The result that caught my eye isn't the dried fruit or the cheese. It's the alcohol one. Not because it's the strongest signal - dried fruit wins that - but because of the specificity: alcohol WITH meals, not alcohol as a category. That distinction matters.

What they did

Researchers took 9 common dietary habits and asked whether any of them causally lowers stroke risk. The method is Mendelian randomization (MR), a way of getting closer to causality without a decades-long trial. Nobody is going to assign people to drink wine with dinner for 20 years. Genes do something similar for free: people carry variants that nudge them toward certain habits, and if those same variants track with lower stroke rates, you have grounds for a causal claim rather than a correlation.

5 MR methods, 3 sensitivity analyses. They pressure-tested the numbers properly.

What they found

Three of the 9 habits came back significant:

  • Alcohol with meals. Odds ratio (OR) = 0.57 (95% confidence interval: 0.37-0.79). Roughly 43% lower stroke odds for people genetically predisposed to drinking with food.
  • Dried fruit. OR = 0.51 (95% CI: 0.37-0.70). Strongest of the three.
  • Cheese. OR = 0.73 (95% CI: 0.58-0.92). Smaller, still consistent.

The other 6 habits showed nothing.

What it means

The alcohol result will attract skepticism. It should. MR leans on an assumption about genetic variants: that they don't reach stroke through some separate pathway nobody accounted for. Horizontal pleiotropy, in the jargon. That assumption gets violated more often than fans of the method like to admit. The abstract also doesn't report a sample size, so I can't gauge statistical power here. Signal worth watching. Not a prescription.

The 'with meals' piece is harder to wave away. Food slows absorption. Quantities tend to be smaller. Drinking at a table is a different act than drinking fast on an empty stomach. And the genetic signal isolates THIS context, not 'drinks alcohol' as a category, which makes the mechanistic story more credible rather than less.

If context matters causally, the practical question is whether you notice it in real time. Without food as an anchor, pace creeps. AlcoBalance shows you the slope of your BAC (blood alcohol concentration) curve while you're still at the table - where your peak is heading, before tomorrow morning turns it into a guess.

Source: Medicine, 10.1097/MD.0000000000048588