The Mediterranean Diet (Med Diet) is based on an eating pattern found mainly in Crete and other parts of Greece and Italy(Reference Willett, Sacks and Trichopoulou1). The dietary pattern was first described by Ancel Keys, an American biologist investigating diets in populations that had lower rates of cardiovascular disease in the Seven Countries Study(Reference Menotti and Puddu2,Reference Keys, Menotti and Karvonen3) . The Med Diet consists of legumes, vegetables, fruits, nuts, wholegrain foods and fish. The main source of fat is olive oil, and the diet encourages smaller quantities of red meat, poultry, eggs and dairy produce (mainly yoghurt and soft cheese)(Reference Willett, Sacks and Trichopoulou1). The focus is on whole foods and historically has included small quantities of wine, which recently has become more contentious(Reference Keys, Menotti and Karvonen3,Reference Simopoulos4) . In the countries from which this dietary pattern is derived, there are additional elements of the lifestyle that are harder to measure, but are likely integral to the overall health benefits associated with the dietary pattern. These include a more physical lifestyle in the garden (contributing both physical activity and home-grown produce) and kitchen (home-cooked meals), foraging for greens and eating meals with family(Reference Willett, Sacks and Trichopoulou1). The literature shows social networks are greater in Mediterranean countries compared with non-Med countries(Reference Litwin5).
It is important to note that the ‘Med Diet’ is not reflective of the eating habits of all countries around the Mediterranean Sea. The particular region the diet comes from has access to more fish and traditionally eats less red meat than other parts of the Mediterranean, where lamb is a staple of the diet. The first food pyramid aiming to encapsulate the Med Diet was published in 1995 by Walter Willet et al. of Harvard University with the WHO(Reference Willett, Sacks and Trichopoulou1). This has since been updated several times, most recently in 2020 to include sustainable food production practices(Reference Serra-Majem, Tomaino and Dernini6).
The objective in this review is to examine the background of the Med Diet, identify barriers to adoption particularly due to cultural differences and examine a selection of studies for their effect on glycaemic and other cardiometabolic outcomes.
Epidemiology of the Med Diet in Mediterranean countries
Many epidemiological observational cohort studies and subsequent meta-analyses and systematic reviews studying outcomes from Mediterranean and non-Mediterranean countries have associated a high adherence to the Med Diet with lower rates of disordered glucose metabolism, such as lower HbA1c, fasting plasma glucose (FPG) and progression to either Type 2 Diabetes or use of glucose-lowering medications, when compared to a low adherence to the Med Diet(Reference Schwingshackl, Chaimani and Hoffmann7,Reference Martín-Peláez, Fito and Castaner8) . Similarly, high adherence to the Med Diet is associated with a lower BMI and blood pressure (BP) and a lower risk of cardiovascular disease, including sudden death(Reference Gregory, Ntailianis and Shannon9–Reference Hareer, Lau and Mole12). The large cohort studies typically record dietary intake using FFQ, whereupon the data are mapped to a Mediterranean Diet Score (Med diet score). They also typically only use baseline data and therefore do not reflect change over time and cannot be used to infer cause. There are a number of Med diet scores used in the literature; some are described later in this review. The Med diet score quantifies the degree to which an individual adheres to the Med Diet. The scores are stratified in tertiles or quartiles with logistic regression applied to estimate the predicted impact of the Med diet score on clinical outcomes. These are not usually diet intervention studies, but are studies with large datasets that can be used to look at relationships outside of the original study aims.
One prospective cohort study in Southern Italy randomly selected 5250 subjects from the electoral roll or GP records in 2005/2006 and followed them until 2018(Reference Campanella, Misciagna and Mirizzi13). Participants completed a study visit and FFQ at baseline, and no further dietary assessment was made during the 12-year follow-up. The primary outcome over the 12-year period was time-to-death from all causes. Those with a medium adherence to the Med Diet at baseline had a lower lifespan of 5.62% (95% CI 1.01, 10.3), and those with a low adherence had a lower lifespan of 9.90% (95% CI 5.30, 5.3) compared with a high adherence.
In an Italian study of 8,291 individuals post-myocardial infarction, diabetes incidence was tracked for a mean follow-up period of 3.2 years(Reference Mozaffarian, Marfisi and Levantesi14). Risk factors for new onset diabetes included higher BMI, smoking, a lower Med diet score and consumption of wine at more than 1L per day. Further, in the Greek arm of the EPIC Cancer prospective study, 22 295 participants were followed for a median of 11.34 years for cases of incident diabetes(Reference Rossi, Turati and Lagiou15). Participants completed an FFQ at baseline, and this was mapped to the Med Diet Score (MDS). A higher MDS was inversely associated with diabetes risk (HR 0.88 [95% CI 0.78, 0.99] for MDS ≥ 6 vs MDS ≤ 3).
Epidemiology of the Med Diet in non-Mediterranean countries
Further evidence supporting the Med Diet pattern comes from observational studies of diet and metabolic or cardiovascular risk factors and outcomes in non-Mediterranean countries. In the PREVENT Dementia study following 700 participants in the United Kingdom, baseline dietary data were used to explore the relationship between Med diet scores and cardiovascular risk factors(Reference Gregory, Ntailianis and Shannon9). Three Med diet scores were calculated – Mediterranean Diet Adherence Screener (MEDAS), the MEDAS continuous score and the Med Diet Pyramid score. Higher adherence to all three diet scores was associated with lower BP, BMI and Framingham and Qrisk3 scores.
In the French Supplémentation en Vitamines et Minéraux AntioXydants study of 3235 participants, 6-year risk of Metabolic Syndrome (MetS) was calculated using three 24-h food recalls at baseline and compared with the MetS score after 6 years. Three Med diet scores were used – the MDS, Med style dietary pattern score (MSPS) and their own developed Med diet score MED. A lower risk of MetS score was observed with increasing Med diet score (P 0.001) in multivariate models(Reference Kesse-Guyot, Ahluwalia and Lassale16). The adjusted odds ratio (95% Confidence Interval) for MetS risk was 0.47 (0.32–0.69) in subjects in the highest versus lowest tertile of MED and MDS scores(Reference Kesse-Guyot, Ahluwalia and Lassale16).
In the USA, the large NHANES III dataset, with observational data from a single study visit and FFQ questionnaire from 13 197 individuals, examined the relationship between the Med Diet and atherothrombotic risk(Reference Carter, Roberts and Salter17). FFQ data were used to calculate a Med diet score, with the exception that the olive oil score (a component of the Med diet score) had to be approximated from 24-h recall data as it was not specifically itemised in the FFQ questionnaire. As the Med diet score increased, the measures of insulin sensitivity and lipid parameters decreased generally, with some variability between genders and age groups. There was no association between the Med diet score and BP. There was a stronger inverse association between Med diet score and coronary heart disease (CHD) risk in men than in women (p = 0.0077 for men under 45 years and p = 0.0005 for men over 45 years, whereas for women, there was no significant trend for 10-year CHD risk). There were some positive results for post-menopausal women with a higher Med diet score resulting in a lower Total Cholesterol:HDL-cholesterol ratio, TAG, insulin and Homeostatic model assessment (HOMA). The lack of positive results for pre-menopausal women may suggest that the effect of female hormones eclipses that of dietary change on cardiovascular risk.
The European Prospective Investigation into Cancer and Nutrition (EPIC) study was a large prospective cohort in ten European countries (Denmark, France, Germany, Greece, Italy, the Netherlands, Norway, Spain, Sweden and the United Kingdom) following 521 448 participants(Reference Lassale, Gunter and Romaguera18). Dietary questionnaires from baseline were used to compute three Med diet scores, (MDS, relative Med diet score (rMED) and the MSPS, among seven other healthy eating scores and indices (US-Healthy Eating Index, Diet Quality Index International, WHO Healthy Diet Indicator, Healthy Lifestyle Index, World Cancer Research Fund guideline score, DASH diet score, Healthy Nordic Food Index). After 10 years of follow-up, there was an inverse relationship between all tested indices and cardiovascular mortality. However, there was no overall greater benefit to adherence to a Med Diet in particular compared with other dietary patterns, except in Mediterranean countries such as Spain. In contrast, in the Netherlands, it was the Healthy Nordic Dietary Pattern that was more strongly associated with lower mortality. This is a very important observation suggesting that different dietary patterns may have improved health outcomes in different populations. Whether that is driven by genetic or cultural factors or both cannot be answered by these observational studies. The application of a range of dietary pattern scores to other large epidemiological datasets from a range of populations in non-Mediterranean countries would be of great interest, as would understanding the typical population dietary intake in each case and targeting the most appropriate dietary intervention accordingly.
Components of the Med Diet
The Med Diet as conceived is a dietary pattern of whole foods. However, specific individual components may have greater or lesser impact on any healthful effects of the diet.
The fatty acid profile of the Med Diet favours MUFA from olive oil, PUFA from nuts and seeds and omega-3 fatty acids from seafood(Reference Simopoulos4). Purslane is a commonly foraged green that also contains omega-3 fats and when fed to chickens results in eggs that are notably high in these PUFAs(Reference Sam-Yellowe19). The health benefits of Med Diet interventions have sometimes been interpreted as resulting from a simple reduction in saturated fat, given that the Med Diet is naturally low in saturated fat by limiting red meat and bakery products. However, in addition, olive oil contains 70–85% MUFA, mainly oleic acid (18:1), and has been singularly linked to lower all-cause and cardiovascular mortality in meta-analyses of MUFA and mortality(Reference Schwingshackl and Hoffmann20,Reference Martínez-González, Sayón-Orea and Bullón-Vela21) . The Med Diet is also high in fibre, nutrients and antioxidants from legumes, wholegrains, fruit and vegetables that likely contribute to its health effects(Reference D’Alessandro, Lampignano and De Pergola22–Reference Marventano, Vetrani and Vitale26).
Alcohol has traditionally been included as a small part of the Med Diet – recommending 1 glass of wine per day for women and 2 for men if one drinks alcohol(Reference Willett, Sacks and Trichopoulou1). Wine, particularly red wine, is a source of antioxidants including resveratrol, which has been linked to reduced effects of aging and decreased oxidative stress(Reference Zhou, Luo and Huang27,Reference Zhang, Li and Kakar28) . Earlier studies linked moderate drinking with slightly lower risk of mortality than those who abstained, but this evidence holds less weight when one considers the abstaining group contains those who were heavy drinkers and have given up alcohol altogether(Reference Tsai, Gao and Wen29–Reference Bergmann, Rehm and Klipstein-Grobusch31). In a Spanish cohort of university graduates, a Med style alcohol pattern of moderate red wine drinking was associated with lower mortality when compared with abstainers and higher or spirit drinkers(Reference Gea, Bes-Rastrollo and Toledo32). However, alcohol is linked to seven cancers, and the US Surgeon General released a report in January 2025 stating that no amount of alcohol is safe from cancer risk(Reference Bagnardi, Rota and Botteri33). Some Med Diet studies have chosen to remove or downplay the alcohol factor so as not to be seen to be encouraging alcohol consumption. Martinez-Gonzalez et al. are currently conducting a study in Spain comparing a Med Diet with alcohol (wine) to a Med Diet with non-alcoholic drinks provided, which may help to answer this question(Reference Martínez-González34).
Med Diet interventions
The Med Diet is currently viewed as the most evidence-based dietary pattern for reducing the risk of metabolic and cardiovascular disease. Such evidence has led to the adoption of a Med Diet by people not traditionally following this pattern as a means of reducing such risks(Reference Nestel and Mori35). However, a closer look at the evidence from intervention trials that test the effectiveness of this approach is warranted. Dietary intervention studies are always difficult to compare because of methodological factors including duration, energy restriction and population characteristics such as obesity, pre-existing diabetes, hypertension or cardiovascular disease. In addition, adherence to dietary patterns that differ widely from usual habits is challenging, particularly for research investigations in free-living individuals. We have sought to summarise the literature here.
We identified 20 Med Diet intervention studies published in the last 25 years of at least 2 months duration and greater than 25 participants that have focused on clinical outcomes of glucose metabolism (progression to diabetes or hypoglycaemic medications, HbA1c, FPG) with or without other metabolic risk factors (weight, BP, blood lipids, metabolic syndrome score) and PubMed indexed. These studies are summarised in Table 1. Of these 20 studies, 13 are randomised controlled trials (RCTs) of Med Diet compared with either usual care or other intervention(Reference Krebs, Parry-Strong and Braakhuis36–Reference Gomez-Huelgas, Jansen-Chaparro and Baca-Osorio48), 4 are crossover studies(Reference Itsiopoulos, Brazionis and Kaimakamis49–Reference Gardner, Landry and Perelman53), and 3 are non-randomised cohort studies of men compared with women on the Med Diet(Reference Bédard, Riverin and Dodin54,Reference Leblanc, Bégin and Hudon55) . Separately, we discuss two studies where all arms are on the Med Diet intervention; the first study by McEvoy et al. tests different ways of providing support for the diet(Reference McEvoy, Moore and Erwin56). The Predimed-plus study has both groups on a Med Diet, but one is energy restricted(Reference Salas-Salvadó, Díaz-López and Ruiz-Canela57). The studies were conducted in the following countries: Spain, Italy, the USA, Canada, Australia, New Zealand, Northern Ireland, Israel and China.
Table 1. Intervention studies of Med Diets on cardiovascular risk factors
Abbreviations: ADA = American Diabetes Association; cf = compared with; WC = waist circumference; FPG = fasting plasma glucose; IR = insulin resistance.
Glycaemia
Of the 17 RCT or crossover studies, eight were in cohorts with type 2 diabetes(Reference Elhayany, Lustman and Abel37,Reference Esposito, Maiorino and Ciotola38,Reference Toobert, Glasgow and Strycker40,Reference Toobert, Strycker and Barrera41,Reference Brehm, Lattin and Summer46,Reference Alonso-Domínguez, García-Ortiz and Patino-Alonso47,Reference Itsiopoulos, Brazionis and Kaimakamis49,Reference Gardner, Landry and Perelman53) and a further five studies were in participants with metabolic syndrome or high fasting glucose(Reference Krebs, Parry-Strong and Braakhuis36,Reference Luo, Wang and Sun42,Reference Montemayor, Bouzas and Mascaró43,Reference Gomez-Huelgas, Jansen-Chaparro and Baca-Osorio48,Reference Bédard, Riverin and Dodin54) . Eight studies demonstrated an improvement in glycaemic control parameters (HbA1c or fasting glucose) in the Med Diet group(Reference Elhayany, Lustman and Abel37–Reference Toobert, Strycker and Barrera41,Reference Alonso-Domínguez, García-Ortiz and Patino-Alonso47,Reference Itsiopoulos, Brazionis and Kaimakamis49,Reference Estruch, Martínez-González and Corella58) , four of which were in participants with type 2 diabetes or metabolic syndrome(Reference Elhayany, Lustman and Abel37,Reference Esposito, Maiorino and Ciotola38,Reference Toobert, Glasgow and Strycker40,Reference Toobert, Strycker and Barrera41) . Of the eight studies demonstrating improvements in glucose metabolism, five of these saw a reduction in weight, which may contribute to improvements in glycaemia(Reference Elhayany, Lustman and Abel37,Reference Esposito, Maiorino and Ciotola38,Reference Toobert, Glasgow and Strycker40,Reference Toobert, Strycker and Barrera41,Reference Alonso-Domínguez, García-Ortiz and Patino-Alonso47) . Two of these eight studies were energy restricted(Reference Esposito, Maiorino and Ciotola38,Reference Shai, Schwarzfuchs and Henkin39) and the rest ad libitum (Reference Toobert, Glasgow and Strycker40,Reference Toobert, Strycker and Barrera41,Reference Alonso-Domínguez, García-Ortiz and Patino-Alonso47,Reference Itsiopoulos, Brazionis and Kaimakamis49,Reference Estruch, Martínez-González and Corella58) . The study by Shai et al. compared three arms – low-fat, low-carbohydrate and Med Diet – and while the Med Diet group lost less weight than the low-carbohydrate group (4.6 kg vs 5.5 kg), they had greater improvements in HbA1c(Reference Shai, Schwarzfuchs and Henkin39). The study by Elhayany also compared three arms, with different weight loss occurring across groups – Med Diet (−7.4 kg), low-carbohydrate Med Diet (−10.1 kg) and the American Diabetes Association (ADA) diet (−7.7 kg)(Reference Elhayany, Lustman and Abel37). The low-carbohydrate Med Diet had the greater reduction in HbA1c. In Northern Ireland, the study by McEvoy et al. was not controlled as all participants were exposed to the Med Diet and all participants experienced improvements in BMI, BP and HbA1c(Reference McEvoy, Moore and Erwin56).
Of the studies that did not find an improvement in response to the Med Diet, Luo et al.(Reference Luo, Wang and Sun42), Montemayor et al.(Reference Montemayor, Bouzas and Mascaró43) and Brehm et al.(Reference Brehm, Lattin and Summer46) restricted energy intake in all groups with resultant significant reductions in weight, HbA1c, fasting glucose and insulin levels and no significant difference between groups. This highlights the importance of weight loss relative to dietary patterns. Gomez-Huelgas et al. was a large trial that concluded poor adherence contributed to their lack of outcomes, despite a MEDAS score increase of 2.1 (out of a total of 17)(Reference Gomez-Huelgas, Jansen-Chaparro and Baca-Osorio48). Our recent RCT, He Rourou Whai Painga, was a 3-month ad libitum trial with provision of food equivalent to 75% energy requirements to the household, with additional dietary change support, that saw no effect on glycaemia or the metabolic syndrome severity score(Reference Krebs, Parry-Strong and Braakhuis36). Despite no instruction to reduce energy intake, there was a small but significant weight loss of −1.2 kg in the intervention group compared with the control. Barnard et al. was a crossover study where HOMA-IR decreased on the vegan diet with a 6 kg weight loss compared to no change on the Med diet(Reference Barnard, Alwarith and Rembert52), and Wade et al. was a shorter 8-week crossover study where no changes were observed in any outcomes, which may have been too short to see a benefit(Reference Wade, Davis and Dyer50,Reference Wade, Davis and Dyer51) . Hershey et al. saw no effect on glucose at 6 or 12 months(Reference Hershey, Chang and Sotos-Prieto44). Lastly, the crossover study by Gardner et al. saw no difference in HbA1c between interventions, but both were carbohydrate restricted, and the lack of a washout period between interventions was a likely factor(Reference Gardner, Landry and Perelman53).
Weight, lipids and blood pressure
Five studies reported greater weight loss for the Med Diet group, two of these were energy restricted(Reference Esposito, Maiorino and Ciotola38,Reference Salas-Salvadó, Díaz-López and Ruiz-Canela57) and three were ad libitum (Reference Krebs, Parry-Strong and Braakhuis36,Reference Toobert, Glasgow and Strycker40,Reference Toobert, Strycker and Barrera41) . The three studies that were ad libitum were compared with usual care rather than another intervention. The weight loss was in the magnitude of 1.2–4.6 kg or 0.38–1.1 BMI points. In the two studies that compared the Med Diet in men and women, one study found a greater weight loss for men(Reference Leblanc, Bégin and Hudon55) and the other for women(Reference Bédard, Riverin and Dodin54). Three studies found a positive effect on BP(Reference Gomez-Huelgas, Jansen-Chaparro and Baca-Osorio48,Reference Barnard, Alwarith and Rembert52,Reference Bédard, Riverin and Dodin54) and six on blood lipids.(Reference Krebs, Parry-Strong and Braakhuis36,Reference Elhayany, Lustman and Abel37,Reference Hershey, Chang and Sotos-Prieto44,Reference Gomez-Huelgas, Jansen-Chaparro and Baca-Osorio48,Reference Gardner, Landry and Perelman53,Reference Bédard, Riverin and Dodin54) Four studies found a greater improvement in waist circumference (WC) in the Med Diet(Reference Montemayor, Bouzas and Mascaró43,Reference Alonso-Domínguez, García-Ortiz and Patino-Alonso47,Reference Gomez-Huelgas, Jansen-Chaparro and Baca-Osorio48,Reference Estruch, Martínez-González and Corella59) . In the Chinese study by Luo et al., a Med Diet was compared with a traditional high fibre Chinese diet and a control diet mainly consisting of white rice and pork(Reference Luo, Wang and Sun42). All meals were provided 5 d a week for 6 months, and each participant had an individualised 25% energy restriction. The three groups all lost around 5 kg and had similar improvements in fasting glucose, insulin and BP. This study in a non-Mediterranean population with a very different traditional dietary pattern, again highlights the importance of weight loss versus dietary pattern in promoting metabolic benefits.
Background – the Predimed and Lyon Heart studies
The largest intervention study of the effect of Med Diet on cardiovascular health has been the well-publicised Predimed study(Reference Estruch, Ros and Salas-Salvadó45,Reference Estruch, Ros and Salas-Salvadó60) . This study recruited 7447 participants in Spain at high cardiovascular risk and randomised to three groups: control diet, Med Diet provided with nuts (walnuts, hazelnuts and almonds) and Med Diet provided with olive oil. The primary outcome was a cardiovascular event (myocardial infarction, stroke or death from cardiovascular causes), and the study was stopped after 4.8 years because of benefits in the Med Diet groups. The main trial was republished after anomalies were found in randomisation in some sites, but cardiovascular events were still significantly lower in both Med Diets (hazard ratios 0.69 for the olive oil group and 0.72 for the nuts group)(Reference Estruch, Ros and Salas-Salvadó45). This impressive reduction in cardiovascular events has underpinned the further widespread promotion of the Med Diet in public health messaging across the world. There was no difference in weight between groups at 5 years, but WC changes favoured the Med diet with nuts group(Reference Estruch, Martínez-González and Corella59). After 3 months in a subset of 772 participants, both Med Diet groups had significant reductions in FPG compared with the low-fat group.(Reference Estruch, Martínez-González and Corella58) In another sub-cohort (Predimed-Reus)(Reference Salas-Salvado, Bullo and Babio61), in 418 participants, diabetes incidence was 10.1% (95% CI 5.1–15.1), 11.0% (5.9–16.1) and 17.9% (11.4–24.4) in the Med Diet with olive oil group, the Med Diet with nuts group and the control group, respectively, after follow-up of 4 years. And finally, in the cohort of 3230 participants who had T2DM at baseline managed by lifestyle, the hazard ratios of starting a first glucose-lowering medication were 0.78 (95% CI 0.62–0.98) for the Med Diet with olive oil group and 0.89 (0.71–1.12) for the Med Diet with nuts group, compared with the control group(Reference Basterra-Gortari, Ruiz-Canela and Martínez-González62).
The Predimed study created the predimed diet score for the study as a way of measuring adherence to the overall Med Diet. The predimed diet score is 14 points, and fewer than 10 is considered a low adherence to the Med Diet. At baseline, the average score was 8.6, and there was an inverse relationship between diet score and obesity(Reference Martínez-González, García-Arellano and Toledo63). After 1 year, in the first 1551 participants, changes in predimed individual food scores reflected the intervention group allocation, and total score increases were as follows: Med Diet + olive oil 1.86 (1.70, 2.03), Med Diet + nuts 2.26 (2.09, 2.43) and control 0.46 (0.28, 0.64)(Reference Zazpe, Sanchez-Tainta and Estruch64). The increases in both Med Diet groups were significant compared with the control group. The studies reporting glycaemic and clinical cardiovascular outcomes did not report changes in predimed score for these sub-cohorts.
As Predimed was such a large study with various cohorts over a number of years, it has generated over 360 publications, making it difficult to find all reported outcomes for diet scores and anthropometric measures. Although it was an intervention trial, many journal articles are cross-sectional analyses or in small subsets of participants. It is also difficult to understand which participants completed which measures, as while blood samples were taken yearly, there is no single publication reporting these measures for the whole cohort.
Predimed-plus was a follow-on study in 6,874 older Spanish adults aged 55–75 with two cohorts – both following a Med Diet, but one group was energy restricted, had behavioural support and encouraged to exercise, while the other was ad libitum (Reference Salas-Salvadó, Díaz-López and Ruiz-Canela57). Both groups received Med Diet education and some supplementary olive oil and nuts. The energy-restricted group saw greater reductions in weight, WC, fasting glucose and TAG after 1 year, suggesting that an ad libitum Med Diet may not translate to improvements in clinical outcomes and that once again, weight loss is a key factor in the benefits to the metabolic syndrome. The energy-restricted group had a higher adherence score to the Med Diet despite both groups receiving the same education and supplementary foods, suggesting the behavioural support helped adherence. But the 17-item Med diet score used in this study included three extra questions aimed at capturing energy restriction. What would the difference have been if both groups had been administered the original 14-item predimed diet score?
The Lyon heart study published in 1994 is another well-cited intervention study, intended to be a Med Diet, but did not solely promote olive oil (instead providing rapeseed oil and margarine) as it was considered this French population would not accept sole olive oil use(Reference de Lorgeril, Renaud and Mamelle65). The intervention did encourage higher fish, bread, vegetable and fruit intake and lower red meat intake but did not encourage nuts or legumes. The study included 605 participants after a first myocardial infarction, to either the intervention or control group (‘prudent diet’) and followed participants yearly for 5 years. Dietary adherence was assessed by dietary fat composition from FFQ and plasma fatty acid concentrations. The primary outcome was a combination of cardiac death or non-fatal MI, and both were lower in the intervention group after 27 months. Notably, there were no significant differences between groups in weight, BP or lipids. While this was intended to be a Med Diet, the lack of three key dietary factors does mean this study doesn’t meet the criteria now understood as common to a Med Diet. Instead, this was the basis of the low saturated fat diet focus of research and public health advice in cardiovascular health for many years.
Therefore, in contrast to the epidemiological studies, the evidence is mixed with respect to the effect of an intervention with a Med Diet on glucose metabolism and cardiovascular disease. There are several factors that might influence this. First may be the population characteristics, including genetic factors, cultural preferences, baseline disease risk and also baseline dietary pattern. Second may be the degree of change in dietary pattern and level of adherence with a Med Diet; that is, do people change their diet enough to make a difference. Third, it may be that there are other characteristics of people living in the Mediterranean countries from which the diet was first conceived that conferred cardiovascular and metabolic protection that are not replicated in a Med Diet intervention in other countries.
Med Diet interventions in Mediterranean countries vs non-Mediterranean countries
As suggested by the Lassale analysis, perhaps the Med Diet is most suited to the countries around the Mediterranean Sea(Reference Lassale, Gunter and Romaguera18). The Predimed study set in Spain was clearly the strongest evidence for the Med diet benefits with both olive oil and nut groups(Reference Estruch, Ros and Salas-Salvadó45). The other studies set in Italy, Spain and Israel still have varied results.
Israel: The studies by Shai and Elhayany et al. in Israel favoured the low-carbohydrate diets (although one was a low-carbohydrate Med diet in Elhayany) and participants in these studies lost more weight than in other trials(Reference Elhayany, Lustman and Abel37,Reference Shai, Schwarzfuchs and Henkin39) .
Italy: Esposito et al. in Italy also compared a low-carbohydrate Med Diet with a low-fat diet and while weight and HbA1c favoured the Med Diet it is hard to know if this was due to the reduction in carbohydrate or the Med Diet per se(Reference Esposito, Maiorino and Ciotola38). Gomez-Huelgas et al.(Reference Gomez-Huelgas, Jansen-Chaparro and Baca-Osorio48) did not find an effect of the Med Diet on glucose or MetS, but Alonso-Dominguez demonstrated a decrease in post-prandial glucose(Reference Alonso-Domínguez, García-Ortiz and Patino-Alonso47).
Spain: The Montemayor study in Spain compared a Med Diet with a Med Diet plus exercise with a conventional diet(Reference Montemayor, Bouzas and Mascaró43). As noted earlier, all groups improved, and the only benefit of the Med Diet was on WC.
Even within Mediterranean countries, intervention results vary, likely due to variations in carbohydrate intake and degree of weight loss.
Med Diet adherence measures
Earlier interventions have mainly used dietary fat ratios from FFQ questionnaires to test adherence to the Med Diet(Reference Elhayany, Lustman and Abel37–Reference Shai, Schwarzfuchs and Henkin39,Reference Toobert, Strycker and Barrera41,Reference Brehm, Lattin and Summer46) . An increase in MUFA and PUFA intake and a reduction in SFA were the main dietary assessments but also an increase in nut and legume consumption. Later studies have used Med diet scoring systems – the 14-point Predimed score, the 44-point Medscore or the 15-point Pyramid Med diet score calculated from FFQ data for adherence. Studies have also removed or added questions to tailor the score to their intervention, population or specific research question. More recently, the MEDAS screener was developed, which is a direct questionnaire of the Med Diet pattern for the participant rather than calculating scores from FFQ data. The variation between scoring systems makes direct comparisons of studies difficult. It is also difficult to be certain whether there is a threshold effect, which must be reached to see benefits, or a quantum change in score that is required; that is, how Mediterranean does the diet need to be or become?
Four treatment RCTs that used Med diet scores found no glycaemic benefit to the Med Diet(Reference Krebs, Parry-Strong and Braakhuis36,Reference Montemayor, Bouzas and Mascaró43,Reference Hershey, Chang and Sotos-Prieto44,Reference Gomez-Huelgas, Jansen-Chaparro and Baca-Osorio48) . Our study increased the PyrMDS (adapted 12-item) by 1.6 (13%) in the intervention group after 3 months from a baseline of 6.0/6.1(Reference Krebs, Parry-Strong and Braakhuis36), Hershey et al. increased the median of the Mediteranean Diet Score (mMDS) (51-item) by 2.01 (4%) at 6 months and 2.67 points (5%) at 12 months from a baseline of 23.41/22.87 points(Reference Hershey, Chang and Sotos-Prieto44). The trial by Montemayor et al. used the direct MEDAS screener (17-item) and increased the Med diet (high meal frequency) group by 6.2 points (36%) compared with the Med diet plus exercise group increase of 4.1 (24%) and the control diet by 3.6 points (21%), from 6.9, 7.7 and 7.3 at baseline respectively(Reference Montemayor, Bouzas and Mascaró43). Yet Alonso-Dominguez saw a reduction in post-prandial glucose after just a 2.2 point (13%) shift in the MEDAS(Reference Alonso-Domínguez, García-Ortiz and Patino-Alonso47). McEvoy et al. reported increases in the MDS (14-item) in their three Med diet cohorts of 4.25 (27%), 5.63 (35%) and 4.72 (30%) from an average score of 2.1, and all cohorts saw improvements in risk factors compared with baseline(Reference McEvoy, Moore and Erwin56). The cohort of McEvoy et al. started with lower scores and increased their scores by a greater magnitude than in the Krebs and Hershey studies. The Montemayor study also analysed adherence to MEDAS regardless of randomisation by tertile at 6 months and found that in those with high adherence compared to no change, there was a larger decrease in weight, BP and WC(Reference Montemayor, Mascaró and Ugarriza66). See Table 2 for a comparison of Med diet scores.
Table 2. Med Diet adherence scores
MDS = Med Diet Score; MEDAS = Mediterranean Diet Adherence Screener.
Self-reported dietary records also have major well-recognised limitations, not least of which is reporting bias. Therefore, biomarkers could be a helpful tool to measure some components of the Mediterranean diet intervention. Whilst there is no single ideal biomarker, plasma carotenoids and plasma fatty acid compositions were also used in two studies for adherence purposes. These measures have not been widely used, but Itsiopolous et al. noted an increase in plasma lycopene and lutein, and Toobert et al. in their 2003 study saw a small increase in EPA from plasma fatty acid analysis in the intervention groups(Reference Toobert, Glasgow and Strycker40,Reference Itsiopoulos, Brazionis and Kaimakamis49) . Biomarkers cannot measure adherence to a dietary pattern overall, and there is inter-individual variability in metabolism and half-life of metabolites. A panel of affordable biomarkers could complement self-reported dietary data in future studies.
Dietary change is notoriously difficult to achieve and sustain; therefore, several studies have provided food to facilitate change and increase adherence. The Predimed study provided olive oil and nuts in the respective groups, and the Predimed-plus study provided some supplementary supplies of both of these but not in the quantities in the original study. Bedard et al. and Gardner et al. provided all meals for the first 4 weeks on the diet, and Krebs et al. provided food equating to 75 % energy requirements of all meals for a whole household for 3 months(Reference Krebs, Parry-Strong and Braakhuis36,Reference Gardner, Landry and Perelman53,Reference Bédard, Riverin and Dodin54) . The Chinese study by Luo et al. set in a workplace provided all meals, 5 d a week at the participant staff cafeteria for 6 months(Reference Luo, Wang and Sun42). In the firefighter study by Hershey et al., samples of Med Diet foods were provided to the intervention group(Reference Hershey, Chang and Sotos-Prieto44). Apart from the Predimed study, none of these interventions achieved a change in glycaemic outcomes, but Bedard et al. and Krebs et al. did achieve increases in Med diet scores(Reference Krebs, Parry-Strong and Braakhuis36,Reference Bédard, Riverin and Dodin54) . Food provision does not seem to have been a defining factor in dietary success or significant increases in Med Diet scores.
Med Diet and cultural and other barriers
A range of barriers to adoption and adherence to the Med Diet have been reported in both Mediterranean and non-Mediterranean populations(Reference Tsofliou, Vlachos and Hughes70). Lower socio-economic status predicts poorer adherence to the Med Diet in multiple countries(Reference Tong, Imamura and Monsivais71–Reference Pribisalić, Popović and Salvatore75), and the Med Diet has been shown to be more expensive than some Western diets for the same energy intake(Reference Lopez, Martinez-Gonzalez and Sanchez-Villegas76,Reference Rydén, Sydner and Hagfors77) . However, the Med Diet promotes lower energy density foods such as pulses and vegetables and reduced animal products, which may partly offset this higher price(Reference Tong, Imamura and Monsivais71,Reference Rydén, Sydner and Hagfors77,Reference Schröder78) . Other demographic factors associated with lower adherence include younger age, lower education level and male gender(Reference Cavaliere, De Marchi and Donzelli72,Reference Olmedo-Requena, Fernández and Prieto79–Reference Knight, Jackson and Rahman81) . Additional support may be needed to improve the feasibility of the Med Diet in these groups. In a systematic review of barriers to adoption of a Med Diet, barriers were categorised as similar to healthy eating in general; taste, finance, accessibility, education, motivation and lifestyle or unique to the Med Diet; acceptability of pulses, legumes and fish, suggestion the Med Diet does not fit with a cooler climate and concern about abandonment of traditions (for example red meat)(Reference Tsofliou, Vlachos and Hughes82).
Perceived behavioural control (motivation, affordability, time/effort, food access, knowledge, food outlets, cooking skills) accounted for the majority of self-perceived barriers to adopting a Med Diet in an Australian study(Reference Scannell, Villani and Mantzioris83). Motivation was the most reported of these, with desire for other food and reluctance to change habitual diets among cited reasons(Reference Scannell, Villani and Mantzioris83). Other perceived barriers included suitability, taste and social norms(Reference Scannell, Villani and Mantzioris83). These findings agree with other literature on implementation of the Med Diet in non-Mediterranean countries(Reference Knight, Jackson and Rahman81,Reference Bottcher, Marincic and Nahay84,Reference Middleton, Keegan and Smith85) . However, in our study where we specifically addressed barriers to access food, knowledge about food preparation and recipe choices, support of family and social context of food, although we did demonstrate an increase in the Pyramid score, this did not translate to reduced cardiometabolic risk.
Cultural adaptation of the Med Diet can improve biological and psychological outcomes. The nutrient profile of the Med Diet can be used to create analogous culturally appropriate meal plans(Reference George, Kucianski and Mayr86). Cultural enablers may help guide the implementation of these meal plans. Some examples of dietary enablers in the Māori population of Aotearoa New Zealand are whānau (family) support/inclusion in interventions and increasing Māori leadership and the use of Māori health models in ratonga hauora (health services)(Reference McKerchar, Barthow and Huria87). Our study (Krebs et al.) was conducted in New Zealand and included a family in the intervention, included a Māori health provider and had both Māori Investigators and a Māori governance panel to ensure cultural responsiveness throughout the study(Reference Krebs, Parry-Strong and Braakhuis36). However, the diet intervention was not co-designed with the community but rather user centred (designed by the investigators and applied to all ethnic groups), which may have impacted outcomes. For some cultures, it is particularly difficult to translate the Med Diet into a cuisine that is acceptable. In a previous cross-sectional study of New Zealanders, adherence to a Med Diet was low among Māori and Pacific people(Reference Lovell, Roy and Klein88), but in our study, there was no difference in PyrMDS score between ethnicities. On the other hand, one critique suggests that imposing a European diet on non-European cultures is disrespectful and more efforts should be made to localise healthy dietary patterns based on usual cultural foods(Reference Burt89).
In the Viva Bien study, a culturally adapted intervention for Latina women including a modified Med Diet resulted in greater improvement of diet composition and biological and psychological outcomes than standard guidelines(Reference Toobert, Strycker and Barrera90). Cultural adaptations made in this study were an initial bilingual 2-d retreat, bilingual handouts with pictures of Latina women, family nights with potluck and adapted recipes for the Med diet from Latin American cuisines.
Conclusion
When implemented with adequate adherence and energy control, the Med Diet can support improvements in glycaemic and metabolic outcomes. However, the effectiveness varies widely across populations and interventions. Despite a wealth of epidemiological evidence for the cardiometabolic benefits of the Med Diet, there have been relatively fewer intervention studies focused on cardiometabolic risk factors rather than cardiovascular events. The Cochrane review of the Med Diet in 2019 concluded that there was uncertainty regarding the effect of the Med Diet on CVD risk factors with only low to moderate evidence for modest benefits.(Reference Rees, Takeda and Martin91) More than 5 years later, this uncertainty persists.
-
• Is weight loss central to the cardiometabolic benefit? Therefore, is energy restriction required as part of a Med Diet? Similarly, there seems to be a benefit from carbohydrate restriction along with a Med diet.
-
• Do baseline factors such as obesity, diabetes or existing CVD modify the effect of a Med Diet intervention?
-
• What degree of change in dietary pattern is required, and is there a threshold effect of Med Diet that must be achieved to see cardiometabolic benefit? A larger increase in adherence would seem to have more effect.
-
• Do ethnicity or cultural factors and/or other lifestyle contexts modify the effect of a Med Diet?
A greater understanding of these questions will help clarify whether widespread adoption of the Med Diet would be expected to reduce cardiometabolic risks in diverse populations and settings. Future research should focus on replicating work with and without carbohydrate and energy restriction to quantify the effects of the Med Diet per se, with a strong focus on adherence.
Acknowledgements
The author list for the He Rourou Whai Painga Consortium is:
Prof Jeremy D. Krebs1,2^, Prof Richard Gearry3, Assoc Prof Troy L. Merry4,5, Assoc Prof Andrea Braakhuis4, Assoc Prof Fiona Lithander4,6, Dr Meika Foster6,7, Dr Anna Rolleston8, Dr Amber Parry-Strong2, Cecilia Ross2, Prof Mark Weatherall1, Dr Denise Conroy9, Cheryl Davies10, Anna Worthington4
1 Department of Medicine, University of Otago, Wellington, New Zealand
2 Centre for Endocrine, Diabetes and Obesity Research (CEDOR), Wellington Regional Hospital –Te Whatu Ora, Wellington, New Zealand
3 Department of Medicine, University of Otago Christchurch, Christchurch, New Zealand
4 Department of Nutrition and Dietetics, School of Medical Sciences, The University of Auckland, Auckland, New Zealand
5 Maurice Wilkins Centre for Molecular Biodiscovery, The University of Auckland, Auckland, New Zealand
6 Liggins Institute, The University of Auckland, Auckland, New Zealand
7 Edible Research Ltd, Canterbury, New Zealand
8 Manawaora Integrated Health and Research Ltd, Tauranga, New Zealand
9 Plant and Food Research, Auckland, New Zealand
10 Tū Kotahi Māori Asthma and Research Trust, Kōkiri Marae, Lower Hutt, New Zealand
^ Principal investigator
Author contributions
APS, KH, FL, AB and JK contributed to the writing of this manuscript. The He Rourou Whai Painga Consortium provided editing and critical revision of the manuscript.
Financial support
The authors declare that financial support was received for the research, authorship and/or publication of this article. This study was supported by High Value Nutrition National Science Challenge Funding through the NZ MBIE.
Competing interests
The author declares no conflicts of interest.
Open access
