Integrated health care data from a broad set of sources is required for many health care purposes including assuring high quality care delivery and enabling patient-centered outcomes research. However, health care data is generated across many independent systems where data is stored as separate islands with different patient identifiers, resulting in fragmented and incomplete patient information. Therefore, effective evidence-based patient matching methods are needed to maximize the accuracy and completeness of health care data. There is a limited body of research focused on patient matching methods and there have been few formal, comprehensive evaluations of consensus-based matching strategy recommendations using real-world, heterogeneous health care data. The ?patchwork quilt? collections of clinical data spanning multiple systems are increasingly common and prior matching studies fail to reflect challenges faced by these data. Thus, evaluating the performance of best-practice recommendations for real-world, robust, accurate patient matching methods in contexts reflected by health information exchanges and other emerging large health care data sources is necessary to provide evidence informing emerging best-practice recommendations for patient matching. While subject matter experts with substantial operational experience informed recent recommendations, there is currently an incomplete peer-reviewed evidence base to fully support the feasibility and effectiveness of recent guidance. Without further formal evaluation to strengthen and refine these recommendations, organizations may be less inclined to pursue improvements or they may implement methods of little benefit. Our long-term goal is to ensure a sustainable learning health care system infrastructure, which includes accurate, consistent and efficient patient identity management. The next step in achieving that goal is to contribute to the current minimal body of patient matching evidence to inform processes, policy discussion, and technology that support consistent, accurate, and efficient patient identity management methods. To address the limited body of knowledge for real-world patient matching, our team has embedded an unparalleled in-vitro patient matching research laboratory in the nation?s largest health information exchange, which contains hundreds of diverse operational clinical data sources. Within this laboratory we have implemented, evaluated and deployed novel and practical methods for improving patient matching that have improved many specific real world clinical, public health, and research processes. Consequently, we are well positioned to evaluate the impact that emerging consensus-based best practice recommendations will have on improving the quality, standardization, and discriminating power of data collected in a broad set of routine health care settings. We will further evaluate the performance of optimized matching methodologies in the same context. Such evidence can meaningfully inform next steps in the formulation of the nationwide patient identity management strategy.
We will implement emerging recommendations for matching data enhancements in combination with novel matching algorithms enhancements and measure the resulting matching accuracy improvements. Such evidence-based outcomes can inform future formulations of the national patient identity management strategy.
