Increase Font Size Decrease Font Size View as PDF Print

What is the relationship between adherence to dietary guidelines/recommendations or specific dietary patterns, assessed using an index or score, and measures of body weight or obesity?

Conclusion

There is moderate evidence that, in adults, increased adherence to dietary patterns scoring high in fruits, vegetables, whole grains, legumes, unsaturated oils, and fish; low in total meat, saturated fat, cholesterol, sugar-sweetened foods and drinks, and sodium; and moderate in dairy products and alcohol is associated with more favorable outcomes related to body weight or risk of obesity, with some reports of variation based on gender, race, or body weight status.
 

Grade

II - Moderate

 

Key Findings:

  • Two major categories of dietary pattern scores were identified in the literature: (1) studies that examined exposure based on a Mediterranean dietary pattern and (2) studies that examined exposure based on dietary guidelines recommendations.
  • In adults, adherence to a Mediterranean diet score or a dietary guidelines-related score is associated with decreased risk of obesity, with some reported variation based on gender or body weight status.
  • This protective association in adults is further supported by consistent evidence indicating that an increased Mediterranean diet score or dietary guidelines-related score is associated with decreased body weight, BMI, waist circumference, or percent body fat, with some variation based on gender and race.

 
Evidence Summary Overview

Description of the Evidence

A total of 14 studies met the inclusion criteria for this systematic review and were categorized based on dietary pattern exposure. Two major categories were identified (appendix A): (1) studies that examined exposure based on a Mediterranean dietary pattern and (2) studies that examined exposure based on dietary guidelines recommendations. Taken together, there were six studies on Mediterranean diet scores (Beunza, 2010; Estruch, 2006; Mendez, 2006; Romaguera, 2010; Rumawas, 2009; Tortosa, 2007), five studies on dietary guidelines-based indices (Berz, 2011; Cheng, 2010; Gao, 2008; Kesse-Guyot, 2009; Zamora, 2010), two studies on Mediterranean scores and dietary guidelines indices (Lassale, 2012; Woo, 2008) and one study that used a trial-based customized score (Jacobs, 2009). Two of the studies were RCTs of positive quality (Estruch, 2006; Jacobs, 2009). Twelve of the studies were prospective cohort studies; of these, ten were of positive quality (Berz, 2011; Beunza, 2010; Cheng, 2010; Gao, 2008; Kesse-Guyot, 2009; Lassale, 2012; Mendez, 2006; Romaguera, 2010; Rumawas, 2009; Zamora, 2010) and two were of neutral quality (Tortosa, 2007; Woo, 2008). The studies were carried out between 2006 and 2012. The sample sizes for the RCTs were from 187 to 769 subjects. The sample sizes for prospective cohort studies ranged from 732 to 373,803 participants (2 studies <1,000, 7 studies >1,000, 2 studies >10,000, and 1 study >100,000). RCT duration ranged from 3 to 12 months and observational study follow-up times from 1.5 to 20 years.
 
Studies were conducted in the United States, Hong Kong, and Europe, including the ten European countries (Denmark, France, Germany, Greece, Italy, the Netherlands, Norway, Spain, Sweden, and the United Kingdom) of the EPIC-PANACEA project. Of the six studies on the Mediterranean dietary pattern alone, four studies were conducted in Spain (Beunza, 2010; Estruch, 2006; Mendez, 2006; Tortosa, 2007), one study was the European multicenter study that was part of EPIC-PANACEA (Romaguera, 2010), and one study was conducted in the United States (Rumawas, 2009). Of the six studies conducted on dietary guidelines-based indices, three studies were conducted in the United States with U.S.-based indices (Berz, 2011; Gao, 2008; Zamora, 2010), one study was conducted in Germany with an index developed in the United States (Cheng, 2010), and one study was conducted in France using a French index (Kesse-Guyot, 2009). The studies that compared Mediterranean diet scores and dietary-guidelines based indices were conducted in France (Lassale, 2012) and Hong Kong (Woo, 2008), and the one study that used a trial-based customized diet score was conducted in Norway (Jacobs, 2009).
 
Ten out of twelve of the prospective cohort studies were conducted with generally healthy adults with a mean age of 25 to 63 years; however, two studies were conducted with children and adolescents (one with girls) (Berz, 2011; Cheng, 2010). The two RCTs were conducted in adults with elevated chronic disease risk: one study with a Mediterranean diet intervention on older adults at increased CVD risk with >90 percent overweight or obese (Estruch, 2006), and one study using an a priori diet intervention on men with metabolic syndrome (Jacobs, 2009). Studies varied in baseline weight status and ranged from <10 percent to >90 percent of subjects being overweight or obese, with the mid-range between 20 to 40 percent overweight or obese. Lastly, one of the studies was focused on Black, Caucasian, Hispanic, and Chinese participants (Gao, 2008), and one study was focused on Black and Caucasian young adults (Zamora, 2010); both of these studies examined ethnic/racial-specific differences in outcomes.
 

Evidence Summary Paragraphs

Mediterranean Dietary Pattern

1. Estruch et al., 2006 (positive quality) conducted a parallel, multicenter randomized controlled trial (RCT) in Spain, the Prevencion con Dieta Mediterranea (PREDIMED) study, to assess the effects of a Mediterranean diet on primary prevention of cardiovascular disease in a high-risk group of men and women. Subjects either had type 2 diabetes or three cardiovascular disease risk factors and 90 percent were overweight or obese. The PREDIMED trial assigned participants to three interventions: (1) Mediterranean diet with olive oil, (2) Mediterranean diet with mixed nuts, and (3) low-fat diet. This was the first report at 3 months of a 4-year clinical trial. The trial included 769 subjects with average age of 69 years. Dietary intake was assessed with a validated 137-item FFQ and the degree of adherence was assessed with a 14-item Mediterranean diet score that is based on a version of the MDS that assesses a cardio-protective Mediterranean diet. The Mediterranean diet score increased in the two Mediterranean diet groups of the trial and remained unchanged in the low-fat group. There were no significant changes in body weight and adiposity within or between groups from baseline to the 3 months. The authors concluded that because the subjects had CVD risk factors, the fact that body weight did not increase was positive because there was no weight gain, even though the Mediterranean diet interventions involved ad libitum diets supplemented with unsaturated fats, such as those contained in olive oil and nuts.

2. Beunza et al., 2010 (positive quality) reported on a prospective cohort study in Spain, the Seguimiento Universidad de Navarra (SUN) study, to examine the association between adherence to a Mediterranean dietary pattern and long-term weight change and incidence of overweight or obesity. This analysis of the SUN cohort of university graduates included 10,376 participants with a mean age of 38±11 years who were followed for approximately 6 years. Dietary intake was assessed with a semi-quantitative 136-item FFQ validated in Spain and adherence to a Mediterranean dietary pattern was assessed using the Trichopoulou MDS (2003). Dietary intake was assessed biennially. Subject exposure was assessed in tertiles of low (0-3), medium (4-6), and high (7-9) adherence to the MDS. Participants with highest adherence to the MDS had lower average yearly weight gain, -0.059 kg/y (95% CI = -0.111 to -0.008 kg/y; P for trend = 0.02), than participants in the lowest adherence group. However, the MDS was not associated with incidence of overweight or obesity in participants who were normal weight at baseline. The authors concluded that adherence to the Mediterranean dietary pattern was significantly associated with reduced weight gain. Further, the authors interpret their results in a highly educated Mediterranean population (i.e., low risk) to indicate that this dietary pattern could be recommended to slow age-related weight gain.
 
3. Lassale et al., 2012 (positive quality) conducted on a prospective cohort study in France to examine the relationship between diet quality and development of obesity by comparing the predictive value of six different dietary scores on weight change and risk of obesity. Subjects were participants in the SUpplementation en VItamines et Minereaux AntioXydants (SU.VI.MAX) study. This analysis included 3,151 adults, aged 45 to 60 years, followed for 13 years. Dietary intake was assessed with 24-hour diet records collected every 2 months and records for the first 2 years of the study were used to determine baseline dietary habits. Diet quality was assessed using the MDS, rMED, MSDPS, the Diet Quality Index-International (DQI-I), the 2005 Dietary Guidelines for Americans Adherence Index (DGAI), and the French Programme National Nutrition Sante-Guidelines Score (PNNS-GS). Overall, better adherence to a Mediterranean diet (except for the MSDPS) or dietary guidelines was associated with lower weight gain in men who were normal weight at baseline (P for trend = <0.05). In addition, among the 1,569 non-obese men at baseline, the odds of becoming obese associated with one standard deviation increase in dietary score ranged from OR = 0.63 (95% CI = 0.51 - 0.78) for the DGAI to OR = 0.72 (95% CI = 0.59 - 0.88) for the MDS, only the MSDPS was non-significant. In women, no association between diet scores and weight gain or incidence of obesity was found. The authors concluded that adherence to a Mediterranean diet or dietary guidelines, except MSDPS score, were associated with lower 13-year weight gain and lower obesity risk for men, while no associations were observed for women.
 
4. Mendez et al., 2006 (positive quality) reported on a prospective cohort study in Spain, the European Prospective Investigation into Cancer and Nutrition (EPIC)-Spain study, to examine if adherence to a Mediterranean diet pattern was associated with reduced incidence of obesity. Analysis of the EPIC-Spain cohort included 27,827 participants, age range 29 to 69 years that were followed for approximately 3 years. Dietary intake was assessed with a validated, computerized diet-history instrument with >600 items. Mediterranean dietary pattern adherence was assessed using a slight modification of the Trichopoulou MDS (2003), with exposure expressed in tertiles of low (0-3), medium (4-5), and high (6-8) adherence. Participants with highest MDS adherence had reduced incidence of obesity when overweight at baseline. Women were 27 percent and men 29 percent less likely to become obese (women OR = 0.69 [95% CI = 0.54 - 0.89]; men OR = 0.68 [95% CI = 0.53 - 0.89]). High MDS adherence was not associated with incidence of overweight in subjects who were normal weight at baseline. When individual MDS score components were assessed, obesity incidence was higher in women who consumed more meat and lower in men who consumed more cereals (P<0.05). The authors concluded that promoting eating habits consistent with a Mediterranean dietary pattern may be useful in efforts to combat obesity.
 
5. Romaguera et al., 2010 (positive quality) reported on a multicenter prospective cohort study conducted in ten countries across Europe, the EPIC-Physical Activity, Nutrition, Alcohol Consumption, Cessation of Smoking, Eating out of Home, and Obesity  (EPIC-PANACEA) study. This study examined the association between adherence to a Mediterranean dietary pattern, prospective weight change, and the incidence of overweight or obesity. Analysis of the EPIC-PANACEA cohort included 373,803 participants, age range 25 to 70 years, followed for 5 years. Dietary intake was assessed with country-specific FFQs and adherence was assessed using the relative Mediterranean Diet Score (rMED), a variation on the Trichopoulou MDS (2003), that assessed 9 components in g/1000 kcal for energy density. Subject exposure was assessed in tertiles of low (0-6), medium (7-10), and high (11-18) adherence to the rMED. Participants with high rMED adherence gained less weight in 5 years than did participants with low rMED adherence (-0.16 kg [95% CI: -0.24 to -0.07 kg]). The contribution of each rMED scoring component was also assessed and it was found that the association between rMED and weight change was no longer significant when meat and meat products were not part of the score. The likelihood of becoming overweight or obese in participants with high rMED adherence was OR = 0.90 (95% CI = 0.82 - 0.96) when compared to the low adherence group. Lastly, a meta-analysis of the OR scores of all 10 European countries showed that a 2-point increase in rMED score was associated with 3 percent (95% CI = 1 - 5%) lower odds of becoming overweight or obese over 5 years. The authors concluded that promoting the Mediterranean dietary pattern as a model of healthy eating may help prevent weight gain and the development of obesity.
 
6. Rumawas et al., 2009 (positive quality) conducted a prospective cohort study using a subset of the Framingham Offspring and Spouse (FOS) study, which was conducted in the United States. The study examined the association between adherence to a Mediterranean dietary pattern and metabolic syndrome, including abdominal adiposity. The study analyzed 2,730 participants with a mean age of 54 years for metabolic syndrome traits including waist circumference, and subgroup of 1,918 participants for incidence of metabolic syndrome, over a 7-year period. Dietary intake was assessed using the Harvard semi-quantitative FFQ of 126 items and adherence was assessed using the Mediterranean-style dietary pattern score (MSDPS) based on the Mediterranean diet pyramid. The MSDPS has 13 components scored 0 to 10 based on servings per day or servings per week. Subject exposure was assessed in quintiles of low to high adherence to the MSDPS. Participants with a higher MSDPS had significantly lower waist circumference (P for trend < 0.001). The authors concluded that their study suggests that consumption of a diet consistent with the principles of a Mediterranean-style diet may protect against metabolic syndrome in Americans.
 
7. Tortosa et al., 2007 (neutral quality) reported on a prospective cohort study in Spain, the Seguimiento Universidad de Navarra (SUN) study, to examine the association between adherence to a Mediterranean dietary pattern and metabolic syndrome, including abdominal adiposity. This analysis of the SUN cohort of university graduates included 2,563 participants initially free of metabolic syndrome (mean ages not reported) who were followed for 6 years. Dietary intake was assessed with a semi-quantitative 136-item FFQ validated in Spain and adherence to a Mediterranean dietary pattern was assessed using the MDS of Trichopoulou. Dietary intake was assessed biennially. Subject exposure was assessed in tertiles of low (0-2), medium (3-5), and high (6-9) adherence to the MDS. Participants in the highest tertile of adherence to the MDS had lower waist circumference, -0.05 cm over 6 years (P for trend = 0.038), compared to the lowest tertile. The authors concluded that this study provided evidence of an inverse relationship between MDS adherence and cumulative incidence of metabolic syndrome.
 
8. Woo et al., 2008 (neutral quality) reported on a prospective cohort study in Hong Kong to examine dietary factors that predispose people to overweight or obesity. The study analyzed 732 participants with a mean age of 45 years over a period of 5 to 9 years. Dietary intake was assessed using a validated FFQ and adherence to the Mediterranean diet was assessed using the Trichopoulou MDS (2003). In addition, food variety was assessed by the ratio of variety of snacks to the variety of grains and meat. Diet quality was also assessed using the Diet Quality Index International (DQI-I). Incidence of overweight or obesity was calculated by dividing the number of subjects with normal BMI at baseline who became overweight or obese (N - >Ow/Ob) by the number with normal BMI at baseline. In multivariate analysis, increased adherence to either the MDS or DQI-I was associated with a slight, but not significant, increase in the risk of becoming overweight (OR =1.35 [95% CI = 0.94 - 1.93] and OR = 1.32 [95% CI = 0.92 - 1.89, respectively) when defined by the Asian criteria (BMI >23 kg/m2). Increased snack and food variety ratios were associated with increased risk of becoming overweight when defined by the Asian criteria (BMI >23 kg/m2), however, there was no association when overweight defined as BMI >25 kg/m2 was used. This study showed no association between Mediterranean diet adherence, or diet quality assessed by DQI-I, and the development of overweight. The authors concluded that increased variety of snack consumption may predispose to weight gain over a 5- to 9-year period.

Dietary Guidelines-Related Patterns

Note: Lassale 2012 et al. (2012) and Woo et al. (2008) (described above) conducted studies that assessed both Mediterranean diet pattern scores and dietary-guidelines related indices.
 
9. Berz et al., 2011 (positive quality) reported on a prospective cohort study to assess the effects of the Dietary Approaches to Stop Hypertension (DASH) eating pattern on BMI in adolescent females in the United States. The Prospective National Growth and Health Study followed 2,327 girls aged 9 to 10 years over a 10-year period. Dietary intake was assessed using a 3-day food record collected for each year of the study. Diet quality was assessed using a modified DASH food group score, reflecting adherence to a DASH eating pattern as described in the 2005 Dietary Guidelines for Americans. The score contained 10 food groups, three of which were excluded for the modified score used in this study: added sugars, discretionary fats and oils, and alcohol. The seven DASH-related groups in the modified score included fruits, vegetables, low-fat dairy, total and whole grains, lean meats, and nuts, seeds, and legumes. Subject exposure was assessed as quintiles of DASH score. Overall, girls in the highest vs. lowest quintile of DASH score had an adjusted mean BMI of 24.4 vs. 26.3 kg/m2 (P<0.05). The strongest individual food component predictors of BMI were total fruit and low-fat dairy. Whole-grain consumption was more weakly, but inversely, associated with BMI. The authors concluded that adolescent girls whose diet more closely resembled the DASH eating pattern had smaller gains in BMI over the 10-year period. This suggests that this eating pattern could help prevent excess weight gain during adolescence.
 
10. Cheng et al., 2010 (positive quality) analyzed data from a prospective cohort study conducted in Germany, the Dortmund Nutritional and Anthropometric Longitudinally Designed (DONALD) study, to examine if the diet quality of healthy children prior to puberty was associated with body composition at onset of puberty. The final sample included 222 children, mean age 7.4 years, who were followed until the onset of pubertal growth spurt (Age at Take-Off = ATO). Dietary intake was assessed with 3-day weighed food records. Adherence to a diet pattern was assessed by the Nutritional Quality Index (NQI) or the Revised Children’s Diet Quality Index (RC-DQI). The NQI determine the extent to which a child meets the nutritional recommendation for particular nutrients. The RC-DQI rated diet quality by scoring childrens’ intakes in relation to the U.S. dietary intake recommendations. Thirteen dietary components were considered, including a proxy for energy balance. Subject exposure was assessed as high, medium, and low diet quality for both indices. Results showed that for both indices, a higher dietary quality was associated with a higher energy intake. Children with a lower diet quality according to their RC-DQI score had lower BMI and Fat Mass Index (FMI) Z-scores at baseline (P<0.01) but not at ATO. Lower NQI in pre-puberty was associated with a ~0.4 year-earlier ATO than children with a higher diet quality (P=0.02). The authors concluded that diet quality was not independently associated with body composition at ATO. Children with lower diet quality according to a nutrient density-based index appear to enter puberty at an earlier age, independent of pre-pubertal body composition.
 
11. Gao et al., 2008 (positive quality) reported on a prospective cohort study of White, African American, Hispanic, and Chinese men and women in the Multi-Ethnic Study of Atherosclerosis (MESA) in the United States. The objective of the study was to assess the diet quality of a multiethnic population using and comparing the HEI-1995 and a modified version, the HEI-05, in ability to predict obesity outcomes, as well as whether their predictive ability differed by ethnicity. A total of 6,236 subjects with mean age of 63 years were recruited from four ethnic populations and followed-up at 18 months. Dietary intake was assessed using a 120-item FFQ that included typical Hispanic and Chinese foods. Diet quality was determined using the HEI-1995 and the HEI-05, a modified HEI developed by the authors and based on the 2005 DGA, but distinct from HEI-2005. The HEI-05 used the same components, weighting, and scoring rules as the original HEI, but further adjusted the food group components to incorporate levels of caloric need based on gender, age, and activity level, as specified in the 2005 DGA. For the overall population, there was an inverse association between quintiles of each HEI score and BMI and waist circumference (WC) (P<0.001). The risk of obesity in normal weight participants was inversely associated with HEI scores only for Whites (P<0.05). A comparison of the HEI-1995 and HEI-05 scores indicated that beta-coefficients, as predictors of body weight and BMI, were higher for the HEI-05 scores in Whites. The authors concluded that their changes to the original HEI resulted in an HEI that was better at predicting obesity outcomes in a multi-ethnic population, although the improvement was primarily seen in the White population. The authors qualified this conclusion in the Discussion and stated that the HEI-05 was only slightly better than the original HEI in predicting both cross-sectional baseline and follow-up obesity outcomes.
 
12. Kesse-Guyot et al., 2009 (positive quality) conducted a prospective cohort study in France to examine the association between adherence to a dietary score based on nutritional guidelines and changes in body weight, body fat distribution, and obesity risk. Subjects were participants in the SUpplementation en VItamines et Minereaux AntioXydants (SU.VI.MAX) study and the score was the Programme National Nutrition Sante´ guidelines score (PNNS-GS), based on assessing adherence to 2001 French national guidelines for the general population. This study followed a total of 3,531 adults, aged 45 to 60 years, for 6 years. Dietary intake was assessed using 24-hour diet records every 2 months, covering all days of the week and seasons. Adherence to dietary guidelines was assessed using the PNNS-GS that includes 12 nutritional components: fruit and vegetables, starchy foods, whole grains, dairy products, meat, seafood, added fat, vegetable fat, sweets, water and soda, alcohol, and salt. The last PNNS-GS component is physical activity. In fully adjusted models, an increase of one PNNS-GS unit was associated with lower weight gain (P=0.004), lower waist circumference gain (P=0.01), lower waist-to-hip ratio gain (P=0.02), and lower BMI gain (P=0.002). An increase of 1 PNNS-GS unit was associated with a lower probability of becoming overweight (including obese), OR = 0.93 (95% CI: 0.88 - 0.99). Similarly, an increase of 1 PNNS-GS unit was associated with a lower probability of becoming obese, OR = 0.89 (95% CI: 0.80 - 0.99). The authors concluded that their study showed that nutritional guidelines can play a role in the prevention of age-related weight increase and the development of obesity.
 
13. Zamora et al., 2010 (positive quality) analyzed data from the prospective cohort study, Coronary Artery Risk Development in Young Adults (CARDIA), which was conducted in the United States, to examine the association between diets consistent with the 2005 DGAs and subsequent weight gain in Black and White young adults. The CARDIA study followed 4,913 participants (average age 25 years) over 20 years. Dietary intake was assessed with a quantitative 100-item diet-history questionnaire. Diet quality was assessed using DQI-2005 to determine adherence to 10 components of the 2005 DGAs: total fat, saturated fat, cholesterol, added sugars, reduced-fat milk, fruit, vegetables, whole grains, nutrient-dense foods, and limited sodium and alcohol intake. The study results differed by race. A 10-point increase in DQI score was associated with a 10 percent lower risk of gaining 10 kg in normal-weight Whites, but with a 15 percent higher risk in obese Blacks (P<0.001). The mean adjusted 20-year weight change was +19.4 kg for Blacks and +11.2 kg for Whites with high DQI (DQI>70) and +17.8 kg for Blacks and +13.9 kg for Whites with a low DQI (DQI<50) (P<0.05). The authors concluded that a diet consistent with the 2005 DGAs was associated with more weight gain in Blacks, particularly obese Blacks, but with less weight gain in Whites. However, even Whites with a high DQI score gained weight over the 20-year study period. Therefore, they concluded that a diet consistent with the 2005 DGAs was not beneficial for long-term weight maintenance in young American adults.

Other Scores:

14. Jacobs et al., 2009 (positive quality) conducted a randomized controlled trial in Norway, the Oslo Diet and Exercise Study (ODES), to examine the effect of changes in diet patterns on body weight and other intermediate metabolic markers of chronic disease risk. The final sample included 187 men, mean age 45 years, who met the criteria for metabolic syndrome. Subjects were randomly assigned to: (1) the diet protocol, (2) the exercise protocol, (3) the diet + exercise protocol, or (4) the control protocol. The trial duration was 12 months. Dietary intake was assessed with a validated 180-item FFQ. An a priori diet score, developed by the authors, was used to determine adherence to the intervention. The diet score was based on summing tertile rankings of 35 food-group variables with postulated beneficial, neutral, or adverse effects on health. A higher score reflected the recommended dietary changes in the ODES trial. Over the year of the study, the diet score increased by 2±5.5 in both diet groups, with a decrease of an equivalent amount in the exercise and control groups. The weight change was -3.5 kg/10-point change in diet score (P<0.0001). Similarly, per 10-point change in diet score, waist circumference decreased by 2.8 cm (P<0.0001) and percent body fat decreased by 1.3 percent (P<0.0001). Additionally, when the results related to diet score were adjusted for intervention group these body size changes were attenuated, but still significant. Subjects with a higher diet score also had more favorable changes in other markers of metabolism. The authors concluded that a more favorable diet pattern was associated with improved body size and metabolic profile.
 

Qualitative Synthesis of the Collected Evidence

Themes and Key Findings for Total Scores and Indices

Intermediate Outcomes: Body Weight, BMI, Waist Circumference, Percent Body Fat

The intermediate outcomes included in this review were body weight, BMI, waist circumference, and percent body fat. Overall, seven studies reported on body weight, four reported on BMI, five reported on waist circumference, and one reported on percent body fat (table 4-A-I-1).
 
Body Weight: Six out of seven studies that examined body weight found that an increase in Mediterranean diet score or dietary guideline index was associated with improved body weight in adults. Based on total scores, four studies reported a protective association, including an association between the MDS (Beunza, 2010), rMED (Romaguera, 2010), PNNS-GS (Kesse-Guyot, 2009), and a customized a priori score (Jacobs, 2009) and body weight. Two studies reported mixed results. Lassale et al. found a protective association between the MDS, rMED, (but not MSDPS), DQI-I, DGAI, and PNNS-GS and body weight in men, but not in women. Zamora et al. found a protective association between the DQI-2005 and body weight in normal weight Caucasians but increased body weight in obese Blacks. One study found no effect of a Mediterranean diet intervention on body weight in at-risk elderly (Estruch, 2006).
 
BMI: Of the four studies that assessed BMI independent of risk of overweight or obesity, two were conducted with adults using HEI scores or the PNNS-GS (Gao, 2008; Kesse-Guyot, 2009) and two were conducted with children and adolescents using the Revised Children’s (RC)-DQI or a DASH score (Cheng, 2010; Berz 2011). The two studies in adults found an association between an increased dietary guidelines index and decreased BMI over time, although the multiethnic MESA study found this primarily in Caucasians (Gao, 2008). The two studies in children and adolescents examined pre-puberty to puberty onset (Cheng, 2010) or the adolescent period (Berz, 2011) with differing results. Although Cheng et al. found no association between the RC-DQI score and BMI at puberty onset in boys and girls, Berz et al. found that girls in the highest quintile of DASH score had the smallest gains in BMI throughout adolescence. 
 
Waist Circumference and Percent Body Fat: Waist circumference was measured as a component of metabolic syndrome and five studies assessed this measure of abdominal adiposity. All of these studies were consistent in that they found an inverse association between a score or index and waist circumference over time (Jacobs, 2009; Kesse-Guyot, 2009; Rumawas, 2009 [MSDPS]; Tortosa, 2007 [MDS]), although the MESA study found this association primarily in Caucasians (Gao, 2008). Jacobs et al. also assessed changes in percent body fat and found that an increase in a customized a priori diet score was associated with a decrease in percent body fat in subjects with metabolic syndrome.
 
Taken together, the majority of studies found that an increase in Mediterranean diet score or dietary guideline-related index was associated with some measure of improved body weight or composition in adults. The MDS, followed by the rMED and PNNS-GS were the most commonly applied scores associated with a protective outcome.
 
Table 4-A-I-1 Summary of Findings
Hypothesis that increased adherence to dietary pattern improves measures of body weight and risk of obesity
 

Endpoint Outcomes: Incidence of Overweight and Obesity

Seven studies included in this review considered the relationship between dietary patterns and the clinical outcome of incident overweight or obesity. The results varied across the studies on the Mediterranean diet scores and dietary guideline indices (tables 4-A-I-1 and 4-A-I-3).
 
Association: Two studies found a protective, inverse relationship between either a Mediterranean score (rMED) or dietary guidelines index (PNNS-GS) and risk of normal weight individuals becoming overweight or obesity over
5 to 6 years (Romaguera, 2010; Kesse-Guyot, 2009). The Romaguera (2010) study included a meta-analysis of the ten EPIC-PANACEA countries and showed an association between a continuous increase in rMED score and a decrease in odds of normal weight individuals becoming overweight or obese.
 
Mixed Association: There were varied results in three studies. Lassale et al. found an inverse association between the MDS, rMED, (but not MSDPS), DQI-I, DGAI, and PNNS-GS scores and odds of non-obese individuals becoming obese over 13 years only in men. For women, there was no association of any diet score with obesity risk; however, there was a non-significant reduction in obesity risk with an increase in rMED or PNNS-GS. Other studies reported no differences by gender or did not assess men and women separately. Gao et al. examined Black, Caucasian, Hispanic, and Chinese Americans in the MESA study to determine if ethnicity impacted the association between either of two HEI scores and incidence of overweight or obesity. They found that the ability of HEI scores to predict risk of obesity was significant only in Caucasians. Mendez et al. examined the 3-year incidence of normal weight individuals developing overweight and overweight individuals developing obesity. They found an inverse association between MDS score and likelihood of overweight individuals becoming obese, but no association for normal weight individuals becoming overweight. The mixed results in these three studies are discussed further below related to methodological differences in weight categories across studies.
 
Null Association: There were null findings reported in two studies. Woo et al. assessed adherence to the MDS and DQI-I and incidence of normal weight individuals becoming overweight or obese in an ethnic Chinese population from Hong Kong. They found no association between the MDS or DQI-I and development of overweight, even by the Asian standard of BMI ≥23 kg/m2, or obesity over 5 to 9 years. Beunza et al. assessed adherence to the MDS and incidence of normal weight individuals becoming overweight or obese and found the MDS was not associated with incidence of overweight or obesity over 6 years. These studies are discussed further below related to methodological differences in weight categories across studies.
 
Differences in weight categories across studies: The method by which these studies assessed risk of overweight or obesity varied in terms of how weight categories at baseline and follow-up were operationalized. As indicated in Table 4-A-I-1, analysis of weight gain included changes from (1) normal weight to overweight; (2) normal weight to obese; (3) normal weight to overweight and obese; or (4) non-obese to obese. Three studies that followed normal weight individuals for development of overweight (BMI ≥25 to <30 kg/m2) found no protective effect of adherence to the MDS (Mendez, 2006; Woo, 2008), HEI (Gao, 2008), or DQI-I (Woo, 2008). Gao et al. and Woo et al. also assessed normal weight individuals becoming obese (BMI ≥30 kg/m2); Gao reported an association only in Caucasians; whereas, Woo reported no association between dietary guidelines index and obesity. The three studies that assessed normal weight individuals for development of overweight or obesity (BMI ≥25 kg/m2) found inconsistent results. Beunza et al., found no association between the MDS and incidence of overweight or obesity. However, Romaguera et al. and Kesse-Guyot et al. found there was an inverse association between the rMED or the PNNS-GS, respectively, and risk of becoming overweight or obese. These studies were all conducted in Europe and compared the same weight status over a similar time period, although they used different scores to assess diet exposure. Additionally, the SUN cohort examined by Beunza et al. was relatively young (mean age 38.4 years) with >90 percent non-obese participants. A fourth weight status comparison was assessing the risk of non-obese (BMI ≤30 kg/m2) individuals becoming obese. Both Lassale et al. and Kesse-Guyot et al. assessed non-obese individuals becoming obese in the same SU.VI.MAX cohort. Kesse-Guyot found an inverse association between the PNNS-GS and obesity incidence; whereas, Lassale found an inverse association between several diet scores and obesity incidence in men but not women.
 
When viewed based on the method by which weight categories were operationalized, three studies showed that adherence to a dietary pattern was not associated with risk of normal weight individuals becoming overweight (Gao, 2008; Mendez, 2006; Woo, 2008), although the Gao study showed that adherence was associated with risk of normal weight individuals (Caucasians) becoming obese (Gao, 2008); two out of three studies showed an association with risk of normal weight individuals developing overweight or obesity (Romaguera, 2010; Kesse-Guyot, 2009); two out of two studies found an association with risk of non-obese individuals becoming obese (Kesse-Guyot, 2009; Lassale, 2012), although Lassale found this only in men; and one study showed an association with risk of overweight individuals becoming obese (Mendez, 2006).
 

Themes and Key Findings for Components of Scores and Indices:

Components of Mediterranean Diet Pattern Scores and Dietary Guidelines-Related Indices

Three studies assessed the association between individual components of a Mediterranean diet score or dietary guidelines index and weight change or incidence of overweight and obesity. Mendez et al. found that when individual components of the MDS were assessed, overweight and obesity incidence was higher in women, but not men, who consumed more meat; whereas, obesity incidence was lower in men, but not women, who consumed more cereals. Romaguera et al. found that the association between rMED adherence and weight change was only significant when the meat and meat products component was included in the score. Both the MDS and rMED are population-based scores, with the cereal component including whole and refined grains, and the meat component defined as meat and poultry in the MDS and total meat including processed meat in the rMED. In the Prospective National Health and Growth study in girls, BMI was assessed throughout adolescence for association with average intake of four DASH food groups: total fruits, vegetables, whole grains, and low-fat dairy over 10 years (Berz, 2011). Total fruit, low-fat dairy, and to a lesser extent whole grains had a beneficial association with BMI over this time period.

Components across All Scores and Indices

The scores or indices that were associated with the clinical endpoint of interest?risk of overweight or obesity?were the MDS, rMED, HEI-1995 and a customized HEI-05, DQI-I, DGAI, and PNNS-GS. (Although, some studies found no association with MDS (Beunza, 2010; Woo, 2008) or DQI-I (Woo, 2008) in low-risk or Asian populations, respectively.) Scores or indices that were associated with decreased risk of overweight or obesity in adults were selected to examine commonalities in components across scores related to the clinically significant outcome, rather than intermediate or surrogate markers of outcomes. (The components of these scores are described in detail in table 4-A-I-2.) These scores include the MDS, rMED, and PNNS-GS that were also the most commonly applied scores associated with a protective outcome in body weight, BMI, or waist circumference in adults. The MSDPS was not associated with either endpoint obesity outcomes or intermediate outcomes including body weight and BMI, although Rumawas et al. found an inverse association between MSDPS and waist circumference. Lastly, although this review included a study that utilized the DASH score (Berz, 2011), they did not consider the clinical outcomes of overweight or obesity in analyses, and thus, the DASH score was not included in the comparison below. Berz et al. reported that the DASH score was associated with BMI in girls throughout adolescence, but they did not assess adults.
 
Table 4-A-I-2 Comparison of Dietary Components Across Dietary Pattern Scores and Indices

Food components that were common across these scores or indices were operationalized differently in that foods were aligned, described, or scored in dissimilar ways. Given this caveat, the food groups, as well as foods or nutrients, that were included as positive components in these scores were fruits (MDS and rMED specified fruits and nuts; DQI-I and DGAI added fruit variety), vegetables (rMED excluded potatoes; DGAI gave positive points for dark-green and orange vegetables and penalized for overconsumption of starchy vegetables; DQI-I and DGAI added vegetable variety); whole grains, in some combination with refined grains in cereals, flour, pasta, rice, and bread; legumes; fish (MDS and rMED) or seafood (PNNS-GS); and dietary fat as olive oil (rMED), MUFA/SFA (MDS), PUFA/MUFA/SFA (DQI-I), percent total fat and percent SFA (HEI, DQI-I, DGAI), or added fats and added vegetable fats (PNNS-GS). Moderate alcohol intake was commonly included as a positive component, with different cut-offs for men and women; exceptions were the HEI that did not include alcohol and the DQI-I that included alcohol as part of the empty calorie foods.
 
There was inconsistency in the way meat and dairy were evaluated. The Mediterranean diet scores assessed meat (MDS included meat and poultry; rMED included meat and processed meats) and dairy (rMED included low- and high-fat milk, yogurt, cheese, and desserts) negatively. For dietary guidelines indices, meat and dairy were scored based on meeting recommended servings, with maximum points for meeting guidelines and proportional negative points for percent deviation (with the exception of the HEI that did not deduct points for overconsumption). The DQI-I included meat and dairy in both the food group and protein variety components.
 
There were additional components included only in some indices. The DGAI, PNNS-GS, and DQI-I included added sugar, sweetened foods, or empty-calorie foods (foods low in nutrient density, including sugar), respectively; these were scored negatively above a specified percent energy intake or below a specified level of nutrient density for the DQI-I. The PNNS-GS also negatively scored sweetened beverages, in relation to water consumption, under a separate beverage component. The dietary guidelines-related indices had additional components including sodium (or salt), cholesterol (not PNNS-GS), and trans fats (only DGAI). Three of the dietary guidelines indices included diet variety (DQI-I, DGAI, and HEI). Furthermore, the DQI-I score that predicted the lowest odds of non-obese men becoming obese in the Lassale et al. study included many components not present in other scores or indices. These were food group variety and within-group variety in dietary protein (both noted above for meat and dairy); adequacy that included vegetables, fruits, cereals, fiber, protein, iron, calcium, and vitamin C; the above empty-energy foods; and in addition to a PUFA/MFA/SFA ratio, a macronutrient carbohydrate/protein/fat ratio. Overall, the DQI-I was more nutrient-based than food-based, compared to the other scores and indices. Woo et al., in addition to the MDS and DQI-I, also looked at food variety and found that variety in snack consumption was associated with weight gain.
 
Taken together, the positive components of the scores or indices that were associated with decreased risk of obesity in one or more studies were fruits (MDS and rMED included nuts with the fruit component), vegetables, whole grains, legumes, unsaturated fats, and fish. Alcohol was commonly included as a positive component when consumed in moderation. Meat and dairy, with some variations, were negative components in Mediterranean scores or recommended within specific ranges for dietary guidelines indices. The dietary guidelines indices also included saturated fat and cholesterol, or added non-vegetable fats, as negative components above a specified level of intake. Lastly, the sugar or sweets component was included and scored negatively in three out of the four dietary guidelines-related indices, although similar components were not included in either the MDS or rMED.
 

Qualitative Assessment of the Collected Evidence

Quality and Quantity

Quality assessment for the studies included in this systematic review involved determining the validity of each study. Validity was assessed by examining the scientific soundness of study design and execution to avoid potential bias in the findings related to outcomes. This can include selection, performance, attrition, detection, or reporting bias. As the preponderance of the evidence consisted of positive quality studies (12 out of 14 studies), this indicates a low risk of bias. In terms of quantity, the majority of these studies were prospective cohort studies with large numbers of participants in nationally recognized cohorts.

Consistency

The evidence of a protective association between a dietary pattern score and change in body weight over time was consistent in the majority of studies that used either a Mediterranean diet score or dietary guidelines index in healthy adult populations, although there were differences reported based on gender and race. The evidence related to other intermediate markers such as changes in BMI and waist circumference was also consistent. However, there was more variation in the endpoint outcome of incidence of overweight or obesity, including within study variation based on gender, race, and weight status, and between study variations in health outcomes. Inconsistency across these studies could be due to differences in the health outcomes measured, variability in the study populations, or differences in assessment of adherence to a given dietary pattern. Although overweight and obesity were similarly defined according to BMI cutoff points (except Woo et al.), there were differences in the way baseline weight status (e.g., normal weight or non-obese) and follow-up weight status were categorized, which could have contributed to inconsistency in results. This variation could also contribute to risk of bias due to outcome measures. Furthermore, some studies calculated overweight and obesity risk separately by gender or race, while others only assessed the pooled population. Across the studies, there were differences in populations as well. The two RCTs were conducted in different at-risk populations. Although the majority of prospective cohort studies were conducted in Europe, there were differences in cohorts; for example, age and weight status of the baseline population in the SUN cohort compared to EPIC cohorts. Lastly, the assessment of dietary exposure in these studies was determined using a large number of different Mediterranean diet pattern scores and dietary guidelines-related indices.

Impact

This body of evidence directly addressed the interventions/exposures and health outcomes of interest for this systematic review. Overall, several large prospective cohort studies found a decrease in incidence of overweight and obesity associated with adherence to a Mediterranean diet score or dietary guidelines index. Although not clinical trials, these cohort studies reported results that are applicable in free-living populations. For example, in the largest study, EPIC-PANACEA, Romaguera et al. reported that a change from lowest to highest rMED adherence reduced the likelihood of normal weight individuals becoming overweight or obese in 5 years by 10 percent. It should be noted, however, that not all studies found this protective association.

Generalizability/External Validity              

The preponderance of the evidence related to the Mediterranean dietary pattern involved large prospective cohort studies conducted in Europe, with four out of eight studies conducted in Spain. In terms of generalizability related to the American population, one study was conducted in the United States, and this study assessed waist circumference as a component of metabolic syndrome (Rumawas, 2009). Therefore, although there was evidence with high quality studies on the Mediterranean dietary pattern in Europe, there was limited data regarding the U.S. population and intermediate outcomes and no data on incidence of overweight and obesity.
 
Related to dietary guidelines patterns, four studies examined U.S. populations, but only two studies focused on healthy adults. Both of these studies included more heterogeneous ethnic/racial groups than the European studies, as one focused on multi-ethnic outcomes from the MESA cohort (Gao, 2008) and one, the CARDIA study, focused on differences between Black and Caucasian young adults (Zamora, 2010). Both of these studies found a protective effect of adherence to an a priori index only in Caucasians. The remaining dietary guidelines related studies were conducted in France, Hong Kong, Germany, and Norway.
 
Given the limited evidence involving U.S. cohorts, the relevance of this body of evidence to U.S. policy on dietary patterns and risk of overweight and obesity would seem to depend on the extent to which results from studies conducted with large European cohorts can be applied to the United States. One of the benefits of doing an index/score-based analysis is that investigators can compare across cohorts because food components are chosen a priori and are independent of specific populations (although the numeric scoring of dichotomous scores, such as the MDS, are based on median population intakes). Studies, such as that conducted by Lassale et al. that assessed both Mediterranean diet scores and U.S.-based (as well as French) dietary guidelines indices, with similar predictive values for obesity, at least in men, provide some further evidence that findings from studies conducted with large European cohorts may be applicable to U.S. populations. This is supported by the fact that the European countries and the United States are classified as “very high” Human Development Index countries. Furthermore, although individual studies may not be representative of the population of interest, consistent findings across studies may suggest broad applicability of the results (Viswanathan, 2012).
 

Limitations of the Evidence

Limitations of the studies included in this systematic review, and potential reasons for differences and inconsistencies in results, include the use of different scores; differences between scores that are based on median population intakes versus indices that are based on recommended intakes; the use of different confounding factors or lack of sufficient adjustment for confounding factors; the problems associated with the use of different FFQs and validation related to other methods of diet assessment; and the handling of underreporting. Furthermore, in the majority of studies, total scores or indices were used and there was no separate analysis of individual score components and their potential association with outcomes. The application of the total score to the diet pattern analysis has the potential to “dilute” the effect of individual components. However, the assessment of individual components without interaction terms assumes that a given component has an independent association which potentially contradicts the theoretical rationale for examining the overall dietary pattern. Lastly, another common limitation was the single measurement of dietary intake at baseline. This does not take into account that diets change over time due to trends in the food supply and population-level and individual-level changes in food choices.
 

Abbreviations

Scores & Indices: Dietary Approaches to Stop Hypertension (DASH); Dietary Guidelines Adherence Index (DGAI); Diet Quality Index (DQI); DQI-International (DQI-I); Revised Children’s (RC)-DQI; Healthy Eating Index (HEI), Alternate Healthy Eating Index (AHEI); Mediterranean Diet Score (MDS); Mediterranean Style Diet Pattern Score (MSDPS); Relative Mediterranean Diet Score (rMED); Programme National Nutrition Santé Guideline Score (PNNS-GS)
Cohorts: Coronary Artery Risk Development in Young Adults (CARDIA); Dortmund Nutritional and Anthropometric Longitudinally Designed (DONALD); European Prospective Investigation into Cancer and Nutrition (EPIC); EPIC-Physical Activity, Nutrition, Alcohol Consumption, Cessation of Smoking, Eating out of Home, and Obesity  (EPIC-PANACEA); Framingham Offspring and Spouse (FOS); Multi-Ethnic Study of Atherosclerosis (MESA); Prevencion con Dieta Mediterranea (PREDIMED); Seguimiento Universidad de Navarra (SUN); SUpplementation en VItamines et Minereaux AntioXydants (SU.VI.MAX)

Table 4-A-I-3 Overview Table: Body Weight and Obesity
 

Research Recommendations

Given the combined evidence from this systematic review, several research recommendations can be advanced. Most striking is the need for consensus on a single index or score that is applicable across populations for a diversity of outcomes. If it is not feasible that one index can adequately assess the diversity of populations related to dietary patterns, research should be conducted to determine the best method by which components are chosen, grouped, and scored and whether or not the research tool is population based or independent of the population, so that there is uniformity across scores. The studies included in this review were focused on total scores, rather than component scores and their association with health outcomes. To strengthen the analysis of component scores, the interaction terms across components need to be assessed in order to maintain a dietary patterns approach. For prospective cohort studies, diet intake should be measured at multiple time points with assessment of dietary changes over the time as they relate to health outcomes.

See Search and Sort Plan for References of Included Articles

REFERENCES (in addition to included articles) 
 

  1. Buckland G, González CA, Agudo A, Vilardell M, Berenguer A, Amiano P, Ardanaz E, Arriola L, Barricarte A, Basterretxea M, et al.  Adherence to the Mediterranean diet and risk of coronary heart disease in the Spanish EPIC Cohort Study. Am J Epidemiol. 2009 Dec 15;170(12):1518-29. PubMed PMID: 19903723.
  2. Estaquio C, Kesse-Guyot E, Deschamps V, Bertrais S, Dauchet L, Galan P, Hercberg S, Castetbon K. Adherence to the French Programme National Nutrition Santé Guideline Score is associated with better nutrient intake and nutritional status. J Am Diet Assoc. 2009 Jun;109(6):1031-41. PubMed PMID: 19465185.
  3. Fogli-Cawley JJ, Dwyer JT, Saltzman E, McCullough ML, Troy LM, Jacques PF. The 2005 Dietary Guidelines for Americans Adherence Index: development and application. J Nutr. 2006 Nov;136(11): 2908-15. PubMed PMID: 17056821.
  4. Kennedy ET, Ohls J, Carlson S, Fleming K. The Healthy Eating Index: design and applications. J Am Diet Assoc. 1995 Oct;95(10):1103-8. PubMed PMID: 7560680.
  5. Kim S, Haines PS, Siega-Riz AM, Popkin BM. The Diet Quality Index-International (DQI-I) provides an effective tool for cross-national comparison of diet quality as illustrated by China and the United States. J Nutr. 2003 Nov;133(11):3476-84. PubMed PMID: 14608061.
  6. 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 Jun;139(6):1150-6. PubMed PMID: 19357215.         
  7. Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to a Mediterranean diet and survival in a Greek population. N Engl J Med. 2003 Jun 26;348(26):2599-608. PubMed PMID: 12826634.
  8. Viswanathan M, Ansari MT, Berkman ND, Chang S, Hartling L, McPheeters LM, Santaguida PL, Shamliyan T, Singh K, Tsertsvadze A, Treadwell JR. Assessing the Risk of Bias of Individual Studies in Systematic Reviews of Health Care Interventions. Agency for Healthcare Research and Quality Methods Guide for Comparative Effectiveness Reviews. March 2012. AHRQ Publication No. 12-EHC047-EF. Available at: www.effectivehealthcare.ahrq.gov/

 


Research Design and Implementation
For a summary of the Research Design and Implementation results, click here.
Worksheets
Berz JP, Singer MR, Guo X, Daniels SR, Moore LL. Use of a DASH food group score to predict excess weight gain in adolescent girls in the National Growth and Health Study. Arch Pediatr Adolesc Med. 2011 Jun; 165(6): 540-546.

Beunza JJ, Toledo E, Hu FB, Bes-Rastrollo M, Serrano-Martínez M,Sánchez-Villegas A, Martínez JA, Martínez-González MA. Adherence to the Mediterranean diet, long-term weight change, and incident overweight or obesity: the Seguimiento Universidad de Navarra (SUN) cohort. Am J Clin Nutr. 2010 Dec; 92(6): 1,484-1,493.

Cheng G, Gerlach S, Libuda L, Kranz S, Günther AL, Karaolis-Danckert N, Kroke A, Buyken AE. Diet quality in childhood is prospectively associated with the timing of puberty but not with body composition at puberty onset. J Nutr. 2010 Jan; 140(1): 95-102.

Estruch R, Martínez-González MA, Corella D, Salas-Salvadó J, Ruiz-Gutiérrez V, Covas MI, Fiol M, Gómez-Gracia E, López-Sabater MC, Vinyoles E, Arós F, Conde M, Lahoz C, Lapetra J, Sáez G, Ros E; PREDIMED Study Investigators. Effects of a Mediterranean-style diet on cardiovascular risk factors: a randomized trial. Ann Intern Med. 2006; 145(1): 1-11.

Gao SK, Beresford SA, Frank LL, Schreiner PJ, Burke GL, Fitzpatrick AL. Modifications to the Healthy Eating Index and its ability to predict obesity: the Multi-Ethnic study of atherosclerosis. Amer J Clin Nutr 2008; 88: 64-69. 

Jacobs DR Jr, Sluik D, Rokling-Andersen MH, Anderssen SA, Drevon CA. Association of one-year changes in diet pattern with cardiovascular disease risk factors and adipokines: Results from the one-year randomized Oslo Diet and Exercise Study. Am J Clin Nutr. 2009 Feb; 89(2): 509-517.

Kesse-Guyot E, Castetbon K, Estaquio C, Czernichow S, Galan P, Hercberg S. Association between the French nutritional guideline-based score and six-year anthropometric changes in a French middle-aged adult cohort. Am J Epidemiol. 2009 Sep 15; 170(6): 757-765.

Lassale C, Fezeu L, Andreeva VA, Hercberg S, Kengne AP, Czernichow S, Kesse-Guyot E. Association between dietary scores and 13-year weight change and obesity risk in a French prospective cohort. Int J Obes (Lond). 2012 Nov; 36(11): 1,455-1,462.

Mendez MA, Popkin BM, Jakszyn P, Berenguer A, Tormo MJ, Sanchéz MJ, QuirósJR, Pera G, Navarro C, Martinez C, Larrañaga N, Dorronsoro M, Chirlaque MD,Barricarte A, Ardanaz E, Amiano P, Agudo A, González CA. Adherence to a Mediterranean diet is associated with reduced three-year incidence of obesity. J Nutr. 2006 Nov; 136(11): 2,934-2,938.

Romaguera D, Norat T, Vergnaud AC, Mouw T, May AM, et al. Mediterranean dietary patterns and prospective weight change in participants of the EPIC-PANACEA project. Am J Clin Nutr. 2010 Oct; 92(4): 912-921.

Rumawas ME, Meigs JB, Dwyer JT, McKeown NM, Jacques PF. Mediterranean-style dietary pattern, reduced risk of metabolic syndrome traits, and incidence in the Framingham Offspring Cohort. Am J Clin Nutr. 2009; 90(6): 1,608-1,614.

Tortosa A, Bes-Rastrollo M, Sanchez-Villegas A, Basterra-Gortari FJ, Nuñez-Cordoba JM, Martinez-Gonzalez MA. Diabetes Care. 2007 Nov; 30(11): 2,957-2,959.

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 Apr; 62(4): 480-487.

Zamora D, Gordon-Larsen P, Jacobs DR Jr, Popkin BM. Diet quality and weight gain among black and white young adults: The Coronary Artery Risk Development in Young Adults (CARDIA) Study (1985-2005). Am J Clin Nutr. 2010 Oct; 92(4): 784-793.