A systematic review published in Public Health in 2026 went through every study on alcohol and cardiovascular outcomes from January 2023 through July 2025. Two and a half years of literature. The question wasn't what those studies concluded, but how they were built.

What they did

Eight publications met the criteria. For each one the reviewers checked how the control group was assembled, whether known sources of bias were handled, and whether the authors said anything about the limits of their own design.

What they found

Of the 8 studies:

  • 5 reported protective effects of low to moderate drinking
  • 3 found none

Same field, same kind of data. The split came from methodology.

The five "positive" studies compared drinkers with non-drinkers or lifetime abstainers. That is where the bias gets in. Among people who don't drink at all there's a disproportionate share who stopped because they got sick. Stand a moderate drinker next to that group and they look healthier automatically - not because alcohol did anything, but because the comparison group was in worse shape to begin with.

One of the eight even mentioned this problem. Just one. None of them tried to measure or correct for it.

The three studies with neutral or negative results went another way:

  • they dropped non-drinkers from the comparison and worked only with drinkers of varying intensity
  • or they used Mendelian randomization (MR), a method that swaps self-reported consumption for genetic variants

MR is one of the few ways to get near causal inference in observational research. Nobody picks their genes, so a genetic tendency toward lower consumption doesn't come bundled with the rest of a person's habits. All three found no protective effect.

One claimed benefit the review calls flatly "implausible": a 40% reduction in cardiovascular events from under one glass per week, compared with non-drinkers. Under one glass. There's no known physiological mechanism that gets you there.

The authors flag one more thing. Some of these datasets could be presented as positive or negative depending on the angle, and the researchers picked which angle went into the main conclusions.

What this means

This isn't a case against particular scientists. Alcohol epidemiology has a handful of well-known traps, and most studies walk straight into them. Some probably avoid them and just never say so in the text.

Two questions worth asking of any study like this:

  • Who is being compared with whom? If the control group is lifetime abstainers, or simply "non-drinkers" lumped together, the result is shaky.
  • Was anything done about confounding? Mendelian randomization, exclusion of former drinkers who quit because of illness. If neither, confidence in a protective effect should stay low.

Or, shorter: who were they actually comparing against?

Caveats

Eight studies over 2.5 years is a thin sample to build broad conclusions on. The review didn't look at funding or conflicts of interest in the original papers. And the reviewers have a leaning of their own - people who care about rigorous methods tend to find sloppy methods everywhere. Still, a 40% benefit at under one glass per week is hard to argue for without explaining the physiology behind it, and nobody has.

Source: Public Health, doi:10.1016/j.puhe.2026.106331