Harvard Ran the Longest Coffee and Brain Study Ever. 43 Years. Here's What They Found

追加: 2026-05-27

[00:00:00]Harvard scientists just released results from the longest coffee and brain study ever run. 43 years, 131,821 people, and the findings are kind of wild. This was published in JAMA in February 2026.

[00:00:14]That is the Journal of the American Medical Association, one of the most trusted medical journals on the planet.

[00:00:20]The researchers tracked nurses and healthcare professionals from the mid-1980s all the way through 2023.

[00:00:28]Every 2 to 4 years, participants filled in detailed questionnaires about exactly how much coffee and tea they drank. And over those decades, 11,033 people in the study developed dementia.

[00:00:39]That is about 8% of the total. So, researchers had a huge amount of real disease data to work with. People who drank 2 to 3 cups of caffeinated coffee a day had up to 35% lower dementia risk compared to people who drank almost none. In adults under 75, that number was especially strong. One to two cups of tea daily showed the same kind of result. And here's something that surprised a lot of people. Decaffeinated coffee had no protective effect at all.

[00:01:05]In fact, people who switched to decaf were slightly worse off probably because they already had early brain changes that made them cut caffeine. The protection was coming from the caffeine itself.

[00:01:16]The protective effect held even in people carrying the APOE4 gene. That is a variant linked to higher Alzheimer's risk, which is interesting because usually genetic risk is hard to modify through lifestyle. But caffeine's effect showed up regardless of whether someone had that variant or not. Two to three cups is the sweet spot. More than that did not add any extra protection. And some evidence suggests that going way over five cups could start to cause problems. So, this is not a reason to drink more. It is a reason not to drink less if you already drink coffee in the morning. One important thing this study is observational. It tracked what people already did. It did not randomly assign coffee to half the group and water to the other. So, it cannot prove cause and effect, but 43 years of data on 131,000 people is about as strong as observational science gets. Do you drink coffee every morning? Drop how many cups below. I am genuinely curious where people land on this.