Stream: implementers
Topic: vaccineCode being a vaccineCode
Dave Barnet (Apr 10 2018 at 14:57):
Hope this is a quick query. After some analysis in the UK, we've found that a lot of systems record immunizations as a completed procedure, rather than holding the actual vaccine used. So for example a system might hold a record that a patient has had (in SNOMED) 247641000000105 | Second pneumococcal conjugated vaccination (procedure) |. This causes an issue trying to use the immunization resource, as within Immunizations the vaccineCode is mandated.
Is there a recommended approach to interoperate this data? - some options might be
Should we use a procedure resource - this takes away from the immunization-ness of the data?
Should we just put the procedure code in vaccineCode - this seems to break the resource?
Should we create an extension for the procedure & put a "Not Known" in the vaccineCode?
Any help appreciated
Lloyd McKenzie (Apr 10 2018 at 17:09):
@Craig Newman @Joginder Madra ?
Craig Newman (Apr 10 2018 at 17:22):
I haven't heard of this before (but I don't have a lot of exposure to workflows outside the US. The important thing is that there is an indication of what was administered. If that can be inferred from the SNOMED code then I personally don't have a problem with using SNOMED codes as the implementation specific value set for vaccineCode, but I don't know if that violates FHIR best practices. An extension would also probably work fine.
Lloyd McKenzie (Apr 10 2018 at 18:47):
It's not "good" practice to send a SNOMED procedure code in an element meant to convey a substance, but we have to work with what systems can do. It would be better to communicate a high-level code for "pneumococcal conjucated vaccine", but it may be that SNOMED doesn't have it or that it's programatically hard for the UK software systems to navigate from the procedure code they currently capture to the high-level vaccine code that we'd like. (The other piece is whether the "first" vs. "second" needs to be conveyed elsewhere as well.)
Grahame Grieve (Apr 10 2018 at 22:01):
technically the binding is an example binding. So you can bind whatever codes you want there, and there's no reason not to allow a snomed procedure code if it infers a vaccine, and that's what systems have, and you can't ask them to infer an actual vaccine code. To me, this is exactly the same as a problem list that includes medications as proxies for problems - something that production healthcare records have a lot of
Alexander Henket (Apr 11 2018 at 22:12):
Looks as if inference of substance is an option?
Joginder Madra (Apr 18 2018 at 17:35):
i would echo Grahame's comment about the binding being an example binding.
Michael Lawley (May 29 2018 at 08:14):
Would it be reasonable to include the linked substance code as an additional coding?
Lloyd McKenzie (May 29 2018 at 12:59):
It would be legal from FHIR's perspective, though if you've got the substance code, it'd probably be best to have just that. I think the original usecase was the system didn't have the substance code, only the procedure code.
Jose Costa Teixeira (Jan 06 2022 at 15:34):
I'd like to confirm - this is a code (procedure) that would appear in vaccine.code, correct?
David Pyke (Jan 06 2022 at 15:49):
yes, vaccineCode would hold the SNOMED code if CVX use isn't possible
Jose Costa Teixeira (Jan 06 2022 at 17:14):
Question is:
There is a SNOMED code for Pertussis vaccine (medicinal product) and Pertussis vaccination (procedure)
Jose Costa Teixeira (Jan 06 2022 at 17:15):
the one I'd consider in the Vaccine.code is the procedure, and not the medicinal product.
Jose Costa Teixeira (Jan 06 2022 at 17:17):
(same applies for substance as for medicinal product)
David Pyke (Jan 06 2022 at 18:36):
Using the medicinal product is preferred, the vaccineCode element is "Vaccine product administered". You'd only use the procedure if there's no product code.
Jose Costa Teixeira (Jan 06 2022 at 20:03):
we have an extension for vaccine product administered because we separate the 3 concepts: vaccine (procedure type, e.g. MMR, DTP), product (the one that has manufacturer and lot etc) and the diseases (e.g. Tetanus)
Jose Costa Teixeira (Jan 06 2022 at 20:04):
I really don't like putting the product code in the place where "MMR" can also go
David Pyke (Jan 06 2022 at 20:04):
It's an example binding so you can define how you want to do it.
Jose Costa Teixeira (Jan 06 2022 at 20:15):
Being an example binding doesn't mean I can use for different concepts
David Pyke (Jan 06 2022 at 20:15):
Nope, but you've not accepted anything I've said.
Jose Costa Teixeira (Jan 06 2022 at 20:16):
I don't know what you said, only what you wrote :)
Jose Costa Teixeira (Jan 06 2022 at 20:16):
(sorry, it's just that my point seems not to be clear)
David Pyke (Jan 06 2022 at 20:17):
The vaccineCode element is for the medicinal product, not the procedure.
David Pyke (Jan 06 2022 at 20:17):
IF you wish to put anything else in it, you're not being conformant
Jose Costa Teixeira (Jan 06 2022 at 20:17):
I don't want to use vaccineCode for product code OR vaccine code, because I need them both. So I use vaccineCode for things like MMR, DTP, HPV
Jose Costa Teixeira (Jan 06 2022 at 20:18):
in SNOMED, this seems to be it:
image.png
Jose Costa Teixeira (Jan 06 2022 at 20:18):
procedure
David Pyke (Jan 06 2022 at 20:19):
Yes, for some SNOMED doesn't have the product as a code so you have no choice but to use the procedure
Jose Costa Teixeira (Jan 06 2022 at 20:20):
that I don't agree - when we don't have a product code, we use procedure.
David Pyke (Jan 06 2022 at 20:20):
Yes, that's forced by a lack of a code that matches
David Pyke (Jan 06 2022 at 20:20):
But if you have a product code, you should use the product code
Jose Costa Teixeira (Jan 06 2022 at 20:20):
lack of codes can be handled by adding codes, that is not a difficulty
Jose Costa Teixeira (Jan 06 2022 at 20:21):
MMR is not a product code, and the example binding is full of those
David Pyke (Jan 06 2022 at 20:21):
Yeah, well, that example binding value set should be updated
Jose Costa Teixeira (Jan 06 2022 at 20:21):
what example?
Jose Costa Teixeira (Jan 06 2022 at 20:22):
no, those are CVX codes
Jose Costa Teixeira (Jan 06 2022 at 20:22):
http://build.fhir.org/valueset-vaccine-code.html
Jose Costa Teixeira (Jan 06 2022 at 20:23):
I think the description of the element is off.
David Pyke (Jan 06 2022 at 20:23):
Yes, CVX has MMR as a single product as it's often set as a single vaccine. However, if you're using SNOMED, then you don't have that option
Jose Costa Teixeira (Jan 06 2022 at 20:24):
I'm lost.
Jose Costa Teixeira (Jan 06 2022 at 20:26):
I have one confirmation pending:
Let's say SNOMED has 2 snomed codes related to "MMR vaccine" - one for medicinal product and one for procedure.
Aligning with CVX, which is "codes for vaccines" and not the product - although there is a correspondence - I'd say that vaccineCode would use procedures, not medicinal products
David Pyke (Jan 06 2022 at 20:27):
Then you should open a ticket to change the description
Jose Costa Teixeira (Jan 06 2022 at 20:27):
that is what this discussion is for - hence starting with the example.
Jose Costa Teixeira (Jan 06 2022 at 20:27):
@Craig Newman is tomorrow a good time to discuss vaccine codes and product codes? I'm being increasingly asked how we fixed that so I want to align what we have and have discussed some times. We postponed this in December, perhaps we can catch up soon?
Jose Costa Teixeira (Jan 06 2022 at 20:29):
David Pyke said:
Then you should open a ticket to change the description
indeed, and that change wouldn't be inocuous so I'm using the example given by someone else to put the finger there. The ticket was already there
Craig Newman (Jan 06 2022 at 22:26):
@Jose Costa Teixeira Tomorrow's call would be fine. I have a conflict for the first half, but if you want to get the party started without me, I'll join as soon as I can
Neha Agarwal (Jan 07 2022 at 05:25):
DrugUse and TobaccoUse can be found as nictiz profile STU3 for observation resource but i cannot see similar profiles for in base profiles of Observation resource for R4 version. What could be the possible reason?
Jose Costa Teixeira (Jan 07 2022 at 07:18):
@Neha Agarwal wrong thread / topic?
Last updated: Apr 12 2022 at 19:14 UTC
