This review investigated the efficacy of brief alcohol interventions (BAIs) for the reduction of alcohol consumption in patients attending primary care. The authors concluded that BAIs can be effective in reducing alcohol consumption, and their effects can last for 48 months. This was a relatively well-conducted review, and it is likely that the authors' conclusions are reliable.

To determine the efficacy of brief alcohol interventions (BAIs) for the long-term reduction of alcohol consumption and related harm in patients attending primary care for reasons other than alcohol-related problems.

The Cochrane CENTRAL Register, MEDLINE and PsycINFO were searched from inception to January 2003. In addition, the ISI Web of Science and ETOH databases were searched. The reference lists of all identified articles and the authors' own bibliographic resources were also checked for further relevant trials. There were no restrictions on language or date of publication.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs) were eligible for inclusion in the review.

Specific interventions included in the review

Studies of interventions delivered individually in a primary care setting that focused on alcohol consumption and had face-to-face contact during the initial session were eligible for inclusion. Interventions had to be defined as a 'brief intervention', a 'motivational intervention', or report the use of feedback or advice to reduce the risk of alcohol consumption. There were no restrictions on repeated interventions or reinforcement sessions. Studies conducted in a hospital ward or emergency department were excluded. The duration of the interventions ranged from 5 to 45 minutes. In most of the studies the intervention was repeated or a booster session given, or participants were offered a follow-up visit. The majority of studies had control groups that consisted of no intervention or usual care; in others it was up to 5 minutes of advice.

Participants included in the review

Studies that included out-patients who were attending primary care for reasons other than alcohol-related problems were eligible for inclusion. Studies of those actively seeking alcohol treatments, including those responding to advertisements or referred for alcohol treatment, were excluded. Studies that selected patients from registers or patient lists, or that brought people together specifically for alcohol screening, were also excluded. At least 75% of the study population had to consist of primary care patients or, if less than 75%, then subanalyses for the different patient populations were to have been performed. The participants included in the review were males and females aged from 15 to 70 years. Most of the studies excluded alcohol-dependent individuals.

Outcomes assessed in the review

The studies had to have at least one outcome related to change in alcohol intake, drinking status, health-related quality of life or functional status, laboratory markers related to alcohol use, utilisation of health care resources, or cost data to be eligible for inclusion. The primary outcome in the review was alcohol consumption. The secondary outcomes included the number of drinking days per month, usual drinking amount per occasion among women, use of health care resources, mental or physical health perception status, well-being, alcohol-related problems and mortality.

How were decisions on the relevance of primary studies made?

Two reviewers independently selected studies for inclusion. If an agreement was not reached, a third reviewer was consulted.

The authors assigned a quality score from 0 (lowest) to 18 (highest) using an instrument adapted from the Cochrane Drugs and Alcohol group. The instrument assessed randomisation, concealment of allocation, blinding in the outcome assessment, attrition during follow-up, intention-to-treat analysis, clear definition of the intervention, selection and performance bias, and the presence of a measure of intervention exposure. Two reviewers independently assessed the validity of the studies. When necessary, decisions were reached by consensus.

Two independent reviewers extracted the data. Any disagreements were resolved by consensus, with a third reviewer consulted when necessary. For studies that reported alcohol consumption or differences in alcohol consumption, data were extracted to calculate a mean net reduction in alcohol consumption (grams of ethanol per week) for the intervention group compared with the control group. When information on weekly alcohol was missing, the authors used the definition of absolute ethanol content in a standard drink used in the country where the study was conducted. Data were extracted in accordance with intention-to-treat principles. Individuals excluded from the follow-up were assigned the mean group baseline alcohol consumption value. In studies that reported change in alcohol consumption, missing individuals were rated as showing no change in alcohol consumption.

How were the studies combined?

Studies that reported alcohol consumption or differences in alcohol consumption at 6 or 12 months' follow-up, and which had confidence intervals (CIs), standard deviations, or standard errors at baseline and other follow-up time points, were combined using a random-effects model. A pooled weighted mean difference (WMD) was calculated, along with 95% CIs, by the weighted average of individual study effects using a DerSimonian and Laird weighting method. The results for the other outcomes were discussed narratively.

Publication bias was assessed using the Begg and Mazumdar adjusted rank correlation test and the Egger regression asymmetry test.

How were differences between studies investigated?

Heterogeneity was assessed using the Cochran chi-squared Q statistic and the I-squared statistic. Potential causes of heterogeneity were investigated using meta-regression models. The variables considered were: year of study publication; gender of study population; mean alcohol consumption in the brief alcohol intervention group at baseline; the length of follow-up; the response rate to follow-up; the duration of the intervention; the technique used in the intervention; and methodological quality.

Nineteen RCTs (3 of which were cluster RCTs) with a total of 5,639 participants were included in the review.

The mean methodological quality score was 9.6 (range: 5 to 14). Studies of lower quality (a score less than 10) were associated with inadequate randomisation or reporting of allocation concealment. High-quality studies (a score of 10 or greater) were more likely to show statistically significant beneficial effects of the intervention in comparison with low-quality studies.

Alcohol consumption.

When the analysis was not adjusted for drop-outs, BAIs were associated with significantly less alcohol consumption per week (10 RCTs; WMD -50 g ethanol, 95% CI: -65, -34). This corresponds to about 5 drinks, or an additional relative mean reduction of 15% in alcohol consumption in a comparison of the intervention group and control groups.

When the analysis only included trials for which intention-to-treat analysis was possible and also adjusted for drop-outs, BAIs were still associated with significantly less alcohol consumption per week, although the effect was smaller (9 RCTs; WMD -38 g, 95% CI: -51, -24).

No evidence of statistical heterogeneity between studies was found (Q=15.1, P=0.24). Heterogeneity was estimated to account for approximately one fourth of the total variance in the outcome (I-squared 25.8%). No evidence of statistical heterogeneity was found in the intention-to-treat analysis (Q=6.7, P=0.82). Heterogeneity was higher in low-quality studies than in high-quality studies. There was no evidence of publication bias, but a cumulative meta-analysis found that trials published after 1996 showed a significantly greater effect size than those published before this time.

The results were similar for studies reporting results at both 6- and 12-month follow-ups, and for males and females. Methodological quality, baseline alcohol consumption and studies conducted in the USA were found to impact on the effect size. Other variables had only a minor impact on the effect of BAIs.

Other outcomes.

The results for the other outcomes were inconclusive. Seven studies evaluated laboratory values as an indicator of alcohol use: four found improvements and three found no significant effect.

Three studies evaluated health care utilisation. One study found a lower number of medical visits in the intervention group than in the control group. One study found no significant differences between the groups in emergency department visits at 6 and 12 months, but did find a significant difference at 48 months; There were significantly less hospital days in the intervention group at 6, 12 and 48 months. One study found no significant differences between the intervention and control groups in health care utilisation during the 2 years after the BAI.

Nine studies evaluated mental or physical well-being-related outcomes. There were significant differences between the intervention and control groups on 9 of the 21 measures of mental or physical health perception status, demonstrating better quality of life for the intervention group participants. One study found a significant decrease in mortality in the intervention group when compared with the control at 36 months, but this difference was not maintained at 48 months.

One study performed a cost-benefit analysis, based on reduced hospitalisations, reduced emergency department visits, and reduced motor vehicle or criminal events compared with the overall cost of screening, assessment and intervention. The study found a cost-benefit ratio for a BAI of 4.3:1 from a medical perspective, and 39:1 from a societal perspective. The net benefits per intervention patient were US$456 (1993) from a medical perspective and US$7,780 (1993) from a societal perspective.

BAIs are effective in reducing alcohol consumption in primary care settings and their effects can last for as long as 48 months. The typical effective BAI lasts up to 15 minutes, has additional written material, and provides patients with the opportunity to make a follow-up appointment.

The authors set out a clear objective and the inclusion criteria were clearly defined in terms of participants, intervention, outcomes and study design. Several relevant sources were searched without any restrictions on language or publication date, which helps reduce the risk of missing relevant articles. The study selection, quality assessment and data extraction processes were carried out in duplicate by two independent reviewers, which also helps reduce the risk of reviewer error and bias. Adequate details of the participants and interventions were provided. Methodological quality was assessed, although the use of a non-validated quality score to rank the study quality might not have been appropriate.

Statistical homogeneity and the impact of several relevant variables were assessed using appropriate techniques. The analysis was undertaken using an intention-to-treat approach, which strengthens the results. The decision to only pool studies of alcohol consumption seemed appropriate given the heterogeneity of the other outcomes. This was a relatively well-conducted review, and it is likely that the authors' conclusions are reliable.

Practice: The authors stated that BAIs that last between 5 and 15 minutes, and are accompanied by written material and the chance for patients to make an appointment for a follow-up visit, have the potential to reduce alcohol consumption in comparison with no intervention, usual care, or interventions which are less than 5 minutes' duration.

Research: The authors stated that future research should focus on determining which components of brief interventions are the most effective in the primary care setting, and should evaluate the effects of BAIs on morbidity, mortality and quality of life-related outcomes.

Clinical Epidemiology Center; Alcohol Treatment Center.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.