Permutation Test Associated with Dynamic Randomization Methods

• Activity Number: 73
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 4:00 PM to 5:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Chunrong Cheng*+
• Companies: FDA
• Keywords: dynamic randomization ; permutation ; minimization

Abstract

Dynamic randomization methods may help to achieve balance across a number of stratification factors simultaneously. However, their applications in clinical trials remain controversial and the users are sometimes required to perform permutation (re-randomization) test as a sensitivity analysis. Under the recommendation that permutation should follow the fixed patient-entry-order, this author found by simulations that the possible number of unique combinations by permutation could be very limited in certain scenarios, especially when deterministic dynamic allocation methods (e.g. Taves minimization) is used. Even with the introduction of p by Pocock and Simon, the improvement may not be substantial. This author conducted extensive simulations to evaluate the performance of the permutation test and provides recommendations for a reliable and meaningful permutation test.