|
Abstract:
|
We propose a semiparametric random effects model for data in the clustered competing risks setting. Specifically, we propose direct modeling of cluster and covariate effects on the cumulative incidence functions of each risk through semiparametric additive regression models containing cluster-specific random effects. A unique feature of our approach is that we model the dependency of failure times both within and across causes among individuals within a cluster by allowing for the correlation of cluster-specific random effects across causes. Further, by decomposing the cause-specific cumulative incidence functions using a mixture model representation, we are able to estimate model parameters associated with all competing risks under consideration, satisfying the constraint that the sum of cumulative incidence functions does not exceed one. We develop estimating equations for parameter estimation and test our estimation procedure via simulations. Using data from the Scientific Registry of Transplant Recipients, we apply our method to evaluate the performance of Organ Procurement Organizations on two competing risks: (i) receipt of a kidney transplant and (ii) death on the wait-list.
|
