Tennessee’s Health Enterprise Warehouse: An Integrated Data Solution to Promote Data Sharing and Rapid Decision Support

BACKGROUND: Although the Tennessee Department of Health (TDH) houses large amounts of health data, like many other government organizations, we lack infrastructure to support data sharing and rapid decision support. This is due in large part to siloed data systems and stewardship. Although data warehousing (DW) is a gold standard in the corporate world, it is relatively new in the public sector. In an effort to integrate these otherwise disparate systems, TDH has begun to lay the groundwork to build a robust health enterprise warehouse (HEW). 

METHODS: DW is the integration of multiple data sources architected to facilitate analytical performance. To implement this work, TDH established a multidisciplinary team including staff from information technology, development/architecture, epidemiology and executive leadership. TDH is employing an Agile development approach for the build utilizing two week sprints. All data will pass through seven phases before they are completed in the HEW: submission, governance, discovery, documentation, architecture, processing, and consumption. The HEW is being built using Microsoft SQL Server 2016 on a server running Microsoft Server 2012 Operating System.

RESULTS: The HEW is initially focused on meeting analytic needs to address the prescription drug overdose (PDO) epidemic. It will integrate prescription drug monitoring data with hospital discharge and vital statistics data, weekly hospital reporting of nonfatal overdoses, and potentially law enforcement and mental health information. The HEW will allow for multiple levels of analytics, including identifying epidemiologic trends and building risk models to identify factors that impact poor health outcomes. Eventually, the HEW will expand to include data necessary for an array of key performance indicators across the department. Some of the biggest benefits of using a DW approach are that it allows for the standardization of common variables, applies linkages at the outset, reduces duplication of effort, decreases turnaround time for analyses, and centralizes data to facilitate rapid decision support. Challenges include addressing data quality, data access from release to delivery, governance, and determining the minimum data elements required. To overcome some of these challenges, TDH established a governance committee to assist with access and the multidisciplinary team identified four use cases to assist with minimum data requirements. 

CONCLUSIONS: Despite the substantial effort required in proper DW (as opposed to data storage), we believe that this solution will allow for yet unobserved data connections and an ability to produce and track indicators across data sources that are critical to our public health mission.