On efficient estimation in additive hazards regression with current status data

The additive hazard regression (AHR) model is known for its convenience in interpretation, as hazard is modeled as a linear function of covariates. One outstanding issue in the application of such a model in the analysis of current status data is that there lacks an efficient and computationally simple approach for parameter estimation. In the current literature, Lin et al.’s (1998) method enjoys the computational ease but it is not semi-parametrically efficient, whereas Martinussen and Scheike’s (2002) method is semi-parametrically efficient but difficult to compute. In this paper, we propose a new estimation approach in the context of Lin et al.’s AHR models where the monitor time process follows a proportional hazard model. We show that not only the proposed estimator achieves semi-parametric information bound, but also its implementation can be done easily using existing statistical software. We evaluate this new method via simulation studies. Also, we illustrate the proposed method through an analysis of renal function recovery data.