Estimating individualized treatment effects using individual participant data meta-analysis

Different approaches can be used to estimate individualized treatment effects with Individual Participant Data Meta-Analyses (IPD-MA). We compared four one-stage models: random effects (RE), stratified intercept (SI), rank-1 (R1) and fully stratified (FS) models, built with two different strategies constructed with a Monte Carlo simulation study in which we explored different scenarios with a binary or a time-to-event outcome. To evaluate the performance of the models, we used the c-statistic for benefit, the calibration of predictions and the mean squared error. The different models were also used on the INDANA IPD-MA, comparing an anti-hypertensive treatment to no treatment or placebo (N = 40 237, 836 events). Simulation results showed that the random effects and the stratified intercept models performed well for both binary and time-to-event outcomes. For the INDANA dataset with a binary outcome, the random effects model had the best performance