Model Checking for Subdistribution Hazards Model Under Case-Cohort Design

• Activity Number: 239 - Study Design and Analysis for Complex Survival Data
• Type: Contributed
• Date/Time: Monday, July 29, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Yayun Xu* and Mei-Jie Zhang and Soyoung Kim
• Companies: Medical College of Wisconsin and Medical College of Wisconsin and Medical College of Wisconsin
• Keywords: Case-cohort design; Competing risks data; Proportional subdistribution hazards; Cumulative sums of residuals; Link function

Abstract

The case-cohort study design has been widely implemented to reduce the cost when assembling covariates in large cohort studies. When analyzing competing risks data, the proportional subdistribution hazards model of Fine and Gray (1999) has been extensively used to directly evaluate the effect of covariates on the cumulative incident function for competing risks data. The proportional subdistribution hazards model may fail in three ways: proportional hazards assumption; the linear functional form, and link function. If any of these assumptions do not hold, it may lead to severe bias in estimating the risk effects. It is important to develop a goodness-of-fit test to check assumptions of the proportional subdistribution hazards under case-cohort design. However, there is limited literature on model checking methods for competing risks data under case-cohort design. In this presentation, we propose diagnostics measures for competing risks data under case-cohort design by using cumulative sums of residuals to check the above model assumptions. Simulation studies are conducted to check the performance of the proposed estimators and we apply our proposed methods to real data examples.