FHIR: The Last, Best Chance to Achieve Interoperability?

Can an impassioned band of savvy, battle-tested techno-optimists save our healthcare system from its worst instincts, and deliver at least a soupcon of real interoperability?

That’s the question I found myself seriously contemplating after last week’s JP Morgan Healthcare Conference in San Francisco, and in particular after an inspiring data science roundtable discussion led by Aneesh Chopra (the first Chief Technology Officer of the United States, now at Hunch Analytics) and Claudia Williams, a senior advisor in the White House Office of Science and Technology Policy (OSTP).

The FHIR Engine

The (not so) secret sauce in question is the FHIR (pronounced “fire”) interoperability standard that has emerged from the non-profit HL7 organization, and functions as a universal adaptor, allowing some types of clinically-relevant data types to be shared easily and securely. Remarkably, most major electronic health record (EHR) vendors – including athenahealth ATHN +0.00%, Cerner CERN +0.00%, and yes, Epic — have signed onto this concept, and agreed to support an early effort at implementation, known as the Argonaut Project.

The broad vision – expressed cogently here — is that the ability to reliably extract select data types from any EHR system will allow the development of high quality applications that will be both widely applicable and commercially viable, since they’ll work across all major EHR platforms. Initially, the apps will aim to benefit individual patients and clinicians, but ultimately this approach may facilitate the rapid data exchange between enterprises, i.e. health systems.

Proponents argue the approach “brings Facebook-, Google- and Amazon-like thinking to healthcare IT,” as John Halamka, a leading physician-informaticist and Chief Information Officer of the Beth Israel-Deaconess Medical Center, told Modern Healthcare’s Joseph Conn when the Argonaut program was announced just over a year ago.

Rather than “moving static, electronic documents—such as care summaries—from one provider to another,” as Halamka explained to Conn, “imagine Epic and Cerner and athena had a hook on their EHRs.” This “hook,” or application programming interface (API), is the key element, and what can enable EHRs to communicate interactively, and in a standardized way, with apps and eventually even other EHRs.

“That’s a different kind of enabler than sending a package of information from place to place,” Halamka told Conn.

“We are clearly now in the era of API access for HIT [health information technology],” agrees another advocate, Cerner Senior Vice President David McCallie.  “Standard-based (or at least widely available, a la Apple) APIs have transformed most every other industry, and now it’s healthcare’s turn.”

Adds Gil Alterovitz, a FHIR champion and core faculty member of Boston Children’s Hospital Computational Health Informatics Program, “FHIR is for real. It will revolutionize medicine much in the same way ATM’s changed banking. Both enabled growth by decentralizing core functionality, enabling novel applications, and expanding the overall system’s reach.”

“Support for FHIR continues to heat up,” notes Epic spokesperson Dana Apfel.  ”We expect FHIR to be adopted across the healthcare IT field, and ultimately to be as widespread as HL7’s other standards.”  Apfel adds that Epic launched a “sandbox for developers” in 2014, and “began supporting FHIR in our 2015 release.”

Government Nudges

As Chopra sees it, at least three powerful regulatory and reimbursement nudges seem to be driving momentum: Meaningful Use 3 (MU3); the Precision Medicine Initiative and other government-sponsored programs; and HIPAA (the Health Insurance Portability and Accountability Act that governs health data security and privacy).

  • Meaningful Use 3: Healthcare providers (under threat of financial penalty) must demonstrate they are using a “certified” EHR that meets particular requirements; the current requirements are called “Meaningful Use 2” (MU2), but the next set, MU3, “explicitly state that the patient can designate an app of his/her choice,” according to Chopra, adding, “In other words, a health system can’t ‘block’ a patient from using a competitor system’s app to consolidate the portals if they happen to see doctors in both.” Due to extensive stakeholder resistance (see here, for example), and recent comments from CMS Administrator Andy Slavit, many now assume MU3 is dead, though Chopra counters the provision enabling patients to designate an app will remain.  Chopra also notes that independent of MU3, the HHS Office of Inspector General recently warned that if a health system subsidized a community doctor’s EHR, the health system, in Chopra’s words, “must open up the APIs as well” or else face financial penalties.
  • Federal initiatives: When President Obama launched the Precision Medicine Initiative last January, he called for “consumers to have access to their own data, and to the applications and services that safely and accurately analyze it.” This year, Chopra says, “NIH will fund work to help patients access their clinical data and make it available to the research effort.” Providers seeking to participate in the program, Chopra believes, would likely allow such “direct enrollment” functionality which could then be useful to connect to other patient-designated apps.  Additional programs expected to leverage FHIR include a “blue button on FHIR” option Medicare reportedly plans to have ready for pilot launch by the end of the year, as well as some health centers receiving federal grants to help meet MU requirements.  While FHIR is not explicitly required in these grants, Chopra says he’s “confident several centers will use the funds to front load MU3 requirements [i.e. adopt FHIR], especially those involved in accountable care organization (ACO) arrangements where access to patient data is a higher order need.”HIPAA: Health systems have long invoked HIPAA-associated concerns to explain their reluctance to share data – to the frustration of many patient and data advocates (eg getmyhealthdata.org). But now, Chopra says, “HIPAA has clarified that an individual has the right to request copies of their data be transmitted in the manner they request, eventually via open API as health systems adopt the capability, and even if the health system fears the patient is putting his/her data at risk.”Mixed ReceptionSome vendors, such as athenahealth and Cerner, seem especially excited about the potential of FHIR.“FHIR is the most promising API for widespread HIT adoption,” Cerner SVP David McCallie tells me.  “I don’t think there are any other close choices.”

    “The conditions are primed for interoperability,” Ed Park, Chief Operating Officer of athenahealth, told me this week, and “FHIR being talked about because it could be the match to light the fuse– to bridge from political pressure to reality.”

    But not everyone is so sanguine. In his 2016 predictions, noted Health IT analyst John Moore explicitly anticipates that FHIR “doesn’t become mainstream.”   Instead, he expects it “will be on the edges rather than where the bulk of patient data resides – in EHRs. FHIR support will continue to pop up in places where few providers can actually use it. No major EHR vendor will release a comprehensive set of production-ready FHIR profiles and resources in 2016.”

    Moore expanded on this, telling me:

    “FHIR is a very interesting standard for the healthcare sector and has the potential to enable an API economy within the sector. Several vendors are aggressively building out a FHIR layer to their platform and spoke to a couple at JP Morgan [Healthcare Conference] that now have it fully built out and are deploying it at customer sites. Note that both of these vendors are not EHR vendors. EHR vendors have more to lose, thus will slowly move in this direction, but move they will – the writing is on the wall.

    That being said, it is one thing to offer the capability and quite another for it to be adopted and deployed. FHIR officially unveils itself in 2017. It will be 2-3 years after that date that we see adoption beyond early innovators/adopters.”

    Pallav Sharda, another health-IT expert, is even more concerned. “I’m pessimistic about FHIR catching on natively,” he tells me. “FHIR is better than anything else that came before it, but it’s being promoted by the same handful of restless free-spirit individuals. They keep talking/arguing about it while the real market moves away into other realms.” Sharda believes the “market is going to be driven by vendor APIs (Athena, Allscripts leading), not a standard.”

    Alterovitz, however, contends that over the last year, FHIR has moved from fringe to mainstream, gaining support from major EHR vendors, and finding expression in key federal initiatives (as Chopra has indicated).

    Several EHR vendors and FHIR proponents, while excited about what FHIR may offer, emphasize the need to modulate expectations.

    “I am pretty sure that FHIR changes the game, but I don’t think it will be as fast a transition as some of us would like,” says Cerner’s McCallie, adding, “FHIR is still young, and may take a few years for a stable set of widely supported FHIR implementation standards to be available.”

    While athenahealth’s Park describes himself as “cautiously optimistic,” he notes “FHIR is far from a complete specification. Over time, if we’re successful as an industry, the FHIR APIs will have to evolve. By way of example, the Google Maps APIs are unrecognizable from version 1.0– as new requirements emerge, Google has continued to change their APIs in response to developer requirements. However, FHIR is a great starting point.”

    Barriers To Adoption

    I suspect there are several important barriers to widespread adoption.

    First, there’s the chicken and egg problem. If there were terrific applications – a FHIR-based application that patients loved and insisted on using – it might drive adoption. But who’s going to seriously invest in building an application when there’s uncertainty about which hospital systems will be ready and able to connect with it.

    While Chopra agrees on the need for compelling apps, he expects “they will come with payment reform” as providers search for effective apps to “help patients stay healthy.” Instead, says Chopra, “the real barrier has been concerns that the app developer will ever get access, or that they might get paid for whatever service they offer.”

    Chopra’s “ideal” is that “a dozen health systems (I don’t imagine small physician practices will be early adopters just yet) adopt the FHIR API and they invite at least some patients to test out connecting the API.” Success here, he believes will motivate others to join.

    A second – and perhaps even more worrisome – barrier to adoption is the persistent doubt that most EHR systems, and most medical centers, are truly committed to reducing barriers to data sharing – carefully crafted public comments to the contrary. The question is whether in their heart of hearts (or more accurately, in the office of the CFO), hospitals and EMR systems really want to share data.

    (In my view, most medical centers, at the senior executive level, really don’t want to make their data easily portable, which is why I was so happy to see Vice President Biden explicitly call out the need for improved data sharing as a core component of his cancer mission. A compelling recent article by Merrill Goozner emphatically highlights the need for improved data sharing as well.)

    As Joy L. Pritts, a former privacy officer at the Department of Health and Human Services, recently told the New York Times, “It may be contrary to the financial interests of health care providers to give patients broad access to their medical records. Once patients have that information, they can share it with competing health care providers.”

    Some experts, however, believe concerns about provider organizations impeding data sharing are overblown and misguided.  Rather, Halamka argues, the “barriers are lack of enabling infrastructure, data governance, uniform policies, appropriately constrained standards, and economic incentives. Focusing on information blocking is a distraction.”

    Park of athenahealth highlights the need for the active engagement of both vendors and providers, rather than the alternative, which might be characterized as “assent without belief.”  Explains Park,

    “Interoperability has always been a political/cultural problem, not a technical one, and without additional political support FHIR will die on the vine like everything before it. To be successful, it needs ongoing, active support from both vendors and health systems. We need both missile keys– just one or the other is not sufficient. This, in turn, will likely continue to require heavy advocacy from the government, since the government is one of the only actors who can get both parties to turn their missile keys at the same time.  Candidly, the political advocacy from government has helped far more than the ‘blessing’ of any particular technical specification… and it’s coming from all quarters these days….”

    Bottom Line

    It may be too early to say for sure whether FHIR will succeed or “die on the vine,” but I desperately want to be optimistic. I appreciate the way FHIR has evolved as a standard generated by the user community (vs imposed by regulators), and I’m impressed by the high degree of bipartisan commitment to interoperability, including not only the Obama administration but also high profile Republican leaders including Lamar Alexander (R-TN) and Bill Cassidy (R-LA).

    My sense is that FHIR may be the last, best hope for interoperability (perhaps reason enough for some stakeholders to secretly root for its demise).   While Yoda might respond “no – there is another,” I suspect the only obvious alternative – government-imposed regulation – is one force no stakeholder seeks to awaken.

Categories: Uncategorized

Tagged as:

3 replies »

  1. It’s all about control or privacy depending on whether you look at it from the institution or patient perspective.

    If the institutions and EHR vendors could avoid FHIR or any other secure web-accessible standard API, they would. A proprietary API puts each vendor in control and allows product differentiation as well as provider lock-in. Proprietary APIs also benefit the largest incumbent provider institutions when they dominate a market through patient lock-in. As Homer S. says: “It works on so many levels.” Proprietary APIs are just data blocking 2.0. But the problem for the vendors is that it’s hard to innovate around proprietary APIs and patient lock-in strategies. Sooner or later, the barbarians get better technology. The providers developing around patient lock-in also have an innovation problem and the cost of services outside the cartel decreases faster.

    The JASON experts saw this quite clearly years ago and called for a Public API. It’s a simple patient control, as in privacy, issue. The feds seem to understand it as well and the latest OCR guidance deals with data blocking as a patient rights issue, covered by HIPAA instead of MU. The OCR guidance stops short of asserting a patient right to the FHIR API but it does mention the MU 3 API.

    When HL7 sorts out the privacy aspects of FHIR, will the FHIR API be accessible for patient-directed connection to third-parties under HIPAA? Or will patients have a lower-grade patient-facing API under MU3? What do you think?

  2. I can’t help but think of the Fogg Behavior Model in every discussion of interoperability.

    Behavior model: http://www.behaviormodel.org/

    We can make the task simple (ala FHIR), but if we don’t change the motivation ($/ROI) we won’t get to long-term change. In that respect, funding health care value and true value-based incentive payments are healthcare’s best and perhaps only chance for long term interoperability.

    When it pays to share, when it pays to collaborate across systems to get patients the right decision as quickly as possible, sharing will happen consistently.

    Artificial incentives like MU, although well-intentioned and at least partially effective short term, are tough to maintain over the long term, as we’re already discovering.

  3. It might be that interoperability would lead to worse medicine.

    Suppose the patient is metaphor like a complex story, the Iliad. Suppose three readers are to evaluate the story and treat the patient based upon the wisdom accumulated from the reading exercise.
    In one case the three readers have their brains connected so that what one sees and hears and believes is communicated simultaneously to the other two. In another case all three readers are separate and isolated until a summary is collected at a meeting just before the patient is to be treated, by using some kind of voting system that assesses the wisdom of 3-crowds.

    Each reader, if independent, would imagine quite different stories. They would also accumulate more variable physical input (images, lab tests, specialty consultations) than the unified three. Perhaps their insights would have more variation and truth gathering scope than the three symbionts tied together in their ITs (read brains).

    I am not saying that I believe all this, only that it is possible. I do believe it is valuable for patients to be presented to clinical meetings–crowds–and to have repeated imaging and lab testing. Too much faith in what others are thinking and seeing might cause more errors and their associated costs than the savings are worth.