Philip Westgate, PhD, Department of Biostatistics, University of Kentucky

"Small-Sample Issues in Cluster Randomized Trials"

Cluster randomized trials (CRTs) are increasingly popular in health-related research due to the need or desire to randomize clusters of subjects to different trial arms as opposed to randomizing each subject individually. As outcomes from subjects within the same cluster tend to be more alike than outcomes from subjects within other clusters, an exchangeable correlation arises that is measured via the intra-cluster correlation coefficient (ICC). The ICC is important because it must be appropriately accounted for in both the design of, and analysis of data from, CRTs. When utilizing generalized estimating equations to account for the correlation, well-known standard error estimators can be biased, thus jeopardizing the validity of inference. Furthermore, the traditionally used method of moments to estimate the ICC can result in a negatively biased estimate, which may be detrimental if a future CRT is designed based on this estimate. Therefore, in this talk both an alternative standard error estimator as well as an alternative ICC estimator are introduced. In a simulation study we assess how much of an improvement these estimators provide, and we demonstrate these estimators in application to a real-life CRT.

Stay connected Youtube