Food store availability and neighborhood characteristics in the United States

Abstract

Objective.

This study provides a multivariate analysis of the availability of food store outlets in the US and associations with neighborhood characteristics on race, ethnicity and socioeconomic status (SES).

Method.

Commercial food store outlet data are linked across 28,050 zip codes to Census 2000 data. Multivariate regression analyses are used to examine associations between the availability of chain supermarkets, non-chain supermarkets, grocery stores and convenience stores and neighborhood characteristics on race, ethnicity and SES including additional controls for population size, urbanization and region.

Results.

Low-income neighborhoods have fewer chain supermarkets with only 75% (p < 0.01) of that available in middle-income neighborhoods. Even after controlling for income and other covariates, the availability of chain supermarkets in African American neighborhoods is only 52% (p < 0.01) of that in White neighborhoods with even less relative availability in urban areas. Hispanic neighborhoods have only 32% (p < 0.01) as many chain supermarkets compared to non-Hispanic neighborhoods. Non-chain supermarkets and grocery stores are more prevalent in low-income and minority neighborhoods.

Conclusion.

The study results highlight the importance of various potential public policy measures for improving access to supermarkets that may serve to reduce systematic local area barriers that are shown to exist by race, ethnicity and income.

Introduction

External environmental, social and economic factors are increasingly recognized as playing an important role in influencing people's lifestyles and risks for developing obesity (Egger and Swinburn, 1997, Hill and Peters, 1998, Swinburn et al., 1999, Allison et al., 2001, French et al., 2001). Examining factors that characterize individuals' local environments can help to provide evidence on the extent to which neighborhood factors are related to behavioral choices and obesity. One such factor relates to the availability of local area food stores.

Larger sized food stores such as supermarkets versus smaller stores and chain versus non-chain supermarkets have been shown to be more likely to stock healthful foods (Sallis et al., 1986, Horowitz et al., 2004) and to offer foods at a lower cost (BLS, 1966, Morris et al., 1990, Kaufman et al., 1997, Mantovani et al., 1997, Chung and Myers, 1999). Food costs are found to be associated with diet quality (French et al., 2001, Drewnowski and Specter, 2004), and studies reveal significant correlations between diet quality and the availability of healthful foods in stores (Cheadle et al., 1991, Fisher and Strogatz, 1999).

Potential barriers to obtaining a variety of healthful foods due to a lack of local area food stores such as supermarkets are likely to adversely affect dietary patterns and contribute to the risk of obesity. The availability of supermarkets has been associated with more fruit and vegetable intake, more healthful diets, and lower rates of obesity (Morland et al., 2002a, Morland et al., 2006, Laraia et al., 2004). Shopping at supermarkets versus independent grocers has been associated with more frequent fruit and vegetable consumption (Zenk et al., 2005a). However, study results based on interventions in the UK aimed at improving local grocery store access have shown mixed results for associated improvements in diet quality (Wrigley et al., 2003, Cummins et al., 2005).

The prevalence of obesity is shown to be significantly higher among Black and Hispanic populations compared to their White counterparts though these relationships differ by gender (Ogden et al., 2006). Evidence also shows higher obesity rates among low- versus high-income and education groups, with associations differing by gender and race (USDHHS, 2001, Paeratakul et al., 2002, Chang and Lauderdale, 2005). Differences in neighborhood socioeconomic (SES) indicators have been related to health outcomes controlling for individual-level social class indicators (Diez Roux et al., 1997, Robert, 1999). Differential rates of local area food store type availability by neighborhood characteristics may contribute to the fact that the problem of obesity does not affect all populations equally (Diez Roux et al., 1999, Cummins and Macintyre, 2006).

The extent to which food store availability differs by local area SES, racial and ethnic characteristics has been examined in several locations across the United States. Low- versus high-poverty, predominantly White versus Black and predominantly Latino versus non-Latino zip codes in LA county were found to have more supermarkets per household (Shaffer, 2002). In Chicago, poor versus non-poor neighborhoods were found to have significantly fewer supermarkets but more small grocery stores (Alwitt and Donley, 1997). Study results based on multi-state samples have found that low- versus high-income neighborhoods and predominantly Black versus White neighborhoods have fewer numbers of available supermarkets but significantly more grocery and convenience stores (Morland et al., 2002b, Moore and Diez Roux, 2006). National studies of metropolitan (Cotterill and Franklin, 1995) and urban (Morris et al., 1990) areas have found that low- versus high-SES neighborhoods have fewer available supermarkets.

Chain versus non-chain grocery stores have been found more likely to be located in non-poor zip code areas (Chung and Myers, 1999). Recent study results showed that, among the poorest tertile of neighborhoods, distance to the nearest chain supermarket increased with a higher proportion of African Americans, but remained similar across race in the least impoverished neighborhoods (Zenk et al., 2005b).

Due to the difficulty of gathering environmental data on a large scale, most studies that have examined food store availability and associations with neighborhood SES, racial and ethnic characteristics have been limited in their geographic coverage and all but one study (Zenk et al., 2005b) perform univariate analyses.

This study provides the first comprehensive multivariate national study of the availability of food stores by zip code across the United States and associations with neighborhood characteristics on race, ethnicity, SES, population size, urbanization and region. Commercial food store outlet data are linked by zip code to Census Bureau population and SES data. This study covers a population of 280,675,874 people living in 28,050 zip codes in the year 2000. For our full sample of zip codes and a sub-sample of zip codes in urban areas, we examine the availability of four types of food stores that include: (1) chain supermarkets, (2) non-chain supermarkets, (3) grocery stores and (3) convenience stores. This study provides evidence on the extent to which different types of food stores are differentially available in low-income communities and in those neighborhoods with higher proportions of minority populations simultaneously accounting for both factors.