The Mediterranean: Diets and Disease Prevention

Marion Nestle. Cambridge World History of Food. Editor: Kenneth F Kiple & Kriemhild Conee Ornelas, Volume 2, Cambridge University Press, 2000.

The basic elements of healthful diets are well established (USDHHS 1988; National Research Council 1989; USDA/USDHHS 1995). They provide adequate amounts of energy and essential nutrients, reduce risks for diet-related chronic diseases, and derive from foods that are available, affordable, safe, and palatable. A very large body of research accumulated since the mid-1950s clearly indicates that healthful diets are based primarily on fruits, vegetables, and grains, with smaller quantities of meat and dairy foods than are typically included in current diets in the United States and other Western countries (James 1988; USDHHS 1988; National Research Council 1989).

Throughout the course of history, societies have developed a great variety of ways to combine the foods that are available to them (as a result of geography, climate, trade, and cultural preferences) into characteristic dietary patterns. In some areas, typical diets have developed patterns so complex, varied, and interesting in taste that they have come to be identified as particular cuisines. Some of these, most notably those of Asia and the Mediterranean, seem to bless the populations that consume them with substantially lower levels of coronary heart disease, certain cancers, diabetes mellitus, and other chronic diseases than those suffered by other peoples. Consequently, such apparent relationships between cuisines and health have created much interest in traditional dietary patterns.

Illustrative is the current interest in Mediterranean diets that has been stimulated by the unusually low levels of chronic diseases and the longer life expectancies enjoyed by adults residing in certain regions bordering the Mediterranean Sea (WHO 1994). Such good health cannot be understood within the context of those factors usually associated with disease prevention in industrialized countries, such as educational levels, financial status, and health-care expenditures. Indeed, the percentages of those who are poor in Mediterranean regions are often quite high relative to those of more developed economies (World Bank 1993). To explain this paradox, researchers have focused on other lifestyle characteristics associated with good health, and especially on the various constituents of the typical Mediterranean diet.

Data from the early 1960s best illustrate the intriguing nature of the paradox. At that time, the overall life expectancy of Greeks at age 45 exceeded that of people in any other nation reporting health statistics to the World Health Organization (WHO/FAO 1993). Subsequently, the ranking of life expectancy in Greece has declined somewhat, at least partly because of undesirable changes in dietary practices that have occurred (Kafatos et al. 1991). But even with such changes, in 1991 life expectancy at age 45 in Greece was an additional 32.5 years, second in rank only to the 33.3 years yet available to Japanese people. By comparison, in the same year, life expectancy at age 45 for adults in the United States, United Kingdom, and Canada was respectively 30.8, 30.9, and 32.1 years (WHO 1994).

Even these brief observations raise interesting historical questions. For example: What, precisely, is a “Mediterranean” diet? When and under what circumstances did it develop? What are the health effects of specific dietary patterns? In what ways do diets change, and what are the health implications of such changes? Should—and could—a Mediterranean-style diet be adopted elsewhere, and if such a diet were to be adopted in, for example, the United States, what would be the impact on agriculture, the food economy, and health patterns? Because such questions address fundamental issues of food and nutrition research and policy, the Mediterranean diet constitutes an especially useful model for studying healthful dietary patterns (Nestle 1994).

Historical Antecedents

Diets of the Ancient Mediterranean

In the absence of written records, knowledge of ancient diets must be inferred from other kinds of evidence. Fortunately, evidence related to Mediterranean diets is extraordinarily abundant, including a vast and extensively documented archaeological record of food debris and a large quantity of food-related art, pottery, tools, and inscribed tablets excavated from prehistoric, Neolithic, Bronze Age, and later sites throughout the region (Fidanza 1979). The evidence also includes information derived from scholarly analyses of the writings of Homer and other classical authors. (The many and varied sources of information about the diets of ancient Egyptians, for example, are summarized in Table V.C.1.1.)

Inferences based on such sources must, however, be tempered by consideration of the difficulties inherent in evaluation: poor preservation of materials, incomplete fragments, errors of oversight, biased opinions, false information, and problems of translation, classification, dating, and interpretation (Darby, Ghalioungui, and Grivetti 1977). Nonetheless, scholars have used these sources over the years to firmly establish the availability in ancient times of an astonishing variety of plant and animal foods, breads, spices, sweets, and beers and wines (Seymour 1907; Vickery 1936; Vermeule 1964).

Discovery of the presence of various foods in a region suggests—but does not prove—that people ate those foods on a routine basis. Reports of actual dietary intake in ancient times are scanty and are especially lacking for the diets of the general population. When classical authors described foods at all, they wrote almost exclusively about those consumed by warriors or noblemen. Such accounts do not seem entirely credible; the writings of Homer, for example, leave the impression that Hellenic heroes consumed nothing but meat, bread, and wine (Seymour 1907). Homeric texts mention vegetables and fruits only rarely, perhaps because such foods were considered inadequate to the dignity of gods and of heroes (Yonge 1909), and olive oil is mentioned only in the context of its use as an unguent (Seymour 1907).

Table V.C.1.1. Sources of information about diets in ancient Egypt Source: Adapted from Darby, Ghalioungui, and Grivetti (1977),1:23.

Archaeology (preserved remains of animals and plants)
Stomach and intestines of human mummies
Tombs (sealed and opened)
Mud bricks
Art (depictions of foods, food preparation, domestic animals)
Temple and tomb paintings and reliefs
Statues, models, dioramas
Papyrus, tomb, or temple texts
Daily food allowances
Lists of food offerings
Foods in medical prescriptions
Cosmology and mythology texts
Greek, Roman, and Arabic texts
Religious texts
Descriptive accounts by travelers, historians, naturalists

Perhaps as a result of such research, scholars have concluded that the typical diet of the common people in ancient times must have been rather sparse, based mainly on plant foods and bread, with meat and seafood only occasional supplements. In fact, such a diet was characteristic of the Mediterranean region even in the early twentieth century (Seymour 1907; Vickery 1936). However, a second- to third-century A.D. review of the food writings of classical poets and authors has provided a vivid contrast. It described foods and drinks in great detail, classifying them by flavor and aroma, means of preparation, and contribution to meals and banquets, suggesting that people of all classes ate and enjoyed a vast array of foods and ingredients (Yonge 1909).

Modern scholars have related dietary practices to the health of ancient Mediterranean populations through inferences from examinations of prehistoric skeletal remains, analyses of sepulchral inscriptions, and other kinds of evidence, as indicated by the listings in Table V.C.1.1. Some of this evidence provides insights into dental lesions, anemia, and other diseases, and taken together, it all suggests that the average life span in ancient Greece and Rome was probably on the order of 20 to 30 years (Wells 1975). The evidence also indicates, however, that this brief life expectancy had much more to do with infection and civil conflict than with malnutrition and starvation (Darby et al. 1977).

Modern History: The Rockefeller Study

The first systematic attempt to investigate dietary intake in the Mediterranean region took place shortly after the end of World War II. In 1948, the government of Greece, concerned about the need to improve the economic, social, and health status of its citizens, invited the Rockefeller Foundation to conduct an epidemiological study on the island of Crete. The aim was to identify factors that would best contribute to raising the standard of living of the Greek population, and Leland Allbaugh, an epidemiologist, was appointed to oversee the study. Allbaugh and his colleagues designed and conducted an extraordinarily comprehensive survey of the demographic, economic, social, medical, and dietary characteristics of the members of 1 out of each 150 households on the island, a sample chosen through a carefully designed randomization process. The foundation published the results of these investigations as a monograph in the early 1950s (Allbaugh 1953).

The report of the survey was remarkable in several important respects. It was, for example, extraordinarily thorough. It included a 75-page appendix that contained descriptions and critical evaluations of statistical methods and a 50-page compendium of the questionnaires used to obtain information. The survey’s numerous dietary components included a review of agricultural data on the Greek food supply; the administration to 128 households of several distinct questionnaires examining cooking practices, daily menus, food expenditures, household food production, and food handling and consumption practices; and three dietary-intake surveys: one of pregnant women and nursing mothers, another of children and adolescents aged 7 to 19 years, and yet another of children aged 1 to 6, with the information in the latter obtained from the parents. These multiple surveys, however, constituted only the most peripheral components of the overall dietary probe.

The core of the survey’s dietary sections consisted of 7-day weighed food inventories collected from the 128 households, 7-day dietary-intake records obtained from more than 500 individuals in those households, and food-frequency questionnaires administered to 765 households. These extensive dietary investigations were conducted in the early fall by volunteer nurses from the Greek Red Cross who, after 5 full days of training, went to live in the survey communities for periods of 7 to 10 days and made daily visits to the sample households. The work of these nurses was closely supervised and their data cross-checked in several ways. Given the extent, complexity, and comprehensiveness of these investigations, it is difficult to imagine that anything like a survey of this magnitude could be initiated—or funded—today.

Table V.C.1.2 compares selected data on Cretan dietary practices obtained through the various dietary survey methods. The methods yielded substantial agreement about the daily amount of energy consumed by the population—an average of 2,500 kilo-calories per day—and the amounts of meat and dairy foods consumed on an average day. Agricultural food “balance” data, which represented the amounts of food available throughout the entire country of Greece on a per capita daily basis, indicated a higher intake of cereals and sugar and a lower intake of potatoes, pulses, nuts, oils, and fats than did data derived from the Crete surveys. However, as is discussed later in this chapter, these differences can be attributed to sources of random and systematic error inherent in methods of dietary-intake measurement. Data on alcohol consumption best illustrate the nature of such errors. Allbaugh was able to explain the discrepancy between the small amount reported in the dietary-intake records and the much larger amount indicated by food balance or household inventory data as a result of systematic underreporting. This was confirmed by his own observations as well as by “an expressed feeling by the respondents that the visiting Americans might be expected to frown upon heavy wine consumption where food was short” (Allbaugh 1953: 106).

Table V.C.1.2. Dietary intake in Crete in 1948 as estimated by three methods Source: Adapted from Allbaugh (1953), p. 107.

Greece Crete
Food 7-day
balance diet Household
1948–9 record inventorya
Energy, MJ (kcal)/d 10.2(2,443) 10.7(2,547) 10.7(2,554)
Foods, kg/person/y
Cereals 158.2 127.7 128.2
Potatoes 30.9 59.1 38.6
Sugar, honey 9.1 5.5 5.5
Pulses, nuts 15.0 20.0 23.2
Vegetables, fruits, olives 120.5 175.9 132.3
Meat, fish, eggs 23.2 28.6 27.7
Milk, cheese 35.0 25.5 34.5
Oils, fats 15.0 30.9 30.9
Wine, beer, spirits 37.7 10.0 38.6

aAdjusted for information obtained from food-frequency questionnaires.

A comparison of the food sources of energy in the diets of people in Crete, Greece, and the United States, as reported in the Rockefeller study, may be found in Table V.C.1.3. It displays data derived from dietary-intake surveys for Crete and reports data taken from food-supply surveys for Greece and the United States, even though these types of data are not truly comparable. The results indicate that plant foods—cereals, pulses, nuts, potatoes, vegetables, and fruits—comprised 61 percent of total calories reported as consumed by people in Crete, whereas plant foods comprised 74 percent of the energy available in the Greek food supply (although not necessarily consumed in the Greek diet) and 37 percent of the energy in the U.S. food supply (again, available but not necessarily consumed).

Table V.C.1.3. Percentage of total energy contributed by major food groups in the diet of Crete as compared to their availability in the food supplies of Greece and the United States in 1948–9 Source: Adapted from Allbaugh (1953), p. 132.

Crete Greece U.S.
(7-day (food (food
record) balance) balance)
Total energy, MJ
(kcal)/d 10.7 (2,547) 10.4 (2,477) 13.1 (3,129)
Energy (%)
Food group
Cereals 39 61 25
Pulses, nuts, potatoes 11 8 6
Vegetables, fruits 11 5 6
Meat, fish, eggs 4 3 19
Dairy products 3 4 14
Table oils, fats 29 15 15
Sugar, honey 2 4 15
Wine, beer, spirits 1 a a

aData not available.

Similarly, foods of animal origin—meat, fish, eggs, and dairy products—comprised only 7 percent of energy in the Cretan diet, in contrast to 19 percent of the energy in the Greek food supply and 29 percent of the energy in the U.S. food supply, but table oils and fats were reported to contribute 29 percent of the energy in the Cretan diet, whereas they constituted only 15 percent of that in the Greek and U.S. food supplies. In Crete, however, 78 percent of the table fats derived from olives and olive oil. The total amount of fat from all sources in the Cretan diet, including that “hidden” in animal foods, was reported as 107 grams per day, or an estimated 38 percent of total energy, a percentage similar to that in the U.S. food supply in the late 1940s (USDA 1968) and considerably higher than that recommended today as a means to reduce chronic disease risk factors (Cannon 1992; USDA 1992; USDA/USDHHS 1995).

The data in Tables V.C.1.2 and V.C.1.3 constitute the basis for the conclusion of the Rockefeller report that “olives, cereal grains, pulses, wild greens and herbs, and fruits, together with limited quantities of goat meat and milk, game, and fish have remained the basic Cretan foods for forty centuries … no meal was complete without bread … Olives and olive oil contributed heavily to the energy intake … food seemed literally to be ‘swimming’ in oil” (Allbaugh 1953: 100). The Rockefeller survey data also indicated that wine was frequently consumed at all meals—midmorning, noon, and evening.

Whether olive oil made such a contribution to the diet for 40 centuries, however, is doubtful. At least one analysis of tree cultivation in southern Italy suggests that olive oil must have been a scarce commodity until at least the sixteenth century and that its principal use in medieval times was in religious rituals (Grieco 1993).

Thus, in attempting to correlate his current observations of dietary intake with the nutritional and general health of the population, Allbaugh noted certain limitations of his study. Few data were available on the nutrient and energy composition of Cretan foods, and virtually no information was available on the clinical and biochemical status of the Cretan population. Nevertheless, the study reported few serious nutritional problems in Crete; those that existed “were limited to a relatively small number of households, living under conditions of very low income and little home production of food” (Allbaugh 1953: 124). Diets generally were nutritionally adequate as measured against the U.S. Recommended Dietary Allowances of that time (National Research Council 1948). The investigators concluded that the diets and food consumption levels observed for most individuals “were surprisingly good. On the whole, their food pattern and food habits were extremely well adapted to their natural and economic resources as well as their needs”(Allbaugh 1953: 31).

This favorable conclusion, however, was one not necessarily shared by the study subjects. Allbaugh reported that only one out of six of the interviewed families judged the typical diet to be satisfactory. He quoted one family as complaining: “We are hungry most of the time” (Allbaugh 1953: 105). When asked what they would most like to eat to improve their diets, survey respondents listed meat, rice, fish, pasta, butter, and cheese, in order of priority. A large majority of respondents (72 percent) listed meat as the favorite food. On the basis of such views, Allbaugh concluded that the diet of Crete could best be improved by providing more foods of animal origin—meat, fish, cheese, eggs—on a daily basis.

Ancel Keys and the Seven Countries Study

Despite the great wealth of information provided by the Rockefeller report, interest in the health implications of Mediterranean diets is more often thought to have begun with the work of Ancel Keys, an epidemiologist from the University of Minnesota. In 1952, impressed by the low rates of heart disease that he had observed on vacations in the Mediterranean (Keys 1995), Keys initiated a series of investigations of dietary and other coronary risk factors with colleagues in seven countries. Keys and his wife, Margaret, have reported the genesis of these investigations in vivid detail:

Snowflakes were beginning to fly as we left Strasbourg on the fourth of February. All the way to Switzerland we drove in a snowstorm
… On the Italian side the air was mild, flowers were gay, birds were singing, and we basked at an outdoor table drinking our first espresso coffee at Domodossola. We felt warm all over. (Keys and Keys 1975: 2)

The two were particularly impressed by the difference between the diet they were eating in Italy and the typical diet consumed by people in the United States. As they described it, the Italian diet included:

[H]omemade minestrone … pasta in endless variety … served with tomato sauce and a sprinkle of cheese, only occasionally enriched with some bits of meat, or served with a little local sea food … a hearty dish of beans and short lengths of macaroni … lots of bread never more than a few hours from the oven and never served with any kind of spread; great quantities of fresh vegetables; a modest portion of meat or fish perhaps twice a week; wine of the type we used to call “Dago red”… always fresh fruit for dessert. Years later, when called on to devise diets for the possible prevention of coronary heart disease we looked back and concluded it would be hard to do better than imitate the diet of the common folk of Naples in the early 1950s. (Keys and Keys 1975: 4)

Keys and his colleagues published the results of their Neapolitan investigations, which found Italian diets to be remarkably low in fat—20 percent of energy, or just half the proportion observed in the diets of comparable American groups (Keys et al. 1954). By that time (and long before such ideas became commonplace), Keys had associated the typical American diet, rich in meat and dairy fats, with higher levels of blood cholesterol and, therefore, with increased risk of coronary heart disease.

In 1959, the the principal lines of evidence for these associations were reviewed in a cookbook designed to help the general public reduce risks for coronary heart disease (Keys and Keys 1959). In a foreword to this volume, the eminent cardiologist Paul Dudley White, who had made several expeditions with the authors “to study the health and the ways of life of native populations” in southern Italy and Crete, extolled both the health benefits and the taste of the lowfat foods—and the wine—that they had routinely consumed during their Mediterranean travels.

In this cookbook, perhaps the first of the “healthy heart” genre, the authors summarized their “best advice” for lifestyle practices to reduce coronary risk (Keys and Keys 1959: 40). Table V.C.1.4 lists their precepts in comparison to the 1995 U.S. dietary guidelines for health promotion and disease prevention (USDA/USDHHS 1995). As is evident, the guidelines closely follow the 1959 advice that Ancel and Margaret Keys derived from their observations of diet and coronary risk in southern Italy and Crete. This comparison demonstrates that the Mediterranean diet of the 1950s can be considered to constitute the original prototype for development of current dietary guidance policy in the United States.

Beginning in the early 1950s, and for more than 20 years thereafter, Keys and his colleagues identified dietary and other risk factors for coronary heart disease through a large-scale study of nearly 13,000 middle-aged men from 7 countries distributed among 16 cohorts (Keys 1970; Keys et al. 1980). The overall results of the Seven Countries Study provided strong epidemiological evidence for the effects of fat and various fatty acids on serum cholesterol levels and on coronary heart disease risk (Kromhout, Menotti, and Blackburn 1994).

Dietary-intake data for foods and food components other than fat, however, were published in English for the first time only in 1989 (Kromhout et al. 1989). That report compared the 16 cohorts in the 7 countries with respect to their intake of bread, cereals, various vegetables, fruit, meat, fish, eggs, dairy foods, table fats, pastries, and alcoholic beverages. These data confirmed that Mediterranean diets in the early 1960s were based primarily on foods from plant sources, but that some versions were higher in fat—mainly olive oil—than might be expected in a population with such good health. The Seven Countries’ data, as confirmed by subsequent investigations (Cresta et al. 1969; Kafatos et al. 1991; Trichopoulou et al. 1992), constituted the principal research basis for the proportions of foods from plant and animal sources proposed recently as a Mediterranean diet pyramid (Willett et al. 1995) or a Greek column (Simopoulos 1995).

Table V.C.1.4. Ancel and Margaret Keys’ 1959 dietary advice for prevention of coronary heart disease compared to the 1995 U.S. dietary guidelines Sources: Adapted from Keys and Keys (1959) and USDA/USDHHS (1995).

The Keyses’ “best advice” 1995 dietary guidelines
Eat a variety of foods.
Do not get fat; if you are fat, Balance the food you eat
reduce. Get plenty of exercise with physical activity–
and outdoor recreation. maintain or improve your
Restrict saturated fats, the fats Choose a diet low in fat,
in beef, pork, lamb, sausages, saturated fat, and
margarine, solid shortenings, fats cholesterol.
in dairy products. Prefer vegetable
oils to solid fats, but keep total
fats under 30 percent of your diet
Favor fresh vegetables, fruits, Choose a diet with plenty
and nonfat milk products. of grain products,
vegetables, and fruits.
Avoid heavy use of salt Choose a diet moderate in
and refined sugar. salt and sodium. Choose
a diet moderate in sugars.
Be sensible about cigarettes, If you drink alcoholic
alcohol, excitement, business beverages, do so in
strain. moderation.


One additional large-scale study, from a rather unexpected source, yielded comparative information about dietary intake in the Mediterranean and other regions of Europe. From 1963 to 1965, the European Atomic Energy Commission (EURATOM) examined household food consumption among 3,725 families in 11 regions of 6 European countries in an effort to identify the foods among those most commonly consumed that were likely to be sources of radioactive contaminants. Investigators conducted dietary interviews for 7 consecutive days in each of the selected households and weighed all foods present in the households on those days. After applying several correction factors, the researchers converted the data on household food consumption to daily average amounts of food consumed per person. These data were published in 1969 (Cresta et al. 1969).

Of the regions selected by EURATOM for the study, nine were in the north of Europe and two in the south. One of the northern regions was in Italy (Friuli). Because both of the southern regions also were in Italy (Campania and Basilicata), the data could be used to compare the typical dietary intake of the Italian north—which was quite similar to dietary patterns throughout the rest of northern Europe—with that of the Mediterranean regions. A detailed comparative analysis of these data is now available (Ferro-Luzzi and Branca 1995).

The EURATOM study revealed distinct differences in dietary-intake patterns between the northern and southern Italian regions. Diets in the Mediterranean areas were characterized by a much greater intake of cereals, vegetables, fruit, and fish, but a much smaller intake of potatoes, meat and dairy foods, eggs, and sweets. Although no consistent differences were observed in overall consumption of table fats, the foods contributing to total fat intake were quite different. Consumption of butter and margarine was much higher in the north, whereas in the south, the principal fat was olive oil, and margarine was not consumed at all. Taken together, the results of the EURATOM study provide further evidence that the Mediterranean diet of the mid-1960s was based predominantly on plant foods and included olive oil as the principal fat.

Recent Observations

In the years following these investigations, the Keys’ description of the role of diet in coronary risk has become more widely accepted (James 1988; USDHHS 1988; National Research Council 1989). Along with this acceptance has come increasing recognition that the traditional dietary patterns of many cultures meet current dietary guidelines and that the cuisines of these cultures—especially those of Mediterranean and Asian countries—could serve as models for dietary improvement (Nestle 1994). In recent years, reports of investigations of the scientific basis and health implications of Mediterranean diets have been published in at least five edited collections of papers (Helsing and Trichopoulou 1989; Spiller 1991; Giacosa and Hill 1993; Serra-Majem and Helsing 1993; Nestle 1995). Public interest in Mediterranean diets has been stimulated by numerous articles in the popular press (Kummer 1993; Hamlin 1994), and their palatability has been celebrated in cookbooks emphasizing the dual themes of good taste and good health (Shulman 1989; Goldstein 1994; Jenkins 1994; Wolfert 1994).

Historical and Research Issues

As noted previously, studies of Mediterranean dietary patterns raise research issues that are also applicable to a more general understanding of the role of diet in health.

Definition of the Mediterranean Diet

The peoples of the 16 or more countries that border the Mediterranean Sea vary greatly in culture, ethnicity, religion, economic and political status, and other factors that might influence dietary intake, and their food supplies vary widely in the quantity used of every item that has been examined. Thus, the identification of common dietary elements within the region has proved a challenging task to researchers (FerroLuzzi and Sette 1989; Giacco and Riccardi 1991; Varela and Moreiras 1991; Giacosa et al. 1993).

Because the studies of Ancel Keys found the typical dietary pattern of the Greek island of Crete in the 1950s and 1960s to be associated with especially good health, this pattern has come to be viewed as the model, and because olive oil was a principal source of fat in the Cretan diet, the model has been extended to include diets consumed in olive-producing Mediterranean regions. In this manner, the generic term “Mediterranean diet” has come to be used, in practice, as referring to dietary patterns similar to those of Crete in the early 1960s and other regions in the Mediterranean where olive oil is the principal source of dietary fat (Willett et al. 1995).

Dietary Epistemology: Research Methods

Knowledge of the content of Mediterranean diets in the early 1960s—or at present—necessarily depends upon the reliability of methods used to determine the typical food intake of the population. National diet surveys, such as those that are conducted regularly in the United States, have not been generally available in Mediterranean countries. The Rockefeller study of Crete was a notable exception, remarkable by any standard of epidemiological investigation (see Table V.C.1.2) in its use of multiple methods, lengthy personal interviews, and critical analysis of results to attempt to define dietary intake (Allbaugh 1953). The Seven Countries Study also used multiple methods. For most of the 16 cohorts, Keys and his colleagues obtained 7-day diet records from small subsamples of each group and corroborated these records by analyzing the energy and nutrient composition of weighed, duplicate meals. For a few cohorts, investigators collected dietary data from 24-hour recalls as verified through food-frequency questionnaires (Keys 1970). Finally, the EURATOM study attempted to corroborate daily reports of household food intake by weighing all foods present in the house on each of the seven consecutive interview days (Cresta et al.1969).

These investigations were designed to overcome fundamental flaws in each of the methods commonly used to evaluate the dietary intake of individuals and populations; all provide opportunities for random and systematic errors in reporting food intake, estimating serving sizes, and determining nutrient content (Mertz 1992; Buzzard and Willett 1994; Young and Nestle 1995). Such problems are compounded in studies that attempt to compare dietary-intake data from one country to another, or within one country over time. If the methods for determination of dietary intake differ, their results are not strictly comparable—a situation similar to comparing apples to oranges (see Tables V.C.1.2 and V.C.1.3).

For purposes of international comparison, investigators must often rely on food-balance data—agricultural data on specific commodities present in the food supply from one year to the next. As already noted, these data are distinctly different from those that describe dietary intake. They reflect the amounts of specific foods produced in a country during a given year, with imports of foods added and food exports subtracted, expressed on a per capita basis through dividing by the population total on a defined day of the year.

Such data are also known by other names: food supply, food availability, food disappearance, and food consumption. Among these terms, “consumption” is a misnomer, because food-balance data are only an indirect estimate of dietary intake. A food that is produced but then wasted, fed to animals, or used for industrial purposes is not consumed; for many foods, therefore, food-balance data overestimate dietary intake. In the case of foods produced at home, however, food-balance data underestimate consumption. The average annual per capita availability of a food commodity only rarely—and accidentally—is an accurate measure of actual consumption by an individual man, woman, or child. These limitations may explain observed discrepancies in study results, and they emphasize the need for caution in interpreting comparative data such as those presented in Tables V.C.1.2 and V.C.1.3.

Despite such limitations, food-balance data are often the best—or only—data available to estimate time trends in dietary practices, and they are used frequently in comparative descriptions of Mediterranean diets (Ferro-Luzzi and Sette 1989; Helsing 1995). Three agencies of the United Nations (UN) produce such data. The Organization of Economic Cooperation and Development (OECD) has published data for the supply and use in 23 countries of specific food items, such as pork, cheese, or olive oil, from 1979 to 1988 (OECD 1991).

The Food and Agriculture Organization (FAO) publishes individual food-balance sheets for 145 countries that include data for per capita supply of major food groups (e.g., meat, legumes, alcohol); its most recent edition provides data in 3-year averages from 1961–3 through 1986–8 (FAO 1991). The World Health Organization (WHO) Regional Office for Europe has established a comprehensive computerized database that incorporates FAO food-balance data as well as the WHO annual health statistics since 1961 for each of the countries that supply such data to the UN (WHO/FAO 1993). This program makes it possible to generate an immediate display of the relationship between the availability of any food and the disease rates in any country of interest (Ferro-Luzzi and Sette 1989; Helsing 1995).

Health Impact

By the definition used here, the Mediterranean diet can be considered a near-vegetarian diet. As such, it would be expected to produce the well-established health benefits of vegetarian diets and to solve any deficiencies of energy or micronutrients (especially vitamin B 12) that are occasionally associated with such diets (Johnston 1994). Vegetarian or near-vegetarian diets are especially plentiful in key nutrients, particularly antioxidant vitamins, fiber, and a variety of phenolic compounds that have been identified as protective against cancer and other chronic diseases (Dwyer 1994; Kushi, Lenart, and Willett 1995a, 1995b). Researchers, however, have yet to establish the relative contribution of any single nutrient or food component, the foods that contain such factors, or physical activity and lifestyle patterns—alone or in combination—to the favorable health indices observed in the Mediterranean region.

In this context, the role of olive oil is of particular interest. The Greek diet, for example, contains a higher proportion of fat than is usually recommended. Yet much of this fat is olive oil, and the diet is associated with very good health. Diets rich in olive oil are associated with exceptionally low rates of coronary heart disease, even when blood cholesterol levels are high (Verschuren et al. 1995). The traditional Greek diet is also associated with an exceptionally low risk for breast cancer (Trichopoulou et al. 1995).

Changing Dietary Patterns

If it is indeed true that Mediterranean diets of the 1960s protected adult populations against premature death, it would seem highly desirable to preserve the protective elements of those diets. Evidence from dietary-intake surveys and from food-balance data indicates, however, that dietary patterns throughout the region are changing rapidly, and generally in an undesirable direction. For example, one dietary-intake study of an urban population on Crete (obtained by 24-hour diet recalls corroborated by food models, photographs, and clinical and biochemical measurements) reported an increase in the intake of meat, fish, and cheese but a decrease in the intake of bread, fruit, potatoes, and olive oil (Kafatos et al. 1991) from levels reported by Keys and his colleagues in the early 1960s (Kromhout, Keys, Aravanis, et al. 1989). Similar changes have been observed in Italy (Ferro-Luzzi and Branca 1995). Food-balance data also document large increases in the availability of meat, dairy foods (FAO 1991), and animal fats (Serra-Majem and Helsing 1993) throughout the region since the early 1960s. Given this situation, the traditional Mediterranean diet may well become a historical artifact.

Increasing evidence suggests that the recent changes in Mediterranean dietary patterns have been accompanied by increases in chronic disease risk factors among the populations. These risk factors include a decline in levels of physical activity, along with higher levels of serum cholesterol (Kafatos et al. 1991), hypertension, and obesity (Spiller 1991). Associated with these changes in risk factors are reports of rising rates of coronary heart disease, diabetes (Spiller 1991), and several types of diet-related cancers (LaVecchia et al. 1993) in several Mediterranean countries. These trends confirm well-established relationships between diet and chronic disease risk (James 1988; USDHHS 1988; National Research Council 1989) and suggest the need to reverse current practices through widespread efforts at preserving and promoting traditional diets within the region (Nestle 1994).

Preservation and Adaptation

Overall dietary patterns in a country are the result of an ongoing interaction between culturally determined food traditions and the assimilation of new foods through economic improvement, foreign contact, or international food marketing. Education also has a role in influencing personal food preferences and dietary change (Heimendinger and Van Duyn 1995). Until recently, Mediterranean dietary patterns were quite resistant to change. Allbaugh and Keys both remarked on the similarity of the foods commonly eaten in Italy and Crete to those produced and consumed in those areas in the ancient past. Despite suggestions that traditional dietary patterns are beginning to be abandoned (Alberti-Fidanza et al. 1994), such foods are still routinely consumed by at least some older population groups (Trichopoulou, Katsouyanni, and Cnardellis 1993).

Issues related to the assimilation of Mediterranean dietary patterns within other countries are best illustrated by the adaptation of southern Italian foods to American tastes (Levenstein and Conlin 1990). Italian immigrants of the late nineteenth and early twentieth centuries retained many of their food traditions despite North American–held views of their diets as insufficiently nutritious, indigestible, unsanitary, and inadequate in amounts of milk and meat. Such views, however, began to change during the economic restrictions of World War I, when Italian pastas became popular as inexpensive, well-balanced alternatives to meat, and since the 1920s, Italian food products have been widely marketed in the United States (Levenstein 1985). But today many Italian-style foods have been “Americanized” to the point that they are far higher in energy, fat, cholesterol, and sodium than the traditional foods from which they were derived (Hurley and Liebman 1994).

Policy Implications

Policies designed to encourage consumption of traditional diets within their country of origin, or to promote the adaptation of traditional models to new locations, will have to address many well-defined cultural, economic, and institutional barriers (Nestle 1994). They will also need to recognize that diet is only one of a great many behavioral factors that influence health and that other determinants may command higher national priorities for action (Jamison and Mosley 1991). Moreover, the transfer of traditional Mediterranean dietary patterns to a country such as the United States would be likely to affect agriculture, the food industry, the overall economy, and the environment in highly complex ways, some of which may be beneficial, but others undesirable (Gussow 1994, 1995; O’Brien 1995).

The role of the Mediterranean diet in U.S. dietary guidance policy is of particular interest. As is demonstrated in Table V.C.1.4, the Mediterranean observations of Keys’ led directly to the formulation of dietary guidelines for the prevention of coronary heart disease. In turn, such guidelines eventually encompassed more general advice for health promotion and disease prevention in American statements of dietary guidance policy, as expressed in the Dietary Guidelines for Americans (USDA/USDHHS 1995). Because animal foods are principal sources of fat, saturated fat, and cholesterol in American diets (Gerrior and Zizza 1994), dietary guidelines necessarily should promote predominantly plant-based diets similar to those traditionally consumed in the Mediterranean region or in Asia. That this may not be evident from standard American food guides (USDA 1992) is, at least in part, a result of political pressures from producers of meat and dairy foods to ensure that their products retain a dominant position in the American food supply and diet (Nestle 1993). Such pressures may well have resulted in dietary recommendations that are ambiguous and confusing to the public (Nestle 1995a).

Research Directions

Traditional Mediterranean diets appear to have been based mainly on plant foods, to contain foods from animal sources in very small amounts, to use olive oil as the principal dietary fat, to feature alcohol in moderation, and to balance energy intake with energy expenditure. Substantial research—in quantity and quality—supports the very great health benefits of just such dietary and activity patterns (Willett 1994; Kushi et al. 1995a, 1995b).

Mediterranean diets are consistent with current food guide recommendations for public-health promotion and disease prevention, as well as with recommendations for nutritionally adequate vegetarian diets (Haddad 1994). Because they also are appreciated for their gastronomic qualities, they are well worth further study as a cultural model for dietary improvement. Several areas of historical and applied research related to Mediterranean diets seem especially worthy of additional investigation; these are listed in Table V.C.1.5. While awaiting the results of such studies, immediate efforts should be instituted to preserve the ancient—and healthful—dietary traditions within the Mediterranean region and to encourage greater consumption of plant foods among industrialized populations as a means to improve health.

Table V.C.1.5. Suggestions for further historical and applied research on the health impact of Mediterranean diets

Historical research needs
Identification of methods to determine the typical dietary intake
of individuals and populations in Mediterranean countries in
the past, present, and future.
Identification of methods to determine time trends in Mediter-
ranean dietary patterns.
Determination of the impact of dietary changes on nutritional
status and health risks in Mediterranean countries in the past
and present.
Identification of cultural, behavioral, economic, and environ-
mental determinants of dietary change in Mediterranean coun-
tries in the past and present.
Determination of the impact of dietary changes on the agricul-
ture, food industry, economy, and environment of Mediter-
ranean countries in the past and present.
Determination of the impact of adoption of Mediterranean
foods or dietary patterns on the agriculture, food industry,
economy, and environment of countries outside the Mediter-
ranean region in the past and present.
Applied research needs
Identification of the roles of specific plant foods characteristic
of Mediterranean diets—fruits, vegetables, legumes, cereals,
nuts, oils, wine—in health promotion and disease prevention.
Identification of the roles of specific plant-food nutrients—vita-
mins, minerals, monounsaturated fatty acids, linolenic acid,
fiber, alcohol, phytochemicals—in the low rates of chronic dis-
eases observed in Mediterranean countries.
Determination of the proportions of plant and animal foods in
Mediterranean diets optimal for reducing disease risk.
Determination of the proportion of energy from fat and specific
fatty acids in Mediterranean diets associated with the lowest
risk of disease.
Development of dietary recommendations and food guides that
best reflect current scientific knowledge of the health benefits
of Mediterranean diets.
Identification of effective methods to educate the public in
Mediterranean countries about traditional dietary practices that
best promote health.