© 1976 by Biometrika Trust
The importance of the patient horizon in the sequential analysis of binomial clinical trials
Department of Mathematics, University of Essex Colchester
The sequential allocation of treatments by an experimenter is considered in the context of the patient horizon. The importance of taking into account the whole of the patient population in designing the trial is illustrated for play-the-winner allocation and vector-at-a-time allocation, both with success lead termination. The optimal scheme is considered to be the one which minimizes the total number of patients who receive the poorer treatment. The difference between the two allocation schemes is found to be small compared with the differences which result from incorrect estimation of the patient horizon, or from the usual procedure of ignoring the patient horizon and selecting an arbitrary significance level. Proposals are made concerning the operation of the termination rule in the absence of complete information about the patient horizon and other parameters.
Key Words: Binomial population Clinical trial Play-the-winner sampling Population horizon Selection procedure Vector-at-a-time sampling