Dataset Information

Pooling individual participant data from randomized controlled trials: Exploring potential loss of information.

ABSTRACT: BACKGROUND:Pooling individual participant data to enable pooled analyses is often complicated by diversity in variables across available datasets. Therefore, recoding original variables is often necessary to build a pooled dataset. We aimed to quantify how much information is lost in this process and to what extent this jeopardizes validity of analyses results. METHODS:Data were derived from a platform that was developed to pool data from three randomized controlled trials on the effect of treatment of cardiovascular risk factors on cognitive decline or dementia. We quantified loss of information using the R-squared of linear regression models with pooled variables as a function of their original variable(s). In case the R-squared was below 0.8, we additionally explored the potential impact of loss of information for future analyses. We did this second step by comparing whether the Beta coefficient of the predictor differed more than 10% when adding original or recoded variables as a confounder in a linear regression model. In a simulation we randomly sampled numbers, recoded those < = 1000 to 0 and those >1000 to 1 and varied the range of the continuous variable, the ratio of recoded zeroes to recoded ones, or both, and again extracted the R-squared from linear models to quantify information loss. RESULTS:The R-squared was below 0.8 for 8 out of 91 recoded variables. In 4 cases this had a substantial impact on the regression models, particularly when a continuous variable was recoded into a discrete variable. Our simulation showed that the least information is lost when the ratio of recoded zeroes to ones is 1:1. CONCLUSIONS:Large, pooled datasets provide great opportunities, justifying the efforts for data harmonization. Still, caution is warranted when using recoded variables which variance is explained limitedly by their original variables as this may jeopardize the validity of study results.

SUBMITTER: van Wanrooij LL

PROVIDER: S-EPMC7217432 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Pooling individual participant data from randomized controlled trials: Exploring potential loss of information.

van Wanrooij Lennard L LL Hoevenaar-Blom Marieke P MP Coley Nicola N Ngandu Tiia T Meiller Yannick Y Guillemont Juliette J Rosenberg Anna A Beishuizen Cathrien R L CRL Moll van Charante Eric P EP Soininen Hilkka H Brayne Carol C Andrieu Sandrine S Kivipelto Miia M Richard Edo E

PloS one 20200512 5

<h4>Background</h4>Pooling individual participant data to enable pooled analyses is often complicated by diversity in variables across available datasets. Therefore, recoding original variables is often necessary to build a pooled dataset. We aimed to quantify how much information is lost in this process and to what extent this jeopardizes validity of analyses results.<h4>Methods</h4>Data were derived from a platform that was developed to pool data from three randomized controlled trials on the ...[more]

PMID: 32396543

Similar Datasets

Project description:OBJECTIVE: To assess the effect of the FTO genotype on weight loss after dietary, physical activity, or drug based interventions in randomised controlled trials. DESIGN: Systematic review and random effects meta-analysis of individual participant data from randomised controlled trials. DATA SOURCES: Ovid Medline, Scopus, Embase, and Cochrane from inception to November 2015. ELIGIBILITY CRITERIA FOR STUDY SELECTION: Randomised controlled trials in overweight or obese adults reporting reduction in body mass index, body weight, or waist circumference by FTO genotype (rs9939609 or a proxy) after dietary, physical activity, or drug based interventions. Gene by treatment interaction models were fitted to individual participant data from all studies included in this review, using allele dose coding for genetic effects and a common set of covariates. Study level interactions were combined using random effect models. Metaregression and subgroup analysis were used to assess sources of study heterogeneity. RESULTS: We identified eight eligible randomised controlled trials for the systematic review and meta-analysis (n=9563). Overall, differential changes in body mass index, body weight, and waist circumference in response to weight loss intervention were not significantly different between FTO genotypes. Sensitivity analyses indicated that differential changes in body mass index, body weight, and waist circumference by FTO genotype did not differ by intervention type, intervention length, ethnicity, sample size, sex, and baseline body mass index and age category. CONCLUSIONS: We have observed that carriage of the FTO minor allele was not associated with differential change in adiposity after weight loss interventions. These findings show that individuals carrying the minor allele respond equally well to dietary, physical activity, or drug based weight loss interventions and thus genetic predisposition to obesity associated with the FTO minor allele can be at least partly counteracted through such interventions. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42015015969.

Project description:BackgroundIncreasingly in network meta-analysis (NMA), there is a need to incorporate non-randomised evidence to estimate relative treatment effects, and in particular in cases with limited randomised evidence, sometimes resulting in disconnected networks of treatments. When combining different sources of data, complex NMA methods are required to address issues associated with participant selection bias, incorporating single-arm trials (SATs), and synthesising a mixture of individual participant data (IPD) and aggregate data (AD). We develop NMA methods which synthesise data from SATs and randomised controlled trials (RCTs), using a mixture of IPD and AD, for a dichotomous outcome.MethodsWe propose methods under both contrast-based (CB) and arm-based (AB) parametrisations, and extend the methods to allow for both within- and across-trial adjustments for covariate effects. To illustrate the methods, we use an applied example investigating the effectiveness of biologic disease-modifying anti-rheumatic drugs for rheumatoid arthritis (RA). We applied the methods to a dataset obtained from a literature review consisting of 14 RCTs and an artificial dataset consisting of IPD from two SATs and AD from 12 RCTs, where the artificial dataset was created by removing the control arms from the only two trials assessing tocilizumab in the original dataset.ResultsWithout adjustment for covariates, the CB method with independent baseline response parameters (CBunadjInd) underestimated the effectiveness of tocilizumab when applied to the artificial dataset compared to the original dataset, albeit with significant overlap in posterior distributions for treatment effect parameters. The CB method with exchangeable baseline response parameters produced effectiveness estimates in agreement with CBunadjInd, when the predicted baseline response estimates were similar to the observed baseline response. After adjustment for RA duration, there was a reduction in across-trial heterogeneity in baseline response but little change in treatment effect estimates.ConclusionsOur findings suggest incorporating SATs in NMA may be useful in some situations where a treatment is disconnected from a network of comparator treatments, due to a lack of comparative evidence, to estimate relative treatment effects. The reliability of effect estimates based on data from SATs may depend on adjustment for covariate effects, although further research is required to understand this in more detail.

Project description:BackgroundThe clinical benefit of aspirin for the primary prevention of cardiovascular disease (CVD) in diabetes remains uncertain. To evaluate the efficacy and safety of aspirin for the primary prevention of cardiovascular outcomes and all-cause mortality events in people with diabetes, we conducted an updated meta-analysis of published randomised controlled trials (RCTs) and a pooled analysis of individual participant data (IPD) from three trials.MethodsRandomised controlled trials of aspirin compared with placebo (or no treatment) in participants with diabetes with no known CVD were identified from MEDLINE, Embase, Cochrane Library, and manual search of bibliographies to January 2019. Relative risks with 95% confidence intervals were used as the summary measures of associations.ResultsWe included 12 RCTs based on 34,227 participants with a median treatment duration of 5.0 years. Comparing aspirin use with no aspirin, there was a significant reduction in risk of major adverse cardiovascular events (MACE)0.89 (0.83-0.95), with a number needed to treat (NNT)of 95 (95% CI 61 to 208) to prevent one MACE over 5 years average follow-up. Evidence was lacking of heterogeneity and publication bias among contributing trials for MACE. Aspirin use had no effect on other endpoints including all-cause mortality; however, there was a significant reduction in stroke for aspirin dosage ≤ 100 mg/day 0.75 (0.59-0.95). There were no significant effects of aspirin use on major bleeding and other bleeding events, though some of the estimates were imprecise. Pooled IPD from the three trials (2306 participants) showed no significant evidence of an effect of aspirin on any of the outcomes evaluated; however, aspirin reduced the risk of MACE in non-smokers 0.70 (0.51-0.96) with a NNT of 33 (95% CI 20 to 246) to prevent one MACE.ConclusionsAspirin has potential benefits in cardiovascular primary prevention in diabetes. The use of low dose aspirin may need to be individualised and based on each individual's baseline CVD and bleeding risk. Systematic review registration PROSPERO: CRD42019122326.

Project description:BACKGROUND:Frailty is common in clinical practice, but trials rarely report on participant frailty. Consequently, clinicians and guideline-developers assume frailty is largely absent from trials and have questioned the relevance of trial findings to frail people. Therefore, we examined frailty in phase 3/4 industry-sponsored clinical trials of pharmacological interventions for three exemplar conditions: type 2 diabetes mellitus (T2DM), rheumatoid arthritis (RA), and chronic obstructive pulmonary disease (COPD). METHODS:We constructed a 40-item frailty index (FI) in 19 clinical trials (7 T2DM, 8 RA, 4 COPD, mean age 42-65?years) using individual-level participant data. Participants with a FI >?0.24 were considered 'frail'. Baseline disease severity was assessed using HbA1c for T2DM, Disease Activity Score-28 (DAS28) for RA, and % predicted FEV1 for COPD. Using generalised gamma regression, we modelled FI on age, sex, and disease severity. In negative binomial regression, we modelled serious adverse event rates on FI and combined results for each index condition in a random-effects meta-analysis. RESULTS:All trials included frail participants: prevalence 7-21% in T2DM trials, 33-73% in RA trials, and 15-22% in COPD trials. The 99th centile of the FI ranged between 0.35 and 0.45. Female sex was associated with higher FI in all trials. Increased disease severity was associated with higher FI in RA and COPD, but not T2DM. Frailty was associated with age in T2DM and RA trials, but not in COPD. Across all trials, and after adjusting for age, sex, and disease severity, higher FI predicted increased risk of serious adverse events; the pooled incidence rate ratios (per 0.1-point increase in FI scale) were 1.46 (95% CI 1.21-1.75), 1.45 (1.13-1.87), and 1.99 (1.43-2.76) for T2DM, RA, and COPD, respectively. CONCLUSION:The upper limit of frailty in trials is lower than has been described in the general population. However, mild to moderate frailty was common, suggesting trial data may be harnessed to inform disease management in people living with frailty. Participants with higher FI experienced more serious adverse events, suggesting screening for frailty in trial participants would enable identification of those that merit closer monitoring. Frailty is identifiable and prevalent among middle-aged and older participants in phase 3/4 drug trials and has clinically important safety implications.

Project description:ObjectivesWe examined major issues associated with sharing of individual clinical trial data and developed a consensus document on providing access to individual participant data from clinical trials, using a broad interdisciplinary approach.Design and methodsThis was a consensus-building process among the members of a multistakeholder task force, involving a wide range of experts (researchers, patient representatives, methodologists, information technology experts, and representatives from funders, infrastructures and standards development organisations). An independent facilitator supported the process using the nominal group technique. The consensus was reached in a series of three workshops held over 1 year, supported by exchange of documents and teleconferences within focused subgroups when needed. This work was set within the Horizon 2020-funded project CORBEL (Coordinated Research Infrastructures Building Enduring Life-science Services) and coordinated by the European Clinical Research Infrastructure Network. Thus, the focus was on non-commercial trials and the perspective mainly European.OutcomeWe developed principles and practical recommendations on how to share data from clinical trials.ResultsThe task force reached consensus on 10 principles and 50 recommendations, representing the fundamental requirements of any framework used for the sharing of clinical trials data. The document covers the following main areas: making data sharing a reality (eg, cultural change, academic incentives, funding), consent for data sharing, protection of trial participants (eg, de-identification), data standards, rights, types and management of access (eg, data request and access models), data management and repositories, discoverability, and metadata.ConclusionsThe adoption of the recommendations in this document would help to promote and support data sharing and reuse among researchers, adequately inform trial participants and protect their rights, and provide effective and efficient systems for preparing, storing and accessing data. The recommendations now need to be implemented and tested in practice. Further work needs to be done to integrate these proposals with those from other geographical areas and other academic domains.

Dataset Information

Pooling individual participant data from randomized controlled trials: Exploring potential loss of information.

Publications

Pooling individual participant data from randomized controlled trials: Exploring potential loss of information.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets