Issues in the meta‐analysis of cluster randomized trials

Abstract

Meta-analyses involving the synthesis of evidence from cluster randomization trials are being increasingly reported. These analyses raise challenging methodologic issues beyond those raised by meta-analyses which include only individually randomized trials. In this paper we review and comment on a selected number of these issues, including problems of study heterogeneity, difficulties in estimating design effects from individual trials and the choice of statistical methods.