Early decision in clinical trials when the treatment differences are small. Experience of a controlled trial in head...

In a long-term clinical trial, it is customary to review the results of the data periodically. Two possible situations suggesting a possible early decision in hypothesis testing may arise. First, the difference in outcome may be sufficiently convincing without further evidence from additional data. Second, the difference may be such that it is not likely to be significant even if additional data are collected. This article is mainly concerned with a method for reaching a decision under the second situation, in particular when inadequate patient acquisition has occurred. A predicted distribution of likely future outcomes on the basis of data already accumulated is used to evaluate what might happen if the trial were carried on to the designed end. The method is intended for the case where the outcome variable is dichotomous and the two groups are compared for the difference in proportions.