Improving alcohol and substance use screening in school-age children: translation, adaptation and psychometric evaluation of the CRAFFT tool for Lumasaaba, Uganda.

Background: Children at risk of substance use disorders (SUD) should be detected using brief structured tools for early intervention. This study sought to translate and adapt the Car, Relax, Alone, Forget, Family/Friends, Trouble (CRAFFT) tool to determine its diagnostic accuracy, and the optimum cut-point to identify substance use disorders (SUD) risk in Ugandan children aged 6 to 13 years.

Methods: This was a sequential mixed-methods study conducted in two phases. In the first qualitative phase, in Kampala and Mbale, the clinician-administered CRAFFT tool version 2.1 was translated into the local Lumasaaba dialect and culturally adapted through focus group discussions (FGDs) and in-depth interviews, in collaboration with the tool's authors. Expert reviews and translations by bilingual experts provided insights on linguistic comprehensibility and cultural appropriateness, while pilot testing with the target population evaluated the tool's preliminary effectiveness. In the second phase, the CRAFFT tool, adapted to Lumasaaba, was quantitatively validated against the Mini International Neuropsychiatric Interview for Children and Adolescents (MINI-KID) for diagnosing SUD in Mbale district, through a survey. Participants, chosen randomly from schools stratified according to ownership, location, and school size, were assessed for the tool's reliability and validity, including comparisons to the MINI KID as the Gold Standard for diagnosing SUD. Data were analyzed using STATA-15. Receiver-operating-characteristic analysis was performed to determine the sensitivity, specificity, and criterion validity of the CRAFFT with the MINI-KID.

Results: Of the 470 children enrolled, 2.1% (n = 10) had missing data on key variables, leaving 460 for analysis. The median age and interquartile range (IQR) was 11 (9-12) years and 56.6% were girls. A total of 116 (25.2%) children had consumed alcohol in the last twelve-month period and 7 (1.5%) had used other substances. The mean CRAFFT score for all the children (n = 460) was 0.32 (SD 0.95). The prevalence of any alcohol use disorder (2 or more positive answers on the MINI KID) in the last 12 months was 7.2% (n = 32). The Lumasaaba version of the CRAFFT tool demonstrated good internal consistency (Cronbach's α = 0.86) and inter-item correlation (Spearman correlation coefficient of 0.84 (p < 0.001). At a cut-off score of 1.00, the CRAFFT had optimal sensitivity (91%) and specificity (92%) (Area Under the Curve (AUC) 0.91; 95% CI 0.86-0.97) to screen for SUD. A total of 62 (13.5%) had CRAFFT scores of > 1.

Conclusion: The Lumasaaba version of the CRAFFT tool has sufficient sensitivity and specificity to identify school-age children at risk of SUD.