Alcohol risk-reduction skills training in a national fraternity: a randomized intervention trial with longitudinal intent-to-treat analysis

Abstract

Objective: The potential effectiveness of two group-administered social-skills training interventions for reducing high-risk drinking behavior was evaluated through a prospective randomized intervention trial with 3,406 members of a national college fraternity.

Method: Ninety eight of 99 chapters of a national fraternity were randomly assigned, within three strata, to receive (1) a 3-hour baseline intervention, (2) the same baseline intervention plus two booster sessions, or (3) assessments only. The current article emphasizes a rigorous intent-to-treat analysis model that compares outcomes among members assigned to receive study interventions (vs assessment-only sites) regardless of whether they actually did receive them; it also includes individuals at intervention sites even if they did not participate. This model allows us to address a social policy issue regarding the effect that introducing such an intervention may have in changing the high-risk normative drinking environment of the fraternity itself.

Results: Frequent heavy drinkers (64.2% of members) assigned to either intervention showed significant reductions at a 6-month follow-up in their frequency of drinking, heavy drinking, and drinking to intoxication; plus, they reported consuming fewer drinks overall. At 12 and 18 months postbaseline, these positive outcomes had largely dissipated. Additionally, there was an increase in drinking among lower-risk members 18 months postbaseline, which may be the result of factors other than differential attrition.

Conclusions: Findings suggest that introducing such a brief intervention can effectively reduce risky drinking behavior on a short-term basis in high-risk members of a national fraternity. Future studies may wish to focus on strategies for sustaining positive outcomes for longer, plus would benefit, in general, from learning more about mechanisms of change.