine production Operations Research 65 1 2017 75 96 Chiu, R. K., Chang, C. M., & Chang, Y. C. (2008, Oct 18-20). A Forecasting Model for Deciding Annual Vaccine Demand. [Icnc 2008: Fourth international conference on natural computation, vol 7, proceedings]. 4th International Conference on Natural Computation (ICNC 2008), Jian, China. Chowdhury P. Paul S.K. Kaisar S. Moktadir M.A. COVID-19 pandemic related supply chain studies: A systematic review Transportation Research Part E: Logistics and Transportation Review 148 2021 102271 Ćirović G. Pamučar D. Božanić D. Green logistic vehicle routing problem: Routing light delivery vehicles in urban areas using a neuro-fuzzy model Expert Systems with Applications 41 9 2014 4245 4258 Cole R. Stevenson M. Aitken J. Jun). Blockchain technology: Implications for operations and supply chain management. Supply Chain Management-an International Journal 24 4 2019 469 483 Cui L. Xiao Z. Wang J. Chen F. Pan Y. Dai H. Qin J. Improving Vaccine Safety Using Blockchain ACM Transactions on Internet Technology (TOIT) 21 2 2021 1 24 Dai T. Song J.-S. Transforming COVID-19 vaccines into vaccination Health Care Management Science 24 3 2021 455 459 34132977 De Boeck K. Decouttere C. Vandaele N. Vaccine distribution chains in low-and middle-income countries: A literature review OMEGA 97 2020 102097 De Villiers C. Kuruppu S. Dissanayake D. Jul). A (new) role for business-Promoting the United Nations' Sustainable Development Goals through the internet-of-things and blockchain technology Journal of Business Research 131 2021 598 609 Decouttere C. De Boeck K. Vandaele N. Advancing sustainable development goals through immunization: A literature review Globalization and Health 17 1 2021 1 29 33397377 Dizbay İ.E. Öztürkoğlu Ö. Determining significant factors affecting vaccine demand and factor relationships using fuzzy DEMATEL method International Conference on Intelligent and Fuzzy Systems 2020 Duijzer L.E. Van Jaarsveld W. Dekker R. Literature review: The vaccine supply chain European Journal of Operational Research 268 1 2018 174 192 Dwivedi Y.K. Hughes D.L. Coombs C. Constantiou I. Duan Y.Q. Edwards J.S. …Upadhyay N. Dec). Impact of COVID-19 pandemic on information management research and practice: Transforming education, work and life International Journal of Information Management 55, 102211 2020 102211 Dwivedi Y.K. Hughes L. Ismagilova E. Aarts G. Coombs C. Crick T. …Williams M.D. Apr). Artificial Intelligence (AI): Multidisciplinary perspectives on emerging challenges, opportunities, and agenda for research, practice and policy International Journal of Information Management 57, 101994 2021 101994 Eccleston-Turner M. Upton H. Jun). International Collaboration to Ensure Equitable Access to Vaccines for COVID-19: The ACT-Accelerator and the COVAX Facility Milbank Quarterly 99 2 2021 426 449 33650737 Edwards B. Gloor C.A. Toussaint F. Guan C. Furniss D. Human factors: The pharmaceutical supply chain as a complex sociotechnical system International Journal for Quality in Health Care 33 Supplement_1 2021 56 59 32991730 El Baz J. Ruel S. Can supply chain risk management practices mitigate the disruption impacts on supply chains’ resilience and robustness? Evidence from an empirical survey in a COVID-19 outbreak era International Journal of Production Economics 233 2021 107972 Ellis S. Morris H.D. Santagate J. IoT-enabled analytic applications revolutionize supply chain planning and execution International Data Corporation (IDC) White Paper 13 2015 259697 Fahimnia B. Jabbarzadeh A. Sarkis J. Greening versus resilience: A supply chain design perspective Transportation Research Part E: Logistics and Transportation Review 119 2018 129 148 Fan Z.P. Che Y.J. Chen Z.Y. May). Product sales forecasting using online reviews and historical sales data: A method combining the Bass model and sentiment analysis Journal of Business Research 74 2017 90 100 Finkenstadt D.J. Handfield R.B. Tuning value chains for better signals in the post-COVID era: Vaccine supply chain concerns International Journal of Operations & Production Management 41 2021 1302 1317 Francisco K. Swanson D. The supply chain has no clothes: Technology adoption of blockchain for supply chain transparency Logistics 2 1 2018 1 13 Georgiadis G.P. Georgiadis M.C. Optimal planning of the COVID-19 vaccine supply chain Vaccine 39 37 2021 5302 5312 34373118 Golan M.S. Trump B.D. Cegan J.C. Linkov I. Supply chain resilience for vaccines: Review of modeling approaches in the context of the COVID-19 pandemic Industrial Management & Data Systems 121 7 2021 1723 1748 Guan D. Wang D. Hallegatte S. Davis S.J. Huo J. Li S. …Coffman D.M. Global supply-chain effects of COVID-19 control measures Nature Human Behaviour 4 6 2020 577 587 Gumus A.T. Guneri A.F. Ulengin F. A new methodology for multi-echelon inventory management in stochastic and neuro-fuzzy environments International Journal of Production Economics 128 1 2010 248 260 Haidari L.A. Brown S.T. Ferguson M. Bancroft E. Spiker M. Wilcox A. …Lee B.Y. The economic and operational value of using drones to transport vaccines Vaccine 34 34 2016 4062 4067 27340098 Hanson C.M. George A.M. Sawadogo A. Schreiber B. Is freezing in the vaccine cold chain an ongoing issue? A literature review. Vaccine 35 17 2017 2127 2133 28364920 Hasan H. AlHadhrami E. AlDhaheri A. Salah K. Jayaraman R. Smart contract-based approach for efficient shipment management Computers & Industrial Engineering 136 2019 149 159 Hobbs J.E. Food supply chains during the COVID-19 pandemic Canadian Journal of Agricultural Economics 68 2 2020 171 176 Hwang S. Kim J. Park E. Kwon S.J. Who will be your next customer: A machine learning approach to customer return visits in airline services Journal of Business Research 121 2020 121 126 Ivanov D. Predicting the impacts of epidemic outbreaks on global supply chains: A simulation-based analysis on the coronavirus outbreak (COVID-19/SARS-CoV-2) case Transportation Research Part E: Logistics and Transportation Review 136 2020 101922 Ivanov D. Dolgui A. OR-methods for coping with the ripple effect in supply chains during COVID-19 pandemic: Managerial insights and research implications International Journal of Production Economics 232 2021 107921 Jarrett S. Wilmansyah T. Bramanti Y. Alitamsar H. Alamsyah D. Krishnamurthy K.R. …Pagliusi S. The role of manufacturers in the implementation of global traceability standards in the supply chain to combat vaccine counterfeiting and enhance safety monitoring Vaccine 38 52 2020 8318 8325 33199075 Karmaker C.L. Ahmed T. Ahmed S. Ali S.M. Moktadir M.A. Kabir G. Improving supply chain sustainability in the context of COVID-19 pandemic in an emerging economy: Exploring drivers using an integrated model Sustainable Production and Consumption 26 2021 411 427 33015267 Kartoglu U. Milstien J. Tools and approaches to ensure quality of vaccines throughout the cold chain Expert Review of Vaccines 13 7 2014 843 854 24865112 Kaufmann J.R. Miller R. Cheyne J. Vaccine supply chains need to be better funded and strengthened, or lives will be at risk Health Affairs 30 6 2011 1113 1121 21653965 Keesara S. Jonas A. Schulman K. Covid-19 and health care’s digital revolution New England Journal of Medicine 382 23 2020 e82 32240581 Kinra A. Ivanov D. Das A. Dolgui A. Ripple effect quantification by supplier risk exposure assessment International Journal of Production Research 58 18 2020 5559 5578 Korpela, K., Hallikas, J., & Dahlberg, T. (2017). Digital supply chain transformation toward blockchain integration. proceedings of the 50th Hawaii international conference on system sciences. Kouzinopoulos, C. S., Spathoulas, G., Giannoutakis, K. M., Votis, K., Pandey, P., Tzovaras, D., Katsikas, S. K., Collen, A., & Nijdam, N. A. (2018). Using blockchains to strengthen the security of internet of things. International ISCIS Security Workshop. Kraus S. Schiavone F. Pluzhnikova A. Invernizzi A.C. Digital transformation in healthcare: Analyzing the current state-of-research Journal of Business Research 123 2021 557 567 Lee S.M. Trimi S. Convergence innovation in the digital age and in the COVID-19 pandemic crisis Journal of Business Research 123 2021 14 22 33012897 Li Z. Liu G. Liu L. Lai X. Xu G. IoT-based tracking and tracing platform for prepackaged food supply chain Industrial Management & Data Systems 117 2017 1906 1916 Lin Q. Zhao Q. Lev B. Cold chain transportation decision in the vaccine supply chain European Journal of Operational Research 283 1 2020 182 195 Liu R. Tan C. Zhao C. Pricing and coordination of vaccine supply chain based on blockchain technology Internet Research 31 2021 2096 2119 Manavalan E. Jayakrishna K. Jan). A review of Internet of Things (IoT) embedded sustainable supply chain for industry 4.0 requirements Computers & Industrial Engineering 127 2019 925 953 Marbouh D. Abbasi T. Maasmi F. Omar I.A. Debe M.S. Salah K. …Ellahham S. Blockchain for COVID-19: Review, opportunities, and a trusted tracking system Arabian Journal for Science and Engineering 45 12 2020 9895 9911 33072472 Mbunge E. Akinnuwesi B. Fashoto S.G. Metfula A.S. Mashwama P. A critical review of emerging technologies for tackling COVID-19 pandemic Human Behavior and Emerging Technologies 3 1 2021 25 39 33363278 Min H. Blockchain technology for enhancing supply chain resilience Business Horizons 62 1 2019 35 45 Mondal S. Wijewardena K.P. Karuppuswami S. Kriti N. Kumar D. Chahal P. Blockchain inspired RFID-based information architecture for food supply chain IEEE Internet of Things Journal 6 3 2019 5803 5813 Monteleone, S., Sampaio, M., & Maia, R. F. (2017). A novel deployment of smart Cold Chain system using 2G-RFID-Sys temperature monitoring in medicine Cold Chain based on Internet of Things. 2017 IEEE International Conference on Service Operations and Logistics, and Informatics (SOLI). Musamih A. Jayaraman R. Salah K. Hasan H.R. Yaqoob I. Al-Hammadi Y. Blockchain-based solution for distribution and delivery of COVID-19 vaccines IEEE Access 9 2021 71372 71387 34812393 Nagurney A. Supply chain game theory network modeling under labor constraints: Applications to the Covid-19 pandemic European Journal of Operational Research 293 3 2021 880 891 33519049 Nandi S. Sarkis J. Hervani A.A. Helms M.M. Redesigning supply chains using blockchain-enabled circular economy and COVID-19 experiences Sustainable Production and Consumption 27 2021 10 22 33102671 Ng W.Y. Tan T.-E. Movva P.V. Fang A.H.S. Yeo K.-K. Ho D. …Wong T.Y. Blockchain applications in health care for COVID-19 and beyond: A systematic review The Lancet Digital Health 3 12 2021 e819 e829 34654686 Nhamo G. Chikodzi D. Kunene H.P. Mashula N. COVID-19 vaccines and treatments nationalism: Challenges for low-income countries and the attainment of the SDGs Global Public Health 16 3 2021 319 339 33317389 Ni D. Xiao Z. Lim M.K. A systematic review of the research trends of machine learning in supply chain management International Journal of Machine Learning and Cybernetics 11 7 2020 1463 1482 Omar I.A. Jayaraman R. Debe M.S. Salah K. Yaqoob I. Omar M. Automating procurement contracts in the healthcare supply chain using blockchain smart contracts IEEE Access 9 2021 37397 37409 Pandey N. Pal A. Impact of digital surge during Covid-19 pandemic: A viewpoint on research and practice International Journal of Information Management 55 2020 102171 Piccialli F. Giampaolo F. Prezioso E. Camacho D. Acampora G. Oct). Artificial intelligence and healthcare: Forecasting of medical bookings through multi-source time-series fusion & nbsp Information Fusion 74 2021 1 16 Preiss S. Garçon N. Cunningham A.L. Strugnell R. Friedland L.R. Vaccine provision: Delivering sustained & widespread use Vaccine 34 52 2016 6665 6671 27884478 Pundir A.K. Jagannath J.D. Ganapathy L. IMPROVING SUPPLY CHAIN VISIBILITY USING IoT-INTERNET OF THINGS [2019 IEEE 9th annual computing and communication workshop and conference (CCWC)]. 9th IEEE Annual Computing and Communication Workshop and Conference (CCWC) 2019 Univ Nevada, Las Vegas, NV Rappuoli R. Black S. Bloom D.E. Vaccines and global health: In search of a sustainable model for vaccine development and delivery Science Translational Medicine 11 497 2019 eaaw2888 31217336 Rejeb A. Keogh J.G. Treiblmaier H. Leveraging the internet of things and blockchain technology in supply chain management Future Internet 11 7 2019 161 Remko V.H. Research opportunities for a more resilient post-COVID-19 supply chain–closing the gap between research findings and industry practice International Journal of Operations & Production Management 40 4 2020 341 355 Ross D.F. Weston F.S. Stephen W. Introduction to supply chain management technologies 2010 Crc Press Saberi S. Kouhizadeh M. Sarkis J. Shen L. Blockchain technology and its relationships to sustainable supply chain management International Journal of Production Research 57 7 2019 2117 2135 Sarkar K. Khajanchi S. Nieto J.J. Modeling and forecasting the COVID-19 pandemic in India Chaos Solitons & Fractals 139 2020 110049 Sarkis J. Supply chain sustainability: Learning from the COVID-19 pandemic International Journal of Operations & Production Management 1 2020 63 73 Secundo G. Shams S.R. Nucci F. Digital technologies and collective intelligence for healthcare ecosystem: Optimizing Internet of Things adoption for pandemic management Journal of Business Research 131 2021 563 572 Sharma A. Adhikary A. Borah S.B. Covid-19′ s impact on supply chain decisions: Strategic insights from NASDAQ 100 firms using Twitter data Journal of Business Research 117 2020 443 449 32834209 Singh R. Dwivedi A.D. Srivastava G. Internet of things based blockchain for temperature monitoring and counterfeit pharmaceutical prevention Sensors 20 14 2020 3951 32708588 Sreedevi R. Saranga H. Uncertainty and supply chain risk: The moderating role of supply chain flexibility in risk mitigation International Journal of Production Economics 193 2017 332 342 Steckel J.H. Gupta S. Banerji A. Supply chain decision making: Will shorter cycle times and shared point-of-sale information necessarily help? Management Science 50 4 2004 458 464 Tajima M. Strategic value of RFID in supply chain management Journal of Purchasing and Supply Management 13 4 2007 261 273 Tanwar S. Bhatia Q. Patel P. Kumari A. Singh P.K. Hong W.-C. Machine learning adoption in blockchain-based smart applications: The challenges, and a way forward IEEE Access 8 2019 474 488 Ting D.S.W. Carin L. Dzau V. Wong T.Y. Digital technology and COVID-19 Nature Medicine 26 4 2020 459 461 Tsolakis N. Niedenzu D. Simonetto M. Dora M. Kumar M. Supply network design to address United Nations Sustainable Development Goals: A case study of blockchain implementation in Thai fish industry Journal of Business Research 131 2021 495 519 Umair M. Cheema M.A. Cheema O. Li H. Lu H. Impact of COVID-19 on IoT adoption in healthcare, smart homes, smart buildings, smart cities, transportation and industrial IoT Sensors 21 11 2021 3838 34206120 Ustundag A. Tanyas M. The impacts of radio frequency identification (RFID) technology on supply chain costs Transportation research part e: Logistics and Transportation Review 45 1 2009 29 38 Vargo D. Zhu L. Benwell B. Yan Z. Digital technology use during COVID-19 pandemic: A rapid review Human Behavior and Emerging Technologies 3 1 2021 13 24 Wamba S.F. Queiroz M.M. Jun). Blockchain in the operations and supply chain management: Benefits, challenges and future research opportunities International Journal of Information Management 52, 102064 2020 102064 Wang H. Wang Y. Cao Z. Li Z. Xiong G. An overview of blockchain security analysis China Cyber Security Annual Conference. 2018 Wang Z. Wang L. Chen Q. Lu L. Hong J. A Traditional Chinese Medicine Traceability System Based on Lightweight Blockchain Journal of Medical Internet Research 23 6 2021 e25946 34152279 Weintraub R.L. Subramanian L. Karlage A. Ahmad I. Rosenberg J. COVID-19 Vaccine To Vaccination: Why Leaders Must Invest In Delivery Strategies Now: Analysis describe lessons learned from past pandemics and vaccine campaigns about the path to successful vaccine delivery for COVID-19 Health Affairs 40 1 2021 33 41 33211554 Whitelaw S. Mamas M.A. Topol E. Van Spall H.G. Applications of digital technology in COVID-19 pandemic planning and response The Lancet Digital Health 2 8 2020 e435 e440 32835201 Wong L.P. Alias H. Wong P.F. Lee H.Y. AbuBakar S. Sep). The use of the health belief model to assess predictors of intent to receive the COVID-19 vaccine and willingness to pay Human Vaccines & Immunotherapeutics 16 9 2020 2204 2214 32730103 Xu S. Zhang X. Feng L. Yang W. Disruption risks in supply chain management: A literature review based on bibliometric analysis International Journal of Production Research 58 11 2020 3508 3526 Yaqoob I. Salah K. Jayaraman R. Al-Hammadi Y. Blockchain for healthcare data management: Opportunities, challenges, and future recommendations Neural Computing and Applications 2021 1 16 Yong B. Shen J. Liu X. Li F. Chen H. Zhou Q. An intelligent blockchain-based system for safe vaccine supply and supervision International Journal of Information Management 52 2020 102024 Zaffran M. Vandelaer J. Kristensen D. Melgaard B. Yadav P. Antwi-Agyei K. Lasher H. The imperative for stronger vaccine supply and logistics systems Vaccine 31 2013 B73 B80 23598495 Zeitouny S. Suda K.J. Mitsantisuk K. Law M.R. Tadrous M. Mapping global trends in vaccine sales before and during the first wave of the COVID-19 pandemic: A cross-sectional time-series analysis Bmj Global Health 6 12 2021 e006874 Zelbst P.J. Green K.W. Sower V.E. Bond P.L. The impact of RFID, IIoT, and Blockchain technologies on supply chain transparency Journal of Manufacturing Technology Management 31 2019 441 457 Zheng, Q., Li, Y., Chen, P., & Dong, X. (2018). An innovative IPFS-based storage model for blockchain. 2018 IEEE/WIC/ACM international conference on web intelligence (WI).	36506475	PMC9718486	NO-CC CODE	2022-12-06 23:23:40	no	J Bus Res. 2023 Feb 2; 156:113480	utf-8	J Bus Res	2,022	10.1016/j.jbusres.2022.113480	oa_other
==== Front Int J Infect Dis Int J Infect Dis International Journal of Infectious Diseases 1201-9712 1878-3511 The Author(s). Published by Elsevier Ltd on behalf of International Society for Infectious Diseases. S1201-9712(22)00590-2 10.1016/j.ijid.2022.11.008 Article Comparative analysis of elderly hospitalized patients with COVID-19 or influenza A H1N1 virus infections Lv Yan # Yu Guodong # Zhang Xiaoli # Gu Jueqing Ye Chanyuan Lian Jiangshan Lu Xiaoqing Lu Yingfeng Yang Yida ⁎ State Key Laboratory for Diagnosis and Treatment of Infectious Diseases, National Clinical Research Center for Infectious Diseases, National Medical Center for Infectious Diseases, Collaborative Innovation Center for Diagnosis and Treatment of Infectious Diseases, The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China ⁎ Corresponding author. # Yan Lv, Guodong Yu, and Xiaoli Zhang contributed equally to this article. 9 11 2022 12 2022 9 11 2022 125 278284 16 12 2021 1 11 2022 4 11 2022 © 2022 The Author(s) 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objectives This study aimed to investigate the differences between elderly patients hospitalized with COVID-19 or influenza A H1N1 virus infections. Methods We contrasted two absolute groups of patients (age ≥60 years) infected with either COVID-19 (n = 222) or influenza A H1N1 virus infections (n = 96). Propensity score matching was used to reduce the imbalance between the two matched groups. The clinical features, imaging presentations, therapies, and prognosis data were compared between the two groups. Results The patients with influenza showed higher proportions of cough, expectoration, fatigue, and shortness of breath. Higher counts of lymphocytes, hemoglobin, and creatine kinase and lower counts of white blood cells, neutrophils, blood urea nitrogen, and C-reactive protein were found in the patients with COVID-19. Regarding the imaging characteristics, bilateral pneumonia was the most abnormal pattern in the two groups of patients. The incidence of acute respiratory distress syndrome or death was lower among the patients with COVID-19. Conclusion The clinical manifestations of patients with COVID-19 are more concealed than those of patients with influenza. Fewer symptoms of sputum production, fatigue, and shortness of breath, combined with lower counts of white blood cells, neutrophils, and C-reactive protein are the possible predictive factors of COVID-19 among elderly patients. Keywords COVID-19 Influenza A H1N1 virus pneumonia Elderly patients Clinical features ==== Body pmcIntroduction The SARS-CoV-2, SARS-CoV, Middle East respiratory syndrome coronavirus (MERS-CoV), and influenza A viruses are major pathogens that damage the respiratory system and can produce outbreaks of SARS, MERS, COVID‐19, and influenza A H1N1 virus pneumonia, respectively. SARS-CoV-2, SARS-CoV, and Middle East respiratory syndrome coronavirus are from the same genus and share many virological and epidemiological similarities. However, COVID-19 shows more similarities with influenza A H1N1 virus infections in the pattern and scale of spread than with SARS or MERS. For example, COVID-19 has higher proportions of asymptomatic and mild infections than SARS and MERS, which is similar to influenza A H1N1 virus infections. Both the COVID-19 and influenza A H1N1 viruses exhibit high viral shedding, which is essential for the spread of infection between hosts at an early stage of infection, which differs from SARS and MERS (Wu et al., 2021). In addition, the clinical manifestations and imaging manifestations of influenza A H1N1 virus pneumonia are similar to those of COVID-19 (Wu et al., 2021). As the diseases develop, some patients may develop acute respiratory distress syndrome (ARDS) and multiorgan failure, leading to death. However, the rate of ARDS is higher in influenza pneumonia, and the fatality rate is lower in COVID-19 (World Health Organization, 2020). Thus, the complications and prognosis of the two pneumonias are diverse. Moreover, previous research has shown that both COVID-19 and influenza A H1N1 viral pneumonia have high morbidity and mortality in elderly individuals (Abdelrahman et al., 2020; Chen et al., 2020; Grasselli et al., 2020). Primary data from Wuhan showed that the death rate among elderly patients (aged ≥60 years) infected with SARS-CoV-2 was 11.0% (Chen et al., 2020). Italian statistics showed that COVID-19 is far more serious among elderly patients, with a mortality rate of 3.6% among those aged between 60 and 69 years, 8.0% among those aged between 70 and 79 years, 14.8% among those aged >80 years, and over 20.0% among those aged >90 years compared with 2.3% in the general population (Grasselli et al., 2020). The mortality rates of H1N1 exhibit a J-shaped curve, and the mortality rate among elderly patients (aged ≥70 years) was 10.3% (Echevarría-Zuno et al., 2009). Due to the different complications, prognoses, and therapies of the two diseases, it is important for clinicians to identify them quickly and administer proper treatments to patients. Therefore, the purpose of this study was to identify the differences between hospitalized elderly patients infected with SARS-CoV-2 or influenza viruses to provide some guidance for their differential diagnoses. Materials and methods Subjects and data In total, 222 elderly (aged ≥60 years) patients infected with SARS-CoV-2 between January 17 and March 10, 2020 in Zhejiang, China were enrolled in this research. All confirmed patients were admitted to a designated hospital in accordance with the requirements of Zhejiang Province. As a comparison group, 96 elderly (aged ≥60 years) patients confirmed to have H1N1 influenza virus infection between November 1, 2017 and March 31, 2018 in Zhejiang who were hospitalized at the First Affiliated Hospital, College of Medicine, Zhejiang University were enrolled. Our study followed the ethical guidelines of the Declaration of Helsinki and obtained the approval of the medical ethics committee of the First Affiliated Hospital, College of Medicine, Zhejiang University. Data collection We retrospectively investigated the characteristics, underlying diseases, clinical symptoms, and lung images of the two groups. The treatments and prognosis were also recorded. All patients were diagnosed with COVID-19 or influenza A H1N1 viral pneumonia by positive polymerase chain reaction results, with a combination of clinical manifestations and imaging presentations. Statistical analysis All data were analyzed by SPSS (version 26.0), and a two-sided α <0.05 was considered statistically significant. Continuous variables with a normal or a non-normal distribution are described as the mean ± standard deviation or median (interquartile range), respectively. Numbers (%) and chi-square tests were used to describe the categorical variables and comparisons between the groups. To reduce the imbalance between the two groups, we performed this research based on propensity score matching (PSM). PSM was performed to obtain a 1:1 matched cohort using the ‘nearest-neighbor’ approach without replacement, with a match tolerance of 0.05. The following factors were included in the PSM: age, sex, current smoking status, and coexisting conditions, such as hypertension, heart disease, diabetes, etc. Results Baseline characteristics and underlying diseases The characteristics of the patients infected with COVID-19 or influenza A H1N1 virus before and after PSM are described in Table 1 .Table 1 Baseline characteristics and underlying diseases of patients with COVID-19 or influenza A H1N1 virus pneumonia. Table 1Characteristics Patients before PSM Patients after PSM COVID-19 (n = 222) H1N1 (n = 96) P-value Estimated difference, (95% CI) COVID-19 (n = 72) H1N1 (n = 72) P-value Estimated difference, (95% CI) Ages, years, median 66.00 (63.00-72.00) 71.00 (64.00-79.00) <0.001 -3.29 (-5.32 to -1.26) 67.00 (62.00-74.00) 70.00 (64.00-77.00) 0.231 -1.79 (-4.74 to 1.15) Male sex, n (%) 102 (45.6) 63 (65.6) <0.001 -0.20 (-0.31 to -0.08) 44 (61.1) 44 (61.1) 1.000 0.00 (-0.16 to 0.16) Current smoker, n (%) 21 (9.5) 24 (25.0) <0.001 -0.16 (-0.25 to -0.06) 15 (20.8) 15 (20.8) 1.000 0.00 (-0.13 to 0.13) Coexisting conditions, n (%) 113 (51.9) 71 (74.0) 0.117 -0.23 (-0.34 to -0.12) 40 (55.6) 49 (68.1) 0.123 -0.13 (-0.28 to 0.03) Hypertension, n (%) 99 (44.6) 52 (54.2) 0.049 -0.10 (-0.21 to 0.02) 37 (51.4) 44 (61.1) 0.314 -0.10 (-0.26 to 0.06) Heart disease, n (%) 31 (14.0) 22 (22.9) 0.291 -0.09 (-0.19 to 0.01) 14 (19.4) 16 (22.2) 0.837 -0.03 (-0.16 to 0.10) Diabetes, n (%) 42 (18.9) 21 (21.9) 0.544 -0.03 (-0.13 to 0.07) 17 (23.6) 17 (23.6) 1.000 0.00 (-0.14 to 0.14) Chronic obstructive pulmonary disease, n (%) 8 (3.6) 6 (6.3) 0.291 -0.03 (-0.08 to 0.03) 5 (6.9) 4 (5.6) 1.000 0.01 (-0.07 to 0.09) Asthma, n (%) 4 (1.8) 1 (1.0) 0.617 0.01 (-0.02 to 0.03) 4 (5.6) 1 (1.4) 0.363 0.04 (-0.02 to 0.10) Cancer, n (%) 4 (1.8) 7 (7.3) 0.014 -0.05 (-0.11 to 0.00) 2 (2.8) 3 (4.2) 1.000 -0.01 (-0.07 to 0.05) Immunosuppression, n (%) 2 (0.9) 9 (9.4) <0.001 -0.08 (-0.14 to -0.03) 2 (2.8) 1 (1.4) 1.000 0.01 (-0.03 to 0.06) Blood disease, n (%) 1 (0.5) 10 (10.4) <0.001 -0.10 (-0.16 to -0.04) 1 (1.4) 1 (1.4) 1.000 0.00 (-0.04 to 0.04) Chronic liver disease, n (%) 10 (4.5) 10 (10.4) 0.046 -0.06 (-0.13 to 0.01) 6 (8.3) 6 (8.3) 1.000 0.00 (-0.09 to 0.09) Chronic renal disease, n (%) 6 (2.7) 5 (5.2) 0.262 -0.03 (-0.07 to 0.02) 4 (5.6) 4 (5.6) 1.000 0.00 (-0.07 to 0.07) PSM, propensity score matching The influenza A H1N1 virus patients were older than the patients with COVID-19, with a median (interquartile range) age of 70.00 (64.00-77.00) versus 67.00 (62.00-74.00) years (P <0.001). The proportions of males and current smokers among the patients with COVID-19 were 46.0% and 9.5%, respectively, which were significantly lower than those among the patients with influenza A H1N1 virus (P <0.001, 95% confidence interval [CI], –0.31 to –0.08, P <0.001, 95% CI –0.25 to –0.06 for each). The patients with influenza had more coexisting diseases than the patients with COVID-19, including heart diseases (22.9% vs 14.0%, P = 0.049, 95% CI, –0.19 to 0.01), cancers (7.3% vs 1.8%, P-value = 0.014, 95% CI, –0.11to 0.00), immunosuppressive diseases (9.4% vs 0.9%, P <0.001, 95% CI, –0.14 to –0.03), blood diseases (10.4% vs 0.5%, P <0.001, 95% CI, –0.16 to –0.04), and chronic liver diseases (10.4% vs 4.5%, P <0.001, 95% CI, –0.13 to 0.01). There were no obvious differences in coexisting hypertension (44.6% vs 54.2%, P-value = 0.117, 95% CI, –0.21 to 0.02), diabetes (18.9% vs 21.9%, P-value = 0.544, 95% CI, –0.13 to 0.07), chronic obstructive pulmonary disease (3.6% vs 6.3%, P-value = 0.291, 95% CI, –0.08 to 0.03), asthma (3.6% vs 6.3%, P-value = 0.617, 95% CI, –0.02 to 0.03), or chronic renal disease (2.7% vs 5.2%, P-value = 0.262, 95% CI, –0.07 to 0.02) between the two groups. After PSM in the cohort, a 1:1 balanced cohort of 144 patients was obtained; in this cohort, 72 were patients with COVID-19, and 72 were patients with influenza. The distribution of the baseline characteristics between the two groups was similar for all covariates before and after PSM. Clinical symptoms and laboratory examinations In summary, the incidence of fever and cough was the highest in the two groups neither before and after the PSM analysis (shown in Table 2 ). After the PSM analysis, more concretely, the percentages of patients with COVID-19 with cough (72.2% vs 88.9%, P-value = 0.021, 95% CI, –0.29 to –0.04), sputum production (40.3% vs 86.1%, P <0.001, 95% CI, –0.60 to –0.32), fatigue (15.3% vs 50.0%, P <0.001, 95% CI, –0.49 to –0.20), and shortness of breath (15.3% vs 69.4%, P <0.01, 95% CI –0.68 to –0.41) were lower than those of patients with influenza. No apparent differences existed in the proportion of fever (86.1% vs 80.6%, P-value = 0.502, 95% CI, –0.07 to 0.18), hemoptysis (1.4% vs 4.2%, P-value = 0.612, 95% CI, –0.08 to 0.03), sore throat (12.5% vs 5.6%, P-value = 0.245, 95% CI, -0.02 to 0.16), nasal obstruction (1.4% vs 1.4%, P-value = 1.000, 95% CI, –0.04 to 0.04), headache (6.9% vs 9.7%, P-value =0.763, 95% CI, –0.12 to 0.06), muscle ache (12.5% vs 8.3%, P-value = 0.585, 95% CI, –0.06 to 0.14) and gastrointestinal symptoms (11.1% vs 13.9%, P-value = 0.801, 95% CI, –0.14 to 0.08) between the two groups. The results were similar to the data before the PSM analysis.Table 2 Clinical symptoms and laboratory examinations of the patients with COVID-19 or influenza A H1N1 virus pneumonia. Table 2Characteristics Patients before PSM Patients after PSM COVID-19 (n = 222) H1N1 (n = 96) P-value Estimated difference, (95% CI) COVID-19 (n = 72) H1N1 (n = 72) P-value Estimated difference, (95% CI) Fever, n (%) 188 (84.7) 80 (83.3) 0.761 0.01 (-0.07 to 0.10) 62 (86.1) 58 (80.6) 0.502 0.06 (-0.07 to 0.18) Cough, n (%) 152 (68.5) 83 (86.5) <0.001 -0.18 (-0.27 to -0.09) 52 (72.2) 64 (88.9) 0.021 -0.17 (-0.29 to -0.04) Sputum production, n (%) 86 (38.7) 81 (84.4) <0.001 -0.46 (-0.55 to -0.36) 29 (40.3) 62 (86.1) <0.001 -0.46 (-0.60 to -0.32) Hemoptysis, n (%) 4 (1.8) 4 (4.2) 0.216 -0.02 (-0.07 to 0.02) 1 (1.4) 3 (4.2) 0.612 -0.03 (-0.08 to 0.03) Sore throat, n (%) 23 (10.4) 8 (8.3) 0.576 0.02 (-0.05 to 0.09) 9 (12.5) 4 (5.6) 0.245 0.07 (-0.02 to 0.16) Nasal obstruction, n (%) 3 (1.4) 2 (2.1) 0.630 -0.01 (-0.04 to 0.03) 1 (1.4) 1 (1.)4 1.000 0.00 (-0.04 to 0.04) Headache, n (%) 11 (5.0) 8 (8.3) 0.243 -0.03 (-0.10 to 0.03) 5 (6.9) 7 (9.7) 0.763 -0.03 (-0.12 to 0.06) Muscle ache, n (%) 24 (10.8) 6 (6.3) 0.201 0.05 (-0.02 to 0.11) 9 (12.5) 6 (8.3) 0.585 0.04 (-0.06 to 0.14) Fatigue, n (%) 38 (17.1) 46 (47.9) <0.001 -0.31 (-0.42 to -0.20) 11 (15.3) 36 (50.0) <0.001 -0.35 (-0.49 to -0.20) Shortness of breath, n (%) 20 (9.0) 67 (69.7) <0.001 -0.61 (-0.71 to -0.51) 11 (15.3) 50 (69.4) <0.001 -0.54 (-0.68 to -0.41) Gastrointestinal symptoms, n (%) 22 (9.9) 11 (11.5) 0.691 -0.02 (-0.09 to 0.06) 8 (11.1) 10 (13.9) 0.801 -0.03 (-0.14 to 0.08) Nausea/vomiting, n (%) 10 (4.5) 7 (7.3) 0.415 -0.03 (-0.09 to 0.02) 3 (4.2) 7 (9.7) 0.325 -0.06 (-0.14 to 0.03) Diarrhea, n (%) 16 (7.2) 5 (5.2) 0.627 0.02 (-0.04 to 0.08) 7 (9.7) 4 (5.6) 0.530 0.04 (-0.04 to 0.13) Blood routine White blood cell, (× 109 per l) 6.13 (5.67-6.59) 11.93 (5.58-18.28) <0.001 -5.35 (-11.12 to 0.41) 5.20 (4.10-7.46) 8.45 (5.18-12.03) <0.001 -3.06 (-4.50 to -1.63) Neutrophils, (× 109 per l) 4.42 (4.00-4.84) 7.19 (6.16-8.22) <0.001 -2.62 (-3.65 to -1.59) 3.40 (2.90-4.64) 6.43 (4.00-10.18) <0.001 -3.29 (-4.58 to -2.01) Lymphocytes, (× 109 per l) 1.20 (1.00-1.40) 3.99 (2.02-10.01) 0.001 -2.54 (-7.99 to 2.91) 1.00 (0.70-1.40) 0.82 (0.45-1.24) 0.030 0.36 (-0.19 to 0.91) Platelets, (× 109 per l) 203.30 (191.75-214.85) 183.84 (154.60-213.08) 0.001 33.02 (6.89 to 59.14) 182.50 (142.50-225.50) 181.00 (112.75-261.75) 0.575 0.33 (-39.66 to 40.32) Hemoglobin, (g/l) 124.68 (122.49-126.87) 115.68 (110.47-120.89) 0.001 9.14 (3.89 to 14.39) 129.50 (120.75-140.25) 117.50 (102.00-130.00) 0.001 11.13 (4.56 to 17.69) Blood biochemistry Alanine aminotransferase, (U/l) 31.34 (27.14-35.54) 40.23 (26.69-53.77) 0.493 -8.22 (-21.24 to 4.80) 23.00 (16.00-31.50) 21.00 (14.00-44.50) 0.853 -15.79 (-33.08 to 1.50) Aspartate aminotransferase, (U/l) 31.73 (28.18-35.28) 43.18 (33.98-52.40) 0.057 -11.22 (-20.33 to -2.10) 25.50 (19.25-32.50) 27.00 (20.00-45.00) 0.322 -15.86 (-27.53 to -4.19) Total bilirubin, (mmol/l) 11.66 (10.05-13.27) 16.38 (8.67-24.09) 0.244 -4.53 (-11.74 to 2.68) 10.00 (7.00-13.20) 8.00 (6.00-12.00) 0.097 -5.64 (-15.26 to 3.99) Blood urea nitrogen, (mmol/l) 6.13 (4.94-7.32) 7.98 (6.68-9.28) <0.001 -1.74 (-3.63 to 0.14) 4.54 (3.55-6.26) 6.22 (4.57-9.37) <0.001 -1.06 (-4.34 to 2.21) Serum creatinine, (mmol/l) 75.61 (65.03-86.20) 102.59 (74.80-130.39) 0.031 -16.72 (-53.61 to 20.17) 68.00 (55.50-85.25) 69.50 (56.00-93.00) 0.431 17.31 (-65.83 to 3.03) Creatine kinase, (U/l) 88.96 (68.60-109.31) 82.10 (46.13-118.07) <0.001 4.00 (-36.10 to 44.10) 67.50 (41.25-108.25) 32.00 (15.00-50.00) <0.001 17.49 (-25.39 to 60.36) Lactate dehydrogenase, (U/l) 258.48 (241.38-275.59) 320.18 (281.86-358.51) <0.001 -59.64 (-100.59 to -18.68) 233.00 (180.00-287.00) 256.00 (220.00-347.00) 0.088 -15.94 (-94.11 to 62.22) C-reactive protein, (mg/l) 29.08 (24.27-33.89) 76.21 (58.06-94.36) <0.001 -44.33 (-57.28 to -31.38) 18.08 (4.65-40.17) 50.50 (13.90-115.10) <0.001 -46.60 (-68.70 to -24.50) Computed tomography findings Unilateral pneumonia, n (%) 24 (10.8) 11 (11.5) 0.502 -0.01 (-0.08 to 0.07) 15 (20.8) 0 (0.0) <0.001 0.21 (0.11 to 0.30) Bilateral pneumonia, n (%) 110 (49.5) 60 (62.5) 0.022 -0.13 (-0.25 to -0.01) 30 (41.7) 65 (90.3) <0.001 -0.49 (-0.62 to -0.35) Multiple mottling and ground-glass opacity, n (%) 81 (36.5) 23 (24.0) 0.019 0.13 (0.02 to 0.23) 27 (37.5) 0 (0.0) <0.001 0.38 (0.26 to 0.49) PSM, propensity score matching After the routine blood laboratory tests, before the PSM analysis, the counts of white blood cells (WBCs) (6.13 vs 11.93 109/l, P <0.001, 95% CI, –11.12 to 0.41), neutrophils (4.42 vs 7.19 × 109/l, P <0.001, 95% CI, –3.65 to –1.59), and lymphocytes (1.20 vs 3.99 × 109/l, P-value = 0.001, 95% CI, –7.99 to 2.91) in the patients with COVID-19 were all lower than those in the patients with influenza A H1N1 virus. However, the counts of hemoglobin (124.68 vs 115.68 g/l, P <0.01, 95% CI, 3.89 to 14.39) and platelets (203.30 vs 183.84 × 109/l, P <0.01, 95% CI, 6.89 to 59.14) were higher in the patients with COVID-19. After the PSM analysis, the lymphocyte counts (1.00 vs 0.82 × 109/l, P-value = 0.030, 95% CI, –0.19 to 0.91) in the patients with COVID-19 were higher, and the platelet counts (182.50 vs 1181.00 × 109/l, P-value = 0.575, 95% CI, –39.66 to 40.32) did not obviously differ between the two groups. Regarding the blood biochemistry, there were no prominent differences in the counts of alanine transaminase, aspartate transaminase, total bilirubin, serum creatinine and C-reactive protein (CRP) between the two groups neither before or after the PSM analysis (P >0.05 for each). Blood urea nitrogen (4.54 vs 6.22 mmol/l, P <0.001, 95% CI, –4.34 to 2.21) in the patients with COVID-19 was lower than that in the patients with influenza, and compared with the patients with influenza, the patients with COVID-19 had higher counts of creatine kinase (67.50 vs 32.00 U/l, P <0.001, 95% CI, –25.39 to 60.36) after the PSM analysis. Computed tomography (CT) scans play a key role in the identification and diagnosis. As shown in Table 2, after the PSM analysis, multiple mottling and ground-glass opacities (37.5% vs 0.0%, P <0.001, 95% CI, 0.26 to 0.49) and unilateral pneumonia (20.8% vs 0.0%, P <0.001, 95% CI, 0.11 to 0.30) were more easily observed in the patients with COVID-19, and bilateral pneumonia (41.7% vs 90.3%, P-value = 0.022, 95% CI, –0.62 to –0.35) was more evident in the patients with influenza. Treatment and prognosis Doctors prescribed antiviral therapy to 97.2% of the patients with COVID-19 and 87.5% of the patients with influenza A H1N1 virus (P-value = 0.060, 95% CI, 0.01 to 0.18). In addition, both groups of patients were administered antibiotics, antifungal drugs, glucocorticoids, and immunoglobulins when necessary. Smaller proportions of patients with COVID-19 received antibiotics (62.5% vs 95.8%, P <0.001, 95% CI, –0.45 to –0.21) and antifungal drugs (2.8% vs 40.3%, P <0.001, 95% CI, –0.49 to –0.26), but a higher proportion underwent immunoglobulin treatment (31.9% vs 12.5%, P-value = 0.009, 95% CI, 0.06 to 0.33). There was no difference in glucocorticoid treatment (40.3% vs 52.8%, P-value = 0.181, 95% CI, –0.29 to 0.04) between the two groups. Regarding respiratory support, 16.7% of the patients with COVID-19 and 23.6% of the patients with influenza received mechanical ventilation (including noninvasive and invasive ventilation) (P-value = 0.406, 95% CI, –0.20 to 0.06). There were no prominent differences in the use of extracorporeal membrane oxygenation or continuous renal replacement therapy (both P-values = 0.243, 95% CI, 0.00 to 0.09, 95% CI, –0.09 to 0.00 for each). Regarding complications, the rate of shock (1.4% vs 2.8%, P-value = 1.000, 95% CI, –0.06 to 0.03) did not significantly differ, but fewer patients with COVID-19 developed ARDS (16.7% vs 40.3%, P-value = 0.003, 95% CI, –0.38 to –0.09) than patients with influenza. In our research, the mortality rate among the patients with COVID-19 was 0.0% between January 17 and March 10, 2020, whereas seven patients with influenza died from November 1, 2017 to March 31, 2018. The mortality rate among the patients with COVID-19 was lower than that among the patients with influenza (0.0% vs 9.7%, P <0.01, 95% CI, –0.17 to –0.03; shown in Table 3 ).Table 3 Complications, treatments, and prognosis of the patients with COVID-19 or influenza A H1N1 virus pneumonia. Table 3Characteristics Patients before PSM Patients after PSM COVID-19 (n = 222) H1N1 (n = 96) P-value Estimated difference, (95% CI) COVID-19 (n = 72) H1N1 (n = 72) P-value Estimated difference, (95% CI) Treatments Antivirus- treatment, n (%) 213 (96.0) 96 (100.0) 0.004 -0.04 (-0.07 to -0.01) 70 (97.2) 63 (87.5) 0.060 0.10 (0.01 to 0.18) Antibiotics- treatment, n (%) 116 (52.3) 92 (95.8) <0.001 -0.44 (-0.51 to -0.36) 45 (62.5) 69 (95.8) <0.001 -0.33 (-0.45 to -0.21) Antifungal- treatment, n (%) 2 (0.9) 36 (37.5) <0.001 -0.37 (-0.46 to -0.27) 2 (2.8) 29 (40.3) <0.001 -0.38 (-0.49 to -0.26) Glucocorticoids, n (%) 74 (33.3) 51 (53.1) <0.001 -0.20 (-0.32 to -0.08) 29 (40.3) 38 (52.8) 0.181 -0.13 (-0.29 to 0.04) Intravenous immunoglobulins therapy, n (%) 57 (25.7) 10 (10.4) 0.002 0.15 (0.07 to 0.24) 23 (31.9) 9 (12.5) 0.009 0.19 (0.06 to 0.33) Mechanical ventilation, n (%) 26(11.7) 24 (25.0) 0.004 -0.13 (-0.23 to -0.04) 12 (16.7) 17 (23.6) 0.406 -0.07 (-0.20 to 0.06) Extracorporeal membrane oxygenator, n (%) 5 (5.3) 0 (0.0) 0.194 0.02 (0.00 to 0.04) 3 (4.2) 0 (0.0) 0.243 0.04 (0.00 to 0.09) Continuous renal replacement therapy, n (%) 3 (1.4) 5 (5.2) 0.057 -0.04 (-0.09 to 0.01) 0 (0.0) 3 (4.2) 0.243 -0.04 (-0.09 to 0.00) Complications Acute respiratory distress syndrome, n (%) 26 (11.7) 37 (38.5) <0.001 -0.27 (-0.37 to -0.16) 12 (16.7) 29 (40.3) 0.003 -0.24 (-0.38 to -0.09) Shock, n (%) 3 (1.4) 3 (3.1) 0.371 -0.02 (-0.06 to 0.02) 1 (1.4) 2 (2.8) 1.000 -0.01 (-0.06 to 0.03) Prognosis Death, n (%) 0 (0.0) 11 (11.5) <0.001 -0.11 (-0.18 to -0.05) 0 (0.0) 7 (9.7) 0.020 -0.10 (-0.17 to -0.03) PSM, propensity score matching Multivariate logistic regression To further identify the differences between the two diseases, we performed a regression analysis of the significantly different data. A multivariate logistic regression was used to discern the main diverse factors of the two pneumonias. Compared with the patients with COVID-19, the patients with influenza tended to have a higher proportion of sputum production (P-value = 0.001, 95% CI, 4.41-247.37), fatigue (P-value = 0.005, 95% CI, 1.64-16.00), and shortness of breath (P <0.001, 95% CI, 3.92-40.79) and the counts of WBC (P-value = 0.004, 95% CI, 1.99-39.67), neutrophils (P-value = 0.01, 0.95% CI, 0.30-0.62), lymphocytes (P-value = 0.012, 95% CI, 0.27-0.64) and CRP (P-value = 0.015, 95% CI, 1.00-1.03) were higher. Moreover, the patients with COVID-19 were less likely to present with an abnormal chest CT (P-value = 0.015, 95% CI, 0.16-0.82; shown in Table 4 ).Table 4 Multivariate regression analysis of the patients with COVID-19 or influenza A H1N1 virus pneumonia. Table 4Variable Coefficient 95% CI P-value Sputum production 33.02 4.41-247.37 0.001 Fatigue 5.13 1.64-16.00 0.005 Shortness of breath 12.65 3.92-40.79 0.000 White blood cell 8.88 1.99-39.67 0.004 Neutrophils 0.14 0.30-0.62 0.010 Lymphocytes 0.13 0.27-0.64 0.012 C-reactive protein 1.01 1.00-1.03 0.015 Computed tomography findings 0.36 0.16-0.82 0.015 Discussion Our study used real-world data from Zhejiang Province, China, and some of the data were obtained from the same hospital. To ensure data quality, the data were collected by the same group of doctors, which was helpful for reducing data collection errors and provided good comparability. COVID-19 and influenza A H1N1 virus have both caused severe pandemics worldwide and have many identical epidemiological characteristics and clinical and imaging manifestations (Huang et al., 2020; Wang et al., 2020a). Hence, it is possible to misdiagnose COVID-19 as influenza A H1N1 viral pneumonia, especially in the early phase of COVID-19. However, COVID-19 is far more contagious than influenza A H1N1 viral pneumonia (Li et al., 2020; Wang et al., 2020b). The basic reproduction number (R0) of COVID-19 was estimated in the initial outbreak to be between 2.2 and 3.6 patients (Zhao et al., 2020). It is estimated that the R0 during the 2009 influenza outbreak in Mexico ranged from 1.3 to 1.7 (Yang et al., 2009). Moreover, elderly patients have more underlying diseases and more easily progress to the critical disease stage (Lian et al., 2020). Previous research has shown that both diseases have high morbidity and mortality in elderly individuals (Abdelrahman et al., 2020; Chen et al., 2020; Grasselli et al., 2020). Therefore, it is very important for clinicians to accurately identify the two diseases, especially in the elderly population. The purpose of the current research was to contrast the distinct clinical manifestations between hospitalized elderly patients infected with COVID-19 and H1N1 to provide some guidance for their differential diagnoses. The results of our research show that the proportion of males among patients with COVID-19 was lower than that among patients with influenza. However, most previous reports showed that the sex ratio of COVID-19 and patients with influenza were similar (Caruso et al., 2020; Han et al., 2020; Wang et al., 2020a). Moreover, the patients with COVID-19 showed lower proportions of underlying diseases than the H1N1 patients, which is similar to previous results (Huang et al., 2020), especially in heart disease, immunosuppression, and blood disease, which have differential diagnostic value. This may be associated with the inconsistent criteria for hospitalization. Therefore, a PSM analysis was used to reduce the imbalance to improve the reliability of our research. Patients with COVID-19 and influenza A H1N1 virus have many similar clinical symptoms, which makes it difficult to distinguish the two only through clinical manifestations before pathogen detection. A previous study reported that fever, fatigue, cough, expectoration, muscular soreness, and rhinorrhea were the most common symptoms of viral pneumonia, with gastrointestinal features, such as diarrhea, nausea, and vomiting (Chen et al., 2020; Han et al., 2020; Wang et al., 2020a). Our research found that patients with COVID-19 had lower rates of cough, expectoration, fatigue, and shortness of breath than those with influenza. Moreover, there was no differential diagnostic value in the digestive symptoms of the two groups in our research. However, previous studies have shown that the frequency of gastrointestinal symptoms is higher in patients with COVID-19 than in patients with influenza (Chan et al., 2020; Jin et al., 2020), which may be related to the damage by the SARS-CoV-2 infection to the gastrointestinal tract (Holshue et al., 2020). As the disease progresses, dyspnea, chest pain, and even ARDS and shock may appear. We found that patients with influenza more easily developed ARDS, which is consistent with previous research (Pormohammad et al., 2021). This result indicates that the clinical manifestation of COVID-19 is more concealed. In addition, we showed that the counts of WBCs, neutrophils, lymphocytes, blood urea nitrogen, and CRP in patients with COVID-19 were lower, and the counts of hemoglobin and creatine kinase were higher, which is consistent with previous studies (Bai et al., 2020; Kuang et al., 2021; Pormohammad et al., 2021; Yin et al., 2020). Because of the limited data, other immunological and inflammatory markers (e.g., erythrocyte sedimentation rate, tumor necrosis factor-α, interleukins [1, 6], coagulation parameters, prolonged prothrombin time, thrombocytopenia, and elevated d-dimer) were not evaluated. In further studies, the inclusion of these indicators may help better differentiate the two diseases. In addition, the most common radiologic abnormalities in patients with COVID-19 were bilateral changes on chest X‐rays (Pormohammad et al., 2021). In our research, most patients with COVID-19 and influenza showed abnormal imaging, with the highest proportion of lung changes on both sides. Multiple mottling and ground-glass opacities and unilateral pneumonia were more frequently observed in the patients with COVID-19, whereas bilateral pneumonia was more common in the patients with influenza pneumonia. Previous research showed the same results as follows: ground-glass opacities were the most abnormal pattern in patients with COVID-19, and bilateral consolidation with or without ground-glass opacities was more common in critical cases of influenza pneumonia on chest CT scans (Marchiori et al., 2011; Rohani et al., 2016; Shi et al., 2020). Therefore, these imaging characteristics may help distinguish the two diseases. Regarding treatment, patients with COVID-19 or influenza A H1N1 virus received a wide variety of treatments, including antibiotics, antifungals, glucocorticoids, and mechanical ventilation, when necessary. Compared with the definitive treatment measures for patients with influenza pneumonia (Uyeki et al., 2019), there is no solid evidence regarding the effectiveness of any remedy for COVID-19. In our research, patients with COVID-19 accepted more immunoglobulins than patients with H1N1. One study showed that glucocorticoids might reduce the death rate in patients with H1N1 (Li et al., 2017), whereas another study showed that glucocorticoids increased the mortality and secondary infection rates in patients infected with SARS and MERS and even complicated corticosteroid therapies in survivors (Russell et al., 2020). Therefore, the application of glucocorticoids should be cautiously assessed in patients with COVID-19, especially in elderly patients. Regarding complications, the incidence of shock did not differ, but fewer patients with COVID-19 developed ARDS, which is similar to a previous research showing that the incidence of ARDS was higher in influenza type A (31.5%; 95% CI 26-38%, P <0.001) than in COVID-19 (26.6%; 95% CI 18-38%, P <0.001) (Pormohammad et al., 2021). The death rate of COVID-19 was 0.00% in our research, which was lower than the fatality rate of H1N1 (9.7%, P-value = 0.020). However, other research found that the fatality rate of COVID-19 was higher (Pormohammad et al., 2021). The previously mentioned differences may be related to the following possible reasons. First, patients with influenza A H1N1 pneumonia may be compared with patients with COVID-19 not necessarily having pneumonia, and there would be an obvious difference in fatality. Second, compared with the management of influenza A H1N1, the pandemic of SARS-CoV-2 triggered more comprehensive life-saving health systems. This, based on the lack of understanding of the spread and severity of SARS-CoV-2 in the early stage, once the pathogen detection was positive, the patients were immediately isolated and admitted to the hospital for good care, which may also be the reason for the zero death rate of the patients with COVID-19 in this study. In summary, in our research, elderly patients infected with H1N1 had a poorer prognosis than patients with COVID-19. In conclusion, there are certain differences between elderly patients with COVID-19 and those with influenza. We report the differences in the clinical features and laboratory examinations between the two groups, including cough, expectoration, fatigue and shortness of breath, and diverse laboratory results. Dormant clinical symptoms of sputum production, fatigue, and shortness of breath, combined with lower counts of WBCs, neutrophils, lymphocytes, and CRP, are possible predictive factors of COVID-19 among elderly patients. In summary, the clinical manifestations of patients with COVID-19 are more concealed than those of patients with influenza in elderly individuals. More attention is needed for elderly individuals, especially those with underlying diseases, which can have a large impact on the prognosis. Because of the lower immune response and concealed clinical manifestations in elderly patients, prevention is still the most important strategy to protect them from infection. Our study had the following limitations. First, this study was retrospective, which may have an unavoidable bias because the data originated from two independent groups. Second, the information of the influenza and COVID-19 groups was collected from 1-year and 2-month time spans, respectively, which could introduce bias. Because of the limited data, other pathogens (e.g., bacteria and parasites) that may result in similar symptoms or poorer prognosis were not analyzed, which is a limitation of this study. The differences between elderly patients with COVID-19 or influenza were not obvious. Therefore, a larger sample size is required for relevant verification in the future. Appendix Supplementary materials Image, application 1 Image, application 2 Image, application 3 Declaration of competing interest The authors have no competing interests to declare. Funding This study received funding from the National Health Council Research Fund for “The study of severe mechanism and new treatment strategy for critical type of older patients with COVID-19” (WKJ-ZJ-2111). Ethical approval The study was approved by the institutional ethical committee. Acknowledgments The authors thank the Health Commission of Zhejiang Province, China for coordinating the data collection. Author contributions Yida Yang supervised the project.Yan Lv, Guodong Yu, Xiaoli Zhang designed the study and Yan Lv drafted the manuscript.Yan Lv, Guodong Yu, Xiaoli Zhang, Jueqing Gu, Chanyuan Ye, Jiangshan Lian, Xiaoqing Lu, Yingfeng Lu participated in the collection of data.All authors scrutinized the manuscript and approved the final version for publication. Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.ijid.2022.11.008. ==== Refs References Abdelrahman Z Li M Wang X. Comparative review of SARS-CoV-2, SARS-CoV, MERS-CoV, and influenza A respiratory viruses Front Immunol 11 2020 552909 Bai HX Hsieh B Xiong Z Halsey K Choi JW Tran TML Performance of radiologists in differentiating COVID-19 from non-COVID-19 viral pneumonia at chest CT Radiology 296 2020 E46 E54 32155105 Caruso D Zerunian M Polici M Pucciarelli F Polidori T Rucci C Chest CT features of COVID-19 in Rome, Italy Radiology 296 2020 E79 E85 32243238 Chan JF Yuan S Kok KH To KK Chu H Yang J A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster Lancet 395 2020 514 523 31986261 Chen N Zhou M Dong X Qu J Gong F Han Y Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study Lancet 395 2020 507 513 32007143 Echevarría-Zuno S Mejía-Aranguré JM Mar-Obeso AJ Grajales-Muñiz C Robles-Pérez E González-León M Infection and death from influenza A H1N1 virus in Mexico: a retrospective analysis Lancet 374 2009 2072 2079 19913290 Grasselli G Zangrillo A Zanella A Antonelli M Cabrini L Castelli A Baseline characteristics and outcomes of 1591 patients infected with SARS-CoV-2 admitted to ICUs of the Lombardy region, Italy JAMA 323 2020 1574 1581 32250385 Han R Huang L Jiang H Dong J Peng H Zhang D. Early clinical and CT manifestations of coronavirus disease 2019 (COVID-19) pneumonia AJR Am J Roentgenol 215 2020 338 343 32181672 Holshue ML DeBolt C Lindquist S Lofy KH Wiesman J Bruce H First case of 2019 novel coronavirus in the United States N Engl J Med 382 2020 929 936 32004427 Huang C Wang Y Li X Ren L Zhao J Hu Y Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China Lancet 395 2020 497 506 31986264 Jin X Lian JS Hu JH Gao J Zheng L Zhang YM Epidemiological, clinical and virological characteristics of 74 cases of coronavirus-infected disease 2019 (COVID-19) with gastrointestinal symptoms Gut 69 2020 1002 1009 32213556 Kuang PD Wang C Zheng HP Ji WB Gao YT Cheng JM Comparison of the clinical and CT features between COVID-19 and H1N1 influenza pneumonia patients in Zhejiang, China Eur Rev Med Pharmacol Sci 25 2021 1135 1145 33577070 Li H Yang SG Gu L Zhang Y Yan XX Liang ZA Effect of low-to-moderate-dose corticosteroids on mortality of hospitalized adolescents and adults with influenza A(H1N1)pdm09 viral pneumonia Influenza Other Respi Viruses 11 2017 345 354 Li Q Guan X Wu P Wang X Zhou L Tong Y Early transmission dynamics in Wuhan, China, of novel coronavirus-infected pneumonia N Engl J Med 382 2020 1199 1207 31995857 Lian J Jin X Hao S Cai H Zhang S Zheng L Analysis of epidemiological and clinical features in older patients with coronavirus disease 2019 (COVID-19) outside Wuhan Clin Infect Dis 71 2020 740 747 32211844 Marchiori E Zanetti G Fontes CA Santos ML Valiante PM Mano CM Influenza A (H1N1) virus-associated pneumonia: high-resolution computed tomography-pathologic correlation Eur J Radiol 80 2011 e500 e504 21035974 Pormohammad A Ghorbani S Khatami A Razizadeh MH Alborzi E Zarei M Comparison of influenza type A and B with COVID-19: a global systematic review and meta-analysis on clinical, laboratory and radiographic findings Rev Med Virol 31 2021 e2179 33035373 Rohani P Jude CM Chan K Barot N Kamangar N. Chest radiological findings of patients with severe H1N1 pneumonia requiring intensive care J Intensive Care Med 31 2016 51 60 24923491 Russell CD Millar JE Baillie JK. Clinical evidence does not support corticosteroid treatment for 2019-nCoV lung injury Lancet 395 2020 473 475 32043983 Shi H Han X Jiang N Cao Y Alwalid O Gu J Radiological findings from 81 patients with COVID-19 pneumonia in Wuhan, China: a descriptive study Lancet Infect Dis 20 2020 425 434 32105637 Uyeki TM Bernstein HH Bradley JS Englund JA File TM Fry AM Clinical practice guidelines by the Infectious Diseases Society of America: 2018 update on diagnosis, treatment, chemoprophylaxis, and institutional outbreak management of seasonal influenzaa Clin Infect Dis 68 2019 895 902 30834445 Wang D Hu B Hu C Zhu F Liu X Zhang J Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan, China JAMA 323 2020 1061 1069 32031570 Wang Y Wang Y Chen Y Qin Q. Unique epidemiological and clinical features of the emerging 2019 novel coronavirus pneumonia (COVID-19) implicate special control measures J Med Virol 92 2020 568 576 32134116 World Health Organization. Q&As on COVID-19 and related health topics. https://www.who.int/emergencies/diseases/novel-coronavirus-2019/question-and-answers-hub, 2020 (accessed 27 January 2021). Wu Z Harrich D Li Z Hu D Li D. The unique features of SARS-CoV-2 transmission: comparison with SARS-CoV, MERS-CoV and 2009 H1N1 pandemic influenza virus Rev Med Virol 31 2021 e2171 33350025 Yang Y Sugimoto JD Halloran ME Basta NE Chao DL Matrajt L The transmissibility and control of pandemic influenza A (H1N1) virus Science 326 2009 729 733 19745114 Yin Z Kang Z Yang D Ding S Luo H Xiao E A comparison of clinical and chest CT findings in patients with influenza A (H1N1) virus infection and coronavirus disease (COVID-19) AJR Am J Roentgenol 215 2020 1065 1071 32452731 Zhao S Lin Q Ran J Musa SS Yang G Wang W Preliminary estimation of the basic reproduction number of novel coronavirus (2019-nCoV) in China, from 2019 to 2020: a data-driven analysis in the early phase of the outbreak Int J Infect Dis 92 2020 214 217 32007643	36371013	PMC9718512	NO-CC CODE	2022-12-06 23:23:40	no	Int J Infect Dis. 2022 Dec 9; 125:278-284	utf-8	Int J Infect Dis	2,022	10.1016/j.ijid.2022.11.008	oa_other
==== Front Value Health Reg Issues Value Health Reg Issues Value in Health Regional Issues 2212-1099 2212-1102 International Society for Health Economics and Outcomes Research. Published by Elsevier Inc. S2212-1099(22)00193-5 10.1016/j.vhri.2022.10.006 Economic Evaluation Cost Analysis of Hospitalization for COVID-19 in a Brazilian Public Teaching Hospital Sousa Fernanda Ferreira de PgDip 1∗ Vieira Bruno Barbosa MSc 12 Reis Augusto da Cunha PhD 2 1 Juiz de Fora Federal University Hospital (HU-UFJF/Ebserh), Juiz de Fora, Minas Gerais, Brazil 2 Postgraduate Programme in Production Engineering and Systems (PPPRO), Federal Centre for Engineering Studies and Technological Education – CEFET/RJ, Rio de Janeiro, Brazil ∗ Correspondence: Fernanda Ferreira de Sousa, PgDip, Juiz de Fora Federal University Hospital, 2505 Dos Timbiras, Santo Agostinho, Belo Horizonte, Minas Gerais, Brazil 30140-063. 2 12 2022 3 2023 2 12 2022 34 4854 14 10 2022 © 2022 International Society for Health Economics and Outcomes Research. Published by Elsevier Inc. 2022 International Society for Health Economics and Outcomes Research Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objectives This study aimed to measure the hospitalization costs for suspect or confirmation cases of COVID-19 and aggregate knowledge in the costing process for future research on related topics. Methods A cost calculation model was applied using absorption costing technique. Cost was allocated into 2 main groups: hospitalization and personnel. The cost analysis considers the hospital perspective. This is a retrospective study whose data were collected between April and September 2020, equivalent to the first wave of the disease in Brazil. This research uses data from Hospital Information System, Brazilian Unified Health System (SUS) Cost Calculation and Management System, and SUS Hospital Information System. Results The average total cost per hospitalization was US$11 260 (R$63 504) for patients suspect or confirmed by COVID-19, and considering only detectable cases, the value was US$17 178 (R$96 886). The profile of hospitalized patients was male (51%), with a mean age of 59 years, white ethnicity (64%), and average length of stay of 9 days. Conclusions The amount approved by SUS for remuneration of hospitalizations by COVID-19 proved to be insufficient to cover the calculated costs. The results of this study collaborate to measure the expenditure of hospital institutions with COVID-19 hospitalizations, contribute as a parameter for health managers to identify whether the values attributed to hospitalization by COVID-19 by the SUS are adequate to cover all costs involved, and provide lessons learned on costs to the public health system in the event of new pandemics. Keywords Brazil cost analysis costs COVID-19 hospitalization public hospital ==== Body pmcIntroduction According to the World Health Organization, 80% of patients who contract the new coronavirus severe acute respiratory syndrome coronavirus 2 (COVID-19) have mild and uncomplicated symptoms, 15% progress to the hospitalization in infirmary beds, and 5% need an intensive care unit (ICU) for treatment.1 Analysis conducted by the Federation of Hospitals, Clinics and Laboratories of the State of São Paulo in August 2020 indicates that, in the first 6 months of the spread of the disease in Brazil, the public health system made available new 8764 adult and pediatric ICU beds and private and charitable hospitals opened 11 061 beds for the treatment of patients with coronavirus.2 Opening new hospitalization beds requires investment in infrastructure, equipment, and operational costs. Considering the context of limited public resources, measuring costs becomes essential for conducting economic health assessments and supporting managers’ decision making with comparative data on costs and benefits. This pandemic disease accounted for 25 million cases in Brazil until January 2022, being the third country with the highest number of cases and the second with the highest number of deaths.3 Due to this magnitude, the cost analysis is relevant because it contributes not only to measuring the financial expenditure but also as a parameter for Brazilian Unified Health System (SUS) managers to identify whether the value attributed to hospitalization for COVID-19 is adequate and sufficient to cover all costs involved. On March 18, 2020, the Brazilian government defined the amount of R$800 (Brazilian currency—real), equivalent to US$141.84 for each COVID-19 ICU daily rate.4 Nevertheless, 1 week later, the amount was changed to R$1600 (US$283.68).5 This episode raises the question of the sufficiency of resources to stimulate the opening of new ICU beds, making it possible to pay for the entire treatment of the disease. Thus, the present study aimed to measure the hospitalization cost for COVID-19 in a public teaching hospital from April to September 2020, the initial 6 months of infection in Brazil, equivalent to the first wave of the disease in the country. As specific objectives, this research identified the sociodemographic and hospital characteristics of hospitalized patients and compared the cost of hospitalization for COVID-19 with the remuneration provided by SUS. Methods A retrospective study with quantitative and qualitative analysis was developed. Data collection was performed with documental research and interview with the application of a semistructured questionnaire. The study was developed in a public teaching hospital located in the state of Minas Gerais, Brazil. This is a midsize hospital6 with 137 inpatient beds. To provide hospitalizations for COVID-19, in March 2020 clinical infirmary was adapted to create an isolation area, in this research called “COVID-19 area” composed of 8 ICU beds and 7 infirmaries totaling 15 COVID-19 hospitalization beds. Internal Regulation Nucleus (NIR) is the area where the admission of the patient in the hospital and the discharge checkout occur. Hospitalizations occur through 2 flows. The first refers to patients whose reason for admission is the suspicion or confirmation of COVID-19. In this case, the initial hospitalization occurs in the COVID-19 area, where reverse transcription polymerase chain reaction test will be conducted to confirm the infection. If the result is positive, the patient remains there, and if not, the patient is moved to an infirmary, where he will be treated according to a new diagnosis. The second flow comprises cases in which admission occurs for a cause other than COVID-19 and during hospitalization there is a manifestation of symptoms of the disease. Thus, initially, the patient will be admitted to a standard infirmary and, in the presence of symptoms, will be sent to the COVID-19 area for tests. Data Collection and Analysis The cost analysis considers the hospital perspective. Retrospective data were collected for the period between April and September 2020 (6 months). Three data sources were consulted: Hospital Information System studied, SUS Cost Calculation and Management System (APURASUS), and SUS Hospital Information System (SIHSUS). The Hospital Information System provided daily the frequency and length of stay of hospitalizations. The National Cost Program of the Brazilian Ministry of Health developed the APURASUS. The hospital implemented this system in July 2018. It is updated monthly through the monitoring of contracts and invoices executed. This system adopts the absorption costing technique that organizes resource data in cost centers according to the hospital structure. The cost data obtained by APURASUS were composed of 3 cost centers: (1) COVID-19 area; (2) Examinations (clinical analysis, pathological anatomy, and imaging); and (3) NIR. The analyzed cost is presented in Figure 1 .Figure 1 COVID-19 hospitalization cost estimation model. APURASUS indicates SUS Cost Calculation and Management System; HIS, Hospital Information System; ICU, intensive care unit; NIR, Internal Regulation Nucleus; SIHSUS, Brazilian Unified Health System Hospital Information System. Due to the absence of individualized cost centers that segregated infirmary and ICU beds in the COVID-19 area, interviews with specialists were conducted using a semistructured questionnaire to distinguish the medical-hospital materials and medicines used in ICU beds and infirmary beds (Appendix 1 in Supplemental Materials found at https://doi.org/10.1016/j.vhri.2022.10.006). Seven interviews with healthcare professionals of medical-hospital supplies and pharmaceuticals sites, which included doctors, nurses, and pharmacists. The number of interviews was sufficient because it included specialists of all categories involved in the process. For the other cost items, the separation between infirmary and ICU beds was performed based on apportionment criteria in the literature, such as square meters (m2), number of medical gas points, number of requisitions, patient per day, and number of employees.7 , 8 The sociodemographic variables (age, sex, ethnicity), the results of the COVID-19 tests, and the outcome of hospitalizations were obtained by crossing the number of Hospital Inpatient Authorizations (AIH) with data from the SIHSUS. To compare the calculated costs with the amount approved by the SUS for hospitalization of COVID-19, the data generated by the SIHSUS were consulted, where the invoiced amounts were presented by the hospital by the AIH obtained, which has the Sistema de Gerenciamento da Tabela de Procedimentos (SIGTAP)9 as a reference. To conduct this comparative analysis, only the hospitalizations of confirmed cases are considered as the value approved by the SUS for the treatment of COVID-19 infection, given that the suspect cases with negative test results, despite adding to the hospital costs with hospitalization by COVID-19, according to the Ministry of Health cannot be billed with the coronavirus infection treatment code.10 Data were recorded, tabulated, and analyzed, and inference of the results was performed using Microsoft Office Excel 365® (Microsoft, Redmond, WA). The exchange rate used to convert the values obtained was R$5.64 for every US$1.00, referring to September 2020. The ethics committee approved the research for Research with Human Beings of the hospital studied under protocol number 4,467,731. Costing Method The absorption costing method was adopted to calculate the hospitalization cost for suspect or confirmed COVID-19. According to Martins (2018), absorption costing is a technique where all production costs are allocated to products or services. Martins also highlights that absorption costing is methodologically accepted, in line with accounting principles.8 , 11 For the development of the method, the first step is to classify the identified costs between direct and indirect. Direct costs are quickly allocated to their respective cost centers, and indirect costs are allocated using apportionment criteria. The calculation of costs begins with the definition of the apportionment criteria and the design of the location map, formed by the cost items (eg, salaries and electricity) and the productive and auxiliary cost centers. After completing the cost allocation, the balances from the auxiliary cost centers are transferred to the production cost centers and, finally, to the products and services.12 , 13 The calculation of the total hospitalization costs comprises from admission to discharge of the patient. Thus, all costs were allocated to the costs of the NIR, COVID-19 area, and examinations and then aggregated into 2 groups: “hospitalization costs” and “personnel costs” according to the model presented in Figure 1. This cost calculation method was based on the model developed by Vieira et al14 adapted for clinical admissions. The costs of the COVID-19 area, examinations, and NIR were prorated based on the proportion of daily hospitalizations in the COVID-19 infirmary and ICU, proportion of examinations requested for suspect or confirmed hospitalizations concerning the total, and proportion of patients admitted to the hospital for hospitalization in the COVID-19 area, respectively. Then, the average cost of the daily infirmary and COVID-19 ICU, the average cost of examinations per hospitalization, and an average cost of NIR per hospitalization were calculated. The “hospitalization cost” results from the sum of the average cost of these areas multiplied by the average length of stay of hospitalization. The “hospitalization cost per procedure” comes from the multiplication between the “cost of hospitalization” and the frequency of hospitalization procedures that generated suspicion or confirmation of COVID-19. The “personnel cost” was estimated using the exact parameters of apportionment of the cost of hospitalization and is also obtained by absorption in the cost centers of the COVID-19 area, examinations, and NIR. Results Hospitalization Costs The results of applying the COVID-19 hospitalization cost calculation model are presented in Table 1 . A total of 156 hospitalizations were performed including suspected and confirmed cases of COVID-19 in the studied period, 64 of them with a positive diagnosis for the new coronavirus (detectable cases) and 92 negatives (undetectable cases). They represented 1426 hospitalizations days, of which 477 are in the infirmary days and 949 in the ICU days. In the evaluated period, the occupancy rate was 30% for the infirmary beds and 73% for the ICU beds. Cost calculation did not consider this parameter.Table 1 COVID-19 hospitalization average costs. Hospitalization procedure Infirmary ICU Exam NIR Personnel Total VS D a b c d e f g h i j k l m n n US$ (b/l) US$ (d/l) US$ (f/l) US$ (h/l) US$ (j/l) (b + d + f + h + j) US$ (m – l) % % % % % US$ US$ Detectable cases (COVID-19 positive) 64 711 4.1 1530 8.9 293 1.7 12 0.1 14 631 85.2 17 178 2403 −14 776 Undetectable cases (COVID-19 negative) 92 313 4.4 509 7.1 293 4.1 12 0.2 6014 84.2 7142 846 −6297 Hospitalization (total cases) 156 477 4.2 928 8.2 293 2.6 12 0.1 9549 84.8 11 260 1484 −9775 D indicates difference between the amount approved by Brazilian Unified Health System and the average total cost; ICU, intensive care unit; NIR, Internal Regulation Nucleus; VS, average Hospital Inpatient Authorizations value approved by Brazilian Unified Health System. Confirmed hospitalizations of COVID-19 were billed by procedure 03.03.01.022-3—treatment of infection by coronavirus—COVID-19, according to SIGTAP (2020). Hospitalizations of suspect cases whose subsequent diagnosis was negative to COVID-19 were reassessed to define the correct diagnosis of the patient and adequate billing. Considering the total number of confirmed or suspect cases of coronavirus (156 hospitalizations), an average total hospitalization cost of R$63 504 (US$11 260) (column l) was computed. Personnel cost is responsible for the most significant portion of the average total hospitalization cost, corresponding to 84.81% (R$53 858) (US$9549). Then there is a more significant share of the cost of ICU daily of R$5233 (US$927.84) (8.24%), infirmary daily of R$2688 (US$476.60) (4.23%), examinations of R$1655 (US$293.44) (2.61%), and NIR of R$70.00 (US$12.41) (0.11%). The average total cost per hospitalization of detectable cases of COVID-19 (64 occurrences), according to column a, resulted in R$96 886 (US$17 178.37) (column l) with an average length of stay of 15 days, of which 5 days in the infirmary and 10 in the ICU. The most represented cost item was personnel, accounting for 85.17%. The second installment comprises the cost of hospitalization, subdivided into ICU daily in the amount of R$8630 (US$1530.14) (8.91%); infirmary daily, calculated at R$4011 (US$711.17) (4.14%); examinations, estimated at 1655 (US$293.44) (1.71%); and NIR, equivalent to R$70 (US$12.41) (0.07%). Considering the other procedures (92 hospitalizations), which comprises suspects with the posterior negative test of COVID-19, the average total cost per hospitalization reached R$40 283 (US$7142.37) (column l), a difference of R$56 603 (US$10 035.99) (−58.42%) of the cost of treating positive cases of COVID-19. The sensitivity analysis was performed with the cost items (infirmary, ICU, examinations, NIR, and personnel) that make up the total cost of hospitalization for detectable cases of COVID-19. Consolidated results are presented in Table 2 considering a total period of hospitalization of 15 days, composed of daily in the infirmary and in the ICU with variations in the number of days spent by nurses between 0 (completely hospitalized in the ICU) and 15 days (completely hospitalized in the infirmary). The results showed that the total cost of hospitalization is more sensitive to variations in the item cost of personnel. In this case, the variation of −15% in the cost of personnel affects the cost of hospitalization by −12.86%, whereas the variation of 15% affects the cost by 12.97%.Table 2 Sensitivity analysis on the cost of hospitalizations for COVID-19 (detectable). Variables Minimum, % Maximum, % Cost variation – infirmary −1.79 2.18 Cost variation – ICU −2.03 1.86 Cost variation – examinations −0.34 0.45 Cost variation – NIR −0.10 0.21 Cost variation – personnel −12.86 12.97 ICU indicates intensive care unit; NIR, Internal Regulation Nucleus. The results obtained for the average cost of hospitalization by type of hospitalization (infirmary, ICU, or mixed) without differentiating detectable from undetectable cases are presented in Table 3 . The highest cost per hospitalization occurred in mixed cases, with a cost of R$80 284 (US$14 234.75). This result refers, on average, to 6 days in the infirmary and 6 days in the ICU. Patients hospitalized exclusively in ICU beds incurred an average hospitalization cost of R$68 687 (US$12 178.54) for 10 days. It was observed the lowest cost in exclusive infirmary hospitalizations, where the average length of stay was 6 days, and the cost reached R$41 701 (US$7393.79).Table 3 Costs by type of hospitalization for suspect and confirmed cases. Type of hospitalization Number of admissions Average length of hospital stay, days Average length of stay in the ICU, days Average cost of hospitalization including staff cost, US$ Infirmary 45 6 0 7394 ICU 76 0 10 12 179 Mixed (infirmary and ICU) 35 6 6 14 235 ICU indicates intensive care unit. The highest hospitalization costs are those with ICU daily. Comparing the cost of exclusive hospitalization in infirmary beds with the cost of exclusive hospitalization in the ICU, the cost difference is 65%. In the COVID-19 scenario, this type of hospitalization is highly requested, including in the hospital in question; the number of hospitalizations in ICU beds was higher than hospitalizations in infirmary beds, which, in turn, reflects in higher cost to the institution. Nevertheless, when analyzing the cost per day in exclusive infirmary and ICU beds, we have that the daily infirmary rate of R$6950 (US$1232.30) overlaps the cost of the ICU bed rate of R$6868 (US$1217.85). This is justified by the low occupancy rate of the infirmary beds (30%) in relation to the expenses evidenced, thus increasing the daily rate of these beds compared with the daily rate of the ICU, which had an occupancy rate of 73%. Sociodemographic and Hospital Variables The variables that characterize hospital admissions are presented in Table 4 . Considering the total number of hospitalizations, the age range of patients ranged between 15 and 98 years, with a mean and median value of 59 years and SD of 17 years. The frequency of individuals by age intervals to all samples is presented in Figure 2 . The mean and median values for detectable cases were 61 years, with SD of 16 years. This result characterizes the sample population as most adults and older people.Table 4 Sociodemographic and hospital variables. Variables Total Detectable (COVID-19 positive) Undetectable (COVID-19 negative) N = 156 n = 64 n = 92 Age, years, mean/median/SD 59/59/17 61/61/16 58/59/18 Male 51% 61% 45% White ethnicity 64% 67% 62% Average length of stay, days 9 15 5 ICU percentile days 71% 75% 68% Detected 41% - - Mortality 17% 22% 14% ICU indicates intensive care unit. Figure 2 Age distribution histogram. The sex representation of the sample proved to be balanced, with 51% of males. For the purposes of the analysis of this study about ethnicity, white and yellow was considered white, which represented a proportion of 64% of patients and 36% of nonwhite ethnicity (black, indigenous, and brown). The mean length of stay for suspect or confirmed COVID-19 resulted in 9 days (SD = 9), with 6 days of ICU beds and 3 days of infirmary. Of the total number of hospitalizations, 71% required at least one day in ICU beds, 41% of those admitted tested positive for COVID-19 and 17% of the patients died. Considering the average length of stay, patients detectable by COVID-19 were hospitalized by triple the time for undetectable patients. Comparison of the Average Hospitalization Cost With the Value From SUS According to the SIGTAP, the reference value to billing COVID-19 hospitalization consists of R$1500 (US$265.95) and each day of COVID-19 accredited ICU beds sums R$1600 (US$283.68) to hospitalization value.9 Thus, considering exclusively confirmed cases of COVID-19 (64 hospitalizations), the total cost calculated was R$6200.69 (US$1099.41) (Table 1, column l × column a), with an average hospitalization cost of R$96 886 (US$17 178.36) (Table 1, column l). The total amount approved by the SUS in AIH was R$867.31 (US$153.77) (Table 1, column m × column a), with an average value per hospitalization of R$13 552 (US$2402.84) (Table 1, column m). Comparing these data results in a total deficit of −R$5 333 387 (R$867 307-R$6 200 694) (−US$945 636) (US$153.77-US$1099.41) for the hospital and −R$83 334 (−US$14 776) (Table 1, column n) per hospitalization. From the perspective that hospitalizations with a negative diagnosis for coronavirus also incur costs of treating COVID-19 by the hospital and prevent other patients from being admitted to these beds, there is a total cost calculated for suspect cases and confirmed cases of COVID-19 (156 hospitalizations) of R$9 906 697 (US$1 756 506) (Table 1, column l × column a). If we also consider the amount approved by the SUS for suspect cases of COVID-19 billed with a code from another procedure after clinical reanalysis, the amount received from the SUS is R$1 306 054 (US$231 570) (Table 1, column m × column A), which makes the comparison with the average total cost of hospitalizations a deficit of −R$8 600 643 (−US$1 524 937). Despite the deficit presented, the accreditation of the hospital of 8 ICU beds for the exclusive treatment of COVID-19 from June 2020,15 comprising 4 months of the researched period, resulted in a fixed daily remuneration of R$1600 (US$283.68) per accredited ICU bed, totaling the amount received of R$1 536 000 (US$27 234.04) referring to the 8 ICU beds available for 120 days. Considering only the fixed amount paid for the ICU daily, it is still not possible to cover the costs with the 64 confirmed hospitalizations of coronavirus (R$6 200 694) (US$1 099 413). Discussion Restricting the analysis of the database to the 64 confirmed hospitalizations, one ICU daily rate costs an average of R$6860.38 (US$1216.38). Comparing the cost of the ICU daily rate calculated with the amount initially planned by the SUS of R$800.004 (US$141.84) to cover the ICU daily rate for the treatment of COVID-19, it is observed that the amount paid would result in a deficit for the institution of −R$6060.38 (−US$1074.54). Not even the change in the daily ICU rate to R$1600.005 (US$283.69) was sufficient to cover the deficit generated by comparing the cost of the ICU daily rate calculated with the SUS reimbursement amount of −R$5260.38 (R$1600.00-R$6860.38) (−US$932.69) (US$283.69-US$1216.38). The financed amount represents only 23% of the calculated cost. In Brazil, the company Planisa, which specialized in health management, developed a study with 7 Brazilian hospitals of reference for the care of COVID-19 to measure the cost of hospitalization for coronavirus in infirmary and ICU beds using the absorption costing method.16 Each daily in the infirmary costs R$1139 (US$201.95) and R$2234 (US$396.10) in an ICU bed. Although the present study uses the same costing technique as the research developed by Planisa, the values obtained proved to be higher, R$6151.81 (US$1090.75) each day in the infirmary and R$6860.38 (US$1216.37) in the ICU. One of the possible explanations for this discrepancy is the volume of resources spent on labor, which in the survey represents 85.17% of the calculated cost. A second point to be noted concerns the fixed working hours of public servants, regardless of the volume of care. The Unicamp from Campinas and Hospital das Clínicas from São Paulo (HCFMUSP) conducted studies on the hospitalization cost for COVID-19 in institutions providing exclusive assistance to SUS users. Unicamp calculated the cost of R$2500 (US$443.26) to R$3000 (US$531.91) per day in ICU.17 The HCFMUSP developed a study in which it analyzed 3254 patients admitted to the hospital from March to June 2020 and observed the average hospitalization cost of R$68 100 (US$12 074.47) for less complex cases and an average expense of R$109 000 (US$19 326.24) for hospitalization in ICU beds.18 The HCFMUSP is similar to the hospital studied because both are teaching hospitals, so that the values found in this study are the ones that most resemble the present work, where the value of hospitalization for COVID-19 in the infirmary was R$57 001 (US$10 106.56) and the average value of hospitalization in the ICU reaches R$104 519 (US$18 531.74). These results are in line with the literature, which finds that the costs of teaching hospitals are higher than those that do not include this function.19 , 20 Comparing the current study with results of some international studies of hospitalization costs for the COVID-19, available in the Web of Science, Scopus, and Google Scholar databases, it is observed that most of the articles analyzed had a hospitalization cost lower than that obtained in this study (US$17 178). The study by Bain et al21 presented cost of US$20 540, higher than that found in this study. The author analyzed the hospitalization cost for COVID-19 in Europe in diabetic patients and those without diabetes (the cost used in this study refers to the hospitalization of patients without diabetes). Their study retrieved information from publicly available cost data from Denmark, France, Spain, and United Kingdom, and for the remaining 28 countries, costs were estimated using the Eurostat cost index. The hospitalization cost that is closest to this study was that realized in Saudi Arabia by Khan et al22 (US$16 283—average of the costs of severe and critical cases), which aimed to assess the survival of hospitalized patients with coronavirus in distinct groups and used the microcosting technique to estimate the direct medical costs associated with hospitalization. In contrast, the cost of hospitalization that was farthest from that measured in this study was that of Lee et al,23 with a cost of US$2192, which represented only 13% of the cost calculated in this research. The cost of hospitalization was calculated from data from the National Health Insurance System of Korea and refers to the average length of stay of 10.38 days. This study analyzed only pediatric hospitalizations, which may contribute to the lower cost observed. The research of Jin et al24 analyzed the economic cost of COVID-19 in China and in 31 administrative regions. The calculated cost of this hospitalization was US$3193. This study was conducted using a bottom-up technique to measure costs and stands out for presenting, in addition to medical costs, the social costs associated with COVID-19. The social and health costs estimated by the study totaled US$0.62 billion and US$383.02 billion, respectively. In addition to the studies mentioned earlier, 2 other studies developed in the United States found costs for hospitalizations for COVID-19 with higher values (US$14 36625 and US$12 04626) than eastern countries, a study in Indonesia found the cost of this hospitalization at US$5347,27 and another study in Iran estimated the cost at US$3755.28 All these had costs lower than the cost determined in this research. In general, it was observed that the reviewed international studies mentioned earlier were developed with different objectives. Regarding the method, a great variation was observed in the methodology applied, and in some studies, the costing technique used was not specified. Most of the studies identified refer to Asian countries, which refer mainly to the continent and country of origin of the disease. The highest costs belonged to the United States and European countries, and the lowest costs were present in Asian countries, as observed in China, Korea, Iran, and Indonesia. Conclusion The main objective of this study was to measure the cost of hospitalization for COVID-19 in a public teaching hospital. Therefore, an average total cost per hospitalization of R$63 504 (US$11 259.57) was obtained for patients suspect or confirmed by COVID-19. Considering only detectable cases, there is an average total cost per hospitalization of R$96 886 (US$17 178.37). The amount approved by the SUS covered 14% of the cost, resulting on a deficit of −R$83 334 (US$14 776) per hospitalization. Therefore, this amount was sufficient to cover infirmary, ICU, examinations, and NIR costs, not covering personnel costs, which are the main ones. The cost results provide important information for future economic assessments of COVID-19 and can be used as a parameter for SUS managers to identify whether the value attributed to hospitalization for COVID-19 in SIGTAP is adequate and sufficient to cover all the costs involved. This research has the limitations of conducting the cost study in a single hospital, difficulty in formatting cost centers, the number of hospitalizations, and the period of 6 months in which they were evaluated. Such limitations may make it difficult to generalize the results, so the following suggestions are proposed for future investigations: apply the method used in referral hospitals to combat the coronavirus, analyze the costs of COVID-19 over a more extended period of data, adjustments in cost centers be made preliminarily in the institutions to better determine costs, and investigate how Brazilian legislation can affect health financing, considering that real expenses have the potential to exceed the reimbursement provided by the SUS. Article and Author Information Author Contributions:Concept and design: Ferreira de Sousa, Vieira Acquisition of data: Ferreira de Sousa Analysis and interpretation of data: Ferreira de Sousa, Vieira, Reis Drafting of the manuscript: Ferreira de Sousa, Vieira, Reis Critical revision of the paper for important intellectual content: Ferreira de Sousa, Vieira, Reis Statistical analysis: Ferreira de Sousa Administrative, technical, or logisticsupport: Vieira Supervision: Vieira, Reis Conflict of Interest Disclosures: The authors reported no conflicts of interest. Funding/Support: The authors received no financial support for this research. Supplemental Material Appendix.1 Supplementary data associated with this article can be found in the online version at https://doi.org/10.1016/j.vhri.2022.10.006. ==== Refs References 1 Oxygen sources and distribution for COVID-19 treatment centres: interim guidance. World Health Organization https://www.who.int/publications/i/item/oxygen-sources-and-distribution-for-covid-19-treatment-centres 2 Federation of Hospitals, Clinics and Laboratories of the State of São Paulo. Almost 20 thousand ICU beds for covid-19 cases were opened in the pandemic. FEHOESP https://fehoesp360.org.br/noticia/7030/quase-20-mil-leitos-de-uti-para-casos-de-covid-19-foram-abertos-na-pandemia 3 WHO coronavirus (COVID-19) dashboard. World Health Organization https://covid19.who.int/table 4 Includes beds and procedures in the Table of Procedures, Medicines, Orthoses, Prostheses and Special Materials (OPM) of the Unified Health System (SUS), for the exclusive care of patients with COVID-19. Brasil Ministry Health Ordinance. https://www.in.gov.br/web/dou/-/portaria-n-237-de-18-de-marco-de-2020-249024782. Accessed October 30, 2020. 5 Authorizes the authorization of beds in the Adult and Pediatric Intensive Care Unit for exclusive care of patients with COVID-19. Brasil Ministry Health Ordinance. https://www.in.gov.br/web/dou/-/portaria-n-568-de-26-de-marco-de-2020-∗-251705696. Accessed October 30, 2020. 6 Filho A.N. Barbosa Z. The role of the hospital in the Health Care Network. Consensus 6 11 2014 42 49 7 Falk J.A. Cost Management for Hospitals: concepts, methodologies and applications 2001 São Paulo Atlas 8 Martins E. Cost Accounting. 11 2018 São Paulo Atlas 9 Brazil. SIGTAP - SUS table management system of procedures, medicines and OPM http://sigtap.datasus.gov.br/tabela-unificada/app/sec/inicio.jsp 10 Technical Guidelines for Operationalization of SIH During the State of Public Health Emergency by Coronavirus. FEHOSP http://fehosp.com.br/app/webroot/files/circulares/307030badf1c79fa6e5f42ec5d942c68.pdf 11 Dubois A. Kulpa L. Souza L.E. Cost Management and Pricing 4th ed. 2019 São Paulo Atlas 12 Souza A. Clemente A. Cost Management: strategic operational applications: exercises solved and proposed using Excel 2nd ed. 2011 São Paulo Atlas 13 Crepaldi S.A. Crepaldi G.S. Cost Accounting 6th ed. 2018 São Paulo Atlas 14 Vieira B.B. da Cunha Reis A. de Paiva L.A. Plácido E.C.R. de Sousa F.F. An integrated cost model based on real patient flow: exploring surgical hospitalization Healthcare 10 8 2022 1458 36011115 15 Ordinance N. 1.516, June 09, 2020. Enables Beds in Intensive Care Units – Adult ICU Type II – COVID-19 of Health Establishments and Establishes a Resource for the Maintenance Block of Public Health Actions and Services – Coronavirus Group (COVID-19), Made Available to the State of Minas Gerais and Municipalities. Brazil Ministry of Health. https://www.in.gov.br/web/dou/-/portaria-n-1.516-de-9-de-junho-de-2020-261040465. Accessed March 1, 2021. 16 COVID-19 Median ICU daily cost is R$2,234. Planisa https://planisa.com.br/site/covid-19-custo-mediano-de-diaria-em-uti-e-de-r-2-234/ 17 Find out how much it costs to face Covid-19 here. UNICAMP https://www.unicamp.br/unicamp/coronavirus/quanto-custa 18 Collucci C. Age and previous illnesses increase the cost of hospitalization for COVID-19 by up to 50%. Folha de São Paulo https://www1.folha.uol.com.br/equilibrioesaude/2021/01/idade-e-doencas-previas-aumentam-em-ate-50-custo-de-internacao-por-covid-19.shtml 19 López-Casasnovas G. Saez M. The impact of teaching status on average costs in Spanish hospitals Health Econ 8 7 1999 641 651 10544329 20 Dallora M.E.L.V. Forster A.C. The Importance of Cost Management in Teaching Hospitals - Theoretical Considerations Med (Ribeirao Preto) 41 2 2014 135 21 Bain S.C. Czernichow S. Bogelund M. Costs of COVID-19 pandemic associated with diabetes in Europe: a health care cost model Curr Med Res Opin 37 1 2021 27 36 33306421 22 Khan A.A. AlRuthia Y. Balkhi B. Survival and estimation of direct medical costs of hospitalized COVID-19 patients in the Kingdom of Saudi Arabia Int J Env Res Pub Health 17 20 2020 7458 33066327 23 Lee J.K. Kwak B.O. Choi J.H. Choi E.H. Kim J.H. Kim D.H. Financial burden of hospitalization of children with coronavirus disease 2019 under the national health insurance service in Korea J Korean Med Sci 35 24 2020 e224 32567260 24 Jin H. Wang H. Li X. Economic burden of COVID-19, China, January-March, 2020: a cost-of-illness study Bull World Health Organ 99 2 2021 112 124 33551505 25 Bartsch S.M. Ferguson M.C. McKinnell J.A. The potential health care costs and resource use associated with COVID-19 in the United States: a simulation estimate of the direct medical costs and health care resource use associated with COVID-19 infections in the United States Health Aff 39 6 2020 927 935 26 Di Fusco M. Shea K.M. Lin J. Health outcomes and economic burden of hospitalized COVID-19 patients in the United States J Med Econ 24 1 2021 308 317 33555956 27 Jati SP, Budiyanti RT, Ginandjar P, et al. Cost estimates related to COVID-19 treatment in Indonesia: what should be concerned? Paper presented at: The 5th International Conference on Energy, Environmental and Information System (ICENIS 2020); November 10, 2020; Jawa Tengah, Indonesia. 28 Darab M.G. Keshavarz K. Sadeghi E. Shahmohamadi J. Kavosi Z. The economic burden of coronavirus disease 2019 (COVID-19): evidence from Iran BMC Health Serv Res 21 1 2021 1 7 33388053	36469989	PMC9718517	NO-CC CODE	2022-12-06 23:23:40	no	Value Health Reg Issues. 2023 Mar 2; 34:48-54	utf-8	Value Health Reg Issues	2,022	10.1016/j.vhri.2022.10.006	oa_other
==== Front eBioMedicine EBioMedicine eBioMedicine 2352-3964 The Authors. Published by Elsevier B.V. S2352-3964(22)00537-0 10.1016/j.ebiom.2022.104355 104355 Comment How sepsis parallels and differs from COVID-19 Herminghaus Anna a Osuchowski Marcin F. b∗ a Department of Anaesthesiology, University of Duesseldorf, Duesseldorf, Germany b Ludwig Boltzmann Institute for Traumatology the Research Centre in Cooperation with AUVA, Vienna, Austria ∗ Corresponding author. Ludwig Boltzmann Institute for Traumatology the Research Centre in Cooperation with AUVA, Donaueschingenstrasse 13, A-1200 Vienna, Austria. 2 12 2022 12 2022 2 12 2022 86 104355104355 24 10 2022 24 10 2022 © 2022 The Authors 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc “Nothing in life is to be feared. It is only to be understood.” Marie Curie-Skłodowska A recent meta-analysis1 indicated that the majority of severely ill COVID-19 patients (78%) met the Sepsis 3.0 criteria for sepsis/septic shock with acute respiratory distress syndrome (ARDS), as the most frequent organ dysfunction (88%). Thus, it is suggestive that COVID-19 in hospitalized patients should be inherently considered as sepsis. This perception is not widely shared, and varying views of COVID-19 and sepsis syndromes cloud the understanding of their pathophysiology. Given that Sepsis 3.0 definition2 is relatively inclusive, it is imperative to understand the similar and distinctive phenotypic features of both conditions, to maximize treatment benefits and reduce harm. In the context of COVID-19, ARDS is a prominent contention element. Namely, to what extent a SARS-CoV-2-induced ARDS is comparable/dissimilar to a bacterial-origin ARDS. Both ARDS forms are paralleled by a decreased lung compliance, inflammation, hypoxemia, hypercarbia and endothelial injury. Conversely, COVID-19 ARDS features a robust alveolar thrombosis accompanied by an excessive fibro-proliferative lung tissue remodeling.3 Another phenomenon for COVID-19 respiratory failure, not observed in other etiologies, is the so-called “silent hypoxemia” (a critically low pO2 accompanied by mild dyspnea).4 Silent hypoxemia is especially detrimental, as it delays timely therapeutic management and facilitates multi-organ failure. Coagulopathy is frequent in both illnesses, yet COVID-19 derangements are far from the typical disseminated intravascular coagulation (DIC) encountered in bacterial sepsis.4 COVID-19-associated coagulopathy features highly elevated circulating fibrinogen, high D-dimers accompanied by a typically non-apparent thrombocytopenia and mildly affected clotting times. Both of those new manifestations constitute a medical terra incognita and require charting of new therapeutic maps. Nearly three years of research have shed some light on the intricacies of the immuno-inflammatory response to COVID-19. It is apparent that the captivating “cytokine storm” label should be downgraded to a “cytokine drizzle”, as the levels of circulating proinflammatory cytokines (e.g. IL-6, IL-8, TNF) are at a fraction of the concentrations recorded in an non-SARS-CoV-2 sepsis/septic shock.5 In contrast to the systemic response, the lung compartment in the severely ill COVID-19 patients typically undergoes a robust, protracted inflammation. At the COVID-19 management level, there is no dominant break-through strategy, which would dramatically differ (apart from the antimicrobials/antivirals) from the established sepsis treatment bundle by the US National Institutes of Health guidelines. One important exception is the dissimilar efficacy of glucocorticoids (GCs). While the current sepsis guidelines feature a weak recommendation for GCs, their use for severe SARS-CoV-2 pneumonia is unequivocally beneficial. Biological mechanisms behind this disparity should be elucidated as the underlying reasons may galvanize a renaissance of GCs in bacterial sepsis and critical care in general.6 There is a striking parallel between bacterial sepsis and COVID-19 phenotypes: the long-term sequelae. In both patient groups, the hospital discharge does not equal full recovery, but it is frequently followed by protracted, incapacitating consequences. While in bacterial sepsis, the post-discharge complications are referred to as post-sepsis syndrome and/or Persistent Inflammation, Immunosuppression, and Catabolism Syndrome (PICS), whereas, in SARS-CoV-2 infected patients, they are known as “long-COVID”. Long-COVID is not very different from post-sepsis syndrome. The most common persistent symptoms include fatigue, muscle pain, poor sleep, cardiac and cognitive disturbances (e.g. arrhythmias, short-term memory loss).7 Remarkably, a new, troubling difference exists: unlike in sepsis, long-COVID is frequently diagnosed in mildly SARS-CoV-2 infected patients (i.e. no hospital stay).8 The presence of the “long-phenotype” in both illnesses strongly indicates a severe and protracted deregulation of the immune-inflammatory (with clear immunosuppression features) and organ homeostasis. In the context of the slowly subsiding severe COVID-19 manifestations, we should re-focus on the long-term sequalae to evaluate a potential risk of increase in chronic debilitations in individuals, repeatedly exposed to the virus, as SARS-CoV-2 becomes seasonal/endemic. A comparison of the pre-clinical research in sepsis and COVID-19 brings several important lessons. A subjective (and largely undeserved) disappointment in pre-clinical bacterial sepsis studies has not been shared by COVID-19 modeling. On the contrary, despite logistic challenges, modeling of SARS-COV-2 infection has demonstrated its robust utility. One of the key advantages of pre-clinical COVID-19 research is the rich palette of species (including non-human primates), while >90% of bacterial sepsis studies are performed in mice and rats. A multi-species approach considerably enhances reproducibility and translatability concurrently reducing idiosyncratic findings. Animal COVID-19 models were well-predictive of both successes (e.g. anti-SARS-CoV-2 monoclonal antibodies, remdesivir, vaccines) and failures (e.g. hydroxychloroquine, lopinavir/ritonavir) of clinically-tested substances.9 Given intuitive drive for benefits, the latter should not be underappreciated; “negative” findings hold a valuable life-/time and cost-saving potential. Notably, anti-TNF treatment in a clinically relevant mouse model of cecal ligation and puncture sepsis predicted failure of that therapy three years before the failed clinical trials.10 A clear pre-clinical parallel for sepsis and COVID-19 models exist: they both can be employed to cover identical research niches: i) mild-to-severe disease phenotypes, ii) defined cohort targeting, iii) selected pathophysiological insights (e.g. compartmentalization of responses). Furthermore, long-term sequalae can be effectively investigated in both bacterial sepsis and COVID-19 models. Given that bacterial sepsis and COVID-19 parallels heavily intertwine with contrasts, it is critical to carefully dissect them into defined, manageable pieces of pathophysiological evidence (eg, by a given system, compartment) before any further therapeutic action is recommended. Equally important is that we avoid a reflexive transplantation of ready-to-use preconceptions (eg, “cytokine storm”) from an existing disease while dealing with any new entity. Well-designed pre-clinical studies can aid in a translationally valid verification of virtually any of the above concepts at the fraction of time/costs required for a clinical trial execution. Contributors MFO and AH conceptualised the content/format of the commentary together; MFO provided the final editing. Both authors read and approved the final version of the manuscript. Declaration of interests None to declare. ==== Refs References 1 Karakike E. Giamarellos-Bourboulis E.J. Kyprianou M. Coronavirus disease 2019 as cause of viral sepsis: a systematic review and meta-analysis Crit Care Med 49 12 2021 2042 2057 34259663 2 Singer M. Deutschman C.S. Seymour C.W. The Third International Consensus definitions for sepsis and septic shock (Sepsis-3) JAMA 315 8 2016 801 810 26903338 3 Batra R. Whalen W. Alvarez-Mulett S. Multi-omic comparative analysis of COVID-19 and bacterial sepsis-induced ARDS PLoS Pathog 18 9 2022 e1010819 4 Osuchowski M.F. Winkler M.S. Skirecki T. The COVID-19 puzzle: deciphering pathophysiology and phenotypes of a new disease entity Lancet Respir Med 9 6 2021 622 642 33965003 5 Stolarski A.E. Kim J. Zhang Q. Remick D.G. Cytokine drizzle-the rationale for abandoning “Cytokine Storm” Shock 56 5 2021 667 672 34652339 6 Winkler M.S. Osuchowski M.F. Payen D. Torres A. Dickel S. Skirecki T. Renaissance of glucocorticoids in critical care in the era of COVID-19: ten urging questions Crit Care 26 1 2022 308 36209188 7 Huang L. Li X. Gu X. Health outcomes in people 2 years after surviving hospitalisation with COVID-19: a longitudinal cohort study Lancet Respir Med 10 9 2022 863 876 35568052 8 Puntmann V.O. Martin S. Shchendrygina A. Long-term cardiac pathology in individuals with mild initial COVID-19 illness Nat Med 28 10 2022 2117 2123 36064600 9 Winkler M.S. Skirecki T. Brunkhorst F.M. Bridging animal and clinical research during SARS-CoV-2 pandemic: a new-old challenge EBioMedicine 66 2021 103291 10 Osuchowski M.F. Remick D.G. Lederer J.A. Abandon the mouse research ship? Not just yet! Shock 41 6 2014 463 475 24569509	36470836	PMC9718536	NO-CC CODE	2022-12-14 23:52:20	no	eBioMedicine. 2022 Dec 2; 86:104355	utf-8	eBioMedicine	2,022	10.1016/j.ebiom.2022.104355	oa_other

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.