FHIR Chat · Supporting LOINC Answer Lists?

Stream: IG creation

Topic: Supporting LOINC Answer Lists?

May Terry (Jul 15 2018 at 22:58):

Working closer with CIMI vocab, we're recognizing a legitimate need for the FHIR IG Publisher dependent terminology server to support LOINC answer lists (LL- and LA- numbers). We suspect it is currently not supported based on the unrecognized value set errors we're receiving. 1) can we confirm this is the case? and 2) How would I go about requesting an enhancement to support this, @Grahame Grieve ? cc: @Susan Matney and @Mark Kramer

Grahame Grieve (Jul 16 2018 at 00:22):

LL and LA codes are supported, though I need to update the Loinc version supported. I'll let you know when this is done

Rob Hausam (Jul 16 2018 at 03:31):

There are a few filename and other changes and new content (like LOINC groups) in 2.64.

Grahame Grieve (Jul 16 2018 at 06:19):

I won't be processing any changes I don't have to until the ballot is published

May Terry (Jul 16 2018 at 12:45):

@Grahame Grieve - is there an example for displaying the LA codes related to the LL in a request?

Grahame Grieve (Jul 16 2018 at 12:57):

well, there's implicit LL value sets - see http://hl7.org/fhir/loinc.html

Chris Moesel (Jul 17 2018 at 18:00):

I think the above conversation is about answer lists (e.g. LL12345-0), which seem to work if we reference the value set with a URI of the form http://loinc.org/vs/LL12345-0. That's good!

I can't seem to successfully include loinc answers (e.g., LA98765-0) in a value set definition however. For example, our shr-oncology-TubuleFormationScoreVS value set is defined with this compose section:

"compose": {
  "include": [
    {
      "system": "http://loinc.org",
      "concept": [
        {
          "code": "LA27216-3",
          "display": "Score 1: >75% of tumor area forming glandular/tubular structures"
        },
        {
          "code": "LA27217-1",
          "display": "Score 2: 10% to 75% of tumor area forming glandular/tubular structures"
        },
        {
          "code": "LA27218-9",
          "display": "Score 3: <10% of tumor area forming glandular/tubular structures"
        }
      ]
    }
  ]
}

This results in a NullPointerException (NPE) on its page in the IG:
la_expansion_npe.png

This NPE, however, doesn't show up in the log when we run the IG publisher:

process res: shr-oncology-StainingIntensityVS                                    (19.0728sec)
process res: shr-oncology-TubuleFormationScoreVS                                 (19.0731sec)
process res: shr-core-YesNoUnknownVS                                             (19.0735sec)

Nor does it show up in the QA:
tubule_vs_qa.png

So... there's not much to go on except that the actual rendered page indicates there was an NPE. Can you confirm that what we're doing should be supported?

Chris Moesel (Jul 18 2018 at 14:34):

@Grahame Grieve -- can you comment on the above? Is it valid for us to create new value sets that contain arbitrary LOINC answer (e.g. LA) codes? And if so, does that compose rule look correct?

Grahame Grieve (Jul 18 2018 at 19:45):

it looks right to me

Chris Moesel (Jul 18 2018 at 20:46):

Thanks, @Grahame Grieve. Then I think there may be a bug in the tooling or the terminology server that is preventing those value sets from being expanded. :-(

Grahame Grieve (Jul 18 2018 at 20:48):

yes indeed. It's on my list. but the list is pretty long right now

Chris Moesel (Jul 18 2018 at 20:53):

I understand. Thanks, Grahame.

Chris Moesel (Jul 18 2018 at 23:09):

I've got good news and bad news.

I don't know if you did something @Grahame Grieve, but I just did another run with the latest jar and I'm getting value sets with LA codes expanded! That's great!

Chris Moesel (Jul 18 2018 at 23:11):

The bad news is that for some reason, the whole list of codes gets repeated in the expansion. See the screenshot below:
la_expansion.png

Chris Moesel (Aug 15 2018 at 04:03):

It looks like some of the LOINC answerlist implicit value sets stuff isn't validating. I'm getting errors like the following in my qa.html:

pasted image

My SD contains this:

"binding": {
  "strength": "preferred",
  "valueSetReference": {
    "reference": "http://loinc.org/vs/LL240-3"
  }
}

It is a valid answerlist (here) and if I go to http://loinc.org/vs/LL240-3 (and type in my credentials) it resolves.

I'm seeing this for three answer lists: LL240-3, LL748-5, and LL4350-6.

Grahame Grieve (Aug 15 2018 at 04:06):

it's committed?

Chris Moesel (Aug 15 2018 at 04:07):

Uh. Not sure. I think it's been around a while, but I'll check.

Chris Moesel (Aug 15 2018 at 04:12):

It is, but let me trigger a rebuild so the qa and published results are using the latest publisher (it's been a while since last commit).

Chris Moesel (Aug 15 2018 at 04:22):

Alright. The build went through. Here are the relevant links:

Note that in the rendered profile, the answer list looks like it was resolved. So it seems like things are working despite the reported errors. That being the case, prioritize this as you wish!

Chris Moesel (Aug 16 2018 at 05:41):

FYI -- this seems to be fixed in the latest builds. Thank you!

Last updated: Apr 12 2022 at 19:14 UTC

Main menu

FHIR Chat · Supporting LOINC Answer Lists? · IG creation