Sample Size and Power Considerations for Futility Trial Design: Buying Power vs. Spending Beta

• Activity Number: 130
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Thomas Dobbins*+
• Companies: Merck & Co., Inc.
• Keywords: Futility ; Group Sequential Test ; Beta Spending

Abstract

In a futility group sequential test (FGST) a trial may terminate early only for lack of treatment effect. A key benefit of FGSTs is that they can lead to savings in trial time and cost, on average, relative to a fixed sample test (FST). However, if a treatment is effective, FGST increases trial time and cost since the current method of incorporating a beta spending function for total Type II error requires larger sample size (N) to maintain power relative to FST. In this talk we consider a FGST with K=2 groups, i.e., one interim analysis, and study total Type II error. We propose a method to construct this FGST that maintains the Type II error of the test of H0 without increase in N relative to FST, and illustrate the method with an example. The proposed FGST is uniformly more efficient than FST since E(N) is less than, and maximum N equals, that of FST always.