Baliunas · Diabetes care 2009 · systematic review and dose-response meta-regression of cohort studies · n=20 studies

Alcohol as a risk factor for type 2 diabetes: A systematic review and meta-analysis.

Cited 709 times in the scientific literature.

Level 2 - randomized trial

Systematic review and meta-analysis of prospective observational cohort studies.

PubMed 19875607 · doi:10.2337/dc09-0227 · record verified 2026-08-29

What was done

Authors conducted a systematic search for longitudinal cohort studies evaluating quantitative alcohol consumption and the incidence of type 2 diabetes. Meta-regression using fractional polynomials was applied to assess dose-response curves separately by sex, using lifetime abstainers as the reference group to adjust for the sick-quitter effect.

What was found

Twenty cohort studies met inclusion criteria, revealing a U-shaped risk curve across both sexes. Compared with lifetime abstainers, the lowest relative risk (RR) for type 2 diabetes occurred at 22 g/day of alcohol for men (RR 0.87, 95% CI 0.76–1.00; risk became deleterious >60 g/day, RR 1.01, 95% CI 0.71–1.44) and at 24 g/day for women (RR 0.60, 95% CI 0.52–0.69; risk became deleterious around 50 g/day, RR 1.02, 95% CI 0.83–1.26).

Why it matters

This analysis clarifies non-linear dose-response thresholds for diabetes risk, showing that apparent protective associations peak at moderate consumption (~2 drinks per day) and reverse at higher consumption levels when controlling for sick-quitter bias.

Limits

Findings rely entirely on observational cohort data subject to residual confounding (such as diet, socioeconomic status, and lifestyle factors) and self-reported alcohol intake. Total participant count, beverage-specific differences, and binge drinking patterns were not detailed in the abstract.

Cited by