The benefits of a Mediterranean diet: thoughts on the new study
by Alan Cohen
A new study in the New England Journal of Medicine claims that a Mediterranean diet (lots of nuts, fish, olive oil, and fruits and vegetables, not too much dairy, red meat etc.) can dramatically lower cardiovascular disease events and mortality. My opinions on this study are a bit schizophrenic – it confirms what I’ve been saying for a while, but I don’t trust the methods. In the end, I think the study is largely correct, but somewhat by luck.
In fact, one of the most interesting things about the study is that individuals on the “control” diet – a standard low-fat diet with counseling – were largely unable to follow the recommendations, whereas subjects on the “treatment” Mediterranean diets were much more successful. So perhaps the strongest conclusion is that behaviour changes work better toward a Mediterranean diet than towards a standard low-fat diet.
One important aspect of the study is that it uses important health events – heart attacks, strokes, death – as the benchmark, rather than intermediate measures such as weight, cholesterol, or blood pressure. I am increasingly convinced that the latter are not causes of problems, but just correlates (or, at worst, they can be aggravating factors under certain conditions, even if they are not reliably causal). So what is nice about this study is that it appears to confirm that what is important is to eat a balanced overall diet, to not be too extreme, and to not make dramatic changes that are hard to sustain. Weight loss may not be important at all if you eat well.
From what I can see, the authors did an admirable job at designing the best possible study. But diet and health is notoriously hard to study well. It is hard to force volunteers to eat something very specific for years of their lives, and even harder to do this if they are randomized (a necessary criterion to avoid the volunteers self-selecting into the diet they want to eat.) Despite the admirable efforts and some creative aspects of the study design, there are still some major weaknesses:
1) The study was done in Spain, where people are used to eating closer to a Mediterranean diet anyway. It may have been less hard for participants to adapt their diets than elsewhere, and their physiologies may have been conditioned to do better on such a diet based on what they have been eating all their lives. In this sense, a Mediterranean country was the worst place to do such a study.
2) It is not clear how successful the researchers were at convincing the control participants that they were treatment participants. There could be an important placebo effect if participants were aware of what their group was. No, I don’t think a placebo effect alone explains the differences in mortality directly. But if participants in the treatment group know they are on the diet the researchers favor, they may be more motivated to try hard to change many little aspects of diet (and exercise) which improve their outcomes for reasons other than the pure Mediterranean aspects of the diet. In this sense, the study may be better as showing that counseling a Mediterranean diet works than at showing that it is the diet itself that works.
3) Adherence to the diets was far from perfect, even if the treatment participants did better than the control participants. This raises the question of whether the effect is that a Mediterranean diet is good, or just that the standard diet of participants (who all had heart disease, and thus may not have been healthy eaters to start with) was bad. It also makes it hard to extrapolate to the effects of a a full Mediterranean diet. I would love to have seen more data on health outcomes broken down by how well participants adhered to their diets – there is just a little part of one figure that shows this, and while the trend is in the right direction, there is no significant difference between those who adhere better or worse to a Mediterranean diet. But with the right analyses it might be possible to show this.
4) This is a randomized controlled trial (RCT). While RCTs are generally considered to be the gold standard of evidence, this is not always true. In particular, RCTs generally have great internal validity (you can trust that there is little chance of an error from study design) but weak external validity (you aren’t sure how well you can extrapolate the results to other contexts). Unfortunately, because of the difficulties of studies such as this one, the researchers have mostly but not perfectly succeeded with the internal validity, but the external validity is very weak. We have no idea how well we can extrapolate these findings to people who are not late-middle-age-to-elderly Spaniards with heart disease who volunteer for studies that modify their diet.
These are just four examples of limits of this study. Again, the researchers have done a good job on a difficult topic; it is just very hard to design a very rigorous study on this kind of topic. I believe their results, but I believe this study is just one small piece of evidence in an increasingly clear overall picture.
For what it’s worth, I eat a diet not too far from the one they recommend, and it tastes great. So there’s relatively little to lose by trying it. Comments? Leave them below…