The place of randomisation in clinical trials

Randomised clinical trials are nowadays the accepted gold standard for treatment comparisons. However, controversies still exist concerning the possibility to perform randomised trials or the adequacy and the ethical appropriateness of randomisation. Randomisation is sometimes credited with advantages it does not possess, and an extremely negative view or even a categorical rejection of non-randomised trials is found. This attitude may be comprehensible from a historical, pragmatic or educational viewpoint, but is not well-founded on epistemological grounds. This article argues that non-randomised clinical trials are needed, clarifies the role of randomisation in clinical trials and analyses the arguments raised against the validity of results from observational studies. It is shown that it has not been demonstrated up to now that well-designed and analysed observational studies would have yielded results that are distinct or even qualitatively different from results of similar randomised clinical trials. Although one is well advised to randomise whenever possible, there is still room for a considerable improvement of observational studies: Randomisation itself should be the only difference between observational studies and randomised clinical trials.