Online Program

Return to main conference page
Thursday, September 13
Thu, Sep 13, 11:45 AM - 1:00 PM
Various Rooms
Roundtable Discussions

TL20: Role of Bayesian Methods for Design and Analysis of Noninferiority Trial (300614)

*Samiran Ghosh, Wayne State University School of Medicine 
*Ram Tiwari, FDA/CDRH 

Keywords: Assay Sensitivity, Two-arm vs. three-arm, Noninferiority margin, Sample Size

With the slow-down of “Blockbuster” drug discovery Noninferiorty (NI) trials are becoming more and more relevant and mainstream, when compared traditional superiority trial. Availability and the use of historical and other subsidiary data provides an exciting opportunity for many legacy disease areas. However, for NI design the usefulness of this “additional” information is even more crucial. This is because NI trials often requires external validation, which involves optimal usage of historical data. NI trials involves legacy drug/s and availability of such historical data more or less guaranteed. However, this also leads to several challenges which statistical community must put into critical attention. For example, 1) Pros and cons of recommending 2-arm vs. 3-arm (with placebo) trial; 2) How much confidence should we put on historical data and what are the principles that should guide us in case prior and current data conflict! 3) Role of Meta-analysis in case of availability of several historical data sources (which goes beyond RCT’s). 4) How the heterogeneity of between and within trials should be balanced? and finally 5) Construction of Noninferiorty margin for different types of outcomes and multiple end-points. Recent FDA guidance (November 2016) for NI trial makes these issues even more relevant and this is high time we should brainstorm on possible future steps. This roundtable will provide a great opportunity to educate some of the recent developments and future collaborations on emerging issues relevant to NI trials.