Low-fat diets lose luster (low-salt too!)

The front pages have been crowded with this story, blogged earlier here, and now the editorial writers are beginning to grasp the breadth of impact -- it goes WAY beyond fat. As USA Today reminds us:

Not long ago, it was generally estimated that 400,000 Americans a year die from obesity. Oops. A new study last year by the Centers for Disease Control and Prevention (CDC) found that obesity accounts for only 26,000 deaths, and that a few extra pounds might add to longevity.

"Similarly, hormone replacement therapy was once thought to protect postmenopausal women against heart attack or stroke. Subsequent studies say it doesn't.

"Yesterday's conventional wisdom is today's myth. No wonder so many are skeptical about whether any study can be believed.

"The latest surprising finding is that low-fat diets don't reduce the rate of heart disease, stroke, breast cancer or colorectal cancer, or even result in greater weight loss. That's the conclusion of a government-sponsored study published Wednesday in the Journal of the American Medical Association. It followed nearly 50,000 overweight, postmenopausal women for an average of eight years.

...

"The low-fat study only means that there's no magic bullet, which leaves an obvious if unpleasant fact: Good health comes from a balanced diet, frequent exercise and avoiding obvious risks. Family history and genes count for a lot also, regardless of diet."

It's not that all of "yesterday's conventional wisdom is today's myth." Only when the conventional wisdom isn't based on sufficiently-rigorous thinking or evidence. Here, as in the salt case, we have a very plausible theory (hypothesis) and a bunch of obvservational studies, though far from consensus the fat studies seemed less controversial than those for salt. What we lacked was a randomized trial. Now we have it. NOW we're prepared to act on the evidence or, in the case of fat, perhaps not act. This study may have cost $415 million, but if we learn the harsh lesson about prejudging before we conduct a controlled trial of the health outcomes of the proposed intervention, it will be money very well spent.