Hospital Networks of Shared Patients and Engagement in Health Information Exchange

Abstract

Although the healthcare delivery system is composed of an array of organizations that are linked through important, enduring, and complex ties, the healthcare delivery system is rarely explicitly conceptualized or measured as a network. In consequence, we know little about how the enduring but often informal relationships between organizations shape their behavior in terms of the decisions that they make, the quality of care that they provide, and the efficiency of that care. Using techniques developed in the multidisciplinary field of network analysis, I sought to better understand two important facets of health care that are intrinsically linked to the network perspective: the fragmentation of patients’ treatments between multiple hospitals, and hospitals engagement in electronically sharing patient information. By analyzing networks of shared Medicare patients treated at multiple hospitals, I first identified dense networks of hospitals that are closely interlinked through many high volume shared patient connections and are therefore likely linked through complex collaborative and competitive relationships. I then characterized these networks to identify arrangements of patient sharing that allowed hospitals to better manage care fragmentation. I found that more concentrated networks, in which hospitals shared most of their patients with few important partners rather than a large number of other hospitals, and more centralized networks, in which the network is arranged in a hub-and-spoke model, were associated with more efficient, higher quality care.

I next described three different approaches to health information exchange and the logic of participation in each approach with specific emphasis on the value of the enterprise approach for connecting a smaller number of providers and the community approach for facilitating broader connections between more partners. I then investigated whether the choice that hospitals made about how to electronically share patient information was shaped by their networks. I found that hospitals with and within more concentrated patient sharing networks were more likely to engage in enterprise exchange while hospitals with and within less concentrated networks engaged in community exchange more frequently.

Together, these findings offer novel insights into the network features of hospitals and how they relate to important healthcare processes and outcomes. More concentrated, centralized networks appear to perform better and these features may be one reason for variation in the cost and quality of care across the nation. Similarly, policy changes designed to shape how healthcare organizations interact and who they interact with—like accountable care organizations, bundled payment initiatives and patient center medical homes—may be more successful if they reinforce beneficial network attributes. Further, as policy efforts designed to facilitate the sharing of information between healthcare providers continue, it will be crucial to allow flexible adoption of different approaches to health information exchange and to support hospitals that engage in an approach to information exchange that benefits communities.