BackgroundDespite substantial evidence supporting the benefits of physical activity (PA) for older adults, participation rates remain low due to physical, psychological, social, and institutional barriers. Recruiting older adults for PA studies thus presents significant challenges, particularly for socially vulnerable populations and those facing digital literacy limitations in technology-assisted PA interventions. The present study aimed to articulate and establish a consensus among multidisciplinary and multinational PA experts regarding key considerations and requirements for designing and implementing recruitment of older adults for PA studies.MethodsThis study employed a mixed-methods consensus approach integrating a structured formal consensus process with a two-round Delphi survey. Through face-to-face and online discussions, an expert panel group formulated 104 initial recommendations regarding the recruitment of older adults for PA studies. In the first Delphi round, 42 external experts assessed the relevance of each recommendation, and recommendations with a ≥ 70% agreement threshold were considered consensual. The second round included 60 refined recommendations that were evaluated by 31 experts. A final consensus was reached through iterative expert panel discussions.ResultsOf the 104 initial recommendations, 52 (50%) met the inclusion threshold during the first round. In the second round of Delphi, 36 of the 60 revised recommendations (60%) achieved consensus, with agreement ranging from 71.0 to 96.8%. The final consensus process established 34 key recommendations categorized into four domains: (1) ethical principles (71.7–93.5%), (2) informed consent (71.0–96.8%), (3) stakeholder engagement (61.3–83.9%), and (4) recruitment strategies for technology-assisted PA interventions (74.2–87.1%). However, there remains a lack of consensus on the extent to which sociocultural diversity and differences in sample representation should be addressed.ConclusionsThis study provides expert-driven, consensus-based recommendations to improve the recruitment of older adults for PA studies. Adherence to ethical considerations, informed consent procedures, stakeholder collaboration, and tailored strategies for technology-assisted interventions were highlighted. These findings offer practical guidance for researchers and practitioners in planning and obtaining adequate participation in PA studies involving older adults.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12874-025-02659-2.

Pre-Post intervention factorial design with multiple observations for each subject before and after the intervention frequently arises in Quality-of-Life (QoL) research. Usually not only a global hypothesis on intervention is of interest, but also the hypotheses of pre-post change and also the interaction between intervention and pre-post change. In most practical situations, the distribution of the observed data is unknown and there may exist a number of atypical measurements and outliers. More importantly, QoL outcomes are measured in rating scales and the multiple Quality-of-Life measurements before and after the intervention present complex within-subject correlations. Hence, use of parametric and semi-parametric procedures that impose restrictive distributional and correlation assumptions becomes questionable. This emphasizes the demand for statistical procedures that enable us to accurately and reliably analyze QoL outcomes with minimal conditions. Nonparametric methods offer such a possibility and thus become of particular practical importance in the field of QoL. In this article, we aim to expose researchers and practitioners in the biomedical and behavioral science to nonparametric methods for the analysis of data collected in clustered randomized design. We also illustrate the use of the R package npclust we developed for an easy and user-friendly access to nonparametric methods for the analysis of Quality-of-Life data. It provides procedures for pre-processing the data, performing various hypothesis tests and computing confidence intervals for the estimated effects. The procedures also contain methods for visual display of the results. We illustrate the implemented procedures with Pediatric Asthma Quality-of-Life data collected from the Asthma Randomized Trial of Indoor wood Smoke (ARTIS).

Structured datasets, such as time-series or survey-based tables, often lack intuitive visualizations that reveal rankings, interrelationships, or leadership dominance among subjects. Traditional parametric statistics fail to capture such relational patterns, especially in ordinal or categorical data. This study proposes a novel nonparametric framework to visualize leadership styles through network diagrams. We introduced a three-tier "basket model"-cells (small baskets), columns (medium baskets), and the full table (large basket)-to transform rectangular data into weighted co-word matrices. Using publicly available 2023 HCAHPS survey data from 52 U.S. states and territories, we applied a follower-leader clustering algorithm (FLCA) implemented in R. Leadership was classified into three types: absolute, relative, and no advantage. Network visualizations were generated using Sankey-style diagrams to highlight dominance and inter-cluster relationships. The weighted approach successfully identified Nebraska as the top leader in the upper 20 states and District of Columbia as a cluster leader among the bottom 20 after data inversion. The network diagrams effectively differentiated between absolute dominance (single strong cluster), relative dominance (sub-cluster formations), and no dominance (multiple independent clusters). Compared to traditional bar charts and choropleths, the method provided deeper insights into inter-state performance dynamics. This study offers an innovative method for visualizing rankings and leadership patterns in rectangular datasets. By combining a multi-level basket model with co-word network analysis and open-source R scripts, users can quickly generate interpretable, cluster-based dominance diagrams. The approach is scalable, customizable, and applicable to a variety of fields, including healthcare, education, and public policy. Future work may extend this model to dynamic visual tools and broader interdisciplinary applications.

BackgroundWhen developing a clinical prediction model (CPM), a case-mix shift could occur in the development dataset where the distribution of individual predictors changes, potentially affecting model performance. This study exploits the case-mix shift that is already observed in the development dataset to address the case-mix shift between the development and deployment phase of a CPM.MethodsWe propose a Membership-based method to correct for case-mix shift in the development phase of CPMs. This method uses a probabilistic similarity metric to re-weight data samples in the source set (before the case-mix shift) to more closely match the target set (after the case-mix shift), assuming the target set reflects the target population. We apply the proposed method in a real-world dataset of myocardial infarction patients with out-of-hospital cardiac arrest within 90 days as the outcome. We design nine scenarios (including case-mix shift and no case-mix shift with a range of target/source sets sample sizes) to explore the impact on predictive performance of CPM developed with the proposed method in comparison to CPMs developed by either using all data samples but ignore the shift, or only using the most recent data. We report calibration and discrimination on development and 200 bootstrap samples.Results and ConclusionsThe proposed method shows promise in accounting for case-mix shift when developing a CPM, particularly when the target set sample size is insufficient. In a partial case-mix shift scenario with an insufficient target sample size, the Membership-based model achieved an optimism-adjusted calibration slope (c-slope) of 0.98, outperforming other models. Conversely, when the target set sample size is sufficient, the Unweighted model on target data only had an optimism-adjusted c-slope of 0.95, compared to 0.92 for the Membership-based model. In complete case-mix shift cases, the Membership-based and Unweighted on target data only models performed similarly. Both achieved an optimism-adjusted c-slope of 0.77 with insufficient target sample size, and optimism-adjusted c-slope of 0.94 with a sufficient target sample size. Further investigation and testing are needed, as well as accounting for other types of data distribution shift to improve model fit for the latest distribution shift in the development dataset.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12874-025-02621-2.

As hypertension and diabetes emerge as co-morbidities among people living with HIV/AIDS (PLWH), the need for robust epidemiological research to inform policy and action is imperative. Proper reporting and handling of missing data are crucial in such studies to avoid loss of statistical power and precision and generate unbiased results. We aimed to assess the reporting and handling of missing data in published studies of co-morbid hypertension and diabetes among PLWH. We searched in PubMed for cross-sectional studies of co-morbid hypertension and diabetes among PLWH published worldwide between January 1990 and June 2023. We extracted data on reporting of missing data (quantity, type, where it occurred, and any bias assessment) and how it was handled. Of 2179 records identified, 154 studies were included among which 53 (34.4%) reported missing data, primarily within exposure variables such as CD4 count and viral load. Only 19 of these studies (37.7%) cited reasons for missingness, predominantly attributed to lack of documentation and non-response. Out of the 24 (45.5%) studies that detailed how they handled missing data, the majority (16 studies; 30.2%) used complete case analysis. Only 5/53 studies (9.43%) adopted multiple imputation methods. The potential biases introduced by missing data were acknowledged in only 12/53 (22.6%) studies. The reporting and handling of missing data in hypertension and diabetes studies among PLWH are currently suboptimal. Enhanced understanding of why data is missing and choosing appropriate methods to address it is paramount to reduce potential biases. Adopting and adhering to comprehensive guidelines for managing missing data is a pressing need and will ensure that more accurate results are better represented in PLWH population.

Test-negative design (TND) studies are increasingly common in evaluating vaccine effectiveness (VE) for various infectious diseases. TND studies are susceptible to bias due to disease outcome misclassification caused by imperfect test sensitivity and specificity. Several bias correction methods have been proposed. However, sample size or power considerations for TND studies incorporating bias correction for such misclassification have not yet been investigated. We used Monte Carlo simulations to assess how bias correction affects the statistical power and sample size for VE estimation in TND studies. Simulations were conducted under varying levels of diagnostic test sensitivities (60%, 80%, and 95%). Bias correction was implemented using the multiple over-imputation method, which accounts for test misclassification through a parametric bootstrapping approach. Using a malaria vaccine as an example, we defined six vaccination status categories based on the time since receipt of the third or fourth vaccine dose. In the simulated target population, vaccination coverage was assumed to be low (< 10%) except for the group vaccinated more than 12 months after dose 4. We assumed relatively low VE (< 50%) against clinical malaria cases and a 30% malaria positivity rate among unvaccinated individuals presenting with malaria-related symptoms. Statistical power to detect VE was estimated for each vaccination status, both with and without bias correction. Estimated VEs based on observed data were consistently underestimated across all vaccination status groups due to diagnostic misclassification. In contrast, bias-corrected estimates were approximately unbiased but displayed wider confidence intervals, with their precision decreasing at lower test sensitivities. Statistical power to detect VE declined substantially when diagnostic test sensitivity was low. For instance, at 80% sensitivity, only three vaccination status groups reached 80% power with a sample size of 10,000, whereas the same power was achieved with just 6,000 individuals under a perfect test. Bias due to imperfect diagnostic testing can substantially reduce the power of TND studies. Power calculations should account for outcome misclassification and potential correction methods. Failure to do so may lead to underpowered studies and misleading VE estimates.

Pre-post designs are widely used in clinical trials and experimental studies to assess the effectiveness of treatments. Common statistical methods for analyzing pre-post data include analysis of variance (ANOVA) using post-treatment or the change from baseline, analysis of covariance (ANCOVA) with homogeneous or heterogeneous slopes, and linear mixed models (LMM). While numerous studies have compared these methods, limited studies have investigated the impact of adjusting for influential baseline covariates under different randomization approaches. In this study, we conducted a series of comprehensive simulation studies to investigate the impact of adjusting baseline covariates under several randomization approaches: simple randomization, stratified block randomization, and covariate adaptive randomization using the minimization method by Pocock and Simon. Results demonstrated that when no covariates were considered in the randomization approach, the two ANCOVA methods always have good performance. Adjusting for relevant baseline covariates led to substantial power gains, with the extent of these gains depending on the size of the covariate effects and the randomization approach employed. Stratified block randomization and covariate adaptive randomization consistently outperformed simple randomization in terms of power gains after adjusting for covariates, with covariate adaptive randomization becoming more superior as the number of covariates increased.

BackgroundLongitudinal studies are essential for understanding health trajectories and determinants over time. Successfully re-engaging and monitoring participants at key stages such as the transition from adolescence to adulthood is crucial. This stage of life is characterized by many changes and challenges and is considered critical for the manifestation of mental health problems. This study addresses challenges of contacting, re-activation and comparability of individuals in transition to adulthood with the initial population-based sample last contacted up to 9 years ago.MethodsIn 2024, former participants of the “German Health Interview and Examination Survey for Children and Adolescents” (KiGGS) aged 16–25 years were invited to register in the new online panel infrastructure “Health in Germany” and participate in the “Study on Mental Health in Emerging Adulthood” (JEPSY) via a push-to-online panel approach. Logistic regression models identified predictors of registration and study participation. Weighting adjustments were applied to analyse and correct for selective participation. To assess potential selection bias, life satisfaction of JEPSY participants was benchmarked against a representative sample.ResultsAmong 11,737 invitees, 4,451 (37.9%) registered, and among these, 3,063 (68.8%) completed the JEPSY survey. For participants aged 11 + years, higher probabilities for registration and participation were evident for females (OR = 2.61 [95% CI: 2.37–2.88]); OR = 2.85, [2.56–3.18]), individuals with higher occupational (OR = 1.08, [1.04–1.13]; OR = 1.06, [1.01–1.11]) or educational status (OR = 1.13, [1.08–1.17]; OR = 1.13, [1.08–1.17]). Individuals with higher emotional problems were less likely to register (OR = 0.98, [0.97–0.99]) and participate (OR = 0.98, [0.97–0.99]). Weighting adjustments reduced biases but increased statistical variance. Benchmarking life satisfaction revealed no significant differences between JEPSY participants and their representative counterparts.ConclusionsThis study demonstrated the feasibility of re-activating former participants and integrating them into a modern online panel. While re-activation was successful for a substantial proportion of invitees, specific challenges (e.g., selective re-participation) remain. The findings provide insights for refining re-contact and retention strategies to improve representativeness and quality of longitudinal mental health research, ultimately enabling more accurate monitoring of health trajectories and their determinants over time.

BackgroundWith the rise of large language models, the application of artificial intelligence in research is expanding, possibly accelerating specific stages of the research processes. This study aims to compare the accuracy, completeness and relevance of chatbot-generated responses against human responses in evidence synthesis as part of a scoping review.MethodsWe employed a structured survey-based research methodology to analyse and compare responses between two human researchers and four chatbots (ZenoChat, ChatGPT 3.5, ChatGPT 4.0, and ChatFlash) to questions based on a pre-coded sample of 407 articles. These questions were part of an evidence synthesis of a scoping review dealing with digitally supported interaction between healthcare workers.ResultsThe analysis revealed no significant differences in judgments of correctness between answers by chatbots and those given by humans. However, chatbots’ answers were found to recognise the context of the original text better, and they provided more complete, albeit longer, responses. Human responses were less likely to add new content to the original text or include interpretation. Amongst the chatbots, ZenoChat provided the best-rated answers, followed by ChatFlash, with ChatGPT 3.5 and ChatGPT 4.0 tying for third. Correct contextualisation of the answer was positively correlated with completeness and correctness of the answer.ConclusionsChatbots powered by large language models may be a useful tool to accelerate qualitative evidence synthesis. Given the current speed of chatbot development and fine-tuning, the successful applications of chatbots to facilitate research will very likely continue to expand over the coming years.

In biomedical research, the calculation of sample size is a critical component of study design. Adequate sample size ensures the reliability of statistical tests, including the chi-square test. This manuscript outlines the use of an online sample size calculator for chi-square tests. The paper includes detailed explanations of the formulas used in the calculations and highlights the importance of power analysis in planning research studies. This tool is designed to assist and guide researchers in determining the optimal sample size for detecting statistically significant differences in categorical data. We describe the theory behind the chi-square test, the statistical principles involved in sample size calculation, and the specific methodology for using the sample size calculator. The calculator is freely available to use at https://hanif-shiny.shinyapps.io/chi-sq/.