A comparison of four methods of allocation for clinical trials with small sample sizes.

Although not without critics, randomised controlled trials are the most powerful method of attributing cause-effect relationships between phenomena. Pysiotherapy is not immune from the need to demonstrate its effect using this methodology. Central to the success of trials is the method of allocation used within randomised controlled trials in physiotherapy, which through necessity are small in terms of sample size. In this study four methods of allocation are used on 48 stroke patients and the results compared. Three of the methods (simple randomisation, minimisation and alternate allocation) showed some form of imbalance. Two of the methods demonstrated potential problems in terms of selection bias (block randomisation and alternate allocation). All forms of randomisation (minimisation, block randomisation and simple randomisation) were capable of adjustment in order to overcome the problems of imbalance and selection bias, whereas with alteration did not appear to be redeemable. The authors strongly suggest that alternate allocation (alternation) is never used in clinical trials. These findings are pertinent, not only to researchers in stroke, but also to those involved in the design and execution of any randomised trials in pysiotherapy and the allied professions. The conclusions reached in this study should also help readers of randomised controlled studies in assessing the suitablility of allocation procedures.