An improved method of evaluating drug effect in a multiple dose clinical trial

In drug development, finding an optimal dose is normally carried out in a phase II trial. A phase III trial will then be conducted to demonstrate that the selected dose is efficacious and safe. As choosing a dose from the phase II trial which has the highest observed response rate could overestimate the true response rate of the selected dose, the data from the phase II study cannot be simply pooled with the data from the phase III study in a final analysis. Therefore, a solution to the overestimation problem needs to be found so that the information obtained from phase II dose finding clinical trials can appropriately be combined with the data in the phase III study. In this paper, the potential overestimation in a multiple dose clinical trial is assessed and a method for correcting this bias is proposed. Simulations show that stepwise over-correction, the proposed method, is better than methods such as Bonferroni's procedure. Copyright (C) 2001 John Wiley & Sons, Ltd.