Doubly Robust Estimators of Treatment-Specific Survival Distributions in Observational Studies with Stratified Sampling

• Activity Number: 424
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Xiaofei Bai*+ and Anastasios (Butch) Tsiatis and Sean M. O'Brien
• Companies: North Carolina State University and North Carolina State University and Duke University Medical Center
• Keywords: Cox proportional hazard model ; Double robustness ; Observational study ; Stratified sampling ; Survival analysis

Abstract

Observational studies are frequently conducted to compare the effects of two treatments on survival. For such studies we must be concerned about confounding; that is, there are covariates that affect both the treatment assignment and the survival distribution. With confounding the usual treatment-specific Kaplan-Meier estimator might be a biased estimator of the underlying treatment-specific survival distribution. This paper has two aims. In the first aim we use semiparametric theory to derive a doubly robust estimator of the treatment-specific survival distribution in cases where it is believed that all the potential confounders are captured. In cases where not all potential confounders have been captured one may conduct a substudy using a stratified sampling scheme to capture additional covariates that may account for confounding. The second aim is to derive a doubly-robust estimator for the treatment-specific survival distributions and its variance estimator with such a stratified sampling scheme. Simulation studies are conducted to show consistency and double robustness. These estimators are then applied to the data from the ASCERT study that motivated this research.