[AQ7] Please clarify assessing also 4 y as the follow-up period.

Transcription

1 QUERIES -AJCN29764q [AQ1] Please confirm the edited statement regarding the receipt of a fellowship by JJB from the La Caixa Foundation is correct. [AQ2] Please confirm Medical School-Clinica Universidad de Navarra is correct in footnote 5. Also confirm whether reprints will be requested. [AQ3] Please clarify the follow-up intervals [eg, does only 2.3-y follow-up denote only 2.3 y of follow-up or only a 2.3-y follow-up (ie, a follow-up at 2.3 y)? Please amend each instance throughout the text accordingly. [AQ4] For clarity, our previous report was changed to the report by Sanchez-Villegas et al (10). [AQ5] For consistency, The metabolic equivalent (MET) index per week (METs-hours/week) was changed to The index of metabolic equivalent task hours per week (MET-h/wk) and subsequent instances of METs-h were changed to MET-h. Please confirm, or amend each instance accordingly. [AQ6] Should MET-h be changed to MET-h/wk? [AQ7] Please clarify assessing also 4 y as the follow-up period. [AQ8] MFP-Sa nchez-villegas was changed to the Mediterranean dietary pattern proposed by Sa nchez-villegas et al (21). Please confirm or amend. [AQ9] Should MET/wk be changed to MET-h/wk? [AQ10] Please spell out DIRECT. [AQ11] Please spell out Medi-RIVAGE and delete the acronym. [AQ12] For clarity, MDS was changed to modified Mediterranean dietary score. Okay as done, or is modified MDS-Trichopoulou meant? [AQ13] Two instances of MDP in the sentence beginning A previous report in our cohort [.] were spelled out as Mediterranean dietary pattern. Okay as done? [AQ14] Changed highest adherence to the MDS to highest adherence to the MDS-Trichopoulou. Okay as done? [AQ15] In the acknowledgments, please clarify the idea, the first idea (2 instances) and the initial idea. [AQ16] In the DS-Panagiotakos column of Appendix A, including like, cheese, yoghurt, and milk was changed to including cheese, yogurt, and milk Okay as done? [AQ17] z values was changed to z scores in the MDP-Sa nchez-villegas row of Appendix A. Okay as done? [AQ18] Clarify what the ranges are in reference to in last row. [AQ19] For 4 2 (highest), Footnote 2 (ref category) as intended? [AQ20] fiber intake ok as edited? [AQ21] Check edits for Journal style to footnotes 4 and 6. [AQ22] In footnote 5 of Table 5, Adjusted for sex, sex was changed to Adjusted for sex, age. Please confirm or amend.

2 Adherence to the Mediterranean diet, long-term weight change, and incident overweight or obesity: the Seguimiento Universidad de Navarra (SUN) cohort 1 5 Juan-José Beunza, Estefanía Toledo, Frank B Hu, Maira Bes-Rastrollo, Manuel Serrano-Martínez, Almudena Sánchez-Villegas, J Alfredo Martínez, and Miguel A Martínez-González ABSTRACT Background: The Mediterranean dietary pattern might be a potential tool for the prevention of obesity. Objectives: We studied the association between adherence to 6 previously published scores used to assess the adherence to the Mediterranean diet and weight change. We also assessed the risk of relevant weight gain (5 kg) or the risk of developing overweight or obesity. Design: The study population included 10,376 Spanish men and women who were university graduates (mean age = 38 y) and were followed-up for a mean (6SD) of y. Diet was assessed at baseline with a 136-item, previously validated food-frequency questionnaire. Weight was assessed at baseline and biennially during follow-up. Results: Participants with the lowest adherence (3 points) to the Mediterranean dietary score (MDS) proposed by Trichopoulou et al (range: 0 9; N Engl J Med 2003;348: ) exhibited the highest average yearly weight gain, whereas participants with the highest (6 points) adherence exhibited the lowest weight gain (adjusted difference: kg/y; 95% CI: , kg/y; P for trend = 0.02). This inverse association was extended to other a priori defined MDSs. The group with the highest adherence to the MDS also showed the lowest risk of relevant weight gain (5 kg) during the first 4 y of follow-up (odds ratio: 0.76; 95% CI: 0.64, 0.90). Conclusions: Adherence to the Mediterranean dietary pattern is significantly associated with reduced weight gain. This dietary pattern can be recommended to slow down age-related weight gain. Am J Clin Nutr 2010;92:1 8. INTRODUCTION Obesity is a priority for public health because of its high worldwide prevalence, its increasing trends (1), and its wellknown relation with higher mortality (2). Obesity is usually related to cumulative effects of small daily energy imbalance, and thus, both food habits and physical inactivity are considered to be key determinants of obesity risk (3). Recent nutrition research and large cohort studies have focused on food groups and food patterns (3) as opposed to individual nutrients. The Mediterranean dietary pattern has been proposed as one of the best models for healthy eating because of its favorable effect on increased survival (4) and the reported inverse associations with cardiovascular disease (5), diabetes (6), inflammation (5), and neurodegenerative disorders (4). Concern has been raised about the potential for weight gain associated with dietary patterns rich in vegetable fat such as the Mediterranean diet. However, some trials have shown a beneficial association of the Mediterranean dietary pattern with weight loss (7), and the European Prospective Investigation into Cancer and Nutrition (EPIC) Spain cohort confirmed a reduction in incident obesity associated with a higher adherence to the Mediterranean diet, although it was restricted only to overweight subjects (8). Another cohort did not find any association (9). In addition, a previous assessment in the Seguimiento Universidad de Navarra (University of Navarra Follow-Up Study) (SUN) cohort with only 2.3-y follow-up showed no significant association between adherence to the Mediterranean diet and weight changes (10). ½AQ3Š However, the follow-up period in the report by Sánchez-Villegas ½AQ4Š et al (10) might have been too short and did not use any of the most widely known scores of adherence to the Mediterranean diet to operationally define this food pattern. The aim of the current study was to assess the association between the most widely known scores used to appraise adherence to the Mediterranean food pattern and weight changes after a median follow-up of 6.2 y in the SUN cohort. 1 From the Departments of Preventive Medicine and Public Health (J-JB, ET, MB-R, and MAM-G) and Nutrition (JAM), University of Navarra, Navarra, Spain; the Departments of Epidemiology (J-JB and FBH) and Nutrition (FBH), Harvard School of Public Health, Boston, MA; the Department of Internal Medicine, Clínica la Luz, Madrid, Spain (MS-M); and the Department of Clinical Sciences, University of Las Palmas de Gran Canaria, Gran Canaria, Spain (AS-V). 2 J-JB and ET contributed equally to this study. 3 The funding entities had no role in the design, implementation, analysis, or interpretation of the data. 4 Supported by the Spanish Government (Instituto de Salud Carlos III, Fondo de Investigaciones Sanitarias; grants PI030678, PI040233, PI070240, PI070312, PI081943, PI080819, RD06/0045, and G03/140), the Navarra Regional Government (project PI41/2005), the University of Navarra (LE/97), and a fellowship from the La Caixa Foundation (to J-JB). 5 Address reprint requests and correspondence to: MA Martinez-Gonzalez, ½AQ1Š Department of Preventive Medicine and Public Health, Medical School-Clinica Universidad de Navarra, Irunlarrea 1, Pamplona, Navarra, Spain. ½AQ2Š mamartinez@unav.es. Received May 3, Accepted for publication September 26, doi: /ajcn Am J Clin Nutr 2010;92:1 8. Printed in USA. Ó 2010 American Society for Nutrition 1

3 2 BEUNZA ET AL SUBJECTS AND METHODS Study population The SUN project is an ongoing, multipurpose, prospective, and dynamic cohort of university graduates conducted in Spain. The study protocol was approved by the Institutional Review Board of the University of Navarra. The study methods were published in detail elsewhere (11). In short, the recruitment of participants started in Participants have been contacted through mailed questionnaires to gather their comprehensive baseline characteristics. Besides the questionnaire, they receive an invitation letter to participate in the study. Voluntary completion of the first questionnaire is considered providing informed consent. The enrollment was permanently open, and each year an average of new participants are admitted in the cohort. Followup is conducted through biennial mailed questionnaires. Nonrespondents receive up to 5 additional mailings requesting their follow-up questionnaire. To warrant a minimum follow-up of 2 y, 17,159 participants recruited before April 2007 were candidates to be eligible for this analysis because they had spent enough time in the study to be able to complete at least the first 2-y follow-up questionnaire. In these participants, the retention rate was 92%. Therefore, we had follow-up information of 15,753 participants. We excluded 3282 participants because of one or more of the following criteria: 1483 (10%) participants had values outside the predefined limits for total caloric intake (,2092 or.14,644 kj/d in women or,3347 or.16,736 kj/d in men) (12), 1799 (12%) female participants were pregnant at baseline or during follow-up, 1975 (13%) participants had prevalent cardiovascular disease or cancer at baseline, and 7 participants did not have information on weight. Thus, the effective sample size for the analyses was 10,376 participants. Exposure assessment Habitual diet was assessed at baseline with a semiquantitative 136-item food-frequency questionnaire previously validated in Spain (13). The validity and reproducibility of this questionnaire were recently replicated (14, 15). Each item in the questionnaire included a typical portion size. Daily food consumption was estimated by multiplying the portion size of each food item by its consumption frequency. The nutrient composition of the food items was derived from Spanish food-composition tables (16, 17). We used the Mediterranean dietary score proposed by Trichopoulou et al (18) (MDS-Trichopoulou), which was the score used for all our main analyses. As sensitivity analyses, we also tested other proposed scores, including the Mediterranean adequacy index proposed by Alberti-Fidanza et al (19), the Mediterranean diet quality index proposed by Scali et al (20), the Mediterranean dietary pattern proposed by Sánchez-Villegas et al (21), the diet score proposed by Panagiotakos et al (22), and the Mediterranean-style dietary pattern score recently proposed by Rumawas et al (23). The way these indexes were calculated is shown in Appendix A. Ascertainment of change in weight Weights of participants were recorded at baseline and every 2 y during follow-up. The reliability and validity of self-reported weight was previously assessed in a subsample of the cohort and was highly correlated with directly measured weight (r = 0.99; 95% CI: 0.99, 0.99) and exhibited a mean relative error in selfreported weight of 1.45% (24). Body mass index (BMI), defined as weight in kilograms divided by the square of the height in meters, was computed in the baseline questionnaire and during follow-up by using the follow-up weight. We carried forward the height value obtained in the baseline questionnaire to compute follow-up values for BMI. The outcomes were 1) average yearly change in body weight (in kg) as a continuous variable, 2) substantial weight gain between biennial questionnaires, defined as an increment 3 kg/y and, alternatively, as an increment 5 kg/y, and 3) incident overweight or obesity [participants with a BMI (in kg/m 2 ),25 at baseline and a BMI 25 during follow-up]. Assessment of nondietary variables The baseline assessment also included other questions (46 items for men and 54 items for women) to assess medical history, health habits, and lifestyle and sociodemographic variables (11). Physical activity was ascertained through a baseline 17-item questionnaire. The index of metabolic equivalent task hours per ½AQ5Š week (MET-h/wk) was computed by using the time spent engaging in 17 activities and multiplying the time spent by the resting metabolic rate (MET score) specific for each activity (25). The MET-h for all activities were combined to obtain ½AQ6Š a value of total MET-h/wk, which adequately correlated with energy expenditure measured by a triaxial accelerometer in a validation study in a subsample of the cohort (26). Statistical analyses Multiple linear regression was used to evaluate the average change in body weight (in kg) per year across categories of adherence to the Mediterranean dietary pattern. We fit an initial model adjusted only for age (continuous) and sex and a multivariable model adjusted for age (continuous), sex, baseline BMI (continuous), physical activity (MET-h/wk categorized in quartiles), sedentary behaviors (hours sitting/wk, categorized in quartiles), smoking (categorized as current, previous smokers, never smokers, and missing), between-meals snacking (dichotomous), and total energy intake (continuous). Beta coefficients, which represented adjusted between-group mean differences in weight gains, and their 95% CIs were calculated by taking participants with the lowest adherence to the Mediterranean dietary pattern as the reference group. We added an analysis stratified by sex and weight gain (,3 or3 kg) during the 5 y before recruitment. To specifically appraise the association between adherence to the Mediterranean diet and early weight-gain changes, we assessed the association between baseline adherence to the MDS- Trichopoulou categorized in 3 groups (with the lowest category of adherence as the reference) and the likelihood of gaining either 3 or5 kg during the first 2 y of follow-up. We repeated these analyses assessing also 4 y as the follow-up period. We used logistic regression analyses to calculate odds ratios (ORs) and ½AQ7Š

4 MEDITERRANEAN DIET AND WEIGHT CHANGE 3 their 95% CIs for each of these 2 periods (from baseline to the 2-y follow-up and from baseline to the 4-y follow-up) and the alternative cutoff points (3 or 5 kg) to define clinically meaningful weight gain. Variables included in the models for adjustment were the same as previously described for linear regression models. Finally, we used Cox regression analyses to evaluate the association between baseline adherence to the MDS-Trichopoulou and the incidence of overweight or obesity (BMI 25). For this final analysis, we excluded all subjects with overweight or obesity (BMI 25) at baseline. We assessed interactions between Mediterranean diet adherence and several variables, including sex and weight gain before enrollment. To assess these interactions, we introduced product terms in the different multivariable models and considered P, 0.05 in the likelihood ratio test as statistically significant. All statistical analyses were performed with SPSS version 15.0 software (SPSS Inc, Chicago, IL). All presented P values were 2- tailed. We conducted tests of linear trend (likelihood ratio test) across increasing categories of the score by using the score as a continuous variable and adjusting for potential confounders. RESULTS The mean (6SD) age of the 10,376 participants was y. Participants were followed up for an average of 5.7 y (median: 6.2 y). The mean yearly weight-change gain was kg. Baseline characteristics of the study participants according to their adherence to the Mediterranean diet (MDS-Trichopoulou) are presented in Table 1. Subjects with the highest adherence tended to be older and had a higher prevalence of hypercholesterolemia, but their high HDL-cholesterol concentrations accounted for this difference, were more likely to be nonsmokers or former smokers, and were physically more active. These subjects had higher total energy intakes with higher proportions of energy from carbohydrates and less proportions of energy from total fat intake; their intakes of saturated fatty acids were lower, but they had higher monounsaturated to saturated fatty acid ratios, which were probably related to reduced intakes of fast food, red meats, and whole-fat dairy products. The consumption of sugar-sweetened beverages was lower in these subjects; they tended to have a higher fiber and alcohol intake and were less likely to practice between-meals snacking. The average yearly weight change (in kg) according to adherence to the different Mediterranean scores is shown in Table 2. Participants with the highest adherence (6 points) to the MDS-Trichopoulou showed smaller weight gains than did the TABLE 1 Baseline characteristics according to adherence to the Mediterranean dietary score of Trichopoulou et al (18) (MDS-Trichopoulou) in the Seguimiento Universidad de Navarra project, MDS-Trichopoulou Low score (0 3) Moderate score (4 6) High score (7 9) P for trend n Sex (% women) Age (y) ,0.001 Weight (kg) ,0.001 BMI (kg/m 2 ) High total cholesterol, 240 mg/dl (%) ,0.001 High HDL cholesterol, 60 mg/dl (%) ,0.001 Smoking (%) Current smokers ,0.001 Former smokers ,0.001 Physical activity (MET-h/wk) ,0.001 Total energy intake (kcal/d) ,0.001 Carbohydrate (% of energy intake) ,0.001 Protein (% of energy intake) ,0.001 Vegetable protein (% of energy intake) ,0.001 Total fat (% of energy intake) ,0.001 MUFA (% of energy intake) ,0.001 SFA (% of energy intake) ,0.001 Ratio of MUFA to SFA ,0.001 PUFA (% of energy intake) ,0.001 Fiber (g/d) ,0.001 Alcohol (g/d) ,0.001 Sodium intake (g/d) ,0.001 Potassium intake (g/d) ,0.001 Snacking (%) ,0.001 Sugar sweetened beverages (drinks/d) ,0.001 Fast food (g/d) , MET-h/wk, metabolic equivalent task hours per week; MUFA, monounsaturated fatty acid; SFA, saturated fatty acid; PUFA, polyunsaturated fatty acid. 2 Mean 6 SD (all such values). 3 Sum of hamburgers, sausages, and pizza.

5 4 BEUNZA ET AL TABLE 2 Yearly weight change according to categories of adherence to the Mediterranean dietary pattern in the Seguimiento Universidad de Navarra project, Values P for trend MDS-Trichopoulou Scores 0 3 (lowest) (highest) n Yearly weight change (kg) Age and sex adjusted 4 0 (ref) (20.083, 0.005) (20.122, ) 0.04 Multivariable adjusted 4,5 0 (ref) (20.072, 0.016) (20.111, ) 0.02 MAI-Alberti-Fidanza Scores 0.95 (lowest) to to (highest) n Yearly weight change (kg) Age and sex adjusted 4 0 (ref) (20.089, 0.016) (20.068, 0.037) (20.137, ) 0.01 Multivariable adjusted 4,5 0 (ref) (20.079, 0.025) (20.057, 0.048) (20.131, ) 0.02 MDQI-Scali 6 Scores 11 (lowest) (highest) n Yearly weight change (kg) Age and sex adjusted 4 0 (ref) (20.097, 0.053) (20.149, 0.004) (20.212, ),0.01 Multivariable adjusted 4,5 0 (ref) (20.075, 0.075) (20.117, 0.037) (20.194, ),0.01 MDP-Sánchez-Villegas Scores (lowest) (highest) n Yearly weight change (kg) Age and sex adjusted 4 0 (ref) (20.064, 0.041) (20.102, 0.005) (20.130, ),0.01 Multivariable adjusted 4,5 0 (ref) (20.052, 0.052) (20.083, 0.024) (20.116, ) 0.02 DS-Panagiotakos Scores 29 (lowest) (highest) n Yearly weight change (kg) Age and sex adjusted 4 0 (ref) (20.049, 0.037) (20.085, 0.013) 0.18 Multivariable adjusted 4,5 0 (ref) (20.043, 0.044) (20.079, 0.021) 0.30 MSDPS-Rumawas Scores 20 (lowest) (highest) n Yearly weight change (kg) Age and sex adjusted 4 0 (ref) (20.081, 0.023) (20.117, 0.015) 0.13 Multivariable adjusted 4,5 0 (ref) (20.063, 0.041) (20.094, 0.038) MDS-Trichopoulou, Mediterranean dietary score proposed by Trichopoulou et al (18); ref, reference; MAI-Alberti-Fidanza, Mediterranean adequacy index proposed by Alberti-Fidanza et al (19); MDQI-Scali, Mediterranean diet quality index proposed by Scali et al (20); MDP-Sánchez-Villegas; Mediterranean dietary pattern proposed by Sánchez-Villegas et al (21); DS-Panagiotakos, diet score proposed by Panagiotakos et al (22); MSDPS-Rumawas, Mediterranean-style dietary pattern score proposed by Rumawas et al (23). Multivariable linear regression models were used for analyses. 2 Ref category. 3 Values are means. 4 Values are means; 95% CIs in parentheses. 5 Adjusted for age, sex, baseline BMI, physical activity, sedentary behavior, smoking (never, former, and current smokers), snacking, and total energy intake. 6 Highest score in the MDQI-Scali means the lowest adherence to the Mediterranean dietary pattern. group with the lowest adherence (3 points) to the MDS- Trichopoulou (adjusted mean difference between extreme categories: kg/y; 95% CI: , kg/y). This inverse association remained statistically significant for 3 of the other Mediterranean scores [the Mediterranean adequacy index of Alberti-Fidanza et al (19), Mediterranean diet quality index of Scali et al (20), and the Mediterranean dietary pattern proposed ½AQ9Š by Sánchez-Villegas et al (21)]. A higher adherence to the last 2 scores [the diet score proposed by Panagiotakos et al (22) and the Mediterranean-style dietary pattern score proposed by Rumawas et al (23)] was also associated with lower weight gain, but the 95% CIs were wide, and the inverse linear trend was not significant. Analyses of yearly weight changes (in kg) according to adherence to MDS-Trichopoulou stratified by sex and weight gain 3 kg before recruitment (5 y) are displayed in Table 3. An interaction was apparent (P, 0.01) between weight gain (3kg) in the previous 5 y before entering the cohort and adherence to the MDS-Trichopoulou as the association was stronger in those who had gained 3 kg in the 5 y before recruitment. We also observed a stronger association in men than in women, but the interaction was only marginally significant (P = 0.06).

6 MEDITERRANEAN DIET AND WEIGHT CHANGE 5 TABLE 3 Yearly weight change according to categories of adherence to the Mediterranean dietary score of Trichopoulou et al (18) (MDS-Trichopoulou) stratified ½AQ20Š according to sex, baseline age, BMI, previous weight gain (5 y), physical activity, and fiber intake in the Seguimiento Universidad de Navarra project, MDS-Trichopoulou We did not find any evidence of a significantly different association across strata of age (,40 or 40 y), BMI (,25 or ½AQ9Š 25), physical activity (,19 or 19 MET/wk), or fiber intake (,25 or 25 g/d) (data not shown). Risk of experiencing an absolute weight gain 3 or5 kg during the first 2 y of follow-up (Table 4) was lowest in participants with the highest adherence (6 points) to the MDS- Trichopoulou with an OR of 0.80 (95% CI: 0.70, 0.92) for 3 kg and an OR of 0.76 (95% CI: 0.62, 0.92) for 5 kg after multivariable adjustment than in participants with the lowest adherence (3 points) to the MDS-Trichopoulou. These values remained similar when we took into account the first 4 y of follow-up instead of the first 2 y of follow-up (Table 4). We did not find any association between adherence to the MDS-Trichopoulou and the incidence of overweight or obesity (BMI 25) in participants who initially were of normal weight (BMI,25) (Table 5). A secondary analysis stratified by weight gain 3 kg before recruitment (5 y) is also displayed in Table 5. We showed no significant interaction between previous weight gain and adherence to the Mediterranean diet on risk of overweight or obesity in this analysis. Finally, when we assessed weight change during follow-up as percentage of change from baseline weight, the multivariable adjusted estimates where 1.21% (95% CI: 0.32%, 2.11%) in the group with the lowest adherence to the MDS-Trichopoulou (3 points), 1.16% (95% CI: 0.27%, 2.06%) in the group with moderate adherence to the MDS-Trichopoulou (4 5 points), and 1.10% (95% CI: 0.20%, 1.99%) in the group with the highest adherence to the MDS (lowest) (highest) P for interaction Baseline (n) Women n Yearly weight change (kg) Multivariable adjusted 4,5 0 (ref) (20.101, 0.018) (20.109, 0.031) Men 0.06 n Yearly weight change (kg) Multivariable adjusted 4,5 0 (ref) (20.072, 0.059) (20.137, 0.015) Weight gain,3 kg before recruitment (5 y) n Yearly weight change (kg) Multivariable adjusted 5,6 0 (ref) (20.064, 0.037) (20.080, 0.037) Weight gain 3 kg before recruitment (5 y),0.01 n Yearly weight change (kg) Multivariable adjusted 5,6 0 (ref) (20.167, 0.007) (20.267, ) 1 ref, reference. Multivariable linear regression models were used for analyses. 2 Ref category. 3 Values are means. 4 ½AQ21Š Adjusted for age, baseline BMI, physical activity, sedentary behavior, smoking (never, former, and current smokers), snacking, and total energy intake. 5 Values are means; 95% CIs in parentheses. 6 Adjusted for sex, age, baseline BMI, physical activity, sedentary behavior, smoking (never, former, and current smokers), snacking, and total energy intake. Trichopoulou (6 points), with P for trend = (data not shown). DISCUSSION The current analyses from the SUN project conducted in a highly educated Mediterranean population with considerably lower risk of overweight or obesity than most American cohorts suggested that the Mediterranean dietary pattern may have a beneficial effect in slowing down the weight gain usually observed with age. Three randomized clinical trials have assessed the effect of dietary interventions by using a Mediterranean dietary pattern and weight change. The DIRECT study included 322 moderately ½AQ10Š obese volunteers (mean BMI of 31) who were randomly assigned to 3 different dietary recommendations and assessed weight change after 24-mo follow-up. The study showed that the recommendation of a low-calorie Mediterranean diet was more effective for weight loss than the recommendation of a lowcalorie, low-fat diet ( kg for the Mediterranean diet compared with kg for the low-fat diet) (7). The Medi-RIVAGE study included 212 participants who were randomly assigned to a low-fat diet or a Mediterranean diet and ½AQ11Š showed no significant weight differences in both groups after 3-mo follow-up; however, the duration of the trial was probably too short (27). The percentage of dropouts was 16% in the Mediterranean diet group and 36% in the low-fat diet group. Finally, McManus et al (28) randomly assigned 101 volunteers to either a low-fat diet or a Mediterranean diet for 18 mo. The authors

7 6 BEUNZA ET AL TABLE 4 Incident weight gain (3 or5 kg) during the first 2 or 4 y of follow-up according to categories of adherence to the Mediterranean dietary score of Trichopoulou et al (18) (MDS-Trichopoulou) in the Seguimiento Universidad de Navarra project, MDS-Trichopoulou First 2 y of follow-up n Weight gain 3 kg Incident cases Age and sex adjusted 3 0 (ref) 1.03 (0.92, 1.15) 0.78 (0.68, 0.89) Multivariable adjusted 3,4 0 (ref) 1.04 (0.93, 1.16) 0.80 (0.70, 0.92) Weight gain 5 kg Incident cases Age and sex adjusted 3 0 (ref) 0.94 (0.81, 1.10) 0.72 (0.60, 0.87) Multivariable adjusted 3,4 0 (ref) 0.96 (0.82, 1.12) 0.76 (0.62, 0.92) First 4 y of follow-up n Weight gain 3 kg Incident cases Age and sex adjusted 3 0 (ref) 0.93 (0.84, 1.03) 0.77 (0.68, 0.87) Multivariable adjusted 3,4 0 (ref) 0.95 (0.85, 1.05) 0.80 (0.71, 0.91) Weight gain 5 kg Incident cases Age and sex adjusted 3 0 (ref) 0.88 (0.77, 1.01) 0.71 (0.60, 0.84) Multivariable adjusted 3,4 0 (ref) 0.90 (0.78, 1.03) 0.76 (0.64, 0.90) 1 ref, reference. Multivariable logistic regression was used for analyses. 2 Ref category. 3 Values are odds ratios; 95% CIs in parentheses. 4 Adjusted for sex, age, baseline BMI, physical activity, sedentary behavior, smoking (never, ex-smokers, current smokers), snacking, and total energy intake. showed an average weight loss of 4.1 kg in the Mediterranean diet group and a weight gain of 2.9 kg in the low-fat diet group. The percentage of dropouts was 46% in the Mediterranean diet group and 80% in the low-fat diet group. Out of 4 trials with different intensity intervention, 3 trials showed a significant reduction in average weight or BMI related to Mediterranean diet (29 31), but the other trial did not (5). No trial has described increased weight gain associated with a Mediterranean dietary pattern. Some observational cohorts have also assessed the relation between adherence to the Mediterranean diet and weight change or incident overweight or obesity. Results from the EPIC-Spain study showed that the OR for incident obesity in initially overweight participants in the highest category of adherence to ½AQ12Š a modified Mediterranean dietary score was 0.68 (95% CI: 0.53, 0.89) for men, and 0.69 (95% CI: 0.54, 0.89) for women compared with the category of the lowest adherence (8). However, the incidence of overweight in initially normal-weight subjects did not change according to baseline adherence to the Mediterranean diet (OR: 1.11; 95% CI: 0.81, 1.52 for men; OR: 0.99, 95% CI: 0.78, 1.25 for women). Woo et al (9) also showed no association between adherence to the MDS-Trichopoulou pattern (highest category compared with the lowest category) and BMI change (OR: 1.35; 95% CI: 0.94, 1.93) in an Asian population. A previous report in our cohort did not find any apparent association between adherence to the Mediterranean dietary pattern score and risk of overweight or obesity after 2.3-y followup, although the point estimate for the average weight gain ½AQ13Š tended to be lower in participants with a higher adherence to the Mediterranean dietary pattern (10). Finally, some cross-sectional studies reported an inverse association with mean body weight (32 34), whereas others reported a null association (35 37). Several mechanisms have been proposed to explain the inverse association between adherence to a Mediterranean dietary pattern and weight gain. The high content in dietary fiber usually present in the Mediterranean dietary pattern that leads to increased satiety, the low energy density of Mediterranean diets, and their low glycemic load (38) are likely to provide a better metabolic control together with lower total calorie intake than are other dietary patterns (eg, a low-fat diet). Finally, despite concerns about the relatively high fat content of a typical Mediterranean diet, most of this fat comes from vegetable sources, and recent trials and cohort studies have repeatedly shown that this property of the Mediterranean dietary pattern does not increase total calorie intake, probably because of the quality of fat in olive oil and nuts and its effect on satiety, in combination with other healthier food components (vegetables, fruit, and legumes). Participants with the highest adherence to the MDS-Trichopoulou in our cohort had ½AQ14Š a lower total fat intake than those who were less adherent. In any case, the amount of energy that came from fat was high in these participants in absolute terms (33%). This characteristic contributes to the high palatability of the Mediterranean dietary pattern, which ensures a high potential for compliance (38). The significant interaction of weight gain before enrollment on weight change observed in this study suggested that the beneficial effect of the Mediterranean diet might be stronger in subjects who are prone to age-related weight gain. The Mediterranean diet indexes are aimed to assess the adherence of individuals to a reallife traditional dietary pattern, which later turned out to be beneficial for health. Nevertheless, one of the assessed scores was mainly based on ideal dietary guidelines (22). Thus, this index may not fully capture adherence to the traditional Mediterranean diet but to what the developers of that score considered to be a beneficial dietary profile (including very low alcohol consumption). This might explain why the inverse association for this score was weaker and statistically nonsignificant. Limitations in our study are the self-reported nature of exposure and outcomes. Although previous validation studies showed acceptable correlations on the food-frequency questionnaire (13) and information on weight (24), some degree of misclassification is likely to exist. However, we would expect it to be nondifferential and, therefore, drive the association toward the null value. Therefore, we do not believe that misclassification might be an alternative explanation of the significant inverse association we identified. Even more, we showed that this inverse association remained apparent in most sensitivity analyses that we conducted by using alternative definitions of the Mediterranean diet. It might be thought that a potential weakness of our cohort is that it was completely made up of university graduates, and therefore, it was not representative of the general Spanish population. This feature may have affected the generalizability of the findings, but it could also have actually enhanced the validity of our study because the high level of education and homogeneity of the cohort reduced the potential confounding related to socioeconomic status. In addition, the high educational level of our participants allowed us to collect high-quality information with

8 MEDITERRANEAN DIET AND WEIGHT CHANGE 7 TABLE 5 Incident overweight or obesity [BMI (in kg/m 2 ) 25] according to categories of adherence to the Mediterranean dietary score of Trichopoulou et al (18) (MDS-Trichopoulou) both general and stratified by 3-kg weight gain in the 5 y previous to baseline in the Seguimiento Universidad de Navarra project, MDS-Trichopoulou questionnaires. The various adherence scores were calculated only at baseline. It is possible that dietary habits change over time, and thus, future studies with a repeated assessment of the food-frequency questionnaire during follow-up are needed to confirm our results. In conclusion, adherence to a Mediterranean dietary pattern was significantly associated with a reduced weight gain in Spanish university graduates with low absolute risk of becoming overweight or obese. This vegetable fat rich dietary pattern with a high palatability and potential for compliance can be recommended to slow down age-related weight gain. We thank participants of the SUN project for their continued cooperation and participation and all members of the SUN project group for their administrative, technical, and material support. ½AQ15Š The authors responsibilities were as follows J-JB and ET: developed the idea, performed analyses, and wrote and corrected the manuscript; FBH: assisted in the development of the first idea, supervised the writing of the manuscript in all its phases, substantially contributed to the editing of the manuscript for intellectual content, and approved the final version of the manuscript; MB-R and AS-V: assisted in the development of the study and approved the final version of the manuscript; MS-M: substantially contributed to the editing of the manuscript for intellectual content and approved the final version of the manuscript; JAM: assisted in the development of the initial idea, provided scientific support, and approved the final version of the manuscript; and MAM-G: obtained funding, created the SUN project, provided the first idea, supervised the statistical analyses, substantially contributed to the editing of the manuscript for intellectual content, and approved the final version of the manuscript. None of the authors declared a conflict of interest General Person-years 13,911 14, Incident cases Age and sex adjusted 3 1 (ref) 1.04 (0.91, 1.21) 0.93 (0.76, 1.10) Multivariable adjusted 3,5 1 (ref) 1.00 (0.87, 1.16) 0.90 (0.75, 1.06) Weight gain 3 kg before enrollment (5 y) Person-years 10,530 11, Incident cases Age and sex adjusted 3 1 (ref) 1.12 (0.94, 1.33) 0.88 (0.71, 1.09) Multivariable adjusted 3,5 1 (ref) 1.05 (0.88, 1.26) 0.85 (0.69, 1.06) Weight gain,3 kg before enrollment (5 y) Person-years Incident cases Age and sex adjusted 3 1 (ref) 0.98 (0.77, 1.24) 1.13 (0.86, 1.48) Multivariable adjusted 3,5 1 (ref) 0.92 (0.72, 1.17) 1.00 (0.75, 1.33) 1 ref, reference. Subjects with overweight or obesity at baseline were excluded. Cox regression was used for analyses. 2 Ref category. 3 Values are hazard ratios; 95% CIs in parentheses. 4 For trend. ½AQ22Š 5 Adjusted for sex, age, baseline BMI, physical activity, sedentary behavior, smoking (never, former, and current smokers), snacking, and total energy intake. 6 For interaction. REFERENCES 1. York DA, Rossner S, Caterson I, et al. Prevention Conference VII: Obesity, a worldwide epidemic related to heart disease and stroke: Group I: worldwide demographics of obesity. Circulation 2004;110: e Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity. JAMA 2005;293: Hu FB. Obesity epidemiology. New York, NY: Oxford University Press, Sofi F, Cesari F, Abbate R, Gensini GF, Casini A. Adherence to Mediterranean diet and health status: meta-analysis. BMJ 2008;337:a Martínez-González MA, García-López M, Bes-Rastrollo M, et al. Mediterranean diet and the incidence of cardiovascular disease: a Spanish cohort. Nutr Metab Cardiovasc Dis (Epub ahead of print 20 January 2010). 6. Martinez-Gonzalez MA, de la Fuente-Arrillaga C, Nunez-Cordoba JM, et al. Adherence to Mediterranean diet and risk of developing diabetes: prospective cohort study. BMJ 2008;336: Shai I, Schwarzfuchs D, Henkin Y, et al. Weight loss with a lowcarbohydrate, Mediterranean, or low-fat diet. N Engl J Med 2008;359: Mendez MA, Popkin BM, Jakszyn P, et al. Adherence to a Mediterranean diet is associated with reduced 3-year incidence of obesity. J Nutr 2006;136: Woo J, Cheung B, Ho S, Sham A, Lam TH. Influence of dietary pattern on the development of overweight in a Chinese population. Eur J Clin Nutr 2008;62: Sánchez-Villegas, A, Bes-Rastrollo M, Martinez-Gonzalez MA, Serra-Majem L. Adherence to a Mediterranean dietary pattern and weight gain in a follow-up study: the SUN cohort. Int J Obes (Lond) 2006;30: Segui-Gomez M, de la Fuente C, Vazquez Z, de Irala J, Martinez-Gonzalez MA. Cohort profile: the Seguimiento Universidad de Navarra (SUN) study. Int J Epidemiol 2006;35: Willett W. Nutritional epidemiology. 2nd ed. New York, NY: Oxford University Press, P

9 8 BEUNZA ET AL 13. Martin-Moreno JM, Boyle P, Gorgojo L, et al. Development and validation of a food frequency questionnaire in Spain. Int J Epidemiol 1993; 22: de la Fuente-Arrillaga C, Ruiz ZV, Bes-Rastrollo M, Sampson L, Martinez-Gonzá lez MA. Reproducibility of an FFQ validated in Spain. Public Health Nutr (Epub ahead of print 28 Janaury 2010). 15. Fernandez-Ballart JD, Pinol JL, Zazpe I, et al. Relative validity of a semi-quantitative food-frequency questionnaire in an elderly Mediterranean population of Spain. Br J Nutr 2010;103: Mataix-Verdú J, Mañas-Almendros M. Tabla de composición de alimentos, 4th ed. [Food composition tables.] Granada, Spain: Universidad de Granada, 2003 (in Spanish). 17. Moreiras F, Carvajal A, Cabrera L. Tablas de composición de alimentos, 9th ed. [Food composition tables.] Madrid, Spain: Pirámide, 2005 (in Spanish). 18. Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to a Mediterranean diet and survival in a Greek population. N Engl J Med 2003;348: Alberti-Fidanza A, Fidanza F. Mediterranean adequacy index of Italian diets. Public Health Nutr 2004;7: Scali J, Richard A, Gerber M. Diet profiles in a population sample from Mediterranean southern France. Public Health Nutr 2001;4: Sánchez-Villegas A, Martinez JA, De Irala J, Martinez-Gonzalez MA. Determinants of the adherence to an a priori defined Mediterranean dietary pattern. Eur J Nutr 2002;41: Panagiotakos DB, Pitsavos C, Stefanadis C. Dietary patterns: a Mediterranean diet score and its relation to clinical and biological markers of cardiovascular disease risk. Nutr Metab Cardiovasc Dis 2006;16: Rumawas ME, Dwyer JT, McKeown NM, Meigs JB, Rogers G, Jacques PF. The development of the Mediterranean-style dietary pattern score and its application to the American diet in the Framingham Offspring Cohort. J Nutr 2009;139: Bes-Rastrollo M, Perez Valdivieso J, Sánchez-Villegas A, Alonso A, Martinez-Gonzalez M. Validation of the self-reported weight and body mass index of the participants in a cohort of university graduates. Rev Esp Obes 2005;3: Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc 2000;32:S Martinez-Gonzalez MA, Lopez-Fontana C, Varo JJ, Sánchez-Villegas A, Martinez JA. Validation of the Spanish version of the physical activity questionnaire used in the Nurses Health Study and the Health Professionals Follow-up Study. Public Health Nutr 2005;8: Vincent-Baudry S, Defoort C, Gerber M, et al. The Medi-RIVAGE study: reduction of cardiovascular disease risk factors after a 3-mo intervention with a Mediterranean-type diet or a low-fat diet. Am J Clin Nutr 2005;82: McManus K, Antinoro L, Sacks F. A randomized controlled trial of a moderate-fat, low-energy diet compared with a low fat, low-energy diet for weight loss in overweight adults. Int J Obes Relat Metab Disord 2001;25: Esposito K, Marfella R, Ciotola M, et al. Effect of a mediterranean-style diet on endothelial dysfunction and markers of vascular inflammation in the metabolic syndrome: a randomized trial. JAMA 2004;292: Esposito K, Pontillo A, Di Palo C, et al. Effect of weight loss and lifestyle changes on vascular inflammatory markers in obese women: a randomized trial. JAMA 2003;289: Toobert DJ, Glasgow RE, Strycker LA, et al. Biologic and quality-of-life outcomes from the Mediterranean Lifestyle Program: a randomized clinical trial. Diabetes Care 2003;26: Panagiotakos DB, Chrysohoou C, Pitsavos C, Stefanadis C. Association between the prevalence of obesity and adherence to the Mediterranean diet: the ATTICA study. Nutrition 2006;22: Schroder H, Marrugat J, Vila J, Covas MI, Elosua R. Adherence to the traditional mediterranean diet is inversely associated with body mass index and obesity in a spanish population. J Nutr 2004;134: Shubair MM, McColl RS, Hanning RM. Mediterranean dietary components and body mass index in adults: the peel nutrition and heart health survey. Chronic Dis Can 2005;26: Romaguera D, Norat T, Mouw T, et al. Adherence to the Mediterranean diet is associated with lower abdominal adiposity in European men and women. J Nutr 2009;139: Rossi M, Negri E, Bosetti C, et al. Mediterranean diet in relation to body mass index and waist-to-hip ratio. Public Health Nutr 2008;11: Trichopoulou A, Naska A, Orfanos P, Trichopoulos D. Mediterranean diet in relation to body mass index and waist-to-hip ratio: the Greek European Prospective Investigation into Cancer and Nutrition Study. Am J Clin Nutr 2005;82: Buckland G, Bach A, Serra-Majem L. Obesity and the Mediterranean diet: a systematic review of observational and intervention studies. Obes Rev 2008;9:

10 MEDITERRANEAN DIET AND WEIGHT CHANGE 9 APPENDIX A Indexes that evaluate the adherence to the Mediterranean dietary pattern 1 MDS-Trichopoulou MAI-Alberti-Fidanza MDQI-Scali MDP-Sánchez-Villegas DS-Panagiotakos MSDPS-Rumawas Positive items Vegetables (g/d) Typical Mediterranean foods: bread, cereal, legumes, potatoes, vegetables, fruit, legumes, fish, red wine, and vegetable oils (percentage of total energy) Vegetables and fruit (g/d) ncluding all types of red, yellow, and green vegetables, both cooked and raw, and all types of fresh fruit Vegetables (g/d) Vegetables (times/d or mo) Potatoes (times/d or mo) Vegetables: 6 servings/d Potatoes and other starchy roots: 3 servings/wk Fruit and nuts (g/d) Fruit (g/d) Fruit (times/d or mo) Fruit: 3 servings/d Legumes (g/d) Legumes (g/d) Legumes (times/d or mo) Legumes, olives, and nuts: 4 servings/wk Cereals, bread, and Cereals (g), including bread Cereals (g/d), including Nonrefined cereals potatoes (g/d) (wheat, barley, plain, bread and potatoes (times/d or mo), and whole grain), pasta including whole (plain and whole grain), bread, pasta, rice, rice (plain and whole other grain, and grain), and breakfast biscuits cereals Ratio of monounsaturated Olive oil (ml) Ratio of monounsaturated Olive oil (times/d fatty acids to saturated fatty acids to saturated or mo) fatty acids fatty acids Moderate alcohol (g/d) Moderate alcohol (g/d) Alcohol (times/d or mo). (10 50 g alcohol/d for (30 men; 20 women) For alcoholic men; 5 25 g alcohol/d beverages, 5 points for women) were scored for consumption of,300 ml alcohol/d, 0 points for.700 ml alcohol/d, and a gradient in the middle. Fish Fish (g), including white and Fish (times/d or mo) fatty fish Poultry (times/d or mo) Negative items Meat and poultry (g/d) Nontypical Mediterranean foods: milk, cheese, meat, eggs, animal fat and margarines, sweet beverages, pastries, cookies, and sugar (percentage of total energy) Meat (g), including processed and fresh-cut beef, veal, mutton, lamb and pork Meat (g/d), including meat products ½AQ16Š Dairy products Milk and dairy products (g/d) Meat and meat products (times/d or mo) Full-fat dairy products (times/d or mo), including cheese, yogurt, and milk Whole grains: 8 servings/d Sweets: 3 servings/wk Dairy products: 2 servings/d Eggs: 3 servings/wk Olive oil as the only culinary fat 3 glasses of wine/d for men; 1.5 glasses of wine/d for women Fish: 6 servings/wk Poultry: 4 servings/wk Meat: 1 servings/wk (Continued)

11 10 BEUNZA ET AL Appendix A (Continued) MDS-Trichopoulou MAI-Alberti-Fidanza MDQI-Scali MDP-Sánchez-Villegas DS-Panagiotakos MSDPS-Rumawas Saturated fatty acids trans Fatty acids (g/d) (percentage of energy) Cholesterol (mg) Formula Adding all points from Division of the sum of the Each item was categorized Energy-adjusted intakes Ratings were produced Ratings were produced for positive items (higher percentage of total in tertiles and scored for all items except for (from 0 to 5 in positive each item (from 0 to 10) than sex- specific energy from typical from 1 to 3. Higher alcohol intake. For items and the reverse in according to its closeness median value) and Mediterranean food consumption of positive moderate alcohol negative items) to the goals negative items groups by the sum of the items scored 1, medium consumption, a according to their (Mediterranean (lower than percentage of total consumption of positive transformation was position in the pyramid). Further sex-specific energy from nontypical items scored 2, and used to obtain the Mediterranean diet adjustment was median value) Mediterranean food lowest consumption of highest value for 30 g pyramid used for issuing added according to the groups positive items scored 3; alcohol/d (for men) dietary guidelines for an proportion of total energy highest consumption of or 20 g alcohol/d (for ideal diet (from daily intake provided by the negative items scored 3, women) and consumption to rare or consumption of medium consumption of progressive lower monthly consumption). Mediterranean foods. negative items scored 2, values as the and lowest consumption consumption was lower of negative items or higher than these scored 1. The 7 scores values. All items were were summed up. The standardized as z scores ½AQ17Š lower the MDQI-Scali, and then summed the higher the adherence (subtracting negative to the Mediterranean diet. items) and transformed into percentages of adherence. ½AQ18Š Range: 0 9 Range: 0 14 Range: Range: 0 55 Theoretical range: MDS-Trichopoulou, Mediterranean dietary score proposed by Trichopoulou et al (18); MAI-Alberti-Fidanza, Mediterranean adequacy index proposed by Alberti-Fidanza et al (19); MDQI-Scali, Mediterranean diet quality index proposed by Scali et al (20); MDP-Sánchez-Villegas, Mediterranean dietary pattern proposed by Sánchez-Villegas et al (21); DS-Panagiotakos, diet score proposed by Panagiotakos et al (22); MSDPS-Rumawas, Mediterranean-style dietary pattern score proposed by Rumawas et al (23).