Jul 25, 2017 · 1 min read
So at what point do you have enough information to do something. Correlation is not causation, but, as Bradford Hill said, you don’t have causation without correlation. I prefer Rothman’s causal model, but even there, we’ve never had an RCT of smoking and lung cancer, as the tobacco industry used to delight to point out. All we have is observational studies. When do we decide we have enough information to recommend not smoking?
What about HPV and cervical carcinoma in situ? No RCTs there, either for obvious reasons. Laura Koutsky found RRs of over 100 for four different HPVs — is that causal? Can we do something, like make a vaccine?
