A Novel Rank Rest for a Time-to-Event Outcome That Incorporates Information on a Surrogate Event

• Activity Number: 405
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Pamela A. Shaw*+ and Michael P. Fay
• Companies: National Institute of Allergy and Infectious Diseases and National Institute of Allergy and Infectious Diseases
• Keywords: Interval censored data ; Rank tests ; Bivariate censoring ; Nonparametric likelihood ; Permutation test ; Progression-free survival

Abstract

We present a new group level test for bivariate time-to-event data, where one event can be considered a surrogate for the other. This type of data arises commonly when an event of interest, such as survival, is infrequently observed and data on a surrogate (time to remission, time to cancer progression, etc.) are instead collected. In many settings, a time-to-first type endpoint may not be adequate or appropriate to summarize patient outcomes. We propose a rank-based test on a clinical severity score which considers information on both a short-term surrogate event and the long-term true event of interest. The proposed test allows for interval-censored outcomes and incorporates information from the bivariate survival distribution. Properties of the test are explored with a simulation study and the method is illustrated with an applied data example.