BMC Medical Research Methodology最新文献

Background: The use of Expert Opinion (EO) in clinical guidelines is highly variable and lacks standardization, leading to ongoing controversy. A clear and universally accepted definition of EO is also lacking. To date, no research has systematically assessed how guideline-developing societies conceptualize and apply EO. This study aims to map methodological manuals, evaluate their rationale for EO use, examine its foundations, and synthesize a comprehensive definition.

Methods: Systematic searches for clinical guidelines were conducted in PubMed to identify guideline-developing societies, supplemented by additional searches. Systematic searches were then conducted to identify methodological manuals from these societies. Screening was performed independently by two reviewers, and data extraction was conducted using piloted forms. Findings were summarized through narrative evidence synthesis using descriptive statistics.

Results: A total of 473 national and international societies were identified, and methodological manuals from 98 societies were mapped and analysed. These manuals included 61 handbooks, 29 journal articles, and 8 websites. EO is mentioned in 65 (66%) manuals, with substantial variation in its utilization and terminology. EO is primarily used in two contexts: (1) filling evidence gaps (72%), and (2) interpreting existing evidence (8%). In the remaining 20%, EO use is unclear. Five main foundations could be identified as a potential basis for EO (clinical experience, indirect evidence, low-quality evidence, mechanism-based reasoning, and expert evidence/witnesses). Based on these findings, a novel comprehensive definition of EO was synthesized.

Conclusions: EO is widely used to address evidence gaps and interpret ambiguous evidence, underscoring its importance in guideline development. However, the variability in its application and conceptualization across societies highlights the need for standardization. We propose a comprehensive EO definition as a first step towards standardization to improve consistency, transparency, and clinical decision-making.

Background: Meta-analysis is a useful method for combining evidence from multiple studies to detect treatment effects that could perhaps not be identified in a single study. While traditionally meta-analysis has assumed that populations of included studies are comparable, over recent years the development of precision medicine has led to identification of predictive genetic biomarkers which has resulted in trials conducted in mixed biomarker populations. For example, early trials may be conducted in patients with any biomarker status with no subgroup analysis, later trials may be conducted in patients with any biomarker status and subgroup analysis, and most recent trials may be conducted in biomarker-positive patients only. This poses a problem for traditional meta-analysis methods which rely on the assumption of somewhat comparable populations across studies. In this review, we provide a background to meta-analysis methods allowing for synthesis of data with mixed biomarker populations across trials.

Methods: For the methodological review, PubMed was searched to identify methodological papers on evidence synthesis for mixed populations. Several identified methods were applied to an illustrative example in metastatic colorectal cancer.

Results: We identified eight methods for evidence synthesis of mixed populations where three methods are applicable to pairwise meta-analysis using aggregate data (AD), three methods are applicable to network meta-analysis using AD, and two methods are applicable to network meta-analysis using AD and individual participant data (IPD). The identified methods are described, including a discussion of the benefits and limitations of each method.

Conclusions: Methods for synthesis of data from mixed populations are split into methods which use (a) AD, (b) IPD, and (c) both AD and IPD. While methods which utilise IPD achieve superior statistical qualities, this is at the expense of ease of access to the data. Furthermore, it is important to consider the context of the decision problem in order to select the most appropriate modelling framework.

Background: Interventional studies are intended to provide robust evidence. Yet poorly designed or conducted studies may bias research results and skew resulting evidence. While there have been advances in the assessment of risk of bias, it is unclear how to intervene against risks of bias during study design and conduct.

Objective: To identify interventions to reduce or predict risk of bias in interventional studies during their design and conduct.

Search strategy: For this scoping review, we searched three electronic bibliographic databases (MEDLINE, Embase, and Cochrane Library) and nine grey literature sources and Google from in September 2024. This was supplemented by a natural language processing fuzzy matching search of the top 2000 relevant publications in the electronic bibliographic databases. Publications were included if they described the implementation and effectiveness of an intervention during study design or conduct aimed at reducing risk of bias in interventional studies. The characteristics and effect of the interventions were recorded.

Result: We identified, and reviewed the title and abstracts of, a total of 41,793 publications, reports, documents and grey literature, with 24,677 from electronic bibliographic databases and 17,140 from grey literature sources. There were 67 publications from bibliographic databases and 24 items from grey literature that were considered potentially eligible for inclusion, and the full-text of these were reviewed. Only three studies met the inclusion criteria. The first intervention was offering education and training to researchers during study design. This training included the implementation of a more rigorous participant screening process and systematic participant tracking program that reduced loss to follow-up and missing data, particularly for long-term follow-up trials. The second intervention was introducing an independent clinical events committee during study conduct. This was intended to mitigate bias due to conflicts of interest affecting the analysis and interpretation of results. The third intervention was to provide participants with financial incentives in randomized controlled trials, so that participants could more actively accomplish the requirements of the trials.

Conclusion: Despite the major impact of risk of bias on study outcomes, there are few empirical interventions to address this during study design or conduct.

Background: Tracking and understanding the progress and experiences of health workers and the outcomes of workforce decisions are essential for evidence-based workforce planning. In this scoping review, we aim to identify longitudinal studies that prospectively tracked healthcare professionals and that specifically focused on workforce issues such as career preferences, choices, and working conditions, and summarise the different approaches and methods used for tracking.

Methods: We searched MEDLINE, Embase, Global Health, PsycINFO, CINAHL, Education Resource Information Center (ERIC), EconLit and the Cochrane Library for articles published between 2000-2022 that longitudinally tracked doctors, nurses, midwives, physician associates/assistants. We further compared articles and conducted a back-and-forward citation search to identify longitudinal tracking studies which sometimes have multiple published articles. We developed a typology of the different tracking approaches, and summarised the major areas assessed and tracked by different studies.

Results: We identified and analysed 263 longitudinal tracking studies. Based on population recruitment and follow-up methods, we grouped studies into seven categories (cohort studies, multiple-cohort studies, baseline and data linkage studies, baseline and short repeated measure studies, baseline-only studies, data linkage-only studies and repeated survey studies). The majority of studies included used a cohort or multiple-cohort design (n = 180), and several others also used data linkage (n = 45) and repeated measure approaches (n = 24). Sixty-two studies recruited participants while they were students and followed them until they became the active workforce, and nearly half of the included studies started directly from the active workforce stage. Most of the included studies examined workforce issues including employment status, preference or intention (to leave/remain/migrate, specific speciality or location etc.), and work environment, however there was a lack of widely used measurement tools for workforce issues. Additionally, nearly 40% examined wellbeing issues and a subset (20%) examined physical health in the context of workforce-related issues.

Conclusion: We described a large number of different healthcare professional longitudinal tracking studies. In order for longitudinal tracking to contribute to effective workforce planning, we recommend employing a mix of cohort and data linkage approaches to collect data across the different stages of the workforce 'working lifespan', and using and continuing to test standardised measurement instruments to better capture experiences related to workforce and wellbeing.

Background: The use of graphical models in the multimorbidity context is increasing in popularity due to their intuitive visualization of the results. A comprehensive understanding of the model itself is essential for its effective utilization and optimal application. This article is a practical guide on the use of graphical models to better understand multimorbidity. It provides a tutorial with a focus on the interpretation of the model structure and of the parameter values. In this study, we analyze data related to a cohort of 214,401 individuals, who were assisted by the Local Health Unit of the province of Padova (north-eastern Italy), collecting information from hospital discharge forms.

Methods: We explain some fundamental concepts, with special attention to the difference between marginal and conditional associations. We emphasize the importance of considering multimorbidity as a network, where the variables involved are part of an interconnected system of interactions, to correct for spurious effects in the analysis. We show how to analyze the network structure learned from the data by introducing and explaining some centrality measures. Finally, we compare the model obtained by adjusting for population characteristics with the results of a stratified analysis.

Results: Using examples from the estimated model, we demonstrate the key differences between marginal and conditional associations. Specifically, we show that, marginally, all variables appear associated, while this is not the case when considering conditional associations, where many variables appear to be conditionally independent given the others. We present the results from the analysis of centrality indices, revealing that cardiovascular diseases occupy a central position in the network, unlike more peripheral conditions such as sensory organ diseases. Finally, we illustrate the differences between networks estimated in subpopulations, highlighting how disease associations vary across different groups.

Conclusion: Graphical models are a versatile tool for analyzing multimorbidity, offering insights into disease associations while controlling for the effects of other variables. This paper provides an overview of graphical models without focusing on detailed methodology, highlighting their utility in understanding network structures and potential subgroup differences, such as gender-related variations in multimorbidity patterns.

Background: This study presents a Bayesian Adaptive Semiparametric approach designed to address the challenges of pediatric randomized controlled trials (RCTs). The study focuses on efficiently handling primary and secondary endpoints, a critical aspect often overlooked in pediatric trials. This methodology is particularly pertinent in scenarios where sparse or conflicting prior data are present, a common occurrence in pediatric research, particularly for rare diseases or conditions.

Method: Our approach considers Bayesian adaptive design, enhanced with B-Spline Semiparametric priors, allowing for the dynamic updating of priors with ongoing data. This improves the efficiency and accuracy of the treatment effect estimation. The Semiparametric prior inherent flexibility makes it suitable for pediatric populations, where responses to treatment can be highly variable. The design operative characteristics were assessed through a simulation study, motivated by the real-world case of the REnal SCarring Urinary infEction Trial (RESCUE).

Result: We demonstrate that Semiparametric prior parametrization exhibits an improved tendency to correctly declare the treatment effect at the study conclusion, even if recruitment challenges, uncertainty, and prior-data conflict arise. Moreover, the Semiparametric prior design demonstrates an improved ability in truly stopping for futility, with this tendency varying with the sample size and discontinuation rates. Approaches based on Parametric priors are more effective in detecting treatment efficacy during interim assessments, particularly with larger sample sizes.

Conclusion: Our findings indicate that these methods are especially effective in managing the complexities of pediatric trials, where prior data may be limited or contradictory. The flexibility of Semiparametric prior design in incorporating new evidence proves advantageous in addressing recruitment challenges and making informed decisions with restricted data.

Background: Surveys are a useful tool for eliciting self-reported health information, but the accuracy of such information may vary. We examined the agreement between self-reported health information and medical record data among 116,288 military service members and veterans enrolled in a longitudinal cohort.

Methods: Millennium Cohort Study participants who separated from service and registered for health care in the Veterans Health Administration (VHA) by September 18, 2020, were eligible for inclusion. Baseline and follow-up survey responses (2001-2016) about 39 medical conditions, health behaviors, height, and weight were compared with analogous information from VHA and military medical records. Medical record diagnoses were classified as one qualifying ICD code in any diagnostic position between October 1, 1999, and September 18, 2020. Additional analyses were restricted to medical record diagnoses occurring before survey self-report and using specific diagnostic criteria (two outpatient or one inpatient ICD code). Positive, negative, and overall (Youden's J) agreement was calculated for categorical outcomes; Bland-Altman plots were examined for continuous measures.

Results: Among 116,288 participants, 71.8% self-reported a diagnosed medical condition. Negative agreement between self-reported and VHA medical record diagnoses was > 90% for most (80%) conditions, but positive agreement was lower (6.4% to 56.3%). Mental health conditions were more frequently recorded in medical records, while acute conditions (e.g., bladder infections) were self-reported at a higher frequency. Positive agreement was lower when analyses were restricted to medical record diagnoses occurring prior to survey self-report. Specific diagnostic criteria resulted in higher overall agreement.

Conclusions: While negative agreement between self-reported and medical record diagnoses was high in this population, positive and overall agreement were not strong and varied considerably by health condition. Though the limitations of survey-reported health conditions should be considered, using multiple data sources to examine health outcomes in this population may have utility for research, clinical planning, or public health interventions.

Understanding causality, over mere association, is vital for researchers wishing to inform policy and decision making - for example, when seeking to improve population health outcomes. Yet, contemporary causal inference methods have not fully tackled the complexity of data hierarchies, such as the clustering of people within households, neighbourhoods, cities, or regions. However, complex data hierarchies are the rule rather than the exception. Gaining an understanding of these hierarchies is important for complex population outcomes, such as non-communicable disease, which is impacted by various social determinants at different levels of the data hierarchy. The alternative of analysing aggregated data could introduce well-known biases, such as the ecological fallacy or the modifiable areal unit problem. We devise a hierarchical causal diagram that encodes the multilevel data generating mechanism anticipated when evaluating non-communicable diseases in a population. The causal diagram informs data simulation. We also provide a flexible tool to generate synthetic population data that captures all multilevel causal structures, including a cross-level effect due to cluster size. For the very first time, we can then quantify the ecological fallacy within a formal causal framework to show that individual-level data are essential to assess causal relationships that affect the individual. This study also illustrates the importance of causally structured synthetic data for use with other methods, such as Agent Based Modelling or Microsimulation Modelling. Many methodological challenges remain for robust causal evaluation of multilevel data, but this study provides a foundation to investigate these.

Background: The integration of historical data into ongoing clinical trials through Bayesian Dynamic Borrowing offers significant advantages, including reduced sample size, trial duration, and associated costs. However, challenges such as ensuring exchangeability between historical and current data and mitigating Type I error inflation remain critical. This study proposes a Bayesian group-sequential design incorporating a Self-Adaptive Mixture (SAM) prior framework to address these challenges in medical device trials.

Methods: The SAM prior combines informative priors derived from historical data with weakly informative priors, dynamically adjusting the weight of historical information based on congruence with current trial data. The design includes interim analyses, with Bayesian decision rules leveraging futility and efficacy boundaries derived using the frequentist spending functions. Effective Sample Size calculations informed adjustments to sample size and allocation ratios between experimental and control arms at each interim. The methodology was evaluated using a motivating example from a cardiovascular device trial with a noninferiority hypothesis.

Results: Four historical studies with substantial heterogeneity were incorporated. The SAM prior showed improved adaptation to prior-data conflicts compared to static methods, maintaining Type I error and Power at their nominal levels. In the motivating trial, the MAP prior was approximated as a mixture of beta distributions, facilitating congruence testing and posterior inference. Simulation studies confirmed the proposed design's efficiency under both congruent and incongruent scenarios.

Conclusions: The proposed Bayesian Group-Sequential Design with SAM prior offers a robust, adaptive framework for medical device trials, balancing statistical rigor with clinical interpretability. This approach enhances decision-making and supports timely, cost-effective evaluations, particularly in dynamic contexts like medical device development.