RESEARCH AND PRACTICE

Relative and Absolute Availability of Healthier Food

and Beverage Alternatives Across Communities
in the United States
Shannon N. Zenk, RN, PhD, MPH, Lisa M. Powell, PhD, Leah Rimkus, RD, MPH, Zeynep Isgor, PhD, Dianne C. Barker, MPH,
Punam Ohri-Vachaspati, RD, PhD, and Frank Chaloupka, PhD

Poor diet is common in the United States,

Objectives. We examined associations between the relative and absolute
particularly among disadvantaged groups,
availability of healthier food and beverage alternatives at food stores and
and is a risk factor for obesity and numerous
community racial/ethnic, socioeconomic, and urban–rural characteristics.
chronic diseases.1,2 The retail food environ-
Methods. We analyzed pooled, annual cross-sectional data collected in 2010 to
ment may present barriers to healthy eating in 2012 from 8462 food stores in 468 communities spanning 46 US states. Relative
Black, low-income, rural, and central-city com- availability was the ratio of 7 healthier products (e.g., whole-wheat bread) to
munities in the United States. Considerable less healthy counterparts (e.g., white bread); we based absolute availability on
research demonstrates that healthier foods and the 7 healthier products.
beverages, such as low-fat dairy and whole- Results. The mean healthier food and beverage ratio was 0.71, indicating
grain products, are less available in Black, low- that stores averaged 29% fewer healthier than less healthy products. Lower
income, rural, and central-city communities relative availability of healthier alternatives was associated with low-income,
than in White, higher-income, and urban and Black, and Hispanic communities. Small stores had the largest differences:
relative availability of healthier alternatives was 0.61 and 0.60, respectively,
suburban communities.3---6 Other studies show
for very low-income Black and very low-income Hispanic communities, and
no systematic differences in healthier food
0.74 for very high-income White communities. We found fewer associations
availability across communities.7,8 Some con-
between absolute availability of healthier products and community charac-
tend that it is not just absolute availability of teristics.
foods and beverages but rather the relative Conclusions. Policies to improve the relative availability of healthier alterna-
availability of less healthy products compared tives may be needed to improve population health and reduce disparities. (Am J
with healthier alternatives that may influence Public Health. 2014;104:2170–2178. doi:10.2105/AJPH.2014.302113)
food choices.9---12 Although studies generally
have not compared relative availability across
communities, relative availability of healthier In one of the first nationwide studies of directly Project.20 We measured aspects of the retail
to less healthy foods was related to lower risk of observed food availability in the United States, we food environment via direct observation in
overweight and obesity in 1 study.9 examined associations between the relative and 468 communities spanning 46 states. The
Differences in store composition across absolute availability of healthier food and bever- observed communities were school enrollment
communities may partially explain observed age alternatives and community characteristics zones for nationally representative samples
community differences in absolute and relative (racial/ethnic composition, socioeconomic char- of grade 8, 10, and 12 public school students
availability of healthier food and beverage acteristics, and urbanicity). Our study was novel in the continental United States from the Mon-
products. Inequities in the spatial accessibility not only because of its focus on relative avail- itoring the Future study. (A more detailed
of supermarkets and other retail food stores, ability of healthier alternatives, but also because it description of the Monitoring the Future
such as convenience stores, are well docu- examined communities nationwide, compared sample is provided elsewhere.21)
mented, with low-income, Black, rural, and the food environment across a continuum of Food stores (i.e., stores that sold snacks and
central-city communities having less access to community characteristics (e.g., urban---rural, pre- drinks at a minimum) were sampled annually
supermarkets, for example.13---16 Supermar- dominant race/ethnicity, high---low income), and to obtain a representative sample of super-
kets have greater availability of healthier had ample statistical power to detect differences markets, grocery stores, and limited-service
products than do convenience stores and because of the large number of stores. stores (convenience stores, specialty stores,
many small grocery stores.10,17,18 Yet research liquor stores, drug stores, dollar or discount
shows that differences in healthier food and METHODS stores, general merchandise stores, or produce
beverage availability across communities markets) in each community.22 We obtained
persist among stores of the same type or after Our cross-sectional study drew on 3 years of business lists from both InfoUSA and Dun
accounting for differences in the types of pooled data (2010---2012) from the Bridging and Bradstreet by using relevant Standard
stores present.19 the Gap Community Obesity Measures Industrial Classification codes23 and specific

2170 | Research and Practice | Peer Reviewed | Zenk et al. American Journal of Public Health | November 2014, Vol 104, No. 11
 RESEARCH AND PRACTICE

keyword searches (e.g., “dollar”; details are in Field staff conducted almost all observations beverage items or healthier alternatives (n = 7),
Table A, available as a supplement to the (99.8%) Monday through Friday and the ma- we assigned 1 as the ratio value, corresponding
online version of this article at [Link] jority (75%) during regular business hours with an equivalent number of healthier to less
[Link]). We compared and deduplicated (8 AM---5 PM). They usually completed data col- healthy products. Sensitivity analyses removing
these lists to create a single list. We screened lection in any given community within 1 week. these 11 recoded observations revealed con-
each business by telephone to confirm its Field staff had discretion regarding whether sistent findings.
existence, ascertain whether it met eligibility to ask permission from an owner or employee We created 2 measures of absolute avail-
criteria, and obtain information for store clas- first before collecting data at a store. They ability of healthier food and beverage alterna-
sification (e.g., availability of fresh meat, pres- carried copies of a letter that explained the tives. The first measure was a count of the 7
ence of service counters [butcher, bakery, study, which they could provide to store healthier alternatives that were part of the
delicatessen]). owners or employees as needed. On average, study’s relative availability measure. The ex-
We then randomly selected a sample of data collection took 37.7 614.2 minutes in panded 13-item version was a count of these 7
stores in each community by store type (su- supermarkets, 24.4 611.1 minutes in grocery healthier alternatives plus availability of at least
permarket, grocery, limited service). Because of stores, and 19.2 610.6 minutes in limited- 1 variety of each of the following: fresh fruit,
limitations previously identified in the validity service stores. fresh vegetables, frozen fruit, frozen vegetables,
of commercial business lists,24,25 we implemented canned fruit, and canned vegetables.
a half-open interval procedure to help address Measures Store type, community characteristics, and
error stemming from an incomplete list of Availability of healthier food and beverage covariates. We classified food stores into 3
businesses.26 This involved field staff identifying alternatives. The study measured relative food types according to operational definitions de-
and observing additional eligible food stores not and beverage availability at each outlet with rived from the Food Marketing Institute29 and
on the list but discovered in the field. the Bridging the Gap Food Store Observation previous observational studies.30---32 Super-
Form.27 The measure of relative availability of markets sold fresh meat (uncooked, unpro-
Data Collection healthier food and beverage alternatives was cessed, not frozen), had at least 4 cash registers,
Across the 468 communities, field staff the ratio of counts of 7 healthier food and and had at least 2 of the following staffed
identified 9226 eligible food stores and com- beverage products to 7 matched less healthy service counters: butcher, bakery, and delica-
pleted data collection at 95.3% (n = 8793). products. The less healthy products consisted tessen. Grocery stores sold fresh meat but
Field staff completed a portion of the data of whole milk, cheese, high-sugar cereal (‡ 6 g did not meet the other supermarket criteria.
collection at 1.8% (n = 166) of eligible stores, sugar/serving), white bread, potato chips, fruit Limited-service food stores, which did not sell
most of which were not finished because store drink (< 50% juice), and regular soda (car- fresh meat, were convenience stores, dollar or
personnel asked field staff to leave after they bonated, sweetened beverage or soft drink). discount stores, drug stores, general merchan-
had initiated data collection (n = 154). The matched healthier alternatives, identified dise stores, liquor stores, or produce markets.
Data collection did not begin at 2.9% (n = according to current dietary guidelines as lower We measured community racial/ethnic and
267) of eligible stores because store personnel in fat, lower in added sugar, or higher in fiber socioeconomic characteristics with 5-year esti-
asked field staff to leave before they had or whole-wheat content, consisted of low-fat mates from the American Community Survey
initiated data collection (n = 222), the store milk (skim or 1%), low-fat cheese (labeled as (2007---2011),33 aggregated according to cen-
was temporarily inaccessible (n = 18), the low fat, reduced fat, or made with skim, 1%, or sus block groups, with centroids falling within
business was not located where it was origi- 2% milk), low-sugar cereal (< 6 g sugar per community boundaries. We classified racial/
nally confirmed to be (n = 16), field staff felt serving), 100% whole-wheat bread, baked or ethnic composition as Hispanic (‡ 50% resi-
unsafe (n = 3), or another reason (n = 8). low-fat potato chips (< 4 g fat/1 oz serving), dents Hispanic or Latino), Black (‡ 50% non-
Each year, field staff, all of whom were based 100% orange juice, and diet soda (carbonated, Hispanic Black), White (‡ 50% non-Hispanic
at 1 location, completed a 3-week training artificially sweetened beverage or soft drink).28 White), and other race/ethnicity (i.e., majority
course on how to conduct systematic observa- Cohen j coefficients for interrater reliability of another race/ethnicity or no group having
tions of multiple aspects of the local community for these items ranged from 0.69 (juice drink) majority representation). We assessed socio-
environment (e.g., parks, fitness facilities, fast- to 1.00 (whole milk), and the proportion of economic characteristics by median household
food restaurants, streetscapes). The course overall agreement (not corrected for chance income in quintiles. We based urbanicity cat-
dedicated 2.5 days to the Bridging the Gap agreement or sensitive to highly skewed dis- egorizaton on National Center for Education
Food Store Observation Form,27 providing tributions) ranged from 0.86 (white bread) to Statistics urban-centric locale codes as urban
classroom exercises and field practice at the 0.99 (regular soda).27 (small, midsize, and large cities), suburban
different types of food stores. We supple- For stores without any of the less healthy (small, midsize, and large suburbs plus distant
mented this initial training with periodic foods or beverages (n = 4; resulting in a divisor and fringe towns), or rural (distant, fringe, and
in-person debriefings and refresher courses of 0 and thus an undefined ratio value), we remote rural areas plus remote towns).34
throughout the approximately 16-week field assigned the highest observed value of the Several other control covariates helped to
season. ratio. For stores with no less healthy food or account for potential variations in social norms

November 2014, Vol 104, No. 11 | American Journal of Public Health Zenk et al. | Peer Reviewed | Research and Practice | 2171
 RESEARCH AND PRACTICE

or food-stocking patterns: census division, to data values. These analyses allowed us to relative to very high-income communities.
capture regional differences (Northeast, Middle report estimated differences in relative avail- Racial/ethnic differences in relative availability
Atlantic, South Atlantic, East North Central, ability for contrasting communities of interest. of healthier alternatives remained after adjust-
East South Central, West North Central, West We conducted all analyses with Stata ver- ment for community income level, although
South Central, Mountain, Pacific); year, day of sion 11.2 (Statacorp LP, College Station, TX). the magnitudes of the Black and Hispanic
week, and start time of the data collection; and We applied sample weights to account for the coefficients were reduced by about one third.
a seasonality control (number of days from the complex sample design and obtain results that Differences by urbanicity were eliminated after
date of data collection to June 1 of the data were nationally representative of communities adjustment for median household income. As
collection year). where 8th-, 10th-, and 12th-grade public expected, relative availability of healthier al-
school students reside in the continental United ternatives was strongly patterned by store type
Data Analysis States. Approximately 6% of stores were miss- (model 3). The addition of store type in model
We calculated descriptive statistics for the ing data on 1 or more of the variables of 3 attenuated associations for racial/ethnic
sample overall and by store type (supermarket, interest for our analysis; we applied complete composition and income.
grocery, or limited service). We used ordinary case analysis and based our results on data Table 3 provides associations between the
least squares regression to estimate multivari- from the remaining 8462 stores. relative availability of healthier alternatives
ate associations. To determine the extent to and community characteristics, stratified by
which demographics and store type accounted RESULTS store type. Among supermarkets (model 1),
for community differences, we used a hierar- relative availability of healthier alternatives
chical approach. Model 1 regressed relative Table 1 shows descriptive statistics of our was 0.05 units lower in Hispanic than in White
availability of healthier alternatives on com- sample. The mean relative availability of communities. By income, only very low-income
munity racial/ethnic composition, urbanicity, healthier alternatives ratio was 0.71, indicating communities had lower relative availability
and covariates. Model 2 added community that stores had, on average, 29% fewer of healthier alternatives (–0.02) than did very
median household income. Model 3 added healthier alternatives than less healthy coun- high-income communities. Among grocery
store type. We estimated stratified models terpart food and beverage products. Overall, stores, the only significant difference was that
regressing relative availability on community very few stores (0.9%) had a greater number of urban stores had lower relative availability
racial/ethnic composition, income, urbanicity, healthier alternatives than less healthy food than did suburban stores (–0.09). Among
and covariates separately for supermarkets, and beverage products; most (77.5%) had limited-service stores, both Hispanic (–0.05)
grocery stores, and limited-service stores to more of the less healthy products than the and Black (–0.04) communities had lower
determine whether any racial/ethnic, socio- healthier alternatives. However, the relative relative availability than did White communi-
economic, or urban---rural associations differed availability of healthier alternatives varied by ties. Moreover, we observed a similar income
across stores types. Substituting absolute store type, with mean ratios of 0.98 at super- gradient for limited-service stores as for the
availability of healthier products as a depen- markets, 0.72 at grocery stores, and 0.68 at overall sample, with lower-income communities
dent variable in model 3 allowed us to examine limited-service stores. having progressively lower relative availability of
whether any community variations were spe- Table 2 presents associations between the healthier alternatives.
cific to the relative availability measure. relative availability of healthier alternatives The fully adjusted models predicted relative
To assist with interpretation of ordinary least and community characteristics. Relative avail- availability of healthier alternatives overall and
squares regression coefficients in the fully ability of healthier alternatives was 0.08 units for 3 different types of communities (very
adjusted models for relative availability, we lower at stores located in Hispanic communi- low-income Black, very low-income Hispanic,
estimated predicted relative availability of ties and 0.06 units lower in Black communities, and very high-income White), shown in the
healthier alternatives for all stores as well as as well as 0.03 units lower in communities of lower panels of Tables 2 and 3. Although the
for supermarkets, grocery stores, and limited- another racial/ethnic composition, than in average predicted relative availability of
service stores. From our regression results, we White communities (model 1). Relative avail- healthier alternatives ratio was 0.71 overall, we
derived the predicted relative availability of ability of healthier alternatives was also lower found a 10 to 12 percentage point difference
healthier alternatives for 4 scenarios: (1) com- in both urban and rural communities than in between the predicted relative availability of
munities with all covariates at their original suburban communities. healthier alternatives for very low-income
data values; (2) communities set as being very In model 2, which added median household Black (0.65) and Hispanic (0.63) communities
low income and majority Black, with all other income, a graded association between income and for very high-income White communities
covariates at their original data values; (3) level and relative availability of healthier al- (0.75; Table 2). Among supermarkets, we
communities set as being very low income and ternatives was evident, with relative availability observed a 6.6 percentage point difference
majority Hispanic, with all other covariates at declining with decreasing income. These dif- between the predicted relative availability for
their original data values; and (4) communities ferences ranged from a 0.10-unit difference in very low-income Hispanic communities and
set as being very high income and majority communities with very low income to a 0.04- very high-income White communities (Table
White, with all other covariates at their original unit difference in high-income communities, 3). The predicted relative availability of

2172 | Research and Practice | Peer Reviewed | Zenk et al. American Journal of Public Health | November 2014, Vol 104, No. 11
 RESEARCH AND PRACTICE

TABLE 1—Weighted Descriptive Statistics for the Full Sample and by Store Type for Relative Availability of Healthier Food and Beverage
Alternatives: Bridging the Gap Community Obesity Measures Project, United States, 2010–2012

All Stores (n = 8462), Supermarkets (n = 928), Grocery Stores (n = 813), Limited-Service Stores (n = 6721),
Food Availability Mean (SD) or % Mean (SD) or % Mean (SD) or % Mean (SD) or %

Relative availability of healthier food and beverage alternatives 0.71 (0.24) 0.98 (0.07) 0.72 (0.32) 0.68 (0.22)
Absolute availability of healthier food and beverage productsa (n = 7) 5.63 (1.53) 6.99 (0.15) 6.2 (1.66) 5.44 (1.50)
Absolute availability of healthier food and beverage productsb (n = 13) 6.58 (3.16) 12.82 (0.64) 9.39 (3.61) 5.69 (2.28)
Absolute availability of less healthy food and beverage products (n = 7) 4.03 (1.80) 6.87 (0.51) 4.61 (2.41) 3.68 (1.53)
Relative availability of healthier alternatives, ratioc
< 1.0 77.5 9.5 71.4 85.0
1.0 21.6 90.3 27.7 14.0
> 1.0 0.9 0.2 0.9 1.0
Region
New England 5.8 4.8 5.3 5.9
Mid-Atlantic 12.0 11.9 12.7 12.0
West North Central 5.0 5.5 4.4 5.0
East North Central 10.8 10.5 12.2 10.7
South Atlantic 24.6 27.9 20.4 24.6
East South Central 5.3 2.8 5.6 5.5
West South Central 16.4 12.4 13.4 17.1
Mountain 4.1 6.6 2.9 3.9
Pacific 16.0 17.6 23.0 15.3
Urbanicity
Urban 41.2 35.9 43.6 41.5
Suburban 40.6 48.1 39.0 40.0
Rural 18.2 16.0 17.4 18.6
Racial/ethnic majority
Non-Hispanic White 69.6 77.6 60.7 69.5
Non-Hispanic Black 4.8 2.6 6.8 4.8
Hispanic 9.2 7.4 13.1 9.1
Non-Hispanic other 16.5 12.4 19.4 16.6
Median household income
Very high 18.8 30.6 18.3 17.6
High 23.3 24.6 22.3 23.3
Middle 18.7 16.1 17.5 19.1
Low 22.1 17.9 21.3 22.6
Very low 17.2 10.8 20.6 17.5
Store type
Supermarket 8.7 ... ... ...
Grocery 7.2 ... ... ...
Limited service 84.1 ... ... ...
a
Low-fat milk, low-fat cheese, low-sugar cereal, 100% whole-wheat bread, baked or low-fat potato chips, 100% orange juice, and diet soda.
b
The 7 healthier alternatives plus availability of at least 1 variety of each of the following: fresh fruit, fresh vegetables, frozen fruit, frozen vegetables, canned fruit, and canned vegetables.
c
Ratio > 1 = more healthier alternatives than less healthy products; 1 = equivalent number of healthier alternatives and less healthy products; < 1 corresponds with fewer healthier alternatives than
less healthy products.

healthier alternatives in limited-service stores differences appeared among limited-service and community characteristics. Stores located
for very low-income Black (0.61) and very stores, with an approximately 13 to 14 per- in non-White (Black, Hispanic, other) com-
low-income Hispanic (0.60) communities was centage point difference. munities had between 0.28 and 0.34 fewer
significantly lower than for very high-income Table 4 presents associations between the of the 7 healthier products, on average, than
White communities (0.74). Indeed, the largest absolute availability of healthier alternatives White communities, and very low-income

November 2014, Vol 104, No. 11 | American Journal of Public Health Zenk et al. | Peer Reviewed | Research and Practice | 2173
 RESEARCH AND PRACTICE

lower relative availability of healthier alterna-

TABLE 2—Associations Between Relative Availability of Healthier Food and Beverage tives even within stores that are present. Find-
Alternatives and Community Characteristics and Predicted Relative Availability: ings that racial/ethnic differences in relative
Bridging the Gap Community Obesity Measures Project, United States, 2010–2012 availability of healthier alternatives across store
Model 3 types were attenuated but persisted after ac-
counting for community income level and store
Predicted
Variable Model 1, b (SE) Model 2, b (SE) b (SE) Value (F) type provide additional evidence of environ-
mental barriers faced by racial/ethnic minority
Urbanicity communities.39---41
Suburban (Ref) 1.000 1.000 1.000 Although we observed an income gradient
Urban –0.027*** (0.009) –0.010 (0.009) –0.009 (0.009) for relative availability of healthier alternatives
Rural –0.027*** (0.010) –0.009 (0.010) –0.005 (0.009) in analyses of the full sample of stores, socio-
Racial/ethnic majority economic differences appeared only between
Non-Hispanic White (Ref) 1.000 1.000 1.000 very low-income and very high-income com-
Non-Hispanic Black –0.064*** (0.010) –0.042*** (0.012) –0.037** (0.012) munities for absolute availability of healthier
Hispanic –0.077*** (0.017) –0.052*** (0.015) –0.048*** (0.015) food and beverage products. Thus, studies that
Non-Hispanic other –0.033*** (0.012) –0.030*** (0.011) –0.023** (0.011) only examine absolute availability of healthier
Median household income food and beverage products and lack sufficient
Very high (Ref) 1.000 1.000 1.000 socioeconomic variability may fail to detect
Very low –0.095*** (0.013) –0.073*** (0.013) socioeconomic differences in healthier food
Low –0.069*** (0.012) –0.048*** (0.011) and beverage availability.6
Middle –0.067*** (0.012) –0.047*** (0.012) When we stratified by store type, the income
High –0.043*** (0.011) –0.029*** (0.011) gradient in relative availability of healthier
Store type alternatives was only evident for limited-
Supermarket (Ref) 1.000 1.000 1.000 service stores. Estimates of predicted relative
Grocery –0.247*** (0.012) availability of healthier alternatives showed
Limited service –0.294*** (0.005) particularly large differences between very
R2 0.038 0.051 0.168 low-income Black and Hispanic communities
Predicted relative availability and very high-income White communities
of healthier alternatives among limited-service stores. These findings
Overall 0.706 support the need for healthy corner store
Very low-income Black communities 0.645 (57.18***) initiatives in very low-income Black and His-
Very low-income Hispanic communities 0.634 (42.81***) panic communities to improve the balance of
Very high-income White communities 0.754 healthier versus less healthy options. Policy
Note. Models controlled for region, time of day, day of week, seasonality indicator, and year. changes modeled after the cost-neutral 2009
a
Adjusted Wald tests of significance for pairwise comparisons, relative to very high-income White communities. Supplemental Nutrition Program for Women,
*P < .05; **P < .01; ***P < .001. Infants, and Children food package revision
and 2014 Farm Bill provisions for the Sup-
communities had an average of 0.28 fewer healthier food availability may make it more plemental Nutrition Assistance Program that
healthier products than very high-income difficult for residents of non-White and lower- increase retailer stocking requirements may also
communities. For the expanded 13-item abso- income communities to make healthier food help to alter product mixes in small stores.42---44
lute availability measure, which included fruits choices and may contribute to racial/ethnic Improving relative availability of healthier
and vegetables, lower availability in Black than and socioeconomic disparities in obesity and alternatives in small stores may be particularly
White communities and very low-income than diet-related diseases.35---38 critical in low-income communities without
very high-income communities remained. Our findings suggest that racial/ethnic and supermarkets and grocery stores and where
socioeconomic inequities in the relative avail- substantial proportions of households lack au-
DISCUSSION ability of healthier alternatives compound pre- tomobiles. Residents of these communities may
viously documented differences in availability be more reliant on smaller local stores, espe-
Our results provide some of the first evi- of food stores across communities.13---16 That is, cially when food stocks run low.45---47 Improv-
dence of racial/ethnic and socioeconomic in- not only do non-White and low-income com- ing relative availability of healthier alternatives
equities in the relative and absolute availability munities often have fewer supermarkets and may also facilitate healthier choices among
of healthier food and beverage products in more convenience stores than White and children, who often frequent small stores on
communities across the United States. Lower higher-income communities, they also have their way to and from school.48,49

2174 | Research and Practice | Peer Reviewed | Zenk et al. American Journal of Public Health | November 2014, Vol 104, No. 11
 RESEARCH AND PRACTICE

TABLE 3—Associations Between Relative Availability of Healthier Food and Beverage Alternatives and Community Characteristics and Predicted
Relative Availability by Store Type: Bridging the Gap Community Obesity Measures Project, United States, 2010–2012

Supermarkets (n = 928) Grocery Stores (n = 813) Limited-Service Stores (n = 6721)

Predicted Predicted Predicted
Variable b (SE) Value F b (SE) Value F b (SE) Value F

Urbanicity
Suburban (Ref) 1.000 1.000 1.000
Urban –0.002 (0.005) –0.091*** (0.026) –0.003 (0.010)
Rural 0.004 (0.005) 0.028 (0.028) –0.010 (0.010)
Racial/ethnic majority
Non-Hispanic White (Ref) 1.000 1.000 1.000
Non-Hispanic Black –0.009 (0.018) < 0.001 (0.046) –0.041*** (0.015)
Hispanic –0.047*** (0.016) –0.073* (0.040) –0.045** (0.017)
Non-Hispanic other –0.012* (0.007) –0.015 (0.029) –0.023* (0.012)
Median household income
Very high (Ref) 1.000 1.000 1.000
Very low –0.019** (0.008) 0.006 (0.041) –0.087*** (0.014)
Low –0.011 (0.007) 0.052 (0.039) –0.063*** (0.013)
Middle –0.007 (0.006) –0.017 (0.042) –0.057*** (0.013)
High –0.003 (0.005) 0.024 (0.045) –0.039*** (0.012)
R2 0.070 0.078 0.052
Predicted relative availability of healthier
alternativesa
Overall 0.983 0.720 0.676
Very low-income Black communities 0.966 1.79 0.724 0.01 0.607 54.03***
Very low-income Hispanic communities 0.928 14.74*** 0.650 2.08 0.603 39.22***
Very high-income White communities 0.994 0.718 0.735

Note. Models controlled for region, time of day, day of week, seasonality indicator, and year.
a
Adjusted Wald tests of significance for pairwise comparisons, relative to very high-income White communities.
*P < .05; **P < .01; ***P < .001.

Although few studies have been able to alternatives at stores in rural than suburban work,13---16 our findings support policies such as
compare food availability by urbanicity, our areas. the Healthy Food Financing Initiative that
findings suggest that urban grocery stores Overall, our results provide additional evi- expand supermarket availability in under-
may have poorer food options than suburban dence that supermarkets are important re- served low-income and non-White communi-
grocery stores. These findings may reflect sources for healthy eating. Similar to other ties.58,59 As suggested by our findings of racial/
urban communities’ higher concentration of research showing that supermarkets have ethnic and socioeconomic differences in rela-
corner grocery stores, which often have less a wide selection of snacks and other energy- tive availability of healthier alternatives in
healthy product mixes.19,50---52 Furthermore, dense, nutrient-poor foods and beverages,55---57 supermarkets, however, it may be important to
we found few differences in relative or abso- our study showed that supermarkets also car- incentivize healthier in-store offerings in all
lute availability of healthier alternatives be- ried a large number of less healthy food stores, including supermarkets, through policy.
tween rural and suburban stores, after ad- products. However, 9 in 10 supermarkets Changing the balance between healthier and
justment for community income level and carried at least an equivalent number of less healthy options in supermarkets may re-
store type. Previous studies reported that rural healthier alternatives. On the one hand, this quire larger retailer stocking requirement in-
communities have fewer supermarkets and suggests that options are available for super- creases than proposed to date for government
grocery stores than do suburban or urban market customers to make healthier choices. food assistance programs.44
communities,15,53,54 so poorer food access On the other hand, the findings imply that
for rural residents may primarily arise from supermarket customers can easily be tempted Strengths and Limitations
fewer local stores and consequent longer by less healthy products and thus must remain Strengths of our study include its national
travel distances, rather than fewer healthier vigilant. Nonetheless, together with previous scope, directly observed measures of food and

November 2014, Vol 104, No. 11 | American Journal of Public Health Zenk et al. | Peer Reviewed | Research and Practice | 2175
 RESEARCH AND PRACTICE

on limiting the availability of less healthy items

TABLE 4—Associations Between Absolute Availability of Healthier Food and Beverage may not have a sufficient impact on the local
Alternatives and Community Characteristics: Bridging the Gap Community Obesity food retail environment. To realize positive
Measures Project, United States, 2010–2012 health benefits, policies and other interventions
Variable Availability of 7 Healthier Items,a b (SE) Availability of 13 Healthier Items,b b (SE) may need to address relative availability of
healthier alternatives, particularly in low-income
Urbanicity communities of color, so that healthier products
Suburban (Ref) prevail over less healthy products more fre-
Urban –0.086 (0.063) –0.001 (0.089) quently in retail stores. These efforts may need to
Rural 0.002 (0.074) 0.056 (0.106) be complemented with pricing policies that make
Racial/ethnic majority healthier products more accessible and less
Non-Hispanic White (Ref) healthy products less appealing.62 j
Non-Hispanic Black –0.285*** (0.073) –0.299** (0.132)
Hispanic –0.337*** (0.120) –0.221 (0.163)
Non-Hispanic other –0.280*** (0.077) –0.138 (0.100) About the Authors
Shannon N. Zenk is with the College of Nursing, Lisa M.
Median household income
Powell is with the School of Public Health, Leah Rimkus and
Very high (Ref) Zeynep Isgor are with the Institute for Health Research and
Very low –0.284*** (0.103) –0.305** (0.142) Policy, and Frank Chaloupka is with the Department of
Low –0.144 (0.099) –0.060 (0.141) Economics, University of Illinois at Chicago. Dianne C.
Barker is with Barker Bi-Coastal Health Consultants Inc,
Middle –0.126 (0.092) –0.053 (0.130) Calabasas, CA. Punam Ohri-Vachaspati is with the School
High –0.092 (0.089) 0.018 (0.127) of Nutrition and Health Promotion, Arizona State Univer-
Store type sity, Phoenix.
Correspondence should be sent to Shannon N. Zenk,
Supermarket University of Illinois at Chicago, College of Nursing, 845 S
Grocery –2.160*** (0.093) –3.353*** (0.135) Damen Ave. 9th floor, Chicago, IL 60612 (e-mail: szenk@
[Link]). Reprints can be ordered at [Link] by
Limited service –3.119*** (0.036) –7.081*** (0.052)
clicking the “Reprints” link.
R2 0.284 0.476 This article was accepted May 28, 2014.

Note. Models controlled for region, time of day, day of week, seasonality indicator, and year.
a
Low-fat milk, low-fat cheese, low-sugar cereal, 100% whole-wheat bread, baked or low-fat potato chips, 100% orange juice, Contributors
and diet soda. S. N. Zenk led the conceptualization of the analysis,
b
The 7 healthier alternatives plus availability of at least 1 variety of each of the following: fresh fruit, fresh vegetables, frozen data analysis, interpretation, and writing of the article.
fruit, frozen vegetables, canned fruit, and canned vegetables. L. M. Powell and D. C. Barker participated in the con-
*P < .05; **P < .01; ***P < .001. ceptualization and design of the overall study, led by
F. Chaloupka. S. N. Zenk, L. M. Powell, L. Rimkus,
D. C. Barker, and P. Ohri-Vachaspati designed the data
collection. L. M. Powell and L. Rimkus interpreted the
beverage availability, large samples of com- We focused on availability and did not examine results. Z. Isgor participated in the data analysis and
munities and stores, considerable variation in price differences, which might have further affected interpretation. All authors revised the article.

community characteristics, and inclusion of the accessibility of healthier alternatives. Despite

store types not typically included in previous these limitations, our study provides new nation- Acknowledgments
Data collection and analysis were supported by the
studies (e.g., drug stores, dollar stores). wide evidence on relative and absolute availability Robert Wood Johnson Foundation (grant 64702 and
Our findings are generalizable to communi- of healthier food and beverage products. 70157 to the Bridging the Gap program at the University
ties where public school students in grades 8, of Illinois at Chicago).

10, and 12 reside in the continental United Conclusions

States, but they may not apply to all US We showed that there is considerable room Human Participant Protection
This study was exempt from institutional review board
communities. The availability measures were for improvement in the relative availability of approval because it did not involve human participants.
not comprehensive, but instead were based on healthier alternatives across food stores. Indeed,
indicator foods and beverages60; their compo- 78% of stores carried a greater number of less References
nent foods and beverages may not be equally healthy foods and beverages than healthier al- 1. Kirkpatrick SI, Dodd KW, Reedy J, Krebs-Smith SM.
relevant across ethnic communities and ternatives. More research is needed to understand Income and race/ethnicity are associated with adherence
to food-based dietary guidance among US adults and
stores.61 The relative availability measure cap- contributions of not only absolute but also relative children. J Acad Nutr Diet. 2012;112(5):624---635.
tured only products with both a healthier and availability of healthier alternatives to improve 2. Thompson FE, McNeel TS, Dowling EC, Midthune
less healthy option and thus did not include dietary intakes and body weight outcomes. D, Morrissette M, Zeruto CA. Interrelationships of added
a comprehensive array of food and beverage Our results suggest that policies and other sugars intake, socioeconomic status, and race/ethnicity in
adults in the United States: National Health Interview
products recommended under current dietary interventions focused solely on increasing the Survey, 2005. J Am Diet Assoc. 2009;109(8):1376---
guidelines. availability of healthier foods and beverages or 1383.

2176 | Research and Practice | Peer Reviewed | Zenk et al. American Journal of Public Health | November 2014, Vol 104, No. 11
 RESEARCH AND PRACTICE

3. Horowitz CR, Colson KA, Hebert PL, Lancaster K. 20. Bridging the Gap. Bridging the Gap: research postmenopausal women in the US: the Women’s Health
Barriers to buying healthy foods for people with diabetes: informing policies and practices for healthy youth. Initiative 1993---2009. Diabetes Care. 2012;35(11):
evidence of environmental disparities. Am J Public Health. Available at: [Link] 2226---2234.
2004;94(9):1549---1554. Accessed February 5, 2014. 37. Shai I, Jiang R, Manson JE, et al. Ethnicity, obesity,
4. Baker EA, Schootman M, Barnidge E, Kelly C. The 21. Bachman JG, Johnston LD, O’Malley PM, and risk of type 2 diabetes in women: a 20-year follow-up
role of race and poverty in access to foods that enable Schulenberg JE, eds. The Monitoring the Future Project study. Diabetes Care. 2006;29(7):1585---1590.
individuals to adhere to dietary guidelines. Prev Chronic After Thirty-Seven Years: Design and Procedures. Ann 38. Jemal A, Siegel R, Ward E, et al. Cancer statistics,
Dis. 2006;3(3):A76. Arbor, MI: Institute for Social Research; 2011. Monitor- 2008. CA Cancer J Clin. 2008;58(2):71---96.
5. Morland K, Filomena S. Disparities in the availability ing the Future occasional paper 76.
39. Maantay J. Mapping environmental injustices: pit-
of fruits and vegetables between racially segregated 22. Barker D, Chaloupka F, Chriqui J, et al, eds. Bridging falls and potential of geographic information systems in
urban neighbourhoods. Public Health Nutr. 2007;10 the Gap Community Obesity Project: Methodology. Chicago, assessing environmental health and equity. Environ
(12):1481---1489. IL: Bridging the Gap; 2014. Bridging the Gap research Health Perspect. 2002;110(suppl 2):161---171.
6. Gustafson A, Hankins S, Jilcott S. Measures of the paper series, 1.
40. Mohai P, Pellow D, Roberts JT. Environmental
consumer food store environment: a systematic review of 23. US Dept of Labor Statistics, Occupational Safety and justice. Annu Rev Environ Resour. 2009;34:405---430.
the evidence 2000---2011. J Community Health. 2012; Health Administration. SIC search. Available at: https://
37(4):897---911. [Link]/pls/imis/[Link]. Accessed August 41. Morello-Frosch R, Lopez R. The riskscape and the
2, 2014. color line: examining the role of segregation in environ-
7. Zenk SN, Schulz AJ, Israel BA, James SA, Bao S, mental health disparities. Environ Res. 2006;102(2):
Wilson ML. Fruit and vegetable access differs by com- 24. Powell LM, Han E, Zenk SN, et al. Field validation 181---196.
munity racial composition and socioeconomic position in of secondary commercial data sources on the retail
Detroit, Michigan. Ethn Dis. 2006;16(1):275---280. food environment in the US. Health Place. 2011;17(5): 42. Andreyeva T, Luedicke J, Middleton AE, Long MW,
1122---1131. Schwartz MB, eds. Changes in Access to Healthy Foods
8. Sharkey JR, Horel S, Dean WR. Neighborhood After Implementation of the WIC Food Package Revisions.
deprivation, vehicle ownership, and potential spatial 25. Liese AD, Colabianchi N, Lamichhane AP, et al. Washington, DC: USDA Economic Research Service
access to a variety of fruits and vegetables in a large rural Validation of 3 food outlet databases: completeness Food Assistance and Nutrition Research Program; 2011.
area in Texas. Int J Health Geogr. 2010;9(1):26. and geospatial accuracy in rural and urban food Contractor and Cooperator Report 66.
9. Hutchinson PL, Nicholas Bodor J, Swalm CM, Rice environments. Am J Epidemiol. 2010;172(11):
1324---1333. 43. Hillier A, McLaughlin J, Cannuscio CC, Chilton M,
JC, Rose D. Neighbourhood food environments and Krasny S, Karpyn A. The impact of WIC food package
obesity in southeast Louisiana. Health Place. 2012; 26. Iannacchione VG, Staab JM, Redden DT. Evaluating changes on access to healthful food in 2 low-income
18(4):854---860. the use of residential mailing addresses in a metropol- urban neighborhoods. J Nutr Educ Behav. 2012;44(3):
10. Raja S, Yin L, Roemmich J, et al. Food environment, itan household survey. Public Opin Q. 2003;67(2): 210---216.
built environment, and women’s BMI: evidence from 202---210.
44. Zenk SN, Odoms-Young A, Powell LM, et al. Fruit
Erie County, New York. J Plann Educ Res. 2010;29(4): 27. Rimkus L, Powell LM, Zenk SN, et al. Development and vegetable availability and selection: federal food
444---460. and reliability testing of a food store observation form. package revisions, 2009. Am J Prev Med. 2012;43(4):
11. Truong K, Fernandes M, An R, Shier V, Sturm R. J Nutr Educ Behav. 2013;45(6):540---548. 423---428.
Measuring the physical food environment and its re- 28. US Dept of Agriculture. Dietary Guidelines for 45. Dubowitz T, Smith-Warner SA, Acevedo-Garcia D,
lationship with obesity: evidence from California. Public Americans, 2010. 7th ed. Washington DC: US Dept of Subramanian SV, Peterson KE. Nativity and duration
Health. 2010;124(2):115---118. Health and Human Services; 2010. of time in the United States: differences in fruit and
12. Searching for Healthy Food: The Food Landscape in 29. Food Marketing Institute. Supermarket facts: in- vegetable intake among low-income postpartum
California Cities and Counties. Detailed Methodology. Davis, dustry overview 2013. Available at: [Link] women. Am J Public Health. 2007;97(10):1787---
CA: California Center for Public Health Advocacy; 2007. org/research-resources/supermarket-facts. Accessed 1790.
13. Moore LV, Diez Roux AV. Associations of neigh- May 23, 2014. 46. Clifton KJ. Mobility strategies and food shopping for
borhood characteristics with the location and type of food 30. Connell CL, Yadrick MK, Simpson P, Gossett J, low-income families: a case study. J Plann Educ Res.
stores. Am J Public Health. 2006;96(2):325---331. McGee BB, Bogle ML. Food supply adequacy in the lower 2004;23(4):402---413.
14. Morland K, Wing S, Diez Roux A, Poole C. Mississippi delta. J Nutr Educ Behav. 2007;39(2):77---83. 47. Zenk SN, Odoms-Young AM, Dallas C, et al. “You
Neighborhood characteristics associated with the location 31. Farley TA, Rice J, Bodor JN, Cohen DA, Bluthenthal have to hunt for the fruits, the vegetables”: environmental
of food stores and food service places. Am J Prev Med. RN, Rose D. Measuring the food environment: shelf space barriers and adaptive strategies to acquire food in a low-
2002;22(1):23---29. of fruits, vegetables, and snack foods in stores. J Urban income African American community. Health Educ Behav.
15. Powell LM, Slater S, Mirtcheva D, Bao Y, Chaloupka FJ. Health. 2009;86(5):672---682. 2011;38(3):282---292.
Food store availability and neighborhood characteristics in 32. Galvez MP, Morland K, Raines C, et al. Race and 48. Borradaile KE, Sherman S, Vander Veur SS, et al.
the United States. Prev Med. 2007;44(3):189-- 195. food store availability in an inner-city neighbourhood. Snacking in children: the role of urban corner stores.
16. Zenk SN, Schulz AJ, Israel BA, James SA, Bao S, Public Health Nutr. 2008;11(06):624---631. Pediatrics. 2009;124(5):1293---1298.
Wilson ML. Neighborhood racial composition, neigh- 33. American Community Survey Office. The 2006--- 49. Dennisuk LA, Coutinho AJ, Suratkar S, et al. Food
borhood poverty, and the spatial accessibility of super- 2010 ACS 5-year summary file technical documentation. expenditures and food purchasing among low-income,
markets in metropolitan Detroit. Am J Public Health. Available at: [Link] urban, African-American youth. Am J Prev Med. 2011;
2005;95(4):660---667. summaryfile/ACS_2006-2010_SF_Tech_Doc.pdf 40(6):625---628.
17. Glanz K, Sallis JF, Saelens BE, Frank LD. Nutrition Accessed February 5, 2014. 50. Laska MN, Borradaile KE, Tester J, Foster GD,
environment measures survey in stores (NEMS-S): develop- 34. National Center for Education Statistics. NCES Gittelsohn J. Healthy food availability in small urban food
ment and evaluation. Am J Prev Med. 2007;32(4):282---289. locale codes. Available at: [Link] stores: a comparison of four US cities. Public Health Nutr.
18. Liese AD, Weis KE, Pluto D, Smith E, Lawson A. handbook/data/pdf/appendix_d.pdf. Accessed February 2010;13(7):1031---1035.
Food store types, availability, and cost of foods in 5, 2014. 51. Lucan SC, Karpyn A, Sherman S. Storing empty
a rural environment. J Am Diet Assoc. 2007;107(11): 35. Flegal KM, Carroll MD, Kit BK, Ogden CL. Preva- calories and chronic disease risk: snack-food products,
1916---1923. lence of obesity and trends in the distribution of body nutritive content, and manufacturers in Philadelphia
19. Franco M, Diez Roux AV, Glass TA, Caballero B, mass index among US adults, 1999---2010. JAMA. corner stores. J Urban Health. 2010;87(3):394---409.
Brancati FL. Neighborhood characteristics and availabil- 2012;307(5):491---497. 52. Cavanaugh E, Mallya G, Brensinger C, Tierney A,
ity of healthy foods in Baltimore. Am J Prev Med. 36. Ma Y, Hebert JR, Manson JE, et al. Determinants Glanz K. Nutrition environments in corner stores in
2008;35(6):561---567. of racial/ethnic disparities in incidence of diabetes in Philadelphia. Prev Med. 2013;56(2):149---151.

November 2014, Vol 104, No. 11 | American Journal of Public Health Zenk et al. | Peer Reviewed | Research and Practice | 2177
 RESEARCH AND PRACTICE

53. Sharkey JR, Horel S. Neighborhood socioeconomic

deprivation and minority composition are associated with
better potential spatial access to the ground-truthed food
environment in a large rural area. J Nutr. 2008;138
(3):620---627.
54. Jilcott SB, Liu H, Moore JB, Bethel JW, Wilson J,
Ammerman AS. Commute times, food retail gaps, and
body mass index in North Carolina counties. Prev Chronic
Dis. 2010;7(5):A107.
55. Cameron AJ, Thornton LE, McNaughton SA,
Crawford D. Variation in supermarket exposure to
energy-dense snack foods by socio-economic position.
Public Health Nutr. 2013;16(7):1178---1185.
56. Thornton LE, Cameron AJ, McNaughton SA,
Worsley A, Crawford DA. The availability of snack food
displays that may trigger impulse purchases in Melbourne
supermarkets. BMC Public Health. 2012;12:194.
57. Farley TA, Baker ET, Futrell L, Rice JC. The
ubiquity of energy-dense snack foods: a national multicity
study. Am J Public Health. 2010;100(2):306---311.
58. Giang T, Karpyn A, Laurison HB, Hillier A, Perry
RD. Closing the grocery gap in underserved communi-
ties: the creation of the Pennsylvania Fresh Food Fi-
nancing Initiative. J Public Health Manag Pract. 2008;
14(3):272---279.
59. US Dept of Health and Human Services. Healthy
Food Financing Initiative. Available at: [Link]
[Link]/programs/ocs/resource/healthy-food-financing-
initiative-0. Accessed May 31, 2013.
60. McKinnon RA, Reedy J, Handy SL, Rodgers AB.
Measuring the food and physical activity environments:
shaping the research agenda. Am J Prev Med. 2009;
36(4 suppl):S81---S85.
61. Grigsby-Toussaint DS, Zenk SN, Odoms-Young A,
Ruggerio L, Moise I. Availability of commonly consumed
and culturally specific fruits and vegetables in African-
American and Latino neighborhoods. J Am Diet Assoc.
2010;110(5):746---752.
62. Powell LM, Chriqui JF, Khan T, Wada R, Chaloupka
FJ. Assessing the potential effectiveness of food and
beverage taxes and subsidies for improving public health:
a systematic review of prices, demand and body weight
outcomes. Obes Rev. 2013;14(2):110---128.

2178 | Research and Practice | Peer Reviewed | Zenk et al. American Journal of Public Health | November 2014, Vol 104, No. 11