Predictors of study setting (primary care vs. hospital setting) among studies of the effectiveness of brief interventions among heavy alcohol users: a systematic review

Issues: The aim of this study is to compare studies by their setting in order to identify design differences between studies on brief interventions (BI) for heavy alcohol use conducted in primary care and those in hospital settings.

Approach: Potential studies were extracted from 16 reviews and from systematically searching literature up to October 2011. We assessed whether the following factors were statistically significant predictors of study setting: exclusion of very heavy/dependent drinkers; mean age of study sample; gender composition of study samples; sample size; total intervention delivery time; number of sessions; interventionist (physician vs. non-physician); various study design and intervention fidelity aspects; accounting for screening/assessment reactivity; and control condition utilised.

Key findings: Seventy-six studies (30 in primary care and 46 in hospital settings) met the inclusion criteria. The following factors were statistically significant predictors of study setting: number of sessions {odds ratio [OR] = 0.281 [95% confidence interval (CI) 0.081, 0.979; P = 0.046]}, exclusion of very heavy/dependent drinkers [OR = 0.052 (95% CI 0.004, 0.716, P = 0.027)] and gender composition of study samples [OR = 1.063 (95% CI 1.005, 1.125; P = 0.033)].

Implications: Researchers developing hospital setting BIs for excessive alcohol consumption should take into account methodological issues that could explain differences in the consistency of findings between hospital setting studies and primary care setting studies where BIs have been more consistently found effective in reducing alcohol use.

Conclusion: The observed study design differences between hospital and primary care settings might partly explain the disparity in the consistency of findings on effectiveness of BIs between these settings