The loss in efficiency from misspecifying covariates in proportional hazards regression models

Abstract

This note considers partial likelihood score tests for association between a covariate process and survival time based on a proportional hazards regression model. We derive the asymptotic efficiency of the test based on a misspecified form of the covariate process relative to the test based on the correct form. The results can be used to assess the consequences of various types of misspecification.