[v3.0] Mapping DiseaseType for combo tests

[MichaelJRussell]

While reviewing how mapping the new disease codes to real-life data from the past year, there's one issue I want to raise to get consensus on the expectation for how this should be handled. Right off the bat, I want to be clear that this represents only a very small percentage of all CVIs.

The issue is with tests that include multiple target diseases. Two examples:

• Species: multiple, Test: Brucella abortus / Brucella suis
• Species: Swine, Test: Brucellosis / Pseudorabies

The two obvious options seem to be:

• Create two tests in the XML
• Use the DiseaseOther element

Also, I hadn't thought about this when looking at the original disease list, but regarding brucellosis, do we only care if a specific bacteria strain (abortis, canis, ovis, suis) is identified in the test name? It's looking like we'll have many "Brucellosis" tests going to DiseaseOther, unless we make assumptions on which strain based on other data from the CVI (can this be determined by species?). Happy to open a new issue if that needs separate discussion.

[ryanscholzdvm]

This is somewhat of a continuation of #49. That issue was included in version 3.0, but is being linked here to provide historical context.

[mkm1879]

In the draft "Easier guide" I try to sell the case that we don't really need precise test methodology because we have the accession information with which to query the lab if we need more detail. I think it might help to think of this element as saying "Has been tested for X" rather than "Has had text X performed.." I think that would favor your first option.

[ryanscholzdvm]

I would tend to agree with @mkm1879. From the SAHO standpoint, I'm not sure that we really care HOW an animal has been tested, but rather that they have been tested and are negative for the given disease.

To @MichaelJRussell's original comment, we could potentially change the structure of the test element to allow multiple diseases to be declared in a single test. If you perform a respiratory panel test on a chicken (at our NAHLN lab that covers AI, MG, MS, and vND), then you could potentially list it as a single test/accession, but then list all of those diseases under the test.

[ryanscholzdvm]

Here is the current disease list. Please review an offer suggestions for additional diseases and/or diseases that should be changed. https://github.com/AAVLD-USAHA-ITStandards/eCVI/blob/5ff49f08f3bce139c810997642f887d01443e6b8/ecvi2.xsd#L1616-L1629

[ryanscholzdvm]

The above proposed commit to the PendingChanges branch would allow multiple diseases to be declared in a single test. As I was drafting it though, I realized that it might have some unintended consequences that should probably be discussed further and clarified.

As the XML standard is written, the Accession information occurs separately from the test information (accession is in the main document section, test info is specific to each animal). In my example above using a panel test, it would still probably be more appropriate to list a multiple tests, all associated with a single accession because a separate results is being provided for each disease that is part of the test panel. The utilization of a single test with multiple diseases is likely less common scenario where a test provides a single result for multiple diseases. I think that in @MichaelJRussell's original post, the scenario of B. suis and B. abortus may be one of these scenarios where there are tests that truly test for both diseases, but only provide a single result. In the case of the brucellosis/pseudorabies, I would suspect that most tests are likely actually providing separate results, even though they are a single test.

Example of single accession with two results (this is already possible with current XML standard): <Test AccessionRef="Acc1"> <Result ResultName="RESULT"> <ResultString>Negative</ResultString> </Result> <DiseaseCode Code="Brucella suis" /> </Test> <Test AccessionRef="Acc1"> <Result ResultName="RESULT"> <ResultString>Negative</ResultString> </Result> <DiseaseCode Code="Pseudorabies" /> </Test>

Example of single accession for test of two diseases with single result (this would be made possible with above proposed commit): <Test AccessionRef="Acc1"> <Result ResultName="RESULT"> <ResultString>Negative</ResultString> </Result> <DiseaseCode Code="Brucella suis" /> <DiseaseCode Code="Brucella abortus" /> </Test>

[mkm1879]

I think you've hit on an important distinction. This comes up in LOINC coding tests. LOINC codes actually identify observations or in the terms we are using here, results. It was sometimes "pulling teeth" to get the lab folks to be explicit about exactly which species, serotypes, etc. a test would pick up. As your B. suis test here, would also pick up B. abortus. Perhaps with this example, we should use Brucella spp. because we don't want to imply that we wouldn't pick up B. melitensis if that were present. But is this level of detail important in the CVI? Maybe just "Brucellosis." Are there use cases for one test result for multiple agents that don't summarize to genus, etc.?

[greenvet2]

@mkm1879 .... I can see how maybe at the CVI level, "Brucellosis" is sufficient to say that we test and it's Brucellosis negative. BUT, if it's sent with just the result of "Brucella spp.", is that going to potentially be interpreted as "negative for all species of Brucella"? If the test truly JUST tests for B. suis and B.abortus, then it would be more accurate to report those two DiseaseCodes, rather than the more general. If the test WOULD pick up all of them, and we just only happen to care about those two in this instance, then the broader would be fine.

[mkm1879]

It is overly specific for this discussion, but I believe most of the screening level Brucellosis tests are less discriminating than we think and would likely react with any but may not have been validated for them. So how to interpret? That's why I think we should rely on lab-specific sources if we need to be sure.

[mmcgrath]

Agree with making the DiseaseCode unbounded for the Jan 2025 release