An Improved Analytic Method for 2x2 Crossover Trials with Potential Missing Data

• Activity Number: 625
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Yu Ding*+ and Yang Liu and Devan V. Mehrotra and John Palcza
• Companies: Merck Research Laboratories and Merck Research Laboratories and Merck Research Laboratories and Merck Research Laboratories
• Keywords: 2x2 Crossover ; superiority ; compound symmetry ; degrees of freedom ; missing data ; sequence effect

Abstract

The 2x2 crossover design is often used to compare within-subject responses of a test (T) to a reference (R) treatment. The resulting data are commonly analyzed using a linear mixed effects model with fixed effect terms for treatment, period and sequence, and random effect terms for subject within sequence and residual error. With no missing values, the standard method generally delivers acceptable results. However, when some values are missing, the performance of the standard analysis can sometimes be adversely affected. In this talk, we (i) explain when and why the standard method breaks down, and quantify the detrimental impact on type I error rate, power, and parameter estimates using simulations, and (ii) propose a simple alternate method of analysis with desirable properties. Our results are applicable for both superiority trials and equivalence trials.