73 – Randomization and Covariate Adjustment
The Risk of Imbalance in Treatment Assignments in Stratified Blocked Randomized Clinical Trials
Bingzhi Zhang
Sanofi
Hui Quan
Sanofi
Lin Wang
Sanofi
Lynn Wei
Sanofi
In clinical trials, the stratification and blocking method is frequently used to achieve balance in prognostic factors across treatment groups. However, such randomization strategy has a disadvantage that may lead to overall imbalance in sample size across treatment groups when the number of strata is large. To measure the risk of such imbalance, Hallstrom and Davis (1988) provided the variance of the sample size difference between two treatment groups. However, the distribution assumptions made in their method has limited its validity and applicability. This limitation was confirmed by the simulation studies conducted by Kundt and Glass (2012). Kundt and Glass then proposed utilizing simulation to assess the likelihood of imbalance inclinical trials where Hallstrom and Davis' assumptions are no longer valid. In this paper, we proposed a new method deriving the exact probability density function (pdf) of the sample size difference between any two treatment groups in any multiple-arm and stratified blocked randomized clinical trials. It will help the clinical trial designers choose an appropriate stratification strategy at the planning stage to avoid serious imbalanced trials at the end.