King M et al. (2005) Impact of participant and physician intervention preferences on randomized trials: a systematic review. JAMA 293: 1089–1099

Although a randomized controlled trial is generally accepted as the most effective way to assess clinical efficacy, patients' preferences for a particular treatment might affect the study's validity. This is particularly relevant when treatments are not blinded, and is likely to become increasingly important as patients take an ever more active interest in their management.

To assess the magnitude of any effect of patient choice on recruitment or outcomes, King and colleagues have carried out a systematic review of clinical studies that recorded patient or physician treatment preference. The selected studies followed up all participants, whether allocated to random or preference cohorts.

The results revealed that a considerable proportion of patients refused randomization because they preferred one treatment over another. In 14 of 27 studies included in the analysis, more than 50% of individuals refused randomization after having agreed to participate in the trial. Reassuringly, however, there was little bias in the characteristics of those who were randomized, and only small differences were found between the randomized and preference groups in terms of outcomes.

King et al. conclude that although recruitment to randomized trials is strongly affected by patients' preferences, neither external nor internal validity appears to suffer significantly. These findings also lend support to the use of observational studies in areas where randomization is not appropriate.