Ye Shen

Background: Older adults with dementia face markedly elevated mortality from COVID-19. Both donepezil and metformin have anti-inflammatory effects and are associated with improved outcomes in respiratory viral infections, but whether their combined use is associated with lower COVID-19 mortality remains unknown. Methods: We conducted a retrospective cohort study using the National Clinical Cohort Collaborative (N3C) data enclave, including adults aged ≥65 years with primary SARS-CoV-2 infection. We evaluated the association between donepezil monotherapy (5 mg or 10 mg) and COVID-19 mortality using 1-to-1 propensity score-matched analyses. Analyses were stratified by diabetes and metformin use. Results: Among patients with dementia, 30-day mortality was lower in patients receiving donepezil 5 mg compared with matched untreated controls (10.3% vs 13.2%; P = .00046) and similarly for 10 mg (10.3% vs 13.4%; P = .0014). Among patients with dementia and diabetes, donepezil–metformin use was associated with the lowest mortality, approaching levels in matched individuals without dementia or diabetes. Metformin or donepezil alone was associated with modest reductions. Among donepezil users, those with diabetes treated with metformin had lower mortality than those without diabetes (30-day: 7.6% vs 10.2%, P = .016; significant only after FDR correction). No significant mortality differences by diabetes or metformin use were observed among dementia patients not receiving donepezil. Conclusions: Donepezil use was associated with reduced COVID-19 mortality in older adults with dementia, with findings consistent with a potential synergistic association with metformin. These findings warrant further study of cholinergic modulation, alone or with metabolic therapies, to reduce severe COVID-19 outcomes.

Background: Noroviruses are a leading cause of acute gastroenteritis worldwide, yet no licensed vaccines currently exist. Antibody-mediated immunity provides protection, but the relationship between norovirus exposure dose and antibody kinetics remains unclear. Methods: We analyzed longitudinal data on histo-blood group antigen–blocking, IgA, and IgG antibodies from a human norovirus infection trial and a candidate vaccine trial. We developed hierarchical nonlinear Bayesian models to characterize antibody kinetics following different amounts of virus or vaccine antigen exposure within infection and vaccination settings. We fit models with exponential and power-law decay terms and compared their performance. Results: Across studies and antibody types, the amount of antigen individuals were exposed to influenced early antibody kinetics, including growth, peak, and decay, but effects differed between infection and vaccination. Doses that produced larger initial antibody boosts also decayed faster, resulting in little association with 1-year antibody persistence. Conclusions: Our findings suggest that the amount and type of antigen exposure shape early antibody kinetics but not lasting durability. This reflects a trade-off between dose-induced initial immune boost and subsequent waning.

While clinical trials are often regarded as the gold standard by medical researchers and regulatory agencies, the statistical properties of typical randomized controlled trials heavily depend on large sample sizes, which may not always be achievable due to budget constraints or other practical limitations. To address this issue, adaptive designs have gained increasing attention since the 1970s. These designs incorporate modifications in trial procedures or statistical methodologies, aiming to fulfill study objectives more efficiently using smaller sample sizes and shorter durations compared to traditional clinical trials. Specifically, covariate-adaptive designs are developed to reduce the risk of significant covariate imbalances among treatment groups, particularly in studies with limited sample sizes. Additionally, patient dropout during trials significantly impacts statistical inference regarding treatment effects and diminishes statistical power. Ignoring dropout events can result in undesirable participant allocation and substantial treatment imbalances. In this article, we propose novel strategies that explicitly incorporate dropout information into covariate-adaptive randomization. We thoroughly investigate the statistical properties of covariate-adaptive designs adjusted for dropouts. Extensive simulation studies highlight the advantages of the proposed approaches. Our research lays a robust foundation for future developments in covariate-adaptive randomization, specifically addressing dropout challenges.

Long COVID (LC) affects millions of individuals worldwide, particularly those with preexisting comorbidities. However, when assessing comorbidity-related risk factors for LC, it remains unclear whether preexisting comorbidities should be defined prior to SARS-CoV-2 infection or LC diagnosis. This methodological choice may substantially bias estimates of LC risk. In this study, leveraging Electronic Health Records (EHR) from 6,130,413 adults with documented COVID-19 across 49 U.S. states in the National Clinical Cohort Collaborative (N3C) from 2020 to 2024, we evaluated the impact of different comorbidity exposure-window definitions on LC risk estimates. We developed ensemble cross-fitted double/debiased machine learning to adjust for individual- and county- level confounders. We found that defining comorbidities before infection rather than before LC diagnosis yielded 8%-314% higher adjusted attributable risks and 3%-34% higher adjusted relative risks across 16 major comorbidities. Comorbidity-associated risks generally declined from 2020 to 2024, while substantial demographic, socioeconomic, geographic, and multimorbidity-related disparities persisted. Exposure-window specification can be a major source of bias in estimates of LC risk, particularly for comorbidities prone to delayed clinical recognition or underascertainment. More broadly, our results highlight how temporal misclassification of exposures in longitudinal EHR studies can distort risk attribution and population-level inference for risk stratification for LC and prioritization of prevention strategies in clinical medicine and public health, respectively.

Background: The 2019 medication regimen complexity‐intensive care unit (MRC‐ICU) score is associated with patient outcomes, ICU complications, and critical care pharmacist workload. This score was developed using heuristic component selection and validated in a single‐center cohort of 130 ICU patients. We sought to apply data‐driven reweighting methodology in a large, multicenter cohort of ICU adults to improve the predictive capabilities of MRC‐ICU. Methods: This was a retrospective, observational cohort study of adults admitted to an ICU between 2015 and 2023 at two academic health systems. Machine learning–based methods, including Principal Component Analysis and Random Forest, were used to create an updated MRC‐ICU score optimized to predict three outcomes: hospital mortality, ICU fluid overload (FO) occurrence, and invasive mechanical ventilation (IMV) use. MRC‐ICU 2.1 used average mortality, FO, and IMV use; MRC‐ICU 2.2 used average mortality and FO and adjusted for prolonged IMV use. Data from one center were used for training and testing, and data from the other for validation. The predictive abilities of MRC‐ICU 2.1 and 2.2 for each outcome were compared to MRC‐ICU 1.0 and to severity of illness scores (i.e., Acute Physiology and Chronic Health Evaluation [APACHE] II and Sequential Organ Failure Assessment [SOFA]). Results: A total of 19,117 patients across training, testing, and validation datasets were included. MRC‐ICU 2.0 scores outperformed MRC‐ICU 1.0 for predicting most outcomes, with improvements in Area Under the Receiver Operating Characteristic (AUROC) ranging from +0.03 to +0.08 across datasets. MRC‐ICU 2.1 and 2.2 did not consistently outperform APACHE II and SOFA in predicting mortality. The addition of MRC‐ICU 2.0 scores to models including APACHE II or SOFA resulted in statistically significant improvements in discrimination in several settings (DeLong p < 0.05), with AUROC increases generally ranging from approximately +0.01 to +0.13 depending on outcome and dataset. Conclusions: The updated MRC‐ICU 2.0 score (MRC‐ICU 2.1 and 2.2) demonstrated consistently improved discrimination compared with the original MRC‐ICU 1.0 across outcomes and datasets. The performance of MRC‐ICU 2.0 (MRC‐ICU 2.1 and 2.2) was generally comparable to established severity‐of‐illness scores (SOFA and APACHE II), although it did not consistently outperform these measures. When incorporated into combined models, MRC‐ICU 2.0 provided additional predictive value, indicating that it captures information complementary to traditional severity‐of‐illness scores. Overall, these findings suggest that MRC‐ICU 2.0 represents an improved and clinically interpretable measure of medication regimen complexity that is useful as a complementary predictor.

Natural disasters fueled by climate change will exacerbate existing health disparities and marginalize affected communities. This process manifests as the extreme weather–climate gap, underscoring the critical importance of preventive and resilience-focused interventions in these communities. This study examines the effectiveness of a pre-post implementation study with a voluntary in-person disaster preparedness exercise on a sample of 106 participants. This exercise involved educating participants on disaster preparedness and a participatory mapping exercise to better understand perceptions about disaster risk. Survey constructs include social resources, social capital, perceived susceptibility, perceived stress, and self-efficacy. Adoption of the target disaster preparedness behavior was evaluated quantitatively using logistic regression. The probability of adopting the disaster preparedness behavior was higher among participants who attended the disaster exercise for three of the four preparedness outcomes and was highest for knowing where to find resources (29 percentage points). Social capital was associated with a decreased probability of having a disaster plan, suggesting it may have an unexpected impact on disaster preparedness by fostering an inflated sense of support. Perceived susceptibility was associated with an increased probability of having an emergency bag and knowing where to find resources following the exercise. The short implementation that included an active learning exercise was associated with changing risk perceptions toward flood disasters, having an emergency bag in the event of a disaster, and knowing where to find critical resources. Barriers to achieving in-person participation including work or family limitations should be addressed for larger-scale implementations.

Background: Long COVID remains a disease of significant public health importance globally, posing unprecedented health and socioeconomic challenges from daily activity limitation and reduced work productivity. While several modifiable risk factors (e.g., hypertension, non-vaccination, smoking) have been associated with Long COVID, their attributable burden remains unknown to inform policy priorities. This study assesses the temporal trends and disparities in the population attributable fractions (PAFs) of major preexisting comorbidity risk factors (PCRFs) for Long COVID during 2021-2024. Methods: We used data from the US National Clinical Cohort Collaborative (N3C) electronic health record with measurements of seven PCRFs for Long COVID (hypertension, cardiovascular disease, diabetes, chronic lung disease, chronic kidney disease, chronic liver disease, and depression), integrated with community- and health system-level social determinants of health data on education and poverty, and healthcare access and quality index, respectively. We included 4,854,205 adults (≥18 years) with polymerase chain reaction test, point-of-care antigen test, and clinical diagnosis (PPC) indicated COVID-19 infections. We employed various analytic methods, including statistical weighting, age adjustment, propensity score matching, and regression modelling to estimate prevalence, communalities, and relative risks for PCRFs, as appropriate. We then used these estimates to calculate individual weighted PAFs (IW-PAFs) for each PCRF and overall weighted PAFs (OW-PAFs) for all PCRFs. We examined temporal trends in PAFs with average percent change, and assessed disparities by age, sex, geographic location, and BMI. Subanalyses were conducted to identify subnational differences in OW-PAF cases based on the expected PPC and antibody seroprevalence test indicated COVID-19 infections for 2021-2024 and 2024 and 2030, respectively. Findings: Among 3,777,439 participants analysed, the overall proportion of Long COVID cases attributable to the seven PCRFs (OW-PAF) was 22.01%, 95% CI [21.89–22.13], with a statistically significant increase from 18.67% in 2021 to 29.34% in 2024. During 2021-2024, the PCRFs with the most contributions were diabetes (IW-PAF 8.41%, 95% CI [8.26–8.56]), chronic lung disease (IW-PAF 6.23%, 95% CI [6.12–6.34]), and depression (IW-PAF 3.95%, 95% CI [3.88–4.01]), followed by hypertension, cardiovascular disease, and other PCRFs. During 2021-2024, all seven PCRFs showed upward trends. These trends were statistically significant for six of the seven PCRFs studied, while the trend for cardiovascular disease was not statistically significant. Young- and middle-aged, female, obesity groups had higher IW-PAFs. Disparity was driven by diabetes, chronic lung disease, depression, hypertension, chronic kidney disease, and chronic liver disease in the young- and middle-aged, by chronic lung disease and depression in females, and by all PCRFs in the obesity group. For the analysed timepoints, the Southeast region had the highest OW-PAF cases. This was followed by Midwest and Northeast in 2021-2024 and 2024, and Midwest and West by 2030. Interpretation: The substantial, sustained increase in PCRFs PAFs highlights the urgent need for Long COVID prevention strategies, targeting leading PCRFs, high-risk subpopulations, and high-burdened geographies with tailored policy interventions.

Despite the positive role of social relationships in behavior change, dyadic interventions for smoking have not been consistently successful. This owes, in part, to the fact that dual-smoking cohabiting couples share similar routines, environments, and relational dynamics that can undermine quitting efforts. We adopted an exploratory and descriptive approach to identify distinct profiles of smoking abstinence within dyads, how relationship dynamics relate to these patterns, and whether these patterns predict smoking status at follow-up. We utilized preexisting data from a pilot RCT examining the effects of partner-involved financial incentives on dyadic abstinence. Both members of 95 dual-smoking couples (52% female, 73% White, M_age = 42.39, SD = 10.57) recruited between 2021 and 2022 provided baseline information about their relational and motivational dynamics, followed by 10 weeks of daily reports of smoking behavior, and biochemically verified quit status at follow-up. Latent growth mixture models suggested four patterns of dyadic abstinence: Concordant abstainers, Discordant abstainers, Discordant nonabstainers, and Concordant nonabstainers. Treatment arm and motivational and relational dynamics predicted the probability of following specific dyadic patterns of behavior change, and dyadic patterns of change predicted individual and couple quitting status at follow-up. Our findings underscore the importance of considering that treatment will facilitate coordinated dyadic behavior change for some couples but not all and emphasize the need to better understand when and how mechanisms support long-term abstinence. More work is needed to investigate whether these patterns generalize across samples with more diverse sociodemographic and health characteristics. (PsycInfo Database Record (c) 2026 APA, all rights reserved).

Background: Several genetic variants have been identified to modify the effects of fish oil supplementation (FOS) on increasing circulating omega-3 fatty acids, but it remains unexplored whether polygenic predisposition to low circulating omega-3 fatty acids modifies these effects. Objective: To test if polygenic scores (PGS) for circulating omega-3 fatty acids modify the associations of FOS with corresponding circulating concentrations. Methods: We developed PGS models for absolute circulating concentrations of total omega-3 fatty acids (Omega-3), docosahexaenoic acid (DHA), and their relative percentages in total fatty acids (Omega-3% and DHA%), using a multi-ethnic genome-wide association study (N=136,016). PGS models were validated in 437,803 UK Biobank participants of European (EUR), Central/South Asian (CSA), African, and East Asian genetic ancestries. Linear models tested PGS-by-FOS interactions on corresponding observed circulating concentrations. Discovery analysis was performed separately in 237,380 EUR participants and each non-EUR group. Replication analyses were performed using oily fish intake and in another 178,935 EUR participants. Results: In EUR participants, PGS explained 5.3-11.1% of the phenotypic variance, and significant PGS-by-FOS interactions were detected across all four circulating omega-3 traits. Among participants in the bottom 5% of the PGS distribution, FOS was significantly associated with a 0.40 SD (95% CI: 0.39–0.44) increase in Omega-3. This association effect was 11.1% larger than the population average (β = 0.36; 95% CI: 0.35–0.37; P _Int = 0.016) and 42.8% larger than that in participants in the top 5% of the PGS distribution (β = 0.28 SD; 95% CI: 0.25–0.32; P _Int = 4.03X10 ^-10 ). These interaction patterns were consistently observed in CSA ancestry and confirmed in replication and sensitivity analyses. Conclusions: PGS modify the associations of FOS with circulating omega-3 fatty acids in EUR and CSA populations, with larger FOS effects in participants with lower PGS. These findings support the development of genome-informed precision nutrition.