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 Int J Syst Assur Eng Manag International Journal of System Assurance Engineering and Management 0975-6809 0976-4348 Springer India New Delhi 1822 10.1007/s13198-022-01822-y Original Article Recognizing elderly peoples by analyzing their walking pattern using body posture skeleton http://orcid.org/0000-0002-1897-1660 Singh Dushyant Kumar [email protected] MNNIT Allahabad, Prayagraj, India 1 12 2022 18 3 2 2022 4 7 2022 26 11 2022 © The Author(s) under exclusive licence to The Society for Reliability Engineering, Quality and Operations Management (SREQOM), India and The Division of Operation and Maintenance, Lulea University of Technology, Sweden 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 increasing age of the population has become a significant concern internationally. During the COVID-19 pandemic situation, it has been seen that the most sensitive and affected class of the population is the class of Elder’s. It is therefore necessary to track the movement and behavior of the old persons. This kind of monitoring could help them in providing assistance in their needy time. Our objective is to develop an approach to classify elderly people using skeleton data for their assistance. OpenPose algorithm is used here to detect human skeletons (joint positions) from the video sequences. OpenPose algorithm with a sliding window of size ‘N’ is used to achieve a real-time posture recognition framework. Posture features from each extracted skeleton are then used to build a classifier for recognizing elderly people. We also introduce here a new dataset that includes old person walk and young person walk video’s. The experimental outcomes reveal that the proposed method has achieved up to 98.45% training accuracy and 96.16% testing accuracy for deep feed-forward neural network (FFNN) classifier. This asserts the effectiveness of the approach. Keywords Feed forward neural network (FFNN) OpenPose Human detection Posture recognition Skeleton ==== Body pmcIntroduction According to the 2011 Population Survey (Velayutham et al. 2016), approximately 104 million people were aged 60 years or older. Another report (Singh and Kaur 2018) reveals that the elderly population in India is supposed to increase to 173 million by 2026. In 2010, Central Intelligence Agency World Factbook (Agency et al. 2010) recorded the elderly dependency ratio of 9.8 of the elderly people of age 65 or above per 100 working people between ages 15–64. Increased aging dependency ratio puts additional pressure on caregivers. Since elderly individuals are more prone to diseases, their health care therefore is a serious concern (Lubeek et al. 2017). The work under this manuscript focuses on enhancing the quality of life of elderly people and preventing them from any unwanted accidents related to their health status. This can be done through real-time monitoring of elder ones either physically by ourselves or through some digital technology in case of our absence. The two prevalent tools of digital technology that can be used for monitoring any movement of elderly people are “Sensors based assistive device(s)” and/or “Computer vision based techniques”. The first way is to deploy wearable devices in different body parts of the person to detect lively activities of daily routine. These devices are usually placed behind the ear lobe, under the axilla, around the wrist, or at the waist. It is a tiresome job to wear them all the time. The problem mostly arises from improper use of such detectors, as people mostly forget to wear them. This behavior of humans limits the performance of these detectors. On the other hand, the computer vision system tries to excerpt some noticeable postures and features from the videos of the elderly people’s movements and analyze these for activity-like patterns. The recognized activity can then help assessing the health condition of elder ones. The computer vision system attracts a great deal of interest, specifically in assistive technology for elderly people. Computer vision systems perform several tasks like detecting human presence and recognizing elderly people based on their movement. This could ensure the safety and comfort of all the elderly people living on their own. The primary health problem of elderly people is that they are prone to fall, which leads to very long-term injuries, fear and even death in some cases. Falling incidents lead to fractures and psychological consequences that lessen their independence. According to the study (Visutsak and Daoudi 2017), 28–34% of older persons fall at least once a year, of which 40–60% of those falls resulting in injury. Therefore, this paper proposed a practical assistive-technology based surveillance system to identify young and old people’s in real-time video sequences. One of the popular pose estimation techniques, namely the OpenPose algorithm (Cao et al. 1812), is used here to derive skeletons in terms of posture joints. Since the body movements of younger and older people are different, they can be identified as younger or older based on joint movements. The list of the contributions presented in this article is as follows:We present an activity recognition framework that analyzes 2D skeletal data and classifies related actions. We present skeleton pre-processing and feature extraction methods to extract relevant features from a sequence of skeletal data. We perform a variety of experiments on a synthesized video dataset to access the walking patterns of elderly people. Computer vision based proposed assistance mechanism is beneficial for the caregivers to take care of older people at homes or hospitals. The remaining of the article is organized as follows. After the brief literature discussion in Sect. 2, Sect. 3 describes the proposed methodology for recognizing older people based on their movements. In addition, this section also includes a brief discussion of the various techniques involved as part of the proposed methodology. The experiments and discussion are presented in Sect. 4. Finally, the last section involves the conclusion of the paper. Related works Nowadays, researchers are widely adopting sensor-based and vision-based approaches to recognize human acts in real-time scenarios. The sensor-based approach uses wearable sensors like accelerometers, gyroscopes, etc., to track an individual’s activity. In contrast, vision-based approaches mostly use a Convolutional neural network (CNN) to classify human activities in real-time video sequences. Convolution neural network (CNN) (Ansari and Singh 2021; Ojha et al. 2017) is an influential innovation in computer vision. A CNN is a deep learning approach that automatically learns spatial hierarchies of features through back propagation by using input, Convolutional, pooling, fully connected, and output layers. Some of the researches based on sensors and vision are discussed as follows: Pienaar and Malekian (2019) suggested human activity recognition (HAR) system to track the basic activities of a person by analyzing raw sensor data. Here, an open-source dataset introduced by Wireless Sensor Data Mining Lab is used that consists of six activity levels. The outcomes show that the model has achieved an accuracy of more than 94. Chernbumroong et al. (2013) suggested a HAR system to detect activities of daily living by analyzing wrist-worn sensor data. The data produced by sensors is first pre-processed and then passed to the feature extraction module to extract relevant features. Next, the extracted features are used to build a classifier to categorize the involved activities. This method shows proficient outcomes in terms of accuracy up to 94%. Putra and Yulita (2019) proposed HAR model based on the bed-wake gesture. Here, a multilayer perceptron network was used to predict activity based on sensor readings. This work has achieved proficient results in terms of accuracy of up to 90.17% for MLP and 84.46% for Naïve Bayesian. Ma et al. (2019) employed two-stream deep ConvNets to build an expert HAR system system using Inception-style temporal Convolutional Neural Network and a Recurrent Neural Network. Both networks are used to extract spatiotemporal information and exploit Spatio-temporal dynamics to enhance the whole system’s performance. This method has provided excellent accuracy, up to 94.1% for on UCF101 dataset and 69.0% for HMDB51 dataset. Ji et al. (2012) developed a novel 3D Convolution Neural Network for recognizing human activities in surveillance videos. The deep 3D-CNN model evaluates appearance and motion features from each video frame. This method has attained superior performance with an average accuracy of 90.2%. Zhang and Tian (2012) study various Spatio-temporal features-based descriptors for activity recognition. They found that the probabilistic graphical models are good enough to recognize activity patterns over time than SVMs. Li et al. (2018) used deep neural networks to recognize various actions based on modeling human body posture. The method integrates RGB and optical flow streams with 2D posture features to perform human activity classification. The literature, as discussed in sensor-based HAR systems, requires wearable sensors such as accelerometers, glucometers, proximity sensors, etc., to recognize human actions. However, the range of detecting human actions is limited in sensors-based HAR systems. Other side, vision-based HAR systems can identify a wide range of human acts using camera-based surveillance. They use complex convolutional neural architectures to learn temporal relations for human acts. However, improvements are still being required to enhance the performance of existing HAR systems. Therefore, this work proposed a cost-effective solution to differentiate human actions by analyzing human 2D body joints. The details related to the proposed system are presented in Sect. 3. Proposed methodology As mentioned earlier in the introduction section, this manuscript proposes an assistive technology based surveillance system to classify young and old persons in real-time scenarios. The overall workflow of the proposed system is presented in Fig. 1. The system takes video stream as input through the camera, examines each frame, and categorizes younger and older people by analyzing their walking styles. So that caregivers can be more attentive in the case of older people. The OpenPose algorithm is used to generate the human skeleton (pair of joints locations) in each frame. The OpenPose algorithm inputs an RGB frame of size “w×h” and provides the joint locations to form a skeleton for each individual within an image. After getting the skeleton joints, the skeleton data is aggregated for the first N frames using a sliding window of size N. Here, N skeletons are first pre-processed and then passed to a feature extraction module that extracts relevant features from them. Further, the extracted features are used to build a classifier for categorizing young and old walks. To achieve a real-time recognition framework, the window slides frame by frame along the video’s time dimension and outputs a label for each video frame.Fig. 1 Workflow of the proposed system Human detection and skeleton generation Human detection and skeleton generation are the primary tasks for identifying old and young walks. The work presented here uses the OpenPose algorithm (Cao et al. 1812) for human detection and skeleton generation from an image. It can jointly detect the human body and involve key points to generate skeleton. The OpenPose provides two Heat Maps, one for evaluating joint positions, i.e. Confidence Map (S), and the other for associating the joints, i.e. Part Affinity Field (PAF) Map (L) in a human skeleton. The OpenPose algorithm takes an image as input and spots skeletons for the person found in that image. An extracted skeleton involves 18 joints, including head, neck, arms, and legs, as shown in Fig. 2a. Each joint position is represented using spatial pair of coordinates, i.e. (x, y). Therefore, each skeleton is represented using 18 pairs of coordinates (a total of 36 values), as shown in Fig. 2b.Fig. 2 a Skeleton representation using OpenPose, b Representation of skeleton in spatial domain Pre-processing for features extraction After extracting the raw skeleton, the pre-processing stage suppresses the unwanted distortion from the skeleton data. The pre-processing stage helps to enhance the characteristics of skeleton data, which helps in more accurate classification at later stage. The pre-processing includes four steps summarized as follows:Considering all head’s joints: Along with the body and limb configurations, the head position can help a lot for the classification. Therefore, the five joints on the head are added manually to make the features more meaningful. Coordinate Scaling: The x and y coordinates for representing joint position do not follow the same scale. Therefore, these points need to be normalized in the same unit to deal with different width and height ratios. Discard frames that do not have neck and thighs: If OpenPose does not recognize a human skeleton or if the identified skeleton does not have a neck or thighbone inside the frame, the frame is considered invalid and dropped. The sliding window slides to the successive frames. Fill the missing joints: OpenPose may fail to recognize a full human skeleton in an occluded environment, which results in blanks at joint positions. To keep a fixed-size feature vector for the classification purpose, these joints must be filled with certain values. Here, the position of the missing joint is determined by its relative position to the neck in the preceding frame. Features extraction After pre-processing, the joint positions are completed and ready for use in the feature extraction process. Therefore, we used the sliding window of size N with N = 5 to extract relevant features from extracted joint positions that help to identify the action types. A better presentation with skeleton from five consecutive frames is illustrated in Fig. 3.Fig. 3 Raw joint positions The salient features are constructed using normalized joint positions by calculating the moving velocity of the joints and the angle of each joint for N window size. Further, a feature vector is created by concatenating these features, and then the extracted vectors are fed into a deep FFNN Classifier for training. The algorithm for finding more salient features from the raw skeleton data is discussed as follows:Algorithm 1. Finding the more salient features from the raw skeleton data Result: A feature vector appended with the following feature. 1. Link all the joints for N frames, Where N = 5 is Sliding Window. 2. Average the height (H) of the skeleton by considering previous N frames. This height is equivalent to the distance between the neck and the thigh. It is used to normalize all of the characteristics listed below. 3. Normalize the joint positions (N), as presented in equation 1. (1) N = [Joint Positions - mean (Joint Positions)]/Height (H) 4. Velocity (V) between joints is calculated as the next joint position minus the previous joint position in the normalized coordinate, as presented in equation 2. (2) V = N[tk] - N [tk-1] 5. Compute the angle of each joint from the joint's positions. Deep feed-forward neural network A deep feed-forward neural network (FFNN) is a deep neural network comprised of two or more layers of neurons. Feed-forward Network consists of an input layer and an output layer. The input layer is responsible for receiving the signal, while the predictions about the input are made in the output layer. There are an uncertain number of hidden layers between input and output layers in which the actual computation has to be performed. Only one hidden layer in FFNNs is proficient in approximating any continuous function. FFNN learns to simulate the correlation between inputs and outputs by training on a collection of input–output pairings. The model parameters, or weights and biases, are adjusted throughout training to reduce the error. Backpropagation is utilized to adjust the weights and biases relative to the error, and root mean squared error is used for measurements. FFNN updates the partial derivatives of the error function for many weights and biases using back propagation and the chain rule of calculus. Figure 4 shows FFNN architecture with hidden layers.Fig. 4 FFNN architecture with hidden layers In this work, the deep FFNN model has been used to detect and recognize elderly people that contain one input layer, three hidden layers, and an output layer. Three times dropout has been used to prevent the model from overfitting. The input layer has 314 nodes which are features extracted from the raw skeleton data. The rectified linear unit (ReLU) is used here as an activation function to deal with non-linear input data. There are 100 nodes in each hidden layer and 2 nodes in the output layer. The output layer has 2 nodes for 2 different classes. The sigmoid function has been used as an activation function that is used to calculate the probabilistic value of each class with a learning rate of 0.0001. The highest probabilistic class will be considered as an output to the corresponding input image. Experimentation Google Colab platform has been used to perform a wide range of experiments. PIL (python imaging library) and OpenCV have been employed to open, save, and manipulate images. Keras library is used for classification purposes by incorporating SVM with linear/RBF kernel and Deep Neural Network. The Matplotlib library is used here to visualize model accuracy and loss curves. Scikit-learn is employed to produce the confusion matrix, and TensorFlow is used as a data flow. Dataset Dataset is synthesized by ourselves in an indoor and outdoor environment using a 16 MP Mobile camera. The dataset contains two types of people walk such as elderly people walk and younger people walk. Each class consists of a variable amount of videos, ranging from 30 s to 2 min in length. Videos are captured at the resolution of 640 × 480. For machine understanding, the created dataset requires proper formatting and labeling for training the model. YAML (Yet Another Markup Language), a comprehensible information serialization language is used for this purpose. The images that are going to be used for training and their label are configured in a text file containing the information like class name, starting and ending index of the video corresponding to that class. The distribution of our dataset is presented in Table 1.Table 1 Dataset’s frame distribution Actions Old people walk Young people walk No. of frames 4526 5246 Figure 5 shows some instances/samples of video sequences representing old people walking and young people walking. The clips are shot in different scenarios like indoor, outdoor, low lighting, etc.Fig. 5 Snapshots of two types of classes in the training data: elderly people and young people walk Training The entire dataset is divided as 70% for training and 30% for testing. After pre-processing and features extraction from the raw skeleton data, the next phase forms a model to classify the data. The classification has been done using different classifiers, including Neural Network (FFNN Classifier), Support Vector Machine (SVM), and SVM with kernel method. Setting up the hidden layer and balancing the learning rate (Putra and Yulita 2019) are worked out for efficient modeling. To obtain the best outcome for each parameter, the experiment was repeated several times to find the best amount of hidden layers and the best learning rate. For updating the parameters, the loss function partial derivate w.r.t the model parameters is computed. Validation is done along with training. Figure 6 illustrates the performance of a posture recognition system on a synthesized dataset where people perform the walking action. The trained model is fine enough for detecting the old walk or young walk.Fig. 6 Sequences outcomes Testing The model is tested for 2930 test images containing two classes, each having 1357 and 1573 images to represent old people walk and young people walk, respectively. The test images are passed to our method proposed for prediction. The model first processes each test image through the OpenPose to detect the human skeleton in that image. The skeleton data is passed to pre-processing module, feature extraction module, and classification module. The model weights are adjusted during the training phase, and the window slides frame by frame along the video’s time dimension in which the prediction has been made. The proposed recognition system’s performance for elderly people has been tested on our dataset. Testing videos are also similar to training videos. Result This section shows the experimentation outcomes of the proposed model. Table 2 shows the recognition accuracy of the proposed method over different classifiers. The outcomes show that the proposed method trained over the FFNN classifier is quite higher than both variants of the SVM classifier.Table 2 Recognition accuracy for N = 5 Method Classifier Accuracy train (%) Accuracy test (%) SVM Linear 88.8 89.1 SVM Kernel 91.6 92.1 FFNN classifier 100 × 100 × 100 98.45 96.16 The Confusion matrix of the Deep Neural Network model is given in Fig. 7. Diagonal values of the matrix represent correctly classified testing outcomes. Non-diagonal values represent miss-classified outcomes. Miss-classified means predicted value and actual value are not matching.Fig. 7 Confusion matrix Different performance measures (Singh 2015, 2017; Reddy and Geetha 2020) like Precision, Recall, F-Measures, and Support are used to evaluate the performance of this proposed model, illustrated in Table 3.Table 3 Evaluated performance measures Metrics name Old people walk Young people walk Precision 0.94 0.98 Recall 0.96 0.96 F-measure 0.95 0.97 Support 223 376 The performance of the proposed method is compared with other existing methods in Table 4. The result drawn in Chernbumroong et al. (2013) works on analyzing the sensors’ data for assisted living and provides an accuracy of 90.23%. In Reddy and Geetha (2020), activities are modeled using video-based classification that offers up to 92.16% accuracy. However, the proposal in Li et al. (2018) classifies activities by modeling body posture using a deep neural network and stands good compared to others with an accuracy of 93.61%. The last and our proposed method achieves around 96% of accuracy for almost the same activities and behavioral modeling.Table 4 Comparison with existing methods Methods Accuracy (%) Chernbumroong et al. (2013) 90.22 Reddy and Geetha (2020) 92.16 Li et al. (2018) 93.61 Proposed method 96.16 Conclusion These manuscripts proposed a system to spot human presence and recognize whether the particular person is an older adult by analyzing the human walking style. This system constructs features from the pre-processed skeleton data constructed over video sequences. The developed method aggregates the skeleton data of a 0.5 s window for feature extraction. The model used raw features from five consecutive frames to improve the models’ performance. We evaluated the developed model over our synthesized dataset. This paper considered a deep neural network model to recognize elderly people in the video. There are also two other variants of the SVM classifier, such as SVM with linear kernel and RBF kernel, which achieves good accuracy in elderly people’s recognition. The result shows that the deep neural network model outperforms Linear SVM and RBF-SVM on our dataset, demonstrating good results in real-world environments. In the future, the system can be used to deploy in different applications like smart homes, theft detection, augmentation reality and many more. Additionally, more adaptive techniques like advanced CNN architecture can be used to upsurge the performance of the proposed system. 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 Agency CI The CIA world factbook 2010 2009 Skyhorse Publishing Inc. Ansari M Singh DK Human detection techniques for real time surveillance: a comprehensive survey Multimed Tools Appl 2021 80 6 8759 8808 10.1007/s11042-020-10103-4 Cao Z, Hidalgo G, Simon T, Wei S-E, Sheikh Y (2018) Openpose: real-time multi-person 2d pose estimation using part affinity fields. arXiv preprint arXiv:1812.08008 Chernbumroong S Cang S Atkins A Yu H Elderly activities recognition and classification for applications in assisted living Expert Syst Appl 2013 40 5 1662 1674 10.1016/j.eswa.2012.09.004 Ji S Xu W Yang M Yu K 3d convolutional neural networks for human action recognition IEEE Trans Pattern Anal Mach Intell 2012 35 1 221 231 10.1109/TPAMI.2012.59 Li C Tong R Tang M Modelling human body pose for action recognition using deep neural networks Arab J Sci Eng 2018 43 12 7777 10.1007/s13369-018-3189-z Lubeek SF van Vugt LJ Aben KK van de Kerkhof PC Gerritsen M-JP The epidemiology and clinicopathological features of basal cell carcinoma in patients 80 years and older: a systematic review JAMA Dermatol 2017 153 1 71 78 10.1001/jamadermatol.2016.3628 27732698 Ma C-Y Chen M-H Kira Z AlRegib G TS-LSTM and temporal-inception: exploiting spatiotemporal dynamics for activity recognition Signal Process Image Commun 2019 71 76 87 10.1016/j.image.2018.09.003 Ojha U, Adhikari U, Singh DK (2017) Image annotation using deep learning: a review. In: 2017 international conference on intelligent computing and control (I2C2). IEEE Pienaar SW, Malekian R (2019) Human activity recognition using LSTM-RNN deep neural network architecture. In: 2019 IEEE 2nd wireless Africa conference (WAC). IEEE, pp 1–5, IEEE Putra D, Yulita I (2019) Multilayer perceptron for activity recognition using a batteryless wearable sensor. In: IOP conference series: earth and environmental science, vol 248. IOP Publishing, p 012039 Reddy GP Geetha MK Video based fall detection using deep convolutional neural network Eur J Mol Clin Med 2020 7 2 5542 5551 Singh DK (2015) Recognizing hand gestures for human computer interaction. In: 2015 international conference on communications and signal processing (ICCSP). IEEE Singh DK (2017) Gaussian elliptical fitting based skin color modeling for human detection. In: 2017 IEEE 8th control and system graduate research colloquium (ICSGRC). IEEE Singh NP Kaur G Urinary tract infection in elderly: to treat or not to treat? J Assoc Physicians India 2018 66 11 Velayutham B Kangusamy B Joshua V Mehendale S The prevalence of disability in elderly in India—analysis of 2011 census data Disabil Health J 2016 9 4 584 592 10.1016/j.dhjo.2016.04.003 27174073 Visutsak P, Daoudi M (2017) The smart home for the elderly: perceptions, technologies and psychological accessibilities: the requirements analysis for the elderly in Thailand. In: 2017 XXVI international conference on information, communication and automation technologies (ICAT). IEEE, pp 1–6 Zhang C Tian Y RGB-D camera-based daily living activity recognition J Comput Vis Image Process 2012 2 4 12	0	PMC9714410	NO-CC CODE	2022-12-02 23:24:45	no	Int J Syst Assur Eng Manag. 2022 Dec 1;:1-8	utf-8	null	null	null	oa_other
==== Front Curr Psychol Curr Psychol Current Psychology (New Brunswick, N.j.) 1046-1310 1936-4733 Springer US New York 3996 10.1007/s12144-022-03996-x Article How the Dark Triad associated with internet gaming disorder? The serial mediation of basic psychological needs satisfaction and negative coping styles Xu Xuan 1 Gao Ling-feng 2 Lian Shuai-lei 3 Chen Qian 1 http://orcid.org/0000-0002-0958-6568 Zhou Zong-kui [email protected] 1 1 grid.411407.7 0000 0004 1760 2614 School of Psychology, Central China Normal University, Wuhan, China 2 grid.453534.0 0000 0001 2219 2654 Institute of Psychological and Brain Sciences, Zhejiang Normal University, Zhejiang, China 3 grid.410654.2 0000 0000 8880 6009 College of Education and Sports Sciences, Yangtze University, Jingzhou, China 1 12 2022 19 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. According to the I-PACE model, this study focused on the role of need satisfaction and negative coping styles in the relationship between the Dark Triad (i.e., Machiavellianism, psychopathy, and narcissism) and internet gaming disorder (IGD). In a sample of 749 emerging adult gamers, a multiple mediation model with Dark Triad as the distal variable, psychological need satisfaction and negative coping style as mediating variables, and IGD as the outcome variable was tested. Results indicated that Machiavellianism and psychopathy were found to be significant predictors of IGD when mediated by psychological need satisfaction and negative coping styles. Narcissism predicts IGD only through the indirect effect of negative coping styles. The findings enhanced our understanding that Machiavellianism and psychopathy are characterized by compensatory use of internet games, as well as added new perspectives to the understanding of addiction mechanisms in narcissists. Keywords Dark Triad Machiavellianism Psychopathy Narcissism Internet gaming disorder Basic psychological needs Negative coping style ==== Body pmcIntroduction Against the backdrop of the digital age, internet games have become a part of daily leisure activities (Kuss et al., 2017). Especially due to the COVID-19 epidemic, many people who stay at home have already participated in internet games (King et al., 2020). Currently, China has 665 million gamers, mainly emerging adults aged 18 to 25, who spend 81 minutes playing daily (Statista, 2021). Although internet games can provide fun and interaction, it also puts people at risk of internet gaming disorder (IGD; King et al., 2019). As a behavioral addiction disorder in the DSM-5, IGD refers to a behavioral pattern of excessive and compulsive use of Internet gaming (Petry et al., 2014). IGD often exposes individuals to significant distress, such as depression, anxiety, sleep disorders, and interpersonal conflicts (Männikkö et al., 2020). The I-PACE model integrates a wide range of theoretical and experimental findings in the field of internet use disorders and provides a comprehensive theoretical framework to describe the onset and maintenance of internet use disorders (Brand et al., 2016). This model emphasizes the role of core characteristics (e.g., personality, specific needs) and cognitive-emotional factors (e.g., coping) in internet use disorders. From the I-PACE model, personality traits are a key factor that induces IGD. Individuals with vulnerable traits may initially indulge in internet games driven by motivational factors, and subsequently become accustomed to internet games as a way to cope with stress, eventually developing into IGD (Brand et al., 2019). For example, with the need to escape emotions, people with anxiety traits gradually turn to internet games as a coping mechanism, eventually leading to IGD (Biolcati et al., 2021; Plante et al., 2019). According to the I-PACE model, we examine the role of the Dark Triad (i.e., Machiavellianism, psychopathy, and narcissism) as distal factors, psychological needs, and negative coping styles as proximal factors, to understand IGD. Dark Triad and internet gaming disorder In recent years, there has been increasing interest in the relationship between personality traits and IGD. The role of the Big Five personality in IGD has been initially revealed (Şalvarlı & Griffiths, 2019), but the same attention needs to be paid to certain dark traits that describe the malevolent side of human nature. We aim to focus on the role of the Dark Triad as its construct is well supported, even though the composition of dark traits is still up for debate (Bonfá-Araujo et al., 2022; Muris et al., 2017). Currently, the relationships between the Dark Triad and IGD are not well understood, which has aroused particular attention. The Dark Triad refers to a group of personality traits with undesirable and antisocial characteristics, namely Machiavellianism, psychopathy, and narcissism (Paulhus & Williams, 2002). Machiavellianism describes a personality tendency to focus on the pursuit of personal goals, with manipulation and exploitation of others as a means to an end (Christie & Geis, 1970). Psychopathy is a personality trait described as persistently antisocial, lacking in empathy and remorse, and bold (Sleep et al., 2019). Narcissism originated in a mythology about Narcissus and was later summarized as a personality trait of vanity, admiration, disregard for others and arrogance (Sedikides, 2021). The Dark Triad was found to be closely related to various behavioral addictions, both offline and online (for a review, see Jauk & Dieterich, 2019; Kircaburun et al., 2019). In cyberspace with abundant and accessible rewards, the Dark Triad was prone to developing problematic internet use (Kircaburun & Griffiths, 2018). Pioneering research has been conducted to explore the unique risk of developing IGD in the Dark Triad (Kircaburun et al., 2018). Specifically, The Dark Triad were found to indulge in perpetrating grief play (trolling behavior in games) and other antisocial behaviors in internet games (Ladanyi & Doyle-Portillo, 2017; Tang et al., 2019). For machiavellianism, internet games can be used not only for entertainment (especially violent games), but also to satisfy the need to exploit and control others (Sindermann et al., 2018; Tang et al., 2020). Individuals with high psychopathology are similarly at risk of excessive use of internet games due to high impulsivity and preference for violent content (Sindermann et al., 2018). Narcissists relish the pursuit of exaggerated power and admiration from others in games and develop IGD (Kim et al., 2008). In particular, individuals with Dark Triad may employ compensatory internet games usage when real-life needs are not fulfilled and gradually form usage expectations to avoid stress (Tang et al., 2020). For instance, in order to gain respect from others and cope with negative emotions, narcissists may continue to use internet games, leading to IGD (Di Blasi et al., 2020). Machiavellians and psychopaths pursue competition in games to satisfy the unfulfilled desire for manipulation and achievement in real life (Kircaburun et al., 2018). Therefore, the Dark Triad may be predisposing variables to IGD. According to the above evidence, we assumed that the Dark Triad was positively associated with IGD. The mediating role of psychological needs satisfaction The I-PACE model suggests that individuals with predisposing traits have compensatory experiences during game usage (Brand et al., 2019). The compensatory perspective helps researchers to explore the motivational factors that underpin IGD in the context of the gamer's real life (Scerri et al., 2019). Recently, a framework of motivation based on psychological needs fulfillment has gained attention (Przybylski & Weinstein, 2019). Meanwhile, the process of need satisfaction is often driven by traits (Costa et al., 2019). For example, the stability trait was positively associated with real-life need satisfaction (Şimşek & Koydemir, 2013). Psychological needs have been suggested as being central to understanding the relationship between traits and IGD (Arpaci et al., 2018). Individuals with a certain trait may favor one or more psychological needs fulfillment (Özteke Kozan et al., 2019). With the multifactorial gamification design, games can continuously meet their psychological needs (Sailer et al., 2017). Eventually, the compensation process will lead to IGD (Kardefelt-Winther, 2014). Especially for individuals with the Dark Triad traits who exhibit psychological need deficiency (Kaufman et al., 2019). Self-determination theory suggests that individuals have three basic psychological needs as relatedness, competence, and autonomy, which drive the various daily activities. Relatedness needs refer a desire for harmonious interpersonal relationships, competency needs reflect a sense of control over the environment, and autonomy needs focus on the freedom of psychology and behavior (Deci & Ryan, 2000). There may be one or more psychological needs deficits in real life for certain maladaptive personalities (e.g., Dark Triad; Dweck, 2017). Specifically, narcissists lacked the autonomy need satisfaction in their life, while machiavellians and psychopaths wished for more competence satisfaction (Jonason & Ferrell, 2016). Given poor interpersonal and environmental adaptability, the individuals with Dark Triad had difficulty meeting psychological needs in real life (Jonason et al., 2015). Psychological need deficiency may cause them to turn to other way (e.g., internet game) for need fulfillment. Internet games usage that fills psychological voids in real life is more likely to turn into addiction (Chen, 2019). Previous researchs have found that internet games can attract gamers with the Dark Triad traits to play consistently by satisfying social, competitive, and other needs (Kircaburun et al., 2018). Thus, it is crucial to examine the compensation model of the Dark Triad traits and IGD in terms of basic psychological needs. Given the lack of research evidence, we will explore how three basic psychological need satisfaction in real life explains the relationship between the Dark Triad and IGD. The mediating role of negative coping styles Coping styles were also often considered as a mediating mechanism for personality and IGD (Zhou et al., 2017). For example, individuals with high conscientiousness are more likely to adopt the problem-solving-oriented coping styles, while maladaptive traits such as neuroticism often adopted the negative coping styles (Segerstrom & Smith, 2019). Correspondingly, for individuals who adopt the negative coping style, an expectation may develop that internet game is an effective way to escape stress (Loton et al., 2016). Negative coping styles were described as a way of dealing with pressure by avoidance and denial (Folkman & Moskowitz, 2000). Recent studies have shown that individuals with the Dark Triad exhibit more negative coping styles (Jonason et al., 2020). Pace et al. (2021) found that negative coping styles could explain why individuals with the Dark Triad are prone to developing various types of behavioral addictions. However, to date no studies examined that the Dark Triad could be connected to IGD via negative coping styles. In addition, psychological need satisfaction may not only be directly related to IGD. Driven by unmet psychological needs, individuals will resort to more negative coping styles (Neufeld & Malin, 2021). In the perspective of self-determination theory, unmet psychological needs can cause individuals to gradually develop rigid patterns of behavior, such as negative coping styles. (Vansteenkiste et al., 2013). Meanwhile, needs dissatisfaction could motivate and reinforce the pursuit of external goals (i.e., internet games) as a coping mechanism (Sheldon & Anderson, 2011). As previously stated, the negative coping styles increases the risk of developing IGD. Thus, negative coping styles can be used as a pattern of maladaptive behavior induced by unmet needs, explaining the relationship between psychological need satisfaction and IGD. Similarly, drawing upon I-PACE, for individuals with susceptible traits, the early stage may lead to IGD due to compensating for unmet psychological needs, and the later stage may orient to IGD due to the formation of habitual negative coping styles (Brand et al., 2019). The developmental stages assumed by the I-PACE model also suggest that basic psychological need satisfaction may be indirectly linked to IGD through negative coping styles. Such a mechanism of intrinsic need compensation followed by external coping change has not yet been tested on the Dark Triad. Therefore, this study examined the question of whether basic psychological needs satisfaction and negative coping styles sequentially mediate the relationship between the Dark Triad and IGD. The present study By and large, the Dark Triad traits may be the emerging core factor in understanding IGD. However, previous studies have considered how the individual with the Dark Triad was pushed into IGD from the perspective of game motivation (Kircaburun et al., 2018; Tang et al., 2020). It is also important to examine the pull toward IGD through the daily basic needs satisfaction (Mills et al., 2018). Following this rationale, this study first focused on the mediating role of basic psychological needs satisfaction between the Dark Triad and IGD. In addition, while considering psychological needs, negative coping styles were also a possible mediating mechanism to explain the relationship between the Dark Triad traits and IGD. Guided by the I-PACE model (Brand et al., 2019), exposure to low satisfaction with daily psychological needs can reinforce the formation of habitual negative coping, thereby facilitating the development of IGD. The multiple mediation model tested in this study was shown in Fig. 1.Fig. 1 Measurement model Methods Participants and procedure Participants were recruited through an online advertisement posted on QQ, WeChat, and Sina Weibo. The inclusion criteria are Chinese internet gamers (used internet games in the last six months) and undergraduate students aged 18–25. The investigation started in March 2020 and ended in August 2021. In total, 849 participants completed the questionnaires. Questionnaires with regular responses, repeated responses, and incorrect answers to trap questions (an instruction to choose a specific answer) were deleted. The final sample consisted of 740 undergraduate gamers (males = 63.6%). Of these, 151 (20.4%) were freshmen, 167 (22.6%) were sophomores, 157 (21.2%) were juniors, and 265 (35.8%) were seniors or higher. It was determined by Monte Carlo method that a statistical power of 0.8 could be achieved at α = 0.05 for a sample of n = 419 (Schoemann et al., 2017). Measures Dark Triad The Dirty Dozen (DD) was used (Jonason & Webster, 2010), which is a validated and proven measurement for Dark Triad. The DD comprises 12 items on a 7-point Likert scale from “1 = strongly disagree” to “7 = strongly agree”, four items for each dimension (i.e., Machiavellianism, psychopathy, narcissism). The higher the score on each dimension, the higher the level of the corresponding dark trait. Basic psychological needs The Chinese version of the Basic Psychological Needs Scales (C-BPNS) was used (Gagné, 2003), which comprises 19 items on a 7-point Likert scale from “1 = strongly disagree” to “7 = strongly agree”. The C-BPNS measures three psychological needs in real life: relatedness, competence, and autonomy. The lower the score of each dimension, the lower corresponding need satisfaction. Negative coping styles The Simplified Coping Style Questionnaire was used (Xie, 1998), which was divided into two subscales: positive coping styles (PCS) and negative coping styles (NCS). For the aim of the current study, only the negative coping styles scale was used. The NCS comprises 8 items on a 4-point Likert scale from “0 = never” to “4 = always”. Higher scores reflect the level of the negative coping style. Internet gaming disorder The Internet Gaming Disorder Questionnaire was used (Petry et al., 2014), which assessed the nine criteria for IGD. This measure comprises 9 items on a 5-point Likert scale from “1 = strongly disagree” to “9 = strongly agree”. Scores reflect the extent of online game addiction. Data analysis SPSS Statistics 25 was used for descriptive statistics and Pearson correlation analysis. The multiple mediation analyses via using Mplus 7.4 software (Model 80; Stride et al., 2015), were conducted with each Dark Triad as independent variables, each psychological need satisfaction as first mediating variables, negative coping styles as subsequent mediator variables, IGD as the outcome variable, gender as a control variable. The bootstrapping method with 95% bias-corrected confidence intervals (N = 10,000) was used to reduce the risk of biased results. Results In the present study, the descriptive statistics, variance inflation factors (VIF), internal consistency coefficient (Cronbach's Alpha) and correlation of each variable are shown in Table 1. The VIF value for all independent variables is less than 5, indicating that the collinearity problem is not serious (Shrestha, 2020).Table 1 Descriptive statistics and correlation 1 2 3 4 5 6 7 8 1. Gaming disorder – 2. Machiavellianism 0.43* – 3. Psychopathy 0.40* 0.64* – 4. Narcissism 0.15* 0.31* 0.11 – 5. Competence satisfaction − 0.30* − 0.29* − 0.38* − 0.04 – 6. Relatedness satisfaction − 0.20* − 0.30* − 0.41* − 0.03 0.61* – 7. Autonomy satisfaction − 0.23* − 0.28* − 0.33* − 0.09* 0.69* 0.59* – 8. Negative coping styles 0.33* 0.37* 0.30* 0.23* − 0.33* − 0.20* − 0.27*** – M 23.75 11.15 9.93 19.59 26.40 33.48 25.07 19.49 SD 7.41 5.97 5.43 5.43 5.17 6.31 4.92 4.46 Skewness 0.17 0.59 1.03 − 0.74 0.48 0.14 0.01 0.15 Kurtosis − 0.47 − 0.53 0.53 0.35 0.41 0.09 0.15 − 0.14 Cronbach’s alpha 0.88 0.90 0.86 0.88 0.64 0.75 0.61 0.76 VIF – 1.95 1.89 1.18 1.88 2.29 2.13 1.27 * p < 0.05, p < 0.01, * p < 0.001 As shown in Fig. 2, the multiple mediation model of the Dark Triad (Machiavellianism, psychopathy, and narcissism) to predict IGD was partially supported. The results of the direct effect show: (I) Machiavellianism negatively predicted competence (β = − 0.10, SE = 0.04, p = 0.02), relatedness (β = − 0.09, SE = 0.04, p = 0.02), and autonomy (β = − 0.10, SE = 0.05, p = 0.02) need satisfaction. Machiavellianism positively predicted negative coping styles (β = 0.22, SE = 0.05, p < 0.001) and IGD (β = 0.22, SE = 0.05, p < 0.001). (II) Psychopathy negatively predicted competence (β = − 0.31, SE = 0.04, p < 0.001), relatedness (β = − 0.31, SE = 0.04, p < 0.001), and autonomy (β = − 0.31, SE = 0.04, p < 0.001) need satisfaction. Psychopathy positively predicted IGD (β = − 0.25, SE = 0.06, p < 0.001), but not negative coping styles (β = 0.06, p = 0.20). (III) The relationship between narcissism and three needs satisfaction was not significant. Narcissism positively predicted negative coping styles (β = 0.15, SE = 0.04, p < 0.001), but not IGD (β = 0.02, p = 0.57). (IV) Of the three needs, only the competence satisfaction negatively predicted IGD (β = − 0.17, SE = 0.05, p < 0.001). Negative coping styles positively predicted IGD (β = 0.14, SE = 0.04, p < 0.001).Fig. 2 The multiple mediation model. Note: All variables in the model are observed variables. The value on the horizontal line is the standardized path coefficient. On the left, the three coefficients describe when the independent variables are machiavellianism, psychopathy, and narcissism. In the dummy code for gender, male = 0, female = 1. For clarity, non-significant paths are hidden. * p < 0.05, p < 0.01, * p < 0.001 The results of the indirect effect show: Machiavellianism was indirectly associated with IGD via three needs satisfaction and negative coping styles. All indirect paths in the model are significant, except for the single mediating effect of relatedness and autonomy need, which is not significant. Similarly, the sequential mediation model for psychopathy predicting IGD was significant. For the single mediation model, only the autonomy need, relatedness need, and negative coping styles were not significant. However, narcissism only indirectly predicts IGD through negative coping styles. Finally, the total mediation model predicted 28.4% of the variance in IGD. Table 2 contains the results of the indirect effects.Table 2 Standardized estimates of indirect effects of the tested model (n = 749) Path Effect SE 95% CI Mach → IGD (total effect) 0.058 0.016 [0.030, 0.094] Mach → CS → IGD 0.017 0.010 [0.004, 0.036] Mach → NCS → IGD 0.037 0.009 [0.018, 0.059] Mach → CS → NCS → IGD 0.001 0.001 [0.001, 0.003] Mach → AS → NCS → IGD 0.001 0.001 [0.001, 0.003] Mach → RS → NCS → IGD 0.001 0.001 [0.001, 0.003] Psych → IGD (total effect) 0.068 0.018 [0.037, 0.107] Psych → CS → IGD 0.056 0.016 [0.025, 0.092] Psych → CS → NCS → IGD 0.004 0.001 [0.002, 0.008] Psych → AS → NCS → IGD 0.004 0.001 [0.002, 0.008] Psych → RS → NCS → IGD 0.004 0.001 [0.002, 0.008] Narci → NCS → IGD (total effect) 0.025 0.009 [0.010, 0.046] Mach Machiavellianism, Psych Psychopathy, Narci Narcissism, CS Competence satisfaction, AS Autonomy satisfaction, RS Relatedness satisfaction, NCS Negative coping styles. Only significant pathways are shown in the table Discussion This study constructed a sequential mediation model to explore the psychological mechanisms underlying the link between the Dark Triad and IGD, taking into account psychological needs and negative coping styles. The results showed that Machiavellianism and psychopathy could predict IGD through sequential mediation of basic psychological needs satisfaction and negative coping styles, and narcissism could predict IGD only via the negative coping styles. Consistent with the I-PACE model (Brand et al., 2019) and previous studies (Kircaburun et al., 2018), the present study identified Machiavellianism, psychopathy and narcissism all as susceptible personalities for IGD. The Dark Triad, which summarized the malevolent side of human nature, demonstrates low satisfaction of psychological needs in daily life (Kaufman et al., 2019). The disinhibitory virtual game world becomes a convenient way for them to get instant gratification. In short, what the individual with the Dark Triad loses in their daily lives, then gets back on the internet. In models of Machiavellianism or psychopathy predicting IGD, we found a compensatory effect due to unmet psychological needs, and subsequent cognitive-behavioral changes (the formation of negative coping styles). In the early stages of IGD, individuals with high Machiavellian or psychopathy, experienced insufficient satisfaction of basic psychological needs due to their personality defects and real-life constraints. In other words, they have difficulty interacting well with the environment/others and gaining sufficient competence due to their indifferent, confrontational and manipulative characteristics. Given the maladaptive character of Machiavellians and psychopaths, for them, these needs are difficult to meet in real life. Therefore, they used the social, competitive, and exploratory functions of games to compensate for their psychological needs, which is consistent with previous studies (Kircaburun et al., 2018). In the later stages of maintaining IGD, with constant repetition and reinforcement, this alternative compensation through Internet gaming was gradually solidified into the negative coping styles. Internet games became the primary way of coping with stress and problems (i.e., unmet psychological needs), contributing to the progressive development of IGD. Similar mechanism processes were found in the study of Ataşalar and Michou (2019). In addition, competence needs satisfaction more robustly predicted IGD than the other needs. This suggests that untalented feelings in real life were more likely to be pulled into internet games, especially for Machiavellianism and psychopathy. It should also be noted that psychopaths have lower psychological needs satisfaction than Machiavellians. Previous studies found that psychopathy is the “darkest” trait in Dark Triad. Of the three personalities, psychopathy may be the only trait that exhibits a rapid life history strategy (Jonason et al., 2010). Psychopathy showed a stronger correlation with antisocial behavior and emotional deficiencies than machiavellianism and narcissism (Jonason & Krause, 2013; Szabó et al., 2018). The present study extends this finding that psychopathy was more likely to be susceptible to IGD due to unmet psychological needs in real life. In addition, there was no significant association between narcissism and unmet psychological needs. Contrary to our hypothesis, psychological needs cannot mediate the relationship between narcissism and IGD or narcissism and negative coping styles. Previous research had found that higher narcissism was not only unrelated to the fast life history strategy but also acted as a performance facilitator for need fulfillment in real life, at least to some extent (Sedikides, 2021). That is, it was difficult for narcissists who are confident, bold and self-centered to experience low satisfaction with basic psychological needs. It makes sense that the compensation model for narcissism and IGD is not supported. But narcissism was not unrelated to IGD. Rather, we found that negative coping styles can mediate the relationship between narcissism and IGD, which is consistent with research on gambling addiction (Pace et al., 2021). This suggests that narcissists may try to use internet games as an opportunity to avoid stress or increase their sense of mastery. Based on this, it is also necessary to examine in the future whether individuals with high narcissism are more engaged in games, especially when they feel stressed or lose a sense of control. At the same time, narcissism can be further divided into grandiose and vulnerable narcissism. Di Blasi et al (2020) found that the compensation model was effective in explaining the relationship between vulnerable narcissism and IGD. We speculated that the vulnerable subtype of narcissism may overuse games because of low need satisfaction in daily life (Sedikides, 2021). Overall, we found mediation mechanisms of IGD in Machiavellianism and psychopathy, which comprise of realistic needs compensation and dysfunctional coping formation. Narcissism causes IGD only because of negative coping styles, without the unmet needs in real life. Such findings offer the possibility of personalized prevention and treatment for IGD, and enrich our understanding of the Dark Triad and IGD. However, this study is not without limitations. First, the data are from the cross-sectional questionnaire, which cannot provide strong support for causality conclusions. The longitudinal or mixed approach could be used in the future. Second, there are more dark traits or subtypes that should be considered, such as sadism or vulnerable narcissism. Third, basic psychological needs can be further distinguished as satisfied and frustrated, or both online and offline satisfaction can be considered. Fourth, the internal consistency coefficients of the C-BPNS were low, especially for the subscales of autonomous needs. This suggests that the results on autonomous needs should be viewed with caution and should receive further validation. Finally, self-report methods may suffer from biases such as social desirability, in particular to the Dark Triad traits measure. Despite its limitations, the present study has some strengths. We initially depict a compensatory trajectory about the development of the dark triad into IGD. Negative coping mechanisms based on unmet real-life needs can lead to IGD, especially for some maladaptive traits such as Machiavellianism and psychopathy. The findings provide evidence for a compensatory rather than a pathologically compulsive model of IGD. Meanwhile, where the motivation to use internet games is grounded in some unmet psychological needs and where games alleviate stress, some individuals with susceptible traits may devote their time to internet games which leads to IGD. This compensation model can be usefully replicated and expanded in the future. Notably, we found that highly narcissistic individuals seem to be less "vulnerable" in real life, which is reflected in the failure of the compensatory effect of unmet real-life needs. This may provide ideas for the advantages and heterogeneity of narcissism compared to other dark traits. Lastly, if the Dark Triad is prominent, the improvement measures for real-life feelings of target restriction, incompetence and interpersonal frustration, as well as intervention that targets coping styles, could be considered by clinicians to reduce the progression to IGD. Authors contribution Zong-kui Zhou, Ling-feng Gao and Xuan Xu conceived the idea of the study; Xuan Xu and Ling-feng Gao analyzed the data; Xuan Xu, Ling-feng Gao, Shuai-lei Lian, Qian Chen and Zong-kui Zhou interpreted the results; Xuan Xu and Ling-feng Gao wrote the paper; all authors discussed the results and revised the manuscript. Data availability All data included in this study are available upon request by contact with the corresponding author. Declarations Ethics approval The research was approved by the Institutional Review Boards at the Central China Normal University, PRC, and APA ethical standards were followed in the conduct of the study. Competing interests The authors have no financial or proprietary interests in any material discussed in this article. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Arpaci I Kesici Ş Baloğlu M Individualism and internet addiction: The mediating role of psychological needs Internet Research 2018 28 2 293 314 10.1108/IntR-11-2016-0353 Ataşalar J Michou A Coping and mindfulness: Mediators between need satisfaction and generalized problematic internet use Journal of Media Psychology 2019 31 2 110 115 10.1027/1864-1105/a000230 Biolcati, R., Passini, S., & Pupi, V. (2021). The role of video gaming motives in the relationship between personality risk traits and Internet Gaming Disorder. Journal of Gambling Issues, 46. 10.4309/jgi.2021.46.12 Bonfá-Araujo B Lima-Costa AR Hauck-Filho N Jonason PK Considering sadism in the shadow of the Dark Triad traits: A meta-analytic review of the Dark Tetrad Personality and Individual Differences 2022 197 111767 10.1016/j.paid.2022.111767 Brand M Young KS Laier C Wölfling K Potenza MN Integrating psychological and neurobiological considerations regarding the development and maintenance of specific Internet-use disorders: An Interaction of Person-Affect-Cognition-Execution (I-PACE) model Neuroscience & Biobehavioral Reviews 2016 71 252 266 10.1016/j.neubiorev.2016.08.033 27590829 Brand M Wegmann E Stark R Müller A Wölfling K Robbins TW Potenza MN The Interaction of Person-Affect-Cognition-Execution (I-PACE) model for addictive behaviors: Update, generalization to addictive behaviors beyond internet-use disorders, and specification of the process character of addictive behaviors Neuroscience & Biobehavioral Reviews 2019 104 1 10 10.1016/j.neubiorev.2019.06.032 31247240 Chen A From attachment to addiction: The mediating role of need satisfaction on social networking sites Computers in Human Behavior 2019 98 80 92 10.1016/j.chb.2019.03.034 Christie R Geis FL Studies in Machiavellianism 1970 Academic Press Costa PJ McCrae RR Lockenhoff CE Personality across the life span Annual Review of Psychology 2019 70 423 448 10.1146/annurev-psych-010418-103244 30231002 Deci EL Ryan RM The "what" and "why" of goal pursuits: Human needs and the self-determination of behavior Psychological Inquiry 2000 11 4 227 268 10.1207/S15327965PLI1104_01 Di Blasi M Giardina A Lo Coco G Giordano C Billieux J Schimmenti A A compensatory model to understand dysfunctional personality traits in problematic gaming: The role of vulnerable narcissism Personality and Individual Differences 2020 160 109921 10.1016/j.paid.2020.109921 Dweck CS From needs to goals and representations: Foundations for a unified theory of motivation, personality, and development Psychological Review 2017 124 6 689 719 10.1037/rev0000082 28933872 Folkman S Moskowitz JT Positive affect and the other side of coping American Psychologist 2000 55 6 647 654 10.1037//0003-066X.55.6.647 10892207 Gagné M The role of autonomy support and autonomy orientation in prosocial behavior engagement Motivation and Emotion 2003 27 3 199 223 10.1023/A:1025007614869 Jauk, E., & Dieterich, R. (2019). Addiction and the Dark Triad of personality. Frontiers in Psychiatry, 10. 10.3389/fpsyt.2019.00662 Jonason PK Ferrell JD Looking under the hood: The psychogenic motivational foundations of the Dark Triad Personality and Individual Differences 2016 94 324 331 10.1016/j.paid.2016.01.039 Jonason PK Krause L The emotional deficits associated with the Dark Triad traits: Cognitive empathy, affective empathy, and alexithymia Personality and Individual Differences 2013 55 5 532 537 10.1016/j.paid.2013.04.027 Jonason PK Webster GD The dirty dozen: A concise measure of the Dark Triad Psychological Assessment 2010 22 2 420 432 10.1037/a0019265 20528068 Jonason PK Koenig BL Tost J Living a fast life Human Nature 2010 21 4 428 442 10.1007/s12110-010-9102-4 Jonason PK Wee S Li NP Competition, autonomy, and prestige: Mechanisms through which the Dark Triad predict job satisfaction Personality and Individual Differences 2015 72 112 116 10.1016/j.paid.2014.08.026 Jonason PK Talbot D Cunningham ML Chonody J Higher-order coping strategies: Who uses them and what outcomes are linked to them Personality and Individual Differences 2020 155 109755 10.1016/j.paid.2019.109755 Kardefelt-Winther D A conceptual and methodological critique of internet addiction research: Towards a model of compensatory internet use Computers in Human Behavior 2014 31 351 354 10.1016/j.chb.2013.10.059 Kaufman, S. B., Yaden, D. B., Hyde, E., & Tsukayama, E. (2019). The light vs. Dark Triad of personality: contrasting two very different profiles of human nature. Frontiers in Psychology, 10. 10.3389/fpsyg.2019.00467 Kim EJ Namkoong K Ku T Kim SJ The relationship between online game addiction and aggression, self-control and narcissistic personality traits European Psychiatry 2008 23 3 212 218 10.1016/j.eurpsy.2007.10.010 18166402 King DL Koster E Billieux J Study what makes games addictive Nature 2019 573 7774 346 10.1038/d41586-019-02776-1 31530926 King DL Delfabbro PH Billieux J Potenza MN Problematic online gaming and the COVID-19 pandemic Journal of Behavioral Addictions 2020 9 2 184 186 10.1556/2006.2020.00016 32352927 Kircaburun K Griffiths MD The dark side of internet: Preliminary evidence for the associations of dark personality traits with specific online activities and problematic internet use Journal of Behavioral Addictions 2018 7 4 993 1003 10.1556/2006.7.2018.109 30427212 Kircaburun K Jonason PK Griffiths MD The Dark Tetrad traits and problematic online gaming: The mediating role of online gaming motives and moderating role of game types Personality and Individual Differences 2018 135 298 303 10.1016/j.paid.2018.07.038 Kircaburun K Demetrovics Z Tosuntaş ŞB Analyzing the links between problematic social media use, dark triad traits, and self-esteem International Journal of Mental Health and Addiction 2019 17 6 1496 1507 10.1007/s11469-018-9900-1 Kuss DJ Griffiths MD Pontes HM Chaos and confusion in DSM-5 diagnosis of Internet Gaming Disorder: Issues, concerns, and recommendations for clarity in the field Journal of Behavioral Addictions 2017 6 2 103 109 10.1556/2006.5.2016.062 27599673 Ladanyi, J., & Doyle-Portillo, S. (2017). The development and validation of the Grief Play Scale (GPS) in MMORPGs. Personality and Individual Differences, 114, 125–133. 10.1016/j.paid.2017.03.062 Loton, D., Borkoles, E., Lubman, D., & Polman, R. (2016). Video game addiction, engagement and symptoms of stress, depression and anxiety: The mediating role of coping. International Journal of Mental Health and Addiction, 14(4), 565–578. 10.1007/s11469-015-9578-6 Männikkö N Ruotsalainen H Miettunen J Pontes HM Kääriäinen M Problematic gaming behaviour and health-related outcomes: A systematic review and meta-analysis Journal of Health Psychology 2020 25 1 67 81 10.1177/1359105317740414 29192524 Mills DJ Milyavskaya M Mettler J Heath NL Exploring the pull and push underlying problem video game use: A Self-Determination Theory approach Personality and Individual Differences 2018 135 176 181 10.1016/j.paid.2018.07.007 Muris P Merckelbach H Otgaar H Meijer E The malevolent side of human nature Perspectives on Psychological Science 2017 12 2 183 204 10.1177/1745691616666070 28346115 Neufeld, A., & Malin, G. (2021). Need fulfillment and resilience mediate the relationship between mindfulness and coping in medical students. Teaching and Learning in Medicine, 1-11. 10.1080/10401334.2021.1960533 Özteke Kozan, H. İ., Baloğlu, M., Kesici, Ş., & Arpacı, İ. (2019). The role of personality and psychological needs on the problematic internet use and problematic social media use. Addicta: The Turkish Journal on Addictions, 6(2). 10.15805/addicta.2019.6.2.0029 Pace, U., D Urso, G., Ruggieri, S., Schimmenti, A., & Passanisi, A. (2021). The Role of Narcissism, hyper-competitiveness and maladaptive coping strategies on male adolescent regular gamblers: Two mediation models. Journal of Gambling Studies, 37(2), 571-582. 10.1007/s10899-020-09980-z Paulhus, D. L., & Williams, K. M. (2002). The Dark Triad of personality: Narcissism, Machiavellianism, and psychopathy. Journal of Research in Personality, 36(6), 556–563. 10.1016/S0092-6566(02)00505-6 Petry NM Rehbein F Gentile DA Lemmens JS Rumpf H Mößle T Bischof G Tao R Fung DSS Borges G Auriacombe M González Ibáñez A Tam P O'Brien CP An international consensus for assessing internet gaming disorder using the new DSM-5 approach Addiction 2014 109 9 1399 1406 10.1111/add.12457 24456155 Plante CN Gentile DA Groves CL Modlin A Blanco-Herrera J Video games as coping mechanisms in the etiology of video game addiction Psychology of Popular Media Culture 2019 8 4 385 394 10.1037/ppm0000186 Przybylski AK Weinstein N Investigating the motivational and psychosocial dynamics of dysregulated gaming: Evidence from a preregistered cohort study Clinical Psychological Science 2019 7 6 1257 1265 10.1177/2167702619859341 Sailer M Hense JU Mayr SK Mandl H How gamification motivates: An experimental study of the effects of specific game design elements on psychological need satisfaction Computers in Human Behavior 2017 69 371 380 10.1016/j.chb.2016.12.033 Şalvarlı Şİ Griffiths MD Internet gaming disorder and its associated personality traits: A systematic review using prisma guidelines International Journal of Mental Health and Addiction 2019 19 5 1420 1442 10.1007/s11469-019-00081-6 Scerri M Anderson A Stavropoulos V Hu E Need fulfilment and internet gaming disorder: A preliminary integrative model Addictive Behaviors Reports 2019 9 100144 10.1016/j.abrep.2018.100144 31193898 Schoemann AM Boulton AJ Short SD Determining power and sample size for simple and complex mediation models Social Psychological and Personality Science 2017 8 4 379 386 10.1177/1948550617715068 Sedikides C In search of Narcissus Trends in Cognitive Sciences 2021 25 1 67 80 10.1016/j.tics.2020.10.010 33229145 Segerstrom SC Smith GT Personality and coping: Individual differences in responses to emotion Annual Review of Psychology 2019 70 1 651 671 10.1146/annurev-psych-010418-102917 30265823 Sheldon KM Anderson J Integrating behavioral-motive and experiential-requirement perspectives on psychological needs: A two process model Psychological Review 2011 118 4 552 569 10.1037/a0024758 21787097 Shrestha, N. (2020). Detecting multicollinearity in regression analysis. American Journal of Applied Mathematics and Statistics, 8(2), 39–42. 10.12691/ajams-8-2-1 Şimşek ÖF Koydemir S Linking metatraits of the big five to well-being and ill-being: Do basic psychological needs matter? Social Indicators Research 2013 112 1 221 238 10.1007/s11205-012-0049-1 Sindermann C Sariyska R Lachmann B Brand M Montag C Associations between the dark triad of personality and unspecified/specific forms of Internet-use disorder Journal of Behavioral Addictions 2018 7 4 985 992 10.1556/2006.7.2018.114 30541336 Sleep CE Weiss B Lynam DR Miller JD An examination of the Triarchic model of psychopathy’s nomological network: A meta-analytic review Clinical Psychology Review 2019 71 1 26 10.1016/j.cpr.2019.04.005 31078055 Statista. (2021). Gaming in China - Statistics & Facts. Retrieved October 12, 2021, from https://www.statista.com/topics/4642/gaming-in-china/ Stride, C. B., Gardner, S., Catley, N., & Thomas, F. (2015). Mplus code for mediation, moderation, and moderated mediation models. Retrieved August 21, 2021, from http://www.offbeat.group.shef.ac.uk/FIO/mplusmedmod.htm Szabó ZP Czibor A Restás P Bereczkei T “The darkest of all” The relationship between the Dark Triad traits and organizational citizenship behavior Personality and Individual Differences 2018 134 352 356 10.1016/j.paid.2018.04.026 Tang WY Reer F Quandt T Investigating sexual harassment in online video games: How personality and context factors are related to toxic sexual behaviors against fellow players Aggressive Behavior 2019 46 1 127 135 10.1002/ab.21873 31736097 Tang WY Reer F Quandt T The interplay of gaming disorder, gaming motivations, and the dark triad Journal of Behavioral Addictions 2020 9 2 491 496 10.1556/2006.2020.00013 32544080 Vansteenkiste M Ryan RM Shahar G On psychological growth and vulnerability: Basic psychological need satisfaction and need frustration as a unifying principle Journal of Psychotherapy Integration 2013 23 3 263 280 10.1037/a0032359 Xie Y Reliability and validity of the simplified coping style questionnaire Chinese Journal of Clinical Psychology 1998 6 114 115 Zhou Y Li D Li X Wang Y Zhao L Big five personality and adolescent Internet addiction: The mediating role of coping style Addictive Behaviors 2017 64 42 48 10.1016/j.addbeh.2016.08.009 27543833	36471813	PMC9714411	NO-CC CODE	2022-12-02 23:24:45	no	Curr Psychol. 2022 Dec 1;:1-9	utf-8	Curr Psychol	2,022	10.1007/s12144-022-03996-x	oa_other
==== Front Neural Comput Appl Neural Comput Appl Neural Computing & Applications 0941-0643 1433-3058 Springer London London 8078 10.1007/s00521-022-08078-4 Original Article Harris hawks optimization for COVID-19 diagnosis based on multi-threshold image segmentation Ryalat Mohammad Hashem [email protected] 1 http://orcid.org/0000-0001-9807-7820 Dorgham Osama [email protected] [email protected] 15 Tedmori Sara [email protected] 2 Al-Rahamneh Zainab [email protected] 1 Al-Najdawi Nijad [email protected] 1 Mirjalili Seyedali [email protected] 34 1 grid.443749.9 0000 0004 0623 1491 Prince Abdullah Bin Ghazi Faculty of Information and Communication Technology, Al-Balqa Applied University, Al-Salt, 19117 Jordan 2 grid.29251.3d 0000 0004 0404 9637 King Hussein School of Computing Sciences, Princess Sumaya University for Technology, Amman, 11941 Jordan 3 grid.449625.8 0000 0004 4654 2104 Centre for Artificial Intelligence Research and Optimisation, Torrens University, Adelaide, SA 5000 Australia 4 grid.15444.30 0000 0004 0470 5454 Yonsei Frontier Lab, Yonsei University, Seoul, South Korea 5 grid.461585.b 0000 0004 1762 8208 School of Information Technology, Skyline University College, Sharjah, United Arab Emirates 1 12 2022 119 8 1 2022 22 11 2022 © The Author(s), under exclusive licence to Springer-Verlag London Ltd., 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. Digital image processing techniques and algorithms have become a great tool to support medical experts in identifying, studying, diagnosing certain diseases. Image segmentation methods are of the most widely used techniques in this area simplifying image representation and analysis. During the last few decades, many approaches have been proposed for image segmentation, among which multilevel thresholding methods have shown better results than most other methods. Traditional statistical approaches such as the Otsu and the Kapur methods are the standard benchmark algorithms for automatic image thresholding. Such algorithms provide optimal results, yet they suffer from high computational costs when multilevel thresholding is required, which is considered as an optimization matter. In this work, the Harris hawks optimization technique is combined with Otsu’s method to effectively reduce the required computational cost while maintaining optimal outcomes. The proposed approach is tested on a publicly available imaging datasets, including chest images with clinical and genomic correlates, and represents a rural COVID-19-positive (COVID-19-AR) population. According to various performance measures, the proposed approach can achieve a substantial decrease in the computational cost and the time to converge while maintaining a level of quality highly competitive with the Otsu method for the same threshold values. Keywords Harris hawks optimization Multilevel thresholding Image segmentation Otsu method Covid-19 CT images ==== Body pmcIntroduction Among the wide variety of sciences that can offer support during global pandemics, digital image processing is the one that can assist many aspects of medicine. Medical imaging systems are used to identify or study the occurrence/development of certain diseases. The domain of medical imaging is developing rapidly due to advances in image processing techniques (e.g., applied machine learning) that include image recognition, segmentation, enhancement, and analysis. Among various modalities, computed tomography (CT) can be used effectively to screen and monitor diagnosed cases and has a direct impact on health outcomes because it assists healthcare professionals to administer rapid, accurate, and large-scale diagnostic screenings [18]. CT scans of the chest may be helpful in diagnosing COVID-19 in patients with a high clinical suspicion of infection but who have not been recommended for routine screening. In this type of medical imaging, cross-sectional tomographic images (slices) of the body are produced by using multiple X-ray measurements. Chest CT scans of patients with COVID-19-associated pneumonia usually show ground-glass opacification, possibly with consolidation. Among various available medical image processing techniques and algorithms, image segmentation methods are the best tools to simplify image representation and analysis. Image segmentation is considered an essential part of the pattern recognition, computer vision, and medical image processing techniques [13, 30]. In image segmentation, the image is divided into regions of interest, and the segmentation process can be conducted by using shape, size, color, texture, illumination, etc. [16]. Image segmentation approaches can be categorized into four main types: clustering, region-based, texture analysis, and histogram thresholding. Among these various segmentation approaches, histogram thresholding is known for its robustness, simplicity, and accuracy and therefore is widely applied in image segmentation. In histogram thresholding, the histogram data are extracted from a given image and the best threshold values are used to categorize pixels in different regions [12]. For automatic image thresholding, traditional statistical approaches such as the Otsu and Kapur methods are the typical benchmark algorithms. Multilevel thresholding methods deliver better results when compared to other traditional thresholding methods [14, 17]. A wide range of multilevel segmentation methods have been proposed in the literature [40, 50]. Furthermore, obtaining optimal threshold values is considered a practical optimization problem. Therefore, many nature-inspired algorithms have been presented in recent years for solving difficult practical optimization issues. Because of their adaptability and efficiency, those algorithms have attracted the interest of both scientists and researchers. Despite their simple design, nature-inspired algorithms are effective at addressing extremely difficult optimization problems. Those algorithms are an essential component of modern global optimization algorithms, artificial intelligence, and informatics in general. Many metaheuristic algorithms have the property of reaching a global optimum for the problem after a relatively small number of iterations. Therefore, various evolutionary and swarm-based strategies have been combined with statistical thresholding methods such as the Kapur method and the Otsu method in order to find the optimal threshold values for multilevel segmentation [52]. The Harris hawks optimization (HHO) algorithm is a well-known swarm-based, gradient-free optimization technique. HHO has received much interest from scientists and researchers in terms of its performance, quality of results, and its acceptable convergence in dealing with different applications in real-world problems. In this work, the Harris hawks optimization technique is combined with Otsu's method to effectively reduce the required computational cost while maintaining optimal results. The proposed method is validated using publically accessible imaging datasets, which include chest scans with clinical and genetic correlates and reflect a COVID-19-positive (COVID-19-AR) population in rural areas. For the same threshold values, the suggested strategy can achieve a significant reduction in computing cost and time to converge while retaining a level of quality that is highly competitive with the Otsu method, according to different performance criteria. The content of the paper can be summarized as follows: Sect. 2 presents a literature review of CT-based methods for efficient lung segmentation. Section 3 describes the implementation of Harris hawks optimization algorithm for multilevel image segmentation. Section 4 outlines the Harris hawks optimization for multi-threshold image segmentation evaluation with some confirmatory ranking results and analysis. Section 5 presents the evaluation of results. Finally, Sect. 6 presents the conclusions of this paper. Literature review In the field of CT imaging, the accurate segmentation of lung can lead to the accurate diagnosis of lung infection and to the correct detection and classification of lung nodules. For this reason, various CT-based methods for efficient lung segmentation have been proposed in the literature. For example, region-growing techniques have been applied in works such as Netto et al. [31], while other works such as Keshani et al. [26] have applied active contours techniques. On the other hand, authors in Messay et al. [32], Pu et al. [43], and Tan et al. [49] have utilized thresholding strategies such as rule-based, fuzzy inference methods, and intensity in order to detect nodules in CT images. Template matching methods have also been widely used in some works such as Akram et al. [3], Jo et al. [24], and Serhat Ozekes and Ucan [47]. Meanwhile, other works such as da Silva Sousa et al. [8] and Narayanan et al. [37] have designed composite discriminative feature techniques to classify and detect lung nodules through the use of different classification methods. In addition, some other machine learning methods have been investigated for lung nodule detection, such as those presented in Furqan et al. [19], Gruetzemacher et al. [21], Jiang et al. [23], Pehrson et al. [42], and Zhang et al. [53]. Recently, contextual image information (i.e., channel and spatial) is shown to be effective for semantic segmentation as proposed by Li et al. [28], which interactively explores the spatial contextual information and the channel contextual information. In their work, the spatial contextual module is exploited to uncover the spatial contextual dependency between pixels by exploring the correlation between pixels and categories, while the semantic features are extracted using the channel contextual module by modeling the long-term semantic dependence between channels. Results show that the two context modules are adaptively integrated to achieve better outcomes, by connecting them in tandem for interactive training for mutual communication between the two dimensions. Image thresholding techniques can be categorized as either bi-level or multilevel. In bi-level techniques, designated objects are differentiated by using a single threshold value. In multilevel techniques, the image is divided into multiple regions using multiple threshold values [17]. Therefore, the multilevel thresholding techniques give more accurate segmentation results as compared to the bi-level ones [27]. Moreover, the optimal threshold values in multilevel thresholding techniques can be classified into parametric and nonparametric approaches [38]. Traditional image segmentation approaches such as the Otsu [39] and the Kapur et al. [25] methods use the maximization of the variance between classes and the histogram entropy to obtain the best possible threshold values. Although those methods obtain optimal results, they impose high computational cost that increases as the number of thresholds is increased. More recently, the use of metaheuristic algorithms in combination with traditional statistical approaches has been proved to work effectively and to reduce the computational cost when multilevel thresholding is required. Well-known automatic thresholding algorithms, such as the algorithms in the Otsu and the Kapur methods, function by performing a set of steps. These steps include processing the input image, obtaining the image histogram, computing the threshold value, and finally converting the image pixels into white in regions where the saturation is greater than the obtained threshold and into black otherwise. However, these algorithms utilize different methods to compute the threshold value. For instance, the algorithm in Otsu’s method processes the image histogram and segments the objects by minimizing the variance on each of the classes. The obtained histogram contains two expressed peaks that represent different ranges of intensity value. In Otsu’s method, the image histogram is separated into two clusters by a threshold that is defined by minimizing the weighted variance of the classes denoted by σw2t, where σw2t=w1tσ12t+w2tσ22t, which is defined as the Otsu model, where w1t,w2t are the probabilities of the two classes divided by a threshold t, where 0≤t≤255. Otsu’s method defines two alternatives to find the threshold. The first alternative is to minimize the within-class variance σw2t, and the second is to maximize the between-class variance using σb2t=w1tw2tμ1t-μ2t2, where μ1 is a mean of class i. The cluster probability function is used to calculate the probability P for each pixel value in the two separated clusters C1,C2 as w1t=∑i=1tPi and w2t=∑i=t+1IPi, respectively [48]. Digital images can be represented as an intensity function fx,y that includes gray-level values, with a total number of pixels n and the total number of pixels with a specified gray-level i. Therefore, the probability of the occurrence of gray-level i is Pi=ni/n. The pixel intensity values for C1 and C2 are within the range of 1,t and t+1,I, respectively, where I is the maximum intensity value (i.e., 255). The means for C1,C2, denoted by μ1t,μ2t, are obtained by μ1t=∑i=1tiPi/w1t and μ2t=∑i=t+1IiPi/w2t, respectively. Therefore the σ12,σ22 values can be obtained by σ12t=∑i=1ti-μ1t2Pi/w1t, and σ22t=∑i=t+1Ii-μ2t2Pi/w2t, respectively [48]. Obtaining optimal threshold values is considered a practical optimization problem. Therefore, many nature-inspired algorithms have been presented in recent years for solving difficult practical optimization problems [5–7, 33–35, 44–46]. Some of the metaheuristic algorithms are designed for image processing applications and image thresholding specifically include [2], which proposed the use of bird-mating optimization together with the Otsu and the Kapur methods as a strategy to find the best thresholds for image segmentation . On the other hand, Mousavirad and Ebrahimpour-Komleh [36] used the human mental search algorithm with Otsu’s and Kapur’s methods. Another nature-inspired approach was put forward in Pare et al. [41], in which Rényi entropy is used for locating the best thresholds using the bat algorithm. Other works have proposed using different algorithms for image segmentation, such as Liang et al. [29], which used Tsallis entropy and the grasshopper optimization algorithm, Baby Resma and Nair [4] which used the krill herd optimization algorithm and [54] which combined the firefly optimization algorithm to reduce the computation time and enhance the accuracy of image segmentation . A comprehensive review of available metaheuristic algorithms can be found in Pare et al. [41]. As presented in this section, there are several approaches that have been applied in the recent years for medical image segmentation. The first group of those approaches employed the concept of region-growing and active contours techniques to segment images and isolate the region-of-interest areas. Although this group of techniques can lead to a good level of accuracy, those approaches need to specific knowledge in the domain, and at the mean time they usually need to interact with human medical experts. This adds a pressure on human experts and also delays the process of treatment. There are other different approaches such as template matching methods and composite discriminative feature that have been also utilized by a few researchers. The main drawback of those approaches is they cannot be generalized to be used in different imaging modalities and they do not lead to convincing outcomes in some cases. On the other hand, many researchers utilized thresholding techniques to detect nodules in CT images as a base for the image segmentation. In these techniques, the maximization of the variance between classes is usually employed to segment images. Otsu method is one of the most popular techniques which is used in this domain. Although utilizing Otsu method in multilevel segmentation leads to competitive results in the medical image segmentation, the high computational cost needed by this approach is a challenging task that researchers are still trying to overcome; hence, there is the importance of employing optimization. The authors of this research employed the HHO with Otsu criterion as an objective function to overcome the high computational cost and in the meantime to achieve optimal outcomes. The Harris hawks optimization (HHO) algorithm proposed in Heidari et al. [22] is a new metaheuristic algorithm that has the potential to be used for multilevel image thresholding. This algorithm is swarm based and has been developed to efficiently handle continuous optimization tasks and produce high-quality solutions. The HHO algorithm is still relatively new and has not been tested sufficiently on real-world problems. In this research, therefore, it is applied to the multilevel image segmentation of chest images of COVID-19 patients. Its segmentation results are then analyzed and compared against those obtained by the Otsu method. The authors in this research claim that the use of metaheuristic algorithms in image segmentation domain lowers the amount of computations required to locate the best threshold configuration. Implementation of Harris hawks optimization algorithm for multilevel image segmentation Harris hawk’s optimization is a population-based swarm intelligence algorithm. It mimics the hunting strategy of Harris hawks, which is mathematically modeled to address different optimization problems. A detailed explanation of the background and fundamentals of the algorithm can be found in Heidari et al. [22]. Similar to other meta-heuristics, HHO consists of two main phases: diversification (exploration) and intensification (exploitation) which mimics the attacking strategy of Harris hawks when they are hunting their prey, where the attacking strategy is changed based on the circumstances of the prey. The attacking strategy that is simulated in HHO is explained in the following subsections. Figure 1 illustrates the phases of the HHO algorithm.Fig. 1 Phases of Harris hawks optimization [22] Diversification phase (exploration) In HHO, the Harris hawks represent the solutions; the best solution in each iteration represents the prey. Harris hawks perch randomly in some places, and they are one of the two strategies to attack their prey. The perch positions of the Harris hawks are based on the positions of other family members, as modeled in Eq. (1) for the condition q < 0.5, or they perch randomly in tall trees, which is modeled by Eq. (1) for the condition q ≥ 0.5. The first rule in the model (see Eq. 1) represents the random generation of solutions, and the second rule represents the difference between the position of the best solution (prey) and the average location of the group (hawks):1 Xt+1=Xrandt-r1\|Xrandt-2r2Xt\|q≥0.5Xrabbitt-Xmt-r3LB+r4UB-LBq≥0.5 where Xt is the position vector of a hawk in the tth iteration, Xrandt is a randomly selected hawk from the current population, Xrabbitt is the position of the rabbit prey, Xt is the current position vector of the hawks, r1, r2, r3, r4, and q are random numbers that are updated in each iteration, UB and LB are the upper bound and lower bound of the variables, respectively, and Xm is the average position of the current population of hawks. The average position Xm can be defined as shown in Eq. (2):2 Xmt=1N∑i=1NXit. Switch between diversification (exploration) and intensification (exploitation) The HHO algorithm can switch between exploration and exploitation according to the escaping energy E of the prey. The mathematical model for the energy of the prey can be defined as shown in Eq. (3):3 E=2E01-t/T In HHO, E0 randomly changes inside the interval (− 1, 1) at each iteration. This requires E to be decreased linearly proportional to the number of iterations. Intensification phase (exploitation) Harris hawks execute a surprise dive to pounce on their prey. However, the prey has the power or capability to escape from this risky situation. The prey’s chance of escaping attack can be represented by r as follows:Escape capability=successfully escapesifr<0.5unsuccessfully escapesifr≥0.5 Soft (smooth) besiege strategy If the prey has some energy, it tries to escape from the hawks by doing random jumps. However, the Harris hawks surround the prey softly to exhaust it and then execute a surprise attack. This process can happen when the chance of the prey escaping, r, equals r ≥ 0.5, (i.e., unsuccessful) and the escaping energy of the prey, E, equals E ≥ 0.5. This process can be modeled by Eqs. (4) and (5) as follows:4 Xt+1=ΔXt-E21-r5Xrabbitt-Xt 5 ΔXt=Xrabbitt-Xt, where ΔXt is the difference between the current location and the vector of the rabbit in iteration t and r5 is a random number between 0 and 1, which represents the random bounce force of the rabbit throughout the escaping criteria. Hard besiege strategy If the prey has a little escaping energy (\|E\|< 0.5) and it becomes exhausted (unsuccessfully escaping, r ≥ 0.5), the Harris hawks surround the prey and perform a surprise attack. This situation can be modeled by Eq. (6) as follows:6 Xt+1=Xrabbitt-EΔXt Soft (smooth) besiege strategy and progressive quick pounce If the prey has some energy to escape (which implies that \|E\|≥ 0.5), it can successfully escape (r < 0.5). In this case, the Harris hawks use a smooth (soft) besiege to attack the prey. During the escaping process, a simulation of the zigzag motion of the prey can be performed using a Lévy flight (LF) operator. Based on the tricky movements of the prey, the Harris hawks try to modify their pouncing strategy gradually. The Harris hawks can perform the soft besiege by deciding their next position as shown in Eq. (7):7 Y=Xrabbitt-E21-r5Xrabbitt-Xt The Harris hawks try to adjust their movement by comparing the current pounce result and the previous one. If the result is not good, they will pounce based on the LF as shown in Eq. (8):8 Z=Y+S×LFD, where D is the dimension of problem, S is a random vector of size 1 × D, and LF is the Lévy flight function. Based on the previous assumption of the soft besiege, the Harris hawks update their position by Eq. (9) as follows:9 Xt+1=YifFY<FXtZifFZ<FXt Hard besiege strategy and progressive quick pounce The Harris hawks apply the hard besiege strategy when the prey has a little energy to escape (\|E\|< 0.5) and it also has a chance to successfully escape (r < 0.5). To perform this strategy, the Harris hawks try to reduce the distance between their average position Xm and that of the prey. The overall process can be modeled by Eq. (10) as follows:10 Xt+1=YifFY<FXtZifFZ<FXt, where Y and Z are obtained using the new rules in Eqs. (11) and (12), respectively:11 Y=Xrabbitt-E21-r5Xrabbitt-Xmt 12 Z=Y+S×LFD Pseudocode of Harris hawks optimization algorithm for multilevel thresholding The pseudocode of the procedures of the HHO algorithm employed for multilevel thresholding in this research is described in Algorithm 1. There are three primary phases in the HHO algorithm. The following are the processes and their computational complexity (i.e., time complexity) [20].Initiation process →ON. Updating the locations of all Harris hawks→OT×N×D. Determining the best location →OT×N. where T is the maximum iteration, N: population size, D: the dimension of the problem. Therefore, the total complexity of the HHO algorithm can be described as O(N×T+TD+1. Experiment and results The World Health Organization declared the outbreak of the novel coronavirus (2019-nCoV) an ongoing global pandemic in March 2020. As of today, tens of millions of cases have been confirmed, and with millions of lives lost due to COVID-19, it is becoming one of the deadliest pandemics in history. This pandemic has shown that public health is not only a medical problem; it has also become the main common concern of all scientific fields. Therefore, interdisciplinary teams of scientists from all over the world have conducted novel research and studies to find efficient solutions that can be used to control the consequences of the pandemic and to prevent its return and the emergence of similar pandemics. Dataset In order to evaluate the efficiency of the proposed approach, a publicly available imaging dataset was obtained from the Cancer Imaging Archive. The obtained dataset consists of a set of CT images (762 × 762 pixels with a spacing 1.08 × 1.08 mm and 0.98 × 0.98 mm and a slice thickness 3.14 mm) [9–11]. The dataset contains chest imaging with clinical and genomic correlates and represents a rural COVID-19-positive population (COVID-19-AR). The dataset contains a total of about 105 patients. For the purpose of this research, 20 randomly selected patient records (hereinafter referred to as 10 individual datasets) were used to test and evaluate the proposed approach. Experimental settings Prior to the experiment, the initial parameter values were set and tuned experimentally (see Table 1). In the experiment, the proposed algorithm starts by generating 30 different possible solutions, where each solution represents the set of possible thresholds. The number of iterations is set to 150 iterations to allow the analysis of the convergence behavior of the algorithm. In order to evaluate the stability and the reliability of the generated outcomes, the experiment is repeated 25 times. The lower bound (LB) and the upper bound (UB) are set to 0 and 255, respectively (for grayscale images), and the value of the dimension parameter in the algorithm is set dynamically in accordance with the number of thresholds in each specific experiment.Table 1 Initial parameter values of the HHO algorithm Parameter name Parameter value Population size 30 Number of iterations 150 Number of runs 25 LB 0 UB 255 Dimension Number of thresholds Results Table 2 presents a sample from each dataset and presents a comparison between the outcomes generated by the Otsu method as a ground truth against the threshold values generated by the proposed approach. To keep Table 2 short and readable, the outcomes of only the first 10 datasets are presented in the table. However, the full details of all the experimental outcomes can be accessed via the URL linked to this paper (https://sites.google.com/view/hhocovid19/home). In Table 2, the third and fourth columns present the Otsu threshold values and the threshold values generated by the proposed approach, respectively. The fourth column presents the most frequent threshold values that appeared after the 25th execution of the algorithm. The last column provides the number of times (NOT), which is the number of times out of 25 that the most frequent threshold values appeared.Table 2 Threshold values generated by the Otsu method versus threshold values generated by the proposed approach ID Image Otsu methodthreshold value Proposed approach threshold value NOT 16406488 56, 173 56, 173 24 45, 110, 187 43, 109, 186 24 39, 100, 150, 213 39, 100, 150, 213 16 33, 85, 116, 156, 216 32, 83, 115, 156, 215 11 16406490 45, 159 45, 159 25 26, 88, 181 26, 88, 180 14 25, 84, 146, 211 24, 69, 113, 187 10 23, 66, 106, 157, 217 23, 66, 106, 157, 217 5 16406498 32, 110 32, 110 25 29, 96, 165 29, 96, 165 18 28, 77, 124, 175 26, 76, 124, 175 8 25, 75, 121, 163, 209 25, 75, 121, 163, 209 13 16406502 36, 143 36, 142 25 24, 81, 170 24, 81, 170 13 24, 76, 138, 206 23, 76, 138, 205 17 22, 62, 100, 150, 212 21, 62, 100, 150, 211 5 16406503 50, 154 50, 154 25 28, 92, 175 28, 92, 175 22 28, 88, 153, 214 27, 88, 152, 212 17 24, 68, 109, 160, 216 23, 68, 109, 160, 216 6 16406513 61, 172 60, 171 25 31, 98, 185 31, 98, 185 24 29, 90, 145, 208 29, 90, 145, 207 8 24, 69, 108, 151, 211 23, 68, 108, 151, 210 13 16407187 61, 170 61, 170 25 31, 97, 181 31, 97, 180 20 30, 92, 143, 201 28, 81, 124, 187 22 27, 75, 115, 155, 210 26, 75, 115, 155, 210 13 16424071 63, 177 63, 177 25 22, 86, 182 22, 86, 181 24 19, 63, 110, 185 19, 63, 110, 184 12 19, 61, 107, 158, 216 19, 61, 107, 157, 215 7 16424081 58, 169 58, 169 25 20, 78, 176 20, 78, 176 25 17, 61, 108, 178 17, 58, 108, 178 17 15, 42, 78, 119, 189 14, 41, 77, 118, 188 14 16424106 29, 104 29, 104 25 26, 86, 164 26, 85, 163 18 23, 67, 113, 177 23, 67, 113, 177 21 20, 55, 92, 137, 195 20, 55, 92, 137, 195 19 It is obvious from Table 2 that the values generated by the proposed approach are either the same or very close to the Otsu threshold values. Moreover, even in the cases where there is no complete match between the outcomes of the Otsu method and those generated by the proposed approach, the difference is negligible. The two main inputs that are fed into the proposed approach are the group of CT images and the number of thresholds. The optimal threshold values are generated as output. The optimizer is employed to maximize the value of the between-class variance, as suggested by the Otsu criterion. Table 3 presents the average fitness values (i.e., between-class variance) that are achieved when the proposed approach is applied to the 10 datasets. Figure 2 displays the histograms of five different images picked randomly from different datasets along with the threshold values which are represented as red vertical lines. These threshold values represent the outcome of applying the proposed approach.Table 3 Average between-class variance value for each dataset over 25 runs # Patient ID # Of thr. Avg. fitness # Patient ID # Of thr. Avg. fitness 1 16406488 2 1,995,968,548.14 11 16424120 2 1,325,763,624.57 3 2,025,402,282.08 3 1,373,514,362.89 4 2,040,028,586.97 4 1,389,606,981.23 5 2,045,857,463.56 5 1,398,038,015.74 2 16406490 2 2,298,088,159.24 12 16434363 2 3,524,501,058.11 3 2,436,038,694.92 3 3,710,226,862.57 4 2,453,919,131.42 4 3,745,843,560.59 5 2,466,483,199.24 5 3,766,274,305.97 3 16406498 2 2,076,199,600.90 13 16434368 2 2,320,389,483.76 3 2,130,226,824.33 3 2,404,024,664.02 4 2,154,189,341.66 4 2,430,796,897.84 5 2,162,355,593.61 5 2,441,260,652.15 4 16406502 2 1,940,207,310.35 14 16434369 2 2,602,673,414.62 3 2,040,921,175.96 3 2,749,944,666.81 4 2,062,553,214.82 4 2,777,323,399.19 5 2,074,264,187.22 5 2,798,949,958.41 5 16406503 2 3,012,563,719.35 15 16434411 2 2,830,098,693.27 3 3,151,883,519.94 3 2,914,157,752.27 4 3,178,821,393.68 4 2,931,386,952.33 5 3,195,958,229.64 5 2,939,713,611.57 6 16406513 2 3,265,136,378.30 16 16434453 2 3,225,924,532.16 3 3,371,021,181.84 3 3,389,121,940.72 4 3,405,659,240.79 4 3,415,523,456.42 5 3,422,308,473.62 5 3,434,394,061.24 7 16407187 2 2,340,736,567.15 17 16445122 2 2,355,216,836.24 3 2,438,359,772.60 3 2,492,331,087.81 4 2,455,628,797.20 4 2,514,181,165.77 5 2,468,027,367.69 5 2,524,959,336.26 8 16424071 2 2,299,778,005.74 18 16445138 2 2,812,376,551.29 3 2,407,441,253.39 3 2,962,616,506.72 4 2,428,542,182.15 4 2,995,377,758.74 5 2,437,592,717.67 5 3,012,410,682.07 9 16424081 2 1,857,758,173.25 19 16445144 2 2,269,545,641.31 3 1,994,137,148.17 3 2,380,108,941.03 4 2,026,029,797.17 4 2,401,661,550.92 5 2,041,342,618.95 5 2,413,854,491.02 10 16424106 2 2,724,258,369.10 20 16445168 2 3,891,245,700.61 3 2,860,700,119.06 3 3,941,055,911.03 4 2,899,143,433.02 4 3,965,460,390.77 5 2,921,678,998.55 5 3,976,172,936.86 Fig. 2 Histograms of five images from different datasets with a different number of thresholds Evaluation of results In order to evaluate the efficiency of the proposed approach, the optimal thresholds obtained from the experiment are supplied to the evaluation process as an input. The evaluation starts, as illustrated in Fig. 3, by generating a segmented image in accordance with the optimal thresholds. The generated segmented image and the Otsu segmented image (i.e., ground truth) are then compared and tested to determine the degree of similarity.Fig. 3 Basic inputs and outputs in the evaluation process Three different similarity metrics are used to evaluate the accuracy of the experiment: the peak signal-to-noise ratio (PSNR), the structural similarity index (SSIM), and the Jaccard index. These similarity metrics are explained in the following subsections. Peak signal-to-noise ratio (PSNR) The complications associated with subjective quality assessment necessitate the use of automated objective quality assessment methods to measure the quality differences between processed images. Among the well-known objective quality assessment methods is the PSNR, which is based on the mean squared error (MSE) and can be calculated as follows [51]:MSE=∑i=1N(xi-yi)2, where N is the number of pixels in the given image (both images must be of the same size) and xi, yi are the pixel values in both of the images to be tested. The PSNR can be calculated as follows:PSNR=20log10K/MSE where K is the maximum range of the image pixel values (K = 255 for grayscale images). The PSNR and MSE are high-quality measures for testing the proposed approach. The PSNR is widely used in digital image processing applications. In this work, the PSNR gives a value of 100 dB under one condition only, i.e., when both images (to be compared) are identical, which leads to a division by 0, that is controlled by a conditional statement. When the PSNR result exceeds 30 dB, the human visual system would find it more difficult to differentiate between the two images. However, when the PSNR result is lower than 30 dB, this indicates that the human visual system would have the ability to notice the differences between the two images. Structural similarity index (SSIM) The SSIM is another well-known measure that can be used to assess the quality of the output generated by the proposed approach. The SSIM is based on the computation of three factors: luminance, contrast, and structure. The SSIM can be defined as follows [1]:SSIMx,y=lx,yα×cx,yβ×sx,yγ, where x,y=2μxμy+C1/(2μx2+C1), cx,y=2σxσy+C2/(2σx2+C2), and sx,y=σxy+C3/(σxσy+C3), where μx,μy,σx,σy, and σxy are the local means, standard deviations, and cross-covariance of the two compared images, respectively. If α=β=γ=1 and C3=C2/2, the equation can be simplified as follows:SSIMx,y=2μxμy+C1(2σxy+C2)μx2+μy+2C1σx2+σy2+C2 The SSIM results are given in values that range between [0, 1], where 1 indicates a perfect match between the two images and 0 indicates no match. Jaccard similarity index The Jaccard similarity index measures the similarity of two sets of data (two images) on a range from 0 to 100%. This similarity index compares the pixels in two images to see which of them are shared and which are distinct. It counts the number of pixels that are shared by the two images, counts the total number of pixels in both images, divides the number of shared pixels by the total number of pixels, and finally multiplies the number by 100. This percentage gives the similarity between the two images. This similarity index is simple and vulnerable to small samples sizes. It can be defined as follows [15]:JX,Y=X∩Y/X∪Y Similarity results Table 4 presents the average value of the PSNR between each segmented image (i.e., generated by the proposed approach) and its respective ground-truth image (i.e., generated by the Otsu method) that was obtained after repeating the same experiment 25 times. The table shows the average PSNR values when 2, 3, 4, and 5 thresholds are applied to segment the datasets.Table 4 Average value of the PSNR (20 datasets, 25 different runs for each experiment) # Patient ID # Of thr. PSNR # Patient ID # Of thr. PSNR 1 16406488 2 99.32582 11 16424120 2 100 3 74.86861 3 81.04659 4 94.14886 4 78.23629 5 72.37319 5 82.44525 2 16406490 2 100 12 16434363 2 100 3 82.19359 3 77.21508 4 59.37486 4 70.61255 5 78.81656 5 73.18433 3 16406498 2 100 13 16434368 2 100 3 94.89771 3 97.4049 4 74.64346 4 76.29251 5 89.6506 5 68.79373 4 16406502 2 82.22882 14 16434369 2 100 3 90.61202 3 100 4 78.26527 4 56.58543 5 76.76858 5 83.73876 5 16406503 2 100 15 16434411 2 100 3 97.62194 3 97.41837 4 73.58808 4 74.54786 5 76.39956 5 57.7601 6 16406513 2 74.13531 16 16434453 2 100 3 80.21965 3 77.90913 4 77.67605 4 68.96586 5 70.98543 5 74.70691 7 16407187 2 100 17 16445122 2 83.64803 3 81.48877 3 97.52143 4 58.98134 4 81.85069 5 81.63743 5 65.45271 8 16424071 2 100 18 16445138 2 100 3 84.32184 3 83.75593 4 76.03031 4 88.94338 5 70.7011 5 78.00936 9 16424081 2 100 19 16445144 2 100 3 100 3 94.8936 4 66.61292 4 77.19462 5 68.36883 5 76.91706 10 16424106 2 100 20 16445168 2 100 3 72.29532 3 82.15853 4 95.68653 4 79.26923 5 92.65646 5 76.03681 The PSNR values presented in Table 4 indicate that the proposed approach is able to achieve very accurate results, where the PSNR value of 100 dB clearly indicates a high level of accuracy. Tables 5 and 6 present the average values of the SSIM and Jaccard index, respectively. Three main observations can be made about these outcomes. First, there is a high match between the segmented images and the ground-truth images when the ratio between the matched pixels and the total number of pixels is calculated. Second, there is a general tendency by the proposed approach to achieve better results when the number of thresholds is relatively small. Third, most of the records in Tables 5 and 6 are segmented at accuracies greater than 98%, which clearly confirms the robustness of the results generated by the proposed approach.Table 5 Average value of the SSIM (20 datasets, 25 different runs for each experiment) # Patient ID # Of thr. SSIM # Patient ID # Of thr. SSIM 1 16406488 2 0.999923 11 16424120 2 1 3 0.992272 3 0.998286 4 0.999545 4 0.996793 5 0.991701 5 0.997164 2 16406490 2 1 12 16434363 2 1 3 0.997395 3 0.993362 4 0.897879 4 0.98207 5 0.991361 5 0.988317 3 16406498 2 1 13 16434368 2 1 3 0.999283 3 0.999411 4 0.991874 4 0.992309 5 0.997574 5 0.980276 4 16406502 2 0.997756 14 16434369 2 1 3 0.997689 3 1 4 0.99476 4 0.818353 5 0.995071 5 0.99308 5 16406503 2 1 15 16434411 2 1 3 0.999434 3 0.999388 4 0.987911 4 0.990381 5 0.990183 5 0.851039 6 16406513 2 0.986831 16 16434453 2 1 3 0.996997 3 0.993229 4 0.994794 4 0.967014 5 0.984312 5 0.982524 7 16407187 2 1 17 16445122 2 0.998781 3 0.997659 3 0.999349 4 0.878351 4 0.997061 5 0.996419 5 0.94744 8 16424071 2 1 18 16445138 2 1 3 0.997648 3 0.998709 4 0.991567 4 0.996321 5 0.959043 5 0.994599 9 16424081 2 1 19 16445144 2 1 3 1 3 0.999081 4 0.954849 4 0.995619 5 0.971602 5 0.994153 10 16424106 2 1 20 16445168 2 1 3 0.978961 3 0.998022 4 0.997128 4 0.996986 5 0.99353 5 0.990996 Table 6 Average value of the Jaccard index (20 datasets, 25 different runs for each experiment) # Patient ID # Of thr. Jaccard # Patient ID # Of thr. Jaccard 1 16406488 2 0.999955 11 16424120 2 1 3 0.981613 3 0.993089 4 0.999077 4 0.993548 5 0.958543 5 0.988738 2 16406490 2 1 12 16434363 2 1 3 0.997526 3 0.992903 4 0.645924 4 0.978777 5 0.986191 5 0.984631 3 16406498 2 1 13 16434368 2 1 3 0.999675 3 0.999716 4 0.989342 4 0.994441 5 0.995199 5 0.959493 4 16406502 2 0.996841 14 16434369 2 1 3 0.999401 3 1 4 0.993744 4 0.544799 5 0.988634 5 0.992978 5 16406503 2 1 15 16434411 2 1 3 0.999844 3 0.999431 4 0.980146 4 0.982615 5 0.981625 5 0.617592 6 16406513 2 0.989879 16 16434453 2 1 3 0.995785 3 0.994461 4 0.992992 4 0.965739 5 0.973302 5 0.973598 7 16407187 2 1 17 16445122 2 0.996711 3 0.995345 3 0.999807 4 0.678648 4 0.995945 5 0.993707 5 0.867911 8 16424071 2 1 18 16445138 2 1 3 0.997077 3 0.997058 4 0.985951 4 0.997613 5 0.889522 5 0.992204 9 16424081 2 1 19 16445144 2 1 3 1 3 0.999021 4 0.93825 4 0.986873 5 0.969069 5 0.993515 10 16424106 2 1 20 16445168 2 1 3 0.986755 3 0.997408 4 0.998627 4 0.993114 5 0.992681 5 0.978976 Discussion and analysis The average total time (in sec) required to process each image is shown in Table 7, from which it can be noted that the proposed approach converges in the early stages before reaching 150 iterations. As shown in the table, there is a significant difference between the total time to complete 150 iterations and the time needed to converge. This further reinforces the findings on the effectiveness and suitability of the proposed approach for medical image segmentation because it can find a fast and accurate solution in a domain where both time and accuracy are crucial factors in real-time medical applications.Table 7 Average total time for 150 iterations and the time to converge (in sec) for each run # Patient ID # Of thr. Total time Time to converge # Patient ID # Of thr. Total time Time to converge 1 16406488 2 99.33306 14.70808 11 16424120 2 103.087 11.59913 3 99.4228 32.68711 3 102.4821 21.29996 4 100.0869 47.43099 4 101.1351 42.59933 5 100.8867 59.37802 5 101.1553 64.76624 2 16406490 2 120.8951 9.951318 12 16434363 2 132.9661 14.82821 3 120.9023 48.20626 3 133.4617 45.54483 4 125.0947 59.53491 4 133.6005 53.71276 5 122.7947 65.88505 5 137.0266 70.90094 3 16406498 2 119.2373 13.70028 13 16434368 2 123.3992 9.125473 3 119.8075 42.18315 3 124.3 42.3352 4 119.7914 62.41246 4 123.3049 59.64479 5 119.052 64.44114 5 121.9602 64.79661 4 16406502 2 122.1041 19.12062 14 16434369 2 140.0901 17.89299 3 120.5432 37.49903 3 135.9724 37.27247 4 121.0931 53.75128 4 135.4112 53.74089 5 121.3464 76.15637 5 135.3814 64.44001 5 16406503 2 129.1814 17.3181 15 16434411 2 116.8461 28.92963 3 128.1719 43.99658 3 116.4991 42.26079 4 129.4383 63.27366 4 116.7608 61.45203 5 129.0465 73.70594 5 116.8804 60.08481 6 16406513 2 120.871 10.91851 16 16434453 2 132.6529 9.854632 3 123.0058 30.695 3 131.05 46.19639 4 125.866 59.49019 4 131.469 45.90509 5 126.3875 73.85112 5 132.551 61.38263 7 16407187 2 118.8549 19.50727 17 16445122 2 118.8132 16.70446 3 118.9825 44.72838 3 116.1046 31.02404 4 118.6215 50.22636 4 115.3362 44.98582 5 118.9379 71.74427 5 116.0373 62.20557 8 16424071 2 127.3556 12.85624 18 16445138 2 147.0102 16.6875 3 127.5319 54.22377 3 147.54 35.463 4 127.1523 53.90645 4 147.6751 53.03595 5 133.0665 89.02275 5 149.806 75.82007 9 16424081 2 146.3925 14.39858 19 16445144 2 113.0392 17.31427 3 159.8886 40.56328 3 111.7535 40.76643 4 147.8028 52.99171 4 112.3271 46.3613 5 150.3438 78.41631 5 112.0688 65.08288 10 16424106 2 158.2036 15.30983 20 16445168 2 114.4999 23.31525 3 165.5247 37.40362 3 114.7055 43.63497 4 162.9541 52.89718 4 115.0052 62.77938 5 162.7715 81.35365 5 115.9229 62.69237 As shown in Table 7, there is a positive correlation between the time to converge and the number of thresholds. Therefore, the time to converge is proportional to the number of segmented clusters. It is remarkable that the time to converge in most images is about 7–15% of the total time when the number of thresholds = 2. Moreover, even when the number of thresholds = 5, which is a case considered to require a longer time for convergence to be achieved, the time to converge in most images is about 50–55% of the total time. Figure 4 shows the convergence plots of three sample images at different threshold values. It is evident that the convergence behavior is present in all of the images regardless of the number of thresholds.Fig. 4 Convergence plots of sample images Table 8 presents the average number of iterations at which the convergence starts for each dataset. It can be seen that the number of iterations required to converge is proportional to the number of thresholds. However, as also shown in the table, a small number of iterations can lead to the required convergence and achieve optimal results.Table 8 Average number of iterations at which convergence starts # Patient ID # Of thr. Avg. # of iter. # Patient ID # Of thr. Avg. # of iter. 1 16406488 2 19.76 11 16424120 2 13.84 3 48.88 3 30.36 4 71.76 4 64.24 5 87.48 5 95.04 2 16406490 2 8.2 12 16434363 2 13.68 3 60.88 3 51.36 4 72.36 4 61.6 5 80.72 5 78.68 3 16406498 2 14.52 13 16434368 2 7.08 3 53.32 3 52 4 79.12 4 73.16 5 81.52 5 80.56 4 16406502 2 22.08 14 16434369 2 16.68 3 47.48 3 41.56 4 66.84 4 60.52 5 93.16 5 72.56 5 16406503 2 17.56 15 16434411 2 34.2 3 51.92 3 54.52 4 73.28 4 79.24 5 86.16 5 77.68 6 16406513 2 9.72 16 16434453 2 7.12 3 36.92 3 53.04 4 71.64 4 53.36 5 87.72 5 71.16 7 16407187 2 22.28 17 16445122 2 18.36 3 56.32 3 40.28 4 64.68 4 59.72 5 90.2 5 80.72 8 16424071 2 11.52 18 16445138 2 14.32 3 62.8 3 35.96 4 64.2 4 54.8 5 99.2 5 76.64 9 16424081 2 11.6 19 16445144 2 20.84 3 38.44 3 54.2 4 55.12 4 63.32 5 79.08 5 87.08 10 16424106 2 11.08 20 16445168 2 29.48 3 33.72 3 56.88 4 49.96 4 82.12 5 75.64 5 81.4 # of thresholds # Of thr. = 2 # Of thr. = 3 # Of thr. = 4 # Of thr. = 5 # of iterations 16.196 48.042 66.052 83.12 The main contribution of this research is to achieve the optimal results provided by Otsu’s algorithm while trying to reduce the massive number of computations it require. A set of experiments have been designed in order to compare and prove (a) the accuracy of outcomes and how they are matching the outcomes generated by Otsu, (b) the degree of similarity between the proposed approach and Otsu, and (c) the required time that the proposed approach need to reach the solution. In Table 2, the threshold values generated by the Otsu method (i.e., the third column) are compared with the threshold values generated by the proposed approach (i.e., the fourth column). The values in the two columns are either the same or very close which proves the accuracy of outcomes generated by the proposed approach. Furthermore, Table 3 presents the average between-class variance which is an indicator to the accuracy of the outcomes. All results and comparisons are presented in details in the URL linked to this paper (https://sites.google.com/view/hhocovid19/home). In Tables 4, 5, and 6, the degree of similarity between each segmented image (i.e., generated by the proposed approach) and its respective ground-truth image (i.e., generated by the Otsu method) is presented using PSNR, SSIM, and Jaccard index, respectively. The outcomes in these tables confirm again the accurate results that are generated by the proposed approach and how they are competitive when compared with the original method. In order to present the required time that the proposed approach needs, Tables 7 and 8 present the average total time and the average number of iterations at which convergence starts since the original method (i.e., Otsu’s original method) follows the brute force approach in which all combinations are evaluated. Conclusion Digital image processing algorithms have been supporting the medical field over the last few decades, and a huge number of algorithms have been specifically designed to assist radiologists and specialists to identify or study the occurrence or the development of diseases. The universal transmission of COVID-19 has encouraged researchers in this domain to develop medical imaging solutions to help medical experts with decision making and diagnosis. Image segmentation methods that are based on thresholding are considered an essential part of various domains including medical image processing. However, multilevel thresholding methods give improved segmentation results when compared to the standard single thresholding methods. Among the wide variety of available segmentation methods, the traditional standard benchmark statistical methods give optimal results at the cost of intensive computation when used for multilevel thresholding. Therefore, various metaheuristic optimization algorithms have been proposed and combined with statistical thresholding algorithms to reduce the number of required computations. In this research, the recent metaheuristic HHO algorithm was combined with the standard benchmark Otsu algorithm to perform image segmentation using multilevel image thresholding. The HHO algorithm is swarm based and has been proposed to efficiently handle continuous optimization tasks and produce high-quality solutions. The proposed approach was tested on a publicly available imaging dataset that contains chest images with clinical and genomic correlates and represents a rural COVID-19-positive population (COVID-19-AR). Three well-known similarity metrics (PSNR, SSIM, and Jaccard index) were used to evaluate the accuracy of the results generated by the proposed approach against those produced by the Otsu method. After running the experiments 25 times, the average value of the PSNR was around 97%, 88%, 76%, and 76% when the number of thresholds was 2, 3, 4, and 5, respectively. The average value of the SSIM showed an overlap match of 99.9%, 99.7%, 97.9%, and 97.2% for 2, 3, 4, and 5 thresholds, respectively, and the Jaccard index showed a similarity of 99.9%, 99.6%, 95.4%, and 93.2% for the same number of thresholds. These results indicate that the proposed approach can achieve a massive reduction in the computational cost and simultaneously match the quality of images produced by the Otsu method for the same threshold values. It is worth mentioning here that the number of iterations in the experiment was limited to 150 due to the limitation of time. Thus, we envisage that forcing the algorithm to iterate a larger number of times will lead to better outcomes in terms of the degree of matching quality. Declarations Conflicts of interest The authors declare no conflict of interest. Data availability Publication of the annotated images database is available through this public repository (https://sites.google.com/view/hhocovid19/data-set). Code availability The source code is available at this public repository (https://sites.google.com/view/hhocovid19/source-code). Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Abdel-Basset M Mohamed R Elhoseny M Chakrabortty RK Ryan M A hybrid COVID-19 detection model using an improved marine predators algorithm and a ranking-based diversity reduction strategy IEEE Access 2020 8 79521 79540 10.1109/ACCESS.2020.2990893 2. Ahmadi M Kazemi K Aarabi A Niknam T Helfroush MS Image segmentation using multilevel thresholding based on modified bird mating optimization Multimed Tools Appl 2019 78 16 23003 23027 10.1007/s11042-019-7515-6 3. Akram S Javed MY Akram MU Qamar U Hassan A Pulmonary nodules detection and classification using hybrid features from computerized tomographic images J Med Imaging Health Inform 2016 6 1 252 259 10.1166/jmihi.2016.1600 4. Baby Resma KP Nair MS Multilevel thresholding for image segmentation using krill herd optimization algorithm J King Saud Univ Comput Inf Sci 2018 10.1016/j.jksuci.2018.04.007 5. Chen H Heidari AA Zhao X Zhang L Chen H Advanced orthogonal learning-driven multi-swarm sine cosine optimization: framework and case studies Expert Syst Appl 2020 144 113113 10.1016/j.eswa.2019.113113 6. Chen H Li S Heidari AA Wang P Li J Yang Y Wang M Huang C Efficient multi-population outpost fruit fly-driven optimizers: framework and advances in support vector machines Expert Syst Appl 2020 142 112999 10.1016/j.eswa.2019.112999 7. Chen H Yang C Heidari AA Zhao X An efficient double adaptive random spare reinforced whale optimization algorithm Expert Syst Appl 2020 154 113018 10.1016/j.eswa.2019.113018 8. da Silva Sousa JRF Silva AC de Paiva AC Nunes RA Methodology for automatic detection of lung nodules in computerized tomography images Comput Methods Programs Biomed 2010 98 1 1 14 10.1016/j.cmpb.2009.07.006 19709774 9. Desai S Baghal A Wongsurawat T Jenjaroenpun P Powell T Al-Shukri S Gates K Farmer P Rutherford M Blake G Nolan T Chest imaging representing a COVID-19 positive rural U.S. population Sci Data 2020 7 1 414 10.1038/s41597-020-00741-6 33235265 10. Desai S, Baghal A, Wongsurawat T, Al-Shukri S, Gates K, Farmer P, Rutherford M, Blake GD, Nolan T, Powell T, Sexton K, Bennett W, Prior F (2020) Data from chest imaging with clinical and genomic correlates representing a rural COVID-19 positive population data set. The Cancer Imaging Archive 11. Desai S, Baghal A, Wongsurawat T, Al-Shukri S, Gates K, Farmer P, Rutherford M, Blake GD, Nolan T, Powell T, Sexton K, Bennett W, Prior F (2020) Chest imaging with clinical and genomic correlates representing a rural COVID-19 positive population 12. Dorgham O Fisher M Laycock S Performance of a 2D–3D image registration system using (lossy) compressed x-ray CT Ann BMVA 2009 3 1 11 13. Dorgham O Ryalat MH Naser MA Automatic body segmentation for accelerated rendering of digitally reconstructed radiograph images Inform Med Unlocked 2020 20 100375 10.1016/j.imu.2020.100375 14. Dorgham OM Alweshah M Ryalat MH Alshaer J Khader M Alkhalaileh S Monarch butterfly optimization algorithm for computed tomography image segmentation Multimed Tools Appl 2021 10.1007/s11042-020-10147-6 15. Eelbode T Bertels J Berman M Vandermeulen D Maes F Bisschops R Blaschko MB Optimization for medical image segmentation: theory and practice when evaluating with dice score or jaccard index IEEE Trans Med Imaging 2020 39 11 3679 3690 10.1109/TMI.2020.3002417 32746113 16. Elaziz MA Lu S Many-objectives multilevel thresholding image segmentation using knee evolutionary algorithm Expert Syst Appl 2019 125 305 316 10.1016/j.eswa.2019.01.075 17. Elaziz MA Oliva D Ewees AA Xiong S Multi-level thresholding-based grey scale image segmentation using multi-objective multi-verse optimizer Expert Syst Appl 2019 125 112 129 10.1016/j.eswa.2019.01.047 18. Fisher M Dorgham O Laycock SD Fast reconstructed radiographs from octree-compressed volumetric data Int J Comput Assist Radiol Surg 2013 8 2 313 322 10.1007/s11548-012-0783-5 22821505 19. Furqan S Gulistan R Alejandro FF Computer-aided detection of lung nodules: a review J Med Imaging 2019 6 2 1 11 10.1117/1.JMI.6.2.020901 20. Gezici H Livatyalı H Chaotic Harris hawks optimization algorithm J Comput Des Eng 2022 9 1 216 245 10.1093/jcde/qwab082 21. Gruetzemacher R Gupta A Paradice D 3D deep learning for detecting pulmonary nodules in CT scans J Am Med Inform Assoc 2018 25 10 1301 1310 10.1093/jamia/ocy098 30137371 22. Heidari AA Mirjalili S Faris H Aljarah I Mafarja M Chen H Harris hawks optimization: algorithm and applications Futur Gener Comput Syst 2019 97 849 872 10.1016/j.future.2019.02.028 23. Jiang H Ma H Qian W Gao M Li Y An automatic detection system of lung nodule based on multigroup patch-based deep learning network IEEE J Biomed Health Inform 2018 22 4 1227 1237 10.1109/JBHI.2017.2725903 28715341 24. Jo HH Hong H Mo Goo J Pulmonary nodule registration in serial CT scans using global rib matching and nodule template matching Comput Biol Med 2014 45 87 97 10.1016/j.compbiomed.2013.10.028 24480168 25. Kapur JN Sahoo PK Wong AKC A new method for gray-level picture thresholding using the entropy of the histogram Comput Vis Graph Image Process 1985 29 3 273 285 10.1016/0734-189X(85)90125-2 26. Keshani M Azimifar Z Tajeripour F Boostani R Lung nodule segmentation and recognition using SVM classifier and active contour modeling: a complete intelligent system Comput Biol Med 2013 43 4 287 300 10.1016/j.compbiomed.2012.12.004 23369568 27. Kumar AS, Kumar A, Bajaj V, Singh, GK (2018) Fractional-order darwinian swarm intelligence inspired multilevel thresholding for mammogram segmentation. Paper presented at the 2018 international conference on communication and signal processing (ICCSP) 28. Li Z Sun Y Zhang L Tang J CTNet: context-based tandem network for semantic segmentation IEEE Trans Pattern Anal Mach Intell 2021 10.1109/TPAMI.2021.3132068 29. Liang H Jia H Xing Z Ma J Peng X Modified grasshopper algorithm-based multilevel thresholding for color image segmentation IEEE Access 2019 7 11258 11295 10.1109/ACCESS.2019.2891673 30. Shapiro LG Stockman GC Computer vision 2001 1 Hoboken Prentice Hall 31. Magalhães Barros Netto S Corrêa Silva A Acatauassú Nunes R Gattass M Automatic segmentation of lung nodules with growing neural gas and support vector machine Comput Biol Med 2012 42 11 1110 1121 10.1016/j.compbiomed.2012.09.003 23021776 32. Messay T Hardie RC Rogers SK A new computationally efficient CAD system for pulmonary nodule detection in CT imagery Med Image Anal 2010 14 3 390 406 10.1016/j.media.2010.02.004 20346728 33. Mirjalili S Gandomi AH Mirjalili SZ Saremi S Faris H Mirjalili SM Salp swarm algorithm: a bio-inspired optimizer for engineering design problems Adv Eng Softw 2017 114 163 191 10.1016/j.advengsoft.2017.07.002 34. Mirjalili S Lewis A The whale optimization algorithm Adv Eng Softw 2016 95 51 67 10.1016/j.advengsoft.2016.01.008 35. Mirjalili S Mirjalili SM Lewis A Grey wolf optimizer Adv Eng Softw 2014 69 46 61 10.1016/j.advengsoft.2013.12.007 36. Mousavirad SJ Ebrahimpour-Komleh H Human mental search-based multilevel thresholding for image segmentation Appl Soft Comput 2020 97 105427 10.1016/j.asoc.2019.04.002 37. Narayanan BN Hardie RC Kebede TM Sprague MJ Optimized feature selection-based clustering approach for computer-aided detection of lung nodules in different modalities Pattern Anal Appl 2019 22 2 559 571 10.1007/s10044-017-0653-4 38. Oliva D Hinojosa S Osuna-Enciso V Cuevas E Pérez-Cisneros M Sanchez-Ante G Image segmentation by minimum cross entropy using evolutionary methods Soft Comput 2019 23 2 431 450 10.1007/s00500-017-2794-1 39. Otsu N A threshold selection method from gray-level histograms IEEE Trans Syst Man Cybern 1979 9 1 62 66 10.1109/TSMC.1979.4310076 40. Pare S Kumar A Singh GK Bajaj V Image segmentation using multilevel thresholding: a research review Iran J Sci Technol Trans Electr Eng 2020 44 1 1 29 10.1007/s40998-019-00251-1 41. Pare S, Bhandar AK, Kumar A, Singh GK (2019) Rényi’s entropy and bat algorithm based color image multilevel thresholding. In: Machine intelligence and signal analysis. Advances in intelligent systems and computing, pp 71–84, Springer, Singapore 42. Pehrson LM Nielsen MB Ammitzbøl Lauridsen C Automatic pulmonary nodule detection applying deep learning or machine learning algorithms to the LIDC-IDRI database: a systematic review Diagnostics 2019 10.3390/diagnostics9010029 43. Pu J Zheng B Leader JK Wang X-H Gur D An automated CT based lung nodule detection scheme using geometric analysis of signed distance field Med Phys 2008 35 8 3453 3461 10.1118/1.2948349 18777905 44. Ryalat MH, Laycock S, Fisher M (2017) Automatic removal of mechanical fixations from ct imagery with particle swarm optimisation. Paper presented at the bioinformatics and biomedical engineering, Cham 45. Ryalat MH, Laycock S, Fisher M (2017) A fast and automatic approach for removing artefacts due to immobilisation masks in X-ray CT. Paper presented at the 2017 IEEE EMBS international conference on biomedical and health informatics (BHI). IEEE, New York 46. Saremi S Mirjalili S Lewis A Grasshopper optimisation algorithm: theory and application Adv Eng Softw 2017 105 30 47 10.1016/j.advengsoft.2017.01.004 47. Serhat Ozekes OO Ucan ON Nodule detection in a lung region that’s segmented with using genetic cellular neural networks and 3D template matching with fuzzy rule based thresholding Korean J Radiol 2007 9 1 1 9 10.3348/kjr.2008.9.1.1 48. Sharma A Kumar S Singh SN Brain tumor segmentation using DE embedded OTSU method and neural network Multidimens Syst Signal Process 2019 30 3 1263 1291 10.1007/s11045-018-0603-3 49. Tan M Deklerck R Jansen B Bister M Cornelis J A novel computer-aided lung nodule detection system for CT images Med Phys 2011 38 10 5630 5645 10.1118/1.3633941 21992380 50. Tarkhaneh O Shen H An adaptive differential evolution algorithm to optimal multi-level thresholding for MRI brain image segmentation Expert Syst Appl 2019 138 112820 10.1016/j.eswa.2019.07.037 51. Tedmori S Al-Najdawi N Lossless image cryptography algorithm based on discrete cosine transform Int Arab J Inf Technol 2012 9 5 471 478 52. Upadhyay P Chhabra JK Kapur’s entropy based optimal multilevel image segmentation using crow search algorithm Appl Soft Comput 2020 97 105522 10.1016/j.asoc.2019.105522 53. Zhang G Jiang S Yang Z Gong L Ma X Zhou Z BaoLiu CQ Automatic nodule detection for lung cancer in CT images: a review Comput Biol Med 2018 103 287 300 10.1016/j.compbiomed.2018.10.033 30415174 54. Zhou C, Tian L, Zhao H, Zhao K (2015) A method of two-dimensional otsu image threshold segmentation based on improved firefly algorithm. Paper presented at the 2015 IEEE international conference on cyber technology in automation, control, and intelligent systems (CYBER). IEEE, New York	36471798	PMC9714421	NO-CC CODE	2022-12-02 23:24:45	no	Neural Comput Appl. 2022 Dec 1;:1-19	utf-8	Neural Comput Appl	2,022	10.1007/s00521-022-08078-4	oa_other
==== Front 0410462 6011 Nature Nature Nature 0028-0836 1476-4687 31019317 10.1038/s41586-019-1119-1 nihpa1525169 Article Speech synthesis from neural decoding of spoken sentences Anumanchipalli Gopala K. 12* Chartier Josh 123* Chang Edward F. 123 1 Department of Neurological Surgery, University of California–San Francisco, San Francisco, California 94143, USA 2 Weill Institute for Neurosciences, University of California–San Francisco, San Francisco, California 94158, USA 3 University of California–Berkeley and University of California–San Francisco Joint Program in Bioengineering, Berkeley, California 94720, USA * Authors contributed equally Author Contributions Conception G.K.A., J.C., and E.F.C.; Articulatory kinematics inference G.K.A; Decoder design G.K.A and J.C.; Decoder analyses: J.C.; Data collection G.K.A., E.F.C., and J.C.; Prepared manuscript all; Project Supervision E.F.C. Correspondence and requests for materials should be addressed to [email protected] 27 10 2022 4 2019 24 4 2019 01 12 2022 568 7753 493498 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 Technology that translates neural activity into speech would be transformative for people unable to communicate as a result of neurological impairment. Decoding speech from neural activity is challenging because speaking requires such precise and rapid multi-dimensional control of vocal tract articulators. Here, we designed a neural decoder that explicitly leverages kinematic and sound representations encoded in human cortical activity to synthesize audible speech. Recurrent neural networks first decoded directly recorded cortical activity into articulatory movement representations, and then transformed those representations into speech acoustics. In closed vocabulary tests, listeners could readily identify and transcribe neurally synthesized speech. Intermediate articulatory dynamics enhanced performance even with limited data. Decoded articulatory representations were highly conserved across speakers, enabling a component of the decoder be transferrable across participants. Furthermore, the decoder could synthesize speech when a participant silently mimed sentences. These findings advance the clinical viability of speech neuroprosthetic technology to restore spoken communication. ==== Body pmcNeurological conditions that result in the loss of communication are devastating. Many patients rely on alternative communication devices that measure residual nonverbal movements of the head or eyes1, or now brain-computer interfaces (BCIs)2,3 to control a cursor to select letters one-by-one to spell out words. While these systems can enhance a patient’s quality of life, most users struggle to transmit more than 10 words/minute3, a rate far slower than the average of 150 words/min in natural speech. A major hurdle is how to overcome the constraints of current spelling-based approaches to enable far higher or even natural communication rates. A promising alternative is to directly synthesize speech from brain activity4,5. Spelling is a sequential concatenation of discrete letters, whereas speech is a highly efficient form of communication produced from a fluid stream of overlapping, multi-articulator vocal tract movements6. For this reason, a biomimetic approach that focuses on vocal tract movements and the sounds they produce may be the only means to achieve the high communication rates of natural speech, and also likely the most intuitive for users to learn7,8. In patients with paralysis, for example from ALS or brainstem stroke, high fidelity speech control signals may only be accessed by directly recording from intact cortical networks. Our goal was to demonstrate the feasibility of a neural speech prosthetic by translating brain signals into intelligible synthesized speech at the rate of a fluent speaker. To accomplish this, we recorded high-density electrocorticography (ECoG) signals from five participants undergoing intracranial monitoring for epilepsy treatment as they spoke several hundred sentences aloud. We designed a recurrent neural network that decoded cortical signals with an explicit intermediate representation of the articulatory dynamics to synthesize audible speech. Speech decoder design The two-stage decoder approach is shown in Figure 1a–d. Stage 1: a bidirectional long short-term memory (bLSTM) recurrent neural network9 decodes articulatory kinematic features from continuous neural activity (high-gamma amplitude envelope10 and low frequency component11,12, see methods) recorded from ventral sensorimotor cortex (vSMC)13, superior temporal gyrus (STG)14, and inferior frontal gyrus (IFG)15 (Figure 1a, b). Stage 2: a separate bLSTM decodes acoustic features (F0, mel-frequency cepstral coefficients (MFCCs), voicing and glottal excitation strengths) from the decoded articulatory features from Stage 1 (Figure 1c). The audio signal is then synthesized from the decoded acoustic features (Figure 1d). To integrate the two stages of the decoder, Stage 2 (articulation-to-acoustics) was trained directly on output of Stage 1 (brain-to-articulation) so that it not only learns the transformation from kinematics to sound, but can correct articulatory estimation errors made in Stage 1. A key component of our decoder is the intermediate articulatory representation between neural activity and acoustics (Figure 1b). This step is crucial because the vSMC exhibits robust neural activations during speech production that predominantly encode articulatory kinematics16,17. Because articulatory tracking of continuous speech was not feasible in our clinical setting, we used a statistical approach to estimate vocal tract kinematic trajectories (movements of the lips, tongue, and jaw) and other physiological features (e.g. manner of articulation) from audio recordings. These features initialized the bottleneck layer within a speech encoder-decoder that was trained to reconstruct a participant’s produced speech acoustics (see methods). The encoder was then used to infer the intermediate articulatory representation used to train the neural decoder. With this decoding strategy, it was possible to accurately reconstruct the speech spectrogram. Synthesis performance Overall, we observed detailed reconstructions of speech synthesized from neural activity alone (See Supplemental Video). Figure 1e,f, shows the audio spectrograms from two original spoken sentences plotted above those decoded from brain activity. The decoded spectrogram retained salient energy patterns present in the original spectrogram and correctly reconstructed the silence in between the sentences when the participant was not speaking. Extended Data Figure 1a,b, illustrates the quality of reconstruction at the phonetic level. Median spectrograms of original and synthesized phonemes showed that the typical spectrotemporal patterns were preserved in the decoded exemplars (e.g. formants F1-F3 in vowels /iː/ and /æ/; and key spectral patterns of mid-band energy and broadband burst for consonants /z/ and /p/, respectively). To understand to what degree the synthesized speech was perceptually intelligible to naïve listeners, we conducted two listening tasks that involved single-word identification and sentence-level transcription, respectively. The tasks were run on Amazon Mechanical Turk (see methods), using all 101 synthesized sentences from the test set for participant P1. For the single-word identification task, we evaluated 325 words that were spliced from the synthesized sentences. We quantified the effect of word length (number of syllables) and the number of choices (10, 25, and 50 words) on speech intelligibility, since these factors inform optimal design of speech interfaces18. Overall, we found listeners were more successful at word identification as syllable length increased, and number of word choices decreased (Figure 2a), consistent with natural speech perception19. For sentence-level intelligibility, we designed a closed vocabulary, free transcription task. Listeners heard the entire synthesized sentence and transcribed what they heard by selecting words from a defined pool (of either 25 or 50 words) that included the target words and random words from the test set. The closed vocabulary setting was necessary because the test set was a subset of sentences from MOCHA-TIMIT20 which was primarily designed to optimize articulatory coverage of English but contains highly unpredictable sentence constructions and low frequency words. Listeners were able to transcribe synthesized speech well. Of the 101 synthesized trials, at least one listener was able to provide a perfect transcription for 82 sentences with a 25-word pool and 60 sentences with a 50-word pool. Of all submitted responses, listeners transcribed 43% and 21% of the total trials perfectly, respectively (Extended Data Figure 2). In Figure 2b, the distributions of mean word error rates (WER) of each sentence are shown. Transcribed sentences had a median 31% WER with a 25-word pool size and 53% WER with a 50-word pool size. Table 1 shows listener transcriptions for a range of WERs. Median level transcriptions still provided a fairly accurate, and in some cases legitimate transcription (eg., “mum” transcribed as “mom” etc.). The errors suggest that the acoustic phonetic properties of the phonemes are still present in the synthesized speech, albeit to the lesser degree (eg., “rabbits” transcribed as “rodents”). This level of intelligibility for neurally synthesized speech would already be immediately meaningful and practical for real world application. We then quantified the decoding performance at a feature level for all participants. In speech synthesis, the spectral distortion of synthesized speech from ground-truth is commonly reported using the mean Mel-Cepstral Distortion (MCD) 21. Mel-Frequency bands emphasize the distortion of perceptually relevant frequency bands of the audio spectrogram22. In Figure 2c, the MCD of neurally synthesized speech was compared to a reference synthesis from articulatory kinematics and chance-level decoding (lower MCD is better). The reference synthesis simulates perfect neural decoding of the kinematics. For our five participants (P1–5), the median MCD scores of decoding speech ranged from 5.14 dB to 6.58 dB (p<1e-18, Wilcoxon signed-rank test (WSRT), for each participant). We also computed the correlations between original and decoded acoustic features. For each sentence and feature, the Pearson’s correlation coefficient was computed using every sample (at 200 Hz) for that feature. The sentence correlation of the mean decoded acoustic features (intensity + MFCCs + excitation strengths + voicing) and inferred kinematics across participants are plotted in Figure 2d. Prosodic features such as pitch (F0), speech envelope, and voicing were decoded well above chance-level (r > 0.6, except F0 for P2: r= 0.49 and all features for P5, p<1e-10, WSRT, for all participants and features in Figure 2d). Correlation decoding performance for all other features is shown in Extended Data Figure 4a,b. Decoder characteristics The following analyses were performed on data from P1. In designing a neural decoder for clinical applications, there are several key considerations that determine model performance. First, in patients with severe paralysis or limited speech ability, training data may be very difficult to obtain. Therefore, we assessed the amount of data necessary to achieve a high level of performance. We found a clear advantage in explicitly modeling articulatory kinematics as an intermediate step over decoding acoustics directly from the ECoG signals. The “direct” decoder was a bLSTM recurrent neural network optimized for decoding acoustics (MFCCs) directly from same ECoG signals as employed in articulatory decoder. We found robust performance could be achieved with as little as 25 minutes of speech, but performance continued to improve with the addition of data (Figure 2e). Without the articulatory intermediate step, the direct ECoG to acoustic decoding MCD was offset by 0.54 dB (0.2 dB is perceptually noticeable21) using the full data set (Figure 3a) (p=1e-17, n=101, WSRT). This performance gap between the two approaches persisted with increasing data sizes. One interpretation is that aspects of kinematics are more preferentially represented by cortical activity than acoustics16, and thereby learned more quickly with limited data. Another aspect that may underlie this difference is that articulatory kinematics lie on a low-dimensional manifold that constrain the potential high-dimensionality of acoustic signals (Extended Data Figure 5)6,7,23. Hence, separating out the high-dimensional translation of articulation to speech, as done Stage 2 of our decoder may be critical for performance. It is possible that with sufficiently large data both decoding approaches would converge with one another. Second, we wanted to understand the phonetic properties that were preserved in synthesized speech. We used Kullback-Leibler (KL) divergence to compare the distribution of spectral features of each decoded phoneme to those of each ground-truth phoneme to determine how similar they were (Extended Data Figure 6). We expected that, in addition to the same decoded and ground-truth phoneme being similar to one another, phonemes with shared acoustic properties would also be characterized as similar to one another. Hierarchical clustering on the KL-divergence of each phoneme pair demonstrated that phonemes were clustered into four main groups. Group 1 contained consonants with an alveolar place of constriction. Group 2 contained almost all other consonants. Group 3 contained mostly high vowels. Group 4 contained mostly mid and low vowels. The difference between groups tended to correspond to variations along acoustically significant dimensions (frequency range of spectral energy for consonants, and formants for vowels). Indeed, these groupings explain some of the confusions reflected in listener transcriptions of these stimuli. This hierarchical clustering was also consistent with the acoustic similarity matrix of only ground-truth phoneme-pairs (Extended Data Figure 7) (cophenetic correlation24 = 0.71, p=1e10). Third, since the success of the decoder depends on the initial electrode placement, we quantified the contribution of several anatomical regions (vSMC, STG, and IFG) that are involved in continuous speech production25. Decoders were trained in a leave-one-region-out fashion where all electrodes from a particular region were held out (Figure 2f). Removing any region led to some decreased decoder performance (Figure 2g) (p=3e-4, n=101, WSRT). However, excluding vSMC resulted in the largest decrease in performance (1.13 dB MCD increase). Fourth, we investigated whether the decoder generalized to novel sentences that were never seen in the training data. Since P1 produced some sentences multiple times, we compared two decoders: one that was trained on all sentences (not the particular instances in the test set), and one that was trained excluding every instance of the sentences in the testing set. We found no significant difference in decoding performance of the sentences for both MCD and correlations of spectral features (p=0.36, p=0.75, n=51, WSRT, Extended Data Figure 8). Importantly, this suggests that the decoder can generalize to arbitrary words and sentences that the decoder was never trained on. Synthesizing mimed speech To rule out the possibility that the decoder is relying on the auditory feedback of participants’ vocalization, and to simulate a setting where subjects do not overtly vocalize, we tested our decoder on silently mimed speech. We tested a held-out set of 58 sentences in which the participant (P1) audibly produced each sentence and then mimed the same sentence, making the same articulatory movements but without making sound. Even though the decoder was not trained on mimed sentences, the spectrograms of synthesized silent speech demonstrated similar spectral patterns to synthesized audible speech of the same sentence (Figure 3a–c). With no original audio to compare, we quantified performance of the synthesized mimed sentences with the audio from the trials with spoken sentences. We calculated the spectral distortion and correlation of the spectral features by first dynamically time-warping the spectrogram of the synthesized mimed speech to match the temporal profile of the audible sentence (Figure 3d,e) and then comparing performance. While synthesis performance on mimed speech was inferior to that of audible speech (likely due to absence of phonation signals during mime), this demonstrates that it is possible to decode important spectral features of speech that were never audibly uttered (p < 1e-11, compared to chance, n = 58; Wilcoxon signed-rank test) and that the decoder did not rely on auditory feedback. State-space of decoded speech articulation Our findings suggest that modeling the underlying kinematics enhances the decoding performance, so we next wanted to better understand the nature of the decoded kinematics from population neural activity. We examined low-dimensional kinematic state-space trajectories, by computing the state-space projection via principal components analysis (PCA) on the articulatory kinematic features. The first ten principal components (PCs) (of 33 total) captured 85% of the variance and the first two PCs captured 35% (Extended Data Figure 5). In Figures 4a,b, the kinematic trajectory of an example sentence is projected onto the first two PCs. These trajectories were well decoded, as seen in the example (r=0.91, r=0.91, Figure 4a,b), and summarized across all test sentences and participants (median r>0.72 for all participants except P5, r represents mean r of first 2 PCs, Figure 4e). Furthermore, state-space trajectories of mimed speech were well decoded (median r=0.6, p=1e-5, n=38, WSRT, Figure 4e). The state-space trajectories appeared to manifest the dynamics of syllabic patterns in continuous speech. The time courses of consonants (grey) and vowels (blue) were plotted on the state-space trajectories and tended to correspond with the troughs and peaks of the trajectories, respectively (Figures 4a,b). In Figures 4c,d, we sampled from every vowel-to-consonant transition (n=22453) and consonant-to-vowel transition (n=22453), and plotted 500 ms traces of the average trajectories for PC1 and PC2 centered at the time of transition. Both types of trajectories were biphasic in nature, transitioning from the “high” state during the vowel to the “low” state during the consonant (white), and vice versa (black). When examining transitions of specific phonemes, we found that PC1 and PC2 retained their biphasic trajectories of vowel/consonant states, but showed specificity toward particular phonemes indicating that PC1 and PC2 are not necessarily just describing jaw opening and closing, but rather global opening and closing configurations of the vocal tract (Extended Data Figure 9). These findings are consistent with theoretical accounts of human speaking behavior, which postulate that high-dimensional speech acoustics lie on a low-dimensional articulatory state-space6. To evaluate the similarity of the decoded state-space trajectories, we correlated productions of the same sentence across participants that were projected into their respective kinematic state-spaces (only P1, P2, and P4 had comparable sentences). The state-space trajectories were highly similar (r>0.8, Figure 4f), suggesting that the decoder is likely relying upon a shared representation across speakers, a critical basis for generalization. A shared kinematic representation across speakers could be very advantageous for someone who cannot speak as it may be more intuitive and faster to first learn to use the kinematics decoder (Stage 1), while using an existing kinematics-to-acoustics decoder (stage 2) trained on speech data collected independently. In Figure 4g, we show synthesis performance from transferring Stage 2 from a source participant (P1) to a target participant (P2). The acoustic transfer performed well, although less than when both stage 1 and stage 2 were trained on the target (P2), likely because the MCD metric is sensitive to speaker identity. Discussion In this paper, we demonstrate speech synthesis using high-density, direct cortical recordings from human speech cortex. Previous strategies for neural decoding of speech production focused on reconstructing spectrotemporal auditory representations26 or direct classification of speech segments like phonemes or words27,28,29 but were limited in their ability to scale to larger vocabulary sizes and communication rates. Meanwhile, decoding of auditory cortex responses has been more successful for speech sounds30,31 in part because of the direct relationship between the auditory encoding of spectrotemporal information and the reconstructed spectrogram. An outstanding question has been whether decoding vocal tract movements from the speech motor cortex could be used for generating high-fidelity acoustic speech output. Previous work focused on understanding movement encoding at single electrodes16, however, the fundamentally different challenge for speech synthesis is decoding the population activity that addresses the complex mapping between vocal tract movements and sounds. Natural speech production involves over 100 muscles and the mapping from movement to sounds is not one-to-one. Our decoder explicitly incorporated this knowledge to simplify the translation of neural activity to sound by first decoding the primary physiological correlate of neural activity and then transforming to speech acoustics. This statistical mapping permits generalization with limited amounts of training. Direct speech synthesis has several major advantages over spelling-based approaches. In addition to the capability to communicate at a natural speaking rate, it captures prosodic elements of speech that are not available with text output, for example pitch intonation32. Furthermore, a practical limitation for current alternative communication devices is the cognitive effort required to learn and use them. For patients in whom the cortical processing of articulation is still intact, a speech-based BCI decoder may be far more intuitive and easier to learn to use7,8. BCIs are rapidly becoming a clinically viable means to restore lost function. Neural prosthetic control was first demonstrated in participants without disabilities33,34,35 before translating the technology to participants with tetraplegia36,37,38,39. Our findings represent one step forward for addressing a major challenge posed by paralyzed patients who cannot speak. The generalization results here demonstrate that speakers share a similar kinematic state-space representation (speaker-independent), and it is possible to transfer model knowledge about the mapping of kinematics to sound across subjects. Tapping into this emergent, low-dimensional representation from coordinated population neural activity in the intact cortex may be a critical for bootstrapping a decoder23, as well facilitating BCI learning7. Our results may be an important next step in realizing speech restoration for patients with paralysis. Methods Participants and experimental task. Five human participants (30 F, 31 F, 34 M, 49 F, 29 F) underwent chronic implantation of high-density, subdural electrode array over the lateral surface of the brain as part of their clinical treatment of epilepsy (right, left, left, left, left) hemisphere grids, respectively, Extended Data Figure 3). Participants gave their written informed consent before the day of the surgery. All participants were fluent in English. All protocols were approved by the Committee on Human Research at UCSF and experiments/data in this study complied with all relevant ethical regulations. Each participant read and/or freely spoke a variety of sentences. P1 read aloud two complete sets of 460 sentences from the MOCHA-TIMIT20 database. Additionally, P1 also read aloud passages from the following stories: Sleeping Beauty, Frog Prince, Hare and the Tortoise, The Princess and the Pea, and Alice in Wonderland. P2 read aloud one full set of 460 sentences from the MOCHA-TIMIT database and further read a subset of 50 sentences an additional 9 times each. P3 read 596 sentences describing three picture scenes and then freely described the scene resulting in another 254 sentences. P3 also spoke 743 sentences during free response interviews. P4 read two complete sets of MOCHA-TIMIT sentences, 465 sentences drawn of scene descriptions and 399 sentences during free response interviews. P5 read one set of MOCHA-TIMIT sentences and 360 sentences of scene descriptions. In addition to audible speech, P1 also read 10 sentences 12 times each alternating between audible and silently mimed (i.e. making the necessary mouth movements) speech. Microphone recordings were obtained synchronously with the ECoG recordings. Data acquisition and signal processing. Electrocorticography was recorded with a multi-channel amplifier optically connected to a digital signal processor (Tucker-Davis Technologies). Speech was amplified digitally and recorded with a microphone simultaneously with the cortical recordings. The grid placements were decided upon purely by clinical considerations. ECoG signals were recorded at a sampling rate of 3,052 Hz. Each channel was visually and quantitatively inspected for artifacts or excessive noise (typically 60 Hz line noise). The analytic amplitude of the high-gamma frequency component of the local field potentials (70 – 200 Hz) was extracted with the Hilbert transform and down-sampled to 200 Hz. The low frequency component (1–30 Hz) was also extracted with a 5th order Butterworth bandpass filter, down-sampled to 200 Hz and parallelly aligned with the high-gamma amplitude. Finally, the signals were z-scored relative to a 30 second window of running mean and standard deviation, so as to normalize the data across different recording sessions. We studied high-gamma amplitude because it has been shown to correlate well with multi-unit firing rates and has the temporal resolution to resolve fine articulatory movements10. We also included a low frequency signal component due to the decoding performance improvements note for reconstructing perceived speech from auditory cortex11,12. Decoding models were constructed using all electrodes from vSMC, STG, and IFG except for electrodes with bad signal quality as determined by visual inspection. We removed 8 electrodes for P1, 7 electrodes for P2, and 16 electrodes for P3. No electrodes were removed for P4 or P5. The decoder uses both high-gamma amplitude and raw low-frequency signals together as input to the model. For instance, n electrodes will result as n * 2 input features. Phonetic and phonological transcription. For the collected speech acoustic recordings, transcriptions were corrected manually at the word level so that the transcript reflected the vocalization that the participant actually produced. Given sentence level transcriptions and acoustic utterances chunked at the sentence level, hidden Markov model based acoustic models were built for each participant so as to perform sub-phonetic alignment40 within the Festvox41 framework. Phonological context features were also generated from the phonetic labels, given their phonetic, syllabic and word contexts. Cortical surface extraction and electrode visualization. We localized electrodes on each individual’s brain by co-registering the preoperative T1 MRI with a postoperative CT scan containing the electrode locations, using a normalized mutual information routine in SPM12. Pial surface reconstructions were created using Freesurfer. Final anatomical labeling and plotting was performed using the img_pipe python package42. Inference of articulatory kinematics. Among the most accurate methods to record vocal tract kinematics is called Electromagnetic Midsagittal Articulography (EMA). The process involves gluing small sensors to the articulators, generally 3 sensors on the tongue, 1 on each lip, 1 on each incisor. A magnetic field is projected at the participant’s head and as the participant speaks, each sensor can be precisely tracked as it moves through the magnetic field. Each sensor has a wire leading out of the participant’s mouth and connected to a receiver to record measurements. Because of the above requirements, we did not pursue using EMA in the setting of our ECoG recordings because potential disruption of medical instruments by the magnetic field, long setup time conflicted with limited recording session time with patients, the setup procedure was too uncomfortable. Instead, we developed a model to infer articulatory kinematics from audio recordings. The articulatory data used to build the articulatory inference models was from MOCHA-TIMIT20 and MNGU0 corpora43. The articulatory kinematics inference model comprises a stacked deep encoder-decoder, where the encoder combines phonological (linguistic and contextual features, resulting from the phonetic segmentation process) and acoustic representations (25 dimensional MFCC vectors sampled at 200 Hz) into a latent articulatory representation (also sampled at 200 Hz) that is then decoded to reconstruct the original acoustic signal. The latent representation is initialized with inferred articulatory movement and appropriate manner features. We performed statistical subject-independent acoustic-to-articulatory inversion16 to estimate 12 dimensional articulatory kinematic trajectories (x and y displacements of tongue dorsum, tongue blade, tongue tip, jaw, upper lip and lower lip, as would be measured by EMA) using only the produced acoustics and phonetic transcriptions. Since EMA features do not describe all acoustically consequential movements of the vocal tract, we append complementary speech features that improve reconstruction of original speech. First, to approximate laryngeal function, we add pitch, voicing (binary value indicating if a frame is voiced or not), and speech envelope, i.e., the frame level intensity computed as the sum total power within all the Mel scale frequencies within a 25 millisecond analysis window, computed at a shift of 5 milliseconds. Next, we added place-manner tuples (represented as continuous [0–1] valued features) to bootstrap the EMA with what we determined were missing physiological aspects in EMA. There were 18 additional values to capture the following place-manner feature tuples (palatal approximant, labial stop etc., see Supplemental Information (a) for the complete list). We used an existing annotated speech database (Wall Street Journal Corpus44) and trained speaker independent deep recurrent network regression models to predict continuous valued place-manner vectors only from the acoustics features, the phonetic labels were used to determine the ground truth values for these labels (e.g., the dimension “labial stop” would be 1 for all frames of speech that belong to the phonemes /p/, /b/ and so forth). However, with a regression output layer, predicted values were not constrained to the binary nature of the input features. The network architecture was 3 feedforward layers followed by one bLSTM layer to predict each time point of these manner descriptors from a 100 millisecond window of acoustic features. Combined with the EMA trajectories, these 33 feature vectors form the initial articulatory feature estimates. To ensure that the articulatory representation has the potential to reliably reconstruct speech for the target subject, we designed a stacked encoder-decoder network to optimize these initial estimates for these values. Specifically, a recurrent neural network encoder is trained to convert phonological and acoustic features to the articulatory representation and then a decoder that converts the articulatory representation back to the acoustic features (original MFCC). The encoder is implemented as 2 feedforward layers followed by 2 bLSTM layers. The decoder is implemented as 3 feedforward layers. Software implementation was done using Keras Functional API within Tensorflow45. The stacked network is re-trained optimizing the joint mean squared error loss on acoustic and EMA parameters using the ADAM optimizer, with an initial learning rate set at 0.001. For regularization 40% dropout was allowed in all feedforward layers. After convergence, the trained encoder is used to estimate the final articulatory kinematic features that act as the articulatory intermediate to decode acoustic features from ECoG. Neural decoder. The decoder maps ECoG recordings to MFCCs via a two stage process by learning intermediate mappings between ECoG recordings and articulatory kinematic features, and between articulatory kinematic features and acoustic features. All data (ECoG, kinematics, and acoustics) are sampled and processed by the model at 200 Hz. We implemented this model using TensorFlow in python. In the first stage, a stacked 3-layer bLSTM9 learns the mapping between 300 ms (60 time points) sequences of high-gamma and LFP signals and a corresponding single time point (sampled at 200 Hz) of the 33 articulatory features. In the second stage, an additional stacked 3-layer bLSTM learns the mapping between the output of the first stage (decoded articulatory features) and 32 acoustic parameters (200 Hz) for full sentences sequences. These parameters are 25 dimensional MFCCs, 5 sub-band voicing strengths for glottal excitation modelling, log(F0), voicing. During testing, a full sentence sequence of neural activity (high-gamma and low-frequency components) is processed by the decoder. The first stage processes 300 ms of data at a time, sliding over the sequence sample by sample, until it has returned a sequence of kinematics that is equal length to the neural data. The neural data is padded with an additional 150 ms of data before and after the sequence to ensure the result is the correct length. The second stage processes the entire sequence at once, returning an equal length sequence of acoustic features. These features are then synthesized into an audio signal. At each stage, the model is trained using the Adam optimizer to minimize mean-squared error. The optimizer was initialized with learning rate=0.001,beta1=0.9, beta2=0.999, epsilon=1e-8. Models were stopped from training after the validation loss no longer decreased. Dropout rate is set to 50% in stage 1 and 25% in stage 2 to suppress overfitting tendencies of the models. There are 100 hidden units for each LSTM cell. Each model employed 3 stacked bLSTMs with an additional linear layer for regression. We use a bLSTM because of their ability to retain temporally distant dependencies when decoding a sequence46. In the first stage, the batch size for training is 256, and in the second stage the batch size is 25. Training and testing data were randomly split based off of recording sessions, meaning that the test set was collected during separate recording sessions from the training set. The training and testing splits in terms of total speaking time (minutes:seconds) are as follows: P1 – training: 92:15, testing: 4:46 (n=101); P2 – training: 36:57, testing: 3:50 (n=100); P3 – training: 107:42, testing: 4:44 (n=98); P4 – training: 27:39, testing 3:12 (n=82).; P5 – training 44:31, testing 2:51 (n=44). n=number of sentences in test set. For shuffling the data to test for significance, we shuffled the order of the electrodes that were fed into the decoder. This method of shuffling preserved the temporal structure of the neural activity. The “direct” ECoG to acoustics decoder described in Figure 2e a similar architecture as the stage 1 articulatory bLSTM except with an MFCC output. Originally we trained the direct acoustic decoder as a 6-layer bLSTM that mimics the architecture of the 2 stage decoder with MFCCs as the “intermediate layer” and as the output. However, we found performance was better with a 4-layer bLSTM (no intermediate layer) with 100 hidden units for each layer, 50% dropout and 0.005 learning rate using Adam optimizer for minimizing mean-squared error. Models were coded using Python’s version 1.9 of Tensorflow. Speech synthesis from acoustic features. We used an implementation of the Mel-log spectral approximation algorithm with mixed excitation47 within Festvox to generate the speech waveforms from estimates of the acoustic features from the neural decoder. Mel-Cepstral Distortion (MCD). To examine the quality of synthesized speech, we calculated the Mel-Cepstral Distortion (MCD) of the synthesized speech when compared the original ground-truth audio. MCD is an objective measure of error determined from MFCCs and is correlated to subjective perceptual judgments of acoustic quality21. For reference acoustic features mc(y) and decoded features mc(y^), MCD=10ln(10)∑0<d<25(mcd(y)−mcd(y^))2 Intelligibility Assessment. Listening tests using crowdsourcing are a standard way of evaluating the perceptual quality of synthetic speech48. To comprehensively assess the intelligibility of the neurally synthesized speech, we conducted a series of identification and transcription tasks on the Amazon Mechanical Turk. The unseen test set from P1 (101 trials of 101 unique sentences, shown in Supplemental Information (b)) was used as the stimuli for listener judgments. For the word level identification tasks, we created several cohorts of words grouped by the number of syllables within. Using the time boundaries from the ground truth phonetic labelling, we extracted audio from the neurally synthesized speech into four classes of 1-syllable, 2-syllable, 3-syllable and 4-syllable words. We conducted tests on each of these groups of words that involve identification of the synthesized audio from a group of i) 10 choices, ii) 25 choices, and iii) 50 choices of what they think the word is. The presented options included the true word and the remaining choices randomly drawn from the other words within the class (see Supplemental Information (c) for class sizes across these conditions). All words within the word groups were judged for intelligibility without any further sub-selection. Since the content words in the MOCHA-TIMIT data are largely low frequency words to assess sentence-level intelligibility, along with the neurally synthesized audio file, we presented the listeners a pool of words that may be in the sentence. This makes it task a limited vocabulary free response transcription. We conducted two experiments where the transcriber is presented with pool of i) 25 word choices, and ii) 50 word choices that may be used the sentence (a sample interface is shown in Supplemental Information (d)). The true words that make up the sentence are included along with randomly drawn words from the entire test set and displayed in alphabetical order. Given that the median sentence is only 7 words long (std=21., min=4, max=13), this task design allows for reliable assessment of intelligibility. Each trial was judged by 10–20 different listeners. Each intelligibility task was performed by 47–187 unique listeners (a total of 1755 listeners across 16 intelligibility tasks, see supplemental information (e) for breakdown per task) making all reported analyses statistically reliable. All sentences from the test set were sent for intelligibility assessment without any further selection. The listeners were required to be English speakers located in the United States, with good ratings(>98% rating from prior tasks on the platform). For the sentence transcription tasks, an automatic spell checker was employed to correct misspellings. No further spam detection, or response rejection was done in all analyses reported. Word Error Rate (WER) metric computed on listener transcriptions is used to judge the intelligibility of the neurally synthesized speech. Where I is the number of word insertions, D is the number of word deletions and S is the number of word substitutions for a reference sentence with N words, WER is computed as WER=I+D+SN Data limitation analysis. To assess the amount of training data affects decoder performance, we partitioned the data by recording blocks and trained a separate model for an allotted number of blocks. In total, 8 models were trained, each with one of the following block allotments: [1, 2, 5, 10, 15, 20, 25, 28]. Each block comprised an average of 50 sentences recorded in one continuous session. Quantification of silent speech synthesis. By definition, there was no acoustic signal to compare the decoded silent speech. In order to assess decoding performance, we evaluated decoded silent speech in regards to the audible speech of the same sentence uttered immediately prior to the silent trial. We did so by dynamically time-warping49 the decoded silent speech MFCCs to the MFCCs of the audible condition and computing Pearson’s correlation coefficient and Mel-cepstral distortion. Phoneme acoustic similarity analysis. We compared the acoustic properties of decoded phonemes to ground-truth to better understand the performance of our decoder. To do this, we sliced all time points for which a given phoneme was being uttered and used the corresponding time slices to estimate its distribution of spectral properties. With principal components analysis (PCA), the 32 spectral features were projected onto the first 4 principal components before fitting the gaussian kernel density estimate (KDE) model. This process was repeated so that each phoneme had two KDEs representing either its decoded and or ground-truth spectral properties. Using Kullback-Leibler divergence (KL divergence), we compared each decoded phoneme KDE to every ground-truth phoneme KDE, creating an analog to a confusion matrix used in discrete classification decoders. KL divergence provides a metric of how similar two distributions are to one another by calculating how much information is lost when we approximate one distribution with another. Lastly, we used Ward’s method for agglomerative hierarchical clustering to organize the phoneme similarity matrix. To understand whether the clustering of the decoded phonemes was similar to the clustering of ground-truth phoneme pairs (Extended Data Figure 7), we used the cophenetic correlation (CC) to assess how well the hierarchical clustering determined from decoded phonemes preserved the pairwise distance between original phonemes, and vice versa24. For the decoded phoneme dendrogram, the CC for preserving original phoneme distances was 0.71 as compared to 0.80 for preserving decoded phoneme distances. For the original phoneme dendrogram, the CC for preserving decoded phoneme distances was 0.64 as compared to 0.71 for preserving original phoneme distances. p<1e-10 for all correlations. State-space kinematic trajectories. For state-space analysis of kinematic trajectories, principal components analysis (PCA) was performed on the 33 kinematic features using the training data set from P1. Figure 4a,b shows kinematic trajectories (original, decoded (audible and mimed) projected onto the first two principal components (PCs). The example decoded mimed trajectory occurred faster in time by a factor of 1.15 than the audible trajectory so we uniformly temporally stretched the trajectory for visualization. The peaks and troughs of the decoded mimed trajectories were similar to the audible speech trajectory (r=0.65, r=0.55) although the temporal locations are shifted relative to one another, likely because the temporal evolution of a production, whether audible or mimed, is inconsistent across repeated productions. To quantify the decoding performance of mimed trajectories, we used the dynamic time-warping approach described above, although in this case, temporally warping with respect to the inferred kinematics (not the state-space) (Figure 4e). For analysis of state-space trajectories across participants (Figure 4f), we measured the correlations of productions of the same sentence, but across participants. Since the sentences were produced at different speeds, we dynamically time-warped them to match and compared against correlations of dynamically time-warped mismatched sentences. Code Availability. All code may be freely obtained for non-commercial use by contacting the corresponding authors. Extended Data Extended Data Figure 1: a,b Median spectrograms, time-locked to the acoustic onset of phonemes from original (a) and decoded (b) audio (n: /i/ = 112, /z/ = 115, /p/ 69, /ae/ = 86). These phonemes represent the diversity of spectral features. Original and decoded median phoneme spectrograms were well correlated (Pearson’s r > 0.9 for all phonemes, p=1e-18) Extended Data Figure 2: Transcription word error rate for individual trials. Word error rates (WER) for individually transcribed trials for 25 (a) and 50 (b) word pool size. Listeners transcribed synthesized sentences by selecting words from a defined pool of words. Word pools included correct words in synthesized sentence and random words from the test set. One trial is one listener transcription of one synthesized sentence. Extended Data Figure 3: Electrode array locations for participants. MRI reconstructions of participants’ brains with overlay of electrocorticographic electrode (ECoG) array locations. Extended Data Figure 4: Decoding performance of kinematic and spectral features. Data from P1. a, Correlations of all 33 decoded articulatory kinematic features with ground-truth (n=101 sentences). EMA features represent X and Y coordinate traces of articulators (lips, jaw, and three points of the tongue) along the midsagittal plane of the vocal tract. Manner features represent complementary kinematic features to EMA that further describe acoustically consequential movements. b, Correlations of all 32 decoded spectral features with ground-truth (n=101 sentences). MFCC features are 25 mel-frequency cepstral coefficients that describe power in perceptually relevant frequency bands. Synthesis features describe glottal excitation weights necessary for speech synthesis. Box plots as described in Figure 2. Extended Data Figure 5: Comparison of cumulative variance explained in kinematic and acoustic state-spaces. For each representation of speech—kinematics and acoustics—principal components analysis (PCA) was computed and variance explained for each additional principal component was cumulatively summed. Kinematic and acoustic representations had 33 and 32 features, respectively. Extended Data Figure 6: Decoded phoneme acoustic similarity matrix. Acoustic similarity matrix compares acoustic properties of decoded phonemes and originally spoken phonemes. Similarity is computed by first estimating a gaussian kernel density for each phoneme (both decoded and original) and then computing the Kullback-Leibler (KL) divergence between a pair of decoded and original phoneme distributions. Each row compares the acoustic properties of a decoded phoneme with originally spoken phonemes (columns). Hierarchical clustering was performed on the resulting similarity matrix. Data from P1. Extended Data Figure 7: Ground-truth acoustic similarity matrix. Compares acoustic properties of ground-truth spoken phonemes with one another. Similarity is computed by first estimating a gaussian kernel density for each phoneme and then computing the Kullback-Leibler (KL) divergence between a pair of a phoneme distributions. Each row compares the acoustic properties of a two ground-truth spoken phonemes. Hierarchical clustering was performed on the resulting similarity matrix. Data from P1. Extended Data Figure 8: Comparison between decoding novel and repeated sentences. Comparison metrics were spectral distortion (a) and correlation between decoded and original spectral features (b). Decoder performance for these two types of sentences was compared to find no difference (p=0.36, p=0.75, n=51 sentences, Wilcoxon signed-rank test). A novel sentence consists of words and/or a word sequence not present in the training data. A repeated sentence is a sentence that has at least one matching word sequence in the training data, although unique production. Comparison was performed on P1 and sentences evaluated were the same across both cases with two decoders trained on differing datasets to either exclude or include unique repeats of sentences in the test set. ns indicates p>0.05. Box plots as described in Figure 2. Extended Data Figure 9: Kinematic state-space trajectories for phoneme-specific vowel-consonant transitions. Average trajectories of PC1 and PC2 for transitions from a either a consonant or vowel to a specific phonemes. Trajectories are 500 ms and centered at transition between phonemes. a, Consonant -> corner vowels (n=1387, 1964, 2259, 894, respectively). PC1 shows separation of all corner vowels and PC2 delineates between front vowels (iy, ae) and back vowels (uw, aa). b, vowel -> unvoiced plosives (n=2071, 4107, 1441, respectively). PC1 was more selective for velar constriction (k) and PC2 for bilabial constriction (p). c Vowel -> alveolars (n=3919, 3010, 4107, respectively). PC1 shows separation by manner of articulation (nasal, plosive, fricative) while PC2 is less discriminative. d, PC1 and PC2 show little, if at all, delineation between voiced and unvoiced alveolar fricatives (n=3010, 1855, respectively). Supplementary Material 1525169_Sup_Info 1525169_RS 1525169_Sup_Video_1_legend 1525169_Sup_Video_1 Acknowledgments We thank Matthew Leonard, Neal Fox, David Moses for their helpful comments on the manuscript. We also thank Ben Speidel for his work reconstructing MRI images of patients’ brains. This work was supported by grants from the NIH (DP2 OD008627 and U01 NS098971-01). E.F.C is a New York Stem Cell Foundation- Robertson Investigator. This research was also supported by The New York Stem Cell Foundation, the Howard Hughes Medical Institute, The McKnight Foundation, The Shurl and Kay Curci Foundation, and The William K. Bowes Foundation. The data that support the findings of this study are available from the corresponding author upon request. All code may be freely obtained for non-commercial use by contacting the corresponding author. Figure 1: Speech synthesis from neurally decoded spoken sentences. a, The neural decoding process begins by extracting relevant signal features from high-density cortical activity. b, A bi-directional long short-term memory (bLSTM) neural network decodes kinematic representations of articulation from ECoG signals. c, An additional bLSTM decodes acoustics from the previously decoded kinematics. Acoustics are spectral features (e.g. Mel-frequency cepstral coefficients (MFCCs)) extracted from the speech waveform. d, Decoded signals are synthesized into an acoustic waveform. e, Spectrogram shows the frequency content of two sentences spoken by a participant. f, Spectrogram of synthesized speech from brain signals recorded simultaneously with the speech in e(repeated 5 times with similar results). Mel-cepstral distortion (MCD) was computed for each sentence between the original and decoded audio. 5-fold cross-validation used to find consistent decoding. Figure 2: Synthesized speech intelligibility and feature-specific performance. a, Listening tests for identification of excerpted single words (n=325) and full sentences (n=101) for synthesized speech from participant P1. Points represent mean word identification rate. Words were grouped by syllable length (n=75, 158, 68, 24). Listeners identified speech by selecting from a set of choices (10, 25, 50). b, Listening tests for closed vocabulary transcription of synthesized sentences (n=101). Responses were constrained in word choice (25, 50), but not in sequence length. Outlines are kernel density estimates of the distributions. c, Spectral distortion, measured by Mel-Cepstral Distortion (MCD) (lower values are better), between original spoken sentences and neurally decoded sentences (n=101, 100, 93, 81, 44, respectively). Reference MCD refers to the synthesis of original (inferred) kinematics without neural decoding. d, Correlation of original and decoded kinematic and acoustic features (n=101, 100, 93, 81, 44 sentences, respectively). Kinematic and acoustic values represent mean correlation of 33 and 32 features, respectively. e, Mean MCD of sentences (n=101) decoded from models trained on varying amounts of training data. The neural decoder with an articulatory intermediate stage (purple) performed better than direct ECoG to acoustics decoder (grey) (all data sizes: p < 1e-5, n = 101 sentences; WSRT). f, Anatomical reconstruction of a single participant’s brain (P1) with the following regions used for neural decoding: ventral sensorimotor cortex (vSMC), superior temporal gyrus (STG), and inferior frontal gyrus (IFG). g, Difference in median MCD of sentences (n=101) between decoder trained on all regions and decoders trained on all-but-one region. Exclusion of any region resulted in decreased performance (p < 3e-4, n = 101 sentences; WSRT). All box plots depict median (horizontal line inside box), 25th and 75th percentiles (box), 25/75th percentiles ±1.5× interquartile range (whiskers), and outliers (circles). Distributions were compared with each as other as indicated or with chance-level distributions using two-tailed Wilcoxon signed-rank tests (WSRT). * indicates p<0.001. All error bars are SEM. Figure 3: Speech synthesis from neural decoding of silently mimed speech. a-c, Spectrograms of original spoken sentence (a), neural decoding from audible production (b), and neural decoding from silently mimed production (c) (repeated 5 times with similar results). d, e, Median spectral distortion (MCD) (d) and correlation of original and decoded spectral features (e) for audibly and silently produced speech (n=58 sentences). Decoded sentences were significantly better than chance-level decoding for both speaking conditions (audible: p=3e-11, mimed: p=5e-11, n = 58; Wilcoxon signed-rank test). Box plots as described in Figure 2. * indicates p<0.001. Figure 4. Kinematic state-space representation of speech production. a, b, A kinematic trajectory (grey-blue) from a single trial (P1) projected onto the first two principal components—PC1 (a) and PC2 (b)—of the kinematic state-space. Decoded audible (dashed) and mimed (dotted) kinematic trajectories also plotted (Pearson’s r, n=510 time samples). The trajectory for mimed speech was uniformly stretched to align with the audible speech trajectory for visualization as it occurred at a faster time scale. c, d, Average trajectories for PC1 (a) and PC2 (b) for transitions from a vowel to a consonant (black, n=22453) and from a consonant to a vowel (white, n=22453). Time courses are 500 ms. e, Distributions of correlations between original and decoded kinematic state-space trajectories (averaged across PC1 and PC2) (n=101, 100, 93, 81, 44 sentences, respectively).. Pearson’s correlations for mimed trajectories were calculated by dynamically time warping (DTW) to the audible production the same sentence and then compared to correlations to DTW of a randomly selected sentence trajectory (p=1e-5, n=58 sentences, Wilcoxon signed-rank test). f, Distributions of correlations for state-space trajectories of the same sentence across participants. Alignment between participants done via DTW and compared to correlations from DTW on unmatched sentence pairs (p=1e-16, n=92; p=1e-8, n=44, respectively, WSRT). g, Comparison between acoustic decoders (Stage 2) (n=101 sentences). “Target” refers to an acoustic decoder trained on data from the same participant that kinematic decoder (stage 1) is trained on (P1). “Transfer” refers to acoustic decoder trained on kinematics and acoustics from a different participant (P2). Box plots as described in Figure 2. *** indicates p<0.001. Table 1. Listener transcriptions of neurally synthesized speech. Examples shown at several word error rate levels. The original text is indicated by “o” and the listener transcriptions are indicated by “t”. Word Error Rate Original sentences (o) and transcriptions of synthesized speech (t) 0% o: is this seesaw safe t: is this seesaw safe ~10% o: bob bandaged both wounds with the skill of a doctor t: bob bandaged full wounds with the skill of a doctor ~20% o: those thieves stole thirty jewels t: thirty thieves stole thirty jewels o: help celebrate brother’s success t: help celebrate his brother’s success ~30% o: get a calico cat to keep the rodents away t: the calico cat to keep the rabbits away o: carl lives in a lively home t: carl has a lively home ~50% o: mum strongly dislikes appetizers t: mom often dislikes appetizers o: etiquette mandates compliance with existing regulations t: etiquette can be made with existing regulations >70% o: at twilight on the twelfth day we’ll have Chablis t: i was walking through chablis The authors declare no competing interests. ==== Refs References: 1. Fager SK , Fried-Oken M , Jakobs T , & Beukelman DR (2019). New and emerging access technologies for adults with complex communication needs and severe motor impairments: State of the science, Augmentative and Alternative Communication, DOI: 10.1080/07434618.2018.1556730 2. Brumberg JS , Pitt KM , Mantie-Kozlowski A , & Burnison JD (2018). Brain–computer interfaces for augmentative and alternative communication: A tutorial. American Journal of Speech-Language Pathology, 27 , 1–12. doi:10.1044/2017_AJSLP-16-0244 29318256 3. Pandarinath C , Nuyujukian P , Blabe CH , Sorice BL , Saab J , Willett FR , … Henderson JM (2017). High performance communication by people with paralysis using an intracortical brain-computer interface. ELife, 6 , 1–27. doi:10.7554/eLife.18554 4. Guenther FH , Brumberg JS , Joseph Wright E , Nieto-Castanon A , Tourville JA , Panko M , … Kennedy PR (2009). A wireless brain-machine interface for real-time speech synthesis. PLoS ONE, 4 (12 ). 10.1371/journal.pone.0008218 5. Bocquelet F , Hueber T , Girin L , Savariaux C , & Yvert B . (2016). Real-time control of an articulatory-based speech synthesizer for brain computer interfaces. PLoS computational biology, 12 (11 ), e1005119. 6. Browman CP , & Goldstein L . (1992). Articulatory phonology: An overview. Phonetica, 49 (3–4 ), 155–180.1488456 7. Sadtler PT , Quick KM , Golub MD , Chase SM , Ryu SI , Tyler-Kabara EC , … & Batista AP (2014). Neural constraints on learning. Nature, 512 (7515 ), 423.25164754 8. Golub MD , Sadtler PT , Oby ER , Quick KM , Ryu SI , Tyler-Kabara EC , … & Yu BM (2018). Learning by neural reassociation. Nat. Neurosci, 21 . 9. Graves A , & Schmidhuber J . (2005). Framewise phoneme classification with bidirectional LSTM and other neural network architectures. Neural Networks, 18 (5–6), 602–610.16112549 10. Crone NE , Hao L , Hart J Jr. , Boatman D , Lesser RP , Irizarry R , and Gordon B . (2001). Electrocorticographic gamma activity during word production in spoken and sign language. Neurology 57 , 2045–2053.11739824 11. Nourski KV , Steinschneider M , Rhone AE , Oya H , Kawasaki H , Howard III MA , & McMurray B . (2015). Sound identification in human auditory cortex: Differential contribution of local field potentials and high gamma power as revealed by direct intracranial recordings. Brain and language, 148 , 37–50.25819402 12. Pesaran B , Vinck M , Einevoll GT , Sirota A , Fries P , Siegel M , … & Srinivasan R . (2018). Investigating large-scale brain dynamics using field potential recordings: analysis and interpretation. Nature neuroscience. 13. Bouchard KE , Mesgarani N , Johnson K , and Chang EF (2013). Functional organization of human sensorimotor cortex for speech articulation. Nature 495 , 327–332.23426266 14. Mesgarani N , Cheung C , Johnson K , & Chang EF (2014). Phonetic feature encoding in human superior temporal gyrus. Science, 343 (6174 ), 1006–1010.24482117 15. Flinker A , Korzeniewska A , Shestyuk AY , Franaszczuk PJ , Dronkers NF , Knight RT , & Crone NE (2015). Redefining the role of Broca’s area in speech. Proceedings of the National Academy of Sciences, 112 (9 ), 2871–2875. 16. Chartier J , Anumanchipalli GK , Johnson K , & Chang EF (2018). Encoding of Articulatory Kinematic Trajectories in Human Speech Sensorimotor Cortex. Neuron, 98 (5 ), 1042–1054.e4. 10.1016/j.neuron.2018.04.031 17. Mugler EM , Tate MC , Livescu K , Templer JW , Goldrick MA , & Slutzky MW (2018) Differential Representation of Articulatory Gestures and Phonemes in Precentral and Inferior Frontal Gyri. J Neurosci. 38 (46 ):9803–9813. doi: 10.1523/JNEUROSCI.30257858 18. Huggins JE , Wren PA , Gruis KL (2011) What would brain-computer interface users want? Opinions and priorities of potential users with amyotrophic lateral sclerosis. Amyotroph Lateral Scler. 2011 Sep;12 (5 ):318–24. doi: 10.3109/17482968.2011.572978.21534845 19. Luce PA & Pisoni DB Recognizing spoken words: the neighborhood activation model. Ear Hear. 19 , 1–36 (1998).9504270 20. Wrench A . (1999). MOCHA: multichannel articulatory database. http://www.cstr.ed.ac.uk/research/projects/artic/mocha.html. 21. Kominek J , Schultz T , and Black A . (2008). “Synthesizer voice quality of new languages calibrated with mean mel cepstral distortion”, In SLTU-2008, 63–68. 22. Davis SB , & Mermelstein P . (1990). Comparison of parametric representations for monosyllabic word recognition in continuously spoken sentences. In Readings in speech recognition (pp. 65–74). 23. Gallego JA , Perich MG , Miller L,E , Solla S,A , (2017) Neural manifolds for the control of movement., Neuron, 94 (5 ), 978–984.28595054 24. Sokal RR , & Rohlf FJ (1962). The comparison of dendrograms by objective methods. Taxon, 33–40. 25. Brumberg JS , Krusienski DJ , Chakrabarti S , Gunduz A , Brunner P , Ritaccio AL , & Schalk G . (2016). Spatio-Temporal Progression of Cortical Activity Related to Continuous Overt and Covert Speech Production in a Reading Task. PloS one, 11 (11 ), e0166872. doi:10.1371/journal.pone.0166872 26. Martin S , Brunner P , Holdgraf C , Heinze H-J , Crone NE , Rieger J , Schalk G , Knight RT , Pasley BN (2014). Decoding spectrotemporal features of overt and covert speech from the human cortex. Front. Neuroeng 7 :14.24904404 27. Mugler EM , Patton JL , Flint RD , Wright ZA , Schuele SU , Rosenow J , Shih JJ , Krusienski DJ , and Slutzky MW (2014). Direct classification of all American English phonemes using signals from functional speech motor cortex. J. Neural Eng 11 , 035015. 28. Herff C , Heger D , de Pesters A , Telaar D , Brunner P , Schalk G , and Schultz T . (2015). Brain-to-text: decoding spoken phrases from phone representations in the brain. 29. Moses DA , Mesgarani N , Leonard MK , & Chang EF (2016). Neural speech recognition: continuous phoneme decoding using spatiotemporal representations of human cortical activity. Journal of neural engineering, 13 (5 ), 056004. 30. Pasley BN , David SV , Mesgarani N , Flinker A , & Shamma SA (2012). Reconstructing Speech from Human Auditory Cortex. PLoS Biol, 10 (1 ), 1001251. 10.1371/journal.pbio.1001251 31. Akbari H , Khalighinejad B , Herrero JL , Mehta AD , & Mesgarani N . (2019). Towards reconstructing intelligible speech from the human auditory cortex. Scientific reports, 9 (1 ), 874.30696881 32. Dichter BK , Breshears JD , Leonard MK , and Chang EF (2018) The Control of Vocal Pitch in Human Laryngeal Motor Cortex. Cell, 174 , 21–31.29958109 33. Wessberg J , Stambaugh CR , Kralik JD , Beck PD , Laubach M , (2000) Real-time prediction of hand trajectory by ensembles of cortical neurons in primates. Nature 408 : 361–365.11099043 34. Serruya MD , Hatsopoulos NG , Paninski L , Fellows MR , Donoghue JP (2002) Instant neural control of a movement signal. Nature 416 : 141–142.11894084 35. Taylor DM , Tillery SI , Schwartz AB (2002) Direct cortical control of 3D neuroprosthetic devices. Science 296 : 1829–1832.12052948 36. Hochberg LR , Serruya MD , Friehs GM , Mukand JA , Saleh M , Caplan AH , … & Donoghue JP (2006). Neuronal ensemble control of prosthetic devices by a human with tetraplegia. Nature, 442 (7099 ), 164 16838014 37. Collinger JL , Wodlinger B , Downey JE , Wang W , Tyler-Kabara EC , Weber DJ , … & Schwartz AB (2013). High-performance neuroprosthetic control by an individual with tetraplegia. The Lancet, 381 (9866 ), 557–564. 38. Aflalo T , Kellis S , Klaes C , Lee B , Shi Y , Pejsa K , … & Andersen RA (2015). Decoding motor imagery from the posterior parietal cortex of a tetraplegic human. Science, 348 (6237 ), 906–910.25999506 39. Ajiboye AB , Willett FR , Young DR , Memberg WD , Murphy BA , Miller JP , … & Peckham PH (2017). Restoration of reaching and grasping movements through brain-controlled muscle stimulation in a person with tetraplegia: a proof-of-concept demonstration. The Lancet, 389 (10081), 1821–1830. 40. Prahallad K , Black AW , & Mosur R . (2006). Sub-phonetic modeling for capturing pronunciation variations for conversational speech synthesis. In Proceedings of the 2006 IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), pp. I–I. 41. Anumanchipalli GK , Prahallad K , & Black AW (2011). Festvox: Tools for creation and analyses of large speech corpora, Workshop on Very Large Scale Phonetics Research, UPenn, Philadelphia. http://www.festvox.org 42. Hamilton LS , Chang DL , Lee MB , & Chang EF (2017). Semi-automated Anatomical Labeling and Inter-subject Warping of High-Density Intracranial Recording Electrodes in Electrocorticography. Frontiers in Neuroinformatics, 11 , 62. 10.3389/fninf.2017.00062 29163118 43. Richmond K , Hoole P , & King S . (2011). Announcing the electromagnetic articulography (Day 1) subset of the mngu0 articulatory corpus Proceedings of Interspeech 2011, Florence, Italy 44. Paul BD , & Baker M,J (1992). The design for the wall street journal-based CSR corpus. In Proceedings of the workshop on Speech and Natural Language (HLT ‘91). Association for Computational Linguistics, Stroudsburg, PA, USA, 357–362. DOI: 10.3115/1075527.1075614 45. Abadi Martín , Agarwal Ashish , Barham Paul , Brevdo Eugene , Chen Zhifeng , (2015). TensorFlow: Large-scale machine learning on heterogeneous systems. http://www.tensorflow.org 46. Hochreiter S , and Schmidhuber J . (1997). Long short-term memory. Neural Comput. 9 , 1735–1780.9377276 47. Maia R , Toda T , Zen H , Nankaku Y , Tokuda K , 2007. An excitation model for HMM-based speech synthesis based on residual modeling. In: Proc. ISCA SSW6, pp. 131–136. 48. Wolters MK , Isaac , Renals S , Evaluating Speech Synthesis intelligibility using Amazon Mechanical Turk. (2010) In proceedings of ISCA speech synthesis workshop (SSW7), 2010. 49. Berndt DJ , & Clifford J . (1994). Us ing dynamic time warping to find patterns in time series. In KDD workshop (Vol. 10, No. 16, pp. 359–370).	31019317	PMC9714519	NO-CC CODE	2022-12-02 23:24:49	no	Nature. 2019 Apr 24; 568(7753):493-498	utf-8	Nature	2,019	10.1038/s41586-019-1119-1	oa_other
==== Front Environ Health Perspect Environ Health Perspect EHP Environmental Health Perspectives 0091-6765 1552-9924 Environmental Health Perspectives 36454223 EHP10103 10.1289/EHP10103 Research Bayesian Estimation of Human Population Toxicokinetics of PFOA, PFOS, PFHxS, and PFNA from Studies of Contaminated Drinking Water https://orcid.org/0000-0002-7575-2368 Chiu Weihsueh A. 1 2 https://orcid.org/0000-0003-0184-5307 Lynch Meghan T. 3 https://orcid.org/0000-0001-9517-4852 Lay Claire R. 3 Antezana Adriana 3 Malek Parker 3 Sokolinski Sara 3 https://orcid.org/0000-0002-9716-1075 Rogers Rachel D. 4 1 Interdisciplinary Faculty of Toxicology, Texas A&M University, College Station, Texas, USA 2 Department of Veterinary Physiology and Pharmacology, School of Veterinary Medicine and Biomedical Sciences, Texas A&M University, College Station, Texas, USA 3 Abt Associates, Cambridge, Massachusetts, USA 4 Centers for Disease Control and Prevention/Agency for Toxic Substances and Disease Registry, Atlanta, Georgia, USA Address correspondence to Meghan T. Lynch, 10 Fawcett St., Cambridge, MA 02138 USA. Email: [email protected] 1 12 2022 12 2022 130 12 12700109 8 2021 03 8 2022 27 10 2022 https://ehp.niehs.nih.gov/about-ehp/license EHP is an open-access journal published with support from the National Institute of Environmental Health Sciences, National Institutes of Health. All content is public domain unless otherwise noted. Background: Setting health-protective standards for poly- and perfluoroalkyl substances (PFAS) exposure requires estimates of their population toxicokinetics, but existing studies have reported widely varying PFAS half-lives (T½) and volumes of distribution (Vd). Objectives: We combined data from multiple studies to develop harmonized estimates of T½ and Vd, along with their interindividual variability, for four PFAS commonly found in drinking water: perfluorooctanoic acid (PFOA), perfluorooctane sulfonate (PFOS), perfluorononanoic acid (PFNA), and perfluorohexane sulfonate (PFHxS). Methods: We identified published data on PFAS concentrations in human serum with corresponding drinking water measurements, separated into training and testing data sets. We fit training data sets to a one-compartment model incorporating interindividual variability, time-dependent drinking water concentrations, and background exposures. Use of a hierarchical Bayesian approach allowed us to incorporate informative priors at the population level, as well as at the study level. We compared posterior predictions to testing data sets to evaluate model performance. Results: Posterior median (95% CI) estimates of T½ (in years) for the population geometric mean were 3.14 (2.69, 3.73) for PFOA, 3.36 (2.52, 4.42) for PFOS, 2.35 (1.65, 3.16) for PFNA, and 8.30 (5.38, 13.5) for PFHxS, all of which were within the range of previously published values. The extensive individual-level data for PFOA allowed accurate estimation of population variability, with a population geometric standard deviation of 1.57 (95% CI: 1.42, 1.73); data from other PFAS were also consistent with this degree of population variability. Vd estimates ranged from 0.19 to 0.43L/kg across the four PFAS, which tended to be slightly higher than previously published estimates. Discussion: These results have direct application in both risk assessment (quantitative interspecies extrapolation and uncertainty factors for interindividual variability) and risk communication (interpretation of monitoring data). In addition, this study provides a rigorous methodology for further refinement with additional data, as well as application to other PFAS. https://doi.org/10.1289/EHP10103 Supplemental Material is available online (https://doi.org/10.1289/EHP10103). The authors declare they have no actual or potential competing financial interests. Note to readers with disabilities: EHP strives to ensure that all journal content is accessible to all readers. However, some figures and Supplemental Material published in EHP articles may not conform to 508 standards due to the complexity of the information being presented. If you need assistance accessing journal content, please contact [email protected]. Our staff will work with you to assess and meet your accessibility needs within 3 working days. ==== Body pmcIntroduction Poly- and perfluoroalkyl substances (PFAS) are man-made chemicals consisting of chains of linked carbon and fluorine atoms. Their structure gives them unique physicochemical properties that have been found useful in a range of consumer and industrial products. However, these properties also lead to high water solubility, bioaccumulation, persistence in the environment, and resistance to degradation, leading them to be called “forever chemicals.” Owing to the widespread use of PFAS and their high solubility in water, PFAS contamination has been reported in drinking water throughout the United States (Cordner et al. 2019; Evans et al. 2020; Hu et al. 2016). Biomonitoring data have confirmed that PFAS exposure is widespread in humans (CDC 2019; Daly et al. 2018; Olsen et al. 2017; Yu et al. 2020). The National Health and Nutrition Examination Survey (NHANES) has measured PFAS in serum samples from the general population since 1999, finding detectable levels in the blood of 99% of the population >12 years of age (CDC 2019). Finally, several well-studied PFAS, including perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS), have been linked to adverse human health effects (ATSDR 2021; IARC 2017; NTP 2016; Sunderland et al. 2019). For the many thousands of other PFAS that have been released into the environment, few toxicological data exist. Some PFAS have been phased out in manufacturing processes in much of the world, and serum concentrations for some long-chain PFAS are declining in the United States (CDC 2019). However, PFAS contamination in water persists, and tools to estimate exposure and health effects from drinking water concentrations (DWCs) are useful for public health and regulatory work. This study used data from published human studies linking DWCs to serum concentrations to produce robust estimates of key pharmacokinetic parameters for four PFAS. Since the early 2000s, the U.S. Environmental Protection Agency (EPA) and many state agencies have embarked on monitoring programs and the development of drinking water advisory levels (e.g., CA Water Boards 2020; MDH 2021; EGLE 2020; NJDEP 2021; U.S. EPA 2016). State guidelines are generally <50 ng/L. For instance, the Minnesota Department of Health (MDH) has set drinking water standards for the state for perfluorohexane sulfonate (PFHxS), PFOA, and PFOS of 47, 35, and 15 ng/L, respectively (MDH 2021). The State of New Jersey has established enforceable drinking water standards for PFOA, PFOS, and perfluorononanoic acid (PFNA) where the maximum contaminant levels are set at 14, 13, and 13 ng/L, respectively (NJDEP 2021). The State of California has established notification levels for PFOS and PFOA of 6.5 and 5.1 ng/L, respectively (CA Water Boards 2020). The State of Michigan has set maximum contaminant levels for PFNA, PFOA, PFOS, and PFHxS at 6, 8, 16, and 51 ng/L, respectively (EGLE 2020). Most recently, the U.S. EPA has set much lower interim health advisory levels for PFOA (0.004 ng/L) and PFOS (0.02 ng/L) (U.S. EPA 2022). A critical component for risk assessment and risk communication regarding PFAS involves quantifying their toxicokinetics in human populations. Several widespread PFAS appear to be eliminated extremely slowly in humans, with half-life (T½, the time it takes for serum concentrations to decline by 50%) measured in years, whereas T½ in experimental animals are on the order of days or weeks. Numerous human toxicokinetics models for individual PFAS have been developed, ranging from simple compartmental models to physiologically based pharmacokinetic (PBPK) models. However, for regulatory and public health applications, the focus has been on the one-compartment model (Egeghy and Lorber 2011; Lorber and Egeghy 2011). The key parameters for this one-compartment model are the T½ and volume of distribution (Vd, representing a chemical’s tendency to remain in the blood or to distribute to other body tissues), given that these parameters together are sufficient to derive chemical-specific adjustment factors for interspecies and interindividual variability (WHO et al. 2005; U.S. EPA 2014). However, existing human studies report wide ranges of T½ even for relatively well-studied PFAS. For example, depending on the study and population, T½ estimates for PFOA have ranged from 3 to 10 y, and estimates for PFOS range even wider, from 3 to 27 y (Bartell et al. 2010; Costa et al. 2009; Harada et al. 2005; Li et al. 2018; Olsen et al. 2007; Seals et al. 2011; Wong et al. 2014; Worley et al. 2017a, 2017b; Zhang et al. 2013). Several studies have also reported substantial interindividual variability in T½ (Li et al. 2018; Olsen et al. 2007). Moreover, the data available on Vd are sparse compared with the data available for T½ (Koponen et al. 2018). Accurate estimates of Vd are important for converting between T½ and clearance (e.g., Zhang et al. 2013). Estimating Vd and T½ for long-lived substances such as PFAS is challenging for several reasons. First, time-course data are required over several years to accurately estimate elimination. However, the ubiquity of PFAS means that ongoing exposure needs to be well characterized to avoid overestimating T½. For Vd it is also necessary to have quantitative information on levels of exposure, which is often unavailable. Further, interindividual variation can make it difficult to distinguish measurement error from true heterogeneity. Characterizing this variation is also important to ensure adequate health protection across the population, including susceptible subgroups. Given that Vd and T½ are the main parameters in one-compartment pharmacokinetic models, including those that include subcompartments addressing placental or lactational transfer, harmonized estimates of their values, along with estimates of the extent of interindividual variability, are needed to support public health actions related to PFAS. In this study, we aimed to estimate the population toxicokinetics of four PFAS often found in drinking water: PFOA, PFOS, PFNA, and PFHxS (U.S. EPA 2017; EWG 2019a). Unlike previous studies that analyzed individual data sets, we took a more integrative approach and combined individual-level data from multiple studies for which detailed information on drinking water contamination, background exposures, and serum concentrations were available. Similar to previous work integrating PFAS toxicokinetics data across nonhuman animal species (Wambaugh et al. 2013), we employed a Bayesian approach, which enabled incorporation of prior knowledge, statistically rigorous incorporation of multiple data sets, better accommodation of unobserved variables, and quantitative characterization of uncertainty (see Dunson 2001; Silver 2012; Nuzzo 2015). Although this work is complementary to Wambaugh et al. (2013), because of the importance of characterizing human interindividual variability, we used a hierarchical Bayesian approach that adds random effects to model population variability. Methods Water and Serum PFAS Data The goal of this data collection was to find published PFAS concentrations in human serum and drinking water levels to predict those serum levels. We located human serum concentrations and corresponding data on DWCs by searching databases such as PubMed, ScienceDirect, and Google Scholar for English language journal articles and reports that presented human serum data for PFOA, PFOS, PFNA, or PFHxS levels. We also conducted a tree search on identified review articles and reports. We did not apply constraints on publication date, but the literature review to identify studies was concluded in 2019. We focused on populations exposed through contaminated drinking water, as defined by study authors. When serum data for additional PFAS were identified in studies where contamination of one PFAS was indicated prior to conducting the study, we did also collect these serum concentrations and include them if we could also identify corresponding drinking water information. In some cases, when PFAS levels in drinking water were reported as below the minimum reporting level (MRL) for the U.S. EPA’s third Unregulated Contaminant Monitoring Rule (UCMR3), but identified in serum of the study participants, we estimated the DWC as below the MRL as part of the analysis. We excluded studies focused on occupational cohorts, and where possible, we excluded individuals in community studies also known to have likely occupational exposures, such as employees of some chemical companies or firefighters, because our goal was to develop a tool that is predictive of community PFAS levels. For individual-level data, this information was only available for the Decatur, Alabama, data set, and so other data sets may include individuals with significant occupational exposure. Although several studies collected some information about occupation, only Bartell et al. (2010) and Emmett et al. (2006) specifically detailed how they excluded those with known or likely occupational exposure to the PFAS. All data were obtained from previous publications, and this analysis did not require institutional review board approval. To be included in our model, a study needed to provide sufficient detail so that we could understand the timing of the PFAS exposure and map the serum levels to relevant DWCs for the study participants if the study did not explicitly give these concentrations. We reviewed the background, methods, results, figures, and tables of each study for reporting of community- and individual-level human serum data. Data extraction included recording the individual serum level if available, as well as the mean, geometric mean (GM), median, minimum, maximum, standard deviation (SD), sample size, dates of sampling, geographic location, and relevant demographic information, such as participant age and sex. When they were provided, we recorded dates of PFAS water contamination and remediation to better estimate participant exposure. If a study provided water consumption or finished water concentration data along with serum levels, we also collected this information. We aimed to develop as complete a picture as possible of DWCs over time prior to the serum measurements. When participant geographic information such as ZIP code, city, or water district service area was reported, we searched PFAS water concentration data to match them to serum data by location and dates. We obtained finished water concentration data for matching locations from journal articles, service districts’ water quality reports, and data collected under the U.S. EPA’s UCMR3. We used the Environmental Working Group’s (EWG’s) National Tap Water Database (U.S. EPA 2017; EWG 2019a) to identify additional sources of drinking water concentration data but verified data contained in the EWG database with the original sources. We did not include source water concentrations, and we included finished water data for all available time periods, including during participant sampling, before remediation, or after remediation, if available. Data recorded included water collection data, water type, arithmetic mean, GM, median, minimum, maximum, SD, and number of samples. The authors of several human serum studies reported the PFAS concentration that participants were exposed to in their drinking water, and in those cases, we recorded the PFAS concentration for the location directly from the serum-level studies. To be included for analysis, studies with time-course serum data had to have available water concentration data during the time period between the first and last serum collection. For studies with only a single serum collection, water concentration data had to precede the serum collection date. If we could not reliably match study participants with reliable water concentrations, we excluded those serum levels. In addition, owing to limited data, no data sets focusing exclusively on children were included, and for a single study with both children and adults >18 years of age, only the adult values were used in estimation. We excluded individuals with a known potential for occupational PFAS exposure, those who could not be linked to specific water concentrations, those who reported filtering their water, and those with missing values for important variables such as weight and sex. We did include populations drinking bottled water in the testing data sets. We ultimately included data from nine sites for parameter estimation. Study data sets, citations, and location-specific water concentration handling are listed in Table 1 and, in further detail, in Tables S1 and S2. Table S1 includes details on why a location was excluded, which was typically due to a lack of sufficient information on water concentrations or serum concentrations. Table 1 Studies of PFAS-contaminated community drinking water and serum concentrations used. City, state/country Studies/data sources Water data dates Serum data dates (time points) (n) PFAS data used Individuals or (populations) (n) Train/test data LOD Comments Arnsberg, Germany Hölzer et al. 2008, 2009 2006–2007 2006–2007 (2) PFOA 220 110/110 10 ng/L Time-varying water concentration between serum collections. PFOS only has summary data, but inadequate to calculate population arithmetic mean. Decatur, Alabama ATSDR 2013, 2016 2010–2016 2010–2016 (2) PFOA, PFOS, PFNA, PFHxS 37 18/19 10 ng/L Time-varying water concentration between serum collections. Horsham, Pennsylvania U.S. EPA 2017; HWSA 2014, 2018; Penn DOH 2019 2014–2015 2018 (1) PFOA, PFOS, PFNA, PFHxS (1) (1/0) 20 ng/L (PFOA; PFNA) 30 ng/L (PFHxS) 40 ng/L (PFOS) Assumed steady state until intervention in July 2016, 2 y prior to serum data collection. Lake Elmo/ Cottage Grove, Minnesota Johnson et al. 2017 2005–2008 2008 (1) PFOA, PFOS PFOA: 95, PFOS: 98 PFOA: 48/47; PFOS: 49/49 NR Matched well water and serum measurements; assumed steady state. Little Hocking, Ohio Bartell et al. 2010 2007–2008 2007–2008 (2) PFOA (2) (1/1) 16 ng/L Public water population used for training, bottled water population for testing; only included data after intervention in November 2007. Little Hocking, Ohio Emmett et al. 2006 2002–2005 2004–2005 (1) PFOA (1) (0/1) 10 ng/L Assumed steady state. Lubeck, West Virginia Bartell et al. 2010 2007–2008 2007–2008 (3) PFOA (2) (1/1) 16 ng/L Public water population used for training, bottled water population for testing; all data after intervention in June 2007. Paulsboro, New Jersey Graber et al. 2019; Post et al. 2013 2009–2013 2016 (1) PFOA, PFOS, PFNA, PFHxS (1) (1/0) 5 ng/L Assumed steady state until intervention in April 2014, 2.2 y prior to serum data collection. Warminster, Pennsylvania U.S. EPA 2017; Penn DOH 2019; WMA 2018 2013–2014 2018 (1) PFOA, PFOS, PFNA, PFHxS (1) 0/1 NR Assumed steady state until intervention in July 2016, 2 y prior to serum data collection. Warrington, Pennsylvania EWG 2019b; Warrington Township 2014, 2018; Penn DOH 2019 2014–2015 2018 (1) PFOA, PFOS, PFHxS (1) 0/1 NR Assumed steady state until intervention in July 2016, 2 y prior to serum data collection. Note: Table shows location, dates, PFAS, population size, training/testing data set size as individuals or (populations), and limit below which water concentrations were reported as nondetects (LOD). Three identified study populations were excluded (see text). When water concentrations are below detection or reporting limits, a uniform prior distribution between 0 and the concentration limit is used. Additional information regarding water concentrations is in Table S2. LOD, level of detection; NR, not reported; PFAS, poly- and perfluoroalkyl substances; PFOA, perfluorooctanoic acid; PFOS, perfluorooctane sulfonic acid; PFNA, perfluorononanoic acid; PFHxS, perfluorohexane sulfonate. Toxicokinetic Model In keeping with previous and current PBPK modeling by the U.S. EPA in adult (nonpregnant) humans (Lorber and Egeghy 2011; U.S. EPA 2016), we used a one-compartment toxicokinetic model for each PFAS. Moreover, the parameters of the one-compartment model are also included in human PBPK models that include pregnancy and lactation (MDH 2021; NJDEP 2021). Thus, this approach will enable direct use of our modeling results in a wide range of PFAS public health assessments. The parameters were defined as follows: Cbgd is the background serum concentration (in micrograms per liter); C0 is the initial serum concentration (in micrograms per liter) at t=0; DWIBW is the drinking water intake on a body weight-adjusted basis (e.g., as liters per kilogram per day); k=ln(2)/T1/2 is the rate constant per day; and Vd is the Vd (in liters per kilogram). All of these parameters are assumed to be constant for the duration of the simulation, except for the DWC, DWC(t) (in micrograms per liter), which may vary over time. The ordinary differential equation (ODE) for the concentration of PFAS in the body is as follows: (1) dC(t)dt=DWIBWDWC(t)Vd+k[Cbgd−C(t)]. As shown in Figure 1, this model is equivalent to the usual one-compartment model used for pharmaceuticals, but with the usual bolus dose (typically modeled as either an initial condition or an exponential input) replaced by a time-varying input that includes both a drinking water component (first term) and a background exposure component. Moreover, the background dose rate (e.g., Dbgd, in micrograms per kilogram per day) has been reparameterized in terms of the background serum concentration, Cbgd=Dbgd/(kVd). This reparameterization improves model fitting by reducing parameter correlations. In addition, because clearance is used as the basis for both chemical-specific interspecies extrapolation and interindividual variability factors (WHO et al. 2005; U.S. EPA 2014), CL (in liters per kilogram per day) =k×Vd was also calculated. Figure 1. Schematic of modeling approach. A one-compartment pharmacokinetic model in the form of an ordinary differential equation (ODE) for serum concentration C(t) is the basis at the individual level. This model accounts for both drinking water exposure, as well as background exposure, and has parameters of body weight (BW), drinking water intake per unit BW (DWIBW), drinking water concentration (DWC, which can be function of time t), background serum concentration (Cbgd), elimination rate k, and volume of distribution Vd. If data are only at a single time point, then steady state is assumed, whereas if there are data at multiple time points, the ODEs are solved numerically. If only summary data are available, then the population mean is predicted for comparison. The Bayesian calibration uses prior distributions for each parameter, some of which are study specific. Markov chain Monte Carlo (MCMC) simulations are used to generate posterior distributions for the population mean and population geometric standard deviation for each parameter. See the “Methods” section for additional details. Note: NHANES, National Health and Nutrition Examination Survey; PFAS, poly- and perfluoroalkyl substances. Figure 1 is a schematic flowchart having five steps. Step 1: Prior distributions with study specific data, including measure drinking water concentrations and background serum levels based on concurrent N H A N E S data lead to study. Step 2: Study includes individual data, single time point, multiple time points, and Only summary data. Individual data includes One-compartment model: change with respect to time of serum concentration equals drinking water intake on a body weight adjusted basis times drinking water concentration open parenthesis lowercase italic t closed parenthesis over volume of distribution plus lowercase italic k open bracket background serum concentration minus serum concentration closed bracket. Single time point leads to Assume steady state. Multiple time points lead to Solve ordinary differential equations. Only summary data: assume steady state and Solve ordinary differential equations lead to population mean. Step 3: One-compartment model leads to drinking water exposure, including body weight times drinking water intake on a body weight adjusted times drinking water concentration open parenthesis lowercase italic t closed parenthesis and background exposure, including body weight times volume of distribution times lowercase italic k times background serum concentration lead to amount of per- and polyfluoroalkyl substances in body: uppercase italic a open parenthesis lowercase italic t close parenthesis equals serum concentration times body weight times volume of distribution. Step 4: amount of per- and polyfluoroalkyl substances in body: uppercase italic a open parenthesis lowercase italic t close parenthesis equals serum concentration times body weight times volume of distribution lead to lowercase italic k times uppercase italic a open parenthesis lowercase italic t close parenthesis which causes first order elimination. Step 5: Study with Bayesian M C M C simulations lead to posterior distributions. For a constant water concentration, Equation 2 has an analytic solution for the serum concentration as a function of time: (2) C(t)=Cbgd+(C0−Cbgd)e−kt+Css(1−e−kt), where C0 is the concentration at time 0, and Css, the steady-state serum concentration (due to drinking water alone, without background), is given by Equation 3: (3) Css=DWIBWDWCkVd. The first term in Equation 2 is simply the background concentration, the second term describes the time-dependence from time 0 to background in the absence of water contamination, and the third term describes the time-dependence of the transition to steady state. If data are available only at a single time point, then steady state is assumed, and DWCs are assumed to be constant: (4) Cbgd+ss=Cbgd+Css. In some cases, the data at a single time point involve an intervention in which the source of contamination was removed at a certain time interval, Δt, in the past. This simplification of intervention is necessary when no water concentrations for after intervention are available. In this case, the serum concentration takes the following form: (5) Cbgd+ss+Δt=Cbgd+Csse−kΔt. Bayesian Population and Statistical Model We employed a Bayesian population approach to estimate model parameters. This approach involves two statistical models: one for population variability and one for the likelihood of the observed data (serum concentrations) given a set of model parameters (such as T½ and Vd). With respect to the population variability model, each individual i is assumed to have their own (unknown) model parameters θi and serum concentration data Ci at time tj. The individual parameters (log-transformed) are drawn from a normal population distribution: (6) log θi=μθ+αθ,iΣθ; αθ,i∼N(0,1). Here, αθ,i represents the z-score of the individual in the population for parameter θ. For some parameters, the population mean, μθ, and SD, Σθ, may be fixed (e.g., background, drinking water intake), whereas for others, μθ and Σθ may have distributions that are due to uncertainty (e.g., T½, Vd). With respect to the statistical model for the likelihood of the observed data, we assume lognormal errors with an additional parameter to specify the SD of the error distribution, so the likelihood function for an observation, Ci,obs(tj), in individual i at time point tj would be defined as follows: (7) Ci,obs(tj)=Ci,pred(tj)eϵij; ϵij∼N(0,σerr); GSDerr=eσerr, where Cpred is the model prediction for serum concentration as a function of time as described above. Different values for the error of the geometric standard deviation (GSDerr) are used for individual time-course data, individual steady-state data, and summary-level steady-state data. For individual data, there are two common situations. In the first, time-course data are available for the individual. In this case, the model sets t=0 at the initial sampling time point, where C(0)=C0. We then evaluate subsequent time points at times after the initial sample point using Equation 1 for time-varying water concentrations and Equation 2 for constant water concentrations. This approach avoids the uncertainty regarding what the background and water concentrations were before the first sampling point but requires data on the water concentrations between the first and last serum sampling point. In the second situation, serum data are available only at a single time point, but historical water concentration data are available. In this case, we use Equation 5, where steady state is assumed until the point of intervention (if any), after which the water concentration level is assumed to be negligible. This simplification allowed us to include data from study locations with no postintervention water concentration data. If only summary data are available, then the only statistic that the model can directly predict at the population level is the population arithmetic mean because of the presence of the background term. For instance, at steady state, taking the mean of Equation 4, we have the following: (8) 〈Cbgd+ss〉=〈Cbgd〉+〈Css〉. The Cbgd term has a population mean as follows: (9) 〈Cbgd〉=exp[μCbgd+12ΣCbgd2]. Because each of the parameters—DWIBW, DWC, k, and Vd—that make up Css are assumed to be lognormally distributed in the population, their product (and quotient) is also lognormally distributed. Thus, the population mean of Css is given as follows: (10) 〈Css〉=〈DWIBWDWCkVd〉=exp[μCss+12ΣCss2], where the population GM (eμ) and geometric standard deviation (GSD) (eΣ) of Css are related to those of the other parameters as follows: (11) μCss=μDWIBW+μCDW−μk−μVd (12) and ΣCss2=ΣDWIBW2+ΣCDW2+Σk2+ΣVd2 . In the case of an intervention at time Δt prior to the serum measurement, taking the population mean of Equation 5 gives the following: (13) 〈Cbgd+ss+Δt〉=〈Cbgd〉+〈Csse−kΔt〉. Because the term e−kΔt cannot be expressed exactly in a formula, we applied an approximation. When appearing in the exponential, k is assumed to be normally, rather than lognormally, distributed, with the mean and variance matched to the actual distribution of k. This implies that e−kΔt is lognormally distributed, with GM e−〈k〉Δt and GSD eΔtVAR(k). Then, we make an additional approximation of independence so that each term above consists of a product of independent lognormal distributions. Thus, the product Y=Cssekt has a population mean as follows: (14) 〈Y〉=〈Csse−kΔt〉=exp[μY+12ΣY2], (15) where μY=μCss−Δt×exp(μk+Σk22) (16) and ΣY2=ΣCss2+(Δt)2×[exp(Σk2)−1]exp[2μk+Σk2]. A similar approach can be used for summary data consisting of multiple time points as long as the DWC is constant. Taking the arithmetic mean of Equation 2 and separating the terms gives the following: (17) 〈C(t)〉=〈Cbgd〉+〈C0e−kt〉−〈Cbgde−kt〉+〈Css〉−〈Csse−kt〉. For the terms 〈C0e−kt〉 and 〈Cbgde−kt〉, we can use the same approximation from Equations 14–16, replacing the μCss and ΣCss2 with the corresponding values for C0 and Cbgd. We checked the approximations in Equations 14–17 with simulations and found them to have errors mostly of <10% for Δt up to about twice the T½. Prior Distributions Prior distributions for each model parameter are summarized in Table 2 (for additional detail, see Table S3). For the background serum concentrations, Cbgd, NHANES data collected closest to the year(s) during which the study data were collected were used as the prior central estimate (Table S3). However, 80% of the appropriate NHANES level was used as Cbgd, corresponding to a relative source contribution from drinking water of 20%, so as to avoid double-counting background exposures with exposures due to the measured DWCs. This assumption is consistent with U.S. EPA drinking water guidance, which as a default assumes that 20% of exposure to a contaminant is from drinking water in the absence of chemical-specific data (U.S. EPA 2018). For time-course data, the prior central estimate for the initial concentration C0 was set to the reported initial value for each individual. For drinking water intake, the population distribution was fixed based on community water source intake data from the Food Commodity Intake Database “What We Eat in America” (FoodRisk 2020) for the years 2005–2010 for those 16–81 years of age (consumers only). For residual error, each data type (individual time-course, individual steady-state, and summary steady-state) involves different assumptions, so each is assumed to have a different residual error. Table 2 Model parameters and prior distributions used in Bayesian parameter calibration. Parameter (units) Description Fixed value or prior for population GM=eμ Fixed value or prior for population GSD=eΣ Comments (see text for details) Cbgd (μg/L) Background serum concentration μ∼ LN (GMy, 1.5) 1.2 Year-specific prior (see Table S3). C0 (μg/L) If >1 time point, initial serum concentration μ∼ LN (GMindiv, 1.5) 1.2 Individual-specific prior. DWIBW (mL/kg per day) Drinking water intake (body weight-adjusted) 12.33 2.43 Fixed, ages 16–81 y, consumers only. k (per year) Elimination rate constant μ∼ LN (GMPFAS, 1.5) PFOA: Σ−2∼Γ (9, 0.75) Other PFAS: based on PFOA posterior PFAS-specific prior (see text). Vd (L/kg) Volume of distribution μ∼ LN (GMPFAS, 1.3) Σ∼ HN (0, 0.2) PFAS-specific prior (see text). GSDerr,t, GSDerr,ss, GSDerr,sum Residual error for individual time-course data (t) or steady-state data (ss), and summary steady-state data (sum) GSDerr∼LUnif (1.1, 10) NA Different data types have different assumptions, so are assumed to have different residual errors. DWC<MRL (μg/L) Drinking water concentration when below minimum reporting level (MRL) DWC<MRL ∼Unif (0, MRL) NA If DWC is below reporting level, assume uniform distribution from 0 to reporting level. Note: The lognormal distribution is specified by LN (GM, GSD). The gamma distribution is specified by Γ(α,β) for shape and rate parameters α and β, respectively. The half-normal distribution is specified by HN (M, SD) and is defined only by positive values. The log-uniform distribution is specified by Lunif (min, max), and the uniform distribution is specified by Unif (min, max). MRL is the minimum reporting level in the drinking water testing. DWC, drinking water concentration; Err, error; GM, geometric mean; GSD, geometric standard deviation; HN, half-normal distribution; indiv, individual; LN, lognormal distribution; LUnif, log-uniform distribution; max, maximum; min, minimum; NA, not applicable; PFAS, poly and perfluoroalkyl substances; PFOA, perfluorooctanoic acid; PFOS, prefluorooctane sulfonic acid; Unif, uniform distribution. For elimination rates, we assigned informative prior distributions for each PFAS. We reviewed the available body of literature on T½ for the four PFAS of interest. We focused on those studies that estimated T½ from consuming contaminated drinking water. We focused exclusively on nonoccupational populations exposed through contaminated drinking water given that this is a common exposure pathway of relevance to public health and eliminates any potential differences in toxicokinetics associated with high occupational exposure levels. Therefore, we excluded occupationally exposed cohorts without drinking water measurements because ongoing exposures in the community can confound T½ estimates (e.g., Costa et al. 2009; Olsen et al. 2007). We also excluded studies that estimated elimination from urine alone because total elimination may include other pathways for some PFAS (e.g., Harada et al. 2005; Zhang et al. 2013) and sufficient data were available for estimation without including those studies. However, the results of these studies were considered in setting the bounds on the prior distributions for T½. For PFOA, prior distributions were based on the estimates from Bartell et al. (2010) and Seals et al. (2011), the smallest and largest T½ values identified, respectively. The population mean was centered on the mean log elimination rate [corresponding to a median T½ of 4.6 y with uncertainty GSD=1.5, so the 95% confidence interval (CI) of 2.1 to 10.2 covers both studies]. Population variation was based on the reported 95% interval of individual T½ from Bartell et al. (2010) (log SD=0.286). Assuming the precision Σ−2 has a prior uncertainty coefficient of variation (CV)=α−½=33%, this implies a shape parameter α=9. The rate parameter β is then derived by matching the prior mean=α/β=0.286−2≈12. For PFOS, this is based on the range of several studies, centered on a GM of 4.8 from Olsen et al. (2007) and the same GSD as PFOA. Based on this range, we dropped a study that found a T½ of 27 y for PFOS in adults >50 years of age (Zhang et al. 2013) given that it is an outlier compared with the other studies. For population variation, because PFOA has the most data, we used the posterior from PFOA as the prior for PFOS and the other PFAS. For PFNA and PFHxS, the prior central estimates for the T½ were 4.3 y (from Zhang et al. 2013) and 5.3 y (from Li et al. 2018), respectively. Limited T½ data were available for PFNA; therefore, we used the data from Zhang et al. (2013) to center our prior distribution even though these data were based on blood–urine pairs. We chose to use the higher estimate (4.3 in all males and females >50 years of age) vs. the lower (2.5 y in younger females) as our central estimate for the prior distribution given that the elimination rate is hypothesized to be longer for the longer-chained PFAS (PFNA is a 9-carbon chain PFAS) (Graber et al. 2019). For PFHxS, we chose to center our estimate on the study by Li et al. (2018) that the MDH used in their PBPK model for PFHxS. T½ for PFHxS also ranged widely in the literature, from 4.7 y (females only) to 15.5 y (Li et al. 2018; Worley et al. 2017a), when excluding the urine-based outlier of 35 y from Zhang et al. (2013). We based Vd priors on the same literature search as was conducted for T½. The prior distribution for the population GM was centered on 0.17L/kg for PFOA and 0.23L/kg for PFOS, based on Thompson et al. (2010) with uncertainty GSD=1.3 for both PFOA and PFOS. The prior distribution for population variation was based on a weakly informative half-normal prior (per a recommendation by Gelman 2006) with mean of 0.16. For PFHxS, the prior central estimate was 0.25L/kg, based on Sundström et al. (2012), which found a Vd range of 0.2 to 0.3L/kg informed by rat, mice, and monkey data, and Koponen et al. (2018), which assumed the same Vd for PFHxS as for PFOS. Owing to a lack of data, the prior central estimate for PFNA was assumed to be the same as for PFOA (0.17L/kg), consistent with the assumption made in Koponen et al. (2018), which was informed by rodent data. For some studies, the DWCs are reported to be “below the minimum reporting level.” In this case, a uniform distribution uncertainty between zero and the reporting level is used as a prior for the actual DWC. Model Implementation and Evaluation Data were separated into “training” and “testing” data sets, with only training data used for calibration (i.e., part of the likelihood function). For studies with individual data, training data consisted of half the individuals of each sex, randomly selected, with the remaining individuals treated as testing data sets. For studies with summary data only, half of the studies were selected for training and half for testing. In both cases, if an odd number of individuals or studies were present, the additional case was randomly assigned to either testing or training. We implemented the model in the open-source software MCSim (version 6.1.0; GNU MCSim; https://www.gnu.org/software/mcsim/), and all analyses were performed in R within RStudio (R Development Core Team; version 3.6.1; Rstudio Team). Plotting and summarization were performed with tidyverse packages (Wickham et al. 2019). Four independent Markov chain Monte Carlo (MCMC) chains were run for each PFAS. We assessed convergence using the potential scale reduction factor (R^), which approaches 1.0 with convergence, and for which a value of ≤1.2 is proposed as acceptable (Gelman et al. 2013). As mentioned above, PFOA was run first, because it has the most individual data. The posterior distribution for the population variance of the elimination rate for PFOA was used as the prior distribution for this parameter for the other PFAS (i.e., replacing the inverse gamma distribution prior used for PFOA). We fit the PFOA posterior for this variance parameter to a lognormal distribution. R codes for all modeling and analyses are included in supplementary materials and available at https://github.com/wachiuphd/2022-Bayes-PFAS-PK. Sensitivity Analysis We performed multiple local sensitivity analyses with respect to our modeling assumptions and approach to validate the parameter estimates produced by the model. These included a) changing the relative source contribution from drinking water in background serum concentration from 20% to either 0% or 80%, b) shifting the prior distributions for Vd to 20% above or below the primary estimate, and c) evaluating the effect of using alternative training vs. testing data sets. Specifically, we analyzed the effect of using only individual-level serum concentrations, using only population-level aggregates, and switching the training and testing data sets from the primary analysis. Finally, we performed an analysis focused only on parameter estimation, where all data were used for training. In all cases, we reran the entire model calibration and evaluation process using the alternative parameter values, prior distributions, or data sets. Results Numeric outputs and data that are not under a data sharing agreement are available in the Supplemental Information. Individual-level data for Decatur, Alabama, cannot be published in detailed form. Serum and Water Data Comprehensive details on the identified populations and studies are provided in Table 1 and in Table S1. Briefly, for the Decatur population (n=37 after excluding eight individuals owing to occupational history), the Agency for Toxic Substances and Disease Registry (ATSDR) supplied the individual serum data. For the Arnsberg population, the study by Hölzer (2008) provided individual serum PFOA levels graphically for 2006 and 2007 (n=151). We derived and estimated those values from the line graph provided in the study to use them as individual serum data inputs in the model at two time periods. For the Minnesota population, the study by Johnson et al. (2017) provided individual-level serum data for PFOS and PFOA correlated with individual drinking water exposure (n=98). These were also derived from digitizing figures with WebPlotDigitizer (version 4.2) because tabular data were not provided. We obtained all other data from text or tables from study publications. Individual water consumption information was not available in any of the identified studies. A number of identified populations were excluded because the publications were missing critical information. We excluded the North Wales, Pennsylvania, cohort because the water data from before the serum measurements were not available and because some of the water was purchased from a neighboring town (North Wales Water Authority 2018). We excluded the Ronneby cohort because serum concentrations were not reported at individual time points (Li et al. 2018). The Uppsala County cohort was excluded because it contained only a cumulative estimate of months of exposure for children. For the Arnsberg cohort, we included only data for adults, given that it was the only study that separated children’s serum levels from adults’. The study populations we used for modeling, along with summary statistics of the serum and water concentrations for each PFAS, are summarized in Table 1. In some cases, there was only a single water or serum level for an individual or a population. In other cases, there were multiple water levels preceding the serum level, or ideally, more than one serum level with corresponding water concentrations over time at the individual level. Model Convergence and Fit Parameter estimates in all four models achieved excellent convergence (R^<1.05) with a reasonable number of MCMC iterations (4 chains; 20,000 iterations per chain). The resulting overall model fits comparing posterior median predictions and data are shown in Figure 2. For training data (Figure 2, left panels), the model was able to match the data very tightly. The GSDerr for individual time-course data was ∼1.1 for all four PFAS, indicating that the model fit had a residual error (difference between predictions and training data) of only ∼10% when time-course data were available. In cases where data are only available at a single time point per individual, such as for the Minnesota data for PFOA and PFOS, the GSDerr was ∼1.5, indicating ∼50% residual error. A larger error such as this one is expected because of the need to approximate steady state in these cases. Moreover, the Minnesota data appeared to tend toward overprediction, which is consistent with the use of a steady-state approximation (i.e., in reality, steady state would not have been reached, so the actual concentration would tend to be lower). Summary time-course data, available only for PFOA, had a GSDerr of ∼1.2, indicating ∼20% error. For summary data at a single time point, however, the residual errors were quite a bit larger: ∼2-fold for PFOA and PFOS, 2.3-fold for PFNA, and 2.8-fold for PFHxS. Remarkably, however, the residual error in all cases was <3-fold. These results show the importance of individual data for accurately estimating PFAS elimination. Figure 2. Overall evaluation of model fit. Comparison of data and median posterior predictions for (A,B) perfluorooctanoic acid (PFOA), (C,D) perfluorooctane sulfonic acid (PFOS), (E,F) perfluorononanoic acid (PFNA), and (G,H) perfluorohexane sulfonate (PFHxS) for both (A,C,E,G) training data and (B,D,F,H) testing data. The solid line represents equality, and the dashed line represents a 3-fold error. In each panel, R2 and root mean square error (RMSE) are also shown in log10 units. The underlying numeric values can be found in Table S5. Figures 2A to 2H are eight line graphs titled perfluorooctanoic acid train, perfluorooctanoic acid test, perfluorooctane sulfonic acid train, perfluorooctane sulfonic acid test, perfluorononanoic acid train, perfluorononanoic acid test, perfluorohexane sulfonate train, and perfluorohexane sulfonate test, plotting Predicted serum concentration (micrograms per liter), ranging from 0.1 to 1.0 in increments of 0.9, 1.0 to 10.0 in increments of 9.0, 10.0 to 100.0 in increments of 90.0, and 100.0 to 1000.0 in increments of 900.0 (y-axis) across Measured serum concentration (micrograms per liter), ranging from 0.1 to 1.0 in increments of 0.9, 1.0 to 10.0 in increments of 9.0, 10.0 to 100.0 in increments of 90.0, and 100.0 to 1000.0 in increments of 900.0 (x-axis) for city (datatype), including Decatur (individual), Arnsberg (individual), Minnesota (individual), Lubeck Bartell, Little Hocking Bartell, Little Hocking Emmett, Paulsboro, Horsham, Warminster, and Warrington, respectively. The comparisons with the testing data (Figure 2, right panels) showed a similar trend. For individual data, the model performed best when we used data sets containing individual time-course data, with residual errors well within 50%. Individual data with a single time point and summary steady-state data performed worse, although residual errors were generally within 3-fold. Figure 3 shows the posterior distributions of predictions for the Decatur time-course data, which were available for all four PFAS. For both training and testing data sets, the data were well within the CIs of the posterior predictions. Results were similarly accurate for the individual data sets across all PFAS (Figures S1–S4). However, for summary data, the predictions were less accurate for both the training and testing data sets. Figure 3. Data and posterior distribution of predictions for Decatur, Alabama, data. Comparison and Decatur data (symbols) and distribution of posterior predictions (box plots) for (A,B) perfluorooctanoic acid (PFOA), (C,D) perfluorooctane sulfonic acid (PFOS), (E,F) perfluorononanoic acid (PFNA), and (G,H) perfluorohexane sulfonate (PFHxS) for both (A,C,E,G) training data and (B,D,F,H) testing data. Shading indicates time (T), in years, since the first serum sample. Samples from the same individuals are paired such that the two left-most bars indicate samples from one individual taken at two time points (the initial time point is T=0, and the second is T=5.802 y later). The underlying numeric values can be found in Table S6. Figures 3A, 3C, 3E, and 3G are dot graphs titled perfluorooctanoic acid Decatur train, perfluorooctane sulfonic acid Decatur train, perfluorononanoic acid Decatur train, and perfluorohexane sulfonate Decatur train, plotting perfluorooctanoic acid serum (micrograms per liter), ranging from 1 to 10 in increments of 9 and 10 to 100 in increments of 90; perfluorooctane sulfonic acid serum (micrograms per liter), ranging from 1 to 10 in increments of 9 and 10 to 100 in increments of 90; perfluorononanoic acid serum (micrograms per liter), ranging from 0.1 to 1.0 in increments of 0.9 and 1.0 to 10.0 in increments of 9.0; and perfluorohexane sulfonate serum (micrograms per liter), ranging from 1 to 3 in increments of 2, 3 to 10 in increments of 7, and 10 to 30 in increments of 20 (y-axis) across posterior distribution of predictions for Decatur data (x-axis) for male and female. Figures 3B, 3D, 3F, and 3H are box plots titled perfluorooctanoic acid Decatur test, perfluorooctane sulfonic acid Decatur test, perfluorononanoic acid Decatur test, and perfluorohexane sulfonate Decatur test, plotting perfluorooctanoic acid serum (micrograms per liter), ranging from 1 to 10 in increments of 9 and 10 to 100 in increments of 90; perfluorooctane sulfonic acid serum (micrograms per liter), ranging from 1 to 10 in increments of 9 and 10 to 100 in increments of 90; perfluorononanoic acid serum (micrograms per liter), ranging from 0.1 to 1.0 in increments of 0.9 and 1.0 to 10.0 in increments of 9.0; and perfluorohexane sulfonate serum (micrograms per liter), ranging from 1 to 3 in increments of 2, 3 to 10 in increments of 7, and 10 to 30 in increments of 20 (y-axis) across posterior distribution of predictions for Decatur data (x-axis) for Decatur uppercase t equals 0 and Decatur uppercase t equals 5.802. Posterior Distributions Posterior distributions for the main toxicokinetic parameters of T½ and Vd, as well as the derived parameter of clearance, are shown in Table 3. These reflect updating of the prior distributions after consideration of the likelihood of the data. In addition to estimates for the population GM and population GSD, the model makes a prediction for a “random individual,” which combines uncertainty and variability. The random individual is relevant to the general public because for any individual, there are two sources of uncertainty: a) the uncertainty in the population distribution (GM and SD), and b) uncertainty in where one is located on the population variability distribution (e.g., does an individual have a higher or lower than typical T½ for a PFAS?). Table 3 Summary of posterior distributions identified through Bayesian parameter calibration. Parameter [prior population GM median (95% CI)] Population GM median (95% CI) Population GSD median (95% CI) Random individual median (95% CI) [98% CI]a PFOA Half-life (y) [4.6 (2.1, 10.2)] 3.14 (2.69, 3.73) 1.57 (1.42, 1.73) 3.13 (1.13, 7.83) [0.90, 9.14] Volume of distribution (L/kg) [0.17 (0.10, 0.28)] 0.43 (0.32, 0.59) 1.12 (1.01, 1.47) 0.43 (0.27, 0.74) [0.23, 0.87] Clearance (L/kg per year) [0.037 (0.014, 0.095)] 0.095 (0.074, 0.126) 1.62 (1.45, 1.85) 0.097 (0.0369, 0.262) [0.0327, 0.341] PFOS Half-life (y) [4.8 (2.2, 10.6)] 3.36 (2.52, 4.42) 1.57 (1.42, 1.76) 3.40 (1.28, 8.42) [1.20, 9.96] Volume of distribution (L/kg) [0.23 (0.14, 0.38)] 0.32 (0.22, 0.47) 1.10 (1.01, 1.38) 0.32 (0.19, 0.51) [0.15, 0.56] Clearance (L/kg per year) [0.048 (0.019, 0.123)] 0.066 (0.048, 0.092) 1.60 (1.45, 1.83) 0.066 (0.0245, 0.176) [0.0203, 0.199] PFNA Half-life (y) [4.3 (1.9, 9.5)] 2.35 (1.65, 3.16) 1.53 (1.40, 1.70) 2.27 (0.83, 5.36) [0.76, 5.94] Volume of distribution (L/kg) [0.17 (0.10, 0.28)] 0.19 (0.11, 0.30) 1.12 (1.01, 1.51) 0.18 (0.10, 0.32) [0.09, 0.40] Clearance (L/kg per year) [0.040 (0.015, 0.102)] 0.056 (0.033, 0.093) 1.57 (1.42, 1.86) 0.056 (0.019, 0.163) [0.0165, 0.199] PFHxS Half-life (y) [5.3 (2.4, 11.7)] 8.30 (5.38, 13.5) 1.57 (1.42, 1.77) 8.12 (2.96, 21.6) [2.19, 24.8] Volume of distribution (L/kg) [0.25 (0.15, 0.42)] 0.29 (0.17, 0.45) 1.11 (1.00, 1.45) 0.28 (0.16, 0.46) [0.14, 0.56] Clearance (L/kg per year) [0.047 (0.018, 0.122)] 0.025 (0.012, 0.039) 1.61 (1.45, 1.86) 0.022 (0.0075, 0.078) [0.0065, 0.10] Note: Model parameters are shown by chemical species with GMs, GSDs, and CIs. CI, confidence interval; GM, geometric mean; GSD, geometric standard deviation; PFHxS, perfluorohexane sulfonate; PFOA, perfluorooctanoic acid; PFOS, perfluorooctane acid; PFNA, perfluorononanoic acid. a 98% CI shows the variation from the 1st percentile random individual to the 99th percentile random individual. Comparisons of prior and posterior distributions for the T½ are shown in Figure 4. For all four PFAS, the posteriors for the population GMs were noticeably shifted and narrower than the prior distributions. This indicates that the data were informative relative to the prior. For PFOA, the posterior for the population GSD was also shifted and indicated more population variation in the T½ than under prior assumptions. However, for the other PFAS, the priors and posteriors for the population variation were similar because there were insufficient numbers of individuals to inform this parameter; in other words, the data were not informative relative to the prior. Nonetheless, this indicates that the data are consistent with the amount of population variation in T½ observed with PFOA. T½ estimates (95% CIs) for the population GM were 3.14 (2.69, 3.73) y for PFOA, 3.36 (2.52, 4.42) y for PFOS, 2.35 (1.65, 3.16) y for PFNA, and 8.30 (5.38, 13.5) y for PFHxS. For PFHxS, there was a noticeable shift from the prior to the posterior, with the central estimate moving from 5.29 to 8.30 y. Although the uncertainty ranges in Figure 4 look similar on the natural scale, the 95% CI range on the log-scale shrank substantially, from 4.9- to 2.5-fold. Figure 4. Prior and posterior distributions for half-life (T½). Comparison of priors (cyan dotted lines) and posteriors (orange lines) for (A,C,E,G) T½ population geometric mean (GM) and (B,D,F,H) population geometric standard deviation (GSD) for (A,B) perfluorooctanoic acid (PFOA), (C,D) perfluorooctane sulfonic acid (PFOS), (E,F) perfluorononanoic acid (PFNA), and (G,H) perfluorohexane sulfonate (PFHxS). Text includes posterior median and confidence intervals (CI). The underlying numeric values are presented in Tables 2 and 3 and in the text. Figures 4A to 4H are line graphs titled perfluorooctanoic acid uppercase t begin subscript 1 by 2 end subscript population geometric mean, Posterior median (95 percent confidence intervals); perfluorooctanoic acid uppercase t begin subscript 1 by 2 end subscript population geometric standard deviation, Posterior median (95 percent confidence intervals); perfluorooctane sulfonic acid uppercase t begin subscript 1 by 2 end subscript population geometric mean, Posterior median (95 percent confidence intervals); perfluorooctane sulfonic acid uppercase t begin subscript 1 by 2 end subscript population geometric standard deviation, Posterior median (95 percent confidence intervals); perfluorononanoic acid uppercase t begin subscript 1 by 2 end subscript population geometric mean, Posterior median (95 percent confidence intervals); perfluorononanoic acid uppercase t begin subscript 1 by 2 end subscript population geometric standard deviation, Posterior median (95 percent confidence intervals); perfluorohexane sulfonate uppercase t begin subscript 1 by 2 end subscript population geometric mean, Posterior median (95 percent confidence intervals); and perfluorohexane sulfonate uppercase t begin subscript 1 by 2 end subscript population geometric standard deviation, Posterior median (95 percent confidence intervals), plotting density, ranging from 0.0 to 1.5 in increments of 0.5; 0 to 6 in increments of 2; 0.0 to 0.8 in increments of 0.2; 0 to 5 in unit increments; 0.00 to 0.75 in increments of 0.25; 0 to 4 in increments of 2; 0.00 to 0.20 in increments of 0.05; and 0 to 4 in unit increments, respectively, (y-axis) across population geometric mean uppercase t begin subscript 1 by 2 end subscript years, ranging from 0 to 15 in increments of 5 (for Figures 4A,C,E,G) and population geometric standard deviation uppercase t begin subscript 1 by 2 end subscript, ranging from 1.0 to 3.0 in increments of 0.5 (for Figures 4B,D,F,H) (x-axis) for prior and posterior, respectively. Comparisons of prior and posterior distributions for the Vd are shown in Figure 5. Vd posterior estimates ranged from 0.19 to 0.428L/kg across the four PFAS. In general, the posterior estimates for the Vd across the four PFAS are shifted to larger values as compared with the prior distributions, indicating that the data were informative for the population mean of Vd. A larger Vd indicates that for a given intake, the ratio between DWCs and serum concentrations would tend to be larger as compared with previous studies; this result is likely due to accounting for background (nondrinking water sources). There are insufficient data, however, to substantially inform the population variability in the Vd given that the prior and posterior distributions are similar. Figure 5. Prior and posterior distributions for the volume of distribution. Comparison of priors based on literature (cyan dotted lines) and posteriors (orange lines) for the volume of distribution (A,C,E,G) population geometric mean (GM) and (B,D,F,H) population geometric standard deviation (GSD) for (A,B) perfluorooctanoic acid (PFOA), (C,D) perfluorooctane sulfonic acid (PFOS), (E,F) perfluorononanoic acid (PFNA), and (G,H) perfluorohexane sulfonate (PFHxS). Text includes posterior median and confidence intervals (CIs). The underlying numeric values are presented in Tables 2 and 3 and in the text. Figures 5A to 5H are line graphs titled perfluorooctanoic acid volume of distribution population geometric mean, Posterior median (95 percent confidence intervals); perfluorooctanoic acid volume of distribution population geometric standard deviation, Posterior median (95 percent confidence intervals); perfluorooctane sulfonic acid volume of distribution population geometric mean, Posterior median (95 percent confidence intervals); perfluorooctane sulfonic acid volume of distribution population geometric standard deviation, Posterior median (95 percent confidence intervals); perfluorononanoic acid volume of distribution population geometric mean, Posterior median (95 percent confidence intervals); perfluorononanoic acid volume of distribution population geometric standard deviation, Posterior median (95 percent confidence intervals); perfluorohexane sulfonate volume of distribution population geometric mean, Posterior median (95 percent confidence intervals); and perfluorohexane sulfonate volume of distribution population geometric standard deviation, Posterior median (95 percent confidence intervals), plotting density, ranging from 0.0 to 7.5 in increments of 2.5; 0 to 4 in unit increments; 0 to 6 in increments of 2; 0 to 4 in unit increments; 0.0 to 7.5 in increments of 0.5; 0 to 4 in unit increments; 0 to 6 in increments of 2; and 0 to 4 in unit increments, respectively, (y-axis) across population geometric mean volume of distribution (liters per kilogram), ranging from 0.00 to 1.00 in increments of 0.25 (for Figures 5A,C,E,G) and population geometric standard deviation volume of distribution, ranging from 1.0 to 3.0 in increments of 0.5 (for Figures 4B,D,F,H) (x-axis) for prior and posterior, respectively. With respect to clearance, which is the product of the rate coefficient k and the Vd, posterior medians ranged from 0.025 to 0.095L/kg per year, with PFHxS< PFNA< PFOS< PFOA. These values tended to be on the higher end of the prior distributions for PFOA, PFOS, and PFNA, and toward the lower end for PFHxS. We also found that posterior distributions for the components of clearance, T½, and Vd were uncorrelated (all R2<0.05). Posteriors for the population variation were slightly larger than those for T½, reflecting the relatively smaller contribution of Vd to interindividual variation. To check for the possibility of systematic biases, we examined the posterior distributions for individual k and Vd parameters and compared them across study cohorts. For instance, if individual posteriors for one location were systematically different from individual posteriors for another location, then that would suggest errors in the model or parameters. As shown in Figure S5, there is no discernable difference across cohorts from different cities/locations, and all individual parameter posterior samples had z-scores that were statistically consistent with the expected standard normal distribution. Sensitivity Analysis The posterior estimates for parameters were generally stable across all sensitivity analyses, and in all cases, there was substantial overlap between the 95% CI from each sensitivity analysis and that from the primary analysis (Table S4, Figures S6–S9). Changing relative source contribution from drinking water and nondrinking water sources had little effect on T½ or Vd estimates. Swapping the data sets used for testing with those used for training produced nearly identical results to the primary analysis. Increasing and decreasing the priors for Vd resulted in nearly identical posterior estimates for all parameters. The largest differences from the primary analysis occurred when removing either the individual- or the population-level data for training. The posteriors for the GSD for variability in T½ were much lower when we used only population summary data, as would be expected because it is more challenging to estimate variability when only summary data are available. Discussion To our knowledge, ours is the largest analysis to date of individual serum data of communities with known and measured PFAS drinking water contamination. We incorporated data from 13 studies performed across widespread geographic locations. We have integrated these multiple data sets in a Bayesian toxicokinetic analysis to estimate the T½ and Vd, as well as the population variability, for four common PFAS. We have also incorporated NHANES estimates of background exposures over time, without which kinetic parameter estimates may be biased. Our model accurately predicts serum data from a large number of individuals across multiple studies, including data not used for calibration. Furthermore, the posterior estimates are insensitive to a variety of changes to the prior inputs and to the design of testing and training data sets, suggesting these estimates are stable given the current data. Our results for the population GM of T½ are in the range of several previous studies that we did not use in our analysis. For instance, Olsen et al. (2007) estimated T½ (95% CIs) of occupationally exposed retirees to be 3.8 (3.1, 4.4) y for PFOA, 5.4 (3.9, 6.9) y for PFOS, and 8.5 (6.4, 10.6) y for PFHxS. The values for PFOA and PFOS are somewhat longer than our estimates, but Olsen et al. (2007) did not account for continued exposure due to drinking water contamination, which was later found to be substantial in the community. Li et al. (2018) reported results from an analysis of residents in Sweden exposed via drinking water contamination and reported shorter T½ (95% CIs) of 2.7 (2.5, 2.9) y for PFOA, 3.4 (3.1, 3.7) y for PFOS, and 5.3 (4.6, 6.0) y for PFHxS, which are concordant with our estimates for PFOA and PFOS but somewhat shorter than our estimates for PFHxS. Overall, our analysis supports the higher estimate for PFHxS T½, though recognizing potential for substantial population variation. Our results for the population GSD are also within the range of population variation reported in previous studies that we did not use in our analysis. Population variation in individual T½ from Olsen et al. (2007) was estimated to be ∼1.49 for PFOA, 1.66 for PFOS, and 1.78 for PFHxS. The values for PFOA and PFOS are similar to those found in our analysis, with the value for PFHxS somewhat higher. Variation reported by Li et al. (2018) was somewhat less, with a GSD of ∼1.4 across all PFAS, at the low end of the CI from our analysis. Our estimates for the Vd of PFOA and PFOS are somewhat larger than values reported in the literature and used as priors in the analysis (Figure 5). For instance, Thompson et al. (2010) found 0.17 L/kg for PFOA; we found 0.43 L/kg (95% CI: 0.32, 0.59). Thompson et al. (2010) found 0.23 L/kg for PFOS, and our prior estimate was slightly larger at 0.32 L/kg (95% CI: 0.22, 0.47). Our findings for PFNA [0.19L/kg (95% CI: 0.11, 0.30)] and PFHxS [0.29L/kg (95% CI: 0.17, 0.45)] were similar to priors (Figure 5). Importantly, none of the previous values used to develop prior estimates for Vd (see the “Methods” section) were based on statistical calibration using multiple data sources. Previous values were estimated based on adjusted values from animal studies (Sundström et al. 2012) or other PFAS (Koponen et al. 2018) or were based on assumed serum and water data from some of the same cohorts as used here (e.g., Lubeck, West Virginia; Little Hocking, Ohio) but with fixed values for other parameters, such as drinking water rates (Thompson et al. 2010). In addition, we made a greater effort to adjust for background exposures compared with previous studies; underestimating background exposure can lead to underprediction of Vd. For a given T½, DWC, and observed serum concentration, underestimating background will require a smaller Vd to fit the observation. Overall, our Bayesian approach provides a rigorous basis for Vd estimates because it integrates multiple individual-level data sets, includes extensive prior information on background exposures, and incorporates population variability. In addition, we have derived posterior estimates for both the population GM and population variation in clearance, which is the key parameter for chemical-specific values for interspecies extrapolation and interindividual variability, as well as for in vitro to in vivo extrapolation (IVIVE). For instance, as discussed in WHO IPCS (2005) and U.S. EPA (2014) guidance documents, the ratio of clearances can be used to replace default uncertainty factors for interspecies and interindividual toxicokinetic differences. For interspecies extrapolation, these results could be combined with those of Wambaugh et al. (2013) for extrapolating experimental animal points of departure to human equivalent doses. In addition, for interindividual variability, the ratio of the median to 1% random individual from Table 3 could replace the default value of UFH,TK=3.16 for toxicity end points in (nonpregnant) adults. Interestingly, the ratios of the median to the 1% random individual fall within a narrow range—3.0 for PFOA, 3.3 for PFOS, and 3.4 for PFNA—and near to the default value for PFHxS. Finally, for IVIVE, clearance estimates can be used to convert from in vitro test concentrations to oral equivalent doses so as to put high-throughput screening results in the context of human exposure (Wetmore et al. 2015). One limitation of our analysis is that our estimates are not specific to age, sex, geographic location, or race/ethnicity, nor do we have geography-specific estimates of background exposures. We did not identify sufficient data to adequately stratify our analysis among these different groups, so we cannot assess the potential for such differences to affect our results. In Arnsberg, the women were all mothers, and we did not have information on their ages. In Decatur, Alabama, the mean age in 2010 was 52 y, and in 2016 the mean age was 63 y, with very few younger women. For the Minnesota data set, we did not have age or sex information for the individuals. Studies have shown that PFAS serum levels can vary by age, sex, race/ethnicity, and geographic location (Park et al. 2019). In addition, Park et al. (2019) found that parity and menstrual bleeding were important predictors of PFAS levels. This is consistent with other studies in both animals and humans, which indicate that serum levels in menstruating women, and in women who have breastfed or given birth, may be lower than in older women and men (Brantsæter et al. 2013; Huang et al. 2019; Singer et al. 2018). Further, Zhang et al. (2013) found shorter T½ in younger women compared with older women and men, hypothesizing that menstrual clearance is important for PFOS, PFNA and PFHxS and less pronounced for PFOA. Future work as more individual serum-level data become available could better distinguish these differences. This work has a number of additional limitations. First, it excludes some studies for which individual serum data could not be readily obtained, or for which corresponding water concentration information was not available (Table S1). This limitation is particularly important for PFNA and PFHxS, for which individual data were available only for the Decatur cohort, where PFAS drinking water levels were below the minimum reporting level. This limitation is less important for PFOA and PFOS, for which more than one study with many individuals is included in the analysis. A corollary limitation is that because only three studies had individual data for PFOA, two studies for PFOS, and one study for PFNA and PFHxS, we could not use separate studies for training and testing, but instead split individuals within each data set. Thus, the training and testing data sets were not completely independent. A sensitivity analysis switching testing and training data sets gave similar results (Figures S6–S9), which could indicate either an issue with independence or robust estimates. A further comparison of individual-level Vd and k posteriors for independent study cohorts of PFOA- and PFOS-containing individual-level data showed no systematic differences across cohorts, as expected for physiological parameters (Figure S5). Thus, the nonindependence of testing and training data sets is not a critical limitation for use of these results. Our main objective here was to estimate model parameters. Second, for many studies, particularly those with summary data, we had to make a steady-state assumption of constant DWCs in the past based on few measurements, or assume negligible concentrations after drinking water interventions due to missing postintervention measurements. This limitation is again more acute for PFNA and PFHxS but is also an issue for PFOS, for which the Decatur cohort was the only data set with time-course data. Thus, we have greatest overall confidence in the PFOA results, moderate confidence in the PFOS results, and greater uncertainty in the PFNA and PFHxS results. Obtaining additional individual-level time-course data, including postintervention concentrations, would be the most effective way to address these limitations. This analysis also did not consider the potential contribution of exposure to precursors or degradation products of the PFAS. Finally, although a one-compartment model has the virtue of simplicity and ease of implementation, the results are ultimately empirical. They do not provide mechanistic insights, nor can they be used to characterize tissue-specific internal dose. Furthermore, they cannot directly account for saturable reabsorption mechanisms; however, unlike exposures in experimental animal studies, human drinking water exposure levels appear to be well below saturation, so reabsorption can be lumped into an overall first-order clearance process. In particular, the concentrations in serum in the studies analyzed here were well below any saturation, and hence an explicit model for this mechanism is not necessary. For example, the KM values for organic anion transporters in the kidney range from 20 to 78mg/L for PFOA (Worley et al. 2017b; Nakagawa et al. 2008). This contrasts with values from experimental animal studies, which are performed at higher exposures and have lower degrees of renal reabsorption, where such a mechanism is needed (e.g., Andersen et al. 2006). In any case, given the need to incorporate individual-level data and complex exposure patterns, this model is a useful first step in combining data from many sources while also estimating interindividual variability. Moreover, given the relatively high data needs for even this one-compartment analysis—extensive individual serum data along with local DWCs—it will be challenging to validate more complex toxicokinetic models for PFAS more generally. These results have a number of important public health implications. The degree of variability in T½ across individuals means that for the same external exposure, some individuals will experience much greater internal exposure than others and, therefore, have a higher effective dose (Table 3). For instance, the 95% CI of the T½ for PFOA of a random individual drawn from the population spans from 1.13 to 7.83 y, with the 98% CI spanning a 10-fold range from 0.90 to 9.14 y. Across the four PFAS, the range from the 1st to the 99th percentile of T½ for a random individual span between 8- and 11-fold (Table 3). In addition, the higher estimate for the Vd as compared with previous studies implies that, for a given serum concentration, the body burden is greater. Although biomonitoring data on selected PFAS suggest that levels are declining across the population overall, drinking water contamination continues to be an issue in many parts of the country. Better estimates of toxicokinetic parameters of PFAS, as well as their variation in the population, will be essential in better characterizing the potential public health effects of PFAS, and the methods we applied here can be readily applied to other PFAS where individual time-course data are available for both serum and water concentrations. Ultimately, these estimates will be essential in supporting risk assessments and risk management decisions aimed at reducing current and future exposures to this ubiquitous and persistent class of compounds. Supplementary Material Click here for additional data file. Click here for additional data file. Acknowledgments Abt Associates authors (M.T.L., C.L., A.A., D.M., and S.S.) and W.A.C. were supported under contracts PO0400247 and GS00F045DA (both to M.T.L.) from the Agency for Toxic Substances and Disease Registry (ATSDR) under a subcontract to Guidehouse LLC. This work was also supported, in part, by grants from the National Institutes of Health/National Institute of Environmental Health Sciences (P42 ES027704, P30 ES029067, both to W.A.C.). Preparation of this paper was also supported by Abt Associates internal funds. The paper was improved by presentation at an Abt Work in Progress Seminar and comments received as part of that process. P. Do and L. Katz at Abt Associates provided valuable support on the literature review, while D. Ferguson and R. Balachandran provided technical editing. C. Welsh provided helpful review comments as the program lead for the Computational Toxicology and Methods Development Laboratory at the ATSDR. M. Shoemaker, F. Sieling, S. Lane, and L. Pogorelov at Guidehouse LLC, as well as D. Hunt, M. Lorie, and E. Chen of Abt Associates, provided project management support. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention/the Agency for Toxic Substances and Disease Registry. ==== Refs References Andersen ME, Clewell HJ III, Tan YM, Butenhoff JL, Olsen GW. 2006. Pharmacokinetic modeling of saturable, renal resorption of perfluoroalkylacids in monkeys—probing the determinants of long plasma half-lives. Toxicology 227 (1–2 ):156–164, PMID: , 10.1016/j.tox.2006.08.004.16978759 ATSDR (Agency for Toxic Substances and Disease Registry). 2013. Health Consultation, Exposure Investigation Report, Perfluorochemical Serum Sampling in the Vicinity of Decatur, Alabama: Morgan, Lawrence, and Limestone Counties. Atlanta, GA: U.S. Department of Health and Human Services. http://www.atsdr.cdc.gov/HAC/pha/Decatur/Perfluorochemical_Serum%20Sampling.pdf [accessed 9 October 2019]. ATSDR. 2016. Health Consultation, Exposure Investigation, Biological Sampling of Per- and Polyfluoroalkyl Substances (PFAS1) in the Vicinity of Lawrence, Morgan, and Limestone Counties, Alabama. Atlanta, GA: U.S. Department of Health and Human Services. https://www.atsdr.cdc.gov/HAC/pha/BiologicalSampling/Biological_Sampling_of_Substances_in_Alabama_EI%20-Report_11-28-2016_508.pdf [accessed 8 July 2019]. ATSDR. 2021. Toxicological profile for perfluoroalkyls. Last updated March 2020. https://www.atsdr.cdc.gov/ToxProfiles/tp200-p.pdf [accessed 10 April 2020]. Bartell SM, Calafat AM, Lyu C, Kato K, Ryan PB, Steenland K. 2010. Rate of decline in serum PFOA concentrations after granular activated carbon filtration at two public water systems in Ohio and West Virginia. Environ Health Perspect 118 (2 ):222–228, PMID: , 10.1289/ehp.0901252.20123620 Brantsæter AL, Whitworth KW, Ydersbond TA, Haug LS, Haugen M, Knutsen HK, et al. 2013. Determinants of plasma concentrations of perfluoroalkyl substances in pregnant Norwegian women. Environ Int 54 :74–84, PMID: , 10.1016/j.envint.2012.12.014.23419425 CA Water Boards (California State Water Resources Control Board). 2022. Perfluorooctanoic acid (PFOA) and Perfluorooctanesulfonic acid (PFOS). https://www.waterboards.ca.gov/drinking_water/certlic/drinkingwater/PFOA_PFOS.html [accessed 22 January 2022]. CDC (Centers for Disease Control and Prevention). 2019. Fourth National Report on Human Exposure to Environmental Chemicals. Atlanta, GA: U.S. Department of Health and Human Services, CDC. Cordner A, De La Rosa VY, Schaider LA, Rudel RA, Richter L, Brown P. 2019. Guideline levels for PFOA and PFOS in drinking water: the role of scientific uncertainty, risk assessment decisions, and social factors. J Expo Sci Environ Epidemiol 29 (2 ):157–171, PMID: , 10.1038/s41370-018-0099-9.30622333 Costa G, Sartori S, Consonni D. 2009. Thirty years of medical surveillance in perfluooctanoic acid production workers. J Occup Environ Med 51 (3 ):364–372, PMID: , 10.1097/JOM.0b013e3181965d80.19225424 Daly ER, Chan BP, Talbot EA, Nassif J, Bean C, Cavallo SJ, et al. 2018. Per- and polyfluoroalkyl substance (PFAS) exposure assessment in a community exposed to contaminated drinking water, New Hampshire, 2015. Int J Hyg Environ Health 221 (3 ):569–577, PMID: , 10.1016/j.ijheh.2018.02.007.29514764 Dunson DB. 2001. Commentary: practical advantages of Bayesian analysis of epidemiologic data. Am J Epidemiol 153 (12 ):1222–1226, PMID: , 10.1093/aje/153.12.1222.11415958 Egeghy PP, Lorber M. 2011. An assessment of the exposure of Americans to perfluorooctane sulfonate: a comparison of estimated intake with values inferred from NHANES data. J Expo Sci Environ Epidemiol 21 (2 ):150–168, PMID: , 10.1038/jes.2009.73.20145679 EGLE (Michigan Department of Environment, Great Lakes, and Energy). 2020. New State Drinking Water Standards Pave Way for Expansion of Michigan’s PFAS Clean-Up Efforts. https://content.govdelivery.com/accounts/MIDEQ/bulletins/2988e74 [accessed 19 January 2022]. Emmett EA, Shofer FS, Zhang H, Freeman D, Desai C, Shaw LM. 2006. Community exposure to perfluorooctanoate: relationships between serum concentrations and exposure sources. J Occup Environ Med 48 (8 ):759–770, PMID: , 10.1097/01.jom.0000232486.07658.74.16902368 Evans S, Andrews D, Stoiber T, Naidenko O. 2020. PFAS Contamination of Drinking Water Far More Prevalent Than Previously Reported. https://www.ewg.org/research/national-pfas-testing/ [accessed 10 March 2020]. EWG (Environmental Working Group). 2019a. EWG’s Tap Water Database 2019 Update. https://www.ewg.org/tapwater/ [accessed 4 December 2019]. EWG. 2019b. Perfluorohexane Sulfonate (PFHxS): Horsham Water and Sewer Authority. https://www.ewg.org/tapwater/system-contaminant.php?pws=PA1460033&contamcode=E204 [accessed 4 December 2019]. FoodRisk. 2020. Food Commodity Intake Database. What We Eat in America. https://fcid.foodrisk.org/percentiles [accessed 10 March 2020]. Gelman A. 2006. Prior distributions for variance parameters in hierarchical models (comment on article by Browne and Draper). Bayesian Anal 1 (3 ):515–534, 10.1214/06-BA117A. Gelman A, Carlin JB, Stern HS, Dunson DB, Vehtari A, Rubin DB. 2013. Bayesian Data Analysis. 3rd ed. Boca Raton, FL: Chapman and Hall/CRC Press. Graber JM, Alexander C, Laumbach RJ, Black K, Strickland PO, Georgopoulos PG, et al. 2019. Per- and polyfluoroalkyl substances (PFAS) blood levels after contamination of a community water supply and comparison with 2013–2014 NHANES. J Expo Sci Environ Epidemiol 29 (2 ):172–182, PMID: , 10.1038/s41370-018-0096-z.30482936 Harada K, Inoue K, Morikawa A, Yoshinaga T, Saito N, Koizumi A. 2005. Renal clearance of perfluorooctane sulfonate and perfluorooctanoate in humans and their species-specific excretion. Environ Res 99 (2 ):253–261, PMID: , 10.1016/j.envres.2004.12.003.16194675 Hölzer J, Göen T, Rauchfuss K, Kraft M, Angerer J, Kleeschulte P, et al. 2009. One-year follow-up of perfluorinated compounds in plasma of German residents from Arnsberg formerly exposed to PFOA-contaminated drinking water. Int J Hyg Environ Health 212 (5 ):499–504, PMID: , 10.1016/j.ijheh.2009.04.003.19464951 Hölzer J, Midasch O, Rauchfuss K, Kraft M, Reupert R, Angerer J, et al. 2008. Biomonitoring of perfluorinated compounds in children and adults exposed to perfluorooctanoate-contaminated drinking water. Environ Health Perspect 116 (5 ):651–657, PMID: , 10.1289/ehp.11064.18470314 Hu XC, Andrews DQ, Lindstrom AB, Bruton TA, Schaider LA, Grandjean P, et al. 2016. Detection of poly- and perfluoroalkyl substances (PFASs) in U.S. drinking water linked to industrial sites, military fire training areas, and wastewater treatment plants. Environ Sci Technol Lett 3 (10 ):344–350, PMID: , 10.1021/acs.estlett.6b00260.27752509 Huang MC, Dzierlenga AL, Robinson VG, Waidyanatha S, DeVito MJ, Eifrid MA, et al. 2019. Toxicokinetics of perfluorobutane sulfonate (PFBS), perfluorohexane-1-sulphonic acid (PFHxS), and perfluorooctane sulfonic acid (PFOS) in male and female Hsd:Sprague Dawley SD rats after intravenous and gavage administration. Toxicol Rep 6 :645–655, PMID: , 10.1016/j.toxrep.2019.06.016.31334035 HWSA (Horsham Water and Sewer Authority). 2014. 2014 Water Quality Report: Horsham Water and Sewer Authority. https://www.horshamwater-sewer.com/sites/default/files/ccr_reports/2014%20CCR.pdf [accessed 23 December 2019]. HWSA. 2018. 2018 Water Quality Report: Horsham Water and Sewer Authority. https://www.horshamwater-sewer.com/sites/default/files/ccr_reports/2018_ccr.pdf [accessed 25 November 2019]. IARC (International Agency for Research on Cancer). 2017. Some Chemicals Used as Solvents and in Polymer Manufacture. Lyon, France: IARC. Johnson J, Kari A, Husest C, Williams A. 2017. Community Exposure to PFCs in Washington County, Minnesota: the East Metro Perfluorochemicals Biomonitoring Pilot Project. https://www.cleanwateraction.org/files/publications/mn/pfc_biomonitoring_study_results.pdf [accessed 4 December 2019]. Koponen J, Winkens K, Airaksinen R, Berger U, Vestergren R, Cousins IT, et al. 2018. Longitudinal trends of per- and polyfluoroalkyl substances in children’s serum. Environ Int 121 (pt 1 ):591–599, PMID: , 10.1016/j.envint.2018.09.006.30308470 Li Y, Fletcher T, Mucs D, Scott K, Lindh CH, Tallving P, et al. 2018. Half-lives of PFOS, PFHxS and PFOA after end of exposure to contaminated drinking water. Occup Environ Med 75 (1 ):46–51, PMID: , 10.1136/oemed-2017-104651.29133598 Lorber M, Egeghy PP. 2011. Simple intake and pharmacokinetic modeling to characterize exposure of Americans to perfluoroctanoic acid, PFOA. Environ Sci Technol 45 (19 ):8006–8014, PMID: , 10.1021/es103718h.21517063 MDH (Minnesota Department of Health). 2021. Comparison of State Water Guidance and Federal Drinking Water Standards. https://www.health.state.mn.us/communities/environment/risk/guidance/waterguidance.html [accessed 19 January 2022]. Nakagawa H, Hirata T, Terada T, Jutabha P, Miura D, Harada KH, et al. 2008. Roles of organic anion transporters in the renal excretion of perfluorooctanoic acid. Basic Clin Pharmacol Toxicol 103 (1 ):1–8, PMID: , 10.1111/j.1742-7843.2007.00155.x.18373647 NJDEP (New Jersey Department of Environmental Protection). 2021. PFAS in Drinking Water. https://www.nj.gov/dep/watersupply/pfas/ [accessed 19 January 2022]. North Wales Water Authority. 2018. Water Quality Report 2018. https://www.nwwater.com/images/documents/ccrs/2018ccrwebversion.pdf [accessed 19 October 2019]. NTP (National Toxicology Program). 2016. NTP Monograph: Immunotoxicity Associated with Exposure to Perfluorooctanoic Acid or Perfluorooctabe Sulfonate. https://ntp.niehs.nih.gov/ntp/ohat/pfoa_pfos/pfoa_pfosmonograph_508.pdf [accessed 19 October 2019]. Nuzzo R. 2015. Chance: peace talks in the probability wars. New Scientist , Physics Features. https://www.newscientist.com/article/mg22530121-200-chance-peace-talks-in-the-probability-wars/ [accessed 20 October 2019]. Olsen GW, Burris JM, Ehresman DJ, Froehlich JW, Seacat AM, Butenhoff JL, et al. 2007. Half-life of serum elimination of perfluorooctanesulfonate, perfluorohexanesulfonate, and perfluorooctanoate in retired fluorochemical production workers. Environ Health Perspect 115 (9 ):1298–1305, PMID: , 10.1289/ehp.10009.17805419 Olsen GW, Mair DC, Lange CC, Harrington LM, Church TR, Goldberg CL, et al. 2017. Per- and polyfluoroalkyl substances (PFAS) in American Red Cross adult blood donors, 2000–2015. Environ Res 157 :87–95, PMID: , 10.1016/j.envres.2017.05.013.28528142 Park SK, Peng Q, Ding N, Mukherjee B, Harlow SD. 2019. Determinants of per- and polyfluoroalkyl substances (PFAS) in midlife women: evidence of racial/ethnic and geographic differences in PFAS exposure. Environ Res 175 :186–199, PMID: , 10.1016/j.envres.2019.05.028.31129528 Penn DOH (Pennsylvania Department of Health). 2019. PFAS Exposure Assessment Technical Toolkit (PEATT) Pilot Project. Final Report. Division of Environmental Health Epidemiology. https://www.health.pa.gov/topics/Documents/Environmental%20Health/PEATT%20Pilot%20Project%20Final%20Report%20April%2029%202019.pdf [accessed 5 June 2020]. Post GB, Louis JB, Lippincott RL, Procopio NA. 2013. Occurrence of perfluorinated compounds in raw water from New Jersey public drinking water systems. Environ Sci Technol 47 (23 ):13266–13275, PMID: , 10.1021/es402884x.24187954 Seals R, Bartell SM, Steenland K. 2011. Accumulation and clearance of perfluorooctanoic acid (PFOA) in current and former residents of an exposed community. Environ Health Perspect 119 (1 ):119–124, PMID: , 10.1289/ehp.1002346.20870569 Silver N. 2012. The Signal and the Noise: Why So Many Predictions Fail—But Some Don’t. New York, NY: Penguin Press. Singer AB, Whitworth KW, Haug LS, Sabaredzovic A, Impinen A, Papadopoulou E, et al. 2018. Menstrual cycle characteristics as determinants of plasma concentrations of perfluoroalkyl substances (PFASs) in the Norwegian Mother and Child Cohort (MoBa study). Environ Res 166 :78–85, PMID: , 10.1016/j.envres.2018.05.019.29879567 Sunderland EM, Hu XC, Dassuncao C, Tokranov AK, Wagner CC, Allen JG. 2019. A review of the pathways of human exposure to poly- and perfluoroalkyl substances (PFASs) and present understanding of health effects. J Expo Sci Environ Epidemiol 29 (2 ):131–147, PMID: , 10.1038/s41370-018-0094-1.30470793 Sundström M, Chang SH, Noker PE, Gorman GS, Hart JA, Ehresman DJ, et al. 2012. Comparative pharmacokinetics of perfluorohexanesulfonate (PFHxS) in rats, mice, and monkeys. Reprod Toxicol 33 (4 ):441–451, PMID: , 10.1016/j.reprotox.2011.07.004.21856411 Thompson J, Lorber M, Toms LML, Kato K, Calafat AM, Mueller JF. 2010. Use of simple pharmacokinetic modeling to characterize exposure of Australians to perfluorooctanoic acid and perfluorooctane sulfonic acid. Environ Int 36 (4 ):390–397, PMID: , 10.1016/j.envint.2010.02.008.20236705 U.S. EPA (U.S. Environmental Protection Agency). 2014. Guidance for Applying Quantitative Data to Develop Data-Derived Extrapolation Factors for Interspecies and Intraspecies Extrapolation. EPA/R-14/002F. Washington, DC: U.S. EPA, Risk Assessment Forum, Office of the Science Advisor. https://www.epa.gov/sites/production/files/2015-01/documents/ddef-final.pdf [accessed 18 November 2022]. U.S. EPA. 2016. Drinking Water Health Advisory for Perfluorooctane Sulfonate (PFOS). https://www.epa.gov/sites/production/files/2016-05/documents/pfos_health_advisory_final_508.pdf [accessed 28 March 2021]. U.S. EPA. 2017. Unregulated Contaminant Monitoring Rule 3 (UCMR 3), (2013–2015) Occurrence Data. https://catalog.data.gov/dataset/unregulated-contaminant-monitoring-rule-3-ucmr-3-2013-2015-occurrence-data [accessed 18 October 2019]. U.S. EPA. 2018. 2018 Edition of the Drinking Water Standards and Health Advisories Tables. EPA 822-F-18-001. https://www.epa.gov/system/files/documents/2022-01/dwtable2018.pdf [accessed 18 November 2022]. U.S. EPA. 2021. PFAS Strategic Roadmap: EPA’s Commitments to Action 2021–2024. EPA-100-K-21-002. https://www.epa.gov/system/files/documents/2021-10/pfas-roadmap_final-508.pdf [accessed 9 November 2021]. U.S. EPA. 2022. Technical Fact Sheet: Drinking Water Health Advisories for Four PFAS (PFOA, PFOS, GenX chemicals, and PFBS). EPA 822-F-22-00. https://www.epa.gov/system/files/documents/2022-06/technical-factsheet-four-PFAS.pdf [accessed 2 August 2022]. Wambaugh JF, Setzer RW, Pitruzzello AM, Liu J, Reif DM, Kleinstreuer NC, et al. 2013. Dosimetric anchoring of in vivo and in vitro studies for perfluorooctanoate and perfluorooctanesulfonate. Toxicol Sci 136 (2 ):308–327, PMID: , 10.1093/toxsci/kft204.24046276 Warrington Township. 2014. 2014 Warrington Township Water Quality Report. https://www.warringtontownship.org/waterqualityreports/ [accessed 4 December 2019]. Warrington Township. 2018. 2018 Warrington Township Water Quality Report. https://www.warringtontownship.org/waterqualityreports/ [accessed 22 November 2019]. Wetmore BA, Wambaugh JF, Allen B, Ferguson SS, Sochaski MA, Setzer RW, et al. 2015. Incorporating high-throughput exposure predictions with dosimetry-adjusted in vitro bioactivity to inform chemical toxicity testing. Toxicol Sci 148 (1 ):121–136, PMID: , 10.1093/toxsci/kfv171.26251325 WHO (World Health Organization), ILO (International Labour Organization), UNEP (U.N. Environment Programme). 2005. Chemical-Specific Adjustment Factors for Interspecies Differences and Human Variability: Guidance Document for Use of Data in Dose/Concentration–Response Assessment. IPCS Harmonization Project Document No. 2. Geneva, Switzerland: World Health Organization. https://apps.who.int/iris/rest/bitstreams/51378/retrieve [accessed 18 November 2022]. Wickham H, Averick M, Bryan J, Chang W, D’Agostino McGowan L, François R, et al. 2019. Welcome to the tidyverse. J Open Source Softw 4 (43 ):1686, 10.21105/joss.01686. WMA (Warminster Municipal Authority). 2018. 2018 Annual Drinking Water Quality Report. PWSID No. 1090069. https://www.warminsterauthority.com/sites/default/files/ccr_reports/2018ccr.pdf [accessed 4 December 2019]. Wong F, MacLeod M, Mueller JF, Cousins IT. 2014. Enhanced elimination of perfluorooctane sulfonic acid by menstruating women: evidence from population-based pharmacokinetic modeling. Environ Sci Technol 48 (15 ):8807–8814, PMID: , 10.1021/es500796y.24943117 Worley RR, Moore SM, Tierney BC, Ye X, Calafat AM, Campbell S, et al. 2017a. Per- and polyfluoroalkyl substances in human serum and urine samples from a residentially exposed community. Environ Int 106 :135–143, PMID: , 10.1016/j.envint.2017.06.007.28645013 Worley RR, Yang X, Fisher J. 2017b. Physiologically based pharmacokinetic modeling of human exposure to perfluorooctanoic acid suggests historical non drinking-water exposures are important for predicting current serum concentrations. Toxicol Appl Pharmacol 330 :9–21, PMID: , 10.1016/j.taap.2017.07.001.28684146 Yu CH, Riker CD, Lu SE, Fan ZT. 2020. Biomonitoring of emerging contaminants, perfluoroalkyl and polyfluoroalkyl substances (PFAS), in New Jersey adults in 2016–2018. Int J Hyg Environ Health 223 (1 ):34–44, PMID: , 10.1016/j.ijheh.2019.10.008.31679856 Zhang Y, Beesoon S, Zhu L, Martin JW. 2013. Biomonitoring of perfluoroalkyl acids in human urine and estimates of biological half-life. Environ Sci Technol 47 (18 ):10619–10627, PMID: , 10.1021/es401905e.23980546	36454223	PMC9714558	NO-CC CODE	2022-12-05 23:21:35	no	Environ Health Perspect. 2022 Dec 1; 130(12):127001	utf-8	Environ Health Perspect	2,022	10.1289/EHP10103	oa_other
==== Front Med Clin (Barc) Med Clin (Barc) Medicina Clinica 0025-7753 1578-8989 Elsevier España, S.L.U. S0025-7753(22)00561-9 10.1016/j.medcli.2022.10.008 Imagen Médica Neumonía lipoidea exógena secundaria al consumo crónico de aceite de parafina Exogenous lipoid pneumonia secondary to chronic paraffin oil consumptionGonzález Castro Sara a⁎ Arrieta Narváez Paola a Gorospe Sarasúa Luis b a Servicio de Neumología, Hospital Universitario Ramón y Cajal, Madrid, España b Servicio de Radiodiagnóstico, Hospital Universitario Ramón y Cajal, Madrid, España ⁎ Autora para correspondencia. 1 12 2022 1 12 2022 © 2022 Elsevier España, S.L.U. All rights reserved. 2022 Elsevier España, S.L.U. 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 pmcUna mujer de 36 años fumadora paucisintomática fue derivada al hospital por alteraciones radiológicas tras una infección aguda leve por SARS-CoV-2. En la tomografía computarizada (TC) de tórax se observaron extensas opacidades bilaterales de predominio basal en las que coexistían zonas de atenuación en vidrio deslustrado asociadas a engrosamiento del intersticio inter e intralobulillar (crazy paving) y consolidaciones peribronquiales con focos de atenuación negativa (fig. 1 A: imagen axial de TC de tórax [ventana pulmón] en la que se identifican varias consolidaciones peribronquiales [asteriscos]) en base pulmonar izquierda. Entre las consolidaciones se observa un patrón en empedrado [coexistencia de opacidades de atenuación en vidrio deslustrado e intersticio engrosado]; 1B: imagen axial de TC de tórax [ventana de mediastino] en la que se demuestra la atenuación negativa de la consolidación dominante del lóbulo inferior izquierdo [−38 unidades Hounsfield];. 1C: imagen axial de TC de tórax [ventana de pulmón] en la que se visualizan un patrón en empedrado bilateral). Dada la discordancia clínico-radiológica y los hallazgos en TC, se emitió el diagnóstico de probable neumonía lipoidea exógena, reconociendo la paciente el consumo crónico culinario de aceite hipocalórico (compuesto por aceite de parafina).Figura 1 La neumonía lipoidea exógena es una entidad poco frecuente definida por la presencia de lípidos alveolares y causada por la aspiración de agentes oleosos. El diagnóstico diferencial radiológico incluye el síndrome de distrés respiratorio agudo, la proteinosis alveolar y el cáncer de pulmón. Aunque clásicamente se ha descrito que el diagnóstico de certeza requiere de la demostración de macrófagos vacuolados cargados de lípidos en esputo o en el lavado broncoalveolar, varios trabajos recientes indican que, en un contexto clínico adecuado, los hallazgos de consolidaciones peribronquiales de atenuación negativa en TC son suficientes para emitir el diagnóstico, evitándose la realización de pruebas invasivas. Conflicto de intereses Ninguno de los autores tiene ningún conflicto de intereses de ningún tipo. Financiación No se ha obtenido ninguna financiación.	36464511	PMC9714629	NO-CC CODE	2022-12-03 23:20:12	no	Med Clin (Barc). 2022 Dec 1; doi: 10.1016/j.medcli.2022.10.008	utf-8	Med Clin (Barc)	2,022	10.1016/j.medcli.2022.10.008	oa_other
==== Front Lancet Reg Health Eur Lancet Reg Health Eur The Lancet Regional Health - Europe 2666-7762 The Authors. Published by Elsevier Ltd. S2666-7762(22)00243-5 10.1016/j.lanepe.2022.100547 100547 Articles Seroprevalence of anti-SARS-CoV-2 antibodies and cross-variant neutralization capacity after the Omicron BA.2 wave in Geneva, Switzerland: A population-based study Zaballa María-Eugenia ap Perez-Saez Javier abp de Mestral Carlos ac Pullen Nick a Lamour Julien a Turelli Priscilla d Raclot Charlène d Baysson Hélène ae Pennacchio Francesco a Villers Jennifer a Duc Julien d Richard Viviane a Dumont Roxane a Semaani Claire a Loizeau Andrea Jutta a Graindorge Clément a Lorthe Elsa a Balavoine Jean-François f Pittet Didier fg Schibler Manuel h Vuilleumier Nicolas fh Chappuis François ei Kherad Omar fj Azman Andrew S. ab Posfay-Barbe Klara M. kl Kaiser Laurent fhmn Trono Didier d Stringhini Silvia aceq∗ Guessous Idris eiq on behalf of the Specchio-COVID19 study groupo a Unit of Population Epidemiology, Division of Primary Care Medicine, Geneva University Hospitals, Geneva, Switzerland b Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States c University Centre for General Medicine and Public Health, University of Lausanne, Lausanne, Switzerland d School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland e Department of Health and Community Medicine, Faculty of Medicine, University of Geneva, Geneva, Switzerland f Department of Medicine, Faculty of Medicine, University of Geneva, Geneva, Switzerland g Infection Control Program and World Health Organization Collaborating Centre on Patient Safety, Geneva University Hospitals, Geneva, Switzerland h Division of Laboratory Medicine, Department of Diagnostics, Geneva University Hospitals, Geneva, Switzerland i Division and Department of Primary Care Medicine, Geneva University Hospitals, Geneva, Switzerland j Division of Internal Medicine, Hôpital de la Tour, Geneva, Switzerland k Department of Woman, Child, and Adolescent Medicine, Geneva University Hospitals, Geneva, Switzerland l Department of Pediatrics, Gynecology & Obstetrics, Faculty of Medicine, University of Geneva, Geneva, Switzerland m Division of Infectious Diseases, Department of Medicine, Geneva University Hospitals, Geneva, Switzerland n Geneva Centre for Emerging Viral Diseases, Geneva University Hospitals, Geneva, Switzerland ∗ Corresponding author. Division of Primary Care, Geneva University Hospitals, 1205, Geneva, Switzerland. o The members of this group are listed at the end of the article. p Contributed equally. q Contributed equally. 1 12 2022 1 12 2022 10054722 8 2022 28 10 2022 2 11 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. Background More than two years into the COVID-19 pandemic, most of the population has developed anti-SARS-CoV-2 antibodies from infection and/or vaccination. However, public health decision-making is hindered by the lack of up-to-date and precise characterization of the immune landscape in the population. Here, we estimated anti-SARS-CoV-2 antibodies seroprevalence and cross-variant neutralization capacity after Omicron became dominant in Geneva, Switzerland. Methods We conducted a population-based serosurvey between April 29 and June 9, 2022, recruiting children and adults of all ages from age-stratified random samples of the general population of Geneva, Switzerland. We tested for anti-SARS-CoV-2 antibodies using commercial immunoassays targeting either the spike (S) or nucleocapsid (N) protein, and for antibody neutralization capacity against different SARS-CoV-2 variants using a cell-free Spike trimer-ACE2 binding-based surrogate neutralization assay. We estimated seroprevalence and neutralization capacity using a Bayesian modeling framework accounting for the demographics, vaccination, and infection statuses of the Geneva population. Findings Among the 2521 individuals included in the analysis, the estimated total antibodies seroprevalence was 93.8% (95% CrI 93.1–94.5), including 72.4% (70.0–74.7) for infection-induced antibodies. Estimates of neutralizing antibodies in a representative subsample (N = 1160) ranged from 79.5% (77.1–81.8) against the Alpha variant to 46.7% (43.0–50.4) against the Omicron BA.4/BA.5 subvariants. Despite having high seroprevalence of infection-induced antibodies (76.7% [69.7–83.0] for ages 0–5 years, 90.5% [86.5–94.1] for ages 6–11 years), children aged <12 years had substantially lower neutralizing activity than older participants, particularly against Omicron subvariants. Overall, vaccination was associated with higher neutralizing activity against pre-Omicron variants. Vaccine booster alongside recent infection was associated with higher neutralizing activity against Omicron subvariants. Interpretation While most of the Geneva population has developed anti-SARS-CoV-2 antibodies through vaccination and/or infection, less than half has neutralizing activity against the currently circulating Omicron BA.5 subvariant. Hybrid immunity obtained through booster vaccination and infection confers the greatest neutralization capacity, including against Omicron. Funding General Directorate of Health in Geneva canton, Private Foundation of the Geneva University Hospitals, 10.13039/501100000780 European Commission (“CoVICIS” grant), and a private foundation advised by CARIGEST SA. Keywords Anti-SARS-CoV-2 antibodies Neutralizing antibodies Variants of concern Omicron Seroprevalence Switzerland ==== Body pmc Research in context Evidence before this study Although prevalence of anti-SARS-CoV-2 antibodies developed through infection and/or vaccination is thought to be high in most settings, the extent of population-level neutralizing capacity against past and current variants of concern (VOCs) is unknown. We searched PubMed, medRxiv, and bioRxiv using no language restrictions and search query [“COVID-19” AND “neutrali∗” AND (“prevalence” OR “seroprevalence”)] in the Title and Abstract fields on July 20, 2022. Most relevant seroprevalence studies we found focused on specific populations, particularly health care workers. Most studies reporting population-level seroprevalence estimates along with data from neutralization assays used the latter for diagnostic confirmation only. We found four studies reporting population-level seroprevalence of neutralizing antibodies, however, these were all conducted before June 2021 and assessed neutralization against the ancestral SARS-CoV-2 strain only. Added value of this study We here quantified both anti-SARS-CoV-2 antibody presence and neutralizing capacity against VOCs, including five Omicron subvariants, through a population-based serosurvey in the canton of Geneva, Switzerland. Our findings reveal that more than nine in ten (93.8%) individuals in the population have antibodies, including seven in ten (72.4%) individuals with antibodies of infection origin. However, neutralizing capacity is significantly lower ranging from 78.3% against the ancestral D614G strain to 46.7% for the currently dominating Omicron BA.5 subvariant, with large differences between age groups. Highest neutralizing capacity against Omicron subvariants was associated with vaccination booster alongside recent infection. These estimates provide an up-to-date and population-level picture of the SARS-CoV-2 immune landscape as shaped by age-specific infection and vaccination patterns. Implications of all the available evidence Our results show that through vaccination and several variant-driven pandemic waves, seroprevalence of anti-SARS-CoV-2 antibodies is high in the population, but that neutralizing capacity is variant-specific and determined by infection history and vaccination status, yielding strong differences between age groups in the general population. They also highlight the importance of booster vaccination, which appears to confer the highest neutralizing capacity, including against Omicron subvariants. These results may help prioritizing public health interventions against current and future variants as the COVID-19 pandemic progresses. Introduction By the end of 2021, most of the world's population had developed antibodies against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), through infection, vaccination, or both.1, 2, 3 In high-income countries, vaccination programs have contributed to high prevalence of anti-SARS-CoV-2 antibodies particularly among elderly people and individuals with chronic conditions,1, 2, 3 two groups with high risk of severe COVID-19, hospitalization, and death.4 , 5 While population-based serosurveys remain important to monitor the pandemic,6 they fall short in shedding light on the state of population-level protection against infection from current SARS-CoV-2 variants.7, 8, 9, 10 In fact, studies on neutralizing antibodies, primarily based on small and non-representative samples,11 have shown that neutralizing capacity depends strongly on how antibodies are mounted (through vaccination and/or infection; and the number, duration and severity of infection episodes) and may differ between variants.8, 9, 10, 11 This has implications for current immune landscapes shaped by complex age-specific vaccination and infection patterns. Vaccination-induced antibodies have demonstrated high neutralizing capacity against previously predominant variants (ancestral D614G, Alpha, and Delta), but substantially less neutralizing capacity against more recent and currently dominant Omicron subvariants8 , 9 , 12 , 13—even though, importantly, protection against severe disease, hospitalization, and death remains high.14 , 15 Simultaneously, antibodies developed through infection by the Omicron BA.1 subvariant have shown reduced neutralizing capacity against most other variants of concern (VOCs).8 , 9 , 13 , 16 , 17 From a public health standpoint, having up-to-date estimates of antibody seroprevalence, their origin, and their neutralizing capacity against diverse circulating VOCs is critically important to disseminate public health messages to the population and to inform and adapt decisions on vaccination strategies, mask requirements, and other preventive measures. Such evidence-based decisions are needed to minimize the number of people falling ill with severe disease, requiring hospitalization, and dying during the current wave, driven by the highly contagious Omicron BA.5 subvariant, as well as future pandemic waves. To our knowledge, at the time of writing, there are no population-based seroprevalence estimates of neutralizing antibodies against main VOCs, particularly after the Omicron variant became predominant. To fill this gap, we recruited a representative sample of the general population of Geneva, Switzerland, and assessed seroprevalence of anti-SARS-CoV-2 antibodies and their neutralizing capacity against SARS-CoV-2 VOCs 28 months after the first confirmed case in the country and 5 months after the Omicron BA.1 subvariant became dominant.18 With a population of about 500,000, the canton of Geneva, Switzerland, has had 261,946 confirmed cases (448 per 1000 inhabitants) and 883 deaths reported by July 21st, 2022.19 Our previous serosurvey revealed that by June–July 2021, around two thirds of the population had developed anti-SARS-CoV-2 antibodies following vaccination and/or infection, half of which having antibodies of infection origin.20 Since then, eligibility for vaccination against COVID-19 has progressively expanded in Geneva to cover most age groups, including ages 12–15 years since June 2021 and ages 5–11 years since January 2022 (appendix p 6). Methods Study design Between April 29 and June 9, 2022, we recruited participants from a random sample of individuals aged ≥6 months provided by the Geneva cantonal population registry (Office cantonal de la population et des migrations), and from an age- and sex-stratified random sample of adults who participated in at least one of our previous serosurveys (appendix p 7).20, 21, 22 Newly selected individuals were invited by letter, while returning participants were invited by letter or email when available. A written reminder was sent to all non-responding individuals and a phone call was made to all whose phone number was available. Children and teenagers (<18 years) were invited to participate with members of their household. Participation rates differed across age groups and depending on previous participation: 12.5% for those aged <18 years, 19.5% for those aged 18–64 years and 37.4% for those aged ≥65 years among newly invited participants; and 59.4% for those aged 18–64 years and 84.2% for those aged ≥65 years among returning participants (in all cases, after excluding not only ineligible individuals but also those hospitalized or in bad health conditions or not in Geneva during the whole study) (appendix pp 7–8). Participants provided a venous blood sample and completed an online or paper questionnaire that collected sociodemographic and vaccination information and COVID-19-related medical history. Informed written consent was obtained from all participants. The Geneva Cantonal Commission for Research Ethics approved this study (Project N° 2020-00881). For seroprevalence analyses, anti-SARS-CoV-2 antibodies presence was assessed in all participants while neutralizing activity against SARS-CoV-2 variants was only tested in a subset of them, the newly selected participants. This newly randomly selected sample included individuals of all ages, while the randomly selected sample of returning participants excluded those aged <18 years (appendix p 7). Immunoassays To detect anti-SARS-CoV-2 antibodies, we used two commercially available immunoassays: the Roche Elecsys anti-SARS-CoV-2 S and anti-SARS-CoV-2 N immunoassays (Roche Diagnostics, Rotkreuz, Switzerland), which detect immunoglobulins (IgG/A/M) against the receptor binding domain of the virus spike (S) protein (#09 289 275 190, Roche-S) and the virus nucleocapsid (N) protein (#09 203 079 190, Roche-N), respectively. Both assays have high accuracy and have been validated in multiple settings, including our previous serosurveys.20, 21, 22 We defined seropositivity using the manufacturer's provided cut-off values of titer ≥0.8 U/mL for the Roche-S, and cut-off index ≥1.0 for the Roche-N immunoassays. S3-cell free neutralization assay To assess anti-SARS-CoV-2 antibody neutralizing activity, we used the S3-ACE2 neutralization assay.23 , 24 Production and purification of trimeric Spike variants (D614G [B.1], Alpha [B1.1.7], Beta [B.1.351], Gamma [P.1], Delta [B.1.617.2], Iota [B.1.526], Kappa [B.1.617.1], Lambda [C.37], and Omicron [BA.1, BA.2, BA.2.12.1 and BA.4/BA.5]) and ACE2 mouse Fc fusion protein as well as Spike protein-beads coupling were performed as previously described in a validation study.23 Variant-specific validation tests against cell-based neutralization experiments are performed as lab routine on a small subset of serum samples if the corresponding live replicating virus lineage/sublineage is available, i.e. for four of the variants here analyzed (D614G, Alpha, Delta and Omicron BA.1). Neutralization assays were done in 96-well plates, where 10 multiplexed Spike variants were incubated with sera, as described.23 Briefly, a volume of 5 μL of serum per well was used for the starting dilution and a total of 6 serial dilutions (1:10, 1:30, 1:90, 1:270, 1:810, 1:7290) of sera were incubated 1 h with the Spike proteins before ACE2 mouse Fc fusion protein was added, and binding detected with an anti-mouse IgG-PE secondary antibody (eBioscience, Thermo Fisher Scientific, catalogue #12-4010-87). Control wells were included on each 96-well plate with each variant Spike-coupled beads alone. Plates were read with a Luminex 200 instrument and mean fluorescence intensity (MFI) for beads without serum was averaged and used as the 100% binding signal for the ACE2 receptor to the bead-coupled Spike trimer. MFI obtained for D614G Spike using a high concentration of imdevimab (RGN10987, 1 μg/mL), a monoclonal antibody known to neutralize the ancestral strain, was used as the maximum inhibition signal.25 The percent blocking of the Spike trimer-ACE2 interaction was calculated using the formula: % Inhibition = (1- ([MFI Test dilution – MFI Max inhibition]/[MFI Max binding – MFI Max inhibition]) × 100). Serum dilution response inhibition curves were generated using GraphPad Prism 8.3.0. NonLinear four-parameter curve fitting analysis of the agonist versus response, and ED50% values (mean serum dilution needed to achieve 50% neutralization) were extracted using an in-house script (https://doi.org/10.5281/zenodo.7124818). Neutralizing capacity was assessed against the ancestral variant (D614G), the Alpha, Beta, Gamma, Delta, Iota, Kappa, and Lambda variants, and the Omicron BA.1, BA.2, BA.2.12.1, and BA.4/BA.5 subvariants (Spike proteins of BA.4 and BA.5 subvariants share identical sequences17 and thus results of the cell-free surrogate neutralization assay apply to both). Statistical analyses To estimate seroprevalence of anti-SARS-CoV-2 antibodies (% and 95% credible intervals [95% CrI]), we used a Bayesian modeling framework jointly inferring anti-N and anti-S presence while accounting for age, sex, immunoassay performance, and household clustering following our previous work.20 Since the vaccines used to date in Geneva do not elicit a response to the SARS-CoV-2 N protein,26 we used participants' two-marker antibody profiles to estimate the proportion of those having anti-SARS-CoV-2 antibodies from any origin (vaccination and/or infection) and those having antibodies due to infection (who could be vaccinated or not). To obtain population-level estimates of total and infection-induced antibodies seroprevalence, we post-stratified model estimates to account for the sex and age distribution of the Geneva general population and for household clustering of infection and vaccination. Separately, in a model including only participants aged ≥18 years, we additionally post-stratified for educational level. We further developed a Bayesian logistic model to estimate the seroprevalence of neutralizing antibodies (% and 95% CrI), accounting for age, sex, and infection (uninfected, latest infected by a pre-Omicron variant, latest infected by an Omicron subvariant) and vaccination status (unvaccinated, vaccinated without booster, vaccinated with booster). The term ‘booster’ refers to a vaccine dose received several months after completion of primary vaccination: it could mean having received a third vaccine dose after two-dose primary vaccination or having received a second vaccine dose after one-dose primary vaccination post-infection. Data on infection dates and vaccination was self-reported by the participants at the moment of the blood drawing (appendix pp 3–5). Since Omicron became the dominant circulating variant in the Geneva region by late December 2021,18 we assumed infections were due to Omicron (no subvariants distinction) if the participant reported having had a COVID-19 diagnostic positive test (PCR or rapid antigen test, including self-tests) after January 1st, 2022. Our base model assumed additive contributions of vaccination and infection and was fit to each variant separately. As sensitivity analysis, we fit three additional sets of models: the first with interaction terms between infection and vaccination status, the second with overdispersion in neutralizing capacity, and the third with differential surrogate neutralization test performance for children under the age of 12 years versus older individuals. Model selection was performed based on estimated leave-one-out cross-validation error.27 To obtain population-level estimates of neutralizing antibodies seroprevalence, we post-stratified model estimates to account for the age and sex distribution and vaccination and infection statuses of the general population of Geneva. Full details of the statistical models are provided in the supplement (appendix pp 3–5). Role of the funding source The funding sources had no role in the study design, methodology, data collection or analysis, results interpretation, manuscript writing or decision to submit manuscript for publication. Results Overall anti-SARS-CoV-2 seroprevalence estimates Our analytical sample for seroprevalence estimation comprised 2521 participants (appendix p 7), of whom 55.2% were women, 21.4% were aged <18 years and 14.2% were aged ≥65 years (appendix p 9). Among adults, 11.3% had a primary education level and 56.9% had a tertiary education level, compared with 26.6% and 41.5%, respectively, in the general population of Geneva (appendix p 19). Overall, 75.4% of participants declared having received at least one COVID-19 vaccine dose at the time of their recruitment in the study, compared with 71.1% in the Geneva general population (appendix p 20); 96.9% of all participants tested positive for anti-S antibodies, and 70.0% tested positive for anti-N antibodies (Table 1 ; appendix p 10).Table 1 Demographic characteristics of sample, serological results, and seroprevalence estimates in Geneva, Switzerland, April 29 to June 9, 2022. Participants Vaccinated (self-reported)a Seropositiveb Seroprevalencec Anti-SARS-CoV-2 S protein Anti-SARS-CoV-2 N protein Antibodies of any origin % (95% CrI) Antibodies of infection origin % (95% CrI) Total 2521 1902 (75.4) 2442 (96.9) 1765 (70.0) 93.8 (93.1–94.5) 72.4 (70.0–74.7) Men 1129 (44.8) 844 (74.8) 1090 (96.5) 775 (68.6) 93.3 (92.4–94.2) 71.7 (68.5–74.9) Women 1392 (55.2) 1058 (76.0) 1352 (97.1) 990 (71.1) 94.2 (93.4–95.0) 73.1 (70.2–75.8) Age, y 0-5 144 (5.7) 3 (2.1) 112 (77.8) 107 (74.3) 77.2 (70.3–83.3) 76.7 (69.7–83.0) 6-11 242 (9.6) 23 (9.5) 231 (95.5) 217 (89.7) 91.0 (87.1–94.4) 90.5 (86.5–94.1) 12-17 155 (6.1) 95 (61.3) 153 (98.7) 126 (81.3) 93.4 (90.3–96.1) 86.4 (80.2–91.9) 18-24 169 (6.7) 150 (88.8) 169 (100.0) 135 (79.9) 95.0 (93.0–96.9) 84.0 (77.8–90.1) 25-34 267 (10.6) 239 (89.5) 263 (98.5) 209 (78.3) 94.9 (93.7–95.9) 78.9 (74.3–83.3) 35-49 787 (31.2) 706 (89.7) 780 (99.1) 554 (70.4) 95.0 (94.2–95.7) 74.4 (70.5–78.1) 50-64 400 (15.9) 349 (87.2) 387 (96.8) 254 (63.5) 95.4 (94.7–96.1) 67.7 (62.5–72.6) 65-74 183 (7.3) 173 (94.5) 179 (97.8) 92 (50.3) 95.1 (94.3–95.9) 55.0 (47.3–62.5) ≥75 174 (6.9) 164 (94.3) 168 (96.6) 71 (40.8) 96.7 (96.2–97.1) 45.9 (38.3–53.7) Data are n (%) unless otherwise stated. CrI: credible interval; N: nucleocapsid protein; S: spike protein; SARS-CoV-2: severe acute respiratory syndrome coronavirus 2. a Self-reported having received at least 1 dose of the COVID-19 vaccine before blood sample was drawn. b Serology based on Roche Elecsys anti-SARS-CoV-2 S immunoassay and N immunoassay, respectively. c Seroprevalence estimates reported as % and 95% credible interval, adjusted for test performance of both immunoassays and post-stratified to account for the sex and age distribution of the Geneva general population and for household clustering of infection and vaccination. Seroprevalence of any antibodies is based on proportion of participants with any anti-SARS-CoV-2 antibodies, additionally post-stratified to the vaccination data in the general population of Geneva; seroprevalence of infection antibodies is based on proportion of participants who were naturally infected (but could also have been vaccinated). After accounting for the Geneva population demographics and vaccination status, the overall seroprevalence of anti-SARS-CoV-2 antibodies developed through vaccination and/or infection was 93.8% (95% CrI: 93.1–94.5), with no difference between men and women (Table 1). Estimates varied only slightly across age groups except the youngest; overall seroprevalence among children aged 0–5 years was 77.2% (70.3–83.3) while it was 91.0% (87.1–94.4) among children aged 6–11 years, progressively increasing with age to reach 96.7% (96.2–97.1) among adults aged ≥75 years. The overall seroprevalence of infection-induced antibodies was 72.4% (70.0–74.7), also being similar between men and women, but varying considerably across age groups. Among children aged 0–5 years, the estimate was 76.7% (69.7–83.0), while it was highest among children aged 6–11 years at 90.5% (86.5–94.1), markedly decreasing with age and being lowest at 45.9% (38.3–53.7) among adults aged ≥75 years (Fig. 1 ; appendix pp 11, 21). We found no meaningful differences in vaccination rate or seroprevalence estimates according to educational level (appendix pp 21–22).Fig. 1 Seroprevalence of anti-SARS-CoV-2 antibodies in the general population of Geneva, Switzerland, April 29 to June 9, 2022. Seroprevalence estimates in total sample and by age group (in years) and origin of antibody response. Symbols indicate the antibody origin: dot indicates antibodies developed after infection; triangle indicates antibodies developed after infection and/or vaccination. Vertical bars represent 95% credible intervals. Seroprevalence estimates and determinants of anti-SARS-CoV-2 antibodies neutralizing capacity Our analytical sample for assessing anti-SARS-CoV-2 neutralization activity included a subset of 1160 participants (appendix p 7), of whom 54.4% were women, 28.2% aged <18 years and 19.3% aged ≥65 years (appendix pp 9, 24). Overall anti-SARS-CoV-2 seroprevalence estimates on this subsample were similar to those obtained on the main study sample (appendix p 24). Distribution of ED50% values obtained on this subsample for all tested SARS-CoV-2 variants using our cell-free surrogate neutralization assay is shown in Fig. S7 (appendix p 12). After accounting for the general population demographic and vaccination and infection status distributions, population-level neutralizing capacity was variant-specific (Fig. 2 ; appendix p 13): while 75–80% of the population had neutralizing antibodies against the ancestral D614G, Alpha, and Delta variants, seroprevalence of neutralizing antibodies against all tested Omicron subvariants was lower than 60% (Fig. 2; appendix pp 13, 25-27).Fig. 2 Seroprevalence of neutralizing antibodies against main SARS-CoV-2 variants in the general population of Geneva, Switzerland, April 29 to June 9, 2022. Panels show the effect of covariates tested in the model: age and infection and vaccination statuses (self-reported)—though sex was included as covariate, it showed no apparent effect, so it is excluded here, but estimates are shown in appendix pp 13, 25–27. Estimates of neutralizing capacity against Beta, Gamma, and Lambda variants are shown in appendix pp 13, 25–27. Global seroprevalence estimates for anti-S and anti-N antibodies are included in black in the two left panels for comparison purposes. Symbols indicate the antibody origin: dot indicates antibodies developed after infection (anti-N); triangle indicates antibodies developed after infection and/or vaccination (anti-S); and square indicates neutralizing antibodies (cell-free surrogate neutralization assay). Vertical bars represent 95% credible intervals. Among children aged 0–5 years, estimated seroprevalence of neutralizing antibodies was 29.3% (18.9–40.4) and 26.3% (16.4–36.7) against Alpha and Omicron BA.1, respectively, but 11.5% (4.7–20.2) against Omicron BA.4/BA.5. For children aged 6–11 years, seroprevalence of neutralizing antibodies ranged from 46.2% (36.7–56.0) against Alpha to 17.3% (9.8–25.5) against Omicron BA.4/BA.5. Starting with age 12 years, seroprevalence of neutralizing antibodies against D614G, Alpha and Delta was markedly higher: around 75% among adolescents aged 12–17 years, and around 90% among individuals aged ≥75 years. However, seroprevalence of neutralizing antibodies against the Omicron subvariants remained relatively similar, and lower, across these age groups, being 48.2% (39.1–57.7) and 49.7% (40.5–59.3) against BA.4/BA.5 among individuals aged 12–17 years and ≥75 years, respectively (Fig. 2; appendix pp 13, 25–27). Seroprevalence of neutralizing antibodies varied considerably according to vaccination and infection statuses (Fig. 2; appendix pp 13, 25–27). In general, neutralizing capacity against D614G, Alpha, and Delta variants was substantial (>90%) among vaccinated individuals having received booster vaccination, regardless of infection status. However, among vaccinated individuals without booster vaccination, neutralizing capacity was decreased (reaching 60%) if uninfected. Consistently, in multivariable analyses, having received booster vaccination showed the strongest association with neutralizing capacity; for instance, boosted individuals had 14.2 (95% Cr: 6.4–28.0) times greater odds of having neutralizing antibodies against Delta, and 2.2 (1.4–3.4) times the odds of having antibodies with neutralizing capacity against Omicron BA.4/BA.5, compared with vaccinated individuals without booster vaccination (Table 2 ). Regarding infection, compared with individuals last infected before 2022, those infected in 2022 had more than three times greater odds of having neutralizing antibodies against the Omicron subvariants. Finally, regardless of variant, unvaccinated individuals had substantially lower odds of having neutralizing antibodies (Table 2).Table 2 Association between individuals’ attributes and neutralizing capacity against main SARS-CoV-2 variants in the general population of Geneva, Switzerland, April 29 to June 9, 2022. SARS-CoV-2 variant D614G Alpha Delta Omicron BA.1 Omicron BA.2 Omicron BA.2.12.1 Omicron BA.4/BA.5 Men 0.83 (0.51–1.26) 0.84 (0.52–1.26) 0.90 (0.59–1.34) 0.97 (0.69–1.31) 0.92 (0.65–1.27) 0.98 (0.67–1.39) 1.07 (0.75–1.49) Women 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) Age group, y 0-5 0.66 (0.28–1.31) 0.74 (0.31–1.46) 0.69 (0.30–1.36) 1.39 (0.59–2.74) 0.99 (0.38–2.05) 0.68 (0.24–1.51) 0.68 (0.24–1.52) 6-11 1.34 (0.66–2.43) 1.22 (0.60–2.20) 1.22 (0.60–2.22) 0.89 (0.42–1.67) 0.98 (0.45–1.82) 1.09 (0.48–2.10) 0.95 (0.41–1.86) 12-17 2.20 (0.98–4.25) 2.05 (0.91–3.98) 2.50 (1.17–4.77) 2.41 (1.22–4.27) 2.75 (1.39–4.89) 2.63 (1.33–4.72) 2.03 (1.01–3.75) 18-24 1.38 (0.49–3.18) 2.40 (0.79–6.05) 2.10 (0.76–4.75) 2.41 (1.11–4.65) 2.91 (1.32–5.68) 1.64 (0.80–3.02) 1.41 (0.67–2.64) 25-34 0.86 (0.32–1.97) 1.00 (0.37–2.27) 1.09 (0.44–2.37) 1.34 (0.70–2.33) 1.56 (0.83–2.72) 1.28 (0.66–2.27) 1.00 (0.52–1.78) 35-49 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 50-64 0.75 (0.33–1.55) 0.72 (0.32–1.44) 0.80 (0.40–1.47) 0.95 (0.55–1.54) 1.03 (0.61–1.65) 0.85 (0.48–1.36) 0.77 (0.44–1.26) 65-74 0.74 (0.22–2.00) 0.85 (0.27–2.25) 0.73 (0.29–1.62) 0.77 (0.41–1.29) 0.72 (0.38–1.24) 0.61 (0.32–1.06) 0.54 (0.27–0.98) ≥75 0.96 (0.28–2.59) 1.19 (0.33–3.24) 0.78 (0.29–1.70) 0.99 (0.52–1.72) 0.74 (0.40–1.30) 1.00 (0.51–1.82) 0.83 (0.43–1.51) Infection statusa Uninfected 0.18 (0.08–0.35) 0.18 (0.08–0.34) 0.10 (0.05–0.18) 0.20 (0.12–0.30) 0.21 (0.12–0.33) 0.25 (0.14–0.38) 0.23 (0.13–0.35) Pre-2022 infection 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 2022 infection 0.89 (0.49–1.56) 0.91 (0.49–1.55) 0.95 (0.52–1.57) 4.62 (2.77–7.39) 3.67 (2.15–6.02) 3.99 (2.35–6.65) 3.91 (2.36–6.43) Vaccination statusb Unvaccinated 0.03 (0.01–0.05) 0.03 (0.02–0.06) 0.05 (0.03–0.09) 0.10 (0.05–0.16) 0.09 (0.05–0.15) 0.08 (0.04–0.14) 0.07 (0.03–0.12) Vaccinated, no booster 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) 1 (ref) Vaccinated, booster 21.99 (7.90–53.43) 25.21 (8.23–62.8) 14.19 (6.44–28.0) 2.89 (1.78–4.56) 3.35 (2.05-5.30) 2.51 (1.56–3.96) 2.20 (1.36–3.39) Data are odds ratio (95% credible interval) of having neutralizing antibodies against a SARS-CoV-2 variant. Only main variants having been detected at significant proportions in the canton of Geneva are included in this table. SARS-CoV-2: severe acute respiratory syndrome coronavirus 2. a Self-reported most recent infection via a COVID-19 diagnostic positive test (PCR or rapid antigen test, including self-tests). b Vaccination status based on self-reported number of doses received and dates. The term 'booster' refers to a vaccine dose received several months after completion of primary vaccination: it could mean having received a third vaccine dose after two-dose primary vaccination or having received a second vaccine dose after one-dose primary vaccination post-infection. Discussion This serosurvey found that by April–June 2022, 93.8% of the Geneva population had developed antibodies against SARS-CoV-2 after vaccination and/or infection. Yet the proportion of the population with neutralizing antibodies varied considerably across SARS-CoV-2 variants, from more than three quarters against Alpha to less than half against the currently circulating Omicron BA.5 subvariant, with particularly lower proportions in children <12 years and unvaccinated individuals who were last infected before Omicron became dominant. The seroprevalence of total antibodies estimated in this study is slightly lower than the 97.5–98.8% found in two other Swiss sample populations by the end of March 2022,28 and the 94.7–96.1% reported in the United Kingdom by mid-July 2022,29 but these estimations only included individuals aged ≥16 years. We also found that 72.4% of the population had been infected—a 42.5 percentage point increase from the 29.9% seroprevalence reported by June–July, 202120 (appendix p 28). This important increase in seroprevalence of infection-induced antibodies within a 11-month period was largest among children aged 0–5 years (55.8 percentage point increase) and 6–11 years (59.5 percentage point increase), indicating that the Delta-dominant and notably the Omicron-dominant pandemic waves in Geneva particularly affected children,30 as observed in other countries.29 , 31 Conversely, this increase in infection-induced antibodies does not appear to have translated to a corresponding level of neutralizing capacity; for instance, while three quarters of children aged 0–5 years had developed antibodies through infection, only around one in four had neutralizing antibodies against the Delta or Omicron BA.1 or BA.2 variants. This proportion was only slightly higher among the 6–11 years old even though nine in ten children in this age group had infection-induced antibodies. These findings indicate that most children, not having received the COVID-19 vaccine and first becoming infected by Delta or Omicron BA.1 or BA.2, did not develop neutralizing capacity against earlier variants and only limited neutralizing capacity against the currently circulating BA.5 subvariant. Our model suggests that this discrepancy can be explained by the significantly lower vaccination rates in these age groups alone, without requiring age-specific effects (95% CrI of odds ratios for all variants and for all age groups cover 1, except for the 12–17 years age group, which had higher odds of neutralization). We however note that most infected-only participants in our sample were children under the age of 12 years, which therefore limits the power to identify age-specific effects within the analysis. Although antibodies in children seem to have a lower neutralization efficacy, their immune response has been observed to resemble that of adults in cases of mild COVID-19, but to differ in moderate to severe COVID-19 and multisystem inflammatory syndrome,32 including a robust mucosal response to the SARS-CoV-2 virus with high levels of interferon and a different T-cell response.32 , 33 While data on these other immunity components were not available in our study, children-specific immune response may explain the consistently lower levels of severe COVID-19, hospitalizations, and death observed among children, despite much higher levels of infections during the Delta- and Omicron-driven waves.20 We found little age differences in seroprevalence of total antibodies among adults, in contrast to what we observed in our previous serosurvey,20 likely reflecting the fact that for adults vaccination has since been widely available. Simultaneously, seroprevalence of infection-induced antibodies differed markedly among age groups, peaking for the 6–11 years age group and reducing gradually with increasing age, reflecting the pattern of age-related infection risk observed across pandemic waves, preventive measures and behavior, as well as the earlier vaccination availability and higher uptake in older people.1 , 3 , 20 , 31 The finding that neutralizing capacity against the D614G, Alpha, and Delta variants was reduced among individuals infected in 2022 compared with those last infected before 2022 is in line with reports of reduced overall neutralizing capacity against pre-Omicron variants after infection by Omicron.8 , 13 , 16 Notably, we also found that having received booster vaccination was associated with increased neutralizing capacity against all variants, including Omicron subvariants, in agreement with previous reports.8 , 9 , 13 , 16 , 17 Finally, we found that being vaccinated (with or without booster) and infected in 2022, a period when Omicron was almost exclusively circulating in Geneva, was associated with increased neutralizing capacity against all Omicron subvariants, in line with previous reports.9 , 13 , 16 , 17 , 34 The aforementioned Swiss serosurvey reported considerably higher neutralizing capacity than what we found, but we are unable to compare findings given a lack of methodological details in the study.28 Potential explanations for the observed disparities include differences in the sample age composition, proportion of boosted participants and of recent Omicron-infections, as well as the threshold used to define neutralizing activity. Despite these differences, we deem our results to reliably reflect the state of neutralization capacity in the Geneva population thanks to the use of a random sample of the population together with post-stratification based on the state's demographics and population-level infection and vaccination statuses. In general, our results revealed that a combination of vaccination (with booster) and recent infection appeared to confer the highest level of neutralizing antibodies against each variant, including Omicron subvariants. Implications for public health and clinical practice Our findings show that, while most of the population have developed anti-SARS-CoV-2 antibodies through infection and/or vaccination, less than half have neutralizing antibodies against the highly contagious, currently circulating Omicron BA.5 variant, including only one in four children aged <12 years. A similar pattern of total antibodies and variant-specific neutralizing antibodies in the population is likely to be present in other settings that have experienced the same successive variant-driven pandemic waves as Geneva, Switzerland. The highest level of neutralizing capacity against each variant was observed among vaccinated individuals who had received a booster dose, indicating that vaccine-induced antibodies confer substantial neutralizing capacity against pre-Omicron variants, while also maximizing neutralizing capacity against Omicron subvariants. At the same time, not surprisingly, we observed a reduced neutralizing capacity against Omicron subvariants relative to pre-Omicron variants. This suggests that updated vaccines specifically targeting the Omicron lineage may be beneficial in containing the spread of infections and their consequent health and socio-economic burden.1 Our findings also revealed that, while less than half of the population show neutralizing capacity against the currently dominant Omicron subvariants, a substantial proportion have neutralizing capacity against less common variants, including Beta, Gamma, and Lambda (appendix pp 12-13, 25-27). Since future VOCs may develop from or share structural characteristics with less frequent variants, monitoring the level of neutralizing capacity against them in the population may help in building scenarios for future pandemic waves. Strengths and limitations This study benefits from several strengths, including the large representative sample, the recent recruitment time-frame post-Omicron BA.1 and BA.2-driven pandemic waves, the measurement of antibodies against both the SARS-CoV-2 S and N proteins as well as neutralizing antibodies against ten SARS-CoV-2 variants/subvariants, and a robust modeling framework. We also acknowledge several limitations. First, like most serosurveys,35 the sample only included formal residents of the canton and had a higher proportion of individuals with tertiary education than in the general population (appendix p 19). While we used education as the only socioeconomic marker, the lack of socioeconomic inequalities in our findings is consistent with previous studies in the population of Geneva and that of neighboring countries in which other socioeconomic markers were examined.36, 37, 38, 39, 40 We chose against using multiple imputation for the 5% of participants (n = 99) with missing education data–while this may introduce a slight bias in the education-stratified analysis, it is unlikely to have affected the conclusions drawn from it. Second, data on infection dates was self-reported by the participants at the moment of the blood drawing and only the latest known infection was included in the analysis. Third, the cut-off value for neutralizing activity was defined using pre-pandemic sera from adult donors only—no children samples were included in the validation study, as previously reported.23 While the level of neutralizing antibodies has been shown to be a strong marker of immune protection, it does not fully describe it, especially among children whose immune response differs from that of adults.32 , 33 Fourth, our seroprevalence estimates assume constant test sensitivity despite potential changes in test performance with time since infection. Previous studies have shown that the performance of both Roche immunoassays used in this study remains high and with a very limited decaying trend several months after infection.41 We therefore expect the impact of these changes to be accounted for within our Bayesian framework, which explicitly models test performance uncertainty allowing for departure from manufacturer-provided values. Lastly, due to lack of data, we did not include in our modeling framework indicators that have been shown to influence neutralizing capacity, including severity and duration of symptoms, number of infections, and interval between last infection/vaccination and blood sampling.9, 10, 11 , 42 , 43 Conclusions This study provides up-to-date seroprevalence estimates of anti-SARS-CoV-2 antibodies in a representative sample of the general population 5 months after Omicron became the dominant circulating SARS-CoV-2 variant in Geneva, Switzerland. It shows that while most of the population (notably ≥12 years of age) have neutralizing antibodies against pre-Omicron variants, the seroprevalence of antibodies with neutralizing capacity against the currently circulating and highly contagious Omicron BA.5 subvariant is low. Our findings suggest that the mass vaccination of older individuals, as well as other preventive measures and behaviors, may have protected them from infection during the Delta- and Omicron BA.1- and BA.2-driven waves. They also show that the highest level of neutralizing capacity against most VOCs is attained through hybrid immunity combining vaccination, notably including a booster dose, and recent infection. As new variants emerge driving new pandemic waves, having up-to-date snapshots of the immune landscape of the population can help develop rational risk mitigation strategies. Contributors IG, SS, LK, DT, MEZ, JPS and NP designed the study. MEZ, JL, PT, CR, HB, FP, JV, JD, VR, RD, CS, AJL, CG, EL, MS, NV, OK, KMPB, LK, DT, SS and IG contributed to participants’ recruitment and/or data acquisition. NP and JPS conducted statistical analyses. CdM conducted literature review and wrote the first draft of the manuscript. All authors contributed to the interpretation of results and read and approved the final manuscript. IG, SS, MEZ, JPS, NP and JL had full access to all data in the study, and the corresponding author had final responsibility for decision to submit for publication. Data sharing statement Our data are accessible to researchers upon reasonable request for data sharing to the corresponding author. All code to reproduce the analysis as well as code for the creation of a synthetic dataset on which to run the analysis code is available at: https://github.com/UEP-HUG/sp4_public. The script to process neutralizing antibody data from the Luminex instrument is available at: https://zenodo.org/record/7124818. Specchio-COVID19 study group Isabelle Arm-Vernez, Andrew S Azman, Delphine Bachmann, Antoine Bal, Jean-François Balavoine, Michael Balavoine, Rémy P Barbe, Hélène Baysson, Lison Beigbeder, Julie Berthelot, Patrick Bleich, Livia Boehm, Gaëlle Bryand, François Chappuis, Prune Collombet, Sophie Coudurier-Boeuf, Delphine Courvoisier, Alain Cudet, Vladimir Davidovic, Carlos de Mestral, Paola D'ippolito, Richard Dubos, Roxane Dumont, Isabella Eckerle, Nacira El Merjani, Antoine Flahault, Natalie Francioli, Marion Frangville, Clément Graindorge, Idris Guessous, Séverine Harnal, Samia Hurst, Laurent Kaiser, Omar Kherad, Julien Lamour, Pierre Lescuyer, Arnaud G L'Huillier, François L'Huissier, Andrea Jutta Loizeau, Elsa Lorthe, Chantal Martinez, Lucie Ménard, Ludovic Metral-Boffod, Alexandre Moulin, Mayssam Nehme, Natacha Noël, Francesco Pennacchio, Javier Perez-Saez, Didier Pittet, Klara M Posfay-Barbe, Géraldine Poulain, Caroline Pugin, Nick Pullen, Viviane Richard, Frederic Rinaldi, Déborah Rochat, Irine Sakvarelidze, Khadija Samir, Hugo Santa Ramirez, Etienne Satin, Philippe Schaller, Manuel Schibler, Stephanie Schrempft, Claire Semaani, Silvia Stringhini, Stéphanie Testini, Didier Trono, Déborah Urrutia-Rivas, Charlotte Verolet, Pauline Vetter, Jennifer Villers, Guillemette Violot, Nicolas Vuilleumier, Ania Wisniak, Sabine Yerly, and María-Eugenia Zaballa. Declaration of interests DT is a founder and co-chair of the Scientific Advisory Board of Aerium Therapeutics, holds stock in that company, and has two patents pending for monoclonal antibodies against SARS-CoV-2. KMPB is a member of the Advisory Boards for pneumococcal vaccine and varicella vaccine at MSD. All other authors declare that they have no competing interests. Appendix A Supplementary data Supplementary appendix Supplementary File S1 Supplementary File S2 Acknowledgments We warmly thank the HEdS students Léa Baettig, Jessica Chavet, Noé Kummer, Manuel Lobato Sineiro and Reza Nazari and the nurses Nassima Sadadou Djouder, Manon Ladouce, Marie Le Belz, Celine Breuil and Marion Figini for their passionate work during this study. We thank Florence Pojer, Kelvin Lau and the Protein Production and Structure Core Facility team at EPFL for the tremendously efficient Spike variants production, and Vanessa Genet at EPFL for her precious technical help. We also thank the members of the sérothèque and the laboratory of virology at HUG for sample preparation and serological testing. We are grateful to the Hôpital des enfants des HUG, the Hôpital de La Tour, the Clinique and Permanence d’Onex and the Centre Médical du Lignon for allowing us to use their premises for the recruitment of participants, and to Dr Cyril Sahyoun and his team for our enriching collaboration aimed at easing the experience of children during blood sampling. Finally, we are deeply grateful to all the participants for their interest and invaluable contribution to the study. Sources of funding: General Directorate of Health in Geneva canton, Private Foundation of the Geneva University Hospitals, European Commission (“CoVICIS” grant), and a private foundation advised by CARIGEST SA. The funding sources had no role in the study design, methodology, data collection or analysis, results interpretation, manuscript writing or decision to submit manuscript for publication. Appendix A Supplementary data related to this article can be found at https://doi.org/10.1016/j.lanepe.2022.100547. ==== Refs References 1 World Health Organization Interim statement on hybrid immunity and increasing population seroprevalence rates 2022 https://www.who.int/news/item/01-06-2022-interim-statement-on-hybrid-immunity-and-increasing-population-seroprevalence-rates 2 Bergeri I. Whelan M. Ware H. Global SARS-CoV-2 seroprevalence from January 2020 to April 2022: A systematic review and meta-analysis of standardized population-based studies. PLOS Medicine 19 11 2022 e1004107 10.1371/journal.pmed.1004107 36355774 3 Rostami A. Sepidarkish M. Fazlzadeh A. Update on SARS-CoV-2 seroprevalence: regional and worldwide Clin Microbiol Infect 27 2021 1762 1771 34582980 4 Bonanad C. García-Blas S. Tarazona-Santabalbina F. The effect of age on mortality in patients with COVID-19: a meta-analysis with 611,583 subjects J Am Med Dir Assoc 21 2020 915 918 32674819 5 Liu H. Chen S. Liu M. Nie H. Lu H. Comorbid chronic diseases are strongly correlated with disease severity among COVID-19 patients: a systematic review and meta-analysis Aging Dis 11 2020 668 678 32489711 6 Theel E.S. Slev P. Wheeler S. Couturier M.R. Wong S.J. Kadkhoda K. The role of antibody testing for SARS-CoV-2: is there one? J Clin Microbiol 58 2020 007977-20 7 European Centre for Disease Prevention and Control Considerations for the use of antibody tests for SARS CoV-2 - first update. Stockholm 2022 8 Turelli P. Zaballa M.-E. Raclot C. Omicron infection induces low-level, narrow-range SARS-CoV-2 neutralizing activity medRxiv 2022 10.1101/2022.05.02.22274436 9 Bekliz M. Adea K. Vetter P. Neutralization capacity of antibodies elicited through homologous or heterologous infection or vaccination against SARS-CoV-2 VOCs Nat Commun 13 2022 3840 35787633 10 Khoury D.S. Cromer D. Reynaldi A. Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection Nat Med 27 2021 1205 1211 34002089 11 Savage H.R. Santos V.S. Edwards T. Prevalence of neutralising antibodies against SARS-CoV-2 in acute infection and convalescence: a systematic review and meta-analysis PLoS Negl Trop Dis 15 2021 e0009551 12 Tang J. Novak T. Hecker J. Cross-reactive immunity against the SARS-CoV-2 Omicron variant is low in pediatric patients with prior COVID-19 or MIS-C Nat Commun 13 2022 2979 35624101 13 Servellita V. Syed A.M. Morris M.K. Neutralizing immunity in vaccine breakthrough infections from the SARS-CoV-2 Omicron and Delta variants Cell 185 2022 1539 1548.e5 35429436 14 Lauring A.S. Tenforde M.W. Chappell J.D. Clinical severity of, and effectiveness of mRNA vaccines against, covid-19 from omicron, delta, and alpha SARS-CoV-2 variants in the United States: prospective observational study BMJ 376 2022 e069761 15 Plumb I.D. Effectiveness of COVID-19 mRNA vaccination in preventing COVID-19–associated hospitalization among adults with previous SARS-CoV-2 infection — United States, June 2021–february 2022 MMWR Morb Mortal Wkly Rep 2022 71 10.15585/mmwr.mm7115e2 16 Reynolds C.J. Pade C. Gibbons J.M. Immune boosting by B.1.1.529 (Omicron) depends on previous SARS-CoV-2 exposure Science 0 2022 eabq1841 17 Hachmann N.P. Miller J. Collier A.Y. Neutralization escape by SARS-CoV-2 omicron subvariants BA.2.12.1, BA.4, and BA.5 N Engl J Med 387 2022 86 88 35731894 18 Swiss national SARS-CoV-2 genomic and variants, surveillance program. Surveillance variants SARS-CoV-2 - Genève et National 2022 https://www.hug.ch/laboratoire-virologie/surveillance-variants-sars-cov-2-geneve-national 19 République et Canton de Genève COVID19 à Genève. Données cantonales 2022 published online July 19 https://infocovid.smc.unige.ch/ 20 Stringhini S. Zaballa M.-E. Pullen N. Seroprevalence of anti-SARS-CoV-2 antibodies 6 months into the vaccination campaign in Geneva, Switzerland, 1 June to 7 July 2021 Eurosurveillance 26 2021 2100830 21 Stringhini S. Wisniak A. Piumatti G. Seroprevalence of anti-SARS-CoV-2 IgG antibodies in Geneva, Switzerland (SEROCoV-POP): a population-based study Lancet 396 2020 313 319 32534626 22 Stringhini S. Zaballa M.-E. Perez-Saez J. Seroprevalence of anti-SARS-CoV-2 antibodies after the second pandemic peak Lancet Infect Dis 21 2021 600 601 33539733 23 Fenwick C. Turelli P. Pellaton C. A high-throughput cell- and virus-free assay shows reduced neutralization of SARS-CoV-2 variants by COVID-19 convalescent plasma Sci Transl Med 13 2021 eabi8452 24 Obeid M. Suffiotti M. Pellaton C. Humoral responses against variants of concern by COVID-19 mRNA vaccines in immunocompromised patients JAMA Oncol 8 2022 e220446 25 Takashita E. Kinoshita N. Yamayoshi S. Efficacy of antibodies and antiviral drugs against covid-19 omicron variant N Engl J Med 386 2022 995 998 35081300 26 Wheeler Sarah E. Shurin Galina V. Mary Yost Differential antibody response to mRNA COVID-19 vaccines in healthy subjects Microbiol Spectr 9 2021 003411-21 27 Vehtari A. Gelman A. Gabry J. Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC Stat Comput 27 2017 1413 1432 28 Amati R. Frei A. Kaufmann M. Functional immunity against SARS-CoV-2 in the general population after a booster campaign and the Delta and Omicron waves, Switzerland, March 2022 Eurosurveillance 27 2022 2200561 29 UK Office for National Statistics Coronavirus (COVID-19) latest insights https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/articles/coronaviruscovid19latestinsights/antibodies 2022 30 Lorthe E. Bellon M. Berthelot J. A SARS-CoV-2 omicron (B.1.1.529) variant outbreak in a primary school in Geneva, Switzerland Lancet Infect Dis 22 2022 767 768 35429994 31 Clarke K.E.N. Jones J.M. Deng Y. Seroprevalence of infection-induced SARS-CoV-2 antibodies — United States, September 2021–February 2022 Morb Mortal Wkly Rep 71 2022 606 608 32 Chou J. Thomas P.G. Randolph A.G. Immunology of SARS-CoV-2 infection in children Nat Immunol 23 2022 177 185 35105983 33 Yoshida M. Worlock K.B. Huang N. Local and systemic responses to SARS-CoV-2 infection in children and adults Nature 602 2022 321 327 34937051 34 Wu M. Wall E.C. Carr E.J. Three-dose vaccination elicits neutralising antibodies against omicron Lancet 399 2022 715 717 35065005 35 Accorsi E.K. Qiu X. Rumpler E. How to detect and reduce potential sources of biases in studies of SARS-CoV-2 and COVID-19 Eur J Epidemiol 36 2021 179 196 33634345 36 Aziz N.A. Corman V.M. Echterhoff A.K.C. Seroprevalence and correlates of SARS-CoV-2 neutralizing antibodies from a population-based study in Bonn, Germany Nat Commun 12 2021 2117 33837204 37 Santa-Ramírez H.-A. Wisniak A. Pullen N. Socio-economic determinants of SARS-CoV-2 infection: results from a population-based cross-sectional serosurvey in Geneva, Switzerland Front Public Health 10 2022 https://www.frontiersin.org/articles/10.3389/fpubh.2022.874252 38 Richard A. Wisniak A. Perez-Saez J. Seroprevalence of anti-SARS-CoV-2 IgG antibodies, risk factors for infection and associated symptoms in Geneva, Switzerland: a population-based study Scand J Public Health 50 2022 124 135 34664529 39 Warszawski J. Beaumont A.-L. Seng R. Prevalence of SARS-Cov-2 antibodies and living conditions: the French national random population-based EPICOV cohort BMC Infect Dis 22 2022 41 35000580 40 Wachtler B. Müters S. Michalski N. Socioeconomic inequalities in the prevalence and perceived dangerousness of SARS-CoV-2 infections in two early German hotspots: findings from a seroepidemiological study BMC Res Notes 14 2021 375 34565448 41 Perez-Saez J. Zaballa M.-E. Yerly S. Persistence of anti-sars-cov-2 antibodies: immunoassay heterogeneity and implications for serosurveillance Clin Microbiol Infect 2021 10.1016/j.cmi.2021.06.040 published online July 21 42 Chia W.N. Zhu F. Ong S.W.X. Dynamics of SARS-CoV-2 neutralising antibody responses and duration of immunity: a longitudinal study Lancet Microbe 2 2021 e240 e249 33778792 43 Dispinseri S. Secchi M. Pirillo M.F. Neutralizing antibody responses to SARS-CoV-2 in symptomatic COVID-19 is persistent and critical for survival Nat Commun 12 2021 2670 33976165	36474728	PMC9714630	NO-CC CODE	2022-12-03 23:20:12	no	Lancet Reg Health Eur. 2022 Dec 1;:100547	utf-8	Lancet Reg Health Eur	2,022	10.1016/j.lanepe.2022.100547	oa_other
==== Front JHEP Rep JHEP Rep JHEP Reports 2589-5559 Elsevier S2589-5559(22)00210-5 10.1016/S2589-5559(22)00210-5 100638 Article Contents 1 12 2022 12 2022 1 12 2022 4 12 100638100638 2020 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	36474962	PMC9714688	NO-CC CODE	2022-12-03 23:20:12	no	JHEP Rep. 2022 Dec 1; 4(12):100638	utf-8	JHEP Rep	2,022	10.1016/S2589-5559(22)00210-5	oa_other
==== Front World J Surg World J Surg World Journal of Surgery 0364-2313 1432-2323 Springer International Publishing Cham 6855 10.1007/s00268-022-06855-9 Original Scientific Report Geographical Inequalities in Access to Bellwether Procedures in Brazil Faleiro Matheus Daniel [email protected] 12 Fernandez Miguel Godeiro 32 Santos Jéssica Moreira 42 Menezes Catarina Ester Gomes 52 Lima João Vitor Sabadine 12 Haddad Júlia Oliveira Dabien 62 Viana Sofia Wagemaker 72 Alonso Nivaldo 8 1 grid.8430.f 0000 0001 2181 4888 Federal University of Minas Gerais, Belo Horizonte, Brazil 2 International Student Surgical Network Brazil, Belo Horizonte, Brazil 3 grid.414171.6 0000 0004 0398 2863 Bahiana School of Medicine and Public Health (EBMSP), Salvador, Brazil 4 grid.419130.e 0000 0004 0413 0953 Faculdade Ciências Médicas de Minas Gerais, Belo Horizonte, Brazil 5 grid.8399.b 0000 0004 0372 8259 State University of Bahia, Salvador, Brazil 6 grid.441787.9 0000 0001 0152 1834 Universidade de Itaúna, Itaúna, Brazil 7 grid.411191.d 0000 0000 9146 0440 Kursk State Medical University, Kursk, Russia 8 grid.11899.38 0000 0004 1937 0722 Division of Plastic Surgery, University of São Paulo, São Paulo, Brazil 1 12 2022 17 5 11 2022 © The Author(s) under exclusive licence to Société Internationale de Chirurgie 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 Brazil is a middle-income country that aims to provide universal health coverage, but its surgical system’s efficiency has rarely been analyzed. In an effort to strengthen surgical national systems, the Lancet Commission on Global Surgery proposed bellwether procedures as quality indicators of surgical workforces. This study aims to evaluate regional inequalities in access to bellwether procedures and their associated mortality across the five Brazilian geographical regions. Methods Using DATASUS, Brazil’s national healthcare database, data were collected on the total amount of performed bellwether procedures—cesarean section, laparotomy, and open fracture management—and their associated mortality, by geographical region. We evaluated the years 2018–2020, both in emergent and elective conditions. Statistical analysis was performed by one-way ANOVA test and Tukey’s multiple comparisons test. Results During this period, DATASUS registered 2,687,179 cesarean sections, 1,036,841 laparotomies, and 648,961 open fracture treatments. The access and associated mortality related to these procedures were homogeneous between the regions in elective care. There were significant geographical inequalities in access and associated mortality in emergency care (p < 0.05, 95% CI) for all bellwether procedures. The Southeast, the most economically developed region of the country, was the region with the lowest amount of bellwether procedures per 100,000 inhabitants. Conclusion Brazil’s public surgical system is competent at promoting elective surgical care, but more effort is needed to fortify emergency care services. Public policies should encourage equity in the geographic allocation of the surgical workforce. Supplementary Information The online version contains supplementary material available at 10.1007/s00268-022-06855-9. ==== Body pmcIntroduction Surgery-manageable diseases account for 28% of the global disease burden [1]. Logically, a scale-up of a functional surgical system could benefit a large portion of the global population, potentially saving millions of lives that lack access to surgical procedures. In this regard, the Lancet Commission on Global Surgery (LCoGS) was established with the goal of integrating surgery into the global health agenda, promoting political change, and defining scalable solutions for providing quality surgery to all [2]. The LCoGS stated that access to bellwether procedures could serve as a quality indicator for surgical systems. Bellwether procedures can be defined as a group of operations that have been identified as markers of a healthcare system’s capability to provide essential and emergency surgical care [3]. They are composed of three surgical procedures, which are cesarean section (CS), a procedure in which the baby is delivered through an incision in the abdominal wall and the uterus [4]; laparotomy, an open procedure used to treat a wide range of surgically manageable diseases such as penetrating abdominal injuries and acute abdomen [5]; and treatment of open bone fractures, injuries in which the body's protective skin barrier has been broken and the potential for contamination is high [6]. Previous research found that the ability to perform bellwether procedures was strongly linked to the ability to perform all obstetric, general, basic, emergency, and orthopedic procedures in low and middle-income countries [7]. Brazil is a developing Latin-American country that aspires to provide universal health coverage through Sistema Único de Saúde (SUS) or Unified Health System. This system is based on the principles of comprehensiveness, equality, and universality, with a strong emphasis on primary care [8]. Providing equal access to SUS is a major challenge for the public healthcare system, given that the country comprises five geographical regions with distinct socioeconomic and historical backgrounds [9]. In fact, Brazil was the world's most unequal country in the 1980s, and this regional inequality became more evident in historically excluded regions of the country, such as the Northeast, Midwest, and the North, where the supply of health services is lower than in the South and Southeast regions. In those three regions, health facilities are usually located at a greater distance, making it difficult to access surgical care [9]. The primary goal of this study was to assess regional differences in access to bellwether procedures between the five Brazilian geographical regions, both in emergency and elective scenarios. We also assessed the associated mortality related to these procedures. As a result, we intend to evaluate the quality of the Brazilian public surgery system and guide future public policies to improve it. Material and methods The total number of bellwether procedures (CS, laparotomy, and open fracture management) performed and their associated mortality for all patient admissions in Brazil’s public hospitals were collected in the period 2018–2020, both in emergent and elective conditions. These data were compared between the five Brazilian geographical regions: Midwest, North, Northeast, South, and Southeast. Source of data The data of this study were collected from DATASUS, a national, open-access database organized and funded by the federal government that contains aggregate data from the Hospital Information System (SIH). In this system, individual hospitals submit a monthly formal report that contemplates a variety of diagnostic and procedural statistics. SIH collects information on 60–70% of hospital admissions in the country within the Unified Health System [3]. These data do not include private healthcare information, for which approximately 28.5% of the population is covered [10]. Data about the number of bellwether procedures and their associated mortality, between 2018 and 2020 per Brazilian geographical region, were collected in elective and emergent contexts. The unique procedure-related codes used to access data on bellwether procedures in DATASUS were the same as cited in a previous study [3]. Microsoft Excel 365 was used to tabulate data and construct a figure summarizing results for Brazilian regions (Fig. 1).Fig. 1 Regional inequalities in access and mortality rate related to bellwether procedures in Brazil. The color gradient represents the mean amount of procedures between 2018 and 2020, per 100,000 inhabitants, with the darkest shade corresponding to the highest registered number. The gray circles represent the mean mortality rate during the same period, per 100,000 inhabitants, and their sizes are proportional to the registered value Given the differences in population between the five Brazilian geographical regions and the biases that it could represent in the research, the results regarding the total annual number of procedures and associated mortality were divided by the total population of the region in 2021, according to Brazilian Institute of Geography and Statistics [11], which is the country's most important source of demographic data. Following, the mean amount for each bellwether procedure per 100,000 inhabitants in each region, both in elective and emergent contexts, and the associated mean mortality rate, between the years 2018 and 2020, were calculated. The standard deviations of the means were also assessed. Statistical analysis Statistical analyses were performed using the statistical program GraphPad Prism version 8.0.0 (GraphPad Software, California, USA). Since the total number of procedures per region and associated death rate between 2018 and 2020 were normally distributed according to the D'Agostino-Pearson normality test (p > 0.05), analysis of variance was performed by one-way ANOVA and Tukey’s multiple comparisons tests. We considered a 95% confidence interval. Differences in data that reached p < 0.05 were considered statistically significant. Results Cesarean section Of the 2,687,179 procedures reported, 4.5% (121,046) were elective, and 95.5% (2,566,133) were emergencies. In terms of mortality, 1232 deaths were reported, with 3.09% (38) occurring during elective and 96.91% (1194) occurring during emergency procedures. Figure 1 synthesizes the mean results in mortality and amount of procedures, both in elective and emergent contexts on the period, for all the bellwether procedures. Regarding elective procedures, there were no differences in access (Fig. 2) and associated mortality (Fig. 3) between the five Brazilian geographical regions (p > 0.05). Relating to emergency procedures, all Brazilian regions presented inequalities in access, as presented in Fig. 4. Associated mortality in emergencies also presented differences between the regions as shown in Fig. 5.Fig. 2 Mean amount of elective procedures per 100,000 inhabitants between 2018 and 2020. Error bars indicate the standard deviation (SD). “” represents the statistically significant difference (p < 0.05) unless otherwise noted. a Cesarean sections. b Laparotomies. c Open fracture management Fig. 3 Mean associated mortality of elective procedures per 100,000 inhabitants between 2018 and 2020. Error bars indicate the standard deviation (SD). “” represents the statistically significant difference (p < 0.05) unless otherwise noted. a Cesarean sections. b Laparotomies. c Open fracture management Fig. 4 Mean amount of emergency procedures per 100,000 inhabitants between 2018 and 2020. Error bars indicate the standard deviation (SD). “” represents the statistically significant difference (p < 0.05) unless otherwise noted. All comparisons were statistically significant. a Cesarean sections. b Laparotomies. c Open fracture management Fig. 5 Mean associated mortality of emergency procedures per 100,000 inhabitants between 2018 and 2020. Error bars indicate the standard deviation (SD). “” represents the statistically significant difference (p < 0.05) unless otherwise noted. a Cesarean sections. b Laparotomies. c Open fracture management Concerning the mean amount of CS performed in the period, the Southeast region was the region that less performed this procedure both in elective and emergency contexts. In elective, the Midwest region was the region that most performed this procedure, 44.83% higher than the Southeast, and in emergency, the North was the region that most performed this procedure, 60.89% higher than the Southeast. Related to the associated mortality of CS, in elective, Midwest was the region with the highest mortality, 331.45% higher than the Northeast, the region with the lowest mortality. In emergency, the North had the highest mortality rate, 193.98% higher than the South, which had the lowest. Laparotomy Of the 1,036,841 procedures reported, 37.5% (388,785) were elective procedures and 62.5% (648,056) emergency procedures. In total, 29,169 deaths were reported in the period, including 10.03% (2,924) during elective procedures and 89.97% (26,245) during emergencies. There were no differences in access and associated mortality in elective laparotomy procedures between the five Brazilian geographical regions, and just one comparison, South versus Midwest, presented a statistically significant difference (p < 0.05) in associated mortality (Fig. 3). Concerning emergency procedures, there were significant inequalities both in access (Fig. 4) and mortality (Fig. 5). In terms of the mean amount of laparotomies performed during the period, the lowest amount was registered for the Southeast, and the highest for the South, both in elective and emergency contexts (76.52% and 48.55% higher than the Southeast, respectively). In elective laparotomies, associated mortality was found to be significantly higher in the South region, 59.23% higher than in the Midwest, with the lowest mortality. In emergencies, again, the South was the region with the highest mortality, 95.31% superior to the North, the region with the lowest mortality. Treatment of open bone fractures In the period of 2018–2020, 648,961 open fracture managements were reported in DATASUS, which included 12.98% (84,217) elective procedures and 87.02% (564,744) emergency procedures. Concerning mortalities reported in this period, 5467 deaths were reported, 4.7% (257) in elective and 95.3% (5,210) in emergency procedures. There were no differences in access and associated mortality in elective open fracture procedures between the five regions (p > 0.05), except when comparing the access between the South and Midwest regions (Fig. 2) and the associated mortality between the Southeast and Midwest regions (Fig. 3). However, there were significant differences both in access (Fig. 4) and mortality (Fig. 5) related to emergency procedures, although associated mortality showed more statistically significant differences than access. Concerning the mean amount in the period, in elective, the North region performed the procedure most, 78.8% higher than the Midwest, region with the lowest amount. Interestingly, in the emergency context, the scenario was opposed: the Midwest was the region that most performed this procedure, 90.02% higher than the North. Related to the associated mortality, in elective, the Southeast was the region with the highest mortality, 216.9% higher than the Midwest, the region with the lowest mortality. In emergency, the South was the region with the highest mortality, 259.73% higher than the North, the region with the lowest number. Discussion In this work, we used the Brazilian public health database to evaluate differences in access and mortality rate related to bellwether procedures between the five Brazilian geographical regions, as these indicators represent the quality of the entire public surgical system. Our findings are important because it was found that while Brazil's elective environment may be effective at promoting surgical care, more has to be done to increase access and reduce mortality related to urgent procedures. In addition to our main findings, we noticed that, despite having the highest surgical workforce density in the country [12], the Southeast region had the lowest mean amount between the three bellwether procedures per 100,000 inhabitants. It was an unexpected result because of the great proportion of academic surgical programs in this region [13]. This result could be explained by not evaluating the private practice since DATASUS represents hospital information regarding the public sector. 37.5% of the health insurance coverage in 2019 was concentrated in the Southeast region, and these unexpected findings may be due to a greater effect of the private sector in this region [10]. The key point of our work is that, despite the existence of a national healthcare system, regional variations are a major contributor to health inequalities in the country. For example, in laparotomy data, the South was the region that most performed this procedure in elective and urgent contexts. This region performed 25.06% more laparotomies by population than the Northeast, one of the poorest regions of the country, during the period evaluated. It can be related to the fact that exploratory laparotomies are high-risk and costly procedures that demand strong healthcare structures [14], which suggests that the differences observed between these two regions can be due to economic and social factors. Additionally, emergency laparotomies had the highest mortality among the procedures analyzed in this study. It supports prior findings of other works that also observed high mortality and morbidity associated with laparotomies in emergencies [15, 16]. The South’s highest mortality rate associated with this procedure, both in emergency and elective contexts, can be due to the high surgical volume registered, but more work is needed to understand this result. The treatment of open fractures is considered an orthopedic emergency due to the risk of contamination and death. The management of these fractures is challenging in the context of trauma care, and there are still controversies regarding its optimal timing [17]. Most of the current literature supports that the surgical treatment of open fractures should be done under urgent circumstances [18]. Our findings reveal that the surgical management regimen for open fractures in Brazil is following most of the current literature, with 87.02% of these procedures being performed as emergencies. Based on CS data, we determined that the North, one of Brazil's poorest regions, was the one that most performed this procedure in the emergency context. The difference in rates of CS compared with the other regions could be explained by the demographic and social context of the North, where there is low coverage of public health services [9], resulting in poor prenatal care and compromising the detection of early maternal–fetal disorders that contraindicate vaginal delivery, which increases CS in an emergency scenario. This region also has the highest mortality in emergency CS, which may also be related to the difficulty of accessing medical services. In addition, the deficiency in health professionals and adequate structure to assist vaginal delivery, often demanding more time and staff, can lead to an indiscriminate amount of CS without strong clinical indication [19]. Despite CS representing a medical advancement, given its potential to save lives and avoid complications [20], there is an excess of the procedure in medical practice. In Latin America and the Caribbean, a study on CS rates in 150 countries showed that 40.5% of births occur by CS, reaching 55.6% in Brazil [19]. It represents a sociocultural problem because of the association of this procedure with increased risk to the health of the mother and the newborn [20]. The major limitation of this work, as an epidemiological study, is that it is necessary to consider reporting bias and the consequent distortion of data due to underreporting or reporting error, which is more common to occur in the North and Northeast of Brazil than in other regions [21]. Another limitation of our study is the emphasis placed on the public healthcare system, even though private insurance covers 28.5% of the population [10]. Regardless, our findings referred to a system that any Brazilian can access at any time and for free across the five Brazilian geographical regions. Also, this study includes the year 2020 in the analysis, which is recognized by the COVID-19 international outbreak that impacted the Brazilian public health system. Previous studies show that there was a huge backlog in the delivery of elective operations during this year in Brazil, although health state-level policies were effective in ensuring minimal reductions in the delivery of emergent surgeries [22]. Our data about laparotomies and treatment of open bone fractures, available in Online Resource 1, are in accordance with these previous results, as we can see a decrease in the total number of elective procedures per 100,000 inhabitants in the year 2020. Despite these limitations, this study utilized statistical techniques to conduct a detailed analysis of public data and provides an evaluation of regional surgical inequalities in Brazil. The access and associated mortality to the three bellwether procedures were described in this country for the first time, and the findings of this study have significant implications for the overall improvement of the public healthcare system. Conclusion Identifying access gaps is critical for evaluating barriers to providing equal surgical care. The current study highlights large geographical differences in emergency procedures, raising concerns about universality and equity, two of the guidelines of the Brazilian National Public Healthcare System [8]. Beyond differences in access, our results could reflect differences in practice, workforce, and the economic development of the Brazilian regions. In light of these findings, there is an urgent need to identify the origin of these differences and to solve them to overcome inequality in surgical care in Brazil. We hope that this research may contribute to the development of public policies that will improve surgical access across the country, benefiting especially those who cannot afford healthcare insurance. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 56 kb) Funding All the authors declare that this study has received no funding. Declarations Conflict of interest MDF holds a voluntary position as National Research Director at InciSioN Brazil, MGF holds a voluntary position with the National Research Team at InciSioN Brazil, JMS holds a voluntary position with the National Research Team at InciSioN Brazil and is a scientific committee member with the Society of Medical Academics of Minas Gerais, CEGM holds a voluntary position with the National Research Team at InciSioN Brazil, JVSL holds a voluntary position with the National Research Team at InciSioN Brazil, JODH holds a voluntary position with the National Research Team at InciSioN Brazil, SWV holds a voluntary position as National Chair at InciSioN Brazil. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Shrime MG Bickler SW Alkire BC Global burden of surgical disease: an estimation from the provider perspective Lancet Glob Health 2015 3 S8 S9 10.1016/S2214-109X(14)70384-5 25926322 2. Meara JG Leather AJM Hagander L Global surgery 2030: evidence and solutions for achieving health, welfare, and economic development Lancet 2015 386 9993 569 624 10.1016/S0140-6736(15)60160-X 25924834 3. Truche P Roa L Citron I Bellwether procedures for monitoring subnational variation of all-cause perioperative mortality in Brazil World J Surg 2020 44 10 3299 3309 10.1007/s00268-020-05607-x 32488666 4. Antoine C Young BK Cesarean section one hundred years 1920–2020: the good, the bad and the ugly J Perinat Med 2020 49 1 5 16 10.1515/jpm-2020-0305 32887190 5. Biffl WL Leppaniemi A Management guidelines for penetrating abdominal trauma World J of Surg 2015 39 6 1373 1380 10.1007/s00268-014-2793-7 25315088 6. Sagi HC Patzakis MJ Evolution in the acute management of open fracture treatment? Part 1 J Orthop Trauma 2021 35 9 449 456 10.1097/BOT.0000000000002094 34415869 7. O'Neill KM Greenberg SL Cherian M Bellwether procedures for monitoring and planning essential surgical care in low- and middle-income countries: caesarean delivery, laparotomy, and treatment of open fractures World J Surg 2016 40 11 2611 2619 10.1007/s00268-016-3614-y 27351714 8. Santos N 30 Years of SUS: the beginning, the pathway and the target. SUS 30 anos: o início, a caminhada e o rumo Ciên Saúde Colet 2018 23 6 1729 1736 10.1590/1413-81232018236.06092018 29972482 9. Moura M Diego L Lack of access to surgery: a public health problem Cad Saúde Pública 2017 33 10 e00151817 10.1590/0102-311X00151817 29091179 10. Souza Júnior P Szwarcwald CL Damacena GN Health insurance coverage in Brazil: analyzing data from the National Health Survey, 2013 and 2019 Ciên Saúde Colet 2021 26 suppl 1 2529 2541 10.1590/1413-81232021266.1.43532020 34133632 11. Brasil (2021) Estimativas da população residente no Brasil e unidades da federação com data de referência em 1o de julho de 2021. Inst Bras Geogr e Estatística [Internet] (3):1–119. https://ftp.ibge.gov.br/Estimativas_de_Populacao/Estimativas_2021/estimativa_dou_2021.pdf. Accessed 13 July 2022 12. Massenburg BB Saluja S Jenny HE Assessing the Brazilian surgical system with six surgical indicators: a descriptive and modelling study BMJ Glob Health 2017 2 2 e000226 10.1136/bmjgh-2016-000226 28589025 13. Alonso N Massenburg BB Galli R Surgery in Brazilian health care: funding and physician distribution Rev Col Bras Cir 2017 44 2 202 207 10.1590/0100-69912017002016 28658340 14. Bampoe S Odor PM Ramani Moonesinghe S A systematic review and overview of health economic evaluations of emergency laparotomy Perioper Med 2017 6 1 1 12 10.1186/s13741-017-0078-z 15. Nally DM Sorensen J Kavanagh DO Emergency laparotomy research methodology: a systematic review Surgeon 2020 18 2 80 90 10.1016/j.surge.2019.06.003 31345681 16. Chua M Chan D Increased morbidity and mortality of emergency laparotomy in elderly patients World J Surg 2020 44 3 711 720 10.1007/s00268-019-05240-3 31646368 17. Giglio PN Cristante AF Pécora JR Avanços no tratamento das fraturas expostas Rev Bras Ortop 2015 50 2 125 130 10.1016/j.rboe.2015.02.009 26229904 18. Ryan SP Pugliano V Controversies in initial management of open fractures Scand J Surg 2014 103 2 132 137 10.1177/1457496913519773 24737846 19. Betrán AP Ye J Moller AB The increasing trend in caesarean section rates: global, regional and national estimates: 1990–2014 PLoS ONE 2016 11 2 e0148343 10.1371/journal.pone.0148343 26849801 20. Torloni MR Brizuela V Betran AP Mass media campaigns to reduce unnecessary caesarean sections: a systematic review BMJ Glob Health 2020 5 2 e001935 10.1136/bmjgh-2019-001935 32296554 21. Departamento de Informática do SUS/Ministério da Saúde. Razão entre óbitos informados e estimados - F.11 - 2006. Brasília: Ministério da Saúde, 2006. Available in: http://fichas.ripsa.org.br/2006/F-11/?l=en_US. Accessed 25 July 2022 22. Truche P Campos LN Marrazzo EB Association between government policy and delays in emergent and elective surgical care during the COVID-19 pandemic in Brazil: a modeling study Lancet Reg Health Am 2021 3 100056 10.1016/j.lana.2021.100056 34725652	36456731	PMC9714764	NO-CC CODE	2022-12-03 23:20:12	no	World J Surg. 2022 Dec 1;:1-7	utf-8	World J Surg	2,022	10.1007/s00268-022-06855-9	oa_other
==== Front J Immigr Minor Health J Immigr Minor Health Journal of Immigrant and Minority Health 1557-1912 1557-1920 Springer US New York 1434 10.1007/s10903-022-01434-5 Original Paper Factors Contributing to West Indian American Depression http://orcid.org/0000-0002-3949-5261 Kanhai Gregory A. [email protected] 1 Chang Doris F. [email protected] 2 1 grid.264933.9 0000 0004 0523 9547 The New School for Social Research, Department of Psychology, 80 Fifth Avenue, New York, NY 10011 USA 2 grid.137628.9 0000 0004 1936 8753 Silver School of Social Work, New York University, 1 Washington Square North, New York, NY 10011 USA 1 12 2022 111 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. This study explored factors that mediate the relationship between subjective wellbeing and depression in a sample of West Indian American immigrants. An intersectional theoretical framework was used to identify the relative contribution of psychological stressors—perceived discrimination, financial strain and acculturative stress—that mediate the relationship between subjective wellbeing and depression. A geographically diverse sample was recruited by an online survey (N = 255), consisting of 138 men, 115 women, 173 Indo-West Indians and 82 Afro-West Indians. Path analysis was used to identify the relative contribution of psychological stressors. Acculturative stress and financial strain were both statistically significant predictors of depression. Financial strain was identified as the major mediator between subjective wellbeing and depression in West Indian Americans. West Indian Americans are vulnerable to financial strain and acculturative stress. These sources of psychological stress are important contributors to depression in the community. More research is needed to clarify these relationships. Keywords West Indian American Financial strain Acculturative stress Perceived discrimination Subjective wellbeing ==== Body pmcIntroduction West Indian Americans (English-speaking Caribbean Americans) are a uniquely at-risk population for negative health outcomes due to multiple life domain pressures. This assessment is based on data from Afro-West Indian immigrants and economic data on West Indians [1, 2]. It is also consistent with the larger body of immigrant mental health literature showing that rates of depression in immigrant communities are more elevated than that in the general U.S. population [1, 2]. West Indian Americans are one of the largest immigrant communities in New York state, making up approximately 13% of the immigrant population. The three largest subgroups are Jamaican, Guyanese, and Trinidadian, who together account for the fourth, fifth and eighth largest immigrant groups in the state, respectively [3]. Despite their large size, West Indian Americans’ English-speaking origins and racial diversity (including Afro-Caribbean, Indo-Caribbean, and multiracial individuals) create statistical invisibility among immigrant health profiles in the literature. This study adopts an intersectional theory approach to deepen our understanding of West Indian American wellbeing by examining the relationships between different intersectional stressors and mental health. Intersectional Theory Intersectional theory, borrowed from feminist and critical race theory [4, 5] posits that the intersectionality of life domains within which an individual resides in the host society allows the examination of the various forms of psychological stressors that are unique to immigrant groups. Two sets of influences are theorized to be related to mental health through intra-psychic and social domains of life that enhance or diminish aspects of wellbeing, such as a person’s sense of belonging [6], social network density [7], language [5, 8, 9], economic barriers and perceived discrimination [4, 9, 10]. Torres et al. [11] describe these two categories of stressors as separate components of perceived stress—that which is due to adaptation (acculturative stress) and that due to social and structural stressors. An intersectional theoretical framework as proposed by Viruell-Fuentes et al. (2012) provides a conceptual approach in which to explore these intersectional stressors and will be used to support understanding of the cultural and socio-structural factors in explaining mental health outcomes of West Indian Americans. Afro-West Indians and Mental Health Afro-West Indians may be a good indicator group of how language and culture may influence mental health of West Indians in general [2]. Using data from the National Study of American Life, Williams found differences in rates of depression between Haitian, Spanish and English-speaking Afro-Caribbeans (Afro-West Indians) [2, 12]. Lifetime and 1-year depressive rates were 19.9% and 10.8% for U.S.-born Afro-West Indians, and 12.9% and 6.9% for their foreign-born counterparts [2, 13, 14]. Afro-West Indians were also less likely to use mental health resources than their Spanish-speaking counterparts [12]. Importantly, the number of years of U.S. residency was associated with lifetime risk for psychiatric disorders, and U.S.-born individuals were at higher risk than foreign-born individuals for all Afro-Caribbeans. Subjective Wellbeing Subjective wellbeing is a measure of the overall assessment of one’s life experience and life trajectory [15]. It has been validated across cultures as a robust assessment of host society integration and quality of life [16]. Subjective wellbeing has been shown to be related to economic welfare [17], social connections [18] and access to public services [19]. As such, subjective wellbeing is an important assessment of an immigrant community’s sense of place in their host society. Most importantly, subjective wellbeing has been validated as a measure of positive health outcomes [20] and inversely, with depression [21]. Acculturative Stress Immigrants’ well-being is also likely to be impacted by stressors associated with adaptation to the host society [22]. The construct of acculturative stress has been described as an agglomeration of various stressful experiences encountered by immigrant communities. An important component of acculturative stress is acculturation pressure from family and society [23]. Perceived Discrimination Perceived discrimination is defined as the belief that one has been unfairly treated because of one’s group identification [24] and has been identified as a potently deleterious factor affecting mental health outcomes in immigrant communities [24–27]. The intersection of being monolingual English-native and a racial minority may expose West Indian Americans to discrimination that may be less readily perceived by immigrant groups with limited English proficiency, with adverse impacts for mental health. Explicit and subtle forms of discrimination are pervasive and normative in U.S. society [26, 28, 29], with 46.5% of non-Hispanic Black Americans reporting regular experiences of perceived discrimination, and almost 25% reporting frequent occurrences. Financial Strain Min [30] found that language and socio-economic position separated South Asians from Indo-West Indians despite cohabitation in the same neighborhoods in New York City. Indo-West Indian average income was reported to be between $39,000–$49,000. This is similar to that reported for Afro-Caribbean immigrants [31]. The average U.S. income during the same period was reported to be $52,000. Financial strain has been linked to psychological distress in minority populations [32, 33]. Perceived Social Support Social network robustness and social support are important protective factors of mental health for immigrant communities [7, 34, 35]. Connection to a social group is known to insulate immigrants from social stressors by providing a sense of safety, understanding and belonging [35–39]. Social support has been explored as a mediator between acculturative stress and depressive symptoms [40, 41]. Present Study Building on literature on intersectionality and immigrant mental health, this study seeks to examine the mediating influence of social support, acculturative stress, financial strain and perceived discrimination between subjective wellbeing and depression in the West Indian American immigrant community. The hypothesis is that host society factors such as financial strain and perceived discrimination will be primary mediators between subjective wellbeing and depression (Fig. 1.)Fig. 1 Predicted relationships among variables Methods The study consisted of a 20-min confidential survey hosted on Qualtrics.com through the New School for Social Research. Data collection was conducted under approval of the institutional review board, during the period of March to July 2016. West Indian immigrants were recruited through snowball sampling in the community in the New York City area, and outside the New York City area through Amazon Mechanical Turk (MTurk). Participants were offered compensation of a $2 Amazon.com gift certificate for community participation, and $0.50 compensation for MTurk participant in accordance with the pay rate on the platform. There were 349 respondents from MTurk, and 139 via the snowball method. The completion rates were 51.8% for MTurk and 32.3% for community. Those who met criteria i.e. first- and second-generation West Indian residing in the U.S. and who had completed the survey on MTurk resulted in a final sample size of 255 participants (196 from MTurk and 59 from snowballing community). The data used for this study consisted of the scales as described: Variables Perceived Discrimination The Everyday Discrimination Scale (EDS; [42]) is a nine-item scale that measures the frequency of perceived discriminatory events in one’s life. The EDS is reported to be a measure of microinsults and micro-assaults in the microaggressions taxonomy [43]. Each item on the EDS is rated on a six-point Likert-like frequency scale, ranging from 5-almost never to 0-almost always. Items are reverse scored, and a total sum score calculated. Higher scores indicate more perceptions of perceived discrimination. The EDS has demonstrated good internal consistency (0.87) and has been found to be correlated with measures of psychological distress [44]. In this study Chronbach’s alpha was (0.94). Financial Strain The Financial strain Scale for Undergraduates (FSS; [45]) is a 13-item scale that measures financial strain in the three domains of stress—credit burden, stress from inability to weather a financial emergency, and current financial insecurity. The questions ask about the frequency of thought about situations that are financially stressful, such as “living paycheck to paycheck,” and “Having to borrow money from family/friends” which were identified as appropriate items for an immigrant population. It has been shown to be correlated with measures of psychological distress such as the Daily Stress Inventory [46]. Items are rated on a four-point scale ranging from, “never” to “all the time.” Higher scores indicate higher levels of stress. Scoring consists of calculating a sum score for the 13 items. The scale was found to have a high internal consistency (0.87) among a young adult population and was (0.92) in the present study. Depression The Center for Epidemiologic Studies Depression-revised (CESD-R; [47]) is a 20-item screening tool for depressive symptoms. The CESD-R asks respondents to choose from four possible responses on a Likert-like format, where “0” is “rarely or none of the time (less than 1 day)”, and “3” is “nearly every day for 2 weeks.” Scores range from 0 to 60 with higher scores reflecting greater levels of depressive symptoms while lower scores reflect lower levels of symptoms. The CESD has very good internal consistency (0.85) for the general population and (0.90) for a psychiatric population. The original scale has been shown to be valid in many diverse community samples [48] and in West Indian Americans [49]. In this study, the internal consistency was (0.96). Acculturative Stress The Societal, Attitudinal, Familial and Environmental Acculturative Stress Scale, revised (SAFE-R; [50]) is an expanded 30-item version of the original 24-item SAFE which includes items related to family acculturation gaps between parents and children. The 24-item SAFE has been used in many studies and has demonstrated internal consistency of (0.89). In this study the SAFE-R demonstrated internal consistency of (0.95). It is scored from 1 (not stressful) to 5 (extremely stressful), with items that are not applicable being rated 0 (not applicable). The scale is scored by calculating a total sum score. Subjective Wellbeing The Satisfaction with Life scale [15] is a measure of subjective wellbeing. It is a five-question global measure of cognitive and affective self-assessment of a respondent’s sense of wellness rated on a seven-point Likert-like scale. It is scored by summing the scores. It has demonstrated strong self-report and peer-reported correlation. It has shown good convergent validity with other measures of subjective wellbeing. The scale has demonstrated a co-efficient alpha of (0.87) and test–retest reliability of (0.82). It has been used across cultures and races with good reliability [51]. In this study, internal reliability was (0.88). Perceived Social Support Multidimensional Scale of Perceived Social Support [52] is a 12-question measure with three dimensions of social support—family, friends, and significant others. It has been validated across different ethnic and national populations and shows good internal reliability and test–retest reliability. Chronbach’s alpha was found to be 0.88–0.92 across samples of pregnant women, people with mental illness and college students [53]. In this study, the internal reliability was (0.69), which falls in the acceptable range of reliability. Scoring consists of calculating the score for the total scale. Analysis Data analyses were conducted using IBM SPSS version 25 and R programming language. Mean differences were identified between subgroups and correlations used to determine whether the hypothesized relationships among variables agreed with the literature. A path analysis was then conducted to identify the major mediating variables between subjective wellbeing and depression (Fig. 1). Results Demographics Two hundred and fifty-five (255) participants (59 Community and 196 MTurk), age 18 and over, completed the survey and self-identified as West Indian by three questions asking about self and parental nativity. The sample consisted of 138 (54.0%) male, 115 (45.1%) female and 2 (0.75%) intersexed individuals, with a median age of 28.7 years (see Table 1). 173 self-identified as Indo-Caribbean and 82, as Afro-Caribbean. Median income was $39, 441, with 127 (49.6%) having less than a bachelor’s degree, and 91 (35.5%) having a bachelor’s degree. 111 (43.5%) were first-generation immigrants and 144 (56.5%) were West Indian-born. The majority, 169 (66.3%), lived with either family or relatives, and 162 (63.5%) reported being single, either dating or not dating. The large majority, 219 (85.9%) reported speaking standard English, and 36 (14.1%) reported speaking English creole. All participants reported English as their primary language. Among survey respondents, 204 (80.0%) indicated that they had never utilized a mental health service provider to cope with problems.Table 1 Reported demographic of study respondents Demographic Category Population Total N = 255 % n % n % Indo-West Indian 173 (61) Afro-West Indian 82 (23) Total (n = 255) Indo- Afro- Median Sex Male 138 (51) 92 (53) 46 (56) Female 115 (46) 79 (46) 36 (44) Intersex 2 (1) 2 (1) 0 (0) Age Range 18–25 102 (40) 73 (42) 29 (35) 26–34* 103 (40) 71 (41) 32 (39) 28.7 35–44 29 (11) 17 (10) 12 (15) 45–54 16 (6) 9 (5) 7 (9) 55–64 3 (1) 1 (1) 2 (2) Income $0–$20,000 58 (23) 44 (25) 14 (17) $21,000–$40,000* 79 (31) 48 (28) 31 (38) $39, 441 $41,000–$60,000 66 (26) 42 (24) 24 (30) $61, 000–$80, 000 39 (15) 30 (17) 9 (11) $81, 000 and over 13 (5) 9 (5) 4 (5) Education High school/GED 29 (11) 19 (11) 10 (12) Some college 68 (27) 46 (27) 22 (27) 2-yr college 28 (11) 17 (10) 11 (13) 4-yr college 90 (35) 68 (39) 22 (27) Master’s degree 32 (13) 17 (10) 15 (18) Doctoral & professional 6 (2) 6 (3) 0 (0) Country of origin United States 144 (57) 99 (57) 45 (55) Trinidad 25 (10) 18 (10) 7 (9) Guyana 28 (11) 26 (15) 2 (7) Jamaica 36 (14) 15 (9) 21 (26) Other 22 (9) 14 (8) 7 (9) Living situation Parents 71 (28) 25 (33) 11 (13) Family 86 (34) 15 (20) 31 (38) Housemates 25 (10) 8 (11) 11 (13) Alone 60 (24) 25 (33) 19 (23) Other 12 (5) 2 (3) 7 (9) Location New York City & LI 85 33 Northeast 10 (4) Southeast 52 (20) Elsewhere 108 (42) Category where median lies Descriptive statistics and mean differences for sex, generational status and sampling group are reported in Table 2. Male respondents (M = 27.271, SD = 11.746.51) reported significantly higher perceived discrimination, t(251) = 2.078, p ≤ 0.05, than female respondents (M = 24.226, SD = 10.967). Afro-West Indians reported lower subjective wellbeing (M = 21.122, SD = 7.337), t(253) = 3.991, p ≤ 0.001, and higher financial strain (M = 1.277, SD = 0.727), t(253) = -2.027, p ≤ 0.005, than Indo-West Indians (M = 24.815, SD = 6.688) and (M = 1.081, SD = 0.688), respectively. First generation individuals (M = 4.881, SD = 1.105) reported lower perceived social support t(253) = 1.960, p ≤ 0.05 than second-generation individuals (M = 4.595, SD = 1.191). MTurk respondents reported higher acculturative stress (M = 49.821, SD = 27.747), t(253) = -2.787, p ≤ 0.001, and perceived discrimination (M = 26.704, SD = 11.863), t(253) = -2.218, p ≤ 0.05, than community-contacted respondents (M = 38.949, SD = 20.525) and (M = 22.966, SD = 9.046). No between-group differences were present for depressive symptoms. Twenty percent [51] of respondents scored as clinically significant for depression on the CESD-R, and these individuals reported lower subjective wellbeing, and higher acculturative stress, perceived discrimination, and financial strain than their non-depressed cohort (see Table 2).Table 2 Descriptive statistics and t-test for group differences Variable M SD n M SD n 95% CI for M. Diff t df Sex Male 138 Fem 115 Subjective wellbeing 23.493 6.955 23.809 7.359 [−2.092, 1.459] −.350 251 Perceived social support 4.736 1.292 4.706 .997 [−,259, .321] .206 251 Acculturative stress 48.065 27.455 45.547 25.035 [−.404, 9.078] .756 251 Perceived discrimination 27.217 11.746 24.226 10.967 [.157, 5.826] 2.078 251 Financial strain 1.203 .706 1.070 .697 [−.419, .307] 1.496 251 Depression 6.920 8.536 6.496 8.970 [−1.748, 2.597] .385 251 Race Indo 173 Afro 82 Subjective wellbeing 24.815 6.688 21.122 7.337 [1.870, 5.516] 3.991** 253 Perceived social support 4.683 1.255 4.797 .934 [−.420, .193] −.730 253 Acculturative stress 46.064 27.748 49.927 24.001 [−10.888, 3.161] −1.083 253 Perceived discrimination 25.445 12.254 26.671 9.492 [−4.247, 1.796] −.799 253 Financial strain 1.081 .688 1.277 .727 [−.375, −.005] −2.027* 253 Depression 6.642 8.752 7.156 9.051 [−2.853, 1.819] −.436 253 Sampling Comm 59 MTurk 196 Subjective wellbeing 22.661 7.510 23.918 6.969 [−3.333, .818] −1.193 253 Perceived social support 4.591 .995 4.758 1.206 [−.507, .172] −.969 253 Acculturative stress 38.949 20.524 49.821 27.747 [−18.554, −3.191] −2.787** 253 Perceived discrimination 22.966 9.046 26.704 11.863 [−7.056, −.419] −2.218* 253 Financial strain 1.121 .679 1.156 .713 [−.238, .167] −.335 253 Depression 6.542 9.243 6.888 8.731 [−.345, 1.355] −.263 253 Immigrant Generation 1st 111 2nd 144 Subjective wellbeing 22.7771 6.902 24.333 7.198 [−3.3804, .137] −1.816 253 Perceived social support 4.881 1.105 4.595 1.191 [−.001, .573] 1.960* 244.5 Acculturative stress 46.568 24.957 47.875 27.898 [−7.939, 5.324] −.388 253 Perceived discrimination 25.135 10.659 26.3819 12.006 [−4.095, 1.599] −.863 253 Financial strain 1.138 .7282 1.156 .689 [−.194, .157] −.204 253 Depression 6.721 8.379 6.875 9.1986 [−2.356, 2.048] −.138 253 Clinically depressed CESDR ≥ 16 No 204 Yes 51 Subjective wellbeing 24.333 7.166 20.804 .502 1.379, 5.679] 3.233 253 Perceived social support 4.718 1.113 4.725 1.349 [−.365, .352] −.035 253 Acculturative stress 40.878 23.970 73.019 20.559 [−39.337, −24.947] −8.798 253 Perceived Discrimination 24.025 11.589 33.098 7.192 [−12.422, −5.725] −5.336 253 Financial strain 1.029 .699 1.624 .501 [−.799, −.389] −5.715 253 Depression 3.059 4.454 21.804 5.407 [−20.181, −17.379] −25.705 253 N = 255. Two-tailed significance P ≤ .01 = , p ≤ .05 = * Correlations among predictor and dependent variables demonstrated that sociological stressors were moderately correlated with depression: acculturative stress (r = 0.595, p ≤ 0.001), perceived discrimination (r = 0.465, p ≤ 0.001), and financial strain (r = 0.500, p ≤ 0.001). As expected, subjective wellbeing was negatively correlated with depression (r = −0.352, p ≤ 0.001). Perceived social support was not correlated with any other predictor variable. (see Table 3).Table 3 Correlations among predictor and dependent variables Variable 1 2 3 4 5 6 Subjective wellbeing 1 1 −.040 −.364 −.275 −.590 −.352 Perceived Social Support 2 1 .045 .063 −.022 .029 Acculturative Stress 3 1 .584 .567 .595 Perceived Discrimination 4 1 .440 .465 Financial strain 5 1 .500 Depression 6 1 Mean 23.628 56.537 47.306 25.839 1.148 12.467 Std. Error .444 .865 1.667 .716 .044 .329 SD 7.103 13.815 26.615 11.434 .704 5.255 Cronbach’s alpha .88 .69 .95 .94 .92 .91 N = 255. Two-tailed significance p ≤ .01 = *, p ≤ .05 = No asterisk indicates no significant correlation To further identify the contribution of each stressor variable to depression, hierarchical regression (Table 4), and path analysis were conducted. The initial model (Table 5) identified financial strain, acculturative stress and perceived discrimination as being significant predictors of depression F (5, 249) = 34.55, p ≤ 0.001 (R2adj = 0.398). Backward elimination regression (Table 5) confirmed that they accounted for the most variance in the model F(3, 251) = 57.34, p ≤ 0.001, (R2adj = 0.399).Table 4 Model summary of hierarchical regression Model R R square Adjusted R square Std. error of the estimate Change statistics R Square Change F Change df1 df2 Sig. F Change Acculturative stress .595a .354 .352 .43946 .354 138.910 1 253 < .001 Acculturative Stress + Financial strain .627b .393 .388 .42689 .039 16.118 1 252 < .001 Acculturative Stress + Financial strain + Perceived discrimination .637c .406 .399 .42317 .013 5.454 1 251 .020 aPredictors acculturative stress bAcculturative stress, financial strain cAcculturative stress, financial strain, perceived discrimination. Dependent variable: depression Table 5 Predictor variables contributing to depression in full model and optimized model Full model. F statistic: 34.55, df: 5 and 249, p-value: < 0.001, adjusted R2: 0.3978 Coefficient Coefficient Standardized coefficient t statistic p-value Constant −0.022 −6.510–17 −0.115 0.909 Financial strain 0.139 0.179 2.589 0.010 Subjective wellbeing −0.005 0.067 −1.104 0.271 Acculturative stress 0.008 0.387 5.826 < 0.001 Perceived discrimination 0.007 0.142 2.325 0.021 Perceived social support 0.002 0.004 0.09 0.928 Optimized model. F statistic: 57.34, df: 3 and 251, p-value: < 0.001, adjusted R2: 0.3996 Coefficient Coefficient Standardized coefficient t statistic p-value Constant −0.171 −6.310–17 −2.487 0.014 Financial strain −0.168 0.217 3.623 < 0.001 Acculturative stress 0.008 0.390 5.889 < 0.001 Perceived discrimination 0.007 0.142 2.338 0.020 Acculturative stress, perceived discrimination, and financial strain are the significant predictors of depression (optimized model results). Based on the standardized coefficients, the most significant predictor is acculturative stress, followed by financial strain, and perceived discrimination Further analysis (Table 6) identified which stressors mediated the relationship between subjective wellbeing and depression. The full regression model identified financial strain as the most significant F(4, 250) = 34.05, p ≤ 0.001, (R2adj = 0.342) mediator between subjective wellbeing and depression. This was confirmed with the optimized model F(1, 253) = 135.5, p ≤ 0.001, (R2adj = 0.346).Table 6 For subjective wellbeing, the full model, and the optimized model Full model. F statistic: 34.05, df: 4 and 250, p-value: < 0.001, adjusted R2: 0.3423 Coefficient Coefficient Standardized coefficient t statistic p-value Constant 32.167 −4.610–20 18.791 < 0.001 Financial strain −5.743 −0.570 −9.083 < 0.001 Acculturative stress −0.011 −0.040 −0.576 0.565 Perceived discrimination 0.002 0.003 0.040 0.968 Perceived social support −0.314 −0.051 −1.004 0.317 Optimized model. F statistic: 135.5, df: 1 and 253, p-value: < 0.001, adjusted R2: 0.3462 Coefficient Coefficient Standardized coefficient t statistic p-value Constant 30.458 −4.810–18 44.26 < 0.001 Financial strain −5.949 −0.591 −11.64 < 0.001 Financial strain is the only independent predictor of subjective wellbeing. While subjective wellbeing has significant correlations with acculturative stress, and perceived discrimination as well, their collinearity with financial strain renders them insignificant in the regression for path analysis Discussion West Indian Americans are an underrepresented immigrant community in the United States due to cultural and racial within-group differences that classify them within other immigrant groups i.e. Afro-Caribbean and Asian American. Their wellbeing is not well documented. Theories on the primary mental health stressors in American immigrant communities have changed over the decades, moving away from theories of cultural identities toward host society structural stressors. Intersectional theory stresses the relationship between host society structural stressors and immigrant community mental health declines i.e. socio-economic stress, host society climate, discrimination, enclave residency, and minoritization [5, 8, 9]. Following on intersectional theory, this exploratory study sought to identify individual sources of host society structural stressors that contribute to West Indian American depression. Specifically, the study weighed acculturative stress against financial strain and perceived discrimination as the primary mediators between subjective wellbeing and depression (Fig. 1). While acculturative stress was identified as the major contributor to depression, financial strain was the predominant mediator between subjective wellbeing and depression (Fig. 2).Fig. 2 Observed mediation and predictors. Bold arrows indicate statistically significant relationships Subjective Wellbeing Within our sample, subjective wellbeing scores were higher for Indo West Indians. This particular finding was not surprising as research by Williams et al. [2] and others have found that Afro-West Indians, especially men, were more likely than their Spanish-speaking counterparts to report less use of mental health resources and higher rates of mood disorders. Furthermore, non-Hispanic Black Americans, which include Afro-West Indians, have reported frequent experiences of discriminatory experiences [26, 28]. A recent study by Gigantesco et al. [54] found subjective wellbeing is negatively correlated with depression in a general population sample. Our findings agree with previous studies, that subjective wellbeing is a reliable measure of depression risk in West Indian Americans. Financial Strain Our findings are consistent with prior literature [33, 55, 56] identifying financial strain as an important contributing factor to mental health stress in immigrant populations, in addition to acculturative stress [5, 8, 9, 22, 23]. We found that financial strain is independent of acculturative stress in mediating the relationship between subjective wellbeing and depression in this population. This finding is supported by the reported median income of $39,441, which is in agreement with the overall reported West Indian American income [31, 57, 58]. Our study identified financial strain as an important factor to study in this population to identify risks to mental health. Acculturative Stress We found that fully anonymous respondents (MTurk) reported higher levels of acculturative stress and perceived discrimination. These data agree with literature indicating that anonymity results in more honest self-reports [59]. Studies on Afro-West Indians suggests that mental health in West Indians may still be taboo [12] and anonymity may have enhanced reporting of psychological distress. Acculturate stress was found to be a major predictor of depression in this sample, indicating that stressors from the process of acculturation may contribute to decreased mental health in West Indian Americans. Perceived Discrimination Our findings were consistent with prior studies—that experiences of discrimination are important predictors of psychiatric illness among U.S. immigrant populations [24–27]. Monolingual English-nativity may also enhance the perception of discrimination in West Indian Americans, adding to the psychological burden of such experiences [29, 60]. The finding that men were more likely to report experiences of discrimination may provide an explanation for previous reports that Afro-West Indian men are at higher risk for mood disorders than women [2]. Depressed Individuals Those who met clinical criteria for depression (CESD-R ≥ 16) reported lower subjective wellbeing and significantly higher rates of acculturative stress, perceived discrimination and financial strain. This supports the overall model that these stressors are predictors of depression and confirms prior findings that lower subjective wellbeing is indicative of increase risk of depression [54]. Effects of Gender, Race, and Immigrant Generation on Depression and Subjective Wellbeing Through the descriptive statistics and t-test for group differences (Table 2), we can conclude that there is no significant difference between groups of gender, race, and immigrant generation when looking at depression data. Further between-group exploration would expand the analysis into a large factor analysis and that is not within the scope of this work. Limitations of Study and Areas for Future Research There were significant limitations to our study that prevent us from generalizing these findings to the larger West Indian American community. An online survey was used to overcome the potential barrier of discussing mental health honestly [59] in a population reported not to use mental health supports despite high risk for mood disorders [12]. However, we found that an online survey excluded many community members who were not familiar with online survey participation and study participation. This resulted in a small sample and fewer members of the Afro-West Indian community than ideal. Perceived Social Support Our study failed to confirm perceived social support as a mediator of subjective wellbeing and depression [61]. Within our sample, perceived social support was not related to any measures of psychological stress or subjective wellbeing. Reasons for this could be related to the size of our sample which skewed toward second-generation immigrants who may experience acculturation pressure (inter-relationship acculturative stress). The contrast between acculturative stress and acculturation pressure is an important distinction when studying immigrant populations. Studies have identified that acculturation pressures are the primary component of acculturative stress that are not structurally embedded in host society. As a result, an area for further study would be to disentangle acculturation pressure and acculturative stress in West Indian Americans to determine whether acculturation pressure is related to the absence of a significant relationship between perceived social support and our model. Contribution to the Literature This is one of the first studies to identify predictors of depression in the West Indian American community. Despite their large presence in the New York metropolitan area and eastern U.S., they have been included in other demographics and have been statistically invisible as a cohesive group. Our study identifies consistencies between the Indo- and Afro- communities. Despite this, they are collectively at risk for mental health stress due to their English-nativity, acculturative stress, socio-economic status and discriminatory events. Our study highlights the significant impact of financial strain in this population prior to the COVID-19 pandemic. With the economic impact of COVID-19 on immigrant communities and the recent declaration of racism as a public health emergency in New York State, continued observation of this immigrant population is necessary to support their wellbeing. Acknowledgements The author would like to thank Prabuddha Bansal, Ph.D. for his invaluable support and guidance with the statistical analyses. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Wong EC Miles JN Prevalence and correlates of depression among new US immigrants J Immigr Minor Health 2014 16 3 422 428 10.1007/s10903-013-9781-0 23400525 2. Williams DR The mental health of Black Caribbean immigrants: results from the National Survey of American Life Am J Public Health 2007 97 1 52 59 10.2105/AJPH.2006.088211 17138909 3. NY Planning The newest New Yorkers: characteristics of the city’s foreign-born population 2013 New York Department of City Planning 4. Shields SA Gender: an intersectionality perspective Sex Roles 2008 59 5 301 311 10.1007/s11199-008-9501-8 5. Schulz AJ Mullings LE Gender, race, class, & health: Intersectional approaches 2006 Jossey-Bass/Wiley 6. Berry JW Phinney JS Sam DL Vedder P Immigrant youth: acculturation, identity, and adaptation Appl Psychol 2006 55 3 303 332 10.1111/j.1464-0597.2006.00256.x 7. Doucerain MM Second language social networks and communication-related acculturative stress: the role of interconnectedness Front Psychol 2015 4 6 1111 8. Cole ER Intersectionality and research in psychology Am Psychol 2009 64 3 170 10.1037/a0014564 19348518 9. Viruell-Fuentes EA Miranda PY Abdulrahim S More than culture: structural racism, intersectionality theory, and immigrant health Soc Sci Med 2012 75 12 2099 2106 10.1016/j.socscimed.2011.12.037 22386617 10. Helms JE Nicolas G Green CE Racism and ethnoviolence as trauma: enhancing professional and research training Traumatology 2012 18 1 65 74 10.1177/1534765610396728 11. Torres L Driscoll MW Voell M Discrimination, acculturation, acculturative stress, and Latino psychological distress: a moderated mediational model Cultur Divers Ethnic Minor Psychol 2012 18 1 17 10.1037/a0026710 22250895 12. Jackson JS Use of mental health services and subjective satisfaction with treatment among Black Caribbean immigrants: results from the National Survey of American Life Am J Public Health 2007 97 1 60 67 10.2105/AJPH.2006.088500 17138907 13. Figueiredo JM Distress, demoralization and psychopathology: diagnostic boundaries Eur J Psychiatry 2013 27 1 61 73 10.4321/S0213-61632013000100008 14. Taylor RJ Chatters LM Nguyen AW Religious participation and DSM IV major depressive disorder among Black Caribbeans in the United States J Immigr Minor Health 2013 15 5 903 909 10.1007/s10903-012-9693-4 22851131 15. Diener ED Emmons RA Larsen RJ Griffin S The satisfaction with life scale J Pers Assess 1985 49 1 71 75 10.1207/s15327752jpa4901_13 16367493 16. Lucas R Subjective well-being in psychology Oxford Handbook of Well-Being Public Policy 2016 21 1 403 423 17. De Neve JE Oswald AJ Estimating the influence of life satisfaction and positive affect on later income using sibling fixed effects Proc Natl Acad Sci 2012 109 49 19953 19958 10.1073/pnas.1211437109 23169627 18. Schiefer D Van der Noll J The essentials of social cohesion: a literature review Soc Indic Res 2017 132 2 579 603 10.1007/s11205-016-1314-5 19. Adler A Seligman ME Using wellbeing for public policy: theory, measurement, and recommendations Int J Wellbeing. 2016 6 1 1 35 10.5502/ijw.v6i1.429 20. Danner DD Snowdon DA Friesen WV Positive emotions in early life and longevity: findings from the nun study J Pers Soc Psychol 2001 80 5 804 10.1037/0022-3514.80.5.804 11374751 21. Strine TW The associations between depression, health-related quality of life, social support, life satisfaction, and disability in community-dwelling US adults J Nerv Ment Dis 2009 197 1 61 64 10.1097/NMD.0b013e3181924ad8 19155812 22. Rudmin F Constructs, measurements and models of acculturation and acculturative stress Int J Intercult Relat 2009 33 2 106 123 10.1016/j.ijintrel.2008.12.001 23. Rodriguez N Development of the multidimensional acculturative stress inventory for adults of Mexican origin Psychol Assess 2002 14 4 451 10.1037/1040-3590.14.4.451 12501570 24. Mesch GS Turjeman H Fishman G Perceived discrimination and the well-being of immigrant adolescents J Youth Adolesc 2008 37 5 592 604 10.1007/s10964-007-9210-6 25. Breslau J Risk for psychiatric disorder among immigrants and their US-born descendants: evidence from the national comorbidity survey-replication J Nerv Ment Dis 2007 195 3 189 10.1097/01.nmd.0000243779.35541.c6 17468677 26. Kessler RC Mickelson KD Williams DR The prevalence, distribution, and mental health correlates of perceived discrimination in the United States J Health Soc Behav 1999 1 208 230 10.2307/2676349 27. Mossakowski KN Coping with perceived discrimination: does ethnic identity protect mental health? J Health Soc Behav 2003 1 318 331 10.2307/1519782 28. Suárez-Orozco C Suárez-Orozco MM Todorova I Learning a new land 2008 Harvard University Press 29. Hunter CD The roles of shared racial fate and a sense of belonging with African Americans in Black immigrants’ race-related stress and depression J Black Psychol 2017 43 2 135 158 10.1177/0095798415627114 30. Min PG The attachments of New York City Caribbean Indian immigrants to Indian culture, Indian immigrants and India J Ethn Migr Stud 2013 39 10 1601 1616 10.1080/1369183X.2013.833688 31. Social P, Trends D. The rise of Asian Americans. Pew Social & Demographic Trends. 2012. https://www.immigrationresearch.org/system/files/PewResearch---Rise-of-Asian-Americans.pdf 32. Bisgaier J Rhodes KV Cumulative adverse financial circumstances: associations with patient health status and behaviors Health Soc Work 2011 36 2 129 137 10.1093/hsw/36.2.129 21661302 33. Peirce RS Financial stress, social support, and alcohol involvement: a longitudinal test of the buffering hypothesis in a general population survey Health Psychol 1996 15 1 38 10.1037/0278-6133.15.1.38 8788539 34. Puyat JH Exposure to deaths and dying and risks of burnout among long-term care staff: a cross-sectional survey Palliat Med 2019 33 6 717 720 10.1177/0269216319833248 30813836 35. Rudolph CW Perceived social support and work-family conflict: a comparison of Hispanic immigrants and non-immigrants Cross Cult Manage. 2014 10.1108/CCM-01-2013-0002 36. Almeida J Ethnicity and nativity status as determinants of perceived social support: testing the concept of familism Soc Sci Med 2009 68 10 1852 1858 10.1016/j.socscimed.2009.02.029 19303184 37. Kong F Zhao J You X Social support mediates the impact of emotional intelligence on mental distress and life satisfaction in Chinese young adults Personality Individ Differ 2012 53 4 513 517 10.1016/j.paid.2012.04.021 38. Berríos-Riquelme J Psychometric properties and factorial invariance of the satisfaction with life scale in Latino immigrants in Chile, Spain, and United States Terapia Psicológica 2021 39 2 199 218 10.4067/s0718-48082021000200199 39. Haber MG The relationship between self-reported received and perceived social support: a meta-analytic review Am J Community Psychol 2007 39 1 133 144 10.1007/s10464-007-9100-9 17308966 40. Crockett LJ Acculturative stress, social support, and coping: relations to psychological adjustment among Mexican American college students Cult Divers Ethnic Minor Psychol 2007 13 4 347 10.1037/1099-9809.13.4.347 41. Sirin SR The role of acculturative stress on mental health symptoms for immigrant adolescents: a longitudinal investigation Dev Psychol 2013 49 4 736 10.1037/a0028398 22563676 42. Williams DR Racial differences in physical and mental health: socio-economic status, stress and discrimination J Health Psychol 1997 2 3 335 351 10.1177/135910539700200305 22013026 43. Panter AT Everyday discrimination in a national sample of incoming law students J Divers Higher Educ 2008 1 2 67 10.1037/1938-8926.1.2.67 44. Forman TA Race, place, and discrimination Perspect Soc Problems. 1997 9 231 261 45. Northern JJ O'Brien WH Goetz PW The development, evaluation, and validation of a financial stress scale for undergraduate students J Coll Stud Dev 2010 51 1 79 92 10.1353/csd.0.0108 46. Brantley PJ Jones GN Boudreaux E Catz S Zalaquett CP Wood RJ Weekly stress inventory Evaluating stress: a book of resources 1997 Scarecrow Education 405 420 47. Eaton WW Muntaner C Smith C Tien A Ybarra M Maruish ME Center for epidemiologic studies depression scale: review and revision (CESD and CESD-R) The use of psychological testing for treatment planning and outcomes assessment 2004 3 Mahwah, NJ Lawrence Erlbaum 363 377 48. Radloff LS The CES-D scale: a self-report depression scale for research in the general population Appl Psychol Meas 1977 1 3 385 401 10.1177/014662167700100306 49. Hosler AS Kammer JR Cong X Everyday discrimination experience and depressive symptoms in urban Black, Guyanese, Hispanic, and White adults J Am Psychiatr Nurses Assoc 2019 25 6 445 452 10.1177/1078390318814620 30569835 50. Mena FJ Padilla AM Maldonado M Acculturative stress and specific coping strategies among immigrant and later generation college students Hisp J Behav Sci 1987 9 2 207 225 10.1177/07399863870092006 51. Pavot W Diener E The affective and cognitive context of self-reported measures of subjective well-being Soc Indic Res 1993 28 1 1 20 10.1007/BF01086714 52. Zimet GD The multidimensional scale of perceived social support J Pers Assess 1988 52 1 30 41 10.1207/s15327752jpa5201_2 53. Osman A The multidimensional scale of perceived social support: analyses of internal reliability, measurement invariance, and correlates across gender J Pers Assess 2014 96 1 103 112 10.1080/00223891.2013.838170 24090236 54. Gigantesco A The relationship between satisfaction with life and depression symptoms by gender Front Psych 2019 14 10 419 10.3389/fpsyt.2019.00419 55. Price RH Choi JN Vinokur AD Links in the chain of adversity following job loss: how financial strain and loss of personal control lead to depression, impaired functioning, and poor health J Occup Health Psychol 2002 7 4 302 10.1037/1076-8998.7.4.302 12396064 56. Gilman SE Socioeconomic status in childhood and the lifetime risk of major depression Int J Epidemiol 2002 31 2 359 367 10.1093/ije/31.2.359 11980797 57. US Census Bureau American factfinder 2004 US Department of Commerce Economics and Statistics Administration, US Census Bureau 58. Brown A Stepler R Statistical portrait of the foreign-born population in the United States 2015 Pew Research Center 59. Conrad FG Schober MF New frontiers in standardized survey interviewing Handbook Emergent Methods 2008 22 173 188 60. Medvedeva M Perceived discrimination and linguistic adaptation of adolescent children of immigrants J Youth Adolesc 2010 39 8 940 952 10.1007/s10964-009-9434-8 20596820 61. Siedlecki KL Salthouse TA Oishi S Jeswani S The relationship between social support and subjective well-being across age Soc Indic Res 2014 117 2 561 576 10.1007/s11205-013-0361-4 25045200	36456840	PMC9714766	NO-CC CODE	2022-12-03 23:20:12	no	J Immigr Minor Health. 2022 Dec 1;:1-11	utf-8	J Immigr Minor Health	2,022	10.1007/s10903-022-01434-5	oa_other
==== Front J Gen Intern Med J Gen Intern Med Journal of General Internal Medicine 0884-8734 1525-1497 Springer International Publishing Cham 7955 10.1007/s11606-022-07955-x Concise Research Report A Virtual-First Telehealth Treatment Model for Opioid Use Disorder http://orcid.org/0000-0002-7380-6203 Williams Arthur Robin [email protected] 12 Aronowitz Shoshana 23 Gallagher Ryan 2 Behar Emily 2 Gray Zack 2 Bisaga Adam 12 1 grid.239585.0 0000 0001 2285 2675 Department of Psychiatry, Columbia University Medical Center, New York, NY USA 2 Ophelia Health Inc., New York, NY USA 3 grid.25879.31 0000 0004 1936 8972 University of Pennsylvania School of Nursing, Philadelphia, PA USA 1 12 2022 13 9 3 2022 15 11 2022 © The Author(s), under exclusive licence to Society of General Internal Medicine 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 Death rates from opioid overdose continue to rise at an alarming pace. Many overdose victims die with untreated opioid use disorder (OUD), despite the availability of effective and lifesaving medications for opioid use disorder (MOUD). Significant barriers to treatment access include transportation, competing responsibilities, stigma, and related costs1. While millions of individuals receive buprenorphine in a given year2, the majority discontinue treatment within a few weeks or months3,4 forfeiting ongoing protective effects5. OUD treatment via virtual care platforms may help overcome these common barriers to retention in care3,4. Until recently, the 2008 Ryan Haight Act required an in-person assessment before prescribing any controlled substance, greatly limiting telehealth for OUD1. However, COVID-19-related reforms largely waived such requirements and hold promise for improving access to care, especially in underserved rural areas. These regulatory changes have attracted innovation among startups as well as incumbent health systems1. Given long-standing challenges in retaining patients in traditional office-based buprenorphine treatment3,4, research is needed to determine how virtual care outcomes can optimize retention. METHODS We analyzed data from a cohort of individuals with OUD treated at Ophelia, a virtual-first telehealth OUD treatment platform. Individuals in New York and Pennsylvania were recruited directly online (e.g., Facebook and Google ads). Medical visits and clinically indicated urine drug screens, organizational structure, custom EHR, and care coordination services were all built explicitly for remote care without requiring any in-person visits. In-network patients, predominantly Medicaid beneficiaries, used insurance. Patients out of network or uninsured paid $195 monthly to receive unlimited real-time, video-based clinical visits. Eligible patients (i.e., those with OUD not requiring a higher level of care) were prescribed buprenorphine at intake, and seen weekly during the stabilization phase, and then stepped down to monthly visits under a nurse care manager model. To investigate 180-day treatment retention, a minimum duration of pharmacotherapy for OUD endorsed by the National Quality Forum6, we analyzed a sample of consecutive new intakes from July 1, 2020, to April 15, 2021. Kaplan-Meier survival analyses determined retention, with discontinuation being defined as a 60+ day gap between clinical visits. Consistent with prior studies, care episodes of ≤ 7 days (9% of total episodes) were excluded3 as they often reflected patients ineligible or unwilling to initiate treatment. Geographic heat maps compared distribution of patients to the SAMHSA locator for buprenorphine x-waivered prescribers at the zip code level with eSpatial mapping technology. Patients’ home addresses were categorized under USDA Rural-Urban Commuting Area Codes, RUCA codes, with 1–3 denoting urban and 4–10 denoting rural locations.7 Secondary outcomes measuring adherence included the proportion of days covered (PDC) and medication possession ratio (MPR) of buprenorphine. The Western (WCG) IRB approved a waiver of consent for the study conducted under STROBE guidelines. RESULTS A total of 475 patients were included, 60.3% male, mean age 36.3 years (SD=7.1 years). Two-thirds (66.5%) self-reported race/ethnicity: 88.3% were white. The majority reported Medicaid coverage, consistent with prior studies.3,4 The 180-day retention was 69.1% (95%CI: 65.0–73.2%) (Fig. 1); 21.9% of patients resided in rural/small town areas reflecting much greater geographic variation than that of x-waivered prescribers (Fig. 2). 28% of urban patients were in a zip code without an x-waivered provider as were 31.96% of rural patients. Patients had a proportion of days covered of 0.96 and a medication possession ratio of 1.05 reflecting high adherence without evidence of stockpiling medication. Figure 1 The 180-day retention among patients with opioid use disorder in buprenorphine maintenance treatment provided by telehealth. Figure 2 Density distribution of x-waivered buprenorphine providers (per SAMHSA locator) vs patients enrolled via a virtual care platform in Pennsylvania and New York (n=475), 2020–2021 data. *Among 475 patients, 43.4% (206) resided in New York and 56.6% (269) in Pennsylvania. DISCUSSION The observed 180-day retention rate of 69.1% is superior to prior observational studies analyzing multi-state Medicaid (27.0%)3 and commercial insurance prescription claims (31.0%)4. Unlike requirements for in-person care, telehealth enables patients to access care that may not be available locally and attend visits with less interruption and more discretion, decreasing stigma. Expanding access to medication-based care and improving treatment retention are vital to address the worsening opioid overdose crisis. Technology-enabled telehealth platforms may be important tools for increasing access, retention, and patient satisfaction with evidence-based OUD care. Telehealth is not a single entity and further research is needed to determine best practices, especially for improving patient retention in OUD treatment with buprenorphine. Further research is also necessary to determine how telehealth interventions can best reach racially and ethnically diverse populations who are underserved by OUD treatment generally and for whom fatal overdose rates are currently rising the fastest. Funding Financial support for this work was provided by Ophelia Health, Inc. Declarations Conflict of interest The authors receive compensation in the form of equity, salary, consulting fees, and/or travel expenses from Ophelia Health, Inc., a telehealth provider for opioid use disorder. ARW also receives consulting fees from the National Quality Forum for work on measure development for the treatment of opioid use disorder. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Verma, S. Early impact of CMS expansion of Medicare telehealth during COVID-19. Health Affairs blog, 2020. 2. 2019 National Survey of Drug Use and Health (NSDUH) Releases \| CBHSQ Data. https://www.samhsa.gov/data/release/2019-national-survey-drug-use-and-health-nsduh-releases. Last accessed October 2021. 3. Samples H Williams AR Olfson M Crystal S Risk factors for premature discontinuation of buprenorphine treatment for opioid use disorders in a multi-state sample of Medicaid enrollees Journal Subst Abus Treat 2018 95 9 17 10.1016/j.jsat.2018.09.001 4. Morgan JR Schachkman BR Leff JA Linas BP Walley A Injectable naltrexone, oral naltrexone, and buprenorphine utilization and discontinuation among individuals treated for opioid use disorder in a United States commercially insured population J Subst Abus Treat 2018 85 90 96 10.1016/j.jsat.2017.07.001 5. Williams AR Samples H Crystal S Olfson M Retention on buprenorphine beyond six months and risk of acute care service utilization, opioid prescription use, and overdose Am J Psych 2019 177 2 117 124 10.1176/appi.ajp.2019.19060612 6. National Quality Forum. Behavioral Health 2016-2017 Final Report. 2017. Accessed 3/6/2022 at https://www.qualityforum.org/Publications/2017/08/Behavioral_Health_2016-2017_Final_Report.aspx 7. Economic Research Service, U.S. Department of Agriculture. Rural-Urban Commuting Area Codes: https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes.aspx last accessed August 20, 2022.	36456841	PMC9714767	NO-CC CODE	2022-12-03 23:20:13	no	J Gen Intern Med. 2022 Dec 1;:1-3	utf-8	J Gen Intern Med	2,022	10.1007/s11606-022-07955-x	oa_other
==== Front Monatsschr Kinderheilkd Monatsschr Kinderheilkd Monatsschrift Kinderheilkunde 0026-9298 1433-0474 Springer Medizin Heidelberg 1660 10.1007/s00112-022-01660-z Pädiatrie aktuell \| Für Sie gelesen Wie häufig ist Long-COVID wirklich? How frequent is long COVID really?Kerbl Reinhold [email protected] grid.508273.b Abteilung für Kinder und Jugendliche, LKH Hochsteiermark/Leoben, Vordernberger Str. 42, 8700 Leoben, Österreich Redaktion Reinhold Kerbl, Leoben Tim Niehues, Krefeld Peter Voitl, Wien 1 12 2022 12 3 11 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. ==== Body pmcOriginalpublikation Global Burden of Disease Long COVID Collaborators, Wulf Hanson S, Abbafati C et al (2022) Estimated global proportions of individuals with persistent fatigue, cognitive, and respiratory symptom clusters following symptomatic COVID-19 in 2020 and 2021. JAMA 328(16):1604–1615. 10.1001/jama.2022.18931 Hintergrund. Die Prävalenz von Long-COVID wird in verschiedenen Studien höchst unterschiedlich angegeben. Sie ist u. a. abhängig von der angewandten Definition, der Art der Studie, der untersuchten Fallzahl, aber auch der jeweiligen Infektionswelle (z. B. Delta vs. Omikron). Insbesondere in Fallserien ohne Kontrollgruppe werden teilweise sehr hohe Prävalenzzahlen berichtet; viele Studien berücksichtigen in ihrer Analyse auch das Vorliegen präexistenter Symptome nicht adäquat. Schließlich werden bis zu 300 verschiedene Symptome mit Long-COVID assoziiert und auch dadurch sehr hohe Prävalenzzahlen erzielt. Eine nunmehr im Journal of the American Medical Association (JAMA) publizierte Modellierungsstudie (Bayes-Metaregression) hat nun versucht, unter Anwendung der WHO-Definition und einheitlicher Kriterien die reale Prävalenz für Long-COVID zu erheben. Methodik. In die Metaregression eingeschlossen wurden 54 Studien aus 22 Ländern, die Gesamtzahl der inkludierten Patienten betrug 1,2 Mio. Für die Einstufung als Long-COVID wurde die WHO-Definition angewandt, die eine Symptomdauer von zumindest 3 Monaten voraussetzt. Berücksichtigt wurden nur Fälle mit vormals symptomatischem Verlauf, asymptomatische SARS-CoV-2-Infektionen wurden nicht eingeschlossen. Weiters wurden nur jene Fälle als „Long-COVID“ eingestuft, die zumindest einen der folgenden 3 Symptomenkomplexe zeigten: andauernde Müdigkeit mit Körperschmerz und Stimmungsschwankungen, kognitive Beeinträchtigung oder anhaltende respiratorische Probleme. Ergebnisse. Insgesamt erfüllten 6,2 % aller Patienten die Kriterien für Long-COVID. Vorherrschende Symptome waren respiratorische Probleme (bei 60,4 %), gefolgt von anhaltender Müdigkeit (51 %) und kognitiven Defiziten (35,4 %). Die Prävalenz war bei Personen nach dem 20. Lebensjahr signifikant höher als bei Personen unter 20 Jahren (Frauen 10,6 % vs. 5,4 %, Männer 5,4 % vs. 2,2 %). Die mittlere Symptomdauer betrug bei im Rahmen der Akuterkrankung hospitalisierten Patienten 9 Monate (Konfidenzintervall 7–12), bei vormals nichthospitalisierten Patienten 4 Monate (Konfidenzintervall 3,6–4,6). Kommentar Die Autoren verweisen auf mehrere Einschränkungen ihrer Modellierungsstudie. So waren die für die Analyse herangezogenen Einzelstudien sehr heterogen; für die „gepoolte“ Analyse mussten entsprechende Algorithmen angewandt werden. Eingeschlossen wurden nur im Rahmen ihrer Erkrankung vormals symptomatische Patienten, allerdings scheint Long-COVID bei vormals nicht symptomatischen Patienten selten aufzutreten. Als weitere Limitation wird angeführt, dass die Studie mit Jänner 2022 endete und somit keine/kaum Omikronfälle inkludiert sind. Für Letztere scheint die Prävalenz für „Long-COVID“ um 50–75 % niedriger zu sein als für frühere Virusvarianten. Dass die Symptomdauer bei vormals Hospitalisierten und insbesondere Intensivpatienten länger war als bei Nicht-Hospitalisierten, verwundert nicht wirklich. Die mittlere Symptomdauer von 4 bzw. 9 Monaten widerlegt aber für beide Gruppen – zumindest für die meisten Fälle – die in vielen Medien transportierte Behauptung des „jahrelangen Krankseins“. Die Autoren weisen auch darauf hin, dass die Symptomatik von „Long-COVID“ durchaus auch bei anderen Infektionserkrankungen vorkommt, als Beispiele nennen sie Influenza, SARS-CoV‑1, Ebola, EBV, Dengue, Q‑Fieber, Lyme-Borreliose und Giardiasis. Schließlich geht die Diskussion auch auf den Geschlechterunterschied und die deutlich höhere Prävalenz bei Frauen ein. Die Autoren spekulieren über mögliche Unterschiede der Immunantwort bei Männern und Frauen, die X‑chromosomal bedingt sein könnte. Diesbezüglich bleiben aber viele Fragen offen, auch was die zahlreichen bisher mit „Long-COVID“ assoziierten Symptome betrifft. Schlussbemerkung. Bekannterweise ist die Prävalenz von „Long-COVID“ im Kindes- und Jugendalter wesentlich niedriger als bei Erwachsenen, exakte Prävalenzzahlen dazu fehlen allerdings noch. Es wäre daher wünschenswert, dass eine derartige Metaanalyse auch mit spezieller Berücksichtigung von Kindern und Jugendlichen erfolgt, wie überhaupt zur Fragestellung „Long-COVID bei Kindern“ mehr Licht ins Dunkel gebracht werden sollte [1–3]. Interessenkonflikt R. Kerbl gibt an, dass kein Interessenkonflikt besteht. QR-Code scannen & Beitrag online lesen ==== Refs Literatur 1. https://pubmed.ncbi.nlm.nih.gov/35994282/. Zugegriffen: 21.11.2022 2. https://pubmed.ncbi.nlm.nih.gov/36194229/. Zugegriffen: 21.11.2022 3. https://www.mdpi.com/journal/children/special_issues/1JJ41U998K. Zugegriffen: 21.11.2022	36475055	PMC9714768	NO-CC CODE	2022-12-03 23:20:13	no	Monatsschr Kinderheilkd. 2022 Dec 1;:1-2	utf-8	Monatsschr Kinderheilkd	2,022	10.1007/s00112-022-01660-z	oa_other
==== Front J Sci Med Sport J Sci Med Sport Journal of Science and Medicine in Sport 1440-2440 1878-1861 Published by Elsevier Ltd on behalf of Sports Medicine Australia. S1440-2440(22)00473-X 10.1016/j.jsams.2022.11.002 Editorial Vaccination in athletes, affordable research tools for all and the opportunities arising from publicly available data Meyer Tim Editor in Chief 1 12 2022 12 2022 1 12 2022 25 12 949949 10 9 2022 15 11 2022 19 11 2022 © 2022 Published by Elsevier Ltd on behalf of Sports Medicine Australia. 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 pmcWhen the first vaccinations against SARS-CoV-2 infections were available in early 2021 (at that time still with the alpha variant being most prevalent) there was a run for getting the shots as quickly as possible because this implied at least some freedom from Covid-related restrictions - and for athletes the opportunity to restart regular competition. However, many countries installed prioritization schedules with young and healthy athletes rather at the end of the row. Only improved vaccine availability and the upcoming Olympic Games were responsible for some mitigation of these rules among athletes. Although vaccination efficacy was somewhat reduced when the delta variant of the virus occurred, the overall immunization situation of athletes did not change very much. Omicron altered the clinical efficacy and at the same time the discussions within the sporting community. Still the vaccine was highly effective to prevent severe courses of Covid-19 but its protection against acquisition of the disease was almost lost. Given that young athletes mostly experienced mild courses, their motivation to get booster shots became very limited mainly because their individual benefit decreased more strongly than for older and more vulnerable persons. This scenario sets the frame for the study from Krzywański et al3. who have investigated within a large athlete cohort from Poland if the individual time lost from training and competition was larger due to vaccination or due to acquired Covid-19. Their results are quite clear and in favour of the vaccination. It has, however, to be taken into account that their study period was before omicron (in fact, under high prevalence of the delta variant) which limits direct transfer to our present situation but still there is potential applicability for infections like influenza and future viruses... In addition, we are very happy to publish a study from a consortium involving a number of differently developed countries addressing physical activity in young children. The first author of this study comes from Malawi (current first affiliation: Strathclyde, UK) and was part of a larger research group. It is one of our main goals to support research from underrepresented countries (like Malawi). However, this has to be done without lowering the quality of papers which is a task not easy to achieve for a journal like ours. Over this year, we have already taken some action but still we have not arrived where we want to be. With the occurrence of more and more "transformative agreements" (facilitating open access to papers) between publishers and governments or university consortia the situation has already become worse for developing countries in terms of visibility for their research. This is because they cannot always afford the amount of money to be paid for such contracts. It will be the responsibility of both parties within transformative agreements to either include developing countries (governments, consortia) or to offer them acceptable open-access conditions (publishers). Besides all these considerations, the paper from Mwase-Vuma et al4. reports data from a broad and multi-national study (13 countries) which utilized accelerometry and compared it to simpler (cheaper) ways of assessing physical activity in 3-4 year old children. However, parent assessment of physical activity in their children (the cheap approach) failed to match the objective measurement sufficiently. Instead, there was a tendency for the parents to overestimate the activity of their offspring. Nevertheless, it is of particular importance to examine and validate affordable tools and methods for research in large populations. Otherwise, such studies (and not only their publication) will soon be limited to a small number of countries where technical devices are more easily available for researchers. The third paper to be highlighted in this issue illustrates a recent trend: the more extensive utilization of publicly available data (Charest et al.)1. With the given media coverage of highly professional sports and the increasing ability of researchers to extract respective information from existing databanks, the opportunity arises to address some research questions more conveniently than before. The authors from Canada and the US have used this to investigate more than 17,000 National Hockey League matches and analyze them with regard to the influence of travelling and time zone changes on match outcome in a convincing manner. Results confirm the widely held belief that the outcome of away matches is negatively influenced by the travelling distance (regardless of the direction). A similar example had already been highlighted in issue 4, 2022: the Hoenig et al2. study about injury epidemiology in soccer which - in contrast to the one in this issue - utilized individual data. This leads to more detailed analyses being possible within the data set but larger concerns around data quality from public sources. ==== Refs References Charest J. Cook J.D. Bender A.M. Associations between time zone changes, travel distance and performance: A retrospective analysis of 2013–2020 National Hockey League Data Journal of Science and Medicine in Sport 25 12 2022 1008 1016 Hoenig T. Edouard P. Krause M. Analysis of more than 20,000 injuries in European professional football by using a citizen-based approach: An opportunity for epidemiological research? J Sci Med Sport 25 4 2022 300 305 34916169 Krzywański J. Mikulski T. Krysztofiak H. Vaccine versus infection – COVID-19-related loss of training time in elite athletes Journal of Science and Medicine in Sport 25 12 2022 950 959 Mwase-Vuma T.W. Janssen X. Okely A.D. Validity of low-cost measures for global surveillance of physical activity in pre-school children: The SUNRISE validation study Journal of Science and Medicine in Sport 25 12 2022 1002 1007	36464483	PMC9714962	NO-CC CODE	2022-12-03 23:20:13	no	J Sci Med Sport. 2022 Dec 1; 25(12):949	utf-8	J Sci Med Sport	2,022	10.1016/j.jsams.2022.11.002	oa_other
==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(22)02325-X 10.1016/S0140-6736(22)02325-X Correspondence Integrated respiratory surveillance after the COVID-19 pandemic Elson William a Zambon Maria b de Lusignan Simon a a Nuffield Department of Primary Care Health Sciences, University of Oxford, Oxford OX2 6EE, UK b Reference Microbiology, UK Health Security Agency, London, UK 1 12 2022 3-9 December 2022 1 12 2022 400 10367 19241925 © 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. ==== Body pmcWe welcome Thedi Ziegler and colleagues’ Comment on the WHO Global Influenza Surveillance and Response System and believe its proposals could strengthen global sentinel surveillance.1 We respond to this Comment from the Oxford-Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC), one of Europe's oldest sentinel networks.2 The RSC is a collaboration between the University of Oxford, RCGP, and the UK Health Security Agency. The UK Health Security Agency's respiratory virus reference laboratory has provided data to the Global Influenza Surveillance and Response System and its predecessors since 1950. The RSC has grown substantially throughout the pandemic and, therefore, we are well placed to comment on the strategic and operational implications of the proposals.3 We support the principle of year-round integrated respiratory surveillance and the broadening of laboratory testing to other respiratory viruses. Furthermore, we agree that measuring disease severity will support effective public health decision making. However, we should not overlook the importance of quality clinical data reporting and sufficient representative sampling to evaluate performance of diverse influenza and COVID-19 vaccine types across primary and secondary care networks. Additionally, longitudinal data enable prompt detection of changes in disease incidence, and routine surveillance with serology can also provide insights into heterogeneous population immunity.4 The RSC's rapid expansion and improved digital maturity through the pandemic have shown that surveillance systems can be a test bed for innovation.3 The integration of data from near-patient diagnostics into the RSC surveillance system's dataset and the evaluation of interventions (eg, antiviral therapies and new vaccines) are examples of such innovation.5 Growth, however, comes at a cost, and if improvements are to be sustained and built upon, necessary investment in personnel and resources is essential. MZ is the chair of the International Society for influenza and other Respiratory Virus Diseases (ISIRV), a charitable organisation. MZ is a member of the government advisory groups Scientific Advisory Group for Emergencies, New and Emerging Respiratory Virus Threats Advisory Group, and the Joint Committee on Vaccination and Immunisation. Through his university, SdeL has received vaccination-related grants from AstraZeneca, GlaxoSmithKline, Sanofi, Seqirus, and Takeda. SdeL has been member of advisory boards for AstraZeneca, Sanofi, and Seqirus. We declares no competing interests. ==== Refs References 1 Ziegler T Moen A Zhang W Cox NJ Global influenza surveillance and response system: 70 years of responding to the expected and preparing for the unexpected Lancet 400 2022 981 982 36154679 2 de Lusignan S Correa A Smith GE RCGP research and surveillance centre: 50 years’ surveillance of influenza, infections, and respiratory conditions Br J Gen Pract 67 2017 440 441 28963401 3 Leston M Elson WH Watson C Representativeness, vaccination uptake and COVID clinical outcomes 2020-21 in the UK's Oxford-RCGP Research and Surveillance Network: cohort profile JPHS 2022 published online Nov 2. 10.2196/preprints.39141 (preprint). 4 Whitaker H Tsang RSM Button E Sociodemographic disparities in COVID-19 seroprevalence across England in the Oxford RCGP primary care sentinel network J Infect 84 2022 814 824 35405169 5 de Lusignan S Hoang U Liyanage H Using point of care testing to estimate influenza vaccine effectiveness in the English primary care sentinel surveillance network PLoS One 16 2021 e0248123 33705452	36463902	PMC9714972	NO-CC CODE	2022-12-03 23:20:13	no	Lancet. 2022 Dec 1 3-9 December; 400(10367):1924-1925	utf-8	Lancet	2,022	10.1016/S0140-6736(22)02325-X	oa_other
==== Front Lancet Respir Med Lancet Respir Med The Lancet. Respiratory Medicine 2213-2600 2213-2619 Published by Elsevier Ltd. S2213-2600(22)00495-7 10.1016/S2213-2600(22)00495-7 Spotlight Patient collaboration in COVID-19 research: translating ideas to reality Kirby Tony 1 12 2022 1 12 2022 © 2022 Published by Elsevier Ltd. 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. ==== Body pmcInvolving patients in all stages of health data research is a noble aim; however, there are not too many examples of it being successfully put into practice—particularly not in real-time pandemic surveillance. In November, 2020, the Early Pandemic Evaluation and Enhanced Surveillance of COVID-19 (EAVE II) team created a unique collaboration, inviting a Public Advisory Group (PAG) of 15 patient and public contributors to work alongside university and Public Health Scotland staff to provide vital perspectives of the public on all areas of their COVID-19 research. “Our PAG has played a crucial role in shaping key aspects of our research into COVID-19, asking questions, and often providing answers to issues that our researchers or analysts might never have thought of”, explains Dr Lana Woolford of the University of Edinburgh, Patient and Public Involvement (PPI) Coordinator for EAVE II (figure ), to The Lancet Respiratory Medicine. “The relevance of research to the public and how it is communicated is absolutely essential. Some members of the advisory group have had COVID-19, some are shielding, some live with multiple health conditions or disabilities—and their unique insights have helped develop EAVE II and subsequent studies in a way that would not have been possible otherwise.”Figure EAVE II patient and public involvement co-lead David Weatherill (far left) and Lana Woolford (second from right) with University of Edinburgh Press and communications staff © 2022 University of Edinburgh 2022 The original EAVE project took place more than a decade ago during the H1N1 swine flu pandemic that gripped the world. Back then, Scottish researchers used data from almost 250 000 patients to track that pandemic. EAVE II has massively scaled up this research to track COVID-19 outcomes in almost the entire Scottish population (5·4 million people). The project has identified population groups most at risk from SARS-CoV-2 infection and COVID-19 hospital admission and death, and monitors the effectiveness and safety of population-wide vaccination. “Working meaningfully alongside patients and the public is at the core of our work”, says Sir Aziz Sheikh, principal investigator for EAVE II, and Professor of Primary Care Research and Development at the University of Edinburgh, UK. “Bridging the gap between the pandemic experiences of people in Scotland and the UK, and the data we analyse, helps to ensure that the evidence we offer is relevant, understandable, acceptable, and accountable to real people in a real-world setting”, adds Sheikh, whose other roles include Director of the Usher Institute and Director of the Asthma UK Centre for Applied Research. Woolford explains that EAVE II originally advertised for nine members of the PAG, expanding to 15 in October, 2021, to further increase their diversity and capacity for input. They meet online every 4–6 weeks, carrying out more work by email in between. Staff engage with individual members to ensure their access needs are met, including providing software, stationery, large-print documents, or the ability to provide input by phone. Getting used to meeting online has also improved collaboration and has allowed people from across the UK to take part. The group includes two co-leads—Sandra Jayacodi (based in London) and David Weatherill (based in Dundee)—who also represent the PAG in each steering group meeting. David is a retired chartered engineer and company director who has had more than 10 years of PPI experience, including as a lay member of the Royal College of Anaesthetists and Royal College of Surgeons. As a patient, he has had three separate cancers that have greatly impacted him, creating a concern to “give something back” to the UK's National Health Service. He explains: “My role as a PPI Lead provides me with the opportunity to contribute to the work of medical researchers and statisticians. I feel privileged to act as a public voice representing the concerns of people. Monitoring the work of researchers at all stages of their work from initial concept to delivery of research papers is amazing...we, the PAG, are welcomed by all of the researchers and are considered as full members of their team.” Jayacodi left her profession as a solicitor due to ill health and has since used her lived experience to help improve care services, including with the Imperial Biomedical Research Centre Public Advisory Panel, and as an improvement research fellow with the Applied Research Collaboration Collaborative Leadership in northwest London. Sandra explains: “Being part of an elite team of researchers advising the UK Government on COVID-19 has been a privilege. EAVE II is a study I felt has had truly meaningful input from us. I am particularly proud of how PAG members pushed the team to get permissions for ethnicity data.” Despite the best intentions of the EAVE II investigators, the rate at which the COVID-19 pandemic disrupted the UK meant that the full PAG was not properly in place until November, 2020, and thus was not able to play as key a part in some of the collaboration's earlier projects as had been hoped. However, working together with researchers, Public Health Scotland, and other relevant agencies, they are now full participants in key COVID-19 research. “One of the first things the PAG felt it was important to work on was our communications activities”, explains Woolford. “This included a focus on our public-facing materials, such as research summaries, infographics, and animations, advising on rewriting where appropriate and making them more easily understandable.” The PAG subsequently moved into the other parts of the research cycle, including grant development (including shaping the project design and writing the lay research summary); analysis design, including protocols for research on COVID-19 vaccines in children and young people, and new COVID-19 treatments; and assistance throughout the actual project or study. This has included project steering, involving networks of contributors with relevant lived experience, and providing public perspectives throughout the research cycle. Where the PAG felt it did not have sufficient experience or representation for a particular issue, it sought help from outside, such as from the Long Covid Scotland Action Group for issues around long COVID. The group also provides analysis and reflection on EAVE II's projects, in lay terms. For one study, PAG member Eve Smyth gave her perspective in a profile in The Lancet Respiratory Medicine, commenting that children with asthma should be prioritised for COVID-19 vaccination, including younger children aged 5–11 years who had just become eligible for vaccination. Data access and quality have also been improved as a direct result of the PAG's work. After they raised concerns about the availability and quality of ethnicity data in Scottish health records, they successfully lobbied the Scottish Government and Public Health Scotland alongside researchers, enabling the EAVE II team to gain access to a wider repository of ethnicity records, which were not previously available to them. This also led to PPI representation from EAVE II on a collaboration study led by the University of Glasgow, looking at the impact of ethnicity and socioeconomic status on COVID-19 outcomes in Scotland. A more recent example of effective PAG input is in a study by EAVE II assessing so-called breakthrough infections, in which people who are fully vaccinated still get COVID-19 and can become very ill. The study results discussed the risk of serious COVID-19 disease outcomes with omicron infection following both two vaccine doses and a subsequent booster dose. It found patients taking immunosuppressants were at 6-times higher risk of severe COVID-19 disease than those not taking them, despite being fully vaccinated. One of the vaccine breakthrough project's PPI co-leads Lynn Laidlaw, in an online interview with Sheikh, described how the study shows “how people like me who are immunosuppressed and live with multiple long-term conditions are still high risk after taking a booster vaccine…it enables me to make informed decisions about my risk and benefits.” Sheikh also explained how the study information was disseminated to key government departments and the media. The second round of the study assesses the latest Moderna bivalent vaccine that includes additional protection against omicron variants. “One of our patient representatives pointed out that these very same patients, vulnerable to breakthrough infections, are also those eligible for COVID-19 antiviral and monoclonal antibody treatments”, explains Woolford. “We realised that the new study would be significantly more beneficial for patients if we also looked at vaccine effectiveness in the context of receiving or being eligible for COVID-19 treatments—the study design was modified to reflect this.” EAVE II is not the only successful collaboration with strong involvement from the public, but the team hope their model can be copied as both researchers and the PAG have learned and contributed so much to the experience. Another project called CO-CONNECT, has worked on standardising data across UK institutions nationwide so that, while they will continue with their own data systems, they will also have health data that can be collected and compiled into other studies. The Public User Group, an advisory group made up of members from the general public, has advised on storing and representing patient data. And in the International Severe Acute Respiratory Infection Consortium 4C-R project, data from England and Scotland are being used to predict hospital readmission for people with COVID-19, with EAVE II's PAG being asked to contribute. Funding for EAVE II's PAG is currently guaranteed until March, 2023, and Woolford says the team hope funding will be extended since there are many unanswered questions they want to continue working on. This includes further research into COVID-19 vaccine immune responses, and studies on long COVID that can help provide policy makers with desperately needed information. “Even diagnosis of long COVID is difficult, with it having similarities to other chronic conditions”, explains Woolford. “Another area we are collaborating on is pregnancy and COVID-19, again a very complex and delicate topic.” She adds: “The EAVE II PAG has a really strong sense that we can do more, including branching into other respiratory infections like influenza and chronic conditions, such as asthma, COPD, and even diabetes.” Businesswoman video conferencing at laptop and computer.© 2022 Paul Bradbury/Caia Image/Science Photo Library 2022	36463910	PMC9714973	NO-CC CODE	2022-12-03 23:20:13	no	Lancet Respir Med. 2022 Dec 1; doi: 10.1016/S2213-2600(22)00495-7	utf-8	Lancet Respir Med	2,022	10.1016/S2213-2600(22)00495-7	oa_other
==== Front Lancet Diabetes Endocrinol Lancet Diabetes Endocrinol The Lancet. Diabetes & Endocrinology 2213-8587 2213-8595 The Lancet, Diabetes & Endocrinology S2213-8587(22)00324-2 10.1016/S2213-8587(22)00324-2 Comment Diabetes after SARS-CoV-2 infection Al-Aly Ziyad ab a Clinical Epidemiology Center, Research and Development Service, VA Saint Louis Health Care System, Saint Louis, MO 63106, USA b Institute for Public Health, Washington University in Saint Louis, Saint Louis, MO, USA 1 12 2022 1 12 2022 Published by Elsevier Ltd. 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. ==== Body pmcThe COVID-19 global pandemic represents a unique opportunity to better understand the post-viral and post-infectious condition. More than 2 years into the pandemic, a large body of evidence makes it clear that infection with SARS-CoV-2 can lead to post-acute sequelae in the pulmonary and broad array of extrapulmonary organ systems—collectively referred to by the umbrella term of long COVID.1, 2 Evidence also suggests that the myriad clinical abnormalities of long COVID might extend to new onset diabetes.3 A US Center of Disease Control (CDC) analysis of a large electronic health-care database of 353 164 adults with COVID-19 and 1 640 776 controls with no evidence of infection, suggested that people with COVID-19 had an increased risk of new onset type 1 diabetes and type 2 diabetes.4 Furthermore, a German cohort study of 35 865 people with COVID-19 showed higher risk of newly diagnosed type 2 diabetes than an equal number of matched controls with acute upper respiratory tract infections.5 I also investigated whether there was an association between new-onset diabetes and COVID-19. Together with my colleague Yan Xie, we used data from the US Department of Veterans Affairs to characterise the risk and 12-month burden of diabetes in 181 280 people with SARS-CoV-2 infection versus two control groups: 4 118 441 contemporary controls who were enrolled during the same time but did not get infected with SARS-CoV-2 and 4 286 911 historical controls from before the pandemic.3 Our findings suggested that compared with both the contemporary and historical controls, people with SARS-CoV-2 had increased risk of incident diabetes and incident use of antihyperglycemic therapy in the post-acute phase.3 Interestingly, compared with non-infected controls, the increased risk of diabetes (>99% was type 2 diabetes) was evident even in people who had very low baseline (pre-COVID-19) risk of diabetes according to traditional risk factors including age, race, sex, body-mass index, hypertension, and hyperlipidaemia). Among people with COVID-19, the risk of diabetes increased in a graded fashion according to baseline risk of diabetes (ie, the traditional baseline characteristics that predict the risk of developing diabetes in an individual). The main limitation of our study was that the participants were mostly White males. Our findings, along with the findings of others, suggest the possible coexistence of two pathways that should be investigated in mechanistic studies: (1) COVID-19 leading to de novo disease in people who might have otherwise not developed diabetes, and (2) COVID-19 as an amplifier of baseline risks and accelerant of disease development. Due to paucity of studies, the evidence base of new onset diabetes following COVID-19 in children is far less well developed. In an analysis of two large health-care databases, researchers from the US CDC suggested that, compared with non-infected controls, people younger than 18 years with SARS-CoV-2 infection had increased risk of a diabetes diagnosis in the post-acute phase of COVID-19; they also showed that COVID-19 was associated with higher risk of diabetes than pre-pandemic acute respiratory infections and that non–SARS-CoV-2 respiratory infection was not associated with an increased risk for diabetes.6 However, this study did not differentiate between type 1 and type 2 diabetes. The evidence for increased risk of diabetes after SARS-CoV-2 infection is not universally consistent. A Scottish study in people younger than 35 years documented a 20% increase in the incidence of type 1 diabetes during the pandemic in general, and increased risk of type 1 diabetes within, but not beyond, the first 30 days of SARS-CoV-2 infection.7 Another study of 428 650 people (median age 35 years) with COVID-19 and matched controls showed a net increase in incidence of diabetes in the first 4 weeks after COVID-19, which remained elevated from 5 to 12 weeks but not from 13 to 52 weeks.8 Most of these studies on COVID-19 and diabetes were conducted before vaccines were available and when reinfections were uncommon. However, recent evidence from a US Department of Veterans Affairs study involving more than 13 million individuals suggests that compared with non-infected controls, both unvaccinated and vaccinated individuals with SARS-CoV-2 infection are at increased risk of diabetes and that the risk of diabetes in the post-acute phase of COVID-19 was not significantly different in people who had a SARS-CoV-2 infection after vaccination than unvaccinated individuals.9 A new study of more than 5 million people (also from the US Department of Veterans Affairs) suggests that reinfections with SARS-CoV-2 (compared with no reinfection) could contribute additional risks of acute and post-acute sequelae including increased risk of diabetes in both phases of the disease.10 A major methodological challenge in studying the post-acute and long-term health effects of SARS-CoV-2 infection—and a main reason for the discordance in evidence—is how to best disentangle the causal effects of the infection itself from other changes that might relate to both exposure and outcomes. For example, increased health-care use following SARS-CoV-2 infection and changes due to the pandemic itself (without SARS-CoV-2 infection) including effects of lockdowns, social isolation, loss of employment, and other factors that might have differentially affected people with SARS-CoV-2 infection might influence the risk of health outcomes (including diabetes). Although, for obvious ethical reasons, randomly exposing people to SARS-CoV-2 versus placebo is not possible, leveraging large-scale observational data and advances in causal inference methodologies to emulate a target trial are indeed feasible and should be actively pursued. The target trial emulation approach might be especially helpful to approximate—by design—a matched comparison between people with COVID-19 and non-infected controls, and estimate the causal effects of COVID-19 exposure. A trial emulation approach would first specify the causal question and lay out the protocol components of the ideal randomised trial that—if conducted—would randomise exposure and answer the causal question. This step would be then followed by specification of the emulation strategy including specification of the target population, eligibility (inclusion and exclusion) criteria, follow up, outcome, causal estimate, and a detailed analysis plan to estimate the causal contrast of interest. Although conceptualisation of an infection as a treatment in a randomised controlled trial might be perceived as unusual, the target trial emulation approach will further elevate the scientific rigor of large epidemiological analyses and enhance the ability to infer the causal long-term health effects of SARS-CoV-2 infection. Additionally, prospective controlled studies with detailed assessment of pre-COVID-19 health status and protocolised longitudinal health assessments are also useful to characterise the health trajectories of people with SARS-CoV-2 infection. Although, these studies might be less powered because they generally include far fewer participants than large observational studies. Robust research agendas to better understand long COVID (and all its components including the increased risk of diabetes), prevent it, and treat it are urgently needed. Several pressing strategic considerations and research questions will need to be answered in the near future (panel ).Panel Strategic considerations and urgent research priorities to address knowledge gaps in long COVID and more broadly infection-associated chronic illnesses Strategic considerations for governments to address the challenge of long COVID and to prepare for future pandemics • Prioritise research funding of post-viral and infection-associated chronic illnesses • Investment in real-world data systems that integrate a broad array of data sources and leverage advanced methodologies in causal inference is needed to address in near real-time major knowledge gaps and to devise public health policies • Surveillance systems for infectious diseases must incorporate longitudinal surveillance with appropriate controls to monitor for development of post-acute and chronic sequelae Urgent research priorities • Experimental studies to elucidate the biological mechanisms of diabetes following SARS-CoV-2 infection and other viral infections • Large well-powered prospective cohort studies with carefully curated controls, adaptive design (to dynamically adapt to the changes in the pandemic) and robust assessment of health status at regular time intervals after infection • Studies integrating both protocolised prospective collection of health information and routinely collected health data (eg, from wearable health trackers, electronic health records, and other data sources) • Studies evaluating genetic, environmental, and other susceptibility factors of post-viral disease (including risk of diabetes after SARS-CoV-2 infection) • Studies with both historical controls (pre-pandemic era) and contemporary (pandemic era) controls that disentangle the effects of the pandemic (effects of lockdowns and behavioural, environmental, and other changes) from those of SARS-CoV-2 infection on the epidemiology of diabetes • Long-term comparative analyses of the risks of diabetes following SARS-CoV-2 versus other viral infections (eg, seasonal influenza) would help contextualise the risk of diabetes following SARS-CoV-2 within the broader post-viral condition • Longitudinal studies to understand health trajectories and outcomes of people with diabetes following COVID-19, including response to treatment, health resource use, and downstream health outcomes • Studies looking at whether antivirals or other therapeutics during the acute or post-acute phase of COVID-19 reduce the risk of diabetes (or other post-acute sequelae) • Longer term studies to determine if diabetes and other cardiometabolic sequelae in people with COVID-19 might remit with time or whether they morph into chronic conditions • Studies evaluating the effect of SARS-CoV-2 variants (and subvariants) and new vaccines and boosters as well as the effect of repeated infections on the epidemiology on post-acute sequelae including diabetes • The effect of new onset diabetes (and other chronic diseases) on health systems, the economy, and society at large The broad implications of SARS-CoV-2 infection on human health are becoming increasingly clear. Before the pandemic, the global burden of diabetes was high and rising; the possible increased incidence of diabetes due to the pandemic could further compound the already staggering pre-pandemic burden. In turn, this could lead to substantial ramifications on health systems, health-care costs, life expectancy and economic indicators such as employment and labour participation. More broadly, the multifaceted long-term consequences of SARS-CoV-2 infection (including the risk of diabetes) should be reflected in the global discussion about non-communicable diseases. Long after the pandemic ends (and we must admit that it has not yet ended), millions of people around the world will still bear its scars. Chronic conditions, including diabetes, require lifelong care and can affect people's lives, livelihood, the economy, and societal wellbeing. A silverlining of this pandemic is the opportunity to more broadly understand the post-viral condition, which has been marginalised and understudied for more than a century. Long COVID, including the possible burden of diabetes, must be further investigated, understood and considered in every health-care and health policy decision we make now and going forward. Young woman suffering from negative thoughts on white copy space background. Medicines, syringe in long hair. Overthinking during coronavirus outbreak, lockdown. Flat cartoon vector illustration.© 2022 Shutterstock 2022 I declare no competing interests. ==== Refs References 1 Al-Aly Z Xie Y Bowe B High-dimensional characterization of post-acute sequelae of COVID-19 Nature 594 2021 259 264 33887749 2 Xie Y Bowe B Al-Aly Z Burdens of post-acute sequelae of COVID-19 by severity of acute infection, demographics and health status Nat Commun 12 2021 6571 34772922 3 Xie Y Al-Aly Z Risks and burdens of incident diabetes in long COVID: a cohort study Lancet Diabetes Endocrinol 10 2022 311 321 35325624 4 Bull-Otterson LBS Baca S Saydah S Post–COVID conditions among adult COVID-19 survivors aged 18–64 and ≥65 years — United States, March 2020–November 2021 MMWR Morb Mortal Wkly Rep 71 2022 713 717 5 Rathmann W Kuss O Kostev K Incidence of newly diagnosed diabetes after Covid-19 Diabetologia 65 2022 949 954 35292829 6 Barrett CE Koyama AK Alvarez P Risk for newly diagnosed diabetes >30 days after SARS-CoV-2 infection among persons aged <18 years - United States, March 1, 2020-June 28, 2021 MMWR Morb Mortal Wkly Rep 71 2022 59 65 35025851 7 McKeigue PM McGurnaghan S Blackbourn L Relation of Incident Type 1 Diabetes to Recent COVID-19 Infection: Cohort Study Using e-Health Record Linkage in Scotland Diabetes Care 2022 published online July 26 10.2337/dc22-0385 8 Rezel-Potts E Douiri A Sun X Chowienczyk PJ Shah AM Gulliford MC Cardiometabolic outcomes up to 12 months after COVID-19 infection. A matched cohort study in the UK PLoS Med 19 2022 e1004052 35853019 9 Al-Aly Z Bowe B Xie Y Long COVID after breakthrough SARS-CoV-2 infection Nat Med 28 2022 1461 1467 35614233 10 Bowe B Xie Y Al-Aly Z Acute and postacute sequelae associated with SARS-CoV-2 reinfection Nat Med 28 2022 1461 1467 35614233	36463908	PMC9714974	NO-CC CODE	2022-12-15 23:18:07	no	Lancet Diabetes Endocrinol. 2023 Jan 1; 11(1):11-13	utf-8	Lancet Diabetes Endocrinol	2,022	10.1016/S2213-8587(22)00324-2	oa_other
==== Front Lancet Infect Dis Lancet Infect Dis The Lancet. Infectious Diseases 1473-3099 1474-4457 Elsevier Ltd. S1473-3099(22)00763-0 10.1016/S1473-3099(22)00763-0 Correspondence Decline of RSV-specific antibodies during the COVID-19 pandemic den Hartog Gerco a van Kasteren Puck B a Schepp Rutger M a Teirlinck Anne C a van der Klis Fiona R M a van Binnendijk Robert S a a Center for Infectious Disease Control, National Institute for Public Health and the Environment, 3721 Bilthoven, Netherlands 1 12 2022 1 12 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. ==== Body pmcHospitalisations due to respiratory syncytial virus (RSV) infections largely decreased after social distancing measures were introduced to control the COVID-19 pandemic. Lifting these measures resulted in out-of-season RSV activity, sometimes exceeding the incidence of hospitalisations observed in regular seasons.1, 2, 3 Declining immunity due to reduced exposure to the virus may contribute to this altered epidemiology.1, 4, 5 Bardsley and colleagues1 showed that the combination of laboratory, clinical, and syndromic data capture the impact of RSV activity, yet did not provide insight into the proposed decline in immunity. To investigate this effect, we analysed sera of 558 randomly selected participants of a prospective nationwide study in the Netherlands for changes in IgG antibody concentrations to the RSV post-fusion F protein.6 Participants were 1–89 years of age (mean age 48 years [SD 20·7]). 236 (42·3%) were male; 245 (43·9%) were female; and 77 (13·8%) were other, missed the question, or did not disclose this information. Samples of the same people were collected in June, 2020 (timepoint 1, several months after the introduction of social restrictions), February, 2021 (timepoint 2, approximately 1 year after the end of the last typical RSV season), and June, 2021 (timepoint 3, the month when social restrictive measures were lifted in the Netherlands and the out-of-season RSV epidemic started; figure 1A ). Concentrations were log10 transformed for all statistical analyses. The repeated-measures general linear model (SPSS version 28; IBM, Armonk, NY, USA) was used to compare antibody concentrations between sampling timepoints of the same subjects and between age groups. p values, adjusted to Bonferroni and Benjamini-Hochberg procedures, are reported for timepoints and age-group differences respectively. Student's t test was used to compare the 2020 IgG concentrations of the people who showed at least a twofold increase in RSV-specific IgG from 2020 to 2021 (n=9) and those who did not (n=549).Figure Changes in IgG concentrations to the post-fusion F protein from 2020 to 2021, relative to notification of RSV infections (A) Weekly RSV surveillance by the Dutch Working Group for Clinical Virology of the Dutch Society for Medical Microbiology and National Institute for Public Health and the Environment (blue line), and sampling periods in June, 2020, February, 2021, and June, 2021 (in grey). (B) Median IgG concentrations in the different age groups of all 558 individuals. General linear models for repeated measurements were used on log10-transformed data to test the decay of antibodies over time (p<0·001), for all age groups except age 31–40 years, and differences between age groups (p<0·001). (C) Individuals (n=9) showing at least a twofold increase in RSV-specific IgG from 2020 to 2021. Each of the individuals that showed at least a twofold increase are represented by a different coloured line. (D) Stratification of people showing at least a twofold increase in IgG, which is indicative of exposure to the virus, and those who did not. The difference in antibody concentrations between the two groups in 2020 was assessed with student's t test on log10-transformed concentrations. The boxplot shows medians and quartiles. AU=arbitrary units. RSV=respiratory syncytial virus. *Outliers. Post-fusion F IgG antibody concentrations declined from 2020 to 2021 (p<0·001) and increased with age (p<0·001; figure 1B). The decrease was greatest for the 1-year interval between timepoints 1 and 3 (p<0·001) when compared with the decrease between timepoints 1 and 2 (p<0·001) and between timepoints 2 and 3 (p=0·182). The decrease in antibodies was significant in all age groups, except for participants aged 31–40 years. Across the 3 timepoints, the age group of 71 years and older had higher antibody concentrations than participants aged 1–10 years (p=0·019), 21–30 years (p<0·001), 31–40 years (p=0·021), 41–50 years (p<0·001), and 51–60 years (p=0·034). In our analysis, we did not find evidence of differences in decay rates between age groups. We found 9 individuals (1·6%) with antibody boosting of at least two-fold during this period, indicative of exposure to the virus (figure 1C). These individuals were all adults of at least 30 years of age, and since two adults showed elevated IgG before the increase in clinical reports of RSV infections, these findings might indicate that circulation initiated in the adult population. On average, these individuals had lower IgG concentrations in 2020 (p=0·028) than those not showing a rise in IgG concentrations (figure 1D). These data support the assumption that RSV-specific antibody concentrations declined during the COVID-19 pandemic in all age groups and are in line with a previous report showing decay of antibodies to RSV.5 We do not have data on RSV-specific antibody kinetics in our cohort before the pandemic and there are relatively large variations between individuals, so the effect on susceptibility to RSV is not clear yet. Antibodies to the F protein, especially in pre-fusion confirmation, have an important role in the neutralisation of RSV and were previously shown to correlate well with virus neutralisation.7 However, the degree to which virus neutralisation is affected and the exact correlation with immune protection are yet to be determined.8 Following this preliminary analysis, additional timepoints, including follow-up samples, are being investigated to support and extend these findings. In conclusion, monitoring changes in antibody concentrations could identify populations susceptible to RSV infection. The study was funded by the Dutch Ministry of Health, Welfare, and Sports. The funder had no role in the generation of the data or writing of the manuscript. ACT received funds from the Respiratory Syncytial Virus Consortium in Europe and Preparing for RSV Immunisation and Surveillance in Europe consortium for grants and travel costs. All other authors declare no competing interests. We thank all study participants, the Dutch Working Group on Clinical Virology from the Dutch Society for Clinical Microbiology, and all participating laboratories for providing the virological data from the weekly laboratory virological report. ==== Refs References 1 Bardsley M Morbey RA Hughes HE Epidemiology of respiratory syncytial virus in children younger than 5 years in England during the COVID-19 pandemic, measured by laboratory, clinical, and syndromic surveillance: a retrospective observational study Lancet Infect Dis 2022 published online Sept 2 10.1016/S1473-3099(22)00525-4 2 Delestrain C Danis K Hau I Impact of COVID-19 social distancing on viral infection in France: a delayed outbreak of RSV Pediatr Pulmonol 56 2021 3669 3673 34473914 3 Eden J-S Sikazwe C Xie R Off-season RSV epidemics in Australia after easing of COVID-19 restrictions Nat Commun 13 2022 2884 35610217 4 Billard M-N Bont LJ Quantifying the RSV immunity debt following COVID-19: a public health matter Lancet Infect Dis 2022 published online Sept 2 10.1016/S1473-3099(22)00544-8 5 Reicherz F Xu RY Abu-Raya B Waning Immunity against respiratory syncytial virus during the COVID-19 pandemic J Infect Dis 2022 published online May 7 10.1093/infdis/jiac192 6 Berbers G Mollema L van der Klis F den Hartog G Schepp R Antibody responses to respiratory syncytial virus: a cross-sectional serosurveillance study in the Dutch population focusing on infants younger than 2 years J Infect Dis 224 2020 269 278 7 Schepp RM de Haan CAM Wilkins D Development and standardization of a high-throughput multiplex immunoassay for the simultaneous quantification of specific antibodies to five respiratory syncytial virus proteins mSphere 4 2019 e00236 e00319 31019002 8 Kulkarni PS Hurwitz JL Simões EAF Piedra PA Establishing correlates of protection for vaccine development: considerations for the respiratory syncytial virus vaccine field Viral Immunol 31 2018 195 203 29336703	36463892	PMC9714975	NO-CC CODE	2022-12-03 23:20:13	no	Lancet Infect Dis. 2022 Dec 1; doi: 10.1016/S1473-3099(22)00763-0	utf-8	Lancet Infect Dis	2,022	10.1016/S1473-3099(22)00763-0	oa_other
==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(22)02396-0 10.1016/S0140-6736(22)02396-0 Comment Monkeypox virus infection in women and non-binary people: uncommon or neglected? Rodriguez-Morales Alfonso J abc Amer Fatma A de a Grupo de Investigación Biomedicina, Faculty of Medicine, Fundación Universitaria Autónoma de las Américas—Institución Universitaria Vision de las Americas, Pereira 660003, Colombia b Master of Clinical Epidemiology and Biostatistics, Universidad Científica del Sur, Lima, Perú c Gilbert and Rose-Marie Chagoury School of Medicine, Lebanese American University, Beirut, Lebanon d Medical Microbiology and Immunology Department, Faculty of Medicine, Zagazig University, Zagazig, Egypt e Viral Infection Working Group, International Society for Antimicrobial Chemotherapy, Zagazig, Egypt 1 12 2022 3-9 December 2022 1 12 2022 400 10367 19031905 © 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. ==== Body pmcIn 2022, amid the COVID-19 pandemic, monkeypox re-emerged globally as a new and additional infectious threat for countries outside Africa, and it was declared a Public Health Emergency of International Concern by WHO.1, 2 During this multicountry monkeypox outbreak, which has comprised more than 80 488 cases to Nov 21, 2022 according to the US Centers for Disease Control and Prevention, new epidemiological and clinical features have been observed, including genital primary presentation, an association with sexual activity, and a predominance among men, particularly adult men who have sex with men.3, 4 Although epidemiological data and various studies have shown that the disease occurs primarily in men,5 monkeypox can also affect other groups, including children,6 older populations, and women.7, 8 In The Lancet,9 John P Thornhill and colleagues report a series of 136 cases of monkeypox virus infection in women and non-binary individuals from 15 countries, showing its clinical importance. 62 individuals in the case series were transgender (trans) women, 69 were cisgender (cis) women, and five were non-binary individuals assigned female at birth (the latter two groups were pooled to form a group of 74 people assigned female at birth for the purpose of comparison). The median age in the whole cohort was 34 years (IQR 28–40; range 19–84), meaning that most individuals were in age groups likely to be sexually active. 43 (69%) trans women and 65 (88%) cis women and non-binary people were heterosexual or had sex with men. In addition to the presence of previously known clinical features of rash (124 [93%] of 134 individuals), anogenital lesions (95 [74%] of 129), and vesiculopustular rash (105 [87%] of 121) in the cohort, monkeypox viral DNA was detected in the vaginal fluid of all tested individuals (n=14). Previous studies have identified viral DNA in semen in 78% of cases.10 Although the implications of such detections are still to be defined, these findings are essential in the understanding and future investigation of potential sexual transmission of this disease. As shown in Thornhill and colleagues’ study,9 prominent genital and mucosal features of the disease, which have been a defining feature of the global outbreak in men,4 are also present in cis and trans women and non-binary individuals (mucosal lesions involving the vagina, anus, oropharynx, or eye occurred in 65 [55%] of 119 individuals with available data), with fewer lesions per individual (median 10 [IQR 5–24; range 1–200]) than previously described; in previous studies, skin lesion severity scores were severe (100–250) or very severe (>250).9, 11 However, a clinical pattern of serious and profuse lesions in the vagina was observed. The absolute number and proportion of monkeypox cases among individuals could be increased by multiple factors, especially different sexual practices. In the current case series, 45 (73%) of 62 trans women and nine (12%) of 74 cis women and non-binary individuals had multiple male sexual partners. The inclusion of trans women in this study could help to close a knowledge gap regarding this under-studied population.12, 13 Although some case reports of trans women have been published, no studies were previously available. Collecting data from this population is especially important because of the higher rates of HIV and other sexually transmitted infections in trans women.2, 9, 13, 14 Co-infections of this type could influence the acquisition and clinical course of monkeypox virus infection, especially in people who are immunosuppressed, and in some cases could lead to fatal outcomes (42 deaths have so far been reported in 12 countries where monkeypox was not previously endemic as of Nov 21, 2022, according to the US Centers for Disease Control and Prevention, with evidence that HIV or other STI co-infection was an important factor in some of these deaths).14 The current case series showed a high prevalence of HIV among infected individuals and particularly trans women (31 [50%], compared with six [8%] cis women and non-binary individuals). Trans women also frequently face barriers to accessing health care and social support, among other related issues. Furthermore, 34 (55%) trans women in this international case series reported engaging in sex work,9 suggesting that this group might have a high level of precarity and vulnerability, which might include factors such as homelessness, injection drug use, migrant status (three [5%] of trans women were migrants), or having multiple sexual partners, which could compound the level of stigma they face.9, 15 More studies are needed to understand the differences in risk, transmission, and clinical consequences of monkeypox virus infection among different populations. Crucially, concerted efforts are needed to determine a comprehensive approach for differentiated integral case management that improves early detection and treatment, when required, especially among people at high risk. In terms of clinical practice, findings from Thornhill and colleagues’ case series could help to expand preparedness and increase the knowledge available to physicians and health-care workers in sexual health clinics, gynaecology and obstetrics (3% of cis women and non-binary individuals were pregnant at the time of infection), and female and reproductive medicine, among other related areas, to improve diagnosis, treatment, and support. During the current monkeypox outbreak, it is important to consider this infection among differential diagnoses, especially given that 25 (34%) of 74 cis women and non-binary individuals in Thornhill and colleagues’ case series were misdiagnosed before monkeypox diagnosis. Such misdiagnosis might be more common in regions where there are still few or no confirmed cases of monkeypox, such in as the Middle East, where additional stigma and discrimination against people with monkeypox, especially the LGBTQ+ population,15 might exist in health services as well as society more broadly. Equitable access to care, treatment, removal of stigma, and prevention for monkeypox is crucial at this time while the world is learning about this re-emerging threat. Finally, prioritisation should consider discussing vaccination and other preventive measures for women and non-binary individuals. Future studies should consider gender differences in transmission routes, viral loads in different sample types, risk of severity, and therapeutic and vaccination responses. female gp with patient We declare no competing interests. ==== Refs References 1 Sah R Reda A Lashin BI Mohanty A Abdelaal A Rodriguez-Morales AJ Public health emergencies of international concern in the 21st century Ann Med Surg 81 2022 104417 2 Farahat RA Sah R El-Sakka AA Human monkeypox disease (MPX) Infez Med 30 2022 372 391 36148174 3 Adler H Gould S Hine P Clinical features and management of human monkeypox: a retrospective observational study in the UK Lancet Infect Dis 22 2022 1153 1162 35623380 4 Thornhill JP Barkati S Walmsley S Monkeypox virus infection in humans across 16 countries—April–June 2022 N Engl J Med 387 2022 679 691 35866746 5 DeWitt ME Polk C Williamson J Global monkeypox case hospitalisation rates: a rapid systematic review and meta-analysis EClinicalMedicine 54 2022 101710 36345526 6 Hennessee I Shelus V McArdle CE Epidemiologic and clinical features of children and adolescents aged <18 years with monkeypox—United States, May 17–September 24, 2022 MMWR Morb Mortal Wkly Rep 71 2022 1407 1411 36331124 7 Zayat N Huang S Wafai J Philadelphia M Monkeypox virus infection in 22-year-old woman after sexual intercourse, New York, USA Emerg Infect Dis 2022 published online Nov 10. 10.3201/eid2901.221662 8 Bruno G Fabrizio C Rodano L Buccoliero GB Monkeypox in a 71-year-old woman J Med Virol 2022 published online July 13. 10.1002/jmv.27993 9 Thornhill JP Palich R Ghosn J Human monkeypox virus infection in women and non-binary individuals during the 2022 outbreaks: a global case series Lancet 400 2022 1943 1965 10 Reda A Abdelaal A Brakat AM Monkeypox viral detection in semen specimens of confirmed cases: a systematic review and meta-analysis J Med Virol 2022 published online Oct 22. 10.1002/jmv.28250 11 Whitehouse ER Bonwitt J Hughes CM Clinical and epidemiological findings from enhanced monkeypox surveillance in Tshuapa province, Democratic Republic of the Congo during 2011–2015 J Infect Dis 223 2021 1870 1878 33728469 12 Silva MST Jalil EM Torres TS Monkeypox and transgender women: the need for a global initiative Travel Med Infect Dis 50 2022 102479 36257591 13 Gandrakota N Lee H Nwosu O Kulshreshtha A Monkeypox coinfection with neurosyphilis in a transgender with HIV in Atlanta, USA Travel Med Infect Dis 50 2022 102454 36126913 14 Sah R Mohanty A Abdelaal A Reda A Rodriguez-Morales AJ Henao-Martinez AF First monkeypox deaths outside Africa: no room for complacency Ther Adv Infect Dis 9 2022 20499361221124027 15 Sah R Mohanty A Reda A Padhi BK Rodriguez-Morales AJ Stigma during monkeypox outbreak Front Public Health 10 2022 1023519 36203672	36463895	PMC9714976	NO-CC CODE	2022-12-15 23:16:11	no	Lancet. 2022 Dec 1 3-9 December; 400(10367):1903-1905	utf-8	Lancet	2,022	10.1016/S0140-6736(22)02396-0	oa_other
==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(22)02414-X 10.1016/S0140-6736(22)02414-X Department of Error Department of Error 1 12 2022 3-9 December 2022 1 12 2022 400 10367 19261926 © 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd 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. ==== Body pmcMitjà O, Ogoina D, Titanji BK, et al. Monkeypox. Lancet 2022; published online Nov 17. https://doi.org/10.1016/S0140-6736(22)02075-X—In this Seminar, the section on Treatment in the unanswered research questions panel should read “Should early initiation of treatment and an extended duration of treatment be recommended for individuals who are highly immunocompromised (eg, HIV with a CD4 count <200)?” This correction has been made to the online version as of Dec 1, 2022, and will be made to the printed version.	36463905	PMC9714977	NO-CC CODE	2022-12-15 23:16:11	no	Lancet. 2022 Dec 1 3-9 December; 400(10367):1926	utf-8	Lancet	2,022	10.1016/S0140-6736(22)02414-X	oa_other
==== Front Brachytherapy Brachytherapy Brachytherapy 1538-4721 1873-1449 Published by Elsevier Inc. S1538-4721(22)00310-5 10.1016/j.brachy.2022.09.149 Article PO43 Presentation Time: 4:45 PM Reduced Duration Intracavitary Brachytherapy in Cervix Malignancies: A Comparative Analysis from Single Institute Chaudhary Shwetima MBBS,MD Goel Varshu MBBS,MD Pareek Vibhay MBBS,DNB Abiramasundari V MBBS,DNB Raut Sagar MBBS,MD Amritt Adhar MBBS,MD Amariyil Adhila MBBS Ravi Ashwin MBBS Samala Sai Kumar MBBS Ghosh Adrija MBBS Dagar Abhilash MBBS Sharma Aman MBBS,MD Mallick Supriya MBBS,MD Patil Pritee P MBBS,MD Sharma Dayanand MBBS,MD Radiation Oncology, All India Institute of Medical Sciences, New Delhi, India 2 12 2022 November-December 2022 2 12 2022 21 6 S93S94 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. Purpose The on-going pandemic has impacted the use of anesthesia and the operating room frequency thereby affecting the brachytherapy treatment in various institutions due to the COVID-19 protocols. This has led to single applications of Intracavitary brachytherapy (ICRT) being used to deliver entire treatment boost in cervix cancer. We present our dosimetric and early clinical outcomes comparing traditional weekly three-fractions ICRT with single application/ two-applications ICRT Materials and Methods In this retrospective analysis conducted in our department, a total of 39 cases, treated between January 2021 to January 2022 were evaluated for the study. Of these, 15 cases were treated with the traditional once a week applicator insertion for 3 fractions and 24 cases underwent lesser application - 20 cases underwent 2 insertions and 4 cases single insertion (all receiving total 3 fractions of 7Gy each). The dosimetric parameters were compared including CTV D90 and D95 along with rectum, sigmoid and bladder D2cc, 1cc and 0.1cc respectively. The acute toxicity assessment was done using the RTOG scale. The follow-up was undertaken as per the institutional protocol and Mann-Whitney U-test were applied to compare the cohorts. Results With a median follow-up of 6 months, the median CTV was D90%: 81.2 vs. 80.9 Gy and the median CTV volume was 44.3 vs 42.9 cc respectively. The 0.1 cm3 and 2 cm3 to bladder, rectum, and sigmoid were 105.6 vs 104.2 Gy and 85.5 vs 85.9Gy, 89.4Gy vs 88.7Gy and 69.1 vs 67.8Gy, and 84.7 vs 84.1Gy and 71.7 vs 69.9Gy, respectively suggesting no significant difference in the dosimetric outcomes with the two forms of applications. The less than three applications had a shorter overall treatment time with median OTT of 43 days vs. 55 days (p = 0.02). On completion of treatment and 6 months follow-up, local control was achieved in all patients. There was no significant difference in the acute toxicities in terms of cystitis and proctitis in both forms of the application. Conclusion The single application/ twice application ICRT procedure showed similar outcomes as the traditional three-week duration treatment in terms of dosimetric outcomes and acute toxicities and ultimately leading to shortened overall treatment time. It also helped reduce the anesthesia burden and various resources associated with the procedure. ==== Body pmc	0	PMC9715012	NO-CC CODE	2022-12-03 23:20:14	no	Brachytherapy. 2022 Dec 2 November-December; 21(6):S93-S94	utf-8	Brachytherapy	2,022	10.1016/j.brachy.2022.09.149	oa_other
==== Front Brachytherapy Brachytherapy Brachytherapy 1538-4721 1873-1449 Published by Elsevier Inc. S1538-4721(22)00307-5 10.1016/j.brachy.2022.09.146 Article PO40 Presentation Time: 4:45 PM Definitive Management of Cervical Cancer Patients at an Urban Institution During the COVID-19 Pandemic - Brachytherapy Treatment During the Surge Lymberis Stella C. MD Lee Sarah S. MD Boyd Leslie MD Hacker Kari E. MD Salame Ghadir MD Pothuri Bhavana MD Schiff Peter B. MD, PhD Radiation Oncology, NYU Langone Medical Center, NYC, NY, USA 2 12 2022 November-December 2022 2 12 2022 21 6 S92S92 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. Purpose Locally advanced cervical cancer was defined by an international consensus panel as a high priority malignancy during the COVID-19 pandemic, recommending prompt initiation of definitive treatment and completion of treatment (PMID 32563593). The objective of this study was to study the clinical outcomes of patients (pts) with cervical cancer treated with definitive chemoradiation (CRT) and brachytherapy (BT) at our institution in 2019 (pre-COVID) and in 2020 (peri-COVID). Materials and Methods This was a retrospective cohort study of pts with FIGO Stage IB2-IVA cervical cancer at our institutions from 1/1/2019 to 12/31/2020. Pts received CRT followed by intracavitary brachytherapy (IC) with two operative insertions one week apart, or interstitial (IS) BT with one operative insertion. BT treatment was planned using image-guided CT or MR delineation. Pre-COVID was defined by initiation of CRT in 1/2019-12/2019, and peri-COVID was defined by initiation in 1/2020-10/2020. Process changes peri-COVID included limited on-site staff (e.g., minimal OR staff, no trainees, remote physics team), universal implementation of COVID-19 testing prior to surgery, and CT instead of MR-delineation based treatment. Outcomes of interest were time to treatment initiation and completion and differences in treatment planning modality or dosimetry. Fisher's exact and Mann Whitney U tests were used with significance p<0.05. Results Thirty-one pts were included, with 18 patients undergoing treatment pre-COVID and 13 peri-COVID. The median age at diagnosis pre-COVID was 57.7 (range 23-77) and for peri-COVID, 45.5 (range 28-62, p=0.06). There were no differences in non-English speaking pts (44% vs 59%, p=0.71) or uninsured pts (11% vs 33%, p=0.184) between the two cohorts. Median time to initiation of treatment from biopsy diagnosis was 52 days (range 13-209) in 2019 and for peri-COVID, 55.5 (range 20-173, p=0.71). During COVID, four pts had delayed initiation to treatment >100 days: two related to fertility, and one due to fear of COVID-19. For this pt, tumor size progressed from 2.3 cm to 4.2 cm maximal dimension. One pt treated in 2020 tested positive following treatment and did not require hospital admission. All pts except one completed CRT with RT: 25 pts pelvic RT (45 Gy), 3 pelvic and para-aortic RT (45 Gy with 57.5 Gy concomitant boost to nodes), 8 pts pelvic RT (45Gy) with sequential parametrial boost (50.4-59.4 Gy) using IMRT with no dose differences between pre and peri-COVID (Table 1). No pts required treatment breaks and the median overall treatment time was 50 days (range 31-85) in 2019 vs 50 days (range 43-63) in 2020 (p=0.710). Conclusions Despite the significant burden of the COVID-19 pandemic on our health care system, all cervical cancer pts receiving CRT met standard of care including CRT and BT within the recommended time frame with no significant differences in dosimetric treatment parameters pre- and peri-COVID. Delays in treatment initiation of treatment initiation were seen in 30% of pts in the peri-COVID period, suggesting that patients may have had increased barriers to access care. More follow-up is needed to determine how the Covid pandemic impacted cervical cancer outcome measures. ==== Body pmc	0	PMC9715013	NO-CC CODE	2022-12-03 23:20:14	no	Brachytherapy. 2022 Dec 2 November-December; 21(6):S92	utf-8	Brachytherapy	2,022	10.1016/j.brachy.2022.09.146	oa_other
==== Front J Allergy Clin Immunol J Allergy Clin Immunol The Journal of Allergy and Clinical Immunology 0091-6749 1097-6825 Published by Elsevier Inc. on behalf of the American Academy of Allergy, Asthma & Immunology. S0091-6749(22)01618-9 10.1016/j.jaci.2022.11.013 Article FUNCTIONALLY IMPAIRED ANTIBODY RESPONSE TO BNT162B2 BOOSTER VACCINATION IN CVID IgG RESPONDERS Sauerwein Kai M.T. MSc 123 Geier Christoph B. MD 1 Stemberger Roman F. Mag. 1 Rossmanith Raphael MSc 1 Akyaman Hüseyin BSc 1 Illes Peter MD 4 Fischer Michael B. MD Prof. 35 Eibl Martha M. MD Prof. 12 Walter Jolan E. MD Prof. 67 Wolf Hermann M. MD Prof. 18∗ 1 Immunology Outpatient Clinic, Vienna, Austria 2 Biomedizinische Forschung & Bio-Produkte AG, Vienna, Austria 3 Department for Biomedical Research, Center of Experimental Medicine, Danube University Krems, Krems an der Donau, Austria 4 USF Health Department of Pediatrics, Division of Allergy/Immunology, Children´s Research Institute, St. Petersburg, FL, USA 5 Clinic for Blood Group Serology and Transfusion Medicine, Medical University of Vienna, Vienna, Austria 6 Division of Allergy and Immunology, Department of Pediatrics, Morsani College of Medicine, University of South Florida, Tampa, FL, USA 7 Division of Allergy/Immunology, Department of Pediatrics, Johns Hopkins All Children's Hospital, St. Petersburg, FL, USA 8 Sigmund Freud Private University - Medical School, Vienna, Austria ∗ Corresponding author: Hermann M. Wolf, MD Immunology Outpatient Clinic, Schwarzspanierstraße 15 1090 Vienna, Austria, Tel: +43-1-4031450, Fax: +43-1-4051046 2 12 2022 2 12 2022 28 7 2022 7 10 2022 4 11 2022 © 2022 Published by Elsevier Inc. on behalf of the American Academy of Allergy, Asthma & Immunology. 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 While previous studies described the production of IgG-antibodies in a subgroup of CVID-patients following mRNA-vaccinations with bnt162b2 SARS-CoV2 (CVID responders), the functionality of these antibodies in terms of avidity as measured by the dissociation rate constant (kdis) and the antibody response to booster immunization has not been studied. Objective In CVID responders and healthy individuals the avidity of anti-SARS-CoV-2 serum-antibodies and their neutralization capacity as measured by surrogate virus neutralizing antibodies were analyzed in addition to IgG-, IgM- and IgA-antibody levels and the response of circulating follicular T-helper cells after a third vaccination with BNT162b2 SARS-CoV2 mRNA-vaccine. Methods Binding IgG, IgA and IgM serum levels were analyzed by ELISA in CVID-patients responding to the primary vaccination (CVID responders, n=10) and healthy controls (n=41). The binding-avidity of anti-spike antibodies was investigated using biolayer interferometry in combination with biotin-labelled receptor-binding-domain (RBD) of SARS-CoV2 spike-protein and streptavidin-labelled sensors. Antigen-specific recall T-cell responses were assessed by measuring activation-induced markers by flow cytometry. Results After the third vaccination with BNT162b2 IgG-, IgM and IgA-antibody levels, sVNT levels and antibody avidity were lower in CVID responders as compared to healthy. In contrast αSpike-avidity was comparable in CVID responders and healthy individuals following primary vaccination. Follicular T-helper cell response to booster vaccination in CVID-responders was significantly reduced when compared to healthy individuals. Conclusion Impaired affinity-maturation during booster-response provides new insight into CVID pathophysiology. KEY WORDS BNT162b2 booster vaccination CVID antibody avidity cTfh biolayer-interferometry ABBREVIATIONS CVID, Common variable immunodeficiency BNT162b2, SARS-CoV-2 mRNA vaccine - BNT162b2 αSpike, Anti-SARS-CoV-2 spike protein sVNT, surrogate virus neutralizing antibodies kdis, dissociation rate constant CVID R, CVID Responders, CVID patients, that responded to primary BNT162b2 vaccination with an IgG-antibody serum concentration at or above 33 RE/ml (three times the detection limit) S, seconds HC, healthy individual(s) cTfh, circulating (peripheral blood) follicular T-helper cells ==== Body pmcCONFLICT OF INTEREST STATEMENT Authors KMTS and MME were employed by the company Biomedizinische Forschung & Bio-Produkte AG that had no role in the design of this study or during its execution, and was not involved in the analyses, interpretation of the data and decision to submit the present manuscript. All other authors have nothing to disclose. CLINICAL IMPLICATIONS Future studies should address whether defective antibody maturation as demonstrated by analysis of antibody kinetics in CVID is associated with susceptibility to infection in other immunodeficient patients. CAPSULE SUMMARY In this study a defect in antibody maturation following BNT162b2 booster-vaccination of CVID patients was associated with defective circulating follicular T-helper cell responses.	36463978	PMC9715258	NO-CC CODE	2022-12-08 23:16:17	no	J Allergy Clin Immunol. 2022 Dec 2; doi: 10.1016/j.jaci.2022.11.013	utf-8	J Allergy Clin Immunol	2,022	10.1016/j.jaci.2022.11.013	oa_other
==== Front Int J Obstet Anesth Int J Obstet Anesth International Journal of Obstetric Anesthesia 0959-289X 1532-3374 The Author(s). Published by Elsevier Ltd. S0959-289X(22)00328-4 10.1016/j.ijoa.2022.103613 103613 Short Report COVID-19 infection and maternal morbidity in critical care units in Scotland: a national cohort study McPeake J. ab⁎ Blayney M.C. cd Stewart N.I. e Kaye C.T. f Seem R.C. d Hall R. d Martin C. d Paton M. d Wise A. g1 Puxty K. hi1 Lone N.I. cg1 on behalf of the Scottish Intensive Care Society Audit Group a Healthcare Improvement Scotland, Scotland, UK b The Healthcare Improvement Studies Institute, University of Cambridge, UK c Usher Institute, University of Edinburgh, UK d Public Health Scotland, UK e NHS Forth Valley, Scotland, UK f NHS Grampian, Scotland, UK g NHS Lothian, Scotland, UK h NHS Greater Glasgow and Clyde, UK i University of Glasgow, School of Medicine, Dentistry and Nursing, Scotland, UK ⁎ Corresponding author at: J. McPeake, 50 West Nile Street, Glasgow, G1 2NP, UK 1 AW, KP and NL contributed equally to this manuscript 2 12 2022 2 12 2022 1036136 7 2022 16 11 2022 22 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. Background Previous research has shown that, in comparison with non-pregnant women of reproductive age, pregnant women with COVID-19 are more likely to be admitted to critical care, receive invasive ventilation, and die. At present there are limited data in relation to outcomes and healthcare utilisation following hospital discharge of pregnant and recently pregnant women admitted to critical care. Methods A national cohort study of pregnant and recently pregnant women who were admitted to critical care in Scotland with confirmed or suspected COVID-19. We examined hospital outcomes as well as hospital re-admission rates. Results Between March 2020 and March 2022, 75 pregnant or recently pregnant women with laboratory-confirmed COVID-19 were admitted to 24 Intensive Care Units across Scotland. Almost two thirds (n=49, 65%) were from the most deprived socio-economic areas. Complete 90-day acute hospital re-admission data were available for 74 (99%) patients. Nine (12%) women required an emergency non-obstetric hospital re-admission within 90 days. Less than 5% of the cohort had received any form of vaccination. Conclusions This national cohort study has demonstrated that pregnant or recently pregnant women admitted to critical care with COVID-19 were more likely to reside in areas of socio-economic deprivation, and fewer than 5% of the cohort had received any form of vaccination. More targeted public health campaigning across the socio-economic gradient is urgently required. Keywords COVID-19 Critical Care Maternal Readmission Vaccination ==== Body pmcIntroduction Risk factors for the development of severe COVID-19 disease such as multimorbidity, socio-economic deprivation, ethnicity and age, are well established.[1], [2] Ongoing research has also shown that, in comparison with non-pregnant women of reproductive age, pregnant women with COVID-19 are more likely to be admitted to critical care, receive invasive ventilation, and die.[3], [4] Despite this, the clinical course of pregnant and recently pregnant women admitted to critical care warrants further investigation, especially in relation to hospital outcomes and healthcare utilisation following hospital discharge. To understand the care trajectory of this patient group, we report a complete national cohort of pregnant and recently pregnant women admitted to critical care in Scotland with confirmed or suspected COVID-19. Uniquely, we also examined re-admission rates in the 90 days following hospital discharge. Methods The Scottish Intensive Care Audit Group (SICSAG) received approval by the Public Benefit and Privacy Panel for Health and Social Care (1920-0093) to undertake work relating to the COVID-19 pandemic. Data sources were linked via the Community Health Index number, a unique patient identifier. This linkage included: the Scottish Morbidity Record (SMR) 01 which captures acute, non-obstetric hospital activity; the Electronic Communication of Surveillance in Scotland which captures virology testing; the National Records of Scotland death records; and the SICSAG database. The SICSAG database prospectively captures all adult intensive care unit (ICU) and general high dependency unit (HDU) activity within Scotland, including pregnancy status on admission, and is subject to regular validation assessments. Using a cohort study design, we examined patients aged 16 years or older, admitted to Scottish critical care units, who were pregnant or recently pregnant (within six weeks of delivery), with a positive polymerase chain reaction test for nucleic acid for SARS CoV-2 before or during critical care admission. Demographic and acute illness variables were obtained from the SICSAG dataset. Ethnicity was derived from categories of the Scottish Census (2011) with low frequencies aggregated. Socio-economic deprivation was defined using quintiles of the Scottish Index of Multiple Deprivation (SIMD), which is an area-based ranking index based on postcode of residence.5 Vaccination status was categorised as: one dose, two or more doses, or unvaccinated. We divided time periods into ‘waves’ of COVID-19 critical care admissions, which were defined by Public Health Scotland: Wave 1 from 1 March, 2020 to 31 July, 2020; Wave 2 from 1 August, 2020 to 18 May, 2021; Wave 3 from 18 May, 2021 to 13 March, 2022.6 We included waves in our descriptive analysis as these reflect the different predominant variants of SARS CoV-2 during the time period.6 Data were analysed using R version 3.6.1 (R Core Team (2018)). Results Between March 2020 and March 2022, 75 pregnant or recently pregnant women with laboratory-confirmed COVID-19 were admitted to 24 ICUs across Scotland. The majority of women (n=51, 68%) were admitted during the third pandemic ‘wave’ (Fig. 1 ).Fig. 1 Number of admissions of pregnant or recently pregnant patients with COVID-19 disease to Scottish Critical Care units by month, stratified by ‘wave’. Waves are defined as: Wave 1 from 1 March, 2020 to 31 July, 2020; Wave 2 from 1 August, 2020 to 18 May, 2021; Wave 3 from 18 May, 2021 to 13 March, 2022. May 2021 therefore incorporates the end of Wave 2 and beginning of Wave 3. The median age of the cohort was 31 (interquartile range (IQR) 27.5-35) years and almost two thirds (n=49, 65%) were from the most deprived socio-economic geographical areas (SIMD 1 and 2). Across the cohort, 48 (64%) women were pregnant on admission to ICU and 27 (36%) were within six weeks of delivery. Less than 5% of the cohort had received any form of vaccination. In total, nine (12%) patients had a recorded comorbidity on admission (Table 1 ).Table 1 Demographics and outcomes of cohort Characteristic n=75 Age, years (median, IQR) 31 (27.5-35) Pregnancy status on admission:- Antenatal - Postnatal 48 (64.0%)27 (36.0%) Socio-economic status quintile (SIMD)- 1 (most deprived) - 2 - 3 - 4 - 5 (least deprived) 31 (41.3%)18 (24.0%)12 (16.0%)9 (12.0%)5 (6.7%) Vaccination status on admission- One dose - Two or more doses - Unvaccinated 1 (1.3%)1 (1.3%)73 (97.3%) Admission diagnosis- COVID-19 confirmed chest infection/viral chest infection - Other 57 (76%)18 (24%) Ethnicity- White - Black/Caribbean/African - Asian - Other - Missing 58 (79.5%)4 (5.5%)8 (11.0%)3 (4.1%)2 (2.7%) Count of previous comorbidity- 0 - 1 or more 66 (88.0%)9 (12.0%) Admission wave- Wave 1 - Wave 2 - Wave 3 2 (2.7%)22 (29.3%)51 (68.0%) Critical care length of stay, days (median, IQR) 4 (1.5-9) Total hospital length of stay, days (median, IQR) 10 (6-17) Requirement for advanced respiratory support 33 (44.0%) Requirement for non-invasive respiratory support 44 (58.7%) Requirement for non-invasive and invasive respiratory support 19 (25.3%) Duration of advanced respiratory support, days (median, IQR) 5 (2-10) Requirement for cardiovascular support 24 (32.0%) Duration of cardiovascular support, days (median, IQR) 2.5 (2.8-5) Requirement for renal replacement therapy 1 (1.3%) Hospital mortality 1 (1.3%) Emergency hospital re-admission for non-obstetric causes (n=74)- within 30 days - within 60 days - within 90 days 7 (9.5%)8 (10.8)9 (12.2%) SICSAG-defined severe comorbidities were combined with Charlson-defined comorbidities in order to report the most prevalent comorbidities. The median critical care length of stay (LOS) was 4 (IQR 1.5-9) days and median hospital LOS was 10 (IQR 6-17) days. Advanced respiratory support was required by 33 (44%) patients, while non-invasive support was required by 44 (59%) and one quarter (n=19, 25%) required a combination of both types of respiratory support. Almost one third (n=24, 32%) required cardiovascular support. Ultimate hospital mortality was 1.3% (n=1). Complete 90-day, acute hospital re-admission data were available for 74 (99%) patients. Nine (12%) women required an emergency non-obstetric hospital re-admission within 90 days. Discussion This complete national cohort of pregnant or recently pregnant women admitted to critical care with COVID-19 has shown a high need for organ support and a non-obstetric re-admission rate of 12%. Despite significant public health campaigning around the benefits of vaccination, vaccination rates were low in this cohort. Less than 5% of this cohort had been vaccinated, findings which support the vital role that vaccination plays in pregnancy. Urgent public health attention is required to inform people of the benefits of vaccination and the significant impact that severe COVID-19 can have on the entire family unit. Moreover, future clinical trials which investigate vaccinations must explicitly include pregnant women to ensure that women can benefit fully from potentially life-saving treatments.7 Nearly two-thirds of the cohort were from areas of socio-economic deprivation. This is higher than that reported for an unselected non-pregnant cohort, where fewer than half of the population were from areas of socio-economic deprivation.1 Lower levels of vaccination uptake have been associated with socio-economic position.8 More targeted public health campaigning across the socio-economic gradient is urgently required. Following discharge, one in eight women in this cohort experienced an emergency re-admission in the 90 days following hospital discharge. Although, this re-admission rate was lower than that of the wider critical care COVID-19 cohort in Scotland (16%) and other critical care cohorts, these non-maternity cohorts are older and have a higher prevalence of comorbidity.9 Previous research has demonstrated that those most at risk of re-admission following critical illness are likely to be those survivors with established frailty or complex comorbidity, a distinctly different group from this current cohort.9 This analysis was unable to delineate reasons for re-admission, so future research should seek to understand the causes of re-admission in this cohort. In parallel with understanding the medical management of this cohort, it is also essential that psychological sequelae are addressed. It is well known that patients can have psychological problems such as anxiety and post-traumatic stress symptomology following a critical illness and following a pregnancy complicated by severe morbidity.[10], [11] Well established psychological interventions in pregnancy or the postpartum period may not have been available for this cohort due to the nature of their illness, which may worsen these psychological issues further. Clinicians should ensure that patients have access to rehabilitation services across the recovery trajectory. The strengths of our study include the complete, nationwide capture of critically ill pregnant or recently pregnant women with COVID-19, and the ability to report hospital re-admission. Limitations include being unable to report more person-centred outcomes, such as psychological sequelae, and neonatal outcomes. Moreover, we do not have data on why women were re-admitted to hospital or which women gave birth during their critical care admission. Finally, this cohort lacks a comparator cohort, and as such we are unable to determine if the problems described were unique to the pregnant or recently pregnant cohort. This national cohort study has demonstrated that pregnant or recently pregnant women admitted to critical care with COVID-19 were more likely to reside in areas of socio-economic deprivation and that <5% of the cohort had received any form of vaccination. More targeted public health campaigning across the socio-economic gradient is urgently required. Research Fellowship (PD-2019-02-16). ==== Refs References 1 Lone N.I. McPeake J.M. Stewart N.I. Influence of socioeconomic deprivation on interventions and outcomes for patients admitted with COVID-19 to critical care units in Scotland: A national cohort study Lancet Reg Health Europe. 1 2020 10.1016/j.lanepe.2020.100005 2 Richardson S. Hirsch J.B. Narasimham D.O. Presenting characteristic, comorbidities and outcomes among 5700 patients hospitalised with COVID-19 in the New York City area J Am Med Assoc. 323 2020 2052 2059 10.1001/jama.2020.6775 3 Stock S.J. Carruthers J. Calvert C. SARS-CoV-2 infection and COVID-19 vaccination rates in pregnant women in Scotland Nature Med. 28 2022 504 512 10.1038/s41591-021-01666-2 35027756 4 Allotey J. Fernandez S. Bonet M. Clinical manifestations, risk factors, and maternal and perinatal outcomes of coronavirus disease 2019 in pregnancy: living systematic review and meta-analysis Br Med J. 370 2020 10.1136/bmj.m3320 5 Scottish Government. Scottish Index of Multiple Deprivation 2020. Available from https://www.gov.scot/collections/scottish-index-of-multiple-deprivation-2020/. Accessed 16 April, 2022. 6 Public Health Scotland (2022) Scottish Intensive Care Society Audit Group report on COVID-19. As at 11th of January 2022. https://publichealthscotland.scot/media/11407/2022-02-02_sicsag_report.pdf. Accessed 16 April, 2022 7 Knight M. Morris R.K. Furniss J. Include pregnant women in research- particularly COVID-19 research Br Med J. 370 2020 10.1136/bmj.m3305 8 Dolby T. Finning K. Baker A. Monitoring sociodemographic inequality in COVID-19 vaccination uptake in England: a national linked data study J Epidemiol Commun Health. 76 2022 646 652 10.1136/jech-2021-218415 9 McPeake J. Bateson M. Christies F. Hospital readmission after critical care survival: a systematic review and meta-analysis Anaesthesia. 77 2022 475 485 10.1111/anae.15644 34967011 10 Wade D. Howell D. Weinman A. Investigating risk factors for psychological morbidity three months after intensive care: a prospective cohort study Crit Care. 16 2013 R192 10.1186/cc11677 11 Hinton L. Locock L. Knight M. Maternal critical care: what can we learn from patient experience? A qualitative study. BMJ Open. 5 2015 10.1136/bmjopen-2014-006676	0	PMC9715259	NO-CC CODE	2022-12-03 23:20:14	no	Int J Obstet Anesth. 2022 Dec 2;:103613	utf-8	Int J Obstet Anesth	2,022	10.1016/j.ijoa.2022.103613	oa_other
==== Front J Neurol Sci J Neurol Sci Journal of the Neurological Sciences 0022-510X 1878-5883 Elsevier B.V. S0022-510X(22)00377-X 10.1016/j.jns.2022.120515 120515 Clinical Short Communication Covid-19 associated free hanging clots in acute symptomatic carotid stenosis Schwartzmann Y. a Leker R.R. a Filioglo A. a Molad J. c Cohen J.E. b Honig A. a⁎ a Departments of Neurology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel b Departments of Neurosurgery, Hadassah-Hebrew University Medical Center, Jerusalem, Israel c Department of Stroke & Neurology, Tel-Aviv Sourasky Medical Center, Tel-Aviv, Israel ⁎ Corresponding author at: Department of Neurology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel. 2 12 2022 15 1 2023 2 12 2022 444 120515120515 28 8 2022 31 10 2022 29 11 2022 © 2022 Elsevier B.V. All rights reserved. 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 Thrombotic complications including stroke were previously described following Covid-19. We aim to describe the clinical and radiological characteristics of Covid-19 related with acutely symptomatic carotid stenosis (aSCS). Method All patients presenting with an aSCS were prospectively enrolled in an ongoing institutional database. Inclusion criteria for the Covid-19-aSCS group were a combination of both antigen test and a positive reverse-transcriptase (PCR) test for Covid-19 upon admission. Patients with additional potential etiologies for stroke including cardioembolism, carotid dissection or patients with stenosis of <50% on CTA were excluded. A cohort of non-Covid-19 related aSCS patients admitted to the same institution before the pandemic during 2019 served as controls. Results Compared to controls (n = 31), Covid-19-aSCS (n = 8), were younger (64.2 ± 10.7 vs 73.5 ± 10, p = 0.027), and less frequently had hypertension (50% vs 90%, p = 0.008) or hyperlipidemia (38% vs 77%, p = 0.029) before admission. Covid-19-aSCS patients had a higher admission NIHSS score (mean 9 ± 7 vs 3 ± 4, p = 0.004) and tended to present more often with stroke (88% vs 55%, p = 0.09) rather than a TIA. Covid-19-aSCS patients had higher rates of free-floating thrombus and clot burden on CTA (88% vs 6.5%, p = 0.002). Covid-19 patients also less often achieved excellent outcomes, with lower percentage of mRS score of 0 after 90-days (13% vs 58%, p = 0.022). Conclusion Covid-19- aSCS may occur in a younger and healthier subpopulation. Covid-19- aSCS patients may have higher tendencies for developing complex clots and less often achieve excellent outcomes. Keywords Covid-19 Symptomatic carotid stenosis Carotid stent Free-floating thrombus ==== Body pmc1 Introduction Thrombotic complications such as acute ischemic stroke (AIS) following Covid-19 infection have been well described [1] and are attributed to large vessel arteriopathy [2,3] cardiac injury including myocarditis and secondary cardioembolism, development of new onset atrial fibrillation, and extra-cranial dissections. Importantly, AIS following Covid-19 infection was found to be associated with higher rates of novel carotid plaque formation and destabilized plaques [3]. The association of symptomatic carotid stenosis (aSCS) and Covid-19 has been previously described in several cohorts [4,5] but without direct comparison to aSCS patients that did not have a recent Covid-19 infection. We aim to highlight the clinical and radiological characteristics of Covid-19-aSCS patients by comparison to a cohort of aSCS patients who underwent carotid artery stenting (CAS) before the Covid-19 era. 2 Methods All patients presenting with an aSCS were prospectively enrolled in an ongoing institutional database. The study was approved by the institutional review board (HMO-0405-20). Symptomatic carotid was defined by a clinical presentation of either TIA or stroke attributed to ipsilateral carotid stenosis of ≥50% by NASCET criteria as adjudicated by a certified stroke neurologist. All patients underwent multiphasic computed tomography angiography (CTA) upon their presentation and degree of stenosis was measured on maximal intensity projection images on axial and sagittal views. All patients with >50% stenosis of the symptomatic carotid indicated by CTA were sent for digital subtraction angiography (DSA). During the Covid-19 pandemic, all patients that were admitted as inpatients were routinely tested for Covid-19. Inclusion criteria for the Covid-19 group patients were a positive Covid-19 test result on a combination of both an antigen test and reverse-transcriptase (PCR) test upon admission indicating an active infection by Covid-19. Consecutive aSCS from our 2019 prospective registry were selected as controls as it was prior to the Covid-19 outbreak. Patients with Covid-19-aSCS patients were treated similarly to those that were Covid-19 negative. All antiplatelet naïve patients were treated according to similar treatment protocols with loading dose of 300 mg of Aspirin and 300 mg of Clopidogrel followed by a daily dose of 100 mg Aspirin and 75 mg of Clopidogrel. Patients that were on single antiplatelet agents prior to admission were loaded with the other antiplatelet agent and continued DAPT. All patients received high dose potent statin therapy with atorvastatin 80 mg/ day and blood pressure was kept at 140–150 systolic. Exclusion criteria in both groups included any other possible additional etiology for stroke including cardioembolism, dissection, etc. All patients underwent workup included trans-thoracic echocardiogram with injection of agitated saline and a 48-h electrocardiogram-holter to rule out atrial fibrillation. Patients who were deemed non-eligible for carotid intervention due to short life expectancy or ongoing dementia were not sent for DSA. As per institutional protocols, all aSCS patients were referred to endovascular treatment with potential stent implantation if feasible. Carotid endarterectomy was avoided in patients with Covid-19-aSCS because of the potential risk of anesthesia and surgery. All CAS procedures were performed by the same interventional neurosurgeon using a similar institutional protocol. 3 Data collection We collected demographic and vascular risk factors. Neurological deficits were measured using the National Institutes of Health Stroke Scale (NIHSS) [6] at admission and discharge. Stroke etiology was classified with the TOAST classification [7]. Functional status was assessed with the modified Rankin Score (mRS) prior to stroke, upon discharge, and 90 days after stroke. Due to the relatively young age of the entire cohort and the Covid-19 patients in particular we anticipated significant functional improvement. Therefore, an excellent functional outcome was defined as mRS 0 on day-90. 4 Results Out of 13 Covid-19-aSCS patients, eight fulfilled entry criteria. Five patients were excluded, of which three patients had concomitant atrial fibrillation, one patient had a carotid dissection and one had severe cardiac failure with poor life expectancy. In comparison to the non-Covid-19 controls (n = 31), Covid-19-aSCS patients were younger (64.2 ± 10.7 vs 73.5 ± 10, p = 0.027), and were less frequently diagnosed with either hypertension (50% vs 90%, p = 0.008) or hyperlipidemia (38% vs 77%, p = 0.029) before admission (Table 1 ) but other risk factors did not differ significantly.Table 1 Characteristics of symptomatic carotids stenosis patients with and without Covid-19. Table 1Characteristics Non-COVID-19 N = 31 COVID-19 N = 8 P Age, mean (SD) 73.5 (10) 64.2 (10.7) 0.027 Sex male (%) 22 (71) 7 (88) 0.340 Hypertension (%) 28 (90) 4 (50) 0.008 Hyperlipidemia (%) 24 (77) 3 (38) 0.029 Smoking (%) 9 (29) 1 (13) 0.340 Diabetes (%) 12 (39) 2 (25) 0.471 Ischemic heart disease (%) 10 (32) 2 (25) 0.692 Clinical presentation 0.090 Stroke (%) 17 (55) 7 (88) TIA (%) 14 (45) 1 (12) Left side (%) 18 (58) 5 (63) 0.820 Stenosis percent, mean (SD) 85 (14) 78 (22) 0.305 Admission NIHSS, mean (SD) 3 (4) 9 (7) 0.004 Discharge NIHSS, mean (SD) 2 (3) 8 (7) 0.002 Delta NIHSS, mean (SD) 1 (2) 1 (4) 0.796 mRS at 3 months, median (IQR) 0 (0–2) 2 (1–3) 0.246 90-day-mRS 0 (%) 18 (58) 1 (13) 0.022 90-day-mRS 0–1 (%) 21 (68) 3 (37) 0.11 Covid-19-aSCS patients tended to present more often with stroke (88% vs 55%, p = 0.09) rather than a transient ischemic attack (TIA) and had higher admission and discharge NIHSS scores (mean 9 ± 7 vs 3 ± 4, p = 0.004) and (mean 8 ± 7 vs 2 ± 3, p = 0.002) respectively. Radiologically, all Covid-19-aSCS patients that presented with a stroke (n = 7, Table 2 ) had a free-floating thrombus in the carotid (100% vs 6.5%, p < 0.001) (Fig. 1 ). Two patients presented with total occlusion of the internal carotid arteries. One (#8) underwent urgent endovascular thrombectomy to occlusive clots in the LICA and left-M1while the other (#6) already suffered from an extensive cerebral ischemic damage with no mismatch on CT perfusion imaging and was not given EVT.Table 2 Covid-19 associated acute symptomatic carotid stenosis. Table 2 Age Sex HTN Lipid* D.M Smoking Previous Anti-PLT treatment Vaccination status Neurological Symptoms from Covid-19 diagnosis NIHSS upon arrival NIHSS upon discharge Affected Carotid % Carotid Stenosis Free hanging thrombus carotid M1 involvement on ipsilateral side EVT Day of CAS from symptom onset MRS 90 1 68 M + − − − − − Day 0 1 1 RICA 60% + + Sub occlusive clot − Day 80 1 2 64 F − − − − − − Day 7 10 10 LICA 50% + + Total occlusion due to clot + − 2 3 75 M − + − − − − Day 7 0 0 LICA 65% − − − Day 7 1 4 80 M + − + − − − Day 0 12 10 LICA 90% + − − Day 7 4 5 64 M + + − − Aspirin + Day 0 8 8 RICA 90% + − − Day 14 4 6 61 M − − − − − − Day 7 15 21 RICA 100% + + Total occlusion due to clot − − 4 7 57 M + + + + − − Day 0 4 1 LICA 90% + − − Day 21 0 8 45 M − − − − − − Day 0 20 9 LICA 100% + + Total occlusion due to clot + − 3 NIHSS- National Institutes of Health Stroke Scale. DSA- Digital subtraction angiography. EVT- Endovascular thrombectomy. CAS- Carotid artery stenting. HTN- Hypertension. D.M.- Diabetes mellitus. MRS- Modified Rankin scale. Lipid = Dyslipidemia. PLT- Platelets. Fig. 1 Free hanging clots in the carotid of Covid-19 symptomatic carotid stenosis patients. A- RICA 90%, patient number 5. B- LICA 90% stenosis, patient number 7. C- LICA 70% stenosis, Patient number 3. Blue arrow- Free floating thrombus, red arrow- atherosclerotic plaque, Fig. 1 Another patient (#2) underwent thrombectomy with an occlusive clot removed from left-M1only. In the remaining patients with stroke, non-occlusive floating thrombus were found on focal atherosclerotic plaques in proximity to the carotid bifurcation and an additional free-floating thrombus was found at a more distal part of the carotid (Fig. 1) or an additional non-occlusive thrombus along the M1 (Table 2). Four Covid-19-aSCS patients with free hanging clots were treated with variable dosages of Low Molecular Weight Heparin (LMWH) in addition to DAPT until clot resolution. In three of them full therapeutic dose of LMWH (1 mg/kg BID) was given while in the fourth a reduced dose of LMWH was administered due to the fear of hemorrhagic transformation of his extensive ischemic lesions. In contrast, in the control cohort, none of the patients were treated with LMWH. Two patients who were treated with DAPT without LMWH has experienced recurrent TIA events in the same carotid territory until CAS was instituted. Five patients underwent CAS, three of them had a near-total occlusion and one of them had a free-floating clot but no occlusion. In these cases, CAS was delayed until imaging proved clot resolution. Consequently, CAS procedure was performed within longer interval from symptom onset in the Covid-19-aSCS patients (26 ± 27 vs 12 ± 7, p < 0.001). All patients included in our study, from both the Covid-19 and the control group, survived after 90 days from stroke onset. Covid-19-aSCS patients less frequently achieved excellent outcome, defined as mRS score of 0 after 90-day (13% vs 58%, p = 0.022), and tended to less frequently achieve good functional outcome, defined as 90-day mRS 0–1 (37% vs 68%, p = 0.11). Notably, 88% (7/8) of the Covid-19-aSCS patients were unvaccinated for Covid-19 upon presentation. Upon admission, three COVID-19 patients were systemically asymptomatic, three had mild respiratory illness and two had substantial respiratory illness requiring oxygen support. One patient required mechanical ventilation at a later stage of the admission due to substantial Covid-19 respiratory illness. In contrast, none of the control group presented with respiratory illness or either required oxygen support or mechanical ventilation during their course of admission. 5 Discussion The main findings of the current analysis are that in comparison to non-Covid-19-aSCS patients, those patients with concomitant Covid-19 infection and aSCS are younger, more often present with stroke, more often harbor complex plaques with free-floating thrombi, and generally have poorer outcomes. In our study, Covid-19 -aSCS patients were younger and this is in accordance with previous studies [3]. Covid-19-aSCS patients had lower rate of cardiovascular risk factors and possibly Covid-19 was a trigger for an active atherosclerotic plaque in patients with milder atherosclerotic disease. Again, this is accordance with a previous study describing AIS in Covid-19-aSCS patients due to fulminant carotid thrombosis overlying mild atherosclerotic plaque [4].Moreover, the affinity of Covid to the carotid endothelial wall is well known and may occur even in young patients without vascular risk factors [8]. Covid-19-aSCS patients tended to present more often with stroke rather than a TIA compared to the non-covid-19 population. Possible explanations to this observation include higher clot burden within the carotid lumen [4,5] and the overall hypercoagulable state found in Covid-19 [6]. Covid-19-aSCS patients more frequently presented with free-floating-thrombus. This radiological finding has been previously described [4,5] and could also contribute to the higher rate of observed strokes rather than TIAs. Possible pathogenesis for accelerated thrombus formation in Covid-19 patients include systemic inflammation that may cause atherosclerotic plaque rupture, and the affinity of Covid-19 to the carotid endothelial cells through ACE-2 receptors causing local endotheliosis [7]. When taken together, these two processes promote thrombi formation. Additionally, the hypercoagulable state seen in Covid-19 may promotes clot extension and disintegration until reaching large sizes [6]. Unfortunately, it is possible that positive Covid-19 status may have resulted in delayed medical attention, imaging and consequently treatment that may contribute to the worse outcome of the Covid-19-aSCS patients. Importantly, the fact that only one of the Covid-19-aSCS patients was vaccinated possibly contributed to an increased systemic inflammation and pro-coagulable state [9]. Our findings favor treating Covid-19-aSCS patients with LMWH in addition to DAPT due to high burden of free hanging clots that predispose to recurrent ischemic events. Further larger scale studies could shed light on the required dosage and duration of LMWH treatment in Covid-19-aSCS patients. Our treatment decision paradigm was not risk free. We have tried to treat conservatively whenever possible with either anticoagulation whenever a clot was visible on CT-angiography or dual antiplatelet treatment whenever clot was not clearly visible. Repeated imaging would allow to observe resolution of the clot thus allowing carotid stenting with reduced risk of periprocedural stroke. Imaging was performed on different time interval on an individual basis. Additionally, we tried to delay intervention to allow the body to overcome the Covid-19 illness and assumed diminished pro-coagulable state. Finally, the minority of the Covid-19-aSCS patients had severe respiratory illness further highlighting the increased risk of stroke in mildly symptomatic Covid-19 patients and may suggest additional treatment with anticoagulants such as low molecular weight heparin in addition to antiplatelet treatment whenever clots are seen in order to dissolve the floating clots [8]. Our study limitations include a single center setting with a small number of included patients. Further, larger scale studies could reinform our findings and shed more-light on the phenomenon and the best suited treated strategies. ==== Refs References 1 Xie Y. Xu E. Bowe B. Al-Aly Z. Long-term cardiovascular outcomes of COVID-19 Nat. Med. 28 2022 583 590 10.1038/s41591-022-01689-3 35132265 2 Vogrig A. Gigli G.L. Bnà C. Morassi M. Stroke in patients with COVID-19: clinical and neuroimaging characteristics Neurosci. Lett. 743 2021 135564 10.1016/j.neulet.2020.135564 3 Shahjouei S. Tsivgoulis G. Farahmand G. Koza E. Mowla A. Vafaei Sadr A. Kia A. Vaghefi Far A. Mondello S. Cernigliaro A. SARS-CoV-2 and stroke characteristics: a report from the multinational COVID-19 stroke study group Stroke. 52 2021 e117 e130 10.1161/strokeaha.120.032927 33878892 4 Esenwa C. Cheng N.T. Lipsitz E. Hsu K. Zampolin R. Gersten A. Antoniello D. Soetanto A. Kirchoff K. Liberman A. COVID-19-associated carotid atherothrombosis and stroke AJNR Am. J. Neuroradiol. 41 2020 1993 1995 10.3174/ajnr.A6752 32819896 5 Cancer-Perez S. Alfayate-García J. Vicente-Jiménez S. Ruiz-Muñoz M. Dhimes-Tejada F.P. Gutiérrez-Baz M. Criado-Galan F. Perera-Sabio M. de Benito-Fernández L. Symptomatic common carotid free-floating thrombus in a COVID-19 patient, case report and literature review Ann. Vasc. Surg. 73 2021 122 128 10.1016/j.avsg.2021.02.008 33689754 6 Brott T. Adams H.P. Jr. Olinger C.P. Marler J.R. Barsan W.G. Biller J. Spilker J. Holleran R. Eberle R. Hertzberg V. Measurements of acute cerebral infarction: a clinical examination scale Stroke. 20 1989 864 870 10.1161/01.str.20.7.864 2749846 7 Adams H.P. Jr. Bendixen B.H. Kappelle L.J. Biller J. Love B.B. Gordon D.L. Marsh E.E. 3rd. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of org 10172 in acute stroke treatment Stroke. 24 1993 35 41 10.1161/01.str.24.1.35 7678184 8 Itsekson Hayosh Z. Schwammenthal Y. Orion D. Can other coronavirus infections cause a cryptogenic stroke in a young patient? BMJ Case Rep. 2021 14 10.1136/bcr-2020-239113 9 Tenforde M.W. Self W.H. Adams K. Gaglani M. Ginde A.A. McNeal T. Ghamande S. Douin D.J. Talbot H.K. Casey J.D. Association between mRNA vaccination and COVID-19 hospitalization and disease severity Jama. 326 2021 2043 2054 10.1001/jama.2021.19499 34734975	36493703	PMC9715260	NO-CC CODE	2022-12-08 23:15:57	no	J Neurol Sci. 2023 Jan 15; 444:120515	utf-8	J Neurol Sci	2,022	10.1016/j.jns.2022.120515	oa_other
==== Front Contemp Clin Trials Commun Contemp Clin Trials Commun Contemporary Clinical Trials Communications 2451-8654 Published by Elsevier Inc. S2451-8654(22)00160-0 10.1016/j.conctc.2022.101043 101043 Article Recruitment challenges for a prospective telehealth cohort study Pertl Kellie a Petluri Ritwika a Wiest Katharina a Hoffman Kim b∗ McCarty Dennis b Levander Ximena A. c Chan Brian c Martin Stephen A. ad Korthuis P. Todd bc a Boulder Care, Portland, OR, USA b OHSU – PSU School of Public Health, Oregon Health & Science University, Portland, OR, USA c Addiction Medicine Section, Department of Medicine, School of Medicine, Oregon Health & Science University, Portland, OR, USA d Department of Family Medicine and Community Health, UMass Chan Medical School, Worcester, MA, USA ∗ Corresponding author. Department of General Internal Medicine, 3181 S.W. Sam Jackson Park Road, Portland, OR, 97239-3098, USA. 2 12 2022 2 12 2022 10104316 9 2022 10 11 2022 27 11 2022 © 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. Background The COVID-19 pandemic presents challenges in participant recruitment strategies for clinical research involving people with opioid use disorders recently engaged in treatment. We describe challenges to participant recruitment in a trial comparing virtual buprenorphine treatment platform to office-based buprenorphine treatment. Methods The parent study was a cohort trial of telehealth delivered buprenorphine treatment compared to office-based buprenorphine treatment, however, due to the pandemic potential participant recruitment for both arms became virtual. Between 9/27/2021 and 7/11/2022, telephone, email, flyers, and word-of-mouth were used to recruit study participants from each treatment setting. Recruitment tracking documents recorded the primary outcomes: number of outreach attempts and most effective contact methods. Results Treatment settings provided contact information for 1485 potential study participants. Information was incorrect or disconnected for 282 (19%) individuals, 695 (47%) did not respond to outreach, and 508 (34%) responded to outreach. Of these responders, 369 were interested in study participation, 259 completed the online informed consent and screening assessment, and 148 met eligibility criteria and enrolled in the study. A total of 3804 virtual outreach attempts across 1485 potential participants were made, resulting in an average of 2.7 attempts per contact and a mean of 25.7 attempts per enrolled participant (n = 148). Conclusion Conducting research during the COVID-19 pandemic required shifting from in-person to virtual recruitment strategies to contact and engage potential study participants. Virtual recruitment for this population during a pandemic appears to be less efficient and hindered efforts to meet recruitment goals. Keywords Recruitment challenges Virtual recruitment COVID-19 Telemedicine Buprenorphine Opioid Use Disorder ==== Body pmcAbbreviations OUD Opioid Use Disorder NIH National Institutes of Health OHSU Oregon Health & Science University NIDA National Institute on Drug Abuse Funding This work was supported by the 10.13039/100000026 National Institute on Drug Abuse [NIDA 4 R44 DA050345]. Role of funding source This project was funded by the 10.13039/100000026 National Institute on Drug Abuse 10.13039/100006370 Small Business Innovation Research (SBIR) grant; therefore, by design, has academics and industry working together. The funder did not participate in the design of the study, nor in the collection, analysis, interpretation of data, or writing of the manuscript. 1 Background During the COVID-19 pandemic, up to 80% of non-COVID-19 trials were interrupted or terminated [1]. Clinic closures, staffing issues, and restrictions on clinical research activities at many sites impeded recruitment [2]. For studies that continued, recruitment strategies focused on online and telephone interactions [3]. In a pre-pandemic meta-analysis comparing online recruitment versus traditional in-person strategies, online recruitment strategies were more effective at recruiting eligible participants but offline/in-person recruitment showed a higher rate of enrolling participants [4]. Barriers to pandemic study recruitment should be addressed early to avoid delays in recruitment and mitigate threats to internal and external study validity [[5], [6], [7]]. Opioid use disorder (OUD) treatment studies typically recruited participants in treatment settings where they had one or more in-person visits [3]. Telehealth makes it easier to connect with potential participants who have difficulty meeting in-person due to the logistical barriers people with OUD often experience [8]. Online engagement, however, can be limited by participants’ access to and comfort with technology [8]. People with substance use disorders often lack access to a phone, wireless connection, or data plan, making online engagement more difficult [9]. The Buprenorphine Evaluation and Telehealth Study (CTN 05529225) awarded prior to the pandemic planned to recruit from two arms: 1) adults receiving buprenorphine for OUD through telehealth only with virtual recruitment and 2) adults receiving buprenorphine for OUD through office-based treatment with in-clinic, in-person recruitment. However, due to pandemic constraints, recruitment for both arms became virtual. We analyzed recruitment data to identify recruitment challenges. 2 Methods The parent study was a cohort trial of telehealth delivered buprenorphine treatment compared to office-based buprenorphine treatment, however, due to the pandemic, potential participant recruitment for both arms became virtual. Between September 27, 2021, and July 11, 2022, eligible participants – adults 18 years of age or older, within 45 days of a new buprenorphine prescription for OUD treatment, and not pending court appearances or incarceration – were recruited from electronic health records (EHR) and provider lists in Oregon and Washington. Eligibility screening, informed consent and baseline assessments were completed online. Participants received compensation for participation via a reloadable gift card at each study visit. The Oregon Health & Science University IRB approved all study procedures. 2.1 Description of contact and recruitment methods 2.1.1 Phone Potential participants were contacted using a toll-free phone number from RingCentral [10] or by Doximity [11] (HIPAA-compliant software that allows medical professionals to use medical office phone numbers from personal telephones). RingCentral displayed a recognized phone number of the telehealth clinic to potential telehealth-only participants. Doximity displayed a recognized phone number of an office-based clinic to potential office-based participants. 2.1.2 Email Two email domains (.edu or.care) were used to recruit and follow study participants. 2.1.3 Fliers With permission from individual clinics who were still operating in-person, paper fliers were posted for patients obtaining in-person treatment. 2.1.4 Word of mouth Participants shared information about the study with friends and family. 2.2 Outcomes We tracked every attempted contact, noting the methods and outcomes for each. The primary outcomes were the number of outreach attempts and most effective contact methods overall, and by each method of recruitment. 2.3 Analysis Recruitment tracking documents were analyzed by each recruitment method using total counts, percentage, and the mean response rate overall. We analyzed the outcomes to understand recruitment rates related to the method of recruitment. 3 Results We received contact information for 1485 potential participants. Nearly half (47%; n = 695) did not respond to repeated contact attempts. Information was incorrect or not in service for 282 (19%) individuals. One in three potential participants (n = 508; 34%) responded to a contact attempt. Among the 508 individuals who responded, 369 (73%) were interested in study participation, 259 (51%) completed the online screening assessment, and 148 (29%) completed informed consent, met eligibility criteria and enrolled in the study (Fig. 1 ). Overall, of 1485 potential participants, 148 (10%) enrolled in the study.Fig. 1 Enrollment of potential participant pool. Fig. 1 Among 369 individuals who responded to outreach efforts and were interested in study participation, 53% responded to email, 45% were contacted by phone, 1% were contacted by word of mouth, and 1% contacted us after viewing a flier. Of 148 eligible participants, 69% were recruited through email, 36% were recruited through phone, 2% were referred by word of mouth, and 1% through flier advertisement. Study personnel made 3804 outreach attempts across 1485 potential participants, yielding an average of 2.7 attempts per contact and a mean of 25.7 attempts per enrolled participant (n = 148). The study planned to recruit 100 participants in each arm. We recruited 100 in the telehealth arm between September 2021 and April 2022 and 34 in the planned office-based arm between November 2021 and April 2022. 4 Conclusion Conducting research during the COVID-19 pandemic required adaptation of study recruitment methods and a shift from in-person to telephone and email strategies to contact and engage potential participants. Recruitment during a pandemic was challenging and the shift to virtual methods led to lower-than-expected recruitment in the planned office-based arm. Traditional pre-pandemic recruitment protocols need to be improved and adapted to support virtual recruitment strategies. Our study differed from similar previous studies [[4], [5], [6], [7]] as we planned to recruit telehealth participants using online strategies and office-based participants in-person in treatment centers. However, access to in-person settings was restricted due to COVID-19. Virtual recruitment differed from in-person recruitment due to the lack of face-to-face communication and the computer literacy needed to access and engage with the study's online survey platforms. Online recruitment was not as effective for enrolling participants receiving planned office-based buprenorphine compared to participants receiving telehealth only. A possible explanation is that telehealth participants were comfortable with engaging virtually and may have had more regular access to cell phones, computers, or tablets with internet access compared to participants receiving planned office-based treatment. Participants with virtual healthcare experience might also be more trusting of phone or internet recruitment. Additionally, many clinics providing what would have been office-based buprenorphine were safety-net clinics serving a more medically complex population who might not have the resources to engage online. 4.1 The study team identified six challenges 4.1.1 COVID-19 related delays We were unable to start office-based recruitment on time due to clinic closures, staffing issues, restrictions on all clinical research activities, and lack of identification of a point person at each clinic to obtain participant contact information. 4.1.2 IRB restrictions We experienced IRB delays during the study start-up phase due to IRB staffing issues. We received IRB approval to start recruitment in September 2021. During the study it became clear that participants preferred text messaging. The IRB initially did not approve text messaging due to confidentiality concerns, however, a compromise was reached where texting, via the secure RingCentral app, was limited to consented participants who agreed to texting. This restriction adversely impacted the number of people reached during the recruitment phase. 4.1.3 Electronic health record and provider lists Electronic health record and provider lists often contained incorrect or missing contact information and we did not know the proportion of the list that would have been eligible to participate in this study. This added to the challenges brought on by COVID-19 and reflects how contact data and diagnoses are stored in the EHR, the data we were able to pull, and how we received the lists. We adjusted the EHR data pull variables frequently throughout recruitment to refine the list of potential participants but gained few additional successful recruitments from this. 4.1.4 Relationships with clinics COVID-19 research restrictions made it difficult to establish working relationships with office-based study sites. The availability of potential participant contact information from clinics was irregular. A clinic providing occasional in-person care was reluctant to participate because of the pandemic and concerns regarding additional staff burden. Study staff met with clinic leadership and staff to facilitate research participation. In contrast, the telehealth clinic was better suited for online recruitment because the pandemic's impact on their care model was minimal. An established relationship with the clinic was essential to recruitment success and to bring the study to completion. In the absence of these partnerships, recruitment would have been more difficult. In-Person Recruitment Less Effective. Because office-based buprenorphine clinics were operating virtually, the fliers were rarely seen, and word of mouth was less effective than expected. 4.1.5 Unrecognized phone numbers and emails Study staff made persistent efforts to contact potential study participants and called at various times of day and days of the week to accommodate their needs. We learned from potential participants that they ignored unrecognized telephone numbers. RingCentral and Doximity were used to mitigate this and display a recognizable telephone number to potential participants in each arm. Additionally, the domain of the email used to recruit mattered. Based on participant feedback, we learned non “.edu” emails ended up in spam or were deleted without opening. The cumulative impact of these challenges during the funded study period was an inability to recruit the planned 100 office-based buprenorphine participants. The recruitment and contact obstacles are important for designing future studies given the time and budgetary impact of delayed recruitment. Author contributions Kellie Pertl: Writing - Original Draft, Formal analysis, Methodology, Investigation, Data Curation, Visualization Ritwika Petluri: Writing - Original Draft, Formal analysis, Methodology, Investigation, Data Curation, Visualization Katharina Wiest: Conceptualization, Methodology, Writing - Review & Editing Kim Hoffman: Writing - Review & Editing Dennis McCarty: Writing - Review & Editing Ximena A. Levander: Writing - Review & Editing, Resources Brian Chan: Writing - Review & Editing, Resources Stephen A. Martin: Funding acquisition, Writing - Review & Editing, Resources P. Todd Korthuis: Supervision, Writing - Review & Editing, Resources. Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Ms. Pertl reported being an employee of Boulder Care. Ms. Petluri reported being an employee of Boulder Care. Dr. Wiest reported being a consultant of Boulder Care. Dr. Martin reported being an employee of Boulder Care. The other authors declare they have no competing interests. To reduce potential conflict of interest, OHSU researchers served as final manuscript reviewers and corresponding author. Data availability The data that has been used is confidential. ==== Refs References 1 van Dorn A. COVID-19 and readjusting clinical trials Lancet Lond. Engl. 396 10250 2020 523 524 10.1016/S0140-6736(20)31787-6 2 COVID19-Response8.0_Clinical-Trials_2020824_v1.pdf https://www.medidata.com/wp-content/uploads/2020/08/COVID19-Response8.0_Clinical-Trials_2020824_v1.pdf 3 Steinhubl S.R. Wolff-Hughes D.L. Nilsen W. Iturriaga E. Califf R.M. Digital clinical trials: creating a vision for the future NPJ Digit Med. 2 2019 126 10.1038/s41746-019-0203-0 31872066 4 Brøgger-Mikkelsen M. Ali Z. Zibert J.R. Andersen A.D. Thomsen S.F. Online patient recruitment in clinical trials: systematic Review and meta-analysis J. Med. Internet Res. 22 11 2020 e22179 10.2196/22179 5 Sullivan-Bolyai S. Bova C. Deatrick J.A. Barriers and strategies for recruiting study participants in clinical settings West. J. Nurs. Res. 29 4 2007 486 500 10.1177/0193945907299658 17538128 6 Hoeflich C.C. Wang A. Otufowora A. Cottler L.B. Striley C.W. Virtual recruitment and participant engagement for substance use research during a pandemic Curr. Opin. Psychiatr. 35 4 2022 252 258 10.1097/YCO.0000000000000794 7 Gul R.B. Ali P.A. Clinical trials: the challenge of recruitment and retention of participants J. Clin. Nurs. 19 1–2 2010 227 233 10.1111/j.1365-2702.2009.03041 20500260 8 Aronowitz S.V. Engel-Rebitzer E. Dolan A. Telehealth for opioid use disorder treatment in low-barrier clinic settings: an exploration of clinician and staff perspectives Harm Reduct. J. 18 2021 119 10.1186/s12954-021-00572-7 34823538 9 Buchheit B.M. Wheelock H. Lee A. Brandt K. Gregg J. Low-barrier buprenorphine during the COVID-19 pandemic: a rapid transition to on-demand telemedicine with wide-ranging effects J. Subst. Abuse Treat. 131 2021 108444 10.1016/j.jsat.2021.108444 34098299 10 Video Phone \| RingCentral Message https://www.ringcentral.com/ 11 Doximity https://www.doximity.com	36475092	PMC9715261	NO-CC CODE	2022-12-09 23:15:18	no	Contemp Clin Trials Commun. 2023 Feb 2; 31:101043	utf-8	Contemp Clin Trials Commun	2,022	10.1016/j.conctc.2022.101043	oa_other
==== Front J Infect J Infect The Journal of Infection 0163-4453 1532-2742 The British Infection Association. Published by Elsevier Ltd. S0163-4453(22)00686-7 10.1016/j.jinf.2022.11.025 Article Impact of the early phase of COVID-19 on the trends of isolated bacteria in the national database of Japan: an interrupted time-series analysis Kakimoto Masaki a Miyamori Daisuke a⁎ Omori Keitaro b Kobayashi Tomoki a Ikeda Kotaro a Kashiyama Seiya c Ohge Hiroki b Ito Masanori a a Department of General Internal Medicine, Hiroshima University Hospital, 1-2-3 Kasumi, Minamiku, Hiroshima 734-8551, Japan b Department of Infectious disease, Hiroshima University Hospital, 1-2-3 Kasumi, Minamiku, Hiroshima 734-8551, Japan c Section of Clinical Laboratory, Department of Clinical Support, Hiroshima University Hospital, 1-2-3 Kasumi, Minamiku, Hiroshima 734-8551, Japan ⁎ Corresponding author. Daisuke Miyamori, Department of General Internal Medicine, Hiroshima University Hospital, 1-2-3 Kasumi, Minamiku, Hiroshima 734-8551, Japan, Tel: +81-82-257-5460 2 12 2022 2 12 2022 28 11 2022 © 2022 The British Infection Association. Published by Elsevier Ltd. All rights reserved. 2022 The British Infection Association 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 During the coronavirus disease 2019 (COVID-19) pandemic, a change in the trend of infections was observed. However, there are few reports comprehensively assessing the impact of the early phase of COVID-19 on the trend of bacteria isolated. Methods We extracted the number of positive cultures of hospitalized patients for approximately 200 institutions using the Japanese national database. The outcome was the ratio of 10 species isolated in comparison to the total isolates for each month. Interrupted time-series analyses were conducted between 13 (from Jan-2019 to Jan-2020) and 8 (from May-2020 to Dec-2020) monthly data points. Results A total of 369,210 isolates were involved. Differences in the level change for Streptococcus pneumoniae, Haemophilus influenzae, and Streptococcus pyogenes decreased significantly by 0.272 (95% confidence interval [CI]:0.192-0.352), 0.244 (95%CI:0.174-0.314), and 0.324 (95%CI:0.06-0.589), respectively. Bacteria transmitted by contact infection, such as Staphylococcus aureus, did not decrease. Differences in slope change were not significant in all species. Conclusions The ratios of isolated bacteria transmitted by droplet infection decreased immediately after the early phase of COVID-19 and maintained the same level. The awareness and behavioral changes toward increased COVID-19 prevention might have a substantial impact on the prevention of bacterial infections, especially droplet infections. Keywords COVID-19 J-SIPHE Positive cultures National database Behavioral change Bacteria Culture Large-scale Behavior change Interrupted time-series analysis ==== Body pmcIntroduction Since the early phase of the coronavirus disease (2019) COVID-19 in Japan, there has been a change in awareness and behavior among the public, including “keeping social distance,” “washing hands,” and “wearing masks,” which have improved public health by preventing infections. The most impactful trigger for the public in Japan for these changes in awareness and behavior was the widespread infection of COVID-19 on cruise ships in early February 2020. At the same time, the number of opportunities for people to come into contact was reduced by the declaration of a state of emergency and the closure of schools 1. The COVID-19 pandemic brought about a significant change in the trends of bacteria, and no increase in the prevalence of community-acquired pneumonia along with the COVID-19 pandemic during the winter of 2020-2021 was observed although there were concerns. Instead, it has been reported that the number of patients with community-acquired pneumonia admitted to hospitals and the number of patients with invasive bacterial infections that are sequels to respiratory tract infections has decreased since the COVID-19 pandemic 2, 3, 4. However, statistics based on the number of cases, such as those in previous reports 3, only measure the statistical results diagnosed from various symptoms caused by bacterial infections. It is difficult to accurately determine bacterial trends, including subclinical infections and asymptomatic carriers, for these cases. It is unknown whether the decrease in the number of patients with bacterial infections caused a relative increase in the number of subclinical infections or a decrease in the number of bacteria. While it is possible that the interruption of the infection route directly affected the reduction in the number of patients, it is also possible that changes in lifestyle, such as diet, affected the human immune system, which in turn affected the onset of infectious diseases even after infection. Second, for bacteria mainly transmitted by droplet infection, a decrease in the number of patients has been reported, but no reports have compared multiple bacteria simultaneously. It is difficult to refer to a decrease in the number of cases due to the different routes of infection because bacteria from multiple routes of infection have not been comprehensively analyzed. A comprehensive comparison and understanding of the changes in the trends of detection ratios for each bacterial isolates before and after the COVID-19 pandemic from based on the characteristics of infection routes are essential for the general prevention of infectious diseases. Analyzing the changes in the detection ratios of isolated bacteria from the culture and considering the route of infection mainly affected will help determine the bacterial groups affected by the infection prevention measures against COVID-19 as a secondary effect. In this study, we investigated the effect of COVID-19 on bacteria using an interrupted time-series analysis (ITSA) of changes in the number of bacteria isolated from the cultures of in-patients during the early phase of COVID-19 using a large Japanese database. Materials and Methods Study design We conducted ITSA of the detection ratios of bacteria isolated during the early phase of the COVID-19 pandemic, one of the most valuable methods for comparing the effects before and after an event using observational data 5. ITSA is used to evaluate long-term trends at the population level under the assumption that the influences of confounding, such as contamination, are maintained at a constant rate 6. This study determined whether the detection ratios of bacteria isolated in any cultures was associated with the early phase of COVID-19 using a large-scale database in Japan. The conduct of this study complied with the aims of Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Data source We used data from the Japan Surveillance for Infection Prevention and Healthcare Epidemiology (J-SIPHE) database, which is collected and managed by the Antimicrobial Resistance (AMR) Clinical Reference Center of the National Center for Global Health and Medicine. The J-SIPHE database has been aimed at improving regional and national infection control and AMR measures since 2017. Information is regularly collected from participating registered facilities throughout Japan. As of July 2020, 518 facilities were participating. Among them, 418 facilities had entered information related to microbes and resistant bacteria, including the following: the amount and rate of antimicrobial use at each facility; the number of bacteria isolated and their antimicrobial susceptibilities; the number of multidrug-resistant bacteria isolated; and Antimicrobial Stewardship Program by consultants specializing in infectious diseases and infection control team activities in each hospital. In the section on microorganisms, the 21 major bacteria to be surveyed are voluntarily registered once detected by bacterial culture. Hospitals that provide information, such as multicenter comparisons, are given feedback to promote AMR activities. The details of J-SIPHE are provided elsewhere 7 , 8. This study involved facilities with antimicrobial testing systems contributing to the above registries. The facilities that contribute to these databases account for 75,000 acute care beds and approximately 10% of Japan's total acute care beds. Study Population Data on bacterial culture tests performed at the hospital for in-patients from approximately 200 facilities were extracted from an open cohort for which information on microorganisms, resistant bacteria, and all detected bacteria was registered in the database. The database used in this study is in an open cohort format, so the number of participating institutions may vary from year to year. Furthermore, to reduce errors in the species and number of bacteria isolated due to differences in healthcare delivery systems and patient backgrounds during infectious disease treatment, the selected data for in-patients satisfied the criteria for additional reimbursement for infection prevention type 1. The reasons for the errors include the size and function of the facilities. The Ministry of Health, Labour, and Welfare in Japan classified the additional reimbursement for infection prevention in medical institutions into several types according to the size and function of the hospital. Specific facility criteria for infection prevention must be met for additional reimbursement for an infection prevention type 9, 10, 11. The number of bacteria isolated from all cultures submitted to the hospital during the study period, including blood, sputum, urine, spinal fluid, ascites/pleural fluid, open pus, non-open pus, bile, endotracheal aspirate, various drain effluents, joint fluid, and stool, were extracted every month. In cases of multiple bacterial culture tests on the same patient, only the first detected bacteria within 90 days were counted, even if they were from different specimens. Among the bacteria registered in the database, those that could be pathogens in community-acquired infections were included in the study in order to assess whether COVID-19 affected the transmission of bacterial infections. The observation period was from January 2019 until December 2020, before and after the early stage of COVID-19 pandemic, respectively. Outcome The primary endpoint of this study was the detection ratios of isolated target bacteria. Since the number of participating institutions could vary by year, it was assumed that basing the outcome on the number of institutions would affect the outcome. Therefore, the number of each bacterium isolated per institution and for each month was calculated, with January 2019 set at “1”, and the ratios for the other months compared to the set January 2019 date. The same ratios as above for all bacterial species were used as the control because the present outcome could be affected by fluctuations in the total number of bacterial isolates due to medical care pressure caused by COVID-19 and refusal to see patients. In this study, We analyzed all the bacterial species registered in the database (Supplemental Table. 1). The protocol for bacterial culture was based on the guidelines of the Clinical and Laboratory Standards Institute. There were no changes in the guidelines regarding the method of bacterial isolation during the observation period. The early phase of the COVID-19 pandemic After the first outbreak of COVID-19 was reported in Wuhan, China, in December 2019, the first case of severe acute respiratory syndrome coronavirus 2 infections was confirmed in Japan on January 15, 2020. In early February 2020, 712 infected persons with 14 deaths were recorded on the Diamond Princess cruise ship that was having 3,700 onboard. This sensitive event was reported on TV every day, and it was one of the most significant events that increased awareness and behavioral change toward COVID-19 infection prevention measures 1. At the same time, the Ministry of Health, Labor, and Welfare requested pharmacies and supermarkets in Japan to restrict purchases of masks and disinfectants due to shortages12. The number of newly infected people continued to increase in March, and the Japanese government declared a state of emergency in all prefectures on April 16. After that, the number of infected people started to decrease, and the emergency declaration was lifted in all prefectures on May 25. Therefore, in this study, we set the period from February 2020, when awareness and behavioral changes started, to May 2020, when the emergency ban was lifted, as the early phase of COVID-19, based on previous reports. Statistical Methods: Interrupted time series analysis We used the ITSA to estimate changes in the level and the pre-existing trends of the study outcomes during the early phase of the COVID-19 pandemic while controlling for pre-pandemic levels and trends. We estimated the following linear segmented regression model:Yt=β0+β1(Time)+β2(Exposure)+β3(Exposure*time)+εt where Yt is the outcome at time point t, time indicates the duration from the start of the study, exposure is a dummy variable representing the periods before and after the early phase of the COVID-19 pandemic, and εt is the error term. In this model, β0 estimates the baseline level of each bacterium at the beginning of the study, β1 represents the underlying pre-early phase of the COVID-19 trend, and β2 and β3 estimate the immediate level change and slope change following the early phase of COVID-19, respectively. To account for the lag between the initial effect of COVID-19, we considered a 3-month “phase-in” period, and three monthly data points for this period were excluded from the analysis. Therefore, after the early phase of COVID-19, we had 13 monthly data points from January 2019 to January 2020 and eight monthly data points from May 2020 to December 2020. We used the “ITSA” command in STATA (Version 16SE; Stata Corporation, College Station, TX, USA), which estimates the model using ordinary least squares (OLS) with Newey-West methods. We set the lag period as 12 months because of the seasonality. A significant threat to the validity of the ITSA is the history of events that occurred concurrently with the exposure, which may have caused changes in the observed outcomes. To account for this threat, we assessed changes in all the bacteria and compared them to the total number so that the phase-in periods was not expected to be influenced. Ethics Since this was an observational study using the AMR control database, which has already been compiled, and not involving human participants, we determined that ethical review was unnecessary according to the Ethical Guidelines for Medical and Health Research Involving Human Participants in Japan. Results The analysis involved 369,210 isolates over 24 months from January 2019 to December 2020 for the 21 strains. The number of isolates per month is presented in Supplemental Table 1. Figs. 1 and 2 show plots of the ratios of the bacteria isolates included. From the 21 species registered in the database, we selected 10 species that could cause community-acquired infections in healthy individuals. The ratios of the isolates detected in January 2020, just before the early phase of COVID-19; May 2020, just after the early phase of COVID-19; and December 2020, at the end of the observation period, were as follows: Streptococcus pneumoniae (S. pneumoniae): 0.633, 0.218, and 0.224; Haemophilus influenzae (H. influenzae): 0.763, 0.253, and 0.252; Streptococcus pyogenes (S. pyogenes): 0.856, 0.337, and 0.27; Staphylococcus aureus (S.aureus): 1.12, 0.79, and 0.872, and Escherichia coli (E.coli): 0.897, 0.718, and 0.758, respectively. Trends in other species of bacteria are described in Supplemental Table 2. Differences in the level and slope change Fig. 3 shows the differences in the level and slope change for each bacteria isolate relative to those of the control group. The level changes of S. pneumoniae, H. influenzae, and S. pyogenes, relative to that of the control group, significantly decreased after the early phase of COVID-19, with the differences in level change being -0.272 (95% confidence interval [CI]; -0.352 to -0.192), -0.244 (95% CI; -0.314 to -0.174), and -0.324 (95% CI; -0.589 to -0.06), respectively. On the other hand, the level changes of Escherichia coli (E. coli) and Enterococcus faecalis (E. faecalis), relative to that of the control group, significantly increased after the early phase of COVID-19, with the differences in level change being 0.077(95% CI; 0.029 to 0.125) and 0.084 (95% CI; 0.031 to 0.136), respectively. Meanwhile, no significant difference in slope change compared to the control group was observed for any bacteria. The differences in the level and slope change for other species of bacteria are described in Supplemental Fig. 1.Fig. 3 Forest plots showing differences in the level and slope change between each bacteria isolate and control group. Bars in the left forest plot indicates difference in the level change with 95% confidence interval between each bacteria isolate and control group. The level for S. pneumoniae, H. influenzae, and S. pyogenes after the phase-in periods were significantly lower than those of the control group. Bars in the right forest plot indicates the slope change with 95% confidence interval between each bacteria isolate and control group. S. pyogenes: Streptococcus pyogenes, S. pneuimoniae: Streptococcus pneumoniae, H. influenzae: Haemophilus influenzae, P. aeruginosa: Pseudomonas aeruginosa, K. pneumoniae: Klebsiella pneumoniae, CNS: coagulase-negative staphylococci, S. agalactiae: Streptococcus agalactiae, S. aureus: Staphylococcus aureus, E. coli: Escherichia coli, E. faecalis: Enterococcus faecalis, CI: Confidence interval. Fig 3 Fig. 1, Fig. 2Fig. 1 The ratios of Streptococcus pneumoniae (S. pneumoniae), Haemophilus influenzae (H. influenzae), Streptococcus pyogenes (S. pyogenes), and Streptococcus agalactiae (S. agalactiae) detected were compared with those of the control group for the period between January 2019 and December 2020. Fig 1 Fig. 2 The ratios of intestinal microflora and indigenous skin bacteria detected were compared with those of the control group between January 2019 and December 2020. Fig 2 Discussion This study analyzed the impact of the early phase of the COVID-19 pandemic on bacteria isolates using the large-scale database of J-SIPHE from January 2019 to December 2020. The four months, from February 2020 to May 2020, were designated as the early phase of the COVID-19 pandemic. The detection ratios of S. pneumoniae and H. influenzae, causing community-acquired pneumonia, and S. pyogenes, causing tonsillitis, showed significant differences in level changes compared to that of the control group of -0.272, -0.324, and -0.244, respectively, and these bacteria were mainly transmitted by droplet infection. We could not identify any differences in level changes compared to that of the control group in the detection ratios of intestinal microflora, indigenous skin bacteria, or S. agalactiae. S. pneumoniae, H. influenzae, and S. pyogenes are causative bacteria of respiratory tract infections, but they are also causative bacteria of invasive bacterial infections with high mortality rates in people of all ages. In this study, the number of causative bacteria isolates decreased, as well as the number of cases, which has been pointed out in previous reports2, 3, 4. Therefore, the decrease in the number of bacteria isolates was behind the decrease in the number of cases of these infections. Among these bacteria, S. pneumoniae and H. influenzae are the first and second most common causative bacteria of community-acquired pneumonia, accounting for 25% and 10%, respectively, in a survey conducted before the COVID-19 pandemic in Japan 13. S. pneumoniae causes fatal bacterial infection in older and immunocompromised adults when it causes community-acquired pneumonia; meanwhile, the target populations for vaccination in Japan are infants, older adults, and high-risk patients with underlying diseases. It is also known to be carried in the nasopharynx of 40-60% of infants and is responsible for several subclinical infections 14. During the period of awareness and behavioral changes, which coincides with the early phase of the COVID-19 pandemic, containment was not enforced as a policy, and voluntary home confinement and the closure of public schools nationwide may have had an impact 15. Awareness and behavioral changes have been reported to result in a marked decrease in pediatric visits 16. Other specific examples of personal infection prevention measures practiced were “keeping social distance,” “washing hands,” and “wearing a mask.” A Japanese Internet survey on infection prevention measures showed that more than three-quarters of the respondents took preventive measures 1. In addition to the impact of awareness and behavioral changes, the activities of RS and influenza viruses were lower than in previous years 17, 18, 19, 20, 21, 22, 23, suggesting that the prevalence of respiratory viral infections other than COVID-19, as well as secondary bacterial pneumonia, decreased. Meanwhile, a little difference in the level change in the contact route of infection compared to all bacterial species was observed. These may be partially due to endogenous transmission by bacterial species that are primarily contact-transmitted with little impact on behavioral change. However, as for endogenous infections, the impact of such infections cannot be assessed because the proportion of endogenous infections and other transmission route among each species of bacteria may varies. Thus, it is possible that there was a certain impact for each species of bacteria. On the other hand, there was a decrease in the ratios of total isolated cultures of bacterial species (Supplemental Fig.2). The level change was significantly decreased before and after the phase-in periods at -0.2 (95% CI; -0.27 to -0.13). This may be due to the combination of various effects, such as infection control, behavioral restrictions, and refraining from medical visits, therefore, the ratio of total isolates was used as control group in this study, and we compared the differences in levels and slopes between the control and each bacterium to identify the bacteria that were significantly affected. A snap poll conducted in 28 countries showed that 71% of the respondents frequently washed their hands, as one of the infection prevention measures for COVID-19 in Japan 24. However, the following considerations are important for the results that did not show a relative decrease: hand washing may have been inadequate; hand washing alone did not directly affect the reduction of bacteria, and long-term evaluation may be necessary. Therefore, it is impossible to make a definitive statement about the specific changes in awareness and behavior that directly affected the bacteria, as a combination of factors may be suggested. However, there was no relative decrease in the number of bacteria isolates mainly due to contact infection, and the relative decrease in the number of bacteria mainly transmitted through droplet infection suggested that “keeping social distance” and “wearing masks” have a specific impact at the individual level. Therefore, awareness and behavioral changes may be leveraged to prevent respiratory bacterial infections in people who cannot take vaccines for various reasons or in older adults at risk of severe infections. This study included patients with all diseases for which hospitalization was indicated for bacterial culture testing at a group of major acute care hospitals in Japan. The validity and reliability of the test results were considered high because these facilities performed the test according to the guidelines of the Clinical and Laboratory Standards Institute, which are used globally. There are several limitations in this study. The limitations of this study include the fact that we were unable to analyze each culture specimen by the type of bacteria because we did not know the number of bacteria isolates for each type of culture specimen. In addition, the absolute number of hospital admissions and culture tests, including sputum tests, may have decreased due to the COVID-19 pandemic because patients could not receive appropriate medical care and undergo appropriate tests due to the reduced workforce and vacant beds. Moreover, there might be an bias due to the increase of false negative in culture tests due to the change of the community-prescribing oral antimicrobial agents. However, it is unlikely that the participating facilities did not perform culture tests because they are hospitals actively implementing measures against antimicrobial resistance and have infectious disease physicians who monitor the administration of unnecessary antimicrobial agents and conduct appropriate bacterial culture tests. In addition, a decrease in the total number of inpatients has been ruled out in previous reports 2. Regarding the possibility of a decrease in the number of respiratory culture specimens submitted, no decrease in the number of specimens submitted were found between 2019 and 2020 in the J-SIPHE database (Supplemental Table 3). Therefore, the impact of changes in the number of submitted specimens after the COVID-19 pandemic was considered negligible. Although culture-positive specimens were analyzed by ITSA, the total number of cultures and the number of culture-negative specimens were unknown, and some effect on the results of the analysis that this may have had, cannot be ruled out. Regarding concerns about bias that false negative culture test may increase due to changes in oral antimicrobial dosing, the total antimicrobial dosing in Japan during the year 2020 decreased compared to 2019 25, which would contribute toward underestimating the results. Finally, the data were compiled by open cohort, and because the data on the number of beds per institution were not available, we used the average number of bacteria isolates by participating institutions in each month. However, it is possible that the number of beds per institution may differ, and that the number of beds at participating institutions may be non-uniform. In conclusion, we conducted a longitudinal retrospective cohort study of bacterial culture test isolates from inpatients using a large-scale nationwide database and compared the bacterial pre- and post-early phase of the COVID-19 pandemic. During the COVID-19 pandemic, the detection ratio of bacteria, transmitted through droplet infections, decreased, and maintained the same level. Our results indicate the changes in public health awareness and the behavior of the public and social communities have a significant impact on bacterial trends. Declaration of Competing Interest The authors declare no conflicts of interest associated with this manuscript. Appendix Supplementary materials Image, application 1 Image, application 2 Image, application 3 Image, application 4 Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Author contributions MK and DM developed the study design and analyzed the data. SK extracted the data. MK. and DM wrote the manuscript. KI, TK, KO, HO, and MI provided advice on the interpretation of data and revised the manuscript. All authors read and approved the manuscript. Acknowledgements we thank the amr clinical reference center, national center for global health and medicine for access to the database used for this study Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jinf.2022.11.025. ==== Refs References 1 Muto K Yamamoto I Nagasu M Tanaka M Wada K. Japanese citizens’ behavioral changes and preparedness against COVID-19: An online survey during the early phase of the pandemic PloS One 15 2020 10.1371/journal.pone.0234292 2 Nagano H Takada D Shin JH Morishita T Kunisawa S Imanaka Y. Hospitalization of mild cases of community-acquired pneumonia decreased more than severe cases during the COVID-19 pandemic Int J Infect Dis 106 2021 323 328 10.1016/j.ijid.2021.03.074 33794382 3 Brueggemann AB van Rensburg MJ Shaw D McCarthy ND Jolley KA Maiden MCJ Changes in the incidence of invasive disease due to Streptococcus pneumoniae, Haemophilus influenzae, and Neisseria meningitidis during the COVID-19 pandemic in 26 countries and territories in the Invasive Respiratory Infection Surveillance Initiative: a prospective analysis of surveillance data Lancet Digit Health 3 2021 e360 e370 10.1016/s2589-7500(21)00077-7 34045002 4 Lastrucci V Bonaccorsi G Forni S D'Arienzo S Bachini L Paoli S. The indirect impact of COVID-19 large-scale containment measures on the incidence of community-acquired pneumonia in older people: a region-wide population-based study in Tuscany, Italy Int J Infect Dis 109 2021 182 188 10.1016/j.ijid.2021.06.058 34216731 5 Harris AD McGregor JC Perencevich EN Furuno JP Zhu J Peterson DE The use and interpretation of quasi-experimental studies in medical informatics J Am Med Inform Assoc 13 2006 16 23 10.1197/jamia.m1749 16221933 6 Bernal JL Cummins S Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial Int J Epidemiol 46 2017 348 355 10.1093/ije/dyw098 27283160 7 AMR Clinical Reference Center, Japan. Japan Surveillance for Infection Prevention and Healthcare Epidemiology: J-SIPHE. Available at https://j-siphe.ncgm.go.jp/. Accessed 8 May 2021. 8 Ohmagari N. National Action Plan on Antimicrobial Resistance (AMR) 2016-2020 and relevant activities in Japan Global Health Med 1 2019 71 77 10.35772/ghm.2019.01017 9 Maeda M Muraki Y Kosaka T Yamada T Aoki Y Kaku M. Impact of health policy on structural requisites for antimicrobial stewardship: A nationwide survey conducted in Japanese hospitals after enforcing the revised reimbursement system for antimicrobial stewardship programs J Infect Chemother 27 2021 1 6 10.1016/j.jiac.2020.09.015 33008739 10 Mi Maeda Muraki Y Kosaka T. Yamada T Aoki Y Kaku M The first nationwide survey of antimicrobial stewardship programs conducted by the Japanese Society of Chemotherapy J Infect Chemother 25 2019 83 88 10.1016/j.jiac.2018.11.001 30473181 11 Ministry of Health, Labour Standards: For Foreign Workers in Japan (Information on Labour Standards) Available at: https://www.mhlw.go.jp/seisakunitsuite/bunya/kenkou_iryou/iryouhoken/dl/chousa_03-37.pdf (in Japanese) Accessed 20 September 2021. 12 Sakamoto H Ishikane M Ueda P. Seasonal Influenza Activity During the SARS-CoV-2 Outbreak in Japan JAMA 323 2020 1969 1971 10.1001/jama.2020.6173 32275293 13 Otsuka T Chang B Shirai T. Iwaya A Wada A Yamanaka N Individual risk factors associated with nasopharyngeal colonization with Streptococcus pneumoniae and Haemophilus influenzae: a Japanese birth cohort study Pediatr Infect Dis J 32 2013 709 714 10.1097/inf.0b013e31828701ea 23411622 14 Peto L Nadjm B Horby P. Ngan TT van Doorn R Kinh NV The bacterial aetiology of adult community-acquired pneumonia in Asia: a systematic review Trans R Soc Tro Med Hyg 108 2014 326 337 10.1093/trstmh/tru058 15 Karako K Song P Chen Y Tang W Kokudo N. Overview of the characteristics of and responses to the three waves of COVID-19 in Japan during 2020-2021 Biosci Trends 15 2021 1 8 10.5582/bst.2021.01019 33518668 16 Sekine I Uojima H Koyama H. Kamio T Sato M Yamamoto T Impact of non-pharmaceutical interventions for the COVID-19 pandemic on emergency department patient trends in Japan: a retrospective analysis Acute Med Surg 7 2020 10.1002/ams2.603 17 Wagatsuma K Koolhof IS Shobugawa Y Saito R. Decreased human respiratory syncytial virus activity during the COVID-19 pandemic in Japan: an ecological time-series analysis BMC Infect Dis 21 2021 10.1186/s12879-021-06461-5 18 Baker RE Park SW Yang W Vecchi GA Metcalf CJ Grenfell BT. The impact of COVID-19 nonpharmaceutical interventions on the future dynamics of endemic infections Proc Natl Acad Sci U S A 117 2020 30547 30553 10.1073/pnas.2013182117 33168723 19 Friedrich F Ongaratto R Scotta MC Veras TN Stein RT Lumertz MS Early impact of social distancing in response to coronavirus disease 2019 on hospitalizations for acute bronchiolitis in infants in Brazil Clin Infect Dis 72 2021 2071 2075 10.1093/cid/ciaa1458 32986818 20 Soo RJ Chiew CJ Ma S Pung R Lee V. Decreased influenza incidence under COVID-19 control measures, Singapore Emerg Infect Dis 26 2020 1933 1935 10.3201/eid2608.201229 32339092 21 Kuo SC Shih SM Chien LH Hsiung CA. Collateral benefit of COVID-19 control measures on influenza activity Taiwan. Emerg Infect Dis 26 2020 1928 1930 10.3201/eid2608.201192 32339091 22 Sakamoto H Ishikane M Ueda P. Seasonal influenza activity during the SARS-CoV-2 outbreak in Japan JAMA 323 2020 1969 1971 10.1001/jama.2020.6173 32275293 23 Olsen SJ Azziz-Baumgartner E Budd AP Brammer L Sullivan S Pineda RF Decreased influenza activity during the COVID-19 pandemic - United States, Australia, Chile, and South Africa, 2020 MMWR Morb Mortal Wkly Rep 69 2020 1305 1309 10.15585/mmwr.mm6937a6 32941415 24 Gallup International Association. 2020 March. The coronavirus: a vast scared majority around the world. Available at: https://www.gallup-international.com/fileadmin/user_upload/surveys/2020/GIA_SnapPoll_2020_COVID_Tables_final.pdf Accessed 22 October 2021. 25 25 AMR Clinical Reference Center. Surveillance of antibiotic sales in Japan. Available at: https://amrcrc.ncgm.go.jp/surveillance/020/20200813162318.html Accessed 10 November 2022.	36463984	PMC9715262	NO-CC CODE	2022-12-14 23:52:35	no	J Infect. 2022 Dec 2; doi: 10.1016/j.jinf.2022.11.025	utf-8	J Infect	2,022	10.1016/j.jinf.2022.11.025	oa_other
==== Front Res Int Bus Finance Res Int Bus Finance Research in International Business and Finance 0275-5319 1878-3384 Elsevier B.V. S0275-5319(22)00210-0 10.1016/j.ribaf.2022.101824 101824 Full length Article Does economic policy uncertainty drive the dynamic spillover among traditional currencies and cryptocurrencies? The role of the COVID-19 pandemic Al-Shboul Mohammad a⁎ Assaf Ata bc Mokni Khaled d a Department of Finance and Economics, School of Business Administration, University of Sharjah, United Arab Emirates b Faculty of Business and Management, University of Balamand, P.O.Box: 100 Tripoli, Lebanon c Cyprus International Institute of Management (CIIM), Cyprus, 2151 Nicosia, P. O. Box 20378 d Institut Supérieur de Gestion de Gabès, Gabès Université, Gabès 6002, Tunisia ⁎ Corresponding author. 2 12 2022 2 12 2022 1018243 5 2022 12 11 2022 23 11 2022 © 2022 Elsevier B.V. All rights reserved. 2022 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. The paper examines the dynamic spillover among traditional currencies and cryptocurrencies before and during the COVID-19 pandemic and investigates whether economic policy uncertainty (EPU) impacts this spillover. Based on the TVP-VAR approach, we find evidence of spillover effects among currencies, which increased widely during the pandemic. In addition, results suggest that almost all cryptocurrencies remain as “safe-haven” tools against market uncertainty during the COVID-19 period. Moreover, comparative analysis shows that the total connectedness for cryptocurrencies is lower than for traditional currencies during the crisis. Further analysis using quantile regression suggests that EPU exerts an impact on the total and the net spillovers with different degrees across currencies and this impact is affected by the health crisis. Our findings have important policy implications for policymakers, investors, and international traders. Graphical abstract Keywords Economic policy uncertainty dynamic spillover traditional currencies TVP-VAR cryptocurrency COVID-19 ==== Body pmc1 Introduction In the aftermath of the 2008 global financial crisis (GFC), economic policy uncertainty (EPU) has raised remarkably, suggesting an increasing influence on the financial markets, especially the foreign exchange markets (Krol, 2014, Arouri et al., 2016, Balcilar et al., 2016, Kido, 2016, Beckmann and Czudaj, 2017, Abid, 2020). In this regard, Baker, Bloom, and Davis (2016) invented new economic uncertainty indices using the information from newspaper articles for major economies, which have been popularly used in the recent literature. For instance, Krol (2014) and Chen, Du, and Hu (2020) argued that economic policy uncertainty adversely affects foreign currency values. Other studies argued that cryptocurrencies are also influenced by changes in EPU (Yen and Cheng, 2020; Demir, Gozgor, Lau, and Vigne, 2018; Paule-Vianez, Prado-Román, and Gómez-Martínez, 2020; Mokni, 2021; Elsayed et al., 2022). Further studies suggest that cryptocurrencies are not only used as means of exchange but also act as investment tools and safe havens to protect against economic uncertainty (Wu et al., 2019, Wu et al., 2021, Paule-Vianez et al., 2020, Hasan et al., 2021, Colon et al., 2021, Mokni et al., 2022). As investors in foreign currency markets have various choices of investment opportunities in different currencies and/or currency futures contracts, any change in the value of a particular currency (or cryptocurrency) might transmit to the value and/or price volatility of other currencies, including cryptocurrencies, leading to a significant effect on portfolio investments, and diversification benefits. The higher connectedness among currencies within each currency system or among currencies in the other system is the higher information transmission among currencies. Therefore, studying the connectedness among both types of currencies can help identify how extreme volatilities and trends in one type of currency are transmitted within and across others. Such an analysis can also be useful for traders and policymakers to make suitable decisions and investors regarding portfolio diversification benefits and hedging opportunities, especially in periods of high uncertainty. The degree of connectedness between traditional and digital currencies may also result from the differences in the characteristics of these two kinds of assets. Like traditional currencies, cryptocurrencies may function as a store of value, a medium of exchange, and a unit account. However, unlike traditional currencies, cryptocurrencies are not issued by central banks and can be transferred electronically between users without the involvement of intermediaries (Andrada-Félix et al., 2020, Bech and Garratt, 2017). They can also generate revenues similar to any regular goods and services and rely on the cryptographic integrity of their network, while traditional currencies depend on political and legal systems for value and legitimacy. Therefore, the difference in the characteristics of these two types of assets would lead to a difference in the degree of connectedness of currencies among each other and within every type of currency. If cryptocurrencies serve as a medium of exchange, there would be systematic traditional currency exposures in cryptocurrencies; when volatility connectedness among traditional currencies is high, this may generate extreme volatility transmission in cryptocurrencies, which in turn might be transmitted within cryptocurrencies and across traditional currencies as well. In the last few years, we observed a significant increase in the cryptocurrency market, accompanied by a significant rise in the demand for traditional currencies, especially over the ongoing COVID-19 pandemic. However, the increase in the demand for both types of currencies has led to high variations in exchange rates and the value of cryptocurrencies. Both currencies are still considered valuable investment opportunities, either in speculation, hedging against risk, or portfolio diversification. In addition, cryptocurrencies are considered as a solution for unexpected economic and financial structure changes (Demir, Gozgor, Lau, & Vigne, 2018). Given the increasing importance of cryptocurrencies as an investment vehicle, modeling cryptocurrencies connectedness becomes necessary for making investment decisions and risk management (e.g., Katsiampa, 2019; Bouri, Molnár, Azzi, Roubaud and Hagfors, 2017; Bouri, Saeed, Vo, and Roubaud, 2021). Most of the literature has reported evidence of connectedness among traditional currencies and/or cryptocurrencies (e.g., Kitamura, 2010; Baruník et al., 2017; Sehgal, Pandey, and Diesting, 2017; Mokni and Ajmi, 2020). However, the literature has left several research gaps. The different studies have generally reached mixed results and inconclusive evidence of such connectedness. Several studies have focused mostly on the connectedness among one-one traditional currency and/or a few currencies, the world's major traditional currencies (Mai et al., 2018, Sehgal et al., 2017). Other studies only examined the connectedness among cryptocurrencies (Wei et al., 2020, Mai et al., 2018, Kumar and Anandarao, 2019, Mensi et al., 2019, Aslan and Sensoy, 2020), and largely concentrated on Bitcoins and Ethereum. Moreover, another stream of studies examined the connectedness among traditional or cryptocurrencies with other financial assets, including commodities, bonds, oil prices, and stocks (Boako and Alagidede, 2017, Grobys, 2015, Malik and Umar, 2019, Antonakakis and Kizys, 2015, Hassen et al., 2022, Cao and Xie, 2022, Conlon et al., 2020). Given the above-uncovered research gaps, it can be seen that the dynamic connectedness among traditional currencies and cryptocurrencies is extremely understudied and still an attractive research venue. Another issue which has been overlooked by the literature is the effect of EPU on the degree of connectedness among traditional currencies and cryptocurrencies. However, although the evidence of the effect of EPU on either type of currency is well established, there is still a need to re-examine such an effect due to the limited research effort on this topic. Mokni, Ajmi, Bouri, and Vo (2020) have examined the effect of EPU on the connectedness among either type of currency but without taking into account the impact of the COVID-19 crisis. Although the effect of EPU has been examined by other studies (Chen et al., 2020, Huynh et al., 2020), they focused only on examining the effect on the connectedness among either cryptocurrencies or traditional currencies, not both. Therefore, it is of great interest to examine the effect of EPU on the connectedness among cryptocurrencies and fiat currencies during the COVID-19 outbreak. The COVID-19 pandemic has caused a higher exchange rate risk exposure to many currencies, which has led to high demand for hard currencies as well as for cryptocurrencies, leading to an appreciation of these currencies. On the other hand, over the COVID-19 period, exchange rates and/or the global stock markets have significantly fluctuated with high downturns in the global economies (Narayan et al., 2020; Njindan, 2020; So, Chu, and Chan, 2021; Samitas, Kampouris, and Polyzos, 2022; Cheng et al., 2022; Guo et al., 2021). Given the negative impacts of the health crisis on the global economy and major global currencies as well as on cryptocurrencies, there has been a limited research investigation directed toward the relationship between EPU and the dynamic connectedness among traditional currencies and cryptocurrencies during the COVID-19 crisis. Therefore, exploring the impact of economic policy uncertainty (EPU) on the dynamic connectedness among currencies during the COVID-19 period is still an appealing research issue. The main objectives of the paper are: 1) analyzing the dynamic connectedness among traditional currencies (the Euro (EUR), the British Pound (UK), the Japanese Yen (JPY), the Indian Rupee (IND), Swiss Franc (SFR), Korean Won (KOR), Chinese Yuan (CHIN), and the Canadian Dollar (CAN)), and cryptocurrencies (Bitcoin (BIT), Ethereum (ETH), Ripple (RIP) and Litecoin (LIT)) and 2) whether such connectedness, among both types of currencies, is affected by uncertainty factors under different levels of spillovers’ distribution. To be specific, this paper contributes to the literature in two folds. First, it analyzes the dynamic connectedness and spillover effects among traditional currencies before and during the COVID-19 crisis. To perform the analysis, several currencies and cryptocurrencies are selected based on long-term trading history and according to their market capitalization. To measure the dynamic connectedness among currencies, the connectedness index and its variant are generated as proposed by Diebold and Yilmaz, 2009, Diebold and Yilmaz, 2012, Diebold and Yılmaz, 2014. This index provides a clear insight into comparing the extent and the nature of interdependencies and spillover between both types of currencies. The variants of the index are based on the time-varying parameter vector autoregressive (TVP-VAR) framework. The TVP-VAR approach is based on a multivariate Kalman filter and is less sensitive to the presence of outliers (e.g., Antonakakis et al., 2019; Gabauer and Gupta, 2018). This approach is more sophisticated than that based on the rolling-window approach. The latter approach outcomes some weaknesses related to the choice of the window size and the loss of information in calculating the dynamic measures of connectedness. We also apply the generalization of the connectedness among these two groups of currencies. Second, our analysis includes the most recently updated data covering the ongoing COVID-19 crisis, which can detect the possible effect of this outbreak on the shock transmission between these two types of currencies. Our updated data sample covers the period between the beginning of 2017 and early October 2022. Third, our study analyzes the potential effect of the economic policy uncertainty index (EPU) on the dynamic connectedness among traditional currencies before and during the COVID-19 pandemic in different levels of dynamic spillovers. To perform this analysis, we implement a quantile regression model to investigate the possible effect of the EPU on the dynamic connectedness to evaluate such an effect for different levels of spillover indices’ distribution. Therefore, this analysis enables us to assess the sensitivity of the information flows between currencies to those uncertainty factors under different phases of connectedness. Our results are summarized as follows. We find strong evidence of dynamic spillover across currency markets, and the spillover level increased widely during the COVID-19 pandemic period. Specifically, almost all cryptocurrencies remain as “safe-haven” tools to hedge uncertainty, showing heterogeneous responses to the presence of the COVID-19 pandemic. The level of connectedness across traditional currencies is higher than the connectedness among cryptocurrencies during the crisis period. However, before the crisis, the connectedness for traditional currencies is lower than connectedness for cryptocurrencies. Based on the quantile regression analysis, we find that the total spillover index is negatively impacted by the economic policy uncertainty through the pre-COVID-19 and during COVID-19, indicating that total spillover among currencies is negatively and highly influenced by the rise in EPU in normal and high connectedness condition. However, EPU has different degrees of influence on the dynamic spillover across different conditions during the crisis. Although this impact fluctuates when the connectedness effects are low, our results show a decreasing pattern when the level of connectedness is medium or high. The total connectedness among all currencies becomes increasingly exposed to the negative impact of EPU in normal and high levels of connectedness. The rest of the paper is organized as follows: The review of the related literature is discussed in Section 2. In Section 3, the methodology and data are explained. The empirical analysis is reported and discussed in Section 4, and the concluding remarks are addressed in Section 5. 2 Literature review The literature examining the impact of EPU on the currency markets has been growing in recent years. However, the existing literature has provided conflicting conclusions. The recent studies can be categorized into two important strands of literature (see Appendix A for the summary of the literature review). The first strand of literature focused on volatility connectedness among global currency markets, mainly examining the volatility transmission among major global traded currencies. Using the M-GARCH model, Kitamura (2010) argued that volatility spillovers from the Euro drastically impact the Swiss franc and Japanese yen during the subprime global financial crisis (GFC) period. Baruník et al. (2017) found evidence of asymmetric volatility connectedness in the foreign exchange markets using a 2N-dimensional VAR model. The most actively traded currencies (AUD, GBP, CAD, EUR, JPY, and CHF) also reported that volatility's negative spillovers dominate positive ones. Positive spillovers are correlated with the subprime crisis (GFC), while negative spillovers are generally caused by the sovereign debt crisis in Europe. Using an empirical network model, Greenwood-Nimmo et al. (2016) examined the risk-return connectedness among the G10 currencies and reported evidence of time-varying connections between such currencies. During the periods of financial crises, volatility connectedness was increasing, and returns became more sensitive to risk measures. By using the VAR model, Salisu and Ayinde (2018) provided evidence of volatility transmission across the Naira and six most-traded currencies (the US Dollar, Euro, Pound Sterling, Yen, Swiss Franc, and the West African Unit of Account (WAUA)), indicating that the election process in Nigeria led to greater spillover effects on the Naira than the GFC. In the context of the Asian markets, Shu, He, and Cheng (2015) found evidence of transmission effects from the Chinese currency on the Asia-Pacific countries’ currencies. The offshore currency exchange of the Chinese currency was found to exert more effect on Asian currencies than the onshore Chinese currency market due to China's monetary policy transmission to the region. Sehgal, Pandey, and Diesting (2017), using constant and time-varying Copula-GARCH models, found that currencies of the South Asian member countries, except for India and Nepal/Bhutan, were weakly connected due to the poor levels of intra-regional trade intensity and portfolio flows. More recently, Wei, Luo, Huang, and Guo (2020) use the time-varying spillover model and showed evidence of volatility spillover effects among "the Belt and Road" currency market during the regional and global crises. The spillover in the RMB exchange rate was affected by internal financial reforms, as well as external economic shocks, but the spillover system for such a market was disrupted during the COVID-19 period. Based on the Correlation matrices (CM) and the information flow graph method, Mai, Chen, Zou, and Li (2018) argued that, in a global context, the correlations of currency rate volatility are present in the network. They also found that the US Dollar dominates the global foreign currency markets while the Euro greatly impacts European currencies. The East Asian currencies were more strongly correlated than the European currencies due to the strong co-movement of currencies in the East Asian region. By implementing the four major cryptocurrencies (Bitcoin, Ethereum, Ripple, and Litecoin), Kumar and Anandarao (2019) found evidence of volatility spillover from Bitcoin to Ethereum and Litecoin, using the multivariate IGARCH-DCC and the pairwise wavelet cross-spectral analysis. In the European region, Antonakakis (2012) found evidence of return co-movements and volatility spillovers between major foreign currency rates before the introduction of the Euro. Yet, this evidence is weak in the post-euro period. Specifically, the Euro was a net transmitter of volatility, while the British pound was a net receiver of volatility in both periods. Using the GARCH and BVAR models, Orlowski (2016) showed evidence of high positive co-movements between the Euro and the non-Euro currencies, suggesting a strong substitution between these currencies and the Euro in foreign exchange markets which significantly rose during the GFC. Kočenda and Moravcová (2019) reported significant differences in the extent of currency co-movements during various periods of financial distress. They concluded that all three currencies in the new EU foreign exchange markets brought hedging benefits during crisis periods but at different costs. During the crisis, volatility spillovers among currencies rose significantly, and the Hungarian currency showed a leading role. Using the LASSO-VAR approach to examine the volatility connectedness among 65 major global currencies, Wen and Wang (2020) found that the US dollar and Euro exhibit volatility transmission, while other currencies, including the Japanese yen and British pound, exhibit evidence of volatility receivers. They argued that volatility connectedness responded sensitively to changes in international economic fundamentals and increased during crisis periods (oil price crashes, exchange rate regimes, and monetary policy changes). However, the directional volatility connectedness of the Renminbi significantly declined after the reforms of the Chinese exchange rate regime (which shifts from a USD-pegged exchange rate regime to a managed floating exchange rate regime)1 . Fung, Jeong, and Pereira (2022) found evidence of volatility persistence with negative leverage effects among cryptocurrencies’ return behavior using the GARCH family models. By analyzing the connectedness between six traditional currencies, and six cryptocurrencies using the quantile cross-spectral approach, Baumöhl (2019) showed evidence of significant negative dependencies between currencies and cryptocurrencies in both the short- and long-term horizons, suggesting that investors can diversify their portfolio by using a mixture of these two asset groups. By examining the effect of variation in the Chinese currency regulatory management system on the co-movement between the Chinese currency (RMB) with regional and other emerging market currencies after the three post-reform periods of RMB management (transition, basket of currencies management, and countercyclical management), McCauley and Shu (2019) revealed that the co-movement with regional and Latin American currencies peaked in the basket period while it declined after the countercyclical period. Using a diagonal BEKK model, Hsu, Sheu, and Yoon (2021) argued that cryptocurrencies acted as a safe haven during the COVID-19 pandemic against risk spillover. The other strand of studies deals with the effect of economic policy uncertainty and/or trade uncertainty on currencies. In general, previous studies on this issue did not succeed in presenting clear-cut evidence of the dynamic connectedness between currencies and Bitcoin in the presence of economic policy uncertainty before and during the COVID-19 pandemic. Mokni, Ajmi, Bouri, and Vo (2020) argued that EPU is negatively associated with the dynamic conditional correlations between Bitcoin and the US stock markets only after the crash of Bitcoin in December 2017, highlighting the usefulness of using Bitcoin as a hedging instrument to the stock portfolio. By exploring the connectedness among the nine US dollar exchange rates of globally traded currencies in association with the effect of trade policy uncertainty, Huynh, Nasir, and Nguyen (2020) found evidence of connectedness among the US dollar exchange rates when trade policy uncertainty is present. Regarding the Chinese case, Chen, Du, and Hu (2020) reported that economic policy uncertainty positively affects China’s exchange rate volatility on all quantiles. Alam, Shahzad, and Ferrer (2019) argued that oil prices are strongly connected to the exchange rates of six major bilateral currencies against the US dollar during the GFC period and the European sovereign debt crisis. Further, Mokni and Ajmi (2020) provided evidence of the linkage between the top-five cryptocurrencies and the US dollar under different market conditions, especially before and during the COVID-19 period. During the COVID-19 crisis, the connectedness between the US dollar and cryptocurrencies was discovered at higher and lower tails of the distribution, and the predictive power of the US dollar was lost in favor of cryptocurrencies, leading cryptocurrencies to be good predictors and acting as hedging instruments against the US dollar variation. When adopting the Quantile-VAR approach, Al-Shboul, Mokni, and Assaf (2022) argued that the impact of cryptocurrency policy and price uncertainties on the dynamic connectedness among cryptocurrenciesrose widely during the COVID-19 outbreak. Bouri et al. (2021) also found that the connectedness among cryptocurrencies in the lower and upper quantiles is substantially higher than those in the mean and median of the conditional distribution, confirming that connectedness increases with an increase in shock size either for both positive or negative shocks. Another study by Kumar, Iqbal, Mitra, Kristoufek, and Bouri (2022) reported evidence of a presence of structural change in the connectedness evolving in 2020 as a result of the COVID-19-pandemic. Overall, the above-referenced existing studies have not fully emphasized the connectedness among currencies, especially major global currencies. Although the existing studies have examined the effect of policy factors (e.g., exchange rate regimes, central bank interventions under inflation or exchange rate targeting, capital controls, and forex reserves), and economic factors (e.g., financial crisis, trade linkage, oil exports, investor behavior, and trade deficits) on volatility transmission across currencies, they failed to provide a comparative analysis between connectedness among the two types of currencies as an example. On the other hand, previous studies in this context have overlooked the effect of economic uncertainty. More precisely, they failed to report clear-cut evidence of the direction of the relationship between economic policy uncertainty and exchange rates. Few studies concluded that economic policy uncertainty has a negative effect on currency markets, while others argued that economic policy uncertainty positively impacts currency markets. Second, there is a general paucity in the literature concerning the impact of economic uncertainty on the connectedness among currencies during normal and distressed market conditions. Third, although the impact of economic policy uncertainty on currency markets is well documented in the literature, existing studies have noticeably overlooked the examination of the effects of economic uncertainty on the connectedness among major traditional currencies with leading cryptocurrencies during the COVID-19 period where significant currencies fluctuations and high recessions have emerged. Lastly, most of them focused on a small number of currencies (mostly hard currencies) but ignored the impact of economic policy uncertainty on the connectedness among a mix of hard and emerging countries’ currencies. 3 Data and methodology 3.1 Data To conduct our study, we collect a database on the daily exchange rates of twelve currencies, consisting of four cryptocurrencies and eight traditional currencies. The cryptocurrencies group includes Bitcoins (BIT), Ethereum (ETH), Litecoin (LIT), and Ripple (RIP), while the traditional currencies are: the Swiss Franc (SFR), Indian Rupee (IND), the Euro (EUR), Great Britain Pound (UK), Chinese Yuan (CHIN), Korean Won (KOR), Canadian Dollar (CAD), and the Japanese Yen (JPY). All the data covers the period from January 2, 2017, to October 6, 2022. The choice of the sample period accounts for two main issues: i) the significant drop in the value of cryptocurrencies after the beginning of 2018 and currency markets indices in general, as a result of the uncertainty generated by the Japanese cryptocurrency exchange Coincheck getting hacked, and was robbed to the tune of half a billion dollars; and ii) the uncertainty created by the ongoing COVID-19 crisis period. The traditional and digital currencies under study are taken based on their equivalency to the US dollar. Traditional currencies are collected from Refinitiv (DataStream database), while cryptocurrencies are obtained from Coinmarket.com. To conduct our analysis, we use the percentage log-returns. To examine the possible effect of the COVID-19 pandemic crisis on the connectedness between currencies, we divide our sample period into two sub-periods based on the declaration of the World Health Organization (WHO) that COVID-19 is a pandemic on March 11th, 2020. To investigate the effect of economic policy uncertainty (EPU) on the connectedness among currencies, we use three global risk indices (EPU, the fear index (VIX), which is the CBOE volatility index, and (OVX) representing the crude oil market volatility). The proxy for EPU is obtained from The Economic Policy Uncertainty index website: www.policyuncertainty.com. This index is developed by Baker et al. (2016) and reflects the growing uncertainty-generated policies that impact economic policy and financial decisions. The data of the other proxies for market uncertainties, such as VIX and OVX, were downloaded from the investing.com website. These indices have been used by several studies (among others, Manela and Moreira, 2017; Fang et al., 2020; Gozgor et al., 2019). Fig. 1 shows the exchange rates and cryptocurrencies’ returns series during the sample period. We observe increased volatility and volatility clustering, especially during the COVID-19 pandemic period for almost all series. However, returns series of Chinese Yuan, Bitcoin, Ethereum and Ripple show less volatility during the pandemic. Table 1, Table 2 provide summary statistics for the returns series before and during the COVID-19 periods, respectively. Generally, cryptocurrencies show positive average returns before the crisis, whereas traditional currencies show mixed signs of average returns. We see that Indian Rupee, the Canadian dollar and Chinese Yuan show positive average return while the Swiss Franc, British Pound, Euro, Korean Won, and Japanese Yen, show average negative returns. However, the results differ during the pandemic crisis in which traditional currencies and cryptocurrencies exhibit positive average returns. During the crisis, there was a noticeable change in the level of standard deviation for all currencies’ series, suggesting that both types of currencies are characterized by different levels of risk. More specifically, cryptocurrencies show higher standard deviations than traditional currencies during the crisis. In general, cryptocurrencies experience a significant drop in value after the beginning of 2018, leading to lower returns.Fig. 1 Return series of conventional and digital currencies. Fig. 1 Table 1 Descriptive Statistics and preliminary analysis before COVID-19. Table 1 BIT ETH LIT RIP SFR IND UK EUR CHIN KOR CAN JPY Descriptive statistics and preliminary tests Mean 0.254 0.380 0.290 0.417 -0.010 0.010 -0.005 -0.008 0.000 -0.001 0.003 -0.013 Variance 24.559 46.774 56.749 71.678 0.142 0.120 0.249 0.152 0.066 0.218 0.150 0.202 Skewness -0.035 0.485* 1.684* 2.074* -0.346* 0.579* -0.173 -0.198 0.346* -0.441* -0.103 -0.646* Ex.Kurtosis 3.035* 4.072* 11.382* 16.403* 1.317* 3.867* 1.451* 1.022* 4.871* 2.186* 1.839* 3.661* JB 187.700* 98.840* 61.694* 1374.614* 2.141 77.005* 11.593* 9.034 166.883* 7.194 94.223* 516.348* ERS -4.907* -9.783* -6.330* -9.496* -13.177* -11.611* -12.317* -12.922* -11.421* -13.330* -12.402* -12.460*** Q(10) 4.737 10.493* 9.805* 41.847*** 8.448 10.418* 6.520 5.002 12.278 2.597 3.972 3.820 Q2(10) 35.568* 30.114* 68.180* 82.492* 13.362 28.954* 4.424 19.715* 11.143 10.873 8.156 114.930* Correlation matrix BIT 1.000 ETH 0.619 1.000 LIT 0.589 0.596 1.000 RIP 0.431 0.424 0.452 1.000 SFR -0.021 -0.066 0.011 0.028 1.000 IND -0.045 -0.060 -0.041 -0.086 0.077 1.000 UK -0.026 -0.020 -0.005 0.000 0.381 0.166 1.000 EUR -0.008 -0.034 0.037 0.020 0.744 0.182 0.502 1.000 CHIN 0.049 0.015 0.005 -0.009 0.218 0.314 0.275 0.349 1.000 KOR -0.031 -0.022 -0.011 -0.068 0.194 0.414 0.202 0.326 0.466 1.000 CAN -0.055 -0.093 -0.083 -0.075 0.259 0.193 0.298 0.362 0.302 0.321 1.000 JPY 0.004 -0.012 0.036 0.017 0.573 0.009 0.210 0.400 0.114 0.072 0.100 1.000 Notes: This table reports the descriptive statistics of currencies (cryptocurrencies and traditional currencies) before the COVID-19 pandemic period. JB is the Jarque-Bera normality test statistics. ERS is the statistics for the Elliot-Rothenberg-Stock unit root test. Q(20) and Q2(20) are the Ljung-Box tests for 20th-order serial correlations for returns and squared returns, respectively. LM(20) is the LM heteroscedasticity test at order 20. (), (), and () indicate the statistical significance, respectively, at the 1%, 5%, and 10% levels. Table 2 Descriptive Statistics and preliminary analysis during COVID-19. Table 2 BIT ETH LIT RIP SFR IND UK EUR CHIN KOR CAN JPY Descriptive statistics and preliminary tests Mean 0.138 0.290 0.017 0.130 0.008 0.016 0.021 0.021 0.003 0.025 0.000 0.048 Variance 23.019 40.981 40.066 58.436 0.215 0.116 0.399 0.228 0.086 0.264 0.229 0.231 Skewness -1.659* -1.145* -1.278* 0.245* -0.220 0.123 0.385* 0.326* 0.000 -0.102 0.328* 0.401* Ex.Kurtosis 15.367* 12.470* 8.087* 14.650* 2.754* 3.171* 5.815* 1.382* 9.888* 1.293* 1.639* 2.768* JB 6909.852* 4494.322* 2011.299* 6007.306* 217.516* 282.884* 961.928* 65.238* 2733.803* 47.885* 87.146* 232.227* ERS -0.504 -0.524 -0.730 -0.845 -1.361 -2.064 -1.040 -1.284 -1.189 -2.065 -2.437 -3.015* Q(10) 13.143 16.265* 13.701 4.576 6.127 5.872 10.904 7.372 25.492* 18.604*** 9.300* 2.920 Q2(10) 3.770 12.319 18.941* 21.068* 14.500* 38.153* 128.080* 60.823* 147.701* 66.687* 50.816* 93.721* Correlation matrix BIT 1.000 ETH 0.827 1.000 LIT 0.829 0.831 1.000 RIP 0.560 0.592 0.654 1.000 SFR -0.061 -0.054 -0.039 -0.003 1.000 IND -0.049 -0.050 -0.043 0.014 0.319 1.000 UK -0.095 -0.085 -0.062 -0.029 0.530 0.409 1.000 EUR -0.099 -0.084 -0.078 -0.021 0.740 0.403 0.630 1.000 CHIN -0.104 -0.065 -0.071 -0.001 0.357 0.308 0.417 0.352 1.000 KOR -0.042 -0.018 0.003 0.022 0.361 0.375 0.460 0.468 0.339 1.000 CAN -0.089 -0.063 -0.056 -0.024 0.433 0.446 0.626 0.537 0.344 0.487 1.000 JPY -0.001 0.017 0.019 0.025 0.530 0.132 0.351 0.405 0.202 0.189 0.177 1.000 Notes: This table reports the descriptive statistics of currencies (cryptocurrencies and traditional currencies) before the COVID-19 pandemic period. JB is the Jarque-Bera normality test statistics. ERS is the statistics for the Elliot-Rothenberg-Stock unit root test. Q(20) and Q2(20) are the Ljung-Box tests for 20th-order serial correlations for returns and squared returns, respectively. LM(20) is the LM heteroscedasticity test at order 20. (), (), and () indicate the statistical significance, respectively, at the 1%, 5%, and 10% levels. Considering the COVID-19 period, we notice that cryptocurrencies have the highest positive returns (Ethereum has the highest return and Litecoin shows the lowest positive return). However, for traditional currencies, the Swiss Franc, Canadian dollar, and Chinese Yuan have the lowest returns, while the British Pound, Korean Won, and Japanese Yen experienced the highest positive returns. Then, testing for normality, skewness, and excess kurtosis values indicate that the return series are asymmetric and fat-tailed. The Jarque-Berra test statistics are all significant at the 1% level for all series, confirming the non-normality issue. Finally, the results from the augmented Dickey-Fuller (ADF) (Dickey and Fuller, 1979) and the Phillip-Perron (PP) (Phillips and Perron, 1988) unit root tests indicate that all return series are stationary. Moreover, the Ljung-Box Q(20) and Q2(20) tests of the autocorrelation of returns and squared returns series indicate that the null hypothesis of no autocorrelation is rejected in almost of cases at the 5% significance level. Table 1, Table 2 also provide the correlation matrix between all currency return series. The correlations between the two types of currencies differ across both sub-periods; correlations between cryptocurrencies and traditional currencies are generally negative during the COVID-19 period. However, cryptocurrencies, as a separate group, are positively correlated in both sub-periods. Traditional currencies, as a separate group, are also significantly and positively correlated with each other during the COVID-19 period. The negative correlations between cryptocurrencies and traditional currencies during COVID-19 indicate the usefulness of both types of currencies to hedge each other during stress periods. 3.2 Methodology This section presents the econometric methodology used in the empirical analysis of the total and directional connectedness between cryptocurrencies and exchange rates. Initially, we outline the methodology proposed by Diebold and Yilmaz, 2009, Diebold and Yilmaz, 2012, Diebold and Yılmaz, 2014 and then present the dynamic connectedness procedure based on the TVP-VAR methods proposed by Antonakakis and Gabauer (2017). This approach is proven to gain a lot of attention and popularity, as evidenced by many studies, including those by Korobilis and Yilmaz (2018), Gabauer and Gupta (2018), Liu and Gong (2020), Youssef et al. (2021), and Mokni et al. (2021), among others. 3.2.1 TVP-VAR approach Researchers have developed and elaborated methods to capture shock transmission mechanisms between macroeconomic variables. Diebold and Yilmaz, 2012, Diebold and Yılmaz, 2014 (DY) introduced a rolling-window VAR-based empirical approach providing various connectedness measures built from pieces of variance decompositions, which they describe as “natural and insightful.” Their approach assesses forecast error variation in the variable i attributed to innovations in other variables in the model. They considered that the time series follows the reduced-form autoregressive (VAR) model with a fixed parameter and variance-covariance matrix. Instead, to improve the accuracy of the dynamic connectedness measures of DY, Antonakakis and Gabauer (2017) employed a time-varying parameter vector autoregressive model (TVP-VAR) with a time-varying covariance structure proposed in Primiceri (2005), as opposed to the constant-parameter rolling-window VAR approach. The drifting coefficients and stochastic volatility are meant to capture the possible nonlinearities or time variation in the lag structure of the model, the possible heteroscedasticity of the shocks, and the nonlinearities in the simultaneous relations among the variables of the model (Primiceri, 2005). The TVP-VAR dynamic connectedness approach improves the Diebold and Yilmaz methodology in many ways. It is argued by Antonakakis and Gabauer (2017) and Antonakakis et al. (2020) that this framework allows “to capture possible changes in the underlying structure of the data in a more flexible and robust manner” and that results indicate that the TVP-VAR estimations are superior to those generated by rolling-windows. First, since the rolling window size is not set arbitrarily, there is no loss of information in calculating the dynamic measures of connectedness. Second, since it is based on a multivariate Kalman filter, it is less sensitive to the presence of outliers and thus adjusts immediately to events (e.g., Antonakakis et al., 2018; Gabauer and Gupta, 2018). The N-variable TVP-VAR(p) model can thus be written as follows,(1) Yt=βtZt−1+εt,εt\|Ωt−1~N(0,Σt) (2) βt=βt−1+ϑt,ϑt\|Ωt−1~N(0,Rt) where Yt = (Y1t,Y2t,…YNt)′ is a vector with N variables and Zt−1 is an Np×1 conditional vector formed of the past p lags, p being the optimal lag length with Zt−1=Yt−1,Yt−2,⋯,Yt−p', and Ωt−1represents all information available through time t−1, whereas βt is an N×Np dimensional time-varying coefficient matrix, which follows a random walk model and can be represented as βt=β1t,β2t,⋯,βpt, βit being an N×N matrix of time-varying coefficients. The error-disturbance vectorsεt and ϑt are N×1 and N×Npdimensional with N×N and Np×Np time-varying variance-covariance matrices, Σt and Rt respectively. The Kalman filter algorithm is employed with forgetting factors chosen based on a Bayesian model selection, as introduced by Koop and Korobilis (2014) and demonstrated in Antonakakis et al. (2018). To calculate the generalized forecast error variance decompositions (GFEVD), the model in Eq. (1) is transformed to its vector moving average (VMA) representation based on the Wold representation theorem. The representation of the system is:(3) Yt=∑j=0∞Θjtεt−j, where Θjt is an N×N dimensional matrix. 3.2.2 Dynamic connectedness measures To measure the dynamic connectedness between different variables, the time-varying parameters and variance-covariance matrices of the TVP-VAR model are used in Diebold and Yilmaz’s measure of connectedness. The elements of the dynamic H-step generalized variance decomposition matrix DtgH = dij,tgH are defined as:dij,tgH=σjj,t−1∑h=0H−1ei'Θh,tΣtej2∑h=0H−1ei'Θh,tΣtΘh,t'ej, where σjj,t is the jth diagonal element of Σt. The normalized terms d~ij,tgH=dij,tgH∑j=1Ndij,tgH are used to determine the dynamic total directional connectedness, net total directional connectedness, and total connectedness as follows. The total connectedness index (TCI), which measures interconnectedness among the time-series, is calculated as:(4) CtgH=∑i,j=1,i≠jNd~ij,tgH∑j=1Nd~ij,tgH×100 The directional spillover received by variable i from all other variables j, is measured as:(5) Ci←jgH=∑j=1,i≠jNd~ij,tgH∑i=1Nd~ij,tgH×100 Likewise, the spillovers received by variable j from all other variables i, is measured as:(6) Ci→jgH=∑j=1,i≠jNd~ij,tgH∑j=1Nd~ij,tgH×100 In order to obtain the net pairwise directional connectedness, we subtract the total directional connectedness to others from the total directional connectedness from others, which can be interpreted as the influencing variable i has on the analyzed network. That is,(7) Cij,tgH=Cj←i,tgH−Ci←j,tgH. At last, the net pairwise directional connectedness is defined as: NPDCijgH=(d~ji,tgH−d~ij,tgH)×100. A result greater than zero indicates that variable i dominates variable j; otherwise, the latter is said to dominate. 4 Empirical analysis In this study, we perform two empirical analyses. First, the dynamic connectedness among currencies (both types of currencies: traditional and cryptocurrencies) is quantified using the procedure introduced by Diebold and Yilmaz, 2009, Diebold and Yilmaz, 2012, Diebold and Yılmaz, 2014 based on the time-varying parameter vector autoregressive (TVP-VAR) methodology. Second, the effect of economic policy uncertainty (EPU) on this connectedness is estimated based on a quantile regression model. 4.1 Dynamic connectedness results Based on the TVP-VAR approach, the results of total connectedness among all currencies before and during the health crisis are reported in Table 3. The table shows that the total connectedness index (TCI) exhibits different behaviors before and during the pandemic. Indeed, the average TCI increased from 51.043 before the pandemic to 57.7 during the crisis. Fig. 2 depicts the dynamic TCI and provides further results supporting those reported in Table 3 . As shown in Fig. 2, total connectedness increased at the beginning of the COVID-19 crisis. It fell somehow during 2021 due to the drop in death cases based on the discovery of the vaccination, but then it started to increase during the 2022 period, marking the impact of the Russian-Ukraine war.Table 3 Total average dynamic connectedness. Table 3 BIT ETH LIT RIP SFR IND UK EUR CHIN KOR CAN JPY FROM Before COVID-19 BIT 35.558 21.942 20.438 15.908 0.625 0.98 0.627 1.031 0.404 0.689 0.5 1.298 64.442 ETH 21.247 33.56 21.648 17.225 0.614 1.186 0.663 0.967 0.508 0.827 0.715 0.841 66.44 LIT 20.219 23.841 33.226 15.816 0.865 0.991 0.9 1.372 0.375 0.856 0.461 1.076 66.774 RIP 17.372 20.93 17.439 37.2 0.761 1.909 0.735 0.793 0.46 0.68 0.664 1.057 62.8 SFR 1.252 0.713 0.703 0.546 49.382 1.387 5.705 23.645 1.842 1.844 3.122 9.859 50.618 IND 0.666 0.775 1.033 1.598 1.282 63.341 2.823 2.783 8.962 12.794 3.032 0.912 36.659 UK 1.416 1.014 1.257 0.681 6.452 2.826 57.758 13.813 3.573 3.041 6.657 1.512 42.242 EUR 0.734 1.387 1.219 0.666 21.368 1.968 10.847 43.422 5.264 4.783 5.72 2.62 56.578 CHIN 1.555 0.704 0.518 1.421 2.457 7.068 3.445 5.748 54.583 14.816 6.826 0.861 45.417 KOR 0.836 0.866 0.929 0.749 1.716 9.668 1.839 5.057 14.588 53.977 8.024 1.752 46.023 CAN 0.717 0.788 0.434 0.655 2.936 2.156 5.712 9.47 8.8 8.503 58.253 1.575 41.747 JPY 2.168 0.691 0.739 0.852 15.678 1.502 2.716 4.845 1.703 1.119 0.761 67.227 32.773 Cont TO others 68.183 73.65 66.355 56.117 54.755 31.642 36.011 69.524 46.478 49.951 36.481 23.364 612.51 Cont incl own 103.741 107.211 99.581 93.317 104.137 94.983 93.769 112.947 101.061 103.929 94.734 90.591 TCI Net spillovers 3.741 7.211 -0.419 -6.683 4.137 -5.017 -6.231 12.947 1.061 3.929 -5.266 -9.409 51.043 During COVID-19 BIT 35.55 22.62 23.96 11.35 0.52 0.59 1.13 1 0.8 0.66 1.5 0.32 64.45 ETH 22.54 35.73 23.71 14.04 0.36 0.48 0.86 0.63 0.2 0.31 0.9 0.25 64.27 LIT 22.98 22.88 34.14 15.04 0.55 0.39 0.83 0.84 0.45 0.42 1.13 0.33 65.86 RIP 14.25 17.58 19.65 45.57 0.24 0.34 0.41 0.31 0.23 0.4 0.6 0.43 54.43 SFR 0.7 0.72 0.91 0.42 36.07 3.85 9.2 21.22 4.75 5.16 6.37 10.63 63.93 IND 1.06 1.46 1.44 1.38 5.03 49.9 8.43 8.34 5.33 6.6 9.69 1.34 50.1 UK 1.74 1.86 1.63 1.07 9.19 6.18 37.67 11 5.59 7.01 13.37 3.7 62.33 EUR 1.22 1.34 1.47 0.59 19.93 5.61 10.27 34.3 4.23 6.27 8.37 6.41 65.7 CHIN 1.21 1.09 1.19 0.71 6.8 4.97 7.79 6.46 55.63 5.15 5.86 3.15 44.37 KOR 1.59 1.92 1.66 1.51 5.64 5.86 9.18 7.67 4.88 48.91 9.55 1.63 51.09 CAN 2.27 3.02 2.14 1.36 6.21 7.87 14.12 8.76 5.01 8.14 40.25 0.84 59.75 JPY 1.16 0.84 1.05 0.96 15.96 2.19 5.81 10.44 3.62 2.51 1.56 53.9 46.1 Cont TO others 70.72 75.33 78.8 48.44 70.42 38.33 68.01 76.69 35.1 42.62 58.9 29.03 692.38 Cont incl own 106.27 111.06 112.94 94.01 106.49 88.23 105.68 110.99 90.73 91.53 99.15 82.93 TCI Net spillovers 6.27 11.06 12.94 -5.99 6.49 -11.77 5.68 10.99 -9.27 -8.47 -0.85 -17.07 57.7 Notes: This table reports the results of the average connectedness measures based on the Diebold and Yilmaz (2012, 2014) estimated from a TVP-VAR model for the full sample before and during the COVID-19 period. We provide the average of total connectedness, the directional spillover received (denoted by “From”), and transmitted (denoted by “Contribution to others”) by each variable. Then, the net directional spillover (denoted by “Net Spillovers”) is obtained as the difference between directional ‘To’ spillovers and directional ‘From’ spillovers. Fig. 2 Total connectedness between currencies over the full period. Fig. 2 Further, Table 3 shows the contribution of each currency to others before the COVID-19 pandemic and emphasizes that cryptocurrencies are the most transmitters of shocks compared to traditional currencies. Ethereum is the highest transmitter, while Ripple is the lowest one. The highest traditional currencies transmitters are the Swiss Franc and the Euro, while the lowest transmitters are the Indian Rupee and the Japanese Yen. Although the level of volatility transmission rose significantly across all types of currencies during the pandemic crisis, the majority of traditional currencies are the highest transmitters (the Euro (76.69), the British Pound (UK) (68.01), the Swiss Franc (SFR) (70.42), the Canadian Dollar (CAD) (58.9), followed by the cryptocurrencies (Bitcoin (70.72), Ethereum (75.33) and Litecoin (78.8)). These results confirm that the dominance of currencies in the shock transmission changes from one currency to another before and during COVID-19. Traditional currencies (EUR, UK, SFR, and CAN) as well as cryptocurrencies (BIT, ETH, and LIT) are the dominant transmitters of shocks during the crisis, with the Euro, British Pound, and Swiss Franc being the most interconnected with other currencies2 . Regarding the amount of shocks received by each currency from the others, Table 3 shows that the most receiver of shocks from others is almost all cryptocurrencies, as well as a few traditional currencies, namely: Euro (56.57) and Swiss Franc (50.61) before the COVID-19 period, and 62.55 and 63.93, for the during COVID-19 period. During the health crisis, although the same cryptocurrencies maintained the same position as the highest receivers of shocks from the system, traditional currencies became the highest receivers as well. Therefore, during the crisis, almost all currencies, either cryptocurrencies or traditional currencies, became more receivers of shocks. Our results are inconsistent with those obtained by Baumöhl (2019) while falling in-line with those of Chemkha et al. (2021). For example, Baumöhl (2019) analyzed the relationship between traditional currencies and cryptocurrencies and found a negative relationship between the two types of currencies in the short-and long-term. While, Chemkha et al. (2021) provided evidence that connectivity between cryptocurrencies has greatly strengthened since 2017 yet, “the dependence is positive and higher for the pairs of the same market than those across markets, and the cryptocurrency market and the fiat currency market are weakly connected,” providing investors with more benefits from holding these assets together, for portfolio diversification purposes. Table 3 also includes the results of the net connectedness. The currency is a net transmitter (receiver) of shocks if this index is positive (negative). The net connectedness index varies over currencies before and during the COVID-19 crisis. Before the crisis, four traditional currencies (SFR, EUR, CHIN, KOR) and two cryptocurrencies (ETH and BIT) were found to be net transmitters. However, the others are found to be net receivers. The highest net transmitters are the Euro and Ethereum, whereas the most net receivers are the Japanese Yen, the British Pound, the Canadian dollar, the Indian Rupee, and Ripple as well. During the COVID-19 crisis, the highest cryptocurrency net transmitters rose to three (BIT, ETH, and LIT). Litecoin has turned from the lowest net receivers to one of the highest net transmitters compared to the period before the crisis. Such a result can be interpreted by the important position taken by Litecoin as a safe haven asset that dominates this of Bitcoin. Our results confirm the previous findings of Yi et al. (2018), Bouri et al. (2019), Ji et al. (2020), Moratis (2021), Al-Shboul et al. (2022) and Raza et al. (2021). In addition, the results show that the level of net transmission increased significantly during the crisis since five traditional currencies (SFR, EUR, UK, CHIN, and CAD) became net transmitters to the system. In particular, the Swiss Franc, the Euro, and the British Pound are the highest net transmitters. The highest net receivers during the pandemic crisis were the Ripple, the Japanese Yen, Indian Rupee, and the Korean Won. Overall, we find an increase in the level of net transmitting and net receiving of shocks during the COVID-19 period. Our results are consistent with those of Shahzad et al. (2021), who considered four cryptocurrencies (Bitcoin, Ethereum, Ripple, and Litecoin) and four traditional currencies, namely the Euro, Japanese yen (JPY), the British pound (GBP), and Chinese yuan (CNY). During the negative explosiveness and non-bubble periods, they found that JPY is the most consistent hedger for the considered cryptocurrencies, followed by GBP and the Euro. All other currencies, except the Euro, have safe-haven properties for Bitcoin and Litecoin. Our results also fall in-line with Baumöhl (2019), who argued that Euro, Japanese Yen, and Chinese yuan share a safe-haven potential during turmoil and extreme market periods3 . We go further to obtain the total and the net spillover among currencies as separate groups. Table 4 presents the results of the dynamic connectedness among cryptocurrencies. On average, the total connectedness among cryptocurrencies during the COVID-19 crisis is higher than before the crisis period. The highest contributors to other cryptocurrencies are Ethereum and Litecoin during and before the crisis periods, with a noticeable increase in contribution levels during the pandemic period. The net spillover among cryptocurrencies indicates that Ethereum is the highest net transmitter before the crisis, while Ripple is found to be the highest net receiver. Overall, during the crisis period, cryptocurrencies show evidence of more net transmitters and net receivers than before COVID-19.Table 4 Total average dynamic connectedness for cryptocurrencies. Table 4 BIT ETH LIT RIP FROM Full sample BIT 45.38 21.34 20.47 12.8 54.62 ETH 20.51 43.27 21.71 14.51 56.73 LIT 20.06 20.94 43.44 15.56 56.56 RIP 14.17 17.39 18.3 50.15 49.85 Cont TO others 54.74 59.67 60.48 42.87 217.76 Contr incl own 100.12 102.95 103.92 93.02 TCI Net spillovers 0.12 2.95 3.92 -6.98 54.44 Before COVID-19 BIT 52.39 19.28 17.28 11.05 47.61 ETH 17.86 50.24 18.79 13.11 49.76 LIT 16.82 19.03 51.07 13.08 48.93 RIP 11.74 15.34 15.83 57.09 42.91 Cont TO others 46.41 53.65 51.91 37.24 189.21 Contr incl own 98.8 103.88 102.98 94.34 TCI Net spillovers -1.2 3.88 2.98 -5.66 47.30 During COVID-19 BIT 38.19 24.27 25.77 11.77 61.81 ETH 23.51 37.26 24.71 14.52 62.74 LIT 24.31 24.11 36 15.59 64 RIP 14.41 18.08 20.14 47.37 52.63 Cont TO others 62.22 66.46 70.63 41.88 241.19 Contr incl own 100.41 103.72 106.63 89.25 TCI Net spillovers 0.41 3.72 6.63 -10.75 60.30 Notes: This table reports the results of the average connectedness measures of cryptocurrencies based on the Diebold and Yilmaz (2012, 2014) estimated from a TVP-VAR model for the full sample before and during the COVID-19 period. We provide the average of total connectedness, the directional spillover received (denoted by “From”), and transmitted (denoted by “Contribution to others”) by each variable. Then, the net directional spillover (denoted by “Net Spillovers”) is obtained as the difference between directional ‘To’ spillovers and directional ‘From’ spillovers. Table 5 presents the results of spillovers among traditional currencies. Interestingly, such results are different from the results of cryptocurrencies. The total spillover among traditional currencies, on average, during the COVID-19 crisis is higher compared to the period before crisis. The highest contributors to other traditional currencies are the Swiss Franc, Euro, Chinese Yuan, Korean Won, and Canadian Dollar in both sub-periods. Yet, during the crisis, there is a noticeable increase in the level of contribution of the traditional currencies compared to the period before the pandemic. We observe that only four traditional currencies are the most receivers of information. Still, the number of traditional currencies receivers is significantly increasing during COVID-19, as almost all traditional currencies exhibit a higher level of information reception.Table 5 Total dynamic connectedness for traditional currencies. Table 5 SFR IND UK EUR CHIN KOR CAN JPY FROM Full sample SFR 41.38 2.06 8.17 23.09 4.56 3.3 4.7 12.74 58.62 IND 3.11 60.33 5.59 5.77 6.38 10.77 6.36 1.7 39.67 UK 9.61 4.14 49.22 13.99 5.52 4.62 9.56 3.34 50.78 EUR 21.8 3.46 11.08 39.04 5.43 5.6 7.03 6.57 60.96 CHIN 6.17 5.54 6.52 7.59 55.16 9.39 6.05 3.57 44.84 KOR 4.29 9.14 5.46 7.56 9.32 53.4 8.38 2.46 46.6 CAN 5.59 5.34 10.21 8.94 5.72 8.44 53.98 1.79 46.02 JPY 18.02 1.34 4.11 9.86 3.75 2.13 1.8 58.99 41.01 Cont TO others 68.59 31.01 51.15 76.8 40.67 44.25 43.87 32.16 388.51 Cont incl own 109.97 91.34 100.38 115.83 95.84 97.65 97.85 91.15 TCI Net spillovers 9.97 -8.66 0.38 15.83 -4.16 -2.35 -2.15 -8.85 48.56 Before COVID-19 SFR 44.39 1.29 6.93 23.36 4.62 2.28 3.37 13.77 55.61 IND 2.05 64.72 3.13 4.18 6.97 13.23 3.34 2.37 35.28 UK 9.07 2.46 56.75 15.25 5.1 2.77 5.36 3.24 43.25 EUR 21.89 2.34 11 41.44 6.49 5.09 5.57 6.17 58.56 CHIN 6.01 5.76 4.94 8.62 52.96 11.36 5.55 4.81 47.04 KOR 3.04 11.13 2.85 7 11.52 54.96 6.62 2.89 45.04 CAN 4.42 3.27 5.61 7.88 6.03 7.52 62.73 2.54 37.27 JPY 17.96 1.51 3.14 8.02 4.57 2.25 2.33 60.22 39.78 Cont TO others 64.44 27.75 37.6 74.3 45.3 44.49 32.15 35.79 361.83 Cont incl own 108.83 92.48 94.35 115.75 98.26 99.45 94.88 96.01 TCI Net spillovers 8.83 -7.52 -5.65 15.75 -1.74 -0.55 -5.12 -3.99 45.23 During COVID-19 SFR 36.09 4.26 9.78 21.4 5.28 5.44 6.98 10.76 63.91 IND 5.55 50.28 9.28 9.17 5.96 7.39 10.88 1.47 49.72 UK 9.9 6.92 38.34 12.08 6 8.15 14.75 3.87 61.66 EUR 20.33 6.26 11.14 34.59 4.79 6.9 9.45 6.54 65.41 CHIN 7.55 5.65 8.42 7.22 55.46 5.85 6.65 3.2 44.54 KOR 6.24 6.78 10.58 8.66 5.54 49.41 11.09 1.72 50.59 CAN 7.03 9.02 15.88 10.17 5.55 9.81 41.54 0.99 58.46 JPY 16.75 2.38 6.21 10.94 3.84 2.64 1.84 55.41 44.59 Cont TO others 73.34 41.26 71.3 79.64 36.96 46.18 61.64 28.56 438.88 Cont incl own 109.43 91.54 109.64 114.23 92.41 95.59 103.19 83.97 cTCI/TCI Net spillovers 9.43 -8.46 9.64 14.23 -7.59 -4.41 3.19 -16.03 62.70/54.86 Notes: This table reports the results of the average connectedness measures of cryptocurrencies based on Diebold and Yilmaz (2012, 2014) estimated from a TVP-VAR model for the full sample before and during the COVID-19 period. We provide the average of total connectedness, the directional spillover received (denoted by “From”), and transmitted (denoted by “Contribution to others”) by each variable. Then, the net directional spillover (denoted by “Net Spillovers”) is obtained as the difference between directional ‘To’ spillovers and directional ‘From’ spillovers. Before the crisis, two traditional currencies were net transmitters (the Swiss Franc and the Euro), while the others were net receivers. The highest net transmitter is the Euro, whereas the Indian Rupee is the highest net receiver. However, the results of net spillover are somewhat different during the COVID-19 period. The net transmitters are the Swiss Franc, the British Pound, the Euro, and the Canadian Dollar, with the Euro being the highest transmitter. The British Pound and Canadian Dollar change their position from being net receivers before the crisis to net transmitters during the crisis. The other currencies maintain their positions as net receivers of shocks during the pandemic period. The highest net receivers are the Indian Rupee and the Japanese Yen. Overall, traditional currencies display a higher level of net spillover during the COVID-19 period. The results from Table 3, Table 4, Table 5 are confirmed by the graphs reported in Fig. 3, Fig. 4, Fig. 5. Fig. 3 provides the behavior of the net dynamic spillover in the full sample periods for both groups of currencies. Moreover, Fig. 4 provides the dynamic net connectedness indices before and over the COVID-19 pandemic period. A close look at this figure shows that the net spillover among currencies is higher during the COVID-19 period than before the crisis period for all currencies. Due to the COVID-19 crisis in March 2020, the spillover level got higher. For traditional currencies, the net spillover shows higher levels of net transmission and receiving of shocks, implying that during the COVID-19 period, the net connectedness among traditional currencies increased. However, the net spillover among cryptocurrencies at the beginning of the crisis showed a lower level of net transmission and net receiving than before the crisis, indicating a significant increase at the end of 2020.Fig. 3 : Dynamic net connectedness over the full sample period. Fig. 3 Fig. 4 : Dynamic net connectedness before the COVID-19 pandemic. Fig. 4 Fig. 5 : Dynamic net connectedness during the COVID-19 pandemic. Fig. 5 For the pre-COVID-19 period, Fig. 4 shows a higher fluctuation in the net spillover of all cryptocurrencies and traditional currencies before the crisis compared to their situation during the crisis. However, the net spillover among cryptocurrencies shows higher levels of net transmission and net reception of shocks during the crisis than before the crisis. The net spillover of cryptocurrencies has a wider range than traditional currencies during the COVID-19 crisis. As indicated by Fig. 5 , during the crisis period, we notice a wider range of net spillover of all currencies in the early beginning of the pandemic, and such range started to relax at the end of the third quarter of 2020, with the beginning of the vaccination programs. However, after the end of 2020, there was an increase in the net spillover among all currencies. In light of the traditional currencies, we see an increase in the net spillover after the third quarter of 2020. Although the net spillover of the traditional currencies became wider, the level of net spillover of cryptocurrencies became higher than the net spillover of the traditional currencies during the crisis period. 4.2 EPU and dynamic connectedness After obtaining the spillover measures, our empirical analysis provides further evidence of the relationship between economic policy uncertainty and connectedness among currencies. We use mainly Economic Policy Uncertainty (EPU). The EPU was developed by Baker et al. (2016) and Baker et al. (2012), respectively, due to the growing uncertainty-generated policies that impact economic policy and financial decisions. These relate to oil price fluctuations, regulatory conflicts, disputes over income distribution inequality, etc., that are happening everywhere. To further support our analysis, we also adopt two other uncertainty factors, namely, the CBOE fear index (VIX), and the crude oil volatility (OVX)4 . To examine this relationship, we run a quantile regression model to obtain the effects of these factors on the dynamic (total and net) connectedness among currencies. The quantile regression is specified via the following regression:(9) conntτ=β0τ+β1τEPUt+βxτXt+εtτ where conntτ refers toτth the quantile of the total connectedness or the net connectedness index. EPUt is the proxy for economic policy uncertainty. Xt represents the two other risk factors, namely, the CBOE volatility index (VIX) and the crude oil volatility index (OVX), and εtτ is the error term.τth refers to the quantile order. If the coefficients on these factors, β1τ and βxτ, are statistically significant, this indicates that those factors significantly affect the dynamic spillover among currencies. This analysis gives more inclusive evidence of the relationship between uncertainty factors and the dynamic connectedness measures at different levels of the spillovers distribution. Table 6 presents the estimated parameters and their corresponding probabilities over the full sample period, while Table 7, Table 8 present the results before and during the COVID-19 health crisis, respectively. Then, Fig. 6, Fig. 7, Fig. 8 depict the impact of EPU, VIX, and OVX on the total connectedness among all currencies across different quantiles over the full and the two sub-periods, respectively5 . As can be observed from these tables and figures, some estimated parameters are statistically significant for some quantiles, indicating that the connectedness between currencies is strongly influenced by these three factors. The results also confirm the presence of a heterogeneous impact of uncertainty on the dynamic connectedness for both types of currencies under different spillover distribution levels (e.g., low, normal, and high). Specifically, the results show a negative effect of the EPU on the total spillover among all currencies at all quantiles for both the full sample and COVID-19 periods. Therefore, an increase in uncertainty levels leads to a decrease in the level of connectedness among currencies (either traditional or crypto). These results can be explained by the reaction of investors to the increase in uncertainty. Investors may interpret uncertainty as bad news, which tends to increase their investment in currencies.Table 6 Effect of EPU VIX and OVX on the total and net connectedness among currencies over the full sample period. Table 6 EPU VIX OVX τ=0.1 τ=0.5 τ=0.9 τ=0.1 τ=0.5 τ=0.9 τ=0.1 τ=0.5 τ=0.9 TOTAL 0.025* 0.015* 0.003* 0.345* 0.411* 0.020* -0.086* 0.0111 0.018 (0.000) (0.000) (0.087) (0.000) (0.000) (0.000) (0.0000) (0.454) (0.735) NET BIT 0.008* 0.007* -0.003 -0.048* 0.028 0.277* -0.008 0.047 0.041* (0.000) (0.000) (0.022) (0.001) (0.363) (0.000) (0.037) (0.027) (0.000) NET ETH 0.003* 0.006* 0.001* 0.009* 0.006*** 0.001 -0.004 0.032* 0.031* (0.000) (0.000) (0.000) (0.000)) (0.000) (0.385) (0.604) (0.076) (0.000) NET LIT 0.011* 0.009 -0.003 0.195* 0.284* 0.571* -0.055* 0.015 0.207 (0.000) (0.000) (0.217) (0.000) (0.000) (0.056) (0.000) (0.385) (322) NET RIP -0.016* 0.003* 0.004* 0.061* 0.069* 0.113* 0.013* 0.013 0.141* (0.000)) (0.000)) (0.000) (0.000) (0.000) (0.000) (0.000) (0.327) (0.000) NET SFR -0.005* -0.009* 0.002 0.053* -0.056 0.001 0.011 0.000 -0.055* (0.000) (0.000) (0.344) (0.000) (0.010) (0.995) (0.021) (0.902) (0.000) NET IND 0.007* -0.001 -0.008* -0.371 -0.094* -0.050 0.002 0.025 0.033 (0.000) (0.025) (0.000) (0.227) (0.000) (0.363) (0.991) (0.338) (0.136) NET UK 0.016* 0.014* 0.004* 0.115* 0.388* 0617* -0.057* -0.062* -0.066 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.034) NET EUR 0.006* -0.006* -0.006* 0.032 0.007 0.032 -0.152 -0.075* -0.059* (0.000) (0.000) (0.000) (0.814) (0.850) (0.267) (0.214) (0.000) (0.000) NET CHIN 0.008 0.001 -0.005*** -0.364* -0.299* -0.029 -0.001 0.034 0.005 (0.021) (0.405) (0.000) (0.090) (0.000) (0.017) (0.710) (0.023) (0.248) NET KOR -0.022* -0.021* -0.023* -0.060 -0.104* -0.323* -0.029 0.044* 0.075* (0.000) (0.000) (0.000) (0.189) (0.000) (0.000) (0.283) (0.000) (0.000) NET CAN 0.021* 0.02* 0.023*** -0.184* -0.013 0.002 -0.146 -0.011 -0.006 (0.000) (0.000) (0.000) (0.086) (0.646) (0.949) (0.024) (0.381) (0.228) NET JPY -0.011* -0.019* -0.021* -0.425* -0.429* -0.272*** -0.134* 0.020 0.131* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.071) (0.357) (0.000) Notes: This table reports the estimated coefficients of the effect of EPU, VIX and OVX on the total and net dynamic connectedness, as in Eq. (9). The numbers between parentheses are the probabilities of the t-statistics. (), (), () indicate the parameter significance at 1%, 5%, and 10% significance level, respectively. Table 7 Effect of EPU VIX and OVX on the total and net connectedness among currencies before the COVID-19 pandemic. Table 7 EPU VIX OVX τ=0.1 τ=0.5 τ=0.9 τ=0.1 τ=0.5 τ=0.9 τ=0.1 τ=0.5 τ=0.9 TOTAL 0.027* 0.022* 0.014* 0.146 0.199 0.161 -0.022 -0.020 0.103 (0.000) (0.000) (0.000) (0.034) (0.029) (0.474) (0.486) (0.233) (0.363) NET BIT 0.009* 0.010* 0.007* -0.065* -0.086* -0.119* -0.002 0.001 0.093* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.212) (0.916) (0.000) NET ETH 0.011* 0.009* 0.009* 0.068* 0.094* 0.036 0.001 0.018 0.073* (0.000)) (0.000) (0.000) (0.000) (0.087) (0.286) (0.597) (0.744) (0.000) NET LIT 0.011* 0.011*** 0.010 0.142* 0.152* 0.061* -0.034 -0.011* 0.049* (0.000) (0.000) (0.000) (0.062) (0.000) (0.002) (0.330) (0.002) (0.000) NET RIP 0.010* 0.012* 0.006* 0.058* 0.042* 0.075* 0.002 0.001 0.131* (0.000) (0.000) (0.000) (0.000) (0.073) (0.000) (0.339) (0.909) (0.000) NET SFR -0.006* -0.014* -0.018*** 0.046* -0.004* 0.075 0.011* -0.007 -0.055* (0.000) (0.000) (0.000) (0.093) (0.000) (0.014) (-0.088) (0.411) (0.000) NET IND -0.001 -0.002* -0.009* 0.059* -0.071 0.005 -0.001 0.042* 0.012*** (0.452) (0.000) (0.000) (0.000) (0.019) (0.702) (0.822) (0.000) (0.000) NET UK 0.011* 0.022* 0.016 0.079* 0.088* 0.289 -0.051* -0.045* -0.044 (0.071) (0.000) (0.000)) (0.000)) (0.000) (0.011) (0.000) (0.000) (0.409) NET EUR 0.002 -0.012* -0.011* 0.197 0.186*** 0.064* -0.103 -0.121* -0.062* (0.297) (0.000) (-0.35) (0.167) (0.000) (0.051) (0.176) (0.000) (0.000) NET CHIN -0.000 -0.009* -0.010* -0.079 0.014 -0.030 0.024* 0.007 0.007 (0.624) (0.000) (0.000) (0.016) (0.382) (0.021) (0.000) (0.155) (0.187) NET KOR -0.011* -0.022*** -0.017 -0.202 -0.053 -0.049 -0.117* 0.027 0.233 (0.000) (0.000) (0.212) (0.102) (0.101) (0.845) (0.066) (0.029) (0.119) NET CAN 0.000 0.015* 0.015* -0.155 -0.186* -0.092* -0.191* -0.105* -0.009* (0.998) (0.000) (0.000) (0.030) (0.000) (0.000) (0.000) (0.000) (0.000) NET JPY -0.012* -0.023* -0.020* -0.594* 0.406* -0.100* -0.064 0.042 0.053 (0.000) (0.000) (0.000) (0.000) (0.000) (0.059) (0.360) (0.230) (0.037) Notes: This table reports the estimated coefficients of the effect of EPU, VIX and OVX on the total and net dynamic connectedness, as in Eq. (9). The numbers between parentheses are the probabilities of the t-statistics. (), (), () indicate the parameter significance at 1%, 5%, and 10% significance level, respectively. Table 8 Effect of EPU VIX and OVX on the total and net connectedness among currencies during the COVID-19 pandemic. Table 8 EPU VIX OVX τ=0.1 τ=0.5 τ=0.9 τ=0.1 τ=0.5 τ=0.9 τ=0.1 τ=0.5 τ=0.9 TOTAL 0.027* 0.022* 0.017* 0.014 0.199 0.161 -0.022 -0.020 0.103 (0.000) (0.000) (0.000) (0.034) (0.029) (0.474) (0.485) (0.233) (0.363) NET BIT 0.009* 0.010* 0.007* -0.065* -0.086* 0.119* -0.002 0.001 0.093* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.212) (0.916) (0.000) NET ETH 0.011* 0.009* 0.009* 0.069* 0.096* 0.036 0.001 0.018 0.073* (0.000) (0.000) (0.000) (0.000) (0.087) (0.286) (0.597) (0.744) (0.000) NET LIT 0.011* 0.011* 0.010* 0.142* 0.152* 0.061* -0.034 -0.011* 0.049* (0.000) (0.000) (0.000) (0.062) (0.000) (0.000) (0.330) (0.000) (0.000) NET RIP 0.010* 0.012* 0.006* 0.058* 0.042* 0.076* 0.002 0.000 0.131* (0.000) (0.000) (0.000) (0.000) (0.073) (0.000) (0.339) (0.903) (0.000) NET SFR -0.046 -0.014 -0.018 0.046* -0.004 0.075** 0.011* -0.007 -0.055* (0.000) (0.000) (0.000) (0.093) (0.856) (0.014) (0.088) (0.411) (0.000) NET IND -0.00079 -0.002* -0.009* 0.059* -0.071 0.005 -0.001 0.042* 0.012*** (0.452) (0.000) (0.000) (0.000) (0.019) (0.702) (0.822) (0.000) (0.000) NET UK 0.011* 0.022* 0.016* 0.079* 0.088* 0.289 -0.0051* -0.045* -0.044 (0.071) (0.000) (0.000) (0.000) (0.000) (0.011) (4.45) (0.000) (0.409) NET EUR 0.002 -0.012* -0.011* 0.197 0.186* 0.064* -0.103 -0.121* -0.062* (0.297) (0.000) (0.000) (0.167) (0.000) (0.051) (.176) (0.000) (0.000) NET CHIN -0.0002 0.009* 0.010* -0.079 0.014 -0.030 0.024* 0.007 0.007 (0.624) (0.000) (0.000) (0.016) (0.382) (0.021) (0.000) (0.155) (0.187) NET KOR -0.010* -0.021* -0.021* -0.177* -0.042 -0.174 0.009 0.027 0.032 (0.000) (0.000) (0.000) (0.000) (0.177) (0.011) (0.638) (0.028) (0.336) NET CAN 0.014* 0.017* 0.019* 0.109*** 0.049 0.065* -0.104*** -0.023* -0.014 (0.000) (0.000) (0.000) (0.000) (0.110) (0.096) (0.000) (0.081) (0.018) NET JPY -0.012* -0.026* -0.0194* -0.561* -0.285* -0.099 -0.050 0.082* 0.062 (0.000) (0.000) (0.000) (0.000) (0.000) (0.106) (0.555) (0.000) (0.000) Notes: This table reports the estimated coefficients of the effect of EPU, VIX and OVX on the total and net dynamic connectedness, as in Eq. (9). The numbers between parentheses are the probabilities of the t-statistics. (*), (), (*) indicate the parameter significance at 1%, 5%, and 10% significance level, respectively. Fig. 6 Quantile process estimates for the impact of EPU, VIX, and OVX on total connectedness for the full sample period. Fig. 6 Fig. 7 Quantile process estimates for the impact of EPU, VIX, and OVX on total connectedness before COVID-19. Fig. 7 Fig. 8 Quantile process estimates for the impact of EPU, VIX, and OVX on total connectedness during COVID-19. Fig. 8 Regarding the impact of VIX and OVX, we observe the following: VIX positively impacts total connectedness at low quantiles and then decreases over high quantiles. This indicates that investors react positively to this risk factor and interpret that as good news under which they tend to increase their investment in these assets. The impact of OVX is negative at the low quantiles but then positive at high quantiles, indicating that the oil volatility exerts a negative effect on total connectedness when the market is in a stress condition, while total connectedness reacts positively when markets are at normal and high levels. The results of the total spillover before COVID-19 are similar to those of the full sample. We see that EPU has a significant negative impact on the total spillover among all currencies at all quantiles. Although this impact fluctuates when the spillover effects are low, it shows a decreasing pattern when the level of connectedness is medium or high. The total spillover among all currencies becomes increasingly exposed to the negative impact of economic policy uncertainty in normal and high levels of connectedness. Looking at the results of the net spillover in the full sample (as in Table 6), we notice more interesting results. The results report that EPU positively affects the net spillover of Bitcoin, Litecoin, and Ethereum at low quantiles but negatively for Ripple. Nevertheless, this impact turns negative at upper quantiles for Bitcoin and Litecoin and positive for Ethereum and Ripple. This means that the net spillover of Bitcoin and Litecoin become more exposed to the negative impact at the upper tail of the net connectedness distribution, making the two cryptocurrencies a good hedge against this market uncertainty in times of high-stress conditions. Regarding the impact of VIX on the net connectedness, we notice a positive impact on the Swiss Franc, Indian Rube, The British pound, and the Euro at low quantiles, while there is a negative impact on the other traditional currencies. Those results also hold at the upper quantile. The negative impact of the fear index on some traditional currencies means that they are highly exposed to this factor, yet, they might play a good hedge against this market uncertainty at times of stress or when the markets are in times of crisis. Then in terms of the impact of OVX, we notice that most traditional currencies' net spillover reacts negatively to this factor at a low quantile level, except for the Swiss franc and Indian Rube. However, at a high quantile level of the distribution, oil volatility has an alternating impact. The Swiss franc, British pound, Euro, Canadian Dollar, and Japanese yen react negatively, while Indian Rube, Chinese Yuan, and Korean Won, react positively. This indicates again that some traditional currencies can play the role of diversifiers in times of stress or when the markets go through turmoil periods, such as those happening during the COVID-19 period. Table 7, Table 8 again present the impact of the three factors on total connectedness and net spillover over the two sub-periods (before and during the COVID-19 crisis). Regarding the subperiod before COVID-19 period, Table 7 indicates that EPU positively affects the net spillover of the four cryptocurrencies at low and high quantile levels, meaning that the four assets react positively to this factor. However, the results are different for the impact of VIX on the four cryptocurrency net spillovers. Bitcoin is the only cryptocurrency that reacts negatively to the fear index under all market conditions, making it a good hedge against this market volatility under all market conditions. Also, Bitcoin reacts negatively to OVX at a low quantile level, along with Litecoin, implying the importance of these two cryptos for hedging and diversification benefits against the changes in fluctuations of oil prices during times of stress Before the COVID-19 pandemic, EPU had a significant positive effect on the net connectedness of the British Pound, and Canadian Dollar, at all quantiles. The EPU positively impacts the Euro in the lower quantile, yet, turns negative at the medium and high quantiles. Swiss France, Indian Rube, Chinese Yuan, Korean Won, and Japanese Yen all react negatively to the EPU under all market conditions. Regarding the fear index (VIX), it is noticed that British Pound, and the Euro, react positively to this factor under all market conditions, while the rest alternate in terms of their reaction, depending on whether the market is in a bearish or bullish condition. For example, the Swiss franc is negatively reacting to VIX at low and high quantiles, making this currency a good hedge against this market uncertainty under extreme market stress. The same applied to Indian Rube, Korean Won, Canadian Dollar, and Japanese Yen. When it comes to OVX, we notice a negative reaction of the British Pound, The Euro, and the Canadian Dollar to the oil risk factor under all market conditions, while the Swiss Franc is affected negatively when the market is in a bullish state. The results during the COVID-19 period are surprisingly different for traditional currencies. Although some of the traditional currencies (Swiss franc, Indian Rupee, Korean Won, and the Japanese Yen) are exposed to a negative impact of EPU, the net spillovers for other traditional currency show evidence of a positive relationship with EPU (e.g., the Pound, and the Canadian Dollar). For the impact of VIX on the traditional currencies' net spillover, Korean Won and Japanese Yen react negatively at all quantiles, while the British Pound, the Euro, and the Canadian Dollar react positively at all quantiles. That makes the Korean Won and Japanese Yen good hedges against this market uncertainty during the COVID-19 period under all market conditions. The Chinese Yuan also reacts negatively to VIX when the market is in bearish and bullish states, making this currency another good candidate for hedging the fear market uncertainty under extreme market conditions. The Swiss Franc and Indian Rube are affected positively under extreme market conditions. When it comes to the oil price volatility (OVX), the British Pound, the Euro, and the Canadian Dollar, are all affected negatively by this factor, providing investors with some hedging opportunities during the COVID-19 period against this market uncertainty. The Indian Rube, Chinese Yuan, Korean Won, and Japanese Yen, react positively to oil price volatility under all market conditions. The Swiss Franc also is negatively impacted at a high quantile, making it a good candidate for hedging this uncertainty during the COVID-19 period. Our results fall in-line with those by Demir et al. (2018), Dybrberg (2016), and Briere et al. (2015), among others. For example, Demir et al. (2018) suggested that Bitcoin is an effective tool for hedging, especially during bullish market conditions, while it showed some diversification benefits during bearish market conditions. Dyhrberg (2016) further suggested that Bitcoin can be an advantageous instrument for portfolio and risk management when it is isolated from other financial assets. Briere et al. (2015), Corbet et al., 2018, Corbet et al., 2021, and Lee et al. (2018) also reported that Bitcoin offers diversification benefits due to being weakly correlated with other traditional assets. Yet, the disadvantage of cryptocurrency, as explained by Kristoufek (2018), was the low liquidity of the cryptocurrency market compared to other financial markets, like the equity and foreign exchange markets. 5 Conclusion The paper examined the dynamic connectedness among eight traditional currencies and four cryptocurrencies before and over the COVID-19 pandemic periods. We found strong evidence of dynamic spillover across currency markets, with a noticeable increase during the COVID-19 pandemic period. During the COVID-19 period, a positive jump in the total spillover of both types of currencies was clearly noticed, reaching its highest level between the end of 2020 and the beginning of 2021. Then, it fell somehow during 2021 but then increased during the 2022 period, marking the impact of the Russian-Ukraine war. The fall in total spillover in the early days of the crisis indicates that investors had less fear and became more confident because of the final confirmation of the COVID-19 vaccination discovery. Our results revealed that the total spillover among traditional currencies is slightly higher than the spillover among cryptocurrencies during the crisis periods. Before COVID-19, cryptocurrencies were found to be the most transmitters of shocks compared to traditional currencies, with Ethereum being the highest transmitter and Ripple the lowest one. The highest traditional currencies transmitters are the Swiss Franc and the Euro, while the lowest transmitters are the Indian Rupee and the Japanese Yen. We also reported evidence of a significant rise in the level of volatility transmission across all types of currencies during the pandemic crisis, where the majority of traditional currencies and cryptocurrencies became the highest transmitters. These results confirm that the dominance of currencies in the shock transmission changes from one currency to another before and during COVID-19. Traditional currencies (EUR, UK, SFR, and CAN) as well as cryptocurrencies (BIT, ETH, and LIT) are the dominant transmitters of shocks during the crisis, with the Euro, British Pound, and Swiss Franc being the most interconnected with other currencies. In addition, Litecoin played an important role as a safe haven asset that dominated Bitcoin during the pandemic crisis. Overall, during the COVID-19 period, the dependence was found to be positive and higher for the pairs of the same market than those across markets. As both markets (cryptocurrencies and fiat currencies) are weakly connected, investors might benefit more from holding these assets together for portfolio diversification purposes. Looking at the two markets as separate groups, total spillover among cryptocurrencies, on average, during the COVID-19 crisis was found to be higher than before the crisis period. Ethereum and Litecoin are the highest contributors to other cryptocurrencies during and before the crisis. That shows cryptocurrencies to be considered as good diversifiers for portfolio construction during the pandemic period, with a noticeable increase in contribution levels during the pandemic period. The net spillover among cryptocurrencies indicates that Ethereum is the highest net transmitter before the crisis, while Ripple is found to be the highest net receiver. Overall, during the COVID-19 period, cryptocurrencies showed evidence of more net transmitters and net receivers than before COVID-19. Similarly, the results obtained for the traditional currencies showed that the total spillover among traditional currencies, on average, during the COVID-19 crisis, is higher than their total spillover before crisis period. The highest contributors to other traditional currencies are the Swiss Franc, Euro, Chinese Yuan, Korean Won, and Canadian Dollar in both sub-periods. Yet, during the crisis, there is a noticeable increase in the level of contribution of the traditional currencies compared to the period before the pandemic. Before the crisis, two traditional currencies were net transmitters (the Swiss Franc and the Euro), while the others were net receivers. The highest net transmitter is the Euro, whereas the Indian Rupee is the highest net receiver. However, the results of net spillover are somewhat different during the COVID-19 period. The net transmitters are the Swiss Franc, the British Pound, the Euro, and the Canadian Dollar, with the Euro being the highest transmitter. The position of British Pound and Canadian Dollar is changed from being net receivers before the crisis to net transmitters during the crisis. The other currencies maintained their positions as net receivers of shocks during the pandemic period. The highest net receivers are the Indian Rupee and the Japanese Yen. Overall, traditional currencies display a higher level of net spillover during the COVID-19 period. The paper also provided evidence on the impact of economic policy uncertainty (EPU) on total spillovers and net spillovers among cryptocurrencies and traditional currencies. Our results confirmed the presence of a heterogeneous impact of EPU on the dynamic connectedness for both types of currencies under different spillover distribution levels (e.g., low, normal, and high). Specifically, the results show a negative effect of EPU on the total spillover among all currencies at all quantiles for both the full sample and COVID-19 periods. These results can be explained by the reaction of investors to the increase in uncertainty. Investors may interpret uncertainty as bad news, which tends to increase their investment in currencies. In general, we found that EPU has a significant negative impact on the total spillover among all currencies at all quantiles. Although this impact fluctuates when the spillover effects are low, it shows a decreasing pattern when the level of connectedness is medium or high. The total spillover among all currencies becomes increasingly exposed to the negative impact of EPU in normal and high levels of connectedness. Regarding the impact of VIX and OVX on the level of connectedness, contradictory results were reported. VIX positively impacted total connectedness at low quantiles, then decreased gradually over the high quantiles. This indicates that investors react positively to this risk factor and interpret that as good news under which they tend to increase their investment in these assets. However, OVX showed a negative impact on the total connectedness at low quantiles but then turned positive at high quantiles. This means that oil market volatility exerts a negative effect on total connectedness when the market is in a stressful condition. In contrast, total connectedness reacts positively to OVX when markets are at normal and high levels. Our findings have many important policy implications. First, it can help investors select the appropriate currencies that lead to better portfolio diversification. Indeed, our results suggest that investors in the foreign exchange market can use cryptocurrency investment as a hedge against the risk associated with the foreign exchange market. Second, the nature of the relationship between economic policy uncertainty and the connectedness of different currencies with cryptocurrencies can offer useful information to promote economic prosperity, especially in periods of extreme market conditions, such as the COVID-19 pandemic. Third, our paper is also of great significance to policymakers. They can rely on our results to properly manage foreign currency risk and other potential risks associated with currency markets. This way, policymakers can control the risks in currencies of significant connectedness and then decide on the timing and extent of foreign currency rate policy intervention. Finally, unveiling the causes of currency price variations leads to a better selection of the source of oscillation that affects economic stability differently. This study could be extended in the future in some ways. First, the sample can be extended to include other cryptocurrencies or traditional currencies to examine their connectedness across the virtual and foreign exchange markets. Second, it can also include other asset classes, such as the recently created NFTs or other virtual assets, to explore their hedging and safe haven properties for cryptocurrencies under different market conditions and their response to policy uncertainty indices. Third, different research methodologies can be used, like copula or multivariate dependence models, to study the dependence structure. Fourth, analyzing the risk benefits of combining the two markets in terms of their risk value, expected shortfall, or other risk measures during normal and crisis periods can be used as another aspect for future research. Finally, applying a tail-dependence approach within the Diebold-Yilmaz framework to analyze the connectedness among cryptos, traditional assets, and other alternative assets might be a fruitful line for future research. Declaration of Interests The authors declare that they have no financial and personal relationships with other people or organizations that could inappropriately influence (bias) their work. Appendix A Literature review summary TableReferences Targeted assets The model Global crises Findings Volatility connectedness among global currency markets Kitamura (2010) AUD, GBP, CAD, EUR, JPY, CHF M-GARCH model Volatility connectedness from the Euro significantly impacts the Swiss franc and Japanese yen during the global financial crisis (GFC) period Baruník et al. (2017) AUD, GBP, CAD, EUR, JPY, CHF 2N-dimensional VAR model Negative volatility connectedness among currencies dominated positive ones Greenwood-Nimmo et al. (2016) G10 currencies Empirical network model The 2008 global financial crisis (GFC) Volatility connectedness among currencies was time-varying and it increased widely during financial crises Salisu and Ayinde (2018) USD, EUR, GBP, JPY, CHF, the West African Unit of Account (WAUA) The VAR model Around the election date in Nigeria and the GFC Volatility connectedness among Naira and six most-traded currencies rose during the election process in Nigeria more than during the GFC Mai et al. (2018) Global currency markets, East Asian and the European currency markets Currency correlation matrix and impact elimination method The East Asian currencies were more strongly connected than the European currencies due to the strong co-movement of currencies in the East Asian region Antonakakis (2012) Major foreign currency rates in Europe Dynamic correlations and VAR-based spillover Before and after the introduction of the Euro currency Volatility connectedness between major foreign currency rates was weak in the post-euro period Kočenda and Moravcová (2019) Three currencies in the new EU foreign exchange markets Dynamic Conditional Correlation (DCC) model and the generalized vector autoregressive (GVAR) variance decomposition Before the GFC and during GFC as well as during the European debt crisis Volatility connectedness among currencies rose significantly rose during the GFC Return connectedness among global currency markets Shu et al. (2015) The Chinese Yuan and Asia-Pacific countries’ currencies Multiple regression with causality Before and after the transmission of China's monetary stance The offshore currency exchange of the Chinese currency exerted more effect on the Asia-Pacific currencies than the onshore Chinese currency market due to China's monetary policy transmission Sehgal et al. (2017) South East Asian currencies Constant and time-varying Copula-GARCH models Connectedness among currencies return of the South Asian member countries was very weak Wei et al. (2020) The Belt and Road currency market Time-varying parameter vector autoregressive (TVP-VAR) framework During the regional and global crises including COVID-19 Connectedness among RMB and "the Belt and Road" currencies was stronger during the regional and the GFC crises but it was disrupted during the COVID-19 period due to internal financial reforms, as well as external economic shocks in the region. Orlowski (2016) The Euro and the non-Euro currencies GARCH and BVAR models, During the GFC The positive connectedness between the Euro and the non-Euro currencies rose significantly during the GFC McCauley and Shu (2019) The Chinese currency (RMB) with regional and other emerging market currencies Simple and multiple regressions After the three post-reform periods of RMB management (transition, basket of currencies management, and countercyclical management) Connectedness between regional and Latin American currencies rose in the basket period but declined after the countercyclical period Volatility connectedness among cryptocurrencies and/or traditional currencies Kumar and Anandarao (2019) Bitcoin, Ethereum, Ripple, and Litecoin Multivariate IGARCH-DCC and the pairwise wavelet cross-spectral analysis Volatility connectedness was very strong Bitcoin and other cryptos (Ethereum and Litecoin). Wen and Wang (2020) 65 major global currencies LASSO-VAR approach Oil price crashes, exchange rate regimes, and monetary policy changes) The sensitivity between volatility connectedness among the US dollar and the Euro and changes in international economic fundamentals increased during crisis periods (oil price crashes, exchange rate regimes, and monetary policy changes) Fung et al. (2022) 254 cryptocurrencies GARCH family models Volatility connectedness among cryptocurrencies persist over time. Return connectedness among traditional currencies and/or cryptocurrencies Baumöhl (2019) Six traditional currencies, and six cryptocurrencies The quantile cross-spectral approach A negative connectedness was found between currencies and cryptocurrencies in both the short- and long-term horizons Hsu et al. (2021) Cryptocurrency and traditional currencies or gold markets Using a diagonal BEKK model During COVID-19 Cryptocurrencies were strongly connected with traditional currencies and gold, acted as a safe haven during the COVID-19 pandemic against risk spillover Mokni and Ajmi (2020) Bitcoin, Ethereum, Ripple, and Litecoin and Bitcoin Cash and the US dollar Granger-causality in quantiles Before and during the COVID-19 period During the COVID-19 crisis, the connectedness between the US dollar and cryptocurrencies was stronger at higher and lower tails of the distribution and cryptocurrencies were good predictors, and acted as hedgers against the US dollar volatility. ElSayed et al. (2022) Bitcoin, Litecoin, Ripple, and nine major foreign currency markets The generalized vector autoregressive (GVAR) variance decomposition and the Bayesian graphical structural vector autoregressive estimations During the 2017-2018 cryptocurrency crash Bitcoin and Litecoin were the most connected currencies during the three quarters of 2017. Except for the Chinese Yuan, major traditional currencies did not significantly impact cryptocurrencies. Bouri et al. (2021) Bitcoin, Ethereum, Ripple, Litecoin, Stellar, Monero, Dash Quantile VAR approach During COVID-19 The connectedness among cryptocurrencies rose widely during the COVID-19 outbreak. The level of connectedness increases with increase in shock size either for both positive or negative shocks Kumar et al. (2022) Bitcoin, Ethereum, Ripple, Litecoin, Bitcoin Cash, EOS, Binance Coin, Tether, Bitcoin SV, and Tron The generalized vector autoregression (VAR) framework and Granger causality During COVID-19 The structural change resulted from the COVID-19-pandemic impacted the connectedness among cryptocurrencies EPU and connectedness among traditional currencies and/or among cryptocurrencies Mokni et al. (2020) EPU and connectedness between Bitcoin and the US stock markets DCC-EGARCH model and simple regression After the crash of Bitcoin in December 2017 EPU negatively affected the connectedness among Bitcoin and the US stock markets only after the crash of Bitcoin in December 2017, suggesting that Bitcoin as a hedging instrument to the stock portfolio Huynh et al. (2020) TPU and connectedness between the US exchange rates of globally traded currencies The generalized vector autoregressive (VAR) variance decomposition Trade policy uncertainty impacted the connectedness among the US dollar exchange rates of globally traded currencies Chen et al. (2020) EPU and China’s exchange rate volatility The quantile regression approach EPU positively affected China’s exchange rate volatility on all quantiles. Alam et al. (2019) Oil, foreign currency futures for AUD, CAD, CHF, EUR, GBP and JPY against the US dollar Wavelet-Granger causality method of Olayeni (2016) During the GFC period and the European sovereign debt crisis Oil prices were strongly connected to exchange rates of six major bilateral currencies against the US dollar Al-Shboul et al. (2022) Bitcoin, Ripple, Litecoin and Ethereum Quantile VAR approach During the COVID-19 pandemic The effect of cryptocurrency uncertainty on the return connectedness increased significantly during the COVID-19 crisis Appendix B Total average dynamic connectedness based on QVAR TableFull Sample BIT ETH LIT RIP SFR IND UK EUR CHIN KOR CAN JPY FROM q=0.05 Cont TO others 84.51 87.35 85.53 81.80 92.23 82.69 90.43 94.31 82.97 88.11 87.16 85.52 1042.61 Cont incl own 97.34 99.97 98.67 95.41 105.08 95.57 103.42 107.32 96.50 101.30 100.39 99.01 TCI Net spillovers -2.66 -0.03 -1.33 -4.59 5.08 -4.43 3.42 7.32 -3.50 1.30 0.39 -0.99 86.88 q=0.5 Cont TO others 65.36 72.13 70.04 51.30 64.31 30.06 50.25 73.32 38.07 41.94 43.74 27.84 628.36 Cont incl own 105.19 110.26 108.43 97.91 106.43 88.94 97.75 112.22 94.63 95.22 95.00 88.02 TCI Net spillovers 5.19 10.26 8.43 -2.09 6.43 -11.06 -2.25 12.22 -5.37 -4.78 -5.00 -11.98 52.36 q=0.95 Cont TO others 87.57 88.60 86.29 85.88 89.80 85.22 87.70 92.35 86.21 85.56 85.92 83.42 1044.53 Cont incl own 100.43 101.93 99.81 98.90 102.51 97.69 100.61 104.94 99.24 98.29 99.04 96.59 TCI Net spillovers 0.43 1.93 -0.19 -1.10 2.51 -2.31 0.61 4.94 -0.76 -1.71 -0.96 -3.41 87.04 Before COVID-19 q=0.05 Cont TO others 87.80 89.86 83.88 80.82 89.81 78.87 91.78 93.97 84.90 89.79 87.26 85.55 1044.30 Cont incl own 100.26 102.31 97.31 94.52 102.30 91.69 104.56 107.02 97.56 103.22 100.49 98.76 TCI Net spillovers 0.26 2.31 -2.69 -5.48 2.30 -8.31 4.56 7.02 -2.44 3.22 0.49 -1.24 87.03 q=0.5 Cont TO others 56.59 65.93 57.79 45.83 59.30 27.11 35.03 70.47 43.90 45.21 32.90 28.15 568.20 Cont incl own 102.04 107.82 101.73 96.52 106.26 91.34 92.60 113.26 98.94 100.46 94.99 94.03 TCI Net spillovers 2.04 7.82 1.73 -3.48 6.26 -8.66 -7.40 13.26 -1.06 0.46 -5.01 -5.97 47.35 q=0.95 Cont TO others 87.10 88.01 87.42 84.46 90.43 85.35 85.80 93.35 89.80 85.26 83.57 84.15 1044.68 Cont incl own 99.75 100.80 101.14 97.05 103.40 97.61 99.40 105.90 102.63 97.74 97.18 97.42 TCI Net spillovers -0.25 0.80 1.14 -2.95 3.40 -2.39 -0.60 5.90 2.63 -2.26 -2.82 -2.58 87.06 During COVID-19 q=0.05 Cont TO others 85.68 85.22 89.17 83.43 93.74 83.31 89.07 92.27 80.23 88.18 87.24 82.46 1040.01 Cont incl own 99.23 98.44 102.43 97.31 106.77 95.47 102.45 105.22 94.56 101.49 100.44 96.18 TCI Net spillovers -0.77 -1.56 2.43 -2.69 6.77 -4.53 2.45 5.22 -5.44 1.49 0.44 -3.82 86.67 q=0.5 Cont TO others 71.46 75.07 81.15 50.18 71.34 33.77 62.54 75.39 29.57 42.51 51.00 27.97 671.95 Cont incl own 108.58 112.71 116.73 98.14 108.76 85.88 101.24 110.91 89.39 91.99 91.86 83.81 TCI Net spillovers 8.58 12.71 16.73 -1.86 8.76 -14.12 1.24 10.91 -10.61 -8.01 -8.14 -16.19 56.00 q=0.95 Cont TO others 90.21 89.64 86.48 86.63 89.28 84.83 88.85 92.77 78.75 85.05 89.46 83.67 1045.63 Cont incl own 103.53 103.43 99.59 99.82 101.60 97.32 100.95 105.11 91.80 97.47 102.47 96.91 TCI Net spillovers 3.53 3.43 -0.41 -0.18 1.60 -2.68 0.95 5.11 -8.20 -2.53 2.47 -3.09 87.14 Notes: This table presents the average of connectedness measures based on Diebold and Yilmaz (2012, 2014) estimated from a Q-VAR model. For each quantile’s order, we provide the directional volatility spillover received (denoted by “From”), and transmitted (denoted by “Cont TO others”) by each variable. Then, we obtain the net directional spillover (denoted by “Net Spillovers”) as the difference between directional ‘To’ spillovers and directional ‘From’ spillovers. Uncited references (Assaf et al., 2021, Balcilar et al., 2015, Boubakri et al., 2019, Cheng and Yen, 2020, Fang et al., 2019, Li et al., 2015, Yen and Cheng, 2021) Data availability Data will be made available on request. 1 Another strand of literature focused on the volatility connectedness between traditional currencies and cryptocurrencies. For example, Andrada-Félix, Fernandez-Perez, and Sosvilla-Rivero (2020) found evidence of volatility connectedness across cryptocurrencies and traditional currencies. The volatility connectedness was found to be time-varying, especially during the periods of economic and financial instability, due to idiosyncratic shocks in the markets. 2 When analyzing the spillover effects based on the TVP– VAR approach, we made the results using several window sizes and forecasting periods. However, the results remain similar, indicating the robustness of our results. Also, we obtained all the obtained all the pair-wise connectedness graphs for the full sample, before COVID-19 and during COVID-19 periods. To save space, we have not reported them, yet, they are available upon request. 3 Based on the request of a referee, we also obtain the extreme spillover based on the QVAR as suggested by Bouri et al. (2021). The results are reported in Appendix B. As can be seen, the results are similar for the before COVID-19 and during COVID-19 periods. The total connectedness at the extreme quantiles are in the range of 86%, while at the median level, they are 52%, 47%, and 56% for the full sample, before COVID-19, and COVID-19 periods, respectively. 4 We thank a referee for the suggestion of including VIX and OVX in the regression analysis. 5 To save space, we only report the figures for the total connectedness and not those related to net connectedness of each cryptocurrency/traditional currency. Yet, they are available upon request. ==== Refs References Abid A. Economic policy uncertainty and exchange rates in emerging markets: Short and long runs evidence Finance Research Letters 37 2020 101378 Alam M.S. Shahzad S.J.H. Ferrer R. Causal flows between oil and forex markets using high-frequency data: Asymmetries from good and bad volatility Energy Economics 84 2019 104513 Al-Shboul M. Assaf A. Mokni K. When bitcoin lost its position: Cryptocurrency uncertainty and the dynamic spillover among cryptocurrencies before and during the COVID-19 pandemic International Review of Financial Analysis 83 2022 102309 Andrada-Félix J. Fernandez-Perez A. Sosvilla-Rivero S. Distant or close cousins: Connectedness between cryptocurrencies and traditional currencies volatilities Journal of International Financial Markets, Institutions and Money 67 2020 101219 Antonakakis N., Gabauer D. (2017) Refined Measures of Dynamic Connectedness based on TVP-VAR. MPRA Paper No. 78282. Antonakakis N. Exchange return co-movements and volatility spillovers before and after the introduction of euro Journal of International Financial Markets, Institutions and Money 22 5 2012 1091 1109 Antonakakis N. Kizys R. Dynamic spillovers between commodity and currency markets International Review of Financial Analysis 41 2015 303 319 Antonakakis N. Chatziantoniou I. Gabauer D. Cryptocurrency market contagion: market uncertainty, market complexity, and dynamic portfolios Journal of International Financial Markets, Institutions and Money 61 2019 37 51 Antonakakis N. Chatziantoniou I. Gabauer D. Refined measures of dynamic connectedness based on time-varying parameter vector autoregressions Journal of Risk and Financial Management 13 4 2020 84 Arouri M. Estay C. Rault C. Roubaud D. Economic policy uncertainty and stock markets: Long-run evidence from the US Finance Research Letters 18 2016 136 141 Aslan A. Sensoy A. Intraday efficiency-frequency nexus in the cryptocurrency markets Finance Research Letters 35 2020 101298 Assaf A. Charif H. Mokni K. Dynamic connectedness between uncertainty and energy markets: Do investor sentiments matter? Resources Policy 72 2021 102112 Baker S.R. Bloom N. Davis S.J. Has economic policy uncertainty hampered the recovery? Becker Friedman Institute for Research In Economics Working Paper 2012 2012-003 Baker S.R. Bloom N. Davis S.J. Measuring economic policy uncertainty The Quarterly Journal of Economics 131 4 2016 1593 1636 Balcilar M. Gupta R. Miller S.M. Regime switching model of US crude oil and stock market prices: 1859 to 2013 Energy Economics 49 2015 317 327 Balcilar M. Gupta R. Kyei C. Wohar M.E. Does economic policy uncertainty predict exchange rate returns and volatility? Evidence from a nonparametric causality-in-quantiles test Open Economies Review 27 2 2016 229 250 Baruník J. Kočenda E. Vácha L. Asymmetric volatility connectedness on the forex market Journal of International Money and Finance 77 2017 39 56 Baumöhl E. Are cryptocurrencies connected to forex? A quantile cross-spectral approach Finance Research Letters 29 2019 363 372 Bech, M.L., & Garratt, R. (2017). Central bank cryptocurrencies. BIS Quarterly Review September. Beckmann J. Czudaj R. Exchange rate expectations and economic policy uncertainty European Journal of Political Economy 47 2017 148 162 Boako G. Alagidede P. Currency price risk and stock market returns in Africa: Dependence and downside spillover effects with stochastic copulas Journal of Multinational Financial Management 41 2017 92 114 Boubakri S. Guillaumin C. Silanine A. Non-linear relationship between real commodity price volatility and real effective exchange rate: The case of commodity-exporting countries Journal of Macroeconomics 60 2019 212 228 Bouri E. Gupta R. Roubaud D. Herding behaviour in cryptocurrencies Finance Research Letters 29 2019 216 221 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 Research Letters 20 2017 192 198 Bouri E. Saeed T. Vo X.V. Roubaud D. Quantile connectedness in the cryptocurrency market Journal of International Financial Markets, Institutions and Money 71 2021 101302 Briere M. Oosterlinck K. Szafarz A. Virtual currency, tangible return: Portfolio 126 diversification with Bitcoin Journal of Asset Management 16 6 2015 365 373 Cao G. Xie W. Asymmetric dynamic spillover effect between cryptocurrency and China's financial market: Evidence from TVP-VAR based connectedness approach Finance Research Letters 49 2022 103070 Chemkha R. BenSaïda A. Ghorbel A. Connectedness between cryptocurrencies and foreign exchange markets: Implication for risk management Journal of Multinational Financial Management 59 2021 100666 Chen L. Du Z. Hu Z. Impact of economic policy uncertainty on exchange rate volatility of China Finance Research Letters 32 2020 101266 Cheng H.P. Yen K.C. The relationship between the economic policy uncertainty and the cryptocurrency market Finance Research Letters 35 2020 101308 Cheng T. Liu J. Yao W. Zhao A.B. The impact of COVID-19 pandemic on the volatility connectedness network of global stock market. Pacific-Basin Finance Journal 71 2022 101678 Conlon T. Corbet S. McGee R.J. Are cryptocurrencies a safe haven for equity markets? An international perspective from the COVID-19 pandemic Research in International Business and Finance 54 2020 101248 Colon F. Kim C. Kim H. Kim W. The effect of political and economic uncertainty on the cryptocurrency market Finance Research Letters 39 2021 101621 Corbet S. Hou Y.G. Hu Y. Oxley L. Xu D. Pandemic-related financial market volatility spillovers: Evidence from the Chinese COVID-19 epicentre International Review of Economics & Finance 71 2021 55 81 Corbet S. Meegan A. Larkin C. Lucey B. Yarovaya Exploring the dynamic 132 relationships between cryptocurrencies and other financial assets Economics Letters, 133 165 2018 28 34 Demir E. Gozgor G. Lau C.K.M. Vigne S.A. Does economic policy uncertainty predict the Bitcoin returns? An empirical investigation Finance Research Letters 26 2018 145 149 Dickey D.A. Fuller W.A. Distribution of the estimators for autoregressive time series with a unit root Journal of the American statistical association 74 366a 1979 427 431 Diebold F.X. Yilmaz K. Measuring financial asset return and volatility spillovers, with application to global equity markets The Economic Journal 119 534 2009 158 171 Diebold F.X. Yilmaz K. Better to give than to receive: Predictive directional measurement of volatility spillovers International Journal of Forecasting 28 1 2012 57 66 Diebold F.X. Yılmaz K. On the network topology of variance decompositions: Measuring the connectedness of financial firms Journal of Econometrics 182 1 2014 119 134 Dyhrberg A.H. Bitcoin, gold and the dollar–A GARCH volatility analysis Finance Research Letters 16 2016 85 92 Elsayed A.H. Gozgor G. Lau C.K.M. Causality and dynamic spillovers among cryptocurrencies and currency markets. International Journal of Finance & Economics 27 2 2022 2026 2040 Fang L. Bouri E. Gupta R. Roubaud D. Does global economic uncertainty matter for the volatility and hedging effectiveness of Bitcoin? International Review of Financial Analysis 61 2019 29 36 Fung K. Jeong J. Pereira J. More to cryptos than bitcoin: A GARCH modelling of heterogeneous cryptocurrencies Finance research letters 47 2022 102544 Gabauer D. Gupta R. On the transmission mechanism of country-specific and international economic uncertainty spillovers: Evidence from a TVP-VAR connectedness decomposition approach Economics Letters 171 2018 63 71 Gozgor G. Tiwari A.K. Demir E. Akron S. The relationship between Bitcoin returns and trade policy uncertainty Finance Research Letters 29 2019 75 82 Greenwood-Nimmo M. Nguyen V.H. Rafferty B. Risk and return spillovers among the G10 currencies Journal of Financial Markets 31 2016 43 62 Grobys K. Are volatility spillovers between currency and equity market driven by economic states? Evidence from the US economy Economics Letters 127 2015 72 75 Guo Y. Li P. Li A. Tail risk contagion between international financial markets during COVID-19 pandemic International Review of Financial Analysis 73 2021 101649 Hasan M.B. Hassan M.K. Karim Z.A. Rashid M.M. Exploring the hedge and safe haven properties of cryptocurrency in policy uncertainty Finance Research Letters 2021 102272 Hassen M.K. Hasan M.B. Halim Z.A. Maroney N. Rashid M.M. Exploring the dynamic spillover of cryptocurrency environmental attention across the commodities, green bonds, and environment-related stocks The North American Journal of Economics and Finance 2022 101700 Hsu S.H. Sheu C. Yoon J. Risk spillovers between cryptocurrencies and traditional currencies and gold under different global economic conditions The North American Journal of Economics and Finance 57 2021 101443 Huynh Luu Duc Toan Nasir Muhammad Ali Nguyen. Duc Khuong "Spillovers and connectedness in foreign exchange markets: The role of trade policy uncertainty The Quarterly Review of Economics and Finance 2020 2020 Ji Q. Liu B.Y. Zhao W.L. Fan Y. Modelling dynamic dependence and risk spillover between all oil price shocks and stock market returns in the BRICS International Review of Financial Analysis 68 2020 101238 Katsiampa P. An empirical investigation of volatility dynamics in the cryptocurrency market Research in International Business and Finance 50 2019 322 335 Kido Y. On the link between the US economic policy uncertainty and exchange rates Economics Letters 144 2016 49 52 Kitamura Y. Testing for intraday interdependence and volatility spillover among the euro, the pound and the Swiss franc markets Research in International Business and Finance 24 2 2010 158 171 Kočenda E. Moravcová M. Exchange rate comovements, hedging and volatility spillovers on new EU forex markets Journal of International Financial Markets, Institutions and Money 58 2019 42 64 Koop G. Korobilis D. A new index of financial conditions European Economic Review 71 2014 101 116 Korobilis D. Yilmaz K. Measuring dynamic connectedness with large Bayesian VAR models Available at SSRN 2018 3099725 Kristoufek L. On Bitcoin markets (in)efficiency and its evolution Physica A: Statistical 171 Mechanics and its Application 503 2018 257 262 Krol R. Economic policy uncertainty and exchange rate volatility International Finance 17 2 2014 241 256 Kumar A.S. Anandarao S. Volatility spillover in crypto-currency markets: Some evidences from GARCH and wavelet analysis Physica A: Statistical Mechanics and its Applications 524 2019 448 458 Kumar A. Iqbal N. Mitra S.K. Kristoufek L. Bouri E. Connectedness among major cryptocurrencies in standard times and during the COVID-19 outbreak Journal of International Financial Markets, Institutions and Money 77 2022 101523 Lee D.K.C. Guo L. Wang Y. Cryptocurrency: A new investment opportunity? 175 Journal of Alternative Investments 20 3 2018 16 40 Li W. Wong M.C. Cenev J. High frequency analysis of macro news releases on the foreign exchange market: A survey of literature. Big Data Research 2 1 2015 33 48 Liu T. Gong X. Analyzing time-varying volatility spillovers between the crude oil markets using a new method Energy Economics 87 2020 104711 Mai Y. Chen H. Zou J.Z. Li S.P. Currency co-movement and network correlation structure of foreign exchange market Physica A: Statistical Mechanics and its Applications 492 2018 65 74 Malik F. Umar Z. Dynamic connectedness of oil price shocks and exchange rates Energy Economics 84 2019 104501 Manela A. Moreira A. News implied volatility and disaster concerns Journal of Financial Economics 123 1 2017 137 162 McCauley R.N. Shu C. Recent renminbi policy and currency co-movements Journal of International Money and Finance 95 2019 444 456 Mensi W. Sensoy A. Aslan A. Kang S.H. High-frequency asymmetric volatility connectedness between Bitcoin and major precious metals markets The North American Journal of Economics and Finance 50 2019 101031 Mokni K. When, where, and how economic policy uncertainty predicts Bitcoin returns and volatility? A quantiles-based analysis The Quarterly Review of Economics and Finance 80 2021 65 73 Mokni, K., & Ajmi, A.N. (2020). Cryptocurrencies vs. US dollar: Evidence from causality in quantiles analysis. Economic Analysis and Policy. Mokni K. Ajmi A.N. Bouri E. Vo X.V. Economic policy uncertainty and the Bitcoin-US stock nexus Journal of Multinational Financial Management 57 2020 100656 Mokni K. Al-Shboul M. Assaf A. Economic policy uncertainty and dynamic spillover among precious metals under market conditions: Does COVID-19 have any effects? Resources Policy 74 2021 102238 Mokni K. Youssef M. Ajmi A.N. COVID-19 pandemic and economic policy uncertainty: The first test on the hedging and safe haven properties of cryptocurrencies Research in International Business and Finance 60 2022 101573 Moratis G. Quantifying the spillover effect in the cryptocurrency market Finance Research Letters 38 2021 101534 Narayan P.K. Devpura N. Wang H. Japanese currency and stock market—What happened during the COVID-19 pandemic? Economic Analysis and Policy 68 2020 191 198 33012962 Orlowski L.T. Co-movements of non-Euro EU currencies with the Euro International Review of Economics & Finance 45 2016 376 383 Olayeni O.R. Causality in continuous wavelet transform without spectral matrix factorization: theory and application Computational Economics 47 3 2016 321 340 Paule-Vianez J. Prado-Román C. Gómez-Martínez R. Economic policy uncertainty and Bitcoin. Is Bitcoin a safe-haven asset?. European Journal of Management and Business Economics.‏ 29 3 2020 347 363 Phillips P.C. Perron P. Testing for a unit root in time series regression Biometrika 75 2 1988 335 346 Primiceri G.E. Time varying structural vector autoregressions and monetary policy The Review of Economic Studies 72 3 2005 821 852 Raza S.A. Shah N. Guesmi K. Msolli B. How does COVID-19 influence dynamic spillover connectedness between cryptocurrencies? Evidence from non-parametric causality-in-quantiles techniques Finance Research Letters 2021 102569 Salisu A.A. Ayinde T.O. Testing for spillovers in naira exchange rates: The role of electioneering & global financial crisis Borsa Istanbul Review 18 4 2018 341 348 Samitas A. Kampouris E. Polyzos S. Covid-19 pandemic and spillover effects in stock markets: A financial network approach International Review of Financial Analysis 80 2022 102005 Sehgal S. Pandey P. Diesting F. Examining dynamic currency linkages amongst South Asian economies: An empirical study Research in International Business and Finance 42 2017 173 190 Shahzad, S.J.H., Balli, F., Naeem, M.A., Hasan, M. and Arif, M. (2021). Do traditional currencies hedge cryptocurrencies?. The Quarterly Review of Economics and Finance. Shu C. He D. Cheng X. One currency, two markets: the renminbi's growing influence in Asia-Pacific China Economic Review 33 2015 163 178 So M.K. Chu A.M. Chan T.W. Impacts of the COVID-19 pandemic on financial market connectedness Finance Research Letters 38 2021 101864 Wei Z. Luo Y. Huang Z. Guo K. Spillover effects of RMB exchange rate among B&R countries: Before and during COVID-19 event Finance research letters 37 2020 101782 Wen T. Wang G.J. Volatility connectedness in global foreign exchange markets Journal of Multinational Financial Management 2020 100617 Wu W. Tiwari A.K. Gozgor G. Leping H. Does economic policy uncertainty affect cryptocurrency markets? Evidence from Twitter-based uncertainty measures Research in International Business and Finance 58 2021 101478 Wu S. Tong M. Yang Z. Derbali A. Does gold or Bitcoin hedge economic policy uncertainty? Finance Research Letters 31 2019 171 178 Yen K.C. Cheng H.P. Economic policy uncertainty and cryptocurrency volatility Finance Research Letters 38 2021 101428 Youssef M. Mokni K. Ajmi A.N. Dynamic connectedness between stock markets in the presence of the COVID-19 pandemic: does economic policy uncertainty matter? Financial Innovation 7 1 2021 1 27	36474632	PMC9715263	NO-CC CODE	2022-12-13 23:16:26	no	Res Int Bus Finance. 2023 Jan 2; 64:101824	utf-8	Res Int Bus Finance	2,022	10.1016/j.ribaf.2022.101824	oa_other
==== Front Oral Surg Oral Med Oral Pathol Oral Radiol Oral Surg Oral Med Oral Pathol Oral Radiol Oral Surgery, Oral Medicine, Oral Pathology and Oral Radiology 2212-4403 2212-4411 The Author(s). Published by Elsevier Inc. S2212-4403(22)01283-4 10.1016/j.oooo.2022.11.012 Review Article Oral lesions in human monkeypox disease and their management: a scoping review Joseph Betsy 1 Anil Sukumaran 23⁎ 1 Department of Periodontics, Saveetha Dental College and Hospital, Saveetha Institute of Medical and Technical Sciences, Chennai, India. 2 Hamad Medical Corporation, Post Box: 3050 Doha, Qatar. 3 College of Dental Medicine, Qatar University, Doha, Qatar ⁎ Corresponding Author. Prof. Dr. Anil Sukumaran, BDS, MDS, Ph.D., FRCPath, FDS RCPS (Glas), FDS RCS (Ed), Professor, Senior Consultant, Hamad Medical Corporation, Post Box: 3050 Doha, Qatar, Phone: +974 50 40 66 70 2 12 2022 2 12 2022 15 9 2022 10 11 2022 27 11 2022 © 2022 The Author(s). Published by Elsevier Inc. 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. Objective Monkeypox (MPX) disease poses a threat to the frontline health workers, including dental practitioners; however, there is limited literature on its dental implications. The objective of this scoping review was to map the oral manifestations of MPX and its management based on existing information. Materials and Methods Articles published up to July 31, 2022, were searched to select relevant observational and experimental studies in humans that reported oral lesions in MPX disease, including case reports. The findings of this review are based on the pooled data of 1136 patients (age range: 2–52 years) reported from different parts of the world. Results Oral lesions included mouth sores, oral mucosal lesions, ulcers on the tongue, tongue swelling, pustular lesions on the gingiva, perioral erosive lesions, oral candidiasis, and oropharyngeal lesions. Oral lesions of MPX infection and their management strategies are relevant to dentists. Dental practitioners may be the first to detect the initial symptoms of MPX disease. Conclusion Oral lesions of may present as initial lesions of MPX suggesting that dentists and dental personnel should be aware of the nature of the disease. Clinicians must be alert to rashes resembling MPX lesions, and distinguish MPX from herpetic and similar vesicular-bullous lesions for differential diagnosis. Symptomatic and supportive care for oral lesions is important. Keywords Monkeypox monkeypox virus oral manifestations oral ulcer disease management ==== Body pmcIntroduction The emergence of monkeypox (MPX) in humans is a global concern during the current COVID-19 crisis. Human MPX is a viral disease of animal origin (zoonosis) caused by a virus of the orthopox family and is closely associated with smallpox viruses. Although the first case was reported in the mid-1900s in Africa, it is now spreading to different parts of the world, which has led the World Health Organization (WHO) to declare it as a "global emergency." Monkeypox virus (MPXV) infections have been reported not only in African countries, but also in high-income settings such as the USA, Canada, Australia, Singapore, and several European countries, including the UK, Sweden, Spain, Portugal, Netherlands, Italy, Germany, Belgium, and France.1 New cases have been reported in Europe without any travel history to Africa or history of contact with infected persons.2 One of the hypotheses regarding the current outbreak is the general decline in the population's immunity to smallpox and similar orthopox virus diseases, as the smallpox vaccination program was discontinued 30 years ago.3 The cases of MPXV infections mostly occurred among young adults and small children who were unvaccinated against smallpox4 and had no other comorbidity.5, 6, 7 In some cases, patients below 18 years were more likely to be hospitalized.8 The incubation period from initial animal exposure to the onset of illness ranged from 11 to 18 days.5 , 8 The clinical features include fever, rash, sweats, chills, lymphadenopathy, headache, stiff neck, red eyes, runny nose, sore throat, cough, wheezing, nausea and/or vomiting, abdominal pain, scrotal lump, itchy maculopapular rashes, and confusion.5 , 8 Concurrent rashes range from 5–25 (benign) to 101–250 (grave) lesions depending on the cases.5 , 7 Oral lesions are found in many cases;7 , 9 however, there are also reports of cases without oral manifestations.10 When present, these lesions can be infectious until all the lesions are crusted.11 Although there is no conclusive evidence of how the infection spreads, a primary animal-to-human infection by direct or indirect contact with MPXV-infected animal's body fluids12 is evident. Broken skin, mucosa of the eyes, nose, mouth, and respiratory tract are possible routes of viral entry to the body.12 Human transmission is often documented through contact with respiratory droplets, body fluids, lesion material, and contaminated clothes.13 This transmission mode has been considered a significant threat to community health after the 2018 Nigerian outbreak. Diagnosis is made based on viremia, samples from upper respiratory tract swabs, and the presence of MPXV DNA by multiple molecular assays, especially polymerase chain reaction (PCR) of the samples collected from the lesions.13 In July 2022, the WHO declared MPX infection a public health emergency of international concern (PHEIC). This should alert the frontline health workers, including dental practitioners, to the high risk of developing MPX disease if they come in contact with infected patients and their body fluids. We have witnessed the catastrophic impact of the COVID-19 pandemic on dentists and dental practices. In this context, it is necessary to discuss the implications of MPXV infections on dental practice. Therefore, the objective of this scoping review was to map the oral manifestations of MPX disease and its management based on existing information. Materials and methods This scoping review aimed to map the oral manifestations of MPX disease and its management, based on existing information. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist was used. The protocol was prepared after a preliminary search of existing literature. Eligibility criteria Articles published up to August 15, 2022, were searched to select appropriate studies using the following (Population/Concept/Context) PCC framework: Population: human patients (all ages); Concept: case reports, observational, and experimental studies; Context: MPX disease in humans. Case reports and observational and experimental studies conducted in English were included in this review. However, in vitro, and in vivo studies, editorials, commentaries, letters to the editor, conference papers, and consensus papers were excluded. Search strategy Prominent literature databases, such as MEDLINE/PubMed, Scopus, Google Scholar, and Web of Science, were extensively searched on August 15, 2022. The search was conducted based on "Monkeypox" and "Oral lesions," which are the key concepts in the research question. Literature that contained the MeSH terms, keywords, and other free terms related to "monkeypox," "monkeypox," "monkeypox virus," "monkeypox virus," "MPX," "MPXV," "oral lesions," "oral ulcers," "mouth," "perioral," "pharynx," "oropharynx," and "nasopharynx" were searched with Boolean operators and included in the initial screening. In addition, preprints (medrxiv and biorxiv) were searched. The full text of the selected articles was read in detail, and the evidence was corroborated in this review. A preliminary search identified 45 eligible studies, of which 12 were excluded during the screening of title and abstract based on the eligibility criteria. Duplicates were removed using the reference manager (Endnote), and two reviewers examined the remaining 33 articles in full length based on the eligibility criteria. In cases of disagreement, a third reviewer resolved the differences through a discussion. Finally, 21 studies were excluded where oral manifestations were not described, and the remaining 12 studies were included where oral lesions in human MPX disease were reported (Figure 1 ).Figure 1 PRISMA flowchart. Figure 1: The relevant data were selected and recorded by two reviewers. The author's name, year of publication, country, mean age, sample size, study design, oral lesions, management of oral lesions, remarks assessed, results, and conclusions were charted for all the included studies. Results The findings of this review were based on the pooled data of 1136 patients (age range: 2 to 52 years) reported in case reports, retrospective observational studies, and prospective cohort studies from different parts of the world, such as the USA, UK, Nigeria, Spain, and Korea (Table 1 ). Only one study included epidemiological reports from multiple (sixteen) countries.7 Table 1 : Oral lesions in human monkeypox disease and their management. Table 1Author; Country Age/ Mean age Sample size; Study design Oral lesions Prevalence of Oral lesions (Percentage) Management of oral lesions Comments Huhn et al.,8 USA Age 6-47 years Median age: 26 years 34 patients; Retrospective observational study Mouth sores and dysphagia (6/34) 17.6% Not mentioned Patients with nausea and/or vomiting and mouth sores had a longer hospitalization duration and pronounced lymphadenopathy. Yinka-Ogunlete et al.,14 Nigeria 11-year-old boy 1 patient Case report Oral mucosal lesions and ulcers Case report Not mentioned Generalized lymphadenopathy. Vaughan et al.,13 UK Middle-aged men 2 patients Case reports Progressive crops of vesicles and lesions on the mucosal surfaces of the mouth (1/2). 50% Not mentioned Not initially suspected of monkeypox (MPX) infection because the first lesions appeared in the groin. Lymphadenopathy present. Yinka-Ogunleye et al.,22 Nigeria Age 2 days -50 years. Median age: 29 years 122 patients Clinical and epidemiological report 58% of cases had a sore throat. 58% Not mentioned 84/122 cases were male. Face was most commonly involved. All 122 cases had vesiculopustular rash. Thornhill et al.7 16 countries, 943 sites Mean age: 38 years 528 patients Observational study Lesions with prodrome occurred in 17/30 cases. Isolated oral lesions in 13/30 cases. Within seven days, ulcers on the tongue, corner of the mouth, and perioral umbilicated and vesicular lesions appeared. Pharyngitis, odynophagia, epiglottitis, and tonsillar lesions. 43.33% Not mentioned specifically. antiviral treatment, hospitalization for management of severe anorectal pain, and tissue superinfection. Oropharyngeal symptoms were reported as the initial symptoms in 26/528 cases. Transmission is suspected through sexual activity in 95% of the infected persons. Generalized lymphadenopathy Patel et al.15 UK Age 12-67 years Median age: 38 years 197 patients Descriptive Case-series 27/197 patients had oral lesions 13.70% No specific management for oral lesions. All 197 participants were men. 196 were gay or bisexual. Tonsillar signs were seen in 9/197 cases, and sore throat in 33/197 cases. Lymphadenopathy in most cases. Girometti et al.21 UK Median age: 41 years (IQR 34–45). 54 patients Observational study 4/54 had an oropharyngeal lesion 7.40% No specific management for oral lesions. Admission to hospital, mainly due to pain or localized bacterial cellulitis requiring antibiotic intervention or analgesia. Most cases in men who have sex with men (MSM). High proportion of concomitant sexually transmitted infections (STIs). Local inoculation occurs during close skin-to-skin or mucosal contact. Matias et al.18 USA Two men in their 20s, A third man in his 40s. 3 patients Case reports Painful, pruritic vesiculo-pustular oropharyngeal lesions. Pustular lesions on his gingiva. Left tonsillar pain with associated odynophagia (2/3) 66.60% Tecovirimat administered. No prominent side effects of Tecovirimat 600 mg twice daily orally for 14 weeks for systemic lesions. Tarin Vincente et al.20 Spain Median age: 37 years (IQR 31-42) 181 patients. Prospective cohort Study 78/181 cases had lesions in the oral and perioral region 43.09% Not mentioned 166/181 cases were MSM. 32/181 had smallpox vaccination. 73/181 cases had HIV. Jang et al. 19 Korea 34-year-old 1 patient Case report Perioral erosive lesions covered with black crusts were found on his face Case report Not mentioned Swabs and crusts were obtained from the perioral erosion after crust removal. Lymphadenopathy present. Ajmera et al.17 USA 26-year-old 1 patient Case report Oral lesions, tongue swelling, burning sensation, multiple umbilicated pox-like perioral and oral lesions and oral candidiasis. Case report Magic mouth wash, IV fluconazole for oral thrush, and an IM dose of penicillin. Past medical history of syphilis. Tenofovir/emtricitabine for HIV pre-exposure prophylaxis (PrEP). Rashes on his tongue and perioral region. Peiro-Mestres et al. 16 Spain Age: 32-52 years Median age: 38.5 years 12 patients Case- series Oral and pharyngeal lesions (2/12) 16.60% Not mentioned Young adult MSM with a previous history of STIs. MPX DNA was detected in saliva samples collected between 4–16 days after the onset of symptoms. Abbreviations: MPX: Monkeypox, IV: intravenous; IM: intramuscular Oral lesions included mouth sores,8 , 14 oral mucosal lesions,7 , 13, 14, 15, 16 ulcers on the tongue,7 tongue swelling,17 epiglottitis,7 pustular lesions on the gingiva,18 perioral erosive lesions,7 , 17 , 19 , 20 oral candidiasis,17 and pharyngeal,16 oropharyngeal,7 , 18 , 21 and tonsillar lesions 7 , 18 (Table 2 ). Generalized lymphadenopathy was a common feature,7 , 8 , 13, 14, 15 while sore throat was also reported in a few cases.15 , 17 , 19 , 22 Table 2 Oral manifestations in human monkeypox disease. Table 2:Oral manifestations in patients with monkeypox disease References Oral ulcers Huhn et al.8, Girometti et al.48, Peiro-Mestres et al.16 Oral mucosal lesions Thornhill et al.49, Huhn et al.8, Yinka-Ogunlete et al.14 Patel et al.15, Ajmera et al.50, Tarin Vincente et al.20 Progressive crops of vesicles intraorally Vaughan et al.,13 Ulcers on the tongue Thornhill et al.49, Ajmera et al.50 Epiglottitis, and tonsillar lesions Thornhill et al.49, Matias et al.18 Pustular lesions on gingiva Matias et al.18 Oral thrush Ajmera et al.50 Perioral lesions Tarin Vincente et al.20, Thornhill et al.49 Perioral erosive lesions Ajmera et al.50, Jang et al.19 Pharyngeal lesions Thornhill et al.49, Huhn et al.8, Girometti et al.48, Peiro-Mestres et al.16, Matias et al.18 The specific management of these lesions has not been mentioned in most reports. Tonsillar edema and odynophagia improved slowly after initiation of tecovirimat for systemic lesions, and the patient recovered after five days.18 The Magic mouthwash, intravenous fluconazole for oral candidiasis, and an intramuscular dose of penicillin were administered along with supportive care.17 Discussion Oral lesions Various presentations of oral lesions have been reported in several cases of human MPX infections (Figures 2 and 3). The oral mucosa may present with lesions that transform from vesicles to pustules, including umbilication and crusting, within 1 to 4 weeks.7 , 14 , 17 , 23 These lesions may appear in the oral cavity and then follow the skin around the extremities in a centrifugal pattern, along with fever and lymphadenopathy.24, 25, 26 Few patients may have oropharyngeal symptoms as the initial symptoms.7 The oral lesions (enanthem) have detectable viral DNA particles in the oral and pharyngeal passages, even in the absence of skin lesions in the prodromal stage.27 , 28 Experiments in the infected animal model also found the highest concentration of virus in oral secretions29 irrespective of the route of infection (intradermally or intranasally).Figure 2 (a) Perioral lesions in a 26-year-old male patient who tested positive for monkeypox PCR test. Arrow depicts umbilicated pox-like lesions of monkeypox disease (Photo credit: Kunal M. Ajmera et al. 2022; Courtesy: Elsevier and Copyright Clearance Center). PCR, polymerase chain reaction. Figure 2 (b): Oral candidiasis and umbilicated lesions on the tongue in the same 26-year-old male patient. (Photo credit: Kunal M. Ajmera et al. 2022; Courtesy: Elsevier and Copyright Clearance Center). Figure 2 Patients with nausea and/or vomiting and mouth sores may have a longer duration of hospitalization.8 The oral lesions on the oral epithelium, tongue, and labial mucosa have shown epithelial hyperplasia, intracellular edema (ballooning degeneration), necrosis, ulceration, scattered and poorly defined eosinophilic intracytoplasmic inclusion bodies.29 The stratified squamous epithelium of the tongue also showed multinucleated syncytial cells.30 Orthopoxvirus antigens were found only in the skin and oral cavity during the immunohistochemistry evaluation.29 Although oral lesions are present in these studies, it can be a manifestation of comorbidities such as HIV infection.31 Mode of transmission The exact mode of MPXV transmission to humans has not yet been confirmed. It can either be transmitted from animals to humans through direct or indirect contact with MPX-infected animals or their body fluids8 , 12 or from human-to-human transmission.32 , 33 The virus enters the body via damaged skin, the respiratory tract, or the mucous membranes of the eyes, nose, or oral cavity; large respiratory droplets; or direct or indirect contact with body fluids, lesion material, and contaminated surfaces, clothing, or linens. Local inoculation occurs during close skin-to-skin or mucosal contact.21 Disease progression in a prairie dog MPXV model suggests that virus shedding from the oral cavity begins before the onset of lesions.34 Dentists should be vigilant when examining a suspected case of MPX because the primary lesions often originate in the oropharynx before manifesting on the skin.3 In some cases, oral samples,26 including saliva,16 have been found to have the highest load of viral particles, and viral shedding could be detected in oropharyngeal secretions26 before the development of generalized skin lesions. Viable MPX virus is seen in oral samples, including saliva,16 from days 9 to 18.26 This is similar to the oral lesions seen in human orthopoxvirus disease26 and increases the risk of MPX infection in dentists. Similarly, in some cases, perioral lesions extending to the face can also5 , 22 increase the risk of transmission. Severe pharyngitis in these patients might hinder the oral intake of food7, which could be of particular significance in children and patients with comorbidities such as diabetes mellitus. A possible gender effect may be suggested based on the findings of the marmoset model of MPX. This experiment found more lesions and a lower viral burden (viremia and oral shedding) in women than in men.35 However, in humans, oral lesions are mostly in men15 , 20, 21, 22 in non-endemic regions such as the UK and Spain. A noteworthy finding was that most infected persons were gay or bisexual men; hence, transmission was suspected to occur through sexual activity7 , 15 , 16 , 21. Many had a history of sexually transmitted infections (STIs).16 A report from 16 countries found that many MPX-infected men had human immunodeficiency virus infection7 , 16 , 20 , 21. However, these well-controlled HIV patients were on ART, and a very low HIV viral load (less than 50 copies per milliliter) was seen in most patients.7 There is also a change in the pattern of presentation of the clinical features of MPX infection.15 Travel history13, history of contact with an individual with an MPX-like rash at a gathering13, and consumption of bush meat13 have also been reported in patients with MPX disease. Management of oral lesions of MPX disease Most reports included in this review did not mention the specific management of oral lesions. Although it is self-limiting, prompt management is essential. Symptomatic treatment is the mainstay of therapy, especially during the prodromal phase27 , 36 , 37. The interim rapid response guidance released on June 10, 2022, by the WHO recommends rinsing the mouth with salt water at least four times a day in case of oral lesions. Oral lesions can cause secondary infections and facilitate person-to-person viral transmissions. A clean, moist microenvironment can mitigate transmission potential by covering infectious sores and promoting re-epithelialization of damaged exanthem. Using an oral antiseptic solution, such as chlorhexidine mouthwash, can keep the lesions clean. Alternatively, cleansing the ulcers with a dilute povidone-iodine solution will help maintain a hygienic microenvironment.27 Application of local anesthetics such as viscous lidocaine will provide symptomatic relief.38 Oral and topical analgesics or acetaminophen can be administered to control the pain in oral mucosal lesions. Patients with secondary bacterial infections should be treated with appropriate antibiotics.27 The use of non-steroidal anti-inflammatory drugs (NSAIDs) for pain relief should be limited owing to the concern of developing hemorrhagic lesions. Mucosal lesions of the mouth can be painful and warrant the use of appropriate analgesics.27 "Magic mouthwash” along with intravenous fluconazole for oral candidiasis and an intramuscular dose of penicillin demonstrated symptomatic relief in a patient.17 Oral acyclovir may be prescribed at a prophylactic dose of 50 mg/kg given twice daily.39 Many oropharyngeal lesions, such as tonsillar edema and odynophagia, can be relieved within five days when tecovirimat is administered for skin lesions.18 Based on the evidence from the previous epidemics, antivirals, smallpox vaccine, and vaccinia immune globulin (VIG) can be effective40 in controlling the outbreak since the MPX and smallpox viruses are related. Cidofovir, brincidofovir, and tecovirimat are antivirals suggested to defend against MPX41 infection and as a post-exposure prophylactic (PEP) agent in exposed immunocompromised individuals who are contraindicated to receive the smallpox vaccine as a PEP measure.42 , 43 Patients with oral lesions should be monitored for dehydration and malnutrition, and should maintain good nutrition and hydration. Complications of illness include low mood and emotional lability, which may expose patients to poor oral health practices.5 Elective surgery should be deferred as long as the patient is infectious. Linens, hospital gowns, towels, and other fabrics should be carefully handled. Precautions in dental clinics Transmission can be prevented in dental care settings by taking standard, contact, and droplet infection control precautions when treating patients with MPX symptoms.3 , 44, 45, 46 The patient should be managed in isolation, and precautions should be taken to minimize the exposure to surrounding individuals, such as covering any exposed skin lesions and placing a surgical mask over the patient's nose and mouth. In patients with probable or confirmed MPX infection, elective dental treatment should be deferred until the patient is no longer infectious.47 Conclusions Ulcers in the oral cavity or oropharynx can be the primary lesions compared with skin lesions in MPX cases. Oral ulcers have been recorded in almost one-quarter of MPX patients. The lesions initially appear as pink macules or papules. Oral ulcers make it difficult for patients to consume food. Perioral blistered and ulcerated lesions have also been reported in several recent cases. The management of oral ulcerations includes symptomatic and supportive care, including the use of mouthwashes. Antiviral treatment and the use of antibiotics are recommended in severe cases as well as in cases with multiple lesions elsewhere in the body. Dental practitioners may be the first to detect the initial symptoms of MPX infection. Therefore, caution should be exercised, particularly when examining patients with fever and lymphadenopathy. Author contribution statement “Conceptualization: B.J. and S.A.; methodology: B.J.; validation: S.A.; data curation: B.J.; writing—original draft preparation: B.J. and S.A; writing—review and editing: S.A. Both authors have read and agreed to the published version of the manuscript. Concept and design; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Software; Supervision; Validation; Visualization; Roles/Writing - original draft; Writing - review & editing. All authors read and approved the final manuscript. Declaration of Competing Interest The authors declare no conflict of interest. Funding This research received no external funding Ethical approval Not applicable Availability of supporting data Not applicable Acknowledgments None ==== Refs References 1 Phoolcharoen W Shanmugaraj B Khorattanakulchai N. Emergence of monkeypox: Another concern amidst COVID-19 crisis Asian Pac J Trop Med 15 2022 193 2 Vivancos R Anderson C Blomquist P Community transmission of monkeypox in the United Kingdom, April to May 2022 Euro Surveill 2022 27 3 Samaranayake L Anil S. The Monkeypox Outbreak and Implications for Dental Practice Int Dent J 72 2022 589 596 35934521 4 Perez Duque M Ribeiro S Martins JV Ongoing monkeypox virus outbreak, Portugal, 29 April to 23 May 2022 Euro Surveill 2022 27 5 Adler H Gould S Hine P Clinical features and management of human monkeypox: a retrospective observational study in the UK Lancet Infect Dis 22 2022 1153 1162 35623380 6 Anderson MG Frenkel LD Homann S Guffey J. A case of severe monkeypox virus disease in an American child: emerging infections and changing professional values Pediatr Infect Dis J 22 2003 1093 1096 discussion 1096-1098 14688573 7 Thornhill JP Barkati S Walmsley S Monkeypox Virus Infection in Humans across 16 Countries - April-June 2022 N Engl J Med 387 2022 679 691 35866746 8 Huhn GD Bauer AM Yorita K Clinical characteristics of human monkeypox, and risk factors for severe disease Clin Infect Dis 41 2005 1742 1751 16288398 9 Sookaromdee P Wiwanitkit V. Mouth sores and monkeypox: A consideration J Stomatol Oral Maxillofac Surg 123 2022 593 594 35760308 10 Sukhdeo SS Aldhaheri K Lam PW Walmsley S. A case of human monkeypox in Canada CMAJ 194 2022 E1031 E1035 35793837 11 Ng OT Lee V Marimuthu K A case of imported Monkeypox in Singapore Lancet Infect Dis 19 2019 1166 12 Petersen E Abubakar I Ihekweazu C Monkeypox - Enhancing public health preparedness for an emerging lethal human zoonotic epidemic threat in the wake of the smallpox post-eradication era Int J Infect Dis 78 2019 78 84 30453097 13 Vaughan A Aarons E Astbury J Two cases of monkeypox imported to the United Kingdom, September 2018 Euro Surveill 2018 23 14 Yinka-Ogunleye A Aruna O Ogoina D Reemergence of Human Monkeypox in Nigeria, 2017 Emerg Infect Dis 24 2018 1149 1151 29619921 15 Patel A Bilinska J Tam JCH Clinical features and novel presentations of human monkeypox in a central London centre during the 2022 outbreak: descriptive case series BMJ 378 2022 e072410 16 Peiro-Mestres A Fuertes I Camprubi-Ferrer D Frequent detection of monkeypox virus DNA in saliva, semen, and other clinical samples from 12 patients, Barcelona, Spain, May to June 2022 Euro Surveill 2022 27 17 Ajmera KM Goyal L Pandit T Pandit R. Monkeypox - An emerging pandemic IDCases 29 2022 e01587 35938150 18 Matias WR Koshy JM Nagami EH Tecovirimat for the treatment of human monkeypox: an initial series from Massachusetts, United States Open Forum Infectious Diseases 2022 19 Jang YR Lee M Shin H The First Case of Monkeypox in the Republic of Korea J Korean Med Sci 37 2022 e224 35818706 20 Tarin-Vicente EJ Alemany A Agud-Dios M Clinical presentation and virological assessment of confirmed human monkeypox virus cases in Spain: a prospective observational cohort study Lancet 400 2022 661 669 35952705 21 Girometti N Byrne R Bracchi M Demographic and clinical characteristics of confirmed human monkeypox virus cases in individuals attending a sexual health centre in London, UK: an observational analysis Lancet Infect Dis 22 2022 1321 1328 35785793 22 Yinka-Ogunleye A Aruna O Dalhat M Outbreak of human monkeypox in Nigeria in 2017-18: a clinical and epidemiological report Lancet Infect Dis 19 2019 872 879 31285143 23 Drago F Ciccarese G Merlo G Oral and cutaneous manifestations of viral and bacterial infections: Not only COVID-19 disease Clin Dermatol 39 2021 384 404 34517997 24 Mahmoud A Nchasi G. Monkeypox virus: A zoonosis of concern J Med Virol 2022 25 Thakur S Kelkar D Garg S Why Should RNA Viruses Have All the Fun - Monkeypox, a Close Relative of Smallpox and a DNA Virus J Glob Infect Dis 14 2022 47 49 35910829 26 Hutson CL Olson VA Carroll DS A prairie dog animal model of systemic orthopoxvirus disease using West African and Congo Basin strains of monkeypox virus J Gen Virol 90 2009 323 333 19141441 27 Koenig KL Bey CK Marty AM. Monkeypox 2022 Identify-Isolate-Inform: A 3I Tool for frontline clinicians for a zoonosis with escalating human community transmission One Health 15 2022 100410 28 Reynolds MG Yorita KL Kuehnert MJ Clinical manifestations of human monkeypox influenced by route of infection J Infect Dis 194 2006 773 780 16941343 29 Falendysz EA Lopera JG Doty JB Characterization of Monkeypox virus infection in African rope squirrels (Funisciurus sp.) PLoS Negl Trop Dis 11 2017 e0005809 30 Zaucha GM Jahrling PB Geisbert TW Swearengen JR Hensley L. The pathology of experimental aerosolized monkeypox virus infection in cynomolgus monkeys (Macaca fascicularis) Lab Invest 81 2001 1581 1600 11742030 31 Betancort-Plata C Lopez-Delgado L Jaen-Sanchez N Monkeypox and HIV in the Canary Islands: A Different Pattern in a Mobile Population Trop Med Infect Dis 7 2022 32 Nolen LD Osadebe L Katomba J Introduction of Monkeypox into a Community and Household: Risk Factors and Zoonotic Reservoirs in the Democratic Republic of the Congo Am J Trop Med Hyg 93 2015 410 415 26013374 33 Kalthan E Tenguere J Ndjapou SG Investigation of an outbreak of monkeypox in an area occupied by armed groups, Central African Republic Med Mal Infect 48 2018 263 268 29573840 34 Hutson CL Kondas AV Mauldin MR Pharmacokinetics and Efficacy of a Potential Smallpox Therapeutic, Brincidofovir, in a Lethal Monkeypox Virus Animal Model mSphere 6 2021 35 Mucker EM Wollen-Roberts SE Kimmel A Shamblin J Sampey D Hooper JW. Intranasal monkeypox marmoset model: Prophylactic antibody treatment provides benefit against severe monkeypox virus disease PLoS Negl Trop Dis 12 2018 e0006581 36 Afshar ZM Rostami HN Hosseinzadeh R The reemergence of monkeypox as a new potential health challenge: A critical review Authorea (Preprint) 2022 37 Durski KN McCollum AM Nakazawa Y Emergence of Monkeypox - West and Central Africa, 1970-2017 MMWR Morb Mortal Wkly Rep 67 2018 306 310 29543790 38 France K Villa A. Acute Oral Lesions Dermatol Clin 38 2020 441 450 32892853 39 Hukkanen RR Gillen M Grant R Liggitt HD Kiem HP Kelley ST. Simian varicella virus in pigtailed macaques (Macaca nemestrina): clinical, pathologic, and virologic features Comp Med 59 2009 482 487 19887033 40 Russo AT Berhanu A Bigger CB Co-administration of tecovirimat and ACAM2000 in non-human primates: Effect of tecovirimat treatment on ACAM2000 immunogenicity and efficacy versus lethal monkeypox virus challenge Vaccine 38 2020 644 654 31677948 41 Siegrist EA Sassine J. Antivirals with Activity Against Monkeypox: A Clinically Oriented Review Clin Infect Dis 2022 42 Baker RO Bray M Huggins JW. Potential antiviral therapeutics for smallpox, monkeypox and other orthopoxvirus infections Antiviral Res 57 2003 13 23 12615299 43 Xiao Y Isaacs SN. Therapeutic Vaccines and Antibodies for Treatment of Orthopoxvirus Infections Viruses 2 2010 2381 2403 21197387 44 Cunha BE. Monkeypox in the United States: an occupational health look at the first cases AAOHN J 52 2004 164 168 15119816 45 Tsagkaris C Eleftheriades A Laubscher L Vladyckuk V Papadakis M. Viruses monkeying around with surgical safety: Monkeypox preparedness in surgical settings J Med Virol 2022 46 Samaranayake L. Essential microbiology for dentistry 2018 Elsevier Health Sciences 47 Samaranayake L Anil S. Monkeypox and the dental team. Dental Update 49 2022 683 687 48 Girometti N Byrne R Bracchi M Demographic and clinical characteristics of confirmed human monkeypox virus cases in individuals attending a sexual health centre in London, UK: an observational analysis Lancet Infect Dis 2022 49 Thornhill JP Barkati S Walmsley S Monkeypox Virus Infection in Humans across 16 Countries - April-June 2022 N Engl J Med 2022 50 Ajmera KM Goyal L Pandit T Pandit R. Monkeypox – An emerging pandemic IDCases 29 2022 e01587 35938150	0	PMC9715264	NO-CC CODE	2022-12-14 23:45:35	no	Oral Surg Oral Med Oral Pathol Oral Radiol. 2022 Dec 2; doi: 10.1016/j.oooo.2022.11.012	utf-8	Oral Surg Oral Med Oral Pathol Oral Radiol	2,022	10.1016/j.oooo.2022.11.012	oa_other
==== Front Cell Genom Cell Genom Cell Genomics 2666-979X The Authors. S2666-979X(22)00190-2 10.1016/j.xgen.2022.100232 100232 Article Altered and allele-specific open chromatin landscape reveals epigenetic and genetic regulators of innate immunity in COVID-19 Zhang Bowen 12319 Zhang Zhenhua 124519 Koeken Valerie A.C.M. 12619 Kumar Saumya 12 Aillaud Michelle 7 Tsay Hsin-Chieh 12 Liu Zhaoli 12 Kraft Anke R.M. 128910 Soon Chai Fen 28 Odak Ivan 11 Bošnjak Berislav 11 Vlot Anna 12 Deutsche COVID-19 OMICS Initiative (DeCOI) Swertz Morris A. 45 Ohler Uwe 12 Geffers Robert 13 Illig Thomas 1415 Huehn Jochen 1016 Saliba Antoine-Emmanuel 17 Sander Leif Erik 1418 Förster Reinhold 91011 Xu Cheng-Jian 1268 Cornberg Markus 128910 Schulte Leon N. 714 Li Yang 1261020∗ 1 Department of Computational Biology for Individualised Infection Medicine, Centre for Individualised Infection Medicine (CiiM), a joint venture between the Helmholtz-Centre for Infection Research (HZI) and the Hannover Medical School (MHH), Hannover, Germany 2 TWINCORE, a joint venture between the Helmholtz-Centre for Infection Research (HZI) and the Hannover Medical School (MHH), Hannover, Germany 3 Beijing Normal University, College of Life Sciences, Beijing, China 4 Genomics Coordination Center, University of Groningen and University Medical Center Groningen, Groningen, the Netherlands 5 Department of Genetics, University of Groningen and University Medical Center Groningen, Groningen, the Netherlands 6 Department of Internal Medicine and Radboud Center for Infectious Diseases (RCI), Radboud University Medical Center, Nijmegen, the Netherlands 7 Institute for Lung Research, Philipps University, Marburg, Germany 8 Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany 9 German Centre for Infection Research (Deutsches Zentrum für Infektionsforschung [DZIF]), Partner Site Hannover-Braunschweig, Hannover, Germany 10 Cluster of Excellence Resolving Infection Susceptibility (RESIST; EXC 2155), Hannover Medical School, Hannover, Germany 11 Institute of Immunology, Hannover Medical School, Hannover, Germany 12 Berlin Institute for Medical Systems Biology (BIMSB), Max Delbrück Center for Molecular Medicine in the Helmholtz Association (MDC), Berlin, Germany 13 Genome Analytics, Helmholtz-Center for Infection Research (HZI), Braunschweig, Germany 14 German Center for Lung Research (DZL), Giessen, Germany 15 Hannover Unified Biobank, Hannover Medical School, Hannover, Germany 16 Department of Experimental Immunology, Helmholtz Centre for Infection Research, Braunschweig, Germany 17 Helmholtz Institute for RNA-based Infection Research (HIRI), Helmholtz-Centre for Infection Research (HZI), Wurzburg, Germany 18 Charité–Universitätsmedizin Berlin, Department of Infectious Diseases and Respiratory Medicine, Charité, Universitätsmedizin Berlin, Berlin, Germany ∗ Corresponding author 19 These authors contributed equally 20 Lead contact 2 12 2022 2 12 2022 10023231 5 2022 21 10 2022 17 11 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. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection causes severe COVID-19 in some patients and mild COVID-19 in others. Dysfunctional innate immune responses have been identified to contribute to COVID-19 severity, but the key regulators are still unknown. Here, we present an integrative single-cell multi-omics analysis of peripheral blood mononuclear cells from hospitalized and convalescent COVID-19 patients. In classical monocytes, we identified genes that were potentially regulated by differential chromatin accessibility. Then, sub-clustering and motif-enrichment analyses revealed disease condition-specific regulation by transcription factors and their targets, including an interaction between C/EBPs and a long-noncoding RNA LUCAT1, which we validated through loss-of-function experiments. Finally, we investigated genetic risk variants that exhibit allele-specific open chromatin (ASoC) in COVID-19 patients and identified a SNP rs6800484-C, which is associated with lower expression of CCR2 and may contribute to higher viral loads and higher risk of COVID-19 hospitalization. Altogether, our study highlights the diverse genetic and epigenetic regulators that contribute to COVID-19. Graphical abstract With a single-cell multi-omics study, Zhang et al. linked both genetic and epigenetic regulation to transcriptional responses in COVID-19 patients. Their findings suggest the identification of underlying regulators of innate immune dysfunction in COVID-19 patients and the regulatory roles of known genetic risk loci in the COVID-19 pathogenesis. Keywords COVID-19 genetic regulation epigenetic regulation scATAC-seq scRNA-seq allele-specific open chromatin Published: December 1, 2022 ==== Body pmcIntroduction COVID-19 is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2),1 and clinical symptoms of patients with SARS-CoV-2 infection range from asymptomatic to severe pneumonia and acute respiratory distress syndrome.2 Although vaccines reduce the risk of major illness and mortality, the molecular mechanisms underlying the heterogeneous outcome in disease presentation remain unclear.3 A number of studies have examined the complex interplay between peripheral blood leukocytes in COVID-19 and linked immune activation and specific cell subsets to disease severity.4 , 5 The adaptive immune system is clearly linked to disease presentation, because prominent lymphopenia is a hallmark of severe disease.6 Alterations in T cell function have also been observed, with T cells from severe patients showing increased signs of migration to inflamed site and apoptosis7 and excessive or suboptimal CD4+ and CD8+ T cell responses detected in severe disease.8 The innate immune system has also been reported to be dysregulated in severe disease, which is characterized by high neutrophil counts,9 likely contributing to tissue damage and hyperinflammation, dysfunctional monocytes with low expression of HLA-DR and interferon (IFN)-stimulated genes (ISGs),4 and functionally impaired NK cells.10 Additionally, long noncoding RNAs are reported to be involved in the regulation of antiviral immune responses in COVID-19 and subsequent disease states.11 These studies have shed light on the detrimental immune responses that contribute to immunopathology in severe COVID-19. In addition to studies of molecular signatures, several genome-wide association studies (GWASs) of severe COVID-19 have been performed.12 , 13 , 14 These revealed the impact of genetic variations on disease severity, improving our understanding of COVID-19 pathology. Moreover, although an epigenetic study on individuals convalescing from COVID-19 revealed remodeling of the chromatin accessibility landscape that established immunological memory,15 a recent study shows the immune responses and cytokine production capacity generally recover without major sequelae after COVID-19.16 More importantly, because the majority of these risk factors were identified in noncoding regions, they are predicted to have functional effects on gene expression via transcription factor (TF) binding and interaction with regulatory elements.17 These regulatory effects are highly cell type specific18 , 19 and are not yet understood in relation to COVID-19 risk factors. Bridging the existing gaps requires an integrative approach that connects genetic variations, epigenetic factors, and immune responses at the cellular level.20 For this reason, we captured both the transcriptome and epigenome of individual peripheral blood mononuclear cells (PBMCs), as well as genome-wide genotypes, from hospitalized and convalescent COVID-19 samples. We identified C/EBP-motif-enriched open chromatin profiles in classical monocytes and illustrated their interaction with the immune-regulatory LUCAT1 RNA locus using single-cell omics and loss-of-function experiments. Additionally, we demonstrate that COVID-19 GWAS risk variants contribute to the disease by regulating chromatin accessibility through allele-specific open chromatin (ASoC) effects. Our ASoC analysis reveals that the COVID-19 GWAS risk SNP rs6800484 is associated with the expression of CCR2 via chromatin accessibility of an enhancer in monocytes. Together, these data indicate that altered chromatin accessibility and ASoC both result in impaired epigenetic regulation that contributes to COVID-19 pathogenesis, while the complex co-action of these factors could lead to heterogeneous and individualized disease outcomes. Our study further provides a broad resource for exploring cell-type-specific genetic and epigenetic regulatory effects that contribute to COVID-19. Results Study overview and patient population Using single-cell RNA sequencing (scRNA-seq), single-cell assay for transposase-accessible chromatin using sequencing (scATAC-seq), and genotype array, we examined the transcriptomics and epigenomics of PBMCs, as well as individual genotypes, across 46 hospitalized COVID-19 and 32 convalescent samples from 48 individuals, including 20 individuals from whom we have samples from multiple time points (Figures 1A, 1B, and S1). Hospitalized COVID-19 patients were further allocated to mild or severe patient categories using World Health Organization (WHO) scores (severe: 5–7, mild: 3–4). Clinical characteristics of all study participants are summarized in Table S1.Figure 1 Study overview and single-cell multi-omics (A) Workflow of the study. Sample numbers in each data layer and disease condition are indicated. (B) Schematic overview of all patients enrolled in the study. Sampling dataset, disease conditions, and convalescent days are indicated. (C and D) UMAP showing the cell distribution of hospitalized and convalescent conditions in scRNA-seq (C) and in scATAC-seq (D); see also Figure S3 and Table S2 for annotation details. (E) Boxplots showing cell proportion of hospitalized and convalescent samples in main cell types of scRNA-seq and scATAC-seq. (F) Scatterplots showing the log-fold-change (log2FC) of DE-Gs identified in monocytes between the comparison of severe versus convalescent and the comparison of mild versus convalescent. More details on DE-Gs can be found in Figure S4 and Table S3. (G) Scatterplots showing the log2FC of DE-Gs and log2FC of differentially accessible peaks (DAPs) identified in classical monocytes between comparison of severe versus convalescent and comparison of mild versus convalescent. Details of the matched DE-Gs and DAPs can be found in Table S4. See also Figure S1 and Table S1. In total, after quality control, we obtained scRNA-seq data for 165,054 cells from 64 samples (n = 37 active, n = 27 convalescent samples) taken from 41 individuals and scATAC-seq for 46,690 cells from 49 samples (n = 25 hospitalized, n = 24 convalescent samples) taken from 39 individuals (Figure S2). We characterized these cells with unsupervised clustering and, based on the marker genes or gene activity scores in each cluster (Table S2), identified 10 major cell types in the scRNA-seq dataset and 8 major cell types in the scATAC-seq dataset (Figures 1C, 1D, and S3). The relative percentage of cell types in the PBMC fractions of each sample reveals a higher abundance of classical monocytes and a lower abundance of non-classical monocytes, as well as CD4+ and CD8+ T cells, in hospitalized COVID-19 compared with convalescent patients in both datasets (Figures 1E and S4A, Dirichlet regression test, false discovery rate [FDR] adjusted p < 0.05), in line with a recent publication.21 Additionally, a high proportion of CD163+ classical monocytes was found in five hospitalized COVID-19 patients (four severe and one mild) exclusively in the scRNA-seq dataset (Figures 1C and 1E). Severe and mild COVID-19 patients show different magnitudes of transcriptional responses Differential expression (DE) tests per cell type between hospitalized and convalescent samples revealed that a large number of differentially expressed genes (DE-Gs) were in NK cells, classical monocytes, and non-classical monocytes (Table S3; Figure S4B), suggesting that these cell types respond most prominently during COVID-19. Within these cell types, a large proportion of DE-Gs were shared between the mild versus convalescent and the severe versus convalescent comparison, especially for classical monocytes (Figure 1F). This suggests that similar transcriptional changes occur in mild and severe COVID-19, and that the difference between mild and severe COVID-19 is due to a difference in the magnitude of the response, rather than different transcriptional programs. For example, we observed several ISGs, such as IFI6, IFI27, IFI30, and IFI44L, to be significantly upregulated in both severe and mild samples compared with convalescent samples, whereas a significantly higher expression of IFI27 and IFITM3 was detected in mild patients compared with severe patients, reminiscent of the results of a previous study.4 Gene-enrichment analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database showed a clear upregulation of oxidative phosphorylation, across different immune cells, in both mild and severe COVID-19, as well as an enrichment of immune-related pathways, such as antigen processing and presentation and phagosome (Figure S4C). Differential open chromatin accessibility contributes to transcriptional differences between hospitalized and convalescent COVID-19 patients To reveal epigenetic alterations at the level of chromatin accessibility in COVID-19, we explored open chromatin signatures of PBMCs across the different disease conditions. Among all 49 samples, 15% of the 157,330 reproducible peaks are in promoter regions, while 32% and 45% are located in intergenic and intronic regions, respectively. When comparing across cell types and conditions, we observed no general enrichment of cell-type- or condition-specific peaks in the promoter or enhancer region (Figure S5A). We noticed that among all cell types, the number of open chromatin peaks is highest in classical monocytes, where it is significantly higher in samples from hospitalized than from convalescent COVID-19 patients (chi-square test, p < 2.22 × 10−16). When comparing peaks from one cell type with all other cell types, a large number of cell-type-specific peaks were identified. This includes 17,105 peaks specific for classical and/or non-classical monocytes (FDR-adjusted p < 0.05 and log-fold-change [log2FC] > 0.5, compared with other cell types), which comprise 10.9% of all peaks in our data. Next, we investigated the condition-specific peaks by comparing open chromatin peaks between disease conditions (hospitalized versus convalescent, mild versus convalescent, and severe versus convalescent) within each cell type. The lack of genome-wide significant condition-specific peaks (FDR-adjusted p < 0.05) suggests that cell types contribute more than disease conditions to variation in open chromatin accessibility. To test the regulatory impact of open chromatin marks on transcriptional responses, we integrated the significant DE-Gs (Bonferroni-corrected p < 0.05) described above with the nominal differential peaks (p < 0.05) through peak-to-gene linkages, i.e., correlating the gene expression from scRNA-seq and peak accessibility from scATAC-seq (Table S4; for details, see STAR Methods). Interestingly, we observed an enrichment of open chromatin peaks that associated DE-Gs in classical monocytes in both severe and mild patients (Fisher exact test, FDR-adjusted p = 6.03 × 10−4 and 4.08 × 10−4, respectively), which includes in total 977 out of 2,367 (41.3%) genes that are upregulated in either severe or mild comparing with convalescent patients (Figures 1G and S5B). These results illustrate that there is a large overlap of changes in chromatin accessibility and gene expression in the monocyte compartment during COVID-19, suggesting the underlying epigenetic regulation on transcriptional responses. Motif enrichment reveals different transcriptional regulation between hospitalized and convalescent COVID-19 in classical monocytes To further characterize the epigenetic regulation of gene expression in COVID-19, we performed TF motif-enrichment analysis for the open chromatin peaks identified in each cell type and condition. In total, we found 60 TFs with significantly enriched motifs among the identified peaks. This included SPI1 (PU.1), JUN/FOS, and C/EBP motifs, which were enriched in classical monocytes in both hospitalized and convalescent COVID-19 patients. Of note, C/EBP motifs (CEBPA, CEBPB, CEBPD, CEBPE, and CEBPG) are even more significantly enriched in hospitalized patients than in convalescent patients (Figure 2A). Given their important role in monocyte differentiation and pro-inflammatory activation,22 , 23 we further investigated the interaction between these TFs and their targets, which were identified based on the genes with motif-binding peaks in classical monocytes. In total, 4,681 genes were associated with peaks harboring either SPI1, JUN/FOS, or C/EBP motifs, of which 1,514 were also DE-Gs between hospitalized and convalescent COVID-19 in classical monocytes (Table S5).Figure 2 TF regulation via motifs in the open chromatin peaks (A) Heatmap showing the significantly enriched TF motifs in the open chromatin peaks of each cell type and condition. Colors represent −log10 p value of enrichment. Rows are significantly enriched TF motifs. (B) Track plots showing the peaks around the LUCAT1 gene. Blue lines indicate inferred linkages between peaks and LUCAT1 expression. (C) Heatmap showing the expression correlation of LUCAT1, IFI27, and IFI30 and TF genes in classical monocytes of hospitalized patients with a dot plot showing the expression of these genes in different conditions. (D) Boxplots showing the molecular responses after knockdown of SPI1 and LUCAT1 or inhibition of C/EBP proteins. (E) Schematic plot summarizing the potential regulating program in LUCAT1, SPI1, and C/EBP, as well as ISG, and COVID-19 severity. See also Figure S5 and Table S5. Interestingly, we found that the long noncoding RNA LUCAT1 was associated with monocyte-specific accessible peaks harboring SPI1, JUN/FOS, and C/EBP motifs (Figure 2B), suggesting a monocyte-specific influence of the SPI1, JUN/FOS, and C/EBP TFs on LUCAT1. Because LUCAT1 has previously been reported as a negative regulator of IFN responses,24 we determined the co-expression patterns of LUCAT1, the TFs, and the two highly expressed ISGs, IFI27 and IFI30, in classical monocytes. As shown in Figure 2C, the expression of LUCAT1 is positively correlated with the expression of SPI1, JUN/FOS, and CEBPD/CEBPE TFs but negatively correlated with IFI27 and IFI30 expression in active COVID-19 patients. A similar co-expression correlation was observed in convalescent individuals, although the correlations between LUCAT1 and CEBPE or IFI30 were no longer significant (Figure S5C). Furthermore, in the DE comparison between disease conditions, SPI1 and LUCAT1 showed significantly higher expression in severe and mild samples compared with convalescent samples in classical monocytes (Bonferroni-corrected p < 0.05; Table S3; Figure 2C), whereas CEBPD showed significantly higher expression in severe samples compared with both mild and convalescent ones (Bonferroni-corrected p = 3.64 × 10−10 [severe versus mild] and 4.15 × 10−43 [severe versus convalescent]). IFI27 and IFI30, as mentioned above, were significantly upregulated in mild samples compared with severe samples (Bonferroni-corrected p = 9.23 × 10−71 [IFI27] and 3.00 × 10−15 [IFI30]; Table S3; Figure 2C). Together, these findings indicate a complex interaction of these genes at the expression level through epigenetic regulation that results in their altered expression under different disease conditions. To validate the interactions between LUCAT1 and C/EBPs in monocytes of COVID-19 patients, we measured the expression of LUCAT1 in isolated monocytes after inhibiting C/EBP using celastrol and betulinic acid (Figure 2D). In unstimulated monocytes, the C/EBP inhibitors enhanced LUCAT1 expression at lower doses but suppressed LUCAT1 expression at higher doses. In monocytes activated with a cocktail of interleukin (IL)-1α, IFNα, and 3p-hairpin RNA (viral mimic), both inhibitors suppressed LUCAT1 expression. In addition, in a stable LUCAT1 knockdown monocyte cell line, we observed increased CEBPE expression (Figure 2D), which indicates strong negative feedback of LUCAT1 to upstream regulatory C/EBP TFs. Because LUCAT1 was previously reported to suppress inflammatory and ISGs,24 and ISGs are suppressed in severe COVID-19 patients,4 we speculate that the interaction of LUCAT1, SPI1, and C/EBPs inhibits ISG responses, resulting in a more severe condition in COVID-19 patients (Figure 2E). Single-cell RNA and ATAC profiles revealed altered C/EBP regulation in a monocyte subset associated with oxygen supply of COVID-19 patients To further investigate the heterogeneity of gene regulation in the monocyte compartment of COVID-19 patients, we explored disease condition-specific subsets. Subsampling the monocytes and sub-clustering them revealed eight transcriptionally distinct cell clusters (R1–R8; Figure 3A), from which R3, R4, and R8 were largely contributed by hospitalized COVID-19 patients. Through the DE tests comparing expression of gene between one cluster with the rest of the clusters and visualization of selected marker gene expression by Uniform Manifold Approximation and Projection (UMAP) (Table S6; Figure 3B), we identified R1 as CD14−CD16+ non-classical monocytes and R2–R8 as CD14+CD16− classical monocytes. In the classical monocytes, CEBPD and SPI1 TFs were expressed similarly among all clusters. However, a more distinctive expression pattern was observed for the previously mentioned TF target genes. LUCAT1 was significantly higher expressed in R2 and R8 compared with the other clusters (Bonferroni-corrected p < 2.22 × 10−16; Table S6), while ISGs such as IFI27, IFI30, and IFITM3 showed expression predominantly in the R3 cluster (Bonferroni-corrected p < 2.22 × 10−16) (Figure 3B; Table S6).Figure 3 RNA and ATAC profiles in monocyte sub-clusters in hospitalized and convalescent COVID-19 patients (A) UMAP showing the cell distribution of hospitalized and convalescent conditions in monocyte sub-clusters of scRNA-seq. (B) Expression of marker genes in monocyte sub-clusters. See also Table S6 for all the markers. (C) Violin plots showing the AUCell-based gene signature scores for each sub-cluster from HLA-DRloS100Ahi monocytes in PBMCs (Schulte-Schrepping et al.4) and infiltrating monocytes (FCN1-Mono) in bronchoalveolar lavage (BAL) fluid (Wendisch et al.25). (D) UMAP showing the monocyte sub-clusters of scATAC-seq. (E) Dot plots and heatmap showing the expression and imputed activity scores of shared marker genes identified in monocyte sub-clusters of scRNA-seq and scATAC-seq, respectively. (F) Boxplots showing the cell proportion of severe, mild, and convalescent patients, as well as oxygen supply needed, not needed, and convalescent patients in each monocytes sub-cluster of scRNA-seq. See the cell distributions in Figure S8. (G) Heatmap showing the significantly enriched TF motifs in the open chromatin peaks of each monocytes sub-cluster; TFs that were also enriched as regulon in R4 cluster cells by SCENIC (single-cell regulatory network inference and clustering) are marked with red asterisks. See also Figures S6–S8 and Table S7. To validate our findings, we compared our monocyte sub-clusters with the previously published transcriptional markers of monocytes from COVID-19 patients.4 By applying the AUCell scores based on the top 30 marker genes from the PBMC datasets, we confirmed that R1 is non-classical monocytes, whereas R2–R5 should be classical monocytes (Figure S6A). Additionally, we identified that the hospitalized COVID-19-specific R4 and R8 sub-clusters were similar to HLA-DRloS100Ahi monocytes (Figure 3C), which are previously found as dysfunctional CD14+ monocytes in severe COVID-19 patients. Moreover, the ISG-predominated R3 sub-cluster was also found similar to another reported severe COVID-19 patient-specific HLA-DRloCD163hi monocyte (Figure S6A).4 In response to SARS-CoV-2 infection, circulating monocytes could be recruited to the lung tissue and participate in tissue immune responses by further differentiating into macrophages.26 We therefore assessed the transcriptional similarity between the monocyte sub-clusters and monocytes/macrophages reported in bronchoalveolar lavage (BAL) fluid samples from COVID-19 patients.25 By applying AUCell scores again to the BAL dataset markers, we identified that the hospitalized COVID-19-specific R3 and R4 were similar to FCN1-Mono in BAL (Figure 3C), which was reported25 as infiltrating monocytes that would later differentiate toward macrophages, whereas the macrophages themselves were not identified in any of our clusters (Figure S6B). These together suggest that both R3 and R4 sub-clusters were associated with COVID-19 severity and play an important role in the immune responses by infiltrating to the patients’ lungs. Next, we performed sub-clustering on the monocytes from the scATAC-seq data. Unsupervised clustering revealed six epigenetically distinct cell clusters (C1–C6; Figures 3D, S7A, and S7B). Of these, the C4 cluster is specific to hospitalized COVID-19 patients, and the C2 cluster is specific to convalescent patients. Through a multi-omics alignment of the transcriptomic and epigenomic profiles across monocyte sub-clusters (see STAR Methods for details), we confidently matched C1 to R1 as non-classical monocytes, as well as C2 to R2 and C4 to R4 as classical monocytes (with >90% of aligned cells matched; Figures S7C and S7D). This can be confirmed by the shared pattern between gene expression levels of marker genes and estimated gene activity scores of the same marker genes based on peak data (Table S6; Figure 3E). When comparing the cell proportions across different disease conditions, the non-classical monocytes (C1/R1) had a higher abundance in convalescent COVID-19 patients (Figures 3F and S8A) that could be seen even before sub-clustering (Figure 1F). More interestingly, the remaining classical monocytes displayed high heterogeneity of cell proportions, with the C2/R2 and C4/R4 sub-clusters varying dramatically across disease conditions. C2/R2, which expresses LUCAT1 and harbors a strong antigen-presentation capacity with high expression of MHC class II components (including HLA-DQA and HLA-DPA), was largely contributed by convalescent patients (Figure 3F). In contrast, the C4/R4 cluster, which is annotated as HLA-DRloS100Ahi monocytes and shows a reverse expression pattern of MHC class II components and suppressed expression of ISGs, is mainly contributed by hospitalized COVID-19 patients (both mild and severe) and has a higher proportion in patients requiring oxygen supply than in those without (Figure 3F), suggesting a potential correlation between these monocytes and impaired lung function in patients. To disclose the epigenetic regulation that underpins the transcriptional differences of these monocyte subsets, especially the condition-specific ones, we performed TF motif enrichment for marker peaks identified in each monocyte subset. The results demonstrate a distinct pattern of enriched motifs in different subsets. SPI1 is enriched in the convalescent-specific R2/C2 subset, together with RUNX1/2, IRF4, STAT2, and BCL11 A/B, whereas C/EBP motifs (CEBPA, CEBPB, CEBPD, CEBPG, and CEBPE) are enriched in the hospitalized patient-specific R4/C4 subset, together with an ATF4 motif (Figure 3G). From the scRNA-seq data, we found that among these TFs, CEBPD, CEBPB, and ATF4 were also widely expressed in R4 cells (Figure S8B). Through an independent TF regulon enrichment analysis in R4 cluster cells,27 we have confirmed that the identified C/EBPs and ATF4 were also high-confidently enriched TFs, together with IRF4, FOS, JUNB, JUND, BACH1, etc. (red asterisks in Figure 3G; see all enriched TFs in Table S7). This result indicates a shift of the regulatory elements between convalescent and hospitalized COVID-19 patients. Additionally, the suppressed expression of IFI27 and IFITM3 in R4 in comparison with R3 (Figure 3B) corresponds to its matched C4 cluster, which was enriched with C/EBP motifs. These results again suggest that altered TF motif accessibility may contribute to the dysregulation of IFN responses in COVID-19, and further indicate a potential correlation with the need for oxygen supply of COVID-19 patients. COVID-19 GWAS variants are overrepresented in open chromatin regions of classical monocytes Previous GWASs have revealed a number of genome regions associated with COVID-19 conditions. Therefore, we tested whether the identified GWAS hospitalization risk variants (“Hospitalized covid vs. population”, release 6)14 have an impact on open chromatin peaks in specific immune cell types. Our data reveal that these variants are significantly enriched in open chromatin peaks of classical monocytes from hospitalized COVID-19 patients (Fisher exact test, p = 2.98 × 10−12) and of CD4+ T cells from convalescent individuals (Fisher exact test, p = 2.68 × 10−6) compared with the other conditions and cell types. In classical monocytes, risk variants on chromosomes (chr) 3, 12, 17, and 21 were found to be located in several open chromatin peaks that were highly accessible in hospitalized patients (Figures 4A, 4B, and S9). When looking at the genes linked to the risk variant peaks mapped using the aforementioned method of peak-to-gene linkage (see STAR Methods), we identified significantly elevated expression (Wilcoxon rank-sum test, Bonferroni-corrected p < 0.05) of CCR1 and CCR2 (chr3), OAS3 (chr12), and IFNAR1 and IFNGR2 (chr21) in hospitalized patients compared with convalescent individuals in classical monocytes (Figure 4C). These results suggest that several GWAS risk variants may impact the expression of linked immune response genes through epigenetic regulation.Figure 4 GWAS risk variants associated with peaks and genes (A) Schematic plot showing the potential regulatory role of a GWAS risk variant located in an open chromatin peak that is bound by the TF motif and associated with gene expression. (B) Heatmap showing chromatin accessibility of peaks detected with hospitalized COVID-19 risk variants. More details can be found in Figure S9. (C) Dot plots showing the expression of DE-Gs associated with peaks from (B). ASoC analysis reveals epigenetic effects of genetic variants To further investigate the epigenetic effects of genetic variants, we evaluated the ASoC, which represents the imbalance of chromatin accessibility between alleles, at heterozygous SNPs by integrating scATAC-seq and SNP data from the same individuals. In total, 292 and 86 ASoC SNPs were identified in hospitalized and convalescent COVID-19 individuals, respectively (FDR-adjusted p < 0.05; Figures 5A and S10A–S10C). Of these identified ASoC SNPs, about 5% were shared by hospitalized and convalescent conditions, which is in contrast with the fact that the majority of heterozygous SNPs (89.18%) available for testing the ASoC effect were shared by participants between conditions. This result suggests there is distinct allele-specific regulation in open chromatin regions between hospitalized and convalescent COVID-19 patients.Figure 5 ASoC analysis reveals the epigenetic effect of COVID-19 GWAS variants (A) Venn diagram of ASoC SNPs identified in six cell types from hospitalized and convalescent participants. ASoC SNPs were merged per disease condition. ASoC SNPs identified in more than one cell type were counted once. (B) Upset plot showing functional annotation of identified ASoC SNPs. Regulatory element annotations were determined based on 25-state models from the Roadmap Epigenomics Project. ASoC SNPs were assigned to eQTL genes and DE genes based on significant variant-gene pairs (GTEx V8) and positions (25 kbp up/downstream of ASoC SNPs), respectively. Numbers at the top of each bar indicate the exact number of ASoC SNPs belonging to the annotation or the gene group. (C) Bar plot showing the enrichment of ASoCs assigned to our DE-Gs. The x axis represents cell types, and the y axis represents odds ratio that ASoCs are assigned to the DE-Gs (i.e., ASoC SNP is located in the promoter of the DEG). Color indicates disease conditions: red for hospitalized COVID-19 and blue for convalescent COVID-19. The numbers on the bar are FDR-adjusted p values and number of ASoC SNPs assigned to DEG out of the number of ASoC SNPs identified for the cell type. (D) Heatmap of correlations between allelic imbalance and TF motif disruption. For each ASoC SNP, the allelic imbalance was represented by log2(reference read counts/alternative read counts), while the motif disruption was the difference between altScore and refScore by motifbreakR. Colors of the heatmap are Spearman’s rho, and multiplication symbols (×) indicate significant correlations (FDR-adjusted p < 0.05). (E) Q-Q plot of COVID-19 GWAS p values for identified ASoC SNPs. The y axis represents observed GWAS p values (converted by −log10) of ASoC SNPs in cMono of hospitalized (red), convalescent COVID-19 (green) participants, and random selected SNPs (blue) with matched minor allele frequency. (F) Allelic reads depth of ASoC SNP rs6800484 at the COVID-19-related CCR locus. (G) Integration of gene-to-peak link, single-cell ATAC-seq, promoter capture Hi-C, eQTL SNPs, and COVID-19 GWAS SNPs around ASoC SNP rs6800484. (H) Schematic plot showing the potential epigenetic and genetic regulating program at CCR2 locus under COVID-19 scenario. (I) CCR2 expression in the differentiation trajectory of monocytes and macrophages of BAL fluid samples of COVID-19 patients (Wendisch et al.25). <NA>, no valid estimation available. See also Figure S10. As shown in Figure 5B, the majority of ASoC SNPs were located in enhancer (>25%) or promoter (>65%) regions, suggesting that epigenetic regulations occur in regulatory DNA sequences. In whole-blood samples, more than 55% of ASoC SNPs were reported to be associated with the expression of nearby genes (expression quantitative trait loci or eQTL),28 which indicates that these ASoC SNPs potentially affect gene expression by controlling allele-specific chromatin accessibility. In addition, in our scRNA-seq analysis for about 10% of ASoC SNPs, the nearby genes (of which promoters overlap with at least one ASoC SNP) were identified to be differentially expressed in at least one cell type (i.e., cell-type-dependent DE-Gs identified by comparisons between conditions). Further enrichment analysis revealed an over-representation of ASoC SNPs assigned to DE-Gs in COVID-19 patients (Fisher exact test, FDR-adjusted p < 0.05; Figure 5C), suggesting that the genetic risk variants have an impact on the transcriptional responses to SARS-CoV-2 infection through allele-specific chromatin accessibilities. Of note, we also observed that the ASoC SNPs were enriched in enhancer regions in hospitalized patients (Fisher exact test, p = 0.047), but not in the convalescent ones, showing the alteration of transcriptional profiles/activities in hospitalized COVID-19 patients compared with convalescent ones. When zooming in on cell subsets, the ASoC SNPs we identified show significant over-representation in open chromatin regions (Figure S10D) and TF binding sites (TFBSs) (Figure S10E). Given that the genetic variants can perturb TF binding affinities by breaking the corresponding TF motifs, resulting in dysregulation of target genes,29 we calculated motif disruption scores (MDSs) for each ASoC SNP.30 We found that the allelic chromatin accessibilities were significantly correlated with MDSs for several TF motifs in classical monocytes from hospitalized COVID-19 patients (Spearman’s rank rho, FDR-adjusted p < 0.05; Figure 5D), suggesting that ASoC SNPs can play regulatory roles by disrupting TF motifs (i.e., affecting TF binding affinities). ASoC of COVID-19 GWAS variants Next, we intersected our ASoC SNPs with the above-mentioned COVID-19 GWAS hospitalization risk variants.14 We found that ASoC SNPs identified in classical monocytes from hospitalized COVID-19 patients were also associated with COVID-19, compared with randomly selected SNPs with matched minor allele frequency (Figures 5E and S10G). Among them, rs6800484 (COVID-19 GWAS p = 6.58 × 10−9) showed an imbalance of chromatin accessibility in classical monocytes in hospitalized COVID-19 patients (binomial test, p < 0.05), but not in convalescent individuals (Figure 5F). Of note, this SNP is located in a classical monocyte-specific open chromatin peak that was annotated as an EnhA1 enhancer (Roadmap Epigenomics Project)31 close to the CCR gene family. This observation led us to further explore this locus by combining our results (ASoC SNPs, scRNA-seq, and scATAC-seq) with publicly available data, including promoter capture Hi-C of monocytes (PCHiCs),32 eQTL of whole-blood samples,33 and COVID-19 GWAS summary statistics.14 As shown in Figure 5G, we illustrated a potential regulatory program showing the effect of this variant underlying the COVID-19 context. Specifically, the publicly available monocyte PCHiC data and our peak-to-gene linkage analysis (see STAR Methods) suggest that the expression of CCR2 is correlated with the regulatory elements pinpointed by rs6800484 in classical monocytes from hospitalized COVID-19 patients. In addition, rs6800484-C is significantly associated with both COVID-19 (p = 6.58 × 10−9) and decreased CCR2 expression (p = 4.29 × 10−13). Meanwhile, in classical monocytes, homozygous risk allele (C/C) carriers show significantly lower CCR2 levels compared with other COVID-19 patients (Wilcoxon test, Bonferroni-corrected p = 3.2 × 10−3; Figure S10H), validating the inhibiting role of the risk allele on CCR2 expression. As summarized in Figure 5H, the COVID-19 risk allele rs6800484-C identified in hospitalized patients is associated with decreased chromatin accessibility of an enhancer at the locus, which further inhibits the CCR2 expression. Of note, a recent study using a mouse-adapted SARS-CoV-2 strain has shown that mice lacking Ccr2 demonstrate higher viral loads and increased lung viral dissemination.34 In addition, our scRNA-seq data of classical monocytes confirms the importance of CCR2 because it was significantly upregulated in classical monocytes both in hospitalized COVID-19 patients (compared with convalescent ones, Bonferroni-corrected p < 2.22 × 10−16) and the disease-related R4 sub-cluster (compared with other monocyte sub-clusters, Bonferroni-corrected p < 2.22 × 10−16). In the public BAL samples from COVID-19 patients,25 CCR2 was also highly expressed in the infiltrating monocytes that would later differentiate toward macrophages (Figure 5I). Another interesting example of potential regulation programs for DPP9, a candidate gene for COVID-19 severity, is depicted in Figures S10I–S10K. The DPP9 gene harbors SNPs associated with COVID-19 (p < 5 × 10−8) and was prioritized as a candidate gene that is involved in host-driven inflammatory lung injury in severe patients.13 Also, this locus has been previously reported to be associated with fibrotic idiopathic interstitial pneumonias,35 which suggests the potential role of dipeptidylpeptidase 9 (the enzyme encoded by DPP9) in severe COVID-19 patients. Moreover, early studies reported the enzyme is involved in antiviral signaling pathways,36 antigen presentation,37 and the activation of inflammasome.38 Taken together, these data depict an epigenetic regulation effect of risk allele of DPP9 locus in severe COVID-19 patients. Discussion Both host response and genetic predisposition affect the course and outcome of COVID-19, although the interplay between the two is not yet fully understood. Our single-cell multi-omics study has revealed numerous insights into the (epi)genetic mechanisms that regulate immune cells in COVID-19. First, we observed that COVID-19 has a pronounced effect on the transcriptional signature of classical monocytes, which was shown to be epigenetically regulated, and that the difference between mild and severe COVID-19 is due to a difference in the magnitude of response rather than differing transcriptional programs. Second, we depicted the regulatory properties of the long noncoding RNA LUCAT1 on the C/EBPs TFs, linking it to COVID-19 severity, and we experimentally validated the regulatory relationships. Finally, we identified a number of ASoC SNPs with potential regulatory effects in hospitalized COVID-19 patients. Interestingly, among these ASoC SNPs, rs6800484-C was associated with COVID-19 risk and linked to decreased chromatin accessibility, as well as reduced expression of CCR2, specifically in classical monocytes from hospitalized COVID-19 patients. Together, these findings shed light on the genetic, epigenetic, and transcriptional regulation of immune cells in COVID-19 (Figure 6 ).Figure 6 Schematic plot summarizing the genetic and epigenetic dysregulation of innate immunity in COVID-19 In our study, we observed a number of changes in cell proportions between hospitalized and convalescent COVID-19 patients, but fewer between severe and mild patients. Lymphopenia has been linked to COVID-19, as is also observed in our data, with hospitalized COVID-19 patients having a lower percentage of CD4+ T cells. We also saw an increased proportion of classical monocytes in hospitalized COVID-19 patients, as also observed earlier.7 In addition, we observed an upregulation of type I IFN signaling, which is crucial for antiviral immunity, in various monocyte subsets in COVID-19 patients compared with recovered individuals, e.g., IFI27, ISG15, and IFI6, which was also observed earlier in monocytes from COVID-19 patients compared with healthy controls.7 Interestingly, most of the observed differences were shared between mild and severe COVID-19 patients compared with convalescent individuals. This suggests that the difference in immunity between mild and severe COVID-19 is a matter of degree rather than reflecting distinct transcriptional profiles. This is in line with a previous observation that there are no immunological endotypes within the spectrum of COVID-1939 like those seen, for example, in sepsis.40 With open chromatin profiles, we observed enrichment of C/EBP, JUN, FOS, and SPI1 motifs in classical monocytes, which are also reported as critical TFs to monocyte development in sepsis.23 Because a subgroup of severe COVID-19 patients also developed a sepsis-like syndrome,40 , 41 , 42 there could be some overlapping immune-regulatory mechanisms at play. Examining co-expression of genes and peak-to-gene linkages, we found that the long noncoding RNA LUCAT1 interacts with all these SPI1, JUN, FOS, CEBPD, and CEBPE TFs. We applied knockout and inhibitor experiments to decode the regulatory and feedback mechanism among these molecules. The sub-clustering of monocytes further illustrated C/EBP motif-enriched classical monocyte subsets specific to hospitalized COVID-19 patients. These results together led us to envision a dysregulated cascade where increased C/EBP regulation enhanced LUCAT1 expression and further suppressed IFN responses to SARS-CoV-2 infection, which finally led to dysfunctional immune responses of COVID-19. Activation of C/EBP TFs was also reported to license the differentiation of profibrotic macrophages and trigger lung fibrosis in COVID-19.25 In our study, we observed the enrichment of open chromatin regions with C/EBP motifs in an oxygen-supply-associated monocyte sub-cluster, suggesting the activation of C/EBP regulation programs in circulating monocytes may also be associated with lung fibrosis and contribute to the need for oxygen in COVID-19 patients. Finally, we identified ASoC SNPs in regulatory elements that potentially disrupt regulation and consequently affect gene expression.43 , 44 Of note, we observed that the COVID-19 risk allele rs6800484-C corresponds to lower chromatin accessibility and lower expression of CCR2 in classical monocytes,13 suggesting the potential genetic and epigenetic regulatory function of rs6800484 in COVID-19 patients. The CCR2 gene encodes the chemokine receptor for monocyte chemoattractant protein-1 (MCP-1/CCL2), which promotes the migration of monocytes to sites of inflammation.45 , 46 MCP-1/CCL2 was reported to be enriched in BAL samples collected from severe COVID-19 patients, indicating active recruitment of CCR2+ monocytes and high inflammation in lung tissues.47 This suggests that the ASoC in the observed variant may impact monocyte recruitment to tissue by reducing CCR2 expression and thereby further influence the innate immune responses in COVID-19 patients. Although the down-regulation of CCR2 expression corresponds to higher viral loads and increased viral dissemination in animal models,34 we observed a significant down-regulation of CCR2 expression only in patients with C/C alleles compared with others, but not in the comparisons between hospitalized/severe COVID-19 patients and convalescent. In summary, our data have improved the understanding of the genetic and transcriptional regulation of dysregulated immune responses in COVID-19 and identified LUCAT1 and CCR2 as key regulators of detrimental immunity. Both factors contribute to COVID-19 pathogenesis in a subset of patients, while the co-action of these factors could bring heterogeneous responses to the SARS-CoV-2 infection. These leads can be used as a starting point for the development of personalized host-directed therapy to treat COVID-19. Limitations of the study Despite our interesting findings, this study also has several limitations. First, we focused our analysis to monocytes in peripheral blood. This choice was motivated by an over-representation of significant differences in the myeloid cell compartment, which also corresponds to our previous results.4 Although our data provided limited observations in lymphocytes, the importance of their role in the immune response in COVID-19 should not be ignored. Considering the large diversity and complexity of T cell populations, it will be interesting to dissect the T cell subsets in COVID-19 through T cell enrichment combined with T cell receptor sequencing in future studies. Second, although we used computational methods to link cells across scRNA-seq and scATAC-seq, the sequencing libraries of the two platforms were constructed independently. Therefore, we were unable to simultaneously profile gene expression and open chromatin from the same cell, which limited our power to characterize the full regulatory programs for different cells. Finally, our ASoC analyses were limited by the number of heterozygous SNPs among our participants. A future study based on a large cohort or a cohort pre-selected to have heterozygous alleles along the COVID-19 GWAS risk variants would address the full picture of genetic regulators of immune responses in COVID-19. STAR★Methods Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Biological samples Human peripheral blood mononuclear cell N/A N/A Chemicals, peptides, and recombinant proteins RPMI 1640 Medium SIGMA MDL# R0883-500ML Fetal Bovine Serum PAN BIOTECH Cat# P30-5500 Dulbecco’S Phosphate Buffered Saline PAN BIOTECH Cat# P04-36500 Buffer EB QIAGEN Cat# 19086 SPRIselect Beckmann Coulter Cat# B23318 Critical commercial assays Chromium Next GEM Single Cell 3′Regent Kits v3.1 10X CG000315 Rev A Chromium Next GEM Single Cell ATAC Regent Kits v1.1 10X CG000209 Rev D High sensitivity DNA kit Agilent Cat# 5067-4626 Deposited data Human reference genome NCBI build 38, GRCh38 Genome Reference Consortium48 http://www.ncbi.nlm.nih.gov/projects/genome/assembly/grc/human/ Human reference epigenomic annotations Roadmap Epigenomics Consortium31 https://egg2.wustl.edu/roadmap/web_portal/index.html Bulk RNA-seq eQTL summary statistics (release 2019-12-11) eQTL-Gen Consortium33 https://www.eqtlgen.org GTEx eQTL summary statistics (v8) GTEx v828 https://www.gtexportal.org/home/datasets Promoter capture HiC results Javierre et al.32 https://osf.io/u8tzp/ COVID-19 GWAS summary statistics (release round 6) The COVID-19 Host Genetics initiative12 https://www.covid19hg.org/results/r6 snRNA-seq data Wendisch et al.25 EGAS00001004928; EGAS00001005634; https://nubes.helmholtz-berlin.de/s/XrM8igTzFTFSoio snRNA-seq data This paper EGAS00001006559 snATAC-seq data This paper EGAS00001006560 Genotypes This paper EGAZ00001823187 Software and algorithms Plink v1.90b6.21 64-bit (19 Oct 2020) Chang et al.49 https://www.cog-genomics.org/plink/ R 4.1 R Core Team https://www.r-project.org/ R package Seurat version 3.2.2 Stuart et al.50 https://cran.r-project.org/web/packages/Seurat/index.html R package ArchR version 1.0.1 Granja et al.51 https://www.archrproject.com/ R package ggplot2 version 3.3.2 Wickham et al.52 https://cran.r-project.org/web/packages/ggplot2/index.html R package Bioconductor version 3.12 Gentleman et al.53 https://bioconductor.org R package clusterProfiler version 4.0.5 Yu et al.54 https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html R package ggpubr version 0.4.0 Kassambara, The Comprehensive R Archive Network (CRAN) https://cran.r-project.org/web/packages/ggpubr/index.html R package pheatmap version 1.0.12 Raivo, The Comprehensive R Archive Network (CRAN) https://cran.r-project.org/web/packages/pheatmap/index.html R package tidyverse version 1.3.0 Wickham et al.52 https://CRAN.R-project.org/package=tidyverse R package org.Hs.eg.db version 3.12.0 Carlson55 https://bioconductor.org/packages/3.12/data/annotation/html/org.Hs.eg.db.html R package ggrepel version 0.9.0 Slowikowski56 https://cran.r-project.org/web/packages/ggrepel/index.html R package data.table version 1.14.0 Dowle57 https://cran.r-project.org/web/packages/data.table/index.html R package GenomicInteractions version 1.30.0 Harmston et al.58 https://bioconductor.org/packages/release/bioc/html/GenomicInteractions.html R package GViz version 1.20.1 Hahne and Ivanek59 https://bioconductor.org/packages/3.15/bioc/html/Gviz.html R package motifbreakR version 2.10.0 Coetzee, Coetzee, and Hazelett30 https://bioconductor.org/packages/release/bioc/html/motifbreakR.html Python version 3.9.6 Van Rossum60 https://www.python.org Python package MACS2 version 2.2.7 Gaspar61 https://pypi.org/project/MACS2/ Bowtie2 version 2.4.4 Langmead and Salzberg62 N/A Python package matplotlib version 3.4 Hunter63 https://matplotlib.org Python package pysam version 0.17.0 Github pysam-developers https://github.com/pysam-developers/pysam BCFtools version 1.12 Danecek et al.64 http://www.htslib.org SAMtools version 1.12 Danecek et al.64 http://www.htslib.org Htslib version 1.12 Danecek et al.64 http://www.htslib.org WASP pipeline version 0.3.4 van de Geijn et al.65 https://github.com/bmvdgeijn/WASP GATK/ASEReadCounter version 4.2.0.0 Castel et al.66 https://gatk.broadinstitute.org VEP (online) McLaren et al.67 https://www.ensembl.org/Tools/VEP TOPMed and Michigan Imputation server Fuchsberger, Abecasis, and Hinds; NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium et al.; Das et al.68,69,70 https://imputation.biodatacatalyst.nhlbi.nih.gov; https://imputationserver.sph.umich.edu Original codes and scripts This paper https://doi.org/10.5281/zenodo.7270242 Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Yang Li ([email protected]). Materials availability This study did not generate new unique reagents. Experimental model and subject details We collected DNA and PBMCs from blood samples of COVID-19 patients. Samples taken on days when patients were hospitalized were considered hospitalized samples and samples taken from discharged patients were considered convalescent samples. Clinical information on age, sex, medication, active days, O2 supply, etc. is recorded for each sample and listed in Table S1. WHO scores were used to allocate the samples to mild (WHO 3–4) or severe (5–7) conditions according to the WHO clinical ordinal scale. The study was approved by the institutional review board at Hannover Medical School (#9001_BO_K2020) and informed consent was obtained from all patients. Method details Single-cell RNA-seq library preparation and sequencing Cells were counted, and an equal number of cells from five or six different individuals were pooled together. In total, 16,000 cells in total were loaded into the 10X ChromiumTM Controller, and libraries were prepared based on the manufacturer’s instructions (Chromium Next GEM Single Cell 3′ Reagent Kits v3.1 (Dual Index) User Guide, Rev A, CG000315 Rev A). Library quality per pool was examined using the Agilent Bioanalyzer High Sensitivity DNA kit. Sequencing was carried out on NovaSeq 6000 (Illumina), with a depth of 50,000 reads per cell. Single-cell ATAC-seq library preparation and sequencing Nuclei isolation was performed based on manufacturer’s instructions from 10X (CG000169 ⋅ Rev D). Briefly, cells were washed and lysed for 3 min on ice. After discarding the supernatant, lysed cells were diluted within 1× diluted nuclei buffer (10x Genomics) and counted using a Countess II FL Automated Cell Counter to validate lysis. An equal number of nuclei from five or six individuals were pooled and then loaded into the Chromium Next GEM Chip H based on the user guides from 10X genomics (Chromium Next GEM Single Cell ATAC Reagent Kits v1.1 User Guide, CG000209 Rev D). After breaking the emulsion, the barcoded tagmented DNA was purified and amplified for sample indexing and generation of scATAC-seq libraries. The final library was quantified using the Agilent Bioanalyzer High Sensitivity DNA kit. Sequencing was performed on NovaSeq 6000 (Illumina) with a depth of 25,000 reads per nuclei. Genotyping Genotyping of DNA samples isolated from subjects in the current study were performed using the GSA-MDv3 array (Infinium, Illumina) following the manufacturer’s instructions. In total, 725,875 variants of 48 individuals were called by Optical 7.0 with default settings. Quantification and statistical analysis Genotype imputation Quality control (QC) for raw variants was performed using PLINK.71 In brief, no sample was excluded initially due to failure in sex-check (--check-sex). Then, low-quality variants and individuals were excluded by parameters --geno 0.1 --mind 0.1. Next, missingness of genetic information and rates of heterozygosity were filtered by --missing and --het, respectively. After QC, 719,942 variants from 48 individuals were retained for the imputation procedure. The clean raw variants were uploaded to the TOPMed Imputation Server and imputed against the TOPMed (Version R2 on GRC38) reference panel.70 The imputed variants (n = 290,971,705) were downloaded and filtered by BCFtools,64 excluding variants with R2 < 0.5, with 14,232,029 variants retained for the downstream analysis. Additional QC and annotation was performed to obtain the genotypes that were used in the ASoC analysis. The variants were assigned reference SNP id (rs) by BCFtools against common variants (b151 GRCh38) downloaded from dbSNP. Subsequently, only variants that have rs numbers and that were heterozygous in at least three individuals were retained for the ASoC analysis. Data pre-processing and demultiplex of 10x genomics Chromium scRNA-seq data BCL files from each library were converted to FASTQ files using bcl2fastq Conversion Software (Illumina) using the respective sample sheet with the 10x barcodes utilized. The proprietary 10x Genomics CellRanger pipeline (v4.0.0) was used with default parameters. CellRanger was used to align read data to the reference genome provided by 10x Genomics (Human reference dataset refdata-cellranger-GRCh38–3.0.0) using the aligner STAR,72 and a digital gene expression matrix was generated to record the number of UMIs for each gene in each cell. The single-cell transcriptome in each library was further demultiplexed by assigning cell barcodes to their donor. The pre-mapped bam files of each library were loaded to Souporcell (v1.3gb)73 for a genotype-free SNP-based demultiplex with default settings, where candidate variants were called for each library and cells from each library were clustered into different samples based on their allele patterns. SNPs called from each sample were then matched with known genotypes of donors to assign a donor ID to each sample. The demultiplex assignments were double-checked using the expression of Y chromosome genes (ZFY, RPS4Y1, EIF1AY, KDM5D, NLGN4Y, TMSB4Y, UTY, DDX3Y, and USP9Y) in male samples. Data pre-processing and demultiplex of 10x genomics Chromium scATAC-seq data BCL files from each library were converted to FASTQ files using bcl2fastq Conversion Software (Illumina) using the respective sample sheet with the 10x barcodes utilized. The proprietary 10x Genomics CellRanger-ATAC pipeline (v1.2.0) was used with default parameters. CellRanger-ATAC was used to align read data to the reference genome provided by 10x Genomics (Human reference dataset refdata-cellranger-atac-GRCh38–1.2.0) and a fragments matrix was generated to record the number of reads for each open chromatin region in each cell. The cells in each library were further demultiplexed by assigning cell barcodes to their donor. The pre-mapped bam files of each library were loaded to Souporcell (v1.3gb) for genotype-free SNP-based demultiplexing. To call robust SNPs from the ATAC-seq samples, candidate variants were first called by freebayes74 with minimal mapping quality = 20, minimal base quality = 20, minimal coverage = 6, and minimal alternative allele = 2. Next, cell allele matrices from each library were generated with vartrix with minimal mapping quality = 20, and cells were clustered based on their allele patterns to identify different samples in one library. SNPs called from each sample were matched with known genotypes of donors to assign the donor IDs. Independent sample set In our cohort, some samples with different disease statuses came from the same donor. To obtain independent samples for pairwise comparison across conditions, we manually selected an independent sample set after QC for both scRNA-seq and scATAC-seq based on the following criteria: one sample per donor, early-stage for multiple stages, with samples shared by scRNA-seq and scATAC-seq data are preferred. The selected samples are marked in Table S1. QC for scRNA-seq data After the demultiplex, the expression matrix from PBMC was loaded to R/Seurat package (v3.2.2)50 for downstream analysis. To control the data quality, we first excluded cells with ambiguous assignments from Souporcell demultiplex. Next, we further excluded low-quality cells with >15% mitochondrial reads, <100 or >3,000 expressed genes, or <500 UMI counts (criteria were chosen according to the overall distribution of samples). In addition, genes expressed in less than three cells were also excluded from further analysis. Dimensionality reduction and clustering for scRNA-seq data After QC, we applied LogNormalization (Seurat function) to each cell, where original gene counts were normalized by total UMI counts, multiplied by 10,000 (TP10K), and then log-transformed by log10(TP10k+1). We then scaled the data, regressing for total UMI counts, and performed principal component analysis (PCA) based on the 2,000 most-variable features identified using the vst method implemented in Seurat. Subsequently, data from each sequencing batch was integrated, using the ‘harmony’ algorithm, based on the first 20 principal components to correct technical differences in the gene expression counts of different libraries. Cells were then clustered using the Louvain algorithm based on the first 20 ‘harmony’ dimensions with a resolution of 0.4. For visualization, we applied UMAP based on the first 20 dimensions of the ‘harmony’ reduction. Annotation of scRNA-seq clusters Clusters were annotated based on a double-checking strategy: 1) checking by automatic annotation with R/SingleR package75 and 2) manually checking the expression of cluster markers or known marker genes. Specifically, automatic annotation was applied with five pre-installed reference datasets in SingleR: HumanPrimaryCellAtlas data (HPCA),76 BlueprintEncode data,77 , 78 ImmuneCellExpression data,79 Novershtern Hematopoietic data,80 and MonacoImmune data.81 Cluster marker genes were identified by comparing gene expression of each cluster to all other clusters of the tested dataset using the FindAllMarkers function in Seurat with the Wilcoxon rank-sum test. Only upregulated genes with a log-fold change >0.25 and a Bonferroni-corrected p-value <0.05, and were expressed in at least 25% of cells were calculated for each cluster, and genes from each cluster of interest were ranked by their log-fold changes. In addition, T and NK cell clusters were further characterized by expression of marker genes related to memory T cells (IL7R), naive/central memory (SELL, CCR7), cell cytotoxic (CD8A, CD8B, NKG7, GZMB), interferon responses (IFI6, ISG15), and other data-derived cluster markers (Table S2). Monocyte clusters were then characterized by classical and non-classical monocyte markers (CD14, FCGR3A) and pro-inflammatory cytokines (TNF, IL1B), and the data-derived cluster markers, such as CD163 (Table S2). DE-Gs across Covid-19 conditions For pairwise comparison between COVID-19 conditions, differential expression (DE) tests were performed using the FindMarkers functions in Seurat with the Wilcoxon rank-sum test. The non-parametric Wilcoxon rank-sum test is distribution-free, but the results may still be biased by age effects. We therefore also performed DE tests using MAST,82 where we fit a hurdle model to the expression of each gene consisting of a linear regression for age as supplementary results. In both tests, genes with a log-fold change >0.05 and a Bonferroni-corrected p-value <0.05, and were expressed in at least 10% of tested groups were regarded as significantly differentially expressed. QC for scATAC-seq data After alignment and demultiplex, we used ArchR,51 a full-featured scATAC-seq analysis package, with minor adaptation to analyze our scATAC-seq data. Briefly, we created an Arrow file for the CellRanger mapped fragments file from each single-cell library and annotated the cells with the Souporcell demultiplex assignments. For QC, we filtered out cells that had fewer than 1,000 unique fragments, a transcription start site enrichment <4, or potential doublets recognized by the ArchR package. We also excluded cells with ambiguous assignments from Souporcell demultiplex. Finally, an ArchRProject combining all of the Arrow files was created for downstream analysis. Dimensionality reduction and clustering for scATAC-seq data After QCl, we used the ArchR function ‘addIterativeLSI’ to process iterative latent semantic indexing using the top 25,000 variable features and top 30 dimensions. We then used the harmony algorithm to correct batch effects from different libraries and clustered cells based on the results with a resolution of 0.8. For visualization, we applied UMAP based on the dimensions of the ‘harmony’ reduction with nNeighbors = 30 and minDist = 0.5. Annotation of scATAC-seq clusters To annotate the scATAC-seq clusters, gene scores were calculated and imputed for each cell, and marker genes from each cluster were detected with functions in ArchR package. Briefly, we firstly use ‘addGeneScoreMatrix’ function to independently compute gene activity scores per cell, then applied ‘addImputeWeights’ to impute gene scores by smoothing the signal across nearby cells using the MAGIC algorithm. Next, we compared independent gene scores between cells from one cluster and all the other clusters using the Wilcoxon rank-sum test to detect cluster-specific genes. The ‘bias’ parameter in ‘getMarkerFeatures’ from ArchR was used to account for transcription start site enrichment scores and the number of unique fragments per cell during the comparison. Finally, we visualized these genes and other cell-type-specific marker genes used for our scRNA-seq data to assign an identity to each cluster. Peaks calling and marker peaks detection To generate a comparable peak matrix for cross-sample comparison of differential open chromatin accessibility, reproducible peaks were called based on the pseudo-bulk replicates for each condition and clustered using the ‘addReproduciblePeakSet’ functions with Macs2 algorithm.83 After adding a peak matrix based on the called reproducible peaks, we applied differential peak detection with the Wilcoxon rank-sum test, again accounting for transcription start site enrichment scores and the number of unique fragments per cell during the comparison. For marker peaks per cell type and disease conditions, peaks were compared between cells from the tested group and cells from all other groups, and peaks with FDR <0.05 were considered as significant cell type- and/or condition-specific peaks. For pairwise comparison between conditions within a cell population, peaks with p-value <0.05 were considered as nominal differential accessible peaks and used for integrative analysis with DE-Gs. TF motif annotation and enrichment After calling peaks, we looked for the motifs that are enriched in peaks that are openly accessible in different cell types and conditions. To do this, we first added motif annotation based on the “CIS-BP” database,84 then, we applied the ‘peakAnnoEnrichment’ function in ArchR to obtain overrepresented motifs in test peak sets. Cross-platform linkage of scATAC-seq data with scRNA-seq data To do an integrative analysis of scATAC-seq and scRNA-seq data, we performed a preliminary integration by aligning all cells from scATAC-seq with cells from scRNA-seq by comparing the above-mentioned scATAC-seq cell-independent gene score matrix with the scRNA-seq expression matrix using the ‘FindTransferAnchors’ function from the Seurat package and the ‘addGeneIntegrationMatrix’ function from the ArchR package. Based on the result of this initial integration and the cell type annotation, we filtered out undefined scATAC-seq clusters and clusters with <100 cells aligned to annotated scRNA-seq clusters. We then annotated remaining scATAC-seq clusters based on the aligned scRNA-seq clusters. Finally, we re-ran the integration process by aligning remaining scATAC-seq cells to cells from the aligned scRNA-seq clusters and created a gene-integration matrix by adding gene integration scores to each cell. Peak-to-gene linkage To find potential regulation from peaks to genes, we inferred a peak-to-gene linkage by calculating the correlation between peak accessibility and gene expression within the above-mentioned integrated scRNA-seq and scATAC-seq cells. A co-accessibility >0.45 and FDR-adjusted p < 0.05 were regarded as regulatory links. TF footprinting with scATAC-seq data To calculate the TF footprint for each motif, we first obtained all the positions from one TF motif. To profile the footprint, cells were grouped again by each condition and each cell type to create pseudo-bulk ATAC-seq profiles. To account for the insertion sequence bias of the Tn5 transposase, which can lead to misclassification of TF footprints, we used the “Substract” normalization method to subtract the Tn5 bias from the footprinting signal. Sub-clustering of monocyte compartments in scRNA-seq In the scRNA-seq dataset, the monocyte subpopulations were investigated by applying sub-clustering on the three monocyte clusters (cMono, CD163+ cMono, and ncMono) identified in PBMC. We first identified the 1,000 most-variable features again in monocytes using the vst method implemented in Seurat. Next, we scaled the data and performed PCA based on these 1,000 most-variable features. Subsequently, the cells were clustered using the Louvain algorithm based on the top-10 PCs with a resolution of 0.3. For visualization, we applied UMAP based on the top-10 PCs. The marker genes for each sub-cluster were calculated by the FindAllMarkers function in Seurat and a contaminated lymphocyte cluster with CD3 gene expression was identified and removed from further analyses. AUCell-based gene signature scores were calculated using the AUCell method.27 We set the threshold for the calculation of the AUC to the top 3% of ranked genes and normalized the maximum possible AUC to 1. Top-30 marker genes reported from monocytes and macrophages reported in BAL fluid and PBMC4 were used to calculate AUC scores for each monocyte sub-clusters respectively. The resulting AUC values were subsequently visualized in violin plots. Sub-clustering of monocyte compartments in scATAC-seq In the scATAC-seq dataset, the two monocyte clusters (cMono and ncMono) identified in PBMC were extracted and investigated for sub-clustering analyses. Again, we used ArchR function ‘addIterativeLSI’ to process iterative latent semantic indexing using the top-25,000 variable features and top-30 dimensions. We then clustered cells based on the IterativeLSI reduced dimensions with a resolution of 0.8 and calculated UMAP with nNeighbors = 30 and minDist = 0.5. The resulting sub-clusters were aligned to scRNA-seq monocyte sub-clusters using the Cross-platform linkage method described above. Cells with a predicted linkage score >0.6 were regarded as aligned cells, and a scATAC-seq sub-cluster with a percentage of aligned cells >90% matched to the same scRNA-seq sub-cluster was regarded as the confidently matched sub-cluster. TF gene expression and regulon enrichment analysis In order to estimate the expression of the motif-enriched TFs in R4/C4 monocyte cluster, we ranked the genes based on their expressed percentages of R4 cells from hospitalized COVID-19 patients in scRNA-seq dataset and marked out TF genes among them. Next, we applied a regulon enrichment analysis across genes that were expressed at least 10% of R4 cells with SCENIC.27 Then, we intersected the enriched TFs, that were marked with 'high confidence' annotations by the algorithm, with the TF-motif enrichment results and marked the overlap TFs on the Heatmap. Chromatin accessibility of COVID-19 risk variants The genetic variants with a reported p-value <5 × 10−8 from the COVID-19 GWAS summary statistics (Hospitalized covid vs. population” release 6) by HGI14 were considered as risk variants of COVID-19. An open-chromatin peak was regarded to be associated with risk variants if its genomic location overlaps with at least one significant variant. The over-representation of “risk-variants-overlapping” peaks was estimated by the Fisher exact test comparing between peaks found in hospitalized patients and convalescent samples in each cell type, respectively. Identification of ASoC SNPs To estimate the allelic open chromatin for each identified cell type, the ATAC-seq reads of each subject were first split into individual BAM files per cell type using an in-house Python script according to the CB barcode which was added by CellRanger pipeline and error-corrected. The resulting BAM files were then calibrated using the WASP65 pipeline with Bowtie262 as aligner (-X 2000) to remove the mapping bias to reference allele at heterozygous sites. Afterward, the GATK/ASEReadCounter tool66 was used to count allelic reads at each heterozygous site with the default parameters. To detect the maximum allelic imbalance, the read counts from each subject were allelicly summed for each cell type at each heterozygous SNP, a pool approach that was justified in the previous study.43 Finally, only biallelic SNP sites with at least 20 read counts and at least 2 read counts for either allele were retained for downstream statistical analyses. To alleviate possible mapping bias to the reference allele, the WASP pipeline was applied, but no apparent bias effect was observed (Figure S10A). Binomial p values were calculated for the allelic read counts per SNP per cell-type by the R function binom.test(), with the alternative read counts as success trials and all read counts as total trials. Next, the R function p.adjust() was exploited to perform a multiple testing correction using the “fdr” method, and an FDR-adjusted p < 0.05 was considered the significant threshold. No obvious mapping bias to reference alleles was observed by visualizing the volcano plot of -log10(p values) and allelic read counts ratio. Annotation and function enrichment of ASoC SNPs To understand the function of the identified allelic imbalances, the ASoC SNPs were annotated against the GRCh38 reference genome using the online version of vep.67 Next, epigenomic annotations from RoadMap epigenomics projects85 were assigned to each identified ASoC SNPs based on their physical position and cell type. These epigenomic annotations were further grouped into promoters (including “TssA”, “PromU”, “PromD1”, and “PromD2”) and enhancers (including “TxReg”, “TxEnh5”, “TxEnh3”, “TxEnhW”, “EnhA1”, “EnhA2”, “EnhAF”, “EnhW1”, “EnhW2”, “EnhAc”, and “DNase”). To evaluate the effects of the identified ASoC SNPs, significant variant-gene pairs of whole blood tissue were downloaded from the GTEx Portal (V8) and allocated to the corresponding ASoC SNPs. Further, the DE genes identified by scRNA-seq between each pair of conditions in the current study were also attached to ASoC SNPs if the transcription start site of the gene is located in a 50Kbp-window of the ASoC SNPs. In addition, ASoC SNPs were allocated to TFs if the corresponding TF motifs were identified by chromVAR in scATAC-seq analysis of the current study. Finally, all the enrichment estimations were performed by R function fisher.test() while the adjustment of p values from multiple tests were done by p.adjust() using the “fdr” method except for those indicated in the context. Correlation between allelic imbalance and motif disruptions To test the effects of ASoC SNPs, i.e. a genetic perturbation, on TF binding motifs, we calulated motif break scores at each ASoC SNP using R package motifbreakR.30 Concretely, for each cell type, we first compiled a set of ASoC SNPs that are in the TF binding footprints identified in the TF binding footprints analysis. Subsequently, the disruptiveness of ASoC SNPs on TFBS were evaluated using motifbreakR() function with parameters: threshold = 1 × 10−4, method = "log", bkg = c(A = 0.25, C = 0.25, G = 0.25, T = 0.25). Next, only SNPs with a “strong” effect were retained, and motif break scores were represented by alleleDiff which is calculated by the difference between the scoreAlt and scoreRef in the motifbreakR results. Then, for each cell type of each condition, we estimated the correlation between motif break scores and allelic imbalance for each TF motif using Spearman’s rank correlation using cor.test() R function. The allelic imbalance was evaluated by the log2-transform ratio between alternative and reference ATAC-seq read counts per ASoC SNP. Finally, the correlations measured by Spearman’s rho were plotted as a heatmap using the ggplot2 package. Multi-omics integration from the public resource and the current study The functions of ASoC SNPs were also evaluated in scenarios of multi-omics integration. We downloaded publicly available GWAS/omics data, including COVID-19 GWAS summary statistics by HGI,14 whole blood eQTL summary statistics from the meta-analysis by eQTLGen,33 and promoter capture Hi-C data from Javierre and colleagues’ study.32 After integrating with scATAC-seq read depth, peak-to-gene links, and ASoC SNPs from the current study, the cross-omics results were visualized by the R/Gviz package59 along the genomic coordinates to show the ASoC SNP examples. Specifically, the track for promoter capture Hi-C and peak-to-gene links were visualized by R/GenomicInteractions.58 CRISPR- and inhibitor-experiments Cells deficient in LUCAT1 were generated using a lentiviral CRISPR interference vector (Addgene #71237). A gRNA inserts targeting the transcriptional start site of LUCAT1 was cloned into the vector followed by lentiviral particle production. To this end, HEK293T cells were transfected with the lentiviral vector, a VSVG pseudotyping plasmid (pVSVG) and a helper-plasmid (psPAX2), using lipofectamine 2000 reagent. Viral particles were collected by passing transfected cell supernatants through a 0.45 μm filter, followed by ultracentrifugation. For transduction, the viral pellet was resuspended in PBS and transferred to THP1 cells, followed by centrifugation at 37°C and 800 g for 2 h. Transduced cells were enriched using an Aria III cell sorter (BD), based on GFP-expression from the lentiviral backbone. THP1 and Hek293T cells were cultivated in RPMI 1640 medium (Thermo Fisher), supplemented with 10% FCS (Biochrom) and 1% penicillin/streptomycin solution (Thermo Fisher). Primary monocytes were isolated from Buffy coats (deidentified prior to use) using Lymphoprep gradient centrifugation and CD14-micoboeads (Miltenyi) and cultivated in X-vivo 15 medium (Lonza). All cells were kept in a 37°C incubator with a humidified atmosphere containing 5% CO2. For inhibitor experiments, cells were pre-incubated with the respective inhibitor for 2 h, followed by further stimulations. IL1α, IFNα were purchased from Preprotech and 3p-hairpin-RNA from Invivogen. Stimulations were carried out for 4 h (100 ng of each factor). RNA was extracted with Trizol reagent and qRT-PCR was done using the High Capacity cDNA Reverse Transcription kit (Thermo Fisher), LUNA Universal qPCR master mix (NEB) and a Quantstudio 3 instrument. Fold-changes based on CT values were calculated using the 2ˆ-DDCT method. Supplemental information Document S1. Figures S1–S10 Data S1. Tables S1–S7 Document S2. Transparent peer review records for Zhang et al Document S3. Article plus supplemental information Data and code availability All raw sequencing and genotypes generated during this study are deposited at the European Genome-phenome Archive (EGA) under the accession numbers EGA: EGAZ00001823187, EGAS00001006559, and EGAS00001006560, which are hosted by the EBI and the CRG. Processed data as Seurat objects with scRNA-seq count matrices and ArchR objects with scATAC-seq peak matrices have been deposited via Nubes: https://nubes.helmholtz-berlin.de/s/wqg6tmX4fW7pci5. Original codes and scripts used for the analyses are available at GitHub: https://github.com/CiiM-Bioinformatics-group/MHH50_COVID19_code and Zenodo: https://doi.org/10.5281/zenodo.7270242. Acknowledgments The authors thank all volunteers from the Hannover Medical School (MHH) for participation in the study. This study was supported by the COFONI (COVID-19 Research Network of the State of Lower Saxony) with funding from the Ministry of Science and Culture of Lower Saxony, Germany (14-76403-184) to Y.L., R.F., T.I., and J.H.; the Network Universities of Medicine (NUM) CODEX+ fund from the 10.13039/501100002347 Federal Ministry of Education and Research (BMBF) to Y.L.; the 10.13039/501100001659 Deutsche Forschungsgemeinschaft (DFG; German Research Foundation) under Germany’s Excellence Strategy—EXC 2155 project number 390874280 to Y.L.; and the Helmholtz Centre for Infection Research network fund (NASAVIR) to Y.L. and J.H. Y.L. was also supported by a European Research Council (ERC) Starting Grant (948207) and the 10.13039/501100001832 Radboud University Medical Center Hypatia Grant (2018) for Scientific Research. Z.Z. was supported by a joint scholarship by the 10.13039/501100001721 University of Groningen and China Scholarship Consortium (CSC201706350277) and Singh-Chhatwal-Postdoctoral Fellowship at the Helmholtz Centre for Infection Research. The Genotype-Tissue Expression (GTEx) Project was supported by the Common Fund of the 10.13039/100000179 Office of the Director of the 10.13039/100000002 National Institutes of Health and by the National Cancer Institute (NCI), National Human Genome Research Institute (10.13039/100000051 NHGRI ), National Heart, Lung, and Blood Institute (10.13039/100000050 NHLBI ), National Institute on Drug Abuse (NIDA), National Institute of Mental Health (10.13039/100000025 NIMH ), and National Institute of Neurological Disorders and Stroke (10.13039/100000065 NINDS ). The data used for the analyses described in this manuscript were obtained from the GTEx Portal on May 28, 2021. A.-E.S. acknowledges FOR-COVID (Bayerisches Staatsministerium für Wissenschaft und Kunst) and the 10.13039/501100009318 Helmholtz Association for support. Author contributions Conceptualization and study design: Y.L.; data analysis and investigation: B.Z., Z.Z., V.A.C.M.K., S.K., A.V., and R.G.; loss-of-function experiments: L.N.S.; discussion and interpretation: B.Z., Z.Z., V.A.C.M.K., Y.L., C.-J.X., S.K., M.C., L.N.S., L.E.S., A.V., U.O., and J.H.; sample collection and biospecimen resources: M.C., L.N.S., R.F., T.I., U.O., H.-C.T., Z.L., C.F.S., B.B., and R.G.; writing – original manuscript: B.Z., Z.Z., and V.A.C.M.K.; review and editing manuscript: all authors. Declaration of interests The authors declare no competing interests. Inclusion and diversity We support inclusive, diverse, and equitable conduct of research. Supplemental information can be found online at https://doi.org/10.1016/j.xgen.2022.100232. ==== Refs References 1 Wu F. Zhao S. Yu B. Chen Y.-M. Wang W. Song Z.-G. Hu Y. Tao Z.-W. Tian J.-H. Pei Y.-Y. A new coronavirus associated with human respiratory disease in China Nature 579 2020 265 269 10.1038/s41586-020-2008-3 32015508 2 Verity R. Okell L.C. Dorigatti I. Winskill P. Whittaker C. Imai N. Cuomo-Dannenburg G. Thompson H. Walker P.G.T. Fu H. Estimates of the severity of coronavirus disease 2019: a model-based analysis Lancet Infect. Dis. 20 2020 669 677 10.1016/S1473-3099(20)30243-7 32240634 3 Polack F.P. Thomas S.J. Kitchin N. Absalon J. Gurtman A. Lockhart S. Perez J.L. Pérez Marc G. Moreira E.D. Zerbini C. Safety and efficacy of the BNT162b2 mRNA covid-19 vaccine N. Engl. J. Med. 383 2020 2603 2615 10.1056/NEJMoa2034577 33301246 4 Schulte-Schrepping J. Reusch N. Paclik D. Baßler K. Schlickeiser S. Zhang B. Krämer B. Krammer T. Brumhard S. Bonaguro L. Severe COVID-19 is marked by a dysregulated myeloid cell compartment Cell 182 2020 1419 1440.e23 10.1016/j.cell.2020.08.001 32810438 5 Ren X. Wen W. Fan X. Hou W. Su B. Cai P. Li J. Liu Y. Tang F. Zhang F. COVID-19 immune features revealed by a large-scale single-cell transcriptome atlas Cell 184 2021 1895 1913.e19 10.1016/j.cell.2021.01.053 33657410 6 Chen G. Wu D. Guo W. Cao Y. Huang D. Wang H. Wang T. Zhang X. Chen H. Yu H. Clinical and immunological features of severe and moderate coronavirus disease 2019 J. Clin. Invest. 130 2020 2620 2629 10.1172/JCI137244 32217835 7 Zhang J.-Y. Wang X.-M. Xing X. Xu Z. Zhang C. Song J.-W. Fan X. Xia P. Fu J.-L. Wang S.-Y. Single-cell landscape of immunological responses in patients with COVID-19 Nat. Immunol. 21 2020 1107 1118 10.1038/s41590-020-0762-x 32788748 8 Chen Z. John Wherry E. T cell responses in patients with COVID-19 Nat. Rev. Immunol. 20 2020 529 536 10.1038/s41577-020-0402-6 32728222 9 Reusch N. De Domenico E. Bonaguro L. Schulte-Schrepping J. Baßler K. Schultze J.L. Aschenbrenner A.C. Neutrophils in COVID-19 Front. Immunol. 12 2021 652470 10.3389/fimmu.2021.652470 33841435 10 Krämer B. Knoll R. Bonaguro L. ToVinh M. Raabe J. Astaburuaga-García R. Schulte-Schrepping J. Kaiser K.M. Rieke G.J. Bischoff J. Early IFN-α signatures and persistent dysfunction are distinguishing features of NK cells in severe COVID-19 Immunity 54 2021 2650 2669.e14 10.1016/j.immuni.2021.09.002 34592166 11 Yang Q. Lin F. Wang Y. Zeng M. Luo M. Long noncoding RNAs as emerging regulators of COVID-19 Front. Immunol. 12 2021 700184 10.3389/fimmu.2021.700184 34408749 12 Severe Covid-19 GWAS GroupEllinghaus D. Degenhardt F. Bujanda L. Buti M. Albillos A. Invernizzi P. Fernández J. Prati D. Baselli G. Genomewide association study of severe covid-19 with respiratory failure N. Engl. J. Med. 383 2020 1522 1534 10.1056/NEJMoa2020283 32558485 13 Pairo-Castineira E. Clohisey S. Klaric L. Bretherick A.D. Rawlik K. Pasko D. Walker S. Parkinson N. Fourman M.H. Russell C.D. Genetic mechanisms of critical illness in COVID-19 Nature 591 2021 92 98 10.1038/s41586-020-03065-y 33307546 14 Callaway E. Mapping the human genetic architecture of COVID-19 Nature 596 2021 472 473 10.1038/s41586-021-03767-x 34417582 15 You M. Chen L. Zhang D. Zhao P. Chen Z. Qin E.-Q. Gao Y. Davis M.M. Yang P. Single-cell epigenomic landscape of peripheral immune cells reveals establishment of trained immunity in individuals convalescing from COVID-19 Nat. Cell Biol. 23 2021 620 630 10.1038/s41556-021-00690-1 34108657 16 Liu Z. Kilic G. Li W. Bulut O. Gupta M.K. Zhang B. Qi C. Peng H. Tsay H.-C. Soon C.F. Multi-omics integration reveals only minor long-term molecular and functional sequelae in immune cells of individuals recovered from COVID-19 Front. Immunol. 13 2022 838132 10.3389/fimmu.2022.838132 35464396 17 Gallagher M.D. Chen-Plotkin A.S. The post-GWAS era: from association to function Am. J. Hum. Genet. 102 2018 717 730 10.1016/j.ajhg.2018.04.002 29727686 18 Nott A. Holtman I.R. Coufal N.G. Schlachetzki J.C.M. Yu M. Hu R. Han C.Z. Pena M. Xiao J. Wu Y. Brain cell type-specific enhancer-promoter interactome maps and disease-risk association Science 366 2019 1134 1139 10.1126/science.aay0793 31727856 19 Corces M.R. Shcherbina A. Kundu S. Gloudemans M.J. Frésard L. Granja J.M. Louie B.H. Eulalio T. Shams S. Bagdatli S.T. Single-cell epigenomic analyses implicate candidate causal variants at inherited risk loci for Alzheimer’s and Parkinson’s diseases Nat. Genet. 52 2020 1158 1168 10.1038/s41588-020-00721-x 33106633 20 Chu X. Zhang B. Koeken V.A.C.M. Gupta M.K. Li Y. Multi-omics approaches in immunological research Front. Immunol. 12 2021 668045 10.3389/fimmu.2021.668045 34177908 21 Patterson B.K. Francisco E.B. Yogendra R. Long E. Pise A. Rodrigues H. Hall E. Herrara M. Parikh P. Guevara-Coto J. Persistence of SARS CoV-2 S1 protein in CD16+ monocytes in post-acute sequelae of COVID-19 (PASC) up to 15 Months post-infection Immunology 12 2021 746021 10.1101/2021.06.25.449905 22 Rosenbauer F. Tenen D.G. Transcription factors in myeloid development: balancing differentiation with transformation Nat. Rev. Immunol. 7 2007 105 117 10.1038/nri2024 17259967 23 Reyes M. Filbin M.R. Bhattacharyya R.P. Billman K. Eisenhaure T. Hung D.T. Levy B.D. Baron R.M. Blainey P.C. Goldberg M.B. Hacohen N. An immune-cell signature of bacterial sepsis Nat. Med. 26 2020 333 340 10.1038/s41591-020-0752-4 32066974 24 Agarwal S. Vierbuchen T. Ghosh S. Chan J. Jiang Z. Kandasamy R.K. The long non-coding RNA LUCAT1 is a negative feedback regulator of interferon responses in humans Nat. Commun. 11 2020 6348 10.1038/s41467-020-20165-5 33311506 25 Wendisch D. Dietrich O. Mari T. von Stillfried S. Ibarra I.L. Mittermaier M. Mache C. Chua R.L. Knoll R. Timm S. SARS-CoV-2 infection triggers profibrotic macrophage responses and lung fibrosis Cell 184 2021 6243 6261.e27 10.1016/j.cell.2021.11.033 34914922 26 Charo I.F. Ransohoff R.M. The many roles of chemokines and chemokine receptors in inflammation N. Engl. J. Med. 354 2006 610 621 10.1056/NEJMra052723 16467548 27 Aibar S. González-Blas C.B. Moerman T. Huynh-Thu V.A. Imrichova H. Hulselmans G. Rambow F. Marine J.-C. Geurts P. Aerts J. SCENIC: single-cell regulatory network inference and clustering Nat. Methods 14 2017 1083 1086 10.1038/nmeth.4463 28991892 28 GTEx ConsortiumLaboratory, Data Analysis &Coordinating Center LDACC—Analysis Working GroupStatistical Methods groups—Analysis Working GroupEnhancing GTEx eGTEx groupsNIH Common FundNIH/NCINIH/NHGRINIH/NIMHNIH/NIDABiospecimen Collection Source Site—NDRI Genetic effects on gene expression across human tissues Nature 550 2017 204 213 10.1038/nature24277 29022597 29 Abramov S. Boytsov A. Bykova D. Penzar D.D. Yevshin I. Kolmykov S.K. Fridman M.V. Favorov A.V. Vorontsov I.E. Baulin E. Landscape of allele-specific transcription factor binding in the human genome Nat. Commun. 12 2021 2751 10.1038/s41467-021-23007-0 33980847 30 Coetzee S.G. Coetzee G.A. Hazelett D.J. motifbreakR : an R/Bioconductor package for predicting variant effects at transcription factor binding sites: fig. 1 Bioinformatics 31 2015 3847 3849 10.1093/bioinformatics/btv470 26272984 31 Bernstein B.E. Stamatoyannopoulos J.A. Costello J.F. Ren B. Milosavljevic A. Meissner A. Kellis M. Marra M.A. Beaudet A.L. Ecker J.R. The NIH Roadmap epigenomics mapping Consortium Nat. Biotechnol. 28 2010 1045 1048 10.1038/nbt1010-1045 20944595 32 Javierre B.M. Burren O.S. Wilder S.P. Kreuzhuber R. Hill S.M. Sewitz S. Cairns J. Wingett S.W. Várnai C. Thiecke M.J. Lineage-specific genome architecture links enhancers and non-coding disease variants to target gene promoters Cell 167 2016 1369 1384.e19 10.1016/j.cell.2016.09.037 27863249 33 Võsa U. Claringbould A. Westra H.-J. Bonder M.J. Deelen P. Zeng B. Kirsten H. Saha A. Kreuzhuber R. Yazar S. Large-scale cis- and trans-eQTL analyses identify thousands of genetic loci and polygenic scores that regulate blood gene expression Nat. Genet. 53 2021 1300 1310 10.1038/s41588-021-00913-z 34475573 34 Vanderheiden A. Thomas J. Soung A.L. Davis-Gardner M.E. Floyd K. Jin F. Cowan D.A. Pellegrini K. Shi P.Y. Grakoui A. CCR2 signaling restricts SARS-CoV-2 infection mBio 12 2021 10.1128/mBio.02749-21 e0274921-21 35 Fingerlin T.E. Murphy E. Zhang W. Peljto A.L. Brown K.K. Steele M.P. Loyd J.E. Cosgrove G.P. Lynch D. Groshong S. Genome-wide association study identifies multiple susceptibility loci for pulmonary fibrosis Nat. Genet. 45 2013 613 620 10.1038/ng.2609 23583980 36 Zhang H. Maqsudi S. Rainczuk A. Duffield N. Lawrence J. Keane F.M. Justa-Schuch D. Geiss-Friedlander R. Gorrell M.D. Stephens A.N. Identification of novel dipeptidyl peptidase 9 substrates by two-dimensional differential in-gel electrophoresis FEBS J. 282 2015 3737 3757 10.1111/febs.13371 26175140 37 Geiss-Friedlander R. Parmentier N. Möller U. Urlaub H. Van den Eynde B.J. Melchior F. The cytoplasmic peptidase DPP9 is rate-limiting for degradation of proline-containing peptides J. Biol. Chem. 284 2009 27211 27219 10.1074/jbc.M109.041871 19667070 38 Griswold A.R. Ball D.P. Bhattacharjee A. Chui A.J. Rao S.D. Taabazuing C.Y. Bachovchin D.A. DPP9’s enzymatic activity and not its binding to CARD8 inhibits inflammasome activation ACS Chem. Biol. 14 2019 2424 2429 10.1021/acschembio.9b00462 31525884 39 Janssen N.A.F. Grondman I. de Nooijer A.H. Boahen C.K. Koeken V.A.C.M. Matzaraki V. Kumar V. He X. Kox M. Koenen H.J.P.M. Dysregulated innate and adaptive immune responses discriminate disease severity in COVID-19 J. Infect. Dis. 223 2021 1322 1333 10.1093/infdis/jiab065 33524124 40 Reyes M. Filbin M.R. Bhattacharyya R.P. Sonny A. Mehta A. Billman K. Kays K.R. Pinilla-Vera M. Benson M.E. Cosimi L.A. Plasma from patients with bacterial sepsis or severe COVID-19 induces suppressive myeloid cell production from hematopoietic progenitors in vitro Sci. Transl. Med. 13 2021 eabe9599 10.1126/scitranslmed.abe9599 34103408 41 Adams T.S. Schupp J.C. Poli S. Ayaub E.A. Neumark N. Ahangari F. Chu S.G. Raby B.A. DeIuliis G. Januszyk M. Single-cell RNA-seq reveals ectopic and aberrant lung-resident cell populations in idiopathic pulmonary fibrosis Sci. Adv. 6 2020 eaba1983 10.1126/sciadv.aba1983 32832599 42 Webb B.J. Peltan I.D. Jensen P. Hoda D. Hunter B. Silver A. Starr N. Buckel W. Grisel N. Hummel E. Clinical criteria for COVID-19-associated hyperinflammatory syndrome: a cohort study Lancet. Rheumatol. 2 2020 e754 e763 10.1016/S2665-9913(20)30343-X 33015645 43 Zhang S. Zhang H. Zhou Y. Qiao M. Zhao S. Kozlova A. Shi J. Sanders A.R. Wang G. Luo K. Allele-specific open chromatin in human iPSC neurons elucidates functional disease variants Science 369 2020 561 565 10.1126/science.aay3983 32732423 44 Atak Z.K. Taskiran I.I. Demeulemeester J. Flerin C. Mauduit D. Minnoye L. Hulselmans G. Christiaens V. Ghanem G.-E. Wouters J. Aerts S. Interpretation of allele-specific chromatin accessibility using cell state–aware deep learning Genome Res. 31 2021 1082 1096 10.1101/gr.260851.120 33832990 45 Ginhoux F. Jung S. Monocytes and macrophages: developmental pathways and tissue homeostasis Nat. Rev. Immunol. 14 2014 392 404 10.1038/nri3671 24854589 46 Serbina N.V. Pamer E.G. Monocyte emigration from bone marrow during bacterial infection requires signals mediated by chemokine receptor CCR2 Nat. Immunol. 7 2006 311 317 10.1038/ni1309 16462739 47 Zhou Z. Ren L. Zhang L. Zhong J. Xiao Y. Jia Z. Guo L. Yang J. Wang C. Jiang S. Heightened innate immune responses in the respiratory tract of COVID-19 patients Cell Host Microbe 27 2020 883 890.e2 10.1016/j.chom.2020.04.017 32407669 48 Schneider V.A. Graves-Lindsay T. Howe K. Bouk N. Chen H.-C. Kitts P.A. Murphy T.D. Pruitt K.D. Thibaud-Nissen F. Albracht D. Evaluation of GRCh38 and de novo haploid genome assemblies demonstrates the enduring quality of the reference assembly Genome Res. 27 2017 849 864 10.1101/gr.213611.116 28396521 49 Chang C.C. Chow C.C. Tellier L.C. Vattikuti S. Purcell S.M. Lee J.J. Second-generation PLINK: rising to the challenge of larger and richer datasets GigaScience 4 2015 7 10.1186/s13742-015-0047-8 25722852 50 Stuart T. Butler A. Hoffman P. Hafemeister C. Papalexi E. Mauck W.M. Hao Y. Stoeckius M. Smibert P. Satija R. Comprehensive integration of single-cell data Cell 177 2019 1888 1902.e21 10.1016/j.cell.2019.05.031 31178118 51 Granja J.M. Corces M.R. Pierce S.E. Bagdatli S.T. Choudhry H. Chang H.Y. Greenleaf W.J. ArchR is a scalable software package for integrative single-cell chromatin accessibility analysis Nat. Genet. 53 2021 403 411 10.1038/s41588-021-00790-6 33633365 52 Wickham H. Averick M. Bryan J. Chang W. McGowan L. François R. Grolemund G. Hayes A. Henry L. Hester J. Welcome to the tidyverse J. Open Source Softw. 4 2019 1686 10.21105/joss.01686 53 Gentleman R.C. Carey V.J. Bates D.M. Bolstad B. Dettling M. Dudoit S. Ellis B. Gautier L. Ge Y. Gentry J. Bioconductor: Open Software Development for Computational Biology and Bioinformatics 2004 54 Yu G. Wang L.-G. Han Y. He Q.-Y. clusterProfiler: an R Package for comparing biological themes among gene clusters OMICS A J. Integr. Biol. 16 2012 284 287 10.1089/omi.2011.0118 55 Carlson M. org.Hs.eg.db: Genome Wide Annotation for Human 2017 10.18129/B9.BIOC.ORG.HS.EG.DB R package version 3.12.0 56 Slowikowski K. Schep A. Hughes S. Dang T.K. Lukauskas S. Irisson J.-O. Kamvar Z.N. Thompson R. Christophe D. Hiroaki Y. R package ggrepel version 0.9.0 2020 CRAN 57 Dowle M. Srinivasan A. Gorecki J. Chirico M. Stetsenko P. Short T. Lianoglou S. Antonyan E. Bonsch M. Parsonage H. R package data.table version 1.14.0 2021 CRAN 58 Harmston N. Ing-Simmons E. Perry M. Barešić A. Lenhard B. GenomicInteractions: an R/Bioconductor package for manipulating and investigating chromatin interaction data BMC Genom. 16 2015 963 10.1186/s12864-015-2140-x 59 Hahne F. Ivanek R. Visualizing genomic data using Gviz and bioconductor Mathé E. Davis S. Statistical Genomics Methods in Molecular Biology 2016 Springer New York 335 351 10.1007/978-1-4939-3578-9_16 60 Van Rossum G. Python version 3.9.6. 2009 Python Software Foundation 61 Gaspar J.M. Improved peak-calling with MACS2 Bioinformatics 2018 10.1101/496521 62 Langmead B. Salzberg S.L. Fast gapped-read alignment with Bowtie 2 Nat. Methods 9 2012 357 359 10.1038/nmeth.1923 22388286 63 Hunter J.D. Matplotlib: A 2D Graphics Environment Computing in Science & Engineering 9 2007 90 95 64 Danecek P. Bonfield J.K. Liddle J. Marshall J. Ohan V. Pollard M.O. Whitwham A. Keane T. McCarthy S.A. Davies R.M. Li H. Twelve years of SAMtools and BCFtools GigaScience 10 2021 10.1093/gigascience/giab008 giab008 65 van de Geijn B. McVicker G. Gilad Y. Pritchard J.K. WASP: allele-specific software for robust molecular quantitative trait locus discovery Nat. Methods 12 2015 1061 1063 10.1038/nmeth.3582 26366987 66 Castel S.E. Levy-Moonshine A. Mohammadi P. Banks E. Lappalainen T. Tools and best practices for data processing in allelic expression analysis Genome Biol. 16 2015 195 10.1186/s13059-015-0762-6 26381377 67 McLaren W. Gil L. Hunt S.E. Riat H.S. Ritchie G.R.S. Thormann A. Flicek P. Cunningham F. The ensembl variant effect predictor Genome Biol. 17 2016 122 10.1186/s13059-016-0974-4 27268795 68 Fuchsberger C. Abecasis G.R. Hinds D.A. minimac2: faster genotype imputation Bioinformatics 31 2015 782 784 10.1093/bioinformatics/btu704 25338720 69 NHLBI Trans-Omics for Precision MedicineTOPMed) ConsortiumTaliun D. Harris D.N. Kessler M.D. Szpiech Z.A. Torres R. Taliun S.A.G. Corvelo A. Gogarten S.M. Kang H.M. Pitsillides A.N. Sequencing of 53, 831 diverse genomes from the NHLBI TOPMed Program Nature 590 2021 290 299 10.1038/s41586-021-03205-y 33568819 70 Das S. Forer L. Schönherr S. Sidore C. Locke A.E. Kwong A. Vrieze S.I. Chew E.Y. Levy S. McGue M. Next-generation genotype imputation service and methods Nat. Genet. 48 2016 1284 1287 10.1038/ng.3656 27571263 71 Purcell S. Neale B. Todd-Brown K. Thomas L. Ferreira M.A.R. Bender D. Maller J. Sklar P. de Bakker P.I.W. Daly M.J. Sham P.C. PLINK: a tool set for whole-genome association and population-based linkage analyses Am. J. Hum. Genet. 81 2007 559 575 10.1086/519795 17701901 72 Dobin A. Davis C.A. Schlesinger F. Drenkow J. Zaleski C. Jha S. Batut P. Chaisson M. Gingeras T.R. STAR: ultrafast universal RNA-seq aligner Bioinformatics 29 2013 15 21 10.1093/bioinformatics/bts635 23104886 73 Heaton H. Talman A.M. Knights A. Imaz M. Gaffney D.J. Durbin R. Hemberg M. Lawniczak M.K.N. Souporcell: robust clustering of single-cell RNA-seq data by genotype without reference genotypes Nat. Methods 17 2020 615 620 10.1038/s41592-020-0820-1 32366989 74 Garrison E. Marth G. Haplotype-based variant detection from short-read sequencing Preprint at arXiv 2012 10.48550/ARXIV.1207.3907 75 Aran D. Looney A.P. Liu L. Wu E. Fong V. Hsu A. Chak S. Naikawadi R.P. Wolters P.J. Abate A.R. Reference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage Nat. Immunol. 20 2019 163 172 10.1038/s41590-018-0276-y 30643263 76 Mabbott N.A. Baillie J.K. Brown H. Freeman T.C. Hume D.A. An expression atlas of human primary cells: inference of gene function from coexpression networks BMC Genom. 14 2013 632 10.1186/1471-2164-14-632 77 Martens J.H.A. Stunnenberg H.G. BLUEPRINT: mapping human blood cell epigenomes Haematologica 98 2013 1487 1489 10.3324/haematol.2013.094243 24091925 78 ENCODE Project Consortium An integrated encyclopedia of DNA elements in the human genome Nature 489 2012 57 74 10.1038/nature11247 22955616 79 Schmiedel B.J. Singh D. Madrigal A. Valdovino-Gonzalez A.G. White B.M. Zapardiel-Gonzalo J. Ha B. Altay G. Greenbaum J.A. McVicker G. Impact of genetic polymorphisms on human immune cell gene expression Cell 175 2018 1701 1715.e16 10.1016/j.cell.2018.10.022 30449622 80 Novershtern N. Subramanian A. Lawton L.N. Mak R.H. Haining W.N. McConkey M.E. Habib N. Yosef N. Chang C.Y. Shay T. Densely interconnected transcriptional circuits control cell states in human hematopoiesis Cell 144 2011 296 309 10.1016/j.cell.2011.01.004 21241896 81 Monaco G. Lee B. Xu W. Mustafah S. Hwang Y.Y. Carré C. Burdin N. Visan L. Ceccarelli M. Poidinger M. RNA-seq signatures normalized by mRNA abundance allow absolute deconvolution of human immune cell types Cell Rep. 26 2019 1627 1640.e7 10.1016/j.celrep.2019.01.041 30726743 82 Finak G. McDavid A. Yajima M. Deng J. Gersuk V. Shalek A.K. Slichter C.K. Miller H.W. McElrath M.J. Prlic M. MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data Genome Biol. 16 2015 278 10.1186/s13059-015-0844-5 26653891 83 Zhang Y. Liu T. Meyer C.A. Eeckhoute J. Johnson D.S. Bernstein B.E. Nusbaum C. Myers R.M. Brown M. Li W. Liu X.S. Model-based analysis of ChIP-seq (MACS) Genome Biol. 9 2008 R137 10.1186/gb-2008-9-9-r137 18798982 84 Weirauch M.T. Yang A. Albu M. Cote A.G. Montenegro-Montero A. Drewe P. Najafabadi H.S. Lambert S.A. Mann I. Cook K. Determination and inference of eukaryotic transcription factor sequence specificity Cell 158 2014 1431 1443 10.1016/j.cell.2014.08.009 25215497 85 Roadmap Epigenomics ConsortiumKundaje A. Meuleman W. Ernst J. Bilenky M. Yen A. Heravi-Moussavi A. Kheradpour P. Zhang Z. Wang J. Integrative analysis of 111 reference human epigenomes Nature 518 2015 317 330 10.1038/nature14248 25693563	36474914	PMC9715265	NO-CC CODE	2022-12-16 23:16:09	no	Cell Genom. 2022 Dec 2;:100232	utf-8	Cell Genom	2,022	10.1016/j.xgen.2022.100232	oa_other
==== Front Comput Biol Med Comput Biol Med Computers in Biology and Medicine 0010-4825 1879-0534 Published by Elsevier Ltd. S0010-4825(22)01091-5 10.1016/j.compbiomed.2022.106383 106383 Article Human monkeypox diagnose (HMD) strategy based on data mining and artificial intelligence techniques Saleh Ahmed I. Rabie Asmaa H. ∗ Computers and Control Dept. Faculty of Engineering Mansoura University, Mansoura, Egypt ∗ Corresponding author. 2 12 2022 1 2023 2 12 2022 152 106383106383 10 7 2022 2 11 2022 28 11 2022 © 2022 Published by Elsevier Ltd. 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. In May 2022, monkeypox re-emerged as a rare zoonotic disease that is an important viral disease for public health. Monkeypox can be transmitted from animals to humans, between humans through close contact with an infected human, or with a virus stained substance. Through this paper, a new detection strategy based on artificial intelligence techniques is provided to early detect monkeypox patients. This strategy is called Human Monkeypox Detection (HMD) strategy and mainly consists of two main phases, which are; (i) Selection Phase (SP) and (ii) Detection Phase (DP). While SP tries to select the best features, DP tries to introduce fast and accurate detection based on valid data from SP. In SP, an Improved Binary Chimp Optimization (IBCO) algorithm as a new feature selection algorithm is introduced to select valuable features before learning an Ensemble Diagnosis (ED) model as a new diagnostic algorithm in the next phase called DP. In fact, the proposed IBCO algorithm is a hybrid selection algorithm that includes both filter and wrapper methods. IBCO consists of a filter layer called Filter Selection Layer (FSL) and a wrapper layer called Wrapper Selection Layer (WSL). At first, monkeypox dataset is entered into FSL to quickly select meaningful features by using ‘m’ filter selection techniques. Then, ‘m’ sets of selected features are fed into WSL to construct the initial population of Binary Chimp Optimization (BCO) algorithm to precisely choose the best set of features for the next phase (DP). Finally, the ED model will be correctly trained on the filtered data from FSL. This model consists of three diagnostic algorithms called Weighted Naïve Bayes (WNB), Weighted K-Nearest Neighbors (WKNN), and deep learning which are combined using a new weighted voting method to provide the best diagnostic results. The weighted values of WNB algorithm are determined by measuring the impact of each feature on the class categories while the Grey Wolf Optimization (GWO) algorithm is used to determine the weighted values of WKNN. Experimental results illustrated that the suggested feature selection algorithm called IBCO outperforms other modern feature selection methods and also the proposed ED model outperforms other modern diagnostic models. At the end, the HMD strategy gives the best results compared to other modern strategies with accuracy, precision, and recall values equal 98.48%, 91.1% and 88.91% respectively. Also, the HMD gives 92.56%,89.01%,88.01%,85.01%, 83.9%, and 5.4 s for micro-average precision, micro-average recall, macro-average precision, macro-average recall, F1-measure, and implementation time values respectively. Keywords Monkeypox Artificial intelligence Ensemble classification Chimp algorithm Deep learning Feature selection ==== Body pmc1 Introduction Monkeypox is a rare zoonotic disease caused by monkeypox virus that belongs to the genus Orthopoxvirus in the family Poxviridae [[1], [2], [3], [4], [5]]. Two outbreaks of a pox-like disease appeared in colonies of monkeys preserved for research in 1958, hence, monkeypox was discovered [1]. In 1970, the first human patient of monkeypox was reported in the Democratic Republic of Congo and then the virus spread to other central and western countries in Africa [1]. Accordingly, the number of reported infected cases has increased. Although monkeypox virus has appeared since ancient times and no longer exists, the World Health Organization (WHO) and Centers for Disease Control and Prevention (CDC) have announced that the monkeypox virus has re-emerged now in May 2022 [[2], [3], [4]]. WHO and CDC announce that monkeypox is a self-limited disease. Additionally, symptoms of monkeypox last from 2 to 4 weeks [[2], [3], [4]]. The symptoms of human monkeypox are fever, body aches, headache, lymphadenopathy (lymph nodes to swell), Pustular Rashes, and exhaustion [1,2,4]. Recently, the case-fatality ratio was about 3–6% [4]. Early monkeypox detection is an essential process to reduce the spread of this infection around the world, isolate infected cases, and follow appropriate treatment of infected cases (see Table 4). Nowadays, Artificial Intelligence (AI) methods are used in many medical system applications such as end-to-end drug discovery and development, transcribing medical documents, patients diagnosis, pre-processing of medical data such as feature selection, and enhancing contact between physician and patient [6,7]. In fact, patients diagnosis and pre-processing processes based on AI techniques are the core of modern medical systems. That is because pre-processing techniques enable the medical system to filter data from useless data and diagnostic techniques can automatically diagnose patients without direct contact to medical staff. Thus, these techniques can reduce the efforts of medical staff, reduce patient waiting for examination, and reduce cost [6,7]. The importance of applying AI techniques to diagnose diseases is especially important in the event of a global pandemic resulting from new diseases that human expertise cannot diagnose with the required accuracy and speed. This may lead to a worsening of the health situation in the affected countries, threatening a catastrophe that the medical systems may be unable to absorb. Recently, serious and rapidly spreading diseases and epidemics have begun to appear in the world, such as Covid-19 disease and other diseases that have begun to appear, such as monkeypox. Therefore, diagnostic methods using AI techniques for serious diseases, that may be difficult for a human to diagnose quickly and accurately, are very important for the early, rapid and accurate detection of the disease to limit its spread. Nowadays, many diagnostic models need to select the informative features before starting to diagnose patients for their correct class category [[6], [7], [8]]. Feature selection is a pre-processing of data that is used to provide valuable features that enables diagnostic models to perform well. Thus, feature selection process aims to prevent overfitting. Two main classes called filter and wrapper can be used to classify feature selection techniques [6,7]. To diagnose disease like monkeypox, diagnostic models based on AI such as neural network, fuzzy inference system, Naïve Bayes, and Association Rules can be applied [6,7]. In fact, most researchers have diagnosed monkeypox patients based on Polymerase Chain Reaction (PCR) and manually. PCR test is fast, sensitive, and reliable but has the risk of getting false-negative and false-positive results. A negative PCR test result does not negate the possibility of monkeypox infection, so PCR test doesn't capture all infections. Hence, PCR test should not be taken into account as the only criterion for diagnosing monkeypox cases. Additionally, manual diagnosis is accurate but time-consuming. On the other hand, automatic diagnosis based on AI techniques can provide fast and accurate results and can also prevent the spread of infection between humans. Accordingly, it is an important to introduce a new strategy based on AI techniques to accurately and quickly diagnose monkeypox patients based on blood tests rather than relaying only on PCR test. During this paper, a new Human Monkeypox Detection (HMD) strategy has been presented to give rapid and more precise detection of monkeypox patients. HMD includes two main phases, which are; SP and DP. After selecting the most effective features of monkeypox patients using a new selection method in SP, the diagnostic process will be performed using a new ensemble diagnostic model in DP for early detection of monkeypox patients. The main objective of SP is to remove irrelevant features from the used dataset before beginning to learn the ED model in DP to enable it to introduce fast and precise diagnosis.• This paper provides two main contributions, which are; IBCO algorithm as a new feature selection method in SP and ED model as a new diagnostic model in DP. • The first contribution called IBCO is a hybrid selection algorithm that composes of two layers, namely; FSL and WSL. The main idea of IBCO is that it can solve the problems of the original version of BCO algorithm by determining the population size and initial values of search agents using many filter selection methods through FSL before implementing the BCO in WSL. • Hence, many filter selection methods are used in FSL to quickly select different sets of features. Then, these sets of features is passed to WSL to produce the initial population of BCO algorithm for accurately selecting the best set of features. At the last, the best set of features is utilized to correctly learn the ED model as a new diagnostic model in DP to provide rapid and precise diagnosis. • The second contribution called ED model is a hybrid diagnostic model that contains three new and different algorithms called WNB using the effect of each feature on the class categories to calculate the weighted values, WKNN using GWO to calculate the weighted values, and deep learning that are combined using a new weighted voting method. While WNB is a modified probabilistic method, WKNN is a modified distance method and deep learning is a machine learning method which are combined to give accurate results. • WNB algorithm is a new diagnostic model that modifies the classical NB model to take in the consideration the effective impact of each feature on the classifier by calculating the weights of features. • WKNN algorithm is a new diagnostic model that modifies the classical KNN model to use the best value of K and the best weights of features obtained from the GWO algorithm before implementing KNN algorithm. • Long Short-Term Memory (LSTM) model as a deep learning structure is used to diagnose monkeypox patients. • Then, the results of WNB, WKNN, and LSTM are combined by a new weighted voting method called Confusion Based Voting (CBV) to accurately take correct decisions. Experimental results showed that the proposed IBCO can provide the best subset of features compared to other recent methods and also the proposed ED model can provide the best results compared to its components based on the features selected by IBCO. Finally, the HMD strategy is superior other recently used strategies because it has the ability to give the best values of accuracy, precision, recall, micro-average, macro-average, F1-measure, and implementation time. The structure of this paper is organized as follows; segment 2 shows the problem definition but segment 3 provides the research motivation. Segment 4 reviews the previous research efforts about medical diagnostic techniques while the human monkeypox detection strategy will be discussed in segment 5. Segment 6 describes the experimental results while segment 7 depicts the conclusions and future directions. 2 Problem definition As the world is trying to get back to normal ignoring new record numbers of Covid-19, a new threat has been suddenly emerged, which is already spreading in the world known as Human MonkeyPox (HMP). Although HMP, which is a zoonotic viral disease, occurs predominantly in the rainforests of central and western Africa, it has recently appeared in the United States in wild rodents imported from Africa. The only escape for any potential pandemic that may arise from an outbreak of monkeypox is the accuracy and speed of diagnosis, so that it can be dealt with in the appropriate way. Accordingly, it is of utmost importance to find a way to accurately diagnose patients to give them the appropriate treatment at the right time. However, there are many challenges to the accurate and rapid diagnosis of HMP, including; (i) HMP is clinically almost identical to ordinary smallpox as both belong to the same group of viruses called orthopoxviruses [9]. They have similar clinical presentation including headache, fever, flulike symptoms, malaise, back pain, and characteristic rash, (ii) HMP is difficult to manage due to the limited knowledge of it among both patients and health staff as well as the huge lack of diagnostic tools and treatment protocols, (iii) the risk of the disease is not only limited to the advanced level of care that should be offered to the affected individuals but may also put other patients and health personnel at risk of infection. This paper keeps raising attention to an urgent need for an AI based methodology for fast and accurate diagnosing the disease [[9], [10], [11], [12], [13], [14]]. Fig. 1 depicts some considerations for HMP, where Fig. 1 (A) depicts 156 Occurrence locations of HMP into Central African (red), West Africa (blue), and unclassified (green) HMP genotypes, on the other hand, Fig. 1(B) illustrates the Overall predicted distribution of HMP based on ecological niche modeling [15]. the model shows the high capacity and speed of spread that characterizes the disease. Dark shades indicate regions with the greatest model agreement in predicting HMP fit, while green dots indicate input occurrences used in model development [15]. Fig. 1(C) shows the number of recorded HMP cases per country (May 26, 2022) and where disease is endemic based of WHO recent reports, finally, Fig. 1(D) presents some cases of HMP patients and what does the disease look like on the skin of the infected people.Fig. 1 (A) 156 Occurrence locations of human monkeypox into Central African (red), West Africa (blue), and unclassified (green) monkeypox genotypes, (B) the Overall predicted distribution of HMP based on ecological niche modeling, (C) Number of recorded HMP cases per country (May 26, 2022) and where disease is endemic, (D) Some cases of HMP patients. Fig. 1 3 Research motivation Currently, in light of the sudden emergence of infectious diseases that are rapidly spreading among humans, such as Covid-19 and monkeypox, it was necessary to find an early and rapid diagnosis of patients without contact with the medical staff. We are motived to work in this area of research to:• Introduce a complete diagnostic strategy for new and rapidly spreading infectious diseases. • Provide early and accurate diagnoses to new emerging diseases such as monkeypox based on AI and machine learning. • Prevent the spread of infection with newly emerged diseases by preventing direct contact between the patient and the medical staff. • Determine the appropriate treatment methods for the patient based on the correct diagnosis. • Reduce the risk of transmission of infection, whether from patients to non-patients or to the treating medical staff. 4 The previous efforts The previous efforts about the diagnostic methodologies in medical systems will be discussed in this segment. In Ref. [9], four diagnostic models called Random Forest (RF), Naïve Bayes (NB), Support Vector Machine (SVM), and Decision Tree (DT) were used to early identify diabetes. Experimental results showed that RF outperformed SVM, NB, and DT based on using two different datasets where RF provided the maximum accuracy, recall, and F-measure values. Although RF proved its effectiveness for diagnosing diabetes, it has not been tested on monkeypox. Additionally, RF is based on the original dataset without initially selecting the most effective features. As depicted in Ref. [1], a Neuro-Fuzzy based technique was provided to early diagnose monkeypox patients. This technique combines the benefits of fuzzy logic and artificial neural network techniques. In fact, fuzzy logic gave Neuro-Fuzzy the ability to handle uncertainty while neural network gave Neuro-Fuzzy the learning capability. In experimental results, Neuro-Fuzzy can effectivity diagnose monkeypox patients but it lacked to use all symptoms in the input. It also lacks the use of feature selection approach before implementing the diagnostic algorithm to enhance its performance further. As introduced in Ref. [8], an Ensemble Learning based Genetic Algorithm (ELGA) method was provided to early diagnose heart disease patients. ELGA method begins to select the most effective features and then diagnose heart disease. According to experimental results, ELGA provided the maximum accuracy value compared to other diagnostic models. Despite the benefits of ELGA, it has not been combined with many common diagnostic models, namely; NB, DT and SVM which may enable ELGA to improve its performance further. It also has not been tested on many different diseases such as breast cancer, lung cancer, Covid-19, and monkeypox. In [6], Distance Based Classification (DBC) strategy was introduced as a new diagnostic model to classify people vulnerability to Covid-19 infection. In fact, the DBC strategy consists of three stages; outlier rejection, feature selection, and classification. Hybrid outlier rejection approach that includes standard division and enhanced particle swarm optimization methods was implemented to reject noise data. Hybrid feature selection approach that includes chi-square and enhanced grey wolf optimization methods has been used to select valuable features. At the last, the filtered data has been passed to a diagnostic method called accumulative K-nearest neighbors to introduce fast and precise results. The description of experimental results showed that DBC provided the maximum accuracy and the minimum error and implementation time. Despite the benefits of DBC, it has not been combined with many heuristics models such as fuzzy logic and deep learning. Additionally, DBC has not been tested on other diseases such as monkeypox. As mentioned in Ref. [7], a new diagnostic model called Covid-19 Prudential Expectation (CPE) strategy has been provided to classify people vulnerability to Covid-19 infection. Outlier rejection, feature selection, and classification are the three main phases of CPE. Outlier rejection task was performed by executing improved grey wolf optimization algorithm while feature selection task has been performed by executing improved genetic algorithm. Finally, the filtered data has been passed to the statistical Naïve Bayes as a new diagnostic model to provide the best results. According to experimental results, CPE outperformed other strategies because it achieved the best accuracy and execution time values. Despite the effectiveness of CPE, it has not been implemented on other diseases such as monkeypox. For the Diagnosis of Monkeypox Patients (DMP) in the UK, clinical features of humans monkeypox have been characterized in Ref. [16]. This study based on 197 patients who had confirmed monkeypox based on PCR test. All 197 patients were men with a mean age of 38 years and had mucocutaneous lesions whereby 111 patients had genital infection or 82 patients had infection in the perianal region. It is concluded that there are several clinical features of monkeypox in humans in the UK that can be used for the early diagnosis of monkeypox patients. Despite the accurate description and in-depth study of the characteristics of monkeypox patients, which help in the early and accurate diagnosis of patients, it takes a great deal of time for diagnosis. Therefore, it was better to rely on AI and machine learning methods for rapid and accurate diagnosis based on data collected from patients. Additionally, it is not sufficient to rely only the PCR test to diagnose patients. In Ref. [17], Monkeypox Diagnosis Process (MDP) in a sexual health center in London, UK, depended on demographic and clinical features of the patients. Confirmed cases were detected using PCR test as 54 cases were infected with the monkeypox virus. Despite the careful examination of the 54 cases, this process took a great deal of time with a small number of cases. Also, depending only on the PCR test reduced the efficiency of the diagnosis. Therefore, it is preferable to use AI techniques to make a quick and accurate diagnosis based on other characteristics in addition to the PCR test. As described in Ref. [18], the Diagnostic Method (DM) based on clinical features of 7 patients with monkeypox characterized in the UK between 2018 and 2021 was performed to accurately diagnose monkeypox patients. PCR test was performed on all patients to identify confirmed cases of monkeypox. The clinical features included clinical data (such as symptoms and signs, complications of illness, demographic variables, and any antiviral treatments received) and laboratory results (such as monkeypox virus PCR results and routine biochemical tests). Despite the good description of the patients' condition and their diagnosis, this study suffers from a small number of samples, as well as the lack of use of artificial intelligence methods. Additionally, it is not sufficient to rely only the PCR test to diagnose patients. Between April and June 2022, 528 cases of monkeypox were diagnosed in 16 countries as presented [19]. To perform Diagnose Monkeypox Individuals (DMI) process, there are many demographic and clinical features mentioned in Ref. [19]. These clinical features included gay or bisexual men, human immunodeficiency virus infection, the median age, sexual activity, rash, anogenital lesions, and mucosal lesions. PCR test was used to confirm the infected cases. Despite the benefits of this diagnosis, it lacks the use of AI technology to deal with a large number of cases and also to give fast and accurate results. Also, depending only on PCR test is not sufficient. The recent diagnostic methodologies in medical systems are presented in Table 1 .Table 1 The recent diagnostic strategies in medical system. Table 1Technique Description Advantages Disadvantages Four diagnostic models [9] Four diagnostic models called RF, NB, SVM, and DT were used to early identify diabetes. RF outperformed SVM, NB, and DT based on using two different datasets where RF provided the maximum accuracy, recall, and F-measure values. - RF has not been tested on monkeypox. - RF is based on the original dataset without initially selecting the most effective features. Neuro-Fuzzy [1] Neuro-Fuzzy based technique was provided to early diagnose monkeypox patients. This technique combines the benefits of fuzzy logic and artificial neural network techniques. It can effectivity diagnose monkeypox patients. - It lacked to use all symptoms in the input. - It lacks the use of feature selection approach before implementing the diagnostic algorithm to enhance its performance further. Ensemble Learning based Genetic Algorithm (ELGA) [8] ELGA was provided to early diagnose heart disease patients. ELGA method begins to select the most effective features and then diagnose heart disease. It provided the maximum accuracy compared to other diagnostic models. - It has not been combined with many common diagnostic models, namely; NB, DT and SVM which may enable ELGA to improve its performance further. - It has not been tested on many different diseases such as breast cancer, lung cancer, Covid-19, and monkeypox. Distance Based Classification (DBC) strategy [6] DBC was introduced to classify people vulnerability to Covid-19 infection. This strategy consists of three stages; outlier rejection, feature selection, and classification. It provided the maximum accuracy and the minimum error and implementation time. - It has not been combined with many heuristics models such as fuzzy logic and deep learning. - It has not been tested on other diseases such as monkeypox. Covid-19 Prudential Expectation (CPE) strategy [7] CPE has been provided to classify people vulnerability to Covid-19 infection. Outlier rejection, feature selection, and classification are the three main phases of CPE. It outperformed other strategies because it achieved the best accuracy and execution time values. - it has not been implemented on other diseases such as monkeypox. Diagnosis of Monkeypox Patients (DMP) [16] Clinical features of humans monkeypox for 197 patients who had confirmed monkeypox based on PCR test were characterized to diagnose patients. It provided accurate description and in-depth study of the characteristics of monkeypox patients, which help in the early and accurate diagnosis of patients. - Diagnosis takes a long time. - Diagnosis is done manually instead of using AI techniques. - It only relies on PCR testing to find confirmed cases. Monkeypox Diagnosis Process (MDP) [17] In a sexual health center in London, UK, demographic and clinical features of the patients were used to diagnose infected cases. Careful examination of 54 cases was conducted to provide an accurate diagnosis. - Diagnosis process took a great deal of time with a small number of cases. - Depending only on the PCR test reduced the efficiency of the diagnosis. - It only relies on PCR testing to find confirmed cases. Diagnostic Method (DM) [18] Humans diagnosis based on clinical features of 7 patients with monkeypox characterized in the UK between 2018 and 2021 was performed to accurately diagnose monkeypox patients. The good description of the patients' condition and their diagnosis were provided. - This study suffers from a small number of samples,. - It also suffers from the lack of use of AI methods. - Additionally, it is not sufficient to rely only the PCR test to diagnose patients. Diagnose Monkeypox Individuals (DMI) process [19] Based on several demographic and clinical features such as gay or bisexual men, human immunodeficiency virus infection, the median age, sexual activity, rash, anogenital lesions, and mucosal lesions, diagnosis was performed. It can accurately diagnose monkeypox patients based on their clinical features. - It takes a long time for diagnosing patients. - Diagnosis is done manually but AI techniques were not used. - It only based on PCR testing to determine confirmed patients. 5 The human monkeypox detection (HMD) strategy This segment describes the HMD strategy for early detection of monkeypox patients. The HMD strategy attempts to quickly and accurately diagnose patients who suffer from monkeypox. In fact, this strategy composes of two main phases which are named Selection Phase (SP) and Detection Phase (DP) as provided in Fig. 2 . While SP aims to choose the most informative features in dataset without any useless features, DP aims to rapidly and accurately diagnose monkeypox patients depending on valid data from SP. Thus, feature selection task will be carried out at first to select valuable features and then the monkeypox detection model will be learned by valid dataset based on informative features to give the desired results. Feature selection algorithms can be categorized into two main classes called filter and wrapper [6,7]. Actually, filter techniques are faster than wrapper but less precise than wrapper [6,7].Fig. 2 The human monkeypox detection (HMD) strategy. Fig. 2 Recently, optimization algorithms have been applied as wrapper selection algorithms to select valuable features [6,7,10]. Hence, a significant amount of time can be consumed by optimization algorithms versus providing an accurate set of features. Accordingly, a new feature selection methodology, namely; Improved Binary Chimp Optimization (IBCO) algorithm will be provided as a hybrid method that include both filter and wrapper algorithms to quickly and carefully choose the best features. Accurate set of features selected from SP allows ED model as a new diagnostic model in DP to be learned correctly and thus can give fast and accurate results. In fact, the ED model is a hybrid diagnostic model consisting of three techniques, which are; WNB, WKNN, and deep learning that are combined together by a new weighted voting method used to provide the best diagnostic results. In WNB, weighted values are calculated by measuring the effect of each feature on the class categories while the GWO is used to determine WKNN weighted values. In the following sub-segments, the IBCO as a new feature selection methodology in SP and the ED model as a new diagnostic model in DP will be explained in details. 5.1 Improved Binary Chimp Optimization (IBCO) algorithm In this segment, the details of the proposed IBCO algorithm as a new selection algorithm in the first phase of the provided strategy called SP will be described. The IBCO algorithm is a hybrid selection method that includes filter and wrapper algorithms to provide fast and accurate group of features. It consists of two layers called Filter Selection Layer (FSL) and Wrapper Selection Layer (WSL) as illustrated in Fig. 3 . FSL aims to quickly choose group of valuable features while WSL aims to accurately select the best subset of features. In fact, Binary Chimp Optimization (BCO) algorithm as a wrapper algorithm can give correct results but it is time consuming [12]. In addition to the fact that BCO is a slow technique, it lacks the determination of the exact number of search agents “chimps” in the population and their initial values. Thus, IBCO aims to enhance the performance of BCO by using FSL as a quick layer before using it in WSL. In FSL, ‘m’ filter selection techniques will be applied in parallel and each technique will separately provide a set of selected features. After that, the sets of features selected by ‘m’ techniques in FSL will be passed to WSL as initial population of BCO. Hence, initial population of BCO consists of ‘m’ search agents that include initial values equal the sets of features selected by ‘m’ filter selection techniques in FSL. Based on this initial population, BCO tries to quickly select an accurate set of features that can give a diagnostic model the ability to introduce quick and correct monkeypox diagnoses.Fig. 3 The steps of IBCO as a feature selection method. Fig. 3 The second difference between the original BCO and the IBCO is that the IBCO depends on a better fitness function that is the average accuracy value produced by many diagnostic methods learned on the same set of features in dataset to produce the fitness of each chimp in the population. Thus, the evaluation values for chimps in IBCO will be generated by using many diagnostic methods rather than using only specific one. The main aim of that is to achieve the generality of selecting useful features that can adapt to any diagnostic model. Implementing IBCO requires many sequential steps as shown in Fig. 3. At first, monkeypox dataset will be passed to ‘m’ filter selection techniques in FSL to provide ‘m’ sets of selected features. Secondly, these sets of selected features will be used to generate the initial population of BCO in WSL. In fact, each chimp in the population will be represented in a q-dimensional space and also in a discrete form to represent a set of meaningful features in each chimp. Hence, each chimp's length equals the same number of features in the monkeypox dataset where the bits of each chimp includes either ‘0’ or ‘1’ value; 0 = not selected features and 1 = selected features. Then, the steps of BCO will be continued until stopping condition (the maximum iterations number) is reached. After generating initial population from FSL as showed in Fig. 3, chimps will be evaluated by using the fitness function in (1).(1) FF(Hk)=∑j=1ndAccj(Hk)nd Where FF(H k ) represents the evaluation value of kth chimp, the accuracy value of jth diagnostic model based on the set of features in kth chimp is Acc j (H k ). j represents an index to the used diagnostic models; j = 1,2, ….,nd where their number is nd. To illustrate the idea, it is assumed that the population size is two (two chimps) and three diagnostic models (nd = 3) will be applied to assess the picked features in every chimp as depicted in Table 2 . In Table 2, the used diagnostic models are SVM [6], Deep Learning Method (DLM) [[20], [21], [22]], and K-Nearest Neighbors (KNN) [13]. It is noted that the maximum accuracy values of SVM and DLM are presented in H 1 while the maximum accuracy value of KNN is presented in H 2. Hence, the best chimp is the first solution (H 1) related on the average accuracy value to introduce a global solution because single diagnostic model cannot generally give the best set of features that can deal with any other diagnostic model (see Table 3).Table 2 Identify the best chimp depending on both every diagnostic model and the average accuracy value. Table 2Diagnostic model # Accuracy of every chimp The best chimp (Hattack) H1 H2 D1 = SVM 0.75 0.7 H1 D2 = DLM 0.9 0.7 H1 D3 = KNN 0.8 0.9 H2 Average accuracy 0.816 0.767 H1 Table 3 CM for WNB. Table 3 Predicted Class Total A B Actual Class A 550 = 92% 50 600 B 240 160 = 40% 400 790 210 1000 Table 4 CM for WKNN. Table 4 Predicted Class A B Total Actual Class A 370 = 62% 230 600 B 180 220 = 55% 400 550 450 1000 After all chimps in the population are evaluated and their fitness values are calculated, the four leaders (H attack , H bar , H chas , and H driv) as the best solutions are determined. In the population, the positions of rest chimps (H k) will be adjusted for the next iteration (t+1) based on the positions of leaders at the current iteration (t) by using (2), (3), (4), (5), (6) [12,14].(2) H→1(t+1)=H→attck(t)−Ah→1.Dh→attck,Dh→attck=\|Ch→1.H→attcker−mhH→k(t)\| (3) H→2(t+1)=H→bar(t)−Ah→2.Dh→bar,Dh→bar=\|Ch→2.H→bar−mhH→k(t)\| (4) H→3(t+1)=H→chas(t)−Ah→3.Dh→chas,Dh→chas=\|Ch→3.H→chas−mhH→k(t)\| (5) H→4(t+1)=H→driv(t)−Ah→4.Dh→driv,Dh→driv=\|Ch→4.H→driv−mhH→k(t)\| (6) H→k(t+1)=H→1+H→2+H→3+H→44 Where the current iteration number is t and the position of each chimp at t iteration is H k (t). Additionally, the distance between a prey and the chimp (H k) is Dh, the positions of the best four chimps are H 1 , H 2 , H 3 ,and H 4 respectively, and mh represents a chaotic value. In fact, mh includes value between [0,1] using quadratic map that refers to the effect of the agents' sexual motivation that can be calculated by using (7).(7) mh=Hk2−g,g=1 Coefficient vectors are Ah and Ch which are adjusted to determine the nearest solution to the optimal. For each leader, Ah and Ch will be calculated by using (8), (9), (10), (11).(8) Ah1=\|2fhrh11−fh\|,Ch1=2rh12 (9) Ah2=\|2fhrh21−fh\|,Ch2=2rh22 (10) Ah3=\|2fhrh31−fh\|,Ch3=2rh32 (11) Ah4=\|2fhrh41−fh\|,Ch4=2rh42 Where fh is decreasing from 2 to 0 linearly. It can be calculated by using (12).(12) fh=2−2(tMT) Where the maximum number of iterations represents MT and random factors between [0,1] which are calculated for each leader chimp are rh 1 and rh 2 using (13), (14), (15), (16), (17), (18), (19), (20) [12].(13) rh11=u1d1Rand(),u1d1=1.95−(2(t14)MT13) (14) rh12=u2d1Rand(),u2d1=(2(t13)MT13)+0.5 (15) rh21=u1d2Rand(),u1d2=1.95−(2(t13)MT14) (16) rh22=u2d2Rand(),u2d2=(2(t3)MT3)+0.5 (17) rh31=u1d3Rand(),u1d3=(−3(t3)MT3)+1.5 (18) rh32=u2d3Rand(),u2d3=(2(t13)MT13)+0.5 (19) rh41=u1d4Rand(),u1d4=(−2(t3)MT3)+0.5 (20) rh42=u2d4Rand(),u2d4=(2(t3)MT3)+0.5 Where uniform distribution between [0,1] is Rand() and the dynamic coefficients applied to determine the values of rh 1 and rh 2 are u 1 d 1 , u 2 d 1 , u 1 d 2 , u 2 d 2 , u 1 d 3 , u 2 d 3 , u 1 d 4 , and u 2 d 4. In the population, it is assumed a probability of 50% to select between either the normal updating position method or the chaotic model (mh) to update the positions of chimps by using (21).(21) Hk(t+1)={H1+H2+H3+H44,if(y<0.5)mh,if(y≥0.5) Where a random value between [0,1] is y. In fact, a new position value for each chimp H k in the population is generated in a continuous form but this form cannot be applied to choose meaningful features. Hence, the converting function called sigmoid function should be applied to transform the continuous value to binary value. Accordingly, every chimp's position in the population; H k = (H k 1 , H k 2 , …..,H k q) will be modified by implementing the sigmoid function to calculate new chimp's position in a discrete form; H bin_k = (H 1 bin_k , H 2 bin_k , …..,H q bin_k) by using (22) [6,7].(22) Hbin_ki(t+1)={1ifRAND(0,1)≥SG(Hki)0Else Where the binary value of kth chimp in the next iteration t+1 at ith position is H i bin_k (t+1) and i is a pointer to the current position (feature); i = 1,2,3, …..,q. Random value between 0 and 1 is RAND(0,1) and the sigmoid function is SG(H k i ). In fact, SG(H k i ) refers to the probability of ith bit that includes one or zero value measured by applying (23) [6,7].(23) SG(Hki)=11+e−Hki Where e is the base of the natural logarithm. Related to H i bin_k (t+1) as a new position of every chimp in the population, the fitness value of every chimp is calculated by applying the fitness function in (1). The steps of BCO will be finished when the stopping condition is satisfied. At the last, the fittest chimp (H 1 or H attack) is the best solution and the algorithm is finished. Then, all bits that includes 1 in H 1 are the most effective features that will be used to enable the diagnosis model to correctly learned for providing quick and more accurate diagnosis. In other words, the filtered dataset without irrelevant features in the SP will be passed to the next phase of the provided strategy called DP to correctly learn the ED model in order to give quick and accurate diagnosis for monkeypox patients. The steps of IBCO algorithm are mentioned in algorithm 1.Algorithm 1 Improved Binary Chimp Optimization (IBCO) Algorithm.Image 1 5.2 The proposed Ensemble Diagnosis (ED) model In this segment, the ED model as a new diagnostic model in the second phase of the HMD strategy called DP will be discussed in detail. The ED model is a hybrid model consisting of three diagnostic algorithms called WNB, WKNN, deep learning implemented on the filtered dataset that is passed from the previous phase called SP without irrelevant features to be accurately diagnose monkeypox patients. The ED model aims to combine the results of these three diagnostic algorithms together through a new weighted voting method to provide more accurate results. The steps of implementing this model are showed in Fig. 4 . As presented in Fig. 4, the filtered monkeypox dataset will be divided into training, testing, validating dataset before starting to implement the ED model. Then, the ED model implementation sequence will be passed through four main stages called training, testing, validation, and voting stages. At the training stage, the three diagnostic algorithms; WNB, WKNN, and deep learning will be trained in parallel on the same training dataset.Fig. 4 The steps of Ensemble Diagnosis (ED) model. Fig. 4 Secondly, these algorithms will be tested in parallel on the same testing dataset during the testing stage. At the testing stage, the class categories will have different weight values according to each diagnostic algorithm, whether it is WNB, WKNN or deep learning. At the third stage called validation stage, each case in the validation dataset will be diagnosed by WNB, WKNN, and deep learning algorithms into different or similar class categories that have different weight values according to each diagnostic algorithm. In the fourth and final stage called voting stage, the weight value of the class category for each validate case will be passed from the three diagnostic algorithm to voting stage for determining the final diagnosis based on a new weighted voting method. In the next sub-segments, WNB, WKNN, deep learning, and weighted voting methods will be described in detail. 5.2.1 The Weighted Naïve Bayes (WNB) algorithm In this segment, WNB is presented as an improved version of NB method to solve the NB problems. In fact, NB is a popular classification method that is characterized by simplicity and it can address real-time problems such as image and pattern recognition, medical diagnosis, and intrusion detection [[23], [24], [25]]. NB can give fast diagnoses rather than other diagnostic models and also it can be used for both small and large dataset. Additionally, NB has the ability to deal with the noise in the dataset as well as it is less sensitive to missing data [[23], [24], [25]]. Although NB is a sufficient technique for real-time applications such as medical diagnosis application, it is considered all features equal and independent during the diagnosis process. Hence, NB should be modified to depend on the effective impact of each feature on the classifier to give more accurate results. In this paper, WNB is provided as a modification of the classical NB that takes into account the weights of features. In the WNB, the diagnosis of each patient in the dataset can be performed based on the different weights of features where each feature has its own weight according to its effectiveness on the class category using (24).(24) Diagnose(Ix)=argmaxcli∈cl[Probability(cli)∏j=1pWeightjProbability(fj\|cli)]whereWeightj∈R+ Where Diagnose (I x ) is the diagnosis of patient I x to the class category that give the highest probability value. Probability (cl i ) is the prior probability of the class cl i while Probability (f j \|cl i ) is the conditional probability of the feature f j according to the class cl i. Additionally, weight j is the weight of the jth feature that represents the impact of this feature on the class category using (25).(25) weightj=Acc(+fj)−Acc(−fj) Where weight j is the weight of the jth feature, the accuracy of the NB method based on the existence of the feature f j in the feature set is Acc(+f j ), and the accuracy of the NB method based on the absence of the feature f j from the feature set is Acc(-f j ). 5.2.2 The weighted K-nearest neighbors (WKNN) algorithm WKNN is provided in this segment as an improved version of traditional KNN method to treat the KNN problems. KNN is a popular and straightforward method that is simple and easy to understand and use. It is used in many real-world applications such as electrical load forecasting, patient diagnosis, and traffic management [13]. Although KNN is simple, it is a lazy learning technique, depends on the value of K, and does not take into account the weight of each feature because each feature has different impact on the classification. Thus, KNN should be modified to use the best value of K and the best weights of features to provide the best classifications. In this paper, WKNN is presented as a new diagnostic algorithm based on using the optimal value of K and the best weight value for each feature obtained from the GWO algorithm before learning KNN algorithm. There are many steps to implement WKNN algorithm as shown in Fig. 5 . The GWO will be implemented to select the best weight values for the features and the best K value and then the WKNN will be implemented on these best values. At first, initial population of GWO will be generated where each wolf includes weights of features and K value. Then, each wolf in the population will be evaluated using (26).Where Evaluation (W i ) is the evaluation value of ith wolf and WKNN_Accuracy(W i ) is the accuracy of implementing WKNN algorithm based on the values of genes in ith wolf. The best three wolves; W α, W β, and W δ as leaders will be decided based on the high accuracy values. Depending on the position of these three wolves, the other wolves in the population including Omega (ω) will modify their position. Coefficient vectors AW and CW for the leaders must be calculated before starting to modify the positions of wolves in population using (27), (28) [6,26].(27) AW→=\|2aW→ran→1−aW→\| (28) CW→=2ran→2 Where ran→1 and ran→2 are random vectors in [0,1]. The encircling coefficient that is used to balance the tradeoff between exploration and exploitation is aW→.In fact, aW→ is linearly decreasing from 2 to 0 over iterations using (29) [6,26].(29) aW→=2−2(itrM_itr) Where the number of iterations is itr and the maximum number of iterations is M_itr. After calculating the coefficient vectors AW and CW for the leaders, each wolf (e.g., ith wolf) in population can modify its position in the next iteration (itr+1) based on W α, W β, and W δ by using (30) [6,26].(30) W→i(itr+1)=W→1+W→2+W→33 Where the positions of W α, W β, and W δ are W→1,W→2 , and W→3 respectively based on the current wolf (W i). In fact, W→1, W→2, and W→3 can be calculated as in (31), (32), (33) [6,26].(31) W→1=W→α−AW→1D→α (32) W→2=W→β−AW→2D→β (33) W→3=W→δ−AW→3D→δ Where the position of the leaders wolfs at iteration itr are W→α, W→β, and W→δ. A→1, A→2, and A→3 are calculated using (27) and D→α, D→β, and D→δ are calculated using (34), (35), (36) [6,26].(34) D→α=\|CA→1.W→α−W→i\| (35) D→β=\|CA→2.W→β−W→i\| (36) D→δ=\|CA→3.W→δ−W→i\| Where C→1, C→2, and C→3 are calculated as in (28). Based on W→i(itr+1) as a new position of every wolf in population, the evaluation function will be implemented on every wolf using (26). Then, these steps continue until the maximum number of generations is reached. In the end, the algorithm terminates and the weights of features and the K value given in the best wolf W α will be used as the best values to implement the steps of the WKNN algorithm to diagnose monkeypox patients. The implementation steps of WKNN are similar to the traditional KNN method but have a different distance method in which the WKNN distance depends on the weights of features. Additionally, WKNN is based on a predefined value of K but KNN is based on undefined value. In the WKNN algorithm, each testing case is passed through many steps to be diagnosed. In step 1, the distance between each testing case and each training case is calculated by using Euclidean distance in weighted form as presented in (37).(37) Distance(TE,TR)=∑i=1qMi(TEi−TRi)2 Where Distance(TE,TR) is the distance between the testing case TE and the training case TR. q is the number of features in the filtered dataset, M i is the weight of ith feature, TE i is the value of ith feature at the testing case TE, and TR i is the value of ith feature at the training case TR. In step 2, the nearest k of neighbors which give the lowest distance between the testing case and everyone of training cases are assigned. In step 3, the diagnostics of k neighbors are used to determine the final diagnosis of the testing case by voting.Fig. 5 The steps of implementing WKNN algorithm.(26) Evaluation(Wi)=WKNN_Accuracy(Wi) Fig. 5 5.2.3 Deep learning algorithm In this segment, Long Short-Term Memory (LSTM) model as a deep learning structure used to diagnose monkeypox patients will be described in detail. LSTM is an evolution of Recurrent Neural Network (RNN) to solve the gradient vanishing and exploding problem by replacing the hidden vectors from RNN with memory cells equipped with gates [[20], [21], [22]]. Thus, LSTM represents a special type of RNN that has the ability to learn long-term dependencies. It also can by default remember information for long periods of time. Accordingly, LSTM is a popular deep learning tool because it has the ability to learn from sequential data [[20], [21], [22]]. LSTM is a sufficient model for several real-time applications such as sequence-to-sequence predictions, medical diagnosis, various tagging problems, language modeling, and classification of sentences. In this paper, the designed model is based on a many-to-one LSTM structure to handle multi-label diagnostics as shown in Fig. 6 .Fig. 6 A many-to-one LSTM structure for multi-label diagnostics. Fig. 6 According to Fig. 6, the input dataset that contains values of ‘p’ features is passed to ‘p’ LSTM cells where the cell state (c i ) and the current output state (h i) of ith LSTM are used as inputs for the next LSTM or (i+1) th LSTM. In other words, the outputs of each LSTM are used as inputs to the next LSTM and then the last LSTM gives the definitive diagnosis. Each LSTM cell consists of three gates called input, forget, and output gates used to update the output value and maintain the cell state as illustrated in Fig. 7 . These gates are intended to control the flow of information from one cell state to another. To give a decision to control the flow information, sigmoid activation (σ) is used by all three gates. In fact, information does not change in cell state but it can be added or omitted via each gate. The input values that should be used to change the cell state are determined by the input gate. The useless information that should be omitted from the cell state is determined by the forget gate while the amount of output is determined by the output gate [[20], [21], [22]].Fig. 7 The structure of LSTM cell. Fig. 7 To construct the LSTM, three main steps are required. Initially, LSTM begins to identify undesired information and then remove it from the cell by the forget gate. In the forget gate, current input (f i) and previous output (h i-1 ) in the cell state (c i-1 ) are used to give output (x i ) between zero and one. Completely forget the information is represented by one while completely retaining it is represented by zero. In the second step, a decision about storing information in the current cell state (c i ) is provided by the input gate by multiplying its output (t i ) with the output of tanh activation layer (ci˜). In the third and final step, the flow of fraction of information(h i ) in the current cell state (c i ) is provided at the output of LSTM cell by the output gate by combining its output (o i ) with the output of another tanh activation layer. The operation of these three gates in an LSTM cell for giving output (h i ) in cell state (c i ) can be mathematically represented using (38), (39), (40), (41), (42), (43) [[20], [21], [22]].(38) xi=σ((wxhi−1)+(wxfi)+bx) (39) ti=σ((wthi−1)+(wtfi)+bt) (40) ci˜=tanh((wchi−1)+(wcfi)+bc) (41) ci=(xici−1)+(tici˜) (42) oi=σ((wohi−1)+(wofi)+bo) (43) hi=oitanhci Where w x , w t , w c , and w o are the weight matrices. b x , b t , b c , and b o are bias factors for different gates of LSTM cell. 5.2.4 The weighted voting method During this segment, Confusion Based Voting (CBV) is provided as a new weighted voting method for combining the ensemble classifier (see Table 5). Based on the validation dataset, the Confusion Matrices (CMs) of the applied three classifier of the ensemble called WNB, WKNN, and Deep Learning are illustrated in tables (3, 4, and 5) respectively. The general accuracy for WNB, WKNN, Deep Learning are; 71%, 59%, and 35% as depicted from such figures. On the other hand, Table 6 presents the output of classifying a new case depending on the used three classifiers. Based on the majority voting, if class B gets two votes and class A votes gets one vote only, the target class will be class B. However, class A gets a weight of 0.92 and class B gets 0.55 + 0.32 = 0.87 based on CBV. Accordingly, the input case belongs to class A.Table 5 CM for deep learning. Table 5 Predicted Class Total A B Actual Class A 220 = 37% 380 600 B 270 130 = 32% 400 490 510 1000 Table 6 ED with confusion based voting. Table 6Classifier Predicted class Vote for Class A B WNB A 0.92 0.40 WKNN B 0.62 0.55 Deep Learning B 0.37 0.32 6 The description of experimental results Through this segment, the HMD strategy that includes SP and DP will be implemented to early detect monkeypox cases. The implementation of HMD starts with the implementation of the IBCO algorithm as a feature selection method in SP to determine a valuable set of features. Then, the valid dataset without useless features is passed to ED as a diagnostic model in DP to give fast and more accurate diagnosis. In fact, ED model is a hybrid model that includes three methods called WNB, WKNN, and LSTM as a Deep Learning technique which are combined using a new weighted voting method called CBV. During this implementation, the fitness function of BCO depends on using SVM [6], DLM [[20], [21], [22]], NB [9], and KNN [13] to calculate the average accuracy of them. Three main scenarios will be followed to implement the HMD strategy. During the first scenario, the proposed IBCO algorithm will be tested and compared to other modern selection algorithms using NB algorithm as a standard method [9]. In the second scenario, WNB, WKNN, LSTM, and the combined model called ED will be tested and their results will be compared. In the third scenario, the HMD strategy that includes both the proposed IBCO as a new feature selection method and ED model as a diagnostic method will be implemented and compared to other modern strategies. This work is based on the use of monkeypox dataset that classifies patients into two classes called “Positive” as an infected case and “Negative” as an uninfected case [27]. In fact, negative or uninfected case does not mean health case but he/she does not suffer from monkeypox infection but may suffer from other diseases or not. The performance of the used methods can be calculated by using recall, accuracy, and precision measurements based on confusion matrix as presented in Table 7 [6,7]. Various formulas of confusion matrix are summarized in Table 8 . The dataset is divided into 10 equal parts based on10-fold cross-validation. Training sets are represented in 9 of parts while a testing set is represented in the other part. Actually, 70% of the used dataset has been assigned as training data while 30% has been assigned as testing data. The values of used parameters are mentioned in Table 9 .Table 7 Confusion matrix which depicts how diagnostic on cases. Table 7 Diagnosed Label Positive Negative Known Label Positive True Positive (TP) False Negative (FN) Negative False Positive (FP) True Negative (TN) Table 8 Confusion matrix formulas. Table 8Measure Formula Meaning Precision TP/(TP + FP) The percentage of positive diagnostics those are already correct. Recall TP/(TP + FN) The percentage of positive diagnostics that were diagnosed as positive. Accuracy (TP + TN)/(TP + TN + FP + FN) The percentage of diagnostics those are correct. Macro-average ∑i=1cPi/c “for Precision” The average of the precision and recall of the system on different c classes. ∑i=1cRi/c “for Recall” Micro-average (TP1 + TP2)/(TP1 + TP2 + FP1 + FP2) “for Precision” the summation up to the individual true positives, false positives, and false negatives of the system for different classes and the apply them to get the statistics. (TP1 + TP2)/(TP1 + TP2 + FN1 + FN2) “for Recall” F1-measure 2PR/(P + R) The weighted harmonic mean of Precision and Recall. Table 9 The values of the used parameters. Table 9Parameter Description Applied value MT The maximum iterations number in BCO 100 Rand () Uniform distribution value in BCO Random (0 ≤ Rand () ≤ 1) Y The random value to choose between the chaotic model or the normal adjusting position method Random (0 ≤ y ≤ 1) K The closed number of neighbors 1 < K < 5 r1 and r2 Two independent random numbers Random (0 ≤ r1,r2 ≤ 1) aW Linearly decrease [2,0] M_itr The maximum number of iterations for GWO 100 6.1 The used hardware and software The proposed HMD strategy has been implemented using hardware and software tools. The used hardware tools are Dell machine with 8 GB RAM and 1 TB hard disk while the used software tools are windows operating system and MATLAB_R2021b_win64. Based on the MATLAB libraries, the implementation codes for the used techniques are represented in m-files. Thus, the components of HMD, which are; IBCO, WNB, WKNN, LSTM, and CBV were established as source codes in m-files. In MATLAB, the original versions of the used methods are available as open source m-files codes. During this work, these m-files were downloaded and then modified to be new versions provided in this paper. Initially, the used dataset in spreadsheet (Excel sheet) has been read in the MATLAB and then stored in a matrix (m-dimensional vector). This dataset has been entered into IBCO code that consists of two m-files where the first m-file contains the filter selection methods that passes their results as initial population values in the second m-file that contains BCO algorithm. The dataset was filtered from irrelevant features and only includes the selected features. The filtered dataset was passed to three m-files include WNB, WKNN, LSTM algorithms. Diagnosis results from these three algorithms were passed to CBV m-file to define the final diagnosis. 6.2 The monkeypox dataset description Monkeypox dataset is an internet data collected from 6-5–2022 to 19-9-2022 [27]. This dataset is a blood test dataset collected from patients of different ages and genders in different regions in different countries such as Nigeria, Spain, UK, etc. Monkeypox dataset contains 500 cases who were classified into two class categories called “Positive” and “Negative”. While the positive cases are patients with monkeypox, the negative cases are patients without monkeypox. In fact, negative case does not mean health case but he/she does not have monkeypox but may or may not have other diseases. This dataset was collected from patients suffering from different diseases, which are; monkeypox, acne, alopecia, normal, psoriasis, and small pox as shown in Fig. 8 that is a snapshot of monkeypox dataset. Monkeypox is a class category of positive cases but normal is a class category of negative cases who are health cases. On the other hand, acne, alopecia, normal, psoriasis, and small pox are classes of negative cases who are uninfected with monkeypox but suffer from other diseases. In fact, this dataset consists of 47 features that include demographic features and features of laboratory blood tests as presented in Table 10 . These features are used to describe causative conditions based on the blood test that gives each patient's status. In fact, the selected features after applying IBCO are 34. According to the 500 cases in the dataset, 296 of them had monkeypox as presented in Table 11 . In fact, monkeypox dataset was divided into 350 cases as a training set of data and 150 cases as a testing set of data.Fig. 8 A snapshot of monkeypox dataset. Fig. 8 Table 10 Descriptions about the features of Monkeypox. Table 10Feature Normal Range Selected Feature Age (>2) – Yes Sex (Male/Female) – No Transmission rank – Yes Country of acquisition – No Smallpox vaccination history – No HIV, hepatitis B, and hepatitis C status (Negative/Positive) Negative Yes Fever (Yes/No/None) No Yes Rectal pain or pain on defecation (Yes/No/None) No Yes Dysuria (Yes/No/None) No Yes Bleeding/discharge per rectum (Yes/No/None) No No Conjunctivitis (Yes/No/None) No Yes Oropharyngeal manifestations – Yes Back pain (Yes/No/None) No Yes Myalgia (Yes/No/None) No Yes headache (Yes/No//None) No Yes Sexually transmitted infections – No Lymphadenopathy (Yes/No/None) No Yes Approximate maximum number of concurrent lesions – Yes Distribution of lesions – Yes Complications of illness No Monkeypox viral DNA detected in Blood (Yes/No/None) No Yes Monkeypox viral DNA detected in Nose or throat swab (Yes/No/None) No Yes Monkeypox viral DNA detected in Urine (Yes/No/None) No Yes Antivirals received – No Day of illness treatment commenced – No Complications of treatment – No Duration of hospitalization with monkeypox (days) – No Sore throat (Yes/No/None) No Yes Chills (Yes/No/None) No Yes White Blood Cell (WBC) count, cells/mm3 400–9000 Yes Hematocrit, % For men (39–49) & For woman (35–45) Yes Platelet count 109 platelets/L. 150–400 Yes Sodium level, mmol/L 136–145 Yes Potassium level, mmol/L 3.5–5.0 Yes Blood urea nitrogen level, mg/dL 10–20 Yes Creatinine level, mg/dL <1.5 Yes Calcium level, mmol/L 9–10.5 Yes Total bilirubin level, mg/dL 0.3–1 Yes Aspartate aminotransferase (AST) level, U/L 0–35 Yes Alanine aminotransferase (ALT) level, U/L 0–35 Yes Alkaline phosphatase (ALP) level, U/L 40–140 Yes Arthralgia 6.7–15.8 Yes Albumin level, mg/dL 3.5–5.5 Yes Hospitalized (Yes/No/None) – No Date_confirmation – No Reverse transcription polymerase chain reaction (RT-PCR) (Yes/No/None) – Yes Outcome of monkeypox infection – No Table 11 Distribution of people in dataset based on infection. Table 11Criteria Value/Description Total number of cases Monkeypox Patients Normal People Cases with other Diseases 297 95 108 Type of other Diseases Acne Alopecia Psoriasis 18 9 15 small pox other 29 37 Sex Male Female Monkeypox 152 145 Normal 48 47 Other Diseases 57 51 6.3 Testing the Improved Binary Chimp Optimization (IBCO) algorithm Through this segment, IBCO will be executed as a new feature selection algorithm and compared to other modern selection methods to ensure its effectiveness in identifying valuable features in the monkeypox dataset. These selection methods are Genetic Algorithm (GA) [8], Improved Genetic Algorithm (IGA) [7], Adjusted Brain Storm Optimization (ABSO) algorithm [28], Hybrid Feature Selection Method (HFSM) [6], and BCO [12]. After implementing these feature selection methods, the NB is used as a diagnostic model to be trained on the filtered dataset using the selected features from each feature selection method separately and then it will be tested to calculate the diagnostic efficiency according to each selection method [9]. Accuracy, precision, and recall calculations are illustrated in Fig. 9, Fig. 10, Fig. 11 . Implementation time measurement also is provided in Fig. 12 . Actually, IBCO provides the best performance values, thus, it outperforms other methods.Fig. 9 Accuracy of selection techniques using NB. Fig. 9 Fig. 10 Precision of selection techniques using NB. Fig. 10 Fig. 11 Recall of selection techniques using NB. Fig. 11 Fig. 12 Implementation time of selection techniques using NB. Fig. 12 Fig. 9, Fig. 10, Fig. 11 show that IBCO outperforms GA, IGA, ABSO, HFSM, and BCO as it introduced 98.05% accuracy value that represents the maximum value at number of training data = 350. At the maximum number of training dataset, the accuracy values of GA, IGA, ABSO, HFSM, BCO, and IBCO are 62.4%, 65.65%, 76.05%, 83.74%, 90.1%, and 98.05% respectively. Based on these results, it is noted that GA gives the lowest accuracy value while IBCO outperforms all methods because it gives the highest accuracy. Additionally, IBCO provides the maximum precision and recall values equal 89.16% and 90.09% respectively. The precision values of GA, IGA, ABSO, HFSM, and BCO are 62.12%, 65%, 73%, 82.5%, and 84.06% respectively but their recall values are 58.2%, 64.99%, 78%, 86%, and 87.9% at the maximum number of training data. From these measurements, it is noted that GA provides the worst results while IBCO provides the best results. According to implementation time in Fig. 12, it is noted that IBCO takes a short execution time but GA takes a long execution time with values reach to 2.51 s and 8.2 s respectively at the number of training data = 350. In fact, BCO is faster than GA, IGA, ABSO, and HFSM but slower than IBCO as it takes 3.1 s to be executed. Hence, IBCO can determine valuable features that can accurately diagnose monkeypox patients. At the last, it is concluded that the performance of IBCO method is superior to GA, IGA, ABSO, HFSM, and BCO. Based on experimental results, the second best feature selection method after IBCO is BCO algorithm because it can provide the maximum accuracy, precision, and recall values and the minimum implementation time. Thus, IBCO can improve the diagnostic performance more than BCO because IBCO can solve two main problems of BCO which are the number of search agents in population and the initial values of each search agents are were randomly generated. This solution is provided by using FSL that contains ‘m’ of filter methods which give the BCO ‘m’ of search agents which include the selected features from filter methods as initial values of these ‘m’ agents. At the end, the selected features from GA = {Age, Transmission rank, Smallpox vaccination history,Fever, Dysuria, Myalgia, headache, Approximate maximum number of concurrent lesions, Monkeypox viral DNA detected in Blood, Day of illness treatment commenced, AST level, Duration of hospitalization with monkeypox, Outcome of monkeypox infection}. The selected features from IGA = {Transmission rank, HIV, hepatitis B, and hepatitis C status, Fever, Dysuria, Myalgia, headache, Monkeypox viral DNA detected in Blood, Day of illness treatment commenced, AST level, Hematocrit, Arthralgia, Outcome of monkeypox infection }. The selected features from ABSO = {Sex, Transmission rank, HIV, hepatitis B, and hepatitis C status, Dysuria, Bleeding/discharge per rectum, Myalgia, headache, Monkeypox viral DNA detected in Blood, Monkeypox viral DNA detected in Nose or throat swab, Day of illness treatment commenced, Hematocrit, AST level, Arthralgia, RT-PCR, Outcome of monkeypox infection}. The selected features from HFSM = {Age, Transmission rank, HIV, hepatitis B, and hepatitis C status, Dysuria, Bleeding/discharge per rectum, Myalgia, headache, Monkeypox viral DNA detected in Blood, Monkeypox viral DNA detected in Nose or throat swab, Oropharyngeal manifestations, Hematocrit, AST level, ALT level, RT-PCR, Outcome of monkeypox infection}. The selected features from BCO = {Age, Transmission rank, HIV, hepatitis B, and hepatitis C status, Dysuria, Bleeding/discharge per rectum, Monkeypox viral DNA detected in Blood, Monkeypox viral DNA detected in Nose or throat swab, Oropharyngeal manifestations, Sore throat, Chills, WBC counts, Hematocrit, Platelet count, Sodium level, Potassium level, AST level, ALT level, RT-PCR, Outcome of monkeypox infection}.The selected features from IBCO are presented in the last column in Table 10. Accordingly, monkeypox dataset with the best subset of features selected from IBCO will be passed to the next segments to train and then test the proposed ED model on the correct dataset without irrelevant features. 6.4 Testing the Ensemble Diagnosis (ED) model In this segment, ED model will be tested against its components which are WNB, WKNN, and deep learning algorithm called LSTM to ensure the effectiveness of the combined model called ED is higher than its component separately. At first, the monkeypox dataset after selecting the most significant features using IBCO will be divided into training, testing, and validation datasets. Then, WNB, WKNN, LSTM, and ED will be trained by using the same training dataset. After that, these algorithms will be tested by using the same testing dataset. Finally, these algorithms will be validated by using validation dataset to measure their performance metrics which are accuracy, precision, and recall as presented in Fig. 13, Fig. 14, Fig. 15 . Additionally, the implementation time measurement is provided in Fig. 16 . In fact, ED model gives the best results, hence, it is superior WNB, WKNN, and LSTM.Fig. 13 Accuracy of different diagnostic models. Fig. 13 Fig. 14 Precision of different diagnostic models. Fig. 14 Fig. 15 Recall of different diagnostic models. Fig. 15 Fig. 16 Implementation time of different diagnostic models. Fig. 16 According to figures (13 → 15), ED model outperforms WNB, WKNN, and LSTM with accuracy values reach to 98.48%, 64%, 66.01%, and 80.99% respectively at the maximum value at number of training data = 350. On the other hand, WNB algorithm introduced the lowest accuracy value while ED model outperforms all methods because it gives the maximum accuracy value. According to precision and recall measurements, ED model provides the maximum precision and recall values equal 91.1% and 88.91% respectively. WNB, WKNN, and LSTM algorithms provide precision values reach to 63.12%, 67.2%, and 80.04% respectively but their recall values are 58%, 66.5%, and 83.25% at the maximum number of training data. From these calculations, it is noted that WNB algorithm provides the worst results while ED model provides the best results. In Fig. 16, LSTM model takes a short execution time but ED takes a long execution time with values reach to 3.5 s and 5.4 s respectively at the maximum number of training data. In fact, WKNN algorithm is faster than WNB and ED but slower than LSTM model as it takes 4.25 s to be executed but WNB takes 5.2 s. It is noted that ED model requires a large diagnostic time because it has to wait for the decision-making time of the three classifiers. Therefore, the time of ED diagnosis is the highest time taken from the three classifiers, in addition to the voting time. However, this time spent does not affect the efficiency of the diagnostic system, as the goal of diagnostic systems is the efficiency of diagnosis and not the time taken for diagnosis, in addition to the combined time for the three classifiers is small and can be neglected in the case of diagnosis. Hence, ED model can accurately diagnose monkeypox patients because it combines three different algorithms to insure the maximum diagnose accuracy. These three algorithms are WNB as a probabilistic classifier, WKNN which combined a distance based classifier (traditional KNN) with one of the most effective bio-inspired optimization technique, which is GWO, and LSTM which the most recently used machine learning method and introduces excellent results. Finally, it is concluded that the performance of ED model is superior to WNB, WKNN, and LSTM. In the next segment, the proposed HMD strategy that includes IBCO to select the best subset of features and ED model as a diagnostic model to provide accurate diagnosis will be tested against many recent strategies. 6.5 Testing the human monkeypox detection (HMD) strategy In this segment, HMD strategy will be executed and compared to other modern diagnostic strategies to ensure that HMD can provide fast and accurate diagnosis. These strategies are RF [9], Neuro-Fuzzy [1], ELGA [8], DBC [6], CPE [7], and Ensemble Diagnosis Strategy (EDS) [29]. In fact, HMD strategy takes many steps to be implemented where it begins with implement IBCO as a new feature selection method. After that, ED model is applied depending on valid data without useless features to give a quick and correct results. Based on Table 8, accuracy, precision, recall, micro-average precision, micro-average recall, macro-average precision, macro-average recall, and F1-measure calculations are illustrated in figures (17 →24). In fact, micro-average precision, micro-average recall, macro-average precision, macro-average recall, and F1-measure will be used to measure the performance of algorithms based on unbalanced data. The implementation time measurement also is provided in Fig. 25. At the end, the diagnosing time according to testing dataset is illustrated in Fig. 26 to prove the computational efficiency of the proposed HMD strategy against other strategies. Actually, HMD provides the best performance values, thus, it outperforms other strategies.Fig. 17 Accuracy of monkeypox diagnostic strategies. Fig. 17 Fig. 17, Fig. 18, Fig. 19, Fig. 20, Fig. 21, Fig. 22, Fig. 23, Fig. 24 show that HMD outperforms RF, Neuro-Fuzzy, ELGA, DBC, CPE, and EDS as it introduced the best results at the number of training data = 350. In Fig. 17, HMD algorithm provides the maximum accuracy value while RF provides the minimum value with values 98.48% and 85.26% respectively at the maximum number of training data. Additionally, the accuracy values of Neuro-Fuzzy, ELGA, DBC, CPE, and EDS are 88.12%, 90.9%, 92.3%, 94.25%, and 95% respectively. According to precision and recall results in Fig. 18, Fig. 19 , it is noted that HMD introduces the maximum precision and recall values reach to 91.1% and 88.91% respectively at the number of training data = 350. In fact, RF, Neuro-Fuzzy, ELGA, DBC, CPE, and EDS provide precision values reach to 62.5%, 64.5%, 70%, 75.25%, 83%, and 90% respectively at the number of training data = 350. On the other hand, these strategies provide recall values in the same order reach to 61%, 65%, 68.01%, 73.02%, 80.05%, and 87% respectively. From these measurements, it is noted that RF provides the worst results while HMD provides the best results. The reason is that RF is implemented on the original dataset without selecting informative features before starting to be learned but HMD begins with selecting the best subset of features using IBCO before learning the diagnostic model. Additionally, EDS can provide the best results after HMD.Fig. 18 Precision of monkeypox diagnostic strategies. Fig. 18 Fig. 19 Recall of monkeypox diagnostic strategies. Fig. 19 In Fig. 20, Fig. 21, Fig. 22, Fig. 23, Fig. 24 , micro-average, macro-average, and F1-measure are measured to test the ability of diagnostic strategies to handle unbalanced data and provide the best performance. As presented in Fig. 20, HMD gives the maximum micro-average precision value but RF gives the minimum value with values 92.56% and 59.99% respectively. The micro-average precision of Neuro-Fuzzy, ELGA, DBC, CPE, and EDS are 68.9%, 75.25%, 79.85%, 83.65%, and 90.32% respectively at the maximum number of training data. Micro-average recall values in Fig. 21 are 60.85%, 62.6%, 65.8%, 75%, 80.25%, 85.9%, and 89.01% for RF, Neuro-Fuzzy, ELGA, DBC, CPE, EDS, and HMD respectively at the number of training data = 350. Thus, RF provides the minimum micro-average recall value while HMD provides the maximum value. Fig. 22 shows that the maximum macro-average precision is provided by HMD while the minimum value is provided by RF with values reach to 88.01% and 62% respectively at the number of training data = 350. The macro-average precision of Neuro-Fuzzy, ELGA, DBC, CPE, and EDS reach to 66.01%, 68.5%, 77.25%, 80.65%, and 83.6% respectively. Hence, the best macro-average precision is provided by HMD but the worst value is provided by RF. The macro-average recall of HMD is the maximum value but the minimum value is given by RF at the number of training data = 350 as shown in Fig. 23 . The macro-average recall of RF, Neuro-Fuzzy, ELGA, DBC, CPE, EDS, and HMD are 54%, 59.5%, 67.5%, 72.85%, 77.68%, 80% and 85.01% respectively.Fig. 20 Micro_average precision of monkeypox diagnostic strategies. Fig. 20 Fig. 21 Micro_average recall of monkeypox diagnostic strategies. Fig. 21 Fig. 22 Macro_average precision of monkeypox diagnostic strategies. Fig. 22 Fig. 23 Macro_average recall of monkeypox diagnostic strategies. Fig. 23 In Fig. 24 , F1-measure of RF, Neuro-Fuzzy, ELGA, DBC, CPE, EDS, and HMD are 64%, 68.85%, 71.85%, 76.9%, 80.65%, 82%, and 83.9% respectively at the maximum number of training data. Based on Fig. 17, Fig. 18, Fig. 19, Fig. 20, Fig. 21, Fig. 22, Fig. 23, Fig. 24, the proposed HMD strategy outperforms other diagnostic strategies because it provides the maximum accuracy, precision, recall, micro-average, macro-average, and F1-measure values. In fact, it is noted that EDS outperforms other strategies after HMD strategy. As shown in Fig. 25, the implementation time of HMD takes a long execution time but RF takes a short execution time with values reach to 5.4 s and 3 s respectively at the number of training data = 350. As presented in Fig. 26, the diagnosing time of HMD is larger than other strategies. The diagnosing time of RF, Neuro-Fuzzy, ELGA, DBC, CPE, EDS, and HMD are 1.8, 2.7, 2.9, 3, 3.2, 3.2, and 3.24 s respectively at the maximum number of testing data = 150. Accordingly, EDS can provide the best results and takes a long execution time after HMD strategy. Hence, HMD can accurately diagnose monkeypox patients which consumes a long execution time but this time was neglected compared to accurate diagnosis. At the last, it is concluded that the performance of HMD method is superior to RF, Neuro-Fuzzy, ELGA, DBC, CPE, and EDS.Fig. 24 F1-measure of monkeypox diagnostic strategies. Fig. 24 Fig. 25 Implementation time of monkeypox diagnostic strategies. Fig. 25 Fig. 26 Diagnosing time of diagnostic strategies. Fig. 26 7 Conclusions and future directions The main core of this paper is to provide a robust strategy using AI techniques to accurately detect monkeypox patients to limit the spread of the virus. Hence, Human Monkeypox Detection (HMD) strategy has been introduced for early detection of infected persons quickly and accurately. This strategy includes two main phases called Selection Phase (SP) and Detection Phase (DP). In SP, monkeypox dataset has been filtered from useless features using Improved Binary Chimp Optimization (IBCO) algorithm that combines two layers called Filter Selection Layer (FSL) as a quick layer and Wrapper Selection Layer (WSL) as an accurate layer. After eliminating irrelevant features as possible in FSL using many filter methods, sets of features provided by these filter methods have been passed to WSL to accurately choose the useful features. The filtered dataset without any irrelevant features has been passed to DP to correctly learn Ensemble Diagnosis (ED) model to accurately diagnose monkeypox patients. ED model is a hybrid model that consists of Weighted Naïve Bayes (WNB), Weighted K-Nearest Neighbors (WKNN), and deep learning which are combined using a new weighted voting method to introduce the best diagnostic results. Experimental results illustrated that the suggested IBCO as a new feature selection outperformed other selection techniques using NB algorithm as a standard diagnostic model. Additionally, the HMD strategy gives the best measurements compared to other strategies in terms of accuracy, precision, recall, micro-average, macro-average, F1-measure, implementation time, and diagnosing time. The HMD strategy provided 98.48%, 91.1% and 88.91% for accuracy, precision, and recall values at the number of training data = 350. Additionally, the micro-average precision, micro-average recall, macro-average precision, macro-average recall, F1-measure, and implementation time of HMD strategy are 92.56%, 89.01%, 88.01%, 85.01%, 83.9%, and 5.4 s respectively at the number of training data = 350. At the number of testing data = 150, diagnosing time of HMD is 3.24 s. Thus, it is concluded that the HMD is superior other strategies because it provided the maximum accuracy. On the other hand, HMD provided the maximum execution time but this time was neglected compared to accurate diagnosis. According to future directions, the proposed HMD strategy should be tested on a large dataset and also should be tested on different datasets. Outlier rejection layer should be added to the proposed HMD strategy to reject noise data before learning ED model for improving the performance of this strategy. Declaration of competing interest The authors declare that they have no conflict of interest. ‘‘This paper does not contain any studies with human participants or animals performed by any of the authors”. ==== Refs References 1 Tom J. Anebo N. A neuro-fussy based model for diagnosis of monkeypox diseases International Journal of Computer Science Trends and Technology (IJCST) 6 Issue 2 2018 143 153 2 CDC, Centers for Disease Control and Prevention Monkeypox [cited 2021 July 16]. Available at: https://www.cdc.gov/poxvirus/monkeypox/symptoms.html 2015 3 CDC, Centers for Disease Control and Prevention Monkeypox [cited 2022 May 27]. Available at: https://www.cdc.gov/poxvirus/monkeypox/outbreak/us-outbreaks.html 2015 4 World Health Organization (WHO) Monkeypox [cited 2022 May 19]. Available at: https://www.who.int/news-room/fact-sheets/detail/monkeypox 2019 5 Oladoye M. Monkeypox: a neglected viral zoonotic disease European Journal of Medical and Educational Technologies 14 Issue 2 2021 1 6 6 Rabie A. Saleh A. Mansour N. A Covid-19's integrated herd immunity (CIHI) based on classifying people vulnerability Comput. Biol. Med. 140 2022 1 29 Elsevier 7 Rabie A. Mansour N. Saleh A. Expecting individuals' body reaction to Covid-19 based on statistical Naïve Bayes technique Pattern Recogn. 128 2022 1 23 Elsevier 8 Abdollahi J. Nouri-Moghaddam B. A hybrid method for heart disease diagnosis utilizing feature selection based ensemble classifier model generation Iran Journal of Computer Science 2022 1 18 10.1007/s42044-022-00104-x Springer 9 Edeh M. Khalaf O. Tavera C. A classification algorithm-based hybrid diabetes prediction model Front. Public Health 10 2022 1 7 10 Rabie A. Ali S. Ali H. Saleh A. A fog based load forecasting strategy for smart grids using big electrical data Cluster Comput. 22 Issue 1 2019 241 270 Springer 11 Shamshirband S. Fathi M. Dehzangi A. A review on deep learning approaches in healthcare systems: taxonomies, challenges, and open issues J. Biomed. Inf. 113 2021 1 17 Elsevier 12 Pashaei E. Pashaei E. An efficient binary chimp optimization algorithm for feature selection in biomedical data classification Neural Comput. Appl. 34 2022 6427 6451 Springer 13 Rabie A. Ali S. Saleh A. Ali H. A fog based load forecasting strategy based on multi-ensemble classification for smart grids J. Ambient Intell. Hum. Comput. 11 Issue 1 2020 209 236 Springer 14 Khishe M. Mosavi M. Chimp optimization algorithm Expert Systems with Applications vol. 149 2020 Elsevier 1 26 15 Levine R. Peterson A. Yorita K. Ecological niche and geographic distribution of human monkeypox in Africa PLoS One 2 Issue 1 2007 1 7 16 Patel A. Bilinska J. Tam J.C.H. Clinical features and novel presentations of human monkeypox in a central London centre during the 2022 outbreak: descriptive case series BMJ 2022 1 17 10.1136/bmj-2022-072410 17 Girometti N. Byrne R. Bracchi M. Demographic and clinical characteristics of confirmed human monkeypox virus cases in individuals attending a sexual health centre in London, UK: an observational analysis Lancet Infect. Dis. 22 Issue 9 2022 1321 1328 Elsevier 35785793 18 Adler H. Gould S. Hine P. Clinical features and management of human monkeypox: a retrospective observational study in the UK Lancet Infect. Dis. 22 Issue 8 2022 1153 1162 Elsevier 35623380 19 Thornhill J. Barkati S. Walmsley S. Monkeypox virus infection in humans across 16 countries — April–June 2022 N. Engl. J. Med. 387 Issue 8 2022 679 691 35866746 20 Yang J. Kim J. An accident diagnosis algorithm using long short-term memory Nucl. Eng. Technol. 50 Issue 4 2018 582 588 Elsevier 21 Le X. Ho H. Lee G. Jung S. Application of long short-term memory (LSTM) neural network for flood forecasting Water 11 Issue 7 2019 1 19 22 Wu X. Wang H. Shi P. Long short-term memory model – a deep learning approach for medical data with irregularity in cancer predication with tumor markers Comput. Biol. Med. 144 2022 1 10 Elsevier 23 Bhatia S. Malhotra J. Naïve Bayes classifier for predicting the novel coronavirus Proceedings of the 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV) 2021 IEEE Tirunelveli, India 880 883 24 Zhang H. Jiang L. Yu L. “Attribute and instance weighted naive Bayes Pattern Recogn. 111 2021 1 11 Elsevier 25 Singh D. Singh B. Feature wise normalization: an effective way of normalizing data Pattern Recognition vol. 122 2022 1 14 Elsevier 26 Mirjalili S. Mirjalili S. Lewis A. Grey wolf optimizer Advances in Engineering Software vol. 69 2014 46 61 Elsevier 27 http://covid19.nilehi.edu.eg/Available_datasets.php 28 Tuba E. Strumberger I. Bezdan T. “Classification and feature selection Method for medical Datasets by Brain Storm optimization Algorithm and Support vector machine Procedia Computer Science vol. 162 2019 Elsevier 307 315 29 Nagavelli U. Samanta D. Chakraborty P. Machine learning technology-based heart disease detection models Journal of Healthcare Engineering 2022 1 9 Hindawi 2022	36481764	PMC9715266	NO-CC CODE	2022-12-05 23:15:21	no	Comput Biol Med. 2023 Jan 2; 152:106383	utf-8	Comput Biol Med	2,022	10.1016/j.compbiomed.2022.106383	oa_other
==== Front Ir J Med Sci Ir J Med Sci Irish Journal of Medical Science 0021-1265 1863-4362 Springer International Publishing Cham 36456718 3238 10.1007/s11845-022-03238-w Original Article Agreement between tele-assessment and face-to-face assessment of 30-s sit-to-stand test in patients with type 2 diabetes mellitus http://orcid.org/0000-0002-3327-461X Aktan Rıdvan [email protected] 1 http://orcid.org/0000-0002-1151-7190 Yılmaz Hayriye 2 http://orcid.org/0000-0001-7787-1443 Demir İsmail 3 http://orcid.org/0000-0002-5528-1036 Özalevli Sevgi 4 1 grid.411796.c 0000 0001 0213 6380 Department of Physiotherapy, Izmir University of Economics, Vocational School of Health Services, Sakarya St. No: 156, 35330 Balcova, Izmir, Turkey 2 grid.414879.7 0000 0004 0415 690X Department of Physical Therapy and Rehabilitation, Health Sciences University İzmir Bozyaka Training and Research Hospital, İzmir, Turkey 3 grid.414879.7 0000 0004 0415 690X Department of Internal Medicine, Health Sciences University İzmir Bozyaka Training and Research Hospital, İzmir, Turkey 4 grid.21200.31 0000 0001 2183 9022 Faculty of Physical Therapy and Rehabilitation, Dokuz Eylül University, İzmir, Turkey 2 12 2022 16 11 11 2022 21 11 2022 © The Author(s), under exclusive licence to Royal Academy of Medicine in Ireland 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 The globalization of healthcare systems, and the aim to lower healthcare costs have all contributed to the growth of telehealth technology in recent years. However, before these systems are put into use, their efficacy should be verified. To the best of our knowledge, this is the first study focusing on the evaluation of functional exercise capacity using the 30-s sit-to-stand (30-s STS) test as a tele-assessment method in patients with type 2 diabetes mellitus (T2DM). Aims The purpose of the study is to investigate the level of agreement between tele-assessment and face-to-face assessment of 30-s STS test in patients with T2DM. Methods Fifty participants performed two times 30-s STS tests separated by 1 h: a face-to-face and an Internet-connected video call examination (tele-assessment). Two physiotherapists conduct these evaluations; each was blinded to the other. The order of the evaluations was designated at random for each participant and physiotherapist. Results There was a good level of agreement between tele-assessment and face-to-face assessment of the 30-s STS test (mean differences = 0.20 ± 0.88, limits of agreement = 1.93 to − 1.53). Excellent interrater reliability was found for scores of the 30-s STS test [ICC = 0.93 (95% CI: 0.88; 0.96)]. In addition, all before and after test parameters show that there was a very good interrater reliability (ρ ≥ 0.75). Conclusions This study shows a good level of agreement between tele-assessment and face-to-face assessment of the 30-s STS test. Our study’s findings indicate that tele-assessment is a potential application to determine the level of physical capacity remotely in patients with T2DM. Keywords Physical assessment Tele-assessment Telehealth Type 2 diabetes 30-s sit-to-stand ==== Body pmcIntroduction Type 2 diabetes mellitus (T2DM) is caused by hyperglycemia brought on by reduced insulin synthesis, impaired insulin utilization, or both, with symptoms such as thirst, polyuria, and weight loss [1]. Because of the high prevalence and high medical costs worldwide, research is continuing in both the diagnosis and treatment of the disease and important developments are being experienced [2]. Maintaining sufficient glycemic control and lowering cardiovascular risk factors is part of the therapy of T2DM, along with the treatment of complications and related comorbidities [3]. Many systemic complications such as cardiovascular, cognitive, psychological, and neuropathy can occur in T2DM [4]. One of the keys to the management of the disease is maintaining a healthy lifestyle that includes maintaining a healthy weight, eating a balanced diet, quitting smoking, and engaging in physical activity [5]. Glycemic control is enhanced by physical activity, which reduces the risk of cardiovascular disease and also reduces mortality [6]. However, patients with T2DM often do not have adequate physical activity habits that may help glycemic balance [6–8]. Physical activity has psychological benefits as well as physical benefits and reduces depressive symptoms [9, 10]. Therefore, the physical conditions of T2DM patients need to be assessed to recommend individualized exercise programs and follow up on their progression by health professionals [7]. However, few studies have investigated the reliability of the 30-s sit-to-stand (30-s STS) test for assessing functional exercise capacity in the T2DM population [11, 12]. The 30-s STS test involves standing up and sitting down from a regular chair as many times as a person can in 30 s, and its validity and reliability have been proven in patients with T2DM [11, 12]. It is commonly used to assess physical performance in the lower extremities and to provide insight into functional exercise capacity [13–15]. The 30-s STS test has many advantages in that the materials required to perform this test are widely used, easy to use, and inexpensive. It also does not require professional personnel to perform the test. The test is very fast as it requires no more than 1 min per assessment. It is also a type of test used in many studies without a ground effect [16]. Nowadays, new options for clinical evaluation are provided by information and communication technology. In particular, it can benefit from Internet-based systems for the evaluation and follow-up of patients with chronic diseases who reside in distant places or lose their independence. Thus, accessibility to health services increases and the economic burden of the disease decreases [17]. However, scientific evidence is needed before recommending their use. Instead of traditional face-to-face evaluation, tele-assessment methods can be used with video conferencing method with any device with a camera, screen, microphone, speaker, and Internet access. To the best of our knowledge, there is no previous study focusing on the evaluation of the functional exercise capacity of individuals with T2DM as a tele-assessment method has been found. The aim of this study is to investigate the level of agreement between tele-assessment and face-to-face assessment of the 30-s STS test in patients with T2DM. Methods Study design and participants The interrater reliability was tested using a descriptive crossover design. T2DM patients, who consult the Internal Medicine Department of Health Sciences University—İzmir Bozyaka Training and Research Hospital, for routine outpatient follow-up and were diagnosed by an internal diseases specialist were included in this study. Inclusion criteria were having been diagnosed with T2DM by an internal diseases specialist (or endocrinologist), being able to walk independently, having a smartphone and the ability to use it, and having a pulse oximeter device (for O2 saturation measurement) and a blood pressure monitor. Exclusion criteria were having a health problem (orthopedic, neurological, internal, or cardiorespiratory) that prevents standing/walking; acute inflammation; intestinal tumor; cognitive impairment; vision and hearing loss; and/or an orthopedic, vascular, neurological, or psychiatric problem affecting balance. Assessment procedure The 30-s STS test was taught once when participants came for the routine control appointment. Participants’ demographic and clinic details (age, gender, educational status, height, body weight, and comorbidities) were recorded after receiving consent for the study. Participants perform two physical tests: a traditional in-person examination (face-to-face) and an Internet-connected video call examination (tele-assessment). Two physiotherapists with an experience of more than 7 years did these evaluations; one physiotherapist always performed the face-to-face assessment while the other always performed the tele-assessment, and each was blinded to the other. The two evaluations were separated by 20 min, and to counterbalance any testing order effects, the order of the evaluations was designated at random for each participant and physiotherapist [18]. Thirty-second STS tests were conducted under the same time frame from 11.00 am to 3.00 pm, Monday to Friday. Participants were informed not to do any vigorous physical activities 3 h before testing and to continue taking their prescription medications. The 30-s STS is a test consisting of getting up from a standard chair (with 45–47 cm seat height) and sitting down, and the participants were instructed to rise up straight and sit down again as many times as they could in 30 s starting from the seated position [13–15]. A smartphone with a built-in microphone and camera and high-speed Internet were used for the tele-assessment. Using the WhatsApp application (WhatsApp Inc., Mountain View, CA), a live video communication was kept running between the patient and the therapist. WhatsApp is an application that has been used frequently in recent years with the coronavirus pandemic and is a promising application as a communication tool between healthcare professionals and patients [19]. When the patient was performed to tele-assessment, he was required to position the smartphone so that all his/her limbs were visible from the camera. Before and after the tests, dyspnea perception and lower limb fatigue were questioned according to the 10-grade modified Borg scale [20], and peripheral oxygen saturation (SpO2), systolic/diastolic blood pressure, and heart rate were measured and recorded. While the perception of dyspnea and fatigue was questioned in the tele-assessment, oxygen saturation and blood pressure measurements were obtained by the patient by self-measurement. The patient’s devices were used in both tele-assessment and face-to-face assessment to eliminate measurement tool differences. Statistical analysis On the recommendation of the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN), a total sample size of 50 participants was selected [21]. The Bland–Altman limits of agreement test were used to investigate the agreement between tele-assessment and face-to-face assessment methods for the 30-s STS test. Additionally, two-way random effect intraclass correlation coefficients (ICC) (ρ) and confidence intervals of the parameters were calculated for the interrater reliability analysis. SPSS Version 21 for Windows (SPSS Inc., Chicago, IL) was used to perform all of the statistical analysis, with a 0.05 α value. The ICC was considered excellent (≥ 0.75), fair to good (0.4–0.75), or poor (≤ 0.4) [22]. Results Fifty-eight of the patients were assessed for eligibility to participate in the study. Four did not meet the inclusion criteria and four did not join in some of the assessment stages. The study was completed with a total of 50 patients (Fig. 1).Fig. 1 Recruitment flowchart The mean + SD age and BMI of patients were 54.5 ± 6.3 years and 28.8 ± 3.5 kg/m2, respectively. Sixty-two percent of the patients were male and 38% female. Table 1 presents the descriptive characteristics of patients.Table 1 Descriptive characteristics of patients Parameters Mean ± SD Gender (% male/female) 62.0/38.0 Age (years) 54.5 ± 6.3 BMI (kg/m2) 28.8 ± 3.5 Educational status, n (%) Primary school 3 (6.0) Secondary school 33 (66.0) High school 2 (4.0) University 10 (20.0) Postgraduate 2 (4.0) 30-s sit-to-stand test score (face-to-face) 12.4 ± 1.83 30-s sit-to-stand test score (tele-assessment) 12.2 ± 1.62 Comorbidity, n (%) HT 25 (50.0) COPD 2 (4.0) CAD 6 (12.0) HF 0 SD Standart Deviation, BMI Body Mass Index, HT Hypertension, COPD Chronic Obstructive Pulmonary Disease, CAD Coronary Artery Disease, HF Heart Failure In Fig. 2, the Bland–Altman plot illustrated a good level of agreement between tele-assessment and face-to-face assessment for the 30-s STS, with no evidence of systematic bias. The mean of the differences was 0.20 ± 0.88. The limits of agreement were 1.93 and − 1.53, respectively. Excellent interrater reliability was found for scores of the 30-s STS test [ICC = 0.93 (95% CI: 0.88; 0.96)] (Table 2).Fig. 2 Bland–Altman plot of the difference between tele-assessment and face-to-face assessment for the 30-s sit-to-stand test. The Y axis shows the difference between the two paired measurements and the X axis represents the average of these measures. The solid line represents the mean difference between the face-to-face and tele-assessment. Dotted lines are the upper and lower limits of agreement (UL and LL, respectively) Table 2 Interrater reliability between tele-assessment and face-to-face assessment of 30-s sit-to-stand test parameters Mean difference between assessments (95% CI) Interrater reliability ICC (95% CI) 30-s sit-to-stand test score (repetitions) 0.20 (− 0.05, 0.45) 0.93 (0.88, 0.96) Resting dyspnea (mBorg) 0.14 (0.03, 0.26) 0.92 (0.85, 0.95) Resting lower limb fatigue (mBorg) 0.14 (− 0.001, 0.28) 0.85 (0.74, 0.92) Resting heart rate (bpm) 0.66 (− 0.62, 1.94) 0.75 (0.56, 0.86) Resting systolic blood pressure (mmHg) 2.16 (− 1.30, 5.62) 0.89 (0.81, 0.94) Resting diastolic blood pressure (mmHg) 1.12 (− 1,24, 3.48) 0.80 (0.64, 0.88) Resting SpO2 (%) − 0.12 (− 0.36, 0.12) 0.79 (0.63, 0.88) End-dyspnea (mBorg) 0.06 (− 0.12, 0.24) 0.85 (0.74, 0.92) End-lower limb fatigue (mBorg) 0.04 (− 0.14, .022) 0.86 (0.76, 0.92) End-heart rate (bpm) 2.06 (0.68, 3.43) 0.76 (0.57, 0.86) End-systolic blood pressure (mmHg) 2.16 (− 1.30, 5.62) 0.84 (0.72, 0.91) End-diastolic blood pressure (mmHg) 0.12 (− 1.80, 2.04) 0.88 (0.79, 0.93) End-SpO2 (%) − 0.04 (− 0.28, 0.20) 0.75 (0.57, 0.86) CI Confidence Interval, ICC Intraclass Correlation Coefficient, SpO2 Peripheral Oxygen Saturation, mBorg Modified Borg Score In Table 2, the mean differences between tele-assessment and face-to-face assessment for the 30-s STS parameters and the interrater reliability values are shown. Interrater reliability was deemed “acceptable” if the score was ≥ 0.75 [22]. All parameters show that there was a very good interrater reliability (ρ ≥ 0.75). Additionally, most of the 95% CI were narrow. Discussion The main finding of this study is that the 30-s STS test tele-assessment can be successfully used to assess functional exercise capacity in patients with T2DM. Our results show that there is high interrater reliability between tele-assessment and the face-to-face test methods. The COVID-19 epidemic, the globalization of healthcare systems, and the aim to lower healthcare costs have all contributed to the growth of telehealth technology in recent years [18, 23]. However, before these systems are put into use, their efficacy should be verified against that of the more traditional health systems. Technological developments in the last decade have increased telehealth opportunities and provided greater flexibility in accessing health services [24]. In this study, we used the WhatsApp application, which can be used freely just that it needed an Internet-connected smartphone or tablet, for performing tele-assessment. WhatsApp is seen as a simple, inexpensive, and effective communication tool in the clinical health sector and its use will increase [25]. The security policy of WhatsApp is one of the important issues for its potential widespread usage in patient evaluation. WhatsApp uses end-to-end encryption for its services. It means that data and calls/video calls are encrypted and secured against third parties and wrong hands, even WhatsApp developers and authorities [26]. The 30-s STS test is often used for the evaluation of physical performance in adults because it is a simple and easy-to-use test [27]. The Centers for Disease Control recommend it as one of the tests for fall-risk screening [15]. The first reliability findings of the 30-s STS test were shown by the authors who developed it [28]. They found a high ICC for men and an excellent ICC for women. The 30-s STS reliability has been investigated in various populations. It was found an excellent ICC for people with total hip arthroplasty [29], a high ICC for people with dementia [30], excellent results in people who suffered a stroke [31], and an excellent ICC for people with multiple sclerosis [32]. To the best of our knowledge, only two studies researched on the reliability of the 30-s STS test in people with T2DM [11, 12]. Relative reliability for the 30-s STS test was excellent in both studies (for both, ICC > 0.90). In addition, Ogawa et al. [33] examined the reliability of virtual physical performance assessments in a group consisting of various diseases, and in this study, they found an excellent ICC for the 30-s STS test performed remotely. However, we think that there is a heterogeneous population in the mentioned study and that patients’ self-reporting of the diagnosis was a weakness of the study. In our study, we investigated the level of agreement between tele-assessment and face-to-face assessment of the 30-s STS test in patients physician-diagnosed with T2DM. We found excellent interrater reliability (ICC = 0.93) for the 30-s STS test of T2DM patients. When tele-assessment studies are investigated, it is shown that different methods are used in various groups [34–37]. In a systematic review, it was shown that the evaluation of some functional and neurodynamic tests with the tele-assessment method is applicable [35]. In a recent study, Güngör et al. [34] showed that functional and core strength-endurance tests assessed using the tele-assessment approach were valid and practicable for assessing the capabilities of healthy young adults. Another recent review emphasized the fact that while tele-assessments are useful for monitoring the progression of neuromuscular illnesses, they still require improvement [38]. In our study, we showed that the 30-s STS test tele-assessment method used in the measurement of functional capacity in patients with type 2 diabetic patients is easy to use, applicable, and reliable. The validity of the measurement of the vital signs before and after the test are crucial outcomes for the tele-assessment procedure. Kagiyama et al. [39] investigated the feasibility and validity of a tele-assessment in which healthcare providers monitor vital signs remotely. They demonstrated that telemedicine-based self-assessment of vital signs was feasible and reliable; additionally, it is a useful alternative to traditional face-to-face vital sign measurements. In their study, the agreement between tele-assessment and face-to-face assessment was excellent for all parameters. The ICC for systolic (0.92), diastolic blood pressure (0.86), heart rate (0.89), and SpO2 (0.92) were statistically significant [39]. In our study, we also found excellent agreement between tele-assessment and face-to-face assessment for vital signs both before and after the tests. Strengths and limitations Being the first study examining the interrater reliability, validity, and feasibility of a 30-s STS tele-assessment test in patients with T2DM, and having a sufficient sample according to COSMIN [21], can be listed as the strengths of our study. We have some limitations in our study. Our study was limited to T2DM who are more familiar with the Internet and be able to use technology. Therefore, we cannot generalize the findings of the study to all T2DM patients. Future research should assess the effectiveness of a complementing home-based treatment centered on functional capacity tele-assessment methods with a minimal clinical difference in light of our findings. Conclusions This study shows a good level of agreement between tele-assessment and face-to-face assessment for the 30-s STS test. Additionally, the 30-s STS test tele-assessment to assess functional exercise capacity in patients with T2DM was found to have high interrater reliability. The study’s findings indicate that tele-assessment is a potential application to conduct physical evaluations remotely. Author contribution All authors contributed to the study conception and design. Material preparation: Rıdvan Aktan, Hayriye Yılmaz, and İsmail Demir. Data collection: Rıdvan Aktan and Hayriye Yılmaz. Analysis of data: Rıdvan Aktan and Sevgi Özalevli. The first draft of the manuscript was written by Rıdvan Aktan and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Data availability On request. Code availability Not applicable. Declarations Ethics approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the local Ethical Committee of the Health Sciences University Izmir Bozyaka Training and Research Hospital with approval number 2022–33 on 23/02/2022, and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Consent to participate Informed consent was obtained from all individual participants included in the study. 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 1. American Diabetes Association. Summary of Revisions: Standards of Medical Care in Diabetes (2021) Diabetes Care 44 (Suppl 1):S4-S6. 10.2337/dc21-Srev 2. Ogurtsova K da Rocha Fernandes JD Huang Y IDF Diabetes Atlas: global estimates for the prevalence of diabetes for 2015 and 2040 Diabetes Res Clin Pract 2017 128 40 50 10.1016/j.diabres.2017.03.024 28437734 3. Reyes-García R Moreno-Pérez Ó Tejera-Pérez C Document on a comprehensive approach to type 2 diabetes mellitus Endocrinol Diabetes Nutr (Engl Ed) 2019 66 443 458 10.1016/j.endinu.2018.10.010 30827909 4. Alberti KG, Zimmet PZ (1998) Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus provisional report of a WHO consultation. Diabet Med 15 (7):539–553. 10.1002/(SICI)1096-9136(199807)15:73.0.CO;2-S 5. Diabetes Association Of The Republic Of China (Taiwan) (2020) Executive summary of the DAROC clinical practice guidelines for diabetes care- 2018. J Formos Med Assoc 119:577–586. 10.1016/j.jfma.2019.02.016 6. Hamasaki H Daily physical activity and type 2 diabetes: a review World J Diabetes 2016 7 243 251 10.4239/wjd.v7.i12.243 27350847 7. Duclos M Oppert JM Verges B Physical activity and type 2 diabetes. Recommandations of the SFD (Francophone Diabetes Society) diabetes and physical activity working group Diabetes Metab 2013 39 205 216 10.1016/j.diabet.2013.03.005 23643351 8. López Sánchez GF Smith L Raman R Physical activity behaviour in people with diabetes residing in India: a cross-sectional analysis Sci Sports 2019 34 1 e59 e66 10.1016/j.scispo.2018.08.005 9. Schram MT Baan CA Pouwer F Depression and quality of life in patients with diabetes: a systematic review from the European depression in diabetes (EDID) research consortium Curr Diabetes Rev 2009 5 112 119 10.2174/157339909788166828 19442096 10. Narita Z Inagawa T Stickley A Physical activity for diabetes-related depression: a systematic review and meta-analysis J Psychiatr Res 2019 113 100 107 10.1016/j.jpsychires.2019.03.014 30928617 11. Alfonso-Rosa RM Del Pozo-Cruz B Del Pozo-Cruz J Test-retest reliability and minimal detectable change scores for fitness assessment in older adults with type 2 diabetes Rehabil Nurs 2014 39 260 268 10.1002/rnj.111 23780835 12. Barrios-Fernández S Pérez-Gómez J Galán-Arroyo MDC Reliability of 30-s Chair Stand Test with and without cognitive task in people with type-2 diabetes mellitus Int J Environ Res Public Health 2020 17 1450 10.3390/ijerph17041450 32102379 13. Millor N Lecumberri P Gómez M An evaluation of the 30-s chair stand test in older adults: frailty detection based on kinematic parameters from a single inertial unit J Neuroeng Rehabil 2013 10 86 10.1186/1743-0003-10-86 24059755 14. Pinheiro PA Carneiro JAO Coqueiro RS “Chair stand testˮ as simple tool for sarcopenia screening in elderly women J Nutr Health Aging 2016 20 56 59 10.1007/s12603-016-0676-3 26728934 15. Chow RB Lee A Kane BG Effectiveness of the “Timed Up and Go” (TUG) and the Chair test as screening tools for geriatric fall risk assessment in the ED Am J Emerg Med 2019 37 457 460 10.1016/j.ajem.2018.06.015 29910184 16. Domínguez-Muñoz FJ Carlos-Vivas J Villafaina S Association between 30-s Chair Stand-Up Test and anthropometric values, vibration perception threshold, FHSQ, and 15-D in patients with type 2 diabetes mellitus Biology (Basel) 2021 10 246 10.3390/biology10030246 33809864 17. Bernard M-M Janson F Flora PK Videoconference-based physiotherapy and tele-assessment for homebound older adults: a pilot study Act Adapt Aging 2009 33 39 48 10.1080/01924780902718608 18. Cabrera-Martos I Ortiz-Rubio A Torres-Sánchez I Agreement between face-to-face and tele-assessment of upper limb functioning in patients with Parkinson disease PM R 2019 11 590 596 10.1002/pmrj.12001 30840363 19. Giordano V Koch H Godoy-Santos A WhatsApp messenger as an adjunctive tool for telemedicine: an overview Interact J Med Res 2017 6 e11 10.2196/ijmr.6214 28733273 20. Wilson RC Jones P A comparison of the visual analogue scale and modified Borg scale for the measurement of dyspnoea during exercise Clin Sci 1989 76 277 282 10.1042/cs0760277 21. Mokkink LB Terwee CB Knol DL The COSMIN checklist for evaluating the methodological quality of studies on measurement properties: a clarification of its content BMC Med Res Methodol 2010 10 22 10.1186/1471-2288-10-22 20298572 22. Fleiss JL Design and analysis of clinical experiments 2011 John Wiley & Sons 23. Heredia-Ciuró A, Lazo-Prados A, Blasco-Valls P et al (2022) Agreement between face-to-face and tele-assessment of upper limb disability in lung cancer survivors during COVID-19 era. J Telemed Telecare 1357633x221079543. 10.1177/1357633x221079543 24. Car J Tan WS Huang Z eHealth in the future of medications management: personalisation, monitoring and adherence BMC Med 2017 15 73 10.1186/s12916-017-0838-0 28376771 25. Mars M Scott RE WhatsApp in clinical practice: a literature review Stud Health Technol Inform 2016 231 82 90 27782019 26. WhatsApp LLC (2021) Privacy policy. https://www.whatsapp.com/legal/privacy-policy/?lang=en. Accessed 01 Sept 2022 27. Collado-Mateo D Madeira P Dominguez-Muñoz FJ The automatic assessment of strength and mobility in older adults: a test-retest reliability study Medicina (Kaunas) 2019 55 270 10.3390/medicina55060270 31212695 28. Rikli RE, Jones CJ (2001) Senior fitness test manual. Human Kinetics 29. Unver B Kahraman T Kalkan S Test-retest reliability of the 50-foot timed walk and 30-second chair stand test in patients with total hip arthroplasty Acta Orthop Belg 2015 81 435 441 26435238 30. Blankevoort CG van Heuvelen MJ Scherder EJ Reliability of six physical performance tests in older people with dementia Phys Ther 2013 93 69 78 10.2522/ptj.20110164 22976448 31. Lyders Johansen K Derby Stistrup R Skibdal Schjøtt C Absolute and relative reliability of the Timed ‘Up & Go’ Test and ‘30second Chair-Stand’ Test in hospitalised patients with stroke PLoS ONE 2016 11 e0165663 10.1371/journal.pone.0165663 27798686 32. Özkeskin M, Özden F, Ar E et al (2022) The reliability and validity of the 30-second chair stand test and modified four square step test in persons with multiple sclerosis. Physiother Theory Pract 1–7. 10.1080/09593985.2022.2070811 33. Ogawa EF Harris R Dufour AB Reliability of virtual physical performance assessments in veterans during the COVID-19 pandemic Arch Rehabil Res Clin Transl 2021 3 100146 10.1016/j.arrct.2021.100146 34589696 34. Güngör F, Ovacık U, Ertan Harputlu Ö et al (2022) Tele-assessment of core performance and functional capacity: reliability, validity, and feasibility in healthy individuals. J Telemed Telecare 1357633x221117335. 10.1177/1357633x221117335 35. Mani S Sharma S Omar B Validity and reliability of Internet-based physiotherapy assessment for musculoskeletal disorders: a systematic review J Telemed Telecare 2017 23 379 391 10.1177/1357633x16642369 27036879 36. Palacín-Marín F Esteban-Moreno B Olea N Agreement between telerehabilitation and face-to-face clinical outcome assessments for low back pain in primary care Spine (Phila Pa 1976) 2013 38 947 952 10.1097/BRS.0b013e318281a36c 23238489 37. Mani S Sharma S Singh DK Concurrent validity and reliability of telerehabilitation-based physiotherapy assessment of cervical spine in adults with non-specific neck pain J Telemed Telecare 2021 27 88 97 10.1177/1357633x19861802 31272309 38. Spina E Trojsi F Tozza S How to manage with telemedicine people with neuromuscular diseases? Neurol Sci 2021 42 3553 3559 10.1007/s10072-021-05396-8 34173087 39. Kagiyama N, Hiki M, Matsue Y et al (2021) Validation of telemedicine-based self-assessment of vital signs for patients with COVID-19: a pilot study. J Telemed Telecare 1357633X211011825. 10.1177/1357633x211011825	36456718	PMC9715279	NO-CC CODE	2022-12-03 23:20:15	no	Ir J Med Sci. 2022 Dec 2;:1-6	utf-8	Ir J Med Sci	2,022	10.1007/s11845-022-03238-w	oa_other
==== Front Atten Percept Psychophys Atten Percept Psychophys Attention, Perception & Psychophysics 1943-3921 1943-393X Springer US New York 36456797 2627 10.3758/s13414-022-02627-8 Article Selection history influences an attentional decision bias toward singleton targets http://orcid.org/0000-0002-3679-8785 Burnham Bryan R. [email protected] grid.267131.0 0000 0000 9464 8561 Department of Psychology, University of Scranton, Scranton, PA 18510 USA 1 12 2022 19 21 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. Selection history effects are ubiquitous findings that show how implicitly encoding a target’s feature or location on a trial can facilitate target activation on the following trial. Although the target-defining feature (e.g., color) is usually unpredictable, it is often relevant to determining the target on a given trial. The present study used a feature priming task, like the three-item oddball search task, but varied the target-defining feature (shape) orthogonal to the priming feature (color) that could influence target activation. On any trial the target could be a color singleton or not, and the target’s feature could repeat or switch between trials. Larger priming effects were seen when the current target was a color singleton than a nonsingleton. Importantly, diffusion analyses showed that pretrial selection bias contributed to these larger priming effects. The results suggest selection history facilitates target activation through an attentional decision bias to select the object with the most recently attended color, and this attentional decision is easier when the current target is also distinct. Keywords Feature priming Selection history Target activation Diffusion model ==== Body pmcSome primary functions of attention include selection of relevant information and suppression of irrelevant and potentially distracting information, which can occur in bottom-up and top-down manners. Additionally, an abundance of research indicates that selection history influences target and distractor activation/processing (e.g., Awh et al., 2012; Kristjánsson & Campana, 2010; Maljkovic & Nakayama 1994, 1996, 2000). The present study used diffusion modelling to examine how selection history via feature priming influences target activation. One of the earliest studies to demonstrate selection history effects on visual search was Maljkovic and Nakayama’s (1994) classic priming of popout (PoP) effect. PoP is the finding that responding to a feature singleton target is facilitated when its identifying feature (e.g., color) on Trial N – 1 is repeated on Trial N, even though the specific color (e.g., red, green) of the target is irrelevant on a given trial and unpredictable across trials. Importantly, it is repetition of the selected feature of the target that produces such priming, and this can occur whether the target is a singleton or a nonsingleton. Several mechanisms have been proposed for selection history effects. According to preattentive or salience-based accounts (Becker, 2008; Bichot & Schall, 2002; Maljkovic & Nakayama, 1994, 1996, 2000), implicitly encoding the target’s feature on Trial N – 1 boosts the gain of that feature on Trial N; that is, selection history effects arise through an increase in signal strength of the primed feature. In contrast, biased decision accounts suggest selection history biases attention toward features associated with the most recent target (Amunts et al., 2014; Lleras et al., 2008; Tseng et al., 2014; Yashar & Lamy, 2010; Yashar et al., 2017). That is, feature salience is unaffected by priming, but the attentional decision for which item to select is biased by a preceding target’s feature. Other accounts suggest selection history influences response selection and only after a target is selected (Hillstrom, 2000; Huang et al., 2004; Huang & Pashler, 2005; Thomson & Milliken, 2011, 2013). In this account, there is no influence on priming on feature salience or decision bias; rather, response retrieval is facilitated when the current target’s feature matches a preceding target. Despite evidence for each account, many agree that selection history effects arise due to several mechanisms (e.g., Ásgeirsson & Kristjánsson, 2011; Ásgeirsson et al., 2015; Kristjánsson & Campana, 2010; Lamy, Yashar & Ruderman, 2010; Yashar et al., 2013). One means of examining the mechanisms behind selection history’s influence on target activation is to apply Ratcliff diffusion modelling (RDM; Ratcliff, 1978, 1981, 2002; Ratcliff & McKoon, 2008; Ratcliff & Rouder, 1998; Ratcliff & Smith, 2010; Ratcliff et al., 1999) to data from PoP tasks. These models assume that evidence accumulates over time until a decision is made, and they can use RT distributions for correct and error responses to estimate parameters that relate to different, ongoing cognitive processes presumed to be involved with a decision. Importantly, these models have been adapted to make specific predictions about the cognitive processes operating during attentive decisions and priming tasks (e.g., Voss et al., 2013; Voss et al., 2004; Voss et al., 2015), which are the focus of the present study. Figure 1 illustrates the typical RDM: On any trial, a decision process begins at point z (zr) and continues until the lower or upper decision boundary (a) is reached and a response is made. The model is defined by several parameters, which are indicators of different cognitive processes: The threshold parameter (a) is the distance between response (evidence) thresholds and corresponds to liberal-conservative response criteria or decision strategies. Starting point (z) or relative starting point (zr; Voss et al., 2015) is the pretrial, a priori bias toward one of the response thresholds. In the present study, zr was the bias toward selecting the correct target. Drift rate (v) is the mean rate of evidence accumulation toward a response and reflects the speed of information accumulation during the decision process (Voss, et al., 2013). Lastly, a nondecision constant (t0) is the duration of nondecision processes including response retrieval or other response-related processes. Additionally, variability for each of the parameters can be estimated (i.e., sz, sv, st0). Fig. 1 The Ratcliff diffusion model. Evidence accumulation begins from a starting point (z) within interval (sz). Evidence accumulates in a noisy manner with a drift rate (v) and intertrial variation (sv) until one of two decision thresholds is reached, which are separated by boundaries (0, a). The response (correct or error) is based on whichever boundary is reached first. A nondecision time t0 with variation st0 is added to the RT The present study used diffusion modelling to examine the influence of selection history on processes involved with feature priming. If priming engages a pretrial attentional decision bias (biased selection), this should be reflected in the zr parameter; that is, a target’s feature on Trial N – 1 should bias selection of the same feature on Trial N. On the other hand, if priming leads to enhanced pre-activation of a previously selected target’s features and speeds processing of that feature, this should be reflected by priming effects in the v parameter. Lastly, if priming activates a motor response system associated with the preceding target, this should be seen in the t0 parameter. Several studies have used diffusion modeling to examine the processes involved with selection history. Tseng et al. (2014) examined the mechanisms underlying PoP by applying RDM to saccadic RTs in a popout search task. Subjects made a saccade toward a color singleton target and withheld saccades if no singleton was present, and they found that only the bias parameter (z) predicted PoP on target processing. This bias parameter reflected a pretrial tendency to make saccades toward the item with the preceding target’s color; that is, priming biased the attentional decision over which item to select. Additionally, Burnham (2018) applied RDM to a standard (manual response) PoP task that included a speed–accuracy manipulation. Results showed several parameters (i.e., z, v, t0) predicted overall PoP effects, but importantly, only specific parameters predicted the speed-accuracy manipulation on PoP. Specifically, decision bias (z) was associated with priming effects under accuracy instructions, whereas drift rate (v) was associated with priming for speed instructions. Hence, RDM logic can be applied to examine how selection history manifests its influence on feature priming. Although PoP and selection history effects are useful for examining target activation, one issue in previous studies is that although a specific color does not per se define the target on any trial (target color is random), color is relevant to identifying and selecting the target, and observers have incentive to attend to color. To examine implicit encoding and priming of target features without top-down incentive, the target-relevant feature can be made orthogonal to the feature that is used to assess selection history effects on target activation. In the present study, observers viewed displays containing three diamonds that included a color singleton and two homogenous nonsingletons, with the colors (red and green) chosen randomly (Fig. 2). On each trial, the target was defined as the diamond missing its left or right corner, whereas the two nontargets were diamonds missing the top or bottom corners. Hence, the target was always defined by its shape, but was unpredictably a color singleton or nonsingleton. This orthogonal relationship between target-relevant feature (shape) and priming feature (color) allows examination of selection history effects on target activation that are uncontaminated by the target-relevant feature being correlated with the priming feature. Indeed, this setup should eliminate top-down selection strategies based on color, and allows examination of priming when the target is a singleton or not on Trial N – 1 and on Trial N. Fig. 2 Example of the eight between-trial transitions created by the factors Trial N – 1 Target (Singleton vs. Nonsingleton), Trial N Target (Singleton vs. Nonsingleton), and Transition (Repeat vs. Switch). The target appears in the upper left in each panel, but the target and nontarget locations and orientations (left/right for the target and up/down for the nontargets) were randomized across trials. (Color figure online) Methods Subjects Previous research from my lab has observed PoP effects with effect sizes in excess of d = 1.50, and a power analysis indicated that six subjects were needed to detect priming effects of that size to achieve power = .80 (α = .05). A total of nine University of Scranton undergraduates participated (six females; all subjects right-handed). Subjects ranged from 18 to 19 years old (M = 18.56 years, SD = 0.53) and reported normal or corrected-to-normal vision. All subjects passed an Ishihra colorblindness test. Apparatus The experiment was programmed and presented using E-Prime software (Version 3.0.3.82) on a Dell OptiPlex 3050 x64 computer. Subjects sat 60 cm from a Dell E178Fpv monitor (1,024 × 768; 60 Hz). A five-button response box was used for responding. Stimuli A white cross (25.77 cd/m2, RGB: 255, 255, 255) appeared throughout each trial to maintain fixation. Search displays (Fig. 2) contained three diamonds (1.1° × 1.1°) on a black background (0.16 cd/m2, RGB: 0, 0, 0). Each diamond was missing one corner (0.14°). The target was missing the left or right corner, and the two nontargets were missing the top or bottom corner. One diamond was a color singleton, and the other diamonds were homogeneously colored nonsingletons. The colors of the singleton and nonsingletons were chosen randomly on each trial to be red (20.44 cd/m2, RGB: 255, 0, 0) or green (20.62 cd/m2, RGB: 10, 177, 31). The target was unpredictably the singleton. Each diamond appeared at a different one of 12 randomly chosen locations on the circumference of an imaginary ellipse (10° wide × 8° high), with no constraints on where the diamonds appeared. Procedures As data collection occurred during the Covid-19 pandemic, subjects wore mouth and nose-covering masks throughout the study. Subjects were informed they would see displays containing three diamonds, one of which (target) was missing the left or right corner and the other two (nontargets) were missing the top or bottom corners. Subjects were told their task was to indicate whether the target was missing its left or right corner as quickly and accurately as possible by pressing a corresponding key on the response box. Subjects were informed that one diamond would be colored differently from the other two (one red among two green diamonds or vice versa). Importantly, subjects were told (a) color would not help identify the target, as the colors were chosen randomly, and (b) the target would unpredictably appear as a color singleton or nonsingleton; hence, features of the target were completely unpredictable. Each subject completed a practice block of 48 trials followed by 10 blocks of 120 trials each with a self-paced break after each block, for a total of 1,248 trials. Each trial began with a fixation display containing a grey ‘+’ for 500 ms. Next the target display appeared for 2,000 ms or until the subject responded. The next trial began after a 100-ms delay. If a subject responded incorrectly or took longer than 2,000 ms, a 500-Hz tone played in the 100-ms delay. Design The target was unpredictably a singleton or nonsingleton and was unpredictably colored green or red on each trial. Considering the target’s color and singleton/nonsingleton status on both Trial N – 1 and Trial N, eight conditions emerged (Fig. 2) in a 2 (Trial N – 1 Target: Singleton vs. Nonsingleton) by 2 (Trial N Target: Singleton vs. Nonsingleton) by 2 (Transition: Target Color Repeat vs. Target Color Switch) fully within-subjects design. Results For RT analyses, only trials with a correct response on the current trial and preceding trial were used, which resulted in the removal of 5.4% of trials. For error analyses, only trials with a correct response on the preceding trial were used. Each subject’s mean RT (MRT) and percent error was calculated for each of the eight conditions. The MRT and percent errors averaged over all nine subjects appear in Table 1 and results are plotted in Fig. 3A. Table 1 Mean response times (RTs), percent errors, and standard deviations in the Trial N – 1 Target by Trial N Target by Transition design Trial N – 1 Target Trial N Target Transition RTs Percent Errors M SD M SD Singleton Singleton Switch 887 174 2.3 1.2 Repeat 779 144 2.6 1.9 Nonsingleton Switch 904 184 2.7 1.3 Repeat 844 167 2.7 2.5 Nonsingleton Singleton Switch 892 171 3.5 2.9 Repeat 796 135 2.6 2.5 Nonsingleton Switch 897 176 3.3 2.8 Repeat 845 175 2.5 2.2 Mean RTs are rounded to the nearest millisecond and percentage errors to the nearest tenth. Fig. 3 Mean resposne times (RTs), decision bias (zr), drift rate (v), and nondecision times (t0) in the Trial N Target by Transision interaction. Error bars are the 95% confidence intervals. (Color figure online) RTs A 2 (Trial N – 1 Target: Singleton vs. Nonsingleton) by 2 (Trial N Target: Singleton vs. Nonsingleton) by 2 (Transition: Repeat vs. Switch) repeated-measures analysis of variance (ANOVA) was conducted on the MRTs. The main effect of Trial N Target was significant, F(1, 8) = 13.23, MSE = 1,581, p = .007, ηp2 = .62, due to responses being faster when the current target was the color singleton (MRT = 838 ms; 95% CI [712, 965]) than a nonsingleton (873 ms; [746, 999]). The main effect of Transition was significant, F(1, 8) = 41.32, MSE = 2700, p < .0001, ηp2 = .84, due to responses being faster when the target color repeated (816 ms; [690, 943]) than switched (895 ms; [768, 1022]): a priming effect of 79 ms [50, 107]. The only interaction was between Trial N Target and Transition, F(1, 8) = 20.25, MSE = 459, p = .002, ηp2 = .72. No other effects were significant (Fs < 2.1, ps > .18). As seen in Fig. 3A, the interaction resulted from a larger priming effect when the Trial N target was the color singleton, 101 ms; t(8) = 6.9, SE = 14.70, p < .0001; d = 2.3, than a nonsingleton, 56 ms; t(8) = 4.82, SE = 11.61, p = .001; d = 1.6. Percent errors A similar repeated-measures ANOVA on percent errors did not yield any effects (Fs < 1.02, ps > .341). Diffusion model analysis Correct responses were assigned to the upper boundary and errors to the lower boundary of the model (Fig. 1). RT distributions for correct and error responses were entered into a diffusion-model analysis using fast-dm (Voss & Voss, 2007; Voss et al., 2015), with parameters estimated separately for each subject. Drift rate (v), nondecision constant (t0), and starting point (zr) were estimated in each of the eight conditions, whereas intertrial variability parameters (sv, szr, st0) were estimated over all conditions and the threshold parameter was constant (a = 1). Kolmogorov–Smirnov optimization was used to estimate the parameters due to moderate number of trials per condition (~125 trials/condition). Table 2 provides parameter estimates averaged over the nine subjects. The intertrial variability estimates were: szr (M = 0.154; SD = 0.116), sv (M = 0.222; SD = 0.149), st0 (M = 0.569; SD = 0.245). Table 2 Means and standard deviations of the estimates for the diffusion model parameters in the Trial N – 1 Target by Trial N Target by Transition design Trial N – 1 Target Trial N Target Transition zr v t0 M SD M SD M SD Singleton Singleton Switch 0.654 0.108 1.441 0.563 0.745 0.222 Repeat 0.794 0.116 1.898 1.043 0.666 0.165 Nonsingleton Switch 0.657 0.114 1.466 0.735 0.768 0.230 Repeat 0.581 0.066 2.005 0.427 0.681 0.184 Nonsingleton Singleton Switch 0.674 0.097 1.316 0.473 0.749 0.191 Repeat 0.783 0.114 1.426 0.936 0.677 0.166 Nonsingleton Switch 0.667 0.137 1.458 0.670 0.758 0.207 Repeat 0.637 0.083 1.886 0.545 0.696 0.190 For estimation, a = 1. A 2 (Trial N – 1 Target) by 2 (Trial N Target) by 2 (Transition) repeated-measures ANOVA was conducted on the zr estimates. The main effect of Trial N Target was significant, F(1, 8) = 25.65, MSE = .006, p < .0001, ηp2 = .76, due to a larger bias estimate toward correct responding when the target was a singleton (.73; [.66, .76]) than a nonsingleton (.64; [.57, .70]). The only other reliable effect was the interaction between Trial N Target and Transition, F(1, 8) = 15.90, MSE = .009, p = .004, ηp2 = .66, which is congruent with the interaction in the RT analysis. No other effects were significant (Fs < 3.0, ps > .109). As seen in Fig. 3B, a priming effect emerged when the Trial N Target was a singleton, priming = .13; t(8) = 5.36, SE = .023, p < .0001, d = 1.79, which was nonsignificant and reversed when the target was a nonsingleton, −.05; t(8) = −1.49, SE = .035, p = .174, d = −.50. A similar ANOVA was conducted on drift rate (v) estimates. Only the effect of Transition was significant, F(1, 8) = 7.32, MSE = .361, p = .027, ηp2 = .48, due to greater drift toward correct responding when the target color repeated (1.80; [1.45, 2.16]) than switched (1.42; [1.07,1.78]). No other effects approached statistical significance (Fs < 2.78, ps > .134). Lastly, an ANOVA was conducted on the nondecision (t0) estimates. The only effect that arose was Transition, F(1, 8) = 23.85, MSE = .004, p = .001, ηp2 = .75, due to greater nondecision time when the target color switched (.76; [.61, .90]) than repeated (.68; [.53, .83]). (All other Fs < 2.77, ps > .135.) Model fit Fit was examined graphically by comparing the empirical RTs and accuracies to a distributions of predicted RTs and accuracies. Predicted response time and error distributions were generated for each condition for each subject using the construct-samples routine from fast-dm (Voss & Voss, 2007). Using the parameter values obtained from each subject separate data sets of N = 1,000 trials each were generated, for a total of 9 (subjects) × 8 (conditions) 72,000 trials.1 Parameters were then re-estimated, from the predicted data, for each subject using fast-dm. The predicted accuracies and MRT in each of the conditions were compared against the empirical accuracies and MRT for the 10th, 30th, 50th, 70th, and 90th quantiles (Voss et al., 2004, 2008, 2013, 2015). As seen in Fig. 4, the points lie close to the line of perfect congruency, suggesting a good fit of the model and no bias in the predicted data. Fig. 4 The figures display the relationship between the empirical versus predicted RTs based on fits to the diffusion model. Each symbol represents the mean of a single subject in a single experimental condition. In the legend, the first and second letter refer to the target on Trial N – 1 and Trial N, respectively (S = singleton; NS = nonsingleton), and the third letter refers to the between-trial transition (S = switch colors; R = repeat colors). (Color figure online) Discussion This study used a variant of the three-item oddball search task by orthogonally manipulating the target feature (shape) and priming feature (color), to examine how decoupling these features, which negated top-down selection, influenced selection history on target activation. The results showed that feature priming occurred whether the current target was a color singleton or nonsingleton (Fig. 3B), but the priming effect was larger when the current target was the singleton. Thus, irrelevant features of a selected target were implicitly encoded and facilitated selection on the following trial when the target appeared with the same feature. Importantly, the priming effect was greater when the current target was the color singleton, revealing the influence of bottom-up activation. The diffusion analysis showed that feature priming influenced attentional selection (decision) bias (zr), target processing time (v), and post-selection processes (t0). That is, priming influenced pretrial selection bias, information accumulation, and response execution. More importantly, the interaction in RTs between transition and target status on Trial N was driven by an influence of pretrial selection bias (zr). Thus, the larger feature priming effect when the current target was a color singleton was driven completely by a larger influence of pretrial bias (zr) and not facilitated processing time (v) or post-selection response processes (t0; see Fig. 3). The results suggest selection history induced a bias to select the most likely target—the item or items with the repeated target color—and when that item was the color singleton, the attentional decision was facilitated by bottom-up salience. However, because the target was the color singleton on half the trials, it may have been rational for observers to select the singleton in a top-down manner. That is, to observers, it may have seemed easier to select the singleton rather than the nonsingletons as a potential target. If so, the larger priming effect when the Trial N target was the color singleton may have been driven by a strategy to select the singleton, rather than bottom-up activation, as noted above. Such a singleton selection strategy predicts carryover when the Trial N – 1 target was the singleton, in the form of a Trial N – 1 Target by Transition interaction and a three-way Trial N – 1 by Trial N by Transition interaction. If observers adopted a singleton selection strategy, larger priming should have been observed when the Trial N – 1 target was the singleton, and when both the Trial N – 1 and Trial N targets were singletons, but neither interaction was statistically reliable. Although null interactions are difficult to interpret, such a singleton selection strategy is plausible. Nonetheless, it is noteworthy that the singleton status of the target on Trial N – 1 had no influence on priming. Specifically, whether the target was the color singleton or a nonsingleton on Trial N – 1 had no influence on RTs, accuracy, or the diffusion parameters. This suggests a target’s features are implicitly encoded only after it has been selected. Or, more specifically, it is the features of the selected target, whether salient or not, that are encoded and lead to priming. Also, interestingly is that ‘inhibitory tagging’ did not occur when the target was a nonsingleton on Trial N – 1, which is somewhat unlike the distractor preview effect (e.g., Ariga & Kawahara, 2004; Goolsby & Suzuki, 2001). In conclusion, the results of the present study add to other studies that suggest several mechanisms are involved in selection history (priming) effects on target activation and selection (PoP). The unique outcome from this study is showing that selection history influenced target selection primarily through a pretrial attentional decision bias to select the most likely target in the current display. That is, implicitly encoding a target’s features biased the attentional decision on the following trial to select the item (or items) with that same feature. Critically, this decision was made easier when the current target was distinct and displayed as a singleton, suggesting bottom-up attention (or possibly top-down strategies) of current displays may be a critical factor in selection history. 1 The number of trials (1,000) generated is based on previous work (Burnham, 2018; Voss et al., 2013). Open practices statement None of the experiments reported in this study was preregistered; however, data or materials used to run the experiments will be made available upon request. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Amunts L Yashar A Lamy D Inter-trial priming does not affect attentional priority in asymmetric visual search Frontiers in Psychology: Cognition 2014 5 1 10 10.3389/fpsyg.2014.00957 Ariga A Kawahara J-I The perceptual and cognitive distractor-previewing effect Journal of Vision 2004 4 10 891 903 10.1167/4.10.5 15595893 Ásgeirsson ÁG Kristjánsson Á Episodic retrieval and feature facilitation in intertrial priming of visual search Attention, Perception, & Psychophysics 2011 73 5 1350 1360 10.3758/s13414-011-0119-5 Ásgeirsson ÁG Kristjánsson Á Bundesen C Repetition priming in selective attention: A TVA analysis Acta Psychologica 2015 160 35 42 10.1016/j.actpsy.2015.06.008 26163225 Awh E Belopolsky A Theeuwes J Top-down versus bottom-up attentional control: A failed theoretical dichotomy Trends in Cognitive Sciences 2012 16 8 437 443 10.1016/j.tics.2012.06.010 22795563 Becker SI The mechanism of priming: Episodic retrieval or priming of pop-out Acta Psychologica 2008 127 324 339 10.1016/j.actpsy.2007.07.005 17868628 Bichot NP Schall JD Priming in macaque frontal cortex during popout visual search: Feature based facilitation and location-based inhibition of return Journal of Neuroscience 2002 22 4675 4685 10.1523/JNEUROSCI.22-11-04675.2002 12040074 Burnham BR Selection and response bias as determinants of priming of popout search: Revelations from diffusion modelling Psychonomic Bulletin & Review 2018 25 2389 2397 10.3758/s13423-018-1482-1 29725950 Goolsby BA Suzuki S Understanding priming of color-singleton search: Roles of attention at encoding and “retrieval” Perception & Psychophysics 2001 63 3 929 944 10.3758/BF03194513 11578055 Hillstrom AP Repetition effects in visual search Perception & Psychophysics 2000 62 800 817 10.3758/BF03206924 10883586 Huang LQ Holcombe AO Pashler H Repetition priming in visual search: Episodic retrieval, not feature priming Memory & Cognition 2004 32 12 20 10.3758/BF03195816 15078040 Huang LQ Pashler H Expectation and repetition effects in searching for featural singletons in very brief displays Perception & Psychophysics 2005 67 150 157 10.3758/BF03195018 15912878 Kristjánsson Á Campana G Where perception meets memory: A review of repetition priming in visual search Attention, Perception, & Psychophysics 2010 72 5 19 10.3758/APP.72.1.5 Lamy, D. Yashar, A., & Ruderman, L. (2010). A dual-stage account of intertrial priming effects. Vision Research, 50, 1396–1401. Lleras A Kawahara J Wan XI Ariga A Inter-trial inhibition of focused attention in popout search Perception & Psychophysics 2008 70 114 131 10.3758/PP.70.1.114 18306966 Maljkovic V Nakayama K Priming of popout: I Role of features. Memory & Cognition 1994 22 657 672 10.3758/BF03209251 7808275 Maljkovic V Nakayama K Priming of popout: II. The role of position Perception & Psychophysics 1996 58 977 991 10.3758/BF03206826 8920835 Maljkovic V Nakayama K Priming of popout: III. A short-term implicit memory system beneficial for rapid target selection Visual Cognition 2000 7 571 595 10.1080/135062800407202 Ratcliff R A theory of memory retrieval Psychological Review 1978 85 59 108 10.1037/0033-295X.85.2.59 Ratcliff R A theory of order relations in perceptual matching Psychological Review 1981 88 552 572 10.1037/0033-295X.88.6.552 Ratcliff R A diffusion model account of response time and accuracy in a brightness discrimination task: Fitting real data and failing to fit fake but plausible data Psychonomic Bulletin & Review 2002 9 278 297 10.3758/BF03196283 12120790 Ratcliff R McKoon G The diffusion decision model: Theory and data for two-choice decision tasks Neural Computation 2008 20 873 922 10.1162/neco.2008.12-06-420 18085991 Ratcliff R Rouder JN Modeling response times for two-choice decisions Psychological Science 1998 9 347 356 10.1111/1467-9280.00067 Ratcliff R Smith PL Perceptual discrimination in static and dynamic noise: The temporal relation between perceptual encoding and decision making Journal of Experimental Psychology: General 2010 139 70 94 10.1037/a0018128 20121313 Ratcliff R Van Zandt T McKoon G Connectionist and diffusion models of reaction time Psychological Review 1999 106 261 300 10.1037/0033-295X.106.2.261 10378014 Thomson DR Milliken B A switch in task affects priming of pop-out: Evidence for the role of episodes Attention, Perception, & Psychophysics 2011 73 318 333 10.3758/s13414-010-0046-x Thomson DR Milliken B Contextual distinctiveness produces long-lasting priming of pop-out Journal of Experimental Psychology: Human Perception and Performance 2013 39 202 215 10.1037/a0028069 22506787 Tseng Y-C Glaser JI Caddigan E Lleras A Modeling the effect of selection history on pop-out visual search PLOS ONE 2014 9 3 e89996 10.1371/journal.pone.0089996 24595032 Voss A Rothermund K Brandtstädter J Interpreting ambiguous stimuli: Separating perceptual and judgmental biases Journal of Experimental Social Psychology 2008 44 1048 1056 10.1016/j.jesp.2007.10.009 Voss A Rothermund K Gast A Wentura D Cognitive processes in associative and categorical priming: A diffusion model analysis Journal of Experimental Psychology: General 2013 142 536 559 10.1037/a0029459 22866687 Voss A Rothermund K Voss J Interpreting the parameters of the diffusion model: An empirical investigation Memory & Cognition 2004 32 1206 1220 10.3758/BF03196893 15813501 Voss A Voss J Fast-dm: A free program for diffusion model analysis Behavior Research Methods 2007 39 767 775 10.3758/BF03192967 18183889 Voss, A., Voss, J., & Lerche, V. (2015). Assessing cognitive processes with diffusion model analyses: A tutorial on fast-dm-30. Frontiers in Psychology, 6(336). 10.3389/fpsyg.2015.00336 Yashar A Lamy D Intertrial repetition affects perception: The role of focused attention Journal of Vision 2010 10 14 3, 1 3, 8 10.1167/10.14.3 Yashar A Makovski T Lamy D The role of motor response in visual encoding during search Vision Research 2013 93 80 87 10.1016/j.visres.2013.10.014 24400358 Yashar A White AL Fang W Carrasco M Feature singletons attract spatial attention independently of feature priming Journal of Vision 2017 17 9 7, 1 7,18 10.1067/17.9.7	36456797	PMC9715281	NO-CC CODE	2022-12-03 23:20:15	no	Atten Percept Psychophys. 2022 Dec 1;:1-9	utf-8	Atten Percept Psychophys	2,022	10.3758/s13414-022-02627-8	oa_other
==== Front Mol Divers Mol Divers Molecular Diversity 1381-1991 1573-501X Springer International Publishing Cham 36456773 10576 10.1007/s11030-022-10576-5 Original Article Evaluation of action of steroid molecules on SARS-CoV-2 by inhibiting NSP-15, an endoribonuclease http://orcid.org/0000-0003-2774-853X Dhanabalan Anantha Krishnan 1 http://orcid.org/0000-0003-3781-3921 Raghavan Sriram Srinivasa 1 http://orcid.org/0000-0002-6679-980X Rajendran Selvakumar 1 Ramasamy Velavan 2 Abdul Shaik Abdul Azeez 1 http://orcid.org/0000-0001-5588-8025 Narayanasamy Nandhagopal [email protected] 1 http://orcid.org/0000-0001-9027-0108 Krishnasamy Gunasekaran [email protected] 1 1 grid.413015.2 0000 0004 0505 215X Centre of Advanced Study in Crystallography and Biophysics, University of Madras, Guindy Campus, Chennai, 600 025 India 2 grid.444347.4 0000 0004 1796 3866 Department of Physics, Bharath Institute of Higher Education and Research, Chennai, 600 073 India 1 12 2022 114 1 9 2022 20 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. Many countries in the world have recently experienced an outbreak of COVID-19, turned out to be a pandemic which significantly affected the world economy. Among many attempts to treat/control infection or to modulate host immunity, many small molecules including steroids were prescribed based on their use against other viral infection or inflammatory conditions. A recent report established the possibility of usage of a corticosteroid against the virus through inhibiting NSP-15; an mRNA endonuclease of SARS-CoV-2 and thereby viral replication. This study aimed to identify potential anti-viral agents for the virus through computational approaches and to validate binding properties with the protein target through molecular dynamics simulation. Unlike the conventional approaches, dedicated data base of steroid like compounds was used for initial screening along with dexamethasone and cortisone, which are used in the treatment of COVID-19 affected population in some countries. Molecular docking was performed for three compounds filtered from data base in addition to dexamethasone and Cortisone followed by molecular dynamics simulation analysis to validate the dynamics of binding at the active site. In addition, analysis of ADME properties established that these compounds have favorable drug-like properties. Based on docking, molecular dynamics simulation studies and various other trajectory analyses, compounds that are identified could be suggested as therapeutics or precursors towards designing new anti-viral agents against SARS-CoV-2, to combat COVID-19. Also, this is an attempt to study the impact of steroid compounds on NSP-15 of SARS-CoV-2, since many steroid like compounds are used during the treatment of COVID-19 patients. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1007/s11030-022-10576-5. Keywords SARS-CoV-2 COVID-19 NSP-15 Endoribonuclease Steroids Dexamethasone Molecular docking Principal component analysis ==== Body pmcIntroduction Severe Acute Respiratory Syndrome out break by SARS-CoV and Middle East Respiratory Syndrome (MERS-CoV) identified in 2002/2003 and 2012, respectively. The infections of SARS-CoV spread to 27 countries and MERS-CoV resulted in 8000 infections worldwide. Presently, SARS-CoV-2, newly identified virus is creating a havoc around the world. Infection causes a disease of Corona Virus disease -2019, a serious pneumonia where mortality is high compared with other viral infections. World Health Organization declared SARS-CoV-2 infection as pandemic. Recent emergence of many variants add complexity in understanding the disease and hence in strategies to control or manage the situation. COVID-19 has worse effect both on human life and world economy. This disease challenge the present day health care system and made the human race to think about the in-adequacy of available medical facilities to fight this invisible particles, SARS-CoV-2. Scientific community and pharma companies have developed vaccines which are used for mass vaccination is taken up by many countries. Since no specific drug is discovered to combat COVID-19, some small molecule drugs such as remdesivir are prescribed time to time. The spike protein which is essential for viral invasion was targeted for vaccine development as well as for drug designing efforts. There are reports on different targets such as Non-Structural proteins with varying functions involved in viral replication. Also, from these reports describe the possibilities aiming to inhibit them in turn to control viral replication. Researchers employ different data bases such as drug data base, small molecule data base and natural compounds data base etc., for screening against different targets identified of SARS-CoV-2. This approach is faster and may result in effective compounds against SARS-CoV-2. Historically, besides different approaches to design anti-virals, steroids have been used against viral diseases for treatment as immuno-suppressant. Systemic use of corticosteroids were practiced against SARS-CoV and MERS-CoV for the symptom of pneumonia resulted in increased viral replication [1]. Recent study of revisiting the usage of corticosteroids divulged the effect on viral replication as well as inflammation of the host. It was shown that the steroid compounds, Ciclesonide and mometasone had concentration dependent effect on viral growth and cytotoxicity [2]. The report suggested that the Ciclesonide interacts with NSP-15 directly or indirectly. NSP-15 is a nidoviral RNA uridylate-specific endoribonuclease (NendoU). It carries C-terminal catalytic domain belonging to EndoU family [3]. Previously, NSP-15 was thought to be directly involving in viral replication but later found that it’s role in viral replication is indirect. Also, the immunomodulatory property of NSP-15 by interfering innate immune response is recently reported [4]. In addition, it was also suggested that NSP-15 is involving in RNA degradation which help to hide from the host defenses. Considering various roles of NSP-15, it is reported as essential in coronavirus biology. By considering the use of steroids and the importance of NSP-15 enzyme, here we have used dedicated steroid compounds data base for screening against the target NSP-15. Also, binding of different steroid compounds with NSP-15 have been analyzed in-detail through MD simulation and tragterory analysis. This study elucidates structure based interactions of various steroid compounds with NSP-15 (endoribonuclease) which will be useful to expand research efforts on the use of steroid compounds against SARS-CoV-2. Computational methods Library preparation A total of 533 steroid like compounds were retrieved from ChemDiv database and eight steroid compounds with anti-viral activity as reported [2] were chosen for the structure based virtual screening against NSP-15 protein target. Protein and ligand preparation Protein preparation The three-dimensional structure of NSP-15 was retrieved from the Protein Data Bank (PDB ID: 6VWW) [5]. The structure of NSP-15 and the active site are shown in Fig. 1. To perform the docking studies, the protein structure was prepared using protein preparation wizard available in Schrödinger suite [6]. There are two steps involved in protein preparation. First one is the preparation; in this step hydrogens were added and side-chain atoms were neutralized [7]. The second step is refinement where the water molecules were removed and hydrogen atoms were geometry minimized until the average root mean square deviation of the non-hydrogen atoms reached 0.3 Å [8].Fig. 1 Crystal structure of NSP-15 of SARS-CoV-2 and the active site details (PDB ID: 6VWW) Ligand preparation Ligprep module included in PHASE was used to convert 2D to 3D structures, and energy minimization was performed using OPLS 2005 force field [9], with implicit GB/SA solvent model. Using the rapid torsion angle approach, the conformers of all the compounds were generated and structures with high estimated energies were eliminated. Using preprocesses, minimization of 100 steps and post-process minimization of 50 steps, maximum of 1000 conformers for each structure were generated. All the minimized conformers were filtered using a relative energy window of 10 kcal mol−1 and a minimum atom deviation of 1.0 Å [10]. Energy threshold value (10 kcal mol−1) was set as to the lowest energy conformer. Conformers having higher energy were discarded. Structure based virtual screening Structure based virtual screening is the simple and best approach to find suitable lead molecules against the target protein. To identify a suitable steroid like ligand against NSP-15, ChemDiv DB chemical database was used and SBVS protocol was employed. Initially, a receptor grid was generated based on the non-structural protein -15 (NSP-15) structure around the active site. The prepared database compounds were docked with structure based docking parameter of extra precision (XP) procedures with default settings [11]. The best compounds were selected based on their glide score and energy. SBVS has been carried out with 541 compounds. Based on key interaction profile and glide energy, 16 compounds were short listed. Finally, by employing induced fit docking (IFD) protocols, five lead compounds were shortlisted based on their binding energetics. The overall lead identification work flow is shown in Fig. S1. Molecular docking (induced fit docking) From the Schrodinger’s Glide extra precision XP docking results, best hit compounds were selected and subjected to induced fit docking (IFD) using Glide Schrödinger, LLC, New York, NY, 2015 [12]13. A mixed molecular docking method where the protein was flexible and ligand was rigid during docking was employed. During docking protocols, van der Waals (vdW) radii scaling of 0.5 for the proteins and ligand nonpolar atoms was used. The NSP-15 protein structure (PDB ID: 6VWW) was subjected to energy minimization, using force field (OPLS-2005) with an implicit solvation model. The Prime module was used to determine Docking score and Glide energy. Finally, the obtained binding poses were evaluated through glide empirical scoring function. The best pose of docked protein–ligand complexes were analyzed using Pymol and ligplot [14, 15]. Molecular dynamics simulation The complexes were solvated with water molecules in a cubic box where a protein molecule is at 10 Å distance from the box edge. TIP3P water model was used for simulations. The system was neutralized by adding counter ions. Then, the system was subjected to maximum force minimization of 1000 kJ/mol/nm. The minimized system was subjected to initial NVT equilibration followed by NPT for 500 ps. The system was maintained at 300 K and 1 bar using modified Berendsen Thermostat [16] and Parinello-Rahman barostat [17] for temperature and pressure, respectively. Subsequently, the ligand-bound protein was subjected to 50 ns molecular dynamics. The MD simulations were carried out using Gromacs 5.1 version [18]. PCA analysis PCA analysis was carried out based on the Cartesian coordinate deviants from the reference crystal structure (6VWW) through diagonalization of covariance matrix. PCA tends to reduce the dimension of the data. In MD, the principal components refers to the eigen vector determining the direction of motion and the eigen value, the extent of residual motion. The low dimensional displacement subspace covered by first few eigen vector representing the PC’s is termed as essential dynamics [19]. The PCA is defined by the coordinates of the trajectories x(t). The correlational atomic motions is expressed in terms of covariance matrix Cα equation below:1 C∝=covx=(⟨x-⟨x⟩x-⟨x⟩⟩T) where ⟨x⟩ represents the positional average over time. The matrix Cα can be normalized by orthogonal transformation T given by:2 x-⟨x⟩=Tqorq=TT⟨x-⟨x⟩⟩ The transformation leads to the determination of eigen value from the diagonal matrix Δ=⟨qqT⟩ofλi. The overall solution of the matrix can be given by:3 Cα=TΔTTorΔ=TTCαT All the calculations were chosen for Cα atoms to find maximum variance in implicit solvent system by stripping the solvent and ions from the trajectories. CPPTRAJ [20] implemented in AMBER was used for computing the PCA. The first three PCA with corresponding eigen values were utilized to understand the conformations. Free energy analysis Protein energy was calculated with molecular mechanics with generalized born surface approximation (MM/GBSA)[21]. The protein energy was estimated using the equation having energy contribution terms given by Eq. 4:4 G=Einternal+EElectrostatics+Evan der Waals+Gsolvation+Gnon-polar solvation-TSMM In the above equation, first three terms represent molecular mechanics energy such as internal, electrostatics and van der Waals energy. Gsolvation and Gnon-polar solvation represents solvation energies, T represents absolute temperature, SMM represent the entropy calculated from molecular mechanics from estimated harmonic frequencies of vibrations. ADME prediction To evaluate pharmacological properties of ligands, QikProp [22] module in Schrödinger 2015–2 was used. The following properties were predicted such as Molecular weight (MW), Hydrogen bond donor (HBA), hydrogen bond acceptor (HBD), water partition (QPlogPo/w), blood–brain barrier (BBB) penetration (QPLogBB) and central nervous system (CNS) activity of the molecules were predicted. Results and discussion The structure based virtual screening was performed to identify new lead(s) of steroid like compounds with therapeutic effectiveness against COVID-19 targeting NSP-15 protein. Molecular docking through induced fit strategies followed by MD simulation revealed the binding mode, strength and stability of protein–ligand interactions. Virtual screening The library of 533 steroids compounds from ChemDIV database was screened against NSP-15 through structure based docking using glide XP module. Compounds were ranked using docking score and glide energy (Table S1a and S1b) and the top 8 hit compounds were shortlisted. In addition, docking studies were carried out for eight steroid compounds having anti-viral activities (tested compounds) as reported [2]. From these analysis, three compounds from ChemDIV (0449-0045, N001-0004, N006-0008) and two compounds (Dexamethasone and Cortisone) from tested compounds were identified as potential compounds based on glide scoring function and ligand–protein interaction profiles. The 2D chemical schemes are shown in Fig. 2.Fig. 2 2D structure of known steroids and best hit compounds Molecular docking (induced fit docking) Induced Fit Docking (IFD) has been carried out for the tested compounds (from earlier reports) and steroid compounds from ChemDIV against NSP-15 of SARS-CoV2. All the compounds are docked in to the active site having interactions with catalytic residues with meaningful binding energetics. The compound 0449-0045 make four hydrogen bond interactions with SER294, VAL292, GLN245 and GLU340 at a distance of 2.83 Å, 3.19 Å, 2.81 Å and 2.75 Å respectively as the best pose with glide score and energy of − 9.74, − 65.61 kcal/mol, respectively. Compound N001-0004 shows binding with docking score of − 7.87 and glide energy of − 45.99 kcal/mol by interactions with residues, HIS235, VAL292, SER292 and GLY248 at a distance of 2.97 Å, 3.30 Å, 2.98 Å and 2.99 Å, respectively. Similarly, N006-0008 compound has five hydrogen bond interactions with SER294, LYS290, HIS250, GLY248 and GLN245 and at a distance of 3.14 Å, 2.79 Å, 3.06 Å, 3.25 Å and 2.86 Å respectively with glide score of − 9.42 and glide energy of − 55.68 kcal/mol. Both dexamethasone and cortisone make two hydrogen interaction with active sites of SER290 and LYS292 at a distance of 3.33 Å and 3.21 Å and 3.14 Å and 3.06 Å, respectively. Also energetics found similar for the protein–ligand binding [− 4.07 and − 31.03 kcal/mol for dexamethasone and − 4.63 and − 28.18 kcal/mol for cortisone]. Interestingly, all the compounds show hydrogen bond interaction with SER294 which actually involving in catalysis by initiating nucleophilic attack on the substrate at the active site of NSP-15. Interactions of the compounds with the target at the active site are shown in Fig. 3 and the hydrogen bonding and hydrophobic interactions are listed in Table 1. All the shortlisted compounds found to have energetically favorable binding. Superposition of the docked compounds at the active site with NSP-15 are shown in Fig. 4. Fig. 3 Ligand interactions of experimental tested steroids and identified best hit steroid compounds with NSP-15. A Dexamethasone, B Cortisone, C 0449-0045, D N001-0004, E N006-0008 Table 1 Molecular docking studies of NSP-15 protein complexed with identified steroid hit compounds Compounds Glide gscore (kcal/mol) Glide energy (kcal/mol) Hydrogen bond interactions (D–H⋯A) Distances (Å) 0449-0045 − 9.74 − 65.61 (O–H⋯O) GLU340 (O–H⋯O) SER294 (O–H⋯O) SER294 SER294 (N–H⋯O) (O–H⋯O) VAL292 (O–H⋯O) VAL292 (O–H⋯O) GLN245 (O–H⋯O) GLN245 2.75 2.83 2.76 3.18 3.19 3.32 2.81 2.79 N001-0004 − 7.87 − 45.99 (O–H⋯O) SER 294 (O–H⋯O) SER 294 (O–H⋯O) VAL 292 GLY 248 (N–H⋯O) HIS 235 (N–H⋯O) 2.83 2.98 3.30 2.99 2.97 N006-0008 − 9.42 − 55.68 (O–H⋯O) SER 294 LYS 290 (N–H⋯O) HIS 250 (N–H⋯O) GLY 248 (N–H⋯O) (O–H⋯O) GLN 245 (O–H⋯O) GLN 245 3.14 2.79 3.06 3.25 2.86 2.70 Dexamethasone (Pubchem ID: 5743) − 4.07 − 31.03 SER294 (N–H⋯O) LYS 290 (N–H⋯O) 3.33 3.21 Cortisone (Pubchem ID: 222786) − 4.63 − 28.18 SER294 (N–H⋯O) LYS 290 (N–H⋯O) 3.14 3.06 Bold represents “active site residues” Fig. 4 Superposition of experimental tested steroids and identified best hit steroid like compounds with NSP-15. a Surface representation of superimposed docked complexes, b binding interaction of identified steroids at the catalytic sites of NSP-15 Molecular dynamics simulations Molecular dynamics simulations for 50 ns time scale were performed for analyzing the dynamic behaviors of the NSP-15-ligand complexes. Detailed trajectory analyses were performed to understand the dynamics of inter molecular interactions. The root mean square deviation (RMSD) profile measures the average distance between superimposed biomolecules (Cα), indicating the degree of closeness with reference to the initial structure. The radius of gyration (Rg) represents the mass-weighted RMS deviation of group of atoms from their center of mass, denoting global measurement of protein dimension. All the ligand-bound complexes were stable over 50 ns. The RMSD is minimal for N006-0008 ligand-bound structure in comparison with that of N001-0004 (Fig. 5). The RGYR (Rg) values of the NSP-15 complexes were constrained, which ranges from 23 to 24 Å, for all the cases. Hydrogen bond interactions around the active of the complexes are uniformly present, though N006-0008 compound has a larger contribution favoring binding in comparison with other compounds (Fig. 6). The fluctuation of amino acids during MDS is an important parameter to evaluate internal residual dynamics. The RMSF profiles of the complexes are similar in the overall pattern while certain active site residues are involved in the interactions behave differently. In general, complexes have linear and correlated RMSF profiles (Fig. 7).Fig. 5 RMSD of C-alpha atoms of best hits and known steroid compound complexes from the MD trajectories Fig. 6 RGYR of best hits and known steroid compound complexes from the MD trajectories Fig. 7 RMSF of C-alpha atoms of best hits and known steroid compound complexes from the MD trajectories All five docked complexes were observed to be stable throughout the MD simulation of 50 ns. Hydrogen bonding interactions with residues Lys290, Val292, His250, Ser294, Leu346, Gln347, Cys293 and Gln245 at the active site vicinity found to be stable for all the complexes through out the duration of simulation. Hydrophobic interactions with the residues Trp333 and Tyr343 are also intact during simulation (Fig. 8).Fig. 8 Protein and ligand contact analysis from the simulated MD trajectory Principal component analysis NSP-15 complexes (0449-0045, N001-0004, N006-0008, cortisone and dexamethasone) were analyzed for dominant motion using principal components analysis (PCA), for which the first few eigenvectors were captured and their combined motions amplitude were identified to evaluate local fluctuations. The first three PCs accounted for the variance of motion observed for the complexes and two-dimensional plot between eigenvectors was drawn to compare the correlated motion. (Fig. S2). The graph represents the variance in conformational distribution, denoting periodic change between the conformations. All the complexes have a similar distribution profile, the subspace of N001-0004 and 0449-0045 has PC 1/2 and 1/3 distinct distribution centered on the origin, making it confined to one local energy basin. The first PCs accounted for 50.9%, 18.6%, 31.2%, 36.4% and 25.8% of conformational variance observed during the trajectories of complexes (N001-0004, N006-0008, cortisone, 0449-0045 and dexamethasone) respectively. Assesment of binding free energy Docked NSP-15 complexes with best hit compounds (0449-0045, N001-0004, N006-0008) from virtual screening and reported steroids cortisone, dexamethasone were analyzed to evaluate binding free energy using the MM/GBSA protocol. The total binding free energy (ΔGbind) for dexamethasone complex is − 14.87 kcal/mol which is the least when compared with other compounds 0449-0045 (− 26.01), N001-0004 (− 15.93), N006-0008 (− 28.12) and cortisone (− 28.78 kcal/mol). Table 2 describes individual contributions of polar (ΔEele) and non-polar (ΔEvdw) components. Both non polar and polar contributions are higher for shortlisted compounds than tested compounds. As a result, based on total binding energy, it can be established that 0449-0045 and N006-0008 are better lead compounds to inhibit NSP-15.Table 2 Free energy calculation of shortlisted hit compounds and steroids Compounds N001-0004 N006-0008 0449-0045 Dexamethasone Cortisone Vdwaals − 30.06 ± 2.59 − 43.58 ± 2.93 − 38.49 ± 3.31 − 22.38 ± 5.23 − 28.78 ± 2.85 Eel − 14.89 ± 7.16 − 12.52 ± 9.36 − 42.27 ± 11.38 − 5.55 ± 4.44 − 26.73 ± 5.33 Egb 32.58 ± 6.66 33.73 ± 7.04 60.09 ± 10.16 15.99 ± 4.40 33.68 ± 4.10 Esurf − 3.55 ± 0.30 − 5.75 ± 0.37 − 5.32 ± 0.32 − 2.93 ± 0.66 − 3.49 ± 0.22 Delta g gas − 44.96 ± 8.01 − 56.11 ± 8.75 − 80.77 ± 12.79 − 27.93 ± 8.30 − 55.51 ± 4.79 Delta g solv 29.02 ± 6.57 27.98 ± 7.07 54.76 ± 9.95 13.06 ± 4.08 30.19 ± 4.04 Delta total − 15.93 − 28.12 − 26.01 − 14.87 − 28.78 ADME properties By the encouragement of docking and dynamics results, the identified potential NSP-15 inhibitor lead compounds were analyzed for their drug likeliness. ADME analysis was performed to predict biological properties of these lead compounds. Using Qikprop module, properties were calculated and tabulated (Table 3). Action of these identified leads on central nervous system were predicted. Molecular weight < 500 Da, Hydrogen bond donor ≤ 5, hydrogen bond acceptor ≤ 10, QPlogPo/w ≤ 5, QPlogBB) range of − 3.0 to 1.2, Central nervous system (CNS) activity − 2 to + 2Table 3 ADME properties of shortlisted hit compounds and steroids Compounds mol MW donorHB acceptHB QPlogPo/w QPlogBB CNS N001-0004 584.659 8 18.1 − 1.997 − 3.246 − 2 N006-0008 885.054 9 27 0.342 − 4.153 − 2 0449-0045 872.956 8 29.3 − 1.866 − 5.94 − 2 5743 392.466 3 8.15 1.83 − 1.071 − 2 222786 360.449 2 8.45 1.468 − 1.275 − 2 Conclusion NSP-15 encodes for an endonuclease, exhibiting EndoU activity which in turn interfere innate immune response. The importance of this target is maximized by its action on dsRNA to degrade so that host RNA sensors could not recognize which resulted effective evasion of host immune system. Binding of a compound at the active site would inhibit the function of NSP-15, and so effective against SARS-CoV-2. Since steroids are widely used to treat inflammatory conditions because of viral invasion, the dedicated steroid like compounds data base was screened and three compounds 0449-0045, N001-0004, N006-0008 were identified with potential NSP-15 binding. Also Cortisone and Dexamethasone were included for Induced-Fit-docking and the MD simulation studies. All the compounds were found to bind with NSP-15 at the active site with favorable binding energetics and conformation. The residue SER294 which was implicated in the catalytic mechanism was having interaction with all compounds and hence these compounds may stop the nucleophilic attack on the RNA substrate. MD simulation and trajectory analyses established the stability of protein–ligand interactions. As revealed by favorable glide score, glide energy, and stable interactions during simulation, the identified steroids are proposed as anti-viral agents against SARS-CoV-2. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 5400 KB) Acknowledgements Anantha Krishnan Dhanabalan, thank the Indian Council of Medical Research, India for Senior Research Fellowship (Grant No: ISRM/11(69)/2017). Authors thank DBT Bioinformatics Infrastructure Facility, University of Madras and CAS in Crystallography and Biophysics for computing facility. Funding The authors declare that no funds, Grants, or other support were received during the preparation of this manuscript. Data availability All data generated or analyzed during this study are included in this published article. 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 1. Yang J Fan L Miao X Corticosteroids for the treatment of human infection with in fl uenza virus : a systematic review and meta-analysis Clin Microbiol Infect 2015 21 956 963 10.1016/j.cmi.2015.06.022 26123860 2. Matsuyama S Kawase M Nao N The inhaled steroid ciclesonide blocks SARS-CoV-2 RNA replication by targeting the viral replication-transcription complex in cultured cells J Virol 2020 10.1128/JVI.01648-20 33055254 3. Kim Y Jedrzejczak R Maltseva NI Crystal structure of Nsp15 endoribonuclease NendoU from SARS-CoV-2 Protein Sci 2020 29 1596 1605 10.1002/pro.3873 32304108 4. Liu X Fang P Fang L Porcine deltacoronavirus nsp15 antagonizes interferon-β production independently of its endoribonuclease activity Mol Immunol 2019 114 100 107 10.1016/j.molimm.2019.07.003 31351410 5. Kim Y Jedrzejczak R Maltseva NI Crystal structure of Nsp15 endoribonuclease NendoU from SARS-CoV -2 Protein Sci 2020 29 1596 1605 10.1002/pro.3873 32304108 6. Sastry GM Adzhigirey M Day T Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments J Comput Aided Mol Des 2013 27 221 234 10.1007/s10822-013-9644-8 23579614 7. Jacobson MP Friesner RA Xiang Z Honig B On the role of the crystal environment in determining protein side-chain conformations J Mol Biol 2002 320 597 608 10.1016/S0022-2836(02)00470-9 12096912 8. Shelley JC Cholleti A Frye LL Epik: a software program for pK a prediction and protonation state generation for drug-like molecules J Comput Aided Mol Des 2007 21 681 691 10.1007/s10822-007-9133-z 17899391 9. Jorgensen WL Maxwell DS Tirado-Rives J Development and testing of the OPLS all-atom force field on conformational energetics and properties of organic liquids J Am Chem Soc 1996 118 11225 11236 10.1021/ja9621760 10. Schrödinger Release 2014–2 (2014) LigPrep, Schrödinger, LLC, New York, NY, 2014–2 11. Friesner RA Murphy RB Repasky MP Extra precision glide: docking and scoring incorporating a model of hydrophobic enclosure for protein−ligand complexes J Med Chem 2006 49 6177 6196 10.1021/jm051256o 17034125 12. Sherman W Beard HS Farid R Use of an induced fit receptor structure in virtual screening Chem Biol Drug Des 2006 67 83 84 10.1111/j.1747-0285.2005.00327.x 16492153 13. Farid R Day T Friesner RA Pearlstein RA New insights about HERG blockade obtained from protein modeling, potential energy mapping, and docking studies Bioorg Med Chem 2006 14 3160 3173 10.1016/j.bmc.2005.12.032 16413785 14. Holm L Laakso LM Dali server update Nucleic Acids Res 2016 44 W351 W355 10.1093/nar/gkw357 27131377 15 Wallace AC Laskowski RA Thornton JM LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions Protein Eng Des Sel 1995 8 127 134 10.1093/protein/8.2.127 16. Bussi G Donadio D Parrinello M Canonical sampling through velocity rescaling J Chem Phys 2007 126 014101 10.1063/1.2408420 17212484 17. Parrinello M Rahman A Polymorphic transitions in single crystals: a new molecular dynamics method J Appl Phys 1981 52 7182 7190 10.1063/1.328693 18. Abraham MJ Murtola T Schulz R GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers SoftwareX 2015 1–2 19 25 10.1016/j.softx.2015.06.001 19. Amadei A Linssen ABM Berendsen HJC Essential dynamics of proteins Proteins Struct Funct Genet 1993 17 412 425 10.1002/prot.340170408 8108382 20. Roe DR Cheatham TE PTRAJ and CPPTRAJ: software for processing and analysis of molecular dynamics trajectory data J Chem Theory Comput 2013 9 3084 3095 10.1021/ct400341p 26583988 21. Bello M Binding free energy calculations between bovine β-lactoglobulin and four fatty acids using the MMGBSA method Biopolymers 2014 101 1010 1018 10.1002/bip.22483 24619557 22. Ioakimidis L Thoukydidis L Mirza A Benchmarking the reliability of QikProp. correlation between experimental and predicted values QSAR Comb Sci 2008 27 445 456 10.1002/qsar.200730051	36456773	PMC9715282	NO-CC CODE	2022-12-03 23:20:15	no	Mol Divers. 2022 Dec 1;:1-14	utf-8	Mol Divers	2,022	10.1007/s11030-022-10576-5	oa_other
==== Front J Bus Res J Bus Res Journal of Business Research 0148-2963 0148-2963 Elsevier Inc. S0148-2963(22)00949-3 10.1016/j.jbusres.2022.113484 113484 Article Exploring the factors that affect user experience in mobile-health applications: A text-mining and machine-learning approach Pal Shounak a Biswas Baidyanath b Gupta Rohit c Kumar Ajay d⁎ Gupta Shivam e a PricewaterhouseCoopers Private Limited, India b Enterprise and Innovation Group, DCU Business School, Dublin City University, Ireland c Operations Management Area, Indian Institute of Management, Ranchi, India d AIM Research Center on Artificial Intelligence in Value Creation, EMLYON Business School, Ecully, France e Department of Information Systems, Supply Chain Management & Decision Support, NEOMA Business School, Reims, France ⁎ Corresponding author. 2 12 2022 2 2023 2 12 2022 156 113484113484 18 2 2022 18 11 2022 19 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. Recent years have witnessed an increased demand for mobile health (mHealth) platforms owing to the COVID-19 pandemic and preference for doorstep delivery. However, factors impacting user experiences and satisfaction levels across these platforms, using customer reviews, are still largely unexplored in academic research. The empirical framework we proposed in this paper addressed this research gap by analysing unmonitored user comments for some popular mHealth platforms. Using topic-modelling techniques, we identified the impacting factors (predictors) and categorised them into two major dimensions based on strategic adoption and motivational association. Findings from our study suggest that time and money, convenience, responsiveness, and availability emerge as significant predictors for delivering a positive user experience on m-health platforms. Next, we identified substantial moderating effects of review polarity on the predictors related to brand association and hedonic motivation, such as online booking and video consultation. Further, we also identified the top predictors for successful user experience across these platforms. Recommendations from our study will benefit business managers by offering an improved service design leading to higher user satisfaction across these m-health platforms. Keywords Mobile health Service-dominant logic Machine learning Text analytics Proportional-odds logit ==== Body pmc1 Introduction and motivation Mobile-healthcare (mHealth) applications are mobile-based platforms that allow providers (i.e., pharmacies, doctors, healthcare institutions, and clinical laboratories) to reach the consumers (i.e., patients). They offer services for chronic conditions, remote monitoring, patient data, electronic records, and prescriptions, sometimes including fitness and wellness applications. According to a Statista Report1 , the global mHealth market was estimated at USD 23 billion in 2016 and is expected to reach nearly USD 190 billion by 2025. The World Health Organization has also emphasised “patient empowerment” among the significant goals in their Health Policy Framework2 , thereby supporting the increased usage of mHealth applications that allow more patient control. In this manner, mHealth platforms expect to improve the communication between healthcare providers and patients, increase their satisfaction levels, and lead to higher success rates of health-related outcomes (Lamprinos et al., 2016, Tartaglione et al., 2018, Wakefield et al., 2018, Ghose et al., 2021, Yang et al., 2022). Nevertheless, several challenges still plague mHealth apps, affecting the overall experience the customers (i.e., patients) enjoy during usage. In a recent survey conducted by Accenture3 across 1800 people in the United States, the users reported that they would be more likely to adopt mHealth platforms only if there were adequate customer support (i.e., brand association), recommendations from the health provider (i.e., perceived quality), or if these technologies enabled them to receive better information about their health (i.e., utilitarian motivation). Using Fig. 1 , we explain a few of these constructs built using keywords extracted from the electronic word-of-mouth (eWOM) (i.e., online reviews) submitted for Practo App in India. These reasons motivate mHealth apps users to post reviews and final star ratings as an outcome of their overall “user experience”.Fig. 2. .Fig. 1 Online reviews for Practo mobile app demonstrate various sources of motivation for the user. Fig. 2 Factors that influence unwillingness to use remote clinical trials (Adapted from PWC Global Top Health Industry Issues 2021)(PwC Global Top Health Industry Issues 2021: https://www.pwc.com/gx/en/industries/healthcare/top-health-industry-issues.html). Past literature has conclusively identified the impact of hedonic motivation that is observed from a hassle-free consultation and medicine delivery booking (Kim and Hwang, 2012, Dwivedi et al., 2016, Alam et al., 2020a, Biduski et al., 2020, Ashraf et al., 2021, Tran et al., 2021), on a consumer’s choice of mHealth app. The past studies also observed utilitarian motivations such as untethered payment and refund system, time and cost-effectiveness ( Kim and Hwang, 2012, Dwivedi et al., 2016, Ashraf et al., 2021, Tran et al., 2021) are also good motivators for app usage. Social motivation (Tran et al., 2021) and perceived quality are also antecedents of mHealth service quality. Our study reiterates the significance of these motivators (from a user’s perspective) while linking the strategic features (from a firm’s perspective) that an app requires to develop an overall conducive “user experience” (measured by the stars received on a mHealth app). Fig. 1 shows that reviews on certain features have stronger sentiments than others (Chatterjee, Goyal, Prakash, & Sharma, 2021). Also, some features of the mHealth apps were more critical to a given group of consumers than others (Tran et al., 2021). Extant studies have demonstrated that the sentiment polarity of eWOMs and social media platforms is a better measure of the direction of the sentiment as well as its strength rather than using the raw values of positive and negative polarities (Stieglitz and Dang-Xuan, 2013, Salehan and Kim, 2016). Such adjustment in the measurement of sentiment polarity prevents the unbalanced effect of one particular polarity (such as highly positive or highly negative) that can be present in the eWOMs related to the mHealth apps. Further, according to the self-construal theory, the variations in self-construals can generate different levels of user motivation to use mHealth apps and lead to differential levels of user satisfaction. So, independent users are typically less dependent upon others and rely upon their internal thoughts, feelings, and actions for motivations. For instance, independent users might write the following eWOMs to express their personal experiences, such as “A fantastic doctor I personally want to recommend” or “My refund not sent”. On the contrary, interdependent users value social relationships and emphasise “fitting in” with other users. Therefore, interdependent users express their interconnectedness with a larger group of users for a mHealth app and might write: “Got doctor's consultation within 10 mins when my brother-in-law was in severe pain and needed doctors’ consultation. We could not go to the hospital due to Covid. This app turned out to be very helpful. Thanks, DocsApp! Cheers”. Our study aims to explain these differential ratings across various features and user groups by proposing two predictors - effective polarity (measured by the difference between “positive” and “negative” polarity) and reviewer-type (independent vs self-construal), thereby allowing scholars to measure the impact of each feature on user rating accurately. 1.1 Identification of gaps in the extant literature A large body of literature has applied the theoretical frameworks of the Technology Acceptance Model (TAM), Unified Theory of Acceptance and Use of Technology (UTAUT and UTAUT2), and Expectation Confirmation Theory (ECT) to examine the adoption and usage continuance of mHealth applications across users (Wu et al., 2011, Lim et al., 2011, Gao et al., 2015, Chang et al., 2016, Cho, 2016, Dwivedi et al., 2016, Hoque and Sorwar, 2017, Kim et al., 2019, Alam et al., 2020a, Alam et al., 2020b, Chiu et al., 2020, Cho et al., 2020, Lim et al., 2021, Nusairat et al., 2021, Prakash et al., 2021, Suroso and Sukmoro, 2021). However, these theoretical frameworks overlooked the growing idea of shared decision-making among the caregivers (e.g., healthcare institutions, doctors) and care receivers (e.g., customers, patients) of mHealth applications. Recently, a growing body of literature (Tartaglione et al., 2018, Russo et al., 2019, Balta et al., 2021, Dahl et al., 2021, Shirazi et al., 2021) examined mHealth applications and the factors responsible for their successful adoption with the application of service-dominant (SD) logic (Vargo and Lusch, 2004, Barrett et al., 2015). This unique view of looking at mHealth apps also matches the call for more scholarly studies by Stocchi, Pourazad, Michaelidou, Tanusondjaja, and Harrigan (2021) to develop “frameworks of value creation, value fusion, and value co-creation (including SD Logic) for apps and via apps.”. Therefore, applying SD Logic in this study will enable us to understand the users’ motivation (i.e., hedonic, utilitarian, or social) and observe patients’ reactions by examining eWOMs. Here, the consumers (i.e., patients) transform themselves into central actors through ongoing information acquisition and exchange with their health providers before, during, and after service encounters (McKinley and Wright, 2014, Dahl et al., 2021). Some studies have also identified “poor user experience” as a significant impediment to repeated usage and continuance of mHealth applications among users (Akter et al., 2013b, Kim et al., 2019). Therefore, examining the significant dimensions of “user experience” and improving them will ensure sustainable growth for these apps through increased usage and user base. This study also addressed a methodological gap in the past literature. Primarily, past researchers mainly relied upon user-based surveys (Azad-Khaneghah, Neubauer, Miguel Cruz, & Liu, 2021) to identify and examine various constructs of user satisfaction, popularity, and adoption of mHealth platforms, such as motivation (Tran et al., 2021), facilitating conditions (Alam et al., 2020a, Alam et al., 2020b), perceived service quality (Akter et al., 2013a, Akter et al., 2013b), service usability (Sadegh, Saadat, Sepehri, & Assadi, 2018), personalisation (Birkmeyer, Wirtz, & Langer, 2021), social networking (Birkmeyer et al., 2021), quality of care (Jannati, Nakhaee, Yazdi-Feyzabadi, & Tjondronegoro, 2021), privacy-protection mechanisms (Rodríguez-Priego, Porcu, & Kitchen, 2022), usage intensity (Veríssimo, 2018), cognitive and social factors (Nusairat et al., 2021, Rahimi and Khoundabi, 2021). However, the excessive reliance of past scholarly studies on surveys as a data collection mechanism poses a methodological challenge by ignoring the vast repository of eWOM data from mHealth apps. Also the use of online reviews from mHealth platforms can serve as an advantage because they are collected from a broad sample pool and therefore diminish the associated instrument bias that could weaken the accuracy of the results (Müller, Junglas, Brocke, & Debortoli, 2016). 1.2 Role of COVID-19 pandemic as a motivator of mHealth app usage The global health and financial crisis caused by the COVID-19 pandemic played a pivotal role in facilitating the growth of mHealth apps and reshaping the features that influence overall “user experiences”. For instance, the Zoom app generated revenues of USD 2.65 billion in 2020, followed by a 55 per cent rise to USD 4.10 billion. Subsequently, the lockdown of essential services that led to a rush to adopt e-commerce channels across all significant businesses4 influenced the healthcare business, too (e.g., mHealth and telemedicine services) (Webster, 2020). According to the Global Health Industry Report published by PwC 5 , 51 % of respondents had received virtual medical care (via mobile, email or text) during the quarantine periods, with 91 % willing to do so again in the future. In this manner, COVID-19 created an inflexion point for consumers to adopt and use mHealth applications. Incorporating COVID-19 as a motivator to examine mHealth apps and related eWOM data from the related timeline also allows us to align this study with the recent calls from high-impact journals in business management - “services” theme (Hashemi, Rajabi, & Brashear-Alejandro, 2022); “COVID-19 & digital healthcare” (Verma & Gustafsson, 2020); consumer-centred healthcare (Kraus, Schiavone, Pluzhnikova, & Invernizzi, 2021); “home healthcare services” (Tsiotsou & Boukis, 2022); “scaling virtual contactless services” (Rai, 2020); “information systems value and success in the context of the COVID-19 pandemic” (Ågerfalk, Conboy, & Myers, 2020). Therefore, in this study, we address the abovementioned research gaps by answering the following three primary research questions:RQ1: What are the major features that impact the user experience in a mHealth platform during the pandemic? RQ2a: How does a review’s effective polarity influence the impact of application features on users’ experience? RQ2b: To what extent does reviewer type (independent/self-construal) influence the impact of hedonic motivation on user experience? RQ3: What strategies can a mHealth platform adopt to improve its users’ experience based on reviews during the pandemic? Our study answers the above research questions by performing an in-depth analysis of users’ reviews. We studied the factors influencing user experience based on an extensive literature review. The factors are then used to perform a textual analysis of reviews we collected from multiple mHealth apps. The resultant impact is then revealed by the relationship between the factors, reviews and user ratings. Our study helps in providing guidelines to mobile app developers, technology consultants, and researchers to improve these mHealth platforms by addressing the existing concerns and enhancing the users’ experience. The remaining part of the paper is structured as follows. Section 2 summarises the background literature on mHealth applications and their theoretical foundations. Section 3 explains the collection of the research data and methodology. Section 4 presents the empirical modelling adopted in our study. Section 5 presents the results, while Section 6 discusses its significant findings. Section 7 discusses the implications of the study while Section 8 concludes the paper and offers the scope for future research. Section 7 concludes the paper and offers the scope for future research. 2 Literature review and theoretical foundation We examined prior academic works on mHealth before and after the pandemic. Most of the research studies before 2020 have been focused on technology adoption, while most of the research work after 2020 is based on value co-creation and customer rating. These findings clearly show a shift in healthcare IT research after 2020, from adoption-centric to consumer-centric. Kraus et al. (2021) claimed that operational efficiencies and a patient-centric approach could achieve digital transformation in healthcare. Operational efficiency is obtained through resource quality and responsiveness, while the patient-centric approach includes hedonic, social, and utilitarian motivation. Verma and Gustafsson (2020) also highlighted the effect of COVID-19 on digital healthcare (i.e., telemedicine, robust surveillance and wearable devices, and diagnostic and clinical decision-making technologies) that will lead to new operating models to cope with the changing demand pattern by remaining agile and productive. Recent literature reviews are also available that (i) examine mHealth apps for individuals with chronic conditions and diseases (Lorca-Cabrera et al., 2021), (ii) compare measurement scales that were used to evaluate the usability and quality of mHealth apps (Azad-Khaneghah et al., 2021), (iii) mHealth interventions (Yang & Kovarik, 2021) and (iv) identify the factors of acceptance and user behaviour of mHealth users (Wang & Qi, 2021). 2.1 Mobile applications for healthcare Scholarly research in healthcare information systems has focused on continued usage and behavioural intention to use mHealth apps. The academic research published before 2020 is mostly focused on “actual usage” (Alam et al., 2020a, Azad-Khaneghah et al., 2021, Veríssimo, 2018) and mHealth app evaluations (Nouri et al., 2018, Sadegh et al., 2018), while the focus on value co-creation (Balta et al., 2021, Shirazi et al., 2021) and customer satisfaction (Chatterjee et al., 2021, Jannati et al., 2021) is relatively new. Survey-based research and literature review have shown extensive dominance in terms of methodology, while text-mining and sentiment analysis have been applied only in a few research studies (Chatterjee et al., 2021, Shah et al., 2021). Similarly, owing to the extensive development in application offering based on sharing economy, research on service-dominant logic (S-D logic) has come to the forefront (Peltier et al., 2020, Tran et al., 2021), while the Technology Acceptance Model (TAM) (Lim et al., 2011, Birkmeyer et al., 2021, Nusairat et al., 2021), Unified Theory of Acceptance and Use of Technology (UTAUT) (Dwivedi et al., 2016, Suroso and Sukmoro, 2021, Wu et al., 2022, Zhang et al., 2022), Service Quality theory (SERVQUAL) (Akter et al., 2013a, Akter et al., 2013b, Shah et al., 2021) continue to be a favourable topic for researchers. Some other theories that were discussed on improving mHealth app usage are Expectation–Confirmation model (ECM) (Chiu et al., 2020, Prakash et al., 2021), Resource-Based View (), Theory of Planned Behavior (TPB) (Kim et al., 2019, Stocchi et al., 2021), Protection Motivation Theory (PMT) (McKinley and Wright, 2014, Gao et al., 2015). We have also analysed existing research to understand the coveted features of users’ expectations from a mHealth application. The most important factor remains the app's overall effectiveness, perceived usefulness and performance (Chang et al., 2016, Veríssimo, 2018, Kim et al., 2019, Biswas et al., 2021, Kraus et al., 2021). However, factors such as (i) trust (Prakash et al., 2021, Wu et al., 2022), (ii) perceived security (Zhao, Ni, & Zhou, 2018), (iii) ease of use (Cho et al., 2020, Biswas et al., 2021), (iv) self-efficacy and control (Dwivedi et al., 2016, Tsiotsou and Boukis, 2022), (v) personalisation (Gimpel, Manner-Romberg, Schmied, & Winkler, 2021), (vi) user satisfaction (Akter et al., 2013a, Akter et al., 2013b, Birkmeyer et al., 2021) and (vii) hedonic and utilitarian motivation (Kim et al., 2019, Tran et al., 2021) have gained importance in the overall evaluation of applications. Customer sentiment is highly important in almost all research (Chatterjee et al., 2021) using eWOMs. User perception about apps’ features that are identified as effective polarity thus influences their motivation to use an application. In Table 1 , we also observe that customer characteristics influence user acceptance of mHealth apps. Further, Tran et al. (2021) explained users with an independent self-construct are more likely to be impacted by hedonic motivations.Table 1 Summary of factors from the existing literature on mHealth applications. Features Period Literature Support Time Pre-COVID Miró and Llorens-Vernet, 2021, Dwivedi et al., 2016 Post-COVID Balta et al., 2021, Chatterjee et al., 2021 Convenience Pre-COVID Azad-Khaneghah et al., 2021, Miró and Llorens-Vernet, 2021 Post-COVID Alam et al., 2020a, Alam et al., 2020b, Balta et al., 2021, Jannati et al., 2021, Webster, 2020, Stocchi et al., 2021 Money / Value Pre-COVID Dwivedi et al., 2016, Nouri et al., 2018 Post-COVID Alam et al., 2020a, Alam et al., 2020b, Chatterjee et al., 2021, Chiu et al., 2020, Cho et al., 2020, Webster, 2020 Transparency / Trust (Refund, Information Security) Pre-COVID Gao et al., 2015, Nouri et al., 2018, Nusairat et al., 2021, Zhao et al., 2018, Sadegh et al., 2018, Kim et al., 2019 Post-COVID Balta et al., 2021, Biswas et al., 2021, Gimpel et al., 2021, Prakash et al., 2021, Wu et al., 2022, Zhang et al., 2022 COVID-19 (lockdown) Pre-COVID Not Applicable Post-COVID Jannati et al., 2021, Webster, 2020 Responsiveness Pre-COVID Azad-Khaneghah et al., 2021, Miró and Llorens-Vernet, 2021, Tan and Yan, 2020 Post-COVID Balta et al., 2021, Chatterjee et al., 2021, Dahl et al., 2021, Gimpel et al., 2021, Paramita and Noviarisanti, 2021, Shah et al., 2021 Effectiveness / Performance / Perceived usefulness Pre-COVID Lim et al., 2011, Akter et al., 2013a, Akter et al., 2013b, McKinley and Wright, 2014, Gao et al., 2015, Chang et al., 2016, Cho, 2016, Dwivedi et al., 2016, Hoque and Sorwar, 2017, Veríssimo, 2018, Zhao et al., 2018, Pal, Mukhopadhyay, and Shukla (2018), Kim et al., 2019, Miró and Llorens-Vernet, 2021 Post-COVID Alam et al., 2020a, Alam et al., 2020b, Chiu et al., 2020, Birkmeyer et al., 2021, Biswas et al., 2021, Chatterjee et al., 2021, Jannati et al., 2021, Kraus et al., 2021, Prakash et al., 2021, Shah et al., 2021 Resource Quality / Reliability Pre-COVID Akter et al., 2013a, Akter et al., 2013b), Nouri et al., 2018, Sadegh et al., 2018, Kim et al., 2019, Kim et al., 2019 Post-COVID Alam et al., 2020a, Alam et al., 2020b, Balta et al., 2021, Biswas et al., 2021, Dahl et al., 2021, Gimpel et al., 2021, Jannati et al., 2021, Paramita and Noviarisanti, 2021, Shirazi et al., 2021, Shah et al., 2021, Tan and Yan, 2020, Tsiotsou and Boukis, 2022 Ease of use (Effort) / App Design Pre-COVID Azad-Khaneghah et al., 2021, Lim et al., 2011, Gao et al., 2015, Chang et al., 2016, Cho, 2016, Dwivedi et al., 2016, Hoque and Sorwar, 2017, Nouri et al., 2018, Tan and Yan, 2020, Veríssimo, 2018, Zhao et al., 2018, Kim et al., 2019 Post-COVID Birkmeyer et al., 2021, Biswas et al., 2021, Cho et al., 2020, Kraus et al., 2021, Stocchi et al., 2021, Suroso and Sukmoro, 2021, Tsiotsou and Boukis, 2022, Wu et al., 2022 Customer support Pre-COVID Miró and Llorens-Vernet (2021) Post-COVID Chatterjee et al., 2021, Kraus et al., 2021, Tsiotsou and Boukis, 2022 Resource Availability Pre-COVID Dwivedi et al. (2015) Post-COVID Peltier et al., 2020, Balta et al., 2021, Biswas et al., 2021, Chatterjee et al., 2021, Paramita and Noviarisanti, 2021 Customer characteristics (Reviewer Type) Pre-COVID Lim et al., 2011, Hoque and Sorwar, 2017, Veríssimo, 2018, Zhao et al., 2018 Post-COVID Alam et al., 2020a, Alam et al., 2020b, Chatterjee et al., 2021, Stocchi et al., 2021, Tran et al., 2021 Self-efficacy / Confidence Pre-COVID Lim et al., 2011, Akter et al., 2013a, Akter et al., 2013b), McKinley and Wright, 2014, Gao et al., 2015, Chang et al., 2016, Dwivedi et al., 2016, Tartaglione et al., 2018, Zhao et al., 2018 Post-COVID Alam et al., 2020a, Alam et al., 2020b, Gimpel et al., 2021, Li et al., 2020, Shirazi et al., 2021, Tsiotsou and Boukis, 2022, Wu et al., 2022, Zhang et al., 2022 Sentiment (WOM) (Polarity) Pre-COVID Chang et al. (2016) Post-COVID Chatterjee et al., 2021, Paramita and Noviarisanti, 2021, Zhang et al., 2022 Emotion / Satisfaction/ Motivation (Hedonic) Pre-COVID Kim et al., 2019, Akter et al., 2013a, Akter et al., 2013b), Cho, 2016, Gao et al., 2015, Tartaglione et al., 2018, Kim et al., 2019, Wang and Qi, 2021 Post-COVID Birkmeyer et al., 2021, Chatterjee et al., 2021, Chiu et al., 2020, Li et al., 2020, Prakash et al., 2021, Stocchi et al., 2021, Suroso and Sukmoro, 2021 Social Influence Pre-COVID McKinley and Wright, 2014, Gao et al., 2015, Chang et al., 2016, Cho, 2016, Dwivedi et al., 2016, Hoque and Sorwar, 2017, Tan and Yan, 2020, Tartaglione et al., 2018, Veríssimo, 2018, Wang and Qi, 2021 Post-COVID Gimpel et al., 2021, Jannati et al., 2021, Nusairat et al., 2021, Shirazi et al., 2021, Suroso and Sukmoro, 2021, Wu et al., 2022, Zhang et al., 2022 Competition Pre-COVID Post-COVID Cho et al., 2020, Chiu et al., 2020, Li et al., 2020, Peltier et al., 2020, Tsiotsou and Boukis, 2022 Further, we observed that certain factors have specifically gained importance in the post-2020 pandemic period. Factors such as (i) convenience (Balta et al., 2021, Jannati et al., 2021), (ii) responsiveness (Dahl et al., 2021, Gimpel et al., 2021), (iii) sentiment analysis (Chatterjee et al., 2021, Zhang et al., 2022), (iv) customer support (Kraus et al., 2021, Tsiotsou and Boukis, 2022), and (v) competition (Li et al., 2020, Chiu et al., 2020) have obtained considerable significance in the success of mHealth apps post COVID-19 pandemic. The impact of the pandemic, however, has never been taken as a significant contributing factor to the digital transformation of healthcare (Webster, 2020, Verma & Gustafsson, 2020). Further, Stocchi et al. (2021) mentioned a pressing need for “new frameworks outlining and evaluating strategies for apps’ introduction.” Table 1 classifies existing literature into pre-COVID and post-COVID eras and their discussion topics. Our study has considered these factors while developing a theoretical model that categorises the discussion themes in most reviews. Our work contributes to the text mining-based research in determining mHealth app features. Earlier, Lim et al. (2021) worked on ride-sharing app services where semi-structured interviews, text-mining and topic modelling of app reviews and PLS-SEM on passenger survey responses were used to obtain the essential mHealth app features. Shah et al. (2021) used text-mining, topic modelling and sentiment analysis to create a strategic SWOT framework for healthcare organisations to improve patient satisfaction. Tran et al. (2021) obtained features based on a survey for S-D logic-based research to enhance the brand equity of branded apps. Our study also aims to contribute further using the factor’s coefficient values comparing a mHealth app against its competitors. This objective is achieved by developing a score for the mHealth apps based on “three different sharing economy differentiation strategies” (Frey, Trenz, & Veit, 2019), i.e., technology, partnership and user experience. Thus, our study contributes to mHealth app managers by identifying the essential features that impact user ratings and presenting a strategic view of their position compared to the competitors. 2.2 Theoretical Foundation: Value co-creation in mobile-health platforms In this study, we apply the theoretical framework of the “Service-Dominant” (S-D) Logic (Vargo and Lusch, 2004, Vargo and Lusch, 2008a, Vargo and Lusch, 2008b) to understand the value co-creation activities that occur when a typical user is consuming the services of mHealth applications by ordering medicines, healthcare services such as telemedicine consultations and diagnostic tests. Vargo and Lusch (2008b) have suggested that S-D logic could provide the foundation for a revised theory of the firm (and other resource-integrating activities), a theory of service systems (Maglio & Spohrer, 2008), and a revised theory of economics and society. We look at how S-D Logic helps us to re-look at these mHealth applications and the related activities that involve multiple actors, including the user(s), the mHealth platform and the caregiver (doctors/consultants). On the one hand, traditional healthcare services that are designed around the “goods-dominant” (G-D) logic (Vargo and Lusch, 2004, Vargo and Lusch, 2008a, Vargo and Lusch, 2008b), where the patient (or care-receiver) visits the physical premises of the healthcare service provider such as the pharmacy or the hospital to receive medical services such as advice, prescription drugs or treatments. According to the G-D logic, the economic exchange of goods (here medicines and related products) acquire value during their design and manufacturing (Vargo & Lusch, 2004). Ideally, in G-D logic, “this output is tangible, produced away (separate) from the interference of customers, standardisable and capable of being inventoried until sold, all to enable maximum efficiency in operations” (see p. 34, Vargo & Akaka, 2009). On the other hand, are the mHealth applications where the technology (i.e. the mHealth platform) enhances the “value co-creation” of the healthcare services offered to the end-user. The next level of value co-creation takes place when the supplier (or the m-health platform and service provider) and the customer (or the mHealth user) join in a reciprocally symbiotic value-exchange mechanism (Sheth, 2019). Further, the foundational premises of S-D Logic (Vargo & Akaka, 2009), i.e., “FP4: Operant resources are the fundamental source of competitive advantage” and “FP6: The customer is always a co-creator of value”, clearly reflect the theoretical motivation behind our research. Therefore, the mHealth platform is not only meant to offer its resources and collaborate with the users but also to co-create value based on the user's approval. Now, let us look at the constructs of our study through the theoretical lens of S-D Logic. Each construct we define in our study builds upon this idea of “value co-creation” rather than the process of singular creation and delivery of value to the customer. For instance, the “utilitarian motivation” construct consists of the following variables: “time and money”, “payment”, “convenience”, “refund”, and “lockdown”. Among those variables, “time and money” is measured by how the mHealth App usage helps to save the user’s time and money and, therefore, includes the perception of the cost of service - which directly links to both the user and the firm (or the mHealth platform). Similarly, the construct “perceived quality” consists of the following variables - “responsiveness”, “effectiveness”, and “resource quality.” Now, “responsiveness” is measured by the agility of the responses by the firm’s resources (such as doctors and clinicians) to user’s needs with diligence - which is again a combination of the values co-created by the user and the firm (or the mHealth platform). The proposed conceptual framework for this study depicting all the constructs and their associated variables is presented in Fig. 3 .Fig. 3 Conceptual framework to explore the predictors of user experience in mHealth applications. Finally, recent academic research (Tartaglione et al., 2018, Russo et al., 2019, Balta et al., 2021, Dahl et al., 2021, Pop et al., 2018, Shirazi et al., 2021) suggests that the “user experience” and the related “customer value” are now generated from a collaborative ecosystem, unlike the traditional channels. The S-D logic also recognises mHealth platforms as unique service systems that behave as “dynamic value co-creation configurations of resources (people, technology, organisations, and shared information)” (Maglio & Spohrer, 2008: p. 19). Therefore, these mHealth platforms should be ideally viewed as a concerted and synchronously designed “value-in-use” service system (Maglio and Spohrer, 2008, Osei-Frimpong et al., 2018, Russo et al., 2019) instead of an isolated service-offering mechanism (represented by value-in-exchange). 3 Data collection We collected users’ reviews and general information of nine mHealth care applications from India, namely, PharmEasy, Tata 1 MG, Apollo 24 X 7, Practo, Netmeds, Medibuddy, MFine, DocsApp and Tata Health, from the Google Play Store for three months starting from 1st June 2021 until 1st September 2021. Table 2 summarises the details (release date, download count, total review count by users, review count in 90 Days, total versions, rank in India, and last version update) for each of these apps till 1st September 2021. While we began our data collection with a total of 41,007 reviews, we removed 156 non-English reviews (such as in regional languages of India), leading to 40,852 reviews for final analysis.Table 2 Summary of mHealth applications used in our study. App Name Release Date Downloads (in 1000 s) Review Count Reviews in last 90 days* Total Versions Released* Rank in India Last Version Update PharmEasy 06/03/2015 5000–10000 39,700 16,835 6 1 12/08/2021 Tata 1MG 24/08/2012 5000–10000 20,400 8426 3 2 11/08/2021 Apollo 24 X 7 31/01/2020 5000–10000 17,100 5811 7 3 18/08/2021 Practo 15/03/2014 5000–10000 13,500 1963 10 4 03/12/2021 Netmeds 08/07/2015 5000–10000 12,300 4120 2 5 11/08/2021 Medibuddy 07/07/2014 1000–5000 21,600 1724 9 9 06/12/2021 MFine 28/09/2017 1000–5000 6000 187 3 10 03/12/2021 DocsApp 14/04/2015 5000–10000 22,800 1601 4 21 10/06/2021 Tata Health 13/02/2018 100–500 550 340 8 – 29/11/2021 * Data was collected from 1st June 2021 until 1st September 2021 for the analysis Next, we categorised the independent variables for our study based on a bi-dimensional grid, as shown in Table 3 . This grid explains how each variable can be mapped to two unique dimensions simultaneously – motivational association and strategic adoption. The first dimension links to RQ1, while the second dimension links to RQ2 of our study. While motivational association can be of four types, namely (i) utilitarian motivation, (ii) hedonic motivation, (iii) brand support and loyalty, and (iv) perceived quality; strategic adoption can be of three types, namely (i) technology, (ii) partnership, and (iii) user experience.Table 3 Mapping of predictors according to their strategic adoption and motivational associations. Strategic Adoption (Firm) Technology Partnership User Experience Motivational Association (User) Utilitarian Motivation Payment Time and Money Convenience Refund Lockdown Hedonic Motivation Video Consultation Online Booking Brand Support and Loyalty Customer Support Availability Perceived Quality Responsiveness Resource Quality Effectiveness 3.1 Using text mining to create new predictors We applied relevant text-mining techniques to build novel motivational association-based predictors from the reviews received in the mHealth apps. That would help us to generate actionable insights for businesses (Tan, 1999). Text-mining often requires important steps such as cleaning the unstructured text data, pre-processing, and applying the text-mining techniques. In this study, data pre-processing included the following steps - removal of “English” stop-words, removal of punctuation marks (such as /, @, //), and removal of numbers and blank spaces to create the pre-processed corpus. We used the tm package in R to complete the data pre-processing steps and clean the unstructured textual data. For example, the review text “Superb experience in booking appointments always” would read as “superb experi in book appoint always” after pre-processing, text-cleaning, and stemming. Next, we applied the stemming technique with the corpus generated from the review texts to find significant keywords and then used the TF-IDF scores (Salton and Buckley, 1988, Schütze et al., 2008, Biswas et al., 2021). We used the tm package in R to create the term-document matrix based on the TF-IDF scores. After this step, we consulted 12 experts from healthcare and e-commerce domains with varied profiles ranging from UX designers, operations managers, senior data analysts, and data engineers (presented in Table 4 ), who assisted us in creating the various service attributes that we presented in a novel wordlist (shown in Table 5 ). This lexicon consisted of 12 themes and associated word stems for each theme. These themes were mapped as predictors, which are binary (0 and 1), where “1″ indicates that word stem(s) for that particular category has been discussed in the eWOM and “0” indicates otherwise. Therefore, it is understandable that each review can belong to more than one category. In this manner, we built the predictors through data pre-processing and text-mining the reviews collected from the mHealth apps.Table 4 Details of experts for developing the lexicon. No. Domain Profile Avg. Exp. (in Years) 2 Healthcare e-commerce Product and UX Designer 3.5 1 Business Analytics Business Analyst 4 2 Product Management Senior Data Analyst 3 2 Retail E-commerce Lead Data Engineer 6.5 1 Healthcare Supply-chain Operations Manager 5 4 Hospital/Medical Center Senior Manager “Patient Experience” 5.5 Table 5 Mapping of predictors with associated word stems extracted from reviews. Predictor Word stems Time and Money time, monei, experi, onlin, appoint, servic, chat, worst, even, wast, refund, paid, take Payment pay, monei, payment, deduct, paid, experi, ask, servic, chat, custom, need, time, fee, book, use, refund, avail, respond, now, month, talk, fee, free, charg Convenience conveni, find, help, appoint, good, thank, easi, connect, prescript, download Refund refund, monei, cancel, deduct, time, experi, appoint, ask, day, wait, onlin, book, paid, chat, bad, repli, amount, membership Lockdown lockdown, use, covid, pandem, day, thank, great, support, respons, pandem, time, help, situat, realli, helpless, hospit, covid, download Online Booking onlin, book, appoint, consult, call, servic, good, monei, chat, time, experi, even, cancel, worst, connect, prescript Video Consultation video, consult, call, appoint, time, onlin, chat, servic, worst, even, book, bad, messag, connect, virtual Customer Support custom, app, care, servic, consult, support, even, onlin, time, experi, appoint, chat, paid, number, call, contact, medicin, ask, cancel, pathet, day, bad, one, wast, feedback Availability availabl, wait, avail, time, experi, day, onlin, servic, medicin, use, appoint, chat, doc, good, monei, call, connect, download Responsiveness respons, quick, servic, time, experi, chat, custom, ask, give, take, support, proper, day, help, even, monei, question, prescript, feedback Effectiveness effectiv, plan, assign, question, wast, refund, paid, take, recommend, feedback, satisfi Resource Quality resourc, qualiti, avail, consult, nice, app, time, experi, question, medicin, review, fake, fair, wast 3.2 Using sentiment analysis to create moderating variables After building the preliminary empirical models, we focussed on creating the moderating variables by applying relevant text-analysis tools and techniques. Each review text can have a significant positive and negative polarity embedded within itself based on its sentence construction which can significantly influence the relationship of the primary predictors towards the user experience (here measured by the ratings received in the mHealth app). We then applied vader 6 lexicon to extract each review text's positive and negative sentiments. Recent studies have confirmed the efficacy of the vader as a powerful analyser of sentiments from online reviews (Kim et al., 2021, Tripathi et al., 2021). Next, we built a measure of effective polarity (“positive polarity” – “negative polarity”) based on extant studies that have demonstrated its superiority compared to using the raw values of positive and negative polarities separately (Stieglitz and Dang-Xuan, 2013, Salehan and Kim, 2016). Such adjustment in the sentiment polarity measurement prevents one particular polarity’s unbalanced effect. Next, we divided the consumers of the mHealth apps according to self-construal behaviour into independent and interdependent types. The independent users were more concerned about their experience. They often used pronouns such as “I”, “we”, or “you” in sentences to express their personal experiences, such as “A fantastic doctor I personally want to recommend” or “My refund not send”. Reviewers with a more generalised opinion or view of application features are interdependent (i.e. social motivation) (Rahimi and Khoundabi, 2021, Tran et al., 2021), who might write reviews such as “Hello MFine Team, …… due to Covid-19 no doctor was available then I have used your services for my father who was in hometown and symptoms were for stroke but timely help within few minutes from the doctor from Bangalore has helped my family..” We completed the semantic parsing with the LIWC (Linguistic Inquiry and Word Count) tool (Pennebaker, Boyd, Jordan, & Blackburn, 2015). Recent studies have established the validity of LIWC in inferring psychometric properties from text (Biswas et al., 2020, Sengupta et al., 2021). The overall methodology is presented in Fig. 4 . Table 6 and Table 7 present the variables used in this study and their descriptive statistics, respectively.Fig. 4 Methodology. Table 6 Predictors of “user experience” in this study. S. No. Variable Brief Description Literature Source Independent Variable Utilitarian Motivation 1 Time and Money (U) App usage helps save user’s time and money also includes the perception about cost of service (B) Tran et al., 2021 2 Payment (T) The payment system is easy to use (B) Tran et al. (2021) 3 Convenience (U) Use of the application is more convenient than the physical experience of service (B) Developed for this study 4 Refund (U) Refund is received conveniently in case of service failure (B) Developed for this study 5 Lockdown (U) Whether review reflects that app has been helpful during lockdown (B) Developed for this study Hedonic Motivation 6 Online Booking (P) Booking an appointment using the app can be done seamlessly according to review (B) Tran et al. (2021) 7 Video Consultation (T) Video consultation is done seamlessly according to review (B) Developed for this study Brand Support and Loyalty 8 Customer Support (P) Active support from staff is available and effective in solving problems (B) Tran et al. (2021) 9 Availability (P) Whether the services are available or not (B) Tran et al. (2021) Perceived Quality 10 Responsiveness (P) Resources such as doctors and clinicians respond to user’s needs with diligence (B) Tran et al. (2021) 11 Effectiveness (U) Service provided by doctors and labs meet users’ requirements (B) Jannati et al. (2021) 12 Resource Quality (P) Perceived quality of solutions/services received is good (B) Jannati et al. (2021) Moderating Variable 13 Positive Polarity Positive polarity within user’s comment. Reflects user’s positive experience (N) Chatterjee et al. (2021) 14 Negative Polarity Negative polarity within the user’s comment. Reflects user’s negative experience (N) Chatterjee et al. (2021) 15 Reviewer Type If the user is specific towards their personal experience, more is the score (N) Developed for this study Dependent Variable 16 User Rating Review rating assigned by the user after experiencing the app (N) Chatterjee et al. (2021), Tran et al. (2021) T - Technology; P - Partnership; U - User Experience; B - Binary variable (0 or 1); N-Numeric/Continuous variable Table 7 Descriptive statistics for the variables used in this study. N Mean Std. Dev. Max Min Time and Money 40,852 0.562 0.496 1.000 0.000 Payment 40,852 0.681 0.466 1.000 0.000 Convenience 40,852 0.434 0.496 1.000 0.000 Refund 40,852 0.538 0.499 1.000 0.000 Lockdown 40,852 0.595 0.491 1.000 0.000 Online Booking 40,852 0.705 0.456 1.000 0.000 Video Consultation 40,852 0.502 0.500 1.000 0.000 Customer Support 40,852 0.847 0.360 1.000 0.000 Availability 40,852 0.798 0.402 1.000 0.000 Responsiveness 40,852 0.854 0.353 1.000 0.000 Effectiveness 40,852 0.219 0.413 1.000 0.000 Resource Quality 40,852 0.768 0.422 1.000 0.000 Positive Polarity 40,852 0.181 0.194 0.826 0.000 Negative Polarity 40,852 −0.089 0.123 0.000 −0.835 Reviewer Type 40,852 0.097 0.125 1.000 0.000 User Rating 40,852 2.871 1.861 5.000 1.000 We create two separate word clouds from the mHealth reviews - one with positive emotions (Fig. 5a ) and the other with negative emotions (Fig. 5b ). Typically with a word cloud, we understand that the relative size of each keyword is proportional to its frequency of usage in the entire word corpus (Chatterjee et al., 2022, Grover et al., 2021, Malik et al., 2021). Similar to recent literature that has examined social media content by segregation based on emotions, we split the entire corpus of eWOMs based on positive and negative polarities (Chatterjee et al., 2022, Grover et al., 2021, Malik et al., 2021). Therefore, based on Figure 5 (a), we find that users of mHealth apps express their positive emotions with the help of the following keywords - “app”, “good”, “experi”, “help”, “best”, “easi”, “great”, “nice”, “thank” and “excel”. Furthermore, based on Figure 5 (b), we find that users of mHealth apps express their negative emotions with the help of the following keywords - “worst”, “app”, “experi”, “bad”, “waste”, “fake”, “fraud” and “support” (see Fig. 6a. and Fig. 6b. ).Fig. 5a Word Cloud of mHealth reviews - positive emotions. Fig. 5b Word Cloud of mHealth reviews - negative emotions. Fig. 6a Feature Importance Plot from the Linear Regression Model (combined dataset). Fig. 6b Feature Importance Plot from the Proportional-odds Logit Regression Model (combined dataset). 4 Empirical modelling 4.1 Using multiple linear regression for app-ratings We applied multiple linear regression to build exploratory models that fit the rating data received from the mobile healthcare apps. We analysed the data as a whole and sub-category-wise according to each app separately. If the rating received in the mobile app by the jth customer be yj such that yj it follows a linear relationship with its predictors as follows:(1) yj=βj0+βj1Xj1+βj2Xj2+⋯+βj15Xj15∀i=1,2,⋯.15 In Eqn. (1), βj0 is the intercept and βj1, βj2, …., βj15 are the multiple linear regression coefficients. Therefore, the modified equation for our study is as follows:Ratingsreceivedi =β0+β1TimeandMoneyi+β2Paymenti +β3Conveniencei+β4Refundi +β5Lockdowni+β6OnlineBookingi+β7VideoConsultationi+β8CustomerSupporti+β9Availabilityi+β10Responsivenessi +β11Effectivenessi+β12ResourceQualityi+β13PositivePolarityi (2) +β14NegativePolarityi+β15ReviewerTypei 4.2 Using proportional odds logit regression for app-ratings Next, we applied the proportional odds logistic regression models due to possible non-linear relationships between the predictors and the dependent variable “rating”, which is categorical with an order of magnitude (values are 1, 2, 3, 4 and 5) (Agresti, 2003). We applied the proportional-logit odds regression in R using the polr package. We applied the proportional odds logistic regression to analyse the combined data and according to each app separately. Therefore, the generalised equation for the proportional-odds logistic regression in our study is as follows:(3) logit[P(Ratingsreceivedi≤j)]=αi-∑i=1MβiXi where j=1,2,⋯,(j-1) and i=1,2,⋯M. Here, j is the level of an ordered category with J levels, and i corresponds to independent variables as follows:j=1 indicates a rating “1″ on the mHealth app; j=2 indicates a rating “2″ on the mHealth app; ….…........... and. j=5 indicates a rating “5″ on the mHealth app. While the independent variables are denoted as follows: i=1 indicates TimeandMoney; i=2 indicates Payment; ….….......; i=15 indicates .ReviewerType 4.3 Moderator-based modelling to build exploratory models Next, we built additional empirical models to examine the moderating effects of “reviewer-type” and the “effective polarity” of the text messages. Therefore, the modified equation with moderator effects for our study is as follows:Ratingsreceivedi =β0+β1TimeandMoneyi+β2Paymenti +β3Conveniencei+β4Refundi +β5Lockdowni+β6OnlineBookingi+β7VideoConsultationi+β8CustomerSupporti+β9Availabilityi+β10Responsivenessi +β11Effectivenessi+β12ResourceQualityi+β13EffectivePolarityi +β14ReviewerTypei+β15TimeandMoneyi∗EffectivePolarityi +β16Conveniencei∗EffectivePolarityi β17Paymenti∗EffectivePolarityi+β18Refundi∗EffectivePolarityi +β19Lockdowni∗EffectivePolarityi +β20OnlineBookingi∗EffectivePolarityi +β21VideoConsultationi∗EffectivePolarityi +β22CustomerSupporti∗EffectivePolarityi +β23Availabilityi∗EffectivePolarityi +β24Responsivenessi∗EffectivePolarityi +β25Effectivenessi∗EffectivePolarityi +β26ResourceQuality∗EffectivePolarityi +β27OnlineBookingi∗ReviewerTypei (4) +β28VideoConsultationi∗ReviewerTypei ##EffectivePolarity=PostivePolarity-NegativePolarity according to Stieglitz and Dang-Xuan (2013). 5 Results We applied an assortment of empirical models in the R environment to estimate the main effects and moderating effects identified in the proposed framework. The detailed results are reported in Tables A1, A2, A3, A4, A5 and A6, available in the online Appendix. 5.1 Feature importance schemes Next, we applied feature importance techniques with the help of the varImp() function from the caret package (see Figure 6). This scheme helped us better understand the effect of the predictors on the response variable, i.e., the “ratings” received by the mobile apps. When we applied the varImp() function in conjunction with the linear regression results, the order of precedence of the predictors was similar to their t-statistic values reported in the regression. 6 Discussion of results We applied various regression techniques to examine the users’ experience measured by the ratings received by the nine mHealth applications. We present the results from the linear regression in Tables A1 and A2 (please see online Appendix) and those from the moderating effects in Tables A3 and A4 (please see online Appendix). Next, we present the results from the proportional-odds logit regression model in Tables A5 and A6 (please see online Appendix). 6.1 Main effects First, among the different measures of utilitarian motivation of mHealth apps, we found that most apps have received significant positive experiences regarding time and money. However, users of Tata Health, MFine, and MediBuddy faced minor issues. Further, payment-related problems included poor payment portals and issues with the refund when the consultations were cancelled or the medicine deliveries were unsuccessful. Thus, users faced problems related to payment and refunds except for Tata Health and Tata 1MG. On the other hand, most satisfied users were impressed by the convenience created by these mHealth apps, leading to an overall positive effect on the ratings. In similar veins, users of all mHealth applications except the DocsApp platform had an overall positive experience during lockdown-related services. Second, among the different measures of hedonic motivation of mHealth apps, our results show that video consultation was an ongoing issue for all platforms and a weak motivation for users of Apollo24x7, DocsApp, Mfine and MediBuddy. Next, we found that mHealth users on most apps had a significant negative experience with online booking except for PharmEasy, Tata Health and Tata 1MG, the only mHealth platforms that provided users with a positive and significant user experience during online bookings. Third, among the different measures of brand support and loyalty for users of mHealth apps, we found that most apps provided significantly poor customer support. However, users of Tata Health and MFine had a relatively better experience with customer support. Next, availability was highly significant and positive for mHealth users across all platforms in our study. Fourth, among the different measures of perceived quality of mHealth apps, we found that users received significant negative experiences regarding responsiveness except those of Tata 1MG and DocsApp, where users faced relatively minor issues. Next, we found that mHealth users on most apps had a poor experience with effectiveness except for TataHealth, which provided users with a strong and effective user experience. Then, while examining the effect of resource quality of mHealth applications, we found that users were mostly satisfied across all platforms. Thus, our study found a significant and positive impact on ratings. 6.2 Moderating effects of effective polarity and reviewer-type Next, we examined the moderating effects of effective polarity on all predictors belonging to utilitarian motivation, perceived quality, hedonic motivation, and brand association, which helped us to determine whether the high polarity content of the reviews could amplify the main effects of these variables across mHealth platforms (Tables A3 and A4, online Appendix). First, we examined the predictors belonging to the utilitarian motivation construct. We found that the effective polarity strongly moderated the effects of (i) time and money towards the user experience for all the mHealth applications except for Netmeds; (ii) payment towards the user experience for all the mHealth applications except for TataHealth; (iii) refund towards the user experience for TataHealth; (iv) convenience towards the user experience for all mHealth apps except for DocsApp, Mfine and Netmeds; (v) lockdown towards the user experience for all mHealth apps except for DocsApp and Mfine. Second, we examined the predictors belonging to the hedonic motivation construct. We found that the effective polarity strongly moderated the effects of (i) online booking towards the user experience for all mHealth applications and, in particular, very strong for TataHealth, Tata 1MG and Apollo24x7; (ii) video consultation towards the user experience for all mHealth applications and, in particular, very strong for TataHealth and PharmEasy. Third, we examined the predictors belonging to the brand support and loyalty construct. We found that the effective polarity strongly moderated the effects of (i) customer support towards the user experience for all the mHealth applications; (ii) availability towards the user experience for all the mHealth applications except for DocsApp, Mfine and Netmeds. Fourth, we examined the predictors belonging to the perceived quality construct. We found that the effective polarity strongly moderated the effects of (i) responsiveness towards the user experience for all the mHealth applications except for Practo; (ii) effectiveness towards the user experience for all the mHealth applications except for Medibuddy and PharmEasy; (iii) resource quality towards the user experience for all the mHealth apps and in particular, very strongly for TataHealth, Mfine, Tata 1MG and Apollo24x7. Fifth, we examined the reviewer type’s moderating effects on the predictors of the hedonic motivation construct. We found that the reviewer type strongly moderated the effects of (i) online booking towards the user experience for all mHealth applications except for Tata 1MG and Netmeds; (ii) video consultation towards the user experience for all mHealth applications and, in particular, very strongly for TataHealth, Mfine, MediBuddy and Tata 1MG . 6.3 Applying the strategy Map: Differentiation strategies for improving user experience on mHealth platforms Next, we presented the effect of overall experience in a 2-by-2 strategy map (see Fig. 7 ) based on Tables A1, A2, A3, A4, A5 and A6 (see online Appendix) across the three dimensions – user-experience, technology and partnerships. Based on the map, we found that most mHealth applications possessed a solid partnership to ensure a positive user experience. However, they remain stressed on providing an excellent technical experience and subsequent user involvement, especially related to payment processing, refunds, customer support, and effectiveness of responses. In the map in Fig. 7, we observe that for most mHealth applications, user perception of partnership factors, i.e., online booking, availability and resource quality, is optimistic. Tata Health, Tata 1MG, Medibuddy and Apollo 24X7 are significant mentions. Furthermore, Netmeds has the weakest user perception in terms of partnerships.Fig. 7 Strategy map showing the nine mHealth apps across the three dimensions. Almost all the apps, except for Tata Health, have poor user perception regarding technology, i.e., ease of payment completion and video consultation quality. The app with the best perception regarding technology is Tata Health, and the worst is MFine. Apps that have developed a good perception in terms of user experience are, PharmEasy, Netmeds, Apollo 24X7 and MFine, in the order of their performance. User experience includes convenience, refund, lockdown impact, time, value for money and effectiveness. Medibuddy has the worst user experience, while other apps also suffer in user experience, especially regarding refund and efficacy. The above findings have shown that most mHealth apps need to improve their technology offerings and also ensure a better experience in terms of refund policies and overall offerings and performance. These mHealth app service firms can use this strategic framework to visualise their strategic position and accordingly monitor the impact of their actions on their competitive position. In addition to the strategic map, we also provide a supporting recommendations table (Frey et al., 2019) that can offer actionable agendas to the mHealth platforms and help them overcome their incumbent challenges (see Table 8 ).Table 8 Strategic recommendations to resolve shortcomings in respective strategic pillars. Technology Partnership User Experience High standardisation External resources needed Low standardisation High efficiency Low budget High usability Regional focus Sharing of expensive or exclusive resources Transactions not planned in advance Low standardisation The quality of human–computer interaction defines the experience Adapted from Frey et al. (2019). 7 Implications of findings 7.1 Contribution to Theory Our study offers two significant theoretical contributions, primarily toward the Value Co-Creation Theory and the differentiation strategies in sharing economy. First, our study identifies a “value co-creation mechanism” using S-D Logic where the mHealth platform and the patients (or users) are engaged through the consumption of mHealth services. Contrary to the incumbent healthcare sectors, where the service design aims essentially at G-D logic, the users’ medical facilities, pharmacies, and other primary interactions generate much consumer value. Our study indicates that customer value primarily depends on the service design and depends less on its pre-existing physical attributes. This finding is the primary theoretical contribution to the existing research on mHealth care applications. Second, our study categorises the identified factors into three differentiation strategies for firms competing in the sharing economy. Mobile health applications depend largely on platform economics to generate successful and enjoyable customer experiences. These firms build partnerships with physicians and clinics that enable patient consultation through mobile platforms. Our study builds a map where it simultaneously draws motivation from the three pillars of strategic adoption and the constructs of motivational association (shown in Table 3). This mapping leads to understanding (i) the current comparative position based on user review and (ii) possible areas of improvement to reach the coveted “aim” where an application meets all the user expectations. 7.2 Contributions to managerial practices Our study offers interesting recommendations for managers and software developers of mHealth applications. First, it proposes an explanatory framework to measure customer ratings given by the mHealth users. It identifies availability, customer support, responsiveness and seamless payment as significant factors for a successful user experience. Using these findings, managers in mHealth firms can now focus their marketing efforts and monitor the specific product features that will help garner positive consumer reactions in the long run (Nouri et al., 2018). The current scenarios in the mHealth market also coincide, where mobile-health firms are continuously trying to improve the customer experience. Their efforts match the high frequency of new version releases as product manufacturers constantly improvise on existing versions to meet the demand of their customers (Yakubu & Kwong, 2021). Second, findings from our study will work as a blueprint for app developers working for these mobile-health firms. Often, app developers and back-end programmers remain disconnected from the front-end users, who are essentially consuming the services of the mHealth app. This scenario may lead to challenges, such as incorrect calculations of the app’s overall rating on the Google Play Store or Apple App Store. Finally, the mHealth app’s popularity may diminish among online users, and its usage may reduce, leading to obsolescence. In this aspect, our study presents the detailed specifications and features (see Figure 6) that could also be used to build visualisation tools such as monitoring dashboards. Businesses widely use such dashboards for application performance monitoring7 and feature importance analysis using text-mining user reviews and ratings. Third, our study identifies a strategy map to help visualise a firm’s relative position and compare its performance with the competitors using the three chosen dimensions – technology, partnership and user experience (see Fig. 7). In this manner, mHealth firms can estimate their relative market positions and plan their business decisions accordingly. In the following sub-section, we explain the map by applying it to the nine mHealth apps in our study and elaborate on their strategy recommendations. 8 Conclusion and future scope of research This study introduces the application of user reviews in identifying features that impact user star ratings. The study also analyses nine active healthcare apps based on their user review and suggests strategic solutions to improve user satisfaction and rating by a healthcare service provider app. We have divided the reviews into the twelve most relevant factors based on literature review and domain knowledge. We observe that a seamless payment system, transparent refund policy, video consultation, and doctor availability through online booking are factors that need serious attention from the service provider. On the other hand, factors such as lockdown/COVID, time and money, resource quality, convenience and responsiveness have significantly influenced digital transformation in healthcare delivery. This study will provide essential insights on factors that influence a user’s experience of a mHealth app. These findings will allow application and business model developers to understand the basic requirements of developing a user-centric app. It will also help researchers by providing them with a methodological guideline for determining new factors in future as the mHealth industry evolves with time. Despite these important findings, our study has a few limitations. First, we examined cross-sectional user reviews that we had scraped from mobile health platforms. Future studies can study the longitudinal effects of the predictors across these platforms and incorporate seasonality and economic shocks such as supply-chain disruptions. Second, we worked with mobile health applications based in India. A consequential extension of this study could also examine mobile healthcare apps across various countries and compare their cultural differences. Third, mixed-method studies could be developed by conducting customer interviews and experiments and combining online reviews as a source of customer experience. CRediT authorship contribution statement Shounak Pal: Writing – original draft, Software, Methodology, Conceptualization. Baidyanath Biswas: Writing – review & editing, Writing – original draft, Visualization, Supervision, Methodology. Rohit Gupta: Writing – review & editing, Validation, Supervision, Resources, Methodology. Ajay Kumar: Writing – review & editing, Validation, Supervision, Conceptualization. Shivam Gupta: Validation, Supervision, Project administration. 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. Dr. Shounak Pal is working as a Senior Consultant at PricewaterhouseCoopers Private Limited India. He received his PhD (FPM) from IIM Lucknow, India. His research interest includes healthcare information systems, digital twin and data science process modelling. Dr. Baidyanath Biswas is an Assistant Professor of Digital Business and Information Systems at DCU Business School. He received his PhD (FPM) from IIM Lucknow, India. His research has appeared in Decision Support Systems, Electronic Markets, Computers in Industrial Engineering, and the Journal of Enterprise Information Management. Baidyanath is also associated with top peer-reviewed international conferences, namely, HICSS and ICIS. He has a rich industry-experience of nine years working as a mainframe and DB2 database analyst at Infosys and IBM. Currently, Baidyanath serves on the editorial boards of the Global Business Review and Electronic Markets journal. Dr. Rohit Gupta is an Assistant Professor of Operations Management Area at Indian Institute of Management Ranchi, Ranchi, India. He received his PhD (FPM) from IIM Lucknow, India. Dr. Ajay Kumar is an Assistant Professor at EMLYON Business School France. His research and teaching interests are in data and text mining, decision support systems, knowledge management, business intelligence and enterprise modeling. He has been Postdoctoral Fellow at Massachusetts Institute of Technology, USA and Harvard Business School, USA. He has published several research papers in reputed journals, including Journal of Business Research, Decision Support Systems, International Journal of Operations & Production Management, International Journal of Production Economics, Industrial Marketing Management, Technological Forecasting & Social Change, Annals of Operation Research, etc. Dr. Shivam Gupta is a full Professor at NEOMA Business School, France and expert in Statistics, Cloud Computing, Big Data Analytics, Artificial Intelligence and Sustainability domains. He has published several research papers in reputed journals, including Decision Support Systems, International Journal of Production Economics, Industrial Marketing Management, Journal of Business Research, Annals of Operation Research, etc. He has been the recipient of the International Young Scientist Award by the National Natural Science Foundation of China (NSFC) in 2017 and winner of the 2017 Emerald South Asia LIS award. Appendix A Supplementary material The following are the Supplementary data to this article:Supplementary data 1 1 Statista Report on Global digital health market by major segment 2015–2025: https://www.statista.com/statistics/387867/value-of-worldwide-digital-health-market-forecast-by-segment/. 2 Health 2020. A European policy framework and strategy for the 21st Century: https://www.euro.who.int/data/assets/pdf_file/0011/199532/Health2020-Long.pdf. 3 Accenture Survey on Digital adoption: Reaction or revolution? https://www.accenture.com/us-en/insights/health/digital-adoption-healthcare-reaction-or-revolution. 4 How COVID-19 triggered the digital and e-commerce turning point:https://unctad.org/news/how-covid-19-triggered-digital-and-e-commerce-turning-point. 5 PWC Global Top Health Industry Issues 2021:https://www.pwc.com/gx/en/industries/healthcare/top-health-industry-issues.html. 6 github “vader” https://github.com/cjhutto/vaderSentiment. 7 What are Application Performance Monitoring Tools? https://www.gartner.com/reviews/market/application-performance-monitoring. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.jbusres.2022.113484. ==== Refs References Ågerfalk P.J. Conboy K. Myers M.D. Information systems in the age of pandemics: COVID-19 and beyond European Journal of Information Systems 29 3 2020 203 207 Akter S. D’Ambra J. Ray P. Development and validation of an instrument to measure user perceived service quality of mHealth Information & Management 50 4 2013 181 195 Akter S. Ray P. D’Ambra J. Continuance of mHealth services at the bottom of the pyramid: The roles of service quality and trust Electronic Markets 23 1 2013 29 47 Alam M.Z. Hu W. Kaium M.A. Hoque M.R. Alam M.M.D. Understanding the determinants of mHealth apps adoption in Bangladesh: A SEM-Neural network approach Technology in Society 61 2020 101255 Alam M.Z. Hoque M.R. Hu W. Barua Z. Factors influencing the adoption of mHealth services in a developing country: A patient-centric study International Journal of Information Management 50 2020 128 143 Agresti A. Categorical data analysis Vol. 482 2003 John Wiley & Sons Ashraf A.R. Tek N.T. Anwar A. Lapa L. Venkatesh V. Perceived values and motivations influencing m-commerce use: A nine-country comparative study International Journal of Information Management 59 2021 102318 Azad-Khaneghah P. Neubauer N. Miguel Cruz A. Liu L. Mobile health app usability and quality rating scales: A systematic review Disability and Rehabilitation: Assistive Technology 16 7 2021 712 721 31910687 Balta M. Valsecchi R. Papadopoulos T. Bourne D.J. Digitalization and co-creation of healthcare value: A case study in Occupational Health Technological Forecasting and Social Change 168 2021 120785 Barrett M. Davidson E. Prabhu J. Vargo S.L. Service innovation in the digital age MIS Quarterly 39 1 2015 135 154 Biduski D. Bellei E.A. Rodriguez J.P.M. Zaina L.A.M. De Marchi A.C.B. Assessing long-term user experience on a mobile health application through an in-app embedded conversation-based questionnaire Computers in Human Behavior 104 2020 106169 Birkmeyer S. Wirtz B.W. Langer P.F. Determinants of mHealth success: An empirical investigation of the user perspective International Journal of Information Management 59 2021 102351 Biswas B. Sengupta P. Chatterjee D. Examining the determinants of the count of customer reviews in peer-to-peer home-sharing platforms using clustering and count regression techniques Decision Support Systems 135 2020 113324 Biswas M. Tania M.H. Kaiser M.S. Kabir R. Mahmud M. Kemal A.A. ACCU3RATE: A mobile health application rating scale based on user reviews PloS one 16 12 2021 e0258050 34914718 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 Chatterjee S. Ghatak A. Nikte R. Gupta S. Kumar A. Measuring SERVQUAL dimensions and their importance for customer-satisfaction using online reviews: A text mining approach Journal of Enterprise Information Management 2022 Chiu W. Cho H. Chi C.G. Consumers' continuance intention to use fitness and health apps: An integration of the expectation–confirmation model and investment model Information Technology & People 2020 Cho J. The impact of post-adoption beliefs on the continued use of health apps International Journal of Medical Informatics 87 2016 75 83 26806714 Cho H. Chi C. Chiu W. Understanding sustained usage of health and fitness apps: Incorporating the technology acceptance model with the investment model Technology in Society 63 2020 101429 Chang I.C. Chou P.C. Yeh R.K.J. Tseng H.T. Factors influencing Chinese tourists’ intentions to use the Taiwan Medical Travel App Telematics and Informatics 33 2 2016 401 409 Dahl A.J. Milne G.R. Peltier J.W. Digital health information seeking in an omni-channel environment: A shared decision-making and service-dominant logic perspective Journal of Business Research 125 2021 840 850 Dwivedi Y.K. Shareef M.A. Simintiras A.C. Lal B. Weerakkody V. A generalised adoption model for services: A cross-country comparison of mobile health (m-health) Government Information Quarterly 33 1 2016 174 187 Frey A. Trenz M. Veit D. Three Differentiation Strategies for Competing in the Sharing Economy MIS Quarterly Executive 18 2 2019 Gao Y. Li H. Luo Y. An empirical study of wearable technology acceptance in healthcare Industrial Management & Data Systems. 2015 Ghose A. Guo X. Li B. Dang Y. Empowering patients using smart mobile health platforms: Evidence from a randomized field experiment MIS Quarterly. 2021 Gimpel H. Manner-Romberg T. Schmied F. Winkler T.J. Understanding the evaluation of mHealth app features based on a cross-country Kano analysis Electronic Markets 2021 1 30 Grover P. Kar A.K. Gupta S. Modgil S. Influence of political leaders on sustainable development goals–insights from twitter Journal of Enterprise Information Management 34 6 2021 1893 1916 Hashemi H. Rajabi R. Brashear-Alejandro T.G. COVID-19 Research in Management: An Updated Bibliometric Analysis Journal of Business Research. 2022 Hoque R. Sorwar G. Understanding factors influencing the adoption of mHealth by the elderly: An extension of the UTAUT model International Journal of Medical Informatics 101 2017 75 84 28347450 Jannati N. Nakhaee N. Yazdi-Feyzabadi V. Tjondronegoro D. A cross-sectional online survey on patients’ satisfaction using store-and-forward voice and text messaging teleconsultation service during the COVID-19 pandemic International Journal of Medical Informatics 151 2021 104474 Kim D.J. Hwang Y. A study of mobile internet user’s service quality perceptions from a user’s utilitarian and hedonic value tendency perspectives Information Systems Frontiers 14 2 2012 409 421 Kim K.H. Kim K.J. Lee D.H. Kim M.G. Identification of critical quality dimensions for continuance intention in mHealth services: Case study of onecare service International Journal of Information Management 46 2019 187 197 Kim M. Lee S.M. Choi S. Kim S.Y. Impact of visual information on online consumer review behavior: Evidence from a hotel booking website Journal of Retailing and Consumer Services 60 2021 102494 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 Lamprinos I. Demski H. Mantwill S. Kabak Y. Hildebrand C. Ploessnig M. Modular ICT-based patient empowerment framework for self-management of diabetes: Design perspectives and validation results International Journal of Medical Informatics 91 2016 31 43 27185507 Li J. Zhang C. Li X. Zhang C. Patients’ emotional bonding with MHealth apps: An attachment perspective on patients’ use of MHealth applications International Journal of Information Management 51 2020 102054 Lim S. Xue L. Yen C.C. Chang L. Chan H.C. Tai B.C. …Choolani M. A study on Singaporean women's acceptance of using mobile phones to seek health information International Journal of Medical Informatics 80 12 2011 e189 e202 21956003 Lim W.M. Gupta G. Biswas B. Gupta R. Collaborative consumption continuance: A mixed-methods analysis of the service quality-loyalty relationship in ride-sharing services Electronic Markets 2021 1 22 Lorca-Cabrera J. Marti-Arques R. Albacar-Rioboo N. Raigal-Aran L. Roldan-Merino J. Ferre-Grau C. Mobile applications for caregivers of individuals with chronic conditions and/or diseases: Quantitative content analysis International Journal of Medical Informatics 145 2021 104310 Malik N. Tripathi S.N. Kar A.K. Gupta S. Impact of artificial intelligence on employees working in industry 4.0 led organizations International Journal of Manpower. 43 2 2021 334 354 Maglio P.P. Spohrer J. Fundamentals of service science Journal of the Academy of Marketing Science 36 1 2008 18 20 McKinley C.J. Wright P.J. Informational social support and online health information seeking: Examining the association between factors contributing to healthy eating behavior Computers in Human Behavior 37 2014 107 116 Miró J. Llorens-Vernet P. Assessing the Quality of Mobile Health-Related Apps: Interrater Reliability Study of Two Guides JMIR mHealth and uHealth 9 4 2021 e26471 33871376 Müller O. Junglas I. Brocke J.V. Debortoli S. Utilising big data analytics for information systems research: Challenges, promises and guidelines European Journal of Information Systems 25 4 2016 289 302 Nouri, R., R Niakan Kalhori, S., Ghazisaeedi, M., Marchand, G., & Yasini, M. (2018). Criteria for assessing the quality of mHealth apps: a systematic review. Journal of the American Medical Informatics Association, 25(8), 1089-1098. Nusairat N. Abdellatif H. Al-Gasawneh J. Akhorshaideh A. Aloqool A. Rabah S. Ahmad A. Determinants of behavioral intentions to use mobile healthcare applications in Jordan International Journal of Data and Network Science 5 4 2021 547 556 Osei-Frimpong K. Wilson A. Lemke F. Patient co-creation activities in healthcare service delivery at the micro level: The influence of online access to healthcare information Technological Forecasting and Social Change 126 2018 14 27 Pal S. Mukhopadhyay A. Shukla G.K. Can IT Improve Cardiac Treatment Quality? A Quantitative Study of Interaction between Technology and External Factors Proceedings of the 2018 ACM SIGMIS Conference on Computers and People Research 2018 18 25 10.1145/3209626.3209704 Paramita N. Noviarisanti S. Service Quality Analysis of Mhealth Services Using Text Mining Method: Alodokter and Halodoc International Journal of Management, Finance and Accounting 2 2 2021 1 21 Peltier J.W. Dahl A.J. Swan E.L. Digital information flows across a B2C/C2C continuum and technological innovations in service ecosystems: A service-dominant logic perspective Journal of Business Research 121 2020 724 734 Pennebaker J.W. Boyd R.L. Jordan K. Blackburn K. The development and psychometric properties of LIWC2015 2015 Pennebaker Conglomerates Austin, TX Pop O.M. Leroi-Werelds S. Roijakkers N. Andreassen T.W. Institutional types and institutional change in healthcare ecosystems Journal of Service Management 2018 Prakash A.V. Das S. Pillai K.R. Understanding digital contact tracing app continuance: Insights from India Health Policy and Technology 10 4 2021 100573 Rahimi R. Khoundabi B. Investigating the effective factors of using mHealth apps for monitoring COVID-19 symptoms and contact tracing: A survey among Iranian citizens International Journal of Medical Informatics 155 2021 104571 Rodríguez-Priego N. Porcu L. Kitchen P.J. Sharing but caring: Location based mobile applications (LBMA) and privacy protection motivation Journal of Business Research 140 2022 546 555 Russo G. Moretta Tartaglione A. Cavacece Y. Empowering patients to co-create a sustainable healthcare value Sustainability 11 5 2019 1315 Sadegh S.S. Saadat P.K. Sepehri M.M. Assadi V. A framework for mHealth service development and success evaluation International Journal of Medical Informatics 112 2018 123 130 29500009 Salehan M. Kim D.J. Predicting the performance of online consumer reviews: A sentiment mining approach to big data analytics Decision Support Systems 81 2016 30 40 Salton G. Buckley C. Term-weighting approaches in automatic text retrieval Information Processing & Management 24 5 1988 513 523 Schütze H. Manning C.D. Raghavan P. Introduction to information retrieval Vol. 39 2008 234 265 Sengupta P. Biswas B. Kumar A. Shankar R. Gupta S. Examining the predictors of successful Airbnb bookings with Hurdle models: Evidence from Europe, Australia, USA and Asia-Pacific cities Journal of Business Research 137 2021 538 554 Shah A.M. Yan X. Tariq S. Khan S. Listening to the patient voice: Using a semantic computing model to evaluate physicians’ healthcare service quality for strategic planning in hospitals Quality & Quantity 55 2021 173 201 Sheth J.N. Customer value propositions: Value co-creation Industrial Marketing Management 87 2019 312 315 Shirazi F. Wu Y. Hajli A. Zadeh A.H. Hajli N. Lin X. Value co-creation in online healthcare communities Technological Forecasting and Social Change 167 2021 120665 Stieglitz S. Dang-Xuan L. Emotions and information diffusion in social media—sentiment of microblogs and sharing behavior Journal of Management Information Systems 29 4 2013 217 248 Stocchi L. Pourazad N. Michaelidou N. Tanusondjaja A. Harrigan P. Marketing research on Mobile apps: Past, present and future Journal of the Academy of Marketing Science 2021 1 31 Suroso J.S. Sukmoro T.C. Factors affecting behavior of the use of healthcare mobile application technology in Indonesian society Journal of Theoretical and Applied Information Technology 99 15 2021 3923 3934 Tan, A. H. (1999). Text mining: The state of the art and the challenges. In Proceedings of the PAKDD 1999 Workshop on Knowledge Discovery from Advanced Databases (Vol. 8, pp. 65-70). Tan H. Yan M. Physician-user interaction and users' perceived service quality: Evidence from Chinese mobile healthcare consultation Information Technology & People 2020 Tartaglione A.M. Cavacece Y. Cassia F. Russo G. The excellence of patient-centered healthcare: Investigating the links between empowerment, co-creation and satisfaction The TQM Journal 2018 Tran T.P. Mai E.S. Taylor E.C. Enhancing brand equity of branded mobile apps via motivations: A service-dominant logic perspective Journal of Business Research 125 2021 239 251 Tripathi S. Deokar A.V. Ajjan H. Understanding the Order Effect of Online Reviews: A Text Mining Perspective Information Systems Frontiers 2021 1 18 Tsiotsou R.H. Boukis A. In-home service consumption: A systematic review, integrative framework and future research agenda Journal of Business Research 145 2022 49 64 Vargo S.L. Lusch R.F. Evolving to a new dominant logic for marketing Journal of Marketing 68 1 2004 1 17 Vargo S.L. Lusch R.F. Service-dominant logic: Continuing the evolution Journal of the Academy of Marketing Science 36 1 2008 1 10 Vargo S.L. Lusch R.F. From goods to service (s): Divergences and convergences of logics Industrial Marketing Management 37 3 2008 254 259 Vargo S.L. Akaka M.A. Service-dominant logic as a foundation for service science: Clarifications Service Science 1 1 2009 32 41 Veríssimo J.M.C. Usage intensity of mobile medical apps: A tale of two methods Journal of Business Research 89 2018 442 447 Verma S. Gustafsson A. Investigating the emerging COVID-19 research trends in the field of business and management: A bibliometric analysis approach Journal of Business Research 118 2020 253 261 32834211 Wang, C., & Qi, H. (2021, March). Influencing factors of acceptance and use behavior of mobile health application users: systematic review. In Healthcare (Vol. 9, No. 3, p. 357). MDPI. Wakefield D. Bayly J. Selman L.E. Firth A.M. Higginson I.J. Murtagh F.E. Patient empowerment, what does it mean for adults in the advanced stages of a life-limiting illness: A systematic review using critical interpretive synthesis Palliative Medicine 32 8 2018 1288 1304 29956568 Webster P. Virtual health care in the era of COVID-19 The Lancet 395 10231 2020 1180 1181 Wu L. Li J.Y. Fu C.Y. The adoption of mobile healthcare by hospital's professionals: An integrative perspective Decision Support Systems 51 3 2011 587 596 Wu P. Zhang R. Luan J. Zhu M. Factors affecting physicians using mobile health applications: An empirical study BMC Health Services Research 22 1 2022 1 14 34974828 Yakubu H. Kwong C.K. Forecasting the importance of product attributes using online customer reviews and Google Trends Technological Forecasting and Social Change 171 2021 10.1016/j.techfore.2021.120983 Yang H. Guo X. Peng Z. Lai K.H. Patient empowerment in an online health platform: Exploring the quadratic effects of patients’ conscious-competence on perceived health status Computers in Human Behavior 107346 2022 Yang X. Kovarik C.L. A systematic review of mobile health interventions in China: Identifying gaps in care Journal of Telemedicine and Telecare 27 1 2021 3 22 31319759 Zhang M. Zhang C. Shi Q. Zeng S. Balezentis T. Operationalising the telemedicine platforms through the social network knowledge: An MCDM model based on the CIPFOHW operator Technological Forecasting and Social Change 174 2022 121303 Zhao Y. Ni Q. Zhou R. What factors influence the mobile health service adoption? A meta-analysis and the moderating role of age International Journal of Information Management 43 2018 342 350	36475057	PMC9715352	NO-CC CODE	2022-12-03 23:20:14	no	J Bus Res. 2023 Feb 2; 156:113484	utf-8	J Bus Res	2,022	10.1016/j.jbusres.2022.113484	oa_other
==== Front Child Abuse Negl Child Abuse Negl Child Abuse & Neglect 0145-2134 1873-7757 Published by Elsevier Ltd. S0145-2134(22)00475-6 10.1016/j.chiabu.2022.105941 105941 Article Using research-practice-policy partnerships to mitigate the effects of childhood trauma on educator burnout Tirrell-Corbin Christy a Klika J. Bart bc⁎ Schelbe Lisa c a University of Maryland, College Park, MD, 20742, United States of America b Prevent Child Abuse America, 228 S. Wabash Ave., Floor 10, Chicago, IL 60604, United States of America c Florida State University, 296 Champions Way, University Center, Building C-Suite 2500, Tallahassee, FL 32306-2570, United States of America ⁎ Corresponding author at: Prevent Child Abuse America, 228 S. Wabash Ave., Floor 10, Chicago, IL 60604, United States of America. 2 12 2022 2 12 2022 1059414 2 2022 21 10 2022 26 10 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. Background The effects of COVID-19 pandemic on children have been immense. Objective In this commentary, we argue for the need to utilize research-practice-policy partnerships to address the issue of educator burnout and Secondary Traumatic Stress. Participants and setting Education systems have the potential to be the site of public health interventions in helping to identify and address the needs of children and families. Methods In this commentary, we review the literature on child trauma and adversity, educator burnout, and research-practice-policy partnerships. Results With the return to in-person learning, educators, and the systems in which they work are overwhelmed by the magnitude of mental health challenges presenting in the classroom due to child trauma. As a result, many educators are reporting high levels of compassion fatigue, secondary trauma, and burnout, which are known predictors of leaving the workforce. Many of the strategies employed to address educator compassion fatigue, secondary trauma, and burnout focus directly on the individual level (e.g., deep breathing, yoga). Yet the compassion fatigue, secondary trauma, and burnout are rooted in larger system failures to address the growing needs of children and families. Conclusions By bringing together key community members, including educators, and utilizing local data to inform policy decisions, actionable, trauma-informed solutions can create the conditions for thriving educators and therefore, thriving children. Keywords Child abuse Burnout Secondary trauma Research-practice-policy partnership ==== Body pmc1 Introduction Childhood trauma and adversity pre-date the COVID-19 crisis; however, the effects of the global pandemic have exacerbated pre-existing social inequities and magnified the fractured nature of our child and family serving systems (Herrenkohl, Scott, Higgins, Klika, & Lonne, 2020). Many experts shared concerns early in the pandemic as to the potential for skyrocketing cases of child abuse and neglect due to formal shelter-in-place orders and worried that children were having to shelter in unsafe places with unsafe people (Wulczyn, 2020). With children no longer under the close physical watch of educators who are legally mandated to report suspected cases of child abuse and neglect, reports of child abuse and neglect dropped precipitously in comparison to prior years (Jonson-Reid, Drake, Cobetto, & Ocampo, 2020). While there is debate as to whether child abuse and neglect actually rose during the COVID-19 pandemic, it is clear that many of the risk factors for maltreatment increased, exacting an enormous toll on the mental health of our nation's children (Agrawal, 2020; Roy, 2020). By the fall of 2022 school districts had fully returned to in-person learning and discussions of achievement loss dominated public discourse. In spite of the American Academy of Pediatrics (AAP), the American Academy of Child and Adolescent Psychiatry (AACAP), and the Children's Hospital Association (CHA) declaring children's mental health a “national emergency” (AAP, AACAP, & CHA, 2021), meeting the mental and emotional needs of children was largely absent from school-related discussions. Prior to the pandemic, a critical mass of educators and education systems had begun to embrace trauma-informed practices in school settings (Herrenkohl et al., 2021). However, the onus of addressing the increases in childhood trauma and adversity had and have largely fallen into the laps of educators themselves. As a result, educators are reporting record rates of compassion fatigue, burnout, and secondary trauma (Bakuli & Levin, 2021; Cardoza, 2021; Streeter, 2021). As we discuss in this commentary, the solutions to childhood trauma are complex and require coordination between multiple child and family serving systems, of which, the education system plays a central role. Trauma-informed practices and pedagogy are a critical component of a trauma-informed education system, especially considering learning losses experienced by children because of COVID-19, yet greater attention is warranted to ensure the health and wellness of educators themselves. The commentary begins with a discussion of the trauma and mental health needs of children and how the COVID-19 pandemic amplified these challenges. Next, we discuss the core components of trauma-informed systems and pedagogy and the ways in which education systems can begin to address childhood trauma. In doing so, we highlight the ways in which educators' secondary trauma, compassion fatigue, and burnout is often given short shrift and the ways in which mitigation strategies are aimed at individuals (i.e., personal self-care) instead of at system challenges. In closing, we introduce research-practice-policy partnerships as a strategy to identify, implement, and monitor solutions within the education system to address educator secondary trauma, compassion fatigue, and burnout. 1.1 Child trauma and adversity Within weeks of the initial March 2020 COVID-19 shelter-in-place orders, survey data began documenting an increase in many of the risk factors for child abuse and neglect including social isolation, parental stress, and mental health challenges (Lee, Ward, Chang, & Downing, 2021; Lee, Ward, Lee, & Rodriguez, 2020; Rodriguez, Lee, Ward, & Pu, 2020). At the same time, reports to official child welfare agencies across US jurisdictions plummeted (Jonson-Reid et al., 2020). Much of the decrease in reports appeared due to school closures, and not necessarily a decline in abuse or neglect (Baron, Goldstein, & Wallace, 2020). While there was also an initial precipitous decrease in calls and text messages to a large national child abuse hotline in March 2020, by May 2020 calls and text messages to the hotline surpassed levels from the prior year (Ortiz et al., 2021). Although we lack sufficient data to confirm that child abuse and neglect got worse during the COVID-19 pandemic, the data do signal that children and families had increased needs because of the pandemic. It is important to note that child abuse and neglect are not the only forms of trauma and adversity that children experience. Data from the Adverse Childhood Experiences (ACE) studies document the highly prevalent and overlapping nature of various traumas and adversities (Merrick, Ford, Ports, & Guinn, 2018). In addition to the traditional ACE's, children experience other challenges, such as poverty, which can ultimately affect their readiness to learn. Data from the U.S. Census Bureau (2021) found that in 2020, approximately 16.1 % of children were living in poverty, an increase from prior years. In 2022, Child Trends (Thomson et al., 2022) reported 29.2 % of Black children, 27.3 % of Latino children and 17.5 % of all children to be living in poverty in 2020. This represented an increase from 2019 poverty rates for Black and Latino children by 2.8 % and 4.3 % respectively. Many children are also dealing with the loss of a caregiver or loved one due to COVID-19. Hillis et al. (2021) found that nearly 140,000 children in the US lost a parent or grandparent because of COVID-19. This is of particular importance due to the number of children in the US living in single-parent households or being raised by grandparents. Of those US children who lost a primary caregiver, nearly 65 % were from a racial or ethnic minority group providing further evidence of the disproportionate toll that COVID-19 has had on communities of color (Hillis et al., 2021). Prior to the COVID-19 pandemic, children and youth mental health was of significant concern (Perou et al., 2013). As noted in a recent US surgeon general report (2021), estimates of youth mental health challenges range between 13 and 20 %. Surveillance data from the Centers for Disease Control and Prevention (2020) show that reports of mental health symptoms increased by approximately 40 % between 2009 and 2019. Examining data from the National Survey of Children's Health, Whitney and Peterson (2019) found that half of youth diagnosed with a mental health problem were not receiving treatment from a mental health provider. The COVID-19 pandemic intensified the youth mental health crisis (AAP, AACAP, & CHA, 2021). Data show that while emergency department (ED) visits generally declined during COVID-19, ED visits for mental health reasons increased for children and youth (24 % and 31 %, respectively) (Leeb et al., 2020). ED visits related to concerns of suicide increased nearly 50 % for girls compared to pre-pandemic estimates (Yard, Radhakrishnan, Ballesteros, et al., 2021). 1.2 Supporting children who experience trauma Considering the compounding and persistent trauma and adversity experienced by children and exacerbated by the COVID-19 pandemic, there is reason for continued concern about children's ability to arrive in the classroom ready to learn. Consequently, educators are often left with the responsibility of not only teaching children but also dealing with the outward manifestation of trauma and adversity, which can present as child externalizing behavior in the classroom. Positive teacher-child relationships are of critical importance to young children as they co-occur with rapid social, emotional, cognitive, and physical development (Ansari, Hofkens, & Pianta, 2020) and because, next to their families, children spend most of their waking hours with educators. Supportive teacher-student relationships allow children to feel safe, develop peer relationships, and take appropriate risks (Hamre & Pianta, 2006; Lippard, La Paro, Rouse, & Crosby, 2018). These relationships are also strong predictors of a student's academic success (Hamre & Pianta, 2001) with individual teacher-child relationships in early childhood identified as critical to children's school readiness and academic trajectories (Blair, McKinnon, & The Family Life Project investigators, 2016; Rudasill, Niehaus, Buhs, & White, 2013). Therefore, the classroom becomes an important microsystem where educators must be supported in order to promote a positive trajectory in these students' development and learning. Furthermore, educators can offer safe spaces for children away from neighborhood/domestic violence and where children feel valued and respected (Cole, Eisner, Gregory, & Ristuccia, 2013). Educators are also able to teach children self-regulation strategies and mindfulness, which reduce stress and, therefore, promote learning. Nonetheless, teacher-child relationships occur in the context of the classroom and the greater school community, all of which are influenced by the lived experiences of those who interact with the students each day. The Trauma Sensitive Pedagogy (TSP) school model put forth by Panlilio and Tirrell-Corbin (2021b) illustrates the complexity of and influences on that relationship. Their TSP model is guided by the Bioecological model of development (Bronfenbrenner & Morris, 2007) to specify multiple levels of development and learning that occur in a classroom/school. As seen in Fig. 1 , the focus is on the teacher-student dyad within the classroom (i.e., microsystem of interest) with contextual factors inside (e.g., other students) and outside (e.g., teacher support, administration, school norms, and policies) the classroom that influence the quality of teacher-student relationships. Within the dyad, teachers' socioemotional competence (SEC) and well-being (personal experiences and secondary traumatic stress) are important in developing a supportive relationship (Jennings & Greenberg, 2009). Teachers who exhibit higher socioemotional competence demonstrate more effective classroom management and model emotional expressions and verbal support to promote engaged learning. Such quality interactions are important in strengthening children's resilience during daily interactions, particularly for children with histories of trauma. However, as illustrated in the model, the interactions between teacher and student are also heavily influenced by the school staff (administrators, specialists, nurses, counselors and paraeducators), families and community members associated with that school. For example, schools in low-resourced communities with high rates of violence are more likely to have children who have experienced trauma and teachers who report symptoms of secondary traumatic stress (Panlilio & Tirrell-Corbin, 2021b).Fig. 1 Trauma sensitive pedagogy model. Fig. 1 2 Personal risks of supporting children through trauma There has been a growing awareness that schools should be “trauma-informed” in order to meet the needs of all learners (Lawson, Caringi, Gottfried, Bride, & Hydon, 2019). In fact, the National Child Traumatic Stress Network (NCTSN), Schools Committee (2017) published a framework with 10 Core Areas of a Trauma-informed School system. Broadly, the framework lays the foundation for understanding and responding to the various childhood traumas and adversities presented in the classroom. Among those core elements is “Trauma Education and Awareness,” to include professional development on the impact of trauma and building student coping and protective skills. While the framework was an important step forward, the US has few evidence-based, trauma-informed schools to serve as models (Lawson et al., 2019). Moreover, few educators have had formal preparation on assessing or addressing their students' traumatic experiences. This is at least in part because the faculty that taught them in educator preparation programs had no training themselves (Farrell & Walsh, 2010; Goldman & Grimbeek, 2014; Lawson et al., 2019). The lack of education leaves educators to support their students in the best ways they know how, which evidence suggests puts their students at risk (Lawson et al., 2019). Evidence also suggests that responding to children's trauma has taken a personal toll on educators across the US (Hydon, Wong, Langley, Stein, & Kataoka, 2015; Kerig, 2019; Lawson et al., 2019). More specifically, educators report experiencing Secondary Traumatic Stress (STS), which is the vicarious trauma resulting from learning about and responding to another's traumatic experiences (Borntrager et al., 2012). STS presents itself in a number of ways, including sadness, insomnia, guilt, inadequacy, substance abuse, and disengagement from one's students and one's family members (Borntrager et al., 2012; Kerig, 2019; Rankin, 2020). Figley (1995) coined these responses as the “cost of caring,” which results in burnout, compassion fatigue, and exiting the profession (Holme, Jabbar, Germain, & Dinning, 2018; Snodgrass Rangel, 2018). While some researchers use the terms as equivalents, Kerig (2019) defined the unique characteristics of burnout and compassion fatigue. More specifically, burnout has been equated with exhaustion (mental and physical) resulting from a perceived lack of control, appreciation, or support, as well as administrative burdens. Compassion fatigue results from active engagement with those who have experienced trauma resulting in emotional exhaustion to the point of decreased compassion. During the COVID-19 pandemic, educators took on a greater role in supporting children and families, notably those residing in low resourced communities and/or those who recently immigrated to the US (Panlilio & Tirrell-Corbin, 2021a; Tirrell-Corbin, Panlilio, & Klika, 2021a; Tirrell-Corbin, Panlilio, & Klika, 2021b). Disparities in service availability and access often placed the burden of care solely on educators, requiring them to go beyond addressing curricular standards in classrooms and, for most, well beyond the preparation they received to become a teacher (Lawson et al., 2019). Educators collected donations for grocery store gift cards for families in need, organized holiday toy drives, contacted state legislators to advocate for the reopening of school-based health clinics to ensure children had access to medical care, facilitated access to virtual mental health screenings, dropped off school supplies to children's homes, and attended virtual funerals (Panlilio & Tirrell-Corbin, 2021a). This suggests that educators are increasingly tasked with responsibilities beyond instructional planning and delivery and more appropriate for other members of the community, such as counselors, social workers, nurses, and community organizers. In addition to fulfilling their instructional duties, educators juggled many intersecting issues that included supporting their students' families, managing their own families' pandemic-driven needs (sometimes a source of primary trauma), and being caught in the middle of the politicization of the pandemic. A Virginia school superintendent wrote that even though teachers are indispensable members of society they were the recipients of “revolting comments at school board meetings, floggings via social media and even being called losers by national leaders” (Jeck, 2021). Taken together, such a balancing act took a toll on educators' mental health and well-being (Bennett, 2022; Hsu, 2021). On the surface, teacher burnout and STS appear to be mental health issues. However, a closer examination of the issue reveals that this is also a serious workforce issue (Diliberti, Schwartz, & Grant, 2021) occurring in the context of a national teacher shortage (García & Weiss, 2019; Jeck, 2021; Sutcher, Carver-Thomas, & Darling-Hammond, 2016). Prior to the pandemic, Goldring, Taie, and Riddles (2014) found that approximately 8 % of educators left the profession each year. In October of 2020, seven months into the pandemic, approximately one-quarter of educators, in a national sample, indicated that they were likely to leave their jobs before the completion of the current academic year (Diliberti & Kaufman, 2020). It is also important to note that women, who continue to dominate the field of education (U.S. Department of Labor, 2019), left the workforce in record numbers in 2020 and 2021 (Bateman & Ross, 2020; Gogoi, 2020; Hsu, 2021). Both before and during the pandemic, stress was cited as the main reason educators were choosing to leave the education profession (Diliberti et al., 2021). In 2021, the stress and burnout percolating in the education community made it to local and national news publications (Baitinger & Travis, 2021; Brown, 2021; Doherty, 2021; Jeck, 2021). By the fall of 2022, the teacher shortage was a crisis across the United States (Bennett, 2022; Cohen, 2022). Teacher attrition is a costly and chronic issue in American education. Costs come in two forms: resources and student achievement (Diliberti & Kaufman, 2020). The costs of replacing a teacher have been estimated to range from $10,000 to $17,000 in the US (Barnes, Crowe, & Schaefer, 2007; DeFeo, Tran, Hirshberg, Cope, & Cravez, 2017). The cost to student learning is immeasurable as instruction is disrupted and often of poor quality in the aftermath of teacher resignations. The situation is worse for students in low-resourced communities where teacher turnover is higher and where school administrators find it difficult to staff classrooms with highly effective, experienced educators (Boyd et al., 2011; Ronfeldt, Loeb, & Wyckoff, 2013). Combined with the protective role that educators can play in the lives of children, teacher burnout and attrition can have significant effects on children's well-being. Historically, many of the strategies employed to address educator compassion fatigue, secondary trauma, and burnout focus directly on the individual level (e.g., deep breathing, taking walks, yoga), in spite of these issues being rooted in larger systemic failures. 2.1 Research-practice-policy partnerships The desire for evidence-based policy and practice is “nearly ubiquitous... across the fields of education, child welfare, mental health, juvenile justice, youth programs and health care” however it is not always realized (Tseng, 2012, p. 1). It is well understood that a significant gap exists between the point of a scientific discovery and the ensuing uptake in practice and policy. Morris, Wooding, and Grant (2011) estimate that it takes nearly 17 years for research findings to take root in real-world settings. This means there is significant “research waste” where the significant financial investments in research to address social problems are not translating to practice and policy (Oliver & Boaz, 2019). It is important to make sure that the billions of dollars spent on research benefit society and to determine how best to ensure this happens (Rosenblatt & Tseng, 2010). The current educator burnout crisis needs immediate solutions and cannot wait decades for scientific discovery to guide practice change. An efficient and viable way to create evidence-based policy and practice is to create partnerships across the sectors of research, policy, and practice (Schelbe, Wilson, Fickler, Williams-Mbengue, & Klika, 2020). These partnerships help bridge the gaps between “what is known” and “what is done” (Leslie, Maciolek, Biebel, Debordes-Jackson, & Nicholson, 2014) by placing an explicit focus on “problems of practice” (Coburn & Penuel, 2016). Schelbe et al. (2020) advanced the idea of research-practice-policy partnerships (RPPPs) as a strategy for cross-sector collaboration to strengthen child welfare practice and the array of child maltreatment prevention strategies. As conceptualized, researchers, state policymakers, and practitioners (e.g., child welfare, maltreatment prevention, home visiting) would utilize principles of design thinking to organize efforts in problem definition, exploration of solutions, implementation, and ongoing evaluation. To facilitate this work, the Design Thinking Framework (Empathize, Design, Ideate, Prototype and Test) provides a vehicle to engage researchers, policymakers and practitioners in breaking down barriers and testing solutions that focus on meeting the needs of children and families early enough to prevent childhood trauma/abuse (Schelbe et al., 2020). The utility of design thinking for policymaking processes has been recognized as it assists with defining problems, identifying solutions, and implementing new ideas (Mintrom & Luetjens, 2016). It is relevant in educational settings to assist educators in addressing various problems (Henriksen, Richardson, & Mehta, 2017) and is promising for addressing the issues of educator burnout and childhood trauma within education systems. In prior work, there has been an emphasis with RPPPs to engage state-level policy makers (Schelbe et al., 2020). However, within education systems, a great deal of power and decision-making authority can be relegated to local jurisdictions suggesting that mid-level administrators (e.g., superintendent, principal) should be involved in the RPPP. Tseng (2012) notes that agency:mid-level administrators and program managers shape the frontline practices—teaching, social work, counseling, policing—of concern to many researchers. They play a critical role in designing staff development systems and adopting new programs and reforms, shaping the process and conditions for their implementation, and allocating resources in support of them. These mid-level decision makers straddle policy and practice and are well-poised to put research to work to benefit youth. In addition, they can be a more stable presence than agency leaders, who have short tenures in many places. (p. 5) While policy at the state and federal level may be helpful in some regards, as local communities have unique needs and resources, the issue can be best addressed by considering the local context with school and district administrators. 2.2 RPPP in practice In creating a RPPP to address childhood trauma within education systems and its demands and impact on educators, there must be great commitment as developing these partnerships requires considerable time and resources (Penuel & Hill, 2019). The process is not always linear but can follow these steps: 1) identify a team, 2) secure funding, 3) build team relationships and a shared vision, 4) engage in a design thinking process to address the problem(s). The first step is to pull together a team across the sectors of research, policy, and practice and to incorporate community members with lived experience. There is an abundance of researchers at universities and colleges that could be part of a RPPP. Potential places to find researchers are at institutions committed to serving the community. Researchers should value community partnerships and understand the realistic time demands of community-based research, especially RPPPs. Policymakers involved in the process should be the administrators in schools and the school districts or those with the capacity to make practice/policy decisions for the school or district. Ideally, schools and school districts can learn from one another and explore adopting or modifying strategies proven to be effective at a different school. The practice sector should include educators and school staff as well as mental health professionals and other professionals who work with children who have experienced traumas. Community members as a category should be broad and integrate parents and students as well as others who are interested in education and reducing the impact of child trauma. Fig. 2 illustrates how researchers, policymakers, practitioners, and community members can work within Panlilio and Tirrell-Corbin's (2021b) Trauma Sensitive Pedagogy framework to address childhood trauma and its impact on educators.Fig. 2 Trauma-informed RPPP. Fig. 2 Funding can be secured as the team is being formed. An important consideration for RPPPs is funding. The amount of work that needs to be done is significant, and it cannot rest solely on volunteers. Ensuring there are resources to support the RPPP efforts means that people have dedicated time to do the work and do not have to spend their own resources to be part of the RPPP. Ideally, there is funding to support those who are doing the administrative work (e.g., scheduling, writing, reviewing). Considering one goal is to reduce the demands on educators, it would be especially inappropriate to simply pile more tasks upon educators and expect them to donate their time. Compensation of RPPP members must also be prioritized. At a bare minimum reducing costs of participation, such as providing childcare for parents with children and transportation for those who need it or offering reimbursements for childcare, transportation, and other expenses related to being part of the RPPP. Ideally, there would also be stipends for those who actively participate in the RPPP to compensate them for their time. Funding to support RPPP may be available through foundations that have supported similar work (e.g., Spencer Foundation, W.T. Grant Foundation). Additionally, school districts or schools may wish to budget resources to support the efforts to develop a RPPP. Considering that this work may reduce teacher attrition and improve student outcomes, it can be justified as a sound financial decision. After the RPPP team is assembled, time must be spent developing relationships and building trust (Coburn & Penuel, 2016). Throughout this process it is important to remember that members of the team bring different perspectives and expertise. Incorporating these differences and leveraging the strengths of the team members to develop a shared vision and goal is central. RPPP's successes are connected to the communication and connection among the team members. To address the issues of childhood trauma and educator burnout, the RPPP team can engage in the design thinking process. (See Schelbe et al., 2020 for details about RPPs using design thinking.) The implementation of design thinking is an iterative process where the RPPP moves swiftly among the steps. The process begins with the RPPP seeking to understand the issue. Getting information from many sources helps the RPPP to empathize with the various constituents and grasp the gravity of the issue. The RPPP team can collect information from educators, students, mental health professionals, and others, which provides a holistic understanding of the issue. With an agreed upon understanding of the issue, the RPPP enters the design phase of the project where the team starts framing the policy issue and identifies research to inform policy and practice (Schelbe et al., 2020). With the gathered information and shared understanding, the RPPP team begins the process of generating ideas about how to address child trauma and educator burnout in schools. The ideas can be at multiple levels including in the classroom, the school, the district, or community. During this process, the RPPP team also identifies research questions and potential data sources. The team then begins to build a plan to pilot some of the ideas and conduct research or evaluation to answer their research questions using local data. The ideas are tested, and the findings can be used to refine ideas which can be further piloted. Multiple ideas can be examined and implemented simultaneously with the goal of finding what can best address the issue. From there, the RPPP continues to identify challenges, create, and test solutions using local data, and using the immediate feedback from research/evaluation to inform practice and/or policy changes. In doing so, the issues of childhood trauma and educator burnout are addressed immediately and do not fall victim to the 17 year “research-to-practice gap.” 2.3 Challenges of RPPP While the need and utility for RPPPs is clear, there are challenges and barriers in place which make cross-sector partnerships challenging to implement. With pressure to publish academic articles, researchers can be disincentivized to spend time building community-based collaborations, especially if these collaborations do not result in numerous, high-impact peer-reviewed research publications. Policymakers, including school administrators, must juggle many competing priorities during and in-between legislative sessions making consistent participation in RPPPs challenging. Finally, those representing the practice community often have job responsibilities that create challenges for and disincentivize participation in RPPPs. For example, educators may not be able to meet with partners of a RPPP during the educators' working hours because of the classroom responsibilities. If they meet after school hours educators are volunteering their time as the work is outside of their job responsibilities, a potential contributor to burnout. 2.4 Setting a research agenda As noted by Coburn and Penuel (2016), there are critical research questions that must be answered prior to the widespread adoption of RPPPs. Central to this research agenda is a better understanding of the critical components of RPPPs that drive positive outcomes across initiatives. What factors, under what conditions, lead to which outcomes? Do RPPPs lead to increased trust among participants? Do RPPPs lead to better use of data in decision making? Do RPPPs facilitate more rapid and/or efficient scale-up of interventions? In addition, it is imperative to understand the costs associated with participation in RPPPs; these can include monetary costs as well as opportunity costs to the involved parties. Having educators participate in an RPPP not only costs money but also diverts attention away from other tasks that educators may need to do (e.g., lesson planning). 3 Conclusion Prior to, and exacerbated by the COVID-19 pandemic, children experience a host of traumas and adverse conditions that place their well-being and learning in the classroom at risk. Educators play a key role for children by providing safety, stability, and nurturing care and environments for children; however, the effects of trauma can wreak havoc on educator well-being. Unfortunately, educators experience STS and burnout, likely the result of the unrelenting responsibilities placed on their plate, including having to address the external manifestations of childhood traumas in the classroom. Few educators have adequate training on how to address the complicated behaviors associated with childhood trauma, leaving many feeling helpless and ineffective. Trauma-informed pedagogical approaches are helpful in the classroom however are rendered useless unless educators are physically, mentally, and emotionally available to implement the practices. By addressing educator STS and burnout using RPPP, the hope is to create the context where educators have the support they need to implement trauma-informed practices therefore, addressing issues of childhood trauma. Declaration of competing interest None. Data availability No data was used for the research described in the article. ==== Refs References American Academy of PediatricsAmerican Academy of Child and Adolescent PsychiatryChildren’s Hospital Association AAP-AACAP-CHA declaration of a National Emergency in Child and Adolescent Mental Health 2021 The American Academy of Pediatrics https://www.aap.org/en/advocacy/child-and-adolescent-healthy-mental-development/aap-aacap-cha-declaration-of-a-national-emergency-in-child-and-adolescent-mental-health/ Agrawal N. The coronavirus could cause a child abuse epidemic 2020 New York Times https://www.nytimes.com/2020/04/07/opinion/coronavirus-child-abuse.html Ansari A. Hofkens T.L. Pianta R.C. Teacher-student relationships across the first seven years of education and adolescent outcomes Journal of Applied Developmental Psychology 71 2020 101200 Baitinger B. Travis S. A ‘serious problem’: Schools grapple with growing shortage of teachers and substitutes 2021, November Sun Sentinel https://www.sun-sentinel.com/local/broward/fl-ne-teacher-subtitute-shortage-classes-20211101-x3poasv5ozbxxeyk4qh2mg26mm-story.html Bakuli E. Levin K. Detroit’s December remote days a sign of how long and challenging the pandemic recovery will be 2021, November 29 Chalkbeat Detroit https://detroit.chalkbeat.org/2021/11/29/22808141/detroit-schools-remote-learning-days-covid-pandemic-recovery Barnes G. Crowe E. Schaefer B. The cost of teacher turnover in five school districts: A pilot study 2007 National Commission on Teaching and America's Future Baron E.J. Goldstein E.G. Wallace C.T. Suffering in silence: How COVID-19 school closures inhibit the reporting of child maltreatment Journal of Public Economics 19 2020 104258 Bateman N. Ross M. Why has COVID-19 been especially harmful for working women? 2020 The Brookings Institution Washington, DC https://www.brookings.edu/essay/why-has-covid-19-been-especially-harmful-for-working-women/ Bennett G. Why teachers in America are leaving the profession in droves 2022, August 20 PBS Newshour https://www.pbs.org/newshour/show/why-teachers-in-america-are-leaving-the-profession-in-droves Blair C. McKinnon R. The Family Life Project investigators Moderating effects of executive functions and the teacher-child relationship on the development of mathematics ability in kindergarten Learning and Instruction 41 2016 85 93 10.1016/j.learninginstruc.2015.10.001 28154471 Borntrager C. Caringi J.C. van den Pol R. Crosby L. O'Connell K. Trautman A. McDonald M. Secondary traumatic stress in school personnel Advances in School Mental Health Promotion 5 1 2012 38 50 Boyd D. Grossman P. Ing M. Lankford H. Loeb S. Wyckoff J. The influence of school administrators on teacher retention decisions American Educational Research Journal 48 2 2011 303 333 Bronfenbrenner U. Morris P.A. The bioecological model of human development Handbook of child psychology 2007 1 Brown C. Surprise labor shortage hits schools 2021, October Axios https://www.axios.com/surprise-labor-shortage-hits-schools-709326d6-de95-410c-84ce-6e0a5ed60b49.html Cardoza K. ‘We need to be nurtured, too’: Many teachers say they’re reaching a breaking point 2021 National Public Radio: All Things Considered https://www.npr.org/2021/04/19/988211478/we-need-to-be-nurtured-too-many-teachers-say-theyre-reaching-a-breaking-point Centers for Disease Control and Prevention Youth risk behavior surveillance data summary & trends report: 2009-2019 https://www.cdc.gov/nchhstp/dear_colleague/2020/dcl-102320-YRBS-2009-2019-report.html 2020 Coburn C.E. Penuel W.R. Research–practice partnerships in education: Outcomes, dynamics, and open questions Educational Researcher 45 1 2016 48 54 Cohen G. Why teachers are burning out and leaving districts scrambling to fill jobs 2022, August 31 CNN https://www.cnn.com/2022/08/31/us/teachers-shortage-burnout-vacancies/index.html Cole S.F. Eisner A. Gregory M. Ristuccia J. Creating and advocating for trauma- sensitive schools 2013 Massachusetts Advocates for Children DeFeo D.J. Tran T. Hirshberg D. Cope D. Cravez P. The cost of teacher turnover in Alaska 2017 Diliberti M. Kaufman J.H. Will this school year be another casualty of the pandemic? Key findings from the American Educator Panels Fall 2020 COVID-19 surveys. Data note: Insights from the American Educator Panels. Research report. RR-A168-4 2020 RAND Corporation Diliberti M. Schwartz H.L. Grant D.M. Stress topped the reasons why public school teachers quit, even before COVID-19 2021 RAND Corporation Doherty E. Teacher burnout leaves schools scrambling 2021, November Axios https://www.axios.com/teacher-burnout-fatigue-pandemic-covid-schools-75df52ab-720c-470fac9b426514e0452a.html?utm_campaign=organic&utm_medium=socialshare&utm_source=email Farrell A. Walsh K. Working together for Toby: Early childhood student teachers engaging in collaborative problem-based learning around child abuse and neglect Australian Journal of Early Childhood 35 2010 53 62 Figley C.R. Compassion fatigue: Toward a new understanding of the costs of caring Stamm B.H. Secondary traumatic stress: Self-care issues for clinicians, researchers, and educators 1995 The Sidran Press 3 28 García E. Weiss E. U.S. Schools Struggle to hire and retain teachers. The second report in" The Perfect Storm in the Teacher Labor Market" series 2019 Economic policy institute Gogoi P. Stuck-at-home moms: The pandemic’s devastating toll on women 2020 National Public Radio https://www.npr.org/2020/10/28/928253674/stuck-at-home-moms-the-pandemics-devastating-toll-on-women Goldman J.D.G. Grimbeek P. Reporting intervention preservice content preferred by student teachers Journal of Child Sexual Abuse 23 2014 1 16 24393087 Goldring R. Taie S. Riddles M. Teacher attrition and mobility: Results from the 2012-13 teacher follow-up survey. First look. NCES 2014-077 2014 National Center for Education Statistics Hamre B.K. Pianta R.C. Student-teacher relationships Bear G.G. Minke K.M. Children's needs III: Development, prevention, and intervention 2006 National Association of School Psychologists 59 71 Hamre B.K. Pianta R.C. Early teacher–child relationships and the trajectory of children's school outcomes through eighth grade Child Development 72 2 2001 625 638 11333089 Henriksen D. Richardson C. Mehta R. Design thinking: A creative approach to educational problems of practice Thinking Skills and Creativity 26 2017 140 153 Herrenkohl T.I. Miller A. Eisman A. Davis E. Price D. Robinson Y. Sherman B.A. Trauma informed programs and practices for schools (TIPPS) program guide https://tipps.ssw.umich.edu/wp-content/uploads/2021/11/Program-Guide_10-25-2021.pdf 2021 Herrenkohl T.I. Scott D. Higgins D.J. Klika J.B. Lonne B. How COVID-19 is placing vulnerable children at risk and why we need a different approach to child welfare Child Maltreatment 26 1 2020 9 16 10.1177/1077559520963916 33025825 Hillis S.D. Blenkinsop A. Villaveces A. Annor F.B. Liburd L. Massetti G.M. COVID-19–associated orphanhood and caregiver death in the United States Pediatrics 148 6 2021 Holme J.J. Jabbar H. Germain E. Dinning J. Rethinking teacher turnover: Longitudinal measures of instability in schools Educational Researcher 47 1 2018 62 75 Hsu A. Millions of women haven’t rejoined the workforce–and may not anytime soon 2021 National Public Radio, Morning Edition https://www.npr.org/2021/06/03/1002402802/there-are-complex-forces-keeping-women-from-coming-back-to-work Hydon S. Wong M. Langley A.K. Stein B.D. Kataoka S.H. Preventing secondary traumatic stress in educators Child and Adolescent Psychiatric Clinics 24 2 2015 319 333 Jeck D. Commentary: Superintendent of schools: “Who would want to be a teacher right now?” 2021, September Fauquier Times https://www.fauquier.com/news/superintendent-of-schoolswhowouldwanttobeateacherrightnow/article_6ee1dc301c7011ecae0e9fd71d412ee7.html?utm_medium=email&_hsmi=165253041&_hsenc=p2ANqtz_J8Inv8C5CPa2X6nPpqftoTYn3cgY_k8SBxcq_kXFo86fx_sm8gW23YUmMjgftgPA4FvkBunO70gFYw1QvAJTSARw&utm_content=165253041&utm_source=hs_email Jennings P.A. Greenberg M.T. The prosocial classroom: Teacher social and emotional competence in relation to student and classroom outcomes Review of Educational Research 79 2009 491 525 10.3102/0034654308325693 Jonson-Reid M. Drake B. Cobetto C. Ocampo M.G. Child abuse prevention month in the context of COVID-19 Retrieved from 2020 Washington University https://cicm.wustl.edu/child-abuse-prevention-month-in-the-context-of-covid-19/ Kerig P.K. Enhancing resilience among providers of trauma-informed care: A curriculum for protection against secondary traumatic stress among non-mental health professionals Journal of Aggression, Maltreatment & Trauma 28 5 2019 613 630 Lawson H.A. Caringi J.C. Gottfried R. Bride B.E. Hydon S.P. Educators' secondary traumatic stress, children’s trauma, and the need for trauma literacy Harvard Educational Review 89 3 2019 421 519 Lee S.J. Ward K.P. Lee J.Y. Rodriguez C.M. Parental social isolation and child maltreatment risk during the COVID-19 pandemic Journal of Family Violence 14 1 2020 1 12 Lee S.J. Ward K.P. Chang O.D. Downing K.M. Parenting activities and the transition to home-based education during the COVID-19 pandemic Children and Youth Services Review 122 2021 105585 Leeb R.T. Bitsko R.H. Radhakrishnan L. Martinez P. Njai R. Holland K.M. Mental Health–Related Emergency Department visits among children aged <18 years during the COVID-19 pandemic — United States, January 1–October 17, 2020 MMWR. Morbidity and Mortality Weekly Report 69 2020 1675 1680 10.15585/mmwr.mm6945a3externalicon 2020 33180751 Leslie L.K. Maciolek S. Biebel K. Debordes-Jackson G. Nicholson J. Exploring knowledge exchange at the research–policy–practice interface in children’s behavioral health services Administration and Policy in Mental Health and Mental Health Services Research 41 6 2014 822 834 24464480 Lippard C. La Paro K.M. Rouse H.L. Crosby D.A. A closer look at teacher–child relationships and classroom emotional context in preschool Child & Youth Care Forum 47 2018 1 21 https://doi-org.proxy-um.researchport.umd.edu/10.1007/s10566-017-9414-1 Merrick M.T. Ford D.C. Ports K.A. Guinn A.S. Prevalence of adverse childhood experiences from the 2011–2014 behavioral risk factor surveillance system in 23 states JAMA Pediatrics 172 1 2018 1038 1044 30242348 Mintrom M. Luetjens J. Design thinking in policymaking processes: Opportunities and challenges Australian Journal of Public Administration 75 3 2016 391 402 Morris Z.S. Wooding S. Grant J. The answer is 17 years, what is the question: understanding time lags in translational research Journal of the Royal Society of Medicine 104 12 2011 510 520 22179294 National Child Traumatic Stress Network, Committee on Schools Trauma-informed schools for children in K-12: a system framework. Washington, D. C https://www.nctsn.org/sites/default/files/resources/fact-sheet/trauma_informed_schools_for_children_in_k-12_a_systems_framework.pdf 2017 Oliver K. Boaz A. Transforming evidence for policy and practice: Creating space for new conversations Palgrave Communications 5 1 2019 1 10 Ortiz R. Kishton R. Sinko L. Fingerman M. Moreland D. Wood J. Venkataramani A. Assessing child abuse hotline inquiries in the wake of COVID-19: Answering the call JAMA Pediatrics 175 8 2021 859 861 33938944 Panlilio C. Tirrell-Corbin C. Our research shows educators are experiencing trauma during the pandemic. Here’s how we can reduce the burden 2021, March EdSurge https://www.edsurge.com/news/2021-03-02-our-research-shows-educators-are-experiencing-trauma-during-the-pandemic-here-s-how-we-can-reduce-the-burden Panlilio C. Tirrell-Corbin C. Examining implementation outcomes for the Trauma Sensitive Pedagogy pilot study for early childhood educators. Paper presented at the Virtual Biennial Meeting of the Society for Research in Child Development 2021, April Penuel W.R. Hill H.C. Building a knowledge base on research-practice partnerships: Introduction to the special topic collection AERA Open 5 4 2019 332858419891950 Perou R. Bitsko R.H. Blumberg S.J. Pastor P. Ghandour R.M. Gfroerer J.C. Hedden S.L. Crosby A.E. Visser S.N. Schieve L.A. Parks S.E. Hall J.E. Brody D. Simile C.M. Thompson W.W. Baio J. Avenevoli S. Kogan M.D. Huang L.N. Centers for Disease Control and Prevention (CDC) Mental health surveillance among children--United States, 2005-2011 MMWR Supplements 62 2 2013 May 17 1 35 Rankin B.A. First Person: Secondary traumatic stress: One teacher’s experience Phi Delta Kappan 102 4 2020 58 59 Rodriguez C.M. Lee S.J. Ward K.P. Pu D.F. The perfect storm: Hidden risk of child maltreatment during the COVID-19 pandemic Child Maltreatment 26 2 2020 139 151 33353380 Ronfeldt M. Loeb S. Wyckoff J. How teacher turnover harms student achievement American Educational Research Journal 50 1 2013 4 36 Rosenblatt A. Tseng V. The demand side: Uses of research in child and adolescent mental health services Administration and Policy in Mental Health and Mental Health Services Research 37 1 2010 201 204 20393795 Roy E.A. New Zealand braces for spike in child abuse reports as Covid-19 lockdown eases 2020 The Guardian https://www.theguardian.com/world/2020/may/18/new-zealandbraces-for-spike-in-child-abuse-reports-as-covid-19-lockdowneases Rudasill K.M. Niehaus K. Buhs E. White J.M. Temperament in early childhood and peer interactions in third grade: The role of teacher-child relationships in early elementary grades Journal of School Psychology 51 2013 701 716 10.1016/j.jsp.2013.08.002 24295144 Schelbe L. Wilson D.L. Fickler W. Williams-Mbengue N. Klika J.B. Bridging the gaps among research, policy, and practice in the field of child maltreatment through cross-sector training and innovation International Journal on Child Maltreatment: Research, Policy and Practice 3 3 2020 293 305 Snodgrass Rangel V. A review of the literature on principal turnover Review of Educational Research 88 1 2018 87 124 Streeter L.G. Why so many teachers are thinking of quitting 2021, October 18 The Washington Post Magazine https://www.washingtonpost.com/magazine/2021/10/18/teachers-resign-pandemic/ Sutcher L. Carver-Thomas D. Darling-Hammond L. A coming crisis in teaching? Teacher supply, demand, and shortages in the U.S 2016 Learning Policy Institute Thomson D. Ryberg R. Harper K. Fuller J. Paschall K. Franklin J. Guzman L. Lessons from a historic decline in child poverty Child Trends 2022 10.56417/1555c6123k Tirrell-Corbin C. Panlilio C. Klika J.B. The epidemic behind the mask: COVID-related education inequities 2021 The Hill https://thehill.com/opinion/education/539180-the-epidemic-behind-the-mask-covid-related-education-inequities Tirrell-Corbin C. Panlilio C. Klika J.B. Educators navigating the epidemic of inequities in low-resourced communities: The importance and power of coalition building. Presentation at the Kempe International Virtual Conference: A Global Call to Action to Change Child Welfare 2021, October Tseng V. The uses of research in policy and practice 2012 Society for Research in Child Development Washington, DC U.S. Census Bureau Income, poverty, and health insurance coverage in the United States: 2020 https://www.census.gov/newsroom/press-releases/2021/income-poverty-health-insurance-coverage.html 2021 U.S. Department of Labor Employment and Earnings by Occupation 2019 Washington, DC. Retrieved from: https://www.dol.gov/agencies/wb/data/occupations Whitney D.G. Peterson M.D. US national and state-level prevalence of mental health disorders and disparities of mental health care use in children JAMA Pediatrics 173 4 2019 389 391 10.1001/jamapediatrics.2018.5399 PMID: 30742204; PMCID: PMC6450272 30742204 Wulczyn F. Looking ahead: The nation’s child welfare system after coronavirus 2020 The Imprint https://imprintnews.org/child-welfare-2/looking-ahead-the-nations-child-welfare-systems-after-coronavirus/41738 Yard E. Radhakrishnan L. Ballesteros M.F. Emergency Department visits for suspected suicide attempts among persons aged 12–25 years before and during the COVID-19 pandemic — United States, January 2019–May 2021 MMWR. Morbidity and Mortality Weekly Report 2021 70 2021 888 894 10.15585/mmwr.mm7024e1	36464510	PMC9715397	NO-CC CODE	2022-12-03 23:20:15	no	Child Abuse Negl. 2022 Dec 2;:105941	utf-8	Child Abuse Negl	2,022	10.1016/j.chiabu.2022.105941	oa_other
==== Front CJEM CJEM Cjem 1481-8035 1481-8043 Springer International Publishing Cham 36456743 405 10.1007/s43678-022-00405-x Commentary EM:POWER: if not us, who? If not now, when? http://orcid.org/0000-0002-5394-8845 Chochinov Alecs [email protected] 1 Petrie David A. 2 Kollek Daniel 3 Innes Grant 4 1 grid.21613.37 0000 0004 1936 9609 Department of Emergency Medicine, Max Rady College of Medicine, University of Manitoba, Winnipeg, MB Canada 2 grid.55602.34 0000 0004 1936 8200 Department of Emergency Medicine, Dalhousie University, Halifax, NS Canada 3 grid.25073.33 0000 0004 1936 8227 Division of Emergency Medicine, Department of Medicine, McMaster University, Hamilton, ON Canada 4 grid.22072.35 0000 0004 1936 7697 Department of Emergency Medicine, University of Calgary, Calgary, AB Canada 2 12 2022 13 8 10 2022 23 10 2022 © The Author(s), under exclusive licence to Canadian Association of Emergency Physicians (CAEP)/ Association Canadienne de Médecine d'Urgence (ACMU) 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 pmcTimes are tough. As we prepare to turn the corner on the COVID-19 pandemic, the world faces serious, inter-dependent challenges to its environment, economy, geopolitical stability and health. All of these are inextricably linked, something to keep in mind as we contemplate the future of healthcare. By way of illustration, some experts believe that over the next 2 decades, over a billion climate refugees will relocate to cooler parts of the world, further stressing all our systems, challenging the notion of sovereignty over closed borders, compelling us to “shed some of our tribal identities and embrace a pan-species identity” [1]. This scenario poses an existential question: as evidence mounts that our global ecosystem is failing, how can humankind collaborate to make it through to the future and thrive as a species, especially when the pain and suffering are so unequally shared? Looking at emergency care through a similar lens, we too face a clear and present existential threat and practice within an unstable ecosystem, in which it already feels like waves of medical refugees are landing on our shores. With the decline and fall of their primary care home, they present to EDs with more co-morbidities and in ever greater numbers. Those requiring admission often face siloed services, quotas, ‘closed borders’; so, they wait, and those behind them wait, with predictable results. Commissioned by CAEP in 2021, the mandate of the EM:POWER* Task Force (*Emergency Medicine: Patient Care-Organizational-Workforce-Ecosystem Redesign) is to propose a systems-based approach to the future of emergency care, where integrated networks with multiple access points–not just emergency departments–are responsive to patient needs and adaptive to changing conditions. Commitment to a common purpose, a new approach to workforce planning, transition to a Learning Health System [2] and accountability from the top down will be required to enable systems-level change. The ‘caps’ in EM:POWER also speak to bottom-up, front-line ownership, imparting a sense of agency to emergency medicine to catalyze meaningful change. If not us, who? Be it the environment, the economy, or healthcare, leading large-scale change [3] is never easy. It differs from the processes we are familiar with and defies the easy “fixes” [4] that are so tempting to political leaders and bureaucrats. We cannot act alone, so our task force is consulting broadly, engaging with leaders and organizations within and beyond medicine. Thus far, the response has been extremely encouraging. “We don’t rise to the level of our goals; we fall to the level of our systems”—James Clear The late Mikhail Gorbachev saw a failing system in the USSR decades ago, and launched a program of reform, widely known as glasnost (openness) and perestroika (redesign). He believed the way to a better future for his country was through increasing transparency, new ideas and empowering organizations and individuals outside their secretive bureaucracy. That courageous approach won him a Nobel Prize but not many friends in the halls of power. Here in Canada, our secret is out—our health care systems are failing and this is having an overwhelming impact on emergency departments and the patients they serve. We experience it up close every day—packed waiting rooms, ambulance offload delays, treatment delays resulting in unnecessary harm, frustrated patients leaving before they are seen, and moral injury to our team of providers. Experienced nurses are leaving for more sustainable lives; staff shortages are causing burnout and, ultimately, ED closures. It is a vicious cycle of demand, dysfunction and delay, system-wide. Canada is next to last among OECD countries for access to family physicians, near last in acute care beds, and worst in terms of waits for specialists, elective surgery, and advanced imaging. As a consequence, Canadians have the highest rate of ED use in the first world and visits are rising rapidly [5]. Critically, our healthcare system is not a system in any cohesive sense, but rather a disconnected set of programs with countless loci of decision-making, and dynamic, often unpredictable responses to simplistic or ill-conceived plans. Governments and health administrators struggle with understanding complex systems, and often make decisions based on narrow assumptions that may be harmful to other patients and the system as a whole. COVID-19 is not a root cause of system dysfunction; it is more the last straw and another call to action. Reports and commissions addressing health system dysfunction date back decades. Despite these, the system has continued its seemingly inexorable decline, leading some to question its viability and core values. The Canada Health Act [6] expresses those values in five foundational principles, but the promise of “reasonable access” for all Canadians to health services “without…barriers” is a long-lost ideal. The Romanow Report [7], which also articulated a value-based future for health care, supported system modernization and recommended a sixth foundational principle—accountability—which we will strongly promote, at all levels of governance. But large numbers of health providers with a solitary focus on individual patients do not make for an effective, responsive or sustainable system. A formal curriculum on Health Systems Science must therefore be a key component of the next generation of health sciences education, to complement basic and clinical sciences [8]. A redesigned healthcare ecosystem must have a clear purpose, to provides direction, coherence, and an overarching WHY to all that we do. This is well articulated by the IHI’s triple aim: improving patient experience, improving population outcomes, and optimizing value (outcomes per dollar spent) [9]. We recommend the adoption of a quadruple aim, recognizing the vital importance of readiness and resilience. The two are linked–readiness is the preparedness to respond while resilience in the ability to recover. Both provider and system readiness/resilience are required to address the inevitable but unpredictable surges that occur during normal times, and to meet the unknown risks of the future. Without these, the other three aims are just empty words, and the system will be buckle under unmet demand/capacity mismatches and a burned-out workforce, as our experience with COVID-19 has illustrated. “What if there was a different starting point–the intended function of the system–and planners worked backward to determine the most suitable form for that function?” [10] The primary function of emergency medicine is the assessment and treatment of unexpected, time-dependent illness and injury. However, our present workplace is incompatible with our prime directive and the myriad other functions we undertake in the midst of intractable access block. Emergency care in the future will be defined less by bricks and mortar, or geography, and increasingly by the range of competencies we offer to patients and other caregivers, even at great distances. New technologies and national licensure will even the playing field for acute care in historically underserviced areas, allowing physically isolated providers to become part of a virtual team practice guided by emergency physicians, and patients to receive care by in their home environment, whenever possible. But emergency medicine cannot achieve its present or future functions in the health system if it also must be the “safety net” for other specialties or sectors that are unable to fulfill their own mandates, be it through poor design, misunderstanding, or lack of accountability. Counterbalancing the paradigm of sovereignty over closed borders with a common good approach, in which resources are equitably allocated to those most in need–including those we have variously called boarders, orphans or refugees– will be a major step forward. We all feel most comfortable and safe in a familiar environment. Physicians are very comfortable in the house of medicine, but it’s really a neighborhood of houses, each one home to a specialty with its own professional identity, evaluating its own set of discrete patients and problems, in the way we have all been taught. Do not look now, but just outside our neighborhood the temperature is rising and there are already medical refugees at our doors. The effects of wildfires, floods and hurricanes are now directly impacting the health of our patients; socioeconomic inequity fuels other fires, even as burnout spreads amongst our own colleagues. If we are to create a better future for ourselves and our patients, we must start by agreeing on core values, accepting the interlinked responsibilities for patient, population and system health, and supporting our workforce. We must be literate in health systems science, establish truly integrated networks of care, and expand collaborative health services research. Leaders within complex health systems must show they are competent in these essential concepts, nurture innovation, be comfortable with uncertainty, but still insist on accountability. These are the some of the ideas that are driving the EM:POWER Task Force and could contribute to a future framework for emergency care, in a redesigned system. Ultimately, it is hoped that this project will be a catalyst for other disciplines and organizations within healthcare to join with us to help plan a better future. If not now, when? Declaration Conflict of interest None of the authors has any conflict of interest to declare. ==== Refs References 1. Vince G Nomad century: how climate migration will reshape our world 2022 Flatiron Books 2. Kraft S Caplan W Trowbridge E Building the learning health system Describing an organizational infrastructure to support continuous learning Learn Health Syst 2017 1 4 e10034 10.1002/lrh2.10034Team 31245569 3. Sustainable Improvement and the Horizons Team. Leading Large Scale Change: A guide to leading large scale change through complex health and social care environments. https://www.england.nhs.uk/wp-content/uploads/2017/09/practical-guide-large-scale-change-april-2018-smll.pdf. Accessed 15 Aug 2022. 4. Kim DH System archetypes 1: diagnosing systemic issues and designing high-leverage interventions 2000 Pegasus Communications 5. Canadian Institute for Health Information. Sources of potentially avoidable emergency department visits. Ottawa, ON: CIHI. 2014. https://secure.cihi.ca/freeproducts/ED_Report_ForWeb_EN_Final.pdf. Accessed 15 Aug 2022. 6. Government of Canada Justice Laws Website, Canada Health Act (R.S.C., 1985, c. C-6) https://laws-lois.justice.gc.ca/eng/acts/C-6/page-1.html. Accessed 15 Aug 2022. 7. Building on values: the future of health care in Canada: final report/Roy J. Romanow, Commissioner. CP32–85–2002E.pdf; 2002. 8. Gonzalo JD Chang A Dekhtyar M Health systems science in medical education: unifying the components to catalyze transformation Acad Med 2020 95 9 1362 1372 10.1097/ACM.0000000000003400 32287080 9. IHI Triple Aim Initiative. 2007. https://www.ihi.org/Engage/Initiatives/TripleAim/Pages/default.aspx. Accessed 15 Aug 2022. 10. Grumbach K Redesign of the health care delivery system a Bauhaus, “Form Follows Function” approach JAMA 2009 302 21 2363 2364 10.1001/jama.2009.1772 19952323	36456743	PMC9715401	NO-CC CODE	2022-12-03 23:20:15	no	CJEM. 2022 Dec 2;:1-3	utf-8	CJEM	2,022	10.1007/s43678-022-00405-x	oa_other
==== Front Pharmaceut Med Pharmaceut Med Pharmaceutical Medicine 1178-2595 1179-1993 Springer International Publishing Cham 36456682 455 10.1007/s40290-022-00455-7 Editorial The Future of Health and Science: Envisioning an Intelligent HealthScience System Kohn Martin S. 12 Kush Rebecca 3 Whalen Matthew 4 Tobin Mary 4 Dori Dov 5 http://orcid.org/0000-0002-5859-3625 Koski Greg [email protected] 46 1 MedPredixAI, LLC, Winston-Salem, NC USA 2 grid.241167.7 0000 0001 2185 3318 Center for Biomedical Informatics, Wake Forest University School of Medicine, Winston-Salem, NC USA 3 Elligo Health Research, Austin, TX USA 4 Alliance for Clinical Research Excellence and Safety (ACRES), 7 Liberty Square, Suite 2772, Boston, MA USA 5 grid.6451.6 0000000121102151 Technion, Israel Institute of Technology, Haifa, Israel 6 The Albright Stonebridge Group, Denton’s Global Advisors, Washington, DC USA 2 12 2022 16 21 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 pmcIntroduction The term “healthcare system” is commonly and loosely used to describe the existing disconnected, inefficient, ineffective and expensive approach to health management, including disease prevention, diagnosis and treatment. The unfortunate reality is a complex array of proprietary enterprises, from individual and group medical practices, hospitals and medical centers, to networks of affiliated centers and practices. These range from small to huge in both size and complexity, all attempting to use technology and best-practices to deliver evidence-based care to a variety of patient populations with varied economic means and accessibility. Simply put, even without delving into technological, administrative and financial realms, clearly, a healthcare enterprise exists, but it falls far short of a true healthcare system. A recent article in the IEEE Systems Journal bluntly notes: “The definition and characteristics of systems have eluded recognition and understanding for a very long time, as different people refer to the concept of system in various ways,” adding that one survey of experts used “100 definitions of system and formed assumptions and hypotheses about the different worldviews represented by different groups of definitions.” [1] The consequences of not understanding a true systems approach to healthcare and biomedical research plague the health endeavor today as evidenced by the ongoing COVID-19 pandemic. From a systems perspective, the COVID-19 pandemic has fostered confusion as to what is and what is not a systemic intervention even while also offering useful insights into how the situation might be improved. Systems thinking, grounded in systems engineering principles, has been utilized by “high hazard” enterprises to deal with identification and prevention of catastrophic events in mission-critical situations, e.g., a nuclear reactor core meltdown or prevention of aviation accidents. Systems thinking and engineering have improved transportation and distribution systems and banking operations, benefiting many. Even today’s automobiles are themselves elaborate systems capable of transporting their occupants in comfort and safety, sometimes without a driver! Despite decades of discussion, application of systems thinking and design principles to health and science remain elusive at best. In some microcosms, success has been achieved by limiting the scope of the size and complexity of the endeavor. Yet, to be effective, a systems approach to health and science must encompass the entirety of healthcare and biomedical research–the people, processes, policies and technologies, and the many stakeholders, each with their own agendas and vested interests. Ways that healthcare and biomedical research currently affect each other and how they should in the future can be improved through enhanced systems development. Looking Beyond the Current State At present, promising work is being done in developing the Learning Health Systems (LHS) concept [2–5]. As conceived, a LHS is an evolving health, research, and technology ecosystem that is potentially transformative for all stakeholders. This requires understanding what exists and how it operates, all while promoting a vision of a future state, leveraging identification of critical pain points, pressures, and operational gaps. Measurements of progress and adaptations learned from failures within the healthcare environment are valuable. The intentions of the LHS approach are laudable, but its impact has as yet been marginal. Expanding and more rigorously applying systems-thinking and systems engineering principles to the conceptual underpinnings of a LHS, one can envision a future state that can be characterized as a comprehensive Intelligent HealthScience System. Through systems thinking and design, health and science can be more than merely integrated. Beyond people, processes and technologies working together in a fully coordinated and synergistic fashion, application of advanced analytics, artificial intelligence and machine learning, error analysis and other tools can render the system “intelligent”--that is, capable of actually augmenting our limited human abilities to search, analyze, interpret and apply data and real world experience. Doing so would likely improve outcomes, identification of risk factors and prevention of disease across diverse populations, optimize care (both precision and personalized), while also optimizing safety, accessibility and economics. Indeed, such an integrated ‘intelligent system’ could revolutionize our approach to healthcare, public health and medicines development, all while saving enormous amounts of resources, enhancing productivity, equity and quality of life. Any design for an Intelligent HealthScience System must include what has traditionally been called “hard” and “soft” systems dimensions. “Hard” systems dimensions are those at the focus of systems engineering, predicated on a clear picture of the problem(s) at hand and the desired future state, which allows for clear steps to be followed to reach this outcome. “Soft” system dimensions identify issues that all agree need resolving but consensus on what the root problem is cannot be found, nor is the desired outcome agreed upon by all parties, e.g., global warming. Soft systems methodologies are useful for solving such complex “wicked” challenges. Soft systems are often equated with and intimately related to human factors. These are exemplified by the difficulties of dealing with an event like the COVID-19 pandemic, where there are significant and diverse social, political, economic, equity, healthcare, and research aspects that cannot be dismissed. Underlying each position there is a worldview that arises from individual and group cultural beliefs, values, assumptions, and personal experiences. Each worldview is the filter through which the world is interpreted and assists in decision-making [6]. A hard systems approach alone is not equipped to deal with this type and extent of complexity. Together, hard and soft system perspectives provide an understanding of what happens among components of a true system, reinforcing its fundamental and essential integrated, i.e., interconnected and interdependent nature. The value proposition of an integrated Intelligent HealthScience System is powerful. Well before the COVID-19 pandemic, the biomedical research and development and healthcare ecosystem had been confronting challenges of leveraging such technologies as artificial intelligence, machine learning, and natural language processing. Difficulties in collaborating, both within and among organizations, overcoming “silos” and resistance to deep process transformation posed challenges to progress, as did slowly-responding regulatory actions to address problems posed by the development and adoption of new technologies. The pandemic merely heightened awareness of these shortcomings and intensified calls for action. Core Challenges Numerous platforms, applications and algorithms have been developed to improve healthcare, but core challenges persist. The abundance of these challenges requires systemic understanding and attention. Creating an Intelligent HealthScience System requires public trust and engagement beyond that currently addressed in non-systematic “patient-centricity” initiatives. The direction of personalized medicine reinforces this necessity in both research and care–the ultimate decentralized clinical trial (DCT), “trials of one”, and analysis of what is now appreciated as “real world evidence.” Historically, the healthcare enterprise as we know it bears little resemblance to a true integrated system. Rather, it is a maze of healthcare and research silos, proprietary and competitive networks and business endeavors that effectively work in opposition to, rather than in support of, system integration. The current state reflects stakeholder self-interest, and intentionally limited interconnectivity, optimizing one part at the others’ expense. An integrated systems approach to health, science and business, supported by technological innovation, could positively impact wellness, disease management, knowledge creation, outcomes, productivity, quality of life, medicines development, and, yes, cost. To securely acquire, share and apply knowledge in the current healthcare environment is challenging. For example, chaotic and episodic acquisition and application of both new knowledge, and confronting misinformation, has greatly challenged an effective response to the COVID-19 pandemic. The trajectory of the pandemic, and of health more broadly, is influenced by multiple agents, each acting upon, and is acted upon, by other agents in the system, often in unknown ways. Systems on the other hand are designed to deal with multiple, potentially flawed data streams affecting complex interactions of humans, hardware, and information to overcome such shortcomings. Multiple data streams and varying types of data from disparate technologies contribute to an ongoing need for standardization. Limited progress toward harmonizing healthcare and data standards with existing global research standards is inadequate. A better understanding of the value and challenges around data formats and data sharing should encourage pressure for vendors to support standards that enable interoperability among technologies to support a true systems approach. Security and trust are essential. A collaborative enterprise-wide therapeutic ecosystem spanning care, research, medicines development, regulatory and ethical oversight, and broad public access requires a robust, secure and trustworthy systems approach by stakeholders, simultaneously addressing comprehensive change management and transformation. Decision makers too often confront vested and short-term (i.e., typical bottom line) interests and impediments to change. Impediments to potential benefits and gains from greater longer-term socially purposeful planning and bridging operations could be overcome through systems thinking. Unfortunately, decision makers in the current environment are not inherently systems thinkers and the disconnected, highly siloed enterprise is not conducive to an integrated systems approach. Elements of Systems Modeling Development and delivery of effective therapeutics are embedded in complicated subsystems that need to be orchestrated into an integrated system, as opposed to the currently fragmented environment. The current state is a collection of loosely integrated entities and functions at best. Too often, one element may be optimized while ignoring, and detrimentally impacting, its connectivity and interoperability with others. For example, most healthcare guidelines and clinical trials are disease-focused rather than person-focused, failing to address multiple health challenges associated with an individual patient. Similarly, well-intentioned efforts to protect the privacy of patient information has unintentionally, but knowingly, impacted the ability to access, share and analyze patient data–what many cancer patients refer to as “protecting us to death”. Most essential to effectively bridging healthcare and research is this fundamental principle: A system is “an arrangement of parts or elements that together exhibit behavior or meaning that the individual constituents do not” [7]. Put simply, the performance of the whole exceeds the capability of the individual parts–together they produce synergy. Modeling an Intelligent HealthScience System One approach to facilitating systems thinking is modeling. A systems model or map can represent the various agents and their interactions, allowing both analysis and aggregation of their individual functions and behaviors using tools such as Agent-Based Models (ABM) and Systems Dynamics Models (SDM). The validity of ABM is well established [8]. The complexity of healthcare makes it difficult to identify and understand all the potential interactions that occur, but “ABM can also incorporate ongoing learning from events whereby patients can be influenced by their interactions with other patients or health workers and by their own personal experience with the health system” [9]. Hybrid models may produce better results by taking advantage of the strengths of different methods and mitigating their weaknesses [9]. Similar models can be created to bridge healthcare and biomedical research. Although such an undertaking may seem overwhelming, there are established systems languages and methodologies, such as Object-Process Methodology (OPM), to support the effort that have been adopted as a standard by the International Organization for Standardization (ISO) [10]. The OPM model integrates the functional, structural, and behavioral aspects of a system in a single, unified view. Although the model is not the solution to a problem, methods like OPM allow expression of all the components of the environment and the different kinds of interactions among them. OPM is based on a minimalistic universal ontology, in which “everything … is either an object or a process, and a process is not necessarily a method of a single object class…open[ing] the door for the possibility of modeling systems so that both their structural and procedural relations are represented within the same frame of reference without suppressing each other” [11]. In brief, visualizing what must be done significantly helps to realize the transformation. As an example of this approach, Figure 1 presents a top-level Object-Process Diagram (OPD) of an Intelligent HealthScience System and the health promoting processes that it would enable. A unique feature of OPM is its bimodal representation in both graphics and natural language text. In this OPM depiction, objects are presented as green rectangles and processes are presented as blue ellipses. Inter-relationships and exchanges are illustrated by directional arrows.Fig. 1 Top-level Object-Process Diagram of an Intelligent HealthScience System. Objects are presented as green rectangles and processes are presented as blue ellipses. Inter-relationships and exchanges are illustrated by directional arrows. Society consists of many Individuals. Intelligent HealthScience System exhibits Human Health Promoting. Human Health Promoting of Intelligent HealthScience System affects Society. Society, which is the assembly of its many individuals, is expected to be affected by the process Human Health Promoting, which is the service enabled by our envisioned Intelligent Health System. Human Health Promoting consists of Analytics, Healthcare Providing and Scientific Research. The Human Health Promoting process comprises three sub-processes: Healthcare Providing, Analytics, and Scientific Research, the integration of which is expected to generate the synergy and promote human health. Intelligent HealthScience System consists of Basic & Medical Research System, Current Siloed Healthcare Environment and Technology. Scientific Research requires Basic & Medical Research System. Healthcare Providing requires Current Siloed Healthcare Environment. Analytics requires Technology. Intelligent HealthScience System has three parts: Basic & Medical Research System, Current Siloed Healthcare Environment, and Technology, each enabling a corresponding sub-process of Human Health Promoting. Individual exhibits Health Level. Health Level of Individual can be current or improved. Society exhibits Quality of Life. Quality of Life of Society can be current or improved. The Society’s Quality of Life is the result of the health level of its Individuals. Each has two states: current and improved. Human Health Promoting of Intelligent HealthScience System changes Health Level of Individual from current to improved. Human Health Promoting of Intelligent HealthScience System changes Quality of Life of Society from current to improved. Human Health Promoting improves an individual’s Health level, and consequently, the whole Society’s Quality of Life improves Data to support this model come from multiple sources, including known interactions among identified agents and components, but many of these can, should and in all likelihood will change. As new agents and interactions are identified, they can be added to the model and their impact (effects of new data, estimating consequences if flawed) assessed. The model does not do the work, rather it creates a framework to represent the thought process brought to the problem. It also allows prediction of the impact of an intervention or a change in one of the agents to the extent the model is valid. Experience with predictions from the models, true or false, provide some of the most essential feedback to amend an approach and the model. Models may not be perfect but properly formed – and informed – models provide valuable insight. Requirements for effective modeling include:Standards for all data collection and formats for patient medical data representations, as well as for managing fundamental processes, beginning with addressing the continuing proliferation and ad hoc development of formatting and collecting; Structures, both internal to and across stakeholders, for interaction and information exchanges; Solution innovations enabling aggregation, integration, analytics, and decision-making; and, Identification of systems dynamics in both hard and soft system/human factor terms considering aspects like scale, randomness, and diversity of perspective. Clearly, effectively modeling a system also requires both an understanding of the current state and a clear vision of a future state, identifying and leveraging critical pain and pressure points, gaps, as well as measurements for progress. Adaptation learned from failures and acceptance of less-than-optimal results promotes continuing evolution of the health-science ecosystem [12]. A fresh look at improving the current status, includes consideration of the following:Data accessibility, variety and analytical capacity; “Silos” both internal to an organization and across organizations and disciplines; Collaboration models spanning business, investment, and science; Person-centeredness/empowerment toward achieving a “social compact” of law, regulations, and ethics; Comprehensive change management rather than the typically narrowly-focused efforts, including social and human dimensions like “de-innovation” (overcoming ingrained practice and process biases); and, Innovation for substantive transformation overall across the multi-stakeholder ecosystem, emphasizing safety and effectiveness mutually. A true systems approach is inherently oriented toward understanding and managing complex and larger-scaled challenges and discovering unintended consequences that well-intentioned “low-hanging fruit” solutions do not address. In systems thinking terms, discussions should be refocused away from simple delivery of services and population-based research protocols toward meeting needs of specific patients on the one hand and those of the entire enterprise on the other. This can be achieved by expanding and more effectively using data generated by the care process, stimulating fresh thinking beyond the basic notion of delivering care. An example of such expanded learning and application of Human Health Promoting, as illustrated in Figure 1, encompasses wellness promotion, disease prevention, care optimization, and patient level research including analysis of real world evidence. Such modeling helps one envision how a fully integrated system operates to constantly improve outcomes. Moving Forward While existing healthcare and research organizations are making progress, the road to a true integrated system is challenging. The COVID-19 pandemic has emphasized understanding health and its intersection with research differently than in the past, for example:Focusing on individuals in addition to larger cohorts – why some victims are asymptomatic while others die, experience such a range of adverse events, recover only partially, or become “long-haulers”; Dealing with the current state of fragmented and inadequate data, in which clinicians are grasping for solutions without the time or ability to undertake “precision COVID-19 medicine,” the extreme duress and intensity of contending with an unknown disease entity, and the uneven distribution of health care resources; Understanding health status as the result of a multitude of agents or influencers – each interacting with the others in known and unknown ways, like, in the case of COVID-19, political views and attitudes toward public health, which played important roles in the pandemic’s development; and, Taking on the challenges of information silos, since meaningful data standards and interoperability are often identified as major barriers to progress in medical care and research. Individual providers cannot hope to evaluate all the different kinds of data that may be relevant to the individual patient. Doing so requires a system-based analytic process to provide patient-specific insights for the patient and clinician to consider in making decisions in addition to sound population and cohort-based recommendations. The potential is enormous, and the motivation is powerful, but commitment to build consensus among essential parties has as yet been insufficient. There are many reasons for this limited commitment to change, including concerns about requisite technical skill, uncertainty about the purpose or value of proposed changes, professional liability, and comfort with the status quo [13–15]. The complexity and magnitude of the challenge is simply overwhelming to many, but these concerns are not compelling reasons for inaction. Computational power and tools such as machine learning and artificial intelligence put an Intelligent HealthScience System in reach technologically. An opportunity to apply systems thinking to these challenges is before us, but success requires bringing together essential stakeholders and processes, in addition to technology. Fostering the will and commitment to do so is essential. Together, systems-oriented leaders can envision a future in which all the essential components are interconnected, data is securely and privately exchangeable, mechanisms for observing and analyzing events and patterns are developed, and the different perspectives, needs, and constraints impacting stakeholders are delineated. A data-driven, standards-based systems approach was recently proposed to manage COVID-19 and better prepare for future pandemics, but a coordinated and engaged effort to realize such an approach remains elusive [16]. The vision we offer for an Intelligent HealthScience System is even more challenging, as its goals are more far-reaching and comprehensive. Energized and informed by the COVID-19 pandemic, stakeholders could come together to further envision and design an Intelligent HealthScience System. Yes, they face formidable challenges. Action requires leadership and shared commitment, and now, with a global pandemic as a catalyst, an opportunity exists to accept the challenge of leadership. Convening a group of committed stakeholders is an essential first step. An initial ‘envisioning exercise’ among diverse stakeholders can set the stage for establishing cross-functional teams of experts to undertake the design and development of the system and to generate resources needed to build it. We propose this as a logical next step and are currently working to make it a reality. Yes, we can more effectively integrate healthcare and research to benefit all–the question remains, will we seize this promising opportunity? The authors gratefully acknowledge the conceptual contributions of Dr. Gary Gottlieb. Declarations Funding This is an unfunded publication. Conflict of interest Authors have no competing interests to report. Ethical approval No ethics approval was necessary as this was not a research project involving patients or patient data. Consent for publication Not applicable. Consent to participate Not applicable. Availability of data and materials All data used in this manuscript is from publications listed in the bibliography. Code availability Not applicable. Authors’ contributions All authors contributed text and citations to the paper. ==== Refs References 1. Dori D Sillitto H Griego RM System definition, system worldviews, and systemness characteristics IEEE Syst J. 2020 14 2 1538 48 10.1109/JSYST.2019.2904116 2. Carey G Malbon E Carey N Systems science and systems thinking for public health: a systematic review of the field BMJ Open 2015 5 e009002 10.1136/bmjopen-2015-009002 26719314 3. Pitt M Monks T Crowe S Systems modeling and simulation in health service design, delivery and decision making BMJ Qual Saf 2016 25 38 45 10.1136/bmjqs-2015-004430 26115667 4. Brand SL Thompson Coon J Fleming LE Whole-system approaches to improving the health and wellbeing of healthcare workers: a systematic review PLoS ONE 2017 12 12 e0188418 10.1371/journal.pone.0188418 29200422 5. Augustsson H Churruca K Braithwaite J Mapping the use of soft systems methodology for change management in healthcare: a scoping review protocol BMJ Open 2019 9 e026028 10.1136/bmjopen-2018-026028 30940758 6. Checkland P Systems thinking, systems practice including soft systems methodology: a 30-year retrospective 1999 1 Hoboken, NJ, USA Wiley 7. INCOSE. Overview. Available from https://www.incose.org/about-systems-engineering/system-and-se-definition. Accessed 21 Nov 2022. 8. Seid M Bridgeland D Bridgeland A A collaborative learning health system agent-based model: computational and face validity Learn Health Syst. 2021 5 3 e10261 34277939 9. Cassidy R Singh NS Schiratti P-R Mathematical modelling for health systems research: a systematic review of system dynamics and agent-based models BMC Health Serv Res 2019 19 845 10.1186/s12913-019-4627-7 31739783 10. International Organization for Standardization. Automation systems and integration — Object-Process Methodology. Available from https://www.iso.org/obp/ui/fr/#iso:std:iso:pas:19450:ed-1:v1:en. Accessed 21 Nov 2022. 11. Dori D Object-process methodology – a holistic systems paradigm 2002 Berlin, Heidelberg, New York Springer Verlag 12. Chapman J System failure: why governments must learn to think differently 2004 2 London UK Demos 13. Ubel PA Asch DA Creating value in health by understanding and overcoming resistance to de-innovation Health Aff 2015 34 2 239 244 10.1377/hlthaff.2014.0983 14. Gupta DM Boland RJ Jr Aron DC The physician’s experience of changing clinical practice: a struggle to unlearn Implement Sci 2017 12 28 10.1186/s13012-017-0555-2 28245849 15. LeTourneau B Managing physician resistance to change J Healthc Manag 2004 49 5 286 288 15499802 16. Ros F Kush R Friedman C Addressing the COVID-19 pandemic and future public health challenges through global collaboration and a data-driven systems approach Learn Health Syst. 2020 5 1 e10253 33349796	36456682	PMC9715402	NO-CC CODE	2022-12-03 23:20:15	no	Pharmaceut Med. 2022 Dec 2;:1-6	utf-8	Pharmaceut Med	2,022	10.1007/s40290-022-00455-7	oa_other
==== Front Environ Dev Sustain Environ Dev Sustain Environment, Development and Sustainability 1387-585X 1573-2975 Springer Netherlands Dordrecht 2742 10.1007/s10668-022-02742-4 Article The regulatory role of sustainable product design media and environmental performance in the impact of the Covid-19 epidemic on corporate sustainability: an application in Turkey http://orcid.org/0000-0003-0332-4185 Durmaz Yakup [email protected] 1 http://orcid.org/0000-0001-5626-8856 Fidanoğlu Ahmet 2 1 grid.440437.0 0000 0004 0399 3159 Hasan Kalyoncu University, Gaziantep, Turkey 2 Şanlıurfa municipality, Gaziantep, Turkey 2 12 2022 116 21 4 2022 24 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. 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 research aims to investigate the effect of the COVID-19 epidemic, which entered the world agenda in 2019 and affected the whole world, on the corporate sustainability of businesses. The mediating effect of sustainable supply on this effect and the regulatory effect of environmental performance were investigated. The research was conducted among 235 businesses operating in Turkey. The data obtained using the survey method were analyzed in SPSS and AMOS analysis programs. As a result of the analyses obtained, it was determined that the COVID-19 epidemic significantly affected the corporate sustainability of the enterprises and that the environmental performance of the enterprises was a regulatory effect, together with the mediation of sustainable supply. It is understood day by day that COVID-19 negatively affects the economies of the countries. However, despite these negative effects; It is expected that the results of this research will contribute to the literature with a significant effect on the institutional sustainability of the COVID-19 epidemic. Keywords COVID-19 outbreak Corporate sustainable Sustainable supply Environmental performance ==== Body pmcIntroduction Corporate sustainability is accepted as a fundamental paradigm and solution for establishing a prosperous future for businesses. However, the social sustainability issues and epidemic problems of Covid-19 affected businesses and interrupted sustainable development plans. To date, corporate sustainability frameworks have had a relatively narrow view of this paradigm. This study aims to identify the effects on corporate sustainability while providing a framework to enable more sustainable business practices as a result of the COVID-19 pandemic. To fill the gap in the literature, it is aimed to develop a framework by drawing attention to important sustainability indicators. Institutional sustainability is the potential to meet the needs of the present without undermining the ability of future generations to meet their needs (Brundtland, 1987). Corporate sustainability has become an important alternative to business as usual, which seems to focus on short-term productivity with little or no long-term social and environmental impacts (Ikram et al., 2020). Despite the COVID-19 pandemic, the business environment is changing rapidly, guided by social, economic, and ecological developments (Initiative, 2018). For these and other related reasons, businesses have become increasingly interested in corporate sustainability. While managers realize that sustainability integration is an opportunity, they rarely take it into account in their tactical management decisions (Gonzalez-Perez & Leonard, 2015). Considering sustainability in corporate strategy and processes has become a promising way to cope with the changing business environment and increasing pressure for social sustainability performance. As sustainability-related decisions are made at a strategic level, there is increasing scientific interest in the field of strategic management in connection with incorporating sustainability into an enterprise's strategy, vision, change management, and culture (Engert et al., 2016). Due to the COVID-19 pandemic, the modern world is experiencing a global economic crisis with more than 180 million cases, resulting in more than 4 million deaths affecting over 200 countries as of June 2021 (Worldometers, 2020). Many businesses closed and the unemployment rate rose. According to the World Health Organization, Covid-19 has turned into a unique global crisis and challenges organizations, which ultimately hurts corporate sustainability and sustainable supply (Hakovirta & Denuwara, 2020). Until now, businesses have not had to plan for the Covid-19 outbreak. Moreover, corporate sustainability has been marginalized due to organizations, governments, and society’s strug, going to survive in times of pandemics. They stated that the COVID-19 outbreak has created new challenges for corporate sustainability, as supply chains and manufacturers have faced negative effects of the pandemic, such as market recession, the bankruptcy of SMEs the, employee resignations, loss of income. Therefore, we see the post-COVID-19 era as an opportunity for organizations to rethink the change in production, sustainable supply chains, and sustainability strategies. In partial response to the recent and growing call to understand the impacts of Covid-19 on businesses, we conducted this research by contributing to the growing literature on the subject by developing a multidimensional framework for corporate sustainability. Conceptual framework COVID-19 outbreak COVID-19 appeared on the world agenda in December 2019, China's Wuhan, and affected the whole world (Hui et al., 2020). As the spread of the virus expanded in geographical regions, the number of deaths increased (Wu & Mcgoogan, 2020). It can be spread from person to person through tiny droplets from the nose or mouth that are spread when a person with COVID-19 coughs, sneezes, or exhales. These droplets fall on objects and surfaces around the person. Other people then catch Covid-19 by touching these objects or surfaces and then touching their eyes, nose, or mouth. People can also catch COVID-19 if they breathe in droplets from someone with COVID-19 who is coughing, sneezing, or exhaling. It is difficult to confirm exact details about the origins and initial spread of Covid-19, but the disease originated in a wholesale market in Wuhan, a city of about 11 million people in eastern China, and some market traders may have contracted the disease after contact with animals at the market. The impact of the COVID-19 crisis on the global economy has been devastating, threatening national and global growth with the closure of many manufacturing plants and manufacturing facilities, massive disruption of supply chains, reductions in production output, and corporate profitability. Decreased trade in consumer goods from large multinational enterprises and less developed economies has been caused (de Carvalho & Senhoras, 2020). The COVID-19 crisis has had a very negative impact on sectoral economies. Increasing restrictions on the movement of people and curfews in every country of the world have deeply affected the service sector, especially industries that involve physical interactions, such as retail trade, entertainment, and hospitality, leisure, and transportation services. In addition to the economic effects of the COVID-19 epidemic, we also see the environmental effects. The relationship between the COVID-19 crisis and corporate sustainability can be seen in various ways. Initially, several environmental improvements were identified, including significant reductions in pollution levels and greenhouse gas emissions, dramatic reductions in air travel volume, and restrictions on the movement of people in motor vehicles, following the closure of many power generation plants and factories. Business travel or shopping trips in towns and cities have decreased. However, such improvements will certainly not be sustained if/when the economy improves. For example, we must consider the environmental signals of Inger Anderson, Director of the United Nations Environment Programme, and what they mean for our future and well-being, because Covid-19 is by no means a silver lining for the environment. Visible positive effects are temporary, whether through air quality improvement (Net, 2020). Corporate sustainability Corporate sustainability is the use of the strategies and approaches that businesses take to eliminate the damage they cause to the environment in the production process, for business continuity in the long term (Yavuz, 2010). Institutional sustainability is not enough for businesses to be companies that only produce and create economic value, but they try their best to eliminate or minimize the negative environmental impacts they cause while continuing their normal activities (Hahn & Scheermesser, 2006). Corporate sustainability has started to gain increasing importance in the business world. Many of the leading players in the retail industry have implemented sustainability programs designed to incorporate environmental, social and, governance issues into their business strategies (Jones et al., 2011). Covid-19 poses several challenges in production sectors. On the one hand, these challenges consist of the need to respond to investors' demands to promote long-term reductions in incorporation emissions and pollution levels, increased use of renewable energy sources, a clearer commitment to waste recycling, and the development of circular economy principles. On the other hand, the COVID-19 outbreak will reduce the availability and access of capital in businesses. The COVID-19 pandemic is forcing retailers across all industries to cut costs and change priorities. This pressure may cause corporate sustainability initiatives to be put on the back burner. It seems unlikely that businesses will continue to try to reduce their environmental impact in the face of the COVID-19 outbreak (Rachel, 2020). That Covid-19 is an opportunity to reduce the prevalence of a lifestyle that relies on large outputs of energy and mater, is in the long run, policymakers should work on this, and the COVID-19 pandemic can contribute to a sustainable consumption transition. Such a transition will require major changes in the current business model of the vast majority of businesses in the manufacturing sectors. Currently, neither the majority of businesses in the industries nor their customers seem very enthusiastic about such an opportunity policymaker sakers advocating such a future (Cohen, 2020). Sustainable supply Widely accepted as a development that meets the needs of the present without compromising the ability of future generations to meet their own needs, sustainability is a dynamic long-term process and poses several challenges for supply chain managers. World Commission on Environment and Development (1987). To ensure sustainability, they must address the interconnected social, environmental, and financial objectives in the supply chain (Mitchell & Walinga, 2017; Raut et al., 2017). Sustainable procurement aims to reduce the negative effects of supply chain operations and improve the social, economic, and environmental performance of organizations (Vargas et al., 2018). Business organizations have also begun to integrate sustainable sourcing as a way to improve their brand image and reputation. In addition, sustainable procurement reduces risks and vulnerabilities such as environmental damage and labor shortages, which can increase business stability and reduce delays and costs in production and distribution processes. However, many businesses in emerging economies today lack the expertise needed to successfully implement and adopt sustainability practices. The main reason for this is that the field of sustainability research is still not well defined, developed, or implemented (Luthra et al., 2017). Sustainable procurement tools have also been badly affected in a new, far-reaching devastating pandemic known as COVID-19 (Ivanov & Das, 2020; Kilpatrick & Barter, 2020). This epidemic is the most devastating disruption in the last few decades (Remko, 2020). The COVID-19 pandemic is no longer just a global health crisis, but also a crisis in terms of the labor market, sustainable supply, and economy. Inappropriate monitoring of these challenges leads to serious disturbances in sustainable supply networks and thus societies. These disturbances can cause permanent economic damage. The COVID-19 pandemic has led to a major disruption in many value chains around the world, particularly in sustainable procurement (Govindan et al., 2020). Looking at the literature; explored operational excellence to improve sustainability (Bag et al., 2020). de Sousa Jabbour et al. (2020) addressed trends and challenges in sustainable supply chains. Gupta et al. (2020) studied sustainability innovation to support policy makers in decision making. Environmental performance Emphasizes that environmental responses to the economic slowdown triggered by the COVID-19 pandemic hurt the environment for people and organizations. (Hallema et al., 2020). Analysis of environmental awareness of water consumption represents an important tool for water efficiency and decision-making procedures, in line with the challenges arising from the scarcity of water resources (Gomez-Lianos et al., 2020). Zambrano-Monserrate et al. (2020), another important factor is air pollution. The COVID-19 outbreak is between emergency measures and improving air quality, cleaning beaches,es and reducing environmental noise.There is an important relationship (Tahir & Batool, 2020). After the 2009 collapse, global carbon dioxide emissions decreased by 0.3% to limit local transport and industry, bringing improving air quality for the next generation with the Covid-19 Pandemic ( Rugani & Caro, 2020). However, there are also negative secondary aspects such as reducing recycling and increasing waste, compromising the pollution of physical spaces, where the greatest reduction of waste and recycling is the negative side effects of Covid-19 (Zambrano-Monserrate et al., 2020). It reports that population density and climatic conditions may affect COVID-19 cases. (Pirouz et al., 2020). However, he emphasizes that the Coronavirus epidemic has positive environmental consequences, namely significant reductions in air pollution due to the large-scale slowdown in economic activity (Sarkis et al., 2020). By affecting the ecological balance, the COVID-19 outbreak may reveal an untapped potential for people's conscious, environmentally conscious and market trends (Sofo & Sofo, 2020). As a result of the Covid-19 outbreak; by offering a small-scale approach to the sustainable use of natural resources, it can lead to self-sufficiency, self-regulation, sustainability and environmental protection. The positive environmental impacts of the COVID-19 pandemic are likely to be temporary but can serve as an example of changes in society's lifestyle. (El Zowalaty et al., 2020). It offers a new existing link, where a window of opportunity is opened to accelerate environmental awareness towards broader sustainability transitions after the COVID-19 pandemic (Cohen, 2020; Sarkis et al., 2020). Methods of the research Researchmodel and hypotheses In this section, information about the research model and research model and the hypotheses of the research will be given (Fig. 1).H1: The perception of COVID-19 has a significant impact on corporate sustainability. H2: Sustainable supply has a mediating effect on the impact of COVID-19 perception on corporate sustainability. H3: The perception of COVID-19 has a significant impact on sustainable supply. H4: Sustainable procurement has a significant impact on corporate sustainability. H5: Environmental performance has a regulatory role in the impact of COVID-19 perception on corporate sustainability. Fig. 1 Model of the research Population and sample of the research 350 enterprises are operating in Turkey. Turkey’s sized industrial zones. The sample size was calculated as 229 enterprises, taking into account the 1% margin of error within the 99% confidence limit from the population (Sekaran, 1992). In this context, a total of 235 business surveys were conducted randomly by cluster sampling method. Questionnaire application Due to the COVID-19 outbreak, several companies were asked to respond by sending questionnaires on the internet. Some of the businesses were interviewed face to face and asked to answer the questionnaire. 240 forms were returned from the questionnaire forms obtained from the senior managers of the enterprises. 2 of the questionnaire forms were not included in the research due to sloppy filling and 3 due to incomplete filling. Only 240 top managers of 340 enterprises that constitute the universe of the research accepted to participate in the research and answered the survey questions. Majority of the senior managers who participated in the survey are 88% male (n = 207), associate degree and undergraduate degree graduates 70% (n = 164) in terms of education level, 87% of them are over 25 years old (n = 205) in terms of age., 79% (n = 215) of the study shows a partial distribution according to years. Business managers participating in the research in terms of demographics; 12% of them are female (n = 28. 2) and 88% (n = 206.8) are male, consisting of business managers. In addition, the education level of business managers is 14% (n = 33) primary education, 16% (n = 37.6) high school, 45% (n = 105) associate degree, 25% (n = 58.75) undergraduate education.. In terms of age range, 13% (n = 30) are 18–25, 32% (n = 73.6) are 26–30 years old, 39% (n = 91) are 31–35 years old, 16% are (n = 37.6) It consists of the age range of 36 years and above. According to the working years; 9% (n = 21) less than 1 year, 25% (n = 58.75) 1–5 years, 22% (n = 51.7) 6–10 years, 26% (n = 61) between 11–15 years and 18% (n = 42.3) of 16 and above. Scales of research In the research, the effect of the COVID-19 perception on corporate sustainability and the mediating effect of sustainable supply in this effect and, the regulatory role of environmental performance were investigated. Goodness-of-fit values for the scales used in the study are given in Table 1. The results regarding the reliability and validity of the scales used in the research are given in each section. KMO and Barlett tests were found to be in accepted standards in all scales. Answers in the scale were used on a 5-point Likert scale (1 = I disagree 5 = I were formed.Table 1 Effects of the Covid-19 outbreak on sustainability Positive effects Adverse effects Effects of Covid-19 on the environment Sharp reductions in travel-related energy consumption and carbon emissions Disruption of the clean energy business, Instant reduction in electricity consumption, Disruption of clean energy supply chains, The collapse of fossil fuel markets (especially coal, oil and, gas), The real and significant risk of recovery in consumption accelerated by incentive and improvement packages, Immediate reduction in global air pollution Disruption of off-grid energy markets and eroded progress in energy access programs COVID-19 perception scale: The scale created to measure the COVID-19 perception of the top managers of the businesses was developed and validated by Arpacı et al. (2020), and the COVID-19 perception scale, which consists of two dimensions (Economic factors and social factors), was used. Questions on this scale worry me about the possibility of under-supply of food due to the causes of the COVID-19 pandemic. The possibility of shortages in cleaning supplies due to the COVID-19 outbreak worries me. I stock up on food for fear of the Covid-19 outbreak.After the COVID-19outbreakoutbrea't feel comfortable with my materials at home unless I constantly check them. After the COVID-19 pandemic, I feel very anxious when I see people coughing. After the COVID-19 outbreak, I actively avoid people I see sneezing from. After the COVID-19 outbreak, I realized that I was spending a long time cleaning my hands. Fear of falling ill with COVID-19 seriously hinders my social relationships. I cannot prevent myanxiety about catching COVID-19 disease from others. It is two-dimensional in total and consists of 9 expressions. As a result of the reliability analysis conducted by Arpacı et al. (2020), the Cronbach alpha reliability coefficient was determined as 0.80 for economic factors and 0.92 for social factors. The translation of the scale into Turkish was done by experts. The translated scale was translated back into English and its accuracy was tested. In this study, confirmatory factor analysis was performed to test the structural validity of the scale. As factor loads, it was found to be two-dimensional by the scale of the research. It was determined that the factor loads were between 0.56–89 for economic factors and between 0.57–90 for social factors. Corporate Sustainability scale: To measure corporate sustainability of enterprises, the scale developed and used by Shashi (2018) in his study titled 'Corporate sustainability orientation, supply chain integration and a causal analysis of SMEs' performance' will be used in the research. The scale consists of four items that aim to measure the orientation of business owners and managers to sustainable procurement. The scale was first translated from English to Turkish by experts. It was translated back into English and compared with the original version of the scale, and it was found to be appropriate. As a result of the reliability analyzes analysis done y Shashi (2018), the Cronbach alpha reliability coefficient was determined as 0.85 for corporate sustainability. In this study, confirmatory factor analysis was performed to test the structural validity of the scale. It has been determined that factor loads take values between 0.54-0.91 for corporate sustainability. In addition, factor loadings were controlled by exploratory factor analysis and confirmatory factor analysis. The scale is one-dimensional and consists of 6 expressions. As a result of the factor analysis, the one-dimensional structure of the data was confirmed and it was determined that the factor loads took values between 0.57-0.90. Goodness-of-fit values of the scale are given in Table 1 together with the scales. Sustainable Procurement Scale:The scale, which was made by Shashi (2018) reliability and validity, Shashi (2018) to measure the sustainable supply levels of business executives, consists of 5 statements. Expressions Our Company shares its production planning and forecasts with its suppliers during the COVID-19 process. Our business attaches importance to communication with its suppliers during the COVID-19 process. Our business has an integrated order system with suppliers during the COVID-19 process. We share inventory information with its suppliers during the COVID-19 process of our business. Our business is planning its activities with our suppliers during the COVID-19 process. English-Turkish and Turkish-English translations of the scale into Turkish were made by experts in the field of Turkish, and exploratory and confirmatory factor analysis was performed by applying it on the pilot and main sample. In the research, exploratory factor analysis was performed primarily to measure the construct validity of the scale. As a result of the analysis, it was seen that the factor loads, which were compatible with the one-dimensional structure of the scale, took values between 0.52–89. The factor loads of the scale, in which the confirmatory analysis was performed in the Amos package program, took values between 0.54 and 0.91, and the goodness of fit values are given in Table 1. Environmental Performance Scale: The scale, which was created to measure the environmental performance of the senior managers of the enterprises and whose reliability and validity was made by Shashi (2018), will be used by Shashi (2018). Questions of the scale Our business is more effective than its competitors in reducing air emissions during the COVID-19 process. Our business makes more efforts than competitors to reduce wastewater during the COVID-19 process. Our business does more than its competitors to reduce its solid waste during the COVID-19 process. Our business makes more efforts than its competitors to reduce the consumption of dangerous/harmful/toxic substances during the COVID-19 process. Our business spends more effort than competitors to reduce environmental accidents during the COVID-19 process. Our business works harder than its competitors for its environmental sensitivity during the COVID-19 process, it consists of statements. The scale is one-dimensional and consists of 6 statements, and in the reliability analysis conducted by Shashi (2018), the Cronbach alpha coefficient was determined as 91 for environmental performance. The translation of the scale into Turkish was done by experts. Exploratory and confirmatory factor analysis was carried out by applying the scale expressions on the pilot and main sample by making English-Turkish and Turkish-English translations. In the research, exploratory factor analysis was first applied to measure the construct validity of the scale. As a result of the analysis, it was seen that the factor loads that were compatible with the one-dimensional structure of the scale took values between 0.49 and 0.90. The factor loads of the scale, in which the confirmatory analysis was performed in the Amos analysis program, took values between 0.50 and 0.89, and the goodness of fit values are given in Table 1. Amos analysis program was used to determine the results of confirmatory factor analysis. As seen in Table 2 of the confirmatory factor analysis results, RMSEA (Root Mean Square Error of Approximation), CFI (Comparative Fit Index), GFI (Goodness of Fit Index), AGFI (Adjusted of Goodness Fit Index) values were found to be within acceptable limits.Table 2 Goodness of fit values of the scales as a result of confirmatory factor analysis Values x2 df CMIN/DF ≤ 5 GFI ≥ .85 AGFI ≥ .80 CFI ≥ .90 NFI ≥ .90 TLI ≥ .90 RMSEA ≤ .08 Covid-19 outbreak 8.6 3 2.93 .93 .89 .92 .94 .92 .06 Corporate Sustainability 8.1 5 .98 .96 .96 .94 .96 .96 .03 Sustainable supply 2.5 2 .95 .92 .93 .91 .93 .91 .07 Environmental performance 2.6 2 1.94 .99 .98 .96 .94 .94 .03 Goodness-of-fit values appear to be compatible with “acceptable” standards Results The data obtained as a result of the research were made in SPSS 26 program and Amos analysis program. In the light of the results obtained, at the first stage, the averages, standard deviations, and correlation values of the data obtained regarding the COVID-19 perception, corporate sustainability, sustainable supply, and environmental performance of the senior managers of the enterprise were examined. In the second stage of the analysis, the mediation effect was investigated and data related to its regulatory role were included. The means, standard deviation, on,s, and correlation values obtained as a result of the analysis are given in Table 2. When the results of the analysis were examined, it was determined that the COVID-19 perception and corporate sustainability were significant affected and that it had a significant effect on sustainable supply. In addition, it has been determined that sustainability has a significant effect on corporate sustainability. As a result of the analysis, when the variables are examined separately, there are indications that sustainable supply may have a mediating effect on the effect of the COVID-19 perception on corporate sustainability. An indirect effect of 0.06 was determined in the effect of COVID-19 perception on institutional sustainability. Within the scope of the analysis, an indirect effect of 0.06 was determined on the effect of the COVID-19 perception on the institutional sustainability and unity for the mediation effect, and based on these findings (Baron & Kenny, 1986; Quoted by Şimşek, 2007) hierarchical regression analysis was used to determine the mediation effect. A three-stage regression analysis proposed by Baron and Kenny (1986) was conducted to explain the mediating effect of sustainable procurement between the perception of Covid-19 and corporate sustainability. To be able to talk about mediation, three conditions must be met.The independent variable must affect the mediating variable. The independent variable must affect the dependent variable. When the mediator variable is included in the second regression analysis, the regression coefficient of the independent variable on the dependent variable should decrease, while the mediator variable should have a significant effect on the institutional sustainability of the dependent variable. To determine the intermediation of sustainable supply within the scope of intermediation, hierarchical regression analysis was used. Data related to the mediation test are given in Table 3 in detail.Table 3 Mean, standard deviation,and correlation values of the variables Variables Cover ss 1 2 3 4 Covid-19 perception 3.9 .72 (.87) Corporate sustainability 3.8 .82 .05* (.90) Sustainable supply 3.7 .69 .06* .03 (81) environmental performance 4.1 .76 .03 .02 .04 (90) It is seen that p < .05 and * p < .01 The Process Macro extension developed by Hayes was used to test the regulatory role of environmental performance in the impact of Covid-19 perception on corporate sustainability. Detailed data of the analysis are given in Table 4. Table 4 The ıntermediary role of sustainable procurement in the ımpact of Covid-19 perception on corporate sustainability Corporate sustainability Test 1-corporate sustainability Test 2 sustainable supply β T β T Covid-19 perception .407 11.67 .387 11.38 R = .407, R2 = .164, Straight. R2 = . 162, F = 136.2, P(F) = 0.00 R = .387, R2 = 136, Straight. R2 = . 156, F = 131, P(F) = 0.000, Sur edrk Setup surβ = .448, R2 = 200, Straight. R2 = 200, F = 176, P(F) = 0.000 The following conclusions have been reached regarding the regulatory role of environmental performance in the impact of COVID-19 perception on corporate sustainability (Table 5).Table 5 The mediation effect of sustainable procurement on the ımpact of Covid-19 perception on corporate sustainability Test 3-Corporate sustainability β T Covid-19 .266* 8.67 Wall supply .326 9.189 R = 506, R2 = .256 Straight. R2 = 255, F = 116, P(F) = 0.000 Sobel Test = 4.61**; p < 0.001 Dependent Variable (Covid-19 perception, Independent Variable (institutional sustainability), p < 0.05, *p < 0.01, p < 0.001 N = 710 Environmental performance has a moderating effect on the perception of COVID-19 in those with low regulatory role (-1SD) (B = 1.02; SE = 0.039; t = 25; p < 0.05). The regulatory role of environmental performance has moderate effects on the perception of COVID-19 (B = 0.99; SE = 0.037; t = 26; p < 0.05). Environmental performance has a moderating effect on the perception of covid-19 in those with a high regulatory role (+ 1SD) (B = 0.80; SE = 0.51; t = 15; p < 0.05). The independent variables used within the scope of the research were obtained in the PROCESS Macro Model output file, which was run by centering (Hayes, 2018). As seen in Fig. 2, the corporate sustainability line (y = 4.02E-3 + 0.77x) and β = 1.02, 0.99 and 0.80 values show that the trend is decreasing, as well as the Kovid19 perception line. (y = 2.72 + 0.25x) and β = 1.02, 0.99 and 0.80, it is seen that it tends to decrease. Environmental performance has a regulatory role in the impact of Covid19 perception on corporate sustainability and the H5 hypothesis was supported. As a result; as the Covid-19 perceptions of the enterprises increase, the corporate sustainability of the enterprises increases, as well as the environmental performance of the enterprises (Table 6).Fig. 2 Regulatory impact of environmental performance on the impact of COVID-19 perception on corporate sustainability Table 6 The regulatory role of environmental performance in the impact of the COVID-19 outbreak on corporate sustainability analysis results The dependent variable β SE t p LLCI ALSO Still .0601 .0380 1.581 .1151 − .0148 .1350 environmental performance 1.0473 .0418 25.040 .000 .9649 1.1297 Covid-19 perception .1885 .534 3.5315 .003 .2937 .0833 Çev_PerCOVID-19perception .1209 .0275 .4.4054 .000 .0669 Effect SE t p LLCI ALSO Environmental performance regulation In the corporate sustainable impact of COVID-19 perception of enterprises (M ± SD) − 1SD 1.0272 .0398 25.8085 .0000 .9487 1.1056 M .9925 .0374 26.5495 .0000 .9188 1.0662 + 1SD .8000 .0511 15.6482 .0000 .6993 .9008 LL lower limit CI upper limit 5000 sample bootstrap method free of bias error was used. Pre-analysis averages of environmental performance and COVID-19 perception variables were standardized H5: The hypothesis of “environmental performance has a regulatory role in the impact of the COVID-19 perception on corporate sustainability” was accepted. Discussion and conclusion The impact of the perception of COVID 19 on sustainability seems to be a temporary situation. With the complete disappearance of the COVID 19 perception, businesses will resume their activities and practices will be abandoned at the point of sustainability. For the continuation of positive practices related to sustainability, which is formed with the perception of COVID 19; Sustainability obtained during the COVID 19 epidemic can be ensured permanently by state administrators' enacting laws supporting sustainable behavior, media and social media managers keeping sustainability on the agenda, consumers turning to the products of businesses that produce sustainable products, and avoiding the use of non-sustainable products. This research was conducted among senior managers of businesses operating in Turkey. Research It has been investigated how the COVID-19 epidemic affects the corporate sustainability of businesses. In addition, it has been investigated whether the environmental performance of enterprises has a regulatory effect with the mediation of sustainable supply in the impact of the COVID-19 epidemic on corporate sustainability. According to the results obtained in the research; As a result of the literature review, it has been understood that the effect of the COVID-19 epidemic on corporate sustainability is positively significant. In the correlation analysis, it was understood that there was a strong relationship between all three variables. In the first stage of the hierarchical regression analysis, it was determined that the COVID-19 epidemic had a positive and significant effect on Corporate Sustainability. The result obtained is consistent with the results of the literature review. In the second phase of the analysis, it was investigated how the COVID-19 epidemic affected sustainable supply. The result obtained is consistent with the results of the literature review and it has been found to have a positive and significant effect. In the third stage of the research; it has been investigated whether sustainable supply has a mediating effect on the impact of the COVID-19 epidemic on corporate sustainability. It has been determined that sustainable supply has a mediating effect on the effect of the obtained COVID-19 epidemic on Corporate sustainability. This result; similar to other studies in the literature. As a result of the data analysis obtained in the research, it was determined that the COVID-19 epidemic had a positive and significant effect on the corporate sustainability of the enterprises. These results are consistent with the results of the study by Sarkis et al. (2020), A Brave new world: Lessons from the COVID-19 outbreak for sustainable supply and transition to manufacturing. In addition, Zhang et al. (2017) are similar to the study he conducted among businesses in China after the 2008 financial crisis. After the 2008 financial crash in China; Regulatory, technological, and cultural changes have occurred to address the shortcomings highlighted by the disaster. For example, China has invested heavily in an a stimulus package that includes a significant focus on renewable energy, resulting in accelerated growth in related industries and reductions in production costs that benefit companies and communities around the world. In addition, it has been determined in the research that sustainable supply has a mediating effect on the effect of COVID-19 on corporate sustainability. In the literature, studies investigating the intermediary effect of sustainable supply on the institutional sustainable effect of the COVID-19 epidemic could not be reached. However, we see it in the literature in studies investigating the mediation effect of sustainable supply. In the literature, Sharma et al. (2008), Hosseinpour et al. (2015), Shashi (2018) and Fidanoğlu and Aytekin (2021), similar studies were conducted and similar results were obtained. In addition, according to the results obtained in the research; It has been determined that the impact of the COVID-19 epidemic on corporate sustainability has a positive and meaningful effect on the environmental performance of enterprises. The COVID-19 outbreak has caused economic and social damage in most parts of the world, and although all sectors of the economy have been badly affected, it has had positive effects on the corporate sustainability of businesses. Although the improvement of air quality and pollution during the COVID-19 epidemic is temporary, it is important in terms of the perception of the environmental damage caused by the enterprises after the COVID-19 epidemic. As a result of this research; it has been determined that the COVID-19 epidemic has achieved gains in terms of corporate sustainability as well as financial losses of businesses. This research contains some limitations. The research does not include businesses operating only in Turkey and other provinces. The same research can be done in different provinces and between different sector groups. The research is expected to provide information to the senior managers of the enterprises in terms of corporate sustainability. In recent years, the concept of corporate sustainability has provided scholars and practitioners with a framework to explore crises, but such studies have traditionally focused on national economies and local environments. Therefore, it has not only posed a complex set of major global challenges for the COVID-19 pandemic, but also some new perspectives on the relationships between these sectors and sustainability. Looking ahead, we will see what the future holds and how these relationships will play out. On the one hand, there is hope for a return to normalcy, but at the time of this writing, the timescale and extent of such a reversal are very uncertain. However, in such a scenario, businesses could effectively try to pick up where they left off as part of a much broader post-COVID-19 recovery. As capital resources focus on economic recovery, the importance of government and corporate sustainability programs is better understood here during the COVID-19 outbreak. On the other hand, the COVID-19 outbreak has opened a window to some of the relationships between businesses and sustainability, some environmental issues that may be central to the transition to a more sustainable future. signaled changes and proposed some radical changes. This research on the COVID-19 pandemic is expected to offer solutions to corporate sustainability challenges. Author contributions The authors contributed equally. Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Declarations Competing interests The authors have no relevant financial or non-financial interests to disclose. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Arpacı, Ö. (2020). Lise Öğrencilerinin Öğretmenlerine Yönelik İletişim Algılarının ve Doyum Düzeylerinin İncelenmesi, Trakya Üniversitesi Sosyal Bilimler Dergisi, 405–417. Bag S Wood LC Xu L Dhamija P Kayikci Y Big data analytics as an operational excellence approach to enhance sustainable supply chain performance Resources, Conservation and Recycling 2020 153 104559 10.1016/j.resconrec.2019.104559 Baron RM Kenny DA The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations Journal of Personality and Social Psychology 1986 51 6 1173 1182 10.1037/0022-3514.51.6.1173 3806354 Brundtland, G. H. (1987). Report of the World Commission on environment and development:" our common future. UN. Carvalho PN Senhoras EM The impact of COVID-19 crisis on the global economy and the North American hegemonic cycle: A reading Agenda Internacional 2020 27 38 9 28 10.18800/agenda.202001.001 Cohen MJ Does the COVID-19 outbreak mark the onset of a sustainable consumption transition? Sustainability: Science Practice and Policy 2020 16 1 1 3 de Sousa Jabbour ABL Jabbour CJC Hingley M Vilalta-Perdomo EL Ramsden G Twigg D Sustainability of supply chains in the wake of the coronavirus (COVID-19/SARS-CoV-2) pandemic: Lessons and trends Modern Supply Chain Research and Applications 2020 2 3 117 122 10.1108/MSCRA-05-2020-0011 El Zowalaty ME Young SG Järhult JD Environmental impact of the COVID-19 pandemic—A lesson for the future Infection Ecology & Epidemiology 2020 10 1 1768023 10.1080/20008686.2020.1768023 32922688 Engert S Rauter R Baumgartner RJ Kurumsal sürdürülebilirliğin stratejik yönetime entegrasyonunu araştırmak: Bir literatür taraması J. Temiz. Ürün 2016 112 2833 2850 Fidanoğlu ve Aytekin, İşletmelerin Sürdürülebilir Oryantasyonun Performansa Etkisinde Sürdürülebilir Tedarik ve Ürün Tasarımının Aracılık Rolü, Balkan ve Yakın Doğu Sosyal Bilimler Dergisi, 2021; 1–7. Gómez-Llanos E Durán-Barroso P Robina-Ramírez R Analysis of consumer awareness of sustainable water consumption by the water footprint concept Science of the Total Environment 2020 721 137743 10.1016/j.scitotenv.2020.137743 32171142 Gonzalez-Perez MA Leonard L The global compact: Corporate sustainability in the post 2015 world 2015 Institutions and regulations. Emerald Group Publishing Limited 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) Transportation Research Part e: Logistics and Transportation Review 2020 138 101967 10.1016/j.tre.2020.101967 32382249 Gupta A Madhavan MV Sehgal K Nair N Mahajan S Sehrawat TS Landry DW Extrapulmonary manifestations of COVID-19 Nature Medicine 2020 26 7 1017 1032 10.1038/s41591-020-0968-3 Hahn T Scheermesser M Approaches to corporate sustainability among German companies Corporate Social Responsibility and Environmental Management 2006 13 3 150 165 10.1002/csr.100 Hakovirta M Denuwara N How COVID-19 redefines the concept of sustainability Sustainability 2020 12 9 3727 10.3390/su12093727 Hallema DW Robinne FN McNulty SG Pandemic spotlight on urban water quality Ecological Processes 2020 9 1 1 3 10.1186/s13717-020-00231-y Hayes AF Partial, conditional, and moderated moderated mediation: Quantification, inference, and interpretation Communication Monographs 2018 85 1 4 40 10.1080/03637751.2017.1352100 Hosseinpour-Mashkani SM Ramezani M Sobhani-Nasab A Esmaeili-Zare M Synthesis, characterization, and morphological control of CaCu3Ti4O12 through modify sol–gel method Journal of Materials Science: Materials in Electronics 2015 26 8 6086 6091 Hui DS Azhar EI Madani TA Ntoumi F Kock R Dar O Ippolito G Mchugh TD Memish ZA Drosten C Zumla A The continuing 2019-nCoV epidemic threat of novel coronaviruses to global health—The latest 2019 novel coronavirus outbreak in Wuhan, China International Journal of Infectious Diseases 2020 91 264 266 10.1016/j.ijid.2020.01.009 31953166 Ikram M Sroufe R Rehman E Shah SZA Mahmoudi A Do quality, environmental, and social (QES) certifications improve international trade? A comparative grey relation analysis of developing vs. developed countries Physica a: Statistical Mechanics and Its Applications 2020 545 123486 10.1016/j.physa.2019.123486 Initiative, G. R. (2018). Sustainability report. Disponvel em. http://www.globalreporting.org]. Acessado em, 17, 01–09. Ivanov D Das A Coronavirus (COVID-19/SARS-CoV-2) and supply chain resilience: A research note International Journal of Integrated Supply Management 2020 13 1 90 102 10.1504/IJISM.2020.107780 Jones P Hillier D Comfort D Shopping for tomorrow: Promoting sustainable consumption within food stores British Food Journal. 2011 113 7 935 948 10.1108/00070701111148441 Kilpatrick, J., & Barter, L. (2020). COVID-19: Managing supply chain risk and disruption. Deloitte. https://www2.deloitte.com/global/en/pages/risk/articles/covid-19-managing-supply-chain-risk-and-disruption.html. Accessed 29 Nov 2020. Kumar, A., & Ramesh, A. Belirsizlik altında yük taşımacılığı endüstrisinin sosyal sürdürülebilirlik göstergelerinin değerlendirilmesi için bir ÇKKV çerçevesi. Çok şirketli bir bakış açısı. J. Enterp. Enf. 2020'yi yönetin. Luthra S Govindan K Kannan D Mangla SK Garg CP An integrated framework for sustainable supplier selection and evaluation in supply chains Journal of Cleaner Production 2017 140 1686 1698 10.1016/j.jclepro.2016.09.078 Mitchell IK Walinga J The creative imperative: The role of creativity, creative problem solving and insight as key drivers for sustainability Journal of Cleaner Production 2017 140 1872 1884 10.1016/j.jclepro.2016.09.162 Net, H. (2020). Coronavirus: impact on the hospitality industry. Hospitality Net. https://www.hospitalitynet.org/hottopic/coronavirus. Pirouz B Shaffiee Haghshenas S Pirouz B Shaffiee Haghshenas S Piro P Development of an assessment method for investigating the impact of climate and urban parameters in confirmed cases of COVID-19: A new challenge in sustainable development International Journal of Environmental Research and Public Health 2020 17 8 2801 10.3390/ijerph17082801 32325763 Rachel, Ł. (2020). An analytical model of covid-19 lockdowns. CFM, Centre for Macroeconomics. https://www.lse.ac.uk/CFM/assets/pdf/CFM-Discussion-Papers-2020/CFMDP2020-29-Paper.pdf Raut RD Cheikhrouhou N Kharat M Bankacılık Sektöründe Sürdürülebilirlik: Stratejik Çok Kriterli Bir Analiz Otobüs. Strateji. Çevre. 2017 26 550 568 Remko VH 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 2020 40 4 341 355 10.1108/IJOPM-03-2020-0165 Rugani B Caro D Impact of COVID-19 outbreak measures of lockdown on the Italian Carbon Footprint Science of the Total Environment 2020 737 139806 10.1016/j.scitotenv.2020.139806 32492608 Sarkis J Cohen MJ Dewick P Schröder P A brave new world: Lessons from the Covıd-19 pandemic for transitioning to sustainable supply and production Resources, Conservation and Recycling 2020 159 104894 10.1016/j.resconrec.2020.104894 32313383 Sekaran, U. (1992). Middle-class dual-earner families and their support systems in urban India. Sharma U Forlin C Loreman T Impact of training on pre-service teachers' attitudes and concerns about inclusive education and sentiments about persons with disabilities Disability & Society 2008 23 7 773 785 10.1080/09687590802469271 Shashi, K. C. (2018). Food cold chain management: from a structured literature review to a conceptual framework and research agenda. International Journal of Logistics Management. Şimşek, Ö. F. (2007). Yapısal Eşitlik Modellemesine Giriş: Temel Ilkeler ve Lisrel Uygulamalaıi, Ekinoks, Ankara. Sofo A Sofo A Converting home spaces into food gardens at the time of Covid-19 quarantine: All the benefits of plants in this difficult and unprecedented period Human Ecology 2020 48 2 131 139 10.1007/s10745-020-00147-3 Tahir MB Batool A COVID-19: Healthy environmental impact for public safety and menaces oil market Science of the Total Environment 2020 740 140054 10.1016/j.scitotenv.2020.140054 32562988 Vargas JRC Mantilla CEM de Sousa Jabbour ABL Enablers of sustainable supply chain management and its effect on competitive advantage in the Colombian context Resources, Conservation and Recycling 2018 139 237 250 10.1016/j.resconrec.2018.08.018 WCED, S. W. S. (1987). World commission on environment and development. Our Common Future, 17(1), 1–91. Worldometers. (2020). COVID-19 Coronavirus pandemic. https://www.worldometers.info/coronavirus/#countries Wu Z McGoogan JM Characteristics of and important lessons from the coronavirus disease 2019 (Covıd-19) outbreak in China: Summary of a report of 72314 cases from the Chinese center for disease control and prevention JAMA 2020 323 13 1239 1242 10.1001/jama.2020.2648 32091533 Yavuz VA Sürdürülebilirlik Kavramı Ve İşletmeler Açısından Sürdürülebilir Üretim Stratejileri/Concept Of Sustainability And Sustainable Production Strategies For Business Practices Mustafa Kemal Üniversitesi Sosyal Bilimler Enstitüsü Dergisi 2010 7 14 63 86 Zambrano-Monserrate MA Ruano MA Sanchez-Alcalde L Indirect effects of COVID-19 on the environment Science of the Total Environment 2020 728 138813 10.1016/j.scitotenv.2020.138813 32334159 Zhang L Kara A Spillan JE Mintu-Wimsatt A Exploring market orientation among Chinese small and medium-sized enterprises Chinese Management Studies 2017 11 4 617 636 10.1108/CMS-08-2016-0158	36474599	PMC9715404	NO-CC CODE	2022-12-03 23:20:15	no	Environ Dev Sustain. 2022 Dec 2;:1-16	utf-8	Environ Dev Sustain	2,022	10.1007/s10668-022-02742-4	oa_other
==== Front Childs Nerv Syst Childs Nerv Syst Child's Nervous System 0256-7040 1433-0350 Springer Berlin Heidelberg Berlin/Heidelberg 36456749 5766 10.1007/s00381-022-05766-3 Original Article Epidemiology of pediatric central nervous system tumors in Uyghur: experience from a single center Wu Xuchao Dangmurenjiafu·Geng Fan Guofeng Zeng Jia Zhao Xiaoyu Sheng Chengjun http://orcid.org/0000-0002-9327-4152 Zhu Guohua [email protected] grid.412631.3 Neurosurgery, The First Affiliated Hospital of Xinjiang Medical University, Urumqi, China 2 12 2022 16 22 9 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. Purpose Retrospective analysis of clinical and epidemiological characteristics of central nervous system (CNS)tumors in Uyghur children from a single center in Xinjiang. Methods Between January 2013 and December 2021, 243 children (0–17 years old) with a clear pathological type of CNS tumor are collected and analyzed for tumor size, grade, and category, as well as their relationship with the child’s gender, age, and region of origin according to the 2021 edition of the new WHO CNS tumor classification. Outcome The 243 cases of CNS tumors in Uyghur children are predominantly from rural areas, with 144 cases (59.26%) of supratentorial tumors and 129 cases (53.09%) of low-grade tumors. With an overall male-to-female ratio of 1.43:1, a peak age of incidence of 6 to 8 years. Concluding The present study is based on a 9-year analysis of pediatric CNS data from a single center, and the center is the largest tertiary hospital in Xinjiang with large numbers of admitted patients, which may reflect some extent the clinical characteristics and epidemiological features characteristics of pediatric CNS tumors in Uyghur in Xinjiang. Keywords Central nervous system tumors Pediatric Epidemiology ==== Body pmcCNS tumors are the second most common cancers in pediatric populations, and among pediatric solid tumors, CNS tumors are the most common and have the highest mortality rate of all pediatric tumors, with incidence rates as high as 20% [1–4]. As the diagnosis and treatment of CNS tumors in pediatric patients continue to develop and improve, improvements have also been made in the detection rate of CNS tumors in pediatric patients [5]. The Uyghurs can be traced back to the third century B.C. and are part of the Turkic language family of the Altaic language family, mostly congregated in the north and northwest of China and south of Lake Baikal and between the Irtysh River and Lake Balkhash. The ancestors of the Uyghurs are thought by some to be related to the Hungarians, and Xinjiang, the main gathering place of the Uyghurs in China, is the home to 11,774,500 Uyghurs, representing more than half of the Uyghurs people in the world. However, there have been few analyses of pediatric CNS tumors in Uyghurs. For this reason, the present study is based on the most recent version of the WHO Classification of Tumors of the Central Nervous System (2021 edition), and the clinical data of Uyghur pediatric patients admitted to our neurosurgery department in the past 9 years are analyzed to supplement the clinical and epidemiological gaps of CNS tumors in Uyghur children. Materials and methods Date from 243 pediatric Uyghur (0–17 years old) with a confirmed diagnosis of CNS tumor admitted to the Department of Neurosurgery at the First Affiliated Hospital of Xinjiang Medical University between January 2013 and December 2021 are counted, including the hospitalization number, name, gender, age, date of admission, location of residence, clinical symptoms, primary diagnosis, pathological diagnosis, and imaging diagnosis. Pediatric patients are strictly checked for information to avoid duplication, and all receive treatment for tumor removal, the diagnosis of tumor tissue obtained is made by histopathology at the First Affiliated Hospital of Xinjiang Medical University, and all are confirmed to be CNS tumors by either light microscopic observation or (and) immunohistochemistry staining. Clinical manifestations In 6 cases, there are no overt CNS tumor symptoms or tumors due to head trauma or physical exam. A total of 237 cases are diagnosed as CNS tumors with clinical symptoms. Including 142 cases of headache, 113 cases of nausea and vomiting, 49 cases of loss of appetite, 44 cases of abnormal vision, 43 cases of disorders of limb movement disorder, 37 cases of balance dysfunction, 36 cases of seizure, 22 cases of dizziness, 21 cases of developmental abnormalities, 13 cases of polydipsia and polyuria, 13 cases of fever, 8 cases of limb pain, 8 cases of somnolence, 8 cases of cognitive dysfunction, 3 cases of abnormal crying, 2 cases of coma, 2 cases of difficulty in micturition, and respiratory difficulty in 1 case. Classification and gradation The pathological diagnosis of 243 pediatric cases is classified and graded in conjunction with the latest version of whom neurological classification criteria (2021 version), the age distribution, sex distribution, gender characteristics, clinical manifestations, tumor sites, and pathological types of the pediatric patients are statistically analyzed. The tumor sites are divided into the supratentorial, infratentorial, and spinal canals. Supratentorial tumors are divided into the cerebral hemisphere, the sella region, three ventricles, the lateral ventricles, and others, the infratentorial tumors are divided into the cerebellum, the brainstem, the fourth ventricle, and others. The tumor grade is divided into low-grade and high-grade, with WHO grades I and II classified as low-grade tumors and grades III and IV classified as high-grade tumors. Statistical methods Data are processed and analyzed using descriptive statistics and count data are expressed as frequency and rate (%). Outcome Histopathological types and subtypes of tumors with the size distribution Of the 243 tumors in this group, 144 are supratentorial, comprising 59.26%, 86 are infratentorial tumors, comprising 35.39%, and 13 are spinal canal tumors, comprising 5.35%. Gliomas, glial neuronal tumors, and neuronal tumors account for the largest number with 107 cases (44.03%) in the type classification of pathologic tumor histology, followed by embryonal tumors in 49 cases (20.16%), tumors of the sellar tumors in 40 cases (16.46%), and mesenchymal non-membranous epithelial tumors in 13 cases (5.35%). In 42 cases (17.28%) medulloblastoma is one of the top 5 tumor subtypes, ependymal tumors in 34 cases (14.00%), craniopharyngioma in 30 cases (12.35%), pilocytic astrocytoma in 28 cases (11.52%), and pituitary tumors in 10 cases (4.12%) (Fig. 1, Table 1).Fig. 1 Basic pathological classification of pediatric CNS tumors Table 1 Pathology and location of CNS tumors in pediatric patients Tumor classification Supratentorial Infratentorial Spinal canal Gliomas, glioneuronal tumors, and neuronal tumors Pilocytic astrocytoma 15 13 0 Glioblastoma 8 0 0 Oligodendroglioma 9 0 0 Astrocytoma 5 4 0 Ependymal tumors 12 21 1 Dysembryolastic neuroepithelial tumor 6 0 0 Others 11 2 0 Choroid plexus tumors 5 3 0 Embryonal tumors CNS embryonal tumor 2 0 0 CNS neuroblastoma 0 0 2 Medulloblastomas 1 40 1 Atypical teratoid/rhabdoid tumor 3 0 0 Pineal tumors 2 0 0 Cranial and paraspinal nerve tumors 2 3 5 Meningiomas 2 0 0 Mesenchymal,non-meningothelial tumors 11 1 1 Melanocytic tumors 0 0 0 Hematolymphoid tumors 1 0 0 Germ cell tumors Mature teratoma 1 0 2 Immature teratoma 1 0 0 Germinoma 4 0 0 Others 2 0 0 Tumors of the sellar region Pituitary adenoma 10 0 0 Craniopharyngioma 30 0 0 Metastases to the CNS 1 0 0 The bold indicates that it is one of the 12 CNS tumor classifications, and those not in bold indicate that it is a pathological subtype Gender and tumor age distribution In our group, 142 (58.44%) of 243 CNS tumor cases are male and 101 (41.56%) are female. The sex ratio is 1.41:1. All patients are between 0 and 17 years old with a mean age of 8.81 years old. The age groups are divided into 0 to 2 years group, 3 to 5 years group, 6–8 years group, 9 to 11 years group, 12 to 14 years group, and 15 to 17 years group in 3-year intervals. Sex and age distribution statistics of the pediatric population have shown that the tumors are predominantly found in pediatrics aged 6–8 years. Among them, male pediatrics are mainly distributed among 6–8 years old and female pediatrics are mainly distributed among 12–14 years old (Fig. 2).Fig. 2 Gender and age distribution of pediatric patients with CNS tumors The Grade and age distribution of the tumors Of the 243 CNS tumors in our group, low-grade is predominant, with 129 cases (53.09%), and high-grade has 114 cases (46.91%). With a ratio of 1.31:1 between the two groups. Based on statistics of tumor grade and the affected children's age show that the most affected patients are 6 to 8 years old and 15 to 17 years old in the lower grades and 3 to 5 years old in the upper grades (Fig. 3).Fig. 3 Grade and distribution of pediatric patients with CNS tumor The region of origin and the year distribution of children with tumors Our group of 243 CNS tumors includes 167 children of rural origin, constituting 68.72%, and 76 children of urban origin, constituting 31.28% of the population. With the exception of 2019, there are more children from rural sources than from urban sources for the period from 2013 to 2021 (Table 2).Table 2 Regional and year distribution of CNS tumors in the pediatric population 2013 2014 2015 2016 2017 2018 2019 2020 2021 City 4 2 6 5 4 9 18 8 20 Country 23 20 20 11 17 15 13 27 21 Discussion 35.9 cases per million children aged 0–15 years in Sweden, 36.1 cases in Kumamoto Prefecture, Japan, 32.7 cases in Yorkshire, UK, 29.9 cases in 59 European cancer registries, and 47.1 cases in the US [6, 7], while according to CBTRUS statistics reporting Whites (6.36 per 100,000), Blacks (4.83 per 100,000), Asians (3.22 per 100,000), and Asian Pacific Islanders (API) (3.48 per 100,000) [8]. We can see that there are some differences between different regions and different ethnicities of pediatric tumors worldwide, despite the lower incidence of CNS tumors in Asian pediatrics, there are some studies that show little difference between the West and the East. The Uyghur people are ethnically close to Central Asia and the Middle East, and a study from Syria showed that the incidence of CNS tumors in Syrian pediatric does differ somewhat from that in Western and Far Eastern countries [7, 9]. This is a single-center study, and accurate incidence of CNS tumors in Uyghur pediatric patients is difficult to obtain, despite a large number of pediatric admission. In this study, the ratio of male to female CNS tumors in pediatric patients in this study is 1.41:1, compared with 1.31:1 in Kumamoto Prefecture, Japan, 1.49:1 in Uganda, 1.4:1 in Pakistan, and 0.98:1 in the USA, which is close to that of Pakistan, a country located in Central Asia, with a higher number of male children than female children, but a higher proportion of female children than male children in the USA [7–10]. This may be due to some differences in the incidence of certain types of tumors between different ethnic groups, for example, germ cell tumors are more prevalent in the East than in the West, and there are significant differences between men and women in the incidence of ependymal tumors, embryonal tumors, germ cell tumors, and pituitary tumors, which are more prevalent in pediatric patients [7, 8]. Research has shown that CNS tumors in pediatric patients can present at any age, with a peak incidence in the late preschool years, but some tumor subtypes, such as medulloblastoma, have a bimodal incidence [11–14], in this group of cases, male pediatric patients are predominantly distributed in the age group of 6 to 8 year age range, with a high incidence of the school-aged patient, while female pediatric patients are predominantly distributed in the age group of 12 to 14 years, with high age. Based on whom tumor grading criteria, the number of low-grade tumors is greater than the number of high-grade tumors in the 243 Uyghur pediatric cases collected, the number of pediatric patients in each age group is also variable, with a ratio of 1.31:1, and there is no clear pattern, among which the most affected children are 6–8 years old and 15–17 years old in the lower level, and the most affected pediatric patients are aged 3 to 5 years old in the upper tier, which may reflect characteristics of the age distribution of Uyghur children to some degree, this may be due to the polymorphism of tumor genes in children and prefer to entail site of tumors [4, 6, 15, 16]. In contrast, as the Chinese government pays greater attention to pediatric medicine, they tend to conduct annual routine medical examinations for school children, which greatly assists in the early detection and diagnosis of illnesses this also prevents some low-grade tumors in children from progressing to high-grade tumors. In our group, the top 5 tumor subtypes are medulloblastoma in 42 cases (17.28%), ependymal tumors in 34 cases (14.00%), craniopharyngioma in 30 cases (12.35%), pilocytic astrocytoma in 28 cases (11.52%), and pituitary tumor in 10 cases (4.12%). Gliomas, medulloblastomas, ependymal tumors, germ cell tumors, and craniopharyngiomas are prevalent in Western countries, and astrocytoma, germ cell tumors, craniopharyngiomas, medulloblastomas, ependymal tumors are prevalent in Kumamoto Prefecture, Japan, gliomas, and embryonal tumors are prevalent in Pakistan [6, 7, 11, 17, 18]. This founding differs somewhat from data reported in Western countries and some Far Eastern nations. Specifically, pituitary tumors replaced germ cell tumors as one of the top 5 tumors with the highest prevalence, this may be because our hospital is located earlier in Xinjiang to perform endoscopic resection of pituitary tumors, which tends to be mature in terms of technology, and patients do not have to choose better medical units in mainland China, and patients from other cities in Xinjiang are more willing to visit our hospital, and therefore patients are more enriched as a result. Thus, geographic differences, ethnic differences, and local levels of medical technology levels may be the primary factors influencing the incidence of tumor types of CNS in pediatric patients. According to recent reports, the predominant site of CNS tumors in pediatric patients is supratentorial, with the brain hemispheres and the saddle region being the most common sites [1, 10, 12], but some reports suggest that tumors in pediatric patients are more likely to be found inferiorly, with the fourth ventricle being the most common [8, 19]. There are 144 cases of supratentorial tumors in this group, of which 59 cases occur in the sella, 50 in the cerebral hemispheres, 18 cases in the lateral ventricles and the third ventricle, and 17 cases in the remainder. 86 cases of infratentorial tumors, including 63 cases in the fourth ventricle, 17 cases in the cerebellar hemispheres, 3 cases in the brainstem, 3 cases in the others, and 13 cases of tumors in the spinal canal, in this group of cases, supratentorial tumors are more common than infratentorial tumors than vertebral canal: 11.08:6.62:1, with supratentorial tumors being more common. It is possible that differences in the region of origin, ethnicity, age group, and local medical resources of the included children may result in different sites of predilection of CNS tumors in pediatric patients [20]. Several studies have reported that headache, nausea and vomiting, seizures, and abnormal vision are the most common presenting symptoms of CNS tumors in children, which may be related to the fact that pediatric painful CNS tumors tend to arise in midline structures and cause symptoms of obstructive hydrocephalus as the tumor grows in size, whereas clinical symptoms in children often tend to begin with the headache. In general, patients with headaches lasting less than 6 months, unresponsive to medication, increasingly severe headaches, personality changes, or associated with abnormal neurological examinations are considered to be at high risk for structural diseases [6, 8, 18, 21]. Most of our patients are from rural areas, so there are 167 (68.72%) cases of rural background and 76 (31.28%) from an urban background, parents of rural patients often have low levels of education and economic hardship, parents of our children and local medical clinics should therefore have sufficient knowledge of patients’ clinical manifestations to make accurate judgments and select the treatment needed to avoid diagnostic errors and waste of medical resources that increase the burden on patients’ parents. Although the overall incidence of brain and other CNS tumors is 11.7% higher in urban areas compared with rural areas in the USA [8], there are more cases in rural areas than in urban cases in this group. However, as of 2019, the urban population is beginning to increase significantly, which may be related to the development of urbanization in China and the decrease in population in rural areas, on the other hand, the rapid development of pediatric neurosurgery in our hospital has resulted in more affluent urban families opting to visit our hospital instead of other medical facilities in mainland China. Due to the new coronavirus outbreak in 2020, our hospital is primarily responsible for treating children in rural areas, and as a result, there are significantly fewer cases of children in urban areas. Concluding Despite the fact that our center is the largest comprehensive tertiary care hospital in Xinjiang, we are looking forward to multi-center, multi-disciplinary, and collaborative research with the greatest number of patients. Combining the geographical characteristics of Xinjiang and the demographic features of Uyghurs, to establish a reasonable diagnostic and treatment system and assessment criteria for tumors evaluation in Uyghur pediatric patients, and the improvement of epidemiological follow-up system is of great importance for improving the prognosis of pediatric CNS tumors in Uyghurs in China. Acknowledgements Many thanks are due to colleagues from the Data Management Department of Xinjiang Medical University for providing the clinical data of pediatric patients. Author contribution All authors contributed to the study’s conception and design. Material preparation, data collection, and analysis were performed by [Guofeng Fan], [Jia Zeng], [Xiaoyu Zhao], and [Cengjun Sheng]. The first draft of the manuscript is written by [Xuchao Wu] and [Dangmurenjiafu·Geng] and [Guohua Zhu] commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding Xinjiang Uygur Autonomous Region Natural Science Foundation Youth Project. Data availability A submission to the journal implies that materials described in the manuscript, including all relevant raw data, will be freely available to any researcher wishing to use them for non-commercial purposes, without breaching participant confidentiality. All clinical information of pediatric patients can be obtained from the Data Processing Center of the First Affiliated Hospital of Xinjiang Medical University. Declarations Ethical approval and consent to participate. Not applicable. Consent for publication Not applicable. Conflict of interest The authors have no competing interests to declare relevant to this article’s content. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Ullrich NJ, Pomeroy SL (2003) Pediatric brain tumors. J Neurol Clin 21(4):897–913 2. Walker D, Wilne S, Grundy R, Kennedy C (2016) A new clinical guideline from the Royal College of Paediatrics and Child Health with a national awareness campaign accelerates brain tumor diagnosis in UK children--"HeadSmart: Be Brain Tumour Aware". J Neuro Oncol 18(3):445–454 3. La Madrid AM, Kieran MW (2018) Epigenetics in clinical management of children and adolescents with brain tumors. J Curr Cancer Drug Targets 18(1):57–64 4. Duke ES, Packer RJ (2020) Update on Pediatric Brain Tumors: the Molecular Era and Neuro-immunologic Beginnings. J Curr Neurol Neurosci Rep 20(8):30 5. Warren KE NMR spectroscopy and pediatric brain tumors J Oncologist 2004 9 3 312 318 10.1634/theoncologist.9-3-312 6. Pollack IF, Jakacki RI (2011) Childhood brain tumors: epidemiology, current management, and future directions. J Nat Rev Neurol 7(9):495–506 7. Makino K, Nakamura H, Yano S et al (2010) Population-based epidemiological study of primary intracranial tumors in childhood. J Childs Nerv Syst 26(8):1029–1034 8. Ostrom QT, Patil N, Cioffi G et al (2020) CBTRUS Statistical Report: Primary Brain and Other Central Nervous System Tumors Diagnosed in the United States in 2013–2017. J Neuro Oncol 22(12 Suppl 2):iv1-iv96 9. Kadri H, Mawla AA, Murad L (2005) Incidence of childhood brain tumors in Syria (1993–2002). J Pediatr Neurosurg 41(4):173–177 10. Maaz AUR, Yousif T, Saleh A et al (2021) Presenting symptoms and time to diagnosis for Pediatric Central Nervous System Tumors in Qatar: a report from Pediatric Neuro-Oncology Service in Qatar. J Childs Nerv Syst 37(2):465–474 11. Glod J, Rahme GJ, Kaur H et al (2016) Pediatric Brain Tumors: Current Knowledge and Therapeutic Opportunities. J J Pediatr Hematol Oncol 38(4):249–260 12. de Robles P, Fiest KM, Frolkis AD et al (2015) The worldwide incidence and prevalence of primary brain tumors: a systematic review and meta-analysis. J Neuro Oncol 17(6):776–783 13. Wilne S, Collier J, Kennedy C et al (2007) Presentation of childhood CNS tumors: a systematic review and meta-analysis. J Lancet Oncol 8(8):685–695 14. Ostrom QT, Francis SS, Barnholtz-Sloan JS (2021) Epidemiology of Brain and Other CNS Tumors. J Curr Neurol Neurosci Rep 21(12):68 15. Akeret K, Staartjes VE, Vasella F et al (2020) Distinct topographic-anatomical patterns in primary and secondary brain tumors and their therapeutic potential. J J Neurooncol 149(1):73–85 16. Reichert JL, Chocholous M, Leiss U et al (2017) Neuronal correlates of cognitive function in patients with childhood cerebellar tumor lesions. J PLoS One 12(7):e0180200 17. Bouffet E (2000) Common brain tumors in children: diagnosis and treatment. J Paediatr Drugs 2(1):57–66 18. Riaz Q, Naeem E, Fadoo Z et al (2019) Intracranial tumors in children: a 10-year review from a single tertiary health-care center. J Childs Nerv Syst 35(12):2347–2353 19. Duffner PK (2007) Diagnosis of brain tumors in children. J Expert Rev Neurother 7(7):875–885 20. Wanner M, Rohrmann S, Korol D et al (2020) Geographical variation in malignant and benign/borderline brain and CNS tumor incidence: a comparison between a high-income and a middle-income country. J J Neurooncol 149(2):273–282 21. Harrup R, White VM, Coory M et al (2021) Treatment and Outcomes for Central Nervous System Tumors in Australian Adolescents and Young Adults: A Population-Based National Study. J J Adolesc Young Adult Oncol 10(2):202–208	36456749	PMC9715407	NO-CC CODE	2022-12-03 23:20:15	no	Childs Nerv Syst. 2022 Dec 2;:1-6	utf-8	Childs Nerv Syst	2,022	10.1007/s00381-022-05766-3	oa_other
==== Front Rev Endocr Metab Disord Rev Endocr Metab Disord Reviews in Endocrine & Metabolic Disorders 1389-9155 1573-2606 Springer US New York 36456777 9772 10.1007/s11154-022-09772-3 Article Role of phase angle in older adults with focus on the geriatric syndromes sarcopenia and frailty Norman Kristina [email protected] 1234 Herpich Catrin 12 Müller-Werdan Ursula 25 1 grid.11348.3f 0000 0001 0942 1117 Institute of Nutritional Science, University of Potsdam, 14558 Nuthetal, Germany 2 grid.6363.0 0000 0001 2218 4662 Department of Geriatrics and Medical Gerontology, Charité-Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt- Universität zu Berlin, Berlin, Germany 3 grid.418213.d 0000 0004 0390 0098 Department of Nutrition and Gerontology, German Institute for Human Nutrition Potsdam-Rehbrücke, Arthur-Scheunert-Allee 114-116, 14558 Nuthetal, Germany 4 grid.452396.f 0000 0004 5937 5237 German Center for Cardiovascular Research (DZHK), Partner Site Berlin, Berlin, Germany 5 Evangelisches Geriatriezentrum Berlin gGmbH, Berlin, Germany 2 12 2022 19 18 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. Age-related changes in body composition reflect an increased risk for disease as well as disability. Bioimpedance analysis is a safe and inexpensive bed side method to measure body composition, but the calculation of body compartments with BIA is hampered in older adults. Phase angle, a raw parameter derived from bioimpedance analysis, is free from calculation-inherent errors. It declines with age and disease and is highly predictive of a variety of clinical outcomes as well as mortality. This review summarizes the current evidence linking the phase angle to geriatric syndromes such as malnutrition, sarcopenia and frailty and also investigates whether the phase angle reacts to interventions. Since the majority of studies show an association between the phase angle and these geriatric syndromes, a low phase angle is not suitable to exclusively indicate a specific condition. It does not inform on the underlying cause and as such, a low phase angle mainly indicates increased risk. Phase angle decline over time is reflected by deterioration of e.g. frailty status. It reacts to physical training and detraining, but studies investigating whether these induced changes are also associated with improved outcome are missing. Keywords Phase angle Body composition Aging Sarcopenia Frailty Malnutrition ==== Body pmcIntroduction Body composition is known to change with aging hallmarked by a decline in skeletal muscle mass as well as by an increase in total and abdominal fat tissue. As changes in body composition are associated with increased risk for disease and disability, monitoring body composition in the old is of clinical importance. Sarcopenia has been referred to as a geriatric syndrome [1] as it describes the age-associated loss of muscle mass and function, which is accompanied by a progressive decline in physical performance and is associated with a higher risk for physical disability and need for care. Sarcopenia can be present without apparent weight changes, hence assessing body composition is required in order to detect it. Moreover, older adults are particularly vulnerable to malnutrition [2] which is frequently characterized by involuntary weight loss resulting in further deleterious changes in body composition. The loss of skeletal muscle mass in malnutrition has been shown to be greater in older compared to younger adults [3]. Both malnutrition and sarcopenia are frequent in higher age with considerable overlap between the two entities, as malnutrition can contribute to or accelerate the development of sarcopenia. Importantly, malnutrition and sarcopenia have been linked to the development of the complex frailty syndrome which is a geriatric syndrome associated with lower resilience against stressors and impaired clinical outcome [4]. Both weight loss and low muscle strength or function are principal phenotypic characteristics of the physical frailty syndrome [5]. Bioimpedance analysis (BIA) which is a simple, non-invasive, inexpensive and safe bedside method, has long been considered an attractive alternative to the cumbersome and expensive imaging methods for body composition assessment such as dual X-ray absorptiometry (DXA) or computed tomography (CT). However, deriving body composition in the old using BIA is challenging, as the required conditions for calculating body compartments such as constant hydration of fat free mass or no fluid imbalance, no body shape abnormalities, are frequently not present in the old. They represent a very heterogeneous population with high inter-individual variation of fat free mass hydration and a higher likelihood of disease and multimorbidity which may further affect body composition [6]. Most regression equations for assessing body composition using single frequency BIA developed in healthy populations are not suitable in disease [7] and have even been found inadequate in the old [8, 9]; moreover, only few age-specific equations for single body compartments have been developed so far. Research on the use of raw bioimpedance parameters such as resistance (R), reactance (Xc) or phase angle without the equation-inherent errors for calculating body composition has increased in the last decade and there is a large body of evidence linking reduced bioimpedance phase angle to a variety of clinical outcomes including mortality in e.g. critically ill, patients with kidney, heart or liver disease, and patients with cancer [10]. Resistance and reactance provide information on hydration of tissues as well as cell membrane mass. R is the pure opposition of the body as a biological conductor to the flow of an alternating electric current; while reactance is the resistive effect produced by the double layer of cell membranes and tissue interfaces. The phase angle as the ratio between Xc and R is therefore interpreted as an indicator of cell membrane integrity and better cell function and health, while lower phase angle values have been associated with impaired cellular structure and greater cell death [11]. While the relationship between phase angle and overall body cell mass (BCM) has been described early, only recently, a study also confirmed an association between phase angle and cell growth and metabolism in healthy older adults. Using a proteomics approach, six protein markers were identified as being strongly associated with the phase angle [12]. When the protein markers were grouped according to their functional characteristics, regulation of the amount and growth of cells emerged as the main biological process which is related to the phase angle [12]. N-terminal pro b-type natriuretic peptide (NT-proBNP) was the key marker of the phase angle. NT-proBNP is an established marker of heart failure which is frequently characterized by overhydration. Studies in patients on hemodialysis show that changes in phase angle over a 6 month period was related to changes in NT-proBNP [13]. Also, age is one of the strongest determinants of phase angle in health, and phase angle prominently declines with higher age. While most studies are cross-sectional consistently showing peak values in young adulthood and lower values in later life [14], a recent study investigated prospective changes in phase angle over time in adults aged over 50 [15]. The authors showed an annual decline in phase angle values which mirrored the decline in muscle quality (hand grip or knee extension strength per kg muscle mass) while overall changes in body composition were not yet detectable. The annual percentage change in this study did not differ between men and women, or between the older (< 65 years) and younger group. However, the precision of the device, which allows evaluation of repeated measurements was not stated in the study. Phase angle in higher age has also been associated with prominent aging biomarkers such as the higher pro-inflammatory status in higher age (e.g. “inflammaging”) and parameters of oxidative stress [16, 17] which is linked to inflammation and cell damage and therefore also has been implicated in the development of age-related disease [18]. Since phase angle reflects both fluid distribution and BCM, an association with the amount and quality of skeletal muscle mass is expected. Moreover, phase angle has also been linked to muscle strength and functional capacity and therefore several studies have explored the relationship between phase angle values and sarcopenia in older adults. This review focusses on the association of phase angle with parameters of impaired nutritional status, with muscle mass, strength and function as well as sarcopenia and frailty in the old. It moreover explores whether a low phase angle is indicative of incident functional decline and whether it reacts to interventions such as physical training. Phase angle and its association with impaired nutritional status (malnutrition) The association between impaired nutritional status and low phase angle is well established [10]. As the calculation of body compartments using BIA in the older adults is not accurate enough, in particular in the presence of disease, the use of raw bioimpedance parameters to indicate impaired nutritional status has gained increasing attention. Malnutrition is characterized by a decrease in BCM and loss of intracellular water with a compensatory increase in extracellular water [10]. Not surprisingly, alterations of electrical tissue properties, reflected by characteristic changes in reactance, occur in malnutrition [19] which is therefore usually accompanied by a decreased phase angle. The gold standard for the diagnosis of malnutrition has long been under debate, but there are several nutrition screening tools which indicate malnutrition or the risk of developing malnutrition and which have been investigated in relation to phase angle. We studied phase angle in octogenarians living in a nursing home and found a stepwise reduction in phase angle with increasing degree of malnutrition assessed by the Mini Nutritional Assessment (MNA) [20]. Similarly, in a small sample of old frail hospital patients, the overall phase angle positively correlated with the MNA short-form score [21]. Additionally, malnutrition indicated by the Nutritional Risk Score (NRS-2002) was associated with a significantly lower phase angle in geriatric inpatients [22]. One study in old rehabilitation patients moreover showed that phase angle was predictive of malnutrition assessed with the Geriatric Nutritional Risk Index, with sensitivity and specificity however different between men and women [23]. Although many studies consistently report associations between various nutrition screening tools which indicate malnutrition and the phase angle, one systematic review failed to conclude that phase angle was an accurate predictor of malnutrition assessed by the Subjective Global Assessment (SGA) in different disease settings [24]. Recently, a study in a large clinical cohort of predominantly older patients [25] also showed that the phase angle was an independent and even better predictor of mortality compared to the SGA, indicating that a low phase angle yields additional information which cannot be attributed to nutritional status alone. Malnutrition is a complex phenomenon, and in disease, of multifactorial origin. Several factors, such as inflammation or disease-specific catabolism, which contribute to malnutrition, also have an adverse impact on phase angle itself [26]. Moreover, both inflammation and malnutrition are frequently accompanied by edema, which per se is known to decrease phase angle. Therefore, while malnutrition and low phase angle are closely linked, the underlying factors will be hard to disentangle. Although malnutrition is most likely accompanied by a reduced phase angle, a low phase angle cannot be interpreted as an exclusive indicator of malnutrition. Phase angle as an indicator of low muscle strength and sarcopenia The electric properties of cell membranes are related to both area and integrity of cell membranes and phase angle has been referred to as an index of cell membrane integrity [27] which is one determinant of membrane potential and, together with area, most likely determines muscle cell function [28]. The impedance parameters reactance and resistance normalized for height have both been shown to be independently associated with hand grip strength [29], so an association between phase angle and strength parameters is expected. In healthy old, phase angle is a predictor of muscle function such as slow gait speed [30] and in healthy older women, moderate associations were obtained between phase angle and muscle quality (strength/kg appendicular lean mass), functional capacity score (composed of various walking and rising tests) [31], gait speed [32] and with 6-meter walking test, forearm flexion and chair stand, dependent on BMI category [33]. In older patients with cancer, low phase angle predicted decreased hand grip strength, knee extension strength and reduced peak expiratory flow, as well as impaired physical function (determined by the European Organization for Research and Treatment of Cancer quality of life questionnaire) [34]. Higher phase angle values suggest higher BCM [35], of which an integral part is skeletal muscle. Not surprisingly, phase angle is increased in athletes reflecting both the higher amount of skeletal muscle mass [36] as well as performance capacity [37] and has been referred to as indicator of muscle quality [38, 39]. Similar to muscle mass, phase angle is affected by age, sex, race, BMI [40] and physical activity [41]. Given these associations, it is likely that sarcopenia, the loss of muscle mass and function in higher age, is accompanied by lower phase angle values. While the definition of sarcopenia has undergone very few changes in the last decades (from age-related loss of muscle mass to loss of muscle mass and function), the diagnostic criteria have been under debate, which in part is due to the challenges of assessing muscle mass in vivo. Whether phase angle values can be used to indicate sarcopenia has been addressed in several studies in the old or in cancer patients. A recent systematic review summarized the evidence and analysed the relationship between sarcopenia and phase angle in old adults with and without disease. A correlation between phase angle and muscle mass was seen in six studies, while phase angle correlated with the sarcopenia related muscle parameters hand grip strength and gait speed in six and three studies, respectively. Di Vincenzo and colleagues found significantly lower values of phase angle in sarcopenia in seven out of eight studies and a higher prevalence of sarcopenia in patient groups below cut offs indicating low phase angle in five out of six studies. In two out of four studies, low phase angle significantly predicted sarcopenia. The phase angle cut off values to indicate sarcopenia in the studies, however, ranged considerably from 4° to 5° [42]. It is not clear whether these differences were due to the population studied or due to the BIA device, as the studies included in the systematic review were different and differences between BIA devices from different manufacturers are well known [43]. Similar to malnutrition, when phase angle and sarcopenia were analysed with regard to survival, they were both independent predictors of death in two studies [44, 45] again implying that phase angle provides additional information and is not solely an indicator of sarcopenia. In summary, there appears to be a consistent association between sarcopenia and low phase angle, however valid phase angle cut off values to reliably indicate sarcopenia are missing. That may in part be due to the lack of gold standard methods used in the studies to assess muscle mass, but also to the multifactoriality of a low phase angle. In the end, a low phase angle cannot be used in a diagnostic way [27] but may be used as a risk factor to indicate low muscle mass or quality. The relation between phase angle, frailty and the dysmobility syndrome Not surprisingly, there is also a close association between phase angle and frailty, an important geriatric syndrome which is described as a complex multifactorial syndrome with increased vulnerability against stressors [4], of which sarcopenia has often been referred to as the biological substrate [46]. In a nationally representative sample of the NHANES study of 4,667 older adults, men and women with phase angle values below the first quintile had a 3-fold higher risk of being frail [47]. In the FraDySMex cohort study (Frailty, Dynapenia, and Sarcopenia in Mexican Adults), low phase angle was associated with frailty, even after controlling for age, sex, BMI, and comorbidities [48]. Similarly, in a large cohort of community-dwelling old in Japan, a low phase angle was linked to frailty as well to the locomotive syndrome, a condition in which mobility is reduced due impairments of the locomotor system [49] with some sex-specific differences as these associations were more pronounced in men. In a small prospective study in older adults, with every one-degree increment in phase angle over a one-year period, the likelihood of improving from frailty was 4-fold higher [50]. In older patients with rheumatoid arthritis, a low phase angle was also associated with significantly higher odds of being frail. Moreover, the change in phase angle over time was also correlated to the change in frailty status. Deterioration in frailty status was associated with a decline in phase angle, while in non-frail patients who maintained at a stable phase angle either improved or did not change [51]. Figure 1 illustrates the relationship between phase angle, malnutrition, sarcopenia and frailty. Fig. 1 Phase angle in the framework of malnutrition, sarcopenia and frailty and age-associated disease, inflammation and oxidative stress Phase angle has also been linked to the dysmobility syndrome, a recently proposed concept which integrates bone, muscle as well as adipose tissue in order to predict future fracture risk. In a large Korean cohort of community-dwelling old, low phase angle was significantly associated with a more 2-fold risk of having the dysmobility syndrome, independent from significant confounders such as sex, body mass index, and inflammation [52]. Is the phase angle a good predictor of subsequent falls, incident disability and mortality in the old? Given the observed associations with nutritional status, muscle mass, strength and function as well as frailty in older adults, the question arises whether phase angle is also a useful tool to predict age-relevant clinical outcomes such as falls, disability or mortality. In a large population of community-dwelling old without disability at baseline, low phase angle was a significant independent predictor of incident disability during a two year follow up period while appendicular lean mass corrected for BMI was not [53]. Also, older community-dwelling adults with low phase angle experienced more falls in a 6-month period [54] and phase angle as well as number of medications were significant predictors of incident fall risk, whereas age, sex, low muscle mass or low muscle function were not. In older patients with rheumatoid arthritis [51], phase angle predicted falls in a two year follow up period, whereas sarcopenia assessed by the Asian Working Group for Sarcopenia 2014 criteria was not a significant risk factor for falls. Similarly, in older patients with cirrhosis, phase angle and the disease severity Model for End-stage liver disease (MELD) score were both independently associated with incident hospitalization and mortality, but phase angle was the only independent predictor of falls [55]. Low phase angle values have also been linked to increased risk of death in many clinical settings [10]. This association of low phase angle and mortality has been observed in community-dwelling older adults as well. In the sub-analysis of the NHANES cohort in older adults described above [47], low phase angle was moreover predictive of long-term mortality independent from sex, age, race or ethnicity and comorbidity. Since these study participants were also more likely frail, this is anticipated, however, even in non-frail individuals with no little or no comorbidity at baseline, lower phase angle was associated with an increased risk of mortality during the follow up period, and the authors therefore suggest that low phase angle can be seen as a global marker of aging [47]. In a large cohort of patients aged over 65 years (n = 1,307), a low phase angle was even associated with mortality in a 10-year period following hospital stay, irrespective of age, sex, comorbidities or BMI category [56]. The presence of acute disease compounds the problem even further. In 1,071 geriatric in-hospital patients, a low phase angle was associated with a 4-fold higher risk of in-hospital mortality [57]. In a cohort of old patients with cancer, a low phase angle was an independent predictor of one-year mortality next to cancer severity, whereas grip strength was not [34]. Also, in older patients undergoing major cardiac surgery, phase angle was associated with frailty and higher risk for overall morbidity and longer hospital stay as well as higher short term and one-year mortality [58]. Moreover, in older patients with COVID-19 [59], low phase angle was an independent predictor of short-term mortality risk irrespective of age, sex, BMI, and comorbidities. In a small study in critically ill older patients, phase angle improved after five days in survivors while it decreased further in non-survivors [60]. Overall, a low phase angle is a strong independent predictor of mortality in various diseases and is also associated with a higher incidence of disability as well as falls in older adults who were healthy at baseline. Phase angle changes with training and detraining Aside from the biological determinants age and sex, phase angle is impacted by various parameters in disease, such as inflammation and catabolism; while physical activity is believed to be the strongest modifiable determinant of phase angle in health. Whether phase angle also reflects acute changes achieved by exercise or by phases of detraining has been studied in older adults. One systematic review summarizing seven studies showed in their meta-analysis that resistance training induced increases in phase angle, which result from an increase in reactance with a concomitant reduction in resistance [61]. The improvements of phase angle after a 12-week resistance training in women aged over 60 were mirrored by increases in muscle quality as assessed by the muscle quality index (MQI) which is calculated as total strength 1 repetition maximum / total body lean mass (by DXA) [62]. Interestingly, the improvement of the phase angle following a 12-week resistance training was also correlated with decreases in the pro-inflammatory cytokines TNF-α and CRP and with a decrease in advanced oxidation protein products [63]. Conversely, a systematic detraining phase of only two weeks was reflected by an acute decline in phase angle values in older trained adults. These changes were due to decreases in reactance with minimal changes in resistance (R) which were observed while they were still no measurable changes in e.g. knee extension strength [64]. While data consistently show an association between phase angle and conditions which are reflected by low BCM such as malnutrition, sarcopenia and impaired functionality, low phase angle itself is not a diagnostic or an exclusive parameter. It can be seen as an indicator of general at-risk conditions which are associated with worsened outcome. While phase angle reacts to training exercises and detraining phases as well as to changes in inflammation over time [65, 66], it has not yet been shown that these improvements are also associated with improved outcome. Conclusion In the old, phase angle is impacted by disease, inflammation and oxidative stress. Current literature shows that a low phase angle is linked to various age relevant conditions such as malnutrition, sarcopenia and frailty and predicts disability and mortality in older adults (Fig. 2). This indicates that the phase angle, while highly predictive of mortality, is not a good indicator of a single entity, as it does not inform on its etiology. A low phase angle in individuals only indicates increased risk, but the underlying cause is unknown and needs to be investigated. This also explains the wide range of proposed cut off values which differ with regard to e.g. disease and the outcome or the condition it was related to. At present, no universal phase angle cut off value exists and caution is also necessary due to the lacking comparability of BIA devices. In healthy older adults, the phase angle increases after training and decreases after detraining, however whether these induced changes also indicate a change in prognosis has not yet been investigated. It however stands to reason, as phase angle increases over time have e.g. been shown to reflect improvement from the frailty syndrome. Fig. 2 Changes in phase angle and its association with outcome in older adults Statements and declarations The authors have no financial or proprietary interests in any material discussed in this article. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Cruz-Jentoft AJ Landi F Topinkova E Michel JP Understanding sarcopenia as a geriatric syndrome Curr Opin Clin Nutr Metab Care 2010 13 1 7 10.1097/MCO.0b013e328333c1c1 19915458 2. Norman K, Hass U, Pirlich M. Malnutrition in older adults-recent advances and remaining Challenges. Nutrients. 2021;13. DOI:10.3390/nu13082764. 3. Hebuterne X Bermon S Schneider SM Ageing and muscle: the effects of malnutrition, re-nutrition, and physical exercise Curr Opin Clin Nutr Metab Care 2001 4 295 300 10.1097/00075197-200107000-00009 11458024 4. Morley JE Vellas B van Kan GA Anker SD Bauer JM Bernabei R Cesari M Chumlea WC Doehner W Evans J Fried LP Guralnik JM Katz PR Malmstrom TK McCarter RJ Gutierrez Robledo LM Rockwood K von Haehling S Vandewoude MF Walston J Frailty consensus: a call to action J Am Med Dir Assoc 2013 14 392 7 10.1016/j.jamda.2013.03.022 23764209 5. Fried LP Tangen CM Walston J Newman AB Hirsch C Gottdiener J Seeman T Tracy R Kop WJ Burke G McBurnie MA Cardiovascular Health Study Collaborative Research G Frailty in older adults: evidence for a phenotype J Gerontol A Biol Sci Med Sci 2001 56 M146-56 10.1093/gerona/56.3.m146 11253156 6. Graf CE Herrmann FR Genton L Relation of disease with standardized Phase Angle among older patients J Nutr Health Aging 2018 22 601 7 10.1007/s12603-018-1034-4 29717760 7. Haverkort EB Reijven PL Binnekade JM de van der Schueren MA Earthman CP Gouma DJ de Haan RJ Bioelectrical impedance analysis to estimate body composition in surgical and oncological patients: a systematic review Eur J Clin Nutr 2015 69 3 13 10.1038/ejcn.2014.203 25271012 8. Bussolotto M Ceccon A Sergi G Giantin V Beninca P Enzi G Assessment of body composition in elderly: accuracy of bioelectrical impedance analysis Gerontology 1999 45 39 43 10.1159/000022053 9852379 9. Lupoli L Sergi G Coin A Perissinotto E Volpato S Busetto L Inelmen EM Enzi G Body composition in underweight elderly subjects: reliability of bioelectrical impedance analysis Clin Nutr 2004 23 1371 80 10.1016/j.clnu.2004.05.005 15556259 10. Norman K Stobaus N Pirlich M Bosy-Westphal A Bioelectrical phase angle and impedance vector analysis–clinical relevance and applicability of impedance parameters Clin Nutr 2012 31 854 61 10.1016/j.clnu.2012.05.008 22698802 11. Baumgartner RN Chumlea WC Roche AF Bioelectric impedance phase angle and body composition Am J Clin Nutr 1988 48 16 23 10.1093/ajcn/48.1.16 3389323 12. Huemer MT Petrera A Hauck SM Drey M Peters A Thorand B Proteomics of the phase angle: results from the population-based KORA S4 study Clin Nutr 2022 41 1818 26 10.1016/j.clnu.2022.06.038 35834914 13. Jacobs LH van de Kerkhof JJ Mingels AM Passos VL Kleijnen VW Mazairac AH van der Sande FM Wodzig WK Konings CJ Leunissen KM van Dieijen-Visser MP Kooman JP Inflammation, overhydration and cardiac biomarkers in haemodialysis patients: a longitudinal study Nephrol Dial Transplant 2010 25 243 8 10.1093/ndt/gfp417 19692417 14. Mattiello R Amaral MA Mundstock E Ziegelmann PK Reference values for the phase angle of the electrical bioimpedance: systematic review and meta-analysis involving more than 250,000 subjects Clin Nutr 2020 39 1411 7 10.1016/j.clnu.2019.07.004 31400996 15. Kolodziej M Ignasiak Z Ignasiak T Annual changes in appendicular skeletal muscle mass and quality in adults over 50 y of age, assessed using bioelectrical impedance analysis Nutrition 2021 90 111342 10.1016/j.nut.2021.111342 34166898 16. da Silva BR Gonzalez MC Cereda E Prado CM Exploring the potential role of phase angle as a marker of oxidative stress: a narrative review Nutrition 2022 93 111493 10.1016/j.nut.2021.111493 34655952 17. Tomeleri CM Cavaglieri CR de Souza MF Cavalcante EF Antunes M Nabbuco HCG Venturini D Barbosa DS Silva AM Cyrino ES Phase angle is related with inflammatory and oxidative stress biomarkers in older women Exp Gerontol 2018 102 12 8 10.1016/j.exger.2017.11.019 29197561 18. Luo J Mills K le Cessie S Noordam R van Heemst D Ageing, age-related diseases and oxidative stress: what to do next? Ageing Res Rev 2020 57 100982 10.1016/j.arr.2019.100982 31733333 19. Norman K Smoliner C Kilbert A Valentini L Lochs H Pirlich M Disease-related malnutrition but not underweight by BMI is reflected by disturbed electric tissue properties in the bioelectrical impedance vector analysis Br J Nutr 2008 100 590 5 10.1017/S0007114508911545 18234142 20. Norman K Smoliner C Valentini L Lochs H Pirlich M Is bioelectrical impedance vector analysis of value in the elderly with malnutrition and impaired functionality? Nutrition 2007 23 564 9 10.1016/j.nut.2007.05.007 17616343 21. Slee A Birc D Stokoe D Bioelectrical impedance vector analysis, phase-angle assessment and relationship with malnutrition risk in a cohort of frail older hospital patients in the United Kingdom Nutrition 2015 31 132 7 10.1016/j.nut.2014.06.002 25466657 22. Varan HD Bolayir B Kara O Arik G Kizilarslanoglu MC Kilic MK Sumer F Kuyumcu ME Yesil Y Yavuz BB Halil M Cankurtaran M Phase angle assessment by bioelectrical impedance analysis and its predictive value for malnutrition risk in hospitalized geriatric patients Aging Clin Exp Res 2016 28 1121 6 10.1007/s40520-015-0528-8 26786583 23. Kubo Y Noritake K Nakashima D Fujii K Yamada K Relationship between nutritional status and phase angle as a noninvasive method to predict malnutrition by sex in older inpatients Nagoya J Med Sci 2021 83 31 40 10.18999/nagjms.83.1.31 33727735 24. Rinaldi S Gilliland J O’Connor C Chesworth B Madill J Is phase angle an appropriate indicator of malnutrition in different disease states? A systematic review Clin Nutr ESPEN 2019 29 1 14 10.1016/j.clnesp.2018.10.010 30661671 25. Plauth M, Sulz I, Viertel M, Hofer V, Witt M, Raddatz F, Reich M, Hiesmayr M, Bauer P. Phase Angle is a stronger predictor of Hospital Outcome than Subjective Global Assessment-Results from the prospective Dessau Hospital Malnutrition Study. Nutrients. 2022;14. DOI:10.3390/nu14091780. 26. Stobaus N Pirlich M Valentini L Schulzke JD Norman K Determinants of bioelectrical phase angle in disease Br J Nutr 2012 107 1217 20 10.1017/S0007114511004028 22309898 27. Lukaski HC Kyle UG Kondrup J Assessment of adult malnutrition and prognosis with bioelectrical impedance analysis: phase angle and impedance ratio Curr Opin Clin Nutr Metab Care 2017 20 330 9 10.1097/MCO.0000000000000387 28548972 28. Stark G Functional consequences of oxidative membrane damage J Membr Biol 2005 205 1 16 10.1007/s00232-005-0753-8 16245038 29. Norman K Pirlich M Sorensen J Christensen P Kemps M Schutz T Lochs H Kondrup J Bioimpedance vector analysis as a measure of muscle function Clin Nutr 2009 28 78 82 10.1016/j.clnu.2008.11.001 19064305 30. Hirano Y Yamada Y Matsui Y Ota S Arai H Lower limb muscle quality and phase angle contribute to the reduced walking speed among older adults Geriatr Gerontol Int 2022 22 603 9 10.1111/ggi.14423 35781752 31. Tomeleri CM Cavalcante EF Antunes M Nabuco HCG de Souza MF Teixeira DC Gobbo LA Silva AM Cyrino ES Phase Angle is moderately Associated with muscle quality and functional capacity, Independent of Age and Body Composition in Older Women J Geriatr Phys Ther 2019 42 281 6 10.1519/JPT.0000000000000161 29210931 32. Bittencourt DCD Schieferdecker MEM Macedo DS Biesek S Silveira Gomes AR Rabito EI Phase Angle reflects loss of functionality in Older Women J Nutr Health Aging 2020 24 251 4 10.1007/s12603-020-1324-5 32115604 33. Oliveira R, Leao C, Silva AF, Clemente FM, Santamarinha CT, Nobari H, Brito JP. Comparisons between Bioelectrical Impedance variables, functional tests and blood markers based on BMI in older women and their Association with Phase Angle. Int J Environ Res Public Health. 2022;19. DOI:10.3390/ijerph19116851. 34. Norman K Wirth R Neubauer M Eckardt R Stobaus N The bioimpedance phase angle predicts low muscle strength, impaired quality of life, and increased mortality in old patients with cancer J Am Med Dir Assoc 2015 16 173 e17 22 10.1016/j.jamda.2014.10.024 25499428 35. Selberg O Selberg D Norms and correlates of bioimpedance phase angle in healthy human subjects, hospitalized patients, and patients with liver cirrhosis Eur J Appl Physiol 2002 86 509 16 10.1007/s00421-001-0570-4 11944099 36. Di Vincenzo O Marra M Scalfi L Bioelectrical impedance phase angle in sport: a systematic review J Int Soc Sports Nutr 2019 16 49 10.1186/s12970-019-0319-2 31694665 37. Genton L Mareschal J Norman K Karsegard VL Delsoglio M Pichard C Graf C Herrmann FR Association of phase angle and running performance Clin Nutr ESPEN 2020 37 65 8 10.1016/j.clnesp.2020.03.020 32359757 38. Akamatsu Y Kusakabe T Arai H Yamamoto Y Nakao K Ikeue K Ishihara Y Tagami T Yasoda A Ishii K Satoh-Asahara N Phase angle from bioelectrical impedance analysis is a useful indicator of muscle quality J Cachexia Sarcopenia Muscle 2022 13 180 9 10.1002/jcsm.12860 34845859 39. Cruz-Jentoft AJ Bahat G Bauer J Boirie Y Bruyere O Cederholm T Cooper C Landi F Rolland Y Sayer AA Schneider SM Sieber CC Topinkova E Vandewoude M Visser M Zamboni M Writing Group for the european Working Group on Sarcopenia in older P, the Extended Group for E. Sarcopenia: revised european consensus on definition and diagnosis Age Ageing 2019 48 16 31 10.1093/ageing/afy169 30312372 40. Gonzalez MC Barbosa-Silva TG Bielemann RM Gallagher D Heymsfield SB Phase angle and its determinants in healthy subjects: influence of body composition Am J Clin Nutr 2016 103 712 6 10.3945/ajcn.115.116772 26843156 41. Mundstock E Amaral MA Baptista RR Sarria EE Dos Santos RRG Filho AD Rodrigues CAS Forte GC Castro L Padoin AV Stein R Perez LM Ziegelmann PK Mattiello R Association between phase angle from bioelectrical impedance analysis and level of physical activity: systematic review and meta-analysis Clin Nutr 2019 38 1504 10 10.1016/j.clnu.2018.08.031 30224304 42. Di Vincenzo O Marra M Di Gregorio A Pasanisi F Scalfi L Bioelectrical impedance analysis (BIA) -derived phase angle in sarcopenia: a systematic review Clin Nutr 2021 40 3052 61 10.1016/j.clnu.2020.10.048 33183880 43. Walowski CO, Braun W, Maisch MJ, Jensen B, Peine S, Norman K, Muller MJ, Bosy-Westphal A. Reference values for skeletal muscle Mass - Current Concepts and Methodological Considerations. Nutrients. 2020;12. DOI:10.3390/nu12030755. 44. Perez Camargo DA Allende Perez SR Verastegui Aviles E Rivera Franco MM Meneses Garcia A Herrera Gomez A Urbalejo Ceniceros VI Assessment and Impact of Phase Angle and Sarcopenia in Palliative Cancer Patients Nutr Cancer 2017 69 1227 33 10.1080/01635581.2017.1367939 29083245 45. Sipers W de Blois W Schols J van Loon LJC Verdijk LB Sarcopenia is related to Mortality in the acutely hospitalized geriatric patient J Nutr Health Aging 2019 23 128 37 10.1007/s12603-018-1134-1 30697621 46. Landi F Calvani R Cesari M Tosato M Martone AM Bernabei R Onder G Marzetti E Sarcopenia as the Biological substrate of physical Frailty Clin Geriatr Med 2015 31 367 74 10.1016/j.cger.2015.04.005 26195096 47. Wilhelm-Leen ER Hall YN Horwitz RI Chertow GM Phase angle, frailty and mortality in older adults J Gen Intern Med 2014 29 147 54 10.1007/s11606-013-2585-z 24002625 48. Rosas-Carrasco O Ruiz-Valenzuela RE Lopez-Teros MT Phase Angle cut-off points and their Association with Sarcopenia and Frailty in adults of 50–64 years old and older adults in Mexico City Front Med (Lausanne) 2021 8 617126 10.3389/fmed.2021.617126 33791322 49. Tanaka S Ando K Kobayashi K Seki T Hamada T Machino M Ota K Morozumi M Kanbara S Ito S Ishiguro N Hasegawa Y Imagama S Low Bioelectrical Impedance Phase Angle is a significant risk factor for Frailty Biomed Res Int 2019 2019 6283153 10.1155/2019/6283153 31281842 50. Zanforlini BM Trevisan C Bertocco A Piovesan F Dianin M Mazzochin M Alessi A Zoccarato F Manzato E Sergi G Phase angle and metabolic equivalents as predictors of frailty transitions in advanced age Exp Gerontol 2019 122 47 52 10.1016/j.exger.2019.04.016 31028839 51. Matsumoto Y Tada M Yamada Y Mandai K Hidaka N Koike T The bioimpedance phase angle is more useful than sarcopenia as a predictor of falls in patients with rheumatoid arthritis: results from a 2-y prospective cohort study Nutrition 2022 102 111729 10.1016/j.nut.2022.111729 35810573 52. Jung YW Hong N Kim CO Kim HC Youm Y Choi J Rhee Y The diagnostic value of phase angle, an integrative bioelectrical marker, for identifying individuals with dysmobility syndrome: the korean Urban-Rural Elderly study Osteoporos Int 2021 32 939 49 10.1007/s00198-020-05708-2 33128075 53. Uemura K Doi T Tsutsumimoto K Nakakubo S Kim MJ Kurita S Ishii H Shimada H Predictivity of bioimpedance phase angle for incident disability in older adults J Cachexia Sarcopenia Muscle 2020 11 46 54 10.1002/jcsm.12492 31436391 54. Uemura K Yamada M Okamoto H Association of bioimpedance phase angle and prospective falls in older adults Geriatr Gerontol Int 2019 19 503 7 10.1111/ggi.13651 30957354 55. Roman E Poca M Amoros-Figueras G Rosell-Ferrer J Gely C Nieto JC Vidal S Urgell E Ferrero-Gregori A Alvarado-Tapias E Cuyas B Hernandez E Santesmases R Guarner C Escorsell A Soriano G Phase angle by electrical bioimpedance is a predictive factor of hospitalisation, falls and mortality in patients with cirrhosis Sci Rep 2021 11 20415 10.1038/s41598-021-99199-8 34650096 56. Genton L Norman K Spoerri A Pichard C Karsegard VL Herrmann FR Graf CE Bioimpedance-Derived Phase Angle and Mortality among older people Rejuvenation Res 2017 20 118 24 10.1089/rej.2016.1879 27796163 57. Wirth R Volkert D Rosler A Sieber CC Bauer JM Bioelectric impedance phase angle is associated with hospital mortality of geriatric patients Arch Gerontol Geriatr 2010 51 290 4 10.1016/j.archger.2009.12.002 20044156 58. Mullie L Obrand A Bendayan M Trnkus A Ouimet MC Moss E Chen-Tournoux A Rudski LG Afilalo J Phase Angle as a biomarker for Frailty and Postoperative Mortality: the BICS Study J Am Heart Assoc 2018 7 e008721 10.1161/JAHA.118.008721 30371163 59. Cornejo-Pareja I Vegas-Aguilar IM Garcia-Almeida JM Bellido-Guerrero D Talluri A Lukaski H Tinahones FJ Phase angle and standardized phase angle from bioelectrical impedance measurements as a prognostic factor for mortality at 90 days in patients with COVID-19: a longitudinal cohort study Clin Nutr 2021 10.1016/j.clnu.2021.02.017 33642143 60. Ellegard LH Petersen P Ohrn L Bosaeus I Longitudinal changes in phase angle by bioimpedance in intensive care patients differ between survivors and non-survivors Clin Nutr ESPEN 2018 24 170 2 10.1016/j.clnesp.2018.02.001 29576357 61. Campa F Colognesi LA Moro T Paoli A Casolo A Santos L Correia RR Lemes IR Milanez VF Christofaro DD Cyrino ES Gobbo LA Effect of resistance training on bioelectrical phase angle in older adults: a systematic review with Meta-analysis of randomized controlled trials Rev Endocr Metab Disord 2022 10.1007/s11154-022-09747-4 35918569 62. Nunes JP Ribeiro AS Silva AM Schoenfeld BJ Dos Santos L Cunha PM Nascimento MA Tomeleri CM Nabuco HCG Antunes M Cyrino LT Cyrino ES Improvements in Phase Angle are related with muscle Quality Index after Resistance Training in Older Women J Aging Phys Act 2019 27 515 20 10.1123/japa.2018-0259 30507279 63. Tomeleri CM Ribeiro AS Cavaglieri CR Deminice R Schoenfeld BJ Schiavoni D Dos Santos L de Souza MF Antunes M Venturini D Barbosa DS Sardinha LB Cyrino ES Correlations between resistance training-induced changes on phase angle and biochemical markers in older women Scand J Med Sci Sports 2018 28 2173 82 10.1111/sms.13232 29858504 64. Freitas SP Judice PB Hetherington-Rauth M Magalhaes JP Correia IR Lopes JM Strong C Matos D Sardinha LB The impact of 2 weeks of detraining on phase angle, BIVA patterns, and muscle strength in trained older adults Exp Gerontol 2021 144 111175 10.1016/j.exger.2020.111175 33279660 65. Barrea L Muscogiuri G Aprano S Vetrani C de Alteriis G Varcamonti L Verde L Colao A Savastano S Phase angle as an easy diagnostic tool for the nutritionist in the evaluation of inflammatory changes during the active stage of a very low-calorie ketogenic diet Int J Obes (Lond) 2022 46 1591 7 10.1038/s41366-022-01152-w 35614205 66. Beberashvili I Azar A Sinuani I Kadoshi H Shapiro G Feldman L Sandbank J Averbukh Z Longitudinal changes in bioimpedance phase angle reflect inverse changes in serum IL-6 levels in maintenance hemodialysis patients Nutrition 2014 30 297 304 10.1016/j.nut.2013.08.017 24484680	36456777	PMC9715408	NO-CC CODE	2022-12-03 23:20:15	no	Rev Endocr Metab Disord. 2022 Dec 2;:1-9	utf-8	Rev Endocr Metab Disord	2,022	10.1007/s11154-022-09772-3	oa_other
==== Front Aquat Sci Aquat Sci Aquatic Sciences 1015-1621 1420-9055 Springer International Publishing Cham 917 10.1007/s00027-022-00917-9 Research Article Spatial ecology of non-native common carp (Cyprinus carpio) in Lake Ontario with implications for management http://orcid.org/0000-0002-0816-3221 Piczak M. L. [email protected] 1 Brooks J. L. 1 Boston C. 2 Doka S. E. 2 Portiss R. 3 Lapointe N. W. R. 4 Midwood J. D. 2 Cooke S. J. 1 1 grid.34428.39 0000 0004 1936 893X Department of Biology, Institute of Environmental and Interdisciplinary Science, Carleton University, 1125 Colonel By Drive, Ottawa, ON Canada 2 grid.23618.3e 0000 0004 0449 2129 Great Lakes Laboratory for Fisheries and Aquatic Science, Fisheries and Oceans Canada, 867 Lakeshore Road, Burlington, ON Canada 3 grid.451491.c Toronto and Region Conservation Authority, 101 Exchange Avenue, Concord, ON Canada 4 grid.453940.f 0000 0000 9795 461X Canadian Wildlife Federation, 350 Michael Cowpland Dr., Ottawa, ON Canada 2 12 2022 2023 85 1 2031 1 2022 10 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. Common carp, Cyprinus carpio, are a non-native species that established within the Laurentian Great Lakes more than a century ago and are abundant in some locations. Common carp have negatively impacted freshwater ecosystems, including in the Great Lakes, by increasing turbidity and uprooting vegetation through foraging and/or spawning activities. Knowledge of spatial ecology is necessary to effectively manage non-native species and aid in the development of remediation strategies. The aim of this study was to examine the spatial ecology of common carp across multiple spatial scales within Lake Ontario using passive acoustic telemetry. First, Residency Index (RI), as a metric for habitat preference, was calculated for common carp in Toronto Harbour (TH) and Hamilton Harbour (HH). Linear mixed modelling revealed that season, as well as the interaction between season and physical habitat conditions significantly affected RI. Specifically, during spring and summer common carp had significantly higher RI at sites with increased submerged aquatic vegetation, which could be associated with spawning activities. All common carp tagged in HH were resident, compared to half of individuals tagged in TH. Larger individuals tagged in TH were more likely to be absent from the array during summer. Non-resident common carp tagged at TH made extensive movements in spring and summer along the nearshore of Lake Ontario and were detected throughout the entire basin. Knowledge of spawning habitat could inform efforts to exclude common carp from these specific locations. Based on our findings, common carp should be managed at a regional level, as opposed to single sites, owing to their extensive movements. Supplementary Information The online version contains supplementary material available at 10.1007/s00027-022-00917-9. Keywords Movement Management Invasive species Control Fish Spawning http://dx.doi.org/10.13039/501100002790 Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada Great Lakes Action Planissue-copyright-statement© Springer Nature Switzerland AG 2023 ==== Body pmcIntroduction Non-native species are organisms introduced to a novel ecosystem, which can have detrimental economic and environment impacts (NISC 2006). In aquatic environments, non-native species can have direct or indirect biological impacts through predation, competition, hybridization, habitat modification, and transmission of novel pathogens or diseases (Gozlan et al. 2010). Non-native fish species have been introduced for various purposes including for sport, aquaculture, and ornamental trade (Welcomme 1988). Once established, these species can expand their geographical range (Lorenzoni et al. 2010) and rapidly colonize new habitats (Penne and Pierce 2008). Eradication of established populations can be difficult and understanding their ecology is key to guiding effective management strategies (Beatty et al. 2017). The Laurentian Great Lakes of North America have seen numerous introductions of non-native species, which have contributed to the decline and even extirpation of native species (Mandrak and Cudmore 2010). Among the non-native fish species present in the Great Lakes (Mills et al. 1994), one cyprinid species, common carp (Cyprinus carpio), first introduced to the American side of Lake Ontario has raised concerns for over 100 years. Common carp, native to Eurasia, have been ranked as one of the 100 worst non-native species on the planet (Lowe et al. 2004). Introduced throughout the world for aquaculture and recreational fisheries purposes, common carp have become a dominant species in many freshwater ecosystems (Bajer and Sorensen 2010). Common carp use shallow, vegetated wetlands and floodplains for spawning and use littoral habitat throughout the rest of their life cycle (Penne and Pierce 2008). Once established, common carp populations can reach high abundances, and can drastically alter ecosystems by causing increased turbidity and nutrient mobilization, decreased density of macrophytes, and ultimately lower abundance of macroinvertebrates and fishes (Miller and Crowl 2006; Matsuzaki et al. 2009). These negative impacts are most commonly seen in small, shallow lakes or specific coastal embayments, and do not necessarily occur in all ecosystems where common carp are introduced. Life-history characteristics of common carp enable this species to expand rapidly and attain high biomasses (Britton et al. 2011). These strategies include relatively early maturation (compared to many native fishes), extended adult longevity (up to 64 years; Koch 2014), long breeding seasons in temperate areas (between water temperatures of 17 and 28 °C; Panek 1987), and repeated spawning events in a single year (Smith and Walker 2004). The Laurentian Great Lakes have long suffered from negative anthropogenic effects, mainly stemming from industry, agriculture, and urbanization (Jones et al. 2006). These deleterious impacts are often concentrated in coastal wetlands (Steedman and Regier 1987). Throughout the Laurentian Great Lakes, coastal wetlands provide spawning, foraging, refugia, or nursery habitat for the majority of native fishes (Jude and Pappas 1992). Despite their ecological importance, over 50% of wetlands within the Great Lakes have been lost (Uzarski et al. 2017), with many of the remaining wetlands impaired or degraded (Chow-Fraser 2006). Additionally, common carp use these wetlands (Lougheed and Chow-Fraser 2001), further contributing to their degradation (Weber and Brown 2009). In light of this impairment, restoration efforts throughout the Great Lakes basin have often included the remediation or creation of aquatic habitat (Hartig et al. 2020); however, the presence of common carp can potentially hinder remediation through removal of aquatic vegetation or increasing turbidity (Lougheed and Chow-Fraser 2001; Miller and Crowl 2006). Management of non-native species requires an integrated and holistic approach rooted in baseline scientific knowledge (Britton et al. 2011). Biotelemetry can provide important information on the spatial ecology and movement patterns of non-native species (Lennox et al. 2016). Telemetry arrays are broadly accepted as an effective means of studying the spatial ecology of fish including their seasonal habitat preferences and movements (Cooke et al. 2013; Hussey et al. 2015; Krueger et al. 2018). While there have been studies examining common carp spatial ecology within the Great Lakes (see Landsman et al. 2011; Rous et al. 2017; Brooks et al. 2017; Kraus et al. 2018), there remains a need to further understanding of spatial extent, timing, and frequency of movements to inform management actions. Previous telemetry studies revealed that common carp access shallow areas to spawn during spring, are capable of complex, extensive movements in summer (Jones and Stuart 2007; Banet et al. 2021), and form aggregations at deep overwintering sites (Bajer et al. 2011; Watkinson et al. 2021). Further, Kim and Mandrak (2016) found that common carp dispersed from Lake Ontario to Lake Erie through a system of locks. Preliminary evidence suggests that common carp move large distances in Lake Ontario (Midwood et al. 2019) in a similar manner as other invaded systems (Jones and Stuart 2007; Banet et al. 2021); however, these results have not been fully explored or confirmed. Currently, it is not known how mobile common carp are within Lake Ontario (Midwood et al. 2019), or the extent to which individuals undertake partial migration (Banet et al. 2021), thereby hindering coordinated management measures throughout the basin. Additionally, some observations indicate that common carp within Toronto Harbour (TH) may be part of a larger metapopulation (Midwood et al. 2019). Though there are common carp exclusion structures in some locations (e.g., the Fishway within Cootes Paradise, Hamilton Harbour; Boston et al. 2016), it remains unclear where common carp may be forming seasonal aggregations for spawning in other areas within Lake Ontario. Moreover, effective control measures such as the placement of exclusion structures, will require the identification of areas accessed by common carp during summer for spawning, thereby minimizing recruitment and controlling populations. This study examined seasonal habitat preference, and movements of common carp within Lake Ontario, in the Laurentian Great Lakes. Specifically, we estimated a residency index (RI) in two harbours in western Lake Ontario: TH and Hamilton Harbour (HH). Using this RI, we then examined how habitat preference varied with season and fish sizes. We also investigated the effect of season and fish size on presence and absence within the TH and HH arrays using general linear modelling. Lastly, we documented broad-scale movements in Lake Ontario to examine evidence for movements between TH and HH and to highlight other areas accessed by common carp within Lake Ontario. Methods Study sites and telemetry array Lake Ontario, the most easterly of the Laurentian Great Lakes, has been subjected to anthropogenic activity for over 200 years, particularly in the densely populated western portion, which is home to both the cities of Toronto and Hamilton (43.631–79.369 and 43.285–79.843, respectively; Fig. 1). Due to historic and ongoing anthropogenic disturbance, habitat impairment and loss, TH and HH were identified as Areas of Concern (AOCs) in 1985 and are the focus of considerable remediation efforts (Hartig et al. 2020). Efforts to remediate fish habitat in both harbours target physical habitat enhancement and creation, with the goal of supporting the recovery of native freshwater fishes and other aquatic organisms (Barnes et al. 2020). To assess the efficacy of these efforts, extensive biotelemetry arrays have been deployed in TH and HH, and a variety of fish species including common carp, have been implanted with acoustic transmitters and tracked (Midwood et al. 2019; Brooks et al. 2019). Fig. 1 The receivers deployed as part of the Great Lakes Acoustic Telemetry Observation System (GLATOS) within Lake Ontario are shown with deployment year (A). Toronto Harbour (TH; B) and Hamilton Harbour (HH; C) are located in the north central and western portion of Lake Ontario, respectively. Acoustic receiver groupings across the TH and HH arrays are denoted with different symbols (see Supplemental 1 for additional receiver group information) Toronto, with a population of over five million people, has experienced widespread loss of littoral and wetland habitat along its waterfront (over 400 ha; Whillans 1982) mainly owing to infilling to support urbanization and the expansion of industry (Barnes et al. 2020). A large system (18 km2) of embayments, TH has four zones; Inner Harbour, Toronto Islands, Outer Harbour, and Tommy Thompson Park (TTP; Fig. 1). TTP is a man-made peninsula consisting of four embayments (A through D; Fig. 1) and a confined disposal facility (CDF) comprised of three cells (1 through 3) that was created and modified to enhance habitat for aquatic species (Barnes et al. 2020). Dredged contaminated materials have been deposited in the CDF cells; however, this has ceased in Cells 1 and 2 (1985 and 1997, respectively), but is ongoing in Cell 3. Subsequently, Cells 1 and 2 were capped and restored with techniques designed to increase shoreline complexity, encourage the establishment of aquatic vegetation, increase structural habitat complexity, and passively exclude common carp with an exclusion structure (Barnes et al. 2020). The majority of the telemetry array within TH was installed in spring 2011, with fluctuations thereafter in coverage due to the loss of receivers or expansion of coverage into new areas of interest (see Supplemental 1 for receiver details). Key movement corridors, as well as various habitat types were strategically instrumented with VR2W 69 kHz acoustic receivers (Innovasea, Bedford, Nova Scotia; Fig. 1). Receivers were combined into 37 groups based on habitat consistency/proximity (Midwood et al. 2019), as well as range-testing results (conservative estimate of 350 m; see Veilleux 2014). Detections in TH were available from fall 2010 to summer 2020, and this entire period was included in the analysis. Hamilton Harbour, a 21 km2 protected embayment, is located at the far western end of Lake Ontario. The south shore of the harbour is dominated by industry (mostly steel or concrete walls), whereas the north and east portions are composed of mostly artificial hard and soft shorelines, with more natural shorelines to the west (Gardner Costa et al. 2020). Cootes Paradise Marsh, situated at the western end of the Harbour, is a large (250 ha) degraded coastal wetland. A physical exclusion structure, the Fishway (operational since 1997), connecting the marsh to the main harbour was designed to exclude common carp. The HH telemetry array has been operational since late summer 2015, with receivers deployed throughout the area covering various habitat types and movement corridors. Similar to TH, the array comprises VR2W 69 kHz receivers (27 initially, expanding to 51; Fig. 1) that have been assigned to one of 15 groups (Supplemental 1). Range testing completed in HH showed considerable variability, particularly when the system was stratified in summer. Detection ranges were approximately 300 m during summer and increased to over 400 m during isothermal conditions (Wells et al. 2021). Detection data were available from summer 2015 to summer 2020. In addition to the TH and HH acoustic telemetry arrays, data from a larger network of receivers deployed in Lake Ontario as part of the Great Lakes Acoustic Telemetry Observation System (GLATOS; Fig. 1) were used. Data sharing through GLATOS network allows tracking of tagged fish tagged throughout much of Lake Ontario. These additional receivers were deployed at various times such that receiver coverage in Lake Ontario was variable with limited coverage prior to 2014. Fish capture and tagging Common carp (n = 102) were captured from both TH and HH (n = 81 and n = 21, respectively; see Supplemental 2). All common carp were collected using boat electrofishing (both models SR-18EH, 7.0 A, 340 and 250 V for TH and HH, respectively; Smith-Root, Inc., Vancouver, WA) between 2010 and 2018. After capture, common carp were placed in live wells with ambient lake water and transported to shore for surgery (TH) or surgery was conducted on the vessel (both TH and HH). Fish were immobilized for surgery using either a Portable Electroanesthesia System (Smith-Root; Rous et al. 2015) or electric fish handling gloves (HH; Smith-Root; Reid et al. 2019). Common carp were put in a trough with ambient lake water passed over the gills to aid respiration. All surgical tools and acoustic transmitters were disinfected with an iodine solution and rinsed. An incision (< 15 mm) was made with a scalpel and the transmitter (see Supplement 3 for transmitter details) was inserted into the coelom. Incisions were closed with two or three interrupted sutures. Fish size (total length) was measured, and fish were returned to a live well with circulating lake water. Common carp were released at their point of capture after ensuring full recovery. Fish handling and surgical procedures were approved and followed a Canadian Council on Animal Care protocol administered by Carleton University (Certificate CU 110,723). Seasonal delineation All analyses were completed in R Studio (version 1.1.456; R Core Team; 2021). Water temperature transitions were used to identify seasonal periods (based on Larocque et al. 2020). For TH, temperature-profile data were collected from a chain of temperature loggers deployed nearby in Lake Ontario (Ajax, Ontario; 43.461–78.584). We delineated seasons by taking an average of the temperature loggers: spring started when water temperatures first exceeded 5 °C, until they surpassed 15 °C, which was then designated as summer. Fall occurred when temperatures consistently decreased below 15 °C until falling below 5 °C, which was designated as winter (Table 1). For HH, a chain of temperature loggers (average across the loggers) deployed in the center of the harbour was used to delineate seasons (Table 1) and permitted a different approach to defining seasons. Spring started when temperatures first warmed above 5 °C, shifting to summer when a clear thermocline was established. Fall started when the harbour system “turned over” and lasted until temperatures were consistently below 5 °C. Common carp spawn between 17 and 28 °C (Panek 1987); therefore, spawning likely occurs in late spring or early summer, with staging starting in spring. Due to the COVID-19 pandemic, data were not downloaded from either logger sites in 2020; therefore, some seasons were defined using the average of previous years. Additionally, some monitoring periods did not cover the complete season (Table 1). Table 1 Season delineation based on water temperatures (based on Larocque et al. 2020) in (A) Toronto Harbour (TH) and (B) Hamilton Harbour (HH) Year Spring Summer Fall Winter TH 2010 NA Sept 131 Oct 3 Nov 12 2011 April 22 June 7 Oct 16 Nov 30 2012 April 13 May 21 Sept 22 Nov 15 2013 April 20 June 12 Oct 18 Nov 19 2014 April 26 June 19 Oct 5 Nov 24 2015 April 21 June 17 Oct 14 Nov 26 2016 April 15 May 28 Oct 23 Nov 21 2017 April 9 June 12 Oct 25 Nov 18 2018 April 20 June 11 Oct 16 Nov 27 2019 April 1 June 27 Oct 15 Nov 8 2020 April 16 June 10 Sept 221 NA HH 2015 NA Aug 121 Oct 1 Nov 21 2016 April 30 June 1 Sept 26 Nov 17 2017 April 17 June 18 Oct 14 Nov 25 2018 May 1 June 1 Oct 3 Nov 17 2019 April 12 June 13 Oct 17 Nov 20 2020 April 22 June 8 July 21 NA 1Indicates start or stop of study period, and grey shading indicates seasons where averages were taken from all previous years due to missing data Telemetry data collection and preparation Data from each telemetry array (TH and HH) were downloaded approximately every six months, once in spring and fall, annually. Receivers were either treated as individual stations or grouped based on their proximity (i.e., overlapping fields of detections) or habitat type (Fig. 1; Supplemental 1), both of which herein are referred to as groups. Erroneous detections were removed if they met criteria for false-positive detections (single occurrences with > 3600 s between successive detections; Pincock et al. 2012). The dataset was also filtered to remove fish that died or expelled their transmitters, which was presumed to have occurred when consistent depth profiles and locations were indicated for an extended period (i.e., stationary horizontal detections at a given station and/or consistent depths; see Klinard and Matley 2020). Additionally, fish that were detected for fewer than 14 days total were removed from the dataset to eliminate those that died following surgery or had malfunctioning tags. To decrease temporal autocorrelation, a reduced dataset was created by randomly selecting one detection per fish once per hour over the course of their period of activity. We also created two different working datasets per harbour: RI (based only on that harbour’s array) and total detections in the reduced dataset (including detections outside of the arrays). As noted, prior to 2014 there was limited receiver coverage in western Lake Ontario outside of TH and HH. For example, if a fish left TH in 2013, it would not have been detected in Lake Ontario due to a lack of receiver coverage, thereby leaving extended gaps in detections. We manually identified and removed these extended absences (greater than seven days) from the harbours from the residency datasets, as well as confirmed absences (i.e., individuals detected outside either the HH or TH arrays). Seasonal residency and habitat conditions Residency indices are often calculated as the number of days an individual fish was detected at group divided by the total number of days the fish was detected anywhere within the acoustic array. Rather than using raw detections, RI reduces the potential bias of a large number of detections at a given station, generated by a small number of individuals (Kessel et al. 2016). However, in the present study we estimated a modified seasonal RI, which was calculated as the time spent at a given receiver group, divided by the total length of a given season, using the residency function in the GLATOS package (Holbrook et al. 2016). We used this modified RI to avoid potential bias from common carp that had no detections for extended periods of time. Further, due to limited receiver coverage outside both TH and HH during the earlier years of the study, we manually identified periods where common carp were not detected for greater than seven days to determine if individuals remained within or departed either array. For example, if an individual fish was detected continuously throughout a given season, the sum of the modified RI values would be 1, compared to an individual that was not detected continuously (e.g., departed the harbour), the modified RI would sum to less than 1. Mean modified RI was calculated for each season-year combination and for each season across all years of study, for each harbour. Zeroes were added for any receiver group or season combination when an individual fish was not detected during that time period but was known to still be active (i.e., detected elsewhere or detected during a later time period); these zeroes were included in the calculation of the mean seasonal RI. To estimate habitat conditions within TH and HH, a 350 m circular buffer was created around each receiver group and these buffers were clipped to not include land (see Supplemental 4). The buffer size was used to reflect an approximate receiver detection range based on range testing in each harbour. Percent cover of submerged aquatic vegetation (SAV) was estimated with a model that used the depth, slope, and mean exposure to produce a static estimate of SAV (Doolittle et al. 2010). This model has been determined to be 80% effective at predicting the presence of SAV within HH (Gardner Costa et al. 2020) and was subsequently applied to both HH and TH to produce estimates of mean cover for each receiver group (87.1% accuracy for TH; see Midwood et al. 2020). We acknowledge that percent cover of SAV is highly dynamic both within and among years given that macrophytes grow during spring and senesce in fall; however, we elected to apply a static model due to limited information to support implementation of a more seasonally dynamic approach. Habitat preference was analyzed by fitting a linear-mixed-effect (LME; package lme4) model with modified RI as the response variable (which was log transformed), as described in Midwood et al. (2018). Each sample in this analysis represented the modified RI (time spent at a given receiver group, divided the total length of a given season) of an individual fish for each season for 1 year. Explanatory variables included season (categorical), total length (mm; continuous), as well as percent cover of SAV (continuous) estimated from a 350 m buffer around receiver groups for both TH and HH (see Supplemental 4). Interaction terms included season by SAV. To account for repeated measures, animal transmitter ID was included as a random effect (categorical). Further, diagnostics were performed for validation and included plotting the residuals (with a Q-Q plot for normality), residuals versus explanatory variables (for independence), and the residuals against fitted values (to verify homogeneity) to visually inspect model fit (Zuur et al. 2009). Spatial autocorrelation was assessed by plotting residuals at receiver coordinates. All procedures were conducted in R statistical environment using the “ggplot2” (Wickham 2016) and “lmer4” (Bates et al. 2015) packages for data visualization and modelling, respectively. Post-hoc Tukey HSD tests were conducted as necessary on categorical variables. Presence/absence within study areas To understand drivers of forays (i.e., individual fish that departed the array and subsequently returned) and dispersal (i.e., fish that departed the array and did not return), we documented and described movements beyond each harbour, including associated details (date, time, season, as well as origin and destination when detected). We produced a presence/absence dataset, whereby common carp were absent if they departed either array within a given season and present if they did not leave the array. Absences were denoted when common carp were detected at points of exit within each array (e.g., the curtain or Western Gap in TH or at the Lake Ontario station in HH) and were absent for greater than seven days or detected outside the array (i.e., elsewhere in Lake Ontario). Common carp were designated as non-resident if they undertook forays or resident if they did not depart the array. Each sample in this analysis represented the presence/absences of an individual fish for one season for one given year. The effects of season (categorical), tagging date (Julian date; continuous) and total length (mm; continuous) on presence/absence from each harbour were tested using generalized linear mixed models (GLMM; package lme4) with a binomial distribution. Interaction terms included season by total length. Diagnostics were performed on the GLMM as per the LME, and animal transmitter ID was included as a random effect (categorical). Diagnostics were performed for validation and included plotting the normalized residuals and the residuals against fitted values. Spatial autocorrelation was assessed by plotting residuals at receiver coordinates. Large-scale movements We documented and described movements undertaken by common carp throughout Lake Ontario. Specifically, we mapped detections throughout Lake Ontario on receivers maintained by members of the GLATOS network to examine spatial extent of movements for each season, as well as monthly from May to September (Supplemental 5), which was the time period that captured the bulk of the movements outside the TH and HH arrays. Specifically, for each season we mapped the total number of detections and total number of individual common carp at receivers across all years of study to identify other areas accessed by common carp within Lake Ontario. Results Across both TH and HH, there were a total of 6,698,378 detections from 102 common carp. Due to death or transmitter malfunctions, 13 common carp were removed from further analyses (n = 8 from TH and n = 5 from HH; Supplemental 3) with subsequent analysis focused on 89 common carp (n = 73 from TH and n = 16 from HH). Within TH, there was consistently high modified RI in TTP (Cells 2 and 3) and the Western receiver group in HH (Fig. 2). During spring, common carp were found in Cells 1 and 2, as well as Embayment D within TH and in Cootes Paradise and the Western receiver group in HH (Fig. 2). Through the LME, we determined that common carp RI across both harbours was significantly influenced by the interaction term between season and SAV (p = 0.002; Table 2). During spring there was a positive correlation between RI and SAV (Fig. 3). During summer and winter, SAV did not influence RI, and there was a negative relationship in fall (Fig. 3). Fig. 2 Mean modified Residency Index of common carp (Cyprinus carpio; n = 89) by season in Toronto Harbour (A; n = 73; 2010–2020) and Hamilton Harbour (B; n = 16; 2015–2020). Residency Index was calculated as time spent at a given receiver group, divided by the total length of a given season. Receivers were either treated as individual stations or grouped based on their proximity or habitat type (see Supplemental 1 for additional details) Table 2 The importance of individual terms and interactions for the linear mixed effects model of common carp residency index (n = 1193) Model term Chi square df P value Total Fish Length (mm) 2.87 1 0.08 Season 42.07 3 < 0.001* SAV (%) 2.76 1 0.09 Season x SAV (%) 14.11 3 0.002* Percent cover of submerged aquatic vegetation is SAV and animal transmitter ID (categorical; n = 89) was included as a random effect Fig. 3 Impact of percent cover of submerged aquatic vegetation (SAV) on common carp (Cyprinus carpio) residency across seasons in Toronto Harbour and Hamilton Harbour as determined by the linear mixed effects model. There was a strong positive relationship between the Residency Index and SAV during the spring and summer No common carp tagged in HH were detected outside of the HH array, whereas more than half of the common carp tagged in TH exhibited forays or dispersals outside of the array. Because no common carp tagged in HH left the HH array, they were not included in the presence/absence analysis. The GLMM revealed that both season and total fish length significantly influenced presence/absence of common carp within the TH array (p = 0.04 and p < 0.001, respectively; Table 3), while tagging date did not (p = 0.8; Table 3). Specifically, absences outside the array increased with body size (Fig. 4), with absences were more common in spring and summer and least common during fall. Table 3 The relative importance of terms in the generalized linear mixed model with a binomial distribution for common carp that were present or absent from the TH array (n = 843) Model term Chi square df P value Total Fish Length (mm) 9.86 1 0.04* Season 8.45 3 < 0.001* Season x Total Fish Length 1.74 3 0.63 Tagging Date 0.06 1 0.80 Animal transmitter ID was included as a random effect (categorical; n = 73) Fig. 4 Impact of season and total length on common carp (Cyprinus carpio) presence/absence on the Toronto Harbour array (as no common carp tagged in Hamilton Harbour departed the array) as determined by the generalized linear mixed model with a binomial distribution. Of the 73 common carp tagged in TH, 39 were non-resident in that they moved outside the array for at least seven days, while 34 carp were resident and only ever detected within the array. For TH common carp, the mean size for non-resident fish was only slightly larger than the resident carp (665 ± 83 mm and 630 ± 103 mm SD, respectively). Of the non-resident common carp that departed TH, 18 dispersed from the TH array and did not return, and 21 undertook forays (i.e., ultimately coming back to the TH array). Most of these movements outside the TH array occurred during summer as evident by higher total number of detections and number of individual common carp on receivers in Lake Ontario during this season (Fig. 5). Fourteen of the common carp tagged in TH were detected elsewhere in Lake Ontario (via the GLATOS network) and six of these individuals were detected within HH (Fig. 5). There were also extended absences outside the TH array (i.e., periods with no detections) prior to the expansion of the GLATOS network in Lake Ontario and the location of common carp during these periods is unknown. Fig. 5 Locations of detections throughout Lake Ontario (outside of the Toronto Harbour array) of common carp (Cyprinus carpio); seasons are shown to highlight differences across spatial extent and distances moved. Colour gradient represents numbers of detections, with size of circle depicting the number of individuals detected at a receiver. Receivers within of the GLATOS array are also displayed (small black points) to show the full extent of coverage as of 2020. More spatially extensive movements by some common carp are evident in the spring and summer compared to fall and winter (Color figure online) The majority of forays outside the TH array occurred during summer (n = 59), followed by spring (n = 27), winter (n = 4), and fall (n = 3). Movement during spring was not as extensive as summer; however, individual common carp were still detected in Credit River, as well as throughout the middle portion of Lake Ontario (albeit infrequently) near Presqu’ile Provincial Park and Braddock Bay, outside of Rochester, USAduring this season (Fig. 5). Additionally, there were detections throughout the Niagara River, Credit River, and Bronte Creek in spring (Fig. 5) for a total of six receiver groups. In summer, common carp were detected at 30 different receiver groups (outside the TH array) throughout Lake Ontario, followed by five in fall, and one in winter (Fig. 5). Multiple common carp undertook extensive movements in summer, with some detections recorded at the eastern end of Lake Ontario in July (over 300 km from the tagging site assuming fish followed the nearshore (i.e., not direct Euclidean distance; Fig. 5). During summer, common carp were also detected throughout the western portion of Lake Ontario, ranging from Duffins Creek to the Niagara River, and as far east on the south shore as Braddock Bay (Fig. 5). During fall, common carp were detected at Duffins Creek and Niagara River receiver groups. Finally, the only location common carp were detected outside the HH and TH arrays during winter was in the Niagara River (Fig. 5). At a finer temporal resolution, there was a gradual increase of distance moved from TH for common carp movements from May to a maximum in July, where individuals were detected in the far eastern portion of Lake Ontario (Supplemental 5). There were also extensive movements undertaken in August and September, where common carp were detected as far away as Braddock Bay (Supplemental 5). Discussion We examined the spatial ecology, seasonal habitat preference, and movements of common carp within Lake Ontario. We found that common carp habitat preference was highly variable across individuals and seasons, but was influenced by season and SAV. Specifically, we determined that common carp were associated with SAV coverage in spring and summer, though many fish also left TH during summer. We also found that larger common carp were more likely to leave TH during summer, with some individuals exhibiting extensive movements. Further, only common carp tagged in TH departed the array (i.e., forays and dispersals), while individuals tagged in HH did not leave the area. The results from our study highlight potential spawning areas within TH and HH, as well as areas within Lake Ontario accessed by common carp during spring and summer. Our study highlighted potential areas within TH and HH where common carp could be spawning during spring and summer. Specifically, within TH and HH, we found that in spring and summer common carp were seeking sites with increased SAV and it has been shown within TH that SAV presence is dictated by fetch and depth (Midwood et al. 2020), suggesting that common carp seek shallow, protected areas that support SAV growth. Indeed, it has been well documented that common carp migrate in spring to shallow, littoral areas (floodplains, shallow lakes, ponds or wetlands) with SAV (Lougheed et al. 2004; Hennen and Brown 2014; Sorensen and Bajer 2020; Banet et al. 2021). Further, Banet et al. (2021) determined that common carp exhibited homing during spawning migrations, with consistent use of sites across multiple years. Although we did not explicitly study homing or site fidelity, we did find consistent preference of specific sites within TH (Cells 1 and 2) and HH (Cootes Paradise Marsh) across years (10 and 5 years, respectively), future studies should look to confirm if these behaviours are representative of homing during spawning. There were increased absences of common carp from the TH array during spring and summer, which coincided with some extensive movements throughout Lake Ontario. This is consistent with a previous study where large-scale movements, likely for the purposes of spawning, were documented during spring in the Murray-Darling Basin in Australia (up to 650 km; Jones and Stuart 2009). It is possible that movements (albeit infrequently) off the TH array during spring and summer in Lake Ontario could be indicative of common carp movements to spawning areas including Hamilton Harbour, Bronte Creek, Credit River, near Presqu’ile Provincial Park, and southern Lake Ontario (near Hamlin Beach State Park, USA). Other individuals may have spawned in their resident harbours, given that most common carp in TH and all individuals in HH remained resident during spring. Management implications of these extensive movements undertaken by common carp during spring include the identification of additional areas where physical exclusion structures could be placed to minimize access to spawning habitats. During summer, common carp were detected over 300 km away from TH, throughout the entire basin of Lake Ontario. In addition to spawning activities, it has been suggested that these extensive summer movements may be associated with environmental factors, such as foraging or prey resources, and/or climatic factors, including temperature or wave action (Hennen and Brown 2014). Banet et al. (2021) found that in the Rice Creek Watershed, Minnesota, common carp movements were extensive throughout June, July, and August, similar to our findings (see Supplemental 5). Further, it has been hypothesized that increased movements during summer by common carp could be related to strategies to avoid eating their own eggs or to forage in different areas than spawning (Banet et al. 2021; Watkinson et al. 2021). In addition to increased absences during spring and summer, we found that larger common carp were more likely to be absent from TH. While the relationship between total fish length of common carp and distance moved has not been previously examined, it may be that larger individuals require greater resources and thus travel further to acquire them, have more energetic resources to support long-distance movement, or more apt to move because they face lower risk of predation. Based on these movements undertaken by common carp throughout Lake Ontario across international borders, management of this non-native species should be carried out at a large spatial scale across jurisdictional boundaries. Our study revealed that individual common carp within Lake Ontario have different movement strategies, with only portions of the population foraying and/or dispersing. Sample sizes differed between tagging sites (73 and 16 for TH and HH, respectively) and other (non-tagged) common carp in HH may undertake movements outside the array. Different movement strategies across TH and HH could be due to spatial and temporal variation in resources (Dingle and Drake 2007), potentially with additional resources and habitat throughout Lake Ontario being accessed by TH individuals. It is also possible that there is higher resource or other habitat availability within HH, as indicated by movements by some TH fish to access the harbour and further supported by a lack of common carp leaving HH. Previous studies have found other common carp populations exhibiting various movement strategies (Stuart and Jones 2006; Chizinski et al. 2016; Banet et al. 2021). Further, through these individualistic movements, common carp can locate periodically shallow habitats, inundated floodplains and forests to seek spawning sites (Jones and Stuart, 2006). We found that there were movements outside the TH array during both spring and summer, which could be indicative of breeding (to access spawning sites) and non-breeding (for foraging; Banet et al. 2021). At this point, the purpose of movements throughout Lake Ontario during spring and summer cannot be explicitly determined, although coincident with spawning and expanded summer foraging, future works could aim to confirm these activities through surveys of recruitment, such as egg and fry collection. Individualistic movements along with other adaptive and flexible life history traits, such as early maturation, extended breeding season, and adult longevity (Jones and Stuart 2009) contribute to the success of common carp (Weber and Brown 2009), as well as challenges with their management. Understanding the spatial ecology of common carp is critical to producing effective management strategies. We identified sites within TH and HH, as well as Lake Ontario, that common carp accessed during spring and early summer, which could be indicative of spawning or spawning staging areas. Currently within HH, adult common carp have decreased access Cootes Paradise Marsh due to a physical exclusion structure (the Fishway) that was installed in 1997 to exclude common carp from shallow spawning habitat (Lougheed et al. 2004; Boston et al. 2016). Within 5 years of installation, there was a 50% decline in common carp biomass within HH, suggesting these efforts were successful (Boston et al. 2016). Other studies have similarly documented success from the use of common carp exclusion structures to not only reduce biomass (e.g., Tempero et al. 2019), but also improve water quality and establishment of aquatic vegetation (Lougheed and Chow-Fraser 2001; Lougheed et al. 2004; Knopik and Newman 2018). Additional exclusion barriers could be constructed in areas across Lake Ontario that common carp accessed during the spring and summer, in efforts to decrease access and improve aquatic habitat conditions. Evidence of extensive movements beyond the TH common carp population during spring and summer suggests that some individuals may be spawning at additional sites throughout Lake Ontario (as observed in Smith and Walker 2004). Our emerging understanding of the scale of common carp movements during spring and summer not only highlights other areas throughout the western portion of Lake Ontario where management efforts may be required, but also indicates that population control measures need to be coordinated throughout Lake Ontario to ensure individuals blocked from accessing one spawning area are restricted from alternate spawning habitat within their movement range. Physical exclusion structures, such as the Cootes Paradise Fishway could be a potential management strategy in other areas where common carp aggregate. In addition to supporting population management, findings from the present work can also inform current restrictions on consumption related to bioaccumulation of contaminants. For example, common carp, among other species, within Great Lakes AOCs Lakes have been shown to accumulate polychlorinated biphenyls (PCBs; Brooks et al. 2017; Visha et al. 2021), which has led to restrictions related to consumption and continued impairment of the AOCs. Determining the extent of movement for fishes, such as common carp, has been identified as a key element in assessing whether actions taken within Great Lakes AOCs aimed at reducing contaminant levels will be successful (Bhavsar et al. 2018) because resident fishes are likely to benefit from local reductions of contaminants while migratory or wide-ranging fishes may still be exposed in other locations. Our finding of limited movement by common carp in HH compared to those in TH suggests that potential sources of PCBs from HH fish likely come from within the harbour, whereas sources for more mobile fish in TH are more difficult to determine. Future studies should attempt to confirm sources of PCB contamination within common carp in Great Lakes AOCs as well as explore other potential sources of contaminants that mobile individuals may be exposed to. To confirm habitat preference, consistent and thorough receiver coverage of all habitats in TH and HH is needed; however, in some shallow areas there was limited coverage because of the need to minimize receiver damage from boat collisions or ice scour and limited detection range because of dense SAV. Therefore, despite efforts to achieve total coverage with receivers and consistent detections, individual fish can “disappear” (i.e., out of line-of-sight of a receiver or in dense SAV; Midwood et al. 2019), therefore impacting RI analyses. We aimed to alleviate this limitation with large sample sizes across multiple years of study (ten years in TH and five in HH). Additionally, we also attempted to account for these periods where fish could “disappear” (i.e., imperfect detections) by manually identifying times when common carp were not detected to determine if they departed the array or not. Next, our estimations of SAV were coarse, in that we assigned one, static mean for a given group; however, these conditions are highly variable over time (i.e., the cycle of SAV growth both within and among years; Tang et al. 2021). Therefore, the estimate of SAV throughout fall and winter is likely an overestimate. Finally, for multiple years of this study there were considerable gaps in receiver coverage outside of TH and HH. Due to these gaps, our coverage of common carp movements outside of our arrays was limited and it is highly likely they are using more parts of Lake Ontario than were documented here. The Lake Ontario array was expanded in 2021 to provide near complete coverage of the main basin of the lake as well as numerous nearshore areas (https://glatos.glos.us/). This expanded infrastructure will support future studies of movements of common carp and other fishes throughout Lake Ontario, which will further inform more targeted and effective management. Conclusion Knowledge of the spatial ecology of non-native species is necessary for effective control and management. With the use of passive acoustic telemetry, we have identified areas within both TH and HH where common carp could be spawning and that sites accessed during spring and summer had increased SAV. We also determined that common carp tagged in Toronto Harbour undertake individualistic, extensive movements throughout the entire Lake Ontario basin, mostly during summer and spring. These extensive movements coincide with increased absence outside of the TH array, which was positively related to the total fish length of the individual. Based on these extensive movements, population control measures for common carp need to be coordinated throughout Lake Ontario to ensure individuals blocked from accessing one spawning area (i.e., with exclusion structures) are restricted from alternate spawning habitat within their movement range. Our study contributes to the identification of places within both harbours and Lake Ontario (i.e., Toronto Islands and the western portion of HH, Bronte Creek and the Credit and Niagara Rivers) where control measures could be implemented, including passive management measures such as exclusion structures that may decrease access during spring and summer. Ideally, minimizing access of common carp to their spawning habitat will decrease recruitment and biomass, thereby bringing balance back to Lake Ontario’s struggling aquatic ecosystems. Supplementary Information Below is the link to the electronic supplementary material. Supplementary material 1 (DOCX 569 kb) Acknowledgements We would like to thank staff members of the Toronto and Region Conservation Authority for their support in the field and planning including Don Little, Brian Graham, Brynn Coey, Gord MacPherson, Adam Weir, Pete Shuttleworth, Bradley Bloemendal, Meg St. John, Kaylin Barnes, Matthew Fraschetti, and Ross Davidson (among others). Thank you to lab members of the Fish Ecology and Conservation Physiology Laboratory at Carleton University including Maxime Veilleux and Andrew Rous. We also extend thanks to those working for the Great Lakes Acoustic Telemetry Observation System, especially Nancy Nate. A special thank you to Chris Holbrook for assisting with coding and R. Thank you to the field crew of Fisheries and Oceans Canada including David Reddick, Erin Budgell, Sarah Larocque, Dallas Linley, and numerous summer students who assisted with array maintenance and fish tagging. Thank you to Laud Matos, Thomas Sciscone, and Kristin O’Connor for their support in planning and implementing these telemetry projects. We sincerely appreciate the help with R and modelling from Dylan Melmer and Robert Lennox. Steven Cooke is supported by the Natural Sciences and Engineering Research Council of Canada and Genome Canada. Funding for both TRCA and Carleton University came from the Great Lakes Sustainability Fund. Funding to Susan Doka and Jon Midwood for these projects came from the Great Lakes Action Plan administered by Environment and Climate Change Canada, as well as support from Fisheries and Oceans Canada through the Freshwater Habitat Initiative. Author contributions MLP wrote the paper and analyzed data, with guidance from NWRL, JDM and SJC. MLP, SED, RP, NWRL, JDM and SJC conceived and designed the study. MLP, JLB, CB and NWRL performed fieldwork, with direction from RP. Equipment was contributed by SED, RP, JDM and SJC. Editing and proofreading was done by all authors. SED, RP, JDM and SJC provided research funding. Data availability The data and source code that support the findings of this study are available from the corresponding author upon reasonable request. Declarations Conflict of interest The authors declare that there are no conflicts of interest. Consent for publication All the authors consent to publication. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Bajer PG Sorensen PW Recruitment and abundance of an invasive fish, the common carp, is driven by its propensity to invade and reproduce in basins that experience winter-time hypoxia in interconnected lakes Biol Invasions 2010 12 1101 1112 10.1007/s10530-009-9528-y Bajer PG Chizinski CJ Sorensen PW Using the Judas technique to locate and remove wintertime aggregations of invasive common carp: USING JUDAS FISH TO REMOVE CARP AGGREGATIONS Fish Manag Ecol 2011 18 497 505 10.1111/j.1365-2400.2011.00805.x Banet NV Fieberg J Sorensen PW Migration, homing and spatial ecology of common carp in interconnected lakes Ecol Freshw Fish eff 2021 10.1111/eff.12622 Barnes K, Cartwright L, Portiss R et al (2020) Evaluating the Toronto Waterfront aquatic habitat restoration strategy. Toronto and Region Conservation Authority. https://trcaca.s3.ca-central-1.amazonaws.com/app/uploads/2021/01/07074413/TWAHRS-assessment-FINAL-technical-document-Nov-2020.pdf Bates D Mächler M Bolker B Walker S Fitting linear mixed-effects models using lme4 J Stat Softw 2015 67 1 48 10.18637/jss.v067.i01 Beatty SJ Allen MG Whitty JM First evidence of spawning migration by goldfish (Carassius auratus); implications for control of a globally invasive species Ecol Freshw Fish 2017 26 444 455 10.1111/eff.12288 Bhavsar SP Drouillard KG Tang WK Assessing fish consumption Beneficial Use Impairment at Great Lakes Areas of concern: Toronto case study Aquat Ecosyst Health Manag 2018 21 318 330 10.1080/14634988.2018.1498272 Boston CM Randall RG Hoyle JA The fish community of Hamilton Harbour, Lake Ontario: Status, stressors, and remediation over 25 years Aquat Ecosyst Health Manag 2016 19 206 218 10.1080/14634988.2015.1106290 Britton JR Gozlan RE Copp GH Managing non-native fish in the environment Fish Fish 2011 12 256 274 10.1111/j.1467-2979.2010.00390.x Brooks JL Boston C Doka S Use of Fish Telemetry in Rehabilitation Planning, Management, and monitoring in Areas of concern in the Laurentian Great Lakes Environ Manage 2017 60 1139 1154 10.1007/s00267-017-0937-x 28939998 Brooks JL Midwood JD Gutowsky LFG Spatial ecology of reintroduced walleye (Sander vitreus) in Hamilton Harbour of Lake Ontario J Great Lakes Res 2019 45 167 175 10.1016/j.jglr.2018.11.011 Chizinski CJ Bajer PG Headrick ME Sorensen PW Different migratory strategies of invasive common carp and native Northern Pike in the american Midwest Suggest an opportunity for Selective Management Strategies North Am J Fish Manag 2016 36 769 779 10.1080/02755947.2016.1167141 Chow-Fraser P (2006) Quality Index (WQI) to assess Effects of Basin-wide land-use alteration. on Coastal Marshes of the Laurentian Great Lakes Cooke SJ Midwood JD Thiem JD Tracking animals in freshwater with electronic tags: past, present and future Anim Biotelem 2013 1 5 10.1186/2050-3385-1-5 Dingle H Drake VA What is migration? BioScience 2007 57 113 121 10.1641/B570206 Doolittle AG, Bakelaar CN, Doka SE (2010) Spatial framework for storage and analyses of fish habitat data in Great Lakes’ areas of concern: Hamilton Harbour Geodatabase Case Study. Canadian Technical Report of Fisheries and Aquatic Sciences 2879 Gardner Costa J, Rémillard CYL, Doolittle A, Doka SE (2020) Hamilton Harbour shoreline survey — 2006. Canadian Technical Report of Fisheries and Aquatic Sciences, 3381, vii, 37 pp Gozlan RE Britton JR Cowx I Copp GH Current knowledge on non-native freshwater fish introductions J Fish Biol 2010 76 751 786 10.1111/j.1095-8649.2010.02566.x Hartig JH Krantzberg G Alsip P Thirty-five years of restoring great lakes areas of concern: gradual progress, hopeful future J Great Lakes Res 2020 46 429 442 10.1016/j.jglr.2020.04.004 Hennen MJ Brown ML Movement and spatial distribution of common carp in a South Dakota Glacial Lake System: implications for management and removal North Am J Fish Manag 2014 34 1270 1281 10.1080/02755947.2014.959674 Holbrook C, Hayden T, Binder T (2016) glatos: a package for the great lakes acoustic telemetry observation system Hussey NE Kessel ST Aarestrup K Aquatic animal telemetry: a panoramic window into the underwater world Science 2015 348 1255642 1255642 10.1126/science.1255642 26068859 Jones MJ Stuart IG Movements and habitat use of common carp (Cyprinus carpio) and Murray cod (Maccullochella peelii peelii) juveniles in a large lowland australian river Ecol Freshw Fish 2007 16 210 220 10.1111/j.1600-0633.2006.00213.x Jones MJ Stuart IG Lateral movement of common carp (Cyprinus carpio L.) in a large lowland river and floodplain Ecol Freshw Fish 2009 18 72 82 10.1111/j.1600-0633.2008.00324.x Jones ML, Shuter BJ, Zhao Y, Stockwell JD (2006) Forecasting effects of climate change on great lakes fisheries: models that link habitat supply to population dynamics can help. 63:12 Jude DJ Pappas J Fish utilization of Great Lakes Coastal Wetlands J Great Lakes Res 1992 18 651 672 10.1016/S0380-1330(92)71328-8 Kessel ST Hussey NE Crawford RE Distinct patterns of Arctic cod (Boreogadus saida) presence and absence in a shallow high Arctic embayment, revealed across open-water and ice-covered periods through acoustic telemetry Polar Biol 2016 39 1057 1068 10.1007/s00300-015-1723-y Kim J Mandrak NE Assessing the potential movement of invasive fishes through the Welland Canal J Great Lakes Res 2016 42 1102 1108 10.1016/j.jglr.2016.07.009 Klinard NV Matley JK Living until proven dead: addressing mortality in acoustic telemetry research Rev Fish Biol Fisheries 2020 30 485 499 10.1007/s11160-020-09613-z Knopik JM Newman RM Transplanting aquatic macrophytes to restore the littoral community of a eutrophic lake after the removal of common carp Lake Reserv Manag 2018 34 365 375 10.1080/10402381.2018.1477885 Koch JD (2014) Source-sink population structure of invasive common carp in a model midwestern watershed: empirical evidence and notes on management. University of Minnesota Kraus RT Holbrook CM Vandergoot CS Evaluation of acoustic telemetry grids for determining aquatic animal movement and survival Methods Ecol Evol 2018 9 1489 1502 10.1111/2041-210X.12996 Krueger CC Holbrook CM Binder TR Acoustic telemetry observation systems: challenges encountered and overcome in the Laurentian Great Lakes Can J Fish Aquat Sci 2018 75 1755 1763 10.1139/cjfas-2017-0406 Landsman SJ Nguyen VM Gutowsky LFG Fish movement and migration studies in the Laurentian Great Lakes: Research trends and knowledge gaps J Great Lakes Res 2011 37 365 379 10.1016/j.jglr.2011.03.003 Larocque S, Boston CM, Midwood JD (2020) Seasonal daily depth use patterns of acoustically tagged freshwater fishes informs nearshore fish community sampling protocols. Canadian Technical Report of Fisheries and Aquatic Sciences, 3409 Lennox RJ Blouin-Demers G Rous AM Cooke SJ Tracking invasive animals with electronic tags to assess risks and develop management strategies Biol Invasions 2016 18 1219 1233 10.1007/s10530-016-1071-z Lorenzoni M Ghetti L Pedicillo G Carosi A Analysis of the biological features of the goldfish Carassius auratus auratus in Lake Trasimeno (Umbria, Italy) with a view to drawing up plans for population control Folia Zool 2010 59 142 156 10.25225/fozo.v59.i2.a9.2010 Lougheed VL, Chow-Fraser P (2001) Spatial variability in the response of lower trophic levels after carp exclusion from a freshwater marsh Lougheed VL Theÿsmeÿer T Smith T Chow-Fraser P Carp exclusion, Food-web interactions, and the restoration of Cootes Paradise Marsh J Great Lakes Res 2004 30 44 57 10.1016/S0380-1330(04)70328-7 Lowe S Browne M Boudjelas S De Poorter M 100 of the World’s worst invasive alien species: a selection from the global invasive species database Encyclopedia of biological invasions 2004 University of California Press 715 716 Mandrak NE Cudmore B The fall of native fishes and the rise of non-native fishes in the Great Lakes Basin Aquat Ecosyst Health Manag 2010 13 255 268 10.1080/14634988.2010.507150 Matsuzaki SS Usio N Takamura N Washitani I Contrasting impacts of invasive engineers on freshwater ecosystems: an experiment and meta-analysis Oecologia 2009 158 673 686 10.1007/s00442-008-1180-1 18941787 Midwood JD Gutowsky LFG Hlevca B Tracking bowfin with acoustic telemetry: insight into the ecology of a living fossil Ecol Freshw Fish 2018 27 225 236 10.1111/eff.12340 Midwood JD, Rous AM, Doka SE, Cooke SJ (2019) Acoustic telemetry in Toronto Harbour: assessing residency, habitat selection, and within-harbour movements of fishes over a five-ear period. Can. Tech. Rep. Fish. Aquat. Sci. 3331: xx + 174 p. https://publications.gc.ca/collections/collection_2019/mpo-dfo/Fs97-6-3331-eng.pdf Midwood JD Tang RWK Doka SE Gardner Costa JM Comparison of approaches for modelling submerged aquatic vegetation in the Toronto and Region Area of concern J Great Lakes Res S0380133020302008 2020 10.1016/j.jglr.2020.08.019 Miller SA Crowl TA Effects of common carp (Cyprinus carpio) on macrophytes and invertebrate communities in a shallow lake Freshw Biol 2006 51 85 94 10.1111/j.1365-2427.2005.01477.x Mills EL Leach JH Carlton JT Secor CL Exotic species and the Integrity of the Great Lakes Bioscience 1994 44 666 676 10.2307/1312510 Panek FM Cooper EL Biology and ecology of carp Carp in North America 1987 Bethesda, Maryland American Fisheries Society 1 16 Penne CR Pierce CL Seasonal distribution, aggregation, and Habitat Selection of Common Carp in Clear Lake Iowa Trans Am Fisheries Soc 2008 137 1050 1062 10.1577/T07-112.1 Pincock D, Welch D, McKinley S, Jackson G (2012) Acoustic Telemetry for Studying Migration Movements of Small Fish in Rivers and the Ocean— Current Capabilities and Future Possibilities. In: Tagging, Telemetry, and Marking Measures for Monitoring Fish Populations. p 15 R Core Team (2021) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.https://www.R-project.org/ Reid CH Vandergoot CS Midwood JD On the electroimmobilization of fishes for Research and Practice: Opportunities, Challenges, and Research needs Fisheries 2019 44 576 585 10.1002/fsh.10307 Rous AM Forrest A McKittrick EH Orientation and position of Fish affects Recovery Time from Electrosedation Trans Am Fish Soc 2015 144 820 828 10.1080/00028487.2015.1042555 Rous AM Midwood JD Gutowsky LFG Telemetry-determined Habitat Use informs Multi-Species Habitat Management in an urban Harbour Environ Manage 2017 59 118 128 10.1007/s00267-016-0775-2 27744518 Smith BB Walker KF Spawning dynamics of common carp in the River Murray, South Australia, shown by macroscopic and histological staging of gonads J Fish Biol 2004 64 336 354 10.1111/j.0022-1112.2004.00293.x Sorensen PW Bajer PG Case Studies demonstrate that common carp can be sustainably reduced by exploiting Source-Sink Dynamics in midwestern lakes Fishes 2020 5 36 10.3390/fishes5040036 Steedman RJ Regier HA Ecosystem Science for the Great Lakes: perspectives on degradaiive and rehabilitative transformations Can J Fish Aquat Sci 1987 44 s95 s103 10.1139/f87-313 Stuart IG Jones MJ Movement of common carp, Cyprinus carpio, in a regulated lowland australian river: implications for management Fisheries Manage 2006 13 213 219 10.1111/j.1365-2400.2006.00495.x Tang RWK Doka SE Midwood JD Gardner Costa JM Development and spatial application of a submerged aquatic vegetation model for Cootes Paradise Marsh, Ontario, Canada Aquat Sci 2021 83 9 10.1007/s00027-020-00760-w Tempero GW Hicks BJ Ling N Fish community responses to invasive fish removal and installation of an exclusion barrier at Lake Ohinewai, Waikato N Z J Mar Freshwat Res 2019 53 397 415 10.1080/00288330.2019.1579101 Uzarski DG Brady VJ Cooper MJ Standardized Measures of Coastal Wetland Condition: implementation at a Laurentian Great Lakes Basin-Wide scale Wetlands 2017 37 15 32 10.1007/s13157-016-0835-7 Veilleux M Spatial Ecology of Fish in Toronto Harbour in response to aquatic Habitat Enhancement 2014 Carleton University Master of Science Visha A Lau A Yang C A probabilistic assessment of the impairment status of areas of concern in the Laurentian Great Lakes: how far are we from delisting the Hamilton Harbour, Lake Ontario Canada? Ecol Inf 2021 62 101271 10.1016/j.ecoinf.2021.101271 Watkinson DA Charles C Enders EC Spatial ecology of common carp (Cyprinus carpio) in Lake Winnipeg and its potential for management actions J Great Lakes Res 2021 47 583 591 10.1016/j.jglr.2021.03.004 Weber MJ Brown ML Effects of Common Carp on aquatic Ecosystems 80 years after “Carp as a Dominant”: ecological insights for Fisheries Management Rev Fish Sci 2009 17 524 537 10.1080/10641260903189243 Wells MG Li J Flood B Speed of sound gradients due to summer thermal stratification can reduce the detection range of acoustic fish tags: results from a field study in Hamilton Harbour, Ontario Can J Fish Aquat Sci 2021 78 269 285 10.1139/cjfas-2020-0078 Whillans TH Changes in Marsh Area along the Canadian Shore of Lake Ontario J Great Lakes Res 1982 8 570 577 10.1016/S0380-1330(82)71994-X Wickham H ggplot2: elegant graphics for data analysis 2016 New York Springer Zuur A Ieno EN Walker N Saveliev AA Smith GM Mixed effects models and extensions in ecology with R 2009 New York Springer	36474645	PMC9715410	NO-CC CODE	2022-12-03 23:20:15	no	Aquat Sci. 2023 Dec 2; 85(1):20	utf-8	Aquat Sci	2,022	10.1007/s00027-022-00917-9	oa_other
==== Front Dialect Anthropol Dialect Anthropol Dialectical Anthropology 0304-4092 1573-0786 Springer Netherlands Dordrecht 9678 10.1007/s10624-022-09678-1 Forum Commentary “Youth speaking truth to power”: intersectional decolonial activism in Namibia http://orcid.org/0000-0002-8079-9152 Becker Heike [email protected] grid.8974.2 0000 0001 2156 8226 Department of Anthropology, University of the Western Cape, Bellville, South Africa 2 12 2022 114 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. This article portrays a recent movement towards intersectional activism in urban Namibia. Since 2020, young Namibian activists have come together in campaigns to decolonize public space through removing colonial monuments and renaming streets. These have been linked to enduring structural violence and issues of gender and sexuality, especially queer and women’s reproductive rights politics, which have been expressly framed as perpetuated by coloniality. I argue that the Namibian protests amount to new political forms of intersectional decoloniality that challenge the notion of decolonial activism as identity politics. The Namibian case demonstrates that decolonial movements may not only emphatically not be steeped in essentialist politics but also that activists may oppose an identity-based politics which postcolonial ruling elites have promoted. I show that, for the Namibian movements’ ideology and practice, a fully intersectional approach has become central. They consciously juxtapose colonial memory with a living vision for the future to confront and situate colonial and apartheid history. Young Namibian activists challenge the intersectional inequalities and injustices, which, they argue, postcolonial Namibia inherited from its colonial–apartheid past: class inequality, racism, sexism, homophobia, and gender-based violence. Keywords Activism Intersectional Decolonial Namibia Windhoek ==== Body pmcOn the 16th of June 2020, a hundred mostly young Windhoekers engaged in a multi-sited protest action. Their handwritten placards expressed their allegiance to the global Black Lives Matter movement, which had arisen after the murder of George Floyd on 25 May 2020. They protested against racism, gender-based violence and Namibian police brutality.1 They also rallied for the decolonisation of the public space calling for the removal of a statue of German colonial officer, Curt von François, in front of the Windhoek Municipality building. In 1890 the German colonial officer Curt von François had designed and laid the foundations of the first European building in Windhoek (still called, the Alte Feste—Old fortress).2 In 1965, the (all-white) Windhoek City Council decided to honour him as the supposed town founder with a 2.4-m high bronze statue, which was modeled and cast by a South African sculptor, Hennie Potgieter. The protesters gathered around the monument, some climbing onto its top. Their quest for a reception by city officials to deliver their demands was unsuccessful; they left, leaving behind their posters. Some among those read, “Rape culture must fall”, “Legalize Abortion”, “Police Brutality must end”, and “Black Lives Matter”. Others addressed historical issues. Some recalled the 1893 German colonial attack on the Witbooi Nama settlement at Hornkranz, southwest of Windhoek. By remembering this brutal act of colonial violence, the demonstrators denounced “white supremacy—an insult to those who water our freedom”, as stated one eloquent poster, alluding to Namibia’s national anthem’s lyrics. Calling on the Hornkranz onslaught was directly connected to the protest’s key target since the same von François, the removal of whose statue they were demanding, was its leader. Well-known activist and journalist, Keith Vries, reads out a petition in which the protesters voiced their grievances. It demanded a review of colonial legislation and public sector policies still promoting racist ideals. The protestors asked for accountability and action from both justice and policing systems in all cases of police and military brutality and for a broad and holistic public re-education on gender-based violence, rape culture and LGBTQ + people’s rights. The petition also called for the removal countrywide of colonial monuments and the change of street names named for German colonial era despots. It concluded: “We understand all these issues to be intersectional and interconnected.” A decolonial intersectional activism The above vignette illustrates a recent movement towards intersectional activism in urban Namibia. Since 2020, young Namibian activists have come together in campaigns to decolonize public space through removing colonial monuments and renaming streets. These have been linked to enduring structural violence and issues of gender and sexuality, especially queer and women’s reproductive rights politics, which have been expressly framed as perpetuated by coloniality (Becker 2020). I argue that such protests amount to new political forms of intersectional decoloniality that challenge the notion of decolonial activism as identity politics. Recently, some authors have claimed that decoloniality is inevitably caught up in identity politics (e.g., Táíwò 2022; Hull 2021).3 I argue that while the broad-based movements, which identity with decolonial aspirations have, in certain moments, espoused problematic essentialist tendencies, the politics of decolonial movements needs to be examined more cautiously. Firstly, it appears problematic to construct a purportedly uniform theory of decoloniality, while drawing on the writings of only a few authors. Hull’s verdict on decoloniality theory as essentialist and “metaphysical”, for instance, is based only on his reading of works by Walter Mignolo and Sabelo Ndlovu-Gatsheni. He entirely ignores other influential authors, such as Nelson Maldonado Torres. Secondly, one must ask those that have argued that the turn to decoloniality has been a problematic turn to identity politics why they ignore an entirely different genealogy of decolonial theory from thinkers such as Frantz Fanon and Walter Rodney, who have been widely received in the movements. The most important questions however relate to practice. Regarding the South African Fallist movements, Kelly Gillespie and Leigh-Ann Naidoo (2019) have argued that, instead of issuing a verdict on students’ turn to blackness as a problematic turn to identity politics, their invocations’ substance should be considered carefully. In this article, I present the case of Namibian movements that avowedly “lean(s) towards decoloniality” (Mushaandja 2021: 192; Fn 2). I contend that the Namibian case demonstrates even more clearly than the South African movmements Gillespie and Naidoo (2019) discuss that decolonial movements may not only emphatically not be steeped in essentialist politics but, instead, that activists may oppose an identity-based politics which postcolonial ruling elites have promoted through memory politics and discourses of policultural nationalism and citizenship (see Becker 2015).4 I show that, for the Namibian movements’ ideology and practice, a fully intersectional approach has become central. Namibian activists consciously juxtapose colonial memory with a living vision for the future to confront and situate colonial and apartheid history. They imagine a postcolonial Namibia that takes up the challenges of full decolonization. They continuously challenge the intersectional inequalities and injustices, which, they argue, postcolonial Namibia inherited from its colonial–apartheid past: class inequality, racism, sexism, homophobia and gender-based violence. Most recently, they have focused on two sets of apartheid era colonial laws that are still in force in Namibia, despite having been abolished in South Africa soon after that country’s transition to democracy.5 These are the extremely narrowly defined grounds allowing legal abortion and discrimination on grounds of sexual orientation. The latter is epitomized in Namibia’s Sodomy Law and in administrative practices discriminating against partners in same sex marriages and children born into such families.6 Activists have redefined their struggles for legal reform in these arenas expressly as part of the efforts to decolonize public space. As self-defined queer and decolonial activist Omar van Reenen commented on Twitter (19 March 2022): “The Divorce Law, the Sodomy law, the Abortion & Sterilization Act: all laws inherited from apartheid South Africa, laws disproportionately affecting women and LGBTQ + persons. All laws not written—or voted on—by democratically elected representatives of the new Namibian people”. Van Reenen thus exemplified the understanding that laws imposed by apartheid South Africa as Namibia’s second colonial ruler until 1990, are a legacy of colonialism, as well as apartheid. While activists conceptualise their struggle as decolonial, they equally emphasise its intersectionality. In an interview recorded by Nicola Brandt, artist and activist Hildegard Titus articulated thoughts about entanglements in space and time activists work with as they attempt to decolonize public space in the built-up environment, in living memory and in everyday democracy.By protesting in public we are drawing attention not just to the single grievance, but to the wider structures and how they connect us as people. For example, although one of the demands was the removal of the Curt von François statue, we were protesting interrelated things. We were protesting gender-based violence (GBV), racial oppression—we were protesting police brutality. It is all of these intersections (Titus 2021: 178). Leading activist Nashilongweshipwe Mushaandja (2021: 196–197) who, on his Twitter account, describes himself as an “undisciplined cultural worker”, embraced the poetics of the intersectional decolonial struggle:Our struggles do not exist in isolation Colonial events such as the Hornkrantz Massacre of 1893… Have everything to do with Namibian Military brutality Colonial and the apartheid projects were also deeply gendered And here we are… Fighting ghosts of our past as Our Fathers of Nation Inherited and preserved these imperial systems … Hence, here we are fighting the past with decolonial futures … Contemporary Black youth speaking truth to power With iconographies of colonialists Towering over them In these examples, two well-known Namibian activists pronounce what they see as the inextricable connections between ostensibly separate issues of contemporary Namibian society: persisting coloniality in public spaces through monuments and street names, state violence in the form of police and military brutality, and the postcolonial Namibian ruling elites’ sexist, patriarchal and homophobic inclinations. During public protest action, Windhoek’s decolonial activists have repeatedly temporarily occupied colonial remnant spaces in the city. These performative actions are best described as embodying forms of political aesthetics, which were previously unknown in Namibia, where protestors consciously put their bodies on display—they sat and lay down to occupy spaces; in some instances the performative embraced what some of the activists referred to as “queering” public space, for instance with bold hip-hop moves. Even though those were rather short-lived demonstrations, the activists conceived them as occupations because of their public acts of taking up—occupying space with the aim of disrupting “business as usual”. Chronology of postcolonial performance and remembrance When protesters climbed on top of the von François memorial in June 2020, it was the first time young Namibians had come out in a public demonstration for the eradication of colonial symbols. It was not the first-ever intervention against traces of colonialism in Windhoek; previous actions had however taken place in the dark of the night, surprising passers-by only the next morning.7 Earlier actions had, in fleeting moments of counter-monumental performance, targeted the city’s (then) most notorious colonial edifice, the Reiterdenkmal, described by Steinmetz & Hell (2006: 177) as the “most aggressive colonial symbol in all of Namibia.”8 Interventions subverted the monument’s colonial claims before the infamous monument was removed in 2009 from its prominent public position. One subversive initiative, in July 2008, saw white wooden crosses planted around the statue, bearing place names and expressions in Otjiherero, the main language spoken by victims of the 1904–8 German colonial genocide. Three months later, passers-by were greeted by the Rider holding up a small flag in post-independence Namibia’s national colours (Kössler 2015: 156). These were thought-provoking initiatives intended to subvert and decolonise public space. Every time protestors re-signified the Rider statue, controversial local media discussions flared up (Niezen 2018: 559). Though less militant than South Africa’s anticolonial RhodesMustFall activism, these Namibian initiatives came from within civil society. They were credited to activists connected to various Ovaherero and Nama victim-descendant pressure groups, which have demanded justice for the communities that suffered most during the colonial war and genocide. Bidding ‘A Curt Farewell’ The 2020 protest against the von François memorial pointed out the significance of that monument (and others) as a painful site of remembrance and memorialisation in several ways that differed from earlier postcolonial Namibian memory politics. The contemporary activists regard their approach as decolonial and intersectional. This makes it distinct from the overnight actions of ethnic-identified genocide reparations activists and from the hegemonic state-centred commemorative practices which celebrate the master narrative of the Namibian postcolonial state: “SWAPO brought us liberation through the barrel of a gun” (Becker 2011, 2018).9 Firstly, it was the first public demonstration demanding eradication of colonial symbols in public spaces. Secondly, it was connected to a sustained campaign that built on new forms of online activism. Inspired by the global Black Lives Matter movement against racism, Windhoek activist and artist, Hildegard Titus, started an online petition entitled “A Curt Farewell”, which garnered over 1,600 signatures by the day of the protest three weeks later. Considering that Windhoek’s population is less than half a million, and that, in 2020, online activism was an entirely new endeavour in Namibian civil society politics; this was quite remarkable. The petition, like the protest that followed, demanded removal of the statue celebrating the purported city founder and its replacement with a statue of Windhoek’s Jonker Afrikaner, the Nama leader who first established a settlement in the area of today’s Windhoek around 1840, and should thus be acknowledged as the city’s real founder. It reads verbatim (excerpts):Continuing to keep Curt von François on his pedestal at the intersection of Sam Nujoma Drive and Independence Avenue is a painful erasure of the city's history and that of its rightful founder, Jonker Afrikaner. This colonial monument continues to feed the incorrect narrative that “this land was empty” until he "discovered" it. It is now time that [the city] … ceases honouring colonial faces. Curt von François was responsible for the building of the Alte Feste, a military fort meant to protect the interests of the German colonial regime, and that is where his statue belongs. He should be confined within the walls that he built, next to the other statue of a bygone and violent era—the Reiterdenkmal—to contemplate their violent colonial legacies until the end of time. Windhoek Mayor Fransina Kahungu eventually accepted the petition in July 2020. However, she and the city council chose not to respond to it. Instead, she said, in September 2020, that the city council did not want to remove the statue although residents of Windhoek had indicated their desire to do so (The Namibian 24 September 2020). Later in 2020, however, a new mayor was elected. Job Amupanda, an academic and activist who had cut his teeth with youth protests a few years earlier, signalled support for the statue’s removal. So did his successor, Sade Gawanas, who took over from Amupanda on 1 December 2021. Unlike Windhoek’s previous post-independence mayors, Amupanda and Gawanas are not members of Namibia’s ruling party, Swapo,10 but belong to a new generation of politicians with roots in civil society activism.11 In mid-2022, discussions considered a new heritage policy for the city. On 27 October 2022 the Windhoek City Council finally voted to remove the statue, and four weeks later, on 23 November 2022, it was removed to the loud cheers of the assembled activists and their allies.12 Thirdly, the activism around the von François statue also reflects an exploration of ‘another’ postcolonial Namibian society, which can be best described with reference to Saidiya Hartman’s (2019: 228) idea of waywardness “as an ongoing exploration of what might be”. Through their temporary occupation of the memorial, the activists embarked on a transgressive act that challenged Namibian society, which, more than thirty years after achieving independence, has remained marked by deep social conservatism. “Full decolonisation”, as the activists have declared on many occasions, challenges conservative Christian perspectives on issues such as gender and sexuality. More about this in a moment. Leading self-declared decolonial activists and artists have on several occasions also emphasised their explicit stance against the ethnicised identity politics which has characterised much of Namibian memory politics, on the one hand, implicitly, on the part of the SWAPO government, which has paid primary attention to the experience of the northern regions, and on the other hand, the Ovaherero and Nama genocide reparations movements (cf. Kössler 2015). Although most decolonial activists are not ethnically Ovaherero or Nama, they have campaigned for reparations by Germany for the colonial genocide of 1904–1908. As my interlocutors have repeatedly told me, justice for the genocide is not just an issue for those identified as descendants of the victims. Instead it should be seen as a Panafricanist and, ultimately, as a concern of global humanism. Finally, activists have frequently emphasised the intersectionality of their decolonial practices. As I showed in my opening vignette, those demonstrating for the von François statue’s removal did not restrict their demands to decolonizing public space; they connected their protest about the coloniality of the public space with concerns about gender-based and state violence, particularly police brutality in the city’s impoverished former black townships. As a leading woman activist told me in August 2020, they considered the police brutality to be yet another expression of coloniality because it related to both race and class. She emphasised that the strict lockdown measures during the early weeks of the pandemic had been enforced brutally only in the townships whereas she had never been stopped while out jogging, together with her white boyfriend, in their affluent (still largely white) neighbourhood. The published statements regarding the intersectionality of the decolonial struggle by leading activists Nashilongweshipe Mushaandja and Hildegard Titus, cited earlier, also demonstrate that they did not regard contemporary Namibian struggles as separate from each other. Instead, activists have pointed out that, in their view, their struggle for the “full decolonisation of Namibia”, as they have often phrased it, integrates various concerns, ranging from decolonisation of public space, through to matters of state violence, and particularly gender and sexuality. The intersectional conceptualisation of decolonial activism in urban Namibia is reminiscent of Maldonado-Torres’s (2016: 10) elaboration of decolonial efforts. Maldonado-Torres presents a perspective on decolonial practice as transcending identity politics by “breaking hierarchies of difference that dehumanize subjects and communities and that destroy nature, and to the production of counter-discourses, counter-knowledges, counter-creative acts and counter-practices that seek to dismantle coloniality and to open up multiple other forms of being in the world.” Maldonado-Torres (2016: 1) identifies the targets of contemporary decolonial movements as the “predominant racist, sexist, homo- and transphobic conservative, liberal and neoliberal politics of today”. In the remainder of this article, I show how Namibian activists have incorporated struggles against the sexist and homophobic politics of the postcolonial Namibian state into their campaigns for the decolonization of public space. Decolonizing public space: performance and counter-memorials Windhoek’s intersectional decolonial activists have protested against colonial monuments; they have also begun to inscribe their politics into the built environment with counter-memorials. In December 2021, activists painted a “Rainbow Sidewalk” in front of the old Namibia Breweries building, which had over recent years housed a ‘queer friendly’ meeting place. Leading queer activist Omar van Reenen tweeted: “Today, we commemorated Namibia’s first LGBTQ + Historic Landmark. …, it is a statement of solidarity against discrimination”. A plaque explains the rainbow flag and the struggles to end discrimination on grounds of sexual orientation. The Rainbow Sidewalk celebrates recent activism. It also provides a penetrating critique of colonial monuments, indeed any monument as concept and idea that fixes memory.13 Instead, it seeks to “confront or disrupt established meanings and tropes” (Mayat and Hart 2021: 203). Even where they materialised in more permanent artistic form, these actions, such as the collective painting of the Rainbow Sidewalk, have been performative. The colours of the rainbow on a public road in central Windhoek demonstrate the presence of a queer community, and thus confront Namibian society where apartheid era sodomy laws remain in force. As Annie Coombes (2004) has argued, monuments are animated and re-animated only through performance. Performance, especially site-specific artistic performance, entered the remembrance of colonialism and of the 1904–1908 genocide a few years earlier through the arts rather than political demonstrations. In 2016, a multimedia dance enactment, The Mourning, directed by choreographer and dance lecturer Trixie Munyama, was performed in the Alte Feste courtyard. The fortress was chosen as the production’s site because of its proximity to the Reiterdenkmal’s previous location on the site of a 1904–8 colonial genocide era concentration camp and now occupied by the Independence Memorial Museum. The Mourning and its sequel, The Mourning Citizen, performed in 2019, were powerful interventions into memories of historical violence, performed at the site of what was once the epicentre of white, male coloniality. Nicola Brandt (2020) emphasised the site-specific-ness of these performances as significant. Being specific to the site of the concentration camp, the performances were profound gestures to honour and acknowledge the dead, to conjure up the site’s traumatic legacies and to bring about healing. Employment of performative activism is one of the decolonial activists’ significant innovations. The following section shows how these have become salient in a succession of recent protests that focused on sexism and homophobia. A new generation of youth activists Despite the restrictions of recurrent COVID lockdowns, in mid-July 2020, protesters again took to Windhoek’s streets. This time they demanded legalization of abortion. The action was organised by a newly formed alliance, Voices for Choices and Rights Coalition (VCRC). By then VCRC had collected 60,000 signatures (in a national population of only 2.5 million) calling for the right to safe abortion and abolition of Namibia’s Abortion and Sterilisation Act of 1975, a legacy of South African colonial presence. In October 2020, another movement galvanized an unprecedented number of young people to reclaim the streets, marching, dancing and unleashing incredible creative energy with their performances. Hundreds of activists, students, working youth, and artists took to Windhoek’s and other towns’ streets to protest against gender-based violence and femicide. The protests started after a young woman’s body was found in the port city of Walvis Bay. Demonstrators blocked busy intersections in downtown Windhoek. As a leading activist in these protests pointed out, in a reflection on protest, performance, publicness and praxis, their unprecedently radical practice and strategy was “embodied through disruptive politics of public life” (Mushaandja 2021: 193). The young protesters were quickly dubbed “Ama2000”, the “people of the twenty-first century”. Mostly in their early twenties or teens, they amazed even the movements’ most seasoned trailblazers. Nashilongweshipwe Mushaandja, who is in his early thirties, commented during a recent conversation with the author that the new generation’s energy and audacity in challenging the “old guard” of Namibian politicians was surprising, even for his generation who, while growing up, had acquiesced much more readily to incomplete decolonisation and authoritarian structures. Triggered by the scourges of femicide and gender-based violence, women and young people made clear that they were tired of living in a violent society. Their rallying cry was #OnsIsMoeg (Afrikaans for “We are tired”), along with #ShutItAllDownNamibia (Becker 2020). The second hashtag expresses their aim to disrupt business-as-usual in a situation of crisis. Protesters marched on various ministries and demanded the resignation of Minister of Gender Equality, Poverty Eradication and Social Welfare, Doreen Sioka. Some carried posters reading, ‘Jou Poes Doreen’ (literally ‘your cunt, Doreen’), which Mushaandja (2021: 13) described as a poetic “gesture of radical rudeness” rooted in a feminist tradition. This transgressive directive confronted the Minister for her conservative, insensitive and ignorant views around sexual and reproductive health rights. Inspired by Cardi B’s recent hit single, “WAP”, young protesters taunted the police force with radical hip-hop moves. During one march on Saturday, 10 October, protesters were forced to scatter in central Windhoek after security forces threw tear gas and shot rubber bullets at them. Twenty-six activists were detained, although charges against them were later dropped. Minister of Home Affairs, Frans Kapofi, eventually apologised for the police brutality during a meeting, on 23 October, with youth activists to discuss issues of gender-based violence (Becker 2020). Following the unfolding 2020 Windhoek protests through social media and interpersonal engagements via Zoom and WhatsApp calls from Berlin, Germany, where I was then based, it became increasingly clear to me that this was no longer just a protest against gender-based violence. A new generation of young Namibians were challenging the vestiges of coloniality and raising pertinent questions regarding the politically and socially incomplete liberation of Namibia in 1990 (Becker 2020). When, in March 2022, another protest took to the streets, again demanding the repeal of the prohibitive abortion legislation, one activist tweeted that the march targeted the “APARTHEID abortion act of 1975” (Twitter, 19/03/22, Callipygian; caps in the original). The same activist added that, “As Namibia celebrates Independence, we march for freedom from archaic laws!” This tweet exemplifies how pro-choice and anti-homophobic protestors explicitly regarded the sexism and patriarchy issues concerning them as enduring legacies of apartheid and colonialism. Concurrently the activists turned their focus onto colonial apartheid laws still in force in Namibia. They include the South African Immorality Act of 1957, which prohibits “unlawful carnal intercourse” and the sexual act of sodomy between men. Whilst these laws have rarely been enforced in recent years,14 there have been repeated homophobic attacks, including those fuelled by former President Sam Nujoma in 1996 when he claimed, in an address to the SWAPO Women’s Council Congress, that homosexuality was a negative foreign influence.15 At the time a document, formally issued by the President’s office, claimed that “gays and lesbians” were “exploiting our democracy” (cited by Frank 1997: 6). In 2021, the newly formed Namibia Equal Rights Movement (known as “Equal Namibia”) campaigned for abolition of the sodomy law. In a vibrant social media campaign and through participation in Namibian television and radio shows, queer activists made clear that they regarded this legislation as integral to coloniality in Namibia. Equal Namibia also mobilised public protests around court challenges regarding recognition of same-sex marriages and queer families with the slogan: “There is no freedom if there is no equality”. Unlike during earlier waves of state-induced homophobic campaigns, queer Namibians and their allies were no longer silent. Hundreds came out in public protest against openly displayed homophobia by members of the country’s political class. On 17 November 2021, a vibrant queer protest march swept down Windhoek’s Independence Avenue, proudly waving rainbow flags and colourful banners in protest against homophobic utterances by veteran SWAPO politician Jerry Ekandjo.16 Two weeks later, the largest ever Namibia Pride Parade marched on the Ministry of Justice and about 300 participants held a vigil on the Independence Memorial Museum’s steps, displaying their newly won confidence with banners such as “abolish sodomy law—#lovewins NA” (The Namibian, 7 December 2021). Conclusion Since 2020, young people have been engaged in an array of intersectional activist work, calling for social justice in Namibia. This they have done with reference to decoloniality, while explicitly rejecting identity politics. They tackle issues of social justice and equality, which have not been addressed since independence. Those include matters of poverty and land restitution (Nghitevelekwa 2020). Immediate attention has gone into campaigns against gender based violence, homophobia and for women’s reproductive rights. Those, in turn, have been identified as intersecting with class and race struggles. The activists thus describe their conceptualisation of decolonial activism consciously as intersectional. In Mushaandja’s (2021: 199–201) eloquent expression, “these recent protests demonstrate a consistent focus on the interconnectedness of these struggles and the need to create responsive interventions that take all these links into consideration”. Namibia’s decolonial activists call for inclusivity in society. They demand that Namibian society should espouse full equality and undertake radical reconfigurations of institutions and citizenship. Thus, they contest the Namibian transition of 1990 as an “elite pact” (Melber 2014: 23) between the former colonial rulers and the new nationalist ruling class; and they challenge the country’s society and politics’ clearly visible “authoritarian features” (Melber 2014: 56). Prominent among the intersectional activists are young women and people who identify as “queer”, who are tired of living in a post-independence society where entangled histories of colonialism are reflected in the coloniality of unfinished progressive constitutionalism. While the Namibian constitution promised liberal freedoms once colonialism and apartheid had ended, and despite comparatively progressive gender politics having been implemented after 1990 (Becker 1995), the activists demonstrate that, in many ways, Namibian society and politics remained steeped in conservatism and authoritarian structures. The Ama2000 are disillusioned with the Swapo government, in power since 1990. Namibia’s decolonial activists claim that decolonisation has been abandoned in the post-independence dispensation. This they see reflected in remnants of colonial apartheid legislation still in force. It is also tangible for observers of Windhoek’s built environment, where the country’s colonial heritage remains visible and overlaid with new constructions of Namibian nationalism and capitalism (Becker 2018). Aiming for another possible world in the Namibian context, decolonial intersectional activists imagine a decolonised Namibia, which is ordered differently to the hierarchies which have been inherited and reproduced from colonial legacies: authoritarianism within the formally democratic dispensation, heteronormativity, and a lack of women’s autonomy over their bodies. The analysis of these issues as a perpetuation of coloniality has led Namibia’s young activists to adopt an intersectional approach that disavows identity politics by integrating struggles for the decolonization of the public space through removing colonial monuments and renaming streets with those combatting structural violence, sexism, patriarchy and homophobia. Acknowledgements I thank participants at lectures and seminar presentations at Technical University (TU) Berlin, Bremen University, Hamburg University, and artco-gallery Berlin, where earlier versions of this article were presented. Some of the material introduced in this article was published previously in blog articles on the websites of the Rosa-Luxemburg Foundation, the Review of African Political Economy, and the Namibian Journal of Social Justice. 1 Police brutality was a major concern during Namibia’s brutally enforced first COVID-19 lockdown in Windhoek’s impoverished townships. 2 The Alte Feste was the German colonial military force (Schutztruppe) headquarters. Together with the Reiterdenkmal (see below) and the Christuskirche (Christ Church) consecrated in 1910, it conveyed an encompassing message of an intertwined political–military and spiritual–cultural domination (Steinmetz & Hell 2006: 175). 3 Similar critiques have appeared regarding the US's Black Lives Matter movement (Haider 2018) 4 I adopt John and Jean Comaroff’s (2005) term “policultural” to describe the foundations of postcolonial citizenship. Namibian official discourse has emphasized “unity in diversity” whilst governmental practice has promoted ethnic-based performances of “heritage” and “identity” in state-sponsored cultural festivals (Akuupa 2015) 5 South Africa’s 1996 Constitution expressly prohibits discrimination on grounds of sexual orientation. South Africa made provision for same sex marriages from 2006. Legal abortion has been widely available since 1997. 6 In a prominent case, which has been repeatedly before the courts in 2021 and 2022, Namibia’s Ministry of Home Affairs has opposed residency and citizenship to the Mexican-born partner of Namibian Phillip Lühl and their three children. The couple were married in South Africa, and the children’s South African birth certificates name both men as their parents. 7 Starting from c 2008, several noteworthy artistic works, films and performances also critically addressed the coloniality of the postcolonial Namibian society (see, eg., Brandt 2020; Lehmann 2021), a detailed discussion of those goes however beyond the present analysis’s focus on decolonial social justice movements; only a brief discussion of the noted 2016 “The Mourning” will be provided in the following section. 8 The Reiterdenkmal (literally Horse-rider Monument but usually referred to in English as the Windhoek Rider or the Equestrian Statue) was designed, sculpted and cast in bronze by Adolf Kürle in Berlin. Sitting on a 5-m high sandstone plinth, the double life size (4.5 m) statue of a mounted German colonial soldier with rifle had been used as the logo for Windhoek lager beer, and served in many other formats as the city’s iconic image. Inaugurated in 1912, its plaque commemorated German military and civilian casualties during the 1904-7 colonial war. Despite its location having previously been the site of a concentration camp incarcerating prisoners of that genocidal war, no mention was made of the about one hundred thousand OvaHerero and Nama who were murdered during the genocide (1904-8) by Germany’s colonial army. The statue forcefully illustrated a claim to perpetual colonial domination. The post-independence government stopped illuminating it at night, but otherwise little changed until the removal of the Reiterdenkmal in 2009. 9 Until 2020 public decolonisation of the Namibian public space remained an official project of the postcolonial state. When Namibia finally gained its much-delayed independence (March 1990), the city’s high street, formerly known as ‘Kaiserstrasse’ (‘Emperor’s Street’), was renamed ‘Independence Avenue’. A handful of other streets were renamed. Otherwise, little changed in the capital’s public space. Most streets retained their colonial eponyms. All Windhoek’s German and South African colonial monuments, with their histories of violence, remained. In the mid-2010s a scholar who has written extensively about the Namibian colonial past, observed that, “in a tangible way, the view over Windhoek presents testimony to the current state of public memory in Namibia. Monuments and representative buildings from the colonial era not only dominate the scenery, but attest to the compromise surrounding the transition to independence” (Kössler 2015: 26) The post-independence SWAPO government’s policy was primarily geared at constructing new memorials, statues and monuments that added another layer of public space commemorative aesthetics and narrative. Post-independence structures, such as the Namibian Heroes Acre and the Independence Memorial Museum, are distinct from the colonial monuments in terms of aesthetics and the historical narrative they prescribe. Yet they too are easily comprehended as affirmative glorification of victory (Becker 2011; 2018). 10 After independence in 1990, the former liberation organization South West Africa People’s Organisation (SWAPO) was officially renamed Swapo Party of Namibia. 11 Amupanda was a founder of Affirmative Repositioning, a youth movement started in 2014 with a campaign for land redistribution (Becker 2016). Gawanas is a member of the Landless People’s Movement (LPM), another movement of young activists since turned political party. 12 On 15 June 2022 the City Council convened a stakeholder workshop to develop a policy framework for the identification and management of heritage properties in the capital. In a remarkably transparent move, the workshop was accessible to all via zoom, and inclusive of critical historians and the non-governmental Museums Association of Namibia. 13 Bayron van Wyk’s forthcoming Masters thesis on Namibian genocide reparations and decolonial movements however analyses Windhoek’s Rainbow Sidewalk initiative more explicitly as a critique of masculinist and homophobic attitudes, understood as an effort towards decoloniality. 14 The Namibian Law Reform and Development Commission noted that between 2003 and 2013 115 sodomy cases had been reported to the police and 64 arrests had been made (The Namibian, 18 May 2021). 15 While there is conclusive historical ethnographic evidence, and common knowledge of (male) homosexual practices in some regions of Namibia, as in many other African countries, state-driven homophobic campaigns have repeatedly emphasized alleged “foreign influence”. 16 During a parliamentary debate Ekandjo had said that, “We cannot, comrade speaker, allow a male person to insert his penis into the anus of another man (Video posted by The Namibian on 17 November 2021).”. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Akuupa Michael National culture in post Apartheid Namibia: State-sponsored cultural festivals and their histories 2015 Basel Basler Afrika Bibliographien Becker Heike Commemorating heroes in Windhoek and Eenhana: Memory culture and nationalism in Namibia 1990–2010 Africa Journal of the International African Institute 2011 81 4 519 543 10.1017/S0001972011000490 Becker Heike From ‘to die a tribe and be born a nation’ towards ‘culture, the foundation of a nation’: the shifting politics and aesthetics of Namibian nationalism Journal of Namibian Studies 2015 18 21 35 Becker Heike Changing Urbanscapes: Colonial and postcolonial monuments in Windhoek Nordic Journal of African Studies 2018 27 1 1 21 Becker, Heike. 1995. Namibian women's movement 1980 to 1992. From anticolonial struggle to reconstruction. Frankfurt: IKO - Verlag für Interkulturelle Kommunikation. Becker, Heike. 2016. ‘Namibia’s moment: Youth and urban land activism’. Review of African Political Economy website/blog; 18 January 2016. https://roape.net/2016/01/18/namibias-moment-youth-and-urban-land-activism/. Accessed 22 July 2022. Becker, Heike. 2020. ‘#ShutItAllDownNamibia: Young Namibians are hitting the streets against gender-based violence and colonial legacies’. Rosa-Luxemburg-Stiftung, Online Dossier Namibia@30, 27 October https://www.rosalux.de/en/news/id/43225/shutitalldownnamibia. Accessed 24 July 2022. Brandt, Nicola. 2020. Landscapes between then and now: Recent histories in Southern African photography, performance and video art. London & New York: Routledge. Comaroff John Comaroff Jean Robins Steven L Reflections on liberalism, policulturalism & ID-logy: Citizenship & difference in South Africa Limits to liberation after Apartheid: Citizenship, governance & culture 2005 Oxford James Currey 33 56 Coombes Annie History after Apartheid: Visual culture and public memory in a democratic South Africa 2004 Johannesburg Wits University Press Gillespie Kelly Naidoo Leigh-Ann Between the cold war and the fire: The student movement, antiassimilation, and the question of the future in South Africa The South Atlantic Quarterly 2019 118 1 226 239 10.1215/00382876-7281744 Haider Asad Mistaken identity: Race and class in the age of trump 2018 London Verso Hartman Saidiya Wayward Lives, beautiful experiments: Intimate histories of social upheaval 2019 New York W. Norton Hull George Some pitfalls of decoloniality theory The Thinker 2021 89 63 74 10.36615/thethinker.v89i4.691 Kössler Reinhart Facing a fragmented past: Memory, culture and politics Journal of Southern African Studies 2007 33 2 361 382 10.1080/03057070701292640 Kössler Reinhart Namibia and Germany: Negotiating the past 2015 Windhoek University of Namibia Press Lehmann, Fabian. 2021. Postkoloniale Gegenbilder: Künstlerische Reflexionen des Erinnerns an den deutschen Kolonialismus in Namibia. Basel: Basler Afrika Bibliographien. Maldonado-Torres, Nelson. 2016. ‘Outline of ten theses on coloniality and decoloniality’. Frantz Fanon Foundation. https://fondation-frantzfanon.com/outline-of-ten-theses-on-coloniality-and-decoloniality/. Accessed 13 May 2022. Mayat Yasmin Hart Brendan Judin Hilton Creating spaces of memorialisation: New Dellville Wood (France) and SS Mendi (South Africa) Falling monuments, reluctant ruins: The persistence of the past in the architecture of Apartheid 2021 Johannesburg Wits University Press 193 211 Melber, Henning. 2014. Understanding Namibia: The trials of independence. Auckland Park: Jacana. Mushaandja Nashilongweshipwe Critical visualities & spatialities: Protest, performance, publicness and praxis Namibian Journal of Social Justice 2021 1 192 201 Nghitevelekwa, Romie. 2020. ‘Namibia after 30 years of Independence: Namibians are still waiting for redistributive justice’. Rosa-Luxemburg-Stiftung, Online Dossier Namibia@30, 21 March 2020. https://www.rosalux.de/en/news/id/41791/namibia-after-30-years-of-independence. Accessed 23 July 2022. Niezen Ronald Speaking for the dead: The memorial politics of genocide in Namibia and Germany International Journal of Heritage Studies 2018 24 5 547 567 10.1080/13527258.2017.1413681 Steinmetz George Hell Julia The visual archive of colonialism: Germany and Namibia Public Culture 2006 18 1 147 183 10.1215/08992363-18-1-147 Táíwò Olúfémi Against decolonisation: Taking African agency seriously 2022 London Hurst Titus Hildegard Brandt Nicola Whorrall-Campbell Frances The right to protest: Intersectional activism in Namibia Conversations across place: Reckoning with an entangled world 2021 Berlin The Green Box 167 186	36474751	PMC9715411	NO-CC CODE	2022-12-03 23:20:15	no	Dialect Anthropol. 2022 Dec 2;:1-14	utf-8	Dialect Anthropol	2,022	10.1007/s10624-022-09678-1	oa_other
==== Front Curr Psychol Curr Psychol Current Psychology (New Brunswick, N.j.) 1046-1310 1936-4733 Springer US New York 4072 10.1007/s12144-022-04072-0 Article Test anxiety in online exams: scale development and validity http://orcid.org/0000-0001-7015-6236 Dikmen Melih [email protected] grid.411320.5 0000 0004 0574 1529 Department of Educational Sciences, Firat University Faculty of Education, Elazığ, Turkey 2 12 2022 113 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. Considering the increasing use of online tests, this study aims to develop an up-to-date and reliable scale to measure university students’ online test anxiety. This study was designed by using mixed research model by combining qualitative and quantitative research methods together. The study consisted of four stages: planning, structuring, quantitative evaluation, reliability and validation. While in the first phase an extensive literature review was conducted, students’ opinions were obtained to create an item pool in the second phase. In the third phase, the 29-item scale was administered to 442 university students for factor and reliability analysis. A total of nine items were dropped out from the pool. The Cronbach’s alpha value was .98. Exploratory factor analysis revealed that the items loaded on two factors: the psychological and physiological anxiety factor (α = .95), the technical anxiety factor (α = .89). The two-factor solution accounted for more than 63% of the total variance. The final version of the scale was administered to 387 university students for confirmatory factor analysis in the fourth stage. The results proved that the scale had two factors and the fit indices were at an acceptable level. The reliability analysis was run and Cronbach’s alpha values were .94 the whole scale, .93 for the psychological and physiological anxiety factor, and .90 for the technical anxiety factor. According to the result, it was concluded that the Test Anxiety Scale for Online Exams is a reliable and valid measurement tool in determining university students’ online test anxiety. Finally, recommendations for future research are provided. Keywords Online test anxiety University students Measurement development Online learning ==== Body pmcIntroduction Today, almost all students experience anxiety and fear before or during exams. Test anxiety is defined as the cluster of phenomenological, physiological, and behavioral responses that accompany anxiety that may emerge due to the possible negative consequences of failure in a test or similar assessment (Zeidner, 1998). Test anxiety is considered as a special fear that is not apparent in a daily life but becomes evident before or during an examination (Sarı et al., 2018). Actually, anxiety protects individuals when it is at a normal level; however, it may cause a decrease or deterioration in performance by negatively affecting one’s daily life when it is at a high level. According to their review, Putwain and Daly (2014) concluded that 15% to 22% of student’s experience high level of test anxiety. While Cassady and Johnson (2002) stated that students with test anxiety demonstrate poor academic performance, Hamilton et al. (2021) puts strong emphasis on university students’ critical problems caused by test anxiety. There exist many factors triggering test anxiety including the duration of an exam, the number of questions, the test technique used, test instructions, and environment in which the test is administered (Wadi et al., 2022). Ohata (2005) also discusses the exam duration and its pressure on students. Wong (2008) reported that cognitive structures such as irrational beliefs, negative autonomous thoughts, and dysfunctional attitudes initiate test anxiety. Another factor that causes students to experience test anxiety is their parents’ attitudes and behaviors (Ringeisen & Raufelder, 2015). In another study, Wong (2008) found that students’ irrational beliefs, dysfunctional attitudes and automatic thoughts were significant predictors of test anxiety. In addition, some studies reported that anxiety was affected by personal characteristics and that test anxiety was a non-threatening and temporary situation that students experience (Hodapp et al., 2011). Segool et al. (2013) stated that test anxiety may occur especially during high-risk exams and may be an obstacle to academic performance. Moreover, some studies examined the gender differences in terms of test anxiety and reported that female students experience higher test anxiety than male students (Zaheri et al., 2012). Test anxiety creates deep thoughts and frustration accompanied with physical pain and enthusiasm, which cause students to experience a sense of panic and failure. This situation leads them to consider exams as potential disasters (Maxfield & Melnyk, 2000). Some studies even reported other types of anxiety disorders due to test anxiety in students (Spielberger, 2010). Overall, test anxiety has association with low academic performance (Segool et al., 2013), lack of motivation (Keller & Szakál, 2021), low self-esteem (Thomas et al., 2022), learning difficulties (Chapell et al., 2005), dropout and high level of depression (Leadbeater et al., 2012). In addition, people with test anxiety may show various psychological and physiological symptoms including amnesia, hypertrophy, increased breathing and hearth beating, nausea, diarrhea, sweating, and difficulty in focusing (Huberty & Dick, 2006). Beers (2003) also stated that students with test anxiety may have difficulty in reading questions and finding correct answers, organizing and expressing their thoughts, and finding appropriate words. Test anxiety is a debilitating variable for academic performance from primary school to higher education (Kader, 2016). Manchado Porras and Hervías Ortega (2021) conducted a study and reported a negative correlation between test anxiety and academic performance. A similar result was found by Brady et al. (2018). If one experiences a high level anxiety, his reasoning and abstract thinking skills are disrupted, which, in turn, negatively affects his academic performance. In addition to these, test anxiety may affect one’s cognitive structure by creating failure and disappointment, which are observed through embarrassment, hypersensitivity, and memory problems (Huberty & Dick, 2006). There are various methods to measure test anxiety level. Test anxiety was first considered as a one-dimensional structure (Sarason et al., 1960). Then, it was determined that it had a two-dimensional (Zeidner & Matthews, 2005) and multidimensional structure in the following years (Putwain & Daniels, 2010). According to Mowbray et al. (2015), two dimensions of test anxiety were worry and emotionality. While the worry dimension includes internal conversations such as thinking about the consequences of failure and doubting about the ability to succeed, the emotionality dimension consists of physiological reactions related to anxiety during an examination. In another study, Zeidner and Matthews (2005) considered test anxiety in two dimensions: emotional-physiological and cognitive. While the cognitive dimension refers to negative thoughts and concerns during the exam period, the emotional- physiological factor refers to the unintentional experiences that occur before the exam period such as not taking enough time to study for the exam and procrastination behaviors. In addition to cognitive and emotional factors, some scales include factors related to thoughts, off-task behaviors, autonomic reactions, and so on. The test anxiety measurement tools in the literature were generally designed for exams that are administered in face-to-face environments. With the emergence of COVID-19 pandemic, there was a direct shift to online education. Due to online exams, students’ test anxiety levels may change (Saadé & Kira, 2009). For instance, Block et al. (2008) reported that students had a high level anxiety in their first online exams due to lack of experience and knowledge about them. A similar conclusion was reported by Wang et al. (2001). Most parents are concerned about missing out on learning as a result of the hasty transition to remote study, whereas students have seen benefits. With the decrease in affected cases, many students and parents expected schools to resume face-to-face instruction. However, given the importance of digital literacy in preparing students for future challenges, it must be maintained (Jamilah & Fahyuni, 2022). Blended learning can be an alternative to post-COVID-19 learning because it preserves online learning developed during the pandemic while also requiring physical presence and social interaction, which are key features of face-to-face learning (Andrew et al., 2021; Peimani & Kamalipour, 2021). Systematic reviews demonstrated that online learning can be modified and combined with offline learning to create a blended learning method that schools can adopt and use post-COVID-19 (Jamilah & Fahyuni, 2022). Because digital skills are twenty-first century abilities that students must master, online learning developed during the COVID pandemic must continue to be developed. According to research, online learning can continue in the post-COVID-19 era (Lockee, 2021). Online learning can be modified and combined with offline learning in the post-COVID-19 era to form a blended learning method. This method can compensate for the shortcomings of both online and face-to-face learning. There is likely to be an increased demand for pedagogically sound and adaptable learning environments, as well as innovations in learning technology and design, in the post-COVID-19 era (Peimani & Kamalipour, 2021). Therefore, it is likely that online exams will be widely used, even as the impact of COVID-19 is waning. The previous studies examined the cognitive, affective, physiological, and social dimensions of test anxiety while focusing on traditional test structures. In addition, due to the widespread use of online learning environments, measurement tools are needed to examine students' online test anxiety. Therefore, this study aims to develop a scale that measures students’ test anxiety levels in online exams. Method This study was designed by using mixed research model by combining qualitative and quantitative research methods. The mixed research method enables researchers to ensure the reliability and validity of the research and to eliminate the deficiencies of both qualitative and quantitative methods by combining them (Gültekin et al., 2020). Participants The participants consisted of 859 university students from three different universities in Turkey (male n = 519, female n = 340). Participants reported that they took the online exam at least once. Each stage of the research was consisted with a different sample. Information about the participants was given in detail at each stage. The research procedure was approved by Ethics Review Board of Fırat University. Data analysis In the structuring phase of the research, the data obtained from the participants were analyzed using descriptive analysis. As a first step, the data collected via the interview protocols, and reports were saved digitally on a computer without any alteration or correction. It was indexed as P1, P2, … to P30 for the anonymity of the participants' names and the confidentiality of their personal information. Then the answers to the open-ended questions were coded to identify topics, issues, similarities, and differences revealed through the participants’ narratives (Braun & Clarke, 2006). After discussing emerged meanings to agree on overemphasized or underemphasized themes (Shenton, 2004), another colleague is also consulted for his inner vision into the emerging codes to decrease or avoid any potential bias. The researcher applied an independent coding process for the data gathered from interviews. According to Patton (1999), the expertise of a different expert working in the field of qualitative data analysis than the researchers was utilized. With the analysts whose expertise was used during the data analysis process, authors who completed their Ph.D. in the field of Assessment and Evaluation in Education. Moreover, for inter-rater agreement, the formula of “[the number of agreement / (the number of agreement + the number of disagreement)] × 100” (Miles & Huberman, 1994) was used. The interrater agreement in the initial case was 88.3%. The coding process is considered in two titles: Technical Anxiety, Physiological and Psychological Anxiety. The analysis process included cyclical and continuous comparisons in the form of code development/refinement and theme development/refinement (Glaser & Strauss, 1967). Finally, as a result of discussions on analyst comments, final themes were introduced (Richards & Morse, 2013). The MS Microsoft Excel program was used for data analysis, as suggested by Bree and Gallagher (2016) and Meyer and Avery (2009). Because coding is an important step in analyzing qualitative data (Ose, 2016), this study's qualitative data analysis was done in MS Excel because it allows systematic coding in a simple and user-friendly manner. Analyzes in the quantitative evaluation, reliability and validation phase of this study were performed using SPSS 22.0 and AMOS 20.0. Explanatory factor analysis (EFA) and confirmatory factor analysis (CFA) were performed to examine the internal structure of the survey. In addition, convergent and discriminant validity were performed. There were no missing data. For EFA and CFA, there exist assumptions to be met: the linear relationship, normality and no multicollinearity problem. In this context, the data was examined for normality and linearity. The skewness and kurtosis values must be between -2 and + 2 for the data to be considered normally distributed, according to Tabachnick and Fidell (2007). The normality assumption was met based on preliminary data analysis. There were no univariate or multivariate outliers found using box plots and Mahalanobis distance. Furthermore, the correlation levels of the variables were examined using the variance inflation factor to determine whether multicollinearity exists in the data set (VIF). An acceptable VIF value needs to be < 5.0 (Hair et al., 2010). There was no multiple correlation discovered between variables. Combining EFA and CFA is recommended because EFA, especially at the beginning of a scale development, can account for “unanticipated, but substantively meaningful, factors influencing subsets of items or unanticipated cross-loadings” (Flora & Flake, 2017); in turn, CFA allows to strengthen EFA results by replicating them on a separate sample. To determine the number of factors to be extracted, two different statistical methods were used: Kaiser’s rule (i.e., number of eigenvalues greater than 1, Kaiser, 1960) and scree test (Cattell, 1966). The factors were then extracted using Principal Component Analysis (PCA) with promax oblique rotation, with the hypothesis that there is some degree of correlation between the factors because they are all related to the study. Importantly, the aim of EFA analysis is to determine whether the factor extracted are consistent with the theoretical aspects they should reflect: in our case, factors were expected to technical anxiety, physiological and psychological anxiety. Cronbach's alpha was used to determine the internal consistency of factors, and item-total correlations were also computed. Comparative fit index (CFI) > 0.95, non-normed fit index (NNFI) > 0.95, root mean square error of approximation (RMSEA) < 0.08 and standardized root mean square residual (SRMR) < 0.08 (Hu & Bentler, 1999). The analysis involved in assessing the internal consistency reliability (Cronbach’s alpha and composite reliability), convergent validity Average Variance Extracted (AVE) and factor loading), and discriminant validity (Fornell-Larcker criterion, heterotrait-monotrait, HTMT ratio of correlations criterion) (Fornell & Larcker, 1981; Hair et al., 2017a, b). Development of the Scale The development process consisted of five phases: (1) including planning, (2) structuring, (3) quantitative evaluation, (4) validation, and (5) convergent and divergent validity. Each phase was explained in detail below. Phase 1: Planning The first phase includes the planning process. In this phase, in order to determine the similarities and differences between traditional and online test anxiety, the theoretical and experimental studies related to test anxiety in the literature were examined. It was observed that the psychological dimension was considered as the most important and common dimension in terms of test performance. In addition to that, early studies mentioned emotional, psychological and somatic, and behavior-related dimensions (see Dusek, 1980; Sieber, 1980). While developing the item pool for the scale, the section on anxiety disorders in a book entitled Diagnostic and Statistical Manual of Mental Disorders (DSM-5) was reviewed to identify the symptoms of anxiety. According to the DSM-5, anxiety and concern are associated with three or more of the following six symptoms: Restlessness or feeling keyed up or on edge, being easily fatigued, difficulty concentrating or mind going blank, irritability, muscle tension, and sleep disturbance. This reveals that online test anxiety scale needs to include items related to psychological and physiological anxiety symptoms (American Psychiatric Association, 2021). In their study, Block et al. (2008) and Wang et al. (2001) put strong emphasis on students’ inexperience in online education, their limited knowledge and skills about online exams, and their negative effects on students’ test anxiety levels. In addition, Oyedele and Simpson (2007) pointed out technical anxiety by stating that people may experience a high level of anxiety or stress when they use new tools and equipment that they are not familiar with. Also, Baker et al., (2015) argued that anxiety negatively affects individuals’ logical judgments, decisions, and behaviors in many situations. According to these statements, in addition to the traditional test anxiety, there is a possibility for individuals to experience technical anxiety in online environments. Therefore, it was decided that items related to psychological, physiological, and technical anxiety to be included in the draft version of the test anxiety scale of online exams. Phase 2: Structuring In the second phase of the study, interviews were conducted with 30 students (male: n = 16, female: n = 14, age: M = 22.1 years, SD = 2.33) who already took online test in order to identify the factors influencing test anxiety. The interviews took approximately 20 min. The participants were in their either second or third year in undergraduate education. They were chosen specifically because their instructor had the impression that they had a high level of test anxiety. Due to the pandemic epidemic, the interviews were conducted via online. After getting their permissions, the interviews were recorded. At the beginning of the interviews, each participant was given the definition of anxiety and asked whether they felt anxious before or during online exams. They all agreed that they had test anxiety. The participants then were asked two questions. While the first question was about their physiological and psychological symptoms, the second question was about technical anxiety symptoms. In this study, interviews were used to develop items on online test anxiety. To ensure content adequacy, three experts' ratings were used to confirm that the developed items addressed all two levels of online test anxiety triggers (physiological and psychological symptoms). Cognitive interviews were conducted with four additional people who were not involved in the interviews or expert ratings to test the comprehensibility of the items (age: M = 33.50 years; gender: male: n = 2, female: n = 2) and asked them to think out loud while reading the items and answering them. In order to analyze the data, a content analysis was conducted. Figure 1 is designed to represent the findings related to the physiological and psychological anxiety symptoms.Fig. 1 Findings related to the physiological and psychological anxiety symptoms As seen in Fig. 1, the findings of the answers received from the participants showed that consisted of eight subcategories in the online exams. Participants stated that they experienced the most physiological and psychological anxiety symptoms such as stress, feeling panic and sweating in online exams. Some example quotes are provided below: K5 reflected the symptoms of physical anxiety during the online exam and put it as follows: "As soon as the online exam started, I feel like a bottle of hot water was dumped on me. I feel that my breathing is accelerating and I cannot control it. I get anxious enough to forget what I already know." Similarly, participant P7 stated that he experienced physical anxiety during the online exam and expressed it as follows: "Just minutes before the online exam starts, my hearth start beating so fast like it will get out of my body in seconds. During the exam, my palms get sweaty. This prevents me from focusing on my exam." In the APA Dictionary of Psychology, conditions such as sweating, tremors, dizziness, and rapid heartbeat are seen as physical symptoms of anxiety. Therefore, there is significant evidence that participants experience physical anxiety during the online exam. In addition, there were important opinions that some of the participants experienced psychological anxiety during the online exam. P11 participant stated the symptoms of psychological anxiety during the online exam as follows: "I do not know what I am doing in the exam due to the stress I feel. Even I read the questions several times, I do not understand. I do not remember whether I am that stressful in face-to-face exams. I experience an incredible panic. I really hate online exams." Moreover, participant P17 explained the psychological anxiety in online exams with the following statements: "Exams always make me nervous. Especially in online exams I panic because of the thoughts that I will not be able to keep up the time or the internet will be cutoff. I think online exams are more stressful than face-to-face exams." It can be said that the participants had difficulty in controlling their feelings of worry. This is explained by the psychological symptoms of anxiety (National Institute of Mental Health, 2022). Thus, it can be said that students who participate in online exams experience some physical and psychological anxiety. In the structuring phase of the research, the views of the participants on technical anxiety during the online exam were examined. Figure 2 is designed to demonstrate the findings in regard to participants’ views about technical anxiety.Fig. 2 Findings related to the technical anxiety symptoms As seen in Fig. 2, the findings obtained from the answers received from the participants showed that technical anxiety in online exams consisted of five subcategories. Participants stated that they were most concerned about the lack of technical skills in online exams. Some sample quotes are given below: Participant P11 stated that she had technical anxiety in online exams and expressed this as follows: "I am stressed by constantly thinking that my internet connection will be cutoff during the online exams. I always worry about what to do if my computer breaks down before the exam. In addition, in some exams, although I click on the answer, the system does not accept it. I think online tests are a big problem for students with no solution." Similarly, participant P7 stated that he experienced technical anxiety during the online exam and expressed it as follows:”My computer got frozen in the last exam and I couldn't do anything. I don't understand much about computers. I had to restart it and therefore my time was spent for rebooting it. So not enough time left for the exam. I know many people have similar problems. Online exams are a disaster for students.” All these situations expressed by the participants were concerns arising from technical situations. Internet connection problems, lack of technical skills, systemic problems and difficulties in reading from the screen during the online exam cause participants to experience anxiety in online exams. According to the literature review and the interview results, a strong evidence was obtained about including items related to physiological, psychological, and technical anxiety in the scale. The draft version of the scale consisted of eleven items for the psychological anxiety dimension, nine items for the physiological dimension, and ten items for the technical anxiety dimension. There were a total of 30 items in the draft version. At the end of the second phase, expert views were obtained. The scale was reviewed by three experts from the field. In addition, it was reviewed by a language expert to ensure items’ reading level. Moreover, twelve undergraduate students were asked to review the draft version. The results of the cognitive interviews showed that one items had a similar content. Based on the obtained views, one item was deleted from the psychological anxiety dimension. There were 26 positive and three negative items in the revised version of the scale. It was designed to be a five-point Likert type, where 1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = very often. Phase 3: Quantitative evaluation In this phase, exploratory factor analysis (EFA) was conducted to examine the factor structure of the revised version of the scale. In this phase, the participants consisted of 442 students (male: n = 266, female: n = 176, age: M = 22.02 years) from three different universities. The participants were sent out a form including questions about their demographic information as well as the items of scale via Google Forms. They were asked to answer questions by considering how they feel or think before and during an online exam. Participants stated that they used computers (N = 229), smartphones (N = 211), and tablets (N = 2) during the online exam. Table 1 provides descriptive information about the participants.Table 1 Demographic information about the participants of Phase-3 Variable Statistics Sex Male = 266 (60.2%), Female = 176 (39.8%) Age age range: 18–36; mean age = 22.02, SD = 2.89 Field Type Natural Applied Sciences = 247 (55.9%), Humanities & Social Sciences = 158 (35.7%), Healthcare Sciences = 37 (8.4%) Device used in online exams Computer = 229 (51.8%), Smart phone = 211 (47.7), Tablet PC = 2 (0.5%) Total 442 Before conducting EFA, Kaiser Meyer Olkin (KMO) and Bartlett tests were performed to ensure whether the data was suitable for a factor analysis. The KMO value was found to be 950 and the Barlett’s test results were χ 2 = 6953.87; df = 190 (p = 0.00). According to these results, the data were suitable for factor analysis. A factor analysis was performed with varimax rotation. According to Bryant and Yarnold (1995), rotation refers to a process in which eigenvectors (factors) are rotated to reach a basic structure. The results revealed two factors that had eigenvalue higher than one. Table 2 is designed to provide the factor loadings, eigenvalues, factor variance, cumulative variance, and the cronbach’s alpha value of the scale.Table 2 The EFA results Item no Factors and item loadings F1 F2 Q1 F1: Physiological and Psychological Anxiety 0.652 Q2 0.735 Q3 0.743 Q4 0.697 Q5 0.726 Q6 0.612 Q7 0.512 Q8 0.664 Q9 0.762 Q10 0.860 Q11 0.819 Q12 0.786 Q13* F2: Technical Anxiety 0.441 Q14 0.777 Q15 0.803 Q16 0.781 Q17 0.676 Q18 0.709 Q19 0.565 Q20 0.606 * Reverse-coded item According to the EFA results, nine items were dropped out due to their correlation values of less than 0.20. As a result, there were a total of 20 items (six items for the physiological dimension, six items for the psychological dimension, and eight items for the technical dimension). The EFA resulted in a two-factor structure explaining 63.4% of the variance with the eigenvalue higher than one. Also, the scree plot indicated a two-factor structure as well (Fig. 3).Fig. 3 Scree plot of 20-item version of the scale In Fig. 3, it was observed that the factors after the second dot were less than one and close to each other, which was concluded that the scale had two factors. While the first factor explained 55.43% of the total variance, the second one explained 8% of it. The total variance explained was 63.43%. The items’ factor loading values ranged between 0.52 and 0.85 for the overall measurement. The factors were named based on the literature, items, and expert views. According to the findings, the items related to physiological and psychological anxiety symptoms gathered together. Thus, the first one was named as the physiological and psychological anxiety (PPA) factor. The second one was named as the technical anxiety (TA) factor since the items under this factor were related to technical anxiety symptoms. The Cronbach’s alpha values for the whole scale and the factors were found to be 0.98, 0.95, and 0.89, respectively. Phase 4: Reliability and validation In the reliability and validation phase of the study consisted 387 students completed a survey (male: n = 237, female: n = 150; age: M = 22 years SD = 2.99). Participants declared that they took the exam of at least once courses online (M = 1.07, SD = 0.251). All participants declared that they have a constant internet connection at home. In addition, the majority of the participants reported that they took the online exam with a computer. Table 3 provides descriptive information about the participants.Table 3 Demographic information about the participants of Phase-4 Variable Statistics Sex Male = 237 (60.2%), Female = 150 (39.8%) Age age range: 18–32; mean age = 22, SD = 2.99 Field Type Natural Applied Sciences = 143(55.9%), Humanities & Social Sciences = 152 (35.7%), Healthcare Sciences = 92 (8.4%) Device used in online exams Computer = 369 (51.8%), Smart phone = 17 (47.7), Tablet PC = 1 (0.5%) Total 387 The data obtained from these students were used in descriptive analysis, CFA, composite reliability, convergent validity and discriminant validity analyses. Descriptive analysis results of the data obtained from these students are given in Table 4.Table 4 Descriptive analysis results N Minimum Maximum Mean SD Skewness Kurtosis PPA 387 1.00 4.00 1.341 0.497 1.213 1.915 TA 387 1.00 4.63 1.692 0.658 0.798 -0.48 Scale (Overall) 387 1.00 3.45 1.481 0.507 1.044 0.269 PPA Physiological and psychological anxiety, TA Technical anxiety As a result of the analysis, it was observed that the mean of the scale was 1.48 (SD = 0.506). In the analysis made in terms of sub-dimensions (factors), the mean of PPA was 1.34 (SD = 0.497) and the mean of TA was 1.69 (SD = 0.658). It was observed that the Skewness and Kurtosis values of the scores obtained from the scale were within the interval of ± 2. Confirmatory factor analysis Although EFA is widely used in developing psychological scales, there are many contradictory situations regarding EFA including which extraction and rotation method are acceptable and how to decide the number of factors (Tabachnick & Fidell, 2007). Therefore, it is suggested to conduct other analysis to ensure the results of EFA (Osborne & Fitzpatrick, 2012). For this particular study, a confirmatory factor analysis (CFA) was performed. Although Schmitt (2011) stated that in order to test the results of EFA, the same data set may be used for CFA, there are opposite views as well (Schumacker & Lomax, 2010). For instance, it is suggested to split the data set into two parts if the data set is large enough to conduct EFA and CFA or to collect data from a different sample to conduct CFA after performing EFA (Schumacker & Lomax, 2010). Since the purpose of CFA is to test the EFA results to make sure the obtained structure is reliable (Brown, 2006), data collection procedure was repeated with a different sample group who had similar characteristics to which EFA was performed (Fig. 4).Fig. 4 Confirmatory model Only one modification was added between e3 and e4. The item loadings ranged between 0.63 and 0.80 for the physiological and psychological anxiety factor and 0.54 and 0.91 for the technical anxiety factor. The standardized parameter estimates and t value of the CFA were significant (p < 0.001). The fit indices are provided in Table 5.Table 5 Results of CFA CMIN Sd p CMIN/SD CFI AGFI GFI IFI TLI RMSEA SRMR 544.644 168 0.000 3.242 0.89 0.88 0.87 0.89 87 0.075 0.062 According to the table, the X2/sd value was 3.242, the CFI, AGFI, GFI, IFI and NFI values were close to one, and the RMSEA and SRMR values were smaller than 0.080. Based on the sample size, the Chi-square values were found to be an acceptable fit (Kline, 2005). Those values indicate a good fit between the model and the observed data (Schreiber et al., 2006). The results proved that the scale has a two-factor structure. In addition, the t-values were between 11.13 and 20.07 for the items in the scale, which were significant (p < 0.01). Internal consistency reliability Internal consistency reliability was used to measure the reliability of survey items in a construct. Internal consistency reliability is achieved when all items of such measures can reflect the same underlying construct (Myrtveit & Stensrud, 2012). Crobach’s alpha (α) and composite reliability are two indicators to measure internal consistency of reliability. To achieve internal consistency reliability, the recommended level of α should be more than 0.70 and composite reliability value should be between 0.70 to 0.95 (Hair et al., 2017a, b). The correlation among technical anxiety, psychological and physiological anxiety structure, cronbach’s alpha, and composite reliability values are presented in Table 6.Table 6 Correlation Matrix, Cronbach’s alpha, and Composite Reliability 1 2 Cronbach’s alpha Composite reliability 1.PPA - 0.934 0.938 2.TA 0.614 0.898 0.901 3.Scale (Overall) 0.898 0.899 0.936 PPA Physiological and psychological anxiety, TA Technical anxiety Based on Table 6, it was observed that technical anxiety, psychological and physiological anxiety correlated significantly with each other. The cronbach’s alpha values for the whole scale and the factors were found to be 0.94, 0.90, and 0.93, respectively. In addition, composite reliability value for PPA and TA was 0.94 and 0.90. This result concluded that all items in this survey study were reliable as they reflected to its own underlying construct. Convergent validity Convergent validity was used to measure the degree of the correlation between items in the same construct (Campbell & Fiske, 1959). Convergent validity is achieved when items in a same construct are strongly correlated to each other (Bagozzi & Yi, 2012). Factor loading and Average Variance Extracted (AVE) are two indicators to measure convergent validity. To achieve convergent validity value of AVE of each construct should be exceeded 0.50 (Hair et al., 2017a, b). The factor loading and AVE are shown in Table 7.Table 7 Factor Loading and Average Variance Extracted (AVE) Values Items Factor Loading S.E C.R p AVE PPA1 0.704 0.046 14.583 * 0.557 PPA2 0.786 0.066 16.168 * PPA3 0.773 0.074 15.888 * PPA4 0.787 0.076 16.169 * PPA5 0.802 0.058 16.459 * PPA6 0.677 0.082 13.693 * PPA7 0.630 0.081 12.570 * PPA8 0.744 0.061 15.152 * PPA9 0.691 0.073 14.243 * PPA10 0.780 0.068 16.427 * PPA11 0.797 0.056 16.691 * PPA12 0.763 0.064 16.691 * TA13 0.664 0.043 14.705 * 0.538 TA14 0.907 0.043 25.079 * TA15 0.827 0.044 20.624 * TA16 0.714 0.041 16.270 * TA17 0.538 0.038 11.130 * TA18 0.641 0.044 14.066 * TA19 0.633 0.057 13.762 * TA20 0.865 0.093 13.762 * * < 0.001, PPA Physiological and psychological anxiety, TA Technical anxiety Based on Table 7, it was determined that all items in a same construct were strongly correlated to each other. Thus, this suggested that the survey items of the study have a good convergent validity. Discriminant validity Discriminant validity was used to measure the degree of the correlation between items in different construct (Campbell & Fiske, 1959). Discriminant validity is achieved when items in a particular construct are not highly correlated with any items in other constructs (Hulland, 1999). Fornell-Larcker criterion and heterotrait-monotrait (HTMT) ratio of correlations criterion are two indicators to measure discriminant validity. To achieve discriminant validity, square root of the construct’s AVE should be the highest correlation with any other constructs and the HTMT value should be lower than 0.90 (Hair et al., 2017a, b). Based on Table 8, the square root of all the constructs’ AVE was larger than the squared correlation with any other constructs. This means that items in the construct of the PPA were not highly correlated with any items in construct of TA. In addition, all the values of construct passed HTMT value of 0.90 tests (HTMT = 0.674). Therefore, with these results, we concluded that discriminant validity issue was not existed in this study.Table 8 The Discriminant Validity Results Fornell-Larcker Criterion HTMT Criterion PPA TA PPA TA PPA 0.746 1 TA 0.624 0.733 0.674 1 PPA Physiological and psychological anxiety, TA Technical anxiety The final version of TAS-OE is provided in Table 9. It has 20 items with two factors. The physiological and psychological factor has 12 items (item 1,2,3,4,5,6,7,8,9,10,11, and 12) and the technical anxiety factor has eight items (item 13,14,15,16,17,18,19, and 20). There is only one reverse-coded item (item 13). It is designed to be a five-point Likert type, where 1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = very often. For analysis, a total score is calculated. A high score indicates a high level of test anxiety.Table 9 The final version of TAS-OE Items PPA1 I experience stomach cramps before or during online exams PPA2 I feel nervous, intense, and restless right before or during online exams PPA3 I get anxious enough to forget what I already know during online exams PPA4 I feel panic and I feel that something terrible will happen during online exams PPA5 Online exams make me feel weak and powerless PPA6 I experience more stress with online exams than with face-to-face exams PPA7 When I fail in the online exams, other people’s thoughts about me irritate me PPA8 My emotions distract me during online exams PPA9 I feel my heart beating gets increased just before or during online exams PPA10 I feel that my breathing is accelerated just before or during online exams PPA11 I feel hot or cold and my palms get sweaty just before or during online exams PPA12 My hands shake just before or during online exams TA13 Time management is easier in online exams TA14 I feel nervous and uneasy when I think that my internet connection may be cutoff at any time during an online exam TA15 I feel nervous and uneasy when I think that my device (e.g., computer and smartphone) that I use for online exams may break down at any time TA16 The thought of not being able to access the Internet before online exams makes me anxious and restless until the exam day TA17 I don't have enough technical skills to complete an online exam, which makes me nervous and uneasy TA18 I am concerned about whether my answers to questions on online exams are being recorded TA19 The limited number of question types used in online exams makes me think that online exams are not suitable for every course TA20 Reading the questions in online exams on the screen causes the questions to be misunderstood PPA Physiological and psychological anxiety, TA Technical anxiety Discussion The goal of this study was to develop and validate TAS-OE to assess students’ test anxiety levels in online exams. The study consisted of four phased. In the first phase, an extensive literature review was conducted and an item pool was developed based on the literature. In the second phase, interviews were conducted with students with a high level test anxiety. They were asked about their anxiety experiences before and during online exams. The interview data were combined with the results of the extensive literature review. As a result, a draft version of TAS-OE with 29 items was developed. The following phase allowed the researcher to conduct exploratory factor analysis to examine the factor structure of the scale. The EFA was conducted with 442 cases and resulted in a two-factor structure with 20 items. There were nine items dropped out from the scale, which left 20 items. In the fourth phase, this structure was tested through confirmatory factor analysis. Overall, the scale showed acceptable fit to the data and loadings in line with the literature, expert views, and students’ real anxiety experiences in online exams. The final version of the scale consists of 20 items with two factors. The first factor is the physiological and psychological anxiety factor with twelve items and the second factor is the technical anxiety factor with eight items. The two-factor structure is consistent with previous studies revealing that test anxiety is a multi-dimensional structure (Alibak et al., 2019; Putwain & Daniels, 2010). More specifically, the physiological and psychological anxiety factor of TAS-OE coincides with the psychological and somatic dimensions in the questionnaires related to exam anxiety (Mowbray et al., 2015). In addition, in some studies on anxiety (Sarason, 1978; Barnes et al., 2019), it is seen that physical and psychological symptoms are considered together. Moreover, it has been suggested that physical and psychological symptoms are a common structure influenced by personality traits (Spangler, 1997). In this context, considering the physical and psychological anxiety dimensions together for the online exams revealed in the current research is compatible with the literature. Also, the results of Folk’s study (2010) support the technical anxiety factor of TAS-OE. Actually, there exist a limited number of scales measuring students’ online test anxiety levels (Alibak et al., 2019). A special feature of TAS-OE that differentiates it from the other scales is that it includes items related to technical anxiety. The reason to include such items is that use of technology in learning process may increase students’ anxiety levels (Matsumura & Hann, 2004), which negatively affect learning outcomes (Brown et al., 2004). Therefore, TAS-OE will enable researchers to identify the online test anxiety levels of university students in terms of physiological, psychological, and technical aspects and to determine ways to diminish negative effects of test anxiety on academic performance. Although the impact of COVID-19 has decreased worldwide, the place and importance of online tests in our lives cannot be denied. The results of the current research provide a reliable tool for determining online test anxiety in terms of improving the effectiveness and efficiency of online learning. Theoretical implications This study conceptualized the structure of online test anxiety. TAS-OE extends existing work of online exam anxiety or stress in the following ways: First, TAS-OE is not related to online exams conducted with specific technologies and therefore is also applicable to future, anticipated technologies. The items refer to online test anxiety in TAS-OE indicate that online tests that individuals are not yet accustomed to can cause anxiety. Second, the TAS-OE addresses online exams as an ongoing process involving the integration of technology into all aspects of the education and training process. This process perspective is reflected in two ways: (a) The TAS-OE includes a unique subscale describing anxiety triggers related to online exams (b) Items are formulated in a way that incorporates a process perspective, mostly by using verbs such as “feel” or “experience”, which describe feelings, skills, and behaviors. Finally, TAS-OE differs somewhat from test anxiety constructs. (1) technical skill anxiety (specific to the nature of online exams) and (2) psychological and physiological anxiety (similar to the construct of test anxiety). The theoretical implications of the research were conceptualized by researching online test anxiety in a concrete way based on qualitative and quantitative data. Practical implications As online exams are increasingly used in education environment, attitudes and fears towards these exams should be continuously monitored with effective measures. The TAS-OE scale can be used as such a measure by educators or students to identify the “top triggers” of online test anxiety in education institutions. Completing the TAS-OE can help individuals develop measures to counteract their online test anxiety. The current study revealed that online test anxiety is associated with technical skill, cognitive and behavioral indicators. Online test anxiety can increase an individual's perceived online test anxiety and make it difficult for them to have positive experiences with learning. In conclusion, the 21-item TAS-OE has been shown to provide satisfactory reliability, composite reliability, criterion validity, content validity, discriminant validity, and convergent validity. Limitations and future research There were several limitations of the study. The first limitation is about students’ high level of test anxiety in online exams. Since this development and validity study was conducted during COVID-19 epidemic, students' anxiety levels about the epidemic may have an impact on the findings. Therefore, future research must examine the effects of students’ perceptions about epidemic on students’ test anxiety in online exams. In addition, in order to ensure the validity and reliability of TAS-OE, this study must be replicated in learning environments in which blended learning approach is adapted. The second limitation is about technical anxiety. Individuals who need to control and manage technological devices are likely to experience anxiety when they have insufficient knowledge about technology and how to use it. Therefore, it is suggested to examine the structure of online exams to determine which factors trigger students’ anxiety. The third limitation is about the sample and the sample size of the study. The data were collected from university students from different majors. Future research must consider replicating the study with participants from different grade levels, majors, and cultures and with a larger sample size. This enables researchers to examine TAS-OE’s psychometric properties. Further, the participants were recruited by using convenient sampling method; future studies must consider using purposeful sampling method in order to determine other possible factors that influence their test anxiety in online exams. Data availability The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Consent to participate Informed consent was obtained from all individual participants included in the study. Consent to publish The authors affirm that human research participants provided informed consent for publication of the research. Conflict of interest/Competing interests On behalf of all authors, the corresponding author states that there is no conflict of interest. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References American Psychiatric Association. (2021, June). What are anxiety disorders?. https://www.psychiatry.org/patients-families/anxiety-disorders/what-are-anxiety-disorders Alibak M Talebi H Neshat-Doost HT Development and validation of a test anxiety inventory for online learning students Journal of Educators Online 2019 16 2 10 25 Andrew L Wallace R Sambell R A peer-observation initiative to enhance student engagement in the synchronous virtual classroom: A case study of a COVID-19 mandated move to online learning Journal of University Teaching & Learning Practice 2021 18 4 14 10.53761/1.18.4.14 Bagozzi RP Yi Y Specification, evaluation, and interpretation of structural equation models Journal of the Academy of Marketing Science 2012 40 1 8 34 10.1007/s11747-011-0278-x Baker, S. R., Bloom, N., & Davis, S. J. (2015). Measuring economic policy uncertainty, kellogg school of management. Stanford University and Chicago Booth School of Business Working Paper. Barnes MC Kessler D Archer C Wiles N Prioritising physical and psychological symptoms: What are the barriers and facilitators to the discussion of anxiety in the primary care consultation? BMC Family Practice 2019 20 1 1 10 10.1186/s12875-019-0996-6 30606122 Beers, K. (2003). When Kids Can't Read: What teacher's can do (Vol. 20, No. 3). Heinemann. Block A Udermann B Felix M Reineke D Murray S Achievement and satisfaction in an online versus a traditional health and wellness course MERLOT: Journal of Online Learning and Teaching 2008 4 1 57 66 Brady ST Hard BM Gross JJ Reappraising test anxiety increases academic performance of first-year college students Journal of Educational Psychology 2018 110 3 395 10.1037/edu0000219 Braun V Clarke V Using thematic analysis in psychology Qualitative Research in Psychology 2006 3 2 77 101 10.1191/1478088706qp063oa Bree RT Gallagher G Using Microsoft Excel to code and thematically analyze qualitative data: A simple, cost-effective approach All Ireland Journal of Higher Education 2016 8 2 2811 2824 Brown TA Confirmatory factor analysis for applied research 2006 Guilford Brown SA Fuller RM Vician C Who’s afraid of the virtual world? Anxiety and computer-mediated communication Journal of the Association for Information Systems 2004 5 2 79 107 10.17705/1jais.00046 Bryant FB Yarnold PR Grimm LG Yarnold PR Principal-components analysis and exploratory and confirmatory factor analysis Reading and understanding multivariate statistics 1995 1 American Psychological Association 99 136 Campbell DT Fiske DW Convergent and discriminant validation by the multitrait-multimethod matrix Psychological Bulletin 1959 56 2 81 10.1037/h0046016 13634291 Cassady JC Johnson RE Cognitive test anxiety and academic performance Contemporary Educational Psychology 2002 27 2 270 295 10.1006/ceps.2001.1094 Cattell RB The scree test for the number of factors Multivariate Behavioral Research 1966 1 2 245 276 10.1207/s15327906mbr0102_10 26828106 Chapell MS Blanding ZB Silverstein ME Takahashi M Newman B Gubi A McCann N Test anxiety and academic performance in undergraduate and graduate students Journal of Educational Psychology 2005 97 2 268 274 10.1037/0022-0663.97.2.268 Dusek JB Sarason IG The development of test anxiety in children Test anxiety: Theory, research, and applications 1980 Lawrence Erlbaum Associates 87 110 Flora DB Flake JK The purpose and practice of exploratory and confirmatory factor analysis in psychological research: Decisions for scale development and validation Canadian Journal of Behavioural Science/revue Canadienne Des Sciences Du Comportement 2017 49 2 78 10.1037/cbs0000069 Folk, J. (2010, November 20). Anxiety Centre. Retrieved from http://www.anxietycentre.com/anxiety-symptoms.shtm Fornell, C., Larcker, D. (1981, May 1). Structural Equation Models With Unobservable Variables and Measurement Error: Algebra and Statistics. Working Paper No. 266. https://deepblue.lib.umich.edu/bitstream/handle/2027.42/35622/b1378752.0001.001.pdf?sequence=2 Glaser BG Strauss AL The discovery of grounded theory: Strategies for qualitative research 1967 Aldine Publishing Company Gültekin, M., Gürdoğan-Bayır, Ö., & Yaşar, E. (2020). Mixed Research methods. In B. Oral & A. Çoban (Eds.), Scientific research methods in education from theory to practice (pp. 317–350). Pegem. 10.14527/9786257880176.12 Hair JF Black WC Babin BJ Anderson RE Multivariate data analysis 2010 7 Prentice-Hall Inc. Hair JF Jr Sarstedt M Ringle CM Gudergan SP Advanced issues in partial least squares structural equation modeling 2017 Sage publications Hair JF Hult GTM Ringle CM Sarstedt M Thiele KO Mirror, mirror on the wall: A comparative evaluation of composite-based structural equation modeling methods Journal of the Academy of Marketing Science 2017 45 5 616 632 10.1007/s11747-017-0517-x Hamilton N Freche R Zhang Y Zeller G Carroll I Test anxiety and poor sleep: A vicious cycle International Journal of Behavioral Medicine 2021 28 2 250 258 10.1007/s12529-021-09973-1 33730347 Hodapp V Rohrmann S Ringeisen T The test anxiety questionnaire 2011 Hogrefe Hu LT Bentler PM Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives Structural Equation Modeling: A Multidisciplinary Journal 1999 6 1 1 55 10.1080/10705519909540118 Huberty, T. J., & Dick, A. C. (2006). Performance and test anxiety. In Bear, G. G. & Minke, K. M. (Eds.), Children’s needs III: Development, prevention, and intervention (pp. 281–291). The National Association of School Psychologists. Hulland J Use of partial least squares (PLS) in strategic management research: A review of four recent studies Strategic Management Journal 1999 20 2 195 204 10.1002/(SICI)1097-0266(199902)20:2<195::AID-SMJ13>3.0.CO;2-7 Jamilah, J., & Fahyuni, E. F. (2022). The future of online learning in the Post-COVID-19 era. Proceedings of KnE Social Sciences, Indonesia, 497–505. 10.18502/kss.v7i10.11251 Kader AA Debilitating and facilitating test anxiety and student motivation and achievement in principles of microeconomics International Review of Economics Education 2016 23 40 46 10.1016/j.iree.2016.07.002 Kaiser HF The application of electronic computers to factor analysis Educational and Psychological Measurement 1960 20 1 141 151 10.1177/001316446002000116 Keller T Szakál P Not just words! Effects of a light-touch randomized encouragement intervention on students’ exam grades, self-efficacy, motivation, and test anxiety PLoS ONE 2021 16 9 e0256960 10.1371/journal.pone.0256960 34525100 Kline TJ Psychological testing: A practical approach to design and evaluation 2005 Sage Publications Leadbeater B Thompson K Gruppuso V Co-occurring trajectories of symptoms of anxiety, depression, and oppositional defiance from adolescence to young adulthood Journal of Clinical Child & Adolescent Psychology 2012 41 6 719 730 10.1080/15374416.2012.694608 22742519 Lockee BB Online education in the post-COVID era Nature Electronics 2021 4 1 5 6 10.1038/s41928-020-00534-0 Manchado Porras M Hervías Ortega F Procrastinación, ansiedad ante los exámenes y rendimiento académico en estudiantes universitarios Interdisciplinaria 2021 38 2 242 258 Matsumura S Hann G Computer Anxiety and students' preferred feedback methods in EFL writing The Modern Language Journal 2004 88 3 403 415 10.1111/j.0026-7902.2004.00237.x Maxfield L Melnyk WT Single session treatment of test anxiety with eye movement desensitization and reprocessing (EMDR) International Journal of Stress Management 2000 7 2 87 101 10.1023/A:1009580101287 Meyer DZ Avery LM Excel as a qualitative data analysis tool Field Methods 2009 21 1 91 112 10.1177/1525822x08323985 Miles MB Huberman AM Qualitative data analysis: An expanded Sourcebook 1994 2 Sage Mowbray T Jacobs K Boyle C Validity of the German test anxiety inventory (TAI-G) in an Australian sample Australian Journal of Psychology 2015 67 2 121 129 10.1111/ajpy.12058 Myrtveit I Stensrud E Validity and reliability of evaluation procedures in comparative studies of effort prediction models Empirical Software Engineering 2012 17 1 23 33 10.1007/s10664-011-9183-7 National Institute of Mental Health. (2022, April). Anxiety disorders. https://www.nimh.nih.gov/health/topics/anxiety-disorders/index.shtml Ohata K Potential Sources of Anxiety for Japanese Learners of English: Preliminary Case Interviews with Five Japanese College Students in the U.S. TESL-EJ 2005 9 3 1 21 Osborne JW Fitzpatrick DC Replication analysis in exploratory factor analysis: What it is and why it makes your analysis better Practical Assessment, Research, and Evaluation 2012 17 1 15 10.7275/h0bd-4d11 Ose SV Using excel and word to structure qualitative data Journal of Applied Social Science 2016 10 2 147 162 10.1177/1936724416664948 Oyedele A Simpson PM An empirical investigation of consumer control factors on intention to use selected self-service technologies International Journal of Service Industry Management 2007 18 3 287 306 10.1108/09564230710751497 Patton MQ Enhancing the quality and credibility of qualitative analysis HSR: Health Services Research 1999 34 1189 1208 10591279 Peimani N Kamalipour H Online education in the post COVID-19 era: Students’ perception and learning experience Education Sciences 2021 11 10 633 10.3390/educsci11100633 Putwain D Daly AL Test anxiety prevalence and gender differences in a sample of English secondary school students Educational Studies 2014 40 5 554 570 10.1080/03055698.2014.953914 Putwain DW Daniels RA Is the relationship between competence beliefs and test anxiety influenced by goal orientation? Learning and Individual Differences 2010 20 1 8 13 10.1016/j.lindif.2009.10.006 Richards L Morse JM Read me first for a user’s guide to qualitative methods 2013 3 Sage Ringeisen T Raufelder D The interplay of parental support, parental pressure and test anxiety–gender differences in adolescents Journal of Adolescence 2015 45 67 79 10.1016/j.adolescence.2015.08.018 26378971 Saadé RG Kira D Computer anxiety in e-learning: The effect of computer self-efficacy Journal of Information Technology Education: Research 2009 8 1 177 191 10.28945/166 Sarason SB Davidson KS Lighthall FF Waite RR Ruebush BK Anxiety in elementary school children 1960 Wiley Sarason, I. G. (1978). The test anxiety scale: Concept and research. In C. D. Spielberger, & I. G. Sarason (Eds.), Stress and Anxiety (4th ed., pp. 193–216). Hemisphere. Sarı SA Bilek G Çelik E Test anxiety and self-esteem in senior high school students: A cross-sectional study Nordic Journal of Psychiatry 2018 72 2 84 88 10.1080/08039488.2017.1389986 29037120 Schmitt TA Current methodological considerations in exploratory and confirmatory factor analysis Journal of Psychoeducational Assessment 2011 29 4 304 321 10.1177/0734282911406653 Schreiber JB Nora A Stage FK Barlow EA King J Reporting structural equation modeling and confirmatory factor analysis results: A review The Journal of Educational Research 2006 99 6 323 338 10.3200/JOER.99.6.323-338 Schumacker RE Lomax RG A beginner's guide to structural equation modeling 2010 3 Routledge Segool N Carlson J Goforth A von der Embse N Barterian J Heightened test anxiety among young children: Elementary school students’ anxious responses to high-stakes testing Psychology in the Schools 2013 50 5 489 499 10.1002/pits.21689 Shenton AK Strategies for ensuring trustworthiness in qualitative research projects Education for Information 2004 22 2 63 75 10.3233/EFI2004-22201 Sieber JE Sarason IG Defining test anxiety: Problems and approaches Test anxiety: Theory, research, and applications 1980 Lawrence Erlbaum Associates 15 40 Spangler G Psychological and physiological responses during an exam and their relation to personality characteristics Psychoneuroendocrinology 1997 22 6 423 441 10.1016/S0306-4530(97)00040-1 9364621 Spielberger, C. D. (2010). Test Anxiety Inventory. https://onlinelibrary.wiley.com/10.1002/9780470479216.corpsy0985 Tabachnick BG Fidell LS Experimental designs using ANOVA 2007 Thomson/Brooks/Cole 724 Thomas T Joseph G Paul S A Study to Assess the Correlation between Academic Test Anxiety and Self-Esteem among Undergraduate Students Journal of Health and Allied Sciences NU 2022 10.1055/s-0042-1742464 Wadi M Yusoff MSB Abdul Rahim AF Lah NAZN Factors affecting test anxiety: A qualitative analysis of medical students’ views BMC Psychology 2022 10 1 1 8 10.1186/s40359-021-00715-2 34980253 Wang A Newlin M Tucker T A discourse analysis of online classroom chats: Predictors of cyber-student performance Teaching of Psychology 2001 28 3 222 226 10.1207/S15328023TOP2803_09 Wong SS The relations of cognitive triad, dysfunctional attitudes, automatic thoughts, and irrational beliefs with test anxiety Current Psychology 2008 27 3 177 191 10.1007/s12144-008-9033-y Zaheri F Shahoei R Zaheri H Gender differences in test anxiety among students of guidance schools in Sanandaj, Iran Wudpecker Journal of Medical Sciences 2012 1 1 001 005 Zeidner M Test anxiety: The state of the art 1998 Plenum Press Zeidner M Matthews G Elliot A Dweck C Evaluative anxiety Handbook of competence and motivation 2005 Guilford Press 141 166	36474484	PMC9715417	NO-CC CODE	2022-12-03 23:20:15	no	Curr Psychol. 2022 Dec 2;:1-13	utf-8	Curr Psychol	2,022	10.1007/s12144-022-04072-0	oa_other
==== Front Endocrine Endocrine Endocrine 1355-008X 1559-0100 Springer US New York 36456885 3273 10.1007/s12020-022-03273-8 Original Article Temporal trends in referrals of RET gene carriers for neck surgery to a tertiary surgical center in the era of international management guidelines http://orcid.org/0000-0002-6796-079X Machens Andreas [email protected] 1 Lorenz Kerstin 1 Huessler Eva-Maria 2 Stang Andreas 23 Weber Frank 4 Dralle Henning 14 1 grid.9018.0 0000 0001 0679 2801 Medical Faculty, Department of Visceral, Vascular and Endocrine Surgery, Martin Luther University Halle-Wittenberg, Ernst-Grube-Str. 40, D-06097 Halle (Saale), Germany 2 grid.410718.b 0000 0001 0262 7331 Institute of Medical Informatics, Biometry and Epidemiology, University Hospital Essen, Essen, Germany 3 grid.189504.1 0000 0004 1936 7558 Department of Epidemiology, School of Public Health, Boston University, Boston, MA USA 4 grid.5718.b 0000 0001 2187 5445 Department of General, Visceral and Transplantation Surgery, Division of Endocrine Surgery, University of Duisburg-Essen, D-45122 Essen, Germany 2 12 2022 111 7 10 2022 21 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. Purpose Thirty years into the genomic era, this study sought to explore events that helped transform the clinical landscape of hereditary medullary thyroid cancer (MTC). Method This retrospective analysis of prospectively collected data included all RET carriers referred to a tertiary center for neck surgery that was performed between 1986 and 2021, using descriptive statistics and Poisson regression analysis. Results Altogether, 496 RET carriers were referred for thyroidectomy (388 carriers) or neck reoperation (108 carriers). Of these, 44 carriers had highest risk mutations (p.Met918Thr), 164 carriers high risk mutations (p.Cys634Arg/Gly/Phe/Ser/Trp/Tyr/insHisGluLeuCys), 116 carriers moderate–high risk mutations (p.Cys609/611/618/620/630Arg/Gly/Phe/Ser/Tyr) and 172 carriers low–moderate risk mutations (p.Glu768Asp, p.Leu790Phe, p.Val804Leu/Met, or p.Ser891Ala). Three event clusters drove referral numbers upward: a string of first reports of causative RET mutations between 1993 and 1998; the international consensus guidelines for diagnosis and therapy of MEN type 1 and type 2 in 2001; and the revised American Thyroid Association guidelines for the management of medullary thyroid carcinoma in 2015. Referrals for neck reoperation declined sluggishly over 30 years, ending in 2018. Index patients continued to be referred into 2021. Referrals for thyroidectomy, grouped in 5-year increments, peaked in 1996–2000 for carriers of highest and high risk mutations, and in 2006–2010 for carriers of moderate–high and low–moderate risk mutations, some 10 years later. Conclusion International management guidelines are critical in building and increasing the pressure towards screening of sporadic-appearing disease and offspring of known gene families by encompassing the complete disease spectrum early on. Keywords RET Protooncogene Genotype-Specific Malignant Transformation Precision Medicine Practice Management Guidelines Medullary Thyroid Carcinoma Preemptive Thyroidectomy ==== Body pmcIntroduction Hereditary medullary thyroid cancer (MTC) is inherited in an autosomal dominant fashion but may also arise de novo. Commonly, MTC is the first clinical manifestation of multiple endocrine neoplasia type 2 (MEN2). Intriguingly, the speed of progression from neoplastic C cell disease to node-negative and node-positive MTC is proportional to the transforming potential of the respective germline mutations in the RET (REarranged during Transfection) proto-oncogene [1–4]. Early diagnosis of MTC, facilitated by calcitonin and RET screening, is associated with improved oncologic outcome [5, 6]. The RET gene is located on chromosome 10q11.2 and codes for a transmembrane RET tyrosine kinase receptor that is essential for proliferation, differentiation, and survival of central and peripheral nerve lineages [7]. Constitutive activation of the RET tyrosine kinase receptor drives the malignant transformation of thyroid C cells in a genotype-specific manner, yielding close genotype-phenotype correlations [1, 2, 4] which open a surgical window of opportunity [8, 9]. The 2015 revised American Thyroid Association guidelines for the management of medullary thyroid carcinoma [3], enhancing the 2001 international consensus guidelines for diagnosis and therapy of MEN type 1 and type 2 [10] and the 2012 European Thyroid Association guidelines for genetic testing and its clinical consequences in medullary thyroid cancer [11] and replacing the 2009 medullary thyroid cancer management guidelines of the American Thyroid Association [12], categorize RET mutations into highest risk (RET p.Met918Thr, producing MEN2B), high risk (RET p.Cys634, predominantly associated with MEN2A) and moderate risk mutations. Subsequent research [4] published in 2018 uncovered that moderate risk mutations are a fairly heterogeneous group of mutations that can be subdivided further into moderate-high (RET p.Cys609/611/618/620/630, associated with MEN2A from time to time) and low-moderate risk mutations (RET p.Glu768Asp, p.Leu790Phe, p.Val804Leu/Met and p.Ser891Ala, infrequently associated with MEN2A). To deplete the pool of unrecognized RET carriers in the general population and enable early pre-emptive thyroidectomy in offspring, DNA-based screening would need to be targeted at sporadic-appearing MTC [13]. In accordance with this concept, systematic genetic screening would eventually (i) eliminate the need for neck reoperations for persistent disease by making the diagnosis earlier, (ii) decrease the number of index patients in the general population, and (iii) reduce the presence of MTC at preemptive thyroidectomy. Thirty years into the genomic era, it remains unclear to what extent these goals have been achieved and which events helped transform the clinical landscape of hereditary MTC. Because not all mutations are created equal, the expected health benefits may not have developed in synchrony across all four RET risk categories. The present investigation, taking stock of temporal trends in the past four decades, was undertaken to address these research questions. Patients and methods Study population This research was based on retrospective evaluation of prospectively collected data on all RET carriers who were referred to a tertiary surgical center for neck surgery that was carried out between 1986 and 2021. All carriers of RET mutations (http://www.arup.utah.edu/database/MEN2/MEN2_display.php) [14] provided written informed consent before undergoing neck surgery at the authors’ institution between 1986 and 2021 [13, 15, 16]. All procedures complied with national and international clinical standards applicable at the time of surgery [3, 10–12] and were in accordance with the amended Declaration of Helsinki and relevant local rules (institutional review board approval reference 2020-237). At the time of first referral to the authors’ institution, operative status was determined for each RET carrier: referral for thyroidectomy or referral for neck reoperation. Subsequent reoperations at the authors’ institution were disregarded so that each RET carrier was counted only once. The diagnosis of MTC was based on histopathologic demonstration of cells immunoreactive with calcitonin extending beyond the basement membrane [17]. RET mutations were subdivided into highest, high, moderate-high or low-moderate risk mutations, as described above [4]. Statistical analysis For statistical analysis, the software package SPSS® version 28 (IBM, Armonk, New York, USA) was used. Categorical and continuous data were tested with the two-tailed Fisher’s exact test and the two-tailed Kruskal-Wallis rank sum test, respectively, to determine differences among RET risk categories, referral intervals, and clinical variables. Multiple testing was corrected for with the Bonferroni method. In doing so, observed (nominal) P values were adjusted by multiplying these by the number of statistical comparisons within each set of tests addressing the same research question [18]. Adjusted P values were compared to the significance level alpha = 0.05. To assess the impact of events on subsequent changes in referral patterns, Poisson regression analysis was performed using R version 4.1.2 [19]. As predictors were considered the number of years, taking 1986 as year 1, and the number of years elapsed since each event. For instance, when an event occurred in 1993, 1994 was taken as post-event year 1. These predictors were included in the Poisson regression model using multivariable fractional polynomials [20] to allow for non-linearity in the data. Backward selection with the Akaike information criterion (AIC) was applied to remove weak predictors and to select variables for inclusion in the final Poisson regression model. These analyses were repeated using forward selection to evaluate the sensitivity of the regression model. Results Baseline characteristics of the study population Altogether, 496 RET carriers were referred for thyroidectomy (388 carriers) or neck reoperation (108 carriers) between 1986 and 2021 (Table 1). Of these, 44 carriers had highest risk mutations (p.Met918Thr), 164 carriers high risk mutations (p.Cys634Arg/Gly/Phe/Ser/Trp/Tyr/insHisGluLeuCys), 116 carriers moderate–high risk mutations (p.Cys609/611/618/620/630Arg/Gly/Phe/Ser/Tyr) and 172 carriers low–moderate risk mutations (p.Glu768Asp, p.Leu790Phe, p.Val804Leu/Met, or p.Ser891Ala).Table 1 Baseline characteristics of the study population referred to the authors’ institution in 1986–2021 RET category Highest risk High risk Moderate–high Low–moderate P* RET mutation p.Met918Thr p.Cys634Arg/Gly/ Phe/Ser/Trp/Tyr/ insHisGluLeuCys p.Cys609/611/618/620/630Arg/Gly/Phe/Ser/Tyr p.Glu768Asp; p.Leu790Phe; p.Val804Leu/Met; p.Ser891Ala No. of carriers (496 in total) 44 164 116 172 Sex, male, n 23 (52) 75 (46) 46 (40) 73 (42) 0.482 Index patients, n 37 (84) 34 (21) 24 (21) 54 (31) <0.001 Carriers with initial thyroidectomy at the authors’ institution, n 30 (68) 129 (79) 90 (78) 139 (81) 0.348 Age at thyroidectomy, yr, median [IQR] 12 [3; 17] 10.5 [5; 30.8] 21.5 [7; 37.8] 38 [20.3; 55] <0.001 Carriers with medullary thyroid cancer, n 42 (95) 114 (70) 67 (58) 102 (59) <0.001 Numbers in parentheses denote percentages unless noted otherwise Ins insertion, IQR interquartile range (in brackets), Ala alanine, Arg arginine, Asp aspartic acid, Cys cysteine, Glu glutamic acid, Gly glycine, His histidine, Leu leucine, Met methionine, Phe phenylanalnine, Ser serine, Trp tryptophan, Thr threonine Nominal (unadjusted) P value Statistically significant after Bonferroni correction for multiple testing adjusting for all five comparisons Significant differences across RET categories (from highest to low–moderate risk) were observed in referrals regarding the proportion of index patients (84% in the highest RET category vs. 21–31% in the other three RET categories; P < 0.001), age at initial thyroidectomy (medians of 12, 10.5, 21.5 and 38 years; P < 0.001), and presence of MTC at thyroidectomy (95, 70, 58, and 59%; P < 0.001). No significant differences were noted for sex (279 female versus 217 male carriers) or in respect to whether thyroidectomy had been performed at the authors’ institution (388 carriers) or elsewhere (108 carriers). Seminal events and referral of carriers to the authors’ institution by year of referral Figure 1 chronicles the ups and downs of the annual number of referrals to the authors’ institution between 1986 and 2021 (no referrals took place in 1987 and 1992).Fig. 1 Seminal events and referral of RET carriers to the authors’ institution by year of referral. A Operative status. B Index status. C Medullary thyroid cancer. White circles denote first mutational reports, black circles other high-end RET publications, and arrows international management guidelines, with numbers indicating references. ATA American Thyroid Association, ETA European Thyroid Association Altogether, there were three event clusters that drove referral numbers upward (Fig. 1):The first seminal event comprised a series of genomic discoveries: two first reports from competing research groups [21, 22] in 1993 demonstrating that high and moderate–high RET mutations caused MEN2A and familial MTC (p.Cys634 and p.Cys609/611/618/620/630)—at a time when referral numbers to the authors’ institution hit a record low (first nadir). This was followed by another first report in 1994 that found that highest risk RET mutations (p.Met918Thr) caused MEN2B [23] and a string of first reports from other institutions revealing that low–moderate risk RET mutations caused familial MTC in 1995 (p.Glu768Asp and p.Val804Leu) [24, 25], 1996 (p.Val804Met) [26], 1997 (p.Ser891Ala) [27], and 1998 (p.Leu790Phe) [28]. This rise in referral numbers was buttressed by high-end publications that appeared in 1994 [29, 30] and 1996 [1]. The second seminal event was the publication of the international consensus guidelines for diagnosis and therapy of MEN type 1 and type 2 [10] in 2001 – at a time when referral numbers to the authors’ institution were down again, though on a higher level (second nadir). The subsequent surge in referrals to a record high at the authors’ institution in 2007 was fueled by two high-ranking publications in 2003 [2] and 2005 [31]. Interestingly, publication of the medullary thyroid cancer management guidelines of the American Thyroid Association [12] in 2009 may have stemmed the accelerating decline of referral numbers that began taking up pace in 2008. In this endeavor, the 2012 European Thyroid Association guidelines for genetic testing and its clinical consequences in medullary thyroid cancer [11] may have been less successful. The third seminal event was the publication of the revised American Thyroid Association guidelines for the management of medullary thyroid carcinoma [3] in 2015—at a time when referral numbers to the authors’ institution hit another record low (the third nadir). These associations were explored further using Poisson regression analysis. The Akaike information criterion, striking a balance between the risks of overfitting and underfitting, did not permit inclusion of the 2009 (medullary thyroid cancer management guidelines of the American Thyroid Association [12]) and 2012 (European Thyroid Association guidelines for genetic testing and its clinical consequences in medullary thyroid cancer [11]) events, neither in the backward nor in the forward Poisson regression model. The final Poisson regression hence focused on the 1993 (first description of the RET proto-oncogene as causative of hereditary MTC and MEN2) [21, 22], 2001 (international consensus guidelines for diagnosis and therapy of MEN type 1 and type 2 [10]) and 2015 (revised American Thyroid Association guidelines for the management of medullary thyroid carcinoma) [3] events. Each of these three events, showing statistically significant effects (P < 0.001) when formally explored, was followed by an abrupt surge in referrals of RET carriers to the authors’ institution. Figure 2, plotting the number of observed against the number of expected referrals of RET carriers to the authors’ institution per year, visualizes these boosts in referrals setting in immediately after the 1993, 2001 and 2015 events.Fig. 2 Number of observed annual referrals of RET carriers versus number of expected annual referrals of RET carriers (final Poisson regression model). Events not included in the final Poisson regression model are represented by dashed vertical lines. ATA American Thyroid Association, ETA European Thyroid Association Referral of carriers with certain clinical features to the authors’ institution by year of referral Figure 1 depicts time trends of referral numbers for each year of referral to the authors’ institution from 1986 to 2021 (no referrals took place in 1987 and 1992). By and large, referral numbers for neck reoperations paralleled total referral numbers over the years, but in recent years were on the wane (Fig. 1A). After 30 years of sluggish decline (1988–2018), referrals for reoperation came to an end, as illustrated by a complete lack of referrals for neck reoperation in 2019, 2020, and 2021. Congruent with this, index patients (Fig. 1B) and carriers with MTC (Fig. 1C) displayed fairly similar referral patterns, although more carriers were affected by these conditions. Index patients, most recently a 3-year-old girl with C cell hyperplasia alone who was found to carry the highest risk p.Met918Thr RET mutation during work-up for stunted growth, continued to be referred to the authors’ institution into 2021 (Fig. 1B). Referral of carriers to the authors’ institution by RET category and referral period When referrals were grouped in 5-year increments (Table 2), the decline of referrals for reoperation was statistically significant for carriers of high, moderate–high and low–moderate risk mutations (P < 0.001), but missed statistical significance for carriers of highest risk mutations.Table 2 Referrals of all 496 RET carriers for thyroidectomy or reoperation to the authors’ institution in 1986–2021 Referrals for thyroidectomy and neck reoperation to the authors’ institution, no. (%) of RET carriers RET risk category (no. of carriers) Year of surgery 1986–1990 1991–1995 1996–2000 2001–2005 2006–2010 2011–2015 2016–2020 2021 P* No. of carriers 15 (3) 30 (6) 108 (22) 90 (18) 118 (24) 65 (13) 67 (14) 3 (1) Highest (44) Thyroidectomy 1 (50) 8 (73) 7 (88) 6 (75) 3 (60) 4 (67) 1 (100) 0.266 Reoperation 3 (100) 1 (50) 3 (27) 1 (12) 2 (25) 2 (40) 2 (33) High (164) Thyroidectomy 4 (50) 8 (40) 28 (74) 25 (86) 26 (87) 18 (95) 18 (100) 2 (100) <0.001 Reoperation 4 (50) 12 (60) 10 (26) 4 (14) 4 (13) 1 (5) Moderate-high (116) Thyroidectomy 2 (50) 2 (40) 20 (65) 20 (91) 26 (90) 6 (55) 14 (100) <0.001 Reoperation 2 (50) 3 (60) 11 (35) 2 (9) 3 (10) 5 (45) Low-moderate (172) Thyroidectomy 23 (82) 23 (74) 40 (78) 25 (83) 28 (97) 0.006** Reoperation 3 (100) 5 (18) 8 (26) 11 (22) 5 (17) 1 (3) Total (496) Thyroidectomy (388) 6 (40) 11 (37) 79 (73) 75 (83) 98 (83) 52 (80) 64 (96) 3 (100) Reoperation (108) 9 (60) 19 (63) 29 (27) 15 (17) 20 (17) 13 (20) 3 (4) Numbers designate RET carriers, whereas numbers in parentheses denote percentages; empty fields denote absence of referrals; owing to rounding, not all percentages match up Nominal (unadjusted) P value Statistically significant after Bonferroni correction for multiple testing adjusting for all four comparisons Referral of carriers to the authors’ institution by reoperation, RET category and referral period Evaluating all annual referrals together can provide valuable clues but may obscure differential referral patterns for carriers who differ in operative status, index status, and thyroid histopathology and harbor RET mutations that confer disparate risks. To control for such confounders, Fig. 3 stratifies 5-year referral patterns of index patients (Fig. 3A), carriers with MTC (Fig. 3B), and age at thyroidectomy (Fig. 3C) by operative status and RET risk category.Fig. 3 Referral of carriers to the authors’ institution by reoperation, RET category and referral period. A Index status. B Medullary thyroid cancer. C Age at thyroidectomy As pictured in Fig. 3, increases in referrals for thyroidectomy peaked in 1996–2000 for carriers of highest and high risk mutations, and in 2006–2010 for carriers of moderate–high and low–moderate risk mutations, some 10 years later. Corresponding trends, although attenuated, were present also in referrals for reoperations, involving mainly index patients (Fig. 3A) and almost always carriers with MTC (Fig. 3B). The 108 carriers referred for reoperation were considerably older at outside thyroidectomy than the 388 carriers referred for thyroidectomy to the authors’ institution (Fig. 3C). Likewise, the spread in age at thyroidectomy across RET risk categories was greater in the former group who were older than in the latter group of younger carriers who were diagnosed earlier. When formally tested, the above referral trends were consistently significant for referrals of carriers of high risk mutations for thyroidectomy (Table 3), and consistently negative for referrals for reoperation (Table 4).Table 3 Referral of all 388 RET carriers to the authors’ institution for thyroidectomy in 1986–2021 Year of first neck surgery at authors’ institution (388 carriers) Variable RET risk category 1986–1990 1991–1995 1996–2000 2001–2005 2006–2010 2011–2015 2016–2020 2021 P No. of carriers 6 (2) 11 (3) 79 (20) 75 (19) 98 (25) 52 (13) 64 (16) 3 (1) No. of index patients among all carriers Highest (30) 1 of 1 (100) 8 of 8 (100) 6 of 7 (86) 5 of 6 (83) 1 of 3 (33) 2 of 4 (50) 0 of 1 (0) 0.129 High (129) 2 of 4 (50) 1 of 8 (13) 8 of 28 (29) 0 of 25 (0) 1 of 26 (4) 1 of 18 (6) 2 of 18 (11) 0 of 2 (0) 0.007 Moderate-high (90) 0 of 2 (0) 0 of 2 (0) 2 of 20 (10) 1 of 20 (5) 4 of 26 (15) 0 of 6 (0) 1 of 14 (7) 0.865 Low-moderate (139) 4 of 23 (17) 2 of 23 (9) 6 of 40 (15) 3 of 25 (12) 9 of 28 (32) 0.244 No. of carriers with medullary thyroid cancer among all carriers Highest (30) 1 of 1 (100) 8 of 8 (100) 7 of 7 (100) 6 of 6 (100) 2 of 3 (67) 4 of 4 (100) 0 of 1 (0) 0.023 High (129) 4 of 4 (100) 5 of 8 (63) 26 of 28 (93) 18 of 25 (72) 11 of 26 (42) 7 of 18 (39) 10 of 18 (56) 0 of 2 (0) <0.001 Moderate-high (90) 2 of 2 (100) 0 of 2 (0) 8 of 20 (40) 10 of 20 (50) 16 of 26 (62) 4 of 6 (67) 2 of 14 (14) 0.024 Low-moderate (139) 15 of 23 (56) 8 of 23 (35) 23 of 40 (58) 10 of 25 (40) 16 of 28 (57) 0.296 Age at thyroidectomy at authors’ institution in years, median (no. of carriers) [IQR] Highest (30) 16 (1) 16 (8) [9.3; 29.8] 12 (7) [1.5; 14] 10 (6) [1.3; 17.8] 0.6 (3) [0.5; N/A] 2 (4) [0.6; 10.5] 3 (1) N/A High (129) 34 (4) [22; 39.3] 11 (8) [7.8; 20.3] 25 (28) [7; 38.5] 6 (25) [4; 24] 5 (26) [3; 6.3] 3.5 (18) [2; 6.3] 7.5 (18) [3.8; 38.5] 3.5 (2) [3; N/A] <0.001** Moderate-high (90) 38 (2) [36; N/A] 15.5 (2) [13; N/A] 10.5 (20) [7; 21.3] 28.5 (20) [5.3; 42.5] 19.5 (26) [3.8, 36] 11.5 (6) [5; 25.5] 6.5 (14) [5: 10.5] 0.170 Low-moderate (139) 33 (23) [8; 47] 17 (23) [7; 33] 39.5 (40) [25.5; 55] 33 (25) [20; 45] 37.5 (28) [14.3; 58.8] 0.020 IQR interquartile range (in brackets), N/A not assessable due to inadequate number of observations Numbers designate RET carriers, whereas numbers in parentheses denote percentages, unless indicated otherwise; empty fields denote absence of referrals; owing to rounding, not all percentages match up Nominal (unadjusted) P value Statistically significant after Bonferroni correction for multiple testing adjusting for four comparisons within each variable category Table 4 Referral of all 108 RET carriers to the authors’ institution for neck reoperation in 1986–2021 Variable Year of first neck surgery at authors’ institution (108 carriers) RET risk category 1986–1990 1991–1995 1996–2000 2001–2005 2006–2010 2011–2015 2016–2020 2021 P No. of carriers 9 (8) 19 (18) 29 (27) 15 (14) 20 (19) 13 (12) 3 (3) No. of index patients among all carriers Highest (14) 3 of 3 (100) 1 of 1 (100) 3 of 3 (100) 1 of 1 (100) 1 of 2 (50) 2 of 2 (100) 2 of 2 (100) 0.571 High (35) 3 of 4 (75) 3 of 12 (25) 7 of 10 (70) 2 of 4 (50) 3 of 4 (75) 1 of 1 (100) 0.178 Moderate-high (26) 1 of 2 (50) 1 of 3 (33) 7 of 11 (64) 1 of 2 (50) 3 of 3 (100) 3 of 5 (60) 0.709 Low-moderate (33) 3 of 3 (100) 3 of 5 (60) 7 of 8 (88) 11 of 11 (100) 5 of 5 (100) 1 of 1 (100) 0.177 No. of carriers with medullary thyroid cancer among all carriers Highest (14) 3 of 3 (100) 1 of 1 (100) 3 of 3 (100) 1 of 1 (100) 2 of 2 (100) 2 of 2 (100) 2 of 2 (100) N/A High (35) 4 of 4 (100) 10 of 12 (83) 10 of 10 (100) 4 of 4 (100) 4 of 4 (100) 1 of 1 (100) 0.789 Moderate-high (26) 2 of 2 (100) 3 of 3 (100) 11 of 11 (100) 2 of 2 (100) 3 of 3 (100) 4 of 5 (80) 0.577 Low-moderate (33) 3 of 3 (100) 4 of 5 (80) 8 of 8 (100) 11 of 11 (100) 5 of 5 (100) 1 of 1 (100) 0.424 Age at thyroidectomy at outside institution in years, median (no. of carriers) [IQR] Highest (14) 17 (3) [6; N/A] 18 (1) 9 (3) [5; N/A] 3 (1) 8.5 (2) [3; N/A] 17.5 (2) [16; N/A] 28.5 (2) [27; N/A] 0.063 High (35) 35 (4) [24.7; 38.5] 21.5 (12) [12.5; 37] 29 (10) [18; 53.8] 37 (4) [18; 38.8] 26 (4) [21; 41.5] 50 (1) 0.606 Moderate-high (26) 37.5 (2) [37; N/A] 26 (3) [15; N/A] 40 (11) [27; 52] 26 (2) [24; N/A] 42 (3) [38, N/A] 32 (5) [21; 61] N/A Low-moderate (33) 56 (3) [47; N/A] 60 (5) [46; 68.5] 49.5 (8) [45; 56.3] 53 (11) [47; 65] 45 (5) [33; 72] 49 (1) N/A IQR interquartile range (in brackets), N/A not assessable due to inadequate number of observations Numbers designate RET carriers, whereas numbers in parentheses denote percentages, unless indicated otherwise; empty fields denote absence of referrals; owing to rounding, not all percentages match up *Nominal (unadjusted) P value Discussion This is the first investigation to explore the impact of international consensus and professional management guidelines on referrals of RET carriers to a tertiary surgical center before and after genomic testing became available in the 1990s. Following a string of first reports linking RET missense mutations to MEN2 and MTC, international management guidelines on MEN2 and MTC highlighted and emphasized the importance of genetic screening of patients with sporadic-appearing MTC and offspring of known RET families, reaching a wider, more diverse clinical audience and sparking off two further waves of referrals (Fig. 2). Because they are more likely to be the first family member to present for genetic counseling, carriers with early-onset disease typically become the index patient who brings the family under study. As a rule, carriers of highest and high risk RET mutations meet the criteria for family cascade testing more frequently than carriers of moderate–high and low–moderate risk RET mutations. Because of this, the former group of patients was referred more often with the first wave, whereas the latter group of patients was more commonly referred with the second and third wave (Fig. 3A, B). A disproportionate number of carriers of highest risk mutations, 37 of 44 carriers or 84% (Table 1), were index patients carrying de novo RET p.Met918Thr mutations. Owing to the burden of disease, many index patients, developing MEN2B in infancy and early childhood, will not pass on the trait to future generations. This limits the pool of nonindex patients inheriting highest risk RET mutations, thwarting efforts by family cascade testing to spot these mutations earlier. In spite of this, RET screening is crucial to make the diagnosis early, both at the level of the individual and the level of the population. The clinical benefits of systematic calcitonin screening before (pre-1993) and systematic molecular screening in the genomic era (starting in 1993), notably reductions in the number of index patients, in the number of patients with MTC and in age at thyroidectomy, were greater among carriers of high risk mutations than among carriers of moderate risk mutations (Table 3). This finding is explained by the fact that carriers of highest risk mutations, which are more penetrant, acquired the trait frequently from a clinically affected MEN2A parent, whereas carriers of moderate risk mutations, which are less penetrant, inherited the trait often from a mildly affected parent. For many RET carriers, timely thyroidectomy alone will be adequate treatment, sparing them additional node dissection with incremental morbidity coming with the procedure [8]. Early preemptive thyroidectomy also helps diminish neck reoperations, which carry even greater operative risks. In this respect, it may be interesting to note that referrals to the authors’ institution for neck reoperations petered out recently (Fig. 1A). The present work, breaking new ground, has strengths and weaknesses. A key asset were the large number of almost 500 RET carriers referred to, and carefully documented at the authors’ institution over a 36-year period, which resulted in complete data sets. Of note, all key performance indicators – operative status at referral, index or nonindex patient status, presence of MTC at thyroidectomy, year of and age at thyroidectomy, and year of referral – were prospectively documented and are robust to temporal change. In view of the long observation period, it cannot be ruled out that referrals to the authors’ institution may have been subject to change in the course of time. Remarkably, the 2020 climax of the Covid-19 pandemic in Germany increased, rather than decreased, the number of referrals to the authors’ institution where patients in need of prophylactic or therapeutic cancer surgery continued to be prioritized. Given the senior author’s longstanding expertise in prophylactic thyroidectomy [32] and compartment-oriented node dissection [33], intervening changes in referrals of RET carriers, if any, may have skewed the disease spectrum at the authors’ institution towards advanced MTC. If so, the present research may underestimate the size of the impact of seminal events on the presentation of hereditary MTC and the magnitude of health benefits associated with routine RET screening. Having said that, cause (publication) and effect (promptly being acted on by the medical community) can be implied, but not proven without a shadow of doubt. This is a challenge inherent in any public health care research. In the era of the internet, though, it is reasonable to presuppose expeditious uptake of seminal publications that appear in computerized biomedical bibliographic retrieval systems upon acceptance. Notoriously hard, if not impossible, to disentangle are additive effects that may be reinforcing referrals to specialist surgical centers. For instance, two key publications, appearing in the New England Journal of Medicine in 2003 [2] and 2005 [31], arguably may have helped sustain the second wave of referrals to the authors’ institution, which set off with the 2001 publication of the international consensus guidelines for diagnosis and therapy of MEN type 1 and type 2 [10]. The present study of 496 MEN2 gene carriers holds useful learning lessons for other monogenic disease. Once genomic screening becomes available, carriers presenting with early onset and more severe disease will be targeted – and referred for pre-emptive surgery as appropriate – in the first place, whereas carriers presenting with late onset and indolent disease stand to be lost sight of at the outset. The international community is called on to quickly develop and promote international management guidelines encompassing the complete spectrum of genetic disease to build and increase the pressure throughout the entire health care system towards screening of sporadic-appearing disease and offspring of known gene families early on. Author contributions A.M.: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Writing—original draft. K.L.: Investigation, Writing—review & editing. E.-M.H.: Poisson regression modeling using multivariable fractional polynomials, Writing—review & editing. A.S.: Poisson regression modeling using multivariable fractional polynomials, Writing—review & editing. F.W: Investigation, Writing—review & editing. H.D.: Conceptualization, Methodology, Investigation, Writing—review & editing, Supervision. Compliance with ethical standards Conflict of interest The authors declare no competing interests. To err on the side of transparency, Andreas Machens and Henning Dralle wish to disclose that they served as unpaid members on the American Thyroid Association Guidelines Task Force on Medullary Thyroid Carcinoma which wrote and authored the 2015 Revised American Thyroid Association Guidelines for the Management of Medullary Thyroid Carcinoma (reference #3). Ethics 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. Consent to participate Informed consent was obtained from each patient and/or legal guardian as applicable before each RET gene test and each operation, all of which represented standard of care. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. C. Eng, D. Clayton, I. Schuffenecker, G. Lenoir, G. Cote, R.F. Gagel, H.K. van Amstel, C.J. Lips, I. Nishisho, S.I. Takai, D.J. Marsh, B.G. Robinson, K. Frank-Raue, F. Raue, F. Xue, W.W. Noll, C. Romei, F. Pacini, M. Fink, B. Niederle, J. Zedenius, M. Nordenskjöld, P. Komminoth, G.N. Hendy, H. Gharib, S.N. Thibodeau, A. Lacroix, A. Frilling, B.A.J. Ponder, L.M. Mulligan, The relationship between specific RET proto-oncogene mutations and disease phenotype in multiple endocrine neoplasia type 2. International RET mutation consortium analysis. J. Am. Med. Assoc. 276, 1575–1579 (1996) 2. Machens A Niccoli-Sire P Hoegel J Frank-Raue K van Vroonhoven TJ Roeher HD Wahl RA Lamesch P Raue F Conte-Devolx B Dralle H European Multiple Endocrine Neoplasia (EUROMEN) Study Group, Early malignant progression of hereditary medullary thyroid cancer N. Engl. J. Med. 2003 349 1517 1525 10.1056/NEJMoa012915 14561794 3. Wells SA Jr Asa SL Dralle H Elisei R Evans DB Gagel RF Lee N Machens A Moley JF Pacini F Raue F Frank-Raue K Robinson B Rosenthal MS Santoro M Schlumberger M Shah M Waguespack SG American Thyroid Association Guidelines Task Force on Medullary Thyroid Carcinoma, Revised American Thyroid Association guidelines for the management of medullary thyroid carcinoma Thyroid 2015 25 567 610 10.1089/thy.2014.0335 25810047 4. Machens A Lorenz K Weber F Dralle H Genotype-specific progression of hereditary medullary thyroid cancer Hum. Mutat. 2018 39 860 869 10.1002/humu.23430 29656518 5. Elisei R Bottici V Luchetti F Di Coscio G Romei C Grasso L Miccoli P Iacconi P Basolo F Pinchera A Pacini F Impact of routine measurement of serum calcitonin on the diagnosis and outcome of medullary thyroid cancer: experience in 10,864 patients with nodular thyroid disorders J. Clin. Endocrinol. Metab. 2004 89 163 168 10.1210/jc.2003-030550 14715844 6. Machens A Dralle H Biomarker-based risk stratification for previously untreated medullary thyroid cancer J. Clin. Endocrinol. Metab. 2010 95 2655 2663 10.1210/jc.2009-2368 20339026 7. Mulligan LM YEARS OF THE DOUBLE HELIX: Exploiting insights on the RET receptor for personalized cancer medicine Endocr. Relat. Cancer 2018 25 T189 T200 10.1530/ERC-18-0141 29743166 8. Machens A Elwerr M Lorenz K Weber F Dralle H Long-term outcome of prophylactic thyroidectomy in children carrying RET germline mutations Br. J. Surg. 2018 105 e150 e157 10.1002/bjs.10746 29341155 9. Prete FP Abdel-Aziz T Morkane C Brain C Kurzawinski TR MEN2 in Children UK Collaborative Group, Prophylactic thyroidectomy in children with multiple endocrine neoplasia type 2 Br. J. Surg. 2018 105 1319 1327 10.1002/bjs.10856 29663329 10. Brandi ML Gagel RF Angeli A Bilezikian JP Beck-Peccoz P Bordi C Conte-Devolx B Falchetti A Gheri RG Libroia A Lips CJ Lombardi G Mannelli M Pacini F Ponder BA Raue F Skogseid B Tamburrano G Thakker RV Thompson NW Tomassetti P Tonelli F Wells SA Jr Marx SJ Guidelines for diagnosis and therapy of MEN type 1 and type 2 J. Clin. Endocrinol. Metab. 2001 86 5658 5671 10.1210/jcem.86.12.8070 11739416 11. R. Elisei, M. Alevizaki, B. Conte-Devolx, K. Frank-Raue, V. Leite, G.R. Williams, 2012 European thyroid association guidelines for genetic testing and its clinical consequences in medullary thyroid cancer. Eur. Thyroid J. 1, 216–231 (2012) 12. Kloos RT Eng C Evans DB Francis GL Gagel RF Gharib H Moley JF Pacini F Ringel MD Schlumberger M Wells SA Jr. Medullary thyroid cancer: management guidelines of the American Thyroid Association Thyroid 2009 19 565 612 10.1089/thy.2008.0403 19469690 13. Machens A Dralle H Therapeutic effectiveness of screening for multiple endocrine neoplasia type 2A J. Clin. Endocrinol. Metab. 2015 100 2539 2545 10.1210/jc.2015-1689 25946031 14. Margraf RL Crockett DK Krautscheid PM Seamons R Calderon FR Wittwer CT Mao R Multiple endocrine neoplasia type 2 RET protooncogene database: repository of MEN2-associated RET sequence variation and reference for genotype/phenotype correlations Hum. Mutat. 2009 30 548 556 10.1002/humu.20928 19177457 15. Dralle H Scheumann GF Kotzerke J Brabant EG Surgical management of MEN 2 Recent Results Cancer Res 1992 125 167 195 10.1007/978-3-642-84749-3_9 1360170 16. Machens A Lorenz K Weber F Dralle H Sex differences in MEN 2A penetrance and expression according to parental inheritance Eur. J. Endocrinol. 2022 186 469 476 10.1530/EJE-21-1086 35130180 17. R.A. DeLellis, R.V. Lloyd, P.U. Heitz, C. Eng (eds), Classification of tumours of endocrine organs. World Health Organisation (2017). 18. Bland JM Altman DG Multiple significance tests: the Bonferroni method Br. Med. J. 1995 310 170 10.1136/bmj.310.6973.170 7833759 19. R Core Team, R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria (2021); https://www.R-project.org/ 20. Sauerbrei W Royston P Building multivariable prognostic and diagnostic models: transformation of the predictors by using fractional polynomials. J. R. Statist Soc. A 1999 162 71 94 21. Donis-Keller H Dou S Chi D Carlson KM Toshima K Lairmore TC Howe JR Moley JF Goodfellow P Wells SA Jr. Mutations in the RET proto-oncogene are associated with MEN 2A and FMTC Hum. Mol. Genet. 1993 2 851 856 10.1093/hmg/2.7.851 8103403 22. Mulligan LM Kwok JB Healey CS Elsdon MJ Eng C Gardner E Love DR Mole SE Moore JK Papi L Ponder MA Telenius H Tunnacliffe A Ponder BAJ Germ-line mutations of the RET proto-oncogene in multiple endocrine neoplasia type 2A Nature 1993 363 458 460 10.1038/363458a0 8099202 23. Hofstra RM Landsvater RM Ceccherini I Stulp RP Stelwagen T Luo Y Pasini B Höppener JW van Amstel HK Romeo G Lips CJM Buys HCM A mutation in the RET proto-oncogene associated with multiple endocrine neoplasia type 2B and sporadic medullary thyroid carcinoma Nature 1994 367 375 376 10.1038/367375a0 7906866 24. Eng C Smith DP Mulligan LM Healey CS Zvelebil MJ Stonehouse TJ Ponder MA Jackson CE Waterfield MD Ponder BA A novel point mutation in the tyrosine kinase domain of the RET proto-oncogene in sporadic medullary thyroid carcinoma and in a family with FMTC Oncogene 1995 10 509 513 7845675 25. Bolino A Schuffenecker I Luo Y Seri M Silengo M Tocco T Chabrier G Houdent C Murat A Schlumberger M Tourniaire L Lenoir GM Romeo G RET mutations in exons 13 and 14 of FMTC patients Oncogene 1995 10 2415 2419 7784092 26. Fink M Weinhäusel A Niederle B Haas OA Distinction between sporadic and hereditary medullary thyroid carcinoma (MTC) by mutation analysis of the RET proto-oncogene. “Study Group Multiple Endocrine Neoplasia Austria (SMENA)” Int. J. Cancer 1996 69 312 316 10.1002/(SICI)1097-0215(19960822)69:4<312::AID-IJC13>3.0.CO;2-7 8797874 27. Hofstra RM Fattoruso O Quadro L Wu Y Libroia A Verga U Colantuoni V Buys CH A novel point mutation in the intracellular domain of the ret protooncogene in a family with medullary thyroid carcinoma J. Clin. Endocrinol. Metab. 1997 82 4176 4178 9398735 28. Berndt I Reuter M Saller B Frank-Raue K Groth P Grussendorf M Raue F Ritter MM Höppner W A new hot spot for mutations in the ret protooncogene causing familial medullary thyroid carcinoma and multiple endocrine neoplasia type 2A J. Clin. Endocrinol. Metab. 1998 83 770 774 9506724 29. Lips CJ Landsvater RM Höppener JW Geerdink RA Blijham G van Veen JM van Gils AP de Wit MJ Zewald RA Berends MJ Beemer FA Brouwers-Smalbraak J Jansen R Ploos van Amstel HK van Vroonhoven T Vroom TM Clinical screening as compared with DNA analysis in families with multiple endocrine neoplasia type 2A N. Engl. J. Med. 1994 331 828 835 10.1056/NEJM199409293311302 7915822 30. Wells SA Jr. Chi DD Toshima K Dehner LP Coffin CM Dowton SB Ivanovich JL DeBenedetti MK Dilley WG Moley JF Norton JA Donis-Keller H Predictive DNA testing and prophylactic thyroidectomy in patients at risk for multiple endocrine neoplasia type 2A Ann. Surg. 1994 220 237 247 10.1097/00000658-199409000-00002 7916559 31. Skinner MA Moley JA Dilley WG Owzar K Debenedetti MK Wells SA Jr. Prophylactic thyroidectomy in multiple endocrine neoplasia type 2A N. Engl. J. Med. 2005 353 1105 1113 10.1056/NEJMoa043999 16162881 32. Dralle H Gimm O Simon D Frank-Raue K Görtz G Niederle B Wahl RA Koch B Walgenbach S Hampel R Ritter MM Spelsberg F Heiss A Hinze R Höppner W Prophylactic thyroidectomy in 75 children and adolescents with hereditary medullary thyroid carcinoma: German and Austrian experience World J. Surg. 1998 22 744 450 10.1007/s002689900463 9606292 33. Dralle H Damm I Scheumann GF Kotzerke J Kupsch E Geerlings H Pichlmayr R Compartment-oriented microdissection of regional lymph nodes in medullary thyroid carcinoma Surg. Today 1994 24 112 121 10.1007/BF02473391 8054788	36456885	PMC9715418	NO-CC CODE	2022-12-03 23:20:15	no	Endocrine. 2022 Dec 2;:1-11	utf-8	Endocrine	2,022	10.1007/s12020-022-03273-8	oa_other
==== Front Process Integr Optim Sustain Process Integration and Optimization for Sustainability 2509-4238 2509-4246 Springer Nature Singapore Singapore 298 10.1007/s41660-022-00298-4 Original Research Paper Green-Resilient Supplier Selection and Order Allocation Under Disruption by Utilizing Conditional Value at Risk: Mixed Response Strategies Taghavi Seyed Mojtaba 1 http://orcid.org/0000-0001-8188-5426 Ghezavati Vahidreza [email protected] [email protected] 1 Bidhandi Hadi Mohammadi 1 Al-e-Hashem Seyed Mohammad Javad Mirzapour 2 1 grid.411463.5 0000 0001 0706 2472 Department of Industrial Engineering, South Tehran Branch, Islamic Azad University, Tehran, Iran 2 grid.411368.9 0000 0004 0611 6995 Department of Industrial Engineering and Management Systems, Amirkabir University of Technology, Tehran, Iran 2 12 2022 122 2 9 2022 6 11 2022 10 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. Two important decisions in supply chains and logistics systems design are the supplier selection and order allocation (SS&OA) problem and the vehicle routing problem (VRP). Supply disruption may reduce the capacity of suppliers, and the transportation network disruption may decrease the number of vehicles in the fleet and disrupt some routes. Also, increasing environmental regulations and environmental awareness makes companies pay more attention to green supply chain management (GSCM). In this paper, we integrate green and resilient supplier selection and order allocation decisions with vehicle routing decisions under disruption. We present a multiproduct two-stage risk-averse mixed-integer stochastic linear programming for the green and resilient supplier selection and order allocation integrated with vehicle routing (G&RSS&OA-V) problem. We consider resilient strategies before disruption, including multiple sourcing, supplier fortification, prepositioned inventory at the protected supplier, and contract with third-party logistics providers (3PLs). The objective function is to minimize the total mean-risk cost and the cost of greenhouse emissions. We use conditional value at risk (CVaR) as a risk measure to control the risk of worst-case cost. The most significant decisions of this model are the strategic decisions of determining the optimal suppliers and the operational decisions of vehicles routing under disruption simultaneously. Other decisions include determining which suppliers should be fortified, the amount to be transported to the hybrid manufacturing-distribution (HMD) center through the supplier or prepositioned emergency inventory, and the amount of lost sales. In order to validate the proposed model and its features, several numerical examples along with sensitivity analysis are performed by GAMS software, which shows the efficiency and application of the developed model, and some managerial insights are reported. The results of the sensitivity analysis show that as α increases from 0.1 to 0.9, the mean-CVaR objective function cost increases to 13.2%. As λ increases from 0.1 to 0.9, the mean-CVaR objective function cost increases to 35.6%. The increase of these two risk factors makes the proposed model more risk-averse. As the expected shortage cost increases by 150%, the mean-CVaR objective function cost increases to 36% while the amount of expected shortage decreases by 56%. Keywords Supplier selection/order allocation problem Vehicle routing problem Green paradigm and greenhouse emissions Fuel consumption Two-stage stochastic programming Disruption risk and resilience ==== Body pmcIntroduction Natural disasters such as floods, earthquakes, and hurricanes and intentional/unintentional human actions such as strikes, fires, terrorist attacks, and epidemic/pandemic outbreaks are some of the disruptions that may occur in the supply chain (Aldrighetti et al. 2021; Azimian et al. 2021). The effects of these major disruptions include the incomplete implementation of companies’ production plans; delays in purchase orders; lost sales; inventory shortage; high supply, production, and transportation costs; disruption of the transport fleet; and disruption of routs. Following the outbreak of the coronavirus in 2020, many factories around the world shut down. The shutdowns had a major impact on the global supply chains. Some major automakers faced the threat of a shortage of parts. There were also concerns about supplies of Apple products as the disruptions continue. In today’s competitive market, companies are looking to outsource the elements needed in their production planning. Therefore, the supplier selection and order allocation (SS&OA) play an essential role in supply chain management. The supplier selection problem is one of the strategic decisions that will be made in the long term. The SS&OA problem is vulnerable to supply disruption, which leads to a reduction or loss of supplier capacity, and among its effects are delays in customer orders and a lack of inventory. Various resilience strategies have been used to deal with these disruptions. Supply chain resilience is the ability of a network to withstand disruptions, adaptation, and recovery until customer demand is met and performance is guaranteed (Hosseini et al. 2019a, b). Multiple sourcing strategies, backup supplier, supplier recovery, supplier fortification, and prepositioned inventory have been reported as effective risk mitigation strategies in the resilient supply chain (Wang et al. 2022; Wofuru-Nyenkeet al. 2022; Esmaeili-Najafabadi et al. 2021). The vehicle routing problem (VRP) is one of the operational decisions that will be made in the short term. The purpose of the VRP problem is to find the optimal set of routes for the vehicle fleet from one warehouse to a specific set of customers to meet customer demand with the objective function of minimizing the total cost of transportation, fixed cost of vehicles, and other related costs (Hamidi Moghaddam et al. 2021). In most vehicle routing problems, all transport routes and vehicles are always available and serving customers. However, once a disruption occurs, the transportation network may lose the capacity of the vehicle fleet or a number of routes may be inactive due to route disruptions, roadwork, quarantine for COVID-19, and heavy traffic. Transportation mitigation strategies can include contracting with 3PLs, backup vehicles, and various modes of transportation. Environmental concerns have led to the enactment of laws and regulations by the government and international institutions. Therefore, companies are required to pay attention to environmental issues when configuring their supply chain network. For example, the use of up-to-date production technologies and transportation modes is one of the most effective ways to reduce environmental damage, which leads to lower greenhouse gas emissions, fuel consumption, and pollution, but at a higher cost (Tahmasebi Zadeh and Boyer 2021). The green paradigm protects the environment by minimizing the environmental wastes/pollution through the purchase of green materials from suppliers and green production using less-pollutant technologies (Panpatilet al. 2022). As a result of this increase in global awareness and government legislation on environmental impacts, it is necessary to integrate environmental issues into the SS&OA problem and VRP. In the green SS&OA problem and green VRP, we seek to reduce greenhouse gas (GHG) emissions in suppliers and reduce fuel consumption in the delivery of goods to customers by vehicles. Traditionally, strategic decisions are made for the SS&OA problem and then operational decisions for the VRP. That is, first the major suppliers and order allocation are identified, and then routing decisions are made. While in the real world, both decisions are made simultaneously. Proactive optimization models under disruptions and developing resilient supply chain network designs can help supply chains and markets survive (Aldrighetti et al. 2021). Also, one of the topics of interest today is green supply chain management issues and environmental concerns. Thus, this paper presents the strategic and operational decision-making in a risk-averse, multiproduct green and resilient supplier selection and order allocation integrated with vehicle routing (G&RSS&OA-V) problem under supplier capacity, transport fleet, and route disruptions (David et al. 2022). Resilient strategies are adopted before disruption. We adopt multiple sourcing, fortification, and prepositioned inventory strategies under supply disruption. We also consider a 3PL contracting strategy to provide transportation services under the disruption of the transport fleet and route. Hence, we propose a new two-stage stochastic programming model for the risk-averse decision-maker with the mean-risk objective function. We use CVaR as a risk measure in optimizing the objective function, which seeks to minimize the worst-case scenario. The structure of this paper is as follows. The “Literature Review” section prepares a review of the related literature. The green and resilient supplier selection and order allocation integrated with vehicle routing (G&RSS&OA-V) problem under disruption with formulation, assumptions, and limitations is elaborated in the “Model Description” section. The “Computational Analysis and Examples” section presents a computational experiment and sensitivity analysis. Finally, the “Conclusion” section states the conclusions and future research of this paper. Literature Review In this article, we aim to combine supply and transportation (vehicles and routs) disruption risk in the G&RSS&OA-V problem. Hence, in the following, the related literature is reviewed in the two research streams: supplier selection and order allocation under disruption and vehicle routing problem under disruption. In the scope of our review, we have ignored articles that only address the SS&OA and VRP without disruption. Supplier Selection and Order Allocation Under Disruption The SS&OA problem is a complex, multicriteria decision problem that deals with selecting the best suppliers and assigning orders to the suppliers. Sawik (2011) presented a risk-averse selection of supply portfolio in a make-to-order manufacturing strategy under disruption risks by using CVaR. Sawik (2013) proposed a risk-neutral, risk-averse, and mean-risk resilient supply portfolio under disruption with CVaR metric to control the risk of worst-case cost. Naqvi and Amin (2021) reviewed the supplier selection and order allocation problem in three categories: literature reviews, deterministic optimization models, and uncertain optimization models. Torabi et al. (2015) designed the resilient supplier selection and order allocation problem in response to uncertainties caused by major disruptions and operational risks of supply. They used a biobjective two-stage mixed possibilistic, stochastic programming model to minimize total expected cost and maximize a resilience objective. They also considered proactive strategies such as suppliers’ business continuity plans and fortifying suppliers. Esmaeili-Najafabadi et al. (2019) studied a joint supplier selection and order allocation problem under disruption risks. They developed a mixed-integer nonlinear programming model to minimize costs of centralized multiproduct supply chains and used two proactive strategies to reduce interruptions, including supplier protection and propositioned emergency inventory policy. Fattahi et al. (2020) developed a mixed-integer two-stage stochastic nonlinear programming in supply chain network design under a disrupted distribution center. They set a new measure of supply chain resilience as the expected amount of the supply chain’s operational cost increase due to a disruption event during its recovery period. They reformulated a mixed-integer nonlinear programming model into a conic quadratic mixed-integer program that can be solved by commercial solvers such as CPLEX. They used the sample average approximation method to manage the large number of disruption scenarios and also examined the criterion of risk-based resilience using CVaR. Kaur and Singh (2021) presented a multiperiod, multiproduct hybrid supplier selection and order allocation model under supply disruption risks and disruptive technologies. Suppliers are divided into efficient and inefficient suppliers using the DEA method, and efficient suppliers are evaluated and ranked using the FAHP-TOPSIS method, and then the risk of noncompliance of each supplier is calculated by TOPSIS. Finally, mixed-integer programming is applied to minimize the total cost of logistics and the associated disruption risk. Esmaeili-Najafabadi et al. (2021) proposed a mixed-integer nonlinear programming model for risk-averse supplier selection and order allocation in the centralized supply chains under local and regional disruption risks. They categorized the suppliers into domestic suppliers and foreign suppliers. Finally, they used value-at-risk (VaR) and conditional value-at-risk (CVaR) to analyze the risk aversion model. Chen et al. (2022) proposed a mixed-integer linear programming model multiperiod and multistage supply chain under supply disruption during COVID-19. They used product design change considering product life cycle and design change time as a proactive strategy. Fahimnia and Jabbarzadeh (2016) integrated supply chain sustainability and resilience and developed a sustainability performance scoring method. They designed a stochastic multiobjective fuzzy goal programming model under supply disruption. Hamdan and Cheaitou (2017) suggested a dynamic green supplier selection and order allocation with quantitative discounts and different supplier availability between planning periods. First, the decision-makers used fuzzy TOPSIS to assign two preferred suppliers’ weights based on traditional and green. Second, top management used AHP to determine the weight of importance to each of the two criteria. Third, they used a biobjective integer linear programming model with all-unit quantity discounts to maximize order quantity to suppliers and minimize the total cost. Zahiri et al. (2017) proposed a multiobjective fuzzy possibilistic-stochastic programming model for a sustainable and resilient supply chain under uncertainty. Vahidi et al. (2018) proposed a sustainable and resilient SS&OA problem under operational and disruption risks. The first objective function has been developed to maximize the sustainability and resilience aspects of selected suppliers, and the second objective function aims to minimize the total expected cost of the biobjective two-stage possibilistic-stochastic programming model. Ghomi-Avili et al. (2021) studied inventory-pricing decisions in a competitive green supply chain network design problem under supplier and distribution center disruptions. They introduced a robust bilevel model integrated by conditional value at risk (CVaR) to maximize the total profit and reduce the CO2 emissions. They also used the Stackelberg game to model the competition and to show the customer response in a price-dependent demand environment with fuzzy coefficients for each supply chain. In their model, they used the strategy of contracting with reliable suppliers to mitigate supply disruption and the sharing strategy to reduce distribution center disruption risks. Yavari and Zaker (2020) presented biobjective linear programming in a resilient green closed-loop supply chain network for perishable products under supply and power network disruption. Their first goal is to minimize the total network costs, and their second goal is to minimize the total network carbon emissions. In order to deal with disruptions, they used intermediate facilities, lateral transshipment, emergency inventory, capacity expansion, and integrating interdependent networks as resilient strategies. Tirkolaee et al. (2020) designed a hybrid fuzzy decision-making and sustainable-reliable SS&OA model. They used the weighted goal programming method with three objective functions to minimize the total cost, maximize the weighted value of products, and to maximize the reliability of the supply chain. Nayeri et al. (2021) proposed a multiobjective fuzzy robust stochastic model for a sustainable-resilient-responsive supply chain network under supply, manufacturer, and distribution center disruption. The purpose of their model is to minimize total costs and environmental damages while maximizing social impacts, responsiveness, and resilience levels. They used node criticality and node complexity as resilience measures. Hasani et al. (2021) proposed a robust green and resilient multiobjective supply chain optimization model under disruption for the global medical equipment manufacturing system. The first goal is to maximize the total profit. The second goal is to minimize the centralization facilities. The third goal is to minimize the CO2 emissions from material transport between facilities. They used four mitigation strategies such as facility fortification, facility dispersion, semifinished products, and multiple sourcing. Yavari and Ajalli (2021) designed a biobjective mixed-integer linear programming model for a green resilient supply chain network under disruption risks to minimize total cost and carbon emissions. In order to deal with disruptions, they applied coalition between suppliers, multiple sourcing, emergency inventory, and capacity expansion. Vehicle Routing Problem Under Disruption The VRP inherently provides significant savings in transportation costs. The green VRP (GVRP) is also an attractive research field that is of interest to many researchers. Lin et al. (2014) reviewed GVRP models in energy consumption, greenhouse gas emissions, and reverse logistics and classified them to green VRP, pollution routing problem, and VRP in reverse logistics. Moghdani et al. (2021) systematically reviewed GVRP in its variants, objective functions, uncertainty, and solution approach. Ahmadi-Javid and Seddighi (2013) designed a location-routing problem under disruption. The capacity of each producer–distributor and the vehicles are vulnerable. They applied the mixed-integer linear programming model to minimize the total cost under three risk-measurement policies: moderate, cautious, and pessimistic. Nasiri et al. (2018) proposed an integrated supplier selection and order allocation problem with vehicle routing and in multi-cross-dock supply chain in order to make a suitable trade-off between cost and responsiveness. They used mixed-integer linear programming to minimize the objective function including purchase, shipping, cross-docking, holding, and early/tardy delivery penalty costs. Yavari et al. (2020) presented a location-inventory-routing problem for perishable products under route disruptions. They integrated the location-inventory-routing problem by price-sensitive demand, a product with a certain lifetime, and disruption in routes. They used a mixed-integer nonlinear programming model to maximize the profits of their entire network. Zhong et al. (2020) introduced a risk-averse, biobjective mixed-integer nonlinear programming model for disaster relief facility location and vehicle routing under stochastic demand. Their model included conditional value at risk with regret (CVaR-R) as a risk measure. They proposed two goals including CVaR-R of the waiting time and the CVaR-R of the network cost. Finally, they solved the proposed model by the hybrid genetic algorithm. Dehghan et al. (2021) proposed a scenario-based mixed-integer linear programming model for the capacitated location routing problem with simultaneous pickup and delivery under disrupted depots to minimize the expected cost of the fixed location, unfulfilled demand, and variable routing costs. They used three tailored metaheuristic algorithms to solve the proposed model. Disruption risks cause customers not to receive their goods or services at scheduled times in a VRP problem. This causes dissatisfaction and loss of customers and over time causes significant financial losses to the transportation network. Therefore, it is absolutely necessary to consider a reliable VRP in the transportation network. Zhang et al. (2015) developed a reliable location-routing problem under depot disruption risks. They also designed a two-stage scenario-based mixed-integer programming model for the location-routing problem with the goal of minimizing costs. Then, they develop an efficient metaheuristic method to solve their proposed problem. Xie et al. (2016) formulated a reliable location-routing problem under depot disruption. Disruption in the depot makes it impossible to send the vehicle. Therefore, customers in that depot must be serviced by additional vehicles from other backup depots. Finally, they applied integer linear programming to minimize the fixed setup cost of depots, transportation cost, and the cost of penalties for missing services. Rayat et al. (2017) presented a reliable multiperiod, multiproduct location-inventory-routing problem under disruption. They used a biobjective mixed-integer nonlinear programming model to minimize the first objective function, including the total locating, routing, and inventory costs. Their model also minimized the second objective function, which includes the total failure costs related to disrupted distribution centers. Cheng et al. (2018) studied a two-stage robust approach for designing a reliable logistic network under supply and transportation disruptions. In the first stage, location decisions are made before disruptions and recourse decisions are made after the disruptions. They solved the proposed model exactly by a column-and-constraint-generation algorithm, which works better than the Benders decomposition method. Elluru et al. (2019) proposed a resilient location routing model with time windows under the disrupted distribution center and route. They used proactive and reactive strategies to deal with the disruptions. In the proactive strategy, the risk factor of each distribution center is considered before the disruption. The reactive strategy identifies the disrupted routes and recalculates the distribution routes to minimize the penalty for time window delays. Then, the proposed model optimizes the facility expansion costs, unmet demand costs, and delay costs. Table 1 classifies more characteristics of the literature in the field of the SS&OA problem and VRP under disruption. We also discuss the features of our work and demonstrate them in the last row of Table 1.Table 1 Review and classification of recent SS&OA and VRP model under disruption risk References (authors) Decision level Mitigation strategy (proactive and reactive) Green paradigm Risk measure Disruption Vulnerable part Objective function Strategic Tactical Operational Emergency inventory Protection/fortification Capacity expansion Multiple sourcing Recovery Spot purchase Backup supplier 3PL CVaR Other Partial Complete Suppliers Manufacturers Distribution centers/depots Transport (Mean) expected cost/profit (Risk) CVaR cost/service level Expected cost + CVaR Other Route Vehicle Ahmadi-Javid and Seddighi (2013) ✓ - ✓ - - - ✓ - - - - - ✓ - ✓ - - ✓ ✓ - ✓ ✓ - - - Zhang et al. (2015) ✓ - ✓ - - - - - - - - - - - - ✓ - - ✓ - - ✓ - - - Torabi et al. (2015) ✓ - - ✓ ✓ - ✓ ✓ - ✓ - - - - ✓ - ✓ - - - - ✓ - - ✓ Xie et al. (2016) ✓ - ✓ - - - - - - - - - - - - ✓ - - ✓ - - ✓ - - - Fahimnia and Jabbarzadeh (2016) ✓ - - - - - ✓ - - - - ✓ - - - ✓ ✓ - - - - ✓ - - ✓ Rayat et al. (2017) ✓ - ✓ - - - - - - - - - - - ✓ - - - ✓ - - ✓ - - ✓ Hamdan and Cheaitou (2017) ✓ ✓ - - - - - - - - - ✓ - - ✓ - ✓ - - - - ✓ - - ✓ Zahiri et al. (2017) ✓ ✓ - - - - ✓ - - - - ✓ - - ✓ - - ✓ - - - ✓ - - ✓ Cheraghalipour and Farsad (2018) ✓ ✓ - - - - ✓ - - - - - - - - ✓ ✓ - - - - ✓ - - ✓ Sabouhi et al. (2018) ✓ - - ✓ ✓ - ✓ - - - - - - - ✓ - ✓ - - - - ✓ - - - Jabbarzadeh et al. (2018) ✓ - - - - ✓ ✓ - - - - - - - ✓ - ✓ ✓ - - - ✓ - - ✓ Ni et al. (2018) ✓ ✓ - ✓ - ✓ - ✓ - - - - - - - - - ✓ - - - ✓ - - Cheng et al. (2018) ✓ ✓ - - - - - - - - - - - ✓ - ✓ ✓ - - ✓ - - - - - Nasiri et al. (2018) ✓ - ✓ - - - - - - - - - - - - - - - - - - ✓ - - ✓ Namdar et al. (2018) ✓ - - - - - ✓ ✓ ✓ ✓ - - ✓ - ✓ - ✓ - - - - ✓ ✓ - - Vahidi et al. (2018) ✓ - - - - - ✓ - - ✓ - ✓ - - ✓ ✓ ✓ - - - - ✓ - - ✓ Azad and Hassini (2019) ✓ ✓ - - - - ✓ ✓ - - - - - - ✓ - ✓ - - - - ✓ - - - Sawik (2019) ✓ ✓ - - - - - ✓ - - - - ✓ - ✓ - ✓ ✓ - - - ✓ ✓ - - Hosseini et al. (2019a) ✓ ✓ - ✓ - - - ✓ - - - - - - ✓ - ✓ - - - - ✓ - - - Esmaeili-Najafabadi et al. (2019) ✓ - - ✓ ✓ - ✓ - - - - - - - - ✓ ✓ - - - - ✓ - - ✓ Elluru et al. (2019) ✓ - ✓ - - ✓ - - - - - - - - - ✓ - - - ✓ - ✓ - - - Yavari et al. (2020) ✓ ✓ ✓ - - - - - - - - - - - - ✓ - - - ✓ - ✓ - - - Yavari and Zaker (2020) ✓ ✓ - ✓ - ✓ - - - - - ✓ - - - ✓ - ✓ - - - ✓ - - ✓ Zhong et al.(2020) ✓ - ✓ - - - - - - - - - ✓ - - - - - - - - - ✓ - - Fattahi et al. (2020) ✓ - - - - - - ✓ - - - - ✓ - - ✓ - - ✓ - - ✓ ✓ - - Ghomi-Avili et al. (2021) ✓ - - - - - - - - ✓ ✓ ✓ - ✓ ✓ ✓ - ✓ - - ✓ - - - Sawik (2021) ✓ - - ✓ - - ✓ ✓ - - - - ✓ - - ✓ ✓ - - - - - ✓ - ✓ Hasani et al. (2021) ✓ ✓ - - ✓ - ✓ - - ✓ - ✓ - - - ✓ ✓ ✓ - - - ✓ - - - Kaur and Singh (2021) ✓ ✓ - ✓ - - ✓ - - - - - - - ✓ - ✓ - - - - ✓ - - ✓ Esmaeili-Najafabadi et al. (2021) ✓ - - - - - ✓ - - - - - ✓ ✓ - ✓ ✓ - - - - ✓ ✓ - - Dehghan et al. (2021) ✓ - ✓ - - - - - - - - - - - - ✓ - - ✓ - - ✓ - - - Yavari and Ajalli (2021) ✓ - - ✓ - ✓ ✓ - - - - ✓ - - - ✓ ✓ - - - - ✓ - - ✓ Nayeri et al. (2021) ✓ - - - - - ✓ - - - - ✓ - - ✓ ✓ ✓ ✓ ✓ - - ✓ - - ✓ Chen et al. (2022) ✓ ✓ - ✓ - - ✓ - - ✓ - - - - - ✓ ✓ - - - - ✓ - - - Proposed model ✓ - ✓ ✓ ✓ - ✓ - - - ✓ ✓ ✓ - ✓ ✓ ✓ - - ✓ ✓ - - ✓ - Gap Analysis Table 1 and the reviewed articles show the gap in the literature, and we try to fill them. The limitation of most existing studies is that most of them consider the SS&OA problem and VRP separately. They assume that supply facilities and transportation networks are always reliable and available. They also do not care about green goals like minimizing greenhouse gas emissions and pollution, while in reality, integrating the SS&OA problem with VRP saves costs. Disruption risks can damage the facilities and transportation networks. Hence, disruption risks can affect the performance of logistics networks. As a result, it is very important to take into account integrated green and resilient SS&OA and VRP under disruption risks in the design phase of the logistics network in order to make decisions at the strategic planning level simultaneously with operational planning decision levels. Also, most of the articles addressed the supply disruption risk, and the number of articles that paid attention to the transportation network disruption risk (disruption in vehicles and network routes) is almost negligible. With regard to mitigation strategies, most articles considered proactive resilience strategies for supplier disruptions, and a few articles evaluated proactive resilience strategies for transportation network disruptions. A limited number of articles considered real-world assumptions, such as complete disruption of the transportation network and partial disruption of supply. In addition, another important limitation of the existing studies is that they assume the logistics network design problem for the risk-neutral decision-maker, while in real life, most decision-makers are risk-averse. Therefore, we conclude that the integrated risk-averse SS&OA and VRP have not been extensively studied and analyzed in the literature with the mean-risk objective function. Based on the mentioned features, the main contributions of our article are as follows:This paper considers stochastic programming for the G&RSS&OA problem integrated with VRP (G&RSS&OA-V) that optimizes strategic supplier selection decisions and operational routing decisions. The proposed model formulates new multiproduct risk-averse mathematical programming for the G&RSS&OA-V problem with the mean-risk objective function. We use CVaR as a risk measure, which is a linear, convex, well-behaved, and coherent risk measure to control the risk of worst-case cost (Sawik 2013). The proposed objective function minimizes the costs of supplier selection and order allocation, greenhouse emissions, fuel consumption (routing), resilience, lost sales, and CVaR simultaneously. The G&RSS&OA-V problem, in addition to supply disruptions, also considers the transportation network disruptions (routes and vehicle transport fleet). Aldrigetti et al. (2021) provided a review of the supply chain network design literature under disruption risks and suggested investing in different proactive resilience strategies in supplier selection and logistics network design. Our proposed model accounts for resilience strategies in the G&RSS&OA-V problem before the disruption. Suppliers do not completely lose their capacity due to supply disruption but lose it partially. The resilient strategies adopted in this area are to fortify suppliers, multiple suppliers, and prepositioned inventory. Some vehicles and some routes are completely deactivated due to transportation disruption. The resilient strategy in this area is to contract with a 3PL to serve the transportation network. Model Description Description of Green and Resilient Supplier Selection and Order Allocation Integrated with Vehicle Routing (G&RSS&OA-V) Problem Under Disruption Risk Supplier selection and order allocation (SS&OA) decisions at the strategic planning level and routing (VRP) decisions at the operational planning level are what we seek to integrate. We use the green paradigm to minimize the total negative environmental impacts and resilience strategies to deal with disruptions. Environmental experts use the life cycle assessment (LCA) method to analyze the environmental impact of activities and processes (Pishvaee et al. 2012). For this purpose, we use the environmental LCA method to measure greenhouse emissions. In this paper, we consider strategic and operational decisions on the G&RSS&OA-V problem under supply and transport network disruption. The purpose of the proposed problem is to select main green suppliers, assign orders to suppliers, and find optimal green vehicle routing to meet customer demand so as to minimize the costs of selecting suppliers and order allocation, greenhouse emissions, fuel consumption (routing), lost sales, resilient strategies, and CVaR. In our study, the customer can be a central depot, a retailer, or an end customer. The capacity of suppliers as well as the transportation network that serves customers is vulnerable to various types of disruption. If a disruption leads to a reduction in the capacity of suppliers, a reduction in the number of vehicles, a route ban, and a route breakdown, a series of costs such as lost sales costs and resilience strategies are imposed on the system to meet unfulfilled customer demand as much as possible. The root of disruptions is natural disasters (floods, earthquakes, hurricanes, etc.) and man-made disruptions (labor strikes, terrorist attacks, quarantine for COVID-19, etc.). In this paper, we use a scenario-based approach that threatens supply and transport network disruption scenarios. Disruptions are usually formulated by a set of scenarios, so that a set of facilities is disrupted together under each disruption scenario (Snyder and Daskin 2006). In the G&RSS&OA-V problem under random disruption, the parameters are usually investigated through a set of discrete scenarios with a definite probability and applied to the model. The characteristics of the disruptions in our problem are:In each scenario, a disruption event may attack each supplier, and the level of disruption in supplier capacity varies. Hence, the remaining capacity of each supplier can be different from other suppliers under any disruption scenario. Depending on the severity of the disruption, the degree of disruption varies in the range [0,1]. For example, bbis = 0.6 means that 60% of the capacity of supplier i is available. In each scenario, each vehicle may be disrupted and completely removed from the transport network, and the parameter related to it is binary. For example, vvks = 0 means that vehicle k is broken in scenario s. In each scenario, each route may be disrupted and completely disabled in the transport network, and the parameter related to it is binary. For example, rdjls = 0 means that the route j to l and vice versa is inactive under scenario s. Estimation of the probability of potential disruptions and their impact on the supply and transportation process under each scenario can be obtained through risk assessment analysis (Torabi et al. 2014). In general, companies can increase resilience by creating redundancy in the entire supply chain (including prepositioned emergency inventory strategy and dual/multiple sourcing strategy), increasing supply chain flexibility, and changing corporate culture (Sheffi 2005). To cope with potential supply and transport network disruptions, we employ the following diverse (proactive) resilience strategies in the proposed model:Employing a multiple sourcing strategy for outsourced materials/parts. One of the solutions to reduce disruption is dual or multiple sourcing instead of single sourcing. Dual or multiple sourcing is more expensive than single sourcing, but in the event of a disruption, it can respond to customer demand and prevent shortages and increase the credibility and reliability of companies (Torabi et al. 2015). Fortifying suppliers. Using prepositioned emergency inventory. Another common resiliency strategy is to maintain prepositioned inventory that is held in fortified suppliers and used after a disruption. Concluding a contract with 3PL for servicing the transportation network. As Fig. 1 shows, our problem network consists of three layers of suppliers, a hybrid manufacturing-distribution (HMD) center, and customers. The HMD center produces different products and dispatches them to customers through vehicle routing. The HMD center outsources materials/parts to a set of selected suppliers. Suppliers are divided into two categories. The first category is that suppliers do not use resilience strategies. The second category is suppliers which use resilience strategies. The transportation network can also use a resilience strategy. In this paper, we use a HMD center, the advantages of which include reducing pollution and saving on transportation network costs. Based on customer demand, we determine the HMD center production–distribution plan and supply plan. We may not meet demand under supply and transportation network disruption, so shortage is allowed as lost sales. The location of suppliers, HMD center, and customers are fixed and predefined.Fig. 1 The proposed network for RSS&OA-V problem Two-Stage Risk Aversion Stochastic Programming Framework with CVaR Criteria Two-stage stochastic programming is a common approach in SS&OA problems because of the two-stage nature of decisions. Birge and Louveaux (1997) defined the general formulation of a two-stage stochastic programming framework. In these models, strategic decisions such as selecting the main suppliers are made in the first stage before knowing the realization of stochastic parameters. However, when stochastic parameters are revealed, the operational and tactical decisions such as production, transportation, and routing should be considered the second-stage decisions (Govindan et al. 2017). Rockafellar and Uryasev (2000) described conditional value risk (CVaR), and in this study, we use a mean-risk model in order to integrate risk parameters in two-stage stochastic programming models. Thus, we can apply a linear programming problem based on phrase (1) (Soleimani and Govindan, 2014, Noyan 2012). 1 min(1+λ)CTX+∑sps(qs)Tys+λ(ξ+11-α∑spszs)st:Wsys=hs-TsX,∀sys≥0,∀szs≥(qs)Tys-ξ,∀szs≥0,∀sξ∈R0≤λ≤1 ξ is a decision variable that illustrates the optimal value of VaR in the risk-averse model. zs is the tail cost in scenario s defined as the nonnegative value that the cost of scenario s exceeds VaR. Phrase (1) is a two-stage stochastic programming model integrated with CVaR in the G&RSS&OA-V problem under disruption. We present the entire model in the next segment. The G&RSS&OA-V Model Decisions of the G&RSS&OA-V problem include identifying major green suppliers, order allocation to suppliers, green production rate per customer, green vehicle routing, resilience strategies, and lost sales. In the proposed problem, we consider scenario-based modeling under supply and transportation network disruption caused by natural and man-made disasters, in which the probability of occurrence of each scenario is definite. The explanations of each scenario are in accordance with “Description of Green and Resilient Supplier Selection and Order Allocation Integrated with Vehicle Routing (G&RSS&OA-V) Problem Under Disruption Risk.” In each scenario, the capacity of the suppliers decreases due to disruption, so that the remaining capacity of each supplier under each scenario is known. Also, some vehicles and some routes under disruption may be inactive. In each scenario, it is clear which vehicles and routes are inactive after the disruption. In addition, proactive resilience strategies can help to satisfy the customer’s unmet demand as much as possible. Therefore, we propose a single-objective mixed-integer two-stage stochastic linear programming model for the G&RSS&OA-V problem under disruption with the mean-risk objective function. We use CVaR as a risk measure; we seek to minimize the sum of the total expected cost and CVaR cost. The G&RSS&OA-V problem decision-making process has a two-stage nature with a scenario-based approach. Here, the variables of selecting supplier, fortifying supplier, determining prepositioned inventory, and contracting with 3PL are scenario-independent. These are the first-stage variables and are fixed under each scenario. While the variables of flow between facilities, the production rate, routing, and sales lost are scenario-dependent. These are second-stage or recourse variables and can change in relation to all disruptions. The G&RSS&OA-V problem is an NP-hard problem in terms of complexity because it is the integration of two NP-hard problems. One is the SS&OA problem, which belongs to the category of integer linear programming problems (Karp 1972), and the other is the VRP (Golden et al. 2008). Despite the complexity of the G&RSS&OA-V problem, its most important superiority is the cost savings resulting from merging the two problems. One of the disadvantages of the stochastic programming approach is the lack of sufficient historical data in most real situations, which makes it difficult to estimate random distributions for uncertain parameters (Pishvaee and Torabi 2010). This issue is also true regarding the scenario-dependent parameters under the supply and transportation network disruption risks. Therefore, it is almost impossible to find probability distributions for such uncertain parameters. Due to the unavailability of the required data, they rely on judgmental data extracted from specialists and field experts (Torabi et al. 2015). In the following, we first describe the assumptions, sets, subsets, parameters, and decision variables and then illustrate our risk-averse two-stage stochastic linear programming model. Model Assumptions and Limitations The assumptions considered in the proposed model are as follows:Each material/part is supplied only by a certain number of suppliers, not all suppliers. Each product contains only a given number of parts, not all parts. We assume multiple potential locations to select and contract with suppliers. The HMD center produces and distributes products and meets customer demand through a routing system. The transport fleet is heterogeneous and the capacity of vehicles is different. Customer demand meet through vehicles in the network. Each vehicle also meets its dedicated customers with the goal of minimizing routing costs. Shortage in customer demand is allowed in the form of lost sales under disruption Suppliers and the transport network may face frequent disruptions that lead to reduced supply capacity and the transport network. Hence some costs such as lost sales and costs of implementing resilience strategies are imposed on the present problem. In this study, supply resilience strategies include multiple sourcing, supplier fortification, and pre-positioned inventory. Transport network resilience strategy is to conclude a contract with a3PL. The cost of transport by 3PL is much higher than the cost of transport by the network itself. 3PLcapacity does not decrease due to disruption and there is also a limited capacity for transporting products. Each random scenario occurs independently of other scenarios with a certain probability. The remaining capacity rate in each supplier, complete disruption of vehicles, and complete disruption of routes are stochastic and scenario-based. Emitted greenhouse gases and fuel consumption depend on distance, cargo weight, type, and speed of a vehicle. The main assumption of our model is the use of a hybrid manufacturing-distribution (HMD) center, the advantages of which are saving space, reducing time, reducing costs, reducing emissions (the main goal of the green paradigm), saving transportation, using common equipment in two manufacturing and distribution centers, and coordinating policy of two manufacturing and distribution centers. To distribute products to customers, routing is used by a heterogeneous transport fleet, which leads to a reduction in costs, including the cost of transportation, and reducing the cost of transportation also reduces pollution (Ostermeier and Hübner 2018). In real life, disruption in suppliers is partial and disruption in vehicles and routes is complete. The resilient strategies employed have different levels of resilience to be more realistic assumptions. In general, the assumptions of the proposed model are based on real life. Model Formulation Sets and subsets:I set of all suppliers i index of suppliers, i∈I L set of all customers j,l index of all customers, j,l∈L A set of all nodes (includes customer nodes plus origin node (HMD center)) a Total number of nodes S set of disruption scenarios s index of disruption scenarios, s∈S M set of products m index of products, m∈M P set of materials/parts p index of materials/parts, p∈P E set of fortification level e index of fortification level, e∈E K′ set of primary vehicles k′ index of primary vehicles, k∈K K″ set of vehicles in the 3PL k″ index of vehicles in the 3PL, k′∈K′ K set of all vehicles, K=K′∪K" Mp set of products in which part p is used, Mp⊆M pi set of materials/parts that supplied through supplier i, pi⊆P pm set of materials/parts used in product m, pm⊆P Parameters: Demand: dml demand of customer l from product m Fixed cost:ci fixed selection and ordering cost in supplier i cgie fixed fortification cost of supplier i at level e cpip fixed supply cost per part/material p in supplier i cv fixed cost of concluding a contract with 3PL for service by the transportation fleet Variable cost:pnip greenhouse emission cost for supply and transportation per part/material p in supplier i by HMD center pfjp greenhouse emission cost for pre-positioned emergency inventory per part/material p in supplier i by HMD center cmjlk fuel consumption cost for transport from node j to l by vehicle k cvmjlk' fuel consumption cost for transport from node j to l by 3PL (vehicle k′) Shortage cost:lsm lost sale cost per unit of product m Capacity constraints:cai maximum capacity of fortified supplier i for holding pre-positioned emergency inventory capi maximum initial capacity of supplier i for supplying parts/materials cape maximum capacity of HMD center for producing and distributing the product cappk maximum capacity of vehicle k Factors:wp weight of part/material p bbis remaining capacity rate of supplier i for supplying parts/materials under scenario s ps probability of occurrence of scenario s zmpm consumption coefficient of part/material p in product m wwm weight of product m, wwm=∑p∈Pmzmpmwp,∀m vvks binary parameter equals “1” if the vehicle k is not disrupted under scenario s, otherwise equals “0” rdjls binary parameter equals “1” if the route of node i to node j is not disrupted under scenario s, otherwise equals “0” α confidence level λ risk weight (factor) Decision variables:XNipms the amount of part/material p supplied and shipped for product m from non-fortified supplier i under scenario s XFipms the amount of part/material p supplied and shipped for product m from fortified supplier i under scenario s Nms the amount of lost sales of product m in the HMD center under scenario s Oipms the amount of part/material p use for product m from pre-positioned emergency inventory of supplier i under scenario s Qmkls the amount of product m assemble by the manufacturer for the customer l shipped by vehicle k under scenario s Moip the amount of pre-positioned emergency inventory of part/material p in fortified supplier i vlk non-negative variable to remove subtours VaR value at risk Cos auxiliary variable for calculating the conditional value at risk under scenario s xi binary variable equals “1” if supplier i is selected, otherwise equals “0” yie binary variable equals “1” if selected supplier i fortified at level e, otherwise equals “0” vl binary variable equals “1” in case of concluding a contract with 3PL for servicing the transportation fleet, otherwise equals “0” zjlks binary variable equals “1” if primary vehicle k travels from node j to node l under scenario s, otherwise equals “0” Objective Function The objective function aims to minimize the total mean-CVaR cost.2 MinTc=1+λ∑i∈Icixi 3 +1+λ∑i∈I∑e∈Ecgieyie 4 +1+λ∑i∈I∑p∈PcpipMoip 5 +1+λcv.vl 6 +∑i∈I∑p∈Pm∑m∈M∑s∈SpspnipXNipms 7 +∑i∈I∑p∈Pm∑m∈M∑s∈SpspnipXFipms 8 +∑i∈I∑p∈Pm∑m∈M∑s∈SpspfipOipms 9 +∑m∈M∑s∈SpslsmNms 10 +∑j∈Jj≠l∑l∈L∑k"∈K"∑s∈Spscmjlk"zjlk"s 11 +∑j∈Jjmath≠l∑l∈L∑k"∈K"∑s∈Spscvmjlk"zjlk"s 12 +λVaR+11-α∑s∈SpsCos The detailed objective function is the following: we consider the total cost of the two-stage mean-risk stochastic mathematical programming model as follows (Rahimi and Ghezavati 2018).Totalcost=1+λ×Fixedcosts+Expectedcosts+λ×CVaRcosts Expressions (1) to (11) represent the objective function of the model, which is described in the following sentences: term (1) includes the fixed cost of selecting and ordering with vulnerable suppliers multiplied by the risk weight plus one. Term (2) shows the fixed fortification cost of suppliers at different levels multiplied by the risk weight plus one. Term (3) indicates the fixed supply cost of prepositioned emergency inventory of fortified suppliers multiplied by the risk weight plus one. Phrase (4) includes the fixed cost of concluding a contract with 3PL for servicing the transportation fleet multiplied by the risk weight plus one. Term (5) is the greenhouse emission cost for supply and transportation of parts/materials from nonfortified suppliers to the HMD center. Phrase (6) calculates the greenhouse emission cost for supply and transportation of parts/materials from fortified suppliers to the HMD center. Term (7) covers the purchase and transportation cost of parts/materials as prepositioned emergency inventory from fortified suppliers for the HMD center. Term (8) calculates the lost sales cost of the HMD center. Term (9) includes the fuel consumption cost for transport by network vehicles from the HMD center to customers. Term (10) includes the fuel consumption cost for transport by 3PL from the HMD center to customers. In term (11), to achieve a better risk estimate of the worst-case scenario in the G&RSS&OA-V problem, we minimize the cost of CVaR under the disruption risk by considering resilience strategies. We use CVaR with the auxiliary function provided by Rockafellar and Uryasev (2000, 2002). λ is the risk factor that indicates the decision-makers’ willingness to risk VaR (value at risk). α is the confidence level that controls the risk of losses due to supply and transportation network disruption, and Cos is the cost of scenario s that exceeds VaR. Constraints 13 s.t.Cos≥∑ieI∑pϵPm∑mϵMpspnipXNipms+∑ieI∑pϵPm∑mϵMpspnipXFipms+∑ieI∑pϵPm∑mϵMpspfipOipms+∑mϵMpslsmNms+∑j∈Jj≠l∑lϵL∑kϵKpscmjlk′Ojlk′s+∑j∈Jj≠l∑lϵL∑k"ϵk"pscmjlk"Ojlk′′s-VaR∀s 14 Cos≥0∀s 15 ∑eϵEyie≤xi∀i∈I 16 ∑pϵPwpMoip≤cai∑eϵEyie∀i∈I 17 ∑pϵPi∑mϵMpwpXNipms≤xi-∑eϵEyiebbiscapi∀i,s 18 ∑pϵPi∑mϵMpwpXFipms≤capi∑eϵEyiebies∀i,s 19 ∑mϵMpOipms≤MOip∀i,p,s 20 ∑kϵK∑lϵLQmkls≤1zmpm∑iϵIXNipms+XFipms+Oipms∀m,p,s 21 ∑kϵKQmkls+Oms≥dml∀m,l,s 22 ∑m∈M∑l∈LwwmQmk′ls≤cappk′.vvk′s∀k′s 23 ∑m∈M∑l∈LwwmQmk"ls≤cappk′′.vl∀k",s 24 ∑j∈Aj≠lzjlks=∑j∈Aj≠lzljks∀l,k,s 25 ∑l∈Lzjlks≤1∀j=0,k,s 26 ∑m∈MQmkls≤M∑j∈Aj≠lzjlks∀l,k,s 27 vjk-vlk+azjlks≤a-1∀j,l∈A:j≠lj≠0,k,s 28 ∑m∈M∑l∈L∑k∈KwwmQmkls≤cape∀s 29 MOip,Nms,XNipms,XFipms,Oipms,Qmkls,vlk,VaR,Cos≥0∀i,j,m,p,k,s 30 xi,yie,zjlks,vlϵ0,1∀i,j,k,e,s Constraints (13) and (14) are risk constraints and used to calculate CVaR costs (Rockafellar and Uryasev 2000, 2002). Constraint (13) is the tail cost for scenario s which is defined as the cost of each scenario minus VaR. Constraint (14) Cos is a nonnegative variable under scenario s. Constraint (15) states that fortification of a supplier depends on the selection of that supplier, and supplier fortification is done maximum in one level. Constraint (16) sets the maximum capacity of prepositioned emergency inventory at each fortified supplier. Constraint (17) indicates the maximum capacity of unfortified suppliers after disruption. Constraint (18) illustrates the maximum capacity of fortified suppliers after disruption. Constraint (19) indicates the maximum prepositioned emergency inventory of each part/material under each scenario in each fortified supplier. Constraint (20) demonstrates the amount of products produced at the HMD center under each scenario. Constraint (21) states that the amount of product produced by the HMD center plus its lost sales is greater than the demand for that product. Constraint (22) indicates the capacity of each vehicle. Under complete disruption, it reaches zero. If not, it is equal to cappk’. Constraint (23) specifies the 3PL capacity for vehicles. In case of concluding a contract with 3PL, the capacity is equal to cappk″; otherwise, it is equal to zero. Equation (24) guarantees that each vehicle (related to the network itself or 3PL) that enters a node must also leave it. Constraint (25) ensures that each vehicle on each route exits the origin node (HMD center) only once. Constraint (26) represents that the amount of products produced in the HMD center per customer depends on the existence of the transportation route to that customer. Equation (27) is the subtour elimination constraint that causes each tour to start from one HMD center and multiple customers (Miller et al. 1960). Constraint (28) expresses the capacity of production and distribution in the HMD center. Constraints (29) and (30) also specify the type of decision variables. Computational Analysis and Examples Considering that the G&RSS&OA-V problem is NP-hard, in order to solve the model in a reasonable time, we present a small- or medium-sized computational example to validate the proposed problem. In fact, due to the limitations of the GAMS software, it can be an acceptable size for the computational example. The objective is to minimize the expected costs plus CVaR costs (mean-CVaR) of the supply network under disruption. The input data are hypothetical for computational examples. The examples are implemented in GAMS software to discover the optimal solution to the proposed G&RSS&OA-V problem. All calculations were performed on a laptop with an Intel Core i7 processor with 8 GB RAM. We consider a problem consisting of four suppliers, two fortification levels, two product types, three parts/material types, one HMD center (origin node), seven customers, five vehicles related to the transportation network, and five vehicles related to 3PL. In the real world, we cannot consider all the parameters definitively, especially the remaining capacity rate in each vulnerable supplier, complete disruption of vehicles, and complete disruption of routes and related costs. Then, we consider them as stochastic parameters. Khalili et al. (2017) described each risk by two parameters: risk probability and risk severity. In this study, we assume four cases for risk severity in stochastic parameters (high, mid, low, and no disruption). Thus, to deal with the uncertainty, we create ten scenarios with a definite probability together with the severity of the disruption of unfortified suppliers and vehicles in Table 2 in the test example. Table 3 illustrates the severity of the disruption in fortified suppliers in the test example. Table 4 shows the severity of route disruption in scenarios. Table 5 demonstrates the generated parameters. In all computational analyses, the parameters are generated by uniform distribution.Table 2 Likelihood and severity of disruption in unfortified suppliers and vehicles in test example Disruption scenario Severity of disruption in unfortified suppliers ps bbis Severity of vehicle disruption vvks i k s 1 2 3 4 1 2 3 4 5 1 High 0.08 0.18 0.23 0.18 0.19 High 0 0 1 0 1 2 High 0.11 0.19 0.21 0.12 0.18 Mid 1 1 0 1 0 3 High 0.08 0.15 0.19 0.25 0.16 Low 1 1 1 1 0 4 Mid 0.09 0.35 0.32 0.33 0.28 High 1 0 1 0 0 5 Mid 0.11 0.32 0.33 0.28 0.28 Mid 0 1 1 0 1 6 Mid 0.1 0.3 0.35 0.34 0.34 Low 1 0 1 1 1 7 Low 0.08 0.5 0.47 0.5 0.36 High 0 0 0 1 1 8 Low 0.12 0.38 0.37 0.38 0.47 Mid 0 1 0 1 1 9 Low 0.11 0.5 0.5 0.37 0.43 Low 1 0 1 1 1 10 No disruption 0.12 1* 1* 1* 1* No disruption 1* 1* 1* 1* 1* When we have no disruption Table 3 Severity of disruption in fortified suppliers in test example Disruption scenario Severity of disruption in fortified suppliers bies (e = 1) bies (e = 2) i i s 1 2 3 4 1 2 3 4 1 High 0.31 0.35 0.3 0.31 0.35 0.44 0.45 0.38 2 High 0.33 0.34 0.26 0.26 0.43 0.45 0.4 0.36 3 High 0.35 0.35 0.27 0.31 0.38 0.44 0.43 0.35 4 Mid 0.38 0.44 0.36 0.44 0.55 0.53 0.53 0.47 5 Mid 0.36 0.44 0.38 0.42 0.54 0.52 0.47 0.53 6 Mid 0.36 0.4 0.45 0.37 0.54 0.51 0.53 0.53 7 Low 0.55 0.57 0.59 0.5 0.66 0.62 0.69 0.67 8 Low 0.5 0.56 0.57 0.59 0.63 0.68 0.7 0.66 9 Low 0.57 0.56 0.58 0.5 0.65 0.62 0.65 0.69 10 No disruption 1 1* 1* 1* 1* 1* 1* 1* When we have no disruption Table 4 Severity of route disruption scenarios (rdjls) in test example Disruption scenario (s) 1 2 3 4 5 6 7 8 9 10 Severity of route disruption route (j.l) High Mid Low High Mid Low High Mid Low No dis 1.2 1 1 1 0 1 1 0 0 0 1 1.3 0 1 1 0 1 1 1 1 1 1* 1.4 1 0 1 0 1 1 1 1 0 1* 1.5 1 1 0 1 1 0 0 1 1 1* 1.6 0 0 1 1 0 1 0 1 1 1* 1.7 1 1 1 1 1 1 1 0 1 1* 1.8 0 1 1 1 0 1 1 1 1 1* 2.1 1 1 1 0 1 1 0 0 0 1* 2.3 0 0 1 1 1 1 1 1 1 1* 2.4 1 0 1 1 0 1 1 1 1 1* 2.5 0 1 1 0 1 1 0 1 1 1* 2.6 1 1 0 1 0 1 0 1 1 1* 2.7 0 1 1 0 1 0 1 0 1 1* 2.8 1 1 1 1 1 1 1 1 0 1* 3.1 0 1 1 0 1 1 1 1 1 1* 3.2 0 0 1 1 1 1 1 1 1 1* 3.4 1 1 1 0 1 0 0 1 1 1* 3.5 1 0 1 1 1 1 1 0 1 1* 3.6 0 1 1 0 0 1 0 0 0 1* 3.7 1 1 0 1 1 1 0 1 1 1* 3.8 1 1 1 1 0 1 1 1 1 1* 4.1 1 0 1 0 1 1 1 1 0 1* 4.2 1 0 1 1 0 1 1 1 1 1* 4.3 1 1 1 0 1 0 0 1 1 1* 4.5 0 1 1 1 0 1 0 0 1 1* 4.6 1 1 0 1 1 1 0 0 1 1* 4.7 0 1 1 0 1 1 0 1 0 1* 4.8 0 1 1 1 1 1 1 1 1 1* 5.1 1 1 0 1 1 0 0 1 1 1* 5.2 0 1 1 0 1 1 0 1 1 1* 5.3 1 0 1 1 1 1 1 0 1 1* 5.4 0 1 1 1 0 1 0 0 1 1* 5.6 1 1 1 1 1 1 1 1 0 1* 5.7 0 0 1 0 0 1 0 1 1 1* 5.8 1 1 1 0 1 1 1 1 1 1* 6.1 0 0 1 1 0 1 0 1 1 1* 6.2 1 1 0 1 0 1 0 1 1 1* 6.3 0 1 1 0 0 1 0 0 0 1* 6.4 1 1 0 1 1 1 0 0 1 1* 6.5 1 1 1 1 1 1 1 1 0 1* 6.7 0 1 1 0 1 1 1 1 1 1* 6.8 1 0 1 0 1 0 0 1 1 1* 7.1 1 1 1 1 1 1 1 0 1 1* 7.2 0 1 1 0 1 0 1 0 1 1* 7.3 1 1 0 1 1 1 0 1 1 1* 7.4 0 1 1 0 1 1 0 1 0 1* 7.5 0 0 1 0 0 1 0 1 1 1* 7.6 0 1 1 0 1 1 1 1 1 1* 7.8 1 0 0 1 0 1 1 1 1 1* 8.1 0 1 1 1 0 1 1 1 1 1* 8.2 1 1 1 1 1 1 1 1 0 1* 8.3 1 1 1 1 0 1 1 1 1 1* 8.4 0 1 1 1 1 1 1 1 1 1* 8.5 1 1 1 0 1 1 1 1 1 1* 8.6 1 0 1 0 1 0 0 1 1 1* 8.7 1 0 0 1 0 1 1 1 1 1* *When we have no disruption Table 5 Parameters and scalars of the test example Parameters Severity Distribution ci - Uniform(20,000,40,000) cai - Uniform(10,000,20,000) capj - Uniform(5000,7000) cappk - Uniform(6000,12,000) lsm - Uniform(20,000,40,000) dml - Uniform(200,500) cpip - Uniform(100,200) cmjlk - Uniform(10,30) cmjlk’ - Uniform(30,60) cgie - Uniform(1000,2000) pnip - Uniform(40,60) pfip - Uniform(70,90) zmpm - Uniform(1,3) wp - Uniform(2,4) bbis High Uniform(0.1,0.25) Mid Uniform(0.25,0.35) Low Uniform(0.35,0.5) bies (e = 1) High Uniform(0.25,0.35) Mid Uniform(0.35,0.45) Low Uniform(0.5,0.6) bies (e = 2) High Uniform(0.35,0.45) Mid Uniform(0.45,0.55) Low Uniform(0.6,0.7) cv - 20,000 cape - 100,000 Computational Results The G&RSS&OA-V problem was solved based on the data in Tables 2, 3, 4, and 5. Table 6 shows different components of objective function, risk measure, and expected lost sale for the described example (α = 0.1, λ = 0.1).Table 6 Different components of the main example First-stage costs Second-stage costs Mean-CVaR CVaR VaR Expected lost sale 2,596,086 5,689,224 8,605,587 606,688 356,079 155 In the first-stage decisions, among the four suppliers, the model selects suppliers 1, 2, 3, and 4. All four suppliers are fortified at level two. Prepositioned emergency inventory is stored in all four suppliers. Also, a contract is concluded with 3PL to serve the transportation network. Regarding the second-stage decisions, we investigate the outputs related to the four scenarios 2, 5, 8, and 10 in which the severity of supply disruption is high, med, low, and no disruption, respectively. Scenario 2 uses the strategy of multiple suppliers to purchase part/material 1, including fortified suppliers 1, 2, and 4. Also, scenario 2 uses the strategy of multiple suppliers to purchase part/material 3, including fortified suppliers 2 and 3. In scenario 2, the prepositioned emergency inventory is stored in suppliers 3 and 4. Table 7 shows the amount of production and lost sales under scenario 2 for products 1 and 2. The network uses vehicles 1 and 2 in routing. Also, 3PL uses vehicle 7 in routing.Table 7 Comparison between production and lost sales Scenario Severity of disruption in supply Severity of transportation network disruption The amount of product (Qmkls) Quantity The amount of lost sale (Nms) Quantity 2 High Mid ∑k,lQm=1.k.l.2 305 N1.2 289 ∑k,lQm=2.k.l.2 813 N2.2 219 5 Mid Mid ∑k,lQm=1.k.l.5 1354 N1.5 136 ∑k,lQm=2.k.l.5 2329 N2.5 0 8 Low Mid ∑k,lQm=1.k.l.8 1611 N1.8 99 ∑k,lQm=2.k.l.8 2329 N2.8 0 10 No disruption No disruption ∑k,lQm=1.k.l.10 2016 N1.10 41 ∑k,lQm=2.k.l.10 2329 N2.10 0 Scenario 5 uses the strategy of multiple suppliers to purchase part/material 1, including fortified suppliers 1, 2, 3, and 4. There is the prepositioned inventory in the fortified suppliers 1, 2, 3, and 4 under scenario 5. Table 7 shows the amount of production and lost sales under scenario 5 for products 1 and 2. The network uses vehicles 2, 3, and 5 in routing. 3PL also uses vehicles 6, 7, 8, and 9. Scenario 8 uses the strategy of multiple suppliers to purchase part/material 1, including fortified suppliers 1, 2, 3, and 4. There is the prepositioned inventory in the fortified suppliers 1, 2, 3, and 4 under scenario 8. Table 7 shows the amount of production and lost sales under scenario 8 for products 1 and 2. The network uses vehicles 2, 4, and 5 in routing. 3PL also uses vehicles 6, 7, 8, 9, and 10. Scenario 10 uses the strategy of multiple suppliers, including fortified suppliers 1, 2, 3, and 4, to purchase part/material 1. There is the prepositioned inventory in the fortified suppliers 1, 2, 3, and 4 under scenario 10. Table 7 shows the amount of production and lost sales under scenario 10 for products 1 and 2. Vehicles 1, 2, 3, 4, and 5 are also used in routing. 3PL also uses vehicles 6, 7, 9, and 10. Table 7 shows that the production rate among the scenarios increases from high to low, and vice versa, the amount of lost sales between the scenarios decreases from high to low. Sensitivity Analysis on the Risk Parameters In this section, we present a complete sensitivity analysis of the risk-averse parameters α and λ. We then apply different values of α and λ in the mean-risk model to understand how they affect the proposed G&RSS&OA-V model. There are usually two well-known financial metrics for controlling the risk of supply disruptions based on α confidence level, which are:Value at risk (VaR) is a decision variable based on α% costs so that for α% scenarios, the result will not exceed VaR. Conditional value at risk (CVaR) is the expected cost of the portfolio in the worst (1-α) % total costs, i.e., (1-α) % of results more than VaR, and the average value of these results (greater than VaR) is represented by CVaR. The mathematical properties of CVaR are superior to VaR. CVaR is a coherent risk measure. For example, the CVaR portfolio is a continuous and convex function, while VaR may even have a discontinuous function (Sarykalin et al.2008; Sawik 2013). A risk-averse decision-maker wants to use CVaR to minimize the worst-case scenario that goes beyond VaR. In the mean-CVaR model, the supply portfolio integrated with routing decisions is selected along with green paradigm and proactive resilience strategies to minimize both expected costs and CVaR costs. α is one of the fundamental parameters of risk-averse decision-making, the effects of which should be analyzed on the proposed model. We put seven cases for alpha (α = 0.1, 0.3, 0.5, 0.6, 0.7, 0.8, 0.9) with a fixed value of lambda (λ = 0.1) to investigate how alpha (α) affects the objective function. According to Fig. 2, with increasing alpha (α), the model’s risk aversion behavior increases; in other words, the model acts more conservatively. As you can see in Fig. 2, VaR and CVaR values increase with increasing alpha. Also, with the increase of alpha, the first-stage costs do not change significantly. It is worth noting that the second-stage costs and the mean-CVaR costs increase with increasing alpha with a slight slope.Fig. 2 Alpha vs. cost, mean-CVaR, VaR, and CVaR Table 8 demonstrates the effect of different alpha on the VaR, CVaR, first-stage cost, second-stage cost, and mean-CVaR. According to Table 8, there is the lowest mean-CVaR in α = 0.1. Also, by increasing α from 0.1 to 0.9, the mean-CVaR objective function value grows from 8,115,034 to 9,187,824, that is, the objective function increases by 13.2%.Table 8 Effect of different alpha on VaR, CVaR, first-stage cost, second-stage cost, and mean-CVaR (λ = 0.1) α 0.1 0.3 0.5 0.6 0.7 0.8 0.9 VaR 349,642 480,390 590,427 576,408 616,189 723,134 746,948 CVaR 533,704 599,208 756,815 709,747 663,818 741,922 1,459,407 First-stage cost 2,622,188 2,470,178 2,436,527 2,473,741 2,337,033 2,514,960 2,572,953 Second-stage cost 5,177,257 5,433,567 6,123,415 5,608,118 5,559,949 5,377,333 6,211,635 Mean-CVaR 8,115,034 8,210,684 8,879,277 8,400,207 8,197,068 8,217,981 9,187,824 λ is also another essential parameter of risk-averse decision-making, the effects of which should be analyzed on the proposed model. To analyze how λ affects the objective function, we set five cases of λ (0.1, 0.5, 0.9, 5, 10) ith a constant value for α = 0.1. Figure 3 and Fig. 4 show the results of this experiment. As shown in Fig. 3, at a constant alpha (α) value by increasing the λ values, the first-stage costs decrease and the second-stage costs increase. According to Fig. 4, at a constant alpha (α) value, as the λ values increases, the Var and CVaR costs increase Fig. 3 Lambda vs. first-stage cost and second-stage cost Fig. 4 Lambda vs. VaR and CVaR cost As shown in Fig. 5, at different α (0.1, 0.5, 0.7), CVaR increases with increasing λ. According to Fig. 6, at different α (0.1, 0.5, 0.7), the cost of mean-CVaR objective function increases while λ increases. For example, at α = 0.1, as λ increases from 0.1 to 0.9, the mean-CVaR objective function value grows from 8,115,034 to 11,004,211; that is, the objective function increases by 35.6%.Fig. 5 CVaR values for different lambda Fig. 6 Mean-CVaR values for different lambda Lambda (λ) is the risk weight, and as λ increases, the degree of risk-averse decision-making increases and the model becomes more conservative. Therefore, by increasing the parameter λ and/or α, we achieve a high degree of risk aversion decision-making. Sensitivity Analysis on the Lost Sale (Shortage) Cost In the supply chain, a demand management strategy can be used to mitigate network risk (Tang 2006). The shortage cost (lsm) in the form of lost sales is a parameter that is effective in regulating demand management. By adjusting the shortage cost (lsm), we examine its effects on this model. To analyze the effect of lsm on the objective function, we set four cases of lsm according to Table 9 with a constant value for α = 0.1, λ = 0.1. Table 9 shows the results of this experiment. The higher the shortage cost (lsm), the higher the mean-CVaR and CVaR cost and the lower the expected shortage quantity. Therefore, increasing lsm leads to a decrease in shortage and an increase in resilience. Also, as the cost of shortages (lsm) increases, the number of selected suppliers does not change. According to Table 9, as the expected shortage cost increases from 12,500 to 30,000 (a 150% increase), the mean-CVaR objective function cost increases from 5,949,329 to 8,115,034 (a 36% increase), while the amount of expected shortage amount decreases from 315 to 138 (a 56% decrease).Table 9 Effect of different ls(m) on selected suppliers, mean-CVaR, CVaR, expected lost sale, fortified suppliers, prepositioned inventory, and multiple sourcing (α = 0.1, λ = 0.1) Distribution associated with shortage costs Shortage cost Selected suppliers Mean-CVaR CVaR Expected lost sale Fortified suppliers (e = 2) Sum of prepositioned inventory Multiple sourcing (i) 1 2 3 4 Uniform(5000,20,000) ls (m = 1) = 11,208 ls (m = 2) = 19,704 1, 2, 3, 4 5,949,329 469,786 315 1, 2, 3, 4 8729 p1,2 p1,2,3 p1,2,3 p1 Uniform(20,000,40,000) ls (m = 1) = 25,943 ls (m = 2) = 30,326 1, 2, 3, 4 8,115,034 533,704 138 1, 2, 3, 4 18,310 p1 p1,2 p1,3 p1 Uniform(40,000,60,000) ls (m = 1) = 43,549 ls (m = 2) = 54,279 1, 2, 3, 4 10,186,617 707,913 115 1, 2, 3, 4 21,045 p1 p1,2 p1,2,3 p1 Uniform(60,000,80,000) ls (m = 1) = 67,374 ls (m = 2) = 79,597 1, 2, 3, 4 11,905,326 884,838 98 1, 2, 3, 4 21,045 p1 p1,2 p1,2,3 p1 p1 include material/part 1. p1,2 include material/part 1and material/part 2. p1,2,3 include material/part 1, material/part 2, and material/part 3 Also, in Table 9, we intend to examine the strategies that help suppliers to become more resilient during disruptions. The strategy of fortifying suppliers at four levels of shortage cost shows that four suppliers are fortified at level 2. The prepositioned inventory strategy at four shortage cost levels states that inventory is increasing. This increase indicates that the assumed model has to increase the inventory of this strategy to reduce the shortage and more resilience, while the objective function cost increases with the increase of this strategy. In the multiple-sourcing strategy, p1 means procurement of material/part 1 with supplier i. p1,2 means procurement of material/part 1 and material/part 2 with supplier i. p1,2,3 means procurement of material/part 1, material/part 2, and material/part 3 with supplier i. As can be seen in Table 9, in some cases, we have single-sourcing, double-sourcing, and triple-sourcing strategies. The multiple-sourcing strategy helps to reduce the scarcity and increase the resilience of the model. In general, double or multiple sourcing is more expensive than single sourcing, but it prevents shortages in the event of a disruption and increases the reliability of the system. Managerial Insights Based on the analyzed results, the following managerial insights are provided:Applying the G&RSS&OA-V problem improves the performance of SS&OA and VRP problems under economic and environmental aspects, and resilience strategies significantly contribute to the problem performance under disruption. Hence, managers can make optimal decisions about proactive resilience strategies and green paradigm. The supplier selection and order allocation integrated with vehicle routing (SS&OA-V) problem is mainly investigated for the risk-neutral decision-maker, and our results show that the supply and transportation network disruption and green requirements significantly affect the structure of the SS&OA-V problem, so the proposed problem should be considered a risk-averse G&RSS&OA-V problem. The type of disruption events and the probability of disruption events have a great impact on the configuration of supplier selection and the transportation network. By increasing the weight factor parameter (λ) and the confidence level (α), the model becomes more risk-averse or, in other words, acts more conservatively. Managers can tune their risk level in CVaR through the confidence level (α) and its weight factor (λ). Also, increasing the shortage cost (lsm) leads to less shortage and more resilience of the model. Managers can tune demand management through parameter lsm. Conclusion In this paper, we proposed the multiproduct green and resilient supplier selection and order allocation integrated with vehicle routing (G&RSS&OA-V) problem under disruption risks to optimize total cost. We tried to consider the most practical environmental objectives and resilience strategies in the problem. To the best of our knowledge, this is the first time that the SS&OA-V problem with an efficient and practical combination of GSCM and proactive resilience strategies with risk aversion decisions is proposed to minimize greenhouse emissions and fuel consumption simultaneously. The objective function of our model includes minimizing the mean-risk (mean-CVaR) costs to optimize the performance of the worst-case scenario of the G&RSS&OA-V problem in a two-stage stochastic programming model. Our proposed model includes three stochastic parameters: the remaining capacity rate in each supplier, complete disruption of vehicles, and complete disruption of routes. We considered multiple sourcing, supplier fortification, prepositioned inventory, and concluding a contract with a 3PL as resilience strategies. In order to validate the proposed model, numerical examples are solved by using GAMS software. We used three important model parameters for sensitivity analysis. Various computational experiments are performed to examine these parameters on the objective function of the proposed model. In future research, we can consider other disruption risks (political and economic crises) and operational risks (cost fluctuations, climate changes) and their impact on suppliers and HMD center. Considering new proactive and reactive resilience strategies and green objectives will be another research avenue, accounting for the occurrence of multiple successive disruptions instead of one. In addition, applying other transportation modes, such as air transportation under the transportation network disruption, will lead to greater resilience. Further, the approach of this model can be used with the concept of sustainability which, in addition to economic factors and environmental concerns, also plans and manages social responsibility. Providing a metaheuristic algorithm to solve large-scale problems and identify an example of a problem that can sufficiently represent a set of all disruption scenarios can be another future area of research. In order to manage the uncertainty of the input data, we propose optimization approaches such as robust and fuzzy. Author Contribution Seyed Mojtaba Taghavi: conceptualization, methodology, investigation, resources, software, formal analysis, writing—original draft, and writing—review and editing. Vahidreza Ghezavati: conceptualization, methodology, investigation, formal analysis, writing—review and editing, validation, and supervision. Hadi Mohammadi Bidhandi: investigation, writing—review and editing, validation, and supervision. Seyed Mohammad Javad Mirzapour Al-e-Hashem: investigation, writing—review and editing, validation, and supervision. Data Availability The random datasets generated during and/or analyzed during the current study are available in the (Google Drive) repository (https://drive.google.com/file/d/1UVd6EX5PI5HfN5Two1nPsYykIeuVAvSO/view?usp=sharing). Declarations 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 Ahmadi-Javid A Seddighi AH A location-routing problem with disruption risk Transp Res Part E: Logist Transp Rev 2013 53 63 82 10.1016/j.tre.2013.02.002 Aldrighetti R, Battini D, Ivanov D, Zennaro I (2021) Costs of resilience and disruptions in supply chain network design models: a review and future research directions. Int J Prod Econ 235:1–21 Azad N Hassini E Recovery strategies from major supply disruptions in single and multiple sourcing networks Eur J Oper Res 2019 275 2 481 501 10.1016/j.ejor.2018.11.044 Azimian M Karbasian M Atashgar K Kabir G A new approach to select the reliable suppliers for one-shot devices Prod Eng 2021 15 3 371 382 10.1007/s11740-021-01032-8 Birge JR Louveaux F Introduction to Stochastic Programming 1997 New York Springer Chen J, Wang H, Fu Y (2022) A multi-stage supply chain disruption mitigation strategy considering product life cycle during COVID-19. Environ Sci Pollut Res 1–15. 10.1007/s11356-022-18931-7 Cheng C Qi M Zhang Y Rousseau LM A two-stage robust approach for the reliable logistics network design problem Transp Res Part B: Methodol 2018 111 185 202 10.1016/j.trb.2018.03.015 Cheraghalipour A Farsad S A bi-objective sustainable supplier selection and order allocation considering quantity discounts under disruption risks: A case study in plastic industry Comput Ind Eng 2018 118 237 250 10.1016/j.cie.2018.02.041 David DU Aikhuele DO Ughehe PO Tamuno EM Multi-echelon, Multi-period Supply Chain Master Planning in the Food Process Industry: A Sustainability Concept Process Integr Optim Sustain 2022 6 2 497 512 10.1007/s41660-022-00229-3 Dehghan M, Hejazi SR, Karimi-Mamaghan M, Mohammadi M, Pirayesh A (2021) Capacitated location routing problem with simultaneous pickup and delivery under the risk of disruption. RAIRO-Operations Research 55(3):1371–1399 Elluru S Gupta H Kaur H Singh SP Proactive and reactive models for disaster resilient supply chain Ann Oper Res 2019 283 1 199 224 10.1007/s10479-017-2681-2 Esmaeili-Najafabadi E Nezhad MSF Pourmohammadi H Honarvar M Vahdatzad MA A joint supplier selection and order allocation model with disruption risks in centralized supply chain Comput Ind Eng 2019 127 734 748 10.1016/j.cie.2018.11.017 Esmaeili-Najafabadi E Azad N Nezhad MSF Risk-averse supplier selection and order allocation in the centralized supply chains under disruption risks Expert Syst Appl 2021 175 114691 10.1016/j.eswa.2021.114691 Fahimnia B Jabbarzadeh A Marrying supply chain sustainability and resilience: A match made in heaven Transp Res Part E: Logist Transp Rev 2016 91 306 324 10.1016/j.tre.2016.02.007 Fattahi M, Govindan K, Maihami R (2020) Stochastic optimization of disruption-driven supply chain network design with a new resilience metric. Int J Prod Econ 230:107755 Ghomi-Avili M Tavakkoli-Moghaddam R JalaliNaeini SG Jabbarzadeh A Competitive green supply chain network design model considering inventory decisions under uncertainty: a real case of a filter company Int J Prod Res 2021 59 14 4248 4267 10.1080/00207543.2020.1760391 Golden BL Raghavan S Wasil EA The vehicle routing problem: latest advances and new challenges 2008 43 New York Springer Govindan K Fattahi M Keyvanshokooh E Supply chain network design under uncertainty: A comprehensive review and future research directions Eur J Oper Res 2017 263 1 108 141 10.1016/j.ejor.2017.04.009 Hamdan S Cheaitou A Dynamic green supplier selection and order allocation with quantity discounts and varying supplier availability Comput Ind Eng 2017 110 573 589 10.1016/j.cie.2017.03.028 Hamidi Moghaddam S Akbaripour H Houshmand M Integrated forward and reverse logistics in cloud manufacturing: an agent-based multi-layer architecture and optimization via genetic algorithm Prod Eng Res Devel 2021 15 6 801 819 10.1007/s11740-021-01069-9 Hasani A Mokhtari H Fattahi M A multi-objective optimization approach for green and resilient supply chain network design: a real-life Case Study J Clean Prod 2021 278 123199 10.1016/j.jclepro.2020.123199 Hosseini S Morshedlou N Ivanov D Sarder MD Barker K Al Khaled A Resilient supplier selection and optimal order allocation under disruption risks Int J Prod Econ 2019 213 124 137 10.1016/j.ijpe.2019.03.018 Hosseini S Ivanov D Dolgui A Review of quantitative methods for supply chain resilience analysis Transp Res Part E: Logist Transp Rev 2019 125 285 307 10.1016/j.tre.2019.03.001 Jabbarzadeh A Fahimnia B Sabouhi F Resilient and sustainable supply chain design: sustainability analysis under disruption risks Int J Prod Res 2018 56 17 5945 5968 10.1080/00207543.2018.1461950 Karp RM Reducibility among combinatorial problems Complexity of computer computations 1972 Boston Springer 85 103 Kaur H Singh SP Multi-stage hybrid model for supplier selection and order allocation considering disruption risks and disruptive technologies Int J Prod Econ 2021 231 107830 10.1016/j.ijpe.2020.107830 Khalili SM Jolai F Torabi SA Integrated production–distribution planning in two-echelon systems: a resilience view Int J Prod Res 2017 55 4 1040 1064 10.1080/00207543.2016.1213446 Lin C Choy KL Ho GT Chung SH Lam HY Survey of green vehicle routing problem: past and future trends Expert Syst Appl 2014 41 4 1118 1138 10.1016/j.eswa.2013.07.107 Miller CE Tucker AW Zemlin RA Integer programming formulation of traveling salesman problems J ACM (JACM) 1960 7 4 326 329 10.1145/321043.321046 Moghdani R Salimifard K Demir E Benyettou A The green vehicle routing problem: A systematic literature review J Clean Prod 2021 279 123691 10.1016/j.jclepro.2020.123691 Namdar J Li X Sawhney R Pradhan N Supply chain resilience for single and multiple sourcing in the presence of disruption risks Int J Prod Res 2018 56 6 2339 2360 10.1080/00207543.2017.1370149 Naqvi MA Amin SH Supplier selection and order allocation: a literature review J Data Inform Manag 2021 3 2 125 139 10.1007/s42488-021-00049-z Nasiri MM Rahbari A Werner F Karimi R Incorporating supplier selection and order allocation into the vehicle routing and multi-cross-dock scheduling problem Int J Prod Res 2018 56 19 6527 6552 10.1080/00207543.2018.1471241 Nayeri S, Torabi SA, Tavakoli M, Sazvar Z (2021) A multi-objective fuzzy robust stochastic model for designing a sustainable-resilient-responsive supply chain network. J Clean Prod 311:127691 Ni N Howell BJ Sharkey TC Modeling the impact of unmet demand in supply chain resiliency planning Omega 2018 81 1 16 10.1016/j.omega.2017.08.019 Noyan N Risk-averse two-stage stochastic programming with an application to disaster management Comput Oper Res 2012 39 3 541 559 10.1016/j.cor.2011.03.017 Ostermeier M Hübner A Vehicle selection for a multi-compartment vehicle routing problem Eur J Oper Res 2018 269 2 682 694 10.1016/j.ejor.2018.01.059 Panpatil SS, Prajapati H, Kant R (2022) Effect of green supply chain practices on sustainable performance indicators: a fuzzy MADM approach. Process Integration and Optimization for Sustainability 1–14. 10.1007/s41660-022-00260-4 Pishvaee MS Torabi SA A possibilistic programming approach for closed-loop supply chain network design under uncertainty Fuzzy Sets Syst 2010 161 20 2668 2683 10.1016/j.fss.2010.04.010 Pishvaee MS Torabi SA Razmi J Credibility-based fuzzy mathematical programming model for green logistics design under uncertainty Comput Ind Eng 2012 62 2 624 632 10.1016/j.cie.2011.11.028 Rahimi M Ghezavati V Sustainable multi-period reverse logistics network design and planning under uncertainty utilizing conditional value at risk (CVaR) for recycling construction and demolition waste J Clean Prod 2018 172 1567 1581 10.1016/j.jclepro.2017.10.240 Rayat F Musavi M Bozorgi-Amiri A Bi-objective reliable location-inventory-routing problem with partial backordering under disruption risks: A modified AMOSA approach Appl Soft Comput 2017 59 622 643 10.1016/j.asoc.2017.06.036 Rockafellar RT Uryasev S Optimization of conditional value-at-risk J Risk 2000 2 21 42 10.21314/JOR.2000.038 Rockafellar RT Uryasev S Conditional value-at-risk for general loss distributions J Bank Finance 2002 26 7 1443 1471 10.1016/S0378-4266(02)00271-6 Sabouhi F Pishvaee MS Jabalameli MS Resilient supply chain design under operational and disruption risks considering quantity discount: A case study of pharmaceutical supply chain Comput Ind Eng 2018 126 657 672 10.1016/j.cie.2018.10.001 Sarykalin S, Serraino G, Uryasev S (2008) Value-at-risk vs. conditional value-at-risk in risk management and optimization. In: State-of-the-art decision-making tools in the information-intensive age. Informs, pp 270–294. 10.1287/educ.1080.0052 Sawik T Selection of supply portfolio under disruption risks Omega 2011 39 2 194 208 10.1016/j.omega.2010.06.007 Sawik T Selection of resilient supply portfolio under disruption risks Omega 2013 41 2 259 269 10.1016/j.omega.2012.05.003 Sawik T Disruption mitigation and recovery in supply chains using portfolio approach Omega 2019 84 232 248 10.1016/j.omega.2018.05.006 Sawik T On the risk-averse selection of resilient multi-tier supply portfolio Omega 2021 101 102267 10.1016/j.omega.2020.102267 Sheffi Y Building a resilient supply chain Harvard Bus Rev Supply Chain Strat 2005 1 5 1 11 Snyder LV Daskin MS A Random-key Genetic Algorithm for the Generalized Traveling Salesman Problem Eur J Oper Res 2006 174 1 38 53 10.1016/j.ejor.2004.09.057 Soleimani H Govindan K Reverse logistics network design and planning utilizing conditional value at risk Eur J Oper Res 2014 237 2 487 497 10.1016/j.ejor.2014.02.030 TahmasebiZadeh H Boyer O A model for integrating green product development strategies and supply chain configuration considering market share Process Integr Optim Sustain 2021 5 3 417 427 10.1007/s41660-020-00152-5 Tang CS Perspectives in supply chain risk management Int J Prod Econ 2006 103 2 451 488 10.1016/j.ijpe.2005.12.006 Tirkolaee EB Mardani A Dashtian Z Soltani M Weber GW A novel hybrid method using fuzzy decision making and multi-objective programming for sustainable-reliable supplier selection in two-echelon supply chain design J Clean Prod 2020 250 119517 10.1016/j.jclepro.2019.119517 Torabi SA Soufi HR Sahebjamnia N A new framework for business impact analysis in business continuity management (with a case study) Saf Sci 2014 68 309 323 10.1016/j.ssci.2014.04.017 Torabi SA Baghersad M Mansouri SA Resilient supplier selection and order allocation under operational and disruption risks Transp Res Part E: Logist Transp Rev 2015 79 22 48 10.1016/j.tre.2015.03.005 Vahidi F Torabi SA Ramezankhani MJ Sustainable supplier selection and order allocation under operational and disruption risks J Clean Prod 2018 174 1351 1365 10.1016/j.jclepro.2017.11.012 Wang D, Ge G, Zhou Y, & Zhu M (2022). Pricing-decision analysis of green supply chain with two competitive manufacturers considering horizontal and vertical fairness concerns. Environ Sci Pollut Res 1–24 Wofuru-Nyenke OK, Briggs TA, Aikhuele DO (2022) Advancements in sustainable manufacturing supply chain modelling: a review. Process Integration and Optimization for Sustainability 1–25. 10.1007/s41660-022-00276-w Xie W Ouyang Y Wong SC Reliable location-routing design under probabilistic facility disruptions Transp Sci 2016 50 3 1128 1138 10.1287/trsc.2015.0630 Xu X Shang J Wang H Chiang WC Optimal production and inventory decisions under demand and production disruptions Int J Prod Res 2016 54 1 287 301 10.1080/00207543.2015.1073402 Yavari M Ajalli P Suppliers’ coalition strategy for green-Resilient supply chain network design J Ind Prod Eng 2021 38 3 197 212 Yavari M Zaker H Designing a resilient-green closed loop supply chain network for perishable products by considering disruption in both supply chain and power networks Comput Chem Eng 2020 134 106680 10.1016/j.compchemeng.2019.106680 Yavari M Enjavi H Geraeli M Demand management to cope with routes disruptions in location-inventory-routing problem for perishable products Res Transp Bus Manag 2020 37 100552 10.1016/j.rtbm.2020.100552 Zahiri B Zhuang J Mohammadi M Toward an integrated sustainable-resilient supply chain: A pharmaceutical case study Transp Res Part E: Logist Transp Rev 2017 103 109 142 10.1016/j.tre.2017.04.009 Zhang Y Qi M Lin WH Miao L A metaheuristic approach to the reliable location routing problem under disruptions Transp Res Part E: Logist Transp Rev 2015 83 90 110 10.1016/j.tre.2015.09.001 Zhong S Cheng R Jiang Y Wang Z Larsen A Nielsen OA Risk-averse optimization of disaster relief facility location and vehicle routing under stochastic demand Transp Res Part E: Logist Transp Rev 2020 141 102015 10.1016/j.tre.2020.102015	0	PMC9715419	NO-CC CODE	2022-12-03 23:20:15	no	Process Integr Optim Sustain. 2022 Dec 2;:1-22	utf-8	null	null	null	oa_other
==== Front J Porous Mater Journal of Porous Materials 1380-2224 1573-4854 Springer US New York 1394 10.1007/s10934-022-01394-z Article The preparation of ultrathin and porous electrospinning membranes of HKUST-1/PLA with good antibacterial and filtration performances Zhu Yanyan Yang Dangsha Li Jiangen Yue Zhenqing Zhou Jingheng Wang Xinlong [email protected] grid.410579.e 0000 0000 9116 9901 School of Chemical Engineering, Nanjing University of Science & Technology, Nanjing, 210094 China 2 12 2022 19 18 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. Developing degradable filter membranes that inhibit bacterial infection for preventing particle matter and infectious disease has been a research hotspot. Here, the fiber membranes of polylactic acid (PLA)/HKUST-1 with porous structure through the entire fiber matrix were prepared by electrospinning method. Due to the HKUST-1 incorporation and the presence of pore through fiber, the hydrophobicity of prepared membranes had been improved. The PLA/HKUST-1 membranes exhibited the good antibacterial activity against Escherichia coli and Staphylococcus aureus, and the antibacterial rate for S. aureus reached 99.99%. The filtration performance of PLA/HKUST-1 membranes was better than that of the melt-blown fabric although their thickness was only about one-third of the thickness of the currently commercial polypropylene melt-blown fabric. Supplementary Information The online version contains supplementary material available at 10.1007/s10934-022-01394-z. Keywords MOFs Antibacterial Filtration Electrospinning Science and Technology Support Program (Social Development) of Jiangsu Province of ChinaBE 2020709 BE 2020709 BE 2020709 BE 2020709 BE 2020709 BE 2020709 Zhu Yanyan Yang Dangsha Li Jiangen Yue Zhenqing Zhou Jingheng Wang Xinlong http://dx.doi.org/10.13039/501100012246 Priority Academic Program Development of Jiangsu Higher Education Institutions PAPD PAPD PAPD PAPD PAPD PAPD Zhu Yanyan Yang Dangsha Li Jiangen Yue Zhenqing Zhou Jingheng Wang Xinlong ==== Body pmcIntroduction The effects of particulate matter (PM) with different sizes and highly infectious diseases such as COVID-19 on the human health have driven mankind to develop more effective protective materials [1–3]. Among them, the filter membrane is one of the most important materials, which can be widely used in various protective equipment such as protective face masks and antismog window screens. The electrospinning fiber membranes in filtration application have attracted considerable attention due to their small diameter, high surface-to-volume ratio, small inter-fiber pore sizes, low-cost and relatively high production rate [4–11], etc. By incorporating appropriate polymers with functional materials, degradable filter membranes can be prepared to solve the environmental problems caused by common non-degradable filter materials, and at the same time, the membranes can be endowed with functions such as antibacterial properties [12–14]. The morphology, hydrophilicity or hydrophobicity, filtration efficiency and pressure drop of the electrospinning composite membranes can be controlled by changing contents of particles added into the polymer matrix because the presence of particles could influence the roughness, chemistry, and porosity of the surface of the prepared membranes [15–17]. The incorporation of particles into electrospinning polymer nanofibers have been explored working in drug delivery [18, 19], water treatment, air filtration and antibacterial applications [14, 20–22]. A lot kind of particles such as SiO2, Ag, TiO2, and graphene oxides have been used for different purposes [23, 24]. Porous structures on the electrospinning fiber surface or through the entire fiber matrix can increase the surface to volume and broaden the range of applications for membranes. The preparation of porous polymer nanofibers with high surface areas is still a challenge. Metal–organic frameworks (MOFs) are formed by the coordination of metal ions with organic ligands, and have the advantages of high specific surface area, large porosity, variable structure, and adjustable channel [24–26]. MOFs have become the attractive antimicrobial materials for applications in which a tunable antibacterial agent is required. Different from the action mechanism of small molecular antibacterial agents, the antibacterial performance of MOFs is mainly due to release of metal ions (such as Ag+, Cu2+, Zn2+, and Co2+) from structure collapse, which destroy the cell membrane, lipid peroxidation, and DNA degradation to kill bacteria [27–30]. In this study, the fiber membranes with a porous structure consisted of polylactic acid (PLA) and HKUST-1 were prepared by electrospinning method. PLA was a biodegradable polymer [31], which could be degraded into CO2 and H2O. As a kind of Cu-MOFs, HKUST-1 had previously been reported to have good antibacterial activity against Candida albicans, Aspergillus niger and Aspergillus oryzae etc. [1, 12]. Therefore, the prepared membranes had biodegradability and antibacterial performance. When they were used as filter material, they could overcome the environmental problems caused by the non-degradable polypropylene melt blown cloth which was widely used as at present. In addition, due to its good antibacterial performance, the membranes could reduce the influence of bacteria and other microorganisms in the air on people’s health. Experimental The preparation of HKUST-1 HKUST-1 was synthesized by the hydrothermal method [30]. Cu(NO3)2⋅3H2O (12 mM) was dissolved in 25 mL of deionized water and 8 mM of H3BTC was dissolved in 25 mL of DMF. At room temperature, the two solutions were mixed and stirred for 10 min. The mixture was transferred into the reaction kettle and kept in the oven at 105 ℃ for 24 h. The HKUST-1 particles were obtained by centrifugation and washed with water and ethanol for three times, respectively. Preparation of the fiber membranes PLA (18 g) and 105.6 g of dichloromethane (DCM) were added to a round-bottomed flask, and the mixture was sealed and stirred for 2 h to dissolve PLA in DCM completely. The HKUST-1 particles in DMF suspension were put it into the above round-bottomed flask and stirred to obtain electrospinning solution (The HKUST-1 content was 0, 0.6, and 3.0 wt% of the PLA mass and the ratio of DCM to DMF was 8:2). Then, the spinning solution was pit into a syringe. The electrospinning process proceeded under 18 kV. The receiving distance was set as 17 cm and the pushing injection speed was 0.004 mm/s. The temperature and the humidity were controlled at about 40 ℃ and 50–80% respectively. The fiber membranes with a thickness of about 0.036 mm were obtained and named as PLA, PLA-K0.6 and PLA-K3 according to the content of HKUST-1 (0, 0.6, and 3.0 wt%) in sequence. Figure 1 showed the preparation process of membranes.Fig. 1 The preparation process of membranes General characterization Field emission scanning electron microscope (FEI Quanta 250FEG) was used to observe the surface morphology of the samples. X-ray Diffraction (XRD, Bruker D8-ADVANCE) was performed on Cu Kα radiation with a range of 5°–70°. The Fourier transform infrared test was performed with Shimadzu 8400S and the scanning range was 400–4000 cm−1. TGA was carried on the GA/SDTA85 (DTG, Shimadzu DTG-60) between 20 and 800 ℃ in a nitrogen atmosphere. The concentration of Cu2+ released from PLA-K3 was obtained by inductively coupled plasma mass spectrometry (ICP-MS, ThermoFisher I CAPQ). The membrane was cut discs with a diameter of 5 cm and immersed in water. The bottle with water and discs was placed in a constant temperature oscillator to shake for different time. Then, the solution was centrifuged for detection. The tensile test of the membrane was carried out with the TY8000A-500N electromechanical universal testing machine (Jiangsu Tianyuan Testing Equipment Co., Ltd) in accordance with the ASTMD638 standard. The sample was 4 cm × 0.5 cm. The speed was 10 N/min and each sample was tested for 5 times. Antibacterial test of fiber membrane The antibacterial abilities of fiber membranes were evaluated by bacterial colony counting method using Escherichia coli and Staphylococcus aureus [7]. The target bacteria were cultivated at 37 ℃ in Luria–Bertani broth until the bacteria concentration was at approximately 109 CFU/mL. The concentration of bacterial suspension was diluted with sterile water to (1–5) × 106 CFU/mL. Then, 2 mL of 106 CFU/mL suspension was put into 40 mL water suspension containing the membrane discs of 5 cm diameter. The bottles with mixture were set on the shaker at 37 ℃ and shaken for 24 h. Subsequently, 1 mL of suspension after shake period was cultivated at 37 ℃. The antibacterial rate (R) was calculated via the Eq. (1):1 R\%=N0-NN0∗100 where N0 and N are the average number of viable bacteria on a reference sample without discs and on the sample containing discs after antibacterial tests, respectively. Each sample was tested for 3 times. PM filtration measurement The PM adsorption experiment was carried out with a self-assembled experimental device as shown in Fig. S1 [4, 16, 20]. The dust detector was the LD-5/J laser dust meter from Nanjing Trinyaer Environmental Protection Technology Co., Ltd. The PM particles were produced by burning cigarette in the upper container. The prepared membrane was cut into 7 cm × 7 cm square and sandwiched between upper and lower containers. The air in the upper container was drawn through the prepared membrane into the lower container using a vacuum pump at a flow rate of 2 L/min. The removal efficiency Re was calculated via Eq. (2):2 Re=Cabove-CbelowCabove×100\% where Cabove was the PM concentration in the upper container (μg/m3) and Cbelow was the PM concentration in the lower container (μg/m3). Statistical analysis One-way analysis of variance (ANOVA) with Tukey’s post hoc test was used for statistical analysis of between-group and within-group data. The p < 0.05, p < 0.01 and **p < 0.001 were accepted as statistically significant. Results and discussion The morphology of fiber membranes The HKUST-1 particles were synthesized through the solvothermal reaction according to the previous report. The SEM image, XRD and IR of the prepared HKUST-1 were displayed in Fig. S2. The HKUST-1 particles had the octahedral morphology and their size was about 16 μm. The results of XRD and IR were consistent with the literature as reported [30]. Figure 2 showed the morphology of the fiber membranes. The fibers with diameter of about 1 μm were stacked together and many interspaces were formed. As seen from the enlarged image in the upper right corner, every fiber contained a large number of nano pores which were through the entire fiber. The formation of pores on the fiber primarily was attributed to the phase separation of the different components of the electrospinning solution, specifically, the separation between the polymers and solvents, the separation between the polymers and the non-solvents [32, 33]. The existence of pores on fiber was beneficial to increase the surface to volume or surface to weight ratio of the fiber membranes. Because the size of HKUST-1 particles was larger than the diameter of fibers, it could be seen that they were wrapped on the membrane by the intersecting fibers. The mapping of Cu showed that the HKUST-1 particles were dispersed uniformly in the membrane containing 0.6 wt% of HKUST-1. However, the agglomeration of HKUST-1 particles could be observed for PLA-K3 containing 3 wt% of HKUST-1 [4].Fig. 2 The SEM of a PLA; b PLA-K0.6; c PLA-K3 The contact angles and mechanical properties of fiber membranes The contact angle was related to the surface roughness and chemical properties of the material, which could reflect the wettability of the fiber membrane and directly affected the adsorption of bacteria or pollution particles on the surface [34, 35]. As shown in Fig. 3a, the contact angle of the PLA membrane was 72.31°, and the values for PLA-K0.6 and PLA-K3 were increased 83.41° and 92.54°, respectively. Both the embedded HKUST-1 particles on membrane surface and the resulting change in the chemical composition in the surface increased the contact angle. As shown in Fig. 3b, the tensile strength of PLA membrane was 1.43 MPa. When 0.6 wt% of HKUST-1 was added, the tensile strength of PLA-K0.6 membrane was 2.28 MPa. However, the tensile strength of PLA-K3 membrane was slightly reduced to 2.00 MPa. It was reported that the MOF particles in the composites could effectively transfer stresses and lead to the enhancement in tensile strength [36, 37]. As the amount of MOF increased, agglomeration of particles occurred, which weakened the inter-fiber forces and reduced the tensile strength [7]. (A typical stress–strain curve was shown in Fig. S3).Fig. 3 Bar diagram of contact angle (a) and mechanical properties (b) of fiber membranes Antibacterial rate of HKUST-1/PLA fiber membranes The antibacterial properties of prepared fiber membranes against both Gram-positive Staphylococcus aureus and Gram-negative E. coli were assessed and shown in Fig. 4. The neat PLA membrane adsorbed bacteria as had frequently been proved [36]. When E. coli contacting with PLA-K3, the number of colonies in the petri dishes was reduced compared with PLA membrane and the antibacterial rate was 91%. For S. aureus, there were no colonies growing in the petri dishes (Fig. 4) and the antibacterial rate reached 99.99% (Fig. 5a) with PLA-K3. The enhancement in bactericidal properties for fiber membranes with HKUST-1 was mainly due to release of Cu2+ from HKUST-1 in the membrane. Figure 5b showed the Cu2+ concentration change in solution when PLA-K3 was soaked in water at different time. The concentration of Cu2+ in solution gradually increased and reached a maximum of 22.06 ppb at 12 h. Then, the concentration of Cu2+ in the solution began to decrease. The Cu2+ from the release of PLA-K3 entered the bacterial cell, and damaged the structure of DNA and some essential enzymes [24, 38–40].Fig. 4 Agar plate experiment of antibacterial activity of PLA and PLA-K a E. coli; b S. aureus Fig. 5 a Antibacterial rate of E. coli and S. aureus by PLA-K; b Cu2+ release curves of PLA-K3 in aqueous solution at different time The bacterial morphologies cultured on PLA membrane and PLA-K3 were compared in Fig. 6 to further understand the antibacterial mechanism of the membranes. On the PLA membrane, E. coli and S. aureus exhibited smooth surface and intact morphology with rod and rounded shapes, respectively. However, they clumped, deformed, collapsed, and even dissolved on the PLA-K3 membrane. Table 1 showed the comparison of the antibacterial rates of the prepared membranes against E. coli and S. aureus with those reported. The presence of pores on the fibers increased the specific surface area of the membrane, which would help bacteria to adhere on the surface of membrane and Cu2+ to find bacteria quickly after release. Therefore, the membrane had a good antibacterial effect when the addition amount of HKUST-1 was only 3 wt%.Fig. 6 SEM images of PLA (a) and PLA-K3 (b) after 24 h in E. coli; SEM images of PLA (c) and PLA-K3 (d) after 24 h in S. aureus Table 1 Comparison of antibacterial rate of HKUST-1/PLA membrane with others Fiber membranes Antibacterial rate (%) References E S PLA-K3 91 99.99 This work 1.7%CuCl2L2/CA 75 80 [41] 9.1%Cu-MOFs/PLA 99 99 [42] 10%HKUST-1/CS/PVA 99 99 [39] GO/SF 64 58 [43] Filtration performance of HKUST-1/PLA fiber membranes The photos of the filtration effect to PM2.5 and PM10 of the HKUST-1/PLA fiber membranes at different time were shown in Fig. S4. It could be seen that the membrane had good filtration performance in the case of high concentration or low concentration of PM. Figure 7a showed the filtration efficiency curves of melt-blown fabric (MB) and PLA-K3 with time. It was obvious that the filtration performance of PLA-K3 was very stable. The filtration efficiency of PLA-K3 for PM2.5 and PM10 was close to 100% within filtration time, while the filtration efficiency of MB fabric was only 24% at 2 min. Then, the filtration efficiency slowly increased with time but it was still lower than that of the PLA-K3. The above results showed that the filtration efficiency of the prepared fiber membrane was higher than that of the melt-blown fabric, especially in the high concentration PM2.5 environment, where the filtration efficiency was four times higher than that of the melt-blown fabric. In addition, PLA-K3 had the advantage of being ultra-thin with a thickness of 0.036 mm, which was only 1/3 of that of melt-blown cloth. This would reduce the amount of polymer used, thereby lightening the environmental burden that may be caused by the large-scale application of non-degradable polymers. However, it can be seen in Fig. 7b that the PLA-K3 had higher pressure drop than that of the melt-blown fabric. Table 2 showed a comparison of the filtration efficiency for PM2.5 between PLA-K3 membrane and other spinning membranes reported. The quality factor of PLA-K3 was 0.063, which was sufficient to demonstrate its good filtration performance further. SEM was used to observe the membrane after filtration as shown in Fig. 7e and f, and a lot of deposits were observed on the membranes.Fig. 7 The filtering efficiency of MB and PLA-K3 at different time (a); the pressure drop of MB and PLA-K3 (b); thickness of melt-blown fabric (c) and PLA-K3 (d); SEM of the membranes after adsorption (e) MB and (f) PLA-K3 Table 2 Comparison of particle filtration efficiency of PLA-K3 with others Fiber membranes Particle (um) η (%) ΔP (Pa) Qf References MB 2.5 80.4 70 0.023 This work 3%HKUST-1/PLA 2.5 99.9 110 0.063 This work 9%ZIF-8/9%PAA 5 99.6 172 0.03 [4] 2.5%ZIF-8/PP/PVA 0.2–4.6 96.5 34 0.099 [17] GO/PI-6/PAN 2.5 99.5 92 0.058 [20] Conclusion HKUST-1 particles were prepared and incorporated into the PLA electrospinning fiber membranes. When the HKUST-1 content was 3.0 wt%, the removal efficiency was clos to 100% for PM2.5 and PM10 within 20 min and the quality factor of PLA-K3 was 0.063. The PLA-K3 had an antibacterial rate of 99.99% for S. aureus. The prepared membrane could be used in the field of masks instead of melt-blown fabric of polypropylene, and solved the problem of non-degradability and non-functionalization of melt-blown fabric. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 1638 kb) Acknowledgements This work was supported by Science and Technology Support Program (Social Development) of Jiangsu Province of China (BE 2020709) and Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). Author contributions All authors contributed to the study conception and design. The manuscript was written by YZ and revised by XW. Material preparation, data collection and analysis were performed by DY and JL. The data verification was done by ZY and JZ. All authors commented on previous versions of the manuscript. Final manuscript read and approved by all authors. Declarations Conflict of interest The authors declared that they have no conflicts of interest to this work. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Khan NA Hasan Z Jhung SH Adsorptive removal of hazardous materials using metal-organic frameworks (MOFs): a review J. Hazard. Mater. 2013 244–245 444 456 10.1016/j.jhazmat.2012.11.011 23195596 2. Tomczak A Miller AB Weichenthal SA Long-term exposure to fine particulate matter air pollution and the risk of lung cancer among participants of the Canadian National Breast Screening Study Int. J. Cancer 2016 139 1958 1966 10.1002/ijc.30255 27380650 3. Jack LSE Ulrik CS Physical activity, air pollution, and the risk of asthma and chronic obstructive pulmonary disease Am. J. Respir. Crit. Care Med. 2016 194 855 865 10.1164/rccm.201510-2036OC 27653737 4. Guo J Hanif A Shang J PAA@ZIF-8 incorporated nanofibrous membrane for high-efficiency PM2.5 capture Chem. Eng. J. 2021 405 126584 10.1016/j.cej.2020.126584 5. Chen H Malheiro A Blitterswijk C Direct writing electrospinning of scaffolds with multidimensional fiber architecture for hierarchical tissue engineering ACS Appl. Mater. Interfaces 2017 9 38187 38200 10.1021/acsami.7b07151 29043781 6. Zhu M Li J Yu J Superstable and intrinsically self-healing fibrous membrane with bionic confined protective structure for breathable electronic skin Angew. Chem. Int. Ed. Engl. 2022 61 e202200226 10.1002/anie.202200226 35212123 7. Zheng X Zhang Y Zou L Robust ZIF-8/alginate fibers for the durable and highly effective antibacterial textiles Colloid Surf. B 2020 193 111127 10.1016/j.colsurfb.2020.111127 8. Yin J Xu L Batch preparation of electrospun polycaprolactone/chitosan/aloe vera blended nanofiber membranes for novel wound dressing Int. J. Biol. Macromol. 2020 160 352 363 10.1016/j.ijbiomac.2020.05.211 32470578 9. Cleeton C Keirouz A Chen X Electrospun nanofibers for drug delivery and biosensing ACS Biomater. Sci. Eng. 2019 5 4183 4205 10.1021/acsbiomaterials.9b00853 33417777 10. Li W Luo T Yang Y Formation of controllable hydrophilic/hydrophobic drug delivery systems by electrospinning of vesicles Langmuir 2015 31 5141 5146 10.1021/la504796v 25897828 11. Xu Y Zhu J-W Fang J-B Electrospun high-thermal-resistant inorganic composite nonwoven as lithium-ion battery separator J. Nanomater. 2020 2020 1 10 10.1155/2020/3879040 12. Cun JE Fan X Pan QQ Copper-based metal-organic frameworks for biomedical applications Adv. Colloid Interface 2022 305 102689 10.1016/j.cis.2022.102686 13. Fleuret C Andreani AS Laine E Complex wing spar design in carbon fiber reinforced composite for a light aerobatic aircraft Mech. Ind. 2016 17 19 10.1051/meca/2016032 14. Lu T Cui J Qu Q Multistructured electrospun nanofibers for air filtration: a review ACS Appl. Mater. Interfaces 2021 13 23293 23313 10.1021/acsami.1c06520 33974391 15 Deng Y Lu T Cui J Morphology engineering processed nanofibrous membranes with secondary structure for high-performance air filtration Sep. Purif. Technol. 2022 294 121093 10.1016/j.seppur.2022.121093 16. Zhu Q-H Zhang G-H Zhang L Self-charge-carrying air filter by in situ polymerization to avoid charge dissipation and potential material poisoning Chem. Eng. J. 2022 449 13778 10.1016/j.cej.2022.137788 17 Li TT Fan Y Cen X Polypropylene/polyvinyl alcohol/metal-organic framework-based melt-blown electrospun composite membranes for highly efficient filtration of PM2.5 Nanomaterials 2020 10 2025 10.3390/nano10102025 33066527 18. Gautam S Singhal J Lee HK Drug delivery of paracetamol by metal-organic frameworks (HKUST-1): improvised synthesis and investigations Mater. Today Chem. 2022 23 100647 10.1016/j.mtchem.2021.100647 19. Horcajada P Chalati T Serre C Porous metal-organic-framework nanoscale carriers as a potential platform for drug delivery and imaging Nat. Mater. 2010 9 172 178 10.1038/nmat2608 20010827 20 Dai H Liu X Zhang C Electrospinning polyacrylonitrile/graphene oxide/polyimide nanofibrous membranes for High-efficiency PM2.5 filtration Sep. Purif. Technol. 2021 276 119243 10.1016/j.seppur.2021.119243 21. Xiao Y Wang Y Zhu W Development of tree-like nanofibrous air filter with durable antibacterial property Sep. Purif. Technol. 2021 259 118135 10.1016/j.seppur.2020.118135 22. Zhang C Yao L Yang Z Graphene oxide-modified polyacrylonitrile nanofibrous membranes for efficient air filtration ACS Appl. Nano Mater. 2019 2 3916 3924 10.1021/acsanm.9b00806 23. Rahmati Z Abdi J Vossoughi M Ag-doped magnetic metal organic framework as a novel nanostructured material for highly efficient antibacterial activity Environ. Res. 2020 188 109555 10.1016/j.envres.2020.109555 32559687 24. Slavin YN Asnis J Hafeli UO Metal nanoparticles: understanding the mechanisms behind antibacterial activity J. Nanobiotechnol. 2017 15 65 10.1186/s12951-017-0308-z 25. Zhu N Zou Y Huang M A sensitive, colorimetric immunosensor based on Cu-MOFs and HRP for detection of dibutyl phthalate in environmental and food samples Talanta 2018 186 104 109 10.1016/j.talanta.2018.04.023 29784336 26. Adam R Mon M Greco R Self-assembly of catalytically active supramolecular coordination compounds within metal-organic frameworks J. Am. Chem. Soc. 2019 141 10350 10360 10.1021/jacs.9b03914 31194534 27 Firouzjaei MD Shamsabadi AA Sharifian Gh M A novel nanocomposite with superior antibacterial activity: a silver-based metal organic framework embellished with graphene oxide Adv. Mater. Interfaces 2018 5 1701365 10.1002/admi.201701365 28. Hasan MN Bera A Maji TK Sensitization of nontoxic MOF for their potential drug delivery application against microbial infection Inorg. Chim. Acta 2021 523 120381 10.1016/j.ica.2021.120381 29. Wang H Yu D Fang J Renal-clearable prphyrinic metal-organic framework nanodots for enhanced photodynamic therapy ACS Nano 2019 13 9206 9217 10.1021/acsnano.9b03531 31408319 30. Bouson S Krittayavathananon A Phattharasupakun N Antifungal activity of water-stable copper-containing metal-organic frameworks R. Soc. Open Sci. 2017 4 170654 10.1098/rsos.170654 29134075 31 Deng Y Lu T Cui J Bio-based electrospun nanofiber as building blocks for a novel eco-friendly air filtration membrane: a review Sep. Purif. Technol. 2021 277 119623 10.1016/j.seppur.2021.119623 32. Chen PY Tung SH One-step electrospinning to produce nonsolvent-induced macroporous fibers with ultrahigh oil adsorption capability Macromolecules 2017 50 2528 2534 10.1021/acs.macromol.6b02696 33. Li CL Wang DM Deratani A Insight into the preparation of poly(vinylidene fluoride) membranes by vapor-induced phase separation J. Membr. Sci. 2010 361 154 166 10.1016/j.memsci.2010.05.064 34. Liu Y Wang D Sun Z Preparation and characterization of gelatin/chitosan/3-phenylacetic acid food-packaging nanofiber antibacterial films by electrospinning Int. J. Biol. Macromol. 2021 169 161 170 10.1016/j.ijbiomac.2020.12.046 33309663 35. Deng L Zhang X Li Y Characterization of gelatin/zein nanofibers by hybrid electrospinning Food Hydrocolloid 2018 75 72 80 10.1016/j.foodhyd.2017.09.011 36. Kiadeh SZH Ghaee A Farokhi M Electrospun pectin/modified copper-based metal-organic framework (MOF) nanofibers as a drug delivery system Int. J. Biol. Macromol. 2021 173 351 365 10.1016/j.ijbiomac.2021.01.058 33450340 37. Ghaee A Karimi M Lotfi-Sarvestani M Preparation of hydrophilic polycaprolactone/modified ZIF-8 nanofibers as a wound dressing using hydrophilic surface modifying macromolecules Mater. Sci. Eng. C 2019 103 109767 10.1016/j.msec.2019.109767 38. Shams S Ahmad W Memon AH Cu/H3BTC MOF as a potential antibacterial therapeutic agent against Staphylococcus aureus and Escherichia coli New J. Chem. 2020 44 17671 17678 10.1039/d0nj04120c 39. Wang S Yan F Ren P Incorporation of metal-organic frameworks into electrospun chitosan/poly (vinyl alcohol) nanofibrous membrane with enhanced antibacterial activity for wound dressing application Int. J. Biol. Macromol. 2020 158 9 17 10.1016/j.ijbiomac.2020.04.116 32353508 40. Pettinari C Pettinari R Di Nicola C Antimicrobial MOFs Coord. Chem. Rev. 2021 446 214121 10.1016/j.ccr.2021.214121 41. Demirdogen RE Kilic D Emen FM Novel antibacterial cellulose acetate fibers modified with 2-fluoropyridine complexes J. Mol. Struc. 2020 1204 127537 10.1016/j.molstruc.2019.127537 42. Liu Z Ye J Rauf A A flexible fibrous membrane based on copper(II) metal-organic framework/poly(lactic acid) composites with superior antibacterial performance Biomater. Sci. 2021 9 3851 3859 10.1039/d1bm00164g 33890581 43. Wang SD Ma Q Wang K Improving Antibacterial activity and biocompatibility of bioinspired electrospinning silk fibroin nanofibers modified by graphene oxide ACS Omega 2018 3 406 413 10.1021/acsomega.7b01210 30023780	0	PMC9715420	NO-CC CODE	2022-12-03 23:20:15	no	J Porous Mater. 2022 Dec 2;:1-9	utf-8	null	null	null	oa_other
==== Front Asia Pac J Manag Asia Pacific Journal of Management 0217-4561 1572-9958 Springer US New York 9861 10.1007/s10490-022-09861-6 Article Toward an institution-based paradigm Peng Mike W. [email protected] 1Mike W. Peng (PhD, University of Washington) is the Jindal Chair of Global Strategy at the Jindal School of Management, University of Texas at Dallas; a National Science Foundation career award winner; and a Fellow of the Academy of International Business and Asia Academy of Management. His research interests are global strategy, international business, and the institution-based view. Wang Joyce C. 2Joyce C. Wang (PhD, University of Texas at Dallas) is an associate professor of management and entrepreneurship at Herberger Business School, St. Cloud State University. Her research centers on an institution-based view of corporate governance. Kathuria Nishant 3Nishant Kathuria (PhD, University of Texas at Dallas) is an assistant professor of management at the College of Business and Economics, Towson University. His research interests include corporate social responsibility, corporate social irresponsibility, and socioeconomic inequalities. Shen Jia 4Jia Shen is a PhD candidate at the Jindal School of Management, University of Texas at Dallas. Her research interests are entrepreneurship, global strategy, and the institution-based view. Welbourne Eleazar Miranda J. 5Miranda J. Welbourne Eleazar (JD, University of Michigan; PhD, University of Texas at Dallas) is an assistant professor of management and entrepreneurship and John L. Miclot Faculty Fellow in Entrepreneurship at Tippie College of Business, University of Iowa. Her research focuses on how entrepreneurs and firm leaders make difficult decisions that significantly impact society, including their responses to adversity and ethical issues. 1 grid.267323.1 0000 0001 2151 7939 University of Texas at Dallas, Richardson, TX USA 2 grid.264047.3 0000 0001 0738 3196 St. Cloud State University, St Cloud, MN USA 3 grid.265122.0 0000 0001 0719 7561 Towson University, Towson, MD USA 4 grid.267323.1 0000 0001 2151 7939 University of Texas at Dallas, Richardson, TX USA 5 grid.214572.7 0000 0004 1936 8294 University of Iowa, Iowa City, IA USA 2 12 2022 130 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. As part of the broader intellectual movement throughout the social sciences that is centered on new institutionalism, the institution-based view has emerged as a leading perspective in the strategic management literature. This article (1) traces the emergence of the institution-based view, (2) reviews its growth in the last two decades, and (3) responds to three of its major criticisms. We also identify four promising research directions—deglobalization and sanctions, competitive dynamics, hybrid organizations, and corporate social responsibility. Overall, we demonstrate that the thriving research on institutions has culminated in an institution-based paradigm, which has significant potential for future growth. Keywords Competitive dynamics Corporate social responsibility Deglobalization Hybrid organizations Institutions Institution-based view New institutionalism Paradigm Sanctions ==== Body pmcAs part of the broader intellectual movement throughout the social sciences that is centered on new institutionalism (North, 1990; Ostrom, 2005; Scott, 1995; Williamson, 2000), the institution-based view has emerged as a leading perspective in the strategic management literature. The institution-based view is characterized by its emphasis on institutions as rules and norms, its quest for dynamic rather than static explanations of firm behavior, and its embrace of interdisciplinary approaches. While the term “institution-based view” was coined by Peng (2002) in the pages of the Asia Pacific Journal of Management, its contributions come from numerous scholars worldwide. Thanks to these endeavors, the institution-based view has gained tremendous legitimacy and flourished in the last two decades (Opper, 2022). At the same time, it has attracted debates and criticisms, necessitating further assessments, responses, and progress. It is in this spirit that the current article—20 years after Peng (2002)—is written. We address the following four questions: (1) What is behind the emergence of the institution-based view? (2) What has fueled its growth? (3) What are the leading criticisms, and how can these criticisms be addressed? (4) What are some of the promising new research directions? Overall, we advance the argument that different lines of research underpinned by the institution-based view have culminated in an integrative paradigm. Prior to emergence According to Kuhn (1970), a paradigm is “universally recognized scientific achievements that, for a time, provide model problems and solutions to a community of practitioners.” Instead of having only one dominant paradigm, the management field has always had several paradigms (Conner, 1991; Pfeffer, 1993). In strategic management, prior to the emergence of the institution-based view, two paradigms can be identified: the industry-based view and the resource-based view. Anchored to the five forces framework, “the essence of this [industry-based] paradigm is that a firm’s performance in the marketplace depends critically on the characteristics of the industry environment in which it competes” (Porter, 1981: 610). Focusing on the valuable, rare, and inimitable resources and capabilities, the resource-based view (Barney, 2001), complemented by a dynamic capabilities perspective (Teece, 2007), has enjoyed ascendancy as the second paradigm. Paradigm shifts appear in response to anomalies that existing paradigms cannot resolve (Kuhn, 1970). Both the industry-based and resource-based views have been criticized for their lack of attention to contexts. Take cost leadership, the most widely practiced (and taken-for-granted) strategy in the industry-based view. Transplanted to foreign markets, a cost leadership strategy may be declared “illegal.” Ignoring the context of host-country antidumping laws, a firm that single-mindedly pursues a cost leadership strategy in exporting its products may be sued by host-country incumbents for dumping (selling below cost). The upshot? Heavy fines for such “illegal” conduct. In other words, institution-based constraints such as antidumping laws have been overlooked by the industry-based view. The resource-based view has similarly been challenged for its “little effort to establish appropriate contexts” (Priem & Butler, 2001: 32). Valuable resources and capabilities in some contexts may become nonvaluable in other contexts. For example, in least developed countries (LDCs), leading multinationals famous for their world-class capabilities are typically not among the most successful foreign firms (Cuervo-Cazurra & Genc, 2008). Instead, multinationals from other less developed economies, with less advanced capabilities, often do better. The reason may be that multinationals from other less developed economies have a much better understanding of how to effectively navigate the context of LDCs. In other words, excellent capabilities honed in the context of developed economies do not go far in the context of LDCs. Barney (2001: 52) acknowledges the validity of this criticism, agreeing that “the value of a firm’s resources must be understood in the specific market context within which a firm is operating.” The challenge is: How can scholars theorize about such a “context”? In the absence of theoretical (or paradigmatic) breakthrough, scholars encountering new anomalies are likely to be constrained by the straightjacket of old paradigms (Kuhn, 1970). For example, in the very first paper on firm strategy in China published in the Strategic Management Journal, Tan and Litschert (1994) are constrained by the then prevailing industry-based and resource-based paradigms. They end up relying on the literature on firm strategy in regulated industries in the West to derive their hypotheses on firm strategies in the electronics industry in China. But during the time of their survey (1990-91), the electronics industry is one of the least regulated industries in China (with the most vibrant market competition and foreign investment) (Tan & Litschert, 1994: 5). It is not industry-specific regulatory changes that drive the strategic changes they find. Rather, such strategic changes are driven by large-scale institutional transitions sweeping throughout the country—“fundamental and comprehensive changes introduced to the formal and informal rules of the game that affect organizations as players” (Peng, 2003: 275). In summary, efforts to understand firm behavior around the world have created anomalies for the industry-based and resource-based paradigms, which originate from research on firms in the United States and which assume a relatively stable, market-friendly environment. Admittedly, the environment has long been featured in the industry-based and resource-based paradigms, but such an environment tends to be task environment, measured by economic variables such as market demand and technological change (Dess & Beard, 1984). What about the larger environment such as the home country (McGahan & Victor, 2010) and the host country (Makino et al., 2004)? What is needed is theoretical sublimation that goes above and beyond the task environment to shed light on the drivers of firm behavior that the two existing paradigms cannot fully explain (Peng et al., 2005). Overall, both the push and pull effects are at work. While the lack of adequate attention to contexts in the industry-based and resource-based views has pushed for a new paradigm (Peng et al., 2009: 65), the development of new institutionalism research throughout the social sciences has pulled scholars to develop an institution-based view—discussed next. Emergence Starting from the idea that institutions can be conceptualized as “the rules of the game” (North, 1990), the institution-based view argues that firm behavior and performance are determined, at least in part, by the institutional conditions and transitions confronting firms (Peng & Heath, 1996; Peng et al., 2009; Peng et al., 2008). The origins and emergence of the institution-based view have been discussed in a series of articles by Meyer and Peng (2016), Peng (2002, 2005, 2014), and Peng et al. (2008, 2009, 2018b). Other reviews of the institution-based literature can be found in Aguilera and Grogaard (2019), Ahuja and Yayavaram (2011), Ahuja et al. (2018), Cuervo-Cazurra et al. (2019b), Dielman et al. (2022), Doh et al. (2017), Greenwood et al. (2011), Heugens and Lander (2009), Ingram and Silverman (2002), Jackson and Deeg (2008, 2019), Kostova et al. (2020), Marquis and Raynard (2015), Opper (2022), Su (2021), Sun et al. (2021), Sun et al. (2020), Zhao et al. (2017), and others. Collectively, these articles document the growth of the institution-based view from a relatively peripheral research stream to a leading pillar in the literature. Virtually all institutionally-minded scholars—regardless of disciplinary backgrounds—share a consensus that “institutions matter” (DiMaggio & Powell, 1983; North, 1990; Oliver, 1997; Ostrom, 2005; Scott, 2008; Williamson, 2000). From this basic proposition, scholars have tackled “the harder and more interesting issues of how they matter, under what circumstances, to what extent, and in what ways” (Powell, 1996: 297). It is such a quest to enhance our understanding of how institutions matter that leads to the proliferation of new institutionalism research throughout the social sciences and management literatures, including work on the institution-based view. Summarizing and extending earlier work, the institution-based view has advanced two most fundamental propositions (Peng et al., 2009: 67–68): Proposition 1: Managers and firms rationally pursue their interests and make strategic choices within the formal and informal constraints in a given institutional framework. Proposition 2: While formal and informal institutions combine to govern firm behavior, in situations where formal constraints are unclear or fail, informal constraints will play a larger role in reducing uncertainty, providing guidance, and conferring legitimacy and rewards to managers and firms. Different from Kuhn’s (1970) model of a new paradigm displacing an old one, the emergence of a new paradigm in management does not necessarily result in such displacement. This may be due to the fact that the management field is diverse enough to accommodate multiple paradigms (Conner, 1991; Pfeffer, 1993). Positioning itself as a third leg for a strategy tripod, the institution-based view has always emphasized that it complements existing theories such as the industry-based and resource-based views (Peng et al., 2009). Studies have leveraged the strategy tripod to generate interesting insights (Gao et al., 2010; Krull et al., 2012; Lahiri et al., 2020; Lu et al., 2010; Su et al., 2016; Yamakawa et al. 2008). Being theoretically eclectic and inclusive, the institution-based view has thrived by integrating a number of theories (Gaur et al., 2014; Hung & Tseng, 2017; Kostova & Hult, 2016; Lin et al., 2009; Mahlich, 2009; Meyer & Peng, 2005, 2016; Shi et al., 2012; Yi et al., 2019; Zoogah et al., 2015). Growth The growth of the institution-based view predates Peng (2002). However, given that Peng (2002) is the first journal publication coining the term “institution-based view” (Ahuja & Yayavaram, 2011: 1649), it is useful to examine how the institution-based view has grown since the initial 2002 article. We use the Web of Science to review the 3,043 articles that have cited Peng (2002, 2003) and Peng et al. (2008, 2009) as of June 30, 2022. We extract the high-frequency keywords from these articles—emerging, innovations, institutions, institution-based, China, view, performance, economies, markets, and management (in descending order of frequency)—to form a word cloud (Fig. 1). Fig. 1 High-frequency keywords in the institution-based view literature. [Sources] 3,043 articles that cited Peng (2002, 2003) and Peng et al. (2008, 2009) reported by Web of Science, as of June 30, 2022 The sampled articles suggest that China is the most frequently-studied emerging economy, followed by (in descending order of frequency) Africa, Russia, India, and other Asian countries. The institution-based view has also significantly benefitted research studying firms in developed economies such as the United States, Japan, South Korea, Western Europe, and New Zealand (in descending order of frequency). It is noteworthy that the institution-based view has penetrated countries that otherwise are rarely covered by management journals—namely, Fiji, Ghana, and Tanzania. Such a diverse and global reach of the institution-based view attests to its growing influence and popularity (Fig. 2). Fig. 2 Geographic coverage of the institution-based view literature. [Sources] Panel A: 522 articles that cited Peng (2002, 2003) and Peng et al. (2008, 2009) reported by Web of Science, as of December 31, 2012. Panel B: 3,043 articles that cited these four articles, as of June 30, 2022 (including the 522 articles used in Panel A) Visually, the institution-based view can be conceptualized as a “tree” that has grown various branches, covering research topics such as corporate diversification, corporate governance, entrepreneurship, intellectual property rights, international business strategy, and large family firms (Fig. 3). The “tree” visualizes how the institution-based view provides a solid base for theorizing and dealing with a variety of topics and phenomena. Extending Meyer and Peng (2016: 14), we argue that this family (or tree) of research topics and phenomena stemming from the institution-based view is converging toward an integrative paradigm as conceptualized by Kuhn (1970). The broad agreement on the most fundamental proposition that “institutions matter” unifies institutionally minded scholars, while different branches explore how institutions matter within the institution-based paradigm. In the social sciences and management literatures, the acceptance and diffusion of paradigms (or schools of thought) depend on their continuity, novelty, and scope (McKinley et al., 1999). It is reasonable to suggest that the institution-based view exhibits these three attributes, which propel its growth (Meyer & Peng, 2016: 14; Peng et al., 2009: 72–73). First, by extending new institutionalism into management research, the institution-based view exemplifies continuity from the larger social sciences literature (Ingram & Silverman, 2002). Drawing primarily from economic institutionalism (North, 1990; Ostrom, 2005; Williamson, 2000) and sociological institutionalism (DiMaggio & Powell, 1983; Scott, 2008), the institution-based view argues that institutions’ most fundamental role is “to reduce uncertainty and provide meaning” (Peng et al., 2009: 66). Therefore, the institution-based view, albeit with differences in disciplinary roots, offers significant insights into how and when institutions matter (Peng et al., 2018b). Second, the institution-based view brings significant novelty to management research by addressing problems that neither the industry-based nor the resource-based paradigms can solve satisfactorily. Within the institutions literature, it reconciles two contrasting ideas about firm behavior—structure versus agency (Heugens & Lander, 2009). The “structure” school posits that firms become increasingly isomorphic over time as they are under collective institutional pressures in search of legitimacy (DiMaggio & Powell, 1983). The “agency” school argues that firms leverage institutional entrepreneurship to deviate from norms in an effort to gain competitive advantages (Oliver, 1991). In other words, institutional pressures “do not just ‘enter’ an organization—they are interpreted, given meaning, and ‘represented’ by occupants of structural positions” (Greenwood et al., 2011: 342).1 Even within the same institutional environment (such as one industry or one country), not all firms would behave the same, resulting in significant heterogeneity (Barney, 2001). Overall, firms strive for optimal distinctiveness—sufficiently differentiated to stand out and sufficiently recognizable to be legitimate (Zhao et al., 2017). Finally, the institution-based view is distinguished by its broad scope. Instead of being divisive—typical of some institutions literature famous for having numerous “family quarrels” (Heugens & Lander, 2009; Hirsch & Lounsbury, 1997)—the institution-based view is integrative and inclusive. Avoiding being partial to a particular line of the institutions literature (such as economic institutionalism or sociological institutionalism), the institution-based view—as part of management scholarship—builds bridges by drawing on the best available insights from the interdisciplinary literature on institutions (Peng et al., 2009: 74). Therefore, its broad scope “allows for numerous ways of theorizing, operationalizing, and testing, resulting in an expanding and cumulative body of knowledge” (Meyer & Peng, 2016: 14). In addition to the three content attributes outlined above—continuity, novelty, and scope—certain context attributes also fuel the growth of the institution-based view (Ofori-Dankwa & Julian, 2005). When research on the institution-based view in management was starting in the 1990s, the rise of rapidly-transitioning emerging economies attracted scholars’ attention (Peng & Heath, 1996). While trained in a number of theories in their repertoire, these scholars often choose to invoke an institutional perspective, which provides the best insights relative to other theories in advancing management research focusing on emerging economies (Cuervo-Cazurra et al., 2019a; Hoskisson et al., 2013; Keister, 2009; Luo et al., 2019; Jiang et al., 2022; Marquis & Raynard, 2015; Meyer & Peng, 2005, 2016; Peng, 2003; Peng et al., 2008; Pezeshkan et al., 2022; Pinkham & Peng, 2017; Sun et al., 2017; Weng et al., 2021; Wright et al., 2005; Young et al., 2014). Ultimately, research on the institution-based view has crossed into areas outside of emerging economies and has been applied to a wider range of economies (see Fig. 2). A number of influential scholars such as John Child, John Dunning, and Michael Hitt, who are not known as “institutional scholars” in their early career, have not only endorsed, but also contributed toward, the institution-based view (Child et al., 2007; Dunning & Lundan, 2008; Hitt et al., 2004). Overall, contextual factors fueling the growth of the institution-based view include: (1) an attractive initial research context (emerging economies), (2) eagerness of many scholars in search of the best theoretical tool, and (3) prestige of some contributors invoking this view (Ofori-Dankwa & Julian, 2005). Criticisms and responses Despite its growth and development toward a paradigm, the institution-based view has faced criticisms as would be expected with any theory. Twenty years after Peng (2002), it is useful to take stock of some leading criticisms and respond to them. This section responds to three criticisms. Criticism 1: The usefulness of the institution-based view will decline as market-supporting institutions progress in emerging economies. Initially focusing on emerging economies, a major stream of the institution-based view spearheaded by Peng and Heath (1996), Peng (2002, 2003), and Peng et al. (2008, 2009) has propelled this view to become “the most dominant” theory when probing emerging economies (Wright et al., 2005: 1). But its usefulness is criticized to be transient. According to Hoskisson et al. (2000: 252), “in the early stages of market emergence, institutional theory is preeminent in helping to explain impacts on enterprise strategy. . As markets mature, transaction cost economics and, subsequently, the resource-based view are more important.” In other words, there was a concern that as emerging economies develop, the institution-based view would become less important relative to other theories. In his decision letter accepting Peng (2002), Chung Ming Lau, one of the guest-editors of the Academy of Management Journal special research forum on emerging economies (Hoskisson et al., 2000—of which Peng and Luo (2000) is a part), challenged the author by asking “how far can we use the institution-based view when we have a developed economy (e.g., China 15 years later)?” Yet, the institution-based view “has become more enduring than anticipated,” and the prediction that other theories “will become more relevant and prominent in research on emerging economies” has only been partially supported (Wright et al., 2005: 22). A key reason is “the development of [market-supporting] institutions in emerging economies has been slower than anticipated and the nature of institutional developments has not been uni-directional” (Wright et al., 2005: 22). Recent political reversals in a number of emerging economies such as Brazil, China, Hungary, Mexico, Poland, Russia, and Turkey have further slowed (and sometimes reversed) the development of market-supporting institutions. The theoretical implications are that the usefulness of the institution-based view is unlikely to decline anytime soon, if scholars endeavor to enhance our understanding of firm behavior and performance in emerging economies (Bruton et al., 2021; Lebedev et al., 2015; Marquis & Raynard, 2015; Meyer & Peng, 2005, 2016; Opper, 2022). Furthermore, shown in Fig. 2, the institution-based view has expanded beyond emerging economies and asserted its influence in developed economies (Greenwood et al., 2011; Peng et al., 2009; Weng & Peng, 2018). This development is not only underpinned by a substantial body of institutional research that has always focused on developed economies (Fligstein, 1996; Oliver, 1997; Scott, 2008), but is also necessitated by the numerous institutional transitions unfolding throughout developed economies, such as the institutional transitions brought by Brexit (2016), Donald Trump’s presidency (2017–2021), and rapid policy U-turns of the UK government (2022). Therefore, designing appropriate strategies in response to institutional transitions—advised by the institution-based view—remains crucial for firms in both emerging and developed economies (Opper, 2022). Overall, in the marketplace for theories, the institution-based view, which originally focuses on the impact of institutional transitions, is likely to be useful throughout the world. This is illustrated in the current world conditions amounting to unprecedented turbulence, ranging from public health crises (COVID-19) to geopolitical conflicts (Russia’s invasion of Ukraine). Such turbulence has unleashed significant changes to the “rules of the game” throughout the world, such as government-imposed lockdowns and sanctions with significant impact on firm behavior, performance, and even survival (Liu et al., 2022). Such institutional transitions naturally trigger changes in firm behavior, which will have significant performance implications (Yiu et al., 2018). Instead of being some interesting events only affecting emerging economies, institutional transitions are likely to become the “new normal” throughout the world (Ahlstrom et al., 2020; Li et al., 2022). Therefore, as a dynamic theory built around the interaction between institutions and organizations (Peng, 2002), the institution-based view will become more important. Criticism 2: As a big tent, the institution-based view has too many different strands, schools, and flavors of what institutions mean, and how they affect or are affected by firms. As a “big tent,” the institutions literature is thriving throughout the social sciences and management disciplines (Aguilera & Grogaard, 2019: 26; Cuervo-Cazurra et al., 2019b: 153). It is also true that there are a number of “family quarrels” within the institutions literature (Heugens & Lander, 2009; Hirsch & Lounsbury, 1997). These quarrels often take place between economics and sociology—for example, the words “institutional theory” cannot be used to describe economics research represented by North (1990) and Williamson (2000), which should be labeled “institutional economics” (but not “institutional theory”). Within “institutional theory” (otherwise known as sociological institutionalism), debates rage between “old” and “new” institutionalism (Hirsch & Lounsbury, 1997). Within new institutionalism, there are further divisions such as institutional logics (Greve & Zhang, 2017), institutional work (Lawrence et al., 2013), and varieties of capitalism (Carney et al., 2009; Hall & Soskice, 2001; Jackson & Deeg, 2008, 2019). The end result, according to Aguilera and Grogaard (2019: 25), is that “we cannot refer simultaneously to Scott, North, Peng, and Jackson and Deeg (2008).”2 Because different strands of the institutions literature emerge from different intellectual traditions, the alleged risk, according to critics, is “a potpourri of arguments from incompatible logics” (Cuervo-Cazurra et al., 2019b: 151). In other words, “scholars use different language to refer to similar, if not the same, concepts or mechanisms” (Aguilera & Grogaard, 2019: 25). Going forward, scholars are advised “to properly anchor their research within the boundaries of a given strand and to be cognizant of the conceptual challenges if these strands are ever to be combined” (Aguilera & Grogaard, 2019: 25; see also Cuervo-Cazurra et al., 2019b: 151). In short, as criticized by Hirsch and Lounsbury (1997), some of the institutions research has conventionally been divisive. Given the integrative nature of the institution-based view, our response is: of course, we can refer simultaneously to Scott, North, Peng, and Jackson and Deeg. Scott (1995) has long acknowledged North’s (1990) influence on economic sociology. North (2005) explicitly discusses “stickiness” (resistance to change) as part of cognition, which bears reciprocal correspondence to Scott’s (1995) cultural-cognitive pillar. In addition to members from economics, the International Society of New Institutional Economics (ISNIE), which has recently rebranded itself as the Society for Institutional and Organizational Economics (SIOE), has members from anthropology, law, management, political science, and sociology (Menard & Shirley, 2014: 554). The institution-based view is indeed “inspired by both the economic and sociological versions of the institutional literature” (Peng et al., 2009: 74). Given the significant cross-fertilization across the institutions space, sticking to one “party line” will be challenging (Marquis & Raynard, 2015: 322). As the larger world—the institutional environment of our scholarly work—marches toward more diversity and inclusion, emphasizing divisiveness and exclusion is unhealthy and impractical. Given its broad scope, the institution-based view is and should be integrative and inclusive (Roberts & Greenwood, 1997). In short, institutional pluralism is preferred (Kraatz & Block, 2017). Criticism 3: Empirical research associated with the institution-based view has overly focused on formal institutions and paid inadequate attention to informal institutions. Following North (1990), the institution-based view claims that “formal and informal institutions combine to govern firm behavior” (Peng et al., 2009: 68; see also Holmes et al., 2013; Marano et al., 2016). However, formal institutions have attracted disproportionately more attention (Seligson & McCants, 2021). Because “it is much easier to describe precisely the formal rules that societies devise than to do the same for the informal ways by which human beings have structured human interaction” (North, 1990: 36), there is indeed an imbalance in the development of the institution-based view. Specifically, the imbalance is reflected in the vastly greater number of publications on formal institutions via-a-vis those on informal institutions. Therefore, it is important to address this criticism by paying more attention to informal institutions (Boddewyn & Peng, 2021; Seligson & McCants, 2021; Voigt, 2018). Although limited, management research on informal institutions has emerged in a number of areas. These include corporate governance (Estrin & Prevezer, 2011; Sauerwald & Peng, 2013), entrepreneurship (Batjargal et al., 2013; Lahiri et al., 2020; Opper & Anderson, 2019; Peng et al., 2018c; Salvi et al., 2022), international market entry (Boddewyn & Peng, 2021; Holmes et al., 2013), multinational subsidiaries (Curchod et al., 2020), offshoring (Sartor & Beamish, 2014), reputation (Stevens & Makarius, 2015), and strategic alliances (Ahlstrom et al., 2014; Cao et al., 2018). In the context of emerging economies, there is substantial research on informal institutions embodied by interpersonal networks and social capital (Burt & Burzynska, 2017; Lebedev et al., 2021; Ledeneva, 2018; Li et al., 2008; Peng, 2003). From an initial focus on China (Burt & Batjargal, 2019; Haveman et al., 2017; Li & Qian, 2013; Mutlu et al., 2018; Opper et al., 2017; Peng & Luo, 2000), such research has been extended to Ghana (Acquaah, 2007), Indonesia (Fisman, 2001), Mongolia (Ulziisukh & Wei, 2022), Russia (Puffer et al., 2010), and South Korea (Horak & Klein, 2016). Despite the progress, two challenges foreshadow research focusing on informal institutions. First, it is challenging to differentiate informal institutions from culture. Some studies have largely treated culture and informal institutions as synonymous, and adopted cultural dimensions such as power distance and collectivism as informal institutions (Cao et al., 2018; Holmes et al., 2013). Admittedly, there is an overlap between culture and informal institutions (Chimenson et al., 2022; Peterson, 2016). However, informal institutions also have important facets that cannot be captured by culture (Cantwell et al., 2010: 578; Singh, 2007: 442). How to break away from solely using cultural measures to reflect informal institutions and thus truly appreciate the broad range of informal institutions is a formidable challenge (Shoham, 2022; Singh, 2007). A second challenge stems from the interdependency between informal and formal institutions (North, 1990; Ostrom, 2005). Formal and informal institutions interact and influence each other, making it difficult to tease out the specific effects of informal institutions (Holmes et al., 2013; Zhu et al., 2019). For instance, firms may embrace market competition in emerging economies at the backdrop of institutional transitions from state to market, which can be traced to the rising normative pressure from rivals or the emerging cognitive shifts within firms—informal institutional adaptations that are catalyzed by formal institutional transitions (Peng, 2003; Peng & Heath, 1996). Likewise, informal institutions can also affect and eventually change formal institutions (North, 1990). Several studies have endeavored to unveil the complex relationships between informal institutions and organizations (Cappellaro et al., 2020; Smith & Besharov, 2019). Overall, how to creatively capture informal institutions and untangle their distinctive impact is challenging, and remains a future research direction. From criticisms to promising research areas Constructive criticisms are both inspiring and motivating. Having outlined the institution-based view’s three major criticisms,3 next we showcase four promising research directions: (1) deglobalization and sanctions, (2) competitive dynamics, (3) hybrid organizations, and (4) corporate social responsibility. These areas are chosen, because they represent new research endeavors to address some of the criticisms and expand further. New research on deglobalization and sanctions as well as competitive dynamics shows that institution-based research is not limited to emerging economies, and has global ramifications, thus addressing Criticism 1. Research on hybrid organizations demonstrates how firms leverage and mix multiple institutional demands, as reflected in their goals and actions. This addresses some aspects of Criticism 2. Responding to Criticism 3, new research on CSR expands on the research on informal institutions and on the interaction between informal and formal institutions. Like previous paradigms, a new paradigm such as the institution-based view may not solve all questions. However, it can generate new questions. Such potential can be called the generative capability of the institution-based view—“a sociotechnical system where social and technical elements interact to facilitate combinatorial innovation” (Thomas & Tee, 2022). This means that the institution-based view can be integrated with other theories to facilitate theoretical innovation (Meyer et al., 2009; Zoogah et al., 2015). Overall, these four areas demonstrate how exciting new research stemming from the institution-based view, often in combination with other views, can grow to generate (and hopefully solve) new puzzles. An institution-based view of deglobalization and sanctions Contrary to the concerns expressed in Criticism 1, while the institution-based view started with an early geographic emphasis on emerging economies, it has subsequently expanded to assert its reach globally—in both emerging and developed economies. A rapidly evolving institution-based view of deglobalization and sanctions embodies such a global approach (Blake et al., 2022; Devinney & Hartwell, 2020; Li et al., 2022; Meyer et al., 2022). Most theories are likely to carry some imprinting of the era during which they are developed (Kriauciunas & Kale, 2006; Peng, 2003). New institutionalism research has largely grown in the post-Cold War era, when globalization—often embodied by market-liberalizing institutional transitions—was increasing (Doh et al., 2017; Hoskisson et al., 2000, 2013; Marquis & Raynard, 2015; Wright et al., 2005). Therefore, it has a great deal of pro-market flavors (Cuervo-Cazurra et al., 2019a; Kathuria et al., 2023; Meyer & Peng, 2005, 2016; Shin et al., 2022). As a result, much of the research on institutions focuses on innovation (Khoury & Peng, 2011; Zhou et al., 2017), market entry (Boddewyn & Peng, 2021; Deng et al., 2020; Lu et al., 2018; Meyer et al., 2009; Yiu & Makino, 2002), and growth of the firm (Peng & Heath, 1996; Peng et al., 2018a). In contrast, there is little research on sanctions, withdrawals, and shrinking of the firm, to name a few. Defined as “the process of weakening economic interdependence among countries” (Witt, 2019: 1054), deglobalization seems to be a new wave that has hit the world (Peng et al., 2021; Petricevic & Teece, 2019). Evidence of deglobalization is everywhere: border closures, COVID-induced lockdowns, immigration controls, investment screenings, military conflicts, nationalism, sanctions, supply chain localization, and trade wars. How managers and firms respond strategically to such changing rules of the game has presented a series of new research opportunities (Ahlstrom et al., 2020; Contractor, 2022; Globerman & Shapiro, 2009; Peng & Kathuria, 2021; Peng et al., 2021; Rodrik, 2018; Young et al., 2022). This can become a new research frontier in the institution-based view (Li et al., 2022). Among numerous topics on deglobalization, an institution-based view can inform research on sanctions. The recent scale, scope, and frequency with which sanctions are imposed is unprecedented. Sanctions can be defined as politically motivated, nonmilitary coercive measures against foreign countries, organizations, and/or individuals (Bapat & Kwon, 2015; Meyer et al., 2022; Mirkina, 2018). Sanctions are key tools of foreign policy, and the West has long imposed sanctions on smaller countries such as Cuba, Iran, Myanmar, and North Korea (Meyer & Thein, 2014). Now larger countries such as China and Russia are frequently targeted, and these countries have launched countersanctions against the West. Yet, “our theoretical understanding of sanctions is woefully underdeveloped” (Felbermayr et al., 2021: 2). Consider Western multinational enterprises (MNEs) operating in Russia since its February 2022 invasion of Ukraine. For foreign firms operating in Russia whose home-country governments are imposing sanctions on Russia, the strategic choices include: (1) permanent withdrawal, (2) temporary closure, and (3) business as usual. In contrast to the tremendous amount of work in the institutions literature on legitimization, the process of delegitimization has rarely been studied (Oliver, 1992). Essentially, if MNEs choose options 1 and 2, they lose legitimacy in the eyes of host-country stakeholders in Russia. But if they choose option 3, they lose legitimacy in the eyes of stakeholders in the home country and third countries (Stevens et al., 2016). Given the thorny nature of each option, the strategic choices are not preordained, thus necessitating more in-depth research. A new generation of research on the institution-based view can leverage the imprinting of the new era (Barry & Kleinberg, 2015; Li et al., 2022; Luo, 2022; Meyer & Li, 2022; Witt, 2019; Witte et al., 2020). Fascinating but underexplored questions in a new institution-based view of deglobalization and sanctions include: How predictable are government actions (Witt, 2019)? How maneuverable is corporate diplomacy (as opposed to country diplomacy) (Li et al., 2022)? How can MNE subsidiaries engage with local stakeholders to minimize the impact of sanctions (Meyer & Li, 2022)? Given the global nature of deglobalization and sanctions, this new area of research can address Criticism 1 by demonstrating that the institution-based view is a globally relevant paradigm. An institution-based view of competitive dynamics Competitive dynamics is another area in which the institution-based view has the potential to reach developed economies as well as emerging economies, countering Criticism 1. Historically, research on competitive dynamics has examined how new entrants and incumbents engage in rounds of competitive attacks and counterattacks (Chen & Miller, 1994; Mutlu et al., 2015). New entrants often use tactics such as novel products to outcompete the pricing, quality, or timing of incumbent products, while incumbents use tactics such as entering new markets, price cutting, or developing new products to fight back. Recently, the emergence of new and disruptive innovation, including digitalized new entrants, has made competitive dynamics more complicated, with some new entrants creating entirely new ways of competing that make it difficult for incumbents to keep up with (Kammerlander et al., 2018). New entrants such as Uber and Airbnb are able to capitalize on platform ecosystems to change the “rules of the game” and counter incumbents’ existing products or services (Kretschmer et al., 2022; Kumaraswamy et al., 2018). From an institution-based view, these disruptive new entrants amount to a fundamental challenge for incumbents, as traditional competitive responses may be less effective. Due to the fact that the new entrants can avoid, or circumvent, existing institutions, both new entrants and incumbents have turned to institutions in their competitive attacks and counterattacks. Consistent with the institutional work literature that considers how firms “create, disrupt, or maintain institutions” (Voronov & Vince, 2012: 59), incumbents can leverage institutions to combat new entrants by lobbying the government (Ridge et al., 2017), mobilizing voters (Bonardi et al., 2005), and bringing lawsuits against new entrants (Bagley, 2008). At the same time, disruptive new entrants such as Uber and Airbnb have found ways to manipulate institutions, allowing them to operate in areas that incumbents cannot (Baron, 2018). Using similar institutional tactics such as lobbying, voter mobilization, and lawsuits (Garud et al., 2022), new entrants’ attacks and counterattacks leverage institutions to enable them to gain an advantage over incumbents, which often operate under antiquated regulations and which do not have the same technological capabilities possessed by new entrants. In response, incumbents may counterattack using their own institutional defenses, while embarking on obtaining similar technology to compete directly with new entrants. The recent literature on disruptive innovation and digitalization acknowledges that new entrants using technology and platforms adopt nonmarket strategies (Baron, 2018; Garud et al., 2022), but has not extensively addressed the role of institutions in competitive dynamics, from both new entrants’ and incumbents’ perspectives. Therefore, there is potential to develop an institution-based view of competitive dynamics focusing on both sides’ jockeying for positions by enlisting help from institutions—in other words, weaponizing institutions (Welbourne Eleazar et al., 2022). Future work may consider: How can new entrants and incumbents leverage both formal and informal institutions in different countries to gain a competitive advantage in digitalized markets? In countries that have greater regulations, are certain institutions more likely to be weaponized than in less regulated countries? In addition, while we focus on disruptive innovation such as digitalization and institutions, there are other topics on which the weaponization of institutions in competitive dynamics has not been clearly identified to date. For examples, how does weaponizing institutions affect new entrants vis-à-vis incumbents respectively, such as the recent legalization of cannabis and sports betting in many states of the United States? When faced with stricter regulations such as the European Union’s General Data Protection Regulation (GDPR), how can incumbents weaponize these regulations against new entrants, and how can new entrants navigate through these regulations? While battles on weaponizing institutions between entrants and incumbents unfold in emerging economies, a great deal of such battles take place in developed economies, thus in part addressing Criticism 1 that the institution-based view is “only relevant” to emerging economies. An institution-based view of hybrid organizations While critics associated with Criticism 2 complain about arguments from incompatible logics within the same study, many organizations increasingly grapple with hybridity, featuring a combination of institutional logics that traditionally do not go together (Cappellaro et al., 2020; Pache & Santos, 2013; Smith & Besharov, 2019; Sun & Liang, 2021). Institutional logics are “overarching belief systems that provide rationales for organizational goals, underpin identities, and shape behaviors” (Wang et al., 2022: 5; see also Thornton, 2004). Nested within the umbrella framework of the institution-based view, institutional logics have emerged as a leading perspective to help us understand the inner workings of hybrid organizations—defined as organizations that incorporate different institutional logics (Pacho & Santos, 2013). For example, Wang et al. (2022) extend an institutional logics lens to delve into how state-owned enterprises (SOEs) leverage hybridity in order to become more innovative. Under the old state (socialism) logic, SOEs are neither interested in, nor capable of, innovating (Peng & Heath, 1996). However, the emerging market (competition) logic sweeping through most emerging economies necessitates an emphasis on innovation to win market battles (Inoue et al., 2013; Raynard et al., 2020). Therefore, the broader institutional context nudges SOEs to combine the competing state and market logics, epitomizing hybrid organizations. As hybrid organizations, not all SOEs are dominantly guided by a state logic, which prioritizes social and political goals at the expense of efficiency and innovation (Wright et al., 2021). Given significant pro-market institutional transitions (Cuervo-Cazurra et al., 2019a; Meyer & Peng, 2005; Peng, 2003), many SOEs simultaneously incorporate a market logic that structures cognition and shapes decision-making in favor of market competitiveness embodied by innovation (Bruton et al., 2015; Huang et al.,, 2017; Peng et al., 2016). In other words, informal institutions that influence SOEs’ strategic choices may gradually feature a coexistence of state and market logics (Peng, 2003; Raynard et al., 2020). The looser the grip of state logic, the easier the acceptance of market logic, and thus the more innovative such SOEs become (Wang et al., 2022). By contrast, some SOEs under the tight grip of a state logic may be institutionally constrained to innovate more. Therefore, heterogeneity in innovation lies in organizational hybridity of SOEs and results from the dynamic interaction between multiple institutional logics governing SOEs (Peng & Heath, 1996; Raynard et al., 2020). Under what circumstances do SOEs interact with institutions to produce varied innovation outcomes? To answer this question, researchers have often turned to the nature and degree of state ownership (Greve & Zhang, 2017; He et al., 2022). SOEs with a low level of state ownership may be less constrained by a state logic, thus leaving room for a market logic that shapes their innovativeness (Zhou et al., 2017). Departing from this dominant research focus, Wang et al. (2022) unveil a structural explanation by examining the ownership distance between the ultimate state owner and SOEs in China. Informed by the corporate pyramid literature (Almeida & Wolfenzon, 2006), Wang et al. (2022) map ownership linkages between the state and SOEs, and turn the spotlight on state-owned pyramids—in which the state directly owns some SOEs, which in turn own other SOEs. As a result, it is the indirect ownership by the state—specifically, pyramidal ownership—that explicates certain SOEs’ incorporation of a market logic with the presence of a state logic. Specifically, SOEs that are indirectly controlled and that are further away from the ultimate state owner along the pyramidal chains are more likely to respond to an emerging market logic that fuels more innovation. Therefore, ownership distance influences how SOEs embrace a market logic, which in turn results in hybrid SOEs. Overall, embedded in similar formal institutions, SOEs experience different informal constraints. The upshot? Heterogeneity in organizational hybridity as well as strategic choices and outcomes concerning innovation. In summary, an institution-based view of hybrid organizations can untangle the complex interactions between institutions and SOEs with innovation as an outcome (Wang et al., 2022). Such a focus, going beyond the primary attention paid to state ownership, innovatively probes the mechanisms underlying the relationship among institutions, organizations, and strategic choices (Peng, 2002; Peng & Heath, 1996). In light of the increasing demands on organizational hybridity, such as combining public with private expectations in medical centers (Cappellaro et al., 2020) and social missions with commercial goals in microfinance organizations (Sun & Liang, 2021), more attention to an institution-based view of hybrid organizations is warranted. Questions remain: How do different types of organizations—such as hospitals, universities, and social enterprises—achieve hybridity? Given the variety of institutional logics, how do organizations decide which logic(s) to accept and which to avoid or even reject? After integrating multiple logics, how do hybrid organizations reconcile the often incompatible demands? If conflicts arise, how can hybrid organizations address these ongoing challenges? Overall, an institution-based view of hybrid organizations can reveal how managers and firms respond to multiple conflicting institutional demands (Pache & Santos, 2013). In a similar spirit, scholars can address Criticism 2 by reaching out higher, wider, and deeper with various strands of the literature to produce integrative and inclusive work, offering “model problems and solutions to a community of practitioners” (Kuhn, 1970), who can in turn blend elements of different institutional demands to construct winning hybrid organizations. An institution-based view of corporate social responsibility (CSR) In the absence of legal mandate, CSR is driven by informal institutions. A firm can be seen as a social entity whose survival depends upon the fulfilment of explicit and implicit contracts with stakeholders (Freeman & Evan, 1990; Matten & Moon, 2008). The very embeddedness in social relations sketches the contours of normative and cognitive framework that defines managers’ sensemaking, preferences, and decisions (Basu & Palazzo, 2008). Because informal institutions significantly impact CSR engagement, developing an institution-based view of CSR will help address Criticism 3 by pushing further the research on informal institutions. A primary driver of CSR engagement has been a need to earn legitimacy from stakeholders (Suchman, 1995). It is therefore not surprising that pressures from local and global nongovernmental organizations (NGOs), environmentalists, labor unions, consumer groups, and human rights activists are instrumental in motivating firms to embrace specific CSR activities (Gond et al., 2011). At the same time, deeper and more enduring informal institutions such as cultures and norms continue to shape CSR engagement. For instance, highlighting one of the most ancient social stratification systems—the caste system—Kathuria (2022) shows that social norms permeate boardrooms to influence the choice of CSR activities. Leveraging an institution-based view, Kathuria (2022) finds that due to powerful informal norms discriminating against lower-caste individuals, lower-caste directors may refrain from selecting CSR activities that benefit their own (lower-caste) communities. Traditionally, some scholars have viewed firms’ CSR activities that deliberately benefit the society as unnecessary (Friedman, 1970). Over time, more scholars have acknowledged CSR as critical to enhancing legitimacy among stakeholders (Campbell, 2007). The society in which the firm operates defines the boundaries of acceptable and unacceptable actions, thereby constituting economic, environmental, and societal “rules of the game” that firms need to respect. For instance, if setting up a factory displaces the native rural population, informal institutions may push managers to hire some of the displaced natives, even if these workers may be less efficient than others available in the labor market. Similarly, out of concerns for public backlash, reprisal, and reputational harm, firms may go above and beyond the legal requirements of formal institutions to comply with informal institutions of their community. They may choose to offer compensation or alternative means of living to native populations that their operations displace or invest in costly equipment to avoid contaminating the nearby environment. Local communities may further pressurize the firms to informally limit their discharge to be well below the formal compliance requirements. Overall, informal institutions are a major driver of CSR engagement. Informal institutions complement formal institutions to influence firms’ CSR engagement. Ample evidence exists that CSR is intricately tied to political motivations that differ across regions (Julian & Ofori-Dankwa, 2013; Kang & Moon, 2010; Matten & Moon, 2008). National institutional frameworks determine what counts as CSR, the extent to which firms engage in CSR, how firms incorporate CSR into their operations, and what roles the government play in motivating firms to be socially responsible. Using the typology of liberal market versus coordinated market capitalism (Hall & Soskice, 2001), we suggest that countries practicing liberal market capitalism (e.g., the United States) primarily view CSR as voluntary in nature (Dahlsrud, 2008; McWilliams & Siegel, 2001). On the other hand, countries practicing coordinated market capitalism (e.g. European countries) often actively push firms to perform CSR activities (Matten & Moon, 2008). The formal institutionalization of CSR in the corporate world has become a global phenomenon (Vogel, 2010). Although many firms have been voluntarily engaging in CSR activities that benefit the society, many governments have formally required firms to disclose or undertake CSR activities (Jackson et al., 2020). Governments in Australia, China, Denmark, France, Malaysia, and South Africa require firms to disclose CSR activities, while governments in India, Indonesia, and Mauritius mandate firms to allocate certain portions of their profits toward CSR. Through formal institutional pressures, CSR activities have become a key part of business instead of a peripheral choice. Overall, institutional frameworks—via informal and formal constraints—largely determine firms’ CSR engagement (Aguilera et al., 2007; Julian & Ofori-Dankwa, 2013; Marano et al., 2017). The institution-based view has been instrumental in advancing CSR research. More questions remain unaddressed. Can CSR be a localized approach to mitigate grand challenges—such as poverty, hunger, inequality, and climate change—that our world faces today? How do informal institutions promote transitions in formal institutions that mandate CSR? What are the short-term and long-term performance implications of mandatory CSR? At the individual level, CSR is found to positively affect employee morale and employee identification with their firms. Would these effects persist when CSR becomes mandatory? By deepening and broadening our understanding of informal institutions and their linkages with formal institutions—thus addressing Criticism 3—scholars can significantly advance an institution-based view of CSR. Discussion Contributions Overall, three contributions emerge. First, this article reviews what has been accomplished in the last two decades since the publication of Peng (2002). Nourished by the broader intellectual movement throughout the social sciences, the institution-based view has synthesized a diverse and scattered body of literature on the dynamic interaction between institutions and firms to form a coherent theoretical contribution. Starting with a geographic focus on emerging economies, the institution-based view has spilled over to inform research on developed economies. The branches of the institution-based view have covered a variety of topics such as corporate diversification, corporate governance, entrepreneurship, family firms, intellectual property rights, and international business strategy. This growing tree (or family of theories) all share a fundamental proposition that institutions matter, but develop in various ways to probe into how institutions matter. Overall, we extend Meyer and Peng (2016) to advance the argument that different lines of research underpinned by the institution-based view are converging toward an integrative paradigm. Second, this article has responded to three major criticisms. The three criticisms allege that the institution-based view (1) is a transient theory whose importance will decline as emerging economies develop, (2) has too many incompatible flavors, and (3) has overly focused on formal institutions at the expense of informal institutions. In response, we argue that as the world becomes more institutionally chaotic and unpredictable, the institution-based view is likely to become more important going forward on a worldwide basis. Representing diversity and pluralism, the institution-based view is characterized by its many interdisciplinary flavors whose insights can be brought together. Furthermore, the institution-based view can indeed benefit from and contribute to the broadening and deepening of research on informal institutions. Finally, responding to the three criticisms, we showcase four promising areas of institution-based research as exemplars of a new generation of scholarship. In addition to the relatively well-developed branches shown in Fig. 3, new branches are sprouting, further strengthening the development of the institution-based paradigm. These new branches include institution-based research on deglobalization and sanctions, competitive dynamics, hybrid organizations, and CSR. These research areas are timely, underexplored, and rapidly growing—so are the larger events that motivate research in these four areas. In other words, the larger institutional context in which research takes place calls for deeper and broader understanding of deglobalization and sanctions, competitive dynamics, hybrid organizations, and CSR—informed by an institution-based research agenda. Adding significant diversity and pluralism to the institution-based view, these and other promising new research areas will make the “big tree” more vibrant and flourishing.Fig. 3 The institution-based view: A family (tree). [Note] Due to space constraints, the literature cited is illustrative (but is not comprehensive). Limitations and future directions Given the mushrooming research on the institution-based view, the first limitation of our article is that only limited work has been reviewed. Narrowing down our Web of Science search to work citing four articles—Peng (2002, 2003) and Peng et al. (2008, 2009)—limits the number of articles using the institution-based paradigm. While institution-based research had been conducted prior to the publication of Peng (2002), even research published after 2002 may further the paradigm without necessarily citing any of these four articles. The diversity and volume of institution-based research is simply tremendous (Aguilera & Grogaard, 2019; Cuervo-Cazurra et al., 2019b), making a review of the entire body of literature difficult. Second, given the large volume of institution-based work since 2002, we cannot concretely differentiate institution-based research from non-institution-based research that merely cites one of the four key articles to acknowledge an existing literature. Including articles that only have cursory notes to Peng (2002, 2003), for example, may overstate the topic coverage of the institution-based research. To partially mitigate this problem, we have reviewed leading articles in detail when summarizing research topics to ensure that the institution-based view is one of the main theoretical frameworks. We have also conducted a cursory review of the remaining articles to confirm that they appear to be generally related to the institution-based view. Our word cloud shown in Fig. 1 provides further support for our method, since the featured words are consistent with the use of the institution-based view and “institution-based” is one of the top listed key words. In terms of future research directions, deepening and broadening the institution-based view is a must (Peng, 2014). Deepening would involve sustained efforts to enrich the existing branches of the family of theories that has become the institution-based paradigm (see Fig. 3). Broadening would entail efforts to grow new branches of the “tree.” For example, given the obvious importance of rule-making in the global battle against climate change, there is potential to develop an institution-based view of climate change (Pinkse & Kolk, 2012). Furthermore, given the criticisms about the lack of progress on research on informal institutions discussed in relation to Criticism 3, future research can develop an institution-based view of emotions and relationships (Voronov & Weber, 2016). Such a focus can help us get to the “heart of institutions,” which is often informal in nature (Voronov & Weber, 2016: 1). Other examples can include developing an institution-based view of diverse leadership in top management teams and boards (Wang et al., 2019), institutional entrepreneurship (Chen & Sun, 2019), real options (Smit, Pennings, & van Bekkum, 2017), slack resources (Tan & Peng, 2003; Vanacker et al., 2017), strategy implementation (Opper, 2022), and tax evasion (Gokalp et al., 2017). Conclusion “One thing for sure,” Peng (2002: 263) concludes, “is that the importance of institutional influences will be increasingly appreciated in the new millennium, thus necessitating more attention from researchers, practitioners, and policymakers.” Clearly, this conclusion has been supported by both the advancement of the institution-based view in the scholarly world and its development in the wider world in the last two decades, during which the only constant seems to be institutional change. In 1987, Scott argues that new institutionalism research is in its adolescence. By 2008, Scott suggests that it has approached adulthood. Similarly, Peng et al. (2009: 77) conclude that the institution-based view has reached its adolescence at that time. Now, we can conclude that the thriving research on institutions has culminated in an institution-based paradigm that has approached adulthood. Acknowledgements Ideas based on this article were presented at Stockholm, Temple, Tongji, and Tsinghua Universities as well as University of Economics in Katowice. We thank Dave Ahlstrom (editor) and two reviewers for guidance, and Jay Choi, Susan Feinberg, Tony Fang, Xufei Ma, Amir Shoham, Todd Schifeling, and En Xie for helpful discussions. We acknowledge the support from the Jindal Chair at UT Dallas and the Herberger Business School at SCSU. 1 Most of the institutional entrepreneurship literature has focused on the top-down process, and the bottom-up process also needs to be addressed (Chen & Sun, 2019). 2 Aguilera and Grogaard’s (2019) article is a commentary on Jackson and Deeg’ (2008) article, which won the Decade Best Paper Award from the Journal of International Business Studies in 2018. 3 Other criticisms exist. For example, Walgenbach, Drori, and Hollerer (2017: 103) point out that “such a reinstatement of the rational actor who makes strategic choices is inherent, for instance, in the article of Kostova, Roth, and Dacin (2008), and comes through even more clearly and explicitly in the writings of Peng and his colleagues (Peng, 2002, 2003; Peng, Wang, & Jiang, 2008; Peng et al., 2009).” The problem according to critics? “Such pronouncements of agency, strategic choice, and interests ignore the rampant isomorphism among corporations” (Walgenbach et al., 2017: 103). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Acquaah M Managerial social capital, strategic orientation, and organizational performance in an emerging economy Strategic Management Journal 2007 28 1235 1255 10.1002/smj.632 Aguilera RV Grogaard B The dubious role of institutions in international business: A road forward Journal of International Business Studies 2019 50 20 35 10.1057/s41267-018-0201-5 Aguilera RV Rupp DE Williams CA Ganapathi J Putting the S back in corporate social responsibility: A multilevel theory of social change in organizations Academy of Management Review 2007 32 836 863 10.5465/amr.2007.25275678 Ahlstrom D Arregle JL Hitt MA Qian GM Ma XF Fraems D Managing technological, sociopolitical, and institutional change in the new normal Journal of Management Studies 2020 57 411 437 10.1111/joms.12569 Ahlstrom D Levitas E Hitt MA Dacin MT Zhu H The three faces of China: Strategic alliance partner selection in three ethnic Chinese economies Journal of World Business 2014 49 572 585 10.1016/j.jwb.2013.12.010 Ahuja G Yayavaram S Explaining influence rents: The case for an institution-based view of strategy Organization Science 2011 22 1631 1652 10.1287/orsc.1100.0623 Ahuja G Capron L Lenox M Yao DA Strategy and the institutional envelope Strategy Science 2018 3 iii xi 10.1287/stsc.2018.0062 Almeida HV Wolfenzon D A theory of pyramidal ownership and family business groups Journal of Finance 2006 61 2637 2680 10.1111/j.1540-6261.2006.01001.x Bagley C Winning legally: The value of legal astuteness Academy of Management Review 2008 33 378 390 10.5465/amr.2008.31193254 Bapat NA Kwon BR When are sanctions effective? A bargaining and enforcement framework International Organization 2015 69 131 162 10.1017/S0020818314000290 Baron D Disruptive entrepreneurship and dual purpose strategies: The case of Uber Strategy Science 2018 3 439 462 10.1287/stsc.2018.0059 Barney JB Is the resource-based “view” a useful perspective for strategic management research? Yes Academy of Management Review 2001 26 41 56 Barry C Kleinberg K Profiting from sanctions: Economic coercion and US foreign direct investment in third-party states International Organization 2015 69 881 912 10.1017/S002081831500017X Basu K Palazzo G Corporate social responsibility: A process andel of sensemaking Academy of Management Review 2008 33 1 122 136 10.5465/amr.2008.27745504 Batjargal B Hitt MA Tsui AS Arregle JL Webb JW Miller TL Institutional polycentrism, entrepreneurs’ social networks, and new venture growth Academy of Management Journal 2013 56 1024 1049 10.5465/amj.2010.0095 Baumol WJ Entrepreneurship: Productive, unproductive, and destructive Journal of Political Economy 1990 98 893 921 10.1086/261712 Blake DJ Markus S Martinez-Suarez J Populist syndrome and nonmarket strategy Journal of Management Studies 2022 10.1111/joms.12859 Boddewyn JJ Peng MW Reciprocity and informal institutions in international market entry Journal of World Business 2021 56 1 101145 10.1016/j.jwb.2020.101145 Bonardi, J. P., Hillman, A., & Keim, G. (2005). The attractiveness of political markets: Implications for firm strategy. Academy of Management Review, 30, 397–413. Bruton GD Ahlstrom D Chen J China has emerged as an aspirant economy Asia Pacific Journal of Management 2021 38 1 15 10.1007/s10490-018-9638-0 Bruton GD Peng MW Ahlstrom D Stan C Xu K State-owned enterprises as hybrid organizations around the world Academy of Management Perspectives 2015 29 92 114 10.5465/amp.2013.0069 Burt RS Batjargal B Comparative network research in China Management and Organization Review 2019 15 3 29 10.1017/mor.2019.8 Burt RS Burzynska K Chinese entrepreneurs, social networks, and guanxi Management and Organization Review 2017 13 221 260 10.1017/mor.2017.6 Campbell JL Why would corporations behave in socially responsible ways? An institutional theory of corporate social responsibility Academy of Management Review 2007 32 946 967 10.5465/amr.2007.25275684 Cantwell J Dunning JH Lundan SM An evolutionary approach to understanding international business activity: The co-evolution of MNEs and the institutional environment Journal of International Business Studies 2010 41 567 586 10.1057/jibs.2009.95 Cao Z Li Y Jayaram J Liu Y Lumineau F A meta-analysis of the exchange hazards-interfirm governance relationship: An informal institutions perspective Journal of International Business Studies 2018 49 303 323 10.1057/s41267-017-0144-2 Cappellaro G Tracey P Greenwood R From logic acceptance to logic rejection: The process of destabilization in hybrid organizations Organization Science 2020 31 415 438 10.1287/orsc.2019.1306 Carney M Gedajlovic E Yang X Varieties of Asian capitalism: Toward an institutional theory of Asian enterprise Asia Pacific Journal of Management 2009 42 1 13 Chen MJ Miller D Competitive attack, retaliation and performance: An expectancy valence framework Strategic Management Journal 1994 15 85 102 10.1002/smj.4250150202 Chen VZ Sun SL Barbarians at the gate of the Middle Kingdom: The international mobility of financing contract and governance Entrepreneurship Theory and Practice 2019 43 802 837 10.1177/1042258717745808 Child J Lu Y Tsai T Institutional entrepreneurship in building an environmental protection system in the People’s Republic of China Organization Studies 2007 19 766 784 Chimenson D Tung RL Panibratov A Fang T The paradox and change of Russian cultural values International Business Review 2022 31 101944 10.1016/j.ibusrev.2021.101944 Conner K An historical comparison of resource-based theory and five schools of thought within industrial organization economics: Do we have a new theory of the firm here? Journal of Management 1991 17 121 154 10.1177/014920639101700109 Contractor FJ The world economy will need even more globalization in the post-pandemic 2021 decade Journal of International Business Studies 2022 53 156 171 10.1057/s41267-020-00394-y 33551513 Cuervo-Cazurra A Gaur AS Singh D Pro-market institutions and global strategy: The pendulum of pro-market reforms and eversals Journal of International Business Studies 2019 50 598 632 10.1057/s41267-019-00221-z Cuervo-Cazurra A Mudambi R Pedersen T Clarifying the relationships between institutions and global strategy Global Strategy Journal 2019 9 151 175 10.1002/gsj.1342 Cuervo-Cazurra, A., & Genc, M. (2008). Transforming disadvantages into advantages: Developing-country MNEs in the least developed countries. Journal of International Business Studies, 39, 957–979. Curchod C Patriotta G Wright M Self-categorization as a nonmarket strategy for MNE subsidiaries: Tracking the international expansion of an online platform Journal of World Business 2020 55 3 101070 10.1016/j.jwb.2019.101070 Dahlsrud A How corporate social responsibility is defined: An analysis of 37 definitions Corporate Social Responsibility and Environmental Management 2008 15 1 13 10.1002/csr.132 Deng P Delios A Peng MW A geographic relational perspective on the internationalization of emerging market firms Journal of International Business Studies 2020 51 50 71 10.1057/s41267-019-00276-y Devinney TM Hartwell CA Varieties of populism Global Strategy Journal 2020 10 32 66 10.1002/gsj.1373 Dess GG Beard D Dimensions of organizational task environments Administrative Science Quarterly 1984 29 52 73 10.2307/2393080 Dieleman M Markus S Rajwani T White GO Revisiting institutional voids: Advancing the international business literature by leveraging social sciences Journal of International Management 2022 28 100935 10.1016/j.intman.2022.100935 DiMaggio PJ Powell WW The ron cage revisited: Institutional isomorphism and collective rationality in organizational fields American Sociological Review 1983 48 147 160 10.2307/2095101 Doh J Rodrigues S Saka-Helmhout A Makhija M International business responses to institutional voids Journal of International Business Studies 2017 48 293 307 10.1057/s41267-017-0074-z Dunning JH Lundan SM Institutions and the OLI paradigm of the multinational enterprise Asia Pacific Journal of Management 2008 25 4 573 593 10.1007/s10490-007-9074-z Duran P van Essen M Heugens P Kostova T Peng MW The impact of institutions on the competitive advantage of publicly listed family firms in emerging markets Global Strategy Journal 2019 9 243 274 10.1002/gsj.1312 Estrin S Prevezer M The role of informal institutionsin corporate governance: Brazil, Russia, India, and China compared Asia Pacific Journal of Management 2011 28 41 67 10.1007/s10490-010-9229-1 Felbermayr G Morgan TC Syropoulos C Yotov YV Understanding economic sanctions: Interdisciplinary perspectives on theory and evidence European Economic Review 2021 135 103720 10.1016/j.euroecorev.2021.103720 Fisman R Estimating the value of political connections American Economic Review 2001 91 1095 1102 10.1257/aer.91.4.1095 Fligstein N Markets as politics: A political-cultural approach to market institutions American Sociological Review 1996 61 656 673 10.2307/2096398 Freeman RE Evan WM Corporate governance: A stakeholder interpretation Journal of Behavioral Economics 1990 19 337 359 10.1016/0090-5720(90)90022-Y Friedman, M. (1970). The social responsibility of business is to increase its profits. New York Times Magazine, September, 13, 122–126. Gao GY Murray JY Kotabe M Lu J A “strategy tripod” perspective on export behaviors: Evidence from domestic and foreign firms based in an emerging economy Journal of International Business Studies 2010 41 377 396 10.1057/jibs.2009.27 Garud R Kumaraswamy A Roberts A Xu L Liminal movement by digital platofrm-base sharing economy ventures: The case of Uber Technologies Strategic Management Journal 2022 43 447 475 10.1002/smj.3148 Gaur AS Kumar V Singh D Institutions, resources, and internationalization of emerging economy firms Journal of World Business 2014 49 12 20 10.1016/j.jwb.2013.04.002 Globerman S Shapiro D Economic and strategic considerations surrounding Chinese FDI in the United States Asia Pacific Journal of Management 2009 26 163 183 10.1007/s10490-008-9112-5 Gokalp ON Lee SH Peng MW Competition and corporate tax evasion: An institution-based view Journal of World Business 2017 52 288 269 10.1016/j.jwb.2016.12.006 Gond JP Kang N Moon J The government of self-regulation: On the comparative dynamics of corporate social responsibility Economy and Society 2011 40 640 671 10.1080/03085147.2011.607364 Greenwood R Raynard M Kodeih F Micelotta ER Lounsbury M Institutional complexity and organizational responses Academy of Management Annals 2011 5 317 371 10.5465/19416520.2011.590299 Greve HR Zhang CM Institutional logics and power sources: Merger and acquisition decisions Academy of Management Journal 2017 60 671 694 10.5465/amj.2015.0698 Hall PA Soskice D Varieties of capitalism 2001 Oxford University Press Haveman HA Jia N Shi J Wang Y The dynamics of political embeddedness in China Administrative Science Quarterly 2017 62 67 104 10.1177/0001839216657311 He, X., Cui, L., & Meyer, K. E. (2022). How state and market logics influence firm strategy from within and outside? Evidence from Chinese financial intermediary firms. Asia Pacific Journal of Management, forthcoming. Heugens P Lander MW Structure! Agency! (and other quarrels): A meta-analysis of institutional theories of organization Academy of Management Journal 2009 52 61 85 10.5465/amj.2009.36461835 Hirsch PM Lounsbury M Ending the family quarrel: Toward a reconciliation of “old” and “new” institutionalisms American Behavioral Scientist 1997 40 406 418 10.1177/0002764297040004004 Hitt MA Ahlstrom D Dacin MT Levitas E Svobodina L The institutional effects on strategic alliance partner selection in transition economies: China vs. Russia Organization Science 2004 15 173 185 10.1287/orsc.1030.0045 Horak S Klein A Persistence of informal social networks in East Asia: Evidence from South Korea Asia Pacific Journal of Management 2016 33 673 694 10.1007/s10490-015-9416-1 Holmes RM Miller T Hitt MA Salmador MP The interrelationships among informal institutions, formal institutions, and inward foreign direct investment Journal of Management 2013 39 531 566 10.1177/0149206310393503 Hoskisson RE Eden L Lau CM Wright M Strategy in emerging economies Academy of Management Journal 2000 43 249 267 10.2307/1556394 Hoskisson RE Wright M Filatotchev I Peng MW Emerging multinationals from mid-range economies: The influence of institutions and factor markets Journal of Management Studies 2013 50 1295 1321 Huang Z Li L Ma G Xu LC Hayek, local information, and commanding heights: Decentralizing state-owned enterprises in China American Economic Review 2017 107 2455 2478 10.1257/aer.20150592 Hung SC Tseng YC Extending the LLL framework through an institution-based view: Acer as a dragon multinational Asia Pacific Journal of Management 2017 34 799 821 10.1007/s10490-016-9494-8 Ingram, P., & Silverman, B. (2002). Introduction. In P. Ingram & B. Silverman (Eds.), The new institutionalism in strategic management (pp. 1–30). Elsevier. Inoue CF Lazzarini SG Musacchio A Leviathan as a minority shareholder: Firm-level implications of state equity purchases Academy of Management Journal 2013 56 1775 1801 10.5465/amj.2012.0406 Jackson G Bartosch J Avetisyan E Kinderman D Knudsen JS Mandatory non-financial disclosure and its influence on CSR: An international comparison Journal of Business Ethics 2020 162 323 342 10.1007/s10551-019-04200-0 Jackson G Deeg R Comparing capitalisms: Understanding institutional diversity and its implications for international business Journal of International Business Studies 2008 39 540 561 10.1057/palgrave.jibs.8400375 Jackson G Deeg R Comparing capitalisms and taking institutional context seriously Journal of International Business Studies 2019 50 4 19 10.1057/s41267-018-0206-0 Jiang, H., Luo, Y., Xia, J., Hitt, M., & Shen, J. (2022). Resource dependence theory in international business: Progress and prospects. Global Strategy Journal. 10.1002/gsj.1467. Jiang Y Peng MW Are family ownership and control in large firms good, bad, or irrelevant? Asia Pacific Journal of Management 2011 28 15 39 10.1007/s10490-010-9228-2 Julian SD Ofori-Dankwa JC Financial resource availability and corporate social responsibility expenditures in a sub-Saharan economy: The institutional difference hypothesis Strategic Management Journal 2013 34 1314 1330 10.1002/smj.2070 Kammerlander N Konig A Richards M Why do incumbents respond heterogeneously to disruptive innovations? The interplay of domain identity and role identity Journal of Management Studies 2018 55 1122 1165 10.1111/joms.12345 Kang, N., & Moon, J. (2010). Variations and change in CSR from a “varieties of capitalism” perspective. Oxford-Archilles Seminar on CSR, Said Business School, University of Oxford. Kathuria, N. (2022). The implications of mandatory CSR on firms. Doctoral dissertation, Jindal School of Management, University of Texas at Dallas. Kathuria, N., Majumdar, S., & Peng, M. W. (2023). Institutional transitions, research and development, and exports from India. Journal of Management Studies forthcoming. Keister LA Organizational research on market transition: A sociological approach Asia Pacific Journal of Management 2009 26 719 742 10.1007/s10490-008-9113-4 Khanna T Palepu K Why focused strategies may be wrong for emerging markets Harvard Business Review 1997 75 41 48 Khoury TA Peng MW Does institutional reform of intellectual property rights lead to more inbound FDI? Evidence from Latin America and the Caribbean Journal of World Business 2011 46 337 345 10.1016/j.jwb.2010.07.015 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 51 467 497 10.1057/s41267-019-00294-w Kostova T Hult GTM Meyer and Peng’s 2005 article as a foundation for an expanded and refined international business research agenda: Context, organizations, and theories Journal of International Business Studies 2016 47 23 32 10.1057/jibs.2015.39 Kostova T Roth K Dacin MT Institutional theory in the study of multinational corporations: A critique and new directions Academy of Management Review 2008 33 994 1006 10.5465/amr.2008.34422026 Kraatz, M. K., & Block, E. S. (2017). Institutional pluralism revisited. In R. Greenwood, C. Oliver, T. B. Lawrence, & R. Meyer (Eds.), The Sage handbook of organizational institutionalism. Sage. Kretschmer T Leiponen A Schilling M Vasudeva G Platform ecosystems as meta-organizations: Implications for platform strategies Strategic Management Journal 2022 43 405 424 10.1002/smj.3250 Kriauciunas A Kale P The impact of socialist imprinting and search on resource change: A study of firms in Lithuania Strategic Management Journal 2006 27 659 679 10.1002/smj.537 Krull E Smith P Ge GL The internationalization of engineering consulting from a strategy tripod perspective Service Industries Journal 2012 32 1097 1119 10.1080/02642069.2012.662758 Kuhn TS The structure of scientific revolution 1970 2 University of Chicago Press Kumaraswamy A Garud R Ansari S Perspectives on disruptive innovations Journal of Management Studies 2018 55 1025 1042 10.1111/joms.12399 Lahiri S Mukherjee D Peng MW Behind the internationalization of family SMEs: A strategy tripod synthesis Global Strategy Journal 2020 10 813 838 10.1002/gsj.1376 Lebedev S Peng MW Xie E Stevens CE Mergers and acquisitions in and out of emerging economies Journal of World Business 2015 50 651 662 10.1016/j.jwb.2014.09.003 Lebedev S Sun SL Markoczy L Peng MW Board political ties and firm internationalization Journal of International Management 2021 27 100860 10.1016/j.intman.2021.100860 Lawrence TB Leca B Zilber TB Instituitonal work: Current research, new directions, and overlooked issues Organization Studies 2013 34 1023 1033 10.1177/0170840613495305 Ledeneva A The global encyclopaedia of informality 2018 UCL Press Lee SH Peng MW Barney JB Bankruptcy law and entrepreneurship development Academy of Management Review 2007 32 257 272 10.5465/amr.2007.23464070 Li J Shapiro D Peng MW Ufimtseva A Corporate diplomacy in the age of US-China rivalry Academy of Management Perspectives 2022 36 4 1 25 10.5465/amp.2021.0076 Li JJ Poppo L Zhou KZ Do managerial ties in China always produce value? Competition, uncertainty, and domestic vs. foreign firms Strategic Management Journal 2008 29 383 400 10.1002/smj.665 Li JT Qian C Principal-principal conflicts under weak institutions: A study of corporate takeovers in China Strategic Management Journal 2013 34 498 508 10.1002/smj.2027 Lin Z Peng MW Yang H Sun SL How do networks and learning drive M&As? An institutional comparison between China and the United States Strategic Management Journal 2009 30 1113 1132 10.1002/smj.777 Liu, Y., Peng, M. W., Wei, Z., Xu, J., & Xu, L. C. (2022). Resources, institutions, and cultures behind firm survival and growth during COVID-19. Policy Research Working Paper 9633. The World Bank. Lu J Liu X Wang H Motives for outward FDI of Chinese private firms: Firm resources, industry dynamics, and government policies Management and Organization Review 2010 7 223 248 10.1111/j.1740-8784.2010.00184.x Lu JW Li W Wu A Huang X Political hazards and entry modes of Chinese investments in Africa Asia Pacific Journal of Management 2018 35 39 61 10.1007/s10490-017-9514-3 Luo, Y. (2022). New connectivity in the fragmented world. Journal of International Business Studies, forthcoming. Luo Y Zhang H Bu J Developed country MNEs investing in developing economies Journal of International Business Studies 2019 50 633 667 10.1057/s41267-019-00230-y Makino S Isobe T Chan C Does country matter? Strategic Management Journal 2004 25 1027 1043 10.1002/smj.412 Mahlich JC Patents and performance in the Japanese pharmaceutical industry: An institution-based view Asia Pacific Journal of Management 2009 27 99 113 10.1007/s10490-008-9128-x Marano V Arregle JL Hitt MA Spadafora E van Essen M Home country institutions and the internationalization-performance relationship: A meta-analytic review Journal of Management 2016 4 1075 1110 10.1177/0149206315624963 Marano V Tashman P Kostova T Escaping the iron cage: Liabilities of origin and CSR reporting of emerging market multinational enterprises Journal of International Business Studies 2017 48 386 408 10.1057/jibs.2016.17 Marquis C Raynard M Institutional strategies in emerging markets Academy of Management Annals 2015 9 291 335 10.5465/19416520.2015.1014661 Matten D Moon J “Implicit” and “explicit” CSR: A conceptual framework for a comparative understanding of corporate social responsibility Academy of Management Review 2008 33 404 424 10.5465/amr.2008.31193458 McGahan AM Victor R How much does home country matter to corporate profitability? Journal of International Business Studies 2010 41 142 165 10.1057/jibs.2009.69 McKinley W Mone MA Moon G Determinants and development of schools in organization theory Academy of Management Review 1999 24 4 634 648 10.2307/259346 McWilliams A Siegel D Profit maximizing corporate social responsibility Academy of Management Review 2001 26 4 504 505 Menard C Shirley MM The future of new institutional economics: From early instututions to a new paradigm? Journal of Institutional Economics 2014 10 541 565 10.1017/S174413741400006X Meyer KE Estrin S Bhaumik SK Peng MW Institutions, resources, and entry strategies in emerging economies Strategic Management Journal 2009 30 61 80 10.1002/smj.720 Meyer, K. E., Fang, T., Panibratov, A., Peng, M. W., & Gaur, A. (2022). International business under sanctions. Working paper. Ivey Business School. Meyer KE Li C The MNE and its subsidiaries at times of global disruptions: An international relations perspective Global Strategy Journal 2022 10.1002/gsj.1436 Meyer KE Peng MW Probing theoretically into Central and Eastern Europe: Transactions, resources, and institutions Journal of International Business Studies 2005 36 600 621 10.1057/palgrave.jibs.8400167 Meyer KE Peng MW Theoretical foundations of emerging economy business research Journal of International Business Studies 2016 47 3 22 10.1057/jibs.2015.34 Meyer KE Thein HH Business under adverse home country institutions: The case of international sanctions against Myanmar Journal of World Business 2014 49 156 171 10.1016/j.jwb.2013.04.005 Mirkina I FDI and sanctions: An empirical analysis of short- and long-run effects European Journal of Political Economy 2018 54 198 225 10.1016/j.ejpoleco.2018.05.008 Mutlu CC van Essen M Peng MW Saleh SF Duran P Corporate governance in China: A meta-analysis Journal of Management Studies 2018 55 943 979 10.1111/joms.12331 Mutlu CC Wu Z Peng MW Lin Z Competing in (and out of) transition economies Asia Pacific Journal of Management 2015 32 571 596 10.1007/s10490-015-9419-y North, D. C. (1990). Institutions, institutional change and economic performance. Cambridge University Press. North, D. C. (2005). Understanding the process of economic change. Princeton University Press. Ofori-Dankwa J Julian SD From thought to theory to school: The role of contextual factors in the evolution of schools of management thought Organization Studies 2005 26 1307 1329 10.1177/0170840605054620 Oliver C Strategic responses to institutional processes Academy of Management Review 1991 16 145 179 10.2307/258610 Oliver C The antecedents of deinstituonalization Organization Studies 1992 13 563 588 10.1177/017084069201300403 Oliver, C. (1997). Sustainable competitive advantage: Combining institutional and resource-based views. Strategic Management Journal, 18, 697–713. Opper S Social network and institution-based strategy research Asia Pacific Journal of Management 2022 10.1007/s10490-021-09798-2 Opper S Anderson FNG Are entrepreneurial cultures stable over time? Historical evidence from China Asia Pacific Journal of Management 2019 36 1165 1192 10.1007/s10490-018-9573-0 Opper, S., Nee, V., & Holm, H. J. (2017). Risk aversion and guanxi activities: A behavioral analysis of CEOs in China. Academy of Management Journal, 60, 1504–1530. Ostrom, E. (2005). Understanding institutional diversity. Princeton University Press. Pache A Santos F Inside the hybrid organization: Selective coupling as a response to competing institutional logics Academy of Management Journal 2013 56 972 1001 10.5465/amj.2011.0405 Peng MW Towards an institution-based view of business strategy Asia Pacific Journal of Management 2002 19 251 267 10.1023/A:1016291702714 Peng MW Institutional transitions and strategic choices Academy of Management Review 2003 28 275 296 10.2307/30040713 Peng MW From China strategy to global strategy Asia Pacific Journal of Management 2005 22 123 141 10.1007/s10490-005-1251-3 Peng MW Boddewyn JJ New research directions in the institution-based view Multidisciplimary insights from new AIB Fellows 2014 Emerald 59 78 Peng MW Ahlstrom D Carraher SM Shi W An institution-based view of global IPR history Journal of International Business Studies 2017 48 893 907 10.1057/s41267-016-0061-9 Peng MW Bruton GD Stan C Huang Y Theories of the (state-owned) firm Asia Pacific Journal of Management 2016 33 293 317 10.1007/s10490-016-9462-3 Peng MW Heath PS The growth of the firm in planned economies in transition: Institutions, organizations, and strategic choice Academy of Management Review 1996 21 492 528 10.2307/258670 Peng MW Jiang Y Institutions behind family ownership and control in large firms Journal of Management Studies 2010 47 253 273 10.1111/j.1467-6486.2009.00890.x Peng MW Kathuria N COVID-19 and the scope of the firm Journal of Management Studies 2021 58 1431 1435 10.1111/joms.12699 Peng, M. W., Kathuria, N., Viana, F. L. E., & Lima, A. C. (2021). Conglomeration, (de)globalization, and COVID-19. Management and Organization Review, 17, 394–400. Peng MW Lee SH Wang DYL What determines the scope of the firm over time? A focus on institutional relatedness Academy of Management Review 2005 30 622 633 10.5465/amr.2005.17293731 Peng MW Luo Y Managerial ties and firm performance in a transition economy: The nature of a micro-macro link Academy of Management Journal 2000 43 486 501 10.2307/1556406 Peng MW Lebedev S Vlas CO Wang JC Shay JS The growth of the firm in (and out of) emerging economies Asia Pacific Journal of Management 2018 35 829 857 10.1007/s10490-018-9599-3 Peng MW Nguyen HW Wang JC Hasenhuttl M Shay J Bringing institutions into strategy teaching Academy of Management Learning and Education 2018 17 259 278 10.5465/amle.2017.0120 Peng MW Sun SL Pinkham BC Chen H The institution-based view as a third leg for a strategy tripod Academy of Management Perspectives 2009 23 63 81 10.5465/amp.2009.43479264 Peng MW Sun W Vlas C Minichilli A Corbetta G An institution-based view of large family firms: A recap and overview Entrepreneurship Theory and Practice 2018 42 187 205 10.1177/1042258717749234 Peng MW Wang DYL Jiang Y An institution-based view of international business strategy: A focus on emerging economies Journal of International Business Studies 2008 39 920 936 10.1057/palgrave.jibs.8400377 Peterson MF A culture theory commentary on Meyer and Peng’s theoretical probe into Central and Eastern Europe Journal of International Business Studies 2016 47 33 43 10.1057/jibs.2015.40 Petricevic O Teece DJ The structural reshaping of globalization: Implications for strategic sectors, profiting from innovation, and the multinational enterprise Journal of International Business Studies 2019 50 1487 1512 10.1057/s41267-019-00269-x Pezeshkan A Smith A Fainshmidt S Nair A A neo-institutional analysis of international venture capital attractiveness and performance: Insights for Asia-Pacific Asia Pacific Journal of Management 2022 39 365 393 10.1007/s10490-020-09727-9 Pfeffer J Barriers to the advance of organizational science: Paradigm development as a dependent variable Academy of Management Review 1993 18 599 620 10.2307/258592 Pinkham BC Peng MW Overcoming institutional voids via arbitration Journal of International Business Studies 2017 48 344 359 10.1057/s41267-016-0026-z Pinkse J Kolk A Multinational enterprises and climate change: Exploring institutional failures and embeddedness Journal of International Business Studies 2012 43 332 341 10.1057/jibs.2011.56 Porter ME The contributions of industrial organization to strategic management Academy of Management Review 1981 6 609 620 10.2307/257639 Powell, W. W. (1996). Commentary: On the nature of institutional embeddedness: Labels vs. explanation. In J. Baum & J. E. Dutton (Eds.), The embeddedness of strategy (pp. 293–300). JAI Press. Priem RL Butleter JE Is the resource-based “view” a useful perspective for strategic management research? Academy of Management Review 2001 26 22 40 Puffer SM McCarthy DJ Boisot M Entrepreneurship in Russia and China: The impact of formal institutional voids Entrepreneurship Theory and Practice 2010 34 441 467 10.1111/j.1540-6520.2009.00353.x Raynard M Lu F Jing R Reinventing the state-owend enterprises? Negotiating change during profound environmental upheaval Academy of Management Journal 2020 63 1300 1335 10.5465/amj.2017.1162 Ridge J Ingram A Hill A Beyond lobbying expenditures: How lobbying breadth and political connecttedness affect firm outcomes Academy of Management Journal 2017 60 1138 1163 10.5465/amj.2015.0584 Roberts PW Greenwood R Integrating transaction cost and institutional theories: Toward a constrained-efficiency framework for understanding organizational design adoption Academy of Management Review 1997 22 346 373 10.2307/259326 Rodrik D Populism and the economics of globalization Journal of International Business Policy 2018 1 12 33 10.1057/s42214-018-0001-4 Salvi E Belz F Bacq S Informal entrepreneurship: An integrative review and future research agenda Entrepreneurship Theory and Practice 2022 10.1177/10422587221115365 Sartor MA Beamish PW Offshoring innovation to emerging markets: Organizational control and informal institutional distance Journal of International Business Studies 2014 45 1072 1095 10.1057/jibs.2014.36 Sauerwald S Peng MW Informal institutions, shareholder coalitions, and principal-principal conflicts Asia Pacific Journal of Management 2013 30 853 870 10.1007/s10490-012-9312-x Scott WR The adolescence of institutional theory Administrative Science Quarterly 1987 32 493 511 10.2307/2392880 Scott WR Institutions and organizations 1995 Sage Scott WR Approaching adulthood: The maturing of institutional theory Theory and Society 2008 37 427 442 10.1007/s11186-008-9067-z Seligson D McCants A Coevolving institutions and the paradox of informal constraints Journal of Institutional Economics 2021 17 259 312 10.1017/S1744137420000600 Shi W Sun SL Peng MW Sub-national institutional contingencies, network positions, and IJV partner selection Journal of Management Studies 2012 49 1221 1245 10.1111/j.1467-6486.2012.01058.x Shin J Mendoza X Choi C Do internationalizing business group affiliates perform better after promarket reforms? Evidence from Korean SMEs Asia Pacific Journal of Management 2022 39 805 841 10.1007/s10490-021-09751-3 Shoham A The fundamental endogeneity of survey-based cultural dimension British Journal of Management 2022 10.1111/1467-8551.12599 Singh, K. (2007). The limited relevance of culture to strategy. Asia Pacific Journal of Management, 24, 421–428. Smit H Pennings E van Bekkum S Real options and institutions Journal of International Business Studies 2017 48 620 644 10.1057/s41267-016-0055-7 Smith WK Besharov ML Bowing before dual gods: How structured flexibility sustains organizational hybridity Administrative Science Quarterly 2019 64 1 44 10.1177/0001839217750826 Stevens CE Makarius EE Overcoming information asymmetry in foreign entry strategy: The impact of reputation Global Strategy Journal 2015 5 3 256 272 10.1002/gsj.1099 Stevens CE Xie E Peng MW Toward a legitimacy-based view of political risk: The case of Google and Yahoo in China Strategic Management Journal 2016 37 945 963 10.1002/smj.2369 Su Z The co-evolution of institutions and entrepreneurship Asia Pacific Journal of Management 2021 38 1327 1350 10.1007/s10490-019-09703-y Su Z Peng MW Xie E A strategy tripod perspective on knowledge creation capability British Journal of Management 2016 27 58 76 10.1111/1467-8551.12097 Suchman MC Managing legitimacy: Strategic and institutional approaches Academy of Management Review 1995 20 571 610 10.2307/258788 Sun P Doh JP Rajwani T Siegel D Navigating cross-border institutional complexity: A review and assessment of multinational nonmarket strategy research Journal of International Business Studies 2021 10.1057/s41267-021-00438-x Sun SL Liang H Globalization and affordability of microfinance Journal of Business Venturing 2021 36 106065 10.1016/j.jbusvent.2020.106065 Sun SL Peng MW Tan W Institutional relatedness behind product diversification and international diversification Asia Pacific Journal of Management 2017 34 339 366 10.1007/s10490-016-9498-4 Sun SL Shi W Ahlstrom D Tian L Understanding institutions and entrepreneurship: The microfoundations lens and emerging economies Asia Pacific Journal of Management 2020 37 957 979 10.1007/s10490-020-09738-6 Tan JJ Litschert RJ Environment-strategy relationship and its performance implications: An empirical study of the Chinese electronics industry Strategic Management Journal 1994 15 1 20 10.1002/smj.4250150102 Tan JJ Peng MW Organizational slack and firm performance during economic transitions: Two studies from an emerging economy Strategic Management Journal 2003 24 1249 1263 10.1002/smj.351 Teece DJ Explicating dynamic capabilities: The nature and microfoundations of (sustainable) enterprise performance Strategic Management Journal 2007 28 1319 1350 10.1002/smj.640 Thomas LDW Tee R Generativity: A systematic review and conceptual framework International Journal of Management Reviews 2022 24 255 278 10.1111/ijmr.12277 Thornton PH Markets from culture: Institutional logics and organizational decisions in higher education publishing 2004 Stanford University Press Ulziisukh S Wei Z Behind the political connections under emerging democracies Management and Organization Review 2022 18 686 716 10.1017/mor.2021.74 Vanacker T Collewaert V Zahra S Slack resources, firm perfromance, and the institutional context: Evidence from privately held European firms Strategic Management Journal 2017 38 1305 1326 10.1002/smj.2583 Vogel D The private regulation of global corporate conduct: Achievements and limitations Business & Society 2010 49 68 87 10.1177/0007650309343407 Voigt S How to measure informal institutions Journal of Institutional Economics 2018 14 1 22 10.1017/S1744137417000248 Voronov M Vince R Integrating emotions into the analysis of institutional work Academy of Management Review 2012 37 58 81 Voronov M Weber K The heart of institutions: Emotional competence and institutional actorhood Academy of Management Review 2016 41 1 33 10.5465/amr.2013.0458 Walgenbach P Drori GS Hollerer MA Between local mooring and global orientation: A neo-institutional theory perspective on the contemporary multinational corporation Research in the Sociology of Organizations 2017 49 99 125 10.1108/S0733-558X20160000049004 Wan WP Hoskisson RE Home country environments, corporate diversification strategies, and firm performance Academy of Management Journal 2003 46 1 27 45 10.2307/30040674 Wang JC Markoczy M Sun SL Peng MW She’-E-O compensation gap: A role congruity view Journal of Business Ethics 2019 159 745 760 10.1007/s10551-018-3807-4 Wang, J. C., Yi, J., Zhang, X., & Peng, M. W. (2022). Pyramidal ownership and SOE innovation. Journal of Management Studies. 10.1111/joms.12803. Welbourne Eleazar, M. J., Peng, M. W., & Pinkham, B. C. (2022). Weaponizing institutions in competitive dynamics. Working paper. University of Iowa. Weng DH Lee SH Yamakawa Y Time to change lanes: How pro-market reforms affect informal ventures’ formalization speed Global Strategy Journal 2021 11 767 795 10.1002/gsj.1421 Weng DH Peng MW Home bitter home: How labor protection influences firm offshoring Journal of World Business 2018 53 632 640 10.1016/j.jwb.2018.03.007 Williamson OE The new institutional economics: Taking stock, looking ahead Journal of Economic Literature 2000 38 595 613 10.1257/jel.38.3.595 Witt MA De-globalization: Theories, predictions, and opportunities for international business research Journal of International Business Studies 2019 50 1053 1077 10.1057/s41267-019-00219-7 Witte CT Burger MJ Pennings E When political instability devaluates home-host ties Journal of World Business 2020 55 4 101077 10.1016/j.jwb.2020.101077 Wright M Filatotchev I Hoskisson RE Peng MW Strategy research in emerging economies: Challenging the conventional wisdom Journal of Management Studies 2005 42 1 33 10.1111/j.1467-6486.2005.00487.x Wright M Wood G Musacchio A Okhmatovskiy I Grosman A Doh JP State capitalism in international context: Varieties and variations Journal of World Business 2021 56 2 101160 10.1016/j.jwb.2020.101160 Yamakawa Y Peng MW Deeds DL What drives new ventures to internationalize from emerging to developed economies? Entrepreneurship Theory and Practice 2008 32 59 82 10.1111/j.1540-6520.2007.00216.x Yi J Meng S Macaulay CD Peng MW Corruption and foreign direct investment phases: The moderating role of institutions Journal of International Business Policy 2019 2 167 181 10.1057/s42214-019-00024-x Yiu D Makino S The choice between joint venture and wholly owned subsidiary: An institutional perspective Organization Science 2002 13 667 683 10.1287/orsc.13.6.667.494 Yiu DW Lam LW Gaur A Lee SH Wong CS Asian relevance, global impact: Asian management research entering a new era Asia Pacific Journal of Management 2018 35 565 571 10.1007/s10490-018-9606-8 Young, M. N., Bruton, G. D., Peng, M. W., & Yu, A. (2022). American corporations are from Mars, Chinese corporations are from Venus. Business Horizons, 65, 505–517. Young MN Peng MW Ahlstrom D Bruton GD Jiang Y Corporate governance in emerging economies: A review of the principal-principal perspective Journal of Management Studies 2008 45 196 220 10.1111/j.1467-6486.2007.00752.x Young MN Tsai T Wang X Liu S Ahlstrom D Strategy in emerging economies and the theory of the firm Asia Pacific Journal of Management 2014 31 331 354 10.1007/s10490-014-9373-0 Zhao EY Fisher G Lounsbury M Miller D Optimal distinctiveness: Broadening the interafce between institutional theory and strategic management Strategic Management Journal 2017 38 93 113 10.1002/smj.2589 Zhou KZ Gao GY Zhao H State ownership and firm innovation in China: An integrated view of institutional and efficiency logics Administrative Science Quarterly 2017 62 375 404 10.1177/0001839216674457 Zhu H Ma X Sauerwald S Peng MW Home country institutions behind cross-border acquisition performance Journal of Management 2019 45 1315 1342 10.1177/0149206317699520 Zoogah D Peng MW Woldu H Institutions, resources, and organizational effectiveness in Africa Academy of Management Perspectives 2015 29 7 31 10.5465/amp.2012.0033	0	PMC9715421	NO-CC CODE	2022-12-03 23:20:15	no	Asia Pac J Manag. 2022 Dec 2;:1-30	utf-8	null	null	null	oa_other
==== Front Int J Mach Learn Cybern Int J Mach Learn Cybern International Journal of Machine Learning and Cybernetics 1868-8071 1868-808X Springer Berlin Heidelberg Berlin/Heidelberg 1723 10.1007/s13042-022-01723-3 Original Article Generative face inpainting hashing for occluded face retrieval Yang Yuxiang [email protected] 1 Tian Xing [email protected] 1 http://orcid.org/0000-0003-0783-3585 Ng Wing W. Y. [email protected] 1 Wang Ran [email protected] 2 Gao Ying [email protected] 1 Kwong Sam [email protected] 3 1 grid.79703.3a 0000 0004 1764 3838 School of Computer Science and Engineering, South China University of Technology, Guangzhou, 510006 China 2 grid.263488.3 0000 0001 0472 9649 College of Mathematics and Statistics, Shenzhen University, Shenzhen, 518060 China 3 grid.35030.35 0000 0004 1792 6846 Department of Computer Science, City University of Hong Kong, Hongkong, 999077 China 2 12 2022 114 3 5 2022 9 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. COVID-19 has resulted in a significant impact on individual lives, bringing a unique challenge for face retrieval under occlusion. In this paper, an occluded face retrieval method which consists of generator, discriminator, and deep hashing retrieval network is proposed for face retrieval in a large-scale face image dataset under variety of occlusion situations. In the proposed method, occluded face images are firstly reconstructed using a face inpainting model, in which the adversarial loss, reconstruction loss and hash bits loss are combined for training. With the trained model, hash codes of real face images and corresponding reconstructed face images are aimed to be as similar as possible. Then, a deep hashing retrieval network is used to generate compact similarity-preserving hashing codes using reconstructed face images for a better retrieval performance. Experimental results show that the proposed method can successfully generate the reconstructed face images under occlusion. Meanwhile, the proposed deep hashing retrieval network achieves better retrieval performance for occluded face retrieval than existing state-of-the-art deep hashing retrieval methods. Keywords Occlusion Face retrieval Inpainting Generative adversarial http://dx.doi.org/10.13039/501100001809 National Natural Science Foundation of China 61876066 62176160 61672443 Ng Wing W. Y. Wang Ran Kwong Sam Chinese Postdoctoral Science Foundation2020M672631 Tian Xing Hong Kong RGC General Research Funds9042489 (CityU 11206317) Grant 9042816 (CityU 11209819) Grant 9042322 (CityU 11200116) Kwong Sam Natural Science Foundation of Guangdong Province of China2022A1515010791 Wang Ran Natural Science Foundation of Shenzhen20200804193857002 Wang Ran ==== Body pmcIntroduction As of this article date, because of the communicable disease (such as COVID-19) or other reasons, people always wear a mask, hat, or glasses outside. These items block most of the face information including eyes, nose, and mouth. Existing face retrieval and face recognition systems cannot perform well when encountering challenges such as large-pose variation, varying illumination, low resolution, different facial expressions, and occlusion [1]. Therefore, how to retrieve large-scale face images efficiently and accurately under occlusion has become a key problem in current human life and scientific research. The previous work to improve the performance of face recognition under occlusion can be generally partitioned into three categories, i.e., occlusion robust feature extraction, occlusion aware face recognition, and occlusion reconstruct based face recognition. The occlusion robust feature extraction methods [2, 3] adopt the data augmentation method to expand the datasets, which alleviate the effect of face occlusion. However, these methods are limited to some special occlusion situations in recognition, which means that they may not perform well under different occlusions. Another possible approach is to add a MaskNet [4] branch in the deep networks to better learn the facial feature representation of the unoccluded region. The MaskNet is used to assign higher weights to hidden units activated by the unoccluded regions. However, there is not enough supervision information to train the MaskNet and the outputs discriminability of middle convolutional layer is not enough. Similarly, a mask learning strategy [5] is proposed to build a mask dictionary that corresponds between occluded regions and missing feature representation. The occlusion reconstruct based face recognition method intends to recover a face without occlusion [6, 7] to improve the performance of face recognition systems. Face retrieval has been widely used in many application areas, such as surveillance, forensics and security. Given a query face image, the target of face retrieval task is to retrieve face images that are similar to it from a large-scale face image dataset. Hashing technique, as an advanced indexing technique, has been widely researched to handle this task due to its high retrieval efficiency and low space cost [8–13]. With compact hash codes generated for face images, similarities between face images can be evaluated efficiently based on the Hamming distances, which can be computed quickly by computers. Generally, feature extraction plays an importance role in the performance of most existing hashing methods. Traditionally, hand-craft visual features are employed for face images, such as HOG [14], LBP [15], GIST [16] and SIFT [17]. With the development of deep learning techniques for feature learning [18–20], some deep hashing methods are proposed to improve the efficiency and retrieval accuracy of hash learning, such as CNNH [21], DH [22], DSH [23], DSHSD [24], DPSH [25], Hashnet [26], and CSQ [27]. Because the retrieval performance could be greatly improved by facial feature learning with deep neural network, many deep hashing networks have been proposed for face retrieval such as DDQH [28], DHCQ [29], DCBH [30], and DFH-GAN [31]. These methods achieve encouraging performance for face image retrieval. However, the retrieval performance has been greatly reduced because some of the face components are blocked under occlusions, which makes them fail to adapt to face retrieval problems in occlusion environments. In this paper, we propose an effective occluded face retrieval method based on a deep generative model and hashing retrieval network, which decomposes the problem of face retrieval under partial occlusion into two stages: face inpainting and generated face retrieval. The occluded face images are reconstructed using a face inpainting model trained with a combination of adversarial loss, reconstruction loss and hash bits loss, which encourages the hash codes of the real face image and the reconstructed face image to be as close as possible. Major contributions of this work are summarized as follows: An occluded face retrieval framework is proposed for face retrieval under several occlusion situations, named Generative Face Inpainting Hashing (GFIH). To the best of our knowledge, GFIH is the first approach combining generative adversarial network and deep hashing network to learn the hash codes for occluded face retrieval. A joint loss function consisting of adversarial loss, reconstruction loss, and hash bits loss is proposed to encourage the generative model to reconstruct a similarity-preserving face image without occlusion. This facilitates the hashing retrieval network to generate compact similarity-preserving hashing codes. Six face occlusion image datasets are created to simulate six different face occlusion situations with different occlusion regions for face retrieval performance evaluation. Quantitative experimental results show that GFIH obtains outstanding occluded face retrieval performance than other comparative methods. The rest of this paper is organized as follows: Sect. 2 introduces related works on existing face inpainting models and hashing-based face retrieval models. The proposed GFIH is introduced in Sect. 3. Experimental results are discussed in Sect. 4. Section 5 concludes our work in this paper. Related work In this section, two most related works to the proposed method are described briefly. Section 2.1 introduces the existing face inpainting models. The hashing-based retrieval models are introduced in Sect. 2.2. Existing face inpainting models Previous face inpainting models can generally be divided into two categories: Non-learning inpainting methods and Learning inpainting methods. The non-learning inpainting methods [32, 33] is traditional diffusion-based or patch-based models with low-level features. The learning inpainting methods [34–36] reconstruct a face without occlusion by using deep learning and generative adversarial networks. Some previous face inpainting models use an autoencoder architecture to generate the occluded face region [6, 7]. Context Encoders [6] firstly propose a deep learning method for image inpainting tasks, which employs a generative adversarial network. The input occluded images are created by adding some masked region on the original normal images. This method can learn the feature representation of the occluded image and generate the coherent contents by optimizing the adversarial loss. However, this method focuses more on unsupervised feature learning rather than image inpainting. It is no clear if the generated content can help improve the image retrieval network to learn the compact similarity-preserving hashing codes sufficiently. An effective object completion algorithm is proposed in [7] using a deep generative model and a face parsing network. Two adversarial loss functions are used to jointly train the autoencoder and discriminator. The first adversarial loss tried to help improve the generated content of occluded region more realistic. The second adversarial loss tries to help improve the entire reconstructed image which consist of the generated content and unoccluded region of the original image more realistic. A face parsing network is proposed as an additional loss to regularize the generation procedure, which facilitates the generator to generate more reasonable and consistent face inpainting images. However, the performance to generate a fine-detailed content is not well enough. A high-resolution image inpainting method is proposed in [34] using a multi-scale neural patch synthesis approach, which jointly optimize the image contents and texture constraints. The output of context encoder is employed to generate a high-resolution image by gradually increasing texture details. However, the optimization in this method significantly increases computational costs. Partial convolutions are used in [35] to help the convolution filters focus on the unoccluded regions. This approach renormalizes the convolution filter to be conditioned on only valid pixels by assigning the convolution weights with mask value. EdgeConnect [36] decomposes the image inpainting problem into two stages: structure prediction and image completion. The image structure of the occluded regions is predicted to guide the image inpainting process. However, the EdgeConnect method tend to perform unsatisfactorily in generating contents from highly textured areas and large occluded images. MAT [40] proposed a novel transformer-based model for large hole inpainting to efficiently process high-resolution images. With the goal of achieving higher retrieval performance for occluded face retrieval by employing the reconstructed face images, an additional hash bits loss is proposed to encourage the hashing retrieval network to generate compact similarity-preserving hashing codes. Hashing-based retrieval models Generally, deep hashing retrieval methods construct a hash function by incorporating a convolution neural network (CNN) model to learn the similarity-preserving feature representation. Deep Supervised Hashing (DSH) [23] learns compact similarity-preserving hashing codes by using the deep feature representation extracted by convolution network of image pairs (similar/dissimilar) and the pairwise similarity. Deep supervised hashing based on stable distribution (DSHSD) [24] is proposed to solve the problem of feature distribution changes caused by the quantization regularizer. A smooth projection is used to help improve the efficiency of the training convergence and make the output binary code preserve more similarity. Deep pairwise-supervised hashing (DPSH) [25] proposes an end-to-end architecture which performs feature learning and hash-code learning simultaneously based on pairwise labels. Hashnet [26] proposes a novel deep architecture for hash code learning by continuation method with convergence guaranteed. It can learn exactly binary hash codes from imbalanced similarity data. The ill-posed gradient problem is solved by optimizing deep networks with non-smooth binary activations. In addition, a new global similarity metric, named as central similarity, is proposed in Central Similarity Quantization (CSQ) [27]. This metric is used to encourage hash codes of similar image pairs to approach a common center and encourage the dissimilar image pairs to converge to different centers. Many deep hashing networks have been proposed for face retrieval. DDQH [28] is proposed to capture the multiscale feature of face images for hashing codes learning. The feature representation is learned by fusing the output of the last convolutional layer and the last pooling layer. Another deep hashing face retrieval method, DHCQ [29] is proposed to retrieve scalable face images. A loss function consists of quantization error and prediction error is used to optimize the by capturing discriminative facial representations retrieve the discriminative facial feature learning. To solve the problems of inter-class similarities and intra-class variations, DCBH [30] is proposed to learn the robust and multi-scale feature representations. The center-clustering loss is used to encourage the face images of intra-class to approach a common center. Besides, a block hashing layer is used to reduce the number of parameters but also can generate the compact similarity-preserving hashing codes. DFH-GAN [31] proposes a deep face hashing retrieval method combined with generative adversarial network. GAN is employed to generate fake images to augment the training dataset, so the hashing network can be trained from both real images and diverse synthesized images to learns compact binary hash codes. However, these hashing methods focus on normal images, which are not effective to handle the occluded face retrieval problem. Hence, we are motivated to propose the GFIH to combining generative adversarial network and deep hashing network to learn the hash codes for occluded face retrieval. Generative face inpainting hashing The proposed Generative Face Inpainting Hashing (GFIH) decomposes the problem of face retrieval under partial occlusion into two stages: face inpainting and hashing retrieval stages. The occluded face images are reconstructed using a face inpainting model firstly which consists of a loss function and two inpainting networks: generator and discriminator. Then, a deep hashing retrieval network is used to perform the face retrieval using reconstructed face images from the previous stage for a better retrieval performance. Figure 1 shows an overview of the GFIH. The face inpainting network, the loss function, and the hashing retrieval network of the GFIH will be described in Sects. 3.1, 3.2, and 3.3 respectively.Fig. 1 Overview of the GFIH Face inpainting networks of GFIH The generator in the face inpainting model is designed to inpaint the masked region in the occluded face image. The overall structure of the generator is an encoder-decoder pipeline, as shown in Fig. 1. Different from the original GAN model [37], the latent feature representation is used to generate new content instead of a random noise vector. The latent feature representation is extracted by the encoder with the input occluded face images. Then, the decoder reconstructs the masked region using aforementioned feature representation. Architectures of the discriminator and the encoder in the generator are similar to the architecture of discriminator in [6, 38], which is a series of four fractionally-stride convolutional layers. Stride convolutional layers allow the network to learn its spatial upsampling by replacing the deterministic spatial pooling functions (such as maxpooling). The encoder features are projected to a small spatial extent convolution representation with many feature maps. Then, five upconvolution layers are employed to reconstruct the occluded region of face image from aforementioned high-level feature representation. The upconvolution layers is a series of transposed convolution which can be consider as upsampling followed by fractionally strided convolutions to reconstruct a higher resolution image. The rectified linear unit (ReLU) activation function is employed in the decoder, while leaky ReLU is employed in the encoder and discriminator. Batch normalization is employed to normalize the input of each unit to zero mean and variance. Generally, there is an explosion problem of the number of network parameters when using the fully connected layers to connect the high-level feature representation and decoder. To solve this problem, channel-wise fully connected layers [6] are employed to propagate the information across feature maps by replacing the fully connected layers. Unlike fully connected layers, there are no parameters to connecting each feature map in the channel-wise fully connected layers, which is followed by a stride 1 convolution. Therefore, the number of parameters in a channel-wise fully connected layer is mn4 form feature maps of size n×n. Because of all the activations are directly connected to each other in the fully connected layers, the number of parameters in a fully connected layer is m2n4 form feature maps of size n×n. The number of parameters is significantly reduced, which help improve the efficiency of training model. The occluded region of face image can be filled using the generator by minimizing the reconstruction errors, but the generator may only learn the rough shape of the unoccluded region of face image, which will result in a fuzzy and rough generated content. To encourage the reconstructed face images to look realistic and coherent, a discriminator is employed to help improve the quality of generated details. The discriminator can be trained to distinguish the real face images and inpainting face images, while help improve the ability of generator to generate a face images that can fool the discriminator. This facilitates the face inpainting model to generate a more realistic face image without occlusion. Loss function of inpainting networks A joint loss consisting of adversarial loss, reconstruction loss and hash bits loss is proposed to learn parameters of both the generator and the discriminator in the inpainting model. The adversarial loss tries to make the reconstructed image more realistic and has the effect of matching the distribution of the reconstructed image with the distribution of the original face image. The reconstruction loss tries to help the generator learn the knowledge of the overall structure of the unoccluded region and keep the generated content consistent. The hash bits loss is a reflection of the Hamming distance between the hash codes of the reconstructed face image and real face image. By optimizing the hash bits loss, it facilitates the network to generate a reconstructed face image whose hash code is similar to the real face image. By using the discriminator, the adversarial loss is employed to measure the ability of the generator to fool the discriminator, and the ability of the discriminator to distinguish the real and fake face images. The adversarial loss of the proposed model is based on generative adversarial networks. A generator G and a discriminator D are jointly trained in the GAN model. The discriminator D can provides loss gradients to generator G. The training process is a two-player game. The discriminator can be trained to distinguish the ground truth samples and the generated samples of generator G, while help improve the ability of generator G to generate the data pixels that can fool the discriminator D. The objective function for discriminator is logistic likelihood, which indicates whether the input face image is real face or generated one:1 minGmaxDEx∼pdata(x)[logD(x)]+Ez∼pz(z)[log(1-D(G(z)))] where pdata(x) and pz(z) denote distributions of real image and noise, respectively. By employing this method, this face inpainting model is adopted for face inpainting by modeling generator. Let M be a binary mask corresponding to the occluded region of face image with a value of 1 for the occluded region and 0 for unoccluded region. For each face image, the occluded region M⊙x is automatically generated for each face image to simulate face occlusion situations. The adversarial loss Ladv is defined as:2 Ladv=minGmaxDEx∼pdata(x)[logD(x)+log(1-D(G((1-M)⊙x)))] where ⊙ and (1-M)⊙x denote the element-wise product operation and the distribution of face image with masked regions, respectively. During training, generator G and discriminator D are optimized jointly using alternating SGD. This objective encourages the reconstructed face images to look realistic and coherent. A reconstruction loss Lrec is another component of the joint loss function for generator, which is the L2 distance between the reconstructed face images and real face images:3 Lrec=‖M⊙(x-G((1-M)⊙x))‖22 Hash bits loss is employed to encourage the generator to generate a reconstructed face image xg=G((1-M)⊙x) whose hash code is close to the real face image. It is defined as:4 Lh=‖h(G((1-M)⊙x))-h(x)‖22+α‖h(G((1-M)⊙x))-1‖1 where h(·) and α denote the hash binary code of the images, and a weighting parameter that controls the strength of the regularizer, respectively. h(·) is generated by using a scaled tanh function tanhβz to binarized the feature representation into a K-bit binary hash code, which will be described in detail in Sect. 3.3. The first term encourages the hash codes of the reconstructed face image and corresponding original face image to be as close as possible. The second term is an additional regularizer to replace the binary constraints. Generally, a sigmoid or tanh function is used as a relaxation method to approximate the thresholding procedure. However, optimizing the generative network with these non-linear functions would cause the convergence of the network become difficult and slow [23]. To alleviate this problem, an additional regularizer is adopted to encourage the output values to approach the binary code in the hash bit loss. The overall loss function, a combination of adversarial loss, reconstruction loss and hash bits loss, is defined as:5 L=λ1Ladv+λ2Lrec+λ3Lh where λ1,λ2 and λ3 are weights to balance the effects of different loss. Hashing retrieval network The proposed occluded face retrieval method focuses on the occluded face inpainting learning. By employing the reconstructed face images, the deep hashing retrieval network is expected to achieve higher precision for face retrieval under occlusion. So, the original hash codes of the real face images directly influence the performance of the proposed occluded face retrieval method. Note that there are no limits on the method for learning the original hash codes of the real face images, which means that all existing hashing methods can be used. In this paper, Hashnet [26] is selected for binary hash codes learning in GFIH. Our proposed method first trains the Hashnet using real face image pairs and pairwise similarity to learn the similarity-preserving hash codes of real face images. Then, the trained Hashnet generates binary hash codes for newly coming inpainting face query images. By employing the computation of Hamming distances among hash codes of query images and real face images, GFIH return top k face images yielding the smallest Hamming distance. The architecture of Hashnet consists of a convolutional neural network and a fully connected hash layer, which accepts the pairwise input faces xi,xj,sij. The convolutional neural network is used to learn discriminative feature representations of each face image xi, and the feature representations are transformed into K dimensional representation zi∈Rk by the fully connected hash layer. Then the K dimensional representation zi is binarized into a K-bits binary hash code hi∈{-1,1}k by an activation function hi=signzi. For a pair of face images xi and xj, let dist Hhi,hj be the Hamming distance between a pair of binary hash codes hi and hj, which is expected to preserve the similarity among image pairs. Note that, there exists a relationship between distHhi,hj and inner product, distHhi,hj=12K-hi,hj. Therefore, inner product can be adopted to represent the similarity. Given the set of pairwise similarity S=sij, the Weighted Maximum Likelihood (WML) estimation of the hash codes H= h1,…,hN for all training points is defined as:6 logP(S∣H)=∑sij∈SwijlogPsij∣hi,hj where P(S∣H) and wij denote the weighted likelihood function and the weight for each training pair xi,xj,sij, respectively. To solve the data imbalance problem caused by the difference in the number of similar image pairs and dissimilar image pairs, the wij is employed to assign the weights of the image pairs according to the importance of misclassifying that pair. For a pair of hash codes hi and hj, Psij∣hi,hj is the conditional probability of pairwise similarity sij. The optimization problem of Hashnet is derived as:7 minθ∑sij∈Swijlog1+expγhi,hj-γsijhi,hj where θ and γ denote the parameter of the feature learning model and the hyper-parameter of adaptive sigmoid function to control its bandwidth, respectively. By optimizing the WML estimation, the hashing network can learn exactly binary hash codes from imbalanced image pairs. However, optimizing deep networks with sign activation may cause the gradient vanishing problem, which makes it difficult to optimize the network using the standard backpropagation. There exists a relationship between the sign function and the scaled tanh function in the concept of limit in mathematics: limβ→∞tanhβz=sign(z). According to this relationship, the scaled tanh function can be used to replace sign function to optimize the Hashnet. The learning procedure starts with a smoothed activation function y=tanhβz. Then, increase the value of β to make the scaled tanh function approach to the original sign function. For face retrieval, there is a big difference in the number of intra-class face pairs and inter-class face pairs, so Hashnet is adopted to generate the compact similarity-preserving hashing codes using reconstructed face images for a better retrieval performance in the proposed method. Experiments Dataset and performance metric Experiments are conducted on two datasets to evaluate the retrieval performance of the proposed method. The input to the face inpainting generator is an image with one or more masked regions. A masked region in the input occluded face image is filled with constant mean value. The masked region could be of any shape. Six different strategies are proposed here to simulate six different face occlusion situations in the actual environment, which consist of people wearing hat, glasses, mask, hat+glasses, glasses+mask and hat+mask. Among these situations, different key components (e.g., eyes, nose, and mouth) that play an important role in retrieval performance are masked. The samples of six different occlusion situations in two datasets are shown in the second row of Fig. 2.Fig. 2 The face inpainting result of six different face occlusion situations on two datasets The CelebA [39] dataset is used to generate two datasets CelebA-1H and CelebA-1K. Each face image in CelebA dataset is cropped, roughly aligned by the position of two eyes, and rescaled to 128×128×3 pixels. The CelebA-1H dataset contains 2752 face images of 97 people, 1541 face images for training, 602 for validation and 609 for testing. The CelebA-1K dataset contains 26963 face images of 954 people, 15104 face images for training, 5921 for validation and 5938 for testing. The MFRD dataset contains 90,000 face images without masks, 2203 face images with masks of 525 people. Different from CelebA-1H and CelebA-1K datasets, most of occluded face images in MFRD for test are real images in the wild. The quality of generated face images plays an important role in improving retrieval performance. Peak Signal to Noise Ratio (PSNR), Structural Similarity Index (SSIM) are used in this work to evaluate the quality of generated face images. We can well measure the similarity between the generated face and the original face at different levels using these quality metrics, the PSNR measure the pixel-level difference well, and SSIM measures the structure similarity of generated images to original ones. The performance metric of face retrieval employed in this work is mean average precision (mAP). mAP is the mean of average precision (AP) of all query data, which amounts to the area under the precision-recall (PR) curve. An image retrieval network with higher mAP corresponds to a larger area under its PR curve, which means that it has a better retrieval performance. For an image dataset, the AP of every images are average to output a value to evaluate the retrieval performance of the entire dataset. Therefore, mAP is used to evaluate occluded face retrieval performance on the CelebA-1h and CelebA-1k datasets. The AP of n query data is defined as:8 AP=1F∑k=1nTkkΔTk where F,k,Tk, and ΔTk denote the total number of samples in the database that have the same label with the query, the total number of returned samples, the number of returned samples which have the same label with the query among k returned samples, and the change in recall from item k-1 to k, respectively. Performance for face inpainting The generator of the proposed method is trained with the joint loss function defined in Equation 5 for the task of face inpainting under partial occlusion. By employing the reconstructed face images, the deep hashing retrieval network is expected to achieve higher precision for occluded face retrieval. The deep hashing retrieval network is learned by exploiting a deep CNN model to extract appropriate feature representation for the real face images and then generate the hashing codes using the supervised information (similarity and dissimilarity) of image pairs. While the parameters of the hashing retrieval network are already learned and the binary hash codes for the real face images are already given, they remain unchanged during the whole generator learning process. The default solver hyper-parameters λ1,λ2 and λ3 are set to be 0.8325, 0.1665 and 0,001 , respectively. And a higher learning rate is used for generator (10 times) to that of adversarial discriminator. The face inpainting result of six different face occlusion situations on two datasets are shown in Fig. 2. The first row is the real face image in two datasets. In the second row, six different face occlusion situations are presented, which can simulate most of the face occlusion situations in the actual environment. The second row of each panel shows some results of the proposed method which are visually realistic and pleasing. The third row presents the corresponding reconstructed face images using the proposed generative model. It can be seen that the inpainting face images look realistic and coherent. The occluded region is filled with generated content that fit well within the context. It shows that the inpainting results of the proposed method are encouraging regardless of the mask locations. The quantitative evaluation results of the generated images are given in Table 1. PNSR and SSIM are calculated to evaluate the quality of generated face images. To evaluate the effectiveness of our proposed inpainting method, the inpainting method Context Encoder (CE) [6] and MAT [40] are employed as a comparison method. As shown in Table 1, the proposed generative model in GFIH achieves a better PNSR and SSIM results against CE in most of the compared experiments on both two datasets. However, compared to MAT, the SSIM results of MAT are better than GFIH, but GFIH achieves a better PNSR. In this paper, our goal is to retrieve face images under several occlusion situations, so further retrieval experimental result are given in Sect. 4.4 to compare the performance of these two inpainting methods for occluded face retrieval.Table 1 PSNR and SSIM comparisons of inpaining methods for CelebA-1H and CelebA-1K dataset Methods CELEBA-1H CELEBA-1K PNSR SSIM PNSR SSIM Hat CE [6] 24.056 0.851 24.207 0.858 MAT [40] 22.785 0.889 24.719 0.905 GFIH 24.285 0.863 24.520 0.866 Glasses CE [6] 29.581 0.945 29.719 0.949 MAT [40] 26.045 0.926 27.320 0.955 GFIH 29.675 0.950 29.789 0.950 Mask CE [6] 27.552 0.916 27.692 0.919 MAT [40] 25.102 0.905 27.301 0.933 GFIH 27.700 0.919 28.419 0.927 Hat+Glasses CE [6] 24.059 0.849 23.938 0.847 MAT [40] 23.287 0.902 23.09 0.901 GFIH 24.369 0.856 24.353 0.858 Glasses+Mask CE [6] 24.778 0.859 25.186 0.866 MAT [40] 24.503 0.903 24.787 0.913 GFIH 25.133 0.862 25.356 0.870 Hat+Mask CE [6] 24.052 0.838 24.535 0.845 MAT [40] 23.780 0.863 23.926 0.876 GFIH 24.993 0.849 25.273 0.855 The bold font indicate the largest values in the corresponding column Comparison with state-of-the-art hashing retrieval methods To evaluate the retrieval performance of the proposed method, several deep hashing methods are compared including DSH [23], DSHSD [24], DPSH [25], Hashnet [26] and CSQ [27]. The mAP is calculated to evaluate the retrieval accuracy of the proposed method and these several competitors. A set of experiments is conducted to confirm the effectiveness of the proposed method in six different occlusion situations. The results of the proposed method compared with existing hashing methods on two datasets are given in Table 2. The third column of Table 2 reports the mAP results of the proposed method (labeled as GFIH) and existing hashing methods using different hash bits length for CelebA-1H dataset, while the fourth column of Table 2 is for CelebaA-1K dataset. It can be seen from Table 2 that the proposed method yields the best results against the state-of-the-art hashing retrieval methods in all occlusion situations, which means that the proposed method can achieve higher precision for occluded face retrieval by employing the reconstructed face images. A combination of inpainting model and the Hashnet retrieval model enables the face retrieval model to perform better under different occlusion situations. Moreover, it can be seen that the mAP of the proposed method remains in a relatively stable and acceptable range under six different occlusion situations. This proves the effectiveness of the proposed method under different occlusion situations, which can simulate most of the face occlusion situations in the actual environment. It also should be noted that the proposed method achieves a best retrieval performance under the occlusion situation that people wearing a mask among these six different occlusion situations. This may imply that the eyes and forehead region play an important role in face inpainting than other regions.Table 2 mAP comparisons of deep hashing methods for CelebA-1H and CelebA-1K dataset Methods CELEBA-1H CELEBA-1K 16bits 32bits 64bits 16bits 32bits 64bits 128bits Hat DSH [23] 0.138 0.168 0.294 0.011 0.014 0.028 0.033 DSHSD [24] 0.152 0.288 0.352 0.005 0.013 0.013 − DPSH [25] 0.143 0.150 0.172 0.013 0.021 0.040 0.045 CSQ [27] 0.076 0.163 0.185 − − 0.026 0.030 Hashnet [26] 0.087 0.138 0.210 0.023 0.040 0.071 0.090 GFIH 0.242 0.351 0.431 0.042 0.104 0.182 0.237 Glasses DSH [23] 0.110 0.201 0.294 0.014 0.012 0.020 0.025 DSHSD [24] 0.173 0.198 0.209 0.004 0.010 0.007 − DPSH [25] 0.141 0.290 0.311 0.028 0.045 0.103 0.105 CSQ [27] 0.119 0.286 0.358 − − 0.015 0.020 Hashnet [26] 0.195 0.257 0.350 0.027 0.060 0.106 0.149 GFIH 0.309 0.475 0.530 0.046 0.136 0.197 0.270 Mask DSH [23] 0.117 0.142 0.409 0.026 0.020 0.029 0.035 DSHSD [24] 0.254 0.350 0.435 0.011 0.023 0.024 − DPSH [25] 0.147 0.280 0.327 0.027 0.040 0.109 0.112 CSQ [27] 0.122 0.293 0.415 - - 0.017 0.160 Hashnet [26] 0.183 0.311 0.435 0.041 0.081 0.141 0.189 GFIH 0.276 0.489 0.579 0.057 0.166 0.285 0.385 Hat+Glasses DSH [23] 0.0782 0.139 0.157 0.006 0.008 0.011 0.012 DSHSD [24] 0.086 0.135 0.150 0.003 0.005 0.006 − DPSH [25] 0.083 0.141 0.160 0.015 0.021 0.053 0.055 CSQ [27] 0.067 0.133 0.166 − − 0.012 0.015 Hashnet [26] 0.084 0.127 0.162 0.017 0.026 0.046 0.056 GFIH 0.201 0.288 0.345 0.035 0.075 0.117 0.147 Glasses+Mask DSH [23] 0.099 0.143 0.240 0.006 0.007 0.008 0.007 DSHSD [24] 0.136 0.169 0.179 0.005 0.009 0.009 − DPSH [25] 0.121 0.210 0.254 0.022 0.031 0.079 0.071 CSQ [27] 0.083 0.252 0.317 − - 0.017 0.033 Hashnet [26] 0.171 0.214 0.281 0.025 0.043 0.065 0.079 GFIH 0.263 0.357 0.436 0.039 0.102 0.149 0.192 Hat+Mask DSH [23] 0.0762 0.153 0.218 0.011 0.007 0.012 0.013 DSHSD [24] 0.085 0.139 0.150 0.002 0.004 0.006 − DPSH [25] 0.065 0.082 0.158 0.011 0.012 0.031 0.036 CSQ [27] 0.097 0.151 0.163 - - 0.019 0.022 Hashnet [26] 0.080 0.128 0.164 0.019 0.020 0.032 0.042 GFIH 0.198 0.337 0.433 0.043 0.114 0.198 0.255 The bold font indicate the largest values in the corresponding column Comparison with other inpainting methods For the same occlusion situation and deep hashing retrieval network, the occluded face and face reconstructed by the Context Encoder (CE) [6] and MAT [40] are compared to confirm the effectiveness of the proposed generative model. The proposed generative model architecture is similar to the CE model except the loss functions. Thus, the effectiveness of the joint loss can be evaluated by using the same deep hashing retrieval network. The results of the proposed method compared with CE and MAT inpainting model on two datasets are given in Table 3. The third column of Table 3 reports the mAP results of combination of Hashnet and non-inpainting model, CE inpainting model, MAT inpainting model or GFIH inpainting model (labeled as Hashnet, CE+Hashnet, MAT+Hashnet and GFIH) using different hash bits length for Celeba-1h dataset, while the fourth column of Table 3 is for Celeba-1k dataset. It can be seen from Table 3 that the proposed generative model achieves a better mAP in most of the compared experiments on both two datasets, which demonstrate the good performance of the proposed generative model in GFIH. This proves that the combination of adversarial loss, reconstruction loss and hash bits loss enable the GFIH to perform better than other inpainting model. The possible reason for this superior performance is that the proposed generative model help generate the reconstructed face images whose hash code is closer to the real face image against other inpainting methods, so the deep hashing retrieval network can generate the compact similarity-preserving hashing codes and achieve higher mAP by employing these reconstructed face images.Table 3 mAP comparisons of inpainting methods for CelebA-1H and CelebA-1K dataset Methods CELEBA-1H CELEBA-1K 16bits 32bits 64bits 16bits 32bits 64bits 128bits Hat Hashnet [26] 0.087 0.138 0.210 0.023 0.040 0.071 0.090 CE+Hashnet 0.219 0.334 0.415 0.036 0.093 0.162 0.226 MAT+Hashnet 0.222 0.348 0.409 0.033 0.086 0.159 0.225 GFIH 0.242 0.351 0.431 0.042 0.104 0.182 0.237 Glasses Hashnet [26] 0.195 0.257 0.350 0.027 0.060 0.106 0.149 CE+Hashnet 0.267 0.414 0.519 0.043 0.115 0.184 0.258 MAT+Hashnet 0.239 0.395 0.500 0.039 0.108 0.173 0.252 GFIH 0.309 0.475 0.530 0.046 0.136 0.197 0.270 Mask Hashnet [26] 0.183 0.311 0.435 0.041 0.081 0.141 0.189 CE+Hashnet 0.251 0.442 0.570 0.054 0.155 0.257 0.368 MAT+Hashnet 0.256 0.458 0.523 0.040 0.123 0.251 0.356 GFIH 0.276 0.489 0.579 0.057 0.166 0.285 0.385 Hat+Glasses Hashnet [26] 0.084 0.127 0.162 0.017 0.026 0.046 0.056 CE+Hashnet 0.196 0.267 0.332 0.033 0.063 0.093 0.132 MAT+Hashnet 0.191 0.331 0.402 0.028 0.065 0.107 0.135 GFIH 0.201 0.288 0.345 0.035 0.075 0.117 0.147 Glasses+Mask Hashnet [26] 0.171 0.214 0.281 0.025 0.043 0.065 0.079 CE+Hashnet 0.212 0.315 0.420 0.037 0.098 0.129 0.185 MAT+Hashnet 0.228 0.337 0.471 0.032 0.086 0.135 0.181 GFIH 0.263 0.357 0.436 0.039 0.102 0.149 0.192 Hat+Mask Hashnet [26] 0.080 0.128 0.164 0.019 0.020 0.032 0.042 CE+Hashnet 0.179 0.283 0.383 0.038 0.097 0.172 0.248 MAT+Hashnet 0.213 0.351 0.430 0.030 0.083 0.163 0.232 GFIH 0.198 0.337 0.433 0.043 0.114 0.198 0.255 The bold font indicate the largest values in the corresponding column Comparison with other hashing methods adopted in GFIH Note that there are no limits on the method for learning the original hash codes of the real face images, which means that existing hashing methods can also be employed. Therefore, several deep hashing methods including DSH, DSHSD, DPSH, CSQ and Hashnet are adopted for illustration. The results of the models combining the proposed inpainting model and several hashing methods (labeled as GFIH-DSH, GFIH-DSHSD, GFIH-DPSH, GFIH-CSQ, and GFIH) on two datasets are given in Table 4. It can be seen from Table 4 that the mAP of all combined models remains in a relatively stable and acceptable range under six different occlusion situations, which proves that the proposed framework can achieve a better retrieval performance against occluded image using different hash retrieval models and is general enough to adopt other deep hashing retrieval model to replace the Hashnet for binary hash codes learning. Figure 3 presents the recall-precision curve of the proposed method, existing hashing methods, the combinations of the proposed inpainting model and several hashing methods with 64-bit length hash codes under six different face occlusion situations on CelebA-1H dataset and Fig. 4 presents the recall-precision curve of the above methods with 128-bit length hash codes under six different face occlusion situations on CelebA-1K dataset. Compared to other methods, GFIH is proved to be efficient to balance the recall and precision while achieving higher recall and precision with the same code length under most of the occlusion situations in two datasets. It can be concluded from Table 4 that the proposed method yields the best results against other combined retrieval methods in most occlusion situations of CelebA-1H dataset and yields the best results against other combined retrieval methods in all occlusion situations of CelebA-1K dataset. This proves the effectiveness of the hashing retrieval methods selected in this paper. The results in Table 4 also imply that the other combined method such as the combination of GFIH and DSH, DSHSD or CSQ achieves a much lower retrieval performance than the proposed method under most occlusion situations on CelebA-1H dataset. The possible reason for this may be due to the increase of face categories, the amount of similar image pairs is much smaller than the amount of dissimilar image pairs in CelebA-1K dataset, which brings an imbalance problem between similar and dissimilar pairs in the face dataset. The data imbalance problem makes the similarity-preserving learning ineffective in these hashing methods. Because Hashnet is designed to learn similarity from imbalanced similarity relationships with a weighted pairwise cross-entropy loss function, GFIH can generate exactly binary hash codes and yield best retrieval performance on CelebA-1K dataset. In summary, the proposed occluded face retrieval method achieves a superior performance comparing to other face inpainting models and non-inpainting models for face retrieval under occlusion. The proposed method employs a novel joint loss, which consists of adversarial loss, reconstruction loss and hash bits loss. It encourages the generative model to reconstruct a reconstructed face image, which helps the hashing retrieval network generate the compact similarity-preserving hashing codes.Table 4 mAP comparisons of combination of generator and deep hashing methods for CelebA-1H and CelebA-1K dataset Methods CELEBA-1H CELEBA-1K 16bits 32bits 64bits 16bits 32bits 64bits 128bits Hat GFIH-DSH 0.194 0.356 0.443 0.023 0.031 0.043 0.050 GFIH-DSHSD 0.262 0.373 0.408 0.015 0.025 0.034 − GFIH-DPSH 0.275 0.280 0.306 0.031 0.06 0.101 0.107 GFIH-CSQ 0.160 0.286 0.416 − − 0.041 0.131 GFIH 0.242 0.351 0.431 0.042 0.104 0.182 0.237 Glasses GFIH-DSH 0.229 0.385 0.443 0.031 0.049 0.032 0.081 GFIH-DSHSD 0.302 0.385 0.457 0.014 0.028 0.035 − GFIH-DPSH 0.211 0.349 0.458 0.039 0.072 0.130 0.142 GFIH-CSQ 0.232 0.355 0.529 − − 0.040 0.046 GFIH 0.309 0.475 0.530 0.046 0.136 0.197 0.270 Mask GFIH-DSH 0.267 0.358 0.455 0.038 0.051 0.093 0.132 GFIH-DSHSD 0.371 0.501 0.585 0.023 0.052 0.060 − GFIH-DPSH 0.213 0.361 0.472 0.037 0.069 0.137 0.151 GFIH-CSQ 0.228 0.396 0.551 − − 0.054 0.216 GFIH 0.276 0.489 0.579 0.057 0.166 0.285 0.385 Hat+Glasses GFIH-DSH 0.177 0.283 0.377 0.017 0.018 0.025 0.030 GFIH-DSHSD 0.187 0.243 0.270 0.007 0.011 0.017 − GFIH-DPSH 0.189 0.260 0.316 0.033 0.057 0.105 0.103 GFIH-CSQ 0.152 0.263 0.381 − 0.031 0.038 GFIH 0.201 0.288 0.345 0.035 0.075 0.117 0.147 Glasses+Mask GFIH-DSH 0.210 0.325 0.410 0.019 0.020 0.024 0.029 GFIH-DSHSD 0.245 0.288 0.331 0.011 0.013 0.017 − GFIH-DPSH 0.201 0.315 0.394 0.039 0.070 0.122 0.126 GFIH-CSQ 0.198 0.323 0.455 − − 0.031 0.039 GFIH 0.263 0.357 0.436 0.039 0.102 0.149 0.192 Hat+Mask GFIH-DSH 0.210 0.306 0.431 0.024 0.026 0.041 0.045 GFIH-DSHSD 0.193 0.302 0.321 0.009 0.017 0.022 − GFIH-DPSH 0.168 0.268 0.361 0.033 0.058 0.118 0.118 GFIH-CSQ 0.161 0.303 0.415 − − 0.033 0.090 GFIH 0.198 0.337 0.433 0.043 0.114 0.198 0.255 The bold font indicate the largest values in the corresponding column Fig. 3 The Recall-Precision curve under six different face occlusion situations on Celeba-1H dataset Fig. 4 The Recall-Precision curve under six different face occlusion situations on Celeba-1K dataset Comparison with other methods on MFRD dataset To validate the effectiveness of knowledge distillation under real occluded face situation, another dataset MFRD is conducted to validate the effectiveness of GFIH against real mask face images. Most of occluded face images in MFRD for test are real images in the wild. Experimental results of the proposed GFIH in comparison to other methods on MFRD are given in Table 5. This proves the effectiveness of the proposed KDH for real occluded face retrieval. The MAP comparison in Table 5 shows that face inpainting methods is effective to improves retrieval performance for real occluded faces retrieval. The occlusion reconstructed based face learning methods GFIH-DSH, GFIH-CSQ and GFIH yields superior results against the original DSH, CSQ, Hashnet models. Moreover, our proposed method outperforms other inpaing methods, which proves the effectiveness of the proposed GFIH for real occluded face retrieval. Conclusion In this paper, an occluded face hash retrieval method is proposed for face retrieval under several occlusion situations. The proposed model consists of generator, discriminator, and deep hashing retrieval network. By optimizing the objective function defined over adversarial loss, reconstruction loss and hash bits loss, the learned generator model can generate the face images under occlusion and enhance the retrieval performance of occluded face retrieval by employing the reconstructed face images. Because the dataset of six different face occlusion situations is established by artificially adding masked region, it cannot fully simulate all the possible situations of face occlusion in the natural environment. Moreover, the position of occluded region in the nature is unknown in advance. In future work, the GFIH may be extended to an effective retrieval method for different occluded face situations in the wild, so the GFIH is capable of improving the occluded face retrieval performance in the practical environment.Table 5 mAP comparison on MFRD dataset Methods MFRD 32bits 64bits 128bits DSH [23] 0.035 0.052 0.057 DSHSD [24] 0.008 0.010 0.010 DPSH [25] 0.003 0.003 0.004 CSQ [27] 0.007 0.008 0.015 Hashnet [26] 0.036 0.061 0.083 CE+Hashnet 0.115 0.259 0.351 MAT+Hashnet 0.135 0.226 0.359 GFIH-DSH 0.109 0.228 0.227 GFIH-CSQ 0.023 0.105 0.157 GFIH 0.123 0.276 0.363 The bold font indicate the largest values in the corresponding column Acknowledgements This work was supported in part by the National Natural Science Foundation of China under Grants 62202175, 61876066, 62176160, and 61672443, the 67th Chinese Postdoctoral Science Foundation (2020M672631), the Hong Kong RGC General Research Funds under Grant 9042489 (CityU 11206317), Grant 9042816 (CityU 11209819) and Grant 9042322 (CityU 11200116), Natural Science Foundation of Guangdong Province of China (2022A1515010791), Hong Kong Innovation and Technology Commission (InnoHK Project CIMDA), and Natural Science Foundation of Shenzhen (20200804193857002). Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Zeng D, Veldhuis R, Spreeuwers L (2021) A survey of face recognition techniques under occlusion. IET Biometr. 10(6):581–606 2. Lv J-J Shao X-H Huang J-S Zhou X-D Zhou X Data augmentation for face recognition Neurocomputing 2017 230 184 196 10.1016/j.neucom.2016.12.025 3. Trigueros DS Meng L Hartnett M Enhancing convolutional neural networks for face recognition with occlusion maps and batch triplet loss Image Vis Comput 2018 79 99 108 10.1016/j.imavis.2018.09.011 4. Wan W, Chen J (2017) Occlusion robust face recognition based on mask learning. In: 2017 IEEE international conference on image processing (ICIP). IEEE, pp 3795–3799 5. Song L, Gong D, Li Z, Liu C, Liu W (2019) Occlusion robust face recognition based on mask learning with pairwise differential siamese network. In: Proceedings of the IEEE/CVF international conference on computer vision, pp 773–782 6. Pathak D, Krahenbuhl P, Donahue J, Darrell T, Efros AA (2016) Context encoders: feature learning by inpainting. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 2536–2544 7. Li Y, Liu S, Yang J, Yang M-H (2017) Generative face completion. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 3911–3919 8. Gong Y Lazebnik S Gordo A Perronnin F Iterative quantization: a procrustean approach to learning binary codes for large-scale image retrieval IEEE Trans Pattern Anal Mach Intell 2012 35 12 2916 2929 10.1109/TPAMI.2012.193 9. Li J Ng WW Tian X Kwong S Wang H Weighted multi-deep ranking supervised hashing for efficient image retrieval Int J Mach Learn Cybern 2020 11 4 883 897 10.1007/s13042-019-01026-0 10. Ng WW Tian X Lv Y Yeung DS Pedrycz W Incremental hashing for semantic image retrieval in nonstationary environments IEEE Trans Cybern 2016 47 11 3814 3826 27390201 11. Zhu J Shu Y Zhang J Wang X Wu S Triplet-object loss for large scale deep image retrieval Int J Mach Learn Cybern 2022 13 1 1 9 10.1007/s13042-021-01330-8 12. Heo J-P Lee Y He J Chang S-F Yoon S-E Spherical hashing: binary code embedding with hyperspheres IEEE Trans Pattern Anal Mach Intell 2015 37 11 2304 2316 10.1109/TPAMI.2015.2408363 26440269 13. Ng WW Jiang X Tian X Pelillo M Wang H Kwong S Incremental hashing with sample selection using dominant sets Int J Mach Learn Cybern 2020 11 12 2689 2702 10.1007/s13042-020-01145-z 14. Déniz O Bueno G Salido J De la Torre F Face recognition using histograms of oriented gradients Pattern Recogn Lett 2011 32 12 1598 1603 10.1016/j.patrec.2011.01.004 15. Huang D Shan C Ardabilian M Wang Y Chen L Local binary patterns and its application to facial image analysis: a survey IEEE Trans Syst Man Cybern Part C (Appl Rev) 2011 41 6 765 781 10.1109/TSMCC.2011.2118750 16. Purandare V, Talele K (2014) Efficient heterogeneous face recognition using scale invariant feature transform. In: 2014 International conference on circuits, systems, communication and information technology applications (CSCITA), pp 305–310. IEEE 17. Oliva A Torralba A Modeling the shape of the scene: a holistic representation of the spatial envelope Int J Comput Vision 2001 42 3 145 175 10.1023/A:1011139631724 18. Li Z Liu J Tang J Lu H Robust structured subspace learning for data representation IEEE Trans Pattern Anal Mach Intell 2015 37 10 2085 2098 10.1109/TPAMI.2015.2400461 26353186 19. Sun Y, Chen Y, Wang X, Tang X (2014) Deep learning face representation by joint identification-verification. In: Proceedings of the 27th International Conference on Neural Information Processing Systems vol 2 pp 1988–1996 20. Wang H, Wang Y, Zhou Z, Ji X, Gong D, Zhou J, Li Z, Liu W (2018) Cosface: large margin cosine loss for deep face recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 5265–5274 21. Xia R, Pan Y, Lai H, Liu C, Yan S (2014) Supervised hashing for image retrieval via image representation learning. In: Twenty-eighth AAAI conference on artificial intelligence 22. Erin Liong V, Lu J, Wang G, Moulin P, Zhou J (2015) Deep hashing for compact binary codes learning. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 2475–2483 23. Liu H, Wang R, Shan S, Chen X (2016) Deep supervised hashing for fast image retrieval. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 2064–2072 24. Wu L Ling H Li P Chen J Fang Y Zhou F Deep supervised hashing based on stable distribution IEEE Access 2019 7 36489 36499 10.1109/ACCESS.2019.2900489 25. Li W-J, Wang S, Kang W-C (2015) Feature learning based deep supervised hashing with pairwise labels. arXiv preprint arXiv:1511.03855 26. Cao Z, Long M, Wang J, Yu PS (2017) Hashnet: deep learning to hash by continuation. In: Proceedings of the IEEE international conference on computer vision, pp 5608–5617 27. Yuan L, Wang T, Zhang X, Tay FE, Jie Z, Liu W, Feng J (2020) Central similarity quantization for efficient image and video retrieval. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 3083–3092 28. Tang J Lin J Li Z Yang J Discriminative deep quantization hashing for face image retrieval IEEE Trans Neural Netw Learn Syst 2018 29 12 6154 6162 10.1109/TNNLS.2018.2816743 29994008 29. Tang J Li Z Zhu X Supervised deep hashing for scalable face image retrieval Pattern Recogn 2018 75 25 32 10.1016/j.patcog.2017.03.028 30. Jang YK, Jeong D-j, Lee SH, Cho NI (2018) Deep clustering and block hashing network for face image retrieval. In: Asian conference on computer vision. Springer, pp 325–339 31. Zhou L, Wang Y, Xiao B, Xu Q (2021) Dfh-gan: a deep face hashing with generative adversarial network. In: 2020 25th international conference on pattern recognition (ICPR). IEEE, pp 7012–7019 32. Criminisi A Pérez P Toyama K Region filling and object removal by exemplar-based image inpainting IEEE Trans Image Process 2004 13 9 1200 1212 10.1109/TIP.2004.833105 15449582 33. Xu Z Sun J Image inpainting by patch propagation using patch sparsity IEEE Trans Image Process 2010 19 5 1153 1165 10.1109/TIP.2010.2042098 20129864 34. Yang C, Lu X, Lin Z, Shechtman E, Wang O, Li H (2017) High-resolution image inpainting using multi-scale neural patch synthesis. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 6721–6729 35. Liu G, Reda FA, Shih KJ, Wang T-C, Tao A, Catanzaro B (2018) Image inpainting for irregular holes using partial convolutions. In: Proceedings of the European conference on computer vision (ECCV), pp 85–100 36. Nazeri K, Ng E, Joseph T, Qureshi F, Ebrahimi M (2019) Edgeconnect: structure guided image inpainting using edge prediction. In: Proceedings of the IEEE/CVF international conference on computer vision workshops, pp 0–0 37. Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A, Bengio Y (2014) Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks. arxiv:1511.06434 38. Radford A, Metz L, Chintala S (2015) Unsupervised representation learning with deep convolutional generative adversarial networks. arXiv preprint arXiv:1511.06434 39. Liu Z, Luo P, Wang X, Tang X (2015) Deep learning face attributes in the wild. In: Proceedings of the IEEE international conference on computer vision, pp 3730–3738 40. Li W, Lin Z, Zhou K, Qi L, Wang Y, Jia J (2022) Mat: mask-aware transformer for large hole image inpainting. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition	36474954	PMC9715423	NO-CC CODE	2022-12-03 23:20:15	no	Int J Mach Learn Cybern. 2022 Dec 2;:1-14	utf-8	Int J Mach Learn Cybern	2,022	10.1007/s13042-022-01723-3	oa_other
==== Front iScience iScience iScience 2589-0042 The Author(s). S2589-0042(22)01974-5 10.1016/j.isci.2022.105701 105701 Article Influenza A virus modulates ACE2 expression and SARS-CoV-2 infectivity in human cardiomyocytes Wu Qian 1∗ Kumar Naresh 1∗ Lafuse William P. 1∗ Santiagonunez Ahumada Omar 1 Saljoughian Noushin 1 Whetstone Elizabeth 1 Zani Ashley 1 Patton Ashley K. 2 El Refaey Mona 3 Webb Amy 4 Pietrzak Maciej 4 Yu Lianbo 4 KC Mahesh 56 Peeples Mark E. 56 Ganesan Latha P. 7 Yount Jacob S. 1 Rajaram Murugesan V.S. 18# 1 Department of Microbial Infection and Immunity, Abigail Wexner Research Institute, Nationwide Children’s Hospital, Columbus, OH, USA, 43209 2 Department of Pathology, Abigail Wexner Research Institute, Nationwide Children’s Hospital, Columbus, OH, USA, 43209 3 Department of Surgery, Abigail Wexner Research Institute, Nationwide Children’s Hospital, Columbus, OH, USA, 43209 4 Department of Biomedical Informatics, Abigail Wexner Research Institute, Nationwide Children’s Hospital, Columbus, OH, USA, 43209 5 Department of Pediatrics, Abigail Wexner Research Institute, Nationwide Children’s Hospital, Columbus, OH, USA, 43209 6 Center for Vaccines and Immunity, Abigail Wexner Research Institute, Nationwide Children’s Hospital, Columbus, OH, USA, 43209 7 Department of Internal Medicine. College of Medicine, The Ohio State University, Wexner Medical Center, Columbus, OH, USA, 43210 # Corresponding author Murugesan V.S. Rajaram 460 West 12th Ave Columbus, OH- 43210, USA Phone: 614-366-4660 Fax: 614-292-9616 Email: 8 Lead Contact ∗ Authors Q.W., N.K and W.P.L. contributed equally 2 12 2022 2 12 2022 10570122 10 2021 22 9 2022 29 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. Influenza A virus (IAV) and SARS-CoV-2 virus are both acute respiratory viruses currently circulating in the human population. This study aims to determine the impact of IAV infection on SARS-CoV-2 pathogenesis and cardiomyocyte function. Infection of human bronchial epithelial cells (HBEC), A549 cells, lung fibroblasts (HLF), monocyte derived macrophages (MDMs), cardiac fibroblasts (HCF) and hiPSC-derived cardiomyocytes with IAV enhanced the expression of ACE2, the SARS-CoV-2 receptor. Similarly, IAV infection increased levels of ACE2 in the lungs of mice and humans. Interestingly, we detected heavily glycosylated form of ACE2 in hiPSC-CMs and poorly glycosylated ACE2 in other cell types. Also, prior IAV infection enhances SARS-CoV-2 spike protein binding and viral entry in all cell types. However, efficient SARS-CoV-2 replication was uniquely inhibited in cardiomyocytes. Glycosylation of ACE2 correlated with enzymatic conversion of its substrate Ang II, induction of eNOS and nitric oxide production, may provide a potential mechanism for the restricted SARS-CoV-2 replication in cardiomyocytes. Graphical abstract Key words Influenza A virus cardiomyocytes SARS-CoV-2 ACE2 glycosylation ==== Body pmcConflict of Interest Statement The authors declare that they have no conflict of interest.	36474635	PMC9715453	NO-CC CODE	2022-12-13 23:17:33	no	iScience. 2022 Dec 22; 25(12):105701	utf-8	iScience	2,022	10.1016/j.isci.2022.105701	oa_other
==== Front Inform Med Unlocked Inform Med Unlocked Informatics in Medicine Unlocked 2352-9148 The Authors. Published by Elsevier Ltd. S2352-9148(22)00275-1 10.1016/j.imu.2022.101138 101138 Article Construction of a nomogram for predicting COVID-19 in-hospital mortality: A machine learning analysis Padilha Daniela M.H. a2 Garcia Gabriel R. b2 Liveraro Gianni S.S. b2 Mendes Maria C.S. ac Takahashi Maria E.S. b Lascala Fabiana a Silveira Marina N. a Pozzuto Lara a Carrilho Larissa A.O. a Guerra Lívia D. a Moreira Rafaella C.L. a Branbilla Sandra R. a Dertkigil Sérgio S.J. a Takahashi Jun b1 Carvalheira José B.C. a1∗ a Department of Anesthesiology, Oncology and Radiology, School of Medical Sciences, University of Campinas, Campinas, SP, Brazil b Institute of Physics “Gleb Wataghin”, University of Campinas, Campinas, SP, Brazil c Department of Internal Medicine, School of Medical Sciences, University of Campinas, Campinas, SP, Brazil ∗ Corresponding author. Division of Oncology, Department of Anesthesiology, Oncology and Radiology, School of Medical Sciences, University of Campinas, Rua Vital Brasil, 80, Cidade Universitária, ZIP Code: 13, 083-888, Campinas, SP, Brazil. 1 D.M.H.P., G.R.G. and G.S.S.L. contributed equally to this paper. 2 J.T. and J.B.C.C. were co-advisors in these studies. 2 12 2022 2023 2 12 2022 36 101138101138 4 10 2022 17 11 2022 25 11 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. Background and objectives We aim to verify the use of ML algorithms to predict patient outcome using a relatively small dataset and to create a nomogram to assess in-hospital mortality of patients with COVID-19. Methods A database of 200 COVID-19 patients admitted to the Clinical Hospital of State University of Campinas (UNICAMP) was used in this analysis. Patient features were divided into three categories: clinical, chest abnormalities, and body composition characteristics acquired by computerized tomography. These features were evaluated independently and combined to predict patient outcomes. To minimize performance fluctuations due to low sample number, reduce possible bias related to outliers, and evaluate the uncertainties generated by the small dataset, we developed a shuffling technique, a modified version of the Monte Carlo Cross Validation, creating several subgroups for training the algorithm and complementary testing subgroups. The following ML algorithms were tested: random forest, boosted decision trees, logistic regression, support vector machines, and neural networks. Performance was evaluated by analyzing Receiver operating characteristic (ROC) curves. The importance of each feature in the determination of the outcome predictability was also studied and a nomogram was created based on the most important features selected by the exclusion test. Results Among the different sets of features, clinical variables age, lymphocyte number and weight were the most valuable features for prognosis prediction. However, we observed that skeletal muscle radiodensity and presence of pleural effusion were also important for outcome determination. Integrating these independent predictors was successfully developed to accurately predict mortality in COVID-19 in hospital patients. A nomogram based on these five features was created to predict COVID-19 mortality in hospitalized patients. The area under the ROC curve was 0.86 ± 0.04. Conclusion ML algorithms can be reliable for the prediction of COVID-19-related in-hospital mortality, even when using a relatively small dataset. The success of ML techniques in smaller datasets broadens the applicability of these methods in several problems in the medical area. In addition, feature importance analysis allowed us to determine the most important variables for the prediction tasks resulting in a nomogram with good accuracy and clinical utility in predicting COVID-19 in-hospital mortality. Keywords Skeletal muscle radiodensity Myosteatosis SARS-CoV-2 Prognosis Machine learning Nomogram ==== Body pmcABBREVIATIONS ALT Alanine aminotransferase AST Aspartate aminotransferase AUC Area under the curve BDT Boosted decision trees CRP C-reactive protein HU Hounsfield unit IQR interquartile range LR Logistic regression ML: Machine learning RF Random forest ROC: Receiver operating characteristic RT-PCR Reverse transcriptase-polymerase chain reaction SAT Subcutaneous adipose tissue area SATI subcutaneous adipose tissue index SATR Subcutaneous adipose tissue radiodensity SMA Skeletal muscle area SMD Skeletal muscle radiodensity SMI skeletal muscle index SVM Support vector machines UNICAMP University of Campinas VAT Visceral adipose tissue area VATI visceral adipose tissue index VATR Visceral adipose tissue radiodensity 1 Introduction Clinical and epidemiological characteristics of coronavirus disease 19 (COVID-19) have been investigated since the beginning of the pandemic, and some risk factors have been shown to be associated with a worse clinical course of the disease. There is clear evidence associating comorbidities and inflammatory states with disease severity [1]. Older patients (>60 years); those with certain pre-existing diseases, such as cardiovascular disease, diabetes, chronic respiratory disease or hypertension; immune-compromised people; and people who smoke tobacco or drink alcohol are at greater risk of developing serious disease and dying [2]. Several inflammatory markers have been observed to be elevated in patients with severe COVID-19, including white blood cell count, procalcitonin level, C-reactive protein (CRP), IL-6, IL-10, D-dimer, and lactate dehydrogenase [3,4]. Sarcopenia is characterized by a progressive loss of skeletal muscle mass, strength and physical capacity, and is also associated with increased mortality from COVID-19 [5]. Myoesteatosis, defined as the infiltration of fat into skeletal muscle [6], is considered a negative prognostic factor in many pathologies, including COVID-19 and non-COVID-19 critical patients [7,8]. CT-derived adipose tissue area measurements have been reported as a possible tool to predict worse prognosis in patients with COVID-19 [1,9]. The same has been documented for pulmonary parenchymal damage and associated pathologic features assessed on chest CT images [10]. A visual semi-quantitative analysis of disease extent at CT is correlated with clinical severity [11]. Some authors have developed nomograms to predict poor outcomes in COVID-19 using clinical characteristics, laboratory findings, and lung abnormalities found on CT scan, but none of them have analyzed the impact of body composition along with these other variables [[12], [13], [14], [15], [16]]. Machine learning algorithms [[17], [18], [19]] are used to create data-driven analytical models to identify correlations and patterns in large datasets. The goal of this study is to investigate the possibility of using machine learning algorithms to predict patient outcomes using a relatively small dataset. In addition, we aim to identify which COVID-19 patient diagnostic features (clinical features, body composition and chest abnormalities) are the most useful in the forecast of poor outcome and to create a nomogram that predicts mortality of in-hospital COVID-19 patients to be incorporated in clinical practice. 2 Materials & methods 2.1 Study population In this retrospectively analyzed observational study, we enrolled patients hospitalized at the Clinical Hospital of the State University of Campinas (UNICAMP) between May 1st and July 31st, 2020. The inclusion criteria were: (a) positive SARS-CoV-2 reverse transcriptase-polymerase chain reaction (RT-PCR) assay; (b) availability of computed tomography (CT) scan at the first lumbar vertebra level in the electronic system of the hospital at the diagnosis of COVID-19; and (c) date of hospital discharge or death available from the medical record. The exclusion criteria were: (a) refusal to participate in the study, (b) contrast-enhanced or suboptimal image quality of the CT scan; and (c) lack of clinical variables. This study was approved by our Institutional Review Board (CAAE: 36276620.2.0000.5404), informed consent was obtained, and all procedures were conducted according to the Declaration of Helsinki. 2.2 Data collection Demographic, biochemical and clinical data were collected from medical records. Biochemical data were collected at the closest date of diagnosis. Variables were categorized into three groups [1]: clinical-related, which includes demographic data [age, race (Black, Black + Caucasian, White/Caucasian), weight and height], biochemical data [hematocrit, lymphocytes, neutrophils, monocytes, eosinophils, platelets, prothrombin time, creatinine, sodium, alanine aminotransferase (ALT), aspartate aminotransferase (AST), CRP, glycemia, symptoms at admission (fever, cough, sputum, myalgia, fatigue, diarrhea, nausea or vomiting, dyspnea, anosmia, dysgeusia, headache, hyporexia, odynophagia, nasal symptoms, abdominal pain, other, asymptomatic), comorbidities (hypertension, dyslipidemia, acquired immunodeficiency syndrome (AIDS), emphysema, asthma, chronic kidney disease, congestive heart failure, coronaropathy, stroke, dementia, hypothyroidism, chronic liver disease, autoimmune rheumatic disease, obesity, cancer, diabetes, other, no comorbidities) [2]; body composition-related, including skeletal muscle radiodensity, intramuscular adipose tissue radiodensity, subcutaneous adipose tissue radiodensity, visceral adipose tissue radiodensity, skeletal muscle area, subcutaneous adipose tissue area, and visceral adipose tissue area; and [3] chest abnormalities-related (Bernheim score, presence of pleural effusion, pericardium effusion and mediastinal lymphadenomegaly). 2.3 Body composition assessment Images from CT scans performed at the time of diagnosis of COVID-19 were obtained from the electronic medical image viewer. A trained evaluator (M.C.S.M.), who was blinded to the clinical data, analyzed the single cross-section image nearest to the inferior border of L1 using SliceOMatic V.5.0 software (Tomovision, Canada). The tissue Hounsfield unit (HU) thresholds used were: (a) −29 to 150 HU for skeletal muscle, (b) −150 to −50 HU for visceral adipose tissue, and (c) −190 to −30 HU for subcutaneous adipose tissue [20,21]. Skeletal muscle area (SMA), skeletal muscle index (SMI), skeletal muscle radiodensity (SMD), visceral adipose tissue area (VAT), visceral adipose tissue index (VATI), visceral adipose tissue radiodensity (VATR), subcutaneous adipose tissue area (SAT), subcutaneous adipose tissue index (SATI), and subcutaneous adipose tissue radiodensity (SATR) were obtained. Tissue areas (SMA, VAT and SAT) were measured in centimeters squared (cm2) and tissue indices (SMI, VATI and SATI) were calculated by normalizing tissue area by height and reported in centimeters squared per square meter (cm2/m2). 2.4 Chest abnormalities assessment The pulmonary parenchymal involvement was determined by a single radiologist (S.S.J.D) assessing the same CT scan images used to evaluate body composition. The extent of disease was classified as proposed by Bernheim et al. [22]: 0% (absent, category 0), 1–25% (minimal, category 1), 26–50% (mild, category 2), 51–75% (moderate, category 3), or over 75% (severe, category 4). The presence of mediastinal lymphadenopathy, and pleural and pericardial effusion were also recorded. 2.5 Data analysis The analyzed data sample had 200 patients, with an outcome proportion of one death to every five discharged. The information from each patient was categorized into three sets: clinical, body composition and chest abnormalities. The clinical set included 56 features, while the body composition and chest abnormalities sets included eight and four features, respectively. The complete list of the features is shown in Table 1 , along with the median and interquartile range (IQR) for continuous variables and the number and relative percentage for the categorical variables.Table 1 COVID-19 patient features used in this study. The asterisk indicates the presence of missing data. Table 1Characteristic (variable type) All patients, N = 200 Median [IQR] for continuous No (%) for categorical Age (continuous) 59.1 [49.9,68.6] years Weight* (continuous) 78 [70,90] kg Height (continuous) 167 [160,172] cm Gender (categorical) Female 84 (42%) Male 116 (58%) Time from first symptom to admission* (continuous) 7 [4,9] days Race (categorical) Black 15 (7.5%) White/Caucasian 139 (69.5%) Black + Caucasian 46 (23.0%) Presence of Comorbidities (categorical) Hypertension 115 (57.5%) Dyslipidemia 33 (16.5%) Emphysema 10 (5.0%) Chronic kidney disease 28 (14.0%) Congestive heart failure 14 (7.0%) Coronaropathy 21 (10.5%) Stroke 7 (3.5%) Chronic liver disease 7 (3.5%) Autoimmune rheumatic diseases 5 (2.5%) Cancer 21 (10.5%) Diabetes 66 (33.0%) AIDS 3 (1.5%) Asthma 3 (1.5%) Dementia 3 (1.5%) Hypothyroidism 3 (1.5%) Obesity 41 (20.5%) Other 93 (46.5%) No comorbidities 22 (11.0%) Presence of Symptoms on admission (categorical) Fever 133 (66.5%) Cough 118 (59.0%) Myalgia 68 (34.0%) Fatigue 24 (12.0%) Diarrhea 39 (19.5%) Nausea or vomiting 38 (19.0%) Dyspnea 122 (61.0%) Anosmia 28 (14.0%) Dysgeusia 22 (11.0%) Headache 39 (19.5%) Sputum 13 (6.5%) Hyporexia 30 (15.0%) Odynophagia 22 (11.0%) Nasal symptoms 32 (16.0%) Abdominal pain 14 (7.0%) Other symptoms 41 (20.5%) Asymptomatic 19 (9.5%) Laboratory findings (continuous) Lymphocytes 1010 [640, 1390] cells/μL Neutrophils 5605 [3710, 8708] cells/μL Eosinophils 0 [0, 20] cells/μL Monocytes 480 [340, 703] cells/μL Platelets 205500 [150000, 262000] cells/μL Hematocrit 39.5 [34.4, 44.4] % Prothrombin time* 12.3 [11.7, 13.2] s Creatinine* 0.96 [0.76, 1.34] mg/dL Sodium* 136 [133, 138] mEq/L C-reactive protein 83.9 [44.7, 146.0] mg/L Glycemia* 122 [103, 179] mg/dL Aspartate aminotransferase* 33 [23, 47] U/L Alanine aminotransferase* 25 [16, 42] U/L Body composition variables (continuous) Skeletal muscle area 119.9 [100.3, 139.6] cm2 Skeletal muscle attenuation 35.72 [29.37, 42.15] HU Visceral adipose tissue area 151.4 [93.0, 196.0] cm2 Visceral adipose tissue attenuation −93.9 [-98.2, −88.0] HU Subcutaneous adipose tissue area* 130.0 [81.8, 216.2] cm2 Subcutaneous adipose tissue attenuation −89.6 [-95.1, −82.0] HU Intramuscular adipose tissue area 10.1 [5.9, 17.2] cm2 Intramuscular adipose tissue attenuation* −60.2 [-64.9, −56.5] HU Chest abnormality variables Bernheim score (continuous) 0 (3.5%); 1 (52%); 2 (24.5%); 3 (16%); 4 (4%) Presence of pleural effusion (categorical) 51 (25.5%) Presence of pericardium effusion (categorical) 22 (11.0%) Presence of mediastinal node enlargement (categorical) 29 (14.5%) All features had less than 10% missing data. For continuous features, the median calculated from the other patients was considered where data was missing. For categorical features, the most frequent category in the other patients was considered. The following machine learning (ML) techniques were used to analyze the data: random forest (RF) [23,24], boosted decision trees (BDT) [25,26], logistic regression (LR) [27,28] and support vector machines (SVM) [29]. Classical ML methods such as tree-based models like the ones we tested are known to have better or equivalent performance compared to Deep Learning techniques when dealing with tabular data [30]. But, for completeness, we also tested artificial neural networks models with different multilayer perceptron (MLP) architectures, varying from a few neurons in only one inner layer to even deep learning structures with several neurons. In our tests, these models failed to converge due to the low volume of input data. This was expected since Deep Learning techniques have a larger number of free parameters to be adjusted and thus usually require a much larger amount of data for the training process and to avoid overfitting. The ML algorithms were trained to solve a classification problem. This was performed using a subgroup of the patients (the training group) for which the information on the patients’ outcome was used to train the algorithm, and then the classification sensitivity (also known as recall) and specificity were calculated using the remaining patients (the test group), for which the expected outcomes calculated by the algorithm were compared to the actual patient outcome. From the dataset of 200 patients, we assigned 125 patients to the training group and 75 to the test group, with no overlap between the two groups. The relative proportion between the two possible outcomes seen in the total sample was kept constant in the two subgroups. The classification sensitivity and specificity obtained from the ML algorithm applied to the test group depends on the outcome probability threshold and is characterized by a ROC curve. To reduce the fluctuations caused by the limited statistics and the risk of bias due to outliers, the procedure described above was repeated 300 times, shuffling the patients, and creating different training and test subgroups, always with 125 and 75 patients and keeping the outcome proportion constant. We have 200 total patients and for the ML analysis they need to be split into a test set and a training set. There are many ways of selecting 75 individuals out of 200 and depending on the selection we observed variations on the performance of the ML models. So, in the context of this work “shuffling” means a random selection of 75 patients (with the constraint of the outcome proportion) to form a test set. The complementary 125 patients form the training set and then we can repeat the analysis and evaluate the average and spread of the obtained performance of each model. The choice of 300 repetitions was somehow arbitrary, however it is large enough to be statistically stable and small enough to be computationally possible. This technique ends up being a particular case of the Monte Carlo Cross Validation method (MCCV) [31] that is used to reduce the chance of overfitting in a given model. Another option commonly used for the same purpose is the k-fold Cross Validation method in which one splits the dataset in k independent folders, train in k-1, test in the remaining set and repeat k times, so every folder is tested exactly once. The smaller possible folder that could have been created would be with six patients (five discharged and one death) with k = 33. The MCCV allowed us to evaluate 300 test sets of size 75. It is 10 times more training and a 10 times bigger test set. Each of the 300 analyses performed in this way was completely independent of each other and was performed to calculate the spread of the ROC curve and used to estimate the uncertainties associated with the limited statistics. To evaluate the importance of the different features to the final class discrimination (ML outcome), we used two different common procedures, feature permutation and feature exclusion tests [32,33]. In both methods, the ML procedure described above was repeated, but altering the data regarding one specific feature. In the feature permutation test, the data of the specific feature analyzed was shuffled between different patients and the performance of the ML algorithm was analyzed. If the ML performance decreased, it was concluded that the specific feature was indeed important for the outcome performance. The exclusion test was performed by calculating the ML algorithm performance without the specific feature. Since we obtained the highest prediction performance with the RF model, feature importance analyses were performed only with the RF model. Finally, we created a nomogram based on the most important features selected by the exclusion test. The nomogram was built with the mean coefficients extracted from the LR model trained and tested over 300 subgroups. We estimated a score interval for each feature using its specific values and the corresponding mean coefficient. The score intervals were re-scaled to 0–10 range. The total score scale was defined as the sum of individual features scores. From this, we calculated the probability of death with a logistic function. 3 Results Between May 1st, 2020, and July 31st, 2020, 421 patients were admitted to UNICAMP with COVID-19 confirmed by laboratory test. We included 200 patients in the analysis, after exclusion of patients without L1 level on CT scan, who refused to participate, or who did not have the clinical variables of interest available in their medical records. In total, 45 patients (22.5%) died during this period (Supplementary Fig. 1). Table 1 summarizes the clinical and physical characteristics of the patients. These patients were predominantly male (58.0%) and white/Caucasian (69.5%), with a median age of 59.1 years. The most frequent comorbidity was hypertension (57.5%), followed by diabetes (33.0%) and dyslipidemia (16.5%). At hospital admission, the most frequent symptoms were fever (66.5%), dyspnea (61.0%) and cough (59.0%), and 9.5% of patients were asymptomatic. Mean skeletal muscle attenuation was 34.96 HU and 25.5% of patients had pleural effusion. The ROC curves showing the variation of sensitivity and specificity with the outcome probability threshold for the different ML techniques are shown in Fig. 1 . Each curve corresponds to the mean ROC curve of the 300 analyses performed, shuffling the patients in the training and testing subgroups. For each curve, the point for 50% of threshold (squares) and the threshold that yields the maximum Youden's index value (circles) are also indicated. The results in Fig. 1 were obtained using the set of clinical features. Comparison of the performance of the different methods shows that RF yields the best predictive result, with a true positive rate (sensitivity) of around 87% and a false positive rate (1 - specificity) of 36% when applying the Youden's index threshold. BDT also yields reasonable results. LR and SVM do not perform well, showing very low predictive capability, similar to a random classifier. The right plot of Fig. 1 shows the dispersion of the RF algorithm ROC curve for the 300 different analyses, shuffling the patients. From this, we estimated the uncertainties and for this case, the AUC of the average ROC curve is 0.84 ± 0.04. The AUC values obtained from the other techniques are also shown in the inset of the plot on the left.Fig. 1 Both figures show the performance of the ML models though the ROC curve, that corresponds to the true positive rate (sensitivity) versus false positive rate (1-specificity) for the different threshold values of the model. The left panel shows the mean ROC curves for different ML techniques: random forest, BDT, logistic regression and SVM, trained with clinical variables alone. The straight purple line corresponds to the ROC curve of a random classifier, (i.e., with a 50% true or false rate). Symbols on the ROC curves correspond to a chosen threshold value of 0.5 (squares) or the threshold value that maximizes the Youden's Index (circles). The right panel shows the mean ROC curve obtained using the random forest technique. The shaded area shows the uncertainty estimated considering the 300 shuffled samples. The area of the average ROC curve is 0.84 ± 0.04. Abbreviations: BDT: boosted decision tree; ML: machine learning; ROC: receiver operating characteristic; SVM: support vector machine. . (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Fig. 1 We evaluated the impact of body composition features and chest abnormalities using the model that performed best with the clinical variables alone (RF). The inclusion of these features in the process of training did not improve the predictive capability of the model, as shown in Fig. 2 , with all of the different configurations yielding similar ROC curves and AUC.Fig. 2 ROC curves of the random forest model for different configurations of features. The model using only clinical features in shown in blue, using clinical and body composition features in red, clinical and chest abnormality features in dark red, and clinical, body composition and chest abnormality features in magenta. All cases show an AUC of 84%. Circular symbols on the ROC curves indicate the maximum Youden's indexes for each curve, all of which return a roughly 80% true positive rate, 30% false positive rate, 41% precision and 71% accuracy. Abbreviations: AUC: area under the curve; ROC: receiver operating characteristic. . (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Fig. 2 Although the predictive capability did not improve with the inclusion of the body composition features and chest abnormalities, the analysis of ML models allowed us to evaluate the relative importance of each variable and how it compares among the different sets through feature permutation and exclusion tests. Fig. 3 presents the results of the feature importance methods for each configuration, showing the 10 most important features in each case. Blue bars show the performance (AUC) drop when the corresponding feature is permuted. Orange bars show the performance (AUC) drop when the corresponding feature is excluded. The complete feature importance results are presented in Supplementary Tables 1 and 2 Fig. 3 Performance drop of the random forest's AUC using the permutation (blue bars) and exclusion (orange bars) methods for the 10 most important features. The analyses were performed for all different feature configurations: only clinical features (upper left panel), clinical and body composition features (upper right panel), clinical and chest abnormality features (lower left panel), and clinical, body composition and chest abnormality features (lower right panel). The black bars represent one standard deviation. In all four configurations, it is clear that age is the most important feature, followed by the lymphocytes since they yield the highest performance drops. Abbreviations: AUC: area under the curve. . (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Fig. 3 The results of both feature importance methods agree that age and lymphocyte number are the two most important variables for the prediction of outcome in all cases. When we combined clinical and body composition features, we observed that skeletal muscle radiodensity is the third most important feature for the prediction of outcome using this ML model. This result shows that skeletal muscle radiodensity is more relevant for the learning process than every other clinical feature, except for age and lymphocytes. When we combine clinical features and chest abnormalities, we observed that the presence of pleural effusion is the fourth or fifth most important variable, as determined by exclusion or permutation tests, respectively. In this case, age, lymphocyte number and weight were most significant clinical features according to both tests. When we combine clinical and body composition features, and chest abnormalities, the five most important features were (in decreasing order of importance) age, lymphocyte number, skeletal muscle radiodensity, weight and pleural effusion, according to feature exclusion test. The feature importance result also shows that, in each trained model, there are features with a negative contribution, i.e., it would be better for the training processes if such features were not present. The most surprising result is that the feature exclusion test applied to the complete dataset returns only five features with positive contribution: three clinical features (age, lymphocyte count and weight), one body composition feature (skeletal muscle attenuation) and one chest abnormalities feature (pleural effusion). To evaluate how the reduction of input variables affects the predictive capability of the models, we have rerun the RF training and testing procedures using only the five most important features indicated by each of the two feature importance methods. Fig. 4 presents the ROC curves and AUC for each configuration using only the five most important features, indicated by the permutation test (left plot) and exclusion test (right plot). Within uncertainties, the AUC values of ROC curves obtained using only five features are equivalent to the curves obtained when using all features (AUC: 0.86 ± 0.04).Fig. 4 Mean ROC curves of the random forest method for different feature configurations after permutation (left panel) and exclusion (right panel) tests. Model results using only clinical features are presented in blue, clinical and body composition features in red, clinical and chest abnormality features in dark red, and clinical, body composition and chest abnormality features in magenta. Circular symbols on the ROC curves indicate the maximum Youden's indexes for each curve, all of them returning a roughly 85% true positive rate, 30% false positive rate, 42% precision and 72% accuracy. Abbreviations: ROC: receiver operating characteristic. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Fig. 4 We created a nomogram to predict in-hospital mortality using these five most important features (Fig. 5 ). In the nomogram, the total score ranged from 11.2 to 21.8, with corresponding mortality risks of 0.001–0.999. Fig. 6 demonstrates the application of the nomogram to predict risk of death in two subjects.Fig. 5 Nomogram model to predict mortality of in-hospital COVID-19 patients. The nomogram was constructed based on 5 independent risk factors. Fig. 5 Fig. 6 Prediction of death using the proposed nomogram. The figure shows two CT images of white men at L1 level post processing using SliceOMatic® with segmented tissues: blue = subcutaneous fat; red = skeletal muscle (SM); yellow = visceral fat; green = intra skeletal muscle fat. (A) 68 y/o; 92 Kg; lymphocyte count 540/μL; SM radiodensity = −29 HU; no pleural infusion on CT. The nomogram score is 18, showing a death probability of 90%. Outcome: death. (B) 67 y/o; 68 Kg; lymphocyte count 1380/μL; SM radiodensity = 34.63 HU; no pleural infusion on CT. The nomogram score is 14.4, showing a death probability of 10%. Outcome: discharge. Abbreviations CT: computerized tomography; HU: Hounsfield Unit; L1: first lumbar vertebra; y/o: years old. . (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Fig. 6 4 Discussion We successfully applied ML algorithms to predict COVID-19 patient outcome based on different sets of features that included clinical, body composition and chest abnormality information. In order to obtain stable and reliable results, we applied a shuffling technique to overcome the low statistics in the input data. We tested different ML techniques and obtained the best performance with the RF algorithm, which yielded an AUC of 86 ± 4%. The inclusion of body composition features or/and chest abnormalities did not lead to substantial improvement (results agree within uncertainties) in the final AUC. Naively, one could say that these features are not important as their addition did not improve the predictive capability (AUC) of the ML; however, looking at the feature importance results, we can see that the ML achieved a similar AUC in completely different ways. Skeletal muscle radiodensity (a body composition feature) and pleural effusion (a chest abnormality feature) are more important than most clinical features according to the applied feature importance tests. Schaffino et al. [10] assessed the impact of clinical variables, muscle status and chest CT features in patients with COVID-19 by evaluating a combination of these features into four models, finding a similar result. According to ROC analysis, the CT-based model (which included muscle status and chest CT features), and the model that included muscle status, chest CT features and clinical variables, were not different in terms of predicting death (AUC, 0.86 vs AUC, 0.87, respectively; P = 0.28). However, the discrimination performance of the model with only clinical variables and the discrimination performance of the model with only muscle status are both significantly inferior to those of the model using muscle status and chest CT features, and the model that included muscle status, chest CT features and clinical variables (all AUC comparisons: P < 0.001). Using permutation and exclusion tests, we evaluated the relative importance of each feature, considering the different datasets and obtained confirmation, finding that the expected, known features are indeed important, such as age, lymphocyte count, skeletal muscle attenuation and pleural effusion. Other authors have described demographic factors, laboratory findings and radiological characteristics as prognostic factors for mortality and/or severity of COVID-19 [34]. In our present study, we confirmed the importance of age, lymphocyte count and lung impairment in predicting poorer outcomes. Another notable point can be observed when evaluating the scores of feature importance obtained using the exclusion test. When using only clinical variables to train the RF algorithm, the exclusion test yielded only five variables with positive mean scores, indicating that the other variables were not truly contributing to the decision of the algorithm. The inclusion of the body composition variables in the ML training increased the positive score features of the clinical set to 56, hence all clinical variables yielded a positive score. The same effect was not achieved with the inclusion of the chest abnormality variables. There are some scores reported in the literature that predict the risk of death from COVID-19 by analyzing patient characteristics on hospital admission [[35], [36], [37]]. None of these included skeletal muscle radiodensity as a feature to be evaluated. In fact, Padilha et al. and other authors have previously described the association between low skeletal muscle radiodensity and poor outcomes in patients with COVID-19 [8,[38], [39], [40], [41]]. In line with these studies, our study endorses the use of skeletal muscle radiodensity as an important feature to be assessed on hospital admission and a possible emerging biomarker of systemic inflammation in patients with COVID-19. The feature importance test also allowed the reduction of the number of features used in the ML models and obtained a final predictive performance equivalent to when using all variables. In our case, with only five features, we were able to obtain an AUC of 86 ± 4%. Furthermore, the five features used are variables with little risk of misinformation: three basic pieces of clinical information (age, weight and lymphocyte count), the skeletal muscle attenuation from tomography, and one chest abnormality feature. Variables such as symptoms at admission and patients’ preexisting comorbidities were not needed to obtain equivalent results. The reduction of features needed for the model also improves the possibility of transferring or applying our model to other datasets that may not be as complete as ours. Altogether, these results demonstrate the importance of integrating body composition assessed by skeletal muscle radiodensity along with lung impairment, age and lymphocyte count as predictive factors for COVID-19 outcomes. We created a nomogram based on the five most important features (age, lymphocyte number, weight, skeletal muscle radiodensity, and presence of pleural effusion) to predict COVID-19 in-hospital mortality. The higher the score, the higher the mortality risk. Some studies also designed a nomogram to predict poor outcome in COVID-19, but none of them included body composition and chest abnormalities, along with clinical features [[12], [13], [14], [15], [16]]. This study has some limitations. First, our data was retrospectively collected from a single hospital. Second, we did not assess the changes in body composition over time. Third, data on dietary intake, nutrition care support, socioeconomic status and physical activity, which may have affected body composition status and outcome, were not evaluated. Therefore, the predictions of the nomogram should be assessed for external validation in different cohorts. 5 Conclusions In our cohort of patients with a COVID-19 diagnosis, the most important features for predicting poor outcome were age, lymphocyte count, weight, low skeletal muscle radiodensity and presence of pleural effusion. To our knowledge, this is the first study using ML algorithms in a small dataset to investigate the impact of clinical characteristics, body composition and chest abnormalities integrated in a nomogram with good accuracy and clinical practicability in predicting COVID-19 in-hospital mortality. Furthermore, we developed a method to reliably obtain an ML algorithm capable of providing stable predictive performance, with a much-reduced number of input variables, indicating that this method can also be applied to other studies, or even be expanded to other datasets. Through the use of a modified version of the Monte Carlo Cross Validation technique that we named shuffling, we obtained ROC curves with uncertainties and a high value of predictability of the model. The success of ML techniques in smaller datasets broadens the applicability of these methods in several problems in the medical area. In addition, feature importance analysis allowed us to determine the most important variables for the prediction tasks, which indeed agrees with the already known risk factors of COVID-19. Funding This study was supported by the 10.13039/501100001807 São Paulo Research Foundation (FAPESP) , Grant/Award Number: 2018/23428‐0 and 2017/05685-2 and 10.13039/501100003593 National Council for Scientific and Technological Development (CNPq) , Grant/Award Number: 303429/2021-6. Author contributions D.M.H.P., G.R.G., G.S.S.L., M.C.S.M., M.E.S.T., J.T., and J.B.C.C. designed the research. D.M.H.P., G.R.G., G.S.S.L., M.C.S.M., M.E.S.T., S.S.J.D. J.T., and J.B.C.C. analyzed the data. D.M.H.P., M.C.S.M., F.L.J., M.N.S., L.P., L.A.O.S., L.D.G., R.C.L.M., and S.R.B collected the data. D.M.H.P., G.R.G., G.S.S.L., M.C.S.M., J.T., and J.B.C.C. wrote the paper. D.M.H.P., G.R.G., G.S.S.L., M.C.S.M., J.T., and J.B.C.C. had primary responsibility for the final content. All authors read and approved the final manuscript. Data availability statement The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding author. 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 are the Supplementary data to this article:Multimedia component 1 Multimedia component 1 Multimedia component 2 Multimedia component 2 Acknowledgement We thank Proof-Reading-Service.com for proof Reading our manuscript. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.imu.2022.101138. ==== Refs References 1 Brodin P. Immune determinants of COVID-19 disease presentation and severity Nat Med 27 1 2021 28 33 33442016 2 Rosenthal N. Cao Z. Gundrum J. Sianis J. Safo S. Risk factors associated with in-hospital mortality in a US National sample of patients with COVID-19 JAMA Netw Open 3 12 2020 e2029058 3 Ji P. Zhu J. Zhong Z. Li H. Pang J. Li B. Association of elevated inflammatory markers and severe COVID-19: a meta-analysis Medicine (Baltim) 99 47 2020 e23315 4 Ou M. Zhu J. Ji P. Li H. Zhong Z. Li B. Risk factors of severe cases with COVID-19: a meta-analysis Epidemiol Infect 148 2020 e175 32782035 5 Kim J.-W. Yoon J.S. Kim E.J. Hong H.-L. Kwon H.H. Jung C.Y. Prognostic implication of baseline sarcopenia for length of hospital stay and survival in patients with coronavirus disease 2019 J Gerontol A Biol Sci Med Sci 76 8 2021 e110 e116 33780535 6 Souza N.C. Gonzalez M.C. Martucci R.B. Rodrigues V.D. de Pinho N.B. Ponce de Leon A. Frailty is associated with myosteatosis in obese patients with colorectal cancer Clin Nutr 39 2 2020 484 491 30833213 7 Loosen S.H. Schulze-Hagen M. Püngel T. Bündgens L. Wirtz T. Kather J.N. Skeletal muscle composition predicts outcome in critically ill patients Crit Care Explor 2 8 2020 e0171 8 Yi X. Liu H. Zhu L. Wang D. Xie F. Shi L. Myosteatosis predicting risk of transition to severe COVID-19 infection Clin Nutr 41 12 2021 3007 3015 10.1016/j.clnu.2021.05.031 34147286 9 McGovern J. Dolan R. Richards C. Laird B.J. McMillan D.C. Maguire D. Relation between body composition, systemic inflammatory response, and clinical outcomes in patients admitted to an Urban teaching hospital with COVID-19 J Nutr 151 8 2021 2236 2244 34159388 10 Schiaffino S. Albano D. Cozzi A. Messina C. Arioli R. Bnà C. CT-Derived chest muscle metrics for outcome prediction in patients with COVID-19 Radiology 300 2 2021 E328 E336 33724065 11 Yang R. Li X. Liu H. Zhen Y. Zhang X. Xiong Q. Chest CT severity score: an imaging tool for assessing severe COVID-19 Radiology: Cardiothoracic Imag 2 2 2020 e200047 12 Yu Y. Wang X. Li M. Gu L. Xie Z. Gu W. Nomogram to identify severe coronavirus disease 2019 (COVID-19) based on initial clinical and CT characteristics: a multi-center study BMC Med Imag 20 1 2020 111 13 Li J. Wang L. Liu C. Wang Z. Lin Y. Dong X. Exploration of prognostic factors for critical COVID-19 patients using a nomogram model Sci Rep 11 1 2021 8192 33854118 14 Yang Y. Zhu X.F. Huang J. Chen C. Zheng Y. He W. Nomogram for prediction of fatal outcome in patients with severe COVID-19: a multicenter study Mil Med Res 8 1 2021 21 33731184 15 Acar H.C. Can G. Karaali R. Börekçi Ş. Balkan İ.İ. Gemicioğlu B. An easy-to-use nomogram for predicting in-hospital mortality risk in COVID-19: a retrospective cohort study in a university hospital BMC Infect Dis 21 1 2021 148 33546639 16 Pan D. Cheng D. Cao Y. Hu C. Zou F. Yu W. A predicting nomogram for mortality in patients with COVID-19 Front Public Health 8 2020 461 32850612 17 Alpaydin E. Introduction to machine learning 2020 MIT press 18 Bishop C.M. Nasrabadi N.M. Pattern recognition and machine learning 2006 Springer 19 Hastie T. Tibshirani R. Friedman J.H. Friedman J.H. The elements of statistical learning: data mining, inference, and prediction 2009 Springer 20 Heymsfield S.B. McManus C.B. Tissue components of weight loss in cancer patients. A new method of study and preliminary observations Cancer 55 1 Suppl 1985 238 249 3965090 21 Miller K.D. Jones E. Yanovski J.A. Shankar R. Feuerstein I. Falloon J. Visceral abdominal-fat accumulation associated with use of indinavir Lancet 351 9106 1998 871 875 9525365 22 Bernheim A. Mei X. Huang M. Yang Y. Fayad Z.A. Zhang N. Chest CT findings in coronavirus disease-19 (COVID-19): relationship to duration of infection Radiology 2020 200463 23 Breiman L. Random forests Mach Learn 45 1 2001 5 32 24 Yin L. Song C. Cui J. Lin X. Li N. Fan Y. A fusion decision system to identify and grade malnutrition in cancer patients: machine learning reveals feasible workflow from representative real-world data Clin Nutr 40 8 2021 4958 4970 34358843 25 Xgboost: a scalable tree boosting system Chen T. Guestrin C. Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining 2016 26 Raita Y. Camargo C.A. Macias C.G. Mansbach J.M. Piedra P.A. Porter S.C. Machine learning-based prediction of acute severity in infants hospitalized for bronchiolitis: a multicenter prospective study Sci Rep 10 1 2020 1 11 31913322 27 Hosmer D.W. Jr. Lemeshow S. Sturdivant R.X. Applied logistic regression 2013 John Wiley & Sons 28 Tolles J. Meurer W.J. Logistic regression: relating patient characteristics to outcomes JAMA 316 5 2016 533 534 27483067 29 Cristianini N. Shawe-Taylor J. An introduction to support vector machines and other kernel-based learning methods 2000 Cambridge university press 30 Shwartz-Ziv R. Armon A. Tabular data: deep learning is not all you need Inf Fusion 81 2022 84 90 31 Xu Q.-S. Liang Y.-Z. Monte Carlo cross validation Chemometr Intell Lab Syst 56 1 2001 1 11 32 Altmann A. Toloşi L. Sander O. Lengauer T. Permutation importance: a corrected feature importance measure Bioinformatics 26 10 2010 1340 1347 20385727 33 Parr T. Turgutlu K. Csiszar C. Howard J. Beware default random forest importances vol. 26 March. 2018 2018 34 Izcovich A. Ragusa M.A. Tortosa F. Lavena Marzio M.A. Agnoletti C. Bengolea A. Prognostic factors for severity and mortality in patients infected with COVID-19: a systematic review PLoS One 15 11 2020 e0241955 35 Garibaldi B.T. Fiksel J. Muschelli J. Robinson M.L. Rouhizadeh M. Perin J. Patient trajectories among persons hospitalized for COVID-19 : a cohort study Ann Intern Med 174 1 2021 33 41 32960645 36 Liang W. Liang H. Ou L. Chen B. Chen A. Li C. 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 8 2020 1081 1089 32396163 37 San I. Gemcioglu E. Baser S. Yilmaz Cakmak N. Erden A. Izdes S. Brescia-COVID Respiratory Severity Scale (BRCSS) and Quick SOFA (qSOFA) score are most useful in showing severity in COVID-19 patients Sci Rep 11 1 2021 21807 38 Padilha D.M.H. Mendes M.C.S. Lascala F. Silveira M.N. Pozzuto L. Santos L.A.O. Low skeletal muscle radiodensity and neutrophil-to-lymphocyte ratio as predictors of poor outcome in patients with COVID-19 Sci Rep 12 1 2022 15718 39 Rossi A.P. Gottin L. Donadello K. Schweiger V. Brandimarte P. Zamboni G.A. Intermuscular adipose tissue as a risk factor for mortality and muscle injury in critically ill patients affected by COVID-19 Front Physiol 12 2021 651167 40 Viddeleer A.R. Raaphorst J. Min M. Beenen L.F.M. Scheerder M.J. Vlaar A.P.J. Intramuscular adipose tissue at level Th12 is associated with survival in COVID-19 J Cachexia Sarcopenia Muscle 12 3 2021 823 827 33939338 41 Yang Y. Ding L. Zou X. Shen Y. Hu D. Hu X. Visceral adiposity and high intramuscular fat deposition independently predict critical illness in patients with SARS-CoV-2 Obesity 28 11 2020 2040 2048 32677752	36474601	PMC9715454	NO-CC CODE	2022-12-05 23:15:31	no	Inform Med Unlocked. 2023 Dec 2; 36:101138	utf-8	Inform Med Unlocked	2,022	10.1016/j.imu.2022.101138	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)00631-2 10.1016/j.ijid.2022.11.038 Letter to the Editor Drug-drug interaction with oral antivirals for early treatment of COVID-19 – Authors’ reply Larsen Carsten Schade ⁎ Department of Infectious Diseases, Aarhus University Hospital, Aarhus, Denmark ⁎ Corresponding author: Carsten Schade Larsen, Dept. of Infectious Diseases, Aarhus University Hospital, 8200 Aarhus N, Tel +4525601510. 2 12 2022 1 2023 2 12 2022 126 181181 20 11 2022 28 11 2022 28 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. Keywords Statins COVID-19 Atherosclerotic cardiovascular disease Nirmatrelvir/ritonavir ==== Body pmcI thank Vuorio et al. [1] for their very relevant concerns and comments on the Danish population-based study estimating the risk of significant drug-drug-interactions (DDIs) with the oral antiviral nirmatrelvir/ritonavir in the elderly Danish population [1,2]. The study showed that simvastatin or lovastatin was used by 15.45 % of people ≥65 years and 17.70 % of people ≥80 years [2]. Coadministration of simvastatin or lovastatin with ritonavir is contraindicated, as ritonavir increases the concentration of these statins 100-fold with risk of severe toxicity, including rhabdomyolysis [3,4]. In cases where coadministration of nirmatrelvir/ritonavir with a drug is contraindicated, there are three options either pause the drug, replace the drug, or consider another antiviral treatment for early COVID-19 [5]. Obviously, this decision should be based on an individual basis, considering the risks and benefits. Concerning the treatment of patients taking simvastatin or lovastatin with nirmatrelvir/ritonavir, most international guidelines and the Danish national guideline recommend withholding these drugs during and at least 2-3 days after treatment [3,6,7,8]. In general, temporarily pausing statins during early treatment of COVID-19 will, in most cases, not cause any clinical harm but will reduce the risk of toxicities due to DDIs [4,5]. The meta-analysis by Wu et al. [9] shows that statins improve the outcome of COVID-19, including patients with severe COVID-19, with the main outcome being the need for intensive care and death. This is not the target population for treatment with nirmatrelvir/ritonavir, which is used for early treatment of COVID-19. I do not question that statins used for primary prevention reduce the risk of major vascular events, nor that an option is to replace simvastatin or lovastatin with pravastatin or fluvastatin. However, the pragmatic approach is to pause any statins during treatment. Although being an infectious disease specialist, I doubt that withholding statins for 8 days will expose the patients to a significantly increased risk of major cardiovascular events. Declaration of competing interests The authors have no competing interests to declare. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Ethical approval Not applicable. ==== Refs References 1 Vuorio A Raal F Kovanen PT Drug-drug interaction with oral antivirals for early treatment of COVID-19 Int J Infect Dis 2022 10.1016/j.ijid.2022.11.039 2 Larsen CS. Assessing the proportion of the Danish population at risk of clinically significant drug-drug interactions with new oral antivirals for early treatment of COVID-19 Int J Infect Dis 122 2022 599 601 10.1016/j.ijid.2022.06.059 35803465 3 Liverpool COVID-19 Interactions. COVID-19 Drug Interactions, https://www.covid19-druginteractions.org/checker; 2022 [accessed 20 November 2022]. 4 Marzolini C Kuritzkes DR Marra F Boyle A Gibbons S Flexner C Prescribing Nirmatrelvir-Ritonavir: how to recognize and manage drug-drug interactions Ann Intern Med 175 2022 744 746 10.7326/M22-0281 35226530 5 Marzolini C Kuritzkes DR Marra F Boyle A Gibbons S Flexner C Recommendations for the management of drug-drug interactions between the COVID-19 antiviral Nirmatrelvir/Ritonavir (Paxlovid) and comedications Clin Pharmacol Ther 112 2022 1191 1200 10.1002/cpt.2646 35567754 6 National Institute of Health COVID-19 Treatment Guidelines 2022 https://www.covid19treatmentguidelines.nih.gov/ [accessed 20 November 2022] 7 Danish Health Authority. Midlertidig retningslinje for visitation og behandling med Paxlovid (nir-matrelvir/ritonavir), https://www.sst.dk/-/media/Udgivelser/2022/Corona/Paxlovid/Paxlovid_midlertidigretningslinje.ashx?sc_lang=da&hash=1DF396365FE15FDD50A33E2603EBF0C0; 2022 [accessed 20 November 2022]. 8 University of Michigan, Michigan Medicine Management of Paxlovid Drug-Drug-Interactions 2022 https://www.med.umich.edu/asp/pdf/outpatient_guidelines/Paxlovid-DDI.pdf [accessed 20 November 2022] 9 Wu KS Lin PC Chen YS Pan TC Tang PL. The use of statins was associated with reduced COVID-19 mortality: a systematic review and meta-analysis Ann Med 53 2021 874 884 10.1080/07853890.2021.1933165 34096808	36470503	PMC9715456	NO-CC CODE	2022-12-15 23:17:45	no	Int J Infect Dis. 2023 Jan 2; 126:181	utf-8	Int J Infect Dis	2,022	10.1016/j.ijid.2022.11.038	oa_other
==== Front Res Social Adm Pharm Res Social Adm Pharm Research in Social & Administrative Pharmacy 1551-7411 1934-8150 Published by Elsevier Inc. S1551-7411(22)00415-6 10.1016/j.sapharm.2022.11.013 Article Technology targeting immunocompromised patients for COVID-19 vaccine in community pharmacies LaFleur Grace E. ab1 Azzi Andrew G. b Schimmel Scott M. b∗ Howard Mitchell S. a a The University of Toledo College of Pharmacy and Pharmaceutical Sciences, 3000 Arlington Ave. Mail Stop 1013, Toledo, OH, 43614, USA b The Kroger Co., 2257 N Holland Sylvania Rd, Toledo, OH, 43615, USA ∗ Corresponding author. University of Toledo College of Pharmacy and Pharmaceutical Sciences, 3000 Arlington Ave., MS 1013, Toledo, OH, 43614, USA. 1 Present Addresses and Affiliations: Clinical Assistant Professor, University of Utah College of Pharmacy, Department of Pharmacotherapy, 30 S 2000 E Room 4963, Salt Lake City, UT 84112; Staff Pharmacist, Smith's Pharmacy of The Kroger Co., 455 S 500 E, Salt Lake City, UT 84102. 2 12 2022 2 12 2022 19 10 2022 29 11 2022 30 11 2022 © 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. Background Medication targeting by community pharmacists may assess medical history of patients for recommendation of clinical services through review of their prescription history. Previous studies have implemented medication targeting to identify patients eligible for vaccine recommendations. Targeting of immunosuppressing medications may impact the rate of third primary doses of COVID-19 vaccine administered to immunocompromised patients. Objectives The primary objective was to determine the impact of medication targeting on the rate of third primary doses of COVID-19 vaccine given to immunocompromised patients. Methods This observational, retrospective cohort study occurred within one division of a large community pharmacy chain. Included patients were greater than 18 years of age with record of at least one immunosuppressing medication dispensed one year prior to study enrollment and 2 primary COVID-19 vaccine doses in the pharmacy dispensing system. An intervention for pharmacist recommendation of a third primary dose of COVID-19 vaccine was automatically loaded into their prescription profiles. The proportion of patients with completed interventions and confirmation of third dose administration was collected with demographic characteristics. Results The pharmacy dispensing system identified 1670 interventions through medication targeting, though 69 interventions met criteria for study inclusion. Baseline characteristics of the included population were a mean age of 51.8 years of primarily female sex (69.6%) and Caucasian race (78.3%). Third primary COVID-19 vaccine dose administration and completed pharmacist recommendation was recorded for 2 (2.9%) patients. Conclusion Medication targeting identified immunocompromised patients for the recommendation of a third primary dose of COVID-19 vaccine. Improved specification for targeting of dosing regimen and route of administration may result in greater accuracy of appropriate recommendations identified. Keywords Medication targeting Community Immunization Technology ==== Body pmc1 Background The rapid transmission and growing severity of coronavirus disease 2019 (COVID-19) involving infections of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) resulted in the declaration of global pandemic by the World Health Organization in March 2020. Development and release of newly developed mRNA vaccines in December 2020 through emergency use authorizations (EUAs) from the Food and Drug Administration (FDA) became the prioritized effort to reduce the number of hospitalizations and deaths related to COVID-19 infection. Current recommendations for the schedule of COVID-19 mRNA vaccines to be administered in the general population by the Centers for Disease Control and Prevention (CDC) include a primary 2 dose series, completed by 76.2% of U.S. adults as of May 2022.1 A third primary dose has been included in the recommendation for immunocompromised patients, to be given 28 days following the second dose.2 Providing a third primary dose of COVID-19 vaccine to patients with immunocompromised status strengthens the weakened immunologic response experienced upon a 2 dose series, with reported improvement from 69% to 88% vaccine effectiveness against COVID-19 associated hospitalization.3 , 4 Additional protection is also experienced against developing viral complications of greater likelihood in immunocompromised conditions. The clinical considerations for the administration of COVID-19 mRNA vaccines provide a description of conditions and treatments conferring additional primary dose eligibility including, but not limited to, active cancer treatment, immunosuppressive therapy following solid organ transplantation, receipt of CAR-T or stem cell transplant, moderate or severe primary immunodeficiency, advanced or untreated HIV infection, and the use of various immunosuppressing medication such as high dose corticosteroids, antimetabolites, and biologic immunomodulators.5 While diagnosis of conditions listed by the CDC may not always be available for use within a community pharmacy, pharmacists can perform a review of medication fill histories for immunosuppressive therapies. Targeted interventions may be offered for third primary doses of a COVID-19 mRNA vaccine to appropriate patients as identified by their prescription histories. The process of medication targeting evaluates a patient's medication history to identify eligibility for personalized recommendations according to underlying conditions inferred by their prescription medications. Medication targeting is especially useful in the community pharmacy setting where patient information available to the pharmacist is often limited to the information provided by prescription details. Previous use of medication targeting for vaccine recommendations has been incorporated within community pharmacies, resulting in increased vaccination rates.6, 7, 8, 9 Coley and colleagues increased doses administered for pneumococcal and tetanus-pertussis vaccine administration in the community pharmacy setting through identification of eligible patients by review of prescriptions for immunosuppressing medications and prenatal vitamins respectively.6 This medication targeting algorithm was incorporated into their pharmacy dispensing software to identify patients with vaccine eligibility. Notification of eligibility was provided to pharmacists through printed note upon a prescription receipt for the targeted patient or within a list generated by the pharmacy dispensing software of patients to be contacted through phone for clinical interventions. Additional studies have also increased vaccination rates within community pharmacies through identifying patients with diabetes, asthma, and COPD medications for recommendation of the pneumococcal and influenza vaccines.7 , 9 Strategies employed for identifying eligible patients in these study designs included manual report generation for pharmacy dispensing history of targeted medications, application of patient notes describing vaccine eligibility within the dispensing software, and completion of patient questionnaires detailing prescription and vaccination history. The use of immunosuppressant medication fill histories to offer interventions for the third primary dose of a COVID-19 mRNA vaccine may utilize similar methods to facilitate the appropriate immunity for immunocompromised individuals at community pharmacies. Programming of the medication targeting algorithm for notification of clinical intervention opportunity and documentation of patient response within the pharmacy dispensing system may offer greater integration of clinical services into prescription dispensing. Optimizing use of current technology capacity to identify, deliver, and document personalized clinical interventions through medication targeting may influence the rate of clinical services provided by community pharmacists. 2 Objectives The primary objective for this study was to determine the impact of community pharmacy medication targeting on the rate of third primary doses of COVID-19 mRNA vaccine given to immunocompromised patients. 3 Methods This observational, retrospective cohort study collected data from 113 pharmacies within a large community grocery store pharmacy chain located throughout Ohio, Michigan, and West Virginia. Approval was granted by the Biomedical Institutional Review Board at the University of Toledo. Patients were identified for inclusion by automated computer programming within the pharmacy dispensing software through record of an initial series of 2 primary doses of COVID-19 mRNA vaccine received greater than 28 days prior to enrollment. Additional inclusion criteria included age 18 years and greater with record of at least one immunosuppressing medication dispensed in the pharmacy dispensing system within one year prior to study enrollment. The selection of inclusion medications was determined by classes of medications previously categorized as immunosuppressing within the pharmacy dispensing system by individuals employed by the pharmacy chain (Appendix A). Programming of immunosuppressing medications followed medication classification by First Databank (San Bruno, CA) and reference of immunocompromising conditions described by the CDC. Study investigators manually reviewed the medication fill histories of initially identified patients for exclusion by confirmation of non-immunosuppressing medication regimens (Fig. 1 ). Targeting of corticosteroid prescriptions was screened for appropriate inclusion of immunosuppressing dosing regimens of prednisone total daily dose of 20 mg or greater for at least 14 days. Exclusion was also applied for patients less than 18 years old or with record of a third primary COVID-19 mRNA vaccine dose recorded in the pharmacy dispensing system.Fig. 1 Screening for inclusion. Fig. 1 The information technology team at the community grocery store pharmacy chain created a computer-generated intervention for pharmacist recommendation of a third primary dose of COVID-19 mRNA vaccine. This was automatically loaded into the prescription profiles of included patients meeting inclusion criteria stated above. Programmed description of the generated intervention available to the pharmacist included a title of COVID-19 vaccine eligibility with supporting details of the date that the primary 2 dose series of COVID-19 vaccine was completed along with description of eligibility for an additional dose due to recorded dispensing of an included immunosuppressing medication. The intervention was identified by pharmacy personnel during the dispensing workflow through computerized notification when accessing the prescription profile of applicable patients. During the prescription dispensing workflow, a patient counseling note could be electronically applied by pharmacists to require the completion of a conversation with the patient at prescription pick-up prior to the release of the prescription. Without use of the counseling note, the intervention was not required to be addressed with the patient prior to release of their prepared medications. Following recommendation delivery by pharmacist or pharmacy intern to the patient or individual picking up the prescription, the result was documented into the pharmacy dispensing software. Patient responses were documented through selection of predetermined answers including the following choices: yes, unsure, not today – unable right now, not today – patient wants to talk to provider, no – previously received vaccine, no – does not want this vaccine, no – uninterested in all vaccines, patient is deceased, or patient is in hospice or long-term care. Positive identification (ID) by means of ID barcode scanning was required of the pharmacist or pharmacy intern who delivered the recommendation and completed the intervention documentation to process completion of the intervention in the pharmacy dispensing software. Documented responses of uncertainty, including “unsure” or “not today”, were programmed to remain as active interventions in the prescription profile of included patients without targeted pharmacist notification upon each appearance in dispensing workflow. Data for the number of completed recommendations and recorded results was collected retrospectively from a report of interventions generated between September 30, 2021 and December 31, 2021. Information retrieved from the pharmacy dispensing system included patient demographic information (age, sex, race, ethnicity, and insurance type), intervention details (immunosuppressing medication dispensed, method of intervention, and recorded result) and date(s) of recorded COVID-19 vaccine doses. The state immunization registry was also retrospectively reviewed for the immunization histories of included patients to record a confirmation of a third primary COVID-19 mRNA vaccine dose. The primary outcome assessed was the proportion of immunocompromised patients who received a third primary COVID-19 vaccine dose and completed pharmacist recommendation out of the total number of interventions generated. Descriptive statistics were performed using SPSS Statistics Version 26 (Armonk, NY). Nominal data are expressed as number and percentage while continuous data are expressed as mean and standard deviation. 4 Results A total of 1670 interventions were generated for immunocompromised patients by the pharmacy dispensing system for pharmacist recommendation of a third primary COVID-19 vaccine dose. A total of 69 patients were confirmed to have eligibility for inclusion. The primary reason for exclusion was a corticosteroid dosing regimen beneath the minimum requirements for producing immunocompromised status (Fig. 1). Baseline characteristics (Table 1 ) of the included population were primarily female sex (69.6%) with a mean age of 51.8 years and Caucasian race (78.3%).Table 1 Baseline characteristics. Table 1Baseline Characteristics n = 69 Age, years (mean ± SD) 51.8 ± 17.8 Female Sex, n (%) 48 (69.6) Race, n (%) Caucasian 54 (78.3) African American 9 (13) Asian 1 (1.4) Other 1 (1.4) Not Specified 4 (5.8) Third Party, n (%) Commercial 35 (50.7) Government 34 (49.3) Receipt of a third primary COVID-19 vaccine dose and completed pharmacist recommendation was recorded for 2 (2.9%) patients, with both patient responses documented as refusal due to previous receipt of the vaccine. Documentation of completed pharmacist recommendation was recorded for 14 (20.3%) generated interventions (Fig. 2 ). The most common patient response recorded upon completed intervention was a refusal due to previous receipt of the vaccine, reported by 6 (42.8%) patients.Fig. 2 Results of completed interventions. Fig. 2 5 Discussion Incorporating medication targeting to provide personalized recommendations for a third primary dose of a COVID-19 mRNA vaccine to immunocompromised patients resulted in two patients with confirmation of a third dose administered and completed pharmacist recommendation. While the pharmacy dispensing software automatically identified 1670 original interventions by medication class categorized as immunosuppressing, further manual screening by study investigators revealed 69 interventions appropriately generated for patients meeting third dose eligibility. The primary reason for inappropriate intervention assignment was the programmed selection of patients with a corticosteroid prescription dispensed within one year prior to study enrollment, many of which were excluded due to dosing regimens that do not impart immunocompromised status. When programming functional medication targeting for future interventions, further specifications such as including dosages and duration of therapy may be required for appropriate intervention eligibility. Narrowing specificity of the targeting algorithm is anticipated to eliminate unnecessary targeting of ineligible patients experienced in this study when extrapolating the methods to additional recommendations and patient populations. The integration of medication targeting within the technology capacity allowed by the pharmacy dispensing system enhances targeting strategies employed in previous studies exploring impact on vaccination rates. Coley and colleagues relied on printed messages on prescription receipts or patient lists generated by their pharmacy software to alert pharmacists of patient eligibility for vaccine recommendations based upon medication dispensing history.6 Through implementation of their targeted interventions for one year, their team demonstrated an increase in administration of pneumococcal vaccines by 7% and administration of tetanus-pertussis vaccines by 31%. While this current study resulted in a smaller impact upon administration rate for third primary doses of COVID-19 vaccine, the intervention notification for vaccine recommendation was electronically incorporated within the prescription profiles of eligible patients, with the ability to document patient answers within the pharmacy dispensing software. This also allows for initial and subsequent contact points with the patient regarding the intervention to be captured for appropriate follow-up as needed. With an extended study duration and recruitment of a larger inclusion population, the resulting impact of programmed targeting for clinical services may develop similarly to the increase in vaccine administration experienced in the study performed by Coley and colleagues. Incorporating technology advancements to facilitate medication targeting by the pharmacy dispensing system will not only minimize pharmacist efforts in identifying eligible patients, but also digitally attach clinical interventions to patient profiles. This would allow pharmacy personnel to alert the patient of the clinical intervention allowing the pharmacist to have a conversation to improve their health outcomes. Digital record of patient eligibility and documented recommendation response collects all pertinent intervention details into the dispensing system, minimizing the need for additional printed documents for pharmacy records or patient takeaways. All of the information would be stored in one location allowing access for multiple pharmacists to promote continuity of care. Expanding the use of targeting technology in delivery of clinical services in the community pharmacy will improve pharmacist efficiency to integrate longitudinal patient care alongside the dispensing of individual prescriptions by incorporating personalized opportunities into a comprehensive digital profile for each patient encountered. Upon review of the state immunization registry, administration of a third dose of COVID-19 mRNA vaccine was confirmed for 21 of the patients with appropriately generated interventions. Record of the third doses within the registry was not integrated into the pharmacy dispensing system, preventing the exclusion of these recipients from the study. Multiple discrepancies between intervention results documented and administration of a third primary dose of COVID-19 vaccine were identified through the data collection process. The state immunization registry was not found to include third COVID-19 vaccine dose documentation for 2 patients that responded with agreement to receive the vaccine upon pharmacist recommendation. Additionally, 3 patients with response of previously receiving the vaccine also were not found to have a third dose recorded within the registry. The discrepancies between the documented intervention result coupled with the large number of incomplete interventions may be a result of misunderstanding of vaccine schedule among pharmacy personnel due to consistent changes to recommendations for the additional and booster doses of COVID-19 vaccine occurring at the time of intervention generation. Misunderstanding of the recommended vaccine schedule by patients may have also contributed to discrepancy with the possibility of patients responding to have already received the third primary dose due to a previous completion of the initial 2 dose primary series. Programming of the intervention documentation options also may not have provided a clear option for pharmacist selection for appropriate documentation of the patient's response to their recommendation. Further evaluation of the current process for technology-driven generation of interventions for clinical services will allow for greater certainty of pharmacy personnel navigating the intervention along with improved ability to accurately document a patient's response to clinical recommendations. The advantage of the medication targeting strategy evaluated is evidenced by the generation of personalized third primary dose COVID-19 vaccine recommendations into the prescription profiles of immunocompromised patients to be recognized during the pharmacy dispensing process. One limitation to the assessment of the effectiveness of this targeting is the retrospective study design, which does not allow guidance or control of the documentation by pharmacists or pharmacy interns completing the interventions. Interventions without documented response also may not account for intangible pharmacist rationale that may have determined an intervention invalid. Future implementation of automated clinical interventions may benefit from improved education for pharmacy personnel on accurate documentation of patient responses. An additional limitation is the small sample size collected, minimizing the effect size that may be reported by the data evaluated. Inclusion criteria for this study have restricted the number of eligible patients identified by the pharmacy dispensing system to patients with immunosuppressing medications dispensed by the included pharmacy locations and record of 2 primary doses of COVID-19 mRNA vaccine within the pharmacy dispensing system. Additional patients who interact with the pharmacies included may also be eligible for recommendation of a third primary dose of COVID-19 vaccine but have filled their immunosuppressing medications at specialty pharmacies or received their first doses of COVID-19 vaccine from alternative health care providers. The time necessary for programing of the targeting algorithm by the information technology team of the community grocery store chain also delayed the identification of eligible patients by six weeks following the recommendation by the CDC for a third primary dose. This allowed for the possibility of eligible patients to receive their third dose prior to release of the targeting technology, which subsequently excluded them from the study. One solution identified for expanding the sample size despite the barriers identified is to expand data collection to the entire population serviced by the community grocery store pharmacy chain instead of pulling a subset of one division of pharmacy locations. With implementation of a more narrowed targeting algorithm applied to a larger population of patients, the sample size may be expanded to include a greater number of included patients eligible to receive the clinical recommendation. Broad selection of medication classes included in the targeting algorithm also resulted in potentially inappropriate intervention generation for patients without immunocompromised status. Inclusion of all corticosteroid regimens resulted in the automatic generation of interventions for patients who did not receive immunosuppressing doses of corticosteroids, considered 20 mg/day of prednisone or equivalent for at least 14 days. Inappropriate intervention generation may have also occurred for patients identified through medication targeting of human immunodeficiency virus (HIV) pharmacotherapy, which may be associated with Pre-Exposure Prophylaxis or Post-Exposure Prophylaxis therapy, or controlled HIV infection not described as immunocompromised status. Manual review by study investigators of medication regimens for patients screened for inclusion allowed inclusion of appropriate interventions for study evaluation, though narrowing of criteria for programmed medication targeting may improve the ability of the pharmacy dispensing system to recognize immunocompromised patients for clinical interventions. Record of third dose vaccination status was also manually confirmed with state immunization registry data to accurately assess the completion of the vaccination recommendations. Due to the timing of the additional CDC recommendation for the third primary dose in immunocompromised patients around the approximate time of general population booster dose recommendations, retrieval of data for third dose administration was not able to be clearly defined as a third primary dose or booster dose received. Future integration of state immunization registry data into the pharmacy dispensing software would allow for more accurate targeting of patients eligible for vaccine recommendations by including a comprehensive vaccination history into screening capabilities at the community pharmacy level. 6 Conclusion The incorporation of medication targeting technology into the pharmacy dispensing system was effective in identifying immunocompromised patients for the recommendation of a third primary dose of COVID-19 vaccine. Further specification within the program targeting for dosing regimen and route of administration along with integration of the state immunization registry into the pharmacy dispensing software may improve the accuracy of appropriate recommendations identified pertaining to any indicated immunization. Additional studies may explore the benefit of various educational strategies for effective pharmacist use of digital intervention documentation and further understanding of updated vaccination schedules in circumstances of rapid change to current recommendations. Author contributions Grace LaFleur: methodology, formal analysis, investigation, writing – original draft, administration Andrew Azzi: methodology, writing – review and editing Scott Schimmel: methodology, writing – review and editing Mitchell Howard: methodology, formal analysis, investigation, writing – review and editing, supervision. Previous presentations of this research American Pharmacists Association Annual Conference, San Antonio, Tx, March 2022; Ohio Pharmacists Association 144th Annual Conference, Columbus, OH, April 2022; Ohio Pharmacy Residency 10th Annual Conference, Ada, OH, May 2022. Declaration of competing interest Authors LaFleur, Azzi, Schimmel and Howard are employees of The Kroger Co. The authors declare no other relevant conflicts of interest or financial relationships. APPENDIX A. Classes of included targeted medications- Immunomodulators - Immunosuppressants - Corticosteroids - Disease-modifying antirheumatic drugs (DMARDs) - Tumor necrosis factor (TNF) blockers - Dihydrofolic acid reductase inhibitors - Antimetabolites - Antineoplastics - Tyrosine kinase inhibitors - Nucleoside reverse transcriptase inhibitors (NRTIs) - Non-nucleoside reverse transcriptase inhibitors (NNRTIs) - Protease inhibitors - Integrase inhibitors - Immune response modifiers - Janus kinase (JAK) inhibitors Acknowledgements The authors acknowledge Olivia Kinney, PharmD, MS, Whitney Dubois, PharmD, CPPS, and the data analytical team of Kroger Health for their collaboration in clinical intervention development and data compilation. ==== Refs References 1 Centers for Disease Control and Prevention COVID Data Tracker 2022, May 06 US Department of Health and Human Services, CDC Atlanta, GA https://covid.cdc.gov/covid-data-tracker 2 Mbaeyi S. Oliver S.E. Collins J.P. The advisory committee on immunization practices' interim recommendations for additional primary and booster doses of COVID-19 vaccines — United States, 2021 MMWR Morb Mortal Wkly Rep 70 2021 1545 1552 34735422 3 Golob J. Immunocompromised People Make up Nearly Half of COVID-19 Breakthrough Hospitalizations. Health & Wellness Topics, Health Tips & Disease Prevention. Published August 23, 2021. Accessed March 30, 2022. 4 Tenforde M.W. Patel M.M. Gaglani M. Effectiveness of a third dose of Pfizer-BioNTech and Moderna vaccines in preventing COVID-19 hospitalization among immunocompetent and immunocompromised adults – United States August – December 2021 MMWR Morb Mortal Wkly Rep 71 2022 118 124 35085218 5 Kotton C.N. Kroger A.T. Freedman D.O. Immunocompromised travelers. CDC yellow book Published August 6, 2021 https://wwwnc.cdc.gov/travel/yellowbook/2020/travelers-with-additional-considerations/immunocompromised-travelers 6 Coley K.C. Gessler C. McGivney M. Richardson R. DeJames J. Berenbrok L.A. Increasing adult vaccinations at a regional supermarket chain pharmacy: a multi-site demonstration project Vaccine 38 24 2020 4044 4049 32093985 7 Page A. Harrison A. Nadpara P. Goode J.V.R. Pharmacist impact on pneumococcal polysaccharide vaccination rates in patients with diabetes in a national grocery chain pharmacy Journal of the American Pharmacists Association vol. 60 2020 Elsevier B.V. S51 S55 e1 8 Murray K. Low C. O’rourke A. A quality improvement intervention failed to significantly increase pneumococcal and influenza vaccination rates in immunosuppressed inflammatory arthritis patients Clin Rheumatol 39 2020 747 754 31820135 9 Klassing H.M. Ruisinger J.F. Prohaska E.S. Melton B.L. Evaluation of pharmacist-initiated interventions on vaccination rates in patients with asthma or COPD J Community Health 43 2018 297 300 28852915	0	PMC9715457	NO-CC CODE	2022-12-08 23:16:25	no	Res Social Adm Pharm. 2022 Dec 2; doi: 10.1016/j.sapharm.2022.11.013	utf-8	Res Social Adm Pharm	2,022	10.1016/j.sapharm.2022.11.013	oa_other
==== Front Biosens Bioelectron Biosens Bioelectron Biosensors & Bioelectronics 0956-5663 1873-4235 Elsevier B.V. S0956-5663(22)01005-3 10.1016/j.bios.2022.114965 114965 Article Affordable on-site COVID-19 test using non-powered preconcentrator Kim Jinhwan a1 Kim Cheonjung ab1 Park Jeong Soo a Lee Na Eun ac Lee Seungmin ad Cho Sung-Yeon ef Park Chulmin e Yoon Dae Sung d∗∗ Yoo Yong Kyoung b∗∗∗ Lee Jeong Hoon a∗ a Department of Electrical Engineering, Kwangwoon University, Seoul, 01897, Republic of Korea b Department of Electronic Engineering, Catholic Kwandong University, Gangneung-si, Gangwon-do, 25601, Republic of Korea c Department of Biotechnology, College of Life Sciences and Biotechnology, Korea University, Seoul, 02841, Republic of Korea d School of Biomedical Engineering, Korea University, Seoul, 02841, Republic of Korea e Vaccine Bio Research Institute, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea f Division of Infectious Diseases, Department of Internal Medicine, Seoul St. Mary's Hospital, Catholic Hematology Hospital, College of Medicine, The Catholic University of Korea, Seoul 06591, Republic of Korea ∗ Corresponding author. ∗∗ Corresponding author. ∗∗∗ Corresponding author. 1 These authors contributed equally. 2 12 2022 15 2 2023 2 12 2022 222 114965114965 2 9 2022 24 11 2022 27 11 2022 © 2022 Elsevier B.V. All rights reserved. 2022 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. A simple, affordable point of care test (POCT) is necessary for on-site detection of coronavirus disease 2019 (COVID-19). The lateral flow assay (LFA) has great potential for use in POCT mainly because of factors such as low time consumption, low cost, and ease of use. However, it lacks sensitivity and limits of detection (LOD), which are essential for early diagnostics. In this study, we proposed a non-powered preconcentrator (NPP) based on nanoelectrokinetics for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Antigen (Ag) lateral flow assay. The non-powered preconcentrator is composed of glass fiber-based composite paper and ion permselective material, and it can be simply operated by force balancing gravitational, capillary, and depletion-induced forces. The proposed approach helps enrich the SARS-CoV-2 viral nucleocapsid (N) proteins based on a 10-min operation, and it improved the LOD by up to 10-fold. The corresponding virus enrichment, which was evaluated using the reverse-transcriptase polymerase chain reaction (RT-PCR), revealed an improvement in ΔCt values > 3. We successfully demonstrated the enhancement of the NPP-assisted LFA, we extended to applying it to clinical samples. Further, we demonstrated an affordable, easy-to-implement form of LFA by simply designing NPP directly on the LFA buffer tube. Keywords COVID-19 LFA Preconcentrator SARS-CoV-2 Sample treatment Enrichment ==== Body pmc1 Introduction The reverse-transcriptase polymerase chain reaction (RT-PCR) is the most widely adopted approach for the rapid and accurate detection of the coronavirus disease 2019 (COVID-19) mainly because of factors such as high accuracy and sensitivity (Carter et al., 2020; Udugama et al., 2020; van Kasteren et al., 2020; Wolfel et al., 2020). However, a few drawbacks associated with RT-PCR include the high cost, long operation time, and complexity (Lai et al., 2021; Pokharel et al., 2022; Shani-Narkiss et al., 2020; Teymouri et al., 2021). The lateral flow assay (LFA) for the point-of-care test (POCT) could be a rapid, simple, and low-cost approach for detecting SARS-CoV-2; self-testing could alleviate the pressure on central laboratories for COVID-19. The LFA platform meets the WHO (World Health Organization) “ASSURED” criteria (sensitive, specific, user-friendly, rapid and robust, equipment-free, and deliverable to end-users) (Chaouch, 2021; Jiang et al., 2021; Otoo and Schlappi, 2022), and it could potentially satisfy the aforementioned criteria for the point-of-care test (POCT) and applicable various applications (Gong Y et al., 2017; Kim et al., 2017; Kim et al., 2022; Kumar et al., 2018; Li et al., 2020a, Li et al., 2020b). Two recently published papers demonstrated new diagnostic strategies for handling pandemics, and they showed that the most accurate and ideal COVID-19 detection could be achieved using frequent testing based on low-cost, simple, and rapid approaches mainly because of the ability of SARS-CoV-2 to spread at an exponential rate (Crozier et al., 2021; Mina et al., 2020). A one-time monitored highly sensitive RT-PCR test can potentially detect viral shedding even for longer infection periods up to 17 days, i.e., the convalescence stage with less transmission. Therefore, rapid testing for COVID-19 with high sensitivity will help minimize the virus transmission significantly, particularly in the early detection stage. However, the low sensitivity of LFA precise diagnostics limits its practical applications. In the case of COVID-19, detecting the SARS-CoV-2 antigen (Ag) in low virus titers (Ct value > 30) using the LFA is difficult, which limits the practical application in cases that involve lower virus titers targeting the early stage of COVID-19.(Escrivá et al., 2021; Gremmels et al., 2021). To overcome the drawback of LFA, several approaches for acquiring enhanced sensitivity and limits of detection (LOD) of LFA have been suggested with signal amplification (Cheng et al., 2017; Wang et al. 2017, Wang et al., 2021a, Wang et al., 2021b), flow control (Alam et al., 2021; Rivas et al., 2014; Sena-Torralba et al., 2020), optimization of LFA (Dighe et al., 2022; Garg et al., 2021; Grant et al., 2020; Yu et al., 2020; Zhang et al., 2021a, Zhang et al., 2021b), and sample enrichment (Kang et al., 2021; Wang et al., 2021a, Wang et al., 2021b; Zhou et al., 2021). Among these approaches, the sample enrichment technique has attracted considerable research attention for enhancing sensitivity and LOD (Deng et al., 2021; Huang et al., 2019; Zhang et al., 2021a, Zhang et al., 2021b). The nanoelectrokinetic (NEK)-based method is a promising sample enrichment technique for the preconcentration of charged biomolecules because of its high preconcentration factor (PF), minimal component requirements, and easy implementation in various materials (Lee et al., 2021). As an example of NEK-based sample enrichments, we propose NEK preconcentration in various forms on a paper, such as 2D lateral flow form (Han et al. 2016, 2018; Kim et al., 2017) and 3D origami form(Han et al., 2019; Kim et al., 2019). As an alternate device structure for NEK-based enrichments, we presented a 3D roll-like device structure, i.e., microfluidic paper-based large-volume preconcentrator (μ-LVP), to enhance the micro ribonucleic acid (miRNA) (Lee et al., 2021) and for COVID-19 antibody test (Kim et al., 2022). Although μ-LVP showed considerable improvements for sample preparations, it requires an external power source or battery to operate the preconcentration step. Moreover, in the practical application of SARS-CoV-2 Ag, we observed severe false positive signals, which, in turn, resulted in a decrease in the selectivity during nanoelectrokinetic operation with an electric field. The main reason could be the electric-field induced pH fluctuation during nanoelectrokinetic operation, denaturing antigen–antibody interaction, or causing protein denaturation. We suspected that a second possible reason could be the electric field given that the SARS-CoV-2 Ag is reportedly vulnerable to electric fields and can be easily denatured by an electric field. (Arbeitman et al., 2021). If we could realize the non-powered preconcentrator on paper materials, the preconcentrator could provide wide applications in paper-based microfluidics and detection systems (Carrell et al., 2019; Hu et al., 2019; Li et al., 2020a, Li et al., 2020b; Nishat et al., 2021; Niu et al., 2021; Ruiz et al., 2022; Tian et al., 2018). In this study, we proposed a non-powered preconcentrator (NPP) that can be simply operated by force balancing gravitational, capillary, and depletion-induced forces, and does not require an external power source or additional equipment. The NPP preconcentrator was combined with an LFA platform to overcome the poor sensitivity and LOD of the LFA (See Fig. 1 ). With the simplicity, low cost, portability, and versatility of the NPP sample enrichment technique, we acquired a 10-fold LOD enhancement using an NPP-assisted commercial SARS-CoV-2 Ag test compared to the commercial LFA (control). In addition to this, the corresponding virus enrichment of NPP was evaluated as ΔCt values > 3 using RT-PCR, which corresponds to a 10-folding enhancement in LOD. Finally, we designed NPP using commercially available sample tubes and integrated it with commercial LFA.Fig. 1 Preconcentration and assay procedure. (a) Human nasal sample assay of COVID-19 with the preconcentration. (b) Preconcentration procedure without external power source. Fig. 1 2 Materials & methods 2.1 NPP devices fabrication To fabricate the NPP, we used glass fiber-based composite paper as a paper-based microfluidic material (Whatman paper, fusion 5TM, GE Healthcare Life Sciences, USA). We used 5% Nafion resin (Sigma-Aldrich, USA) to ensure permselectivity. The fabrication step of the NPP devices is shown in Fig. S1. The Nafion strips were precisely printed on pre-defined Fusion 5 paper (3 × 5 cm) using a custom-made dispenser (BID-02AV, BTM Co., South Korea) that was equipped with a syringe pump (NE-1000 Single Syringe Pump, USA, flow rate setting: 0.6 mL/min). The Nafion resins were then dried on a hot plate at 70 °C for 15 min. We optimized the patterning of Nafion using the optimal infiltration processing, which involved a four-time coating and drying process (Fig. S2). The Nafion coated Fusion 5 paper was then rolled into discs; the rolled discs were sealed with transparent tape (3M Scotch, USA). The assay protocol of NPP-assisted LFA assay is shown in Fig. 1. In the NPP operation, we did not use any external power sources, unlike other NEK-based preconcentrator that operated with a power source. We only injected samples onto the top of NPP and then waited 10 min for the sample enrichment. The main design parameter maintains the force balances between gravimetric, capillary, and depletion-induced forces. We will discuss the detailed analysis with simulation and experimental results in a separate paper. We achieved the sample enrichment by maintaining the force balance, cutting the enriched plug, and extracting the enriched samples using a commercial extraction buffer. Following the manufacturer's guideline, we monitored the color intensity of the test line after 15 min. 2.2 Fluorescence and LFA colorimetric image analysis We used Alexa Fluor™ 594 conjugated albumin from bovine serum (BSA, Thermo Fisher Scientific, USA), 1X phosphate buffered saline buffer solution (PBS, Corning, USA) to visualize and assess the overall enrichment. We acquired fluorescence images from a microscope (IX-71, Olympus, Japan) with a charged-coupled device camera (Hamamatsu Co., Japan), and we analyzed these images using ImageJ software (Wayne Rasband, National Institute of Health, Bethesda, USA). In addition to this, we analyzed the impact of the NPP sample enrichment on the performance of the device. For control experiments, we tested the commercial SARS-CoV-2 Ag kit. The color intensities of the test and control lines were analyzed using a custom-made National Instrument (NI) controlled optical system that was operated on the LabVIEW™ software (National Instrument Co., USA) (Fig. S3). We first measured color signals to set LOD values using the NI-controlled optical system. Then, five individually-trained engineers (Calth Inc. http://www.thecalth.com) observed the colorimetric signal using the standard color chart under the manufacturer's guidelines with 19/20 criteria, which represents the LOD is determined as the lowest concentration where ≥95% (19/20) are positive (https://www.fda.gov/media/137302/download). Finally, we determined the LOD color intensity values in the reader/optical system using labeled information. The LOD line was indicated as an invisible line in Fig. 4b–c. 2.3 Preparation of SARS-CoV-2 Ag samples We used commercial SARS-CoV-2 Ag rapid kits (Calth Inc., Republic of Korea) to test the impact of the NPP on the overall performance. We prepared a COVID-19 nucleocapsid protein recombinant antigen (45 kDa, FPZ0516, Fapon Biotech Inc., China, see Table S1), which is regarded as the best COVID-19 Ag's test target protein, using 1 × PBS buffers (LB004, DUKSAN, Republic of Korea) to examine the sensing performance (i.e., LOD and sensitivity curve). We also used Influenza A and B virus samples (Zeptometrix, USA) and the human nasal sample (Trina Bioreactives AG, Swiss) for the cross-reactivity tests. In addition to this, casein (Sigma-Aldrich, USA) of 0.1% to block non-specific binding and 0.1X PBS buffer solution was mixed to SARS-CoV-2 Ag preconcentration and assay. 2.4 RT-PCR for clinical sample assay For the molecular analysis, we extracted the RNA from the collection disc of NPP and controls using a 30 μL extraction buffer (PCR Direct extracting transport medium, ARCIS, UK). The extracted RNA samples were analyzed using RT-PCR. The RNA was transformed as complementary deoxyribonucleic acid (cDNA) and amplified as DNA using AccuPower® RT-PCR PreMix (BIONEER, South Korea), according to a previous report (the forward primer sequence: TTCGGAAGAGACAGGTACGTTA, the reverse primer sequence: AGCAGTACGCACACAATCG) (Park et al., 2020). The reactions were performed in a thermal cycler (Gentier Mini, Tianlong, China) for 15 min at 50 °C, followed by 94 °C for 3 min, 40 cycles of 94 °C for 30 s, 62 °C for 40 s, 72 °C for 1 min, and a final extension at 72 °C for 10 min. We prepared the real COVID-19 samples with patient samples from the college of medicine, the Catholic University of Korea (Seoul, Republic of Korea), and the Trina Bioreactives AG (Nänikon, Switzerland). All the samples from Catholic University were collected under the Institutional Review Board of Seoul St. Mary's hospital approved the research 6 protocol with a waiver of informed consent (KC21TIDI0134K). Table S3 shows the detailed sample information. 2.5 Prototype design For designing the prototype for POCT sample enrichment, we designed the sample enrichment part, which can be easily connected onto the extraction buffer tube, and they are generally used in commercial LFA. We designed the prototype using two components; one for the absorbent part, and the other for the collection part (enrichment part), using Rhinoceros 3D software. The design will be discussed in detail in the results sections (Fig. 6). 3 Results & discussion 3.1 Sample enrichment using a non-powered preconcentrator (NPP) The two important controllable parameters required to balance the gravimetric, capillary, and depletion-induced forces are the depletion-induced and capillary-driven forces. First, we checked the extracting performance of several paper materials because the extraction after enrichment has a significant impact on the performance enhancement. We tested three paper materials, i.e., Whatman grade 1 ™, Whatman grade 6 ™, and Whatman fusion 5 ™. Then we observed that the extraction efficiencies were 95.7, 37, and 17.2% for Whatman Fusion 5, Grade 6, and Grade 1 paper, showing that the Whatman Fusion 5 ™ paper is the best candidate for NPP operations with low sample absorption and excellent sample delivery (Gan et al., 2014; Kim et al., 2022). We confirmed the minimized non-specific binding using Whatman fusion 5 ™. Second, we fabricated an NPP device using a Whatman Fusion 5™ paper and controlled the depletion-induced force by controlling the portion of Nafion on the cross-sectional area (Fig. S1). We coated Nafion strips a constant width of 1 mm, and we then controlled the thickness using the coating number. The cross-sectional view showed the unrolled NPP with the layered Nafion/Fusion 5™ paper (Fig. S2). The increased coating number indicates the increase in Nafion portion through the cross-sectional area since we rolled Whatman Fusion 5™ paper after Nafion coating. Therefore, by simply increasing the Nafion coating number, we could increase the depletion-induced forces, and this helped prevent the charged particle from freely passing out while decreasing the capillary forces acting on the water transported through the cross-sectional area by capillary force. As shown in Fig. S2, we optimized the optimal coating number as four. The Whatman Fusion 5™ paper has a higher wicking speed (0.105 cm/s) than Grade 1(0.007 cm/s) by 15-fold. The force balancing with a higher wicking speed accelerated the enrichment, and thus, the NPP process could be completed within 10 min. In Fig. 2 , we show the enrichment process of the NPP and rolled cylinder structure (RCS). As a control experiment, we prepared the RCS without Nafion strip lines, and we therefore performed RCS for the control experiment (Fig. 2a). We prepared 400 μL Alexa Fluor 594TM conjugated bovine serum albumin (594-BSA) of 0.1 mg/mL using a 0.1X PBS buffer solution to study visualize the enrichment process. After the injection of the sample onto the top of the RCS (Fig. 2a) and NPP (Fig. 2b), we monitored the enrichment of BSA using fluorescence microscopy.Fig. 2 BSA sample preconcentration. Image of preconcentration (a) without and (b) with Nafion pattern. Fluorescence images of the BSA for observation of accumulation (a) without and (b) with Nafion pattern. (c) Fluorescence intensity graph about preconcentrated BSA sample in the collection disc. (d) Distribution graph according to distance with BSA preconcentration on RSC and NPP surface. Fig. 2 The 594-BSA revealed clear enhanced signals on the upper side of NPP (Fig. 2b) whereas RCS showed no significant fluorescent signals after 10 min of operation (Fig. 2a). We observed slight fluorescence signals across the NPP due to the absorption of fluorescence. With NPP operating time (10 min), we acquired the preconcentration factor over 10-fold (Fig. 2c). Each sample was tested five times. We observed that the accumulation of BSA is only located on the top collection disc part by NPP, indicating that the force balancing the samples contributes to the enrichment without significant losses, even without external electrical and mechanical components (Fig. 2d). 3.2 NPP optimization We optimized the design parameters that would help achieve the best performance, i.e., a collection disc, layered Nafion strip, and absorbent region (Fig. 3 ). First, we tested the impact of the length of the collection disc on the preconcentration factor. By fixing the total length of NPP as 3 cm, we designed four different collection disc lengths—1, 2, 3, and 4 mm. We injected 400 μL of 594-BSA sample and monitored the fluorescence intensity of BSA accumulation after 10 min. Fig. 3a showed that the 1 mm length of the collection disk showed higher fluorescence intensity, which represents the preconcentrator factor (PF).Fig. 3 Optimization of the preconcentration. Fluorescence intensity graph according to (a) collection disc length, (b) device length, and (c) sample volume. Fig. 3 Fig. 4 Detection sensitivity of COVID-19 antigen test. (a) Preconcentration and the assay procedure of nucleocapsid protein recombinant (SARS-CoV-2) antigen. (b) Color intensities of the preconcentration with different SARS-CoV-2 Ag concentrations. (c) Selectivity test of COVID-19 assay with the preconcentration. (d–e) Test and the control line images of (d) commercial COVID-19 LFA and (e) NPP-assisted LFA. Fig. 4 We designed four different absorbent region lengths as 2, 3, 4, and 5 cm, with a fixed collection disc length of 1 mm (Fig. 3b). After injecting the BSA sample of 300 μL onto the top of NPP, we observed a higher fluorescence of BSA accumulation by fluorescence intensity for the length of 3 cm. Finally, we evaluated the impact of the injecting sample volume on preconcentration performance. We tested the injection volume of BSA with 200, 300, 400, and 500 μL, with a fixed total length (3 cm) and collection disc (1 mm), After 10 min, we observed that the fluorescence intensity reached a maximum at 400 μL (Fig. 3c), indicating that this volume is sufficient for acquiring higher preconcentration factors. 3.3 SARS-CoV-2 Ag preconcentration and assay To validate the performance enhancement via NPP sample enrichment, we carried out a COVID-19 LFA assay with NPP sample enrichment (Fig. 4). As a control experiment, we also conducted the assay using a commercial COVID-19 Ag kit. We prepared nucleocapsid protein recombinant (SARS-CoV-2) antigens using five different concentrations (0.01 ng/mL to 1 ng/mL) and 0.1X PBS buffer solution (Fig. 4). We enriched 400 μL of each sample into a final volume of 13 μL, which corresponds to a 30-fold enrichment. Using the positively charged nucleocapsid proteins, we were able to enhance LOD ∼10-fold. We injected 400 μL SARS-CoV-2 nucleocapsid Ag onto the top of NPP (Fig. 4a). After 10 min of the enrichment operation, we cut the collection disc (enriched part) with a hand-held cutter. Then, the collection disc was transferred to commercial LFA. Finally, 100 μL commercial extraction buffer was passed through the pipette and assayed using commercial LFA following the manufacturer's guidelines. We conducted the control experiment using a commercial LFA. To evaluate the analytical sensitivity and LOD enhancement facilitated by the NPP, we showed assay results both from commercial LFA and NPP-assisted LFA (Fig. 4b). We observed the test/control line after 15 min according to the manufacturer's guidelines to determine the sensitivity and LOD. The total assay time of NPP-assisted LFA assay was within 25 min, which included an additional 10-min enrichment process with a 15-min commercial assay time. In Fig. 4b, the red square represents the color intensities of the NPP-assisted LFA test line, and the gray circle indicates that of commercial LFA. Using the NPP-assisted LFA, we observed the increase in color intensity for all concentrations, showing the enhanced assay performance. To set the LOD values that are represented by the invisible line in Fig. 4b, we first measured color signals using custom-built imaging system. Next, five individually trained engineers (Calth Inc. http://www.thecalth.com) observed the colorimetric signal using the standard color chart to determine the LOD. Finally, we determined the LOD color intensity values in the reader/optical system using labeled information. We observed a clear enhancement of the colorimetric signals with NPP-assisted enrichment process (Fig. 4e) compared with commercial one (Fig. 4d). The LOD from the color intensity versus concentration plot shows a higher LOD of NPP-assisted LFA (>10-fold) compared to the LFA without preconcentration, and this indicates the massive potential of the NEK-assisted enrichment process for POCT. 3.4 SARS-CoV-2 Ag specificity test To study the specificity and undesirable false positive signal from NPP operations, we prepared one protein (BSA), two viruses (Influenza A and Influenza B), and two buffers (normal nasal and buffer only), and we tested five different samples that did not contain any SARS-CoV-2 Ag (Fig. 4c). First, we evaluated the color intensity of the commercial LFA using a 0.1 ng/mL SARS-CoV-2 Ag. We did not observe any visible color signals. Then, using the NPP-assisted preconcentration, we acquired visible color signals, which indicate the signal enhancement without any undesirable denaturation of target nucleocapsid protein given that we performed NPP-assisted LFA without any electric field. Second, to further examine whether cross-reactivity happens after NPP-assisted enrichment, we assayed using NPP-assisted LFA with one protein (BSA), two viruses (Influenza A and Influenza B), and two buffers (normal nasal and buffer only), and we did not observe any cross-reactivity, thus indicating good selectivity. 3.5 Clinical sample test We validated the assay enhancement using an NPP-assisted LFA in the previous section. Although we successfully demonstrated the efficacy of the NPP-assisted LFA with diluted standard samples, we needed to verify if the NPP-assisted LFA could be extended onto real patient samples (Fig. 5 ). In Fig. 5a, we show the assay protocol, which included sample collection, NPP enrichment, and commercial LFA assay. The only step added to the commercial LFA test is the 10-min NPP enrichment process. We diluted the real COVID-19 patient samples to demonstrate the performance in low Ag concentration. Table S2 shows the detailed sample information. We included low concentration samples to validate the performance of the NEK-LFA because the commercial LFA cannot detect those low Ag concentrations.Fig. 5 Clinical sample test of COVID-19 antigen test. (a) Sample enrichment and the assay procedure of human nasal sample. (b) Color intensity graph according to diluted sample concentration with the preconcentration. (c) Test and the control line images by enhanced signal with NPP-assisted LFA. (d) Fluorescence intensities of the cycles with RT-PCR test of the preconcentrated COVID-19 sample. (e) Clinical sample evaluations using 10 samples (n = 5 for positive and n = 5 for healthy controls), showing the enhanced assay performance with NPP enrichment. Fig. 5 Fig. 6 Preconcentration kit test. (a) Preconcentration kit. Absorbent part assembled with the NPP and a plastic case. Collection disc is located in the center of collection part. Releasing buffer (sky blue) and the extraction buffet reservoir (blue) in releasing part. (b) Buffer tube in a commercialized LFA. (c) Preconcentration (d) the extraction and the assay process. (e) Comparison of 0.05 and 0.1 ng/mL SARS-CoV-2 Ag test with and without the preconcentration. Fig. 6 We increased the colorimetric signal with the NPP-assisted LFA by concentrating target sample, and we enhanced the LOD by adapting the NPP enrichment step (Fig. 5b–d). First, we demonstrated the ability of NPP-assisted LFA using a patient (female, 63 years). With two diluted samples (500-fold and 100-fold dilutions), we significantly enhanced the colorimetric signal via NPP enrichment, indicating the ability to extend the sensitivity and LOD of the NPP-assisted LFA with real human nasal sample (Fig. 5b–c). Moreover, we tested the NPP-assisted function for the virus concentration using RT-PCR (Fig. 5d). Generally, a lower Ct value indicates higher virus titers. Without NPP, the samples with dilution from the patient sample (male, 59 years) showed lower virus titers (Ct = 33.51). However, they significantly increased the concentrations (Ct = 30.43), representing an 8-fold increase in the overall concentration. Clinical samples include inhibitors, and this may have an adverse impact on the performance of the immunoassay (LFA test) and molecular assay (RT-PCR). We expect that removing inhibition could be another parameter that enhances the Ct values in RT-PCR. To show the clinical ability, we tested the assay with NPP-assisted LFA using 10 clinical samples (n = 5 for positive and n = 5 for healthy controls) (Fig. 5e). We observed that the colorimetric signals are significantly enhanced with NPP-assisted LFA, compared with those of commercial LFA (sample #6∼#10 in Fig. 5e). In contrast to SARS-CoV-2 positive patient samples, samples from healthy controls (sample #1∼#5) showed no signal enhancements with NPP-assisted LFA, representing no false-positive signals via NPP enrichment process. 3.6 Prototype of NPP for POCT In most commercialized LFAs, a flexible tube that contains extraction buffers is generally used based on the manufacturer's guidelines. In a routine assay process, nasal/nasopharyngeal swab samples or saliva samples that extracted the viral protein in an extraction buffer tube were collected, and the diluted samples were applied to the commercial LFA. To suffice low-cost requirements, we attempted to design sample enrichment gadgets directly onto the extraction tube (Fig. 6). Consequently, we do not require any external power source, battery, or pump for sample enrichment. The only additional action that is required is turning the test-tubes upside down. In Fig. 6a, we illustrated four components of NPP. The commercial LFA includes the buffer tube (Fig. 6b) and injection part (part 1 in Fig. 6a). For the NPP enrichment, we added two parts—the absorbent part (2) and the collection part (3). We designed the releasing part (4) by modifying the commercial buffer tube to deliver the two buffers sequentially. For the operation (Fig. 6c), we collected the swabbed samples, released them onto a tube, and then turned the tube upside down after connecting the filter parts. Finally, after the 10-min NPP operation, we replaced the absorbent part with a tube cap; we then tested the LFA assay (Fig. 6d). To verify the feasibility of the prototype, we tested the NPP-assisted LFA assay on two prepared samples (0.05 and 0.1 ng/mL), and we showed the signal enhancement on these samples (Fig. 6e). A further optimization is needed to maximize NPP design. However, we could suggest the first nanoelectrokinetic device with a simplified design. 4 Conclusion We demonstrated a non-powered nanoelectrokinetic preconcentration that can be directly used in the COVID-19 LFA kit. To meet the acceptable criteria of WHO, we suggested an inexpensive sample enrichment technique directly applicable to commercial LFA. We tested the NPP enrichment directly onto a commercial buffer tube and demonstrated the enrichment without any electrical component or external pump. As a result, we acquired an enhancement of LOD by up to 10-fold in COVID-19 detection with the 10 min-NPP operations; moreover, we acquired an enhanced Ct value (ΔCt ∼3) from the RT-PCR. We expected the additional cost to be less than 10 cents, given that NPP requires a small budget for sample preparation. Thus, it is highly affordable compared to other paper-based and microfluidic chips. Through further investigation, we expect that the NPP enrichment technique could meet WHO's desirable criteria for the point of care test (POCT) with a sensitivity of >90%, a specificity of >99%, Ct ≥ 30, and an assay time <20 min. In order for this NPP enrichment technique to be commercialized, the tests for application to other analysis equipment and the tests for application to other samples will be needed. CRediT authorship contribution statement Jinhwan Kim: Conceptualization, Methodology, Investigation, Formal analysis, Writing – review & editing. Cheonjung Kim: Methodology, Validation, Formal analysis, Writing – review & editing. Jeong Soo Park: Methodology, Formal analysis. Na Eun Lee: Formal analysis, Investigation. Seungmin Lee: Formal analysis. Sung-Yeon Cho: Clinical samples. Chulmin Park: Clinical samples. Dae Sung Yoon: Supervision, Writing – review & editing. Yong Kyoung Yoo: Conceptualization, Supervision, Writing – review & editing. Jeong Hoon Lee: Conceptualization, Supervision, Project administration, Writing – review & editing. 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 related to this article:Multimedia component 1 Multimedia component 1 Data availability No data was used for the research described in the article. Acknowledgements This research was supported by the Bio & Medical Technology Development Program of the 10.13039/501100001321 National Research Foundation (NRF) funded by the Korean government (MSIT) (No. 2021M3E5E3080743). J.H. Lee was supported by a research grant from 10.13039/501100002643 Kwangwoon University in 2022. Appendix A Supplementary data related to this article can be found at https://doi.org/10.1016/j.bios.2022.114965. ==== Refs References Alam N. Tong L. He Z. Tang R. Ahsan L. Ni Y. Cellulose 28 13 2021 8641 8651 34305338 Arbeitman C.R. Rojas P. Ojeda-May P. Garcia M.E. Nat. Commun. 12 1 2021 5407 34518528 Carrell C. Kava A. Nguyen M. Menger R. Munshi Z. Call Z. Nussbaum M. Henry C. Microelectron. Eng. 206 2019 45 54 Carter L.J. Garner L.V. Smoot J.W. Li Y. Zhou Q. Saveson C.J. Sasso J.M. Gregg A.C. Soares D.J. Beskid T.R. Jervey S.R. Liu C. ACS Cent. Sci. 6 5 2020 591 605 32382657 Chaouch M. Rev. Med. Virol. 31 6 2021 e2215 33476080 Cheng Z. Choi N. Wang R. Lee S. Moon K.C. Yoon S.-Y. Chen L. Choo J. ACS Nano 11 5 2017 4926 4933 28441008 Crozier A. Rajan S. Buchan I. McKee M. BMJ 372 2021 n208 33536228 Deng Y. Jiang H. Li X. Lv X. Microchim. Acta 188 11 2021 379 Dighe K. Moitra P. Alafeef M. Gunaseelan N. Pan D. Biosens. Bioelectron. 200 2022 113900 Escrivá B.F. Mochón M.D.O. González R.M. García C.S. Pla A.T. Ricart A.S. García M.M. Aranda I.T. García F.G. Cardona C.G. J. Clin. Virol. 143 2021 104941 Gan W. Zhuang B. Zhang P. Han J. Li C.-X. Liu P. LChip 14 19 2014 3719 3728 Garg M. Sharma A.L. Singh S. Biosens. Bioelectron. 171 2021 112703 Gong Y H.J. Choi J.R. You M. Zheng Y. Xu B. Wen T. Xu F. Int. J. Nanomed. 12 2017 4455 4466 Grant B.D. Anderson C.E. Williford J.R. Alonzo L.F. Glukhova V.A. Boyle D.S. Weigl B.H. Nichols K.P. Anal. Chem. 92 16 2020 11305 11309 32605363 Gremmels H. Winkel B.M.F. Schuurman R. Rosingh A. Rigter N.A.M. Rodriguez O. Ubijaan J. Wensing A.M.J. Bonten M.J.M. Hofstra L.M. EClinicalMedicine 31 2021 100677 Han S.I. Hwang K.S. Kwak R. Lee J.H. Lab Chip 16 12 2016 2219 2227 27199301 Han S.I. Lee D. Kim H. Yoo Y.K. Kim C. Lee J. Kim K.H. Kim H. Lee D. Hwang K.S. Yoon D.S. Lee J.H. Anal. Chem. 91 16 2019 10744 10749 31340120 Han S.I. Yoo Y.K. Lee J. Kim C. Lee K. Lee T.H. Kim H. Yoon D.S. Hwang K.S. Kwak R. Lee J.H. Sensor. Actuator. B Chem. 268 2018 485 493 Hu J. Xiao K. Jin B. Zheng X. Ji F. Bai D. Biotechnol. Bioeng. 116 10 2019 2764 2777 31282991 Huang Y. Xu T. Wang W. Wen Y. Li K. Qian L. Zhang X. Liu G. Microchim. Acta 187 1 2019 70 Jiang N. Tansukawat N.D. Gonzalez-Macia L. Ates H.C. Dincer C. Güder F. Tasoglu S. Yetisen A.K. ACS Sens. 6 6 2021 2108 2124 34076428 Kang J. Jang H. Yeom G. Kim M.-G. Anal. Chem. 93 2 2021 992 1000 33296598 Kim C. Yoo Y.K. Han S.I. Lee J. Lee D. Lee K. Hwang K.S. Lee K.H. Chung S. Lee J.H. Lab Chip 17 14 2017 2451 2458 28613296 Kim C. Yoo Y.K. Lee N.E. Lee J. Kim K.H. Lee S. Kim J. Park S.J. Lee D. Lee S.W. Hwang K.S. Han S.I. Lee D. Yoon D.S. Lee J.H. Biosens. Bioelectron. 212 2022 114385 Kim H. Lee K.H. Han S.I. Lee D. Chung S. Lee D. Lee J.H. Lab Chip 19 23 2019 3917 3921 31650155 Kumar S. Bhushan P. Krishna V. Bhattacharya S. Biomicrofluidics 12 3 2018 034104 Lai C.-C. Wang C.-Y. Ko W.-C. Hsueh P.-R. J. Microbiol. Immunol. Infect. 54 2 2021 164 174 32513617 Lee D. Lee J.W. Kim C. Lee D. Chung S. Yoon D.S. Lee J.H. Biosens. Bioelectron. 176 2021 112904 Li J. Ma B. Fang J. Zhi A. Chen E. Xu Y. Yu X. Sun C. Zhang M. Foods 9 1 2020 27 Li Z. You M. Bai Y. Gong Y. Xu F. Small Methods 4 4 2020 1900459 Mina M.J. Parker R. Larremore D.B. N. Engl. J. Med. 383 22 2020 e120 Nishat S. Jafry A.T. Martinez A.W. Awan F.R. Sensor. Actuator. B Chem. 336 2021 129681 Niu J. Bao Z. Wei Z. Li J.X. Gao B. Jiang X. Li F. Anal. Chem. 93 8 2021 3959 3967 33595273 Otoo J.A. Schlappi T.S. Biosensors 12 2 2022 124 35200384 Park M. Won J. Choi B.Y. Lee C.J. Exp. Mol. Med. 52 6 2020 963 977 32546849 Pokharel S. White L.J. Sacks J.A. Escadafal C. Toporowski A. Mohammed S.I. Abera S.C. Kao K. Melo Freitas M.D. Dittrich S. PLOS Global Public Health 2 3 2022 e0000293 Rivas L. Medina-Sánchez M. de la Escosura-Muñiz A. Merkoçi A. Lab Chip 14 22 2014 4406 4414 25241662 Ruiz R.A. Gonzalez J.L. Vazquez-Alvarado M. Martinez N.W. Martinez A.W. Anal. Chem. 94 25 2022 8833 8837 35694851 Sena-Torralba A. Ngo D.B. Parolo C. Hu L. Álvarez-Diduk R. Bergua J.F. Rosati G. Surareungchai W. Merkoçi A. Biosens. Bioelectron. 168 2020 112559 Shani-Narkiss H. Gilday O.D. Yayon N. Landau I.D. medRxiv 2020 20052159 Teymouri M. Mollazadeh S. Mortazavi H. Naderi Ghale-noie Z. Keyvani V. Aghababaei F. Hamblin M.R. Abbaszadeh-Goudarzi G. Pourghadamyari H. Hashemian S.M.R. Mirzaei H. Pathol. Res. Pract. 221 2021 153443 Tian T. Bi Y. Xu X. Zhu Z. Yang C. Anal. Methods 10 29 2018 3567 3581 Udugama B. Kadhiresan P. Kozlowski H.N. Malekjahani A. Osborne M. Li V.Y.C. Chen H. Mubareka S. Gubbay J.B. Chan W.C.W. ACS Nano 14 4 2020 3822 3835 32223179 van Kasteren P.B. van der Veer B. van den Brink S. Wijsman L. de Jonge J. van den Brandt A. Molenkamp R. Reusken C. Meijer A. J. Clin. Virol. 128 2020 104412 Wang C. Shi D. Wan N. Yang X. Liu H. Gao H. Zhang M. Bai Z. Li D. Dai E. Rong Z. Wang S. Analyst 146 12 2021 3908 3917 33970172 Wang L. Wang X. Cheng L. Ding S. Wang G. Choo J. Chen L. Biosens. Bioelectron. 189 2021 113360 Wang X. Choi N. Cheng Z. Ko J. Chen L. Choo J. Anal. Chem. 89 2 2017 1163 1169 28194991 Wolfel R. Corman V.M. Guggemos W. Seilmaier M. Zange S. Muller M.A. Niemeyer D. Jones T.C. Vollmar P. Rothe C. Hoelscher M. Bleicker T. Brunink S. Schneider J. Ehmann R. Zwirglmaier K. Drosten C. Wendtner C. Nature 581 7809 2020 465 469 32235945 Yu S. Nimse S.B. Kim J. Song K.-S. Kim T. Anal. Chem. 92 20 2020 14139 14144 32967427 Zhang C. Zheng T. Wang H. Chen W. Huang X. Liang J. Qiu L. Han D. Tan W. Anal. Chem. 93 7 2021 3325 3330 33570399 Zhang Q. Fang L. Jia B. Long N. Shi L. Zhou L. Zhao H. Kong W. TrAC, Trends Anal. Chem. 144 2021 116427 Zhou Y. Wu Y. Ding L. Huang X. Xiong Y. TrAC, Trends Anal. Chem. 145 2021 116452	36493723	PMC9715458	NO-CC CODE	2022-12-07 23:15:46	no	Biosens Bioelectron. 2023 Feb 15; 222:114965	utf-8	Biosens Bioelectron	2,022	10.1016/j.bios.2022.114965	oa_other
==== Front J Infect J Infect The Journal of Infection 0163-4453 1532-2742 The British Infection Association. Published by Elsevier Ltd. S0163-4453(22)00688-0 10.1016/j.jinf.2022.11.027 Letter to the Editor Viral Dynamics of Omicron BA.2.76 Variant of SARS-CoV-2 in a Cohort of COVID-19 Patients Li Kangguo 1# Wang Demeng 2# Qu Huimin 1# Rui Jia 1# Abudunaibi Buasiyamu 1 Guo Zhinan 2 Wu Sihan 2 Abudurusuli Guzainuer 1 Yang Zimei 1 Fang Kang 1 Zhang Yidun 2 Su Chenghao 3⁎ Chen Tianmu 1⁎ 1 State Key Laboratory of Molecular Vaccinology and Molecular Diagnostics, School of Public Health, Xiamen University, Fujian, China 2 Xiamen Center for Disease Control and Prevention, Fujian, China 3 Zhongshan Hospital, Fudan University (Xiamen Branch), Xiamen City, Fujian Province, People's Republic of China ⁎ Corresponding author # These authors contributed equally to this study. 2 12 2022 2 12 2022 29 11 2022 © 2022 The British Infection Association. Published by Elsevier Ltd. All rights reserved. 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. Keywords COVID-19 Omicron variant viral load Ct value ==== Body pmcDear Editor, In this Journal, Christopher Kandel et al. analyzed the viral dynamics of the Omicron variant and found vaccination has no impact on viral load, regardless of peak concentration and positive duration (1). However, as Christopher Kandel et al. discussed in the limitation section, they used self-collected swabs outcome, which could be easily affected by such factors as the not standardized operators or even the external environment (2). Besides, specimens from participants were refrigerated at home for 1-14 days which may impact the activity of the virus (3) and increase the risk of contamination with each other (4). Furthermore, this study may underestimate pre-symptomatic viral dynamics because the first swab of some patients was positive (1). To address this concern, we designed a cohort study among contacts of Omicron BA.2.76 variants infections to describe the dynamics of the Omicron variant. Difference from self-collected swabs and self-retrieved, our study used contact data provided by eligible hospitals and public health laboratories according to Guidelines for Coronavirus Testing (second edition) (5). This not only could reduce the risk of contamination and decreased activity but also avoided the impact of operating and environmental factors as much as possible because all specimens were sampled, transferred, and tested by professionals. Pre-symptomatic viral dynamics are observed in this study because oropharyngeal and nasopharyngeal swabs were collected from each participant daily before onset or test positive. From August 6 to August 19, 2022, a total of 15,730 contacts were introduced by 69 cases, of whom 61 were infected with the Omicron BA.2.76 variant before September 3, 2022 (Fig. 1 A), and 1 infection was excluded because of unclear vaccination. Patients were identified and classified according to the Diagnosis and treatment plan for COVID-19 (6) and were described in Supplementary materials 1. The analytic population included 60 symptomatic COVID-19 cases (median age: 44 years [IQR: 28-52]); 32 (53%) were male, and none with immunocompromising or any common chronic medical conditions.Fig. 1 The Ct value distribution reflects viral dynamics across the outbreak of Omicron BA.2.76 of SARS-CoV-2. (A) Daily incidence of infections in the outbreak. (B) Observed Ct value of N gene of each infection. (C) Observed Ct value of ORF gene of each infection. (D) Fitting Ct value of N gene and ORF gene, respectively. (E) and (F) Fitting Ct value under varied vaccination status of N gene and ORF gene, respectively. Fig 1 Among 60 COVID-19 symptomatic patients, 23 received 2 doses of inactivated vaccine (CoronaVac and COVILO) or 1 dose of the CanSino vaccine, and 27 further received the booster dose. Compared with unfully vaccinated patients (unvaccinated or received 1 dose inactivated vaccine patients), fully vaccinated and booster-vaccinated patients tended to be older (median age: 38 vs. 44 vs. 46 years). Still, the difference did not attain statistical significance (P=0.357) (Table 1 ). Among 23 fully vaccinated patients, 22 (96%) received inactivated vaccine, and only 1 received the CanSino vaccine. Among 27 booster-vaccinated patients, 24 (89%) received inactivated vaccine, and 3 (11%) received the CanSino vaccine. All specimens collected from patients were tested by polymerase chain reaction (PCR) and recorded the expression of the N gene (Fig. 1B) and ORF gene (Fig. 1C) when positive test. For N gene expression, peak virus load was identified at 2-3 days after onset with the median of cycle threshold (Ct) value of 16-17, regardless of whether patients received the booster dose. This is consistent with the previous study that there is no difference in viral trajectories between fully vaccinated and booster-vaccinated patients (1). However, the median range between onset and negative test after receiving a booster dose was 8 days (IQR: 7-11 days), which was shorter than 10 days (IQR: 8-11 days) for fully vaccinated patients and 12 days (IQR: 10-13 days) for unfully vaccinated patients. ORF gene expression was normalized to that of the N gene.Table 1 Characteristics of symptomatic patients with COVID-19 by vaccination status Table 1 No.(%) Characteristic Unfully vaccinated Fully vaccinated Booster dose P-value (N=10) (N=23) (N=27) Age, median (Q1-Q3), year 38(27-53) 44(19-50) 46(32-53) 0.357KW Gender 0.615Fish Male 4(40.0) 12(52.2) 16(59.3) Female 6(60.0) 11(47.8) 11(40.7) Type of vaccine 0.850Fish SinoVac 2(20.0) 8(34.8) 10(37.0) SinoPharm 1(10.0) 6(26.1) 8(29.6) SinoVac & SinoPharm 0(0.0) 8(34.8) 6(22.2) CanSino 0(0.0) 1(4.3) 3(11.1) (Missing) 7(70.0) 0(0.0) 0(0.0) Range between last vaccine and onset, median (Q1-Q3), days 387(134-443) 355(344-374) 174(142-221) <0.001KW N gene Peak cycle threshold, median (Q1-Q3) 17(15-17) 16(15-17) 16(15-17) 0.483KW Day of the peak, median (Q1-Q3) 3(2-5) 2(2-5) 3(2-3) 0.611KW Pre-symptomatic positivity 0.702Fish No 9(90.0) 22(95.7) 24(88.9) Yes 1(10.0) 1(4.3) 3(11.1) Range between onset and negative, median (Q1-Q3), days 12(10-13) 10(8-11) 8(7-11) 0.167KW ORF gene Peak cycle threshold, median (Q1-Q3) 17(16-21) 17(15-19) 16(15-18) 0.438KW Day of the peak, median (Q1-Q3) 3(2-5) 2(2-5) 3(2-3) 0.566KW Pre-symptomatic positivity 0.818Fish No 9(90.0) 22(95.7) 25(92.6) Yes 1(10.0) 1(4.3) 2(7.4) Range between onset and negative, median (Q1-Q3), days 12(7-13) 8(7-10) 8(7-11) 0.650KW KWKruskal-Wallis's one-way ANOVA FishFisher's exact test In order to reduce individual variations and estimate the positive duration (Ct value < 30), B-spline basis functions (using the 4th-degree basis function) were selected to model the change of Ct value after infected (Table S1, Supplementary materials 3). In our model, ORF gene expression was slightly higher than N gene expression, with a longer positive duration (9.85 vs. 8.80 days) (Fig. 1D). Pre-symptomatic positive duration was 0.5 days and 0.15 days for N gene and ORF gene, respectively (Fig. 1D). For N gene expression, the positive duration after receiving booster dose was 9.55 days, which was slightly shorter than fully vaccinated (10.15 days) and unfully vaccinated (10.70 days) (Fig. 1E). And similar results were observed for ORF gene expression in Fig. 1F (mean duration: 8.65 vs. 9.00 vs. 9.40 days). This is consistent with what was previously described in Table 1, completed full vaccination and received booster reduces the duration of positivity, regardless of N gene or ORF gene expression. Our study is limited by the insufficient sample size in the cohort. Specifically, patients who received the booster dose had higher viral loads may be accused of older and individual variability. In addition, differential vaccine types between fully vaccinated and booster patients would bias the analysis of viral dynamics. Funding This study was partly supported by the National Key Research and Development Program of China (2021YFC2301604), the Self-supporting Program of Guangzhou Laboratory (ZL-SRPG2200702), Medical and Health Guidance Project of Xiamen (3502Z20214ZD1297) and Guidance Project of Fujian Provincial Science & Technology Department (2019D014). Ethical statement This study was approved by the institutional ethics committee of the Xiamen Center for Disease Control and Prevention (CDC), Fujian, China. Written consent was obtained from patients or their guardians (s) when samples were collected. Author Contributions Conceptualization: TMC, CHS, ZNG, KGL. Investigation: HMQ, DMW, ZNG, ZMY, KF, SHW, YDZ. Methodology: KG Li, HMQ, TMC. Software: KGL, HMQ, GA. Validation: KGL, JR. Writing - original draft: KGL, HMQ, JR, BA. Writing - review & editing: TMC, SHW, JR, BA. Data and source code availability The source code and data of the analysis procedure are accessible at the GitHub repository (https://github.com/xmusphlkg/ct_analysis). Role of funding source The funder of the study had no role in study design, data collection, data analysis, data interpretation, manuscript prepared and reviewed. The corresponding authors had full access to all the data in the study and had final responsibility for the decision to submit for publication. Declaration of Competing Interest The authors declare no competing interests. Appendix Supplementary materials Image, application 1 Image, application 2 Image, application 3 Acknowledgments We thank the Xiamen Center for Disease Control and Prevention, China staff, for accessing the various data sources. The opinions expressed are those of the authors and not necessarily the institutions to which they are affiliated. Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jinf.2022.11.027. ==== Refs Reference 1 Kandel C Lee Y Taylor M Llanes A McCready J Crowl G Viral dynamics of the SARS-CoV-2 Omicron Variant among household contacts with 2 or 3 COVID-19 vaccine doses J Infect 2022 October 22 PubMed PMID: 36283495. Pubmed Central PMCID: PMC9595488. Epub 2022/10/26 2 Patriquin G Davidson RJ Hatchette TF Head BM Mejia E Becker MG Generation of False-Positive SARS-CoV-2 Antigen Results with Testing Conditions outside Manufacturer Recommendations: A Scientific Approach to Pandemic Misinformation Microbiol Spectr 9 2 2021 Oct 31 e0068321 PubMed PMID: 34668722. Pubmed Central PMCID: PMC8528119. Epub 2021/10/21 3 Scarica C Parmegiani L Rienzi L Anastasi A Cimadomo D Klinger FG SARS-CoV-2 persistence at subzero temperatures J Assist Reprod Genet 38 4 2021 Apr 779 781 PubMed PMID: 33544317. Pubmed Central PMCID: PMC7863614. Epub 2021/02/06 33544317 4 Caggiano G Triggiano F Apollonio F Diella G Lopuzzo M D'Ambrosio M SARS-CoV-2 RNA and Supermarket Surfaces: A Real or Presumed Threat? International journal of environmental research and public health 18 17 2021 Sep 6 PubMed PMID: 34501993. Pubmed Central PMCID: PMC8430590. Epub 2021/09/11 5 PRC NHC Guidelines for Coronavirus Testing (second edition) 2021 (cited 2021/9/14). Available from http://www.gov.cn/xinwen/2021-09/14/content_5637134.htm 6 PRC NHC. Diagnosis and treatment plan for COVID-19(Trial Version 9) (cited 2022/3/14). Available from: http://www.gov.cn/zhengce/zhengceku/2022-03/15/content_5679257.htm 2022.	36470410	PMC9715459	NO-CC CODE	2022-12-08 23:16:26	no	J Infect. 2022 Dec 2; doi: 10.1016/j.jinf.2022.11.027	utf-8	J Infect	2,022	10.1016/j.jinf.2022.11.027	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)00632-4 10.1016/j.ijid.2022.11.039 Letter to the Editor Drug-drug interaction with oral antivirals for early treatment of COVID-19 Vuorio Alpo 12⁎ Raal Frederick 3 Kovanen Petri T. 4 1 Mehiläinen Airport Health Centre, Vantaa, Finland 2 University of Helsinki, Department of Forensic Medicine, Helsinki, Finland 3 Faculty of Health Sciences, University of Witwatersrand, Johannesburg, South Africa 4 Wihuri Research Institute, Helsinki, Finland ⁎ Correspondence: University of Helsinki and Mehiläinen Airport Health Centre, 01530 Vantaa, Finland 2 12 2022 2 12 2022 24 10 2022 24 11 2022 28 11 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. Keywords statins atherosclerotic cardiovascular disease COVID-19, nirmatrelvir/ritonavir ==== Body pmcWe read with interest the Danish population-based study estimating the risk of significant drug-drug interactions (DDIs) with the antiviral component nirmatrelvir of the drug combination nirmatrelvir/ritonavir (NMV/r) in the age groups ≥65 years and ≥ 80 years (Larsen, 2022). The study highlights the potentially detrimental effects of DDIs if this antiviral treatment is used as part of polypharmacy in this elderly population at high risk for the progression of SARS-CoV-2 infection to severe COVID-19. Regarding statin treatment and NMV/r DDIs, it was found that 15.45% of the study population in the age group ≥ 65 years used simvastatin or lovastatin and that the respective percentage among ≥80 years was 17.70%. The percentages for atorvastatin in the age groups ≥65 and ≥80 years were 19.91% and 15.85%, respectively. As simvastatin and lovastatin are contraindicated during NMV/r treatment, Larsen (2022) recommends that “patients at low risk of atherosclerotic events could potentially pause the statin treatment during NMV/r administration”. For several reasons, we are concerned about the potential risks for recommending discontinuation of statin treatment in the age group ≥65 years. Firstly, according to current guidelines, statins are not routinely prescribed for patients having a low risk for atherosclerotic cardiovascular disease (ASCVD) (Mach et al., 2020). Moreover, the decision to use statin in older individuals is not evidence-based and, therefore, must be made individually (Strandberg et al., 2014). Secondly, it is very difficult to estimate the true ASCVD risk in older patients, especially among those aged ≥75 years who are free of clinically overt ASCVD (Saeed and Mehta, 2020). Among patients aged <75 years, the efficacy of statins used for primary prevention is well-proven, and the relative risk reduction of major vascular events is about 20-30% for every 1 mmol/L reduction in LDL-cholesterol (Cholesterol Treatment Trialists' (CTT) Collaborators, 2012). Thirdly, in COVID-19 patients, several studies have shown that the use of statins is associated with an improved prognosis (Wu et al. 2021). Additionally, the main protease (Mpro) of the SARS-CoV-2 virus adversely affects microvascular endothelial cells in the brain, and statins may directly inhibit Mpro activity (Vuorio et al. 2022a). Fourthly, withdrawal from statin therapy may acutely worsen the prognosis of patients with non-ST-segment elevation myocardial infarction, even in patients without the additional cardiovascular burden caused by a viral infection (Spencer et al., 2004). In addition, statin withdrawal may easily remain permanent. Based on the above considerations, we recommend that, rather than discontinuing simvastatin treatment, simvastatin should be substituted by either pravastatin or fluvastatin (Vuorio et al., 2022b). By taking into account the prognosis of both COVID-19 and ASCVD, the patient would then be at low risk of side effects from statin therapy and will be guaranteed the best possible benefits from statin treatment. Conflict of Interest AV has received consultancy fees from Amgen and Novartis. PTK has received consultancy fees, lecture honoraria, and/or travel fees from Amgen, Novartis, Raisio Group, and Sanofi. FR has received research grants, honoraria, or consulting fees for professional input and/or lectures from Sanofi, Regeneron, Amgen, and Novartis. Ethical Approval Statement Not applicable. Funding Source Not applicable. Conflict of Interest AV has received consultancy fees from Amgen and Novartis. PTK has received consultancy fees, lecture honoraria, and/or travel fees from Amgen, Novartis, Raisio Group, and Sanofi. FR has received research grants, honoraria, or consulting fees for professional input and/or lectures from Sanofi, Regeneron, Amgen, and Novartis. References Cholesterol Treatment Trialists' (CTT) Collaborators, Mihaylova B, Emberson J, Blackwell L, Keech A, Simes J,et al. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012;380:581-90. Larsen CS. Assessing the proportion of the Danish population at risk of clinically significant drug-drug interactions with new oral antivirals for early treatment of COVID-19. Int J Infect Dis 2022;122:599-601. Mach F, Baigent C, Catapano AL, Koskinas KC, Casula M, Badimon L, et al. 2019 ESC/EAS Guidelines for the management of dyslipidaemias: lipid modification to reduce cardiovascular risk. Eur Heart J 2020;41:111-88. Saeed A, Mehta LS. Statin therapy in older adults for primary prevention of atherosclerotic cardiovascular disease: The balancing act. American College of Cardiology. 2020. https://www.acc.org/latest-in-cardiology/articles/2020/10/01/11/39/statin-therapy-in-older-adults-for-primary-prevention-of-atherosclerotic-cv-disease (accessed 20 October 2022). Spencer FA, Fonarow GC, Frederick PD, Wright RS, Every N, Goldberg RJ, et al. Early withdrawal of statin therapy in patients with non-ST-segment elevation myocardial infarction: national registry of myocardial infarction. Arch Intern Med 2004;164:2162–8. Strandberg TE, Kolehmainen L, Vuorio A. Evaluation and treatment of older patients with hypercholesterolemia: A clinical review. JAMA 2014;312:1136–44. Vuorio A, Kovanen PT, Raal F. Statin needs to be continued during Paxlovid therapy in COVID-19. Clin Inf Dis 2022a:ciac667. Vuorio A, Kovanen PT, Raal F. Cholesterol-lowering drugs for high-risk hypercholesterolemia patients with COVID-19 while on Paxlovid™ therapy. Future Virol 2022b:10.2217/fvl-2022-0060. Wu KS, Lin PC, Chen YS, Pan TC, Tang PL. The use of statins was associated with reduced COVID-19 mortality: a systematic review and meta-analysis. Ann Med 2021;53:874–84. Article to which the letter relates Larsen CS. Assessing the proportion of the Danish population at risk of clinically significant drug-drug interactions with new oral antivirals for early treatment of COVID-19. Int J Infect Dis 2022;122:599-601.	36470504	PMC9715460	NO-CC CODE	2022-12-03 23:20:16	no	Int J Infect Dis. 2022 Dec 2; doi: 10.1016/j.ijid.2022.11.039	utf-8	Int J Infect Dis	2,022	10.1016/j.ijid.2022.11.039	oa_other
==== Front Comput Ind Eng Comput Ind Eng Computers & Industrial Engineering 0360-8352 1879-0550 The Author(s). Published by Elsevier Ltd. S0360-8352(22)00847-6 10.1016/j.cie.2022.108859 108859 Article Supply chain risk management with machine learning technology: A literature review and future research directions Yang Mei ad Lim Ming K. c Qu Yingchi ad Ni Du b Xiao Zhi ad⁎ a School of Economics and Business Administration, Chongqing University, Chongqing 400030, PR China b School of Management, Nanjing University of Posts and Telecommunications, Jiangsu 210003, PR China c Adam Smith Business School, University of Glasgow, Glasgow G14 8QQ, UK d Chongqing Key Laboratory of Logistics, Chongqing University, Chongqing 400030, PR China ⁎ Corresponding author. 2 12 2022 1 2023 2 12 2022 175 108859108859 12 8 2021 13 10 2022 28 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. Coronavirus disease 2019 (COVID-19) has placed tremendous pressure on supply chain risk management (SCRM) worldwide. Recent technological advances, especially machine learning (ML) technology, have shown the possibility to prevent supply chain risk (SCR) by decreasing the need for human labor, increasing response speed, and predicting risk. However, the literature lacks a comprehensive analysis of the relationship between ML and SCRM. This work conducts a comprehensive review of the relatively limited literature in this field. An analysis of 67 shortlisted articles from 9 databases shows that this area is still in the rapid development stage and that researchers have shown extraordinary interest in it. The main purpose of this study is to review the current research status so that researchers have a clear understanding of the research gaps in this area. Moreover, this study provides an opportunity for researchers and practitioners to pay attention to ML algorithms for SCRM during the COVID-19 pandemic. Keywords Machine learning COVID-19 Supply chain risk management Algorithm Research status ==== Body pmc1 Introduction Past outbreaks of infectious diseases, earthquakes, and other natural catastrophic events have put the supply chain at risk on a small scale (Govindan et al., 2020, Kaur and Singh, 2020). However, coronavirus disease 2019 (COVID-19) has caused tremendous and continuous negative effects worldwide. According to a report from the Asian Development Bank, due to the COVID-19 pandemic, global economic losses have reached US$5.8 trillion to US$8.8 trillion, which is equivalent to 6.4–9.7 % of the global gross domestic product (ADB, 2020). The associated panic and lockdown policies have increased hoarding and herding behaviors, shut down resource suppliers, and led to bankruptcies. During the COVID-19 pandemic, people can no longer easily find vegetables, fruit, or toilet paper on supermarket shelves (Frances, 2020, Hobbs, 2021). The shortage of these resources and the surge in demand have left the supply chain vulnerable and triggered supply chain risk (SCR), leading to irreparable losses to enterprises. The COVID-19 pandemic has also caused commodity shortages and aggravated researchers' concerns about supply chain deglobalization. Manufacturers and factory owners have found that they cannot receive a sufficient amount of supplies from domestic suppliers. For example, even if American companies increase the production of mask materials to fight the pandemic, the output of these mask materials by Chinese companies is still 15 times that of American companies (Bradsher, 2020). This situation has prompted managers in various countries to realize the shortcomings of supply chain globalization, which has led to increasing discussions about deglobalization after the COVID-19 pandemic. Moreover, some countries have introduced restrictive economic policies, such as export control, support for domestic enterprise relocation, and the tightening of foreign investment security reviews, during the COVID-19 pandemic. These economic policies exhibit strong economic nationalism and protectionism, which further promote economic nationalism and weaken support for supply chain globalization (Hameiri, 2021; Wang, 2021b). In addition, various relationships have formed a tight risk contagion network within the supply chain (Agca et al., 2021). Once an enterprise experiences a crisis, the risk spillover effect causes other enterprises in the supply chain to suffer losses, thus triggering a chain of SCR disasters (Roukny et al., 2018, Wu et al., 2021). This precarious environment not only destroys the enterprise supply chain system but also causes enormous losses to upstream and downstream enterprises. Therefore, enterprises must implement practical risk management programs to prevent SCR and achieve effective SCR management (SCRM). To implement more effective SCRM programs, leading artificial intelligence technologies, such as machine learning (ML), can provide future predictions for SCRM. ML algorithms performs well in identifying abnormal risk factors and deriving predictive insights from historical datasets (Guo et al., 2021, Mohanty et al., 2021, Uthayakumar et al., 2020). Enterprises can use ML to identify risk factors and predict market demand and possible risk situations (Punia et al., 2020, Wu et al., 2021), which means that when the supply chain encounters challenges in terms of time, cost, and resource constraints, ML is an excellent approach to address them. Because ML has excellent ability in risk identification and assessment, it has made breakthrough in other research fields, such as data mining and image recognition (Li et al., 2020, Saygili, 2021). However, this technology is not frequently used in SCRM. This phenomenon occurs because no ML guidelines are provided for supply chain researchers. Most SCRM researchers are in the field of management, while most ML researchers are in the field of engineering, such as computer science. There is a large gap between the two in terms of subject categories. As a result, supply chain researchers are far less familiar with ML than they are with other aspects, such as mathematical programming and stochastic optimization (Liu et al., 2019, Shahed et al., 2021). There are many types of ML algorithms, but a clear classification of these algorithms is lacking (Janiesch et al., 2021, Xu and Jackson, 2019). Therefore, when researchers attempt to apply ML algorithms to SCRM, they often cannot immediately find the appropriate ML algorithm. The knowledge gap between the two disciplines prevents researchers from making any further attempts. A literature review is necessary for the interdisciplinary integration of the two fields to close this research gap and identify the value of ML for SCRM. Several literature reviews on SCRM-related topics have been performed in recent years (Gurtu et al., 2021; Ho et al., 2015). However, to the best of the author's knowledge, no attempt has been made to examine the SCRM literature in an ML environment. Previous SCRM literature reviews have focused mainly on risk definition (Ribeiro et al., 2018), risk classification (Shahbaz et al., 2019), and risk management strategies (Gurtu & Johny, 2021). These articles have not specifically examined the application of ML in SCRM and conducted a comprehensive analysis. In view of the above shortcomings, this study reviews the academic literature on the application of ML algorithms in SCRM. The purpose of this study is to help researchers and practitioners learn more about the application of ML to SCRM by exploring the state of development of previous research. Moreover, this study analyzes the potential direction of the application of ML algorithms in the SCRM process by evaluating the application status of ML algorithms in SCRM. The present research aims to answer the following questions. (1) What is the current research status of ML algorithms in the SCRM field? (2) Which SCRM processes have used ML thus far, which algorithms have been used, and how effective have these algorithms been? (3) What are the current research gaps in ML algorithms in SCRM, and which areas need more attention? To answer the above questions, this study conducts a comprehensive analysis of the research situation of ML algorithms in SCRM. The rest of this study is organized as follows. The second section is the methods of the literature review. The third section describes the results of bibliometric analysis realized by computer software. The fourth section introduces commonly used ML algorithms and their application scenarios in detail. The fifth section describes the application of ML in SCRM. The sixth section presents future research directions. Finally, the seventh section presents the conclusions of this study. 2 Methods To allow researchers to have a clearer understanding of the analytical process of this paper, this section describes the article selection process and provides the descriptive analysis of the selected article. 2.1 Article selection process In this study, we obtained articles from major journal databases to cover relevant fields as comprehensively as possible. There are mainly 6 steps. First, as indicated in Fig. 1 , this study used nine academic databases, namely, Science Direct, Scopus, Wiley Online Library, Springer Nature, JSTOR, Taylor & Francis, IEEE Xplore, Emerald, and Google Scholar. Second, we defined the search terms. The keywords used in the search were “machine learning” and “supply chain risk”. Third, to fully include the reviewed articles on the target topics, the search scope of this study did not take into account publication time limits and limited the end date of the article search to June 30, 2021. In other words, we collected all articles on ML algorithms and SCRM published in these authoritative databases before June 30, 2021. The number of targeted articles amounted to 48,000 articles. The search results were stored in the Research Information Systems (RIS) format and included all important article information, such as the title, author name, abstract, keywords, and references. Fourth, to achieve the highest level of relevance, only articles written in English and published in international journals were selected, thus excluding conference papers, editorials, textbooks, master's theses, etc., from the analysis. And to ensure that we captured every study relevant to ML and SCRM, we did not impose restrictions on the journal lists regardless of the journal in which an article was published. This selection approach reduced the number of articles collected to 8,000. Fifth, we identified and used several criteria to filter the articles. We examined abstracts and keywords to determine whether they covered one or more SCRM topics, including SCR factors and risk management processes. If an article did not meet one of these filtering criteria, it was excluded. We also carefully evaluated the list of references of the shortlisted articles to ensure that no other relevant articles were omitted from the search. The content of each article was thoroughly reviewed to ensure that it was studied in an ML environment, and 93 articles remained. Finally, we screened out duplicate articles that appeared simultaneously in multiple databases for a total of 67 articles.Fig. 1 Paper selection process. 2.2 Descriptive analysis To more clearly display the 67 articles selected for this study, this section conducts a descriptive analysis of annual publications, regional distribution, and research categories. 2.2.1 Annual publications As mentioned, 67 articles were selected in our analysis and were written over a 14-year period (2008–2021). Although ML concepts have been emerging since the late 1990 s, there have been fruitful results in other areas (Li et al., 2020a). However, ML algorithms have seldom been used in SCRM. As shown in Fig. 2 , the first research paper on the application of ML in SCRM was published in 2008. There was a small spike in 2016, which was billed as the “hottest year for machine learning” in history, indicating that researchers were beginning to realize that the use of ML would make supply chains smarter and able to perform well beyond human limitations. Since 2018, researchers who have realized the benefits of ML have further applied it to SCRM, and relevant articles have shown a trend of a rapid increase in the number of such applications. It is worth noting that in the first half of 2021 alone, the number of related articles reached the number of articles published in 2020, showing that the new insights and knowledge that ML brings to enterprise supply chain operations are revolutionizing the SCRM approach.Fig. 2 Annual publication trend. 2.2.2 Regional distribution Determining the geographical distribution of articles can help to clarify the situation of researchers conducting SCRM studies in each region. This section analyzes the geographical distribution of the author's correspondence addresses of all articles. For each affiliation, we obtained the city in which the organization was based and used GPS Visualizer to visually depict these city coordinates. As shown in Fig. 3 , existing ML studies in SCRM are concentrated mainly in Southeast Asia, Western Europe, and eastern North America. Because North America and Western Europe are famous supply chain centers, they contain most of the world's developed economies. The Southeast Asian countries that have benefited from the globalization of the supply chain due to their unique geographical advantages, coupled with their dual advantages of population and resources, have begun to transfer many labor-intensive industries to other Southeast Asian countries. The supply chain transfer of many companies has promoted Southeast Asia as an indispensable and important link in the global market. Therefore, these well-known global supply chain centers are taking the lead in using emerging technologies such as ML to solve SCRM problems to better prevent and mitigate the adverse effects of risk events.Fig. 3 Geographic distribution of contributing organizations. 3 Research categories The analysis of research categories on this topic can help researchers clearly understand the research situation in this field. The specific research categories are shown in Fig. 4 . SCR can exist in all business activities, and supply chain enterprises are interdependent. Once a crisis occurs in one enterprise, it can spread and affect other enterprises. Although ML and the supply chain seem to be two unrelated disciplines, ML can realize a more intelligent SCRM by learning the data information in the supply chain through computer programs. Therefore, the most common research direction in this field is computer science interdisciplinary applications. Because interdisciplinary research transcends the previous categorical research methods, it can address the integration of problems to achieve better decision-making effects. The second most common research direction is operations research management science, which is closely related to the supply chain. This direction combines the optimization methods commonly used in supply chains with ML to guide production decisions. Because sophisticated ML tools can help managers predict production requirements accurately, they can also prevent inappropriate requirement assessments from causing managers to make incorrect decisions (Ni, 2019, Priore et al., 2019). It is worth noting that although the benefits of ML for SCRM are beginning to be recognized by many practitioners and managers, some business processes can exhibit misleading or missing information. Some algorithms cannot exclude incorrect information, leading to distorted learning results. Therefore, some researchers have focused on computer science artificial intelligence and are committed to writing more advanced and effective algorithms to improve ML algorithm accuracy.Fig. 4 Research categories and number of published articles. 4 Bibliometric analysis To make the analysis of this paper more objective and scientific, this section uses computer software to quantitatively analyze the articles. Software, such as VOSviewer, is widely used in bibliometric analyses and has gradually become an important way for many scholars to conduct literature reviews. This section uses software to conduct bibliometric analysis, including citation and keyword analyses, to obtain convincing results. A description of this analysis is presented below. 4.1 Publication sources An analysis of the journals in which the reviewed articles are published is helpful for understanding the journal acceptance of papers in the field. Journals in various fields (e.g., computing, management, and production) publish articles on SCRM and ML algorithms. A total of 54 journals are included in this review. Fig. 5 shows the top 10 contributing journals from the 67 articles reviewed. Only 9 of the 67 journals have published more than two articles. These journals are scattered in the fields of management, computers, expert systems and other areas, indicating that there is no specific trend of journal publication in this field. Supply chain researchers have just begun to note the benefits of ML algorithms and have attempted to solve the SCRM problem with ML algorithms. Therefore, the International Journal of Production Research, European Journal of Operational Research and other management journals related to the supply chain appear in the top five journals. Moreover, it is possible that some ML algorithm researchers are very interested in SCRM problems. Their views on the field have appeared in several computer and interdisciplinary journals, such as the Journal of Intelligent & Fuzzy Systems and Computers & Industrial Engineering. In general, the concerns of supply chain and ML algorithm researchers in this field provide guidance in terms of solving crises and drive other researchers and practitioners to note the advantages of ML algorithms and apply them in practice.Fig. 5 Journal sources. 4.2 Citation analysis A citation is considered an important indicator to measure the influence of an article and to determine the acceptability of research in the field. Table 1 shows the ten most frequently cited articles, including their cumulative and annual number of citations in the Web of Science (WOS) based on how often a particular article in the WOS is cited by other articles (the search scope includes the Science Citation Index, Social Science Citation Index, and Arts and Humanities Citation Index). These articles average approximately 13 citations per year. The most cited study is Sun et al. (2008), which has been cited 279 times. The second most cited article, Yu et al. (2011), has been cited 106 times. The third most cited article is the study by Garvey et al. (2015), with 97 citations in total. It is important to note that the article with the highest number of citations was published in 2008. However, during the ten-year time period of this study, no article received more than 300 citations. These articles are scattered in the fields of management, computer science and other areas across many disciplines and are not concentrated in a specific field, indicating that ML research in SCRM did not become a mainstream research direction in the first ten years of the study period. Moreover, most of the articles cited in the top 10 were published in 2015, 2017, 2019 and 2020, indicating that although ML and SCRM have been studied for a long time, the field has only recently gained attention. This may be due to the explosion in popularity of ML since 2016. In particular, the study of Nikolopoulos et al. (2021) was published in 2021 but has been cited 18 times on average, indicating that the application of ML algorithms in SCRM is attracting great interest among researchers.Table 1 Top 10 cited articles. R Paper title Publication source Year TC TC/Y 1 Sales forecasting using extreme learning machine with applications in fashion retailing Decision Support Systems 2008 279 21.46 2 An intelligent fast sales forecasting model for fashion products Expert Systems with Applications 2011 106 9.64 3 An analytical framework for supply network risk propagation: A Bayesian network approach European Journal of Operational Research 2015 97 13.86 4 A supervised machine learning approach to data-driven simulation of resilient supplier selection in digital manufacturing International Journal of Information Management 2019 55 18.33 5 A data mining-based framework for supply chain risk management Computers & Industrial Engineering 2020 41 10.25 6 Forecasting SMEs' credit risk in supply chain finance with an enhanced hybrid ensemble machine learning approach International Journal of Production Economics 2019 36 12 7 Banking credit worthiness: Evaluating the complex relationships OMEGA - The International Journal of Management Science 2019 35 11.67 8 Comparison of individual, ensemble and integrated ensemble machine learning methods to predict China's SME credit risk in supply chain finance Neural Computing & Applications 2017 31 6.2 9 A credit risk assessment model based on SVM for small and medium enterprises in supply chain finance Financial Innovation 2015 25 3.57 10 Forecasting and planning during a pandemic: COVID-19 growth rates, supply chain disruptions, and governmental decisions European Journal of Operational Research 2021 18 18 Note: R: Ranking based on the total number of citations per paper. TC: Total number of citations per paper. TC/Y: Total number of citations per paper/year. 5 Keyword analysis Coword analysis is a content analysis method based on the extant literature and is a common bibliometric method. Because the research content of each document cannot be summarized with a single word, multiple keywords are needed. At the same time, keywords in the same field are roughly limited to a certain range, and different kinds of literature often have similar keywords. Therefore, the coword analysis of high-frequency keywords can reflect the development trend of the research field. In this paper, VOSviewer was used to extract and analyze 523 keywords from 67 publications. Fig. 6 shows the co-occurrence network of keywords in recent years and indicates that researchers in the first few years had not yet formed a specific concept of ML algorithms. This is because some researchers find ML to be a better tool when performing SCRM, and these researchers choose this tool for risk management. Therefore, the application of the ML algorithm in SCRM is distributed across various industries, such as engineering, logistics, and the food industry. In fact, in the past two years, keywords such as “machine learning”, “supply chain risk”, “forecasting” and “risk assessment” have begun to receive widespread attention, showing that researchers in previous years did not know enough about the benefits of ML algorithms in SCRM and have only recently fully realized that ML algorithms can help business managers deal with supply chain crises. However, with the manifestation of this advantage, researchers and practitioners have developed stronger requirements for risk management and have begun to actively integrate new technologies for its implementation. For example, the new technology blockchain combined with ML algorithms has begun to be applied to SCRM decisions.Fig. 6 Co-occurrence network of keywords. 6 Commonly used ML algorithms By analyzing the articles in this study, we find that 11 out of 40 commonly used ML algorithms are applied to SCRM (Sarker, 2021). Specifically, these ML algorithms include deep learning (DL), support vector machine (SVM), decision tree (DT), neural networks (NN), bayesian network (BN), logistic regression (LR), random forest (RF), ensemble learning (EL), clustering, extreme learning machine (ELM) and naive bayes (NB). These ML algorithms play very important roles in reducing risk and improving supply chain performance. To provide SCRM researchers with a clear introduction to ML algorithms, the characteristics of these 11 ML algorithms based on frequency are shown below. Deep learning. The DL algorithm has recently become the most popular ML algorithm (Du et al., 2016). The DL algorithm originated from artificial NN. Its main idea is to simulate the structure of the human brain so that it can intelligently calculate (Schmidhuber, 2015). The most distinctive advantage of the DL algorithm is to quantify unstructured data such as images, sounds and text compared with other ML algorithms. In addition, compared to other ML algorithms that rely on feature selection, DL sometimes does not require feature engineering. It can directly learn efficient data representations from raw data, which saves the effort of manually constructing features. In a supply chain, the DL algorithm can detect risk information before a supply chain crisis through the automatic learning of comments, news and other unstructured data to achieve management in advance (Lee et al., 2021). Neural networks. The NN algorithm has been a research hotspot in artificial intelligence since the 1980 s (Anderson, 1995). It has outstanding performance in processing nonlinear problems (Jordan et al., 2015). Compared with other algorithms, it finds the optimal solution with a higher accuracy. Moreover, the NN algorithm has good fault tolerance; the misfunction of one neuron does not affect the operation of the entire algorithm. The current supply chain environment is changing quickly because of COVID-19, technologies and policies. Enterprises must assess possible risk in advance to minimize the negative consequences of unexpected changes. In the process of risk assessment, the NN algorithm obtains the optimal solution by continuously adjusting the weights, and the predicted risk coefficient is very close to the actual risk coefficient (Lu, 2021). Support vector machine. The SVM algorithm was developed by Dr. Vapnik et al. (1995). It maps the classification problem to a higher dimension space through the kernel function, which has a definite solution, unlike other black box algorithms. In particular, its optimization goal is to minimize the structural risk instead of experiencing minimization. Thus, the SVM algorithm shows unique advantages in solving small sample, nonlinear and high-dimensional problems and overcomes the dimension problem of the traditional statistical methods. Therefore, SVM is widely used in classification and prediction problems (Kianmehr and Alhajj, 2006, Ni et al., 2018). In the supply chain, the SVM algorithm can process small sample data in supply chain finance, which effectively reduces the probability of classification errors and improves the accuracy of risk prediction (Zhang et al., 2015). Clustering. The clustering algorithm is a typical unsupervised learning algorithm (Jain, 2010). It does not have clear requirements for output data from the clients’ side and mainly classifies data with more similarity into one category according to the sample distance (Zhang et al., 2018). In SCRM, researchers mainly use KNN and K-means algorithms for clustering. The KNN algorithm can classify suppliers' on-time delivery and assign orders to suppliers with the highest on-time probability without requiring client data (Cavalcante et al., 2019). Moreover, the K-means algorithm is used to address the complex relationships among the characteristics of the agricultural supply chain and to detect the core factors to improve farmers' credit level (Bai et al., 2019). Bayesian network. The BN algorithm is a network structure that describes the dependencies between variables with conditional probability (Friedman et al., 1997). It can directly calculate the probability of the posterior between variables without any assumptions. In addition, the BN algorithm reveals the causal relationship between variables by extracting the most relevant features (Gopnik et al., 2007). In the supply chain, previous risk measures were almost specific to a particular problem and lacked a standard. Therefore, many researchers have ignored their own network structure and lack of measurement on risk transmission throughout the network (Garvey et al., 2015). However, the BN algorithm can describe the interdependence of different risks and quantify risk transmission in the supply chain (Chu et al., 2020). Random forest. The RF algorithm is a classifier that contains multiple DTs, and its output is determined by the random selection of individual trees (Breiman, 2001). Instead of simply averaging the predictions of all trees, the algorithm combines a certain number of DTs. It obtains the prediction results by averaging the predictions of each tree. The RF algorithm can provide an effective method to deal with unbalanced datasets by randomly selecting trees. In the supply chain, since crisis events are rare, there may be an imbalanced distribution of datasets. The RF algorithm can better address the imbalance of supply chain data and improve the accuracy of risk prediction (Brintrup et al., 2020). Decision tree. The DT algorithm is a traditional data mining method to display the choices and probability in the training set by constructing a DT (Coadou, 2010), which generates multiple branches and child nodes with different probabilities. In practical applications, most studies have focused on the predictive performance and ignored the importance of interpretability. Therefore, the DT algorithm can help supply chain practitioners understand the results of risk assessment and make decisions to mitigate or prevent risk (Baryannis et al., 2019, Priore et al., 2019). In addition, it is easy for DTs to cooperate with other algorithms to build a powerful model for risk prediction in supply chain management. For example, Zhu et al. (2017b) used the DT algorithm as a basic classifier to predict credit risk in the supply chain. Because of the powerful model performance, it can help financial institutions improve cash flow. Ensemble learning. The EL algorithm is not an individual ML algorithm but combines several individual models (Dietterich, 2000). The EL algorithm is mainly used to improve the performance of a model by choosing the best result of the basic classifiers. In the supply chain, when enterprises cannot obtain supply chain data information in a timely manner, this can cause the algorithm to misclassify. In addition, the EL algorithm can combine multiple classifiers to increase the fault tolerance of the model (Islam et al., 2020b). Logistic regression. The LR algorithm is generated from a linear regression, but the LR algorithm introduces nonlinear factors through the sigmoid function; therefore, it can easily handle the relationships among variables (Gillet et al., 2011). The feature data can be continuous, classified variables. Moreover, LR allows these variables to have dummy variables. The LR algorithm is currently a commonly used ML algorithm in industry and is mainly used to estimate the possibility of latency. In the supply chain, the LR algorithm estimates the impact of a bank's potential risk factors on loans by considering the nonlinear relationships among assets, cash flow and other characteristics (Ying et al., 2020). Extreme learning machine. The ELM algorithm is a new fast-learning algorithm (Huang et al., 2004). This algorithm is a single-layer NN, and its greatest characteristic is that it can randomly initialize input weights and bias and obtain corresponding output weights at a faster speed (Huang et al., 2012, Huang et al., 2006). Because there are many uncertain factors in the supply chain, such as supply and demand, managers especially need to quickly understand consumer’ demand. Thus, they can use the ELM algorithm to predict demand data quickly, which greatly reduces supply and demand uncertainty (Yu et al., 2011). Naive bayes. The NB algorithm is a supervised ML algorithm, and each training sample requires a corresponding label (Murphy, 2006). For a given item to be classified, this algorithm classifies the category into the highest possibility. Compared to other classification methods, it performs best when features are less related. In practical applications, the NB algorithm can be used to classify real and fake chips with low correlation in the supply chain to reduce a product’s flaw rate (Frazier et al., 2018). In addition to the characteristics of ML algorithms in the SCRM process introduced above, the number of times that they are used is also described in Fig. 7 , which shows the number and frequency of ML algorithm use in the articles reviewed. The DL algorithm is the most commonly used algorithm for the period from 2008 to 2021, with a 17.5 % frequency. Because of recent improvement in calculation ability, DL has more stages to perform in SCRM. It can not only process numerical information but also interpret unstructured data such as images, sound and text, which traditional algorithms cannot handle. The NN and SVM algorithms are used 18 and 17 times, respectively. This may be because many articles set them as benchmarks, but in SCRM, the good fault tolerance rate of the NN algorithm (Lu, 2021) and the advantage of the SVM algorithm in handling small sample datasets (Zhang et al., 2015) cannot be ignored. Additionally, other new algorithms, such as ELM, are used less frequently. However, the lack of the application of less-used algorithms does not mean that they are meaningless for SCRM, as they can become a niche to supply chain researchers.Fig. 7 Algorithm usage. 7 Application of ML in SCRM From the analysis of the commonly used ML algorithms, the potential of ML to solve complex problems in SCRM has not been fully explored. However, the benefits of ML for SCRM have indeed been fully recognized. Based on the analysis in the previous three sections, we find that ML algorithms are mainly applied in the supply chain for risk identification, risk assessment, risk mitigation, risk monitoring, etc. The specific categories can be found in Table 2 . In this section, we describe which SCRM processes require ML most, and the specific details are shown below.Table 2 Application of ML in SCRM. Risk identification Risk assessment Risk mitigation Risk monitoring NN (Feng et al., 2021; Han et al., 2021; Lau et al., 2021, Sun, 2021, Sun et al., 2008) (Feng et al., 2021, Han and Zhang, 2021, Lau et al., 2021, Lu, 2021, Ni, 2019, Sang and o. C., & Mathematics, A. , 2021, Wang, 2021a, Wang, 2021, Zhang et al., 2015, Zigiene et al., 2019) (Chatzidimitriou et al., 2008, Han and Zhang, 2021, Piccialli et al., 2021, Punia et al., 2020, Rishehchi et al., 2021, Sun et al., 2008, Wichmann et al., 2020) (Priore et al., 2019) DL (Handfield et al., 2020, Thota et al., 2020, Xuan, 2021) (Abdollahnejadbarough et al., 2020; Hajian Heidary et al., 2019; Handfield et al., 2020; Li et al., 2020b; Liu et al., 2020b) (Chatzidimitriou et al., 2008; Fu et al., 2019; Piccialli et al., 2021, Punia et al., 2020, Wichmann et al., 2020) (Goldberg et al., 2020, Lee et al., 2021, McClements et al., 2021, Palander et al., 2018, Wang et al., 2020) SVM (Cao et al., 2016; Li et al., 2021) (Liu & Huang, 2020b; Sang & Mathematics, 2021; Sun, 2020, Zhang et al., 2020, Zhang et al., 2015, Zigiene et al., 2019) (Baryannis et al., 2019, Berloco, et al., 2021, Cao and Zhang, 2016, Chatzidimitriou et al., 2008, Frazier et al., 2018, Fu and Chien, 2019, Piccialli et al., 2021, Wichmann et al., 2020; Zhang et al., 2021a) (Priore et al., 2019, Zhang et al., 2020) DT (Li et al., 2021, Zhu et al., 2019) – (Baryannis et al., 2019, Berloco, et al., 2021, Rishehchi et al., 2021, Zhu et al., 2016, Zhu et al., 2017a, Zhu et al., 2019) (Priore et al., 2019) BN (Bouzembrak et al., 2019; Brintrup et al., 2020, Chu et al., 2020; Feuerriegel et al., 2019; Marvin et al., 2020, Rodgers et al., 2020) (Feuerriegel and Gordon, 2019, Garvey et al., 2015; Rodgers & Singham, 2020) (Bouzembrak and Marvin, 2019, Brintrup et al., 2020, Frazier et al., 2018, Ji et al., 2017, Kouadio et al., 2021, Marvin et al., 2020) – LR (Brintrup et al., 2020, Li et al., 2021, Merve et al., 2020, Ying et al., 2020) (Cavalcante et al., 2019, Frey et al., 2018, Merve et al., 2020, Sun, 2020) (Berloco, et al., 2021, Brintrup et al., 2020, Merve et al., 2020, Rishehchi et al., 2021) (Palander et al., 2018, Priore et al., 2019, Ying et al., 2020) RF (Brintrup et al., 2020, Feuerriegel and Gordon, 2019, Li et al., 2021) (Feuerriegel and Gordon, 2019, Zigiene et al., 2019) (Berloco, et al., 2021, Brintrup et al., 2020; Islam & Amin, 2020b; Kouadio et al., 2021, Piccialli et al., 2021, Punia et al., 2020, Rishehchi et al., 2021) – EL (Feuerriegel and Gordon, 2019, Zhu et al., 2019) (Feuerriegel & Gordon, 2019; Liu & Huang, 2020b) (Chen et al., 2020; Islam & Amin, 2020b; Zhu et al., 2016, Zhu et al., 2017a, Zhu et al., 2019) (Priore et al., 2019) KNN (Brintrup et al., 2020, Merve et al., 2020) (Cavalcante et al., 2019, Merve et al., 2020) (Berloco, et al., 2021, Brintrup et al., 2020, Chatzidimitriou et al., 2008, Frazier et al., 2018, Fu and Chien, 2019, Merve et al., 2020, Nikolopoulos et al., 2021, Zhao et al., 2020) (Priore et al., 2019) Clustering (Bai et al., 2019, Nayal et al., 2021, Ying et al., 2020) (Bai et al., 2019, Belhadi et al., 2021, Nayal et al., 2021) (Coyle et al., 2016, Nikolopoulos et al., 2021) (Priore et al., 2019, Ying et al., 2020) ELM (Sun et al., 2008) – (Sun et al., 2008, Yu et al., 2016, Yu et al., 2011) NB – – (Rishehchi et al., 2021) (Priore et al., 2019) 7.1 Risk identification As the first step of SCRM, the correct identification of potential risks is very important for enterprises to actively avoid losses. However, in previous studies, the Delphi method, system analysis, and interviews were used by practitioners and researchers to identify risks. In fact, most of these empirical methods cannot accurately quantify risk probability. In this case, ML algorithms can autonomously identify risk factors and quantify risk through mathematical modeling. For example, Li et al. (2021) used the RF algorithm to identify risk factors such as web page color, information, and some redundant risk factors in the online service supply chain. The RF algorithm reduces the time overhead for suppliers. With the rapid development of internet technology, consumers tend to express their true opinions about products and services on social media. This network information is often more authentic than the information obtained from questionnaires (Liu and Zhang, 2020, Liu and Huang, 2020). Meanwhile, these network data are normally unstructured, with a low value density, and are large in both variety and quantity. Therefore, it is difficult to identify risk easily with these data. However, ML algorithms can use computers to process unstructured data such as text and voice data and extract valuable information from them. For example, Ying et al. (2020) used the DL algorithm and 4,014 bank evaluation reports to analyze the risk situation of bank financial service providers. In this way, they identified four important risk management factors, such as supply chain collaboration and cash flow monitoring, to achieve economically optimal risk management. In addition, Chu et al. (2020) proposed an SCRM framework based on text mining and used the BN model combined with news to classify risk and guide supplier selection. In general, the ML algorithm can be used in the first step of SCRM to determine the risk factors, which can prevent potential risks in advance. 7.2 Risk assessment Accurate risk assessment helps managers take timely actions to reduce SCR. However, risk assessment generally relies on experienced practitioners. In the actual evaluation, practitioners’ experience and bias can have a great impact on the evaluation accuracy. This evaluation lacks objectivity and sometimes cannot be repeated. However, ML by computers, instead of inferences from experience, provides neutrality and objectivity. For example, Lu (2021) used the NN algorithm to build a multilevel supply chain inventory control model. The risk coefficient estimated is close to the actual risk coefficient, which effectively improves supply chain efficiency. In addition, the supply chain is a dynamic system. Static evaluation methods, such as the fuzzy evaluation method and analytic hierarchy process, cannot accurately describe the dynamic complexity. Therefore, researchers urgently need a tool to overcome the shortcomings of subjectivity. ML can fit the complex characteristics of the supply chain. For example, Sang and Mathematics (2021) used SVM and back-propagation NN to evaluate banks’ multiple risk characteristics, such as policy, operation and credit risks. This approach greatly reduces the probability of negative effects on bank profits. It can be seen from the above factors that the ML algorithm can objectively and quickly assess risk events. 7.3 Risk mitigation The modern supply chain is faced with unavoidable risks while it must meet customer demand in a timely manner. Natural disasters, sudden equipment failures, transportation delays, and international policies can challenge the expected lead time and cost for enterprises. People around the world have been affected by the shortages caused by COVID-19, which have posed enormous challenges to the supply chain. Therefore, the estimation of risk in advance is more effective than remedying them after they occur. The ML algorithm is considered an excellent method that can provide effective advice for SCRM. For example, Nikolopoulos et al. (2021) correctly predicted excessive demand for daily necessities and electronic products during COVID-19 using the DL algorithm, which can help policy makers make better decisions during current pandemics. In addition, Berloco et al. (2021) used the EL, RF, and SVM algorithms to capture the risk spillover in supply chain. Similarly, Zhao et al. (2020) used the KNN algorithm to predict the legitimacy of online drugstores with 98.6 % accuracy. Accurate prediction helps government departments crack down on illegal drugstores and provides patients with reliable choices. In general, the ML algorithm can help enterprises develop appropriate strategies to avoid risk and improve the robustness of the supply chain. 7.4 Risk monitoring Risk monitoring refers to taking regular measures to detect adverse events in the supply chain. Compared with the studies on risk identification and risk assessment, the research on risk monitoring is fewer in terms of the number of papers. However, the small amount of studies in this area does not mean that they are meaningless; rather, it is possible that not enough advanced technology has been found to replace risk monitoring. In practice, supply chain enterprises often need to recruit a large number of workers to carry out risk monitoring. Some products even require 24/7 inspections to prevent sudden equipment failures. ML can also make the daily monitoring of the production process intelligent. For example, Zhang et al. (2021a) adopted an SVM algorithm to monitor purchased fake electronic components, with an accuracy of 100 %. This approach prevented fake electronic components from flowing into the subsequent production process. In addition, it is often impossible to immediately perceive the occurrence of abnormal conditions. For example, when monitoring hazardous products, workers often observe the surface of the products, but it is difficult to monitor the quality of the products with only eyes. ML can help enterprises automate risk monitoring with “intelligent eyes”. For example, Goldberg et al. (2020) proposed a safety monitoring system for the food industry by using the DL algorithm. Through real-time monitoring, the algorithm was able to discover information on foods that consumers consider dangerous and detect dangerous products with a high accuracy of 90 %. Similarly, Lee et al. (2021) used the DL algorithm and blockchain technology to conduct real-time data monitoring on Twitter, which helps enterprises share emergency information and achieve risk management in advance. In general, the ML algorithm can reduce the costs of manual monitoring. In this study, Table 2 shows the distribution of the ML algorithm in the above four SCRM processes. As shown in the table, ML algorithms significantly favor risk mitigation processes. Risk mitigation contains the largest number of articles in the four SCRM processes, followed by risk assessment, risk identification and risk monitoring. Researchers are most concerned with risk mitigation. They are not limited to mining risk information from the supply chain’s internal data but actively use advanced ML algorithms to mine valuable information from external data to actively avoid risks (Feuerriegel & Gordon, 2019). Notably, ML algorithms do not apply to just one SCRM process. The application of the ML algorithm in a supply chain can achieve many benefits. For example, researchers can predict consumer demand while mining potential risk factors (Marvin et al., 2020, Merve et al., 2020). This section not only helps managers identify potential risk factors, but also provides a guide and map for researchers and practitioners with appropriate references. 8 Future research directions In this section, we summarize the potential future research directions based on previous analyses of ML algorithms and SCRM. Specifically, this section mainly includes the four points below. Enhancement of ML interpretability. From the analysis of Fig. 1, Fig. 2, Fig. 3 in Section 2, compared with the period prior to 2016, the article publication trend has risen sharply in recent years. It demonstrates that supply chain managers have recognized the benefits of ML algorithms in SCRM and have been actively involved in research in this field. However, judging from the total number of articles, which is only 67, there is not much application in this field. This can be because even though ML algorithms are becoming mainstream research tools, they still confuse many supply chain practitioners and researchers, and these algorithms create a black box that can be understood only by designers (Weinan, 2020). Many researchers argue that ML algorithms from data input to result comprise a very confusing process (Handelman et al., 2019). Specifically, in the field of high-dimensional prediction, black-box models represented by the NN and EL algorithms can more accurately predict complex datasets than white-box models such as DT (Zhang et al., 2021b). Thus, although ML algorithms in risk assessment or risk prediction achieve accurate results, the lack of interpretability is unacceptable for managers with no technical background. Since 2019, the explainable AI (XAI) has become a heated topic (Fernandez et al., 2019). The sensitivity (how results change according to input) has been explained by the Shapley value of game theory for tree models (Lundberg et al., 2020). Moreover, more ML algorithms have been recently explained by XAI (Tallón-Ballesteros and Chen, 2020, Zhang, et al., 2021b). Therefore, researchers can consider building XAI algorithms by combining visualization, proxy models, knowledge graphs, etc., to enhance managers’ trust in the algorithms’ results in the future. More advanced technologies. From the analysis in Section 3, we find that some of the latest technologies, including blockchain, have been introduced into SCRM. In practice, although companies in the supply chain have long-term partnerships, they also compete with one another. These companies sometimes deliberately misreport data because of self-interest (Islam et al., 2020a). Entering this inaccurate data can mislead algorithms. Moreover, these new technologies, such as blockchain, are believed to help increase the transparency of supply chain enterprises. Notably, some studies suggest that these techniques may present data security and privacy risks (Narwane et al.; Vafadarnikjoo et al.). Therefore, in future research, we must give more attention to the safety assessment of new technologies and then apply them to SCRM. Involvement of more external data. From the analysis in Section 4, we confirm that ML algorithms can process unstructured data such as texts and comments. With the technological development, from the enterprise’s perspective, any enterprise can focus not only on itself but also on the external environment, such as the market, competitors, and partners. Therefore, in addition to obtaining risk information from internal supply chain data, external data such as the finance status, online news, social media comments of the market, competitors, and partners should also be considered. Moreover, these data can help quickly detect risky behaviors and give early warnings. However, the current studies that process these data give more attention to the algorithm’s effectiveness, such as its accuracy, recall and other indicators, instead of applying these indicators appropriately. Therefore, in future research, researchers and practitioners can consider exploiting the advantages of ML algorithms in processing external data and improving the efficiency of the algorithm. Increased targeting of COVID-19. The impact of the COVID-19 pandemic has made supply chain practitioners deeply aware of the importance of SCRM. From the analysis in Section 5, it is found that with the development of ML, blockchain and other technologies, enterprises can combine these advanced technologies to realize intelligent risk management. Specifically, enterprises can monitor supply chain status in real time through the ML algorithm and then adjust supply chain inventory and production plans. In this way, the supply chain can start a timely emergency plan. Moreover, enterprises can evaluate suppliers through the ML algorithm in advance, which decreases supply chain disruptions. Furthermore, since the COVID-19 outbreak, researchers have given more attention to supply chain interruption, but few scholars have considered supply chain production in an emergency. When COVID-19 first emerged, there was an unexpected surge in the demand for masks, and original mask manufacturers were faced with great pressure. At the same time, some automobile manufacturers also began to produce masks to meet the surge in requirements (Park et al., 2020). Therefore, in future risk management, enterprises can continuously enhance their flexibility with the support of new technologies such as ML and blockchain to deal with “black swan” events. 9 Conclusions This study systematically reviews the current state of the development of ML algorithms in SCRM and answers the three questions raised in Section 1. This study collected 67 articles on the application of ML algorithms in SCRM published by 9 authoritative databases as of the first half of 2021. 2, 3 provide a comprehensive analysis of the years, journals, citations, keywords, etc. and answer the research status of ML algorithms in the SCRM field in question 1. 4, 5 respond to question 2 and show that ML algorithms have been applied to SCRM, such as for risk identification, risk assessment, risk mitigation and risk monitoring, and have proven to be an excellent tool to help supply chain managers better deal with risk. Section 6 gives the future research directions in question 3 that require more attention to ML interpretability, more advanced technologies and external supply chain data, etc. Furthermore, from the analysis in Section 6, it can be concluded that the impact of the COVID-19 pandemic has made supply chain practitioners deeply aware of the importance of SCRM. Enterprises can continuously enhance their flexibility with the support of new technologies, such as ML and blockchain, to better respond to the changes in the supply chain environment brought about by the COVID-19 pandemic. There are still limitations to this study that need to be addressed before concluding. First, although a total of 9 relevant scholarly databases were included for article screening, this study reviewed only international journal articles written in English and excluded conference papers, textbooks, and unpublished articles and notes. Second, this study was based on a keyword search, so the results are limited to keyword combinations, and some work may not have been captured by the keywords. Despite these limitations, this study summarizes the current research status of this field. For researchers, this study demonstrates the application of various algorithms in this field and can provide a reference for choosing further research topics. For practitioners, this study provides an opportunity to understand how ML algorithms can benefit SCRM, helping them better respond to the supply chain changes caused by the COVID-19 pandemic. CRediT authorship contribution statement Mei Yang: Investigation, Writing – original draft, Validation, Visualization. Ming K. Lim: Conceptualization, Project administration, Validation. Yingchi Qu: Validation, Visualization. Du Ni: Conceptualization, Investigation, Validation. Zhi Xiao: Conceptualization, Methodology, Project administration. 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. Data availability Data will be made available on request. Acknowledgement This work was supported by the National Natural Science Foundation of China [grant No. 72071021, 71671019], the Graduate Research and Innovation Foundation of Chongqing, China [grant No. CYS21047], and the Chongqing Social Science Planning Program [grant No.2022NDQN44]. The authors declare no competing interests. ==== Refs References Abdollahnejadbarough H. Mupparaju K.S. Verizon Uses Advanced Analytics to Rationalize Its Tail Spend Suppliers Interfaces 50 3 2020 197 211 ADB. (2020). COVID-19 Economic Impact Could Reach $8.8 Trillion Globally — New ADB Report. Asian Development Bank. Agca, S., Babich, V., et al. (2021). Credit Shock Propagation Along Supply Chains: Evidence from the CDS Market. Management Science. Anderson, J. A. (1995). An introduction to neural networks: MIT press. Bai C.G. Shi B.F. Banking credit worthiness: Evaluating the complex relationships Omega-International Journal of Management Science 83 2019 26 38 Baryannis G. Dani S. Predicting supply chain risks using machine learning: The trade-off between performance and interpretability Future Generation Computer Systems-the International Journal of Escience 101 2019 993 1004 Belhadi A. Mani V. Artificial intelligence-driven innovation for enhancing supply chain resilience and performance under the effect of supply chain dynamism: An empirical investigation Ann. Oper. Res. 2021 1 26 Berloco, C., de Francisci Morales, G., et al. (2021). Predicting corporate credit risk: Network contagion via trade credit. PLoS ONE, 16(4 April). Bouzembrak Y. Marvin H.J.P. Impact of drivers of change, including climatic factors, on the occurrence of chemical food safety hazards in fruits and vegetables: A Bayesian Network approach Food Control 97 2019 67 76 Bradsher K. China Dominates Medical Supplies, in This Outbreak and the Next. The New York Times. 2020 Breiman L. Random forests Machine Learning 45 1 2001 5 32 Brintrup A. Pak J. Supply chain data analytics for predicting supplier disruptions: A case study in complex asset manufacturing International Journal of Production Research 58 11 2020 3330 3341 Cao W. Zhang X. Supply chain risk assessment based on support vector machine Revista Ibérica de Sistemas e Tecnologias de Informação(E5) 2016 310 Cavalcante I.M. Frazzon E.M. A supervised machine learning approach to data-driven simulation of resilient supplier selection in digital manufacturing International Journal of Information Management 49 2019 86 97 Chatzidimitriou K.C. Symeonidis A.L. Agent Mertacor: A robust design for dealing with uncertainty and variation in SCM environments Expert Systems with Applications 35 3 2008 591 603 Chen C.-S. Narani A. Ensemble models of feedstock blend ratios to minimize supply chain risk in bio-based manufacturing Biochemical Engineering Journal 107896 2020 Chu C.-Y. Park K. A global supply chain risk management framework: An application of text-mining to identify region-specific supply chain risks Advanced Engineering Informatics 45 2020 101053 Coadou, Y. (2010). Decision trees. In A. Lucotte & F. Melot (Eds.), Sos 08: In2p3 School of Statistics (Vol. 4). Coyle J.J. Ruamsook K. Weatherproofing Supply Chains: Enable Intelligent Preparedness with Data Analytics Transportation Journal 55 2 2016 190 207 Dietterich, T. G. (2000). Ensemble methods in machine learning. In J. Kittler & F. Roli (Eds.), Multiple Classifier Systems (Vol. 1857, pp. 1-15). Du, X. D., Cai, Y. H., et al. (2016). Overview of Deep Learning. Feng J.Y. Yuan B.Y. Evaluation on risks of sustainable supply chain based on optimized BP neural networks in fresh grape industry Computers and Electronics in Agriculture 183 2021 11 Fernandez A. Herrera F. Evolutionary fuzzy systems for explainable artificial intelligence: Why, when, what for, and where to? IEEE Computational intelligence magazine 14 1 2019 69 81 Feuerriegel S. Gordon J. News-based forecasts of macroeconomic indicators: A semantic path model for interpretable predictions European Journal of Operational Research 272 1 2019 162 175 Frances M. Pneumonia epidemic: Why is toilet paper a rush to buy? 2020 BBC Frazier P.D. Gilmore E.T. A novel counterfeit detection approach for integrated circuit supply chain assurance Journal of Hardware 2 3 2018 240 250 Frey G.P. West T.A.P. Simulated Impacts of Soy and Infrastructure Expansion in the Brazilian Amazon: A Maximum Entropy Approach Forests 9 10 2018 23 Friedman N. Geiger D. Bayesian network classifiers Machine Learning 29 2–3 1997 131 163 Fu W. Chien C.F. UNISON data-driven intermittent demand forecast framework to empower supply chain resilience and an empirical study in electronics distribution Computers and Industrial Engineering 135 2019 940 949 Garvey M.D. Carnovale S. An analytical framework for supply network risk propagation: A Bayesian network approach European Journal of Operational Research 243 2 2015 618 627 Gillet A. Brostaux Y. Main models used in logistic regression Biotechnologie Agronomie Societe Et Environnement 15 3 2011 425 433 Goldberg D.M. Khan S. Text Mining Approaches for Postmarket Food Safety Surveillance Using Online Media Risk Analysis 00 2020 1 20 Gopnik A. Tenenbaum J.B. Bayesian networks, Bayesian learning and cognitive development Developmental Science 10 3 2007 281 287 17444969 Govindan K. Mina H. A decision support system for demand management in healthcare supply chains considering the epidemic outbreaks: A case study of coronavirus disease 2019 (COVID-19) Transportation Research Part E-Logistics and Transportation Review 138 2020 Guo Y.H. Fu Y.S. Integrated phenology and climate in rice yields prediction using machine learning methods Ecological Indicators 120 2021 Gurtu A. Johny J. Supply Chain Risk Management: Literature Review Risks 9 1 2021 Hajian Heidary M. Aghaie A. Risk averse sourcing in a stochastic supply chain: A simulation-optimization approach Computers & Industrial Engineering 130 2019 62 74 Hameiri S. COVID-19: Is this the end of globalization? International Journal 76 1 2021 30 41 Han C. Zhang Q. Optimization of supply chain efficiency management based on machine learning and neural network Neural Computing and Applications 33 5 2021 1419 1433 Handelman G.S. Kok H.K. Peering into the black box of artificial intelligence: evaluation metrics of machine learning methods. 212 1 2019 38 43 Handfield R. Sun H. Assessing supply chain risk for apparel production in low cost countries using newsfeed analysis Supply Chain Management-an International Journal 25 6 2020 803 821 Ho W. Zheng T. Supply chain risk management: A literature review International Journal of Production Research 53 16 2015 5031 5069 Hobbs J.E. The Covid-19 pandemic and meat supply chains Meat Science 181 2021 Huang G.B. Zhou H.M. Extreme Learning Machine for Regression and Multiclass Classification Ieee Transactions on Systems Man and Cybernetics Part B-Cybernetics 42 2 2012 513 529 Huang G.B. Zhu Q.Y. Extreme learning machine: Theory and applications Neurocomputing 70 1–3 2006 489 501 Islam, S., & Amin, S. H. J. J. o. B. D. (2020a). Prediction of probable backorder scenarios in the supply chain using Distributed Random Forest and Gradient Boosting Machine learning techniques. 7(1), 1-22. Islam, S., & Amin, S. H. J. J. o. B. D. (2020b). Prediction of probable backorder scenarios in the supply chain using Distributed Random Forest and Gradient Boosting Machine learning techniques. Journal of Big Data, 7(1), 1-22. Jain A.K. Data clustering: 50 years beyond K-means Pattern Recognition Letters 31 8 2010 651 666 Janiesch C. Zschech P. Machine learning and deep learning Electronic Markets 31 3 2021 685 695 Ji G. Hu L. A study on decision-making of food supply chain based on big data Journal of Systems Science 26 2 2017 183 198 Jordan M.I. Mitchell T.M. Machine learning: Trends, perspectives, and prospects Science 349 6245 2015 255 260 26185243 Kaur H. Singh S.P. Disaster resilient proactive and reactive procurement models for humanitarian supply chain Production Planning & Control 2020 Kianmehr, K., & Alhajj, R. (2006). Support vector machine approach for fast classification. In A. M. Tjoa & J. Trujillo (Eds.), Data Warehousing and Knowledge Discovery, Proceedings (Vol. 4081, pp. 534-543). Kouadio L. Byrareddy V.M. Probabilistic yield forecasting of robusta coffee at the farm scale using agroclimatic and remote sensing derived indices Agricultural and Forest Meteorology 306 2021 Lau H. Tsang Y.P. Risk quantification in cold chain management: A federated learning-enabled multi-criteria decision-making methodology Industrial Management & Data Systems 20 2021 Lee C.H. Yang H.C. Enabling blockchain based scm systems with a real time event monitoring function for preemptive risk management Applied Sciences (Switzerland) 11 11 2021 Li L. Ma S.J. Data-driven online service supply chain: A demand-side and supply-side perspective Journal of Enterprise Information Management 34 1 2021 365 381 Li X. Zhang W. Diagnosing Rotating Machines With Weakly Supervised Data Using Deep Transfer Learning Ieee Transactions on Industrial Informatics 16 3 2020 1688 1697 Li Z. Guo H. A sustainable production capability evaluation mechanism based on blockchain, LSTM, analytic hierarchy process for supply chain network International Journal of Production Research 58 24 2020 7399 7419 Liu X. Zhang N. Research on customer satisfaction of budget hotels based on revised IPA and online reviews Science Journal of Business and Management 8 2 2020 50 Liu Y. Huang L. Supply chain finance credit risk assessment using support vector machine–based ensemble improved with noise elimination International Journal of Distributed Sensor Networks 16 1 2020 Liu Z.B. Gao R. Two-period pricing and strategy choice for a supply chain with dual uncertain information under different profit risk levels Computers & Industrial Engineering 136 2019 173 186 Lu S. Enterprise supply chain risk assessment based on improved neural network algorithm and machine learning Journal of Intelligent and Fuzzy Systems 40 4 2021 7013 7024 Lundberg S.M. Erion G. From local explanations to global understanding with explainable AI for trees Nature machine intelligence 2 1 2020 56 67 Marvin H.J.P. van Asselt E. Expert-driven methodology to assess and predict the effects of drivers of change on vulnerabilities in a food supply chain: Aquaculture of Atlantic salmon in Norway as a showcase Trends in Food Science & Technology 103 2020 49 56 McClements D.J. Barrangou R. Building a Resilient, Sustainable, and Healthier Food Supply through Innovation and Technology Annual Review of Food Science and Technology 12 2021 1 28 Merve E.K. Oktay Fırat S.Ü. A data mining-based framework for supply chain risk management Computers & Industrial Engineering 139 2020 105570 Mohanty D.K. Parida A.K. Financial market prediction under deep learning framework using auto encoder and kernel extreme learning machine Applied Soft Computing 99 2021 Murphy K.P. Naive bayes classifiers University of British Columbia 18 60 2006 1 8 Narwane, V. S., Raut, R. D., et al. Risks to Big Data Analytics and Blockchain Technology Adoption in Supply Chains. Annals of Operations Research. Nayal K. Raut R.D. Are artificial intelligence and machine learning suitable to tackle the COVID-19 impacts? An agriculture supply chain perspective International Journal of Logistics Management. 2021 Ni D. Xiao Z. Multiple Human-Behaviour Indicators for Predicting Lung Cancer Mortality with Support Vector Machine Scientific Reports 8 2018 Ni Y. Comprehensive Assessment of Logistics Financial Risk in Port Regional Supply Chain Journal of Coastal Research 2019 198 204 Nikolopoulos K. Punia S. Forecasting and planning during a pandemic: COVID-19 growth rates, supply chain disruptions, and governmental decisions European Journal of Operational Research 290 1 2021 99 115 32836717 Palander T. Eronen J. Development of a wood damage monitoring system for mechanized harvesting Annals of Forest Research 61 2 2018 243 258 Park, C.-Y., Kim, K., et al. (2020). Global shortage of personal protective equipment amid COVID-19: supply chains, bottlenecks, and policy implications: Asian Development Bank. Piccialli F. Giampaolo F. Artificial intelligence and healthcare: Forecasting of medical bookings through multi-source time-series fusion & nbsp Information Fusion 74 2021 1 16 Priore P. Ponte B. Applying machine learning to the dynamic selection of replenishment policies in fast-changing supply chain environments International Journal of Production Research 57 11 2019 3663 3677 Punia S. Singh S.P. From predictive to prescriptive analytics: A data-driven multi-item newsvendor model Decision Support Systems 136 2020 11 Ribeiro J.P. Barbosa-Povoa A. Supply Chain Resilience: Definitions and quantitative modelling approaches - A literature review Computers & Industrial Engineering 115 2018 109 122 Rishehchi F.M. Rasouli M.R. A data-driven and network-aware approach for credit risk prediction in supply chain finance Industrial Management and Data Systems 121 4 2021 785 808 Rodgers, M., & Singham, D. J. J. o. P. i. (2020). A framework for assessing disruptions in a clinical supply chain using Bayesian belief networks. Journal of Pharmaceutical innovation, 15(3), 467-481. Roukny T. Battiston S. Interconnectedness as a source of uncertainty in systemic risk Journal of Financial Stability 35 2018 93 106 Sang, B. J. J. o. C., & Mathematics, A. (2021). Application of genetic algorithm and BP neural network in supply chain finance under information sharing. Journal of Computational Applied Mathematics, 384, 113170. Sarker I.H. Machine learning: Algorithms, real-world applications and research directions SN Computer Science 2 3 2021 1 21 Saygili A. A new approach for computer-aided detection of coronavirus (COVID-19) from CT and X-ray images using machine learning methods Applied Soft Computing 105 2021 Schmidhuber J. Deep learning in neural networks: An overview Neural Networks 61 2015 85 117 25462637 Shahbaz M.S. Sohu S. A Novel Classification of Supply Chain Risks Engineering Technology & Applied Science Research 9 3 2019 4301 4305 Shahed K.S. Azeem A. A supply chain disruption risk mitigation model to manage COVID-19 pandemic risk Environmental Science and Pollution Research 2021 Sun G. Quantitative analysis of enterprise chain risk based on SVM algorithm and mathematical fuzzy set Journal of Intelligent and Fuzzy Systems 39 4 2020 5773 5783 Sun H. Analysis of risk factors in financial supply chain based on machine learning and IoT technology Journal of Intelligent and Fuzzy Systems 40 4 2021 6421 6431 Sun X.L. Choi T.M. Sales forecasting using extreme learning machine with applications in fashion retailing Decision Support Systems 46 1 2008 411 419 Tallón-Ballesteros A. Chen C. Explainable AI: Using Shapley value to explain complex anomaly detection ML-based systems Machine learning and artificial intelligence 332 2020 152 Thota M. Kollias S. Multi-source domain adaptation for quality control in retail food packaging Computers in Industry 123 2020 Uthayakumar J. Metawa N. Intelligent hybrid model for financial crisis prediction using machine learning techniques Information Systems and E-Business Management 18 4 2020 617 645 Vafadarnikjoo, A., Ahmadi, H. B., et al. Analyzing blockchain adoption barriers in manufacturing supply chains by the neutrosophic analytic hierarchy process. Annals of Operations Research. Vapnik V. Cortes C. SUPPORT-VECTOR NETWORKS Machine Learning 20 3 1995 273 297 Wang J. Swartz C.L.E. Supply Chain Monitoring Using Principal Component Analysis Industrial & Engineering Chemistry Research 59 27 2020 12487 12503 Wang, Y. (2021a). Research on Supply Chain Financial Risk Assessment Based on Blockchain and Fuzzy Neural Networks. Wireless Communications and Mobile Computing, 2021. Wang Z.Y. From Crisis to Nationalism? The Conditioned Effects of the COVID-19 Crisis on Neo-nationalism in Europe Chinese Political Science Review 6 1 2021 20 39 Weinan E. Machine Learning and Computational Mathematics Communications in Computational Physics 28 5 2020 1639 1670 Wichmann P. Brintrup A. Extracting supply chain maps from news articles using deep neural networks International Journal of Production Research 58 17 2020 5320 5336 Wu J. Zhang Z.C. Credit Rating Prediction Through Supply Chains: A Machine Learning Approach Production and Operations Management. 2021 Xu C. Jackson S.A. Machine learning and complex biological data In Vol. 20 2019 Springer 1 4 Xuan F. Regression analysis of supply chain financial risk based on machine learning and fuzzy decision model Journal of Intelligent & Fuzzy Systems 40 4 2021 6925 6935 Ying, H., Chen, L., et al. (2020). Application of text mining in identifying the factors of supply chain financing risk management. Industrial Management Data Systems. Yu L.A. Yang Z.B. Prediction-Based Multi-Objective Optimization for Oil Purchasing and Distribution with the NSGA-II Algorithm International Journal of Information Technology & Decision Making 15 2 2016 423 451 Yu Y. Choi T.M. An intelligent fast sales forecasting model for fashion products Expert Systems with Applications 38 6 2011 7373 7379 Zhang F. Paul S.D. On Database-Free Authentication of Microelectronic Components Ieee Transactions on Very Large Scale Integration (Vlsi) Systems 29 1 2021 149 161 Zhang G. Li G. Risk assessment and monitoring of green logistics for fresh produce based on a support vector machine Sustainability (Switzerland) 12 18 2020 Zhang L. Hu H. A credit risk assessment model based on SVM for small and medium enterprises in supply chain finance Financial Innovation 1 1 2015 1 21 Zhang S.C. Li X.L. Efficient kNN Classification With Different Numbers of Nearest Neighbors Ieee Transactions on Neural Networks and Learning Systems 29 5 2018 1774 1785 28422666 Zhang, Y., Xu, F., et al. (2021b). XAI Evaluation: Evaluating Black-Box Model Explanations for Prediction. Paper presented at the 2021 II International Conference on Neural Networks and Neurotechnologies (NeuroNT). Zhao H. Muthupandi S. Managing illicit online pharmacies: Web analytics and predictive models study Journal of Medical Internet Research 22 8 2020 Zhu Y. Xie C. Predicting China's SME Credit Risk in Supply Chain Finance Based on Machine Learning Methods Entropy 18 5 2016 8 Zhu Y. Xie C. Comparison of individual, ensemble and integrated ensemble machine learning methods to predict China's SME credit risk in supply chain finance Neural Computing & Applications 28 2017 S41 S50 Zhu Y. Xie C. Comparison of individual, ensemble and integrated ensemble machine learning methods to predict China’s SME credit risk in supply chain finance Neural Computing and Applications 28 2017 41 50 Zhu Y. Zhou L. Forecasting SMEs' credit risk in supply chain finance with an enhanced hybrid ensemble machine learning approach International Journal of Production Economics 211 2019 22 33 Zigiene G. Rybakovas E. Artificial Intelligence Based Commercial Risk Management Framework for SMEs Sustainability 11 16 2019 23	36475042	PMC9715461	NO-CC CODE	2022-12-05 23:15:31	no	Comput Ind Eng. 2023 Jan 2; 175:108859	utf-8	Comput Ind Eng	2,022	10.1016/j.cie.2022.108859	oa_other
==== Front Arch Psychiatr Nurs Arch Psychiatr Nurs Archives of Psychiatric Nursing 0883-9417 1532-8228 Elsevier Inc. S0883-9417(22)00149-2 10.1016/j.apnu.2022.11.001 Article The relationship between fear of COVID-19 and psychological resilience according to personality traits of university students: A PATH analysis Eroglu Ayse a⁎ Suzan Ozge Karakaya a Hur Gulsah a Cinar Nursan b a Sakarya University, Institute of Health Sciences, Department of Nursing, Esentepe Campus, 54187 Serdivan, Sakarya, Turkey b Sakarya University, Faculty of Health Sciences, Department of Nursing, Esentepe Campus, 54187 Serdivan, Sakarya, Turkey ⁎ Corresponding author. 2 12 2022 2 2023 2 12 2022 42 18 20 5 2022 20 9 2022 24 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. Purpose The purpose of the current research was to identify the influence of university students' personality traits on their fear of COVID-19 and psychological resilience levels. Design and methods A cross-sectional trial was completed with 690 students. Descriptive statistics and correlations were calculated, and a path analysis was employed with the objective of assessing the model fit and investigating direct and indirect impacts. Findings Among personality traits, conscientiousness and neuroticism were observed to affect fear of COVID-19, and conscientiousness, neuroticism, and openness to experience had an effect on psychological resilience. The tested model has a good fit and explains the direct effects of the study variables. Practice implications Nurses should improve university students' psychological resilience by supporting them with protective and improving factors. The role of the psychiatric nurse is important in providing conscious and need-oriented support in extraordinary events such as pandemics. Keywords COVID-19 Fear Nursing Personality traits Psychological resilience University student ==== Body pmcIntroduction Human beings have faced many epidemics in the historical process, and COVID-19 has been the last epidemic they must fight (Cucinotta & Vanelli, 2020). Human beings, who have been confronted with many social and individual limitations with the declaration of the disease a pandemic by the WHO because of the rapid spread of the virus worldwide, have experienced new circumstances, which they have not experienced in the recent past. In addition to the risk of virus transmission and death, the pandemic has also brought about an intense psychological pressure due to reasons such as deterioration of the routine functioning of life, feeling of uncertainty, fear of death, and fear of losing relatives. People's anxious and fearful state of mind originating from their fear of being infected with COVID-19 is defined by the concept of “fear of COVID-19” (Ahorsu et al., 2020). There is a lot of research in the literature indicating that fear of COVID-19 leads to mental, cognitive, and behavioral problems in people of all ages (Alici & Copur, 2021; Fitzpatrick et al., 2020; Zhang et al., 2020; Yue et al., 2020; Feng et al., 2020; Evren et al., 2020; Wang et al., 2020). Thus, fear of COVID-19 can be said to lead to significant psychological problems as well as health problems (Gundogan, 2021). While trying to protect themselves from the disease, people are also trying to stay psychologically healthy during this period (Karal & Biçer, 2020). The concept of psychological resilience is described as a set of characteristics and protective mechanisms that facilitate the individual's successful adaptation to the existing process in the face of difficult conditions (Gundogan, 2021). Psychological resilience has been suggested to protect against fear of COVID-19, which manifests itself as a mental outcome of the pandemic, and studies have shown that people with high psychological resilience experience less depression, stress, and anxiety related to COVID-19 (Albott et al., 2020; Barzilay et al., 2020; Labrague & De los Santos, 2020). Therefore, psychological resilience can be considered an important component in coping with the fear resulting from COVID-19 (Albott et al., 2020). Although the pandemic usually affects the whole society, the severity and intensity of reactions vary from person to person. Meanwhile, there are personal differences with respect to psychological resilience (Polatcı & Tınaz, 2021). The most important reason for individual differences is the individual's personality, which makes him/her unique among billions of people. Personality, which includes the traits that the individual brings from birth and acquires through his/her life experiences, can be defined as “the unique pattern of factors affecting feelings, thoughts, and ways of behavior that distinguish a person from others.” Due to these traits, the mental reactions in this traumatic process caused by the pandemic take place at different levels individually (Baymur, 2017; Polatcı & Tınaz, 2021). Particularly university students who try to complete their adolescent development processes and adapt to developmental duties specific to adulthood, on one hand, and face academic and social requirements in the university environment, on the other hand, are a more fragile group in this process. Research in the literature has stated that the negativities caused by the pandemic affect the young population more, and university students are the group at risk (Brooks et al., 2020; Kaya, 2020; Luo et al., 2020; Torales et al., 2020; Wang et al., 2020; Xiao, 2020). In addition to their anxiety (Khodabakhshi-koolaee, 2020) and social problems arising from the quarantine, university students were also adversely affected by the sudden and radical changes in their daily lives. With the closure of universities and the transition to distance education, many of them had to move to the houses of their families, working students lost their jobs, and even some of them had to discontinue their education. It is crucial to monitor the mental health of university students, who will carry the effects of the pandemic from today to the future and form social memory, during the pandemic period (Yorguner et al., 2021; Zhai & Du, 2020; Alici & Copur, 2021). The COVID-19 pandemic has led to common fear and anxiety due to the nature of the pandemic. The fear induced by this traumatic life experience in people and the psychological resilience levels that represent their coping skills will differ within the scope of personality traits (Ahorsu et al., 2020; Kaya, 2020; Xiao, 2020). Although there are studies in the literature examining the correlation between fear of COVID-19 and psychological resilience, the predictive role of personality traits has not been investigated. It is thought that determining how these factors are associated with each other will contribute to the literature regarding the management of mental health outcomes. In this respect, the research was done to reveal the effects of university students' personality traits on their fear of COVID-19 and psychological resilience levels. Materials and methods Aim The current research has the following objectives: (a) to identify the impacts of personality traits of students studying in two state universities in the Marmara and Western Black Sea regions of Turkey on their fear of COVID-19 and emotional resilience levels, and (b) to examine the relationships between the relevant results through a path analysis. The hypothesized model is shown in Fig. 1 .Fig. 1 Hypothesized path model of the correlation between fear of COVID-19 and emotional resilience in accordance with personality traits. Fig. 1 Therefore, our initial hypotheses were as follows:H1a The extraverted personality trait of university students positively affects their fear of COVID-19. H1b The agreeable personality trait of university students positively affects their fear of COVID-19. H1c The conscientious personality trait of university students positively affects their fear of COVID-19. H1d The neurotic personality trait of university students positively affects their fear of COVID-19. H1e University students' openness to experience positively affects their fear of COVID-19. H2a The extraverted personality trait of university students positively affects their emotional resilience. H2b The agreeable personality trait of university students positively affects their emotional resilience. H2c The conscientious personality trait of university students positively affects their emotional resilience. H2d The neurotic personality trait of university students positively affects their emotional resilience. H2e University students' openness to experience positively affects their emotional resilience. Design and sampling The current research had a cross-sectional design. In the research, the sample size was computed at a 95 % confidence level by utilizing the “G. Power-3.1.9.2” software. According to the analysis, the standardized effect size was found to be 0.160 at the level of α = 0.05 based on previous research (Lü et al., 2014), and the minimum total number of samples was calculated as 501 with 0.95 theoretical power. Participants The study population consisted of all students studying at Sakarya University and Karabük University. For this cross-sectional study, 720 students were accessed and grouped with the stratified sampling method. Thirty participants were excluded because of missing data. Therefore, the study sample comprised 690 university students. The inclusion criteria were determined in the following way: (a) Volunteering to participate in the research, (b) Speaking and understanding Turkish well. There are health units (MEDIKO-Social Center) within the scope of both universities where the study was conducted, and counseling and support was provided by psychologists to university students who requested during the pandemic process in these units. Data collection Because of the COVID-19 outbreak, data collection was carried out via Google Forms. The survey was shared in the electronic environment by utilizing Google Drive's online service system (https://docs.google.com/forms/d/e/1FAIpQLSdNKpYaOABW4woORu5RQZiQyIG6k8ETzuH4MBnjKMO08E8SAw/viewform?usp=sf_link) and then on Facebook, WhatsApp, and Instagram. It was published between October 5 and November 22, 2021, for 7 weeks. Individuals with access to the survey link responded to the questions. Filling out the scales took about 10–15 min. The data were downloaded in the CSV format, and their analysis was performed after their revision and standardization. Data collection tools The online survey was prepared by the researchers by making use of available studies (Polatcı & Tınaz, 2021; Nazari et al., 2021) and comprised five parts. In the first part, the purpose, scope, and stages of the study were explained. In the second part of the survey, 10 questions were asked about the descriptive characteristics of the participants, including age, gender, department and class, catching COVID-19 infection, and vaccination status. In the third part, the Big Five Personality Traits Scale was used. The 10-item Big Five Personality Traits Scale developed by Rammstedt and John (2007) was adapted to Turkish by Horzum et al. (2017). The scale comprises five sub-dimensions: extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience, and there are a total of two items for each dimension, one of which is reverse. These two items are combined to form the sub-dimension score. It was revealed that Cronbach's alpha internal consistency values for the sub-dimensions of the scale ranged between 0.81 and 0.90, and the composite reliability coefficients ranged between 0.73 and 0.85. Scale assessment was performed according to the sub-dimensions. As the scores for each sub-dimension increase, the characteristics of that dimension increase in the individual (Horzum et al., 2017). The fourth part includes the Fear of COVID-19 Scale. This scale, which was developed by Ahorsu et al. (2020) to determine individuals' levels of fear of coronavirus, was adapted to Turkish by Ladikli et al. (2020). Cronbach's alpha internal consistency coefficient of the Fear of COVID-19 Scale is 0.86. High scores acquired from the scale demonstrate that individuals have high fear of coronavirus. In the last part, the Connor-Davidson Resilience Scale was benefited from. The 10-item short version of the 25-item resilience scale developed by Connor and Davidson (2003), which was applied to university students by Campbell-Sills and Stein (2007) for testing, was employed. The scale's Turkish validity and reliability study was carried out by Kaya and Odacı (2021). The scale's internal consistency coefficient is 0.81. The 10-item scale contains five-point Likert-type scoring. High scores acquired from the scale are considered to demonstrate high psychological resilience. Data analysis Descriptive statistics were utilized to address the study hypotheses. Statistical analysis was performed using SPSS 24 software (IBM Corporation) and SmartPLS. Descriptive statistics (mean, median and standard deviation [SD]) were employed to show the distribution of the study variables in the overall sample (n = 690). The path analysis was conducted to test our conceptual model that associates fear of COVID-19 and psychological resilience with the personality traits of university students (e.g., extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience) (using SmartPLS software). Within the framework of the measurement model, the composite reliability (CR) value was calculated for composite reliability, the average variance extracted (AVE) value was computed for convergent validity, and the Fornell-Larcker criterion and HTMT (Heterotrait-Monotrait Ratio) values were calculated for discriminant validity (Doğan, 2019: 82). Data analysis for the research uses the following guideline: firstly, confirmatory factor analysis is conducted to measure the reliability and validity of the research instrument. A Smart PLS path model analysis is conducted for the purpose of testing the three hypotheses of the present study. The results of the mentioned test will clearly show whether significant relationships exist between the independent variables and the dependent variable. Whereas the R2 value is utilized as an indicator for the overall predictive strength of the model according to these values, the value of 0.67 is considered substantial, 0.33 moderate, and 0.19 weak (Hair et al., 2017; Henseler et al., 2009). Whereas the f2 value is employed as a measure to establish the effect size of predicting variable in the model based on these values, the value of 0.35 is considered as large, 0.15 as medium, and 0.02 as weak (Hair et al., 2017). Finally, if the Q2 value for a dependent variable is above zero, it shows that the model has predictive relevance. The reported statistical significance was two-sided and set at the 5 % level. In our model, we assumed that students' personality traits had direct and indirect impacts on their fear of COVID-19 and psychological resilience (see Fig. 1). Ethical approval Ethical approval was received from the Health Ethics Committee of Sakarya University (Date: 17.09.2021 Issue: E-71522473-050.01.04-64314419). Institutional permission was acquired from the universities where the study was carried out. Furthermore, written permission was received from the Republic Ministry of Health, General Directorate of Health Services, Scientific Research Platform. The current research was done in line with the principles of the World Medical Association Declaration of Helsinki. The consent form was on the survey's first page. The participants were assured that they had the right to refuse to take part in the study and that all information to be given would be kept confidential. The students who participated in the research stated that they had read, understood, and agreed to take part on a voluntary basis by marking the “I agree” option, then completed the other parts of the questionnaire. Google Forms has privacy standards that involve protecting, not using, data; not sharing data without permission; and not selling personal information. Results Of the participants, 42.9 % are between 20 and 21 years of age, and 64.2 % are female. When the distribution of the participants in accordance with their scientific fields is reviewed, it is observed that 34.5 % study Science, 33.9 % study Social Sciences, and 31.4 % study Health Sciences. Of the participants, 91.7 % did not have any chronic disease, 24.9 % had a history of COVID-19 infection (diagnosed with PCR), and 90.2 % had two doses of vaccine. Families of 47.4 % had had COVID-19, and 23.5 % had losses in their families due to COVID-19. The validity and reliability findings of the data collection tools are shown in Table 1 .Table 1 Validity and reliability of the measures. Table 1 Factor Load CR AVE Cronbach's alpha Extraversion K1 0.791 0.820 0.696 0.818 K6 0.875 Agreeableness K2 0.930 0.753 0.614 0.719 K7 0.604 Conscientiousness K3 0.735 0.817 0.693 0.807 K8 0.920 Neuroticism K4 0.972 0.757 0.626 0.700 K9 0.554 Openness to experience K5 0.551 0.723 0.582 0.676 K10 0.927 Fear of Covid-19 CKÖ1 0.661 0.904 0.574 0.905 CKÖ2 0.708 CKÖ3 0.875 CKÖ4 0.742 CKÖ5 0.735 CKÖ6 0.754 CKÖ7 0.810 Psychological resilience PSÖ1 0.598 0.871 0.509 0.870 PSÖ2 0.726 PSÖ3 0.402 PSÖ4 0.642 PSÖ5 0.482 PSÖ6 0.724 PSÖ7 0.632 PSÖ8 0.718 PSÖ9 0.737 PSÖ10 0.650 An evaluation of the measurement model is presented in Table 1. For all constructs, Cronbach's alpha (varying between 0.700 and 0.905) indicates good reliability over 0.60. Moreover, as shown in Table 1, all factor loadings are above 0.4, and the CR value for all constructs (ranging from 0.723 to 0.904) is >0.7 and higher than the relevant AVE, which indicates good construct reliability and convergent validity. AVEs are >0.5 for all constructs. As observed in Table 2 , the correlation coefficients of the variables are lower than the square root of AVE values, and the Fornell-Larcker criteria are met. Furthermore, when the values in the table are reviewed, it is observed that the HTMT values are lower than the threshold value (<0.90). According to the findings in Table 2, discriminant validity can be said to be provided.Table 2 Fornell-Larcker Criterion Analysis and Heterotrait-Monotrait (HTMT) Ratio. Table 2 Fear of Covid-19 Openness to experience Extraversion Neuroticism Psychological resilience Agreeableness Conscientiousness Fornell-Larcker criterion Fear of Covid-19 0.758 Openness to experience −0.019 0.763 Extraversion −0.074 0.263 0.834 Neuroticism 0.267 −0.234 −0.247 0.791 Psychological resilience −0.410 0.356 0.482 −0.535 0.713 Agreeableness −0.103 0.111 0.256 −0.226 0.310 0.784 Conscientiousness −0.145 0.259 0.438 −0.192 0.511 0.348 0.833 Heterotrait-Monotrait (HTMT) ratio Fear of Covid-19 Openness to experience 0.043 Extraversion 0.071 0.276 Neuroticism 0.289 0.223 0.239 Psychological resilience 0.423 0.366 0.482 0.564 Agreeableness 0.100 0.106 0.258 0.275 0.314 Conscientiousness 0.144 0.282 0.447 0.184 0.502 0.361 The Varience Inflation Faktor (VIF) values related to personality traits sub-dimensions are presented in Table 3 . All values are below the critical value (<3). This shows that there is no linearity between the relevant variables.Table 3 The VIF values related to personality traits sub-dimensions. Table 3 VIF values Extraversion 1.921 Agreeableness 1.460 Conscientiousness 1.842 Neuroticism 1.408 Openness to experience 1.353 When the model analysis results are examined in Table 4 and Fig. 2 , conscientiousness and neuroticism are seen to have a statistically significant impact on fear of COVID-19 (p < 0.05). When the R2 values of the model are examined, conscientiousness and neuroticism explain 8.6 % of the change in fear of COVID-19. Conscientiousness, neuroticism, openness to experience and extraversion are seen to have a statistically significant impact on psychological resilience (p < 0.05). When the R2 values of the model are examined, conscientiousness, neuroticism, and openness to experience are revealed to explain 52 % of the change in psychological resilience.Table 4 Model analysis results. Table 4Hypotheses β t p f2 R2 Q2 Conclusion Extraversion➔ Fear of Covid-19 0.030 0.635 0.526 0.001 0.086 0.040 ✖ Agreeableness➔ Fear of Covid-19 −0.017 0.204 0.839 0.000 ✖ Conscientiousness ➔ Fear of Covid-19 −0.198 2.046 0.041⁎ 0.011 ✔ Neuroticism➔ Fear of Covid-19 0.264 5.125 0.000⁎ 0.070 ✔ Openness to experience➔ Fear of Covid-19 0.068 1.375 0.170 0.004 ✖ Extraversion ➔ Psychological Resilience 0.211 4.391 0.000⁎ 0.069 0.520 0.189 ✔ Agreeableness ➔ Psychological Resilience 0.053 1.368 0.172 0.005 ✖ Conscientiousness ➔ Psychological Resilience 0.293 6.517 0.000⁎ 0.129 ✔ Neuroticism ➔ Psychological Resilience −0.385 8.874 0.000⁎ 0.276 ✔ Openness to experience ➔ Psychological Resilience 0.129 3.226 0.001⁎ 0.031 ✔ ⁎ p < 0.05. Fig. 2 Path model of the study. Fig. 2 In accordance with the observed results and the comparison of fit indices with the required criteria, the said model can be considered a fit model. As observed in Fig. 2, hypotheses (except H1a, H1b, H1e, H2a, and H2b) were accepted. Of the five personality traits, agreeableness has no significant and direct influence on fear of COVID-19 and emotional resilience. Discussion The findings of the current study, which was conducted to reveal the impacts of personality traits on fear of COVID-19 and psychological resilience levels in university students, show that conscientious and neurotic personality traits have a statistically significant effect on fear of COVID-19 and psychological resilience. It is crucial to concentrate on the mental health of university students during crises and pandemics (Fuentes et al., 2021). Studies indicate a substantial increase in psychological problems in young people during the COVID-19 pandemic (McGinty et al., 2020; Pierce et al., 2020). A study conducted with 1653 participants aged 18 years and over from 63 countries showed that young age groups were more vulnerable to mental health problems, e.g., anxiety, stress, and depression, during the COVID-19 pandemic and needed more support (Varma et al., 2021). Students may experience more stress during the pandemic due to interruptions in education, concerns about personal or family health, and social isolation. The above-mentioned changes can lead to increased behavioral changes, concentration problems, and the use of negative coping strategies (Fuentes et al., 2021). With the effect of some factors such as personality traits, differences are observed in the way each individual copes with negative situations and their attitudes and behaviors toward circumstances. In this study, whose sample consisted of university students, the conscientious personality trait was revealed to have a reducing impact on fear of COVID-19. The conscientious personality trait refers to being planned, attentive, and showing a degree of resulting self-control. Conscientious individuals have the ability to control their impulses (Horzum et al., 2017). During this period, when people are faced with the threat of COVID-19, conscientiousness improves individuals' compliance with the rules and taking precautions. A study by Mula et al. (2021) involving university students elucidated that the threat of COVID-19 increased conscientiousness (Mula et al., 2021). There are other studies showing the influence of high conscientiousness on buffering the negative effect of the pandemic (Li et al., 2020; Rodriguez et al., 2021; Schnell & Krampe, 2020). Considering the findings of the studies, fear of COVID-19 may decrease as conscientiousness increases, and conscientiousness may increase with the effect of the threat and stress of COVID-19. The relationship between these factors may not be the same as in the early days when there was more uncertainty about the pandemic. People's reactions may change over time because of adaptation to big adverse life experiences. In a study from Slovakia, behavioral and emotional data were collected during the first and second waves of the pandemic, and it was observed that the influence of personality traits and fear of COVID-19 decreased statistically significantly (Kohút et al., 2021). It is believed that the perception of the situation may also evoke responses or behaviors stimulated by some personality traits (Zajenkowski et al., 2020). The neurotic personality trait is associated with the individual's emotional balance. Due to the problems experienced in ensuring emotional balance, these individuals are prone to experiencing negative emotions, e.g., depression and anxiety. They feel anxious, restless, and upset and cope with stress poorly (Horzum et al., 2017). In the current research, a positive association was determined between neuroticism and fear of COVID-19. According to this result, fear of COVID-19 rises as the level of neuroticism increases. In the literature, studies also support this result (Fink et al., 2021; Kohút et al., 2021). Neuroticism is thought to increase fear of COVID-19 due to the tendency to experience negative emotions including anxiety, anger, and depression and weak coping power. In a study, which involved 50 male participants aged 18–25 years, took 4.5 years, and was completed with 28 participants who were evaluated before, during, and after military service, one of the stressful life experiences, an increase was observed in neuroticism in participants during military service. After military service, however, there was a decrease in neuroticism (Magal et al., 2021). This situation coincides with the assumption that the effect of life experiences on personality traits may be temporary (Ormel et al., 2017). To test this assumption in the processes experienced during the COVID-19 period, there is a need for studies on the same group to research the relationships before, during, and after the pandemic. Psychological resilience means the ability to withstand difficulties and positive adaptation in the face of difficulties (Lutha & Cicchetti, 2000). In a study analyzing university students' psychological resilience during COVID-19 restrictions, Serrano Sarmiento et al. (2021) revealed that psychological resilience levels were usually high (highest in men and those over 25 years of age), and students studying Health Sciences had a higher ability to adapt to change and overcome the difficulties brought by the pandemic (Serrano Sarmiento et al., 2021). Conscientiousness, a stable personality trait, is described as the ability to regulate, direct, or control a person's impulsive thoughts, emotions, and behaviors (Telzer et al., 2011). Assuming that there is a supportive positive correlation between conscientiousness and psychological resilience based on the aforesaid characteristics, the findings of this research, which indicate an increase in psychological resilience as conscientiousness increases, are considered to be consistent. In a study performed on nursing students, the rate of neuroticism was found to increase in women and as age decreased (17–24 years) (Cuartero & Tur, 2021). A high level of neuroticism is correlated to a negative or excessively uncontrolled emotional approach, poor coping, and impulsive problems. Therefore, it is an expected result that it has a negative effect on psychological resilience. In the literature, there are studies indicating that neuroticism is associated with decreased psychological resilience (Findyartini et al., 2021) or that there is no association (Kocjan et al., 2021). The findings of the current research support the findings of studies in the literature indicating that neuroticism is correlated to decreased psychological resilience. Openness to experience, which is accepted as a positive personality trait, contains sensitive, flexible and creative features, while, extraversion is a personality trait that has features such as being lively, cheerful, sociable and social (Erkuş & Tabak, 2010; Karduz & Şar, 2019). In a study conducted with university students, extraversion and openness to experience were found to have a positive effect on resilience (Polatcı & Tınaz, 2021), and there are similar studies in different groups in the literature (Eroğlu, 2022, Yazıcı Çelebi, 2021). During the COVID-19 period, children and adolescents aged 18 years and under are also vulnerable groups at high risk of being adversely affected. A systematic review revealed that children and adolescents aged 18 years and under, who were affected by the pandemic, experienced fear, anxiety and disruptions in their daily routines but exhibited resilience with the right support (Berger et al., 2021). In this process, factors such as the way parents, healthcare providers, and the media reflect the event, previous experiences and support are considered to affect the fear experienced by children and the developing resilience (Berger et al., 2021). Psychological resilience is a significant variable to decrease and prevent the adverse impacts of the pandemic. It contributes to post-traumatic positive development (Finstad et al., 2021). Some personality traits such as extraversion are seen to be associated with post-traumatic development during the COVID-19 period with the effect of social support and coping strategies (Xie & Kim, 2022). Conclusion In the current results, conscientiousness, one of the personality traits of university students, adversely affects fear of COVID-19, and neuroticism has a positive and direct influence on fear of COVID-19. Meanwhile, neuroticism was observed to adversely affect psychological resilience, and conscientiousness and openness to experience to have a positive direct impact on psychological resilience. Implications for nursing practice The COVID-19 outbreak leads to numerous physical and mental health conditions in different groups. It is an undeniable fact that, among these groups, university students are exposed to the stress induced by the pandemic. The present research emphasizes that nurses, who play a role in maintaining and improving physical mental health, should plan their approaches considering the personal differences regarding university students' psychological resilience. Nurses should support and improve university students' psychological resilience with protective and improving factors. Nurses in health centers of universities and health personnel in health research and application councils should implement activities to strengthen psychological resilience by adopting a student-centered approach according to students' personality traits. Our current students will form the adult memory of the future. Therefore, nurses play an important role in raising their awareness and supporting them in their needs during the pandemic period. In addition, interventions to increase psychological resilience should be added to emergency measures to be prepared for extraordinary crisis situations such as pandemics and natural disasters. It can be said that psychological support activities should be increased and psychological counseling centers should be made more active in order to reduce the negative effects on university students, especially in crisis situations. It is the responsibility of educational institutions to employ psychologists, nurses and social workers in order to provide adequate service with the arrangements to be made in this regard. Limitations The data in this study were analyzed cross-sectionally. The correlational relationships among the variables do not imply cause and effect. There is also a need for studies investigating the long-term, causal effects of personality traits on fear of COVID-19 and psychological resilience levels. The findings are based on data acquired from self-report measures with the risk of bias. Despite the said limitations, the research findings provide useful data to health professionals who deal with the mental health problems of university students arising from the COVID-19 pandemic. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. CRediT authorship contribution statement The conception and design of the study: NÇ, GH, AE, ÖKS; acquisition of data: AE, ÖKS, GH; analysis and interpretation of data: C; drafting the article or revising it critically for important intellectual content: NÇ, GH, AE, ÖKS; final approval of the version to be submitted: NÇ, GH, AE, ÖKS. Ethical considerations Ethical approval was obtained from the Health Ethics Committee of Sakarya (17.09.2021 Issue: E-71522473-050.01.04-64314419). Declaration of competing interest The authors declared no conflict of interest. ==== Refs References Ahorsu D.K. Lin C.Y. Imani V. Saffari M. Griffiths M.D. Pakpour A.H. The fear of COVID-19 scale: Development and initial validation International Journal of Mental Health and Addiction 1–9 2020 10.1007/s11469-020-00270-8 Albott C.S. Wozniak J.R. McGlinch B.P. Wall M.H. Gold B.S. Vinogradov S. Battle buddies: Rapid deployment of a psychological resilience intervention for health care workers during the coronavirus disease 2019 pandemic Anesthesia and Analgesia 131 2020 43 54 10.1213/ANE.0000000000004912 32345861 Alici N.K. Copur E.O. Anxiety and fear of COVID-19 among nursing students during the COVID-19 pandemic: A descriptive correlation study Perspectives in Psychiatric Care 58 2021 141 148 10.1111/ppc.12851 Barzilay R. Moore T.M. Greenberg D.M. DiDomenico G.E. Brown L.A. White L.K. Gur R.C. Gur R.E. Resilience, COVID-19-related stress, anxiety and depression during the pandemic in a large population enriched for healthcare providers Translational Psychiatry 10 1 2020 1 8 10.1038/s41398-020-00982-4 32066695 Baymur F.B. Genel psikoloji (26. Baskı) s:275 2017 İnkılap Kitabevi İstanbul Berger E. Jamshidi N. Reupert A. Jobson L. Miko A. Review: The mental health implications for children and adolescents impacted by infectious outbreaks - A systematic review Child and Adolescent Mental Health 26 2 2021 157 166 10.1111/camh.12453 33733620 Brooks S.K. Webster R.K. Smith L.E. Woodland L. Wessely S. Greenberg N. Rubin G.J. The psychological impact of quarantine and how to reduce it: Rapid review of the evidence The Lancet 395 10227 2020 912 920 10.1016/S0140-6736(20)30460-8 Campbell-Sills L. Stein M.B. Psychometric analysis and refinement of the connor–davidson resilience scale (CD-RISC): Validation of a 10-item measure of resilience Journal of Traumatic Stress: Official Publication of The International Society for Traumatic Stress Studies 20 6 2007 1019 1028 10.1002/jts.20271 Connor K.M. Davidson J.R. Development of a new resilience scale: The Connor-Davidson resilience scale (CD-RISC) Depression and Anxiety 18 2 2003 76 82 10.1002/da.10113 12964174 Cuartero N. Tur A.M. Emotional intelligence, resilience and personality traits neuroticism and extraversion: Predictive capacity in perceived academic efficacy Nurse Education Today 102 2021 2021 1 6 10.1016/j.nedt.2021.104933 104933 Cucinotta D. Vanelli M. WHO declares COVID-19 a pandemic Acta Bio Medica: Atenei Parmensis 91 1 2020 157 10.23750/abm.v91i1.9397 Doğan D. SmartPLS ile veri analizi 2019 Zet Yayınları Ankara 89. Sayfa Erkuş A. Tabak A. Beş Faktör Kişilik Özelliklerinin Çalışanların Çatışma Yönetim Tarzlarına etkisi: Savunma sanayiinde bir Araştırma Retrieved from Atatürk Üniversitesi İktisadi ve İdari Bilimler Dergisi 23 2 2010 213 242 https://dergipark.org.tr/en/pub/atauniiibd/issue/2696/35515 Eroğlu M. The effect of features of personal features of workig women and factors such as age, education status, income level on psychological resistance levels Retrieved from International Journal of Primary Education Studies 3 1 2022 34 42 https://dergipark.org.tr/en/pub/ijpes/issue/70020/1048503 Evren C. Evren B. Dalbudak E. Topcu M. Kutlu N. Measuring anxiety related to COVID-19: A turkish validation study of the coronavirus anxiety scale Death Studies 1–7 2020 10.1080/07481187.2020.1774969 Feng L.S. Dong Z.J. Yan R.Y. Wu X.Q. Zhang L. Ma J. Zeng Y. Psychological distress in the shadow of the COVID-19 pandemic: Preliminary development of an assessment scale Psychiatry Research 291 2020 2020 1 6 10.1016/j.psychres.2020.113202 Findyartini A. Greviana N. Putera A.M. Sutanto R.L. Saki V.Y. Felaza E. The relationships between resilience and student personal factors in an undergraduate medical program BMC Medical Education 21 1 2021 113 10.1186/s12909-021-02547-5 33602176 Fink M. Bäuerle A. Schmidt K. Rheindorf N. Musche V. Dinse H. Moradian S. Weismüller B. Schweda A. Teufel M. Skoda E.M. COVID-19-fear affects current safety behavior mediated by neuroticism-results of a large cross-sectional study in Germany Frontiers in Psychology 12 2021 671768 10.3389/fpsyg.2021.671768 Finstad G.L. Giorgi G. Lulli L.G. Pandolfi C. Foti G. León-Perez J.M. Cantero-Sánchez F.J. Mucci N. Resilience, coping strategies and posttraumatic growth in the workplace following COVID-19: A narrative review on the positive aspects of trauma International Journal of Environmental Research and Public Health 18 18 2021 9453 10.3390/ijerph18189453 34574378 Fitzpatrick K.M. Harris C. Drawve G. Fear of COVID-19 and the mental health consequences in America Psychological Trauma: Theory, Research, Practice, and Policy 12 S1 2020 17 21 10.1037/tra0000924 Fuentes A.V. Jacobs R.J. Ip E. Owens R.E. Caballero J. Coping, resilience, and emotional well-being in pharmacy students during the COVID-19 pandemic The Mental Health Clinician 11 5 2021 274 278 10.9740/mhc.2021.09.274 34621602 Gundogan S. The mediator role of the fear of COVID-19 in the relationship between psychological resilience and life satisfaction Current Psychology 40 12 2021 6291 6299 10.1007/s12144-021-01525-w 33716474 Hair J.F. Hult G.T.M. Ringle C.M. Sarstedt M. A primer on partial least squares structural equation modeling (PLS-SEM) 2nd ed. 2017 Sage Publications Inc United States of America Henseler J. Ringle C.M. Sinkovics R.R. The use of the partial least squares path modeling in international marketing New Challenges to International Marketing Advances in International Marketing 20 1 2009 277 319 10.1108/S1474-7979(2009)0000020014 Horzum M.B. Ayas T. Padır M.A. Beş faktör kişilik ölçeğinin Türk kültürüne uyarlanması adaptation of big five personality traits scale to turkish culture Sakarya University Journal of Education 7 2 2017 398 408 10.19126/suje.298430 Karal E. Biçer B.G. Salgın hastalık döneminde algılanan sosyal desteğin bireylerin psikolojik sağlamlığı üzerindeki etkisinin incelenmesi Birey ve Toplum Sosyal Bilimler Dergisi 10 1 2020 129 156 10.20493/birtop.726411 Karduz F.F.A. Şar A.H. The effect of psycho-education program on increase the tendency to forgive and five factor personality properties of forgiveness tendency International Journal of Psychology and Educational Studies 6 3 2019 89 105 10.17220/ijpes.2019.03.010 Kaya B. Pandeminin ruh sağlığına etkileri Klinik Psikiyatri Dergisi 23 2 2020 123 124 10.5505/kpd.2020.64325 Kaya F. Odacı H. Connor-Davidson Psikolojik Sağlamlık Ölçeği Kısa Formu: Türkçe’ye uyarlama, geçerlik ve güvenirlik çalışması HAYEF: Journal of Education 18 1 2021 38 54 Khodabakhshi-koolaee A. Living in home quarantine: Analyzing psychological experiences of college students during Covid-19 pandemic Journal of Military Medicine 22 2 2020 130 138 10.30491/JMM.22.2.130 Kocjan G.Z. Kavčič T. Avsec A. Resilience matters: Explaining the association between personality and psychological functioning during the COVID-19 pandemic International Journal of Clinical and Health Psychology 21 2021 2021 1 9 10.1016/j.ijchp.2020.08.002 100198 Kohút M. Kohútová V. Halama P. Big five predictors of pandemic-related behavior and emotions in the first and second COVID-19 pandemic wave in Slovakia Personality and Individual Differences 180 2021 110934 10.1016/j.paid.2021.110934 Labrague L.J. De los Santos J.A.A. COVID-19 anxiety among front-line nurses: Predictive role of organisational support, personal resilience and social support Journal of Nursing Management 28 7 2020 1653 1661 10.1111/jonm.13121 32770780 Ladikli N. Bahadir E. Yumuşak F.N. Akkuzu H. Karaman G. Türkkan Z. The reliability and validity of Turkish version of coronavirus anxiety scale Retrieved from International Journal of Social Science 3 2 2020 71 80 https://dergipark.org.tr/en/pub/injoss/issue/56160/774887 Li J.B. Yang A. Dou K. Cheung R. Self-control moderates the association between perceived severity of coronavirus disease 2019 (COVID-19) and mental health problems among the Chinese public International Journal of Environmental Research and Public Health 17 13 2020 4820 10.3390/ijerph17134820 32635495 Lü W. Wang Z. Liu Y. Zhang H. Resilience as a mediator between extraversion, neuroticism and happiness, PA and NA Personality and Individual Differences 63 2014 128 133 10.1016/j.paid.2014.01.015 Luo Y. Chua C.R. Xiong Z. Ho R.C. Ho C.S. A systematic review of the impact of viral respiratory epidemics on mental health: An implication on the coronavirus disease 2019 pandemic Frontiers in Psychiatry 11 2020 1 21 10.3389/fpsyt.2020.565098 32116830 Lutha S.S. Cicchetti D. The construct of resilience: Implications for interventions and social policies Development and Psychopathology 12 4 2000 857 885 10.1017/s0954579400004156 11202047 Magal N. Hendler T. Admon R. Is neuroticism really bad for you? Dynamics in personality and limbic reactivity prior to, during and following real-life combat stress Neurobiology of Stress 15 2021 2021 1 13 10.1016/j.ynstr.2021.100361 100361 McGinty E.E. Presskreischer R. Han H. Barry C.L. Psychological distress and loneliness reported by US adults in 2018 and april 2020 JAMA 324 1 2020 93 94 10.1001/jama.2020.9740 32492088 Mula S. Di Santo D. Gelfand M.J. Cabras C. Pierro A. The mediational role of desire for cultural tightness on concern with COVID-19 and perceived self-control Frontiers in Psychology 12 2021 713952 10.3389/fpsyg.2021.713952 Nazari N. Safitri S. Usak M. Arabmarkadeh A. Griffiths M.D. Psychometric validation of the ındonesian version of the fear of COVID-19 scale: Personality traits predict the fear of COVID-19 International Journal of Mental Health and Addiction 1–17 2021 10.1007/s11469-021-00593-0 Ormel J. VonKorff M. Jeronimus B.F. Riese H. 9-set-point theory and personality development: Reconciliation of a paradox Personality Development Across the Lifespan 2017 117 137 10.1016/B978-0-12-804674-6.00009-0 Pierce M. Hope H. Ford T. Hatch S. Hotopf M. John A. Kontopantelis E. Webb R. Wessely S. McManus S. Abel K.M. Mental health before and during the COVID-19 pandemic: A longitudinal probability sample survey of the UK population The Lancet. Psychiatry 7 10 2020 883 892 10.1016/S2215-0366(20)30308-4 32707037 Polatcı S. Tınaz Z.D. The effect of personality traits on psychological resilience abstract OPUS International Journal of Society Researches 17 36 2021 2890 2917 Rammstedt B. John O.P. Measuring personality in one minute or less: A 10-item short version of the big five inventory in English and German Journal of Research in Personality 41 1 2007 203 212 10.1016/j.jrp.2006.02.001 Rodriguez J.E. Holmes H.L. Alquist J.L. Uziel L. Stinnett A.J. Self-controlled responses to COVID-19: Self-control and uncertainty predict responses to the COVID-19 pandemic Current Psychology 1–15 2021 10.1007/s12144-021-02066-y Schnell T. Krampe H. Meaning in life and self-control buffer stress in times of COVID-19: Moderating and mediating effects with regard to mental distress Frontiers in Psychiatry 11 2020 582352 10.3389/fpsyt.2020.582352 Serrano Sarmiento Á. Sanz Ponce R. González Bertolín A. Resilience and COVID-19. An analysis in university students during confinement Education Sciences 11 9 2021 533 10.3390/educsci11090533 Telzer E.H. Masten C.L. Berkman E.T. Lieberman M.D. Fuligni A.J. Neural regions associated with self control and mentalizing are recruited during prosocial behaviors towards the family NeuroImage 58 1 2011 242 249 10.1016/j.neuroimage.2011.06.013 21703352 Torales J. O'Higgins M. Castaldelli-Maia J.M. Ventriglio A. The outbreak of COVID-19 coronavirus and its impact on global mental health The International Journal of Social Psychiatry 66 2020 317 320 10.1177/0020764020915212 32233719 Varma P. Junge M. Meaklim H. Jackson M.L. Younger people are more vulnerable to stress, anxiety and depression during COVID-19 pandemic: A global cross-sectional survey Progress in Neuro-Psychopharmacology & Biological Psychiatry 109 2021 2021 1 8 10.1016/j.pnpbp.2020.110236 Wang C. Pan R. Wan X. Tan Y. Xu L. McIntyre R.S. 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 Xiao C. A novel approach of consultation on 2019 novel coronavirus (COVID19) related psychological and mental problems: Structured letter therapy Psychiatry Investigation 17 2 2020 175 176 10.30773/pi.2020.0047 32093461 Xie C.S. Kim Y. Post-traumatic growth during COVID-19: The role of perceived social support, personality, and coping strategies Healthcare 10 2 2022 224 10.3390/healthcare10020224 35206839 Yazıcı Çelebi G. Analysis of the relationship between women’s personality traits and psychological resilience levels Mavi Atlas 9 1 2021 132 146 10.18795/gumusmaviatlas.832657 Yorguner N. Bulut N.S. Akvardar Y. COVID-19 salgını sırasında üniversite öğrencilerinin karşılaştığı psikososyal zorlukların ve hastalığa yönelik bilgi, tutum ve davranışlarının incelenmesi Noro-Psikyatri Arsivi 58 1 2021 3 10 10.29399/npa.27503 Yue J. Zang X. Le Y. An Y. Anxiety, depression and PTSD among children and their parent during 2019 novel coronavirus disease (COVID-19) outbreak in China Current Psychology 1–8 2020 10.1007/s12144-020-01191-4 Zajenkowski M. Jonason P.K. Leniarska M. Kozakiewicz Z. Who complies with the restrictions to reduce the spread of COVID-19?: Personality and perceptions of the COVID-19 situation Personality and Individual Differences 166 2020 2020 1 6 10.1016/j.paid.2020.110199 110199 Zhai Y. Du X. Addressing collegiate mental health amid COVID-19 pandemic. Psychiatry Research 288 113003 2020 10.1016/j.psychres.2020.113003 Zhang S.X. Wang Y. Rauch A. Wei F. Unprecedented disruption of lives and work: Health, distress and life satisfaction of working adults in China one month into the COVID-19 outbreak Psychiatry Research 288 112958 2020 10.1016/j.psychres.2020.112958	0	PMC9715462	NO-CC CODE	2022-12-14 23:45:34	no	Arch Psychiatr Nurs. 2023 Feb 2; 42:1-8	utf-8	Arch Psychiatr Nurs	2,022	10.1016/j.apnu.2022.11.001	oa_other
==== Front Res Social Adm Pharm Res Social Adm Pharm Research in Social & Administrative Pharmacy 1551-7411 1934-8150 Elsevier Inc. S1551-7411(22)00413-2 10.1016/j.sapharm.2022.11.011 Article Biopsychosocial analysis of antibiotic use for the prevention or management of COVID-19 infections: A scoping review Nortey Radolf Ansbert a∗ Kretchy Irene Akwo b Koduah Augustina b Buabeng Kwame Ohene a a Department of Pharmacy Practice, Faculty of Pharmacy and Pharmaceutical Sciences, Private Mail Bag, University Post Office, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana b Department of Pharmacy Practice and Clinical Pharmacy, School of Pharmacy, University of Ghana, PO Box LG 43, Legon, Accra, Ghana ∗ Corresponding author. 2 12 2022 2 12 2022 19 4 2022 12 11 2022 27 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 The novelty and complexity of the COVID-19 pandemic has resulted in various coping mechanisms adopted by individuals as a means of averting the perceived fatalities of the pandemic. The use of antibiotics in the management of COVID-19 is clinically recommended under specific conditions. However, there are increasing trends of non-adherence to the recommended criteria resulting in the unwarranted use of antibiotics as an adaptative approach to the ongoing pandemic. Objective The objective was to identify and classify factors associated with the unwarranted use of antibiotics in the management of COVID-19 from published literature and the perspectives of key stakeholders along a Biopsychosocial model. Methods Literature was searched in the following databases: PubMed/MEDLINE, Scopus, Embase and Google Scholar for studies published between 31st December 2019 and 31st January 2022. The Arskey and O'Malley framework modified by Levac in the six-stage methodological process was adopted for this review and included: a) identification of research questions, b) identification of relevant research articles, c) selection of studies, d) data charting and synthesis, e) summary, discussion and analysis, and f) stakeholder consultations. Results Out of 10,252 records identified from all sources, 12 studies were selected for inclusion in this scoping review. The selected articles reflected both antibiotic use and COVID-19 whilst capturing the biological (medical) and psychosocial perspectives. Most of the studies reported the overuse or abuse of Azithromycin especially in hospital settings. Common themes across the review and stakeholder consultations included fear, anxiety, media influences and deficits in public knowledge. Conclusion The findings of the study highlight the complexity of antibiotic control especially in the context of a pandemic. The identified determinants of antibiotic use provide the necessary framework to simulate health emergencies and be better positioned in the future through the development of targeted and comprehensive policies on antibiotic stewardship. Keywords COVID-19 Antibiotic Misuse Overuse Antimicrobial resistance ==== Body pmc1 Background COVID-19 is a viral infectious disease caused by the coronavirus which manifests as a mild, moderate to severe respiratory illness in victims.1 Similar to the profile of most viruses, the coronavirus is constantly mutating and resulting in genetic variation which may be potentially more virulent.2 The COVID-19 syndrome is caused by a novel SARS-CoV-2 with self-limiting clinical manifestations mostly mild in children.3 As a typical viral disease, antibiotics are not the primary anti-infective agents adopted in the clinical management of COVID-19. However, the use of antibiotics in the clinical management of COVID-19 is not out of place since atypical antibiotics may improve the prognosis of viral infections through indirect immunomodulatory and anti-inflammatory mechanisms.4 , 5 The virus can also predispose a patient to secondary bacterial infections which may require the general use of antibiotics.6 , 7 The antibiotics become necessary due to the history of bacterial co-infections heightening the gravity of respiratory viral infections and subsequently causing death. However, some antibiotics have been found to possess antiviral effects exhibited by their ability to modulate the immune response and decrease inflammatory cytokines.8, 9, 10 These effects have generated significant interest in their potential utilization in the treatment of COVID-19 infections.11 Bacterial co-infections occur in less than 4% of patients hospitalized with COVID-19. More so, the relationship between COVID-19 and bacterial secondary infections is still a grey area.7 , 12 Notwithstanding, most patients with COVID-19 infection receive antibiotics regardless of the rarity of bacterial co-infections.13 Common amongst the frequently used antibiotics for COVID-19 infections are fluoroquinolones, macrolides and cephalosporins.14 Even though antibiotics are mostly not needed in the treatment or prevention of COVID-19, a behavioral insights research in nine countries within the European region validates the increased use of antibiotic with increase in COVID-19 infection cases. The study goes on to report that 79%–96% of the antibiotic consumers are not infected with COVID-19 but were taking it as prophylaxis under the misconception that it could save them from COVID-19 15 The novelty and perceived fatalities of the coronavirus infection in the wake of the pandemic, has significantly advanced the unwarranted use of antibiotics.15 A cause for concern as the misuse and overuse of antimicrobials is a significant factor found to be accelerating the life-threatening process of antimicrobial resistance.16 The World Health Organization (WHO) describes antimicrobial resistance (AMR) as the “ability of a microorganism (bacteria, viruses, parasites etc) to stop an antimicrobial (antibiotics, antivirals and antimalarials) from working against it”.17 The culminating effect of AMR advanced by the unwarranted use of antibiotics is that conventional treatments are no longer effective. Thus, unresolved infections may result in death or may easily spread to others.18 The unwarranted use of antibiotics in this context can be described as the inappropriate use of antibiotics through the overuse, unlicensed use, overprescribing and non-prescription use of antibiotics. Emerging scientific literature on COVID-19 suggested various factors as key drivers of the unwarranted use of antibiotics in managing COVID-19 infection. These factors include medical overuse in hospital settings,19 psychological distress causing fear and social reasons such as the limited knowledge about antibiotics.20 Generally, the prevalence of the unwarranted use of antibiotics and its corresponding drivers and barriers have been well established in literature.21,22,23 However, there have been limited reviews identifying the predisposing factors precipitating this trend of antibiotic misuse for the management of COVID-19 within the context of the pandemic613. This study sought to identify and classify the factors associated with the use of antibiotics in the management of COVID-19 from published literature and the perspectives of key stakeholders within the framework of the Biopsychosocial model. The Biopsychosocial model which was primarily conceptualized by George Engel posits that comprehending an individual's medical situation does not rest solely on the biological factors but also on the psychological and social factors.24 The model is commonly used in addressing disease states and improving clinical outcomes by sensitizing clinicians of to the interaction among the biological, psychological, sociocultural, and spiritual factors in the management of diseases.25 2 Methods 2.1 Study design A scoping review was conducted between November 2021 and January 2022 following the methodological framework proposed by Arskey and O'Malley26 and advanced by Levac et al.27 The six-stage methodological process included: a) identification of research questions, b) identification of relevant research articles, c) selection of studies, d) data charting and synthesis, e) summary, discussion and analysis, and f) stakeholder consultations. The data was collated following the systematic collection and analysis of literature using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Scoping Reviews (PRISMA-ScR).28 The study was approved by the Ghana Health Service Ethics Committee on 11th October 2021 (GHS-ERC: 008/05/21) and the review protocol was registered in Open Science Framework29 2.1.1 Identification of research questions The scoping review focused on investigating the inappropriate use of antibiotics in the management of COVID-19 from published literature and stakeholder perspectives. This was guided by the following research questions: (i) What are the types of antibiotics frequently used for COVID-19 prevention and treatment? (ii) What is the geographical distribution of unwarranted antibiotic use for COVID-19? (iii) What are the associated factors facilitating the unwarranted use of antibiotics in managing COVID-19 based on the bio-psychosocial framework? 2.1.2 Identification of relevant research articles Literature for this review were identified and assessed by searching the following relevant databases: PubMed/MEDLINE, Scopus, Embase and Google Scholar. Scientific papers published after 31st December 2019 which was the date the World Health Organization (WHO) received its first communication on COVID-1930 up until January 2022 were searched. The strategy adopted the use of the following Medical Subject Headings (MeSH) to search for articles in PubMed.• COVID-19: This included the various terms for describing COVID-19; coronavirus, SARS-CoV-2, covid, pandemic • Unwarranted: This included terms/synonyms for: unwarranted, misuse, overuse, abuse, self-medication, self-treatment, non-prescription • Antibiotics: This included terms/synonyms for: antibiotics, antimicrobial, antibacterial, dispensing, medication, treatment • Antibiotic Resistance: This included the terms used to describe antibiotic resistance; antimicrobial resistance, AMR, resistance, antibiotic failure • Factors: This will include terms/synonyms for: factors, facilitators, barriers, drivers, and determinants. These terms also formed part of the appropriate keywords adopted to search for articles in Google Scholar, Scopus and EMBASE. Boolean operators were employed in the combination of the various key search terms to ensure that the selected articles reflected both antibiotic use and COVID-19 whilst capturing both biological (medical) and psychosocial perspectives. 2.2 Inclusion and exclusion criteria For this review, unwarranted use of antibiotics which is the inappropriate use of antibiotics31,32 was recognized as the misuse, overuse, unlicensed use, overprescribing and non-prescription use of antibiotics. The pre-set criteria for the selection of articles comprised: 2.2.1 Inclusion criteria • Empirical studies: randomised controlled trials (RCTs), quasi-experimental studies, cross-sectional, case-control or cohort studies. 2.2.2 Exclusion criteria • Opinion pieces, theoretical publications, and review articles • Studies that were unavailable in English 2.3 Selection of studies Two members of the research team (RAN and IAK) independently assessed the eligibility status of the various study papers. The assessment was done procedurally by a title and abstract screening followed by a full text screening (Fig. 1 ). Discrepancies on the eligibility of an article for study inclusion, were resolved through discussion. The total number of articles identified during the preliminary search was 10,252. After the removal of duplicates and the screening process in line with the pre-determined eligibility criteria, twelve (12) articles were selected for analysis.Fig. 1 Presents a Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Scoping Reviews (PRISMA-ScR) flow diagram showing the process of searching and selecting the research articles. Fig. 1 2.4 Data charting and synthesis The data was extracted from the selected articles using a predesigned extraction sheet developed in Microsoft Excel. The information extracted into the adopted form included the citation, study type, study location, population, sample size, methods, forms of unwarranted antibiotic use, prevalence and COVID-19 associated reasons for antibiotic misuse. Two members of the research team conducted the data extraction, and the data chart was verified independently by another member of the team (Data charting: Additional file 3). The selected studies were assessed for quality and risk of bias using the Mixed Methods Appraisal Tool (MMAT) version 28.33 2.5 Stakeholder consultations Consultations with stakeholders were conducted as part of the six-stage methodological framework proposed by Arskey and O'Malley.34 The consultations were held as a means of validating the study findings and contextualizing the study within the jurisdiction of study. The discussions also offered the additional advantage of communicating the findings of the review and creating the necessary foundation for the development of interventional strategies in line with advancements proposed by Levac.35 The stakeholders were purposively selected based on their institutional affiliation or position they occupied and the relevance to antibiotic stewardship and COVID-19 infection management. The professionals included doctors and pharmacists from the Ministry of Health, Food and Drugs Authority, academia, Pharmacy Council, Teaching Hospitals and the National COVID-19 Committee. The participants were invited to partake in either a telephone interview or face-to-face interview. The time allotted for both interview formats were the same and lasted approximately ten (10) minutes using the same questioning guide. Stakeholders were asked open-ended questions centred on the utilization of antibiotics in the context of the COVID-19 pandemic. Participants reached via the telephone offered verbal consent whilst those reached via a face-to-face interaction signed an informed consent form for the interview. The consultative meetings were recorded and transcribed. Finally, the main thematic areas of the consultations were synthesized and categorized using the elements of the biopsychosocial framework. 2.6 Quality assessment of the included studies The studies selected as part of the review were assessed for quality and risk of bias using the Mixed Methods Appraisal Tool (MMAT). The MMAT was a suitable appraisal tool for mixed studies reviews that include qualitative, quantitative, and mixed methods studies. All the included studies received a ‘Yes’ response in Part I of the MMAT checklist as a part of the prerequisite for selection. This implies that, all the selected studies had clear research questions and collected the requisite data to address these questions. Part II of the MMAT appraisal was specific and tailored to the type of study. Amongst the included studies, only one study was qualitative in design and the remaining were quantitative. The qualitative study met the entire MMAT checklist for qualitative studies. The majority of the quantitative non-randomized studies (n = 3) failed to account for the confounders in the study design and analysis. While an overall score from the ratings of each criterion was not calculated, most of the included studies met at least three of the corresponding specific MMAT criteria. The summary of the MMAT quality assessment is presented in Additional file 2. 3 Results 3.1 Biological (clinical) factors In this categorization, fever, dyspnoea and a productive cough were associated with antibiotic use for COVID-19 treatment11,39 Other studies also demonstrated similar relationships between the use of antibiotics for COVID-19 treatment and clinical markers like a confirmed positive PCR test.38 , 39 , 47 In the case of Baghdadi38 et al., a study in the United States illustrated the likely overuse of Ceftriaxone (48.5%) and Azithromycin (46%) in treating COVID-19 hospital. in-patients not diagnosed with bacterial infection. Similarly, Van Laethem39 et al. reported the unwarranted presumptive use of penicillin/beta-lactamase combinations for treating hospitalized COVID-19 patients in Belgium. The hospitalization of COVID-19 patients also appeared to be a basis warranting the use of antibiotics. In a study conducted in the Netherlands, the unwarranted use of antibiotics amongst hospitalized patients was reported, despite the low incidence of bacterial co-infections.47 3.2 Psychological factors The fear of leaving home, going to crowded places and getting infected were reported to be enablers of antibiotic self-medication in a study conducted in Iran.42 Similar patterns were observed in Nigeria where aside the fear of medical centres, participants also reported the fear of stigmatization, discrimination, quarantine and delays at the hospital.44 In this regard, the anxiety resulting from the various COVID-induced fears had caused Nigerians to resort to self-medication for purported prophylactic reasons. The fear of death and infections coupled with the need to manage emotions in the context of the pandemic also contributed to inappropriate antibiotic consumption patterns.46 An Australian study seeking to explore the relationship between psychological distress and perceived health risks; also reported the cardinal role of COVID-19 associated psychological distress as an enabler of self-medication with antibiotics.45 Perceived superiority of antibiotics as the perfect solution to inexplicable clinical infections also drove the inappropriate use of antibiotics for COVID-19. Fear and the desire for patient compliance were amongst the several factors influencing the antibiotic dispensing patterns of healthcare providers during the pandemic. This resulted in the substantial dispensing of antibiotics without the requisite clinical indications and for longer durations than appropriate.36 3.3 Social factors The preponderance of unwarranted antibiotic use as an antidote to the COVID pandemic was equally enabled by various structural factors such as, the congestion and delays in receiving treatment at medical centres.44 Heydargoy42 corroborated this observation by reporting COVID-induced fears generated from the crowding at medical centres. The media plays a critical role in the social environment influencing unwarranted antibiotic use. For example, social media deliberations on the efficacy of taking antibiotics was a pivotal determinant on unwarranted antibiotic use.46 Wegbom et al. also highlights the stress-inducing influence of the media in addition to the shortage of medicines, and stigmatization as external factors surrounding the inappropriate use of antibiotics.44 There were conflicting reports on the role of gender in the use of antibiotics during the pandemic. Oikonomou48 reported a lack of association between gender and parents' attitude on antibiotic use for children in the course of the pandemic. However, Sadio37 et al. records a correlation between being female and self-medication. The observation of females exhibiting a higher propensity to self-medication with antibiotics is equally reported in a study conducted by Wegbom44 et al. in Nigeria. Self-medication was also found to positively correlate with education and the professional affiliation as a health worker. Health professionals and people with a higher level of education were identified to most likely practice self-medication with antibiotics.37 3.4 Stakeholder consultations Two (2) participants consented to a face-to-face interview and five (5) partook in a telephone interview. All the participants had at least a university degree and occupied focal positions relevant to the control, stewardship of antibiotics and COVID management. The participants represented the following institutions in Ghana: (i) Food and Drugs Authority, (ii) Ministry of Health, (iii) Pharmacy Council, (iv) Mental Health, (v) Academia, (vi) Quaternary Hospital and (vii) the National Guidelines for COVID-19 Committee. Various themes emerged in the interview accounts of antibiotic utilization within the context of the ongoing COVID-19 pandemic. The thematic analyses was conducted in line with the six-step approach proposed by Braun and Clarke.49 Identifying the elements of the Biopsychosocial framework as the main thematic areas, the emerging themes were classified in accordance with this framework as summarized in Table 3 and narrated next. 3.5 Thematic results Common to the various interview accounts were the issues of fear and anxiety driving the overuse, misuse, or abuse of antibiotics in the ongoing pandemic. Participants described the pandemic situation as a life and death situation which had no room for rational antibiotic use. The fear and anxiety were not restricted to the lay community but also health workers at large. This resulted in a subtle disregard for rational assessment. A health worker commented; “We all didn't seem to know what was working. If today we hear that Azithromycin is good, we would use it a bit more freely especially when we find that these antibiotics are safe.” (Respondent 3). “I recall my own COVID experience. I knew I didn't need antibiotics despite all my knowledge and what I have been telling people; I said, You know what, I'm coughing and the cough wasn't getting any better. Okay, the sputum has changed color so guess what, I went for antibiotics. Whether it made a difference or not. But definitely COVID has changed how people use antibiotics” (Respondent 6). Another healthcare professional in justifying antibiotic supply remarked: “We didn't know what we were fighting against, so antibiotics had to be given out. The state in which patients came in, we wouldn't want to wait for last minute before prescribing antibiotics. Also, we didn't have proper guidelines, everyone just wanted to save their patient.” (Respondent 7). The confusion on clinically viable options resulted in antibiotic use not only for the perceived microbiological action but as a means of placating patient and health worker anxiety. The weak regulatory system was another common factor propelling inappropriate antibiotic stewardship. A participant commented, “And we know pharmacies are not regulated as effectively as we would wish they would. So, people can definitely go and get these antibiotics easily.” (Respondent 3). The re-classification of some antibiotics from Prescription Only Medicine (POM) to Pharmacy Only Medicines was also described as a justification for unwarranted antibiotic access. “There is unnecessary easy access to prescription only medicine. Now, Azithromycin has moved from prescription only medicine to pharmacy only medicine. You can buy without a prescription because of the COVID. And even before COVID, I am sure people could get it without a prescription.” (Respondent 2). All the respondents alluded to the existence of an AMR policy. However, some participants described this policy as impractical whereas others opined that the inability to implement this policy was the biggest hindrance to antibiotic stewardship. “The existing AMR policy is not reflecting current practices and would need to be updated.” (Respondent 7). “It is an AMR policy and should include viruses/viral infections, but I don't think the AMR policy is positioned to handle a viral outbreak to the level of a pandemic.” (Respondent 4). “Existing AMR policy in Ghana. We need to review the existing AMR policy in Ghana and see if it is working or are we really working with it.” (Respondent 5). The nationally endorsed framework of antibiotic use in COVID management was also another issue reported as contributing to antibiotic misuse. “There was this issue too, that well even though the evidence doesn't support it, it is still in the treatment guidelines for Ghana. It wasn't in the WHO one, yet still Ghana adopted the routine use of antibiotics. For most other places you gave antibiotics when it was indicated i.e., when there is evidence of a bacterial infection. Because Ghana had adopted this routine use, people who hadn't tested positive but had the least cough, went in for antibiotics.” (Respondent 6). 3.6 Stakeholder recommendations • Educate prescribers more on the need practise targeted means of infection management. • Ensure that nationally endorsed protocols for antibiotic usage are in line with evidence-based best practices. • Address mental health issues during crisis of such nature: educate people on basic mental health strategies to deal with fear. • Implement existing AMR policy. • Simulate future public health emergencies and develop interventional strategies. • Incorporate antimicrobial surveillance and resistance policy as a core part of the curricula in training health professionals. • Review the AMR policy to meet the realities of clinical practice in Ghana. • Increase public education on rational antibiotic use 4 Discussion This review highlights several relationships between the custodial responsibilities and use of antibiotics in line with the ongoing COVID-19 pandemic (Table 1 ). Various studies have sought to establish the appropriate use of antibiotics in COVID-19 whilst highlighting the effectiveness and safety of antibiotics in Sars-Cov-2 treatment.14 , 50, 51, 52, 53 Table 1 Characteristics of the included studies in this Scoping Review. Table 1Article Study Location Study Design COVID Diagnosis Source of Antibiotic Misuse Common Antibiotics Used Factors associated with inappropriate antibiotic use Elsayed et al.36 Egypt Cross-Sectional (Questionnaires) Unconfirmed (Presumptive) Physician/Pharmacist Recommendation ⁃ Azithromycin (40%) ⁃ Ceftriaxone ⁃ Linezolid ⁃ Fear and seeking patient compliance Estrada et al.11 Spain Observational Retrospective Study Confirmed Doctor Prescriptions (Empirical) ⁃ Beta-lactams (72.0%) ⁃ Macrolides (60.2%) ⁃ Fluoroquinolones (13.3%), ⁃ Symptomatic profile of patient especially fever, dyspnea and a productive cough Sadio et al.37 Togo Cross-Sectional (Questionnaires) Unconfirmed Non-Prescription (Self-medication) ⁃ Azithromycin (1.2%) ⁃ Gender (Female) ⁃ Working in the health sector ⁃ Educational Level Baghdadi et al.38 United States Retrospective Observational Cohort Study Confirmed Doctor's Prescription (Overuse) ⁃ Ceftriaxone (48.5%) ⁃ Azithromycin (46.0%) ⁃ Vancomycin (22.9%) ⁃ First wave of the COVID-19 pandemic Van Laethem et al.39 Belgium Retrospective Quantitative Study Confirmed Doctor's Prescription ⁃ Penicillin with beta-lactamase inhibitor ⁃ Longer hospital stay ⁃ Presence of fever and low SpO2 ⁃ Pre-existing pulmonary disease Karami et al.40 Netherlands Retrospective Observational Cohort Study Confirmed Doctor Prescriptions (Empirical) Second and third generation cephalosporins Akhtar et al.41 Pakistan Retrospective observational study Confirmed Doctor Prescriptions ⁃ Azithromycin (88.6%) ⁃ Ceftriaxone (23.6%). Heydargoy42 Iran Online Questionnaire posted on all social networks of target group Unconfirmed Non-Prescription/OTC Not indicated ⁃ Fear of leaving home ⁃ Fear of going to crowded places especially medical centres Abdela et al.43 Ethiopia Retrospective Cohort Study Design Confirmed Not indicated ⁃Amoxicillin/Clavulanate and Azithromycin (Most used oral antibiotics) ⁃ Ceftriaxone and Vancomycin (most used iv antibiotics) Not Indicated Wegbom et al.44 Nigeria Web-based cross-sectional survey using a self-reported questionnaire Not Indicated Over the Counter Not Indicated ⁃ Fear of stigmatization or discrimination ⁃ Fear of being quarantined ⁃ Fear of contact with an infected person ⁃ Emergency illness and delay of hospital services ⁃ Gender ⁃ Educational attainment ⁃ Knowledge level on Self Medication Zhang et al.45 Australia Online Survey Unconfirmed Non-Prescription (Self-Medication) Not Indicated ⁃ Psychological distress associated with COVID-19 (panic and fear) ⁃ Lack of knowledge about the correct therapeutic role of antibiotics ⁃ Previous inappropriate use of antibiotics ⁃ Doctor-patient relationship Kalam et al.46 Bangladesh Qualitative telephone interviews Confirmed (n = 20) and Unconfirmed (n = 20) Non-Prescription (Self-Medication) Not indicated ⁃ Perceived superiority of antibiotics ⁃ Informal sources of treatment advice ⁃ Lack of access to COVID-19 testing and healthcare services ⁃ Dealing with social repercussions of symptoms, diagnosis, and isolation However, not many studies have thoroughly examined the inappropriate use of antibiotics and the corresponding threat it poses to health security. In the context of the ongoing pandemic, the misuse of antibiotics can be characterized in many forms including the overuse of antibiotics, incorrect dosing, incorrect antibiotic combinations and wrong indications.36 The few studies that sought to explore this upsurge of antibiotics misuse were mainly commentaries, opinion pieces and editorial letters which were excluded from the scope of this study54 , 55 , 56 , 57 , 58 , 59. This study does not only elucidate the misappropriate use of antibiotics but also presents the utilization of antibiotics vis a vis the social complexities of the ongoing pandemic. The identification of the interplay between social factors and antibiotic use for COVID-19 is in line with the findings of Toro-Alzate, Hofstraat, & de Vries, (2021) in their study of the social relationships between COVID-19 and antimicrobial resistance based on the SPECIAL SOC AMR Framework. This current scoping review using the Biopsychosocial model has illustrated the inappropriate utilization of antibiotics through the lens of a biological (clinical), social and psychological categorization (Table 2 ).Table 2 Key findings following the Biopsychosocial framework. Table 2Results (Factors) Sub-themes Biopsychosocial Theme ⁃ Symptom profile suggesting COVID infection. ⁃ Presence of a fever ⁃ Positive PCR test for COVID-19 ⁃ Presence of co-morbidities. ⁃ Hospital Admission ⁃ Wave of COVID-19 pandemic ⁃ Prescription practices of General Practitioners encouraging antibiotic hoarding Clinical signs and symptoms, laboratory markers, medical protocols Biological (Clinical) ⁃ Fear of infection or contact with an infected person ⁃ Fear of crowds and infections at medical centres ⁃ Fear of being quarantined ⁃ Stigmatization or Discrimination ⁃ Emotional anxiety ⁃ Social repercussions of symptoms, diagnosis, and infection ⁃ COVID-19 associated psychological distress ⁃ Fear and desiring patient compliance ⁃ Perceived superiority of antibiotics as the ultimate solution Anxiety, fear, perceptions Psychological ⁃ Uncertainty and difficulty in accessing COVID-19 and microbiological tests ⁃ Career in the health sector ⁃ Delays receiving hospital services ⁃ Distance to the health facility ⁃ Knowledge of antibiotics ⁃ High School Education or higher ⁃ Experiences in using antibiotics ⁃ Relationship with doctors ⁃ Gender (Female) ⁃ Influence of media ⁃ Informal sources of treatment advice ⁃ Medicine shortages Education, legislation, health systems, gender, career, peer pressure, community structures, media Social Table 3 Stakeholder thematic results characterized by the Biopsychosocial framework. Factors (enablers) of Antibiotic Misuse for COVID-19 Management Table 3Biopsychosocial Theme Sub-theme Explanation BIOLOGICAL (CLINICAL) Drug Repurposing The repurposing of already existing drugs such as antibiotics justifies its sporadic use. Confusion on viable clinical options In the absence of any other viable clinical antidotes to COVID, antibiotics are used not only for a microbiological action but also a placebo effect. Antibiotic Cocktails People were combining different antibiotics or modifying antibiotic dosing schedules as a curative approach to COVID-19. Perceived Safety of Antibiotics Antibiotics are safe and there is no harm taking it even when you are unsure about the COVID infection. PSYCHOLOGICAL Heightened Fear The fear of infection surrounding the pandemic motivated the use of antibiotics as an antidote. Anxiety People were anxious with the least symptom mimicking a cough or cold. Antibiotics acted as anxiolytics in this regard. Treatment Panic The reports of COVID-related deaths created a sense of panic which resulted in the non-rational use of antibiotics as a kneejerk reaction. SOCIAL Treatment Publicity Widespread publicity of COVID treatment protocols reinforced public decision to use antibiotics for self-medication. Weak Regulatory System Pharmacies are not regulated as effectively as required. Hence, the creation of an unwarranted access pathway to antibiotics. Drug Classification Antibiotics such as Azithromycin have moved from prescription only medicine to pharmacy only medicine. Hence, can now be bought without a prescription because of the COVID Azithromycin Hype The social hype about Azithromycin characterizes it as a COVID wonder drug. Public Knowledge Deficit Most people are unable to differentiate between a bacterial infection and a viral infection. COVID is an infection, so they treat with antibiotics. Weak AMR Policy Implementation The AMR policy is not well positioned to address current practices in antibiotic utilization and is currently facing challenges in implementation. Economic Incentives Pharmacy business owners found an opportunity to cash in with the sale of antibiotics. National Endorsement Not entirely in line with evidence-based practice, Ghana made routine antibiotic use a core part of their COVID management protocol. Media Influence Media especially social media was awash with so many things including antibiotics that were purported to help people recover or even prevent COVID. The geographical representation of the included studies is sparsely distributed across a wide region (Egypt, Bangladesh, Spain, Togo, United States, Belgium, Netherlands, Greece, Pakistan, Iran, Ethiopia, Nigeria, and Australia). The scarcity of primary studies that outline the enablers of antibiotic misuse as a coping strategy to the COVID-19 pandemic is evident. Approximately 67% of the included studies were from the African and European Region. The studies within the European region were retrospective in design and could not comprehensively identify the factors associated with the misuse of antibiotics. The study conducted by Baghdadi et al. in the United States also exhibited a similar profile to that in the European region.38 These are developed settings with robust data systems which may explain the feasibility of the identified retrospective data reports. Trends in the African region as evidenced from studies by Wegbom et al. (2021) and Sadio et al. (2021) mostly reflected the non-prescription use of antibiotics as a means of self-medication. This could be attributed to the relatively weaker regulatory structures facilitating the unlicensed access to prescription medicines in Africa.61 Azithromycin was the most common antibiotic used as a prophylactic and curative means of managing COVID-19 infections. This may be due to its reported potential antiviral and anti-inflammatory properties.62 Trends of increasing azithromycin overuse during the pandemic especially as a result of media posts were also reported by Bogdanić, Močibob, Vidović, Soldo, & Begovać, (2022). Despite the promising evidence supporting the use of Azithromycin as a prospective therapeutic agent for COVID-19, concerns are raised on the widespread unreasonable antibiotic therapy in COVID-19 patients.64 Despite the established clinical basis for antibiotic utilization, the use of antibiotics in this context characterized as unwarranted or misused is as a result of various factors. These include clinical markers such as the existence of co-morbidities,39 productive coughs65 and a fever.39 , 65 Beyond clinical signs, noteworthy is the psychological distress associated with the COVID-19 pandemic which enabled trends of substance or medicines abuse.66 Substance abuse which can also be characterized as self-medication is a common coping strategy adopted by individuals experiencing a form of psychological distress.67 The resultant effects are various health and social implications for a huge number of individuals.68 These patterns of self-medication do not differ significantly from the trends of substance use as a result of anxiety disorders. The self-medication hypothesis is posited as a common explanation for the observed comorbidity of anxiety disorders and substance use disorders.69 This hypothesis posits that, individuals with mood or anxiety disorders will use substances to cope with the difficult symptoms prevailing at that time. Nevertheless, the psychological disposition influencing the misuse as a coping strategy to the pandemic is not restricted only to the antibiotic user but also the suppliers of antibiotics as evidently corroborated by the stakeholder consultative discussions. The health system structure, delays at the hospitals and socio-economic implications have also been reported as established determinants of self-medication.70, 71, 72 This mimics the current observed trend of unwarranted antibiotic use in response to the external factors such as, the congestion and delays in receiving treatment at medical centres.44 Most especially in low- and middle-income countries (LMICs) where the inadequacies of a resource-constrained healthcare environment contribute strongly to the different and emerging factors associated with the non-prescription use of antibiotics.73 , 74 Beyond health systems, the stress-inducing influence of the media coupled with challenges such as shortage of essential commodities and stigmatization are also determinants underpinning the inappropriate use of antibiotics.44 Generally, the COVID-19 pandemic has created a seemingly hostile social environment that triggers a turmoil of distress as well as a myriad of considerable fears.75 , 76 The COVID-19 related media reportage is a critical aspect of this hostile environment and that unearths inexplicable anxieties, depression and fears.76 The role of gender in enabling the misuse of antibiotics cannot be excluded in the social categorization of this review. The association between being female and self-medication is not foreign to existing literature. Carrasco-Garrido in offering a gender perspective on self-medication in Spain reports a prevalence of self-medication amongst women superseding that of men.77 This reported trend may be explained by the preponderance of females demonstrating greater fear and anxiety than their male counterparts.78 These fears and anxieties have heightened amongst women in the context of the ongoing COVID-19 outbreak.79 , 80 Self-medication was also found to positively correlate with education and the professional affiliation as a health worker. Health professionals and people with a higher level of education were identified to most likely practice self-medication with antibiotics.37 Conclusively, the observations of this review correspond with that of Buckner et al. in a study outlining the biopsychosocial model of social anxiety and substance use.81 In their study, chronically elevated negative affective states and fear of scrutiny were aspects of social anxiety working in concert to place vulnerable individuals at risk for substance use.81 In the case of this study, the multiple reported fears and anxieties characterized by the novel coronavirus pandemic coupled with biological indicators ambiguously suggesting an infection and social impediments such as hospital congestions have resulted in the inappropriate supply and use of antibiotics as a coping strategy to the pandemic. 5 Limitations Common to the limitations of the Biopsychosocial model, the degree of influence exerted by each component of factors whether biological, social, or psychological is unaccounted for. The majority of the cross-sectional studies included in this review were conducted using web-based or online platforms which skewed respondents towards those who were educated and/or belonging to an economic class capable of affording an internet-enabled phone from which data were collected. Despite these limitations, this review has shown that interventions such as policies, guidelines and strategies targeted at the control of antibiotics in the course of the COVID-19 pandemic and future pandemics should simultaneously address the three major thematic areas of the biopsychosocial model while taking into cognizance country-specific peculiarities. In addition, including the stakeholder consultations as part of this review also highlighted key clinical practice and policy perspectives to the review findings on the use and misuse of antibiotics during the COVID-19 pandemic. 6 Conclusion As maintained by the biopsychosocial model, the COVID-necessitated use of antibiotics is a result of an extensive interplay of social, psychological, and biological (clinical) factors. The pandemic-enabled determinants of unwarranted antibiotic use common to the review and stakeholder consultations include fear, emotional anxiety, media influences and public knowledge deficits. The stakeholder consultations also outlined critical issues such as the need to evaluate the effectiveness of the antimicrobial policy, the importance of mental health in pandemics and the national call to action for evidence-based medical practice despite prevailing anxieties. The study findings serve as the basis for further research into the sociocultural and psychological enablers of inappropriate antibiotic use as a rationale for developing targeted and comprehensive policy interventions in antibiotic stewardship. Authors' contributions RAN designed the study, screened the studies, searched literature, appraised the quality of the papers, conducted the stakeholder consultations, analyzed the data, and drafted the manuscript. IAK screened the studies, analyzed the data, and supervised the study. AK appraised the quality of the papers, analyzed the data, and revised the manuscript. KOB revised the study design, revised the manuscript, and supervised the study. 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. Appendix A Supplementary data The following are the supplementary data to this article:Multimedia component 1 Multimedia component 1 Multimedia component 2 Multimedia component 2 Multimedia component 3 Multimedia component 3 Acknowledgements The authors are grateful to the stakeholders for the consultations. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.sapharm.2022.11.011. ==== Refs References 1 World Health Organization (WHO) Coronavirus https://www.who.int/health-topics/coronavirus#tab=tab_1 2020 2 Devi S. SARS-CoV-2 evolution Lancet Infect Dis 21 4 2021 467 https://www.who.int/news-room/q-a-detail/sars-cov-2-evolution 3 Who Clinical Management Clinical Management Living Guidance COVID-19 2021 World Heal Organ (January) 4 Yacouba A. Olowo-okere A. Yunusa I. Repurposing of antibiotics for clinical management of COVID-19: a narrative review Ann Clin Microbiol Antimicrob 20 1 2021 37 10.1186/s12941-021-00444-9 34020659 5 Poddighe D. Aljofan M. Clinical evidences on the antiviral properties of macrolide antibiotics in the COVID-19 era and beyond Antivir Chem Chemother 28 2020 10.1177/2040206620961712 6 Chedid M. Waked R. Haddad E. Chetata N. Saliba G. Choucair J. Antibiotics in treatment of COVID-19 complications: a review of frequency, indications, and efficacy J Infect Public Health 14 5 2021 570 576 10.1016/j.jiph.2021.02.001 33848886 7 Feldman C. Anderson R. The role of co-infections and secondary infections in patients with COVID-19 Pneumonia 13 1 2021 10.1186/s41479-021-00083-w 8 Ohe M. Shida H. Jodo S. Macrolide treatment for COVID-19: will this be the way forward? Biosci Trends 14 2 2020 159 160 10.5582/bst.2020.03058 32249257 9 Min J.Y. Jang Y.J. Macrolide therapy in respiratory viral infections Mediat Inflamm 2012 2012 10.1155/2012/649570 10 Bosseboeuf E. Aubry M. Nhan T. Azithromycin inhibits the replication of zika virus J Antivir Antiretrovir 10 1 2018 10.4172/1948-5964.1000173 11 Bendala Estrada A.D. Calderón Parra J. Fernández Carracedo E. Inadequate use of antibiotics in the covid-19 era: effectiveness of antibiotic therapy BMC Infect Dis 21 1 2021 1 23 10.1186/s12879-021-06821-1 33390160 12 Westblade L.F. Simon M.S. Satlin M.J. Bacterial coinfections in coronavirus disease 2019 Trends Microbiol 2021 1 12 10.1016/j.tim.2021.03.018 13 Calderón-Parra J. Muiño-Miguez A. Bendala-Estrada A.D. Inappropriate antibiotic use in the COVID-19 era: factors associated with inappropriate prescribing and secondary complications. Analysis of the registry SEMI-COVID PLoS One 16 5 May 2021 e0251340 10.1371/journal.pone.0251340 14 Chedid M. Waked R. Haddad E. Chetata N. Saliba G. Choucair J. Antibiotics in treatment of COVID-19 complications: a review of frequency, indications, and efficacy J Infect Public Health 14 5 2021 570 576 10.1016/j.jiph.2021.02.001 33848886 15 WHO Regional Office for Europe Preventing the COVID-19 pandemic from causing an antibiotic resistance catastrophe WHO news release November 2020 15 17 https://www.euro.who.int/en/health-topics/disease-prevention/antimicrobial-resistance/news/news/2020/11/preventing-the-covid-19-pandemic-from-causing-an-antibiotic-resistance-catastrophe 16 World Health Organization (Who) Antimicrobial resistance. Global action plan on AMR https://www.who.int/news-room/fact-sheets/detail/antimicrobial-resistance 2020 17 World Health Organization WHO report on surveillance of antibiotic consumption https://apps.who.int/iris/bitstream/handle/10665/277359/9789241514880-eng.pdf 2018 18 World Health Organization (Who) Global Antimicrobial Resistance Surveillance System (GLASS) Report: Early Implementation 2019 132 136 https://apps.who.int/iris/bitstream/handle/10665/259744/9789241513449-eng.pdf;jsessionid=E86D72149B83FABC12ADD26972D1FDA1?sequence=1 19 Langford B.J. So M. Raybardhan S. Antibiotic prescribing in patients with COVID-19: rapid review and meta-analysis Clin Microbiol Infect 27 4 2021 520 531 10.1016/j.cmi.2020.12.018 33418017 20 Zhang A. Hobman E.V. De Barro P. Young A. Carter D.J. Byrne M. Self-medication with antibiotics for protection against COVID-19: the role of psychological distress, knowledge of, and experiences with antibiotics Antibiotics 10 3 2021 1 14 10.3390/antibiotics10030232 21 Torres N.F. Chibi B. Middleton L.E. Solomon V.P. Mashamba-Thompson T.P. Evidence of factors influencing self-medication with antibiotics in low and middle-income countries: a systematic scoping review Publ Health 168 2019 92 101 10.1016/j.puhe.2018.11.018 22 Mahmoud M.A. Aldhaeefi M. Sheikh A. Aljadhey H. Community pharmacists perspectives about reasons behind antibiotics dispensing without prescription: a qualitative study Biomed Res 29 21 2018 3792 3796 10.4066/biomedicalresearch.29-18-1112 23 Barker A.K. Brown K. Ahsan M. Sengupta S. Safdar N. Social determinants of antibiotic misuse: a qualitative study of community members in Haryana, India BMC Publ Health 17 1 2017 1 9 10.1186/s12889-017-4261-4 24 Borell-Carrió F. Suchman A.L. Epstein R.M. The biopsychosocial model 25 years later: principles, practice, and scientific inquiry Ann Fam Med 2 6 2004 576 582 10.1370/afm.245 15576544 25 Kusnanto H. Agustian D. Hilmanto D. Biopsychosocial model of illnesses in primary care: a hermeneutic literature review J Fam Med Prim Care 7 3 2018 497 10.4103/jfmpc.jfmpc_145_17 26 Arksey H. O'Malley L. Scoping studies: towards a methodological framework Int J Soc Res Methodol Theory Pract 8 1 2005 19 32 10.1080/1364557032000119616 27 Levac D. Colquhoun H. O'Brien K.K. Scoping studies: advancing the methodology Implement Sci 5 1 2010 1 9 10.1186/1748-5908-5-69 20047652 28 Moher D. Liberati A. Tetzlaff J. Altman D.G. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement BMJ 339 7716 2009 332 336 10.1136/bmj.b2535 29 Nortey R. Kretchy I. Buabeng K.O. Prevalence and biopsychosocial analysis of the use of antibiotics as a coping strategy for COVID-19 infections: a scoping review OSFPREPRINTS 2022 10.31219/OSF.IO/EJHZG 30 World Health Organization Novel coronavirus (2019-NCoV) 10 2020 10.13070/mm.en.10.2867 Geneva 31 Vilkman K. Lääveri T. Pakkanen S.H. Kantele A. Stand-by antibiotics encourage unwarranted use of antibiotics for travelers' diarrhea: a prospective study Trav Med Infect Dis 27 2019 64 71 10.1016/j.tmaid.2018.06.007 32 MacGeorge E.L. Caldes E.P. Smith R.A. Hackman N.M. San Jose A. Reducing unwarranted antibiotic use for pediatric acute otitis media: the influence of physicians' explanation and instruction on parent compliance with ‘watchful waiting J Appl Commun Res 45 3 2017 333 345 10.1080/00909882.2017.1320575 33 Hong Q.N. Fàbregues S. Bartlett G. The Mixed Methods Appraisal Tool (MMAT) version 2018 for information professionals and researchers Educ Inf 34 4 2018 285 291 10.3233/EFI-180221 34 Levac D. Colquhoun H. O'Brien K.K. Scoping studies: advancing the methodology Implement Sci 5 1 2010 1 9 10.1186/1748-5908-5-69 20047652 35 Levac D. Colquhoun H. O'Brien K.K. Scoping studies: advancing the methodology Implement Sci 5 1 2010 1 9 10.1186/1748-5908-5-69 20047652 36 Elsayed A.A. Darwish S.F. Zewail M.B. Mohammed M. Saeed H. Rabea H. Antibiotic misuse and compliance with infection control measures during COVID-19 pandemic in community pharmacies in Egypt Int J Clin Pract 75 6 2021 10.1111/ijcp.14081 37 Sadio A.J. Gbeasor-Komlanvi F.A. Konu R.Y. Assessment of self-medication practices in the context of the COVID-19 outbreak in Togo BMC Publ Health 21 1 2021 1 9 10.1186/s12889-020-10145-1 38 Baghdadi J.D. Coffey K.C. Adediran T. Antibiotic use and bacterial infection among inpatients in the first wave of covid-19: a retrospective cohort study of 64,691 patients Antimicrob Agents Chemother 65 11 2021 10.1128/AAC.01341-21 39 Van Laethem J. Wuyts S. Van Laere S. Antibiotic prescriptions in the context of suspected bacterial respiratory tract superinfections in the COVID-19 era: a retrospective quantitative analysis of antibiotic consumption and identification of antibiotic prescription drivers Intern Emerg Med 17 1 2022 141 151 10.1007/s11739-021-02790-0 34185257 40 Hadi M.A. Karami N.A. Al-Muwalid A.S. Community pharmacists’ knowledge, attitude, and practices towards dispensing antibiotics without prescription (DAwP): a cross-sectional survey in Makkah Province, Saudi Arabia Int J Infect Dis 47 2016 95 100 10.1016/j.ijid.2016.06.003 27343987 41 Akhtar H. Akhtar S. Rahman F.U. An overview of the treatment options used for the management of COVID-19 in Pakistan: retrospective observational study JMIR Public Heal Surveill 7 5 2021 10.2196/28594 42 Heydargoy M.H. The effect of the prevalence of covid-19 on arbitrary use of antibiotics Iran J Med Microbiol 14 4 2020 374 384 10.30699/ijmm.14.4.374 43 Abdela S.G. Liesenborghs L. Tadese F. Antibiotic overuse for COVID-19; are we adding insult to injury? Am J Trop Med Hyg 105 6 2021 1519 1520 10.4269/AJTMH.21-0603 34715676 44 Wegbom A.I. Edet C.K. Raimi O. Fagbamigbe A.F. Kiri V.A. Self-medication practices and associated factors in the prevention and/or treatment of COVID-19 virus: a population-based survey in Nigeria Front Public Health 9 2021 606801 10.3389/fpubh.2021.606801 45 Zhang A. Hobman E.V. De Barro P. Self-medication with antibiotics for protection against COVID-19: the role of psychological distress, knowledge of, and experiences with antibiotics Antibiotica 10 3 2021 232 10.3390/ANTIBIOTICS10030232 Page 232. 2021;10 46 Kalam A. Shano S. Khan M.A. Understanding the social drivers of antibiotic use during COVID-19 in Bangladesh: implications for reduction of antimicrobial resistance PLoS One 16 12 December 2021 e0261368 10.1371/journal.pone.0261368 47 Karami Z. Knoop B.T. Dofferhoff A.S.M. Few bacterial co-infections but frequent empiric antibiotic use in the early phase of hospitalized patients with COVID-19: results from a multicentre retrospective cohort study in The Netherlands Inf Disp 53 2 2021 102 110 10.1080/23744235.2020.1839672 48 Oikonomou M.E. Gkentzi D. Karatza A. Fouzas S. Vervenioti A. Dimitriou G. Parental knowledge, attitude, and practices on antibiotic use for childhood upper respiratory tract infections during covid-19 pandemic in Greece Antibiotics 10 7 2021 10.3390/antibiotics10070802 49 Clarke V. Braun V. Hayfield N. Qualitative psychology: a practical guide to research methods https://books.google.com.gh/books?hl=en&lr=&id=lv0aCAAAQBAJ&oi=fnd&pg=PA222&dq=Virginia+Braun+and+Victoria+Clarke&ots=eOLLcykqLw&sig=eRpoCLbwHw49_8sU2uEmlyzybfE&redir_esc=y#v=onepage&q=Virginia Braun and Victoria Clarke&f=false 2015 50 Popp M. Stegemann M. Riemer M. Antibiotics for the treatment of COVID-19 Cochrane Database Syst Rev 10 2021 2021 10.1002/14651858.CD015025 51 Ai J. Li Y. Zhou X. Zhang W. COVID-19: treating and managing severe cases Cell Res 30 5 2020 370 371 10.1038/s41422-020-0329-2 32350393 52 Adebisi Y.A. Jimoh N.D. Ogunkola I.O. The use of antibiotics in COVID-19 management: a rapid review of national treatment guidelines in 10 African countries Trop Med Health 49 1 2021 1 5 10.1186/s41182-021-00344-w 33397511 53 Liu C. Wen Y. Wan W. Lei J. Jiang X. Clinical characteristics and antibiotics treatment in suspected bacterial infection patients with COVID-19 Int Immunopharm 90 2021 107157 10.1016/j.intimp.2020.107157 54 Shin D.H. Kang M. Song K.H. Jung J. Kim E.S. Kim H Bin A call for antimicrobial stewardship in patients with COVID-19: a nationwide cohort study in Korea Clin Microbiol Infect 27 4 2021 653 655 10.1016/j.cmi.2020.10.024 33137513 55 Rizk N.A. Moghnieh R. Haddad N. Challenges to antimicrobial stewardship in the countries of the arab league: concerns of worsening resistance during the COVID-19 pandemic and proposed solutions Antibiotics 11 2021 10 10.3390/antibiotics10111320 35052885 56 Jahanshahlou F. Hosseini M.S. Antibiotic resistance: a disregarded concern for misuse of azithromycin in COVID-19 treatment J Res Med Sci 26 1 2021 101 10.4103/jrms.JRMS_1124_20 34899939 57 Garg S.K. Antibiotic misuse during COVID-19 pandemic: a recipe for disaster Indian J Crit Care Med 25 6 2021 617 619 10.5005/jp-journals-10071-23862 34316138 58 Ukuhor H.O. The interrelationships between antimicrobial resistance, COVID-19, past, and future pandemics J Infect Public Health 14 1 2021 53 60 10.1016/j.jiph.2020.10.018 33341485 59 Wagner C.E. Prentice J.A. Saad-Roy C.M. Economic and behavioral influencers of vaccination and antimicrobial use Front Public Health 8 2020 614113 10.3389/fpubh.2020.614113 61 Ocan M. Obuku E.A. Bwanga F. Household antimicrobial self-medication: a systematic review and meta-analysis of the burden, risk factors and outcomes in developing countries BMC Publ Health 15 1 2015 1 11 10.1186/s12889-015-2109-3 62 Butler C.C. Dorward J. Yu L.M. Azithromycin for community treatment of suspected COVID-19 in people at increased risk of an adverse clinical course in the UK (PRINCIPLE): a randomised, controlled, open-label, adaptive platform trial Lancet 397 10279 2021 1063 1074 10.1016/S0140-6736(21)00461-X 33676597 64 Bendala Estrada A.D. Calderón Parra J. Fernández Carracedo E. Inadequate use of antibiotics in the covid-19 era: effectiveness of antibiotic therapy BMC Infect Dis 21 1 2021 1 23 10.1186/s12879-021-06821-1 33390160 65 Bendala Estrada A.D. Calderón Parra J. Fernández Carracedo E. Inadequate use of antibiotics in the covid-19 era: effectiveness of antibiotic therapy BMC Infect Dis 21 1 2021 1144 10.1186/s12879-021-06821-1 34749645 66 Blanco C. Compton W.M. Volkow N.D. Opportunities for research on the treatment of substance use disorders in the context of COVID-19 JAMA Psychiatr 78 4 2021 357 358 10.1001/jamapsychiatry.2020.3177 67 Alexander A.C. Ward K.D. Understanding postdisaster substance use and psychological distress using concepts from the self-medication hypothesis and social cognitive theory J Psychoact Drugs 50 2 2018 177 186 10.1080/02791072.2017.1397304 68 Lorant V. Smith P. Van den Broeck K. Nicaise P. Psychological distress associated with the COVID-19 pandemic and suppression measures during the first wave in Belgium BMC Psychiatr 21 1 2021 1 10 10.1186/s12888-021-03109-1 69 Turner S. Mota N. Bolton J. Sareen J. Self-medication with alcohol or drugs for mood and anxiety disorders: a narrative review of the epidemiological literature Depress Anxiety 35 9 2018 851 860 10.1002/da.22771 29999576 70 Afari-Asiedu S. Kinsman J. Boamah-Kaali E. To sell or not to sell; the differences between regulatory and community demands regarding access to antibiotics in rural Ghana J Pharm Policy Pract 11 1 2018 1 10 10.1186/s40545-018-0158-6 29372061 71 Horumpende P.G. Said S.H. Mazuguni F.S. Prevalence, determinants and knowledge of antibacterial self-medication: a cross sectional study in North-eastern Tanzania PLoS One 13 10 2018 e0206623 10.1371/journal.pone.0206623 72 Khan H. Maheen S. Alamgeer Determinants of increasing trend of self-medication in a Pakistani community Trop J Pharmaceut Res 13 3 2014 437 443 10.4314/tjpr.v13i3.19 73 Shah J.J. Ahmad H. Rehan B.B. Self-medication with antibiotics among non-medical university students of Karachi: a cross-sectional study BMC Pharmacol Toxicol 15 1 2014 1 7 10.1186/2050-6511-15-74 24417770 74 Sakeena M.H.F. Bennett A.A. McLachlan A.J. Non-prescription sales of antimicrobial agents at community pharmacies in developing countries: a systematic review Int J Antimicrob Agents 52 6 2018 771 782 10.1016/j.ijantimicag.2018.09.022 30312654 75 Qiu J. Shen B. Zhao M. Wang Z. Xie B. Xu Y. A nationwide survey of psychological distress among Chinese people in the COVID-19 epidemic: implications and policy recommendations Gen Psychiatry 33 2 2020 e100213 10.1136/gpsych-2020-100213 76 Bendau A. Petzold M.B. Pyrkosch L. Associations between COVID-19 related media consumption and symptoms of anxiety, depression and COVID-19 related fear in the general population in Germany Eur Arch Psychiatr Clin Neurosci 271 2 2021 283 291 10.1007/s00406-020-01171-6 77 Carrasco-Garrido P. Hernández-Barrera V. López De Andrés A. Jiménez-Trujillo I. Jiménez-García R. Sex-Differences on self-medication in Spain Pharmacoepidemiol Drug Saf 19 12 2010 1293 1299 10.1002/pds.2034 20872823 78 McLean C.P. Anderson E.R. Brave men and timid women? A review of the gender differences in fear and anxiety Clin Psychol Rev 29 6 2009 496 505 10.1016/j.cpr.2009.05.003 19541399 79 Moghanibashi-Mansourieh A. Assessing the anxiety level of Iranian general population during COVID-19 outbreak Asian J Psychiatr 51 2020 102076 10.1016/j.ajp.2020.102076 80 Casagrande M. Favieri F. Tambelli R. Forte G. The enemy who sealed the world: effects quarantine due to the COVID-19 on sleep quality, anxiety, and psychological distress in the Italian population Sleep Med 75 2020 12 20 10.1016/j.sleep.2020.05.011 32853913 81 Buckner J.D. Heimberg R.G. Ecker A.H. Vinci C. A biopsychosocial model of social anxiety and substance use Depress Anxiety 30 3 2013 276 284 10.1002/da.22032 23239365	36496334	PMC9715464	NO-CC CODE	2022-12-07 23:16:36	no	Res Social Adm Pharm. 2022 Dec 2; doi: 10.1016/j.sapharm.2022.11.011	utf-8	Res Social Adm Pharm	2,022	10.1016/j.sapharm.2022.11.011	oa_other
==== Front Cell Cell Cell 0092-8674 1097-4172 The Authors. Published by Elsevier Inc. S0092-8674(22)01505-7 10.1016/j.cell.2022.11.030 Article SARS-CoV-2 replication in airway epithelia requires motile cilia and microvillar reprogramming Wu Chien-Ting 111 Lidsky Peter V. 211 Xiao Yinghong 211 Cheng Ran 1311 Lee Ivan T. 456 Nakayama Tsuguhisa 67 Jiang Sizun 4 He Wei 1 Demeter Janos 1 Knight Miguel G. 2 Turn Rachel E. 1 Rojas-Hernandez Laura S. 8 Ye Chengjin 9 Chiem Kevin 9 Shon Judy 10 Martinez-Sobrido Luis 9 Bertozzi Carolyn R. 10 Nolan Garry P. 4 Nayak Jayakar V. 46 Milla Carlos 8 Andino Raul 2∗ Jackson Peter K. 1412∗ 1 Baxter Laboratory, Department of Microbiology & Immunology, Stanford University School of Medicine, Stanford, CA, USA 2 Department of Microbiology and Immunology, University of California, San Francisco, San Francisco, CA, USA 3 Department of Biology, Stanford University, Stanford, CA, USA 4 Department of Pathology, Stanford University School of Medicine, Stanford, CA, USA 5 Division of Allergy, Immunology, and Rheumatology, Department of Pediatrics, Stanford University School of Medicine, Stanford, CA, USA 6 Department of Otolaryngology–Head and Neck Surgery, Stanford University School of Medicine, Stanford, CA, USA 7 Department of Otorhinolaryngology, Jikei University School of Medicine, Tokyo, Japan 8 Department of Pediatric Pulmonary Medicine, Stanford University School of Medicine, Stanford, CA, USA 9 Disease Intervention and Prevention and Population Health Programs, Texas Biomedical Research Institute, San Antonio, TX, USA 10 Department of Chemistry, Stanford University, Stanford, CA, USA ∗ Corresponding author (P.K.J.), (R.A) 11 These authors contributed equally 12 Lead Contact 2 12 2022 2 12 2022 10 4 2022 15 9 2022 23 11 2022 © 2022 The Authors. 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. How SARS-CoV-2 penetrates the airway barrier of mucus and periciliary mucins to infect nasal epithelium remains unclear. Using primary nasal epithelial organoid cultures, we found the virus attaches to motile cilia via the ACE2 receptor. SARS-CoV-2 traverses the mucus layer, using motile cilia as tracks to access the cell body. Depleting cilia blocks infection for SARS-CoV-2 and other respiratory viruses. SARS-CoV-2 progeny attach to airway microvilli 24 hours post-infection and trigger formation of apically extended and highly branched microvilli that organize viral egress from the microvilli back into the mucus layer, supporting a model of virus dispersion throughout airway tissue via mucociliary transport. Phosphoproteomics and kinase inhibition reveal microvillar remodeling is regulated by PAK kinases. Importantly, Omicron variants bind with higher affinity to motile cilia and show accelerated viral entry. Our work suggests that motile cilia, microvilli, and mucociliary-dependent mucus flow are critical for efficient virus replication in nasal epithelia. Graphical abstract Respiratory viruses including SARS-CoV-2 surpass the defensive mucus layer of the lung by entering and exiting epithelial cells via their protruding, motile cilia. ==== Body pmcPeter K. Jackson, PhD, Professor, Baxter Laboratory for Stem Cell Biology, Departments of Microbiology & Immunology and Pathology, Center for Clinical Sciences Research, 269 Campus Drive, Stanford, California 94305-5175 Tel: (650) 302-3581 Raul Andino, Ph.D., Professor, Department of Microbiology and Immunology, University of California, San Francisco, 600 16th Street, Room S572E, Box 2280 San Francisco, CA 94143-2280 Ph. 415 502 6358	0	PMC9715480	NO-CC CODE	2022-12-06 23:15:41	no	Cell. 2022 Dec 2; doi: 10.1016/j.cell.2022.11.030	utf-8	Cell	2,022	10.1016/j.cell.2022.11.030	oa_other
==== Front Educacio´n Me´dica 1575-1813 1579-2099 The Authors. Published by Elsevier España, S.L.U. S1575-1813(22)00068-7 10.1016/j.edumed.2022.100776 100776 Original E-learning in Pathology courses in the times of COVID-19 El e-learning en los cursos de patología en los tiempos de COVID-19Sahraoui Ghada 123 Sassi Farah 123⁎ Doghri Raoudha 123 Charfi Lamia 123 Driss Maha 123 Mrad Karima 123 1 Service d’Anatomie et Cytologie pathologiques, Institut Salah Azaiez, Tunis, Tunisie 2 Precision Medicine, Personalized Medicine and Oncology Investigation laboratory, LR21SP01, Tunis, Tunisia. 3 Faculty of Medicine of Tunis, University of Tunis El Manar, Tunis, Tunisia. ⁎ Corresponding author. 2 12 2022 2 12 2022 10077610 10 2022 23 11 2022 © 2022 The Authors. Published by Elsevier España, S.L.U. 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 Distance learning (DL) is a promising educational approach for teaching medical courses. Our Pathology College was faced with the difficulty of a complete online transition of the classes because of the public health measures implemented during the COVID-19 pandemic. The objective is to evaluate this teaching method of the Pathology College with reference to the learner's point of view. Methods After attending DL sessions in the Pathology College, pathology residents completed the questionnaire using Google Forms. Twenty-six out of 33 initially enrolled in the pathology course returned fully filled out, valid questionnaires. Results Twenty-four residents (92.3%) had already an E-learning experience. Almost 70% of participants were satisfied with their DL experience. Thirty percent of the participants thought that DL should replace face-to-face courses. Technical difficulties were encountered in 42% of cases with the most common one related to internet connection (66.7%). Interaction with teachers during DL courses was considered more difficult than face-to-face courses by 61.5% of participants. Participants found that learning via the virtual slide websites was better than learning in the histology workshop in 53.8% of cases. The main weaknesses of DL were the dependence on technical means (42.3%), the lack of interactivity with colleagues (26.9%) and teachers (26.9%). Conclusion Pathology lessons were successfully taught via DL, which was highly embraced by the students. Our findings shed light on a variety of areas of the students' DL experiences, and we strongly urge the faculty to take the students' opinions into account when formulating guidelines for higher-quality medical education. Introducción El aprendizaje a distancia (AD) es un enfoque educativo prometedor para la enseñanza de cursos de medicina. Nuestra Facultad de Patología se enfrentó a la dificultad de una transición completa de las clases en línea debido a las medidas de salud pública aplicadas durante la pandemia de COVID-19. El objetivo es evaluar este método de enseñanza de la Facultad de Patología con referencia al punto de vista del alumno. Métodos Después de asistir a las sesiones del AD en el Colegio de Patología, los residentes de patología completaron el cuestionario utilizando Google Forms. Veintiséis de los 33 matriculados inicialmente en el curso de patología devolvieron los cuestionarios válidos y totalmente cumplimentados. Resultados Veinticuatro residentes (92,3%) ya habían tenido una experiencia de E-learning. Casi el 70% de los participantes estaban satisfechos con su experiencia de AD. El 30% de los participantes pensaba que el AD debería sustituir a los cursos presenciales. Se encontraron dificultades técnicas en el 42% de los casos, siendo la más común la relacionada con la conexión a Internet (66,7%). La interacción con los profesores durante los cursos del AD fue considerada más difícil que los cursos presenciales por el 61,5% de los participantes. Los participantes consideraron que el aprendizaje a través de los sitios web de diapositivas virtuales era mejor que el aprendizaje en el taller de histología en el 53,8% de los casos. Los principales puntos débiles del AD fueron la dependencia de los medios técnicos (42,3%), la falta de interactividad con los colegas (26,9%) y los profesores (26,9%). Conclusión Las clases de patología se impartieron con éxito a través del AD, que fue muy aceptado por los estudiantes. Nuestros hallazgos arrojan luz sobre una variedad de áreas de las experiencias del AD de los estudiantes, e instamos encarecidamente al profesorado a tener en cuenta las opiniones de los estudiantes a la hora de formular directrices para una educación médica de mayor calidad. Palabras clave residentes de medicina evaluación de e-learning curso de patología Keywords medical residents e-learning evaluation pathology course ==== Body pmcIntroduction All facets of human existence, including healthcare facilities, business prospects, travel and transportation, and social structure, have undergone significant changes as a result of the global spread of the coronavirus illness 2019 (COVID-19) brought on by SARS-CoV 2 [1]. Because of COVID-19's effects on the education sector, schools and universities including health area have had to close in the majority of the affected countries as well as teaching hospitals which are solely handling "on-duty" shifts [2]. Distance learning (DL) is a technique that is frequently utilized as a tool to spread information and culture. The use of DL in the field of health is a topic that is frequently discussed. Many nations have used this tool, and in Tunisia, some specialty colleges have adopted it to ensure the sustainability of medical resident training. It has demonstrated its effectiveness of sharing knowledge [3]. Given the particularity of the Pathology speciality, the importance of analyzing slides through traditional histology workshop in acquiring diagnostic skills and the manipulation of light microscopes in the training of residents, the substitution of face-to-face training devices by digital means is an unprecedented experience as pathology residents had to quickly adapt to learning exclusively online. Due to the absence of double-scoping, educational innovation has been required to keep teaching microscopy. Whole slide imaging (WSI) is one alternative for digital pathology, however due to funding restrictions experienced by many departments this option is frequently unaffordable, especially in Tunisia. Alternatively, a trend toward teaching pathology through dynamic virtual microscopy provides an easily available, physically distant, and cost-effective option. A standard light microscope, a mounted digital camera, laptops, and videoconferencing software are necessary tools to share a slide image with the student (s) [4]. Thus, the question of DL impact on the pathology residency curriculum was raised. In this study, we propose to evaluate this teaching method of the Pathology College with reference to the learner's point of view. Methods 1. Study design and sample This is a descriptive cross-sectional study including pathology residents. It was based on a questionnaire created via "Google Forms" and sent to residents after attending synchronous online teaching sessions with, at the end of the course, accompanying images slide sets from surgical specimens anonymized by removing all identifiable information, including the coded specimen number assigned for internal department tracking purposes from the Pathology College, during the period between March 2020 and December 2021. Following the presentation, students were given a quiz consisting of 5 questions based on digital pathology image slides. The questions in several modules were just multiple choice of "what is the most likely diagnosis?" This session was also recorded and is available for later playback. The questionnaire consisted of multiple choice and short free response questions. The questionnaire focused on the following items:- Organization of courses - Organization of slide pathology images - Overall assessment of online teaching - Supervision and interaction with individual teachers - Supervision and interaction with the course coordinator - Learner motivation to attend online versus face-to-face instruction - Strengths of video-conferencing compared to face-to-face teaching - Difficulties and constraints encountered by students during this period of distance learning - Learners' choice of teaching mode for future college courses - Suggested ways to improve the distance learning system The coordinator of each session is required to create the event via google calendar. He/she would send an invitation to attend a "google meet" to the learners involved in the session. At the beginning of each session, the coordinator accepted participants, introduced the speakers, and noted any questions asked via chat. The coordinator also ensured the coordination between the different speakers, in particular the respect of time limit and the closing of the session with a summary of the content presented and the announcement of possible future sessions. 2. Study population Inclusion criteria We included all questionnaires completed by pathology residents assigned to pathology departments in Tunisian university hospitals, who attended the online teaching of the college courses. Non-inclusion criteria We did not send a questionnaire to residents of other specialties assigned to a pathology department for optional stage. Exclusion Criteria We excluded pathology residents who are assigned to hospitals and did not attend online courses (1st year residents in 2022). Statistical Analysis The statistical analysis was performed by Microsoft Office 2016 Excel software. It was descriptive focusing first on the study population and then on the learners' responses to the different questions in the online questionnaire. Ethical Considerations Consent was obtained from all students included in this study to conduct this work. There were no conflicts of interest. Results A total of 26 pathology residents participated in the study among the 33 invited. The characteristics of the cohort are presented in Table 1 .Table 1 Characteristics of the study participants Table 1 Number (%) Gender Male 4 (15.4%) Female 22 (84.6%) University Tunis 10 (38.5%) Monastir 7 (26.9%) Sousse 6 (23.1%) Sfax 3 (11.5%) Year of residency 4th year 16 (61.5%) 3rd year 6 (23.1%) 2nd year 4 (15.4%) Previous E-learning experience Yes 24 (92.3%) No 2 (7.7%) Technological means used to follow the E-learning Laptop 23 (88.5%) Tablet 2 (7.7%) Smartphone 1 (3.8%) Access to a personal internet connection Wifi 22 (88%) 4G 4 (12%) Location of e-learning course Home 25 (96%) Stage 1 (4%) Time spent for E-learning per month ≥ 4 hours 14 (54%) < 4 hours 12 (46%) Out of the 33 pathology residents, 22 (84.6%) were females. They were from Faculty of Medicine of Tunis in 10 (38.5%) cases and Monastir in 7 (26.9%) cases. Regarding the level of residency, 16 (61.5%) were in 4th year. Twenty-four residents (92.3%) had already an E-learning experience. The most used tool to follow DL was the laptop in 23 (88.5%) cases. All residents had access to internet connection, and it was through WIFI in 22 (88%) cases. Residents followed the courses at home in 25 (96%) cases and only one at a stage (4%). The time spent for DL was ≥ 4 hours in 14 (54%) cases. PowerPoint was the most used presentation program (81%). Almost 70% of participants were satisfied with their DL experience. Thirty percent of the participants thought that DL should replace face-to-face theoretical courses. All participants reported that they were informed of the date of the online course in time. Nearly 54% of participants said they did not have the resources available in time. Technical difficulties were encountered in 42% of cases. The most common difficulty was related to internet connection (66.7%). More than half of the residents (53.8%) felt that DL should be complementary to face-to-face teaching. Student involvement in DL was rated low to moderate in 69% of cases. Most of participants agreed, either totally (38.5%) or partially (34.6%), that the environment offered in the DL was as realistic as the one offered in the classroom/amphitheater. Over 73% of participants rated the level of technical support provided by course coordinators as at least satisfactory. More than 90% of participants were at least satisfied with the level of teacher involvement in the online training. Interaction with teachers during DL courses was considered more difficult than face-to-face by 61.5% of participants. Only 38.5% of the participants thought that the time management of DL course by the teachers was more difficult than in the classroom. Participants felt that face-to-face courses were presented in a clearer manner compared to online courses in 73% of cases. Nearly 90% of participants found the course materials used by teachers interactive. Concerning virtual pathology slides, participants found that acquiring diagnostic skills via the virtual slide websites was better than learning in the traditional double or multi-headed scope in 53.8% of cases. The main two reasons were: more time to analyze virtual slides than in the traditional histology workshop (42.9%) and the ability to view virtual slides at any time (35.7%). Only 11.5% felt that DL was not beneficial for life skills. More than 90% of the participants felt confident in making diagnoses after attending DL via the virtual slides. More than 60% of participants felt able to take exams on courses taught online in actual practice. The main strength identified by participants was the flexibility of learning locations and schedules (73.1%). The main weaknesses of DL were the dependence on technical means (42.3%), the lack of interactivity with colleagues (26.9%) and the lack of interactivity with teachers (26.9%). The main conditions for a successful DL experience identified by participants were willingness to adapt to a new technology (38.5%) and acceptability by students and teachers (34.6%). The suggestions and proposals of the participants were the creation of a media library gathering all the video presentations and virtual slides. Finally, learners' choices for how to teach future college courses was not in favor of online courses (70%). Discussion Since E-learning is the predominate method of academic activities among medical residents during the COVID19 pandemic, it is essential to evaluate their perception and satisfaction with the method to find weaknesses, enhance the program's quality, and make necessary revisions [5]. Thus, the current study sought to evaluate how pathology residents perceived and were satisfied with DL during the COVID-19 epidemic. The main results of the study showed that DL is a satisfactory method from the learner's point of view. Students believe in 70% of cases that DL is a useful tool for carrying out academic activities. The main benefits identified by the residents were accessibility to the sessions from any location and schedules and the possibility of reviewing virtual slides at any time. However, only 30% of the participants think that DL should replace face-to-face theoretical courses. The lack of face-to-face connection with the teachers and colleagues and the incidence of technological issues that occasionally hindered the smooth running of the sessions were the shortcomings highlighted by pathology residents. The Pathology College chose to carry on its training programs for residents at various levels during the period of confinement imposed by our nation's government due to the worldwide health crisis. The only viable choice in these circumstances was DL. Several authors have compared real audience teaching versus DL [[6], [7], [8]]. They have concluded that they were equivalent in terms of knowledge improvement. In our study, participants felt that face-to-face courses were presented in a clearer manner compared to online courses in 73% of cases. On contrary, according to several reports, the instructor's presentation style and visual presence play a key role in getting students interested in the content [9]. In traditional courses, pathologists typically focus their teaching efforts on tumor boards, conferences on medical autopsies, and directly under the microscope with visiting clinical teams. However, in a synchronous e-learning environment, the professor has additional responsibilities as a teacher, including helping students transition smoothly from traditional classroom settings to online classes, engaging the audience in conversation, and ensuring that everyone has a positive and active e-learning experience. To overcome some of the drawbacks of e-pathology courses, one study proposed to combine the two teaching methods into a so-called hybrid "bent learning" [5]. Recent studies proposed game-based learning, like "Kahoot!" which is a mobile game-based online digital formative assessment tool that has recently been examined by medical students [10,11]. Since gaming technologies are thought to be suitable for opportunistic learning and offer helpful learning techniques that promote continuing academic results [12], pathology department should take this new tool into account to promote e-pathology courses. In our study, residents suggested the creation of a media library gathering all the video presentations and virtual slides which can be viewed at any time. There are several obvious and evident benefits to using WSI instead of glass slides in acquiring diagnostic skills in DL pathology lessons. Scanning the original glass slide allows for the dissemination of a precisely replicated reproduction of the slide without having to resort to cutting more sections (which may or may not be representative of the original slide/pathology), depletion of a tissue block, which is important for ancillary studies that frequently affect patient treatment, obviating the necessity for packaging and posting slides, the possibility of glass slide breakage and delays/losses in transit, and the time WSI offers the students and teachers the chance to view the slide in real time while geographically apart, despite certain technical difficulties [13]. The latter capacity strengthens and even encourages distance education because glass slides do not allow both participants to see the slide at the same time. Therefore, there are many and significant benefits of adopting WSI that much outweigh any drawbacks [13]. However, the deployment of WSI is within the means of several departments in Tunisia. The routine currency in all pathology departments are glass slides and, for teaching purpose, scanning slides via a mounted digital camera and a standard light microscope. In our study, participants found that acquiring diagnostic skills via the scanned images slides was better than learning in the traditional double or multi-headed scope in 53.8% of cases as they had more time to analyze virtual slides than in the traditional histology workshop and were able to view them at any time. More than 90% of the participants felt confident in making diagnoses after attending DL via the virtual slides. In order to encourage and enable ongoing support and motivation of the learner, a continuous formative evaluation procedure is also to be carefully considered in this learning process. The main limitation of our study is the small number of participants included. Indeed, this is a preliminary experiment, and the study should be extended to other groups of students with an inclusion of a control group for more objectivity. Conclusion Despite some limitations, the experience in e-pathology courses was efficient and welcomed by residents. In order to improve our understanding of the online teaching process and its quality, we also think that more research in the area of e-learning pathology should be done in the future. Sources of funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Ethics: The present study has been conducted according to the principles of the declaration of Helsinki. Ethical approval for the study was not required. Participants were made aware of the purpose of the study, the anonymous nature of the purpose, the anonymous nature of the dataset generated and the option to not respond if they so wished. Provenance and peer review: Not commissioned, externally peer-reviewed. Credit authorship contribution statement: All the authors read and approved the final version of the manuscript. Declaration of competing interest: The authors report no declarations of interest. Guarantor: Sassi Farah ==== Refs References 1 Montemurro N. The emotional impact of COVID-19: From medical staff to common people Brain Behav Immun 87 2020 23 24 10.1016/j.bbi.2020.03.032 32240766 2 Bayham J. Fenichel E.P. Impact of school closures for COVID-19 on the US health-care workforce and net mortality: a modelling study Lancet Public Health 5 2020 e271 e278 10.1016/S2468-2667(20)30082-7 32251626 3 Machado Júnior A.J. Pauna H.F. Distance learning and telemedicine in the area of Otorhinolaryngology: lessons in times of pandemic Braz J Otorhinolaryngol 86 2020 271 272 10.1016/j.bjorl.2020.03.003 32295739 4 Christian RJ, VanSandt M. Using Dynamic Virtual Microscopy to Train Pathology Residents During the Pandemic: Perspectives on Pathology Education in the Age of COVID-19. Acad Pathol 2021;8:23742895211006820. doi: 10.1177/23742895211006819. 5 L’enseignement à distance à l’aire de la covid 19: un saut vers l’avenir? - PMC n.d. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8759320/ (accessed October 3, 2022). 6 Spickard A. Alrajeh N. Cordray D. Gigante J. Learning about screening using an online or live lecture: does it matter? J Gen Intern Med 17 2002 540 545 10.1046/j.1525-1497.2002.10731.x 12133144 7 Fordis M. King J.E. Ballantyne C.M. Jones P.H. Schneider K.H. Spann S.J. Comparison of the instructional efficacy of Internet-based CME with live interactive CME workshops: a randomized controlled trial JAMA 294 2005 1043 1051 10.1001/jama.294.9.1043 16145024 8 Harris J.M. Elliott T.E. Davis B.E. Chabal C. Fulginiti J.V. Fine P.G. Educating generalist physicians about chronic pain: live experts and online education can provide durable benefits Pain Med 9 2008 555 563 10.1111/j.1526-4637.2007.00399.x 18266811 9 (PDF) Examining Interactivity in Synchronous Virtual Classrooms n.d. https://www.researchgate.net/publication/272151781_Examining_Interactivity_in_Synchronous_Virtual_Classrooms (accessed October 3, 2022). 10 Manou E, Lazari E-C, Lazaris AC, Agrogiannis G, Kavantzas NG, Thomopoulou G-E. Evaluating e-Learning in the Pathology Course During the COVID-19 Pandemic. AMEP 2022;13:285–300. doi: 10.2147/AMEP.S353935. 11 Elkhamisy F.A.A. Wassef R.M. Innovating pathology learning via Kahoot! game-based tool: a quantitative study of students` perceptions and academic performance Alexandria Journal of Medicine 57 2021 215 223 10.1080/20905068.2021.1954413 12 Wynter L. Burgess A. Kalman E. Heron J.E. Bleasel J. Medical students: what educational resources are they using? BMC Med Educ 19 2019 36 10.1186/s12909-019-1462-9 30683084 13 Evans A.J. Depeiza N. Allen S.-G. Fraser K. Shirley S. Chetty R. Use of whole slide imaging (WSI) for distance teaching Journal of Clinical Pathology 74 2021 425 428 10.1136/jclinpath-2020-206763 32646928	0	PMC9715481	NO-CC CODE	2022-12-15 23:18:03	no	2022 Dec 2;:100776	utf-8	null	null	null	oa_other
==== Front J Fr Ophtalmol J Fr Ophtalmol Journal Francais D'Ophtalmologie 0181-5512 1773-0597 Elsevier Masson SAS. S0181-5512(22)00421-1 10.1016/j.jfo.2022.07.007 Article Perforation cornéenne sur ulcère de mooren en post vaccination anti covid -19: à propos d’un cas Corneal perforation on Mooren ulcer after anti-covid 19 vaccination: a case reportAlliti F. ⁎ Mchachi A. Benhmidoune L. Chakib A. Rachid R. El Belhadji M. Service d’ophtalmologie adultes, hôpital 20 août1953, CHU ibn Rochd, faculté de médecine et de pharmacie, université Hassan II, Casablanca, Morocco ⁎ Corresponding author 2 12 2022 2 12 2022 6 6 2022 © 2022 Elsevier Masson SAS. All rights reserved. 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	36517313	PMC9715482	NO-CC CODE	2022-12-13 23:16:28	no	J Fr Ophtalmol. 2022 Dec 2; doi: 10.1016/j.jfo.2022.07.007	utf-8	J Fr Ophtalmol	2,022	10.1016/j.jfo.2022.07.007	oa_other
==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 The Author(s). Published by Elsevier Ltd. S0264-410X(22)01492-X 10.1016/j.vaccine.2022.11.068 Article Menstrual disturbances in 12- to 15-year-old girls after one dose of COVID-19 Comirnaty vaccine: population-based cohort study in Norway Henriette Caspersen Ida a1⁎ Juvet Lene K b1 Feiring Berit b Laake Ida b Hayman Robertson Anna b Mjaaland Siri b Magnus Per a Trogstad Lill b a Centre for Fertility and Health, Norwegian Institute of Public Health, Postbox 222 Skøyen, N-0213 Oslo, Norway b Division of Infection Control, Norwegian Institute of Public Health, Postbox 222 Skøyen, N-0213 Oslo, Norway ⁎ Corresponding author. 1 Contributed equally. 2 12 2022 2 12 2022 29 6 2022 8 11 2022 28 11 2022 © 2022 The Author(s). 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. ==== Body pmc1 Introduction A worldwide COVID-19 mass vaccination campaign targeting adults was launched in late December 2020. Subsequently, the Comirnaty (BNT162b2) vaccine was recommended for children aged 12-15 years in May 2021 [1]. In Norway, only one dose of the Comirnaty vaccine was recommended to children aged 12-15 years. Vaccination was not recommended for children who had been infected with SARS-CoV-2. In line with findings in older age groups, the most prevalent adverse events after vaccination that have been reported in 12- to 15-year-old adolescents are injection site pain (in 79 to 86% of participants), fatigue (in 60 to 66%), and headache (in 55 to 65%) [2]. Adolescents aged 12-17 years have been found to have a moderately higher risk of adverse reactions than adults [3]. For new vaccines, clinical trials typically collect data on commonly recognized adverse events and safety profiles. However, questions about the menstrual cycle have not been included in clinical studies. A significant number of reports on menstrual disturbances after COVID-19 vaccination have been registered in spontaneous adverse events surveillance systems in several countries (USA, UK, Norway, the Netherlands) [4], [5], [6], [7]. Since June 2021, the Norwegian Medicines Agency has received such reports through their routine surveillance system for adverse effects of medications and vaccines [4]. Upon media attention, an increasing number of reports were received during the summer of 2021. Recently, the Netherlands Pharmacovigilance Centre Lareb reported a signal for COVID-19 vaccine-related menstrual disturbances and for postmenopausal bleeding after COVID-19 vaccination [7]. In October 2022, The European Medicines Agency (EMA) also recommended to add heavy menstrual bleeding to the product information as a possible side effect of the mRNA COVID-19 vaccines Comirnaty and Spikevax [8]. Since unverified claims of adverse effects may give rise to vaccine hesitancy or refusal, accurate scientific investigation of these phenomena is imperative for public health, also in young teenage girls. In a recent study, we showed that women aged 18-30 years experienced menstrual cycle changes after COVID-19 vaccination [9], but importantly, such changes were also common prior to vaccination. The aim of the current study was to estimate the association between COVID-19 vaccination and menstrual disturbances in girls aged 12-15 years using maternal questionnaire responses in a large population-based cohort. 2 Methods 2.1 Study population The Norwegian Mother, Father and Child Cohort Study (MoBa) is an ongoing population-based pregnancy cohort study established by the Norwegian Institute of Public Health. Participants were recruited during pregnancy from all over Norway during 1999-2008 [10]. The participation rate was 41% among all invited women. The cohort now includes approximately 114 000 children and their parents, who are followed up with regular questionnaires and registry linkages. MoBa was established with the overall aim to understand causes of diseases. Since March 2020, all active adult cohort participants (about 149 000) have also been invited to answer short electronic questionnaires every 14 days regarding symptoms related to COVID-19 disease and vaccination, life situation during the pandemic, and more. The response rate over the first 50 survey rounds during the pandemic was 46-80%, with an average (mean) response rate of 66%. Participation has been stable throughout the pandemic. For instance, the average participation rate was 71% for the first ten rounds (March to August, 2020), and 68% for rounds 31-40 (May to September, 2021). In the 41st questionnaire round in autumn 2021, a subsample of participating MoBa mothers with one child aged 12-15 years (n=29 959) were invited to answer an electronic questionnaire on behalf of their child. The questionnaire included questions about (in chronological order) COVID-19 symptoms, testing, and adverse events after vaccination. The questionnaire was distributed via text message to the mothers’ mobile phones on October 13, 2021. The response rate was 64%. Using each participant’s (mother’s and child’s) unique national identification number, the questionnaire data and information from MoBa were linked to the Norwegian Immunization Registry (SYSVAK) and the National Surveillance System for Communicable Diseases (MSIS) [11]. It is mandatory for health care workers providing the vaccine to report vaccinations against COVID-19 electronically to SYSVAK, and for laboratories to report polymerase chain reaction (PCR)-confirmed SARS-CoV-2 infection to MSIS. During the fall of 2021, when children went back to school after the summer holidays in Norway, extensive testing and screening against SARS-CoV-2 was performed, due to a surge in circulation of the Delta variant. From the group of 19 310 questionnaire respondents, we excluded 9947 boys and 1646 non-menstruating girls (Figure 1 ). We also excluded 152 individuals who had received two doses of COVID-19 vaccine or were vaccinated before August 1, 2021, under the assumption that they likely belonged to a risk group due to underlying disease. Thus, the study sample consisted of 7565 girls who had started to menstruate (Figure 1).Figure 1 Inclusion of individuals in the study. 2.2 Exposure Date of vaccination for each girl was obtained from the records in the national vaccine registry SYSVAK. We categorized subjects into two main groups: Vaccinated subjects included girls who were vaccinated between August 1st and the date of questionnaire completion, while unvaccinated subjects were girls who were not vaccinated upon completion of the questionnaire. 2.3 Outcomes Questions about menstrual disturbances were posed to all mothers of girls who had started to menstruate (Supplementary Table 1). For vaccinated girls, the mothers were asked whether their daughters had experienced any of the following disturbances in their last menstruation before the first vaccine dose: 1) heavier bleeding than usual, 2) prolonged menstruation, 3) shorter interval between menstruations than usual, 4) longer interval between menstruations than usual, 5) spot bleedings between menstruations, 6) stronger pain during menstruation, 7) period pain without bleeding, and 8) any other symptoms from the pelvic region. Subsequently, they were asked the same list of questions for their first menstrual cycle after the first vaccine dose (Supplementary Table 1). Mothers of girls who had not yet had a menstrual cycle after vaccination were recommended to answer “don’t know” for any of the post-vaccination disturbances. Mothers of unvaccinated girls were asked to answer the same list of questions for their daughter’s last menstruation. 2.4 Other variables Other variables included PCR-confirmed SARS-CoV-2 infection (MSIS), the mother’s vaccination status (obtained from SYSVAK) and her education level, which was self-reported in August 2021. For those with missing information on education from 2021 (about 20%), we used the most recent available self-reported record of education level from the existing MoBa database. The girl’s birth year was obtained from The Medical Birth Registry of Norway. From a questionnaire subsequently distributed to mothers in April 2022, we obtained information about use of a period tracker app (or similar). The question was asked as follows: “Is she using an app/calendar/diary/other method to track her menstruation?” (No/Yes/Not sure/Not relevant) and if Yes, “For how long has she been using this?” (<1 year, 1-2 years, >2 years, Not sure). Those answering “Yes” and duration of use “1-2 years” or “>2 years” were defined as menstrual app users to ensure that the girls had been using the app for several months prior to vaccination. 2.5 Statistical Analyses We used a self-controlled case series (SCCS) analysis to estimate associations between COVID-19 vaccination and menstrual disturbances [12]. In this design, only vaccinated cases with the outcome in question (i.e., a menstrual disturbance event before and/or after vaccination) were included. Cases who reported “don’t know” or with no answer (missing) before and/or after vaccination were excluded. The cases were their own control in the sense that we compared the girl’s risk of the outcome within a specified exposure window against the risk in a non-exposed window. We used the first cycle after vaccination as the window of exposure and the last cycle prior to vaccination as the non-exposed window. We chose the SCCS analysis and not a comparison between vaccinated and unvaccinated subjects because we considered the SCCS design less prone to selection and information bias in this setting. Log-binomial regression was used to estimate risk ratios (RRs) and 95% confidence intervals (CIs). The model was fitted with generalized estimating equations to account for the within-individual dependencies. We also performed the SCCS analysis stratified by age (12-13 and 14-15 years) and on a subsample (n=1006) who had been using an app, calendar, diary or other method to track their menstrual cycle prospectively. In a sensitivity analysis, we excluded n=22 subjects with a previous SARS-CoV-2 infection from the SCCS analysis. The median time between vaccination and completion of the questionnaire was 27 days (interquartile range 22-30, 95th percentile 37 days). Between 19.7% and 22.3% of mothers of vaccinated girls answered “don’t know” or did not answer questions about irregularities for their first period after vaccination. Absolute and relative risks were therefore estimated after excluding these subjects, based on the assumption that a large proportion of these girls had not yet experienced a menstrual period after vaccination. Statistical analyses were done in Stata/SE 16.0 (Stat Corp) and R [13], version 4.1.0 [14]. 2.6 Ethics The establishment of MoBa and initial data collection was based on a license from the Norwegian Data Protection Agency and approval from The Regional Committees for Medical and Health Research Ethics. The MoBa cohort is now based on regulations related to the Norwegian Health Registry Act. The current sub-study was approved by The Regional Committee for Medical and Health Research Ethics, South East Norway C, no. 127708. 3 Results About four out of five (6196 out of 7565, 81.9%) menstruating girls had received one dose of a COVID-19 vaccine at the time of questionnaire completion (Table 1). Nearly all vaccinated girls (99.9%) were registered with the Comirnaty vaccine, while only nine girls were registered with Vaxzevria or Spikevax as the first dose. Almost all mothers of vaccinated girls were themselves vaccinated (99.6%). Among mothers of unvaccinated girls, about 90% were vaccinated (Table 1 ). Only 1.5% of the vaccinated girls had previously had a laboratory confirmed SARS-CoV-2 infection, compared to 28.0% of the unvaccinated girls. In mother-daughter pairs where both were unvaccinated (n=144 pairs), the mother tended to be younger and have a lower level of education than in pairs where both were vaccinated (n=6173 pairs), Supplementary Table 2.Table 1 Background characteristics of the study population, including 7 565 girls aged 12-15 years. Vaccinated with 1 dose of COVID-19 vaccinea(n=6196) Unvaccinated(n=1369) n % n % Birth year 2006 2393 38.6 451 32.9 2007 2198 35.5 484 35.4 2008 1368 22.1 354 25.9 2009 223 3.6 79 5.8 Missing 14 0.2 1 0.1 Maternal level of education < High school 251 4.1 79 5.8 High school 1349 21.8 367 26.8 College ≤4 years 2700 43.6 569 41.6 >4 years college 1758 28.4 320 23.4 Missing 138 2.2 34 2.5 Maternal COVID-19 vaccination status Vaccinated 6173 99.6 1225 89.5 Unvaccinated 23 0.4 144 10.5 History of laboratory confirmed SARS-CoV-2 infection Yes 94 1.5 384 28.0 No 6102 98.5 985 72.0 a Nearly all (99.9%) were vaccinated with the mRNA Comirnaty (BNT162b2) vaccine, which was recommended for children with no history of SARS-CoV-2 infection. Table 2 Number of menstrual disturbance events reported for menstruating girls aged 12-15 years, according to COVID-19 vaccination (single dose) and SARS-CoV-2 infection status. Vaccinateda(n=6196) Unvaccinated, no SARS-CoV-2 infection(n=985) Unvaccinated, history of SARS-CoV-2 infection(n=384) Event registered for last cycle before vaccine Event registered for first cycle after vaccineb Event registered for last cycle Event registered for last cycle n % c n % c n % c n % c Any menstrual disturbance eventd Yes 1054 22.6 1091 25.1 165 22.3 95 33.0 No 3619 3251 576 193 Don't know 1385 1733 217 90 Missinge 138 121 27 6 Heavier bleeding Yes 235 4.7 333 7.3 22 2.8 15 5.1 No 4757 4198 773 280 Don't know 1147 1635 178 86 Missing 57 30 12 3 Prolonged bleeding Yes 199 3.9 245 5.4 16 2.0 12 4.0 No 4844 4302 789 291 Don't know 1086 1617 165 78 Missing 67 32 15 3 Shorter interval Yes 269 5.5 285 6.3 39 4.9 21 7.0 No 4658 4227 755 278 Don't know 1194 1643 176 81 Missing 75 41 15 4 Longer interval Yes 376 7.7 339 7.5 65 8.2 51 16.7 No 4516 4160 725 255 Don't know 1238 1650 178 75 Missing 66 47 17 3 Spot bleeding Yes 153 3.1 142 3.1 24 3.0 9 3.0 No 4836 4470 785 289 Don't know 1139 1534 162 82 Missing 68 50 14 4 Stronger period pains Yes 333 6.4 326 6.9 59 6.9 23 7.3 No 4879 4381 795 294 Don't know 918 1449 118 64 Missing 66 40 13 3 Period pains without bleeding Yes 292 5.8 244 5.2 42 5.1 25 8.3 No 4741 4411 782 277 Don't know 1096 1500 148 79 Missing 67 41 13 3 Other symptoms from the pelvic region Yes 42 0.8 38 0.8 8 1.0 4 1.3 No 4998 4645 804 294 Don't know 1096 1470 160 82 Missing 60 43 13 4 a Vaccinated subjects had received one dose of COVID-19 vaccine. b Median time between vaccination and completion of the questionnaire was 27 days (interquartile range 22-30). c Proportions were calculated based on valid answers (“yes” and “no”), excluding “don’t know” and missing answers. d “Yes” includes subjects with at least one event reported. “No” includes subjects answering “no” for all events. e “Missing” includes subjects with at least one missing answer across all events. Table 3 Risk of menstrual disturbances during the first cycle after vaccination compared to the last cycle prior to vaccination among n=1468 vaccinated girls aged 12–15 years and in a subsample who reported use of a menstruation app (or similar) at least 6 months prior to vaccinationa. Relative risks (RR) were estimated using a self-controlled case series analysis, for girls who experienced any menstrual changes prior to and/or after vaccination. Age group (years) No. of subjects No. of events RR (95% CI) Prior to vaccination After vaccination Heavier bleeding 12-15 b 369 197 316 1.60 (1.43 to 1.80) 12-13 c 87 46 76 1.65 (1.30 to 2.10) 14-15 d 282 151 240 1.59 (1.39 to 1.82) 12-15, app users 74 43 68 1.58 (1.27 to 1.97) Prolonged bleeding 12-15 279 163 227 1.39 (1.22 to 1.59) 12-13 73 46 61 1.33 (1.05 to 1.67) 14-15 205 116 166 1.43 (1.22 to 1.68) 12-15, app users 51 32 45 1.41 (1.09 to 1.82) Shorter interval 12-15 338 228 272 1.19 (1.07 to 1.32) 12-13 87 57 70 1.23 (0.99 to 1.52) 14-15 250 170 201 1.18 (1.05 to 1.33) 12-15, app users 74 49 59 1.20 (0.96 to 1.51) Longer interval 12-15 423 284 328 1.15 (1.05 to 1.27) 12-13 120 79 88 1.11 (0.91 to 1.36) 14-15 302 204 239 1.17 (1.05 to 1.31) 12-15, app users 81 54 65 1.20 (0.97 to 1.50) Spot bleeding 12-15 176 125 133 1.06 (0.92 to 1.23) 12-13 41 27 31 1.15 (0.82 to 1.60) 14-15 133 96 101 1.05 (0.89 to 1.24) 12-15, app users 38 24 29 1.21 (0.85 to 1.72) Stronger period pains 12-15 388 271 310 1.14 (1.04 to 1.26) 12-13 71 51 60 1.18 (0.97 to 1.43) 14-15 316 219 249 1.14 (1.02 to 1.27) 12-15, app users 67 48 55 1.15 (0.93 to 1.42) Period pains without bleeding 12-15 315 240 240 1.00 (0.90 to 1.11) 12-13 80 56 61 1.09 (0.87 to 1.36) 14-15 235 184 179 0.97 (0.87 to 1.09) 12-15, app users 59 42 46 1.10 (0.86 to 1.40) Other symptoms from the pelvic region 12-15 49 38 37 0.97 (0.76 to 1.25) 12-13 18 12 14 1.17 (0.72 to 1.88) 14-15 31 26 23 0.88 (0.66 to 1.18) 12-15, app users 4 3 3 1.00 (0.40 to 2.52) a Among n=7565 menstruating subjects in the study sample, n=4455 (59%) had answered a questionnaire in April 2022 (i.e., approximately 6 months after collection of menstrual data), which included questions about menstruation registration. The question asked was “Is she using an app/calendar/diary/other method to log her menstruation?” and if Yes, “For how long has she been using this?”. Subjects who reported use of a menstruation registration method for at least 1 year were included in this subsample, referred to as “app users”. b Birth year 2006–2009 c Birth year 2008–2009 d Birth year 2006–2007 Notably, menstrual irregularities were relatively common in this sample of 12-15-year-old girls, independent of vaccination and infection status (Table 2). The proportion of vaccinated subjects who reported one or more menstrual irregularities in their last period prior to vaccination was 22.6%. In comparison, 25.1% of this group reported at least one event for the first cycle after vaccination. Of these, one single menstrual irregularity (for instance, unusually heavy bleeding) was reported by 14.2%, two different irregularities were reported by 6.1%, and three irregularities by 2.8%. Only 2.0% reported four irregularities or more in the first cycle after vaccination (data not shown). In a subsample using an app or similar method to track their menstruation, the proportions reporting irregularities were slightly higher, both before and after vaccination (Supplementary Table 3). Unvaccinated girls with a history of SARS-CoV-2 infection reported more menstrual disturbances compared to unvaccinated girls with no reported infection (Table 2). Vaccinated girls had more than two-fold increased risk of reporting heavy menstrual bleeding and prolonged bleeding after vaccination compared to the last cycle for unvaccinated subjects (Table 2). In the SCCS analyses, we found that the risk of heavier menstrual bleeding and prolonged bleeding was higher in the menstrual cycle after vaccination than in the cycle before vaccination, RRs 1.61 (95% CI 1.43 to 1.81) and 1.40 (95% CI 1.23 to 1.60), respectively (Table 3). Vaccination was also associated with increased risk of shorter interval, longer interval, and stronger period pains (RRs 1.14 to 1.19). The effect sizes were of similar magnitude among girls aged 12-13 years and girls aged 14-15 years, and among those who were prospectively (with regard to vaccination) tracking their menstruation using an app or other method. There was no association between vaccination and spot bleeding, period pains without bleeding, or other symptoms from the pelvic region (Table 3). The risks were similar after excluding n=22 subjects with a history of SARS-CoV-2 from the analysis sample (Supplementary Table 4). 4 Discussion In this study of adolescent girls, we found increased risks of menstrual disturbances in the first cycle after receiving one dose of COVID-19 vaccine. The risks of heavier or more prolonged bleeding than usual were significantly higher in the first menstrual cycle after vaccination than in the last cycle prior to vaccination. We also found increased risks of shorter interval, longer interval and more pain during periods following vaccination. Notably, menstrual irregularities were common also prior to vaccination and among unvaccinated subjects. For most girls, no changes were reported following COVID-19 vaccination. Our study design has several strengths. We use a large population-based cohort, with recruitment many years prior to the exposures. The representativeness of MoBa has previously been studied, indicating a somewhat higher socio-economic status among participants than in the general Norwegian population [15], [16]. In the SCCS analyses, differences in time invariant factors, such as genetics, socio-economic status, or underlying diseases, are cancelled out, which reduces bias [12]. This design should therefore provide more valid risk estimates than a design comparing different subgroups (i.e., vaccinated vs. unvaccinated subjects). Another strength of the study is that information on both vaccination and SARS-CoV-2 infection status was obtained from nationwide registries with mandatory reporting. Although self-reported data collection could be subjected to bias, using digital surveys is a safe way to perform research during the pandemic and valuable in a situation where many are vaccinated in a short period. Moreover, responding mothers have participated in MoBa for more than a decade and are accustomed to answering questionnaires, also on behalf of their child. The questionnaire used in the current study was not solely focused on vaccination and menstruation but covered other aspects of the pandemic, which may have reduced selection bias for menstruation-related questions. This study also has some limitations that need to be addressed. First, mothers were reporting on behalf of their daughters, which may have introduced misclassification, likely underreporting, of the outcomes. Still, the questions covered a range of symptoms potentially related to both infection and vaccination (such as respiratory symptoms, headache, dizziness, unwellness, skin rash, and more), which warranted close communication between mother and daughter as a basis for answering the questionnaire. Second, the occurrence of menstrual disturbances both before and after vaccination was reported by mothers at the same timepoint. Therefore, we cannot rule out that the mother’s report of irregularities before vaccination is biased by the outcome after vaccination. Still, our findings seem relatively robust to recall bias, as similar relative risks were estimated for a subsample who used a menstruation app or similar method to track their menstruation. Third, the media’s emphasis on a potential link between menstrual cycle irregularities and COVID-19 vaccination may have increased the awareness among vaccinated participants, thus increasing the likelihood of them reporting such events after vaccination [17]. Indeed, in this study, some menstrual disturbances were reported more often for vaccinated girls compared to unvaccinated girls even before they were vaccinated. This may indicate that mothers of vaccinated girls, or the vaccinated girls themselves, are more aware about menstrual irregularities and the potential adverse events after COVID-19 vaccination. The increased awareness may potentially have introduced some information bias to our analysis but probably to a limited extent for the estimated relative risks since our subjects act as their own controls. Also, the lack of association between vaccination and spot bleeding, period pains without bleeding, or other symptoms from the pelvic region, may support that bias does not solely explain the associations seen for the other outcomes. Lastly, the interval between vaccination and the date of responding to the questionnaire may have been too short to allow detection of the outcomes in all girls, especially the outcome “longer interval”, and some girls probably did not have time to experience a menstruation after vaccination. We cannot rule out that this outcome may be prone to information bias due to this limitation. Also, the exclusion of subjects answering “don’t know” in the SCCS analysis may potentially introduce bias if the occurrence of menstrual disturbances is different in this group before and after vaccination. The reports concerning irregularities in the menstrual cycle after receiving mRNA COVID-19 vaccines among women have been widely discussed [18], [19]. Menstrual disturbances are more common after COVID-19 vaccines than after non-COVID-19 vaccines in VAERS reports [19]. A study from the US reported that COVID-19 vaccination was associated with a small change in cycle length but not menses length [20]. In an online questionnaire with participants from Saudi-Arabia, one per cent of female participants (broad age groups) who had received the Comirnaty vaccine reported abnormal menstrual cycle as a post-vaccinal short term adverse event, in an open field (i.e., the outcome was not listed as a response alternative) [21]. In data collected in the UK prior to the widespread media attention to menstrual disturbances following COVID-19 vaccination, one study found that among menstruating, pre-menopausal, vaccinated individuals, 20% reported changes to their menstrual cycles up to 4 months after receiving their first vaccine dose [22]. A prospectively recruited cohort with 79 individuals showed that COVID-19 vaccine is associated with a delay to the subsequent period, and interestingly detected no association between menstrual changes and other commonly-reported side effects [23]. In a large survey from the US (median age 33 years), increased bleeding appeared to be the most common post-vaccination adverse events [24]. Moreover, a recent large study from our group observed an increased risk of menstrual disturbances after vaccination in women aged 18-30 years, both after the first and after the second vaccine dose [9]. However, another study from the UK using a retrospective recruitment did not find any association between COVID-19 vaccination and menstrual disturbances [25]. The results in our current study therefore lie within the (wide) range of estimates reported in the existing, though limited, data on adults. Given the lack of data for adolescents in particular, our findings need to be confirmed in other studies. There have also been indications that infection with SARS-CoV-2 may cause changes in the menstruation cycle [18], [26], [27]. It has been proposed that since ACE2 receptors are present on ovarian and endometrial tissue, SARS-CoV-2 infection may exert a direct impact on the female reproductive system [26], [27]. In the current study, we observed a higher occurrence of menstrual disturbances among unvaccinated girls with a previous SARS-CoV-2 infection compared to uninfected girls in the unvaccinated group. However, our study was not designed to evaluate menstrual disturbances after SARS-CoV-2 infection and this observation needs to be confirmed in other studies. Also, whether menstrual disturbances are equally common after SARS-CoV-2 infection and vaccination should be elucidated. COVID-19 vaccination and infection may potentially influence the menstruation cycle through a similar mechanism involving ACE2 receptors. The vaccine activates the immune system, possibly attacking immune cells and inflammatory molecules in the uterus [28]. Thus, some studies suggest that vaccination is less likely to affect menstruation via ovarian hormone pathways, and more likely along the inflammatory pathways [20], [24], [29]. Still, the pathophysiological mechanisms are yet unknown. Assessment of the safety aspects are important when deciding on whether to vaccinate children or not since children seldom get seriously ill with SARS-CoV-2-infection [2]. After the rapid development and emergency authorization of SARS-CoV-2 vaccines, serious adverse reactions have been reported more frequently for the SARS-CoV-2 vaccines in comparison to other vaccines, such as influenza vaccines [6]. Careful evaluation of the short- and long-term effects of both the infection itself, as well as the vaccine used for prevention, should always be performed when evaluating new vaccines. This is highly actualized during the COVID-19 pandemic where long-term consequences of SARS-CoV-2 infection are becoming evident and mass vaccination campaigns with vaccines based on new technologies have been rolled out quickly and in the similar time frame. Although the COVID-19 vaccine has become available to children, not all countries have recommended a second dose or have delayed the recommendation for safety issues. Understanding the risk of adverse reactions associated with the vaccine will assist parents in making an informed decision on whether their child should get vaccinated or not, or how many doses. The safety risk must be weighed against the risk of disease from the virus and may change depending on the circulating viral variant. Despite many reports of altered menstrual bleeding patterns after vaccination, it is still not possible to exclude the possibility that these might reflect normal variation amongst the millions of individuals that have had SARS-CoV-2 infections and received COVID-19 vaccine. To our knowledge, this is the first study to estimate the risk of menstrual disturbances after COVID-19 vaccination in adolescents. More data collected in real-time is needed to confirm our findings and to assess the risk of recurrence after a second dose in this age group. Also, studies exploring potential mechanisms are warranted. Studies that assess the direct effect of vaccination on the menstrual cycle are few and far between. Therefore, a continuous monitoring of COVID-19 vaccines is essential to enhance the reassurance and acceptance of COVID-19 vaccinations in all age groups. 5 Conclusions In this population-based study conducted in Norway, vaccination against COVID-19 for adolescent girls was associated with increased risk of experiencing menstrual disturbances in the first cycle after vaccination. However, most young girls reported no changes to their menstrual cycle following COVID-19 vaccination. Nearly all adolescent girls in Norway received one dose of the Comirnaty vaccine, which should be considered in the interpretation of our findings. Contributors IHC and LKJ are co-first authors and contributed equally to this study. LT and PM designed the study and collected the data. All authors contributed to the study conceptualization and methods. IHC, LKJ, BF, and IL had full access to all the data and performed statistical analyses. All authors interpreted the data. IHC, LKJ and LT prepared the first draft of the manuscript. All authors revised the manuscript, approved the final version. Co-first authors IHC and LKJ had final responsibility for the decision to submit for publication. Funding This work was supported by The Norwegian Research Council’s Centres of Excellence Funding Scheme (no. 262700) and the Norwegian Institute of Public Health (NIPH). Data statement Data from the cohort is available for analysis after approval from a Norwegian ethics committee and application to the Norwegian Institute of Public Health. 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. Data availability Data from the cohort is available for analysis after approval from a Norwegian ethics committee and application to the Norwegian Institute of Public Health. Acknowledgements The Norwegian Mother, Father and Child Cohort Study is supported by the Norwegian Ministry of Health and Care Services and the Ministry of Education and Research. We are grateful to all the participating families in Norway who take part in this on-going cohort study. We also thank all NIPH staff involved in collection and preparation of data and follow up of cohort participants. ==== Refs References 1 First COVID-19 vaccine approved for children aged 12 to 15 in EU. https://www.ema.europa.eu/en/news/first-covid-19-vaccine-approved-children-aged-12-15-eu. 2021. 2 Frenck R.W. Klein N.P. Kitchin N. Gurtman A. Absalon J. Lockhart S. Safety, Immunogenicity, and Efficacy of the BNT162b2 Covid-19 Vaccine in Adolescents New England Journal of Medicine. 385 2021 239 250 34043894 3 Chan E.W.W. Leung M.T.Y. Lau L.K.W. Leung J. Lum D. Wong R.S. Comparing self-reported reactogenicity between adolescents and adults following the use of BNT162b2 (Pfizer-BioNTech) messenger RNA Covid-19 vaccine: a prospective cohort study Int J Infect Dis 2021 4 Reported suspected adverse reactions of covid-19 vaccines. 2021. 5 Medicine and Healthcare Products Regulatory Agency. Coronavirus vaccine—weekly summary of Yellow Card reporting. 13 Jan 2022. https://www.gov.uk/government/publications/coronaviruscovid-19-vaccine-adverse-reactions/coronavirus-vaccine-summary-of-yellow-card-reporting#annex-1-vaccine-analysis-print. 2022. 6 Montano D. Frequency and Associations of Adverse Reactions of COVID-19 Vaccines Reported to Pharmacovigilance Systems in the European Union and the United States. Front Public Health 2021 756633 7 Reactions Weekly. 1887 2022 5 - 8 EMA. Meeting highlights from the Pharmacovigilance Risk Assessment Committee (PRAC) 24 - 27 October 2022. 9 Trogstad L. Increased Occurrence of Menstrual Disturbances in 18- to 30-Year-Old Women after COVID-19 Vaccination (January 1, 2022). Available at SSRN: https://ssrn.com/abstract=3998180 or http://dx.doi.org/10.2139/ssrn.3998180. 2022. 10 Magnus P. Birke C. Vejrup K. Haugan A. Alsaker E. Daltveit A.K. Cohort Profile Update: The Norwegian Mother and Child Cohort Study (MoBa) International Journal of Epidemiology. 45 2016 382 388 27063603 11 Trogstad L. Ung G. Hagerup-Jenssen M. Cappelen I. Haugen I.L. Feiring B. The Norwegian immunisation register–SYSVAK Euro Surveill. 17 2012 12 Whitaker H.J. Farrington C.P. Spiessens B. Musonda P. Tutorial in biostatistics: the self-controlled case series method Stat Med. 25 2006 1768 1797 16220518 13 R core team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. 2021. 14 Wickham H, Averick M, Bryan J, Chang W, McGowan LDA. Welcome to the tidyverse. Journal of Open Source Software, 4(43), 1686, https://doi.org/10.21105/joss.01686. 2019. 15 Nilsen R.M. Vollset S.E. Gjessing H.K. Skjaerven R. Melve K.K. Schreuder P. Self-selection and bias in a large prospective pregnancy cohort in Norway Paediatr Perinat Epidemiol. 23 2009 597 608 19840297 16 Biele G. Gustavson K. Czajkowski N.O. Nilsen R.M. Reichborn-Kjennerud T. Magnus P.M. Bias from self selection and loss to follow-up in prospective cohort studies Eur J Epidemiol. 34 2019 927 938 31451995 17 Katz A, Tepper Y, Eran A, Birk O. Web and social media searches highlight menstrual irregularities as a global concern in COVID-19 vaccinations. medRxiv. 2022:2022.01.30.22270125. 18 Male V. Menstrual changes after covid-19 vaccination Bmj. 374 2021 n2211 19 Zhang B. Yu X. Liu J. Liu P. COVID-19 vaccine and Menstrual conditions in female: data analysis of the Vaccine Adverse Event Reporting System Research Square 2022 10.21203/rs.3.rs-1388159/v1 20 Edelman A, Boniface ER, Benhar E, Han L, Matteson KA, Favaro C, et al. Association Between Menstrual Cycle Length and Coronavirus Disease 2019 (COVID-19) Vaccination: A U.S. Cohort. Obstetrics & Gynecology. 2022:10.1097/AOG.0000000000004695. 21 Alghamdi A.N. Alotaibi M.I. Alqahtani A.S. Al Aboud D. Abdel-Moneim A.S. BNT162b2 and ChAdOx1 SARS-CoV-2 Post-vaccination Side-Effects Among Saudi Vaccinees Front Med (Lausanne). 8 2021 760047 22 Alvergne A. Kountourides G. Argentieri A. Agyen L. Rogers N. Knight D. COVID-19 vaccination and menstrual cycle changes: A United Kingdom (UK) retrospective case-control study 2021 medRxiv 23 Von Woon E. Male V. Effect of COVID-19 vaccination on menstrual periods in a prospectively recruited cohort 2022 medRxiv 24 Lee KM, Junkins EJ, Fatima UA, Cox ML, Clancy KB. Characterizing menstrual bleeding changes occurring after SARS-CoV-2 vaccination. medRxiv. 2021:2021.10.11.21264863. 25 Male V. Effect of COVID-19 vaccination on menstrual periods in a retrospectively recruited cohort. medRxiv. 2021:2021.11.15.21266317. 26 Phelan N. Behan L.A. Owens L. The Impact of the COVID-19 Pandemic on Women's Reproductive Health Front Endocrinol (Lausanne). 12 2021 642755 27 Sharp G.C. Fraser A. Sawyer G. Kountourides G. Easey K.E. Ford G. The COVID-19 pandemic and the menstrual cycle: research gaps and opportunities Int J Epidemiol 2021 28 Sağsöz N. Does COVID-19 Infection Affect Female Reproductive System? International Journal of Women’s Health and Reproduction Sciences. 9 2021 158 159 29 Ricke D. Etiology Model for Elevated Histamine Levels Driving High Reactogenicity Vaccines (including COVID-19) Associated Menstrual Adverse Events Research Square 2022	36517325	PMC9715483	NO-CC CODE	2022-12-12 23:20:58	no	Vaccine. 2022 Dec 2; doi: 10.1016/j.vaccine.2022.11.068	utf-8	Vaccine	2,022	10.1016/j.vaccine.2022.11.068	oa_other
==== Front J Dent Sci J Dent Sci Journal of Dental Sciences 1991-7902 2213-8862 Association for Dental Sciences of the Republic of China. Publishing services by Elsevier B.V. S1991-7902(22)00311-7 10.1016/j.jds.2022.11.027 Letter to the Editor Eliminating Candida albicans for endodontic treatment purposes during the SARS-CoV-2 pandemic Nasiri Kaveh Independent Researcher a∗ Wrbas Karl-Thomas b a Essen, Germany b Department of Operative Dentistry and Periodontology, Center for Dental Medicine, Oral and Maxillofacial Surgery, Medical Center, University of Freiburg, Freiburg i.Br., Germany; Division of Endodontics, Department of Dentistry, Faculty of Medicine and Dentistry, Danube Private University, Krems, Austria ∗ Corresponding author. Koenigraetzstrasse, Essen 45138, Germany. 2 12 2022 2 12 2022 26 11 2022 28 11 2022 © 2022 Association for Dental Sciences of the Republic of China. Publishing services by Elsevier B.V. 2022 Association for Dental Sciences of the Republic of China 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 pmcDue to the specific structure of the spike proteins of the severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2), the virus has a high potential for mutation. Also, it is a challenge for current vaccines to provide satisfactory protection against the virus. Therefore, preventing the virus from spreading during a pandemic is a priority. Dental procedures, as a risk factor, can generate dental aerosol and spread the virus. As a result, all treatments that generate dental aerosol should be reduced or avoided unless they are absolutely necessary. Furthermore, eradicating microorganisms from the root canal can be considered a priority for a successful treatment. Candida albicans (C. albicans) is the most common fungus that causes endodontic treatment failure.1 , 2 This brief letter focuses on the topic of C. albicans eradication for endodontic treatment purposes. C. albicans is a diploid, Gram-positive fungus attaching to teeth dentin and colonizing in dentinal walls of root canals. As a result, it penetrates into dentinal tubules and is implicated in cases of persistent or refractory root canal infections. The fungus could remain in dental tubules despite intracanal medicament (i.e., calcium hydroxide) or conventional irrigation solutions. Long-term successful root canal treatment is attributed in eliminating microorganisms from infected root canals. Hence, it is crucial to use appropriate disinfecting methods, including irrigations and materials for root canal therapy to eliminate C. albicans.2 , 3 In the clinical study, Horlenko et al. investigated the anti-microbiological efficacy of two techniques for eliminating microorganisms from infected root canals in 64 patients with chronic apical periodontitis. The infected root canals of 34 patients (36 teeth) were used as the main group and treated using the ultrasonic method and multicomponent antimicrobial gels. The gels included metronidazole benzoate, chlorhexidine diacetate 2%, and hydrocortisone acetate. Only 2% chlorhexidine gel was applied in the canals of 30 patients (35 teeth), i.e., the control group. In the main group, the results showed that C. albicans decreased from 23.5% to 5.6% before and after treatment, respectively. In comparison, the fungus population decreased from 25.3% to 16.1% in the control group before and after the treatment, respectively. Also, the success rate of the treatment was 86% and 63% in the main and control groups, respectively. Thus, the combination of the ultrasonic method (for penetrating gels into dentinal tubules) and multicomponent antimicrobial gels is recommended for eradicating microorganisms from infected root canals.4 In another study, Mustafa et al. assessed the efficacy of antimicrobial photodynamic therapy, mechanical instrumentation, and combining both methods to remove microorganisms in 60 teeth with C-shaped canals. The results demonstrated that combining antimicrobial photodynamic therapy using a 600 nm diode laser with mechanical instrumentation is more effective in reducing microorganisms (i.e., C. albicans and bacterium), particularly in root canal curvature.5 In addition, Yasini et al. investigated the efficacy of nano-curcumin as a sonodynamic antimicrobial chemotherapy in reducing the biofilm of C. albicans and another microorganism from infected root canals of 80 extracted teeth. The study reported that combining ultrasonic waves and nano-curcumin is more effective than other methods. Thus, applying the novel combination method (curcumin and ultrasonic) is recommended to eliminate C. albicans from infected root canals.6 To eliminate C. albicans from infected root canals, Reddy et al. compared the antifungal effectiveness of Octenisept 100% with other solutions (i.e., 17% EDTA + 5.25% NaOCl, 17% EDTA + 5.25% NaOCl + 1% clotrimazole, and phosphate buffer saline) based on 80 extracted teeth (40 teeth from young individuals between 12 and 25 years and the other 40 teeth from elderly people above 60 years). The results showed that Octenisept has the highest antifungal effect against C. albicans. Octenisept also showed better antifungal effect on dental samples in the younger population than in the older population. Thus, using 2 mL of Octenisept for one minute can be useful to eradicate C. albicans from infected root canals.7 The effectiveness of the antimicrobial activity of alpha-mangostin against endodontic pathogens in a multi-species bacterial-fungal biofilm model was evaluated. The results showed that the novel 0.2% alpha-mangostin can inhibit the metabolic activity of bacterial-fungal biofilms, including C. albicans, and has the potential for endodontic therapy.8 Furthermore, Alexidine digluconate has fungicidal effect and can inhibit the biofilm formation of diverse fungi. This chemical compound has also demonstrated longer-lasting antimicrobial activity than chlorhexidine. Even when combined with NaOCl, no precipitation or side effects have been reported for Alexidine. Thus, it is recommended as more effective and safer for endodontic irrigating solutions, particularly in the case of C. albicans.2 To eradicate C. albicans, Kerlikowski et al. investigated the effect of cold atmospheric pressure plasma on C. albicans in root canals using mono and combination methods with other irrigation solutions (NaOCl, chlorhexidine, and octenidine). The researchers applied the cold atmospheric-pressure plasma jet kINPen 08 with argon gas flow at 5 slm and an admixture of 1% oxygen (Plasma/O2). Based on 150 extracted teeth, the results indicated that Plasma/O2 reduced the highest log10 CFU of C. albicans at 6 and 12 min of treatment times. Plasma/O2 alone provided the best antiseptic properties compared to the other groups. Accordingly, it is suggested for the eradication of C. albicans during root canal therapy.9 Moradi Eslami et al. evaluated the efficacy of Ca(OH)2, triple antibiotic paste (i.e., ciprofloxacin, minocycline, and metronidazole), toluidine blue, light emitting diode (630 nm LED exposure), diode laser (940 nm), and photodynamic therapy (1 mg/ml toluidine blue + 630 nm LED exposure) in eradicating microorganisms from 84 teeth. The three experimental groups including photodynamic therapy, light emitting diode, and triple antibiotic paste showed satisfactory results in eradicating C. albicans and another microbial biofilm. Overall, the triple antibiotic paste showed better results in decreasing the biofilm thickness of microorganisms.3 As root-filling materials, epoxy resin-based sealers (e.g., AH Plus) exhibited satisfactory results against C. albicans.10 Due to the ability of C. albicans to form biofilms, invade dentinal tubules, and resist common irrigation solutions, eradication of C. albicans from infected root canals is a challenging issue, particularly during the pandemic. Therefore, clinicians should consider the possibility of C. albicans in the case of retreatment or apical periodontitis.2 Based on the information provided in this brief letter, the eradication of C. albicans requires additional treatment strategies: 1) using combined methods (e.g., ultrasonic technique + antimicrobial gels (metronidazole benzoate, chlorhexidine diacetate 2%, and hydrocortisone acetate), photodynamic therapy (600 nm diode laser) + mechanical instrumentation, and novel curcumin + ultrasonic), 2) endodontic disinfectants (e.g., Octenisept, alpha-mangostin, and Alexidine), 3) Plasma/O2, 4) triple antibiotic paste (i.e., ciprofloxacin, minocycline, and metronidazole), and 5) AH Plus as root filling martial. Hence, appropriate treatment methods are essential in optimizing endodontic therapy, especially during the COVID-19 pandemic. ==== Refs References 1 Nasiri K. Wrbas K.T. Successful root canal therapy during COVID-19 pandemic J Dent Sci 17 2022 1079 1080 35222838 2 Yoo Y.J. Kim A.R. Perinpanayagam H. Han S.H. Kum K.Y. Candida albicans virulence factors and pathogenicity for endodontic infections Microorganisms 8 2020 1300 32858856 3 Moradi Eslami L. Vatanpour M. Aminzadeh N. Mehrvarzfar P. Taheri S. The comparison of intracanal medicaments, diode laser and photodynamic therapy on removing the biofilm of Enterococcus faecalis and Candida albicans in the root canal system (ex-vivo study) Photodiagnosis Photodyn Ther 26 2019 157 161 30708091 4 Horlenko I.M. Gadzhula N.G. Cherepakha O.L. Kurdysh L.F. Pylypiuk O.Y. Clinical and microbiological assessment of root canal decontamination in chronic apical periodontitis using the ultrasound Wiad Lek 73 2020 1119 1123 32723937 5 Mustafa M. Alamri H.M. Almokhatieb A.A. Alqahtani A.R. Alayad A.S. Divakar D.D. Effectiveness of antimicrobial photodynamic therapy as an adjunct to mechanical instrumentation in reducing counts of Enterococcus faecalis and Candida albicans from C-shaped root canals Photodermatol Photoimmunol Photomed 38 2022 328 333 34748657 6 Yasini Z. Roghanizad N. Fazlyab M. Pourhajibagher M. Ex vivo efficacy of sonodynamic antimicrobial chemotherapy for inhibition of Enterococcus faecalis and Candida albicans biofilm Photodiagnosis Photodyn Ther 40 2022 103113 7 Reddy N.B.N. Sridhar D. Rajkumar A. Murugesan S. Selvaraj K. Sankar S. Comparative evaluation of antifungal activity of Octenidine: an in vitro confocal laser study J Contemp Dent Pract 21 2020 905 909 33568613 8 Leelapornpisid W. Efficacy of alpha-mangostin for antimicrobial activity against endodontopathogenic microorganisms in a multi-species bacterial-fungal biofilm model Arch Oral Biol 133 2022 105304 9 Kerlikowski A. Matthes R. Pink C. Effects of cold atmospheric pressure plasma and disinfecting agents on Candida albicans in root canals of extracted human teeth J Biophotonics 13 2020 e202000221 10 Rathod R.K. Taide P.D. Dudhale R.D. Assessment of antimicrobial efficacy of bioceramic sealer, epiphany self-etch sealer, and AH-Plus sealer against staphylococcus aureus and Candida albicans: an in vitro study Niger J Surg 26 2020 104 109 33223806	36475058	PMC9715484	NO-CC CODE	2022-12-12 23:20:05	no	J Dent Sci. 2022 Dec 2; doi: 10.1016/j.jds.2022.11.027	utf-8	J Dent Sci	2,022	10.1016/j.jds.2022.11.027	oa_other
==== Front J Emerg Nurs J Emerg Nurs Journal of Emergency Nursing 0099-1767 1527-2966 Published by Elsevier Inc. on behalf of Emergency Nurses Association S0099-1767(22)00317-8 10.1016/j.jen.2022.11.013 Research Role Delineation of the Code Blue Team: A Quasi-experimental Study During Covid-19 DeGroot Danika MSN, RN, CEN, PCCN 1 Callis Annette PhD, MSN, RN, CNS 1 1 School of Nursing, Vanguard University 2 12 2022 2 12 2022 14 7 2022 12 11 2022 27 11 2022 © 2022 Published by Elsevier Inc. on behalf of Emergency Nurses Association. 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. Objective The purpose of this study was to assess if implementing a code role delineation intervention in an Emergency Department would improve the times to defibrillation and medication administration, and also improve the nurse perception of teamwork. Method A quantitative quasi-experimental study used a retrospective chart review to gather data. A pre- and post- test was used to measure nurse perception of teamwork in a code using the Mayo High Performance Teamwork Scale (MHPTS) after a code role delineation intervention using a paired samples t-test. Nurse participant (N=30) demographics were also collected. Pearson r correlations were utilized to determine relationships between nurse demographics and the results of the MHPTS scores. Results There was a significant increase in teamwork noted in five of the sixteen items on the MHPTS regarding improved communication and identified roles in a code: the team leader assures maintenance of an appropriate balance between command authority and team member participation (t =-5.607, p < .001), each team member demonstrates a clear understanding of his or her role (t =-5.415, p < .001), team members repeat back or paraphrase instructions and clarifications to indicate that they heard them correctly (t =-2.400, p =.029), all members of the team are appropriately involved and participate in the activity (t =-2.236, p =.041), and disagreements or conflicts among team members are addressed without a loss of situation awareness (t =-2.704, p =.016). There was also significance between total pre- and post-test scores (t = -3.938, p = .001). No statistically significant findings were determined related to improved time to defibrillation and medication administration. Conclusion Implementation of code role delineation identifiers is an effective method of improving teamwork in a code in an acute care hospital setting. Keywords CODE BLUE NURSE ROLE DELINEATION NURSE TEAMWORK EMERGENCY DEPARTMENT ==== Body pmcIntroduction Cardiac arrest is a frequent occurrence in the emergency department, with approximately 200,000 cardiac arrest cases occurring every year in the hospital setting in the USA.5 While the American Heart Association (AHA) has clear recommended guidelines for Advanced Cardiovascular Life Support (ACLS), these guidelines are sometimes difficult to follow due to a myriad of extenuating circumstances. The highly stressful nature of an arrest situation warrants specific well-defined guidelines and protocols related to role responsibilities, to promote organization and positive outcomes of the code.1 Optimizing nurse competencies and confidence levels play a significant role during a code.4 In a code situation, having clearly defined role delineations may be essential to optimal patient care outcomes by decreasing the time in which patients are able to receive life saving measures, such as medication administration and defibrillation.7 This research was aimed at establishing a standard of care for a code blue team related to role delineation. Literature Review Studies have shown that having clearly distinguished roles within a code blue can improve patient outcomes by improving communication amongst the code blue team members and decreasing the time to defibrillation and medication administration.7 It is shown that clearly delineated responsibilities improved nurse confidence in initiating actions which improved overall efficiency and speed of their own actions.4 It was shown that a consistent layout and role definition amongst code participants showed improvement when ACLS guidelines were followed and roles and responsibilities of the team were clearly defined.3 The code team also maintained a professional environment and overall improved effectiveness of the code team was reported. The use of physical role delineation lanyards to clearly state individual roles has shown "improved confidence in their role specific skills, clarity in their role positions, and team leadership, as well as a decrease in the time-to-defibrillation."7 In addition, the use of clearly delineated team identifiers has shown trends towards improved patient outcomes and speed to defibrillate shockable rhythms.8 Roles must work together to form a team in a code situation. Study participants rated their teamwork abilities higher after having a role assigned to them by the facilitator prior to a simulation code blue.2 Without clear role definitions, "role ambiguity and confusion for code team members often exists, possibly creating poor communication and ineffective teamwork that lead to poor patient outcomes.”6 By clearly delineating roles and having a team leader excuse those without a role from the code, the number of providers in the room decreased and code members had a more positive outlook on the nurse leader.5 Materials and Methods This quasi-experimental study took place within a 36-bed emergency department at a 320-bed hospital in Southern California, as complete randomization of data would not be practical in a small sample size. Data collection included a retrospective chart review and a pre- and post-test utilizing a psychometrically tested instrument called the Mayo High Performance Teamwork Scale. The convenience sample was drawn from a target population of emergency department registered nurses (N=90) employed at the study location. Prior to the intervention, the standard flow of the department was to have one nurse designated to specific rooms. If a patient in cardiac arrest or who arrested later was placed in a room assigned to a nurse, that nurse would automatically be assigned as the primary nurse. Beyond the primary nurse, however, there was little guidance on who would assist and in what role. If it was a less busy day in the department, there could be more than the necessary number of nurses present while a busy day may yield too few nurses. For example, the primary nurse would randomly assume any one of the three roles depending on the needs of the patient. The other two roles were typically assumed by another staff nurse, the charge nurse, or the rapid response nurse. There was no defined standard regarding who would assume each role. Without definitive roles, crowd control could be difficult to manage. Within this department, a standard of care related to role delineation in a code utilizing role delineation badges was created in order to provide optimal care for all patients who arrest. Role delineation badges were placed on each crash cart to be utilized in every code during the research period. Nurses who participated in code situations during the span of the six-month intervention period were eligible for inclusion and were invited to participate in the study. Study volunteers were asked to complete a teamwork assessment instrument before and after implementation of the role delineation standard of care. A Cronbach Alpha was calculated in order to determine reliability of the study survey items, with a result of .88. This is comparable to what was found in the literature, in which the authors of the scale found a Cronbach Alpha of .81-.85. This signifies appropriate reliability of the survey instrument. Significance was found in five questions; the team leader assures maintenance of an appropriate balance between command authority and team member participation, each team member demonstrates a clear understanding of his or her role, team members repeat back or paraphrase instructions and clarifications to indicate that they heard them correctly, all members of the team are appropriately involved and participate in the activity, and disagreements or conflicts among team members are addressed without a loss of situation awareness. Select demographics on patients were collected to determine potential influence on code blue data. Items collected included age, outcome, and sex. Nurse demographics collected included age, sex, ethnicity, years of ED employment, certification, shift worked, type of employment, and highest level of education. Study Procedures After obtaining Institutional Review Board (IRB) approval (STUDY2020000644 and 2020FA0003), labels were created by printing specific roles on color-coded cards. These labels were placed in plastic card holders and affixed to a badge clip for nurses to wear. The labels included the RN roles of "Defibrillation," "IV/Medications" and "Documentation." These roles were determined by the hospital's code blue protocol and ACLS guidelines.1 A participant email was sent to each nurse within the ED explaining the study and included a link to the survey pre-test to be completed via Microsoft Forms called the Mayo High Performance Teamwork Scale. The scale is a 16-question Likert scale which allows participants to assess different aspects of teamwork, such as leadership, role clarification, and communication, on a scale of 0-2. This email list was provided by the education department and included the nurses employed at the time of this intervention launch date. Next, each crash cart was supplied with three badges, one for each role. After the dispersion of the labels, a notice of the standard of care change was provided in the daily huddle for one week. All RNs were required to sign that they attended huddle at least once each week. Most of the RNs were documented as having heard this update. To promote intervention fidelity, all rapid response team members were asked to aid in enforcing the use of the labels and in defining the roles during an arrest. In addition, a tracking sheet was created for each crash cart and a patient label was affixed to the sheet after utilizing the badges in the code. After six months of utilizing the new standard of care, volunteers who had participated in a code utilizing the intervention were asked to fill out a short instrument using the Mayo High Performance Teamwork Scale as a post-test in order to determine if there was a difference in nurses' perception of teamwork in a code blue after the role delineation standard of care implementation. The instrument was dispersed on Microsoft Forms via the same email address list. The initial participants remained anonymous. A chart review was then conducted utilizing all 47 charts available from patients in the Emergency Department that were dated in the six months before the intervention and after the initiation of Covid-19 isolation precautions (between March and November 2020). The time to defibrillation and time to initial medication administration was then determined. The medication administration time was recorded by the administered time of several different medications including but not limited to Epinephrine, Atropine, Amiodarone, Bicarbonate, etc. The same was done utilizing 17 charts after the intervention was initiated which had documented nurses using the intervention. While there were more codes during this time frame, only the 17 with documented intervention usage were studied. Each code that is run within this facility is recorded on paper for official documentation. One copy is held with the patient's chart and a carbon copy is submitted to clinical excellence for review. The copies reviewed were requested from the clinical excellence department. The time to defibrillation and medication administration were determined from when the code button was pressed (or the door time of arrival) to the time of the documented defibrillation. These times were documented in minutes according to the current standard at the facility. Select patient demographics and nurse demographics were collected. After the study implementation was initiated, the investigator waited five months to send out an email requesting post-test survey participation. Data Analysis The data was analyzed using SPSS 24. A paired sample t-test was performed on the pre and post-test data from the Mayo High Performance Teamwork Scale to measure teamwork in emergency nurses before and after a role delineation intervention. This analysis compared the item mean scores and total mean scores from the 16 items from the pre-test to the post-test. The items were paired by using an anonymous participant identifier. Demographic data was analyzed using descriptive statistics. Correlational analysis was run to determine relationships between demographic data and pre and post-test Performance Teamwork Scale scores. Nurse Demographics A power analysis showed that 30 nurses should fill out the Mayo High Performance Teamwork Scale. In a department that employed 90 nurses at the time of investigation, 33.3% of ED nurses should participate in the scale. As this intervention involved a standard of care change, all nurses in the department would participate in the role delineation intervention. From the sample (N = 30), few nurses (n = 2, 6.7%) held associate’s degrees, most (n=20, 66.7%) held bachelor's degrees, and some (n = 8, 26.7%) held a master's degree. Ages ranged from 24-63 years-old (M = 37.3, SD = 9.71). The majority were female (n = 25), with 60% Caucasian, 20% Hispanic or Latino, 16.7% Asian or Pacific Islander, and 3% other. The participants worked in the Emergency Department for 0.5 years to 24 years. Of the participants, 14 held an MICN (Mobile Intensive Care Nurse) certification, eight earned their CEN (Certified Emergency Nurse) certification, one had an SCRN (Stroke Certified Registered Nurse), one a CMSRN (Certified Medical-Surgical Registered Nurse), and one was a PHN (Public Health Nurse). Nurse demographics can be found on Table 1 .Table 1 Descriptive statistics of nurse demographics (Total N = 30) Demographic Variable M SD n % Age 37.3 9.71 - - Sex Male Female 5 25 16.7 83.3 Ethnicity Asian or Pacific Islander Hispanic or Latino White or Caucasian Other 5 6 18 1 16.7 20.0 60.0 3 Years employed in ED 8.2 6.36 - - Certifications MICN CEN SCRN CMSRN PHN 14 8 1 1 1 46.6 26.6 3.3 3.3 3.3 Shift worked Dayshift Midshift Nightshift 11 10 9 36.7 33.3 30.0 Type of employment Full-time Part-time Per Diem 27 3 0 90 10 0 Highest level of education Associate degree Bachelor’s degree Master’s degree 2 20 8 6.7 66.7 26.7 Results Twenty-one post tests were collected after five months of the code intervention implementation. Total mean teamwork performance scores improved from a total score of 24.8 to 28.9, out of a maximum score of 32. Of the 21 posttests, 17 were able to be paired to the pretest utilizing the four-digit participant identifier. After analyzing the data of individual items with a paired samples t-test, five items showed a significant difference from pre to post-test at the p = < .05 level. This significance was found in questions two, three, seven, eight, and nine and the total scores. These items read as follows: the team leader assures maintenance of an appropriate balance between command authority and team member participation (t =-5.607, p < .001), each team member demonstrates a clear understanding of his or her role (t =-5.415, p < .001), team members repeat back or paraphrase instructions and clarifications to indicate that they heard them correctly (t =-2.400, p =.029), all members of the team are appropriately involved and participate in the activity (t =-2.236, p =.041), and disagreements or conflicts among team members are addressed without a loss of situation awareness (t =-2.704, p =.016). An independent samples t-test revealed a significant improvement from pre-test to post-test total scores (t = -3.938, p = .001). No correlations were found between certification, number of years in the ED, or age, and the Perceived Teamwork survey scores. Data on all questions for the Mayo High Performance Teamwork Scale is shown in Table 2 .Table 2 Paired t-test of Mayo High Performance Teamwork Scale in emergency nurses (Total N = 17) MHPTS Question Pretest M Posttest M SD t df Sig. (2-tailed) 1 1.41 1.70 .59 -2.063 16 .056 2 1.12 1.88 .56 -5.607 16 .000 3 1.12 1.76 .49 -5.416 16 .000 4 1.59 1.94 .70 -2.073 16 .055 5 1.47 1.82 .70 -2.073 16 .055 6 1.53 1.76 .83 -1.167 16 .260 7 1.53 1.88 .60 -2.400 16 .029* 8 1.50 1.75 .44 -2.236 15 .041* 9 1.47 1.94 .72 -2.704 16 .016* 10 1.88 2.00 .33 -1.461 16 .163 11 1.65 1.82 .53 -1.376 16 .188 12 1.75 1.68 .77 .324 15 .751 13 1.82 1.64 .53 1.376 16 .188 14 1.76 1.76 .50 .000 16 1.000 15 1.69 1.88 .54 -1.379 15 .188 16 1.69 1.94 .58 -1.732 15 .104 Total 24.8 28.9 4.24 -3.938 16 .001** Note. Significance at p = < .05 level. * Significance at p = < .01 level. Table 3 Descriptive Statistics of Patient Demographics (Total N =63) Demographic variable M SD n % Pre-intervention age 65.83 14.39 - - Outcome Survived Expired 18 28 38.3 59.6 Sex Male Female 34 13 72.3 27.7 Post-intervention age 63.38 14.50 - - Outcome Survived Expired 7 8 46.7 17.0 Sex Male Female 10 6 62.5 37.5 Forty-seven code blue charts were collected prior to the intervention, ranging from March 2020 to November 2020. Of these charts, all patients received medications and five were defibrillated during the code. An additional 17 patients were documented as nurses utilizing the code role badges during the code blue from November 2020 to May 2021. These physical charts were obtained from the clinical excellence department. One chart was unable to be located by the PI or the clinical excellence department, so 16 charts were reviewed for analysis. Of these patients, 15 received medication and two were defibrillated. Select demographics were collected on the patients before and after the intervention. The ages of those in the pre-intervention group ranged from 27-96 years of age. In that group, 72.3% were male and 27.7% female. In addition, 39.1% survived Return of Spontaneous Circulation (ROSC) and 60.9% expired. The ages of those in the post-intervention group ranged from 40-82 years. Male patients made up 62.5% while 37.5% were female. In the post intervention group, 46.7% survived after the code while 53.3% expired. Long term survival was not investigated. Prior to the intervention, mean time from time of code blue called to medication administration was 1.55 minutes. Post intervention showed a mean time of 2.08 minutes. An independent samples t-test was conducted to compare the time of medication administration between the pre- and post-intervention groups. There was no apparent difference in time to medication administration between groups. In addition, a comparison of patient survival pre- and post-intervention did not show a significant difference. A Pearson r correlation analysis showed no significance between select patient demographics and medication administration timing. As only five patients pre-intervention and two patients post-intervention were defibrillated, there was not adequate data to assess pre- and post- data. While having role delineation badges did not change the time to medication administration or defibrillation, there was a significant difference in the perception of teamwork overall by the nurses after the intervention. The areas showing most improvement were those areas related to the level of involvement of the nurses and their ability to communicate effectively. Discussion This study complemented what has been viewed in the literature review, that having a physical means of determining nurse roles in a code blue may help to improve the nurse perception of a code blue. As stated in the study findings, nurses found that teamwork aspects such as communication were improved through the use the code role delineation badges. Implications for Emergency Nurses Overall, this intervention has offered an improved standard of care to help Emergency Department nurses be clearer on their roles in code situations. Role delineation may help to improve the overall performance of nurses in code situations. After the implementation of this project in the ED and seeing the positive outcomes, the Code Blue Committee at the study site moved to implement a code blue role intervention over the entire hospital. It is the hope of the principal investigator that the role delineation standard of care will continue to improve nursing teamwork performance during code situations in various hospital settings, as well as the Emergency Department. Other hospital populations may also benefit from such an intervention. Recommendations Future studies are needed to investigate the effectiveness of a hospital wide role delineation intervention in codes. Role confusion during inpatient codes at the study site has been observed. These codes require the arrival of a code team including a rapid response nurse, an ED nurse, the House Supervisor RN, and a charge nurse from ICU. Exploring a nurse leader role was not investigated in this study. Having a method of clearly delineated roles for nurses who may not know each other well may potentially be beneficial in improving communication and teamwork. It is recommended that a similar longitudinal study be performed with a larger sample size. There was not enough data regarding time for medication or defibrillation to determine significance as this project ran for a short period, only six months. A power analysis indicated that at least forty code blue charts would be needed to obtain enough data for an adequate effect size. While over 40 code blue charts were able to be obtained for the pre-review, there were only 16 charts documented as having used the intervention available post-intervention and therefore not enough data to determine all the differences in pre-intervention group and post-intervention group items due to the smaller post-review size. In addition, the total number of code blues in the intervention period was not recorded, but the data may have offered insight into code blue performance as a whole during the study intervention. Similarly, while 30 nurses filled out the initial surveys, only 17 were able to be paired with post-intervention surveys. Both events may have caused a Type II error, as there could have been greater significance if the sample of charts collected post-intervention was larger. If this research study is repeated or continued, it should be introduced to multiple hospital units to gain more participants from a wider variety of specialties and to have a higher incidence of a code blue. It may also be beneficial to create an easier method of tracking whether the role delineation badges were used, as most code blues did not document usage and were unable to be included in the data collection. Limitations Initially the nurses were hesitant to participate. There was pushback as nurses did not want to wear role badges or they forgot to wear them. Many of the nurses who wore the badges did not put them back onto the crash cart after using them. In addition, nurses stated they forgot to put patient labels on the tracking forms. The chart review revealed a need for education for nurses on how to fill out a code documentation form correctly. Occasionally, vital information was lacking from the forms. It is the policy of the facility at the time of the code to have all the information from the code documented on paper alone. No information needs to be converted into the electronic medical record by the RN. It was noted during the chart review that many patients were missing information such as initial cardiac rhythm, pulse checks, and patient outcome. Limited data and timing of the study contributed to the overall study limitations. This study pre-test was distributed prior to the Covid-19 surge that took place November 2020 to February 2021. During and after the surge, it was noted that many ED nurses chose to move to a different specialty or quit entirely. As a result, there were fewer nurses who participated in the post-survey than who were working in the department at the time of the pre-survey distribution. In addition, during the post-survey period, the hospital stated a record high number of nurses on LOA (leave of absence) for various reasons. It is the policy of the hospital to not check emails and respond while on LOA. This also impacted the number of nurses who may have filled out the pre-survey but were not available during the post-survey period. Launching the initial intervention during a pandemic created several obstacles. There was some hesitation due to the ED nurses not having the energy to add to an already draining workload. Crash carts occasionally were not used for codes during the first few months of the intervention, as they occurred so frequently that respiratory trays and a defibrillator were sometimes used in lieu of bringing the crash cart into a room with a Covid positive patient. As the badges were stored on top of the crash cart, any code ran in this method would likely not have been recorded as using the intervention. It would be more conducive to implement the intervention at a time when the ED is well staffed with a more consistent patient census and set of resources. This may also serve to eliminate some of the extraneous variables that occurred because of the pandemic such as unusual numbers of codes overall and occurring simultaneously, severe contamination issues, intensified short staffing, and extremely high patient acuity levels. Strengths One strength of this research was the willingness of the rapid response nurses to maintain intervention fidelity and ensure the proper labeling of the nurses and the code tracking form. The compliance of the rapid response team is especially helpful if this intervention moves to a hospital wide setting, as the rapid response nurse responds to all codes in the hospital. Anecdotally, the PI received positive feedback from the rapid response nurses that they perceived improvement of the organization of the code with the role delineation intervention. Conclusion While time to medication administration and defibrillation did not show statistical improvement, nurses stated that their perception of teamwork did improve compared to the original practice of nurses volunteering for a code and helping with several roles within the code blue. This study regarding code role identification for nurses has offered meaningful evidence as to its effectiveness in improving nurse teamwork. The Mayo High Performance Teamwork Scale showed improvement in the following items after the role delineation intervention: the team leader assures maintenance of an appropriate balance between command authority and team member participation, each team member demonstrates a clear understanding of his or her role, team members repeat back or paraphrase instructions and clarifications to indicate that they heard them correctly, all members of the team are appropriately involved and participate in the activity, and disagreements or conflicts among team members are addressed without a loss of situation awareness. Total mean scores of the Mayo High Performance Teamwork Scale improved from 24.8 to 28.9 out of 32 points. It is recommended that this study be repeated involving nurses from a wider variety of specialties. In addition, a longitudinal study could provide more data regarding time to medication administration and defibrillation in a code. This study intervention suggests role delineation is an effective method to improve nursing teamwork in times of patient arrest in the Emergency Department and may lead to improved overall nursing performance and therefore may improve patient outcomes. Header: Research Contribution to Emergency NursingPoor communication between nurses can occur during code situations within the Emergency Department. Clearly delineated roles during codes are encouraged according to ACLS guidelines but may not consistently be utilized in practice. Extensive research has been done regarding communication and teamwork in nursing during codes. This manuscript offers evidence regarding nurse perception of teamwork in relation to the use of role delineation badges during codes. The use of code blue role delineation badges may be a simple and inexpensive way for emergency departments to improve their communication and teamwork within a code blue. Conflict of Interest We have no conflicts of interest to disclose. Conflict of Interest Statement We have no conflicts of interest to disclose regarding this research. There is no financial benefit to either author as a result of this study. In addition, there are no family or personal relationships that may result in a conflict of interest. Danika DeGroot, MSN, RN, CEN, PCCN Annette Callis, PhD, MSN, RN, CNS ==== Refs References 1 American Heart Association. Advanced Cardiovascular Life Support (ACLS) Provider Manual. 16th ed. 2016. 2 Ballangrud R. Persenius M. Hedelin B. Hall-Lord M.L. Exploring intensive care nurses’ team performance in a simulation-based emergency situation, - expert raters’ assessments versus self-assessments: an explorative study BMC nursing 13 1 2014 47 10.1186/s12912-014-0047-5 25606023 3 Dorney P. Code blue: chaos or control, an educational initiative Journal for nurses in staff development: JNSD: official journal of the National Nursing Staff Development Organization 27 5 2011 242 244 10.1097/NND.0b013e31822d6ee4 21946796 4 Lanfranchi J.A. Instituting Code Blue Drills in the OR AORN Journal 97 4 2013 428 434 10.1016/j.aorn.2013.01.017 23531309 5 Leary M. Schweickert W. Neefe S. Tsypenyuk B. Falk S.A. Holena D.N. Improving Providers’ Role Definitions to Decrease Overcrowding and Improve In-Hospital Cardiac Arrest Response American journal of critical care: an official publication, American Association of Critical-Care Nurses 25 4 2016 335 339 10.4037/ajcc2016195 27369032 6 O’Donoghue S.C. DeSanto-Madeya S. Fealy N. Saba C.R. Smith S. McHugh A.T. Nurses’ Perceptions of Role, Team Performance, and Education Regarding Resuscitation in the Adult Medical-Surgical Patient Medsurg nursing: official journal of the Academy of Medical-Surgical Nurses 24 5 2015 309 317 26665866 7 Prince C.R. Hines E.J. Chyou P.-H. Heegeman D.J. Finding the key to a better code: code team restructure to improve performance and outcomes Clinical medicine & research 12 1-2 2014 47 57 10.3121/cmr.2014.1201 24667218 8 Spitzer C.R. Evans K. Buehler J. Ali N.A. Besecker BY. Code blue pit crew model: A novel approach to in-hospital cardiac arrest resuscitation Resuscitation 143 2019 158 164 10.1016/j.resuscitation.2019.06.290 31299222	0	PMC9715485	NO-CC CODE	2022-12-13 23:16:28	no	J Emerg Nurs. 2022 Dec 2; doi: 10.1016/j.jen.2022.11.013	utf-8	J Emerg Nurs	2,022	10.1016/j.jen.2022.11.013	oa_other
==== Front J Stroke Cerebrovasc Dis J Stroke Cerebrovasc Dis Journal of Stroke and Cerebrovascular Diseases 1052-3057 1532-8511 Elsevier Inc. S1052-3057(22)00610-3 10.1016/j.jstrokecerebrovasdis.2022.106918 106918 Article Evaluation of ABCD2 score during the development of stroke in COVID-19 patients diagnosed with transient ischemic attack in the emergency department Yurtsever Güner MD ⁎ Bora Ejder Saylav MD, PhD Karaali Rezan MD Department of Emergency Medicine, Izmir Ataturk Training and Research Hospital, Izmir, Turkey ⁎ Correspondence author: Güner Yurtsever, MD, Izmir Ataturk Training and Research Hospital, Department of Emergency Medicine, Basın Sitesi-Izmir 35360/ TURKEY, Tele: +90 (232) 243 43 43, +90 (232) 244 44 44, Fax: +90 (232) 243 15 30,GSM: +90 506 680 45 94 2 12 2022 2 12 2022 10691823 6 2022 22 11 2022 27 11 2022 © 2022 Elsevier Inc. All rights reserved. 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 The aim of the present study is to reveal the association between the risk of stroke using ABCD2 score and COVID-19 in patients who presented to our emergency department during the pandemic and were diagnosed with TIA. Methods According to the recommendations of the European Stroke Association, patients with an ABCD2 score of <4 were classified as low-risk, and patients with an ABCD2 score of ≥4 were classified as high-risk. Within 90 days of the patient's admission to the emergency room, the development of stroke was tracked and recorded on the system. Results Stroke occurred in 35.78% of the patients. Regarding COVID-19, 75.34% of stroke patients were positive for COVID-19 and 65.75% had COVID-19 compatible pneumonia on 'thoracic CT'. Regarding mortality, 16.4% of the patients who were positive for COVID-19 and developed a stroke died. The presence of COVID-19 compatible pneumonia on thorax CT, PCR test result and ABCD2 score were determined as independent risk factors for the development of stroke. According to the PCR test results, the probability of having a stroke decreases 0.283 times in patients who are negative for COVID-19. According to the PCR test results, the probability of having a stroke increased 2.7 times in COVID-19 positive patients. Conclusions Adding the presence of COVID-19 and the presence of COVID-19 pneumonia to the ABCD2 score, based on the information about the increased risk of stroke in TIA patients, improves the predictive power of the score. More studies are needed in this regard. Keywords SARS-CoV-2 Pandemic Transient Ischemic Attack ABCD2 Score ==== Body pmcINTRODUCTION Transient ischemic attack (TIA) is defined by the American Stroke Association (ASA) as a transient episode of neurological dysfunction caused by an inability to supply blood to the brain for a certain period (1, 2). Some 240,000 TIAs are reported in the United States annually (3). Transient ischemic attacks are associated with a 5% permanent stroke risk within 48 hours and 10% within 3 months, unless specifically treated (4). In other words, TIAs can be considered a precursor to an ischemic CVE, making it important to subject patients presenting with TIAs to careful examination. The main purpose of evaluating patients presenting with TIAs in the emergency department is to identify any risk factors that may cause stroke and to prevent potential strokes with appropriate treatment (1, 5). Studies have reported that the SARS-CoV-2 virus, which has been affecting the whole world for over two years, can cause cerebrovascular events due to the increased risk of thrombosis and neurotropic effects (6, 7, 8). The ABCD2 score, as recommended in guidelines, is the scoring system most commonly used for the determination of the risk of stroke in TIA patients (9, 10). An easy and practical tool, the ABCD2 score has been found to accurately distinguish between TIA patients who develop a stroke within 7–90 days and those who do not (11). To date, however, the presence of COVID-19 has not been included in any risk scoring system used to predict the risk of stroke in TIA patients. The present study investigates the association between the risk of stroke identified based on the ABCD2 score and COVID-19 in patients who presented to our emergency department during the pandemic and were diagnosed with TIA. MATERIALS AND METHOD Study design and setting This observational and retrospective study was conducted at a single center, and was approved by the Turkish Ministry of Health (Date: 01.10.2021 and No: T15-17-23) and the Ethics Committee of Izmir Katip Çelebi University (Date: 21.10.2021 and No: 0434). Included in the study were patients who presented to the emergency department of our hospital between 30.04.2019 and 30.09.2021 and who met the inclusion criteria. Study population The data of patients who presented to the emergency department of our hospital during the pandemic and who were diagnosed with TIA based on examination, tests and a neurological assessment were reviewed retrospectively within the hospital automation system. Included in the study were patients who were over the age of 18, with no trauma, who were not pregnant and whose accessed files contained the necessary data. Data collection The age, gender, blood pressure and fever of the patients were recorded. Heart rhythm and any abnormalities on the electrocardiograms of the patients were recorded. The complaints of the patients at admission to the emergency department were classified as loss of strength and/or sensation (lower extremities, upper extremities, lower and upper extremities), speech disturbances, and other findings (amnesia, dizziness, clouding of consciousness, vision loss, facial paralysis). TIA is defined by the World Health Organization (WHO) as an acute loss of focal cerebral or ocular function with symptoms lasting less than 24 hours, and which after adequate investigation, are presumed to be due to embolic or thrombotic vascular disease (5, 12,13). As per the recommendations (5, 1) in TIA management guidelines of the American Heart Association/American Stroke Association (AHA/ASA) and the European Stroke Association (ESO), the CT brain, CT brain-neck angiography (GE Revolution EVO® 128-slice) and diffusion MRI (Siemens Magnetom Aera® 1.5 tesla) results were recorded. The patients without acute ischemic restricted diffusion on MRI were considered to have TIA, while those with unilateral/bilateral, complete/partial occlusion/plaque in the internal carotid artery, external carotid artery or carotid artery on CT brain/neck angiography were defined as having a vascular pathology. The ABCD2 score was calculated as follows: age >60 (1 point); blood pressure >140/90 mmHg (1 point); clinical features – unilateral weakness (2 points) or speech disturbance without weakness (1 point); duration of symptoms >60 minutes (2 points) or 10–59 minutes (1 point); and diabetes mellitus (1 point) (5). Blood pressure was identified based on a systolic pressure of >140 mmHg and/or a diastolic pressure of >90 mmHg at admission. Diabetes mellitus was defined based either on existing history or a new diagnosis (9, 10). As per the recommendations of the European Stroke Association, patients with an ABCD2 score of <4 were classified as low-risk, and those with an ABCD2 score of ≥4 as high-risk (5). COVID-19, following the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia (Trial Version 7), was diagnosed based on the viral nucleic acid determination through the reverse transcription-polymerase chain reaction (Biospeedy® RT-PCR test) testing of nasopharyngeal swab samples (13). Patients with two negative test results were considered negative. The patients were divided into COVID-19 positive and COVID-19 negative groups based on the RT-PCR test results. All acute patients with respiratory system complaints underwent a low-dose thoracic computed tomography (CT). The Thoracic Computed Tomography (TBT) findings for COVID-19 pneumonia were assessed according to the Radiological Society of North America (RSNA) criteria (14). Patients reported as type 1 and type 2 were considered COVID-19 compatible (Group 1), and those reported as type 3 and type 4 were considered COVID-19 incompatible (Group 2). The hospitalization type (intensive care/ward) and mortality of the patients were recorded, and the e-Nabız system of the Turkish Ministry of Health was used to learn whether the patients had any CVE within 90 days. Patients who presented to any healthcare facility, including our hospital, within 90 days of the TIA diagnosis, who had focal neurological symptoms and signs persisting for at least 24 hours, who had no signs of mass/bleeding on a brain CT scan, who had MRI findings of acute ischemia and who were diagnosed with acute ischemic stroke after assessment by a neurologist according to the World Health Organization criteria were defined as patients with stroke (12). Statistical Analysis The study data was assessed using IBM SPSS Statistics (Version 20.0. Armonk, NY: IBM Corp.). Frequency and percentage distribution were calculated for descriptive statistics, and mean, standard deviation, and minimum and maximum values for continuous variables. Kolmogorov-Smirnov and Shapiro-Wilk (p<0.05) tests were used to analyze the normality of the continuous variables, based on which it was selected whether to use either parametric or non-parametric tests. Chi-Square test statistics were used to compare the categorical variables between groups, while Mann-Whitney U statistical analyses were used for comparisons of two groups when the continuous data consisted of non-normally distributed values. A Binary Logistic Regression model was used to examine the effect of the variables on stroke and the cause-effect relationship between the dependent variable and independent variables. RESULTS The study included 204 patients, 46.08% of which were male, and the mean age was 63±13 years. Stroke occurred in 35.78% of the patients, and 53.43% of the stroke patients were male. There was a statistically significant difference in the patients’ blood pressure, fever, ECG findings, presence of DM, and mortality according to stroke status (p<0.05). The majority of patients had symptoms lasting 10–60 min, and the risk of stroke was higher in patients with this duration of symptoms. CT angiography revealed no pathology in the extracerebral arteries in 65.2% of the patients and in 36.98% of the stroke patients. Concerning COVID-19, 75.34% of the stroke patients were COVID-19 positive and 65.75% had COVID-19-compatible pneumonia on “thoracic CT” (table 1 ).Table 1 Comparison of general characteristics of patients and parameters in terms of stroke Table 1Variables All patients Developing stroke Non- stroke p n=204 n=73 n=131 Mean ± SD Mean ± SD Mean ± SD (Min -Max) (Min -Max) (Min -Max) Age 63±13 65±15 62±12 0,1 (32 - 90) (32- 90) (33 -86) Systolic BP 137±25 143±22 135±27 0,02 (96 - 240) (100 - 206) (96 - 240) Diastolic BP 78±15 81±14 77±15 0,01 (54 - 135) (58 - 135) (54 - 126) Fever 36,9 ± 0,8 37,16 ±0,73 36,81 ± 0,73 0 (36,0 - 38,7) (36 - 38,6) (36,0 - 38,7) n (%) n (%) n (%) p Gender Male 94 (46,08) 39 (53,43) 55 (41.98) 0,11 Female 110 (53,92) 34 (46,57) 76 (58.01) ECG AF 32 (15,69) 16 (21,91) 16 (12,22) 0,07 SR 172 (84,31) 57 (78,09) 115 (87,78) Motor Weakness Lower Extremity 2 (0.98) 0 (0) 2 (1.56) 0,47 Upper Extremity 55 (26.96) 21 (28.76) 34 (25.95) Lower and Upper Extremities 41 (20.09) 15 (20.54) 26 (19.84) None 106 (51,96) 37 (50,68) 69 (38,69) Speech Disorder Exist 92 (45,1) 39 (53,43) 53 (40,45) 0,74 None 112 (54,9) 34 (46,57) 78 (59.54) Other Findings Amnesia 16 (7,84) 5 (6.84) 11 (8.39) 0,34 Dizziness 16 (7.84) 5 (6.84) 11 (8.39) Consciousness 8 (3.92) 3 (4.1) 5 (3.81) Vision Loss 20 (9.8) 4 (5.47) 16 (12.23) Facial Paralysis 17 (8.3) 6 (8.2) 11 (8.39) None 127 (62,25) 50 (68,49) 77 (52,28) Duration of Symptoms 10 min< 72 (35,29) 17 (23,28) 55 (17,77) 0,01 10-60 min 83 (40,69) 31 (42,46) 52 (32,41) 0,7 >60 min 50 (24,51) 26 (35,61) 24 (27,18) 0,01 HT Exist 102 (50) 41 (56,17) 61 (46.57) 0,19 None 102 (50) 32 (43,83) 70 (53.43) DM Exist 50 (24,51) 31 (42,46) 19 (14.51) 0 None 154 (75,49) 42 (57,53) 112 (85.49) Mortality Exist 13 (6,37) 9 (12,33) 4 (3.05) 0,01 None 191 (93,63) 64 (87,67) 127 (96.94) Results of CT Angiography Carotis 6 (2.94) 3 (4.1) 3 (2.29) 0 No vascular occlusion 133 (65.19) 27(36.98) 106 (80.91) ECA 14 (6.86) 9 (12.32) 5 (3.81) ICA 51 (25) 34 (46.57) 17(12.97) Thoracic CT Covid Compatible 72 (35,29) 48 (65,75) 24 (18.32) 0 Covid Noncompatible 70 (34,31) 19 (26,03) 51 (38.93) None 62 (30,39) 6 (8,22) 56 (42.74) PCR Covid Negative 108 (52,94) 18 (24,65) 90 (68.70) 0 Covid Positive 96 (47,06) 55 (75,34) 41 (31.29) ABCD2 Score <4 112 (54,9) 24 (32,87) 88 (67.17) 0 SD: Standart Deviation BP: Blood Pressure ECG: Electrocardiogram AF: Atrial Fibrillation SR: Sinus Rhythm HT: Hypertension DM: Diabetes Mellitus CT: Computed Tomography ECA: External Carotid Artery PCR: Polymerase Chain Reaction ICA: Internal Carotid Artery Regarding COVID-19 and stroke, the patients who were positive for COVID-19 and who developed stroke were in the advanced age group (67±14), 32.7% had no pathology in the extracerebral arteries, 69.1% had COVID-19 compatible pneumonia and 74.5% had an ABCD2 score of >4. Concerning mortality, 16.4% of the patients who were positive for COVID-19 and who developed stroke died (table 2 ).Table 2 Evaluation of patients in terms of COVID-19 positivity and stroke development status Table 2 No stroke Developing stroke Covid Negative Covid Positive Covid Negative Covid Positive Mean + Standard Deviation Mean + Standard Deviation Mean + Standard Deviation Mean + Standard Deviation (Min - Max) (Min - Max) (Min - Max) (Min - Max) Age 61±12 63±13 60±16 67±14 (33-86) (36-84) (32-85) (32-90) Systolic BP 132±25 140±30 133±22 146±22 (96-206) (110-240) (100-188) (110-206) Diastolic BP 76±15 78±16 78±13 82±15 (54-116) (57-126) (60-110) (58-135) Fever 37±1 37±1 37±1 37±1 (36-39) (36- 38) (36-39) (36-39) n (%) n (%) n (%) n (%) ECG AF 6 (6.7) 10 (24.4) 1 (5.6) 15 (27.3) SR 84 (93.3) 31 (75.6) 17 (94.4) 40 (72.7) Motor Deficit Lower Extremity 1(1.11) 1 (2.43) 0 (0) 0 (0) Upper Extremity 21(23.33) 13 (31.7) 7(38.88) 14(25.45) Lower+Upper Extremity 18 (20) 8(19.51) 2(11.11) 13 (23.63) None 50 (55.6) 19 (46.3) 9 (50) 28 (50.9) Aphasia Exist 35 (38.9) 18 (43.9) 11 (61.1) 28 (50.8) None 55 (61.1) 23 (56.1) 7 (38.9) 27 (49.1) Other Findings Amnesia 8 (8.9) 3 (7.31) 0 5 (9.1) Dizziness 6(6.66) 4(9.75) 0 (0) 5(9.1) Consciousness 3(3.33) 2(4.87) 0 (0) 3(5.45) Vision Loss 11 (12.2) 5 (12.19) 2 (11.1) 2 (3.6) Facial Paralysis 10(11.11) 2(4.87) 2(11.1) 3(5.45) None 52 (57.8) 25 (61) 14 (77.8) 36 (65.5) Duration of Symptoms 10 min< 41 (45.6) 14 (34.1) 7 (38.9) 10 (18.2) 10-60 min 32 (35.6) 20 (48.8) 11 (61.1) 20 (36.4) >60 min 17 (18.9) 7 (17.1) 0 (0) 25 (45.5) HT Exist 41 (45.6) 20 (48.8) 11 (61.1) 30 (54.5) None 49 (54.4) 21 (51.2) 7 (38.9) 25 (45.5) DM Exist 8 (8.9) 11 (26.8) 5 (27.8) 26 (47.3) None 82 (91.1) 30 (73.2) 13 (72.2) 29 (52.7) Mortality Exist 2 (2.2) 2 (4.9) 0 9 (16.4) None 88 (97.8) 39 (95.1) 18 (100) 46 (83.6) Vascular Pathology Carotis 0 (0) 3(7.31) 1(5.55) 2(3.63) No Vascular Occlusion 74 (82.2) 32 (78) 9 (50) 18 (32.7) ECA 4(4.44) 1 (2.43) 3(16.66) 6(10.90) ICA 12(13.33) 5(12.19) 5(27.77) 29(52.72) Thoracic CT Covid compatible 10 (11.1) 14 (34.1) 10 (55.6) 38 (69.1) Covid noncompatible 31 (34.4) 20 (48.8) 4 (22.2) 15 (27.3) None 49 (87.5) 7 (17.1) 4 (22.2) 2 (3.6) ABCD2 Score Low Risk 67 (74.4) 21 (51.2) 10 (55.6) 14 (25.5) High Risk 23 (25.6) 20 (48.8) 8 (44.4) 41 (74.5) ABCD2: Age,Blood Pressure, Clinical Features, Duration of TIA, Diabetes BP: Blood Pressure ECG: Electrocardiogram AF: Atrial Fibrillation SR: Sinus Rhythm HT: Hypertension DM: Diabetes Mellitus CT: Computed Tomography Finally, a Binary Logistic Regression analysis was conducted to identify the variables that may affect stroke development in patients diagnosed with TIA. “Stroke Status” was determined as the response variable and the effects of variables that may affect stroke were examined. The presence of COVID-19-compatible pneumonia on thoracic CT, PCR test results and ABCD2 scores were identified as independent risk factors determining the development of stroke in the patients (p<0.05). As the ABCD2 score (p= 0.008, Exp (B)= 0.399) decreases, so does the probability of stroke up to 0.399 times. The higher the ABCD2 score, the higher the probability of stroke. Patients with high ABCD2 scores were 1.5 times more likely to have a stroke than those with low scores. The absence of COVID-19-compatible pneumonia on thoracic CT (p= 0.004, Exp (B)= 0.240) reduces the probability of having a stroke up to 0.240 times. The probability of stroke increases approximately 3 times in patients identified with COVID-19-compatible pneumonia on thoracic CT. The probability of stroke decreases 0.283 times in patients who are COVID-19-negative based on PCR test results (p=0.001, Exp (B)=0.238). The probability of stroke increases by 2.7 times in COVID-19-positive patients identified from PCR test results (table 3 ).Table 3 Binary Logistic Regression analysis for variables that may affect the risk of stroke development in patients Table 3Variables in the Equation B S.E. Wald df Sig. Exp(B) 95% C.I.for EXP(B) Lower Upper Step 2 ABCD2 SCORE (1) -,919 ,346 7,064 1 ,008 ,399 ,203 ,786 CT(1) -1,428 ,496 8,276 1 ,004 ,240 ,091 ,634 PCR(1) -1,264 ,364 12,087 1 ,001 ,283 ,139 ,576 Constant ,767 ,255 9,036 1 ,003 2,153 Based on these findings, the parameters with the highest level of effect on stroke risk in TIA patients were identified as ABCD2 score, PCR test result and the presence of COVID-19-compatible pneumonia on thoracic CT. After determining the variables to be included in the model, first, the effect of the ABCD2 score, estimating the risk of stroke in patients presenting with TIA, was analyzed (Step 1). Then, two variables (PCR test and thoracic CT) that are used to determine the presence of COVID-19 were added to the ABCD2 score to assess their effects on stroke (Step 2). The effect of the score variable on stroke status (Step 1) was found to be significant with a specificity of 87.0%, a sensitivity of 42.8% and a correct classification rate of 72.6%. The Nagelkerke R2 value was 0.217 and the percentage of variation in the response variable (stroke status) explained by the model was 21.7%. The model was then extended with the addition of PCR test results and the presence of COVID-19-compatible pneumonia on thoracic CT next to the ABCD2 score, and this model (Step 2) was found to be significant with a specificity of 87.0%, a sensitivity of 54.8% and a correct classification rate of 77.5%. The Nagelkerke R2 value was 0.331 and the percentage of variation in the response variable (stroke status) explained by the model was 33.1% (Step 2) (table 4 ).Table 4 Model created using parameters found to be effective on the risk of stroke development Table 4Step Specificity Sensitivity Overall Percentage Step 1 87,0 42,8 72,6 Step 2 87,0 54,8 75,5 The Hosmer–Lemeshow test value, which indicates the effectivity of the final model (step 2) in defining the response variable, that is, how well the model fits the data, was found to be significant (p=0.906), which is indicative of a good fit. DISCUSSION Stroke is a medical emergency for which prompt treatment is crucial, as early diagnosis and treatment can prevent permanent brain damage. About a quarter of stroke patients have a TIA prior to the stroke (5), and TIAs are associated with a permanent stroke risk of 5% at 48 hours and 8–12% at 3 months unless specifically treated. All patients presenting to the emergency department with symptoms suggestive of a TIA require extensive research, aggressive treatment and/or hospital admission, and the identification of those who are most likely to benefit from the treatment. The best-known transient ischemic attack scoring tool for triage is the ABCD2 score, and some guidelines recommend that all patients with a suspected TIA should undergo triage in the acute phase using the ABCD2 score (12). In the present study, the risk of stroke in TIA patients increased 1.5 times with increases in the ABCD2 score, three times in the presence of COVID-19-compatible pneumonia on thoracic CT, and 2.7 times with COVID-19 positivity. Based on these findings, we believe that the presence of COVID-19 infection should also be considered when determining the risk of stroke in TIA patients. The model we created by adding COVID-19 positivity and the presence of pneumonia on thoracic CT to the ABCD2 risk score predicted the risk of stroke with a specificity of 87%, a sensitivity of 54.8% and an accuracy of 77.5%. The rate of stroke risk estimated by this model was 11.4% higher than that by the ABCD2 score alone (without adding COVID-19 positivity or thoracic CT findings). Among the clinical features identified in the present study, as one of the parameters assessed when calculating the ABCD2 score, speech disturbance and weakness in the upper extremities were the most common presenting symptoms at the time of admission to the emergency department with TIA. Similarly, the study by Ilstdat et al. of TIA patients reported speech disturbance (aphasia and dysarthria 46.2%) and weakness in the upper extremities (34.1%) to be the most common presenting symptoms (15). Purroy et al. also reported speech disturbance and motor weakness to be the most common presenting symptoms, but underlined that the presenting symptom alone was not a determining factor for the risk of stroke (16). The present study revealed no difference in symptoms between those with and without stroke, although the duration of symptoms, blood pressure and presence of DM, which are among the parameters measured in ABCD2 scoring, were statistically different between those with and without stroke. When the calculated ABCD2 score was assessed based on these parameters, stroke was seen to occurred in 53.26% of the patients with an ABCD2 score of ≥4 in the present study. Similar to our findings, Johnston et al. reported the ABCD2 score to be an independent predictor of 2, 7 and 90-day stroke risk (9). According to our findings, and in line with these studies, these parameters should be considered when assessing TIA patients in the emergency department, and it should also be considered that TIA patients are at risk of SARS-CoV-2 infection during the pandemic. The present study is the first to evaluate the impact of COVID-19 on the risk of stroke in TIA patients. The CNS effects of the SARS-CoV2 virus have yet to be clarified, although it has been suggested that the virus causes neuronal damage by directly passing through the blood-brain barrier through neurotropism and increases the thrombotic tendency, leading to venous and arterial thromboembolism (7, 17, 18). A clinical study reported that cerebrovascular events occurred in the early stages of the disease in all four of their stroke cases and were also positive for COVID-19 (19). Oxley et al. reported a case of stroke due to thrombosis in five large vessels among COVID-19 patients (20), and further studies have suggested that hypoxia due to cardiac and respiratory involvement and the cytokine storm caused by the virus may impair blood flow to the brain, causing ischemia in stroke patients with no vascular pathology who are considered cryptogenic (21, 22, 23). AHA and European Stroke Organization (ESO) guidelines recommend the imaging of the brain-neck arteries (1, 5). Siegler et al. assessed stroke in COVID-19 patients and reported the detection of no extracerebral artery occlusion in 42% of the 156 stroke patients examined (23). Evaluating the presence of COVID-19 in stroke patients, Dhamoon et al. also reported a high rate of stroke of undetermined origin (cryptogenic stroke) among COVID-19-positive patients (22). In a similar study, Topçuoğlu et al. compared 355 COVID-19 negative and 37 COVID-19 positive patients with ischemic stroke, and reported intensive care unit admission, the need for mechanical ventilation and mortality to be significantly higher in COVID-19-positive patients, and that COVID-19 was associated with poor prognosis. In the study, 110 patients were diagnosed with TIA, and of the four who were positive for COVID-19, three were classified as cryptogenic (21). The present study identified no arterial pathology in 32.7% of the COVID-19-positive patients who had a stroke, which leads us to believe that the presence of COVID-19 should be considered when estimating the risk of stroke in TIA patients with normal CT brain-neck angiography. There have been a number of studies reporting cerebrovascular events to be associated with a poorer prognosis in severe COVID-19 cases with pneumonia (18, 22). Görgülü et al. assessed the neurological findings accompanying COVID-19 and reported that 62% of the patients were positive for COVID-19, while 59.5% developed ischemic stroke. The common feature of all patients was the presence of COVID-19-compatible pneumonia (18). Dhamoon et al., on the other hand, reported 38% of 277 stroke patients to be positive for COVID-19, and further, that 67.7% of these patients had COVID-19-compatible pneumonia, with stroke, hemorrhage and TIA being more common in COVID-19-positive patients. The authors also reported a poor prognosis in these patients (22). In the present study, 65.75% of the stroke patients had COVID-19-compatible pneumonia, compared to 69.1% in those with COVID-19-compatible pneumonia in addition to COVID-19 positivity. It can be argued that the presence of COVID-19-compatible pneumonia in addition to COVID-19 positivity further increases the risk of stroke. We believe, thus, that patients who classified as TIA during the pandemic should also be examined for COVID-19 positivity and COVID-19 pneumonia. COVID-19 has been identified by the World Stroke Organization as a risk factor for stroke (24). The first symptoms of COVID-19 can be neurological, and can be the precursor of stroke (19). Studies have reported varying levels of stroke risk in COVID-19 patients. Mao et al. reported that approximately 6% of their COVID-19 patients experienced a stroke (25). Qureshi et al. examined 8,163 COVID-19 in patients from multiple centers in the United States and detected ischemic stroke in 1.3% (26), while study by Marcus et al. reported that 2.8–6% of COVID-19 patients had neurological symptoms that manifested as acute stroke, and most (80%) were ischemic (24). Consistent with these studies, the present study identified the ABCD2 score, the presence of COVID-19 (PCR) and the presence of COVID-19-compatible pneumonia (on thoracic CT finding) as the primary factors affecting the risk of stroke in patients with TIA. CONCLUSION The inclusion of COVID-19 and COVID-19 pneumonia to the ABCD2 score, which is used for the prediction of risk in TIA patients, based on information about the increased risk of stroke in TIA patients, improves the predictive power of the tool. We believe that PCR test results and the presence of COVID-19 pneumonia should be considered in the stroke risk assessment of TIA patients, although more comprehensive studies are needed in this regard. Conflicts of interest Each author declares that he or she has no commercial associations (e.g. consultancies, stock ownership, equity interest, patent/licensing arrangement, etc.) that may be considered a conflict of interest in connection with the submitted article. Acknowledgment We would like to thank emergency medicine specialists Prof. Dr. Fatih Esad Topal and Prof. Dr. Cemil Kavalcı for their contributions to our study, Conflict of Interest statement: The authors had no conflicts of interest to declare in relation to this article. ==== Refs REFERENCES 1 AHA GUİDE Dawn O. Kleindorfer, MD, FAHA, Chair; AmytisTowfighi, MD, FAHA, Vice Chair; Seemant Chaturvedi et al. 2021 Guideline for the Prevention of Stroke in Patients With Stroke and Transient Ischemic Attack. 2 Sacco RL Kasner SE Broderick JP An updated definition of stroke for the 21st century: a statement for healthcare professionals from the American Heart Association/American Stroke Association [published correction appears in Stroke. 2019 Aug;50(8):e239] Stroke 44 7 2013 2064 2089 23652265 3 Kleindorfer D Panagos P Pancioli A Incidence and short- term prognosis of transient ischemic attack in a population-based study Stroke 36 4 2005 720 723 15731465 4 Wu CM McLaughlin K Lorenzetti DL Early risk of stroke after transient ischemic attack: a systematic review and meta-analysis Arch Intern Med 167 22 2007 2417 2422 18071162 5 Fonseca AC Merwick Á Dennis M European Stroke Organisation (ESO) guidelines on management of transient ischaemic attack [retracted in: Eur Stroke J. 2022 Mar;7(1):NP1] Eur Stroke J 6 2 2021 V 6 Gómez-Mesa JE Galindo-Coral S Montes MC Thrombosis and Coagulopathy in COVID-19 Curr Probl Cardiol 46 3 2021 100742 7 Abboud H Abboud FZ Kharbouch H COVID-19 and SARS-Cov-2 Infection: Pathophysiology and Clinical Effects on the Nervous System World Neurosurg 140 2020 49 53 10.1016/j.wneu.2020.05.193 32474093 8 Baig AM Khaleeq A Ali U Evidence of the COVID-19 Virus Targeting the CNS: Tissue Distribution, Host-Virus Interaction, and Proposed Neurotropic Mechanisms ACS Chem Neurosci 11 7 2020 995 998 32167747 9 Johnston SC Rothwell PM Nguyen-Huynh MN Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack Lancet 369 9558 2007 283 292 17258668 10 Sanders LM Srikanth VK Blacker DJ Jolley DJ Cooper KA Phan TG. Performance of the ABCD2 score for stroke risk post TIA: meta-analysis and probability modeling Neurology 79 10 2012 971 980 22700810 11 Tsivgoulis G Stamboulis E Sharma VK Multicenter external validation of the ABCD2 score in triaging TIA patients Neurology 74 17 2010 1351 1357 20421579 12 Meng Xia Wang Yilong Liu Liping Validation of the ABCD2-I score to predict stroke risk after transient ischemic attack Neurological Research 33 5 2011 482 486 21669116 13 Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia (Trial Version 7) Chin Med J (Engl) 133 9 2020 1087 1095 32358325 14 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 4 2020 219 227 32324653 15 Ildstad F Ellekjær H Wethal T ABCD3-I and ABCD2 Scores in a TIA Population with Low Stroke Risk Stroke Res Treat 2021 2021 8845898 10.1155/2021/8845898 Published 2021 Feb 25 16 Purroy Francisco Begue´ Robert Quílez Alejandro The California, ABCD, and Unified ABCD2 Risk Scores and the Presence of Acute Ischemic Lesions on Diffusion-Weighted Imaging in TIA Patients Stroke 40 2009 2229 2232 19372450 17 Mahalakshmi AM Ray B Tuladhar S Does COVID-19 contribute to development of neurological disease? Immun Inflamm Dis 9 1 2021 48 58 33332737 18 Gorgulu U Bayındır H Bektas H Coexistence of neurological diseases with Covid-19 pneumonia during the pandemic period J Clin Neurosci 91 2021 237 242 34373034 19 Avula A Nalleballe K Narula N COVID-19 presenting as stroke Brain Behav Immun 87 2020 115 119 32360439 20 Oxley TJ Mocco J Majidi S Large-Vessel Stroke as a Presenting Feature of Covid-19 in the Young N Engl J Med 382 20 2020 e60 32343504 21 Topcuoglu MA Pektezel MY Oge DD Stroke Mechanism in COVID-19 Infection: A Prospective Case-Control Study J Stroke Cerebrovasc Dis 30 8 2021 105919 22 Mandip S. Dhamoon, Alison Thaler, Kapil Gururangan, et al. Acute Cerebrovascular Events With COVID-19 Infection Stroke. 2021;52:48–56. 23 Siegler JE Cardona P Arenillas JF Cerebrovascular events and outcomes in hospitalized patients with COVID-19: The SVIN COVID-19 Multinational Registry International Journal of Stroke 16 4 2021 437 447 32852257 24 Markus HS Brainin M. COVID-19 and stroke-A global World Stroke Organization perspective Int J Stroke 15 4 2020 361 364 32310017 25 Mao L Jin H Wang M Neurologic Manifestations of Hospitalized Patients With Coronavirus Disease 2019 in Wuhan, China JAMA Neurol 77 6 2020 683 690 32275288 26 Qureshi AI Baskett WI Huang W Acute Ischemic Stroke and COVID-19: An Analysis of 27 676 Patients Stroke. 52 3 2021 905 912 33535779	0	PMC9715486	NO-CC CODE	2022-12-03 23:20:16	no	J Stroke Cerebrovasc Dis. 2022 Dec 2;:106918	utf-8	J Stroke Cerebrovasc Dis	2,022	10.1016/j.jstrokecerebrovasdis.2022.106918	oa_other
==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 The Authors. Published by Elsevier Ltd. S0264-410X(22)01489-X 10.1016/j.vaccine.2022.11.065 Short Communication Effectiveness of vaccines in preventing hospitalization due to COVID-19: A multicenter hospital-based case-control study, Germany, June 2021 to January 2022 Stoliaroff-Pepin Anna a1⁎ Peine Caroline a1⁎ Herath Tim a Lachmann Johannes a Perriat Delphine a Dörre Achim b Nitsche Andreas c Michel Janine c Grossegesse Marica c Hofmann Natalie c Rinner Thomas c Kohl Claudia c Brinkmann Annika c Meyer Tanja d Dorner Brigitte G. d Stern Daniel d Treindl Fridolin d Hein Sascha e Werel Laura e Hildt Eberhard e Gläser Sven f Schühlen Helmut g Isner Caroline h Peric Alexander i Ghouzi Ammar j Reichardt Annette k Janneck Matthias l Lock Guntram m Schaade Lars n Wichmann Ole a Harder Thomas a a Department for Infectious Disease Epidemiology, Immunization Unit, Robert Koch Institute b Department for Infectious Disease Epidemiology, Robert Koch Institute c Centre for Biological Threats and Special Pathogens, ZBS1 Highly Pathogenic Viruses, Robert Koch Institute d Centre for Biological Threats and Special Pathogens, ZBS3 Biological Toxins, Robert Koch Institute e Division Virology, Paul-Ehrlich-Institute f Klinik für Innere Medizin - Pneumologie und Infektiologie, Vivantes Klinikum Neukölln und Spandau, Berlin g Vivantes Netzwerk für Gesundheit GmbH, Direktorat Klinische Forschung & Akademische Lehre, Berlin h Klinik für Innere Medizin - Infektiologie, Vivantes Auguste-Viktoria-Klinikum, Rubensstr. 125, 12157, Berlin i Klinik für Pneumologie und Infektiologie, Vivantes Klinikum im Friedrichshain, Landsberger Allee 49, 10249 Berlin j Schön Klinik Düsseldorf, Interdisziplinäre Notaufnahme, Am Heerdter Krankenhaus 2, 40549 Düsseldorf k Helios Klinikum Berlin-Buch, Schwanebecker Chaussee 50, 13125 Berlin l Klinik für Kardiologie, Sektion Nephrologie, Albertinen Krankenhaus, Süntelstraße 11a, 22457 Hamburg m Klinik für Innere Medizin, Albertinen Krankenhaus, Süntelstraße 11a, 22457 Hamburg n Centre for Biological Threats and Special Pathogens, Robert Koch Institute ⁎ Corresponding authors. 1 equal contribution. 2 12 2022 2 12 2022 10 8 2022 16 11 2022 27 11 2022 © 2022 The Authors. 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. We included 852 patients in a prospectively recruiting multicenter matched case-control study in Germany to assess vaccine effectiveness (VE) in preventing COVID-19-associated hospitalization during the Delta-variant dominance. The two-dose VE was 89% (95% CI 84-93%) overall, 79% in patients with more than two comorbidities and 77% in adults aged 60-75 years. A third dose increased the VE to more than 93% in all patient-subgroups. Keywords COVID-19 vaccine effectiveness hospitalization case-control study SARS-CoV-2 Delta ==== Body pmc1 Introduction 1.1 Background In spring 2021, a new variant of concern (VOC), the highly contagious SARS-CoV-2 Delta variant, was identified and gained global dominance by summer 2021. As of August 2021, the weekly incidence in Germany steadily increased, reaching more than 300,000 newly infected persons per week by November 2021. A combination of non-pharmaceutical interventions and a booster vaccination campaign (three vaccine doses) led to decreasing case numbers until the Omicron variant of SARS-CoV-2 took over in 2022 [1]. High efficacy of COVID-19 vaccines has been demonstrated in randomized, placebo-controlled clinical trials (73% for vector vaccines and 85% for mRNA vaccines in meta-analyses) [2], yet real world data may differ depending on population characteristics and the dominant viral strain during the observation period. In addition, heterologous vaccination schemes i.e. using a vector vaccine for the first and an mRNA vaccine for the second injection, or vice versa, commonly occurred. For booster vaccination, only mRNA vaccines (Comirnaty®, Spikevax®) were recommended in Germany. Furthermore, there is limited data on the duration of protection under the Delta-variant [3], [4], [5]. Early in 2021, we set up a study in 13 hospitals across Germany to analyze the effectiveness of the COVID-19 vaccines in Germany in preventing COVID-19 associated hospitalization in the adult population. In addition, we aimed to perform subgroup analyses for vaccine effectiveness (VE) by age, sex, severity of disease, underlying comorbidities and time since vaccination. Here we present VE results assessed during the Delta-variant wave. 2 Methods COViK is a prospectively recruiting matched case-control study in Germany that is led by Germany's Public Health Institute (Robert Koch Institute, RKI). Study sites are 13 hospitals in Berlin, Hamburg, Wuppertal, Duesseldorf, Erfurt and Chemnitz. The study was commenced on 1 June 2021 and is expected to end in June 2023. We here present an interim analysis based on the data of patients recruited from 1 June 2021 until 31 January 2022. A case was defined as an adult aged 18 to 90 years who had a positive SARS-CoV-2 PCR test and was hospitalized in one of the 13 study hospitals. To be included, cases had to be either hospitalized due to a severe COVID-19 infection or COVID-19 complications, or with a severe nosocomial COVID-19 infection (see supplementary material). Controls were hospitalized patients who were tested negative for SARS-CoV-2 by PCR and were recruited at surgical, orthopaedical, urological and gynecological wards, preferably with acute diseases (e.g. fracture, tendon rupture). Two controls per case were recruited from the same hospital that the respective case was admitted to, or, if not available, from a hospital in the same city (1:2 matching). Controls were matched to cases based on their hospital admission date (+/- 14 days), age (+/- 10 years) and sex. All study nurses were trained by the COVIK study centre team regularly. Repeated on-site visits and quality controls were performed to ensure adherence to the study protocol. Cases and controls with a previous SARS-CoV-2 infection were excluded from the analysis to avoid misclassification (e.g. risk of infection, indication for vaccination; Supplementary Material Table 1). A current SARS-CoV-2 infection was identified through a positive result of a SARS-CoV-2-real-time PCR performed on a naso-oropharyngeal swab [6]. The virus variant was determined by sequencing using the AmpliCoV protocol [7], or by PCR typing assays if RNA load was too low for sequencing. In addition, antibodies against SARS-CoV-2 and other coronaviruses were determined and virus-neutralization tests were performed for future immunological analyses. Data on basic socio-demographic factors (e.g. age, sex, level of graduation), COVID-19 disease (e.g. previous infection, date of symptom onset), COVID-19-vaccination (e.g. vaccination status, vaccine type and number of vaccine doses administered), risk factors for COVID-19 infection and for severe course of disease (e.g. comorbidities) were collected in individual interviews conducted by research nurses. Clinical and laboratory data were extracted from medical records (e.g. sequencing results, admission to an intensive care unit (ICU)). All participants (or their legal guardians) provided a written informed consent to participate in the study. For this first pre-specified interim analysis, we computed the 2-dose and 3-dose VE regardless of the individual matching for the following subgroups: males and females, aged 18-59, 60-75, and 76-90 years, with <3 or ≥ 3 pre-existing comorbidities, last vaccine dose administered in the past 3 months or 3-6 months ago, admitted to intensive care or not (Figure 1 , Supplementary Table 3). Patients vaccinated only once were excluded from the analysis of vaccine effectiveness. The analysis was restricted to cases infected with the Alpha or Delta variant.Figure 1 Vaccine effectiveness for two and three vaccination doses, endpoint severe COVID-19 (hospitalization). Vaccine effectiveness for all participants and subgroups is shown. Calculation of VE was performed as described in the Supplementary Methods. aICU treated cases versus controls with no ICU treatment Based on logistic regression modeling, we determined VE, adjusted for age, education and pre-existing comorbidities. Variable selection was performed on the basis of predictive accuracy, the AIC and context-related information. Confounder variables taken into consideration were age, age group, comorbidities (general, of the immune system, with high risk for severe course of disease), educational level, sex, region, time period, profession, housing situation, contact to COVID-19 positive persons, activities without mask, self-estimated probability of infection, close contact to persons with high risk for severe course of disease and compliance with infection protection measures; see Supplementary Material). Comparison of the case and control groups in Supplementary Table 2, was performed with appropriate significance tests (t-tests for continuous variables and chi-squared tests or Fisher's exact test for categorical variables; see Supplementary Material). Data were analyzed using the statistical software R, version 4.1.2. The study was approved by the Ethics Committee of the Charité Universitätsmedizin, Berlin (EA1/063/21) and was registered at “Deutsches Register Klinischer Studien” (DRKS00025004). 3 Results During the study period, 852 participants were recruited, including 244 cases and 608 matched controls. Median age of cases was 57 years (interquartile range 45-70 years), median age of controls was 59 years (interquartile range 48-72 years), 45 % of cases and 43% of the controls were female. Cases had a lower education status than controls (42% of cases vs 28% of controls with nine or less school years, (p< 0.05)). The majority of the cases was infected with the Delta variant (79%), and 5% of them were infected with the Alpha variant (missing information 16%). Nearly a quarter of cases (n=56, 23%) was admitted to the ICU and 13 patients (5,3%) deceased (Supplementary Table 2). Cases were significantly less often vaccinated against COVID-19 than controls: More than half of the cases (58%, n=142/244) were not vaccinated at all, compared to 11% of the controls. Nearly a third (30%, 73/244) of the cases and more than half of the controls (53%, 323/608) had received two vaccine doses (first vaccination series) and 9/244 cases (4%) and 192/316 controls (32%) had received a third dose (booster dose). Additionally, three controls had received a fourth vaccine dose. The most frequently administered vaccine among study participants was Comirnaty® (Biontech/Pfizer), 1119 doses, followed by Vaxzevria® (Astra-Zeneca, 151 doses) and Spikevax® (Moderna, 141 doses), while Jcovden® (Johnson& Johnson/Janssen) was administered 28 times. Without further adjustment, the two-dose VE was 89% (95% CI 84-93%) overall and a third dose increased the VE to more than 93% in all patient-subgroups (Figure 1). After adjustment for age, education and pre-existing comorbidities, overall VE (all groups) was 93.5% (CI 89.1 – 96.2%) after two vaccine doses and 99.4% (CI 98.1-99.9%) for three doses. VE after two vaccine doses was significantly lower for adults with three or more pre-existing comorbidities as compared to adults with less than three comorbidities (95.7% vs 78.7%), while VE after three vaccine doses was similar for both groups (Figure 1, Supplementary Table 3). The VE was lower among adults aged 60-75 years but this reduction was again compensated when a third dose was administered. When the individual matching of the pairs was considered, similar results were obtained (Supplementary Table 4). 4 Discussion To our knowledge, this is the first study in Germany that assessed VE of COVID-19 vaccines. We decided to use a case control design instead of test negative design because we assumed that recruiting controls with respiratory symptoms would be difficult due to strict infection protection measures (e.g. lock down) and these controls might not be representative, e.g. bedridden patients with endogenous pneumonia due to microaspiration. Furthermore, the advantage of avoiding selection bias for health care seeking, that is important in influenza VE calculations with outpatients, is not as relevant in our study, as the endpoint (hospital admission) is less likely influenced by health care seeking behavior. Our choice of study design is supported by a recent study showing that syndrome-negative and test-negative controls result in comparable VE [8]. Our analysis suggests that the COVID-19-vaccines licensed in the European Union were highly effective in preventing hospitalization due to COVID-19. The vaccine effectiveness was 93.5% after two vaccine doses and 99.4% after three doses. Recently, studies from Canada, the UK and others have reported comparably high VE after two doses against hospitalization [4] with a slow decrease in protection against hospitalization over time. Importantly, waning of protection especially affected clinically vulnerable groups [9]. We found that vaccination protected from severe disease for at least six months and the moderately reduced two-dose protection after three to six months (90%) raised again to 98% after a booster shot, emphasizing the necessity of the third vaccine dose. Patients between 60 and 75 years of age had a significantly reduced two-dose effectiveness. This group benefitted in particular from a third dose. The reduced VE in this group may be due to a weaker immune response upon vaccination compared with younger people but could be also a chance finding as patients older than 75 years showed a higher VE in comparison. Furthermore the comorbidities in the control group (usually surgery) might be documented less carefully compared to the case group (internal medicine). Almost a quarter of our study participants was admitted to an ICU. However, this number may underestimate the ratio of very severe cases as some patients, e.g. elderly people, may refuse ICU treatment. A main advantage of our study design stems from the ability to collect detailed high-quality information in a prospective manner. Every COVID-19 diagnosis was confirmed by clinical records and -if necessary- by direct consultation of the attending physician. Only patients requiring hospitalization due to COVID-19 were included. Many post-deployment studies rely on clinical data registries, resulting in fast reporting which is valuable to guide policy decisions during a pandemic. However, not all patients are included in such registries and their main diagnosis may not be COVID-19. Unlike many other studies, we explicitly examined the pre-existing immunity by serology early in the course of infection and excluded all participants with pre-existing antibodies or with a previous laboratory-confirmed SARS-CoV-2 infection. The detailed verification leads to a better data quality: With relying on patient’s information only we would have excluded six patients due to previous SARS-CoV-2 infection, the antibody test revealed 24 further cases. Participants with a history of SARS-CoV-2 infection were excluded from the analysis to avoid a biased analysis as previous infections potentially prevent subsequent infections and vaccination was not recommended at least for three months after infection after SARS-CoV-2 infection in Germany. Furthermore, we determined the virus variant for each patient. A limitation of our study is the low number of participants. We could not gain precise results for matched pairs and triplet analysis for some subgroups for this reason. A potential source of bias might be a higher likeliness of study participation for vaccinated patients. Here, we took countermeasures such as incentivation and we recorded the response rates of cases and controls and reasons for non-participation. No bias was visible when comparing cases and controls. We will continue to carefully assess this as unvaccinated persons may form a special group in settings of high vaccination coverage. This first interim analysis provides encouraging results and warrants follow-up analyses to assess the evolving COVID-19 vaccine effectiveness, in a changing epidemiologic landscape in terms of circulating variants, available vaccines, and increasing population-wide immunity. In the future course of the study, we plan to analyze the Omicron wave, combinations of natural infection and vaccination, longer time intervals since vaccination, long-term COVID-19 symptoms in vaccinated vs unvaccinated individuals (long COVID), and the immune response after COVID-19 breakthrough infections. 5 Conclusion The COVID-19 vaccines were highly protective against hospitalization in real-world settings in Germany during the Delta-variant predominance. Reduced vaccine effectiveness observed in subgroups after two doses was compensated after three doses. This finding would support efforts to maximize vaccine uptake to three doses among vulnerable populations. Potential conflicts of interest S. G. received payment/honoraria from Astra Zeneca, Boehringer Ingelheim, Roche Pharma and Berlin Chemie, this had no influence on this work; all other authors reported no conflicts of interest. Funding This work awas supported by the German Federal Ministry of Health [2520COR416]. S. G. received payment/honoraria from Astra Zeneca, Boehringer Ingelheim, Roche Pharma and Berlin Chemie, this had no influence on this work; all other authors reported no conflicts of interest. Meetings where this study has been previously presented Joint Annual Meeting DZIF/DGI 2022, June 1-3, 2022, Stuttgart, Germany Nationale Impfkonferenz, June 14-15, 2022, Wiesbaden, Germany 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. Data availability Data will be made available on request. Acknowledgments The authors thank all study nurses for the valuable contribution, namely Sawsanh Al-Ogaidi, Nancy Beetz, Belgin Esen, Rola Khalife, Marie-Kristin Kusnierz, Katja Lange, Luise Mauer, Antje Micheel, Marlies Schmidt, Yvonne Weis, Franziska Weiser, Aysete Yencilek. We thank Wiebke Hellenbrand for her important contribution in designing and setting up the study as well as Anna Meier, Swetlana Muminow, Richard Schensar, Ellen Busch, Hanna Buck, and Moritz Gehring for their support in organizing the project and Vincent Stoliaroff-Pépin for support with R. ==== Refs References 1 Robert-Koch-Institute. Wöchentlicher Lagebericht des RKI zur Coronavirus-Krankheit-2019 (COVID-19). 2022; Available from: https://www.rki.de/DE/Content/InfAZ/N/Neuartiges_Coronavirus/Situationsberichte/Wochenbericht/Wochenberichte_Tab.html. 2 Sharif N. Efficacy, Immunogenicity and Safety of COVID-19 Vaccines: A Systematic Review and Meta-Analysis Front Immunol 12 2021 714170 3 Skowronski D.M. Two-dose SARS-CoV-2 vaccine effectiveness with mixed schedules and extended dosing intervals: test-negative design studies from British Columbia and Quebec Canada. Clin Infect Dis 2022 4 Feikin D.R. Duration of effectiveness of vaccines against SARS-CoV-2 infection and COVID-19 disease: results of a systematic review and meta-regression Lancet 399 10328 2022 924 944 35202601 5 Mahumud R.A. Effectiveness of COVID-19 Vaccines against Delta Variant (B.1.617.2): A Meta-Analysis Vaccines (Basel) 10 2 2022 6 Michel J. Resource-efficient internally controlled in-house real-time PCR detection of SARS-CoV-2 Virol J 18 1 2021 110 34078394 7 Brinkmann A. AmpliCoV: Rapid Whole-Genome Sequencing Using Multiplex PCR Amplification and Real-Time Oxford Nanopore MinION Sequencing Enables Rapid Variant Identification of SARS-CoV-2 Front Microbiol 12 2021 651151 8 Turbyfill C. Comparison of test-negative and syndrome-negative controls in SARS-CoV-2 vaccine effectiveness evaluations for preventing COVID-19 hospitalizations in the United States Vaccine 40 48 2022 6979 6986 36374708 9 Andrews N. Duration of Protection against Mild and Severe Disease by Covid-19 Vaccines N Engl J Med 386 4 2022 340 350 35021002	36509640	PMC9715487	NO-CC CODE	2022-12-09 23:15:06	no	Vaccine. 2022 Dec 2; doi: 10.1016/j.vaccine.2022.11.065	utf-8	Vaccine	2,022	10.1016/j.vaccine.2022.11.065	oa_other
==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 Published by Elsevier Ltd. S0264-410X(22)01496-7 10.1016/j.vaccine.2022.11.072 Article Exploring key informants’ perceptions of Covid-19 vaccine hesitancy in a disadvantaged urban community in Ireland: Emergence of a ‘4Cs’ model Ingram Carolyn a⁎ Roe Mark a Downey Vicky a Phipps Lauren b Perrotta Carla a a School of Public Health, Physiotherapy, and Sports Science, University College Dublin, D04 V1W8 Dublin, Ireland b College of Science, University College Dublin, D04 V1W8 Dublin, Ireland ⁎ Corresponding author. 2 12 2022 2 12 2022 21 1 2022 4 11 2022 29 11 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. Aim The aim of this study was to explore key informants’ views on and experiences with Covid-19 vaccine hesitancy in a Dublin community with a high concentration of economic and social disadvantage and to identify feasible, community-centred solutions for improving vaccination acceptance and uptake. Methods Qualitative, semi-structured interviews were carried out at a local community-centre and a central hair salon. Twelve key informants from the target community were selected based on their professional experience with vulnerable population groups: the unemployed, adults in recovery from addiction, the elderly, and Irish Travellers. Inductive thematic framework analysis was conducted to identify emergent themes and sub-themes. Results Drivers of vaccine hesitancy identified by key informants largely fell under the WHO ‘3Cs’ model of hesitancy: lack of confidence in the vaccine and its providers, complacency towards the health risks of Covid-19, and inconvenient access conditions. Covid-19 Communications emerged as a fourth ‘C’ whereby unclear and negative messages, confusing public health measures, and unmet expectations of the vaccine’s effectiveness exacerbated anti-authority sentiments and vaccine scepticism during the pandemic. Community-specific solutions involve the provision of accurate and accessible information, collaborating with community-based organizations to build trust in the vaccine through relationship building and ongoing dialogue, and ensuring acceptable access conditions. Conclusions The proposed Confidence, Complacency, Convenience, Covid-19 Communications (‘4Cs’) model provides a tool for considering vaccine hesitancy in disadvantaged urban communities reacting to the rapid development and distribution of a novel vaccine. The model and in-depth key informants’ perspectives can be used to compliment equitable vaccination efforts currently underway by public health agencies and non-governmental organizations. Keywords Covid-19 Vaccine hesitancy Hesitancy drivers Qualitative Low-SES community Socioeconomic disadvantage ==== Body pmc1 Introduction Vaccine hesitancy refers to delay in acceptance or refusal of vaccination despite availability of vaccination services. It is complex and context specific, varies across time, place and vaccines, and is influenced by factors such as complacency, convenience and confidence [1]. Though not a new phenomenon [2], a variety of factors have fuelled an increase in Covid-19 vaccine hesitancy specifically, including concerns about rapid production and use of new messenger RNA (mRNA) technology and potential side effects [3]; the propagation of misinformation and disinformation on social media [4]; disapproval of Covid-19 mitigation measures implemented by the government [5]; and confusion about natural immunity, vaccine effectiveness and the need for repeat vaccination [6], [7], [8]. Resulting hesitancy or refusal to be vaccinated can and does have dire consequences. Covid-19 mortality over a 2-year period is up to eight times higher in populations with high vaccine hesitancy compared to those with an ideal vaccination uptake [9]. An individual who is fully vaccinated with a messenger RNA (mRNA) vaccine is up to 90 % less likely to get infected [10], 94 % less likely to be hospitalized [11], and 90 % less likely to die from Covid-19 [12]. Concentrated disadvantage – the phenomenon of spatial clustering of economically or socially disadvantaged individuals within a set of neighbourhoods and the resulting feedback effects that exacerbate the problems of poverty and poor health [13] – is associated with vaccine hesitancy [14]. Individual predisposing and behavioural factors intersect with place-based economic, social and health inequalities and hinder vaccination willingness and uptake [15], [16], resulting in that the communities most vulnerable to Covid-19 are more likely to have low vaccination coverage [14], [15]. Vaccine hesitancy may be particularly consequential in urban settings where factors such as overcrowded housing, large numbers of essential workers, and exposure to air pollution increase residents’ risk of Covid-19 infection and severe outcomes [17], [18], [19], [20]. Experts have called for the prioritisation of densely populated deprived areas during Covid-19 vaccination rollout [21]. However, ensuring equitable vaccination first requires understanding and addressing challenges associated with vaccination willingness and uptake in disadvantaged urban communities [22]. In the Republic of Ireland, 91 % of the eligible population were fully vaccinated for Covid-19 as of January 2022, and 56 % had received a booster vaccine [23]. Nevertheless, an estimated third of the adult population had experienced some Covid-19 vaccine hesitancy, and 9 % were opposed to the vaccine, with trends in resistance steadily increasing as the pandemic progressed [24], [25]. As reported internationally, adults in Ireland who are vaccine hesitant are more likely to live in urban settings and be in a lower income bracket [24], the same areas at increased risk for Covid-19 incidence and comorbidities [21]. Interventions to address vaccine hesitancy are most successful if they are based on empirical data and situational assessment, and adapted to a specific target group in a culturally sensitive manner [26]. Preliminary findings on the drivers of Covid-19 vaccine hesitancy in Ireland show that national trends match those found internationally [24], [26]: vaccine hesitant individuals in Ireland are more likely to have conspiracy beliefs; lower levels of trust in scientists, health care professionals, and the state; and to consume significantly less information from formal information sources and more from social media [24]. Nevertheless, at the onset of the national Covid-19 booster campaign, lower reported vaccination coverage in Irish urban centres [27] highlighted a need to consult disadvantaged urban community residents on local drivers of vaccine hesitancy in order to tailor the design and rollout of interventions to those disproportionately at risk of declining a highly effective vaccine [28]. The aim of this study was to explore key informants’ views on and experiences with Covid-19 vaccine hesitancy in a Dublin community with a high concentration of economic and social disadvantage and to identify feasible, community-centred solutions for improving vaccination acceptance and uptake. Key informants were individuals from the target community working with population groups identified as disproportionately vulnerable to Covid-19: the socioeconomically disadvantaged and/or unemployed, adults in recovery from addiction, the elderly, and Irish Travellers, an indigenous ethnic minority group [29]. 2 Methods 2.1 Defining the target community We used an area-based approach to identify a densely populated, disadvantaged Dublin community with low relative rates of vaccination (Fig. 1 ). In Ireland, area deprivation is measured by the 2016 Pobal HP Deprivation Index which considers three dimensions of affluence/disadvantage: demographic profile (e.g., population loss, social and demographic effects of emigration), social class composition (e.g., social class composition, education, housing quality), and labour market situation (e.g., unemployment, lone parents, low skills base) [30]. Each Small Area (SA) – the smallest spatial units for which population data is available in Ireland (∼100 households) – receives a Relative Index Score ranging from ‘Very Affluent’ to ‘Very Disadvantaged’. Our target community was a cluster of eight ‘Disadvantaged’ SAs with below average rates of vaccination as of autumn 2021 constituting a population of approximately 30,000. The eight SAs, hereinafter referred to as the target community, are linked by a central community partnership organization charged with implementing government-funded Social Inclusion and Community Activation Programmes [31]. Our research team met with representatives from this organization in September 2021 who confirmed lingering hesitancy towards the Covid-19 vaccine in the target community and the need to better understand and address its driversFig. 1 Comparison of Pobal HP Deprivation Index 2016 rankings and Covid-19 Vaccination Rates by national decile as of March 2022 in Dublin, Ireland. Note: The target community is a cluster of eight ‘Disadvantaged’ Small Areas (SA) within a Local Electoral Area (LEA) in the lowest decile for national vaccination rates. In Ireland, Covid-19 vaccination rates are not publicly available in spatial units smaller than LEAs. The target community is also home to a high concentration of population subgroups experiencing social disadvantage as a result of a particular theme or issue which is common between them (e.g., Irish Travellers, low-income workers, the unemployed, adults in recovery from addiction). We sought to explore the barriers to vaccination specific to diverse groups served by the community partnership organization, as well as how place-based vulnerabilities shared across the target community (i.e., poverty, hardship, and social exclusion) related to vaccine hesitancy. 2.2 Study design and setting Qualitative, semi-structured interviews were carried out in collaboration with (1) the community partnership organization that addresses long-term unemployment and poverty through education and social inclusion initiatives, and (2) a long-standing hair salon located on the target community’s main road. These specific collaborations were formed to enhance understanding of vaccine hesitancy through community-based organizational representatives’ knowledge of local social and cultural dynamics [32], and the fact that clients regularly disclose information about health and identity to hairdressers [33]. A semi-structured interview format was selected to allow participants to freely express themselves while providing reliable, comparable data [34]. The study’s qualitative methodology was preregistered on the Open Science Framework (https://osf.io/n5jch). 2.3 Sample and recruitment A non-probability purposive sampling method was used to identify potential key informants from the two partnering organizations that met the following criteria: knowledge of and experience with community vaccine hesitancy based on their professional role and having lived and/or grown up in the target community; ability to communicate that knowledge to the researchers; and willingness to take part in the study [35], [36]. To ensure diversity of opinion, key informants of various ages, genders, and occupational roles were selected. Prior to participant recruitment, the researchers met with managers of the respective establishments to introduce themselves and explain the study objectives. After agreeing to the research collaboration, managers identified and invited eligible staff members to participate in semi-structured interviews. A first round of twelve in-depth interviews were scheduled and completed, at which point the researchers found that no new information relevant to the study objectives was emerging and data saturation was achieved [37]. 2.4 Data collection Semi-structured interviews were conducted at the hair salon on 17 September 2021, and at the community partnership on 19 November 2021, in private rooms provided by management. Interviewers had formal qualitative research training (CI, MR) and were accompanied by an undergraduate research intern from the local community acting as a liaison between the researchers and collaborating organizations (LP). Participants were presented with an information sheet and gave informed oral consent in the full knowledge that interviews would be audio recorded, transcribed, and anonymized as approved by the researchers’ institutional Human Research Ethics Committee (UCD-HREC-LS-E-21–222). Interviews lasted around 20 min and followed a set of research questions developed through consultation with community representatives and review of existing literature on Covid-19 vaccine hesitancy in Ireland [24], [29]. Both individual and community perspectives were sought on (1) attitudes towards the vaccine, (2) thoughts on the current pandemic situation given widespread first-round vaccinations, and (3) attitudes towards the booster (Fig. 2 ).Fig. 2 Covid-19 vaccine topics included in the semi-structured interview guide. Note: key informants were asked about their own perceptions and experiences as well as those of the wider community. 2.5 Data analysis Data were coded and analysed using an inductive thematic framework method according to the following recommended stages of trustworthy, thematic analysis [38], [39]:• Transcription: Audio recordings were transcribed verbatim by two researchers (CI, VD). • Familiarisation: Two researchers (CI,VD) familiarized themselves with the data by re-listening to audio recordings and re-reading the transcripts. Each researcher recorded analytical notes, thoughts, and impressions in the transcript margins. • Initial coding: The same researchers independently coded three transcripts line by line, identifying potential themes and subthemes relating to vaccine hesitancy through an ‘open coding’ process. Once results were compared and an initial coding framework constructed, CI completed line by line open coding of the remaining nine transcripts. This allowed for further revision before the research team met to discuss, refine, and agree to a working thematic framework. During peer debriefing, researchers recognized that key themes largely fell under the World Health Organization’s (WHO) ‘3Cs’ model of vaccine hesitancy [26]. Thus, determinants of hesitancy were re-categorized under Confidence, Complacency, Convenience and – unique from the WHO model – Covid-19 Communications and Community-Centred Solutions, as defined in Fig. 3 .Fig. 3 ‘4Cs’ Model of Covid-19 Vaccine Hesitancy in a disadvantaged urban community: inductive analysis results from key informant interviews, Dublin, Ireland 2021. From the WHO SAGE Working Group 3Cs model of Vaccine Hesitancy [26]. Covid-19 Communications emerged through inductive analysis as a separate theme driving hesitancy. • Applying the thematic framework: The working thematic framework was systematically applied to all transcripts by CI using Nvivo software V.11. Overlapping themes were combined, and necessary refinements made until three layers of distinct themes were finalized and approved by all researchers. • Charting and interpreting the data: A matrix was used to summarize data for each participant, code, and theme. Connections within and between codes and cases were made in order to fulfil the original research objectives and highlight findings generated through inductive analysis. 3 Results 3.1 Key informant demographics and personal views on the vaccine The study sample (n = 12) was made up of three key informants from a central hair salon and nine from a local community centre including: two guidance counsellors for adults in recovery from addiction, three working with job seekers and/or those on social welfare, one community health officer working with the local Irish Traveller population, two administrators, and one centre manager. Key informants’ characteristics and their personal stances on Covid-19 vaccination and booster shots are described in Table 1 . Five key informants were completely accepting of Covid-19 vaccines, citing motivators such as personal safety and that of loved ones, returning to normalcy and going on holidays, and being adequately informed. Of the seven key informants who experienced little to great vaccine hesitancy, fear of side effects, especially in the case of underlying health conditions, and guilt at receiving the vaccine before those more vulnerable were the most-cited barriers. Only two key informants were entirely accepting of the booster shot. Those who were resistant (n = 5) were largely discouraged by unmet expectations of vaccine efficacy.Table 1 Key informant characteristics and stances on the Covid-19 vaccine and booster shot: Dublin, Ireland, 2021. Key Informant ID Key Informant Employment Age Gender Vaccine Received Stance on Vaccine Explanation Stance on Booster Explanation Hair Salon 1 Hairdresser in a centrally located salon 30–39 F Pfizer No hesitancy F: Watching other, older people go first without bad side effects. N/A Hair Salon 2 Hairdresser in a centrally located salon 50–59 F Pfizer No hesitancy F: Wanted to get back to work. Everybody else was doing it. N/A Hair Salon 3 Hairdresser in a centrally located salon 50–59 F Pfizer Some hesitancy B: Worried about reaction with underlying health condition. F: Found information online, at vaccination centre. N/A Adults in Recovery 1 One-on-one education, guidance, and support for adults in recovery from alcohol or substance abuse 40–49 M Astra Zeneca Some hesitancy B: Worried about reaction with underlying health condition. Unanswered questions. Unsure B: Not as fearful of Covid as before. F: Will get the booster if it’s an organizational policy. Management 1 Manages team providing business services, education supports, pre-employment and personal development courses, and health programs for Travellers 30–39 F Astra Zeneca Little hesitancy B: Guilt at receiving vaccine before those more vulnerable. F: Personal safety as a front facing worker. No hesitancy F: Best way out of Covid in the long term. Employment Services 1 Guidance counsellor for job seekers and those on social welfare 20–29 M Astra Zeneca No hesitancy F: Safety of older parents. Going on holidays. Some hesitancy B: People with underlying health conditions should have first access. Community Health Officer 1 Promotes health services within local Traveller community 20–29 F Pfizer Some hesitancy B: Fear of side effects. Lack of information. F: Going on holidays. Reduced risk of severe infection. Great hesitancy B: Fatigue. “How many times do we have to keep doing it, you know?” Employment Services 2 Guidance counsellor for job seekers and those on social welfare 40–49 F Astra Zeneca No hesitancy F: Return to normalcy. Personal safety as front facing worker. N/A Admin 1 Front facing community center employee 40–49 F Pfizer Great hesitancy B: Unanswered questions. Inability to make an informed decision. F: Paid to meet with a GP and got desired info. Great hesitancy B: Unmet expectations of vaccine effectiveness. Adults in Recovery 2 One-on-one education, guidance, and support for adults in recovery from alcohol or substance abuse 20–29 F Pfizer Some hesitancy B: Fear of side effects. F: No reports of allergic reactions in the news or through HSE. No hesitancy F: Trusts the science and government intentions. Employment Services 3 Guidance counsellor for job seekers and those on social welfare 40–49 M Astra Zeneca No hesitancy F: Return to normalcy. Safety of children. N/A Admin 2 Community centre receptionist 30–39 F Astra Zeneca Some hesitancy B: Belief that others deserved to go first. F: Needed access to hospital services. Obtention of vaccine passport. Great hesitancy B: Unmet expectations. Belief that the booster will do no good. B = barrier. F = facilitator. GP = general practitioner. 3.2 Community context Key informants believed community Covid-19 vaccination rates to be in line with national rates at the time of data collection (∼90 % of the eligible population). Though the local population was broadly accepting of the vaccine, participants noted “very strong anti-vaccination feelings in a small number of people” (Management 1). Two participants commented on a pattern of resistance whereby ‘anti-vaxxers’ tended to be ‘anti-maskers’ and harbour conspiracy beliefs. Stances on the vaccine varied by population group. Middle-aged and older clients of the hair salon were “very happy to get it done [and] get back out” (Hair Salon 2 and 3). Attitudes of unemployed community centre clients met “an absolute extreme on both sides, and in the middle” (Employment Services 3), though Employment Services 2 found that many clients who “moaned” about the vaccine still got it eventually. A genuine resistance was noted amongst community centre clients in recovery from addiction; whereas in the Traveller community, vaccine acceptance was possible under the right conditions (e.g., Pfizer instead of Johnson and Johnson vaccine, seeing others be vaccinated first, increased convenience for everyday life, the disease being “on the doorstep”). 3.3 The 4Cs In the following sections, we focus on four main themes explaining Covid-19 vaccine hesitancy in the disadvantaged Dublin community. The first three – Confidence, Complacency, Convenience – are in line with the WHO ‘3Cs’ model of Vaccine Hesitancy [26]. A separate theme of Covid-19 Communications emerged through inductive analysis to explain local hesitancy, as did Community-Centred strategies for improving vaccination willingness and uptake. Sample quotes for each theme and sub-theme are presented in Table 2 . Because key informants are themselves members of the target community, reported results integrate their insights into community perceptions of the vaccine with their own vaccination experiences.Table 2 Key informants’ perceptions on drivers of Covid-19 vaccine hesitancy in a disadvantaged Dublin community by theme and sub-theme. Theme/Sub-theme Sample Quote(s) Confidence Fear of side effects Novelty, speed of development “There's a bit of hesitancy towards it amongst some of the clients that I would have encountered. They would have kind of been like, ‘Oh, I'm not getting a vaccine that's after coming around that fast. I wouldn't know what they're going to be putting into our bodies.’” (Employment Services 2) Underlying health conditions “My daughter wasn't going to get it because she was a bit concerned 'cause she's epileptic.” (Hair Salon 2) “I have asthma and obviously we’re all aware of the difficulty of respiratory illness or whatever, so, I wanted to kind of see would that impact me in any way.” (Adults in Recovery 1) Close proximity to negative vaccination experience “I mean the odds were so low. But I suppose it’s been such an unusual couple of years that I think anxiety levels are probably heightened anyway. Then a good few had very bad side effects the day after so that didn’t help matters either.” (Management 1, speaking on Ireland’s pause of the AstraZeneca jab in March 2021 following reporting of blood clots) Cultural norms “For Traveller women, being infertile was a huge concern because a Traveller woman sees her life made when she’s married and has children. There’s a lot of women in the local Traveller community that aren’t vaccinated as a result and trying to talk them out of that is very difficult. Very, very difficult.” (Community Health Officer 1) Distrust in government and health services Feeling like the world is against you “They just think it's all a big scam. You know, we work in a disadvantaged area and there's a lot of people that have grown up feeling that the world is against them, that the government is against them. So, they already have that kind of mentality and mindset and are very easily swayed as to go against the grain while they're living in disadvantage and poverty.” (Employment Services 3) Fear of conspiracy “An article came out from the government saying that they’re gonna vaccinate homeless people and Travellers with Johnson & Johnson vaccine because it’s more practical for people that move around. Very common sense, but this was seen as an ethnic cleansing. That’s basically the way they saw it. Then a public figure came out and said ‘they’re trying to get rid of us Travellers’ and it was a nightmare trying to debunk that. The only thing we could do was offer other vaccines.” (Community Health Officer 1) Social pressure Community pressure “At the time it was Pfizer that was being given out [instead of Johnson and Johnson] and a lot of other people ended up jumping on. But there's a huge secrecy around it. You know, you’re standing in the garden and getting called in on the side, ‘Hey, can you get me a vaccine? But don’t tell anyone. I don’t want people knowing that I took it.’ So, I think in that kind of mob mentality, people are afraid to say, ‘No, actually, I did take it and I’m grand.’ That’s what we’re dealing with a lot.” (Community Health Officer 1) Family pressure “There was a family I was working with where the son wanted it but his Mam was completely anti-vaxx so he felt like he couldn't get it because he'd be going against her.” (Community Health Officer 1) Inadequate information Exposure to misinformation “I don’t think people are coming at it from a negative perspective necessarily. I think they’re getting bad information. And it’s making them very, very anxious and worried.” (Management 1) Lack of accurate information (individual) “I had a lot of questions that I wanted answered and they weren't answered, so I wasn't going to actually go and have something that I didn't know what I was dealing with.” (Admin 1) “I contacted my GP who was very unwilling to give me information and directed me to the HSE website… but specific information around my asthma was not there. I actually had to go on Google the NHS website and find out more information.” (Adults in Recovery 1) Lack of accurate information (GP) “I was having a conversation with people yesterday who were asking ‘Why would we have to get a third, like that's ridiculous.’ I was just saying it wears off, it's probably not as effective, you know. That's the only reason I can give people at the minute as a healthcare worker because I really don't know myself to be honest.” (Community Health Officer 1) “I don’t know why my GP was reluctant to give information. My own opinion is that there’s a lack of knowledge on their end as well. I don’t think they had the answers.” (Adults in Recovery 1) Complacency Low perceived risk of Covid-19 Lack of first-hand experience with severe Covid-19 “There was quite high incidence of Covid in [this community] at one stage, so I think generally they could have maybe known a lot of people that would have had Covid. But maybe the people that they knew weren't in hospital… so it could be based on a little bit of that. Because the people that they knew didn't have bad symptoms. Therefore, they feel that they don't need a vaccine to protect them from Covid.” (Employment Services 1) “[The local Traveller community] hasn’t been hugely impacted by Covid. There’s been one or two people that had it. One person was in hospital but a few days later was out standing in his garden. I think it’s just not taken seriously, whereas there was an outbreak of Hepatitis, and it was affecting children. So, there was an immediate response at the time. People were petrified.” (Community Health Officer 1, explaining high community uptake of hepatitis vaccine) Feeling less at-risk than others “I wouldn't be in a rush to get [the booster] because I feel like there is other like more vulnerable people that would benefit from it probably more than me.” (Employment Services 1) Reduced fear over time “For me the fear of Covid, it’s kind of, I’m not as fearful as I would have been maybe in April May, June of 2020. So, would take the booster if I had an option? If It was an organizational policy and I had to get it, I would get it.“ (Admin 2) Counterproductive vaccination incentives Resistance to vaccine passports “They might have missed the boat on [motivating people to be vaccinated]. You know, like the flu jab came out every year and you’d say to your friend, ‘Are you getting the jab? No? Grand.’ Every-one moves on, nobody is penalized. Where now this roll out is pushed in people’s faces, like they can’t go to McDonald’s. They can’t go and have a meal because they don’t have the Covid cert. So, I don’t think we can turn it around.” (Admin 2) Prioritizing freedom of choice “I do feel it is a little bit forced. Like, I know you still have an option whether to get it or not, but it restricts you a lot if you don't get it. I think it has to be, at this point, it has to be personal responsibility anyway. I don't think they should be telling us what to do.” (Admin 2) Unsuitable, divisive incentives “There are people who don't need a [vaccine certificate]. There are people out there who probably will not travel because they don't have the means. They don’t have the luxury of going on holiday or to a foreign country. We're a certain kind of cohort of the population that needs this certificate to function in daily life. We need it go for a meal, as I said, but there are people out there who don’t need it. Then for people that don't work, there's no incentive, there's no pressure for them to get [the vaccine], you know…Some of those people have already got health complications as well. So, they're saying to themselves, ‘well, I’ll be fine.’“ (Employment Services 1) “I think they’ve managed it very poorly with the Covid passports. I mean, there’s a 2-tier society going on, you know, and I think that pisses people off more than brings people with. You need to bring people with you, rather than get two sides kind of fighting against each other.” (Employment Services 3) Convenience Access barriers Transport/financial barriers “My mother is a family support worker with the HSE. One of the families she goes to is a lady that is on her own with four or five children, so she wouldn’t have the means for a car or anything like that. And one of her children actually had symptoms of Covid and the GP had suggested that they get her to go to [the HSE testing and vaccination centre] to have the child tested. Now, first of all, she has children that she couldn’t get minded. Her only means of getting to [the centre] was through a taxi. You know, she didn’t have the money for that.” (Admin 1) Lack of access to preferred vaccine “I know within this population there was a lot of like discourse over the right vaccine to get. One individual didn’t want the Astra Zeneca, just straight up refused to get vaccinated up until a couple weeks ago where he could actually go in and get Pfizer.“ (Adults in Recovery 1) Lack of IT/literary skills “I see a lot of clients that don’t have good literacy or IT skills, so they might not have the skills to go online and register on the vaccine portal, like a lot of older members of the community.” (Employment Services 1) Communications Communications breakdown Mixed messages “There's too many leaders saying too many different things. If they got one person to speak… I find they were saying different things throughout Covid, and that was confusing, especially a lot of the older people were very confused.” (Hair Salon 3) Confusing statistics “You know, NPHET is supposed to be the backbone of the pandemic. And I'm sorry but the muppet show… And I don't mean to be smart, I know they're well-educated men but, you know, the statistics and stuff they put up, a lot of people wouldn't get what that means.” (Admin 1) Overreporting of case numbers “People do watch the news and have radios on and all they’re hearing is case numbers. And I think that’s a massive problem because they’re not seeing any improvement. They’re just saying, ‘What’s the point?’ and I don’t blame them.” (Employment Services 2) Lack of encouragement “You ask any old person what they do on a daily basis. Sit down with their cup of tea and watch the 6o'clock news, and they've been like that for 40–50 years. And there was all this information that didn't necessarily need to be [communicated] to them. Where information on how well people were doing on the vaccine, or how the vaccine was going to help people, or you know the benefits of it, didn't happen, unfortunately.” (Admin 1) Illogical rules and regulations Public health measures without explanation “Closing nightclubs at 12:00o'clock when they only open at 11. Does Covid only come out at 12:01? All this stuff drives me bloody crazy. Like all these rules make no sense. You could go to a pub last year and you could stay there if you bought a meal because the meal saved you from COVID. Like it's just crazy, none of it makes sense.” (Admin 2) Disjointed approach “You've so many different stakeholders, my impression of it is that they're trying to please every-one and achieving nothing, you know, and I think that comes out in the communications. I just don't think there is a singular vision for how we’re going to get out of this. Or perhaps there is, but it's just not coming across, you know, so I think that's really damaging…I think if we do get it to a point where we have to reintroduce restrictions or anything like that, I think they’ll really struggle with it this time around.“ (Management 1) Unmet expectations Sense that the vaccine doesn’t work “I suppose they've always been telling us 'get as many people vaccinated as possible’ and now, I think over 90 % the population over 16 is vaccinated and obviously the case numbers are spiking again. So, it’s frustrating and I think probably for the people that were hesitant about getting a vaccine in the first place, it's maybe adding to their suspicions or concerns about it now that they see that all these people are vaccinated but they're still getting Covid, and the case numbers are still going up.” (Employment Services 1) Being sold the wrong story “I just think that perceptions were kind of wrong. People thought that the vaccine was going to stop people getting the virus, which it actually doesn't. It just stops people getting really sick from the virus and I'm not sure that message was put across properly.” (Employment Services 2) Scepticism stemming from false hope “How many times were we told, ‘two weeks to flatten the curve’? And, ‘just another two weeks’? It's been a while now at this stage and it's hasn’t flattened. So, I think there is probably a sense that maybe people don't know what they're doing at government level… I think it's harder to convince people to make sacrifices in their own lives when they don't actually feel like it's really going to have an impact.” (Management 1) Pandemic fatigue Wanting to move on “I think it's a lot more difficult this time with the boosters, 'cause we were sold a story that we'd be grand once we're all vaccinated, and we're not. So, it is going to be harder. People are Covid-fatigued, and just tired after the last few years. I think it will be hard enough to hit the numbers that we need. But, I mean, just keep a consistent message I think would be a good way forward.” (Management 1) HSE = Health Service Executive Ireland. NHS = National Health Service England. 3.4 Confidence “There’s probably-two reasons why people are hesitant. One: being genuinely afraid of putting something into their body, and two: being anti-establishment.” (Employment Services 2). 3.4.1 Fear of side effects Participants acknowledged that lack of trust in the effectiveness and safety of vaccines, and lack of trust in the system and authorities that deliver them were primary drivers of hesitancy in the community. Fear surrounding the vaccine’s safety stemmed from how “fast” (Adults in Recovery 1, Employment Services 1) it was rolled out, and its perceived “trial” status (Hair Salon 1). Safety concerns were heightened in individuals with underlying health conditions and those who witnessed and/or heard reports of serious side effects. In the Traveller community, fear of infertility was a concern amongst women due to the cultural weight placed on having a family. 3.4.2 Distrust in government and health services Anti-establishment sentiments and distrust in government and health services stemming from economic disadvantage further impeded Covid-19 vaccine uptake. Key informants working in employment services noted that clients felt “left behind”, “angry” (Employment Services 2), “poorly treated by government departments”, and that “the government doesn’t care” (Employment Services 3). Though some clients simply needed space to “rant” (Employment Services 2) before getting vaccinated, for others, the consequences of “paranoia” and “lack of trust in the government” (Adults in Recovery 2) were further reaching. Some would not engage with health services as a result or did not have a good relationship with their general practitioner (GP). A history of social inequities and poor community health outcomes left clients feeling that a vaccine wasn’t “gonna change much” (Employment Services 3). Combining anecdotes from Adults in Recovery 1 and 2, a picture emerges of how a history of being let down by health services compiled with lack of information on Covid-19 has created distrust towards the vaccine and its providers amongst former drug users (Fig. 4 ).Fig. 4 Drivers of vaccine resistance amongst adults in recovery from drug addiction as reported by community centre Guidance Counsellors (1) and (2): 19 November 2021, Dublin. GP = General Practitioner. A&E = Accident and emergency department. HSE = Health Service Executive Ireland. In some instances, distrust went as far as to instil fear of conspiracy. Hair salon 2 and 3 both heard rumours circulating in the community of microchip injections, noting a “genuine fear” (Hair salon 3). Community Health Officer 1 outlined the extent to which local Travellers feared malicious intent: the single dose Johnson & Johnson vaccine, prioritised over two dose vaccines for vulnerable groups to support efficiency and coverage in complex environments [29], was believed to be a means of ethnic cleansing. 3.4.3 Social pressure When describing fears circulating in the Traveller community, Community Health Officer 1 spoke of a tendency towards “mob mentality.” Travellers based their vaccination stance on that of trusted community leaders who spoke out against the Johnson & Johnson vaccine. Those who disagreed were afraid to speak out against popular opinion. The phenomenon of “jumping on the bandwagon” to be “outwardly against something” was observed by Employment Services 3, crediting the tendency for negative stories to gather more weight than positive stories. This type of social pressure affected families. Three participants mentioned instances of a parent discouraging their adolescent child to be vaccinated: two participants heard of adult children discouraging elderly parents. 3.4.4 Inadequate information Five participants emphasized the role that misinformation spread via social media and word-of-mouth played in fuelling fears of side effects and conspiracy. They noted that community members may lack the resources to challenge misinformation shared by trusted personal contacts. Participants themselves found it difficult to debunk rumours and make informed decisions due to a lack of accessible, accurate information. Adults in Recovery 1 and Admin 1 found no information on the Health Service Executive Ireland (HSE) website on how the vaccine would react with their underlying health conditions and turned to their GPs for answers. Adults in Recovery 1 never got an appointment: Admin 1 paid 60€ for one. Even healthcare professionals lacked adequate information. Community Health Officer 1 never received specific training on Covid-19 as part of their healthcare role, relying on independent research and, in some instances, “literally just assuming.” 3.5 Complacency “People have relaxed a little bit and I don’t think there’s that same sense of life and death that was there very early on.” (Management 1). 3.5.1 Low perceived Covid-19 risk Complacency refers to factors supporting a view that the risks of Covid-19 are low, and vaccination is not considered a necessary preventive action. Employment Services 1 explained that low perceived risk manifested in the community early in the pandemic because most people had experienced and/or witnessed only mild cases of Covid-19. Conversely, participants noted how a first-hand experience with severe Covid-19 or other illness amplified perception of risk and increased vaccination uptake. Four participants thought their personal level of risk did not merit receiving the Covid-19 vaccine before other more vulnerable people, expressing guilt at going before those who needed it more. Participants felt that fear of Covid-19 had waned over the course of the pandemic, acknowledging that people “weren’t scared anymore” (Admin 2), had grown “complacent” (Management 1), and “were just getting on with it” (Admin 2). 3.5.2 Counterproductive vaccination incentives The theme of complacency emerged indirectly in attitudes towards the vaccine that implied low perceived risk of the virus. At the time of data collection, a vaccine certificate (i.e., proof of full vaccination or recovery from Covid-19) was required for indoor hospitality and events, and for most international travel [40]. The restrictions led many community members to be vaccinated out of social or professional convenience rather than as a necessary preventive action. Participants highlighted potential push back from those who disagreed with restrictions for the unvaccinated, emphasizing people’s right to and preference for making their own medical decisions. Of five participants who mentioned feeling pressurized to get the vaccine either through work or in order to avoid restrictions, none were planning on getting a booster shot at the time of data collection. For many, with fear of Covid-19 waning over time, upholding freedom of choice took precedence over worries about the virus and its health consequences. Relying on non-health related incentives for Covid-19 vaccination may also inadvertently discourage immunization in disadvantaged community members who are frustrated by divisive social and occupational restrictions. 3.6 Convenience “Her only means of getting to [the vaccination centre] was through a taxi. You know, she didn't have the money for that.” (Admin 1). 3.6.1 Access barriers At the time of data collection, the closest HSE vaccination centre was located approximately 20 min on public transport from the local area. This could pose a challenge for elderly people who remained “nervous about getting on a bus” (Hair salon 3), and/or for those without the financial means for a taxi or to have children minded. Some community members were unable to access their preferred vaccine; others had trouble registering for an appointment online due to limited IT and/or literacy skills. Community Health Officer 1 spoke of a one-day mass vaccination campaign initiated for the local Traveller community. Beyond this, participants were unaware of vaccination campaigns being brought to the local area. 3.7 Covid-19 Communications “There’s hostility and fear there because of the lack of communication, and lack of support, and a lack of trying to get people to understand what’s going on here, why this is happening.” (Adults in Recovery 1). 3.7.1 Communications breakdown While identified subthemes generally fell under the WHO 3cs framework for vaccine hesitancy, a separate theme emerged relating to government and media communications. Participants shared a view that communication failures reinforced local vaccine hesitancy during the pandemic. A breakdown of communication was described whereby “mixed messages”, “lack of clarity” (Employment Services 3), and “contradictions” (Community Health Officer 1) from the government and media led to “hostility”, “fear” (Adults in Recovery 1) and “damaged trust” (Management 1) in the community. Contradictory messages from multiple leaders, and the tendency to use big words and statistics were confusing for local community members. Participants attributed some of the communications breakdown to the pandemic’s increasing complexity over time and the dilution of accurate messages due to the quantity of false information on social media. Nevertheless, they felt that unsatisfactory government and media communications, particularly the overreporting of case numbers and lack of encouraging vaccination updates, further deterred vaccine hesitant individuals from seeking out immunization. 3.7.2 Illogical rules and regulations More than half of participants were frustrated by a sense that some public health measures – for example, a closing time of midnight instead of 2am for all on-licensed premises in November 2021 and a requirement that pubs serve a meal of the value of €9 per customer in order to reopen in June 2020 – “made no sense.” (Admin 2). The lack of clarity behind specific approaches “planted seeds in people’s heads” (Admin 1) that they needn’t follow restrictions. One participant made a direct connection between diminished trust in the government’s ability to lead due to confusing regulations and struggling to get every-one “on board” (Admin 1) with vaccination. 3.7.3 Unmet expectations of vaccine effectiveness Unsatisfactory communications also led to unmet expectations of the vaccine’s effectiveness. Ten of twelve participants believed that the pandemic situation would be under control once vaccinations were rolled out and expressed disappointment that case numbers were rising at the time of data collection. Participants described how confusion, frustration, and anger due to perceived lack of effectiveness of the vaccine led to the entrenchment of community scepticism. For those who had been initially accepting of the vaccine, unmet expectations contributed to Covid-19 booster resistance as participants and community members were left with a feeling of, “what’s the point?” (Adults in Recovery 2, Admin 2). Examples of miscommunications that led to disillusionment with the vaccine included selling the vaccine as preventive against all Covid-19 infection, rather than severe Covid-19 infection, and creating false hope by continuously reassuring the population that things would improve in “just another few weeks.” (Hair salon 1). 3.7.4 Pandemic fatigue The culmination of unmet expectations, confusing regulations, and a general breakdown of communication was a sense of community-wide fatigue. Participants described a sense of “apathy” (Admin 1), being “fed up” (Employment Services 1), and “wanting to move on” (Admin 2) with the pandemic. These sentiments had negative implications for the local booster campaign. Some community members that had their two vaccinations felt they had “done their duty” (Admin 1) and weren’t having any more. 3.8 Community-centred solutions 3.8.1 Providing accurate, accessible information To establish confidence in the vaccine and address complacency, participants underlined the importance of providing communities with “the right information to make an informed choice” (Employment Services 2) through conversation and upscaled Covid-19 information resources. Recommended information providers varied by population group. Generally, participants found that conversations with health professionals can “put minds at ease” (Hair Salon 3). For the elderly, public health nurses and community registered general nurses providing in-home care were identified as effective providers of Covid-19 information. For populations with distrust in health professionals, “it would be useful to appoint someone independent with a scientific background to a Covid response role where they go around to different community centres and answer peoples’ questions.” (Adults in Recovery 2). Setting up information stands, providing leaflets at the local chemist, implementing a Covid-19 helpline, and – for the digitally literate – conducting informational zoom meetings, webinars, and podcasts in understandable language came up as feasible ways to improve local knowledge and acceptance of the vaccine. 3.8.2 Building trust in the vaccine and its providers Participants suggested bringing regular Covid-19 question and answer sessions and vaccine campaigns into the community via trusted community-based organizations like youth groups and medical charities. Specific trust-building techniques emerged through inductive analysis:• Ongoing dialogue: “Bringing people together to ask questions and get answers” (Management 1) and “having conversations about initial concerns or reservations in [understandable language].” (Employment Services 1) • Relationship building: “Building a rapport with people who may feel backed into a corner and are used to fighting” (Employment Services 3) by “identifying specific goals”, shifting from a “one-size-fits-all” approach to address individual concerns, and “actively listening” (Adults in Recovery 1). • Erasing preconceptions: “Becoming familiar with vaccine concerns” (Employment Services 1), “being empathetic”, “not talking [down] to people that are not vaccinated” (Management 1), and “understanding it’s a process, that you can’t flip a switch” (Employment Services 3). • Communicating effectively: “Providing real evidence to debunk misinformation” (Adults in Recovery 1), and “letting [community members] know what you’re aiming for, how you’re trying to do it, and being honest and upfront” (Employment Services 3). 3.8.3 Improving vaccine access Along with upscaling local vaccination campaigns and awareness efforts, participants recommended “being more inclusive of communities where general and digital literacy are an issue” (Employment Services 1). Providing marginalized community groups (i.e., Travellers, adults in recovery from addiction) with a choice of vaccine and facilitating private vaccination requests to overcome social pressure and vaccine stigmatization could also improve vaccine uptake. To reduce viral transmission and, by slowing the spread of Covid-19 in the local community, improve perceptions of the vaccine’s effectiveness, two participants suggested simultaneously expanding access to affordable antigen tests. 4 Discussion This qualitative study was the first to examine drivers of Covid-19 vaccine hesitancy in Ireland through consultation with community representatives. While results confirmed that drivers of hesitancy in a disadvantaged urban community largely fell under the WHO Confidence, Complacency, Convenience model [26], the Irish government and media’s handling of Covid-19 communications emerged as a novel barrier to vaccination acceptance and uptake. Prior to Covid-19 vaccination roll-out in Ireland, Murphy et al. (2021) suggested that public health messaging should be clear, direct, repeated, and positively orientated to target the psychological characteristics of those prone to vaccine hesitance or resistance [24]. Our study outlines how pandemic communications missed these objectives, contributing to the entrenchment of anti-authority sentiments and offering one explanation for increased resistance to Covid-19 vaccination in Ireland during the pandemic [25]. While vaccine-safety related concerns have been identified as the main determinant of vaccine hesitancy in Europe and the UK [41], [42], key informants identified anti-established sentiments stemming from a history of being let down by the government and health services as a primary local challenge. Barriers to vaccination uptake specific to adults in recovery from addiction were foreshadowed in a 2019 review on methadone treatment protocol in Ireland [43]. Service users described negative program aspects including patient lack of choice, humiliating experiences consuming methadone in a public space, engaging with uncaring service providers, and being treated with a one size fits all approach [43]; all identified in this study as drivers of Covid-19 vaccine hesitancy. Complacency may also prevent uptake in this group. The primary barrier to vaccination amongst 872 surveyed people who inject drugs in Australia was lack of perceived vaccine utility [44]. Identified barriers to Covid-19 vaccine uptake amongst Irish Travellers (e.g., cultural concerns about vaccines offered during pregnancy, misinformation spread via social media and ‘word of mouth’) have been cited in relation to other vaccines, as have potential facilitators including sufficient understanding of the vaccine and trust in health professionals [45]. The reported negative reaction of the Traveller community towards receiving a single rather than double dose Covid-19 vaccine underlines the importance of applying key informants’ recommendations for trust-building (e.g. ongoing dialogue, erasing pre-conceptions) before the implementation of well-intentioned public health measures, as well as after. Participants expressed negative community sentiments and resistance towards non-health related vaccination incentives and ‘being told what to do’. This is in line with findings from a UK study demonstrating that vaccine passports may induce a lower vaccination inclination in socio-demographic groups that are less confident in Covid-19 vaccines [46]. Social and professional restrictions make those who already intend to get vaccinated even more inclined to do so, potentially explaining surges in vaccination following implementation of a national vaccine passport policy [46], [47]. But research shows, as do our own study findings, that pressurizing those with doubts about the vaccine to vaccinate reinforces resistance, particularly for those who are economically deprived and/or unemployed [46]. Prioritisation of education and outreach initiatives to combat vaccine scepticism and misinformation emerges as a better-suited strategy for encouraging vaccination in disadvantaged communities. Encouragingly, results from this study confirm the effectiveness of many strategies already used by the HSE and non-governmental organizations (NGOs) for ensuring equitable vaccination in Ireland. The HSE’s comprehensive vaccine approach for vulnerable groups, including Travellers and those in addiction settings, recommends a hands-on approach using trusted sources within each population group to listen, alleviate individual concerns, and encourage vaccine participation [29]. Our study findings suggest that this type of ‘champion’ – or someone with a scientific background appointed to a Covid response role, as suggested by one key informant – would be of value at the wider community level in disadvantaged areas. Vaccine communication plans for vulnerable groups are in progress at the HSE, who has called for targeted approaches for meeting information needs [29]. Key informants’ perspectives can again be of value: strategies like Q&A sessions with scientists, healthcare professionals and community representatives that facilitate relationship building and ongoing dialogue should be prioritized. GPs, pharmacists, and community health care workers working with disadvantaged populations should receive additional training on Covid-19 vaccine effectiveness and potential side effects, and on how to communicate this knowledge effectively and empathetically to individuals with limited trust and/or health literacy. Irish NGOs are leading crucial community-level vaccination initiatives in collaboration with the HSE. Pavee Point, an NGO addressing Traveller issues and promoting Traveller rights, has an online ‘Travellers Take the Vaccine’ page with community member video testimonies addressing many of the vaccine fears and concerns outlined in this research study and linking viewers with the HSE website and vaccine helpline [49]. The medical charity Safetynet’s Covid Cluster Rapid Response Teams facilitate pop-up testing, vaccination, and health promotion clinics. Coordinating outreach and communication between federal, state, and local partners such as these will enhance trust in the national vaccination strategy and prevent the breakdown of communication described by key informants [50]. Expanding access to tailored Covid-19 information resources – local helplines, leaflets in familiar language, information sessions conducted in community centres— can ensure that members of disadvantaged communities who are prone to vaccine scepticism understand what makes the vaccine safe and protective. The success of expanded vaccine information efforts will require bolstering trust in the government’s ability to lead during this and future pandemics. Considerations include accompanying new public health measures with clear explanations of the scientific rational behind them; creating realistic expectations of vaccine effectiveness; and expanding access to supplementary preventive resources like antigen tests in disadvantaged communities. During ongoing and future public health crises, communications from government and health officials should not be overly reassuring or foster false illusions of certainty that further erode trust. 4.1 Public health implications The proposed ‘4Cs’ model of Covid-19 vaccine hesitancy provides a tool for considering vaccine hesitancy in disadvantaged urban areas in Ireland in the context of Covid-19 and future pandemics requiring the rapid development and distribution of a novel vaccine. The model can be tested, adapted, and validated in comparative sites nationally and internationally, particularly in high-income countries experiencing community-level Covid-19 vaccine disparities and stalled booster campaigns (e.g., United States, United Kingdom, France). Validation within specific marginalised communities at risk of vaccine hesitancy will also be important (e.g., individuals experiencing homelessness, people with disabilities, ethnic minority groups, migrants). Study findings demonstrate a need for transparent and targeted communication about first-round and repeated Covid-19 vaccination at the local and national level. Whilst overcoming Covid-19 vaccine hesitancy is critical for ensuring that vulnerable communities are adequately protected from the ongoing pandemic and its consequences, preventing future vaccine inequalities and related health disparities will require addressing the systemic neglect and marginalisation experienced by economically and socially disadvantaged individuals. As one small step in this process, our research team aims to conduct community-based participatory research with this study’s target community to facilitate ongoing dialogue on priority health needs and to strengthen relationships between health care providers, researchers, and marginalised community members. 4.2 Limits This study holds potential for information bias as the views of key informants regarding community perceptions on vaccine hesitancy may be influenced by their own experiences with and feelings toward the Covid-19 vaccine. As well, a relatively small number of key informants were interviewed to represent all vulnerable population groups in the community. Nevertheless, the fact that testimonies were similar enough across participants to achieve data saturation after a first round of interviews, and that identified community drivers of hesitancy closely reflected findings from international research efforts [26] confirm the internal and external validity of the study. 5 Conclusion This qualitative study was the first to gather empirical evidence on Covid-19 vaccine hesitancy in a disadvantaged urban community in Ireland. A Confidence, Complacency, Convenience, Communications (‘4Cs’) model of Covid-19 vaccine hesitancy emerged through inductive analysis of key informant interviews. While many drivers of hesitancy in the disadvantaged Dublin community fell under the WHO ‘3Cs’ model, Covid-19 Communications emerged as a separate theme whereby unclear messages, confusing public health measures and unmet expectations of the vaccine’s effectiveness entrenched vaccine scepticism and distrust in the government’s ability to lead during the pandemic. Community-centred strategies for improving information resources, rebuilding trust, and expanding vaccine access were identified by key informants. The emergent ‘4Cs’ model of hesitancy provides key insights and strategies for tackling vaccine hesitancy in disadvantaged urban communities and can be used to compliment equitable vaccination efforts currently underway by public health agencies and NGOs. CRediT authorship contribution statement Carolyn Ingram: Conceptualization, Methodology, Formal analysis, Writing – original draft, Visualization. Mark Roe: Conceptualization, Methodology, Writing – review & editing. Vicky Downey: Formal analysis, Writing – review & editing. Lauren Phipps: Conceptualization, Methodology, Writing – review & editing. Carla Perrotta: Conceptualization, Methodology, Resources, Writing – review & editing, Supervision, Project administration, 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. Data availability The authors do not have permission to share data. Acknowledgements The authors are grateful to management and staff of the collaborating community partnership organization and hair salon, without whom this research project would not have been possible. Funding This work was supported by Science 10.13039/100016163 Foundation Ireland [grant number 20/COV/8539]. Ethical approval Informed consent to participate in this study was obtained from all participants. All participants were 18 years of age or older. All methods were performed in accordance with the guidelines and regulations outlined in the Declaration of Helsinki. This study was approved for exemption from full ethical review by the University College Dublin Human Research Ethics Committee – [Sciences (HREC-LS)]. Research Ethics Exemption Reference Number (REERN): UCD-HREC-LS-E-21-222-Perrotta ==== Refs References 1 MacDonald N.E. Vaccine hesitancy: Definition, scope and determinants Vaccine 33 2015 4161 4164 10.1016/j.vaccine.2015.04.036 25896383 2 Riedel S. Edward Jenner and the History of Smallpox and Vaccination Baylor University Medical Center Proc 18 2005 21 25 10.1080/08998280.2005.11928028 3 Leong C. Jin L. Kim D. Kim J. Teo Y.Y. Ho T.-H. Assessing the impact of novelty and conformity on hesitancy towards COVID-19 vaccines using mRNA technology Commun Med 2 2022 1 6 10.1038/s43856-022-00123-6 35603280 4 Mascherini M. Nivakoski S. Social media use and vaccine hesitancy in the European Union Vaccine 40 2022 2215 2225 10.1016/j.vaccine.2022.02.059 35249775 5 Weitzer J. Birmann B.M. Steffelbauer I. Bertau M. Zenk L. Caniglia G. Willingness to receive an annual COVID-19 booster vaccine in the German-speaking D-A-CH region in Europe: a cross-sectional study Lancet Regional Health – Europe 2022 18 10.1016/j.lanepe.2022.100414 6 Goldberg Y. Mandel M. Bar-On Y.M. Bodenheimer O. Freedman L. Haas E.J. Waning Immunity after the BNT162b2 Vaccine in Israel N Engl J Med 385 2021 e85 10.1056/NEJMoa2114228 7 Zhong D. Xiao S. Debes A.K. Egbert E.R. Caturegli P. Colantuoni E. Durability of antibody levels after vaccination with mRNA SARS-CoV-2 vaccine in individuals with or without prior infection JAMA 326 2021 2524 2526 10.1001/jama.2021.19996 34724529 8 Andrews N. Stowe J. Kirsebom F. Toffa S. Sachdeva R. Gower C. Effectiveness of COVID-19 booster vaccines against COVID-19-related symptoms, hospitalization and death in England Nat Med 28 2022 831 837 10.1038/s41591-022-01699-1 35045566 9 Imperial College London. Report 43 - Quantifying the impact of vaccine hesitancy in prolonging the need for Non-Pharmaceutical Interventions to control the COVID-19 pandemic. 2021. 10 Thompson MG. Interim Estimates of Vaccine Effectiveness of BNT162b2 and mRNA-1273 COVID-19 Vaccines in Preventing SARS-CoV-2 Infection Among Health Care Personnel, First Responders, and Other Essential and Frontline Workers — Eight U.S. Locations, December 2020–March 2021. MMWR Morb Mortal Wkly Rep 2021;70. https://doi.org/10.15585/mmwr.mm7013e3. 11 Tenforde MW. Effectiveness of Pfizer-BioNTech and Moderna Vaccines Against COVID-19 Among Hospitalized Adults Aged ≥65 Years — United States, January–March 2021. MMWR Morb Mortal Wkly Rep 2021;70. https://doi.org/10.15585/mmwr.mm7018e1. 12 Sheikh A. Robertson C. Taylor B. BNT162b2 and ChAdOx1 nCoV-19 vaccine effectiveness against death from the delta variant NEJM 385 2021 2195 2197 10.1056/NEJMc2113864 34670038 13 Jargowsky PA, Tursi NO. Concentrated Disadvantage. In: Wright JD, editor. International Encyclopedia of the Social & Behavioral Sciences, 2nd Ed. Oxford: Elsevier; 2015, p. 525–30. https://doi.org/10.1016/B978-0-08-097086-8.32192-4. 14 Crane M.A. Faden R.R. Romley J.A. Disparities in county COVID-19 vaccination rates linked to disadvantage and hesitancy Health Aff (Millwood) 40 2021 1792 1796 10.1377/hlthaff.2021.01092 34724416 15 Bucyibaruta G. Blangiardo M. Konstantinoudis G. Community-level characteristics of COVID-19 vaccine hesitancy in England: a nationwide cross-sectional study Eur J Epidemiol 2022 10.1007/s10654-022-00905-1 16 Gaughan C.H. Razieh C. Khunti K. Banerjee A. Chudasama Y.V. Davies M.J. COVID-19 vaccination uptake amongst ethnic minority communities in England: a linked study exploring the drivers of differential vaccination rates J Public Health 2022 fdab400 10.1093/pubmed/fdab400 17 Ingram C. Min E. Seto E. Cummings B. Farquhar S. Cumulative Impacts and COVID-19: implications for low-income, minoritized, and health-compromised communities in King County, WA J Racial and Ethnic Health Disparities 2021 10.1007/s40615-021-01063-y 18 Andersen L.M. Harden S.R. Sugg M.M. Runkle J.D. Lundquist T.E. Analyzing the spatial determinants of local Covid-19 transmission in the United States Sci Total Environ 754 2021 142396 10.1016/j.scitotenv.2020.142396 19 Kulu H. Dorey P. Infection rates from Covid-19 in Great Britain by geographical units: a model-based estimation from mortality data Health Place 67 2021 102460 10.1016/j.healthplace.2020.102460 20 Walsh B, Redmond P, ESRI, Roantree B, ESRI. Differences in risk of severe outcomes from COVID-19 across occupations in Ireland. ESRI; 2020. https://doi.org/10.26504/sustat93. 21 Madden J.M. More S. Teljeur C. Gleeson J. Walsh C. McGrath G. Population mobility trends, deprivation index and the spatio-temporal spread of coronavirus disease 2019 in Ireland Int J Environ 18 2021 6285 10.3390/ijerph18126285 22 Webb Hooper M. Nápoles A.M. Pérez-Stable E.J. No populations left behind: vaccine hesitancy and equitable diffusion of effective COVID-19 vaccines J Gen Intern Med 36 2021 2130 2133 10.1007/s11606-021-06698-5 33754319 23 Government of Ireland, Ordnance Survey Ireland. Vaccinations 2020. https://covid-19.geohive.ie/pages/vaccinations (accessed January 8, 2022). 24 Murphy J. Vallières F. Bentall R.P. Shevlin M. McBride O. Hartman T.K. Psychological characteristics associated with COVID-19 vaccine hesitancy and resistance in Ireland and the United Kingdom Nat Commun 12 2021 29 10.1038/s41467-020-20226-9 33397962 25 Hyland P. Vallières F. Shevlin M. Bentall R.P. McKay R. Hartman T.K. Resistance to COVID-19 vaccination has increased in Ireland and the United Kingdom during the pandemic Public Health 195 2021 54 56 10.1016/j.puhe.2021.04.009 34052508 26 World Health Organization. Report of the SAGE Working Group on Vaccine Hesitancy. 2014. 27 Central Statistics Office. COVID-19 Vaccination Statistics Series 2. https://www.cso.ie/en/releasesandpublications/fp/fp-cvac/covid-19vaccinationstatisticsseries2/ (accessed October 3, 2022). 28 European Centre for Disease Prevention and Control. Catalogue of interventions addressing vaccine hesitancy. European Centre for Disease Prevention and Control. 2017. 29 Fitzgerald D.M. McKenna J.-A. HSE Vaccine Approach for Vulnerable Groups in Ireland Health Services Ireland 2021 30 Haase T. Pratschke J. The 2016 Pobal HP Deprivation Index for Small Areas (SA) Pobal 2017 31 Department of Rural and Community Development. Social Inclusion & Community Activation Programme Requirements 2018-2022. 2017. 32 Israel B.A. Schulz A.J. Parker E.A. Becker A.B. Review of community-based research: assessing partnership approaches to improve public health Annu Rev Public Health 19 1998 173 202 10.1146/annurev.publhealth.19.1.173 9611617 33 Page S.M. Chur-Hansen A. Delfabbro P.H. Hairdressers as a source of social support: a qualitative study on client disclosures from Australian hairdressers’ perspectives Health Soc Care Community 2021 1 8 10.1111/hsc.13553 34 Robert Wood Johnson Foundation. Semi-structured Interviews 2008. http://www.qualres.org/HomeSemi-3629.html (accessed June 29, 2021). 35 Palinkas L.A. Horwitz S.M. Green C.A. Wisdom J.P. Duan N. Hoagwood K. Purposeful sampling for qualitative data collection and analysis in mixed method implementation research Adm Policy Ment Health 42 2015 533 544 10.1007/s10488-013-0528-y 24193818 36 Marshall M.N. The key informant technique Fam Pract 13 1996 92 97 8671109 37 Bradley E.H. Curry L.A. Devers K.J. Qualitative data analysis for health services research: developing taxonomy, themes, and theory Health Serv Res 42 2007 1758 1772 10.1111/j.1475-6773.2006.00684.x 17286625 38 Gale N.K. Heath G. Cameron E. Rashid S. Redwood S. Using the framework method for the analysis of qualitative data in multi-disciplinary health research BMC Med Res Methodol 13 2013 117 10.1186/1471-2288-13-117 24047204 39 Nowell L.S. Norris J.M. White D.E. Moules N.J. Thematic analysis: striving to meet the trustworthiness criteria Int J Qual Methods 16 2017 1 13 10.1177/1609406917733847 40 Department of the Taoiseach, Department of Health. Public health measures in place right now 2021. https://www.gov.ie/en/publication/3361b-public-health-updates/ (accessed January 8, 2022). 41 Adel Ali K. Pastore Celentano L. Addressing vaccine hesitancy in the “Post-Truth” Era Eurohealth 23 2017 16 20 42 Freeman D. Loe B.S. Chadwick A. Vaccari C. Waite F. Rosebrock L. COVID-19 vaccine hesitancy in the UK: the Oxford coronavirus explanations, attitudes, and narratives survey (Oceans) II Psychol Med 2020 1 15 10.1017/S0033291720005188 43 Delargy I. Crowley D. Van Hout M.C. Twenty years of the methadone treatment protocol in Ireland: reflections on the role of general practice Harm Reduct J 16 2019 5 10.1186/s12954-018-0272-4 30654803 44 Price O. Dietze P. Sullivan S.G. Salom C. Peacock A. Uptake, barriers and correlates of influenza vaccination among people who inject drugs in Australia Drug Alcohol Depend 226 2021 108882 10.1016/j.drugalcdep.2021.108882 45 Jackson C. Dyson L. Bedford H. Cheater F.M. Condon L. Crocker A. UnderstaNding uptake of Immunisations in TravellIng aNd Gypsy communities (UNITING) A qualitative interview study Health Technol Assess Rep 20 2016 1 176 10.3310/hta20720 46 de Figueiredo A. Larson H.J. Reicher S.D. The potential impact of vaccine passports on inclination to accept COVID-19 vaccinations in the United Kingdom: Evidence from a large cross-sectional survey and modeling study EClinicalMedicine 40 2021 10.1016/j.eclinm.2021.101109 47 McGarvey E. Covid-19: Irish vaccine passports “accelerated” jab uptake BBC News 2021 49 Pavee Point. Travellers Take the Vaccine 2021. https://www.paveepoint.ie/travellers-take-the-vaccine/ (accessed January 11, 2022). 50 Hunter C.M. Chou W.-Y.S. Webb Hooper M. Behavioral and social science in support of SARS-CoV-2 vaccination: National Institutes of Health initiatives Transl Behav Med 11 2021 1354 1358 10.1093/tbm/ibab067 34080616	36496286	PMC9715488	NO-CC CODE	2022-12-07 23:16:36	no	Vaccine. 2022 Dec 2; doi: 10.1016/j.vaccine.2022.11.072	utf-8	Vaccine	2,022	10.1016/j.vaccine.2022.11.072	oa_other
==== Front Hematol Transfus Cell Ther Hematol Transfus Cell Ther Hematology, Transfusion and Cell Therapy 2531-1379 2531-1387 Associação Brasileira de Hematologia, Hemoterapia e Terapia Celular. Published by Elsevier España, S.L.U. S2531-1379(22)01460-2 10.1016/j.htct.2022.10.001 Original Article The course of COVID-19 in patients with hematological malignancies and risk factors affecting mortality: A cross-sectional study Eryilmaz-Eren Esma a⁎ Ture Zeynep b Kilinç-Toker Ayşin a Korkmaz Serdal a Çelik İlhami a a Kayseri City Education and Research Hospital, Kayseri, Turkey b Erciyes University, Faculty of Medicine, Kayseri, Turkey ⁎ Corresponding author at: Kayseri City Education and Research Hospital, Department of Infectious Diseases and Clinical Microbiology, Kayseri, Turkey. 2 12 2022 2 12 2022 25 1 2022 25 10 2022 © 2022 Associação Brasileira de Hematologia, Hemoterapia e Terapia Celular. Published by Elsevier España, S.L.U. 2022 Associação Brasileira de Hematologia, Hemoterapia e Terapia Celular 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. Objective This study aimed to determine the clinical outcomes and risk factors affecting mortality in patients with COVID-19 following hematological malignancy (HM). Methods Patients diagnosed with HM and hospitalized for COVID-19 were included in this retrospective study. The age, demographic and clinical characteristics, prognosis and treatment of surviving and non-surviving patients were compared. Results A total of 49 patients were included in this study, 17 (34.6%) of whom died within 28 days of being diagnosed with COVID-19. Older age (p = 0.001), diabetes (p = 0.001), chronic obstructive pulmonary disease (p = 0.002), secondary infection (p < 0.001) and secondary bacterial infection (p = 0.005) were statistically significantly higher in non-survivors. The remission status of HM was higher in surviving patients (p < 0.001). In multivariate regression analysis, age (OR: 1.102, p = 0.035) and secondary infection (OR: 16.677, p = 0.024) were risk factors increasing mortality, the remission status of HM (OR: 0.093, p = 0.047) was a protective factor from mortality. Conclusion The older age, the remission status of HM and secondary infection due to COVID-19 were determined as prognostic factors predicting mortality in HM patients with following COVID-19. Keywords Hematological malignancy COVID-19 Older age Remission ==== Body pmcIntroduction Hematological malignancies (HMs) are one of the leading causes of mortality and morbidity worldwide.1 The patient with HM usually has long-term immunodeficiency due to the malignancy, anti-cancer treatments or procedures, such as hematopoietic stem cell transplantation (HSCT).2 These people are very concerned about the high risk of morbidity and mortality from COVID-19. According to the literature, the overall mortality associated with COVID-19 among hematologic patients ranges from 32% to 40%.3 Recent studies have demonstrated that COVID-19 has a two-phase illness process. The classic upper respiratory tract infection symptoms are observed during the first phase. If the disease does not improve, it progresses to the second phase and a 'cytokine storm' develops, in which increased cytokine levels characterize an enhanced immune response. There are dyspnea and hypoxia, with infiltrates in the lung.4, 5, 6 It was determined that activated and depleted CD8+ T cells were associated with the severity of the disease in COVID-19.7 In the early stages of the pandemic, the cytokine storm seen in the course of COVID-19 was identified. It was thought that the existing immunodeficiency in patients with HM would not allow for the development of a cytokine storm and would limit the progression to severe/critical COVID-19. However, real-life data indicate that morbidity and mortality rates are variable. The subtype and activation status of HM, age and comorbidities were identified as factors influencing mortality.8 , 9 This study aimed to determine the factors that affect the clinical development and prognosis of HM patients treated for COVID-19. Methods This retrospective study was conducted in a tertiary hospital with a total of 1,600 beds and a hematology unit with a capacity for 24 patients. Patients and definitions Patients with a diagnosis of HM and COVID-19 who were hospitalized in the Kayseri City Training and Research Hospital between April 2020 and October 2021 were included in this retrospective study. Demographic characteristics, HM diagnoses and clinical conditions (new diagnosis, remission and recurrence) of the patients and the clinical course and treatments for COVID-19 were recorded. During the COVID-19 process, 28-day mortality cases were evaluated. The patients were divided into two groups: survivors and non-survivors and the factors predicting mortality were determined. Patients with PCR-positive SARS-CoV-2 were diagnosed with COVID-19. According to the clinical severity classification of COVID-19, patients with fever and cough, but with normal O2 saturation and no change in X-rays and/or CT scans, were classified as mild COVID-19; patients with ground glass opacities, lung consolidation and O2 saturation < 94%, as moderate COVID-19, (Is this correctly placed???) and; patients with PaO2/FiO2 < 300 mmHg and respiratory frequency > 30 breaths/min., as severe/critical COVID-19.6 Diagnosis and treatment strategies for HM were determined according to the National Comprehensive Cancer Network (NCCN).10 The COVID-19 treatment was carried out in accordance with the COVID-19 treatment management guidelines of the Ministry of Health. 11 Favipiravir is given as an antiviral drug to patients who have received treatment for COVID-19. The IL-6 inhibitor tocilizumab, steroids and high-dose steroids were used in accordance with the guidelines for patients with increased inflammatory markers. 12 Secondary bacterial infections were defined as bacterial infections occurring during the COVID-19 disease course or hospital stay (> 48 - 72 h after hospitalization).13 Secondary fungal infections were defined according to the CDC.14 Statistical analysis The statistical analysis was performed using the SPSS 22.0 (IBM Corp., Armonk, NY, USA) package program. The Shapiro-Wilk test was performed to check the normality assumption of the data. Categorical variables were expressed as numbers and percentages and Chi-square or Fisher's Exact Test analysis was used for comparisons. Variables with a p-value ≤ 0.05 were included in the multivariate logistic regression analysis. A p-value ≤ 0.05 was considered statistically significant in all analyses. Ethical approval This study was approved by the Kayseri City Hospital, Clinical Research Ethics Committee (Date: 24.12.2020 and Approval Number: 251) and was conducted in accordance with the Principles of the Declaration of Helsinki. As it was performed retrospectively by the file scanning method, an 'informed voluntary consent form' was not obtained from the patients. Results A total of 49 patients were included in the study. The clinical and demographic data of the patients are presented in Table 1 . Twenty-eight (57.1%) patients were male and the mean age was 57.90 ± 16.43 years. Of the patients, 17 (34.6%) died within 28 days of being diagnosed with COVID-19 and 32 (65.3%) of them recovered. The mean age of the non-survivor group was 67.23 (± 13.33) and of the survivor group, 48.68 (± 15.95), this difference being statistically significant (p =< 0.001). (Should this be “≤”???)At the time of admission, complaints of fever, cough and shortness of breath were similar in both groups (p = 1.000, p = 0.202 and p = 0.096, respectively). The prevalence of diabetes was 76.5% in the survivor group and 25.0% in the non-survivor group and this difference was statistically significant (p = 0.001). When the frequency of diagnosis of HM was evaluated, it was observed that 22 (44.9%) patients were followed up for lymphoma, 18 (36.7%) patients, for leukemia, and 9 (18.4%) patients, for multiple myeloma. No statistically significant difference was found between the survivor and non-survivor groups in the type of HM (p > 0.05). The HM was evaluated for disease activation status and 81.3% of the survivors and 23.5% of the non-survivors were in remission (p < 0.001). A total of 12 (24.5%) patients were HSCT recipients. Allogeneic HSCT was performed in three patients (9.1%) and autologous SCT was performed in six patients (18.8%). The frequency of patients receiving HSCT and the frequency of performing allogeneic and autologous HSCT were found to be statistically similar between survivor and non-survivor patients (p = 0.175, p = 0.646, and p = 0.397, respectively).Table 1 Demographic and clinical characteristics of patients Table 1 Survivor Non-survivor Total p n=32 (%) n=17 (%) n=49 (%) Age (mean±SD) 48.6 (±15.9) 67.2 (±13.3) 56.4 (±16.9) 0.001 Age (median, min-max) 54.5 (18.0-78.0) 71.0 (41.0-87.0) 58.0 (18.0-87.0) 0.001 Male gender 18 (56.3) 10 (58.8) 28 (57.1) 1.000 COVID-19 symptoms Fever 16 (50.0) 9 (52.9) 25 (51.0) 1.000 Cough 20 (62.5) 14 (82.4) 34 (69.4) 0.202 Shortness of breath 20 (62.5) 15 (88.2) 35 (71.4) 0.096 Comorbidities Diabetes Mellitus 8 (25.0) 13 (76.5) 21 (42.9) 0.001 Hypertension 7 (21.9) 8 (47.1) 15 (30.6) 0.104 Chronic Obstructive Pulmonary Disease - 6 (35.3) 6 (12.2) 0.002 Coronary arter disease 5 (15.6) 6 (35.3) 11 (22.4) 0.156 Hematological malignancy diagnosis Leukemia 12 (37.5) 6 (35.3) 18 (36.7) 1.000 Acute Myeloid Leukemia 4 (12.5) 3 (17.6) 7 (14.3) 0.681 Acute Lymphoblastic Leukemia 4 (12.5) 1 (5.9) 5 (10.2) 0.646 Hairy-Cell Leukemia 2 (6.3) - 2 (4.1) 0.537 Chronic Lymphocytic Leukemia 2 (6.3) 2 (11.8) 4 (8.2) 0.273 Lymphoma 12 (37.5) 10 (58.8) 22 (44.9) 0.228 Diffuse Large B Cell Lymphoma 4 (12.5) 6 (35.3) 10 (20.4) 0.075 Non- Hodgkin Lymphoma, Nos 4 (12.5) 1 (5.9) 5 (10.2) 0.646 Hodgkin Lymphoma 2 (6.3) 2 (11.8) 4 (8.2) 0.602 Marginal Zone Lymphoma 1 (3.1) 1 (2.0) 1.000 Mantle Cell Lymphoma 1 (3.1) 1 (5.9) 2 (4.1) 1.000 Multiple Myeloma 8 (25.0) 1 (5.9) 9 (18.4) 0.136 Hematological malignancy condition Remission 26 (81.3) 4 (23.5) 30 (61.2) <0.001 New diagnosis 4 (12.5) 6 (35.3) 10 (20.4) 0.075 Refractory disease - 3 (17.6) 3 (6.1) 0.037 Recurrence 2 (6.3) 4 (23.5) 6 (12.2) 0.164 Hematopoietic Stem Cell Transplantation 10 (31.3) 2 (11.8) 12 (24.5) 0.175 Allogeneic 4 (12.5) 1 (5.9) 5 (10.2) 0.646 Autolog 6 (18.8) 1 (5.9) 7 (14.3) 0.397 Vaccination Inactivated vaccine 2 (6.2) 1 (5.9) 3 (6.1) 0.380 m-RNA Vaccine 4 (12.5) - 4 (8.2) - COVID-19 severity Mild COVID-19 11 (34.4) 3 (17.6) 14 (28.6) 0.323 Moderate COVID-19 11 (34.4) 4 (23.5) 15 (30.6) 0.526 Severe/critical COVID-19 10 (31.3) 10 (58.8) 20 (40.8) 0.075 COVID-19 treatment Favipiravir 32 (100) 17 (100) 49 (100) 1.000 Corticosteroid 27 (84.4) 16 (94.1) 43 (87.8) 0.650 Duration of corticosteroid 8.5±3.5 7.7±3.3 8.2±3.4 0.439 Metilprednisolone 24 (75.0) 13 (76.5) 37 (75.5) 1.000 Daily dose mean (±sd) (mg) 55.8±19.5 69.2±13.2 60.5±18.5 0.058 Total dose mean (±sd) (mg) 362.5±301.3 410.5±326.4 379.1±307.7 0.542 Dexametasone 3 (9.4) 3 (9.4) 6 (12.2) 0.405 Daily dose mean (±sd) (mg) 13.3±4.6 10.6±4.6 12.0±4.3 0.519 Total dose mean (±sd) (mg) 34.4±9.7 47.9±83.5 39.2±12.2 0.900 Pulse- steroid 10 (31.3) 9 (52.9) 19 (38.8) 0.218 Intravenous immunoglobulin 2 (6.3) - 2 (4.1) 0.537 Tocilizumab 7 (21.9) 3 (17.6) 10 (20.4) 1.000 Outcomes Secondary Infection 4 (12.5) 11 (64.7) 15 (24.5) <0.001 Bacterial Infection 1 (3.1) 6 (35.3) 7 (14.3) 0.005 Bacterial pnemonia - 3 (17.7) 3 (6.1) Isolates Acinetobacter baumannii 1 (5.9) Klebsiella pneumonie 1 (5.9) Klebsiella oxytoca 1 (5.9) Bacteremia 1 (3.1) 3 (17.7) 4 (8.2) Enterococcus gallinarum 1 (3.1) Staphylococcus aureus 1 (5.9) Klebsiella pneumoniae 1 (5.9) Enterobacter cloacae 1 (5.9) Invasive fungal infection 2 (6.3) 2 (11.8) 4 (8.2) 0.602 Rinocerebral mucormycosis 2 (6.3) 1 (5.9) 3 (6.1) Candidemia - 1 (5.9) 1 (2.0) According to the clinical severity of COVID-19, a total of 14 (28.6%) patients were mild, 15 (30.6%) patients, moderate, and 15 (40.8%) patients, severe/critical COVID-19. There was no difference between the two groups in the classification according to the clinical severity of COVID-19 (p = 0.323, p = 0.526 and p = 0.075, respectively). The rate of corticosteroid and tocilizumab used for the treatment of covid-19 was similar in the survivor and non-survivor groups (p = 0.650 and p = 1.000, respectively). In a total of 15 patients (24.5%), secondary infection developed while being followed up for COVID-19. Seven (14.3%) of these were bacterial infections, while four (8.2%) were invasive fungal infections. Secondary bacterial infection was seen in 6 (35.3%) in non-survivors and in one (3.1%) patient from the survivors (p = 0.005). Bacterial pneumonia was observed in three patients and bacteremia was observed in four patients as the secondary bacterial infection. An Acinetobacter baumannii strain causing bacterial pneumonia and a Klebsiella pneumoniae strain had carbapenem resistance. In addition, the Klebsiella pneumoniae strain, isolated as a bacteremia agent, was resistant to carbapenem and the Staphylococcus aureus strain was resistant to methicillin. The frequency of invasive fungal infections was similar in both groups (11.8% vs. 12.5%, p = 1.000). Rhinocerebral mucormycosis was seen in three patients. The only patient with candidemia died. In multivariate regression analysis (Table 2 ), older age (OR: 1.150, p = 0.024), secondary infection (OR: 16.677, p = 0.024) were evaluated as factors predicting mortality in patients with HM. Being in remission of the HM was found to be a factor reducing mortality (OR: 0.093, p = 0.047).Table 2 Risk factors of 28-day mortality Table 2 OR (95% Cl), p Age 1.102 (1.007-1.205), 0.035 Diabetes mellitus 7.893 (0.655-95.073), 0.104 Remission of hematologic malignancy 0.093 (0.009-0.973), 0.047 Secondary infection 16.677 (1.449-191.895), 0.024 Discussion This study evaluated the clinical features and risk factors affecting the mortality of patients with HM diagnosed and followed up for COVID-19. Older age and secondary infection have been identified as risk factors that increase mortality. Diabetes was observed more frequently in patients who died, but it was not considered a risk factor. The remission of HM was found more in surviving patients and was evaluated as a factor reducing the risk of mortality. During the COVID-19 pandemic, patients with HM are a potentially high-risk population due to immunosuppressive therapies.15 The immune dysfunction of patients due to HM also affects the course of COVID-19 and there are important risk factors affecting mortality in these patients. It has been reported that older age increases mortality 1 to 13 times in COVID-19 patients without malignancy.16 , 17 In addition, infectious complications in patients with HM increase with age and older age is associated with a poor prognosis.18 There are few studies on COVID-19 patients with HM and older age has been reported as a risk factor that increases mortality and morbidity.3 , 9 According to the results of our study, it was determined that the increased mean age increased mortality 1.1 times. Diabetes and impaired glucose tolerance have been associated with poor prognosis in COVID-19 disease.19 In a meta-analysis of 30 studies and 6,452 patients, it was stated that the presence of DM led to a two-fold increased mortality risk and severe/critical clinic.20 (Is something missing here???) According to our study results, diabetes was found with a higher frequency in patients who died. The remission of HM is known to be a good prognostic factor, especially in infectious complications.21 By controlling the malignancy, it is possible to regain the immune system functions. In a study conducted in Spain, in which 367 patients with HM who had COVID-19 were evaluated, uncontrolled HM was associated with an increased 45-day mortality rate.3 In another study, conducted in the early period of the COVID-19 pandemic and which evaluated 34 patients with HM, it was found that patients in remission without active malignancy had a better survival rate. In our study, the remission of HM was similarly found to be a factor in reducing mortality due to COVID-19. Bacterial/fungal secondary infections contribute to the increased morbidity and mortality of viral respiratory infections.22 It has been reported that the frequency of infections caused by resistant bacteria has increased due to inappropriate antibiotic use and extended hospitalization in the intensive care unit.23 In addition, it has been reported that respiratory tract damage caused by SARS-CoV-2 and alveolar damage caused by the cytokine storm increase the risk of invasive fungal infection.24 Conclusion Older age, remission of HM and secondary infection were determined as prognostic factors predicting mortality in HM patients followed up with COVID-19. There are some limitations in this study. This study was retrospectively performed at a single center. The details of chemotherapy regimens could not be fully reached. In addition, the small number of groups prevented sub-analysis between different hematological malignancies and thus, prospective studies with large patient groups are needed. Ethics statement The clinical research ethics committee of Kayseri City Hospital approved this research (Date: 12/24/2020 Number: 251). Conflicts of interest The authors of this manuscript have no conflicts of interest to declare. ==== Refs References 1 Dizon DS Krilov L Cohen E Gangadhar T Ganz PA Hensing TA Clinical cancer advances 2016: annual report on progress against cancer from the American Society of Clinical Oncology J Clin Oncol. 34 9 2016 987 1011 26846975 2 Atkins S He F. Chemotherapy and beyond: infections in the era of old and new treatments for hematologic malignancies Infect Dis Clin North Am 33 2 2019 289 309 30935703 3 Piñana JL Martino R García-García I Parody R Morales MD Benzo G Risk factors and outcome of COVID-19 in patients with hematological malignancies Exp Hematol Oncol 9 2020 21 32864192 4 Siddiqi HK Mehra MR. COVID-19 illness in native and immunosuppressed states: A clinical-therapeutic staging proposal J Heart Lung Transplant 39 5 2020 405 407 32362390 5 Wong HYF Lam HYS Fong AH Leung ST Chin TW Lo CSY Frequency and distribution of chest radiographic findings in patients positive for COVID-19 Radiology 296 2020 E72 E78 32216717 6 Baj J Karakuła-Juchnowicz H Teresiński G Buszewicz G Ciesielka M Sitarz R COVID-19: specific and non-specific clinical manifestations and symptoms: the current state of knowledge J Clin Med 9 6 2020 Jun 5 1753 10.3390/jcm9061753 PMID: 32516940; PMCID: PMC7356953 32516940 7 Mathew D Giles JR Baxter AE Oldridge DA Greenplate AR Wu JE Deep immune profiling of COVID-19 patients reveals distinct immunotypes with therapeutic implications Science 2020 369 8 Yigenoglu TN Ata N Altuntas F Bascı S Dal MS Korkmaz S The outcome of COVID-19 in patients with hematological malignancy J Med Virol 93 2 2021 1099 1104 32776581 9 Passamonti F Cattaneo C Arcaini L Bruna R Cavo M Merli F Clinical characteristics and risk factors associated with COVID-19 severity in patients with haematological malignancies in Italy: a retrospective, multicentre, cohort study Lancet Haematol 7 10 2020 e737 e745 32798473 10 National Comprehensive Cancer Network (NCCN). Available from: https://www.nccn.org/. 11 COVID-19 (SARS-CoV-2 INFECTION), Management of adult patients 2020 Ministry of Health Ankara, Turkey July 31.Available from: https://www.ekmud.org.tr/files/uploads/files/COVID-19_REHBERI_ERISKIN_HASTA_TEDAVISI.pdf 12 COVID-19 (SARS-CoV-2 INFECTION) Anticytokine-Anti-Inflammatory Treatments, Coagulopathy Management 2020 Ministry of Health Ankara, Turkey November 7. Available from: https://covid19.saglik.gov.tr/Eklenti/39296/0/covid-19rehberiantisitokin-antiinflamatuartedavilerkoagulopatiyonetimipdf.pdf 13 Russell CD Fairfield CJ Drake TM Turtle L Seaton RA Wootton DG Co-infections, secondary infections, and antimicrobial use in patients hospitalised with COVID-19 during the first pandemic wave from the ISARIC WHO CCP-UK study: a multicentre, prospective cohort study Lancet Microbe 2 8 2021 Aug e354 e365 34100002 14 Fungal Diseases and COVID-19 – CDC. Available from: https://www.cdc.gov/fungal/covid-fungal.html. 15 Goldman JD Robinson PC Uldrick TS Ljungman P. COVID-19 in immunocompromised populations: implications for prognosis and repurposing of immunotherapies J Immunother Cancer 9 6 2021 Jun e002630 16 Ho FK Petermann-Rocha F Gray SR Jani BD Katikireddi SV Niedzwiedz CL Is older age associated with COVID-19 mortality in the absence of other risk factors? general population cohort study of 470,034 participants PLoS One 15 2020 e0241824 17 Kılınç Toker A Ulu Kılıç A Eren E Beştepe Dursun Z Toker İ Saatçi E Confronting a pandemic in early stages: a retrospective analysis from a pandemic hospital Hamidiye Med J 1 2020 106 118 18 de Montmollin E Tandjaoui-Lambiotte Y Legrand M Lambert J Mokart D Kouatchet A Outcomes in critically ill cancer patients with septic shock of pulmonary origin Shock 39 2013 250 254 23364436 19 Pranata R Henrina J Raffaello WM Lawrensia S Huang I. Diabetes and COVID-19: the past, the present, and the future Metabolism 121 2021 154814 20 Huang I Lim MA Pranata R. Diabetes mellitus is associated with increased mortality and severity of disease in COVID-19 pneumonia – a systematic review, metaanalysis, and meta-regression: diabetes and COVID-19 Diabetes Metab Syndr Clin Res Rev 14 2020 395 403 21 Cheng LC Yang T Kuang HH Yu S Guan LX Gu ZY Single center analysis of bloodstream infection clinical characteristics and prognosis in patients with hematological malignancies in the tropics Zhongguo Shi Yan Xue Ye Xue Za Zhi 29 1 2021 265 271 Chinese 33554832 22 Del Pozo JL. Respiratory co-and superinfections in COVID-19 Rev Esp Quimioter 34 Suppl 1 2021 Sep 69 71 10.37201/req/s01.20.2021 Epub 2021 Sep 30. PMID: 34598432 34598432 23 Rangel K Chagas TPG De-Simone SG. Acinetobacter baumannii infections in times of COVID-19 pandemic Pathogens 10 8 2021 1006 10.3390/pathogens10081006 Published 2021 Aug 10 34451470 24 Prattes J Wauters J Giacobbe DR Salmanton-García J Maertens J Bourgeois M ECMM-CAPA Study Group. Risk factors and outcome of pulmonary aspergillosis in critically ill coronavirus disease 2019 patients-a multinational observational study by the European Confederation of Medical Mycology Clin Microbiol Infect 2021 10.1016/j.cmi.2021.08.014 S1198-743X(21)00474-2	36474859	PMC9715489	NO-CC CODE	2022-12-05 23:15:27	no	Hematol Transfus Cell Ther. 2022 Dec 2; doi: 10.1016/j.htct.2022.10.001	utf-8	Hematol Transfus Cell Ther	2,022	10.1016/j.htct.2022.10.001	oa_other
==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 Elsevier Ltd. S0264-410X(22)01491-8 10.1016/j.vaccine.2022.11.067 Article Inequalities in infant vaccination coverage during the COVID-19 pandemic: a population-based study in Peru Al-kassab-Córdova Ali a⁎ Silva-Perez Claudia b Mendez-Guerra Carolina b Sangster-Carrasco Lucero b Arroyave Iván c Cabieses Báltica d Mezones-Holguin Edward ae a Universidad San Ignacio de Loyola, Centro de Excelencia en Investigaciones Económicas y Sociales en Salud, Lima, Perú b Universidad Peruana de Ciencias Aplicadas, Facultad de Ciencias de la Salud, Lima, Perú c Universidad de Antioquia, National School of Public Health, Medellin, Colombia d Universidad del Desarrollo, Programa de Estudios Sociales en Salud, Instituto de Ciencias e Innovación en Medicina, Facultad de Medicina Clínica Alemana, Santiago, Chile e Epi-gnosis Solutions, Piura, Perú ⁎ Corresponding autor at: Universidad San Ignacio de Loyola, Lima, Perú. 2 12 2022 2 12 2022 16 9 2022 6 11 2022 28 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. Objective To identify the associated factors and assess the inequalities of full vaccination coverage (FVC) among Peruvian infants aged 12–23 months during the COVID-19 pandemic in a nationally representative sample. Methods We carried out a population-based cross-sectional study based on a secondary data analysis using the 2021 Peruvian Demographic Health Survey (DHS) in infants aged 12 to 23 months. The sampling design was probabilistic, multistage, stratified, and independent at both departmental and area of residence levels. FVC was defined according to the WHO definition. We performed generalized linear models (GLM) Poisson family log link function to estimate crude (aPR) and adjusted prevalence ratios (aPR). Also, for inequality assessment, we calculated the concentration curve (CC), concentration index (CI), and Erreygerś normalized concentration index (ECI). Results We included 4,189 infants in our analysis. Nationwide, the prevalence of FVC was 66.19% (95% CI: 64.33–68). Being younger, having a mother with no education or primary education, belonging to a large family, having no access to mass media, having had six or fewer ANC visits, and having a mother whose age was under 20 at first delivery were inversely associated with FVC. Meanwhile, living in the Highlands or on the rest of the coast, and living in rural areas were directly associated with FVC. We found a pro-rich inequality in FVC based on wealth-ranked households (CI: 0.0066; ECI: 0.0175). Conclusion FVC has dropped among Peruvian infants aged between 12 to 23 months. There were several factors associated with FVC. It was more concentrated among the better-off infants, although in low magnitude. Keywords COVID-19 Vaccination Coverage Healthcare Disparities Peru Low- and middle-income countries ==== Body pmc1 Introduction Vaccination is one of the most powerful public health interventions worldwide. Vaccines are a highly cost-effective measure, especially in low- and middle-income countries (LMICs) [1], [2]. The positive effects of vaccines encompass the reduction of the incidence, severity, and mortality of different infections called vaccine-preventable diseases (VPD), thereby reducing the disease burden and the healthcare services demand [3], [4]. It is paramount to note that the major effect of vaccines is at the population level through direct and indirect protection, where vaccination coverage is essential [5]. In fact, national vaccination programs are crucial collective health strategies, particularly those aimed at children. Infantś vaccination coverage is crucial. Nevertheless, it faces numerous challenges depending on the context, especially considering the current situation due to the COVID-19 pandemic. Globally, vaccination coverage has dropped from 86% in 2019 to 83% in 2020 [6]. Moreover, a relative reduction of 7.7% was estimated for the delivery of the third dose of the diphtheria-tetanus-pertussis vaccine and 7.9% for the first dose of the measles-containing vaccine (MCV1) [7]. This reduction conditions the resurgence of VPD, particularly, in Latin American (LATAM) regions [7]. Prior to the COVID-19 pandemic, infants’ vaccination coverage was jeopardized by various factors such as antivaccine groups, and sociocultural aspects, among others [8]. Worse still, the COVID-19 pandemic had a negative impact on several public health interventions, including immunization programs. Social distancing and isolation were primary measures established by governments to mitigate the COVID-19 pandemic, challenging the delivery of vaccines to the target populations [9]. Nationwide, the temporary closure of primary healthcare centers aggravated the delivery of vaccines [10], [11]. All of this probably contributed to situations like a case of diphtheria reported in Peru after 20 years of epidemiological silence on the disease [12]. According to the above, it is pivotal to evaluate infants’ vaccination coverage during the COVID-19 pandemic. In the context of the COVID-19 pandemic, it is worthwhile to evaluate the factors associated with vaccination coverage. The World Health Organization (WHO) is concerned about fair access to safe and effective vaccines; thus, its 2030 immunization agenda focuses on improving coverage and reducing inequities worldwide[13]. Before the pandemic, studies conducted on full vaccination coverage (FVC), a WHO-standardized health indicator [14], found that it has increased over the latest decades, but the coverage in the LATAM region remained low with high-income inequality gaps [15]. Several factors such as maternal education, antenatal care (ANC) visits, knowledge, and attitudes about immunization, ethnicity, and mass media access, among others, have shown to be directly associated with FVC [16], [17], [18], [19], whereas home delivery, younger maternal age, urban residence, and no post-natal care visits, among others, were inversely associated with FVC [20], [21]. By implementing policies addressing the factors associated in the Peruvian setting, FVC would broaden. In addition, there are regional differences in FVC, which might be explained by the presence of clusters with different geographic, geopolitical, and demographic characteristics[12], [22]. Therefore, these variations should be studied when it is a priority to identify inequalities in access to vaccines. The National Vaccination Program is universally administered country-wide in healthcare centers without charge. It was last updated in 2018, and certain vaccines have been added over the years, such as the chickenpox vaccine. It includes seventeen vaccines protecting against at least 26 diseases. Its target population primarily includes infants under five years old, but it also includes pregnant women and the elderly. Its comprehensive character profiles it as one of the most complete vaccination plans in LATAM [23]. Nationwide, regional disparities and high-risk clusters for VPD [12], [22] have been reported, suggesting potential sociodemographic factors and inequalities in FVC. Despite Peru has a very comprehensive and free-of-charge immunization schedule for infants under five years of age [23], it does not ensure broad coverage. To the best of our knowledge, no scientific publications assessed FVC in the pandemic period in LATAM infants. Hence, we aimed to identify the associated factors and assess the inequalities of FVC among Peruvian infants aged 12–23 months during the COVID-19 pandemic in a sample with national representativeness. 2 Materials and methods 2.1 Study design, population, sampling, sample size, and data source We conducted a cross-sectional study based on the secondary data analysis of the 2021 Peruvian Demographic Health Survey (DHS). This is a national and continuous survey executed yearly by the National Institute of Statistics and Informatics of Peru (INEI, from the Spanish acronym). The sampling design employed was probabilistic, multistage, stratified, and independent at both departmental and area of residence levels. Peruvian DHS has national, departmental, and area of residence representativeness. In urban areas, the sampling unit included conglomerate and private dwellings, while in rural areas, it included rural census regions and private dwellings. Peru is an upper middle-income country, administratively divided into 24 departments and one constitutional province. A table describing the main socioeconomic characteristics of the departments is provided in Supplementary 1. The sample size was 36,760 dwellings: 24,100 from urban areas and 12,660 from rural areas [24]. Specifically, for our analysis, we included infants aged 12 to 23 months. We excluded those subjects without complete data regarding the variables of interest. 3 Outcome definition According to the WHO, crude FVC is defined by those infants aged one year old that have completely received: one dose of Bacille Calmette Guerin (BCG) vaccine; three doses of the combined diphtheria, tetanus, and pertussis vaccine (DTP3); three doses of the polio vaccine (Pol3); and MCV1 [14]. In the Peruvian National Vaccination Schedule, currently ongoing, DPT3 is administered with the pentavalent vaccine, which also includes the Haemophilus b conjugate and hepatitis B (HepB) vaccine. Likewise, MCV1 is administered in combination with the rubella and mumps vaccine (MMR) [23]. (See Supplementary 2) FVC was collected by reviewing infants’ vaccination cards, but in its absence, it was collected from the mothers’ verbal report. This binary variable was coded as 0 for incomplete vaccination coverage and one for FVC. 4 Exposure variables The variables were grouped into four groups: demographic, socioeconomic status (SES), socio-demographic, and delivery. Demographic variables included sex (male, female), age (in years), and region of origin (Metropolitan Lima, the rest of the coast, the Highlands, the jungle). SES included wealth index (poor, middle, rich), and mother’s educational level (no education/primary, secondary, higher). Sociodemographic variables included the area of residence (urban, rural), family size (< 4, 4–6, > 6 members), and mass media access (yes, no), which included having a TV, internet, or radio in the household. Finally, delivery-related variables included ANC visits (0–6, > 6 visits), mother’s age at first delivery (< 20, 20–34, > 34 years), and place of delivery (home, health facility). (See Figure 1 )Figure 1 Scheme of variables To assess SES, we employed the wealth index as a proxy measure, which measures households’ assets and holdings such as automobiles, home appliances, access to basic services, and access to healthcare services, among others. It places the individual interviewed on a continuous scale of wealth, which is then divided into five quintiles. It is calculated through a statistical analysis known as principal components analysis [25]. The wealth index has been previously used in other international studies, even in the Peruvian context [26], [27]. 4.1 Statistical analysis Descriptive, bivariate, and multiple regression analysis All the analyses were performed in STATA version 16.0 (Stata Corporation, College Station, TX, USA) using complex survey design modules (svy). The baseline characteristics of the categorial variables were described through absolute and relative frequencies, whereas age was described through the mean and its standard deviation (SD). For bivariate analysis, the Pearson chi-square test was used to compare the proportion of the outcome variable over each category of the independent variables, while an adjusted Wald test was used for numerical variables. All the variables were adjusted in the multiple regression model, but we also conducted models adjusting by each group of variables. Crude (aPR) and adjusted prevalence ratios (aPR) were estimated through generalized linear models (glm) Poisson family log link function. In addition, we performed a sensitivity analysis according to the source of vaccination data. Confidence intervals were computed to 95% (CI 95%), and p values < 0.05 were considered statistically significant. 5 Inequality analysis We calculated the concentration curve (CC), concentration index (CI), and Erreygers normalized concentration index (ECI) [28], [29]. The CC provides a graphical view of inequality assessing the degree of income-related inequality in the distribution of a health variable. It plots FVC (y-axis) against the wealth index (x-axis). Meanwhile, the ECI is the corrected version of the concentration index (CI), which measures the magnitude of inequality and is defined as twice the area between the CC and the line of equality (45° line). The CI ranges from -1 to +1. A positive value implies that FVC is concentrated among the richest, while a negative value implies that it is concentrated among the poorest. The correction of CI by Erreygers standardizes the uncorrected index by adjusting the CI to allow for the binary nature of the health variable. The ECI is the CI multiplied by four times the outcome of interest [30]. According to international guidelines, we reported both CI and ECI [31]. Ethical considerations Peruvian DHS is a database in the public domain[32], and it does not collect personal information that could reveal identities. Thus, it did not require approval by an ethics committee. 6 Results 6.1 Characteristics of participants A total of 4,334 infants aged 12–23 months were surveyed by the 2021 DHS. Of these, 145 (3.35%) were excluded due to incomplete data for the variables of interest. We included 4,189 infants in our analysis. The mean age of the participants was 17.46 (SD: ± 3.47) months, and 50.83% were male. Most of the population lived in urban areas (73.95%), had a medium family size (62.86%), had access to mass media (90.05%), and was born at health facilities (94%). Likewise, 32.99% and 35.48% had a poor and rich wealth index, respectively. Additionally, 55.2% of mothers were aged between 20 and 34 years old at first delivery, 72.43% had more than six ANC visits, and 35.48% had a higher educational level. The main source of vaccination data was the direct observation of vaccination cards (90.24%). (See Table 1 )Table 1 Characteristics of Peruvian infants aged 12–23 months and their FVC Variables n (%) FVC % Non-FVC p-value Demographic Age (months) < 0.001 12-15 1,524 (34.07) 54.55 45.45 16-19 1,386 (32.35) 69.4 30.6 20-23 1,424 (33.58) 74.93 25.07 Mean (SD) 17.46 (±3.47) 17.91 (±3.37) 16.57 (±3.51) < 0.001b Sex 0.0551 Male 2,212 (50.83) 67.9 32.1 Female 2,122 (49.17) 64.43 35.57 Region of origin 0.0017 Lima 537 (25.29) 61.65 38.35 Rest of the coast 1,308 (28.82) 67.55 32.45 Highlands 1,39 (27.73) 70.86 29.14 Jungle 1,096 (18.15) 63.25 36.75 Socioeconomic status Wealth index 0.5077 Poor 2,475 (32.99) 65.44 34.56 Middle 834 (31.53) 65.51 34.49 Rich 1,025 (35.48) 67.94 32.06 Mother’s education < 0.001 No education/Primary 1,484 (32.99) 60.3 39.7 Secondary 1,356 (31.53) 66.31 33.69 Higher 1,494 (35.48) 71.57 28.43 Sociodemographic Area of residence 0.4188 Urban 2,995 (73.95) 65.77 34.23 Rural 1,339 (26.05) 67.4 32.6 Family size < 0.001 Small (< 4 members) 836 (18.36) 70.98 29.02 Medium (4-6 members) 2,743 (62.86) 66.92 33.08 Large (< 6 members) 755 (18.77) 59.09 40.91 Mass media access (internet, TV, radio) 0.0026 Yes 3,826 (90.05) 67.01 32.99 No 508 (9.95) 58.78 41.22 Delivery-related ANC visits < 0.001 06 1,109 (27.57) 56.09 43.91 > 6 3,124 (72.43) 70.16 29.84 Mother’s age at first delivery < 0.001 < 20 years 1,960 (42.4) 59.79 40.21 20-34 years 2,284 (55.2) 71.39 28.61 > 34 years 90 (2.41) 59.8 40.2 Place of delivery 0.0012 Home 247 (6) 54.69 45.31 Health facility 4,040 (94) 66.98 33.02 All proportions were weighted. a Chi square test. b Adjusted Wald test. 7 Vaccination coverage Nationwide, the frequency of FVC was 66.19% (95% CI: 64.33–68) and it ranged from 51.50% to 78.30% among departments. However, when stratifying the analysis according to the source of vaccination data, FVC was 70.62% (95% CI: 68.71–72.46) in those who owned the vaccination card and 25.25% (95% CI: 20.74–30.35) in those whose mothers self-reported the vaccination data. Concerning the components of FVC, the highest coverage was from BCG (91.89%), followed by Pol3 (82.97%), DPT3 (81.76%), and MCV1 (75.51%). After stratifying by region of origin, the highest coverage of each vaccine was found in the Highlands, whereas the lowest coverage of BCG and DPT3 was found in the Lima Metropolitan area, and of Pol 3 and MCV1 in the Jungle. (See Figure 2 and Figure 3 ) Fig 3A. Figure 2 Proportion of FVC and its components among Peruvian infants aged 12–23 months amid the COVID-19 pandemic, by region of origin. BCG: Bacille Calmette Guerin vaccine. DPT3: three doses of combined diphtheria, tetanus, and pertussis vaccine. Pol3: three doses of polio vaccine. MCV1: measles-containing vaccine. Figure 3 Proportion of FVC and its components among Peruvian infants aged 12–23 months amid the COVID-19 pandemic according to departments Figure 3A Proportion of FVC according to departments. Figure 3B. Coverage of BCG vaccine according to departments. Figure 3C. Coverage of DPT3 vaccine according to departments. Figure 3D. Coverage of Pol3 vaccine according to departments. Figure 3E. Coverage of MCV1 vaccine according to departments. 8 FVC according to sociodemographic factors Most of the evaluated variables showed statistically significant associations (p < 0.05), except for sex, wealth index, and area of residence. FVC showed a higher proportion among mothers whose age at first delivery was 20–34 years old (71.39%), whose mothers had higher education (71.57%), had attended more than six ANC visits (70.16%), had a small family size (70.98%), had mass media access (67.01%), and lived in the Highlands (70.86%). Meanwhile, the lowest FVC was found among those fewer than 6 ANC visits (56.09%) and born at home (54.69%). (See Table 1) 9 Multiple regression analysis Factors associated with FVC are shown in Table 2 . In comparison with infants aged 20–23 months, infants aged 12–15 months (aPR = 0.74, 95%CI: 0.69–0.79) and 16–19 months (aPR = 0.94, 95% CI: 0.88–0.99) had less probability of having FVC. Living outside of the Lima Metropolitan area, either in the rest of the Coast (aPR = 1.10, 95%CI: 1.01–1.20) or in the Highlands (aPR = 1.15, 95%CI: 1.05–1.26), was directly associated to FVC. Having had a mother with primary or no education was associated with fewer probabilities of FVC (aPR = 0.87; 95%CI: 0.80–0.95) when compared to mothers with a higher educational level. Additionally, living in a rural setting (aPR = 1.08; 95%CI: 1.00–1.16) was directly associated with FVC, compared to living in an urban area. Other variables inversely associated were not having access to mass media (aPR = 0.89; 95%CI: 0.81–0.98), belonging to a large family (aPR=.91; 95%CI: 0.83–0.99), and having had less than six ANC visits (aPR = 0.87; 95%CI: 0.81–0.93).Table 2 Factors associated with FVC in Peruvian infants aged 12–23 months Crude model Adjusted model (c)PRa 95% CI p-value (a)PRb* 95% CI p-value Demographic Age (months) 1215 0.72 0.68-0.77 < 0.001 0.74 0.69-0.79 < 0.001 16-19 0.92 0.87-0.98 0.010 0.94 0.88-0.99 0.042 20-23 Ref Ref Ref Ref Ref Ref Sex Male Ref Ref Ref Ref Ref Ref Female 0.94 0.89-1.00 0.055 0.96 0.91-1.01 0.148 Region of origin Lima Metropolitan Ref Ref Ref Ref Ref Ref Rest of the coast 1.09 1.00-1.19 0.044 1.10 1.01-1.20 0.022 Highlands 1.14 1.05-1.25 0.001 1.15 1.05-1.26 0.001 Jungle 1.02 0.93-1.12 0.598 1.08 0.98-1.19 0.097 Socioeconomic status Wealth index Poor 0.96 0.90-1.02 0.259 0.99 0.91-1.07 0.813 Middle 0.96 0.88-1.05 0.402 0.97 0.90-1.06 0.813 Rich Ref Ref Ref Ref Ref Ref Mother’s education No education/Primary 0.84 0.79-.89 < 0.001 0.87 0.80-0.95 0.001 Secondary 0.92 0.87-.98 0.016 0.94 0.88-1.01 0.120 Higher Ref Ref Ref Ref Ref Ref Sociodemographic Area of residence Urban Ref Ref Ref Ref Ref Ref Rural 1.02 0.96-1.08 0.414 1.08 1.00-1.16 0.035 Family size Small (< 4 members) Ref Ref Ref Ref Ref Ref Medium (4-6 members) 0.94 0.88-1-00 0.071 0.97 0.91-1.03 0.364 Large (> 6 members) 0.83 0.76-0.75 < 0.001 0.91 0.83-0.99 0.037 Mass media access (internet, TV, radio) Yes Ref Ref Ref Ref Ref Ref No 0.87 0.79-0.96 0.006 0.89 0.81-0.98 0.024 Delivery-related ANC visits 0-6 0.79 0.74-0.85 < 0.001 0.87 0.81-0.93 < 0.001 < 6 Ref Ref Ref Ref Ref Ref Mother’s age at first delivery < 20 years 0.83 0.79-0.88 < 0.001 0.89 0.84-.95 0.001 20-34 years Ref Ref Ref Ref Ref Ref > 34 years 0.83 0.66-1.06 0.144 0.88 0.72-1.09 0.269 Place of delivery Health facility Ref Ref Ref Ref Ref Ref Home 0.81 0.70-0.94 0.005 0.88 0.76-1.02 0.100 *Multiple regression based on generalized linear model (glm) Poisson family log link function. a (c)PR: crude prevalence ratio; b(a)PR adjusted prevalence ratio The multiple regression adjusted by groups of variables is shown in Supplementary 3. Likewise, when stratifying the analysis according to the source of vaccination data, the associations remained similar to those who owned the vaccination card, except for the area of residence, family size, and mass media access, which became not statistically significant. On the other hand, almost all variables were not statistically significant to those whose vaccination data was obtained from the self-report of the mother (See Supplementary 4). 10 Inequality analysis At a national level, there was a pro-rich distribution of FVC among Peruvian infants (CI: 0.0066; ECI: 0.0175) (Figure 4A). When stratifying the analysis by departments, the highest values were found in Ucayali (CI: 0.13231; ECI: 0.30013), San Martin (CI: 0.11447; ECI: 0.29993), and Loreto (CI: 0.07555; ECI: 0.16199). Meanwhile, the lowest values were found in Puno (CI: -0.09839; ECI: -0.21532), Huánuco (CI: -0.04295; ECI: -0.13161), and Ayacucho (CI: -0.09287; ECI: -0.03418) (Figure 4 B).Figure 4 Concentration curve of FVC among Peruvian infants aged 12–23 months amid the COVID-19 pandemic Figure 4A Concentration curve of FVC nationwide Figure 4B. Concentration curve of FVC according to departments 11 Discussion After a high FVC during the 2010–2019 period (albeit stagnant) [22], there was a substantial decline evidenced in 2021 among Peruvian infants. In fact, according to our findings, at least three out of ten Peruvian infants had incomplete vaccination coverage in 2021. Certain factors were inversely associated with FVC, such as being younger, having a mother with no education or primary, belonging to a large family, having no access to mass media, having had six or fewer ANC visits, and having a mother whose age at first delivery was before the age of 20. Meanwhile, living outside Lima, either in the Highlands or on the rest of the coast, and living in rural areas were directly associated with FVC. Furthermore, a pro-rich inequality in FVC was unveiled, albeit low in magnitude. Our results are based on data measured throughout 2021; thus, our population included infants whose first year of age coincided with the first year of the pandemic, depicting the deleterious effect of the pandemic on the childhood vaccination program at a national level. Therefore, our study points to the importance of an in-depth understanding of the factors associated with FVC in the Peruvian setting to implement timely interventions. 12 Comparison with previous studies Nationwide, FVC was 66.19% among infants aged 12 to 23 months in 2021, which represents a decrease of 7.97% from 2019 (pre-pandemic era) [22]. Several studies conducted in the post-pandemic era have also revealed a marked decrease in routine vaccination coverage among infants and a slow recovery afterwards [33], [34], [35]. We did not find studies measuring FVC with national data in the post-pandemic period. Our study emerges as one of the first assessing FVC during the pandemic. Unfortunately, FVC in Peru, an upper middle-income country, in 2021 (post-pandemic era) are comparable to coverages reported in certain low-income countries in the pre-pandemic era [18], [36]. Hence, arduous efforts are needed at various levels to prevent vaccination coverage from sinking into suboptimal coverage. Our results revealed that maternal education and ANC visits were significantly associated with FVC among Peruvian infants. Like other studies [17], [19], we found that having a mother with a higher education level was associated with a higher likelihood of FVC among infants when compared to not having any education or primary. Educational interventions that have been shown to be effective in promoting childhood vaccination would be more beneficial to women with a higher educational level [37]. FVC was also associated with attending at least one ANC visit, and its probability increased when attending more than six visits, as seen in the previous studies [18], [19]. Region of origin would play a key role in having FVC. It was observed that infants living on the rest of the coast or in the Highlands were more likely to be fully immunized in comparison to infants from Metropolitan Lima. The geographic disparities mentioned above agree with studies from other LMICs [17], [20], [38] and others from Peru [12]. Other diseases affecting infants have also shown a regional pattern of inequality countrywide [39]. On the other hand, contrary to several studies [18], [21], [38], our study found that living in rural areas was associated with a higher probability of being fully vaccinated. We found a pro-rich distribution of FVC on ranked-wealth households, albeit of low magnitude. Likewise, several studies have elucidated a pro-rich and pro-urban inequality concerning vaccination coverage [41]. Moreover, another study revealed inequality in such coverage according to maternal educational status [42]. 13 Explanation and plausibility of the results The decrease in FVC could be partially explained by the detrimental effect exerted by the COVID-19 pandemic as it led to lockdowns and restrictions on freedom of movement, temporary disruptions in routine medical care delivery, and avoidance of medical care centers [10], [43]. In addition, there were significant regional differences in FVC, especially as it was higher in regions of origin other than the capital city of Peru (Lima). As foreseen, it may be related to the fragmentation and segmentation of the Peruvian healthcare system [44] and, in turn, centralization. As well, there are wide gaps in health insurance coverage [45] and access to healthcare services [46]. To the best of our knowledge, this is the first study in Peru assessing FVC and its associated factors among infants after the beginning of the pandemic period. This must be highlighted as factors related to vaccination coverage could behave differently since the origin of the pandemic. As expected, the older the child the more chances they have of being between 12 and 23 months. This direct association is explained by the fact that each month that passes, the chances of receiving more vaccines and, subsequently, of completing the vaccination schedule increase. Indeed, by the age of 24 months, most vaccines are administered in accordance with the Peruvian immunization schedule [23]. This could mean that vaccination strategies and campaigns (plus promotion and communication actions) are successful over time and greatly reduce the risk associated with under-vaccination in the first months of life. Both the mother’s education and antennal care visits would improve the vaccination status in tandem. Formal education could provide satisfactory knowledge concerning infants’ preventive health, including crucial information about immunization. We believe that highly educated women may have a better understanding during each ANC visit [47]. This finding could be explained by active health promotion through preventive care during each visit since pregnant women receive their respective scheduled immunizations. In fact, the Ministry of Health (MINSA, from the Spanish acronym) sets a minimum of six ANC visits to ensure maternal and child well-being, which, according to our results, is independently associated with FVC [48]. The Peruvian government should guarantee adequate child healthcare indirectly through strategies toward delivery-related factors. The regional gaps in FVC could be related to multiple geographically distributed factors, as Peru is a multicultural country where various beliefs and customs can influence vaccination intention, but it could also be due to rugged geography where the presence of the health system is burdensome. A recent study reported the presence of high-risk clusters of low vaccination coverage throughout the national territory [12]. Overall, these geographic inequalities should be taken into consideration to target those regions and strengthen the vaccination strategies. In our study, we found that infants living in rural areas were more likely to be fully vaccinated than those in urban areas, which contrasts with several studies from other LMICs [18], [21], [40]. This relationship can be explained by some factors inherent in the Peruvian context. Conditional cash transfer programs, aimed at poor people living mainly in rural areas, provide economic incentives to those families whose children have completed the vaccination schedule [49]. As well, people living in urban areas are more exposed to the media, which is nowadays a source of misinformation on vaccination, thereby generating distrust in the urban population and, consequently, boosting anti-vaccine groups [50]. All in all, this finding is in line with the coverage found in the Highlands, which is significantly higher than in other regions. Many areas of the Highlands are rural and lack mass media access. On the contrary, the Lima Metropolitan area (the capital of Peru), which is mostly urbanized and home to a third of the Peruvian population, had the lowest vaccination coverage. FVC was more concentrated among the better-off infants in Peru, although of low magnitude. The economic support programs such as JUNTOS could contribute to the absence of major inequalities. This is a conditional cash transfer program founded by the Ministry of Economy. The Ministry of Development and Social Inclusion provides money to mothers who, among other requirements, have their infants vaccinated [49]. Conversely, as MINSA faces shortages, few of the most well-to-do families opt for acquiring vaccines privately, which may generate gaps. Considering this, joint efforts must be continued to maintain this scenario and to continue tackling inequalities. 14 Strengths and limitations Our study must be interpreted considering its limitations. First, due to the cross-sectional design of our study, it was not possible to establish causality. Second, almost 10% of the FVC information was obtained from the mothers’ reports, from which social desirability bias and recall bias may arise. However, we stratified the analysis by the source of vaccination status data and, as stated by Modi RN et al., to enhance coverage and the effect of vaccines, the mothers’ recall could be used, particularly in LMICs [51]. Third, due to the lack of a direct measurement of SES in the Peruvian DHS (such as income, expenditure, or consumption), we used an asset-based wealth index as a proxy measurement instead, which is suitable for inequality studies in the absence of a direct measure [52]. Fourth, as it was a secondary data analysis, it was not possible to include other variables that would be interesting to analyze. Fifth, our study sought to assess the factors associated to FVC amid the COVID-19 pandemic, instead of comparing vaccination coverages with the pre-pandemic period. On the other hand, our study has several strengths. It was based on a large sample size, and great statistical power was obtained. Also, the database is nationally representative. Even though the FVC definition does not include all the vaccines administered by the National Vaccination Program, such as pneumococcal and rotavirus vaccines [23], FVC is a WHO standardized definition in several studies [14], [53], which confers greater comparability to our study. Indeed, it was the most employed definition according to a scoping review on vaccination assessments with DHS data [53]. Our study is the first to address infants’ vaccination coverage countrywide after the beginning of the pandemic in Peru and could be useful as a basis to redirect health policies. Public health implications and recommendations for future studies It is of the utmost importance that strategies be based on scientific evidence to rationalize the existing health resources. Indeed, our article may serve as a basis to redirect the ongoing national vaccination program. It is necessary to continue providing universal access to vaccination so that it remains equitable between different socioeconomic levels. Multidisciplinary and inter-institutional efforts are needed to tackle the dramatic drop in FVC. By articulating the institutions, social programs may be boosted, and better vaccination strategies could be designed. It is not sufficient to offer a comprehensive vaccination scheme; it is vital to administer it efficiently. Besides, we encourage the implementation of quantitative and qualitative studies that assess the determinants of poor adherence to vaccination in the Peruvian context. Nonetheless, access to electronic records is limited, which hinders research. Thus, it is of paramount importance to improve the collection of primary data, which might be possible through the unification and articulation of the information sources of the MINSA. Conclusion FVC has dropped to 66,19% among Peruvian infants aged 12 to 23 months. Infants’ age, their mothers’ education, ANC visits, region of origin, and source of vaccination data have been associated with FVC. A pro-rich inequality in FVC was found, but low in magnitude as vaccines are universal, administered free of charge, and in turn, broadly accessible. Further studies should closely monitor FVC in the upcoming years in Peru. In addition, we encourage the measurement of household incomes rather than assets in future surveys as it provides a better figure for SES. Author contributions AAC and EMH conceptualized and designed the study. Data cleaning, merging and statistical analysis were performed by AAC. AAC, CSP, CMG, LSG and EMG wrote the first draft of the manuscript. EMH, IAZ and BC supervised and edited the manuscript. All authors contributed to data interpretation, revision and approvement of the final version of the manuscript. 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. Data availability Data will be made available on request. Acknowledgments None. ==== Refs References 1 D’Angiolella L.S. Lafranconi A. Cortesi P.A. Rota S. Cesana G. Mantovani L.G. Costs and effectiveness of influenza vaccination: a systematic review Ann Ist Super Sanita 54 2018 49 57 10.4415/ANN_18_01_10 29616674 2 Haider S. Chaikledkaew U. Thavorncharoensap M. Youngkong S. Islam M.A. Thakkinstian A. Systematic Review and Meta-Analysis of Cost-effectiveness of Rotavirus Vaccine in Low-Income and Lower-Middle-Income Countries. Open Forum Infect Dis 2019 6 10.1093/OFID/OFZ117 3 McGovern M.E. Canning D. Vaccination and All-Cause Child Mortality From 1985 to 2011: Global Evidence From the Demographic and Health Surveys Am J Epidemiol 182 2015 791 10.1093/AJE/KWV125 26453618 4 Andre F.E. Booy R. Bock H.L. Clemens J. Datta S.K. John T.J. Vaccination greatly reduces disease, disability, death and inequity worldwide Bull World Health Organ 86 2008 140 10.2471/BLT.07.040089 18297169 5 Mallory M.L. Lindesmith L.C. Baric R.S. Vaccination-induced herd immunity: Successes and challenges J Allergy Clin Immunol 142 2018 64 66 10.1016/j.jaci.2018.05.007 29803799 6 World Health Organization. Immunization coverage 2021. https://www.who.int/news-room/fact-sheets/detail/immunization-coverage (accessed June 2, 2022). 7 Causey K. Fullman N. Sorensen R.J.D. Galles N.C. Zheng P. Aravkin A. Estimating global and regional disruptions to routine childhood vaccine coverage during the COVID-19 pandemic in 2020: a modelling study Lancet 398 2021 522 534 10.1016/S0140-6736(21)01337-4/ATTACHMENT/E2A80C09-5727-4E6E-864E-1D8F7F0DDD15/MMC1.PDF 34273292 8 MacDonald N.E. Eskola J. Liang X. Chaudhuri M. Dube E. Gellin B. Vaccine hesitancy: Definition, scope and determinants Vaccine 33 2015 4161 4164 10.1016/J.VACCINE.2015.04.036 25896383 9 Hungerford D. Cunliffe N.A. Coronavirus disease (COVID-19) – impact on vaccine preventable diseases Eurosurveillance 25 2020 2000756 10.2807/1560-7917.ES.2020.25.18.2000756/CITE/PLAINTEXT 32400359 10 Atamari-Anahui N. Conto-Palomino N.M. Pereira-Victorio C.J. Atamari-Anahui N. Conto-Palomino N.M. Pereira-Victorio C.J. Actividades de inmunización en el contexto de la pandemia por la COVID-19 en Latinoamérica Rev Peru Med Exp Salud Publica 37 773–5 2020 11 Herrera-Añazco P. Uyen-Cateriano A. Mezones-Holguin E. Taype-Rondan A. Mayta-Tristan P. Malaga G. Some lessons that Peru did not learn before the second wave of COVID-19 Int J Health Plann Manage 36 2021 995 998 10.1002/hpm.3135 33595137 12 Mezones-Holguin E. Al-kassab-Córdova A. Maguiña J.L. Rodriguez-Morales A.J. Vaccination coverage and preventable diseases in Peru: Reflections on the first diphtheria case in two decades during the midst of COVID-19 pandemic Travel Med Infect Dis 40 2021 101956 10.1016/J.TMAID.2020.101956 13 World Health Organization. Immunization Agenda 2030: A Global Strategy To Leave No One Behind. 2020. 14 World Health Organization. Full immunization coverage among one-year-olds (%) (Health Equity Monitor) n.d. https://www.who.int/data/gho/indicator-metadata-registry/imr-details/3317 (accessed June 2, 2022). 15 Colomé-Hidalgo M, Campos JD, de Miguel ÁG. Monitoring inequality changes in full immunization coverage in infants in Latin America and the Caribbean. Rev Panam Salud Pública 2020;44. 10.26633/RPSP.2020.56. 16 Kawakatsu Y. Honda S. Individual-, family- and community-level determinants of full vaccination coverage among children aged 12–23 months in western Kenya Vaccine 30 2012 7588 7593 10.1016/J.VACCINE.2012.10.037 23102973 17 Kibreab F, Lewycka S, Tewelde A. Impact of mother’s education on full immunization of children aged 12-23 months in Eritrea: population and health survey 2010 data analysis. BMC Public Health 2020;20. 10.1186/S12889-020-8281-0. 18 Biset G. Woday A. Mihret S. Tsihay M. Full immunization coverage and associated factors among children age 12–23 months in Ethiopia: systematic review and meta-analysis of observational studies Hum Vaccin Immunother 17 2021 2326 2335 10.1080/21645515.2020.1870392 33760689 19 Afolabi R.F. Salawu M.M. Gbadebo B.M. Salawu A.T. Fagbamigbe A.F. Adebowale A.S. Ethnicity as a cultural factor influencing complete vaccination among children aged 12–23 months in Nigeria Hum Vaccin Immunother 17 2021 2008 2017 10.1080/21645515.2020.1870394 33605835 20 Moran E.B. Wagner A.L. Asiedu-Bekoe F. Abdul-Karim A. Schroeder L.F. Boulton M.L. Socioeconomic characteristics associated with the introduction of new vaccines and full childhood vaccination in Ghana, 2014 Vaccine 38 2020 2937 2942 10.1016/J.VACCINE.2020.02.065 32139314 21 Budu E. Seidu A.A. Agbaglo E. Armah-Ansah E.K. Dickson K.S. Hormenu T. Maternal healthcare utilization and full immunization coverage among 12–23 months children in Benin: a cross sectional study using population-based data. Arch Public Health 2021 79 10.1186/S13690-021-00554-Y 22 Al-kassab-Córdova A. Silva-Perez C. Maguiña J.L. Spatial distribution, determinants and trends of full vaccination coverage in children aged 12–59 months in Peru: a subanalysis of the Peruvian Demographic and Health Survey BMJ Open 2022 e05211 23 Ministerio de Salud. Norma técnica que establece el Esquema Nacional de Vacunación. Lima: 2018. 24 INEI. Informe Principal de la Encuesta Demográfica y de Salud Familiar - ENDES 2021. Lima: 2022. 25 Rutstein SO, Johnson K. The DHS wealth index . DHS Comparative Reports. No. 6. Calverton: 2004. 26 Guerrero-Díaz D.V. Hernández-Vásquez A. Montoya-Rivera W.C. Rojas-Roque C. Chacón Díaz M.A. Bendezu-Quispe G. Undiagnosed hypertension in Peru: analysis of associated factors and socioeconomic inequalities, 2019 Heliyon 7 2021 e07516 34296015 27 Shifti D.M. Chojenta C. Holliday E.G. Loxton D. Socioeconomic inequality in short birth interval in Ethiopia: a decomposition analysis BMC Public Health 20 2020 1504 10.1186/s12889-020-09537-0 33023567 28 Jann B. Estimating Lorenz and concentration curves Stata J 16 2016 837 866 10.1177/1536867X1601600403 29 O’Donnell O. O’Neill S. Van Ourti T. Walsh B. conindex: Estimation of concentration indices Stata J 16 2016 112 138 10.1177/1536867X1601600112 27053927 30 Erreygers G. Correcting the concentration index J Health Econ 28 2009 504 515 10.1016/J.JHEALECO.2008.02.003 18367273 31 Erreygers G. Van Ourti T. Measuring socioeconomic inequality in health, health care and health financing by means of rank-dependent indices: A recipe for good practice J Health Econ 30 2011 685 694 10.1016/J.JHEALECO.2011.04.004 21683462 32 INEI. ENCUESTA DEMOGRÁFICA Y DE SALUD FAMILIAR - ENDES 2021. Microdatos 2021. http://iinei.inei.gob.pe/microdatos/. 33 Silveira M.M. Conrad N.L. Leivas Leite F.P. Effect of COVID-19 on vaccination coverage in Brazil J Med Microbiol 70 2021 001466 10.1099/JMM.0.001466/CITE/REFWORKS 34 Ji C. Piché-Renaud P.P. Apajee J. Stephenson E. Forte M. Friedman J.N. Impact of the COVID-19 pandemic on routine immunization coverage in children under 2 years old in Ontario, Canada: A retrospective cohort study Vaccine 40 2022 1790 1798 10.1016/J.VACCINE.2022.02.008 35164987 35 SeyedAlinaghi SA, Karimi A, Mojdeganlou H, Alilou S, Mirghaderi SP, Noori T, et al. Impact of COVID‐19 pandemic on routine vaccination coverage of children and adolescents: A systematic review. Heal Sci Reports 2022;5. 10.1002/HSR2.516. 36 Tesema G.A. Tessema Z.T. Tamirat K.S. Teshale A.B. Complete basic childhood vaccination and associated factors among children aged 12–23 months in East Africa: a multilevel analysis of recent demographic and health surveys BMC Public Health 20 2020 1837 10.1186/s12889-020-09965-y 33256701 37 Saeterdal I. Lewin S. Austvoll-Dahlgren A. Glenton C. Munabi-Babigumira S. Interventions aimed at communities to inform and/or educate about early childhood vaccination Cochrane Database Syst Rev 2014 10.1002/14651858.CD010232.pub2 38 Clouston S. Kidman R. Palermo T. Social inequalities in vaccination uptake among children aged 0–59 months living in Madagascar: an analysis of Demographic and Health Survey data from 2008 to 2009 Vaccine 32 2014 3533 3539 10.1016/J.VACCINE.2014.04.030 24814558 39 Al-Kassab-Córdova A. Mendez-Guerra C. Quevedo-Ramirez A. Espinoza R. Enriquez-Vera D. Robles-Valcarcel P. Rural and urban disparities in anemia among Peruvian children aged 6–59 months: a multivariate decomposition and spatial analysis Rural Remote Health 22 6936 2022 40 Debie A. Lakew A.M. Tamirat K.S. Amare G. Tesema G.A. Complete vaccination service utilization inequalities among children aged 12–23 months in Ethiopia: A multivariate decomposition analyses Int J Equity Health 19 2020 65 10.1186/s12939-020-01166-8 32398089 41 Restrepo-Méndez M.C. Barros A.J.D. Wong K.L.M. Johnson H.L. Pariyo G. França G.V.A. Inequalities in full immunization coverage: Trends in low-and middle-income countries Bull World Health Organ 94 2016 794 805A 10.2471/BLT.15.162172 27821882 42 Acharya K. Paudel Y.R. Dharel D. The trend of full vaccination coverage in infants and inequalities by wealth quintile and maternal education: analysis from four recent demographic and health surveys in Nepal BMC Public Health 19 2019 1673 10.1186/s12889-019-7995-3 31830944 43 Czeisler MÉ, Marynak K, Clarke KEN, Salah Z, Shakya I, Thierry JM, et al. Delay or Avoidance of Medical Care Because of COVID-19–Related Concerns — United States, June 2020. MMWR Morb Mortal Wkly Rep 2022;69:1250–7. 10.15585/MMWR.MM6936A4. 44 Sanchez-Moreno F. The national health system in Peru Rev Peru Med Exp Salud Publica 31 2014 747 753 25597729 45 Mezones-Holguín E. Amaya E. Bellido-Boza L. Mougenot B. Murillo J.P. Villegas-Ortega J. Health insurance coverage: the peruvian case since the universal insurance act Rev Peru Med Exp Salud Publica 36 196–206 2019 46 Mezones-Holguín E. Solis-Cóndor R. Benites-Zapata V.A. Garnica-Pinazo G. Marquez-Bobadilla E. Tantaleán-Del-Águila M. Institutional differences in the ineffective access to prescription medication in health care centers in Peru: analysis of the National Survey on User Satisfaction of Health Services (ENSUSALUD 2014) Rev Peru Med Exp Salud Publica 33 2016 205 214 27656918 47 Miller L.C. Joshi N. Lohani M. Rogers B. Mahato S. Ghosh S. Women’s education level amplifies the effects of a livelihoods-based intervention on household wealth, child diet, and child growth in rural Nepal Int J Equity Health 16 2017 183 10.1186/s12939-017-0681-0 29047376 48 de Salud M. Norma técnica de salud para la atención integral de salud materna Lima 2013 49 Social M de D e I. Programa Nacional de Apoyo Directo a los más Pobres - JUNTOS 2022. https://www.gob.pe/juntos. 50 Germani F. Biller-Andorno N. The anti-vaccination infodemic on social media: A behavioral analysis PLoS One 16 2021 e0247642 33657152 51 Modi R.N. King C. Bar-Zeev N. Colbourn T. Caregiver recall in childhood vaccination surveys: Systematic review of recall quality and use in low- and middle-income settings Vaccine 36 2018 4161 4170 10.1016/j.vaccine.2018.05.089 29885771 52 McKenzie D.J. Measuring inequality with asset indicators J Popul Econ 18 2005 229 260 10.1007/S00148-005-0224-7 53 Shenton L.M. Wagner A.L. Ji M. Carlson B.F. Boulton M.L. Vaccination assessments using the Demographic and Health Survey, 2005–2018: a scoping review BMJ Open 10 2020 e039693	36509638	PMC9715490	NO-CC CODE	2022-12-03 23:20:16	no	Vaccine. 2022 Dec 2; doi: 10.1016/j.vaccine.2022.11.067	utf-8	Vaccine	2,022	10.1016/j.vaccine.2022.11.067	oa_other
==== Front Ann Tour Res Ann Tour Res Annals of Tourism Research 0160-7383 1873-7722 Elsevier Ltd. S0160-7383(22)00173-6 10.1016/j.annals.2022.103522 103522 Article Changes in tourist mobility after COVID-19 outbreaks☆ Yu Ling a Zhao Pengjun ab⁎ Tang Junqing a Pang Liang a a School of Urban Planning and Design, Peking University Shenzhen Graduate School, Shenzhen, 518055, China b School of Urban and Environmental Sciences, Peking University, Beijing, 100871, China ⁎ Corresponding author at: School of Urban and Environmental Sciences, Peking University, Beijing 100871, China. 2 12 2022 1 2023 2 12 2022 98 103522103522 18 4 2022 7 11 2022 13 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. We comparatively examined tourist mobility changes in the entire country and explicitly covered two distinct waves of COVID-19 outbreaks, based on mobile phone data from 277.15 million tourists from 2019 to 2021 in China. The results show that domestic tourism in Beijing was even higher after the pandemic than prior to it. In addition, we found that female and elderly groups had a slower recovery after the first wave, whereas this was the opposite one year later, after the second wave. Additionally, wealthier, larger cities were notably hit the hardest. Overall, our findings provide a better understanding of tourism management in public health crises and policy-making during post-pandemic recovery and for future outbreaks. Graphical abstract Unlabelled Image Keywords The COVID-19 pandemic Tourist mobility Human travel behavior Big data China Handling Editor: Yang Yang ==== Body pmcIntroduction Tourism plays an indispensable role in the United Nations Sustainable Development Goals (World Tourism Organization, 2022a). As the fastest and largest growing sector during the last few decades, tourism has become a key driver for economic growth and social development (Lenzen et al., 2018; Rastegar et al., 2021). According to the World Travel & Tourism Council (WTTC) report (2021), tourism created 10.6 % of all jobs (334 million) worldwide, contributing to 10.4 % of the global GDP (US$9.2 trillion) prior to the pandemic in 2019. For many countries, domestic tourism contributes a significant portion to the tourism sector. For example, in China, tourism's contribution to the country's economic GDP was $1.72 trillion, accounting for 11.05 % of the total GDP in 2019 (Ministry of Culture and Tourism of the People's Republic of China, 2020). All over the world, tourist mobility has been significantly affected by the ongoing COVID-19 pandemic (Gössling et al., 2020; He et al., 2022; Sharma et al., 2021; Uglis et al., 2022). According to United Nations World Tourism Organization (UNWTO) data (2021), global international arrivals dropped by 74 % in 2020 compared with 2019, leading to a potential $2.4 trillion loss in international tourism revenue (United Nations News, 2021). In many countries, assessing tourist mobility changes due to COVID-19 is important for reaching sustainable development goals and implementing policies for post-pandemic economic recovery (Guerriero et al., 2020; Matsuura & Saito, 2022). However, tourist mobility changes following the outbreak of COVID-19 remain unclear due to a lack of accurate data from before and after the outbreaks. In addition, due to the ongoing pandemic's multiple waves and variants appearing successively (Callaway, 2021), the majority of countries have experienced recurring outbreaks (Aleta et al., 2020). During the past two years, tourism has experienced continuous ‘rises and falls’ that have become the new norm. Thus, there is a current need to explore the changes in tourists' mobility after multiple outbreak waves using mobile phone data. Data-driven research strongly hinges on the quality of data. The cell phone mobility data used in this study has benefits regarding long-term (i.e., three years), high temporal resolution (i.e., daily basis), and large-scale (i.e., all over China) human mobility behavior (Huang et al., 2022; Kar et al., 2021; Levin et al., 2021), which enables us to comprehensively understand the changes in tourist mobility before and after COVID-19 outbreaks. Here, we take Beijing as the destination and compare two waves during the same months in a row - during the first wave at the start of 2020, and during the second wave at the start of 2021 - by using three years mobile phone data from 277.15 million tourists. The data span the pre- and post-pandemic periods, covering 254 cities in China and capturing the daily inter-city tourist movements with socio-demographic information. China's capital city, Beijing, was selected as the case study to answer these questions, since it suffered from both waves during the same months in 2020 and 2021, creating an effective chance of comparing the first- and second-wave impacts on tourist mobility. As well, Beijing is the center of the country's economy and culture and is one of the main domestic tourist-friendly destinations. Furthermore, major cities are more likely to experience recurring outbreaks due to extensive human mobility. Three main research questions will be addressed in this exploratory study:• RQ1: How did overall domestic tourist mobility change in response to COVID-19 during different outbreak waves? • RQ2: How did such changes vary among different socio-demographic tourists? • RQ3: How did such changes differ from one city to another? The main contributions of this study can be summarized as follows: Firstly, by using unparalleled mobile phone data, we monitor the dynamic changes in China's tourist mobility in response to COVID-19 during the whole outbreak, as well as during multiple waves. Secondly, we reveal the social demographic disparities and spatial heterogeneity in tourist mobility changes after the outbreaks, which reveals information about tourism equality. Thirdly, we offer comprehensive and national insights for the tourism sector and for stakeholders in order to provide targeted support for future tourism crises such as public health emergencies. As well, this offers timely knowledge for evidence-based decision-making for future waves and can guide equitable tourism recovery in the post-COVID-19 era. Literature review COVID-19 impacts on tourist mobility Mobility is an important part of tourism (Shoval & Isaacson, 2007). Tourist mobility involves a sequence of the spatial movement of tourists (Hannam et al., 2014; Hardy et al., 2020), and is the key issue within tourism geographies (Lin et al., 2020). Understanding and modeling how tourist mobility in time and space plays an important role in tourism planning and management, such as the ways in which tourism facilities and routes design, tourism destination marketing, and management strategies (Zheng et al., 2022). The main topics in tourist mobility studies include monitoring and modeling spatio-temporal patterns and heterogeneity (Han et al., 2021; Liu et al., 2019), revealing factors that affect tourist movement and mobility (Henok, 2021; Jin et al., 2019), and predicting future tourist mobility trends (Mertzanis & Papastathopoulos, 2021; Zheng et al., 2017). Tourist mobility is easily influenced by external political and crisis events (Jin et al., 2019; Zhou et al., 2021). At the beginning of the COVID-19 outbreak, researchers mainly captured how tourist mobility affected the spread of the virus because mobile people were regarded as potential vectors of viral transmission (Iaquinto, 2020). With COVID-19 continuing, a growing number of studies have explored how tourist behaviors have changed (i.e., tourism desire, travel behavior, consumption patterns) (Kock et al., 2020; Ren et al., 2022), and how these changes have been impacted by government-controlled policies (Collins-Kreiner & Ram, 2021; Zha et al., 2022) and tourists' psychological factors (Arbulú et al., 2021). These studies can be divided into two mainstream categories according to the research paradigm: data-driven research, which highly relies on data availability and computational modeling, and explanatory research, which is designed for causal and mechanism analyses that identify the driving force behind tourist mobility and behavior changes. Most studies are based on statistics such as by using the government's official tourism statistics, one study compared the volume and structural changes of the flow of tourism in Finland and Estonia from 2019 to 2020 (Ivanov et al., 2021). Some studies have tried to use big data to monitor how COVID-19 has impacted tourist mobility behavior. For example, based on social media comment data using the travel contact trajectory and spatial trajectory approach, Gao et al. (2021) explored the changes in urban tourists' spatial behaviors before, during, and after the pandemic in Nanjing, China. Another study using Google Destination Insights data examined how moderating distance factors (i.e., geographic, cultural, economic, social, political) affected the relationship between COVID-19 cases and origin-destination countries' bilateral tourism demands (Yang et al., 2022). Given that there are a collection of factors that influence tourist dynamics, unpacking the mechanism of COVID-19's impact on tourist mobility is a long-term challenge. Previous studies have mainly been based on related theories, such as travel risk perception (Neuburger & Egger, 2021) and tourist personalities (Morar et al., 2021), creating an empirical analysis of the determinant factors that might influence the tourist mobility response to the pandemic. One study used the tourist trust, travel constraint, and extended theory of planned behavior as a theoretical basis to explain the influence of travel promoting, restricting, and attitudinal factors on tourist travel decisions (Shin et al., 2022). Data and tools used in tourist mobility studies Previous studies on tourist mobility involve various types of data and spatial analysis methods (i.e., complex network, geoinformatics) (Kádár & Gede, 2021). Over the last few decades, tourism geography scholars have mainly used actively solicited tourists' travel behavior data, including, for example, the self-reported questionnaire survey data (Han et al., 2021), and GPS data requires that tourists carry GPS sensors during their travels (Liu et al., 2022). However, because tourist mobility information is collected via actively collected, these studies are restricted to relatively small samples and at the micro spatial scale (i.e., within a destination) (Domenech et al., 2020; Zheng et al., 2019). Additionally, statistical data have also been widely used in tourist mobility studies (Kulshrestha et al., 2020), but this data has had low temporal and spatial precision issues. In recent years, the fast development of tracking and big data technologies (i.e., machine learning models) (Hardy & Aryal, 2020) have allowed us to obtain massive positioning information of tourists' travel behaviors (i.e., mobile phone signaling data, smart-card data, social media check-ins data) (Chen et al., 2021; Türk et al., 2021), attracting the greater interests of tourism geographers. These data are passively generated without the users' awareness instead of being actively collected like survey data (Schmücker & Reif, 2022). Mobile phone positioning data has the advantage of generating real-time information about tourists with high temporal and spatial resolution, such as who they are, where they visit, and when (Saluveer et al., 2020). This kind of data have been widely used in monitoring the spatial structures and patterns of tourist mobility (Park et al., 2020; Xu et al., 2021). These studies have greatly broadened our understanding of tourist mobility, but many studies have been inhibited by shortages in the availability of mobile phone positioning data, such as time precision and long-time series span, spatial scale, and coverage amounts. These studies have primarily been based on short-term temporal data (Kraemer et al., 2020; Tan et al., 2021) and cannot comprehensively assess the long-term changes occurring before and post-pandemic (Miao et al., 2021; Weaver, 2021; Xie et al., 2021). Research gaps analysis Based on the aforementioned literature, we identify the following gaps that we will address in this paper: Firstly, current studies on changes in tourist mobility after the COVID-19 outbreaks either use traditional data (i.e., statistic data, surveying data) or focus only on the early stage after the initial outbreak. It is necessary to understand the long-term changes of tourist mobility in order to bolster tourism recovery. More importantly, given the recurrence of the pandemic, establishing how multiple waves can affect tourist mobility is becoming more urgent. Nevertheless, current literature suggest that to date, such contributions are few due to the difficulty of data acquisition. Secondly, it would be valuable for policymakers and stakeholders in crisis management to understand the socio-demographic disparities in tourist mobility changes after the outbreaks. While some studies have gained a glimpse of social-demographic gaps in tourist mobility by using questionnaire surveys, they are faced with issues of representation. Therefore, it is necessary to conduct a study using big data with demographic information. Thirdly, even though we acknowledge that different cities have different responses to the public health crisis, there is a lack of empirical evidence regarding how these tourist mobility variations appear in diverse territorial contexts in response to COVID-19. Fourthly, it is relatively underdeveloped for researchers by using a national scale daily mobility dataset to explore the long-term dynamics of tourist mobility after COVID-19, this provides constrained information about the nation's emergency management policy. In this vein, it is necessary to analyze how these changes differ between cities on a national scale to holistically identify the disparities of the pandemic in different cities. Methods and data Study methodology Time series tourist mobility pattern cluster analysis. The time series cluster analysis helped us identify tourist mobility patterns using the daily number of tourists in each city. The cluster algorithm method is the K-means of the machine learning method, which is the distance measurement clustering algorithm of unusual scenic spots (Gibbs et al., 2020). Given a set of observations (x1, x2, …, xn), where each observation is a d-dimensional real vector, K-means clustering aims to partition the n observations into k (≤ n) sets S = {S1, S2, …, Sk} to minimize the within-cluster sum of squares (i.e., variance). Formally, the objective is to find the Eq. (1). The K value was identified by the “Elbow method” (Eyre et al., 2020; Tian et al., 2019) and “Silhouette Coefficient” (Hie et al., 2019).(1) argminS∑i=1k∑x∈Six−μi2 Assessing the dynamic changes of tourists and the impact of COVID-19. In this study, the tourist change rate was constructed to measure the year-on-date tourist change percentage (i.e., 1st Jan 2020 vs. 1st Jan 2019) before and after the COVID-19 outbreaks. The year 2019 was defined as the base year to compare the difference between the first and second outbreak waves. The tourist change rate was calculated using the following Eq. (2):(2) TCRy,d=TNy,d−TN2019,dTN2019,d×100% where TCR is the number of tourist year-to-date change rates after COVID-19. For the pre-pandemic levels ratio, y is the year, d is the date, and TN is the number of tourists. When TCR = 0, the number of tourists is the same as at pre-pandemic levels; when TCR > 0, the number of tourists is higher than at pre-pandemic levels; when TCR < 0, the number of tourists fell below that of pre-pandemic levels. The relationship of tourist change rates and city features in two waves.Model1:TCRwave1=FXgXpXdXk Model2:TCRwave2=FXgXpXdXk where TCR is the tourist change rate function; X g is the GDP per capita of the city; X p is the population size; X d is the distance to Beijing; X k is the average daily COVID-19 cases; F(·) is the linear regression formulation. Data description Tourist data We followed the United Nations World Tourism Organization's broad definition of ‘tourist’ in our study and defined a tourist “as a traveler taking a trip to the main destination outside his/her usual environment, for less than a year, for any main purpose (business, leisure or other personal purpose) other than to be employed by a resident entity in the country or place visited” (World tourism organization, 2008). In this study, we aggregate daily tourist mobility by the city of origin, which can be calculated from Beijing's total inflow of human mobility, minus non-tourism mobility (travel to work or a place of residence). Where total human mobility is extracted from the location of the mobile phone signaling base station. Work and place of residence are identified by the algorithm of the longest dwelling time during the day and night according to the time-constrained detection method (Vanhoof et al., 2018). Here is a sample of mobile data that belongs to the tourist in Table S9 of the Supplementary Data file. The data are anonymized mobile phone data provided by a third party, China Unicom, which is one of the three major providers in China. By 2021, China Unicom had a market share of 19.26 % (China Mobile Annual Report, 2021; China Telecom Annual Report, 2021; China Unicom Annual Report, 2021). We obtained three years of data during the same period between Jan 1st and May 31st, from 2019 to 2021, and three variables of tourist behavior, including the daily number of tourist mobility, their age groups, and gender groups. To ensure anonymity, every piece of data must be grouped by at least 15 individuals in the database. The study used mobility data aggregated by city and only considered cities with at least 15 tourists per day. For our analysis, we derived the daily tourist data in 254 cities in mainland China. Daily COVID-19 data We mainly collected the historical daily number of confirmed cases and risk levels under the Hierarchical Containment Measures in each city. All these data were openly available on the official website of the National Health Commission of the People's Republic of China. Others Data used to identify the factors influencing tourist change rates in cities, such as population size and GDP per capita, are from the China City Statistical Yearbook (2020). The city data used in this study are from 2019, pre-COVID-19. Distances were measured from the central location of each city to the central location of Beijing. Results How did overall domestic tourist mobility change in response to COVID-19 during different outbreak waves? Fig. 1a illustrates the main stages of the process following the first wave in 2020 and the second wave in 2021. Based on the daily number of tourist change trends, we divided the period after the initial COVID-19 outbreak into a downturn and the subsequent recovery stages, respectively. As the number of confirmed cases gradually decreased due to improvements in massive testing, contact tracing, and a series of effective control measures, many cities began relaxing travel restrictions and resumed their tourism businesses, with tourist mobility gradually reaching pre-pandemic levels (Fig. 1a). Fig. 1b shows the daily total number of domestic tourists in Beijing during the same months in a row over three years, from 2019 to 2021. From 2019 to 2021, the number of domestic tourists in Beijing witnessed a dramatic decline with a rate of 35.49 % in 2020 and a noticeable increase with a rate of 62.61 % in 2021. On average, the number of tourists showed a slight growth of 2.45 % during the same period. The maps in Fig. 1c and d show the year-to-date tourist change rate and its corresponding COVID-19 cases in pairs for 254 cities in four stages (see Tables S1, and S2 of Supplementary Data). We can see that a non-correlated relationship between the tourist change rate and confirmed cases in the origin cities in terms of tourist mobility.Fig. 1 The correlations between tourist mobility and COVID-19. a, changes in the time series of the number of tourists and COVID-19 cases in four stages; b, the daily number of domestic tourists in Beijing, which shows the breakdown numbers of tourists in weeks; c, the spatial distribution of the percentage of the tourist change rate for 254 cities during four stages; d, the spatial distribution of the average daily number of COVID-19 cases for 254 cities during four stages. Fig. 1 By comparing the downturn stages of two waves - the downturn stage of the first wave (D1, from Jan 10th, 2020, to Feb 14th, 2020) and the downturn stage of the second wave (D2, from Jan 1st, 2021, to Feb 12th, 2021) - we found that the number of tourists in D1 declined more than in D2. The volume of tourists dropped sharply - around a 90 % decrease in the first wave and around a 70 % in the second wave compared to pre-pandemic levels. The relatively smaller percentage in the second wave is partly due to the precision and refined control policies by the central government that greatly reduced the impact of the pandemic on population mobility, such as intensive Hierarchical Containment Measures (HCMs) (National Health Commission of the People's Republic of China, 2021). The control area was also gradually narrowed down to specific residential communities and buildings. Recently, more precise control measures have emerged, such as the “Time and Space Companion” in November 2021, which uses big data to locate the possible contacts of potential virus carriers. The temporal resolution reached 10 min and space accuracy reached 800 m × 800 m. By comparing the recovery stage of the first wave (R1, from Feb 15th, 2020, to May 31st, 2020) with the recovery stage of the second wave (R2, from Feb 13th, 2021 to May 31st, 2021), we found that the number of tourists in R2 recovered (106 days) faster than those in R1 (30 days) in terms of time spent rehabilitating themselves to pre-pandemic levels. Possible reasons for this improvement could be a) after a long-term decline in the tourism economy, policymakers and tourism practitioners enacted a series of strategies to expand domestic demand, leading to the fast-paced recovery of domestic tourism; or b) lessons were learned from the previous wave, and the government and stakeholders were better prepared both physically and psychologically. Notably, along with the introduction of vaccines after March 2021, the number of tourists had soared to 40 % higher than at pre-pandemic levels in 2019. This might be partially explained by the increasing number of people getting vaccinated and the irrational consumption activities of individuals after the virus hit. How did tourist mobility changes vary among different socio-demographic tourists? By gender Since the categories of gender do not fit a normal distribution with 95 % confidence (see Table A3 in appendix), we chose nonparametric tests (De Biasi et al., 2020) to examine disparities in the gender categories in four stages. Fig. 2a and b illustrate how the number of tourists changed by gender during the two waves. Subplots a and b introduce similar trends observable in female and male groups during the downturn stages of the two waves (Table A4 shows no statistical differences between females and males for tourist change rate in the two decline stages). However, the trends differed during the recovery stages (Table A4 shows statistical differences between females and males for tourist change rate in the two increase stages). Females recovered more slowly than males during the first wave, whereas the opposite occurred during the second wave. During R1, the pandemic affected females more than males (the red line below the blue line), implying that it was more difficult for female tourists to recover to pre-pandemic levels after the first wave. During R2, both female and male groups recovered to pre-pandemic levels at the same time, whereas after they reached that level, female tourists recovered far better and faster (the red line went beyond the blue line).Fig. 2 The socio-demographic disparities in the tourist change rate. a, the tourist change rate compared with pre-pandemic levels measured by gender groups in the first wave; b, the tourist change rate compared with pre-pandemic levels measured by gender groups in the second wave; c, the tourist change rate compared with pre-pandemic levels measured by age groups in the first wave; d, the tourist change rate compared with pre-pandemic levels measured by age groups in the second wave. Note: The imaginary lines are the actual daily number of tourists. The solid line is the 7-day moving average. To filter the fluctuations between workdays and weekends, we conducted a 7-day moving average. Fig. 2 By age Since the categories of age do not conform to a normal distribution with 95 % confidence (see Table A5 in appendix), we used nonparametric tests to examine the disparities in the age categories in four stages. Fig. 2c and d illustrate how the number of tourists changed by age during the two waves. Subplot c introduces a severe issue of tourist loss for all groups in the first wave: a lower-than-normal level of tourists persisted for about four months (Table A6 shows no statistical differences between females and males for tourist change rate in the D1 stage). The lowest daily decrease in the ratio of tourist change rate reached around 80 % for the number of tourists. During the R1 stage, tourists aged 19 to 39 years old recovered relatively faster than other groups, and tourists over 50 and under 18 years old recovered at a relatively slower pace (Table A6 shows statistical differences between females and males for tourist change rate in the R1 stage). Subplot d shows that during the second wave, D2, those under 18 years old suffered the most serious stress, while those over 60 years old were the least affected (Table A6 shows statistical differences between females and males for tourist change rate in the D2 stage). During R2, tourists over 50 years old recovered relatively faster than other groups (Table A6 shows statistical differences between females and males for tourist change rate in the R2 stage). A comparison of Fig. 2c and d reveals that, contrary to any intuitive projection, those most infected by COVID-19 in the first wave were the elderly tourists, whereas they suffered relatively lower rates, with the highest post-event recovery performance level occurring during the second wave. Control strategies such as the testing certificate and the “Health Quick Response code (HQR)1 ” used for accessing airport and tourist attractions, reinstated the elderly's lost travel accessibility during the first wave. Many government and public spaces are still actively trying to solve the problems faced by the elderly by using the HQR, such as the “Scan ID card2 ” and “Offline code.3 ” These measures alleviate the COVID-19 control measures that discriminated against the elderly during the second wave. How did tourist mobility changes differ from one city to another? The spatial variation in inter-city tourist mobility change remains largely unknown. Based on the time-series tourist change rate dynamics using the K-means cluster analysis method, we identified the tourist mobility change patterns at the city level. The K value was decided by using the “Elbow method” and “Silhouette Coefficient.” In the first wave, as shown in Fig. A1 in the appendix, the K value of the elbow is 3 (the highest curvature), so the best K value for clustering this data set should be 3, and the silhouette score with three clusters is 0.487, which is relatively higher. In the second wave, as shown in Fig. A2, the K value of the elbow is 3 (the highest curvature), so the best K value for clustering this data set also should be 3, and the silhouette score with three clusters is 0.371, which is relatively higher. Finally, we decided on three kinds of tourist mobility change patterns in the first (Fig. 3a) and second waves (Fig. 3b).Fig. 3 Tourist mobility change patterns in response to the first wave of the outbreak in 2020, and the second wave of the outbreak in 2021. a, b, pattern map showing the spatial distribution of time-series clusters; c, d, tourist mobility change pattern clusters correspond with the daily COVID-19 cases in each city. Fig. 3 Notably, in the fluctuation shown in Fig. 3c and d, the turning points of the lines were rather similar, but the fluctuation magnitudes were heterogeneous, which indicates that COVID-19 outbreaks have impacted cities at the same time, but in varying degrees. The cities from the pattern A cluster (such as Zhumadian; see the characteristics of these cities in Tables A7 and A8 in the appendix) displayed the quickest recovery and strongest increase ratio in tourist mobility; however, such cluster cities accounted for only 13.78 % in the first wave and 7.87 % in the second wave. The cities from the pattern C cluster (such as Shanghai; see the characteristics of these cities in Tables A7 and A8 in the appendix) suffered a relatively more severe hit in tourist mobility, which occupies a large proportion, with a ratio of 72.05 % in the first wave and 61.02 % in the second wave. We then quantified how tourist mobility change patterns vary with the origin city's population size, economic development level, and distance to Beijing. Through correlation analysis (see Tables A9 and A10), the three clusters were significantly associated with the city's population size (first wave: r = 0.180, p = 0.004; second wave: r = 0.191, p = 0.002), GDP per capita (first wave: r = 0.336, p < 0.001; second wave: r = 0.268, p < 0.001), and distance to Beijing (first wave: r = −0.162, p < 0.01; second wave: r = −0.338, p < 0.001). As shown in Fig. 4a and b, we found that although pattern cluster cities were associated with their population size, economic development level, and distance to Beijing, there were also substantial spatial disparities in the two waves.Fig. 4 Tourist mobility patterns in response to local-scale epidemic wave. Each dot represents a city in here and the coloring scheme is shared with Fig. 3. a, b, box plot of the characteristics of the city for tourist mobility patterns. All value scaling compressed to 0–1; c, d, quadrant analysis for addressing the tourist change rate in GDP per capita of the city, units are Chinese yuan; e, f, quadrant analysis for addressing the tourist change rate in the city's population size; g, h, quadrant analysis for addressing the tourist change rate in the city's distance to Beijing. Fig. 4 Firstly, there are obvious variations in the pandemic's effects on inter-city tourist travel between cities in terms of their geographical distance to Beijing. As shown in Fig. 4g, the tourist change rate was not association with distance during the first wave, which likely had a massive and widespread impact. Yet, during the second wave, as shown in Fig. 4h, the tourist change rate exhibits linear attenuation with the distance to Beijing. The closer city was hit hardest of all, while the city farthest away showed a relatively better increase rate. Tobler's first law of geography (Joo et al., 2017) may explain the aforementioned facts to some extent. Because geographical distance is a key determinant of the uneven spreading of the virus (Tsiotas & Tselios, 2022), cities nearest to Beijing may be more likely to be infected by the pandemic, leading to a wider-range outbreak. For instance, after the second wave, cities around Beijing such as Shijiazhuang were quickly infected by their proximity to the high-risk area. This aggravated the issue of tourists' travel behaviors in the vicinity of Beijing. The main COVID-19 pressure points have been in cities near Beijing, rather than being dispersed all over the country during the second wave. Secondly, urban economic development levels were considered to be the main factor in determining their tourist travel possibilities. We referred to the GDP per capita of each city to compare the tourist change rate variations among them. We found that the tourist change rate depends nonlinearly on the GDP per capita. The higher the city's GDP per capita is, the lower their tourist change rate. Moreover, the correlative relationship between the city's GDP per capita and tourist change rate in the first wave (p < 0.001, R2 = 0.235, see model 1 in Table A11 of appendix) is more significant than in the second wave (p = 0.012, R2 = 0.247, see model 2 in Table A11 of appendix). As shown in Fig. 4c, the most developed and wealthy cities experienced the hardest pandemic shock in the first wave (most of the dots in the fourth quadrant are orange). In addition, as shown in Fig. 4c and d, both in the first and second waves, most tourists originating from poorer cities suffered a lower negative impact from the pandemic (most of the blue dots representing pattern A are located in the second quadrant). Both richer and poorer cities were hit hard, and the main differences were in the recovery speed and level. It is also evident that the recovery speed is inversely proportional to the maturity level of economic development. Thirdly, urban population size was the key variable in measuring the volume of inter-city travel mobility. However, the correlation between tourist change rate and the city's population size was not always significant. The correlation was not significant in the first wave (p < 0.190, R2 = 0.235; see model 1 in Table A12 of appendix), but as shown in Fig. 4e, tourists in the largest cities suffered the most (in orange, which represents pattern C). While the correlation was negatively significant in the second wave (p = 0.098, R2 = 0.247; see model 2 in Table A13 of appendix), the best recovery-level cities were smaller (see Fig. 4f), showing that small cities can actually be more robust than larger ones. This may be because these cities usually have less human mobility, which, to some extent, may have protected them from the spread of the virus. Moreover, the number of tourists in those cities was relatively small during the previous year, so even a slight increase in tourists would lead to a large tourist change rate. Discussion and conclusion With the help of this long-term, wide-scale, high temporal resolution using highly reliable mobile phone data, this study contributes a comprehensive understanding of the dynamic changes in tourist mobility both before and post outbreak, and following the multiple waves. We observed that compared with the previous year's level, the number of domestic tourists in Beijing suffered a severe loss during 2020, then increased to even better than pre-pandemic levels one year after COVID-19. The number of domestic tourists in 2020 dropped sharply by 35.49 % compared with 2019. However, there was a slight average growth rate of 2.45 % from 2019 to 2021. Groups such as females and the elderly were more vulnerable during the first outbreak, whereas the opposite occurred during the second wave one year later. The same groups exhibited a relatively small external impact and recovered faster and more robustly than other groups, which demonstrates a distinct pattern of ‘better-than-before’ performance. Tourists under 18 years old suffered the most constant and serious impact of COVID-19. We also observed that the pandemic impacted cities simultaneously but to various degrees. This depended on the travel distance and socio-economic development of these cities, with larger, wealthier cities suffering from the most severe hit of all. This study's findings are discussed in the following ways. Firstly, given that the tourism sector was severely hit by the pandemic over the past two years (Chica et al., 2021; United Nations News, 2021; World Tourism Organization, 2021), many studies are pessimistic about tourism's recovery (Fotiadis et al., 2021). Here, we demonstrated the new situation for Beijing, China, where the number of domestic tourists in the post-pandemic time was substantially greater than in the pre-pandemic stage, which could partly be explained by the strong international travel restrictions promoting domestic tourist mobility recovery (Yang et al., 2021). This means that the rapid increase in the number of tourists may have originated due to the substantial decrease in outbound international travel during the outbreaks. For example, tourists preferring international travel may choose domestic tourist destinations instead (Organisation for Economic Co-operation and Development, 2020). This result highlights the strong resilience and potential for domestic tourism to bolster economic recovery in the post-pandemic era (World Tourism Organization, 2020), especially for major cities in populated countries. If the global pandemic persists, these changes may have far-reaching consequences for the future development of tourism as a whole. Secondly, recognizing tourist segmentation features is a fundamental issue in tourism research. Our results revealed that while the COVID-19 impact on tourist mobility varies for different socio-demographic subgroups, there were also substantial disparities during the two waves. We found that the elderly and female groups, which are normally recognized as vulnerable groups, were particularly affected by slow recovery speeds during the first wave. While their abilities to fight off the pandemic were limited in the first outbreak, during the second wave in the following year, these groups showed a relatively strong and faster recovery in terms of tourism-related activities. This might be explained by the fact that COVID-19's impact on tourist mobility was derived from external factors (i.e., travel restrictions of government strategies) and the internal psychology of spontaneous behavior (i.e., fear of contagion). The gender disparities might be explained by the shift in female tourists' health risk perceptions (Neuburger & Egger, 2021) after the two waves. Our findings supported the previous studies that female tourists tended to be more sensitive to external disturbances after the first outbreak (Bhopal & Bhopal, 2020). However, one year after the pandemic, female tourists got used to the crisis and their risk perception gradually fell. This, coupled with suppressed tourist demands, led to massively increased tourist mobility after the second outbreak. The difference in the elderly tourist mobility response between the first and second waves demonstrated high capabilities of ‘coping’ and ‘learning from the past.’ Previous studies indicated that access to information and the ability to respond to said information were associated with the subjective probability of risk preferences (Weill et al., 2020). Even if the elderly had lower risk perception due to their limited contact with the internet to access crisis information (Qiu et al., 2020), they were directly or indirectly constrained by government policies and restrictions in many ways during the early stage, such as the inability to show their “Health Quick Response code (HQR)” using a smartphone. With the help of family members and related organizations (i.e., the National health and wellness commission, and the National office for aging) facilitating “smart help for the elderly,” more and more elderly people have improved their ability to use a smartphone (Xinhua News, 2022). Thirdly, young people under 18 found it harder to recover to pre-pandemic levels, which might be explained by the fact that in mainland China, COVID-19 vaccinations were initially implemented for adults over 18 years old, and were not offered to those under 18 during our study period (Zheng et al., 2021). The latest research and media reports suggest that although vaccinations were offered to individuals under 18 after our observation period, their actual vaccination rate was very small (Bramer et al., 2020). This might be primarily because of “vaccine hesitancy,” as many parents were concerned about the possible complications of vaccines and their unclear explanations for youngsters (Bell et al., 2020). This implies a relatively negative post-pandemic performance for tourists under 18, suggesting that the effects of the crisis are likely to persist for young people. Finally, our study indicated that tourists originating from larger and richer cities suffered more due to the pandemic. As populated cities have been the main contributors to domestic tourism in Beijing, the high tourist decrease rate implies a very serious gap in the number of tourists in these cities. In contrast, the number of tourists originating from small cities suffered lower losses. In addition, we found that the tourist change ratewas not associated with distance during the first wave, which might be explained by the stronger and almost national impact of the pandemic during the first wave. However, during the second wave, the tourist change rate exhibited linear performance with the distance to Beijing; the largest tourist decline rate was more likely to happen as the remote cities performed better. It may be inferred that in the case of the second local outbreak, the associated restrictions, policies, and government surveillance on tourist mobility were generally stricter for the surrounding cities due to a higher-perceived contagion risk, which induced the highest decline in short-radius mobility around the city. Our findings could provide several practical implications to promote the sustainable development of tourism in the post-pandemic era as follows:(1) Given the overall growth of current domestic tourism after the outbreaks and during current weak tourism investments, we suggest and call for the government and stakeholders to strengthen investment in local tourism and provide strong support for the merger and recombination of the tourism industry in the next few years (Pham et al., 2021). This will not only meet tourists' needs and desires for travel, but can also encourage the benefits of tourism to spill over to other sectors. (2) As the pandemic hit tourism mobility disproportionally in different demographic groups, this inspires us to rethink the issue of travel fair (Alarcón & Cole, 2019) and guide tourism toward an equitable recovery (Rastegar et al., 2021). Based on the data in our study, the impacts of the pandemic were larger in vulnerable groups, so special care for disadvantaged groups would narrow down the inequity that COVID-19 caused. If this issue is not fully resolved, the tourism industry might lose a large number of tourist markets during the recovery process, causing greater social and well-being inequalities. (3) Our findings imply that the loss rate in the number of tourists was greater in larger and richer cities, which might help allocate more investment and offer prioritization policies for tourist markets in those major cities. Governments are supposed to allow some flexibility on control measurements (Bouman et al., 2021) rather than a “one-size-fits-all” solution. For example, each city should have a specific control policy and tourism recovery plan to better promote the balanced recovery of the different cities. (4) We did not find a correlative relationship between tourist change rates and COVID-19 confirmed cases after the outbreaks by using mobile phone data from China. Despite mobility restrictions being one of the most common strategies to combat the virus spreading (Schlosser et al., 2020), it is necessary to reexamine the effectiveness of government travel restrictions and policies to avoid the spread of the virus, given the huge economic losses and consequences. Additionally, new evidence revealed from our analysis confirms that under the premise of effective pandemic control (i.e., offering vaccination, testing) (Aleta et al., 2020; Russell & Greenwood, 2021) and gradually precise control measures (i.e., China's National top-down Hierarchical Containment policy) (National Health Commission of the People's Republic of China, 2021), it is possible to restore domestic tourism and even accelerate its growth (Iaquinto, 2020). Therefore, there is a need for some new and more precise strategies to seek balance between tourist economics and tourist mobility. For example, related management sectors in tourist destinations have implemented strategies to increase social distancing and avoid overcrowding in places. (5) As demonstrated in our findings, there were substantial disparities during the two waves. Pandemic prevention policies and tourist mobility management measures are meant to be gradual processes that opportunely adjust according to the wave features with different social-demographic subgroups. In addition, we suggest putting balanced recovery and equality development at the core of related control policies and the tourism recovery plan, instead of narrowly focusing on the number of tourist recoveries. (6) Our findings imply that the tourist mobility change differentiates across cities and varies with the city's population size, economic development level, and geographical distance to Beijing. As well, the social-demographic disparities in tourist mobility have not previously been fully captured in tourist modeling. Researchers should be aware of such heterogeneity and disparities in their model assumptions when developing tourist mobility models to further study the effects of COVID-19. We acknowledge that a major limitation in this paper is that our research used Beijing as its case study due to the limitation of data availability, which has specific characteristics regarding what measures to take when dealing with the pandemic. Therefore, we suggest that care be taken when generalizing our results. In addition, in big data studies, it is common to suffer from unbalanced observation issues. This is also a very cutting-edge research question in statistical analysis that we cannot address by ourselves. However, we believe the representativeness of our data is relatively high in the context of China. A research report on internet usage of underage residents in China during 2020 shows that the proportion of those under 18 years old who own mobile phones accounts for 65 % (China Internet Network Information Center, 2021). For future studies, researchers should be aware of tourism equity issues that have garnered a lot of interest in recent years (Rastegar et al., 2021) and should conduct a deep investigation to capture the inequities related to tourist mobility in the post-COVID-19 context. As well, according to UNWTO predictions (World Tourism Organization, 2022b), global tourism will recover to pre-pandemic levels by 2024 or later. International tourism's future recovery could lead to a new hypothesis regarding whether or not domestic tourism will decline due to the effects of the recovering number of people traveling internationally. CRediT authorship contribution statement Ling Yu: Conceptualization, Methodology, Writing – original draft, Visualization. Pengjun Zhao: Supervision, Writing – review & editing, Funding acquisition. Junqing Tang: Writing – review & editing. Liang Pang: Data curation. 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 Supplementary material Image 1 Supplementary tables Image 2 Data availability Source information and data to reproduce figures of this study are attached as a separate file in Supplementary Data. Acknowledgments We are grateful for the constructive comments from the editors as well as the reviewers. This study was supported by 10.13039/501100001809 National Natural Science Foundation of China (Grant numbers: 41925003, 42130402), and 10.13039/501100003453 Guangdong Provincial Natural Science Foundation (Grant number: 2022A1515010696). ☆ The authors are all currently affiliated at the School of Urban Planning and Design, Peking University, China. They are interested in multiple topics of human geography and urban studies including application of big data in tourism research, human mobility, transport planning, and resilient cities. 1 The mobile internet-based “Health Quick Response Code” has played a crucial role in helping China control the pandemic and resume work, production, and businesses. 2 “Scan ID card” means the elderly can check their “Health Quick Response Code” using a professional reading device. 3 “Offline code” means the elderly can take a printed version of offline “Health Quick Response Code”. Appendix A Appendix and supplementary data to this article can be found online at https://doi.org/10.1016/j.annals.2022.103522. ==== Refs References Alarcón D.M. Cole S. No sustainability for tourism without gender equality Journal of Sustainable Tourism 27 7 2019 903 919 Aleta A. Martin-Corral D. Pastore y Piontti A. Ajelli M. Litvinova M. Chinazzi M. …Moreno Y. Modelling the impact of testing, contact tracing and household quarantine on second waves of COVID-19 Nature Human Behaviour 4 9 2020 964 971 Arbulú I. Razumova M. Rey-Maquieira J. Sastre F. Can domestic tourism relieve the COVID-19 tourist industry crisis? The case of Spain Journal of Destination Marketing & Management 20 2021 100568 Bell S. Clarke R. Mounier-Jack S. Walker J.L. Paterson P. Parents’ and guardians’ views on the acceptability of a future COVID-19 vaccine: A multi-methods study in England Vaccine 38 49 2020 7789 7798 33109389 Bhopal S.S. Bhopal R. Sex differential in COVID-19 mortality varies markedly by age The Lancet 396 10250 2020 532 533 Bouman T. Steg L. Dietz T. Insights from early COVID-19 responses about promoting sustainable action Nature Sustainability 4 3 2021 194 200 Bramer C.A. Kimmins L.M. Swanson R. Kuo J. Vranesich P. Jacques-Carroll L.A. Shen A.K. Decline in child vaccination coverage during the COVID-19 pandemic—Michigan care improvement registry, May 2016–May 2020 American Journal of Transplantation 20 7 2020 1930 32596921 Callaway E. Heavily mutated omicron variant puts scientists on alert Nature 600 7887 2021 21 34824381 Chen J. Becken S. Stantic B. Using Weibo to track global mobility of Chinese visitors Annals of Tourism Research 89(C 2021 Chica M. Hernández J.M. Bulchand-Gidumal J. A collective risk dilemma for tourism restrictions under the COVID-19 context Scientific Reports 11 1 2021 1 12 33414495 China Internet Network Information Center Research Report on Internet Usage of Under-age in China in 2020 http://www.cnnic.cn/hlwfzyj/hlwxzbg/qsnbg/202107/t20210720_71505.htm 2021, July 20 China Mobile (2022, March). China Mobile Annual Report 2021 China Telecom (2022, March). Annual Report 2021 China Unicom (2022, March). China Unicom Annual Report 2021 Collins-Kreiner N. Ram Y. National tourism strategies during the Covid-19 pandemic Annals of Tourism Research 89 2021 103076 De Biasi S. Meschiari M. Gibellini L. Bellinazzi C. Borella R. Fidanza L. …Cossarizza A. Marked T cell activation, senescence, exhaustion and skewing towards TH17 in patients with COVID-19 pneumonia Nature Communications 11 1 2020 1 17 Domenech A. Gutierrez A. Clavé S.A. Built environment and urban cruise tourists’ mobility Annals of Tourism Research 81 2020 102889 Eyre R. De Luca F. Simini F. Social media usage reveals recovery of small businesses after natural hazard events Nature Communications 11 1 2020 1 10 Fotiadis A. Polyzos S. Huan T.C.T. The good, the bad and the ugly on COVID-19 tourism recovery Annals of Tourism Research 87 2021 103117 Gao Y. Sun D. Zhang J. Study on the impact of the COVID-19 pandemic on the spatial behavior of urban tourists based on commentary big data: A case study of Nanjing, China ISPRS International Journal of Geo-Information 10 10 2021 678 Gibbs H. Liu Y. Pearson C.A. Jarvis C.I. Grundy C. Quilty B.J. …Eggo R.M. Changing travel patterns in China during the early stages of the COVID-19 pandemic Nature Communications 11 1 2020 1 9 Gössling S. Scott D. Hall C.M. Pandemics, tourism and global change: A rapid assessment of COVID-19 Journal of Sustainable Tourism 29 1 2020 1 20 Guerriero C. Haines A. Pagano M. Health and sustainability in post-pandemic economic policies Nature Sustainability 3 7 2020 494 496 Han Y. Yang G. Zhang T. Spatial-temporal response patterns of tourist flow under entrance tourist flow control scheme Tourism Management 83 2021 104246 Hannam K. Butler G. Paris C.M. Developments and key issues in tourism mobilities Annals of Tourism Research 44 2014 171 185 Hardy A. Aryal J. Using innovations to understand tourist mobility in national parks Journal of Sustainable Tourism 28 2 2020 263 283 Hardy A. Birenboim A. Wells M. Using geoinformatics to assess tourist dispersal at the state level Annals of Tourism Research 82 2020 102903 He L.Y. Li H. Bi J.W. Yang J.J. Zhou Q. The impact of public health emergencies on hotel demand-estimation from a new foresight perspective on the COVID-19 Annals of Tourism Research 94 2022 103402 Henok B.G. Factors determining international tourist flow to tourism destinations: A systematic review Journal of Hospitality Management and Tourism 12 1 2021 9 17 Hie B. Bryson B. Berger B. Efficient integration of heterogeneous single-cell transcriptomes using Scanorama Nature Biotechnology 37 6 2019 685 691 Huang X. Lu J. Gao S. Wang S. Liu Z. Wei H. Staying at home is a privilege: Evidence from fine-grained mobile phone location data in the United States during the COVID-19 pandemic Annals of the American Association of Geographers 112 1 2022 286 305 Iaquinto B.L. Tourist as vector: Viral mobilities of COVID-19 Dialogues in Human Geography 10 2 2020 174 177 Ivanov I.A. Golomidova E.S. Terenina N.K. Influence of the COVID-19 pandemic on the change in volume and spatial structure of the tourist flow in Finland and Estonia in 2020 Regional Research of Russia 11 3 2021 361 366 Jin X.C. Qu M. Bao J. Impact of crisis events on Chinese outbound tourist flow: A framework for post-events growth Tourism Management 74 2019 334 344 32287753 Joo D. Woosnam K.M. Shafer C.S. Scott D. An S. Considering Tobler’s first law of geography in a tourism context Tourism Management 62 2017 350 359 Kádár B. Gede M. Tourism flows in large-scale destination systems Annals of Tourism Research 87 2021 103113 Kar A. Le H.T. Miller H.J. What is essential travel? Socioeconomic differences in travel demand in Columbus, Ohio, during the COVID-19 lockdown Annals of the American Association of Geographers 2021 1 24 Kock F. Nørfelt A. Josiassen A. Assaf A.G. Tsionas M.G. Understanding the COVID-19 tourist psyche: The evolutionary tourism paradigm Annals of Tourism Research 85 2020 103053 Kraemer M.U. Sadilek A. Zhang Q. Marchal N.A. Tuli G. Cohn E.L. …Brownstein J.S. Mapping global variation in human mobility Nature Human Behaviour 4 8 2020 800 810 Kulshrestha A. Krishnaswamy V. Sharma M. Bayesian BILSTM approach for tourism demand forecasting Annals of Tourism Research 83 2020 102925 Lenzen M. Sun Y.Y. Faturay F. Ting Y.P. Geschke A. Malik A. The carbon footprint of global tourism Nature Climate Change 8 6 2018 522 528 Levin R. Chao D.L. Wenger E.A. Proctor J.L. Insights into population behavior during the COVID-19 pandemic from cell phone mobility data and manifold learning Nature Computational Science 1 9 2021 588 597 Lin J. Cui Q. Xu H. Guia J. Health and local food consumption in cross-cultural tourism mobility: An assemblage approach Tourism Geographies 2020 1 19 Liu P. Zhang H. Zhang J. Sun Y. Qiu M. Spatial-temporal response patterns of tourist flow under impulse pre-trip information search: From online to arrival Tourism Management 73 2019 105 114 Liu W. Wang B. Yang Y. Mou N. Zheng Y. Zhang L. Yang T. Cluster analysis of microscopic spatio-temporal patterns of tourists’ movement behaviors in mountainous scenic areas using open GPS-trajectory data Tourism Management 93 2022 104614 Matsuura T. Saito H. The COVID-19 pandemic and domestic travel subsidies Annals of Tourism Research 92 2022 103326 Mertzanis C. Papastathopoulos A. Epidemiological susceptibility risk and tourist flows around the world Annals of Tourism Research 86 2021 103095 Miao L. Im J. Fu X. Kim H. Zhang Y.E. Proximal and distal post-COVID travel behavior Annals of Tourism Research 88 2021 103159 Ministry of Culture and tTourism of the People'’s Republic of China Basic situation of tourism market in 2019 https://www.mct.gov.cn/whzx/whyw/202003/t20200310_851786.htm 2020, March 10 Morar C. Tiba A. Basarin B. Vujičić M. Valjarević A. Niemets L. …Lukić T. Predictors of changes in travel behavior during the COVID-19 pandemic: The role of tourists’ personalities International Journal of Environmental Research and Public Health 18 21 2021 11169 34769688 National Health Commission of the People'’s Republic of China New Coronavirus Pneumonia Prevention and Control Plan (Fifth Edition) http://www.gov.cn/zhengce/zhengceku/2020-02/22/content_5482010.htm 2021, February 21 Neuburger L. Egger R. Travel risk perception and travel behaviour during the COVID-19 pandemic 2020: A case study of the DACH region Current Issues in Tourism 24 7 2021 1003 1016 Organisation for Economic Co-operation and Development Tourism Policy Responses to the coronavirus (COVID-19) https://www.oecd.org/coronavirus/policy-responses/tourism-policy-responses-to-the-coronavirus-covid-19-6466aa20/#p-d1e27 2020, June 2 Park S. Xu Y. Jiang L. Chen Z. Huang S. Spatial structures of tourism destinations: A trajectory data mining approach leveraging mobile big data Annals of Tourism Research 84 2020 102973 Pham T.D. Dwyer L. Su J.J. Ngo T. COVID-19 impacts of inbound tourism on Australian economy Annals of Tourism Research 88 2021 103179 Qiu R.T. Park J. Li S. Song H. Social costs of tourism during the COVID-19 pandemic Annals of Tourism Research 84 2020 102994 Rastegar R. Higgins-Desbiolles F. Ruhanen L. COVID-19 and a justice framework to guide tourism recovery Annals of Tourism Research 91 2021 103161 Ren M. Park S. Xu Y. Huang X. Zou L. Wong M.S. Koh S.Y. Impact of the COVID-19 pandemic on travel behavior: A case study of domestic inbound travelers in Jeju, Korea Tourism Management 92 2022 104533 Russell F.M. Greenwood B. Who should be prioritised for COVID-19 vaccination? Human Vaccines & Immunotherapeutics 17 5 2021 1317 1321 33141000 Saluveer E. Raun J. Tiru M. Altin L. Kroon J. Snitsarenko T. …Silm S. Methodological framework for producing national tourism statistics from mobile positioning data Annals of Tourism Research 81 2020 102895 Schlosser F. Maier B.F. Jack O. Hinrichs D. Zachariae A. Brockmann D. COVID-19 lockdown induces disease-mitigating structural changes in mobility networks Proceedings of the National Academy of Sciences 117 52 2020 32883 32890 Schmücker D. Reif J. Measuring tourism with big data? Empirical insights from comparing passive GPS data and passive mobile data Annals of Tourism Research Empirical Insights 3 2 2022 100061 Sharma A. Shin H. Santa-María M.J. Nicolau J.L. Hotels’ COVID-19 innovation and performance Annals of Tourism Research 88 2021 103180 Shin H. Nicolau J.L. Kang J. Sharma A. Lee H. Travel decision determinants during and after COVID-19: The role of tourist trust, travel constraints, and attitudinal factors Tourism Management 88 2022 104428 Shoval N. Isaacson M. Tracking tourists in the digital age Annals of Tourism Research 34 1 2007 141 159 Tan, S., Lai, S., Fang, F., Cao, Z., Sai, B., Song, B., Dai B., Guo, S., Liu, C., Cai, M., Wang, T., Wang, M., Li, J., Chen S., Qin, S., Floyd, J. R., Cao, Z., Tan, J., Sun, X., Zhou, T., Zhang, W., Tatem, A. J., Holme, P., Chen, X., & Lu, X. (2021). Mobility in China, 2020: A tale of four phases. National Science Review, 8(11), nwab148. Tian T. Wan J. Song Q. Wei Z. Clustering single-cell RNA-seq data with a model-based deep learning approach Nature Machine Intelligence 1 4 2019 191 198 Tsiotas D. Tselios V. Understanding the uneven spread of COVID-19 in the context of the global interconnected economy Scientific Reports 12 1 2022 1 15 34992227 Türk U. Östh J. Kourtit K. Nijkamp P. The path of least resistance explaining tourist mobility patterns in destination areas using Airbnb data Journal of Transport Geography 94 2021 103130 Uglis J. Jęczmyk A. Zawadka J. Wojcieszak-Zbierska M.M. Pszczoła M. Impact of the COVID-19 pandemic on tourist plans: A case study from Poland Current Issues in Tourism 25 3 2022 405 420 United Nations News The COVID-19 epidemic will cost the global tourism industry $2 trillion this year https://news.un.org/zh/story/2021/11/1095142 2021, November 29 Vanhoof M. Reis F. Ploetz T. Smoreda Z. Assessing the quality of home detection from mobile phone data for official statistics Journal of Official Statistics 34 4 2018 935 960 Weaver A. Tourism, big data, and a crisis of analysis Annals of Tourism Research 88 2021 103158 Weill J.A. Stigler M. Deschenes O. Springborn M.R. Social distancing responses to COVID-19 emergency declarations strongly differentiated by income Proceedings of the National Academy of Sciences 117 33 2020 19658 19660 World Tourism Organization Glossary of tourism terms https://www.unwto.org/glossary-tourism-terms 2008 World Tourism Organization UNWTO highlights potential of domestic tourism to help drive economic recovery in destinations worldwide https://www.unwto.org/news/unwto-highlights-potential-of-domestic-tourism-to-help-drive-economic-recovery-in-destinations-worldwide 2020, September 14 World Tourism Organization 2020: Worst year in tourism history with 1 billion fewer international arrivals https://www.unwto.org/news/2020-worst-year-in-tourism-history-with-1-billion-fewer-international-arrivals 2021, January 28 World Tourism Organization Tourism in the 2030 agenda https://www.unwto.org/tourism-in-2030-agenda 2022, January 29 World Tourism Organization Tourism grows 4% in 2021 but remains far below pre-pandemic levels https://www.unwto.org/taxonomy/term/347 2022, January 18 Xie G. Qian Y. Wang S. Forecasting Chinese cruise tourism demand with big data: An optimized machine learning approach Tourism Management 82 2021 104208 Xinhua News Carry out the "wisdom to help the elderly" action http://www.gov.cn/xinwen/2022-06/16/content_5696072.htm 2022, Jun 16 Xu Y. Li J. Belyi A. Park S. Characterizing destination networks through mobility traces of international tourists—A case study using a nationwide mobile positioning dataset Tourism Management 82 2021 104195 Yang Y. Altschuler B. Liang Z. Li X.R. Monitoring the global COVID-19 impact on tourism: The COVID19tourism index Annals of Tourism Research 90 2021 103120 Yang Y. Zhang L. Wu L. Li Z. Does distance still matter? Moderating effects of distance measures on the relationship between pandemic severity and bilateral tourism demand Journal of Travel Research 2022 10.1177/00472875221077978 Zha J. Tan T. Ma S. He L. Filimonau V. Exploring tourist opinion expression on COVID-19 and policy response to the pandemic’s occurrence through a content analysis of an online petition platform Current Issues in Tourism 25 2 2022 261 286 Zheng W. Huang X. Li Y. Understanding the tourist mobility using GPS: Where is the next place? Tourism Management 59 2017 267 280 Zheng W. Li M. Lin Z. Zhang Y. Leveraging tourist trajectory data for effective destination planning and management: A new heuristic approach Tourism Management 89 2022 104437 Zheng W. Yan X. Zhao Z. Yang J. Yu H. COVID-19 vaccination program in the mainland of China: A subnational descriptive analysis on target population size and current progress Infectious Diseases of Poverty 10 1 2021 1 10 33397494 Zheng W. Zhou R. Zhang Z. Zhong Y. Wang S. Wei Z. Ji H. Understanding the tourist mobility using GPS: How similar are the tourists? Tourism Management 71 2019 54 66 Zhou B. Zhang Y. Zhou P. Multilateral political effects on outbound tourism Annals of Tourism Research 88 2021 103184	36474961	PMC9715491	NO-CC CODE	2022-12-15 23:15:23	no	Ann Tour Res. 2023 Jan 2; 98:103522	utf-8	Ann Tour Res	2,022	10.1016/j.annals.2022.103522	oa_other
==== 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. on behalf of Canadian Association of Medical Radiation Technologists. S1939-8654(22)00673-7 10.1016/j.jmir.2022.11.014 Research Article Anxiety, depression, and stress among radiography undergraduates during the COVID-19 pandemic Weerakoon Bimali Sanjeevani ⁎ Chandrasiri Nishadi Rangana Department of Radiography/ Radiotherapy, Faculty of Allied Health Sciences, University of Peradeniya, Peradeniya 20400, Sri Lanka ⁎ Corresponding author. 2 12 2022 2 12 2022 © 2022 Published by Elsevier Inc. on behalf of Canadian Association of Medical Radiation Technologists. 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 The COVID-19 pandemic has severely impacted education and other aspects of life, causing psychological distress. The current study aims to identify anxiety, depression, and stress among radiography undergraduates during the COVID-19 pandemic. Method A descriptive, cross-sectional study was conducted between November and December 2021 on a sample of 140 radiography undergraduates at the Department of Radiography/Radiotherapy, Faculty of Allied Health Sciences, University of Peradeniya. An online survey with two sections: demographic characteristics and a psychometric scale (DASS-42) was used for data collection. Results A total of 107 undergraduates responded to the questionnaire giving a response rate of 76.2%. The results revealed that the majority of radiography undergraduate students suffered from mild to extremely severe depression (87.85%), anxiety (92.52%), and stress (73.83%) levels. In addition, more than two-thirds of the students (>73% of participants) reported at least one symptom of depression, anxiety, or stress to varying degrees. Scores for depression, anxiety, and stress did not differ significantly across gender and academic year. However, a significant difference was observed between the two age groups, 23–26 years and > 27 years, regarding depression. The older students reported severe depression, whereas younger students reported moderate depression. Conclusion A high prevalence of negative psychological impact was observed among radiography undergraduates during the COVID-19 pandemic. This necessitates taking proactive steps to address, safeguard, and nurture undergraduates’ mental health and well-being during the current and future pandemic crises to mitigate the negative impacts. Résumé Introduction La pandémie de COVID-19 a eu de graves répercussions sur l'éducation et d'autres aspects de la vie, provoquant une détresse psychologique. La présente étude vise à identifier l'anxiété, la dépression et le stress chez les étudiants de premier cycle en radiographie pendant la pandémie de COVID-19. Méthode : Une étude descriptive et transversale a été menée entre novembre et décembre 2021 sur un échantillon de 140 étudiants de premier cycle en radiographie au Département de radiographie/radiothérapie, Faculté des sciences paramédicales, Université de Peradeniya. Une enquête en ligne comportant deux sections : les caractéristiques démographiques et une échelle psychométrique (DASS-42) a été utilisée pour la collecte des données. Résultats Un total de 107 étudiants de premier cycle ont répondu au questionnaire, soit un taux de réponse de 76,2%. Les résultats ont révélé que la majorité des étudiants de premier cycle en radiographie souffraient de dépression légère à extrêmement sévère (87,85 %), d'anxiété (92,52 %) et de stress (73,83 %). En outre, plus des deux tiers des étudiants (>73 % des participants) ont signalé au moins un symptôme de dépression, d'anxiété ou de stress à des degrés divers. Les scores de dépression, d'anxiété et de stress ne différaient pas significativement selon le sexe et l'année universitaire. Cependant, une différence significative a été observée entre les deux groupes d'âge, 23-26 ans et > 27 ans, concernant la dépression. Les étudiants les plus âgés ont déclaré une dépression sévère, tandis que les plus jeunes ont déclaré une dépression modérée. Conclusion Une prévalence élevée d'impact psychologique négatif a été observée chez les étudiants de premier cycle en radiographie pendant la pandémie de COVID-19. Il est donc nécessaire de prendre des mesures proactives pour aborder, protéger et entretenir la santé mentale et le bien-être des étudiants de premier cycle pendant les crises pandémiques actuelles et futures afin d'atténuer les impacts négatifs. Keywords Depression Anxiety Stress DASS-42 Radiography Undergraduates ==== Body pmcIntroduction COVID-19 is a global health emergency announced by the World Health Organization (WHO) on 11 March 2020 [1]. Many countries initiated strict public health measures to prevent the rapid spread of the COVID-19 pandemic [2,3]. At the initial stage of this pandemic in Sri Lanka, the government placed the country under lockdown and introduced new guidelines to minimise travel and public gatherings [4,5]. All public and private universities were closed, and face-to-face teaching was entirely halted to prevent the spread of the COVID-19 disease among university undergraduate students [6]. Setting a fixed date for the reopening of the universities was uncertain due to the highly contagious second and third waves of COVID-19 in Sri Lanka. In this context, educational institutes were compelled to devise proactive plans to minimise disruptions at all levels. As a result, most Sri Lankan universities began using digital platforms and online technologies like Moodle-based learning management systems and other types of video communication software to continue educational activities [6], [7], [8]. These measures made the undergraduates socially isolated, which could be a difficult time for them as they are a dynamic and energetic group [9], [10], [11], [12]. Sri Lankan university students had not extensively used the online education system before the pandemic [13,14]. Therefore, during the transition to online education, the students in the university community had to face numerous challenges: the lack of technical knowledge and limited financial, infrastructural, and human resources [6,13,14]. Further, they confronted difficulties in continuing the practical classes and clinical training in health-related fields like radiography, which requires hands-on experience [15], [16], [17]. In particular, COVID-19 has affected the psychological well-being of undergraduates who are already under pressure, especially in health-related sectors [18,19]. Therefore, it is significant to understand the psychological dynamics of undergraduates during a public health emergency to manage the situation effectively, take appropriate interventions, and minimise causalities [20]. Depression is classified as a mood disorder that hinders daily functioning and is accompanied by symptoms of persistent sadness, feelings of guilt, frustration and loss of interest. Anxiety is a psychological condition that causes an unpleasant feeling associated with uneasiness about future events or fear of respecting current conditions. Stress is a condition in which a person's ability to adapt to their environment is surpassed, leading to psychological and biological alternations that could increase the risk of disease [21,22]. Previous studies [16,19,23,24]. demonstrated the psychological well-being of radiography students, mostly from developed and well-resourced countries. A study done in Greece [19] has shown that radiography undergraduates were suffering from moderate levels of depression, anxiety and stress during the COVID-19 pandemic. However, limited studies to date have reported the experiences in low-resource countries [25]. Hence, this study assessed anxiety, depression, and stress among radiography students in Sri Lanka during the COVID-19 pandemic. The study also aimed to identify the associations between anxiety, depression, and stress with demographic variables of radiography students. Methods A descriptive, cross-sectional study was conducted between November and December 2021, at the end of the peak of the third wave of the COVID-19 pandemic in Sri Lanka. This study was performed at the Department of Radiography/Radiotherapy, Faculty of Allied Health Sciences, University of Peradeniya, the only department in Sri Lanka under the Ministry of Higher Education dedicated to delivering radiography education. A total of 140 radiography undergraduates from the first to the fourth year were invited to participate in the study. Data were collected through a self-administered structured online questionnaire developed by Google Forms online survey tool. The students received three reminder emails in the weeks that followed the initial invitation, along with a link to the survey. The questionnaire consisted of two sections. The first section assessed the demographic characteristics of the respondents (age, gender, and academic year). The second section consisted of the standard version of the Depression Anxiety Stress Scale-42 (DASS-42), obtained from the DASS official website [26]. The DASS-42 consists of three self-report scales designed to assess individuals’ negative emotional states of depression, anxiety, and stress. Each of these three scales (anxiety, depression, and stress) contains 14 items (a total of 42 items). Each item in the DASS-42 scale consists of four statements scored on a four-point Likert scale ranging from 0 to 3 [0 = did not apply to me at all; 1 = applied to me to some degree, or sometimes; 2 = applied to me to a considerable degree or a good part of the time; 3 = applied to me very much, or most of the time]. Scores for each subscale are calculated by summing the scores for the relevant items, classified as normal, mild, moderate, severe, and extremely severe, as shown in Table 1 [26].Table 1 Severity levels of depression, anxiety and stress [26]. Table 1Interpretation Depression Anxiety Stress Score Score Score Normal 0–9 0–7 0–14 Mild 10–13 8–9 15–18 Moderate 14–20 10–14 19–25 Severe 21–27 15–19 26–33 Extremely severe ≥28 ≥20 ≥34 Ethical approval was obtained from the Ethics Review Committee of the Faculty of Allied Health Sciences, University of Peradeniya (AHS/ERC/2021/106). A detailed introduction to the study was provided and voluntary participation was emphasised to the participants at the time of enrolment. Participants were asked to provide their informed consent to participate in the study by completing and submitting the consent form along with the questionnaire. Identifying information was not collected to ensure the participants’ confidentiality, and all collected data were securely stored. After excluding incomplete responses, data were analyzed using the Statistical Package for Social Sciences version 20.0 [27]. The results of the descriptive data analysis were presented in the form of frequency tables. Levels of anxiety, depression, and stress were presented in means (SD) and percentages. The normality of the data was assessed by the Shapiro-Wilk test. Based on the results of the normality test, differences between mean scores with gender were assessed using the t-test or Mann-Whitney U test, while ANOVA or Kruskal-Wallis test was used to determine the differences between mean scores with different age groups and academic years. The effect size was reported using eta squared (η 2) to assess the magnitude of differences between the groups. Effect sizes were described as small (η 2 = 0.01), medium (η 2 = 0.06), and large (η 2 = 0.14) [28]. The unique contribution of independent variables on depression, anxiety, and stress scores was assessed via Multiple linear regression analysis. The significance level was set at 0.05 (two-tailed) in all data analyses. Results Demographic characteristics A total of 107 undergraduates responded to the questionnaire giving a response rate of 76.2%. Table 2 presents their demographic characteristics. The majority of respondents were female (60.7%) and belonged to the age group of 23 to 26 years (74.8%). Most respondents who contributed to the questionnaire were in their fourth year (29.0%). First and third-year undergraduates responded equally to the questionnaire, with a response rate of 23.4%.Table 2 Demographic characteristics of the respondents. Table 2Variables Levels Frequency (%) Gender Male 42 (39.3%) Female 65 (60.7%) Age 19 – 22 17 (15.9%) 23 – 26 80 (74.8%) > 27 10 (9.3%) Academic year First-year 25 (23.4%) Second-year 26 (24.2%) Third-year 25 (23.4%) Fourth-year 31 (29.0%) Levels of depression, anxiety and stress As shown in Table 3 , 87.85% of undergraduates suffered from depression at varying degrees; 14.02%, 40.19%, 28.03%, and 5.61% were mild, moderate, severe, and extremely severe, respectively. The majority of undergraduates (92.52%) experienced mild to extremely severe anxiety levels. A significant proportion of the undergraduates (73.82%) suffered from varying degrees of stress levels, and most (41.12%) suffered from mild stress.Table 3 The distribution of severity levels of depression, anxiety and stress among respondents. Table 3Interpretation Depression Anxiety Stress Score No (%) Score No (%) Score No (%) Normal 0–9 13 0–7 8 0–14 29 (12.15%) (7.48%) (27.10%) Mild 10–13 15 8–9 1 15–18 29 (14.02%) (0.93%) (27.10%) Moderate 14–20 43 10–14 17 19–25 44 (41.12%) (40.19%) (15.89%) Severe 21–27 30 15–19 46 26–33 5 (28.04%) (42.99%) (4.67%) Extremely severe ≥28 6 ≥20 35 ≥34 1 (5.61%) (32.71%) (0.93 %) Table 4 demonstrates the average distribution of depression, anxiety, and stress among the respondents. When compared with Table 1, the results indicate that the undergraduates, in general, had moderate depression 17.12 (±6.59), severe anxiety 16.82 (±5.68), and mild stress 17.40 (±5.87) symptoms.Table 4 The average distribution of depression, anxiety and stress levels among the respondents. Table 4 Mean(SD) Interpretation Maximum Minimum Depression 17.12 (6.59) Moderate 34 0 Anxiety 16.82 (5.68) Severe 33 0 Stress 17.40 (5.87) Mild 34 0 Associations between depression, anxiety, and stress of undergraduates with demographic characteristics Table 5 shows the distribution and differences in depression, anxiety, and stress levels between males and females. The mean scores for males’ depression and stress were higher than for females. However, males and females did not demonstrate significantly different depression, anxiety, or stress levels. Both genders reported moderate depression, severe anxiety, and mild stress levels. In addition, no significant difference was noted between age groups regarding anxiety or stress levels. Although, the age group of > 27 years has a significantly higher average level of depression than the 23–26 age group (p < 0.05), the actual mean difference between those two age groups was medium (η 2 = 0.075) (Table 6 ).Table 5 Distribution and differences of depression, anxiety and stress levels among gender. Table 5Distribution of depression, anxiety and stress levels among gender Scale Gender Normal No (%) Mild No (%) Moderate No (%) Severe No (%) Extremely Severe No (%) Depression Male 5 (11.9) 4 (9.5) 16 (38.1) 16 (38.1) 1 (2.4) Female 8 (12.3) 11 (16.9) 27 (41.5) 14 (21.5) 5 (7.7) Anxiety Male 3 (7.1) 2 (4.8) 7 (16.7) 17 (40.5) 13 [31] Female 5 (7.7) 0 (0.0) 10 (15.4) 28 (43.1) 22 (33.8) Stress Male 10 (23.8) 9 (21.4) 22 (52.4) 1 (2.4) 0 (0.0) Female 19 (29.2) 19 (29.2) 22 (33.8) 4 (6.2) 1 (1.5) Differences in depression, anxiety and stress levels among gender Scale Gender Mean (SD) Interpretation Test statistics P-value η 2 Depression Male 17.50 (6.37) Moderate 0.476 a 0.635 0.002 Female 16.88 (6.77) Moderate Anxiety Male 16.74 (5.09) Severe -0.400 b 0.689 0.0015 Female 16.88 (6.08) Severe Stress Male 18.14 (5.29) Mild -1.289 b 0.197 0.016 Female 16.92 (6.21) Mild a t-test b Mann-Whitney test Table 6 Distribution and difference in depression, anxiety and stress levels among age groups. Table 6Distribution of depression, anxiety and stress levels among age groups Scale Age group Normal No (%) Mild No (%) Moderate No (%) Severe No (%) Extremely Severe No (%) Depression 19 – 22 0 (0.0) 0 (0.0) 12 (70.6) 5 (29.4) 0 (0.0) 23 – 26 13 (16.2) 15 (18.8) 29 (36.2) 18 (22.5) 5 (6.2) > 27 0 (0.0) 0 (0.0) 2 (20.0) 7 (70.0) 1 (10.0) Anxiety 19 – 22 0 (0.0) 0 (0.0) 3 (17.6) 10 (58.8) 4 (23.5) 23 – 26 8 (10.0) 2 (2.5) 13 (16.2) 29 (36.2) 28 (35.0) > 27 0 (0.0) 0 (0.0) 1 (10.0) 6 (60.2) 3 (30.0) Stress 19 – 22 3 (17.6) 2 (11.8) 11 (64.7) 1 (5.9) 0 (0.0) 23 – 26 25 (31.2) 23 (28.8) 28 (35.0) 3 (3.8) 1 (1.2) > 27 1 (10.0) 3 (30.0) 5 (50.0) 1 (10.0) 0 (0.0) Differences in depression, anxiety and stress levels among age groups Scale Age group Mean (SD) Interpretation Test statistics P-value Post Hoc test η 2 Depression 19 – 22 18.47 (3.64) Moderate 4.210 a 0.017 P (23 – 26 vs > 27) = 0.02 0.075 23 – 26 16.21 (7.05) Moderate > 27 22.1 (3.67) Severe Anxiety 19 – 22 17.47 (3.41) Severe 0.687 b 0.709 - 0.004 23 – 26 16.49 (6.26) Severe > 27 18.40 (3.44) Severe Stress 19 – 22 19.76 (4.10) Moderate 2.533 a 0.084 - 0.046 23 – 26 16.68 (6.20) Mild > 27 19.20 (4.39) Moderate a ANOVA b Kruskal-Wallis test Table 7 depicts the distribution and differences in depression, anxiety, and stress levels by academic year. The first-year students demonstrated the lowest level of depression and anxiety. However, no significant difference was observed in depression, anxiety, and stress levels among undergraduates concerning their academic year (Table 7).Table 7 Distribution and differences in depression, anxiety and stress levels across the academic year. Table 7Distribution of depression, anxiety and stress levels across the academic year Scale Academic year Normal No (%) Mild No (%) Moderate No (%) Severe No (%) Extremely Severe No (%) Depression First year 4 (16.0) 3 (12.0) 13 (52.0) 4 (16.0) 1 (4.0) Second year 4 (15.4) 4 (15.4) 10 (38.5) 6 (23.1) 2 (7.7) Third year 2 (8.0) 3 (12.0) 9 (36.0) 9 (36.0) 2 (8.0) Fourth year 3 (9.7) 5 (16.1) 11 (35.5) 11 (35.5) 1 (3.2) Anxiety First year 3 (12.0) 1 (4.0) 6 (24.0) 10 (40.0) 5 (20.0) Second year 2 (7.7) 0 (0.0) 4 (15.4) 11 (42.3) 9 (34.6) Third year 1 (4.0) 0 (0.0) 4 (16.0) 11 (44.0) 9 (36.0) Fourth year 2 (6.5) 1 (3.2) 3 (9.7) 13 (41.9) 12 (38.7) Stress First year 8 (32.0) 2 (8.0) 13 (52.0) 2 (8.0) 0 (0.0) Second year 7 (26.9) 6 (23.1) 12 (46.2) 1 (3.8) 0 (0.0) Third year 7 (28.0) 9 (36.0) 8 (32.0) 0 (0.0) 1 (4.0) Fourth year 7 (22.6) 11 (35.5) 11 (35.5) 2 (6.5) 0 (0.0) Differences in depression, anxiety and stress levels across the academic year Scale Academic year Mean (SD) Interpretation Test statistics P-value η 2 Depression First year 15.56 (6.33) Moderate 1.29 a 0.282 0.036 Second year 16.54 (6.93) Moderate Third year 19.08 (6.65) Moderate Fourth year 17.29 (6.37) Moderate Anxiety First year 14.76 (6.19) Severe 2.697 b 0.441 0.068 Second year 17.19 (5.93) Severe Third year 18.08 (4.92) Severe Fourth year 17.16 (5.45) Severe Stress First year 17.52 (7.08) Mild 1.118 b 0.773 0.012 Second year 17.69 (5.33) Mild Third year 17.32 (5.05) Mild Fourth year 17.13 (6.10) Mild a ANOVA b Kruskal-Wallis test Factors affecting depression, anxiety and stress of undergraduates The regression results (Table 8 ) revealed that all three independent variables of gender, age, and academic year could not statistically significantly predict depression, anxiety, or stress levels of the students (p > 0.05).Table 8 Multiple linear regression model for predicting undergraduate's depression, anxiety and stress levels. Table 8Variables Unstandardized coefficients Standardized coefficients t Sig. 95.0% Confidence interval for B B Std. Error Beta Lower Bound Upper Bound Depression Age .078 1.631 .006 .048 .962 -3.157 3.313 Gender -.609 1.316 -.045 -.463 .644 -3.218 2.000 Academic year .689 .714 .119 .965 .337 -.727 2.106 Anxiety Age -1.416 1.396 -.125 -1.014 .313 -4.184 1.352 Gender .238 1.126 .021 .211 .833 -1.995 2.470 Academic year 1.134 .611 .228 1.856 .066 -.078 2.347 Stress Age -.881 1.454 -.075 -.606 .546 -3.764 2.002 Gender -1.175 1.173 -.098 -1.002 .319 -3.500 1.151 Academic year .072 .637 .014 .113 .910 -1.191 1.335 *Correlation is significant at the 0.05 level (2-tailed). Discussion The COVID-19 pandemic has significantly impacted the emotional well-being and psychological health of people worldwide [12,29]. Although the studies have demonstrated that undergraduates, in general, suffered from mental health problems regardless of the pandemic [30], it is evident that COVID-19 has increased these levels significantly [31]. University undergraduates are invariably at elevated risk of psychological problems due to the closure of universities worldwide and difficulties in completing academic work [31]. Students in health-related fields such as radiography have been severely affected by this COVID-19 outbreak as their academic education and clinical-based practices were interrupted [15], [16], [17]. In addition, the pandemic caused a decline in the number of imaging procedures performed by about 50%, which has also impacted student training [32]. Therefore, this study attempted to identify the psychological distress level of radiography students during a global pandemic. The research findings denote that radiography students suffered from moderate depression, severe anxiety, and mild stress symptoms. More than two-thirds of the students (> 73% of participants) reported at least one symptom of depression, anxiety, or stress in varying degrees. A study in Greece [19] produced similar results for the same category of students, revealing that more than half of the students were affected. The current pandemic places demand on students due to the closure of universities, the expectation to adapt to online education despite the shortage of resources, uncertainty about continuing academic programmes, including exams, and increased economic pressure. Restricted social and physical interactions and tension about jeopardising future careers could further worsen this situation [33], [34], [35], [36]. However, the present study shows that a significantly higher proportion of undergraduates suffered from psychological distress compared to previous studies [37,38]. conducted with the same research tool. The study discipline, facilities available to educators and learners, characteristics of the education system in each country, and the time of the study conducted would possibly explain these differences. During the COVID-19 pandemic, Sri Lanka rapidly shifted its traditional learning strategies without considering existing curricula and available facilities because of the obligation to continue student education [6]. Further, the higher prevalence might be reflected in the absence of well-organised mentor-mentee programmes and psychological support systems adopted by the institution. The resulting psychological distress could be associated with the burden of the potential impact of the disease on families and themselves because this study was conducted at the end of the peak of the pandemic's third wave. Nonetheless, this study provides evidence of the importance of having timely access to psychological support systems and stress management programmes for students. According to the findings, anxiety was the most seriously affected psychological distress. As reported in the previous literature [39]. the severity of the anxiety could be reduced if the students were well aware of the disease and the precautionary measures. This highlights the significance of conducting regular institution-based awareness sessions that help students to clarify their doubts during a pandemic. Poor self-confidence could be associated with increased psychological distress [40]. Thus, it is important to set up remote psychological support systems through hotlines, chatlines, or virtual platforms for students where they can boost their self-confidence when needed. In particular, developing guidelines for establishing remote psychological support systems for students can be identified as a necessary measure. The results further emphasise the necessity of regular screening for the psychological distress of undergraduates during a pandemic. Previous studies [19]. conducted with a similar group of students and other types of undergraduates [41,42]. have shown that females are more psychologically vulnerable than males during the COVID-19 pandemic. In the present study, both male and female students experienced similar psychological effects. Therefore, gender was not a critical factor in the development of psychological distress during the COVID-19 pandemic in this study. Several studies [43,44] have demonstrated that pre-clinical students had a greater psychological impact due to their lesser experience in the healthcare system. Pre-clinical students were less resilient compared to clinical students and had more difficulty with self-adaptation to newly-introduced learning patterns [45]. In contrast, the present study revealed no significant difference in mental health among students during the pre-clinical and clinical years of the COVID-19 pandemic. These observations could be due to the lockdown, disruptions to the academic programme, and the negative impact on family well-being caused by COVID-19 is so pervasive and evokes similar feelings among everyone, regardless of gender or educational status. According to the present study findings, the older students (> 27 years) reported severe depression, whereas the younger students (23-26 years) reported moderate depression. In general, the older age group may have a higher prevalence of depressive symptoms due to increasing workload and future uncertainties as they approach graduation and the requirement to seek a job [46]. A few studies, on the contrary, have shown that younger students experienced significantly more emotional health issues than older students during the COVID-19 pandemic [47,48]. Strengths and limitations This study is the first attempt of its kind to assess mental health symptoms of depression, anxiety, and stress in radiography undergraduates during the COVID-19 pandemic in Sri Lanka. It produced significant findings that can help develop a psychological intervention for undergraduates during a global pandemic. Still, the study has certain limitations. First, this study used an online data collection tool, which may be underprivileged for the participation of undergraduates who lack adequate internet facilities. Second, it only focused on the post-pandemic mental health status of the radiography students; no data are available on the pre-pandemic status. Hence it is difficult to identify the exact impact of the COVID-19 pandemic on the mental health status of undergraduates. Third, this research used a self-assessment questionnaire for data collection; the results could not be confirmed by a clinical assessment. Conclusions It is impossible to determine the full impact of the COVID-19 pandemic on radiography education until it is deemed over. This was the first study of its kind in Sri Lanka to evaluate the psychological well-being of radiography undergraduates during the COVID-19 pandemic. The results signify that most radiography undergraduates experienced mild to extremely severe depression, anxiety, and stress levels. Therefore, this study supports the mounting evidence that the COVID-19 pandemic can negatively impact the mental health of radiography undergraduates. The study also demonstrates the risk associated with radiography undergraduates’ mental well-being during a global pandemic. In particular, this highlights the importance of taking the necessary proactive steps to identify, address, safeguard, and nurture radiography undergraduates’ mental well-being during the current and future pandemic crises to mitigate the negative impacts. The findings support relevant parties to recognise the age-related need for psychological support in undergraduates. Future studies could explore the coping strategies incorporated to ensure the sound mental health of radiography undergraduates. Contributors: All authors contributed to the conception or design of the work, the acquisition, analysis, or interpretation of the data. All authors were involved in drafting and commenting on the paper and have approved the final version. Funding: This study did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Competing interests: All authors declare no conflict of interest. Ethical approval: Ethical approval was obtained from the Ethics Review Committee of the Faculty of Allied Health Sciences, University of Peradeniya (AHS/ERC/2021/106). Consent for participation was taken at the time of enrollment. Data availability statement: Data are available upon request. ==== Refs References 1 World Health Organization, Statement on the second meeting of the International Health Regulations (2005) Emergency committee regarding the outbreak of novel coronavirus (2019-nCoV), World Health Organization. 2020 [cited 2022 Sep 26]; Available from: https://www.who.int/news/item/30-01-2020-statement-on-the-second-meeting-of-the-international-health-regulations-(2005)-emergency-committee-regarding-the-outbreak-of-novel-coronavirus-(2019-ncov) 2 Ayouni I. Maatoug J. Dhouib W. Zammit N. Ben F.S. Ghammam R. Effective public health measures to mitigate the spread of COVID-19: a systematic review BMC Public Health 21 1 2021 [Internet]Dec 1 [cited 2022 Sep 26]Available from/pmc/articles/PMC8164261/ 3 Farsalinos K. Poulas K. Kouretas D. Vantarakis A. Leotsinidis M. Kouvelas D. Improved strategies to counter the COVID-19 pandemic: Lockdowns vs. primary and community healthcare Toxicol Rep. 8 2021 1 9 Jan 1 33294384 4 COVID-19 Sri Lanka strategic preparedness & response plan 2021-ministry of health Sri Lanka 2021 [Internet]Available from http://www.health.gov.lk 5 Erandi K. Mahasinghe A.C. Perera S.S.N. Jayasinghe S. Effectiveness of the strategies implemented in Sri Lanka for controlling the COVID-19 outbreak J Appl Math 2020 2020 6 R. Hayashi, M. Garcia, A. Maddawin, KP. Hewagamage Online Learning in Sri Lanka's higher education institutions during the COVID-19 pandemic [Internet]. Asian Develpment Bank BRIEFS NO. 151. 2020. Available from: https://www.adb.org/publications/series/adb-briefs 7 Subashini N. Udayanga L. De Silva L.H.N. Edirisinghe J.C. Nafla M.N. Undergraduate perceptions on transitioning into E-learning for continuation of higher education during the COVID pandemic in a developing country: a cross-sectional study from Sri Lanka BMC Med Educ 22 1 2022 1 12 [Internet]Dec 1 [cited 2022 Sep 26]Available from https://bmcmededuc.biomedcentral.com/articles/10.1186/s12909-022-03586-2 34980091 8 Chandrasiri N.R. Weerakoon B.S. Online learning during the COVID-19 pandemic: perceptions of allied health sciences undergraduates Radiography 28 2 2022 545 549 May 1 34893435 9 Ghazawy E.R. Ewis A.A. Mahfouz E.M. Khalil D.M. Arafa A. Mohammed Z. Psychological impacts of COVID-19 pandemic on the university students in Egypt Health Promot Int 36 4 2021 1116 1125 33367587 10 Debowska A. Horeczy B. Boduszek D. Dolinski D. A repeated cross-sectional survey assessing university students’ stress, depression, anxiety, and suicidality in the early stages of the COVID-19 pandemic in Poland Psychol Med 2020 2020 3 6 11 Maia B.R. Dias PC. Anxiety, depression and stress in university students: the impact of COVID-19 Estud Psicol 37 2020 1 8 12 Wang C. Zhao H. The impact of COVID-19 on anxiety in Chinese university students Front Psychol 11 May 2020 1168 May 22 32574244 13 Haththotuwa P. Rupasinghe R. Adapting to online learning in higher education system during the COVID-19 pandemic: a case study of universities in Sri Lanka Sri Lanka J Soc Sci Humanit 1 2 2021 147 14 Nafrees A.C.M. Roshan A.M.F. Baanu A.N. Nihma M.N.F. Shibly F.H.A. Awareness of online learning of undergraduates during COVID 19 with special reference to South Eastern University of Sri Lanka J Phys Conf Ser 1712 1 2020 15 Tay Y.X. Sng L.H. Chow H.C. Zainuldin M.R. Clinical placements for undergraduate diagnostic radiography students amidst the COVID-19 pandemic in Singapore: Preparation, challenges and strategies for safe resumption J Med Imaging Radiat Sci 51 4 2020 560 566 [Internet]Dec 1 [cited 2022 Sep 27]Available from/pmc/articles/PMC7434406/ 32868260 16 Rainford L.A. Zanardo M. Buissink C. Decoster R. Hennessy W. Knapp K. The impact of COVID-19 upon student radiographers and clinical training Radiography 27 2 2021 464 474 [Internet]May 1 [cited 2022 Sep 27]Available from/pmc/articles/PMC7834574/ 33223416 17 Teo L.W. Pang T. Ong Y.J. Lai C. Coping with COVID-19: perspectives of student radiographers J Med Imaging Radiat Sci 51 2020 358 360 Elsevier[cited 2022 Sep 27]Available from/pmc/articles/PMC7256620/ 32571652 18 Li Y. Wang Y. Jiang J. Valdimarsdóttir U.A. Fall K. Fang F. Psychological distress among health professional students during the COVID-19 outbreak Psychol Med 51 11 2021 1952 1954 [Internet]Aug 1 [cited 2022 Sep 27]Available from https://pubmed.ncbi.nlm.nih.gov/32389148/ 32389148 19 Gkatzia N. Dousi M. Syrgiamiotis V. Kechagias D. Stress GL. Anxiety and depression among radiography students during the Covid-19 pandemic Int J Sci Res 2021 74 76 November 20 R.M. López, V. Carales Supporting college students through a public health crisis: lessons learned from hurricane harvey [Internet]. 2020 [cited 2022 Sep 30]. Available from: https://www.higheredtoday.org/2020/04/20/supporting-college-students-public-health-crisis-lessons-learned-hurricane-harvey/ 21 Abdel Wahed W.Y. Hassan S.K. Prevalence and associated factors of stress, anxiety and depression among medical Fayoum University students Alex J Med 53 1 2017 77 84 Mar 1 22 Mirzaei M. Ardekani S.M.Y. Mirzaei M. Dehghani A. Prevalence of depression, anxiety and stress among adult population: results of yazd health study Iran J Psychiatry 14 2 2019 137 146 [Internet][cited 2022 Nov 25]Available from/pmc/articles/PMC6702282/ 31440295 23 Lawson Jones G. York H. Lawal O. Cherrill R. Mercer S. McCarthy Z. The experience of diagnostic radiography students during the early stages of the COVID-19 pandemic – a cross-sectional study J Med Radiat Sci 68 4 2021 418 425 [Internet]Dec 1 [cited 2022 Sep 29]Available from https://onlinelibrary.wiley.com/doi/full/10.1002/jmrs.544 34482617 24 Elshami W. Abuzaid M.M. McConnell J. Stewart S. Floyd M. Hughes D. The radiography students’ perspective of the impact of COVID-19 on education and training internationally: a across sectional survey of the UK devolved nations (UKDN) and the United Arab Emirates (UAE) Radiography 2022 [Internet][cited 2022 Sep 30]Available from/pmc/articles/PMC9293787/ 25 Verma H. Verma G. Kumar P. Depression, anxiety, and stress during times of COVID-19: an analysis of youngsters studying in higher education in India Rev Socionetw Strateg 15 2 2021 471 488 [Internet]Sep 27 [cited 2022 Oct 2]Available from https://link.springer.com/article/10.1007/s12626-021-00089-2 26 T. Dass, T. Dass, T. Dass, D. Download, D. Order Depression, anxiety, stress scales (DASS) [Internet]. 2020 [cited 2022 Apr 8]. p. 2020. Available from: http://www2.psy.unsw.edu.au/dass/ 27 IBM Corp. Released IBM SPSS statistics for windows 2011 IBM Corp Armonk, NY Version 20.0[Internet]Available from https://www-01.ibm.com/support/docview.wss?uid=swg21476197 28 Miles J. Shevlin M. Watson P. Applying regression and correlation: a guide for students and researchers Rules of thumb on magnitudes of effect sizes: MRC cognition and brain sciences unit 2019 Sage 2001 29 Shuja K.H. Aqeel M. Jaffar A. Ahmed A. Covid-19 pandemic and impending global mental health implications Psychiatr Danub 32 1 2020 32 35 32303027 30 Holm-Hadulla R.M. Koutsoukou-Argyraki A. Mental health of students in a globalized world: prevalence of complaints and disorders, methods and effectivity of counseling, structure of mental health services for students Ment Heal Prev 3 1–2 2015 1 4 10.1016/j.mhp.2015.04.003 [Internet]Available from 31 R. Hewitt The impact of coronavirus on higher education [Internet]. 2020 [cited 2022 Oct 3]. Available from: https://www.timeshighereducation.com/hub/keystone-academic-solutions/p/impact-coronavirus-higher-education 32 Cavallo J.J. Forman HP. The economic impact of the COVID-19 pandemic on radiology practices Radiology 296 3 2020 E141 E144 [Internet]Sep 1 [cited 2022 Oct 4]Available from/pmc/articles/PMC7233391/ 32293225 33 Rehman U. Shahnawaz M.G. Khan N.H. Kharshiing K.D. Khursheed M. Gupta K. Depression, anxiety and stress among indians in times of COVID-19 Lockdown Community Ment Health J 57 1 2021 42 48 10.1007/s10597-020-00664-x [Internet]Available from 32577997 34 Basheti I.A. Mhaidat Q.N. Mhaidat HN. Prevalence of anxiety and depression during COVID-19 pandemic among healthcare students in Jordan and its effect on their learning process: a national survey PLoS One Public Libr Sci 16 2021 e0249716 [cited 2022 Apr 14]Available from https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0249716 35 Cao W. Fang Z. Hou G. Han M. Xu X. Dong J. The psychological impact of the COVID-19 epidemic on college students in China Psychiatry Res 287 March 2020 112934 10.1016/j.psychres.2020.112934 [Internet]Available from 36 Aristovnik A. Keržič D. Ravšelj D. Tomaževič N. Umek L. Impacts of the COVID-19 pandemic on life of higher education students: a global perspective Sustain 12 20 2020 1 34 [Internet]Aug 19 [cited 2022 Apr 21]Available from https://www.preprints.org/manuscript/202008.0246/v2 37 Fawaz M. Samaha A. E-learning: depression, anxiety, and stress symptomatology among lebanese university students during COVID-19 quarantine Nurs Forum 56 1 2021 52 57 [Internet]Jan 1 [cited 2022 Apr 21]Available from https://onlinelibrary.wiley.com/doi/full/10.1111/nuf.12521 33125744 38 Guo K. Zhang X. Bai S. Minhat H.S. Nazan A. Feng J. Assessing social support impact on depression, anxiety, and stress among undergraduate students in Shaanxi province during the COVID-19 pandemic of China PLoS One 16 July 2021 e0253891 [Internet]Jul 1 [cited 2022 Apr 21]Available from https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0253891 39 Saravanan C. Mahmoud I. Elshami W. Taha M.H. Knowledge, anxiety, fear, and psychological distress about COVID-19 among university students in the United Arab Emirates Front Psychiatry 11 2020 Oct 22 40 Perera B. Wickramarachchi B. Samanmalie C. Hettiarachchi M. Psychological experiences of healthcare professionals in Sri Lanka during COVID-19 BMC Psychol 9 1 2021 1 9 [Internet]Dec 1 [cited 2022 Oct 4]Available from https://bmcpsychology.biomedcentral.com/articles/10.1186/s40359-021-00526-5 33388086 41 Torun F. Torun S.D. The psychological impact of the COVID-19 pandemic on medical students in Turkey Pak J Med Sci 36 6 2020 1355 1359 32968408 42 Lyons Z. Wilcox H. Leung L. Dearsley O. COVID-19 and the mental well-being of Australian medical students: impact, concerns and coping strategies used Australas Psychiatry 28 6 2020 649 652 32772729 43 Puttakiaw P. Tangjittiporn T. Sakboonyarat B. Hirunviwatgul N. Wittayasai W. The psychological impact and coping among medical students in phramongkutklao college of medicine during the COVID-19 pandemic J Southeast Asian Med Res 6 2022 e0106 September 2021 44 Nguyen H.T. Do B.N. Pham K.M. Kim G.B. Dam H.T.B. Nguyen T.T. Fear of COVID-19 scale—associations of its scores with health literacy and health-related behaviors among medical students Int J Environ Res Public Health 17 11 2020 1 14 45 Alsoufi A. Alsuyihili A. Msherghi A. Elhadi A. Atiyah H. Ashini A. Impact of the COVID-19 pandemic on medical education: medical students’ knowledge, attitudes, and practices regarding electronic learning PLoS One 15 11 November 2020 1 20 10.1371/journal.pone.0242905 [Internet]Available from 46 Shamsuddin K. Fadzil F. Ismail W.S.W. Shah S.A. Omar K. Muhammad N.A. Correlates of depression, anxiety and stress among Malaysian university students Asian J Psychiatr 6 4 2013 318 323 [Internet]Aug [cited 2022 Oct 22]Available from https://pubmed.ncbi.nlm.nih.gov/23810140/ 23810140 47 Appleby J.A. King N. Saunders K.E. Bast A. Rivera D. Byun J. Impact of the COVID-19 pandemic on the experience and mental health of university students studying in Canada and the UK: a cross-sectional study BMJ Open 12 1 2022 50187 [Internet]Jan 24 [cited 2022 Oct 22]Available from/pmc/articles/PMC8787844/ 48 Aslan H. Pekince H. Nursing students’ views on the COVID-19 pandemic and their percieved stress levels Perspect Psychiatr Care 57 2 2021 695 701 [Internet]Apr 1 [cited 2022 Oct 22]Available from/pmc/articles/PMC7461415/ 32808314	0	PMC9715492	NO-CC CODE	2022-12-14 23:52:35	no	J Med Imaging Radiat Sci. 2022 Dec 2; doi: 10.1016/j.jmir.2022.11.014	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.11.014	oa_other
==== Front J Stroke Cerebrovasc Dis J Stroke Cerebrovasc Dis Journal of Stroke and Cerebrovascular Diseases 1052-3057 1532-8511 Elsevier Inc. S1052-3057(22)00614-0 10.1016/j.jstrokecerebrovasdis.2022.106922 106922 Case Report A neonatal case of cerebral venous sinus thrombosis with intrauterine onset after COVID-19 infection during pregnancy: Cause or coincidence? Ozdil Mine a⁎ Cetin Ipek Dokurel b a Department of Pediatrics, Division of Neonatology, Atatürk City Hospital, Balikesir, Turkey b Department of Pediatrics, Division of Pediatric Neurology, Atatürk City Hospital, Balikesir, Turkey ⁎ Corresponding author at: Atatürk City Hospital, Neonatal Intensive Care Unit, Gaziosmanpasa, 209/26, 10100, Altieylul, Balikesir, Turkey. 2 12 2022 2 2023 2 12 2022 32 2 106922106922 24 10 2022 25 11 2022 29 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. Coronavirus 19 disease (COVID-19) is known to predispose patients to increased thrombotic events and the risk is higher in pregnancy which is already a hypercoagulable state. Vertical transmission of the disease during pregnancy was neglected according to data early in the pandemic, however, despite conflicting results from different studies, there is an increasing suspicion of vertical transmission with the rise of new fetal and neonatal cases and perinatal transmission can be higher than expected. An early term neonate, with the history of maternal COVID-19 infection in the start of third trimester, was diagnosed as cerebral venous sinus thrombosis and chronic hemorrhagic ischemia, with intrauterine onset. Keywords Covid-19 Pregnancy Newborn Intracranial thrombosis ==== Body pmcIntroduction Neonatal cerebral venous sinus thrombosis ( CVST) is a rare cause of seizures in the neonatal period and is associated with significant mortality and morbidity.1 , 2 Intrauterine fetal thrombosis is a rarer condition that is generally caused by hereditary thrombophilia, hypoxia, dehydration, or placental malfunction.3 , 4 Coronavirus disease 19 (COVID-19), declared as a pandemic by the World Health Organization in March 2020, has been associated with increased coagulopathy and thromboembolism with adult and pediatric cases of CVST reported in the literature.5 , 6 , 7 , 8 COVID-19 infections during pregnancy have been linked to preterm birth, growth restriction, miscarriage and hypercoagulability compromising fetal placental perfusion leading to thrombotic changes.9 , 10 , 11 Since neonatal COVID-19 infections appear to be mostly the result of horizontal transmission during the postnatal period, it has been assumed that vertical transmission is uncommon, but it has become conceivable in the advanced stages of the pandemic. Herein we present a 3 days-old-neonate diagnosed postnatally with CVST, of a mother with COVID-19 infection during pregnancy. Case report A three-day-old female neonate was transferred to our neonatal intensive care unit due to focal clonic seizures. She was born by cesarean section at 38 gestational weeks, weighing 3350g, to a 42-year-old multiparous mother whose pregnancy was affected by gestational diabetes mellitus and COVID-19 infection. The infection occurred in the initial weeks of the third trimester and the symptoms did not require hospitalization. There was no history of traumatic birth and the child's APGAR scores were 8 and 9 at 1st and 5th minutes, respectively. On the 2nd day of life, the patient was observed to have poor feeding and fever of 37.9°C. On the 3rd day of life, the patient was slightly lethargic and had clonic seizures on the left arm. Hemogram and sepsis work-up including blood and urine were normal. Serum biochemistry data were within the normal range except for magnesium (1,3 mg/dl) which was replaced immediately. Lumbar puncture revealed a xanthochromic cerebrospinal fluid with a cell count of 6 × 10³ cells/μl (mononuclear), protein of 100 mg/dl and glucose of 60 mg/dl. Cerebrospinal fluid remained sterile upon incubation. Intravenous antibiotics were commenced. The blood and urine cultures grew no bacteria. Phenobarbital treatment was started with a loading dose of 20 mg/kg/day and a maintenance dose of 5mg/kg/day was continued thereafter. Recurrent seizures required addition of levetiracetam as a second antiepileptic drug, with the dose of 40 mg/kg/day and seizures were controlled. Electroencephalogram (EEG) showed repetitive focal spike discharges dominant in the right parietal region in compliance with the cortical lesion. Magnetic resonance imaging (MRI) detected a hypointense lesion in T1 weighted images and partial restricted diffusion in diffusion-weighted images in the right parietal lobe, indicating chronic hemorrhagic ischemia (Fig. 1 A and B). The lesion was less visible on FLAIR and T2-weighted-FSE images and could not be identified as an area of infarction on these sequences. Magnetic resonance venography showed lack of flow in the connection of right transverse sinus and confluence sinuum (Fig. 2 A and B). The coagulation tests revealed D-dimer: 1797ng/ml (0-243), PT: 10sec, PT activity: 125%, INR: 0.84, aPTT: 32sec and fibrinogen: 252mg/dl (200-393). The patient and the mother were negative for nasopharyngeal swab COVID-19 RT-PCR testing, but IgM and IgG antibody tests could not be performed due to financial constraints. Evaluation for thrombophilia showed heterozygous A1298C mutation in the methylene-tetrahydrofolate reductase (MTHFR) gene both in the patient and the mother. Homocysteine levels, protein C, protein S and antithrombin levels were all in normal ranges. Antiphospholipid antibodies were negative. The patient was diagnosed as having right transverse sinus venous thrombosis with chronic hemorrhagic ischemia in the right parietal lobe, that occurred in utero at least 14-28 days before birth. Low molecular weight heparin (enoxaparin) was commenced immediately with the dose of 1,7 mg/kg twice a day. The clinically well appearing and seizure-free baby was discharged at 15 days of age with anticonvulsant monotherapy (levetiracetam with a dose of 40 mg/kg/day) and anticoagulation therapy. At 3 months, enoxaparin was ceased and levatirecetam was discontinued after gradually tapering the dose with normalization of EEG at 7 months of age. The control MRI at 7 months was detected to have normal imaging findings (Fig. 1C). In the follow-up period to date, the neurodevelopment of the patient, at 11 months of age, appears to be normal. Informed consent has been obtained from the family of the patient to share the details of their condition.Fig. 1 Magnetic Resonance images of the neonate with the suspicion of stroke. A. Axial Apparent diffusion coefficient (ADC) Magnetic Resonance Diffusion-weighted image showing partial restricted diffusion in the right parietal lobe adjacent to the occipital horn of the lateral ventricle,consistent with chronic ischemic hemorrhage (arrow). B. Axial T1-weighted Magnetic Resonance images showing approximately 16 × 16 mm in size, heterogeneous hypointense lesion with blurred borders in the right parietal lobe adjacent to the occipital horn of the lateral ventricle, consistent with chronic ischemic hemorrhage (arrow). C. The control Axial Apparent diffusion coefficient (ADC) Magnetic Resonance Diffusion-weighted image was normal at 7 months of age. Fig 1 Fig. 2 A. Sagittal non-contrast Magnetic Resonance Venography Inhance 3D Velocity image showing lack of flow in the connection of right transverse sinus and confluence sinuum (arrows). B. Axial T2-weighted image Magnetic Resonance image showing partially absence of flow in the right transverse sinus causing peripheral narrowing of the vein lumen (arrow) Fig 2 Discussion Neonatal CVST is a rare disease, with an incidence ranging from 12 to 47/100,000 term neonates/year, however, comprises nearly 50% of all pediatric CVST causing a remarkable morbidity and mortality.12 , 13 The diagnosis is challenging with a variable clinical presentation and needs a high degree of suspicion. It is likely to result from the combination of predisposing maternal, fetal and neonatal risk factors. These risk factors include congenital thrombophilia, preeclampsia, pregnancy induced/preexisting diabetes, perinatal asphyxia, difficult delivery, sepsis, meningitis, and placental disorders such as thrombosis, infection and fetomaternal hemorrhage.14 , 15 , 16 COVID-19 infection, known to exacerbate thrombophilia, can be speculated in the pathogenesis of neonatal thrombosis owing to further exaggerating the hypercoagulability of pregnancy and probably infecting the fetus itself. The neonate with CVST in our report was born to an unvaccinated mother with a history of COVID-19 infection in the beginning of the third trimester, which may have been the cause of vertical transmission. Several aspects of COVID-19 infection in pregnancy are not yet fully understood, including long-term effects and the possibility of an embryopathy.17 Some placental pathology studies related to SARS-COV-2 during pregnancy have suggested that COVID-19 is related with a tendency towards coagulopathy with possible transplacental impact on the fetus.10 , 11 Furthermore, a systematic review including 1063 pregnant women with COVID-19 during pregnancy reported higher risk of the thromboembolic complications.18 Also a very recent review has demonstrated that COVID-19 placentitis, which may have both an infectious and immunologic basis, cause severe and diffuse placental destruction, interfering with function of oxygenation and leading to stillbirth and neonatal deaths. All mothers reported to have SARS-COV-2 placentitis were unvaccinated demonstrating viremia at some time during pregnancy.19 The possibility of vertical transmission of COVID-19 disease during pregnancy is a topic of big debate. There is insufficient evidence to rule out vertical transmission, and data suggesting its existence have increased as the pandemic has progressed. Cases of fetal demise in the first trimester, IgM positivity at birth, neonatal limb ischemia and CVST have been reported in last two years, pointing to possible vertical transmission.20 , 21 , 22 , 23 The inability to conduct IgM and IgG antibodly tests to confirm previous COVID-19 infection in our patient and the absence of placental pathology data were the primary limitations of the present case. However, we believe IgM antibodies would have been negative, as we hypothesized that the infection took place at least 3 months before birth and presence of IgG antibodies would not be able to distinguish whether it was due to an infection or maternally derived antibodies through the placenta. Our patient was diagnosed as being heterozygous for MTHFR A1298C mutation. The published literature on intrauterine fetal thrombosis is limited to case reports and case series, it is believed that genetic prothrombotic mutations and maternal diseases such as diabetes may play a role.3 , 4 However, MTHFR A1298C heterozygosity is prevalent in Turkish population and is not linked to elevated homocysteine levels and hypercoagulability unless combined with MTHFR C677T mutation.24 Possible effects of COVID-19 on vascular circulation raise concerns for the current neonatal case of CVST due to the possible vertical transmission of COVID-19 infection, increased hypercoagulability and the superposition of gestational diabetes. Author contributions Mine Özdil, MD, contributed to the design of the work; or the acquisition, analysis, interpretation of data for the work and drafting the work revising it critically for important intellectual content; and final approval of the version to be published; and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. İpek Dokurel Çetin, MD, also contributed to the design of the work; the acquisition, analysis, interpretation of data for the work and drafting the work revising it critically for important intellectual content; and final approval of the version to be published; and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Declaration of Competing Interest The authors report no conflict of interest. ==== Refs References 1 Ramenghi LA Cardiello V Rossi A. Neonatal cerebral sinovenous thrombosis Neonat Neurol 2019 267 280 10.1016/B978-0-444-64029-1.00012-6 2 Fitzgerald KC Williams LS Garg BP Cerebral sinovenous thrombosis in the neonate Arch Neurol 63 2006 405 409 10.1001/archneur.63.3.405 16533968 3 Abdelrazeq SA Alkhateeb A Saleh H Intrauterine upper limb ischemia: an unusual presentation of fetal thrombophilia-a case report and review of the literature Case Rep Pediatr 670258 2013 10.1155/2013/670258 4 Khriesat WM Al-Rimawi HS Lataifeh IM Intrauterine upper limb ischemia associated with fetal thrombophilia: a case report and review of the literature Acta Haematol 124 2010 1 4 10.1159/000314680 20501986 5 Thompson A Morgan C Smith P Cerebral venous sinus thrombosis associated with COVID-19 Pract Neurol 21 2020 75 76 10.1136/practneurol-2020-002678 6 Hughes C Nichols T Pike M Cerebral venous sinus thrombosis as a presentation of COVID-19 Eur J Case Rep Intern Med 7 5 2020 001691 10.12890/2020_001691 7 Guendouz C Quenardelle V Riou-Comte N Pathogeny of cerebral venous thrombosis in SARS-Cov-2 infection: case reports Medicine 100 10 2021 e24708 10.1097/MD.0000000000024708 33725828 8 Ippolito Bastidas H Márquez-Pérez T García-Salido A Cerebral venous sinus thrombosis in a pediatric patient with COVID-19 Neurol Clin Pract 11 2 2021 208 210 10.1212/CPJ.0000000000000899 9 Barrero-Castillero A Beam KS Bernardini LB Harvard neonatal-perinatal fellowship COVID-19 working group. COVID-19: neonatal-perinatal perspectives J Perinatol 41 5 2021 940 951 10.1038/s41372-020-00874-x 33293665 10 Baergen RN Heller DS. Placental pathology in Covid-19 positive mothers: preliminary findings Pediatr Dev Pathol 23 2020 177 180 10.1177/1093526620925569 32397896 11 Shanes ED Mithal LB Otero S Placental pathology in COVID-19 Am J Clin Pathol 154 2020 23 32 10.1093/ajcp/aqaa089 32441303 12 deVeber G Andrew M Adams C Canadian Pediatric Ischemic Stroke Study Group. Cerebral sinovenous thrombosis in children N Engl J Med 345 6 2001 417 423 10.1056/NEJM200108093450604 11496852 13 Raets MM Sol JJ Govaert P Serial cranial US for detection of cerebral sinovenous thrombosis in preterm infants Radiology 269 3 2013 879 886 10.1148/radiol.13130401 23985276 14 Berfelo FJ Kersbergen KJ van Ommen CH Neonatal cerebral sinovenous thrombosis from symptom to outcome Stroke 41 7 2010 1382 1388 10.1161/STROKEAHA.110.583542 20522810 15 Chabrier S Husson B Dinomais M Landrieu P Nguyen The Tich S. New insights (and new interrogations) in perinatal arterial ischemic stroke Thromb Res 127 1 2011 13 22 10.1016/j.thromres.2010.10.003 21055794 16 Rutherford MA Ramenghi LA Cowan FM. Neonatal stroke Arch Dis Child Fetal Neonatal Ed 97 5 2012 377 384 10.1136/fetalneonatal-2010-196451 17 Morhart P Mardin C Rauh M Maternal SARS-CoV-2 infection during pregnancy: possible impact on the infant Eur J Pediatr 181 1 2022 413 418 10.1007/s00431-021-04221-w 34355278 18 Servante J Swallow G Thornton JG Haemostatic and thrombo-embolic complications in pregnant women with COVID-19: a systematic review and critical analysis BMC Pregnancy Childbirth 21 1 2021 108 10.1186/s12884-021-03568-0 33546624 19 Schwartz DA Mulkey SB Roberts DJ. SARS-CoV-2 placentitis, stillbirth, and maternal COVID-19 vaccination: clinical-pathologic correlations Am J Obstet Gynecol 2022 10.1016/j.ajog.2022.10.001 S0002-9378(22)00800-6 20 Shende P Gaikwad P Gandhewar M Persistence of SARS-CoV-2 in the first trimester placenta leading to transplacental transmission and fetal demise from an asymptomatic mother Hum Reprod 36 2021 899 906 10.1093/humrep/deaa36 33346816 21 Bandyopadhyay T Sharma A Kumari P Possible early vertical transmission of COVID-19 from an infected pregnant female to her neonate: a case report J Trop Pediatr 0 2020 1 4 10.1093/tropej/fmaa094 22 Campi F Longo D Bersani I Neonatal cerebral venous thrombosis following maternal SARS-CoV-2 infection in pregnancy Neonatology 119 2 2022 268 272 10.1159/000520537 35220305 23 Dong L Tian J He S Possible vertical transmission of SARS-CoV-2 from an infected mother to her newborn JAMA 323 18 2020 1846 1848 10.1001/jama.2020.4621 32215581 24 Sazci A Ergul E Kaya G Genotype and allele frequencies of the polymorphic methylenetetrahydrofolate reductase gene in Turkey Cell Biochem Funct 23 2005 51 54 10.1002/cbf.1132 15386535	36493705	PMC9715493	NO-CC CODE	2022-12-06 23:15:47	no	J Stroke Cerebrovasc Dis. 2023 Feb 2; 32(2):106922	utf-8	J Stroke Cerebrovasc Dis	2,022	10.1016/j.jstrokecerebrovasdis.2022.106922	oa_other
==== Front Eur J Intern Med Eur J Intern Med European Journal of Internal Medicine 0953-6205 1879-0828 Published by Elsevier B.V. on behalf of European Federation of Internal Medicine. S0953-6205(22)00428-9 10.1016/j.ejim.2022.11.033 Letter to the Editor Lack of efficacy of Interferon β-1a in COVID-19 patients with mild to moderate pneumonia Bosi Carlo 1 Ferrarese Roberto 2 De Lorenzo Rebecca 13 Mancini Nicasio 2 Bosi Emanuele 13⁎ for the INTERCOP study group# 1 Unit of Internal Medicine, IRCCS San Raffaele Hospital, Milan, Italy 2 Laboratory of Medical Microbiology and Virology, Vita-Salute San Raffaele University 3 School of Internal Medicine, Vita-Salute San Raffaele University, Milan, Italy ⁎ Corresponding author. # INTERCOP study group members are listed in the acknowledgments 2 12 2022 2 12 2022 31 10 2022 28 11 2022 30 11 2022 © 2022 Published by Elsevier B.V. on behalf of European Federation of Internal Medicine. 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. Keywords SARS-CoV-2 COVID-19 IFNβ-1a type-I interferons repurposed drugs ==== Body pmcDear Editor, Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome virus 2 (SARS-CoV-2), poses an unprecedented threat for global health. Even in the context of intensified vaccine rollout, identifying efficacious therapeutics remains a priority in case of severe forms of COVID-19. During the early phases of the pandemic, efforts were addressed at investigating repurposed therapeutic agents with antiviral and/or immuno-modulating properties, culminating in a broad multi-arm international trial (WHO SOLIDARITY) testing, among others, Interferon β-1a (IFNβ-1a) 1, a recombinant pharmaceutical compound of the human cytokine IFNβ belonging to the type I interferon family. In contrast to evidence from smaller retrospective studies and clinical trials 2 , 3, the SOLIDARITY and the NIH-sponsored ACTT-3 trial 4 reported no added benefits with IFNβ-1a in hospitalized patients. A subsequent trial testing higher doses of IFNβ-1a also demonstrated no efficacy 5. In the first place, IFNβ-1a stood as a promising candidate for different reasons. First, IFNβ-1a displays antiviral activity against SARS-CoV-2 in vitro6. Second, defective or dysregulated type I interferon responses have emerged as a hallmark of SARS-CoV-2 infection7. Moreover, observational and functional studies have suggested that loss of type I IFN responses, either caused by neutralizing autoantibodies directed against type IFNs8, or by inherited gene mutations9, is associated with higher risk of severe COVID-19. Based on this rationale, between 2020 and 2021, in concomitance with the Solidarity and the ACTT-3 trials, we conducted a single-center, randomized, controlled, open-label, phase 2 clinical trial to evaluate efficacy and safety and antiviral activity of IFNβ-1a in hospitalized patients with COVID-19 radiologically diagnosed pneumonia (INTERCOP study, NCT04449380; EudraCT 2020-002458-25). The study protocol was previously reported10. People aged 18 years or older, diagnosed at the Emergency Department of San Raffaele Hospital (Milan, Italy) with COVID-19 by RT-PCR and pneumonia by chest radiography or computed tomography, with a clinical status of non-severe disease as defined by categories 3 and 4 of a 7-point clinical severity ordinal scale, were considered for enrolment. The scale consisted of the following: 1) not hospitalized, with resumption of normal activities; 2) not hospitalized, but unable to resume normal activities; 3) hospitalized, not requiring supplemental oxygen; 4) hospitalized, requiring supplemental oxygen; 5) hospitalized, requiring high-flow oxygen therapy, non-invasive mechanical ventilation, or both; 6) hospitalized, requiring ECMO, invasive mechanical ventilation, or both; 7) death. The protocol and consent documents were approved by the Ethics Committees of the San Raffaele Hospital, Milan and of the Lazzaro Spallanzani Institute for Infectious Diseases, Rome, Italy (the national center oversighting COVID-19 clinical trials in Italy) and by the Italian Medicines Agency (AIFA). All patients or their legally authorized representative provided written informed consent. The trial has been conducted in compliance with the principles of the Declaration of Helsinki. Eligible patients were randomly assigned 2:1 to either receive interferon IFNβ-1a (Rebif®, Merck Serono, Rome, Italy), 44 micrograms 3-times per week for two weeks, at least 48 hours apart in addition to standard of care, or standard of care alone. The experimental drug was administered over a maximum range of 14 days or until negative conversion of nasal swab. The primary outcome measure was the time to negative conversion of nasopharyngeal swabs for SARS-CoV-2, defined as the date of the first negative, confirmed, nasal swab. Nasopharyngeal swabs were taken every other day from baseline until day 15, then at day 21 and at day 29, as per protocol. The full set of the secondary and exploratory outcomes is reported in the Protocol10. Efficacy and safety were evaluated up to day 29. Evaluation was performed during hospitalization and, in case of discharge in the meanwhile, in a dedicated outpatient clinic at weekly intervals. In patients still positive for SARS-CoV-2 infection at day 29, additional nasopharyngeal swabs were offered as extra-follow-up tests and included in the analyses. Albeit unplanned, an interim analysis was requested by the Principal Investigator at the end of the second epidemic wave and in coincidence with the publication of the interim results of the Solidarity Trial showing no significant clinical effects of IFNβ-1a in COVID-19 patients. The trial was prematurely terminated for futility: 53 patients (44% of the planned cohort) were recruited at the time of trial termination: median age was 67 yr (IQR 51-80), 32 were male, 50 (94%) with no oxygen supplementation and 3 (6%) on low flow oxygen at baseline. There were 4 deaths (7.5% of the total cohort): 2 (5.7%) in the IFNβ-1a intervention and 2 (11.1%) in the control arm. Median time to negative conversion of nasopharyngeal swabs (primary outcome of the study) was 23 and 20 days in the intervention and control arms, respectively (p=0.19; hazard ratio for negative conversion 0.63 [95% CI 0.32-1.26]) (figure 1 ). No significant differences emerged in the secondary outcomes. Overall, adverse events occurred in 28 (82%) and 6 (33%) patients in the intervention and control arms, respectively, being fever the most commonly event reported.Figure 1 Time to negative conversion of nasopharyngeal swabs in Covid-19 patients treated with IFNβ-1a in addition to standard of care vs standard of care alone (shaded areas indicate 95% C.I.) Figure 1 In conclusion, subcutaneous IFNβ-1a does not reduce the duration of infection of SARS-CoV-2 and does not ameliorate the clinical course of COVID-19. Acknowledgments INTERCOP study group Collaborators: Luigi Di Filippo, Nicola Farina, Luigi Nocera, Nicola Clementi, Benedetta Sposito, Diego Palumbo, Francesco De Cobelli, Patrizia Rovere-Querini, Marco Bregni, Fabio Ciceri, Moreno Tresoldi, Antonella Castagna, Andrea Giustina, Massimo Clementi, Sofia Sisti, Marco Tonelli, Giordano Vitali, Sabina Martinenghi, Giuseppe Di Lucca, Junaid Mushtaq, Renato Pennella, Stefania Del Rosso. Merck Serono (Rome, Italy), a division of Merck KGaA, Darmstadt, Germany, provided the experimental medicinal product Rebif® free of charge. The company had no role in the study design, conduct, data analysis and interpretation. Funding This work was supported by IRCCS Ospedale San Raffaele with internal funds dedicated to COVID-19. ==== Refs References 1 Repurposed Antiviral Drugs for Covid-19 — Interim WHO Solidarity Trial Results. N Engl J Med. 2021;384(6):497-511. doi:10.1056/nejmoa2023184 2 Alavi Darazam I Shokouhi S Pourhoseingholi MA Role of interferon therapy in severe COVID-19: the COVIFERON randomized controlled trial Sci Reports 2021 111 11 1 2021 1 11 10.1038/s41598-021-86859-y 3 Fallahzadeh M Pourhoseingholi MA Boroujeni MG Study of the effects of interferon β−1a on hospitalized patients with COVID-19: SBMU Taskforce on the COVIFERON study J Med Virol 94 4 2022 1488 1493 10.1002/JMV.27475 34821387 4 Kalil AC Mehta AK Patterson TF Efficacy of interferon beta-1a plus remdesivir compared with remdesivir alone in hospitalised adults with COVID-19: a double-bind, randomised, placebo-controlled, phase 3 trial Lancet Respir Med 2021 1 12 10.1016/s2213-2600(21)00384-2 33341156 5 Alavi Darazam I Hatami F Mahdi Rabiei M An investigation into the beneficial effects of high-dose interferon beta 1-a, compared to low-dose interferon beta 1-a in severe COVID-19: The COVIFERON II randomized controlled trial Int Immunopharmacol 99 2021 107916 10.1016/J.INTIMP.2021.107916 6 Clementi N Ferrarese R Criscuolo E Interferon-β-1a Inhibition of Severe Acute Respiratory Syndrome–Coronavirus 2 In Vitro When Administered After Virus Infection J Infect Dis 222 5 2020 722 725 10.1093/INFDIS/JIAA350 32559285 7 Galani IE Rovina N Lampropoulou V Untuned antiviral immunity in COVID-19 revealed by temporal type I/III interferon patterns and flu comparison Nat Immunol 22 1 2021 32 40 10.1038/s41590-020-00840-x 33277638 8 Bastard P Rosen LB Zhang Q Autoantibodies against type I IFNs in patients with life-threatening COVID-19 Science (80-) 370 6515 2020 10.1126/SCIENCE.ABD4585/SUPPL_FILE/ABD4585_MDAR-REPRODUCIBILITYCHECKLIST.PDF 9 Zhang Q Bastard P Liu Z Inborn errors of type I IFN immunity in patients with life-threatening COVID-19 Science (80-) September 2020 10.1126/science.abd4570 10 Bosi E Bosi C Rovere Querini P Interferon β-1a (IFNβ-1a) in COVID-19 patients (INTERCOP): study protocol for a randomized controlled trial Trials 21 1 2020 10.1186/S13063-020-04864-4	36517370	PMC9715494	NO-CC CODE	2022-12-12 23:20:30	no	Eur J Intern Med. 2022 Dec 2; doi: 10.1016/j.ejim.2022.11.033	utf-8	Eur J Intern Med	2,022	10.1016/j.ejim.2022.11.033	oa_other
==== Front Environ Int Environ Int Environment International 0160-4120 1873-6750 Published by Elsevier Ltd. S0160-4120(22)00602-X 10.1016/j.envint.2022.107675 107675 Full Length Article Air Pollution and Meteorology as Risk Factors for COVID-19 Death in a Cohort from Southern California Jerrett Michael a⁎ Nau Claudia L. b Young Deborah R. b Butler Rebecca K. b Batteate Christina M. a Su Jason c Burnett Richard T. d Kleeman Michael J. e a Department of Environmental Health Sciences, University of California, Los Angeles 650 Charles Young Dr. S., 56-070 CHS Box 951772, Los Angeles, CA, 90095 b Department of Research & Evaluation, Kaiser Permanente Southern California 100 S. Los Robles Ave., 5th Floor, Pasadena, CA 91101 c Division of Environmental Health Sciences, School of Public Health, University of California, Berkeley 2121 Berkeley Way, Room 5302, Berkeley, CA 94720 d Population Studies Division, Environmental Health Directorate, Health Canada 251 Sir Frederick Banting Driveway, Ottawa, Ontario, Canada K1A 0K9 e Department of Civil and Environmental Engineering, University of California, Davis 1 Sheilds Avenue, Davis CA 95616 ⁎ Corresponding author. 2 12 2022 2 12 2022 10767511 7 2022 18 11 2022 1 12 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. Background Recent evidence links ambient air pollution to COVID-19 incidence, severity, and death, but few studies have analyzed individual-level mortality data with high quality exposure models. Methods We sought to assess whether higher air pollution exposures led to greater risk of death during or after hospitalization in confirmed COVID-19 cases among patients who were members of the Kaiser Permanente Southern California (KPSC) healthcare system (N = 21,415 between 06-01-2020 and 01-31-2022 of whom 99.85% were unvaccinated during the study period). We used 1 km resolution chemical transport models to estimate ambient concentrations of several common air pollutants, including ozone, nitrogen dioxide, and fine particle matter (PM2.5). We also derived estimates of pollutant exposures from ultra-fine particulate matter (PM0.1), PM chemical species, and PM sources. We employed Cox proportional hazards models to assess associations between air pollution exposures and death from COVID-19 among hospitalized patients. Findings We found significant associations between COVID-19 death and several air pollution exposures, including: PM2.5 mass, PM0.1 mass, PM2.5 nitrates, PM2.5 elemental carbon, PM2.5 on-road diesel, and PM2.5 on-road gasoline. Based on the interquartile (IQR) exposure increment, effect sizes ranged from hazard ratios (HR) = 1.12 for PM2.5 mass and PM2.5 nitrate to HR ∼ 1.06-1.07 for other species or source markers. Humidity and temperature in the month of diagnosis were also significant negative predictors of COVID-19 death and negative modifiers of the air pollution effects. Interpretation Particulate air pollution exposures associated with greater risk of COVID-19 death in a cohort of patients in Southern California. These findings have implications for prevention of death from COVID-19 and for future pandemics. Keywords COVID-19 death cohort study air pollution Southern California ==== Body pmc1 Introduction The COVID-19 pandemic represents one of the largest threats to population health in more than a century. Currently, more than 624 million people worldwide have been diagnosed with COVID-19, resulting in more than 6.5 million deaths. (World Health Organization. WHO Coronavirus Disease (COVID-19) Dashboard \| WHO Coronavirus Disease (COVID-19) Dashboard., 2022) Researchers have extensively investigated the etiology of COVID-19, yet considerable uncertainties remain on how risk factors may influence COVID-19 incidence, severity, and death. Recent evidence from the North America, Asia, and Europe implicates air pollution as a risk factor for COVID-19 incidence, prognosis, and death. (Brandt et al., 2020, Li et al., 2020, Wu et al., 2020, Zhang et al., 2020, Zhu et al., 2020, Lippi et al., 2019, Coker et al., 2020, Wang et al., 2020, Travaglio et al., 2021, Yao et al., 2021, Huang et al., 2021, Chen et al., 2021, Berg et al., 2021, Zhou et al., 2021) Biologically plausible mechanisms suggest air pollution exposure may render people more susceptible to contracting COVID-19, and once they have the infection, higher air pollution exposure may worsen their prognosis. (Rebuli et al., 2021, Wang et al., 2020, De Angelis et al., 2021, Kifer et al., 2021, SanJuan-Reyes et al., 2021) Nitrogen dioxide (NO2), a marker for traffic-related air pollution (Zeldovich, 2015, Quiros et al., 2013), likely increases the risk of lung infections by impairing the function of alveolar macrophages and epithelial cells in the lung. (Neupane et al., 2010)The findings from these epidemiological and toxicological studies align with a large body of research linking air pollution to risk of respiratory viral and bacterial infection (Wang et al., 2020, Ciencewicki and Jaspers, 2007), respiratory chronic morbidities (e.g., asthma, chronic pulmonary disease, lung cancer) (Jerrett et al., 2008, Bai et al., 2018, Sydbom et al., 2001), hospitalizations (Neupane et al., 2010) , and mortality. (Jerrett et al., 2005, Beelen et al., 2008, Bozack et al., 2021) In reviewing the growing literature on air pollution exposure and COVID-19 outcomes, we found only four other mortality studies have used individual level data with some levels of control for potential confounders. (Chen et al., 2021, Bozack et al., 2021, Elliott et al., 2021, Nobile et al., 2022, Chen et al., 2022) These studies focused on the early phases of the pandemic, which may have led to lower statistical power due to a relatively small number of deaths. While some of the mortality studies used high-quality exposure estimates, none assessed source contributions or ultrafine particle concentrations. In addition, none of these studies examined interactions between air pollution and meteorological variables such as temperature and humidity. Here we expand the evidence base with a large sample of individual data, a longer study period, exposure models capable of assessing particle species and sources, and meteorological variables. In this context, we addressed two research objectives. Firstly, we assessed whether higher air pollution exposures led to greater risk of death in confirmed COVID-19 cases among patients who were members of the Kaiser Permanente Southern California (KPSC) healthcare system. Secondly, we investigated whether meteorology variables influenced the risk of COVID-19 death or modified associations between air pollution and COVID-19 death. 2 Materials And Methods 2.1 KPSC Cohort and Health Data KPSC is a large integrated health care system with a racially/ethnically and socioeconomically diverse membership of 4.7 million people, living across nine southern California counties. KPSC’s membership approximately represents the underlying population of the second largest urban region in the United States; further details of the KPSC membership are described elsewhere. (Koebnick et al., 2012) KPSC’s Electronic Health Record (EHR) is an integrated data system that captures all aspects of patient care, including diagnoses, inpatient and outpatient encounters, pharmacy encounters, and laboratory tests. Clinical care changed rapidly during the first months of the pandemic. We therefore began our observation period on 06/01/2020 when new standards of COVID care, such as lying patients in the prone position, had become more common. We identified patients with KPSC COVID-19 molecular diagnostic tests and diagnoses (ICD-10 codes: J12.89, J20.8, J22, J80, B34.2, B97.29, U07.1) from 06/01/2020 to 1/30/2021. We include both diagnoses and COVID-19 tests because patients may have been tested outside of KPSC and received a diagnosis at KPSC without being re-tested. The study cohort is comprised of patients who were 18 years or older at the time of diagnoses or positive COVID-19 test. We limited our sample to members who had at least 1 year of membership before their COVID-19 diagnoses/test to reliably assess co-morbidities. We defined COVID-19 hospitalizations as hospitalizations occurring within 21 days of COVID-19 diagnoses or positive test (N = 316,224). (Nau C, Bruxvoort K, Navarro RA, et al. COVID-19 Inequities Across Multiple Racial and Ethnic Groups: Results From an Integrated Health Care Organization. Annals of internal medicine, 2021) We used hospitalized patients from the cohort rather from all those who tested positive because testing could have occurred after possible contact with an infected person or after the onset of severe illness at the point of hospital admission. This would result in uncertainty about the time window at which the test could have occurred among different study patients that would introduce substantial errors in our follow up times, which would lead to biased results in the statistical models. Restricting to those hospitalized eliminated this potential problem, as no uncertainty existed in the time of hospitalization. After applying eligibility and exclusion criteria, the analytic cohort consisted of 21,415. Deaths were included up to 90 days after the initial hospitalization (see Online Data Supplement [ODS] for further details on death ascertainment). We excluded patients who lost membership during our 90-observation window and who were hospitalized for childbirth. This study was approved by the Kaiser Permanente Institutional Review Board. The KPSC EHR provides information on patient age and sex. Member race/ethnicity categories have been created using a validated algorithm that uses multiple data sources. (Nau C, Bruxvoort K, Navarro RA, et al. COVID-19 Inequities Across Multiple Racial and Ethnic Groups: Results From an Integrated Health Care Organization. Annals of internal medicine, 2021) Body mass index (BMI: kg/m2) has been found to be an important risk factor for COVID-19 mortality. (Tartof et al., 2020) The most recent BMI available in the EHR was used to adjust for this potential confounder. (Tartof et al., 2020) We cleaned BMI data using validated algorithms to delete biologically implausible values. Five broad comorbidity categories that have been used in prior COVID-19 research were created to identify co-morbidities that may increase a person’s risk of severe COVID-19 outcomes. (Nau C, Bruxvoort K, Navarro RA, et al. COVID-19 Inequities Across Multiple Racial and Ethnic Groups: Results From an Integrated Health Care Organization. Annals of internal medicine, 2021, Quan et al., 2005) We use Elixhauser disease categories to create COVID-19 relevant disease categories (see ODS for further details). Smoking status and the Exercise Vital Sign (EVS) data are collected during each KPSC in-person outpatient health care encounter. Smoking status was coded (ever or never) based on the information provided during the last encounter before the COVID-19 test/diagnoses reaching back up to four years. The EVS queries on usual exercise is coded in the EHR in minutes/week of moderate to vigorous exercise. All EVS information for the past four years was identified for every patient. The median value of minutes of exercise per week was calculated and used in our analysis. (Young et al., 2018, Zhou et al., 2021) We identified patients who were enrolled at KPSC via MediCal to identify patients with very low income. In sum, four individual-level confounders were considered: smoking status, BMI, Medicaid (low income), and EVS. We queried vaccination status and found only 33 members of our cohort were vaccinated prior to hospitalization; thus about 99.85% of the cohort was unvaccinated during the study period. We also followed common practice in analyses of EHR data and added predictors of community-level SES to help proxy individual SES and adjust for community level effects of social determinants of health. (Krieger, 1992, Diez-Roux et al., 2001, Geronimus and Bound, 1998) Community-level predictors at the census block-group level were drawn from the American Community Survey 2018. (Messer et al., 2006) They include a validated neighborhood deprivation index (NDI), a measure of crowding (the proportion of households with more than one occupant per home), and the proportion of workers aged 16 and older who commute to work via public transportation. (Messer et al., 2006) GridMET meteorological data are high-spatial resolution (∼4-km) surface meteorological data covering the contiguous U.S. We acquired GridMET daily maximum temperature and daily maximum relative humidity for our entire study period through Google Earth Engine (https://developers.google.com/earth-engine/datasets/catalog/IDAHO_EPSCOR_GRIDMET?hl=en). We aggregated the GridMET data to monthly means for the home address of every study participant up to the month where they were hospitalized with COVID-19. 2.2 Exposure Assessment: Chemical Transport Model Exposure simulations were carried out across California using the source-oriented UC Davis-California Institute of Technology (UCD-CIT) 3D reactive chemical transport model (CTM). (Venecek et al., 2019) The UCD/CIT model predicts the evolution of gas and particle phase pollutants in the atmosphere in the presence of emissions, transport, deposition, chemical reaction, and phase change. The pressing timeline for the current study during an ongoing public health crisis necessitated leveraging past efforts that prepared and validated CTM inputs. We previously reported CTM exposure fields with 4km resolution over California for the years 2000-2016. (Yu et al., 2019) The most recent year (i.e., 2016) of this time window was selected as the starting point to characterize chronic exposure in the current study. Meteorology and emissions inputs for the year 2016 were downscaled to improve spatial resolution to 1km. Bias in the raw CTM output fields was removed using a constrained regression model based on source apportionment tags and the difference between predicted and measured concentrations. See ODS for further details on CTM methods. CTM predictions include a wide range of pollutants. For our study area, we estimated PM2.5 total mass, PM2.5 nitrates, PM2.5 organic carbon (OC), PM2.5 elemental carbon (EC), ultra-fine particle mass or PM0.1 for particles with diameters of 100 nanometers or less, nitrogen dioxide (NO2), and ozone (O3). We also extracted source tracers for on-road diesel, on-road gasoline, and biomass burning. These exposure fields were assigned to the geocoded home address of the cohort members. Although the exposure fields were restricted to 2016, we accounted for population mobility by assigning exposures to each address for any member of the cohort who moved in the past five years. We then did time-weighted averaging of the exposures to account for mobility effects for those who had moved in the preceding 5 years. 2.3 Statistical Models We used Cox proportional hazards models with adjustment for potential individual and neighborhood confounders. All models were stratified at baseline for age, sex, and race-ethnicity. Age was included in 5-year intervals. We controlled for potential non-independence at the census-tract level with a sandwich estimator, which allowed for robust variance estimation. All analyses were run in the R package version (4.0.4 (2021-02-15). The Cox model estimates the instantaneous hazard of dying during the follow up as:hijt=h0stexp(βPij+δXij+ζZij+ϕWitj) where, hij(t): hazard function for the ith subject in jth census tract neighborhood; h0s(t) : the baseline hazard function for stratum s (i.e., age, race and sex); Pij : air pollution exposure metric of interest (e.g., PM2.5) standardized to the interquartile range for individual i in census tract j; Xij : individual risk factors (i.e., smoking, exercise, BMI, poverty) for individual i in census tract j; Zij : neighborhood risk factors (i.e., deprivation index, proportion taking public transit, crowding) for individual i in census tract j; and Witj : weather conditions (i.e., maximum temperature and humidity) for individual i at the tth month of admission in census tract j. Equation 1 above represents the general form of the model. Confounders, however, were selected for each pollutant with the following procedure: We ran unadjusted models stratified by age, race/ethnicity, and sex for each pollutant exposures. We tested every possible confounder (BMI, smoking, etc) one at a time with each pollution estimate. We included any confounder that changed the unadjusted pollution coefficient by at least 10%. We subsequently ran the adjusted models for all pollution exposures that included variables meeting the 10% criterion. Exposures were standardized for comparison across pollutants by dividing each by their respective interquartile ranges (IQR). For pollutants that had statistically significant effects at conventional levels after adjustment (i.e., p < 0.05), we then conducted stratified analyses on variables that could modify the association between COVID-19 death and air pollution, including race/ethnicity, sex, age, and number of chronic diseases categories. We also tested for interaction by running models with a multiplicative term with one pollutant and one meteorological variable. When significant interactions were present based on the p-value of the interaction term, we stratified the HR estimates for the pollutant by tertile of the meteorological variable. We examined two-pollutant models (i.e., O3 and NO2, NO2 and PM2.5 mass, and O3 and PM2.5 mass). We also explored the concentration-response (CR) functions for each pollutant that had a significant individual effect in a fully adjusted model. The CR functions were estimated via the pspline function in the gam library in R. We also investigated the potential contribution of different SARS-COV2 variants by performing sensitivity analyses that were restricted to periods when the Delta variant was dominant. The California Department of Public Health has done retrospective genomic analyses on specimens from all stages of the pandemic (https://covid19.ca.gov/variants/). In the early part of our study, five different variants were circulating. For much of the study period, the Delta variant was dominant. In the last month or so of our study, the Omicron variant became dominant. It is likely, however, that many of the hospitalizations that would have occurred in the last weeks to month of our study would have resulted from Delta due to latency in the infection time and the time required for a person to become ill enough to be hospitalized. Our sensitivity analyses focused on the period of 06/19/2020 to 01/03/2021. The first date corresponds to the initial timepoint at which the Delta variant accounted for more than 50% of the cases. We then identified the point at which Delta lost dominance (i.e., greater than 50%) as 12/19/2020. We added a two week buffer to this end date on the assumption that many of the hospitalizations and subsequent deaths from Delta would have taken at least two weeks to occur. This yields a conservative estimate for the end of the sensitivity analysis to be 01/03/2021. We reran our analyses for PM2.5 during this restricted time period so that results could be compared to the main analysis. 2.4 Role of the Funder The California Air Resources Board (CARB) funded most of this research and oversaw peer review of an unpublished final report documenting the methods and results, which was required by the terms of the funded contract. CARB staff also offered comments intended to improve the clarity of presentation in the final report. The Health Effects Institute also funded some of the study, but had no active role in the research. Neither funder had any role in the decision to publish this manuscript. 3 Results 3.1 Descriptive Statistics Table 1 displays the descriptive statistics for the cohort of the 21,415 KPSC patients who were hospitalized with COVID-19, of whom 4,815 died. Cohort characteristics of were as follows: median age 64 (IQR: 52, 75), 58% male, 56% Hispanic origin, 23% white, 11% Asian/Pacific Islanders, 8.6% Black, and 1.6% were of other or unknown race/ethnicity. Some 37% were ever smokers, and 13% had health insurance through MediCal, a government health program for low-income people.Table 1 Descriptives of hospitalized patients, by outcome (overall, survived or died) Died within 90 days Characteristic Overall, N = 21,4151 0, N = 16,6001 1, N = 4,8151 Age at diagnosis (years) 64 (52, 75) 61 (50, 72) 74 (64, 83) Race/ethnicity White 4,861 (23%) 3,550 (21%) 1,311 (27%) Asian-Pacific Islander 2,281 (11%) 1,801 (11%) 480 (10.0%) Black 1,851 (8.6%) 1,444 (8.7%) 407 (8.5%) Hispanic 12,077 (56%) 9,541 (57%) 2,536 (53%) Other/Multiple/Unknown 345 (1.6%) 264 (1.6%) 81 (1.7%) Gender F 9,067 (42%) 7,284 (44%) 1,783 (37%) M 12,348 (58%) 9,316 (56%) 3,032 (63%) Smoking status Never Smoker 13,392 (63%) 10,825 (66%) 2,567 (54%) Ever Smoker 7,738 (37%) 5,542 (34%) 2,196 (46%) Unknown 285 233 52 BMI 31 (27, 36) 31 (27, 36) 29 (25, 35) Unknown 608 502 106 Medicaid N 18,722 (87%) 14,596 (88%) 4,126 (86%) Y 2,693 (13%) 2,004 (12%) 689 (14%) Exercise Vital Sign (median) 0 (0, 90) 0 (0, 100) 0 (0, 65) Unknown 748 625 123 Percent housing units with >1 occupants per room 0.09 (0.03, 0.18) 0.09 (0.03, 0.18) 0.08 (0.03, 0.18) Unknown 598 466 132 NDI_ACS2013 0.42 (-0.28, 1.25) 0.43 (-0.27, 1.25) 0.40 (-0.30, 1.26) Unknown 6 5 1 Percent workers age 16+ commute by public transportation 0.02 (0.00, 0.05) 0.02 (0.00, 0.05) 0.02 (0.00, 0.05) Unknown 599 465 134 BMI category Normal weight 2,777 (13%) 1,876 (12%) 901 (19%) Overweight 5,933 (29%) 4,468 (28%) 1,465 (31%) ObeseClass1 5,669 (27%) 4,543 (28%) 1,126 (24%) ObeseClass2or3 6,193 (30%) 5,075 (32%) 1,118 (24%) Underweight 235 (1.1%) 136 (0.8%) 99 (2.1%) Unknown 608 502 106 Frailty (Lancet index) 5 (2, 12) 5 (2, 10) 9 (4, 18) Unknown 4,608 4,008 600 Elixhauser comorbidities 3.0 (1.0, 5.0) 2.0 (1.0, 5.0) 5.0 (2.0, 7.0) Cardiovascular Disease 8,637 (41%) 5,625 (35%) 3,012 (63%) Unknown 410 349 61 Hypertension 12,369 (59%) 8,738 (54%) 3,631 (76%) Unknown 410 349 61 COPD 4,519 (22%) 3,276 (20%) 1,243 (26%) Unknown 410 349 61 Diabetes 9,524 (45%) 6,887 (42%) 2,637 (55%) Unknown 410 349 61 Other Elixhauser dx 13,627 (65%) 9,878 (61%) 3,749 (79%) Unknown 410 349 61 Skilled Nursing Facility flag 293 (1.4%) 136 (0.8%) 157 (3.3%) County of residence Kern 543 (2.6%) 435 (2.7%) 108 (2.3%) Los Angeles 10,580 (51%) 8,226 (51%) 2,354 (50%) Orange 2,142 (10%) 1,744 (11%) 398 (8.5%) Riverside 2,372 (11%) 1,755 (11%) 617 (13%) San Bernardino 2,890 (14%) 2,131 (13%) 759 (16%) San Diego 1,874 (9.0%) 1,512 (9.4%) 362 (7.7%) Ventura 423 (2.0%) 338 (2.1%) 85 (1.8%) Unknown 591 459 132 Surge: First hospital admission after 2020-NOV-16 15,090 (70%) 11,378 (69%) 3,712 (77%) Days of follow-up 7 (4, 16) 6 (4, 11) 17 (10, 27) 1Median (IQR); n (%) 1. 0 = not dead after 90 days; 1 = dead within 90 days Figure 1 shows the hospitalizations over the entire time period, which included a major surge in admissions in November 2020 to the end of our study period.Figure 1 Hospital admission dates, among patients in study Most hospitalized patients were overweight or obese, with 29% meeting criteria for overweight, 27% for obesity class 1, and 30% for obesity class 2 or higher. Some 41% had a history of cardiovascular disease, 59% had hypertension, 22% had chronic obstructive pulmonary disease, 45% had diabetes, and 65% had another chronic condition. Patients who died within 90 days of their first hospitalization were older than those who did not (median age 74 vs 61 years), more likely to be male (63% vs 56%), and more likely to be ever smokers (46% vs 34%). Patients who died had more comorbidities (median Elixhauser index 5.0 vs 2.0) and greater prevalence of chronic diseases. Of those who died, 63% had prevalent cardiovascular disease (vs. 35% in survivors), 76% had hypertension (vs 54%), 55% had diabetes (vs 42%), and 26% had chronic obstructive pulmonary disease (COPD) (vs 20%). Descriptive statistics for the pollutants are shown in Table 2 .Table 2 Descriptive statistics for pollutants shown by overall sample and sample by event Died within 90 days Characteristic Overall, N = 21,4151 0, N = 16,600 1, N = 4,815 NO2 Median (IQR) 21 (13, 25) 21 (13, 25) 20 (14, 25) Mean (SD) 19 (7) 19 (7) 19 (7) Range 1, 39 1, 39 2, 36 O3 (maximum) Median (IQR) 66 (60, 72) 66 (60, 72) 67 (60, 73) Mean (SD) 66 (8) 66 (8) 66 (8) Range 40, 84 40, 84 43, 83 PM2.5 (mass) Median (IQR) 12.30 (10.50, 14.00) 12.30 (10.50, 14.00) 12.40 (10.60, 14.00) Mean (SD) 12.34 (2.40) 12.33 (2.39) 12.39 (2.44) Range 5.77, 27.70 5.77, 27.70 6.05, 23.80 PM2.5 (nitrates) Median (IQR) 3.81 (2.88, 4.54) 3.80 (2.86, 4.53) 3.84 (2.93, 4.56) Mean (SD) 3.64 (1.18) 3.63 (1.17) 3.67 (1.20) Range 0.19, 7.16 0.19, 7.16 0.26, 7.02 PM2.5 (organic carbon) Median (IQR) 2.07 (1.56, 2.60) 2.08 (1.56, 2.60) 2.05 (1.57, 2.56) Mean (SD) 2.08 (0.69) 2.08 (0.69) 2.07 (0.68) Range 0.31, 8.24 0.31, 8.24 0.32, 7.59 PM0.1 (mass) Median (IQR) 0.90 (0.72, 1.07) 0.90 (0.72, 1.07) 0.91 (0.74, 1.06) Mean (SD) 0.89 (0.29) 0.89 (0.29) 0.90 (0.29) Range 0.22, 6.63 0.26, 6.63 0.22, 4.20 PM2.5 (elemental carbon) Median (IQR) 0.47 (0.33, 0.59) 0.47 (0.33, 0.60) 0.46 (0.34, 0.58) Mean (SD) 0.47 (0.19) 0.47 (0.19) 0.47 (0.19) Range 0.05, 1.53 0.06, 1.52 0.05, 1.53 On-road diesel PM2.5 Median (IQR) 0.30 (0.19, 0.41) 0.30 (0.19, 0.41) 0.29 (0.20, 0.40) Mean (SD) 0.32 (0.18) 0.32 (0.18) 0.32 (0.18) Range 0.01, 1.78 0.01, 1.76 0.02, 1.78 On-road gasoline PM2.5 Median (IQR) 0.071 (0.052, 0.093) 0.072 (0.052, 0.094) 0.071 (0.052, 0.091) Mean (SD) 0.073 (0.029) 0.073 (0.030) 0.072 (0.029) Range 0.003, 0.213 0.003, 0.213 0.003, 0.194 Biomass combustion PM2.5 Median (IQR) 1.01 (0.73, 1.26) 1.01 (0.73, 1.27) 0.99 (0.72, 1.25) Mean (SD) 1.02 (0.46) 1.02 (0.45) 1.02 (0.49) Range 0.01, 9.93 0.01, 9.93 0.01, 9.03 Relative humidity (%) Median (IQR) 70 (58, 82) 71 (59, 82) 67 (57, 79) Mean (SD) 70 (14) 70 (14) 68 (14) Range 25, 99 25, 99 31, 98 Unknown 6 6 0 Temperature (C) Median (IQR) 21.1 (20.0, 25.0) 21.1 (20.0, 25.9) 20.8 (19.9, 22.5) Mean (SD) 22.9 (5.2) 23.1 (5.2) 22.3 (4.9) Range 5.9, 44.6 5.9, 44.5 7.2, 44.6 Unknown 6 6 0 1c(“Median (IQR)”, “Mean (SD)”, “Range”); all gaseous pollutants presented in ppb and all particle species in μ/m3 here and throughout the report Table 3 shows many of the pollutants had moderate to high correlations with one another (e.g., PM2.5 and PM2.5 nitrate r ∼ 0.9). Ozone was the least correlated with the other pollutants and, as expected, had negative associations with NO2 (r ∼ 0.25) and some of the particle species or source tracers.Table 3 Correlations among pollutants and meteorological variables NO2 O3 (maxi-mum) PM2.5 (mass) PM2.5 (nitrates) PM2.5 (organic carbon) PM0.1 (mass) PM2.5 (elemental carbon) On-road diesel PM2.5 On-road gasoline PM2.5 Biomass comb-ustion PM2.5 Relative humidity (%) Temp-erature (C) NO2 1.000 -0.255 0.715 0.615 0.843 0.691 0.849 0.731 0.842 0.522 0.232 0.077 O3 (maximum) -0.255 1.000 0.263 0.304 -0.286 0.090 -0.066 0.090 -0.175 -0.291 -0.584 0.093 PM2.5 (mass) 0.715 0.263 1.000 0.898 0.683 0.839 0.885 0.893 0.804 0.253 -0.021 0.114 PM2.5 (nitrates) 0.615 0.304 0.898 1.000 0.519 0.659 0.728 0.705 0.693 0.095 -0.002 0.125 PM2.5 (organic carbon) 0.843 -0.286 0.683 0.519 1.000 0.797 0.857 0.742 0.847 0.793 0.248 0.047 PM0.1 (mass) 0.691 0.090 0.839 0.659 0.797 1.000 0.817 0.751 0.716 0.464 -0.032 0.062 PM2.5 (elemental carbon) 0.849 -0.066 0.885 0.728 0.857 0.817 1.000 0.929 0.933 0.414 0.154 0.089 On-road diesel PM2.5 0.731 0.090 0.893 0.705 0.742 0.751 0.929 1.000 0.866 0.352 0.046 0.081 On-road gasoline PM2.5 0.842 -0.175 0.804 0.693 0.847 0.716 0.933 0.866 1.000 0.449 0.270 0.076 Biomass combustion PM2.5 0.522 -0.291 0.253 0.095 0.793 0.464 0.414 0.352 0.449 1.000 0.193 -0.006 Relative humidity (%) 0.232 -0.584 -0.021 -0.002 0.248 -0.032 0.154 0.046 0.270 0.193 1.000 0.237 Temperature (C) 0.077 0.093 0.114 0.125 0.047 0.062 0.089 0.081 0.076 -0.006 0.237 1.000 Figure 2 shows spatial distribution of several pollutants across Southern California, including: PM2.5 mass, PM2.5 nitrates, PM2.5 EC, PM0.1 as well as on-road gasoline and diesel tracers. Substantial differences exist in the spatial patterns among several pollutants. For example, on-road gasoline displayed variation consistent with highways that carry large volumes of traffic, while PM2.5 and PM2.5 nitrate had more smoothly-varying exposures across the region, likely due to a large contributions to the mass from secondary formation in the atmosphere. All pollutants had relatively higher concentrations in the inland areas of San Bernardino and Riverside.Figure 2 Predicted PM2.5 mass exposure fields during four seasons in the year 2016. All units µg m-3. 3.2 Results from Adjusted Models The confounders selected for each pollutant in our adjusted models are shown Table 3 of the ODS. Figure 3 below and ODS Table 5 show the main results on associations between air pollution and COVID-19 death. After confounding adjustment, we found several air pollutants were related to COVID-19 death among hospitalized patients including: PM2.5 mass (HR = 1.12, 95% CI 1.06, 1.17); PM2.5 nitrates (HR = 1.12, 95% CI 1.07, 1.17); PM2.5 EC (HR = 1.07, 95% CI 1.03, 1.12); PM0.1 mass (HR = 1.06, 95% CI 1.02, 1.10); PM2.5 on-road diesel (HR = 1.06, 95% CI 1.03, 1.10); and PM2.5 on-road gasoline (HR = 1.07, 95% CI 1.02, 1.13). Effects of PM2.5 mass were partly confounded by NO2 in the two pollutant models, but remained significantly elevated (Figure 3). For the Delta variant-dominant period, results were similar to those from the main model for PM2.5 mass (HR = 1.13, 95% CI 1.07, 1.20).Figure 3 Risk plots showing hazard ratios of all pollutants based on the interquartile range increment Effects for gaseous species were sensitive to co-pollutant adjustment. NO2 had a significant association with the risk of death (HR = 1.10, 95% CI 1.04, 1.16), while ozone had positive but insignificant effects (HR = 1.02, 95% CI 0.96, 1.08). Because the inverse spatial pattern that can lead to positive confounding, (Quiros et al., 2013) we also ran co-pollutant models with ozone and NO2 included. In these models, NO2 remained significantly elevated, but ozone remained null. PM2.5 confounded the NO2 effect to null when both were included in the same model (Figure 3). Higher temperatures (HR = 0.92, 95% CI 0.89, 0.95) and higher humidity (HR = 0.82, 95% CI 0.78, 0.86) in the month of diagnosis were significantly associated with lower risks for COVID-19 death. 3.3 Stratification Analyses All of the subgroup analyses were insignificant based on the Q statistic shown at the bottom of each table, meaning these variables had no significant impact on the air pollution concentration-response association with COVID-19 death (see Tables 6-9 for stratification analyses in the ODS). 3.4 Interaction Models with Meteorological Variables After determining that temperature and humidity significantly modified the effects of air pollution on COVID-19 death, we stratified by tertile for these variables to visualize the effect modification for PM2.5 (see Figure 4 ). See ODS Figure 13 for other pollutants. For most of the pollutants, elevated risks appear only in the lower two tertiles of temperature. Effect modification was particularly pronounced for humidity, with most pollutants showing a graded decline in effects as humidity went up. PM0.1 mass and PM2.5 nitrates followed a slightly different trend with the largest effect in the middle tertile. Overall, most effects were present only in the lowest two tertiles of humidity.Figure 4 Risk plots showing PM2.5 stratified by tertile maximum temperature and relative humidity during the month of diagnosis. See ODS for stratified risk plots for other pollutants 3.5 Concentration-Response Analysis Concentration-response curves are shown in the Figure 5 . For most of the pollutants, we observed fairly linear curves when sufficient data was available to support the spline derivation. Some pollutants such as EC and on-road diesel displayed a supra-linear response with a steeper response curve at the low exposure levels. This supra-linear function has been observed in many air pollution-mortality studies. (Burnett et al., 2018) Humidity displayed a clear linear negative association with risk of COVID-19 death. Temperature had a U-shape risks appear to be higher at lower temperatures, although where there was sufficient data to support the spline derivation, the inverse curve appeared linear.Figure 5 Dose-response Functions for Pollutants and Metrological Variables. All pollutant concentrations expressed in μ/m3, temperature in degrees C, and relative humidity in percent. 4 Discussion And Conclusion Here we evaluated whether chronic exposure to air pollution and meteorology at the time of diagnosis affected the risk of death in patients with a COVID-19-related hospitalization. We found significant associations between the risk of COVID-19 death following hospitalization and PM2.5 mass, PM0.1 mass, and several of the particle species or source tracers. Effects for PM2.5 mass were reduced when NO2 was included in the model, but remained significantly elevated, while NO2 was confounded to null in the two-pollutant model. Meteorology has been associated with COVID-19 transmission in some studies, (Zoran et al., 2022) and recent studies show that meteorology likely affected COVID-19 death rates in Europe. (Kifer et al., 2021) These researchers proposed that humidity can interfere with viral defenses of nasal mucosa tissues and with the sputum deeper in the airway, which can lead to more severe infection and subsequently contribute to a poor prognosis after the virus establishes itself in the respiratory tract, particularly in the nose. (Kifer et al., 2021, Weaver et al., 2022) Temperature and humidity can also affect size of the viral droplets and its persistence in ambient air, but the extent to which this would affect severity is unknown. (Bourdrel et al., 2021) Significant negative effects were present for both temperature and humidity in our study. We also found significant effect modification of the air pollution associations with lower temperature and humidity being associated generally with larger air pollution effects. If the viral defenses are influenced by meteorology, both direct effects of humidity and temperature and the effect modification of the pollution effect have biological plausibility. In comparing our results to other mortality studies, Chen et al. (Chen et al., 2021) investigated the association of air pollution on COVID-19 severity and mortality using data from KPSC members with a CALINE dispersion model, which estimated traffic exposures as Nox (non-freeway and freeway). The odds of intensive care admission were 1.11 (95% CI: 1.04, 1.19) and death were 1.10 (95% CI: 1.03, 1.18) for each SD increase in non-freeway Nox. Several other freeway exposures, however, had protective effects. (Chen et al., 2021) Including regional PM2.5 and NO2 as confounders attenuated the effects by 19-26%, and this adjustment caused the freeway Nox to become significantly protective for mortality (HR = 0.93, 95% CI: 0.87-1.0). Possibly, exposure measurement error may have been present due to the inability of the CALINE dispersion model to deal with complex traffic, terrain, and meteorological conditions, all which exist in Southern California. (Jerrett et al., 2005, Dhyani et al., 2013) Our findings may have also differed due to the longer follow up in the present study (with about 4.5 times as many deaths). A follow up to this study using the same health data, but relying on inverse-distance averaging to interpolate from government monitors, found significant chronic effects associated with PM2.5 exposure and sub-chronic effects from NO2. (Chen et al., 2022) This study, however, also had a high probability of exposure measurement error given the likely level of spatial variation in these pollutants and the sparse data support available from the government monitors of which there are relatively few covering thousands of square kilometers. (Chen et al., 2022) Another study from the UK relied on the Biobank data and used an agnostic exposomic statistical approach to evaluate many factors for risk of COVID-19 incidence and death. (Elliott et al., 2021) Although mild associations were present in univariate models with PM2.5, these were eliminated in multivariate models, leading the authors to conclude there was little evidence of an independent association between air pollution and COVID-19 death. This study, however, had relatively few deaths and may have lacked power to detect subtle effects from air pollution. A study using data from hospitalized patients in New York City reported an association between PM2.5 and risk of mortality (risk ratio, 1.11, 95% CI: 1.02–1.21) per 1 μg/m3 increase). (Bozack et al., 2021) Evaluated across the reported IQR of 0.7 μg/m3 the rate ratio would be ∼ 1.08. Neither black carbon nor NO2 had a significant association with COVID-19 death. This study also found Hispanic ethnicity significantly modified the air pollution risk for COVID-19 death, which differs from our finding of no significant subgroup interaction. This study lacked individual information on some potential risk factors for COVID-19 death, including obesity and smoking. Consequently, residual confounding cannot be ruled out. In a large administrative cohort from Rome Italy, significant associations with COVID-19 mortality were found with both NO2 and PM2.5. (Nobile et al., 2022) Associations found in the Rome study were somewhat smaller than what we have found here, but the confidence intervals for the two studies overlap. (Nobile et al., 2022) The range of exposure for PM2.5 in Rome was much smaller than what we observed in Southern California, which may in part explain the smaller effects observed in Rome. On limitations, while we did control for several individual confounders such as smoking and obesity, the KPSC I data did not include potentially important confounding variables such as occupational status. Nascent research suggests increased risk of mortality in some occupational groups in California, particularly in the farming, material moving, transportation, and construction sectors, all of which could have higher air pollution due to occupational exposures. (Cummings et al., 2021) While numerous complexities exist in analyzing and interpreting COVID-19 mortality risk in different occupations, (Cummings et al., 2021, Pearce et al., 2021) it is plausible that lack of occupational status could have biased our results. In addition, a temporal mismatch existed between the exposure fields from 2016, which predated the study by some three years; however, overall spatial patterns of exposure are unlikely to change during this period. Some portions of our study, however, overlapped with the “lock down” period when traffic emissions were lower. (Liu et al., 2021) Cohort members may therefore have experienced lower exposures than they would have if normal conditions had prevailed, which would not have been accounted for in our exposure or statistical models. The impact would have been to overestimate their exposures near-source traffic exposures, which may have biased some results toward null. The near-road pollutants such as PM2.5 EC, PM0.1 on-road diesel and on-road gasoline had risks that were smaller than PM2.5 and PM2.5 nitrate, which might have been due to the lock down effect not captured in our exposure model. Nevertheless, despite this limitation, several near-source pollutants still displayed significant associations with COVID-19 death. We were also unable to account for acute effects, which may have contributed to risks of COVID-19 death. Currently, we are extending the CTM exposure modeling to derive contemporaneous estimates of acute and chronic exposure. Another concern with observational studies of COVID and mortality rests in the different variants that emerged and gained dominance through the pandemic. If certain variants were more virulent than others as appears likely (Adjei, 2022) and these emerge coincidentally at times when air pollution is high, then associations between air pollution and mortality could be confounded by the virulence of the variant. In this study, the Delta variant was dominant for the majority of the study period. We performed sensitivity tests on the PM2.5 model by restricting the analysis period to times when Delta was dominant. The results from the restricted analysis were virtually the same as the results from the full analysis. Based on this similarity, we conclude that it was unlikely that our results are confounded by variants with different virulence. We used time-to-event data after hospitalization to avoid having bias in our follow up times, which could have varied considerably if we began the study at the point of COVID-19 diagnosis. Although necessary for unbiased statistical inference, this restriction reduces the generalizability of the results to hospitalized individuals rather than the general population. Other environmental variables have also been implicated in the spread and severity of COVID-19, including wind speed and ultraviolet radiation. Both variables were explicitly included in our CTM exposure model. Wind speed in particular has a major impact on ambient concentrations of several pollutants, and we were concerned that including wind speed as its own variable would induce collinearity into the model. In reviewing the literature on wind speed we also found that most of the influence of this meteorological parameter affected the spread of COVID-19, not the severity of symptoms or risk of death. (Weaver et al., 2022, Bourdrel et al., 2021) UVB potentially operates through vitamin D deficiency, which has been identified as a risk factor for more extreme COVID-19 outcomes. (Xu et al., 2020) We visually explored the 1 km UVB fields used as inputs in the CTM modeling. UVB levels were higher inland and lower near the Pacific Coastline likely due to fog and cloud cover. Recent UVB exposure modeling, however, estimates that personal behavior and occupation are much more important predictors of UVB exposure than ambient levels alone, which often account for little of the explained variation in objectively measured UVB. (Dadvand et al., 2011, Soueid et al., 2022) Thus, the ambient levels are unlikely to be reasonable proxies for exposure and subsequent deficiency. We also queried our data base to identify patients who were vitamin D deficient and run stratification analyses to assess whether air pollution contributed to worse outcomes in these patients. Some 4,142 (19.64%) of the hospitalized cohort had a vitamin D lab test within 1 year prior to COVID test date and out of that group, 1,524 (7.23% of total cohort) were vitamin D deficient (25-HYDROXYVITAMIN D lab result <30 ng/mL) based on their most recent vitamin D lab prior to hospitalization. Because this is relatively small proportion of the cohort and likely represents an underestimate that may bias results, we were unable to stratify the analysis. Virucidal activity also decreases in the presence of higher UVB radiation, (Weaver et al., 2022, Bourdrel et al., 2021) but this would be more likely to affect the spread of the virus and not the severity. In addition, it is likely that most of the infections occurred from contact in the indoor environment where ambient UVB would likely have a minimal impact on the virucidal activity. (Weaver et al., 2022) Future research is nevertheless needed to assess whether vitamin D deficiency modifies air pollution effects on COVID-19 severity. Saturation of capacity of the ICU care is another possible factor affecting survival that may have acted as a confounder. Internal data and consultation with attending physicians indicated that despite this surge (see Figure 1), at no time were the ICU units saturated beyond capacity. KPSC did not run out of ventilators or physical space for admitting seriously ill COVID-19 patients. An over-flow facility that could have accepted KPSC patients was never used. Consequently, saturation of the ICU is unlikely to confound our results. The observational nature of this study precludes causal interpretation. Based on our results, however, we can conclude that chronic exposure to air pollution in Southern California is associated with increases in the risk of death from COVID-19. Better knowledge about environmental variables such as air pollution and meteorology could be used by communities and local governments to target neighborhoods with higher risks for COVID-19 death. Such information could also be brought into healthcare systems to assist clinicians with better estimating the likely severity of disease in patients from high air pollution areas. Minimizing the spread and reducing the severity of COVID-19 through non-pharmaceutical interventions (NPI) such as masking and economic shutdowns remains problematic over the longer term (Imai et al., 2020) given the social and environmental costs involved. In addition, modeling suggests that NPI measures have the potential to increase the severity of other respiratory viral outbreaks in the future. (Baker et al., 2020) Pharmaceutical measures such as vaccines continue to have mixed results in part due to vaccine hesitancy in some high prevalence locations and population groups. (Sallam, 2021) In contrast, air pollution is a modifiable environmental risk factor that could affect disease severity across the entire population. Reducing air pollution may thus provide a more sustainable means of reducing COVID-19 severity that would have substantial population benefits. It may also reduce the risks for catastrophic outcomes from future pandemics fueled by novel viruses, while also having beneficial effects on a wide array of other health endpoints. Funding California Air Resources Board and the California Environmental Protection Agency, agreement no. 19RD030 and the Health Effects Institute award no. #4979-RFA20-1B/21-2. 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. Data availability The data that has been used is confidential. ==== Refs References World Health Organization. WHO Coronavirus Disease (COVID-19) Dashboard \| WHO Coronavirus Disease (COVID-19) Dashboard. 2022. https://covid19.who.int/ (accessed October 25, 2022). Brandt E.B. Beck A.F. Mersha T.B. Air pollution, racial disparities, and COVID-19 mortality Journal of Allergy and Clinical Immunology. 2020 146 10.1016/j.jaci.2020.04.035 Li H. Xu X.L. Dai D.W. Huang Z.Y. Ma Z. Guan Y.J. Air pollution and temperature are associated with increased COVID-19 incidence: A time series study International Journal of Infectious Diseases 2020 97 10.1016/j.ijid.2020.05.076 Wu X. Nethery R.C. Sabath M.B. Braun D. Dominici F. Air pollution and COVID-19 mortality in the United States: Strengths and limitations of an ecological regression analysis. Science Advances 6 2020 eabd4049 Zhang Z. Xue T. Jin X. Effects of meteorological conditions and air pollution on COVID-19 transmission: Evidence from 219 Chinese cities Science of the Total Environment 2020 741 10.1016/j.scitotenv.2020.140244 Zhu Y. Xie J. Huang F. Cao L. Association between short-term exposure to air pollution and COVID-19 infection: Evidence from China Science of the Total Environment 2020 727 10.1016/j.scitotenv.2020.138704 Lippi G, Sanchis-Gomar F, Henry BM. Association between environmental pollution and prevalence of coronavirus disease 2019 (COVID-19) in Italy. 2020. DOI:10.1101/2020.04.22.20075986. Coker E.S. Cavalli L. Fabrizi E. The Effects of Air Pollution on COVID-19 Related Mortality in Northern Italy Environmental and Resource Economics 2020 76 10.1007/s10640-020-00486-1 Wang B. Chen H. Chan Y.L. Oliver B.G. Is there an association between the level of ambient air pollution and COVID-19? American journal of physiology Lung cellular and molecular physiology. 2020 319 10.1152/ajplung.00244.2020 Travaglio M. Yu Y. Popovic R. Selley L. Leal N.S. Martins L.M. Links between air pollution and COVID-19 in England Environmental Pollution 268 2021 115859 Yao Y. Pan J. Liu Z. Ambient nitrogen dioxide pollution and spreadability of COVID-19 in Chinese cities Ecotoxicology and Environmental Safety 208 2021 111421 Huang G. Blangiardo M. Brown P.E. Pirani M. Long-term exposure to air pollution and COVID-19 incidence: A multi-country study Spatial and Spatio-Temporal Epidemiology 39 2021 100443 Chen Z. Huang B.Z. Sidell M.A. Near-roadway air pollution associated with COVID-19 severity and mortality – Multiethnic cohort study in Southern California Environment International 157 2021 106862 Berg K. Romer Present P. Richardson K. Long-term air pollution and other risk factors associated with COVID-19 at the census tract level in Colorado Environmental Pollution 2021 287 10.1016/j.envpol.2021.117584 Zhou X. Josey K. Kamareddine L. Excess of COVID-19 cases and deaths due to fine particulate matter exposure during the 2020 wildfires in the United States Science Advances 7 2021 8789 8802 Rebuli M.E. Brocke S.A. Jaspers I. Impact of inhaled pollutants on response to viral infection in controlled exposures J Allergy Clin Immunol 148 2021 1420 1429 34252446 Wang B. Chen H. Chan Y.L. Oliver B.G. Is there an association between the level of ambient air pollution and COVID-19? American Journal of Physiology-Lung Cellular and Molecular Physiology 319 2020 L416 L421 32697597 De Angelis E. Renzetti S. Volta M. COVID-19 incidence and mortality in Lombardy, Italy: An ecological study on the role of air pollution, meteorological factors, demographic and socioeconomic variables Environ Res 195 2021 110777 Kifer D. Bugada D. Villar-Garcia J. Effects of Environmental Factors on Severity and Mortality of COVID-19 Frontiers in Medicine 7 2021 1088 SanJuan-Reyes S. Gómez-Oliván L.M. Islas-Flores H. COVID-19 in the environment Chemosphere 263 2021 127973 Zeldovich YB. 26. Oxidation of Nitrogen in Combustion and Explosions. In: Selected Works of Yakov Borisovich Zeldovich, Volume I. 2015. DOI:10.1515/9781400862979.404. Quiros D.C. Zhang Q. Choi W. Air quality impacts of a scheduled 36-h closure of a major highway Atmospheric Environment 67 2013 404 414 Neupane B. Jerrett M. Burnett R.T. Marrie T. Arain A. Loeb M. Long-term exposure to ambient air pollution and risk of hospitalization with community-acquired pneumonia in older adults American Journal of Respiratory and Critical Care Medicine 2010 181 10.1164/rccm.200901-0160OC 19833826 Ciencewicki J. Jaspers I. Air Pollution and Respiratory Viral Infection Inhalation Toxicology 19 2007 1135 1146 17987465 Jerrett M. Shankardass K. Berhane K. Traffic-related air pollution and asthma onset in children: A prospective cohort study with individual exposure measurement Environmental Health Perspectives 116 2008 1433 1438 18941591 Bai L. Chen H. Hatzopoulou M. Exposure to ambient ultrafine particles and nitrogen dioxide and incident hypertension and diabetes Epidemiology 2018 29 10.1097/EDE.0000000000000798 Sydbom A, Blomberg A, Parnia S, Stenfors N, Sandström T, Dahlén S-E. Health effects of diesel exhaust emissions. European Respiratory Journal 2001; 17: 733 LP – 746 Jerrett M. Burnett R.T. Ma R. Spatial Analysis of Air Pollution and Mortality in Los Angeles Epidemiology 16 2005 727 736 16222161 Beelen R. Hoek G. van den Brandt P.A. Long-term effects of traffic-related air pollution on mortality in a Dutch cohort (NLCS-AIR study) Environmental Health Perspectives 116 2008 196 202 18288318 Bozack A, Pierre S, DeFelice N, et al. Long-Term Air Pollution Exposure and COVID-19 Mortality: A Patient-Level Analysis from New York City. Am J Respir Crit Care Med 2021; published online Dec 9. DOI:10.1164/rccm.202104-0845OC. Elliott J. Bodinier B. Whitaker M. COVID-19 mortality in the UK Biobank cohort: revisiting and evaluating risk factors Eur J Epidemiol 36 2021 299 309 33587202 Nobile F. Michelozzi P. Ancona C. Air pollution, SARS-CoV-2 incidence and COVID-19 mortality in Rome – a longitudinal study European Respiratory Journal 2022 10.1183/13993003.00589-2022 published online Jan 1. Chen Z. Sidell M.A. Huang B.Z. Ambient Air Pollutant Exposures and COVID-19 Severity and Mortality in a Cohort of Patients with COVID-19 in Southern California Am J Respir Crit Care Med 206 2022 440 448 35537137 Koebnick C. Langer-Gould A.M. Gould M.K. Sociodemographic characteristics of members of a large, integrated health care system: comparison with US Census Bureau data The Permanente Journal 16 2012 37 Nau C, Bruxvoort K, Navarro RA, et al. COVID-19 Inequities Across Multiple Racial and Ethnic Groups: Results From an Integrated Health Care Organization. Annals of internal medicine 2021; published online April 20. DOI:10.7326/m20-8283. Tartof S.Y. Qian L. Hong V. Obesity and Mortality Among Patients Diagnosed With COVID-19: Results From an Integrated Health Care Organization Ann Intern Med 173 2020 773 781 32783686 Quan H. Sundararajan V. Halfon P. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data Medical care 43 2005 1130 1139 16224307 Young D.R. Fischer H. Arterburn D. Associations of overweight/obesity and Socioeconomic Status With Hypertension Prevalence Across Racial and Ethnic Groups. 20 2018 532 540 Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology American journal of public health 82 1992 703 710 1566949 Diez-Roux A.V. Kiefe C.I. Jacobs D.R. Jr Area characteristics and individual-level socioeconomic position indicators in three population-based epidemiologic studies Annals of epidemiology 11 2001 395 405 11454499 Geronimus A.T. Bound J. Use of census-based aggregate variables to proxy for socioeconomic group: evidence from national samples American journal of epidemiology 148 1998 475 486 9737560 Messer L.C. Laraia B.A. Kaufman J.S. The development of a standardized neighborhood deprivation index Journal of urban health : bulletin of the New York Academy of Medicine 83 2006 1041 1062 17031568 Venecek M.A. Yu X. Kleeman M.J. Predicted ultrafine particulate matter source contribution across the continental United States during summertime air pollution events Atmospheric Chemistry and Physics 19 2019 9399 9412 Yu X. Venecek M. Kumar A. Regional sources of airborne ultrafine particle number and mass concentrations in California Atmospheric Chemistry and Physics 19 2019 14677 14702 Burnett R. Chen H. Szyszkowicz M. Global estimates of mortality associated with long-term exposure to outdoor fine particulate matter PNAS 115 2018 9592 9597 30181279 Zoran M.A. Savastru R.S. Savastru D.M. Tautan M.N. Baschir L.A. Tenciu D.V. Assessing the impact of air pollution and climate seasonality on COVID-19 multiwaves in Madrid Spain. Environ Res 203 2022 111849 Weaver A.K. Head J.R. Gould C.F. Carlton E.J. Remais J.V. Environmental Factors Influencing COVID-19 Incidence and Severity Annual Review of Public Health 43 2022 271 291 Bourdrel T. Annesi-Maesano I. Alahmad B. Maesano C.N. Bind M.-A. The impact of outdoor air pollution on COVID-19: a review of evidence from in vitro, animal, and human studies European Respiratory Review 2021 30 10.1183/16000617.0242-2020 Jerrett M. Arain A. Kanaroglou P. A review and evaluation of intraurban air pollution exposure models J Expo Sci Environ Epidemiol 15 2005 185 204 Dhyani R. Singh A. Sharma N. Gulia S. Performance evaluation of CALINE 4 model in a hilly terrain – a case study of highway corridors in Himachal Pradesh (India) International Journal of Environment and Pollution 52 2013 244 262 Cummings KJ, Beckman J, Frederick M, et al. Disparities in COVID-19 Fatalities among Working Californians. 2021; : 2021.11.10.21266195. Pearce N. Rhodes S. Stocking K. Occupational differences in COVID-19 incidence, severity, and mortality in the United Kingdom: Available data and framework for analyses Wellcome Open Res 6 2021 102 34141900 Liu J. Lipsitt J. Jerrett M. Zhu Y. Decreases in Near-Road NO and NO2 Concentrations during the COVID-19 Pandemic in California Environ Sci Technol Lett 8 2021 161 167 Adjei S. Mortality Risk Among Patients Hospitalized Primarily for COVID-19 During the Omicron and Delta Variant Pandemic Periods — United States, April 2020–June 2022 MMWR Morb Mortal Wkly Rep 2022 71 10.15585/mmwr.mm7137a4 Xu Y. Baylink D.J. Chen C.-S. The importance of vitamin d metabolism as a potential prophylactic, immunoregulatory and neuroprotective treatment for COVID-19 Journal of Translational Medicine 18 2020 322 32847594 Dadvand P. Basagaña X. Barrera-Goómez J. Diffey B. Nieuwenhuijsen M. Measurement errors in the assessment of exposure to solar ultraviolet radiation and its impact on risk estimates in epidemiological studies Photochem Photobiol Sci 10 2011 1161 1168 21465050 Soueid L. Triguero-Mas M. Dalmau A. Estimating personal solar ultraviolet radiation exposure through time spent outdoors, ambient levels and modelling approaches* British Journal of Dermatology 186 2022 266 273 34403140 Imai N. Gaythorpe K.A.M. Abbott S. Adoption and impact of non-pharmaceutical interventions for COVID-19 Wellcome Open Res 5 2020 59 32529040 Baker RE, Park SW, Yang W, Vecchi GA, Metcalf CJE, Grenfell BT. The impact of COVID-19 nonpharmaceutical interventions on the future dynamics of endemic infections. Proceedings of the National Academy of Sciences 2020; 117: 30547–53. Sallam M. COVID-19 Vaccine Hesitancy Worldwide: A Concise Systematic Review of Vaccine Acceptance Rates Vaccines 9 2021 160 33669441	0	PMC9715495	NO-CC CODE	2022-12-08 23:16:14	no	Environ Int. 2022 Dec 2;:107675	utf-8	Environ Int	2,022	10.1016/j.envint.2022.107675	oa_other
==== Front Chaos Solitons Fractals Chaos Solitons Fractals Chaos, Solitons, and Fractals 0960-0779 0960-0779 Elsevier Ltd. S0960-0779(22)01143-2 10.1016/j.chaos.2022.112964 112964 Article Mutation and SARS-CoV-2 strain competition under vaccination in a modified SIR model Ahumada M. a Ledesma-Araujo A. b Gordillo Leonardo b Marín Juan F. b⁎ a Departamento de Física, Universidad Técnica Federico Santa María, Casilla 110 V, Valparaíso, Chile b Departamento de Física, Facultad de Ciencia, Universidad de Santiago de Chile, Usach, Av. Víctor Jara 3493, Estación Central, Santiago, Chile ⁎ Corresponding author. 2 12 2022 2 12 2022 11296412 7 2022 27 10 2022 27 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 crisis caused by the COVID-19 outbreak around the globe raised an increasing concern about the ongoing emergence of variants of the virus that may evade the immune response provided by vaccines. New variants appear due to mutation, and as the cases accumulate, the probability of the emergence of a variant of concern increases. In this article, we propose a modified susceptible, infected, and recovered (SIR) model with waning immunity that captures the competition of two strain classes of an infectious disease under the effect of vaccination with a highly contagious and deadly strain class emerging from a prior strain due to mutation. When these strains compete for a limited supply of susceptible individuals, changes in the efficiency of vaccines may affect the behaviour of the disease in a non-trivial way, resulting in complex outcomes. We characterise the parameter space including intrinsic parameters of the disease, and using the vaccine efficiencies as control variables. We find different types of transcritical bifurcations between endemic fixed points and a disease-free equilibrium and identify a region of strain competition where the two strain classes coexist during a transient period. We show that a strain can be extinguished either due to strain competition or vaccination, and we obtain the critical values of the efficiency of vaccines to eradicate the disease. Numerical studies using parameters estimated from publicly reported data agree with our theoretical results. Our mathematical model could be a tool to assess quantitatively the vaccination policies of competing and emerging strains using the dynamics in epidemics of infectious diseases. Graphical abstract Keywords COVID-19 Multi-strain SIR models Mutation Vaccination strategies ==== Body pmc1 Introduction The global outbreak of the COVID-19 disease has become a major challenge for multidisciplinary scientific research, even though it is not as lethal as other diseases. While most symptoms among the population are mild, the new severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can cause life-threatening pneumonia among some patients. Furthermore, variations in the viral strain contribute to disease severity and spreading efficiency. As a response, harsh periodical and social isolation was implemented in many countries with severe economic disruptions as a consequence [1], [2]. Under this scenario, vaccination emerge as the ultimate solution to release the population. Only a few months after the first reported case of COVID-19 in Wuhan, China (December 2019) [3], independent variants of SARS-CoV-2 were reported, such as Alpha (B.1.1.7) [4], Beta (B.1.351) [5], Gamma (P.1) [6], Delta (B.1.617), and Omicron (B.1.1.529) [7]. There is thus a rising concern about the effectiveness of the currently developed vaccines in pandemic control [8]. Since SARS-CoV-2 virus attaches to human ACE2 cell surface via its surface S protein, most vaccines were developed to stimulate responses that target only the S protein of the virus [9]. However, mutations are causing amino acid alterations in the S protein that significantly increase the virus effectiveness to bind to the human receptors [10]. According to preliminary studies, this turns into new strains that may spread more rapidly and efficiently, with consequent human losses. This has led to concerns that new strains may escape the immune response generated after vaccination [11]. Kermack and McKendrick proposed in their pioneering work a simplified mathematical model for the evolution of an epidemic [12], widely known as the SIR model. The model divides the population into three classes: Susceptibles, Infected and Recovered (SIR). A set of nonlinear differential equations govern the time evolution of the number of individuals within each class. Since the SIR model gives a good description of the Bombay plague of 1906 [12], many variations have been proposed to adapt the model to other infectious diseases, such as influenza, HIV, measles, malaria, SARS, foot-and-mouth disease, and whooping cough [13]. In the context of the ongoing COVID-19 pandemics, several adaptations of the SIR model have been proposed in the last two years to understand and predict the behaviour of this new coronavirus [14], [15], [16], [17]. Mathematical modelling of multi-strain dynamics [13] and different approaches aiming to guide public health policies and strategies on COVID-19 vaccination [18] have lately achieved significant progress in many aspects. However, any model or approach has filled the gap of singly combining the effects of strain competition and vaccination with ongoing mutations in a coherent mathematical model, although evidence of their interplay is compelling. In this article, we study the effect of virus mutation and vaccination in populations with two competing classes of SARS-CoV-2 strains. We use a modified SIR model that assumes the existence of two classes of strains, one more contagious and slightly deadly than the other one. We consider the effect of vaccination in the model, assuming that vaccines have different efficiencies against each virus species. We contrasted our theoretical results with numerical simulations, with parameter values estimated from public datasets on the COVID-19 pandemics in Chile, one of the pioneering countries in massive vaccination campaigns against SARS-CoV-2 [19], [20], [21]. The article is organised as follows: In Section 2, we introduce our modified SIR model and discuss the epidemiological assumptions made for this study. In Section 3, we study the bifurcations of the system, as well as the equilibria and their stability. We demonstrate that the new terms in our SIR model accounting for strain competition and mutation captures a transient regime during which the stronger strains extinguish the weaker ones. In Section 4, we show and discuss the results from numerical simulations of the time evolution of the pandemic. Our final remarks and conclusions are presented in Section 5. 2 The modified SIR model for strain competition The modified epidemiological SIR model schematised in Fig. 1 accounts for infections for which there is no permanent immunity [13], as it is the case of COVID-19 [22], [23]. This model implicitly assumes an acute infection, i.e. a “fast” infection where a relatively rapid immune response removes pathogens after days or weeks. Our model assumes an infectious disease spreading through one host species in a densely distributed population. Competition between strains is fundamental to disease evolution, and the scientific community has recently begun to understand the range of complex outcomes when multiple strains compete for a limited supply of susceptible individuals. Ir our model, we classify all possible strains of SARS-CoV-2 into two classes:Fig. 1 Flow diagram of the modified SIR model summarising the main epidemiological assumptions of this study. The circulation of two strain classes of the virus is tracked among vaccinated and non-vaccinated individuals. • Strain class 1 (SC1): composed of those strains with moderate contagiousness and relatively low lethality. These strains can be usually controlled by vaccination with relatively high efficiency. • Strain class 2 (SC2): composed of those strains that evade the immune response provided by vaccines. These strains are more contagious and deadly than strains in class 1. Among all possible mutations, we will focus on those of strain class 1 that turns them into a strain of class 2. That is, we are only interested in mutations that increase the contagiousness and the risks of dying from the disease. As shown in Fig. 1, the S-class, composed of susceptible individuals, is subdivided into two classes: SG and V. Unvaccinated individuals are in the SG-pool. Susceptible individuals who receive vaccination are removed from the SG-pool and added to the vaccinated pool, the V-class. Individuals from the V-pool are not entirely immune to strain class 1 and can still be infected by strain class 2. The transition SG→IiS(i=1,2) denotes susceptible individuals infected by the ith strain class. The transition I1S→I2S can occur due to mutations within I1S-individuals. Following the known behaviour of individuals infected with SARS-CoV-2, we neglect the probability of an individual being infected simultaneously by the two strain classes: COVID-19 patients usually isolate after becoming ill, and it is unlikely that they can be infected again during the isolation period by another individual infected with another strain class. Thus, we assume that mutation turns I1S-individuals into I2S-individuals. The IiV-class (i=1,2) corresponds to vaccinated individuals infected by the ith strain class. The complete class of infected individuals is I=I1S∪I2S∪I1V∪I2V. Class RS (RV) comprises unvaccinated (vaccinated) individuals recovered from the disease caused by any of the two strain classes. Infected individuals can recover after an infectious period, which is given by the transitions IiV→RV and IiS→RS (i=1,2). The complete class of recovered individuals is R=RS∪RV. Eventually, R-individuals return to the S-pool due to non-permanent immunity. The level of the infectious disease influences the rate at which SG (V)-individuals move into any of the IiS (IiV)-classes (i=1,2), as indicated by the dotted arrows in Fig. 1. For the transition to any of the recovered classes, we consider a recovery rate γi(i=1,2) for each strain class. Infections occurring in a small time interval dt, under the hypothesis that underlying epidemiological probabilities are constant are thus modelled by the following nonlinear system of ODEs, (1a) S˙G=ν+wRS−dSG−υΘ(SG)−1Nβ1I1+β2I2SG, (1b) V˙=υΘ(SG)+wRV−dV−1Nβ¯1I1+β¯2I2V, (1c) I˙1S=−δ1I1S−μI1S+1Nβ1I1SG, (1d) I˙2S=−δ2I2S+μI1S+1Nβ2SGI2, (1e) I˙1V=−δ1I1V+1Nβ¯1I1V, (1f) I˙2V=−δ2I2V+1Nβ¯2I2V, (1g) R˙S=γ1I1S+γ2I2S−dRS−wRS, (1h) R˙V=γ1I1V+γ2I2V−dRV−wRV, where SG, V, IiS,V(i=1,2), and RS,V are the number of individuals within their respective classes, and dots denote derivatives with respect to time. The total number of infected from the ith strain class is Ii=IiS+IiV, with i=1,2. The total (time-dependent) population size is N=S+I+R, with S=SG+V, I=I1+I2 and R=RS+RV. The effects of demographic processes in the populations are considered with parameters ν and d, which are the rate at which individuals in any epidemiological class are incorporated (e.g. by births or immigration) or removed (e.g. by emigration or death by causes independent of the disease). The term υΘ(SG) considers the vaccination, where υ is the vaccination rate and Θ is the Heaviside step function. Notice that the vaccination campaign is maintained only if SG≠0, i.e., as long as there are still non-vaccinated individuals. If the entire population gets vaccinated at some time TV, then υ=ν for t>TV. Thus, the vaccination rate is adjusted to vaccine individuals at the same rate as they are incorporated into the system. We assume a directly transmitted pathogen, so the disease transmission modelled by the transition S→Ii(i=1,2) is determined by only three factors: the prevalence of the infected, the underlying population contact structure, and the probability of transmission given contact. Hence, we account for homogeneous mixing that dismiss intricate patterns of contacts. The terms proportional to βi(i=1,2) in Eq. (1) are the transmission terms for each strain class, where βi is the transmission coefficient for the ith strain class. Such terms are proportional to the product of the contact probability rates between susceptible and infected individuals [13]. We assume that vaccination provides protection against SC1 and SC2 with different efficiency. The latter assumption is included in the model through a modulated transmission parameter (2) β¯i=βi(1−ηi)(i=1,2), where ηi is the efficiency of the vaccine against strains from the ith class. The parameter μ in Eq. (1) gives the mutation rate per capita at which the virus in I1S-individuals mutates to a strain from the second class. Parameter ρi∈[0,1](i=1,2) represents the per-capita probability of dying from the infection before recovering. The parameter δi in Eq. (1) accounts for the combined effect of the recovery rate γi, d and ρi, and is given by [13] (3) δi=γi+d1−ρi(i=1,2). Parameter w is the rate at which immunity is lost by recovered individuals and moves to either the SG-class or the V-class. Table 1 shows a summary of all parameters in our model. Notice that our model does not consider a latent period, usually present in SEIR-type models. However, it is known that the dynamic properties of the SEIR model are qualitatively similar to those of the SIR model [13]. It is also known that there are asymptomatic cases of COVID-19 where it is unknown how much time infected individuals are transmitting the disease. The estimation from clinical data of the average time of such a latent period is controversial. Therefore, for this study, we assume the worst scenario where the infectious period begins instantly after exposure. It is also generally important to model memory effects as accurate as possible when dealing with public health issues such as the control of SARS-CoV-2. However, it is known that the qualitative dynamics are comparable in models with and without memory [13]. Therefore, only quantitative differences could be important to public health planning in a specific population. In this work, we will not consider memory effects on the behaviour of the infection since we are interested in showing the qualitative behaviour of the system under different vaccination efficiencies. The increased number of parameters of the modified SIR model of Eq. (1) allows a detailed description of the population dynamics and is shown to lead to a manageable and relevant theoretical understanding of the bifurcations occurring in the system.Table 1 Parameters of the modified SIR model (1) estimated from public data-sets [21] (see Appendix). Name Description Value Units ν Population growth rate 3.3×10−5 day−1 d Population decline rate 1.7×10−5 day−1 υ Vaccination rate 0.0043 vaccination/day μ Mutation rate 0.003 mutation/day w Coefficient of waning immunity 0.0046 day−1 β1 Transmission coefficient for the strain-class 1 0.2751 day−1 β2 Transmission coefficient for the strain-class 2 0.6641 day−1 γ1 Recovery rate for the strain-class 1 0.15 day−1 γ2 Recovery rate for the strain-class 2 0.05 day−1 ρ1 Probability of dying due to infection with the strain-class 1 0.013 dimensionless ρ2 Probability of dying due to infection with the strain-class 2 0.028 dimensionless η1 Efficiency of vaccines against the strain-class 1 Variable dimensionless η2 Efficiency of vaccines against the strain-class 2 Variable dimensionless 3 Endemic and disease-free equilibria: stability analysis and bifurcations 3.1 Fixed points We determine the long-term behaviour of the dynamical system (1) by calculating its fixed points and stability. System (1) has a disease-free (DF) fixed point given by (4) Sf∗,If∗,Rf∗=Vf∗,0,0=νd,0,0, where all the populations are zero except for the vaccinated class. There are also two endemic fixed points: one endemic equilibrium for strain class 1 (EE1) and another endemic equilibrium for strain class 2 (EE2), given by (5) Si∗,Ii∗,Ri∗=V(i)∗,IiV∗,RV(i)∗=δid+γi+wΛi,β¯i−δid+wΛi,β¯i−δiΛi, where Λi=ν/δid−δi+γid+w+dδiβ¯i+β¯iwδi−γi with i=1,2. To eradicate the disease, the parameters of the system must be tuned to have a stable DF fixed point. On the other hand, if the endemic fixed point for the ith strain class is stable, then IiV approaches asymptotically to a constant value in time, and the disease remains bounded in numbers within the population. In this latter scenario, IiV∗ must be as small as possible to bound the accumulated number of deaths caused by the disease and avoid the collapse of hospitals and the economic impact due to isolation and quarantines. Also, Eq. (5) shows that IiV∗ is proportional to the difference β¯i−δi. Hence, from Eq. (2), we conclude that it is convenient to tune ηi to decrease IiV∗ as much as possible, thus keeping the infection under control. One immediate conclusion that can be drawn from Eq. (5) is that the two strain classes cannot coexist within the same population in the long term. There are two fixed points that clearly separate the two strain classes. Thus, the dynamical system naturally selects the lasting strain as the population approaches an endemic equilibrium, where one of the strains becomes extinct. If a strain is more contagious and deadly than others and is also more resistant to vaccines, it will extinguish other weaker strains during a transient period and remain endemic in the population in the long term. This behaviour has already been reported in genomic-sequence studies of COVID-19 cases worldwide. For instance, in Chile, the ancestral strain with the Spike-D614G protein mutation was the dominant variant in sequenced cases between March 2 and April 5, 2020 [24]. The Delta variant was detected in October 2020 in India and became dominant worldwide in November 2021, conforming more than 99% of the total cases of SARS-CoV-2 [25]. Nowadays, the ancestral strain is almost extinct worldwide, and the circulation of the Delta strain has significantly decreased after competition with the Omicron variants. Now the Omicron are the only circulating strains categorised as variants of concern by the WHO [26]. Individuals within the non-vaccinated categories vanish as one approaches one of the endemic fixed points. Notice that our model does not consider the effects of anti-vax attitudes in individuals within the SG pool. Thus, there are two possible fates for the non-vaccinated population: (i) they are vaccinated and thus contribute to the transition SG→V, or (ii) they die due to the infection or by other independent causes. As we will show later, the system can approach an endemic fixed point in a time of the order of months, so most of the deaths before approaching one of the endemic fixed points are due to infection rather than other causes. 3.2 Stability of fixed points, basic reproductive ratios and bifurcations The linear stability analysis performed after linearising the system (1) around fixed points is algebraically challenging. We overcame the problem by rewriting Eq. (1) as a reduced model. As we show in the following, the reduced model is equivalent to the original system if some conditions are provided. From Eq. (1), the global dynamics of the S, I and R populations are governed by (6a) S˙=ν+wR−dS−1Nβ1S−η1VI1+β2S−η2VI2, (6b) I˙=−δ1I1−δ2I2+1Nβ1S−η1VI1+β2S−η2VI2, (6c) R˙=γ1I1+γ2I2−d+wR. We consider two cases where each endemic equilibrium can collide with the DF point via transcritical bifurcations. 3.2.1 Case 1: Endemic equilibrium 2 bifurcating transcritically with the disease-free equilibrium First, we assume that SC2 is much more contagious than SC1, i.e. β2≫β1. Assuming that N∼SG∼V and η1∼η2, Eqs. (6a), (6b) reduce to (7a) S˙=ν+wR−dS−1Nβ2S−η2VI2, (7b) I˙=−δ1I1−δ2I2+1Nβ2S−η2VI2. If SC2 is more deadly or displays a shorter infectious period than SC1, then δ2≫δ1 and I1 evolves in a different time scale than I2. Notice that the nonlinear term in Eq. (7b) does not counteract the linear term −δ1I1, which gives an exponential decay of I1. Thus, after an initial transient characterised by an increase of I1(t) up to some maximum value, we expect an exponential decay for I1(t) with a characteristic time given by (8) τ1=1−ρ1γ1+d. Thus, under the assumptions β2≫β1 and δ2≫δ1, the SC1 decays exponentially and becomes extinct after competition with SC2 at a time approximately equal to 2τ1. Notice from Eq. (8) that the characteristic time τ1 is small if individuals infected with the SC1 have a high probability of dying from the disease or have a short infectious period. This remark is in accordance with the intuitive idea that successful mutations of a strain will try to increase their contagious period as much as possible or decrease their lethality to avoid a fast extinction. Given that I1(t) decays exponentially after a transient, the contribution γ1I1 in Eq. (6c) is negligible for t≫2τ1. Thus, for t≫2τ1, the system simplifies to (9a) S˙=ν+wR−dS−1Nβ2S−η2VI2, (9b) I˙=−δ2I2+1Nβ2S−η2VI2, (9c) R˙=γ2I2−d+wR, whose dynamics are represented by the reduced diagram shown in Fig. 2(a). Eventually, all individuals within the SG-class will receive vaccination after some time Tυ. Thus, for t≫Tυ>2τ1, we have SG→0, S→V, and the system (9) is reduced to the following third-order dynamical system (10a) V˙=ν+wRV−dV−1Nβ¯2VI2V, (10b) I˙2(V)=−δ2I2V+1Nβ¯2VI2V, (10c) R˙V=γ2I2V−d+wRV, which is the conventional SIR model with demography and waning immunity [13], represented by the flow diagram depicted in Fig. 2(b). The reduced dynamical system (10) captures the long-term asymptotic behaviour of the complete system (1) if β1≪β2 and δ1≪δ2. Indeed, Eqs. (4), (5) for i=2 are also fixed points of the reduced dynamical system (10). In this case, the EE1 is not stable, and the SC1 becomes extinct.Fig. 2 Reduced SIR models capturing the dynamics of the original system of Eq. (1) for β2≫β1, δ2≫δ1, and (a) 2τ1<t<Tυ, following Eq. (9), and (b) t≫Tυ, following Eq. (10). With the reduced system (10), it is possible to compute a relevant parameter: the basic reproductive ratio for SC2, denoted here as R0,2. Such parameter measures the maximum reproductive potential of infectious disease [27] and is defined as the number of secondary cases arising from a primary case in average in an entirely susceptible population. It is also interpreted as the rate at which new cases are produced by an infectious individual multiplied by the average infectious period. Note from Eq. (10b) that infected individuals spend an average (1−ρ2)/(γ2+d) time units within the infectious class 2. Thus, the average number of new infections caused per infected individual is the transmission rate times the infection period, i.e. (11) R0,2=β¯2δ2. Performing the linear stability analysis of the system (10) around the DF equilibrium, we obtain three eigenvalues, (12) λ1=−d,λ2=−d−w,λ3=β¯2−δ2. Given that d>0 and w>0, from Eqs. (11), (12) we conclude that there is a bifurcation point at R0,2=1, where the EE1 collides with the DF point in a transcritical bifurcation. If R0,2<1 (R0,2>1), the DF point is a stable (saddle) point. Fig. 3 shows the position of the equilibria of the system in the three-dimensional SIR space, as given by Eqs. (4), (5). Here, the position of the ith endemic equilibrium is shown as a parametric curve of ηi, with i=1,2. We have complemented our theoretical results on the linear stability analysis with numerical continuation studies performed on the reduced three-dimensional system of Eq. (10). In Fig. 3, we show in solid (dashed) lines the points where each endemic equilibrium is stable (saddle). Notice that each endemic equilibrium can be stable only in the first octant of the space SIR. When ηi=ηi(c), where ηi(c) is some critical value, the ith endemic equilibrium collide with the DF point, and their stability properties are exchanged. After the bifurcation condition, the DF remains in the same position. In contrast, the endemic point leaves the first octant and moves away from the DF equilibrium, which is a clear evidence of a transcritical bifurcation.Fig. 3 Parametric curves showing the position of the equilibria of the system, with the vaccine efficiency η1 (η2) as the parameter of the EE1(EE2) curve. Transcritical bifurcations occur at ηi=ηi(c), where the ith endemic equilibrium collides with the disease-free equilibrium (i=1,2). The intrinsic parameter values are shown in Table 1. 3.2.2 Case 2: Endemic equilibrium 1 bifurcating transcritically with the disease-free equilibrium We have shown the case where the more contagious strain SC2 extinguishes SC1. Remarkably, here we show that under certain conditions, the endemic fixed points can switch roles. If the efficiency of vaccines against SC1 decreases far below a critical value (for example, due to waning immunity from vaccines or a sudden decrease in their quality), SC2 may become extinct. The system would be characterised by an exponential decay of I2(t) and a stable endemic equilibrium for SC1. This scenario can happen even if β2 is still greater than β1. Suppose in Eq. (6) that η1<η2 such that β¯1≫β¯2 and simultaneously β2≫β1. Keeping the dominant terms in Eq. (6) and considering that after a time Tυ all population within the SG-class becomes vaccinated, Eq. (6) reduces to (13a) S˙=ν+wR−dS−1Nβ¯1VI1, (13b) I˙=−δ1I1−δ2I2+1Nβ¯1VI1, (13c) R˙=γ1I1+γ2I2−(d+w)R. Here, the nonlinear term in Eq. (13b) does not counteract the linear decay term −δ2I2. Thus, in this scenario, we expect the extinction of the SC2, where I2(t) describes an exponential decay for t≥Tυ with characteristic time τ2=1/δ2 [in analogy with Eq. (8)]. Finally, we arrive at a completely analogous third-order system as Eq. (10), replacing β¯2→β¯1, δ2→δ1, γ2→γ1, and I2V→I1V. Eqs. (4), (5) for i=1 are now the fixed points of the reduced dynamical system, which now captures the endemic equilibrium for the SC1. Thus, we have a basic reproductive ratio for SC1 analogous to Eq. (11), given by (14) R0,1=β¯1δ1. Similarly, the EE1 will bifurcate transcritically with the DF equilibrium at R0,1=1. Fig. 4 (a) Parameter space of the system for χ<χmax. The stable fixed point in each region is indicated in parentheses (the remaining equilibria are saddle points). The shaded regions comprises the strain competition region, where both endemic equilibria coexist and compete. Along the orange solid line of slope χ, there is a fold where the endemic equilibria exchange their stability. (b) Diagram with the different types of observed bifurcations. One fold and four transcritical bifurcations can occur, where one endemic equilibrium collides with the DF point. Different lines indicate different bifurcations (see explanation in the text). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 3.2.3 Case 3: Fold between endemic equilibria The reproductive ratio of a given strain class is one of the parameters determining if it will remain endemic in the population in the long term. From the results discussed above, if only one of the two reproductive ratios is above unity, the corresponding strain will extinguish the other and remain endemic within the population. Notwithstanding, there is an intermediate situation where both reproductive ratios are above unity. Given that the two strain classes cannot coexist within the population in an equilibrium state, both strains will compete for the limited population of susceptible individuals, and the system will approach the endemic equilibrium corresponding to the strain with the largest reproductive ratio. Indeed, both endemic fixed points coexist if R0,1>1 and R0,2>1, being stable (saddle) the one with the largest (smallest) reproductive ratio. A fold occurs at R0,1=R0,2, where both fixed points exchange their stability properties. 3.3 Bifurcations and strain competition In previous sections, we have shown that by changing parameters η1 and η2, each endemic equilibrium can change its stability and collide independently with the DF equilibrium via a transcritical bifurcation. The bifurcation point for each strain determines a critical value of the vaccine efficiency against the corresponding strain, given by (15) ηi(c)=1−γi+dβi(1−ρi),i=1,2. Suppose that ηi<ηi(c) with i=1,2. As it has been discussed, strain competition leads to the extinction of the strain with the lowest reproductive ratio. After that, increasing ηi above the critical value ηi(c), where i denotes the surviving strain with the largest reproductive ratio, collides the associated endemic equilibrium with the DF equilibrium, and consequently exchanging their stability properties via a transcritical bifurcation. Moreover, the ith endemic equilibrium becomes inaccessible to the system after the bifurcation since at least one of its coordinates becomes negative and loses biological sense. Thus, having ηi>ηi(c) for both strains is favourable for the population if we wish to eradicate the disease. Note that a highly contagious and lethal strain will have large values of both βi and ρi, requiring a vaccine efficiency near unity to eradicate the disease. Fig. 4 shows the parameter space of the system, summarising our theoretical predictions on the long-term behaviour of the dynamical system. In Fig. 4(a), we enumerate the different regions of the parameter space, which are separated by boundaries where different bifurcations can occur. The possible bifurcations are shown in the diagram of Fig. 4(b), where half-filled circles (◐) denote saddle points, and filled circles (●) denote attractors. In regions I and II, the endemic equilibrium with the largest (smallest) reproductive ratio is a stable (saddle) point. Such regions comprise what we labelled as the strain competition region, and is depicted as a shaded region in Fig. 4(a). The condition R0,1=R0,2 is fulfilled along the orange solid line separating regions I and II, as shown in Fig. 4(a), whose locus are solutions to the equation (16) η2=χ(η1−1)+1,χ=β1δ2β2δ1, where 0<χ<χmax. Here, χmax is the maximum value that χ can have and is given by (17) χmax=η2(c)−1η1(c)−1. Fig. 4(a) shows five possible bifurcations for χ<χmax. Below we summarise the properties of the system in each region of parameter space: • Region I: The EE1 is the attractor of the system, whereas the EE2 and the DF equilibrium are saddle points. This region is the upper part of the strain competition region, and is bounded from below by the fold line (16) where endemic points exchange their stability properties. The region is bounded from above by the bifurcation type 1, where the saddle EE2 and the saddle DF equilibrium collide via a transcritical bifurcation. The region is also bounded from the right by the bifurcation type 2. Approaching the bifurcation type 2 from region I, the stable EE1 collides with the saddle DF point in a transcritical bifurcation. • Region II: The EE2 is the attractor of the system, whereas the EE1 and the DF equilibrium are saddles. This region is the lower part of the strain competition region, is bounded from above by the fold line, and from the right by the bifurcation type 3, where the saddle EE1 collides with the saddle DF equilibrium via a transcritical bifurcation. • Region III: The EE1 is the attractor of the system, whereas the EE2 and the DF equilibrium are saddles. The region is bounded from below by the transcritical bifurcation type 1 and from the right by the bifurcation type 2. The same bifurcation is observed approaching the bifurcation type 2 either from region I or from region III. • Region IV: The EE2 is the attractor of the system, whereas the EE1 and the DF equilibrium are saddles. The region is bounded from the left by the bifurcation type 3 and from above by the bifurcation type 4. Approaching the bifurcation type 4 from region IV, the stable EE2 collides with the saddle DF point via a transcritical bifurcation. • Region V: The DF equilibrium is the attractor, whereas the EE1 and the EE2 are saddles. This region is bounded from the left by the bifurcation type 2 and from below by the bifurcation type 4. Approaching the bifurcation type 2(4) from region V, the saddle EE1(EE2) collides with the stable DF point in a transcritical bifurcation. This region of parameter space is the most favourable for the population. It is worth noting that a double-bifurcation point can occur if χ=χmax, where the crossing point indicated with a blue diamond in Fig. 4 coincide with the crossing point between the thresholds η1=η1(c) and η2=η2(c). However, the parameters of the system should be fine-tuned to observe such a special bifurcation: the reproductive ratio of both strains must be equal, and the efficiency of vaccines must match their critical values. Moreover, although it is realistic to tune vaccine efficiencies, it is still an open question whether it is possible to alter the recovery rate of COVID-19 patients, which would be the only realistic way to change the reproductive ratio of both strain classes with ηi fixed. Hence, the double-bifurcation will not be observed in practical situations. 4 Numerical simulations and continuation of stationary solutions: time evolution of cases and bifurcations We perform a numerical study of the dynamical system (1) in parameter space (η1,η2) through two methods of analysis: using numerical continuation techniques [28], [29], [30] and using direct numerical simulations of Eq. (1). We estimate the parameters of the system based on the public database on the COVID-19 pandemic from the European Centre for Disease Prevention and Control (ECDC) [21]. In Appendix, we give the details on the methods for the estimation of such parameters, which are summarised in Table 1. We have used these values of parameters for the parametric curves of Fig. 3 and for the numerical simulations discussed below. 4.1 Numerical continuation Using numerical continuation, we analyse the stationary solutions and bifurcations of the dynamical system (1). The bifurcation diagrams are shown in Fig. 5. To analyse all the bifurcations, we follow a closed loop the in parameter space from A to B [Fig. 5(a)], B to C [Fig. 5(b)], C to D [Fig. 5(c)], and D to A [Fig. 5(d)]. • PathA→B: Starts in region IV and crosses to region II through the bifurcation type 3. In accordance with our predictions in Section 3, the EE1 and the DF equilibrium are saddle points and collide via a transcritical bifurcation. The EE2 remains stable and unchanged along this path. Notice that the I and R components of the EE1 become negative for η1>η1(c). Thus, the EE1 leaves the first octant of the SIR space after the transcritical bifurcation, as we already noticed in our theoretical analysis in Section 3 [see Fig. 3]. • PathB→C: Starts in region II, passes through the region I, and ends up in region III. Thus, the path crosses two boundary lines: the fold and the bifurcation type 1. As expected, the DF equilibrium is still a saddle point and remains unchanged along this path. Strain competition takes place in regions I and II. Since point B is in region II (below the fold condition), the EE2 is an attractor, whereas the EE1 is a saddle point. At the fold condition, the endemic equilibria exchange their stability properties, and the EE1 becomes the attractor of the system. Remarkably, the bifurcation diagrams in Fig. 5(b) confirm that such exchange occurs at distance: both points are separated by a small distance at the fold condition. Further increasing the value of η2 above the fold condition, we observe the transcritical bifurcation type 1. Similarly, as in the case discussed above, the EE2 leaves the first octant of the SIR space after the transcritical bifurcation, in accordance with Fig. 3. • PathC→D: Goes from region III to region V crossing the transcritical bifurcation type 2. The bifurcation diagrams shown in the rightmost panels of Fig. 5(c) confirm that the EE2 remains outside the first octant of the SIR space as a saddle point all along this path. At point C, the EE1 is stable, whereas the DF equilibrium is a saddle point. As we approach the threshold, the EE1 approaches the DF point and collides via a transcritical bifurcation. For η1 above the threshold, the saddle EE1 leaves the first octant of the SIR space, and the DF equilibrium becomes the attractor of the system. • PathD→A: Goes from region V to region IV, crossing the transcritical bifurcation type 4. The EE1 remains outside the first octant of the SIR space as a saddle point. At point D, η2>η2(c) and the DF equilibrium is the attractor of the system, whereas the EE2 is a saddle point. Crossing the bifurcation type 4, the EE2 collides with the DF equilibrium and exchanges their stability in a transcritical bifurcation. Fig. 5 Bifurcation diagrams of the dynamical system (1). Results obtained using numerical continuation along a closed-loop joining regions (a) IV and II (path A→B, with 0.4≤η1≤0.6 and η2=0.9), (b) II, I, and III (path B→C, with 0.9≤η2≤0.95 and η1=0.4) (c) III and V (path C→D, with 0.4≤η1≤0.6 and η2=0.95), and (d) V and IV (path D→A, with η1=0.6 and 0.9≤η1≤0.95). The left panel show the path in parameter space. The rightmost panels show the SIR components of the stationary solutions. The theoretical thresholds of the fold and transcritical bifurcations are shown with vertical solid lines. Stable (saddle) points are denoted with solid (dashed) lines. Fig. 6 Numerical simulations of the time evolution of the modified SIR model. (a) Region I: η1=0.20 and η2=0.90. (b) Region II: η1=0.20 and η2=0.85. (c) Region III: η1=0.20 and η2=0.95. (d) Region IV: η1=0.80 and η2=0.90. (e) Region V: η1=0.60 and η2=0.97. The parameters used are shown in Table 1. Arrows in the inset denote time direction. The initial condition for each case is indicated with a green triangle. 4.2 Time evolution of cases We also performed direct numerical simulations of Eq. (1) to characterise the time evolution of the modified SIR model given some initial conditions. For the time integration, we used the Dormand-Prince algorithm as an explicit Runge–Kutta of orders 4 and 5 with an adaptive time step [31]. Based on our assumptions and public-data analysis, we use at t=0 the values I1S=0.0042, SG=0.9958, and V=I1V=I2S=I2V=RS=RV=0 [see Appendix]. Fig. 6 shows the trajectory of the dynamical system in each of the five regions indicated in the parameter space of Fig. 4. The trajectories are shown along with the position of the endemic equilibria, EE1 and EE2, and the DF point. In each case, we indicate the corresponding region in parameter space. Fig. 6(a) shows the resulting trajectory for (η1,η2) in region I. After an initial burst in the number of infections, the system approaches the EE1, where the number of infections with strain-class 1 exhibits damped oscillations around the EE1. In Fig. 6(b), we cross the fold line in the parameter space of Fig. 4, and the trajectory approaches the EE2, the attractor of the system in the corresponding region. After the initial burst of infections, the trajectory exhibits a short-lived oscillation before approaching the EE2 monotonically. Fig. 6(a) and (b) illustrate the phenomenon of strain competition, where both endemic equilibria coexist and compete for resources. The stable equilibrium corresponds to the strain with the largest reproductive ratio. In Fig. 6(c), the resulting trajectory is qualitatively similar to Fig. 6(a). Indeed, the EE1 is the attractor of the system in regions I and III. Similarly, the trajectory in Fig. 6(d) is qualitatively similar to Fig. 6(b). However, in contrast to regions I and II, the outcome of the system in regions III and IV are not due to strain competition. In Fig. 6(c), η2>η2(c) and the EE2 is a saddle point outside the first octant. Similarly, in Fig. 6(d), η1>η1(c) and the EE1 is outside the first octant.Fig. 7 Numerical simulation and characteristic decay time with η1=0.5 and η2=0.9, where η1 and η2 are in region IV. The fitted values of the parameters in panel (e) are a=0.4567 with a∈[0.4561,0.4573], b=4.606 with b∈[4.605,4.607], and R-squared 0.9998. Fig. 6(e) shows the trajectory in region V, where the DF equilibrium is the attractor of the system. In this case, there is an initial burst in the number of infections, reaching a maximum. Later, the number of infections decreases to zero, and the trajectory passes near the initial condition of the simulation, as shown in the zoom-in of the boxed region in the inset of Fig. 6(e). After that, the trajectory goes monotonically towards the DF equilibrium. This scenario corresponds to the case where the vaccine efficiency is high against both strain classes, reducing the number of infected individuals to zero in a finite time. Finally, we have estimated from numerical simulations the characteristic time τ1 given by Eq. (8). In Fig. 7, we show the time evolution of the total population N, SG and V [Fig. 7(a)], individuals infected with strain-class 1 [Fig. 7(b)], individuals infected with strain-class 2 [Fig. 7(c)], and recovered individuals [Fig. 7(d)]. The values of the parameters correspond to region IV. Notice that the SC1 becomes extinct due to competition with SC2. The latter remains endemic in the population in the long term. This behaviour is observed in regions from I to IV, where one of the endemic equilibria is stable. In Section 3.2.1, we predicted that I1(t) decays exponentially after an initial burst, as evidenced in Fig. 7(b). In Fig. 7(e), we fit the numerical outcome for the I1S individuals with an exponential decay law for t∈[0.6,2]month. We obtain τ1fit=0.21711±10−5month, which is near the theoretical value τ1=0.21931month. 5 Conclusions Motivated by the ongoing COVID-19 pandemic caused by multiple strains of SARS-CoV-2, we pose a modified SIR model with waning immunity capturing strain competition between two classes of strains of an infectious virus under the effect of vaccination. We consider a mutation parameter that characterises the rate at which the SC1 mutates into a more deadly and contagious strain class, the SC2. We hypothesise that vaccination modulates the transmission coefficient of each strain, decreasing its value by a proportion given by their efficiency. Our model assumes that vaccination does not provide full immunity against any of the strain classes. We characterise the parameter space of the system and bifurcations using the vaccine efficiencies as control parameters. The basic reproductive ratio is determined for each strain class. We determined a region in parameter space where the disease-free equilibrium is stable, which is the most favourable scenario for eradicating the disease. We also found a region in which both strain classes coexist and compete. Our model shows that strain competition always leads to the extinction of one of the strain classes. In the region of strain competition, we show that after a transient period, the strain with the largest reproductive ratio remains endemic in the long term. We obtain the minimum value of the vaccine efficiency against each of the strain classes to eradicate the disease. Such critical efficiency depends on the transmission rate, the infectious period, and the probability of dying due to the infection from each strain class. We also derive an expression for the characteristic exponential decay time at which the strain extinguishes. We estimated the values of the parameters of our modified SIR model based on public databases on the COVID-19 pandemic from the ECDC. The estimation from public data of some key parameters of SIR models, such as the transmission coefficients and the mutation rate, is generally challenging and remains as an open question [13]. An alternative approach is to preliminarily estimate parameters that can be easily inferred from clinical and demographic data through our methods. Then, once the unknown number of parameters is reduced, the remaining ones can be estimated using optimisation algorithms for ordinary differential equations [32], [33], [34] Direct numerical simulations of the time evolution of the dynamical system and numerical characterisations of bifurcations using numerical continuation techniques are in good agreement with theoretical results. We conclude that the efficiency of vaccines controls the stability of the endemic equilibria. Thus, vaccines are determinant in the long-term behaviour of the pandemic. A combination of vaccine efficiencies against both strain classes could yield an endemic SC1, an endemic SC2, or even eradicate the disease. Moreover, we showed that the infectious period controls the characteristic decay time of the exponentially decaying strain. Our model is versatile and can be extended to include more complex features. For example, seasonality in disease transmission and vaccination campaigns can be modelled through a time-periodic modulation on the corresponding parameters [13], [35], mutation and demography can be considered stochastic introducing random variables in the corresponding terms [36], [37], and memory effects in the behaviour of the disease can be considered by replacing nonlinear terms by their integrodifferential formulations [38], [39]. Our study emphasises the importance of SARS-CoV-2 genomic surveillance in regions where the virus is freely transmitted, a condition that increases the probability of emergent variants of concern due to mutation. Surveillance programmes are key in combination with a continuous improvement of vaccine technology facing the emergence of new variants of concern. CRediT authorship contribution statement M. Ahumada: Conceptualization, Software, Investigation, Formal analysis, Visualization, Writing – review & editing. A. Ledesma-Araujo: Software, Validation, Investigation, Formal analysis. Leonardo Gordillo: Investigation, Validation, Writing – review & editing. Juan F. Marín: Methodology, Supervision, Investigation, Formal analysis, Conceptualization, Data curation, Visualization, Writing – original draft. 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 Data analysis and parameterization We rely on freely available data on the COVID-19 pandemic from the ECDC [21]. The public data used in this study comprises the number of people fully vaccinated against COVID-19, the number of active cases per day, and the infection fatality rate (IFR) as a function of time. The estimate of the population growth is retrieved from the World Population Prospects 2019 Revision by the United Nations (UN) population division [40], whereas the demographic parameters ν and d, are gathered from the World Bank collection (WBC) of development indicators [41]. For the initial conditions, we considered that infected individuals join the population under study at t=0, and at the same time, the vaccination campaign starts. Thus, V(t=0)=I1V(t=0)=I2S(t=0)=I2V(t=0)=RS(t=0)=RV(t=0)=0. According to UN sources, the total population in Chile by 2021 was 19.116.209 inhabitants [40]. This population correspond to the initial condition N(t=0)=1 in our numerical simulations. To estimate a realistic value for the initial infected population, I(t=0), we retrieved the number of active cases as a function of time from public data [42], depicted in Fig. A.8(a). We chose for I(t=0) the condition corresponding to the maximum number of active cases, which occurred on June 3, 2020, with 80.933 active cases across the country. This corresponds to I(t=0)=0.0042, which is assigned to the sub-class I1S at t=0. This is how infections with strain-class 2 are seeded by mutation. Thus, our initials conditions are I1S(t=0)=0.0042, SG(t=0)=0.9958, with the remaining sub-classes equal to zero.Fig. A.8 Data retrieved from public databases on the COVID-19 pandemic in Chile. (a) Number of active cases per day. (b) Number of individuals who received all prescribed doses of vaccines (dotted line). The third-order polynomial fitting of the data is shown in solid red line. (c) The infection fatality rate as a function of time used to estimate the mortality risk of each strain class. To estimate the vaccination rate υ, we retrieved the number of people who received all doses prescribed by the vaccination protocol in Chile as a function of time. The number of vaccinated people shows a marked increase after March 2, 2021 [43], as shown in Fig. A.8(b). We fit the data from such date on, using a third-order fitting polynomial pv(t), the highest order polynomial that gives a well-posed fitting of the data. Finally, we estimate the vaccination rate as p˙vt=tl=81723[people/day], where tl denotes the time of lecture of the data. Normalising the vaccination rate to the total population in Chile by 2021 gives υ=0.0043[vaccination/day]. We estimate the value of the mutation rate μ as the inverse of the time Tμ=10[month] between the first reported case of COVID-19 and the first detection of the Delta variant [21]. The first case was notified in China on December 31, 2019 [3], whereas the first case of the Delta variant was notified in India on October 2020 [7]. Thus, mutations leading to a VOC happens at a rate μ=0.003[mutation/day], approximately. Given that the population decline rate d excludes the deaths due to the infection, we estimate d from the number of deaths just after the COVID-19 breakout in Chile. The number of deaths in 2019 was 117490 [44]. Normalising by the total population in Chile, the population decline rate is d=1.7×10−5[1/day]. The birth rate in Chile in 2020 was 227040 births [40]. Similarly, after normalisation by the total population in Chile, we obtain a population growth rate ν=3.3×10−5[1/day]. To estimate the mortality risk of each strain class, we analysed the data of the infection fatality rate (IFR), defined as the number of deaths from the disease divided by the total number of cases. Although the total number of cases of COVID-19 is not precisely known, mainly because not every infected individual is tested [45], the IFR calculated from observed data is a good estimation of how likely it is for someone infected to die from the disease [21], [46]. Fig. 6(c) shows the IFR as a function of time observed in Chile [21]. We estimate the value of ρ2 as the global maximum of the IFR, which was reached soon after 200 days from February 24, 2020. We estimate the value of ρ1 as the minimum IFR reached after the global maximum. We considered that immunity remained seven months after infection, based on a recent study of the risk for reinfection after SARS-CoV-2 with wild-type or Alpha variants [47]. Following the study of Byrne et al. [48], the duration of the infection period for the classes I1 and I2 are T1=6.5[day] and T2=18.1[day], respectively. From these values, we obtain the recovery rate from the ith strain class as γi=(1−ρi)/Ti, with i=1,2. The resulting values are depicted in Table 1. Finally, to estimate the transmission constants β1 and β2, we analyse the effective reproduction rate (ERR) from the epidemiological data [21]. The largest observed value of the ERR in Chile was the first measure, R0,2max=2.91, reported on March 16, 2020. On January 13, 2022, during the rising of the Omicron variant in Chile, the ERR reached the local maximum R0,1max=1.81. Following Eqs. (11), (14), we compute the transmission constants as βi=R0,imaxδi with i=1,2. Data availability Data will be made available on request. Acknowledgments We are thankful to the two anonymous reviewers of a prior manuscript who provided insightful comments and suggestions. A.L.A. thanks ANID-Subdirección de Capital Humano/Doctorado Nacional /2021-21211330 for the financial support. J.F.M. acknowledges the financial support of ANID , through the grant FONDECYT /POSTDOCTORADO/3200499. The authors dedicate this article to the memory of Prof. Enrique Tirapegui. ==== Refs References 1 Keogh-Brown M.R. Jensen H.T. Edmunds W.J. Smith R.D. The impact of Covid-19, associated behaviours and policies on the UK economy: A computable general equilibrium model SSM - Popul Health 12 2020 100651 2 Aucouturier M. Herrmann H.J. Modeling the coupling between a COVID-19-like epidemic and the economy Internat J Modern Phys C 32 11 2021 2150150 3 Ji W. Wang W. Zhao X. Zai J. Li X. Cross-species transmission of the newly identified coronavirus 2019-nCoV J Med Virol 92 4 2020 433 440 31967321 4 Tang J.W. Tambyah P.A. Hui D.S. Emergence of a new SARS-CoV-2 variant in the UK J Infect 82 2021 E27 E28 5 Tang J.W. Toovey O.T. Harvey K.N. Hui D.D. Introduction of the South African SARS-CoV-2 variant 501Y. V2 into the UK J Infect 82 2021 E8 E10 6 Voloch C.M. Genomic characterization of a novel SARS-CoV-2 lineage from Rio de Janeiro, Brazil J Virol 95 10 2021 7 o. S. A. I. D (GISAID) G.I. Tracking of variants 2022 Available online at: https://www.gisaid.org/hcov19-variants/ (accessed June 10, 2022) 8 Callaway E. Fast-evolving COVID variants complicate vaccine updates Nat News 607 2022 18 19 9 Zahradník J. Marciano S. Shemesh M. Zoler E. Chiaravalli J. Meyer B. Rudich Y. Dym O. Elad N. Schreiber G. SARS-CoV-2 RBD in vitro evolution follows contagious mutation spread, yet generates an able infection inhibitor Nature Microbiol 6 2021 1188 1198 34400835 10 Wang Q. Zhang Y. Wu L. Niu S. Song C. Zhang Z. Lu G. Qiao C. Hu Y. Yuen K.-Y. Wang Q. Zhou H. Yan J. Qi J. Structural and functional basis of SARS-CoV-2 entry by using human ACE2 Cell 181 2020 894 904.e9 32275855 11 Wibmer C.K. Ayres F. Hermanus T. Madzivhandila M. Kgagudi P. Oosthuysen B. Lambson B.E. De Oliveira T. Vermeulen M. Van der Berg K. SARS-CoV-2 501Y. V2 escapes neutralization by South African COVID-19 donor plasma Nature Med 27 4 2021 622 625 33654292 12 Kermack W.O. McKendrick A.G. A contribution to the mathematical theory of epidemics Proc R Soc Lond Ser A Math Phys Eng Sci 115 772 1927 700 721 13 Keeling M.J. Rohani P. Modeling infectious diseases in humans and animals 2011 Princeton University Press 14 Fanelli D. Piazza F. Analysis and forecast of COVID-19 spreading in China, Italy and France Chaos Solitons Fractals 134 2020 109761 15 Ndaïrou F. Area I. Nieto J.J. Torres D.F. Mathematical modeling of COVID-19 transmission dynamics with a case study of Wuhan Chaos Solitons Fractals 135 2020 109846 16 Cooper I. Mondal A. Antonopoulos C.G. A SIR model assumption for the spread of COVID-19 in different communities Chaos Solitons Fractals 139 2020 110057 17 Sarkar K. Khajanchi S. Nieto J.J. Modeling and forecasting the COVID-19 pandemic in India Chaos Solitons Fractals 139 2020 110049 18 Wagner C.E. Saad-Roy C.M. Grenfell B.T. Modelling vaccination strategies for COVID-19 Nature Rev Immunol 22 2022 139 141 35145245 19 Shepherd A. Covid-19: Chile joins top five countries in world vaccination league Br Med J News 718 2021 372 20 Mallapaty S. Callaway E. Kozlov M. Ledford H. Pickrell J. Van Noorden R. How COVID vaccines shaped 2021 in eight powerful charts Nat News 600 2021 580 583 21 Roser M. Ritchie H. Ortiz-Ospina E. Coronavirus disease (COVID-19) – Statistics and research Our World Data 2021 https://ourworldindata.org/coronavirus (accessed: April 16th 2021) 22 Altmann D.M. Boyton R.J. Waning immunity to SARS-CoV-2: implications for vaccine booster strategies Lancet Respir Med 9 2021 1356 1358 34688435 23 Suryawanshi R.K. Limited cross-variant immunity from SARS-CoV-2 Omicron without vaccination Nature 2022 10.1038/s41586-022-04865-0 24 Castillo A.E. Parra B. Tapia P. Lagos J. Arata L. Acevedo A. Andrade W. Leal G. Tambley C. Bustos P. Fasce R. Fernández J. Geographical distribution of genetic variants and lineages of SARS-CoV-2 in Chile Front Public Health 8 2020 25 Organization W.H. Weekly epidemiological update on COVID-19 - 2 november 2021 2021 Available online at: https://www.who.int/publications/m/item/weekly-epidemiological-update-on-covid-19--2-november-2021 (accessed June 10, 2022) 26 Organization W.H. Weekly epidemiological update on COVID-19 - 8 june 2022 2022 Available online at: https://www.who.int/publications/m/item/weekly-epidemiological-update-on-covid-19--8-june-2022 (accessed June 10, 2022) 27 Diekmann O. Heesterbeek J.A.P. Mathematical epidemiology of infectious diseases: model building, analysis and interpretation, Vol. 5 2000 John Wiley & Sons 28 Allgower E.L. Georg K. Numerical continuation methods: an introduction, Vol. 13 2012 Springer Science & Business Media 29 Krauskopf B. Osinga H.M. Galán-Vioque J. Numerical continuation methods for dynamical systems, Vol. 2 2007 Springer 30 Doedel E. AUTO 2019 Software for continuation and bifurcation problems in ordinary differential equation: http://indy.cs.concordia.ca/auto/ 31 Shampine L.F. Reichelt M.W. The MATLAB ODE suite SIAM J Sci Comput 18 1 1997 1 22 32 Bonnans J.-F. Gilbert J.C. Lemaréchal C. Sagastizábal C.A. Numerical optimization: theoretical and practical aspects 2006 Springer Science & Business Media 33 Garvie M.R. Maini P.K. Trenchea C. An efficient and robust numerical algorithm for estimating parameters in Turing systems J Comput Phys 229 19 2010 7058 7071 34 Glasner K. Optimization algorithms for parameter identification in parabolic partial differential equations Comput Appl Math 40 2021 146 35 Bailey N.T. The mathematical theory of infectious diseases and its applications 1975 Charles Griffin & Company Ltd 5a Crendon Street, High Wycombe, Bucks HP13 6LE 36 Bjørnstad O.N. Finkenstädt B.F. Grenfell B.T. Dynamics of measles epidemics: estimating scaling of transmission rates using a time series sir model Ecol Monograph 72 2 2002 169 184 37 Dexter N. Stochastic models of foot and mouth disease in feral pigs in the Australian semi-arid rangelands J Appl Ecol 40 2 2003 293 306 38 Donnelly C.A. Ghani A.C. Leung G.M. Hedley A.J. Fraser C. Riley S. Abu-Raddad L.J. Ho L.-M. Thach T.-Q. Chau P. Chan K.-P. Lam T.-H. Tse L.-Y. Tsang T. Liu S.-H. Kong J.H. Lau E.M. Ferguson N.M. Anderson R.M. Epidemiological determinants of spread of causal agent of severe acute respiratory syndrome in Hong Kong Lancet 361 9371 2003 1761 1766 12781533 39 Riley S. Fraser C. Donnelly C.A. Ghani A.C. Abu-Raddad L.J. Hedley A.J. Leung G.M. Ho L.-M. Lam T.-H. Thach T.Q. Chau P. Chan K.-P. Lo S.-V. Leung P.-Y. Tsang T. Ho W. Lee K.-H. Lau E.M.C. Ferguson N.M. Anderson R.M. Transmission dynamics of the etiological agent of SARS in Hong Kong: Impact of public health interventions Science 300 5627 2003 1961 1966 12766206 40 Max Roser H.R. Ortiz-Ospina E. World population growth Our World Data 2021 https://ourworldindata.org/world-population-growth (accessed: August 4th 2021) 41 World development indicators database 2021 The World Bank (n.d.). Retrieved from https://datacatalog.worldbank.org/dataset/world-development-indicators (accessed: May 17th 2021) 42 Covid-19 coronavirus pandemic 2021 (n.d.). Retrieved from https://www.worldometers.info/coronavirus/country/chile/ (accessed: August 17th 2021) 43 Mathieu E. Ritchie H. Ortiz-Ospina E. Roser M. Hasell J. Appel C. Giattino C. Rodés-Guirao L. A global database of COVID-19 vaccinations Nat Hum Behav 2021 1 7 33473201 44 Roser M. Ortiz-Ospina E. Ritchie H. Life expectancy Our World in Data 2013 https://ourworldindata.org/life-expectancy (accessed June 10, 2022) 45 Read J.M. Bridgen J.R.E. Cummings D.A.T. Ho A. Jewell C.P. Novel coronavirus 2019-nCoV (COVID-19): early estimation of epidemiological parameters and epidemic size estimates Philos Trans R Soc B 376 1829 2021 20200265 46 Kobayashi T. Jung S.-m. Linton N.M. Kinoshita R. Hayashi K. Miyama T. Anzai A. Yang Y. Yuan B. Akhmetzhanov A.R. Suzuki A. Nishiura H. Communicating the risk of death from novel coronavirus disease (COVID-19) J Clin Med 9 2 2020 47 Helfand M. Fiordalisi C. Wiedrick J. Ramsey K.L. Armstrong C. Gean E. Winchell K. Arkhipova-Jenkins I. Risk for reinfection after SARS-CoV-2: A living, rapid review for American college of physicians practice points on the role of the antibody response in conferring immunity following SARS-CoV-2 infection Ann Internal Med 175 4 2022 547 555 35073157 48 Byrne A.W. McEvoy D. Collins A.B. Hunt K. Casey M. Barber A. Butler F. Griffin J. Lane E.A. McAloon C. O’Brien K. Wall P. Walsh K.A. More S.J. Inferred duration of infectious period of SARS-CoV-2: rapid scoping review and analysis of available evidence for asymptomatic and symptomatic COVID-19 cases BMJ Open 10 8 2020	36474823	PMC9715496	NO-CC CODE	2022-12-10 23:15:27	no	Chaos Solitons Fractals. 2023 Jan 2; 166:112964	utf-8	Chaos Solitons Fractals	2,022	10.1016/j.chaos.2022.112964	oa_other
==== Front J Plast Reconstr Aesthet Surg J Plast Reconstr Aesthet Surg Journal of Plastic, Reconstructive & Aesthetic Surgery 1748-6815 1878-0539 Published by Elsevier Ltd on behalf of British Association of Plastic, Reconstructive and Aesthetic Surgeons. S1748-6815(22)00688-X 10.1016/j.bjps.2022.11.060 Correspondence and Communications Surgical Treatment of Sacral Pressure Wounds in Patients with COVID-19: a Case Series Ferreira Joao 1 Nicolas Gregory 1⁎ Valente Daniel 1 Milcheski Dimas 1 Saliba Marita 2 Gemperli Rolf 1 1 Department of Plastic & Reconstructive Surgery, Hospital Das Clinicas of the University of Sao Paulo 2 Faculty of Medicine, University of Balamand, Lebanon ⁎ Corresponding author: Gregory Nicolas. Department of Plastic & Reconstructive Surgery, Hospital Das Clinicas of the University of Sao Paulo, São Paulo, Brazil 2 12 2022 2 12 2022 15 4 2021 29 11 2022 © 2022 Published by Elsevier Ltd on behalf of British Association of Plastic, Reconstructive and Aesthetic Surgeons. 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 The COVID 19 pandemic has resulted in an increased number of patients requiring intubation and intensive care. This has led to an increased incidence of sacral pressure ulcers requiring surgical management. We report our experience of COVID 19 related sacral pressure ulcers requiring surgical reconstruction. Methods A case series study was performed with 12 patients who presented grade IV sacral pressure ulcers after hospitalization for COVID-19 in a single institution. The mean age was 49.8 years and the most frequent comorbidities were arterial hypertension, diabetes and obesity, each present in 6 patients. All of them were submitted to surgical reconstruction with fasciocutaneous flaps after improvement of their clinical status. Follow up time was of at least 30 days after reconstruction. Preoperative laboratory tests and surgical outcomes were compared to data available in the literature. Results No major dehiscence was observed and minor dehiscence happened in 2 cases (16.7%). Out of the 12 patients, 8 (66.7%) had hemoglobin levels less than 10.0 and 5 (41.7%) had albumin levels less than 3.0, though this did not lead to a higher rate of complications. Conclusion This study showed that ambulating patients with grade IV pressure ulcer after COVID- 19 infection may undergo debridement, negative-pressure wound therapy and closure with local flaps with adequate results and minimal complication rate. Keywords COVID-19 Negative-pressure wound therapy flaps sacral pressure lesions ==== Body pmcIntroduction As of September 2020, Brazil had the third highest number of COVID-19 cases in the world, with more than 4 million confirmed cases. As a reference in the treatment of the disease in the city of São Paulo, and because it has about 7% of the total ICU beds in the municipality (1), the Hospital of Clinics, Faculty of Medicine, University of São Paulo (HC-FMUSP) complex concentrates the most severe cases and consequently has a higher proportion of intubated patients in their ICUs. A rapid increase of sedated and intubated patients happened during the COVID-19 pandemic. This led to a higher incidence of hospital acquired pressure sores in intensive care units. In the largest center of COVID-19 treatment in Brazil, an immediate response of the Plastic Surgery department was necessary for the surgical treatment of grade IV sacral pressure lesions. Methods A case series study was performed with patients who presented grade IV sacral pressure lesions after hospitalization for COVID-19 in a single institution. Patients treated between March and June were recruited, and cases with grade IV sacral pressure ulcers (2) that developed during hospitalization for COVID-19 in our institution and that were treated surgically were included. Patients with positive rt-PCR for COVID-19 at the time of reconstruction, sacral ulcers unrelated to COVID-19 and those who lost follow-up (minimum of 30 days after the operation) were excluded. All of them were submitted to surgical reconstruction with fasciocutaneous flaps after improvement of their clinical status, 18 to 45 days (average of 31,1 days) after they were extubated Preoperative laboratory tests and surgical outcomes were compared to data available in the literature. Surgeries performed consisted of debridement associated with negative-pressure wound therapy followed by wound closure through fasciocutaneous flaps performed from the unilateral or bilateral gluteal region. Results No major dehiscence was observed and minor dehiscence happened in 2 cases (16.7%). Out of the 12 patients, 8 (66.7%) had hemoglobin levels less than 10.0 and 5 (41.7%) had albumin levels less than 3.0. Even so, this did not lead to a higher rate of complications. Discussion Certain populations are at an increased risk of developing pressure injuries such as patients with hip fractures (range 8.8 to 55%), spinal cord injuries (range 33 to 60%) (3) elderly patients with immobility and cachexia, as well as trauma patients in the lower limbs. Although previous studies have shown that the development of pressure injuries in the hospital environment does not result in higher mortality in ICU patients, they can indirectly contribute to mortality in certain patients (4). In a cohort of 99 patients admitted due to COVID-19, the length of hospital stay averaged 22 days in patients with moderate pulmonary conditions and 25 days in patients with severe conditions (5). The average length of stay presented in our paper was 59 days. This may imply that these were more severe cases, since the meantime of intubation was 14 days and the meantime in the ICU was 21 days. As for the best time to reconstruct sacral lesions, in this study, of the 12 patients, 8 (66.7%) had hemoglobin levels less than 10.0 and 5 (41.7%) had albumin levels less than 3.0. This did not lead to a higher rate of complications. Chronic pressure injuries can have a relatively well-defined bursa in continuity with the base of the injury and methylene blue can be used to demarcate its margins. The approach commonly performed by plastic surgeons in the surgical treatment consists of complete excision of the injury, including devitalized tissue, scar and bursa; removal of possibly exposed bone and padding of any bony prominences; filling dead space; and covering the lesion with large pedicled regional flaps. The flap design should be as large as possible, placing the suture line as far away from the direct pressure zone as possible. The design must also preserve territories of adjacent flaps and allow for new advancement or rotation in cases of complications or recurrence. Recent innovations that include NPWT combined with instillation have further increased the arsenal against difficult-to-treat wounds or high-risk complications cases. In this series of cases, it was used just the conventional NPWT. The patients in this study, who had hospitalization and prolonged immobilization due to COVID-19, did not have paraplegia. They were ambulating, with preserved sensitivity and had the possibility of frequent decubitus changes, allowing for an easier postoperative period with fewer complications. Conclusion This study showed that ambulating patients with grade IV pressure injury after COVID- 19 infection may undergo debridement, negative-pressure wound therapy and closure with local flaps with adequate results and minimal complication rate. These findings led us to conduct a prospective cohort to investigate rates of surgical complication and preoperative optimization in ambulating patients with grade IV sacral pressure injury after COVID-19 infection. Consent Written consent of the patient taken for the publication of this case report and images. Ethical approval Not required Uncited Link Figure 1, table 1Figure 1 Intraoperative flap dissection and Immediate post-op. Figure 1: Table 1   Table 1PATIENT GENDER AGE COMORBIDITIES CT SCAN SUGESTIVE OF COVID-19 POSITIVE PCR-SARS-COV 2 OROTRACHEAL INTUBATION DAYS OF INTUBATION DAYS OF ICU STAY BEFORE RECONSTRUCTION DAYS OF HOSPITALIZATION 1 M 69 HYPERTENSION, SMOKING YES NO YES 20 23 57 2 M 41 HIPERTENSION, DIABETES, OBESITY, SMOKING YES YES YES 14 14 57 3 M 56 HYPERTENSION, SMOKING YES NO YES 11 34 64 4 M 50 NONE YES YES YES 16 21 63 5 M 43 NONE YES NO YES 24 28 73 6 M 82 HYPERTENSION, DIABETES YES YES NO 0 10 54 7 M 29 NONE YES NO YES 19 29 55 8 M 28 NONE YES YES YES 17 20 45 9 M 47 HYPERTENSION, DIABETES, OBESITY YES YES YES 20 34 84 10 M 50 DIABETES, OBESITY YES YES YES 7 9 46 11 M 48 HYPERTENSION, DIABETES, OBESITY YES YES YES 19 32 52 12 F 54 OBESITY, SMOKING YES YES YES 9 11 47 PATIENT DEBRIDEMENTS BEFORE RECONSTRUCTION DAYS OF NEGATIVE PRESSURE THERAPY TYPE OF FASCIOCUTANEOUS FLAP RECONSTRUCTION COMPLICATION COMPLICATION: DAYS AFTER SURGERY NEED FOR REOPERATION DAYS OF PLASTIC SURGERY FOLLOW UP 1 2 7 UNILATERAL NONE NO 27 2 2 8 ROTATION + CONTRALATERAL ADVANCEMENT NONE NO 23 3 1 7 BILATERAL MINOR DEHISCENCE 7 NO 21 4 1 5 ROTATION + CONTRALATERAL ADVANCEMENT NONE NO 18 5 4 15 UNILATERAL NONE NO 25 6 1 7 UNILATERAL NONE NO 19 7 0 0 UNILATERAL NONE NO 22 8 1 4 UNILATERAL NONE NO 23 9 3 10 UNILATERAL NONE NO 53 10 0 0 UNILATERAL NONE NO 21 11 1 11 BILATERAL MINOR DEHISCENCE 6 NO 41 12 1 8 BILATERAL NONE NO 41 Declaration of Competing Interest None declared Funding None ==== Refs References 1 Paulo. G, do E de S. Boletim SP Contra o Novo Coronavírus. Gov do Estado São Paulo [Internet]. 2020; Available from: https://www.seade.gov.br/coronavirus/ 2 Edsberg L.E. Black J.M. Goldberg M. McNichol L. Moore L. Sieggreen M. Revised National Pressure Ulcer Advisory Panel Pressure Injury Staging System: Revised Pressure Injury Staging System J Wound Ostomy Continence Nurs 43 6 2016 585 597 10.1097/won.0000000000000281 27749790 3 Lindholm C, Sterner E, Romanelli M, Pina E, Torra Y Bou J, Hietanen H, et al. Hip fracture and pressure ulcers - The Pan-European Pressure Ulcer Study - Intrinsic and extrinsic risk factors. In: International Wound Journal [Internet]. Blackwell Publishing Ltd; 2008 [cited 2020 Jun 23]. p. 315 –28. Available from: https://scinapse.io/papers/2131934040 4 Moore Z. US Medicare data show incidence of hospital-acquired pressure ulcers is 4.5%, and they are associated with longer hospital stay and higher risk of death Evid Based Nurs 16 4 2013 118 –9 23321277 5 Liu X, Zhou H, Zhou Y, Wu X, Zhao Y, Lu Y, et al. Risk factors associated with disease severity and length of hospital stay in COVID-19 patients [Internet]. Vol. 81, Journal of Infection. W.B. Saunders Ltd; 2020 [cited 2020 Jul 1]. p. e95 –7. Available from: https://linkinghub.elsevier.com/retrieve/pii/S0163445320302164	0	PMC9715574	NO-CC CODE	2022-12-03 23:20:18	no	J Plast Reconstr Aesthet Surg. 2022 Dec 2; doi: 10.1016/j.bjps.2022.11.060	utf-8	J Plast Reconstr Aesthet Surg	2,022	10.1016/j.bjps.2022.11.060	oa_other
==== Front J Ethnopharmacol J Ethnopharmacol Journal of Ethnopharmacology 0378-8741 1872-7573 Published by Elsevier B.V. S0378-8741(21)00930-2 10.1016/j.jep.2021.114701 114701 Article Xuanfei Baidu Decoction protects against macrophages induced inflammation and pulmonary fibrosis via inhibiting IL-6/STAT3 signaling pathway Wang Yuying ab1 Sang Xiaoqing ab1 Shao Rui ab1 Qin Honglin ab Chen Xuanhao abc Xue Zhifeng ab Li Lin abd Wang Yu e Zhu Yan ab Chang Yanxu abc Gao Xiumei abd Zhang Boli abd Zhang Han abd∗∗ Yang Jian ab∗ a State Key Laboratory of Component-based Chinese Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin, 301617, China b Institute of Traditional Chinese Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin, 301617, China c Tianjin Key Laboratory of Phytochemistry and Pharmaceutical Analysis, Tianjin University of Traditional Chinese Medicine, Tianjin, 301617, China d Key Laboratory of Pharmacology of Traditional Chinese Medical Formulae, Tianjin University of Traditional Chinese Medicine, Ministry of Education, Tianjin, 301617, China e School of Integrative Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin, 301617, China ∗ Corresponding author.Tianjin State Key Laboratory of component-based Chinese Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin, 301617, China ∗∗ Corresponding author. Tianjin State Key Laboratory of component-based Chinese Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin, 301617, China. 1 First Author at: Yuying Wang, Xiaoqing Sang and Rui Shao, Tianjin State Key Laboratory of component-based Chinese Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin, 301,617, China. 1 10 2021 30 1 2022 1 10 2021 283 114701114701 19 5 2021 25 9 2021 28 9 2021 © 2021 Published by Elsevier B.V. 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. Ethnopharmacological relevance Xuanfei Baidu Decoction (XFBD), one of the “three medicines and three prescriptions” for the clinically effective treatment of COVID-19 in China, plays an important role in the treatment of mild and/or common patients with dampness-toxin obstructing lung syndrome. Aim of the study The present work aims to elucidate the protective effects and the possible mechanism of XFBD against the acute inflammation and pulmonary fibrosis. Methods We use TGF-β1 induced fibroblast activation model and LPS/IL-4 induced macrophage inflammation model as in vitro cell models. The mice model of lung fibrosis was induced by BLM via endotracheal drip, and then XFBD (4.6 g/kg, 9.2 g/kg) were administered orally respectively. The efficacy and molecular mechanisms in the presence or absence of XFBD were investigated. Results The results proved that XFBD can effectively inhibit fibroblast collagen deposition, down-regulate the level of α-SMA and inhibit the migration of fibroblasts. IL-4 induced macrophage polarization was also inhibited and the secretions of the inflammatory factors including IL6, iNOS were down-regulated. In vivo experiments, the results proved that XFBD improved the weight loss and survival rate of the mice. The XFBD high-dose administration group had a significant effect in inhibiting collagen deposition and the expression of α-SMA in the lungs of mice. XFBD can reduce bleomycin-induced pulmonary fibrosis by inhibiting IL-6/STAT3 activation and related macrophage infiltration. Conclusions Xuanfei Baidu Decoction protects against macrophages induced inflammation and pulmonary fibrosis via inhibiting IL-6/STAT3 signaling pathway. Graphical abstract Image 1 Keywords COVID-19 Xuanfei baidu decoction Pulmonary fibrosis Macrophage polarization IL-6/STAT3 Abbreviations IPF, Idiopathic pulmonary fibrosis COVID-19, coronavirus disease 2019 XFBD, Xuanfei Baidu Decoction LPS, lipopolysaccharide BLM, bleomycin TNF-α, tumor necrosis factor-alpha IL-6, Interleukin 6 IL-1β, Interleukin 1β IL-10, Interleukin 10 iNOS, inducible nitric oxide synthase TGF-β, transforming growth factor-β TCM, Traditional Chinese Medicine Arg-1, Arginase 1 μCT, Micro CT α-SMA, α-smooth muscle actin Dex, Dexamethasone FBS, Fetal bovine serum F4/80, Mouse EGF-like module-containing mucin-like hormone receptor-like 1 STAT3, Signal transducer and activator of transcription 3 EMT, Epithelial-mesenchymal transition ==== Body pmc1 Introduction In December 2019, there was an outbreak of new coronavirus pneumonia (Coronavirus disease 2019, COVID-19) caused by a new type of coronavirus (SARS-CoV-2). There is no clinically targeted specific medicine. In addition to direct viral damage, hyperinflammatory responses also known as cytokine storms is a major reason of disease severity and death. Pro-inflammatory response can solve the virus infection in most cases, however, if inflammation persists and immune cells dysfunction, this recovery response cannot be completed, which may lead to further accumulation of immune cells in the lungs, forming an inflammatory storm and damaging the lung infrastructure (Tay et al., 2020). There are numerous medications in clinical trials or practice against COVID-19, such as a monoclonal neutralizing antibody binding IL-6 receptors tocilizumab (Xu et al., 2020), an anti-Ebola drug remdesivir (Beigel et al., 2020) and so on. Among these drugs, only dexamethasone has a confirmed clinical benefit reducing 28-day mortality rate by 17% compared to that in the remdesivir-treated group (Ledford, 2020). In China, Traditional Chinese Medicine (TCM) has been proven effective for COVID-19 treatment (Liu et al., 2020a). According to the report of State Council of the People's Republic of China (2020), more than 74,000 patients were treated with TCM, and clinical observation shows that the overall effective rate of TCM, (2020) reached above 90%. The National Administration of TCM recommended “three medicines and three prescriptions” for clinically treatment of COVID-19. “Three medicines” are Jinhua Qinggan Granule, Lianhua Qingwen Capsule and Xuebijing Injection, “Three prescriptions” refer to Qingfei Paidu Decoction, HuaShi BaiDu Formula, and XuanFei BaiDu Granule (Huang et al., 2020). An empirical study from Wuhan showed that Qingfei Baidu decoction contributed to the recovery of various disease progresses in COVID-19 patients (Luo et al., 2020). Furthermore, XFBD can significantly improve the clinical symptoms of COVID-19 patients, decrease the number of white blood cells and lymphocytes, and play an anti-inflammatory effect by significantly reducing C-reactive protein and erythrocyte sedimentation rate (Xiong et al., 2020). Furthermore, the 7th edition of the “Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia of P.R. China” mentioned that Xuanfei Baidu Decoction (XFBD) is one of the three clinically used TCM remedies for treating common COVID-19 patients, it showed very promising clinic outcomes. XFBD integrates 13 classical herbs including Ephedrae Herba from Ephedra sinica Stapf, Ephedra intermedia Schrenk & C.A.Mey. and Ephedra equisetina Bunge, Armeniacae Semen Amarum from Prunus sibirica L., Prunus armeniaca L. and Prunus mandshurica (Maxim.) Koehne, Coicis Semen from Coix lacryma-jobi L.var.ma-yuen (Roman.) Stapf, Polygoni Cuspidati Rhizoma et Radix from Polygonum cuspidatum Siebold & Zucc., Gypsum Fibrosum, Atractylodis Rhizoma from Atractylodes lancea (Thunb.) DC., Pogostemonis Herba from Pogostemon cablin (Blanco) Benth., Artemisia Annua Herba from Artemisia annua L., Verbenae Herba from Verbena officinalis L., Phragmitis Rhizoma from Phragmites australis subsp. australis, Descurainiae Semen Lepidii Semen from Descurainia sophia（L.）Webb. ex Prantl. and Lepidium apetalum Willd., Citri Grandis Exocarpium from Citrus maxima (Burm.) Merr., Glycyrrhizae Radix et Rhizoma from Glycyrrhiza uralensis Fisch. ex DC., Glycyrrhiza inflata Batalin., Glycyrrhiza glabra L. and Gypsum Fibrosum. It is reported that multiple herbs of XFBD and their main components have an effect in balancing immune inflammatory response, resisting viral infection and viral protein transcription, and restoring the balance of liver and gallbladder metabolism and energy metabolism in the body by regulating the biological processes of viral infection, immune inflammation, liver and gallbladder metabolism and energy metabolism (Wang et al., 2020). However, the underlying mechanism regulating immune responses by XFBD still remains unknown. Based on the published data, COVID-19 patients can lead to pulmonary fibrosis, and the prognosis of this serious complication is also worthy of our attention (Zhang et al., 2021). The morphological features of pulmonary fibrosis are thickened alveolar septum, collagen deposition and fibroblast proliferation, and diffuse inflammation (Chanda et al., 2019; Craig et al., 2015). It is generally believed that the development of fibrotic diseases includes two stages: inflammatory response and fibrosis stage (Li et al., 2016; Martinez et al., 2017). In the early stage of lung injury, the body undergoes inflammatory response. Inflammatory cells such as neutrophils and macrophages infiltrate and secrete large number of inflammatory factors to promote the repair of lung injury, such as: tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), interleukin-1β (IL-1β), etc.(Heukels et al., 2019). In the fibrosis stage, there will be excessive proliferation of fibroblasts and myofibroblasts and excessive accumulation of extracellular matrix (ECM), which will further lead to impaired lung function. Accordingly, the control of inflammation in the early stage of pulmonary fibrosis can effectively slow down the occurrence and development of pulmonary fibrosis. At present, there have been a lot of studies on molecular signal transduction pathways that regulate the different stages of pulmonary fibrosis development. Multi-signaling pathways such as transforming growth factor-β (TGF-β)/PI3K/AKT, Wnt-β-catenin, hedgehog and Notch pathways have a relation with lung development (Chanda et al., 2019). For example, the activation of the PI3K/AKT signaling pathway induced by TGF-β1 can lead to EMT, fibroblast proliferation and collagen accumulation, which is a key step in the development of fibrosis (Liu et al., 2016a). Wnt-β-catenin signal induces the activation of lung fibroblasts into a fibrotic phenotype, promotes the proliferation of fibroblasts, and the differentiation of myofibroblasts (King et al., 2011). IL-6 mediates many complications in the lungs, and its release imbalance is related to the pathogenesis of various respiratory diseases. IL-6 promotes the phosphorylation and translocation of STAT3 (Pulivendala et al., 2020). STAT3 is associated with activation, proliferation of fibroblasts and ECM deposition, which in turn contributes to fibrotic disease progression. In this study, we focuse on exploring the potential anti-fibrosis role of XFBD, using fibroblasts and macrophages cell models, as well as a bleomycin-induced mouse lung fibrosis model. The anti-fibrosis pharmacodynamics and mechanism of XFBD have been studied, hoping to provide more choices and basis for the applications of traditional Chinese medicine in the treatment of IPF. 2 Materials and methods 2.1 Chemicals and reagents XFBD was provided by TianJin Modern TCM Innovation Center (TRT, 200302). Reference standards including amygdalin, sinapine, verbenalin, hastatoside, liquiritin, glycyrrhizic acid, acteoside and naringin were bought from Chengdu Desite Bio-Technology Co., Ltd. (Chengdu, China), and ephedrine and polydatin were purchased from National Institutes for Food and Drug Control (Beijing, China). HPLC-grade methanol (MeOH) and acetonitrile were obtained from Fisher (Leicestershire, UK). LC-MS/HPLC-grade formic acid (FA) was obtained from Anaqua Chemicals Supply (Wilmington, USA). LPS was purchased from Sigma-Aldrich (Shanghai) trading company Ltd. Bleomycin hydrochloride was purchased from Dalian Meilun Biotech (Dalian, China). Enzyme linked immunosorbent assay (ELISA) kits of IL-6, TNF-α, IL-10 and Arg-1 were purchased from Shanghai ZCi BiO Science & Technology Co., Ltd. F4/80, and α-SMA primary antibodies were purchased from Abcam (Cambridge, MA, USA). MTT cell proliferation and cytotoxicity detection kit were purchased from Beijing Solarbio Science & Technology Co., Ltd. 2.2 Gene networks analysis Prediction of targets of XFBD was performed by BATMAN-TCM (a Bioinformatics Analysis Tool for Molecular Mechanism of Traditional Chinese Medicine) (Liu et al., 2016b). BATMAN-TCM is the first online bioinformatics analysis tool dedicated to analyzing and predicting the molecular mechanism of TCM. In the BATMAN-TCM, the compounds of the herbs in XFBD were obtained from TCMID database (http://119.3.41.228:8000/tcmid/). And all these compounds’ data was curated from literatures. The online tools such as Gene Cards (http://www.genecards.org/) were used to predict potential therapeutic targets for XFBD in the treatment of IPF (Stelzer et al., 2016). The Kyoto Encyclopedia of Genes and Genomics (KEGG) pathway analysis in the DAVID database were used for investigating gene function. We constructed a PPI (protein–protein interaction) network to clarify the molecular mechanisms of anti-inflammation effects of XFBD by using the Cytoscape software (version 3.7.2; http://www.cytoscape.org) (Kohl et al., 2011) and the STRING website (version 11.0, http://www.string-db.org/) with a required confidence >0.4 (Szklarczyk, 2017). Next, we used Cytoscape software (version 3.7.2) to analyze the degree of connectivity in the PPI network and obtained the hub genes. 2.3 UHPLC analysis The freeze-dried powder of XFBD (0.4000 g) was extracted with ultrapure water (1:25, g/mL) in an ultrasonic water bath for 30 min. The solution was diluted with 50% methanol at the ratio of 1:1 and vortex-mixed for 5 min. Then the solution was centrifuged at 14,000 rpm for 10 min before filtered with a 0.22 μm filter membrane. Aliquot (2 μL) of the supernatant solution was injected into UHPLC-PDA for analysis. A Waters Acquity UHPLC System (Waters Co., Milford, MA) equipped with a photodiode array detector (PDA) was used to separate the multiple components in Xuanfei Baidu Decoction. All separations were performed a ZORBAX RRHD Eclipse XDB-C18 column (2.1 × 100 mm, 1.8 μm, Agilent Technologies). The flow rate was 0.3 mL/min. The column temperature was 40 °C. 2.4 Cell culture Mouse macrophage cell line (RAW264.7) and mouse fibroblast cell line (NIH-3T3) were purchased from ATCC. RAW264.7 macrophages and Murine embryo fibroblast (NIH-3T3 cell line) were maintained in Dulbecco's Modified Eagle's Medium (DMEM, Gibco) supplemented with 10% heat-inactivated FBS, penicillin (100 units/mL), streptomycin (100 mg/mL), followed by incubation at 37 °C in a humidified atmosphere containing 5% CO2 and 95% air. 2.5 MTT assay The RAW264.7 cells were seeded into 96-well plates at a density of 4 × 104 cells per well at 37 °C in a humidified atmosphere containing 5% CO2 and 95% air. After inoculation, the cells are then treated with the different concentrations of XFBD (0, 10, 25, 50, 100, 200 μg/mL) for 24 h. NIH-3T3 cells were seeded into 96-well plates at a density of 4 × 103 cells per well. After 24 h, cells were incubated for 48 h in the presence of different concentrations of XFBD (0, 1, 2.5, 5, 10, 25, 50 μg/mL). Cell viability was detected using MTT (methylthiazole diphenyl-tamazole). After discarding the supernatant, adding MTT (10%) and incubated 4 h. The MTT solution was removed, then the precipitation dissolved in the DMSO solution. Measure the absorbance value at 490 nm using the enzyme-labeled instrument (Tecan, Austria). 2.6 Imunofluorescence assay The cells were treated with IL-4 (20 ng/mL), TGF-β (5 ng/mL) and XFBD for 48 h. After fixed with paraformaldehyde (4%) for 30 min, the cells were treated with BSA (2%) for 1.5 h at 37 °C and followed by incubation with the α-SMA primary antibodies (1:500, Abcam, #ab5694) overnight at 4 °C. the cells were stained with the corresponding secondary antibodies after washed 3 times in PBS. Nuclei were stained with Hoechst 33342 for 10 min. Visual images and quantitative analysis were obtained with Operetta High Content Analysis (HCA) System (PerkinElmer, Boston, MA, USA). We selected three multiple wells in each group for the experiment, each well selected three fields for quantification, and each field selected 10 cells for fluorescence intensity quantification, and finally took the average of the fluorescence intensity of each well as the fluorescence data of the entire well. 2.7 Flowcytometry assay A suspension containing 5 × 105 cells was incubated with an APC-anti-CD206 antibody (Biolegend Co., 2.5 μg/106 cells) for 30 min at 4 °C in the dark. Un-administered cells staining with APC-anti-CD206 was used as a negative control. Then, cells were washed with PBS and analyzed by Attune® NxT, Acoustic Focusing Cytometer (Invitrogen). 2.8 Enzyme linked immunosorbent assay (ELISA) The level of IL-6 and Arg-1 is determined by ELISA. Cytokine concentration is calculated using standard curve. Specific operations according to the manual provided by the manufacturer. 2.9 Collagen deposition detection by picro-sirius red (PSR) assay Total collagen content was determined by PSR assay. A total of 100 μL NIH-3T3 cells (5000 cells/well) were seeded into 96-well plates. Cells were then treated with XFBD or SB431542 (5 μmol/L) with or without TGF-β1 (5 ng/mL) for 48 h at 37 °C. The medium was removed from 96 wells and fixation was conducted by iced methanol overnight at −20 °C. After washing 3 times with 200 μL PBS each well, 100 μL Sirius red reagent (Shanghai yuanye Bio-Technology, Shanghai, China) was added each well and incubated for 4 h. Then free Sirius red was removed and cells were washed with 0.01% acetic acid for 3 times. After that, 200 μL 0.1 M NaOH was added. After 4 h, the absorbance at 540 nm was determined by multiplate reader (Tecan, Austria). 2.10 Animal experiments SPF healthy male C57BL/6 mice (22–25 g) were provided by SPF (Beijing) Biotechnology Co., LTD. The research was conducted with the Guidelines for Animal Experiments of Tianjin University of Traditional Chinese Medicine for laboratory animal use. Mice were raised in Tianjin International Biopharmaceutical Joint Research Institute, and the temperature was maintained at 20–25 °C, and the relative humidity was 40–60%. 5 mice in each cage, they were given a regular feed and free water. After 1 week of adaptive feeding, the mice were randomly divided into control group (Control), model group (BLM), XFBD low dose group (XFBD-L) and XFBD high group (XFBD-H) (n = 6). The control group was intragastrically administered with clear water after sham operation; the model group was intragastrically administered with clear water after BLM tracheal instillation; the XFBD-L group was intragastrically administered with XFBD solution (4.6 g/kg) after BLM tracheal instillation, and the XFBD-H group was intragastrically administered with XFBD solution (9.2 g/kg) after BLM tracheal instillation. The dosage of XFBD used in mice was converted from clinical dosage. Mice in the model group received bleomycin hydrochloride (1.5 mg/kg) 50 μL by intratracheal instillation, and mice in the control group received the same volume of saline. 2.11 Micro CT scanning (μCT) Mice were scanned with a μCT scanner (Micro-CT QuantumFX μCT Software, PerkinElmer). Each mouse was accepted a CT scanning and then euthanized. The mice treated by BLM were scanned on day 10 and the control group mice were scanned. The mice are positioned in the CT scanner and use the following parameters to obtain the impact of the chest for anatomical positioning and attenuation correction: 90 kV, 160 mA and 40 mm. The X-ray system uses a 5 mm micro focusing tube with a spot size and a conical beam geometry to generate X-rays. The maximum width of the image field is 68 mm and the voxel size is 35 × 35 × 35 mm. In this section, detailed methods for quantitative measurement of pulmonary fibrosis in mice were described. Mice with a fibrosis score greater than 3 were successfully modeled. 2.12 Tissue sampling and histopathological observation After anesthesia by tribromoethanol, the whole blood, alveolar lavage fluid, lung and other major organs of mice were collected. Lung tissues were washed with iced saline. Part of lung tissues were degassed with paraformaldehyde and fixed in 4% paraformaldehyde. Pathological sections and staining were performed after 72 h. The remaining lung tissue was quickly frozen in liquid nitrogen and transferred to refrigerator at −80 °C until use. The lung tissues that fixed in 4% paraformaldehyde fixing solution were dehydrated, embedded in paraffin and sliced into 5 μm sections. The sections were stained with hematoxylin and eosin (H&E) reagent and visualized under a light microscope. The whole pathological changes were observed at 100× and photographed under 200× by Leica Microsystems CMS GmbH Ernst-Leitz-Str.17–7 (Leica, Germany). A semi-quantitative histological score was used to assess the severity of IPF with double-blind method. Inflammation was scored from 0 (normal) to 5 (extremely severe damage) according to the degree of lung injury, including alveolar inflammation, congestion or bleeding of alveolar wall, proliferation of lymphocytes, emphysema, and degeneration or necrosis of bronchial epithelial cells. Extravascular area, vascular area, vascular outer perimeter, and ascular inner perimeter of arteries with a diameter < 50 μm were also measured. The following formula was used for calculation: the thickness of vascular wall (μm) = outer diameter of the pulmonary arterioles − inner diameter of the pulmonary arterioles of the pulmonary arterioles); the ratio of vascular wall area (WA%) = 100% × (transection area of the walls of pulmonary arterioles)/ (cross-sectional area of pulmonary arterioles). Masson staining kit was utilized to detect collagen deposits according to the manufacturer's instructions. 2.13 Immunohistochemistry (IHC)/immunofluorescence (IF) staining For IHC analysis, the prepared paraffin sections are deparaffinized and hydrated in organic reagents such as xylene. Sodium citrate buffer was used for antigen retrieval, and the slices were treated with 3% H2O2 in order to eliminate endogenous peroxidase activity. Incubate the sections with 10% FBS to block the binding of non-specific antibodies, after which slides were immunostained using the α-SMA primary antibody (1:100; Abcam, #ab5694), the CD206 primary antibody (1:150; Abcam, #ab64693), the IL-6 primary antibody (1:200; Abcam, #ab208113) and the STAT3 (1:150; Abcam, #ab68153) primary antibody. Incubate horseradish peroxidase (HRP) conjugated goat anti-rabbit secondary antibody, and use DAB substrate kit (Boster Biological Technology Co., Ltd; #AR1022) for color development of positive results. Then the slides were counter-stained with Mayer's hematoxylin staining solution and differentiated with 1% hydrochloric acid and ethanol. Finally, the slides are dehydrated and sealed. Image J software was utilized to analyze the photos. The intensity of each stained slide contains at least the average of five non-overlapping fields. For IF analyses, there was no need to quench endogenous peroxidase activity. After incubating slides with 10% FBS, slides were immunostained using the F4/80 primary antibody (1:200; Abcam, #ab6640). Alexa Fluor®488-conjugated goat anti-Rat secondary antibody was utilized for detected and visualized of antibody-antigen complexes. (1:200; Abcam, #150157). The nuclei were stained using Hoechst 33,342, after that slides were sealed. 2.14 Real-time quantitative PCR (RT-qPCR) analysis After 10 days of continuous administration, the lung tissues of each group of mice were collected, quick-frozen in liquid nitrogen and stored at −80 °C. According to the manufacturer's instructions, TRIzol® reagent (Invitrogen, Waltham, MA, USA) was utilized to extract total RNA samples from lung tissue. The RNA sample was reverse transcribed into complementary DNA (cDNA) use the Transcriptor First Strand cDNA Synthesis Kit (Roche, Germany). Using Bestar® SybrGreen qPCR mastermix (DBI®Bioscience, Germany) for RT-qPCR. The mRNA expression of IL-6 and STAT3 was verified by reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR). Gyceraldehyde 3-phosphate dehydrogenase (GAPDH) was utilized as an endogenous reference gene to quantify the expression levels of key genes. Gene-specific primers were synthesized by Sangon Biotech (Shanghai, China). Using LightCycler®480 Software Version 1.5.0.39 (Roche, Germany) for 45 cycles amplification and analysis. After normalization with the GAPDH gene, the relative mRNA expression level of the target gene was calculated using the 2−ΔΔCT method. The sequences of primers for qPCR are shown in Table 1 .Table 1 Primer sequences for RT-qPCR. Table 1Primer name Forward primer (5′–3′) Reverse primer (5′–3′) qMouse STAT3 AATCTCAACTTCAGACCCGCCAAC GCTCCACGATCCTCTCCTCCAG qMouse Il6 TAGTCCTTCCTACCCCAATTTCC TTGGTCCTTAGCCACTCCTTC qMouse Gapdh TGGTGAAGCAGGCATCTGAG TGCTGTTGAAGTCGCAGGAG 2.15 Statistical analysis All experiments were performed in triplicate. GraphPad (GraphPad Prism 5, San Diego, California, USA) was used for data statistics and analysis. All data were presented as the mean ± SD. Statistical analysis was performed by one-way analysis of variance (ANOVA) test with post hoc Tukey's test. A value of P < 0.05 was considered statistically significant. 3 Results 3.1 Identification and enrichment analysis of candidate targets for XFBD against idiopathic pulmonary fibrosis We conducted a virtual study to explored the underlying mechanism both involved in the anti-fibrosis activity of XFBD by using the online BATMAN-TCM server and Gene Cards databases. XFBD is composed of 13 traditional Chinese medicines, among these medicines and related components, 1939 potential targets of 188 compounds were predicted using BATMAN-TCM server and a total of 3054 idiopathic pulmonary fibrosis (IPF) targets were retrieved from the Gene Cards databases. Venn Diagram (http://bioinformatics.psb.ugent.be/webtools/Venn/) was used to find genes that were both the potential therapeutic targets of XFBD and the IPF-related targets. Total of 763 potential therapeutic targets of XFBD on IPF were obtained. The biological classification of the 763 potential targets of XFBD in the treatment of IPF was analyzed by using the functional enrichment analysis of the DAVID website. Subsequently, KEGG analysis confirmed that 763 genes were enriched in 167 pathways. Here we used FDR corrected p < 0.05 as the enrichment screening criterion and obtained the top 15 enriched functional clusters (Fig. 1 A). The results showed that the 763 genes mainly enriched in tumor, virus infection and immune pathways. Furthermore, gene ontology (GO) analysis of biological processes showed significant enrichment of biological process related to inflammation and 82 genes were involved in the biological process of inflammatory response (Fig. 1B). It was suggested that XFBD might alleviate IPF by regulating inflammatory response. Since the protein-protein interaction (PPI) networks are relevant to visualize the role of various key proteins in disease, a visual PPI network of the 82 genes was subsequently constructed using the Cytoscape software (Fig. 1C). And the 10 genes with the highest degree of nodes were identified as hub genes, including IL6, TNF, IL1B, TLR4, IL10, AKT1, PTGS2, CXCR4, CCL5 and CCR7 (Fig. 1D and E). The gene with highest node degrees was IL6, which was 65.Fig. 1 Network Pharmacology analysis on the potential mechanisms of XFBD against Idiopathic Pulmonary fibrosis (IPF). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis (A) and gene ontology (GO) analysis of biological processes (B) were investigated using the DAVID website. PPI interactions and hub genes of XFBD targets related to inflammatory response were analyzed by Cytoscape software (the larger the node, the higher the degree) (C–E). Fig. 1 3.2 Identification of active ingredients in XFBD The contents of the ten ingredients were analyzed by using the UHPLC-PDA method. There are shown representative chromatograms of the mixed ten standards and XFBD extracted solution (Fig. 2 ). The contents of the investigated analyses are as follows: 1.19 mg/g for ephedrine, 4.97 mg/g for amygdalin, 3.63 mg/g for sinapine, 5.04 mg/g for hastatoside, 4.49 mg/g for verbenalin, 8.45 mg/g for polydatin, 3.40 mg/g for liquiritin, 3.46 mg/g for acteoside, 54.91 mg/g for naringin and 4.80 mg/g for glycyrrhizic acid in the freeze-dried powder of XFBD (Table 2 ).Fig. 2 Identification of active ingredients in Xuanfei Baidu Decoction. The typical chromatograms of standard compounds of 210 nm (A), 254 nm (B) and sample of 210 nm (C), 254 nm (D). (1) ephedrine, (2) amygdalin, (3) sinapine, (4) hastatoside, (5) verbenalin, (6) polydatin, (7) liquiritin, (8) acteoside, (9) naringin and (10) glycyrrhizic acid. Fig. 2 Table 2 Linear regression and contents of 10 compounds (n = 6). Table 2Peak No. Rt (min) Compounds Regression equation R2 Linearity range (μg/mL) Content (mg/g) RSD (%) 1 3.3 ephedrine y = 131 94x+37.738 0.999 7 2.00–50.0 1.19 1.64 2 7.1 amygdalin y = 669 3.6x-50.185 0.999 2 2.50–100 4.97 0.70 3 8.4 sinapine y = 133 0x+107.5 0.999 8 1.25–50.0 3.63 3.09 4 8.7 hastatoside y = 259 6.2x+94.754 0.999 9 1.00–100 5.04 3.47 5 9.8 verbenalin y = 554 2.7x+5.5247 0.999 9 1.00–100 4.49 3.89 6 13.7 polydatin y = 362 5.7x+1160 0.999 8 1.50–150 8.45 3.43 7 14.4 liquiritin y = 444 1.4x+119.17 0.999 8 0.50–50.0 3.40 3.15 8 17.1 acteoside y = 381 9.9x+317.12 0.999 9 0.50–50.0 3.46 2.13 9 18.8 naringin y = 233 5.8x+4000.1 0.999 9 10.0–1000 54.91 1.15 10 28.8 glycyrrhizic acid y = 444 2x+1227.1 0.999 9 1.00–100 4.80 1.42 3.3 XFBD inhibits the activation and migration of fibroblasts Fibroblasts can be induced into myofibroblasts by transforming growth factor-β1 (TGF-β1), which will cause excessive collagen production and overexpression of α-Smooth muscle actin (α-SMA) (Liu et al., 2019). Therefore, inhibiting its activation is one of the targets of pulmonary fibrosis treatment (Liu et al., 2019; Lu et al., 2018). First, mouse embryonic fibroblasts NIH-3T3 cells were treated with a series of concentrations (1, 2.5, 5, 10, 25, 50 μg/mL) of XFBD, and the results proved that cells were well tolerance with 50 μg/mL XFBD (Fig. S1). In the TGF-β1 induced fibroblast activation experiment, collagen deposition results showed that the amount of collagen deposition induced by TGF-β1 was increased around 1.7 times (1 ± 0.31 vs 1.67 ± 0.06, p < 0.001)，compared with the blank control group, TGF-β1 inhibitor SB431542 can effectively reduce collagen deposition to normal level (1.67 ± 0.06 vs 1.07 ± 0.24, p < 0.001)，XFBD can effectively inhibit collagen deposition of fibroblasts at a concentration of 10 μg/mL (1.67 ± 0.06 vs 1.39 ± 0.03, p < 0.001) (Fig. 3 A and B). Immunofluorescence analysis results further suggested that XFBD can inhibit the expression of a-SMA, which is the biomarker of the fibroblasts activation at the concentration of 25 μg/mL (70.31 ± 4.09 vs 58.07 ± 3.36, p < 0.01) (Fig. 3C and D). Similarly, the migration assay showed that 10 μg/mL drug already has a good inhibitory effect on activated fibroblasts (12.05 ± 1.05, p < 0.01), and the inhibitory effect is more significant at 25 μg/mL (9.99 ± 1.26, p < 0.01) (Fig. 3E and F).Fig. 3 XFBD inhibits the activation and migration of fibroblasts. (A, B) The inhibitory effect of XFBD on TGF-β1-induced fibroblast activation (magnification × 200). (C, D) The effect of XFBD on TGF-β1-induced α-SMA expression was evaluated by immunofluorescence analysis. (E, F) Inhibition of XFBD on the migration of activated fibroblasts. #p < 0.05, ##p < 0.01, ###p < 0.001, vs. control group, p < 0.05, p < 0.01, *p < 0.001, vs. Model group. Scale bar = 100 μm. Fig. 3 3.4 XFBD inhibits macrophages produced inflammation Macrophages cause the secretion of inflammatory factors and pro-fibrosis factors, which accelerate the development of fibrosis (Li et al., 2019; Mari and Crestani, 2019). Particularly, phenotypic changes of macrophages can affect the process of lung fibrosis (Chandrasekaran et al., 2019; Dong and Ma, 2018). We furtherly investigated its effects on the polarized macrophages and the expression of CD206+ of macrophages after treating XFBD, which is the mannose receptor, a high specificity M2 macrophage marker. As shown in Fig. 4 A and B, XFBD at a concentration of 25 μg/mL could effectively down-regulate the fluorescence expression of CD206+ of M2 polarized macrophages (28.39 ± 14.32, p < 0.001). Flow cytometry analysis also demonstrated that XFBD can reduce the proportion of CD206 positive cells in RAW264.7 cells ( Fig. 4 C & Fig. S2 ). Fig. 4 XFBD inhibits the inflammatory response of macrophages (A, B) XFBD inhibits IL-4 induced polarization of M2 macrophages in vitro. (C) Flowcytometry result of the proportion of CD206 positive cells (D) The inhibitory effect of XFBD on LPS-induced IL-6 expression. (E) The inhibitory effect of XFBD on LPS-induced iNOS expression. #p < 0.05, ##p < 0.01, ###p < 0.001, vs. control group, p < 0.05, p < 0.01, p < 0.001, vs. Model group. Scale bar = 100 μm. Fig. 4 To investigate the mechanism of XFBD against the development of lung injury, murine macrophage RAW264.7 were treated with LPS (100 ng/mL) and different concentrations of XFBD. Then the expression of proinflammatory factor IL-6 was tested. The ELISA results showed that XFBD could effectively inhibit the upregulation of IL-6, when compared with the model group which treated with LPS (11.06 ± 0.36 vs 9.69 ± 0.22, p < 0.01) (Fig. 4D). Similarly, the activity of iNOS was also tested after LPS inducement. When incubated with XFBD at a dose of 5 μg/mL, iNOS activity can be inhibited effective (3.80 ± 0.11 vs 3.36 ± 0.22, p < 0.05) (Fig. 4E). Altogether, these results suggest that XFBD could inhibit the proinflammatory cytokines produced by macrophages. 3.5 XFBD has protective effect against bleomycin-induced pulmonary fibrosis in mice The bleomycin model of in vivo lung injure is a well-described model, which can reflect the pathological characteristics of human pulmonary fibrosis (Della Latta et al., 2015; Tashiro et al., 2017). Here, we established a bleomycin-induced pulmonary fibrosis model in mice, mimicking the pre-inflammatory stage for the purpose of suppressing the occurrence of inflammation. On the 10th day after BLM tracheal instillation, the mice were killed by overdose anesthesia (Fig. 5 A). In the BLM-induced pulmonary fibrosis model, XFBD at high dosage (XFBD-H group) improved weight loss and survival rate of pulmonary fibrosis mice, compared with the model and XFBD-L administration groups (Fig. 5B and C). Additionally, μCT scanning results showed that patch-shaped ground-glass shadows appeared in the middle and lower lungs of the model group, presenting extensive, double-lung, symmetrical gridded changes, and honeycomb changes at the 10th day, while XFBD-H can significantly inhibit the occurrence and development of pulmonary fibrosis (Fig. 5D). A four-point ranking scale was used to semi-quantitatively evaluate the quality of μCT images, and the scoring data further confirmed the effectiveness of XFBD (Fig. 5E).Fig. 5 XFBD has protective effect on mice treated with bleomycin. (A) Timeline of XFBD administration. (B) Body weight changes of mice in each group. (C) Survival rates of mice after different treatment. (D) Scanning of pulmonary fibrosis after BLM treatment at 10th day. (E) μCT score of pulmonary fibrosis. p < 0.05, p < 0.01, p < 0.001, vs. Model group. Scale bar = 100 μm. Fig. 5 3.6 XFBD ameliorated the pathologenesis of pulmonary fibrosis After the administration of XFBD in the early inflammatory stage, the occurrence and development of pulmonary fibrosis was significantly inhibited, and the alveolar cavity of the lung tissue of the normal control group mice was clearly visible, with few inflammatory cell infiltrating. H&E staining of lung tissue in the model group showed that alveolar structure was disordered and collapsed, with prominent inflammatory cell infiltration. The damage degree of lung tissue in mice treated with XFBD-H (2.17 ± 0.75, p < 0.001) was significantly reduced compared with that in the model group (3.67 ± 0.52, p < 0.001) (Fig. 6 A&6 E). We also found that the occurrence of pulmonary fibrosis will result in thickening of the pulmonary blood vessel wall. The pulmonary vascular thickness in the model group (11.6 ± 0.78, p < 0.001) increased by 3 times compared with the control group, and XFBD-H (6.84 ± 1.41, p < 0.001) has a significant inhibitory effect on the increase of pulmonary vascular thickness (Fig. 6B and F). To test the effect of XFBD on collagen deposition, Masson's trichrome staining was carried out. The data demonstrated that XFBD at low or high dosage (XFBD-L and XFBD-H) significant inhibited the deposition of collagen in the IPF mice. Compared with the model group, the XFBD-H (3.26 ± 0.68 vs 2.67 ± 0.94, p < 0.01) group has a significant improvement effect (Fig. 6C and G). Through immunohistochemical staining of fibroblast activation marker α-SMA, we found that compared with the control group, the BLM-induced model group showed a large amount of positive expression of α-SMA in the lungs, and the XFBD administration group at different doses had a significant effect on collagen deposition. The results showed that XFBD-H (20.75 ± 3.02 vs 12.28 ± 3.06, p < 0.001) can significantly down-regulate the level of α-SMA compared with the model group (Fig. 6D and H). In addition, we tested the expression of arginine-1 (Arg-1) in mouse serum and lung tissue homogenate (Fig. 6I). Compared with the control group, the expression level of Arg-1 in the lung tissue of the model group increased by about 4 times (7.86 ± 0.41 vs 1.93 ± 1.11, p < 0.001), after treatment with XFBD-H, the level of Arg-1 was significantly decreased (7.86 ± 0.41 vs 3.13 ± 1.36, p < 0.001), which further presented the anti-fibrosis effect of XFBD.Fig. 6 XFBD inhibits pathological changes of pulmonary fibrosis. (A) The histopathological changes of lung tissues were examined using H&E staining (magnification × 200) and (B) Alveolar structural changes of pulmonary vascular thickness and the value changes (magnification × 400). (C) Masson staining to assess the deposition of collagen in the lungs (Magnification 200 × ). The detect (D) and quantify (H) of α-SMA by Immunohistochemical staining. (E) The morphological damage score for the lung tissues. (F) Improvement of XFBD on vascular remodeling in pulmonary fibrosis mice. (G) The collagen deposition. (I) The level of Arg-1 in lung tissue. #p < 0.05, ##p < 0.01, ###p < 0.001, vs. control group, p < 0.05, p < 0.01, p < 0.001, vs. Model group. Scale bar = 100 μm. Fig. 6 3.7 XFBD inhibits pulmonary fibrosis by down-regulating M2 polarization and IL-6/STAT3 pathway On the 10th day after BLM tracheal instillation, the mice were killed by overdose anesthesia, the left lung in the mice was prepared for further investigation. Observed by immunofluorescence staining of left lung sections of pulmonary fibrosis mice consults showed that the XFBD high-dose group reduced the infiltration of total macrophages and M2 macrophages in the lungs (Fig. 7A, B & 7 E). Here, the total macrophages were labeled with F4/80 (9.24 ± 1.24 vs 4.16 ± 1.89, p < 0.05), CD206+ was used to mark the M2 macrophages (10.56 ± 3.81 vs 3.5 ± 1.01, p < 0.05). Immunohistochemical staining experiments were performed to observe the expression of IL-6 and STAT3 protein in the lungs of mice. Compared with the control group, the positive expression of IL-6 and STAT3 in the lung tissue of the model group induced by BLM was significantly up-regulated (Fig. 7C and D). Quantitative results showed that XFBD-H (7.83 ± 2.48 vs 4.45 ± 1.57, p < 0.01) group inhibited IL-6 upregulation better than XFBD-L (7.83 ± 2.48 vs 5.82 ± 2.6, p < 0.05), and XFBD-H also showed a significant inhibitory effect on the expression of STAT3 (7.08 ± 2.25 vs 2.98 ± 1.32, p < 0.001) ( Fig. 7 F).Fig. 7 XFBD inhibits pulmonary fibrosis by down-regulating M2 polarization and IL-6/STAT3 pathway. (A, B, E) XFBD inhibits fibrosis-induced macrophage infiltration and M2 polarization in mouse lungs in vivo. (C, D, F) Representative microphotographs of immunohistochemical analysis for expression of IL-6 and STAT3 in lung tissue sections showing reduced immunopositivity upon XFBD treatment as compared to BLM alone treated lung sections (Magnification: 20 × ). (G, H) Quantitative analyses of mRNA of the IL-6 and STAT3 genes in mice. (I) Quantitative analyses of cytokine content of IL-6 by ELISA in mice serum and lung tissue. (J) Western blot result of the expression of total STAT3 in polarized macrophages group and XFBD group. #p < 0.05, ##p < 0.01, ###p < 0.001, vs. control group, p < 0.05, p < 0.01, *p < 0.001, vs. Model group. Scale bar = 100 μm. Fig. 7 We prepared mice lung tissue homogenate and further investigated the mechanism via evaluating the gene expression of IL-6 and STAT3. Compared with the control group, the expression of IL-6 and STAT3 genes in the lung tissue of the model group induced by BLM was significantly up-regulated. The XFBD-H group inhibited the expression of IL-6 (4.86 ± 0.69 vs 1.27 ± 0.75, p < 0.01) ( Fig. 7 G) and STAT3 (3.68 ± 1.77 vs 0.86 ± 0.31, p < 0.01) ( Fig. 7 H). The results showed that XFBD could down-regulate the expression of IL-6 and STAT3 in mice at the gene level. In addition, the IL-6 protein expression in mouse serum and lung tissue was detected, and both XFBD-L and XFBD-H groups could significantly inhibit the upregulation of IL-6. It is worth noting that in mouse serum, compared to the model group, XFBD-H (5.93 ± 1.70 vs 18.55 ± 4.16, p < 0.01) groups showed more significant regulation ( Fig. 7 I). Western blot results also showed that polarized M2 macrophage cells with decreased total STAT3, while after XFBD treatment, was back to normal expression level (Fig. 7J). 4 Discussion Pulmonary fibrosis caused by SARS-CoV was an important clinical feature which seriously affected the quality of SARS-CoV patients' life. Pathological examination results showed that there were more fibrotic components in the patient's alveolar cavity edema fluid, and fibroblast proliferation in the patient's alveolar septum, which led to pulmonary interstitial fibrosis (Zhang et al., 2021). The clinical trial data of XFBD showed that it satisfactorily shortens the duration for virus clearance as well as the hospitalization period compared to the control group only treated with anti-viral drugs (Li et al., 2021). It is reported that, multiple herbs of XFBD and their main components have an effect in balancing immune inflammatory response, resisting viral infection and viral protein transcription, and restoring the balance of liver and gallbladder metabolism and energy metabolism in the body by regulating the biological processes of viral infection, immune inflammation, liver and gallbladder metabolism and energy metabolism (Wang et al., 2020). This study is based on the significant clinical antiviral pneumonia efficacy of XFBD. First, this traditional Chinese medicine compound was analyzed by the network pharmacological pharmacodynamic material basis analysis and their ingredients were tested, and then the efficacy of XFBD against fibroblast activation and macrophage inflammation in vitro and in vivo were discussed. Last, the pharmacodynamic test and the mechanism were explored. Some of the compounds quantified by UHPLC were reported to have anti-inflammatory or anti-fibrotic effects. For example, glycyrrhizic acid inhibited the proliferation of 3T3 fibroblasts and down-regulated the expression of IL-6 in LPS-induced RAW264.7 macrophages; glycyrrhizic acid also reduced the inflammation in BLM-induced pulmonary fibrosis rats in vivo via inhibiting the activation of TGF-β signaling pathway (Gao et al., 2015; Wu et al., 2015). Polydatin could significantly reduce the levels of IL-6 and TNF-α in pulmonary fibrosis tissue (Liu et al., 2020b). We established a TGF-β1-induced fibroblast activation model and interleukin 4 (IL-4)/Lipopolysaccharide (LPS)-induced macrophage inflammation models in vitro. The results proved that XFBD can effectively inhibit fibroblast collagen deposition, down-regulate the level of α-SMA and inhibit the migration of fibroblasts. In vivo experiments, the results proved that XFBD improved the weight loss and survival rate of the mice. The XFBD high-dose administration group had a significant effect in inhibiting collagen deposition and the expression of α-SMA in the lungs of mice. Macrophages are a key factor affecting pulmonary fibrosis. M2 type macrophages are thought to cause the activation of TGF-β/Smad and IL-6/STAT3 signaling pathways. In this study, we found that the anti-fibrotic effect of XFBD was achieved by regulating the inflammatory response of macrophages. In a macrophage inflammation model, XFBD could down-regulate the expression of IL-4-induced type 2 macrophage marker protein CD206+, as well as LPS-induced macrophage IL-6 and iNOS. Similarly, we proved that XFBD could down-regulate the expression of macrophages and M2 macrophages in vivo, which proved that XFBD could inhibit the type 2 polarization of macrophages and down-regulate the secretion of inflammatory factors and proteins. In addition, PPI interactions and hub genes of XFBD targets related to inflammatory response were analyzed by Cytoscape software (the larger the node, the higher the degree), Among these genes, IL-6 demonstrated the highest node degrees, which was 65. IL-6 is a pleiotropic cytokine that sends inflammatory signals throughout the body from local lesions, it stimulates immune cell recruitment including monocyte/macrophage, in turn, these immune cells produce more inflammatory factor to recruit more macrophage cells, driving the inflammatory cascade reaction, which cause persistent tissue damage and finally lead to pathological fibrosis development (Tanaka et al., 2016; Wynn and Vannella, 2016). The experimental results proved that XFBD effectively down-regulated the gene expression of IL-6 and STAT3 and the protein expression of both. STAT3 can mediate gene transcription in response to the IL-6 cytokine family (Shieh et al., 2019; Waters et al., 2019). Thus, targeting IL-6/STAT3 would be an effective approach to regulate fibroblast activation and differentiation. Studies have shown that the IL-6/STAT3 signaling pathway was activated in M2 macrophages (Yin et al., 2018). IL-6 secreted by trophoblast cells activates the STAT3 pathway to promote the polarization of M2 macrophages. In addition, activated M2 macrophages promote the invasion and migration of trophoblast cells in a feedback-regulated manner (Ding et al., 2021). Therefore, in this study, the effect of XFBD on the polarization of M2 macrophages may be through inhibiting the activation of the IL-6/STAT3 pathway. 5 Ethics statement This study was conducted in accordance with the recommendations of the “Guidelines for Animal Experiments of Tianjin University of Traditional Chinese Medicine”. The program has been approved by the Institutional Animal Care and Use Committee of Tianjin International Joint Academy of Biotechnology and Medicine (TJAB- JY-2011-002). Funding This work was supported by 10.13039/501100012166 National Key R&D Program of China [2020YFA0708000]; and the grant from 10.13039/100014717 National Natural Science Foundation of China (No. 82074032). Author contributions J.Y., Y.X.C. and H.Z. designed the experiments. X.Q.S., Y.Y.W. and H.L.Q. performed the biological experiments of pulmonary fibrosis model in vitro and in vivo. R.S. performed network pharmacology analysis. X.H.C. performed UHPLC analysis. L.L. and Y.W. participates in sample preparation. Y.Z., B.L.Z. and X.M.G. provided guidance to use Xuanfei Baidu prescription and participated in the discussion, Y.Y.W., R.S., X.Q.S., X.H.C. and Z.F.X. drafted the manuscript. All authors read and approved the final manuscript. 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 Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.jep.2021.114701. ==== Refs References Beigel J.H. Tomashek K.M. Dodd L.E. Mehta A.K. Zingman B.S. Kalil A.C. Hohmann E. Chu H.Y. Luetkemeyer A. Kline S. Lopez de Castilla D. Finberg R.W. Dierberg K. Tapson V. Hsieh L. Patterson T.F. Paredes R. Sweeney D.A. Short W.R. Touloumi G. Lye D.C. Ohmagari N. Oh M.D. Ruiz-Palacios G.M. Benfield T. Fatkenheuer G. Kortepeter M.G. Atmar R.L. Creech C.B. Lundgren J. Babiker A.G. Pett S. Neaton J.D. Burgess T.H. Bonnett T. Green M. Makowski M. Osinusi A. Nayak S. Lane H.C. Members A.-S.G. Remdesivir for the treatment of covid-19 - preliminary report N. Engl. J. Med. 383 2020 1813 1826 10.1056/NEJMoa2007764 32445440 Chanda D. Otoupalova E. Smith S.R. Volckaert T. De Langhe S.P. Thannickal V.J. Developmental pathways in the pathogenesis of lung fibrosis Mol. Aspect. Med. 65 2019 56 69 10.1016/j.mam.2018.08.004 Chandrasekaran P. Izadjoo S. Stimely J. Palaniyandi S. Zhu X. Tafuri W. Mosser D.M. Regulatory macrophages inhibit alternative macrophage activation and attenuate pathology associated with fibrosis J. Immunol. 203 2019 2130 2140 10.4049/jimmunol.1900270 31541024 Craig V.J. Zhang L. Hagood J.S. Owen C.A. Matrix metalloproteinases as therapeutic targets for idiopathic pulmonary fibrosis Am. J. Respir. Cell Mol. Biol. 53 2015 585 600 10.1165/rcmb.2015-0020TR 26121236 Della Latta V. Cecchettini A. Del Ry S. Morales M.A. Bleomycin in the setting of lung fibrosis induction: from biological mechanisms to counteractions Pharmacol. Res. 97 2015 122 130 10.1016/j.phrs.2015.04.012 25959210 Ding J. Yang C. Cheng Y. Wang J. Zhang S. Yan S. He F. Yin T. Yang J. Trophoblast-derived IL-6 serves as an important factor for normal pregnancy by activating Stat3-mediated M2 macrophages polarization Int. Immunopharm. 90 2021 106788 10.1016/j.intimp.2020.106788 Dong J. Ma Q. Macrophage polarization and activation at the interface of multi-walled carbon nanotube-induced pulmonary inflammation and fibrosis Nanotoxicology 12 2018 153 168 10.1080/17435390.2018.1425501 29338488 Gao L. Tang H. He H. Liu J. Mao J. Ji H. Lin H. Wu T. Glycyrrhizic acid alleviates bleomycin-induced pulmonary fibrosis in rats Front. Pharmacol. 6 2015 215 10.3389/fphar.2015.00215 26483688 Heukels P. Moor C.C. von der Thusen J.H. Wijsenbeek M.S. Kool M. Inflammation and immunity in IPF pathogenesis and treatment Respir. Med. 147 2019 79 91 10.1016/j.rmed.2018.12.015 30704705 Huang Y.F. Bai C. He F. Xie Y. Zhou H. Review on the potential action mechanisms of Chinese medicines in treating Coronavirus Disease 2019 (COVID-19) Pharmacol. Res. 158 2020 104939 10.1016/j.phrs.2020.104939 32445956 King T.E.J. Pardo A. Selman M. Idiopathic pulmonary fibrosis Lancet 378 2011 1949 1961 10.1016/s0140-6736(11)60052-4 21719092 Kohl M. Wiese S. Warscheid B. Cytoscape: software for visualization and analysis of biological networks Methods Mol. Biol. 696 2011 291 303 10.1007/978-1-60761-987-1_18 21063955 Ledford H. Coronavirus breakthrough: dexamethasone is first drug shown to save lives Nature 582 2020 469 10.1038/d41586-020-01824-5 32546811 Li M. Luan F. Zhao Y. Hao H. Zhou Y. Han W. Fu X. Epithelial-mesenchymal transition: an emerging target in tissue fibrosis Exp. Biol. Med. 241 2016 1 13 10.1177/1535370215597194 Li G. Jin F. Du J. He Q. Yang B. Luo P. Macrophage-secreted TSLP and MMP9 promote bleomycin-induced pulmonary fibrosis Toxicol. Appl. Pharmacol. 366 2019 10 16 10.1016/j.taap.2019.01.011 30653976 Li F. Li Y. Zhang J. Li S. Mao A. Zhao C. Wang W. Li F. The therapeutic efficacy of Xuanfei Baidu Formula combined with conventional drug in the treatment of coronavirus disease 2019: a protocol for systematic review and meta-analysis Medicine (Baltim.) 100 2021 e24129 10.1097/md.0000000000024129 Liu Q. Chu H. Ma Y. Wu T. Qian F. Ren X. Tu W. Zhou X. Jin L. Wu W. Wang J. Salvianolic acid B attenuates experimental pulmonary fibrosis through inhibition of the TGF-β signaling pathway Sci. Rep. 6 2016 27610 10.1038/srep27610 27278104 Liu Z. Guo F. Wang Y. Li C. Zhang X. Li H. Diao L. Gu J. Wang W. Li D. He F. BATMAN-TCM: a bioinformatics analysis tool for molecular mechANism of traditional Chinese medicine Sci. Rep. 6 2016 21146 10.1038/srep21146 26879404 Liu S.S. Lv X.X. Liu C. Qi J. Li Y.X. Wei X.P. Li K. Hua F. Cui B. Zhang X.W. Yu J.J. Yu J.M. Wang F. Shang S. Zhao C.X. Hou X.Y. Yao Z.G. Li P.P. Li X. Huang B. Hu Z.W. Targeting degradation of the transcription factor C/EBPbeta reduces lung fibrosis by restoring activity of the ubiquitin-editing enzyme A20 in macrophages Immunity 51 2019 522 534 10.1016/j.immuni.2019.06.014 e527 31471107 Liu M. Gao Y. Yuan Y. Yang K. Shi S. Zhang J. Tian J. Efficacy and safety of integrated traditional Chinese and western medicine for corona virus disease 2019 (COVID-19): a systematic review and meta-analysis Pharmacol. Res. 158 2020 104896 10.1016/j.phrs.2020.104896 32438037 Liu Y. Chen B. Nie J. Zhao G. Zhuo J. Yuan J. Li Y. Wang L. Chen Z. Polydatin prevents bleomycin-induced pulmonary fibrosis by inhibiting the TGF-β/Smad/ERK signaling pathway Exp Ther Med 20 2020 62 10.3892/etm.2020.9190 32952652 Lu H. Wu L. Liu L. Ruan Q. Zhang X. Hong W. Wu S. Jin G. Bai Y. Quercetin ameliorates kidney injury and fibrosis by modulating M1/M2 macrophage polarization Biochem. Pharmacol. 154 2018 203 212 10.1016/j.bcp.2018.05.007 29753749 Luo E. Zhang D. Luo H. Liu B. Zhao K. Zhao Y. Bian Y. Wang Y. Treatment efficacy analysis of traditional Chinese medicine for novel coronavirus pneumonia (COVID-19): an empirical study from Wuhan, Hubei Province, China Chin. Med. 15 2020 34 10.1186/s13020-020-00317-x 32308732 Mari B. Crestani B. Dysregulated balance of lung macrophage populations in idiopathic pulmonary fibrosis revealed by single-cell RNA seq: an unstable "menage-a-trois Eur. Respir. J. 54 2019 10.1183/13993003.01229-2019 Martinez F.J. Collard H.R. Pardo A. Raghu G. Richeldi L. Selman M. Swigris J.J. Taniguchi H. Wells A.U. Idiopathic pulmonary fibrosis Nat Rev Dis Primers 3 2017 17074 10.1038/nrdp.2017.74 29052582 Pulivendala G. Bale S. Godugu C. Honokiol: a polyphenol neolignan ameliorates pulmonary fibrosis by inhibiting TGF-beta/Smad signaling, matrix proteins and IL-6/CD44/STAT3 axis both in vitro and in vivo Toxicol. Appl. Pharmacol. 391 2020 114913 10.1016/j.taap.2020.114913 32032644 Shieh J.M. Tseng H.Y. Jung F. Yang S.H. Lin J.C. Elevation of IL-6 and IL-33 levels in serum associated with lung fibrosis and skeletal muscle wasting in a bleomycin-induced lung injury mouse model Mediat. Inflamm. 2019 1 12 10.1155/2019/7947596 2019 Stelzer G. Rosen N. Plaschkes I. Zimmerman S. Twik M. Fishilevich S. Stein T.I. Nudel R. Lieder I. Mazor Y. Kaplan S. Dahary D. Warshawsky D. Guan-Golan Y. Kohn A. Rappaport N. Safran M. Lancet D. The GeneCards suite: from gene data mining to disease genome sequence analyses Curr. Protoc. Bioinformatics. 54 2016 1301 13033 10.1002/cpbi.5 Szklarczyk D. Morris J.H. Cook H. Kuhn M. Wyder S. Simonovic M. Santos A. Doncheva N.T. Roth A. Bork P. Jensen L.J. von Mering C. The STRING database in 2017: quality-controlled protein-protein association networks, made broadly accessible Nucleic Acids Res. 45 2017 D362 D368 10.1093/nar/gkw937 27924014 Tanaka T. Narazaki M. Kishimoto T. Immunotherapeutic implications of IL-6 blockade for cytokine storm Immunotherapy 8 2016 959 970 10.2217/imt-2016-0020 27381687 Tashiro J. Rubio G.A. Limper A.H. Williams K. Elliot S.J. Ninou I. Aidinis V. Tzouvelekis A. Glassberg M.K. Exploring animal models that resemble idiopathic pulmonary fibrosis Front. Med. 4 2017 118 10.3389/fmed.2017.00118 Tay M.Z. Poh C.M. Rénia L. MacAry P.A. Ng L.F.P. The trinity of COVID-19: immunity, inflammation and intervention Nat. Rev. Immunol. 20 2020 363 374 10.1038/s41577-020-0311-8 32346093 TCM treatment Effective on over 90% of COVID-19 Patients on Mainland 2020 [cited 2020 17 December];Availablefrom:http://english.www.gov.cn/news/topnews/202003/23/content_WS5e787a9dc6d0c201c2cbf3b4.html Wang H. Song H.X. Wang D.F. Ma X.R. Zou D.X. Miao J.X. Wang Y.L. Yang W.P. The molecular mechanism of Xuanfei Baidu Formula in the treatment of COVID-19 antiviral effect based on network pharmacology and molecular docking Journal of Hainan Medical University 26 2020 1361 1372 10.13210/j.cnki.jhmu.20200617.003 Waters D.W. C Blokland K.E. Pathinayake P.S. Wei L. Schuliga M. Jaffar J. Westall G.P. Hansbro P.M. Prele C.M. Mutsaers S.E. Bartlett N.W. Burgess J.K. Grainge C.L. Knight D.A. STAT3 regulates the onset of oxidant-induced senescence in lung fibroblasts Am. J. Respir. Cell Mol. Biol. 61 2019 61 73 10.1165/rcmb.2018-0328OC 30608861 Wu C. He L. Guo H. Tian X. Liu Q. Sun H. Inhibition effect of glycyrrhizin in lipopolysaccharide-induced high-mobility group box 1 releasing and expression from RAW264.7 cells Shock 43 2015 412 421 10.1097/SHK.0000000000000309 25526376 Wynn T.A. Vannella K.M. Macrophages in tissue repair, regeneration, and fibrosis Immunity 44 2016 450 462 10.1016/j.immuni.2016.02.015 26982353 Xiong W.Z. Wang G. Du J. Ai W. Efficacy of herbal medicine (Xuanfei Baidu decoction) combined with conventional drug in treating COVID-19:A pilot randomized clinical trial Integr Med Res 9 2020 100489 10.1016/j.imr.2020.100489 32874913 Xu X. Han M. Li T. Sun W. Wang D. Fu B. Zhou Y. Zheng X. Yang Y. Li X. Zhang X. Pan A. Wei H. Effective treatment of severe COVID-19 patients with tocilizumab Proc. Natl. Acad. Sci. U. S. A. 117 2020 10970 10975 10.1073/pnas.2005615117 32350134 Yin Z. Ma T. Lin Y. Lu X. Zhang C. Chen S. Jian Z. IL-6/STAT3 pathway intermediates M1/M2 macrophage polarization during the development of hepatocellular carcinoma J. Cell. Biochem. 119 2018 9419 9432 10.1002/jcb.27259 30015355 Zhang C. Wu Z. Li J.W. Tan K. Yang W. Zhao H. Wang G.Q. Discharge may not be the end of treatment: pay attention to pulmonary fibrosis caused by severe COVID-19 J. Med. Virol. 93 2021 1378 1386 10.1002/jmv.26634 33107641	34606948	PMC9715986	NO-CC CODE	2022-12-03 23:20:52	no	J Ethnopharmacol. 2022 Jan 30; 283:114701	utf-8	J Ethnopharmacol	2,021	10.1016/j.jep.2021.114701	oa_other
==== Front J Ethnopharmacol J Ethnopharmacol Journal of Ethnopharmacology 0378-8741 1872-7573 Elsevier B.V. S0378-8741(21)00989-2 10.1016/j.jep.2021.114760 114760 Article Advanced applications of mass spectrometry imaging technology in quality control and safety assessments of traditional Chinese medicines Jiang Haiyan a Zhang Yaxin b Liu Zhigang c Wang Xiangyi b He Jiuming be∗∗ Jin Hongtao ade∗ a New Drug Safety Evaluation Center, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China b State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China c School of Biological Science and Engineering, South China University of Technology, Guangzhou, 510006, China d Beijing Union-Genius Pharmaceutical Technology Development Co., Ltd., Beijing 100176, China e NMPA Key Laboratory for Safety Research and Evaluation of Innovative Drug, Beijing 100050, China ∗ Corresponding author. NMPA Key Laboratory for Safety Research and Evaluation of Innovative Drug, Beijing 100050, China. ∗∗ Corresponding author. NMPA Key Laboratory for Safety Research and Evaluation of Innovative Drug, Beijing 100050, China. 19 10 2021 10 2 2022 19 10 2021 284 114760114760 13 8 2021 5 10 2021 18 10 2021 © 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. Ethnopharmacological relevance Traditional Chinese medicines (TCMs) have made great contributions to the prevention and treatment of human diseases in China, and especially in cases of COVID-19. However, due to quality problems, the lack of standards, and the diversity of dosage forms, adverse reactions to TCMs often occur. Moreover, the composition of TCMs makes them extremely challenging to extract and isolate, complicating studies of toxicity mechanisms. Aim of the review: The aim of this paper is therefore to summarize the advanced applications of mass spectrometry imaging (MSI) technology in the quality control, safety evaluations, and determination of toxicity mechanisms of TCMs. Materials and methods Relevant studies from the literature have been collected from scientific databases, such as “PubMed”, “Scifinder”, “Elsevier”, “Google Scholar” using the keywords “MSI”, “traditional Chinese medicines”, “quality control”, “metabolomics”, and “mechanism”. Results MSI is a new analytical imaging technology that can detect and image the metabolic changes of multiple components of TCMs in plants and animals in a high throughput manner. Compared to other chemical analysis methods, such as liquid chromatography-mass spectrometry (LC-MS), this method does not require the complex extraction and separation of TCMs, and is fast, has high sensitivity, is label-free, and can be performed in high-throughput. Combined with chemometrics methods, MSI can be quickly and easily used for quality screening of TCMs. In addition, this technology can be used to further focus on potential biomarkers and explore the therapeutic/toxic mechanisms of TCMs. Conclusions As a new type of analysis method, MSI has unique advantages to metabolic analysis, quality control, and mechanisms of action explorations of TCMs, and contributes to the establishment of quality standards to explore the safety and toxicology of TCMs. Graphical abstract Image 1 Keywords MSI TCMs Quality control Spatial metabolomics Pharmacology and toxicology ==== Body pmcAbbreviations: TCMs traditional Chinese medicines MSI mass spectrometry imaging IHC immunohistochemistry LC-MS liquid chromatography-mass spectrometry NMR nuclear magnetic resonance m/z mass to charge ratio MALDI matrix-assisted laser desorption/ionization DESI-MSI desorption electrospray ionization-mass spectrometry imaging SIMS secondary ion mass spectrometry LA-ICP-MSI laser ablation-inductively coupled plasma-mass spectrometry imaging CMC carboxymethyl cellulose GC gas chromatography LC liquid chromatography TLC thin-layer chromatography CE capillary electrophoresis FT-IR fourier-transform infrared PTFE polytetrafluoroethylene SA sinapic acid CHCA α-cyano-4-hydroxycinnamicacid 2-MBT 2-mercaptobenzothiazole DHB 2,5-dihydroxybenzoicacid DHAP 2,5-dihydroxyacetopheno 9-AA 9-aminoacridine DAN 1,5-diaminonaphthalene MCA 3,4-dimethoxycinnamic acid PLL poly-L-lysine MCAEF matrix coating assisted by an electric field AP-SMALDI atmospheric pressure-scanning microprobe matrix-assisted laser desorption/ionization PALDI-MS plasma assisted laser desorption ionization mass spectrometry GALDI colloidal graphite-assisted laser desorption/ionization TIAs terpenoid indole alkaloids ICs idioblast cells LCs laticifer cells Q-makers quality markers PCA principal component analysis OPLS-DA orthogonal partial least squares-discriminant analysis LDA linear discriminate analysis LLS local least square HELP heuristic evolving latent projections OPA orthogonal projection analysis LLF ligustri lucidi fructus QWBA quantitative whole body autoradiography GD graphite dot PNGL notoginseng leaf triterpenes NG-R1 notoginsenoside R1 MCAO/R middle cerebral artery occlusion/reperfusion ATP adenosine triphosphate 1 Introduction Traditional Chinese medicines (TCMs) have been used in the clinic for thousands of years and have shown good therapeutic effects. Due to the complexity of components and the characteristics of multi-target actions, TCMs can be used for broad opportunities, but face severe challenges. Given their various types, qualities, and efficacies, the key to the modernization of TCMs is to study their material bases, discover their therapeutic or toxic components, control their qualities, and clarify their targets and mechanisms of action. A variety of analytical methods have been used for the identification and mechanistic evaluation of individual TCM components, and can be mainly divided into two categories: chromatographic methods (including gas chromatography (GC) and hyphenated techniques (Zhang et al., 2013), liquid chromatography (LC) and hyphenated techniques (Wang et al., 2021a), thin-layer chromatography (TLC)(Chen et al., 2021), capillary electrophoresis (CE)) and spectroscopic methods (fourier-transform infrared (FT-IR)(Mukrimin et al., 2019), near-infrared spectroscopy (NIR)(Li et al., 2017) and nuclear magnetic resonance (NMR)(Zhao et al., 2020)). GC-MS and LC-MS in chromatographic methods are two very popular chromatographic detection methods with high resolutions and sensitivities. The GC is appropriate for the determination of volatile components and LC is suitable for the identification of liquid ingredients in TCMs. However, the premise of these two methods requires complex pre-processing of samples, which will not only destroy information on the distribution of compounds in tissues, but may cause the loss of substances in low abundance (Prideaux and Stoeckli, 2012). FT-IR and NIR are spectroscopic methods that are non-invasive, rapid, and require simple sample preparations. However, their accuracies are lower than that of GC-MS and LC-MS. NMR has high accuracy and stability, but its sensitivity is poor, which renders it incapable of analyzing a large number of low abundance metabolites (Jiang et al., 2010). As an emerging analytical method, mass spectrometry imaging (MSI) overcomes the above technical defects. Without requiring complicated sample pre-processing steps, MSI can detect known or unknown compounds in high-throughput, while achieving high sensitivities and resolutions. In addition, this technology can convert a large volume of mass spectral data into images, retaining in situ information to show the distribution of drugs and small molecule metabolites (Nilsson et al., 2012; Nimesh et al., 2013; Prideaux and Stoeckli, 2012). In recent years, spatially resolved metabolomics derived from MSI technology has been widely used in quality control and mechanistic studies of TCMs, and was first proposed by Sumner's research group at the Joint Annual Meeting of the American-Fern-Society in 2007 (Watson et al., 2007). Compared to traditional MS methods (LC-MS/GC-MS), MSI can retain the in situ spatial information of metabolites. The former “Spatially” of spatially resolved metabolomics can be used to accurately identify and locate the differential distribution of various metabolites in Chinese herbal medicines in tissues and cells, and perform rapid quality screening of drugs. The latter “metabolomics” can be used for in-depth metabolic analyses of target micro-regions to identify the types and contents of metabolites and discover potential efficacy or toxicity biomarkers of various components of TCMs. Such studies lay the foundation for understanding the possible medicinal and toxic mechanisms of TCMs (Bjarnholt et al., 2014; Ganesh et al., 2021). This article reviews the principles and characteristics of MSI technology, as well as its application to the identification, distribution, quality control, the discovery of efficacy/toxicity biomarkers, and possible mechanisms of action of TCM components. This review aims to promote the application of MSI technology in Chinese herbal medicine and provide new directions for the discovery of drugs and the establishment of quality control standards for TCMs. 2 MSI: insights into the principles, indicators, and experimental processing As a new type of molecular imaging technology, MSI performs mass spectrometry analysis and image visualization with high sensitivity, wide coverage, and strong identification ability. A variety of ions on the surface of tissue samples can be ionized point-by-point according to the spatial and multi-dimensional data of the mass to charge ratio (m/z), intensity, and position of ionized molecules obtained by mass spectrometry. Such data can be reconstructed and visualized using software (such as MassImager (He et al., 2018)) with the MSI functions of qualitative, quantitative, and positioning (Qin et al., 2018; Römpp and Spengler, 2013; Takahashi et al., 2015). Compared to LC-MS and immunohistochemistry (IHC), MSI can perform high-throughput detection of substances (endogenous and exogenous metabolites) in tissue sections, without requiring special labeling or complex pre-treatment, which can not only identify and analyze substances but also reveal their spatial distributions and relative contents in tissues (Schwamborn and Caprioli, 2010). MSI was originally developed based on matrix-assisted laser desorption/ionization (MALDI). Therefore, MALDI-MSI is the most widely used mass spectrometry imaging method (Caprioli et al., 1997). In addition, related technologies include desorption electrospray ionization-mass spectrometry imaging (DESI-MSI), secondary ion mass spectrometry (SIMS) imaging, and laser ablation-inductively coupled plasma-mass spectrometry imaging (LA-ICP-MSI), etc.(de Souza et al., 2020; Oppenheimer and Drexler, 2011; Parrot et al., 2018). These technologies are mainly classified according to their ionization mode: SIMS imaging uses a primary ion beam to bombard the surface of the sample, and then introduces secondary ions sputtered from the surface into the mass spectrometer for mass separation and determination (Yoon and Lee, 2018). MALDI-MSI mainly makes use of a matrix to absorb the laser energy and then transfers energy to sample molecules for ionization (Knochenmuss, 2006). DESI-MSI uses atomized charged droplets to hit the surface of the sample. After being hit by high-speed droplets, the sample is sputtered and then subjected to the mass spectrometer (Ifa et al., 2007; Takáts et al., 2004). The ion source type, spatial resolution, sample preparation requirements, and other information of these three mass spectrometry imaging technologies are summarized in Table 1 .Table 1 Comparison of the three most commonly used MSI techniques. Table 1Ionization type Ionization source Environment Resolution Characteristic Ref. MALDI IR/UV High vacuum/low vacuum IR：150 μm UV: 10–250 μm Need matrix, wide detection range Heyman and Dubery (2016) SIMS Primary ion beam High vacuum 50 nm–5 μm High resolution, high vacuum, easy to produce fragments of ions Behrens et al. (2012) DESI Charged corpuscle Atmospheric pressure 100–200 μm No matrix, atmospheric pressure Ifa et al. (2007); Takáts et al. (2004); Wiseman et al. (2006) 2.1 Critical indicators Speed, spatial resolution, and sensitivity are critical indicators of MSI(Vestal et al., 2020). Speed is the main factor affecting the experimental time, and the scanning rate mainly depends on the influence of the laser frequency, mobile platform speed, and signal acquisition. The increase in scanning speed leads to a decrease in ionized ions (Tillner et al., 2017). In this case, high sensitivity is key to ensuring the imaging results of low abundance ions. Spatial resolution and sensitivity are negatively correlated and an improvement in sensitivity will inevitably lead to a decrease in the mass resolution (Vestal et al., 2020). Sensitivity is also closely related to ionization efficiency, ion transport efficiency, and ion detection (Merdas et al., 2021), while the mass resolution is mainly dependent on the specific desorption/ionization method used (Handberg et al., 2015; Römpp and Spengler, 2013) (Table 1). Therefore, MSI is a systematic project, in which the limit value of indicators should be selected according to the experiment. 2.2 Experimental process We will use the most widely used technology, MALDI-MSI, as an example to describe the specific experimental process. First, the appropriate sample preparation method is selected according to the nature of the animal/plant tissue sample; a suitable matrix is selected for spraying based on the type and nature of the test object; a laser beam is used to desorb and ionize each sampling point. Subsequently, the analyte ions are separated and detected by the mass spectrometer to obtain the mass spectra associated with the sample space position. Finally, the MSI map is obtained by matching and reorganizing all of the mass spectral data with their corresponding two-dimensional spatial positions using software (Fig. 1 ) (Dong et al., 2016; Grassl et al., 2011; Sturtevant et al., 2016). The following is an additional introduction to the key experimental steps to enhance the readers’ understanding.Fig. 1 The Experimental process of MALDI-MSI (kidney). Fig. 1 2.2.1 Sample preparation Sample processing is the most critical step in MSI and the material basis for obtaining experimental results. The pretreatment method varies according to the type and location of the sample. For plant samples, a section of the roots, stems, and fruits is generally sliced using a cryostat microtome. Generally, such samples must also be embedded with gelatin (Beck and Stengel, 2016; Gemperline et al., 2014), 2% carboxymethyl cellulose (CMC) (Enomoto, 2020; Li et al., 2020b), or ice (Gorzolka et al., 2014), and frozen in liquid nitrogen prior to slicing into frozen sections (5–20 μm) at −20 °C. However, for plant stem slices with higher water contents or a young and small surface area, the sample is easily deformed or migration of the material occurs due to the blowing of spray gas. Thus, imprinting can be used for sample pretreatment in such situations. This technology utilizes external pressure to transfer a thin layer of plant tissue cell contents in situ to an adsorbent TLC plate (Liao et al., 2019) or the polytetrafluoroethylene (PTFE) membrane (Thunig et al., 2011) for imaging. For the petals and leaves, the surface must be kept as flat as possible, which can be directly blown or imprinted for indirect imaging. 2.2.2 Matrix selection In MALDI-MS analysis, the image quality depends in large part on the establishment and optimization of the matrix system, and thus, the choice and spray type for the matrix is very important. Commonly used matrices include sinapic acid (SA)(Chaurand et al., 2008), α-cyano-4-hydroxycinnamicacid (CHCA)(Grassl et al., 2011; Lemaire et al., 2006), 2-mercaptobenzothiazole (2-MBT)(Astigarraga et al., 2008), 2,5-dihydroxybenzoicacid (DHB)(Li et al., 2016b), 2,5-dihydroxyacetopheno (DHAP)(Jovanović and Peter-Katalinić, 2016), 9-aminoacridine (9-AA)(Morikawa-Ichinose et al., 2019), and 1,5-diaminonaphthalene (DAN)(Korte and Lee, 2014). Among them, SA and DHAP are suitable for the detection of high molecular weight biomolecules (proteins, oligosaccharides, etc.), CHCA and 2-MBT are fit for the detection of medium molecular weight analytes (peptides, lipids), and DHB, DAN, and 9-AA are preferred for the detection of low molecular weight molecules (fatty acids, amino acids, nucleotides, etc.). In addition, some novel matrices such as quercetin (Wang et al., 2014), N-phenyl-2-naphthylamine (Liu, H. et al., 2018), graphene oxide (Wang et al., 2017), 3,4-dimethoxycinnamic acid (DMCA)(He, H. et al., 2019) and poly-L-lysine (PLL)(He, Y. et al., 2019) have been successfully used for MALDI-MSI. After selecting the suitable matrix according to the sample type, it is necessary to evenly cover the matrix solution on the surface of the tissue section to form good co-crystallization with the tissue surface molecules. There are three main methods of matrix covering, including manual spraying, automatic spraying, and vacuum sublimation (Bjarnholt et al., 2014). Furthermore, matrix coating assisted by an electric field (MCAEF) has also been proven to enhance tissue imaging (Wang et al., 2015). 2.2.3 Data processing MSI will obtain large volumes of mass spectral data during high-throughput detection, which can be reconstructed and visualized into image information using imaging software (such as MassImager (He et al., 2018), R Packages (Ràfols et al., 2020), MSiReader (Desbenoit et al., 2018), etc.). Imaging software can image the ions individually or simultaneously to show the distribution of the target molecule in the sectioned tissue. The identification of target molecules can be based on the accurate mass value in commonly used mass spectrometry databases such as METLIN (http://metlin.scripps.edu/), HMDB (http://hmdb.ca/), MassBank (https://massbank.eu/MassBank/), and Lipid Maps (http://www.lipidmaps.org/.) for preliminary search matching. Then the verification of the compound is performed according to the specific fragment ions of the compound in the MS/MS experiment and other experimental support materials (such as nuclear magnetic or ultraviolet spectroscopy). In addition, the mass spectral data can be screened according to the experimental design and compared with KEGG (https://www.kegg.jp/) and other databases to explore the drug mechanisms of action. MSI has a wide detection range from exogenous drugs to endogenous metabolites (lipids, peptides, etc.) and metals (Aichler and Walch, 2015). Sample preparation, parameter settings, data processing, and other MSI operations are detailed in the literature (Gessel et al., 2014; Kaletaş et al., 2009; Schulz et al., 2019). To date, MSI has been widely used in the fields of medicine (Schulz et al., 2019; Végvári, 2015), environment (Böhme et al., 2015), food (Morisasa et al., 2019), and plant biology (Kaspar et al., 2011; Korte et al., 2015; Qin et al., 2018). The MSI methods, research drugs, tissue types, and imaged molecules involved in this article are summarized in their order of appearance in Table 2 .Table 2 Published literature showing the application of MSI for the composition, quality control, and mechanisms of action of TCMs and natural products. Table 2Drug Tissue type Technical method Imaged molecules Ref. Salvia miltiorrhiza Whole plant MALDI-MSI Functional metabolites Sun et al. (2020) Salvia miltiorrhiza Roots, stems and leaves MALDI-MSI Phenolic acids and tanshinones Li et al. (2020b) Tripterygium Roots MALDI-MSI Triterpenoids and sesquiterpene alkaloids Lange et al. (2017) Paeonia lactiflora Roots AP-SMALDI MSI Gallotannins and monoterpene glucosides Li et al. (2016a) Maple Xylem TOF-SIMS imaging Syringyl and guaiacyl lignin Saito et al. (2012) Putterlickia pyracantha Stems and roots MALDI-MSI Maytansinoids Eckelmann et al. (2016) Scutellaria baicalensis Roots PALDI-based MSI Baicalein and wogonin Feng et al. (2014) Asclepias curassavica Injury site 3D-surface MALDI MSI Plant defensive cardiac glycosides Dreisbach et al. (2021) Glycyrrhiza glabra Rhizome AP-MALDI-MSI(Koestler et al., 2008) Flavonoids, flavonoid glycosides and saponins Li et al. (2014) Ginkgo biloba L. Leaves AP-MALDI-MSI Flavonoid glycosides and biflavonoids Beck and Stengel (2016) Catharanthus roseus Stem tissue MALDI-MSI TIAs Yamamoto et al. (2016) Catharanthus roseus Leaves MALDI-MSI TIAs and precursors Yamamoto et al. (2019) Panax ginseng Roots MALDI-MSI Ginsenosides Bai et al. (2016); Lee et al. (2017); Taira et al. (2010) Ginseng Roots DESI-MSI Ginsenosides Yang et al. (2021) Panax ginseng, Panax quinquefolius, and Panax notoginseng Roots MALDI-MSI Saponins Wang et al. (2016) Aconitum carmichaeli Debx Roots MALDI-MSI Aconitum alkaloids Wang et al. (2009) Paeonia suffruticosa and Paeonia lactiflora Roots MALDI-MSI Monoterpene and paeonol glycosides, tannins, flavonoids, saccharides and lipids Li et al. (2021) Ligustri Lucidi Fructus (LLF) LLF fruits MALDI-MSI Q-markers Li et al. (2020a) Vinblastine The whole body of rats MALDI-IMS-MSI Sinblastine and metabolites Trim et al. (2008) Salidroside Multiple organs MALDI-MSI Salidroside Meng et al. (2020) Puerarin Mice kidney tissue GD-4-assisted MSI Puerarin and its two metabolites (daidzein and dihydrodaidzein) Shi et al. (2017) Scutellarin Mice kidney tissue MALDI-MSI Scutellarin and scutellarein Wang et al. (2021c) Notoginseng leaf triterpenes (PNGL) Rat brain MALDI-MSI Endogenous metabolites Wang et al. (2021b) Notoginsenoside R1 Rat brain MALDI-MSI Endogenous metabolites Zhu et al. (2020) Thymoquinone Rat brain MALDI-MSI Endogenous metabolites Tian et al. (2020) Radix Aconiti Lateralis Preparata extracts Rat heart MALDI-MSI Endogenous metabolites Wu et al. (2019) 3 MSI: A camera for showing the distribution of multiple components in a plant Investigations of the basal metabolism of TCMs are the premise for identifying new drug candidates, increasing the clinical range of drugs, and improving quality control. Secondary metabolites (such as flavonoids, mushrooms, alkaloids, etc.) are the main components of TCMs that can prevent or cure diseases. The types, contents, and relative proportions of secondary metabolites are key to determining the effectiveness and quality of TCMs (Zhang et al., 2018) and MSI is suitable for detecting the content and distribution of primary/secondary metabolites in various plant structures (petals, roots, stems, leaves, seeds, seedlings)(Enomoto, 2020; Enomoto and Nirasawa, 2020; Qin et al., 2018; Sagara et al., 2019). The conventional mass spectrometry method used to study the multiple components of TCMs is LC-MS. Complex pretreatment is generally required for LC-MS, including solvent extraction and chromatographic column separation before structural characterization. Such work not only requires substantial investigator energy and wastes a considerable amount of chemical reagents for sample preparation, but may also cause the loss of analytes or damage to the active ingredient (Wu et al., 2007). Furthermore, LC-MS fails to provide location information for the analyte in the tissue. Conversely, MSI can directly analyze the solid sections of plant tissues, without labeling and pre-processing and many studies have confirmed the advantages of direct analysis of plant tissues (Talaty et al., 2005; Wu et al., 2007). MSI can detect and identify the metabolic distribution of various components of TCMs while retaining in situ information, which is especially suitable for showing the material differences among different tissue parts of TCMs and the distribution characteristics of multiple components in the tissue (Hemalatha and Pradeep, 2013). In studies of Salvia miltiorrhiza, MALDI-MSI was used to visualize the spatial dynamics of functional metabolites (such as amino acids, phenolic acids, fatty acids, oligosaccharides, cholines, etc.) (Sun et al., 2020) and MALDI-MSI was used to determine the distribution of metabolites in the tissue structures of roots, stems, and leaves. In this study, the characteristic constituents of the medicinal plant Salvia miltiorrhiza were identified as phenolic acids and tanshinones, which was consistent with the LC-MS data (Li et al., 2020b). MALDI-MSI was also used to identify and show the location of specific metabolites in Tripterygium roots (Lange et al., 2017). In a study of Paeonia lactiflora, atmospheric pressure-scanning microprobe matrix-assisted laser desorption/ionization mass spectrometry imaging (AP-SMALDI MSI, 10 μm/30 μm resolution) was used to detail the specific distribution of the major secondary metabolites, gallotannins and monoterpene glucosides, in root samples (Li et al., 2016a). SIMS imaging was used to characterize the morphological distribution of syringyl and guaiacyl lignin in the xylem of maple samples, which revealed a clear difference in the annual distribution of lignins between the fiber and vessel (Saito et al., 2012). Take Putterlickia Pyracantha as an example to illustrate in detail, the maytansinoids of Putterlickia pyracantha were visualized by AP-SMALDI MSI in the rhizome and were highly distributed in the vascular cambium region and the phloem. Such compounds were also widely distributed in the xylem and extremely low in the outer bark (periderm) of the stem. In addition, maytansine and maytanprine were also mainly detected in the central cylinder of the root (Fig. 2 )(Eckelmann et al., 2016).Fig. 2 MSI results from Putterlickia pyracantha stems (Eckelmann et al., 2016). A. Chemical structures of maytansinoids occurring in Putterlickia pyracantha. B. (a) Anatomical imaging of the cross section of Putterlickia pyracantha stems stained with phloroglucinol/HCL. (b–d) MALDI-imaging-HRMS of different Putterlickia pyracantha stem cuttings (spatial resolution: 40 μm; scan area: b: 3720 × 2600 μm; c: 3520 × 4120 μm; d: 3720 × 2600 μm). Localization of maytansine ([M+K]+; m/z 730.2503), maytanprine ([M+K]+; m/z 744.2659), maytanbutine ([M+K] +; m/z 758.2816), maytansine precursor 6 ([M+K] +; m/z 716.2347), maytanvaline ([M+K]+; m/z 772.2973), normaytancyprine ([M+K]+; m/z 770.2816), maytansine ([M+K]+; m/z 730.2503), hydroxylated maytansine ([M+K]+; m/z 746.2452), and hydroxylated maytanprine ([M+K]+; m/z 760.2609). Fig. 2 Due to high background noise in the low mass (<500 Da) region and the spatial inhomogeneity of matrix crystals formed on plant tissues, the application of MSI to the analysis of small molecule metabolites in plant tissues is more challenging than that in animal tissues. To improve the spatial resolution of MSI, some new ion sources were constructed for plant tissue imaging. Plasma assisted laser desorption ionization mass spectrometry (PALDI-MS) combines multiwavelength laser desorption and heated metastable plasma ionization of analytes, and does not require solvents to decrease ion suppression, reduce the pH effect, or simplify complicated spectra caused by adducts to a high spatial resolution of 60 μm × 60 μm. PALDI-based MSI for tissue section imaging of Scutellaria baicalensis showed that the two active components, baicalein, and wogonin, were mainly distributed in the epidermis of the root (Feng et al., 2014). To solve the problem of uneven matrix distribution in MDLDI-MSI, colloidal graphite was introduced as an alternative matrix that can be evenly distributed on the sample surface. Colloidal graphite-assisted laser desorption/ionization (GALDI) MS imaging was developed to analyze the metabolites of Arabidopsis, showing the specific distribution of flavonoids in Arabidopsis in the whole flower and a single petal (Cha et al., 2008). In addition, 3D-MSI has been developed as cutting-edge technology for plant imaging. The 3D-surface MALDI-MSI is the most recent instrumental approach in AP-SMALDI MSI and was developed to characterize the specific distribution of plant defensive cardiac glycosides at injury sites in Asclepias curassavica (Dreisbach et al., 2021). Most of the MALDI imaging experiments performed on plant tissues have a spatial resolution of 50–200 μm. With high resolutions in mass and space, this technology has been applied to cell-level imaging in plants. The AP-MALDI-MSI approach (Koestler et al., 2008) that was independently developed by Li's laboratory achieves 10 μm resolution in cell level imaging in plants, thus, showing the distribution of the main natural products (flavonoids, flavonoid glycosides, and saponins) of Glycyrrhiza glabra (licorice)(Li et al., 2014). The technology was also used to detect and identify the distribution of flavonoid glycosides and biflavonoids in Ginkgo biloba L (Beck and Stengel, 2016). A study used MALDI-MSI based on the FT-ICR-MS detector (with a spatial resolution of 20 μm) to show that most of the terpenoid indole alkaloids (TIAs) in the stem tissue of Catharanthus roseus were accumulated in idioblast cells (ICs) and laticifer cells (LCs) (Yamamoto et al., 2016). Another study also used the FT-ICR-MS detector to image the leaves of Catharanthus roseus at a resolution of 10 μm, and was combined with single cell MS analysis to detail the biosynthesis of TIAs and determine the cell-specific localization of TIAs in leaf tissue (Yamamoto et al., 2019). MSI technology can achieve high resolution cell and tissue imaging, showing the specific distribution of the functional metabolites of TCMs and laying a foundation for subsequent mechanistic exploration. 4 MSI: A simple and quick way to discover the quality markers of TCMs Due to the polymorphism of medicinal plants, the quality control of drugs is a complicated process and includes a detailed characterization of the appearance, active ingredients, and physical and chemical properties of TCMs, as well as the quantification (absolute dry weight, yield, etc.), manufacturing (temperature, solvent, extraction and drying time), impurity testing, and chemical content determinations of the final active pharmaceutical ingredients (Liu, C. et al., 2018). In recent years, to improve the consistency and quality control of TCMs, quality markers (Q-makers) have been introduced ; (Guo, 2017; Liu et al., 2016). Q-markers of TCMs refer to substances that can be characterized qualitatively and quantitatively and are closely related to the function of the TCM in raw materials or during the processing and preparation of TCMs. The image of Q-makers plays an important role in the authenticity identification and quality assessment of TCMs (including raw materials, extracts, products, and compound preparations)(Yang et al., 2017). There are a variety of methods and strategies used for the discovery of Q-markers, including genomics, metabolomics, system pharmacology, pharmacokinetic analyses, and spider-web mode (Ren et al., 2020). As mentioned, MSI can detect the content and distribution of multiple components of TCMs in a high throughput manner. As a new analytical method, this technique has been used to discover the quality markers of TCMs. In this application, massive volumes of mass spectral data are generated and subsequently analyzed and processed by chemometric methods. Such methods mainly include principal component analysis (PCA), orthogonal partial least squares discriminant analysis (OPLS-DA), linear discriminate analysis (LDA), local least square (LLS), heuristic evolving latent projections (HELP), and orthogonal projection analysis (OPA)(Bansal et al., 2014). Compared to other chemical analyses such as LC-MS and UV, MSI does not require complicated sample extraction and separation steps, and does not lose low-abundance components. Thus, MSI quickly distinguishes the active ingredients and metabolic characteristics of different drugs, as well as readily identifies Q-markers. All such capabilities are suitable to rapidly and semi-quantitatively perform quality screening of TCMs (Huang et al., 2016). Panax ginseng is a type of precious Chinese medicine, known as the king of medicines. However, as there are multiple species of Panax ginseng, the origin, age, efficacy, and nutritional value of ginseng medicines are also different, and counterfeit or substandard products often exist in the market. Ginsenoside is the main active component in Panax ginseng, and the content of ginsenoside increases with plant age. Many studies have used MSI to reveal that ginsenosides are mainly distributed in the sebaceous layer and part of the cortex of Panax ginseng tissue located in the center of the root. Dozens of ginsenoside analytes have been identified by MS/MS as specific markers for quickly distinguishing different varieties, ages, and organs of Panax ginseng based on their specific distributions in tissues (Fig. 3 )(Bai et al., 2016; Lee et al., 2017; Taira et al., 2010; Wang et al., 2016; Yang et al., 2021). In one study, UPLC-QTOF MS and DESI-MSI were simultaneously used to detect and characterize the age and parts of ginseng to identify the common biomarkers across different age groups using the OPLS-DA method. The results showed that compared to UPLC-QTOF MS, DESI-MSI was a novel and stable method for the rapid evaluation of ginseng root slices (Yang et al., 2021). In addition, LC-MS and MALDI-MSI were also used to analyze Aconitum alkaloids in the Chinese herbal medicine, Aconitum carmichaeli Debx. The results between the two analytical methods were consistent and revealed significant differences in the contents of alkaloids between different samples. The comparative study using two analytical methods showed that MALDI-MSI was a more rapid and robust analytical method than LC-MS for semi-quantitative analyses of high concentration alkaloids (Wang et al., 2009). In addition, spatial metabolomics based on MALDI-MSI was also used to comprehensively and accurately detect the differential distribution of metabolites in Paeonia suffruticosa and Paeonia lactiflora (both belonging to genus Paeonia), including monoterpenes and paeonol glycosides, tannins, flavonoids, carbohydrates, and lipids, and it was also used to further visualize the gallotannins biosynthesis pathway in the roots of Paeonia suffruticosa and Paeonia lactiflora (Li et al., 2021). Most TCMs are crude drugs and the majority of which must be processed to reduce their toxicity in clinical medications. A strategy integrating multi-component characterization, non-target metabolomics, and MSI was proposed for quality control during processing. MSI was used to visualize the spatial distribution of four main biomarkers in the Ligustri Lucidi Fructus (LLF) based on steaming time (Li et al., 2020a).Fig. 3 MALDI-MSI distinguishing ginseng of different ages based on the localization of ginsenosides (Bai et al., 2016) A. (a) Optical scan image of ginseng. (b) Overlay of ion images: red, m/z 805.5 (Rg8/Rk/Rz1); yellow, m/z 955.6 (unidentified); blue, m/z 917.8 (unidentified). (c–g) Five localization modes of signals: xylem-medulla type (c1 and c2); xylem-only type (d1 and d2); cork-xylem type (e1 and e2); cork-phloem-cambium-medulla type (f1 and f2); and cork-only type (g1 and g2). B. PCA score plot and 2D peak distribution plot of m/z 1117.5 and m/z 1147.5: a and c, whole tissue; b and d, cork. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Fig. 3 MSI technique was used to discover Q-makers, providing a new direction and insights for the quality control of TCMs. As a rapid evaluation method, MSI has a broad applicability for the quality control of TCMs. 5 MSI: A tool for studying the metabolic distribution and therapeutic/toxic mechanisms of TCMs MSI can be applied to entire animal bodies or multiple tissue sections to observe the distribution of metabolites of active components in each organ, and to determine the target organ and toxicity. A study on the anticancer drug, vinblastine, performed MALDI-IMS-MS whole body imaging. The results showed that most of the product ions of vinblastine were highly distributed in the liver, renal cortex, and surrounding the gastric intestinal tract. The accuracy of the MSI results was verified by quantitative whole body autoradiography (QWBA)(Trim et al., 2008). By collecting multiple organ samples from mice at various time points after intravenous administration of salidroside, MALDI-MSI visualized the temporal and spatial distribution of salidroside showing that salidroside was heterogeneously distributed throughout the kidney and heart, and could be quickly eliminated with 5 min (Meng et al., 2020). The distribution of drugs in tumor tissues or organs is heterogeneous. The possible metabolic pathways of TCMs can also be predicted by MSI to analyze the distribution of active components and their metabolites in microregions of tissues or organs. It has been found that hydroxyl-group-dominated graphite dots (GD) are an ideal matrix with extremely low background noise and ultra-high sensitivity. GD-4-assisted MSI has been used to show the distribution characteristics of puerarin and its metabolites in renal microregions showing that puerarin was primarily distributed in the renal pelvis and major calyx. However, its metabolites (daidzein and dihydrodaidzein) were also detected in the renal pelvis, major calyx, and partly in the minor calyx, but were nearly absent in the medulla (Shi et al., 2017). In another study, MALDI-MSI was also used to identify the in situ localization of scutellarin (traditional Chinese botanic drug of Erigeron breviscapus extract) and its metabolites to show metabolic differences in the kidney (Wang et al., 2021d). Imaging the distribution characteristics after drug administration facilitates an understanding of the biological activity and metabolism of drugs in various animal organs. In recent years, various cutting-edge omics technologies (genomics, transcriptomics, proteomics, metabolomics, lipidomics) have been applied to diverse fields of TCM research, including screening, quality control, research and development, mechanistic research, and clinical verification. Taking metabolomics as an example, metabolism reflects the changes of small molecule metabolites in the body. Metabolomics with high-throughput monitoring can identify the metabolic network of molecules following drug administration, which has become a powerful tool effectively breaking through the application bottleneck of the study of the multi-component mechanisms of TCMs. The discovery of metabolic markers provides a foundation for the early identification of toxicity, quality control, and clinical utility of TCMs (Han et al., 2020; Shi et al., 2016; Sun et al., 2012; Wang et al., 2021a). As MSI is a high-throughput and label-free technology, it can obtain drug metabolism distribution information and also endogenous small molecule metabolism information (that is, metabonomics data) from the same animal tissue. Compared to traditional metabolomics methods, spatial high resolution metabolomics studies based on MSI can preserve tissue integrity and visualize the distribution of metabolites. Researchers can also superimpose MS images with optical/HE scanning images and focus on the tissue microregions or lesions of interest to accurately extract mass spectral data for the target area for metabolic research; thus, avoiding the challenges associate with difficult separations of research specimens. Panax notoginseng is a traditional Chinese medicine and is widely used for the treatment and prevention of ischemic cerebrovascular diseases (Yan et al., 2018). Notoginseng leaf triterpenes (PNGL) and notoginsenoside R1 (NG-R1, Fig. 4 ) extracted from Panax notoginseng were visualized by MALDI-MSI to study the effect on small molecule metabolism after perfusion injury. According to the results, the two drugs had a callback effect on the tricarboxylic acid (TCA) cycle and adenosine triphosphate (ATP) metabolism pathway, and also played a role in improving the malate-aspartate shuttle; thus, improving the antioxidant capacity and maintaining the homeostasis of Na+ and K+ (Wang et al., 2021b; Zhu et al., 2020). In the similar disease model, Fang et al. also used MALDI-MSI to explore the role of Thymoquinone, the main active ingredient in Nigella sativa, in regulating abnormal metabolism in injured brain areas by promoting the aerobic oxidation of glucose, regulating intracellular energy metabolism, improving the phospholipid molecular level, increasing the content of small antioxidant molecules, and balancing sodium homeostasis (Tian et al., 2020). According to the results of metabolomics studies in the same model, it is known that the mechanisms of related drugs for the treatment of stroke and other central nervous system diseases begin with mitochondrial oxidative damage, energy metabolism, lipid metabolism disorders, and Na+ homeostasis. In another study, MALDI-MSI was used to study anti-myocardial infarction effects of Radix Aconiti Lateralis Preparata extracts. Pharmacodynamics results showed that Radix Aconiti Lateralis Preparata extracts can improve the hemodynamic status and organ weight index and inhibit myocardial injury of rats with myocardial infarction. The corresponding MALDI-MSI results elucidated the possible mechanism of action by presenting Radix Aconiti Lateralis Preparata extracts to reverse metabolic changes of related small molecules (energy metabolism-related molecules, phospholipids, potassium ions, and glutamine in the heart) to produce anti-myocardial infarction effects (Wu et al., 2019). The identification of potential biomarkers of TCMs based on changes in the metabolic networks of small molecules in vivo, thus, lays a foundation for further exploration of the mechanisms of action.Fig. 4 Spatially resolved metabolomics based on MSI to elucidate the pharmacodynamic mechanisms of NG-R1 (Zhu et al., 2020). The rats in this study were divided into four groups: Sham, MCAO/R, NG-R1 (20 mg/kg, 7 days), and NBP (20 mg/kg, 7 days). Pharmacodynamic studies (included neurological score, TTC staining, histopathology staining, immunofluorescence staining, and TUNEL staining) conducted 7 days after ischemic-reperfusion showed that NG-R1 can reduce infarction volumes and neurologic deficits in MCAO/R rats and attenuate neuronal loss 7 d after MCAO/R surgery, while also inhibiting neuronal apoptosis and astrocyte activation. To clarify the mechanisms by which those events occur, the study further used spatially resolved metabolomics based on MALDI-MSI and found that NG-R1 can regulate the abnormal accumulation of glucose and citric acid, increase the content of glutamate and malate-aspartic acid shuttle components, increase antioxidant content, increase ATP metabolism, and maintain the homeostasis of Na+ and K+ to achieve anti-ischemia/reperfusion injury effects. Fig. 4 Spatial metabolomics based on MSI can detail the interactions between metabolites, and further screen and identify biomarkers with significant changes by comparing the correlation between metabolomics spectra and histopathological/biochemical indicators. Finally, the analysis of related metabolic pathways can reveal the possible effects or toxic mechanisms of TCMs. The above studies illustrate that spatial metabolomics analyses based on MSI methods are powerful in exploring the therapeutic effects of TCMs and provide insights into the potential mechanisms of action of TCMs. 6 Summary and conclusion In recent years, MSI has attracted the attention of many researchers and was rapidly developed. Currently, the quality control of most TCMs is limited to the identification and analysis following extraction and separation, and the process is cumbersome and time-consuming. The ingredients with lower concentrations are often overlooked and are not the focus of studies. MSI provides a new method for the rapid screening and control of the quality of TCMs. The understanding of modern medicine in TCMs has developed from macroscopic to microscopic considerations. In particular, the discovery and identification of active components of TCMs in the body is a key research topic. MSI technology has become a powerful tool for the analysis of metabolites in animal/plant tissues, as well as single cells, providing a means to study transport pathways, metabolic pathways, and the accumulation of exogenous drugs in animal tissues and endogenous metabolites in plant tissues. The multi-component and multi-target synergistic characteristics of TCMs have been advantageous for the treatment of certain chronic diseases. Extracting active ingredients from TCMs and isolating monomers is a key approach to the identification of new drugs. MSI also offers a new visual perspective and provides multi-dimensional information for metabolomics analysis. However, MSI technology has faced many challenges, such as its limited spatial resolution and insufficient sensitivity. By improving sample preparation methods, matrix replacement, algorithm optimization, and instrument improvements (Abdelmoula et al., 2018; Alexandrov et al., 2011; He et al., 2015; Morikawa-Ichinose et al., 2019; Song et al., 2017), MSI technology has achieved substantial breakthroughs in its sensitivity, resolution and sample suitability. With the integration of MSI with other technologies (Porta Siegel et al., 2018), such as LC-MS(Desbenoit et al., 2013), microscopic imaging (Tian et al., 2019; Van de Plas et al., 2015), Raman spectroscopy (Bocklitz et al., 2015), and magnetic resonance imaging (Verbeeck et al., 2017), the application of MSI technology to TCMs research will also become broader. Declaration of competing interests The authors report no conflicts of interest. Funding This work was supported by the 10.13039/501100001809 National Natural Science Foundation of China [grant numbers 81773996, 81773678, 81973476, and 82074104]; the National Major Scientific and Technological Special Project for “Major New Drugs Development” [grant numbers 2018ZX09735006, 2018ZX09711001-002-001]; and the Research Project of Clinical Toxicology Transformation from the Chinese Society of Toxicology [grant number CST2019CT105]. Declaration of interest The authors declare no conflict of interests. CRediT authorship contribution statement Haiyan Jiang: conceived the idea of the topic scope, wrote the manuscript, performed the literature search, and analyzed the data. Yaxin Zhang: wrote the manuscript, performed the literature search, and analyzed the data. Zhigang Liu: performed experiments and data collection. Xiangyi Wang: performed experiments and data collection. Jiuming He: conceived the idea of the topic scope, performed data analysis and critically revised the manuscript. Hongtao Jin: conceived the idea of the topic scope, performed data analysis and critically revised the manuscript. Acknowledgments The authors are grateful to International Science Editing (http://www.internationalscienceediting.com) for editing this manuscript. ==== Refs References Abdelmoula W.M. Pezzotti N. Hölt T. Dijkstra J. Vilanova A. McDonnell L.A. Lelieveldt B.P.F. Interactive visual exploration of 3D mass spectrometry imaging data using hierarchical stochastic neighbor embedding reveals spatiomolecular structures at full data resolution J. Proteome Res. 17 3 2018 1054 1064 10.1021/acs.jproteome.7b00725 29430923 Aichler M. Walch A. MALDI Imaging mass spectrometry: current frontiers and perspectives in pathology research and practice Lab. Invest. 95 4 2015 422 431 10.1038/labinvest.2014.156 25621874 Alexandrov T. Meding S. Trede D. Kobarg J. Balluff B. Walch A. Thiele H. Maass P. Super-resolution segmentation of imaging mass spectrometry data: solving the issue of low lateral resolution J Proteomics 75 1 2011 237 245 10.1016/j.jprot.2011.08.002 21854879 Astigarraga E. Barreda-Gómez G. Lombardero L. Fresnedo O. Castaño F. Giralt M.T. Ochoa B. Rodríguez-Puertas R. Fernández J.A. Profiling and imaging of lipids on brain and liver tissue by matrix-assisted laser desorption/ionization mass spectrometry using 2-mercaptobenzothiazole as a matrix Anal. Chem. 80 23 2008 9105 9114 10.1021/ac801662n 18959430 Bai H. Wang S. Liu J. Gao D. Jiang Y. Liu H. Cai Z. Localization of ginsenosides in Panax ginseng with different age by matrix-assisted laser-desorption/ionization time-of-flight mass spectrometry imaging J. Chromatogr. B 1026 2016 263 271 10.1016/j.jchromb.2015.09.024 Bansal A. Chhabra V. Rawal R.K. Sharma S. Chemometrics: a new scenario in herbal drug standardization J Pharm Anal 4 4 2014 223 233 10.1016/j.jpha.2013.12.001 29403886 Beck S. Stengel J. Mass spectrometric imaging of flavonoid glycosides and biflavonoids in Ginkgo biloba L Phytochemistry 130 2016 201 206 10.1016/j.phytochem.2016.05.005 27233155 Behrens S. Kappler A. Obst M. Linking environmental processes to the in situ functioning of microorganisms by high-resolution secondary ion mass spectrometry (NanoSIMS) and scanning transmission X-ray microscopy (STXM) Environ. Microbiol. 14 11 2012 2851 2869 10.1111/j.1462-2920.2012.02724.x 22409443 Bjarnholt N. Li B. D'Alvise J. Janfelt C. Mass spectrometry imaging of plant metabolites – principles and possibilities Nat. Prod. Rep. 31 6 2014 818 837 10.1039/C3NP70100J 24452137 Bocklitz T. Bräutigam K. Urbanek A. Hoffmann F. von Eggeling F. Ernst G. Schmitt M. Schubert U. Guntinas-Lichius O. Popp J. Novel workflow for combining Raman spectroscopy and MALDI-MSI for tissue based studies Anal. Bioanal. Chem. 407 26 2015 7865 7873 10.1007/s00216-015-8987-5 26374565 Böhme S. Stärk H.J. Kühnel D. Reemtsma T. Exploring LA-ICP-MS as a quantitative imaging technique to study nanoparticle uptake in Daphnia magna and zebrafish (Danio rerio) embryos Anal. Bioanal. Chem. 407 18 2015 5477 5485 10.1007/s00216-015-8720-4 25943260 Caprioli R.M. Farmer T.B. Gile J. Molecular imaging of biological Samples: localization of peptides and proteins using MALDI-TOF MS Anal. Chem. 69 23 1997 4751 4760 10.1021/ac970888i 9406525 Cha S. Zhang H. Ilarslan H.I. Wurtele E.S. Brachova L. Nikolau B.J. Yeung E.S. Direct profiling and imaging of plant metabolites in intact tissues by using colloidal graphite-assisted laser desorption ionization mass spectrometry Plant J. 55 2 2008 348 360 10.1111/j.1365-313X.2008.03507.x 18397372 Chaurand P. Latham J.C. Lane K.B. Mobley J.A. Polosukhin V.V. Wirth P.S. Nanney L.B. Caprioli R.M. Imaging mass spectrometry of intact proteins from alcohol-preserved tissue specimens: bypassing formalin fixation J. Proteome Res. 7 8 2008 3543 3555 10.1021/pr800286z 18613713 Chen Y. Li L. Xu R. Li F. Gu L. Liu H. Wang Z. Yang L. Characterization of natural herbal medicines by thin-layer chromatography combined with laser ablation-assisted direct analysis in real-time mass spectrometry J. Chromatogr. A 1654 2021 462461 10.1016/j.chroma.2021.462461 34438305 de Souza L.P. Borghi M. Fernie A. Plant single-cell metabolomics-challenges and perspectives Int. J. Mol. Sci. 21 23 2020 8987 10.3390/ijms21238987 33256100 Desbenoit N. Schmitz-Afonso I. Baudouin C. Laprévote O. Touboul D. Brignole-Baudouin F. Brunelle A. Localisation and quantification of benzalkonium chloride in eye tissue by TOF-SIMS imaging and liquid chromatography mass spectrometry Anal. Bioanal. Chem. 405 12 2013 4039 4049 10.1007/s00216-013-6811-7 23430186 Desbenoit N. Walch A. Spengler B. Brunelle A. Römpp A. Correlative mass spectrometry imaging, applying time-of-flight secondary ion mass spectrometry and atmospheric pressure matrix-assisted laser desorption/ionization to a single tissue section Rapid Commun. Mass Spectrom. : RCM (Rapid Commun. Mass Spectrom.) 32 2 2018 159 166 10.1002/rcm.8022 29105220 Dong Y. Li B. Malitsky S. Rogachev I. Aharoni A. Kaftan F. Svatoš A. Franceschi P. Sample preparation for mass spectrometry imaging of plant tissues: a review Front. Plant Sci. 7 60 2016 10.3389/fpls.2016.00060 Dreisbach D. Petschenka G. Spengler B. Bhandari D.R. 3D-surface MALDI mass spectrometry imaging for visualising plant defensive cardiac glycosides in Asclepias curassavica Anal. Bioanal. Chem. 413 8 2021 2125 2134 10.1007/s00216-021-03177-y 33544161 Eckelmann D. Kusari S. Spiteller M. Occurrence and spatial distribution of maytansinoids in Putterlickia pyracantha, an unexplored resource of anticancer compounds Fitoterapia 113 2016 175 181 10.1016/j.fitote.2016.08.006 27521896 Enomoto H. Mass spectrometry imaging of flavonols and ellagic acid glycosides in ripe strawberry fruit Molecules 25 20 2020 10.3390/molecules25204600 Enomoto H. Nirasawa T. Localization of flavan-3-ol species in peanut testa by mass spectrometry imaging Molecules 25 10 2020 2373 10.3390/molecules25102373 32443878 Feng B. Zhang J. Chang C. Li L. Li M. Xiong X. Guo C. Tang F. Bai Y. Liu H. Ambient mass spectrometry imaging: plasma assisted laser desorption ionization mass spectrometry imaging and its applications Anal. Chem. 86 9 2014 4164 4169 10.1021/ac403310k 24670045 Ganesh S. Hu T. Woods E. Allam M. Cai S. Henderson W. Coskun A.F. Spatially resolved 3D metabolomic profiling in tissues Sci Adv 7 5 2021 eabd0957 10.1126/sciadv.abd0957 Gemperline E. Jayaraman D. Maeda J. Ané J.-M. Li L. Multifaceted investigation of metabolites during nitrogen fixation in Medicago via high resolution MALDI-MS imaging and ESI-MS J. Am. Soc. Mass Spectrom. 26 1 2014 149 158 10.1007/s13361-014-1010-0 25323862 Gessel M.M. Norris J.L. Caprioli R.M. MALDI imaging mass spectrometry: spatial molecular analysis to enable a new age of discovery J Proteomics 107 2014 71 82 10.1016/j.jprot.2014.03.021 24686089 Gorzolka K. Bednarz H. Niehaus K. Detection and localization of novel hordatine-like compounds and glycosylated derivates of hordatines by imaging mass spectrometry of barley seeds Planta 239 6 2014 1321 1335 10.1007/s00425-014-2061-y 24671626 Grassl J. Taylor N.L. Millar A. Matrix-assisted laser desorption/ionisation mass spectrometry imaging and its development for plant protein imaging Plant Methods 7 1 2011 21 10.1186/1746-4811-7-21 21726462 Guo D.A. Quality marker concept inspires the quality research of traditional Chinese medicines Chin. Herbal Med. 9 1 2017 1 2 10.1016/S1674-6384(17)60069-8 Han Y. Sun H. Zhang A. Yan G. Wang X.J. Chinmedomics, a new strategy for evaluating the therapeutic efficacy of herbal medicines Pharmacol. Ther. 216 2020 107680 10.1016/j.pharmthera.2020.107680 32956722 Handberg E. Chingin K. Wang N. Dai X. Chen H. Mass spectrometry imaging for visualizing organic analytes in food Mass Spectrom. Rev. 34 6 2015 641 658 10.1002/mas.21424 24687728 He H. Qin L. Zhang Y. Han M. Li J. Liu Y. Qiu K. Dai X. Li Y. Zeng M. Guo H. Zhou Y. Wang X. 3,4-Dimethoxycinnamic acid as a novel matrix for enhanced in situ detection and imaging of low-molecular-weight compounds in biological tissues by MALDI-MSI Anal. Chem. 91 4 2019 2634 2643 10.1021/acs.analchem.8b03522 30636403 He J. Huang L. Tian R. Li T. Sun C. Song X. Lv Y. Luo Z. Li X. Abliz Z. MassImager: a software for interactive and in-depth analysis of mass spectrometry imaging data Anal. Chim. Acta 1015 2018 50 57 10.1016/j.aca.2018.02.030 29530251 He J. Luo Z. Huang L. He J. Chen Y. Rong X. Jia S. Tang F. Wang X. Zhang R. Zhang J. Shi J. Abliz Z. Ambient mass spectrometry imaging metabolomics method provides novel insights into the action mechanism of drug candidates Anal. Chem. 87 10 2015 5372 5379 10.1021/acs.analchem.5b00680 25874739 He Y. Guo W. Luo K. Sun Q. Lin Z. Cai Z. Poly-l-lysine-based tissue embedding compatible with matrix-assisted laser desorption ionization-mass spectrometry imaging analysis of dry and fragile aristolochia plants J. Chromatogr. A 1608 2019 460389 10.1016/j.chroma.2019.460389 31378528 Hemalatha R.G. Pradeep T. Understanding the molecular signatures in leaves and flowers by desorption electrospray ionization mass spectrometry (DESI MS) imaging J. Agric. Food Chem. 61 31 2013 7477 7487 10.1021/jf4011998 23848451 Heyman H.M. Dubery I.A. The potential of mass spectrometry imaging in plant metabolomics: a review Phytochemistry Rev. 15 2 2016 297 316 10.1007/s11101-015-9416-2 Huang Y. Wu Z. Su R. Ruan G. Du F. Li G. Current application of chemometrics in traditional Chinese herbal medicine research J. Chromatogr. B 1026 2016 27 35 10.1016/j.jchromb.2015.12.050 Ifa D.R. Wiseman J.M. Song Q. Cooks R.G. Development of capabilities for imaging mass spectrometry under ambient conditions with desorption electrospray ionization (DESI) Int. J. Mass Spectrom. 259 1 2007 8 15 10.1016/j.ijms.2006.08.003 Jiang Y. David B. Tu P. Barbin Y. Recent analytical approaches in quality control of traditional Chinese medicines—a review Anal. Chim. Acta 657 1 2010 9 18 10.1016/j.aca.2009.10.024 19951752 Jovanović M. Peter-Katalinić J. Negative ion MALDI-TOF MS, ISD and PSD of neutral underivatized oligosaccharides without anionic dopant strategies, using 2,5-DHAP as a matrix J. Mass Spectrom. 51 2 2016 111 122 10.1002/jms.3727 26889927 Kaletaş B.K. van der Wiel I.M. Stauber J. Dekker L.J. Güzel C. Kros J.M. Luider T.M. Heeren R.M.A. Sample preparation issues for tissue imaging by imaging MS Proteomics 9 10 2009 2622 2633 10.1002/pmic.200800364 19415667 Kaspar S. Peukert M. Svatos A. Matros A. Mock H.P. MALDI-imaging mass spectrometry - an emerging technique in plant biology Proteomics 11 9 2011 1840 1850 10.1002/pmic.201000756 21462348 Knochenmuss R. Ion formation mechanisms in UV-MALDI Analyst 131 9 2006 966 986 10.1039/b605646f 17047796 Koestler M. Kirsch D. Hester A. Leisner A. Guenther S. Spengler B. A high-resolution scanning microprobe matrix-assisted laser desorption/ionization ion source for imaging analysis on an ion trap/Fourier transform ion cyclotron resonance mass spectrometer Rapid Commun. Mass Spectrom. : RCM (Rapid Commun. Mass Spectrom.) 22 20 2008 3275 3285 10.1002/rcm.3733 18819119 Korte A.R. Lee Y.J. MALDI-MS analysis and imaging of small molecule metabolites with 1,5-diaminonaphthalene (DAN) J. Mass Spectrom. 49 8 2014 737 741 10.1002/jms.3400 25044901 Korte A.R. Yagnik G.B. Feenstra A.D. Lee Y.J. Multiplex MALDI-MS imaging of plant metabolites using a hybrid MS system Methods Mol. Biol. 1203 2015 49 62 10.1007/978-1-4939-1357-2_6 25361666 Lange B.M. Fischedick J.T. Lange M.F. Srividya N. Šamec D. Poirier B.C. Integrative approaches for the identification and localization of specialized metabolites in Tripterygium roots Plant Physiol. 173 1 2017 456 469 10.1104/pp.15.01593 27864443 Lee J.W. Ji S.-H. Lee Y.-S. Choi D.J. Choi B.-R. Kim G.-S. Baek N.-I. Lee D.Y. Mass spectrometry based profiling and imaging of various ginsenosides from Panax ginseng roots at different ages Int. J. Mol. Sci. 18 6 2017 1114 10.3390/ijms18061114 28538661 Lemaire R. Tabet J.C. Ducoroy P. Hendra J.B. Salzet M. Fournier I. Solid ionic matrixes for direct tissue analysis and MALDI imaging Anal. Chem. 78 3 2006 809 819 10.1021/ac0514669 16448055 Li B. Bhandari D.R. Janfelt C. Römpp A. Spengler B. Natural products in Glycyrrhiza glabra (licorice) rhizome imaged at the cellular level by atmospheric pressure matrix-assisted laser desorption/ionization tandem mass spectrometry imaging Plant J. 80 1 2014 161 171 10.1111/tpj.12608 25040821 Li B. Bhandari D.R. Römpp A. Spengler B. High-resolution MALDI mass spectrometry imaging of gallotannins and monoterpene glucosides in the root of Paeonia lactiflora Sci. Rep. 6 2016 10.1038/srep36074 36074-36074 Li S. Zhang Y. Liu J.a. Han J. Guan M. Yang H. Lin Y. Xiong S. Zhao Z. Electrospray deposition device used to precisely control the matrix crystal to improve the performance of MALDI MSI Sci. Rep. 6 1 2016 37903 10.1038/srep37903 27885266 Li K. Wang W. Liu Y. Jiang S. Huang G. Ye L. Near-infrared spectroscopy as a process analytical technology tool for monitoring the parching process of traditional Chinese medicine based on two kinds of chemical indicators Phcog. Mag. 13 50 2017 332 337 10.4103/pm.pm_416_16 28539730 Li M. Wang X. Han L. Jia L. Liu E. Li Z. Yu H. Wang Y. Gao X. Yang W. Integration of multicomponent characterization, untargeted metabolomics and mass spectrometry imaging to unveil the holistic chemical transformations and key markers associated with wine steaming of Ligustri Lucidi Fructus J. Chromatogr. A 1624 2020 461228 10.1016/j.chroma.2020.461228 32540070 Li S. Zhu N. Tang C. Duan H. Wang Y. Zhao G. Liu J. Ye Y. Differential distribution of characteristic constituents in root, stem and leaf tissues of Salvia miltiorrhiza using MALDI mass spectrometry imaging Fitoterapia 146 2020 104679 10.1016/j.fitote.2020.104679 32619463 Li B. Ge J. Liu W. Hu D. Li P. Unveiling spatial metabolome of Paeonia Suffruticosa and Paeonia Lactiflora roots using MALDI MS imaging New Phytol. 231 2 2021 892 902 10.1111/nph.17393 33864691 Liao Y. Fu X. Zhou H. Rao W. Zeng L. Yang Z. Visualized analysis of within-tissue spatial distribution of specialized metabolites in tea (Camellia sinensis) using desorption electrospray ionization imaging mass spectrometry Food Chem. 292 2019 204 210 10.1016/j.foodchem.2019.04.055 31054666 Liu C. Chen S. Xiao X. Zhang T. Hou W. Liao M. A new concept on quality marker of Chinese materia medica: quality control for Chinese medicinal products Chin. Tradit. Herb. Drugs 47 9 2016 1443 1457 10.7501/j.issn.0253-2670.2016.09.001 Liu C. Guo D.-a. Liu L. Quality transitivity and traceability system of herbal medicine products based on quality markers Phytomedicine : international journal of phytotherapy and phytopharmacology 44 2018 247 257 10.1016/j.phymed.2018.03.006 29631807 Liu H. Zhou Y. Wang J. Xiong C. Xue J. Zhan L. Nie Z. N-Phenyl-2-naphthylamine as a novel MALDI matrix for analysis and in situ imaging of small molecules Anal. Chem. 90 1 2018 729 736 10.1021/acs.analchem.7b02710 29172460 Meng X. Fu W. Huo M. Liu Y. Zhang Z. Wei J. Wang Z. Abliz Z. In situ label-free visualization of tissue distributions of salidroside in multiple mouse organs by MALDI-MS imaging Int. J. Mass Spectrom. 453 2020 116347 10.1016/j.ijms.2020.116347 Merdas M. Lagarrigue M. Vanbellingen Q. Umbdenstock T. Da Violante G. Pineau C. On-tissue chemical derivatization reagents for matrix-assisted laser desorption/ionization mass spectrometry imaging J. Mass Spectrom. 2021 10.1002/jms.4731 JMS, e4731 Morikawa-Ichinose T. Fujimura Y. Murayama F. Yamazaki Y. Yamamoto T. Wariishi H. Miura D. Improvement of sensitivity and reproducibility for imaging of endogenous metabolites by matrix-assisted laser desorption/ionization-mass spectrometry J. Am. Soc. Mass Spectrom. 30 8 2019 1512 1520 10.1007/s13361-019-02221-7 31044355 Morisasa M. Sato T. Kimura K. Mori T. Goto-Inoue N. Application of matrix-assisted laser desorption/ionization mass spectrometry imaging for food analysis Foods 8 12 2019 10.3390/foods8120633 Mukrimin M. Conrad A.O. Kovalchuk A. Julkunen-Tiitto R. Bonello P. Asiegbu F.O. Fourier-transform infrared (FT-IR) spectroscopy analysis discriminates asymptomatic and symptomatic Norway spruce trees Plant Sci. : an international journal of experimental plant biology 289 2019 110247 10.1016/j.plantsci.2019.110247 Nilsson A. Forngren B. Bjurström S. Goodwin R.J. Basmaci E. Gustafsson I. Annas A. Hellgren D. Svanhagen A. Andrén P.E. Lindberg J. In situ mass spectrometry imaging and ex vivo characterization of renal crystalline deposits induced in multiple preclinical drug toxicology studies PLoS One 7 10 2012 e47353 10.1371/journal.pone.0047353 Nimesh S. Mohottalage S. Vincent R. Kumarathasan P. Current status and future perspectives of mass spectrometry imaging Int. J. Mol. Sci. 14 6 2013 11277 11301 10.3390/ijms140611277 23759983 Oppenheimer S.R. Drexler D.M. Tissue analysis by imaging MS Bioanalysis 4 1 2011 95 112 10.4155/bio.11.282 Parrot D. Papazian S. Foil D. Tasdemir D. Imaging the unimaginable: desorption electrospray ionization - imaging mass spectrometry (DESI-IMS) in natural product research Planta Med. 84 9–10 2018 584 593 10.1055/s-0044-100188 29388184 Porta Siegel T. Hamm G. Bunch J. Cappell J. Fletcher J.S. Schwamborn K. Mass spectrometry imaging and integration with other imaging modalities for greater molecular understanding of biological tissues Mol. Imag. Biol. 20 6 2018 888 901 10.1007/s11307-018-1267-y Prideaux B. Stoeckli M. Mass spectrometry imaging for drug distribution studies J Proteomics 75 16 2012 4999 5013 10.1016/j.jprot.2012.07.028 22842290 Qin L. Zhang Y. Liu Y. He H. Han M. Li Y. Zeng M. Wang X. Recent advances in matrix-assisted laser desorption/ionisation mass spectrometry imaging (MALDI-MSI) for in situ analysis of endogenous molecules in plants Phytochem. Anal. 29 4 2018 351 364 10.1002/pca.2759 29667236 Ràfols P. Heijs B. Del Castillo E. Yanes O. McDonnell L.A. Brezmes J. Pérez-Taboada I. Vallejo M. García-Altares M. Correig X. rMSIproc: an R package for mass spectrometry imaging data processing Bioinformatics 36 11 2020 3618 3619 10.1093/bioinformatics/btaa142 32108859 Ren J.-l. Zhang A.-H. Kong L. Han Y. Yan G.-L. Sun H. Wang X.-J. Analytical strategies for the discovery and validation of quality-markers of traditional Chinese medicine Phytomedicine : international journal of phytotherapy and phytopharmacology 67 2020 153165 10.1016/j.phymed.2019.153165 31954259 Römpp A. Spengler B. Mass spectrometry imaging with high resolution in mass and space Histochem. Cell Biol. 139 6 2013 759 783 10.1007/s00418-013-1097-6 23652571 Sagara T. Bhandari D.R. Spengler B. Vollmann J. Spermidine and other functional phytochemicals in soybean seeds: spatial distribution as visualized by mass spectrometry imaging Food Sci. Nutr. 8 1 2019 675 682 10.1002/fsn3.1356 31993191 Saito K. Watanabe Y. Shirakawa M. Matsushita Y. Imai T. Koike T. Sano Y. Funada R. Fukazawa K. Fukushima K. Direct mapping of morphological distribution of syringyl and guaiacyl lignin in the xylem of maple by time-of-flight secondary ion mass spectrometry Plant J. 69 3 2012 542 552 10.1111/j.1365-313X.2011.04811.x 21978273 Schulz S. Becker M. Groseclose M.R. Schadt S. Hopf C. Advanced MALDI mass spectrometry imaging in pharmaceutical research and drug development Curr. Opin. Biotechnol. 55 2019 51 59 10.1016/j.copbio.2018.08.003 30153614 Schwamborn K. Caprioli R.M. Molecular imaging by mass spectrometry — looking beyond classical histology Nat. Rev. Cancer 10 9 2010 639 646 10.1038/nrc2917 20720571 Shi J. Cao B. Wang X.W. Aa J.Y. Duan J.A. Zhu X.X. Wang G.J. Liu C.X. Metabolomics and its application to the evaluation of the efficacy and toxicity of traditional Chinese herb medicines Journal of chromatography. B, Analytical technologies in the biomedical and life sciences 1026 2016 204 216 10.1016/j.jchromb.2015.10.014 26657802 Shi R. Dai X. Li W. Lu F. Liu Y. Qu H. Li H. Chen Q. Tian H. Wu E. Wang Y. Zhou R. Lee S.-T. Lifshitz Y. Kang Z. Liu J. Hydroxyl-group-dominated graphite dots reshape laser desorption/ionization mass spectrometry for small biomolecular analysis and imaging ACS Nano 11 9 2017 9500 9513 10.1021/acsnano.7b05328 28850220 Song X. Luo Z. Li X. Li T. Wang Z. Sun C. Huang L. Xie P. Liu X. He J. Abliz Z. In situ hydrogel conditioning of tissue samples to enhance the drug's sensitivity in ambient mass spectrometry imaging Anal. Chem. 89 12 2017 6318 6323 10.1021/acs.analchem.7b00091 28517936 Sturtevant D. Lee Y.-J. Chapman K.D. Matrix assisted laser desorption/ionization-mass spectrometry imaging (MALDI-MSI) for direct visualization of plant metabolites in situ Curr. Opin. Biotechnol. 37 2016 53 60 10.1016/j.copbio.2015.10.004 26613199 Sun C. Liu W. Ma S. Zhang M. Geng Y. Wang X. Development of a high-coverage matrix-assisted laser desorption/ionization mass spectrometry imaging method for visualizing the spatial dynamics of functional metabolites in Salvia miltiorrhiza Bge J. Chromatogr. A 1614 2020 460704 10.1016/j.chroma.2019.460704 31753480 Sun H. Zhang A. Wang X. Potential role of metabolomic approaches for Chinese medicine syndromes and herbal medicine Phytother Res. : PT 26 10 2012 1466 1471 10.1002/ptr.4613 Taira S. Ikeda R. Yokota N. Osaka I. Sakamoto M. Kato M. Sahashi Y. Mass spectrometric imaging of ginsenosides localization in Panax ginseng root Am. J. Chin. Med. 38 3 2010 485 493 10.1142/s0192415x10008007 20503467 Takahashi K. Kozuka T. Anegawa A. Nagatani A. Mimura T. Development and application of a high-resolution imaging mass spectrometer for the study of plant tissues Plant Cell Physiol. 56 7 2015 1329 1338 10.1093/pcp/pcv083 26063395 Takáts Z. Wiseman J.M. Gologan B. Cooks R.G. Mass spectrometry sampling under ambient conditions with desorption electrospray ionization Science 306 5695 2004 471 10.1126/science.1104404 15486296 Talaty N. Takáts Z. Cooks R.G. Rapid in situ detection of alkaloids in plant tissue under ambient conditions using desorption electrospray ionization Analyst 130 12 2005 1624 1633 10.1039/b511161g 16284661 Thunig J. Hansen S.H. Janfelt C. Analysis of secondary plant metabolites by indirect desorption electrospray ionization imaging mass spectrometry Anal. Chem. 83 9 2011 3256 3259 10.1021/ac2004967 21473636 Tian F. Liu R. Fan C. Sun Y. Huang X. Nie Z. Zhao X. Pu X. Effects of Thymoquinone on small-molecule metabolites in a rat model of cerebral ischemia reperfusion injury assessed using MALDI-MSI Metabolites 10 1 2020 27 10.3390/metabo10010027 31936061 Tian X. Xie B. Zou Z. Jiao Y. Lin L.-E. Chen C.-L. Hsu C.-C. Peng J. Yang Z. Multimodal imaging of amyloid plaques: fusion of the single-probe mass spectrometry image and fluorescence microscopy image Anal. Chem. 91 20 2019 12882 12889 10.1021/acs.analchem.9b02792 31536324 Tillner J. Wu V. Jones E.A. Pringle S.D. Karancsi T. Dannhorn A. Veselkov K. McKenzie J.S. Takats Z. Faster, more reproducible DESI-MS for biological tissue imaging J. Am. Soc. Mass Spectrom. 28 10 2017 2090 2098 10.1007/s13361-017-1714-z 28620847 Trim P.J. Henson C.M. Avery J.L. McEwen A. Snel M.F. Claude E. Marshall P.S. West A. Princivalle A.P. Clench M.R. Matrix-assisted laser desorption/ionization-ion mobility separation-mass spectrometry imaging of vinblastine in whole body tissue sections Anal. Chem. 80 22 2008 8628 8634 10.1021/ac8015467 18847214 Van de Plas R. Yang J. Spraggins J. Caprioli R.M. Image fusion of mass spectrometry and microscopy: a multimodality paradigm for molecular tissue mapping Nat. Methods 12 4 2015 366 372 10.1038/nmeth.3296 25707028 Végvári Á. Drug localizations in tissue by mass spectrometry imaging Biomarkers Med. 9 9 2015 869 876 10.2217/bmm.15.64 Verbeeck N. Spraggins J.M. Murphy M.J.M. Wang H.D. Deutch A.Y. Caprioli R.M. Van de Plas R. Connecting imaging mass spectrometry and magnetic resonance imaging-based anatomical atlases for automated anatomical interpretation and differential analysis Biochim. Biophys. Acta Protein Proteonomics 1865 7 2017 967 977 10.1016/j.bbapap.2017.02.016 Vestal M. Vestal C. Li S. Parker K. The seven S criteria for evaluating the performance of a MALDI mass spectrometer for MSI J. Am. Soc. Mass Spectrom. 31 12 2020 2521 2530 10.1021/jasms.0c00216 32877189 Wang J. van der Heijden R. Spijksma G. Reijmers T. Wang M. Xu G. Hankemeier T. van der Greef J. Alkaloid profiling of the Chinese herbal medicine Fuzi by combination of matrix-assisted laser desorption ionization mass spectrometry with liquid chromatography–mass spectrometry J. Chromatogr. A 1216 11 2009 2169 2178 10.1016/j.chroma.2008.11.077 19095240 Wang X. Han J. Pan J. Borchers C.H. Comprehensive imaging of porcine adrenal gland lipids by MALDI-FTMS using quercetin as a matrix Anal. Chem. 86 1 2014 638 646 10.1021/ac404044k 24341451 Wang X. Han J. Yang J. Pan J. Borchers C.H. Matrix coating assisted by an electric field (MCAEF) for enhanced tissue imaging by MALDI-MS Chem. Sci. 6 1 2015 729 738 10.1039/c4sc01850h 28706636 Wang S. Bai H. Cai Z. Gao D. Jiang Y. Liu J. Liu H. MALDI imaging for the localization of saponins in root tissues and rapid differentiation of three Panax herbs Electrophoresis 37 13 2016 1956 1966 10.1002/elps.201600027 26990111 Wang Z. Cai Y. Wang Y. Zhou X. Zhang Y. Lu H. Improved MALDI imaging MS analysis of phospholipids using graphene oxide as new matrix Sci. Rep. 7 2017 44466 10.1038/srep44466 28294158 Wang G. Hao R. Liu Y. Wang Y. Man S. Gao W. Tissue distribution, metabolism and absorption of Rhizoma Paridis Saponins in the rats J. Ethnopharmacol. 273 2021 114038 10.1016/j.jep.2021.114038 33746004 Wang L. Zhu T. Xu H.-B. Pu X.-P. Zhao X. Tian F. Ding T. Sun G.-B. Sun X.-B. Effects of notoginseng leaf triterpenes on small molecule metabolism after cerebral ischemia/reperfusion injury assessed using MALDI-MS imaging Ann. Transl. Med. 9 3 2021 10.21037/atm-20-4898 246-246 Wang T. Lee H.K. Yue G.G.L. Chung A.C.K. Lau C.B.S. Cai Z. A novel binary matrix consisting of graphene oxide and caffeic acid for the analysis of scutellarin and its metabolites in mouse kidney by MALDI imaging Analyst 146 1 2021 289 295 10.1039/d0an01539c 33140762 Wang T. Liu J. Luo X. Hu L. Lu H. Functional metabolomics innovates therapeutic discovery of traditional Chinese medicine derived functional compounds Pharmacol. Ther. 224 2021 107824 10.1016/j.pharmthera.2021.107824 33667524 Watson B. Urbanczyk-Wochniak E. Lei Z. Spatially Resolved Metabolomics and Proteomics of Medicago Truncatula Border Cells and Root Tips 2007 Joint Annual Meeting of the American-Fern-Society-American-Society-of-Plant-Biologists/American-Society-of-Plant-Taxonomists/Botanical-Society-of-America Wiseman J.M. Ifa D.R. Song Q. Cooks R.G. Tissue imaging at atmospheric pressure using desorption electrospray ionization (DESI) mass spectrometry Angew. Chem. Int. Ed. 45 43 2006 7188 7192 10.1002/anie.200602449 Wu H. Liu X. Gao Z.-Y. Dai Z.-F. Lin M. Tian F. Zhao X. Sun Y. Pu X.-P. Anti-myocardial infarction effects of Radix Aconiti Lateralis Preparata extracts and their influence on small molecules in the heart using matrix-assisted laser desorption/ionization-mass spectrometry imaging Int. J. Mol. Sci. 20 19 2019 4837 10.3390/ijms20194837 31569464 Wu W. Liang Z. Zhao Z. Cai Z. Direct analysis of alkaloid profiling in plant tissue by using matrix-assisted laser desorption/ionization mass spectrometry J. Mass Spectrom. 42 1 2007 58 69 10.1002/jms.1138 17149797 Yamamoto K. Takahashi K. Caputi L. Mizuno H. Rodriguez-Lopez C.E. Iwasaki T. Ishizaki K. Fukaki H. Ohnishi M. Yamazaki M. Masujima T. O'Connor S.E. Mimura T. The complexity of intercellular localisation of alkaloids revealed by single-cell metabolomics New Phytol. 224 2 2019 848 859 10.1111/nph.16138 31436868 Yamamoto K. Takahashi K. Mizuno H. Anegawa A. Ishizaki K. Fukaki H. Ohnishi M. Yamazaki M. Masujima T. Mimura T. Cell-specific localization of alkaloids in Catharanthus roseus stem tissue measured with Imaging MS and Single-cell MS Proc. Natl. Acad. Sci. U. S. A 113 14 2016 3891 3896 10.1073/pnas.1521959113 27001858 Yan Y.-T. Li S.-D. Li C. Xiong Y.-X. Lu X.-H. Zhou X.-F. Yang L.-Q. Pu L.-J. Luo H.-Y. Panax notoginsenoside saponins Rb1 regulates the expressions of Akt/mTOR/PTEN signals in the hippocampus after focal cerebral ischemia in rats Behav. Brain Res. 345 2018 83 92 10.1016/j.bbr.2018.02.037 29501622 Yang W. Zhang Y. Wu W. Huang L. Guo D. Liu C. Approaches to establish Q-markers for the quality standards of traditional Chinese medicines Acta Pharm. Sin. B 7 4 2017 439 446 10.1016/j.apsb.2017.04.012 28752028 Yang Y. Yang Y. Qiu H. Ju Z. Shi Y. Wang Z. Yang L. Localization of constituents for determining the age and parts of ginseng through ultraperfomance liquid chromatography quadrupole/time of flight-mass spectrometry combined with desorption electrospray ionization mass spectrometry imaging J. Pharmaceut. Biomed. Anal. 193 2021 113722 10.1016/j.jpba.2020.113722 Yoon S. Lee T.G. Biological tissue sample preparation for time-of-flight secondary ion mass spectrometry (ToF-SIMS) imaging Nano Converg 5 1 2018 10.1186/s40580-018-0157-y 24-24 Zhang T. Bai G. Han Y. Xu J. Gong S. Li Y. Zhang H. Liu C. The method of quality marker research and quality evaluation of traditional Chinese medicine based on drug properties and effect characteristics Phytomedicine : international journal of phytotherapy and phytopharmacology 44 2018 204 211 10.1016/j.phymed.2018.02.009 29551645 Zhang X.J. Qiu J.F. Guo L.P. Wang Y. Li P. Yang F.Q. Su H. Wan J.B. Discrimination of multi-origin Chinese herbal medicines using gas chromatography-mass spectrometry-based fatty acid profiling Molecules 18 12 2013 15329 15343 10.3390/molecules181215329 24335614 Zhao K. Li B. He D. Zhao C. Shi Z. Dong B. Pan D. Patil R.R. Yan Z. Guo Z. Chemical characteristic and bioactivity of hemicellulose-based polysaccharides isolated from Salvia miltiorrhiza Int. J. Biol. Macromol. 165 Pt B 2020 2475 2483 10.1016/j.ijbiomac.2020.10.113 33098893 Zhu T. Wang L. Tian F. Zhao X. Pu X.-P. Sun G.-B. Sun X.-B. Anti-ischemia/reperfusion injury effects of notoginsenoside R1 on small molecule metabolism in rat brain after ischemic stroke as visualized by MALDI–MS imaging Biomed. Pharmacother. 129 2020 110470 10.1016/j.biopha.2020.110470 32768957	34678417	PMC9715987	NO-CC CODE	2022-12-03 23:20:52	no	J Ethnopharmacol. 2022 Feb 10; 284:114760	utf-8	J Ethnopharmacol	2,021	10.1016/j.jep.2021.114760	oa_other
==== Front Respir Physiol Neurobiol Respir Physiol Neurobiol Respiratory Physiology & Neurobiology 1569-9048 1878-1519 Elsevier B.V. S1569-9048(21)00150-6 10.1016/j.resp.2021.103765 103765 Article Does wearing a facemask decrease arterial blood oxygenation and impair exercise tolerance? Ade Carl J. a* Turpin Vanessa-Rose G. a Parr Shannon K. a Hammond Stephen T. a White Zachary a Weber Ramona E. ab Schulze Kiana M. ab Colburn Trenton D. ab Poole David C. ab a Departments of Kinesiology, Kansas State University, Manhattan, KS, 66506, USA b Anatomy and Physiology, Kansas State University, Manhattan, KS, 66506, USA ⁎ Corresponding author at: Department of Kinesiology, College of Health and Human Sciences, Kansas State University, Manhattan, KS, 66506, USA. 3 8 2021 12 2021 3 8 2021 294 103765103765 29 4 2021 12 7 2021 25 7 2021 © 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. Introduction Concerns have been raised that COVID-19 face coverings compromise lung function and pulmonary gas exchange to the extent that they produce arterial hypoxemia and hypercapnia during high intensity exercise resulting in exercise intolerance in recreational exercisers. This study therefore aimed to investigate the effects of a surgical, flannel or vertical-fold N95 masks on cardiorespiratory responses to incremental exercise. Methods This investigation studied 11 adult males and females at rest and while performing progressive cycle exercise to exhaustion. We tested the hypotheses that wearing a surgical (S), flannel (F) or horizontal-fold N95 mask compared to no mask (control) would not promote arterial deoxygenation or exercise intolerance nor alter primary cardiovascular variables during submaximal or maximal exercise. Results Despite the masks significantly increasing end-expired peri-oral %CO2 and reducing %O2, each ∼0.8−2% during exercise (P < 0.05), our results supported the hypotheses. Specifically, none of these masks reduced sub-maximal or maximal exercise arterial O2 saturation (P = 0.744), but ratings of dyspnea were significantly increased (P = 0.007). Moreover, maximal exercise capacity was not compromised nor were there any significant alterations of primary cardiovascular responses (mean arterial pressure, stroke volume, cardiac output) found during sub-maximal exercise. Conclusion Whereas these results are for young healthy recreational male and female exercisers and cannot be applied directly to elite athletes, older or patient populations, they do support that arterial hypoxemia and exercise intolerance are not the obligatory consequences of COVID-19-indicated mask-wearing at least for cycling exercise. Keywords COVID-19 facemask N95 Surgical facemask Cycle exercise Submaximal Maximal Exhaustion Exercise-induced arterial hypoxemia Cardiovascular responses Rating of perceived exertion Dyspnea Edited by M Dutschmann ==== Body pmc1 Introduction Wearing a mask that covers the mouth and nostrils, along with social distancing and frequent handwashing, represents the first line of defense against the spread of COVID-19 (Leung et al., 2020) and WHO guidelines (Organization, 2020)). Moreover, health experts have indicated the potential need for COVID-19 face coverings through 2022 (Crist, 2021). Whereas some countries report that mask use is ∼80 % or higher (Jang et al., 2020) concern have been raised that lung function and pulmonary gas exchange are compromised by mask wearing to the extent that they produce arterial hypoxemia and hypercapnia during high intensity exercise resulting in a decreased peak exercise capacity (Davis and Tsen, 2020; Sinkule et al., 2013). The premise for concerns surrounding arterial hypoxemia and impaired exercise capacity have largely been based on earlier, pre−COVID-19 reports using a specialized breathing apparatus, which may not be an appropriate surrogate for the types of face coverings used against COVID-19. For instance, the work of Schulte (1964), which evaluated the effects of a respiratory muscle training device that purposefully increases breathing resistance and can reduce effective inspired O2 to 17 % (from ∼20.94 %), is often referenced when discussing the potential effects of N95 facemasks (Davis and Tsen, 2020; Schulte, 1964). In contrast to COVID-19 recommended masks, the respiratory muscle training masks are specifically designed so that the wearer “dials in” a very high breathing resistance that is specifically intended to overcome the ability of the respiratory muscles to elevate ventilation sufficiently to regulate alveolar and thus arterial CO2 pressures (PCO2). As such, they induce a state of hypoventilation that drives down alveolar PO2. Therefore statements made during the COVID-19 pandemic stating that the standard N95 mask reduces the inspired O2 causing “…headache, lightheadedness, drowsiness, muscular weakness, dyspnea on exertion, nausea and vomiting” (Davis and Tsen, 2020; Schulte, 1964) or that the increased inspiratory and expiratory resistance for greater than 10 min causes “…increased lactate levels, fatigue and impaired physical work capacity.”(Álvarez-Herms et al., 2018; Davis and Tsen, 2020), based on non-N95 specialized breathing apparatus may not be accurate. Recent work Lassing et al. (2020) and Fikenzer et al. (2020) have revealed significant exercise performance impairments with surgical and N95 masks during exercise (Fikenzer et al., 2020; Lassing et al., 2020). However, these two studies superimposed a spirometry mask on top of each COVID-19 mask, resulting in a relatively leak-proof seal that is at odds with the fitting of COVID-19 masks in real-world setting and thus may not accurately reflect what occurs with normal mask wearing. This is a critical point since the interaction between each face mask and the spirometry valve assemblies required for pulmonary gas exchange measurements will have markedly variant resistances, dead-space and other qualities that impact the subject’s breathing pattern, total ventilation, cardiopulmonary function, gas exchange, and facial skin temperature (Askanazi et al., 1980; Cerretelli et al., 1969; Scarano et al., 2020). This becomes more evident when compared to the recent work by Epstein et al. (2021) which revealed that wearing an N95 facemask, without the additional spirometry mask, during incremental exercise did not alter peak exercise workload, but did increase end-tidal CO2 compared to a non-mask control (Epstein et al., 2021). Using a similar incremental exercise protocol, Shaw et al. (2020) evaluated the effects of surgical and cloth masks alone on arterial oxygen saturation and muscle tissue oxygenation levels (Shaw et al., 2020). They revealed that wearing a mask had no effect on exercise performance, arterial oxygen saturation, tissue oxygenation index. These initial formative studies have provided some of the first evidence regarding the physiological response to exercise with COVID-19 face coverings, but given the conflicting findings and controversy surrounding this topic (Fikenzer and Laufs, 2020a, b), highlights the need for additional work. Moreover, given that increase in the work of breathing, which may occur with mask wearing, can alter cardiovascular responses to exercise (Harms et al., 1997), evaluation of the cardiac responses to different COVID-19 masks is warranted. Therefore, we tested two hypotheses. Namely, for recreational exercisers, that despite modest effects on “expired” O2 (decrease) and CO2 (increase), wearing a surgical, flannel or vertical-fold N95 mask would: 1. Have little or no impact on arterial oxygenation (i.e., <3% decrease in arterial O2 saturation from rest, (Dominelli et al., 2013; Harms et al., 2000)) or exercise tolerance. 2. Not alter primary cardiovascular variables (heart rate, cardiac output, blood pressure) during submaximal or maximal exercise. In addition to these hypotheses, the present investigation experimentally tested the resistance to flow across each mask at physiologic flow rates. 2 Methods 2.1 Participants Eleven healthy, recreationally active participants (5 men and 6 women, age 30 ± 11 yrs [mean ± SD], height 175 ± 11 cm, body mass 73.0 ± 12.9 kg) who were experienced with laboratory exercise testing and maximal exercise tests completed the experiments. Sample size was estimated using a population mean ± SD of 98 ± 2 and 100 ± 5 for percentages of SpO2 and peak power, with α = 0.05 and β = 0.1, to detect a 3% and 6% decrease, respectively (Rosner, 2011). The number of subjects required to show a 3% decrease in SpO2 = 7 and 6% decrease in peak power = 16. Importantly, a 3% decrease in SpO2 has been defined in the literature as the definition of exercise-induced arterial hypoxemia (Harms et al., 2000). Participants were free from known cardiovascular, pulmonary, or metabolic disease and were non-smokers as determined from a health history questionnaire. All procedures were approved by the Institutional Review Board of Kansas State University (#9954) and conformed to the standards set by the Declaration of Helsinki. Written informed consent was obtained from all participants. Subjects were instructed to abstain from vigorous activity 12 h. prior and caffeine and food consumption 2 h. prior to the scheduled testing times. A minimum of 24 h. was mandated between each test with all test performed at a similar time of day (±3 h). 2.2 Facemasks Surgical mask (USA ASTM F2100): Size: 3.7 × 6.9 in.(9.5 × 17.5 cm), mass: 3.2 g. Dead space: depends on wearer and fitting, estimated <40 mL. Kunshan jiehong (Kunshan City, China) disposable 3-layer face masks feature an inner and outer layer of spun-bound polypropylene, and a middle filter layer made of melt-blown polypropylene. These materials are the industry standard for disposable 3-ply facemasks. They provide level 1 protection with filtration >95 % for 3 μm and 0.1 μm particles. Pore size is 0.1 μm. Flannel facemask: Champion (Rural Hall, North Carolina, 50 % cotton, 50 % polyester, 2 layers, ∼30 stitches/in). Size: 5.5 × 8.5 in.(14 × 19 cm), mass: 17.3 g. Dead space: depends on wearer and fitting, estimated <40 mL. Authorized by the Federal Drug Administration (not FDA cleared or approved) under an Emergency Use Authorization for use by health care professionals as personal protective equipment under section 564(b)(1) of the Act, 21 U.S.C. Section 360bbb-3(b)(1). N95 NIOSH approved 1570 respirator: Horizontal-fold, non-valved. Size: 7 × 7.75 in.(19 × 20 cm), mass 10.0 g. Dead space: depends on wearer and face shape (140 mL + 15 for subjects herein, measured by water displacement). Dukal Corp (Ronkonkoma, New York). Provides >95 % filter efficiency for particulate oil-free aerosols 0.3 μm. Bacterial filtration efficiency 99.9 %. By comparison, the dead space for the standard Hans-Rudolph (adult – large) two-way non-rebreathing valve and mask assembly is 123.5 mL (https://www.rudolphkc.com/pdf/691151%201215%20K.pdf). The resistance to flow across each mask material was determined at constant flow rates of 24, 48, 72 and 96 L min−1 through a Medgraphics pneumatachograph, which allowed for real-time evaluation of flow (Fig. 1 A). Measurements of the differential pressure on the down- and up-stream side of each mask were taken using a pressure manometer and used to calculate resistance.Fig. 1 Illustration of experimental set-up for evaluation of the resistance to flow across each mask (A). The pressure difference across each mask for multiple flow rates demonstrates that a substantial difference exists for all masked conditions compared to control, with the greatest difference for the N95 mask (B). Calculation of resistance at 48 L min−1 revealed a higher resistance to air flow for all masks (C). Fig. 1 2.3 Study design A randomized cross-over study design was utilized in which participants completed a series of exercise tests while wearing a surgical mask, flannel facemask, and N95 respirator (described below), or no face covering (control). All masks were securely fastened by a trained member of the laboratory according to manufacturers’ specifications and Centers for Disease Control and Prevention (CDC) recommendations. All testing procedures were performed in a temperature-controlled laboratory (21−22 °C), with all participants in a well hydrated state and having abstained from vigorous activity for 24 h prior to testing. At least 48 h was given between adjacent testing sessions. Participants first completed four incremental ramp exercise tests to exhaustion on a cycle ergometer (Lode, Groningen, Netherlands), under each experimental condition. Following a 2 min resting baseline and 2 min unloading cycling, the power output was progressively increased at a rate of 20 W/min until the participant could not maintain the pedal cadence of 60 rpm for 5 consecutive revolutions despite verbal encouragement. Pedal cadence was maintained constant because pedal cadence can affect ventilatory, cardiac, and pulmonary gas exchange responses during incremental exercise (Broxterman et al., 2015). Participants were blinded to power output and test duration. Seat height was recorded for the first test and reproduced for all subsequent tests. During each ramp test, arterial O2 saturation (SpO2) was measured continuously via two independent pulse oximetry units (Datex Ohmeda (GE) S/5 Light Patient Monitor and Innovo pulse oximeter, Innovo Medical, Stafford, Texas, U.S.A.). If measurements were more than 3% different the sensors were repositioned. Heart rate was measured via photoplethysmography. Continuous beat-by-beat blood pressure (systolic, diastolic, and mean) was measured via photoplethysmography (Finometer Pro; Finapres Medical Systems, Amsterdam, The Netherlands). To minimize hand movement artifact during exercise, the right arm was placed on foam padding slightly below heart level on an adjustable stand. In addition, the raw arterial pressure waveforms were continuously monitored for any movement artifacts and these were processed with the Modelflow method, incorporating age, height, and weight, to obtain measurements of stroke volume and cardiac output. This method provides a reliable estimate of the relative changes in stroke volume and cardiac output during exercise in healthy men and women but cannot be used to provide absolute values unless corrected to a standard method (Sugawara et al., 2003). As such, only the change in these variables relative to rest is reported. Borg ratings of perceived exertion and dyspnea were recorded at each minute of exercise as previously described (Borg, 1982; Mancini et al., 1992). A 6−20-point Borg dyspnea scale, versus the modified 10-point version, was used to allow for greater fidelity in identifying different levels of perceived dyspnea between conditions. Subjects were well versed in using these scales to assess exertion and dyspnea. Since the interaction between each face mask and the mouthpiece–breathing valve assemblies required for pulmonary gas exchange measurements will have markedly variant resistances, dead-space and other qualities that impact the subject’s breathing pattern, total ventilation, cardiopulmonary function and gas exchange (Askanazi et al., 1980; Cerretelli et al., 1969) as well as facial skin temperature (Scarano et al., 2020), gas exchange measurements were not performed. Compounded by the fact that the devices used to measure ventilatory variables may themselves induce alterations in the measurement of tidal volume, breathing frequency and thus ventilation (Gilbert et al., 1972), a critical aspect of this study is the absence of mouthpiece–breathing valve assembly placed over each respective facemask. Since the primary goal of this work is to provide real-world ecologically valid results that translate to individuals wearing only a facemask, this represents a key aspect of the study design. Peak exercise workload was used as the primary variable for exercise capacity. Peri-oral End-Expired CO2 and O2. In a subset of individuals (n = 5, 3 F/2 M, Age: 24.8 + 1.6 yr, 175.7 + 9.6 cm, 71.7 + 13.4 kg) additional sub-maximal exercise tests were performed to evaluate changes in expired peri-oral CO2 and O2. Experiments could only be performed in a sub-set due to equipment availability. Within 1 week of completing the ramp tests, each participant returned to the laboratory to perform measurements at rest and during ∼3 min of constant-load cycling exercise at 95 W and 127 W (Monark, Ergomedic 828E, Varberg, Sweden) under each experimental condition whilst wearing a nose clip. The order of these tests was randomized. These work rates were chosen as they are associated with MET ranges of 5–7 METs that are common for moderate-to-heavy physical activities (Ainsworth et al., 2011). The short duration was selected purposefully to capture the primary component and avoid development of a slow component in those subjects who may have been > GET at the higher work load. This strategy has been used previously (Poole et al., 1992). During each test, HR and arterial O2 saturation were measured by pulse oximetry (Proven, OXI-27BL, Beaverton, Oregon) while expired O2 and CO2 were measured using Ametek Applied Electrochemistry Inc., CD-3A and S-3A/I analyzers (Oak Ridge, Tennessee) calibrated using precision gasses that spanned the expected range of measured values. These analyzers are accurate to within +/-0.01 % (O2) and 0.02 % (CO2) with response times of 100 ms for O2 and 25 ms for CO2 to 90 % final response with a sensitivity of 0.001 % over the ranges measured. For the constant-load exercise tests performed herein breath-to-breath variation is typically <10 % of the room air to peri-oral end-expired value. With 10 breaths averaged to provide final values as presented the coefficient of variation (CV) on repeated analyses within condition (i.e., control or masked) was 5% (O2) and 4% (CO2) of the delta from inspired to peri-oral end-tidal giving absolute values for CV of 0.025 % (O2) and 0.016 % (CO2). Breathing frequency was determined by timed (30 s interval) observation. To measure the expired gas concentrations exhaled “mask” gas was sampled between the subject’s chin and lower lip at 0.5 L/min and end-expired CO2 was detected automatically and the corresponding end-expired O2 recorded at that precise time. This procedure collected principally, but not exclusively, the expired flow stream and could be replicated exactly in the control and mask trials without altering mask geometry or inspiratory/expiratory resistances in any way. We accepted that the expirate would be contaminated to some degree by room air in the control condition by streaming effects and, in the mask trials, by gas trapping and streaming within the additional dead space. The values of 10 breaths beyond 2 min 30 s of exercise were measured and averaged. Preliminary studies confirmed that the primary measurements had stabilized after 2 min 30 s for these subjects. 2.4 Statistical analyses For all statistical analysis, the Prism (version 7.04, Graphpad software, INC., La Jolla, CA) data analysis software package was utilized. The effect of each facemask condition on peak exercise responses (peak power, SpO2, dyspnea, HR) were assessed using one-way repeated measures ANOVA with Dunnet’s post hoc-analysis. Sub-maximal exercise responses were assessed via two-way repeated measures ANOVA (condition × workrate) with Dunnet’s post hoc-analysis. In cases were the assumption of sphericity was violated Geisser-Greenhouse correction was performed. All primary outcome variables were normally distributed as determined using the Kolmogorov-Smirnov normality test. Moreover, for the ANOVA test herein the following assumptions were made: 1. Each group sample is drawn from a normally distributed population. 2. All populations have a common variance. 3. All samples are drawn independently of each other. 4. Within each sample, the observations are sampled randomly and independently of each other. 5. Factor effects are additive. To minimize the chances of a type II error due to a modest sample size, effect sizes were calculated as Eta squared (η2) for primary comparisons, which provides information on the magnitude of the difference between the groups. The threshold values for η2 were defined as small, moderate, and large effects as 0.01, 0.06, and 0.14, respectively (Lakens, 2013). Spearman correlation coefficients were used to assess the relationship between dyspnea and HR. Data are presented as mean ± SD unless otherwise stated. 3 Results The resistance to flow across each mask material at physiologic flow rates is illustrated in Fig. 1B. The pressure drop increased linearly with flow. There was a noticeable difference in the pressure drop and calculated resistance (Fig. 1C) to air flow caused by all masks relative to a non-masked control, with a markedly higher resistance for the N95. Note that the non-masked control resistance was determined across the standard Medgraphics exercise testing pneumotachograph. By comparison, the differential pressure for the standard Hans-Rudolph (adult – large) two-way non-rebreathing valve and mask assembly at a flow rate of 100 L/min is 2.1 cmH20 (https://www.rudolphkc.com/pdf/691151%201215%20K.pdf). 3.1 Ramp exercise: rest and submaximal work rates At rest, MAP (P = 0.923, η2 = 0.04), HR (P = 0.213, η2 = 0.02), SpO2 (P = 0.422, η2 = 0.03), and dyspnea score were not different across all conditions. Fig. 2 illustrates the submaximal responses during the ramp exercise test. Ratings of dyspnea were significantly higher during the submaximal work rates of the ramp test for all mask types compared to the no mask condition (P < 0.0001, η2 = 0.083). During exercise there was a main effect on HR across conditions (P = 0.041, η2 = 0.009), with surgical and N95 masks eliciting a higher HR compared to the no mask condition at all submaximal work rates above 60 W. However, this increase was less than 10 bpm in all instances, but was not present at maximal exercise (P = 0.667, η2 = 0.006). The difference in dyspnea significantly correlated with the difference in HR between no mask and N95 conditions at 120 W (P = 0.038, R2 = 0.4). Submaximal exercise SpO2 was not different across conditions (P = 0.087, η2 = 0.053). Rating of perceived exertion was not different across conditions (P = 0.286). During exercise the MAP (P = 0.897, η2 = 0.004), SV (P = 0.576, η2 = 0.017), and CO (P = 0.831, η2 = 0.003) responses were not altered by the presence of any of the face masks.Fig. 2 Average (±SD) values during the incremental exercise test up to 120 W for arterial oxygen saturation (SpO2), Borg dyspnea scores, heart rate (HR), mean arterial pressure (MAP), stroke volume (SV), and cardiac output (CO). Dyspnea rating was significantly increased with each facemask, but, with the exception of a small increase in HR for the surgical and N95 masks, no other cardiovascular variables were impacted significantly by any mask across multiple submaximal exercise work rates. * P < 0.05 for N95 vs. no mask condition. + P < 0.05 for all masked conditions vs. no mask condition. Fig. 2 3.2 Ramp exercise: maximal work rate Fig. 3 illustrates that SpO2 at peak power (P = 0.623, η2 = 0.01) was not different across conditions. Moreover, the absolute change in SpO2 from baseline to peak power was also not different across each mask condition (P = 0.744, η2 = 0.005). HR at peak power was not different across conditions (P = 0.667, η2 = 0.006). At peak exercise, the magnitude of dyspnea was significantly increased for the N95 (P = 0.011) facemask compared to the no-mask control, but not the flannel (P = 0.104), or surgical (P = 0.110) facemasks compared to the no mask control (η2 = 0.369). RPE was not different among conditions. Importantly, the present investigation included healthy men and women with a range of maximal exercise capacities (range: 168−328 W; 2.42–5.57 W/kg). Absolute peak power from the no-mask control was not different compared to the flannel (P = 0.246), surgical (P = 0.168), or N95 (P = 0.077) conditions (Fig. 3). Similarly, peak power normalized to body mass in the no-mask control was not different compared to the flannel (P = 0.271), surgical (P = 0.229), or N95 (P = 0.071) conditions.Fig. 3 Average (±SD) values for peak power, arterial oxygen saturation (SpO2) and dyspnea scores at peak exercise for each facemask condition. Note the lack of difference for peak power and SpO2 among conditions, but the presence of increased dyspnea. * P < 0.05 vs. no mask condition. Fig. 3 3.3 Constant-load submaximal exercise The constant-work load exercise bouts at 95 and 127 W, substantially confirmed the lack of effect of wearing any mask on cardiovascular control (Table 1 ), with the exception of HR at 95 W, which was significantly elevated above the no mask condition only for the surgical mask (i.e., 129 ± 25 versus 119 ± 18 bpm, P < 0.05). With respect to peak respiratory rate, there was a significant increase for both the surgical and flannel masks at rest and the flannel mask only at 95 W. SpO2 did not change significantly from the no mask condition either at rest or for either work rate for any mask. For all measurements made, the greatest effect of the masks was on the end-expired peri-oral gas concentrations where the surgical mask decreased peak O2% and increased peak CO2% significantly at rest and the N95 significantly increased peak CO2% at rest. During exercise, however, each mask consistently and significantly decreased end-expired perioral O2 and elevated CO2 between ∼0.8−2% compared with the resting no-mask condition.Table 1 Cardiorespiroatry, arterail oxygenation (SpO2), and peri-oral end-expiratory O2 and CO2 at rest and in response to constant work rate exercise with and without masks. Table 1 Rest 95 W 127 W No Mask Surgical Flannel N5 No Mask Surgical Flannel N5 No Mask Surgical Flannel N5 Respiratory Rate (breaths/min 11 ± 2 15 ± 1 * 13 ± 2 * 15 ± 3 18 ± 6 21 ± 10 25 ± 5 24 ± 8 Heart rate (beats/min) 71 ± 10 70 ± 10 73 ± 16 75 ± 14 119 ± 18 129 ± 25* 126 ± 24 118 ± 36 135 ± 26 138 ± 27 128 ± 22 137 ± 25 SpO2 (%) 98 ± 1 98 ± 1 98 ± 1 98 ± 1 98 ± 0 98 ± 1 98 ± 1 98 ± 1 98 ± 1 97 ± 1 98 ± 1 98 ± 1 O2 (%) 16.63 ± 1.29 15.61 ± 0.76* 16.25 ± 0.67 15.77 ± 0.95 17.35 ± 1.42 15.93 ± 1.40* 15.79 ± 1.40* 15.72 ± 1.57* 17.37 ± 1.0 15.86 ± 1.30* 15.60 ± 1.62* 15.31 ± 1.51* PO2 (%) 123 ± 10 116 ± 6* 121 ± 5 117 ± 7 129 ± 11 119 ± 10* 117 ± 10* 117 ± 12* 129 ± 12 118 ± 10* 116 ± 12* 114 ± 11* CO2 (%) 3.96 ± 1.00 4.63 ± 0.60* 4.35 ± 0.58 4.78 ± 0.50* 3.95 ± 0.95 4.90 ± 1.01* 4.48 ± 0.88* 4.96 ± 0.99* 3.86 ± 1.27 4.98 ± 0.94* 5.09 ± 1.14* 5.22 ± 0.98* PCO2 (%) 29 ± 7 34 ± 4* 32 ± 4 36 ± 4 29 ± 7 36 ± 8* 36 ± 7* 37 ± 7* 29 ± 9 37 ± 7* 38 ± 8* 39 ± 7* Mean ± SD. * P < 0.05. 4 Discussion The principal original findings of this investigation support our hypotheses that wearing either the surgical, flannel or N95 mask at rest and during submaximal and maximal exercise did not induce arterial hypoxemia, compromise maximal cycle exercise capacity nor substantially impact major cardiovascular exercise responses. This was true despite the significant elevation of peak end-expired peri-oral CO2 and depression of O2 within the mask of 0.8–2 %. However, despite the intransigence of arterial O2 saturation, subjects rated their dyspnea significantly higher during mask wearing. The importance of this issue cannot be overstated. The U.S. and the World population is facing an unprecedented challenge to health and longevity. Regular physical exercise avoids, and inactivity promotes, a deterioration in cardiovascular health (Chakravarthy and Booth, 2003) that, over time, contributes to an elevated cardiovascular disease risk (Peçanha et al., 2020). If mask wearing is perceived as noxious or contraindicated for healthy or patient populations, especially during exercise, an unavoidable consequence will be greater home isolation (quarantine), decreased physical activity and the health deterioration sequelae (e.g., reduced maximal O2 uptake (Nolan et al. 2018) and, impaired insulin sensitivity and metabolic health (Krogh-Madsen et al., 2010; Mikus et al., 2012; Thyfault and Krogh-Madsen, 2011)). Moreover, low cardiorespiratory fitness accounts for more overall deaths than hypertension, smoking, high cholesterol and diabetes (Blair, 2009). A telling exemplar from Fitbit Inc. data demonstrates that, for the week ending March 22 (2020) average step counts, for over 30 million people, in most countries monitored, decreased significantly (up to 38 %) relative to 2019 (News, 2020; Peçanha et al., 2020). Given this scenario it is imperative that the impact of mask wearing be scientifically evaluated and health experts, public health officials, exercise specialists and scientists as well as the broader public have access to accurate data to guide public policy. 4.1 Dyspnea and rating of perceived exertion (RPE) Dyspnea is the result of integration among multiple factors, including central command, feedback from a variety of receptors throughout the respiratory system as well as the inspiratory pressor response which may increase both dyspnea and RPE (i.e., Borg-dyspnea and Borg-legs, Romer et al., 2006a, 2006b Parshall et al., 2012; de Morree and Marcora, 2015; Laviolette and Laveneziana, 2014). While hypoxia and elevated PCO2 can increase the intensity of dyspnea, the unchanged SpO2 suggests that other factors may be involved in mediating the increased dyspnea with mask wearing during exercise. Specifically, increased central command related to the inspiratory pressor response (higher ventilatory resistance) in combination with psychological factors as related to peri-oral (lips and surrounding skin) temperature elevation (Scarano et al., 2020). Wearing surgical and N95 masks while at rest elicits ∼0.7−1.9 °C increase in skin temperature, which parallels significant differences in discomfort (Scarano et al., 2020). This is an important finding in that changes in face temperature have been shown to alter levels of dyspnea in clinical populations, suggesting that changes in receptor firing with increased facial temperatures from mask wearing may increase the sensation of respiratory discomfort, particularly during exercise (Parshall et al., 2012; Qian et al., 2019). Our findings of an increased level of dyspnea with each mask during exercise is consistent with previous reports at maximal exercise with surgical masks and FFP2/N95 masks (Fikenzer et al., 2020). Not surprisingly exhaustion corresponds to a high RPE and, depending to a degree on the type of test performed (e.g. constant-load, ramp, submaximal/exhaustive), the RPE and dyspnea ratings may converge towards a high value at exhaustion that may or may not be impacted by increased or decreased work of breathing (see Fig. 3 in Harms et al., 2000). 4.2 Cardiopulmonary responses Accounts of increased HR whilst mask wearing are largely anecdotal (Times, 2020), but do support the reported elevated HR by 8–10 bpm during submaximal exercise herein. This directional change, but often of less magnitude, is supported in the peer reviewed literature and may reflect stimulation of the trigeminal reflex resulting in a mild tachycardia at rest and during activity/exercise (Laird et al., 2002; Li et al., 2005). Moreover, laboratory-induced increases in dyspnea elicited with mild inspiratory threshold loading increased resting HR by 8–10 bpm, but in that instance, did not correlate breathing discomfort rating and HR (Nierat et al., 2017). The present study, however, observed a similar increase in HR that did correlate with the increased dyspnea with mask wearing during exercise. In total these findings implicate a putative role for both stimulation of the trigeminal reflex and breathing discomfort in the observed mask-induced elevated HR during submaximal exercise. That neither arterial oxygenation nor exercise capacity were compromised for our subjects by wearing the N95 (or other) masks is in direct contrast to the assertions of Davis and Tsen (Davis and Tsen, 2020), but in agreement with the recent work of Samannan et al. in which COPD patients did not experience any clinically significant mask-induced changes in SpO2 during a standard six-minute walk test (Samannan et al., 2020). It is also pertinent that, at the low flow rates achievable in COPD patients, the elevated resistance provided by the mask would be very modest. These findings are critical in that citing work by Sinkule et al. (Sinkule et al., 2013), Davis and Tsen (Davis and Tsen, 2020) stated that wearing the N95 mask whilst exercising at 2 metabolic equivalents (METS) similar to walking slowly (V̇O2 ∼500 mL/min for a 70 kg individual) elevates inspired CO2 to between 3 and 4% above the normal in fresh inspired air (i.e., 0.03−0.04) and decreases inspired O2 from normal of 20.94 % to 17 %. There are pertinent considerations that suggest Sinkule et al.’s (Sinkule et al., 2013) results, obtained using an automated breathing and metabolic simulator (ABMS, Ocenco, Inc., Pleasant Prairie, WI), are not applicable to humans wearing an N95 or other COVID-19 protective mask as studied herein. Specifically:1 Had the N95 or other mask evaluated herein reduced inspired O2 and increased inspired CO2 as contended by Sinkule et al., we can use the approximate alveolar gas equation to calculate the impact on alveolar and arterial PO2 (PAO2 and PaO2, (West, 1990)): PAO2 = [(PB – 47)xFIO2] –(PACO2/R) Where PB is barometric pressure, FIO2 is inspired O2 fraction, PACO2 is alveolar PCO2 (considered synonymous with arterial PCO2 in these circumstances) and R is the respiratory exchange ratio (V̇CO2/V̇O2, ∼0.8 on a mixed diet at rest). Assuming PaCO2 and PACO2 rise commensurately to ∼60 mmHg, PAO2 will fall from its control (non-mask) value ∼100 mmHg to only ∼30 mmHg! Such perturbations of the arterial blood gasses would drive a massive hyperpneic response elevating both respiratory frequency and tidal volume, neither of which were evident grossly in the present investigation (Kronenberg and Severinghaus, 1971; West, 1990). Even neglecting the hypercapnic and acidotic rightward shift (Bohr effect) of the O2 dissociation curve, arterial O2 saturation would be reduced to just above 50 %; a most dire and life-threatening clinical condition and one not observed herein. But, is it possible that the subjects might hyperventilate sufficiently to restore alveolar PO2 and elevate SpO2 sufficiently to obscure these effects from the observer? Using the alveolar gas equation above, for an inspired PO2 of 115 mmHg to elevate PAO2 to its normal value of ∼100 mmHg, PACO2 (assuming an R of 0.8) would have to fall to ∼18 mmHg which would necessitate more than a doubling of alveolar and thus total ventilation supposing (reasonably) that the inspired CO2 was 0.03−0.04% breathing room air. If inspired CO2 did indeed rise to 3.5 % there is no amount of additional ventilation that could lower PACO2 to the requisite value. Moreover, whilst there was a slightly increased respiratory rate, there was no evidence for the masks substantially elevating the exercise hyperpnea.2 What would the impact of achieving the levels of arterial hypoxemia (PaO2 ∼30 mmHg) and hypercapnia (∼60 mmHg) estimated above from Sinkule et al. (4) be? Ventilation increases some 2−3 L/min for each mmHg PaCO2 rise by elevation of breathing frequency and tidal volume via stimulation of the peripheral (i.e., carotid bodies) and central chemoreceptors (Ainslie and Poulin, 2004; Cormack et al., 1957; Hirshman et al., 1975; Kronenberg and Severinghaus, 1971; Ogoh et al., 2009; West, 1990) and this response is massively potentiated by concomitant hypoxemic stimulation of the carotid bodies (Kronenberg and Severinghaus, 1971; West, 1990). As above, breathing frequency at rest and 97 W (constant-load exercise, Table 1), whilst increased slightly in certain instances (Table 1), did not evidence the substantial tachypnea expected from any massive derangement of blood gasses; had such occurred. 3 Unlike rebreathing in a closed system, the small-to-modest dead space increases provided by the mask (<40−140 ml) would only increase the CO2 load (and decrease inspired O2) very little. Thus, using the upper extreme of the mask end-expiratory peri-oral CO2 elevation (Table 1) would theoretically increase inspired CO2 by only ∼7.0−10 ml/breath for the N95 (140 mL dead space) and less than 2.0 mL/breath for the surgical and flannel masks. These values would correspond to a mean elevation of only 0.4–1.5 % in inspired CO2 for the N95 and 0.1−0.4% for the surgical/flannel masks. In comparison with the estimated 200 mL CO2/min at rest and up to ∼4,000 mL CO2/min or more exhaled during maximal exercise, clearing these tiny additional CO2 loads would not require substantial additional ventilation compared to the no-mask condition. Whereas the young, healthy males and females studied herein are expected to evince a small drop in SpO2 at maximal exercise related to the Bohr-induced (temperature, acidity) rightward shift in the O2 dissociation curve, they are far less likely to develop exercise-induced arterial hypoxemia (EIAH) than clinical populations, for example, with COPD or other pulmonary disease (Andrianopoulos et al., 2014; Casaburi et al., 1991; Wagner et al., 1977). It is true, however, that EIAH can occur without underlying pathology in adults (Dempsey et al., 1984; Dempsey and Wagner, 1999; Dominelli and Sheel, 2019; Powers et al., 1989) and children (Nourry et al., 2004). Furthermore, EIAH may be present in females at far lower metabolic rates than in their male counterparts (Harms et al., 1998a) and reduce V̇O2max significantly (Harms et al., 2000) thus predicating muscle fatigue and exhaustion (Romer and Dempsey, 2006; Romer et al., 2006a). Directly pertinent to the present investigation is that EIAH is both more prevalent and more extreme in highly trained or fitter individuals (Dempsey and Wagner, 1999; Dominelli et al., 2013; Dominelli and Sheel, 2019). Indeed, this population may be far more sensitive to even modest reductions in inspired (and thus alveolar) PO2 (Gore et al., 1996). For instance, when exercising maximally in a mildly hypobaric chamber (50 mmHg below sea level PB, i.e., ∼710 versus 760 mmHg) equivalent to ∼580 m above sea level or inspiring 19.4 % O2 at sea level, trained male cyclists (V̇O2max77 ml/kg/min) desaturated to 86.5 %. This desaturation compared with a value of 93.7 % in their untrained counterparts (V̇O2max51 ml/kg/min) and, crucially, decreased V̇O2max significantly only in the trained cyclists (see also Lawler et al., 1988). In the present investigation, our subjects included two endurance-trained individuals (>6 h training per week) whose relative maximal work rates (Watts/kg) on the incremental work test exceeded that of the other subjects by almost 50 % (range: 2.42–5.57 W/kg). Despite that the expectation was for these individuals to be extremely sensitive to even small reductions in inspired O2%, had such occurred, no mask-induced decreased O2 saturation or work capacity was evident. 4.3 Work of breathing Of course, a reduction in arterial O2 saturation is not the only mechanism by which mask wearing can potentially decrease locomotory muscle O2 delivery and thus exercise performance. As Harms and colleagues initially demonstrated, increasing the work of breathing by, for example, elevating inspiratory resistance during maximal cycling exercise, elevates leg vascular resistance and lowers blood flow, O2 delivery and leg V̇O2 (Harms et al., 1997; Wetter et al., 1999). These changes, in response to augmented respiratory muscle work, do not elevate cardiac output but, rather, redistribute (or “steal”) blood flow to the respiratory muscles that would normally have gone to the exercising leg muscles (Harms et al., 1998b). As such, during maximal exercise, a 28 % increase in the work of breathing, imposed by adding inspiratory resistance, elicits a ∼1.3 L/min reduction in leg blood flow, and decreases maximal exercise capacity (Harms et al., 1997). Unfortunately, we were unable to measure the increased work of breathing induced by the three masks evaluated herein because doing so would have confounded our primary outcomes. However, the absence of any decrement in maximal exercise capacity argues strongly that any increase in the work of breathing had minimal impact on leg muscle(s) perfusion and work capacity. 4.4 Experimental considerations 4.4.1 Statistical power Our statistical power to detect a 3% decrease in SpO2 or 5% in peak power was >0.9. The very low effect sizes reported for SpO2, peak power, and the cardiovascular variables support the lack of substantial physiological differences with mask wearing in these recreational exercisers. Sample size calculations based on our findings indicate that a sample size of >250 would be required to detect significant differences for SpO2 (G*Power 3.1.9.2). With respect to SpO2 it is pertinent that these results are for young, healthy active individuals for whom the respiratory system has arguably been considered "overbuilt" for exercise (Dempsey and Johnson, 1992; Dempsey et al., 2003) and in whom the incidence of expiratory flow limitation or arterial hypoxemia attending maximal cycling exercise is expected to be low. That said, the 6 young, healthy active women (54 % of the tested sample size) would be more likely to experience expiratory flow limitation compared to men (Guenette et al., 2007; Guenette and Sheel, 2007). We acknowledge that our results likely do not apply to elite athletes or patient populations, especially the latter with emphysema/COPD or the elderly, in whom pulmonary function limits or is close-to-limiting exercise performance, who do experience expiratory flow limitation, respiratory muscle fatigue and/or erosion of arterial blood gasses during exercise (e.g., Yoshimura et al. (Yoshimura et al., 2014); Kalinov et al. (Kalinov et al., 2019)). Moreover, whereas elite athletes may be more physiologically susceptible to modest increases in breathing resistance from a mask they also demonstrate greater motivation that may diminish performance decrements (Martin et al., 2016). 4.4.2 Duration of exercise A second experimental consideration is the duration of exercise. The exercise bouts evaluated herein were designed to explore the full range of whole-body O2 transport such as can be achieved by the maximal ramp protocol to exhaustion in sufficiently well-motivated subjects (Poole and Jones, 2017). Including resting and baseline measurements, subjects wore each mask for 20−25 min. We did not employ extended (at least beyond 3 min) square-wave exercise protocols which, in the moderate or heavy exercise intensity domains, could have been sustained for far longer durations. Thus, our investigative protocol, whereas it effectively stressed the upper extremities of oxidative demands did not evaluate the effect of mask wearing on endurance capacity per se. 4.4.3 Ecological validity As detailed in the Methods, to the extent possible, we specifically designed a measurement protocol that did not in-and-of itself change what was being measured. This included avoiding application of a one-way non-rebreathing valve or respiratory mask necessary for determination of ventilatory volumes and flows as well as V̇O2, V̇CO2 and respiratory patterns. This is in contrast to recent work by Fikenzer et al. that evaluated physiological variables during incremental exercise with standard spirometry mask placed over the COVID mask in a leak-proof manner (Fikenzer et al., 2020). Similar to our findings they observe no differences in peak arterial PO2 and key cardiovascular variables. However, the translatability of their findings to real settings is limited by the addition of the leak proof spirometry mask, as this combination can substantially alter pulmonary gas exchange efficacy and arterial blood gasses. It may also alter peri-oral temperature, which may exacerbate the intensity of dyspnea compared to wearing each mask in the prescribed manner and as described by the CDC (i.e., not leak proof). Thus, although the work of breathing (WOB) was undoubtedly increased herein that increase was not sufficient to divert enough cardiac output away from the locomotory muscles to have a major impact on the maximal work rate/exercise tolerance. As mentioned above, Harms and colleagues (Harms et al. JAPPL, 82:1573−83, 1997; 85:609−18, 1998; Harms et al. 89:131−8, 2000) previously demonstrated that increasing the WOB via elevated inspiratory resistance reduced leg blood flow (Harms et al., 1997) and maximal exercise tolerance (Harms et al., 2000) significantly but the subjects were able to work at the same “maximal” work rate under all conditions (albeit for different durations). We believe that how the masks tested herein operate at high ventilations is that there is effective leakage of inspired and expired ventilation around the mask periphery: Such that the in vivo resistance may well have been reduced below that presented in Fig. 1 where no leakage was allowed. We believe that this aspect of the study adds ecological validity to the present investigation as this is how the public actually uses these masks. 4.4.4 Different masks As is evident from the opinions expressed in the literature (Davis and Tsen, 2020), when protective “masks” are considered, there are many different types (e.g., gas masks) and configurations reflecting specific environments for which they are designed. Results from wearing those masks, or even multiple layers of COVID-19 masks, whilst relevant to their particular circumstance should not be confused with the evaluation of the COVID-19 relevant masks evaluated herein. 4.5 Conclusions The findings that neither arterial oxygenation, maximal exercise capacity nor submaximal cardiovascular responses are compromised by surgical, flannel, or N95 masks worn to help prevent COVID-19 transmission has important implications. Specifically, because physical inactivity and the resultant decrease in cardiorespiratory fitness (e.g., maximal O2 uptake, (Blair, 2009)) and metabolic health (Krogh-Madsen et al., 2010; Mikus et al., 2012; Thyfault and Krogh-Madsen, 2011) are major cardiovascular and all-cause risk factors, any additional roadblock or disincentive to exercise conveys a major public health burden. Thus, at least for the young healthy subjects herein, mask wearing does not present a physiological barrier to physical exercise, from a cardiovascular and arterial oxygenation perspective. However the increased dyspnea rating that attended exercising with any of the masks investigated herein, may still present a deterrent to exercise for some individuals. We acknowledge that there is a pressing urgency to investigate the impact of mask wearing on at-risk individuals including the aged and patient populations. Disclosure statement We state that this manuscript is not under consideration elsewhere and that the research reported will not be submitted for publication elsewhere until a final decision is made as to the acceptability of the manuscript. There is no financial or other relationship that influenced the outcome of this paper. In addition, this manuscript represents original work without fabrication, fraud or plagiarism and has been read and approved by all authors. Funding None. Declaration of Competing Interest The authors declare no conflict of interests. Acknowledgements The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The authors gratefully acknowledge Mr. Alec Butenas for his contributions to this research. ==== Refs References Ainslie P.N. Poulin M.J. Ventilatory, cerebrovascular, and cardiovascular interactions in acute hypoxia: regulation by carbon dioxide J. Appl. Physiol. 97 2004 149 159 15004003 Ainsworth B.E. Haskell W.L. Herrmann S.D. Meckes N. Bassett D.R. Jr. Tudor-Locke C. Greer J.L. Vezina J. Whitt-Glover M.C. Leon A.S. 2011 Compendium of Physical Activities: a second update of codes and MET values Med. Sci. Sports Exerc. 43 2011 1575 1581 21681120 Álvarez-Herms J. Julià-Sánchez S. Corbi F. Odriozola-Martínez A. Burtscher M. Putative role of respiratory muscle training to improve endurance performance in hypoxia: a review Front. Physiol. 9 2018 1970 30697170 Andrianopoulos V. Franssen F.M. Peeters J.P. Ubachs T.J. Bukari H. Groenen M. Burtin C. Vogiatzis I. Wouters E.F. Spruit M.A. Exercise-induced oxygen desaturation in COPD patients without resting hypoxemia Respir. Physiol. Neurobiol. 190 2014 40 46 24121092 Askanazi J. Silverberg P.A. Foster R.J. Hyman A.I. Milic-Emili J. Kinney J.M. Effects of respiratory apparatus on breathing pattern J. Appl. Physiol. Respir. Environ. Exerc. Physiol. 48 1980 577 580 6769880 Blair S.N. Physical inactivity: the biggest public health problem of the 21st century Br. J. Sports Med. 43 2009 1 2 19136507 Borg G.A. Psychophysical bases of perceived exertion Med. Sci. Sports Exerc. 14 1982 377 381 7154893 Broxterman R.M. Ade C.J. Barker T. Barstow T.J. Influence of pedal cadence on the respiratory compensation point and its relation to critical power Respir. Physiol. Neurobiol. 208 2015 1 7 25523595 Casaburi R. Patessio A. Ioli F. Zanaboni S. Donner C.F. Wasserman K. Reductions in exercise lactic acidosis and ventilation as a result of exercise training in patients with obstructive lung disease Am. Rev. Respir. Dis. 143 1991 9 18 1986689 Cerretelli P. Sikand R.S. Farhi L.E. Effect of increased airway resistance on ventilation and gas exchange during exercise J. Appl. Physiol. 27 1969 597 600 5360428 Chakravarthy M.V. Booth F.W. Exercise - Hot Topics 2003 Hanley & Belfus Philadelphia, Pennsylvania Cormack R.S. Cunningham D.J. Gee J.B. The effect of carbon dioxide on the respiratory response to want of oxygen in man Q. J. Exp. Physiol. Cogn. Med. Sci. 42 1957 303 319 13494669 Crist C. Fauci: Americans May Need to Wear Masks Through 2022 WebMD 2021 Davis B.A. Tsen L.C. Wearing an N95 Respiratory Mask: An Unintended Exercise Benefit? Anesthesiology 2020 de Morree H.M. Marcora S.M. Guido H.E. Gendolla Mattie Tops Koole Sander L. Psychobiology of Perceived Effort During Physical Tasks. In Handbook of Biobehavioral Approaches to Self-Regulation 2015 Springer New York New York, NY 255 270 Dempsey J.A. Johnson B.D. Demand vs. Capacity in the healthy pulmonary system Schweiz. Z. Sportmed. 40 1992 55 64 1626272 Dempsey J.A. Wagner P.D. Exercise-induced arterial hypoxemia J. Appl. Physiol. 87 1999 1997 2006 10601141 Dempsey J.A. Hanson P.G. Henderson K.S. Exercise-induced arterial hypoxaemia in healthy human subjects at sea level J. Physiol. (Lond.) 355 1984 161 175 6436475 Dempsey J.A. Sheel A.W. Haverkamp H.C. Babcock M.A. Harms C.A. [The John Sutton Lecture: CSEP, 2002]. Pulmonary system limitations to exercise in health Can. J. Appl. Physiol. 28 Suppl 2003 S2 24 14768314 Dominelli P.B. Sheel A.W. Exercise-induced arterial hypoxemia; some answers, more questions Appl. Physiol. Nutr. Metab. 44 2019 571 579 30412430 Dominelli P.B. Foster G.E. Dominelli G.S. Henderson W.R. Koehle M.S. McKenzie D.C. Sheel A.W. Exercise-induced arterial hypoxaemia and the mechanics of breathing in healthy young women J. Physiol. (Lond.) 591 2013 3017 3034 23587886 Epstein D. Korytny A. Isenberg Y. Marcusohn E. Zukermann R. Bishop B. Minha S. Raz A. Miller A. Return to training in the COVID-19 era: the physiological effects of face masks during exercise Scand. J. Med. Sci. Sports 31 2021 70 75 32969531 Fikenzer S. Laufs U. Response to Letter to the editors of Hopkins et al.: effects of surgical and FFP2/N95 face masks on cardiopulmonary exercise capacity: the numbers do not add up Clin. Res. Cardiol. 2020 Fikenzer S. Laufs U. Response to letter to the editors referring to fikenzer, S., uhe, T., Lavall, D., Rudolph, U., Falz, R., Busse, M., Hepp, P., & laufs, U. (2020). Effects of surgical and FFP2/N95 face masks on cardiopulmonary exercise capacity Clin. Cardiol. Off. J. German Cardiac Soc. 2020 1 9 10.1007/s00392-020-01704-y Advance online publicationClin Res Cardiol. Fikenzer S. Uhe T. Lavall D. Rudolph U. Falz R. Busse M. Hepp P. Laufs U. Effects of surgical and FFP2/N95 face masks on cardiopulmonary exercise capacity Clin. Res. Cardiol. 2020 Gilbert R. Auchincloss J.H. Jr. Brodsky J. Boden W. Changes in tidal volume, frequency, and ventilation induced by their measurement J. Appl. Physiol. 33 1972 252 254 5054434 Gore C.J. Hahn A.G. Scroop G.C. Watson D.B. Norton K.I. Wood R.J. Campbell D.P. Emonson D.L. Increased arterial desaturation in trained cyclists during maximal exercise at 580 m altitude J. Appl. Physiol. 80 1996 2204 2210 8806931 Guenette J.A. Sheel A.W. Physiological consequences of a high work of breathing during heavy exercise in humans J. S. Med. Sport / Sports Med. Australia 10 6 2007 341 350 Guenette J.A. Witt J.D. McKenzie D.C. Road J.D. Sheel A.W. Respiratory mechanics during exercise in endurance-trained men and women J. Physiol. (Lond.) 581 Pt 3 2007 1309 1322 17412775 Harms C.A. Babcock M.A. McClaran S.R. Pegelow D.F. Nickele G.A. Nelson W.B. Dempsey J.A. Respiratory muscle work compromises leg blood flow during maximal exercise J. Appl. Physiol. 82 1997 1573 1583 9134907 Harms C.A. McClaran S.R. Nickele G.A. Pegelow D.F. Nelson W.B. Dempsey J.A. Exercise-induced arterial hypoxaemia in healthy young women J. Physiol. (Lond.) 507 Pt 2 1998 619 628 9518719 Harms C.A. Wetter T.J. McClaran S.R. Pegelow D.F. Nickele G.A. Nelson W.B. Hanson P. Dempsey J.A. Effects of respiratory muscle work on cardiac output and its distribution during maximal exercise J. Appl. Physiol. 85 1998 609 618 9688739 Harms C.A. McClaran S.R. Nickele G.A. Pegelow D.F. Nelson W.B. Dempsey J.A. Effect of exercise-induced arterial O2 desaturation on V̇O2max in women Med. Sci. Sports Exerc. 32 2000 1101 1108 10862536 Hirshman C.A. McCullough R.E. Weil J.V. Normal values for hypoxic and hypercapnic ventilaroty drives in man J. Appl. Physiol. 38 1975 1095 1098 1141125 Jang W.M. Jang D.H. Lee J.Y. Social distancing and transmission-reducing practices during the 2019 coronavirus disease and 2015 middle east respiratory syndrome coronavirus outbreaks in Korea J. Korean Med. Sci. 35 2020 e220 32537955 Kalinov R.I. Marinov B.I. Stoyanova D.I. Hodgev V.A. Vladimirova-Kitova L.G. Nikolov F.P. Kostianev S.S. Desaturation during physical exercise in COPD patients - a stable-over-time phenomenon Folia Med. (Plovdiv) 61 2019 204 212 31301664 Krogh-Madsen R. Thyfault J.P. Broholm C. Mortensen O.H. Olsen R.H. Mounier R. Plomgaard P. van Hall G. Booth F.W. Pedersen B.K. A 2-wk reduction of ambulatory activity attenuates peripheral insulin sensitivity J. Appl. Physiol. 108 2010 1034 1040 20044474 Kronenberg R.S. Severinghaus J.W. Chemical control of ventilation: man Altman P.L. Dittmer D.S. Handbook of Respiration and Circulation. 1971 Federation of American Societies for Experimental Biology Bethesda, Md 102 Laird I.S. Goldsmith R. Pack R.J. Vitalis A. The effect on heart rate and facial skin temperature of wearing respiratory protection at work Ann. Occup. Hyg. 46 2002 143 148 12074023 Lakens D. Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs Front. Psychol. 4 2013 863 24324449 Lassing J. Falz R. Pokel C. Fikenzer S. Laufs U. Schulze A. Holldobler N. Rudrich P. Busse M. Effects of surgical face masks on cardiopulmonary parameters during steady state exercise Sci. Rep. 10 2020 22363 33349641 Laviolette L. Laveneziana P. Dyspnoea: a multidimensional and multidisciplinary approach Eur. Resp. J. Off. J. Eur. Soc. Clin. Respiratory Physiol.ogy 43 6 2014 1750 1762 Lawler J. Powers S.K. Thompson D. Linear relationship between VO2max and VO2max decrement during exposure to acute hypoxia J. Appl. Physiol. 64 1988 1486 1492 3378983 Leung N.H.L. Chu D.K.W. Shiu E.Y.C. Chan K.H. McDevitt J.J. Hau B.J.P. Yen H.L. Li Y. Ip D.K.M. Peiris J.S.M. Seto W.H. Leung G.M. Milton D.K. Cowling B.J. Respiratory virus shedding in exhaled breath and efficacy of face masks Nat. Med. 26 2020 676 680 32371934 Li Y. Tokura H. Guo Y.P. Wong A.S. Wong T. Chung J. Newton E. Effects of wearing N95 and surgical facemasks on heart rate, thermal stress and subjective sensations Int. Arch. Occup. Environ. Health 78 2005 501 509 15918037 Mancini D.M. Henson D. LaManca J. Levine S. Respiratory muscle function and dyspnea in patients with chronic congestive heart failure Circulation 86 1992 909 918 1516204 Martin K. Staiano W. Menaspà P. Hennessey T. Marcora S. Keegan R. Thompson K.G. Martin D. Halson S. Rattray B. Superior Inhibitory control and resistance to mental fatigue in professional road cyclists PLoS One 11 7 2016 e0159907 Mikus C.R. Oberlin D.J. Libla J.L. Taylor A.M. Booth F.W. Thyfault J.P. Lowering physical activity impairs glycemic control in healthy volunteers Med. Sci. Sports Exerc. 44 2012 225 231 21716152 News F. The Impact of Coronavirus on Physical Activity 2020 Nierat M.C. Laviolette L. Hudson A. Similowski T. Sevoz-Couche C. Experimental dyspnea as a stressor: differential cardiovegetative responses to inspiratory threshold loading in healthy men and women J. Appl. Physiol. 123 2017 205 212 28473608 Nourry C. Fabre C. Bart F. Grosbois J.M. Berthoin S. Mucci P. Evidence of exercise-induced arterial hypoxemia in prepubescent trained children Pediatr. Res. 55 2004 674 681 14739360 Ogoh S. Ainslie P.N. Miyamoto T. Onset responses of ventilation and cerebral blood flow to hypercapnia in humans: rest and exercise J. Appl. Physiol. 106 2009 880 886 19131474 Organization W.H. Response H.E.Pa Advice on the Use of Masks in the Context of COVID-19 2020 16 Parshall M.B. Schwartzstein R.M. Adams L. Banzett R.B. Manning H.L. Bourbeau J. Calverley P.M. Gift A.G. Harver A. Lareau S.C. Mahler D.A. Meek P.M. O’Donnell D.E. American Thoracic Society Committee on, D An official American Thoracic Society statement: update on the mechanisms, assessment, and management of dyspnea Am. J. Respir. Crit. Care Med. 185 2012 435 452 22336677 Peçanha T. Goessler K.F. Roschel H. Gualano B. Social isolation during the COVID-19 pandemic can increase physical inactivity and the global burden of cardiovascular disease Am. J. Physiol. Heart Circ. Physiol. 318 2020 H1441 h1446 32412779 Poole D.C. Jones A.M. Measurement of the maximum oxygen uptake V̇02max: V̇02peak is no longer acceptable J. Appl. Physiol. 122 2017 997 1002 28153947 Poole D.C. Gaesser G.A. Hogan M.C. Knight D.R. Wagner P.D. Pulmonary and leg V̇O2 during submaximal exercise: implications for muscular efficiency J. Appl. Physiol. 72 1985 1992 805 810 1559962 Powers S.K. Lawler J. Dempsey J.A. Dodd S. Landry G. Effects of incomplete pulmonary gas exchange on VO2 max J. Appl. Physiol. 66 1989 2491 2495 2745310 Qian Y. Wu Y. Rozman de Moraes A. Yi X. Geng Y. Dibaj S. Liu D. Naberhuis J. Bruera E. Fan therapy for the treatment of Dyspnea in adults: a systematic review J. Pain Symptom Manage. 58 2019 481 486 31004769 Romer L.M. Dempsey J.A. Effects of exercise-induced arterial hypoxaemia on limb muscle fatigue and performance Clin. Exp. Pharmacol. Physiol. 33 2006 391 394 16620307 Romer L.M. Haverkamp H.C. Lovering A.T. Pegelow D.F. Dempsey J.A. Effect of exercise-induced arterial hypoxemia on quadriceps muscle fatigue in healthy humans Am. J. Physiol. Regul. Integr. Comp. Physiol. 290 2006 R365 375 16166208 Romer L.M. Lovering A.T. Haverkamp H.C. Pegelow D.F. Dempsey J.A. Effect of inspiratory muscle work on peripheral fatigue of locomotor muscles in healthy humans J. Physiol. (Lond.) 571 Pt 2 2006 425 439 16373384 Rosner B. Fundamentals of Biostatistics 7th ed. 2011 Brooks/Cole Boston, MA Samannan R. Holt G. Calderon-Candelario R. Mirsaeidi M. Campos M. Effect of face masks on gas exchange in healthy persons and patients with COPD Ann. Am. Thorac. Soc. 2020 Scarano A. Inchingolo F. Lorusso F. Facial skin temperature and discomfort when wearing protective face masks: thermal infrared imaging evaluation and hands moving the mask Int. J. Environ. Res. Public Health 17 2020 Schulte J.H. Sealed environments in relation to health and disease Arch. Environ. Health 8 1964 438 452 14097291 Shaw K. Butcher S. Ko J. Zello G.A. Chilibeck P.D. Wearing of cloth or disposable surgical face masks has no effect on vigorous exercise performance in healthy individuals Int. J. Environ. Res. Public Health 2020 17 33375123 Sinkule E.J. Powell J.B. Goss F.L. Evaluation of N95 respirator use with a surgical mask cover: effects on breathing resistance and inhaled carbon dioxide Ann. Occup. Hyg. 57 2013 384 398 23108786 Sugawara J. Tanabe T. Miyachi M. Yamamoto K. Takahashi K. Iemitsu M. Otsuki T. Homma S. Maeda S. Ajisaka R. Matsuda M. Non-invasive assessment of cardiac output during exercise in healthy young humans: comparison between Modelflow method and Doppler echocardiography method Acta Physiol. Scand. 179 2003 361 366 14656373 Thyfault J.P. Krogh-Madsen R. Metabolic disruptions induced by reduced ambulatory activity in free-living humans J. Appl. Physiol. 111 2011 1218 1224 21636564 Times T.N.Y. Exercising While Wearing a Mask 2020 The New York Times New York, NY Wagner P.D. Dantzker D.R. Dueck R. Clausen J.L. West J.B. Ventilation-perfusion inequality in chronic obstructive pulmonary disease J. Clin. Invest. 59 1977 203 216 833271 West J.B. Respiratory Physiology--the Essentials 4th ed. 1990 Williams and Wilkins Baltimore Wetter T.J. Harms C.A. Nelson W.B. Pegelow D.F. Dempsey J.A. Influence of respiratory muscle work on VO(2) and leg blood flow during submaximal exercise J. Appl. Physiol. 87 1999 643 651 10444624 Yoshimura K. Maekura R. Hiraga T. Miki K. Kitada S. Miki M. Tateishi Y. Mori M. Identification of three exercise-induced mortality risk factors in patients with COPD Copd 11 2014 615 626 24914923	34352384	PMC9715989	NO-CC CODE	2022-12-03 23:20:52	no	Respir Physiol Neurobiol. 2021 Dec 3; 294:103765	utf-8	Respir Physiol Neurobiol	2,021	10.1016/j.resp.2021.103765	oa_other
==== Front Respir Physiol Neurobiol Respir Physiol Neurobiol Respiratory Physiology & Neurobiology 1569-9048 1878-1519 Elsevier B.V. S1569-9048(21)00150-6 10.1016/j.resp.2021.103765 103765 Article Does wearing a facemask decrease arterial blood oxygenation and impair exercise tolerance? Ade Carl J. a* Turpin Vanessa-Rose G. a Parr Shannon K. a Hammond Stephen T. a White Zachary a Weber Ramona E. ab Schulze Kiana M. ab Colburn Trenton D. ab Poole David C. ab a Departments of Kinesiology, Kansas State University, Manhattan, KS, 66506, USA b Anatomy and Physiology, Kansas State University, Manhattan, KS, 66506, USA ⁎ Corresponding author at: Department of Kinesiology, College of Health and Human Sciences, Kansas State University, Manhattan, KS, 66506, USA. 3 8 2021 12 2021 3 8 2021 294 103765103765 29 4 2021 12 7 2021 25 7 2021 © 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. Introduction Concerns have been raised that COVID-19 face coverings compromise lung function and pulmonary gas exchange to the extent that they produce arterial hypoxemia and hypercapnia during high intensity exercise resulting in exercise intolerance in recreational exercisers. This study therefore aimed to investigate the effects of a surgical, flannel or vertical-fold N95 masks on cardiorespiratory responses to incremental exercise. Methods This investigation studied 11 adult males and females at rest and while performing progressive cycle exercise to exhaustion. We tested the hypotheses that wearing a surgical (S), flannel (F) or horizontal-fold N95 mask compared to no mask (control) would not promote arterial deoxygenation or exercise intolerance nor alter primary cardiovascular variables during submaximal or maximal exercise. Results Despite the masks significantly increasing end-expired peri-oral %CO2 and reducing %O2, each ∼0.8−2% during exercise (P < 0.05), our results supported the hypotheses. Specifically, none of these masks reduced sub-maximal or maximal exercise arterial O2 saturation (P = 0.744), but ratings of dyspnea were significantly increased (P = 0.007). Moreover, maximal exercise capacity was not compromised nor were there any significant alterations of primary cardiovascular responses (mean arterial pressure, stroke volume, cardiac output) found during sub-maximal exercise. Conclusion Whereas these results are for young healthy recreational male and female exercisers and cannot be applied directly to elite athletes, older or patient populations, they do support that arterial hypoxemia and exercise intolerance are not the obligatory consequences of COVID-19-indicated mask-wearing at least for cycling exercise. Keywords COVID-19 facemask N95 Surgical facemask Cycle exercise Submaximal Maximal Exhaustion Exercise-induced arterial hypoxemia Cardiovascular responses Rating of perceived exertion Dyspnea Edited by M Dutschmann ==== Body pmc1 Introduction Wearing a mask that covers the mouth and nostrils, along with social distancing and frequent handwashing, represents the first line of defense against the spread of COVID-19 (Leung et al., 2020) and WHO guidelines (Organization, 2020)). Moreover, health experts have indicated the potential need for COVID-19 face coverings through 2022 (Crist, 2021). Whereas some countries report that mask use is ∼80 % or higher (Jang et al., 2020) concern have been raised that lung function and pulmonary gas exchange are compromised by mask wearing to the extent that they produce arterial hypoxemia and hypercapnia during high intensity exercise resulting in a decreased peak exercise capacity (Davis and Tsen, 2020; Sinkule et al., 2013). The premise for concerns surrounding arterial hypoxemia and impaired exercise capacity have largely been based on earlier, pre−COVID-19 reports using a specialized breathing apparatus, which may not be an appropriate surrogate for the types of face coverings used against COVID-19. For instance, the work of Schulte (1964), which evaluated the effects of a respiratory muscle training device that purposefully increases breathing resistance and can reduce effective inspired O2 to 17 % (from ∼20.94 %), is often referenced when discussing the potential effects of N95 facemasks (Davis and Tsen, 2020; Schulte, 1964). In contrast to COVID-19 recommended masks, the respiratory muscle training masks are specifically designed so that the wearer “dials in” a very high breathing resistance that is specifically intended to overcome the ability of the respiratory muscles to elevate ventilation sufficiently to regulate alveolar and thus arterial CO2 pressures (PCO2). As such, they induce a state of hypoventilation that drives down alveolar PO2. Therefore statements made during the COVID-19 pandemic stating that the standard N95 mask reduces the inspired O2 causing “…headache, lightheadedness, drowsiness, muscular weakness, dyspnea on exertion, nausea and vomiting” (Davis and Tsen, 2020; Schulte, 1964) or that the increased inspiratory and expiratory resistance for greater than 10 min causes “…increased lactate levels, fatigue and impaired physical work capacity.”(Álvarez-Herms et al., 2018; Davis and Tsen, 2020), based on non-N95 specialized breathing apparatus may not be accurate. Recent work Lassing et al. (2020) and Fikenzer et al. (2020) have revealed significant exercise performance impairments with surgical and N95 masks during exercise (Fikenzer et al., 2020; Lassing et al., 2020). However, these two studies superimposed a spirometry mask on top of each COVID-19 mask, resulting in a relatively leak-proof seal that is at odds with the fitting of COVID-19 masks in real-world setting and thus may not accurately reflect what occurs with normal mask wearing. This is a critical point since the interaction between each face mask and the spirometry valve assemblies required for pulmonary gas exchange measurements will have markedly variant resistances, dead-space and other qualities that impact the subject’s breathing pattern, total ventilation, cardiopulmonary function, gas exchange, and facial skin temperature (Askanazi et al., 1980; Cerretelli et al., 1969; Scarano et al., 2020). This becomes more evident when compared to the recent work by Epstein et al. (2021) which revealed that wearing an N95 facemask, without the additional spirometry mask, during incremental exercise did not alter peak exercise workload, but did increase end-tidal CO2 compared to a non-mask control (Epstein et al., 2021). Using a similar incremental exercise protocol, Shaw et al. (2020) evaluated the effects of surgical and cloth masks alone on arterial oxygen saturation and muscle tissue oxygenation levels (Shaw et al., 2020). They revealed that wearing a mask had no effect on exercise performance, arterial oxygen saturation, tissue oxygenation index. These initial formative studies have provided some of the first evidence regarding the physiological response to exercise with COVID-19 face coverings, but given the conflicting findings and controversy surrounding this topic (Fikenzer and Laufs, 2020a, b), highlights the need for additional work. Moreover, given that increase in the work of breathing, which may occur with mask wearing, can alter cardiovascular responses to exercise (Harms et al., 1997), evaluation of the cardiac responses to different COVID-19 masks is warranted. Therefore, we tested two hypotheses. Namely, for recreational exercisers, that despite modest effects on “expired” O2 (decrease) and CO2 (increase), wearing a surgical, flannel or vertical-fold N95 mask would: 1. Have little or no impact on arterial oxygenation (i.e., <3% decrease in arterial O2 saturation from rest, (Dominelli et al., 2013; Harms et al., 2000)) or exercise tolerance. 2. Not alter primary cardiovascular variables (heart rate, cardiac output, blood pressure) during submaximal or maximal exercise. In addition to these hypotheses, the present investigation experimentally tested the resistance to flow across each mask at physiologic flow rates. 2 Methods 2.1 Participants Eleven healthy, recreationally active participants (5 men and 6 women, age 30 ± 11 yrs [mean ± SD], height 175 ± 11 cm, body mass 73.0 ± 12.9 kg) who were experienced with laboratory exercise testing and maximal exercise tests completed the experiments. Sample size was estimated using a population mean ± SD of 98 ± 2 and 100 ± 5 for percentages of SpO2 and peak power, with α = 0.05 and β = 0.1, to detect a 3% and 6% decrease, respectively (Rosner, 2011). The number of subjects required to show a 3% decrease in SpO2 = 7 and 6% decrease in peak power = 16. Importantly, a 3% decrease in SpO2 has been defined in the literature as the definition of exercise-induced arterial hypoxemia (Harms et al., 2000). Participants were free from known cardiovascular, pulmonary, or metabolic disease and were non-smokers as determined from a health history questionnaire. All procedures were approved by the Institutional Review Board of Kansas State University (#9954) and conformed to the standards set by the Declaration of Helsinki. Written informed consent was obtained from all participants. Subjects were instructed to abstain from vigorous activity 12 h. prior and caffeine and food consumption 2 h. prior to the scheduled testing times. A minimum of 24 h. was mandated between each test with all test performed at a similar time of day (±3 h). 2.2 Facemasks Surgical mask (USA ASTM F2100): Size: 3.7 × 6.9 in.(9.5 × 17.5 cm), mass: 3.2 g. Dead space: depends on wearer and fitting, estimated <40 mL. Kunshan jiehong (Kunshan City, China) disposable 3-layer face masks feature an inner and outer layer of spun-bound polypropylene, and a middle filter layer made of melt-blown polypropylene. These materials are the industry standard for disposable 3-ply facemasks. They provide level 1 protection with filtration >95 % for 3 μm and 0.1 μm particles. Pore size is 0.1 μm. Flannel facemask: Champion (Rural Hall, North Carolina, 50 % cotton, 50 % polyester, 2 layers, ∼30 stitches/in). Size: 5.5 × 8.5 in.(14 × 19 cm), mass: 17.3 g. Dead space: depends on wearer and fitting, estimated <40 mL. Authorized by the Federal Drug Administration (not FDA cleared or approved) under an Emergency Use Authorization for use by health care professionals as personal protective equipment under section 564(b)(1) of the Act, 21 U.S.C. Section 360bbb-3(b)(1). N95 NIOSH approved 1570 respirator: Horizontal-fold, non-valved. Size: 7 × 7.75 in.(19 × 20 cm), mass 10.0 g. Dead space: depends on wearer and face shape (140 mL + 15 for subjects herein, measured by water displacement). Dukal Corp (Ronkonkoma, New York). Provides >95 % filter efficiency for particulate oil-free aerosols 0.3 μm. Bacterial filtration efficiency 99.9 %. By comparison, the dead space for the standard Hans-Rudolph (adult – large) two-way non-rebreathing valve and mask assembly is 123.5 mL (https://www.rudolphkc.com/pdf/691151%201215%20K.pdf). The resistance to flow across each mask material was determined at constant flow rates of 24, 48, 72 and 96 L min−1 through a Medgraphics pneumatachograph, which allowed for real-time evaluation of flow (Fig. 1 A). Measurements of the differential pressure on the down- and up-stream side of each mask were taken using a pressure manometer and used to calculate resistance.Fig. 1 Illustration of experimental set-up for evaluation of the resistance to flow across each mask (A). The pressure difference across each mask for multiple flow rates demonstrates that a substantial difference exists for all masked conditions compared to control, with the greatest difference for the N95 mask (B). Calculation of resistance at 48 L min−1 revealed a higher resistance to air flow for all masks (C). Fig. 1 2.3 Study design A randomized cross-over study design was utilized in which participants completed a series of exercise tests while wearing a surgical mask, flannel facemask, and N95 respirator (described below), or no face covering (control). All masks were securely fastened by a trained member of the laboratory according to manufacturers’ specifications and Centers for Disease Control and Prevention (CDC) recommendations. All testing procedures were performed in a temperature-controlled laboratory (21−22 °C), with all participants in a well hydrated state and having abstained from vigorous activity for 24 h prior to testing. At least 48 h was given between adjacent testing sessions. Participants first completed four incremental ramp exercise tests to exhaustion on a cycle ergometer (Lode, Groningen, Netherlands), under each experimental condition. Following a 2 min resting baseline and 2 min unloading cycling, the power output was progressively increased at a rate of 20 W/min until the participant could not maintain the pedal cadence of 60 rpm for 5 consecutive revolutions despite verbal encouragement. Pedal cadence was maintained constant because pedal cadence can affect ventilatory, cardiac, and pulmonary gas exchange responses during incremental exercise (Broxterman et al., 2015). Participants were blinded to power output and test duration. Seat height was recorded for the first test and reproduced for all subsequent tests. During each ramp test, arterial O2 saturation (SpO2) was measured continuously via two independent pulse oximetry units (Datex Ohmeda (GE) S/5 Light Patient Monitor and Innovo pulse oximeter, Innovo Medical, Stafford, Texas, U.S.A.). If measurements were more than 3% different the sensors were repositioned. Heart rate was measured via photoplethysmography. Continuous beat-by-beat blood pressure (systolic, diastolic, and mean) was measured via photoplethysmography (Finometer Pro; Finapres Medical Systems, Amsterdam, The Netherlands). To minimize hand movement artifact during exercise, the right arm was placed on foam padding slightly below heart level on an adjustable stand. In addition, the raw arterial pressure waveforms were continuously monitored for any movement artifacts and these were processed with the Modelflow method, incorporating age, height, and weight, to obtain measurements of stroke volume and cardiac output. This method provides a reliable estimate of the relative changes in stroke volume and cardiac output during exercise in healthy men and women but cannot be used to provide absolute values unless corrected to a standard method (Sugawara et al., 2003). As such, only the change in these variables relative to rest is reported. Borg ratings of perceived exertion and dyspnea were recorded at each minute of exercise as previously described (Borg, 1982; Mancini et al., 1992). A 6−20-point Borg dyspnea scale, versus the modified 10-point version, was used to allow for greater fidelity in identifying different levels of perceived dyspnea between conditions. Subjects were well versed in using these scales to assess exertion and dyspnea. Since the interaction between each face mask and the mouthpiece–breathing valve assemblies required for pulmonary gas exchange measurements will have markedly variant resistances, dead-space and other qualities that impact the subject’s breathing pattern, total ventilation, cardiopulmonary function and gas exchange (Askanazi et al., 1980; Cerretelli et al., 1969) as well as facial skin temperature (Scarano et al., 2020), gas exchange measurements were not performed. Compounded by the fact that the devices used to measure ventilatory variables may themselves induce alterations in the measurement of tidal volume, breathing frequency and thus ventilation (Gilbert et al., 1972), a critical aspect of this study is the absence of mouthpiece–breathing valve assembly placed over each respective facemask. Since the primary goal of this work is to provide real-world ecologically valid results that translate to individuals wearing only a facemask, this represents a key aspect of the study design. Peak exercise workload was used as the primary variable for exercise capacity. Peri-oral End-Expired CO2 and O2. In a subset of individuals (n = 5, 3 F/2 M, Age: 24.8 + 1.6 yr, 175.7 + 9.6 cm, 71.7 + 13.4 kg) additional sub-maximal exercise tests were performed to evaluate changes in expired peri-oral CO2 and O2. Experiments could only be performed in a sub-set due to equipment availability. Within 1 week of completing the ramp tests, each participant returned to the laboratory to perform measurements at rest and during ∼3 min of constant-load cycling exercise at 95 W and 127 W (Monark, Ergomedic 828E, Varberg, Sweden) under each experimental condition whilst wearing a nose clip. The order of these tests was randomized. These work rates were chosen as they are associated with MET ranges of 5–7 METs that are common for moderate-to-heavy physical activities (Ainsworth et al., 2011). The short duration was selected purposefully to capture the primary component and avoid development of a slow component in those subjects who may have been > GET at the higher work load. This strategy has been used previously (Poole et al., 1992). During each test, HR and arterial O2 saturation were measured by pulse oximetry (Proven, OXI-27BL, Beaverton, Oregon) while expired O2 and CO2 were measured using Ametek Applied Electrochemistry Inc., CD-3A and S-3A/I analyzers (Oak Ridge, Tennessee) calibrated using precision gasses that spanned the expected range of measured values. These analyzers are accurate to within +/-0.01 % (O2) and 0.02 % (CO2) with response times of 100 ms for O2 and 25 ms for CO2 to 90 % final response with a sensitivity of 0.001 % over the ranges measured. For the constant-load exercise tests performed herein breath-to-breath variation is typically <10 % of the room air to peri-oral end-expired value. With 10 breaths averaged to provide final values as presented the coefficient of variation (CV) on repeated analyses within condition (i.e., control or masked) was 5% (O2) and 4% (CO2) of the delta from inspired to peri-oral end-tidal giving absolute values for CV of 0.025 % (O2) and 0.016 % (CO2). Breathing frequency was determined by timed (30 s interval) observation. To measure the expired gas concentrations exhaled “mask” gas was sampled between the subject’s chin and lower lip at 0.5 L/min and end-expired CO2 was detected automatically and the corresponding end-expired O2 recorded at that precise time. This procedure collected principally, but not exclusively, the expired flow stream and could be replicated exactly in the control and mask trials without altering mask geometry or inspiratory/expiratory resistances in any way. We accepted that the expirate would be contaminated to some degree by room air in the control condition by streaming effects and, in the mask trials, by gas trapping and streaming within the additional dead space. The values of 10 breaths beyond 2 min 30 s of exercise were measured and averaged. Preliminary studies confirmed that the primary measurements had stabilized after 2 min 30 s for these subjects. 2.4 Statistical analyses For all statistical analysis, the Prism (version 7.04, Graphpad software, INC., La Jolla, CA) data analysis software package was utilized. The effect of each facemask condition on peak exercise responses (peak power, SpO2, dyspnea, HR) were assessed using one-way repeated measures ANOVA with Dunnet’s post hoc-analysis. Sub-maximal exercise responses were assessed via two-way repeated measures ANOVA (condition × workrate) with Dunnet’s post hoc-analysis. In cases were the assumption of sphericity was violated Geisser-Greenhouse correction was performed. All primary outcome variables were normally distributed as determined using the Kolmogorov-Smirnov normality test. Moreover, for the ANOVA test herein the following assumptions were made: 1. Each group sample is drawn from a normally distributed population. 2. All populations have a common variance. 3. All samples are drawn independently of each other. 4. Within each sample, the observations are sampled randomly and independently of each other. 5. Factor effects are additive. To minimize the chances of a type II error due to a modest sample size, effect sizes were calculated as Eta squared (η2) for primary comparisons, which provides information on the magnitude of the difference between the groups. The threshold values for η2 were defined as small, moderate, and large effects as 0.01, 0.06, and 0.14, respectively (Lakens, 2013). Spearman correlation coefficients were used to assess the relationship between dyspnea and HR. Data are presented as mean ± SD unless otherwise stated. 3 Results The resistance to flow across each mask material at physiologic flow rates is illustrated in Fig. 1B. The pressure drop increased linearly with flow. There was a noticeable difference in the pressure drop and calculated resistance (Fig. 1C) to air flow caused by all masks relative to a non-masked control, with a markedly higher resistance for the N95. Note that the non-masked control resistance was determined across the standard Medgraphics exercise testing pneumotachograph. By comparison, the differential pressure for the standard Hans-Rudolph (adult – large) two-way non-rebreathing valve and mask assembly at a flow rate of 100 L/min is 2.1 cmH20 (https://www.rudolphkc.com/pdf/691151%201215%20K.pdf). 3.1 Ramp exercise: rest and submaximal work rates At rest, MAP (P = 0.923, η2 = 0.04), HR (P = 0.213, η2 = 0.02), SpO2 (P = 0.422, η2 = 0.03), and dyspnea score were not different across all conditions. Fig. 2 illustrates the submaximal responses during the ramp exercise test. Ratings of dyspnea were significantly higher during the submaximal work rates of the ramp test for all mask types compared to the no mask condition (P < 0.0001, η2 = 0.083). During exercise there was a main effect on HR across conditions (P = 0.041, η2 = 0.009), with surgical and N95 masks eliciting a higher HR compared to the no mask condition at all submaximal work rates above 60 W. However, this increase was less than 10 bpm in all instances, but was not present at maximal exercise (P = 0.667, η2 = 0.006). The difference in dyspnea significantly correlated with the difference in HR between no mask and N95 conditions at 120 W (P = 0.038, R2 = 0.4). Submaximal exercise SpO2 was not different across conditions (P = 0.087, η2 = 0.053). Rating of perceived exertion was not different across conditions (P = 0.286). During exercise the MAP (P = 0.897, η2 = 0.004), SV (P = 0.576, η2 = 0.017), and CO (P = 0.831, η2 = 0.003) responses were not altered by the presence of any of the face masks.Fig. 2 Average (±SD) values during the incremental exercise test up to 120 W for arterial oxygen saturation (SpO2), Borg dyspnea scores, heart rate (HR), mean arterial pressure (MAP), stroke volume (SV), and cardiac output (CO). Dyspnea rating was significantly increased with each facemask, but, with the exception of a small increase in HR for the surgical and N95 masks, no other cardiovascular variables were impacted significantly by any mask across multiple submaximal exercise work rates. * P < 0.05 for N95 vs. no mask condition. + P < 0.05 for all masked conditions vs. no mask condition. Fig. 2 3.2 Ramp exercise: maximal work rate Fig. 3 illustrates that SpO2 at peak power (P = 0.623, η2 = 0.01) was not different across conditions. Moreover, the absolute change in SpO2 from baseline to peak power was also not different across each mask condition (P = 0.744, η2 = 0.005). HR at peak power was not different across conditions (P = 0.667, η2 = 0.006). At peak exercise, the magnitude of dyspnea was significantly increased for the N95 (P = 0.011) facemask compared to the no-mask control, but not the flannel (P = 0.104), or surgical (P = 0.110) facemasks compared to the no mask control (η2 = 0.369). RPE was not different among conditions. Importantly, the present investigation included healthy men and women with a range of maximal exercise capacities (range: 168−328 W; 2.42–5.57 W/kg). Absolute peak power from the no-mask control was not different compared to the flannel (P = 0.246), surgical (P = 0.168), or N95 (P = 0.077) conditions (Fig. 3). Similarly, peak power normalized to body mass in the no-mask control was not different compared to the flannel (P = 0.271), surgical (P = 0.229), or N95 (P = 0.071) conditions.Fig. 3 Average (±SD) values for peak power, arterial oxygen saturation (SpO2) and dyspnea scores at peak exercise for each facemask condition. Note the lack of difference for peak power and SpO2 among conditions, but the presence of increased dyspnea. * P < 0.05 vs. no mask condition. Fig. 3 3.3 Constant-load submaximal exercise The constant-work load exercise bouts at 95 and 127 W, substantially confirmed the lack of effect of wearing any mask on cardiovascular control (Table 1 ), with the exception of HR at 95 W, which was significantly elevated above the no mask condition only for the surgical mask (i.e., 129 ± 25 versus 119 ± 18 bpm, P < 0.05). With respect to peak respiratory rate, there was a significant increase for both the surgical and flannel masks at rest and the flannel mask only at 95 W. SpO2 did not change significantly from the no mask condition either at rest or for either work rate for any mask. For all measurements made, the greatest effect of the masks was on the end-expired peri-oral gas concentrations where the surgical mask decreased peak O2% and increased peak CO2% significantly at rest and the N95 significantly increased peak CO2% at rest. During exercise, however, each mask consistently and significantly decreased end-expired perioral O2 and elevated CO2 between ∼0.8−2% compared with the resting no-mask condition.Table 1 Cardiorespiroatry, arterail oxygenation (SpO2), and peri-oral end-expiratory O2 and CO2 at rest and in response to constant work rate exercise with and without masks. Table 1 Rest 95 W 127 W No Mask Surgical Flannel N5 No Mask Surgical Flannel N5 No Mask Surgical Flannel N5 Respiratory Rate (breaths/min 11 ± 2 15 ± 1 * 13 ± 2 * 15 ± 3 18 ± 6 21 ± 10 25 ± 5 24 ± 8 Heart rate (beats/min) 71 ± 10 70 ± 10 73 ± 16 75 ± 14 119 ± 18 129 ± 25* 126 ± 24 118 ± 36 135 ± 26 138 ± 27 128 ± 22 137 ± 25 SpO2 (%) 98 ± 1 98 ± 1 98 ± 1 98 ± 1 98 ± 0 98 ± 1 98 ± 1 98 ± 1 98 ± 1 97 ± 1 98 ± 1 98 ± 1 O2 (%) 16.63 ± 1.29 15.61 ± 0.76* 16.25 ± 0.67 15.77 ± 0.95 17.35 ± 1.42 15.93 ± 1.40* 15.79 ± 1.40* 15.72 ± 1.57* 17.37 ± 1.0 15.86 ± 1.30* 15.60 ± 1.62* 15.31 ± 1.51* PO2 (%) 123 ± 10 116 ± 6* 121 ± 5 117 ± 7 129 ± 11 119 ± 10* 117 ± 10* 117 ± 12* 129 ± 12 118 ± 10* 116 ± 12* 114 ± 11* CO2 (%) 3.96 ± 1.00 4.63 ± 0.60* 4.35 ± 0.58 4.78 ± 0.50* 3.95 ± 0.95 4.90 ± 1.01* 4.48 ± 0.88* 4.96 ± 0.99* 3.86 ± 1.27 4.98 ± 0.94* 5.09 ± 1.14* 5.22 ± 0.98* PCO2 (%) 29 ± 7 34 ± 4* 32 ± 4 36 ± 4 29 ± 7 36 ± 8* 36 ± 7* 37 ± 7* 29 ± 9 37 ± 7* 38 ± 8* 39 ± 7* Mean ± SD. * P < 0.05. 4 Discussion The principal original findings of this investigation support our hypotheses that wearing either the surgical, flannel or N95 mask at rest and during submaximal and maximal exercise did not induce arterial hypoxemia, compromise maximal cycle exercise capacity nor substantially impact major cardiovascular exercise responses. This was true despite the significant elevation of peak end-expired peri-oral CO2 and depression of O2 within the mask of 0.8–2 %. However, despite the intransigence of arterial O2 saturation, subjects rated their dyspnea significantly higher during mask wearing. The importance of this issue cannot be overstated. The U.S. and the World population is facing an unprecedented challenge to health and longevity. Regular physical exercise avoids, and inactivity promotes, a deterioration in cardiovascular health (Chakravarthy and Booth, 2003) that, over time, contributes to an elevated cardiovascular disease risk (Peçanha et al., 2020). If mask wearing is perceived as noxious or contraindicated for healthy or patient populations, especially during exercise, an unavoidable consequence will be greater home isolation (quarantine), decreased physical activity and the health deterioration sequelae (e.g., reduced maximal O2 uptake (Nolan et al. 2018) and, impaired insulin sensitivity and metabolic health (Krogh-Madsen et al., 2010; Mikus et al., 2012; Thyfault and Krogh-Madsen, 2011)). Moreover, low cardiorespiratory fitness accounts for more overall deaths than hypertension, smoking, high cholesterol and diabetes (Blair, 2009). A telling exemplar from Fitbit Inc. data demonstrates that, for the week ending March 22 (2020) average step counts, for over 30 million people, in most countries monitored, decreased significantly (up to 38 %) relative to 2019 (News, 2020; Peçanha et al., 2020). Given this scenario it is imperative that the impact of mask wearing be scientifically evaluated and health experts, public health officials, exercise specialists and scientists as well as the broader public have access to accurate data to guide public policy. 4.1 Dyspnea and rating of perceived exertion (RPE) Dyspnea is the result of integration among multiple factors, including central command, feedback from a variety of receptors throughout the respiratory system as well as the inspiratory pressor response which may increase both dyspnea and RPE (i.e., Borg-dyspnea and Borg-legs, Romer et al., 2006a, 2006b Parshall et al., 2012; de Morree and Marcora, 2015; Laviolette and Laveneziana, 2014). While hypoxia and elevated PCO2 can increase the intensity of dyspnea, the unchanged SpO2 suggests that other factors may be involved in mediating the increased dyspnea with mask wearing during exercise. Specifically, increased central command related to the inspiratory pressor response (higher ventilatory resistance) in combination with psychological factors as related to peri-oral (lips and surrounding skin) temperature elevation (Scarano et al., 2020). Wearing surgical and N95 masks while at rest elicits ∼0.7−1.9 °C increase in skin temperature, which parallels significant differences in discomfort (Scarano et al., 2020). This is an important finding in that changes in face temperature have been shown to alter levels of dyspnea in clinical populations, suggesting that changes in receptor firing with increased facial temperatures from mask wearing may increase the sensation of respiratory discomfort, particularly during exercise (Parshall et al., 2012; Qian et al., 2019). Our findings of an increased level of dyspnea with each mask during exercise is consistent with previous reports at maximal exercise with surgical masks and FFP2/N95 masks (Fikenzer et al., 2020). Not surprisingly exhaustion corresponds to a high RPE and, depending to a degree on the type of test performed (e.g. constant-load, ramp, submaximal/exhaustive), the RPE and dyspnea ratings may converge towards a high value at exhaustion that may or may not be impacted by increased or decreased work of breathing (see Fig. 3 in Harms et al., 2000). 4.2 Cardiopulmonary responses Accounts of increased HR whilst mask wearing are largely anecdotal (Times, 2020), but do support the reported elevated HR by 8–10 bpm during submaximal exercise herein. This directional change, but often of less magnitude, is supported in the peer reviewed literature and may reflect stimulation of the trigeminal reflex resulting in a mild tachycardia at rest and during activity/exercise (Laird et al., 2002; Li et al., 2005). Moreover, laboratory-induced increases in dyspnea elicited with mild inspiratory threshold loading increased resting HR by 8–10 bpm, but in that instance, did not correlate breathing discomfort rating and HR (Nierat et al., 2017). The present study, however, observed a similar increase in HR that did correlate with the increased dyspnea with mask wearing during exercise. In total these findings implicate a putative role for both stimulation of the trigeminal reflex and breathing discomfort in the observed mask-induced elevated HR during submaximal exercise. That neither arterial oxygenation nor exercise capacity were compromised for our subjects by wearing the N95 (or other) masks is in direct contrast to the assertions of Davis and Tsen (Davis and Tsen, 2020), but in agreement with the recent work of Samannan et al. in which COPD patients did not experience any clinically significant mask-induced changes in SpO2 during a standard six-minute walk test (Samannan et al., 2020). It is also pertinent that, at the low flow rates achievable in COPD patients, the elevated resistance provided by the mask would be very modest. These findings are critical in that citing work by Sinkule et al. (Sinkule et al., 2013), Davis and Tsen (Davis and Tsen, 2020) stated that wearing the N95 mask whilst exercising at 2 metabolic equivalents (METS) similar to walking slowly (V̇O2 ∼500 mL/min for a 70 kg individual) elevates inspired CO2 to between 3 and 4% above the normal in fresh inspired air (i.e., 0.03−0.04) and decreases inspired O2 from normal of 20.94 % to 17 %. There are pertinent considerations that suggest Sinkule et al.’s (Sinkule et al., 2013) results, obtained using an automated breathing and metabolic simulator (ABMS, Ocenco, Inc., Pleasant Prairie, WI), are not applicable to humans wearing an N95 or other COVID-19 protective mask as studied herein. Specifically:1 Had the N95 or other mask evaluated herein reduced inspired O2 and increased inspired CO2 as contended by Sinkule et al., we can use the approximate alveolar gas equation to calculate the impact on alveolar and arterial PO2 (PAO2 and PaO2, (West, 1990)): PAO2 = [(PB – 47)xFIO2] –(PACO2/R) Where PB is barometric pressure, FIO2 is inspired O2 fraction, PACO2 is alveolar PCO2 (considered synonymous with arterial PCO2 in these circumstances) and R is the respiratory exchange ratio (V̇CO2/V̇O2, ∼0.8 on a mixed diet at rest). Assuming PaCO2 and PACO2 rise commensurately to ∼60 mmHg, PAO2 will fall from its control (non-mask) value ∼100 mmHg to only ∼30 mmHg! Such perturbations of the arterial blood gasses would drive a massive hyperpneic response elevating both respiratory frequency and tidal volume, neither of which were evident grossly in the present investigation (Kronenberg and Severinghaus, 1971; West, 1990). Even neglecting the hypercapnic and acidotic rightward shift (Bohr effect) of the O2 dissociation curve, arterial O2 saturation would be reduced to just above 50 %; a most dire and life-threatening clinical condition and one not observed herein. But, is it possible that the subjects might hyperventilate sufficiently to restore alveolar PO2 and elevate SpO2 sufficiently to obscure these effects from the observer? Using the alveolar gas equation above, for an inspired PO2 of 115 mmHg to elevate PAO2 to its normal value of ∼100 mmHg, PACO2 (assuming an R of 0.8) would have to fall to ∼18 mmHg which would necessitate more than a doubling of alveolar and thus total ventilation supposing (reasonably) that the inspired CO2 was 0.03−0.04% breathing room air. If inspired CO2 did indeed rise to 3.5 % there is no amount of additional ventilation that could lower PACO2 to the requisite value. Moreover, whilst there was a slightly increased respiratory rate, there was no evidence for the masks substantially elevating the exercise hyperpnea.2 What would the impact of achieving the levels of arterial hypoxemia (PaO2 ∼30 mmHg) and hypercapnia (∼60 mmHg) estimated above from Sinkule et al. (4) be? Ventilation increases some 2−3 L/min for each mmHg PaCO2 rise by elevation of breathing frequency and tidal volume via stimulation of the peripheral (i.e., carotid bodies) and central chemoreceptors (Ainslie and Poulin, 2004; Cormack et al., 1957; Hirshman et al., 1975; Kronenberg and Severinghaus, 1971; Ogoh et al., 2009; West, 1990) and this response is massively potentiated by concomitant hypoxemic stimulation of the carotid bodies (Kronenberg and Severinghaus, 1971; West, 1990). As above, breathing frequency at rest and 97 W (constant-load exercise, Table 1), whilst increased slightly in certain instances (Table 1), did not evidence the substantial tachypnea expected from any massive derangement of blood gasses; had such occurred. 3 Unlike rebreathing in a closed system, the small-to-modest dead space increases provided by the mask (<40−140 ml) would only increase the CO2 load (and decrease inspired O2) very little. Thus, using the upper extreme of the mask end-expiratory peri-oral CO2 elevation (Table 1) would theoretically increase inspired CO2 by only ∼7.0−10 ml/breath for the N95 (140 mL dead space) and less than 2.0 mL/breath for the surgical and flannel masks. These values would correspond to a mean elevation of only 0.4–1.5 % in inspired CO2 for the N95 and 0.1−0.4% for the surgical/flannel masks. In comparison with the estimated 200 mL CO2/min at rest and up to ∼4,000 mL CO2/min or more exhaled during maximal exercise, clearing these tiny additional CO2 loads would not require substantial additional ventilation compared to the no-mask condition. Whereas the young, healthy males and females studied herein are expected to evince a small drop in SpO2 at maximal exercise related to the Bohr-induced (temperature, acidity) rightward shift in the O2 dissociation curve, they are far less likely to develop exercise-induced arterial hypoxemia (EIAH) than clinical populations, for example, with COPD or other pulmonary disease (Andrianopoulos et al., 2014; Casaburi et al., 1991; Wagner et al., 1977). It is true, however, that EIAH can occur without underlying pathology in adults (Dempsey et al., 1984; Dempsey and Wagner, 1999; Dominelli and Sheel, 2019; Powers et al., 1989) and children (Nourry et al., 2004). Furthermore, EIAH may be present in females at far lower metabolic rates than in their male counterparts (Harms et al., 1998a) and reduce V̇O2max significantly (Harms et al., 2000) thus predicating muscle fatigue and exhaustion (Romer and Dempsey, 2006; Romer et al., 2006a). Directly pertinent to the present investigation is that EIAH is both more prevalent and more extreme in highly trained or fitter individuals (Dempsey and Wagner, 1999; Dominelli et al., 2013; Dominelli and Sheel, 2019). Indeed, this population may be far more sensitive to even modest reductions in inspired (and thus alveolar) PO2 (Gore et al., 1996). For instance, when exercising maximally in a mildly hypobaric chamber (50 mmHg below sea level PB, i.e., ∼710 versus 760 mmHg) equivalent to ∼580 m above sea level or inspiring 19.4 % O2 at sea level, trained male cyclists (V̇O2max77 ml/kg/min) desaturated to 86.5 %. This desaturation compared with a value of 93.7 % in their untrained counterparts (V̇O2max51 ml/kg/min) and, crucially, decreased V̇O2max significantly only in the trained cyclists (see also Lawler et al., 1988). In the present investigation, our subjects included two endurance-trained individuals (>6 h training per week) whose relative maximal work rates (Watts/kg) on the incremental work test exceeded that of the other subjects by almost 50 % (range: 2.42–5.57 W/kg). Despite that the expectation was for these individuals to be extremely sensitive to even small reductions in inspired O2%, had such occurred, no mask-induced decreased O2 saturation or work capacity was evident. 4.3 Work of breathing Of course, a reduction in arterial O2 saturation is not the only mechanism by which mask wearing can potentially decrease locomotory muscle O2 delivery and thus exercise performance. As Harms and colleagues initially demonstrated, increasing the work of breathing by, for example, elevating inspiratory resistance during maximal cycling exercise, elevates leg vascular resistance and lowers blood flow, O2 delivery and leg V̇O2 (Harms et al., 1997; Wetter et al., 1999). These changes, in response to augmented respiratory muscle work, do not elevate cardiac output but, rather, redistribute (or “steal”) blood flow to the respiratory muscles that would normally have gone to the exercising leg muscles (Harms et al., 1998b). As such, during maximal exercise, a 28 % increase in the work of breathing, imposed by adding inspiratory resistance, elicits a ∼1.3 L/min reduction in leg blood flow, and decreases maximal exercise capacity (Harms et al., 1997). Unfortunately, we were unable to measure the increased work of breathing induced by the three masks evaluated herein because doing so would have confounded our primary outcomes. However, the absence of any decrement in maximal exercise capacity argues strongly that any increase in the work of breathing had minimal impact on leg muscle(s) perfusion and work capacity. 4.4 Experimental considerations 4.4.1 Statistical power Our statistical power to detect a 3% decrease in SpO2 or 5% in peak power was >0.9. The very low effect sizes reported for SpO2, peak power, and the cardiovascular variables support the lack of substantial physiological differences with mask wearing in these recreational exercisers. Sample size calculations based on our findings indicate that a sample size of >250 would be required to detect significant differences for SpO2 (G*Power 3.1.9.2). With respect to SpO2 it is pertinent that these results are for young, healthy active individuals for whom the respiratory system has arguably been considered "overbuilt" for exercise (Dempsey and Johnson, 1992; Dempsey et al., 2003) and in whom the incidence of expiratory flow limitation or arterial hypoxemia attending maximal cycling exercise is expected to be low. That said, the 6 young, healthy active women (54 % of the tested sample size) would be more likely to experience expiratory flow limitation compared to men (Guenette et al., 2007; Guenette and Sheel, 2007). We acknowledge that our results likely do not apply to elite athletes or patient populations, especially the latter with emphysema/COPD or the elderly, in whom pulmonary function limits or is close-to-limiting exercise performance, who do experience expiratory flow limitation, respiratory muscle fatigue and/or erosion of arterial blood gasses during exercise (e.g., Yoshimura et al. (Yoshimura et al., 2014); Kalinov et al. (Kalinov et al., 2019)). Moreover, whereas elite athletes may be more physiologically susceptible to modest increases in breathing resistance from a mask they also demonstrate greater motivation that may diminish performance decrements (Martin et al., 2016). 4.4.2 Duration of exercise A second experimental consideration is the duration of exercise. The exercise bouts evaluated herein were designed to explore the full range of whole-body O2 transport such as can be achieved by the maximal ramp protocol to exhaustion in sufficiently well-motivated subjects (Poole and Jones, 2017). Including resting and baseline measurements, subjects wore each mask for 20−25 min. We did not employ extended (at least beyond 3 min) square-wave exercise protocols which, in the moderate or heavy exercise intensity domains, could have been sustained for far longer durations. Thus, our investigative protocol, whereas it effectively stressed the upper extremities of oxidative demands did not evaluate the effect of mask wearing on endurance capacity per se. 4.4.3 Ecological validity As detailed in the Methods, to the extent possible, we specifically designed a measurement protocol that did not in-and-of itself change what was being measured. This included avoiding application of a one-way non-rebreathing valve or respiratory mask necessary for determination of ventilatory volumes and flows as well as V̇O2, V̇CO2 and respiratory patterns. This is in contrast to recent work by Fikenzer et al. that evaluated physiological variables during incremental exercise with standard spirometry mask placed over the COVID mask in a leak-proof manner (Fikenzer et al., 2020). Similar to our findings they observe no differences in peak arterial PO2 and key cardiovascular variables. However, the translatability of their findings to real settings is limited by the addition of the leak proof spirometry mask, as this combination can substantially alter pulmonary gas exchange efficacy and arterial blood gasses. It may also alter peri-oral temperature, which may exacerbate the intensity of dyspnea compared to wearing each mask in the prescribed manner and as described by the CDC (i.e., not leak proof). Thus, although the work of breathing (WOB) was undoubtedly increased herein that increase was not sufficient to divert enough cardiac output away from the locomotory muscles to have a major impact on the maximal work rate/exercise tolerance. As mentioned above, Harms and colleagues (Harms et al. JAPPL, 82:1573−83, 1997; 85:609−18, 1998; Harms et al. 89:131−8, 2000) previously demonstrated that increasing the WOB via elevated inspiratory resistance reduced leg blood flow (Harms et al., 1997) and maximal exercise tolerance (Harms et al., 2000) significantly but the subjects were able to work at the same “maximal” work rate under all conditions (albeit for different durations). We believe that how the masks tested herein operate at high ventilations is that there is effective leakage of inspired and expired ventilation around the mask periphery: Such that the in vivo resistance may well have been reduced below that presented in Fig. 1 where no leakage was allowed. We believe that this aspect of the study adds ecological validity to the present investigation as this is how the public actually uses these masks. 4.4.4 Different masks As is evident from the opinions expressed in the literature (Davis and Tsen, 2020), when protective “masks” are considered, there are many different types (e.g., gas masks) and configurations reflecting specific environments for which they are designed. Results from wearing those masks, or even multiple layers of COVID-19 masks, whilst relevant to their particular circumstance should not be confused with the evaluation of the COVID-19 relevant masks evaluated herein. 4.5 Conclusions The findings that neither arterial oxygenation, maximal exercise capacity nor submaximal cardiovascular responses are compromised by surgical, flannel, or N95 masks worn to help prevent COVID-19 transmission has important implications. Specifically, because physical inactivity and the resultant decrease in cardiorespiratory fitness (e.g., maximal O2 uptake, (Blair, 2009)) and metabolic health (Krogh-Madsen et al., 2010; Mikus et al., 2012; Thyfault and Krogh-Madsen, 2011) are major cardiovascular and all-cause risk factors, any additional roadblock or disincentive to exercise conveys a major public health burden. Thus, at least for the young healthy subjects herein, mask wearing does not present a physiological barrier to physical exercise, from a cardiovascular and arterial oxygenation perspective. However the increased dyspnea rating that attended exercising with any of the masks investigated herein, may still present a deterrent to exercise for some individuals. We acknowledge that there is a pressing urgency to investigate the impact of mask wearing on at-risk individuals including the aged and patient populations. Disclosure statement We state that this manuscript is not under consideration elsewhere and that the research reported will not be submitted for publication elsewhere until a final decision is made as to the acceptability of the manuscript. There is no financial or other relationship that influenced the outcome of this paper. In addition, this manuscript represents original work without fabrication, fraud or plagiarism and has been read and approved by all authors. Funding None. Declaration of Competing Interest The authors declare no conflict of interests. Acknowledgements The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The authors gratefully acknowledge Mr. Alec Butenas for his contributions to this research. ==== Refs References Ainslie P.N. Poulin M.J. Ventilatory, cerebrovascular, and cardiovascular interactions in acute hypoxia: regulation by carbon dioxide J. Appl. Physiol. 97 2004 149 159 15004003 Ainsworth B.E. Haskell W.L. Herrmann S.D. Meckes N. Bassett D.R. Jr. Tudor-Locke C. Greer J.L. Vezina J. Whitt-Glover M.C. Leon A.S. 2011 Compendium of Physical Activities: a second update of codes and MET values Med. Sci. Sports Exerc. 43 2011 1575 1581 21681120 Álvarez-Herms J. Julià-Sánchez S. Corbi F. Odriozola-Martínez A. Burtscher M. Putative role of respiratory muscle training to improve endurance performance in hypoxia: a review Front. Physiol. 9 2018 1970 30697170 Andrianopoulos V. Franssen F.M. Peeters J.P. Ubachs T.J. Bukari H. Groenen M. Burtin C. Vogiatzis I. Wouters E.F. Spruit M.A. Exercise-induced oxygen desaturation in COPD patients without resting hypoxemia Respir. Physiol. Neurobiol. 190 2014 40 46 24121092 Askanazi J. Silverberg P.A. Foster R.J. Hyman A.I. Milic-Emili J. Kinney J.M. Effects of respiratory apparatus on breathing pattern J. Appl. Physiol. Respir. Environ. Exerc. Physiol. 48 1980 577 580 6769880 Blair S.N. Physical inactivity: the biggest public health problem of the 21st century Br. J. Sports Med. 43 2009 1 2 19136507 Borg G.A. Psychophysical bases of perceived exertion Med. Sci. Sports Exerc. 14 1982 377 381 7154893 Broxterman R.M. Ade C.J. Barker T. Barstow T.J. Influence of pedal cadence on the respiratory compensation point and its relation to critical power Respir. Physiol. Neurobiol. 208 2015 1 7 25523595 Casaburi R. Patessio A. Ioli F. Zanaboni S. Donner C.F. Wasserman K. Reductions in exercise lactic acidosis and ventilation as a result of exercise training in patients with obstructive lung disease Am. Rev. Respir. Dis. 143 1991 9 18 1986689 Cerretelli P. Sikand R.S. Farhi L.E. Effect of increased airway resistance on ventilation and gas exchange during exercise J. Appl. Physiol. 27 1969 597 600 5360428 Chakravarthy M.V. Booth F.W. Exercise - Hot Topics 2003 Hanley & Belfus Philadelphia, Pennsylvania Cormack R.S. Cunningham D.J. Gee J.B. The effect of carbon dioxide on the respiratory response to want of oxygen in man Q. J. Exp. Physiol. Cogn. Med. Sci. 42 1957 303 319 13494669 Crist C. Fauci: Americans May Need to Wear Masks Through 2022 WebMD 2021 Davis B.A. Tsen L.C. Wearing an N95 Respiratory Mask: An Unintended Exercise Benefit? Anesthesiology 2020 de Morree H.M. Marcora S.M. Guido H.E. Gendolla Mattie Tops Koole Sander L. Psychobiology of Perceived Effort During Physical Tasks. In Handbook of Biobehavioral Approaches to Self-Regulation 2015 Springer New York New York, NY 255 270 Dempsey J.A. Johnson B.D. Demand vs. Capacity in the healthy pulmonary system Schweiz. Z. Sportmed. 40 1992 55 64 1626272 Dempsey J.A. Wagner P.D. Exercise-induced arterial hypoxemia J. Appl. Physiol. 87 1999 1997 2006 10601141 Dempsey J.A. Hanson P.G. Henderson K.S. Exercise-induced arterial hypoxaemia in healthy human subjects at sea level J. Physiol. (Lond.) 355 1984 161 175 6436475 Dempsey J.A. Sheel A.W. Haverkamp H.C. Babcock M.A. Harms C.A. [The John Sutton Lecture: CSEP, 2002]. Pulmonary system limitations to exercise in health Can. J. Appl. Physiol. 28 Suppl 2003 S2 24 14768314 Dominelli P.B. Sheel A.W. Exercise-induced arterial hypoxemia; some answers, more questions Appl. Physiol. Nutr. Metab. 44 2019 571 579 30412430 Dominelli P.B. Foster G.E. Dominelli G.S. Henderson W.R. Koehle M.S. McKenzie D.C. Sheel A.W. Exercise-induced arterial hypoxaemia and the mechanics of breathing in healthy young women J. Physiol. (Lond.) 591 2013 3017 3034 23587886 Epstein D. Korytny A. Isenberg Y. Marcusohn E. Zukermann R. Bishop B. Minha S. Raz A. Miller A. Return to training in the COVID-19 era: the physiological effects of face masks during exercise Scand. J. Med. Sci. Sports 31 2021 70 75 32969531 Fikenzer S. Laufs U. Response to Letter to the editors of Hopkins et al.: effects of surgical and FFP2/N95 face masks on cardiopulmonary exercise capacity: the numbers do not add up Clin. Res. Cardiol. 2020 Fikenzer S. Laufs U. Response to letter to the editors referring to fikenzer, S., uhe, T., Lavall, D., Rudolph, U., Falz, R., Busse, M., Hepp, P., & laufs, U. (2020). Effects of surgical and FFP2/N95 face masks on cardiopulmonary exercise capacity Clin. Cardiol. Off. J. German Cardiac Soc. 2020 1 9 10.1007/s00392-020-01704-y Advance online publicationClin Res Cardiol. Fikenzer S. Uhe T. Lavall D. Rudolph U. Falz R. Busse M. Hepp P. Laufs U. Effects of surgical and FFP2/N95 face masks on cardiopulmonary exercise capacity Clin. Res. Cardiol. 2020 Gilbert R. Auchincloss J.H. Jr. Brodsky J. Boden W. Changes in tidal volume, frequency, and ventilation induced by their measurement J. Appl. Physiol. 33 1972 252 254 5054434 Gore C.J. Hahn A.G. Scroop G.C. Watson D.B. Norton K.I. Wood R.J. Campbell D.P. Emonson D.L. Increased arterial desaturation in trained cyclists during maximal exercise at 580 m altitude J. Appl. Physiol. 80 1996 2204 2210 8806931 Guenette J.A. Sheel A.W. Physiological consequences of a high work of breathing during heavy exercise in humans J. S. Med. Sport / Sports Med. Australia 10 6 2007 341 350 Guenette J.A. Witt J.D. McKenzie D.C. Road J.D. Sheel A.W. Respiratory mechanics during exercise in endurance-trained men and women J. Physiol. (Lond.) 581 Pt 3 2007 1309 1322 17412775 Harms C.A. Babcock M.A. McClaran S.R. Pegelow D.F. Nickele G.A. Nelson W.B. Dempsey J.A. Respiratory muscle work compromises leg blood flow during maximal exercise J. Appl. Physiol. 82 1997 1573 1583 9134907 Harms C.A. McClaran S.R. Nickele G.A. Pegelow D.F. Nelson W.B. Dempsey J.A. Exercise-induced arterial hypoxaemia in healthy young women J. Physiol. (Lond.) 507 Pt 2 1998 619 628 9518719 Harms C.A. Wetter T.J. McClaran S.R. Pegelow D.F. Nickele G.A. Nelson W.B. Hanson P. Dempsey J.A. Effects of respiratory muscle work on cardiac output and its distribution during maximal exercise J. Appl. Physiol. 85 1998 609 618 9688739 Harms C.A. McClaran S.R. Nickele G.A. Pegelow D.F. Nelson W.B. Dempsey J.A. Effect of exercise-induced arterial O2 desaturation on V̇O2max in women Med. Sci. Sports Exerc. 32 2000 1101 1108 10862536 Hirshman C.A. McCullough R.E. Weil J.V. Normal values for hypoxic and hypercapnic ventilaroty drives in man J. Appl. Physiol. 38 1975 1095 1098 1141125 Jang W.M. Jang D.H. Lee J.Y. Social distancing and transmission-reducing practices during the 2019 coronavirus disease and 2015 middle east respiratory syndrome coronavirus outbreaks in Korea J. Korean Med. Sci. 35 2020 e220 32537955 Kalinov R.I. Marinov B.I. Stoyanova D.I. Hodgev V.A. Vladimirova-Kitova L.G. Nikolov F.P. Kostianev S.S. Desaturation during physical exercise in COPD patients - a stable-over-time phenomenon Folia Med. (Plovdiv) 61 2019 204 212 31301664 Krogh-Madsen R. Thyfault J.P. Broholm C. Mortensen O.H. Olsen R.H. Mounier R. Plomgaard P. van Hall G. Booth F.W. Pedersen B.K. A 2-wk reduction of ambulatory activity attenuates peripheral insulin sensitivity J. Appl. Physiol. 108 2010 1034 1040 20044474 Kronenberg R.S. Severinghaus J.W. Chemical control of ventilation: man Altman P.L. Dittmer D.S. Handbook of Respiration and Circulation. 1971 Federation of American Societies for Experimental Biology Bethesda, Md 102 Laird I.S. Goldsmith R. Pack R.J. Vitalis A. The effect on heart rate and facial skin temperature of wearing respiratory protection at work Ann. Occup. Hyg. 46 2002 143 148 12074023 Lakens D. Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs Front. Psychol. 4 2013 863 24324449 Lassing J. Falz R. Pokel C. Fikenzer S. Laufs U. Schulze A. Holldobler N. Rudrich P. Busse M. Effects of surgical face masks on cardiopulmonary parameters during steady state exercise Sci. Rep. 10 2020 22363 33349641 Laviolette L. Laveneziana P. Dyspnoea: a multidimensional and multidisciplinary approach Eur. Resp. J. Off. J. Eur. Soc. Clin. Respiratory Physiol.ogy 43 6 2014 1750 1762 Lawler J. Powers S.K. Thompson D. Linear relationship between VO2max and VO2max decrement during exposure to acute hypoxia J. Appl. Physiol. 64 1988 1486 1492 3378983 Leung N.H.L. Chu D.K.W. Shiu E.Y.C. Chan K.H. McDevitt J.J. Hau B.J.P. Yen H.L. Li Y. Ip D.K.M. Peiris J.S.M. Seto W.H. Leung G.M. Milton D.K. Cowling B.J. Respiratory virus shedding in exhaled breath and efficacy of face masks Nat. Med. 26 2020 676 680 32371934 Li Y. Tokura H. Guo Y.P. Wong A.S. Wong T. Chung J. Newton E. Effects of wearing N95 and surgical facemasks on heart rate, thermal stress and subjective sensations Int. Arch. Occup. Environ. Health 78 2005 501 509 15918037 Mancini D.M. Henson D. LaManca J. Levine S. Respiratory muscle function and dyspnea in patients with chronic congestive heart failure Circulation 86 1992 909 918 1516204 Martin K. Staiano W. Menaspà P. Hennessey T. Marcora S. Keegan R. Thompson K.G. Martin D. Halson S. Rattray B. Superior Inhibitory control and resistance to mental fatigue in professional road cyclists PLoS One 11 7 2016 e0159907 Mikus C.R. Oberlin D.J. Libla J.L. Taylor A.M. Booth F.W. Thyfault J.P. Lowering physical activity impairs glycemic control in healthy volunteers Med. Sci. Sports Exerc. 44 2012 225 231 21716152 News F. The Impact of Coronavirus on Physical Activity 2020 Nierat M.C. Laviolette L. Hudson A. Similowski T. Sevoz-Couche C. Experimental dyspnea as a stressor: differential cardiovegetative responses to inspiratory threshold loading in healthy men and women J. Appl. Physiol. 123 2017 205 212 28473608 Nourry C. Fabre C. Bart F. Grosbois J.M. Berthoin S. Mucci P. Evidence of exercise-induced arterial hypoxemia in prepubescent trained children Pediatr. Res. 55 2004 674 681 14739360 Ogoh S. Ainslie P.N. Miyamoto T. Onset responses of ventilation and cerebral blood flow to hypercapnia in humans: rest and exercise J. Appl. Physiol. 106 2009 880 886 19131474 Organization W.H. Response H.E.Pa Advice on the Use of Masks in the Context of COVID-19 2020 16 Parshall M.B. Schwartzstein R.M. Adams L. Banzett R.B. Manning H.L. Bourbeau J. Calverley P.M. Gift A.G. Harver A. Lareau S.C. Mahler D.A. Meek P.M. O’Donnell D.E. American Thoracic Society Committee on, D An official American Thoracic Society statement: update on the mechanisms, assessment, and management of dyspnea Am. J. Respir. Crit. Care Med. 185 2012 435 452 22336677 Peçanha T. Goessler K.F. Roschel H. Gualano B. Social isolation during the COVID-19 pandemic can increase physical inactivity and the global burden of cardiovascular disease Am. J. Physiol. Heart Circ. Physiol. 318 2020 H1441 h1446 32412779 Poole D.C. Jones A.M. Measurement of the maximum oxygen uptake V̇02max: V̇02peak is no longer acceptable J. Appl. Physiol. 122 2017 997 1002 28153947 Poole D.C. Gaesser G.A. Hogan M.C. Knight D.R. Wagner P.D. Pulmonary and leg V̇O2 during submaximal exercise: implications for muscular efficiency J. Appl. Physiol. 72 1985 1992 805 810 1559962 Powers S.K. Lawler J. Dempsey J.A. Dodd S. Landry G. Effects of incomplete pulmonary gas exchange on VO2 max J. Appl. Physiol. 66 1989 2491 2495 2745310 Qian Y. Wu Y. Rozman de Moraes A. Yi X. Geng Y. Dibaj S. Liu D. Naberhuis J. Bruera E. Fan therapy for the treatment of Dyspnea in adults: a systematic review J. Pain Symptom Manage. 58 2019 481 486 31004769 Romer L.M. Dempsey J.A. Effects of exercise-induced arterial hypoxaemia on limb muscle fatigue and performance Clin. Exp. Pharmacol. Physiol. 33 2006 391 394 16620307 Romer L.M. Haverkamp H.C. Lovering A.T. Pegelow D.F. Dempsey J.A. Effect of exercise-induced arterial hypoxemia on quadriceps muscle fatigue in healthy humans Am. J. Physiol. Regul. Integr. Comp. Physiol. 290 2006 R365 375 16166208 Romer L.M. Lovering A.T. Haverkamp H.C. Pegelow D.F. Dempsey J.A. Effect of inspiratory muscle work on peripheral fatigue of locomotor muscles in healthy humans J. Physiol. (Lond.) 571 Pt 2 2006 425 439 16373384 Rosner B. Fundamentals of Biostatistics 7th ed. 2011 Brooks/Cole Boston, MA Samannan R. Holt G. Calderon-Candelario R. Mirsaeidi M. Campos M. Effect of face masks on gas exchange in healthy persons and patients with COPD Ann. Am. Thorac. Soc. 2020 Scarano A. Inchingolo F. Lorusso F. Facial skin temperature and discomfort when wearing protective face masks: thermal infrared imaging evaluation and hands moving the mask Int. J. Environ. Res. Public Health 17 2020 Schulte J.H. Sealed environments in relation to health and disease Arch. Environ. Health 8 1964 438 452 14097291 Shaw K. Butcher S. Ko J. Zello G.A. Chilibeck P.D. Wearing of cloth or disposable surgical face masks has no effect on vigorous exercise performance in healthy individuals Int. J. Environ. Res. Public Health 2020 17 33375123 Sinkule E.J. Powell J.B. Goss F.L. Evaluation of N95 respirator use with a surgical mask cover: effects on breathing resistance and inhaled carbon dioxide Ann. Occup. Hyg. 57 2013 384 398 23108786 Sugawara J. Tanabe T. Miyachi M. Yamamoto K. Takahashi K. Iemitsu M. Otsuki T. Homma S. Maeda S. Ajisaka R. Matsuda M. Non-invasive assessment of cardiac output during exercise in healthy young humans: comparison between Modelflow method and Doppler echocardiography method Acta Physiol. Scand. 179 2003 361 366 14656373 Thyfault J.P. Krogh-Madsen R. Metabolic disruptions induced by reduced ambulatory activity in free-living humans J. Appl. Physiol. 111 2011 1218 1224 21636564 Times T.N.Y. Exercising While Wearing a Mask 2020 The New York Times New York, NY Wagner P.D. Dantzker D.R. Dueck R. Clausen J.L. West J.B. Ventilation-perfusion inequality in chronic obstructive pulmonary disease J. Clin. Invest. 59 1977 203 216 833271 West J.B. Respiratory Physiology--the Essentials 4th ed. 1990 Williams and Wilkins Baltimore Wetter T.J. Harms C.A. Nelson W.B. Pegelow D.F. Dempsey J.A. Influence of respiratory muscle work on VO(2) and leg blood flow during submaximal exercise J. Appl. Physiol. 87 1999 643 651 10444624 Yoshimura K. Maekura R. Hiraga T. Miki K. Kitada S. Miki M. Tateishi Y. Mori M. Identification of three exercise-induced mortality risk factors in patients with COPD Copd 11 2014 615 626 24914923	0	PMC9715990	NO-CC CODE	2022-12-09 23:15:06	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S30-S31	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.100	oa_other
==== 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)00449-0 10.1016/j.jmir.2022.10.070 Article Analysis and management of the COVID-19 pandemic impact on a multispecialty diagnostic imaging department Roletto Andrea 1 Zanardo M 1 Cozzi A 1 Schiaffino S 2 Tritella S 2 Susini F 2 Gerra F 2 Sardanelli F 1 1 Department of Biomedical Sciences for Health, University of Milan, Milan, Italy 2 Radiology Unit, IRCCS Policlinico San Donato, San Donato Milanese, Italy 2 12 2022 12 2022 2 12 2022 53 4 S21S21 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 The propose of this study was to evaluate the impact of the first two waves of the COVID-19 pandemic on a multispecialty radiology department in a large tertiary university hospital in Northern Italy. Methods The numbers of all radiological exams performed in the radiology department of Scientific Institute for Research, Hospitalization and Healthcare (namely, IRCCS) Policlinico San Donato (San Donato Milanese, Italy) from March 2019 to March 2021 were collected and compared, subdividing them both temporally, modality, sub-specialty, and setting. Results Comparing the first 12 months of the COVID-19 pandemic (from March 2020 to February 2021) with the previous 12 months (from March 2019 to February 2020), there was an overall decrease in total radiological examinations equal to 26% (from 127,998 to 94,550). The most affected modality was DXA (from 4,706 to 2,989, -36%), followed by ultrasonography (from 17,212 to 11,644, -32%), digital radiography (from 66,050 to 47,374, -28%), MRI (from 13,332 to 10,140, -24%), CT (from 19,208 to 15,746, -18%), and mammograms (from 7,490 to 6,657, -11%). Chest CTs of inpatients saw a +15% surge (from 1,087 to 1,144), with far larger sizable increments being observed for chest X-ray examinations of outpatients (from 3,032 to 7,536, +131%). Further sub-analysis according to pandemic waves highlighted an overall -65% decrease of radiological services during the first wave (from March to May 2020), curtailed to -3% during the June–October period and then again rising to -23% during the second wave (from November 2020 to February 2021). Conclusion The COVID-19 pandemic led to a marked decrease of total radiological examinations during the two pandemic waves, limited to -26% by the implementation of safety protocols during the second wave and by increased activity during the inter-wave period. Keywords COVID-19 Management Radiology Diagnostic Imaging Data analysis ==== Body pmc	0	PMC9715991	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S21	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.070	oa_other
==== 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)00429-5 10.1016/j.jmir.2022.10.050 Article Experiences of newly qualified therapeutic radiographers who transitioned to work during Covid-19 Courtier N 1 Williamson K 1 Brown P 1 Pope E 1 Chivers E 1 Mundy LA 1 1 Department of Radiotherapy & Oncology, Cardiff University, Cardiff. Wales, UK 2 12 2022 12 2022 2 12 2022 53 4 S15S15 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 The Covid-19 pandemic continues to impact on how radiotherapy is delivered, how staff do their job and how patients are cared for. Part of the UK NHS response to the covid-19 crisis was to accelerate final year radiotherapy students into work as therapeutic radiographers. The study objective is to explore the experiences of a cohort of new registrants who started work in May 2020. Methods In depth interviews were conducted remotely with newly qualified therapeutic radiography registrants regarding their first 12 months working in UK NHS cancer centres. Data were analysed within and across cases using a framework analysis and synthesised thematically. Results Eleven radiographers were interviewed, working across six different sites. Key generated themes are the risk of impaired professional socialisation due to incongruence between students’ expectations and the reality in clinical departments. We use Bridges Transitional Model to show how a combination of the disrupted/undefined end to university and a perceived lack of recognition of professional knowledge, skills and values evident in our data may leave participants stuck in a middle stage of the transition process. Slower than expected professional development led to demotivation, which was also associated with rising covid-19 case numbers. Conclusion The covid-19 pandemic accentuated and heightened the existing challenge of professional integration and socialisation faced by new therapeutic radiography staff. Demotivation and potentially attrition are more likely in this environment. Compassionate leadership that fosters the mentorship of junior cohorts as part of a flexible preceptorship package could mitigate these risks. Keywords Covid-19 transition new staff ==== Body pmc	35909060	PMC9715992	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S15	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.050	oa_other
==== 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)00532-X 10.1016/j.jmir.2022.10.153 Article Radiological manifestations of COVID-19 patients aged 50 years and above treated at Mbeya Zonal Referral Hospital – Tanzania Langu Joel 1 Mwakyusa Ngwilo 1 Alphonce Byx 1 1 Radiology Department, Mbeya Zonal Referral Hospital, Mbeya, Tanzania 2 12 2022 12 2022 2 12 2022 53 4 S47S47 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 COVID-19 as a pandemic disease has claimed many lives due to its high mortality rate causing social and economic impact to the societies. This has drawn a greater global attention on the health sector; therefore, its diagnosis is of great importance. Chest radiography has been used as a baseline investigation in many lower- and middle-income countries because of its availability and lower cost so we aim to compare the radiological findings and Polymerase Chain Reaction (PCR) results among patients aged 50 years and above who had chest radiography findings. Methods This is a retrospective study using chest X-ray data collected in April 2021. Adults aged 50 years and above with viral pneumonia suspects were included in this analysis. Patients’ demographics and chest X-ray findings were reported. Radiographic findings were correlated with PCR results. Descriptive statistics were reported. Results A total of 154 participants were included; 95(61%) were male and the median age was 60 years old. The most common finding on chest x-rays was peripheral ground glass opacification involving more to the middle and lower zone in 82(53.2%) patients. Of the patients with ground glass opacification 22 (26.8%) had confirmed PCR results of them 16 (72.7%) were male and 6 (27.3%) were female. Conclusion Chest X-ray has limited sensitivity for COVID-19, however pattern detected on Chest X-ray can help to diagnose and guide treatment. Keywords PCR Chest X-ray COVID-19 ==== Body pmc	0	PMC9715993	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S47	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.153	oa_other
==== 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)00547-1 10.1016/j.jmir.2022.10.168 Article Post COVID 19 Experience among Radiographers in Tanzania Mkoloma S. S MTech, BTech, DDR 12 Burambo A. B MSc,BSc,BA (CAND), DDR 1 1 Ocean Road Cancer Institute (ORCI), Dar es Salaam; Tanzania 2 Muhimbili University of Health and Allied Sciences (MUHAS), Dar es Salaam; Tanzania 2 12 2022 12 2022 2 12 2022 53 4 S51S51 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 Radiography is essential for the initial diagnosis and monitoring patients affected with COVID 19. Despite that Tanzania did not experience a disastrous situation during pandemic, radiographers all around were left bewildered with safe working environment. This study aims to assess the experience of radiographers after the novel pandemic. Methods A questionnaire composed of 16 multiple choice questions was sent to 403 registered radiographers through emails and whatsApp where 169 responses were received. The data analysis was performed using SPSS software with the statistical significance assumed as p-value < 0.05 and quantitative data were presented in percentage, graphs and pie charts. Results Of the 169 responses; 79% were male, 82% were diploma holders and 45% rotate in multiple modalities where 44% are under 5 years’ experience .45% declared not impacted directly with the pandemic while 13% declared to know radiographers who were deceased of the pandemic. 68% were satisfied with country response to pandemic, 43% still receive patient suspected with COVID19 where 74% of hospitals have pandemic safe plan. Despite that 74% are vaccinated, 71% feels Tanzania is safe for radiographers. Conclusion Tanzania has so few to call a new normal as the pandemic did not affect the working culture. Majority of radiographers feels secured, however they expressed that there is a need to have enough PPE in store in case something else erupts and that more information is needed about safety of COVID 19 vaccines. Keywords 3-5 Pandemic COVID 19 Safety Radiographers Vaccines ==== Body pmc	35210177	PMC9715994	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S51	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.168	oa_other
==== 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)00394-0 10.1016/j.jmir.2022.10.015 Article Compliance with infection control and radiation protection measures during COVID-19 in the UAE's radiology department Abuzaid Mohamed Elshami Wiam Tekin HO Department of Medical Diagnostic Imaging, College of Health Sciences, University of Sharjah, Sharjah, United Arab Emirates 2 12 2022 12 2022 2 12 2022 53 4 S4S4 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 To guarantee patient safety and prevent the future transmission of the COVID-19 virus, radiology staff members must have solid knowledge, expertise and commitment to radiation protection and infection control practices. Compliance and adherence to infection control and radiation protection during COVID-19 radiography were examined in this study. Methods Online cross-sectional research was conducted using an electronic survey. The survey collected participants’ demographics, radiation protection compliance, and infection control practices during COVID-19 patients' radiography procedures. Results The participants adhered to patient protection and self-protection by 89.2% and 90.2%, respectively. The total adherence to radiation protection practices score was 80.2%. Older participants with more experience had significantly higher adherence scores (P = 0.0001). However, there was no discernible difference in adherence scores between the participants’ educational backgrounds. The individuals’ mean and standard deviation for infection control were 87.5% ± 16.28, respectively. Additionally, a large percentage of participants (95%) demonstrated good adherence to infection-control measures. Conclusion To promote adherence to and compliance with radiation safety and infection control, ongoing guidance, training, and follow-up is advised. To close the practice and knowledge gap, educational institutions and professional organisations must work together to offer structured training programmes for radiology practitioners. Keywords ALARA infection control radiation protection radiology department ==== Body pmc	0	PMC9715995	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S4	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.015	oa_other
==== 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)00394-0 10.1016/j.jmir.2022.10.015 Article Compliance with infection control and radiation protection measures during COVID-19 in the UAE's radiology department Abuzaid Mohamed Elshami Wiam Tekin HO Department of Medical Diagnostic Imaging, College of Health Sciences, University of Sharjah, Sharjah, United Arab Emirates 2 12 2022 12 2022 2 12 2022 53 4 S4S4 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 To guarantee patient safety and prevent the future transmission of the COVID-19 virus, radiology staff members must have solid knowledge, expertise and commitment to radiation protection and infection control practices. Compliance and adherence to infection control and radiation protection during COVID-19 radiography were examined in this study. Methods Online cross-sectional research was conducted using an electronic survey. The survey collected participants’ demographics, radiation protection compliance, and infection control practices during COVID-19 patients' radiography procedures. Results The participants adhered to patient protection and self-protection by 89.2% and 90.2%, respectively. The total adherence to radiation protection practices score was 80.2%. Older participants with more experience had significantly higher adherence scores (P = 0.0001). However, there was no discernible difference in adherence scores between the participants’ educational backgrounds. The individuals’ mean and standard deviation for infection control were 87.5% ± 16.28, respectively. Additionally, a large percentage of participants (95%) demonstrated good adherence to infection-control measures. Conclusion To promote adherence to and compliance with radiation safety and infection control, ongoing guidance, training, and follow-up is advised. To close the practice and knowledge gap, educational institutions and professional organisations must work together to offer structured training programmes for radiology practitioners. Keywords ALARA infection control radiation protection radiology department ==== Body pmc	36441101	PMC9715996	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S61	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.209	oa_other
==== 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)00394-0 10.1016/j.jmir.2022.10.015 Article Compliance with infection control and radiation protection measures during COVID-19 in the UAE's radiology department Abuzaid Mohamed Elshami Wiam Tekin HO Department of Medical Diagnostic Imaging, College of Health Sciences, University of Sharjah, Sharjah, United Arab Emirates 2 12 2022 12 2022 2 12 2022 53 4 S4S4 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 To guarantee patient safety and prevent the future transmission of the COVID-19 virus, radiology staff members must have solid knowledge, expertise and commitment to radiation protection and infection control practices. Compliance and adherence to infection control and radiation protection during COVID-19 radiography were examined in this study. Methods Online cross-sectional research was conducted using an electronic survey. The survey collected participants’ demographics, radiation protection compliance, and infection control practices during COVID-19 patients' radiography procedures. Results The participants adhered to patient protection and self-protection by 89.2% and 90.2%, respectively. The total adherence to radiation protection practices score was 80.2%. Older participants with more experience had significantly higher adherence scores (P = 0.0001). However, there was no discernible difference in adherence scores between the participants’ educational backgrounds. The individuals’ mean and standard deviation for infection control were 87.5% ± 16.28, respectively. Additionally, a large percentage of participants (95%) demonstrated good adherence to infection-control measures. Conclusion To promote adherence to and compliance with radiation safety and infection control, ongoing guidance, training, and follow-up is advised. To close the practice and knowledge gap, educational institutions and professional organisations must work together to offer structured training programmes for radiology practitioners. Keywords ALARA infection control radiation protection radiology department ==== Body pmc	0	PMC9715997	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S48-S49	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.159	oa_other
==== 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)00394-0 10.1016/j.jmir.2022.10.015 Article Compliance with infection control and radiation protection measures during COVID-19 in the UAE's radiology department Abuzaid Mohamed Elshami Wiam Tekin HO Department of Medical Diagnostic Imaging, College of Health Sciences, University of Sharjah, Sharjah, United Arab Emirates 2 12 2022 12 2022 2 12 2022 53 4 S4S4 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 To guarantee patient safety and prevent the future transmission of the COVID-19 virus, radiology staff members must have solid knowledge, expertise and commitment to radiation protection and infection control practices. Compliance and adherence to infection control and radiation protection during COVID-19 radiography were examined in this study. Methods Online cross-sectional research was conducted using an electronic survey. The survey collected participants’ demographics, radiation protection compliance, and infection control practices during COVID-19 patients' radiography procedures. Results The participants adhered to patient protection and self-protection by 89.2% and 90.2%, respectively. The total adherence to radiation protection practices score was 80.2%. Older participants with more experience had significantly higher adherence scores (P = 0.0001). However, there was no discernible difference in adherence scores between the participants’ educational backgrounds. The individuals’ mean and standard deviation for infection control were 87.5% ± 16.28, respectively. Additionally, a large percentage of participants (95%) demonstrated good adherence to infection-control measures. Conclusion To promote adherence to and compliance with radiation safety and infection control, ongoing guidance, training, and follow-up is advised. To close the practice and knowledge gap, educational institutions and professional organisations must work together to offer structured training programmes for radiology practitioners. Keywords ALARA infection control radiation protection radiology department ==== Body pmc	0	PMC9715998	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S45-S46	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.148	oa_other
==== 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)00394-0 10.1016/j.jmir.2022.10.015 Article Compliance with infection control and radiation protection measures during COVID-19 in the UAE's radiology department Abuzaid Mohamed Elshami Wiam Tekin HO Department of Medical Diagnostic Imaging, College of Health Sciences, University of Sharjah, Sharjah, United Arab Emirates 2 12 2022 12 2022 2 12 2022 53 4 S4S4 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 To guarantee patient safety and prevent the future transmission of the COVID-19 virus, radiology staff members must have solid knowledge, expertise and commitment to radiation protection and infection control practices. Compliance and adherence to infection control and radiation protection during COVID-19 radiography were examined in this study. Methods Online cross-sectional research was conducted using an electronic survey. The survey collected participants’ demographics, radiation protection compliance, and infection control practices during COVID-19 patients' radiography procedures. Results The participants adhered to patient protection and self-protection by 89.2% and 90.2%, respectively. The total adherence to radiation protection practices score was 80.2%. Older participants with more experience had significantly higher adherence scores (P = 0.0001). However, there was no discernible difference in adherence scores between the participants’ educational backgrounds. The individuals’ mean and standard deviation for infection control were 87.5% ± 16.28, respectively. Additionally, a large percentage of participants (95%) demonstrated good adherence to infection-control measures. Conclusion To promote adherence to and compliance with radiation safety and infection control, ongoing guidance, training, and follow-up is advised. To close the practice and knowledge gap, educational institutions and professional organisations must work together to offer structured training programmes for radiology practitioners. Keywords ALARA infection control radiation protection radiology department ==== Body pmc	0	PMC9715999	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S58	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.200	oa_other
==== 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)00394-0 10.1016/j.jmir.2022.10.015 Article Compliance with infection control and radiation protection measures during COVID-19 in the UAE's radiology department Abuzaid Mohamed Elshami Wiam Tekin HO Department of Medical Diagnostic Imaging, College of Health Sciences, University of Sharjah, Sharjah, United Arab Emirates 2 12 2022 12 2022 2 12 2022 53 4 S4S4 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 To guarantee patient safety and prevent the future transmission of the COVID-19 virus, radiology staff members must have solid knowledge, expertise and commitment to radiation protection and infection control practices. Compliance and adherence to infection control and radiation protection during COVID-19 radiography were examined in this study. Methods Online cross-sectional research was conducted using an electronic survey. The survey collected participants’ demographics, radiation protection compliance, and infection control practices during COVID-19 patients' radiography procedures. Results The participants adhered to patient protection and self-protection by 89.2% and 90.2%, respectively. The total adherence to radiation protection practices score was 80.2%. Older participants with more experience had significantly higher adherence scores (P = 0.0001). However, there was no discernible difference in adherence scores between the participants’ educational backgrounds. The individuals’ mean and standard deviation for infection control were 87.5% ± 16.28, respectively. Additionally, a large percentage of participants (95%) demonstrated good adherence to infection-control measures. Conclusion To promote adherence to and compliance with radiation safety and infection control, ongoing guidance, training, and follow-up is advised. To close the practice and knowledge gap, educational institutions and professional organisations must work together to offer structured training programmes for radiology practitioners. Keywords ALARA infection control radiation protection radiology department ==== Body pmc	0	PMC9716000	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S37	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.121	oa_other
==== 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)00394-0 10.1016/j.jmir.2022.10.015 Article Compliance with infection control and radiation protection measures during COVID-19 in the UAE's radiology department Abuzaid Mohamed Elshami Wiam Tekin HO Department of Medical Diagnostic Imaging, College of Health Sciences, University of Sharjah, Sharjah, United Arab Emirates 2 12 2022 12 2022 2 12 2022 53 4 S4S4 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 To guarantee patient safety and prevent the future transmission of the COVID-19 virus, radiology staff members must have solid knowledge, expertise and commitment to radiation protection and infection control practices. Compliance and adherence to infection control and radiation protection during COVID-19 radiography were examined in this study. Methods Online cross-sectional research was conducted using an electronic survey. The survey collected participants’ demographics, radiation protection compliance, and infection control practices during COVID-19 patients' radiography procedures. Results The participants adhered to patient protection and self-protection by 89.2% and 90.2%, respectively. The total adherence to radiation protection practices score was 80.2%. Older participants with more experience had significantly higher adherence scores (P = 0.0001). However, there was no discernible difference in adherence scores between the participants’ educational backgrounds. The individuals’ mean and standard deviation for infection control were 87.5% ± 16.28, respectively. Additionally, a large percentage of participants (95%) demonstrated good adherence to infection-control measures. Conclusion To promote adherence to and compliance with radiation safety and infection control, ongoing guidance, training, and follow-up is advised. To close the practice and knowledge gap, educational institutions and professional organisations must work together to offer structured training programmes for radiology practitioners. Keywords ALARA infection control radiation protection radiology department ==== Body pmc	35618399	PMC9716001	NO-CC CODE	2022-12-09 23:15:06	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S3-S4	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.012	oa_other
==== 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)00394-0 10.1016/j.jmir.2022.10.015 Article Compliance with infection control and radiation protection measures during COVID-19 in the UAE's radiology department Abuzaid Mohamed Elshami Wiam Tekin HO Department of Medical Diagnostic Imaging, College of Health Sciences, University of Sharjah, Sharjah, United Arab Emirates 2 12 2022 12 2022 2 12 2022 53 4 S4S4 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 To guarantee patient safety and prevent the future transmission of the COVID-19 virus, radiology staff members must have solid knowledge, expertise and commitment to radiation protection and infection control practices. Compliance and adherence to infection control and radiation protection during COVID-19 radiography were examined in this study. Methods Online cross-sectional research was conducted using an electronic survey. The survey collected participants’ demographics, radiation protection compliance, and infection control practices during COVID-19 patients' radiography procedures. Results The participants adhered to patient protection and self-protection by 89.2% and 90.2%, respectively. The total adherence to radiation protection practices score was 80.2%. Older participants with more experience had significantly higher adherence scores (P = 0.0001). However, there was no discernible difference in adherence scores between the participants’ educational backgrounds. The individuals’ mean and standard deviation for infection control were 87.5% ± 16.28, respectively. Additionally, a large percentage of participants (95%) demonstrated good adherence to infection-control measures. Conclusion To promote adherence to and compliance with radiation safety and infection control, ongoing guidance, training, and follow-up is advised. To close the practice and knowledge gap, educational institutions and professional organisations must work together to offer structured training programmes for radiology practitioners. Keywords ALARA infection control radiation protection radiology department ==== Body pmc	0	PMC9716002	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S20	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.068	oa_other
==== 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)00524-0 10.1016/j.jmir.2022.10.145 Article The new normal in clinical radiography education and practice in a Nigerian population post-COVID 19 Okpaleke M Okechukwu L Ugwuanyi D Ugwu A Chiegwu H Department of Radiography, Nnamdi Azikiwe University, Nnewi Campus, Anambra State, Nigeria 2 12 2022 12 2022 2 12 2022 53 4 S45S45 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 The COVID 19 pandemic affected various sectors of the global economy with consequences for clinical radiography practice and training. The objective of the study is to determine the availability of equipment/accessories and the attitude of radiographers toward infection control/ hygiene before and after the second wave of the COVID 19 pandemic. Materials and Methods This cross-sectional prospective research survey used questionnaires divided into two sections. Sections A and B were used to elicit the participants’ socio-demographic data and the availability of equipment/accessories as well as the hygiene/infection control practices by radiographers in two tertiary hospitals in Anambra state of Nigeria. The items in the questionnaire were reviewed by a panel of two experts for reproducibility and validity. The questionnaires were distributed to a sample of 80 radiographers selected by convenience sampling from a population of 90 radiographers that met the inclusion criteria. Descriptive statistics were used for analyzing the questionnaires at a 5% level of significance using the SPSS version 23. Results 30(33.3%) and 60 (66.7%) of the respondents were female and males respectively.80(89%), 0(0%), 10(11%), 20(22%) and 77(85.6%), 85(94.4%), 24(26.7%), and 69(76.7%) of the respondents used running water, alcohol sanitizers and disposable towels/wipes to clean their hands or disinfected the radiography equipment/accessories after attending to a patient before and after Covid 19 pandemic respectively. The availability and use of accessories/personnel protective equipment when attending to a suspected Covid 19 or infectious patient were 20(22%) pre- covid 19 and 84(93%) post- covid 19. Clinical radiography lectures and examinations for student radiographers involve social distancing 90(100%), use of face masks 90(100%), and online 70(78%), respectively. Conclusion There is a transition to online lectures for clinical radiography students and improved personnel hygiene and cleanliness of radiography equipment/ accessories in the studied population post- COVID 19. Keywords radiographers COVID -19 practice training ==== Body pmc	34922879	PMC9716003	NO-CC CODE	2022-12-09 23:15:06	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S44-S45	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.145	oa_other
==== 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)00445-3 10.1016/j.jmir.2022.10.066 Article The Influence of COVID-19 on Ultrasound Practice Anthea Chiang BMRSc 12 Marcia Smoke RTT, ACT, MSc 1 Emily Ho MRT(T), RTT, BMRSc 3 Meaghan Jefferson MRT(DMS) 2 Tom Dr. Farrell PhD, FCCPM 13 1 McMaster University, Hamilton, Ontario, Canada 2 Mohawk College, Hamilton, Ontario, Canada 3 Juravinski Cancer Centre, Hamilton, Ontario, Canada 2 12 2022 12 2022 2 12 2022 53 4 S19S19 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 COVID-19 had a major impact on ultrasound practice. The recommended “safe distance” of 2 meters (to limit COVID-19 infection) cannot be maintained during standard ultrasound procedures. Thus, special precautions, proper infection prevention, and control measures were implemented to limit the spread. The purpose of this study was to assess the impact of the pandemic on ultrasound technologists and their work practices and to suggest future changes that may be implemented. Methods This Research Ethics Board (REB) approved study included a quantitative survey which was designed based on a literature review. The survey used questions with a 5-point Likert scale along with multiple-choice questions. For the statistical analyses, the Wilcoxon signed-rank test, the Kruskall-Wallis test and Spearman's coefficient correlation test were used. Results There were 40/100 respondents. All but five questions produced significant results. Enhanced cleaning (p < 0.02) and increased wait time (p < 0.02) were found to be dependent on specific institutions. Additionally, the majority of case types performed by a sonographer (p < 0.02) and the years of experience (p = 0.006) influenced perception of a permanent change of practice. Although procedures have taken longer during the pandemic, sonographers do not anticipate an increase in procedure time for future scheduling. Sonographers experienced an increase in stress levels, causing it to be less manageable than before the pandemic (p = 0.0003). Wearing PPE was identified as a permanent change in practice whereas increasing the use of mobile ultrasound was not highlighted as future practice. Conclusion Respondents reported challenges getting support for initiatives to relieve sonographer stress. If factors such as adequate communication and workload are not properly addressed, there can be negative psychological effects for sonographers. Many changes were suggested to be implemented to ensure sonographers feel supported so that their workload is manageable. Keywords practice change psychological stress PPE ==== Body pmc	0	PMC9716004	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S19	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.066	oa_other
==== 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)00452-0 10.1016/j.jmir.2022.10.073 Article National survey for investigating the diagnostic imaging departments reorganization and management during the COVID-19 pandemic Roletto Andrea 1 Catania D 2 Ciaralli C 3 Cozzi A 1 Di Feo D 4 Durante S 5 Pasini D 6 Raiano N 7 Zanardo M 1 1 Department of Biomedical Sciences for Health, University of Milan, Milan, Italy 2 Directorate General for Healthcare Professions, San Raffaele Research Hospital, Milan, Italy 3 Radiology department, Lazzaro Spallanzani National Institute for Infectious Diseases, Rome, Italy 4 Radiology Department, Meyer Children's Hospital, Florence, Italy 5 Directorate of the Nursing, Technical and Rehabilitation Assistance Service, Rizzoli Orthopaedic Institute, Bologna, Italy 6 Directorate General for Healthcare Professions, Agostino Gemelli University Policlinic, Rome, Italy 7 Radiology and Radiotherapy Department, Research Cancer Center “Pascale Foundation”, Naples, Italy 2 12 2022 12 2022 2 12 2022 53 4 S21S22 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 The COVID-19 pandemic has had a profound impact on radiography services globally. The objective of this study was to evaluate the impact of the COVID-19 pandemic on the management of radiology departments in Italy. Methods An online survey with 32 questions was developed and promoted by the Italian Federation of Scientific Radiographer Societies (FASTeR) and sent to all Radiology Service Managers (RSM) identified in the RSM committee of the Italian Federation of Radiographers and Health Professionals, counting for 39 Italian RSM, representing more than 1,200 radiographers. The survey included questions regarding RSM demographics data, the number of radiographers and specialties managed, the effects of the pandemic on the diagnostic imaging service, and any reorganizations that had been implemented, such as the partial or total suspension of diagnostic activities and the number of radiographers tested as positive to COVID-19. Results Twenty (52%) RSM from different Italian regions completed the questionnaire. A total of 70% of respondents had implemented reorganizations in terms of space, equipment, and pathways dedicated to COVID-19-infected patients, including an extension of the timing of acquisition of the exams. More than half of the respondents reported breast and DXA imaging unit had suffered the most suspension of activities. 70% of respondents reported that more than 50% of radiographers were resulted as COVID-19 positive. Conclusion These data show how challenging was of the reorganization of Italian diagnostic imaging departments during the COVID-19 pandemic, with impact on the suspension of some exams and the rescheduling of breast and DXA imaging. The reorganization of the services also had to consider the high number of radiographers suspended from activity due to the positivity to COVID-19, and the lengthening of the duration of the examinations due to the sanitation of the spaces. Keywords COVID-19 Emergency Management Radiology Survey ==== Body pmc	0	PMC9716005	NO-CC CODE	2022-12-09 23:15:06	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S22	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.073	oa_other
==== 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)00452-0 10.1016/j.jmir.2022.10.073 Article National survey for investigating the diagnostic imaging departments reorganization and management during the COVID-19 pandemic Roletto Andrea 1 Catania D 2 Ciaralli C 3 Cozzi A 1 Di Feo D 4 Durante S 5 Pasini D 6 Raiano N 7 Zanardo M 1 1 Department of Biomedical Sciences for Health, University of Milan, Milan, Italy 2 Directorate General for Healthcare Professions, San Raffaele Research Hospital, Milan, Italy 3 Radiology department, Lazzaro Spallanzani National Institute for Infectious Diseases, Rome, Italy 4 Radiology Department, Meyer Children's Hospital, Florence, Italy 5 Directorate of the Nursing, Technical and Rehabilitation Assistance Service, Rizzoli Orthopaedic Institute, Bologna, Italy 6 Directorate General for Healthcare Professions, Agostino Gemelli University Policlinic, Rome, Italy 7 Radiology and Radiotherapy Department, Research Cancer Center “Pascale Foundation”, Naples, Italy 2 12 2022 12 2022 2 12 2022 53 4 S21S22 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 The COVID-19 pandemic has had a profound impact on radiography services globally. The objective of this study was to evaluate the impact of the COVID-19 pandemic on the management of radiology departments in Italy. Methods An online survey with 32 questions was developed and promoted by the Italian Federation of Scientific Radiographer Societies (FASTeR) and sent to all Radiology Service Managers (RSM) identified in the RSM committee of the Italian Federation of Radiographers and Health Professionals, counting for 39 Italian RSM, representing more than 1,200 radiographers. The survey included questions regarding RSM demographics data, the number of radiographers and specialties managed, the effects of the pandemic on the diagnostic imaging service, and any reorganizations that had been implemented, such as the partial or total suspension of diagnostic activities and the number of radiographers tested as positive to COVID-19. Results Twenty (52%) RSM from different Italian regions completed the questionnaire. A total of 70% of respondents had implemented reorganizations in terms of space, equipment, and pathways dedicated to COVID-19-infected patients, including an extension of the timing of acquisition of the exams. More than half of the respondents reported breast and DXA imaging unit had suffered the most suspension of activities. 70% of respondents reported that more than 50% of radiographers were resulted as COVID-19 positive. Conclusion These data show how challenging was of the reorganization of Italian diagnostic imaging departments during the COVID-19 pandemic, with impact on the suspension of some exams and the rescheduling of breast and DXA imaging. The reorganization of the services also had to consider the high number of radiographers suspended from activity due to the positivity to COVID-19, and the lengthening of the duration of the examinations due to the sanitation of the spaces. Keywords COVID-19 Emergency Management Radiology Survey ==== Body pmc	0	PMC9716006	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S57-S58	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.199	oa_other
==== 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)00452-0 10.1016/j.jmir.2022.10.073 Article National survey for investigating the diagnostic imaging departments reorganization and management during the COVID-19 pandemic Roletto Andrea 1 Catania D 2 Ciaralli C 3 Cozzi A 1 Di Feo D 4 Durante S 5 Pasini D 6 Raiano N 7 Zanardo M 1 1 Department of Biomedical Sciences for Health, University of Milan, Milan, Italy 2 Directorate General for Healthcare Professions, San Raffaele Research Hospital, Milan, Italy 3 Radiology department, Lazzaro Spallanzani National Institute for Infectious Diseases, Rome, Italy 4 Radiology Department, Meyer Children's Hospital, Florence, Italy 5 Directorate of the Nursing, Technical and Rehabilitation Assistance Service, Rizzoli Orthopaedic Institute, Bologna, Italy 6 Directorate General for Healthcare Professions, Agostino Gemelli University Policlinic, Rome, Italy 7 Radiology and Radiotherapy Department, Research Cancer Center “Pascale Foundation”, Naples, Italy 2 12 2022 12 2022 2 12 2022 53 4 S21S22 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 The COVID-19 pandemic has had a profound impact on radiography services globally. The objective of this study was to evaluate the impact of the COVID-19 pandemic on the management of radiology departments in Italy. Methods An online survey with 32 questions was developed and promoted by the Italian Federation of Scientific Radiographer Societies (FASTeR) and sent to all Radiology Service Managers (RSM) identified in the RSM committee of the Italian Federation of Radiographers and Health Professionals, counting for 39 Italian RSM, representing more than 1,200 radiographers. The survey included questions regarding RSM demographics data, the number of radiographers and specialties managed, the effects of the pandemic on the diagnostic imaging service, and any reorganizations that had been implemented, such as the partial or total suspension of diagnostic activities and the number of radiographers tested as positive to COVID-19. Results Twenty (52%) RSM from different Italian regions completed the questionnaire. A total of 70% of respondents had implemented reorganizations in terms of space, equipment, and pathways dedicated to COVID-19-infected patients, including an extension of the timing of acquisition of the exams. More than half of the respondents reported breast and DXA imaging unit had suffered the most suspension of activities. 70% of respondents reported that more than 50% of radiographers were resulted as COVID-19 positive. Conclusion These data show how challenging was of the reorganization of Italian diagnostic imaging departments during the COVID-19 pandemic, with impact on the suspension of some exams and the rescheduling of breast and DXA imaging. The reorganization of the services also had to consider the high number of radiographers suspended from activity due to the positivity to COVID-19, and the lengthening of the duration of the examinations due to the sanitation of the spaces. Keywords COVID-19 Emergency Management Radiology Survey ==== Body pmc	0	PMC9716007	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S50-S51	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.165	oa_other
==== 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	PMC9716008	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S49	utf-8	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.160	oa_other
==== 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	PMC9716009	NO-CC CODE	2022-12-03 23:20:52	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S59	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.203	oa_other
==== 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	PMC9716010	NO-CC CODE	2022-12-09 23:15:06	no	J Med Imaging Radiat Sci. 2022 Dec 2; 53(4):S9	latin-1	J Med Imaging Radiat Sci	2,022	10.1016/j.jmir.2022.10.030	oa_other

Subsets and Splits

No community queries yet

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