Do Hamilton depression scale items have the capacity to differentiate melancholic and non-melancholic depressive sub-types?

Background: While there have been several Hamilton depression sub-scales named as to indicate that they have the capacity to differentiate melancholic depression, all but one actually measure depression severity. As a valid Hamilton-derived melancholia scale would have high utility in evaluating the comparative effectiveness of differing treatments for melancholia we designed a study with such an objective.

Methods: Items from the 17-item Hamilton depression scale over-represented in those with melancholia in five clinical studies generated two potential scales with nine and four items respectively, and were superior in differentiating melancholic and non-melancholic depression in comparison to a 6-item Hamilton measure. As melancholia is viewed as showing a superior response to antidepressant medication and a poor response to placebo we evaluated our newly derived scale measures against the total Hamilton measure in five randomized controlled trials.

Results: There was minimal and non-consistent support for our new measures in demonstrating that they showed greater superiority than total Hamilton depression scores in meeting one of melancholia's ascriptions - a superior response to antidepressant drug than to placebo.

Limitations: We assume that melancholia is a categorical type as against simply being a more severe expression of depression.

Conclusions: Confirming the ascription that melancholia shows a superior response to antidepressant drug and a lower response to placebo has not been able to be demonstrated from generating a melancholia index from Hamilton scale items.