Wine Drinkers Died Less Often in a 340,000 Person Study, and the Most Likely Explanation Is Dinner

It is observational. It compares people who chose different drinks, and people who choose differently differ in many ways that no statistical adjustment fully removes. Alcohol intake was self reported at baseline and does not account for how consumption changed over the following thirteen years, which for a great many people in their fifties and sixties is a lot.

There is also a selection problem specific to this cohort. UK Biobank participants are healthier than the general population, better educated and more likely to be from higher income groups. Effects estimated in that group do not transfer cleanly to everyone, and the authors note that higher risk populations may face greater risks than these figures suggest.

And this was presented at a conference. Conference presentations have not been through the same peer review as a published paper, and a meaningful fraction of them change substantially before publication or never appear at all.

The comparison group is the quiet problem

Look again at who the wine drinkers were compared with: occasional drinkers, not non drinkers.

That framing sidesteps the oldest known trap in alcohol epidemiology. The abstainer group in these studies has historically been contaminated by people who stopped drinking because they were already ill, which makes any level of drinking look protective by comparison. Modern analyses that separate lifelong abstainers from former drinkers tend to shrink or erase the apparent benefit of moderate consumption entirely.

This study's design is better than the old ones on that point. It still does not license the sentence that a lot of readers want, which is that they should start drinking wine.

What is reasonable to take from it

Three things, none of them a licence.

If you do not drink, nothing here is a reason to start. Major health bodies now hold that there is no risk free level of alcohol consumption, and a 21 per cent relative difference between two groups of drinkers says nothing about whether either group is better off than someone who drinks nothing.

If you do drink, the pattern appears to matter as much as the quantity, and drinking with food rather than without it is the cheapest change available. That conclusion does not depend on the polyphenol argument being right.

And the finding fits a shape we keep seeing. Studies of single foods and single drinks tend to resolve, on inspection, into studies of the diets and lives they sit inside, which is the same conclusion we reached about why the Mediterranean diet keeps winning and about the difficulty of isolating any single ingredient in the ultra processed food debate. The wine is rarely the variable. The dinner it came with usually is.