ABSTRACT
Multivariate survival data are abundant in sociological studies and biomedical research. This chapter shows how the random-effects frailty models and marginal survival models can be adapted to accommodate correlated survival times with a possible cure fraction and recurrent survival data with a cure fraction. It discusses competing risks data with a cure fraction. The chapter is concluded with a demonstration of the applications of some of the cure models to real data set. A linear transformation model can also be considered as the latency model in the marginal mixture cure model. Unlike marginal models, random-effects models account for the correlation within clusters explicitly by using shared random effects.
