This [old] article does a good job of describing some of the problems with studies that conflate correlation and causation. I’m sure that the fact that it specifically deals with smoking will make it controversial, but it shouldn’t. The same problem exists all over the place. It doesn’t discredit science, it discredits the hucksters. It takes science out of the hands of of the activists that use science as a convenient cudgel, cleans the muck off of it, and helps us view it as the scalpel that it is.