A regression analysis of discrete time competing risks data using a vertical model approach

Abstract

Over the years, the standard regression analysis method for discrete time competing risks data has been to model the data with discrete time cause-specific hazards. While a few continuous time competing risks models have been proposed in the literature, it is a well documented fact that these models are not appropriate for application in discrete time as is. The vertical regression model of Nicolaie et al. (2010) is the latest of these continuous time competing risks models. We reformulate this regression model for the purpose of application in discrete time. We demonstrate that the proposed model can easily be implemented by using existing software for discrete time models. We apply the proposed model together with some of the existing discrete time models to real discrete time competing risks data and find that the proposed model and these models compare favourably.