CLUSTER VERSUS INDIVIDUAL RANDOMIZATION IN ADOLESCENT TOBACCO AND ALCOHOL STUDIES - ILLUSTRATIONS FOR DESIGN DECISIONS

Background. The decision to randomize by clusters of subjects such as a classroom or clinic versus individual randomization where some conta mination may occur is examined within the framework of sample size iss ues, Estimates for background rates and intraclass correlations are al so provided for adolescent tobacco and alcohol outcomes derived from a recent study using cluster randomization. Methods. A ratio of adjuste d sample sizes is derived which is a function of the intraclass correl ation and cluster size for cluster randomization and total amount of c ontamination for individual randomization, Using estimated incidence r ates and intraclass correlations, we provide a comparison of sample si zes for two plausible study outcomes. Results. Small clusters such as a family or small classroom tend to have stronger within cluster depen dence and cluster randomization would be clearly favoured over individ ual randomization. For moderately sized clusters, if contamination lev els are likely to be high then cluster randomization would be a better choice. However in some situations where lower levels of contaminatio n are expected, individual randomization may be preferred. With larger clusters, individual randomization should be considered when contamin ation rates are expected to be low. Conclusions. Investigators must ca refully consider the choice of cluster randomization versus individual randomization in the context of likely contamination. In this paper w e provided a basis for making this decision as well as examples to ill ustrate these decisions, and parameter estimates that will be especial ly useful for investigators in adolescent tobacco and alcohol studies.