Estimation of cutoff for the Severity of Dependence Scale (SDS) for opiate dependence by ROC analysis

Abstract

Background. The Severity of Dependence Scale (SDS) is a five-item scale that has been reported to be a reliable and valid screening instrument for dependence in several types of substances. Optimal cutoff points on the SDS indicative of clinically significant dependence have been determined for a large range of substance types, however, to date no data have been reported on its performance in a population with opiate dependence.

Sample. A structured interview was administered to 315 opiate-dependent patients in treatment.

Method. The diagnostic performance of the SDS was measured via Receiver Operating Characteristic (ROC) analysis according to the DSM-IV diagnosis of heroin dependence, as measured by section 12 of the Schedule for Clinical Assessment in Neuropsychiatry (SCAN).

Results. ROC analysis revealed the SDS to be a test of high diagnostic utility for the measurement of opiate dependence (Area Under Curve =0.8875). The cut-off point on the SDS at which there is optimal discrimination between the presence and absence of a diagnosis of heroin dependence was found to be 5 (i.e. a score of 5 or more). This score provides the best trade-off between sensitivity (83.15%) and specificity (84.51%). Similar results were found for heroin current consumption (AUC = 0.8325; cut-off = 5; sensitivity = 77.94 and specificity = 77.33).

Conclusion. The SDS can be recommended as an effective short instrument for the discrimination of the degree of dependency and heroin consumption in the clinical area.