Predictors of drinking and driving in healthy young men: a prospective study.

This paper explores the ability of information about alcohol use and problem patterns in men's early 20s to predict drinking and driving by their early 30s. The sample consisted of 231 healthy young men who were students or nonacademic staff at a university. Subjects were evaluated initially with questionnaires and interviews, and subsequently followed-up using interviews with themselves and with a resource person. The reported rate of drinking and driving at follow-up was nearly 38%. Using logistic regression, the combination of variables that best predicted drinking and driving included prior alcohol-related auto accidents, a measure of typical drinking, and a measure of drinking-related problems. The overall prediction employing this combination was 72%, the specificity was 86%, and the sensitivity was 48%.