Effectiveness of brief interventions to reduce alcohol intake in primary health care populations: a meta-analysis.

BACKGROUND: Earlier meta-analyses have not made a distinction between very brief (5- to 20-min) interventions and extended (several visits) brief interventions.
METHODS: Literature searches identified seven publications, comprising 14 data sets, meeting the inclusion criteria: sampling from primary care populations, random allocation to intervention and to control groups, and follow-up time 6-12 months.
RESULTS: For very brief interventions, the change in alcohol consumption was not significant among men nor among women. For extended brief interventions, the pooled effect estimate of change in alcohol intake was -51 g of alcohol per week (95% confidence interval -74, -29) among women. Among men the estimate was of similar magnitude, but significant lack of statistical homogeneity implied that the summary estimate was not meaningful. Significant statistical heterogeneity was observed when data on very brief interventions among men and women were pooled. That was the case also for gamma-glutamyltransferase activity.
CONCLUSIONS: Extended brief interventions were effective among women. Other brief interventions seem to be effective sometimes, but not always, and the average effect cannot be reliably estimated. The reasons for the lack of uniform effectiveness should be explored. Copyright 1999 American Health Foundation and Academic Press.