Depression rating scales demonstrate significant correlations but systematic differences: a multicenter prospective cohort study using equipercentile linking

To evaluate correlations and score equivalence among four widely used clinician-rated and self-report depression scales-the 17-item Hamilton Rating Scale for Depression (HAMD-17), 6-item Hamilton Rating Scale for Depression (HAMD-6), 16-item Quick Inventory of Depressive Symptomatology-Self-Report (QIDS-SR16), and Patient Health Questionnaire-9 (PHQ-9)-to support cross-walk-based interpretation and comparison across studies and research settings.

This national, multicenter prospective cohort study included 1418 adults with major depressive disorder recruited from 12 sites across 11 regions in China. Depression severity was assessed at baseline and at months 1, 2, 3, 4, and 6, yielding 5597 assessments. All four scales were administered concurrently at each visit. Agreement was evaluated using Spearman correlation coefficients and Bland-Altman plots. Equipercentile linking with log-linear smoothing was applied to derive score conversions across scales.

All pairwise baseline correlations between scales were statistically significant, with the strongest correlation observed between HAMD-6 and HAMD-17 (r = 0.73). Clinician-rated scales consistently yielded higher response and remission rates than self-report scales. Bland-Altman analyses revealed systematic differences across measures, with the widest limits of agreement observed between HAMD-6 and HAMD-17 (-3.47 to 14.81), followed by PHQ-9 (-6.14 to 10.61) and QIDS-SR16 (-5.56 to 11.02). Equipercentile linking curves and conversion table were generated to translate HAMD-6, QIDS-SR16, and PHQ-9 scores into HAMD-17 equivalents.

Despite significant correlations, commonly used depression scales exhibit systematic differences in measurement and outcome classification across instruments. Equipercentile linking provides a practical framework for score translation, facilitating cross-study comparisons, secondary data analyses, and harmonization of depression outcomes.

Different scales are commonly used to measure depression severity, but their scores are not directly interchangeable. In this large multicenter study from China, we examined how four widely used depression scales relate to one another when completed at the same time. Although the scales showed moderate to strong correlations, they differed systematically in how treatment response and remission were classified. By developing conversion tables between these instruments, our findings help clinicians and researchers compare results across studies and better interpret depression scores when different scales are used.