In clinical trials, blinding is usually employed to prevent bias that may be introduced due to the knowledge of the identity of the treatment codes. This bias could alter the conclusion of statistical inference on the treatment effect. The purpose of this article is to propose a method for analysing clinical data with breached blindness. The example regarding the study of the effectiveness of an appetite suppressant in weight loss in obese woman as described in Brownell and Stunkard (Am. J. Psychiatry 1982; 139:1487-1489) is used to illustrate the application of the proposed methods.
Copyright 2004 John Wiley & Sons, Ltd.