Your link is available by subscription only. This study is not listed in Pubmed, and the USA Today article seems to be discussing a conference presentation.
As I explained in post 43, this kind of study is the worst kind of “research” possible. I’ve read hundreds of these correlational studies, and my annoyance with them never decreases. Correlation is not causation, yet these studies are used to “prove” all kinds of things, or, worse, used to “prove” a pre-conceived bias.
The worst possible?
I don't think so. It's very subject to misuse, but spotting correlations is useful, even when there is no causation, as it can point to underlying relationships and root causes.
Finding corrections is a good starting point for investigation, the real problem, as you pointed out, is that the studies are presented to the public as the be all and end all of Conclusive Science.
I am not a subscriber. Apparently I bypassed the paywall by clicking on a Google search result. That works surprisingly often.
Try a Google search (quotes included) for:
"A few earlier studies reported that coffee consumption was associated with fewer suicides,"
It might help to delete cookies before doing the search.