Individual participant data meta-analysis of continuous outcomes: A comparison of approaches for specifying and estimating one-stage models

One‐stage individual participant data meta‐analysis models should account for within‐trial clustering, but it is currently debated how to do this. For continuous outcomes modeled using a linear regression framework, two competing approaches are a stratified intercept or a random intercept. The stratified approach involves estimating a separate intercept term for each trial, whereas the random intercept approach assumes that trial intercepts are...