But we are? The article describes an extremely significant effect because dementia rates are already very high at, for instance, ~10% in the 70+ age group. So twice as likely to develop dementia is an extreme effect, as opposed to e.g. twice as likely for something that occurs at a rate of 1 in 10,000 or whatever.
The bigger question would be causality. Observational studies are generally trash, especially with dementia, because it correlates with practically everything in one way or the other. So is it the drug, or is there a common confounding variable among people who use the drug?
That's the problem that makes proving causality so hard for many drugs. Long-term effects are subject to practically endless possible confounders. Like the National Academy of Sciences is rather fond of saying, you end up with endless scenarios where there is neither enough evidence to accept nor reject a possible causality. People that want to spin things one way then claim there's no evidence, which is plainly false - as there is, but its insufficient to confidently claim causality. In the same way that there's insufficient evidence to reject causality.
The bigger question would be causality. Observational studies are generally trash, especially with dementia, because it correlates with practically everything in one way or the other. So is it the drug, or is there a common confounding variable among people who use the drug?
That's the problem that makes proving causality so hard for many drugs. Long-term effects are subject to practically endless possible confounders. Like the National Academy of Sciences is rather fond of saying, you end up with endless scenarios where there is neither enough evidence to accept nor reject a possible causality. People that want to spin things one way then claim there's no evidence, which is plainly false - as there is, but its insufficient to confidently claim causality. In the same way that there's insufficient evidence to reject causality.