Pop Health on FLAT FHIR: A SMART Approach to Universal Healthcare Reporting

In September of 2018 the Boston Children’s Hospital Computational Health Informatics Program (CHIP) was awarded a Leading Edge Acceleration Project (LEAP) in Health Technology Cooperative Agreement from the Office of the National Coordinator for Health Information Technology (ONC) for “Pop Health on FLAT FHIR: A SMART Approach to Universal Healthcare Reporting”. 

Eight years after our original SMART project under ONC’s Strategic Health IT Advanced Research Projects (SHARP) Program, we propose the next step. We catalyze an ecosystem for accessing and analyzing, without special effort, data on whole populations rather than one patient at a time. We have made real progress toward this vision, working with HL7 and ONC to define the population health analog to the SMART API—the FHIR Bulk Data Export API. The output is “Flat FHIR,” an easily consumable flat file. Currently, a provider using EHR data to meet reporting requirements on population-level quality or cost measures requires a customized and prohibitively complex process to extract, transform, and load data into a separate analytic engine.

Our vision is seamless data exchange, via an API, between provider organizations and third parties. We propose a use case of exchange of EHR and claims data and derivative metrics between a provider and a payor. Toward this end, we design, develop, and test a substitutable population health analytics app, SMART-PopHealth, which enables a payor to access permitted data and metrics on covered populations, directly through the API. We test it in a real-world Accountable Care Organization.

The immediate benefit is to reduce or eliminate burdens on at-risk providers for reporting performance metrics to payors. But downstream benefits at the health system level are far greater. If our research shows that the Flat FHIR format provides a rich, production-grade substrate for effective, easily reproducible dataset creation, sharing, and analysis, it could become a lingua franca for a learning health system based on apps and analytic engines for uses as diverse as infectious disease surveillance and pragmatic clinical trials. Planning toward this next stage, we propose, in the out years of the project, implementation of federated data structures enabling query across multiple sites for health system-level intelligence.