OUP user menu

Evaluating and adapting the Mediterranean diet for non-Mediterranean populations: a critical appraisal

Richard Hoffman, Mariette Gerber
DOI: http://dx.doi.org/10.1111/nure.12040 573-584 First published online: 1 September 2013


This review outlines the limitations of current techniques for evaluating the Mediterranean diet in Mediterranean versus non-Mediterranean populations. Differences between the two populations with regard to the foods that are available, food processing and preparation techniques, and eating and lifestyle habits may influence the implementation and effects of a Mediterranean diet in non-Mediterranean regions. For example, the composition of food groups may vary significantly, due to differences in the specific foods within a food group and to differences in aspects of food production and preparation. Notable differences between the diets of Mediterranean versus non-Mediterranean populations include the source of monounsaturated fatty acids (olive oil versus meat), the amount of vegetables consumed and their manner of preparation, the source of alcohol (wine versus other) and the pattern of intake, and the types of meat and dairy products consumed. Lifestyle factors such as meal patterns and exposure to sunlight may also act as confounding factors when the overall benefits of a Mediterranean diet are assessed. Improving the calculation of Mediterranean diet scores and measuring plasma nutrient levels may help mitigate the effects of confounders. These considerations could have important health implications when a Mediterranean diet is implemented by non-Mediterranean populations.

  • confounding factors
  • dietary pattern
  • Mediterranean diet
  • Mediterranean diet score


Although multiple forms of the Mediterranean diet have emerged as a consequence of the diverse food habits across the Mediterranean region, nutritionists have nevertheless established a model Mediterranean diet. This model Mediterranean diet is characterized by a large quantity and diversity of plant-derived foods (whole-grain cereals, raw and cooked vegetables, fresh and dried fruits, legumes, and nuts), fish, a moderate intake of meat and dairy products (with both types of products preferably from goats and sheep), olive oil as the added fat, and a moderate intake of wine during meals.1,2 This dietary pattern typically represented the food habits of southern Italy and Greece (especially Crete) around the 1970s. These countries were part of the Seven Countries Study, which was the first study to demonstrate the association between a Mediterranean diet and a reduction in all-cause mortality, especially a reduction in cardiovascular mortality, when comparing a traditional Mediterranean diet with diets from the United States and northern European countries.3 Later, the health effects of the Mediterranean diet were studied within a single population by classifying the subjects into groups according to their adherence to a Mediterranean diet4 and by evaluating their disease risk relative to how closely their food habits were aligned with the traditional Mediterranean diet.5,6 All of these studies were conducted within Mediterranean populations and used a priori dietary patterns that were based on the main characteristics of the Mediterranean diet.

Subsequently, the application of Mediterranean diet scores (MDSs) to non-Mediterranean populations was proposed. There are, however, a number of differences in the eating and lifestyle habits of Mediterranean versus non-Mediterranean populations that make using the original MDS either impossible (e.g., when the MDS includes olive oil, since only a few populations in the world use olive oil) or misleading (e.g., if the foods being scored have different compositions in Mediterranean and non-Mediterranean countries). Differences in food composition could potentially influence the evaluation of the effectiveness of a Mediterranean diet in non-Mediterranean populations. Analysis of the results of studies that have applied an MDS (either the original one, or a derived or modified score) allows those features of the Mediterranean diet that appear most stringent to be pinpointed. This is important because of the interest in encouraging a Mediterranean diet in non-Mediterranean populations in order to manage coronary heart disease (CHD) and other chronic diseases.7,8

This review discusses some of the differences between the eating and lifestyle habits of Mediterranean versus non-Mediterranean populations that could potentially impact the health benefits of a Mediterranean diet. Three main issues are addressed: 1) Are the a priori MDSs that have been constructed from literature data or dietary assessments obtained in Mediterranean populations appropriate for evaluating adherence to a Mediterranean diet and its health effects in non-Mediterranean countries? 2) Are there factors that are associated with a Mediterranean diet in Mediterranean countries that are not present in non-Mediterranean countries and which may affect the health benefits obtained from a Mediterranean diet (e.g., differences in lifestyle and meal structure)? 3) Could differences in the types and varieties of foods consumed and their methods of preparation also play a role in the health benefits obtained from a Mediterranean diet?

These issues are illustrated by comparing the Western diet, especially in the UK, with the diet in Mediterranean countries.

Dietary Assessments of Adherence to a Mediterranean Diet

A priori dietary patterns

By the 1990s, nutritional epidemiology had foreseen the limits of using an assessment system based on nutrient intake. The ability to understand the relationship between food and health was restricted because of incomplete food composition tables and because of the difficulty in ascertaining the specificity of a given nutrient, or even a given food, to an observed health effect. It became evident that a holistic approach should be used in nutritional epidemiology,9,10 and this applied especially to the Mediterranean diet, a dietary pattern whose beneficial effects were becoming widely acknowledged.11 Several MDSs were therefore designed.6,12,13 Only the MDS of Trichopoulou et al.5,6 will be discussed here because it is the one used most widely and because it has been modified to fit various populations.

The original MDS was devised on the basis of the food consumption pattern of an elderly Greek population who had maintained a traditional Mediterranean diet up until the 1990s,5 and this score was later modified.6 A nutritional survey of this population provided a quantitative assessment of the characteristic foods of the Mediterranean diet. A point scale was established to assess the degree of adherence to the traditional Mediterranean diet. The authors arbitrarily decided that the median sex-specific value of the consumption of nine selected characteristic components of the Mediterranean diet would be the cutoff point that determined adherence. A value of 0 or 1 was assigned for each of the five presumed beneficial food groups: vegetables, legumes, fruits and nuts, cereals, and fish. A value of 0 was assigned if consumption was below the median, and a value of 1 was assigned if consumption was above the median. It was decided that olive oil would be replaced with the ratio of monounsaturated fatty acids (MUFAs) to saturated fatty acids (SFAs) because Ancel Keys had established that this ratio was the main factor contributing to low cardiovascular disease (CVD) mortality.3 Thus, a value of 1 was assigned to individuals with a MUFA:SFA ratio at or above the median observed in the population, and a value of 0 was assigned for a ratio below the median. For dairy products and for meat and meat products (whose consumption is typically low or moderate in the Mediterranean diet), a value of 0 was assigned for consumption at or above the median, and a value of 1 was assigned for consumption below the median. To indicate the moderate intake of ethanol in the Mediterranean diet, a value of 1 was assigned to men who consumed between 10 g and <50 g per day and to women who consumed between 5 g and <25 g per day. Thus, the final MDS ranged from 0 (minimal adherence) to 9 (maximal adherence).

This MDS scoring system was in very close alignment with the dietary patterns of the Greek population and was consistent with previous results on the benefits of a Mediterranean diet on mortality.5,6 It was therefore proposed that this scoring system be applied to other populations, e.g., populations in the European Prospective Investigation into Cancer and Nutrition (EPIC) study.14 However, a few modifications were necessary in order to adapt the MDS to other European populations. The scoring system for the EPIC study required two major adaptations.14

Adaptations of the original Mediterranean Diet Score

Adaptation 1 for EPIC

In the original MDS scoring system, the major source of MUFAs is olive oil, and the high MUFA:SFA ratio reflects a high consumption of olive oil and a low consumption of animal products. Because relatively few people in northern Europe use olive oil, it was necessary to modify the scoring system by using another marker for vegetable oil consumption. Hence, the MUFA:SFA ratio became the unsaturated fat:SFA ratio. This had two consequences. First, it did not take into consideration substances in the nonsaponifiable fraction of olive oil (especially the triterpene squalene and the phenolic compounds hydroxytyrosol and oleocanthal), many of which have antioxidant and anti-inflammatory actions and are potentially protective against chronic degenerative diseases and, hence, may reduce mortality. Second, MUFAs are, to a large extent, markers of olive oil consumption in the original MDS because MUFA consumption was applied to a population that mainly used olive oil for cooking and as a salad dressing. When applied to northern European populations, however, MUFAs reflect, to a large extent, the consumption of animal fat, and it has been demonstrated that these MUFAs are associated with CVD15 and breast cancer,16 as discussed by Gerber and Richardson.17 It is interesting to note that, in an a posteriori adaptation of the MDS, the amount of olive oil consumed replaced the MUFA:SFA ratio.18

Adaptation 2 for EPIC

A significant difference between Greek and northern European populations lies in the intake of plant-derived foods. Retaining the Greek level of consumption for the MDS would have resulted in the majority of northern Europeans scoring 0 for the intake of vegetables and legumes. Hence, the median consumption of each of the nine components of the MDS was used as the cutoff value in each population. Under these conditions, the cutoff values differed markedly among the EPIC populations. For example, the median Greek consumption of vegetables (excluding potatoes) in the EPIC study was 500 g/day for men and 550 g/day for women,6 whereas the median value for all populations of the EPIC study was 157 g/day for men and 184 g/day for women.14 As might be anticipated, such a modified MDS was not associated with a reduced mortality risk in the northern European populations analyzed.14 The a posteriori adaptation of the MDS of Sofi et al.18 proposed absolute values for food consumption to be reached in order to describe adherence to a Mediterranean diet. These values were derived from a segmented regression analysis of the consumption of each food component of the MDS and overall mortality of the Greek EPIC cohort. The “change-point” on the segment represented the food consumption cutoff value to be used in the MDS. Thus, for vegetables, the change-point is =500 g, very close to what could be deduced from the literature and similar to the amount generally proposed in public health recommendations.

Several other modified MDSs have subsequently been proposed. In the United States, Fung et al.19 excluded potato products from the vegetable group, separated fruit and nuts into two groups, eliminated the dairy group because low-fat milk is consumed predominantly in the United States, included whole-grain products only, included only red and processed meats in the meat group, and allocated one point for alcohol intake between 5 and 15 g/day. It was called the alternate Mediterranean Diet Score (aMed) and was not associated with overall postmenopausal breast cancer risk, although an association with risk was found in subclasses of breast cancer.

In 2010, another modification of the MDS was developed and applied to the EPIC cohorts.20 Each component (apart from alcohol) was calculated as a function of energy density (g/1,000 kcal/day) and was then divided into tertiles of intake. A score of 0, 1, or 2 was assigned to the first, second, and third tertiles of intake for the five components presumed to fit the Mediterranean diet, namely fruit (including nuts and seeds), vegetables (excluding potatoes), legumes, fish (fresh or frozen, excluding fish products and preserved fish), and cereals. The scoring was inverted for the two components presumed not to fit the Mediterranean diet, namely total meat and dairy products. This scoring recognized the importance of olive oil: 0 was assigned to nonconsumers, 1 to subjects who consumed an amount below the median (calculated only within olive oil consumers), and 2 to subjects who consumed an amount equal to or above the median. Alcohol consumption was evaluated either as 2 for moderate consumers (range: 5–25 g/day for women and 10–50 g/day for men) or 0 for subjects who consumed an amount outside (above or below) the sex-specific range. This modification was called the relative Mediterranean diet (rMED). The rMED scores were then grouped into low (0–6), medium (7–10), and high (11–18) scores. This score compensated for the differences in food consumption between the cohorts by using energy density and took into consideration the specificity of olive oil. The results for individual cohorts are not shown, but the overall results show a 33% reduction in the risk of gastric adenocarcinoma with increasing adherence to the rMED.

Correlating and confounding factors associated with a Mediterranean diet

It has been shown that the order of courses in a meal and the pattern of meals throughout the day are distinct characteristics of a Mediterranean diet.21 Lunch is the main meal, providing not only the required energy but also a sufficient quantity of plant-derived nutrients, i.e., fiber, micronutrients, and phytochemicals. In addition, besides the expected typical food intake, sharing of meals with family or colleagues and an absence of snacking were found to be major features in Sardinia, where there is high adherence to a traditional Mediterranean diet, compared with Malta, where many traditional features have been lost.21 Although these different factors might play an additional role besides food components in preserving health, they are not currently assessed in dietary questionnaires and hence cannot be controlled for by statistical techniques.

Several other factors may be reduced or absent in the lifestyles of people living in non-Mediterranean countries. A healthy energy balance derived from physical activity was an intrinsic part of the Mediterranean diet in Mediterranean populations following a traditional lifestyle. In addition, taking a siesta – still current practice in some Mediterranean countries – has been associated with a lower risk of CVD,22 but this is rarely considered in epidemiological studies. Another potentially confounding factor in relation to several chronic degenerative disorders is the possible protective effects conferred by vitamin D.23,24 The ultraviolet-induced synthesis of vitamin D will supplement nutritional intake to a greater extent for many people in Mediterranean countries, where the sunlight is more intense, than for those in more northerly countries.

Comparison of Foods and Their Preparation Methods in Mediterranean and Non-Mediterranean Countries

The precise composition of a food and how it is produced and prepared can differ significantly between Mediterranean and non-Mediterranean countries. These differences may have an impact on the overall health benefits of a Mediterranean diet in non-Mediterranean countries.

Olive oil

The major difference between northern and Mediterranean countries in relation to olive oil consumption is simply whether it is consumed at all, as discussed in a previous section. The health benefits of olive oil, however, may be affected by both the quality and the culinary use of olive oil. MUFA content does not vary significantly between different qualities of olive oil, but olive oil also contains a “nonsaponifiable” fraction comprising various triterpenes (mostly squalene), phytosterols (mostly β-sitosterol), tocopherols (mostly vitamin E), and phenolic compounds. The amounts of these components may vary among different types of olive oil and might also be influenced by culinary practices. Levels of phytosterols, tocopherols, and squalene, a putative protective factor against breast cancer,25 are reduced with increasing refinement of virgin olive oil. In addition, phenolic compounds are present in significant quantities only in virgin olive oil, and so their health benefits are lost when nonvirgin olive oils are consumed. Potentially important phenolic compounds in virgin olive oil include lignans, which are associated with reduced breast cancer risk,26,27 hydroxytyrosol, which has cardioprotective and anticancer activity in experimental systems,28 and the anti-inflammatory substance oleocanthal. Frying with virgin olive oil reduces the phenolic compound content,29 and this highlights the potentially important role that consumption of raw virgin olive oil may play in the Mediterranean diet.


Populations in Mediterranean and non-Mediterranean countries can have very different preferences for vegetables and for food preparation methods, and this may influence the effect of vegetable consumption on health outcomes.30 Dutch university students were reported to prefer “Brussels sprouts, green peas and carrots often with apple sauce,” whereas Greek students chose “mostly fresh salads of tomatoes, cucumber, cabbage, rocket [arugula], radishes, spinach, and lettuce with olive oil, vinegar, or balsamic vinegar and herbs, or green vegetables in baked pies (spinach pie, leek pie, etc.).”31 Salads were found to be consumed several times per week in an analysis of the traditional Cretan Mediterranean diet.32 In EPIC cohorts from northern countries, consumption of raw vegetables as a proportion of total vegetable consumption was reported to be lower compared with their southern counterparts.33 This was particularly striking for UK men, who consumed only half the amount of raw vegetables as cooked vegetables, whereas the proportions of raw and cooked vegetables were fairly similar in Mediterranean countries. Data from the UK Living Costs and Food Survey indicate that the main fresh vegetables purchased in the United Kingdom in 2010 included cabbage, Brussels sprouts, cauliflower, salad leaves, carrots, alliums, and tomatoes.34

These wide variations in preferred types of vegetables may have a significant impact on phytochemical intake and, hence, on any correlated disease risk. The low consumption of dark green leafy vegetables (e.g., broccoli, spinach, kale) in the United Kingdom33 is noteworthy because these represent a major dietary source of vitamins C and K, folate, β-carotene, flavones, and the carotenoids lutein and zeaxanthin.35 Garlic consumption, common in Mediterranean countries, is low in the United Kingdom, and the 2007 report of the World Cancer Research Fund/American Institute for Cancer Research considered it “probable” that garlic consumption contributes to protective effects against stomach and colorectal cancers.36 Differences in the phytochemical content between different varieties of the same vegetable can be substantial. For example, the flavonol content of lettuce varieties ranged from 0.5 μg/g fresh weight for iceberg lettuce – a variety commonly purchased in northern European countries – to 207 μg/g fresh weight for the Italian variety lollo rosso.37 Dietary flavonol intake is linked to a decreased risk of CVD, including stroke.38 Of course, low consumption of one vegetable may be compensated for by consumption of another vegetable containing the same beneficial nutrients, and, moreover, many phytochemicals present in vegetables are also found in fruits. Thus, a mixed and diverse diet can help ensure an optimum intake of a wide range of healthy phytochemicals. The potential importance of a varied diet was highlighted in an EPIC study showing that the diversity of fruit and vegetables in the diet was associated with a decreased risk of lung cancer, an effect over and above the inverse association with quantity.39 It is noteworthy that a traditional Mediterranean diet includes a particularly wide diversity of fruits and vegetables.

The preparation method can influence both nutrient levels and nutrient bioavailability. Consumption of raw vegetables preserves heat-labile nutrients – such as vitamins A and C and folates – that otherwise can be lost when vegetables are cooked. In addition, the use of an oil-based dressing – olive oil is traditional in Mediterranean countries – was shown to increase the bioavailability of carotenoids from salad ingredients.40 Breaking down the food matrix by cooking or pureeing can also increase the bioavailability of carotenoids.41 The common practice in northern European countries of boiling a single type of vegetable and discarding the cooking water can result in significant nutrient loss due to leaching of water-soluble nutrients such as folates and glucosinolates into the cooking water.42,43 This practice is less common in Mediterranean countries, where soups and stews are preferred, and since the cooking medium is retained when soups and stews are prepared, there is no loss of water-soluble nutrients. The frying of vegetables can lead to significant losses of fat-soluble nutrients such as carotenoids, probably into the cooking fat,44 but this practice is not very common in Mediterranean countries. On the other hand, when the entire contents of the pan are consumed, for example, in Mediterranean stews, the overall nutritional value of the dish is not compromised.

The emphasis in a traditional Mediterranean diet is on seasonal, field-grown vegetables, whereas for a northern European market, “Mediterranean” vegetables are frequently grown under glass in greenhouses. Greenhouse cultivation reduces ultraviolet B exposure due to the limited light-absorbing properties of glass. It is also often accompanied by increased fertilizer use. Both of these factors can have adverse effects on the production of phytochemicals, although this appears to depend on both the type of vegetable and the specific phytochemical.45

In conclusion, the nutritional benefits that different populations receive from consuming vegetables may vary widely and are not accurately assessed by simply determining absolute levels of consumption.

Fruits and Nuts

Total fruit consumption in the United Kingdom is low compared with that in Mediterranean countries, and across Europe there is a north-south gradient for total consumption.33 Apples, bananas, and citrus fruits together accounted for about two-thirds of all fresh fruit purchased in 2010 by UK households, and processed fruits and fruit products (excluding fruit juices) accounted for one-third of total fruit purchases.34 Consumption of summer fruits popular in Mediterranean countries – such as pomegranates, figs, grapes, and “orange fruits” (e.g., apricots, peaches, nectarines, and cantaloupes) – is relatively low in the United Kingdom.

When eaten raw, many fruits are a good source of vitamin C. In contrast, the relative amounts of various phytochemicals in different fruits can vary widely: citrus fruits are good sources of flavanones and flavones, berries are rich in anthocyanins and flavan-3-ols, and Mediterranean orange fruits are important sources of α-carotene and β-carotene. Due to the difficulty of using epidemiological studies to identify the effects of individual nutrients within a diet, it is unclear whether the particular nutrient content of a specific fruit affects health outcomes. Chong et al.,46 however, concluded that there was some limited evidence that fruits rich in flavonols, anthocyanins, and procyanindins, such as pomegranates, purple grapes, and berries, are more effective at reducing the risk of CVD.

Fruit is a typical way to end a meal in Mediterranean countries. Pro-oxidant and proinflammatory processes increase at the end of a meal, and this is linked to increased cardiovascular damage. Postprandial hyperlipidemia and hyperglycemia are also risk factors for a number of metabolic disorders, including type 2 diabetes, CVD, and metabolic syndrome. Some studies have shown that consumption of phenol-rich fruits during the postprandial phase increases the antioxidant capacity of the blood.47 Hence, consuming fruit at the end of a meal is a prudent strategy for healthy eating. In conclusion, consumption of a wide range of fruits is advisable, especially if the phytochemical intake from vegetables is limited. Eating fruit, rather than a pastry, at the end of a meal not only reduces caloric intake, it may also help counteract oxidative stress and other pathological events during the postprandial phase.

An analysis of nut consumption in the EPIC study found that cohorts from central European countries (northern France, Germany, The Netherlands, United Kingdom) and Mediterranean countries (southern France, Greece, Italy, Spain) had similar overall levels of consumption.48 However, higher proportions of peanuts relative to tree nuts (mainly walnuts, almonds, and hazelnuts) were consumed in the central European countries. For example, in the United Kingdom, peanuts and tree nuts constituted 40.4% and 36.4%, respectively, of total nuts and seeds consumed, whereas in Spain these figures were 26.8% and 54.9%, respectively.

All types of nuts have been shown to reduce the risk of CHD, although only very limited data are available for peanuts.49 Nuts have hypocholesterolemic effects, and a number of intervention studies have demonstrated that nuts lower both low-density lipoprotein (LDL) cholesterol and the ratio of LDL cholesterol to high-density lipoprotein (HDL) cholesterol.49 Participants of the Prevención con Dieta Mediterránea (PREDIMED) study (a multicenter intervention study) who consumed nuts as part of a Mediterranean diet had a statistically significant reduction (P<0.05) in LDL cholesterol and in the LDL:HDL cholesterol ratio.50 The cardioprotective effects of nuts may be related to their relatively high proportion of unsaturated fatty acids such as MUFA and linoleic acid.51 The hypocholesterolemic effects of nuts may also be related, in part, to their high phytosterol content. Pistachio nuts – a common aperitif nut in Mediterranean countries – have the highest phytosterol content of all nuts (279 mg/100 g), and oil- and dry-roasted peanuts and peanut butter also contain moderate levels of phytosterols (135 mg/100 g in oil-roasted peanuts).52 Tree nuts, which constitute a higher proportion of nuts consumed in Mediterranean countries, have additional nutrients that include high vitamin E content in almonds53 and high levels of alpha-linolenic acid in walnuts.51 Moreover, when eaten with their skins, tree nuts provide high levels of phenolic antioxidants.54

In conclusion, although there is good evidence that tree nuts have hypocholesterolemic properties, there is currently no strong evidence of hypocholesterolemic properties for peanuts. Moreover, high levels of salt in many peanut products preclude high intake, especially in subjects with high blood pressure, and peanuts are also relatively high in saturated fat, including palmitic acid.


It is difficult to appropriately assess the effects of cereals by using a nutritional questionnaire, since cereals can be refined, whole grain, salted, or sweetened, all factors that strongly influence health. In one MDS that did not score for either sweetened cereals or bread from fast-food items, it was still not possible to clearly evaluate the effects of cereals.12 Whole-grain cereals are very likely to be a better marker for health benefits, but this evaluation can only be applied when whole-grain cereals are consumed by a high percentage of the subjects under study.55 Another approach is to score refined cereals negatively.56


Legume consumption, excluding peas and green beans (i.e., pulses), showed a gradient of consumption in cohorts from the EPIC study, with higher levels of consumption in southern countries and lower levels in northern countries.33 Besides quantitative differences, the preferences for legumes in these two regions also differ. Commonly consumed legumes in Mediterranean countries include chickpeas, lentils, and fava (broad) beans, although there are national differences, e.g., chickpea consumption is high in Spain and intake of fava beans is high in Egypt. In the United Kingdom, chickpeas, fava beans, and lentils are consumed mainly by ethnic minorities, and the major types of legumes in the UK diet are canned “baked beans in sauce” and garden peas.34

Legume consumption is associated with a decreased risk of CHD and CVD,57 and consumption of various legumes, including baked beans (which are made from haricot beans),58 was shown in a recent meta-analysis to lower cholesterol levels.59 Legumes have excellent nutritional value and were ranked in the US diet as an important source of fiber, phytosterols, folate, vitamin B6, flavonols, flavan-3-ols, and various minerals.35 Both fiber and phytosterols may be linked to the hypocholesterolemic effects of legumes.60 Garden peas, a common legume in the UK diet, contain 134 mg/100 g phytosterols,61 which is comparable to amounts found in other pulses, and haricot beans are a good source of fiber. This suggests that legumes commonly consumed in the United Kingdom may have some of the cardioprotective effects of the pulses more commonly associated with a Mediterranean diet.30


There is substantial evidence that fish consumption lowers the risk of cardiovascular mortality.62 The most important bioactive nutrients in fish are generally considered to be the n-3 long-chain (LC) polyunsaturated fatty acids (PUFAs) eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA). A recent systematic review concluded that marine n-3 LC-PUFAs are effective in preventing cardiovascular events, cardiac death, and coronary events, especially in persons with high cardiovascular risk.63 There is less evidence for a role of LC-PUFAs in the prevention of cancer.64

Levels of n-3 LC-PUFAs are considerably higher in oily fish than in white fish. Since there is variability in the relative proportions of oily and non-oily fish among countries, measuring total fish intake may not reflect the intake of n-3 LC-PUFAs for a given population. The relative amounts of oily and non-oily fish do not follow a north-south gradient since consumption of “very fatty fish,” defined by Welch et al.65 as including herring, kipper, and mackerel, is high in Scandinavian countries. In contrast, the proportion of these types of fish consumed in the United Kingdom is relatively low, and so total fish consumption in the United Kingdom will be associated with a proportionally lower intake of n-3 LC-PUFAs from fish than in some other countries.

A wide range of factors, including the food consumed by the fish, influence the LC-PUFA content of oily fish. This is particularly important when considering farmed fish, an increasingly important dietary source of fish. Farmed salmon is a major source of n-3 PUFAs in the UK diet. Salmon, like other saltwater fish, has only a limited capacity to synthesize LC-PUFAs and instead obtains LC-PUFAs from its feed. When fed fish oils, farmed salmon are an excellent source of LC-PUFAs, but there are increasing environmental and commercial pressures on fish farmers to use nonmarine sources of oils. This practice can drastically reduce DHA levels. For example, levels of 17 g DHA/100 g total fatty acids were measured in salmon that were fed fish oils, whereas levels of 5 g DHA/100 g total fatty acids were found in salmon fed plant oils.66 Feeding fish oils to fish during the last few months before marketing is one technique that can restore EPA and DHA to fish raised on a predominantly vegetable-based diet.67 Hence, the food supply of farmed fish is an important consideration when assessing the health benefits of oily fish.

Pan frying with olive oil is one of the most popular ways of preparing fish in Mediterranean countries, and fish fried in virgin olive oil has been found to absorb significant quantities of antioxidant phenolics, terpenic acids, and vitamin E from the oil.68 Hence, there may be incidental benefits of frying fish in olive oil. On the other hand, there can also be an exchange of fatty acids between those in the fish and those in the frying oil. For example, the amount of n-3 fatty acids in sardines fell between two- and threefold when sardines were fried in either sunflower oil or olive oil, and there was an accompanying rise in n-6 fatty acids.69

In conclusion, the nutritional content of fish is influenced by the type of fish, the diet of the fish, and the method of preparation, and these factors may vary significantly among Mediterranean and non-Mediterranean countries.

Dairy products

The preferred types of dairy products consumed in many Mediterranean countries are significantly different from those consumed in non-Mediterranean countries. In most Mediterranean countries, with Spain being the main exception, proportionally less milk and milk beverages and more cheese and yogurt are consumed than in northern European countries.70 Significantly, a large number of the cheeses in Mediterranean countries are made from sheep's milk (e.g., Roquefort and tomme from the Pays Basque region of southern France, Manchego from Spain, and feta from Greece) and goat's milk (e.g., various chévres from southern France). In contrast, the main dairy products in northern Europe are cow's milk and cow's milk cheese such as cheddar-type hard cheese in the United Kingdom.

Although cheeses made from goat's and sheep's milk have a similar total saturated fat content as cheeses made from cow's milk, the composition of the saturated fats is different because both goat's milk and sheep's milk are richer in medium-chain fatty acids (MCFAs), i.e., those with <12 carbon atoms (<12C). These MCFAs include caproic acid (C6:0), caprylic acid (C8:0), and capric acid (C10:0) (the names are derived from the Latin caper for “goat”) (Table 1).71 For example, fresh goat cheese with 40% fat contains 15% MCFAs, whereas MCFAs constitute only 7% of a comparable cow's milk cheese. Similarly, the fat content of Roquefort cheese contains 15% MCFAs and 23% palmitic acid. By comparison, a cow's milk fatty cheese contains 33% palmitic acid (16C), which is the most atherogenic SFA.72 MCFAs are nonatherogenic and are oxidized directly in the liver, thus reducing their accumulation in adipose tissue. Some epidemiological evidence supporting the beneficial effects of MCFAs came from the Nurses' Health Study (a prospective cohort study that included more than 80,000 US females), which showed that, in contrast to intake of long-chain fatty acids, intake of MCFAs was not significantly associated with the risk of CHD.73

View this table:
Table 1

Medium-chain fatty acid content of sheep's, goat's, and cow's milk.71

Fatty acidFatty acid composition (percentage of total fats)
Sheep's milkGoat's milkCow's milk
Caproic C6:0 2.9 2.4 1.6
Caprylic C8:0 2.6 2.7 1.3
Capric C10:0 7.810.0 3.0
Total13.315.1 5.9

The composition of milk is influenced by the animal's diet. Goats and sheep are more likely than cows to be raised on natural pasture. Pasture is rich in alpha-linolenic acid and gives rise to higher levels of alpha-linolenic acid in the animal's lipids, which can be desaturated to EPA, an n-3 LC-PUFA with anti-inflammatory properties.

High consumption of dairy products was considered to be detrimental in the MDS devised by Trichopoulou et al.5 This is because the dairy products were rarely low in fat, and longer-chain SFAs have detrimental effects on cholesterol levels. There is now a trend in many countries towards low-fat dairy products. In the United Kingdom, weekly purchases of whole milk have steadily decreased year after year (from 2,655 mL/week in 1974, down to 352 ml/week in 2010), with a concomitant rise in the purchase of skimmed milk (mostly semi-skimmed).34 Similarly, milk consumed in the United States is mostly low-fat milk. Not all saturated fats are harmful, however. Myristic acid (14C), present in the milk of ruminants, is necessary for the myristoylation of several functional proteins,74 and it is not atherogenic when the exogenous source constitutes ≤2% of the total energy intake. In addition, the natural trans-fatty acid trans-palmitoleic acid (cis-16:1 n-7), levels of which correlate strongly with whole-fat dairy consumption, was shown to be associated with lower metabolic risk factors.75 These observations cast doubt on the relevance of the low-fat milk recommendations in many countries.


A recent meta-analysis concluded that red meat consumption is associated with a small increase in the risk of colorectal cancer,76 although other analyses have concluded that, because of possible confounding, the current evidence for such an association is weak.77 It has been suggested that the procarcinogenic effects of heme iron, a putative carcinogen in red meat, can be suppressed by various dietary constituents such as chlorophyll, calcium, and antioxidant vitamins (C and E),78,79 although the possible relevance of these interactions to the effects of the Mediterranean diet is not known. Marinating, a technique traditionally used in Mediterranean cuisine to tenderize cheap cuts of meat, may also have beneficial health effects. Cooking meat at high temperatures, such as frying and grilling, generates carcinogenic heterocyclic aromatic amines, and marinades that contain virgin olive oil, onions, garlic, herbs, or red wine have high antioxidant capacity and have been shown to inhibit the formation of heterocyclic aromatic amines.8082

The geographical characteristics of the Mediterranean region favor small livestock with specialized feeding habits: sheep and goats can take advantage of the hilly landscape and of Mediterranean grazing, whereas pigs prefer woodland spaces. Because the pasture of Mediterranean animals used to produce dairy and meat is richer in PUFAs than the animal feeds given to equivalent animals, the fatty acid profile of Mediterranean animals is healthier. This is especially true for pigs running in open spaces in Corsica and Sardinia – their meat is leaner, and their fat consists of 40–50% MUFA. Together with the higher content of linolenic acid, this results in a lower proportion of SFA in the fat composition (Table 2).72,83

View this table:
Table 2

Fat composition of various meats

MeatFatty acids (percentage of total fat)
Lamb52.140.55.8 (LA 5.0 + ALA 0.8)
Pork43.247.69.2 (LA 8.6 + ALA 0.6)
Beef56.440.33.2 (LA 2.5 + ALA 0.7)
  • Data from McCance et al.72 and Favier et al.83

  • Abbreviations: ALA, alpha-linolenic acid; LA, linoleic acid; MUFAs, monounsaturated fatty acids; PUFAs, polyunsaturated fatty acids; SFAs, saturated fatty acids.

Other foods

A number of other foods not generally scored in the MDS may have important health benefits for a Mediterranean diet. These include herbs (also consumed as herbal teas) and wild greens as well as pumpkin, sunflower, and other types of seeds. Local consumption of these foods can be high, and many are very rich sources of phytochemicals (e.g., pumpkin seeds contain high levels of phytosterols) and other nutrients.2

Notable by their absence in a traditional Mediterranean diet are modern fast foods and sugar-sweetened drinks.21 Fast food can be a major source of salt and trans fats, both well-known risk factors for CVD, and both fast foods and sugar-sweetened drinks are positively associated with long-term weight gain.84 An evaluation of fast foods and sugar-sweetened drinks is only rarely included in the MDS. In one such study from rural Lebanon, it was found that, when food consumption deviated from a traditional Mediterranean diet by including refined cereals and pastries and sugar-sweetened drinks, there was an increase in obesity and visceral adiposity.56


Moderate alcohol consumption (defined as 10 g to <50 g per day in men and 5 g to <25 g per day in women) is assumed to be beneficial in the MDS6 and is strongly linked to a reduced risk of CVD. The most compelling mechanism to explain the cardioprotective effects of moderate alcohol consumption is the increase in levels of HDL cholesterol; beneficial effects on the vasculature may also be involved.

A number of studies from non-Mediterranean countries have shown that even moderate alcohol consumption increases the risk of cancer at some sites, such as the breast.85,86 By contrast, a Mediterranean diet reduces overall cancer risk, and there is no evidence from specific analysis of the alcohol component that moderate levels of consumption in the context of a Mediterranean diet increase the risk of cancer.87 One important factor that might contribute to these disparities between Mediterranean and non-Mediterranean countries with regard to cancer risk and moderate alcohol consumption is drinking pattern.88 Whereas the custom in Mediterranean countries is to drink in moderation with a meal, drinking outside mealtimes and binge drinking are more prevalent in northern European countries.89,90 Drinking with a meal slows the rate of absorption of alcohol from the gut, whereas drinking on an empty stomach raises absorption rates and may increase blood alcohol concentrations to levels that saturate alcohol metabolic pathways, resulting in the production of carcinogenic metabolites. Dietary folates may also influence the cancer risks associated with drinking alcohol. Some studies,91 but not all,85 have shown that folates reduce the cancer-associated effects of moderate alcohol consumption. It is noteworthy that the Mediterranean diet has particularly high levels of folate consumption, and there is a good correlation between folate consumption and adherence to a Mediterranean diet.92

Alcohol consumption in Mediterranean countries is typically associated with wine (mostly red wine), which is drunk with a meal, whereas beer and spirits are consumed in greater quantities in some non-Mediterranean countries.89 There is some evidence that over and above the effects of alcohol, phenolic compounds in red wine may have specific cardioprotective effects. Distinguishing between the effects of alcohol and phenolic compounds in wine can be accomplished by using dealcoholized wine. For example, dealcoholized red wine has been shown to retain the ability of complete wine to modulate leukocyte adhesion molecules, important inflammatory biomarkers related to atherosclerosis in subjects at high risk of CVD.93 Studies also suggest that drinking wine with a meal may confer additional cardioprotective effects.94 A large number of studies have shown that dealcoholized wine increases postprandial total antioxidant capacity and reduces postprandial increases in oxidized LDL cholesterol, an important risk factor for CVD.47,95,96

In summary, drinking moderate amounts of red wine with a meal may confer superior health benefits compared with other types of drinking, and this is not assessed when measuring consumption of alcohol in the MDS. Hence, factors not monitored in the MDS – such as drinking pattern, other dietary constituents, and type of alcohol – are important factors to consider when weighing the risks and benefits associated with moderate alcohol consumption.


In the traditional concept of the Mediterranean diet, there are various food habits and lifestyle aspects that may be absent in non-Mediterranean populations. The a priori MDS first described by Trichopoulou et al.6 in 2003, therefore, needed to be adapted for specific populations; in effect, however, the Mediterranean diet is no longer Mediterranean because it has a number of nutritional characteristics that are different from the traditional Mediterranean diet. Determining whether these differences influence health outcomes is important when assessing the effects of a Mediterranean diet in non-Mediterranean populations. One lifestyle factor that may be important is physical activity, which is generally taken into consideration in questionnaires. For other factors, however, such as meal structure, the organization of meals during the day, conviviality, and taking a siesta, it would be necessary to extend the scope of questionnaires undertaken as part of evaluating the MDS. Although longer questionnaires would be more time consuming, it would be helpful to include questions on the source of meat and dairy products in order to assess the importance of specific categories of SFAs and the levels of PUFAs. The information obtained could be supported with biomarker measurements of fatty acids and other nutrients.12 Another biological measurement of interest is vitamin D status.

It became evident during the early development of the MDS that the first MDS, originally developed for a Mediterranean population, would need to be adapted for non-Mediterranean populations in order to take into consideration the eating habits of non-Mediterranean populations.20,55,56 In fact, taking a nutritional survey a posteriori in order to identify dietary patterns is the best way to reveal the healthiest type of diet among this population. These patterns can then be compared with those of the original Mediterranean diet, and differences between the two patterns can be identified. Using this approach to adapting the MDS to non-Mediterranean populations, several important features have been identified. For example, the rMED score20 and the new a posteriori MDS18 identified the importance of measuring olive oil consumption and the need to consider an absolute amount of food consumed when assessing the beneficial effects of olive oil.

Measuring plasma levels of selected nutrients that are specific to a unique source or class of food (e.g., n-3 LC-PUFAs and fish) may offer a more precise way than food frequency questionnaires to assess the contribution of specific nutrients to the health benefits of the traditional Mediterranean diet. When assessing the effects of the Mediterranean diet in non-Mediterranean populations, it is important to examine those nutrients whose levels are likely to be influenced by culinary and lifestyle factors that vary significantly between Mediterranean and non-Mediterranean populations. One candidate nutrient is folate, since dietary intake of folate is associated with a reduced risk of a range of chronic disorders such as colon cancer.97 Folate levels in foods are influenced by culinary practices. Since folate is water soluble and heat labile, boiling of folate-containing foods results in the leaching of folate into cooking water.42 Moreover, folate levels vary widely between vegetables. Since folate is found mainly in green leafy vegetables, folate levels will not be specifically assessed by quantifying total vegetable intake. Hence, folate plasma levels may not necessarily be reflected in the MDS.

Sufficient plasma levels of other nutrients that are linked to adherence to a Mediterranean diet may also be important in order for non-Mediterranean populations to achieve the full benefits of a traditional Mediterranean diet. Gerber et al.12 found a significant correlation between the Mediterranean diet quality index and plasma levels of α-tocopherol, β-carotene, EPA, and DHA, especially when these nutrients were combined in a composite index, thus establishing that plasma values of these nutrients correlated with good adherence to a Mediterranean diet. Subjects in northern Spain with higher Mediterranean diet adherence, as measured by two dietary indexes, had significantly higher plasma concentrations of β-carotene, folates, vitamin C, α-tocopherol, and HDL cholesterol.92 Carotenoids may also be useful biomarkers because of their association with risk of cancer at certain sites and because food preparation can influence bioavailability. However, correlating plasma concentrations of a particular nutrient with disease reduction is not straightforward, since food plants contain many nutrients that are consumed at the same time; for example, most of the fruit and vegetables containing carotenoids also contain lignans.


Although there is extensive epidemiological evidence supporting the health benefits of a Mediterranean diet, most of this evidence is derived from MDSs that do not fully consider the many potentially confounding differences between Mediterranean and non-Mediterranean populations. This review identifies a number of factors that may confound the assessment of the health benefits of a Mediterranean diet in Mediterranean countries, both because of the nature of the food itself and because of aspects related to food production and preparation. Lifestyle factors such as meal patterns and exposure to sunlight may also act as confounders. Improving the calculation of MDSs and measuring plasma nutrient levels may help mitigate the effects of these confounders. Consideration of these confounding factors could have important health implications when assessing the effects of a Mediterranean diet in non-Mediterranean populations.


Declaration of interest

The authors have no relevant interests to declare.


View Abstract