Sensitivity and specificity of the CRAFFT to identify heavy cannabis use: Evidence from a large statewide adolescent sample

Background and Aim: Screening adolescents at risk for cannabis use disorders is critical. The CRAFFT is a screening tool designed to address both alcohol and drug use among youth. Current study tests the psychometric properties of the CRAFFT and attempts to modify one of the screening items to compare the efficiency. Design: We examined the ideal cut-off point of the CRAFFT for identifying persons with heavy cannabis use and compared the utility of the original and revised CRAFFT. Participants: Sample (N = 132,555) averaged 16.19 (±1.21) years of age; 51.0% were female, 59.7% were White, 15.2% were Latino/Latina, and 6.7% were African-American. Majority resided in non-rural area and 34.5% were receiving free or reduced lunch at school. Measurements: Heavy cannabis use was defined as using cannabis 10 or more times in the past 30 days. Sensitivity, specificity, the area under the receiver operating characteristic curve, and Youden value were analyzed to determine the ideal cut-off point. Findings: Maximum overall predictive accuracy was at a cutoff score of 2 or higher when using the original CRAFFT questions. At a cutoff score of 2, sensitivity was 82.0%, specificity was 83.7%, with an AUC of 0.880. On the contrary, when an alternative CAR question was used, maximum predictive accuracy was at a cutoff score of 1 or higher when predicting heavy cannabis use. At a cutoff score of 1, sensitivity was 92.7%, specificity was 75.5%, with an AUC of 0.900. Conclusions: The results provide evidence that the CRAFFT is a promising brief diagnostic instrument for heavy cannabis use among youth. Modification to Car item may have potential in reducing disparities in sensitivity among different racial ethnic groups, as well as those who with low socioeconomic status.