All Times EDT
Keywords: Stratification, Dynamic Allocation, Randomization
Aligned with the theme of this year’s RISW, we will be discussing an old method first published in the 1970’s about dynamic allocation for assigning treatment arms in clinical trials where there is a large number of potential confounding baseline characteristics. For this audience, some historical examples from the past 15 years with regulatory interactions will be shared. Presently, this topic has re-emerged with interest in controlling confounders due to personalized medicine, which has led to increased heterogeneity of patient populations in later line therapies and combination therapies after failure on monotherapies. In addition, when the outcome of a trial is not positive, the first exercise often is to examine the demographics and baseline characteristics by arm to identify explanations. If differences are found, the healthcare community calls into question the interpretability of the trial results. Execution of dynamic allocation methods, and resulting appropriate analysis using re-randomization tests, could avoid this uncertainty. Of course, there are strengths and drawbacks of using dynamic allocation and the popular approach (permuted block stratification), and we will be summarizing these from existing literature. Looking forward, we propose novel hybrid approaches which hopefully combines the strengths of each existing approach without their drawbacks, for potential future implementation and analysis.