Adjusting for Baseline Measurements in Crossover Studies

• Activity Number: 625
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Chengcheng Liu*+ and Lingling Han and Xiaodong Li and Peng Sun
• Companies: Merck & Co., Inc. and Merck & Co., Inc. and Merck & Co., Inc. and Merck & Co., Inc.
• Keywords: Baseline Measurements ; Crossover Design ; arbitrary correlation structure ; under-coverage ; underestimation ; misspecification

Abstract

Crossover designs are very commonly used in pharmaceutical studies, especially in the early stages of drug development. Retaining the baseline measurements as part of the response vector in the model is recommended because of more power and more flexibility when handling missing values. The application of this method in crossover settings with an arbitrary correlation structure is investigated and compared with three other commonly used methods. Some issues, such as underestimation of the standard error of mean treatment change from baseline when baseline is used as a covariate, impact of misspecifications of period effect on inference, and under-coverage problem with small samples are highlighted. Simulation studies are conducted to illustrate our findings, and applications to the real data are discussed in support of our recommendations.