Shuffle the deck, flip that coin: randomization comes to medicine.

Abstract

There is a lot of confusion about the first use of randomization in an experiment and the first use of randomization in medicine. Some writers think that the great statistician R A Fisher carried out the first randomized trial (in agriculture)1 and published the results in 1923.2–5 Others think that the first medical trial was carried out by the British National Research Council in the 1940s.6,7 But there are earlier examples and we think we have found the oldest. Readers of this series will know that controlled trials are as old as Biblical Daniel,8 the evaluation of smallpox inoculation in 1721,9 the evaluation of Mesmerism in 1784,10 and Semmelweiss’s trials to reduce hospital infections.11 To resolve this debate we have to be precise in our definition of randomization in experiments. The word “randomization” does not exist in the Oxford English Dictionary or its supplement.12,13 It has two frequent uses in social science and clinical research. The first is a “random sample” from a larger population. Here randomization is used to ensure that the sample is representative of the population as a whole. The second use of randomization in experimental design is our focus here. There is an experiment, purposefully and prospectively designed; there are control and experimental interventions on patients purposefully designed to eliminate every observable and unobservable difference between the two except for the experimental intervention; and there is a method of assignment which is out …