Stream: terminology
Topic: CodeSystem Resources
Mark Kramer (Nov 18 2021 at 14:00):
Are the CodeSystem resources in the IG package hl7.terminology#2.1.0 valid for use in other IGs? I want to use the CodeSystem resources in the terminology package as a dependency for my IG. However, there are a number of conflicts between the published CodeSystem resources and the confluence pages maintained by HTA. A summary of those differences as they affect my IG are listed here. This is confusing.
Lloyd McKenzie (Nov 18 2021 at 14:21):
@Carol Macumber
Grahame Grieve (Nov 18 2021 at 18:30):
they are supposed to be valid. And you don't even need to explicitly reference hl7.terminology - it's always in scope
Grahame Grieve (Nov 18 2021 at 18:32):
I think that HTA is wrong about HGVS, but I don't know for NCIT - where did you get http://ncithesaurus-stage.nci.nih.gov from?
Mark Kramer (Nov 18 2021 at 20:57):
It is the url in the published CodeSystem resource named NciThesaurus: https://terminology.hl7.org/CodeSystem-v3-nciThesaurus.html
Mark Kramer (Nov 18 2021 at 21:00):
That was part of the FMG discussion yesterday: do the CodeSystems published in FHIR take precedence over those indexed by HTA on their Confluence pages which are under development and being updated on a regular basis. I went with the former and I think that is correct stance.
Robert McClure (Nov 20 2021 at 01:41):
@Carol Macumber Here we go again with NCIT
Grahame Grieve (Nov 20 2021 at 02:57):
this is not quite the same as NCIT - I checked before it was raised here
Patrick Werner (Nov 20 2021 at 10:53):
NCIT was set to http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl by the HTA
Patrick Werner (Nov 20 2021 at 10:54):
now it is published in the fhir package differently. Sorry this is messy and hard to follow.
Patrick Werner (Nov 20 2021 at 10:54):
I recently switched multiple IGs to http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl (not liking the url but this was announced to be the correct url by HTA multiple times)
Patrick Werner (Nov 20 2021 at 10:57):
@Mark Kramer what is you position on HGNC? In the genomics IG we started using: http://www.genenames.org/geneId and http://www.genenames.org/genegroup after discussing the issue with HUGO.
GeneIDs and GeneGroup IDs aren't guaranteed to be distinct.
Mark Kramer (Nov 20 2021 at 11:52):
@Robert McClure ThIs it isn’t about NCIT, it is about discrepancies between what is published in FHIR and what HTA has on its Confluence page. When FHIR updates, it becomes official, but not until then, IMO.
Mark Kramer (Nov 20 2021 at 11:57):
@Patrick Werner I don’t have an opinion on HGNC. At least FHIR and HTA agree on that one.
Alexander Zautke (Nov 20 2021 at 12:18):
Is there at least agreement to update the canonicals in the FHIR package wherever there is a mismatch with the HTA canonical? At least then projects could have the possibility to pre-adopt the change. I don’t follow why there can’t be a pragmatic solution here.
Grahame Grieve (Nov 20 2021 at 20:36):
umm. we couldn't even talk about that unless HTA raises the issue. But I can't issue new packages for past specifications without going through a technical correction. Having said that, where is this:
now it is published in the fhir package differently
?
Mark Kramer (Nov 20 2021 at 21:54):
I think Patrick was just reiterating the point that the FHIR package has http://ncithesaurus-stage.nci.nih.gov/ for NCI Thesaurus.
Robert McClure (Nov 20 2021 at 23:35):
@Grahame Grieve Seems the solution is obvious, FHIR should not publish without confirming that all the HTA canonicals are what is used in the FHIR build.
Grahame Grieve (Nov 21 2021 at 09:34):
@Mark Kramer, Patrick generally uses his language very carefully. I'm not aware of where http://ncithesaurus-stage.nci.nih.gov/ appears in a base FHIR Package anywhere, and I need to review if it does
Grahame Grieve (Nov 21 2021 at 09:41):
@Robert McClure I strongly believe that HTA should not decide and publish FHIR canonical URLs without first consulting with the FHIR community to see what's already happening.
So it makes sense that we won't add external canonicals to FHIR core without first checking with HTA, and won't add in disagreement with HTA. And that's our policy now. (though mainly our policy is that we won't add external canonicals at all now, except in examples)
however most of the things we are talking about were added to FHIR a long time before HTA got interested. Sure, it's totally reasonable that we do a review against HTA before publishing. But as far as I know, that would be on an entirely manual basis right now, so that's a kind of once off before publishing. That's bad timing - too late to do much about disagreements.
For that reason, it would be good for HTA to check and raise issues proactively
Robert McClure (Nov 22 2021 at 01:10):
We'll add that to the list of to-do's for our staff
Patrick Werner (Nov 22 2021 at 08:15):
Hi @Grahame Grieve i meant the term package 2.1.0: https://terminology.hl7.org/CodeSystem-v3-nciThesaurus.html
NCIT canonical here: http://ncithesaurus-stage.nci.nih.gov
NCIT canonical on the HTA page: http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl
Grahame Grieve (Nov 22 2021 at 08:17):
ok thanks. HTA - makes sense. @Robert McClure your move :grinning:
John Moehrke (Nov 22 2021 at 13:51):
OIDS are so much more clear
Carol Macumber (Nov 22 2021 at 14:19):
@Davera Gabriel copied in as she is the HTA member who worked on NCIT.
@Reuben Daniels as he has done a comparison as part of HTA in preparation for the THO work to establish it as authoritative source
With regards to HTA registering new URI's... @Grahame Grieve , agreed. The policy is and has been to announce them to the community and seek feedback. I believe there was a thread and/or ticket for NCIT.
With regards to the order of precedence. The joint HTA, FMG, Vocab task force working on the policy for requesting and resolving issues with external code system URIs is getting ever closer to a set of guiding principles (green have been approved by the group, https://confluence.hl7.org/display/TA/2021-11-08+Meeting+Notes) and a resulting policy and process (https://confluence.hl7.org/display/TA/Validating+and+Requesting+Code+System+Identifiers+for+an+External+Code+System)
Lloyd McKenzie (Nov 22 2021 at 14:26):
I would think that before raising a thread to discuss with the community, the first step would be to manually check to see if a URL was present in the FHIR spec and, if so, to just go with that...
Patrick Werner (Nov 22 2021 at 16:34):
John Moehrke said:
OIDS are so much more clear
how?
Mark Kramer (Nov 22 2021 at 16:40):
I think that was a joke. :laughing: BTW, we're covering the same or similar ground as a previous thread. It seems the outcome of that thread was that previous implementations have to take precedence over any renaming that comes from the code system owner.
John Moehrke (Nov 22 2021 at 16:45):
mostly joking.. but OIDs must only be seen as URI, and can't be misunderstood by humans that are reading the human readable text of the URL and guessing...
John Moehrke (Nov 22 2021 at 16:45):
so an OID is clearly something that must be looked up.
Alexander Zautke (Nov 22 2021 at 17:11):
Lloyd McKenzie said:
I would think that before raising a thread to discuss with the community, the first step would be to manually check to see if a URL was present in the FHIR spec and, if so, to just go with that...
For NCIT this is now too late, right? The new canonical has been discussed with the code system owners. All specifications within HL7 Germany and the ones that @Patrick Werner maintain have switched to the new canonical. @Mark Kramer If your IG would also use and recommend the new canonical, I think we could move towards a solution. At least for one part of the issues presented here.
Mark Kramer (Nov 22 2021 at 17:56):
I must do what FMG and TSC recommend for this situation. They need to sign off on the IG publication.
Lloyd McKenzie (Nov 22 2021 at 18:08):
The fact it's been discussed with the code system owners doesn't much matter if it's already been in use in FHIR implementations with the published one. According to HTA process, they're required to use the existing one in the event of a conflict between the desires of the code system owner and what's in use by the community. The only question is - is the current one in use by the community?
Alexander Zautke (Nov 22 2021 at 19:09):
Would anyone be aware of a published IG using the canonical? It's in the core spec, so it's kind of hard to estimate where it's really being used.
Alexander Zautke (Nov 22 2021 at 19:12):
That brings back the question if it's possible to update the canonical in the core build (so future FHIR versions are aligned with the HTA canonical) and all projects interested currently in this discussion agree to use the HTA canonical?
Mark Kramer (Nov 22 2021 at 19:17):
It's the OID (2.16.840.1.113883.3.26.1.1) that seems to be in use. Maybe the solution is to sidestep the issue of canonical URL, and use the OID since everyone agrees on it. Well done, @John Moehrke
Alexander Zautke (Nov 22 2021 at 19:22):
And it doesn't contain the .owl part :sweat_smile: Sorry...
Lloyd McKenzie (Nov 22 2021 at 19:23):
The intention eventually is to drive all of this stuff off what appears in terminology.hl7.org and have a process that ensures that once a URL appears there, it won't change without broad implementer consultation and agreement. (Exception is where we've done something clearly erroneous like registering to URLs for the same code system or assigning a URL to something that isn't actually a single namespace.) However, that process (and the priority of stability above all) wasn't clear when HTA first started their work, so we have some initial clean-up we need to work through.
Mark Kramer (Nov 22 2021 at 19:28):
So, what is the bottom line, @Lloyd McKenzie? When mCODE goes to FMG again, would you personally support:
1) The OID everyone agrees on: 2.16.840.1.113883.3.26.1.1
2) NCIT canonical in FHIR: http://ncithesaurus-stage.nci.nih.gov
3) NCIT canonical on the HTA page: http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl
John Moehrke (Nov 22 2021 at 20:00):
canonical URI given the OID - urn:oid:2.16.840.1.113883.3.26.1.1
Lloyd McKenzie (Nov 22 2021 at 20:13):
I would support #2. The OID should definitely not be used as it shouldn't have been used in FHIR at all, and we want to minimize as much as possible the number of places OIDs take hold in FHIR. (They're less-implementer friendly for a number of reasons.) If you're going to ballot with mCode, we should highlight the open issue. If you're going to final publication, then we need a resolution between HTA and the FMG on this particular issue before publication occurs.
Robert McClure (Nov 23 2021 at 00:06):
@Lloyd McKenzie I think you get to be the dog HTA brings to all the code system owner dog-fights.
Jean Duteau (Nov 23 2021 at 00:12):
I've been using NCIT terminologies for quite a while and I've been using #3 for all of them because that was a) on the HTA webpage, b) nothing was in the FHIR Terminologies page and c) that was what NCI said they wanted.
Lloyd McKenzie (Nov 23 2021 at 00:35):
Ok. Having looked for myself:
- I can find a reference for the NCI meta-thesaurus in the FHIR core spec, but not the NCI thesaurus. I'm assuming these are two different things?
- There is a URL for the thesaurus in http://terminology.hl7.org. It doesn't in any way indicate that it is 'temporary'
- Given that we've agreed that THO is our source of truth and there's no existing entry in the FHIR core spec, then that's what we should be using (as I understand agreed process)
- The HTA page is (or should be) a temporary holding ground before net new things get migrated into THO (after first checking that they don't already exist in THO or the FHIR core spec)
- Unless a project has looked in the FHIR spec and THO and hasn't found what they're looking for, they shouldn't be looking at the HTA pages. HTA pages are primarily for people who are closely monitoring the HTA pages
- Given that the data is already in the HTA Jira trackers, I'm not sure why it needs to exist. If we close the HTA Jira tracker only when the UTG proposal is applied, then there should be no need for HTA to maintain a list at all.
That's my understanding/opinion.
Woof. ;)
Carol Macumber (Nov 23 2021 at 01:04):
Lloyd McKenzie said:
I would think that before raising a thread to discuss with the community, the first step would be to manually check to see if a URL was present in the FHIR spec and, if so, to just go with that...
yes, it is expected that the requestor check the FHIR spec and THO prior to requesting a new one. HTA members fielding requests are also asked to do the same. Once that leg work is done, which is more than just the URL for new external code systems, the HTA process includes posting to Zulip just to make sure the code system wasn't already used unbeknownst to the requestor or HTA.
Grahame Grieve (Nov 23 2021 at 02:05):
I can find a reference for the NCI meta-thesaurus in the FHIR core spec, but not the NCI thesaurus. I'm assuming these are two different things?
yes, different things. The nci-meta part is being deleted in embarrassment (I'm the embarrassed one)
Grahame Grieve (Nov 23 2021 at 02:06):
I must do what FMG and TSC recommend for this situation. They need to sign off on the IG publication
As an FMG member, I have no opinion on this - it's an internal TSMG issue - they own both pages in disagreement with each other, and how can I say anything other than 'they need to say'?
Grahame Grieve (Nov 23 2021 at 02:07):
but if I did vote, I'd vote with the HTA page (this time) on the grounds that (a) we've made a mess and someone is going to be wrong whatever, (b) nothing about this code system suggests to me it's widely adopted, unlike some others we've considered and (c) the code system owner wants what's on the HTA page
Patrick Werner (Nov 23 2021 at 12:04):
Jean Duteau said:
I've been using NCIT terminologies for quite a while and I've been using #3 for all of them because that was a) on the HTA webpage, b) nothing was in the FHIR Terminologies page and c) that was what NCI said they wanted.
same for me and my projects
Patrick Werner (Nov 23 2021 at 14:52):
one more comment:
My projects which are using NCIT are quite new - so a change wouldn't be that hard. Only thing that is important to me: That we get consensus on the correct url soon.
Changing it in a few month will be much harder than changing it now.
Diane (Nov 30 2021 at 19:35):
Not trying to hijack this thread, but it appears that the discussion on this topic never reached a resolution and now I am another person needing clarity.
For our Adverse Event Questionnaire, the FDA requested during the balloting process that we change some of the code systems. After a call last week, the FDA directed us to use (for our Adverse Event questionnaire), code systems from the NCI Thesaurus.
We are in the midst of updating our code systems so that our IG can be published. My question concerns the url to use in the answerOption.valueCoding.system property when we are using the NCI Thesaurus code systems such as
https://ncit.nci.nih.gov/ncitbrowser/ajax?action=create_src_vs_tree&vsd_uri=http://evs.nci.nih.gov/valueset/CDISC/C66769
A. Do we use the url that leads directly to the code system that we are using (eg. https://ncit.nci.nih.gov/ncitbrowser/ajax?action=create_src_vs_tree&vsd_uri=http://evs.nci.nih.gov/valueset/CDISC/C66769)? Does this meet FHIR standards for being a permanent URL?
B. Or do we use the high level https://ncithesaurus-stage.nci.nih.gov/ncitbrowser/ and will this url be acceptable for FHIR standards in light of the discussion on this thread?
C. Why use the staging url instead of https://ncit.nci.nih.gov/ncitbrowser/?
Grahame Grieve (Nov 30 2021 at 20:01):
A. No - very much not appropriate. It's a NCI description of a CDISC value set
B. no. that's a registry of code systems
C. because that's also a registry of code systems.
Grahame Grieve (Nov 30 2021 at 20:01):
the correct URL will be related to https://evs.nci.nih.gov/ftp1/CDISC/SDTM/
Grahame Grieve (Nov 30 2021 at 20:01):
@Carol Macumber has HTA worked on these CDISC code systems?
Carol Macumber (Nov 30 2021 at 20:50):
@Grahame Grieve No, as far as i can tell, the CDISC Controlled terminologies, including SDTM, have not yet been requested. Happy to fast track them to the next HTA call if someone opens a JIRA request.
Grahame Grieve (Nov 30 2021 at 20:51):
@Diane best if you create the Jira request - you can be specific about which of those code systems you care about
Diane (Nov 30 2021 at 22:12):
I will work with the rest of my team to get the code systems list ready and then open the jira request.
Diane (Nov 30 2021 at 22:15):
@Patrick Werner Which code systems in the NCIT are you using in your project? Did you open a jira request for the specific code lists or are you referencing the full thesaurus?
Davera Gabriel (Dec 01 2021 at 19:12):
Patrick Werner said:
one more comment:
My projects which are using NCIT are quite new - so a change wouldn't be that hard. Only thing that is important to me: That we get consensus on the correct url soon.
Changing it in a few month will be much harder than changing it now.
on 1APR2021 this was explained in JIRA via this ticket (https://jira.hl7.org/browse/HTA-36) I also made an announcement regarding the disposition of the HTA activity to Zulip per the request of the HTA... but I am unable to search for /find the history here in Zulip to support that assertion.
Patrick Werner (Dec 02 2021 at 09:14):
Grahame Grieve said:
C. because that's also a registry of code systems.
https://ncithesaurus.nci.nih.gov/ncitbrowser/ is only searching using NCIT. I don't think thats a url pointing to a registry.
Patrick Werner (Dec 02 2021 at 09:14):
Diane said:
Patrick Werner Which code systems in the NCIT are you using in your project? Did you open a jira request for the specific code lists or are you referencing the full thesaurus?
We/I am using NCIT as a CodeSystem so whole thesaurus. We defined our own VS on it for specific bindings.
Patrick Werner (Dec 02 2021 at 09:19):
just to summarize:
there is consensus that http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl is the correct SDI for the NCIT CS (urn:oid:2.16.840.1.113883.3.26.1.1)
Patrick Werner (Dec 02 2021 at 09:20):
Is there already a ticket to fix the HL7 THO package and change the url NCIT url in it to http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl as well?
Davera Gabriel (Dec 02 2021 at 12:54):
Hi Patrick: it should be https://jira.hl7.org/browse/HTA-36 what seems to be of issue is that its currently unassigned to anyone. I do not have permissions to make that assignment, but I will get that addressed. Thank you for your patience & persistence !
Davera Gabriel (Dec 02 2021 at 13:00):
Additionally for everyone who is engaged in this thread the announcements made via Zulip regarding both the NCiM and NCiT some time ago are here. I see now there's a typ-o in the title of the NCiM announcement, perhaps this created confusion re: the disposition of this resource.
https://chat.fhir.org/#narrow/stream/179166-implementers/topic/NCI-T.20URI
https://chat.fhir.org/#narrow/stream/179166-implementers/topic/NIC-M
Patrick Werner (Dec 02 2021 at 14:16):
Thanks for the info, i just added the linkt to the wrong url: https://terminology.hl7.org/CodeSystem-v3-nciThesaurus.html
Patrick Werner (Dec 02 2021 at 14:16):
as a comment to the tracker item
Last updated: Apr 12 2022 at 19:14 UTC
