Modeling spatiotemporal associations of obesity prevalence with biking, housing cost and green spaces in Chicago, IL, USA, 2015–2017

Abstract

IntroductionObesity is an epidemic that threatens global wellbeing and contributes to an array of illnesses. The prevalence of obesity in the United States has increased substantially in past decades. Focusing on the city of Chicago, IL, USA as a case study, we investigate spatiotemporally varying relationships between obesity and three characteristics of urban areas: green space, biking, and housing cost. We address socio-environmental interventions to control obesity using spatiotemporal data on obesity prevalence. MethodsWe illustrate the spatiotemporal distribution of obesity using the space-time scan statistic to supplement choropleth maps of obesity prevalence in Chicago. In addition, we employ geographically and temporally weighted regression (GTWR), an extension of the geographically weighted regression (GWR) model, to simultaneously address both, the spatial and temporal non-stationarity of obesity prevalence. Lastly, we compare GTWR to GWR and the traditional ordinary least squares (OLS) model to assess the effect of spatial and temporal dependency in the data. ResultsOur findings show that the prevalence of obesity varies over time and space in Chicago, with clusters in the city's south and west. Prediction error of our regression models (AIC) decreases with model complexity (OLS: 15,931; GWR: 14,351; GTWR: 14,266). Obesity is negatively associated with green space in the city center but positively associated in the outskirts of Chicago. In addition, obesity is related to poor biking infrastructure and high rent throughout the city. ConclusionsThe findings may encourage planners and policymakers to improve active urban transportation, access to green spaces, housing affordability, and geographic surveillance of obesity as a means of identifying the most vulnerable communities.