Proportional hazards models with time-dependent frailties are constructed by glueing together classical proportional hazards models for counting processes, which include the frailty proportional hazards models in Nielsen et al. (1992) and Holt & Prentice (1974) as special cases, A regular efficient estimator for the relative risk coefficient is obtained. The theory is applied to discuss a matched-pair data of remission lengths of leukemia patients, A simulation study is also presented.
