Case-cohort design is widely used in biomedical studies of rare diseases as an efficient way to reduce cost. Relevant covariate histories, which are costly or difficult to obtain, are observed only on cases and a random subcohort in such studies. It is often that a lag period exists before the treatment or other covariates is fully effective. This phenomenon may be described well by an accelerated hazards model. Existing methods for the accelerated hazards model do not handle case-cohort data. This paper proposes a semiparametric inference method for the accelerated hazards model with data from a case-cohort design. The proposed estimators are shown to be consistent and asymptotically normally distributed. The finite sample properties of proposed case-cohort estimator and its relative efficiency to full cohort estimator are assessed via simulation studies. An application to a burn study demonstrates the utility of the proposed method in practice.
展开▼
