|
Abstract:
|
Acknowledging competing risks nature of time-to-event data when failure can occur from different causes and analyzing it as such is becoming increasingly popular, especially now that most statistical software packages offer standard procedures to carry out competing risks analyses. Until recently the focus has been on continuous time-to-event, but in a 2018 publication Lee, Feuer, and Fine described a method to analyze competing risks data on discrete time scale. However, model evaluation tools such as concordance index for competing risks in discrete time are lacking.
In this presentation we provide an estimator of the concordance index for competing risks in discrete time, prove that this estimator is unbiased, assess its performance in simulations, and illustrate it using the data on diabetes onset and death from the Offspring and Omni cohorts of the Framingham Heart Study.
|
