A Marginal Approach for Multivariate Survival with Longitudinal Covariates

• Activity Number: 169
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Yi-Kuan Tseng*+ and Ya-Fang Yang
• Companies: National Central University and National Central University
• Keywords: Joint model ; MCEM ; mixed model ; misspecification ; Cancer vaccine

Abstract

In clinical trials and other medical studies, it has become increasingly common to observe multiple event times of interest and longitudinal covariates simultaneously. In the literature, joint modeling approaches have been employed to analyze both survival and longitudinal processes and to investigate their association. Early attention has mostly been placed on developing adaptive and flexible longitudinal processes based on a prespecified univariate survival model, most commonly chosen as the Cox proportional model. We propose a marginal likelihood approach to handle multivariate survival time in joint model framework which implements the similar idea of marginal methods used in literature by ignoring the dependency among event times. The marginal likelihood could be easily incorporated various survival model in the likelihood function including two popular survival models, Cox and AFT models, or others such as extended hazard model. The maximization of the marginal likelihood is conducted through Monte Carlo EM and the standard error estimates are obtained via bootstrap method. The performance of the procedure is demonstrates through simulation study and case study.