Patient preferences and randomised clinical trials

incorporates a judgment that is likely to be made in daily clinical practice. An alternative strategy is to use a crossover design with patients serving as their own controls, a strategy often forgotten for evaluating remedies for chronic diseases. Drs Brewin and Bradley have correctly identified the difficulties concerned in designing an explanatory study when psychological and motivational factors are prominent, but we do not agree with their solution. Perhaps the answer is that there is no solution (randomisation or no randomisation) to the extreme problem of determiining the relative efficacies of two interventions within a population that is effectively defined by its willingness to comply with only one. It would be unfortunate if Drs Brewin's and Bradley's criticism of the randomised controlled trial was accepted with its implication that discarding randomisation in favour of self selection would prove a methodological advance.