Statistical models for recurrent events and death: Application to cancer events

In many biomedical studies, subjects may experience the outcome of interest more than once over a period of observation; outcomes of this sort have been termed recurrent events. A model that is becoming increasingly popular for modeling association between recurrent survival times is the use of a frailty model. In recent years a number of papers appeared, extending the survival models to models that are suitable to handle more complex survival data as recurrent events. We present here frailty model extensions to analyze recurrent events data: cure frailty models for a mixture of susceptible and insusceptible subjects for the event of interest; nested frailty models when data are clustered at several hierarchical levels and joint frailty models for the joint analysis of recurrent events and death. We performed a semi-parametric penalized likelihood approach to estimate the different parameters. Those different models can be fitted using the R package “frailtypack”.