Many studies have investigated the health effects of the Mediterranean diet
In a previous post I introduced the history of the Mediterranean diet, which is probably one of the most popular diets worldwide. Here I would like to describe with more detail the studies that have been performed so far to evaluate the health effects of this dietary pattern.
Let me start by saying that the Mediterranean diet is only one of the many traditional diets that has been studied by epidemiologists. Other examples include the Nordic healthy diet (Mithril et al., 2012) (sometimes called Baltic Sea diet (Kanerva et al., 2014)), the Okinawa traditional diet (Sho, 2001) and the Seventh-day Adventist diet (Orlich et al., 2013). All of these patterns have shown to be beneficial to human health, and share some common grounds with the Mediterranean diet. Indeed, these patterns share a low (or null) intake of animal fat, a high intake of vegetables, grains and, with the exception of the Adventist diet (which is 100% vegetarian), a moderate intake of fish. However, the popularity of these patterns is smaller compared to the Mediterranean diet, probably because the amount of studies that have been performed on the latter so far is far more extensive.
It is not easy to summarize all the studies that have been performed to evaluate the health effects of the Mediterranean diet. However, I will give it a try. The first studies that are related to the Mediterranean diet are the 7-country study (Keys et al., 1986) and the Lyon diet heart study (de Lorgeril et al., 1999) which were both performed several years ago. Other intervention studies, performed later, include the one performed by Esposito and her colleagues in Italy (Esposito et al., 2014), the Medi-RIVAGE (Vincent-Baudry et al., 2005), the DIRECT (Shai et al., 2008) and the PREDIMED studies (Estruch et al., 2013). Finally, a great deal of publications came from the application of the so-called Mediterranean diet score, developed by the Greek researcher Trichopoulou in 1995 (Trichopoulou et al., 1995), which was followed by another score, specifically designed for the application to children: the KIDMED index (Iaccarino Idelson, Scalfi and Valerio, 2017).
The 7-country study
Let’s start with the 7-country study, probably the most cited study about the Mediterranean diet. This intervention study included 16 cohorts in 7 countries, of which 5 were European (Finland, Italy, the Netherlands, former Yugoslavia and Greece), plus the US and Japan.
The 7-country study has, in some cases, continued with different names, depending on the specific cohort. These include the FINE study (Finland, Italy, Netherlands Elderly) (Menotti et al., 2001), the Zutphen study (Streppel et al., 2008), the SENECA study (de Groot et al., 2004) and, finally, the HALE (Healthy Aging Longitudinal Study in Europe) (Knoops et al., 2004). All of these include subjects recruited during the 7-country study and followed-up over the years.
The dietary assessment used in the 7-country study has not always been well described and was probably much less developed than it is in the studies performed nowadays. Anyway, the dietary habits were investigated by either questionnaires or by duplicate diet studies. In particular, only in the Zutphen cohort the questionnaires were administered to all participants during all follow-ups. Re-examinations were performed at 5 and 10 years when anthropometry, lifestyle, medical history, blood pressure, ECG, and laboratory tests were performed. Cardiovascular risk factors were re-examined by repeat surveys in 9 European cohorts at 25, 30, 35 and 40 years follow-up (and this will continue until a follow-up of 50 years).
Among the most relevant results obtained by the 7-country study, I can list the fact that the diet of the subjects recruited in different countries tended to differ in terms of saturated fat, serum cholesterol and the incidence of Coronary Heart Disease (CHD). Notably, the countries with higher saturated fat intakes also showed higher cholesterol levels. The average baseline saturated fat intake also correlated with CHD incidence at 5 and 10-year follow-up. Also, all-cause and coronary heart disease death rates at 15-years follow-up were low in cohorts with olive oil as the main fat (Keys et al., 1986). However, the causality of this association cannot be assumed, since this is what we defined as an “ecological association” which suffers from a potential bias called “ecological fallacy”. When this occurs, the association observed at the ecological level cannot be confirmed at the individual level.
More recently, in the Zutphen cohort it was found that fiber intake inversely predicted CHD & mortality (Streppel et al., 2008), that stable high fruit and vegetable intakes predicted a lower lung cancer incidence, that fish and flavonoid intakes predicted lower stroke risks, whereas no association was found between dietary habits and the risk of T2D.
In the SENECA & HALE cohorts the researchers used the Mediterranean diet score, although we will get back to this later (Knoops et al., 2004).
The 7-country study suffers from some important limitations, in particular, the dietary assessment was not always comprehensive: indeed, it was performed only in groups and it is not clear what aspects of the diet have been estimated. This limits our ability to assess whether the ecological associations, which are the only results available at the beginning of the study and that I mentioned before, can be confirmed at the individual level.
The Lyon Diet Heart Study
Let’s move on to another relevant study, that has assessed the effects of a Mediterranean diet: the Lyon diet heart study (de Lorgeril et al., 1999). The latter is a randomized secondary prevention trial which was aimed to test the association between a Mediterranean-type diet and a recurrent myocardial infarction (MI) risk. This study was composed by two arms: a control group and an intervention group.
At the beginning of this study, both a cardiologist and a dietician instructed the participants recruited in the intervention group about how to follow a Mediterranean-like diet (which included margarine and allowed wine in moderation with meals). Instead, the participants recruited in the control group were given no specific dietary advice.
The study was interrupted because the subjects in the intervention group had a 50 to 70% risk reduction of recurrent MI and the Mediterranean diet was recommended to all participants.
However, what might seem an exciting result, again, suffers from some important limitations. You have probably already guessed that the first one is that only the subjects in the experimental group met a dietician, which introduces a bias. In addition to this, the baseline diet was assessed only in the intervention group, and the diets of the subjects recruited in the control group were presumed to be comparable to the diets of the subjects included in the intervention group. The dietary assessment in the control group was done only at the end of the study.
Despite the design of this study was far from ideal, we can at least conclude that giving dietary advice to a group of subjects who previously suffered from myocardial infarction can sensibly reduce the recurrence of the disease.
The Mediterranean Diet Score
The discussion about the health benefits of the Mediterranean diet became mainstream when a score to assess adherence to this diet in epidemiological studies was developed by the Greek researcher Dimitrios Trichopoulos in 1995 (Trichopoulou et al., 1995). Originally described in a paper published in the BMJ, the Mediterranean Diet Score (MDS) was calculated by giving an increasing number of points to the study participants whose energy-adjusted intakes of typical Mediterranean foods (such as fruit, vegetables, legumes, nuts and grains ) were higher than the population median or when the ratio between mono vs saturated fats was higher than the median and finally when the intakes of either meat or dairy products (considered atypical Mediterranean foods) were below the population-specific median.
This score has been later applied (and continues to be applied) in a long series of observational studies where different variants of the score were applied. Many studies have adapted or modified the score by including fish as a typical Mediterranean food, or by using tertiles instead of medians as cut-off for high (or low) intakes, or simply referred to specific traditional Italian foods. Together with my colleagues at Gothenburg University, I also published some studies where we readapted and applied the MDS for the use in Scandinavian population studies, and in the IDEFICS study in children.
Another score, more specific for the application in children, was also developed and called KIDMED. So far, 58 studies have evaluated the adherence to a Mediterranean-like diet in children, and these have been summarized in a recent review (Iaccarino Idelson, Scalfi and Valerio, 2017). Most of these studies have used the KIDMED and they have frequently showed that the latter is positively associated with the level of physical activity and that the index is inversely associated with the prevalence of a sedentary behavior. However, the results for gender, age, socioeconomic status and weight status are still contradictive. Further validation of the KIDMED in terms of reproducibility and consistency with the Mediterranean diet is needed.
The results obtained by applying the MDS to several cohort studies are summarized by the systematic reviews and meta-analysis published by Francesco Sofi and his colleagues (Sofi et al., 2008, 2010, 2014) that I encourage you to read in case you want to know more about this topic. More recently, Sofi and colleagues also published an umbrella meta-analysis (Dinu et al., 2018), where they summarized and critically evaluated the results of 13 meta-analyses of observational studies which assessed the association between the MDS and 35 different outcomes. The results showed that the MDS is convincingly inversely associated with all-cause mortality, as well as with the incidence of cardiovascular and neurodegenerative diseases.
As it happens for any study, the application of the MDS also has important limitations to consider. The first one is the high variety of dietary indexes (22 different ones) which limit the comparability across studies. Also, the MDS was developed to rest pure observational associations, which need to be confirmed by intervention studies in order to withdraw firm conclusions in terms of public health. Finally, the MDS does not really measure adherence to the Mediterranean diet, since the cut-offs for high and low intakes are population-specific and the score is based on food groups, whose composition can vary from population to population.
Intervention studies (RCTs)
However, as I mentioned at the beginning of this post, several intervention studies have also been performed. These have also been assessed in the umbrella meta-analysis published by Sofi, who identified 16 meta-analyses of RCTs which had analyzed the effects of a dietary intervention with the Mediterranean diet on 26 outcomes. Among the studies described here, I can list:
- The Medi-RIVAGE (France) (Vincent-Baudry et al., 2005);
- The intervention studies performed by Esposito and colleagues (Italy)(Esposito et al., 2014);
- The DIRECT trial (Israel) (Shai et al., 2008);
- The Mediterranean eating study (USA)
- The PREDIMED (Spain) (Estruch et al., 2013);
- The ESTEEM multi-centre study about diet and pregnancy (UK) (Al Wattar et al., 2016).
Unfortunately, most of the evidence from intervention studies is either weak or null, with only a few suggestive associations (for instance with lower blood pressure and waist circumference). A suggestive evidence indicates a possible inverse association with a better glycemic control and reduced insulin resistance in comparison with a control diet.
However, it is also fair to report that, for most outcomes, only a few studies were performed. Therefore, future studies will probably tell a more detailed picture of the effects of intervention studies focused on the Mediterranean diet.
Among the studies performed, one that was particularly interesting was the DIRECT trial, which compared the effect of three diets on weight change: a low-fat diet, a low-carbohydrate diet and a Mediterranean diet. As you can see in this graph, all diets were effective at reducing body weight (Shai et al., 2008). However, it is interesting to note that the low-fat diet was the least effective, whereas the low-carbohydrate diet was the most effective during the first 6 months and then its effects became comparable to those obtained with a Mediterranean diet, where weight loss was more stable.
A discussion about the studies on the Mediterranean diet would not be complete without mentioning the biggest intervention study performed to date: the PREDIMED study.
The PREDIMED is a multicenter Randomized Clinical Trial (RCT) carried out in Spain between 2003 and 2011, which enrolled over 7,000 participants (aged 67 years on average) and who were at high risk for cardiovascular disease (meaning they had high blood lipids, T2D, or other cardiovascular risk factors), but no history for cardiovascular disease.
The PREDIMED was designed with three arms: an arm for the Mediterranean Diet supplemented with extra-virgin olive oil, a second arm for the Mediterranean Diet supplemented with nuts and finally a third arm where a control diet was tested (and where the participants were given advice on reducing fat intake). The adherence to the Mediterranean Diet was evaluated with a 14-point score, the Mediterranean Diet Adherence Screener (MEDAS).
Notably, this graph shows that the 2 arms were the Mediterranean diet (either plus nuts or extra-virgin olive oil) was applied gave considerably better results (in terms of cardiovascular incidence during the follow-up) compared to the control diet (Estruch et al., 2013). A major problem of this design was that, by comparing a low-fat diet with two diets supplemented with either nuts or olive oil (both sources of fats) the results can mostly be explained in terms of the effects of the supplemented foods (either nuts or olive oil) rather than in terms of adherence to the Mediterranean diet. Indeed, the subjects recruited in the control group ended up consuming a diet very similar to the Mediterranean one (Appel and Van Horn, 2013). Moreover, the first PREDIMED publication (which was published by the NEJM) was recently retracted, due to concerns about the validity of some of the data as well as errors in the randomization procedures which affected a portion of the participants.
A corrected version of the original study has been published (Estruch et al., 2017) where the data has been re-analyzed to correct for the previous errors. However, as a consequence of this, the PREDIMED can now only be considered an observational study and not an intervention study.
As it happened for many other topics, the research on the health effects of the Mediterranean diet had several bumps along the way. In general, I think we can conclude that there is a robust observational evidence about the association between the MD and total mortality, cardiovascular disease and neurodegenerative disease (such as Alzheimer Disease). A potential similar association exists for the risk of type 2 diabetes.
However, there is a lot of heterogeneity among the studies and a very limited evidence from intervention studies, especially after the PREDIMED has been retracted. However, the fact that the data from this study needed to be reanalyzed also implies that we still lack a large intervention study on the health effects of the Mediterranean diet, since this gap had been filled by the PREDIMED itself.
Therefore, we can only conclude that a healthy diet has an important role in several disease, and that a low-fat diet is not the most effective one. However, in order to confirm the superiority of the Mediterranean diet compared to other dietary patterns, a large, well-designed RCT is still needed. Such a study, in my opinion, should compare the Mediterranean Diet with other dietary patterns and it should also test the independent health effects of each Mediterranean food group.
Appel, L. J. and Van Horn, L. (2013) ‘Did the PREDIMED Trial Test a Mediterranean Diet?’, New England Journal of Medicine, 368(14), pp. 1353–1354. doi: 10.1056/NEJMe1301582.
Dinu, M. et al. (2018) ‘Mediterranean diet and multiple health outcomes: an umbrella review of meta-analyses of observational studies and randomised trials’, European Journal of Clinical Nutrition, 72(1), pp. 30–43. doi: 10.1038/ejcn.2017.58.
Esposito, K. et al. (2014) ‘The Effects of a Mediterranean Diet on the Need for Diabetes Drugs and Remission of Newly Diagnosed Type 2 Diabetes: Follow-up of a Randomized Trial’, Diabetes Care, 37(7), pp. 1824–1830. doi: 10.2337/dc13-2899.
Estruch, R. et al. (2013) ‘Primary Prevention of Cardiovascular Disease with a Mediterranean Diet’, New England Journal of Medicine, 368(14), pp. 1279–1290. doi: 10.1056/NEJMoa1200303.
Estruch, R. et al. (2017) ‘Primary Prevention of Cardiovascular Disease with a Mediterranean Diet’, N Engl J Med, 36814368(4), pp. 1279–90. doi: 10.1056/NEJMoa1200303.
de Groot, L. C. P. M. G. et al. (2004) ‘Lifestyle, nutritional status, health, and mortality in elderly people across Europe: a review of the longitudinal results of the SENECA study.’, The journals of gerontology. Series A, Biological sciences and medical sciences, 59(12), pp. 1277–84. Available at: http://www.ncbi.nlm.nih.gov/pubmed/15699526 (Accessed: 15 July 2018).
Iaccarino Idelson, P., Scalfi, L. and Valerio, G. (2017) ‘Adherence to the Mediterranean Diet in children and adolescents: A systematic review.’, Nutrition, metabolism, and cardiovascular diseases : NMCD, 27(4), pp. 283–299. doi: 10.1016/j.numecd.2017.01.002.
Kanerva, N. et al. (2014) ‘Associations of the Baltic Sea diet with obesity-related markers of inflammation’, Annals of Medicine, 46(2), pp. 90–96. doi: 10.3109/07853890.2013.870020.
Keys, A. et al. (1986) ‘The diet and 15-year death rate in the seven countries study.’, American journal of epidemiology, 124(6), pp. 903–15. Available at: http://www.ncbi.nlm.nih.gov/pubmed/3776973 (Accessed: 15 July 2018).
Knoops, K. T. B. et al. (2004) ‘Mediterranean Diet, Lifestyle Factors, and 10-Year Mortality in Elderly European Men and Women’, JAMA, 292(12), p. 1433. doi: 10.1001/jama.292.12.1433.
de Lorgeril, M. et al. (1999) ‘Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon Diet Heart Study.’, Circulation, 99(6), pp. 779–85. Available at: http://www.ncbi.nlm.nih.gov/pubmed/9989963 (Accessed: 15 July 2018).
Menotti, A. et al. (2001) ‘Cardiovascular risk factors and 10-year all-cause mortality in elderly European male populations. The FINE study’, European Heart Journal, 22(7), pp. 573–579. doi: 10.1053/euhj.2000.2402.
Mithril, C. et al. (2012) ‘Guidelines for the New Nordic Diet’, Public Health Nutrition. doi: 10.1017/S136898001100351X.
Orlich, M. J. et al. (2013) ‘Vegetarian Dietary Patterns and Mortality in Adventist Health Study 2’, JAMA Internal Medicine, 173(13), p. 1230. doi: 10.1001/jamainternmed.2013.6473.
Shai, I. et al. (2008) ‘Weight Loss with a Low-Carbohydrate, Mediterranean, or Low-Fat Diet’, New England Journal of Medicine, 359(3), pp. 229–241. doi: 10.1056/NEJMoa0708681.
Sho, H. (2001) ‘History and characteristics of Okinawan longevity food.’, Asia Pacific journal of clinical nutrition, 10(2), pp. 159–64. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11710358 (Accessed: 15 July 2018).
Sofi, F. et al. (2008) ‘Adherence to Mediterranean diet and health status: meta-analysis.’, BMJ (Clinical research ed.), 337, p. a1344. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18786971 (Accessed: 15 July 2018).
Sofi, F. et al. (2010) ‘Accruing evidence on benefits of adherence to the Mediterranean diet on health: an updated systematic review and meta-analysis’, The American Journal of Clinical Nutrition, 92(5), pp. 1189–1196. doi: 10.3945/ajcn.2010.29673.
Sofi, F. et al. (2014) ‘Mediterranean diet and health status: an updated meta-analysis and a proposal for a literature-based adherence score’, Public Health Nutrition, 17(12), pp. 2769–2782. doi: 10.1017/S1368980013003169.
Streppel, M. T. et al. (2008) ‘Dietary fiber intake in relation to coronary heart disease and all-cause mortality over 40 y: the Zutphen Study’, The American Journal of Clinical Nutrition, 88(4), pp. 1119–1125. doi: 10.1093/ajcn/88.4.1119.
Trichopoulou, A. et al. (1995) ‘Diet and overall survival in elderly people.’, BMJ (Clinical research ed.), 311(7018), pp. 1457–60. Available at: http://www.ncbi.nlm.nih.gov/pubmed/8520331 (Accessed: 15 July 2018).
Vincent-Baudry, S. et al. (2005) ‘The Medi-RIVAGE study: reduction of cardiovascular disease risk factors after a 3-mo intervention with a Mediterranean-type diet or a low-fat diet’, The American Journal of Clinical Nutrition, 82(5), pp. 964–971. doi: 10.1093/ajcn/82.5.964.
Al Wattar, B. H. et al. (2016) ‘Effect of simple, targeted diet in pregnant women with metabolic risk factors on maternal and fetal outcomes (ESTEEM): study protocol for a pragmatic multicentre randomised trial.’, BMJ open, 6(10), p. e013495. doi: 10.1136/bmjopen-2016-013495.