A Causal Effect Model with Stochastic Monotonicity Assumption for Clinical Trials with Incomplete Longitudinal Outcome

• Activity Number: 324
• Type: Topic Contributed
• Date/Time: Tuesday, July 31, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Chenguang Wang*+ and Michael Daniels and Daniel Scharfstein
• Companies: and University of Florida and The Johns Hopkins University
• Keywords: causal inference ; survivor average causal effect ; missing data ; stochastic assumption ; clinical trial ; longitudinal study

Abstract

In randomized clinical trials, the causal effect of treatment is generally not identifiable when outcome data are unobserved because of early study exit caused by protocol defined events (e.g. death). Since the unobserved responses after early study exit are often undefined, standard missing data analysis methods which impute these responses are not applicable. We propose a Bayesian approach to estimate the causal treatment effect among those who would not have had protocol defined events regardless of treatment assignment. In this approach, we consider longitudinal stochastic assumptions that are weaker than commonly applied deterministic assumptions. For inference, we compute the upper and lower bounds of the causal effects and characterize the uncertainty associated with the estimated intervals using Bayesian methods.