FHIR Chat · A&E triage/prioritization · implementers

Stream: implementers

Topic: A&E triage/prioritization


view this post on Zulip Martin Grundberg (Nov 13 2019 at 19:39):

We have a use case to be able to from an ambulance post a triage/triage outcome so that it is visible at the A&E before the patient comes in.

It can have several observations in it, like BP, pulse, oxygen saturation etc, potentially also a score (NEWS or similar). This will be summarized with some type of priority category (1-5, blue, green, yellow, orange, red...).

What resource is suitable to capture the full context of the triage, including the observations and the outcome priority category? Would it be a Composition? Other suggestions?

Thx!

view this post on Zulip Jens Villadsen (Nov 13 2019 at 19:40):

@Ann Merete Duedal Jensen

view this post on Zulip Ludvig Eek Hofmann (Feb 07 2020 at 11:33):

I have the same question regarding to send a Priority/Triage from a ambulance system to the A&E.

Could the resource RiskAssessment (https://www.hl7.org/fhir/riskassessment.html) be used to capture the several observation and priority?

view this post on Zulip Joe Kesselman (Feb 07 2020 at 18:31):

Could the resource RiskAssessment (https://www.hl7.org/fhir/riskassessment.html) be used to capture the several observation and priority?

My understanding (admittedly from other systems) is that risk assessment is the output of a model that evaluates other data and indicates whether this patient is at elevated risk of a condition (eg, heart attack), rather than being raw observations. And I'm not sure "priority" is or should be a concept in FHIR...

view this post on Zulip Lloyd McKenzie (Feb 07 2020 at 21:48):

Priority would typically be a characteristic of a ServiceRequest or Encounter

view this post on Zulip Brian Postlethwaite (Feb 07 2020 at 23:13):

An ambulance encounter would make sense.

view this post on Zulip Brian Postlethwaite (Feb 07 2020 at 23:13):

Will add that to the list of example scenarios PA is planning to document on the use of encounter.

view this post on Zulip Martin Grundberg (Mar 12 2020 at 15:39):

The triage is something done during an encounter, so I dont agree that the triage itself is an encounter. The stay within the ambulance, with a start and end time etc is the encounter during which several observations and assessments can be performed, one of those being one or several triages.

As there are standardized methods of triage, MTS (https://www.triagenet.net/classroom/), RETTS (https://predicare.eu/about-retts/), SATS (https://emssa.org.za/special-interest-groups/the-south-african-triage-scale-sats/) etc I feel like a something which you can represent using a template that gets filled in is suitable, so to me it could map well to a Questionnaire and QuestionnaireResponse. You can see it as completing a "triage form".

view this post on Zulip Pétur Valdimarsson (Mar 12 2020 at 16:36):

Observation resource with defined components representing the details?

view this post on Zulip Lloyd McKenzie (Mar 12 2020 at 19:46):

QuestionnaireResponse for the raw form. Observation if you want the data to be subsequently comparable/queriable

view this post on Zulip Martin Grundberg (Mar 16 2020 at 15:22):

We have done another round of reviews of different resources.

The issue with the Questionnaire and the QuestionnaireResponse is that a consumer needs to be aware and use the Questionnaire. We looked a bit at RiskAssessment, but we dont feel a triage is a risk assessment. And a risk assessment can often be completed using some type of form (ie Questionnaire), so what do you then use, RiskAssessment or Questionnaire/QuestionnaireResponse (that question we can raise for several use cases)?

We then looked at ClinicalImpression, it seems to fit the use case. There are several key factors that we can use:

Clinical Assessment metadata

  • code 0..1 CodeableConcept Kind of assessment performed ---> Triage (procedure) SCTID: 225390008
  • protocol 0..* uri Clinical Protocol followed ---> Here we can bind to triage protocol followed which is good! Eg Manchester Triage System (assessment scale) SCTID: 713009001

Triage assessment common data

  • problem 0..* Reference(Condition) Relevant impressions of patient state --> Here we can put the actual problem, might have been easier with a codeableConcept.

Repeating Investigation level

  • investigation.code 1..1 CodeableConcept A name/code for the set ---> Blood pressure (observable entity) SCTID: 75367002
  • item 0..* Reference(Observation) Record of a specific investigation--->Here we can include the actual Observation resource
    We can repeat this for all the necessary observations for one particular type of triage method.
    We can also slice the investigations needed for the triage method using the investigation.code as discriminator.

Triage outcome/patient prioritization/classification
Only question is where we would put the actual triage priority "Urgent priority", "Color Yellow" etc, but I think we could use the following:

  • prognosisCodeableConcept 0..* CodeableConcept Estimate of likely outcome Clinical Impression Prognosis (Example)
    Or maybe add some type of extension.

@Pétur Valdimarsson , as I see it we cant really use Observation as it doesnt support the overall triage assessment context. The triage assessment has several individual observations, and a prioritization/categorization of urgency. We need a resource that groups all that together. But we have gone through several resources for this use case, and changed our mind several times, so who knows where we end up :D

Maybe we should only use Questionnaire if it is important to have the form structure and logic, e.g. the enableWhen feature and similar.

view this post on Zulip Lloyd McKenzie (Mar 16 2020 at 15:40):

Questionnaire/QuestionnaireResponse is useful as an initial data capture tool (for anything). It gives tight control over user-interface and allows simple systems to capture data of any type. It's also easy to change the data collected. However, QuestionnaireResponse is not terribly useful for query/comparison/analysis. So the usual workflow is capture using QuestionnaireResponse, then extract into other resources for subsequent use.


Last updated: Apr 12 2022 at 19:14 UTC