Use of Probability of Success to Improve Study Planning and Forecast Probability of Noninferiority

• Activity Number: 653
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Peter Hu*+
• Companies: Bristol-Myers Squibb
• Keywords: Probability of success ; Conditional power ; Predictive power ; Non-inferiority margin ; Relative risk ; Inferiority

Abstract

Unless the chance for an experimental therapy to be superior to an active control is high, study power configured for non-inferiority at design stage is often exaggerated if no difference between the two therapies under alternative hypothesis is assumed, especially when the assumed incidence rate for the active control turns out too high to reach. Probability of success (PoS), defined as Bayesian expected power, may help mitigate such an impact resulted from this assumption. In addition, confidence interval-based Frequentist's approach can be inconclusive or suggest equivocally both inferiority and non-inferiority but without telling straightforward how likely non-inferiority at the end of the study can be achieved based on data already observed. PoS, again, can be used to fill in such a gap. Summary data during various stages of follow-up from the SYNTAX (Serruys, et al (2009)) and PRECOMBAT (Park, et al (2011)) trials will be used to illustrate the above points in a retrospective fashion, without the need of individual patient level data.