Abstract
In rare disease study, the failure rate is low in a large cohort of subjects due to constraint by funding and research time. Case-cohort design is a widely used sampling scheme for cost-effective study. The weighted pseudo-score (PS) estimators are proposed in the additive hazards model with data generated under case-cohort design. In this article, we present two more effective estimators: the generalized moment estimations with two different weights. Combining the subgroup information with the PS estimator, the generalized methods of moment are applied for the regression parameters. We establish the asymptotic properties of the proposed estimators and conclude that the estimators are more efficient than the weighted PS estimator. We also show the asymptotic properties for the estimated cumulative baseline hazard function. Simulation and real data studies demonstrate the finite-sample performances of the proposed method and the corresponding theories.