Geographic Information Systems (GIS) are an innovative and powerful tool that helps analysts and decision makers organize, visualize, analyze, present, and understand complex layers of data. The key to spatial analysis is that most data contain a geographic component that can be tied to a specific location, such as a state, county, zip code, Census block, or single address. The geographic analysis enables users to explore and overlay data by location, revealing hidden trends that are not readily apparent in traditional spreadsheet and statistical packages. Additionally, GIS contains advanced capabilities to generate clear and accessible maps and data reports that can serve as powerful community outreach and policy design tools.
GIS-based research projects can be developed specifically to address the special vulnerabilities and exposures of children and pregnant women and can target traditionally underserved populations. For example, GIS projects can incorporate birth certificate registries, day care center locations, and elementary school addresses as well as Census demographic variables to more effectively document where vulnerable populations reside. Alternatively, projects could overlay Superfund sites and Toxics Releases Inventory sites to better understand chemical exposures. Geologic features influencing exposure to environmental contaminants, such as flood plains, lakes, rivers, and topography, can also be incorporated.
These large and varied datasets -- particularly in combination -- create special challenges for the statistical analyst. A statistical approach particularly well-suited to spatial analysis is Bayesian hierarchical modeling, a popular methodological tool for spatial analysis that provides powerful model-based inference. Many datasets do not arise from designed experiments. Rather, various forms of observational data exist that may contain clues to explain complex environmental health questions. This requires the construction of models that provide reasonable approximations to environmental health research questions. These models must take into account the spatial and temporal elements of the data. To make best use of such data, methods must (1) facilitate data reduction; (2) accommodate missing and sparse data; (3) appropriately handle misalignment (in space and in time) of data sources; and (4) accurately and efficiently capture uncertainty in inference drawn from this modeling. Bayesian hierarchical modeling meets all these needs.
CEHI has four specialized approaches to conducting research:
An example of CEHI's application of GIS technology is the Mapping for Prevention project, which uses GIS spatial analysis of county tax assessor, U.S. Census, and North Carolina blood lead screening data to categorize housing risk levels in multiple North Carolina counties. Presented below in Figure 1 is a sample priority mapping drawn from the Durham County GIS project to demonstrate how spatial analysis can help identify children at high risk for exposure to lead. Dark blue areas represent Priority One parcels, estimated to be most likely to contain lead paint hazards. Priority Two and Three parcels are colored medium and light green, respectively, and are less likely than Priority One parcels to contain lead paint hazards. Priority Four parcels are yellow and least likely to contain lead paint hazards. White areas represent commercial or industrial properties.
Figure 1. Lead Priority Model for a Durham County, NC Neighborhood
Figure 1 demonstrates how the GIS model might shape preventive intervention programs. The intersections of Enterprise Street with South Street and Fayetteville Street with Dunbar Street (note the highlighted stars) represent two neighborhoods with high risk for lead exposure. The detail provided by these maps allows for block or even house-level planning for intervention programs.