The present study aims at validating the Cannabis Abuse Screening Test (CAST) in a clinical sample of adolescent and young adult cannabis users seeking treatment. Applying a classical test theory approach using DSM-IV diagnoses as gold standard, two versions of the CAST questionnaire are compared. The sample consisted of 140 subjects aged 15-26 years (mean 18.9) recruited from two cannabis treatment centers. Gold standard diagnoses were assessed using the Adolescent Diagnostic Interview-Light. Internal structure and consistency of the CAST were assessed by principal component analysis and Cronbach's α. Optimal thresholds were defined using receiver operating characteristic analysis. Both the binary and the full test version revealed unidimensional structures with moderate to satisfactory internal consistency (α = 0.66 and 0.73). Screening properties were unsatisfactory when the CAST was compared against cannabis dependence. With regard to cannabis use disorders, both test versions yielded comparable and good sensitivity and specificity at cut-off 3 (binary: 92.2%, 66.7%) and 6 (full: 93.0%, 66.7%). Overall, the full CAST may be used for screening cannabis use disorders in clinical settings. Further research may use validation methods that do without gold standard.
Copyright © 2012 S. Karger AG, Basel.