Associations between dietary pattern and lifestyle, anthropometry and other health indicators in the elderly participants of the EPIC-Italy cohort

Nutrition, Metabolism & Cardiovascular Diseases (2006) 16, 186e201

www.elsevier.com/locate/nmcd

Associations between dietary pattern and lifestyle, anthropometry and other health indicators in the elderly participants of the EPIC-Italy cohort Valeria Pala a,*, Sabina Sieri a, Giovanna Masala b, Domenico Palli b, Salvatore Panico c, Paolo Vineis d, Carlotta Sacerdote d, Amalia Mattiello c, Rocco Galasso e, Simonetta Salvini b, Marco Ceroti b, Franco Berrino a, Elisabetta Fusconi a, Rosario Tumino f, Graziella Frasca f, Elio Riboli g, Antonia Trichopoulou h, Nikolaos Baibas h, Vittorio Krogh a a

Epidemiology Unit, Istituto Nazionale per lo Studio e la Cura dei Tumori, Milan National Cancer Institute, 20133 Milan, Italy b Moleculare and Nutritional Epidemiology Unit, CSPO, Istituto Scientifico della Reg. Toscana, Florence, Italy c Dipartimento di Medicina Clinica e Sperimentale, Universita` Federico II Naples, Naples, Italy d Servizio di Epidemiologia dei Tumori, Dipartimento di Scienze Biomediche e Oncologia Umana, Universita` di Torino, Turin, Italy e Ospedale Oncologico Regionale, Rionero in Vulture, Potenza, Italy f Cancer Registry, Azienda Ospedaliera ‘‘M.P. Arezzo’’, Ragusa, Italy g Unit of Nutrition and Cancer, International Agency for Research on Cancer, Lyon, France h Department of Hygiene and Epidemiology, School of Medicine, University of Athens, Greece Received 8 November 2004; received in revised form 16 May 2005; accepted 18 May 2005

KEYWORDS Dietary patterns; Exploratory factor analysis;

Abstract Introduction: Epidemiological studies have shown that dietary behaviour is an important aetiological factor in various chronic diseases. We used principal component factor analysis to identify dietary patterns and to examine the associations of these patterns with health-related variables in a sample of elderly

* Corresponding author. Tel.: C39 02 2390 3505; fax: C39 02 2390 3510. E-mail address: [email protected] (V. Pala). 0939-4753/$ - see front matter ª 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.numecd.2005.05.009

Dietary patterns in elderly Italians Elderly; Cross-sectional study; Dietary recommendations

187

(R60 years) Italians participating in the European Prospective Investigation into Cancer and Nutrition (EPIC). Methods and results: Exploratory factor analysis was applied to the intake of food groups as estimated by semi-quantitative food questionnaires. Individual participants were assigned factor scores, indicating the extent to which their diet conformed to each of the four dietary patterns identified: prudent (cooked vegetables, pulses, cabbage, seed oil and fish); pasta & meat (pasta, tomato sauce, red meat, processed meat, bread and wine); olive oil & salad (raw vegetables, olive oil, soup and chicken); and sweet & dairy (sugar, cakes, ice cream, coffee and dairy). Highly educated people had high scores on prudent and low scores on pasta & meat. The pasta & meat and prudent patterns were strongly positively associated with body mass index (BMI) and waist-to-hip ratio (WHR) in men and women. Hyperlipidaemic men and women consumed more of the prudent and olive oil & salad patterns and less of the sweet & dairy pattern than those with normal lipids. The olive oil & salad was significantly higher and the pasta & meat and sweet & dairy patterns significantly lower in men and women who had dieted over the previous year, suggesting awareness of the health consequences of these patterns. Conclusions: Dietary pattern analysis provides a characterization of recurrent dietary behaviour in elderly people, and can be used to provide tangible dietary advice to elderly people. ª 2005 Elsevier B.V. All rights reserved.

Introduction Epidemiological studies have shown that the risk of many chronic diseases varies with diet: dietary behaviour is an aetiological factor in various types of cancer [1e6], cardiovascular [7e11] and metabolic [12,13] diseases. However, the variability of diet, and the complexity of ways in which individual food items may be associated with other items and different dietary habits, make it difficult to identify the effects of individual dietary components on health and disease. The difficulty in describing this complexity may be one reason why studies that have sought to relate individual nutrients or food items to the risk of a disease have produced inconsistent results [14e16]. In addition, studies on diet typically use complex questionnaires listing a large number of foods, food components and other dietary variables, analysis of which carries the risk that significant associations may arise by chance alone resulting in false associations between diet and disease. There is also the risk that concentration on the minutia of dietary components may result in missing associations between more general characteristics of diet and disease. Delineating associations between diet and health is even more challenging in elderly people, since dietary behaviour is likely to have been recently modified due to age-related difficulties in mastication, digestion, etc., or following the diagnosis of a disease. For these reasons it is

important to be able to identify key combinations of foods, less prone to modification compared to individual components of diet, and more likely to reflect long-term eating behaviour. Exploratory factor analysis has recently emerged as a method of examining dietary behaviour and relationships between diet and the risk of disease [17,18]. It is a statistical method that analyses the covariate structure of a range of variables to identify a restricted number of underlying variables. The method was first developed and applied in the field of psychology, but has been extended to etiological studies on cardiovascular diseases [19,20], diabetes [21] and cancer [22e24]. When applied to the analysis of food group consumption, exploratory factor analysis reduces data into patterns based upon correlations between dietary items [25], thus making it possible to summarise in a few variable scores the characteristics of diet, and allowing classification of people according to their overall dietary behaviour. Exploratory factor analysis is being used in nutritional epidemiology because of limitations in the traditional single food item approach (which include inability to account for interactions between foods) and because its findings can be used to provide tangible dietary advice. Factor analysis has also been used recently to assess diet and its relationships to health status in the elderly [26e28]. In the present study we used exploratory factor analysis to identify dietary patterns in a sample of elderly subjects. We then

188 examined the relations of these patterns to socioeconomic, demographic and anthropometric variables. The subjects were Italians aged 60 and over participating in the ongoing EPIC-ELDERLY study, part of the European Prospective Investigation into Cancer and Nutrition (EPIC) [29].

Methods Subjects EPIC is being conducted in ten European countries on populations which differ markedly in terms of dietary habits and cancer risk. The Italian EPIC cohort consists of 47,749 people recruited in the study centres of Ragusa, Florence, Turin, Naples (women only) and Varese. Detailed information on this cohort is available elsewhere [30]. The climate, environment, history, and wealth of these centres differ considerably and contribute to the variations in diet, eating behaviour, lifestyle and cancer risk that are almost as great as those characterising the EPIC cohort as a whole [31]. We initially identified 6031 people from the Italian EPIC cohort of age 60 or over who had fully completed the dietary questionnaires. We then excluded those who before recruitment had had cancer (except for non-melanoma skin cancer) or major cardiovascular disease, were lost to follow up. We also excluded those who died during the first year of follow up as dietary patterns were likely to be influenced by the fact that a considerable proportion were in the terminal stages of illnesses. Following these exclusions 5611 people (1536 men and 4075 women) constituted the study population.

Dietary data and food grouping Dietary information for the Italian EPIC cohort was obtained using three validated semi-quantitative food frequency questionnaires [32,33]. One questionnaire was used for the centres of Varese, Turin and Florence; another for Ragusa; and another again for Naples. The questionnaires contained questions about 188, 217 and 140 food items respectively [34]. We grouped the food items in the three questionnaires into food groups based on similarities in ingredients, nutrient profile or culinary usage. Other sub-groupings possibly representing alternative dietary habits were also retained wherever possible: Thus the fish group was separated into tinned and fresh; bread was separated into whole-

V. Pala et al. meal and white; and tomatoes and leafy vegetables were separated into cooked and raw. When items were grouped differently in the three questionnaires, we adopted the most general categorisation. Thus for alcoholic drinks, spirits, liqueurs and fortified wines were grouped together, since these distinctions was not made in all the questionnaires. Thirteen food groups were excluded as they were absent from at least one questionnaire, however, these 13 groups contributed less than 3% of the total intake in the cohort. After this grouping and exclusion procedure, we had 57 food groups for factor analysis. These were: potatoes; leafy vegetables e cooked; leafy vegetables e raw; tomatoes e raw; tomatoes cooked; root vegetables; cabbage; mushrooms; onion and garlic; mixed salad and mixed vegetables; legumes (pulses) including legumes for soup; other vegetables; citrus fruits; fresh fruit (noncitrus); seeds and nuts; milk; yoghurt; cheese; pasta (including polenta, couscous etc.); rice; bread excluding wholemeal bread; wholemeal bread; beef; veal; pork; rabbit (farmed); chicken and turkey; other meat (goat, game, horse, lamb); processed meat; offal; fish except tinned; crustaceans and molluscs; fish-tinned; eggs, seed oils; olive oil; butter; margarine; other animal fats; sugar; honey and jam; chocolate-based confectionery; non-chocolate confectionery; ice-cream; cakes; sweet pies and pastries; puddings (non-milk based); patisserie and biscuits; fruit and vegetable juices; black coffee; tea; coffee with milk; wine; beer; spirits, fortified wines and aperitifs; soups (excluding legume soup); snacks; and pizza.

Anthropometric data Body mass index (BMI) was calculated as weight (kg) divided by height squared (m2); participants were divided into four BMI categories: underweight or normal !25 kg/m2 (category 1); overweight (category 2) between 25.0 and 29.9 kg/m2; obese between 30 and 35 kg/m2 (category 3); and severely obese O35 kg/m2 (category 4). Waist-to-hip ratio (WHR) was calculated as waist circumference (cm) divided by hip circumference (cm). WHR is used as an indicator of central obesity. Three WHR categories were defined based on tertiles of the WHR distribution calculated for men and women separately. Tertile ranges for men were 0.66e0.92 (tertile 1), 0.92e 0.97 (tertile 2), 0.97e1.27(tertile 3); tertile ranges for women were 0.56e0.72 (tertile 1), 0.79e0.85 (tertile 2), 0.85e1.34 (tertile 3).

Dietary patterns in elderly Italians

Education A section of the EPIC-Italy life style questionnaire enquired about education. We dichotomised the replies into those who completed and those who did not complete high school.

Physical activity A section of the life style questionnaire enquired about physical activity at work and leisure. Time spent in various physical activities was transformed into an index of physical activity e physical activity level (PAL) e as described elsewhere [35]. Tertiles of PAL were calculated based on the distribution in men and women combined: 1.44 and less (tertile 1), from 1.44 to 1.57 (tertile 2), 1.57 and more (tertile 3).

Hypertension and hyperlipidaemia A section of the life style questionnaire enquired about hypertension and hyperlipidaemia. Respondents were asked whether or not a doctor had ever diagnosed high blood pressure, or high blood cholesterol/triglycerides, thus the responses were already categorised as yes or no.

Dieting A section of the dietary questionnaire asked whether the respondent had gone on a diet during the previous year. The responses were already categorised as yes or no.

Statistical analyses Factor analysis and factor characterisation Exploratory factor analysis was applied to the 57 food groups, so as to find a few composite factors that explained the maximum fraction of the variance present in the original groups. An orthogonal rotation procedure, the varimax rotation, was then applied to simplify the factor structure and render it more easily interpretable. Use of a graphical method called the Scree test [36] indicated that four factors should be retained. The next step was to name the retained factors. Names were given that reflected the food groups having the highest loadings on that factor (Table 1). Food groups having a positive loading on a factor contributed directly to that dietary pattern; food groups with negative loading were inversely associated with a given dietary pattern. When a food group loaded on more than one dietary pattern,

189 only the pattern with the highest loading was considered for factor naming. Factor loadings can be considered correlation coefficients between food groups and dietary patterns and take values between ÿ1 and C1. We next calculated the factor score for each of the four dietary patterns. Briefly, factor scores are formed by standardising each variable (food group) to have zero mean and standard deviation of 1, weighting it with a corresponding factor score coefficient, and then summing the terms. Thus, for each participant the factor score indicated the extent to which her/his diet conformed to one of the dietary patterns identified. A high factor score for a given dietary pattern indicated high intake of the food groups constituting that food pattern, and a low score indicated low intake of those food groups. Calculation and testing of means The crude means of the four dietary pattern scores were calculated for men and women separately. Sex-standardised dietary pattern mean scores were then calculated for men and women. Differences between standardised means were tested using the t-test for dichotomous variables (high school education, presence of hypertension, presence of hyperlipidaemia and whether on diet), and by univariate regression for quantitative categorised variables (BMI, WHR and PAL) inserting the category number into the model. Differences between recruitment centres (categorical variable) in terms of dietary pattern were tested by the F test. The analyses were always conducted using the maximum number of subjects for whom data were available, although this number could vary as some data items were missing for some subjects. All analyses were performed using the Stata 7.0 statistical package.

Results Four dietary patterns that explained 21% of the variance in the original dietary variables were identified; their characteristics are summarised in Table 1. The first dietary pattern, which we called prudent, was characterised by high loadings of cooked vegetables, legumes, cabbage and fish, and also seed oil as added fat. The second pattern, pasta & meat had high loadings of pasta, tomato sauce, red meat, processed meat, bread and wine. The third pattern, olive oil & salad, was characterised by high consumption of raw vegetables, olive oil as added fat, soup, and chicken. The last

190 Table 1

V. Pala et al. Factor analysis in Italian EPIC elderly cohort: structures of the four rotated factors

Dietary factor

Loading coefficients of food groups

Factor 1

OTHER VEGETABLES LEGUMES (PULSES) LEAF VEGETABLES-COOKED ONIONS, GARLIC CABBAGE FISH CRUSTACEANS, MOLLUSCS MUSHROOMS SEED OILS TOMATOES e COOKED FRESH FRUIT (non-citrus) NUTS AND SEEDS SNACKS

Factor 2

PASTA, POLENTA COUS-COUS, ETC BEEF OTHER ANIMAL FATS TOMATOES-COOKED WINE BREAD PROCESSED MEAT PORK YOGHURT

Factor 3

OLIVE OIL TOMATOES-RAW LEAF VEGETABLES-RAW ROOT VEGETABLES SOUPS CHICKEN/TURKEY

Factor 4

SUGAR, HONEY, JAM ICE CREAM CHOCOLATE-BASED CONFECTIONERY CAKES, PUDDINGS COFFEE PROCESSED MEAT EGGS MILK BUTTER CHEESES PATISSERIE, BISCUITS

Food name

Loading coefficient 0.694 0.617 0.541 0.529 0.438 0.425 0.420 0.370 0.332 0.327 0.326 0.306 0.298 0.624 0.611 0.588 0.503 0.495 0.437 0.433 0.382 ÿ0.301

Interpreted dietary pattern Prudent

Cumulative variance explained (%) 8

Pasta & meat

13

0.769 0.671 0.617 0.552 0.332 0.312

Olive oil & salad

17

0.461 0.438 0.436 0.430 0.391 0.362 0.352 0.337 0.329 0.310 0.305

Sweet & dairy

21

Food groups are sorted by dimension of loading coefficient. Negative loading are listed at the foot. When a food group loads more than one dietary pattern, it is normally listed in the pattern with the highest loading and it is listed in italic in the other patterns.

pattern, sweet & dairy was characterised by high consumption of sugar, cakes, ice cream, coffee, eggs, butter, milk and cheese. Table 2 shows crude mean scores for the four dietary patterns in men and women, and P values for differences between them. It is evident that the sweet and diary pattern differed least between men and women. Tables 3 and 4 show, for men and women respectively, the crude mean factor scores of the

four dietary patterns, for each covariate. The mean factor scores for each sex were standardized to zero mean and unit SD and have the same units; as a consequence they are directly comparable between patterns, covariates, and also between men and women. Only crude mean scores are shown as they differed little from age- and energy-adjusted ones (data not shown). The descriptions that follow of associations between dietary patterns and covariates do not

Dietary patterns in elderly Italians

191

Table 2 Comparison of crude mean values of dietary pattern scores according to sex Dietary pattern

Crude mean score

P values for between Men Women a (nZ1536) (nZ4075) sexes

Prudent ÿ0.16 Pasta & meat 0.76 Olive oil & salad 0.16 Sweet & dairy ÿ0.04 a

0.06 ÿ0.29 ÿ0.06 0.02

!0.001 !0.001 !0.001 0.046

From t-test.

imply causal relationships, in view of the crosssectional nature of the associations.

Recruitment centre All factor scores differed significantly between the Italian centres. In particular the prudent pattern was strongly characteristic (high factor scores) of women in the southern Italian centres of Naples and Ragusa, with a difference between the highest (Naples) and lowest score (Varese, northern Italy) of C2.18. The pasta & meat and sweet & dairy patterns differed less markedly between centres, particularly in men.

Education People who completed high school had high scores on the prudent pattern, an association that remained significant when participants from each centre were analysed separately (data not shown). Score differences between those with and without a high school education were C0.18 (PZ0.002) for men and C0.34 (P!0.001) for women. By contrast, people with a high-school education had low scores on the pasta & meat pattern. Score differences between those without and with a high school education were ÿ0.29 in men (P!0.001) and ÿ0.08 (PZ0.05) in women. The olive oil & salad dietary pattern had no significant associations with education, whereas the sweet & dairy dietary pattern in women was weakly associated with completion of high school C0.08 (PZ0.04).

Dieting About 15% of men and 19% of women said they had dieted over the previous year. In men and women who had dieted, the olive oil & salad pattern was significantly higher than in those who had not dieted (C0.31 and C0.45 changes respectively,

both P!0.001); while the pasta & meat and sweet & dairy patterns were significantly less prominent (both P!0.001) in those who had dieted (changes of ÿ0.40 and ÿ0.27 respectively for men and ÿ0.41 and ÿ0.23 for women compared to those who had not dieted). In women only, having dieted was also significantly associated with non-adherence to the prudent pattern (ÿ0.11; PZ0.003).

Hypertension and hyperlipidaemia About 32% of men and 37% of women said they had been diagnosed as hypertensive by a doctor. Hypertension was not related to any dietary pattern in men and was weakly associated with higher prudent type foods in women (C11). When the analysis was carried out in women after excluding those who had been on a diet, hypertension was associated with adherence to the pasta & meat patterns (difference vs. normotensive C0.06; PZ0.08). About 32% of men and 36% of women said they had been diagnosed as hyperlipidaemic by a doctor. Hyperlipidaemia was associated with dietary pattern in both sexes. Hyperlipidaemic men and women consumed less pasta & meat and sweet & dairy and more olive oil & salad patterns than those with normal blood lipids. When the analysis was carried out in men after excluding those who had dieted, the pasta & meat and olive oil & salad patterns were no longer associated with hyperlipidaemia.

Body mass index The pasta & meat pattern (both sexes) increased with increase in BMI: Pasta & meat score differences between the lowest and highest BMI categories were C0.55 and C0.39, respectively (P!0.001 for both). In men only, the olive oil & salad dietary pattern increased significantly with BMI with a C0.42 score change passing from obese to severely obese (BMI O35); however when those who had dieted were excluded from the BMI analysis the association was no longer evident. Obese women tended to consume less foods of the sweet & dairy pattern: the difference in score between the lowest and highest category of BMI was ÿ0.25 (P!0.01). The association of the prudent pattern with BMI, and its relation to dieting differed in men and women: Overweight men consumed this pattern less but after excluding those who dieted this negative association was no longer evident.

192

Table 3

Crude mean valuesa of sex-standardised dietary pattern scores, in men according to covariate categories (nZ1536)

Covariate

Categories

Dietary pattern Prudent

Pasta & meat

Olive oil & salad

Sweet & dairy

Standardised mean score

P value

Standardised mean score

P value

Standardised mean score

P value

Standardised mean score

P value

Italian recruitment centre

Florence (nZ374) Varese (nZ417) Ragusa (nZ206) Turin (nZ539)

0.02 ÿ0.30 0.76 ÿ0.76

!0.001b

0.16 0.14 ÿ0.22 ÿ0.13

!0.001b

0.09 ÿ0.05 ÿ0.65 0.23

!0.001b

0.05 0.28 ÿ0.09 ÿ0.22

!0.001b

High school education

No (nZ1108) Yes (nZ422)

ÿ0.05 0.13

0.002c

0.08 ÿ0.21

!0.001c

ÿ0.01 0.03

0.52c

ÿ0.03 0.07

0.11c

Recent modification of habitual diet

No (nZ1304) Yes (nZ225)

ÿ0.01 ÿ0.05

0.38c

0.06 ÿ0.34

!0.001c

ÿ0.05 0.26

!0.001c

0.04 ÿ0.23

!0.001c

Hypertension

No (nZ1043) Yes (nZ491)

0.00 0.00

0.90c

ÿ0.01 0.02

0.59c

0.00 0.00

0.90c

0.01 ÿ0.03

0.50c

Hypertension (on-diet excluded)

No (nZ892) Yes (nZ410)

ÿ0.02 0.00

0.79c

0.04 0.08

0.49c

ÿ0.05 ÿ0.04

0.86c

0.06 ÿ0.01

0.25c

Hyperlipidemia

No (nZ1050) Yes (nZ484)

ÿ0.02 0.04

0.31c

0.05 ÿ0.11

0.005c

ÿ0.03 0.07

0.06c

0.09 ÿ0.19

!0.001c

Hyperlipidemia (on-diet excluded)

No (nZ942) Yes (nZ361)

ÿ0.02 0.01

0.61c

0.08 ÿ0.01

0.16c

ÿ0.06 ÿ0.02

0.51c

0.12 ÿ0.16

!0.001c

Body mass index

!25 (nZ455) R25 & !30 (nZ838) R30 & !35 (nZ209) R35 (nZ34)

ÿ0.06 ÿ0.01

0.09d

ÿ0.07 ÿ0.01

0.001d

ÿ0.05 ÿ0.01

0.014d

!25 (nZ455) R25 & !30 (nZ838)

0.13

ÿ0.33 ÿ0.05 ÿ0.01

0.44

ÿ0.03 ÿ0.07

ÿ0.07

0.06

0.48 d

ÿ0.02

0.48 0.004

d

ÿ0.07 ÿ0.05

0.35d

d

0.30

0.08 0.05

0.13d

V. Pala et al.

Body mass index (on diet excluded)

0.21

0.02 0.00

R30 & !35 (nZ209) R35 (nZ34) e

Waist-hip ratio

Waist-hip ratio (on diet excluded)e

Physical activity level

Physical activity level (on diet excluded)

a b c d e

Tertile 1 (range 0.66e0.92) Tertile 2 (range 0.92e0.97) Tertile 3 (range 0.97e1.27) Tertile 1 (range 0.66e0.92) Tertile 2 (range 0.92e0.97) Tertile 3 (range 0.97e1.27)

0.13 ÿ0.42 0.02

ÿ0.03

0.14 0.54 0.42

d

ÿ0.11

ÿ0.08 ÿ0.02

0.21 !0.001

d

0.05

0.015

d

0.03

ÿ0.10

0.00

ÿ0.01

ÿ0.03

0.07

0.11

ÿ0.04

0.00

0.01

0.48d

ÿ0.06

0.001d

0.00

0.18d

0.07

ÿ0.09

0.07

ÿ0.05

0.02

0.06

0.16

ÿ0.09

0.03

Tertile 1 (range 1.28e1.44) Tertile 2 (range 1.44e1.57) Tertile 3 (range 1.57e2.3)

ÿ0.08

Tertile 1 (range 1.28e1.44) Tertile 2 (range 1.44e1.57) Tertile 3 (range 1.57e2.3)

ÿ0.10

!0.001d

ÿ0.05

0.003d

ÿ0.04

0.29d

ÿ0.01

0.08

0.04

0.17

ÿ0.07

0.19

0.16

ÿ0.04

0.13

!0.001d

0.00

0.003d

ÿ0.09

0.35d

0.03

0.04

0.13

0.11

ÿ0.03

0.25

0.21

ÿ0.08

0.17

0.73d

0.57d

Dietary patterns in elderly Italians

Table 3 (continued)

0.16d

0.15d

Significant differences in bold (P!0.05). From F test, one-way ANOVA. From two tailed t-test. P for trend. Tertiles based on distribution in men.

193

194

Table 4

Crude mean valuesa of sex-standardised dietary pattern scores, in women according to covariate categories (nZ4075)

Covariate

Categories

Dietary pattern Prudent

Pasta & meat

Standardised Mean Score

P value !0.001b

Standardised Mean Score

Olive oil & salad P value !0.001b

Standardised Mean Score

Sweet & dairy P value !0.001b

Standardised Mean Score

P value !0.001b

Italian recruitment centre

Florence (nZ1444) Varese (nZ1395) Ragusa (nZ178) Turin (nZ500) Naples (nZ558)

ÿ0.19

High school education

No (nZ3106) Yes (nZ963)

ÿ0.08 0.26

!0.001c

0.02 ÿ0.06

0.05c

0.01 ÿ0.03

0.23c

ÿ0.02 0.06

0.04c

Recent modification of habitual diet

No (nZ3278) Yes (nZ775)

0.02 ÿ0.09

0.003c

0.08 ÿ0.33

!0.001c

ÿ0.09 0.36

!0.001c

0.04 ÿ0.19

!0.001c

Hypertension

No (nZ2579) Yes (nZ1485)

ÿ0.04 0.07

0.001c

ÿ0.01 0.02

0.37

c

ÿ0.00 0.01

0.68c

0.02 ÿ0.04

0.06c

Hypertension (on-diet excluded)

No (nZ2127) Yes (nZ1141)

ÿ0.03 0.11

!0.001c

0.06 0.12

0.08c

ÿ0.08 ÿ0.10

0.66c

0.06 0.01

0.17c

Hyperlipidemia

No (nZ2596) Yes (nZ1466)

ÿ0.02 0.04

0.07c

0.06 ÿ0.11

!0.001c

ÿ0.05 0.09

!0.001c

0.06 ÿ0.11

!0.001c

Hyperlipidemia (on-diet excluded)

No (nZ2242) Yes (nZ1027)

ÿ0.02 0.01

0.02c

0.08 ÿ0.01

0.001c

ÿ0.06 ÿ0.02

0.001c

0.12 ÿ0.16

0.001c

Body mass index

!25 (nZ1418) R25 & !30 (nZ1851) R30 & !35 (nZ627) R35 (nZ179)

ÿ0.13 0.00

!0.001d

ÿ0.13 0.03

!0.001d

ÿ0.02 0.01

0.56d

0.05 ÿ0.01

0.003d

ÿ0.45 0.32 ÿ0.25 1.73

0.11 ÿ0.13 ÿ0.13 ÿ0.32 0.37

0.15 0.05 ÿ0.24 0.42 ÿ0.81

0.05 0.23 ÿ0.25 ÿ0.11 ÿ0.50

0.13

0.02

ÿ0.02

0.34

0.26

ÿ0.03

ÿ0.20

V. Pala et al.

0.19

Body mass index (on diet excluded)

Waist-hip ratio

e

Waist-hip ratio (on diet excluded)e

Physical activity level

Physical activity level (on diet excluded)

a b c d e

!25 (nZ1164) R25 & !30 (nZ1492) R30 & !35 (nZ485) R35 (nZ137)

ÿ0.13 0.02 0.27

ÿ0.17

Tertile 1 (range 0.56e0.79) Tertile 2 (range 0.79e0.85) Tertile 3 (range 0.85e1.34)

ÿ0.17

Tertile 1 (range 1.28e1.44) Tertile 2 (range 1.44e1.57) Tertile 3 (range 1.57e2.3)

0.53

ÿ0.07 0.13

!0.001d

!0.001

ÿ0.15

0.26d

0.12 0.02

!0.001

d

0.08

ÿ0.19 !0.001

d

0.10

ÿ0.03

0.00

0.01

0.01

0.19

0.15

ÿ0.10

ÿ0.11

!0.001d

ÿ0.09

!0.001d

0.01

!0.001d

0.15

0.00

0.08

ÿ0.06

0.06

0.25

0.25

ÿ0.22

ÿ0.08

!0.001d

0.09

0.22d

ÿ0.35

!0.001d

ÿ0.14

ÿ0.23

ÿ0.06

0.04

0.05

ÿ0.06

0.01

0.15

0.03

0.62

!0.001d

0.16

0.39d

ÿ0.43

!0.001d

ÿ0.01

ÿ0.14

0.31 d

ÿ0.07 ÿ0.09 ÿ0.11

0.22

0.42

Tertile 1 (range 0.56e0.79) Tertile 2 (range 0.79e0.85) Tertile 3 (range 0.85e1.34)

Tertile 1 (range 1.28e1.44) Tertile 2 (range 1.44e1.57) Tertile 3 (range 1.57e2.3)

!0.001d

!0.001d

0.03

ÿ0.24

0.01

0.11

ÿ0.03

ÿ0.06

0.10

ÿ0.08

0.17

!0.001d

Dietary patterns in elderly Italians

Table 4 (continued)

!0.001d

!0.001d

!0.001d

Significant differences in bold (P!0.05). From F test, one-way ANOVA. From two tailed t-test. P for trend. Tertiles based on distribution in women.

195

196 Overweight women consumed this pattern more and this was also evident after excluding those who had dieted.

Waist to hip ratio Women with a prominent waist (high WHR) had high scores on the prudent pattern, with differences between the lowest and highest WHR tertiles of C0.36 (P!0.001). Pasta & meat pattern scores were significantly (P!0.001) associated with WHR in both sexes, with differences between those with and without a prominent waist of C0.22 in men and C0.30 women (P!0.001 for both). The olive oil & salad and sweet & dairy dietary patterns were less characteristic of women with a prominent waist: Differences between lowest and highest scores were ÿ0.18 and ÿ0.21 respectively (P!0.001 for both). Men with a prominent waist consumed less of the olive oil & salad pattern (ÿ0.9, PZ0.015) but when those who dieted were excluded this association disappeared.

Physical activity In men, the prudent and the pasta & meat patterns were directly associated with high PAL, with an increase by C0.27 (P!0.001) and C21 (PZ0.003) respectively, passing form the lowest to the highest PAL tertile. Women behaved differently: The prudent pattern was associated with the most sedentary lifestyle. In particular there was a marked decrease in prudent score (ÿ0.76) passing from the lowest to intermediate activity levels. The pasta & meat pattern was not associated with PAL in women. In women, the olive oil & salad and sweet & dairy patterns were both associated with an active lifestyle, with a increase in score passing from the lowest to highest activity levels of C0.50 (P!0.001), and C0.17 (P!0.001), respectively.

Discussion We identified four major dietary patterns in our sample of Italian elderly. Two of these patterns, prudent and olive oil & salad, are variants of the traditional Mediterranean diet. The third, pasta & meat is an Italian variant of a western type energydense diet. The fourth, sweet & dairy reflects a preference for sweet, dairy and easily prepared foods.

V. Pala et al. The prudent pattern is characterised by high consumption of cooked vegetables including cabbage and pulses, with significant consumption also of fruit and fish and important use of seed oils. Thus it resembles the traditional Mediterranean diet in many aspects [37e40] except for the replacement of olive oil by seed oil. Examination of Tables 3 and 4 suggests that the prudent pattern was perceived differently by men and women, in that it was avoided by overweight men who had dieted but was prominent in overweight women who had dieted. The prudent pattern was characteristic of sedentary, educated women from the south of Italy, who had high BMI and considerable abdominal fat (high WHR); it was particularly characteristic of such women who were also hypertensive and hyperlipidaemic. It was surprising to find that a dietary pattern characterised by seed oil consumption was prominent in the south of Italy where much olive oil is produced. However in a recent study on Greek participants of EPIC, it was also found that one of the principal dietary components identified, based on fruit, vegetables and seed oil, was very close to this prudent pattern [37]. Following the study of Keys et al. [41] which indicated that a low polyunsaturated fat (PUFA) to saturated fat ratio was associated with coronary artery disease, the consumption of seed oil instead of butter and other animal fats came to be recommended by cardiologists for middle-aged people as a means of reducing the risk of cardiovascular disease. Thus, the prominence of seed oils in the prudent dietary pattern may appear to be the healthy choice. However, studies carried out in the 1990s indicated that oils rich in monounsaturated fats (MUFA), and in particular olive oil, may be effective in reducing the risk of cardiovascular disease [42] and the role of olive oil in the prevention and treatment of hyperlipidaemia and insulin-resistance has since been clarified [43]. More recently, increased intake of linoleic acid (the dominant component of seed oil) has been associated with chronic aetherogenic inflammation in arterial walls [44,45] via a mechanism involving enhanced production of leucotrienes [46,47]. Notwithstanding accumulating evidence that high n-6 PUFA consumption is not the healthy choice, medical advice to eat more seed oils has not changed: even recent health promotion programs advise cutting out butter and other animal fats and eating more seed oils rich in PUFAs to reduce the risk of cardiovascular disease [7]. This inertia may be partly due to the perception that it is important to give simple messages that should not be changed too often. Clearly a considerable

Dietary patterns in elderly Italians educational and promotional effort e directed at physicians and other health professionals, as well as the public e will be necessary to overturn the perception that seed oils are good and instil the idea that there are healthier alternatives. Another plausible reason for the dominance of seed oils in the prudent dietary pattern is cost. Although southern Italy is a major olive oilproducing area, our investigations indicate that the most popular brand of extra-virgin olive oil is about 11 times more expensive than the most popular brand of seed oil (source Delta-Piu ` Group, personal communication). Furthermore a recent study [48] on women from Naples (the city in which the prudent dietary pattern was most prominent) showed that seed oil consumption was significantly characteristic of people of low socio-economic status. The olive oil & salad pattern, with high consumption of raw vegetables and olive oil, and white meat as major protein source, also has much in common with the traditional Mediterranean diet. It may be viewed as a modern healthy diet [49] and was characteristic of people living in northern and central Italy, especially the metropolitan areas of Turin and Florence. Neither of these cities are near the sea and this may be the reason for the low fish consumption in the pattern. The olive oil & salad pattern was significantly more evident in people who had dieted, suggesting it was perceived as healthy. Men with a prominent waist, and men and women who were overweight tended to choose the olive oil & salad pattern if they dieted (Tables 3 and 4). Hyperlipidaemic women and men were also inclined to consume foods belonging to this pattern, and this was true even in the subgroup who said they had not dieted recently, suggesting a long-standing perception that the diet is healthy. In women hyperlipidaemia typically arises in early post-menopause [50] so the hyperlipidaemic women in our cohort (all above 60 years) will probably have changed to this pattern some years previously. The pasta & meat pattern, with high consumption of red meat, salami and other cured meats and animal fat, combined with spaghetti and tomato sauce, white bread and red wine, was characteristic of less educated people, particularly men, in all the Italian centres we investigated. The unhealthy characteristics of this pattern seem to be well appreciated, since it was significantly reduced among those of both sexes who had dieted. The sweet & dairy pattern was characterised by high consumption of foods typical of the continen-

197 tal breakfast (milk, sugar, coffee, biscuits, butter, and cakes) as well as quickly available and quickly eaten foods like ice-cream, chocolate cakes and eggs. Cohort members in northern and central Italy, especially women, were most likely to adhere to this pattern. Women consuming it were more educated and slimmer. The latter remained true after excluding Neapolitan women (data not shown) who had very low scores on this pattern, which may have created a centre-BMI interaction. The unhealthy characteristics of this pattern seem to be well appreciated, since it was significantly reduced among those of both sexes who had dieted. In both sexes, this pattern was negatively associated with hyperlipidaemia, even in the subgroup that had not dieted. The likely explanation is that hyperlipidaemic people have moved away from this pattern of eating following advice from their doctors. A number of methodological limitations of the present study should be noted. First, the EPIC population consists of people who volunteered to participate and have their health status followed for the rest of their lives. Thus the cohort is selected and unrepresentative of the Italian elderly population as a whole. In particular, people over 70 years are underrepresented (none were older than 77 years) and hence the dietary patterns revealed almost certainly are not those consumed by very old people. Second, the cross-sectional design of the study provided a snapshot of eating habits and other characteristics at a point in time, thus the possibility of inverse causation bias arises [51]. This is particularly likely for the observed associations between dietary patterns and anthropometric variables, hyperlipidaemia, and hypertension, and it is not possible for our study to determine whether a given pattern influenced these variables or whether the variables are determiners of a particular type of diet. We analysed out data from an analytical epidemiological point of view, looking for cross-sectional associations between variables of socioeconomic status, anthropometry, and health, and the dietary patterns revealed. We are aware that patterns derived by factor analysis should only be related with prudence to health outcomes, since factor analysis does not presuppose prior information about a relation between diet and disease [25]. Another possible limitation is that food intake was obtained in the Italian EPIC centres using three different food frequency questionnaires. We consider it an advantage of this study, and of EPIC in general, that the different questionnaires were developed and validated to capture the local

198 dietary characteristics of the areas studied. Nevertheless use of different questionnaires raises the possibility that differences between centres were emphasized. However dietary patterns are, by their nature, less prone (compared to estimates of intakes of different foods) to questionnaire bias since modifying the instrument (questionnaire) is unlikely to modify relations between food groups, though it is likely to affect estimates of quantities. Furthermore, patterns closely resembling those of present study have been identified in other populations [25] suggesting they are robust and reliable. A common criticism of pattern analysis is that collapsing the primary data into a smaller number of items is an inherently subjective procedure [52]. To overcome this we followed widely-accepted rules for choosing the number of factors to be retained [53,54]. Furthermore we incorporated the maximum possible information available from the dietary questionnaires, by grouping items that differed in the three questionnaires into more general categories. We acknowledge that the three Italian questionnaires tended to overemphasize the role of certain food groups in pattern construction (e.g. inclusion ten sub-groups for vegetables could overestimate the role of vegetables). However the choice of these food groupings was motivated by disease aetiology hypotheses when the questionnaires were being developed. A further justification of the groupings is that a given food item may belong to different food group depending on how it is prepared, and in turn associated with different lifestyles. We now compare the characteristics of the four dietary patterns identified in the present study with those that distinguish the traditional Mediterranean diet [55] with a view to defining strategies that may improve the diet of elderly Italians.

V. Pala et al. not the central and northern parts of Italy. The lower predominance of the pulse-rich pattern among the younger elderly women in the cohort is of concern, particularly since the sweet & dairy pattern was the second most prominent pattern among the women. It might seem difficult to persuade old people to eat more pulses since they are likely to have problems with digestion. Fortunately, Mediterranean cuisine has numerous ways of preparing pulses that are both appetising and digestible and likely to be highly acceptable to old people.

Cereal-based foods Pasta, polenta, rice and other grains were consumed fairly widely by the Italian EPIC cohort [34] but always in their highly refined forms: none of the four retained patterns had positive loadings for wholemeal bread, while wholemeal bread had negative loading (coefficient ÿ0.24) on the pasta & meat pattern (not shown) only shows loadings with absolute coefficients (O0.30). Low glycaemic index foods [56e58] and in particular wholemeal grains, are almost universally advocated as part of a healthy diet [59]. Again however, elderly people may have problems with these foods since they may reduce the mineral bioavailability [60e62]. However, sourdough bread is a tradition of Mediterranean areas and the use of sourdough yeast cultures with wholemeal flour can reduce phytate content and hence improve mineral bioavailability [63e66]. Thus elderly people should by encouraged to gradually increase their consumption of low glycaemic index foods such as the sourdough bread common in southern Italy, and also pasta poor in cellulose but rich in resistant starch.

Consumption of fruit and vegetables Monounsaturated-saturated fatty acid ratio We found that the pasta & meat pattern, characterised by high consumption of foods rich in saturated fatty acids, was strongly related to poorly educated males. Thus elderly men (rather than women) would be good targets for health campaigns that emphasise the use of fish, vegetables and pulses as a traditional, tasty and healthy choice, and advise reduced consumption of red meat, cured meat and animal fats.

Pulses as soluble fibre source and alternative to red meat protein source Pulses formed part of the prudent dietary pattern, that characterised the southern Italian centres but

Analysis of the entire Italian EPIC cohort [34,67] and studies on cohorts from other Mediterranean countries [68] have revealed that the young, particularly the less educated young are consuming less fruit and vegetables. This trend is particularly evident in southern Italy [67]. By contrast, fruit was an important component of one, and vegetables of two, of the four dietary patterns identified, while high consumption of cooked tomatoes (mainly as a component of sauces) also characterised the less healthy pasta & meat pattern. Furthermore, about 18% of our cohort reported they had dieted during the year preceding the interview. These dieters showed a clear tendency to change from the pasta & meat and sweet & dairy patterns to the olive oil & salad

Dietary patterns in elderly Italians pattern. Such a change is likely to have been beneficial even though most of the advantages of dietary change are considered to accrue in the long-term [69]. We can therefore conclude that the positive effects of fruit and vegetables are well-appreciated by our elderly cohort and this is a reassuring finding. Nevertheless the diet of elderly Europeans is of increasing concern. The European population as a whole is ageing, and monotonous diets, poor in essential nutrients have been reported in a previous study on elderly Mediterranean’s [68]. Paradoxically also, overweight and obesity are reported in high proportions of elderly people from western countries [70,71]. It is also noteworthy that authoritative advice specifically targeted to old people is lacking, while dietary recommendations for the over fifties are currently the same as those for 25e50 year-olds [72e76]. In such a context, realistic, culturally acceptable information campaigns based on information from studies such as the present one are likely to make a positive contribution to the health of the elderly.

199

[2]

[3]

[4]

[5]

[6]

[7]

[8]

Acknowledgements [9]

The authors wish to thank all who participated in or collaborated with EPIC Italy. Thanks are also due to Don Ward for helping with the English and to Sara Grioni for critical reading. EPIC Italy is supported by a generous grant from the Associazione Italiana per la Ricerca sul Cancro (AIRC, Milan). EPIC is also supported by the European Union. This study was supported in part by the Quality of Life and Management of Living resources Programme of the European Commission (DG Research, contract No QLK6-CT-2001-00241) for the project EPIC-Elderly, coordinated by the Department of Hygiene and Epidemiology, University of Athens Medical School and by the Europe against Cancer Programme, of the European Commission, (DG SANCO) for the project EPIC coordinated by the International Agency for Research on Cancer (WHO). The authors are solely responsible for the publication and the publication does not represent the opinion of the Community. The Community is not responsible for any use that might be made of data appearing in this work.

References [1] World Cancer Research Fund, American Institute for Cancer Research. Food, Nutrition and the Prevention of

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

Cancer: a global perspective. Washington, DC ed. World Cancer Research Fund American Institute for Cancer Research, 1997. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Actual causes of death in the United States, 2000. J Am Med Assoc 2004;291(10):1238e45. Larsson SC, Rafter J, Holmberg L, Bergkvist L, Wolk A. Red meat consumption and risk of cancers of the proximal colon, distal colon and rectum: The Swedish Mammography Cohort. Int J Cancer 2004;113(5):829e34. Dixon LB, Balder HF, Virtanen MJ, Rashidkhani B, Mannisto S, Krogh V, et al. Dietary patterns associated with colon and rectal cancer: results from the Dietary Patterns and Cancer (DIETSCAN) Project. Am J Clin Nutr 2004;80(4):1003e11. English DR, MacInnis RJ, Hodge AM, Hopper JL, Haydon AM, Giles GG. Red meat, chicken, and fish consumption and risk of colorectal cancer. Cancer Epidemiol Biomarkers Prev 2004;13(9):1509e14. Romieu I, Lazcano-Ponce E, Sanchez-Zamorano LM, Willett W, Hernandez-Avila M. Carbohydrates and the risk of breast cancer among Mexican women. Cancer Epidemiol Biomarkers Prev 2004;13(8):1283e9. Srinath RK, Katan MB. Diet, nutrition and the prevention of hypertension and cardiovascular diseases. Public Health Nutr 2004;7(1A):167e86. Millen BE, Quatromoni PA, Nam BH, O’Horo CE, Polak JF, Wolf PA, et al. Dietary patterns, smoking, and subclinical heart disease in women: opportunities for primary prevention from the Framingham Nutrition Studies. J Am Diet Assoc 2004;104(2):208e14. Sesso HD, Liu S, Gaziano JM, Buring JE. Dietary lycopene, tomato-based food products and cardiovascular disease in women. J Nutr 2003;133(7):2336e41. Pereira MA, Pins JJ. Dietary fiber and cardiovascular disease: experimental and epidemiologic advances. Curr Atheroscler Rep 2000;2(6):494e502. Liu S, Manson JE, Lee IM, Cole SR, Hennekens CH, Willett WC, et al. Fruit and vegetable intake and risk of cardiovascular disease: the Women’s Health Study. Am J Clin Nutr 2000;72(4):922e8. Esposito K, Marfella R, Ciotola M, Di Palo C, Giugliano F, Giugliano G, et al. Effect of a Mediterranean-style diet on endothelial dysfunction and markers of vascular inflammation in the metabolic syndrome: a randomized trial. J Am Med Assoc 2004;292(12):1440e6. Yoo S, Nicklas T, Baranowski T, Zakeri IF, Yang SJ, Srinivasan SR, et al. Comparison of dietary intakes associated with metabolic syndrome risk factors in young adults: the Bogalusa Heart Study. Am J Clin Nutr 2004; 80(4):841e8. Bingham SA, Luben R, Welch A, Wareham N, Khaw KT, Day N. Are imprecise methods obscuring a relation between fat and breast cancer? Lancet 2003;362(9379):212e4. Boyd NF, Stone J, Vogt KN, Connelly BS, Martin LJ, Minkin S. Dietary fat and breast cancer risk revisited: a meta-analysis of the published literature. Br J Cancer 2003;89(9):1672e85. Prentice RL, Sugar E, Wang CY, Neuhouser M, Patterson R. Research strategies and the use of nutrient biomarkers in studies of diet and chronic disease. Public Health Nutr 2002;5(6A):977e84. Darnton-Hill I, Nishida C, James WP. A life course approach to diet, nutrition and the prevention of chronic diseases. Public Health Nutr 2004;7(1A):101e21. Fung TT, Rimm EB, Spiegelman D, Rifai N, Tofler GH, Willett WC, et al. Association between dietary patterns

200

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36] [37]

V. Pala et al. and plasma biomarkers of obesity and cardiovascular disease risk. Am J Clin Nutr 2001;73(1):61e7. Hu FB, Rimm EB, Stampfer MJ, Ascherio A, Spiegelman D, Willett WC. Prospective study of major dietary patterns and risk of coronary heart disease in men. Am J Clin Nutr 2000;72(4):912e21. Schulze MB, Hu FB. Dietary patterns and risk of hypertension, type 2 diabetes mellitus, and coronary heart disease. Curr Atheroscler Rep 2002;4(6):462e7. van Dam RM, Rimm EB, Willett WC, Stampfer MJ, Hu FB. Dietary patterns and risk for type 2 diabetes mellitus in U.S. men. Ann Intern Med 2002;136(3):201e9. Slattery ML, Boucher KM, Caan BJ, Potter JD, Ma KN. Eating patterns and risk of colon cancer. Am J Epidemiol 1998; 148(1):4e16. Terry P, Hu FB, Hansen H, Wolk A. Prospective study of major dietary patterns and colorectal cancer risk in women. Am J Epidemiol 2001;154(12):1143e9. Fung T, Hu FB, Fuchs C, Giovannucci E, Hunter DJ, Stampfer MJ, et al. Major dietary patterns and the risk of colorectal cancer in women. Arch Intern Med 2003; 163(3):309e14. Newby PK, Tucker KL. Empirically derived eating patterns using factor or cluster analysis: a review. Nutr Rev 2004; 62(5):177e203. Diehr P, Beresford SA. The relation of dietary patterns to future survival, health, and cardiovascular events in older adults. J Clin Epidemiol 2003;56(12):1224e35. Knoops KT, de Groot LC, Kromhout D, Perrin AE, MoreirasVarela O, Menotti A, et al. Mediterranean diet, lifestyle factors, and 10-year mortality in elderly European men and women: the HALE project. J Am Med Assoc 2004; 292(12):1433e9. Ledikwe JH, Smiciklas-Wright H, Mitchell DC, Miller CK, Jensen GL. Dietary patterns of rural older adults are associated with weight and nutritional status. J Am Geriatr Soc 2004;52(4):589e95. Riboli E, Kaaks R. The EPIC Project: rationale and study design. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol 1997;26(Suppl. 1):S6eS14. Palli D, Berrino F, Vineis P, Tumino R, Panico S, Masala G, et al. A molecular epidemiology project on diet and cancer: the EPIC-Italy Prospective Study. Design and baseline characteristics of participants. Tumori 2003;89(6):586e93. Mariotto A, Dally LG, Micheli A, Canario F, Verdecchia A. Cancer prevalence in Italian regions with local cancer registries. Tumori 1999;85(5):400e7. Pisani P, Faggiano F, Krogh V, Palli D, Vineis P, Berrino F. Relative validity and reproducibility of a food frequency dietary questionnaire for use in the Italian EPIC centres. Int J Epidemiol 1997;26(Suppl. 1):S152e60. Panico S, Dello IR, Celentano E, Galasso R, Muti P, Salvatore M, et al. Progetto ATENA, a study on the etiology of major chronic diseases in women: design, rationale and objectives. Eur J Epidemiol 1992;8(4):601e8. Pala V, Sieri S, Palli D, Salvini S, Berrino F, Bellegotti M, et al. Diet in the Italian EPIC cohorts: presentation of data and methodological issues. Tumori 2003;89(6):594e607. Salvini S, Saieva C, Sieri S, Vineis P, Panico S, Tumino R, et al. Physical activity in the EPIC cohort in Italy. IARC Sci Publ 2002;156:267e9. Cattell RB. The Scree test for the number of factors. Multivariate Behav Res 1966;1:245e76. Costacou T, Bamia C, Ferrari P, Riboli E, Trichopoulos D, Trichopoulou A. Tracing the Mediterranean diet through principal components and cluster analyses in the Greek population. Eur J Clin Nutr 2003;57(11):1378e85.

[38] Gerber M. Biofactors in the Mediterranean diet. Clin Chem Lab Med 2003;41(8):999e1004. [39] Contaldo F, Pasanisi F, Mancini M. Beyond the traditional interpretation of Mediterranean diet. Nutr Metab Cardiovasc Dis 2003;13(3):117e9. [40] Serra-Majem L, dlC Ngo, Ribas L, Tur JA. Olive oil and the Mediterranean diet: beyond the rhetoric. Eur J Clin Nutr 2003;57(Suppl. 1):S2e7. [41] Keys A. Effects of different dietary fats on plasma-lipid levels. Lancet 1965;17:318e9. [42] Trevisan M, Krogh V, Freudenheim J, Blake A, Muti P, Panico S, et al. Consumption of olive oil, butter, and vegetable oils and coronary heart disease risk factors. The Research Group ATS-RF2 of the Italian National Research Council. J Am Med Assoc 1990;263(5):688e92. [43] Rivellese AA, Auletta P, Marotta G, Saldalamacchia G, Giacco A, Mastrilli V, et al. Long term metabolic effects of two dietary methods of treating hyperlipidaemia. Br Med J 1994;308(6923):227e31. [44] Lotzer K, Spanbroek R, Hildner M, Urbach A, Heller R, Bretschneider E, et al. Differential leukotriene receptor expression and calcium responses in endothelial cells and macrophages indicate 5-lipoxygenase-dependent circuits of inflammation and atherogenesis. Arterioscler Thromb Vasc Biol 2003;23(8):e32e6. [45] Spanbroek R, Grabner R, Lotzer K, Hildner M, Urbach A, Ruhling K, et al. Expanding expression of the 5-lipoxygenase pathway within the arterial wall during human atherogenesis. Proc Natl Acad Sci USA 2003;100(3):1238e43. [46] Ferretti A, Nelson GJ, Schmidt PC, Kelley DS, Bartolini G, Flanagan VP. Increased dietary arachidonic acid enhances the synthesis of vasoactive eicosanoids in humans. Lipids 1997;32(4):435e9. [47] Kelley DS, Taylor PC, Nelson GJ, Mackey BE. Arachidonic acid supplementation enhances synthesis of eicosanoids without suppressing immune functions in young healthy men. Lipids 1998;33(2):125e30. [48] Panico S, Celentano E, Galasso R. Cardiovascular risk in women: role of social status. Ann Ig 2000;12(1 Suppl. 1):31e5. [49] Willett WC. The dietary pyramid: does the foundation need repair? Am J Clin Nutr 1998;68(2):218e9. [50] Dalal D, Robbins JA. Management of hyperlipidemia in the elderly population: an evidence-based approach. South Med J 2002;95(11):1255e61. [51] Kleibaum DG KLMK. Applied Regression Analysis and Other Multivariate Methods. Pacific Grove, CA: Duxbury; 1988. [52] Martinez ME, Marshall JR, Sechrest L. Invited commentary: Factor analysis and the search for objectivity. Am J Epidemiol 1998;148(1):17e9. [53] Balder HF, Virtanen M, Brants HA, Krogh V, Dixon LB, Tan F, et al. Common and country-specific dietary patterns in four European cohort studies. J Nutr 2003;133(12): 4246e51. [54] Balder HF, Tan F, Brants HA, Dixon LB, Virtanen M, Krogh V, et al. Dietscan: a common approach for analysing dietary patterns. IARC Sci Publ 2002;156:27e9. [55] Willett WC, Sacks F, Trichopoulou A, Drescher G, FerroLuzzi A, Helsing E, et al. Mediterranean diet pyramid: a cultural model for healthy eating. Am J Clin Nutr 1995; 61(Suppl. 6):1402Se6S. [56] Muti P, Quattrin T, Grant BJ, Krogh V, Micheli A, Schunemann HJ, et al. Fasting glucose is a risk factor for breast cancer: a prospective study. Cancer Epidemiol Biomarkers Prev 2002;11(11):1361e8. [57] Riccardi G, Clemente G, Giacco R. Glycemic index of local foods and diets: the Mediterranean experience. Nutr Rev 2003;61(5 Pt 2):S56e60.

Dietary patterns in elderly Italians [58] Roberts SB, Pittas AG. The role of glycemic index in type 2 diabetes. Nutr Clin Care 2003;6(2):73e8. [59] Kushi LH, Meyer KA, Jacobs Jr DR. Cereals, legumes, and chronic disease risk reduction: evidence from epidemiologic studies. Am J Clin Nutr 1999;70(3 Suppl.):451Se8S. [60] Bohn T, Davidsson L, Walczyk T, Hurrell RF. Phytic acid added to white-wheat bread inhibits fractional apparent magnesium absorption in humans. Am J Clin Nutr 2004; 79(3):418e23. [61] Hurrell RF. Influence of vegetable protein sources on trace element and mineral bioavailability. J Nutr 2003;133(9): 2973Se7S. [62] Sandberg AS. Bioavailability of minerals in legumes. Br J Nutr 2002;88(Suppl. 3):S281e5. [63] Lopez HW, Krespine V, Guy C, Messager A, Demigne C, Remesy C. Prolonged fermentation of whole wheat sourdough reduces phytate level and increases soluble magnesium. J Agric Food Chem 2001;49(5):2657e62. [64] Navert B, Sandstrom B, Cederblad A. Reduction of the phytate content of bran by leavening in bread and its effect on zinc absorption in man. Br J Nutr 1985;53(1):47e53. [65] Yoon JH, Thompson LU, Jenkins DJ. The effect of phytic acid on in vitro rate of starch digestibility and blood glucose response. Am J Clin Nutr 1983;38(6):835e42. [66] Daniels DG, Fisher N. Hydrolysis of the phytate of wheat flour during bread making. Br J Nutr 1981;46(1):1e6. [67] Pala V, Berrino F, Vineis P, Palli D, Celentano E, Tumino R, et al. How vegetables are eaten in Italian EPIC centers. Still setting a good example? 01; Lyon ce ´dex 08. France: IARC PRESS; 2002.

201 [68] Schroder H, Marrugat J, Covas M, Elosua R, Pena A, Weinbrenner T, et al. Population dietary habits and physical activity modification with age. Eur J Clin Nutr 2004;58(2):302e11. [69] Willett WC. Nutritional epidemiology. New York: Oxford University Press; 1990. [70] Jensen GL, Friedmann JM. Obesity is associated with functional decline in community-dwelling rural older persons. J Am Geriatr Soc 2002;50(5):918e23. [71] Millen BE, Silliman RA, Cantey-Kiser J, Copenhafer DL, Ewart CV, Ritchie CS, et al. Nutritional risk in an urban homebound older population. The nutrition and healthy aging project. J Nutr Health Aging 2001;5(4): 269e77. [72] Touyz RM, Campbell N, Logan A, Gledhill N, Petrella R, Padwal R. The 2004 Canadian recommendations for the management of hypertension: Part III e lifestyle modifications to prevent and control hypertension. Can J Cardiol 2004;20(1):55e9. [73] Murphy SP, Poos MI. Dietary reference intakes: summary of applications in dietary assessment. Public Health Nutr 2002;5(6A):843e9. [74] Fernandes J. Nutrition and healtherecommendations of the Health Council of the Netherlands regarding energy, proteins, fats and carbohydrates. Ned Tijdschr Geneeskd 2002;146(47):2226e9. [75] Dietary reference intakes. Nutr Rev 1997;55(9):319e26. [76] SINU. Livelli di Assunzione Raccomandati di Energia e Nutrienti per la Popolazione Italiana (LARN); 1996.