Stream: terminology
Topic: LOINC CodeSystem
Alexander Henket (Sep 05 2017 at 03:26):
Did anyone do a LOINC CodeSystem? I'm struggling with translated properties, and I'm wondering about relatednames2/long_common_name/shortname.
My current mapping: long_common_name [en-US] goes into concept.display, other language long_common_name goes into concept.designation [type FSN]. Any shortname goes into concept.designation [type SYN]. Any relatednames2 are split by semicolon and go into concept.designation [type SYN]. These three columns are not duplicated as property.
All en-US columns go into concept.property as string. All other language columns also go into concept.property and I've added the language to the property [currently as an extra element since it doesn't get exposed]. E.g.:
<property>
<language value="it-IT"/>
<code value="COMPONENT"/>
<valueString value="Onda R, durata.derivazione III"/>
</property>
How have others handled all that? Would it make sense to have an optional language on concept.property in the core?
Grahame Grieve (Sep 05 2017 at 12:44):
@Daniel Vreeman
Grahame Grieve (Sep 05 2017 at 12:45):
here's my working loinc code system which I did with Dan, and is planned to be posted to loinc.org:
Rob Hausam (Sep 05 2017 at 13:50):
@Alexander Henket I think your mapping looks pretty reasonable. The "Using LOINC with FHIR" page (http://build.fhir.org/loinc.html) gives options of either SHORTNAME or LONG_COMMON_NAME for display, but I agree with your choice. I don't think that the SNOMED CT codes in the current designation-use value set really fit well for LOINC. The LONG_COMMON_NAME isn't really the same idea as a "Fully specified name", so I'm not sure if I would use it for that. I think I would probably use "Synonym" for all of the designations - or maybe consider extending the value set with LOINC-specific codes? I guess I've not really looked at non-English columns (and I'm not even sure where to get them). Possibly an optional core property.language element would make sense (assuming that it's in the 80%).
Alexander Henket (Sep 05 2017 at 13:56):
@Grahame Grieve You have one done? That sounds good. There is no link to it in your response, is that correct?
Alexander Henket (Sep 05 2017 at 13:59):
@Rob Hausam I was struggling with FSN vs Synonym too. Actually FSN is strictly speaking only applicable to SNOMED so in a sense the extensible binding https://www.hl7.org/fhir/valueset-designation-use.html is somewhat limited. The addition of a "preferred" marker would also do nicely. I'll check what Grahame/Dan came up with and see from there
Grahame Grieve (Sep 05 2017 at 14:43):
my apologies - I got distracted
Grahame Grieve (Sep 05 2017 at 16:51):
Jim Steel (Sep 06 2017 at 00:01):
Ontoserver's LOINC is here: https://ontoserver.csiro.au/stu3-latest/CodeSystem/LOINC-2.61
Alexander Henket (Sep 06 2017 at 14:42):
Thanks @Grahame Grieve. I see a typo that you probably want fixed here (leading space):
<valueSet value=" http://loinc.org/vs"/>
I also see you don't intend to import all property columns like EXAMPLE_UCUM_UNITS etc. That's fine, I can probably just choose to add those anyway.
This doesn't yet tell me what a particular concept would look like, or how I would handle the LinguisticVariants of those properties. Do you or @Jim Steel have a version with one or more concepts?
Jim Steel (Sep 27 2017 at 23:55):
Sorry, just caught this. How do you mean "with one or more concepts"? Ours has all the concepts. They just don't come back in a CodeSystem read. You need to do a $expand, $lookup, or similar
Grahame Grieve (Oct 03 2017 at 05:44):
$Lookup should give you what you want
Christof Gessner (Jul 03 2020 at 08:47):
@Alexander Henket Do you have any information on access to LOINC language variants (de-DE etc.) via https://fhir.loinc.org/CodeSystem/$lookup?system=http://loinc.org&code=4544-3? @Simone Heckmann @Daniel Vreeman @Christine Haas
Alexander Henket (Jul 06 2020 at 10:48):
I don't know if linguistic variants (whoever came up with that term....) are implemented through that. I do know that we are still struggling to get everything we have on localization (which is the better term...) to Regenstrief. Currently our Long Common Name does not leave The Netherlands
Jim Steel (Jul 06 2020 at 10:50):
Are you using CodeSystem supplements in the meantime?
Cora Drenkhahn (Jul 06 2020 at 11:04):
Alexander Henket said:
I don't know if linguistic variants (whoever came up with that term....) are implemented through that.
Looking at this discussion, they are probably not implemented yet, but Regenstrief seems to be working on it.
Lin Zhang (Sep 27 2020 at 01:16):
LOINC's FSN is a concatenation of the values of its six axes/fields in tandem and using colons (:) as delimiters.
Lin Zhang (Sep 27 2020 at 01:19):
All the other names (long common name and short name) and the linguistic variants are essentially synonyms. But notably, short names are not neccessorily unique IIRC.
Lin Zhang (Sep 27 2020 at 01:23):
My Question: is it sensible to make them as designations with different "uses" and different languages?
Lin Zhang (Sep 27 2020 at 01:24):
In addition to the official FSN, there would be other non-English FSN(s) for a LOINC code.
Grahame Grieve (Sep 27 2020 at 21:42):
I think so.
Swapna Abhyankar (Oct 08 2020 at 18:35):
Hi - correct, we haven't implemented linguistic variants (I'm also not sure who came up with that term!) in our FHIR TS yet but it's on our to-do list. If I understand correctly, the designation.language element would hold the language, designation.value would be the actual translation, and designation.use would be shortname, FSN, etc. Does that sound right?
And Lin is correct, the LOINC FSN is just the 6 parts delimited by colons.
Lin Zhang (Oct 08 2020 at 23:38):
@Grahame Grieve @Swapna Abhyankar Thanks.
Last updated: Apr 12 2022 at 19:14 UTC
