The Wine Study Measured Wealth, Not Wine

On August 9, a team led by Zhangling Chen of Central South University presented an analysis of 340,924 UK Biobank participants at the American College of Cardiology’s ACC.26 conference. ScienceDaily ran it under the headline “Wine, beer, or spirits? Massive study finds a surprising health divide.” Over an average follow-up of more than 13 years, moderate wine drinkers showed a 21% lower risk of dying from cardiovascular disease than people who never drank or drank only occasionally. Low-to-moderate beer, cider, and spirits drinkers showed a 9% higher risk of cardiovascular death against the same comparison group.

By the weekend, that will be a wellness headline about red wine. It should not be.

This study did not find that wine protects hearts. It found the kind of person who reaches for wine, and then measured their advantages.

Here is the distinction that decides everything else in this piece. The ACC.26 analysis is an observational cohort study: researchers enrolled a few hundred thousand people, asked what they drank, and waited more than a decade to see who died and of what. Nobody assigned anyone to a beverage. Compare that to a randomized trial, where a coin flip decides who gets the wine and who gets the water — which is the only design that can separate the drink from the drinker, because the coin flip scrambles income, diet, and everything else evenly across both groups. Observational cohorts can only tell you that two things traveled together. They cannot tell you which one did the work.

And in beverage-choice research, an enormous amount of other stuff travels with wine. In the UK Biobank population, wine drinkers skew wealthier and better-educated than beer and spirits drinkers. They eat more like the Mediterranean template that independently lowers cardiovascular risk. They exercise more. They get to a GP faster and get statins earlier. Every one of those variables lowers your odds of dying of heart disease on its own, with no help from a grape. Statistical adjustment can shave the edges off that gap, but it cannot remove what nobody measured — and diet quality, wealth trajectory, and healthcare access are notoriously badly measured in a single baseline questionnaire.

The comparison group has its own problem. The reference here is people who never drank or drank only occasionally, and that group is always contaminated by former drinkers who quit because something was already wrong — a bad liver panel, a new prescription, a drinking problem that got scary. Load your baseline with sick people and every drinker in the study looks healthier by contrast. The researchers were candid about the limits: the design is observational, drinking was self-reported once at baseline and never re-checked across 13 years of life changes, and UK Biobank participants are healthier than the general British population to begin with.

What survives all that caution is the part nobody will put in a headline. Heavy drinkers in this cohort were 24% more likely to die of any cause, 36% more likely to die of cancer, and 14% more likely to die of heart disease. Those are the sturdiest numbers in the study, because they run in the direction confounding usually works against — and they don’t care what’s in the glass.

Look at where the “heavy” line actually sits: above roughly 40 grams of alcohol a day for men, about three drinks, and above 20 grams for women, about a drink and a half. That is a Tuesday for a lot of people reading this. Chen’s own framing is worth quoting: “These results come from the general population, and in certain high-risk groups, such as people with chronic diseases or cardiovascular conditions, the risks could be even higher.”

The wine finding also arrives at an awkward moment for the “actually it’s good for you” genre. Gallup reported on August 20 that 54% of U.S. adults say they drink, matching last year as the lowest figure since the poll began in 1939, down from 62% in 2023. Eighty-six percent have now heard something about recent research on alcohol’s long-term health effects, up from 79% in 2024. Fifty-one percent believe one or two drinks a day is bad for you. Americans are drinking less because they have been reading, and one conference abstract about beverage subtypes is not a reason to stop.

Eighty-six percent have now heard something about recent research on alcohol’s long-term health effects, up from 79% in 2024.

The institutions have moved the same direction. The WHO’s position is that no level of alcohol consumption is safe for our health, with alcohol classified as a Group 1 carcinogen alongside asbestos and tobacco. The U.S. Surgeon General’s advisory ties alcohol to nearly 100,000 cancer cases and about 20,000 cancer deaths a year, with breast cancer risk climbing at one drink a day, and it asked for cancer warning labels on the bottles. A 21% cardiovascular signal in an unrandomized cohort does not outweigh a causal carcinogen classification, and the study’s own 36% cancer-mortality figure for heavy drinkers is pointed the same way as the advisory.

So: nothing here should change what you do tonight. If you are sober, this study offers no argument for a glass of anything. If you are moderating, it offers no reason to trade a beer for a Malbec — the difference it measured lives in tax brackets and grocery carts, not in the bottle.

Cardiology conferences do not hand out permission slips. They hand out hazard ratios, and this one spelled out, in three separate causes of death, exactly what heavy drinking costs. The wine was never the finding. The wine drinker was.