Seventh-grade cigarette, alcohol, and marijuana use: distribution in a north central U.S. metropolitan population.

Tobacco, alcohol, and marijuana use were examined in a seventh-grade population from a north central U.S. metropolitan community. Participation exceeded 94% and biological testing was used to increase the disclosure of drug use during the survey. Drug use was related to age, gender, race, family structure, and parental occupation in a series of logit analyses to identify high-risk groups. While univariate analyses uniformly supported earlier studies, the logit analyses revealed patterns not previously reported and at times contrary to previous reports. Significant associations were observed between drug use and each of the demographic factors; the likelihood of drug use increased by as much as 80-fold in subgroups defined by interactions among these variables. Native Americans, Hispanics, Black females, and adolescents whose mothers held white-collar jobs were substantially more likely to report drug use compared to other groups; females were generally similar to males in their level of use after adjustment for other factors. These results suggest that multivariate analyses which consider higher-order interactions may more adequately model the distribution of drug use in adolescent populations, compared to those based on univariate or first-order multivariate techniques.