Does Adherence to Mediterranean Diet Mediate the Association Between Food Environment and Obesity Among Non-Hispanic Black and White Older US Adults? A Path Analysis

Abstract

Purpose:

This study aims to test the hypothesis that in addition to a direct effect of food environment on obesity, food environment is indirectly associated with obesity through consuming Mediterranean diet (MD).

Design:

Cross-sectional secondary data analysis.

Setting:

Nationwide community-dwelling residency.

Sample:

A total of 20 897 non-Hispanic black and white adults aged ≥45 years who participated in the REasons for Geographic and Racial Differences in Stroke study and completed baseline assessment during January 2003 and October 2007.

Measures:

The Modified Retail Food Environment Index (mRFEI; 0-100) was used as food environment indicator. The MD score (0-9) was calculated to indicate the dietary pattern adherence. Body mass index (BMI; kg/m²) was used to estimate obesity.

Analysis:

Path analysis was used to quantify the pathways between food environment, MD adherence, and obesity. Proper data transformation was made using Box–Cox power transformation to meet certain analysis assumptions.

Results:

The participants were from 49 states of the United States, with the majority (64.42%) residing in the South. Most of the participants were retired, female, white, married, having less than college graduate education, having annual household income ≤75 000, and having health insurance. The means of mRFEI was 10.92 (standard deviation [SD] = 10.19), MD score was 4.36 (SD = 1.70), and the BMI was 28.96 kg/m² (SD = 5.90). Access to healthy food outlets (β = .04, P < .0001) and MD adherence (β = .08, P < .0001) had significant and inverse relationships with BMI, respectively. Mediterranean diet adherence mediated the relationship between food environment and obesity among a subpopulation who had an annual household income of <$75 000 (β = −.02, P = .0391).

Conclusion:

Population-tailored interventions/policies to modify food environment and promote MD consumption are needed in order to combat the obesity crisis in the United States.

Keywords

obesity BMI dietary pattern Mediterranean diet food environment built environment path analysis population health health disparities epidemiology

Purpose

Obesity continues to be a significant public health issue in the United States.¹ The Mediterranean diet (MD) has been recognized increasingly for its protective effects against obesity and recently recommended as a healthy diet pattern for Americans by US Department of Agriculture (USDA) and the US Department of Health and Human Services in the Dietary Guidelines for Americans 2015-2020.^2,3 Meanwhile, interests of researchers and policy makers continue to rise in employing an ecological approach to better understand how community food environment influences individuals’ diet and obesity status, allowing for the development of more effective interventions and policies to promote healthy diet and reduce obesity.^4,5 The overall results of studies that examined the relationship between community food environment and diet suggested that access to healthful food outlets was positively associated with healthful food intake.⁶ However, findings were inconsistent in studies that examined the relationship between community food environment and obesity status.⁷ Some studies reported that access to healthful food outlets (eg, supermarket) was inversely associated with obesity status, while few studies found the association was positive. Similarly, some studies found that access to unhealthful food outlets (eg, convenience store, fast-food restaurant) were positively associated with obesity status among the adults; however, others found null or even, opposite associations.⁷

Overall, published research suggested that a network between community food environment, individual’s diet, and obesity status interrelates with each other. However, little has been published which examines this network as a system. Most previous research in this field has used regression analysis, focusing on the examination of a single part of pathway, either direct association for community food environment with dietary behavior/intake or with obesity-related parameters (eg, body mass index [BMI]).^8
-10 The extent to which the community food environment contributes to obesity through diet has not been well understood. Moreover, as a relatively new dietary pattern in the United States, it remains unknown about the role of consuming an MD in combating obesity within the context of the prevailing community food environment in the United States; in another word, if consuming an MD can mediate the relationship between community food environment exposure and obesity status. Using statistical approaches that allow more explicit consideration of complex interrelationships (eg, path analysis) may provide new and important insights and help in the design of more efficient and effective interventions by identifying the mechanism through which interventions can have an effect.

This study aims to examine the mediating role of consuming MD in the relations between community food environment and obesity status among US adults. It is hypothesized that in addition to a direct relationship, the community food environment is indirectly associated with obesity status through consuming MD.

Methods

Design

This cross-sectional secondary data analysis study used data from the REasons for Geographic and Racial Differences in Stroke (REGARDS) study, a national population-based longitudinal cohort study of non-Hispanic black and white community-dwelling residents in the United States.¹¹ The overall goal of REGARDS is to better understand the contributors to the substantial racial and geographic disparities in stroke.¹¹ Individuals were recruited from commercially available nationwide list of residents purchased through Genesys Inc using a combination of mail and telephone contact during 2003 to 2007. Additional methodological details have been published previously.¹¹ Community food environment data were retrieved from the Children’s Food Environment State Indicator Report (2011) developed by the Division of Nutrition, Physical Activity and Obesity of Centers for Disease Control and Prevention, available to the public.¹²

Sample

A total of 20 897 adults aged ≥45 years who participated in the REGARDS study and completed baseline assessment during January 2003 and October 2007 were in analysis. The participants were from 49 states of the United States, with the majority (64.42%) residing in South Carolina, North Carolina, Georgia, California, Louisiana, Alabama, Tennessee, Mississippi, and Ohio. The participants were aged 65 on average, with about half retired, slightly more female, about two-thirds white, married, and having less than college graduate education, about four-fifths having an annual household income ≤75 000, and almost all having health insurance.

Measures

The Modified Retail Food Environment Index (mRFEI; range: 0-100) was used as food environment indicator. The mRFEI represented the percentage of food retailers that were healthy out of the total number of food retailers considered healthy or less healthy in a census tract. Healthy food retailers include supermarkets, larger grocery stores, supercenters, and produce stores (as defined by the North American Industry Classification System [NAICS]: supermarkets defined as stores with ≥50 annual payroll employees and larger grocery stores defined as stores with 10-49 employees) within census tracts or ½ mile from the tract boundary. Less healthy food retailers include fast-food restaurants, small grocery stores, and convenience stores (as defined by NAICS: convenience stores or small groceries defined as where the number of employees was 3 or fewer) within census tracts or ½ mile from the tract boundary.¹² The lower mRFEI scores indicate that census tracts contain many convenience stores and/or fast-food restaurants compared to the number of healthy food retailers. For example, an mRFEI score of 10 means that only 10 of every 100 stores in the community are likely to offer healthy foods, while the other 90 stores were unlikely to provide access to healthy foods.¹² Mediterranean diet score (MD score; range: 0-9) was calculated to indicate the dietary pattern adherence. The food intake data were collected using the self-administered Block 98 Food Frequency Questionnaire (including 110 food items) at REGARDS study baseline. The calculation of MD score followed Trichopoulou et al method.¹³ The intake of each of 9 food categories (dairy, meat, fruit, vegetables, legumes, cereals, fish, fat, and alcohol) was computed, and a value of 0 or 1 was assigned to each of 9 food components with the use of sex-specific medians as the cutoffs. For beneficial components (fruit, vegetables, legumes, cereals, and fish), persons whose consumption was below the median were assigned a value of 0, and persons whose consumption was at or above the median was assigned a value of 1. For meat, fat, and dairy consumption, a value of 1 was assigned to persons whose intake was less than the median. Regarding alcohol consumption, persons with moderate alcohol use (0-7 drinks/wk for women and 0-14 drinks/wk for men) were assigned a value of 1, otherwise (heavy alcohol use: having >7 drinks/wk for women and >14 drinks/wk for men) a value of 0 was assigned. Then the MD adherence score scale was generated by summing up the scores of the 9 food categories. The scale ranged from 0 to 9, with a higher score indicating higher MD adherence.¹³ Body mass index (kg/m²) was calculated to estimate obesity status. The BMI was calculated using height and weight measured during REGARDS study home visit at baseline. Height was obtained utilizing an 8-ft metal tape measure without shoes. Weight was measured using a standard 300-lb calibrated digital scale.¹¹

Analysis

The REGARDS study data and food environment data were integrated using the spatial join function in ArcGIS 10.4 (ESRI Inc, Redlands, California). The joined data were then exported and analyzed in SAS version 9.4 for Windows (SAS Institute, Inc, Cary, North Carolina). Descriptive analysis of sociodemographic characteristics of the participants was conducted. Path analysis using PROC CALIS procedure was conducted to examine the relationship between food environment, MD adherence scores, and obesity status (transformed BMI values) among the whole sample population as well as subgroups stratified by the sociodemographic factors that were possibly related to MD adherence as suggested in previous studies.^14
-16 These factors included gender (female or male), race (white or black), and annual household income (≤$75 000 or >$75 000). Proper data transformation (eg, BMI) was made using Box–Cox power transformation to meet certain analysis assumptions (eg, multivariate normality). The α level denoting statistical significance was set at .05 (2 tailed). Standardized path coefficients and P values for paths and explained variation for endogenous variables (R ²) were reported.

Results

Descriptions of the characteristics of the participants are summarized in Tables 1 and 2. The participants were from 49 states of the United States, with the majority (64.42%) residing in the South. Most of the participants were retired, female, white, married, having less than college graduate education, having annual household income ≤75 000, and having health insurance. The means of mRFEI was 10.92 (standard deviation [SD] = 10.19), MD score was 4.36 (SD = 1.70), and the BMI was 28.96 kg/m² (SD = 5.90). The results of the path analysis for the effects of community food environment on MD adherence and obesity status among the whole sample population are presented in Figure 1. Obesity status was measured by 1/sqrt (BMI) (continuous), transformed from the original BMI values as suggested by Box–Cox transformation (λ = −.5). The model was a good fit, as indicated by the fit indices (χ² = .0000, goodness-of-fit index [GFI] = 1.000, root mean square residual [RMR] = 0.0000, standardized root mean square residual [SRMR] = 0.0000). The results of the path analysis among the whole sample population showed that access to healthful food outlets (β = .04, P < .0001) and consuming MD (β = .08, P < .0001) had significant direct effects on obesity. Specifically, 1 unit increase in mRFEI and MD score resulted in 0.04 unit and 0.08 unit increase in 1/sqrt (BMI), respectively. However, no significant indirect effect of access to healthful food outlets on obesity through adherence to MD was found.

Table 1.

Number of the REasons for Geographic and Racial Differences in Stroke Study Participants by State.^a

State Name	n (%)
North Carolina	2440 (11.68)
South Carolina	2425 (11.60)
Georgia	2123 (10.16)
California	1516 (7.25)
Louisiana	1417 (6.78)
Alabama	1078 (5.16)
Tennessee	878 (4.20)
Mississippi	811 (3.88)
Ohio	776 (3.71)
Florida	720 (3.45)
Michigan	715 (3.42)
Arkansas	597 (2.86)
New York	557 (2.67)
Texas	533 (2.55)
Illinois	522 (2.50)
Pennsylvania	422 (2.02)
Maryland	408 (1.95)
Virginia	309 (1.48)
Missouri	302 (1.45)
Wisconsin	246 (1.18)
Indiana	243 (1.16)
Minnesota	181 (0.87)
Kentucky	170 (0.81)
New Jersey	155 (0.74)
Massachusetts	148 (0.71)
Oklahoma	135 (0.65)
District of Columbia	107 (0.51)
Iowa	106 (0.51)
Colorado	99 (0.47)
Washington	97 (0.46)
Connecticut	86 (0.41)
Oregon	84 (0.40)
Arizona	70 (0.33)
Kansas	65 (0.31)
Nebraska	53 (0.25)
Utah	43 (0.21)
Delaware	36 (0.17)
West Virginia	33 (0.16)
Idaho	28 (0.13)
Nevada	24 (0.11)
South Dakota	21 (0.10)
Maine	19 (0.09)
New Mexico	19 (0.09)
North Dakota	19 (0.09)
New Hampshire	15 (0.07)
Rhode Island	14 (0.07)
Montana	13 (0.06)
Wyoming	13 (0.06)
Vermont	6 (0.03)

^a N = 20 897.

Table 2.

Summary of Characteristics of the REGARDS Study Participants.^a

Characteristics	REGARDS Participants
Sociodemographics
Age, years, mean (SD)	64.88 (9.26)
Male, % (n)	44.22 (9241)
White, % (n)	66.71 (13 941)
Education, % (n)
Less than college graduate	62.43 (13 040)
College graduate and above	37.57 (7849)
Relationship, % (n)
Married	61.74 (12 901)
Single/widowed/devoiced	38.26 (7996)
Income, % (n)
≤75 000	80.54 (14 859)
>75 000	19.46 (3590)
Employment, % (n)
Employed for wages	27.09 (3565)
Self-employed	9.00 (1184)
Unemployed for ≥1 year	1.47 (194)
Unemployed for <1 year	1.48 (195)
Homemaker	6.08 (800)
Student	0.19 (25)
Retired	47.72 (6279)
Unable to work	6.95 (914)
Refused	0.02 (3)
Health insured, % (n)	93.95 (19 620)
Time lived in current address, year, mean (SD)	28.63 (20.62)
mRFEI, mean (SD)	10.92 (10.19)
MD score, mean (SD)	4.36 (1.70)
BMI, kg/m², mean (SD)	28.96 (5.90)

Abbreviations: BMI, body mass index; MD, Mediterranean diet; mRFEI, Modified Retail Food Environment Index; REGARDS, REasons for Geographic and Racial Differences in Stroke; SD, standard deviation.

^a N = 20 897.

Figure 1.

Path analytic model of the effects of community food environment on MD adherence and obesity among the REGADRS study participants (N = 20 897). Box–Cox transformation was used to transform original BMI to meet the multivariate normality assumption. Values shown are standardized path coefficients (β), explained variations (R ²). * Statistical significance at P < .05. BMI indicates body mass index; MD score, Mediterranean diet score; mRFEI, Modified Retail Food Environment Index; REGADRS, REasons for Geographic and Racial Differences in Stroke.

The path analysis was also conducted among subgroups stratified by sociodemographic features to test the direct and indirect effects of community food environment on obesity status. The study hypothesis was supported among a subgroup whose annual household income was ≤$75 000 (n = 14 859). Obesity status was measured by 1/sqrt (BMI), transformed from the original BMI values as suggested by Box–Cox transformation (λ = −.5). The model was a good fit, as indicated by the fit indices (χ² = .0000, GFI = 1.000, RMR = 0.0000, SRMR = 0.0000). The result showed that besides the significant direct effects of access to healthy food outlets (β = .03, P < .0001) and adherence to MD (β = .07, P < .0001) on obesity status, access to healthful food outlets had a significant indirect effect on obesity status through consuming MD among this subpopulation (β = −.02, P = .0391). Specifically, 1 unit increase in mRFEI resulted in 0.02 unit decrease in MD score and 0.07 unit increase in 1/sqrt (BMI) (Figure 2). Among other subgroups, no significant indirect effect of access to healthy food outlets on BMI through adherence to MD was found.

Figure 2.

Path analytic model of the effects of community food environment on MD adherence and obesity among a subgroup of REGADRS study participants whose annual household incomes were less than $75 000 (n = 14,859). Box–Cox transformation was used to transform original BMI to meet the multivariate normality assumption. Values shown are standardized path coefficients (β) and explained variations (R ²). * Statistical significance at P < .05. BMI indicates body mass index; MD score, Mediterranean diet score; mRFEI, Modified Retail Food Environment Index; REGADRS, REasons for Geographic and Racial Differences in Stroke.

Discussion

Summary

This study examined the hypothesis that in addition to a direct association, food environment is indirectly associated with obesity through consuming MD. The results of the path analysis among the whole sample showed that community food environment had a significant, inverse relationship with obesity. Specifically, greater access to healthy food outlets (eg, supermarkets) was related to lower BMI. Although the findings from previous studies were inconsistent, the result from this study supported the significant contribution of community food environment to obesity among US adults.⁷ The results also showed that MD adherence had a significant direct and inverse relation with BMI, which aligns with previous studies findings that MD has a protective effect against obesity.³ These findings provide evidence to support that promoting access to healthy food outlets and MD adherence could be effective approaches to reduce obesity among the US adult population.

We did not find evidence that consuming MD mediates the effect of community food environment exposure on obesity status among the whole sample population. One potential explanation for this nonsignificant finding is that factors rather than individual’s dietary behavior (eg, consuming MD) may play important roles in the relationship between food environment exposure and obesity status. Such potential factors could be, for instance, food-related belief, preference, and culture; certain biological and sociopsychological factors (eg, aging, self-efficacy, stress, anxiety); and structural factors (eg, car ownership, access to public transportation, neighborhood walkability).^17

-20 Future studies need to explore factors that reveal the mechanisms underlying between food environment exposure and obesity status among US adults.

Examining the hypotheses among the sociodemographic subgroups, we found that the hypothesis was held among a subgroup whose annual household income was ≤$75 000. Besides significant direct, inverse effect, the community food environment exposure had a significant indirect effect on obesity through consuming MD among this subpopulation. Interestingly, the result showed that greater access to healthy food outlets was related to lower adherence to MD and higher BMI among this population. One potential explanation for this finding is that this segment of population has such a strong preference for a typical “Western diet”—a diet that is mainly characterized by more frequent intakes of red/processed meat, saturated/trans fat, refined grains, sugar, beer, and spirits and less frequent intakes of vegetables, fruit, cooking/dressing oil, cereals and legumes, whole grains, rice/pasta, fish, low-fat dairy, poultry, and water, the features of which are almost opposite of those of the MD dietary pattern—that even given better healthy food access they tend to continue to make suboptimal decisions.^2,21,22 In this study, greater access to healthy foods means a greater access to healthy food retailers such as supermarkets, larger grocery stores, and supercenters. Although these types of stores are typically considered to offer healthy foods, they also carry numerous less healthy foods. Given the preference for typical “Western diet,” it is no wonder that facing more food choices, individuals tends to maintain their preferred dietary pattern, consuming more typical “Western diet” foods and shifting away from the MD dietary pattern, which in turn may increase body weight.²³ Moreover, the MD requires a higher number of daily servings of fruits, vegetables, and in leaner protein sources, all these food categories may be more expensive compared to other processed or convenient food options.²⁴ The published literature has clearly reported that food choice is highly impacted by cost, taste, and convenience.^25,26 This finding among this segment of population suggests that future obesity prevention interventions targeting low- and middle-income population should not simply seek to increase the availability of healthy food outlets in the neighborhoods, but instead efforts should be made to understand how individuals respond to their food environment and structure the food environment to make it easier to shift their current dietary patterns toward healthier patterns (eg, MD). For instance, interventions, such as providing point-of-choice nutrition information, increasing the variety and convenient access to healthy foods, and decreasing prices for healthy foods (eg, fruits and vegetables), could be implemented to increase customers’ awareness and stimulate demands for healthier food products.²⁷ Future studies are needed to support the findings from the current study and test possible hypotheses.

Limitations

First, although the use of a composite measure, like mRFEI, to characterize the food environment have many strengths (eg, capturing the complexity of food environment, reducing data amount), mRFEI may not accurately represent individuals’ actual healthy food access experience. The mRFEI calculation was based on the classification of types of food outlets, instead of the actual in-store foods availability or the actual purchase and use of foods among the participants. Moreover, it did not include the food shopping habits (eg, transportation, car ownership, grocery shopping on the way home from work) in the index calculation. According to findings from the USDA’s National Household Food Acquisition and Purchase Survey, the average household primarily shops for groceries at a store 3.79 miles from home.²⁸ Therefore, food outlets far away from the individual’s residence, even in the same census tract, may not affect their daily grocery shopping experience, especially for those who do not have car or access to public transportation. Furthermore, the mRFEI was developed using secondary data from private companies.¹² Previous studies that validated commonly used secondary retail food outlets data indicated concerns regarding the accuracy of such data sources. For instance, a study comparing retail food outlets data from Dun & Bradstreet, Inc; InfoUSA, Inc, and the South Carolina Department of Health and Environmental Control to field census food outlets in 8 countries in South Carolina reported that field census identified 26% more outlets than the 3 secondary data sources, the sensitivities were fair to moderate (55%-68%), the positive predictive values were moderate (78%-89%), and the geospatial accuracy was moderate with over 80% of outlets geocoded to the correct US census tract.²⁹

Second, although this study is among the first few that use the path analysis to extend current understanding of the mechanism underlying between the obesogenic food environment exposure and obesity-related health outcomes, some limitations of path analysis are still worth to be noted. Path analysis only examines linear, 1-way relationship between the variables, so the alternative possibilities of the complex and reciprocal relationship between the food environment, MD consumption, and obesity were ruled out in the present analysis. Moreover, the transformation of the original BMI to meet certain analysis assumptions (eg, multivariate normality) made the estimates harder to interpret. Third, the cross-sectional design of this study precludes establishing causal inference. Experimental and longitudinal studies to establish causal inference are needed to extend the findings from the current study. Forth, although the relatively large sample from the national REGARDS study allows us to establish precise estimates, the generalizability of the study may be limited. The participants in the analysis were mid- to older age non-Hispanic white and black adults and the data were collected during 2003 to 2007, so that the findings of the study may not apply to other age and ethnic/racial groups, and future study is needed to confirm the consistency of the findings from the present study.

Significance

The findings from the present study suggest that increasing healthy food access and promoting MD adherence could be effective approaches to combat the obesity crisis in the United States. Moreover, it emphasizes the importance of developing population-tailored interventions/policies in future obesity prevention programs.

So What?

What is already known on this topic?

The Mediterranean diet has been increasingly recognized for its protective effects against obesity and recently recommended as a healthy diet pattern for Americans. As the Mediterranean diet is a relatively new dietary pattern in the United States, its role in combating obesity within the context of the prevailing food environment in the United States remains unknown.

What does this article add?

This study examined the mediating role of reported consumption of a Mediterranean diet pattern in relation to community food environment and obesity status among non-Hispanic black and white older adults. Reported consumption of a Mediterranean diet pattern mediated the relationship between community food environment and obesity status among this elderly population with an annual household income of <$75 000.

What is the implication for health promotion practice or research?

This study suggests that future obesity prevention programs should develop population-tailored interventions/policies to modify food environment and promote Mediterranean diet adherence, in order to effectively combat the obesity crisis in the United States.

Footnotes

Acknowledgments

The authors thank the investigators, the staff, and the participants of the REGARDS study for their valuable contributions.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Mid-South Transdisciplinary Collaborative Center for Health Disparities Research (U54MD008176) and the National Institute of Neurological Disorders and Stroke, National Institutes of Health, Department of Health and Human Service (U01NS041588).

ORCID iD

Meifang Chen, PhD, MPH, MCHES

References

Centers for Disease Control and Prevention. Adult Obesity Facts. Overweight and Obesity, Data and Statistics. 2015. http://www.cdc.gov/obesity/data/adult.html. Accessed August 13, 2018.

US Department of Health and Human Services, US Department of Agriculture Economic Research Service. Shifts needed to align with healthy eating patterns: current eating patterns in the United States. In: US Department of Agriculture and the US Department of Health and Human Services, eds. 2015–2020 Dietary Guidelines for Americans. 8th ed. Chap 2, 2015.

Bendall

Mayr

Opie

Bes-Rastrollo

Itsiopoulos

Thomas

. Central obesity and the Mediterranean diet: a systematic review of intervention trials. Crit Rev Food Sci Nutr. 2018;58(18):3070–3084.

Díez

Gullón

Sandín Vázquez

, et al. A community-driven approach to generate urban policy recommendations for obesity prevention. Int J Environ Res Public Health. 2018;15(4):635.

Hawkes

Jewell

Allen

. A food policy package for healthy diets and the prevention of obesity and diet-related non-communicable diseases: the NOURISHING framework. Obes Rev. 2013;14(suppl 2):159–168.

Giskes

van Lenthe

Avendano-Pabon

Brug

. A systematic review of environmental factors and obesogenic dietary intakes among adults: are we getting closer to understanding obesogenic environments? Obes Rev. 2011;12(5):e95–e106.

Cobb

Appel

Franco

Jones-Smith

Nur

Anderson

. The relationship of the local food environment with obesity: a systematic review of methods, study quality, and results. Obesity (Silver Spring). 2015;23(7):1331–1344.

Clary

Lewis

Flint

Smith

Kestens

Cummins

. The local food environment and fruit and vegetable intake: a geographically weighted regression approach in the ORiEL study. Am J Epidemiol. 2016;184(11):837–846.

Micha

Karageorgou

Bakogianni

, et al. Effectiveness of school food environment policies on children’s dietary behaviors: a systematic review and meta-analysis. PLoS One. 2018;13(3):e0194555.

10.

Pacheco

Balam

Archibald

Grant

Skafida

. Exploring the relationship between local food environments and obesity in UK, Ireland, Australia and New Zealand: a systematic review protocol. BMJ Open. 2018;8(2):e018701.

11.

Howard

Cushman

Pulley

, et al. The Reasons for Geographic and Racial Differences in Stroke study: objectives and design. Neuroepidemiology. 2005;25(3):135–143.

12.

Centers for Disease Control and Prevention (CDC). Census Tract Level State Maps of the Modified Retail Food Environment Index (mRFEI). Atlanta GA: CDC. Division of Nutrition, Physical Activity, and Obesity; 2011.

13.

Trichopoulou

Costacou

Bamia

Trichopoulos

. Adherence to a Mediterranean diet and survival in a Greek population. N Engl J Med. 2003;348(26):2599–2608.

14.

Whalen

Judd

McCullough

Flanders

Hartman

Bostick

. Paleolithic and Mediterranean diet pattern scores are inversely associated with all-cause and cause-specific mortality in adults. J Nutr. 2017;147(4):612–620.

15.

Papadaki

Wood

Sebire

Jago

. Adherence to the Mediterranean diet among employees in South West England: formative research to inform a web-based, work-place nutrition intervention. Prev Med Rep. 2015;2:223–228.

16.

Koyama

Houston

Simonsick

, et al. Association between the Mediterranean diet and cognitive decline in a biracial population. J Gerontol A Biol Sci Med Sci. 2015;70(3):354–359.

17.

Tzanetakou

Katsilambros

Benetos

Mikhailidis

Perrea

. “Is obesity linked to aging?”: adipose tissue and the role of telomeres. Ageing Res Rev. 2012;11(2):220–229.

18.

Durand

Andalib

Dunton

Wolch

Pentz

. A systematic review of built environment factors related to physical activity and obesity risk: implications for smart growth urban planning. Obes Rev. 2011;12(5):e173–e182.

19.

Gariepy

Nitka

Schmitz

. The association between obesity and anxiety disorders in the population: a systematic review and meta-analysis. Int J Obes (Lond). 2010;34(3):407–419.

20.

Lanfer

Knof

Barba

, et al. Taste preferences in association with dietary habits and weight status in European children: results from the IDEFICS study. Int J Obes (Lond). 2012;36(1):27–34.

21.

Adlercreutz

. Western diet and Western diseases: some hormonal and biochemical mechanisms and associations. Scand J Clin Lab Invest Suppl. 1990;201:3–23.

22.

Shakersain

Santoni

Larsson

, et al. Prudent diet may attenuate the adverse effects of Western diet on cognitive decline. Alzheimer Dementia. 2016;12(2):100–109.

23.

Vaughan

Cohen

Ghosh-Dastidar

Hunter

Dubowitz

. Where do food desert residents buy most of their junk food? Supermarkets. Public Health Nutr. 2017;20(14):2608–2616.

24.

Rao

Afshin

Singh

Mozaffarian

. Do healthier foods and diet patterns cost more than less healthy options? A systematic review and meta-analysis. BMJ Open. 2013;3(12):e004277.

25.

Glanz

Basil

Maibach

Goldberg

Snyder

DAN

. Why Americans eat what they do: taste, nutrition, cost, convenience, and weight control concerns as influences on food consumption. J Am Diet Assoc. 1998;98(10):1118–1126.

26.

Aggarwal

Rehm

Monsivais

Drewnowski

. Importance of taste, nutrition, cost and convenience in relation to diet quality: evidence of nutrition resilience among US adults using National Health and Nutrition Examination Survey (NHANES) 2007–2010. Prev Med. 2016;90:184–192.

27.

Glanz

Yaroch

. Strategies for increasing fruit and vegetable intake in grocery stores and communities: policy, pricing, and environmental change. Prev Med. 2004;39(suppl 2):S75–S80.

28.

Ver Ploeg

Mancino

Jessica

Clay

Scharadin

. Where Do Americans Usually Shop for Food and How do They Travel to Get There? Initial Findings From the National Household Food Acquisition and Purchase Survey. EIB–138. Washington, DC: US Department of Agriculture ERS; 2015.

29.

Liese

Colabianchi

Lamichhane

, et al. Validation of 3 food outlet databases: completeness and geospatial accuracy in rural and urban food environments. Am J Epidemiol. 2010;172(11):1324–1333.