The Minimum Required Number of Clusters in Cluster-Randomized Trials

• Activity Number: 124
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Zhiying You*+ and Qing Li
• Companies: and Michigan State University
• Keywords: cluster randomized trials ; sample size ; power ; minimum required number of clusters

Abstract

The study power of two-level cluster randomized trials depends on two sample size components, the number of clusters (unit of level-2) per group and the numbers of individuals within clusters (cluster size; unit of level-1). The number of clusters dominates the power and increasing the cluster size can only increase the power to a certain upper bound for a given number of clusters. This implies there is: (1) an upper bound in the power for a given number of clusters ; (2) a minimum detectable effect size for a given number of cluster , given a spepcified study power; and (3) a minimum required number of clusters for a specified study power. This argumant is applicable to three-level cluster randomized trials where there are three sample size components, the number of level-3 units per group, the number of level-2 units within level-3 units, and the number of level-1 units within level-2 units, and the number of level-3 units dominates the power most. In this study, we derived the formula for the minimum required number of clusters in two-level trails as well as the formulas of the minimum required number of level-3 and level-2 units in three-level trials.