Aim: To examine if melancholic depression is over-represented in those with
'bipolar depression' and, if confirmed, to use that phenomenon to assist t
he clinical definition of melancholia. Methods: We contrast 83 bipolar and
904 unipolar depressed patients on three melancholic sub-typing systems (DS
M, Clinical and CORE system) and compare representation of their clinical d
epressive features. Results: By all three melancholic sub-typing systems, t
he bipolar patients were more likely to receive diagnoses of 'melancholia'
and of psychotic depression. To the extent that this differential prevalenc
e of depressive sub-types was reflected in varying patterns of clinical fea
tures, we so indirectly identified a set of items defining 'melancholia'. B
y such a strategy, melancholia was most clearly distinguished by behavioura
lly-rated psychomotor disturbance. While a number of 'endogeneity symptoms'
were significantly over-represented, logistic regression analyses refined
the set to psychomotor disturbance (both as a symptom and as a sign) and pa
thological guilt. We also established a distinctly higher prevalence of bip
olar depression in those where a refined diagnosis of melancholia was made.
Conclusions: Bipolar depression appears to be more likely to be 'melanchol
ic' in type, thus providing an indirect strategy for the clinical definitio
n of melancholia. (C) 2000 Elsevier Science B.V. All rights reserved.