Treatment effect heterogeneity in cluster randomized trials

Abstract

Cluster randomized trials (CRTs) refer to a popular class of experiments in which randomization is carried out at the group level. In recent years, stepped wedge CRTs, which involve unidirectional crossover to treatment for all clusters, have increased in popularity. Treatment effect heterogeneity refers to differential treatments effect between subpopulations. When such heterogeneity exists, treatment may be beneficial in certain groups and harmful in others. Existing design and analysis methods for stepped wedge and parallel CRTs generally assume the aim of the trial is to detect an overall treatment effect, averaged over the study population, which may cause incorrect inferences and underpowered studies. The dissertation addresses this by proposing a novel model formulation, randomization tests, and design methods for stepped wedge CRTs (Chapter 1), deriving sample size procedures for testing treatment effect heterogeneity in two-level CRTs under a generalized linear mixed model (Chapter 2), and developing a randomization-based inference procedure for detecting treatment effect heterogeneity in CRTs (Chapter 3).