TL31: Role of Futility Analysis in an Unblinded Interim Analysis
*Peter Hu, Janssen Pharmaceutical companies of Johnson & Johnson 


When conducting pre-planned unblinded interim analysis (IA), for example, for efficacy claim or sample size adjustment, sometimes a futility analysis is also being conducted. If the futility analysis is included NOT to reduce type 1 error, rather it is simply for sponsor to monitor the study, regulatory statisticians in general do not have concern of its inclusion and this practice is well accepted within the industry. However, it is quite attractive to use the option of terminating the study for futility to reduce the alpha, or offset the inflation of alpha caused by stopping for efficacy claim and / or sample size increment. This use of the futility analysis has not been as commonly accepted, primary due to regulatory concern that the sponsor may not enforce terminate the futility rule. This roundtable discussion will focus on the experience / opinion of futility analysis, especially the one used to offset the type 1 error rate inflation. In particular, input from regulatory statistician under what circumstance they may feel more comfortable about the use in the case is welcome.