Identifying adolescents at risk for hard drug use: racial/ethnic variations.

PURPOSE: To examine early risk factors for initiation of hard drug use by 10th grade in a sample of adolescents drawn from diverse high schools and communities, compares the results across different racial/ethnic groups, and to evaluate the predictive performance of a user-friendly risk scale against the more complex logistic model.
METHODS: Using longitudinal data from 4347 adolescents from California and Oregon, we developed and cross-validated logistic and additive prediction models for non-Hispanic white students (the largest group) and assessed how well each model worked for black, Hispanic, and Asian adolescents. We also developed a best logistic model for each group. Predictor variables were measured at Grade 7; the hard drug use outcome was measured at Grade 10.
RESULTS: Major risk factors for initiation of hard drug use included early marijuana and cigarette use, deviant behavior, poor parent-child communication, being offered drugs, and prodrug attitudes and intentions. White adolescents had the most risk factors, followed by Hispanics, Asians, and Blacks. Specific risk factors played more important roles for some groups than others. Early marijuana use provided the strongest warning signal for all groups except Blacks, while exposure to drug offers increased the risk for all but Hispanic youth. Poor communication with parents was particularly important for Hispanic and Asian adolescents, whereas doing poorly in school was a key predictor only for Asians. Social influences to use drugs and intentions to use them were the only predictors for Blacks. Although family disruption and limited parental education were associated with an increase in risk for white adolescents, the latter had the opposite effect for Hispanics and Blacks. The simple additive model worked almost as well as the most complicated logistic model in predicting hard drug use for each group except Blacks.
CONCLUSIONS: These results suggest that curbing early initiation of marijuana and cigarettes and reducing prodrug influences and attitudes may dampen initiation of other substances for most youth. They also suggest that drug prevention programs need to be sensitive to differences across racial/ethnic groups and that using social background characteristics as indicators of risk can be very misleading. Carefully constructed risk scales based on simple additive models could help guide program development and provide clinicians with useful information about a troubled adolescent's likely trajectory.