Are panels to be mapped based on their contained LOINC codes or can they be mapped by human-interpreted definition?

I have a longstanding question about LOINC panels. I’ve always interpreted panels as a strict relationship between the panel LOINC and its constituent LOINCs. Even when a panel LOINC might make sense from a “visual definition” perspective, if the LOINCs in the panel don’t match, I don’t map to that panel. This makes sense from the “letter of the law” perspective, which makes sense from a computational standpoint, but it’s a recipe for combinatorial explosion on a massive scale.

Is it proper to map LOINC panels by name rather than their precise structure?

Could some panels be modified to include more than one LOINC per panel component – with some sort of conditional modifier so it’s clear it’s “one or the other”?

An example is LOINC panel 93715-1 Francisella tularensis IgG and IgM panel - Serum. It consists of a nice panel name, and 3 components that don’t list method.

  • 7889-9 Francisella tularensis IgG Ab [Presence] in Serum
  • 7890-7 Francisella tularensis IgM Ab [Presence] in Serum
  • 93718-5 Francisella tularensis IgG and IgM [Interpretation] in Serum

I have an identical panel, except we send this to a reference lab who has assigned method-specific LOINC codes to the components of the panel (other than the interp, since there was no method-specific LOINC for it).

  • 93717-7 Francisella tularensis IgG Ab [Presence] in Serum by Immunoassay
  • 93716-9 Francisella tularensis IgM Ab [Presence] in Serum by Immunoassay
  • 93718-5 Francisella tularensis IgG and IgM [Interpretation] in Serum
1 Like

Hi Doug,
Thank you for the question; it’s a good one too. Originally, panels were structured by components and given constraints to not deviate, as you mention. In 2014 a group of SME’s got together and created the content you find in Chapter 12 of the Knowledge Base (used to be User’s Guide). Knowledge Base – LOINC . I believe those contents will help with some of your questions.

I can say when I’m mapping orders for a client that I’ve first assured all results level LOINCs are complete. I search the details page for ONE of the results and look to see if it has a link to any panel. If one of the observations isn’t in a linked panel, then I know I can’t use any linked panel I may find…my second search is by the name of the panel. In that case I’m looking for unassociated panels.

This topic is on the horizon in 2025 for further discussion from the LOINC chairs and committees. Stay tuned for other details in future LOINC releases.

2 Likes

Thanks Pam. That’s the same approach I take when looking for panels, or even to try to assign LOINCs within a panel. It’s been very helpful.

The KB had my exact use case as an example. here: 12.1.4 Substitutions related to Method. The reasoning seems to be it’s okay (at least for now) to go narrower for components within a panel, but not broader.