Stream: Orders and Observation WG
Topic: Micro / Culture
Grahame Grieve (Jun 25 2018 at 14:53):
@Rob Hausam @Eric Haas : this example looks wrong to me? http://build.fhir.org/diagnosticreport-micro1.xml.html
Grahame Grieve (Jun 25 2018 at 14:55):
and the lri one has problems too :
<code>
<coding>
<system value="http://loinc.org"/>
<code value="624-7"/>
<display value="Bacteria Spt Resp Cul"/>
</coding>
<coding>
<system value="https://acme.org/codes"/>
<code value="CULTID"/>
<display value="Culture ID"/>
</coding>
<text value="Aerobic Respiratory Culture, Sputum"/>
</code>
Grahame Grieve (Jun 25 2018 at 14:55):
??
Grahame Grieve (Jun 25 2018 at 15:00):
So @Alexander Ivanov has just asked me about these examples. In particular, the example given, the growth observation is not a contained resource, but it has not enough information to interpret it independently - you don't know what the organism it is about is.
Grahame Grieve (Jun 25 2018 at 15:01):
more generally, I feel as though we need to augment the lovely diagrams on the page build.fhir.org/diagnosticreport-examples.html with tables that explain how the codes work, and then make sure the example codes and structures are aligned with this
Eric Haas (Jun 25 2018 at 15:42):
Well with focus we can now reference the Observation. We had talked about this problem 2-3 years ago, surprised nobody has brought it up again til now. Can somebody make a tracker? No guarantee though that will get not get stuck in the workgroup for some reason this is one of those topics that seems to generate endless discussion. (I am sure the examples align with the lovely diagrams since I did them.)
Grahame Grieve (Jun 26 2018 at 03:23):
why is focus relevant?
Eric Haas (Jun 26 2018 at 03:43):
you could add the prior observation as the observation focus to get back the context. I have not looked at these examples in a while and probably won't get to them soon btw.
Grahame Grieve (Jun 26 2018 at 03:44):
so the focus of an observation can be another observation?
Eric Haas (Jun 26 2018 at 03:49):
focus = reference(Any)
Rob Hausam (Jun 26 2018 at 14:35):
@Grahame Grieve @Eric Haas I agree that this can still use some work (including updating the diagrams to use .hasMember instead of .related). The 'hasMember' link to the susceptibility panel ("obr") I think is in the wrong direction, and maybe 'focus' would work better for that "child-parent" linking (although it's pretty loose semantically). Eric and I have discussed and worked on all of this in the past, but not so much recently. I can take a further look at it.
Grahame Grieve (Jun 27 2018 at 04:29):
please do. In particular, the examples do not have any of the links from the diagrams that they claim to be an example for
Jay Lyle (Mar 18 2021 at 20:54):
new question: EHR surfacing resulted cultures in parasitology, virology, mycology, with multiple organisms. No AMR tests to worry about. Is that multiple tests, multiple values, multiple components? @Riki Merrick
Riki Merrick (Mar 19 2021 at 16:47):
Well that is why it is important to have the linkage between the parent result and the following tests performed on those isolates - so that would be more than 1 result to the culture and then you need to link all other tests to the essentially new specimen (now more than 1 isolate) - in v2 we do that by creating a new order that links to the result it is based on (OBR-26) and parent order (OBR-29).
Riki Merrick (Mar 19 2021 at 16:54):
I like the thought of focus and referencing the observation that has the organism name in observation.value there that Rob mentioned above - I think we need to put this on the agenda for more discussion on a call - BUT BEFORE THAT we should have some true examples.
Craig Newman (Mar 19 2021 at 17:03):
The OO on FHIR calls the last few weeks have been discussing parent/child relationships in FHIR in general and micro specifically. The minutes from those calls might be of interest. If you have additional workflows that need to be considered, we'd like to hear about them.
Rob Hausam (Mar 23 2021 at 14:02):
Even though it's really important, this topic was dormant for nearly 2 years. We do need to get it on the agenda again and actually finish working through it, and definitely if @Jay Lyle (and anyone else) can provide examples of what they are seeing, that would be very helpful.
Jay Lyle (Mar 23 2021 at 18:01):
Attaching the ER diagram for the domain; working on some instances Lab-Microbiology_3342.jpg
Jay Lyle (Apr 02 2021 at 16:43):
VistA examples attached. I tried pasting in and discovered long posts not navigable. VistAMicroExamples.txt
We have a ParasitologyReport table with potentially many child records in the Parasitology table. The ParasitologyReport table does not include a status (required in DiagnosicReport but supports 'unknown'). The OrderedTest table does not seem to include tests for this domain, which is just as well, since the cardinality doesn't seem resolvable.
In both examples, the "proposed change" is intended to address the question of how to include multiple results. These instances lacks the required component.code, but that would just be "parasitology." The other option would be to have two distinct Observations in a DiagnosticReport.
Notes on Example 1:
We have no test name, but we do know that the result is from the Parasitology table.
Hence:
"code": {
"text": "PARASITOLOGY"
}
Results are from the associated Organism table, which contains references to SNOMED organism codes as well as text results for other kinds of resulte. Hence,
"valueCodeableConcept": {
"text": "TRICHROME EXAM :NEGATIVE"
}
That test name and that result text both exist only in that text.
Notes on Example 2:
Same test code situation.
Organism results are actually encoded and will be included as encoded. The quantification (many, moderate) only occurs in the comment.
Example 1
Current:
{
"cdwId": "12345:P",
"status": "final",
"category": "laboratory",
"code": {
"text": "PARASITOLOGY"
},
"subject": {
"type": "Patient",
"reference": "1015",
"display": "PatientName"
},
"effectiveDateTime": "2009-10-07T16:24:32Z",
"issued": "2009-10-13T20:51:00Z",
"performer": [
{
"type": "Organization",
"reference": "456:I",
"display": "ANCHORAGE VA MEDICAL CENTER"
}
],
"valueCodeableConcept": {
"text": "TRICHROME EXAM :NEGATIVE"
},
"comment": "Performing Lab:QUEST DIAG.INC.2211 E.NO.LIGHTS.BLVD.ANCH.AK 99508 \rRESULT RECEIVED AND VERIFIED 10/13/09 \r",
"specimen": {
"type": "Specimen",
"reference": "800000094",
"display": "FECES"
}
}
Proposed change:
{
"cdwId": "12345:P",
"status": "final",
"category": "laboratory",
"code": {
"text": "PARASITOLOGY"
},
"subject": {
"type": "Patient",
"reference": "1015",
"display": "PatientName"
},
"effectiveDateTime": "2009-10-07T16:24:32Z",
"issued": "2009-10-13T20:51:00Z",
"performer": [
{
"type": "Organization",
"reference": "456:I",
"display": "ANCHORAGE VA MEDICAL CENTER"
}
],
"comment": "Performing Lab:QUEST DIAG.INC.2211 E.NO.LIGHTS.BLVD.ANCH.AK 99508 \rRESULT RECEIVED AND VERIFIED 10/13/09 \r",
"specimen": {
"type": "Specimen",
"reference": "800000094",
"display": "FECES"
},
"components": [
{
"valueCodeableConcept": {
"text": "CONCENTRATE EXAM:NEGATIVE"
}
},
{
"valueCodeableConcept": {
"text": "TRICHROME EXAM :NEGATIVE"
}
}
]
}
Example 2
Current:
{
"cdwId": "1234:P",
"status": "final",
"category": "laboratory",
"code": {
"text": "PARASITOLOGY"
},
"subject": {
"type": "Patient",
"reference": "1003",
"display": "PatientName"
},
"effectiveDateTime": "2009-10-02T11:30:00Z",
"issued": "2009-10-09T07:00:00Z",
"performer": [
{
"type": "Organization",
"reference": "123:I",
"display": "PALO ALTO VA MEDICAL CENTER"
}
],
"valueCodeableConcept": {
"text": "DIENTAMOEBA FRAGILIS"
},
"comment": "MANY BLASTOCYSTIS HOMINIS \rMODERATE DIENTAMOEBA FRAGILIS \rTEST PERFORMED AT:SPECIALTY LABS, 27027 TOURNEY RD., VALENCIA, CA \r",
"specimen": {
"type": "Specimen",
"reference": "800000106",
"display": "FECES"
}
}
Vassil Peytchev (Apr 02 2021 at 17:19):
At first glance, this seems to be a significant step-down from what is currently available when using HL7v2 messages... Is the lack of discrete data for micro results inherent in Vista, or is there something else that is behind this format (e.g. this is specifically for display purposes, and not for actual system-system exchange)?
Jay Lyle (Apr 05 2021 at 16:21):
Right: this is not the lab-to-VistA data; this is data stored in VistA and then moved into a a data warehouse, where the API team can get it.
Rob Hausam (Apr 06 2021 at 19:13):
So what is the difference between the two? Can we see both?
Jay Lyle (Apr 12 2021 at 16:21):
This project has no lab system representation: we're just surfacing what's in the EHR.
Jay Lyle (Oct 13 2021 at 17:54):
Still looking at this. The rule for Components seems to disqualify Component for almost anything. So it looks like separate Observations (code, e.g., = 42807-8 Parasite identified in Isolate). A hasMember panel might help out with the UI; would the parent have the same code?
Jay Lyle (Dec 02 2021 at 21:29):
Re R4 Micro example:
1. I’d expect a question “what did you find?” with answer “Staph aureus”; or “did you find Staph aureus?” with “Yes”
a. Having distinct observations for “what organism are we talking about” and “was it there or not” seems hazardous: the first one might be found independently of the second and a negative finding taken for presence. Can the Organism identification be one Observation, like this?
b. If we must have separate code/value pairs for organism and presence, this seems like a good place to use Component (“they are not separable”).
2. The AMR results also seem too independent, lacking a definitional context (this specimen, this organism). It doesn’t seem that “member” asserts a sufficiently precise or restrictive semantic context. Maybe it is intended to, but non-component Observations are supposed to be independently searchable, right?
a. Perhaps each set of organism tests should point to their respective isolates.
b. Like this
3. When we converge, we need an Micro/AMR example in R5.
4. Is this the last discussion?
Eric Haas (Dec 06 2021 at 16:26):
- When we converge, we need an Micro/AMR example in R5.
There are 2 already there - do you want an additional one ?
Andrea Pitkus, PhD, MLS(ASCP)CM, CSM (Dec 23 2021 at 03:37):
So this looks to be Quest's Ova and Parasites Order (#681 order code for the valencia lab: https://testdirectory.questdiagnostics.com/test/test-detail/681/ova-and-parasites-concentrate-and-permanent-smear?p=r&q=trichrome&cc=SEA ) Lab order information would be modeled as a FHIR Service Request with Specimen info in Specimen resource.
The order according to Quest's catalog info indicates there are two results reported each with their own result code/component and LOINC. The Trichrome smear result would indicate any parasites identified, while the concentrate would indicate the same. These results are likely a short answer/nom or sentence/narrative type field (text values).
Andrea Pitkus, PhD, MLS(ASCP)CM, CSM (Dec 23 2021 at 03:39):
You're likely missing some key "connecting info" such as susceptibility panel orders as they are often built in the order-result model in the LIS, but not reported in most EHRs so EHRs only get part of the total picture.
Andrea Pitkus, PhD, MLS(ASCP)CM, CSM (Dec 23 2021 at 04:07):
Nope. As when the clinician places the order for most microbiology (and also blood bank type and screen, and initial surgical pathology orders), tests they don't know if an organism will grow, what blood antigens or neoplasia may be identified before the laboratory testing commences. Furthermore, most microbiology testing (broadly including virology, parasitology, mycology, etc sections) starts on the specimen collected (which would be in Service request and specimen resources), but 42807-8 you'll note lists isolate which is a derived specimen so linkages of specimens/slides are needed to the parent collected.
So the simple base pattern/model for the Quest Ova and Parasites panel is Order----> result 1 Trichome stain---> result value (name of parasites identified or none seen/negative depending on how lab reports) This result is mapped to LOINC 43227-8
Order----> result 2 Concentrate result with value of (name of parasites identified or none seen/negative depending on how lab reports) this result is mapped to LOINC 10701-1
See https://testdirectory.questdiagnostics.com/test/test-detail/681/ova-and-parasites-concentrate-and-permanent-smear?p=r&q=trichrome&cc=SEA
The order could be LOINCed to 10704-5 although Quest may want to request an order panel code with results for the trichrome and concentrate methods.
Andrea Pitkus, PhD, MLS(ASCP)CM, CSM (Dec 23 2021 at 05:01):
Regarding your example (thanks for the data/examples as very helpful).,
-
Where is the specimen information upon which these results originated? What was collected, looks like child isolate specimens were derived upon which susceptibility results were obtained?
-
The LOINC for Culture, MRSA is incorrect. Would expect 13317-3 with the specimen source as an Ask at Order entry question that the clinician would be required to send to the performing lab so they could do the testing. IT's vital as MRSA is a reportable condition by law (ELR) with criteria for reporting of MRSA from sterile sites in most jurisdictions. This key info appears missing from the report.
-
Glad you have been asking about CLIA. Are the reports from the performing lab CLIA compliant (or in WA and NY, their requirements or if government lab their equivalent?)
-
Regarding your question 1. A service request for a culture is expected, linked to the specimen collected and source (all correctly encoded). Then if there's growth, there'd be an organism identified (or similar named) result with a value of the organism (encoded with SCT organism code) per PH/ELR requirements. I would not expect a yes no response.
If a MRSA pcr screening test were performed, then I'd expect LOINC https://loinc.org/35492-8/ with detected/not detected (or similar results as indicated on the IVD vendor's package insert to be CLIA compliant) is reported. (again simple lab structure or order-result-value ). -
Regarding structure, LISs where these lab data most often originate usually have separate data dictionaries for orders and results where these test terms are built and coded. Values may be in another dictionary, while organisms are usually in yet another.
-
Not sure what you mean by organisms tests pointing to their isolates. An organism is an isolate (a colony on a culture plate is isolated and identified as to which organism it is). Are you indicating tests performed on an isolate should point to it, then yes agree. A gram pos susceptibility panel order could be ordered on Org A, while a Gram neg susceptibility panel could be ordered on Org B. Results need to be kept on the respective organism from a quality and patient safety issue (and avoiding error).
If it is a MRSA screening pCR test, results reported differ from a culture in that the organism isn't "identified" per se and thus no SCT organism codes would be used (and the test is rarely performed on an isolate, but rather the collected specimen).
Last updated: Apr 12 2022 at 19:14 UTC
