As the Mark Twain observed: "Figures don't lie, but liars figure." You may have seen the headline recently heralding research that concluded that a third of all serious heart attacks can be attributed to the "fried and salty foods" in the "Western diet." Not so fast says Sandy Szwarc in Junk Food Science .
The analysis is classic "dredge data," Szwarc points out. The researchers excluded more than half the sample without explaining how the excluded subjects compared to those retained in the reported results. The data themselves were derived from a highly inaccurate post-heart attack dietary recall survey. And that typified a series of built-in biases that included "information bias," "selection bias," "observation bias," "recall bias and reverse causation." The analysis, too, confined to risk factors, was faulty with many key factors not controlled. She commends, as we do, a further read of the work of John P. Ioannidis, M.D., with the Institute for Clinical Research and Health Policy Studies at Tufts-New England Medical Center on "Why most published research findings are false ."
The way forward, she avers and we wholeheartedly agree, is with an evidence-based approach as pioneered by the Cochrane Collaboration. She explains how the process worked in a Cochrane Review of the role of anti-oxidants in cardiovascular disease . She doesn't mention it, but the Cochrane Review on salt and cardiovascular disease concluded there is no scientific basis for a population reduction of dietary sodium. First issued in 2003, that finding was reviewed and affirmed earlier this year.
This blog post should be required reading for every medical journal editor. Go figure.
