Understanding Mechanisms of Change for Brief Alcohol Interventions Among Youth: Examination of Within-Session Interactions

Abstract

Objective: Alcohol brief interventions (BIs) for youth are efficacious, but effect sizes found have been modest. This article presents secondary data analyses from a randomized controlled trial of motivational interviewing-based (MI) alcohol BIs among youth in an emergency department, examining within-session predictors of alcohol outcomes at 3 months among those receiving BIs.

Method: Risky drinkers (ages 14-20) received either a computer BI (n = 277) or therapist BI (n = 278). Within-session measures were obtained via computer metrics (therapist and computer BIs) and coding of audiotapes (therapist BI only). Parallel structural equation models examined direct and indirect effects of BI components on 3-month alcohol consumption and consequences, with posttest precursors of behavior change (e.g., importance, readiness, and intentions) as potential mediators. Components examined included reasons to avoid or reduce drinking, strengths, strategies for change, benefits of change, and leisure activities (computer BI only), and MI global ratings and MI-adherent behaviors (therapist BI only).

Results: For both BIs, greater number of strategies directly predicted greater posttest scores, with posttest scores mediating the relationship between strategies and alcohol consumption at 3 months. Greater number of strengths directly predicted fewer alcohol-related consequences at 3 months for both BIs; however, strengths also indirectly predicted greater consumption, mediated by posttest scores. Greater number of reasons to avoid or reduce drinking directly predicted greater alcohol consumption for the therapist BI only.

Conclusions: These findings suggest ways to optimize the content and efficacy of alcohol BIs. Reviews of strategies such as refusal skills, protective behavioral strategies, and coping with negative affect may be particularly salient for reducing alcohol misuse.

Figures

Figure 1.

Significant prospective direct effects of components of the therapist brief intervention on alcohol outcomes. Model fit statistics: %2(393) = 648.04, p < .001; root mean square error of approximation [90% CI] = .05 [.04, .06]; comparative fit index = .92; standardized root mean square residual = .06. Notes: Coefficients represent standardized betas. Model controls forT0 precursors of behavior change on T2 precursors of behavior change, and T0 alcohol consumption on T3 alcohol consumption. T = time; T1 = intervention; T2 = posttest; T3 = 3-month follow-up. *p < .05; **p < .01; ***p < .001.

Figure 2.

Significant prospective direct effects of components of the computer brief intervention on alcohol outcomes. Model fit statistics: %2(336) = 548.99, p < .001; root mean square error of approximation [90% CI] = .05 [.04, .06]; comparative fit index = .93; standardized root mean square residual = .06. Notes: Coefficients represent standardized betas. Model controls for T0 precursors of behavior change on T2 precursors of behavior change, and T0 alcohol consumption on T3 alcohol consumption. T = time; T1 = Intervention; T2 = Posttest; T3 = 3-month follow-up. *p < .05; **p < .01; ***p < .001.