One meta-theory I had is that the caffeine has been determined to help with headaches only because studies done on it are for people who are ill and need pain management, and ill people tend to change their coffee habits.
When I'm healthy and working, I regularly drink about three espresso shots a day. If I'm sick, that plummets to zero while I'm lying in bed and feeling miserable. Invariably, I'll get a killer headache, which the paracetamol alone doesn't fix. I know this, and either I just ignore the headache or make myself some instant coffee at home.
But a badly run scientific study would not control for this quantifiable effect: Sick people get headaches, and dosing them with caffeine reduces the headaches! Ergo... caffeine reduces headaches.
However, I very much doubt that caffeine has a headache reducing effect in sick people that aren't also having caffeine withdrawal.
Imagine a world where most office workers would regularly partake in, say, Opiates to "take the edge off" their stressful jobs. In that world, researchers would confidently conclude the OTC pain relief medication should include Naloxone!
This isn't even that outlandish a scenario. There are countries where people living at high altitude habitually chew Coca leaves to counteract altitude sickness. Should OTC pain killers in those countries include Propranolol? Is it right to force tourists wanting ordinary Paracetamol to also take drugs for acute Cocaine withdrawal just because most locals feel better if they get it?
It's absurd to mix drugs like this, if you think about it a little...
Also, IIRC, caffeine specifically has been identified to help with migraines, not headaches more generally.