Abstract Body: Background: Observational studies on the cardiometabolic effects of alcohol are inconsistent. Long-term randomized trials on alcohol use are currently unavailable, and trials on increasing intake are unlikely.

Methods: We emulated a target trial using data from the UK Biobank cohort (502,178 individuals aged 40–69). Eligible participants were all stable current drinkers, irrespective of intake level, who had no myocardial infarction, revascularization procedure, stroke, or liver disease in the past six months, no history of heart failure, cancer (breast, colon, or liver), dementia, or alcohol-related conditions, and were not on antithrombotic drugs or reporting poor health. Alcohol use was assessed at baseline (2006–2010) and again during follow-up (2012–2022). The primary endpoint was time to death or first cardiovascular disease (CVD) event (myocardial infarction, ischemic stroke, angina hospitalization, or coronary revascularization). Secondary endpoints were CVD, type 2 diabetes, and alcohol-related disease or death. We used clone-censor-weighting to estimate risks of endpoints, comparing participants who were moderate drinkers (>0–16g alcohol daily or almost daily) during follow-up with those who were social drinkers (>0–16g on 1–3 days/week or only monthly), adjusting for potential baseline confounders. We chose appendicitis and rheumatoid arthritis (RA) as negative control outcomes to assess unmeasured confounding.

Results: Of the 65,594 eligible participants at baseline, 12,470 were moderate drinkers during follow-up and 53,124 were social drinkers, with 1,041 and 4,224 CVD events or deaths, respectively. Moderate drinkers had a 14% lower risk of CVD or death compared to social drinkers (adjusted risk ratio, 0.86; 95% confidence interval, 0.76–0.97). Negative control outcome results indicated similar relative risks for appendicitis (0.87; 0.70–1.08) and RA (0.80; 0.67–0.97) comparing moderate to social drinkers, indicating strong potential for unmeasured confounding. Relative risks for CVD, type 2 diabetes, and alcohol-related disease or death did not differ between the groups.

Conclusions: Our analyses suggest that moderate alcohol use was associated with a lower risk of CVD or death compared to social drinking. However, even after comprehensively accounting for potential biases, unmeasured confounding may explain this lower risk, highlighting the need for trials to assess the health impacts of reducing intake. (ClinicalTrials.gov: NCT06583161).