Psychometric properties of the Brazilian version of the drug use screening inventory

Background: Adolescent involvement with alcohol and other drugs is rising in Brazil, and there is an increasing need for psychometrically sound assessment tools to detect early drug involvement. Methods: The psychometric properties of the Brazilian version of the Drug Use Screening Inventory (DUST) were examined in a sample composed of 71 non-drug-dependent adolescents and 142 adolescents who met DSM-III-R criteria for drug dependence. Results: With a cutoff score of 13% or lower for the absolute density index of the substance use area, DUST correctly classified 80% of the drug-dependent adolescents and 90% of the non-drug-dependent adolescents, thus correctly classifying 83.6% of the sample. Factor analysis applied to each of the 10 DUST areas indicated their unidimensionality, with substantial percentages of variance on the first factor. The Brazilian version of DUST presented strong internal consistency reliability for the whole sample (drug-dependents and non-drug-dependent adolescents) with an average across all 10 scales for Cronbach's alpha reliability coefficient of 0.96 (standard deviation = 0.02) and for the split-half reliability coefficient of 0.88 (standard deviation = 0.08). Conclusions: These results suggest that the Brazilian version of DUST preserves its original psychometric properties and is a sensitive and useful screening instrument for drug use.