FHIR Idea Burns Bright. Implementation Flickers

Nobody likes to think about taxes. But when asked to explain the importance of efficient data exchange, Paul L. Wilder, MBA, says electronic tax filing offers an apt analogy. Not so long ago, when a customer of a commercial tax filing business used the business’s software, the facts and figures they entered into the software were funneled into data fields on Internal Revenue Service (IRS) tax forms.

Paul Wilder, MBA

“And for a while, that’s probably how they sent it to the IRS,” says Wilder, the executive director of the Boston-based CommonWell Health Alliance. “...It probably was nuts. It was a printout form, and they probably used OCR [optical character recognition] to put it back in.”

“Eventually it became a data blob,” he says. “And they send just a data blob over [to the IRS].”

Nowadays, tax filing software has largely broken free of forms and PDFs.

Wilder likes that analogy because it demonstrates a shift that the healthcare industry has yet to make. It has one foot in a bygone era of data exchange and the other in a new one. The result is a system rife with inefficiency and the potential for errors.

That may soon change. In September 2024, the Assistant Secretary for Technology Policy/Office of the National Coordinator for Health IT (ASTP/ONC) announced a draft action plan designed to spur federal agencies and other healthcare entities to adopt the Health Level 7 (HL7) Fast Healthcare Interoperability Resources (FHIR) — the acronym is pronounced “fire” — standard for healthcare data exchange more quickly.

The FHIR standard is a set of rules and specifications governing how clinical and administrative healthcare data are exchanged among different entities. The standard is necessary because in the sprawling American healthcare system, health data are stored in a dizzying array of formats, systems and different types of software. ASTP hopes the standardization will help make it easier, for instance, to gather real-world evidence for clinical trials or to relay product surveillance data to federal regulators.

The FHIR standard enables a receiver to ask for specific data from a database rather than having to sift through a stack of documents that may or may not provide a clear answer.

Wilder’s CommonWell Health Alliance is one of seven qualified health information
networks designated by ASTP to help implement better healthcare interoperability. The alliance connects more than 36,000 healthcare providers, and it recently added the laboratory services firm Labcorp. Wilder says Labcorp’s addition was partly designed to push members to use more advanced data-exchange standards.

Harmonizing stakeholders

Chris Decker, M.S.

It takes no special insight to see the benefits of standardization, but making it actually happen is an obstacle course. For example, harmonizing data exchange also requires some harmonization between the world of healthcare and the world of clinical research, says Chris Decker, M.S., president and chief executive of the Clinical Data Interchange Standards Consortium (CDISC).

“As we try to bring those two worlds together, we must understand the needs and use cases, put aside past control of each space, and be willing to work together to find the solution that works best for what we are trying to accomplish,” he explains.

Decker notes that current FDA regulations require that regulatory submissions for new medical products align with CDISC’s global research standards as a means to ensure consistent terminology. Those standards have evolved over 20 years and are used by investigators worldwide, he notes. In its action plan, ASTP calls for connection or harmonization between FHIR and
CDISC standards.

“However, they have not yet addressed the need to standardize the content to ensure semantic interoperability, which is required to realize the true efficiency of standard data,” he says. Decker hopes ASTP will leverage existing standards from organizations like CDISC rather than trying to “reinvent the wheel.”

“The different key stakeholders must put aside their fiefdoms of standards and terminology, leverage the work that has been done in the past by existing organizations, and be willing to come together to align on the best approach that can support the various use cases, regardless of their own past work,” he says.

Decker adds that it will be important to get electronic health record makers on board as well to ensure semantic interoperability and to export a standard set of data elements and terminology that can be integrated with clinical research data.

Jennifer Stoll

Jennifer Stoll, chief external affairs officer at OCHIN, a network of 300 independent community health organizations and approximately 2,000 care delivery sites, says the action plan also fails to address the cost of
upgrading systems.

“[If] providers don’t have the modernized and interoperable tools they need to meet these standards and a technology partner to help them get the most out of their system, large swaths of our country will be left behind and the gap between the haves and have-nots will only widen,” she says.

Providers in OCHIN’s network serve a large number of patients who live below the poverty line and have social risks such as food insecurity or housing instability, she notes. While OCHIN supports the goal of transitioning to a more efficient healthcare data exchange system, Stoll says switching to HL7 FHIR will require significant time and money. “For providers in rural and medically underserved communities who are already caring for a disproportionate number of medically complex patients with fewer resources than providers in more affluent communities, this is especially untenable,” she says.

AI implications

Stakeholders are also concerned about how the current fragmentation in healthcare data exchange will work with artificial intelligence (AI). Wilder notes that current data-exchange inconsistencies can lead to data duplication and data gaps, which harm the ability of AI and machine learning systems to generate meaningful insights.

“In an age of AI, gaps in data can lead to data hallucinations and inaccurate insights,” he says.

Stoll pointed out that if underserved communities are left behind in data exchange, their data will be left out of AI insights. “Representation is even more important as the AI revolution stands to both offer enormous potential and great risk, including increased costs, to the communities in our nation that are already left behind,” she notes.

It is also possible, though, that ASTP’s action plan itself could be left behind. Wilder says he is glad to see ASTP take the lead in pushing for FHIR. “The question is really going to come down to, will those other departments and agencies listen?” he says. The draft plan encourages federal agencies and partners to utilize the action plan, but given the costs, the technical hurdles and ASTP’s lack of authority to force changes, it is unclear whether other agencies will follow the plan.

The push for FHIR might be affected by the change in presidential administrations. Micky Tripathi, Ph.D., M.P.P., left his role as assistant HHS secretary for technology policy in January. At press time, no replacement had been named. HHS did not respond to a request for comment on the action plan’s status.

Decker says industry stakeholders may not be able to rely on the government to push for a full implementation of FHIR. “It may be necessary for the [healthcare and clinical research] industries to drive instead of government in creating robust processes and high-quality data for the sake of their own efficiencies and cost savings and support accurate results using AI,” he says.