FHIR HL7v3 comparison

I think there are different ways of evaluating "failure". Certainly, it's proven possible (with sufficient time and investment) to implement version 3 and get at least some degree of interoperability out of it. So from that perspective, it can be seen as successful.

However, there are other areas where it's hard to argue that it's a success:
- market uptake has been much more limited than HL7 had hoped to see. Aside from CDA, v3 has pretty much been constrained to large government-sponsored projects with both a lot of money, a lot of time and a lot of clout.
- interoperability has been challenging. Canada, the UK and the Netherlands all implemented v3-based e-prescribing systems, but it would require a significant amount of transformation to get to interoperability between them. To a lesser extent, we have similar interoperability barriers between regions in Canada. Some of that is driven by different decisions, but the fact that introducing a new element breaks wire compatibility unnecessarily compounds the problem
- from an implementer mindshare perspective, it's pretty hard to find implementers who are enthusiastic about the ease, speed or cost of interoperability with v3 (or CDA). Those who do wax enthusiastic tend to be those who are v3 consultants. (I can say that - I've done more than my fair share of it :>)

I actually like v3 from a modeling perspective. It's a very powerful modeling tool. But I've come to conclude that the methodology and the standards that come out of it aren't producing the necessary results for a reasonable amount of investment.

I'm not suggesting that FHIR is a panacea. And FHIR is still evolving. But it does address some of the problems with v3 - learning curve, overly complex instances, wire format differences between implementers and lack of extensibility. More importantly, it's gaining significant traction with implementers in environments where there's no-one pushing and saying "thou shalt", but simply people stumbling across it on the web and saying "yeah, I can do that". That's not something that ever occurred with v3 and occurred only rarely with CDA.

Given the investment we have in v3 in Canada and the fact that FHIR isn't normative yet, it's premature to talk about migrating anything that exists, but it probably does make sense for us to start exploring using it for new work and verifying that it will meet our needs when/if we do decide to migrate. (I think FHIR is the only way we'll ever see significant interoperability across jurisdictional boundaries.)

Disclosures:
- I've created numerous v3 models and implementation guides, taught v3 all over the world and assisted several v3 implementation projects. I've made a pretty good living as a v3 consultant
- I now spend way too much of my volunteer time (and my employer's time) working on FHIR, which is making me a bit of a living as a consultant now, but I'm not expecting to be earning a premium for too long. Implementers just don't need anywhere near as much support with FHIR as they did with v3. On the other hand, there's lots of consulting work around creating good quality implementation guides, facilitating consensus among stakeholders (interoperability is, after all, about getting people to agree on what to share), terminology, etc.

Oh, and the views above are mine, not necessarily my employers or those of my customers :>

I think that FHIR may be a capable "v4," but I disagree that v3 is a "failure." Projects may fail, but, from my experience working with v2, v3 and other standards, it is not because of anything inherently wrong with or complicated about the standards employed. I have also witnessed many organizations specify non-standard, proprietary and v2 solutions for reasons completely unrelated to technology, with the natural consequences.

No doubt, others have more experience than do I and other opinions.

P. S. If these new forums are to work, I suggest that people voluntarily reveal their interests (not that I am suggesting anything untoward in your post, Ken). We formally declared our conflicts of interest at the beginning of every SCWG meeting and obviously noted our affiliations in the minutes so as to fully inform participants and future reviewers.

thanks Ken. Its a very informative paper. I'm interested to see what other members of this group have to say about this topic.

Intelliware's Chief Technologist, BC Holmes, recently published a whitepaper comparing FHIR to HL7v3. You can read it here:

www.intelliware.com/hl7-games-catching-fhir/