Model misspecification in proportional hazards regression

摘要: The proportional hazards model is frequently used to evaluate the effect of treatment on failure time events in randomised clinical trials. Concomitant variables are usually available and may be considered for use primary analyses under assumption that incorporating them reduce bias or improve efficiency. In this paper we consider two approaches including covariate information : regression modelling stratification. We focus setting where effects nonproportional compare bias, efficiency coverage properties these approaches. These results indicate our intuition based linear analysis covariance misleading. Covariate adjustment models has little variance but significantly accuracy estimator.