Regression Extension of the Win Ratio Approach for Composite Endpoints with Clinical Priorities

• Activity Number: 489
• Type: Topic Contributed
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: International Chinese Statistical Association
• Author(s): Xiaodong Luo*+ and Ken Cheung and Wei Yann Tsai
• Companies: Mount Sinai School of Medicine and Columbia University and Columbia University
• Keywords: win ratio ; composite endpoints ; proportional hazards model ; regression extension

Abstract

In two-sample case, Pocock et al. (2012) proposed a win ratio approach to analyze composite endpoints consisting of both fatal and non-fatal outcomes with different clinical priorities, and Luo et al. (2014) derived the null hypothesis and proposed a closed-form variance estimator for the win ratio statistic. In this talk, we propose a proportional-hazards-type regression model to extend the win ratio approach. We will also propose and compare three estimation methods for such model: Gehan, Hybrid and Log-rank, and found that the Log-rank method will provide best precision through empirical and theoretical studies.