Stream: terminology
Topic: Infer LOINC?
Jay Lyle (Mar 09 2022 at 18:39):
We wish to expose lab tests from the EHR in FHIR. Some of these tests are recorded with LOINC test codes; some are not. For those that are not, is it permissible to infer a generic code from the source table? E.g., for an organism recorded in the Parasitology table, infer code of 42807-8 Parasite identified in Isolate?
Rob Hausam (Mar 10 2022 at 03:22):
I don't see any reason that you couldn't do that - just try to make reasonable inferences on which LOINC code to use (as I expect you would). If you have a local or other "non-LOINC" code (which it seems that you would), you should at least consider sending that, too.
Davera Gabriel (Mar 10 2022 at 13:10):
Jay Lyle said:
We wish to expose lab tests from the EHR in FHIR. Some of these tests are recorded with LOINC test codes; some are not. For those that are not, is it permissible to infer a generic code from the source table? E.g., for an organism recorded in the Parasitology table, infer code of 42807-8 Parasite identified in Isolate?
Hello @Jay Lyle there's a project underway here at Hopkins in partnership with the Regenstrief which will, among other things, achieve that. The work is not yet complete - but I'll be sure to post and update.
Cheers
Davera
Jay Lyle (Mar 10 2022 at 14:34):
Many thanks.
One detail: I see methodless "41461-5 Virus identified in Specimen" and "42808-6 Virus identified in Isolate". (Also true for bacteria, mycology, mycobacteria; only Isolate for parasitology.) My understanding is that the organism is typically identified in the isolate from the specimen, so I don't know if there's a good rule for which one to use.
Robert McClure (Mar 11 2022 at 17:12):
@Jay Lyle I assume you know that your OP is asking for a different type of thing then this last one, which would have a semantic type akin to Finding. As for the two latter choices, I believe the difference is one of specific context. If a specimen is "refined" in some way, then the result could be considered an "isolate" versus testing the specimen directly. If you are looking for something general, I'd use "Specimen."
Rob Hausam (Mar 11 2022 at 17:51):
The term "isolate" is normally (probably always) used in practice when we are talking about cultures - i.e. the organism (most often bacterial, but also can be viral) is specifically grown in order to assist with identification and/or provide samples from that particular organism for further testing (susceptibility or others). So an isolate is a derived specimen (although I think that usually isn't stated). If the organism isn't grown to obtain the specimen, then the specimen (whether derived or processed by staining, aliquoting or anything else) isn't an isolate. PCR and other molecular microbial tests are usually done on non-cultured (therefore not "isolate") specimens. So we should choose and use the specific LOINC codes (for "Specimen" or "Isolate") accordingly. @Riki Merrick?
Riki Merrick (Mar 11 2022 at 22:12):
What Rob says is correct - the more generic term is the LOINC that uses XXX - originally this used to mean the code is silent on the system, but it shifted over time to mean specimen (clinical and environmental) while the speicmen term is usually reserved for clinical speicmen - and LOINC does have a system called specimen/isolate - so if you have a methodless LOINC with that system I would use that.
Jay Lyle (Mar 12 2022 at 14:20):
Thanks; we do have an "isolate #" so I think Isolate is right. Just not sure if there are edge cases where you might id the bug without an isolate step, and have to make up a fake isolate #. I think that's rare enough not to matter, and inconsequential for the current case (patient view). If that did happen, what could go wrong? Inappropriate implication of confidence in the ID?
Last updated: Apr 12 2022 at 19:14 UTC
