Adolescent Alcohol Use and Violence: Title and subTitle BreakAre Brief Interventions the Answer?

Violence and alcohol use are risk factors for the 3 leading causes of death among individuals aged 12 to 20 years: unintentional injuries, homicide, and suicide.¹ Thus, the devastating health effects of alcohol and violence on youth lead to an appropriately overwhelming desire to intervene in clinical practice.

Brief counseling could be an answer. In adults, such interventions reduce drinking among those with nondependent unhealthy alcohol use who are identified by screening in primary care settings.² These brief counseling interventions are among the most cost-effective but least performed preventive services; are recommended by professional groups, including those that require the highest levels of evidence; and are now reimbursable services.^{2 - 3}

However, the effects of alcohol brief interventions in other settings (eg, emergency departments, hospitals), among adolescents, and on any outcomes besides consumption are less certain.^{2 ,4 - 5} The US Preventive Services Task Force has found insufficient evidence to recommend for or against screening and counseling to prevent or reduce alcohol misuse by adolescents. Even less evidence is available to support screening and intervention for violence, and results in adults are not promising.⁶

Systematic reviews of studies of alcohol brief interventions for adults in emergency departments find substantial heterogeneity in designs and results, with only about half of studies finding beneficial effects on consumption or consequences.^{4 - 5} Although level I trauma centers are required by accreditation standards to provide alcohol screening and brief intervention, the evidence for efficacy in that setting is limited. One randomized trial had substantial loss to follow-up and a nonsignificant effect on the primary outcome of reinjury. Of 3 subsequent randomized trials, 2 were negative and 1 was positive for reducing the incidence of any arrest for driving under the influence of alcohol in secondary adjusted analyses, but not in primary unadjusted analyses.^{4 - 5 ,7}

Even less convincing evidence is available for the benefits of brief interventions for alcohol use or violence among youth. For alcohol, 3 randomized trials have tested brief intervention after screening in emergency departments among young people, and results have been inconsistent. One trial found a decrease in drinking but not alcohol consequences, while another found a decrease in consequences but not in drinking.^{8 - 9} A third trial (http://www.clinicaltrials.gov; identifier: NCT00183157) has recently been completed, but the results have not yet been published. The prevention of youth violence has not been sufficiently studied in emergency settings,¹⁰ and the results of the few available studies have been mixed. One trial reported a reduction in self-reported (but not trauma registry–documented) reinjury,¹¹ suggesting that self-reported outcomes for violence may be biased in this population.

Despite the lack of evidence for the effectiveness of interventions to reduce alcohol use and violence separately, it does seem logical to address these risky behaviors together, because they so often co-occur. This was the approach taken by Walton et al in this issue of JAMA.¹² Adolescents receiving emergency services for illness or injury who reported both alcohol use and physical aggression were randomized to receive a brochure, a 35-minute motivational intervention delivered by a therapist, or a self-administered animated computer intervention.

Overall, positive results were noted for few outcomes, effect sizes were small, and none of the observed 3-month benefits were sustained at 6 months. The 24 comparisons of binary outcomes (2 intervention groups compared with controls, 2 outcome time points, 3 alcohol and 3 violence outcomes) did not show statistically significant group differences in “alcohol misuse” (ie, unhealthy alcohol use) or “binge drinking” (ie, ≥5 drinks) at 3 or 6 months; however, both intervention groups had significantly larger (by 6%-8%) decreases in alcohol-related consequences vs the control group at 6 (but not 3) months. Computer-delivered interventions did not affect any of the violence-related outcomes except being the recipient of peer violence at 6 months. Therapist counseling had a modest effect (eg, 13% absolute reduction in peer aggression compared with controls) for all 3 violence-related outcomes at 3 months but not at 6 months. Of the additional 24 comparisons of frequency-based outcomes, only 1 was statistically significant: a lower number of violence consequences at 3 months in the therapist group.

In addition to the mixed results and modest effects, the study's trial registration suggests that additional primary outcomes were measured, such as drug use, injury, delinquency, and weapon carrying,¹³ raising the concerns that type I error in the setting of multiple comparisons (7 of 48 reported comparisons were statistically significant at a 1-tailed P < .05) or a focus on more positive outcome domains might explain the results. If this study had measured more objective outcomes such as physician-documented injury events or school-based reports of violent incidents, rather than self-reported risk behaviors, the findings might have been more convincing. Although the authors used validated survey instruments, social desirability bias is of concern with self-reported outcomes, because study participants in the counseling groups might be less likely to report behaviors in follow-up that they were told were unsafe or undesirable. This might be a greater concern for study participants who received the therapist intervention than those who received the computer intervention or the brochure. Violence and alcohol consequences might be seen by adolescents as most embarrassing, and these accounted for the observed differences, whereas behaviors that were possibly perceived as less embarrassing, such as alcohol use, were not less reported in the intervention groups. Taken together, these concerns support the authors' recommendation that their study should be replicated in other settings.

Even a modest benefit of a relatively inexpensive intervention for common health problems might be worth pursuing on a large scale. However, there are barriers to the widespread implementation of screening and brief intervention programs in clinical settings. Although brief intervention for alcohol has been recommended for decades in primary care settings, levels of implementation have been dismal for a variety of reasons, such as inadequate training, lack of clinician time, and inadequate reimbursement.³ Although computerized intervention seems more likely to be disseminated successfully than a 35-minute expert therapist–delivered session, the computer intervention used in the study by Walton et al¹² affected only 1 secondary violence outcome (a decrease in being a recipient of peer violence) at 1 time point. However, the computer intervention was associated with a decrease in alcohol consequences, which is a potentially promising use for this intervention.

The most proven and effective method to reduce youth drinking, and likely alcohol-related violence, is to implement population-based strategies such as raising alcohol excise taxes and enforcing minimum legal drinking age laws.¹⁴ These strategies have been neglected. For example, the federal beer tax, which is based on a fixed amount per volume, has eroded by almost 40% in real terms since it was last adjusted in 1991. Until the findings of Walton et al¹² can be replicated (and hopefully improved), brief interventions for violence prevention in emergency departments do not seem promising. However, existing evidence supports the implementation of screening and brief intervention for unhealthy alcohol use in adult primary care settings, especially for young adults in whom the prevalence of this risky behavior is greatest.

For adolescents, alcohol screening is currently recommended by the American Academy of Pediatrics.¹⁵ While specific screening strategies for underage youth have not been fully elucidated, the focus of screening should be on any use of alcohol (with appropriate assessment of those screening positive), irrespective of their experience with violence, because any use of alcohol by adolescents, not just binge drinking, is associated with a variety of adverse outcomes.¹⁶ However, the ultimate benefit of such screening remains dependent on the development, testing, and implementation of effective clinical strategies to reduce youth alcohol consumption and violence and their adverse consequences.