BMC Medical Research Methodology最新文献

{"title":"Co-creation methods for public health research - characteristics, benefits, and challenges: a Health CASCADE scoping review.","authors":"Danielle Marie Agnello, Vinayak Anand-Kumar, Qingfan An, Janneke de Boer, Lea Rahel Delfmann, Giuliana Raffaella Longworth, Quentin Loisel, Lauren McCaffrey, Artur Steiner, Sebastien Chastin","doi":"10.1186/s12874-025-02514-4","DOIUrl":"10.1186/s12874-025-02514-4","url":null,"abstract":"<p><strong>Background: </strong>Co-creation engages diverse stakeholders, including marginalized populations, in collaborative problem-solving to enhance engagement and develop contextually appropriate solutions. It is increasingly recognized as a way to democratize research and improve the impact of interventions, services, and policies. However, the lack of synthesized evidence on co-creation methods limits methodological rigor and the establishment of best practices. This review aimed to identify co-creation methods in academic literature and analyze their characteristics, target groups, and associated benefits and challenges.</p><p><strong>Methods: </strong>This scoping review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews. The search was conducted in the Health CASCADE database v1.5 (including CINAHL, PubMed, and 17 additional databases via ProQuest) from January 1970 to March 2022. Data was aggregated and summarized, with qualitative data analyzed using Braun and Clarke's six-phase thematic analysis approach.</p><p><strong>Results: </strong>The review included 266 articles, identifying 248 distinct co-creation methods published between 1998 and 2022. Most methods were rooted in participatory paradigms (147 methods), with 49 methods derived from co-approaches like co-creation, co-design, and co-production, and 11 from community-based health promotion and action research. Methods were applied across 40 target populations, including children, adults, and marginalized groups. Many methods (62.3%) were delivered face-to-face, with 40 articles incorporating digital tools. Thematic analysis revealed nine benefits, such as enhanced creativity, empowerment, and improved communication, and six challenges, including resource constraints and systemic and structural barriers.</p><p><strong>Conclusion: </strong>This review emphasizes the importance of robust documentation and analysis of co-creation methods to inform their application in public health. Findings support the development of collaborative co-creation processes that are responsive to the needs of diverse populations, thereby enhancing the overall effectiveness and cultural sensitivity of the outcomes. This review highlights the potential of co-creation methods to promote equity and inclusion while emphasizing the importance of evaluating and selecting methods tailored to specific objectives, offering a critical resource for planning, conducting, and evaluating co-creation projects.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"60"},"PeriodicalIF":3.9,"publicationDate":"2025-03-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11884017/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143572067","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Validity of using a semi-automated screening tool in a systematic review assessing non-specific effects of respiratory vaccines.","authors":"Charlie Holland, Daniel B Oakes, Mohinder Sarna, Kevin Ek Chai, Leo Ng, Hannah C Moore","doi":"10.1186/s12874-025-02511-7","DOIUrl":"10.1186/s12874-025-02511-7","url":null,"abstract":"<p><strong>Background: </strong>The abstract screening process of systematic reviews can take thousands of hours by two researchers. We aim to determine the reliability and validity of Research Screener, a semi-automated abstract screening tool within a systematic review on non-specific and broader effects of respiratory vaccines on acute lower respiratory infection hospitalisations and antimicrobial prescribing patterns in young children.</p><p><strong>Methods: </strong>We searched online databases for Medline, Embase, CINAHL, Scopus and ClinicalTrials.gov from inception until 24th January 2024. We included human studies involving non-specific and broader effects of respiratory vaccines and excluded studies investigating live-attenuated vaccines. The RS trial compared relevant abstracts flagged by RS to manual screening. RS ranks abstracts by relevance based on seed articles used to validate the search strategy. Abstracts are re-ranked following reviewers' feedback. Two reviewers screened RS independently with a third reviewer resolving conflicts; three reviewers screened manually with a fourth reviewer resolving conflicts.</p><p><strong>Results: </strong>After removal of duplicates, 9,727 articles were identified for abstract screening. Of those, 3,000 were randomly selected for screening in RS, with 18% (540) screened in RS and 100% manually. In RS, 99 relevant articles were identified. After comparing RS to manual screening and completing full-text review on 26 articles not captured by RS, 4 articles were missed by RS (2 due to human error, 2 not yet screened). Hence, RS captured articles accurately whilst reducing the screening load.</p><p><strong>Conclusions: </strong>RS is a valid and reliable tool that reduces the amount of time spent screening articles for large-scale systematic reviews. RS is a useful tool that should be considered for streamlining the process of systematic reviews.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"59"},"PeriodicalIF":3.9,"publicationDate":"2025-03-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11884005/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143572083","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Addressing challenges with Matching-Adjusted Indirect Comparisons to demonstrate the comparative effectiveness of entrectinib in metastatic ROS-1 positive Non-Small Cell Lung Cancer.","authors":"Cyril Esnault, Louise Baschet, Vanessa Barbet, Gaëlle Chenuc, Maurice Pérol, Katia Thokagevistk, David Pau, Matthias Monnereau, Lise Bosquet, Thomas Filleron","doi":"10.1186/s12874-025-02500-w","DOIUrl":"10.1186/s12874-025-02500-w","url":null,"abstract":"<p><strong>Background: </strong>Matching Adjusted Indirect Comparison (MAIC) is a statistical method used to adjust for potential biases when comparing treatment effects between separate data sources, with aggregate data in one arm, and individual patients data in the other. However, acceptance of MAIC in health technology assessment (HTA) is challenging because of the numerous biases that can affect the estimates of treatment effects - especially with small sample sizes, increasing the risk of convergence issues. We suggest statistical approaches to address some of the challenges in supporting evidence from MAICs, applied to a case study.</p><p><strong>Methods: </strong>The proposed approaches were illustrated with a case study comparing an integrated analysis of three single-arm trials of entrectinib with the French standard of care using the Epidemio-Strategy and Medical Economics (ESME) Lung Cancer Data Platform, in metastatic ROS1-positive Non-Small Cell Lung Cancer (NSCLC) patients. To obtain convergent models with balanced treatment arms, a transparent predefined workflow for variable selection in the propensity score model, with multiple imputation of missing data, was used. To assess robustness, multiple sensitivity analyses were conducted, including Quantitative Bias Analyses (QBA) for unmeasured confounders (E-value, bias plot), and for missing at random assumption (tipping-point analysis).</p><p><strong>Results: </strong>The proposed workflow was successful in generating satisfactory models for all sub-populations, that is, without convergence problems and with effectively balanced key covariates between treatment arms. It also gave an indication of the number of models tested. Sensitivity analyses confirmed the robustness of the results, including to unmeasured confounders. The QBA performed on the missing data allowed to exclude the potential impact of the missing data on the estimate of comparative effectiveness, even though approximately half of the ECOG Performance Status data were missing.</p><p><strong>Conclusions: </strong>To the best of our knowledge, we present the first in-depth application of QBA in the context of MAIC. Despite the real-world data limitations, with this MAIC, we show that it is possible to confirm the robustness of the results by using appropriate statistical methods.</p><p><strong>Trial registration: </strong>NA.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"57"},"PeriodicalIF":3.9,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11872303/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143536471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The reporting quality and methodological quality of dynamic prediction models for cancer prognosis.","authors":"Peijing Yan, Zhengxing Xu, Xu Hui, Xiajing Chu, Yizhuo Chen, Chao Yang, Shixi Xu, Huijie Cui, Li Zhang, Wenqiang Zhang, Liqun Wang, Yanqiu Zou, Yan Ren, Jiaqiang Liao, Qin Zhang, Kehu Yang, Ling Zhang, Yunjie Liu, Jiayuan Li, Chunxia Yang, Yuqin Yao, Zhenmi Liu, Xia Jiang, Ben Zhang","doi":"10.1186/s12874-025-02516-2","DOIUrl":"10.1186/s12874-025-02516-2","url":null,"abstract":"<p><strong>Background: </strong>To evaluate the reporting quality and methodological quality of dynamic prediction model (DPM) studies on cancer prognosis.</p><p><strong>Methods: </strong>Extensive search for DPM studies on cancer prognosis was conducted in MEDLINE, EMBASE, and the Cochrane Library databases. The Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) and the Prediction model Risk of Bias Assessment Tool (PROBAST) were used to assess reporting quality and methodological quality, respectively.</p><p><strong>Results: </strong>A total of 34 DPM studies were identified since the first publication in 2005, the main modeling methods for DPMs included the landmark model and the joint model. Regarding the reporting quality, the median overall TRIPOD adherence score was 75%. The TRIPOD items were poorly reported, especially the title (23.53%), model specification, including presentation (55.88%) and interpretation (50%) of the DPM usage, and implications for clinical use and future research (29.41%). Concerning methodological quality, most studies were of low quality (n = 30) or unclear (n = 3), mainly due to statistical analysis issues.</p><p><strong>Conclusions: </strong>The Landmark model and joint model show potential in DPM. The suboptimal reporting and methodological qualities of current DPM studies should be improved to facilitate clinical application.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"58"},"PeriodicalIF":3.9,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11872325/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143536489","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Handling missing values in patient-reported outcome data in the presence of intercurrent events.","authors":"Doranne Thomassen, Satrajit Roychoudhury, Cecilie Delphin Amdal, Dries Reynders, Jammbe Z Musoro, Willi Sauerbrei, Els Goetghebeur, Saskia le Cessie","doi":"10.1186/s12874-025-02510-8","DOIUrl":"10.1186/s12874-025-02510-8","url":null,"abstract":"<p><strong>Introduction: </strong>As patient-reported outcomes (PROs) are increasingly used in the evaluation of medical treatments, it is important that PROs are carefully analyzed and interpreted. This may be challenging due to substantial missing values. The missingness in PROs is often closely related to patients' disease status. In that case, using observed information about intercurrent events (ICEs) such as disease progression and death will improve the handling of missing PRO data. Therefore, the aim of this study was to develop imputation models for repeated PRO measurements that leverage information about ICEs.</p><p><strong>Methods: </strong>We assumed a setting in which missing PRO measurements are missing at random given observed measurements, as well as the occurrence and timing of ICEs, and potentially other (baseline or time-varying) covariates. We then showed how these missingness assumptions can be translated into concrete imputation models that also account for a longitudinal data structure. The resulting models were applied to impute anonymized PRO data from a single-arm clinical trial in patients with advanced lung cancer.</p><p><strong>Results: </strong>In our trial example, accounting for death and other ICEs in the imputation of missing data led to lower estimated mean health-related quality of life (while alive) compared to an available case analysis and a naive linear mixed model imputation.</p><p><strong>Conclusion: </strong>Information about the timing and occurrence of ICEs contribute to a more plausible handling of missing PRO data. To account for ICE information when handling missing PROs, the missing data model should be separated from the analysis model.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"56"},"PeriodicalIF":3.9,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11872335/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143536485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Handling of missing component information for common composite score outcomes used in axial spondyloarthritis research when complete-case analysis is unbiased.","authors":"Christos Polysopoulos, Stylianos Georgiadis, Lykke Midtbøll Ørnbjerg, Almut Scherer, Daniela Di Giuseppe, Merete Lund Hetland, Michael John Nissen, Gareth T Jones, Bente Glintborg, Anne Gitte Loft, Johan Karlsson Wallman, Karel Pavelka, Jakub Závada, Ayten Yazici, Maria José Santos, Adrian Ciurea, Burkhard Möller, Brigitte Michelsen, Pawel Mielnik, Johanna Huhtakangas, Heikki Relas, Katja Perdan Pirkmajer, Ziga Rotar, Ross MacDonald, Bjorn Gudbjornsson, Irene van der Horst-Bruinsma, Marleen van de Sande, Myriam Riek","doi":"10.1186/s12874-025-02515-3","DOIUrl":"10.1186/s12874-025-02515-3","url":null,"abstract":"<p><strong>Background: </strong>Observational data on composite scores often comes with missing component information. When a complete-case (CC) analysis of composite scores is unbiased, preferable approaches of dealing with missing component information should also be unbiased and provide a more precise estimate. We assessed the performance of several methods compared to CC analysis in estimating the means of common composite scores used in axial spondyloarthritis research.</p><p><strong>Methods: </strong>Individual mean imputation (IMI), the modified formula method (MF), overall mean imputation (OMI), and multiple imputation of missing component values (MI) were assessed either analytically or by means of simulations from available data collected across Europe. Their performance in estimating the means of the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), the Bath Ankylosing Spondylitis Functional Index (BASFI), and the Ankylosing Spondylitis Disease Activity Score based on C-reactive protein (ASDAS-CRP) in cases where component information was set missing completely at random was compared to the CC approach based on bias, variance, and coverage.</p><p><strong>Results: </strong>Like the MF method, IMI uses a modified formula for observations with missing components resulting in modified composite scores. In the case of an unbiased CC approach, these two methods yielded representative samples of the distribution arising from a mixture of the original and modified composite scores, which, however, could not be considered the same as the distribution of the original score. The IMI and MF method are, thus, intrinsically biased. OMI provided an unbiased mean but displayed a complex dependence structure among observations that, if not accounted for, resulted in severe coverage issues. MI improved precision compared to CC and gave unbiased means and proper coverage as long as the extent of missingness was not too large.</p><p><strong>Conclusions: </strong>MI of missing component values was the only method found successful in retaining CC's unbiasedness and in providing increased precision for estimating the means of BASDAI, BASFI, and ASDAS-CRP. However, since MI is susceptible to incorrect implementation and its performance may become questionable with increasing missingness, we consider the implementation of an error-free CC approach a valid and valuable option.</p><p><strong>Trial registration: </strong>Not applicable as study uses data from patient registries.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"55"},"PeriodicalIF":3.9,"publicationDate":"2025-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11869558/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143531058","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A discrete-time split-state framework for multi-state modeling with application to describing the course of heart disease.","authors":"Ming Ding, Haiyi Chen, Feng-Chang Lin","doi":"10.1186/s12874-025-02512-6","DOIUrl":"10.1186/s12874-025-02512-6","url":null,"abstract":"<p><p>In chronic disease epidemiology, the investigation of disease etiology has largely focused on an endpoint, while the course of chronic disease is understudied, representing a knowledge gap. Multi-state models can be used to describe the course of chronic disease, such as Markov models which assume that the future state depends only on the present state, and semi-Markov models which allow transition rates to depend on the duration in the current state. However, these models are unsuitable for chronic diseases that are largely non-memoryless. We propose a Discrete-Time Split-State Framework that generates a process of substates by conditioning on past disease history and estimates discrete-time transition rates between substates as a function of duration in a (sub)state. Specifically, as the substates are created by conditioning on past history, they satisfy the Markov assumption, regardless of whether the original disease process is Markovian; and the transition rates are approximated by competing risks in a short time interval estimated from cause-specific Cox models. In the simulation study, we simulated a Markov process with an exponential distribution, a semi-Markov process with a Weibull distribution, and a non-Markov process with an exponential distribution. The coverage rate of transition rates estimated using our framework was 94% for the Markov process and 93% for the non-Markov process. However, the estimated transition rates were under coverage (72%) for the semi-Markov process, which is likely due to the approximation of transition rates in discrete time. In the application, we applied the framework to describe the course of heart disease in a large cohort study. In summary, the framework we proposed can be applied to both Markov and non-Markov processes and has potential to be applied to semi-Markov processes. For future research, as substates created using our framework track past disease history, the transition rates between substates have the potential to be used to derive summary estimates that characterize the disease course.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"54"},"PeriodicalIF":3.9,"publicationDate":"2025-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11869649/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143530972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing non-inferiority for binary matched-pairs data with missing values: a powerful and flexible GEE approach based on the risk difference.","authors":"Johannes Hengelbrock, Frank Konietschke, Juliane Herm, Heinrich Audebert, Annette Aigner","doi":"10.1186/s12874-025-02497-2","DOIUrl":"10.1186/s12874-025-02497-2","url":null,"abstract":"<p><strong>Background: </strong>Clinical studies often aim to test the non-inferiority of a treatment compared to an alternative intervention with binary matched-pairs data. These studies are often planned with methods for completely observed pairs only. However, if missingness is more frequent than expected or is anticipated in the planning phase, methods are needed that allow the inclusion of partially observed pairs to improve statistical power.</p><p><strong>Methods: </strong>We propose a flexible generalized estimating equations (GEE) approach to estimate confidence intervals for the risk difference, which accommodates partially observed pairs. Using simulated data, we compare this approach to alternative methods for completely observed pairs only and to those that also include pairs with missing observations. Additionally, we reconsider the study sample size calculation by applying these methods to a study with binary matched-pairs setting.</p><p><strong>Results: </strong>In moderate to large sample sizes, the proposed GEE approach performs similarly to alternative methods for completely observed pairs only. It even results in a higher power and narrower interval widths in scenarios with missing data and where missingness follows a missing (completely) at random (MCAR / MAR) mechanism. The GEE approach is also non-inferior to alternative methods, such as multiple imputation or confidence intervals explicitly developed for missing data settings. Reconsidering the sample size calculation for an observational study, our proposed approach leads to a considerably smaller sample size than the alternative methods.</p><p><strong>Conclusion: </strong>Our results indicate that the proposed GEE approach is a powerful alternative to existing methods and can be used for testing non-inferiority, even if the initial sample size calculation was based on a different statistical method. Furthermore, it increases the analytical flexibility by allowing the inclusion of additional covariates, in contrast to other methods.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"53"},"PeriodicalIF":3.9,"publicationDate":"2025-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11866877/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143522596","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Accounting for extent of non-compliance when estimating treatment effects on an ordinal outcome in randomized clinical trials.","authors":"Junxian Zhu, Jialiang Li, A Mark Richards, Mark Y Chan, Bee-Choo Tai","doi":"10.1186/s12874-025-02493-6","DOIUrl":"10.1186/s12874-025-02493-6","url":null,"abstract":"<p><strong>Background: </strong>In randomized clinical trials (RCTs) with non-compliance, evaluating the causal effects of interventions would lead to a more precise estimation of treatment effect when the estimand of interest is the effect of treatment amongst compliers. While there is a large body of literature addressing the issue of non-compliance for continuous, binary, and time-to-event outcomes, this issue is seldom discussed for ordinal outcomes.</p><p><strong>Methods: </strong>In this paper, we consider one-sided non-compliance. We introduce an extension of the inverse probability weighting (IPW) method for handling non-compliance involving an ordinal outcome by fully utilizing the information of non-compliance and defining it as a categorical variable to describe the extent of non-compliance. This is in contrast to the usual convention where compliance is regarded as a binary variable. We provide the identification and asymptotic distribution of the proposed method. We compare the proposed method (IPW_Dnew) with intention-to-treat (ITT), per protocol (PP), instrumental variable (IV), and IPW method via a simulation study and real-life data from the JOBS II intervention trial and the IMMACULATE trial.</p><p><strong>Results: </strong>Simulation results demonstrate that the proposed method performs better than other methods in terms of bias, coverage, mean squared error, power and Type I error under various scenarios, particularly in situations with selection bias and partial compliance. In the empirical study, a substantial estimate of partial compliance by IPW_Dnew implies that there may be a partial compliance effect.</p><p><strong>Conclusion: </strong>For ordinal outcome in the presence of non-compliance, we suggest using the proposed method to estimate the causal effect of treatment amongst compliers and partial compliers, especially when there exists selection bias.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"52"},"PeriodicalIF":3.9,"publicationDate":"2025-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11852586/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143499193","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Application of causal forests to randomised controlled trial data to identify heterogeneous treatment effects: a case study.","authors":"Eleanor Van Vogt, Anthony C Gordon, Karla Diaz-Ordaz, Suzie Cro","doi":"10.1186/s12874-025-02489-2","DOIUrl":"10.1186/s12874-025-02489-2","url":null,"abstract":"<p><strong>Background: </strong>Classical approaches to subgroup analysis in randomised controlled trials (RCTs) to identify heterogeneous treatment effects (HTEs) involve testing the interaction between each pre-specified possible treatment effect modifier and the treatment effect. However, individual significant interactions may not always yield clinically actionable subgroups, particularly for continuous covariates. Non-parametric causal machine learning approaches are flexible alternatives for estimating HTEs across many possible treatment effect modifiers in a single analysis.</p><p><strong>Methods: </strong>We conducted a secondary analysis of the VANISH RCT, which compared the early use of vasopressin with norepinephrine on renal failure-free survival for patients with septic shock at 28 days. We used classical (separate tests for interaction with Bonferroni correction), data-adaptive (hierarchical lasso regression), and non-parametric causal machine learning (causal forest) methods to analyse HTEs for the primary outcome of being alive at 28 days. Causal forests comprise honest causal trees, which use sample splitting to determine tree splits and estimate treatment effects separately. The modal initial (root) splits of the causal forest were extracted, and the mean value was used as a threshold to partition the population into subgroups with different treatment effects.</p><p><strong>Results: </strong>All three models found evidence of HTE with serum potassium levels. Univariable logistic regression OR 0.435 (95%CI [0.270, 0.683]. p = 0.0004), hierarchical lasso logistic regression standardised OR: 0.604 (95% CI 0.259, 0.701), lambda = 0.0049. Hierarchical lasso kept the interaction between the treatment and serum potassium, sodium level, minimum temperature, platelet count and presence of ischemic heart disease. The causal forest approach found some evidence of HTE (p = 0.124). When extracting root splits, the modal split was on serum potassium (mean applied threshold of 4.68 mmol/L). When dividing the patient population into subgroups based on the mean initial root threshold, risk differences in being alive at 28 days were 0.069 (95%CI [-0.032, 0.169]) and - 0.257 (95%CI [-0.368, -0.146]) with serum potassium ≤ 4.68 and > 4.68 respectively.</p><p><strong>Conclusions: </strong>The causal forest agreed with the data-adaptive and classical method of subgroup analysis in identifying HTE by serum potassium. Whilst classical and data-adaptive methods may identify sources of HTE, they do not immediately suggest subgroup splits which are clinically actionable. The extraction of root splits in causal forests is a novel approach to obtaining data-derived subgroups, to be further investigated.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"50"},"PeriodicalIF":3.9,"publicationDate":"2025-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11846376/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143476241","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}