Professional Documents
Culture Documents
Ajcn/82 5 964
Ajcn/82 5 964
ABSTRACT various populations (2, 3), which suggests the protective effect of
Background: Epidemiologic studies link Mediterranean-type diets traditional Mediterranean-type diets. The more recent MONICA
to a low incidence of cardiovascular disease; however, few dietary survey, conducted in Europe, confirmed the South-North gra-
intervention studies have been undertaken, especially in primary dient in CHD mortality and its relation to dietary habits (4).
prevention. Even more recently, 3 cohort studies in Spain and Greece have
Objectives: In the Mediterranean Diet, Cardiovascular Risks and provided data supporting a cardiovascular protective effect of
Gene Polymorphisms (Medi-RIVAGE) study, the effects of a Mediterranean diets (5-7). These studies, together with other
Mediterranean-type diet (Med group) or a low-fat diet (low-fat reliable epidemiologic studies (8), provide the basis for the
group) on risk factors were evaluated in 212 volunteers (men and
well-known traditional Mediterranean diet pyramid and
women) with moderate risk factors for cardiovascular disease.
healthy eating model (9).
Design: After the 3-mo dietary intervention, changes in many risk
Nevertheless, drastic changes in food habits in Mediterranean
factors were evaluated. Dietary questionnaires and plasma nutri-
populations have recently occurred, and the traditional Mediter-
tional markers were used to test compliance.
Results: Although the dietary goals were only partially reached, ranean model now seems restricted to elderly people and rural
changes in dietary habits were observed in both groups (n ҃ 169): areas (10). Indeed, even in Crete, the progressive upward trend in
protein, carbohydrate, and fiber intakes increased and fat quality total and saturated fat intakes is concomitant with a marked
(decreased saturated fat and increased monounsaturated or polyun- increase in obesity and cardiovascular disease mortality (11).
saturated fat) improved. BMI, total and triacylglycerol-rich lipopro- Although individual nutrients common in the traditional Med-
tein (TRL) cholesterol, triacylglycerols, TRL triacylglycerols, apo- iterranean diet have been individually tested in some studies
lipoproteins A-I and B, insulinemia, glycemia, and the homeostasis (12–14), very few intervention studies have looked at a global
model assessment score were significantly lower after 3 mo. The diet (15). The Lyon Heart Study for secondary prevention of
reductions in total cholesterol, triacylglycerols, and insulinemia re- CHD (16, 17) showed the positive effect of a Mediterranean-type
mained significant after adjustment for BMI. There was a trend for
a diet-by-time interaction for LDL cholesterol (P ҃ 0.09). Our data 1
From the Human Nutrition and Lipids Joint Research Unit 476-
predicted a 9% reduction in cardiovascular disease risk with the INSERM, National Institute of Health and Medical Research/1260-National
low-fat diet and a 15% reduction with this particular Mediterranean Institute of Agronomic Research, Faculty of Medicine Timone, Université de
diet. la Méditerranée, Marseille, France (SV-B, CD, OH, RP, M-JA-C, and DL);
the Analytical Chemistry Laboratory, Faculty of Pharmacy, Marseille,
Conclusion: After a 3-mo intervention, both diets significantly re-
France (CD and HP); the Cancer Research Centre, INSERM-CRLC Val
duced cardiovascular disease risk factors to an overall comparable
d’Aurelle, Montpellier, France (MG); the Centre for Detection and Preven-
extent. Am J Clin Nutr 2005;82:964 –71. tion of Arteriosclerosis, Timone University Hospital, Marseille, France
(M-CB and PV); the Health Regional Observatory, Epidemiology and Social
KEY WORDS Intervention trial, primary prevention, blood Sciences Applied to Medical Innovation Unit-379-INSERM, Marseille,
pressure, lipid metabolism, apolipoproteins France (PV); and the UMMM Metabolic Disease and Macronutrients Lab-
oratory, Clermont-Ferrand, France (PG).
2
Supported by the French Research Ministry (AQS grant, SV’s salary),
INTRODUCTION INSERM (IDS grant), the Provence-Alpes-Côte d’Azur Regional Council,
Cardiovascular disease (CVD) is one of the leading causes of the Bouches du Rhône General Council, the CRITT-PACA, and the follow-
death in the Western world and is the third cause of mortality in ing companies: Rivoire & Carret Lustucru, Jean Martin, Le Cabanon, Bou-
langerie Coagulation Surgelés, Distplack Mariani, and Minoterie Giraud.
developing countries, accounting for 앒25% of all deaths (1). 3
Reprints not available. Address correspondence to D Lairon, UMR 476-
Many studies have already shown the effect of specific nutrients
INSERM/1260-INRA, Human Nutrition and Lipids, Faculty of Medicine
or diets on cardiovascular disease risk factors or mortality. The Timone, 27 Boulevard Jean Moulin, 13385 Marseille Cedex 5, France. E-
Seven Countries Study by Keys et al in the 1960s showed a mail: denis.lairon@medecine.univ-mrs.fr.
relation between saturated fat intake, fasting blood cholesterol Received January 5, 2005.
concentrations, and coronary heart disease (CHD) mortality in Accepted for publication July 2, 2005.
964 Am J Clin Nutr 2005;82:964 –71. Printed in USA. © 2005 American Society for Nutrition
FIGURE 1. Mean intakes (g/d) of important food items at baseline and after a 3-mo dietary intervention in subjects fed a Mediterranean-type diet (Med group;
n ҃ 88) or a low-fat diet (low-fat group; n ҃ 81) at baseline and at 3 mo.
in blood cholesterol, with a 5% level of significance (two-tailed clinical baseline data, with the exception of smoking [more
test). smokers in the Med group (27.3%) than in the low-fat group
The intention-to-treat principle was used to compare the Med (17.3%); P ҃ 0.07] and LDL cholesterol (4.4 mmol/L in the Med
and low-fat groups. The groups were defined according to the group and 4.1 mmol/L in the low-fat group; P 쏝 0.05) as previ-
initial random assignment rather than according to reported com- ously reported (22).
pliance with the protocol (n ҃ 169 at 3 mo). Continuous data are
expressed as mean (앐SD) values. Variables not normally dis- Nutritional data
tributed were tested with the Mann-Whitney U test. Associations As illustrated in Figure 1, the subjects noticeably altered their
were examined by using Pearson’s (parametric data) and Spear- food consumption pattern in line with recommendations to the 2
man’s (non parametric data) correlation analysis. groups. For almost all dietary variables, there were no significant
The data were analyzed by using repeated-measures 2-factor time-by-group interactions; dietary changes in the Med group
analysis of covariance. Time effect and diet interaction were were not significantly different from those in the low-fat group
determined as reported in the tables (P values for time and time- (Table 1). Both groups decreased energy intake over the 3-mo
by-group interactions, respectively). The results were adjusted study period. Alcohol intake decreased significantly in both
for baseline values if they were different and also for age, sex, groups, but the decrease was greater in the low-fat group than in
tobacco, and BMI variation. Comparisons within groups were the Med group (P for interaction ҃ 0.02). In both groups, the
made by using the paired-samples t test. Comparisons between percentage of energy provided by proteins and carbohydrates
groups, at baseline, were made by using the independent-samples increased significantly, whereas that provided by total fat de-
t test. Two-tailed P values 쏝 0.05 were judged significant. SPSS creased. The percentage of energy provided by saturated fat
for WINDOWS version 12.0 (SPSS Inc, Chicago, IL) was used decreased over time, whereas that provided by polyunsaturated
for the statistical analysis. fat increased. There was a significant interaction for percentage
of fat provided by monounsaturated fat, which increased with the
Med diet and decreased with the low-fat diet (P for interaction ҃
RESULTS 0.012). Ingested cholesterol decreased in both groups over the
A total of 232 subjects were invited to participate in the study 3-mo period. Total, soluble, and nonsoluble dietary fiber intakes
(22); 212 (91.4%) subjects were included in the study (n ҃ 102 did not increase significantly in either group. In both groups,
in the Med group and 110 in the low-fat group). During the first there was no variation in -carotene, folate, vitamin C, and vi-
3 mo, 43 subjects dropped out [n ҃ 14 (15.9%) in the Med group tamin B-12 intakes. Calcium intake decreased in both groups.
and 29 (35.8%) in the low-fat group]. The characteristics of the There was a significant interaction for vitamin B-6 and vitamin
dropouts were not significantly different from those of the other E intakes (P for interaction ҃ 0.011 and 0.027, respectively);
subjects. Of the subjects (n ҃ 169) followed at month 3, 42% vitamin B-6 increased with the Med diet and decreased with the
were men and 58% women in the Med group, and 39.5% were low-fat diet, whereas vitamin E showed the reverse trend. Iron
men and 60.5% were women in the low-fat group. The mean age intake decreased in both groups, and there was a significant
of the subjects was 50.8 (앐 10.8) and 51.6 (앐 10.3) in the 2 interaction for iron (P for interaction ҃ 0.032).
groups, respectively. Overall, 35% of the subjects were over-
weight (BMI: 25-30), 38% were obese (BMI 쏜 30), 93% were Nutritional biomarkers
hypercholesterolemic, 46% were hypertriacylglycerolemic, and Changes in plasma nutritional biomarkers after the 3-mo diet
21% were hypertensive. There was no significant difference are shown in Table 2. After adjustment for BMI variation, we
between the Med (n ҃ 88) and low-fat (n ҃ 81) groups with observed a tendency for increased plasma -carotene and lyco-
regard to energy intake and demographic, anthropometric, and pene concentrations over time (P ҃ 0.10). Total plasma phenolic
Baseline 3 mo Mean change Baseline 3 mo Mean change Time Group Time ҂ group
TABLE 2
Plasma nutritional markers of the sample by treatment group at baseline and after the 3-mo diet1
Baseline3 3 mo Mean change Baseline3 3 mo Mean change Time Group Time ҂ group
for a time-by-group interaction in LDL cholesterol and a trend for covariance. After adjustment for BMI variation, the time effect
a main effect of group (P ҃ 0.074): Med group (Ҁ0.5 mmol/L) remained significant for total cholesterol (P ҃ 0.032), triacylg-
and low-fat group (Ҁ0.2 mmol/L). After 3 mo, no change in HDL lycerols (P ҃ 0.039), and insulinemia (P ҃ 0.014).
cholesterol was observed, whereas TRL cholesterol decreased
significantly in both groups.
In both groups Apo-AI and Apo-B were significantly lower DISCUSSION
over the 3-month period with no time-by-group interactions. The Medi-RIVAGE study aimed to determine the effect of a
Apo-E and Apo-CIII did not change after 3 mo. Plasma triacyl- Mediterranean diet and of a commonly prescribed low-fat diet on
glycerol and TRL-triacylglycerol concentrations significantly risk factors for cardiovascular disease. Aimed at a reduction in
decreased after 3 mo. cardiovascular disease risks in our population, the study strategy
Glycemia, insulinemia, and HOMA score decreased signifi- was to globally modify dietary habits without particular focus on
cantly after 3 mo, with no time-by-group interactions (Table 4). any specific nutrient, as is usual in intervention studies. This
Moreover, significant effects of BMI were found on total cho- strategy was evaluated by using dietary questionnaires and mea-
lesterol, triacylglycerols, glycemia, insulinemia, and HOMA suring plasma nutritional markers. In both diet groups, we ob-
score: P ҃ 0.028, ҃ 0.014, 쏝 0.001, 쏝 0.001, and 쏝 0.001 by served noticeable changes in food intake (Figure 1). The most
using a 2-factor (time and BMI) repeated-measures analysis of important changes in the Med group were increases in fruit,
TABLE 3
Mean BMI and blood pressure of the sample by treatment group at baseline and after the 3-mo diet1
Baseline3 3 mo Mean change Baseline3 3 mo Mean change Time Group Time ҂ group
BMI (kg/m ) 2
28.7 앐 4.6 4
27.2 앐 4.2 Ҁ1.5 28.7 앐 5.4 27.5 앐 5.1 Ҁ1.2 0.010 NS NS
Systolic blood pressure (mm Hg) 128 앐 17 127 앐 12 Ҁ1 126 앐 17 124 앐 11 Ҁ2 NS NS NS
Diastolic blood pressure (mm Hg) 78 앐 11 80 앐 8 2 79 앐 21 79 앐 8 0 0.049 NS NS
1
Med, Mediterranean-type diet.
2
Adjusted for age, sex, tobacco use, and BMI variation; P values were obtained by using a 2-factor repeated-measures ANOVA.
3
No significant differences at baseline between groups, P 쏝 0.05 (independent-samples t test).
4
x 앐 SD (all such values).
Baseline 3 mo Mean change Baseline 3 mo Mean change Time Group Time ҂ group
vegetables (including legumes), olive oil, nuts, and fish and re- that we observed a 2-fold lower dropout rate in the Med group
ductions in other vegetable oils, dairy products, and meat. In the after 3 mo, which highlighted a better compliance with the Med-
low-fat group, the main changes observed were increases in fruit, iterranean diet pattern, as recently checked by others (23).
vegetables, and fish and decreases in meat and dairy products. Nevertheless, changes in risk factors in the low-fat group were
These changes in food intake had the effects of reducing protein comparable and even more satisfactory than those reported by
intake, of increasing the relative proportion of carbohydrates, of others during short-term intervention studies with the low-fat
decreasing ingested cholesterol, and of reducing total fat with a American Heart Association–recommended diet (24). In the
marked improvement in fat quality (Table 1). For some fatty Medi-RIVAGE study, several variables were significantly dif-
acids we found a significant variation after a 3-mo diet in the Med ferent after a 3-mo dietary intervention, such as total and TRL
and low-fat groups, and the increase in fish consumption ob- cholesterol, triacylglycerols, TRL triacylglycerols, apo A-I, apo
served was corroborated by higher plasma docosahexaenoic ac- B, glucose, insulin, and HOMA score. Interestingly, the reduc-
ids. Fiber intakes tended to increase after a 3-mo diet, especially tion in LDL cholesterol after 3 mo tended to be greater with the
in the Med group (2.1 g/d), as consistent with recommendations Med diet (P ҃ 0.09 for time-by-group interaction) than with the
to increase intakes of whole-grain food products and legumes. low-fat diet, which agreed with the observations of the MAR-
We observed a trend toward increases in total carotenoids in- GARIN (20) and the Indo-Mediterranean (18) studies. Con-
gested, which was confirmed by the measurement of the plasma versely, the biochemical variables did not change noticeably
nutritional markers, which reflected the increase in vegetable and during the Lyon Diet Heart study (16). Dietary factors may ex-
fruit intakes. To summarize, whereas the 2 diets compared were plain these results. The Med diet provided greater intakes of
based on somewhat different consumption patterns of some soluble fibers (from fruit, vegetables, legumes, and cereals) and
foodstuffs (especially olive oil, whole grains, legumes, and nuts), monounsaturated fatty acids (oleic acid from olive oil), which are
they did not generate markedly different supplies of macro- or known independently to have a cholesterol- and LDL cholester-
micronutrients. ol–lowering effect (25–28). Olive oil intake was recently shown
Although we observed noticeable changes in dietary habits in to be related to reduced rates of myocardial infarction (29). Fiber-
both groups, the objectives set for the subjects were not fully met. rich diets (28) and a high consumption of fruit and vegetables
Indeed, dietary records and nutritional markers showed insuffi- have been related to a reduced occurrence of CHD (5, 6).
cient intakes of several nutrients. Lipid intakes, especially mono- The beneficial changes in total and LDL-cholesterol that we
unsaturated fatty acids, remained insufficient in the Med group, observed in both groups agree with those reviewed by Sacks and
whereas lipid intakes remained too high in the low-fat group; Katan (30), for both a low-fat and a Mediterranean-type diet.
fiber intake did not reach the intake recommended for the Med However, we found no increase in fasting triacylglycerols, as
group (25 g/d). This can be explained by the fact that subjects in observed with high-carbohydrate, low-fat diets (30), which was
both groups did not fully follow the dietary recommendations. It likely due to the limited nature of the modification observed in
is difficult to recruit subjects who will be strictly compliant with dietary intakes and with the limited reduction in dietary energy
dietary recommendations. Moreover, because the subjects lived and related weight loss, all of which are limiting factors for
in the Mediterranean area, those in the low-fat group may have carbohydrate-induced hypertriglyceridemia. The reduction in
unintentionally altered the recommended low-fat diet toward a plasma triacylglycerol (mainly in the form of TRLs) observed
kind of “moderate” Mediterranean diet pattern, which resulted in only in the Med group may have been due mainly to a high intake
a reduction in differences between the 2 groups. It is noteworthy of fish and long-chain nҀ3 fatty acids (31, 32) and more probably
to changes in postprandial lipoprotein metabolism induced by habits on the prevalence of the metabolic syndrome among Greek adults
meals rich in olive oil (33) or dietary fibers (34). from the ATTICA study. Am Heart J 2004;147:106 –12.
8. Hu FB, Willett WC. Diet and coronary heart disease: findings from the
An important point to consider in data interpretation is body Nurses’ Health Study and Health Professionals’ Follow-up Study. J Nutr
weight and BMI reduction, which, although moderate, may im- Health Aging 2001;5:132– 8.
prove some cardiovascular disease risk factors (35). After ad- 9. Willett WC, Sacks F, Trichopoulou A, et al. Mediterranean diet pyramid:
justment for BMI variation for variables showing an interaction a cultural model for healthy eating. Am J Clin Nutr 1995;61(suppl):
with BMI, decreases in total cholesterol, triacylglycerols, and 1402S– 6S.
10. Scali J, Richard A, Gerber M. Diet profiles in a population sample from
insulin after 3 mo remained significant. This suggests that the Mediterranean southern France. Public Health Nutr 2001;4:173– 82.
significant changes still observed for these variables, after ad- 11. Ferro-Luzzi A, James WP, Kafatos A. The high-fat Greek diet: a recipe
justment for weight loss, result from an independent effect of diet for all? Eur J Clin Nutr 2002;56:796 – 809.
quality alteration. This agrees with the recent observation that the 12. Haban P, Klvanova J, Zidekova E, Nagyova A. Dietary supplementation
with olive oil leads to improved lipoprotein spectrum and lower nҀ6
Mediterranean diet is associated with decreased insulin resis-
PUFAs in elderly subjects. Med Sci Monit 2004;10:PI49 –54.
tance in patients with metabolic syndrome (21). 13. Perez DD, Strobel P, Foncea R, et al. Wine, diet, antioxidant defenses,
We observed no improvement in systolic or diastolic blood and oxidative damage. Ann NY Acad Sci 2002;957:136 – 45.
pressure with the dietary interventions. It is possible that the large 14. Gerber MJ, Scali JD, Michaud A, et al. Profiles of a healthful diet and its
number of normotensive subjects present (79% of total) or the relationship to biomarkers in a population sample from Mediterranean
southern France. J Am Diet Assoc 2000;100:1164 –71.
limited duration of the interventions prevented such an observa- 15. Trichopoulou A. Mediterranean diet: the past and the present. Nutr
tion, because several epidemiologic studies have reported re- Metab Cardiovasc Dis 2001;11:1– 4.
duced blood pressure linked to Mediterranean dietary patterns 16. de Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J, Mamelle N.
(36, 37). Mediterranean diet, traditional risk factors, and the rate of cardiovascular
In conclusion, the diets tested during the Medi-RIVAGE complications after myocardial infarction: final report of the Lyon Diet
Heart Study. Circulation 1999;99:779 – 85.
study, a Mediterranean-type diet or commonly prescribed low- 17. de Lorgeril M, Renaud S, Mamelle N, et al. Mediterranean alpha-
fat diet, show measurable efficiency in reducing key cardiovas- linolenic acid-rich diet in secondary prevention of coronary heart dis-
cular risk factors. In line with most recent theories on optimal ease. Lancet 1994;343:1454 –9.
eating patterns for prevention of CHD (38) and according to 18. Singh RB, Dubnov G, Niaz MA, et al. Effect of an Indo-Mediterranean
Law’s concept (39), the observed lowering in plasma cholesterol diet on progression of coronary artery disease in high risk patients (Indo-
Mediterranean Diet Heart Study): a randomised single-blind trial. Lan-
actually indicates a reduction in cardiovascular risk of 9% with cet 2002;360:1455– 61.
the low-fat diet and 15% with the Mediterranean-type diet. The 19. Bemelmans WJ, Broer J, Feskens EJ, et al. Effect of an increased intake
Medi-RIVAGE intervention study thus provides support to early of alpha-linolenic acid and group nutritional education on cardiovascular
(2– 4) or recent (5–7, 21) observational studies linking adherence risk factors: the Mediterranean Alpha-linolenic Enriched Groningen
Dietary Intervention (MARGARIN) study. Am J Clin Nutr 2002;75:
to the traditional Mediterranean diet and reduced cardiovascular 221–7.
disease mortality and reinforces the notion of its efficacy in 20. Bemelmans WJ, Broer J, de Vries JH, Hulshof KF, May JF, Meyboom-De
primary as well as secondary prevention. Jong B. Impact of Mediterranean diet education versus posted leaflet on
dietary habits and serum cholesterol in a high risk population for cardiovas-
We are indebted to N Rousset and S Loviconi (CDPA dietitians); MF cular disease. Public Health Nutr 2000;3:273– 83.
Alliaud and AM Lorec (laboratory supervisors); M Senft, J Leonardi, B Lyan, 21. Esposito K, Marfella R, Ciotola M, et al. Effect of a Mediterranean-style
C Bideau, D Iniesta, N Peyrol, and JF Lesgard (laboratory technicians); and diet on endothelial dysfunction and markers of vascular inflammation in
A Lanteaume and A Loundou (statisticians). the metabolic syndrome: a randomized trial. JAMA 2004;292:1440 – 6.
DL, MG, and PV coordinated the study. SV-B, CD, and DL wrote the 22. Vincent S, Gerber M, Bernard MC, et al. The Medi-RIVAGE study
manuscript. MG was responsible for the nutritional strategy. M-CB recruited (Mediterranean Diet, Cardiovascular Risks and Gene Polymorphisms):
the subjects and conducted the clinical evaluations. CD, SV-B, PG, and rationale, recruitment, design, dietary intervention and baseline charac-
teristics of participants. Public Health Nutr 2004;7:531– 42.
M-JA-C were involved in the laboratory work and data interpretation. PV,
23. Sanchez-Villegas A, Martinez JA, De Irala J, Martinez-Gonzalez MA.
OH, and SV-B were involved in the statistical analyses. HP and RP were Determinants of the adherence to an “a priori” defined Mediterranean
involved in the blood analyses. All authors contributed to the final report. dietary pattern. Eur J Nutr 2002;41:249 –57.
None of the authors declared a conflict of interest. 24. Walden CE, Retzlaff BM, Buck BL, McCann BS, Knopp RH. Lipopro-
tein lipid response to the National Cholesterol Education Program step II
diet by hypercholesterolemic and combined hyperlipidemic women and
men. Arterioscler Thromb Vasc Biol 1997;17:375– 82.
REFERENCES 25. Mensink RP, Katan MB. Effect of dietary fatty acids on serum lipids and
1. Lopez AD. Assessing the burden of mortality from cardiovascular dis- lipoproteins. A meta-analysis of 27 trials. Arterioscler Thromb 1992;12:
eases. World Health Stat Q 1993;46:91– 6. 911–9.
2. Keys A, Menotti A, Karvonen MJ, et al. The diet and 15-year death rate 26. Perez-Jimenez F, Lopez-Miranda J, Mata P. Protective effect of dietary
in the seven countries study. Am J Epidemiol 1986;124:903–15. monounsaturated fat on arteriosclerosis: beyond cholesterol. Athero-
3. Keys A, Blackburn HW, Menotti A, et al. Coronary heart disease in sclerosis 2002;163:385–98.
seven countries. Summary. Circulation 1970;41:I186 –95. 27. Mekki N, Dubois C, Charbonnier M, et al. Effects of lowering fat and
4. Thom TJ, Epstein FH, Feldman JJ, Leaverton PE. Trends in total mor- increasing dietary fiber on fasting and postprandial plasma lipids in
tality and mortality from heart disease in 26 countries from 1950 to 1978. hypercholesterolemic subjects consuming a mixed Mediterranean-
Int J Epidemiol 1985;14:510 –20. Western diet. Am J Clin Nutr 1997;66:1443–51.
5. Martinez-Gonzalez MA, Fernandez-Jarne E, Serrano-Martinez M, 28. Rimm EB, Ascherio A, Giovannucci E, Spiegelman D, Stampfer MJ,
Marti A, Martinez JA, Martin-Moreno JM. Mediterranean diet and re- Willett WC. Vegetable, fruit, and cereal fiber intake and risk of coronary
duction in the risk of a first acute myocardial infarction: an operational heart disease among men. JAMA 1996;275:447–51.
healthy dietary score. Eur J Nutr 2002;41:153– 60. 29. Fernandez-Jarne E, Martinez-Losa E, Prado-Santamaria M, Brugarolas-
6. Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to a Brufau C, Serrano-Martinez M, Martinez-Gonzalez MA. Risk of first non-
Mediterranean diet and survival in a Greek population. N Engl J Med fatal myocardial infarction negatively associated with olive oil consump-
2003;348:2599 – 608. tion: a case-control study in Spain. Int J Epidemiol 2002;31:474 – 80.
7. Panagiotakos DB, Pitsavos C, Chrysohoou C, et al. Impact of lifestyle 30. Sacks FM, Katan M. Randomized clinical trials on the effects of dietary