Covid-19 vaccination

Hi, everyone

I realised I started this topic, and just disappeared. Shame on me! :crazy_face:

I think both Ianā€™s and Heatherā€™s template will work. And Ianā€™s is already in use, I suppose. My comments:

First, the COMPOSITION:

I like the new Certificate COMPOSITION, as linked to the CKM above . There are a lot of use cases for certificate: Death certificate, Birth certificate, Medical condition certificate (i.e. needing assistance), Medication certificate (i.e. bringing opioid medications with you abroad and prevent you from being arrested for drug traffic), Vaccination status, Stating you are fit to (still) drive a car, etc. Keeping certificates apart from other COMPOSITIONS is a wise thing to do, IMO. Likewise to add elements relevant to the certificate within that context is a good idea, i.e. the certificate issuer, validity, ā€¦

I donā€™t know if it should be only one general certificate COMP, and specialize them, or having more specific certificate COMPOSITIONS. Any thoughts on that?

Secondly, EVALUATION.Immunisation_summary vs ACTION.medication-vaccination:

I see that the health service provider need an ACTION to record that a specific vaccine has been given. I might demonstrate my lack of understanding, but is it the ACTON that leads to the vaccination certificate? The ACTION.medication is for recording a given medication, regardless of if it is a vaccine or another product. It even says so in the Concept description.

Shouldnā€™t the ACTION populate a persistent archetype which keeps track of which number in a (potential) sequence of jabs, and when the next vaccination is due? This is what the EVAL.Immunisation_summary does.

It would be nice to agree upon how to deal with vaccinations within the community. I fear that implementations are deviating, and will cause trouble later.

4 Likes

Hi VebjĆørn,
Thank you for bringing up these good questions. Late last year we tried to get a common approach to vaccinations of the ground just before covid started. But in the end everybody did his own thing with only loos correlation. I for Nedap created a quick and dirty archetype. The only semi decent pattern is it uses a ā€œcovid vaccineā€ CLUSTER that specialises CLUSTER.medication (it includes Dutch batchid valuesets and rules to set the manufacturer name based on the valueset). With my current level of understanding maybe this should have been a template cluster.
What we can learn from this is using medication.cluster can work for a vaccination usecase.

Furthermore Iā€™d like to share that Iā€™m concerned about the certificate based on EHR data. Especially if it has legal status (not just health information for other caregivers), as it does in the Netherlands and many other countries. This fundamentally changes/adds to the usecases of the EHR data. And Iā€™m concerned that it will impact the way we uses the EHR since itā€™s no longer just used to record information to the benefit of the individual patient. Please let me know if my concern resonates, Iā€™m struggling to explain itšŸ˜…
I guess we canā€™t (shouldnā€™t) be against secondary use of the EHR data, but letā€™s think carefully before adding certificates to the EHR.

Leaving that for now, semantically I think youā€™re right the process should be vaccination action ā†’ immunisation evaluation ā†’ immunisation certificate composition. Maybe itā€™s possible to generate a evaluation based on a decision logic that takes (only) the action into account. But sementicaly youā€™re interested in the immunisation status, not the vaccination action. Although the difference for covid is non existent atm for the certificate. But for other vaccinations eg hep b itā€™s common to judge the immunisation status with a blood (antigen titer) test, not just the action.

https://github.com/nedap/archetypes/blob/master/COVID/openEHR-EHR-CLUSTER.covid_vaccin.v1.38.0.adls

2 Likes

Hi Joost
Thanks for your support and views!

Excactly, thank you for wording that better than I did!

I guess that will vary between legislations. If there is a need to issue a ā€œcertificate-likeā€ statement from a clinician, and from the EHR, then we should provide a COMPOSITION for that. And if it is not suitable in a country, then donā€™t use it. Mayby the name ā€œcertificateā€ is the issue here? We can change to something different, if it helps. ā€œMedical statementā€ or similar?

I guess how and by whom issuing of COVID-19 certificate is done, is different between countries. In Norway itā€™s issued from a ā€œCore EPRā€ run by a governmental organisation, which recieves messages from different parts of the health system. Nevertheless, all data has to be captured , stored and sent from each potential health service provider. Whether this is sent using a ā€œCertificateā€ COMP, or ā€œMedical statementā€ COMP, or simply by creating a message directly from ACTION and/or OBS and/or EVAL archetypes is not important. Either way will work, AFAIK. But it will be easier to trace what has been sent if it is within a COMPOSITION. Any vendor, please correct me if Iā€™m wrong.

1 Like

And yes, I do remember. Everyone suddenly got very busy! I think the next pandemic should be better planned. I suggest that it will be notified at least one year ahead, so we can prepare better :wink:

4 Likes

My concern is more about the goal and context of a certificate, e.g. a covid vaccination certificate is used to allow e.g. entrance to restaurants (in the Netherlands) and should be issued by a (government) authority. Stuff like chain of evidence, verifiability, trust in the authority etc is very important in that context. And moreover itā€™s very black and white. Itā€™s either valid or invalid, while most of the info in the EHR is on a much more subtle scale between true and not true.
The goal of the EHR is to record information used to care for an individual. And Iā€™m concerned that the requirements for a legal piece of evidence will interfere with that main goal of the EHR.

Ho Joost,

Thatā€™s a perfectly reasonable cultural perspective, but AFAIK, in the UK the source of truth is ultimately the original vaccination record or Covid test record, which would , at least in the UK, definitely be stored in the EHR (may be cached elsewhere).

And it 's not just vaccination info that is ā€˜black and whiteā€™, at least in terms of decision support. As an example UK Covid shielding/ vaccination prioritisation was very largely based on coded diagnosis entries in GP systems.

Iā€™m not against an approach that has a specific certification composition but in other contexts in can imagine that this may be ā€˜assembledā€™ dynamically from the core EHR records, albeit via s specific service which ā€˜assuresā€™ the certification.

2 Likes

Hi Joost,
Iā€™ve been down with a airway infection some days, havenā€™t seen your reply until now. Sorry. Regarding:

Yes, you are of course right, with the primary goal of the EHR. However, information from the EHR serves, or can serve, more purposes, Secondary use, i.e. local or national registries, de- or pseudoanonymized data for research, and - in some countries for different types of ā€œcertificatesā€. What the owner of the data actually uses the data for, is up to local regulations and implementations to decide. I really donā€™t see the problem of facilitating compositions for certificate-like usecases.

1 Like

Yep. Data of both vaccination and test results are ā€œbornā€ in various EHR-systems in Norway as well, and fed to a central hub (ā€œCore EHRā€), and the certificate is issued by it. I guess it will be something similar in other countries as well.

I think this COMPOSITION archetype has a potential for broad re-use in all sorts of certificates, for example death certificates. My only real question about it is about the ā€˜Vaccinated individualā€™ SLOT which I donā€™t understand the purpose of.

@siljelb it was created to facilitate a mock up (for us) , but would not be necessary in real world usage in the CDR.

I have been curious, about how Countries that are adopting the EC Digital Covid certificate, have been capturing ā€œimmunity dataā€

Is it a combination of the positive covid test and number of days without symptoms? or number of days past the positive covid test? How do we account for patients who were presumed positive without a test, maybe based on symptoms?

Interested to see how this is being done

Keisha

1 Like

Hi Keisha,

Weā€™ve not implemented the immunity data. But my initial thought are around using lab result archetypes, sign/symptom, and problem/diagnosis.
Afaik patients who are presumed to have recovered, without a positive pcr, are not eligible for the certificate:

  • Recovered persons, holding an EU Digital COVID Certificate should be exempt from travel-related testing or quarantine during the first 180 days after a positive PCR test.

But for modelling it wouldnā€™t be a problem, you could probably use the problem/diagnosis archetype based on the clinical decision alone.
You could also have a look at the covid templates on CKM for inspiration, or maybe you know them well?

Hope this helps.

1 Like

@heidi.koikkalainen any update on your work? We discussed my design ideas in person. It would be nice to share them here, otherwise I would have to built them just to show my ideas. (My current design is a terrible model, due to haste and technical limitations)

Iā€™d strongly be in favour of storing those certificates outside of the EHR/CDR. Because the goal of the EHR is so different: legal prove vs information to take care of a patient. Iā€™m really worried about the requirements around chains of evidence and authorisations creeping into openEHR/ the CDR. And mixing up our basic design principles and assumptions. Iā€™m struggling to come up with good examples. Maybe @thomas.beale shares my concern and has some examples.

A related example is in my country physical checks for drivers license can never be performed by a doctor that has a normal treatment relationship with a patient, for it will mix up the incentives.

Itā€™s off course fine to share clinical data as input for a certificate, but the decision logic and attributes of a certificate should stay out of the CDR.

1 Like

Thank you for your response.

I will definitely take a look at the covid templates on the CKM, I would not say iā€™m very familiar with them.

1 Like

Hi Joost,
I built a Vaccine Record template for a Finnish use case taking the same approach as NDS in Scotland and DIPS in Norway. Itā€™s based on the ACTION.medication.v1 archetype with some international CLUSTER archetypes and local extensions nested in. The scope of it is to record details of vaccination administration so it doesnā€™t include immunity data. However, Iā€™m not sure at the moment whether the template will actually be implemented as the client might choose to take a more localised approach. Iā€™ll keep you updated! Btw, I actually havenā€™t looked at your models yet as I couldnā€™t figure out how to import ADL2 archetypes to AD :sweat_smile:

3 Likes

So iā€™m unclear, is protocol is something that we would populate in these kinds of templates?

The Scottish template is at Clinical Knowledge Manager and carries the Country, AppointmentID, vaccination Centre and Target disease in a CLUSTER inside a slot in protocol