FHIR Chat · Valid display for LOINC codes

Stream: terminology

Topic: Valid display for LOINC codes

Monique van Berkum (May 10 2021 at 14:00):

There are LOINC display terms being used in Gravity Questionnaire instances that are eliciting warnings in the Trifolia validator. The Questionnaire is being generated by the NLM LHC tool. For code.coding.display, the LHC tool uses the LOINC Survey Question Text - if it exists (e.g., “Do you consider yourself Hispanic/Latino?” for 56051-6). If there is no Survey Question Text in LOINC (e.g., for codes that are headers in the PRAPARE survey), the LHC tool appears to use the Fully Specified Name Component for the display (e.g., “Social and emotional health” for 93040-4:). In both cases, Trifolia QA generates a warning that appears to indicate that the display should be the LOINC Long common name or the Short name (if one exists).
Warning Examples:
o The display “Do you consider yourself Hispanic/Latino?" is not a valid display for the code {http://loinc.org}56051-6 - should be one of ["Hispanic or Latino", "Hispanic or Latino"] (from http://tx.fhir.org/r4) for 'http://loinc.org#56051-6'
o The display "Social and emotional health" is not a valid display for the code {http://loinc.org}93040-4 - should be one of ["Social and emotional health [PRAPARE]",] (from http://tx.fhir.org/r4) for 'http://loinc.org#93040-4'
Per the LOINC User Guide, Survey Question Text seems for be valid for display text. Is there a reason that that these display terms are not valid?

Lloyd McKenzie (May 10 2021 at 15:02):

@Rob Hausam @Grahame Grieve

Rob Hausam (May 10 2021 at 17:57):

@Monique van Berkum The long and short common names are considered in the tx.fhir.org server representation to be the the official LOINC 'display' terms, and in this case they are both "Hispanic or Latino". The survey question text is provided in a separate column (SURVEY_QUEST_TEXT) and it seems to be considered as something that is different from a "display term" by LOINC (although I'm sure you could argue that point), and the terminology server is reflecting that. But we should almost certainly be including the survey question text, when it exists, as a separate attribute of the LOINC concept, and currently we are not doing that. But the LOINC terminology API at fhir.loinc.org (which is the "canonical" :OINC representation in FHIR) also does not contain the survey question text and return it in a $lookup query:

https://fhir.loinc.org/CodeSystem/$lookup?system=http://loinc.org&code=56051-6

So this probably needs to be brought up directly with the LOINC team, as well, and likely HL7 and FHIR would follow their lead.

Grahame Grieve (May 10 2021 at 19:21):

this is in answerList.csv ?

Rob Hausam (May 10 2021 at 20:34):

No, it's in Loinc.csv.

Lloyd McKenzie (May 10 2021 at 20:51):

Key thing is that we want the validator to treat the 'question text' as a valid display for the code - given that that's what Questionnaire tools will typically give us.

Rob Hausam (May 10 2021 at 20:57):

According to our rules and standard practice, we would want to have LOINC agree with that (particularly since they now publish a canonical FHIR representation). It seems to me that they probably would agree, but I think we should at least check with them.

Lloyd McKenzie (May 10 2021 at 21:04):

Well, Regenstrief is publishing a questionnaire tool that does it, so I'd hope they'd agree with themselves... :)

Rob Hausam (May 10 2021 at 21:05):

Yes, you would hope so. They probably need to update their FHIR representation.

Lin Zhang (May 11 2021 at 14:37):

(deleted)

Lin Zhang (May 11 2021 at 14:38):

Generally, the question text is valid, and more importantly proper/authorative one that should be used to display.

Lin Zhang (May 11 2021 at 14:38):

For professional questionnaires and scales, I mean.

Lin Zhang (May 11 2021 at 14:38):

(deleted)

Lin Zhang (May 11 2021 at 14:38):

(deleted)

Lin Zhang (May 11 2021 at 14:40):

It's weird. Why my replies would duplicate sometimes.

Daniel Vreeman (May 17 2021 at 19:12):

Section 10.2.5 of the LOINC Users Manual addresses this pretty directly: https://loinc.org/kb/users-guide/standardized-assessment-measures/#loinc-representation There's a step-wise algorithm for understanding which field(s) should be used for an item display in an instrument (form) and this is what Regenstrief was using to spit out its Questionnaire representations...e.g (http://loinc.org/q/93025-5)

Lloyd McKenzie (May 17 2021 at 19:28):

Ok, so it sounds like when we load the LOINC code system into tx.fhir.org, we should be including the question text as alternate designations for the codes. @Rob Hausam What's the best place to capture that this needs to happen?