Introduction
Cardiovascular diseases are the first death cause in European countries1. It is well known that hypercholesterolaemia, hypertension, tobacco consumption and other factors increase the risk of developing cardiovascular disease2. In Spain, a ‘Spanish paradox’ has been describedReference Serra-Majem, Ribas, Tresserras, Ngo and Salleras3, consisting of a high prevalence of cardiovascular risk factors but low incidence and mortality for ischaemic heart diseaseReference Medrano, Cerrato, Boix and Delgado-Rodríguez4. The existence of protective factors in Spanish subjects was considered to be a possible explanation of this paradox. One of the proposed protective factors was a dietary pattern close to the Mediterranean diet (MD). In the Canary Islands, as opposed as to what is found in the rest of Spain, a high prevalence of cardiovascular risk factors coexists with one of the highest cardiovascular mortality rates of the countryReference Serra-Majem, Navarro, Laínez and Ribas5. The present paper is intended to clarify some issues concerning dietary habits of the Canarian population that could influence on their cardiovascular morbidity and mortality.
The metabolic syndrome (MS) is a cluster of cardiovascular risk factors2 that could affect one of every four SpaniardsReference Álvarez, Ribas and Serra-Majem6. The ATP III establishes that a subject has MS when he/she fulfils more than or equal to three of the five following criteria: high blood pressure (or medical treatment for hypertension), impaired fasting glycaemia (or medical treatment for diabetes), hypertriglyceridaemia, low HDL cholesterol levels and abdominal obesity.
Insulin resistance has been linked to MS developmentReference Ceriello and Motz7. It must be pointed out that there is no scientific consensus as to how to diagnose an insulin-resistance state. Nevertheless, for the purposes of this work the HOMA (Homeostasis Model Assessment) model has been chosen. The HOMA is an index that relates fasting insulin and glucose levelsReference Matthews, Hosker, Rudenski, Naylor, Treacher and Turner8. There is no standardised threshold to determine when a subject has insulin resistance. In Spain, Ascaso et al. Reference Ascaso, Romero, Real, Priego, Valdecabres and Carmena9 have established a threshold based on the percentile 90 of fasting insulinaemia and the same percentile of HOMA index, of a control population (subjects without clinical or analytical parameters of insulin resistance).
Early diagnosis of MS and a preventive approach could diminish cardiovascular disease. A healthier lifestyle, with moderate physical activity and adequate nutritional habits, were the initial steps to prevent MS and cardiovascular complications2. In spite of its high prevalence and importance, there is no agreement about the dietary factors that may attenuate or promote it.
There is some evidence of the beneficial role of the MD on MS. In a randomised controlled study carried out in Naples (Italy)Reference Esposito, Marfella, Ciotola, Di Palo, Giugliano, Giugliano, D'Armiento, D'Andrea and Giugliano10 with 180 MS patients, it was observed that in the intervention group (following a Mediterranean-style diet), the prevalence of MS was reduced by 55.6%, as compared with a reduction of 13.3% in the control group (following a prudent diet) in a 2-year period. In another randomised controlled study carried out in GermanyReference Michalsen, Lehmann, Pithan, Knoblauch, Moebus, Kannenberg, Binder, Budde and Dobos11 with 101 patients with established coronary artery disease, an MD approach could not change metabolic risk factors in a 1-year period. Nevertheless, it must be pointed out that the nutritional intervention developed in the German study was far from being a real MD pattern in some aspects (high intake of red meat and meat products, low use of olive oil, etc.).
The definition of ‘MD’ is not easyReference Bach, Serra-Majem, Carrasco, Roman, Ngo, Bertomeu and Obrador12. It could be considered as a food pattern that characterises a way of life and a culture, and that could improve health and quality of life of those subjects following itReference Serra-Majem, Ngo, Ribas and Tur13. Some food components of the MD could be: high intake of vegetables, legumes, fruit, nuts and cereal (mostly whole grain); high to moderate intake of fish; low intake of saturated fatty acids but high intake of unsaturated fatty acids (specially olive oil); moderate to low intake of dairy products (specially cheese and yoghurt); low intake of red meat and moderate intake of alcohol (mostly red wine)Reference Trichopoulou, Orfanos, Norat, Bueno-de-Mesquita, Ocke and Peeters14.
In epidemiological studies assessing the relation of the MD and the health parameters, indexes of adherence to a previously established MD food pattern were usedReference Bach, Serra-Majem, Carrasco, Roman, Ngo, Bertomeu and Obrador12. The indexes were based on a single score that results from the combination of measured intakes of the different individual components of the MD food pattern.
The present work analysed the relationship between the individual index of adherence to a pre-established MD score, and the prevalence of MS, its criteria and insulin resistance, in a representative sample of a Spanish population.
Material and methods
Subjects
In the Canarian Nutrition Survey (ENCA 1997–1998)Reference Serra-Majem, Armas and Ribas15, a representative sample of the Canarian general population between 6- and 75-year-old was selected by a two-stage stratified sampling method. A total of 1747 individuals participated (67.2% of the original sample). ENCA included two individual questionnaires about diet, lifestyle and health status. Anthropometric variables and blood pressure were also measured. Participants in these home surveys were solicited to have a blood extraction in order to determine biochemical parameters. Seven hundred and eighty-two subjects participated in the biochemical phase (44.8% of the ENCA participants). The present study is based on a sub-sample of 578 adults ≥ 18 years (249 men and 329 women).
The study was approved by the Institutional Committee of Ethics of the Canarian Health Service, and all the study patients gave written informed consent. Details on the methodology have been published elsewhereReference Serra-Majem, Navarro, Ribas and Laínez16.
Nutritional assessment
The usual food intake in the past 12 months was registered by a semiquantitative Food Frequency Questionnaire (FFQ), including 81 food items. Participants were asked to report how many times and how frequently they consumed these items, selecting one out of five frequency categories (ranging from ‘never’ to ‘daily’). Referred intakes were transformed into grams per day. The FFQ is available onlineReference Serra-Majem, Armas and Ribas15. Total energy intake and macronutrient intakes were calculated from the FFQ using the Spanish food composition database developed by Mataix et al. Reference Mataix, Mañas and Martínez de Vitoria17, complemented with French tablesReference Favier, Ireland-Ripert, Toque and Feinberg18. The specific methodology has been published elsewhereReference Henríquez, Doreste, Díaz-Cremades, López-Blanco, Álvarez-León and Serra-Majem19.
In order to assess the adherence to an MD pattern, a specific score was calculated, based on ten food items. The ten food items were considered as follows:
(1) Cereal: white bread, whole grain bread, pasta, rice, breakfast sweetened cereals, gofio (roasted wheat or corn flour) and cookies.
(2) Fruit: apples, avocadoes, oranges, bananas, papayas, natural fruit juice and jam.
(3) Vegetables: vegetable soup, lettuce, boiled vegetables, tomatoes, onions and other vegetables.
(4) Legumes.
(5) Fish: blue fish, white fish, seafood and octopus.
(6) Nuts.
(7) Alcohol from red wine (one glass equivalent to 100 ml of ethanol at 12%).
(8) Monounsaturated to saturated fatty acids ratio (MUFA/SFA ratio).
(9) Whole-fat dairy products: whole-fat milk, caramel custard, yoghurt, cream, cheese, margarine and butter.
(10) Red meat and derivatives: beef, pork, ham, sausages and similar food items, liver and tripe.
Daily intakes of these foods items (in g day− 1) were adjusted to 2500 kcal in men and 2000 kcal in women. Then, sex-specific tertiles of the intakes of these ten food items were calculated. Three points were given to the subjects in the third tertile of cereal, fruit, vegetables, fish, nuts and MUFA/SFA ratio; two points to the subjects in the second tertile and one point to the subjects in the first tertile. Whole-fat dairy products and red meat intake received an inverse marking system (three points to the first tertile and one point to the third one). With respect to the alcohol intake from red wine, three points were assigned to moderate consumers (0.1–19.9 and 0.1–39.9 day− 1 in women and men, respectively) and one point to non-consumers. There were no subjects consuming higher amounts of alcohol from red wine. Finally, the total sum of individual marks in every food item was calculated given the total Mediterranean score (ranging from 10 to 30).
Metabolic Syndrome
The definition of MS used in our work was the one established by the National Cholesterol Education Program's Adult Treatment Panel III (ATP III)2. A participant was considered to have MS if three or more of the following criteria were present:
1 Abdominal obesity, defined by waist circumference: Men>102 cm; Women>88 cm. Waist circumference was collected with the subject standing, taking as reference half of the distance between the iliac crest and the lowest rib margin. A non-stretch metric measure tape was utilised. There were 2.8% of missing data for this variable.
2 Triglycerides ≥ 150 mg dl− 1 in serum (missing data = 1.2%).
3 HDL cholesterol < 40 and < 50 mg dl− 1 for men and women, respectively. All lipid variables were obtained by standardised chemical methods and spectrophotometry (missing data = 1.3%).
4 Blood pressure ≥ 130/85 mmHg. Blood pressure determination was made in the dominant arm of a seated subject with an automatic tensiometer, separating the first and the second readings by 10 min. The mean of both readings was determined. This criterion was also applied to subjects who were receiving medical treatment with antihypertensive medication (missing data = 2.3%).
5 Fasting glucose ≥ 110 mg dl− 1. Subjects had been fasting for 12 h and the glucose was also obtained by standardised chemical methods. This criterion was also applied to subjects who were receiving antidiabetic medication (insulin or oral agents) (missing data = 1.2%).
Insulin resistance
Insulin resistance was considered with two variables: high fasting insulinaemia or high HOMA index. The HOMA index was calculated with the formula = ![[(insulin\times glycaemia)/405]](https://static.cambridge.org/binary/version/id/?pub-status=live)
. Using the methodology described by Ascaso et al. Reference Ascaso, Romero, Real, Priego, Valdecabres and Carmena9, a control population without insulin resistance was used. The control population was selected from those ENCA subjects that fulfil the inclusion criteria (age 30–60 years; absence of family antecedents of diabetes, hypertension or cardiovascular events (acute myocardial infarction/ictus); absence of personal antecedents of diabetes or cardiovascular disease; fasting glycaemia levels < 110 mg dl− 1; blood pressure < 140/90 mmHg; absence of antihypertensive medication; triglycerides level < 150 mg dl− 1; body mass index < 25 kg m− 2; waist circumference < 102 cm (men) or < 88 cm (women)). Percentile 90 of fasting insulinaemia in that control population was 14.0195 μU ml− 1, whereas percentile 90 of HOMA index was 2.8590 UI.
Sociodemographic and lifestyle variables
The variables assessed were: sex, age, educational level (less than elementary, elementary, high school and university), physical activity in free time (sedentary, light, moderate or vigorous), body mass index (BMI, weight in kilogram per height in square metre categorised in < 25, 25–29.9 and ≥ 30 kg m− 2), tobacco consumption (smoking, no smoking and ex-smokers), having followed any kind of regimen diet in the past 12 months (yes or no) and total energy intake.
Statistical analysis
The categorical variables were expressed as percentages. Differences between MS and non-MS subjects were evaluated with χ2 analyses. Continuous variables were expressed as means and standard deviations (SDs) if they followed a normal distribution, or as medians (25th–75th percentiles) in the case of skewed distributions. Differences between MS and non-MS were evaluated with the Student's t-test or, if not normally distributed, with the Mann–Whitney U test.
A multivariate logistic regression was used to calculate the odds ratios (ORs) for MS prevalence, and for the prevalence of every MS criterion. For these calculations, first tertile of adherence of MD and first tertile of intake of each ten items of the MD index were used as reference category. P < 0.05 was considered to be significant. All the statistics were performed with the SPSS software (release 11.0; SPSS Inc., Chicago, IL, USA).
Results
From the total study population, 141 patients (24.4%) met the ATP III definition of MS. The prevalence of every criterion is summarised in Table 1. The most frequent criteria were elevated blood pressure (50.2%), followed by central obesity and low HDL cholesterol. Men showed a higher prevalence of hypertriglyceridaemia and elevated blood pressure, while women showed higher prevalence of abdominal obesity and low HDL cholesterol (all differences P < 0.05, except for central obesity, P = 0.06).
Prevalence of metabolic syndrome, MS criteria and insulin resistance by sex. ENCA Study, n=578
n, number of subjects; NS, non-significative; HDL, high-density lipoprotein cholesterol; M, men; W, women; SBP, systolic blood pressure; DBP, diastolic blood pressure.
1 Antidiabetic medication, insulin or oral hypoglycemic agents.
The prevalence of every criterion by age group is shown in Fig. 1. In younger groups, the most prevalent criteria were low HDL cholesterol and hypertension. From 55 years on, more than 80% of the subjects fulfil the elevated blood pressure criterion, and more than 50% fulfil the central obesity criterion.
Prevalence of metabolic syndrome criteria by age group. ENCA Study, n = 578
The insulin-resistance prevalence was 32.3% when high insulinaemic level was used and 27.2% when high HOMA index was used. No relevant differences by sex were detected. The prevalence of insulin resistance increases significantly with age.
In an MD score ranging from 10 to 30 points, the mean score of ENCA subjects was 19.7 points (SD 3.15). The score was higher in men and older subjects, as shown in Fig. 2.
Mediterranean diet (MD) score, by sex, age group and educational level. ENCA Study, n = 578
The score was higher in MS subjects [mean score 20.2 (SD 3.2)] than in non-MS subjects [mean score 19.5 (SD 3.1)], P < 0.05. In order to consider all the influential variables, a multivariable regression model was developed. Once adjusted, MD adherence was not related to MS prevalence, with an OR (95% CI) of 1.39 (0.75–2.59) for subjects in the second tertile of adherence and an OR 1.37 (0.76–2.46) for subjects in the third tertile of adherence, with respect to the first tertile.
Individual criteria of MS were also considered. Crude analysis showed that the subjects who presented glycaemia criterion (high fasting glycaemia or antidiabetic treatment) showed a higher MD score [mean score 21.0 (SD 2.9)] than the subjects without this criterion [mean score 19.5 (SD 3.1)] P < 0.05. Once adjusted (Table 2), the subjects in the third tertile of adherence to the MD pattern (score of adherence ≥ 21) presented a 70% lower prevalence of the blood pressure criterion (P for tendency 0.051) but 2.5 times more prevalence of the glycaemia criterion with respect to the first tertile (score of adherence ≤ 18). The rest of MS criteria and the insulin-resistance prevalence were not significantly associated with the MD score.
Odds ratio of presence of metabolic syndrome (MS) criteria and insulin resistance, by tertile of adherence to Mediterranean diet (MD). ENCA Study, n=532
NS, non-significative; P tend, P-value for tendency test.
1 Adjusted by sex, age, educational level, physical activity level, BMI, tobacco consumption, diet in the past 12 months and energy intake.
The association of the ten food items of the MD pattern with the five MS criteria and insulin-resistance prevalence were considered separately (Table 3).
Adjusted odds ratio1 of presence of metabolic syndrome (MS) criteria and insulin resistance, by tertile of intake of the food groups of the Mediterranean diet (MD) pattern. ENCA Study, n=532
T1, first tertile; T2, second tertile; T3, third tertile; NA, not applicable (only two groups); MUFA/SFA, monounsaturated to saturated fatty acids ratio.
1 Adjusted by sex (except red wine alcohol intake), age, educational level, physical activity level, BMI, tobacco consumption, diet in the past 12 months, energy intake and intake of rest of food groups included in the MD score. *P < 0.05; **P < 0.01.
Red meat intake was associated with a higher prevalence of the blood pressure criterion (elevated blood pressure or antihypertensive treatment), P for tendency 0.028. Moderate alcohol intake from red wine was associated with a lower prevalence of this criterion in women.
With respect to the two dislipemic criteria (high triglyceridaemia and low HDL cholesterol), moderate alcohol intake was associated with a lower prevalence of both criteria, but it was significant only for HDL cholesterol in men. Fruit intake also showed a protective effect on both dislipemic criteria. On the contrary, vegetable intake was associated with a higher prevalence of dislipemia. In both cases, it was significant only in the case of triglyceride criterion.
With respect to the glycaemic homeostasis (glycaemia criterion and insulin-resistance prevalence), cereal intake showed a protective effect, significative in the case of insulin resistance measured by high insulinaemia levels (P for tendency 0.020). Fruit intake showed a relevant protective effect over high HOMA index. Whole-fat dairy products showed a protective effect on the glycaemia criterion (high fasting glycaemia or antidiabetic treatment), but it showed the reverse effect on insulin resistance, although not significant in later. High monounsaturated fatty acid intake (with respect to saturated intake) showed a protective effect on the glycaemia criterion and insulin resistance when measured by the two variables (insulinaemia and HOMA index), P for tendency = 0.06 for both. Finally, moderate intake of alcohol from red wine showed a protective effect on the glycaemic homeostasis, especially in men (not significant).
No noticeable association was observed for the central obesity criterion.
Discussion
One in every four adults presented MS in this representative sample of Spanish population. This prevalenceReference Álvarez, Ribas and Serra-Majem6 is one of the highest in the European population-based studies and similar to prevalence described in the USAReference Ford, Giles and Dietz20. A higher adherence to the MD was related with a lower prevalence of the blood pressure criterion but a higher prevalence of the glycaemia one.
Subjects with high adherence to the MD index (third tertile, score ≥ 21) showed 70% less prevalence of the blood pressure criterion (elevated blood pressure or antihypertensive treatment) than the subjects with low adherence (first tertile, score ≤ 18), P for tendency 0.051. In the Greek-EPIC cohortReference Psaltopoulou, Naska, Orfanos, Trichopoulos, Mountokalakis and Trichopoulou21, an inverse association between MD adherence and blood pressure has also been described.
The food items involved in the protective effect of the MD on the blood pressure criterion in the ENCA Study were lower red meat intake ( < 40 g day− 1) and moderate alcohol intake from red wine (0.1–19.9 g day− 1 in women). Their protective effects remained even when macro- and micronutrients intake (including proteins, saturated fatty acids, sodium, potassium and alcohol) were included in the model (data not shown).
Recently, a 7-year follow-up study of Japanese–Brazilians aged 40–79 yearsReference Damiao, Castro, Cardoso, Gimeno and Ferreira22 showed that men with the highest tertile of red meat consumption presented a 4.7-fold increase in the risk of developing MS. The red meat effect on blood pressure has extensively been described. In a 7-year follow-up study carried out in 1700 US men, it was observed that those men eating >80 g day− 1 of meat showed higher blood pressure levels than men eating < 80 g per dayReference Miura, Greenland, Stamler, Liu, Daviglus and Nakagawa23. In the Oxford-EPIC cohortReference Appleby, Davey and Key24 more than 11 000 subjects, aged 20–79 years, were studied, and those who did not refer meat, fish, eggs or dairy products consumption showed 2–4 mmHg less in their mean blood pressure levels than regular meat consumers. In the Nurses's Health StudyReference Ascherio, Hennekens, Willett, Sacks, Rosner, Manson, Witteman and Stampfer25, 0.11 mmHg increased in systolic blood pressure was observed for every quintile of increase in meat intake. Similar results were described in the Coronary Artery Risk Development in Young Adults (CARDIA) StudyReference Steffen, Kroenke, Yu, Pereira, Slattery, Van Horn, Gross and Jacobs26, while plant food intake (grains, fruit, vegetables, nuts and legumes) was inversely related to elevated blood pressure. In the ENCA Study, vegetables, nuts and legumes intakes were also related with lower prevalence of the blood pressure criterion (no significant).
With respect to the protective effect showed by red wine, the ENCA Study showed that the women who refer moderate alcohol intake (0.1–19.9 g day− 1 of alcohol from red wine) presented a two-fold decrease prevalence of the blood pressure criterion with respect to non-consumers. A protective effect of moderate alcohol intake on cardiovascular profile in women has also been described in another Spanish population-based studyReference Schroder, Ferrández, Jiménez Conde, Sánchez-Font and Marrugat27.
Nevertheless, it must be highlighted that a deleterious effect of excessive alcohol intake on blood pressure has also been extensively described. Thus, in the National Heart, Lung and Blood Institute Family Heart Study carried out with a USA adult population groupReference Djousse, Arnett, Eckfeldt, Province, Singer and Ellison28, a protective effect of 0.1–12 g day− 1 of alcohol intake on the blood pressure criterion was described, especially for women. However, when the alcohol intake was >24 g day− 1, an incremental risk was described, especially in men.
The food items included in the MD score showed relevant associations with the two dislipemic criteria included in the MS definition (high triglyceridaemia and low HDL cholesterol). Moderate alcohol intake was related to a lower prevalence of low HDL cholesterol, of significance only for men. This effect was maintained when macro- and micronutrients were included in the model, even when total alcohol intake was included [OR 0.56 (95% CI 0.37–0.85)].
The protective effect of alcohol intake on HDL cholesterol has been described in other studies. In the National Heart Study, previously mentionedReference Djousse, Arnett, Eckfeldt, Province, Singer and Ellison28 subjects with high alcohol intakes presented a lower prevalence of the HDL criterion than non-alcohol consumers. In the DESIR French studyReference Vernay, Balkau, Moreau, Sigalas, Chesnier and Ducimetiere29, alcohol consumption was also associated with higher HDL cholesterol serum levels in both sexes. Similar results were mentioned in a studyReference Schmid, Salas Martins, Hernán, Velásquez-Meléndez and Ascherio30 carried out in Brazil. Finally, in the other article of the Spanish population-based study previously mentionedReference Sentí, Masiá, Pena, Elosua, Aubó, Bosch, Sala and Marrugat31, alcohol consumption showed a positive effect over HDL cholesterol serum levels in men. This protective effect could be partially explained by a direct effect of alcohol on HDL cholesterol synthesis, and by a modification of the activity of the lipoprotein lipase, hepatic lipase and cholesterol–ester transfer proteinReference Sentí, Masiá, Pena, Elosua, Aubó, Bosch, Sala and Marrugat31.
Fruit intake also showed a protective effect on both dislipemic criteria. Moderate fruit intake (mean intake 375 g day− 1) was associated with a twofold decrease in triglyceride criterion prevalence with respect to low fruit intake ( < 250 g day− 1), whereas high fruit intake (>450 g day− 1) showed no significant association. The protective effect of moderate fruit intake remained even when macro- and micronutrients were included in the model (1.85-fold decrease for the second tertile of fruit intake). In the Baltimore Longitudinal Study of AgeingReference Newby, Muller and Tucker32, subjects with a ‘healthier’ food pattern, where 10% of their energy was provided by fruit, presented 16% lower triglyceride serum levels than subjects whose fruit-energy intake was < 5%. Some randomised controlled studiesReference Gorinstein, Caspi, Libman, Katrich, Lerner and Trakhtenberg33 have shown that fruit supplementation for one month decreases triglycerides levels by 12% in a high risk population (hypercholesterolaemic patients after coronary bypass surgery), and lower levels of total cholesterol and LDL cholesterol were also observed. In the ENCA Study, fruit intake showed a protective effect on low HDL cholesterol levels (Table 3), and also on high LDL cholesterol level (LDL ≥ 130 mg dl− 1), with and OR 0.59 (95% CI 0.35–0.98) for the second tertile and 0.53 (95% CI 0.32–0.90) for the third tertile, with respect to the first fruit intake tertile (all P < 0.05).
With respect to the vegetable intake, an unexpected higher prevalence of the triglyceridaemia criterion was observed in subjects referring medium intakes of vegetables (mean vegetable intake 400 g day− 1) with respect to the lowest intake ( < 200 g day− 1), even when macro- and micronutrients intake were considered. Once again, the highest tertile showed no relevant association. In the Framingham Offspring Cohort StudyReference McKeown, Meigs, Saltzman, Wilson and Jacques34, fibre intake from vegetables was associated with higher (although non-significant) risks of developing MS (OR 1.15). The association observed in the ENCA Study could be due to the cross-sectional design of the study, and subjects with diagnosed hypertriglyceridaemia could be following a healthier diet with increased intakes of vegetables. Nevertheless, the exclusion of subjects that refer being on a diet in the past 12 months did not materially change the results, with an OR of present triglyceridaemia criterion of 1.84 (95% CI 0.99–3.41) for the second tertile and 1.42 (95% CI 0.70–2.87) for the third tertile, with respect to the first tertile. Those results should be assessed in larger prospective studies.
Finally, the MD influence on glucose homeostasis was assessed. As Table 2 shows, the subjects in the highest tertile of MS adherence presented a 2.5-fold increase in glycaemic criteria (high fasting glycaemia or antidiabetic treatment), although no worth-mentioning effect was seen on the insulin-resistance criterion. Subjects with higher intakes of whole-fat dairy products (third tertile, mean intake 600 g day− 1) showed a 3.7-fold decreased prevalence of the glycaemic criterion with respect to subjects with lower intakes (first tertile, intakes < 140 g day− 1). The results remained even when macro- and micronutrients intakes were considered. An opposite effect was seen on the insulin-resistance criterion (not relevant).
With respect to the glycaemia criterion, these results must be taken with caution, because the protective effect of whole-fat dairy products could also be due to the cross-sectional design of the study. Most subjects that fulfil glycaemic criteria were previously known to be diabetic patients. Most of them were following a diet, and they have probably received nutritional counselling regarding a healthier diet (including low whole-fat dairy products intake). Nevertheless, high intake of whole-fat dairy products has been associated with lower weight gainReference Sánchez-Villegas, Bes-Rastrollo, Martínez-González and Serra-Majem35, and this factor would decrease the risk of developing diabetes. In the Women's Health Study carried out in the USAReference Liu, Song, Ford, Manson, Buring and Ridker36, dairy products intake was associated with lower prevalence of MS, and the protective effect was stronger for whole-fat dairy products (P for tendency 0.002) than for low-fat dairy products (P for tendency 0.07). Those results should be assessed in larger prospective studies.
Higher cereal intake was related with a lower prevalence of the glycaemic criterion and the insulin-resistance criterion. The subjects referring cereal intake >150 g day− 1 presented half the risk of hyperinsulinaemia than subjects with lower intakes. Moderate fruit intake (mean intake 375 g day− 1) was associated with a twofold decrease in high HOMA index prevalence. It was also associated with lower hyperinsulinaemia prevalence, although not significantly. The results remained even when macro- and micronutrients intakes were considered (OR 0.53 for the second tertile and 0.56 for the third tertile).
There are numerous studies analysing the relationship between carbohydrate intake and glucose homeostasis. With a similar methodology as the one used for the ENCA (cross-sectional population-based study), the Framingham Offspring Cohort StudyReference McKeown, Meigs, Saltzman, Wilson and Jacques34 showed that whole-grain cereals' and fibre intakes (especially, fibre from cereal and fibre from fruit) were associated with a lower mean HOMA index. In the Insulin Resistance Atherosclerosis Study (IRAS)Reference Liese, Roach and Sparks37, whole-grain intake was associated with higher insulin sensitivity and lower insulin levels. In the Inter99 StudyReference Lau, Færch, Glümer, Tetens, Pedersen, Carstensen, Jørgensen and Borch-Johnsen38, every 100 g day− 1 of increase in fruit and vegetable intake was related with a 0.99 UI decrease in HOMA index, the same decrease that was described for every 1% of energy provided by fructose (fruit sugar with low glycaemic index).
One possible pathophysiology explanation could be that whole-grain cereals' intake and low glycaemic index foods result in a significant decrease in the after-meal insulin response. This effect could be partially explained by the nutrient composition of those foods (fibre content), which results in a lower rate of gastric emptying and an increase satiationReference Delzenne and Cani39.
Other components of cereals and fruit (like magnesium and vitamin E) have also been involvedReference Liese, Roach and Sparks37. The IRAS authors mentioned that the inclusion of fibre and magnesium in their model attenuated the relationship between cereal intake and insulin sensitivity. Nevertheless, in the ENCA Study, the inclusion of these nutrients did not materially affect the described association.
The other food item that was related with glucose homeostasis was the monounsaturated to saturated fatty acids ratio. The subjects that refer intakes of MUFA higher than SFA (MUFA/SFA ratio >1) showed a lower prevalence of elevated HOMA index, and a lower prevalence of hyperinsulinaemia. This protective role persists even when macro- and micronutrients were considered (OR for the second and the third tertiles 0.57 and 0.36 for HOMA and 0.49 and 0.41 for hyperinsulinaemia, respectively, compared with the first tertile). In an interesting review on carbohydrates and fats' effects on insulin resistanceReference Hung, Sievenpiper, Marchie, Kendall and Jenkins40, it was remarked that the substitution of a high SFA diet (MUFA/SFA ratio 0.74) for a high MUFA diet (MUFA/SFA ratio 2.21) was related with a 10% increase in insulin sensitivity. Olive oil is one of the main MUFA sources in the ENCA subjects, being consumed by 80% of the populationReference Serra Majem, Ribas Barba, Armas Navarro, Álvarez León and Sierra41.
In a Spanish study, the prevalence of insulin resistance was 50% lower in subjects that cooked with olive oil, compared with those that used other vegetable oils (sunflower seed or mixed)Reference Soriguer, Esteva, Rojo-Martínez, Ruiz de Adana, Dobarganes, García-Almeida, Tinahones, Beltrán, González-Romero, Olveira and Gómez-Zumaquero42. It has been mentioned that SFA worsens insulin resistance, whereas MUFA and PUFA improve it, due to a modification in the lipid composition of cell membranes in liver and muscle tissues, where the oleic acid concentration is highReference Riccardi, Giacco and Rivellese43. Other possible effects could be the regulation of gene expression related to insulin sensitivityReference Riccardi, Giacco and Rivellese43.
In conclusion, following a diet approaching the MD model, especially with low intake of red meat, moderate alcohol consumption and high intakes of fruit, cereals and monounsaturated fatty acids, has been related with a lower prevalence of MS criteria and insulin resistance in the general population.
Acknowledgements
This work was supported by the Canarian Health Service through a research agreement with the University of Las Palmas de Gran Canaria, and by the financial support of the INNOVA University Patronage Program 2004, from the Fundación Universitaria de Las Palmas-Clínica San Roque.
Authors acknowledge Dr. Iasías Naranjo who checked the English version.
