Neighborhood Environments

A growing body of evidence indicates that residential segregation by income, race, and ethnicity contributes to health disparities in the U.S. ^{,  ,  ,  ,  ,  ,}  In addition, substantial research shows disparities in health between urban and rural areas. Poor dietary patterns and obesity, established risk factors for chronic disease, have been linked to neighborhood deprivation, neighborhood minority composition, and low area population density (as found in more rural areas). ^{,  ,  ,  ,  ,  ,}  Neighborhood differences in access to foods may be an important influence on these relationships. Establishing the presence, nature, and implications of neighborhood differences in the physical availability of more- and less-healthy foods is necessary to properly inform the development of responsive public health policies and interventions that may help reduce inequalities in health.

Recent reviews have drawn attention to relationships among neighborhood food availability, dietary intake, and obesity, but to our knowledge, a detailed review of disparities in food access has not been completed. ^,  For this review, neighborhood is broadly defined to be “the area around one's place of residence,” and research relating to both micro- (e.g., stores within walking distance from home) and macro-level characteristics (e.g., restaurants within county boundaries) of the physical food environment are considered. This article briefly reviews the current evidence base regarding (1) the relationship between neighborhood access to more- and less-healthy foods and dietary intake and (2) the relationship between neighborhood access to foods and weight status. In addition, this article presents a comprehensive review of disparities across the U.S. according to income, race, ethnicity, and urbanization in neighborhood access to more- and less-healthy foods. Studies relating to both food stores and restaurants are summarized separately and discussed in terms of their limitations and implications for future research and practice.

A snowball strategy was used to identify relevant research studies completed in the U.S. and published between 1985 and April 2008. Searches were completed in PubMed and MEDLINE using the following key words: neighborhood; environment; food store; supermarket; restaurant; dietary intake; obesity; overweight; disparity; inequality; deprivation; income; poverty; rural; race; and ethnicity. The references cited in articles (n=43) indexed in these search engines were also checked; and all potentially relevant articles were retrieved. To be included in the review; articles had to describe research that addressed neighborhood access to food stores; restaurants; or the types of food and beverage products available in such outlets. This search strategy identified 54 articles relevant to the review objectives.

The aim of this review is to describe and evaluate research relating to neighborhood differences in the physical availability of food stores, restaurants, and healthy foods. Associations of neighborhood food access with dietary intake and obesity were found to vary according to the type of food store or restaurant. In general, research suggests that neighborhood residents who have better access to supermarkets and limited access to convenience stores tend to have healthier diets and lower levels of obesity. ^{,  ,  ,  ,  ,  ,  ,  ,  ,}  Studies that examined the accessibility of restaurants produced less consistent findings; however, some evidence suggests that neighborhood residents who have better access to full-service restaurants and greater cost barriers to fast-food consumption have healthier diets and lower levels of obesity. ^{,  ,  ,  ,}  Much of the evidence suggesting a relationship between restaurant accessibility and obesity levels in the U.S. has come from research that focused on land area units larger than neighborhoods (e.g., states, counties). National and local studies across the U.S. have shown disparities according to income, race, ethnicity, and urbanization in neighborhood access to food stores and restaurants. Although additional studies are needed to address various research limitations, completed studies indicate a need for policy action and other intervention strategies to ensure equitable access to healthy foods across the U.S. Recommendations for future research on disparities in access to healthy foods are summarized in Appendix C (online at www.ajpm-online.net).

Common limitations of the studies reviewed here relate to: the validity and reliability of measures; the complexity of defining a relevant neighborhood; and the cross-sectional, observational nature of most research designs. The majority of studies included in this review used commercially (e.g., Dun and Bradstreet, InfoUSA) or publicly available lists (e.g., telephone directories, state departments of agriculture) to identify food stores and restaurants. Few studies identified stores or restaurants by ground truthing or with walking surveys, and only one study was found that reported a detailed comparison of these two methods for measuring residential food environments. That single study reported a considerable discrepancy between the stores found on publicly available lists and the results of ground truthing in rural Texas; nearly 20% of stores on publicly available lists could not be verified, and 36% of stores in the area were missing from those lists. Using data from only publicly available lists would have led to overestimations of the distance to the nearest supermarket in nearly 34% of the census block groups and to the nearest convenience store in 13% of the census block groups. Additional studies are needed to determine the potential impact of errors in commercially and publicly available lists on research findings. Research conducted to validate a commercial database of physical activity facilities found only moderate to poor agreement when the database locations were compared to locations determined by a field census; the authors concluded that the patterns of error observed were likely to bias environment–health associations toward the null. It may be advisable to use a combination of measures and data sources, when feasible, to reduce errors in the identification of food stores and restaurants.

The common use of automatic geocoding tools (i.e., ArcGIS, Automatch, ZP4, Geolytics, and other software that perform batch or interactive address matching) is of further concern. The geocoding process is vulnerable to various types of error that may be introduced during the preprocessing and matching of addresses to spatial areas. A few studies have reported address match rates, ^,  but very little is known about matching accuracy or the potential impact of positional errors on research findings relating to the food environment. Studies comparing automatic geocoding tools have found considerable variation in the assignment of addresses to census block groups and census tracts. For example, one study found that 28%–36% of addresses geocoded by different tools were not assigned to the same census block group, and 33%–51% of residential addresses were not assigned to the same census tract. Other studies have further examined the positional accuracy of automated geocoding tools by comparing the assigned latitude and longitude coordinates to assignments determined using the gold-standard method of aerial orthoimagery and found sizable error, especially in rural areas. ^, 

Discrepancies between researcher- and resident-defined neighborhood boundaries may be a source of bias in studies relating to food access. ^,  Research regarding perceptions of neighborhood boundaries has shown that resident-defined neighborhoods are, on average, four times larger than a census block group. The average size of a resident-defined neighborhood is comparable to a census tract, but neighborhoods typically include portions of at least two census tracts. Clearly, researchers cannot assume that resident- and census-defined neighborhoods are similar. Discrepancies could result in measurable differences in the social conditions (e.g., ethnic/racial composition, average income) occurring within a neighborhood.

A limited number of studies have considered the distance that residents typically travel beyond their neighborhood to purchase food. ^{,  ,  ,  ,}  Among these studies, the definition of a relevant shopping area has ranged from 0.5 to 15 miles. The availability of food along routes to work, school, and other destinations may further influence dietary intake and obesity. Different methods of geocoding and varied buffer sizes could result in substantially different datasets on which study results and recommendations are based. Future research needs to illuminate which food environment areas are most influential and what factors, other than physical proximity, influence people's choices of where to purchase food. Further development of standard measurement protocols and greater reporting on details of the geocoding process are needed in future research studies in order to provide a basis for comparing study outcomes (Appendix C, online at www.ajpm-online.net).

Comparability across studies also depends on the use of standard definitions to classify food stores and restaurants into categories. As summarized in Appendix A, Appendix B (online at www.ajpm-online.net), the studies described in this review have defined categories of food stores and restaurants in various ways. For example, some studies have defined categories of food stores according to the number of cash registers whereas others have categorized stores according to the number of staff employed, the types of food sold, name recognition, or annual sales data. ^{,  ,  ,  ,  ,}  Several studies have similarly used the North American Industry Classification System (NAICS) or predecessor Standard Industrial Classification (SIC) system codes to define categories of stores and restaurants; however, few studies have reported the code numbers included within each category. ^,  Further, some research indicates that the business databases from which NAICS and SIC codes are typically drawn may contain many errors. Although different data sources do not always provide the same details relevant to the categorization of food outlets, researchers should work toward a consensus on the use of a standard set of definitions, and future studies should aim to fully describe how food store and restaurant categories are defined.

Research studies that complete observations within stores and restaurants can most directly address differences in the physical availability of healthy foods and beverages. It is critical that studies employing direct observation report on methodologic details and the reliability of employed measures. Just half of the studies in this review that reported on direct observations in stores (n=5 of 11) and neither study that reported on direct observations in restaurants (n=2) were found to include assessments of inter-rater or test–retest reliability. ^{,  ,  ,  ,}  However, growing efforts are being made to design psychometrically sound measures of food environments within stores and restaurants and to develop strategies to promote high measurement reliability. The implementation of these tools and strategies in future research will contribute to a better understanding of differences in residential access to healthy food and their implications for health.

The majority of studies included in this review used cross-sectional designs; few longitudinal, multilevel, or intervention studies have been reported. Studies examining relationships among residential food access, dietary intake, and obesity have considered several characteristics of neighborhoods and individuals, including gender, race and ethnicity, income, education, and physical activity. However, several factors contribute to dietary intake and the development of obesity. In order to better understand the relative importance of environmental, demographic, psychological, and social factors, as well as the interaction of these factors, it will be necessary to examine hypothesized pathways of influence and the contributions of each factor within the same study.

Future studies investigating disparities in access to more- and less-healthy food also need to systematically consider covarying characteristics of neighborhoods such as population size, urbanization, region, and commercialization. Several studies included in this review indicate that disparities in access to healthy food vary according to these characteristics of neighborhoods. ^{,  ,}  For example, one study found no significant differences in the relative availability of fast-food restaurants by income, race, or ethnicity within a national sample of ZIP codes. However, when only urban ZIP codes were examined, ZIP codes in lower- versus higher-income and predominantly black areas versus predominantly white areas were found to have a higher proportion of fast-food restaurants among all available restaurants. International comparisons using standardized measures may further help to illuminate qualities of the broad social and political environment that contribute to disparities. In stark contrast to research conducted in the U.S., studies in areas of New Zealand, Australia, Canada, and Scotland show greater availability of supermarkets in lower- versus higher-income neighborhoods. ^{,  ,  ,} 

Additionally, more comprehensive assessments of neighborhood food environments are needed. When only one aspect of the food environment is studied, the results are difficult to interpret, as neighborhoods with greater access to unhealthy food options may also have greater access to healthy food options. A limited number of studies have investigated disparities in access to both food stores and restaurants within the same geographic region and suggest that the nature of neighborhood disparities may differ according to the type of food outlet examined. ^{,  ,  ,}  Further, the extent to which other factors, such as economic and social characteristics of neighborhoods, may influence accessibility above and beyond physical proximity is an important area for future research.

Despite these research limitations, the studies reviewed here indicate a need for policy action and other intervention strategies to ensure more equitable access to healthy foods across the U.S. Very few studies have evaluated strategies for reducing disparities or improving physical access to healthy, affordable food. Several strategies and policy actions have been proposed to attract supermarkets to underserved neighborhoods, improve the availability of healthy foods such as fruits and vegetables, and reduce access to calorie-dense foods in restaurants (Appendix D, online at www.ajpm-online.net). ^{,  ,  ,  ,}  However, many challenges remain in implementing interventions of this nature. Research designed to evaluate proposed interventions, build broad support for their implementation, and identify other effective means for improving neighborhood access to healthy food should be made a priority.

This paper was supported in part by the Robert Wood Johnson Foundation Healthy Eating Research Program.

No financial disclosures were reported by the authors of this paper.