Reliability of the Hamilton Rating Scale for Depression: A meta-analysis over a period of 49 years

The aim of this study was to provide a comprehensive meta-analytic review of the reliability of the Hamilton Rating Scale for Depression (HRSD) for the period 1960–2008, taking into consideration all three types of reliability: internal consistency, inter-rater, and test–retest reliability. This is the first such meta-analytic study of a clinician-administered psychiatric scale. A thorough literature search was conducted using MEDLINE and PsycINFO. The total number of collected articles was 5548, of which 409 reported one or more reliability coefficients. The effect size was obtained by the z-transformation of reliability coefficients. The meta-analysis was performed separately for internal consistency, inter-rater and test–retest reliability. A pooled mean for alpha coefficient in random effects model was 0.789 (95%CI 0.766–0.810). The meta-regression analysis revealed that higher alpha coefficients were associated with higher variability of the HRSD total scores. With regard to inter-rater reliability, pooled means in random effects model were 0.937 (95%CI 0.914–0.954) for the intraclass correlation coefficient, 0.81 (95%CI 0.72–0.88) for the kappa coefficient, 0.94 (95%CI 0.90–0.97) for the Pearson correlation coefficient, and 0.91 (95%CI 0.78–0.96) for the Spearman rank correlation coefficient. A meta-regression analysis showed positive association between inter-rater reliability and publication year. Test–retest reliability of HRSD ranged between 0.65 and 0.98 and generally decreased with extending the interval between two measurements (Spearman r between the duration of interval and test–retest reliability figures = –0.74). Results suggest that HRSD provides a reliable assessment of depression. Figures indicate good overall levels of internal consistency, inter-rater and test–retest reliability, but some HRSD items (e.g., “loss of insight”) do not appear to possess a satisfactory reliability.