Such “Other” concept codes reflect one of the key differences between the"closed-world" system (classifications such as ICD-x) and an “open-world” system(terminologies/ontologies such as SNOMED CT) . These codes hardly fit into the latter even some of them are already inluded by the latter (for example, for mapping). The use of Such codes should be restricted explicitly.
I think two types of ‘Other’ are needed, one which just allows a DV_TEXT and one called e.g. “Other missing from list” that also allows a DV_TEXT. Then, when the new entry is eventually added to (e.g. SNOMED) it could be post-coordinated. Maybe I’m too optimistic about such a possibility.
Noting, of course, that the free text values inserted under the banner of ‘Other’ will often be reviewed at a later date and may be used to expand the original coded value set.
This is a challenge we often meet building models for clinical applications. We are building models for a colonscopy report these days. The data will be exported into (at least) two different national registries. For the new national screening program they have developed FHIR profiles for the data.
I will share a few examples.
Further plan/follow up
As part of the procedure the clinician should make a statement about the follow up.
This is the FHIR valueset for this statement: nocolonoscopyfollowup
As you can see the valueset combines SNOMED-CT codes with the FHIR null flavor statement “OTH::other”.
To make the openEHR model clean we decided to use the SNOMED-CT other qualifier.
We ended up with the following valueset for our openEHR template:
[
"SNOMED-CT::10031000202109::No follow-up (situation)",
"SNOMED-CT::10041000202103::Histology result not back (situation)",
"SNOMED-CT::703993001::Colonoscopy planned (situation)",
"SNOMED-CT::10051000202100::Referral to further evaluation and/or treatment (situation)",
"SNOMED-CT::10061000202102::Virtual computed tomography colonoscopy planned (situation)",
"SNOMED-CT::74964007::Other (qualifier value) "
]
Current medication
The registry need a list of relevant drug categories, for use in reporting in connection with colonoscopy.
The FHIR valueset is defined here: nocolonoscopycurrentmedication
This is a mixed valueset from multiple systems. We wanted a clean openEHR model. First we wanted to create a specific archetype for this. The idea was to model the valueset as an archetyped DV_CODED_TEXT.
But - the use of such medications is a precaution related to the colonscopy procedure. There is a reviewed archetype for precaution: openEHR-EHR-EVALUATION.precaution.v1. We decided to use this. Then we had to make a valueset to add to the element Condition.
We ended up creating a registry specific terminology:
SCREENIT-MEDICATION::M00::Nil Known
SCREENIT-MEDICATION::M01::Aspirin (substance)
SCREENIT-MEDICATION::M02::Other platelet inhibitors
SCREENIT-MEDICATION::M03::Warfarin (substance)
SCREENIT-MEDICATION::M04::DOAK
SCREENIT-MEDICATION::M05::Low molecular weight heparin (substance)
SCREENIT-MEDICATION::M06::Other anticoagulants
SCREENIT-MEDICATION::M99::Other
SCREENIT-MEDICATION::M100::Unknown / missing information
This is kind of dirty modelling approach combining different terminologies into a single valueset. Anyway we solved the problem at hand within a reasonable use of time and resources