Investigation of Risk of Bias in Studies on Alcohol Intake and Blood Apolipoprotein Levels

Abstract Background Alcohol consumption is associated with apolipoproteins (Apos), key biomarkers for cardiovascular health. This meta-analysis assesses the effect of moderate alcohol intake (up to 40 g/day) on Apos concentration and evaluates the risk of bias in the available studies. Methods Twenty-five human intervention studies were analysed. Randomised studies (n = 15) were evaluated using the Cochrane Risk of Bias 2 (ROB 2) tool across five domains: the randomisation process, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of reported results. Non-randomised studies (n = 10) were evaluated using the Risk of Bias Assessment tool for non-randomised studies (ROBINS-I), across seven domains: bias due to confounding, participant selection, intervention classification, deviations, missing data, outcome measurement, and result selection. Each study was rated as low, moderate, high or critical risk of bias for each domain. Results Moderate alcohol intake increased ApoA-I (MD: 7.77 mg/dl; 95% CI: 4.95-10.59) and ApoA-II (MD: 1.61 mg/dl; 95% CI: 0.33-2.90) levels compared to controls. However, in randomised studies, only one was rated low risk across all domains. Most studies showed inadequate randomisation or allocation concealment. Nine crossover trials failed to address carry-over effects due to an absent or insufficient washout period. All non-randomised studies (n = 10) were at critical risk of bias, mainly from unaddressed confounding. Six studies showed critical bias due to unclear adherence, affecting the domain of deviation from the intervention. Conclusions Significant risk of bias is present in the current literature assessing the effect of moderate alcohol intake on Apos. Future research must prioritise rigorous study design, better reporting, and control for confounders to improve the quality of evidence. Key messages • Moderate alcohol intake increases ApoA-I and ApoA-II levels, but study quality compromises the validity of the evidence. • Bias-controlled studies are needed for accurate assessment.