Hello everyone,
I’m currently working on mapping qualitative lab results (e.g., +, 1+, 2+, Negative) to value_as_concept_id in the MEASUREMENT table, and I’ve hit a practical design question that I’d love to hear your thoughts on.
In OMOP vocabularies, some LOINC measurement concepts have a defined set of answer concepts (via CONCEPT_RELATIONSHIP with Has answer), typically from the LOINC Answer vocabulary. However, many do not.
That leaves us with two potential strategies:
- Respect the defined answer relationships – if
Has answerexists for a test, use it. If not, fall back on appropriate standard concepts (often SNOMED Meas Value concepts likePositive[9191],Negative[9192]). - Ignore those relationships and directly choose the standard concept that best matches the source value, regardless of linkage.
Theoretically, (1) seems cleaner, but in practice, things get messy.
Even for LOINC tests that have defined answers, the answer variations (e.g., Negative, 1+, 2+, 3+) might not fully cover source values like +, +/-, ++++, etc. So you may end up mixing SNOMED and LOINC concepts within the same test — which might be fine, but it feels semantically inconsistent.
For example, here is a source mapping case I’m working with:
As you can see, some values like 2+ and 3+ have matching LOINC Answer concepts via Has answer, but others like + or +/- do not — so I’m not sure if it’s better to:
- Stay consistent by using only SNOMED (e.g.,
Positive,Equivocal), or - Allow mixing vocabularies in
value_as_concept_id.
I’d love to hear how others are handling this!
- Do you use
Has answerrelationships actively, or do you disregard them for consistency? - How much do you care about vocabulary consistency (SNOMED vs. LOINC) in this context?
- Do you prioritize semantic alignment or practical maintainability?
- Are there any official docs or community recommendations on best practices for qualitative result mapping?
Thanks in advance 
