Random assignment of individual participants in clinical trials entails two separatesteps: (1) generating an unbiased allocation schedule; and (2) using the schedule,without foreknowledge of upcoming allocations, to assign participants to treatmentcomparison groups. Both of these steps were implemented in two controlled trialsdone under the aegis of the British Medical Research Council in the 1940s1: the trial of patulin for the common cold,2,3 and the better known trial ofstreptomycin for pulmonary tuberculosis.1,4 However, the two separate stepsneeded to achieve unbiased treatment assignment were not identified clearly until1955, when Austin Bradford Hill distinguished them in the sixth edition of his bookPrinciples of Medical Statistics.5 Shortly thereafter, David Cox6 and Peter Armitage7 emphasised the importance of implementing an unbiased allocation by usingsealed envelopes to conceal an upcoming allocation until after a participant hadbeen irrevocably entered into a trial. Despite early recognition of the importanceof the two separate steps in unbiased allocation, however, no widely accepted termdenoting the process of concealing upcoming allocations had been adopted more thanhalf a century later.
展开▼
