the answer is "no", mostly
...Has epidemiology been pursued as a strategy for mitigating GA accidents?
So, having read a bit further and taken a bit (okay, a lot), more time to think and even calculate a bit, napkin-back style, I conclude that the answer is: No, it hasn't really been tried, but it probably isn't economically feasible in the real world.
The trouble is that, as the drug companies know well, it's real expensive to do real science to find novel interventions to old diseases. Furthermore, and this is the part I had to figure out so apologies to those who already knew, it gets even more expensive when the disease is rare.
See, if you take a real common disease, say bad breath, it's fairly economical to recruit some subjects who suffer the malady and try some simple treatment and measure its efficacy. Then, if you find an effective intervention and the price of development isn't too high you also have a real big potential market.
But if you look at a rare disease, you have two major problems: obvioulsy a smaller market, but if the disease is also short-lived and fatal, then you have another problem; how do you find subjects to do the tests? Let's say there is an infection which randomly infects only one person in a million in any given year and is always fatal. You have a drug you think might work, but we don't know how the disease is transmitted so we can't put the fix in for testing our drug candidates (anyway you can't infect them on purpose, it's always fatal). How do you test the drug? Simple; if you want a clear answer regarding efficacy of your drug you have to recruit millions of test subjects and give half the drug, half a placebo, and wait for statistics to happen, or some similar variation. With even passable luck within a couple years the outcomes of the two groups should diverge if the drug is doing anything. If you're really lucky a couple in the placebo die and none in the drug group do. If you're less lucky the numbers are inconclusive. In the worst case, the drug group does even worse than the placebo group. Only the first outcome is commercially viable.
As odd as it may seem, fatal ga accidents are rare in this context. We have a few hundred thousand active pilots in the U.S. and a few hundred a year suffer the, well.., worst outcome. So, you can see that in order to try an intervention with a hope of conclusive results, you need to recruit at least hundreds of pilots into your project, and do so with minimal self-selection. That is, If I ask you "hey want to participate in my aviation safety experiment" and you answer "great but only if it's really safe" or worse "no I don't care about safety" that's a big problem for the study because whatever the numbers in the outcomes look like people will just say "duh, you probably flushed out all the pilots at highest risk".
So, let me try to clarify with a thought experiment. I'll recruit a cohort of pilots and give half some additional training and the other half some sort of "placebo". Then wait and see what happens. Problem is that for that to work we'd need about a thousand pilots in each group and we'd have to do everything we could to minimize self-selection, aka washouts. So how do you force two thousand pilots to get spin training whether they like the idea or not? Even with unlimited funds that's a tough one. With zero dollars, it's hopeless.
Back to the drawing board...
