Validation of a new prototypic measure of melancholia

Multiple approaches have been adopted in an attempt to effectively identify and discriminate melancholic and non-melancholic depressive subtypes. We recently developed the Sydney Melancholia Prototype Index (SMPI) which incorporates antecedent and illness course variables as well as symptoms, with clinician-rated and self-rated SMPI versions, and with the former having been shown to have superior sensitivity and specificity in discriminating melancholic from non-melancholic depression. The aim of this study was to further evaluate the capacity of the SMPI to identify melancholia in comparison to DSM-based and clinician-judged assignments. The sample comprised 214 patients diagnosed with melancholic or non-melancholic depression according to a detailed clinical assessment and by the Mini International Neuropsychiatric Structured Interview (MINI) assessing formal DSM-IV melancholia criteria. DSM-IV assignment to melancholic versus non-melancholic depression was contrasted with clinician-judged allocation, the combination of these two strategies ("concordant diagnoses"), and to the SMPI (CR or clinician-rated and SR or self-report versions), with the likely validity of each approach examined against historical ascriptions for melancholia. DSM-IV criteria assigned the highest percentage of the sample with a melancholic diagnosis (64%), whereas the SMPI-SR assigned the smallest percentage with a melancholic diagnosis (37%). DSM-IV assignment was associated with the fewest number of validating variables, whilst SMPI-CR and independent clinician diagnosis were associated with the greatest number of differentiating variables including negative childhood experiences, past and recent stressors, satisfaction with life and perceived social support. These comparative analyses provide further support for the SMPI-CR in identifying and discriminating melancholic depression from non-melancholic depression. Replication of these findings in other samples with independent raters is recommended.