Abusive alcohol consumption among adolescents: a predictive model for maximizing early detection and responses.

OBJECTIVE: To present a predictive model of alcohol abuse among adolescents based on prevalence projections in various population subgroups.

STUDY DESIGN: Cross-sectional study.

METHODS: The sample consisted of 785 adolescents enrolled in the second year of high school in Rio de Janeiro, Brazil. Alcohol consumption was assessed using the Alcohol Use Disorder Identification Test. Socio-economic, demographic, family, individuals, and school-related variables were examined as potential predictors. The logit model was used to estimate the prevalence projections. Model fitting was examined in relation to the observed data set, and in a subset, that was generated from 200 subsamples of individuals via a bootstrap process using general fit estimators, discrimination, and calibration measures.

RESULTS: About 25.5% of the adolescents were classified as positive for alcohol abuse. Being male, being 17-19 years old, not living with mothers, presenting symptoms suggestive of binge eating, having used a strategy of weight reduction in the last 3 months, and, especially, being a victim of family violence were important predictors of abusive consumption of alcohol. While the model's prevalence projection in the absence of these features was 8%, it reaches 68% in the presence of all predictors.

CONCLUSIONS: Knowledge of predictive characteristics of alcohol abuse is essential for screening, early detection of positive cases, and establishing interventions to reduce consumption among adolescents.