Inter-rater reliability of the French version of the core index for melancholia

Abstract

Background Assuming that psychomotor disturbances represent the core and are specific of melancholia, Parker et al. have developed the CORE, an 18-item scale assessing retardation, agitation and non-interactivity by behavioural observation which is able to distinguish melancholia from other depressive disorders. We report an inter-rater reliability study of the French version of the CORE. Methods 35 French-speaking in-patients, with ICD-10 criteria for major depression underwent a video-recorded interview aimed to rate the CORE. Each patient's recording was rated by 5 psychiatrists. We tested the inter-rater reliability of the total CORE score and of each of its three subscales' scores using the intra-class correlation coefficient (ICC). A cut-off score for melancholia was established against the opinion of the psychiatrist in charge of the patient using a ROC curve. We used Cohen's kappa to assess the agreement between raters as to rate patients above the cut-off, namely the allocation of melancholia. Results The global ICC for the total score was 0.88 and ranged from 0.97 to 0.75 for the varying rater dyads. A ROC curve yielded an optimal cut-off of 5 for melancholia. The global kappa for the agreement on melancholia allocation was good (0.65). Limitations The five raters had not exactly the same condition of quotation. The agreement for the “agitation” subscale was poor. Conclusion The French version of the CORE, has a good to excellent inter-rater reliability for the total score as for the allocation of melancholia according to a cut-off. Further validation studies are required to allow research application.