Comparing MTI randomization procedures to blocked randomization.

Randomization is one of the cornerstones of the randomized clinical trial, and there is no shortage of methods one can use to randomize patients to treatment groups. When deciding which one to use, researchers must bear in mind that not all randomization procedures are equally adept at achieving the objective of randomization, namely, balanced treatment groups. One threat is chronological bias, and permuted blocks randomization does such a good job at controlling chronological bias that it has become the standard randomization procedure in clinical trials. But permuted blocks randomization is especially vulnerable to selection bias, so as a result, the maximum tolerated imbalance (MTI) procedures were proposed as better alternatives. In comparing the procedures, we have somewhat of a false controversy, in that actual practice goes uniformly one way (permuted blocks), whereas scientific arguments go uniformly the other way (MTI procedures). There is no argument in the literature to suggest that the permuted block design is better than or even as good as the MTI procedures, but this dearth is matched by an equivalent one regarding actual trials using the MTI procedures. So the 'controversy', if we are to call it that, pits misguided precedent against sound advice that tends to be ignored in practice. We shall review the issues to determine scientifically which of the procedures is better and, therefore, should be used.

RCT Entities:   Population Interventions Outcomes