Inference in Cox Proportional Hazards Model with Covariates Missing in Nonmonotone Patterns

• Activity Number: 27
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: SSC
• Author(s): Yang Zhao*+ and Wei Tang
• Companies: University of Regina and University of Regina
• Keywords: Auxiliary information ; Cox proportional hazards model ; Missing covariates ; Nonmonotone missing data pattern ; Surrogate models ; Unified approach

Abstract

Missing covariates often occur in regression analysis. In this research we propose a method for estimating in Cox proportional hazards models with covariates missing in nonmonotone patterns. We assume that the data are missing at random and the missing data probabilities are known or can be parametrically estimated. We described an idea of using surrogate models to extract partial information from incomplete observations and auxiliary variables to compute an efficient estimator of the parameter in the model. The method can be applied to deal with nonmonotone missing data patterns directly. It is computational simple compared to many methods proposed in the literature. We show the asymptotic properties of the estimator. We use simulation studies to examine the performance of the proposed method.