More recently a post from Retired Doc made the point better than I did:
Over thirty years ago another medical publication dealt with a purported increase in cardio vascular deaths from an oral diabetic medication. The University Group Diabetes Program (UGDP) was presented at the June 1970 meeting of the American Diabetes Association. Data presented indicated that tolbutamide increased cardiovascular deaths by over two and half times. (RR of 2.61, CI-1.29-5.27)…….
The UGDP study in many ways presented more robust evidence that the Nissen article. It was a randomized clinical trial (Yes Virginia, clinical trials were done before the folks from Canada descended from the mountain with the precepts of EBM carved into stone) and the R.R. was greater than 2, a number less than which many epidemiologists put little weight as small RR's can be pushed around by small biases. Putting too much emphasis on medical analyses which derive RR's less than 2 has been a issue of interest to me for some time and I have ranted on about that before.
It’s a shame about medical reporting in the popular media. Medical information, more often than not, is presented superficially, as a series of isolated sound bites without perspective, proportionality or regard for prior research.