Some Key Issues in Noninferiority Trials

• Activity Number: 33
• Type: Invited
• Date/Time: Sunday, August 3, 2003 : 4:00 PM to 5:50 PM
• Sponsor: ENAR
• Author(s): Thomas R. Fleming*+
• Companies: University of Washington

Abstract

In clinical settings where therapeutic regimens have been established to provide an effective standard of care (SOC), there often is interest in determining whether an experimental intervention (EXP) would provide efficacy that is not meaningfully inferior to that of the SOC. From the perspective of improving clinical care, such "noninferiority trials" are particularly well motivated when EXP is expected to have a more favorable profile in safety, convenience of administration, or cost. Noninferiority (NI) trials, by design, provide valuable direct insights regarding the relative efficacy of the EXP and SOC regimens. However, the NI trial usually only provides indirect evidence about the actual level of efficacy (relative to a placebo) of the EXP. This leads to many complex and controversial issues in the design, conduct, analysis, and interpretation of such studies. We will discuss a number of these controversial issues in NI trials. We consider why it is critical that the SOC have substantial levels of efficacy that are precisely established, and where the data providing such precise estimates come from.