Improving Efficiency of Analysis of Generalized Semiparametric Regression Models for Cumulative Incidence Functions with Missing Covariates

• Activity Number: 402 - Statistical Methods for New Challenges in Lifetime/Complex Data
• Type: Topic Contributed
• Date/Time: Wednesday, August 5, 2020 : 1:00 PM to 2:50 PM
• Sponsor: Lifetime Data Science Section
• Author(s): Yanqing Sun* and Fei Heng and Unkyung Lee and Peter B Gilbert
• Companies: University of North Carolina At Charlotte and University of North Florida and Texas A&M University and University of Washington and Fred Hutchinson Cancer Research Center
• Keywords: augmented inverse probability weighted of complete-cases ; cumulative incidence function; competing risks; generalized case-cohort design; semiparametric models; varying coefficients

Abstract

The generalized semiparametric regression models for cumulative incidence functions (CIFs) are investigated for the competing risks data when covariates are missing by two-phase sampling designs. The inverse probability of weighted complete-case (IPW) approach has been developed under two-phase sampling designs. Although the IPW estimators are consistent, the information of subjects with phase two/missing covariates are not fully utilized, which leads to the loss of efficiency. This paper studies an augmented inverse probability weighted of complete-cases (AIPW) estimation under the generalized semiparametric regression models for CIFs. This approach modifies the IPW estimation equations with additional terms that exploit the key features in the relationship between the missing covariates and the phase-one data to improve efficiency. The asymptotic properties of the proposed estimators are established. The finite sample performances of the proposed estimators are examined in a simulation study. The proposed method is applied to the RV144 vaccine efficacy trial to investigate the associations between the immune response biomarkers and the cumulative incidence of HIV-1 infection.