Which diet is better: Mediterranean or Vegan?

Mediterranean diets are popular — even a bit romantic, evoking images of a glass of wine at sunset on the Italian coast. But do they actually work?

Our research team measured the effect of a Mediterranean diet on body weight and cholesterol in a head-to-head comparison with a low-fat vegan diet. The results were published in the Journal of the American College of Nutrition.

First, a bit of background: The term “Mediterranean diet” was coined in the 1950s by University of Minnesota researcher Ancel Keys, who modeled the diet after that of Nicotera, a southern Italian village where vegetables and fruits were in and steak, cheese and ice cream were mostly out. Olive oil was favored over butter and chicken fat, but it was not overused. Nicotera’s fat intake averaged just 23% of calories — far below that in the United States. Exercise and overall lifestyle were different, too. And indeed, heart problems were rarer than in the U.S.

Many research teams have tested variants of the Mediterranean diet. In the Prevención con Dieta Mediterránea (PREDIMED) study, Spanish researchers enrolled 7,447 participants at high cardiovascular risk, of whom 90% were overweight. The results were modest: no weight loss, and no significant reduction in cardiovascular deaths or total mortality. Only when researchers combined myocardial infarction, stroke and cardiovascular death into a composite score did benefits of the diet show up: over the follow-up period, one or more of these events occurred in 4.4% of controls, edging downward to 3.8% for those on a Mediterranean diet supplemented with extra virgin olive oil and to 3.4% for those on a Mediterranean diet supplemented with nuts. But these cardiovascular benefits were real, if small.

The PREDIMED findings echoed those of the classic 1994 Lyon Diet Heart Study: a Mediterranean diet supplemented with a special alpha-linolenic-acid-rich margarine reduced cardiac events but caused slight weight gain.

Is a Mediterranean diet really useless for weight loss? A 2016 systematic review asserted that Mediterranean diets could trim body weight by 4 kg to 10 kg over a 12-month period. However, a closer look shows that all five included studies that used specific calorie restrictions or added exercise, confounding the effects of the diet change.

Even so, the PREDIMED researchers made an important discovery. They developed a “provegetarian” scale to rate how well plant-based foods figured in their participants’ diets. It turned out that those PREDIMED participants who swapped chicken for chickpeas and had mostly plant-based diets experienced large, statistically significant reductions in cardiovascular and all-cause mortality — cutting their risk to about half that of their fish-and-yogurt-eating counterparts.

In our clinical trial, 62 people who wanted to lose weight were randomly assigned to either a Mediterranean diet or a low-fat vegan diet. After 16 weeks, participants switched to the opposite diet for another 16 weeks. In other words, the study did not test the diets in groups of similar individuals; it tested them in precisely the same individuals. Such studies are methodologically powerful, but must be done carefully, because the first diet will inevitably affect the experience of the second diet. It is critical to take order effects into account in the statistical analysis, and we did so.

The weight loss results were just like PREDIMED. That is, while the vegan diet led to a net weight loss of 6 kg, weight loss on the Mediterranean diet was 0 kg. The vegan diet lowered total cholesterol by 18.7 mg/dl and LDL cholesterol by 15.3 mg/dl, compared with 3.1 mg/dl and 0.5 mg/dl on the Mediterranean diet. The vegan diet also boosted insulin sensitivity, which the Mediterranean diet did not. To its credit, the Mediterranean diet did have an advantage for BP.

In two commentaries published on Healio, David S. Seres, MD, ScM, PNS, FASPEN, nutrition section editor for UpToDate, professor of medicine at the Institute of Human Nutrition and director of medical nutrition at Columbia University Irving Medical Center, and Heather Hutchins-Wiese PhD, RD, associate professor of dietetics and human nutrition programs at Eastern Michigan University, weighed in with questions. They asked about sustainability. Can a person actually stick with a vegan diet? The answer is clearly yes. First, in cohort studies, thousands of individuals have followed vegan diets for many years — or for life. In the Oxford cohort of the European Prospective Investigation into Cancer and Nutrition, 1,553 participants followed vegan diets long term, and their average BMI was well below that of their meat-eating counterparts. Similarly, in the Adventist Health Study-2, there were 2,731 participants who followed vegan diets. Their BMI averaged 23.6 kg/m², compared with 26.3 kg/m² for pesco-vegetarians and 28.8 kg/m² for meat-eaters. These studies not only show that many people follow vegan diets long term, but they also confirm their effects on body weight. Among clinical trials, our vegan diet studies have gone for more than 2 years; others have gone longer.

An important question is, what does “sustainable” mean for a Mediterranean diet, if all it sustains is one’s current girth and dyslipidemia? Our participants did not mince words about it. After losing weight on the vegan diet, the crossover design led them into a Mediterranean diet, which countered their weight loss and angered many. For those doing the Mediterranean diet first, the diet proved a tedious disappointment — many found the emphasis on fish and oil to be unpalatable and the absence of weight loss unacceptable. The change to a vegan diet was a relief, because it worked.

Seres also noted that the study did not attempt to control for the number of calories consumed. By design, it aimed to see how the diets — prepared by the participants themselves, just as real-life patients would — would influence consumption patterns which, in turn, influence weight. The Mediterranean pattern, rich in fish, oil and other calorie-dense foods, leads people to pack in calories, blocking weight loss unless they specifically restrain their appetites and portion sizes. A low-fat vegan diet favors high-fiber, lower-calorie foods, making portion control unnecessary. In addition, as reported in our recent metabolic study in JAMA Network Open, a vegan diet also significantly increases postprandial energy expenditure — the “after-meal burn.”

Hutchins-Wiese suggested that the weight loss on a vegan diet must have been due to the diet being “restrictive,” and that it would all even out over time. Clearly this is not the case. People following vegan diets over either the short or long run are significantly slimmer, on average, than their not-yet-vegan friends.

Both Seres and Hutchins-Wiese raised the important question of possible bias. In our study, all participants were offered weekly in-person classes. To minimize bias, the Mediterranean diet was taught by registered dietitians who believed in the diet, used it in their practice and followed it personally. We did the same for the vegan group. The endpoint analyses were conducted by a statistician who was impartial to the study hypotheses and was masked to group assignment.

Regarding bias in comparisons of meaty diets and plant-based diets, researchers find themselves in much the same position that tobacco researchers found themselves in in the 1960s. Quitting was clearly shown to be better than smoking, and all researchers came to accept that tobacco poses serious risks. This did not mean that researchers had to stop studying its effects. But it did mean they had to take steps to control bias.

In the nutrition world, it is now beyond question that meat and dairy products have serious harmful effects and that individuals who adopt low-fat vegan diets improve body weight, plasma lipids and glycemic control. But that does not mean we do not continue to do diet studies to understand more about how they work and how they can be applied, building in measures to handle bias.

The diets’ effects on cholesterol were predictable by basic physiology. Plasma cholesterol concentrations are increased by saturated fat and, to a lesser extent, dietary cholesterol, and are reduced by soluble fiber. Add up the numbers in 100-g servings of animal-derived products: chinook salmon (85 mg of cholesterol, 3.2 g of saturated fat, 0 fiber), roast beef (83 mg of cholesterol, 3.4 g of saturated fat, 0 fiber), and skinless chicken (89 mg of cholesterol, 2 g of saturated fat, 0 fiber). Vegetables, fruits, beans and grains have zero cholesterol, virtually no saturated fat and abundant fiber. In a word, animal products are expected to raise plasma cholesterol, on average, while plant products lower it, and that is exactly what studies show.

Based on the totality of evidence, it is a clinical error to prescribe a Mediterranean diet to patients for whom weight loss or lipid control are clinical goals; the diet does not work. It is also a mistake for clinicians to veto vegan diets based on the mistaken notion that such diets feel “extreme” or “restrictive.” Patients are well aware that many people are going vegan, profiting from it and sticking with it. There are simple techniques for introducing healthful diets, and the benefits reward continued adherence.

For anyone following a Mediterranean diet, it pays to top your angel hair pasta with arrabbiata sauce instead of meat sauce, and build the rest of your menu from the bounty of plant-based foods that the Mediterranean region has to offer. In other words, make your Mediterranean-themed diet vegan, and you’ll have a diet that works.

For more information:

The Physicians Committee for Responsible Medicine. 21-day vegan kickstart. https://kickstart.pcrm.org/en.

The Physicians Committee for Responsible Medicine and Unbound Medicine. Nutrition guide for clinicians. https://nutritionguide.pcrm.org/nutritionguide.

NutritionCME.org. Free, noncommercial nutrition continuing educational credits for healthcare professionals. https://www.nutritioncme.org/.

References:

Barnard ND, et al. J am Coll Nutr. 2021;doi:10.1080/07315724.2020.1869625.

de Lorgeril M, et al. Lancet. 1994;doi:10.1016/s0140-6736(94)92580-1.

Estruch R, et al. N Engl J Med. 2018;doi:10.1056/NEJMoa1800389.

Kahleova H, et al. JAMA Netw Open. 2020;doi:10.1001/jamanetworkopen.2020.25454.

Mancini JG, et al. Am J Med. 2016;doi:10.1016/j.amjmed.2015.11.028.

Martinez-Gonzalez MA, et al. Am J Clin Nutr. 2014;doi:10.3945/ajcn.113.071431.

Spencer EA, et al. Int J Obesity. 2003;doi:10.1038/sj.ijo.0802300.

Tonstad S, et al. Diabetes Care. 2009;doi:10.2337/dc08-1886.