OHDSI Home | Forums | Wiki | Github

Concept_id 35207362 is not mapped correctly

Right now, Concept_id 35207362 (Degenerative disease of nervous system, unspecified) is mapped to 2 following standard concepts:

376337 (Disorder of nervous system)
4179873 (Degenerative disorder )

I believe neither is correct. It should be an ‘Is a’ relationship, not a 'Maps to ’ relationship. “Disorder of nervous system” includes all diseases in CNS, degenerative or non-degenerative. “Degenerative disorder” includes all degenerative diseases regardless if it is in CNS or other parts of the body, such as cataract, osteoarthritis etc,

Would it be more accurate to map the G31 ICD10 (the 35207362) as 4213310: Degenerative disease of the central nervous system? I looked at the approximate synonyms here: https://www.icd10data.com/ICD10CM/Codes/G00-G99/G30-G32/G31- and I wonder if 'nervous system is implying CNS in ICD10. If so, then I think the 4213310 (which explicitly states centeral nervous system) would be a good ‘Maps to’ relationship there.

Yeah that seems like a good mapping. Interestingly, the codes within G31 that correspond to specific diseases all seem to be brain disorders, so one could consider the SNOMED codes for brain disorder or degenerative brain disorder. But to ensure that the ‘not elsewhere classified’ from the original ICD10 code is included, Chris’s proposal is probably best

“35207362 G31.9 Degenerative disease of nervous system, unspecified” does not have an equivalent SNOMED mapping. Although the suggested SNOMED concept of “80690008 Degenerative disease of the central nervous system” seems relevant, from a medical standpoint, it is much more granular that G31.9. Such an approach can lead to considerable data loss. According to the nervous system specification, G31.9 should include such compartments as central, peripheral, and autonomic nervous systems. Consequently, different types of degeneration can occur in different places. That is why G31.9 is mapped to the closest semantic ancestors, which as a combination can reflect its real meaning.
Also, we have empirically established, that (being made by some people) ICD’s Approximate Synonyms may be imprecise.

It should be an ‘Is a’ relationship, not a 'Maps to ’ relationship.

That is where you are totally right. Semantically these multiple “Maps to” mean “Is a”. But, currently, an “Is a” relationship is not allowed to use in mappings of the ICD9CM, ICD10 and ICD10CM vocabularies due to technical reasons. However, we do want to change this by the creation of the SNOMED Extension.

1 Like

Actually, it’s an interesting case:
When you look only at the description of the concept it stands for Unspecified.
But when you look at

you can see

  • Acquired cerebral atrophy
  • Cerebellar paraneoplastic degeneration
  • Cerebral atrophy, acquired
  • Cerebral degeneration in childhood
  • Degenerative brain disorder
  • Paraneoplastic cerebellar degeneration
    and everything here relates to the CNS only.

So, technically, @Chris_Knoll is right.
@Polina_Talapova before creating SNOMED Extension, we need to analyze all such a cases when we derive the meaning of a concept by it’s synonyms in ICD10.

Personally, I am not aware of any degenerative diseases in peripheral nervous system.

@QI_omop, they exist. At least, there are different axonopathies in peripheral neuropathies caused by metabolic diseases, traumas (well-known Wallerian degeneration), and toxicity. Moreover, multifocal acquired motor axonopathy is also characterized by degeneration of PNS (although it has a specific code of G61.82 in ICD-10-CM, in ICD-10, it does not).

t