We set out to examine the psychometric properties of the MDI in comparison to the BDI in a mixed group of patients with primary depression.
At the Department of Biological Psychiatry in Vienna currently depressed inpatients with either a depressive or a schizo-affective disorder filled out both MDI and BDI on day of admission and at a time-point two weeks later during their treatment. Furthermore the Hamilton Depression Scale (HAM-D) was administered by the treating clinician at both time-points.
In total, 51 patients were included in the study. The non-parametric item response analysis was preferred to the classical Cronbach coefficient α as the latter is influenced by the number of items in a questionnaire. MDI obtained a Mokken analysis coefficient above 0.40, indicating unidimensionality. To determine external validity severely depressed patients with psychotic symptoms (N = 10) were compared to the remaining non-psychotic depressed patients (N = 41). Although BDI and MDI showed a lower score for psychotic than for non-psychotic inpatients, the standard deviations for both were greater for psychotic inpatients. On the intercorrelations between the different scales, MDI showed for all coefficients values above 0.70. On the other hand BDI and MDI both showed the same degree of linear relationship as the usual versions of HAM-D.
Our results demonstrate that the MDI had the highest coefficients values and was sufficient as a measure for depressive disorders in psychiatric patients.