Year-end Sale % OFF 
Abstract
Meta‐analysis of individual patient data (IPD) is increasingly used to synthesize data from multiple trials. IPD meta‐analysis offers several advantages over meta‐analyzing aggregate data, including the capacity to individualize treatment recommendations. Trials usually collect information on many patient characteristics. Some of these covariates may strongly interact with treatment (and thus be associated with treatment effect modification) while others may have little effect. It is currently unclear whether a systematic approach to the selection of treatment‐covariate interactions in an IPD meta‐analysis can lead to better estimates of patient‐specific treatment effects. We aimed to answer this question by comparing in simulations the standard approach to IPD meta‐analysis (no variable selection, all treatment‐covariate interactions included in the model) with six alternative methods: stepwise regression, and five regression methods that perform shrinkage on treatment‐covariate interactions, that is, least absolute shrinkage and selection operator (LASSO), ridge, adaptive LASSO, Bayesian LASSO, and stochastic search variable selection. Exploring a range of scenarios, we found that shrinkage methods performed well for both continuous and dichotomous outcomes, for a variety of settings. In most scenarios, these methods gave lower mean squared error of the patient‐specific treatment effect as compared with the standard approach and stepwise regression. We illustrate the application of these methods in two datasets from cardiology and psychiatry. We recommend that future IPD meta‐analysis that aim to estimate patient‐specific treatment effects using multiple effect modifiers should use shrinkage methods, whereas stepwise regression should be avoided.
Highlights
Individual patient data (IPD) meta-analysis of randomized clinical trials (RCTs) is often considered to be the best approach in evidence synthesis, despite being more resource-intensive than the standard meta-analysis based on aggregate data.[1]
We report the mean squared error (MSE) for γm averaged across covariates that were effect modifiers (“true” effect modifiers), and we report the MSE averaged across covariates that were not effect modifiers (“false” effect modifiers)
Besides generalized linear mixed effects model (GLMM)-oracle, we found that all shrinkage methods, that is, least absolute shrinkage and selection operator (LASSO), ridge, adaptive LASSO, Bayesian LASSO, and stochastic search variable selection (SSVS) performed better than models that do not use shrinkage, that is, GLMM-full and STEP
Summary
Individual patient data (IPD) meta-analysis of randomized clinical trials (RCTs) is often considered to be the best approach in evidence synthesis, despite being more resource-intensive than the standard meta-analysis based on aggregate data.[1]. Standard meta-analysis can be used to explore treatment-covariate interaction (ie, by performing a “meta-regression” on study-level characteristics, eg, length of trial or mean age of participants), such analyses are hindered by the usually small number of studies in the analysis, by the typically small variation in study-level characteristics and the threat of aggregation bias.[2] By contrast, in an IPD meta-analysis we model the individual outcome across many patients, with a usually much wider range in values of covariates. IPD meta-analysis is less prone to aggregation bias as within trial information can be directly used to estimate how patient-level characteristics modify treatment effects.[1,3,4]. The one-stage approach simultaneously models the individual participant data from all studies while respecting the randomization by accounting for the clustering of patients within each trial. One-stage approaches may offer generally greater flexibility in modeling,[5] both approaches may lead to similar results in many cases.[6]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
