FHIR Chat · Clarification on immunizationRecommendation recommendations · implementers

Stream: implementers

Topic: Clarification on immunizationRecommendation recommendations

view this post on Zulip Giorgio Cangioli (Feb 16 2021 at 07:55):

I have some doubts about the usage of the recommendation(s) elements. Established that they are describing recommendations and not the events, I was wondering if they shall always describe future events; or, if they can describe the status of the recommendations at a certain point of time. (@Craig Newman )

view this post on Zulip Giorgio Cangioli (Feb 16 2021 at 07:56):

In practice, assuming to have on 2021-01-05 the following version of the instance
(as fsh fragment..)

  • recommendation[0].vaccineCode.coding[0] = $sct#1119349007 "Vaccine product containing only Severe acute respiratory syndrome coronavirus 2 messenger ribonucleic acid"
  • recommendation[0].forecastStatus = http://terminology.hl7.org/CodeSystem/immunization-recommendation-status#due
  • recommendation[0].dateCriterion[0].code = $loinc#30980-7 "Date vaccine due"
  • recommendation[0].dateCriterion[0].value = "2021-01-10"
  • recommendation[0].doseNumberPositiveInt = 1
  • recommendation[0].seriesDosesPositiveInt = 2
  • recommendation[1].forecastStatus = http://terminology.hl7.org/CodeSystem/immunization-recommendation-status#due
  • recommendation[1].dateCriterion[0].code = $loinc#30980-7 "Date vaccine due"
  • recommendation[1].dateCriterion[0].value = "2021-02-07"
  • recommendation[1].doseNumberPositiveInt = 2
  • recommendation[1].seriesDosesPositiveInt = 2

view this post on Zulip Giorgio Cangioli (Feb 16 2021 at 07:56):

on 2021-01-15, after the first vaccination, can I capture the current status as follows ?

plus optionally the reference (supportingImmunization) to the 2021-01-10 Immunization event for the details

view this post on Zulip Giorgio Cangioli (Feb 16 2021 at 07:56):

OR it shall contain only

  • recommendation[0].forecastStatus = http://terminology.hl7.org/CodeSystem/immunization-recommendation-status#due
  • recommendation[0].dateCriterion[0].code = $loinc#30980-7 "Date vaccine due"
  • recommendation[0].dateCriterion[0].value = "2021-02-07"
  • recommendation[0].doseNumberPositiveInt = 2
  • recommendation[0].seriesDosesPositiveInt = 2
    with the reference to the 2021-01-10 Immunization event

view this post on Zulip Craig Newman (Feb 16 2021 at 13:14):

My understanding of the ImmunizationRecommendation resource is that it should only deal with future recommended vaccinations and that .supportingImmunization is where you reference the patient's history of events

view this post on Zulip Craig Newman (Feb 16 2021 at 13:17):

I'll also say that I've heard people talk about "retrospective forecasts" in situations like testing a new CDS engine or figuring out why a patient was or wasn't vaccinated in the past. So sometimes the recommendation may be relative to a date in the past. So even though today is Feb 16th, it's possible (with some systems) to ask for a set of the recommendations for a patient as of Jan 1 (given the patient history up to that date). In this case the "future" recommendation may actually be in the past if they were due for something between Jan 1 and Feb 16. Hope that makes sense.

view this post on Zulip Craig Newman (Feb 16 2021 at 15:07):

I'll also add that often the forecast is just for the next dose in the series and not all future doses. The reason is that the interval between the first forecasted dose and the second forecasted dose is usually dependent on when the first dose is actually given. For COVID, you could forecast dose 1 on Jan 10 and dose 2 on Jan 31 (assuming a 3 week interval) but if dose 1 is actually given on Jan 12, then the due data on dose 2 then changes from Jan 31 to Feb 2 (because the 3 week interval still needs to be enforced). So while ImmunizationRecommendation doesn't preclude multiple forecasted doses in a single series, in practice it may not happen a lot.

view this post on Zulip Giorgio Cangioli (Feb 16 2021 at 16:36):

Craig Newman said:

I'll also add that often the forecast is just for the next dose in the series and not all future doses. <...> So while ImmunizationRecommendation doesn't preclude multiple forecasted doses in a single series, in practice it may not happen a lot.

Thanks Craig :-)

so If I capture correctly what you are saying usually one ImmunizationRecommendation includes only one recommendation and this because typically the next dose is what is managed.

In principle, if more doses are planned, more recommendations items can be included in the same instance updating that instance as needed (e.g. the date of the second dose can be updated after the administration of the first dose and the first recommendation disappear ).
..Or, the expected usage is in any case to have two distinct ImmunizationRecommendation instances one for each planned dose.

view this post on Zulip Craig Newman (Feb 17 2021 at 13:32):

A single ImmunizationRecommendation resource can contain multiple recommendations, it's just that typically they've be for different target diseases. For example, one resource could contain recommendations for both COVID and Influenza if the patient is due for both, but the COVID recommendation would only be for the first dose (assuming they haven't had any doses yet). Then once any due vaccines have been given, a new set of recommendations based on the updated patient history could be generated which would then no longer recommend Influenza (or at worst recommend it for 1 year or so in the future) and the second dose of COVID in 3 or 4 weeks. I always assumed that this new set of recommendations would be a different instance of ImmunizationRecommendations, but I don't know enough about how resources are versioned in the real world to know whether or not systems would update the instance or create a second resource.

Last updated: Apr 12 2022 at 19:14 UTC


An HL7 Website -