- Email: [email protected]
Accessibility Landscapes of Supplemental Nutrition Assistance Program−Authorized Stores
RESEARCH
Original Research
Accessibility Landscapes of Supplemental Nutrition Assistance ProgramAuthorized Stores Elizabeth F. Racine, DrPH, RD; Eric Delmelle, PhD; Elizabeth Major, MA; Corliss A. Solomon, MPH ARTICLE INFORMATION Article history: Submitted 15 March 2017 Accepted 3 November 2017 Available online 22 January 2018
Keywords: Supplemental Nutrition Assistance Program (SNAP) Food access Social determinants Food stamps Geographic Information Systems (GIS) 2212-2672/Copyright ª 2018 by the Academy of Nutrition and Dietetics. https://doi.org/10.1016/j.jand.2017.11.004
ABSTRACT Background The Supplemental Nutrition Assistance Program (SNAP) is the largest food assistance program in the United States. Participants receive electronic benefits that are redeemable at a variety of food stores. Previous research notes that low-income neighborhoods often lack supermarkets with high-quality, affordable food. Objective The first aim of this study was to explore the number and spatial distribution of SNAP stores by type and to assess how SNAP benefit redemption is linked to store type in North Carolina in 2015. The second aim was to compare the demographics of populations living in areas with a high concentration of SNAP participants vs areas with a lower concentration of SNAP participants. The third aim was to test for disparities in the availability of and access to SNAP-authorized stores in areas with high vs low concentration of SNAP participants stratified by rural/urban status. Design US Department of Agriculture and US Census data were used to explore the spatial distribution of SNAP stores at the census block group level utilizing a Geographic Information System. Participants The 9,556 North Carolina SNAP stores in 2015 categorized into full-variety and limited-variety stores. Outcome measures Proximity to limited-variety SNAP food stores and full-variety SNAP food stores within access range (1 mile in urban areas and 10 miles in rural areas). Statistical analyses Wilcoxon rank sum and c2 tests are used to compare the distance to and concentration of SNAP stores by rurality and SNAP participant concentration at census block group scale. Results Among the SNAP stores in North Carolina, 83% are limited-variety stores and 17% are full-variety stores. There are disparities in the demographics of individuals living in census block groups with a high proportion of SNAP participants compared to census block groups with a lower proportion of SNAP participants. More households in higher SNAP participant census block groups were non-white, did not have a car, and had children compared to census block groups with lower SNAP participation. Residents in high SNAP participant census block groups typically had access to 0 full-variety stores and 4 limited-variety stores in urban areas and 3 full-variety stores and 17 limitedvariety stores in rural areas. Conclusions SNAP participant access to a variety of stores should be considered when approving food stores for SNAP authorization. More research is essential to disentangle the relationship between access to SNAP store type and SNAP food choice and to estimate geographical disparities. J Acad Nutr Diet. 2018;118(5):836-848.
The Continuing Professional Education (CPE) quiz for this article is available for free to Academy members through the MyCDRGo app (available for iOS and Android devices) and through www.jandonline.org (click on “CPE” in the menu and then “Academy Journal CPE Articles”). Log in with your Academy of Nutrition and Dietetics or Commission on Dietetic Registration username and password, click “Journal Article Quiz” on the next page, then click the “Additional Journal CPE quizzes” button to view a list of available quizzes. Non-members may take CPE quizzes by sending a request to [email protected]. There is a fee of $45 per quiz (includes quiz and copy of article) for non-member Journal CPE. CPE quizzes are valid for 1 year after the issue date in which the articles are published.
836
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
T
HE SUPPLEMENTAL NUTRITION ASSISTANCE PROgram (SNAP) is the largest food assistance program in the United States, serving more than 44 million lowincome Americans.1 SNAP participants receive electronic benefits that are redeemable at an assortment of food stores, such as grocery stores, convenience stores, and drug stores. In 2016, approximately 19% of American households participated in SNAP, receiving on average $255 per month in benefits, with an annual cost of approximately $75 billion.1 Most (84%) SNAP benefits are redeemed at supermarkets and super stores, while 64% of SNAP transactions take place ª 2018 by the Academy of Nutrition and Dietetics.
RESEARCH at supermarkets and super stores, therefore, about 36% of SNAP transactions take place at other types of SNAP stores, such as smaller grocery stores, and convenience stores.2 While research has explored the relationship between the food environment and obesity among low-income communities, one common finding is that low-income neighborhoods often lack access to supermarkets that offer high-quality affordable food.3 Reduced access to these full-variety supermarkets may increase exposure to energy-dense food and lead to diets high in processed foods, fat, sugar, and sodium.4-7 In addition, low-income communities typically have access to smaller food outlets,8 pay more for shopping locally,9 and have fewer healthy food options compared to higher-income communities.10 The topic of access to food has received considerable attention in the literature, and can be measured from several perspectives, including geographic, economic, and informational.11 Geographic accessibility examines the relationship between the location of the store where food can be purchased and the individual in need. It incorporates structural barriers to food outlets, including transportation resources, travel time and distance, and travel costs.12 Such geographic measures can further be employed to identify areas that are low income with limited access to healthy food retailers, referred to as “food deserts.”13 Recent advances in Geographic Information Systems have facilitated the modeling of geographic measures of food access and food availability.14,15 Three types of methods have been pervasively adopted to estimate food accessibility with Geographic Information Systems: the container method, the buffer method, and the network method.16 Although the first two techniques can be loosely categorized as “exposure” approaches, the latter one, which is used in this article, attempts to determine the ease of access to a food outlet. To date there has been very little research exploring SNAP participants access to SNAP-authorized stores. Wood and Horner17 examined availability of SNAP stores among different types of neighborhoods in Leon County, FL. They found that most low-car-access census block groups (99%) contained a SNAP store accessible within a 20-minute drive of a SNAP store, while less than half (40%) of these census block groups contained a SNAP stores within a 20-minute walk.17 Two other studies, both conducted in urban counties, found that low-income communities with a high percentage of SNAP participants were more likely than higher-income communities to be populated with limitedvariety SNAP-authorized stores and less likely to include full-variety stores.18,19 Limited-variety stores typically lack healthy products like fresh produce and lean, fresh meats.18,19 It is unclear whether this association between SNAPauthorized store type and concentration of residents participating in SNAP is limited to these counties. The first aim of this study was to explore the number and spatial distribution of SNAP stores by type and to assess how SNAP benefit redemption is linked to store type in North Carolina in 2015. The second aim was to compare the demographics of populations living in areas with a high concentration of SNAP participants vs areas with a lower concentration of SNAP participants. The third aim was to test for disparities in the availability of and access to SNAPauthorized stores in areas with a high vs low concentration of SNAP participants stratified by rural/urban status. May 2018 Volume 118 Number 5
MATERIALS AND METHODS Data Two primary data sources are used for this study. First, the US Department of Agriculture (USDA) Food and Nutrition Service provided a geolocated dataset of all SNAP-authorized stores in North Carolina from 2015 (hereafter “USDA data”). This dataset is available by request from the USDA. Second, 2011-2015 data estimates from the American Community Survey (hereafter “Census data”) were used at the census block group level.20 The US Census also publishes populationweighted geographic centers, or centroids, after each decennial census. The 2010 population-weighted centroids are used as points of origin to estimate geographic access for each block group.21,22 This study was exempt from Institutional Review Board review because it is not human subjects research.
USDA Data USDA data includes store type, latitude, longitude, address, and county. The USDA categorizes SNAP-authorized stores by 17 different types, which are summarized in Table 1. When stores apply for SNAP authorization, the store self-selects store type based on instructions provided by USDA. USDA then reviews the store application and, at that time, can change the store type designation to align with the USDA definition. For this study, store types are grouped into binary categorizes: “full variety,” defined as offering a “wide” or “extensive” selection of fresh meat, dairy, produce, and grains (eg, super stores, large grocery stores, and supermarkets) and “limited variety,” defined as not offering a “wide” or “extensive” selection of fresh meat, dairy, produce, and grains (eg, drugstores, convenience stores, small grocery stores, and dollar stores) or all other SNAP-authorized stores (see Table 1).
Census Data The US Census American Community Survey provides 5-year estimates information at the census block group level for the years 2011 to 201520; the unit of analysis for this study. The census block group level was selected to best identify neighborhood characteristics, as census block group is the smallest geographic unit available with demographic information pertinent to this study. The following variables for each census block group in North Carolina are utilized for this analysis: population size, household size, number of households participating in SNAP (recoded into percent of households in the census block group on SNAP), urban/rural status, median household income, number of residents by race/ ethnicity (recoded as percent of census block group that was non-Hispanic black, non-Hispanic white, Hispanic, and other based on the population size), percent of population with access to a vehicle, and percent of households with children.
Geographic Categories To examine whether the SNAP food environment varies according to the concentration of SNAP participants, the North Carolina census block groups are divided into four categories based on the percentage of SNAP participants in each census block group and whether these census block groups are urban or rural. First, to examine the concentration of SNAP participants within the 6,092 census block groups in North Carolina by quartile; those 75th percentile are considered census JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
837
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
USDAb SNAPc Vendor Categoriesd
Binary Groups
Store type
USDA definition
Redemptions, $ Stores, n Groups
Redemption, $ Stores, n
Large grocery store
44,688,127 “A store that carries a wide selection of all four staple food categories. They may sell ineligible items as well, but their primary stock is food items.”
82
Full variety of foods
2,062,944,181
1,614
Supermarket
“Establishments commonly known as supermarkets, food 1,016,738,434 stores, grocery stores and food warehouses primarily engaged in the retail sale of an extensive variety of grocery and other store merchandise.”
1,090
—
—
—
Super store
“Very large supermarkets, “big box” stores, super stores and food warehouses primarily engaged in the retail sale of a wide variety of grocery and other store merchandise.”
1,001,517,620
442
—
—
—
Bakery specialty
“Food stores specializing in the sale of bread/cereal products”
1,586,762
67
Limited variety of foods
Meat/poultry specialty
“Food stores specializing in the sale of meat products”
13,217,395
72
—
—
—
300,091,812
7,942
May 2018 Volume 118 Number 5
Seafood specialty
“Food stores specializing in the sale of seafood products” 4,000,093
81
—
—
—
Fruits/vegetable specialty
458,175 “Food stores specializing in the sale of fruits and/or vegetables that operates in a fixed or semi-permanent location”
43
—
—
—
Combination grocery/other
“Primary business is sale of general merchandise but also 128,983,222 sell a variety of food products. Such stores include independent drug stores, dollar stores, and general stores.”
2,664
—
—
—
Convenience store
92,027,366 “Self-service stores that offer a limited line of convenience items and are typically open long hours to provide easy access for customers.”
4,192
—
—
—
Farmers’ market
119,312 “A single or multi-stall market that sells agricultural products, particularly fresh fruit and vegetables, to the general public at a single or multiple locations.”
110
—
—
—
(continued on next page)
RESEARCH
838
Table 1. Categorization of Supplemental Nutritional Assistance Programauthorized stores, US Department of Agriculture categories, and groupings for a study in North Carolina in 2015a
May 2018 Volume 118 Number 5
Table 1. Categorization of Supplemental Nutritional Assistance Programauthorized stores, US Department of Agriculture categories, and groupings for a study in North Carolina in 2015a (continued) USDAb SNAPc Vendor Categoriesd
Binary Groups
Store type
USDA definition
Direct marketing farmer
“Designation applies to direct marketing farmers; these 124,862 are individual producers of agricultural products, particularly fresh fruit and vegetables, as well as meat, fish, dairy, and/or grains that are sold to the general public through a direct marketing venue such as a roadside farm stand, pick-your own operation, and/or market stall within a farmers’ market.”
174
—
—
—
Medium grocery store
“A store that carries a moderate selection of all four staple 31,388,388 food categories.”
198
—
—
—
Small grocery store
“A store that carries a small selection of all four staple food categories.”
300
—
—
—
Nonprofit food buying cooperative “Any store that operates as a “cooperative”.”
Redemptions, $ Stores, n Groups
20,587,550
Redemption, $ Stores, n
—
—
—
—
—
—
“Designation applies to all retail food entities, located on 5,577,143 military installations that sell food and non-food products”
7
—
—
—
“Statutory/regulatory definition: an establishment that sells eligible food to meal services for resale to households.”
1
—
—
—
“A store that does not have a permanent store location, this includes delivery routes that deliver food at set locations and times, as well as rolling routes.”
Military commissary
Wholesaler
454,098 1,567,446
Redacted
a
The number of stores and redemptions by category are included. Total number of stores is 9,556 and total redemption is $2,363,035,993. USDA¼US Department of Agriculture. c SNAP¼Supplemental Nutrition Assistance Program. d Stores applying for SNAP select the appropriate category, then USDA reviews and may change the category to better align with the definition. The definitions were requested from the Food and Nutrition Services, USDA. North Carolina SNAP stores by type and redemption were provided to the first author by the USDA, Food and Nutrition Service. b
839
RESEARCH
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
9 24
Delivery route
RESEARCH block groups with a high concentration of SNAP participants. At the 75th percentile, 23% of the census block group households participate in SNAP. Hereafter census block groups with a high concentration of SNAP households, those with 23% on SNAP are referred to as “high SNAP block groups,” while census block groups with a lower concentration of SNAP households, those with <23% of households participating in SNAP are referred to as “low SNAP block groups.” Second, census block groups are classified by access to a full-variety SNAP food store using the USDA definition (ie, 1 mile or more from a full-service store in urban areas and 10 miles from a full-service store in rural areas).23 To account for the different definitions by rurality, census block groups were stratified by rural status. The Census Bureau classifies geographies into three categories: urban areas (>50,000 residents), urban cluster areas (between 2,500 and 50,000 residents), and rural areas (any area not meeting the criteria for urban or urban cluster).24 The decennial Census describes how many households in each tract are found within each area.25 The category in which the majority of households were found was used to describe each block group, so a block group with 220 urban households, 630 urban cluster households, and 11 rural households would be coded as an urban cluster block group. For this analysis, urban and urban clusters are defined as urban, with urban cluster census block groups traditionally encompassing suburbs. As a result, the four geographic categories are: rural census block groups with a high concentration of SNAP households, urban census block groups with a high concentration of SNAP households, rural census block groups with a lower concentration of SNAP households, and urban census block groups with a lower concentration of SNAP households.
Measures The following variables are used to examine the SNAP food store environment: comparison of proximity to full-variety vs limited-variety stores; access to one or more SNAP food store by SNAP store type (yes or no) as defined by USDA as a SNAP store located within 1 mile in urban areas and within 10 miles in rural areas23; average distance to accessible SNAP food stores; concentration of SNAP store type within the USDA defined access range; and concentration of SNAP store type within the USDA defined access range by 1,000 population.
Estimating Proximity to SNAP Stores Proximity to SNAP retailers is estimated at the census block group level for 2015. The weighted population centroid of each block group is used as a point of origin to measure the proximity to each SNAP retailer. Following the USDA Food Access Research Atlas,23 only SNAP retailers within 1 and 10 Euclidean miles for urban and rural block groups were retained, respectively. Travel is estimated using the ArcGIS Online Application Programming Interface.26 The travel routing results are then aggregated to each block group using MySQL,27 producing a table of demographic information and shortest distances from the block group’s centroid to each type of store. These results are further aggregated according to SNAP participation levels. The procedure and its results are illustrated for an urban census block group in Figure 1A and a rural census block group in Figure 1B and demonstrate that the use of Euclidean 840
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
distance overestimates the number of available SNAP stores. The distance to all SNAP retailers within 1 and 10 Euclidean miles is estimated; retailers in the blue-shaded regions are within 1 or 10 miles network distance. In the urban scenario (Figure 1A), a total of 5 SNAP retailers of limited variety are available within 1-mile network distance (average distance¼0.70 mile, range¼0.41 to 0.96 mile), and another SNAP store is located within the 1-mile Euclidean distance but outside the 1-mile network distance. The closest available full-variety SNAP store (a supermarket) is located at 1.49 miles from the block group centroid. In the rural scenario (Figure 1B), a total of 14 SNAP retailers are available within a 10-mile network distance (average distance¼6.48 miles, range¼3.58 to 9.06 miles), and three of those are full-variety stores (average distance¼7.45 miles, range¼4.78 to 8.82 miles). Note that 3 full-variety stores and 13 limited-variety stores are located outside the 10-mile network distance, but well within the 10-mile Euclidean distance.
Statistical Analysis Descriptive statistics are presented for all North Carolina census block groups with at least one household (n¼6,092) and for each of the four geographic groups. Wilcoxon ranksum and c2 tests are used to compare the distance to and concentration of SNAP stores by rurality and SNAP participant concentration. Statistical significance is determined at a<.05. Analyses were conducted in SAS software.28
RESULTS In North Carolina, more than $2.3 billion in SNAP redemptions were spent at food stores in 2015; 87% of SNAP redemptions (about $2 billion) were made at full-variety stores and 13% (about $300 million) at limited-variety stores (Table 1). There were 9,556 SNAP stores in North Carolina—17% were full-variety stores and 83% were limitedvariety stores. Approximately 14% of North Carolina households participated in SNAP in 2015 (Table 2). The majority lived in urban areas (65%) and about half live in census block groups with a high concentration of SNAP households (49%). In urban areas, most of the SNAP households (53%) live in census block groups with a high concentration of SNAP households. There are disparities in the demographics of individuals living in high SNAP block groups compared with low SNAP block groups. A greater proportion of non-Hispanic blacks, (47% vs 12%), Hispanics (7% vs 5%), households without a car (16% vs 3%), and households with children (35% vs 30%); and a lower proportion of non-Hispanic whites (30% vs 74%) are located in high SNAP block groups. As expected, in high SNAP block groups, the median household income is significantly lower, approximately $26,000 per year compared to approximately $52,000 per year in low SNAP block groups. In high SNAP block groups about 35% of households participate in SNAP compared to 7% of households living in low SNAP block groups. In rural areas, there is a disparity in the racial and ethnic composition of residents by SNAP household concentration. These differences are statistically significant in all the demographic variables tested, but are particularly divergent among Non-Hispanic Blacks where in rural low SNAP block groups, 5% of households are non-Hispanic black compared to high SNAP block groups where 25% of the households are May 2018 Volume 118 Number 5
May 2018 Volume 118 Number 5 841
RESEARCH
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
Figure 1. Spatial distribution of Supplemental Nutrition Assistance Program stores around the weighted centroid of an urban block group in (A) and a rural block group in (B).
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
Urban CBGsb (n[3,950)
Rural CBGs (n[2,142)
Concentration of SNAPa All of North households Carolina (n[6,092)
Low (n[2,891)
High (n[1,059)
Population, n
9,726,032
5,141,535
1,379,539
2,529,611
675,347
Households, n
3,741,614
2,003,629
517,967
972,512
247,506
No. of households on SNAP (% of SNAP population)
539,057
164,366 (30)
187,251 (35)
111,404 (21)
76,036 (14)
c
P value Low (n[1,659)
d
e
median (IQRf) of CBGs! % of Household receiving SNAP Median CBG annual income, $
12.8 (5.5-23.2)
7.4 (2.6-13.7)
43,223 (32,189-57,309)
52,232 (40,071-71,667)
34.5 (28.0-44.3)
High (n[483)
P valuee
median (IQR) of CBGs! <0.001
26,223 (21,069-33,696) <0.001
11.3 (6.6-16.5)
29.1 (25.6-35.1)
<0.001
45,375 (36,957-54,255) 32,777 (26,849-38,561) <0.001
% Non-Hispanic black
13.5 (2.7-34.3)
11.6 (3.1-26.5)
46.7 (25.0-71.6)
<0.001
5.2 (0.2-16.6)
24.8 (4.7-45.9)
<0.001
% Non-Hispanic white
72.0 (45.6-88.3)
74.0 (53.2-87.7)
30.3 (11.5-54.3)
<0.001
85.5 (70.3-94.5)
56.3 (35.4-80.9)
<0.001
% Hispanic
3.9 (0.6-11.2)
4.5 (1.2-11.4)
6.8 (1.1-17.8)
<0.001
2.3 (0-7.1)
2.8 (0-11.1)
0.030
% Other race
2.7 (0.5-6.4)
3.7 (1.1-7.9)
2.9 (0.4-7.0)
<0.001
1.5 (0-4.0)
1.8 (0-5.8)
0.001
% of Households with no car
4.3 (1.4-10.2)
3.0 (0-7.0)
16.1 (8.5-25.4)
<0.001
3.2 (0.9-6.2)
7.2 (3.6-12.0)
<0.001
% of Households with children
30.7 (23.1-39.1)
35.0 (26.4-42.9)
<0.001
29.4 (22.9-36.0)
32.4 (26.1-39.7)
<0.001
30.0 (21.8-39.3)
CBGs with Accessible SNAP Store,g Full-Service Store, and Limited-Service Store Proximity Comparison n[4,619
Lowc (n[1,594)
Highd (n[905)
P valuee
Lowc (n[1,637)
n (%)!
May 2018 Volume 118 Number 5
Limited-variety SNAP store closer than full-variety Full-variety SNAP store closer than limited-variety a
4,072 (88.2)
1,305 (81.9)
547 (11.8)
289 (18.1)
838 (92.6) 67 (7.4)
P valuee
n (%)! <0.001
1,490 (91.0)
439 (90.9)
147 (9.0)
44 (9.1)
SNAP¼Supplemental Nutrition Assistance Program. CBG¼census block group. c Low concentration of SNAP households are CBGs with the percent of SNAP households <75th percentile. In these CBGs, <23% of household participate in SNAP. d High concentration of SNAP households are CBGs with the percent of SNAP households 75th percentile. In these CBGs, 23% of households participate in SNAP. e Difference in low to high CBGs. f IQR¼interquartile range. g Access is defined as a SNAP store located within 1 mile in urban areas and within 10 miles in rural areas. b
Highd (n[483)
0.930
RESEARCH
842
Table 2. Description of census block groups in North Carolina in 2015 (n¼6,092)
RESEARCH non-Hispanic black (P<0.001). About 7% of households in high SNAP block groups did not have a car compared with 3% in low SNAP block groups (P<0.001).
SNAP Store Access Proximity to Limited-Variety SNAP Stores vs FullVariety SNAP Stores. In North Carolina, residents often (in 88% of census block groups) live closer to limited-variety SNAP stores compared to full-variety SNAP stores. Among urban census block groups, residents in high SNAP block groups have closer access to limited-variety SNAP stores in 93% of the high SNAP block groups compared with low SNAP block groups, where 82% of the census block groups have a limited-variety SNAP store closer (Table 2).
Access. Based on Table 3 and Figure 2, the majority of urban census block groups do not have a full-variety SNAP store within 1 mile (73% of low SNAP block groups and 58% of high SNAP block groups). For larger cities, such as Charlotte, Raleigh, and Greensboro, census block groups with high SNAP participation and no access to a full-variety store are clustered and indicate areas where the need for a full-variety store is greatest. Limited-variety SNAP stores are not accessible in about half (48%) of low SNAP block groups. Among high SNAP block groups, about 16% do not have a limitedvariety SNAP store accessible. In rural areas, about 16% of census block groups did not have a full-variety SNAP food store accessible (within 10 miles), while <2% of census block groups did not have a limited-variety SNAP food store accessible. There was no difference in access by SNAP concentration in rural areas.
Distance if Accessible. In urban areas, as mentioned, the majority of census block groups do not have an accessible fullservice SNAP store. However, among the census block groups that do have a full-service store, the distance to that store for residents is about 0.7 miles. This distance does not differ by the concentration of SNAP households. Among the census block groups with an accessible limited-variety store, the distance to the limited-variety store was shorter for residents in high SNAP block groups (0.4 miles) compared with 0.6 miles for low SNAP block groups. In rural areas, about 85% of the census block groups have a full-service store accessible. Among these accessible census block groups, the median distance to a full-service store is 4.9 miles in low SNAP block groups and 4.4 miles in high SNAP block groups. Virtually all of the rural census block groups have access to a limited-variety SNAP store; the median distance to the store is 2.7 miles in low SNAP block groups and 2.1 miles in high SNAP block groups. Concentration of SNAP Stores. In urban areas, there is a difference in the number of accessible SNAP stores. Across both high and low SNAP block groups, the median number of accessible full-variety SNAP stores is zero. However, among limited-variety SNAP stores, there is typically one store in low SNAP block groups (0 stores/1,000 residents) and four in high SNAP block groups (3 stores/1,000 residents). In rural areas, there are typically three full-variety (2 stores/1,000 residents) and 18 limited-variety (13 stores/1,000 residents) SNAP stores accessible in both high and low SNAP block groups. Figure 3 compares the concentration of full-variety stores to low-variety stores within range for each block group. In May 2018 Volume 118 Number 5
urban areas, there is a great deal of variation between access to full and limited stores. Results in Charlotte, Greensboro, and Raleigh suggest pockets of census block groups with a high concentration of limited-variety stores and a low concentration of full-service stores. Few urban census block groups exhibit a high concentration of both full- and limitedvariety stores, although the rural areas in the immediate vicinity of these cities often do. These rural census block groups tend to have access to both full-variety stores and limitedvariety stores within the allotted 10 miles, although exceptions can be found throughout the state. Finally, urban census block groups at the periphery of those cities suffer from a lack of limited- and full-services stores.
DISCUSSION About half of SNAP households in North Carolina live in census block groups with a high concentration of SNAP households. It is important that SNAP participants living in both high and low SNAP block groups have adequate access to SNAP stores. In North Carolina, the SNAP food environment landscape exhibits significant differences among rural and urban areas. Within rural areas, there were small differences in store accessibility between low SNAP block groups and high SNAP block groups. In general, SNAP households live about 2.5 miles from a limited-variety SNAP store and 4.5 miles from a full-variety SNAP store. Within a 10-miles drive, those in rural areas likely have access to 3 full-variety SNAP stores and 18 limited-variety SNAP stores. This is a sharp contrast from urban areas, where a disparity in the availability of SNAP stores within a 1-mile distance between low and high SNAP block groups is observed. Individuals living in a low SNAP block groups typically do not have access to a full-variety food store, but do have access to 1 limited-variety food store. However, individuals living in high SNAP block groups do not have access to a full-variety food store, but have access to a higher number (ie, four) of limited-variety SNAP food stores. The differences in the SNAP food environment between rural and urban areas are partly due to the difference in the definition of the “accessible” range: 1 mile in urban areas and 10 miles in rural areas. In urban areas, households in high SNAP block groups were five times more likely to not have access to a car (16% compared to 3% for households living in low SNAP block groups). It is unclear whether those households without a car also participated in SNAP. In rural areas, 7% of households in high SNAP block groups did not have a car, double the percent in low SNAP block groups. For those in rural areas without vehicle access, the accessibility definition of 10 miles may not be a reasonable measure of accessibility. Previous research by Ver Ploeg and colleagues29 found that 34% of SNAP households did not use their own vehicle to get to their grocery store; the majority of these households (61%) were able to carpool with someone else and the other 39% “walk, bike, or take public transit or a shuttle to the store.” However, Ver Ploeg and colleagues’ research did not stratify the SNAP households by rural/urban status.29 Additional research to determine the percent of SNAP households without a car or other modes of reliable transportation in urban and rural areas is warranted. In 2015, 87% of SNAP redemptions (about $2 billion) were made at full-variety stores, even though the majority of SNAP households in urban areas did not have adequate access to fullJOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
843
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
Minimum Distance (in Miles) to SNAP Store for CBGs with Accessb
CBGsa without Accessb to SNAPc Store Variable Concentration of SNAP households in urban CBGs
Lowd (n[2,891)
Highe (n[1,059) P valuef Lowd
%!
Highe
No. of Stores within Access Range/per 1,000 Residents
No. of Stores within Accessb Range P valuef Lowd
median (IQRg)!
Highe
P valuef Lowd
median (IQRg)!
Highe
P valuef
median (IQRg)!
Full variety of roods
72.7
58.0
<0.001
0.71 (0.55-0.86) 0.70 (0.53-0.85)
0.267
0 (0-1)
0 (0-1)
<0.001
0 (0-27)
0 (0-10)
<0.001
Limited variety of roods
48.1
15.6
<0.001
0.61 (0.43-0.79) 0.44 (0.28-0.63) <0.001
1 (0-3)
4 (1-8)
<0.001
0 (0-108)
3 (0-75)
<0.001
Lowd (n[1,659)
Highe (n[483)
P valuef Lowd
Highe
P valuef Lowd
Highe
P valuef
Concentration of SNAP households in rural CBGs Full variety of foods May 2018 Volume 118 Number 5
Limited variety of foods a
%!
Highe
P valuef Lowd
median (IQRg)!
median (IQRg)!
16.2
16.1
0.97
4.88 (3.21-6.77) 4.44 (2.60-6.84)
0.019
1.3
0.4
0.09
2.66 (1.67-4.00) 2.08 (1.20-3.37) <0.001
3 (1-8)
3 (1-6)
19 (9-38) 17 (8-34)
CBG¼census block group. Access is defined as a SNAP store located within 1 mile in urban areas and within 10 miles in rural areas. c SNAP¼Supplemental Nutrition Assistance Program. d Low concentration of SNAP households are CBGs with the percent of SNAP households <75th percentile. In these CBGs, <23% of household participate in SNAP. e High concentration of SNAP households are CBGs with the percent of SNAP households 75th percentile. In these CBGs, 23% of households participate in SNAP. f Difference in low to high CBGs. g IQR¼interquartile range. b
median (IQRg)!
0.002 0.05
2 (0-560)
2 (0-33)
0.128
13 (0-1,680) 13 (0-230) 0.181
RESEARCH
844
Table 3. Supplemental Nutrition Assistance program food store availability among neighborhoods with a high and low concentration of Supplemental Nutrition Assistance Program participants by urban/rural status in North Carolina
RESEARCH
Figure 2. Spatial distribution of census block groups without a full-service food store in range (urban census block group: within 1 mile; rural census block groups: within 10 miles). variety stores. Based on the high redemption rate at fullvariety stores, it is assumed that SNAP participants prefer to use their benefits at full-variety stores and will overcome accessibility barriers to do so. Shannon30 examined the relationship between location of SNAP recipient and “outflow” of SNAP benefits from the participants neighborhood. He found that many of the full-variety stores were located in middleclass suburban areas and that about 80% of all SNAP benefits are redeemed at full-variety stores. However, he also found that SNAP participants living in dense urban areas with limited availability to full-variety stores only spent 51% of their SNAP benefits at full-variety stores. The remaining 49% of SNAP benefits were spent at limited-variety stores. Recent research examining the food and beverage purchases made at a supermarket chain in 2011 found that SNAP participants spent the most on the following 5 items: soft drinks, fluid milk products, ground beef, bagged snacks, and cheese.31 Four of these items are likely available at most SNAP stores (whether full-variety stores or limited-variety stores): soft drinks, fluid milk products, bagged snacks, and cheese. Because fresh meat is rarely available at limited-variety stores,32-35 the demand for meat and other foods may draw SNAP participants to fullvariety stores. Limited SNAP stores are much more accessible to the majority of SNAP households. Better accessibility may account for 13% (about $300 million) of SNAP redemptions at limited-variety stores. As the majority of SNAP households redeem their SNAP benefits at SNAP stores outside of their immediate vicinity, May 2018 Volume 118 Number 5
this could be an issue in two ways. First, SNAP participants that live relatively far from a full-variety SNAP store may feel pressured to purchase more shelf-stable, processed foods at a limited-variety SNAP store, which may result in less-healthy food purchases. Second, should SNAP participants be expected to travel long distances in order to redeem their benefits at full-variety stores? This is a question that should be debated among policy makers. While SNAP food stores are required to offer a minimum number of food types and varieties, there is currently no consideration for the need for that particular type of SNAP food store when determining SNAP food store authorization. In urban areas, SNAP participants in high SNAP block groups have access to more limited service stores compared to those in low SNAP block groups. In addition, high SNAP census block groups are disproportionally populated with more non-Hispanic black residents (47% vs 12%), Hispanic residents (7% vs 5%), lower-income households ($26,223 vs $52,232), households without a car (16% vs 3%), and households with a child (35% vs 30%). Research to study how exposure to more limited-variety stores impacts health among these populations is warranted. The USDA Food and Nutrition Service has posted new SNAP retailer stocking requirements.36 These requirements increase the number of stocking units by staple food variety from 1 to 3. It is unclear when this requirement will take effect and, when it does, it is not clear how it will impact the SNAP-authorized stores. Perhaps some limited SNAPJOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
845
RESEARCH
Figure 3. Density of full-service vs limited-service food stores in range for each census block group (urban census block group: within 1 mile; rural census block groups: within 10 miles).
authorized stores will discontinue participation in SNAP. Once implemented, research is warranted to study the impact of this regulation on SNAP store access. There are some limitations of this study. First, the USDA definition for “low access” was used; that may not give a true assessment of accessibility from the SNAP participant’s perspective. Second, the availability of foods and beverages at limited-variety food stores was not assessed, assumptions about the foods/beverages available were made based on USDA SNAP store definitions and previous research on the foods/beverages typically available at convenience stores, drugstores, and dollar stores.32-34,37 Some of the store types (like small and/or medium grocery stores) may have had a greater variety of options than others. Third, proximity from the weighted centroid of each census block group was estimated. This level of geographic simplification may bias travel estimates and future research is needed to evaluate food accessibility at finer geographic scales. Fourth, the route between the centroid and the SNAP retailer is assumed to be the shortest one, which may not reflect reality. Individuals may prefer more familiar routes to combine their trips (tripchaining). Fifth, it is assumed that individuals either drive or walk to each SNAP retailer and do not use public transportation, such as bus or subway. Recent research by Widener38 has shown that the use of public transit can dramatically affect the geographic access to food retailers. Sixth, this analysis is limited to SNAP-authorized food stores, this limits the inclusion of other food stores that do not 846
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
participate in SNAP. Therefore, communities likely have access to additional food stores than are presented in this analysis. To the authors’ knowledge, no research has been done to compare the characteristics of SNAP-authorized and nonSNAP-authorized food stores. This is an area for further research. Seventh, the descriptive statistics (Wilcoxon ranksum and c2 tests) assumed spatial independence among observation units. Spatial dependence (or autocorrelation) suggests that nearby census block groups are likely to exhibit similar results, and the assumption of independence is likely to overestimate the statistical significance of the results, such as P values.39,40 Eighth, although there exists substantial variation in the characteristics of low- vs high-SNAP census block groups in urban areas, more research is needed to evaluate the sensitivity of this analysis at different spatial scales. Additional variables, such as population density, may shed some light on such variation. Finally, this analysis was based on North Carolina, therefore, results may not be generalizable to all states.
CONCLUSIONS In closing, the results from this study highlight the importance of considering location and type of SNAP stores. While there are more than 9,500 SNAP stores in North Carolina, few full-variety stores are easily accessible to SNAP participants in urban areas. Also, >80% of the SNAP stores are limited variety, which tend to offer a poor selection of healthy food May 2018 Volume 118 Number 5
RESEARCH options—particularly fresh fruits and vegetables. In the future, researchers are encouraged to consider access to SNAP stores by type and availability of reliable transportation when evaluating the impact of SNAP on food security and health.
20.
US Census Bureau. American Community Survey, 2011-2015; American Community Survey 5-year estimates. http://factfinder2. census.gov/. Accessed October 10, 2017.
21.
US Census Bureau. Centers of population. https://www.census.gov/ geo/reference/centersofpop.html. Accessed October 11, 2017.
22.
US Census Bureau, Geography Division. Centers of Population Computation for the United States 1950-2010. https://www2.census. gov/geo/pdfs/reference/cenpop2010/COP2010_documentation.pdf. Published 2011. Accessed October 11, 2017.
23.
US Department of Agriculture, Economic Research Service. Food environment atlas: Data access and documentation. http://www.ers. usda.gov/data-products/food-access-research-atlas/documentation. aspx. Published 2013. Accessed October 10, 2017.
24.
Dressler H, Smith C. Food choice, eating behavior, and food liking differs between lean/normal and overweight/obese, low-income women. Appetite. 2013;65:145-152.
US Census Bureau. 2010 Census urban and rural classification and urban area criteria. https://www.census.gov/geo/reference/ua/ urban-rural-2010.html. Accessed October 31, 2017.
25.
US Census Bureau. 2010 Census Data. https://www.census.gov/201 0census/data/.
Poston WS 2nd, Foreyt JP. Obesity is an environmental issue. Atherosclerosis. 1999;146(2):201-209.
26.
ArcGIS [computer program]. Version 10.4. Redlands, CA: ESRI; 2016.
27.
MySQL [computer program]. Version 5.7. Redwood Shores, CA: Oracle Corporation; 2015.
28. 29.
SAS [computer program]. Version 9.3. Cary, NC: SAS Institute; 2011. Ver Ploeg M, Mancino L, Todd JE, Clay DM, Scharadin B. Where do Americans usually shop for food and how do they travel to get there? Initial findings from the National Household Food Acquistion and Purchase Survey. Econ Inform Bull. 2015;138.
30.
Shannon J. What does SNAP benefit usage tell us about food access in low-income neighborhoods? Soc Sci Med. 2014;107:89-99.
31.
Garasky S, Mbwana K, Romualdo A, Tenaglio A, Roy M. Foods typically purchased by Supplemental Nutrition Assistance Program (SNAP) households—Appendices 2016. http://www.fns.usda.gov/ sites/default/files/ops/SNAPFoodsTypicallyPurchased-Appendices.pdf. Accessed October 11, 2017.
References 1.
2.
3.
4.
US Department of Agriculture, Food and Nutrition Service. Supplemental Nutrition Assistance Program (SNAP) participants and costs. http://www.fns.usda.gov/pd/supplemental-nutrition-assistance-programsnap. Published 2017. Accessed October 10, 2017. Castner L, Henke J. Benefit redemption patterns in the Supplemental Nutrition Assistance Program. http://www.fns.usda.gov/sites/ default/files/ARRASpendingPatterns.pdf. Published 2011. Accessed October 10, 2017.
5.
Block JP, Scribner RA, DeSalvo KB. Fast food, race/ethnicity, and income: A geographic analysis. Am J Prev Med. 2004;27(3):211-217.
6.
Drewnowski A, Specter SE. Poverty and obesity: The role of energy density and energy costs. Am J Clin Nutr. 2004;79(1):6-16.
7.
Walker RE, Keane CR, Burke JG. Disparities and access to healthy food in the United States: A review of food deserts literature. Health Place. 2010;16(5):876-884.
8.
Alwitt LF, Donley TD. Retail stores in poor urban neighborhoods. J Consum Aff. 1997;3(1):139-164.
9.
Chung C, Myers SL. Do the poor pay more for food? An analysis of grocery store availability and food price disparities. J Consum Aff. 1999;33(2). 2776-2296.
10.
Lewis LB, Galloway-Gilliam L, Flynn G, Nomachi J, Keener LC, Sloane DC. Transforming the urban food desert from the grassroots up: A model for community change. Fam Community Health. 2011;34(suppl 1):S92-S101.
32.
McEntee J, Agyeman J. Towards the development of a GIS method for identifying rural food deserts: Geographic access in Vermont, USA. Appl Geogr. 2010;30(1):165-176.
Caspi CE, Pelletier JE, Harnack L, Erickson DJ, Laska MN. Differences in healthy food supply and stocking practices between small grocery stores, gas-marts, pharmacies and dollar stores. Public Health Nutr. 2016;19(3):540-547.
33.
Racine EF, Batada A, Solomon CA, Story M. Availability of foods and beverages in Supplemental Nutrition Assistance Program-authorized dollar stores in a region of North Carolina. J Acad Nutr Diet. 2016;116(10):1613-1620.
34.
Racine EF, Kennedy A, Batada A, Story M. Foods and beverages available at SNAP-authorized drugstores in sections of North Carolina. J Nutr Educ Behav. 2017;49(8):674-683 e671.
11.
12.
Apparicio P, Cloutier MS, Shearmur R. The case of Montréal’s missing food deserts: Evaluation of accessibility to food supermarkets. Int J Health Geogr. 2007;6(4).
13.
Shaw HJ. Food deserts: Towards the development of a classification. Geogr Ann A. 2006;88(2):231-247.
14.
Forsyth A, Lytle L, Van Riper D. Finding food: Issues and challenges in using Geographic Information Systems to measure food access. J Transp Land Us. 2010;3(1):43-65.
35.
Laska MN, Borradaile KE, Tester J, Foster GD, Gittelsohn J. Healthy food availability in small urban food stores: A comparison of four US cities. Public Health Nutr. 2010;13(7):1031-1035.
15.
Kirby RS, Delmelle E, Eberth JM. Advances in spatial epidemiology and geographic information systems. Ann Epidemiol. 2017;27(1):1-9.
36.
16.
Chen X, Clark J. Interactive three-dimensional geovisualization of spaceetime access to food. Appl Geogr. 2013;43:81-86.
US Department of Agriculture. Final Rule: Enhancing Retailer Standards in the Supplemental Nutrition Assistance Program (SNAP). https://www.fns.usda.gov/snap/fr-121516. Accessed December 11, 2017.
17.
Wood BS, Horner MW. Understanding accessibility to SNAPaccepting food store locations: Disentangling the roles of transportation and socioeconomic status. Appl Spat Anal Policy; 2015: 1-19.
37.
Laska MN, Caspi CE, Pelletier JE, Friebur R, Harnack LJ. Lack of healthy food in small-size to mid-size retailers participating in the Supplemental Nutrition Assistance Program, Minneapolis-St. Paul, Minnesota, 2014. Prev Chronic Dis. 2015;12:E135.
18.
Rigby S, Leone AF, Kim H, et al. Food deserts in Leon County, FL: Disparate distribution of Supplemental Nutrition Assistance Program-accepting stores by neighborhood characteristics. J Nutr Educ Behav. 2012;44(6):539-547.
38.
Widener MJ. Comparing measures of accessibility to urban supermarkets for transit and auto users. Prof Geogr; 2016:1-10.
39.
Datta S, Satten GA. A signed-rank test for clustered data. Biometrics. 2008;64(2):501-507.
Racine EF, Wang Q, Laditka SB, Johnson CR, Mignery A. The characteristics and concentration of SNAP approved stores and community health. J Hunger Environ Nutr. 2013;8(3):350-361.
40.
Lettenmaier DP. Detection of trends in water quality data from records with dependent observations. Water Resources Res; 1976: 1037-1046.
19.
May 2018 Volume 118 Number 5
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
847
RESEARCH AUTHOR INFORMATION E. F. Racine and E. Delmelle are associate professors, E. Major is a graduate student, and C. A. Solomon is a doctoral student, University of North Carolina at Charlotte. Address correspondence to: Elizabeth F. Racine, DrPH, RD, University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte, NC 28223. E-mail: [email protected]
STATEMENT OF POTENTIAL CONFLICT OF INTEREST No potential conflict of interest was reported by the authors.
FUNDING/SUPPORT This research was funded by Healthy Eating Research, a national program of the Robert Wood Johnson Foundation.
848
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
May 2018 Volume 118 Number 5
Original Research
Accessibility Landscapes of Supplemental Nutrition Assistance ProgramAuthorized Stores Elizabeth F. Racine, DrPH, RD; Eric Delmelle, PhD; Elizabeth Major, MA; Corliss A. Solomon, MPH ARTICLE INFORMATION Article history: Submitted 15 March 2017 Accepted 3 November 2017 Available online 22 January 2018
Keywords: Supplemental Nutrition Assistance Program (SNAP) Food access Social determinants Food stamps Geographic Information Systems (GIS) 2212-2672/Copyright ª 2018 by the Academy of Nutrition and Dietetics. https://doi.org/10.1016/j.jand.2017.11.004
ABSTRACT Background The Supplemental Nutrition Assistance Program (SNAP) is the largest food assistance program in the United States. Participants receive electronic benefits that are redeemable at a variety of food stores. Previous research notes that low-income neighborhoods often lack supermarkets with high-quality, affordable food. Objective The first aim of this study was to explore the number and spatial distribution of SNAP stores by type and to assess how SNAP benefit redemption is linked to store type in North Carolina in 2015. The second aim was to compare the demographics of populations living in areas with a high concentration of SNAP participants vs areas with a lower concentration of SNAP participants. The third aim was to test for disparities in the availability of and access to SNAP-authorized stores in areas with high vs low concentration of SNAP participants stratified by rural/urban status. Design US Department of Agriculture and US Census data were used to explore the spatial distribution of SNAP stores at the census block group level utilizing a Geographic Information System. Participants The 9,556 North Carolina SNAP stores in 2015 categorized into full-variety and limited-variety stores. Outcome measures Proximity to limited-variety SNAP food stores and full-variety SNAP food stores within access range (1 mile in urban areas and 10 miles in rural areas). Statistical analyses Wilcoxon rank sum and c2 tests are used to compare the distance to and concentration of SNAP stores by rurality and SNAP participant concentration at census block group scale. Results Among the SNAP stores in North Carolina, 83% are limited-variety stores and 17% are full-variety stores. There are disparities in the demographics of individuals living in census block groups with a high proportion of SNAP participants compared to census block groups with a lower proportion of SNAP participants. More households in higher SNAP participant census block groups were non-white, did not have a car, and had children compared to census block groups with lower SNAP participation. Residents in high SNAP participant census block groups typically had access to 0 full-variety stores and 4 limited-variety stores in urban areas and 3 full-variety stores and 17 limitedvariety stores in rural areas. Conclusions SNAP participant access to a variety of stores should be considered when approving food stores for SNAP authorization. More research is essential to disentangle the relationship between access to SNAP store type and SNAP food choice and to estimate geographical disparities. J Acad Nutr Diet. 2018;118(5):836-848.
The Continuing Professional Education (CPE) quiz for this article is available for free to Academy members through the MyCDRGo app (available for iOS and Android devices) and through www.jandonline.org (click on “CPE” in the menu and then “Academy Journal CPE Articles”). Log in with your Academy of Nutrition and Dietetics or Commission on Dietetic Registration username and password, click “Journal Article Quiz” on the next page, then click the “Additional Journal CPE quizzes” button to view a list of available quizzes. Non-members may take CPE quizzes by sending a request to [email protected]. There is a fee of $45 per quiz (includes quiz and copy of article) for non-member Journal CPE. CPE quizzes are valid for 1 year after the issue date in which the articles are published.
836
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
T
HE SUPPLEMENTAL NUTRITION ASSISTANCE PROgram (SNAP) is the largest food assistance program in the United States, serving more than 44 million lowincome Americans.1 SNAP participants receive electronic benefits that are redeemable at an assortment of food stores, such as grocery stores, convenience stores, and drug stores. In 2016, approximately 19% of American households participated in SNAP, receiving on average $255 per month in benefits, with an annual cost of approximately $75 billion.1 Most (84%) SNAP benefits are redeemed at supermarkets and super stores, while 64% of SNAP transactions take place ª 2018 by the Academy of Nutrition and Dietetics.
RESEARCH at supermarkets and super stores, therefore, about 36% of SNAP transactions take place at other types of SNAP stores, such as smaller grocery stores, and convenience stores.2 While research has explored the relationship between the food environment and obesity among low-income communities, one common finding is that low-income neighborhoods often lack access to supermarkets that offer high-quality affordable food.3 Reduced access to these full-variety supermarkets may increase exposure to energy-dense food and lead to diets high in processed foods, fat, sugar, and sodium.4-7 In addition, low-income communities typically have access to smaller food outlets,8 pay more for shopping locally,9 and have fewer healthy food options compared to higher-income communities.10 The topic of access to food has received considerable attention in the literature, and can be measured from several perspectives, including geographic, economic, and informational.11 Geographic accessibility examines the relationship between the location of the store where food can be purchased and the individual in need. It incorporates structural barriers to food outlets, including transportation resources, travel time and distance, and travel costs.12 Such geographic measures can further be employed to identify areas that are low income with limited access to healthy food retailers, referred to as “food deserts.”13 Recent advances in Geographic Information Systems have facilitated the modeling of geographic measures of food access and food availability.14,15 Three types of methods have been pervasively adopted to estimate food accessibility with Geographic Information Systems: the container method, the buffer method, and the network method.16 Although the first two techniques can be loosely categorized as “exposure” approaches, the latter one, which is used in this article, attempts to determine the ease of access to a food outlet. To date there has been very little research exploring SNAP participants access to SNAP-authorized stores. Wood and Horner17 examined availability of SNAP stores among different types of neighborhoods in Leon County, FL. They found that most low-car-access census block groups (99%) contained a SNAP store accessible within a 20-minute drive of a SNAP store, while less than half (40%) of these census block groups contained a SNAP stores within a 20-minute walk.17 Two other studies, both conducted in urban counties, found that low-income communities with a high percentage of SNAP participants were more likely than higher-income communities to be populated with limitedvariety SNAP-authorized stores and less likely to include full-variety stores.18,19 Limited-variety stores typically lack healthy products like fresh produce and lean, fresh meats.18,19 It is unclear whether this association between SNAPauthorized store type and concentration of residents participating in SNAP is limited to these counties. The first aim of this study was to explore the number and spatial distribution of SNAP stores by type and to assess how SNAP benefit redemption is linked to store type in North Carolina in 2015. The second aim was to compare the demographics of populations living in areas with a high concentration of SNAP participants vs areas with a lower concentration of SNAP participants. The third aim was to test for disparities in the availability of and access to SNAPauthorized stores in areas with a high vs low concentration of SNAP participants stratified by rural/urban status. May 2018 Volume 118 Number 5
MATERIALS AND METHODS Data Two primary data sources are used for this study. First, the US Department of Agriculture (USDA) Food and Nutrition Service provided a geolocated dataset of all SNAP-authorized stores in North Carolina from 2015 (hereafter “USDA data”). This dataset is available by request from the USDA. Second, 2011-2015 data estimates from the American Community Survey (hereafter “Census data”) were used at the census block group level.20 The US Census also publishes populationweighted geographic centers, or centroids, after each decennial census. The 2010 population-weighted centroids are used as points of origin to estimate geographic access for each block group.21,22 This study was exempt from Institutional Review Board review because it is not human subjects research.
USDA Data USDA data includes store type, latitude, longitude, address, and county. The USDA categorizes SNAP-authorized stores by 17 different types, which are summarized in Table 1. When stores apply for SNAP authorization, the store self-selects store type based on instructions provided by USDA. USDA then reviews the store application and, at that time, can change the store type designation to align with the USDA definition. For this study, store types are grouped into binary categorizes: “full variety,” defined as offering a “wide” or “extensive” selection of fresh meat, dairy, produce, and grains (eg, super stores, large grocery stores, and supermarkets) and “limited variety,” defined as not offering a “wide” or “extensive” selection of fresh meat, dairy, produce, and grains (eg, drugstores, convenience stores, small grocery stores, and dollar stores) or all other SNAP-authorized stores (see Table 1).
Census Data The US Census American Community Survey provides 5-year estimates information at the census block group level for the years 2011 to 201520; the unit of analysis for this study. The census block group level was selected to best identify neighborhood characteristics, as census block group is the smallest geographic unit available with demographic information pertinent to this study. The following variables for each census block group in North Carolina are utilized for this analysis: population size, household size, number of households participating in SNAP (recoded into percent of households in the census block group on SNAP), urban/rural status, median household income, number of residents by race/ ethnicity (recoded as percent of census block group that was non-Hispanic black, non-Hispanic white, Hispanic, and other based on the population size), percent of population with access to a vehicle, and percent of households with children.
Geographic Categories To examine whether the SNAP food environment varies according to the concentration of SNAP participants, the North Carolina census block groups are divided into four categories based on the percentage of SNAP participants in each census block group and whether these census block groups are urban or rural. First, to examine the concentration of SNAP participants within the 6,092 census block groups in North Carolina by quartile; those 75th percentile are considered census JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
837
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
USDAb SNAPc Vendor Categoriesd
Binary Groups
Store type
USDA definition
Redemptions, $ Stores, n Groups
Redemption, $ Stores, n
Large grocery store
44,688,127 “A store that carries a wide selection of all four staple food categories. They may sell ineligible items as well, but their primary stock is food items.”
82
Full variety of foods
2,062,944,181
1,614
Supermarket
“Establishments commonly known as supermarkets, food 1,016,738,434 stores, grocery stores and food warehouses primarily engaged in the retail sale of an extensive variety of grocery and other store merchandise.”
1,090
—
—
—
Super store
“Very large supermarkets, “big box” stores, super stores and food warehouses primarily engaged in the retail sale of a wide variety of grocery and other store merchandise.”
1,001,517,620
442
—
—
—
Bakery specialty
“Food stores specializing in the sale of bread/cereal products”
1,586,762
67
Limited variety of foods
Meat/poultry specialty
“Food stores specializing in the sale of meat products”
13,217,395
72
—
—
—
300,091,812
7,942
May 2018 Volume 118 Number 5
Seafood specialty
“Food stores specializing in the sale of seafood products” 4,000,093
81
—
—
—
Fruits/vegetable specialty
458,175 “Food stores specializing in the sale of fruits and/or vegetables that operates in a fixed or semi-permanent location”
43
—
—
—
Combination grocery/other
“Primary business is sale of general merchandise but also 128,983,222 sell a variety of food products. Such stores include independent drug stores, dollar stores, and general stores.”
2,664
—
—
—
Convenience store
92,027,366 “Self-service stores that offer a limited line of convenience items and are typically open long hours to provide easy access for customers.”
4,192
—
—
—
Farmers’ market
119,312 “A single or multi-stall market that sells agricultural products, particularly fresh fruit and vegetables, to the general public at a single or multiple locations.”
110
—
—
—
(continued on next page)
RESEARCH
838
Table 1. Categorization of Supplemental Nutritional Assistance Programauthorized stores, US Department of Agriculture categories, and groupings for a study in North Carolina in 2015a
May 2018 Volume 118 Number 5
Table 1. Categorization of Supplemental Nutritional Assistance Programauthorized stores, US Department of Agriculture categories, and groupings for a study in North Carolina in 2015a (continued) USDAb SNAPc Vendor Categoriesd
Binary Groups
Store type
USDA definition
Direct marketing farmer
“Designation applies to direct marketing farmers; these 124,862 are individual producers of agricultural products, particularly fresh fruit and vegetables, as well as meat, fish, dairy, and/or grains that are sold to the general public through a direct marketing venue such as a roadside farm stand, pick-your own operation, and/or market stall within a farmers’ market.”
174
—
—
—
Medium grocery store
“A store that carries a moderate selection of all four staple 31,388,388 food categories.”
198
—
—
—
Small grocery store
“A store that carries a small selection of all four staple food categories.”
300
—
—
—
Nonprofit food buying cooperative “Any store that operates as a “cooperative”.”
Redemptions, $ Stores, n Groups
20,587,550
Redemption, $ Stores, n
—
—
—
—
—
—
“Designation applies to all retail food entities, located on 5,577,143 military installations that sell food and non-food products”
7
—
—
—
“Statutory/regulatory definition: an establishment that sells eligible food to meal services for resale to households.”
1
—
—
—
“A store that does not have a permanent store location, this includes delivery routes that deliver food at set locations and times, as well as rolling routes.”
Military commissary
Wholesaler
454,098 1,567,446
Redacted
a
The number of stores and redemptions by category are included. Total number of stores is 9,556 and total redemption is $2,363,035,993. USDA¼US Department of Agriculture. c SNAP¼Supplemental Nutrition Assistance Program. d Stores applying for SNAP select the appropriate category, then USDA reviews and may change the category to better align with the definition. The definitions were requested from the Food and Nutrition Services, USDA. North Carolina SNAP stores by type and redemption were provided to the first author by the USDA, Food and Nutrition Service. b
839
RESEARCH
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
9 24
Delivery route
RESEARCH block groups with a high concentration of SNAP participants. At the 75th percentile, 23% of the census block group households participate in SNAP. Hereafter census block groups with a high concentration of SNAP households, those with 23% on SNAP are referred to as “high SNAP block groups,” while census block groups with a lower concentration of SNAP households, those with <23% of households participating in SNAP are referred to as “low SNAP block groups.” Second, census block groups are classified by access to a full-variety SNAP food store using the USDA definition (ie, 1 mile or more from a full-service store in urban areas and 10 miles from a full-service store in rural areas).23 To account for the different definitions by rurality, census block groups were stratified by rural status. The Census Bureau classifies geographies into three categories: urban areas (>50,000 residents), urban cluster areas (between 2,500 and 50,000 residents), and rural areas (any area not meeting the criteria for urban or urban cluster).24 The decennial Census describes how many households in each tract are found within each area.25 The category in which the majority of households were found was used to describe each block group, so a block group with 220 urban households, 630 urban cluster households, and 11 rural households would be coded as an urban cluster block group. For this analysis, urban and urban clusters are defined as urban, with urban cluster census block groups traditionally encompassing suburbs. As a result, the four geographic categories are: rural census block groups with a high concentration of SNAP households, urban census block groups with a high concentration of SNAP households, rural census block groups with a lower concentration of SNAP households, and urban census block groups with a lower concentration of SNAP households.
Measures The following variables are used to examine the SNAP food store environment: comparison of proximity to full-variety vs limited-variety stores; access to one or more SNAP food store by SNAP store type (yes or no) as defined by USDA as a SNAP store located within 1 mile in urban areas and within 10 miles in rural areas23; average distance to accessible SNAP food stores; concentration of SNAP store type within the USDA defined access range; and concentration of SNAP store type within the USDA defined access range by 1,000 population.
Estimating Proximity to SNAP Stores Proximity to SNAP retailers is estimated at the census block group level for 2015. The weighted population centroid of each block group is used as a point of origin to measure the proximity to each SNAP retailer. Following the USDA Food Access Research Atlas,23 only SNAP retailers within 1 and 10 Euclidean miles for urban and rural block groups were retained, respectively. Travel is estimated using the ArcGIS Online Application Programming Interface.26 The travel routing results are then aggregated to each block group using MySQL,27 producing a table of demographic information and shortest distances from the block group’s centroid to each type of store. These results are further aggregated according to SNAP participation levels. The procedure and its results are illustrated for an urban census block group in Figure 1A and a rural census block group in Figure 1B and demonstrate that the use of Euclidean 840
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
distance overestimates the number of available SNAP stores. The distance to all SNAP retailers within 1 and 10 Euclidean miles is estimated; retailers in the blue-shaded regions are within 1 or 10 miles network distance. In the urban scenario (Figure 1A), a total of 5 SNAP retailers of limited variety are available within 1-mile network distance (average distance¼0.70 mile, range¼0.41 to 0.96 mile), and another SNAP store is located within the 1-mile Euclidean distance but outside the 1-mile network distance. The closest available full-variety SNAP store (a supermarket) is located at 1.49 miles from the block group centroid. In the rural scenario (Figure 1B), a total of 14 SNAP retailers are available within a 10-mile network distance (average distance¼6.48 miles, range¼3.58 to 9.06 miles), and three of those are full-variety stores (average distance¼7.45 miles, range¼4.78 to 8.82 miles). Note that 3 full-variety stores and 13 limited-variety stores are located outside the 10-mile network distance, but well within the 10-mile Euclidean distance.
Statistical Analysis Descriptive statistics are presented for all North Carolina census block groups with at least one household (n¼6,092) and for each of the four geographic groups. Wilcoxon ranksum and c2 tests are used to compare the distance to and concentration of SNAP stores by rurality and SNAP participant concentration. Statistical significance is determined at a<.05. Analyses were conducted in SAS software.28
RESULTS In North Carolina, more than $2.3 billion in SNAP redemptions were spent at food stores in 2015; 87% of SNAP redemptions (about $2 billion) were made at full-variety stores and 13% (about $300 million) at limited-variety stores (Table 1). There were 9,556 SNAP stores in North Carolina—17% were full-variety stores and 83% were limitedvariety stores. Approximately 14% of North Carolina households participated in SNAP in 2015 (Table 2). The majority lived in urban areas (65%) and about half live in census block groups with a high concentration of SNAP households (49%). In urban areas, most of the SNAP households (53%) live in census block groups with a high concentration of SNAP households. There are disparities in the demographics of individuals living in high SNAP block groups compared with low SNAP block groups. A greater proportion of non-Hispanic blacks, (47% vs 12%), Hispanics (7% vs 5%), households without a car (16% vs 3%), and households with children (35% vs 30%); and a lower proportion of non-Hispanic whites (30% vs 74%) are located in high SNAP block groups. As expected, in high SNAP block groups, the median household income is significantly lower, approximately $26,000 per year compared to approximately $52,000 per year in low SNAP block groups. In high SNAP block groups about 35% of households participate in SNAP compared to 7% of households living in low SNAP block groups. In rural areas, there is a disparity in the racial and ethnic composition of residents by SNAP household concentration. These differences are statistically significant in all the demographic variables tested, but are particularly divergent among Non-Hispanic Blacks where in rural low SNAP block groups, 5% of households are non-Hispanic black compared to high SNAP block groups where 25% of the households are May 2018 Volume 118 Number 5
May 2018 Volume 118 Number 5 841
RESEARCH
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
Figure 1. Spatial distribution of Supplemental Nutrition Assistance Program stores around the weighted centroid of an urban block group in (A) and a rural block group in (B).
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
Urban CBGsb (n[3,950)
Rural CBGs (n[2,142)
Concentration of SNAPa All of North households Carolina (n[6,092)
Low (n[2,891)
High (n[1,059)
Population, n
9,726,032
5,141,535
1,379,539
2,529,611
675,347
Households, n
3,741,614
2,003,629
517,967
972,512
247,506
No. of households on SNAP (% of SNAP population)
539,057
164,366 (30)
187,251 (35)
111,404 (21)
76,036 (14)
c
P value Low (n[1,659)
d
e
median (IQRf) of CBGs! % of Household receiving SNAP Median CBG annual income, $
12.8 (5.5-23.2)
7.4 (2.6-13.7)
43,223 (32,189-57,309)
52,232 (40,071-71,667)
34.5 (28.0-44.3)
High (n[483)
P valuee
median (IQR) of CBGs! <0.001
26,223 (21,069-33,696) <0.001
11.3 (6.6-16.5)
29.1 (25.6-35.1)
<0.001
45,375 (36,957-54,255) 32,777 (26,849-38,561) <0.001
% Non-Hispanic black
13.5 (2.7-34.3)
11.6 (3.1-26.5)
46.7 (25.0-71.6)
<0.001
5.2 (0.2-16.6)
24.8 (4.7-45.9)
<0.001
% Non-Hispanic white
72.0 (45.6-88.3)
74.0 (53.2-87.7)
30.3 (11.5-54.3)
<0.001
85.5 (70.3-94.5)
56.3 (35.4-80.9)
<0.001
% Hispanic
3.9 (0.6-11.2)
4.5 (1.2-11.4)
6.8 (1.1-17.8)
<0.001
2.3 (0-7.1)
2.8 (0-11.1)
0.030
% Other race
2.7 (0.5-6.4)
3.7 (1.1-7.9)
2.9 (0.4-7.0)
<0.001
1.5 (0-4.0)
1.8 (0-5.8)
0.001
% of Households with no car
4.3 (1.4-10.2)
3.0 (0-7.0)
16.1 (8.5-25.4)
<0.001
3.2 (0.9-6.2)
7.2 (3.6-12.0)
<0.001
% of Households with children
30.7 (23.1-39.1)
35.0 (26.4-42.9)
<0.001
29.4 (22.9-36.0)
32.4 (26.1-39.7)
<0.001
30.0 (21.8-39.3)
CBGs with Accessible SNAP Store,g Full-Service Store, and Limited-Service Store Proximity Comparison n[4,619
Lowc (n[1,594)
Highd (n[905)
P valuee
Lowc (n[1,637)
n (%)!
May 2018 Volume 118 Number 5
Limited-variety SNAP store closer than full-variety Full-variety SNAP store closer than limited-variety a
4,072 (88.2)
1,305 (81.9)
547 (11.8)
289 (18.1)
838 (92.6) 67 (7.4)
P valuee
n (%)! <0.001
1,490 (91.0)
439 (90.9)
147 (9.0)
44 (9.1)
SNAP¼Supplemental Nutrition Assistance Program. CBG¼census block group. c Low concentration of SNAP households are CBGs with the percent of SNAP households <75th percentile. In these CBGs, <23% of household participate in SNAP. d High concentration of SNAP households are CBGs with the percent of SNAP households 75th percentile. In these CBGs, 23% of households participate in SNAP. e Difference in low to high CBGs. f IQR¼interquartile range. g Access is defined as a SNAP store located within 1 mile in urban areas and within 10 miles in rural areas. b
Highd (n[483)
0.930
RESEARCH
842
Table 2. Description of census block groups in North Carolina in 2015 (n¼6,092)
RESEARCH non-Hispanic black (P<0.001). About 7% of households in high SNAP block groups did not have a car compared with 3% in low SNAP block groups (P<0.001).
SNAP Store Access Proximity to Limited-Variety SNAP Stores vs FullVariety SNAP Stores. In North Carolina, residents often (in 88% of census block groups) live closer to limited-variety SNAP stores compared to full-variety SNAP stores. Among urban census block groups, residents in high SNAP block groups have closer access to limited-variety SNAP stores in 93% of the high SNAP block groups compared with low SNAP block groups, where 82% of the census block groups have a limited-variety SNAP store closer (Table 2).
Access. Based on Table 3 and Figure 2, the majority of urban census block groups do not have a full-variety SNAP store within 1 mile (73% of low SNAP block groups and 58% of high SNAP block groups). For larger cities, such as Charlotte, Raleigh, and Greensboro, census block groups with high SNAP participation and no access to a full-variety store are clustered and indicate areas where the need for a full-variety store is greatest. Limited-variety SNAP stores are not accessible in about half (48%) of low SNAP block groups. Among high SNAP block groups, about 16% do not have a limitedvariety SNAP store accessible. In rural areas, about 16% of census block groups did not have a full-variety SNAP food store accessible (within 10 miles), while <2% of census block groups did not have a limited-variety SNAP food store accessible. There was no difference in access by SNAP concentration in rural areas.
Distance if Accessible. In urban areas, as mentioned, the majority of census block groups do not have an accessible fullservice SNAP store. However, among the census block groups that do have a full-service store, the distance to that store for residents is about 0.7 miles. This distance does not differ by the concentration of SNAP households. Among the census block groups with an accessible limited-variety store, the distance to the limited-variety store was shorter for residents in high SNAP block groups (0.4 miles) compared with 0.6 miles for low SNAP block groups. In rural areas, about 85% of the census block groups have a full-service store accessible. Among these accessible census block groups, the median distance to a full-service store is 4.9 miles in low SNAP block groups and 4.4 miles in high SNAP block groups. Virtually all of the rural census block groups have access to a limited-variety SNAP store; the median distance to the store is 2.7 miles in low SNAP block groups and 2.1 miles in high SNAP block groups. Concentration of SNAP Stores. In urban areas, there is a difference in the number of accessible SNAP stores. Across both high and low SNAP block groups, the median number of accessible full-variety SNAP stores is zero. However, among limited-variety SNAP stores, there is typically one store in low SNAP block groups (0 stores/1,000 residents) and four in high SNAP block groups (3 stores/1,000 residents). In rural areas, there are typically three full-variety (2 stores/1,000 residents) and 18 limited-variety (13 stores/1,000 residents) SNAP stores accessible in both high and low SNAP block groups. Figure 3 compares the concentration of full-variety stores to low-variety stores within range for each block group. In May 2018 Volume 118 Number 5
urban areas, there is a great deal of variation between access to full and limited stores. Results in Charlotte, Greensboro, and Raleigh suggest pockets of census block groups with a high concentration of limited-variety stores and a low concentration of full-service stores. Few urban census block groups exhibit a high concentration of both full- and limitedvariety stores, although the rural areas in the immediate vicinity of these cities often do. These rural census block groups tend to have access to both full-variety stores and limitedvariety stores within the allotted 10 miles, although exceptions can be found throughout the state. Finally, urban census block groups at the periphery of those cities suffer from a lack of limited- and full-services stores.
DISCUSSION About half of SNAP households in North Carolina live in census block groups with a high concentration of SNAP households. It is important that SNAP participants living in both high and low SNAP block groups have adequate access to SNAP stores. In North Carolina, the SNAP food environment landscape exhibits significant differences among rural and urban areas. Within rural areas, there were small differences in store accessibility between low SNAP block groups and high SNAP block groups. In general, SNAP households live about 2.5 miles from a limited-variety SNAP store and 4.5 miles from a full-variety SNAP store. Within a 10-miles drive, those in rural areas likely have access to 3 full-variety SNAP stores and 18 limited-variety SNAP stores. This is a sharp contrast from urban areas, where a disparity in the availability of SNAP stores within a 1-mile distance between low and high SNAP block groups is observed. Individuals living in a low SNAP block groups typically do not have access to a full-variety food store, but do have access to 1 limited-variety food store. However, individuals living in high SNAP block groups do not have access to a full-variety food store, but have access to a higher number (ie, four) of limited-variety SNAP food stores. The differences in the SNAP food environment between rural and urban areas are partly due to the difference in the definition of the “accessible” range: 1 mile in urban areas and 10 miles in rural areas. In urban areas, households in high SNAP block groups were five times more likely to not have access to a car (16% compared to 3% for households living in low SNAP block groups). It is unclear whether those households without a car also participated in SNAP. In rural areas, 7% of households in high SNAP block groups did not have a car, double the percent in low SNAP block groups. For those in rural areas without vehicle access, the accessibility definition of 10 miles may not be a reasonable measure of accessibility. Previous research by Ver Ploeg and colleagues29 found that 34% of SNAP households did not use their own vehicle to get to their grocery store; the majority of these households (61%) were able to carpool with someone else and the other 39% “walk, bike, or take public transit or a shuttle to the store.” However, Ver Ploeg and colleagues’ research did not stratify the SNAP households by rural/urban status.29 Additional research to determine the percent of SNAP households without a car or other modes of reliable transportation in urban and rural areas is warranted. In 2015, 87% of SNAP redemptions (about $2 billion) were made at full-variety stores, even though the majority of SNAP households in urban areas did not have adequate access to fullJOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
843
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
Minimum Distance (in Miles) to SNAP Store for CBGs with Accessb
CBGsa without Accessb to SNAPc Store Variable Concentration of SNAP households in urban CBGs
Lowd (n[2,891)
Highe (n[1,059) P valuef Lowd
%!
Highe
No. of Stores within Access Range/per 1,000 Residents
No. of Stores within Accessb Range P valuef Lowd
median (IQRg)!
Highe
P valuef Lowd
median (IQRg)!
Highe
P valuef
median (IQRg)!
Full variety of roods
72.7
58.0
<0.001
0.71 (0.55-0.86) 0.70 (0.53-0.85)
0.267
0 (0-1)
0 (0-1)
<0.001
0 (0-27)
0 (0-10)
<0.001
Limited variety of roods
48.1
15.6
<0.001
0.61 (0.43-0.79) 0.44 (0.28-0.63) <0.001
1 (0-3)
4 (1-8)
<0.001
0 (0-108)
3 (0-75)
<0.001
Lowd (n[1,659)
Highe (n[483)
P valuef Lowd
Highe
P valuef Lowd
Highe
P valuef
Concentration of SNAP households in rural CBGs Full variety of foods May 2018 Volume 118 Number 5
Limited variety of foods a
%!
Highe
P valuef Lowd
median (IQRg)!
median (IQRg)!
16.2
16.1
0.97
4.88 (3.21-6.77) 4.44 (2.60-6.84)
0.019
1.3
0.4
0.09
2.66 (1.67-4.00) 2.08 (1.20-3.37) <0.001
3 (1-8)
3 (1-6)
19 (9-38) 17 (8-34)
CBG¼census block group. Access is defined as a SNAP store located within 1 mile in urban areas and within 10 miles in rural areas. c SNAP¼Supplemental Nutrition Assistance Program. d Low concentration of SNAP households are CBGs with the percent of SNAP households <75th percentile. In these CBGs, <23% of household participate in SNAP. e High concentration of SNAP households are CBGs with the percent of SNAP households 75th percentile. In these CBGs, 23% of households participate in SNAP. f Difference in low to high CBGs. g IQR¼interquartile range. b
median (IQRg)!
0.002 0.05
2 (0-560)
2 (0-33)
0.128
13 (0-1,680) 13 (0-230) 0.181
RESEARCH
844
Table 3. Supplemental Nutrition Assistance program food store availability among neighborhoods with a high and low concentration of Supplemental Nutrition Assistance Program participants by urban/rural status in North Carolina
RESEARCH
Figure 2. Spatial distribution of census block groups without a full-service food store in range (urban census block group: within 1 mile; rural census block groups: within 10 miles). variety stores. Based on the high redemption rate at fullvariety stores, it is assumed that SNAP participants prefer to use their benefits at full-variety stores and will overcome accessibility barriers to do so. Shannon30 examined the relationship between location of SNAP recipient and “outflow” of SNAP benefits from the participants neighborhood. He found that many of the full-variety stores were located in middleclass suburban areas and that about 80% of all SNAP benefits are redeemed at full-variety stores. However, he also found that SNAP participants living in dense urban areas with limited availability to full-variety stores only spent 51% of their SNAP benefits at full-variety stores. The remaining 49% of SNAP benefits were spent at limited-variety stores. Recent research examining the food and beverage purchases made at a supermarket chain in 2011 found that SNAP participants spent the most on the following 5 items: soft drinks, fluid milk products, ground beef, bagged snacks, and cheese.31 Four of these items are likely available at most SNAP stores (whether full-variety stores or limited-variety stores): soft drinks, fluid milk products, bagged snacks, and cheese. Because fresh meat is rarely available at limited-variety stores,32-35 the demand for meat and other foods may draw SNAP participants to fullvariety stores. Limited SNAP stores are much more accessible to the majority of SNAP households. Better accessibility may account for 13% (about $300 million) of SNAP redemptions at limited-variety stores. As the majority of SNAP households redeem their SNAP benefits at SNAP stores outside of their immediate vicinity, May 2018 Volume 118 Number 5
this could be an issue in two ways. First, SNAP participants that live relatively far from a full-variety SNAP store may feel pressured to purchase more shelf-stable, processed foods at a limited-variety SNAP store, which may result in less-healthy food purchases. Second, should SNAP participants be expected to travel long distances in order to redeem their benefits at full-variety stores? This is a question that should be debated among policy makers. While SNAP food stores are required to offer a minimum number of food types and varieties, there is currently no consideration for the need for that particular type of SNAP food store when determining SNAP food store authorization. In urban areas, SNAP participants in high SNAP block groups have access to more limited service stores compared to those in low SNAP block groups. In addition, high SNAP census block groups are disproportionally populated with more non-Hispanic black residents (47% vs 12%), Hispanic residents (7% vs 5%), lower-income households ($26,223 vs $52,232), households without a car (16% vs 3%), and households with a child (35% vs 30%). Research to study how exposure to more limited-variety stores impacts health among these populations is warranted. The USDA Food and Nutrition Service has posted new SNAP retailer stocking requirements.36 These requirements increase the number of stocking units by staple food variety from 1 to 3. It is unclear when this requirement will take effect and, when it does, it is not clear how it will impact the SNAP-authorized stores. Perhaps some limited SNAPJOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
845
RESEARCH
Figure 3. Density of full-service vs limited-service food stores in range for each census block group (urban census block group: within 1 mile; rural census block groups: within 10 miles).
authorized stores will discontinue participation in SNAP. Once implemented, research is warranted to study the impact of this regulation on SNAP store access. There are some limitations of this study. First, the USDA definition for “low access” was used; that may not give a true assessment of accessibility from the SNAP participant’s perspective. Second, the availability of foods and beverages at limited-variety food stores was not assessed, assumptions about the foods/beverages available were made based on USDA SNAP store definitions and previous research on the foods/beverages typically available at convenience stores, drugstores, and dollar stores.32-34,37 Some of the store types (like small and/or medium grocery stores) may have had a greater variety of options than others. Third, proximity from the weighted centroid of each census block group was estimated. This level of geographic simplification may bias travel estimates and future research is needed to evaluate food accessibility at finer geographic scales. Fourth, the route between the centroid and the SNAP retailer is assumed to be the shortest one, which may not reflect reality. Individuals may prefer more familiar routes to combine their trips (tripchaining). Fifth, it is assumed that individuals either drive or walk to each SNAP retailer and do not use public transportation, such as bus or subway. Recent research by Widener38 has shown that the use of public transit can dramatically affect the geographic access to food retailers. Sixth, this analysis is limited to SNAP-authorized food stores, this limits the inclusion of other food stores that do not 846
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
participate in SNAP. Therefore, communities likely have access to additional food stores than are presented in this analysis. To the authors’ knowledge, no research has been done to compare the characteristics of SNAP-authorized and nonSNAP-authorized food stores. This is an area for further research. Seventh, the descriptive statistics (Wilcoxon ranksum and c2 tests) assumed spatial independence among observation units. Spatial dependence (or autocorrelation) suggests that nearby census block groups are likely to exhibit similar results, and the assumption of independence is likely to overestimate the statistical significance of the results, such as P values.39,40 Eighth, although there exists substantial variation in the characteristics of low- vs high-SNAP census block groups in urban areas, more research is needed to evaluate the sensitivity of this analysis at different spatial scales. Additional variables, such as population density, may shed some light on such variation. Finally, this analysis was based on North Carolina, therefore, results may not be generalizable to all states.
CONCLUSIONS In closing, the results from this study highlight the importance of considering location and type of SNAP stores. While there are more than 9,500 SNAP stores in North Carolina, few full-variety stores are easily accessible to SNAP participants in urban areas. Also, >80% of the SNAP stores are limited variety, which tend to offer a poor selection of healthy food May 2018 Volume 118 Number 5
RESEARCH options—particularly fresh fruits and vegetables. In the future, researchers are encouraged to consider access to SNAP stores by type and availability of reliable transportation when evaluating the impact of SNAP on food security and health.
20.
US Census Bureau. American Community Survey, 2011-2015; American Community Survey 5-year estimates. http://factfinder2. census.gov/. Accessed October 10, 2017.
21.
US Census Bureau. Centers of population. https://www.census.gov/ geo/reference/centersofpop.html. Accessed October 11, 2017.
22.
US Census Bureau, Geography Division. Centers of Population Computation for the United States 1950-2010. https://www2.census. gov/geo/pdfs/reference/cenpop2010/COP2010_documentation.pdf. Published 2011. Accessed October 11, 2017.
23.
US Department of Agriculture, Economic Research Service. Food environment atlas: Data access and documentation. http://www.ers. usda.gov/data-products/food-access-research-atlas/documentation. aspx. Published 2013. Accessed October 10, 2017.
24.
Dressler H, Smith C. Food choice, eating behavior, and food liking differs between lean/normal and overweight/obese, low-income women. Appetite. 2013;65:145-152.
US Census Bureau. 2010 Census urban and rural classification and urban area criteria. https://www.census.gov/geo/reference/ua/ urban-rural-2010.html. Accessed October 31, 2017.
25.
US Census Bureau. 2010 Census Data. https://www.census.gov/201 0census/data/.
Poston WS 2nd, Foreyt JP. Obesity is an environmental issue. Atherosclerosis. 1999;146(2):201-209.
26.
ArcGIS [computer program]. Version 10.4. Redlands, CA: ESRI; 2016.
27.
MySQL [computer program]. Version 5.7. Redwood Shores, CA: Oracle Corporation; 2015.
28. 29.
SAS [computer program]. Version 9.3. Cary, NC: SAS Institute; 2011. Ver Ploeg M, Mancino L, Todd JE, Clay DM, Scharadin B. Where do Americans usually shop for food and how do they travel to get there? Initial findings from the National Household Food Acquistion and Purchase Survey. Econ Inform Bull. 2015;138.
30.
Shannon J. What does SNAP benefit usage tell us about food access in low-income neighborhoods? Soc Sci Med. 2014;107:89-99.
31.
Garasky S, Mbwana K, Romualdo A, Tenaglio A, Roy M. Foods typically purchased by Supplemental Nutrition Assistance Program (SNAP) households—Appendices 2016. http://www.fns.usda.gov/ sites/default/files/ops/SNAPFoodsTypicallyPurchased-Appendices.pdf. Accessed October 11, 2017.
References 1.
2.
3.
4.
US Department of Agriculture, Food and Nutrition Service. Supplemental Nutrition Assistance Program (SNAP) participants and costs. http://www.fns.usda.gov/pd/supplemental-nutrition-assistance-programsnap. Published 2017. Accessed October 10, 2017. Castner L, Henke J. Benefit redemption patterns in the Supplemental Nutrition Assistance Program. http://www.fns.usda.gov/sites/ default/files/ARRASpendingPatterns.pdf. Published 2011. Accessed October 10, 2017.
5.
Block JP, Scribner RA, DeSalvo KB. Fast food, race/ethnicity, and income: A geographic analysis. Am J Prev Med. 2004;27(3):211-217.
6.
Drewnowski A, Specter SE. Poverty and obesity: The role of energy density and energy costs. Am J Clin Nutr. 2004;79(1):6-16.
7.
Walker RE, Keane CR, Burke JG. Disparities and access to healthy food in the United States: A review of food deserts literature. Health Place. 2010;16(5):876-884.
8.
Alwitt LF, Donley TD. Retail stores in poor urban neighborhoods. J Consum Aff. 1997;3(1):139-164.
9.
Chung C, Myers SL. Do the poor pay more for food? An analysis of grocery store availability and food price disparities. J Consum Aff. 1999;33(2). 2776-2296.
10.
Lewis LB, Galloway-Gilliam L, Flynn G, Nomachi J, Keener LC, Sloane DC. Transforming the urban food desert from the grassroots up: A model for community change. Fam Community Health. 2011;34(suppl 1):S92-S101.
32.
McEntee J, Agyeman J. Towards the development of a GIS method for identifying rural food deserts: Geographic access in Vermont, USA. Appl Geogr. 2010;30(1):165-176.
Caspi CE, Pelletier JE, Harnack L, Erickson DJ, Laska MN. Differences in healthy food supply and stocking practices between small grocery stores, gas-marts, pharmacies and dollar stores. Public Health Nutr. 2016;19(3):540-547.
33.
Racine EF, Batada A, Solomon CA, Story M. Availability of foods and beverages in Supplemental Nutrition Assistance Program-authorized dollar stores in a region of North Carolina. J Acad Nutr Diet. 2016;116(10):1613-1620.
34.
Racine EF, Kennedy A, Batada A, Story M. Foods and beverages available at SNAP-authorized drugstores in sections of North Carolina. J Nutr Educ Behav. 2017;49(8):674-683 e671.
11.
12.
Apparicio P, Cloutier MS, Shearmur R. The case of Montréal’s missing food deserts: Evaluation of accessibility to food supermarkets. Int J Health Geogr. 2007;6(4).
13.
Shaw HJ. Food deserts: Towards the development of a classification. Geogr Ann A. 2006;88(2):231-247.
14.
Forsyth A, Lytle L, Van Riper D. Finding food: Issues and challenges in using Geographic Information Systems to measure food access. J Transp Land Us. 2010;3(1):43-65.
35.
Laska MN, Borradaile KE, Tester J, Foster GD, Gittelsohn J. Healthy food availability in small urban food stores: A comparison of four US cities. Public Health Nutr. 2010;13(7):1031-1035.
15.
Kirby RS, Delmelle E, Eberth JM. Advances in spatial epidemiology and geographic information systems. Ann Epidemiol. 2017;27(1):1-9.
36.
16.
Chen X, Clark J. Interactive three-dimensional geovisualization of spaceetime access to food. Appl Geogr. 2013;43:81-86.
US Department of Agriculture. Final Rule: Enhancing Retailer Standards in the Supplemental Nutrition Assistance Program (SNAP). https://www.fns.usda.gov/snap/fr-121516. Accessed December 11, 2017.
17.
Wood BS, Horner MW. Understanding accessibility to SNAPaccepting food store locations: Disentangling the roles of transportation and socioeconomic status. Appl Spat Anal Policy; 2015: 1-19.
37.
Laska MN, Caspi CE, Pelletier JE, Friebur R, Harnack LJ. Lack of healthy food in small-size to mid-size retailers participating in the Supplemental Nutrition Assistance Program, Minneapolis-St. Paul, Minnesota, 2014. Prev Chronic Dis. 2015;12:E135.
18.
Rigby S, Leone AF, Kim H, et al. Food deserts in Leon County, FL: Disparate distribution of Supplemental Nutrition Assistance Program-accepting stores by neighborhood characteristics. J Nutr Educ Behav. 2012;44(6):539-547.
38.
Widener MJ. Comparing measures of accessibility to urban supermarkets for transit and auto users. Prof Geogr; 2016:1-10.
39.
Datta S, Satten GA. A signed-rank test for clustered data. Biometrics. 2008;64(2):501-507.
Racine EF, Wang Q, Laditka SB, Johnson CR, Mignery A. The characteristics and concentration of SNAP approved stores and community health. J Hunger Environ Nutr. 2013;8(3):350-361.
40.
Lettenmaier DP. Detection of trends in water quality data from records with dependent observations. Water Resources Res; 1976: 1037-1046.
19.
May 2018 Volume 118 Number 5
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
847
RESEARCH AUTHOR INFORMATION E. F. Racine and E. Delmelle are associate professors, E. Major is a graduate student, and C. A. Solomon is a doctoral student, University of North Carolina at Charlotte. Address correspondence to: Elizabeth F. Racine, DrPH, RD, University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte, NC 28223. E-mail: [email protected]
STATEMENT OF POTENTIAL CONFLICT OF INTEREST No potential conflict of interest was reported by the authors.
FUNDING/SUPPORT This research was funded by Healthy Eating Research, a national program of the Robert Wood Johnson Foundation.
848
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
May 2018 Volume 118 Number 5