Linking Nutrition to Health Part 1: The Mediterranean Diet
What are we supposed to think now about our precious Mediterranean diet? Somebody took the randomized out of our favorite randomized controlled trial. Yesterday, authors of the landmark 2013 PREDIMED study retracted it from the New England Journal of Medicine. They did this because of problems with the randomization.
In its place, they published a new analysis with a softer conclusion than the original paper.
Prodded to Take a Closer Look
Five years ago, this study was a big deal. That’s because long-term RCTs of dietary prescriptions with hard outcomes are so rare. That study design allowed the authors to legitimately (we all thought) say that a Mediterranean diet reduces the incidence of major cardiovascular events. No equivocation. Just a simple claim of cause and effect.
But subsequently, this study became caught up in a sweeping analysis of randomization problems by a British anesthesiologist, John Carlisle. He checked thousands of papers in eight journals and found that two percent might have problems. He flagged 11 papers in NEJM and the PREDIMED study was one of them.
No More Cause and Effect
So the original authors audited the study’s data and execution. And they found problems. For about 14 percent of the study’s 7,447 subjects, assignment to a treatment group was not random. Taking those problems into account, they still found lower rate of cardiovascular events for people in the Mediterranean diet groups (compared to a lower-fat diet).
But the bold claim of cause and effect evaporated. The original conclusions said the diet “reduced the incidence” of cardiovascular events. The new conclusion says “the incidence was lower.” This difference might seem small. But it’s the difference between causality and association.
How compelling is the study now? That depends on whom you ask. Stanford statistics professor Bradley Elfron says the revised results would not convince him to follow a Mediterranean diet. Cardiologist Steven Nissen feels reassured that the conclusions are correct. NEJM says that the findings are still strong, but not as strong as findings from a flawlessly randomized trial would be.
We like the perspective that David Allison, dean of the IU School of Public Health in Bloomington, offers:
I don’t know anybody who would turn around from this and say, “Now that this has been revealed, we should all eat cotton candy and turn away from the Mediterranean diet.”
But the degree to which empirical data are convincing of a proposition is, in most cases, subjective.
Click here for the new analysis and here for the retraction. For more perspective, click here, here, and here.
And stay tuned tomorrow for Part 2 on linking nutrition to health. Does overeating cause obesity? Or does obesity cause overeating?
Still Life with Jar of Olives, painting by Jean-Baptiste-Simeon Chardin / WikiArt
Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.
June 14, 2018