Electronic Health Records

Integration & Interoperability News

Longstanding IT Challenges Still Limit Potential of FHIR

A new Regenstrief Institute pilot should demonstrate the potential of FHIR to aggregate patient data as well as highlight challenges to using the health IT standard, API.

By Kyle Murphy, PhD

- Members of the Regenstrief Institute showed off the potential of Fast Healthcare Interoperability Resources (FHIR) to aggregate and merge patient health data from separate data sources as a proof of concept for a new pilot of the health IT standard.

Regenstrief on FHIR implementation and use

"We can really stitch together information in various sources using FHIR in a way that is user-centered and would be accepted by physicians and patients," Regenstrief Institute investigator and Clem McDonald Professor of Biomedical Informatics at Indiana University School of Medicine Titus Schleyer, MD, PhD, recently told HealthITInteorperability.com.  

The Regenstrief approach required FHIR implementation on both sides of the handshake — in this case, between an Epic EHR using the open.epic API and the Indiana Network for Patient Care (INPC) using a previous version of FHIR.

"It's not a full implementation on the INPC of the current release of the FHIR standard, but it was enough to do this proof of concept to connect in this case an Epic instance (the open.epic API) and connect that with data in an INPC sandbox," Schleyer explained. "We made sure that the patient identifiers were the same and that we had some overlapping data that you could merge."

The genesis of the project actually dates back several years when members of Regenstrief came in touch with FHIR architect Grahame Grieve.

READ MORE: 3 Programs Supporting Health IT Standard FHIR Development

"About two and a half years ago when FHIR got started, we talked to Grahame Grieve who pushed FHIR in a major way," Schleyer recalled. "We've had close contact with Grahame and followed very closely the development of the specification and other implementations. The conclusion then was to put FHIR one the Indiana Network for Patient Care because it takes two to tango. You need your EHR and other information source — it doesn't have to be an HIE."

Funded internally by the institute, Schleyer and his team set their sights to operationalizing the FHIR standard and API for the purposes of assembling health information from different EHR systems. According to Schleyer, the growing interest of EHR vendors and health IT developers in FHIR has helped promote interest in the group of specifications and APIs.

"Many EHR vendors have committed to FHIR and implement it as much as they can at this time — Cerner, Epic, and all the major ones. That is one end of the equation," he added.

Deploying FHIR at least partially in INPC completed the equation and served as a means of testing the role of FHIR as health data aggregator.

Tempered expectations

READ MORE: ONC Celebrates FHIR-Powered Interoperability Demonstrations

Despite its early successes with assembling health information from different EHR systems, the team at Regenstrief is aware of the limitations of any health IT standards, including FHIR.

"Obviously, FHIR is a very young and dynamic standard. Everyone thinks it's going to solve all the world's problems, which it will not. But it has a lot of promise," said Schleyer.

That dynamism could easily led FHIR down the road travelled by many another health IT standard, a worry for Indiana professor. "My worry is that there is the potential for FHIR to be balkanized — too many implementations and versions of it — that I'm hoping that the Argonaut Project will help us avoid that potential," he noted.

That being said, the potential of FHIR has already compelled EHR vendors and competitors Epic Systems and Cerner Corporation to join forces via the privately-funding Argonaut Project.

"If you compare that with the various versions of HL7, there has been a level of enthusiasm about FHIR that you usually don't see in the standards world," said Schleyer. "There is rapid uptake. The vendors are definitely committing to it. When you see competitors like Cerner and Epic jumping on the boat and being in the Argonaut Project when they usually fight each other, it's actually a good sign."

READ MORE: EHR Data Viewing Tops List of Most Common Healthcare App Uses

Another challenge for FHIR implementation and use is varying in data vocabularies, something the Regenstrief team experienced in a recent connectathon.

"One of the challenges we had was that the data on either side weren't always the same," Schleyer explained. "For instance with lab results, unless you have a common mapping terminology such as LOINC, you have to match a vendor- or EHR-specific test vocabularies to whatever your target vocabulary is. We had some data-matching issues, and as you know they are universal. It is an ongoing challenge."

But more so than either of these is the challenge of patient matching. In FHIR sandboxing tests, this challenge was avoidable.

"One of the challenges we didn't have was knowing which patient records matched each other," Schleyer acknowledged. "In the Indiana Network for Patient Care, we have a global patient identifier that is unique to the INPC. Underneath that is a collection of all the identifiers that a patient has in the contributing health system. As soon as we with our matching algorithm connect a newer patient record sent from a new institution in with the global identifier, we have pretty good confidence that it's actually the same patient."

In a real-world setting, patient matching represents a significant roadblock to information sharing, in large part the result of the federal government continuing to withhold funding for a universal patient identifier.

"The federal government made the decision a while ago not to allow a national patient identifier," said Schleyer. "If you look around the world, that makes us fairly unique. Most of Europe, for instance, has standardized identifiers for healthcare, including Germany where I come from. We have to struggle with that matching."

The lack of common patient identifiers makes having confidence in matching patient records nearly impossible.

"In the end, there's only so much confidence you can have in a match depending on what your source data are. If you have a lot of fields and they all are identical, then you can say you're pretty sure that's the same patient," Schleyer stated.

According to the investigator and biomedical informaticist, the solution to this challenge is much likelier to come from the ground up.

"What's going to happen is that the movement toward using unique identifiers for patients will come from the grassroots," he maintained. "There are some movements across the country where patients can opt-in to a patient identifier. Is there anyone who would object to that — because the risk that you have is either information is missing from your record or maybe worse you get the information from somebody else if you have a common name. People intuitively get that."

FHIR has a ways to go before its potential is achieved, but researchers such as Schleyer are already showing glimpses of its ability to unlock valuable health data in a secure and efficient way.

Dig Deeper:

Health IT Standard FHIR Ready to Advance Interoperability
Will FHIR, APIs Help or Hinder Health Information Exchanges?