Few areas of medical research are as challenging to study as nutrition. Randomized controlled nutrition trials are very difficult to conduct, and individual variation among participants can be much greater than in other areas. Add to that the urgency of the “obesity epidemic” and the multibillion dollar industry of diets, supplements and other weight-loss schemes, and it becomes clear how competing ideologies make it tough to parse the evidence.
Yet that reality is precisely why a recent piece in Scientific American by Melinda Wenner Moyer is so impressive. Moyer took on the claims of Dean Ornish, M.D., a well-known physician whose op-ed “The Myth of High-Protein Diets” appeared in The New York Times. As he has done for years, Ornish vigorously argued in favor of “very low-fat, high-carbohydrate vegetarian diets for preventing and reversing heart disease,” and he presented what seemed to be a pretty convincing case to back up his assertions.
But that’s where Moyer comes in, explaining why his list of studies has actually been cherry-picked so much he could bake a high-carbohydrate pie. She states plainly, “it’s possible to cherry-pick observational studies to support almost any nutritional argument.” After explaining how Ornish’s opening statistic misleads readers, Moyer launches into an elegant discussion of all the ways that Ornish overreaches or falls short in trying to use evidence to make his points.
For example, Moyer points out that Ornish relies on observational studies to support his claim that a diet high in protein and saturated fat causes greater mortality and chronic disease. As she notes, observational studies alone cannot show causality, except in very rare cases where the Hill Criteria are met. Using observational studies to show causality is “inappropriate” and “misleading,” Moyer explains. But she doesn’t stop there, going on to explain the reason – correlation is not causation, and confounding variables might explain the effects seen in an observational study. Then she offers up a couple of examples of better designed studies – more reliable, potentially more definitive randomized, controlled trials – that contradict Ornish’s arguments.
Her full piece is a case study in how to read about medical research claims with a skeptical eye and what kinds of weaknesses to look for.
