To the Editor: The review of the psychometric properties of the Hamilton depression scale by Dr. Bagby et al. included the most relevant studies published from January 1980 to May 2003 that examine both interrater reliability and validity. It clearly demonstrated that the 17-item version, which in this period has been the gold standard as the outcome measure in trials with antidepressive therapies, is a multidimensional scale.

One potential evolutionary solution for a one-dimensional gold standard, as suggested by the authors, would be to use the six core items of the dimension of depression comprising depressed mood, guilt, work and interests, psychomotor retardation, psychic anxiety, and general somatic symptoms—the 6-item Hamilton depression scale—when we measure the outcome of antidepressive interventions because this subscale has been proven to be more effective than the 17-item version in detecting differences between active drug therapy and placebo in trials on the acute therapy of depression. Also, this subscale fulfills criteria for unidimensionality. However, the authors rejected this solution, arguing that the truncated set of six items seems limited in that these items do not "permit capture of the full depressive syndrome" (p. 2174). Exactly this argument was behind the development of the Bech-Rafaelsen Melancholia Scale (1), which contains 11 items covering the various aspects of the full depressive syndrome and is the opposite of what is seen in the full manic syndrome, as measured by the Bech-Rafaelsen Mania Scale (2), that, together with the Bech-Rafaelsen Melancholia Scale, is still the only scale for the bipolar spectrum to fulfill the item-response theory model for unidimensionality (1, 2). In the light of this, we wonder why the authors did not mention this scale in their discussion of a replacement for the Hamilton depression scale.

In a recent study (3), we confirmed in a large sample of depressed patients that the Bech-Rafaelsen Melancholia Scale and the 6-item—but not the 17-item—Hamilton depression scale have been accepted by item-response theory models to be unidimensional depression scales.

Both the 6-item Hamilton depression scale and the Bech-Rafaelsen Melancholia Scale can serve as gold standards for measuring pure antidepressive activity, whereas the other 17-item Hamilton depression scale dimensions (e.g., anxiety and sleep) might serve to identify other aspects of the treatment being examined.