You act like people never did a valid causal analysis of the data in the Nurses' health study.
I know I overstated things. There are such things as natural experiments, having some causal information already, etc.
I'm not familiar with the Nurses' health study, and a quick google only turns up its conclusions. What methods did they use?