Adaptive Monitoring of HIV Prevention Trials in the Presence of Extreme Treatment Effect

• Activity Number: 700
• Type: Contributed
• Date/Time: Thursday, August 13, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): William Koh* and Scott Emerson
• Companies: and University of Washington
• Keywords: adaptive design ; prespecified unblinding ; time-to-event ; group sequential design ; calendar time ; prevention trials

Abstract

In recent years, there has been considerable interest in the possibility of incorporating adaptive features in clinical trials of new prevention strategies. Recent HIV preventions trials such as HPTN052 and Partners Pre-Exposure Prophylaxis dealt with operational issues related to the observation of extremely low event rates at blinded analyses. In such a setting, trial investigators may need to consider increasing accrual rates, prolonging calendar time of follow-up, and/or accepting a lower number of events. The best strategy from among those options may depend on whether such an observation arises primarily from a low background event rate, primarily from an extremely beneficial treatment effect, or from a combination of both. In this research, we compare our ability to preserve study precision solely through the use of blinded adaptation within a pre-specified group sequential design versus the use of an unblinded adaptation that might better distinguish between low background event rates and extreme treatment effects. In particular, we consider the constraints that calendar time might impose on the scientific interpretation and statistical credibility of the chosen strategies.