POSSIBILITY OF INCORPORATING PATIENTS FROM NON-RANDOMIZING CENTERS INTO A RANDOMIZED CLINICAL-TRIAL

The implications of incorporating patients from non-randomising centres into the design of a 2 arm randomised multicentre clinical trial are investigated from the point of view of Bayesian statistics. It is concluded that although the inclusion of such patients may contribute to the reduction in recruitment time to a trial in the special circumstances when any biases introduced by such centres can be assumed very small, this is offset by a larger number of patients to be followed than in the corresponding fully randomised trial. © 1979.