A Bayesian Joint Model for DAS28 Scores and Time-to-Dropout Data

• Activity Number: 321
• Type: Topic Contributed
• Date/Time: Tuesday, July 31, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Violeta Hennessey*+
• Companies: Amgen, Inc.
• Keywords: Competing risk ; Joint modeling ; Multi-level/hierarchical modeling ; Selection models

Abstract

We developed a data-driven Bayesian joint model for longitudinal and time-to-event data. Motivation comes from a double-blinded, randomized, three-arm comparative clinical trial for the treatment of rheumatoid arthritis. The efficacy endpoint was disease activity score based on 28 joints (DAS28). DAS28 is a composite score measuring disease activity that includes physician and patient reported assessment of disease activity. In this presentation we consider a joint model for DAS28 and time-to-dropout where time-to-dropout is assumed to be associated with disease activity. We developed a Bayesian multi-level model for the longitudinal component and adopted a competing risk survival analysis methodology for the dropout process. We explore a mixture of latent variables proposed in the literature to link the longitudinal and dropout process. Separate and joint analyses are provided for the motivating rheumatoid arthritis study.