Similarity of the treatment effects of a drug product among different intrinsic and extrinsic ethnic, geographic, or demographic factors is important not only to regulatory agencies in the approval of the drug but also to the public health. Examples include bridging studies in different regions, the extrapolation from the adult population to the pediatric subpopulation, or comparisons between different geographic regions within a global trial. It is an issue of evaluation of the treatment-by-factor interaction. However, assessment of similarity is not to detect the existence of treatment-by-factor interaction but rather to evaluate whether the magnitude of treatment-by-factor interaction is within a clinically allowable margin. As a result, we propose two testing procedures for the noninferiority hypothesis of the treatment-by-factor interaction to assess the similarity of the treatment effects among ethnical or demographic factors. Two numeric data sets illustrate applications of the proposed methods to different scenarios under different circumstances. Copyright ? 2009 Drug Information Association, Inc.
