Stream: implementers
Topic: Validatior Error ICD-10 codes
Noemi Deppenwiese (Jul 30 2019 at 09:38):
I ran the official validator on some resources i created, and encountered an error with ICD-10 codes.
e.g. i have this code in on of my conditions:
"code": {
"coding": [
{
"code": "A00.9",
"system": "http://hl7.org/fhir/sid/icd-10",
"version": "2016"
}
]
},
And the validator returns
Error @ Bundle.entry[4].resource.code.coding[0] (line 165, col14) : Code "A00.9" not found; The code "A00.9" is not valid in the system http://hl7.org/fhir/sid/icd-10; The code provided (http://hl7.org/fhir/sid/icd-10#A00.9) is not valid in the value set All codes known to the system
Is this an error in the validator, a problem with ICD-10 or...? And is there any way to tell the validator to ignore all errors connected to a specific code system?
Rob Hausam (Jul 30 2019 at 13:20):
HI, @Noemi Deppenwiese The tx.fhir.org terminology server right now only supports a limited (essentially hand-picked) set of ICD-10 codes, because we don't have a straightforward way to get a complete set from WHO that can be readily imported (if you have that and can share or know how to get it, let us know). In the absence of the full code set, if you have a subset of the codes that you need to have supported for your work then send that to me and I should be able to get them added (possibly today).
John Silva (Jul 30 2019 at 13:36):
@Rob Hausam -- does this site have what you need? https://www.cms.gov/Medicare/Coding/ICD10/2018-ICD-10-PCS-and-GEMs.html
Rob Hausam (Jul 30 2019 at 13:41):
Not exactly, John. I'm able to get the set of US codes for ICD-10-CM (although, unfortunately, the distribution doesn't actually enumerate all of the billable leaf codes), but what we're looking for here is the full set of the "base" ICD-10 codes from WHO.
John Silva (Jul 30 2019 at 13:52):
Oh well, I tried! :-) (I'm sure we use ICD-10 in one of our products; not sure where we get it from though.)
Alexander Kiel (Jul 30 2019 at 16:38):
I was able to download a 9 MB ClaML XML file with the 2016 ICD Codes from http://apps.who.int/classifications/apps/icd/ClassificationDownload/DLArea/Download.aspx
One has to register there, describe the research project and agree with the non-commercial research license.
In order to upload the ICD-10 codes to an official FHIR server, I'm not sure a non-commercial research license is sufficient.
Noemi Deppenwiese (Jul 31 2019 at 10:19):
If there is no complete set in FHIR, is it possible to tell the validator to ignore these codes for now? Or maybe downgrade the error to a warning, since this is more a case of validation not possible, not invalid code.
Rob Hausam (Jul 31 2019 at 12:18):
@Noemi Deppenwiese Probably we can consider that (but I don't know how immediately that could happen if we would decide to ignore it or change to a warning). If you do have a set of codes that need to be supported I can see about adding them. We also need to look at the licensing if we move toward hosting the full (or a substantially complete) set. I think with a small subset we should be OK with that for now (although we should further explore the details of that, too).
Noemi Deppenwiese (Jul 31 2019 at 12:46):
Thanks for the offer, but unfortunately we need the full ICD-10 set for our project . We will look into parsing a complete FHIR CodeSystem from the ClaML and start hosting it locally for our project until the licensing issues you mentioned are sorted out.
Rob Hausam (Jul 31 2019 at 13:01):
Probably that's going to be the best approach to take for now. If you do develop a parser for the ClaML files and would happen to be able to share that, that would of course be very useful to have assuming that we are able to clarify the licensing arrangement with WHO. @Grahame Grieve What about changing these errors to warnings, or having a way to ignore them (similar to the examples)? I'm not sure if the current error suppression mechanism would be sufficient for this.
Patrick Werner (Jul 31 2019 at 14:45):
Thanks for the offer, but unfortunately we need the full ICD-10 set for our project .
i suppose you'll need the whole ICD-10-GM ?
Patrick Werner (Jul 31 2019 at 14:49):
my colleague @Hauke Hund implemented ICD-10-GM in his project medicats: https://github.com/hhund/medicats
https://medicats.gecko.hs-heilbronn.de/
Unfortunately not FHIR, but with REST.
Patrick Werner (Jul 31 2019 at 14:50):
If there is no complete set in FHIR, is it possible to tell the validator to ignore these codes for now? Or maybe downgrade the error to a warning, since this is more a case of validation not possible, not invalid code.
you have to filter the errors on your side. The terminology server doesn't know wether the code you are sending is wrong or just not in his DB.
Noemi Deppenwiese (Jul 31 2019 at 15:52):
you have to filter the errors on your side. The terminology server doesn't know wether the code you are sending is wrong or just not in his DB
Isn't the content attribute used to indicate that not all valid codes are listed in the CodeSystem?
i suppose you'll need the whole ICD-10-GM ?
We in fact need the international version since we want to connect biobanks across Europe with BBMRI. But thanks for the links, they still may help us.
Jim Steel (Jul 31 2019 at 23:23):
I have a ClaML to FHIR CodeSystem converter, that I've used with ICD10-GM, ICD10-EN, and a few others
Noemi Deppenwiese (Aug 01 2019 at 07:33):
I have a ClaML to FHIR CodeSystem converter, that I've used with ICD10-GM, ICD10-EN, and a few others
Is your program by any chance open source / on github?
Rob Hausam (Aug 01 2019 at 07:49):
@Jim Steel That sounds good. It looks like I may also have a fully enumerated XML format file available for the US ICD-10-CM variant. Have you imported ICD-10-CM in Ontoserver? I think now the main (or only) barrier to implementing these codes (multiple ICD-10 variants) in the build terminology server is making sure that the licensing issues are worked out. I'm curious what the Ontoserver experience has been with that?
Jim Steel (Aug 01 2019 at 07:50):
We don't distribute Ontoserver pre-loaded with any ICD content, so we haven't really dealt with licensing
Jim Steel (Aug 01 2019 at 07:50):
There are also a couple of ClaML features I haven't gotten around to, mostly significantly modifiers
Rob Hausam (Aug 01 2019 at 07:53):
OK. I probably should look again at how the modifiers are expressed. But it sounds like that's not a showstopper for you.
Jim Steel (Aug 01 2019 at 07:53):
I think we probably could use some clarity whether modifiers mean new child codes, or whether they are encoded as extensions or something
Jim Steel (Aug 01 2019 at 07:54):
I recall you mentioning them in Montreal, but I don't think we really got into it
Rob Hausam (Aug 01 2019 at 08:00):
Right. Maybe we can both do some further research on it and see where it ends up. From my recollection, the US CM variant tends to have the most complexity, with up to 7 character codes (I don't recall the other variants having that many). But if the available distribution files fully enumerate the valid (i.e. billable in the US) codes, then the length of the code doesn't really matter.
Alexander Kiel (Aug 01 2019 at 10:12):
We (@Noemi Deppenwiese and I) are interested in having the international ICD-10 version 2016 as CodeSystem. We like to use it for validation and also possibly for UI generation and search. @Jim Steel can we use your converter? We have the ClaML file already.
Regarding the licensing issues. Who from the FHIR community would be most suited to talk with the WHO about this?
Grahame Grieve (Aug 01 2019 at 10:13):
@Rob Hausam
Grahame Grieve (Aug 01 2019 at 10:13):
it would logically live in the fhir.tx.support package
Rob Hausam (Aug 01 2019 at 13:58):
This one just needs to be loaded from a data file specified in fhirserver.ini. But we need to know that we (FHIR/HL7) have or can get a license to host the full set of codes.
Rob Hausam (Aug 01 2019 at 14:12):
@Alexander Kiel @Noemi Deppenwiese @Grahame Grieve I can begin investigating this with WHO. If anyone is aware of a particular contact or place to start, that would be useful.
Alexander Kiel (Aug 01 2019 at 17:21):
Unfortunately I don't know any contact at the WHO.
Grahame Grieve (Aug 01 2019 at 22:11):
I do.
Rob Hausam (Aug 01 2019 at 23:59):
@Grahame Grieve How do you suggest we proceed with it?
Grahame Grieve (Aug 02 2019 at 01:37):
Wayne and I are talking to them now. I've added this to the list of things we're talking about
Rute Martins (Aug 02 2019 at 03:47):
Wayne and I are talking to them now. I've added this to the list of things we're talking about
Is ICD-O-3 on that list as well?
Grahame Grieve (Aug 02 2019 at 04:02):
what's that?
Rute Martins (Aug 02 2019 at 04:14):
International Classification of Diseases for Oncology: https://www.who.int/classifications/icd/adaptations/oncology/en/. The licensing issues are similar, and we would love to see ICD-O-3 codes in the FHIR terminology server. It's THE standard used by cancer registries to represent cancer histology, location and behavior (at least in North America).
Rob Hausam (Aug 02 2019 at 15:20):
We've also already added initial skeleton support for ICD-O-3, as we use it in our IPS example. So I agree that it should be included in the discussion along with ICD-10.
Michael Lawley (Aug 05 2019 at 05:46):
@Alexander Kiel @Jim Steel is on leave now for a couple of weeks. I'm not sure what the licensing status is of the converter code, but I'll look into it and let you know. I don't imagine it will be a blocker, just some hoops to jump through organisationally.
Alexander Kiel (Aug 05 2019 at 07:49):
@Michael Lawley Thanks. I would like to have access to the converter. So we could at least use ICD-10 internally until global licensing issues are sorted out.
Alexander Kiel (Aug 05 2019 at 07:50):
@Grahame Grieve @Rob Hausam We also have a strong need for ICD-O-3. So it would be very good to have it in the Terminology Server.
Grahame Grieve (Aug 05 2019 at 08:03):
so looking at the WHO licenses - they have no license for us to use ICD-10 or ICD-O-3: we simply don't fit into any of the categories they describe. The only way I can see this moving forwards is directly negotiating with WHO. We have such a discussion going on, but given that we don't fit ito any box, no one should expect that this will resolve any time soon.
Rob Hausam (Aug 05 2019 at 12:25):
That's a bit unfortunate, but it's where we are. Hopefully progress will be made.
Robert McClure (Aug 15 2019 at 04:34):
I've not read through this all but since it's an external terminology you need to send the needs and issues to HTA for them to help sort things out. HTA
Grahame Grieve (Aug 15 2019 at 04:35):
well, perhaps you should read the thread then
Halina Labikova (Nov 08 2019 at 21:18):
Jumping onto this thread since I don't find anything more recent - have there been any developments in the discussions with WHO re ICD-O-3 in the terminology server?
Grahame Grieve (Nov 08 2019 at 21:19):
yes some but the earth moves slowly
Halina Labikova (Nov 08 2019 at 21:40):
It's cruising at 30 kilometers per second, that's seems speedy to me! :P (any channels you would recommend me to follow to be up to date with future developments?)
Grahame Grieve (Nov 08 2019 at 21:41):
not really. Currently we are dsicussing MoU with WHO, and I'm visiting WHO for a meeting in a couple of weeks where I can push things along a bit
Halina Labikova (Nov 08 2019 at 21:50):
Ok, thank you!
Alexander Kiel (Nov 09 2019 at 07:36):
@Noemi Deppenwiese
Lin Zhang (Nov 09 2019 at 11:05):
Right. Maybe we can both do some further research on it and see where it ends up. From my recollection, the US CM variant tends to have the most complexity, with up to 7 character codes (I don't recall the other variants having that many). But if the available distribution files fully enumerate the valid (i.e. billable in the US) codes, then the length of the code doesn't really matter.
Much more complex in China. There are more than a dozen of regional variants of ICD-10, including recently published version by the National Healthcare Security Administration (NHSA, Equivalent to US CMS) . And the number of digits can reach up to 11 (e.g., A00.000x001 for Cholera due to Vibrio cholerae 01, biovar cholerae).
Grahame Grieve (Nov 09 2019 at 11:33):
what is the license on the chinese versions?
Lin Zhang (Nov 09 2019 at 14:32):
It's up to the WHO Collaborating Center based in the Peking Union Medical College Hospital. It is estimated that most people don’t really understand or care the license issue of WHO ICD, possibly because it's a governmental action to adopt and mandate the classification system. The extension/version proliferation problem is challenging, especially for our national CHS-DRGs' current piloting efforts and future implementations.
Lin Zhang (Nov 09 2019 at 14:40):
But this is getting more and more concerns as the nation-wide awareness of IP protections increases.
Grahame Grieve (Nov 26 2019 at 19:26):
ok, talking to WHO terminology term about this: I can (probably) get a license to host WHO code systems on tx.fhir.org without waiting for a formal MoU - which we expect to take quite some time. But no one can publish actual IGs that include any WHO classifications unless they have an agreement with WHO
It's not clear to me whether this is the case - the people who are asking for support for WHO classifications: do you have a license from WHO to publish IGs that include them? If so,where did you get the license from? if not, are you asking for the HL7 license to cover you? Is this for an afilliate project?
Alexander Henket (Nov 26 2019 at 20:41):
I've always been told (by @Frank Oemig for example) that each country has its own version of ICD-10 each having its own peculiarities. For NL you need to license from RIVM (organization comparable to US CDC). This is also why each country has its own system identifier (oid or uri).
I'm sure WHO could give you permission for some international set, but would that help for country XYZ? Hopefully they could shed some light there too.
Frank Oemig (Nov 26 2019 at 21:13):
IMO, you can only approach the national centers who are in charge trying to convince them to publish the national versions in different formats. One should be ClaML which is used already, and FHIR for the more advanced.
But that highly depends on the maturity of the country with regard to their understanding of the problem. Germany is little bit behind here...
Grahame Grieve (Nov 27 2019 at 04:51):
I'm not worried about formats so far, I'm trying to work through the minefield that is licensing
Igor Sirkovich (Dec 19 2019 at 02:18):
@Grahame Grieve IG Publisher generates errors "The code ... is not valid in the system http://hl7.org/fhir/sid/icd-10-cm" for all diagnosis and "Unknown Code http://hl7.org/fhir/sid/ex-icd-10-procedures... in http://hl7.org/fhir/sid/ex-icd-10-procedures" for all procedure elements in my sample ExplanationOfBenefit resource.
I just wanted to confirm whether this is the expected behavior at this stage and whether we can safely ignore these errors until full tx.fhir.org starts hosting all ICD-10 codes.
Grahame Grieve (Dec 19 2019 at 02:40):
those are ICD-10-cm codes, not ICD-10 codes
Igor Sirkovich (Dec 19 2019 at 03:00):
Yes, ICD-10-cm and ICD-10 Procedures. Should we ignore these errors or ICD-10-cm & ICD-10 Procedures codes are supposed to be supported by the IG Publisher, in which case we need to investigate and fix the errors on our end?
Grahame Grieve (Dec 19 2019 at 03:05):
you should use the right system url
Grahame Grieve (Dec 19 2019 at 03:06):
still won't be supported though, which you can't fix. So you have to be extra careful to check the codes.... but not even using the right URL isn't confidence boosting
Igor Sirkovich (Dec 19 2019 at 03:09):
Actually, we were using http://hl7.org/fhir/sid/icd-10 initially, but were told to change to http://hl7.org/fhir/sid/icd-10-cm, which is a US version of ICD-10 based on the description of ICD-10 variants at https://www.hl7.org/fhir/icd.html#variants
Grahame Grieve (Dec 19 2019 at 03:10):
yes that's what it should be: http://hl7.org/fhir/sid/icd-10-cm
Last updated: Apr 12 2022 at 19:14 UTC
