Multivariate meta-analysis using individual participant data

摘要: When combining results across related studies, a multivariate meta-analysis allows the joint synthesis of correlated effect estimates from multiple outcomes. Joint can improve efficiency over separate univariate syntheses, may reduce selective outcome reporting biases, and enables inferences A common issue is that within-study correlations needed to fit model are unknown published reports. However, provision individual participant data (IPD) them be calculated directly. Here, we illustrate how use IPD estimate correlations, using linear regression for continuous outcomes bootstrapping methods binary, survival mixed In 10 hypertension trials, then show these enable address novel clinical questions about continuous, binary outcomes; treatment–covariate interactions; adjusted risk/prognostic factor effects; longitudinal data; prognostic multiparameter models; treatment comparisons. Both frequentist Bayesian approaches applied, with example software code provided derive models. © 2014 The Authors. Research Synthesis Methods by John Wiley & Sons, Ltd.