Explained Variation in Correlated Survival Data

• Activity Number: 117
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Ronghui Xu* and Gordon Honerkamp-Smith
• Companies: UC San Diego and UC San Diego
• Keywords: proportional hazards mixed-effects model ; genetic epidemiology ; multi-center clinical trials ; recurrent events ; residual sum of squares ; likelihood ratio

Abstract

Explained variation in survival data has attracted much attention in recent years. In this talk we consider explained variation in correlated survival data, specifically under the proportional hazards mixed-effects modeling (PHMM) of such data, which provides a natural setting for the decomposition of different sources of variation. A motivation of the concept originated from genetic epidemiology. More generally, explained variation can be formulated in different ways, and we discuss the formulations that are often encountered in the literature and that are interpretable in practice. We study the proposed measures both in theory and through simulation, and discuss some common pitfalls in using and understanding such measures in practice. To conclude, we show some interesting applications to both multi-center clinical trials and recurrent events.