Joint Modeling of Longitudinal Zero Inflated Count and Survival Data

• Activity Number: 608
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Huirong Zhu*+ and Sheng Luo and Stacia M. DeSantis
• Companies: and University of Texas Health Science Center at Houston and University of Texas Health Science Center at Houston
• Keywords: Joint model ; zero-inflated generalized Poisson ; Bayesian inference ; Longitudinal data ; Random effect ; Informative censoring

Abstract

Longitudinal zero-inflated count outcome data arise frequently in substance use research when assessing the effects of behavioral and pharmacological interventions. Often in those studies, the longitudinal repeated measures are censored by a terminal event (e.g., drop out) and the time to the terminal event may depend on the longitudinal outcomes. In this study, we expand the commonly applied joint model of longitudinal and survival data to accommodate the zero-inflated counts and the Cox proportional hazard model with piecewise constant baseline hazard. Additionally, the Zero-inflated generalized Poisson model is applied in the longitudinal sub-model part to capture the possible over-dispersion. Inference is conducted using a Bayesian framework via Markov Chain Monte Carlo simulation implemented in BUGS languages. We demonstrated the more accurate estimates for zero inflated counts and cox regression parameters and the random effect parameters with simulation study. The proposed method is applied to a prospective study of substance use.