Using the data of a positive d.b.c.t. comparing an hypericum extract (W55570) to placebo in depressed patients we explored whether the Ham D6 was unidimentional and in case of a positive answer whether the total score was as sensitive as the total score of the Ham D17.

The study was a 6 weeks double blind placebo controlled trial comparing 300 mg of hypericum t.i.d (n = 186), to placebo (n = 189), in patients with a single or recurrent depression according to DSM-IV. Superiority of hypericum versus placebo on the main outcome criterion (HDRS 17) was already published.

The unidimensionality of the Hamilton depression scale 6 and 17 items were tested using a Mokken scale analysis. The effect size according to the initial severity of depression was calculated on the ITT last observation carried forward population.

The Ham D6, covering the core symptoms of depression was unidimensional, implying that improving this score reflects a true antidepressant effect. The Ham D17 was not unidimensional.

Hypericum was an effective antidepressant in patients with a pre-treatment score of 12 or more (n = 208) on the Ham D6, the effect size was 0.46. No difference with placebo was observed for those with a score of less than 12 (n = 167).

For the evaluation of an antidepressant effect, because of its specificity and sensitivity, the Ham D6 should be used as a primary outcome measure rather than the Ham D17.