ABSTRACT We evaluated the role of a Mediterranean-style diet in preventing acute coronary syndromes (ACS). Data from CARDIO2000, a multi-centre retrospective case–control study investigating the association between ACS and demographic, nutritional, lifestyle and medical risk factors were used. We studied 661 patients hospitalized for a first ACS event and 661 matched controls without clinical suspicion of cardiovascular disease. The Mediterranean diet significantly reduced (by 16%) the risk of developing ACS. The association remained significant in the presence of hypertension, hypercholesterolaemia, sedentary lifestyle, diabetes mellitus or a combination of two of these cardiovascular risk factors. Our findings illustrate the importance of the Mediterranean diet in the primary prevention of acute coronary events.

RESUME Nous avons évalué le rôle d’un régime alimentaire de type méditerranéen dans la prévention des syndromes coronariens aigus. Les données de CARDIO2000, une étude cas-témoins rétrospective, multicentrique examinant l’association entre les syndromes coronariens aigus et des facteurs de risque démographiques, nutritionnels, médicaux et liés aux modes de vie, ont été utilisées. Nous avons étudié 661 patients hospitalisés pour un premier événement lié à un syndrome coronarien aigu et 661 témoins appariés sans suspicion clinique de maladie cardio-vasculaire. Le régime méditerranéen réduit significativement (de 16 %) le risque de survenue des syndromes coronariens aigus. L’association demeure significative en présence d’hypertension, d’hypercholestérolémie, de sédentarité, de diabète sucré ou d’une combinaison de deux de ces facteurs de risque cardio-vasculaire. Nos conclusions montrent l’importance du mode d’alimentation méditerranéen dans la prévention primaire des événements coronaires aigus.

There is extensive scientific evidence of the association between diet and incidence of coronary heart disease, various types of cancer and other diseases [1–4]. Dietary factors exert their influence largely through their effect on blood lipids and lipoproteins, as well as on other established modifiable risk factors, with the exception of cigarette smoking [2,3]. Based on the results of the Seven Countries Study of the early 1970s [1] and the Lyon Diet Heart Study of the late 1990s [2], many investigators have recognized the beneficial role of a Mediterranean-style diet (Mediterranean diet) in cardiovascular diseases, metabolic disorders and several types of cancer [2,3,5–7].

The effects of a Greek-style Mediterranean diet on overall mortality or morbidity have been evaluated using the results of the Seven Countries Study (which included two rural male populations from Crete and Corfu) or other local studies (including sub-groups from rural and urban Greek areas) [3,8,9]. However, due principally to the small number of enrolled subjects, these latter studies lacked the power to clarify the relationship between a Mediterranean diet and the risk of acute coronary syndromes.

We evaluated whether the adoption of a Mediterranean diet was associated with a reduction in the risk of developing acute coronary syndromes, independent of the conventional cardiovascular risk factors. Stratified, random sampling from all Greek regions ensured representation of various cultural, socioeconomic and behavioural characteristics that clustered in the population and might have influenced the results.

The CARDIO2000 was a multicentre retrospective case–control study that investigated the association between several demographic, nutritional, lifestyle and medical risk factors with the risk of developing acute coronary syndromes. From January 2000 to March 2001, 661 cardiac patients and 661 hospitalized controls without any past or present clinical symptoms or suspicions of cardiovascular disease were enrolled in the study. This number of subjects was used to give the study sufficient power to evaluate differences in coronary risk > 7% (statistical power > 0.80; significance level, P < 0.05).

In order to reduce any unbalanced distribution of measured or unmeasured confounders, patients and controls were randomly selected. A sequence of random numbers (1 or 0) was applied to hospital admission listings. Coronary patients assigned a ‘1’ were enrolled in the study. The same procedure was used for controls, after taking into account matching criteria. Only patients hospitalized for an acute first event of coronary heart disease (acute myocardial infarction or unstable angina) were eligible as cases. Stable angina patients were excluded. Specific inclusion criteria for cardiac patients were a first event of acute myocardial infarction diagnosed by two or all of typical electrocardiographic changes, compatible clinical symptoms or specific diagnostic enzyme elevations or by unstable angina, not previously diagnosed, corresponding to class III of the Braunwald classification.

For each patient, a control was randomly selected from hospitalized patients without any clinical symptoms and signs or suspicions of cardiovascular disease in their medical history. Controls were matched to cases by age (± 3 years), sex and region. Controls were drawn mainly from patients in surgical clinics (urology, ophthalmology or orthopaedic clinics). They attended, as far as possible, the same hospital at the same time as the coronary cases. In a few cases, particularly in country hospitals, where the available number of hospitalized controls was insufficient for the matching procedure, we enrolled into the study friends or colleagues of the coronary patients. We used hospitalized controls to obtain more accurate medical information, to eliminate the potential adverse effect of several unknown confounders, to increase the likelihood that cases and controls would share the same study base and to reduce the problem of misclassification [5]. Medical information was retrieved through hospital or insurance records. Demographic and lifestyle data were obtained through a confidential questionnaire that included structured questions concerning lifestyle habits and sociodemographic background factors. Interviews took place after the third day of hospitalization. The same physicians who took the medical history and gave the physical examination also evaluated the controls.

Hypertension was defined by systolic blood pressure > 140 mmHg and/or diastolic blood pressure > 90 mmHg or by special anti-hypertensive medication [6]. Hyper-cholesterolaemia was defined as cholesterol levels > 220 mg/dL [6] and diabetes as glucose concentrations greater than 125 mg/dL. Body mass index (BMI) was calculated at entrance by dividing the subject’s weight by height squared (kg/m2). According to the collected medical records, the majority of controls (78%) and patients (72%) had at least one laboratory measurement during the past 12 months. In addition, we took total cholesterol and blood glucose measurements during the first 12 hours of hospitalization. Quantification of smoking status was based on the calculation of pack years adjusted for nicotine content equal to 0.8 mg per cigarette. The physically active were defined as those who reported engaging in non-occupational physical activity more than once per week. The rest were considered physically inactive, i.e. leading a sedentary lifestyle.

At least 16 countries that border the Mediterranean Sea can be defined as ‘Mediterranean’ countries. Between these populations, many cultural, ethnic, religious, economic and agricultural differences may exist, resulting in different dietary habits and precluding a single definition of a ‘Mediterranean diet’. The evaluation of nutritional habits was based on guidelines from the Department of Nutrition of the National School of Public Health [8]. In particular, we investigated the average consumption during the past year of red meat, chicken, fish, vegetables, legumes, pasta, salads, cereals, diary products, sweets and fruits. For each of the fourteen food items, the frequency of consumption was quantified approximately in terms of the number of times per month the food was consumed [9]. Thus, weekly consumption was multiplied by a factor of 4 and a value of 0 was assigned to food items rarely or never consumed. In order to perform an objective assessment of nutritional characteristics, food frequencies were calculated as quantities (g/day) on the basis of standard portion size estimations [9].

The traditional Mediterranean diet is characterized by olive oil as an important fat source, high intake of fruits, vegetables, bread, other cereals, potatoes, poultry, beans, nuts, fish, little red meat, dairy products and moderate consumption of alcohol. The pattern is based on food patterns typical of many regions in Greece and southern Italy [2,3]. To describe the total diet, we used composite scores, which are necessary for the evaluation of epidemiological associations [9].

The Mediterranean diet can be scored in terms of eight component characteristics:

We defined subjects as consumers of this type of diet by using as cut-off points the median values of the monthly food consumption score [9]. Finally, alcohol consumption was measured by daily ethanol intake in wine glasses (100 cc, 12% ethanol).

Continuous variables are presented as mean values ± one standard deviation, while qualitative variables are presented as absolute and relative frequencies. The Pearson correlation coefficient was applied to measure associations between the continuous variables. Contingency tables with calculation of chi-squared test and application of Student t-test evaluated associations between categorical and continuous variables. The chi-squared test for homogeneity was applied to check differences within the investigated groups. Estimation of the relative risks of developing acute coronary syndromes under several hypotheses was performed by calculating the odds ratio (OR) and corresponding confidence intervals (CI) through multiple conditional logistic regression analysis. A final model was developed through stepwise elimination procedures for the selection of variables, using 5% for the probability for entering a variable and 10% for the probability of removing it from the model. Deviance residuals were used to evaluate the model’s goodness-of-fit [10]. All reported P-values are two-sided and have a significance level of 5%. STATA 6 software was used for the calculations (STATA Corp., College Station, Texas, United States of America).

Table 2 shows the effect of the Mediterranean diet on the prevalence of hypertension, hypercholesterolaemia, diabetes mellitus, sedentary life and BMI among the groups of study.

Adoption of the Mediterranean diet has been shown to be inversely associated with the incidence of acute coronary syndromes and the prevalence of conventional cardiovascular risk factors in both groups in the study. To test the hypothesis that the effect of the Mediterranean diet on coronary risk is independent of other risk factors, we developed multivariate risk models, adjusted for sex, age and region by design, BMI, physical activity level, smoking habit, blood pressure and cholesterol levels. Consuming a Mediterranean diet led to an adjusted 23% reduction of coronary risk (OR = 0.77, 95% CI: 0.69–0.85, P = 0.003). However, stratifying our analysis by region (rural, rural-urban and urban), we found significant differences in the estimated OR (F-test = 4.59, P = 0.012). Thus, the corrected OR was equal to 0.84 (95% CI: 0.73–0.96, P = 0.041).

Further analysis showed that adoption of the Mediterranean diet was associated with a reduction in the risk of developing acute coronary syndromes in the presence of several unhealthy conditions such as hypertension, hypercholesterolaemia, diabetes mellitus and sedentary lifestyle. Conversely, in current smokers, adoption of the Mediterranean diet did not significantly influence coronary risk. Further, in the presence of a combination of three or more risk factors, the effect of Mediterranean diet lost its significance.

In this study, we evaluated the effect of the Mediterranean diet on the risk of developing acute coronary syndromes. Coronary patients and controls that consistently consumed this type of diet had a lower prevalence of hypertension, hypercholestero- laemia and diabetes mellitus and had lower BMI (Table 1). These findings may partially explain the beneficial effect of diet on cardiovascular disease. However, multivariate analysis, after taking into account the presence of the aforementioned risk factors, provided evidence that an a priori defined nutritional pattern, i.e. the Mediterranean diet, favourably affected coronary risk, independent of the presence of several cardiovascular risk factors (Figures 1 and 2).

World Health Organization mortality statistics have documented the long survival of people in southern European Mediterranean countries despite a high prevalence of smoking and gaps in available health services [11]. Many investigators have attributed this outcome to several cultural and behavioural differences between the populations, including nutritional particularities, to provide a plausible explanation for what is widely considered an ‘ecological paradox’ [9]. Based on the results of the Seven Countries study in the late 1970s [1,3,10–13], the protective role of the Mediterranean diet against atherosclerosis has been partially explained by lower cholesterol and blood pressure levels. Similar results were observed in the present study, where subjects who had adopted this type of diet were observed to have a lower BMI and lower blood glucose concentrations. Recent findings of the Lyon Diet Heart Study highlight the potential importance of the Mediterranean dietary pattern within the context of the American Heart Association (AHA) Step I diet, compared to other recommended diets [2,14]. The Lyon study concluded that subjects who consumed a Mediterranean diet had a 50%–70% lower risk of recurrent heart disease, compared to those who followed a diet similar to the AHA Step I diet [2,11]. These results accord with results from the Seven Countries study [1,14,15], the Italian Nine Communities study [16] and from studies involving Finnish [17] and Scottish groups [18]. All confirm a plausible pathway by which diet might influence coronary risk.

The Greek type of Mediterranean diet is low in saturated fat, high in monounsaturated fat (mainly from olive oil), high in complex carbohydrates from legumes and high in fibre, mostly from vegetables and fruits. Total fat intake may be high (about 40% of total energy intake), but the monounsaturated:saturated fat ratio is around 2. Daily foods include large quantities of bread, pasta, legumes, vegetables, cooked meals, soups and salads rich in olive oil. Intake of milk is rather low, but the consumption of feta cheese and yoghurt is high. Also, people prefer to consume fish rather than meat, especially in the rural and rural-urban areas. The high content of vegetables, fresh fruits, cereals and olive oil guarantee a high intake of beta-carotene, vitamins C and E, polyphenols and various important minerals. Finally, wine is consumed in moderation and almost always during meals [9,11–13]. According to several investigators, these are the key elements responsible for the beneficial effects of diet on human health [1–3,9]. Others believe that the effect of the Mediterranean diet on coronary risk may be explained through a number of confounders, such as geographic and other non-measured cultural and social differences of the particular populations under study [19].

In summary, the present study, in addition to confirming the previously established relationship between diet and conventional cardiovascular risk factors, raises the issue of an independent association between the Mediterranean diet and lower risk of acute coronary syndromes. Prospective studies are needed in order to confirm our findings by identifying the factors, i.e. the pathway, by which Mediterranean diet and olive oil are associated with lower coronary risk.

In this type of study, two main errors may occur: selection and recall bias. To eliminate selection bias, we sought to set objective criteria for both patients and controls. To minimize recall bias, we made use of detailed data from patients’ medical records. However, recall bias may still exist, especially in the measurement of nutritional habits, smoking and the onset of cardiovascular risk factors. We sought to minimize the potential effect of uncontrolled or unknown confounders through multivariate analysis and by using the same study base for patients and controls.

This study was supported by research grants from the Hellenic Heart Foundation (11/1999–2001). The authors would like to thank the physicians and the specialists who coordinated this study: Dr C. Tzioumis (Athens, Crete, Pelloponisos), Dr I. Papaioannou (Athens, Thessalia), Dr I. P. Starvopodis (Ionian Islands), Dr L. Karra (Aegean Islands), Dr D. Antoniades (Macedonia), Dr G. Rembelos (Aegean Islands), Dr D. Markou (Athens), A. Moraiti (Athens), D. Evagelou (Crete), Dr S. Vellas (Attica, Hpeirous), B. Meidanis (Macedonia, Sterea Hellas, Thessalia), Dr S. Loggos (Attica), Dr I. Elefsiniotis (Athens), Dr N. Marinakis (Aegean Islands), Dr G. Koutsimbanis (Thrace), and Dr G. Kyratzoglou (East Macedonia).