Optimal Augmented GEE for Estimation of Marginal Treatment Effects in Randomized Trials with Multivariate Outcomes

• Activity Number: 656
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Alisa Jane Stephens*+ and Eric Tchetgen Tchetgen and Victor De Gruttola
• Companies: Harvard School of Public Health and Harvard School of Public Health and Harvard School of Public Health
• Keywords: semiparametric theory ; clustered data ; randomized trials ; longitudinal data

Abstract

Semiparametric estimators developed to accommodate incomplete data have been adapted to increase efficiency of inferences in randomized trials through incorporation of baseline covariates. Optimal estimators of marginal treatment effects have been established for univariate outcomes. We derive semiparametric locally efficient estimators under a restricted mean model for marginal treatment effects when outcomes are multivariate; such outcomes arise in clinical trials with clustered or repeated-measures data. The resulting estimating equations alter standard GEE by identifying the efficient score under a restricted mean model for marginal effects when data contain baseline covariates, and include augmentation for enhanced efficiency. The optimal augmented GEE derived are evaluated through simulation and application to the EXPLORE study, a randomized behavioral HIV intervention trial in which subjects were followed over time.