Design and Analysis Methods for Randomized Clinical Trials with Clustered Survival Endpoints

• Activity Number: 136 - Recent Advances in Clustered Time-to-Event Data
• Type: Topic Contributed
• Date/Time: Monday, August 8, 2022 : 10:30 AM to 12:20 PM
• Sponsor: Lifetime Data Science Section
• Author(s): Sin-Ho Jung* and Jianghao Li
• Companies: Duke University School of Medicine and Eli Lilly and Company
• Keywords: cluster randomization; subunit randomization; intra-cluster correlation ; sample size calculation

Abstract

In clustered data, outcome data are collected from multiple subunits consisting of each cluster. Since subunits in each cluster share the common frailties of the cluster, the outcome data from subunits of each cluster tend to be correlated while those from subunits in different clusters are independent. In a clinical trial comparing two treatment arms, we may randomize either clusters or subunits between arms. Dependency of outcome data exists only within each arm in the former case, while it exists both within each arm and between arms. In this talk, we will discuss analysis methods and their sample calculation methods for such trials with a survival endpoint.