Sample Size Requirement in Clustered Matched-Pair Design for Noninferiority Test

• Activity Number: 187
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Jun-mo Nam*+
• Companies: National Cancer Institute
• Keywords: clustered matched-pair ; non-inferiority ; power ; sample size

Abstract

Consider clustered matched-pair studies for non-inferiority where clusters are independent but units in a cluster are correlated and an inexpensive new method and the expensive standard one are applied to each unit. Proper test statistics for non-inferiority have recently developed by Durkalski etal and Nam and Kwon. In this paper, we investigate power and sample size requirement of a clustered matched pair study for non-inferiority. Power of a test is primarily related to the number of clusters and the effect of a cluster size on power is secondary. A cluster size of more than four has little or no effect in reducing the number of clusters for a given power. We provide a formula for the number of clusters given cluster size and the cluster size given number of clusters for a specific power. An example involving sample size requirement of a clustered matched-pair study is illustrated.