Comparison of Methods for Handling Missing Data in NonInferiority Psychiatric Trials

• Activity Number: 515
• Type: Contributed
• Date/Time: Thursday, August 10, 2006 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Isaac Nuamah*+ and Guohua Pan and Kwang-Shi Shu and Pilar Lim
• Companies: Johnson & Johnson Pharmaceutical R&D and Johnson & Johnson Pharmaceutical R&D and Johnson & Johnson Pharmaceutical R&D and Johnson & Johnson Pharmaceutical R&D
• Keywords: non-inferiority ; per-protocol population ; psychiatric trials ; intent-to-treat ; missing data

Abstract

Noninferiority (NI) analysis is used to demonstrate a new treatment is no worse than a known effective treatment by some predefined margin. While the intent-to-treat (ITT) principle is accepted as an approach to demonstrate superiority, an NI comparison requires consideration of the analysis population (AP). The inclusion of subjects with protocol deviations usually leads to reduced treatment differences, therefore resulting in an anti-conservative bias that would favor the NI conclusion. Analyses based on both the per-protocol (PP) population and the ITT population may need to be consistent to support a claim of NI. Regardless of the AP, missing data and methods to address them would have to be addressed during analysis. We compare the type I error rate and power of several methods for handling missing data in NI analysis based on ITT or PP via simulation.