Centre For Medical Gerontology, School of Medicine, TCD

The Implications of Differential Clustering for the Analysis of Binary Outcomes in Cluster Randomised Trials

Cluster randomised trials usually assume homogeneity of the clustering effect. In trials of interventions involving, for example, professional behaviour change, the post-randomisation clustering within groups might plausibly be influenced by the intervention itself. This may cause differential clustering between intervention arms. We look at some possible examples of this, and examine the robustness of a range of standard estimators for the analysis of binary outcomes in cluster randomised trials in the presence of this differential clustering.