Relative Power and Robustness of Mixed Model, GEE, and MMRM Analysis of Alzheimer's Disease Treatment Trials

• Activity Number: 129
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Steven D. Edland*+
• Companies: University of California, San Diego
• Keywords: Clinical Trial ; Power ; Random Effects Model ; Mixed Effects Model ; GEE ; MMRM

Abstract

The group by time term in a mixed effects regression analysis with random slopes and intercepts can be used to test for differences in mean rate of progression of quantitative trait outcome measures in clinical trials comparing treatment versus control. Alternative analyses include covariance structure robust marginal models with estimation by generalized estimating equations (GEE) and model robust mixed model repeated measures (MMRM) analysis of change scores. We use derivations and Monte Carlo simulations informed by extensive placebo arm data from completed trials performed by the Alzheimer's Disease Cooperative Study to characterize the relative power and bias of these methods under a range of plausible models and study subject dropout rates.