I think reading the entire article is important here. My initial response was that the headline writer was sloppy about risk, correlation and causation, but to

I think reading the entire article is important here. My initial response was that the headline writer was sloppy about risk, correlation and causation, but towards the end it notes some studies where one or more nights of poor sleep appeared to cause measurable increases in substances that are present in elevated levels in dementia patients. If those studies did a reasonable job of showing causation, then this could be an extension of that work.