A Markov Compliance Class and Outcome Model for Causal Analysis in the Longitudinal Setting

• Activity Number: 518
• Type: Contributed
• Date/Time: Wednesday, August 5, 2009 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Xin Gao*+ and Michael Elliott
• Companies: University of Michigan and University of Michigan
• Keywords: causal inference ; noncompliance ; potential outcome ; principal stratification ; latent class model ; randomized trial

Abstract

We propose a Markov compliance class and outcome model for analyzing longitudinal randomized studies when non-compliance is present. We solve the problem in the potential outcome framework, and provide causal estimates on the treatment effect via principal stratification. Previous research (Lin, Ten Have, and Elliott, JASA 2007) considered the effect of subjects' joint compliance behavior on the joint distribution of the longitudinal outcomes, but not the effect of outcomes at time t-1 on the compliance behaviors at time t. The proposed model provides estimates both on the effect of the adherence on the following outcome, and on the effect of the outcome on the following adherence. The proposed model also relaxes the independence assumption between the potential outcomes at the same time. We analyze the Suicide CBT Study using Markov chain Monte Carlo (MCMC) methodology.