A cause-specific hazard spatial frailty model for competing risks data

Competing risks data is often grouped into clusters, such as geographic regions. Shared frailty models can be used to model the correlation between individuals within each cluster. For the better fitting of the model in the competing risks data, one random effect can be used for every type of event in each cluster. But the correlation between random effects in each cluster should also be taken into account. On the other hand, there may also be a spatial correlation between random effects during geographic regions. In this paper, we use cause-specific hazard spatial frailty model with multivariate conditional autoregressive distribution for frailties via Bayesian approach. Simulation studies are used to assess the regression coefficient estimators as well as the variance and correlation of random effects within the clusters. We apply the proposed model to the analysis of the gastrointestinal cancer data of Iran's National Institute of Health Research and the Louisiana breast cancer data from the Surveillance Epidemiology and End Results database of the National Cancer Institute.