Sample Size Consideration in Noninferiority Designs

• Activity Number: 245
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Chengxing Lu*+ and Xing Sun
• Companies: Novartis and Merck
• Keywords: Non-inferiority ; Probability of Success ; Historical Data

Abstract

Non-inferiority trails are intended to show the experimental arm is not inferior to the active control arm in a clinical unacceptable extent. With the nature of such trials, how to design and interpret such studies so that they can support a (valid) conclusion is a formidable challenge (FDA Draft Guidance for Industry Non-Inferiority Clinical Trials). This presentation is aiming to discuss the challenges in deciding the sample size based on the pre-specified non-inferiority margin. A commonly used assumption in addressing this challenge is to assume equality between experimental arm and the active control arm. However, this assumption may or may not be optimal (X Xu, FDA workshop 2011 round table discussion). Examples will be provided in the presentation when trials designed under this assumption are sub-optimal. Suggestions based on calculation of Probability of Success (POS) will be proposed, in cases with absolutely no historical data, and with historical data from one single study, or multiple studies. Simulation results will also be presented for different scenarios.