Abstract
Objective. To re-analyse the Spanish outpatient study on the various versions of the Hamilton Depression Scale (HAM-D) by using item-response theory models for measuring unidimensionality. Methods. The Mokken analysis, based on the coefficient of homogeneity, was used. A coefficient of 0.40 or higher was accepted for unidimensionality. The three HAM-D versions were compared: HAM-D6 (core items of depression), HAM-D17 and HAM-D21. Results. In total, 113 of the 116 patients included in the study completed the planned treatment period of 6 weeks. As the Mokken analysis depends as much on the sample of patients under examination (dispersion of scores on the item) as on the scale itself, the endpoint was used. For HAM-D6, but not for HAM-D17 or HAM-D21, the coefficient of homogeneity was acceptable (0.44). Conclusion. The HAM-D6 is a valid version of the Hamilton Scale to be recommended for monitoring depressed patients in antidepressant therapy in the outpatient setting.
Notes
*Although an abstract of the article in [4] is already indexed in English, the paper was originally written entirely in Spanish. Please see Appendix 1 for a briefing of relevant information concerning patients and methods in this study. More in-depth details are fully available upon request to the corresponding author.