A daily drink looked healthy. Until they fixed the comparison.

Meta-analysis of cohort studies. 4,838,825 people. 107 studies.

Many studies, added up. People were watched, not told what to do. A clue, not proof.

Who: 107 cohort studies of alcohol intake and all-cause mortality published 1980 to July 2021; 4,838,825 participants, 425,564 deaths; compared against lifetime nondrinkers

• Light drinkers, raw data-15%
• After fixing the flawsno clear benefit
• Flawed 'non-drinker' groups86 of 107

More to know, from our read of the paper:

• Observational cohorts only: associations, not causation
• Alcohol intake was mostly self-reported and usually measured at only 1 point in time; the authors say self-reported intake is underreported
• Even 'lifetime abstainers' can be misclassified: several studies found most self-reported lifetime abstainers were in fact former drinkers
• Too few studies controlled for binge drinking to analyse it
• High heterogeneity across studies (I2 all >85.87%)

107 studies, 4.8 million people. At first, light drinkers looked 15% less likely to die than non-drinkers. But 86 of the 107 studies put former or occasional drinkers in the 'non-drinker' group, and some had quit because they were already sick. After fixing that, the benefit was no longer clear. Note: the study looked at alcohol overall, not wine separately.

We read the abstract and the full text of this paper and checked every number on this page against it (15 facts, checked on 2026-09-26). The study design is named above. Cohort and other observational studies are described as linked, not proven. Found an error? Email theis@onestudyaday.com.