ABSTRACT
In studies where the end point is the death of a patient, individuals may face death from a number of different risks. These then compete with one another to be the actual cause of death, giving rise to competing risks. Death from any one cause precludes death from other causes, and this causes difficulties in the estimation of a survivor function. In this chapter, it is shown that the usual Kaplan-Meier estimate of the survivor function is generally inappropriate, and alternative methods for summarising competing risks data are described. The cumulative incidence function is highlighted, and a method for estimating this function is presented. A standard model for cause-specific hazard functions is summarised, together with a discussion of its relative merits. The Fine and Gray model for cause-specific incidence is then presented, and its interpretation in terms of the impact of explanatory variables on cause-specific cumulative incidence illustrated.
