Competing Risks Models

Abstract

In the previous chapters we considered various statistical techniques that model the time to a particular event of interest. In this chapter, we consider competing risks models, which extend the previously described models to survival data with several distinct types of target events. Competing risks models are often used in survival analysis to model several causes of death. Similarly, competing risks models are useful to model the duration of unemployment, where one often wants to distinguish between full-time and part-time jobs that end the unemployment spell. We will first consider parametric competing risks models for discrete time-to-event data (Sect. 8.1). These include the popular multinomial model and the cumulative model for ordered responses, which can both be embedded into the binary-response framework. Parameter estimation and variable selection methods for discrete-time competing risks models are described in Sects. 8.2 and 8.3, respectively.