ABSTRACT
OBJECTIVES: The present study investigates the psychometric adequacy of the Hamilton Depression Scale-Self-Report(HAMD-Self-report) with respect to the measuring the severity of depressive symptom by the use of item-response models(two parameter logistic model). METHODS: 114 patients with depressive disorder as diagnosed by DSM-IV criteria were studied. All depressed patients were aged 22 years or older, with the mean age of 33. They were requested to fill up the questionnaires of Hamilton Rating Scale for Depression-Self-Report. The data was analysed by Bayesian 1.0 for windows for item analysis. RESULTS: 1) Of the 76 items that consist of HAMD-Self-report, only 66 fitted the Item response model. 2)The 66 items reflected the degree of depression necessary for each symptom to shift from 'low intensity' to 'moderate intensity', and from 'moderate' to 'high intensity'. 3) The subscales of feeling of guilty and weight loss were mainly consisted of items with high item severity. Therefore only the severely depressed subjects may response. The subscale work and activities was made of items with low item severity, and the subjects with much less severe may response. CONCLUSION: The 66 items that were fitted to the item response model represented the different level of depressive severity. By weighing each item differently with respect to the symtom severity parameter(item severity), each item on the subscales represents a different level of depressive severity. Therefore the total score computed by summing responses to the highest weighted item of each scale can be used as an adequate index of depressive severity.