Stream: implementers
Topic: FHIR and the "Gartner Hype Cycle"?
John Silva (Jun 22 2018 at 00:57):
I just attended the FHIR DevDays in Boston and was very excited about all the great work going on around FHIR and related "derivatives". I started to write up a "trip report" and I did an "innocent Google search" looking for article that talk about where FHIR might be on the Gartner Hype Cycle" curve; after one of the keynote talks it seems like it is past the "trough of disillusionment" and maybe starting on the "slope of enlightenment". Much to my surprise one of the first articles I found was from a 'FHIR dissenter"; argh, how do we answer his arguments, especially when we are talking about (or talking up) FHIR to build the business (or technical) case for embracing it? The article is titled "Straight Talk about FHIR" by Dave Levin M.D. -- https://www.healthcare-informatics.com/blogs/david-levin/interoperability/straight-talk-about-fhir
His opinions aside, where do people feel that FHIR is on the Gartner Hype Cycle? - https://www.healthcare.digital/single-post/2018/02/20/The-Digital-Health-Hype-Cycle-2018
Bryn Rhodes (Jun 22 2018 at 04:52):
I think the analysis in that article is wrong on at least three points. First, FHIR definitely is a standard, it's an STU, he seems to want it to be normative now, I think we'd all like that, but that's not where it is. I think it's wrong to characterize it as a "suggestion" though, it's a framework that allows you to be as permissive or as constrained as you need to be. Will it solve all interoperability out of the box? Of course not, that's a completely unrealistic expectation; implementers still have to do the hard work of specifying to get real interoperability. Second, I think the comparison between vendors is a little unfair at this point, they are at different levels of maturity. Yes there is variability, and that can be frustrating, but remember the "T" in STU. And lastly, saying that extensions undermine interoperability in the same way that a Z segment does is missing the point. Extensions give you a way to formally negotiate content that is outside of the standard, and they do so in a way that enables a broad range of additional functionality without compromising the integrity of the base standard. To compare that to Z segments is, I think, to misunderstand the problem space.
René Spronk (Jun 22 2018 at 08:47):
I think his assessments (first half of the blog post) are largely correct. Personally I don't follow how he reaches the conclusion "Healthcare needs a mix of legacy technology, FHIR APIs, EHR vendor APIs and commercial APIs." - a standardized framework (or: suggestion, to use his words) is better than nothing as a starting point. I'm sure we'll see a mixture of APIs, but that is certainly not a must to achieve interoperability.
As for the question where FHIR is in the hypecycle: In my book FHIR is "near - peak", but whether it's still on its way up, or going down is not clear to me. As Grahame has explained in the past: once a thing is hyped, pundits (or people claiming to be pundits) will gain attention and clicks if they start posting blogs that cover FHIR in a negative way. As such this is not something to worry about, it's a sign of the increasing maturity and use of FHIR.
Charlie McCay (Jun 22 2018 at 09:16):
The underlying claim that interoperability is not just a technical issue - and will not just be "solved" by signing up to using FHIR (or any other API framework) is valid. The article should be titled "Straight Talk about Interoperability" ... but then it would not get the clicks... fwiw it also does not mention the importance of sharing processes and practices, and a wider framework of expectations and responsibilities for interoperability to be useful.... thus I would say that he has missed the main reason why FHIR on its own will not "provide a rapid path to robust interoperability".
I agree with Rene that having a standard framework available, and a mixture of other interfaces in the wild is not incompatible - and it appears that the market is converging on using FHIR for many things... I would also agree that the trough of disillusionment is still to come... but I expect that there will be some pragmatic bridges to help maintain productivity en route to the plateau
John Silva (Jun 22 2018 at 12:39):
I think you each raise valid points about what he says in his article. I suspect the author is trying to get 'clicks' and 'stir controversy' though some of his points are hard to refute. Any standard, at any level of maturity, has some level of uptake, some do not or can not use it for valid business or technical reasons, but that doesn't mean it's not a useful standard. Look at how the TCP/IP stack evolved over time; in the 'early days' folks were discrediting it because the ISO network stack was "the pure model" of how networking should be. Of course we all know what happened there -- too much over-engineering and slow uptake so the TCP/IP stack 'slipped in' to meet the real world needs because it was 'right sized'. It feels very similar with FHIR; the V3 models were 'purist models' but didn't get much traction so the practicality and ease of 'getting onboard' with FHIR has given rise to its wide use in healthcare while V3 seems to be left 'in academia' with limited uptake. (yes, a little exaggeration)
His analysis about EHR vendors being at different level of maturity and 'buy in' is to be expected. Each vendor needs to weigh the value and business use cases for moving to a standards-based API over their own API, where they can control things much better. There's also the 'existing customer base' that has to be maintained and moved forward in a systematic way; it's not all 'green field' implementations. I suspect over time though that the EHR vendors will incrementally add FHIR to their stacks as the customer demand drives it. It does seem like there is quite a bit of involvement from the EHR vendors working with the FHIR community to move things forward. As FHIR starts to move towards the 'slope of enlightenment' the EHR vendors will almost surely have a FHIR capability enabled in their products; the question is mostly about the timeline (and maybe urgency). I suppose this is where governments can help, not by heavy-handed legislation but by their 'buying power' in influencing decisions about technologies and standards for government contracts and grant projects (like the NIH grants in the US).
Lloyd McKenzie (Jun 22 2018 at 13:33):
FHIR isn't a panecea. It doesn't guarantee interoperability. What it does is reduce the technical barriers to interoperability where there's agreement on how systems should interoperate. It doesn't, however, magically cause all countries to start using the same terminologies, cause clinicians to agree on what information should be in a particular type of referral, or even provide reliable identifiers to all stakeholders. FHIR makes is more likely that more and better data will flow between systems that aren't operating in a negotiated environment, but if you want all data to flow and be understood, you'll still need a negotiated environment. FHIR makes it easier to establish that negotiated environment though. FHIR also enables architectural approaches to sharing that intrinsicly need less negotiation and opens up new modes for sharing. I think that's what really is driving the greatest excitement. For practical, technical reasons, technologies like SMART and CDS Hooks would have been very hard to execute under older standards like v2 or CDA (or SDTM or X12 or other legacy standards)
Grahame Grieve (Jun 22 2018 at 16:35):
I thought that article was amazing. I also agreed with the first half, but then, when he's about to offer a solution he says.. the solution to the fact that FHIR doesn't standardise enough is to.... standardise less....
Abbie Watson (Jun 22 2018 at 22:38):
Punditry. He’s basically making a bet against internet standards. All his arguments apply to OAuth implementations among Facebook, Google, GitHub and other websites... yet, billions of people use hundreds of thousands of services that rely on those protocols to handle them everyday. As a medical doctor, he has a knowledge gap around proxy servers, content delivery networks, application routers, and other services that auto-negotiate differences between API specs. Case in point: Zapier.
John Meredith (Jun 24 2018 at 20:40):
Without being embedded in the US healthcare market, I do find the article very insightful, echoing many of my misgivings about how FHIR is being hyped out of its core context. I don’t agree on the market forces stuff, nor the diversification. He’s missed the point considerably by not drawing out the underlying information model that should drive the requirement for interoperability. And this is where the issue with extensions should be raised. I know why extensions exist, and I know how they should be used. But I also have concerns with the total unchecked flexibility of them and that they could in turn become a barrier if not implemented in a considerate way.
The clinical model is king, and FHIR can provide a representation of that for specific use cases. I think part of the problem is in the use of the term interoperability. It just does not cut it to define the breadth of an electronic patient record encomassing the process and semantic aspects as well as technical interop. Now as a Doctor, the author would implicitly know all of the former aspect but needs to articulate it better. I don’t think it is a click-baityas some of you suggest though.
Grahame Grieve (Jun 25 2018 at 10:24):
the head line was click-baity. and the conclusion was weak. Of course, things would be better if everyone would just abandon their existing records and we could have a clean information model. We've got plenty of those to chose from ;-)
Grahame Grieve (Jun 25 2018 at 10:25):
as for extensions.... the question isn't 'will they become a barrier' but 'did they act to introduce a barrier that wasn't inherent in the actual task'
John George (Jun 29 2018 at 10:47):
@John Silva please correct me if I am wrong but I was under the impression that the Gartner Hype Cycle, was not scientifically/mathematically proven. See: <https://ieeexplore.ieee.org/document/5603442/>
Certainly, that was the way it was described on a MSc Health Informatics course I was studying a few years ago. Therefore I pay little attention to it, as I believe technologies and external influences like financial and political influences make each case unique, and whilst it's interesting to look for common patterns between the take up of different technologies, it could be dangerous to think that all follow the same pattern.
Grahame Grieve (Jun 29 2018 at 11:08):
indeed it is not guaranteed - but like Kuhn's Cycle of Scientific development, you ignore the underlying sociological factors at your peril
John Silva (Jun 29 2018 at 13:15):
I suppose it's not scientifically/mathematically proven but organizations like Gartner (and marketeers who use these to tout their 'wares') have a lot of influence on the perception of the marketplace and then the purchasing decisions made based on them.
Last updated: Apr 12 2022 at 19:14 UTC
