Interim Analyses and Publication of Clinical Trials

In Reply.—  The issues raised by Dr Paykel are relevant and worthy of discussion. First he asks whether knowledge of the results may prejudice blindness and evaluation of remaining cases in the study. It is suggested this is a particular problem in our study since the two drugs have different side-effect profiles. It is true that the profile of side effects of these drugs are different enough so that a portion of the time the rater might guess which drug the patient is taking. However, it is unclear how knowledge of outcome enhances this potential source of bias. Keeping raters blind to outcome will not diminish the likelihood of their assuming that certain side-effect profiles are associated with treatment with phenelzine and others with imipramine.We have attempted to deal with this possible bias by extending the trial for responders. Patients who are considered to have benefited from the treatment, defined as