Development and implementation of a distributed health data network

Background:The Institute of Medicine, the Federal Coordinating Council for Comparative Effectiveness Research, and others promote use of routinely collected electronic health information for “secondary” purposes such as comparative effectiveness research, medical product safety, and public health surveillance. Additionally, the Patient Protection and Affordable Care Act proposes to these data to generate evidence to help improve care. Gathering sufficient data for these purposes will require the ability to efficiently use electronic health data held by multiple organizations. There is no consensus regarding whether to create multi-payer databases for these purposes, or to use a distributed approach that keeps data under the control of the data holders. Objectives: Enhance our existing, open-source distributed data network software (PopMedNetTM) based on feedback from a pilot study and implement a fully-functional system across multiple partners. Methods: After implementing a pilot distributed network, we solicited data partner and user feedback regarding system functionality, usability, and acceptability. We then enhanced the system architecture and specifications to address the feedback. The new architecture was implemented and tested using a rapid-cycle prototyping approach. Results: Feedback from the pilot identified several enhancements. These included the query interface design, query features (e.g., calculation of prevalence rates), security and hosting changes (e.g., a Federal Information Security Management Act compliant hosting center), development of new system “roles”, and improved query management for data partners. These and other changes were successfully implemented, tested, and released for use in several different distributed data network projects, including the FDA's Mini-Sentinel project, a public health surveillance network in Massachusetts, and an AHRQ-funded comparative effectiveness network. A “future state” document details scheduled software enhancements. Conclusions: The PopMedNetTM design addresses data partner and user needs for implementation of secure distributed data networks that obviate the need for centralized databases. Continued use will inform future develop priorities and directions.