title
stringlengths 1
1.19k
| keywords
stringlengths 0
668
| concept
stringlengths 0
909
| paragraph
stringlengths 0
61.8k
| PMID
stringlengths 10
11
|
|---|---|---|---|---|
Limitations | cancer, death | CANCER, TUMOR GROWTH, RARE DISEASE, COMPLICATION | This study has some limitations. The trial was initially planned for enrollment of 30 patients to have complete data for 24 patients, providing 80% power for the primary end point. However, accrual in this rare disease was slow, with only 24 patients enrolled after 8 years (2013 to 2021). When designed in 2012, there was concern that delaying start of treatment would harm patients, thus the only prespecified interim analyses concerned increased death or severe complication in the observation first group. Although not prespecified, since the trial was not blinded, interim efficacy analysis was performed after 8 years; given complete lack of response to chemotherapy and no difference in tumor growth between observation and treatment periods, the trial was closed, as it was felt unethical to continuing treating patients with low-grade AA with fluorouracil-based chemotherapy. This trial was under the institutional data and safety monitoring board oversight for annual review of safety and efficacy. With the data and safety monitoring board’s permission, the trial was administratively closed. All of the patients in the study were enrolled at a tertiary referral cancer center, which may not be representative of patients in a community oncology practice. Despite these limitations, our study represents the first prospective, randomized trial for low-grade AA, to our knowledge. | PMC10236240 |
Conclusions | CRS, CRC | DISEASE, APPENDICEAL CANCER | The findings of this randomized crossover trial, taking into consideration the lack of benefit from fluoropyrimidine-based chemotherapy seen in this trial and prior retrospective studies with similar conclusion, suggest that fluoropyrimidine-based chemotherapy should not be considered a standard-of-care treatment for patients with low-grade AA who are not candidates for CRS. Clinical trials investigating novel therapeutics should be considered for these patients. These prospective clinical data highlight the differences between AA and CRC and demonstrate the need for the development of appendiceal cancer–specific guidelines as well as more preclinical and clinical investigation for this disease. | PMC10236240 |
2. Materials and Methods | PMC10744547 |
|||
2.1. Study Design | This prospective randomized controlled trial protocol was registered with ClinicalTrials.gov under the registration number NCT03310060. Additionally, it received approval from the institutional review board and support from Chang Gung Memorial Hospital (IRB no. 201700271A3). Before surgery, informed consent was obtained from all participants. The manuscript was prepared following the Consolidated Standards of Reporting Trials (CONSORT) guidelines (as detailed in | PMC10744547 |
||
2.2. Inclusion and Exclusion Criteria | rheumatoid arthritis, abnormal coagulation function, abnormal coagulation, end-stage arthritis, allergic reaction, liver cirrhosis, anaphylactic shock, knee joints, non-OA degenerative, end-stage renal failure | RHEUMATOID ARTHRITIS, ABNORMAL COAGULATION, ALLERGIC REACTION, SEPTIC ARTHRITIS, LIVER CIRRHOSIS, ANAPHYLACTIC SHOCK, END-STAGE RENAL FAILURE, SEVERE ANEMIA | Between September 2017 and August 2019, we consecutively assessed a series of 119 patients who underwent unilateral primary TKA for initial eligibility in this study. The inclusion criteria encompassed adult patients aged 20 to 80 years with end-stage arthritis of the knee joint necessitating TKA. The exclusion criteria for this study were as follows: patients with non-OA degenerative knee joints (e.g., rheumatoid arthritis or septic arthritis), patients with severe anemia (hemoglobin/hematocrit (Hb/Hct) < 12 g/dL/ < 36% before surgery), patients with end-stage renal failure undergoing dialysis treatment, liver cirrhosis leading to abnormal coagulation function (Child A/B/C conditions, platelets < 100,000/µL), patients using any oral anticoagulants or antiplatelets, patients with abnormal coagulation function (INR > 1.5 s), body mass index (BMI) exceeding the standard (35 < BMI < 18), a history of anaphylactic shock or severe allergic reaction to bovine protein (aprotinin), and women who were pregnant and/or breastfeeding. | PMC10744547 |
2.3. Intervention Protocol | TKAs | HAAS | The patients were randomly assigned to two groups: the TXA group and the Tisseel@ + TXA group. All TKAs were performed by the same experienced surgical team (H.N.S., Y.H.C, and C.C.H.) using the mini-midvastus approach described by Haas et al., under general anesthesia [In the TXA group, patients received an intravenous injection of 15 mg/kg of TXA (Transamin 100 mg/mL; China Chemical and Pharmaceutical Co., Taipei City, Taiwan) before the surgical incision in the operating theater. An additional dose of IV TXA was administered 3 h after the surgery. Patients in the Tisseel@ + TXA group received the same IV TXA protocol before and after surgery. Concurrently with the surgery, a 4 mL Tisseel@ (Baxter | PMC10744547 |
2.4. Outcomes | blood loss, hypovolemic shock | BLOOD LOSS, HYPOVOLEMIC SHOCK | Originally, the protocol for the study designated only the total blood loss as the primary outcome. However, after team discussions, it was deemed meaningful to explore potential differences between the two groups in terms of blood transfusion rate, decrease in Hb level, and calculated blood loss. Consequently, we decided to include blood transfusion rate, decrease in Hb level, calculated blood loss, and invisibly estimated total blood loss as primary outcomes in our study design. Allogeneic blood transfusion of red blood cells was triggered when the hemoglobin (Hb) level dropped below 8 g/dL, or when any signs of hypovolemic shock were observed. We calculated the patient’s blood volume (PBV) based on Nadler’s formula [To further assess the practical significance of the observed differences between the TXA group and the Tisseel@ + TXA group, we calculated Cohen’s d as a measure of effect size. This effect size will provide additional insights into the magnitude of the observed differences in blood loss between the two treatment groups. | PMC10744547 |
2.5. Randomization and Blinding | BLIND | The randomization was conducted by an independent research assistant using a computer-generated method, based on the sequence of operation dates. The clinical investigators remained blind to the randomization and allocation of all patients until complete data had been collected. | PMC10744547 |
|
2.6. Minimum Sample Size | blood loss, TKA | BLOOD LOSS | The determination of the sample size was based on the research conducted by Molloy et al., who conducted a prospective randomized trial to assess perioperative blood loss following TKA [ | PMC10744547 |
2.7. Statistical Analysis | Statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) software (Version 22.0; SPSS Inc., Chicago, IL, USA). The analysis of continuous variables utilized Student’s | PMC10744547 |
||
3. Results | PMC10744547 |
|||
3.1. Participants’ Flow | The participant flow is illustrated in | PMC10744547 |
||
3.2. Baseline Data | The preoperative demographics of the patients, including age, gender, body mass index (BMI), preoperative hemoglobin (Hb) level, hematocrit (Hct), international normalized ratio (INR), platelet count, and American Society of Anesthesiologists (ASA) grade, underwent comparative analysis between the two groups ( | PMC10744547 |
||
3.3. Primary and Secondary Outcomes | swelling, blood loss, periprosthetic joint infection, pain, ecchymosis, hematoma | BLOOD LOSS, ECCHYMOSIS, HEMATOMA, DEEP-VEIN THROMBOSIS, COMPLICATIONS | Among the participants, none required blood transfusions perioperatively, resulting in a transfusion rate of zero in both groups. In the Tisseel@ + TXA group, the average estimated blood loss was 0.463 ± 0.2422 L, which was comparable to the TXA group’s value of 0.455 ± 0.2522 L (The assessment of perioperative wound conditions, from immediately after the surgery to six weeks post-operation, included evaluating ecchymosis, swelling caused by hematoma, wound-healing issues, and isolated occurrences of periprosthetic joint infection. The incidence of these conditions was found to be comparable between the studied groups. No patient in either group required a return to the operating theater due to any potential complications until two years postoperatively. A single patient experienced calf swelling with minimal pain at the 2-week postoperative follow-up, suspected to be deep-vein thrombosis. However, the patient declined aggressive management, even after duplex ultrasonography confirmation. Fortunately, the symptoms and signs had significantly improved at the 6-week follow-up, with no further complications (The effect size, as indicated by Cohen’s d, for the estimated total blood loss was 0.032382, while the effect size for calculated blood was 0.094433. Both of these findings suggest a small effect ( | PMC10744547 |
4. Discussion | blood loss, postoperative blood loss, TKA, trauma | WORLD WAR II, POSTOPERATIVE COMPLICATIONS, BLOOD LOSS, BLOOD CLOTS, COMPLICATION, BLEEDING REQUIRING TRANSFUSION | Making every effort to improve the reduction in blood loss during and after surgical procedures is a fundamental aspect of all surgeries. According to Gao et al., IV TXA has shown a potential association with reduced blood transfusion requirements and higher hemoglobin levels in patients undergoing TKA. Importantly, this benefit was observed without an increase in the incidence of postoperative complications compared to the use of topically applied fibrin sealants. Moreover, there was no significant difference noted in the total calculated blood loss between the two groups [The findings from our prospective randomized controlled trial indicated that the incidence of blood transfusion was zero in both the combined Tisseel@ + IV TXA group and the IV TXA-only group. In the Tisseel@ + TXA group, the average estimated blood loss was 0.463 ± 0.2422 L, which was comparable to the TXA group’s value of 0.455 ± 0.2522 L. Additionally, postoperative calculated blood loss was comparable between the two groups, with values of 0.259 ± 0.1 L and 0.268 ± 0.108 L, respectively. These results suggest that no synergistic effect was observed when adding fibrin sealants (Tisseel@) to IV TXA in patients undergoing TKA. The absence of statistically significant differences in estimated total blood loss and calculated total blood loss between the two groups may be further clarified by examining Cohen’s d. The effect size is negligible and reinforces the idea that the addition of Tisseel@ may not provide a meaningful advantage over IV TXA in the context of blood conservation during TKA.The primary objective of achieving ERAS for TKA is to implement a multimodal approach that minimizes surgical-related trauma and stress. This approach involves employing the least invasive surgical techniques and practices. Nevertheless, the reduction in blood loss during and after surgical procedures is one of the key elements of ERAS. Proper intraoperative management significantly contributes to ensuring safety, minimizing blood loss, and promoting early recovery. In the American College of Surgeons’ National Surgical Quality Improvement Program database, bleeding requiring transfusion was the most common complication in TKA patients who failed the ERAS protocol [TXA is a synthetic anti-fibrinolytic medication that acts as a competitive inhibitor, specifically blocking the lysine-binding sites of plasminogen, plasmin, and tissue plasminogen activator. By doing so, TXA effectively retards the process of fibrinolysis and prevents the breakdown of blood clots [The utilization of fibrin sealants, which are created by mixing plasma fibrinogen with thrombin, was initially reported during World War II [In the current study, only a single dosage of the fibrin sealant Tisseel@ was used in our trial. According to a meta-analysis published by Yang et al., the dosages in nine randomized control trials and four prospective comparative studies ranged from 2 mL to 10 mL of fibrin sealant. Their conclusion was that the utilization of fibrin sealant may lead to a reduction in the transfusion rate and the number of transfusion units required after TKA. However, the impact in terms of reducing the total blood loss was not significant [Our study has several limitations. First, the sample size in each group was relatively small. Nevertheless, in numerous prospective randomized controlled trials examining the variance in postoperative blood loss after TKA between groups, significant findings have been observed with fewer than 50 patients in each group [ | PMC10744547 |
Supplementary Materials | The following supporting information can be downloaded at: Click here for additional data file. | PMC10744547 |
||
Author Contributions | Conceptualization, Y.-H.C., H.-N.S. and C.-C.H.; methodology, Y.-H.C., H.-N.S. and C.-C.H.; formal analysis, C.-H.C.; investigation, C.-H.C.; data curation, C.-H.L., C.-H.C. and C.-C.H.; writing—original draft preparation, C.-H.L., C.-H.C. and C.-C.H.; writing—review and editing, C.-H.L., C.-H.C. and C.-C.H.; visualization, C.-H.L.; supervision, H.-N.S.; project administration, Y.-H.C. and C.-C.H. All authors have read and agreed to the published version of the manuscript. | PMC10744547 |
||
Institutional Review Board Statement | The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Chang Gung Memorial Hospital (protocol code 201700271A3 and date of approval 5 July 2017). | PMC10744547 |
||
Informed Consent Statement | Informed consent was obtained from all subjects involved in the study. | PMC10744547 |
||
Data Availability Statement | The data presented in this study are available upon request from the corresponding author. The data are not publicly available due to privacy. | PMC10744547 |
||
Conflicts of Interest | The authors declare no conflict of interest. | PMC10744547 |
||
References | SECONDARY, PERIOPERATIVE COMPLICATION | CONSORT 2010 participant flow diagram.Demographic data.Primary and secondary outcomes in both groups.Wound condition and perioperative complications.Effect size measured by Cohen’s d. | PMC10744547 |
|
Rationale | ECOPD, COPD | PULMONARY HYPERTENSION, COPD | Pulmonary hypertension (PH) in COPD confers increased risk of exacerbations (ECOPD). Electrocardiogram (ECG) indicators of PH are prognostic both in PH and COPD. In the Beta-Blockers for the Prevention of Acute Exacerbations of COPD (BLOCK-COPD) trial, metoprolol increased risk of severe ECOPD through unclear mechanisms. | PMC10634074 |
Objective | ECOPD | We evaluated whether an ECG indicator of PH, P-pulmonale, would be associated with ECOPD and whether participants with P-pulmonale randomized to metoprolol were at higher risk of ECOPD and worsened respiratory symptoms given the potential detrimental effects of beta-blockers in PH. | PMC10634074 |
|
Methods | ECOPD | ENLARGEMENT | ECGs of 501 participants were analyzed for P-pulmonale (P wave enlargement in lead II). Cox proportional hazards models evaluated for associations between P-pulmonale and time to ECOPD (all and severe) for all participants and by treatment assignment (metoprolol vs. placebo). Linear mixed-effects models evaluated the association between treatment assignment and P-pulmonale on change in symptom scores (measured by CAT and SOBQ). | PMC10634074 |
Results | ECOPD | We identified no association between P-pulmonale and risk of any ECOPD or severe ECOPD. However, in individuals with P-pulmonale, metoprolol was associated with increased risk for ECOPD (aHR 2.92, 95% CI: 1.45–5.85). There was no association between metoprolol and ECOPD in individuals without P-pulmonale (aHR 1.01, 95% CI: 0.77–1.31). Individuals with P-pulmonale assigned to metoprolol experienced worsening symptoms (mean increase of 3.95, 95% CI: 1.32–6.58) whereas those assigned to placebo experienced a mean improvement in CAT score of -2.45 (95% CI: -0.30- -4.61). | PMC10634074 |
|
Conclusions | In individuals with P-pulmonale, metoprolol was associated with increased exacerbation risk and worsened symptoms. These findings may explain the findings observed in BLOCK-COPD. | PMC10634074 |
||
Supplementary Information | The online version contains supplementary material available at 10.1186/s12890-023-02748-2. | PMC10634074 |
||
Keywords | PMC10634074 |
|||
Introduction | exacerbations, hypoxia, PH, ECOPD, deaths, Shortness of Breath, COPD | ISCHEMIC HEART DISEASE, CHRONIC OBSTRUCTIVE PULMONARY DISEASE, HYPOXIA, CARDIOVASCULAR MORBIDITY, COPD, RHD, PULMONARY HYPERTENSION | Chronic obstructive pulmonary disease (COPD) afflicts over 65 million people worldwide, resulting in 3 million deaths annually, and in the United States, direct medical costs average over ten thousand dollars annually per patient [Cardiovascular morbidity in COPD patients extends beyond the well-established link with ischemic heart disease. In COPD patients, pulmonary hypertension (PH) related to destruction of lung tissue and chronic hypoxia is relatively common with estimates of PH prevalence using cardiac catheterization ranging from 20 to 36%, and up to 45% with exercise [The Beta-Blockers for the Prevention of Acute Exacerbations of COPD (BLOCK-COPD) trial randomized participants to metropolol vs. placebo for the prevention of ECOPD. Metoprolol did not affect the overall risk of ECOPD, but increased the risk of severe exacerbations requiring hospitalization and worsened respiratory symptoms as assessed by the COPD Assessment Test (CAT) and the San Diego Shortness of Breath Questionairre (SOBQ) [Previous work revealed that P-pulmonale, an ECG indicator of RHD, is both prevalent and associated with increased mortality in PH [We used data from BLOCK-COPD to test the hypothesis that P-pulmonale, an indicator of RHD in PH, is associated with ECOPD and would identify individuals at further risk for increased symptoms and ECOPD when randomized to metoprolol. | PMC10634074 |
Methods | PMC10634074 |
|||
BLOCK-COPD | COPD | COPD | BLOCK-COPD (NCT02587351) was a prospective, multi-center, placebo-controlled, double-blind, randomized trial of spirometry confirmed COPD patients aged 40–85 years [ | PMC10634074 |
Participants | All BLOCK-COPD participants had a standard 12-lead ECG performed at enrollment. We excluded participants with ECGs that were uninterpretable due to baseline artifact that obscured P waves ( | PMC10634074 |
||
Exacerbations of COPD | chest tightness, dyspnea, wheezing, cough, ECOPD, COPD | COPD, EVENTS | An exacerbation of COPD was defined as an increase in or new onset of two or more of the following symptoms: cough, dyspnea, sputum production, wheezing, or chest tightness leading to treatment with either antibiotics or systemic glucocorticoids for at least three days. ECOPD severity was graded as mild (outpatient management with or without contact with a healthcare provider), moderate (resulting in emergency department visits), severe (resulting in hospitalization), or very severe (requiring mechanical ventilation). For this analysis, we evaluated ECOPD of any severity and severe ECOPD (which included both severe and very severe events). | PMC10634074 |
Procedures | RIGHT ATRIAL ENLARGEMENT, ENLARGEMENT | Clinical assessments included demographics, CAT and SOBQ questionnaires, 12-lead electrocardiography, and pre- and post- bronchidilator spirometry [Trained technicians collected 12-lead ECGs in a supine position at enrollment. ECGs were analyzed using electronic calipers. A group of trained physicians analyzed all ECGs, and all physicians analyzed a standard subset of ECGs to assess interobserver agreement on the presence or absence of P-pulmonale. Baseline ECGs of 501 participants were analyzed for P-pulmonale, a marker for right atrial enlargement that is evidenced by P wave enlargement > 2.5 mm in lead II [ | PMC10634074 |
|
Statistical analysis | ECOPD, COPD | COPD | Continuous values were summarized using means and standard deviations (SD) and compared between groups using ANOVA tests. Categorical variables were summarized using proportions and compared between groups using chi-square tests. Interobserver agreement was assessed using Cohen’s kappa statistic. Cox proportional hazards models were used to test the association between P-pulmonale and time to first ECOPD (any ECOPD and severe ECOPD) and to test the interaction between P-pulmonale and treatment assignment on time to first ECOPD. Adjusted models included covariates utilized in the parent BLOCK-COPD trial; age, sex, Black race, FEVWe evaluated the relationship between P-pulmonale and treatment assignment and change in COPD symptoms (CAT and SOBQ) using linear mixed-effects models. These models were parameterized with a three-way interaction between P-pulmonale, treatment assignment, and study visit and included a subject-specific random intercept. The mixed-effects models included a subject-specific random effect, which accounts for participants’ baseline characteristics, including sex. If the three-way interaction was not-significant (All | PMC10634074 |
Results | PMC10634074 |
|||
Interobserver agreement | SECONDARY | There was excellent agreement (Kappa ≥ 0.90) in P-pulmonale classification between the primary and secondary ECG observers (Supplementary Table S | PMC10634074 |
|
Discussion | exacerbations, right heart dysfunction, RV dysfunction, exacerbation-prone COPD, ECOPD, COPD, shortness of breath | RHD, COPD | In this post-hoc analysis of the BLOCK-COPD trial, we found that patients with P-pulmonale assigned to treatment with metoprolol were at increased risk of ECOPD of any severity. In addition, metoprolol assignment was associated with worsening in respiratory symptoms in the P-pulmonale group that was both statistically and clinically significant (minimal clinically indicated differed: change in CAT > 2) with no observed change in symptoms in participants without P-pulmonale [Our findings of increased CAT scores and exacerbation risk in those with P-pulmonale assigned to metoprolol highlight the importance identifying concomitant RHD in COPD. Overall, these findings suggest that including P-pulmonale in the risk assessment for metoprolol therapy may be helpful in this select group of COPD patients. While speculative, it is possible that metoprolol use in COPD patients with ECG signs of RHD could lead to a decrease in both inotropy and chronotropy which would reduce heart function andmanifest clinically as increased shortness of breath, a key symptom of ECOPD [Previous investigators have demonstrated that the presence of PH in COPD has functional and prognostic implications and is associated with adverse COPD outcomes, increased risk for exacerbations, and more rapid lung function decline [The parent BLOCK-COPD trial demonstrated that metoprolol does not reduce ECOPD risk in exacerbation-prone COPD patients with no guideline-based indication for beta-blocker therapy. In fact, metoprolol led to an increased risk of hospitalized or mechanically ventilated exacerbations as well as increased respiratory symptoms. The reasons for these findings remain unclear with subsequent post-hoc analyses of BLOCK-COPD revealing no explanatory worsening in lung function with metoprolol treatment [While there has been significant progress in understanding the heart-lung interface, much remains unknown. Structural and mechanical changes in pulmonary vasculature in COPD can increase right ventricular (RV) afterload and, in some cases, subsequently result in RV dysfunction [A better understanding of RV mechanics overall would allow for utilization of multimodal assessments of RV function and morphology and the identification of individuals at heightened risk of COPD morbidity and mortality. This would also allow for targeted investigations of PH therapies in a population enriched for individuals most likely to receive benefit.Our study has several limitations. When compared to right heart catheterization, echocardiography, or cardiac magnetic resonance imaging (cMRI), ECG changes are insensitive to identifying RHD and PH. We focused our analysis on P-pulmonale as it has been shown to be prevalent and have a strong association with PH based on existing literature, but our lack of utilization of more sensitive measures of PH and RHD could lead to some persons with PH being incorrectly analyzed in the ‘no PH’ group which would bias our estimates towards the null. Second, only ECGs obtained at baseline were analyzed which precluded evaluation for longitudinal changes in cardiac conduction, particularly during the an exacerbation; however, this allowed us to standardize for ECGs being obtained only when patients were several weeks exacerbation free and not on study drug. Third, ECG data were analyzed retrospectively, and while observers were well qualified, blinded to patient outcomes, and utilized an electronic caliper system, there may be varying interrater assessment of our ECG indicator of PH, P-pulmonale, which may affect our results. However, in our measurement of interrater reliability, there was strong agreement in P-pulmonale classification between observers. Finally, our study differed from the parent study findings in that metoprolol was associated with an increased risk of severe and very severe exacerbations in BLOCK-COPD; whereas, in our study, we additionally found an association between metoprolol and all exacerbations in individuals with P-pulmonale. We did not find that metoprolol was associated with a greater risk of severe exacerbations in those with p-pulmonale compared to those without p-pulmonale. However, in our analysis a relatively small number of participants with P-pulmonale experienced severe/very severe exacerbations limiting our statistical power to detect associations with this outcome.In conclusion, we found that metoprolol is associated with an increased risk of exacerbation and worsening COPD symptoms in individuals with P-pulmonale. Further investigations are needed using widely available and novel tools targeted at the impact of right heart dysfunction in COPD. | PMC10634074 |
Acknowledgements | None. | PMC10634074 |
||
Authors' contributions | DMM | All authors contributed to the drafting and editing of the manuscript. Conception and design: MTD, MRL, KMK; acquisition of data for the work: TM, RCW, DMM, KP, HV; preparation of figures and tables: SXL, ESH; analysis and interpretation: SXL, ESH, TM, VB, KMK, MRL, MTD; and drafting the manuscript for important intellectual content: RCW, TM, SXL, ESH, MRL KP, VB, KMK, MTD. | PMC10634074 |
|
Funding | Department of Defense W81XWH-15-1-0705. | PMC10634074 |
||
Availability of data and materials | The datasets used and/or analysed during the current study available from the corresponding author on reasonable request. | PMC10634074 |
||
Declarations | PMC10634074 |
|||
Ethics approval and consent to participate | The University of Alabama at Birmingham Institutional Review Board (IRB) approved the study procedures with subsequent review and adoption by each study site. Participants or legal guardians provided informed consent for the parent study, BLOCK-COPD (UAB IRB # 150609007), which included permission for use of data in subsequent analyses. Study procedures involving human participants/data was performed in accordance with the Declaration of Helsinki. | PMC10634074 |
||
Consent for publication | Not applicable. | PMC10634074 |
||
Competing interests | LUNG | ESH has received grant support from the National Institutes of Health and the Department of Defense and clinical trial support from Fisher & Paykel Healthcare; MTD reports grants from the American Lung Association, Department of Defense, and National Institutes of Health, consulting fees from AstraZeneca, GlaxoSmithKline, Novartis, Pulmonx, and Teva. The remaining authors have declared that no conflict of interest exists. | PMC10634074 |
|
References | PMC10634074 |
|||
Background | The implementation of new and complex interventions in mental health settings can be challenging. This paper explores the use of a Theory of Change (ToC) for intervention design and evaluation to increase the likelihood of complex interventions being effective, sustainable, and scalable. Our intervention was developed to enhance the quality of psychological interventions delivered by telephone in primary care mental health services. | PMC10243249 |
||
Methods | A ToC represents how our designed quality improvement intervention targeting changes at service, practitioner, and patient levels was expected to improve engagement in, and the quality of, telephone-delivered psychological therapies. The intervention was evaluated following implementation in a feasibility study within three NHS Talking Therapies services through a qualitative research design incorporating semi-structured interviews and a focus group with key stakeholders (patients, practitioners, and service leads) ( | PMC10243249 |
||
Results | CFIR analysis highlighted a set of challenges encountered during the implementation of our service quality improvement telephone intervention that appeared to have weakened the contribution to the change mechanisms set out by the initial ToC. Findings informed changes to the intervention and refinement of the ToC and are expected to increase the likelihood of successful future implementation in a randomised controlled trial. | PMC10243249 |
||
Conclusions | Four key recommendations that could help to optimise implementation of a complex intervention involving different key stakeholder groups in any setting were identified. These include: 1-developing a good understanding of the intervention and its value among those receiving the intervention; 2-maximising engagement from key stakeholders; 3-ensuring clear planning and communication of implementation goals; and 4-encouraging the use of strategies to monitor implementation progress. | PMC10243249 |
||
Keywords | PMC10243249 |
|||
Background | Evidence shows psychological interventions delivered by telephone can be as effective as those delivered face-to-face [The COVID-19 pandemic forced a sudden shift towards remote therapy delivery, with no preparation time in which to train practitioners in telephone-specific skills [Before the implementation of any new and complex health care intervention, the Medical Research Council (MRC) framework [The Theory of Change (ToC) approach is a pragmatic framework, which describes how and why an initiative/intervention works and can be useful in both intervention planning and in empirical testing [The aim of this article is to explore the use of a Theory of Change (ToC) and an implementation framework of analysis to increase the likelihood of complex interventions being effective, sustainable, and scalable. A ToC was developed to represent how changes of a complex intervention, aimed at enhancing the quality of telephone-delivered psychological therapies, were expected to occur to trigger the intended outcomes. We provide a case example of how the development of a ToC has been informed and amended following evidence from the feasibility study reported here. We also describe potential solutions to overcome implementation challenges and improve likelihood of future implementation success. | PMC10243249 |
||
Method | PMC10243249 |
|||
Description of the EQUITy intervention | The EQUITy intervention was a service quality improvement intervention designed to enhance engagement and the quality of psychological interventions delivered by telephone in primary care in the UK. In brief, the EQUITy intervention targets change at service, practitioner, and patient levels and includes three interlinked components that are equally important: 1-Telephone Recommendations for Services, 2-Telephone Skills Training for Practitioners and, 3-Telephone Information Resources for Patients.The Telephone Recommendations for Services comprised a booklet including guidelines for telephone delivery covering five areas (i.e., Promoting Telephone Work, Key Elements of Telephone Work, Telephone Working Environment and Resources, Boosting Telephone Skills, and Promoting Reflection on Telephone Treatment). Services were asked to decide on a set of goals that could be achieved immediately, and another set that may take longer or need more effort to accomplish. This booklet was circulated to the Service Lead/Team Manager. This component of the EQUITy intervention was designed to ensure availability of the most suitable working environment and resources for the delivery of psychological interventions by telephone, enhance the provision of clinical support for remote delivery and facilitate opportunities for professional development.The Telephone Skills Training for Practitioners included two 3-h online training sessions, combining teaching and interactive activities (e.g., role-play, live demonstration of good practice, group exercises). This component of the EQUITy intervention was designed to develop and/or enhance practitioner telephone skills, address negative preconceptions, and increase engagement and confidence in the delivery of psychological interventions by telephone.The Telephone Information Resources for Patients comprised a leaflet containing key information about telephone interventions delivered by psychological well-being-practitioners, an appointment card which also included tips on how to prepare for telephone sessions, and a poster to be displayed at services. This element of the EQUITy intervention was designed to increase awareness of cognitive behavioural therapy and guided-self-help, and its effectiveness when delivered via telephone; it also aimed to address beliefs and pre-conceptions about remote delivery being inferior in quality to face-to-face delivery.The EQUITy intervention was delivered by a team comprising the two principal investigators (PBo, PBe), the programme manager (JG), a clinical academic with robust experience of delivering training and providing therapy as a Cognitive Behavioural Therapist (KL), a clinical psychologist, two psychological wellbeing practitioners, one clinical academic involved in delivering training for PWPs, one researcher with clinical experience in different evidenced-based therapies including Cognitive Behavioural Therapy (CF) and one patient with experience in supporting training of psychological wellbeing practitioners. | PMC10243249 |
||
Ethical approval | WEST | Ethical approval for the EQUITy feasibility study was granted by the North West—Preston Research Ethics Committee (REF: 20/NW/0082; IRAS ID: 271710cif). All participants completed the consent form, which they returned via email. | PMC10243249 |
|
Study design & recruitment | depression, anxiety | This was a qualitative study using semi-structured interviews and a focus group with stakeholders (patients, practitioners, and service leads). Interviews [This feasibility study was conducted in three UK NHS IAPT services (currently known as Talking Therapies services) delivering guided-self-help for anxiety and/or depression by telephone. NHS IAPT services were approached via email direct invitation to service managers prior to the outbreak of the COVID-19 pandemic, and their commitment remained during the pandemic, when the research study commenced. | PMC10243249 |
|
Stakeholder informants: patients and professionals | depression, anxiety | Patients receiving guided-self-help for anxiety and/or depression by telephone at the services participating in the study were invited to take part in a research interview. Information packs containing an invitation letter, participant information sheet, and a consent to contact form were distributed by the service administration team. Following expressions of interest (return of consent to contact form or direct contact with the research team by email or phone), participants were contacted to confirm willingness to participate, complete the consent form and to schedule a telephone interview. Examples of interview questions for patients included: What did you know about IAPT when you were first referred? What was the information you received at the point of referral and what can you tell me about that information? What did you think about the EQUITy patient resources? How did the EQUITy resources influence (or not) your views about the effectiveness of the treatment you were referred to? How did the EQUITy resources change (or not) your beliefs about how you will be able to successfully engage in your treatment? Thinking about your experience of treatment, how did you find communicating with you practitioner without being able to see her/him? What can you tell me about the therapeutic relationship with your practitioner?Professionals, including practitioners and service leads from each of the three services, were invited to take part in a research interview/focus group. Practitioners delivering guided-self-help by telephone who had attended the EQUITy telephone skills training were invited to take part in the study. All eligible professionals received an invitation letter, a participant information sheet, and a consent to contact form by email, distributed by the service lead/manager or research team. Participants expressing interest to take part were contacted by the research team to confirm willingness to participate and an interview/focus group date was arranged. Examples of interview questions for professionals included: What were your initial thoughts about the EQUITy intervention and taking part in the feasibility study when you heard about it? What aspects of the EQUITy intervention were most helpful for patients, practitioners and for services? What aspects of the EQUITy intervention would you change? How did you feel about delivering telephone treatment following receipt of the EQUITy intervention? What were the challenges you face to implementing the EQUITy intervention? What has been the most significant change in your practice following the reception of the EQUITy intervention? | PMC10243249 |
|
Data collection | depression, anxiety | The study took place in England, UK. The EQUITy intervention was delivered to each of the three NHS IAPT services independently between November 2020 and January 2021. Data collection took place post-intervention between February and April 2021.All participants provided informed consent prior to taking part in the study. All interviews/focus group were audio-recorded and transcribed verbatim by an independent company approved by the University of Manchester. Any identifiable information was removed from the transcripts to protect participants’ anonymity and data were securely stored.Fifteen stakeholder informants across the three NHS services took part in the study (i.e., six patients, five practitioners, and four service leads). All six patient participants had received two or more sessions of guided-self-help by telephone from a practitioner who had received the EQUITy telephone skills training. All patient participants were females, aged between 20 to 57 years old (M=35.33, SD=15.67), five self-described as White and one as Asian, four were employed full-time, one was not employed due to health issues, and one was retired. Three patient participants were receiving treatment for anxiety and depression and the rest for depression only. Three had completed telephone treatment at the time of the interview and the other three were still in treatment.The professional participants included five practitioners with varying degrees of experience from trainees to senior practitioners with supervisory responsibilities, and four service leads. Eight out of the nine professionals were female, all were White British, aged between 24 and 59 years old (M=36.44, SD=11.46). The professionals’ experience within mental health ranged from more than one year up to 20 years.All participants were interviewed and two service leads from one NHS IAPT service participated in a focus group. A focus group was particularly pertinent as they were undergoing a service restructure at the time, enabling a more meaningful and fuller discussion about service implementation . The interviews and the focus group were conducted by telephone and lasted between 40 and 69 minutes. | PMC10243249 |
|
Data analysis | ToC. | Data were analysed incorporating inductive and deductive approaches with deductive approaches informed by the CFIR framework [Firstly, researchers familiarised themselves with the transcripts and the CFIR framework. Secondly, data were coded into the corresponding domains/sub-domains of the CFIR framework and data outside of the framework were coded inductively. CF coded service leads’ data and JC coded data from patients and practitioners. Thirdly, findings were regularly presented and discussed with the wider research team to facilitate further reflection and inform the changes (if needed) to the ToC. The EQUITy intervention was a service quality improvement intervention and targeted change at three different levels (patients, practitioners and service leads); however, the immediate recipients of the intervention were practitioners and services. Consequently, the ToC was used to explore implementation barriers and facilitators, and the patient data aided understanding of the reach of the intervention. Data analysis was supported by QSR International’s NVivo-12 qualitative software [ | PMC10243249 |
|
Results | Table CFIR domains and sub-domains, and key challenges identified during implementation | PMC10243249 |
||
The theory of change for the EQUITy research | Following findings from this feasibility study, the ToC depicted in Fig. EQUITy Programme: Theory of Change updated following findings from this feasibility study and aimed to be evaluated in a randomised controlled trialNote: White boxes in grey lines indicate elements removed from the intervention prior to the commencement of the feasibility study. Light grey boxes highlight elements that are potentially diluted/uncertain and indicate additional uncertainty/risk. Grey dash lines indicate diluted pathways/linkages. White boxes with a dotted pattern indicate challenges on the implementation of components weakening anticipated linkages. The dotted line represents an additional linkage. Thick black arrows indicate pathways to be assessed by mainly quantitative methods; the rest of the pathways will be mainly assessed by qualitative methods (via a process evaluation)Summary of changes to the EQUITy intervention and implementation strategies following findings from this feasibility study and aimed to be implemented and evaluated in a randomised controlled trial | PMC10243249 |
||
Discussion | The assumption that face-to-face systems and processes may simply transfer to a different medium such as the telephone, has hampered service development and delivery in the past [ | PMC10243249 |
||
1-Developing a good understanding of the intervention and its value | Many factors can impede uptake of service quality improvement interventions for telephone working including competing demands on frontline providers, lack of knowledge, skills and resources, and misalignment with service priorities. Therefore, more emphasis should be placed on developing a good understanding of the intervention, identifying its value, exploring resources available, and pinpointing the best way to integrate the intervention within existing procedures to prevent further burden on services. In addition, it is important to overcome any negative beliefs or resistance from staff to prevent implementation failure. | PMC10243249 |
||
2-Increasing engagement from key stakeholders | Findings from our study indicate that a service quality improvement telephone intervention comprised of different components targeting changes at different levels (service, professionals, and patients) requires engagement, commitment, and accountability of different key stakeholders, including service leads, team managers, practitioners, administrative personal, and other staff. Engagement from appropriate individuals could be improved by providing support from leads and using organisational incentives such as goal sharing awards or extrinsic incentives could prove beneficial to increase receptivity and increase engagement of individuals to use the intervention. It is anticipated that high levels of commitment and involvement of leads and managers with the implementation of each of the elements of a complex intervention could enhance the likelihood of a successful implementation. Due to the high turnover of staff in mental health services, it is important to develop strategies to immerse new staff in the awareness and implementation of the intervention, which could ensure sustainability of the intervention over time. | PMC10243249 |
||
3-Clear planning and communication of implementation goals | The lack of time allocated to planning behaviours and implementation tasks, alongside the lack of identification of service quality improvement telephone goals and how to incorporate the intervention in day-to-day work tasks may explain lack of implementation success. Clear planning and communication around implementation goals and what is needed, from whom, and how those would be achieved, are essential to improve the likelihood of a successful outcome. Nominating an implementation lead and identifying champions [ | PMC10243249 |
||
4-Formal strategies to monitor implementation progress | Developing models to guide and identify implementation strategies such as audit feedback [ | PMC10243249 |
||
Further reflections on the ToC | The ToC originally developed was in line with a pre-COVID context where the use of telephone to deliver assessment and treatment sessions in NHS Talking Therapies services varied, but many low intensity practitioners (Step 2 care) were delivering approximately 60% of their appointments by telephone [Findings from the feasibility study presented in this paper have provided information on individual and group influences on behaviour and the ability that social context, as much as clinical context, has influenced the ToC success of our service quality improvement intervention. Findings from this feasibility study additionally resulted in changes to the content of our EQUITy intervention and its implementation. It is anticipated that changes will increase the likelihood of implementation success for its further evaluation in a randomised controlled trial. | PMC10243249 |
||
Conclusions | Concerted attention to evidence-based implementation strategies is needed to reduce the likelihood of implementation failures and optimise success. Implementation strategies should emphasise the value and benefits related to implement an intervention and facilitate awareness and good understanding of what is needed and from whom to implement the intervention. In addition, ensuring engagement and monitoring of progress from key stakeholders is also paramount to improve the likelihood of success and sustainability over time. Understanding implementation challenges for service quality improvement interventions could prove beneficial to increase chances of implementation success. | PMC10243249 |
||
Acknowledgements | The authors would like to thank the service leads, practitioners, and patients that participated in the study and their organisations. The authors would like to specially thank Elinor Hopkin for proofreading the final manuscript. | PMC10243249 |
||
Authors’ contributions | CF | CF, JG, HB, and PBe were involved in the conceptualization of study research goals and aims. PBe and PBo are the Principal Investigators for this research programme and were responsible of the funding acquisition. CA was a grant co-applicant. CF led the formal data analysis, and all authors were involved in regular discussions providing reflective feedback to redefine, revise, and finalise themes within CFRI sub-domains. JG was responsible for the project administration and supervision of the research activity. CF led the writing up of the manuscript, i.e., contributing to the development of the original and subsequent drafts of the manuscript. All authors contributed to reviewing and editing the manuscript. All authors have read and approved the final manuscript. | PMC10243249 |
|
Funding | This study is funded by the National Institute for Health Research (NIHR) Programme Grants for Applied Research (project reference: RP-PG-1016–20010). Armitage is supported by NIHR Manchester Biomedical Research Centre and NIHR Greater Manchester Patient Safety Translational Research Centre. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care. The funders had no role in study design, data collection, analysis or interpretation, decision to publish, or preparation of the manuscript. | PMC10243249 |
||
Availability of data and materials | The dataset generated and analysed during the current study are not publicly available due to privacy and ethical restrictions (i.e. potential for breach of anonymity); but are available from the corresponding author upon reasonable request. | PMC10243249 |
||
Declarations | PMC10243249 |
|||
Ethics approval and consent to participate | WEST | The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Ethical approval was granted by the North West—Preston Research Ethics Committee (REF: 20/NW/0082; IRAS ID: 271710cif). All participants provided written informed consent before participation and consented for the use of anonymised direct quotes. | PMC10243249 |
|
Consent for publication | Not applicable. | PMC10243249 |
||
Competing interests | The authors declare that they have no competing interests. | PMC10243249 |
||
References | PMC10243249 |
|||
Subject terms | death | REGRESSION | Hospital readmission prediction models often perform poorly, but most only use information collected until the time of hospital discharge. In this clinical trial, we randomly assigned 500 patients discharged from hospital to home to use either a smartphone or wearable device to collect and transmit remote patient monitoring (RPM) data on activity patterns after hospital discharge. Analyses were conducted at the patient-day level using discrete-time survival analysis. Each arm was split into training and testing folds. The training set used fivefold cross-validation and then final model results are from predictions on the test set. A standard model comprised data collected up to the time of discharge including demographics, comorbidities, hospital length of stay, and vitals prior to discharge. An enhanced model consisted of the standard model plus RPM data. Traditional parametric regression models (logit and lasso) were compared to nonparametric machine learning approaches (random forest, gradient boosting, and ensemble). The main outcome was hospital readmission or death within 30 days of discharge. Prediction of 30-day hospital readmission significantly improved when including remotely-monitored patient data on activity patterns after hospital discharge and using nonparametric machine learning approaches. Wearables slightly outperformed smartphones but both had good prediction of 30-day hospital-readmission. | PMC10203290 |
Introduction | Nearly 1 in 5 patients discharged from the hospital are readmitted within 30 daysThe use of remote patient monitoring (RPM) devices such as smartphones and wearables is increasing, and prior work has demonstrated they are accurate for tracking daily activity patternsThe objective of our study was to evaluate whether hospital readmission prediction models could be improved by incorporating RPM data on activity patterns after hospital discharge and by using machine learning approaches. To evaluate whether there were differences in prediction based on the type of RPM device, we randomly assigned patients discharged from medicine services at two hospitals in Philadelphia to use either their smartphone or a wearable device to track daily activity patterns. Many patients already have a smartphone and these devices have been demonstrated to lead to longer-term utilization for RPM than wearables | PMC10203290 |
||
Standard model with traditional regression | REGRESSION | These models used electronic health record data until the time of hospital discharge. As expected, since these models did not include RPM data, prediction was similar between device arms for logistic regression (AUC in smartphone arm, 0.63, 95% CI 0.61 to 0.65; AUC in wearable arm, 0.63, 95% CI 0.60 to 0.65; | PMC10203290 |
|
Standard model with machine learning regression | REGRESSION | Prediction increased when using nonparametric machine learning approaches than when using traditional parametric regression techniques (Table | PMC10203290 |
|
Enhanced prediction models with RPM data | REGRESSION | For all 5 regression approaches, prediction increased significantly when RPM data on physical activity and sleep patterns was included in the models (Figure Observed readmission rate by predicted risk quartile for the enhanced model with RPM data and ensemble machine learning. Depicted is the observed rate of hospital readmission at the patient-day level by predicted risk quartile (1 is lowest risk; 4 is highest risk) and by study arm. Data presented use the enhanced model that includes remotely-monitored data and the ensemble machine learning model. | PMC10203290 |
|
Wearable versus smartphone | In the enhanced models that incorporated RPM data, 4 of the 5 models found that the wearable arm had significantly better prediction than the smartphone arm (Table | PMC10203290 |
||
Discussion | REGRESSION | In this clinical trial, we found that hospital readmission prediction models that used traditional parametric regression approaches with electronic health record data until the time of hospital discharge performed similar to previously published models with AUCs ranging from 0.60 to 0.70Our findings reveal several important insights for future implementation and research efforts. Prediction in the wearable arm was significantly greater than the smartphone arm, but the difference was small (AUC 0.85 for wearable vs. 0.84 for smartphone). Prior work has found that physical activity data collected by these devices is similar in accuracyA recent systematic review indicated that machine learning methods showed promise for improving hospital readmission models but that more rigorous evaluation was neededThis study has several strengths. First, we evaluated an important clinical outcome and used several data sources including electronic health records and state databases to identify hospital readmissions. Second, we enrolled a diverse patient sample that of which 47% were 50 years or older, 53% were Black or Hispanic, and 26% had Medicaid. Third, we evaluated 5 different modeling techniques including traditional parametric and nonparametric machine learning approaches and found consistent results across these models. Fourth, we collected remote patient monitoring data on patient behavior after hospital discharge, which has not traditionally been integrated into hospital readmission prediction models. Fifth, we conducted a pragmatic, randomized trial to provide a rigorous assessment of the difference in prediction between two of the most common devices used for remote patient monitoring. | PMC10203290 |
|
Limitations | This study also has limitations. First, the sample included patients with a mean age of about 46 years from medicine services at one health system who were being discharged to home and less than 10% of those approached agreed to participate, which limits generalizability. Future studies should evaluate a broader sample of patients. Second, the analyses evaluated hospital readmissions within 30 days. Future studies should evaluate prediction of rehospitalization at longer periods of time after discharge. Third, data on hospital readmissions was obtained from Penn Medicine and a database on hospitalizations in the State of Pennsylvania. We did not have access to readmissions that occurred outside of Pennsylvania. Fourth, while participant characteristics between study arms were similar, the readmission rate was lower in the wearable arm. This imbalance may have been due to random chance and the small sample size. However, it does not impact the main findings that RPM data with machine learning improved prediction. Fifth, RPM data was limited to measures of physical activity and sleep. We did not have data on heart rate or other biometrics which can be captured by some wearable devices. Nonetheless, we did find improvements in prediction with these RPM measures and future studies could evaluate if additional biometric data can further improvement. | PMC10203290 |
||
Conclusions | Prediction of 30-day hospital readmission significantly improved when including remotely-monitored patient data on activity patterns after hospital discharge and using nonparametric machine learning approaches. Wearables slightly outperformed smartphones but both had good prediction. Since many patients use smartphones and wearables, data from remote patient monitoring devices could be incorporated more broadly into prediction models to identify the patients at highest risk of hospital readmission. | PMC10203290 |
||
Methods | PMC10203290 |
|||
Study design | PREDICT (Prediction using a Randomized Evaluation of Data collection Integrated through Connected Technologies) was a 2-arm randomized clinical trial conducted remotely after patients were discharged from inpatient medicine services at two Penn Medicine hospitals in Philadelphia to their home (ClinicalTrials.gov number, NCT02983812). The trial was conducted was conducted from January 23, 2017 to December 7, 2019. Patients were randomly assigned to use either a smartphone application alone or with a wearable device to collect data on activity patterns for 6 months. Trends in device utilization during this time period have been previously publishedThe trial interventions were conducted using Way to Health, a research technology platform at the University of Pennsylvania used previously for remote-monitoring of activity patterns | PMC10203290 |
||
Participants | Patients admitted to medicine services at two hospitals at Penn Medicine in Philadelphia (Hospital of the University of Pennsylvania and Penn Presbyterian Medical Center) were identified as potential participants using the electronic health record (EPIC) and approached in the hospital by the study team.Patients were eligible for the trial if they were 18 years or older, had a smartphone compatible with the Withings HealthMate smartphone application, had no current medical condition which prohibited them from ambulating or plan for a medical procedure over the next 6 months that would prohibit them from ambulating, planned to be discharged to home, able to speak and read English, and able to provide informed consent. Patients were excluded if pregnant, already participating in another physical activity study or if they did not reside in the States of Pennsylvania or New Jersey.Interested patients used a computer from the study team to create an account on the Way to Health technology platform, provide informed consent, and selected whether to receive study communications by text message, email, interactive voice recording, or a combination. Patients then completed series of survey assessments including baseline information and validated questionnaires. | PMC10203290 |
||
Data | Comorbidity | Patient data obtained from Way to Health included demographics collected by survey during initial hospitalization (age, gender, race/ethnicity, education, marital status, annual household income, and body mass index) and RPM activity patterns transmitted from the smartphone application and wearable including daily measures of step counts, distance, calories burned (active and total), minutes of activity (soft and moderate intensity), minutes of sleep (total, light, and deep), minutes to fall asleep, minutes awake, and number of times awakened. The smartphone application alone did not track sleep but patients in that arm could manually input sleep on their own. Data obtained from the electronic health record included insurance type, the Charlson Comorbidity Index | PMC10203290 |
|
Randomization | coronary heart disease, pneumonia, diabetes | ACUTE MYOCARDIAL INFARCTION, PNEUMONIA, CHRONIC OBSTRUCTIVE PULMONARY DISEASE, CONGESTIVE HEART FAILURE, CORONARY HEART DISEASE, DIABETES | Patients were randomized electronically using block sizes of two and stratified by one of the following six primary conditions for admission: acute myocardial infarction or coronary heart disease, chronic obstructive pulmonary disease, congestive heart failure, diabetes, pneumonia, or any other condition. These conditions were selected to represent the leading primary diagnoses associated with hospital readmissions in Pennsylvania for which we hypothesized activity patterns could inform readmission rates. All investigators, statisticians, and data analysts were blinded to arm assignments until the study and analysis were completed. | PMC10203290 |
Interventions | Patients assigned to the smartphone arm were set up in the hospital with the Withings Health Mate smartphone application which used accelerometers in the smartphone to track physical activity patterns. They were asked to open the application at least once a day to sync the device. These patients were told they would receive a wearable device after their 6-month period completed. Patients assigned to the wearable device arm were also setup in the hospital with the Withings Steel which had a battery that lasted up to 8 months without recharging. In addition to tracking physical activity patterns, the wearable also tracked sleep patterns. Patients were asked to wear the device as much as possible including while sleeping and to sync it with the Withings Health Mate Smartphone application at least once a day.In both arms throughout the 6-month period, patients were sent a reminder to sync their device if data had not been transmitted for four consecutive days. All patients received $50 to enroll and $50 to complete the trial. | PMC10203290 |
||
Outcome measures | death | The primary outcome was prediction of readmission to the hospital or death within 30 days of discharge. | PMC10203290 |
|
Statistical analysis | death, Comorbidity | REGRESSION | All randomly assigned patients were included in the intention-to-treat analysis. For each patient and on each day of the study (patient-day level), we obtained RPM activity data. Data could be missing for any day if the patient did not use the device, did not sync it to upload data, or the device did not capture it (e.g. smartphones did not track sleep data). We also excluded outliers (top and bottom 1 percentile) in activity data and coded them as missing. To account for missing values in the RPM data, we included a dichotomous missing-value indicator variable for each RPM data variableWe used a discrete-time survival analysis and evaluated hospital readmission or death within 30 days of discharge at the patient-day level as the unit of observation. The sample was split into training and testing folds for each arm that were mutually exclusive at the patient level in a 3:2 ratio. The training set used fivefold cross-validation and then final model results are from predictions on the test set.We fit a standard model that comprised data available up to the time of hospital discharge including demographic information (age, gender, race/ethnicity, education, marital status, annual household income, and body mass index), time (calendar month and year), hospital length of stay, vitals near discharge (temperature, heart rate, systolic and diastolic blood pressures, oxygen saturation, respirations per minute), and Charlson Comorbidity Index. We fit an enhanced model with RPM data that included the same variables as the standard model as well as the following: daily measures of step counts, distance, calories burned (active and total), minutes of activity (soft and moderate intensity), minutes of sleep (total, light, and deep), minutes awake, and number of times awakened. The enhanced model with RPM also including missing-value and lagged indicators for each RPM measure, as previously described. For all models, we compared traditional parametric regression (logit and lasso) to nonparametric machine learning approaches (random forest, gradient boosting, and ensemble machine learning). The ensemble machine learning model used a combination of logit, lasso, random forest, and gradient boosting. Hyperparameter specification is described in Supplementary Table Model performance was assessed using the area under the receiver operator characteristic curve (AUC) | PMC10203290 |
Supplementary Information | The online version contains supplementary material available at 10.1038/s41598-023-35201-9. | PMC10203290 |
||
Author contributions | M.P. drafted the manuscript and supervised the study. K.V. and D.P. obtained funding for the study. D.S. and G.K. provided oversight for the analysis. S.P. conducted the analysis. C.E. managed participant enrollment and study operations. All authors reviewed the manuscript. | PMC10203290 |
||
Funding | This project is funded, in part, under a grant with the Pennsylvania Department of Health through the Commonwealth Universal Research Enhancement (CURE) Program. The Department specifically disclaims responsibility for any analyses, interpretations or conclusions. This trial was also supported by the University of Pennsylvania Health System through the Penn Medicine Nudge Unit. | PMC10203290 |
||
Data availability | The datasets generated and/or analysed during the current study are not publicly available because we do not have IRB approval to share them since they contain patient information. Dr. Patel had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. | PMC10203290 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.