Forging Translational Research Networks with FHIR
Forging Translational Research Networks with FHIR
Christopher Chute, MD, DrPH, Johns Hopkins University
Data is the foundation for all of modern science—from hypotheses to inference, conclusions, and reproducibility. Electronic health records (EHRs) have emerged as an important source for interventional, as well as observational, clinical research. The CTSA program aspires to forge synergy across its hubs by supporting federated research at scale, amplifying the precision and power of clinical analytics and discovery over a broad spectrum of clinical problems. The trick to making this federated network work, is to ensure that the clinical data is comparable and consistent—in a word—interoperable.
Research data standards to address the interoperability challenges of translational research have been proposed; however, the joke historically has been that there are so many standards to choose from. Moreover, implementing sometimes idiosyncratic research standards has been expensive, tedious, and ultimately of limited value in the face of standards diversity. Communities could not agree on which parochial data standards should prevail, rending ineffective most efforts expended on data normalization.
The clinical data standards world has witnessed the rise of a new effort for the exchange of comparable and consistent data that has three unique attributes: 1) it is relatively simple and modular, 2) it is based upon conventional computer industry technologies such as REST, and 3) it enjoys unprecedented acceptance, endorsement, and enthusiasm across health providers, payers, academia, government, EHR vendors, and health information technology standards developers. The Fast Healthcare Interoperability Resources (FHIR® fhir.hl7.org ) specification is becoming to clinical data what HTML is to the web—a scalable mechanism to render clinical data in a way that people and machines alike can understand. It is no accident that FHIR embraces many of the most successful and scalable web technologies as its underlying platform. Although FHIR, much like HTML, is not up to the task of single-handedly solving all underlying data incompatibility, its strength is its ability to work in tandem with other more specific standards and services.
The implications and advantages of FHIR for translational research are profound. While developed and supported by the healthcare community, the use of any clinical data for translational research immediately inherits this investment; moreover, the size of this investment dwarfs by orders of magnitude previous efforts to develop clinical research standards. More importantly, EHR vendors have all created FHIR interfaces to the clinical data within them, enormously reducing the cost and burden of data transforms to and from EHRs. More complete and robust FHIR interfaces will likely to be required by draft regulation from the 21st Century Cures Act. The now well-understood advantages of leveraging FHIR for clinical research were ratified by the unprecedented recent notice from the Office of the Director at NIH :
“To encourage NIH researchers to explore the use of the Fast Healthcare Interoperability Resources (FHIR®) standard to capture, integrate, and exchange clinical data for research purposes and to enhance capabilities to share research data.”
This may be the first time the NIH has ever issued such a notice about a data standard.
It is important to recognize that FHIR exists as part of an ecosystem, and should not be construed as a competitor to clinical research common data models (CDMs) such as OHDSI , PCORNet , or ACT . Quite the contrary, FHIR can form a practical hub from which the logic for transforming between these models can be written once and published—dramatically reducing the cost of establishing or maintaining one or more repositories in CDM formats. FHIR is also modular, as demonstrated by the development of many special purpose “resources” (as the modules are called) to support new data types, such as genomic results data .
FHIR is likely to have an enabling effect on translational research, contributing significantly toward the capacity of CTSA hubs to operate as a federated network and accelerating programs such as the ACT network. An intriguing demonstration exists in South Carolina Translational Research Institute leveraging FHIR as a state-wide hub. The Center for Data to Health (CD2H) coordinates tools, maps, and resources, many involving FHIR. For example, we developed tools to transform EHR lab results to standard phenotypes; this work is described in NPJ Digit Med. 2019 (PMID:31119199) and a preliminary demo is at octri.ohsu.edu/hpo_on_fhir . These and other resources will allow CTSA hubs to reduce the cost of achieving interoperability goals, meet emerging data sharing requirements, and support new, analytical applications using big data at scale across the CTSA program.