Psychometric properties and diagnostic accuracy of the DUDIT and ASSIST-Y for cannabis use in adolescents in Sweden

Aims Adolescent substance use presents significant public health concerns, leading to risk of developing substance use disorders later in life. Early identification through reliable screening tools is essential. This study evaluated the psychometric properties and diagnostic screening accuracy, for identifying cannabis use, of two widely used tools: the DUDIT and the ASSIST-Y cannabis use subscale, in a Swedish adolescent population.Methods Data were collected from 1450 randomly selected adolescents and a sample of 60 treatment-seeking adolescents, all 15 to 18 years old. Internal consistency was measured using Cronbach’s alpha, and test–retest reliability assessed with intraclass correlation coefficients (ICC). Construct validity was investigated through exploratory factor analysis, and diagnostic screening accuracy determined via receiver operating characteristic (ROC) analysis.Result The DUDIT and ASSIST-Y demonstrated good (α = 0.89) and moderate (α = 0.55) internal consistency, respectively. Test–retest reliability was good for the DUDIT (ICC = 0.88) and moderate for the ASSIST-Y (ICC = 0.68). Factor analysis indicated an ambiguous structure for both instruments, where the two instruments measure different aspects of substance use-related problems. Both instruments showed moderate to good diagnostic screening accuracy. The optimal cut-off scores for boys and girls on the DUDIT are 4 and 1, respectively, and for the ASSIST-Y, 3 for boys and 2/3 for girls.Conclusions Both scales demonstrated ambiguous construct validity and results suggest that the ASSIST-Y cannabis use subscale requires refinement, particularly concerning its internal reliability. In sum, the DUDIT performed better than the ASSIST-Y, indicating that the DUDIT is a reliable screening tool for substance use among an adolescent general population.