A behavioural Bayes approach for sample size determination in cluster randomized clinical trials

Abstract

Cluster randomized clinical trials are increasingly popular to evaluate disease control interventions for communities. In these trials health interventions are allocated randomly to complete clusters or groups rather than to individual subjects. Sample size calculation for cluster randomized clinical trials has been largely based on classical theory, taking account of between-cluster variation, and of type I and II errors. It is desirable to use an approach which maximizes the expected net benefit, but there is as yet no established methodology along these lines. Gittins and Pezeshk presented an expected net benefit approach to sample size determination. We extend that approach to cluster randomized clinical trials. © 2010 Royal Statistical Society.