[f5ve]

From the conclusion of the study the article cites:

> In this survey study, all types of objectively measured VI were associated with a higher dementia prevalence. As most VI is preventable, prioritizing vision health may be important for optimizing cognitive function.

In other words, the association is dementia and uncorrected visual impairment.

EDIT: I'll add that the study does not show causality. Even people like myself who are well aware that correlation != causality often still assume it does. It doesn't help that the authors themselves imply that correcting VI would ward off dementia. It might; it might not. People with dementia might just be less likely to get glasses.

[smeej]

I think people get away with that kind of thinking more easily when the intervention being proposed seems generally unobjectionable anyway.

"Correcting vision problems for older adults" sounds like a good idea in its own right, so when someone adds a "and maybe it'll reduce dementia!" thought to the equation, even though the relationship might very well not be causal, people kind of shrug out a, "Sure, that'd be nice," in reply.