Recurrent Events Data Sample Clauses

Recurrent Events Data. For recurrent events data without covariate measurement error, a number of regression models have been proposed in the literature as an extension of the Xxx proportional hazards model. Xxxxxxxx & Xxxx (1982) introduced the counting process model with a Xxx-type intensity function for recurrent events and established an elegant large sample asymptotic theory through martingale theory. Wei et al. (1989) proposed a marginal approach for the analysis of recurrent failures. Later, Pepe & Cai (1993) and Xxxxxxx et al. (1997) proposed multiplicative models for the rate and mean functions of arbitrary counting processes. More recently, Lin et al. (1998) proposed to extend the univariate accelerated failure time model for the analysis of recurrent events. They developed a class of consistent and asymptotic normal rank estimators. Comparing to the Xxx model, the accelerated failure time model has an appealing feature of a direct physical interpretation. In the presence of covariate measurement error, Jiang et al. (1999) developed a method using replicated measurements of error-prone covariates and applied their method to the NPC trial data. Their method is based on a discrete-time proportional means model (see Xxxxxxx & Xxxxxx, 1995). Parametric assumptions are imposed on both the true covariates and the errors. Later, Hu & Lin (2004) considered an extended proportional hazards model. Though no parametric assumption is imposed on the covariates, their approach requires the errors to be symmetrically distributed. To our knowledge, there is little development under the accelerated failure time model framework for recurrent events data with covariate measurement error.