The 1990 Darwin Lecture was delivered by Sir Richard Doll, FRS at the Royal Society of Medicine on 7 November 1990.
Sir Richard described the progress which had been made in the design and interpretation of controlled clinical trials over the last fifty years. Before 1937 the evaluation of new treatments had been based on subjective assessments of their efficacy by senior practitioners, an approach which worked well where there was a dramatic and quickly detectable effect but which was incapable of yielding reliable conclusions with less effective but nevertheless potentially beneficial treatments.
Although the underlying ideas were not new, and indeed had already been adopted as the basic principles of experimental design in agriculture, it was not until Austin Bradford Hill published a series of articles on the ‘‘Principles of Medical Statistics’’ in the Lancet in 1937 that the concept of controlled trials was taken seriously by the medical profession. The early trials involved the treatment of alternate patients which allowed some bias in a doctor’s decision whether to include a patient in the trial at all. Randomisation was the answer to this problem and in 1946 Bradford Hill persuaded a Medical Research Council committee to use it in a trial aimed at testing the value of pertussis vaccine. A few months later he extended the method to the clinical field to test the efficacy of streptomycin in the treatment of pulmonary tuberculosis.
The need to obtain an objective assessment of a patient’s progress, ideally by having both patient and assessing clinician ignorant of the treatment being administered, had to be balanced against ethical considerations such as the unpleasantness of any placebo given to patients in a control group and the right of any doctor to change a treatment if judged to be in his patient’s best interests. Indeed, it is noteworthy how much attention was paid to ethical considerations long before ethical vetting committees were thought of.
Developments in recent years have centred on better trial design and ever more refined statistical techniques for extracting the maximum information from the available evidence. Inevitably an increasing proportion of new treatments shows benefits which are far from easy to demonstrate conclusively. Nevertheless, they are of immense importance in the relief of human misery - a reduction in the fatality of breast cancer from 40% to 35%, for example, would avoid more premature deaths in a year than the total elimination of multiple sclerosis but it might take a randomised sample of at least 10,000 patients to demonstrate such an effect. Not surprisingly, inconclusive clinical trials have become common.
Two developments have been aimed at solving this problem. One is the statistical overview which pools the results of all trials of the same treatment, taking due account of the different ways in which they have been conducted and permitting conclusions which none of the trials in isolation would have justified. The successes of this method include the demonstration of the beneficial effects on the survival of patients with breast cancer from treatment with both cytotoxic chemotherapy and tamoxifen. The other development is the large scale collaborative trial involving several hundred hospitals in several countries. Among the successes of this method is the demonstration of the efficacy of aspirin and streptokinase, administered within a few hours of myocardial infarction, in reducing subsequent fatality.
In conclusion Sir Richard celebrated the fact that over the last fifty years the useless remedies which hitherto filled dispensaries throughout the world have largely disappeared from developed countries. There now exist techniques for ensuring that the true effects of new therapies are quickly discovered, largely due to the innovations of British academics.
Robert Peel