Maximum Tolerated Imbalance Randomization: Theory and Practice

Maximum tolerated imbalance (MTI) randomization is a class of restricted randomization procedures that control treatment imbalance within pre-specified limits while lowering allocation predictability in the randomization sequence. The literature has amply demonstrated that MTI randomization is objectively and demonstrably better, by several criteria, than the widely used permuted block randomization (PBR). Despite this, PBR remains the norm, so much so that its use may contribute to its own perpetuation. The perception that MTI randomization has never been used in practice contributes to its not being used in actual clinical trials, a catch 22. The purpose of this chapter is to dispute the contention that MTI randomization has never been used in actual clinical trials. We document numerous cases of MTI randomization and review the parameters and the rationale for the choice of MTI randomization methods in these trials. We argue that MTI randomization can be, and is, successfully implemented in practice and advocate for its continued use in clinical trial research.