Cluster Randomized Trials: Sample Size
In a cluster randomization trial, clusters consisting of multiple subjects are randomized to two or more treatment arms, and observations are collected from the subjects of each cluster while receiving the same intervention. In such a trial, inferences are usually made at subject level while randomization is done at cluster level. Since the subjects in each cluster share some common frailties, their responses are usually positively correlated, both the sample size calculation and the subsequent statistical analysis must take into account the correlation. In this chapter, we investigate some estimators of binary response rate and sample size method for comparing the response rates between two arms using these estimators when the number of subjects is random among clusters.
