Risk of selection bias in randomized trials: Further insight

Abstract

The quality of randomization is an under-appreciated facet of trial design. The present piece represents an advance in our collective understanding of how allocation concealment and randomization relate to risk of selection bias in randomized trials, and other measures are also considered. Though the overwhelming majority of the advice given is timely and correct, it is more instructive to focus on the relatively narrow sliver of advice that is incorrect (namely, that trials should not stratify by site, and that unrestricted randomization is a solution to the problem of selection bias), so it is in this context that the comments here must be understood. In no way is this intended to be a rebuttal of the excellent work we have before us. Rather, it is a refinement.