Stream: implementers
Topic: COVID Immunization IG
Kumar Satyam (Jan 16 2021 at 08:14):
Are there any IGs for Covid vacination. Tried searching could not find any however I remember seeing some discussions around it.
René Spronk (Jan 16 2021 at 09:29):
In the US, there's SANER: http://build.fhir.org/ig/HL7/fhir-saner/ - see also https://fire.ly/2020/04/03/fighting-covid-19-with-fhir/ for initiatives in other countries.
René Spronk (Jan 16 2021 at 09:29):
Those are not specific to immunization however.
Jose Costa Teixeira (Jan 16 2021 at 10:48):
in Belgium we made a Vaccination profile, and added a few covid examples to it.
IHE is starting a Vaccination IG - including some covid-driven requirements, such as the different vaccination schemes for different brands
Kumar Satyam (Jan 16 2021 at 14:17):
Thank you. Will look into these.
Jens Villadsen (Jan 16 2021 at 14:26):
Given my upcomming engagement in WHO's WG I'll welcome any links to IG's that profiles Immunization @René Spronk @Jose Costa Teixeira
René Spronk (Jan 16 2021 at 14:37):
https://www.google.com/search?q=immunization+structuredefinition+implementation+guide ?
Jose Costa Teixeira (Jan 16 2021 at 14:58):
we're starting the IHE IG here - i'll post a first logical model soon
https://github.com/IHE/pharm-vaccination
Jose Costa Teixeira (Jan 16 2021 at 14:58):
(soon - in a few hours, I hope . We want to capture some use cases)
Jose Costa Teixeira (Jan 16 2021 at 14:59):
note that the scope of IHE IG is about the immunization, not the certificate
Jose Costa Teixeira (Jan 16 2021 at 15:03):
I had stumbled upon this some time ago,
http://build.fhir.org/ig/ehealthsuisse/ch-vacd/
Jose Costa Teixeira (Jan 16 2021 at 15:03):
(not sure if useful)
John Moehrke (Jan 18 2021 at 15:58):
I think this confluence page is intended to be a place where these activities can be grouped. https://confluence.hl7.org/display/CR/COVID-19+Response+Home
Craig Newman (Jan 20 2021 at 20:11):
At next week's WGM, the Public Health WG has an Immunizations session scheduled for Friday Q2 if this topics needs some discussion.
Peter Jordan (Jan 20 2021 at 22:14):
It almost goes without saying that this is a critical topic.
The latest version of the IPS Profile on the Immunization Resource can be found here
This was discussed in a session on the IPS Track at last week's Connectathon and recorded as such in the Report Out
"Vaccination data and IPS
Discussion of how the IPS Immunization profile and other structures can be used for representing and exchanging this data
Potential opportunities for IPS profiles in "digital vaccination card" with WHO, etc. "
Josh Mandel (Jan 20 2021 at 22:16):
There's also work underway at https://github.com/dvci/vaccine-yellow-card-ig (FYI @Reece Adamson @Max Masnick )
Peter Jordan (Jan 20 2021 at 22:41):
Thanks @Josh Mandel - perhaps someone, with time to spare :), might reconcile the IPS and Vaccine Credential IG profiles on Immunization.
From an international perspective, I note that the later derives from the US Core IG which has an extensible binding from Vaccine Code to the US Core Vaccine Administered Value Set (CVX)( http://hl7.org/fhir/us/core/ValueSet/us-core-vaccines-cvx.
The IPS Profile has a preferred binding from Vaccine Code to the http://hl7.org/fhir/uv/ips/ValueSet/vaccines-gps-uv-ips Value Set which contains SNOMED CT concepts.
Jose Costa Teixeira (Jan 21 2021 at 03:38):
An interesting case noted in the IHE profile is that the type of protocol depends on the (brand) product code, not on the 'covid-19' code. Pfizer has one protocol, J&J another. This reinforces the separation we made between vaccine and product
Christof Gessner (Jan 23 2021 at 14:00):
Where can I find the discussions on http://build.fhir.org/ig/dvci/vaccine-credential-ig/branches/main/StructureDefinition-vaccine-credential-immunization.html ? Especially: why put dose numbers in Immunization.protocolApplied - instead of using ImmunizationRecommendation for that purpose?
Christof Gessner (Jan 23 2021 at 14:03):
For considerations on implementing this with regards to specific European regulations and requirements, I started a topic https://chat.fhir.org/#narrow/stream/242580-europe/topic/vaccination.20certificates
Jose Costa Teixeira (Jan 23 2021 at 22:52):
Christof Gessner said:
Where can I find the discussions on http://build.fhir.org/ig/dvci/vaccine-credential-ig/branches/main/StructureDefinition-vaccine-credential-immunization.html ? Especially: why put dose numbers in Immunization.protocolApplied - instead of using ImmunizationRecommendation for that purpose?
We plan to do the same in Belgium (using protocolApplied). One thing is to say "this is dose 1/2", another thing is saying "the patient should be immunized". Perhaps I do not really grasp the ImmunizationRecommendation, but I am not using it.
Jose Costa Teixeira (Jan 23 2021 at 22:55):
Another thing I miss - and now it's more evident - Medication resource should have a Class. It is a normal element, so I don't know why it didn't make it to core yet (which requires abusing other attributes like MedRequest.category). We could link to MedicationKnowledge, but we can't because there is no link from a Medication to a MedicationKnowledge (like Device has to DeviceDefinition)
Jose Costa Teixeira (Jan 23 2021 at 22:57):
(maybe @Jean Duteau has a workaround for the above)
Jose Costa Teixeira (Jan 23 2021 at 22:58):
Question on Covid, following something I saw on the news today:
What is the resource to say "Doctor X Says that this patient has a legitimate reason not to be vaccinated"?
Christof Gessner (Jan 24 2021 at 00:15):
@Jose Costa Teixeira ImmunizationRecommendation uses an example value set that includes the value "contraindication", if I remember correctly.
Jean Duteau (Jan 24 2021 at 00:25):
Jose Costa Teixeira said:
Another thing I miss - and now it's more evident - Medication resource should have a Class. It is a normal element, so I don't know why it didn't make it to core yet (which requires abusing other attributes like MedRequest.category). We could link to MedicationKnowledge, but we can't because there is no link from a Medication to a MedicationKnowledge (like Device has to DeviceDefinition)
Medication resource does not need a class. It is not a "normal" element that is needed to identify the specific medication being ordered/dispensed/administered.
Christof Gessner (Jan 24 2021 at 00:33):
@Jose Costa Teixeira http://build.fhir.org/valueset-immunization-recommendation-status.html
Jose Costa Teixeira (Jan 24 2021 at 00:50):
@Christof Gessner that is one option - although what it means is, I believe, "patient was not vaccinated because" - where I think we're after "patient is exempt/ cannot be vaccinated because"
Jose Costa Teixeira (Jan 24 2021 at 00:52):
@Jean Duteau why do you say it is not needed and it is not normal? And it is not needed to identify a medication, it is to indicate it class/category. Medication category is very much needed and normal.
Jose Costa Teixeira (Jan 24 2021 at 00:52):
Class: "mRNA" / "weakened virus" ...
Jean Duteau (Jan 24 2021 at 00:53):
a) because the scope and purpose of the Medication resource is to identify the medication that has been ordered/dispensed/administered
b) because we have lots of implementer experience with this resource and they haven't been asking for it.
Jose Costa Teixeira (Jan 24 2021 at 00:54):
Ok, you hadn't found this requirement yet. I'm presenting the requirement
Jose Costa Teixeira (Jan 24 2021 at 00:55):
The answer should be to evaluate the requirement (unless there is a reason to dismiss it)
Jean Duteau (Jan 24 2021 at 00:55):
sorry, it is clearly out of the scope of the resource. you yourself said it is not needed for identifying a medication
Christof Gessner (Jan 24 2021 at 00:56):
@Jose Costa Teixeira could be a result of ImmunizationRecommendation, which (ideally) considers the overall situation of the patient.
Jean Duteau (Jan 24 2021 at 00:57):
you are acting as if you haven't raised this requirement before. it has been raised and the committee has consistently rejected it as being out of scope of the resource.
Jose Costa Teixeira (Jan 24 2021 at 00:57):
Well, this seems not very constructive. Atc codes are not identifiers of medication. Manufacturer is not an identifier of a medication. Amount are not identifiers of a medication. Still they are all used in identifying a medication
Jose Costa Teixeira (Jan 24 2021 at 00:59):
I didn't see a discussion besides a zulip chat without much context. I don't recall any other discussion - I may be wrong
Jean Duteau (Jan 24 2021 at 00:59):
manufacturer is needed if you are identifying a specific brand of medication
amount is needed to indicate a specific package type of medication
Jose Costa Teixeira (Jan 24 2021 at 01:00):
Manufacturer, like class, is not an identifier of a medication
Jose Costa Teixeira (Jan 24 2021 at 01:01):
And in some projects (epSOS, I think), the product class is actually used as an identifier - which it technically isn't. But it is normal (to my chagrin)
Jean Duteau (Jan 24 2021 at 01:01):
yes it is. If I am telling you that it has to be the Amoxicilin manufactured by Sanis Health
Christof Gessner (Jan 24 2021 at 01:01):
Ideally, in the use case of checking vaccination status, you might want to trust the result of ImmunzationRecommendation. And not review the various Immunization resources.
Jose Costa Teixeira (Jan 24 2021 at 01:02):
It is not an identifier. It is used in specifying a medicine
Jean Duteau (Jan 24 2021 at 01:02):
i never said it was an identifier. I said that is was used to identify (along with other attributes of the resource) a medication
Jose Costa Teixeira (Jan 24 2021 at 01:03):
Yes. Like class (atc)
Christof Gessner (Jan 24 2021 at 01:03):
(are you still focused on the topic ??)
Jose Costa Teixeira (Jan 24 2021 at 01:03):
That is the point.
Jose Costa Teixeira (Jan 24 2021 at 01:05):
Sorry @Christof Gessner I need to go there after this because there are big projects that I think require an attribute in Medication but it seemed the resource group is not willing to accept that feedback - so I just need to see how to approach it
Christof Gessner (Jan 24 2021 at 01:05):
start a new topic, maybe?!
Jose Costa Teixeira (Jan 24 2021 at 01:07):
So, @Jean Duteau atc and other categories are common. I argue this is needed for example for COVID (where I found this need for the Nth time). If the answer is "others we looked at don't need it" - I think nothing like a global interoperability challenge to maybe allow us to have such discussions.
Jose Costa Teixeira (Jan 24 2021 at 01:09):
Sorry @Christof Gessner . I don't think ImmunizationRecommendation is mature and well defined in meaning, so I am perhaps biased in trying not to force it. It could well work as "here's a recommendation for the patient NOT to be vaccinated because she is pregnant"
Jose Costa Teixeira (Jan 24 2021 at 01:12):
But is that the right way? Do we need that affirmation? I think "patient is immune" is a Condition. Do we want a "recommendation not to vaccinate because the patient is immune?"
Christof Gessner (Jan 24 2021 at 01:15):
In the currently discussed "vaccination certificate" this may be the relevant resource, summarizing the situation, without disclosing all the details that led to this recommendation?
Peter Jordan (Jan 24 2021 at 02:15):
Jose Costa Teixeira said:
But is that the right way? Do we need that affirmation? I think "patient is immune" is a Condition. Do we want a "recommendation not to vaccinate because the patient is immune?"
I would have thought that 'patient is immune' is an Observation based on a Diagnostic Report. Is that correct @Russell Leftwich? Certainly, I've never seen 'immune to influenza' on my health record after my annual 'flu jab and that may be an insightful comparison if the coronavirus mutates as rapidly and significantly as the flu virus. Immunity to COVID-19 is likely to be relative to strains, vaccination timelines and infection history.
Lloyd McKenzie (Jan 24 2021 at 02:52):
class can be conveyed as a coding within Medication.code. You're free to represent the medication at whatever of granularity you need. There is no need for a separate element for each 'degree' of granularity that might be represented (because there's probably 12+ different levels somewhere, sometime might want).
Jose Costa Teixeira (Jan 24 2021 at 10:52):
Peter Jordan said:
Jose Costa Teixeira said:
But is that the right way? Do we need that affirmation? I think "patient is immune" is a Condition. Do we want a "recommendation not to vaccinate because the patient is immune?"
I would have thought that 'patient is immune' is an Observation based on a Diagnostic Report. Is that correct Russell Leftwich? Certainly, I've never seen 'immune to influenza' on my health record after my annual 'flu jab and that may be an insightful comparison if the coronavirus mutates as rapidly and significantly as the flu virus. Immunity to COVID-19 is likely to be relative to strains, vaccination timelines and infection history.
I think I recall some discussions. From my perspective, there could be an observation that says "patient has been found with antibodies, hence immune", but IMO "Patient is found to be immune for X months" is an assertion of a Condition. This is also the result of a clinical finding...
I'm looking not at the entire process, but a final assertion that can be used to say "this one can enter (...) because they are immune".
Jose Costa Teixeira (Jan 24 2021 at 10:55):
I think Christof's idea of simply using a immunization(recommendation) with a forecastStatus = "immune" could be one way.
But in that case, ImmunizationEvaluation would be a better fit.
Jose Costa Teixeira (Jan 24 2021 at 10:56):
this is where I see resource overlaps, and I personally do think these cases are essential in maturing our standards or our documentation.
Jose Costa Teixeira (Jan 24 2021 at 10:59):
Lloyd McKenzie said:
class can be conveyed as a coding within Medication.code. You're free to represent the medication at whatever of granularity you need. There is no need for a separate element for each 'degree' of granularity that might be represented (because there's probably 12+ different levels somewhere, sometime might want).
Ok if we get to this point. I think a Classification is not a different level of product code. It does not convey any specific notion of product. You can have a catalog of products with characteristics, but you can't have a catalog of ATCs. I recall this and I don't agree they are the same thing. Personally, I'd use an extension.
Jose Costa Teixeira (Jan 24 2021 at 11:02):
An extension of the medication, I mean.
The fact that we have medication classification as an attribute not of the medication, but of the request and the dispense, is in my opinion incorrect and not justified. The use cases seem to be "in a prescription you need to convey the class of a medication" - and "let's just add it to the prescription, not to the medication" is strange. I think our agility requires the ability to step back on a few issues.
Jose Costa Teixeira (Jan 24 2021 at 11:02):
Anyway, to summarize:
Jose Costa Teixeira (Jan 24 2021 at 11:11):
We may need to codes to say
- "Patients that have a vaccine against COVID-19 (ATC J07BX03)",
- "This specific vaccine is of type mRNA"
- "This patient had COMIRNATY (4989539) as a vaccine, not Moderna (5426867)"
Christof Gessner (Jan 24 2021 at 11:13):
ImmunizationEvaluation refers to only one specific Immunization, indicating things like recall, broken coolingchain etc. that may be observed at a later point in time.
Jose Costa Teixeira (Jan 24 2021 at 11:14):
I think we need to handle the codes well (and COVID is bringing forth the need for univocal identification and description of products across borders, which is something we've been working on, hopefully the issues deserve more attention).
btw, @Christof Gessner this is interesting to see some of these codes. https://www.e-impfpass.gv.at/fileadmin/user_upload/e-impfpass/210111_e-Impfpass_COVID-19_PZN_Infosheet.pdf
Jose Costa Teixeira (Jan 24 2021 at 11:17):
Christof Gessner said:
ImmunizationEvaluation refers to only one specific Immunization, indicating things like recall, broken coolingchain etc. that may be observed at a later point in time.
Ok. From the structure it seemed it could be otherwise (and I think it should). We should ignore this (and I will launch discussion on that in a separate thread)
Russell Leftwich (Jan 24 2021 at 14:21):
Immunity for an individual has not been defined clinically. There can be no valid statement "the patient is immune". What has been shown in vaccine trials is that the individuals who were vaccinated had a 95% decrease (Moderna and Pfizer) in their chance of contracting severe COVID-19. That's what 95% effective means. A recent observational study of thousands showed that those who had SARS-CoV-2 antibodies on an initial sample had a 0.3% chance of a subsequent infection while those who did not have antibodies in the initial sample had a 3% chance of subsequent infection. So, evidence that natural antibodies protect, but it says nothing about what level of antibody would be needed to label an individual "immune". Patient is immune, I would agree would be an observation, but there is no clinically valid observation of immunity. The only observation that is valid at this time is "patient is vaccinated" or "patient had documented COVID-19". It is possible that a level of a particular antibody (different tests measure different antibodies) will be associated with a high probability of "likely immunity". But that determination is not possible at present.
Peter Jordan (Jan 24 2021 at 19:55):
Many thanks @Russell Leftwich . Please note @Jose Costa Teixeira and @Christof Gessner .
Peter Jordan (Jan 24 2021 at 20:11):
Jose Costa Teixeira said:
We may need to codes to say
- "Patients that have a vaccine against COVID-19 (ATC J07BX03)",
- "This specific vaccine is of type mRNA"
- "This patient had COMIRNATY (4989539) as a vaccine, not Moderna (5426867)"
New SNOMED-CT Codes for what you term the "specific vaccine" in the 20210131 version of the International Edition...
1119349007 | COVID-19 mRNA vaccine |
1119305005 | COVID-19 antigen vaccine |
Are the 3rd pair of codes from NCBI (National Center for Biotechnology Information)?
Jean Duteau (Jan 24 2021 at 21:45):
Jose Costa Teixeira said:
We may need to codes to say
- "Patients that have a vaccine against COVID-19 (ATC J07BX03)",
- "This specific vaccine is of type mRNA"
- "This patient had COMIRNATY (4989539) as a vaccine, not Moderna (5426867)"
you would use the Medicataion.code for #2 to say "I'm ordering a mRNA vaccine" and had a code like SNOMED's 1119349007 | COVID-19 mRNA vaccine |
1119305005 | COVID-19 antigen vaccine |
Jose Costa Teixeira (Jan 24 2021 at 21:56):
Peter Jordan said:
Are the 3rd pair of codes from NCBI (National Center for Biotechnology Information)?
No, just some national codes I found - which are those that matter in national records.
Jose Costa Teixeira (Jan 24 2021 at 21:58):
Jean Duteau said:
Jose Costa Teixeira said:
We may need to codes to say
- "Patients that have a vaccine against COVID-19 (ATC J07BX03)",
- "This specific vaccine is of type mRNA"
- "This patient had COMIRNATY (4989539) as a vaccine, not Moderna (5426867)"
you would use the Medicataion.code for #2 to say "I'm ordering a mRNA vaccine" and had a code like SNOMED's 1119349007 | COVID-19 mRNA vaccine |
1119305005 | COVID-19 antigen vaccine |
I'm just expressing the 3 different concepts. I understand that this is not an easy discussion. IMO two of them are classification, not product identifiers.
Jose Costa Teixeira (Jan 24 2021 at 22:00):
if medication.code is "any other code concerning the medication", this seems inconsistent, but I don't insist on that.
Jean Duteau (Jan 24 2021 at 22:01):
no, medication.code is "any code that identifies this medication". if you are saying "I've ordered a mRNA Covid-19 vaccine", then the code for "mRNA Covid-19 vaccine" is what you put there.
Jean Duteau (Jan 24 2021 at 22:03):
if you're trying to say "i'm ordering this medication with code X to vaccinate against Covid-19", you wouldn't put the "Covid-19" in the medication.code. You would put that on the MedicationRequest as the reason why you are ordering the medication. (I think that was your #1 statement)
Jose Costa Teixeira (Jan 24 2021 at 22:15):
This is not about the medication order, but I think you are making my point about immunization.code not being the same as product code
Lin Zhang (Jan 25 2021 at 00:10):
Often there would be more than one way to express the same thing.
Jean Duteau (Jan 25 2021 at 01:47):
Jose Costa Teixeira said:
This is not about the medication order, but I think you are making my point about immunization.code not being the same as product code
I'm sorry. I just assumed you were talking about Medication Order since Immunization doesn't use the Medication resource so it doesn't really matter what attributes are on Medication.
Last updated: Apr 12 2022 at 19:14 UTC
