Primary Prevention of Cardiovascular Disease with a Mediterranean Diet. Ramon Estruch et al. The New England Journal of Medicine 2013;368:1279-90
This paper was presented by Ramon Estruch (MD, PhD) from Barcelona at the Cardiovascular Research Trust Symposium on Diet, Active Lifestyle and Cardiovascular Health. I was particularly interested to read it closely as my talk on the Lifestyle Heart Trial followed immediately on from his. I wanted to know what messages the audience would have received by the time I came to speak.
I am told that this study cost 10 million euros and was funded largely by the Spanish government. This is relevant when you consider that two of Spain’s major exports are extra-virgin olive oil and almonds. It is very much in their interest that the world considers these foods to be healthy. I ask you to bear this in mind as we consider the design of this study, and its results.
The study claims to be a randomised controlled trial of a dietary intervention for the primary prevention of cardiovascular disease (mainly heart attacks and strokes). It enrolled high-risk, but as yet “healthy”, volunteers and sought to determine whether or not the prescribed intervention could prevent cardiovascular disease from occurring in the first instance.
In fact all those enrolled had either type 2 diabetes or three or more of the following risk factors: smoking, high blood pressure, high LDL cholesterol (the bad stuff), low HDL cholesterol (the good stuff), overweight or obesity, or a family history of premature coronary heart disease. None of them were fit, healthy and free of diet-related disease at the time of enrollment. This was most definitely a secondary prevention trial, seeking to identify those with early disease and prevent further morbidity or mortality. Primary prevention for cardiovascular disease needs to begin in childhood, as this is when the pathological changes begin.
7447 persons were enrolled, aged between 55 and 80 years. They were randomly assigned to one of three groups: mediterranean diet + extra-virgin olive oil (EVOO), mediterranean diet + mixed nuts, and a low-fat control group. EVOO and mixed nuts were provided by various corporate sponsors. The EVOO group received 1 litre of EVOO per week and the mixed nut group received 30g of mixed nuts per day (15g walnuts, 7.5g almonds, 7.5g hazelnuts), both with instructions to consume these in addition to a mediterranean diet.
Dietary analysis revealed that the total fat consumption as a percentage of calories was 39% for all three groups at baseline. At the end of the trial it had increased to 41% for both experimental groups, and fallen to 37% for the “low-fat” control group. 37% of calories is not a low-fat diet by anyone’s estimation, indicating that the control intervention was not successful. Saturated fatty acids fell from 10% to 9% of calories for all three groups, while cholesterol intake (mg/day) also fell in all three groups, from 363 to 339 for EVOO, from 367 to 338 in mixed nuts, and from 356 to 324 in the controls. Bear in mind that the recommendations from the US National Cholesterol Education Programme (NCEP) are for no more than 7% saturated fat and no more than 200mg per day of cholesterol.
By the way, 1 litre of EVOO contains 8181 calories, or about four days worth of calories for an average woman (a bit less for men). 100% of these calories come from fat: 14.3% saturates, 73.0% mono-unsaturates and 8.2% polyunsaturates. A weeks worth of mixed nuts (210g) would, if they were all walnuts (I’m looking at my walnut packet here for a rough estimate), contain 1462 calories, of which 68.5 would come from fat: 5.6% saturates, 12.4% mono-unsaturates, 47.5% polyunsaturates.
The “low-fat” control group received small non-food gifts, along with instructions to reduce the amount of fat in their diet. They saw a dietician at enrollment and thereafter, during the first three years of the trial, they received an annual leaflet explaining the low-fat diet. Meanwhile those in the experimental groups received dietician-delivered individual and group training sessions at baseline and every three months thereafter. Three years into the trial its organisers realised that the difference in the amount of personalised contact received by the experimental and control groups might be a limitation of the study, and after that the control group received an equivalent level of dietician input.
An interim analysis was carried out after a median follow-up of 4.8 years, and the trial was stopped as a statistically significant difference had been found between the experimental and control groups. In total there were 288 primary end-point cardiovascular events: 96 (3.8% of participants) in the EVOO group (hazard ratio 0.70, 95% CI 0.54-0.92) and 83 (3.4%) in the mixed nut group (hazard ratio 0.72, 95% CI 0.54-0.96), vs. 109 (4.4%) in the low-fat control group.
The study concluded that among persons at high cardiovascular risk, a mediterranean diet supplemented with EVOO or nuts reduced the incidence of major cardiovascular events. But let’s just consider this statement in more detail.
What was in fact found was that all groups went on to experience multiple cardiovascular events, so it could be concluded that none of these diet-interventions was successful in preventing progression to clinically significant disease. All study groups increased their use of all pharmacological agents monitored (blood pressure drugs, lipid-lowering drugs, anti-platelet agents, insulin and oral anti-diabetic agents) between years 3 and 5 of follow-up.
A difference in actual incidence of cardiovascular end-points of 1% between the mixed nut group and the control group may be statistically significant, but may not be clinically significant. Would you go to all the effort of changing your diet in order to achieve a 1% reduction in risk? We can do better than that.
Of course had the study continued, the difference may have increased, but equally it may have evened out or even reversed, as there will always be some natural variability in rates over time. It was interesting that the study team decided to stop the study at a point at which a statistically significant difference in the desired direction had been observed!
It is quite possible that this difference could be explained entirely by the fact that the low-fat control group did not receive the intensive individual and group input that the experimental group did for 3 out of the 4.8 years of the study. Simply having someone interested in you and your health may well be sufficient motivation for people to make modest positive lifestyle changes, regardless of the exact nature of the advice.
Furthermore, it is not possible to say whether a greater or lesser difference would have been seen without the addition of EVOO or mixed nuts. To properly examine the effect of adding EVOO or mixed nuts to the diet, you need to do this without making any other changes. You need to keep all other factors the same for both experimental and control groups. It is no good to also make one set of dietary recommendations for the experimental group and another for the control group. You should just give them the nuts, or the olive oil, without any other stipulations, and see what happens.
Similarly if you want to test the hypothesis that the mediterranean diet is better at reducing cardiovascular risk that a low-fat diet, then you must assign participants accordingly to two groups that receive an otherwise identical intervention.
The study group has concluded that EVOO and mixed nuts are beneficial in reducing cardiovascular risk, but it is quite conceivable that they are quite the opposite. If a greater effect would have been seen had participants adhered to a mediterranean diet alone, then these additions may in fact have had a negative effect, significantly attenuating the results. On the other hand, if a large proportion of the observed difference was due to the EVOO and/or mixed nuts, then the mediterranean diet-intervention itself resulted in a relatively small reduction in overall risk.
Sadly it impossible to determine which scenario is the case from these results. Some kind of intervention resulted in a small reduction in absolute risk, certainly, but it would be impossible to replicate clinically, and unwise to derive conclusions as to exactly which element of this trial was responsible for the observed effect. It is a shame, given the huge amount of resources that went into this study, because this outcome should have been quite clear at the design stage. Even before that, it could be argued that the study was founded on limited information. In their first paragraph, the authors state that a systematic review ranked the mediterranean diet as the most likely dietary model to provide protection against coronary heart disease.
The authors of this systematic review had clearly never entertained the concept of a low-fat (less than 10% of calories) plant-based diet as a valid contender. The traditional mediterranean diet is characterised by high intakes of fruits and vegetables, alongside nuts, grains, olive oil, wine, fish and poultry. Red meat, dairy products and refined and processed foods are minimal. Much has been made of the type of dietary fat commonly present in this diet, stating that mono-unsaturates and polyunsaturates are “heart healthy”. But perhaps it is actually the fruits and vegetables that are “heart healthy”? Perhaps the incidence of cardiovascular disease would be even lower in mediterranean countries if they gave up their meat, nuts and olive oil? After all, there are many examples of peoples around the world for whom cardiovascular disease is simply not a part of their reality. People eating low-fat, plant-based diets. We need to remember this in future, and begin there.