Most studies reporting measures of criterion validity used a dependence diagnosis (DSM-IV or ICD-10) retrieved from a clinical interview as an external criterion. Altogether, SDS and PUM were found to have a good ability to discriminate between dependent and non-dependent individuals, correctly classifying at least 85% of cases (Table 3). Though different cut-off points have been reported depending on the study sample, all evaluations revealed high values of specificity and usually lower, but still high, levels of sensitivity. As an exception, the study by Steiner et al. showed much higher sensitivity than specificity [24]. Moreover, the positive predictive power of the SDS is very low.
