Evaluation of a Design with Sample Size Re-Estimation Based on Conditional Power

• Activity Number: 534
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Alison Pedley*+
• Keywords: adaptive design ; sample size re-estimation ; group sequential design

Abstract

Sample size calculations are a fundamental design input for any trial. Unfortunately, at the design phase, limited information leads to uncertainty in the estimation of key parameters used to calculate trial sample size. Inaccurate estimates of trial parameters may result in an undersized (lacking statistical power) or oversized (consuming valuable resources) trial. Inclusion of an adaptation allowing for a mid-trial correction to the sample size based on accumulating study data, also known as Sample Size Reassessment or Re-estimation (SSR), is attractive in that it offers "protection" against unnecessarily large or under-powered trials. However, this "protection" also comes at a cost. In this presentation, the operating characteristics of an unblinded sample size re-estimation design based on conditional power are compared with those of more traditional group sequential designs via simulations. The optimal design is determined for a variety of potential situations/various design constraints. The results also give insight into the considerations that should be made in exploring/selecting adaptive designs including SSR based on conditional power.