Comparative evaluation of cut-off criteria of the Hamilton rating scale for depression developed using ROC-analysis and Rasch model

Interpretability of the psychometric scale is based on the presence of certain threshold criteria that allow to divide all scores into levels of the construct measured. The purpose of this study was to conduct the comparative evaluation of the reproducibility and diagnostic efficiency of cut-off criteria of The Hamilton Rating Scale for Depression (HRSD) developed using the ROC-analysis and Rasch model. Three samples of research protocols obtained using HRSD included healthy subjects and patients with depression were established for comparative study. As a result of the study, it was found that cut-off ROC-criteria had low reproducibility and were characterized by high accuracy only if they were used in the same sample of data which was used for their developing. Unlike ROC-criteria, cut-off criteria developed on the base of Rasch model characterized by a high degree of reproducibility and were effectively consistent with clinical data regardless of the sample characteristics. These results were explained by the conceptual and methodological differences of both approaches in defining of cut-off criteria.