AMIA: FHIR is not suitable for research, needs NIH R&D funding

The American Medical Informatics Association wants the National Institutes of Health to fund research and development to advance HL7’s Fast Healthcare Interoperability Resources standard.

AMIA submitted comments to NIH in response to the agency’s request for information seeking input from stakeholders on how FHIR could be used to capture and share clinical data for biomedical research purposes.

According to AMIA, it is critical that NIH assume a leadership position to coordinate a research and development strategy for using FHIR for research and that the agency devote “substantial resources” to the effort.

Specifically, AMIA recommended that NIH directly fund FHIR research and development through grants; indirectly fund FHIR through special emphasis notices and project requirements that prioritize projects that will use FHIR; and educate the research community and help represent it in activities supported by HL7, the Office of the National Coordinator for Health IT and other standards developing organizations that have an interest in FHIR.

“Today, most clinical data used for research is extracted from EHRs, transformed, and then made available for research,” wrote AMIA’s President and CEO Doug Fridsma, MD, in a letter to NIH. “The variable implementation of EHRs and the idiosyncratic mappings of clinical data within EHRs have a cascading effect on data quality for research and this challenge will not be obviated by FHIR. Different EHR developers are using different versions of FHIR and even within the same version of FHIR, there may be varying degrees of consistency and constraints applied to the same FHIR Resource across implementations.”


Fridsma warned that “until there is more uniformity in how EHRs generate clinical data using FHIR, those data will need a high degree of additional, potentially manual, curation, and data quality will remain a central constraint to fully appreciate FHIR’s capacity to be used for research.”

At the same time, Fridsma noted that a limitation of the current version of FHIR is that it only enables extraction of data for a single patient at one time, while a large number of research use cases call for analyzing data from a larger population of patients.

“A new set of capabilities in FHIR, referred to by HL7 as ‘bulk FHIR,’ is currently being developed, but has not reached the standard stage,” according to Fridsma. “Adoption of ‘bulk FHIR’ by the vendor community is essential to fulfilling the promise of FHIR as a way to democratize clinical data for research.”

HL7 CEO Chuck Jaffe, MD, agrees with Fridsma that FHIR is in need of NIH’s financial support to ensure that the standard can serve the needs of researchers. “The bulk (FHIR) data transfer allows you to prospectively look at hundreds of thousands of patients,” says Jaffe.

Kenneth Mandl, MD, director of the Computational Health Informatics Program at Boston Children’s Hospital, leads the bulk (flat) FHIR initiative in collaboration with HL7.

“Right now, as you know, EHRs can generally export data in FHIR across the SMART on FHIR API—an API which is required under the 21st Century Cures Act—but that is one patient at a time,” says Mandl. “A new standard we developed, flat FHIR is for export of population cohorts from EHRs. CMS has already begun to use it extensively for sharing claims data. We are hoping to get flat FHIR regulated as an EHR certification requirement, to enable ‘push button’ population health.”