Summarizing and Visualizing Neighborhood Vulnerability to Environmental Exposures Using ToxPi

Abstract

BACKGROUND AND AIM: Toxicological environmental exposures can cluster by neighborhood and contribute to health outcomes. A greater understanding of which neighborhood characteristics increase vulnerability to environmental exposures can guide resource allocation across neighborhoods. The Toxicological Prioritization Index (ToxPi) is an analytic framework that integrates data across different domains to create a summarized index measure. Originally developed for chemical exposures, ToxPi has been applied to social and physical determinants of health in past studies to characterize their relative contributions to neighborhood vulnerability to various health outcomes. We aimed to characterize neighborhood vulnerability to environmental exposures by applying this approach to New York City (NYC) data. METHODS: The index score is constructed from four equally-weighted domains using census-tract level data from the U.S. Census American Community Survey and the CDC 500 Cities Project. The domains included the following features we conceptualized as contributors to vulnerability to environmental exposures: 1) Demographics: age, sex, race/ethnicity, residential segregation, immigrant composition, disability status, occupation, mobility; 2) Economic characteristics: income, poverty, unemployment, vehicle availability; 3) Residential density: population density and housing characteristics (vacancy, age, type, density/overcrowding); and 4) Health behaviors, status, and access: health behaviors, health outcomes, preventive health and screening visits, health insurance coverage. Additionally, the relative contributions of each domain are accessed within census tracts in visual profiles. RESULTS:In total, 59 features are used to construct the vulnerability index. Generated profiles of vulnerability detail contributions of different data domains for the index score within each census tract, providing greater information than typical indices which provide only a single summarized metric of vulnerability. CONCLUSIONS:Vulnerability index scores at the neighborhood level can be driven by different domains, as demonstrated by different observed drivers of vulnerability across NYC census tracts. Methods such as ToxPi integrate and visualize the contributions of demographic, economic, residential density, and health domains to neighborhood vulnerability. KEYWORDS: neighborhood vulnerability index, data integration, data visualization, social determinants of health