Semiparametric Estimation of Cumulative Treatment Effect in the Presence of Dependent Censoring

• Activity Number: 506
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Qi Gong*+ and Douglas E. Schaubel
• Companies: University of Michigan and University of Michigan
• Keywords: cumulative hazard ; stratification ; dependent censoring ; survival analysis ; inverse weighting

Abstract

In time to event data observed in medical studies, nonproportional hazards and dependent censoring are common issues when comparing group-specific mortality. The group effect on mortality may vary over time, as opposed to being constant. One remedy is adopting a parametric form to model the time-dependent pattern. However, it is generally difficult to verify the correctness of the chosen parametric function. Moreover, investigators tend to be more interested in cumulative effects on mortality (e.g., if and when the survival curves cross) rather than the instantaneous effect. Estimators are no longer consistent in the presence of dependent censoring, which may occur when both censoring and death depend on the same time dependent covariates. Therefore, we propose an estimator for the cumulative group effect on survival in the presence of nonproportional hazards and dependent censoring. The proposed estimator is based on the cumulative hazard function, assumed to follow a stratified Cox model. No functional form needs to be assumed for the nonproportionality. Asymptotic properties are derived and evaluated in simulation studies. The proposal method is applied to SRTR data.