#FHIR and the Gartner Hype Cycle

Sep 4, 2018

On a forum for FHIR Foundation members, I raised the subject of where FHIR is on the Gartner Hype Cycle (see Gartner write up, or Wikipedia). FHIR Foundation member Wes Rishel (@wrishel), who’s a FHIR user, and also was a Gartner Analyst before he retired, graciously made this contribution that I could post here.

From 2000-2014 I was the person positioning health IT standards on Gartner’s Hype Cycle. In the last few years I created and positioned a spot for FHIR. Given my enthusiasm for FHIR it was always a struggle to maintain my professionalism. The professional side was assisted by peer review and a methodology that had evolved tracking tens of thousands of technologies for many years. If I were still evaluating FHIR for the Gartner hype cycle I would be looking for the following factors to move it to the slope of enlightenment.

  • One or more well-defined, application-specific market segments.
  • Full adoption by a small portion of the market.

  • Predictability in assessing project risk and timing, based on experience with the technology.

  • Reasonable labor pool of experienced developers.
  • Conformance testing.
  • The development of at least an informal body of knowledge of what it takes to really interop with certain important FHIR servers or certain important EHR clients.
  • Having been through the “trough of disillusionment”

One or more well-defined market segments.

There are many markets and a wide variety of uses of FHIR, but the usage pattern that seems to have the widest adoption is “sidecar applications for EHRs”. Many of the SMART applications would fall into that category. Are there other segments that can be defined by usage patterns of participants in the marketplace? For example, are analytics products switching to FHIR rather than ad hoc interfaces? I acknowledge I’m out of the loop but I’m not aware of much that could be defined as having a common pool of vendors and purchasers (or in FHIR terms a common pool of server-apps and client-apps).

Full adoption by a small portion of the market.

As shown in the Wikipedia article that Grant cited, the first solid sign that a technology has turned the corner is ful” adoption by 5% of the market. “Full adoption” does not mean replacing an older technology. FHIR could be at the top of the slope while V2 messages still flow. I think I would have found this crieterion to be met if I could see that some group of healthcare organizations that care for at least 5% of the patients in industrialized countries were using FHIR for multiple independently developed apps in one of the market segments in mission-critical apps. I am skeptical about this. As an indirect measure of widespread adoption for the sidecar market, I would be looking at how far the major EHR vendors had progressed at rolling out FHIR support. It seems to me that the biggest two vendors have active programs promoting FHIR adoption including standard business relationships with third-party vendors. That’s reasonable progress. The last I heard, though, the data available through the FHIR interfaces was read-only and limited to registration data. Third-parties were being encouraged to use non-FHIR proprietary interfaces for update and access to more data types.

Predictability in assessing project risk and timing, based on experience with the technology.

FHIR is easy to start to use and to get working in the lab or a Connectathon. If I were looking at project risk it would be towards the end of a project where testing can’t begin or is delayed because of one of the partner’s schedules or where unexpected semantic issues arise. These unfortunate events can happen in any project and, in theory, are less likely in RESTful interfaces. However, as products climb the slope of enlightenment they get less frequent. Are we there yet?

Reasonable labor pool of experienced developers.

This is one major source of technology risk. I would be looking at FHIR books, courses in school curricula including not only Universities but high-volume purveyors of video tutorials. I may be wrong but it seems like a limiting factor in FHIR education is that it takes a time of the most productive developers.

Conformance testing and certification.

A sign of a maturing interface technology is when the vendors that use the technology recognize that they need to make an investment in funding a common testing lab and slog through the hard technical and political issues of what it even means. FHIR has been developed from the start with import provisions for making and testing conformance and there are more than one places to operate such tests. There is a lot of potential. But it doesn’t rise to the level of a service mark that would give the consumer of Health IT products the assurance that if they buy two certified interoperable products, the product will interoperate. AFAIK we’ve never aspired to that level of conformance testing or certification in healthcare. It took Bluetooth and USB technologies years to appproach that level. Having having independent, industry-wide certification is a definite mark of a maturing technology. One can argue that this is not an important criterion for being on the slope. HL7 v2 and CDA clearly made it to the slope without this. I agree, but I would argue that the demonstrated success of conformance testing would be a substantial argument for moving FHIR more rapidly to the slope.

The development of at least an informal body of knowledge of what it takes to really interop with certain important FHIR servers or certain important EHR clients.

This is a definite sign of the maturity of the technology. It means that management and system integrators can assess project risk on a much more nuanced basis. 

Having been through the “trough of disillusionment”.

If FHIR has gone through the trough it’s one of the shallowest troughs on record. Where are the click-bait postings on problems? Where are the failures of early startup vendors and the requirement for advanced rounds of capital?

The closest I’ve heard is some grumbling among sidecar-app vendors that they thought using FHIR would make it easy to sell. They describe problems getting big company CIOs to take their calls. One of the signs or a technology on the slops is a community of vendors that understand not only the technology but the practical issues of selling and delivery.

So, as much as I would like to be at Gartner positioning FHIR on the slope I would almost surely still place it pre-trough. 

Thanks for the contribution, Wes. A few clarifications from me:

  • Wes asks “Are there other segments that can be defined by usage patterns of participants in the marketplace? For example, are analytics products switching to FHIR rather than ad hoc interfaces?” - my answer is, yes, that’s happening, but it’s not yet at the point where you could call it ‘a market’
  • In general, there is lots of FHIR adoption, in lots of different ways, but they don’t constitute a identifiable market in the sense that Wes is talking about here. 
  • I’ve always been aiming for a shallow trough: if the fundamentals are in place, it won’t be so deep. But we still have fundamentals to get in place
  • I think, at least, that we are no longer in the rocket-mode growth phase of the very earliest part of the hype cycle.