Two Approaches to Noninferiority Margin Derivation

• Activity Number: 355
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Kaihong Jiang*+ and Xuezhou Mao and William Stager and Hui Quan and Marilyn Maroni
• Companies: Sanofi and Columbia Univ. and Consultant and Sanofi and Sanofi
• Keywords: Non-Inferiority ; Tuberculosis ; Meta-Analysis

Abstract

Non-inferiority margin calculation entails a rigorous analytic and often laborious process involving historical trials search and selection, data pooling and modeling, and margins derivation and selection. Both EMEA and FDA have guidelines on how to approach the problem, one of which also contains working examples of M1 and M2 derivations. Once established and accepted by the agencies, the margin would be applicable to the subsequent non-inferiority clinical trials in the same area as defined by the disease, indication and active control. Generally speaking, there are two approaches to non-inferiority margin derivation: a widely used meta-analysis approach and a less well-known single-study approach. In this presentation, we share a recent experience in which both approaches are taken to derive the non-inferiority margin for clinical trials in the treatment of latent tuberculosis infection. The key features and requirements of each approach are summarized, and their pros and cons are compared.