- Healthcare organizations participating in the value-based care environment are in need of strong, secure data pipelines that can help them communicate across disparate health systems to coordinate and streamline patient care.
As providers build out the infrastructure required to support meaningful health information exchange, they are keeping their eyes on the ultimate goal of complete semantic interoperability: the ability to seamlessly exchange information that can be accepted, read, and understood across any system created by any health IT vendor.
Using the Fast Healthcare Interoperability Resources, now commonly known by the acronym FHIR, to build out data exchange capabilities has become a popular approach. The internet-based standard offers common ground for providers and developers who wish to move away from proprietary architecture and a “walled garden” approach to data management.
However, FHIR presents its own set of challenges that may not move the industry as quickly towards true semantic interoperability as it may like, cautions John D’Amore, President and Chief Strategy Officer at Diameter Health.
While FHIR is a step in the right direction, the standard must be developed and deployed in a manner that avoids duplication of a fragmented and siloed technical interoperability landscape.
“Large health systems, HIEs, and payers have essentially solved the challenge of transporting clinical data. But when they tried to integrate the transported data from multiple sources, they’ve had difficulty achieving semantic levels of interoperability,” D’Amore explained to HITInfrastructure.com.
There is enormous variation in the way information is collected, coded, and processed, he said. From ICD-10 and SNOMED to LOINC, RxNorm, and NDC, data can be created and stored in a dizzying array of formats that each include their own quirks and rules.
“That’s where the problem lies in healthcare,” stated D’Amore who is also an editor of HL7 and CCDA standards. “For example, a patient is on a medication and that medication data is collected from one source that uses National Drug Classification (NDC) standard, but another source with the same medication information about the same patient uses the standard put out by the National Library of Medicine, RxNorm. There are three or four other proprietary ways medications can be referred to as well.”
Different electronic health record (EHR) systems may prefer to ingest and display medication data in only one of these formats. While many systems offer crosswalking capabilities to translate one coding set into another, the process can be imprecise and leave providers with confusing, incorrect, or missing information.
“Each data standard is referring to the same drug,” D’Amore said, “but all the different terminologies and ways to express the same concept can be very noisy and impede an organization’s ability to use that data for analytics or population health.”
FHIR offers a potential solution to the problem of technical interoperability - getting any data in any format from system A to system B, D’Amore says, and it’s a positive development for the industry.
“FHIR uses much more modern ways to express data than prior standards,” he said. “The CCDA, for example, is an XML standard that really formed the basis of what the industry was doing before FHIR came along.”
“It formed a core part of the meaningful use requirements for EHR adoption and interoperability. You had to be able to send a CCD or CCDA document that includes a summary of a patient from your system to a host of other systems.”
While the CCD and CCDA standards did help to increase health information exchange, FHIR is a more flexible and modern approach that has quickly become the standard of choice for providers and developers.
“Most would agree that FHIR is an inevitable progression for where health interoperability will go,” he said. “It's just a matter of timing in terms of how fast we get there.”
However, FHIR might not be the ultimate answer to the semantic questions of ensuring that the data payload is understood on both sides of the divide.
“The first thing organizations need to understand is that the transport of data is not the fundamental problem,” D’Amore explained. “While FHIR is a great standard for transporting data, a CCDA and HL7 will also get data from point A to point B. However, none of these standards will currently solve the semantic issue with data exchange.”
FHIR has been developed through a community effort guided and overseen by HL7 International. The private sector initiative has strong support from federal entities like the Office of the National Coordinator (ONC), but the government has not mandated a particular approach or implementation strategy.
As a result, different versions of the standard - which only recently came out of “draft” status - can compound the challenges of both technical and semantic interoperability, D’Amore asserted.
“Many major EHR vendors are using an API like FHIR, but they aren’t required to have common support for resources or clinical demands,” he continued. “Medications may be sent one way using one clinical domain, but another EHR may not have it available in that way. The receiving EHR may try to patch medications into a different resource. That’s one of the impediments of FHIR.”
“It’s not good to have multiple versions of FHIR because you can’t achieve the ultimate goal of getting different health systems to communicate consistently across different platforms.”
Organizations concerned about semantic interoperability should focus first on basic issues of data integrity and quality, D’Amore advised.
“Clinical projects require key clinical domains to be normalized to a level where they can be acted upon reliably, regardless of whether that data came from Epic or Cerner, Allscripts or MEDITECH - and regardless of what technical standard is used,” he said.
Organizations may also wish to put some pressure on their EHR vendors to collaborate more effectively around implementing uniform and widely-adopted FHIR standards, he said.
“The inability to write back information to an EHR has been a significant limitation in terms of actually delivering things that have a positive ROI or a positive impact on patient care,” said D’Amore. “It’s unclear how that problem is going to be solved in the near-term because most providers are skeptical of external applications writing information into their EHRs.”
“How do we get to a scenario where you can simply and easily identify that this patient needs a flu vaccine and that patient needs a mammogram? How do you inject that data into every EHR in a way that complements the workflow of the physician and allows them to see that information when and where they need it?”
The technical components of FHIR are only one part of solving those fundamental issues of data analytics, exchange, and management.
“FHIR has a way to support that vision from a standards point of view, but almost no one's implementing FHIR in such a way that you can actually execute upon that,” D’Amore observed.
Both vendors and providers should adopt approaches that treat technical and semantic interoperability as inseparable, he advised.
“Almost every problem within every project in the interoperability space has to appreciate the semantics of the clinical data that you're talking about,” he stressed. “The magnitude of that problem is a lot bigger than other industries, and it leaves healthcare stakeholders with a lot of major issues to work through.”
“We will have to start small when trying to solve for these problems. Find your low-hanging fruit and figure out what you can do to create short-term wins that will bring immediate ROI. Then you can start rolling those small successes into larger ones, and building on your knowledge as you go.”