Key points are not available for this paper at this time.
In most randomised trials each subject is individually assigned at random to an intervention group. The intervention is applied directly to the subject and observations are made on each individual to determine the outcome of the intervention. Sometimes subjects cannot be allocated independently, or they may interact with one another during the treatment period. In trials of population screening, for example, screening centres may be set up in some districts and not in others. This may be necessary because widespread publicity is needed to encourage subjects to come for screening, or because members of the screening group might pass on information to neighbours who have been allocated to the control arm (no screening), leading them to demand screening. In the Swedish two county trial of breast cancer screening the county of Kopparberg was divided into seven geographical areas.1 2 Each was then subdivided into three units, …
Bland et al. (Sat,) studied this question.