The "moderate drinking protects the heart" message has been standard health advice for decades. Official guidelines repeated it. So did health journalism, and half the dinner tables I've sat at. This March 2026 umbrella review is the most systematic audit of those claims I've come across - 49 existing studies, 37 specific health questions. Its answer: "less is better," with no defensible safe threshold.

What they did

An umbrella review is a review of reviews, not a new trial. The team searched PubMed/MEDLINE and Scopus for systematic reviews and meta-analyses published between January 2015 and March 2026.

From 14,991 initial records they kept 49: 46 systematic reviews or meta-analyses, 2 WHO documents, 1 cross-sectional study. Those fed 37 pre-specified questions: dose-response risks for cancer and liver disease, clinical interventions, policy measures.

Each question got one anchor synthesis for the main analysis, with supporting studies kept for triangulation. Quality was checked with tools matched to each study design. Not every question came back with a confident answer. For some, the underlying evidence was thin.

What they found

Injuries. Any drinking raised injury odds by a factor of 2.80 (odds ratio). Not heavy drinking. Any drinking.

Cirrhosis. The dose-response gradient is steep, and it splits by sex. At 40 g/day - roughly 3 standard drinks - relative risk for cirrhosis came out at 9.35 in women and 2.82 in men.

Breast cancer. Light drinking, small but measurable increase: RR 1.05.

Low-dose cardiovascular claims. This is the load-bearing wall of the "moderate drinking is fine" argument. Apparent cardiovascular benefits at low doses didn't survive bias-aware analysis: once researchers corrected for the way observational studies overestimate benefit, the signal vanished.

Brief interventions. Short conversations with a clinician in primary care cut weekly consumption by 20 g at 12 months. Pricing measures lined up with lower consumption too. The rest of the policy toolkit gave mixed results.

What it means

"Less is better" isn't a new public health position. The scale is what's new here: 37 specific questions, 14,991 records screened, bias correction applied to the low-dose claims. That reads as more persuasive to me than one more meta-analysis arguing the same point.

The cirrhosis gap is striking. RR 9.35 versus 2.82 at the same 40 g/day is not a rounding error. I doubt the full explanation is metabolic - the size of that gap probably comes from several factors stacking. Either way, guidance written for "adults in general" understates the risk for women.

Policy is messier. Pricing pointed toward less harm; everything else was heterogeneous enough that I wouldn't lean on it. That's what policy research usually delivers.

The most concrete number in the whole review is the small one. A short doctor visit, 12 months later, 20 g less per week. One drink fewer every few days. Modest. But it looks real.

Source: European Journal of Internal Medicine, DOI