Hazards regression for length-biased and right-censored data

When survival data are collected on a cohort of prevalent cases with constant incidence, it is well known that the survival function corresponding to these data is length-biased. For length-biased data, [Wang, M.-C., 1996. Hazard regression analysis with length-biased data. Biometrika 83, 343–354] developed semiparametric estimation procedures for the proportional hazards model. In this article, for length-biased and right-censored data, we construct semiparametric estimation procedures for the proportional and additive hazards models. Simulation results show that semiparametric estimators can outperform nonparametric estimators when censoring is not heavy and truncation is severe.