[irthomasthomas]

That one graph from one epidemiological study is not enough to refute 60 years of data from many clinical trials. There is probably a confounding variable.