Minimizing predictability while retaining balance through the use of less restrictive randomization procedures

Abstract

The interpretation of between-group comparisons is facilitated by the creation of treatment groups that are similar to each other in baseline composition. To prevent treatment effects from being confounded with time effects, most trials use restricted randomization to force balance. An unintended consequence of these restrictions is that they create patterns that allow for the prediction of future treatment allocations, and hence selection bias, especially in unmasked trials. In fact, the more restrictive the allocation procedure, the greater the potential for selection bias. It was decided, in the context of a recent clinical trial comparing two dosing schedules of paclitaxel and carboplatin for advanced stage IIIB/IV non-small-cell lung cancer, that the randomized block procedure could not simultaneously protect sufficiently against both selection and chronological bias. In this paper we detail our development of the maximal procedure. The maximal procedure takes as input the extent of chronological bias allowed by the randomized block procedure, then matches it, but does so with fewer restrictions. This feature makes the maximal procedure more resistant to selection bias than the randomized block procedure is.

Letter to the Editor: Efficacy and Degree of Bias in Knee Injury Prevention Studies: A Systematic Review of RCTs

“…Given the nature of the treatments involved, it is unlikely that even the one study that claimed masking could have been truly masked, although de facto masking [2] is always an option, and probably should have been used to at least conceal which treatment was the active one. Therefore with or without the claim of masking, none of the studies can be taken as perfectly masked, and any unmasking at all, coupled with any form of restricted randomization (such as permuted blocks), precludes the possibility of allocation concealment, because knowledge of even some of the prior allocations can allow for prediction of future ones [1,4]. Unless the trials used unrestricted randomization, which almost no trials use in practice, there was no allocation concealment.…”

“…Unless the trials used unrestricted randomization, which almost no trials use in practice, there was no allocation concealment. This does not imply that the randomization was improper, as there are also drawbacks to using unrestricted randomization [1,4]. However, the use of an excessively restricted procedure, such as permuted blocks, is inappropriate when masking is uncertain.…”

“…However, the use of an excessively restricted procedure, such as permuted blocks, is inappropriate when masking is uncertain. Grimm et al should have displayed in Table 2 the specific randomization procedure used in each trial, but without this, we cannot assume that all 10 research teams knew to use minimally restrictive procedures such as the (paradoxically named, given its introduction here) maximal procedure [1,4].…”

Letters to EBD

“…We remain therefore somewhat sceptical of the Cochrane recommendations for assessing selection bias risk, and adopted the position by Berger et al 13 Of course, such a position differs from the mainstream line of thought and seems, due to unfamiliarity with it, not easy been understood.…”

“…14,15 Since 2005, Berger et al advocate the inclusion of tests into the methodology of RCTs that provide empirical evidence as to whether randomisation and allocation concealment were indeed effective in protecting against selection bias. [13][14][15] We believe that only the reporting of such evidence merits the judgement of a trial as being of 'low selection bias risk'. According to such a rather stringent standard, we agree with Dr Hurst that all of our appraised evidence is to be regarded as 'poor'.…”