A review of randomization methods in clinical trials

Author(s): Vance W Berger & Olga Antsygina

In many respects, clinical trials can be seen as an art as well as a science, in that there is ample discretion for investigators to select research methods reflecting their own individual preferences. In fact, new research methods are developed on a fairly regular basis, not all of them improvements over existing methods. But the opposite trend also remains in effect, as researchers often follow established precedent, rather than thinking through the issues relevant to the current trial so as to come up with the research methods that are optimal in this case. These two forces pulling in opposite directions, individuality and inertia, compete in many aspects of clinical research, including the specific methods of randomization. New randomization methods are constantly proposed, while at the same time more and more researchers seem to be using the established standards of permuted blocks randomization or minimization (which, in its most extreme form, is not even true randomization at all). A comprehensive review of all randomization methods is beyond the scope of this work, but we will review these two established standards, as well as the newer (and vastly better) maximum tolerated imbalance procedures, including the big stick, Chen’s procedure and the maximal procedure.