The Willis Lecture 2003: evaluating treatments for stroke patients too slowly: time to get out of second gear.

Abstract

Thomas Willis was born in 1621, received his classical education and then medical degree from Oxford in 1646, became a highly successful and prosperous physician, and, in 1660, was appointed Professor of Natural Philosophy at the University of Oxford.1 Although at that time there were no neurologists as such, he coined the word neurology in his Cerebri Anatome :2 “We should deliver an exact neurology, or doctrine of the nerves” and “without the perfect knowledge of the nerves the doctrine of the brain and its appendix would be left wholly lame and imperfect.” Of course, his pharmacopoeia was primitive in the extreme and he had no reliable way of proving that any treatment interventions worked, unless they had a very obvious and immediate effect. But these days we have the most remarkable tool to evaluate our interventions, the randomized controlled trial, although it is not yet 60 years since the first trial in the modern era was published in the BMJ on October 30, 1948, of streptomycin for pulmonary tuberculosis3 (Figure 1). It was conducted in the UK just after World War II when the supply of the antibiotic was so limited that Austin Bradford Hill, the statistician, was able to persuade clinicians to use randomization between the new treatment and control as a form of fair rationing and to construct 2 groups of patients who were so similar in their prognosis at baseline that any definite difference in their outcome must be because of the new treatment. Figure 1. The title page of the trial of streptomycin from BMJ in 1948 (with permission). And in the past 3 decades we have had another tool, meta-analysis, largely developed in Oxford by Richard Peto the statistician and Iain Chalmers who devised the Cochrane Library, and in the US …