FHIR Chat · AMT ValueSet bindings · australia

Stream: australia

Topic: AMT ValueSet bindings

Michael Lawley (Jul 03 2018 at 05:53):

I note that the various Medication-oriented AU Base profiles bind to explicit but indirect AMT-specific ValueSets like http://hl7.org.au/fhir/ValueSet/amt-ctpp-codes rather than more directly to implicitly defined ValueSets like http://snomed.info/sct/32506021000036107?fhir_vs=refset/929360051000036108

Is there a reason for this approach?

Brett Esler (Jul 03 2018 at 06:41):

i can't think of a reason not to use implicit set....

Brian Postlethwaite (Sep 03 2020 at 07:09):

The explicit one is easier to look at :wink: and now that I'm here, from an instance perspective is there any way we can tell which property is an MPP, MP, TP etc without a terminology server?
@Danielle Bancroft @Brett Esler @Michael Lawley
This is so that when I see this

<coding>
   <system value="http://snomed.info/sct" />
   <version value="http://snomed.info/sct/32506021000036107/version/20200601" />
   <code value="36900011000036109" />
</coding>

I can know which of the things it is, now that I've pasted it I don't even know which it was.

Brian Postlethwaite (Sep 03 2020 at 07:31):

That's not what this extension is for is it?
http://hl7.org/fhir/extension-coding-sctdescid.html

Brian Postlethwaite (Sep 03 2020 at 07:33):

Or is this a better extension where we could point to the valueset that you've pointed at Michael?
http://hl7.org/fhir/extension-valueset-reference.html

<coding>
   <extension url="http://hl7.org/fhir/StructureDefinition/valueset-reference">
      <valueUri value="http://hl7.org.au/fhir/ValueSet/amt-ctpp-codes"/>
   </extension>
   <system value="http://snomed.info/sct" />
   <version value="http://snomed.info/sct/32506021000036107/version/20200601" />
   <code value="36900011000036109" />
</coding>

Brian Postlethwaite (Sep 03 2020 at 07:41):

And where did they go for R4? (they are in STU3)
http://hl7.org.au/fhir/aubase1.1/ValueSet-amt-tpp-codes.html
Did we move them to terminology.au?
Are we favouring the less pleasant to look at (and harder to QA)
https://healthterminologies.gov.au/fhir/ValueSet/sctau-reference-set-929360041000036105

Richard Townley-O'Neill (Sep 03 2020 at 23:50):

There is an extension defined in AU Base R4 to say whether a medication id a UU or a TP etc: http://build.fhir.org/ig/hl7au/au-fhir-base/StructureDefinition-medication-type.html

Brian Postlethwaite (Sep 04 2020 at 01:48):

I think we're going to go with the approach I've noted above, so the valueset used is incldued, as it's the only way to split them apart.
The MedicationType doesn't indicate which of the valuesets was used (or at least I couldn't work it out)

Richard Townley-O'Neill (Sep 04 2020 at 01:51):

It does not indicate the value set. It indicates, indirectly, whether it was a trade product, or a containered trade product, or whether it was a medicinal product etc. There is a mapping somewhere that shows that. The original idea was that its codes were not Australian specific. @Dion McMurtrie is the expert.

Brett Esler (Sep 04 2020 at 01:58):

Like @Richard Townley-O'Neill said http://build.fhir.org/ig/hl7au/au-fhir-base/StructureDefinition-medication-type.html can be used for that; problem with Valuesets is that depending on what you are doing this can change (subsets, refsets or otherwise) so not that reliable for exchanging in every scenario - so went for an explicit coded extension where you don't want to have receivers requiring a terminology lookup

Brett Esler (Sep 04 2020 at 01:59):

we got rid of the AMT valueset in AU base and stopped using them in AMT coded content - simplified

Brett Esler (Sep 04 2020 at 02:01):

the extension can also be used with other terminologies to indicate the broad class of the content; @Dion McMurtrie has done some POC work with other drug terminologies also similarly...

Brian Postlethwaite (Sep 04 2020 at 03:24):

So which ones map to which ones?
MP
MPP
TPP
MPUU
TP
TPUU
CTPP

Richard Townley-O'Neill (Sep 04 2020 at 03:34):

Good question.

Brian Postlethwaite (Sep 04 2020 at 03:38):

Brett filled me in it is:
MP = UPD; MPUU = UPDSF; MPP=UPG; TP = BPD, TPUU = BPDSF, TPP=BPG

Richard Townley-O'Neill (Sep 04 2020 at 03:40):

I tried mapping using the definitions and got the same. I was not sure though. I also got CTPP=BPGC

Brian Postlethwaite (Sep 04 2020 at 03:49):

Here is a sample with both in it

        <coding>          <!-- MP -->
          <extension url="http://hl7.org.au/fhir/StructureDefinition/medication-type">
            <valueCoding>
              <system value="http://hl7.org.au/fhir/CodeSystem/medication-type"/>
              <code value="UPD"/>
            </valueCoding>
          </extension>
          <extension url="http://hl7.org/fhir/StructureDefinition/valueset-reference">
            <valueUri value="https://healthterminologies.gov.au/fhir/ValueSet/sctau-reference-set-929360061000036106"/>
          </extension>
          <system value="http://snomed.info/sct" />
          <version value="http://snomed.info/sct/32506021000036107/version/20200601" />
          <code value="14207011000036103" />
        </coding>

Brian Postlethwaite (Sep 04 2020 at 03:55):

Thanks, Now that I have the full mapping there I'll go with that, and drop the valueset reference that is shown above.
Any reason we went with Coding rather than Code with the required binding?

Brian Postlethwaite (Sep 04 2020 at 04:00):

Now that we have that, how can I slice the Coding based on the extension value to make this possible?
Or do we have to wander into the expression based slicing - extension still to be defined for this...

Richard Townley-O'Neill (Sep 04 2020 at 04:01):

I don't know about coding vs codeable concept.

Richard Townley-O'Neill (Sep 04 2020 at 04:03):

http://build.fhir.org/ig/hl7au/au-fhir-base/StructureDefinition-au-medication.html uses the extension on the amt slice

Richard Townley-O'Neill (Sep 04 2020 at 04:05):

I'm not confident that that slicing is a good way to go.

Peter Jordan (Sep 04 2020 at 04:50):

Interesting approach. Thus far, for NZMT, which is structurally very similar to AMT, we've treated Medication Type as a Code System property (NZULM Type) representing one of the '7 boxes' (MP, MPP, MPUU, TP, TPP, TPUU, CTPP). It would be insightful to know the use cases for creating an extension beyond the obvious one of a transparent distinction between a generic and branded drug - because whether it's a product, unit of use or pack is usually (fairly) obvious from the description.

Grahame Grieve (Sep 04 2020 at 05:05):

as I understand it the use case is:

• you need to pick out the different codes and put them in different slots
• you don't want to take the time to consult a terminology server that can tell you which is which

Brian Postlethwaite (Sep 04 2020 at 21:07):

So this is where I ended up in the profile
https://fhir.medicationknowledge.com.au/dev/StructureDefinition-mysl-medication.html
With an example here
https://fhir.medicationknowledge.com.au/dev/MedicationRequest-1XPR352VP3QCX7BJD6.xml.html

Richard Townley-O'Neill (Sep 08 2020 at 06:13):

Thanks.
I suggest that Medication.code.coding.medication-type should be 0..1 instead of 0..*
Also, that you include an example that gives Medication.ingredient.strength as mg/1.
I, a consumer, think of meds as having so many mg per tablet, not so many mg per mg.

Michael Lawley (Sep 08 2020 at 06:55):

If you included the FSN in the display text, then you'd know (as a human) what was what.
If you're going to use an extension to indicate which hierarchy, why use a ValueSet? Why not just the MP MPUU CTPP, etc code itself?

Michael Lawley (Sep 08 2020 at 06:58):

@Dion McMurtrie would have a take on this too, as it is something he tackled for MedServ (with a view to a non-AMT specific solution). I think this might be the source.

Brian Postlethwaite (Sep 08 2020 at 10:54):

I'll update that cardinality, good pickup.

Brian Postlethwaite (Sep 08 2020 at 10:56):

@Michael Lawley we're putting the type which matches. Removes need for terminology server to pick which heirachy it is.

Michael Lawley (Sep 08 2020 at 11:02):

Yep - I believe those are the "neutral" codes that Dion invented to assist in aligning AMT, RxNorm, DM+D, etc

Dion McMurtrie (Sep 09 2020 at 22:27):

Sorry to be late to this, but yes in essence those codes were an attempt to create a general code mapping to the levels exposed in the various drug models. All the models have heavily intersecting sets of these levels, but call them different things. So I came up with something agnostic of any particular model (not AMT's MP, CTPP, TPP... not dm+d's AMP, VTM...etc).

I can supply a ConceptMap if that's useful?

Richard Townley-O'Neill (Sep 09 2020 at 22:28):

I can supply a ConceptMap if that's useful?

Yes please.

Peter Jordan (Sep 09 2020 at 22:38):

That would be useful - R4 and/or R5 versions?

Michael Lawley (Sep 10 2020 at 01:32):

What is this "R5" of which you speak? ;-)

Brian Postlethwaite (Jun 15 2021 at 22:47):

Adding to the next PA meeting a discussion on how we can move forward NCTS dependencies in the AU base IG (among others)
So we can remove the validation info/warning messages that those aren't in the IG - such as countries, languages, IHI statuses etc.

Richard Townley-O'Neill (Jun 16 2021 at 03:49):

If you do not want to validate the examples, the warnings can be removed by adding 50 odd rows to the file suppressed-warnings.txt. (Many valueset warnings appear twice, so only need one entry.) This was used for SRA, see the bottom of https://sra.digitalhealth.gov.au/fhir/currentdraft/qa.html

Jim Steel (Jun 16 2021 at 05:20):

@Brian Postlethwaite For those who might not be regular PA meeting attendees but are interested in that topic, when is said next meeting?

Brian Postlethwaite (Jun 16 2021 at 08:00):

Next Wednesday, will send you the details.

Jim Steel (Jun 16 2021 at 22:59):

Thanks

Last updated: Apr 12 2022 at 19:14 UTC