Balance in cluster randomized trials

This paper explores the role of balancing covariates between treatment grou ps in the design of cluster randomized trials. General expressions are obta ined for two criteria to evaluate designs for parallel group studies with t wo treatments. The first is the variance of the estimated treatment effect and the second is the extent to which the estimated treatment effect is cha nged by adjusting for covariates. It is argued that the second of these is more important for cluster randomized trials. Methods of obtaining balanced designs from covariates which are available at the start of a study are pr oposed. An imbalance measure is used to compare the extent to which designs balance important covariates between the arms of a trial. Several approach es to selecting a well balanced design are possible. A method that randomly selects one member from the class of designs with acceptable bias will all ow randomization inference as well as model-based inference. The methods ar e illustrated with data from a trial of school-based sex education. Copyrig ht (C) 2001 John Wiley & Sons, Ltd.