A Simulation Study Comparing Linear Mixed Models and Generalized Estimation Equations for Use in Longitudinal Biomarker Study for Exposure to Cigarette Smoke

• Activity Number: 154
• Type: Contributed
• Date/Time: Monday, July 30, 2007 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Qiwei Liang*+ and Huajiang Li and Hans Roethig
• Companies: Philip Morris USA and Quintiles Inc and Philip Morris USA
• Keywords: Monte Carlo simulation ; linear mixed models ; generalized estimation equations ; biomarkers ; Type I error ; cigarette smoke

Abstract

The objective of this study is to use Monte Carlo simulation to compare linear mixed models and generalized estimation equations for application in clinical studies investigating human exposure to cigarette smoke. Unlike some other simulation studies for model comparisons, we specify the simulation study design, study variables, population covariance structures, study group sizes, study group means, study group variances, series correlations and missing data rates to resemble the real world situations in studies investigating human exposure to cigarette smoke. The resulting Type I error rates for testing the differences in study group means are compared. Overall, generalized estimation equations show superior Type I error control (mostly around 5% versus considerably larger than 5% for linear mixed models) in detecting the differences in study group means.