Active Low-Carber Forums - View Single Post - The Atlantic diet — could this be the healthiest way to eat?

From Nina Teicholz, Unsettled Science

Quote:

For those who find the Mediterranean diet not quite to their liking, how about the “Atlantic Diet?” The Washington Post tells us that this “variation on Mediterranean eating” is "getting some buzz after a study found adherents to the diet had a significantly lower risk of chronic health problems.”

The good news: this is a randomized clinical trial.

The “Atlantic diet” is the local cuisine of Northwestern Spain and Northern Portugal. These folks eat more seafood, dairy and meat than their brethren in the Mediterranean region, but less pasta. Instead they eat “other starches such as chestnuts, potatoes and bread.”

The researchers hailed from Santiago de Compostela, in Galicia in Northwestern Spain. They randomized 250 local families to the Atlantic diet group (AD) or to a control group. They counseled the families in the AD group how to cook and eat the right foods, and the AD group had “food baskets (free of charge) delivered every 3 weeks with a variety of local foods characteristic of the traditional Atlantic diet.”

As for the controls, they got nada. No food baskets, nor lessons on how to cook. They simply “followed their habitual lifestyle.

Welcome to what is known in the clinical trial business as performance bias, which reflects different levels of care given to the intervention vs. the control groups. If you deliver food baskets every three weeks to your Atlantic diet group and nothing to the controls, you’re not only providing free food, which increases adherence to a diet, but you’re reminding the AD people that they’re enrolled in a clinical trial and should be eating healthfully. Not so for your controls. Maybe it’s these reminders that matter, not actual diet followed? We’ll never know.

This kind of performance bias is the nutritional equivalent of doing a drug study without providing a placebo to the control group.²

Self-respecting journal editors would never publish the results from such a trial, because they’d know they’re unreliable. Without a placebo, there’s no way to know if the health benefits observed were due to the food itself or the greater attention lavished on the intervention group.

This failure to have parallel interventions in both the intervention and control groups is the bad news of this trial. In scientific lingo, the trial was “poorly controlled.” Yet the researchers don’t seem to understand; the editors at JAMA Network Open don’t seem to care, nor (apparently) do the peer reviewers, and certainly not the WP reporter writing this up. The same seems true for the very influential epidemiologists whom the WP reporter contacted for comments.

Here at Unsettled Science, we’re wondering if anybody actually reads these papers anymore or cares about trial design?

But now you’re asking, surely if the Atlantic diet made subjects healthier , then that’s a good thing? Can’t we just go ahead and tell people to eat that way? Yes, but… look at the components of the Atlantic diet. You’ll need a different link to see those. The “food consumption recommendations” (shown in Table 1) included eating “sweets, pastries, cakes, candies, ice cream etc.” only “a few times per month” and “sugary drinks” only “a few times per month.” Beer seems off the menu entirely.

So, maybe the Atlantic diet reduces chronic disease risk due to the olive oil and 3-4 servings of fish and cheese every week? Or maybe just telling people to avoid sugar (and beer) does the trick. This kind of study will never tell us.

2 Our favorite example of performance bias was in the study, called PREDIMED, published in the New England Journal of Medicine in 2013, which cemented the Mediterranean diet’s haloed spot in the nutrition world. Like the Atlantic Diet trial, PREDIMED was conducted in Spain, and it also neglected to provide a placebo similar to the intervention. When the PREDIMED investigators realized this flaw, three years into the study, they tried to address the problem. As the authors wrote in the supplementary material published with their paper, “The initial dietary protocol for the Control group started with the delivery of a leaflet summarizing the recommendations to follow a low-fat diet and scheduled one yearly visit. In October 2006, 3 years into the trial, we realized that such a low-grade intervention might potentially represent a weakness of the trial and amended the protocol to include quarterly individual and group sessions with delivery of food descriptions, shopping lists, meal plans and recipes (adapted to the low-fat diet) in such a way that the intensity of the intervention was similar to that of the Mediterranean diet groups, except for the provision of supplemental foods for free.” It was appropriate of these researchers to acknowledge this problem in their study, but we take exception with their choice of words: it wasn’t that such a low-grade intervention “might potentially represent a weakness of the trial;” it was, in fact, a fundamental flaw. Ironically, PREDIMED was retracted five years later for a different reason (incomplete randomization) and simultaneously republished, with the authors claiming that correcting for their errors had no influence on their original results. The performance bias problem was never addressed. PREDIMED remains the largest and most important clinical trial ever conducted on the Mediterranean diet, and its fundamental design flaws have in no way impeded its continued 5-star status among nutrition experts.