Perspective: Data Governance—Making Data Great

Abstract

Health caredelivery in theUnitedStates is undergoingmassive changes. Although the pace at the local level can often be imperceptible, the overall momentum is relentless. Multiple factors are driving these changes—the rate of cost increases for care in theU.S. isunsustainable, theoutcomesofcareareworse than other countries that spend less, and a large portion of the populationdoesnothave real access toongoing care, leading to enormous societal costs. On thehealth caredelivery side, there has been a shift from stand-alone care delivery through stages of collaborative and value-driven health care toward a more accountable care/populationmanagementapproach (►Fig. 1). Politicians, policy experts, and health care administrators long recognized that changing models of health care would require significant amounts of data. This was a major factor in the founding of theOffice of theNational Coordinator forHealth InformationTechnology (ONC), and later theHealth Information Technology for Economic and Clinical Health (HITECH) Act which brought about the Meaningful Use program.1 With the increaseduseofelectronichealthrecords (EHRs), theavailability of electronic data in health care has risen dramatically. Organizations increasingly see their data as a strategic asset. Tomost effectively leverage this asset, data from diverse sources (e.g., revenue cycle, supply chain, quality measures) must be combined for analysis. The use of data from diverse sources has always occurred, but in many organizations, this was often accomplished through establishment of locally curated data collections. Unsurprisingly, organizational leaders found they could get different answers for the same question depending on which group they asked. Because data definitions can be quite nuanced, a team can become reluctant to share datawith the rest of the enterprise. There is a concern that otherswouldnot understand “mydata.” This can then lead to users on other teams to be creative in finding sources for the data they need. By acquiring data sets from diverse acquaintances across the organization, analysts may be using data with somewhat unclear provenance (often many steps away from the original sources) for important organizational reports that drive executive decision making. This can exacerbate the issue of getting different answers from different groups. Technical approaches to improving the situation include establishing a “single source of truth” for enterprise data, for example, via anenterprise datawarehouseor adata federation approach. This can helpwith some of the notions of “my data” versus “our data” in an organization, but does not in and of itself ensure that the data are well characterized and of high quality. As health care organizations struggle to use their data effectively, they realize they need an approach to addressing these issues. Whether they formally label it as such, what the organizations are grappling with is data governance.2,3