The Ratio of Median or Mean Change Scores for Treatment Comparisons

• Activity Number: 423
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Joseph T. Wang*+ and Kelly H. Zou and Martin Ove Carlsson
• Companies: Pfizer Inc. and Pfizer Inc. and Pfizer Inc.
• Keywords: Ratio statistic ; Chang Score ; Effect Size ; Bootstrap ; Normal-Ratio Distribution ; Ratio T-Test

Abstract

When comparing mean change scores of diary endpoints in clinical trials for treating subjects with overactive bladder (OAB) syndrome between two different groups, two-sample tests and the analysis of covariance with adjustments for baseline and additional covariates are frequently used to assess treatment effect. However, the reduction of diary endpoints in the OAB trials is often within a narrow range (e.g., -5 to 0), and a small mean change may not be a convincing treatment effect to physician/patient. When it is of interest to make a superiority claim of one treatment over another, the ratio of the mean or median change scores of two treatments is useful for the claim. The nonparametric bootstrap may be conducted. Alternatively, a normal-ratio distribution of the ratio statistic may be assumed and a parametric ratio t-test conducted. Finally, an extension to the ratio of two correlated random variables is briefly considered. The above methods for analyzing the ratio statistic are illustrated by pooling two identical randomized clinical trials in treating OAB. Monte-Carlo simulations are performed to compare the performances of these methods.