There seems to be a combination of translation difficulties and underlying modelling issues at play, and conflating them makes the problem harder to solve.
Lesion as morphologic abnormality
Jim’s point that lesion is “a general term for a morphologic abnormality typically used when the precise morphology or nature of the abnormality is not yet known” is crucial.
The clearest use cases I see are preliminary imaging findings and initial findings of skin changes (not yet diagnosed) – including those noticed by the patient themselves. In those contexts, lesion concepts are valid (and deliberately broad).
That said, there are probably more lesion concepts than would be clinically useful.
For example, 10691941000119103 |Lesion of skin of nose| or 298004008 |Lesion of nasal mucosa| are the kinds of concepts more likely to be recorded in practice, rather than less specific 301191003 |Lesion of nose|. Identifying which concepts to retire would be a significant undertaking, but the volume of lesion grouper concepts at varying levels of specificity likely creates noise around those with genuine utility - and I imagine a challenge for translations.
It is also worth noting that the vast majority of concepts referring to lesions via the 116676008 |Associated morphology| property are subtypes of 64572001 |Disease| and carry the “(disorder)” semantic tag – roughly 41,000 concepts, compared to around 1,400 that are findings but not disease. Concepts like 12402003 |Scar| and 103605005 |Pimple| illustrate the types of things lesions subsume. Not all lesions are disorders in any meaningful clinical sense (a freckle is a lesion but not really a disease), and this practical imbalance may contribute to translator confusion – especially given SNOMED CT’s inconsistent handling of “Disorders” (a whole other discussion).
Injury versus traumatic injury
There seems to be a conflation between “injury” and “traumatic injury” running through parts of this discussion. As Jim notes, injury can be traumatic or non-traumatic. The same applies to lesions – brain plaques and calcifications are pathological, non-traumatic lesions. Neither term is inherently traumatic, so the trauma/non-trauma axis does not cleanly separate them.
I suspect – and I say this as a native English speaker, not a translator – that many speakers intuitively read “injury” as implying external force, even though the clinical definition does not require it.
- Not all lesions are injuries.
- Not all injuries are traumatic.
- Not all lesions are traumatic.
Damage and injury
My feeling is that “damage” functions in everyday English as causally neutral – not implying trauma in the way “injury” intuitively might. If correct, “damage” could be understood as a supertype of “injury” (silent on mechanism) rather than a synonym. That would make pairs like 243978007 |Liver damage (disorder)| and 39400004 |Injury of liver (disorder)| potentially meaningful as distinct concepts – but only if the modelling reflects that consistently, which it currently does not.
Genuine duplicates versus modelling inconsistencies
I think it is important to distinguish these two situations:
- 1357798008 |Injury of globe of eye due to birth trauma (disorder)| and 206241005 |Eye damage due to birth trauma (disorder)| look to me like genuine duplicates – same clinical entity, inconsistent choice of finding site and terms. Merging or inactivation seems appropriate here.
- 212400009 |Scalp injury| and 871000119100 |Lesion of scalp| are not duplicates. Scalp injury is not currently classified under Lesion of scalp, but this is because the morphology is not explicitly modelled – not because the concepts are clinically distinct. As Jim notes in post 19, these are modelling inconsistencies that the classifier cannot resolve without complete defining relationships.
Merging or inactivating parent groupers without first addressing incomplete descendant modelling would obscure these inconsistencies rather than fix them. I would strongly prefer modelling gaps to be addressed before, or at minimum alongside, any inactivation or merging work.
The Dutch NRC approach
I agree that “abnormal tissue” as the preferred term for lesion concepts is the right approach – it maps cleanly to the morphologic abnormality sense without importing causal connotations. For translators, I suspect the harder problem is actually the damage/injury distinction, where everyday English usage, clinical definitions, and current modelling are not well aligned.