UC Riverside

Motivated by the United States Renal Data System (USRDS), we propose a joint modeling framework for longitudinal and survival outcomes that accounts for time-dynamic associations. In this population of patients, two outcomes are of interest, hospitalization, a longitudinal binary outcome, which is a major source of death risk, and mortality, which is higher in this population than in other comparable populations, including Medicare patients with cancer. Therefore, it is of interest to identify the patient-and dialysis facility-level risk factors that jointly affect these outcomes. Furthermore, studies have shown the effect of risk factors changes as a patient undergoes dialysis; therefore, it is necessary to model the associations as a function of time. Additionally, we incorporate multilevel random effects and multilevel covariates, at both the patient and facility levels, to account for the hierarchical data structure. An approximate Expectation-Maximization algorithm is developed for estimation and inference, where the fully exponential Laplace approximation is employed to address the hierarchical structure, and spline models are utilized to incorporate a time-dynamic association. We demonstrate the finite sample performance of our approach via simulation studies. We apply our proposed model to USRDS data to identify significant time-varying associations.