Improving Health Outcomes by Georeferencing Patient Data Warehouses within a Large Academic Medical Center

Background

The increasing size and complexity of health data demand new approaches for organizing, viewing, and analyzing multiple data entities in an integrated format. While traditional relational database queries and reports are valuable for creating useful information from large data repositories, additional tools for analysis, visualization, and knowledge discovery are imperative. The highly complex temporal, spatial, and attribute relationships among data entities within health data are well-complemented by the analytical tools and visual simplicity of geographic information systems (GIS) and associated spatial analysis.

Objective/Purpose

Demonstrate a comprehensive spatial data architecture anchored to data collected in the Duke University Health System Decision Support Repository (DSR).

Provide a better analytical substrate to clinicians, researchers, health services administrators, and community members interested in health and health care decision-making.

Methods and Results

We are building a comprehensive spatial data architecture within the Duke Decision Support Repository (DSR). The DSR is an Oracle database containing data associated with encounters at Duke University Hospital and owned or affiliated outpatient clinics. The spatial data architecture links the DSR patient data to georeferenced demographic, environmental, social, administrative, and community data. New data access protocols and systems have also been established. Multiple users are constructing new applications based on the spatial data architecture.

Discussion/Conclusions

Creating a spatial database from an existing health data repository encompasses significant technological, system design, and policy challenges. Integrated spatial data architectures can support interactive map applications, new spatially-based research, and improved health care decision-making.