Multiple Comparisons of Repeatedly Measured Response: Issues of Validation Testing in Thorough QT/QTc Clinical Trials

Abstract

In order to validate the results of a thorough QT/QTc clinical trial, ICH E14 recommended that a concurrent positive control treatment be included in the trial. Zhang (2008) recommended that the study results are validated if the positive control establishes assay sensitivity, i.e., has an effect on the mean QT/QTc interval of 5 ms or more. Zhang (2008) and Tsong et al. (2008) discussed the intersection–union test approach and an alternative global average test approach for testing assay sensitivity during the validation process. In this article, we further discuss the multiple comparison issues of the repeatedly measured QT difference between positive control treatment and placebo in the validation test. We describe and discuss several approaches for type I error rate adjustment that are applicable to the situation.

Key Words:

• Assay sensitivity
• Bonferroni adjustment
• Controlling type I error rate
• Hochberg adjustment
• Holm method
• Length theorem
• Multiple comparisons

ACKNOWLEDGMENT

We thank Prof. Bin Cheng of Columbia University for the comments and suggestions. This paper is greatly improved with his input.

This article represents the points of view of the authors. It does not necessarily represent the official position of the U.S. FDA.

Notes

Note: ρ _xy denotes Pearson correlation coefficient between time (hour) X and time (hour) Y.

Note: ρ _xy denotes Pearson correlation coefficient between time (hour) X and time (hour) Y.