ABSTRACT

The U.S. healthcare system lacks incentives and quantitative evaluation tools to assess coordination in a patient’s care transition process. This is needed because poor care coordination has been identified by many studies as one of the major root causes for the U.S. health system’s inefficiency, for poor outcomes, and for high cost. Despite efforts dedicated to improve care coordination, technical gaps still exist on how to understand and assess care coordination in a quantitative, effective, and methodological manner. Existing literature primarily focuses on applying case studies and qualitative measures to evaluate care coordination activities and their related outcomes. However, utilizing care transition dynamics to quantitatively evaluate the effectiveness of care coordination in terms of outcomes remains an open problem.

This dissertation proposes an integrated data-driven analytical framework for quantitative evaluation of care coordination effectiveness under care transition dynamics. The main objective is to develop quantitative metrics for decision support for identifying care coordination opportunities. First, data structuring, data processing, and aggregation techniques are proposed to extract inpatient episodes that serve as the fundamental units of the proposed framework. Next, a set of metrics are developed to assess care coordination effectiveness from three perspectives, including care transition dynamics, major interactions among patients and providers, and defined patient outcomes – specifically, 30-day hospital readmissions. Finally, an integrated metric model is developed to validate the feasibility and effectiveness of the proposed metrics as well as to identify the most relevant set of metrics for defined system features.

For evaluation and validation, the proposed framework has been applied to a healthcare claims dataset. A number of metrics have been identified that have significant impacts on the corresponding outcomes. In addition, a set of well-performing metrics has been identified by the integrated metric model for defined system features. Outputs of the proposed framework can be utilized by healthcare professionals and administrators as decision support for understanding the unique features of a specified patient-provider network as well as identifying the opportunities for care coordination activities to improve care transition quality and associated patient outcomes.