February 28th, 2013

The Mediterranean Diet in Clinical Practice: Three Experts Weigh In

, , and

This week the New England Journal of Medicine published the PREDIMED study, a large randomized trial showing that a Mediterranean diet can reduce the risk for cardiovascular disease (for study details, see CardioExchange’s news coverage.) Associate Editor John Ryan has asked three nutrition experts — Walter Willett, Arthur Agatston, and Alice Lichtenstein —  to comment on the study’s findings.

Will the PREDIMED findings have a big effect on practice?

Walter Willett, MD: I hope so. Many practitioners have not given enough emphasis to diet for prevention and management of cardiovascular disease.  It is good to keep in mind that the effect of this diet was comparable to that of statins, and there are many beneficial side effects compared with statins, which can cause diabetes, among other side effects.  Of course, the Mediterranean diet and statins are not mutually exclusive approaches.

Arthur Agatston, MD: I hope this trial will encourage physicians and the general public to embrace the principles of the Mediterranean diet. Prospective diet-intervention trials are naturally difficult to perform in the real world because subjects cannot be blinded to what they eat and confounding variables are common. In the PREDIMED study, the interventions were simple, making compliance practical and limiting the potential effect of confounders.

I believe that the totality of evidence has favored the Mediterranean diet for some time. Large observational studies, small interventional studies (e.g., Lyon Heart Study) and studies of individual elements of the Mediterranean diet such as olive oil, fish, nuts, fruits, and vegetables have been concordant in their positive results. The nutrient-rich and antioxidant content of the diet give it biologic plausibility.

Diets recommended to the general public should be based on principles that stand the test of time. That means relying on traditional diets not associated with the chronic diseases that are so common today in the Western world. High-fat Eskimo diets, medium-fat traditional Asian diets, and low-fat sub-Saharan diets were all marked by an absence of Western chronic disease among people who survived starvation, infection, or violent death and who lived to an age when today’s chronic diseases typically begin.

Unintended experiments — such as America’s low-fat, high-carbohydrate movement that ushered in low-fat, processed carbohydrate foods with no precedent in traditional diets — turned out to play an important role in today’s obesity and diabetes epidemics. It is just not feasible to perform trials of the efficacy of truly new diets, as they would have to last too long and blinding is impossible. The Mediterranean diet, which has stood the test of time, has the advantage of being palatable and, thus, being adopted as a lifestyle with sustained positive outcomes.

Alice Lichtenstein, DSc: This study, which focuses on dietary patterns and hard endpoints, has strong statistical power because of its number of subjects and length of follow-up. But it’s also important to consider that it was initially promoted to test the Mediterranean dietary pattern (MetDiet), plus either extra virgin olive oil or nuts, and a low-fat diet. However, all diets were what would currently be considered high-fat: 41% of energy from fat for the MetDiet plus extra virgin olive oil, 42% of energy from fat for the MetDiet plus nuts, and 37% of energy from fat for what turned out to be a habitual diet. The study confirms what we have known for more than a decade — namely, that the total fat content of the diet has little effect on cardiovascular outcomes and that the important variable is the type of fat. Dietary guidelines from the AHA, DHHS, and others have already shifted from endorsing low-fat diets to moderate-fat diets.

How does the trial change your views?

Walter Willett: It doesn’t change my views. I believe there was already sufficient evidence for the benefits of the Mediterranean diet.

Arthur Agatston: The early separation of events between the recipients of the Mediterranean interventions and the controls was impressive and consistent with findings from the secondary-prevention Lyon Heart Study. Therefore, the Mediterranean diet’s effects may be anti-inflammatory rather than just antiatherogenic. That possibility tends to make me more aggressive in recommending the diet for both short- and long-term benefits.

Alice Lichtenstein: In this study, extra virgin olive oil and nuts each had a similar, beneficial effect when integrated with a guideline-supported Mediterranean diet — which is rich in fruits, vegetables, fish (particularly oil fish), and legumes, and also limited in sugar-sweetened beverages, meat, and baked goods. Therefore, the study’s findings cannot be attributed to unique compounds in extra virgin olive oil per se. As the authors note, walnuts are rich polyunsaturated fatty acids (PUFAs), including alpha-linolenic acid (ALA). Soybean oil, too, is high in PUFA and ALA. The other two nuts, hazelnuts and almonds, are high in monounsaturated fatty acids (MUFAs), as is olive oil. We don’t know from this study whether other vegetable oils or foods rich in PUFAs or MUFAs, when integrated with a characteristically Mediterranean diet, would have had the same benefit. It is likely they would.

Should we recommend the Mediterranean diet to patients?

Walter Willett: Certainly. Moreover, we now know enough about the elements of the Mediterranean diet, including abundant fruits and vegetables, healthy fats, whole grains, and limited intakes of red meat and potatoes, that we can combine these in many ways with many flavors to create a variety of healthy meals.  This diet stands as the gold standard.  One of the real advantages of the Mediterranean diet is that it is enjoyable and offers great variety, so people are able to stay with it for many years, in contrast to most more restrictive diets.

Arthur Agatston: We should be recommending the Mediterranean diet because of its record of efficacy and excellent compliance, but I vary my particular approach across patient subgroups. For patients with an atherogenic lipid profile (high triglycerides, low HDL, small LDL particles, high insulin levels), I am very aggressive about fairly strict diet recommendations and consultation with a nutritionist. For patients with the apolipoprotein E 3/4 genotype, I insist more on avoiding saturated fat; for those with an apoE 2/3 genotype, I emphasize avoiding high-glycemic carbohydrates. Further progress in genomics and other advanced testing will enable us to refine our recommendations further.

Alice Lichtenstein: Given that the study participants did not gain weight, we can assume they were not supplementing their Mediterranean diet with extra virgin olive oil or nuts but, rather, replacing an equivalent number of calories from other foods. Presumably, as part of the dietary intervention, the participants were counseled to substitute, rather than add, the extra virgin olive oil or nuts. It is important to note that any diet exceeding a person’s energy needs will result in weight gain. Hence, the results of this study do not provide license for people to start snacking on nuts — or adding nuts to salads, yogurt, and so on — without removing something with an equivalent number of calories from one’s diet. The same goes for extra virgin olive oil: If intake is increased, it should be at the expense of something with a similar number of calories, preferably a fat high in saturated fatty acids, such as meat and dairy fat.

Should this trial be replicated before full translation of the results?

Walter Willett: There is no need to wait. There are layers of evidence, including controlled feeding studies with metabolic outcomes and prospective epidemiologic studies, that support the benefit of this diet (e.g., Circulation 2009;119:1093-1100).  Of course, more research is needed to better understand the basis of the benefits, but we don’t need to defer the benefits in the meantime.

Arthur Agatston: Given the totality of evidence in favor of the Mediterranean diet and the problems with alternative diets, such as lack of evidence and/or poor compliance, it should be our diet of first choice for the general population at this time. Fine-tuning for subgroups, as I mentioned above, is appropriate. As we learn more from this trial and others, diet prescriptions will become more precise. For example, the subgroup with dyslipidemia seemed to do much better than the subgroup without dyslipidemia. Fortunately, no groups appeared to have been harmed.

7 Responses to “The Mediterranean Diet in Clinical Practice: Three Experts Weigh In”

  1. Jean-Pierre Usdin, MD says:


    Who can believe that more than 2 500 subjects were able to absorb one liter of olive oil every week and during about 5 years!? i.e. #250 liters of olive oil!
    I agree with dr Agatston “diet intervention trials are difficult to conduct”
    A new style of publications is arising: the Diet Studies. Is it due to possible conflicts of interests with manufacturers of medical products?

    Calcium and Vitamin D or Calcium alone: beware of cardio vascular diseases in women (and also in men)
    Coffees prevent cardiac failure.
    Chocolate (Swiss made) offer Nobel prizes.

    These studies are not only difficult to conduct but so difficult to believe.

    It is already very hard to convince, in the real life, the patient to follow low cholesterol, low caloric, low salt diets etc. during one year long…

    Yes hard to conduct, to believe and to adhere.

  2. Louis Krut, MB.ChB. MD says:

    What on earth are our experts talking about. We have an end point of only cardiac events, not survival, which is the only relevant measure from the patient’s point of view. The end point in the control group affected 1.46% of that population. In the group with extra-virgin olive oil the end point affected 1.29%, and in the group eating nuts 1.11% were affected. The orders of “benefit” were 0.17% and 0.35% respectively. Surely it is time to stop our persistent self delusion and see what we do for what it is. These orders of difference are devoid of biological meaning. Our experts, like the rest of us, should acknowledge that.

  3. Mark Perlroth Mark, md says:

    Agree with Dr.Krut!.

    “Low-fat” diet was not really low fat (which should have been 10% fat).

    Differences in outcome were limited to stroke (great!) but exceedingly small.

    By all means enjoy the diet (although fish in my neighborhood groceries are twice as expensive as beef) but aspirin will also ward off cancer as well as affecting the incidence of MI.


    • Dan Hackam, BSc, MD, PhD, FRCPC says:

      The trial was powered to the primary endpoint, which like virtually all CV trials (and other trials in other domains) was a composite. You cannot infer anything about the secondary endpoints as they were all underpowered.

      Regarding criticism that the low fat diet was not low fat enough — consuming lipid is essential to satiety and I recommend plant-based oils and fats. Unfortunately, it is very difficult to adhere in the long-run to truly fat-free or ultra low fat diets, and an experiment conducted on polar explorers in the 1920s in which they were maintained at Bellevue Hospital in fat-free conditions for 12 months suggested instantiation of several severe health problems. Don’t forget that the brain is at least 60% lipid – there is a reason we need fat. Nerves are also lipid-rich.

      Regarding the fact that the absolute risk reduction was very small, the authors show that the event rates were far lower than they had anticipated. Gauging that relative risk reduction (about -30%) to a North American or Northern European population where vascular events tend to be much higher (than in the Mediterranean basin) will undoubtedly yield much more ‘juicy’ ARRs (assuming the diet is ‘transferrable’, and I don’t see why not). Also don’t forget that clinical trial participants as a whole tend to be much healthier than the patients we see in clinic, and therefore have low baseline risk to begin with (which for any given RRR will produce much lower ARR’s). Therefore I wouldn’t take much away from the ARR/NNT in this study.

      Regarding survival – how many studies in primary prevention have demonstrated improvements in all cause mortality within this time frame? But you eat a diet and adopt a lifestyle for decades. Don’t worry. If you are seeing reductions in stroke and composite CV events then you are inevitably and eventually going to affect mortality in the long-term. Don’t people die from fatal stroke and fatal MI? Unfortunately all cause mortality in a primary prevention population studied for less than a decade is going to be a melange of contaminating causes such as car accidents, suicides, malignancies, etc that have nothing to do with the intervention under study (although the Lyon trial did show a very large reduction in malignancies as well).

  4. Susan Rhodus, R.Ph., Pharmacist says:

    I agree that this is not a low fat diet for any of the subjects. Are there studies showing cardiac changes for people eating a 10% fat diet or a 20% fat diet versus the diets eaten here?

    • Dan Hackam, BSc, MD, PhD, FRCPC says:

      Two such trials to my knowledge are the STARS – St Thomas Atherosclerosis Research Study RCT published in The Lancet and the trial by Dean Ornish published in The Lancet and JAMA some years ago. In STARS, a low saturated fat diet caused regression of atherosclerosis and a steep reduction in cardiovascular events. In the Ornish trial, randomization to a group consisting of chef-cooked vegetarian meals, social and psychological support (including group therapy), smoking cessation, etc, reduced cardiovascular events and regressed coronary atherosclerosis compared with a control group that did not receive any of these interventions. So yes there are some data suggesting that ultra-low fat diets do regress events.

      There is also a cohort followed by Caldwell Esselstyn at the Cleveland Clinic of patients with coronary disease who have been treated with veganism for about 25 years and have a staggeringly low rate of coronary events (and plaque remission when tested with repeat coronary angiography). However, it is unclear to me what happened to drop-outs in this study or whether there are true historical controls or are we just seeing a survivor effect of the most adherent type patients (healthy adherer bias).

      Finally, the palatability and tolerability, as well as the long-term health consequences of extremely low fat diets is not at all clear to me. In particular, the effects of membrane lipid depletion in nerves and other neural tissues is unclear. But there could be more than one type of optimal diet just as there are several types of optimal medications for atherosclerosis. This is the burgeoning field of nutrinomics.

  5. Tina Dobsevage, MD says:

    Ames of the Ames test for carcinogenicity of pesticides, ate a Mediterranean diet and recommended it for others as the healthiest approach. His aim was to increase the availability of fresh fruits and vegetables to the largest number of people possible.