FHIR Chat · ObservationDefinition discrete value references based on age · implementers

Stream: implementers

Topic: ObservationDefinition discrete value references based on age

view this post on Zulip Sergey Gerasimov (Mar 12 2020 at 13:27):

How to specify normality or abnormality of discrete value in ObservationDefinition based on the age/gender of patients?
For continuous values we have ObservationDefinition.qualifiedInterval where we can define for what age/gender this can be applied, but there is no such possibility for discrete values.
Examples, where it can be used, include:

  1. Babkin reflex (normal when age < 3 months, abnormal in other cases)
  2. The presence of secondary sexual characteristics in early age is abnormal.
  3. Teeth Eruption schedule is characterized not only by the number of teeth. Different types of teeth erupt in different ages.
  4. The presence of Patent ductus arteriosus (PDA) (in age older than 8 weeks is abnormal)

view this post on Zulip Sergey Gerasimov (Mar 17 2020 at 13:08):

@Grahame Grieve, could you say who is working on ObservationDefinition resource now?

view this post on Zulip Lloyd McKenzie (Mar 17 2020 at 13:55):

@Eric Haas

view this post on Zulip Eric Haas (Mar 17 2020 at 22:16):

first to be clear qualifiedInterval applies to "continuous" (decimal) measures as well as to "ordinal" measures (aka discrete ordered values - numbers, letters or any ordered discrete list.)

view this post on Zulip Eric Haas (Mar 17 2020 at 22:17):

so if your measure results are a discrete list then proceed as with continuous measures...

view this post on Zulip Eric Haas (Mar 17 2020 at 22:18):

I assume however you have nominative list of results ( ie a valueset ) of normal and abnormal values.

view this post on Zulip Eric Haas (Mar 17 2020 at 22:22):

You could define the ObervationDefinition.subject to reference a Group which defines a set of possible patients and define several ObservationDefinitions by target Group. ( e.g. patients under <3months vs patients > 3 months)

view this post on Zulip Eric Haas (Mar 17 2020 at 22:29):

I don't think that is very satisfying solution however to have quantitative measures be able to define these characteristics inline and the others using defined via group. I personally would rather have the the qualifying elements apply to both ranges and valuesets so a single OD can be used for all populations for a measurement. I will raise an issue. Meanwhile your only real option right now is to use multiple ODs , one for each population.

view this post on Zulip Eric Haas (Mar 17 2020 at 22:53):

https://jira.hl7.org/browse/FHIR-26603

view this post on Zulip Sergey Gerasimov (Mar 18 2020 at 16:35):

Thank you!
Your proposal looks very good :)
What is the chance that this will be in FHIR R5?

view this post on Zulip Lloyd McKenzie (Mar 18 2020 at 17:04):

Monitor the Jira item. If it gets approved, chances are very good.

view this post on Zulip Sergey Gerasimov (Mar 18 2020 at 17:14):

Thank you

view this post on Zulip François Macary (Sep 24 2020 at 17:03):

I just stepped on this ticket and commented it - @Eric Haas and @Sergey Gerasimov
A more straightforward but with more impact on the resource would be:
place the 4 elements validCodedValueSet, normalCodedValueSet, abnormalCodedValueSet, criticalCodedValueSet under a common backbone "qualifiedValueSets" conveying the same criterion as qualifiedInterval: {context, appliesTo, gender, age, gestationalAge, condition}

Like this:

qualifiedValueSets

  • context
  • appliesTo
  • gender
  • age
  • gestationalAge
  • condition
  • validCodedValueSet
  • normalCodedValueSet
  • abnormalCodedValueSet
  • criticalCodedValueSet

Last updated: Apr 12 2022 at 19:14 UTC


An HL7 Website -