The scientific article is about early onset dementia, not all dementia.
The scientific article uses a large sample size, but the entire population is French.
I would rewrite the title "Alcohol use disorder is the largest risk factor for early onset dementia in French population."
While there are a lot of posters here eager to point out the limits to what this study formally investigates, the inclusion of the French dimension underscores my suspicion that finding a reason to regard this study as irrelevant to one's specific circumstances is not the best way of looking at it.
It's not that those things make this completely irrelevant to you, it's that the blanket statement "Alcohol use largest risk factor for dementia" is not proven by this study, due to the earlier commentor's points.
> finding a reason to regard this study as irrelevant to one's specific circumstances
It's more about being precise about what exactly the study found. There's a real problem with legitimate studies being reported as something sensational, and then when the headline is proven to be false the public loses respect for the scientific process.
I'm the author of the root comment. This phenomenon is one of my greatest disappointments in the modern internet. More than drm, surveillance, and commercialization. The internet is the best avenue for public education we've seen, but instead it's used to sell high sodium processed food and $15/month entertainment subscriptions. The education (on the internet) has to be watered down to be entertainment, because the entire platform is becoming entertainment.