The Neighborhood Food Environment and Change in Body Mass Index: A Systematic Review and Meta-Analysis of Longitudinal Studies.

Introduction: The neighborhood food environment - locations and density of different types of stores and restaurants - is of great interest to reduce obesity. However, prior meta-analyses have focused on cross-sectional studies, for which reverse causation is a major limitation. Meta-analyses restricted to longitudinal study designs evaluating the food environment and change in BMI have yet to be conducted. Hypothesis: We hypothesized that increased exposure to fast food restaurants would be associated with increased change in BMI. Methods: Eleven databases were searched for longitudinal studies evaluating convenience stores, fast food restaurants, grocery stores, or supermarkets and change in BMI. Inclusion criteria consisted of: (1) interventional, quasi-experimental, or cohort studies; (2) adult or children populations; (3) any country; (4) geographic density or distance of food outlets; (5) a multivariable-adjusted change in BMI over time and corresponding uncertainty. Data were extracted independently and in duplicate, and studies pooled using fixed and random-effects models. Heterogeneity was quantified using Cochrane's Q, and publication bias using funnel plots, Egger's test, and Begg's test. Results: 13 studies including 98,268 subjects were identified. In random effects models, no significant association was seen between food environments and change in BMI (Figure). A nonsignificant trend was seen toward lower BMI with increasing density of supermarkets; however, a similar nonsignificant trend was also observed between increasing density of fast food restaurants and lower BMI. Evidence for publication bias was not identified. Conclusion: In available longitudinal studies, no significant relationships were identified between food environments and change in BMI. Our novel findings support the need for further longitudinal and especially interventional studies of how the built food environment, including new measures related to cost and accessibility, may influence health and weight. [ABSTRACT FROM AUTHOR]