Three papers reported the validity of the SDS und CUDIT against DSM diagnoses in cannabis users (Table 3). Again, different cut-off points were found depending on the study sample. Sensitivity, specificity and predictive power of the SDS are comparable to those reported in general population samples [20,25]. CUDIT performed better in an alcohol-dependent sample [13] as compared to the general population study mentioned above. There are no studies evaluating possible subgroup differences (sex or age) or applying other external criteria than dependence diagnoses.
