+44 7868 792050
Background: Randomised controlled trials are widely favoured in research design as the most rigorous way of determining the effectiveness of a treatment. For assigning a small number of participants who are identified before the start of the randomisation into treatment and control, the simple randomization technique can lead to imbalance of covariates among the groups. Furthermore while the stratified randomization method can control for the effect of covariates, in smaller clinical trials, the allocation of participants to groups by flip of a coin can result in uneven arms when the number of participants in each stratum in low. Despite the ability of covariate adaptive randomization technique in minimising the difference in covariate between the arms, the techniques comes with an unnecessary increase in the
computational process specifically when number of covariates increases, and when all participants are identified prior to the randomisation. The purpose of this study was to propose a method of assigning small number of participants (68) who are identified before the start of randomisation, into treatment and control arms.
Methods: The participants were first assigned into strata. For strata with even number of participants, the
participants are sequentially pulled out of the strata on a random basis and assigned to arms by flip of a coin until half of the participants are assigned to any of the two arms. Then the remaining participants were assigned to the other arm. When the number of participants in a stratum is odd the first participants was pulled out of the stratum on a random basis and kept separate, then the remaining even number of participants were assigned to arms according to the method for strata that contain even number of participants. The first participants that were pulled out of the strata with odd number of participants were assigned sequentially using covariate adaptive randomisation method.
Results: Two arms were created with minimal difference between the two arms and with the sum of absolute difference equal to 12.
Conclusions: The method showed to be able to assign small number of participants into balanced arms with minimal computational costs when a number of covariates exist.