In a meta-analysis of short-term (four weeks) trials of antidepressants, two response criteria were compared separately for low- (less-than-or-equal-to 21), medium- (22-27) and high- (greater-than-or-equal-to 28) scores on the 17-item version of the Hamilton Rating Scale for Depression: (HAMD-17): (a) 50 per cent reduction of the baseline HAMD- 1 7 score and (b) 50 per cent reduction plus a cut-off HAMD score of less-than-or-equal-to 10. The material consisted of the data pool of comparative trials with moclobemide provided by Hoffmann-La Roche, Basle. The analysis showed that, on average, the HAMD-17 scores decrease uniformly in all groups independently of the initial severity of depression. The percentage of score differences between high-score, medium-score and low-score groups remained virtually unchanged over the whole observation period. Furthermore, the criterion combining a 50 per cent decrease with a cut-off score of less-than-or-equal-to 10 gives clear disadvantage for high-scores to become responders since cases of severe depression require a much longer time to reach the threshold of 10 than do milder or moderately severe cases. As to the question of evaluating differences between placebo and drugs, our analyses provided evidence that the simple criterion of 50 per cent reduction of baseline HAMD-17 score is better suited to discriminate between placebo and drug responders in acute trials.
