Does acute alcohol consumption increase risk-taking while gambling? A systematic review and meta-analysis

Aims To estimate the effect of acute alcohol consumption on risk-taking while gambling, examine blood alcohol concentration as a moderator and explore possible moderators of this effect. Design, Setting and Participants A systematic review and meta-analysis was completed. A Boolean search strategy was used to identify studies that included (a) alcohol consumption as an independent variable; (b) a gambling or risk-taking task; (c) a control or placebo comparison; (d) human participants; and (e) English publications. Descriptive information, sample characteristics and experimental data were extracted from each study. Searched databases included: PsycINFO, Web of Science, Medline, Cochrane Library and ProQuest Dissertations and Theses. Included as participants were experiments that compared the effects of alcohol and non-alcoholic or placebo beverages on risk-taking while gambling. Measurement Comprehensive Meta-Analysis version 3.3.070 was used. Standardized mean differences of risk-taking while gambling between the experimental and control conditions were calculated when studies did not report effect sizes. Random-effects models were used for overall effect and meta-regressions while mixed-effects models were used for subgroup analyses. Findings Twenty articles containing 47 alcohol versus control comparisons met inclusion criteria. The overall Hedges'g for the difference between groups consuming alcohol and groups consuming a placebo or non-alcoholic drink control was 0.03, 95% confidence interval (CI) = -0.07, 0.12, p = 0.60, indicating no significant difference. Larger effect sizes were found for studies using non-alcoholic control drinks (Hedges' g = 0.30, 95% CI = 0.01, 0.58) compared to placebo beverages (Hedges' g = -0.03, 95% CI = -0.13, 0.06), Cochran's Q = 4.67, p = 0.03. Conclusions Finding that acute alcohol consumption had no reliable effect on risk-taking while gambling was consistent with existing animal research. No support was found for the relation between alcohol dose and risk-taking. The significantly larger effect size for experiments using non-alcoholic versus placebo beverages suggests the potential role of expectancy effects.