Online Program Home
My Program

Abstract Details

Activity Number: 409 - Survival Analysis I
Type: Contributed
Date/Time: Tuesday, July 31, 2018 : 2:00 PM to 3:50 PM
Sponsor: Biometrics Section
Abstract #329465 Presentation
Title: Choosing the Primary Efficacy Analysis for a Randomized Clinical Trial with Competing Risks
Author(s): Eric Leifer* and James Troendle and Lauren Kunz
Companies: National Heart, Lung, and Blood Institute and National Institutes of Health and National Heart, Lung, and Blood Institute
Keywords: competing risk; cause specific hazard; cumulative incidence; treatment effect

Consider a two-armed treatment vs. control randomized clinical trial with a time to event primary outcome that is subject to a competing risk. A popular approach for analyzing the primary outcome's treatment effect uses the cause specific hazard (CSH) Cox model which considers the competing event as an independent censoring. Another popular approach uses the cumulative incidence (CI) function. We compare the CSH and CI approaches by simulating the primary and competing events' CSHs where there is unobserved shared frailty connecting the two events. We show that the CSH is preferable to the CI function as it is reasonably robust with respect to controlling type I error and treatment effect bias even when the shared frailty is moderately prognostic for the primary and competing events. However, when the frailty is strongly prognostic, the CSH is not reliable and it is preferable to combine the primary and competing events into a composite endpoint, even if there is a loss in power. We compare our results to those of Freidlin and Korn (Statistics in Medicine 2005; 24:1703-1712) who considered a similar problem by simulating from a latent failures model.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program