Was the randomised controlled trial the most important development in medicine this century? Some say yes. Others scoff. Whatever your view, it has clearly been essential for moving to a type of medicine where treatment is expected to be based more on firm evidence of benefit than on the treating doctor’s opinion. That transition is far from complete and hotly disputed. But it may be as important a change as that in the renaissance, when medicine began to base itself on experimental evidence rather than on reinterpreting the teachings of the ancients.
This theme issue marks the 50th anniversary of the publication in the BMJ of 30 October 1948 of the world’s “first” randomised trial (which is reproduced in full on our website). But the content is far from empty celebration. The importance of randomised trials is almost taken for granted. Rather, most papers concentrate on their many deficiencies and the huge scope for doing better.
The controversy begins with the history. Was the BMJ’s randomised controlled trial of streptomycin in pulmonary tuberculosis really the first? People had long had the idea that treatments ought to be tested by giving them to some patients and not others, usually by giving the test treatment to alternate patients (p ). Johannes Fibiger published a trial in Danish in 1898 in which he allocated patients to treatment by day of admission (p ) (large chunks of this paper are available on the BMJ website in a new, more accurate translation). The BMJ trial can claim to be the first to describe explicitly the method of randomisation, but, suggests historian Alan Yoshioka (p ), this may have been less for scientific reasons and more to relieve clinicians of the responsibility for deciding who would be treated when only a tiny amount of streptomycin was available.
Then comes a flood of deficiencies. Trials are often too small, too short, of poor quality, and poorly presented, and they address the wrong question (p 1181). Methodological inadequacies distort results (p 1185). Few trials include adequate measures of quality of life (p 1191). Cost data are poorly presented (p 1195). The ethical aspects of trials are often neglected (p 1209). The views of patients are either not sought or forgotten, and participants in trials often have limited understanding of what is happening (p 1177). Trials are usually poorly managed (p 1236). Politics have hijacked the conclusions of some trials (p 1224). Marketeers can use trials to further their own profit making ends (p 1231).
How exciting that there is so much room to improve. But no doubt the 100th anniversary edition will find as many defects. Just in case you need reminding, perfection is unattainable.