Stream: fmg
Topic: For discussion (and hopefully approval) prior to Wed call
Lloyd McKenzie (Jul 16 2021 at 22:09):
Hi everyone,
For those who monitor the #terminology stream, you may be aware of this issue already. For those who don't, quick summary: The HTA's policy for the issuing of URI's does not currently take into account whether a URI has already been assigned and mentioned in the FHIR specification or is otherwise in use in the FHIR community. More importantly, they do not feel the policy should take this into account and that the URIs used to reference code systems should be determined solely by the preference of the terminology author and should change, as necessary, according to the desires of the terminology source. Obviously this is contrary to our intended behavior and has serious consequences for implementers. However, discussion with the HTA chair, including bringing in Wayne has not resulted in a change to policy. As a result, there are HTA published URIs that diverge from this FHIR-published URIs for the same code systems and the prospect of more in the future, as well as the prospect of HTA-published URIs changing over time.
I've written up a detailed letter that describes the issue, the background and the criticality of the HTA not issuing conflicting URIs, as well as removing those they've already posted. The intention is to pass this to the HTA prior to their Wed. call - which happens the hour before our call. If we can't come to agreement on that call, the follow-up plan is to escalate the issue to the TSC.
Austin, Wayne and David have already had a chance to review and comment and I've made a few changes as a result. I'd now like all of the other members of the FMG to review the document and - if you're in agreement - to endorse it. Feedback on content also welcome. Obviously, I'd appreciate review sooner than later so that we can get it to the HTA in time to consider prior to their Wed. call.
https://docs.google.com/document/d/1_QJ4tc4qufQVGsvIWruCjIMfxT87ABrHwWVWzG8AwO4/edit
Hans Buitendijk (Jul 16 2021 at 22:42):
I included two minor edits and one suggested word change, but no concerns. I support the letter.
Lloyd McKenzie (Jul 17 2021 at 02:06):
Thanks Hans - applied
Brian Postlethwaite (Jul 17 2021 at 05:47):
I have read and support it.
Lloyd McKenzie (Jul 17 2021 at 17:03):
@Brian Pech @Grahame Grieve @John Moehrke @Josh Mandel
Josh Mandel (Jul 17 2021 at 17:44):
I also support the recommendation not to second guess established URIs; I don't have all the history on this so the tone of the letter seems surprisingly combative, but I certainly support the conclusion.
Lloyd McKenzie (Jul 18 2021 at 16:05):
@Paul Knapp - did my response make sense?
John Moehrke (Jul 19 2021 at 12:31):
This blindsided me. I have no background to understand this conflict. I have a tendency to not change canonical URI once they are applied. Many of you have seen my anguish with IHE having good reason to change (i.e. formatCode vocabulary).
As I read this, it seems strange to me. We have been willing to break many systems that are using happily OIDs, and have happily changed over to terminology.hl7.org root... we have broken many systems. so this seems really conflicting to be complaining now, simply because HTA is defining a rule that does similar (lesser) breaking?
Lloyd McKenzie (Jul 19 2021 at 12:47):
Changing from OIDs to URIs happens at a translation point - you have to convert and translate all sorts of things moving between CDA and FHIR (element structures, codes, etc.) Translating OIDs to URIs is just one more, and is something that was carefully considered in terms of the cost and benefit.
Changing to terminology.hl7.org was not something "happily" done. It came about after considerable discussion, including consulting with the implementer community. It was also a coordinated change that was dependent on FHIR version - and thus something that could also be managed at a layer that involves translation. The URI used in STU3 is one thing, the URI used in R4 is another. Also, when we did it, we promised it would never happen again.
What the HTA is saying is that, regardless of what release of FHIR you're using (and irrespective of what that release of FHIR might declare), as of a few months ago, the URI used for ATC codes is now different and using the wrong one is non-conformant. There's no translation point there. It's just a random breaking change done without discussing with the community or regard for implementer impact.
John Moehrke (Jul 19 2021 at 15:58):
ATC?
Lloyd McKenzie (Jul 19 2021 at 15:58):
A WHO drug terminology
Lloyd McKenzie (Jul 19 2021 at 15:59):
Which is the most recent one that's surfaced as an issue. However, the more general issue is the policy. From a policy perspective, the URL that changes could as easily be LOINC or SNOMED
John Moehrke (Jul 19 2021 at 16:00):
okay, well then the issue I had with IHE FormatCode might fall into that?
Lloyd McKenzie (Jul 19 2021 at 16:00):
I don't recall what that issue was?
John Moehrke (Jul 19 2021 at 16:01):
In that case IHE changed the URL that FHIR picked.. but coordinated because the FHIR picked one was just ugly
Lloyd McKenzie (Jul 19 2021 at 16:04):
If the change is coordinated with the implementer community and reflects implementer desire, that's fine. The issues here are:
- not checking to see if there's already an existing URL
- not striving to avoid change to the existing URL
- not checking with the community to see whether they'd prefer to stay with the "in use" URL as opposed to moving
John Moehrke (Jul 19 2021 at 16:09):
agreed
Brian Pech (Jul 19 2021 at 20:05):
I agree with the intent of the letter.
Lloyd McKenzie (Jul 19 2021 at 21:48):
@John Moehrke I've added a paragraph (highlighted blue) that I think covers the situation you'd highlighted. With that change, do you endorse the letter?
Lloyd McKenzie (Jul 19 2021 at 21:49):
@Paul Knapp - are you comfortable endorsing the letter as it stands or are there changes you'd like to see?
Last updated: Apr 12 2022 at 19:14 UTC
