Case identification of depression with self-report questionnaires

Abstract

Many self-report measures that are used to identify cases of depression are symptom severity measures that are adopted for diagnostic purposes by use of cutoff scores. A troublesome problem with this approach is that optimal cutoff scores often vary across studies, which increases the difficulty of cross-study comparisons. This study evaluated the performance of a DSM-IV based depression screening scale, the Diagnostic Inventory for Depression. We compared the diagnostic performance of two different approaches to scoring the DID: a cutoff scoring approach and a standardized DSM-IV symptom-summation algorithm. Clinical diagnosis based on a semi-structured interview was the standard of comparison. Receiver operating characteristic (ROC) analysis indicated that a DID cutoff score performed comparably to the DID algorithmic approach in identifying cases. This finding is in contrast to prior research which suggested that algorithmic approaches might improve test performance over the cutoff score approach. The manner by which a user might choose the appropriate scale-scoring method for case identification is discussed.