Abstract
Regression models with mixture (random) components are proposed for the statistical analysis of recurrent events when waiting times between successive events are unknown. These models allow adjustment of parameter estimates for unobserved heterogeneity in the population (due for example to missing covariates) or overdispersion resulting from inexact distributional assumptions. The models are illustrated by a study of recurrence rates of superficial bladder cancer in men.

This publication has 8 references indexed in Scilit: