In contrast with our hypothesis, we found that neighborhood fast food availability was not related to fast food consumption in our large, national sample of young adults residing in neighborhoods throughout the U.S. Our findings suggest that targeting neighborhood fast food availability may not reduce consumption or obesity among young U.S. adults.
Our findings, of no relationship between fast food availability on fast food consumption among adults, are consistent with prior research [11
]. One study reports a positive association between fast food availability and fast food consumption, but only among a subset of 404 adults living in Montreal who were "reward sensitive" [10
Null results may reflect: (1) that young adults more often purchase and consume fast food in settings other than their residential neighborhoods, such as school or workplace locations, (2) that lifestyle factors such as family structure or employment status are stronger determinants of fast food consumption [35
] or (3) the possibility that unmeasured neighborhood and social preferences, such as location selection factors, more strongly influence dietary behaviors. That is, the social, economic, cultural factors that affect where a person is able or wishes to live may also influence dietary behaviors [38
]. Much would be gained by broadening research to focus on environmental contexts beyond the residential location, such as college and workplace neighborhoods as well as commuting routes. In addition, future research should incorporate other neighborhood- and individual-level factors to determine which settings and individual and neighborhood characteristics are most salient for dietary behaviors.
The majority of published research has focused on the indirect relationship between fast food availability and obesity, rather than investigating direct effects on fast food consumption. Greater neighborhood fast food availability is consistently related to higher obesity [1
] yet it is possible that this relationship reflects processes other than higher fast food consumption. In particular, fast food restaurants may cluster with other environmental characteristics that influence obesity [20
]. For example, automobile access may be important for fast food restaurants because of their dependence on drive-thru business. Therefore, neighborhoods with more fast food restaurants may promote obesity as a result of dominating road structures that hinder active transportation. Our findings that neighborhood fast food availability is unrelated to fast food consumption suggest that associations between fast food availability and obesity may reflect this or similar processes.
Our study expands current literature by comparing fast food restaurants scaled by roadway kilometers within a street network buffer - a measure of availability that improves upon widely used restaurant count measures by accounting for urban development and street access - across multiple geographic and sociodemographic characteristics. We present data from a large sample of U.S. young adults that uses a more refined urban/rural classification than the traditional urban/rural dichotomy. Furthermore, we observed vast differences in the availability of fast food for non-urban compared to urban respondents. While these differences precluded the use of comparable availability measures across urbanicity levels, they suggest that the built environment may operate quite differently in rural, suburban, and urban areas, and thus may necessitate different measurement approaches. This might also underlie some of the mixed findings in the literature [10
Strengths and limitations
We present data from a unique and large national cohort that includes a variety of detailed environmental data combined with individual-level diet data in a cohort of young adults from around the U.S. To our knowledge ours is the only large analysis of fast food availability and consumption that accounts for multiple environmental features and individual characteristics simultaneously.
Yet, our study has some limitations. Our measure of fast food consumption is based on self-report and, like all self-report diet measures, has inherent recall and reporting error. Respondents were not instructed how to report meals versus snacks; that is, snacking at a fast food restaurant may be undercounted if the respondent did not consider it as a visit. Further, because the question asks about the number of days that respondents ate fast food in the last week, fast food consumption may be under-represented if the respondent ate more than one fast food meal in a day. Yet, this is a measure that is commonly used to assess fast food consumption in large population-based studies such as the Panel Study of Income Dynamics and Coronary Artery Risk Development in Young Adults [39
]. Moreover, we are unable to ascertain what foods were available and consumed at each fast food visit. This is increasingly an issue as fast food restaurants are including healthier options in response to consumer health concerns [43
]. In addition, our study is cross-sectional and thus does not capture changes in the food environment or consumption over time. We were also unable to control for factors related to selection of residential neighborhoods, however by including Wave I school indicator variables, we attempted to address unmeasured characteristics associated with baseline neighborhood. Our 3 km network neighborhood buffer may not accurately reflect food-purchasing areas for different urban settings and sociodemographic subgroups, however estimates for the 1 km network buffer were very similar. Significant differences in characteristics of respondents included versus excluded in our sample may have biased our results.
There may be error in our roadway (StreetMap Pro) and food resource data that we are unable to investigate in our national sample. In addition, the relatively narrow range of fast food availability may limit our ability to detect effects of fast food availability in relation to fast food consumption. We were not able apply spatial interaction models to our national sample, but our models account for spatial clustering of respondents. The trade-off is our large, national sample and the ability to compare individuals living in non-urban, low density urban, and high density urban environments, within the context of the same cohort. There are no other datasets in which such a study is possible at the small, geographic unit used in our study.
Despite these limitations, our study is an essential step in understanding the allocation and consumption of fast food restaurants across geographic space over the entire U.S. and within urbanicity levels, and our findings can inform measurement and design in future individual-level and longitudinal studies.
Our findings are significant in light of the recent efforts to reduce obesity though policies targeting the fast food environment. For example, a one-year ban on fast food restaurants was unanimously put forth in South Los Angeles (LA) in an effort to reduce obesity in this low SES area, despite lower fast food restaurant per capita relative to more affluent West LA [44
]. Given evidence that eating fast food increases BMI and obesity risk [39
], reducing fast food consumption may be a valuable aim. However, specific environmental factors that influence fast food consumption among young adults are not well understood. Our findings suggest that greater residential neighborhood fast food availability may not be an important driver of fast food consumption. Greater understanding of how lifestyle factors and neighborhood food resources interact to influence fast food consumption is needed to inform effective policy.