ABSTRACT
The usual evaluations of the methodological quality of randomized trials, or the more modern and trendy term “risk of bias,” tend to produce the appearance that most randomized trials are rigorous and only a very few exceptional ones are flawed to the point of constituting a concern. But more careful scrutiny reveals a far darker reality, which is that randomized clinical trials are not, in general, conducted especially well. Nor is this commentary limited in its scope to only a small minority of trials. To a large extent, this collective failure is due to the fact that the parties with the greatest vested interest in the outcomes of the trials are the same ones who get to influence these trials by conducting them and, in some cases, also influencing the manner in which the quality of this trial is to be evaluated. In many cases, even the appearance of impartiality is absent. A complete enumeration of the various methods used to bias trials would be a worthwhile endeavor, but would also fill several volumes of an encyclopedia. Even if we focus exclusively on randomization methods, we still find almost all trials conducted to the lowest standards imaginable. Though there may be some individuals who use flawed randomization methods, such as blocked randomization, with the specific intention of biasing the trial, there are many other investigators with no such ulterior motive who also use fatally flawed randomization methods, while believing that they are using valid procedures. A program of improved education regarding randomization methods is needed to root out the rampant misinformation being taught in the classrooms and in the journals, so that it will no longer be possible, or likely, to use flawed methods by mistake. Once this is done, it will also be correspondingly that much harder to use flawed randomization methods intentionally. Once improved methods are used, this will reign in trial results so that they better reflect reality and, in turn, are more helpful to the patients who depend on trials to inform their medical decisions. The purpose of this chapter is to point out some of the more common misconceptions that are “out there,” and that continue to be taught and understood as if they were facts, and replace them with a more correct understanding of the interplay among the randomization method, allocation concealment, masking, and selection bias.
