Abstract
Background
Obesity, a major risk factor for hypertension, diabetes, and other chronic diseases is influenced by a person’s local environmental setting. Accessibility to supermarkets has been shown to influence nutritional behaviors and obesity rates; however the specific local environmental conditions and behavioral mechanisms at work in this process remain unclear.
Purpose
To determine how individual fruit and vegetable consumption behavior was influenced by a distance decay based gravity model of neighborhood geographic accessibility to supermarkets, across neighborhoods in Los Angeles County, independent of other factors that are known to influence nutritional behaviors.
Methods
A distance decay based accessibility model (gravity model) was specified for a large sample (n=7,514) of urban residents. The associations between their fruit and vegetable consumption patterns and their local accessibility to supermarkets were explored, while controlling for covariates known to influence eating behaviors.
Results
Significant variation in geographic accessibility and nutritional behavior existed by age, gender, race and ethnicity, education, marital status, poverty status, neighborhood safety and knowledge of nutritional guidelines. Logistic regression showed an independent effect of geographic accessibility to supermarkets, even after the inclusion of known controlling factors.
Conclusion
A basic gravity model was an effective predictor of fruit and vegetable consumption in an urban population, setting the stage for inclusion of supply and demand parameters, and the ability to estimate local directions and magnitudes of the factors that contribute to the differential obesity rates found in United States urban areas. This knowledge will facilitate more targeted interventions that can help eliminate health disparities.
Introduction
In recent years there has been a steady rise in obesity rates in the United States. For example, between 2006 and 2007 there was an increase from 19 to 28 states reporting more than a quarter of their adults were obese and obesity rates have continued to rise in nationally through 2009.1 These trends have continued and by 2010, 37% of all adult Americans were obese.2 Obesity has also been linked to many chronic diseases including type 2 diabetes, certain cancers, cardiovascular disease, and hypertension.3,4 In understanding the causes of this obesity epidemic many investigators have researched how dietary behaviors are influenced by the local neighborhood contexts within which an individual lives.5,6 In United States metropolitan areas in particular, the characteristics of the local retail food environment are an important determinant of individual dietary intake and can be critical factors in the regulation of body weight in individuals.7,8
Supermarket accessibility: an important indicator of localized nutritional behaviors
A key neighborhood characteristic that has been associated with greater adherence to recommended dietary practices by individuals is their geographic accessibility to chain supermarkets.9 Supermarkets carry many affordable choices of fruits, vegetables and other products and as such, are important indicators of the robustness of the local food environment in a given locality. In a study of the Atherosclerosis Risk in Communities (ARIC) research cohort, African Americans showed an average increase of 32% in consumption of fruits and vegetables for each additional supermarket located in their census tract of residence and the increased intake of fruit and vegetables was associated with a decreased incidence of obesity.10 In two other studies of a sample of ~2600 residents across 65 neighborhoods in Los Angeles County, greater supermarket accessibility was associated with lower BMI and higher self reported health status.11,12 However, an adequate supply of supermarkets are not located in every neighborhood, and in general, the local food accessibility environment varies along with the socio-demographics of the neighborhood.13 The uneven retail food availability landscape in the United States is the product of historical practices of discrimination and economic disinvestment in urban areas occupied by African Americans and other minorities.14 Urban land use patterns and suburbanization trends associated with increased automobile ownership have created multi-nodal cities with distinct sets of social and spatial relationships governing the distribution of retail food vendors in a given metropolitan environment. This situation severely hampers the availability of affordable and nutritious foods for individuals living in many urban contexts, contributing to entrenched health disparities.15–18
This study addresses these issues by using fruit and vegetable consumption data from the 2002 and 2005 Los Angeles County Health Survey, along with a geographically derived measure of respondent’s spatial accessibility to chain supermarkets to develop a predictive model of the impact of geographic accessibility on nutritional behavior. Our main research hypothesis was that a simple distance decay based spatial accessibility model of proximity to large supermarkets will independently influence fruit and vegetable consumption even after controlling for other relevant individual level factors that are known to affect fruit and vegetable intake. An additional goal of this project was to demonstrate that the incorporating a distance decay method allows us to better quantify the impact of adding supermarkets on individual nutritional behaviors, thus opening the door to better decision making at the policy level.
Understanding the importance of Geographic Accessibility on nutritional behaviors
It has been reported that that the potential for interaction between two locations declines with increasing distance between them and is also associated with the amount of demand and supply (i.e. population vs. stores) at each location.19,20 Individual use patterns of public facilities such as supermarkets have been shown to follow a distance decay pattern, where utilization declines with increasing distance from place of residence, place of work, or some other highly frequented place. Various statistical models of this relationship, often termed “gravity models” are commonly used in transportation planning, trade market analysis, retail location theory, and increasingly in health services research as a method of quantifying the distance decay relationship and including demand/supply parameters.21 Distance decay based gravity models have been demonstrated to approximate underlying functional relationships that describing the aggregate use patterns of facilities of all types.22,23 The rate of declining interaction through geographic space varies by context, and thus these functions require parameters, such as the supply of goods or services, and a distance decay parameter.24 The gravity model is an effective way of measuring “potential accessibility”, which can be defined as the “potential” for an individual living in a given locale to access a given amenity, such as a supermarket produce section. Potential accessibility must be distinguished from actual or “realized” accessibility, yet the two are highly correlated. Measuring potential accessibility is very important as it provides the baseline from which to understand actual accessibility and utilization. We operationalize this within the context of access to chain supermarkets and fruit and vegetable consumption within a large metropolitan population.
Previous literature on local environments and nutritional behavior
Larsen et al, presented a comprehensive review of the United States literature on neighborhood food environments, dietary behavior and obesity conducted between 1985 and early 2008.25 This review showed that the strongest evidence exists for the positive effects of greater accessibility to supermarkets on nutritional behaviors and outcomes. The evidence for independent effects of small markets and restaurants was more mixed, with inconsistent findings, yet some evidence that neighborhood accessibility to fast food restaurants having a negative effect on dietary behaviors and obesity. The Larsen review also identified a number of common limitations that limit the interpretability of the results and impair cross study comparison. The major limitations identified were 1) the complexity in defining a relevant neighborhood, 2) the cross sectional, observational nature of most research designs, 3) reliance on commercially (e.g. Dunn and Bradstreet, InfoUSA) or publicly availably (telephone directories, corporate websites) data sources on food outlets. 3) Lack of "ground truthing" and reliable in-store observation of product availability 4) GIS/Geocoding related errors 5) Poor conceptualization of neighborhood boundaries and distances travelled to shop for food. 6) Few longitudinal and/or multilevel studies 7) The need to consider co-varying characteristics of neighborhoods such as population size, urbanization, region and commercialization. 8) Need for studies that consider the entire food availability context (both food stores and restaurants) together.
Other recent national level studies have found 1) persistent disparities in the distribution of food resources by neighborhood race and income across the U.S.26 and 2) associations between geographic accessibility to supermarkets and fruit and vegetable consumption nationally.27 Both of these studies noted, however that there appeared to be differences in the strength and significance of the relationships based upon the type of neighborhood (rural, vs. suburban vs. urban). Finally, some recent studies have shown lack of association between local food environments, nutritional behavior and obesity28,29 but these studies used questionable methods,and/or had limited sample sizes and remain as outliers within the larger body of literature on food accessibility.
In this paper, we contribute to problem areas 1, 3, 4, and 5 identified by Larsen above, including suggesting a straightforward distance decay based method of establishing a robust score for an individual's residential and/or workplace locations that can captures their accessibility to key features in the environment. This distance decay base approach to modeling the local effects on behaviors and outcomes is well established in diverse fields (business analytics for example), but has yet to become widely utilized in epidemiological, health disparities and health outcomes research. We also diverge from much of the existing work by making use of a comprehensive local administrative database of food vendors (LA County Dept of Environmental Health) to explore the influence of geographic accessibility to one specific class of large food vendors, chain supermarkets, upon fruit and vegetable among consumption by adult residents in a large urban setting. Much of the existing work has relied upon commercially available databases.
In this projects supermarket accessibility is defined by geographic access to multi-site grocery stores (supermarket chains operating two or more stores), an more expansive category of chain supermarkets, that also captures full service local grocery stores that operate on a smaller scale than the large regional chain supermarkets, yet whose stores are still likely to carry similar products lines as the major chain supermarkets. Other stores that may carry limited food products, such as (CVS, Walgreens, Target, Walmart etc…) were excluded unless they had a full service Supermarket. Below the design and implementation of the research are presented.
Study Design, Data Sources and Analytical Methods
Given the importance of multi-site grocery store accessibility in shaping the local retail food environment in large U.S. metropolitan areas, a research strategy was designed: 1) to construct a basic gravity model that captured potential accessibility of Los Angeles County Health Survey respondents to multi-site grocery stores. 2) to estimate the relationships between our gravity model derived accessibility to multi-site grocery stores and self reported fruit and vegetable intake in a large sample (n=7,514) of 2002 and 2005 LA County Health Survey respondents, after controlling for the effects of other individual factors known to influence dietary behaviors.
Spatial Contextualization of the LA Health Survey
The Los Angeles County Adult Health Survey is a single-stage, geographically stratified, equal probability sample conducted every two – three years via random digit dialing of each telephone exchange area in the county. Designed to be representative of the entire adult population of Los Angeles County, the LACHS is one of the largest population base health surveys conducted in the United States, and contains self reported data on health related behaviors and outcomes for LA County residents. In order to better understand the role of accessibility to chain grocers on individual dietary habits, the nearest cross streets of the survey respondents for the 2002 and 2005 waves of the LACHS were obtained. We then implemented a logistic behavioral model to assess the independent influence of grocery store accessibility on fruit and vegetable consumption in a large population-based sample, the LA County Health Survey (LACHS),30 using the basic form of the function for calculating gravity-based accessibility:
Ai = ∑j(sj/dijβ)
Where; Ai is the spatial accessibility from population point i, (nearest cross street to the LA health survey respondent's home), Sj is the service capacity at chain grocery store location j, and d is distance between the respondent's nearest cross street and the chain grocery store location, weighted by β, a distance decay exponent.
In the absence of data on supermarket store size and product offering, the value of Sj was set equally for all the chain supermarkets to 1, and the distance decay Beta was also set to 1, yielding a simple arithmetic distance decay slope as shown in Figure 1.
Figure 1.
Arithmetic distance decay function used to weight supermarkets by distance to respondent's nearest cross-street
ArcGIS 9.3.1 and ArcINFO 7.2.1 geographic information systems software were used to calculate all distance matrices and to generate the individual respondent’s accessibility scores.
Existing literature indicated that the more geographically accessible ones preferred place(s) to shop for fruits and vegetables is, the greater one's fruit and vegetable consumption, with individuals living at distances greater than 5 miles from their preferred place to shop consuming significantly less fruit and vegetables than living in areas with a supermarket within 1 mile.31 Based on this limited but useful available knowledge on distance effects on fruit and vegetable consumption, as well as pilot empirical testing of an arithmetic vs. exponential decay parameter, an arithmetic distance decay (power) function was found to be most appropriate for this study.
Outcome Variable – approaching or exceeding USDA fruit and vegetable consumption guidelines
The outcome variable utilized in this study was the probability of a respondent either approaching or exceeding USDA recommended daily dietary guidelines. During the questionnaire LA Health survey respondents were asked about how many servings of fruits and vegetables they ate in the previous day. Current USDA guidelines for adults recommend eating five or more servings of fruits and vegetables daily. We used this guideline as the initial threshold for our dichotomous variable of fruit and vegetable consumption, and worked down to 3 servings.
Predictor variable – Gravity derived geographic accessibility to multi-site grocery stores
The source of multi-site supermarket locations was a longitudinal food licensure file that is maintained in an ongoing collaboration with the LA County Department of Public Health - Environmental Health Division, who shared their longitudinal database of food licenses by location with the Charles Drew Medical GIS lab. This database was used to extract all multi-site grocery stores as defined as having two or more stores operating in Los Angeles County in 2002 and 2005, the years of the study. Spatial accessibility scores based on the inverse weighted Euclidean distances from the nearest cross street of residence to all multi-site grocery stores in the county were generated for each of the 2002 and 2005 LA Health Survey respondents. The score was constructed using the basic gravity model formula detailed above. Respondents living in the cities of Pasadena and Long Beach were excluded from the analysis as food license data for those cities are not reported to LA County Department of Public Health. The outcome measure and the independent variables are derived directly from the survey. The sample was limited to only those survey respondents who had nearest cross street of residence (n=7,514). leaving out those whose census tract centroid of residence only was known.
Controls and Confounding variables
Age, gender, education, income, race/ethnicity, neighborhood safety and knowledge of the USDA dietary guidelines have been shown to be associated with neighborhood environment and physical proximity to healthy food sources.32 To account for these associations, the individual survey respondents socio-demographic characteristics were included in the model, along with answers to 1) a question regarding how safe respondents felt in their local neighborhood, and 2) a question about their knowledge of the importance of nutritional guidelines. Neighborhood poverty levels were disaggregated down to the individual level to control for neighborhood effects associated with high poverty neighborhoods using data from the 2000 census.
Statistical Analysis
Statistical analysis was performed using Stata 11.0 (Stata Corporation, College Station, Tx). Analysis of variance and chi-square tests were used for comparisons of means and proportions. Multivariate logistic regression analyses were used to examine the relationship between accessibility score and nutritional behavior (eating >= 3 serving of fruit and vegetables a day or not). All logistic regression models were adjusted for age, race or ethnicity, sex, marital status, education, poverty level, neighborhood safety and USDA guideline. Odds ratios (OR) and 95% confidence intervals (95% CI) and marginal effects were used to report associations obtained from the multivariate logistic regression models. Possible multiplicative interactions between meeting USDA guideline, accessibility score, and education were investigated. All reported P values were 2-tailed, and statistical significance was defined at the 0.05 level.
Results
There were pronounced differences in both nutritional behavior and in geographic accessibility observable in the sample. Analysis of Variance indicated that statistically significant variation in geographic accessibility to supermarket existed by age, race and ethnicity, education, marital status, poverty status, neighborhood safety and knowledge of nutritional guidelines. Chi Squared tests showed significant differences between those whose diets approached or exceeded the USDA fruit and vegetable intake guidelines and those who did not for the same variables as above. In addition gender also exhibited significant differences in the Chi Squared tests. These descriptive and bivariate relationships are reported in Table 1.
Table 1.
Descriptive factors associated with geographic accessibility and fruit and vegetable consumption
Sample Characteristics, Fruit and Vegetable Consumption
| Patient characteristics | Overall Sample (N= 7514) |
< 4 Servings of Fruit & vegetables/day N (%) |
>= 4 Servings of Fruit & Vegetables/day N (%) |
|---|---|---|---|
| Gender | |||
| Male | 3413(45.4) | 2612(80.0) | 661 (20.0) |
| Female | 4101(54.6) | 2756 (69.7) | 2028 (30.3) |
| P*** | |||
| Race/Ethnicity | |||
| White | 3222(42.2) | 2099(67.4) | 1014(32.6) |
| Latino | 2687(38.7) | 2064(79.6) | 528(20.4) |
| African American | 658(8.9) | 505(78.8) | 136(21.2) |
| Asian/pacific islander | 762(8.7) | 567(81.0) | 133(19.0) |
| Other | 101(1.5) | 74(75.5) | 24(24.5) |
| P*** | |||
| Age | |||
| >=65 | 920(12.2) | 569(66.5) | 287(33.5) |
| 60–64 | 419(5,6) | 295(74.1) | 103(25.9) |
| 50–59 | 1250(16.6) | 876(72.4) | 334(27.6) |
| 40–49 | 1597(21.3) | 1172(75.8) | 375(24.2) |
| 30–39 | 1835(24.4) | 1345(75.8) | 429(24.2) |
| 25–29 | 736(9.8) | 548(77.2) | 162(22.8) |
| 18–24 | 757(10.1) | 563(77.0) | 168(23.0) |
| P*** | |||
| Education | |||
| College/post Grad/Trade | 4697(62.6) | 3212(70.7) | 1313(29.3) |
| High School | 1583(21.1) | 1198(79.4) | 311(20.6) |
| <High School | 1218(16.3) | 949(82.0) | 208(18.0) |
| P*** | |||
| Marital Status | |||
| Married | 4444(59.3) | 3137(73.5) | 1133(26.5) |
| Not Married | 3047(40.7) | 2219(75.6) | 716(24.4) |
| P* | |||
| Federal Poverty Level | |||
| >= 300 FPL | 3346(44.5) | 2266(69.9) | 977(30.1) |
| 200–299% FPL | 1425(20.0) | 1031(75.3) | 338(24.7) |
| 100–199% FPL | 1473(19.6) | 1112(79.2) | 292(20.8) |
| 0–99% FPL | 1270(16.9) | 959(79.3) | 251(20.7) |
| P*** | |||
| Neighborhood Safety | |||
| Very Safe | 2446(32.8) | 1667(70.9) | 683(29.1) |
| Somewhat Safe | 3661(49.1) | 2652(75.1) | 879(24.9) |
| Unsafe | 1352(18.1) | 1013(78.2) | 283(21.8) |
| P*** | |||
| Total Serving of Fruit & Vegetables you should eat everyday | |||
| Don’t know the Guideline | 3200(48.2) | 2791(89.9) | 314(10.1) |
| Knows the Guideline | 3306(51.8) | 1966(58.3) | 1404(41.7) |
| Accessibility Score | P*** | ||
| Mean(SD) = 1.31(0.37) | |||
| Range =1.71 |
P* <.05,
P** <.01,
P*** <.001
chi-square tests for the comparison of ≤2 servings per day group with the ≥3 servings per day group
African Americans and Latinos, poorer populations, and less educated populations tended to have higher overall accessibility scores, due to the fact that the wealthier, predominantly White and Asian suburban areas having lower population densities and more open space. However, this greater overall accessibility did not mitigate the racial and ethnic effects. Despite the overarching trends in accessibility however, there were significant differences in nutritional behavior for the relevant individual characteristics that are available in L.A. Health survey. Whites, females, older people, more educated people, wealthier people, and people who live in safer neighborhoods all are more likely to have reported approaching or exceeding the USDA guidelines. Based on these findings the analysis was extended to consider predictive capability of a gravity model derived accessibility score when accounting for variation in the individual factors. Table 2 reports the results of a logistic regression analysis with gravity modeled accessibility to multi-site grocery stores as the predictor variable.
Table 2.
Fruit and vegetable consumption and its association with geographic accessibility, while controlling for known covariates
Logistic regression analysis [OR (95% CI) for Adults meeting USDA fruit and vegetable guideline by Accessibility to grocery store and socio demographic characteristics
| Variable | Beta | OR | 95% CI | P-Value |
|---|---|---|---|---|
| Accessibility Score(Marginal Effect =.031) | .0.181 | 1.20 | 1.00 – 1.43 | 0.040 |
| Gender | ||||
| Male | referent | |||
| Female | 0.268 | 1.31 | 1.15 – 1.49 | <0.0001 |
| Race/Ethnicity | ||||
| White | referent | |||
| Hispanic/Latino | −0.246 | 0.78 | 0.65 – 1.01 | 0.007 |
| African-American | −0.371 | 0.69 | 0.58 – 0.80 | 0.002 |
| Asian/Pacific Islander | − 0.195 | 0.82 | 0.70 – 0.98 | 0.104 |
| Other | −0.208 | 0.81 | 0.62 – 1.24 | 0.437 |
| Age | ||||
| >=65 | Referent | |||
| 60–64 | −0.395 | 0.67 | 0.49– 0.92 | 0.013 |
| 50–59 | −0.468 | 0.63 | 0.50 – 0.78 | <0.0001 |
| 40–49 | −0.682 | 0.51 | 0.41 – 0.63 | <0.0001 |
| 30–39 | −0.686 | 0.50 | 0.30 – 0.63 | <0.0001 |
| 25–29 | −0.705 | 0.49 | 0.37 – 0.65 | <0.0001 |
| 18–24 | −0.684 | 0.52 | 0.40 – 0.70 | <0.0001 |
| Education | ||||
| College/post Grad/Trade | referent | |||
| High School | −0.269 | 0.76 | 0.64 – 0.91 | 0.002 |
| <High School | −0.306 | 0.74 | 0.58 – 0.93 | 0.010 |
| Marital Status | ||||
| Married | referent | |||
| Single | −0.191 | 0.83 | 0.74 – 0.89 | 0.004 |
| Federal Poverty Level | ||||
| >= 300 FPL | Referent | |||
| 200–299% FPL | −0.104 | −0.90 | 0.76 – 1.08 | 0.232 |
| 100–199% FPL | −0.049 | 0.95 | 0.78 – 1.16 | 0.628 |
| 0–99% FPL | −0.001 | 1.00 | 081.– 1.09 | 0.995 |
| Neighborhood Safety | ||||
| Very Safe | referent | |||
| Somewhat Safe | −0.082 | 0.92 | 0.80 – 1.06 | 0.254 |
| Unsafe | −0.164 | 0.84 | 0.69 – 1.04 | 0.122 |
| Total Serving of Fruit & Vegetables you should eat everyday | ||||
| Don’t know the Guideline | referent | |||
| Knows the Guideline | 1.789 | 6.0 | 5.20 – 6.90 | <0.0001 |
| PCT 100 POV | 0.001 | 0.99 | −0.03 – 0.03 | 0.968 |
| PCT 200 POV | 0.002 | 1.00 | −0.98 – 1.02 | 0.984 |
CI, confidence interval; OR, odds Ratio
A simple distance decay based accessibility score was able to predict fruit and vegetable intake of 4 or more servings per day in our sample of over 7,514 individuals from the L.A. Health survey. Although not signifigant in predicting 5 or more fruits and vegetables (achieved by only 14% of the sample), our model did predict changes in consumption of 4 or more servings, a slight relaxation of USDA standards that was achieved by 25% of the sample.. This relationship persisted even when race, age, education, marital status, poverty status, neighborhood safety, and knowledge of USDA guidelines were controlled for. Despite one's individual characteristics, geographic accessibility to multi-site grocery markets, as defined by a very basic gravity model is associated with increased fruit and vegetables intake in the Los Angeles County population that the LA County Health Survey is designed to represent.
Interpreting the marginal effect of the geographic accessibility score
The marginal effect of the main predictor variable, a basic geographic accessibility score to chain supermarkets was .031. This means that adding a store adjacent (100 feet) to the nearest cross street increases the odds of a respondant eating four or more fruits and vegetables by three percent. This allows us to quantify the geographic relationships described in our model. Our accessibility score was a log10 of the actual score and ranged from ~ .03 to 3.0. We unitized our score into 100 foot interval thus we used ∑ 100/di where di is the distance to each store in feet. The average accessibility score for the cross streets reported by individuals in our sample is 1.307. Accordingly if we added a hypothetical 10 additional stores all at 1 mile (5280 feet) from the average individual, then this would add 100 × 10/5280 = 0.19 to the score or about 15%. This translates into about a 15% additional increase in the odds of eating 4 or more servings from 1.031 to about 1.035.
Discussion
The validity of a basic gravity model in understanding retail market shopping behaviors has been well established outside of the health and nutrition literature, and is one of the dominant paradigms in the economics of business location. Given that food consumption in most large American metropolitan areas is driven by retail food markets, it follows that gravity modeling as an approach will be helpful in untangling the complex relationships between the built environment, energy balance related behaviors, obesity and chronic disease outcomes. This study was limited in scope, and did not address restaurants, small markets and other factors that may contribute to fruit and vegetable intake. Also the data on nutritional behavior was self reported, and is subject to inaccuracies and biases related to measurement error. Another limitation is the lack of complexity in the gravity model, rooted in the assumption that all multi- site grocery stores exhibit the same influence on their surroundings. This implicit assumption of an unlimited supply of groceries would be more problematic were our observations not chain supermarkets. However, the high volumes and complex multinational food markets that this class of store are able to tap into guarantees a near limitless supply of food types, if not each individual food products themselves. The lack of supply side data for each grocery store is being addressed by this team as part of ongoing research projects, and these results will be reported in the future. Finally, we had no information on individuals work place locations, or other places they might frequent, thus our model is based solely on place of residence.
Much of the previous research on the association between retail food availability and nutritional behavior has relied on aggregated administrative units such as zip codes or census tracts to capture environmental and geographical relationships. In this project each respondent’s local nutritional environment was modeled based upon real world data obtained from local government food sale licensure records, using a simplified form of a standard gravity function that has been shown to reliably explain many retail behavioral interactions, including food consumption. This permitted us to quantify the geographic relationship observed, using hypothetical examples.
There is promising space for further development of gravity style distance decay based accessibility model used in this research. Supply side parameters can be constructed via in-store surveys of product type and amount. This would likely strengthen the explanatory power of the model and provide new insight on the nature and magnitude of the relationships between local environmental contexts and individual nutritional behavior. Additionally, although a simple arithmetic gravity function was found to have predictive capability in this study, there is a still a need for further calibration and understanding of the distance parameter. This study found that distance for accessibility to grocery stores Beta = 1 (1/x1) an arithmetic Decay (power) function, has explanatory power and that Beta = 2 (1/x2), an exponential distance decay function does not. This work cannot, however, establish that a basic decay function of (1/x1) is the the most appropriate decay function until further model calibration can be completed. Research projects are underway by the Authors that will permit the inclusion of empirically observed supply and demand parameters, as well as model calibration in the Los Angeles county context. With an optimized local accessibility model, it will be possible to delve deeper into questions of what the differential behavioral impacts of local geographies are on specific age, gender, race, or income groups nutritional behaviors. Ultimately, the goal is to provide information on the critical interventional entry points that move beyond the crude targeting of entire age/gender/racial groups or entire classes of neighborhoods, and instead focus in on the key vulnerable groups who are likely to be most impacted by living in compromised states of accessibility to key amenities such as supermarkets. This knowledge can guide intervention efforts designed to encourage healthier behaviors. By understanding how local environments and individual characteristics interact to influence behavior related to obesity and its related chronic conditions, a "roadmap" to intervention can be developed that will help chart the path to the translation of medical knowledge into communities and the eventual elimination of health disparities.
Acknowledgments
Support for this research project was provided by grants:
1R03HL088622-01 from the NIH - National Heart Lung and Blood Institute
1SC3GM087224-01 from the NIH - National Institute of General Medical Sciences
U54RR026138-01 from the National Center for Research Resources
We also thank Daniel Ryan, Deyu Pan, and Cambria Broadway for their valuable research assistance.
Footnotes
No financial disclosures were reported by the authors of this paper.
References
- 1.Trust for America's Future. Robert Wood Johnson Foundation; 2010. F as in Fat: How Obesity Threatens America's Future 2010. [Google Scholar]
- 2.Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of obesity in the United States, 2009–2010. Hyattsville, MD: National Center for Health Statistics; 2012. NCHS data brief, no 82. [Google Scholar]
- 3.Wang Y, Chen X, Song Y, Caballero B, et al. Association between Obesity and Kidney Disease: A Systematic Review and Meta-analysis Kidney Int. 2008;1(73):19–33. doi: 10.1038/sj.ki.5002586. [DOI] [PubMed] [Google Scholar]
- 4.Weight Information Network. NIH/NIDDK; 2007. Do You Know the Health Risks of being Overweight? [Google Scholar]
- 5.Zick CD, Smith KR, Fan JX, Brown BB, Yamada I, Kowaleski-Jones L. Running to the Store? The relationship between neighborhood environments and the risk of obesity. Soc Sci Med. 2009;10(69):1493–1500. doi: 10.1016/j.socscimed.2009.08.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wang MC, Cubbin C, Ahn D, Winkleby MA. Changes in neighbourhood food store environment, food behaviour and body mass index. Public Health Nutr. 1981–1990;9(11):963–970. doi: 10.1017/S136898000700105X. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gittelsohn J, Song HJ, Suratkar S, Kumar MB, Henry EG, Sharma S, et al. An Urban Food Store Intervention Positively Affects Food-Related Psychosocial Variables and Food Behaviors. Health Educ Behav. 2010;3(37):390–402. doi: 10.1177/1090198109343886. [DOI] [PubMed] [Google Scholar]
- 8.Ford PB, Dzewaltowski DA. Disparities in obesity prevalence due to variation in the retail food environment: three testable hypotheses. Nutr Rev. 2008;4(66):216–228. doi: 10.1111/j.1753-4887.2008.00026.x. [DOI] [PubMed] [Google Scholar]
- 9.Laraia BA, Siega-Ruiz AM, Kaufman JS, Jones SJ. Proximity of supermarkets is positively associated with diet quality index for pregnancy. Prev Med. 2004;5(39):869–875. doi: 10.1016/j.ypmed.2004.03.018. [DOI] [PubMed] [Google Scholar]
- 10.Morland K, Wing S, Diez-Roux A. The contextual effect of the local food environment on residents' diets: the atherosclerosis risk in communities study. Am J Public Health. 2002;11(92):1761–1767. doi: 10.2105/ajph.92.11.1761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Brown AF, Vargas RB, Ang A, Pebley A. The Neighborhood Food Resource Environment and the Health of Residents with Chronic Conditions. J Gen Intern Med. 2008;8(23):1137–1144. doi: 10.1007/s11606-008-0601-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Inagami S, Cohen DA, Finch B, Asch S. You are Where You Shop: Grocery Store Locations, Weight, and Neighborhoods. Am J Prev Med. 2006;1(31):10–17. doi: 10.1016/j.amepre.2006.03.019. [DOI] [PubMed] [Google Scholar]
- 13.Larson NI, Story MT, Nelson MC. Neighborhood environments: disparities in access to healthy foods in the U.S. Am J Prev Med. 2009;36(1):74–81. doi: 10.1016/j.amepre.2008.09.025. [DOI] [PubMed] [Google Scholar]
- 14.Galvez MP, Morland K, Raines C, Kobil J, Siskind J, Godbold J, et al. Race and food store availability in an inner-city neighbourhood. Public Health Nutr. 2008;6(11):624–631. doi: 10.1017/S1368980007001097. [DOI] [PubMed] [Google Scholar]
- 15.Bovell-Benjamin AC, Hathom CS, Ibrahim S, Gichuhi PN, Bromfield EM. Healthy food choices and physical activity opportunities in two contrasting Alabama cities. Health Place. 2009;2(15):429–438. doi: 10.1016/j.healthplace.2008.08.001. [DOI] [PubMed] [Google Scholar]
- 16.Odoms-Young AM, Zenk S, Mason M. Measuring food availability and access in African-American communities: implications for intervention and policy. Am J Prev Med. 2009;4(36):S145–S150. doi: 10.1016/j.amepre.2009.01.001. [DOI] [PubMed] [Google Scholar]
- 17.Powell LM, Slater S, Mirtcheva D, Bao Y, Chaloupka FJ. Food store availability and neighborhood characteristics in the United States. Prev Med. 2007;3(44):189–195. doi: 10.1016/j.ypmed.2006.08.008. [DOI] [PubMed] [Google Scholar]
- 18.Morland K, Filomena S. Disparities in the availability of fruits and vegetables between racially segregated urban neighbourhoods. Public Health Nutr. 2007;10(12):1481–1489. doi: 10.1017/S1368980007000079. [DOI] [PubMed] [Google Scholar]
- 19.Reilly WJ. Methods of the Study of Retail Relationships. University of Texas Bulletin 2944; 1929. pp. 1–50. [Google Scholar]
- 20.Stewart JQ. Demographic Gravitation: Evidence and Applications. Sociometry. 1948;11(1/2):31–58. [Google Scholar]
- 21.Isard W. Location and Space Economy: A General Theory Relating to Industrial Location, Market Areas, Land Use, Trade, and Urban Structure. Cambridge: MIT Press/John Wiley & Sons; 1956. [Google Scholar]
- 22.Cervero R. America’s Suburban Centers: The Land-Use Transportation Link. Boston: Unwin Hyman; 1989. [Google Scholar]
- 23.Levinson D, Kumar A. Activity, Travel, and the Allocation of Time. J Am Plann Assoc. 1995;4(61):458–470. [Google Scholar]
- 24.Guagliardo MF. Accessibility of Primary Care: Concepts, Methods, and Challenges. Int J Health Geogr. 2004;1(3) doi: 10.1186/1476-072X-3-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Larson NI, Story MT, Nelson MC. Neighborhood environments: disparities in access to healthy foods in the U.S. Am J Prev Med. 2009;36(1):74–81. doi: 10.1016/j.amepre.2008.09.025. [DOI] [PubMed] [Google Scholar]
- 26.Richardson AS, Boone-Heinonen J, Popkin BM, Gordon-Larsen P. Are neighbourhood food resources distributed inequitably by income and race in the USA? Epidemiological findings across the urban spectrum. BMJ Open. 2012 Apr 13;2(2) doi: 10.1136/bmjopen-2011-000698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Michimi A, Wimberly MC. Associations of supermarket accessibility with obesity and fruit and vegetable consumption in the conterminous United States. Int J Health Geogr. 2010 Oct 8;9:49. doi: 10.1186/1476-072X-9-49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lee H. The role of local food availability in explaining obesity risk among young school-aged children. Social Science & Medicine. Volume 74(Issue 8):1193–1203. doi: 10.1016/j.socscimed.2011.12.036. [DOI] [PubMed] [Google Scholar]
- 29.An R, Sturm R. School and Residential Neighborhood Food Environment and Diet Among California Youth. American journal of preventive medicine. 2012 Feb 1;volume 42(issue 2):129–135. doi: 10.1016/j.amepre.2011.10.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Los Angeles Department of Public Health The LA Health Survey. http://publichealth.lacounty.gov/ha/hasurveyintro.htm.
- 31.Rose D, Richards R. Food store access and household fruit and vegetable use among participants in the US Food Stamp. Program Public Health Nutr. 2004;7(8):1081–1088. doi: 10.1079/PHN2004648. [DOI] [PubMed] [Google Scholar]
- 32.Baker EA, Schootman M, Barnidge E, Kelly C. Role of race and poverty in access to foods that enable individuals to adhere to dietary guidelines. Prev Chronic Dis. 2006;3(3):A76. [PMC free article] [PubMed] [Google Scholar]