I recently ran across this question closed as “primarily opinion based.” At the time, it read:
Some people don't sleep enough due to work, school, or personal issues. Regardless of the cause, what are the long-term impacts of losing sleep?
I left a comment (now deleted as obsolete):
This question is not primarily opinion based. In my opinion! The problem is not that there are no studies about this; there are too many -- I think it’s too broad. I suggest choosing an endpoint. Or at least a category of endpoints. Cardiovascular? Psychiatric? Immunological? Plenty of data all around.
It turns out, the question at the time it was closed was actually:
Is there scientific evidence for the possibility of addiction to sleep deprivation?
This is also an area where there is plenty of interesting research, albeit more indirect and less obvious.
That happens to be a research niche that I know something about, so it was clear to me that, in the form I found it, it was not primarily opinion based. However, for many topics here, I wouldn't know either. I suspect that most of us are like that — we know a lot about a narrow bit of health, and about the rest of it we just have some general ideas. In many instances, we’re going to see specialized questions as “opinion based” simply because we don’t know what data are available.
Now I have a mod hammer (and I hear it also comes with a crowbar), so I can go around re-opening such things, but I don't know most of the relevant literature well enough to figure out what is and is not opinion-based, and we can't expect other mods or high-rep users with closing powers to know either.
How do we close questions as “opinion based” if we don't know what we don't know?
This Q&A was inspired by a post at meta.hermeneutics.SE that raised similar issues in a different context.