Background: The Hamilton depression rating scale (HAMD) has been criticised for its multidimensionality, sensitivity to change, and discriminative power to define remission. To overcome these limitations several short scales have been devised but they have had limited use. We compared the performance of five HAMD short scales and their parental 17-item HAMD on sensitivity to change and discriminative power.
Methods: A local multicenter study was conducted with depressed outpatients (n=113). Depression severity was appraised at baseline and at 6 weeks since inception with the HAMD-17 and the clinical global impression scale (CGI). Sensitivity to change was calculated by a within-group standardised effect size (dw). Discriminative power (against a clinical remission criterion [CGI=1]) was assessed by receiver operating characteristic (ROC) analysis and the areas under the ROC curves (AUC).
Results: There were no differences among the five short scales on sensitivity to change (HAMD-17 dw: 1.6, 95% CI: 1.3-2.0; subscales range: 1.5-1.7), and discriminative power (HAMD-17 AUC: 0.93, 95% CI: 0.86-0.99; subscales range: 0.86-0.99). Appropriate cutoff points to define remission with short scales are suggested.
Limitations: The non-independence of the scales may have overestimated their performance. Nevertheless their comparisons seem fair as we do not expect a differential bias among them.
Conclusions: The short scales showed similar performance when compared with the parental HAMD. Since some were devised as unidimensional depression severity measures, and others to be sensitive to change, their use could circumvent previous criticisms raised to the canonical HAMD.