OHDSI Home | Forums | Wiki | Github

Mapping Data in ICD10CA

Implementers
#1

Hi folks,

I am about to map a data source that’s in ICD10CA (Canadian version of ICD10), but I noticed that Athena has ICD10, ICD10CM, ICD10GM, ICD10CN but no ICD10CA. In this case, my questions are:

  1. Is it doable? ( I guess the answer is YES)
  2. What is the best/proper way to do the ETL mappings

ICD10CA dictionary like this:

code        class         short_desc              long_desc
80000     ICD10CA         Neoplasm, Benign        NEOPLASM, BENIGN
80003     ICD10CA         Neoplasm, Malignant    NEOPLASM, MALIGNANT
....

Any suggestions are highly appreciated.

Sincerely,
R. Z.

#2

Tagging @mik to help.

#3

Thank you @MPhilofsky
Thanks @mik in advance. Look forward to your reply @mik.

#4

Hi @zohdsi ,
first, please accept my apologies for the very delayed response. Apparently I missed that @MPhilofsky tagged me here and had to be made aware of this post by a colleague.
The inclusion of a certain flavor of ICD10 into the OHDSI standardized vocabularies is demand-driven and might require a small bit of funding for initial processing and maybe maintenance for the following years. In the case of ICD10-CA to me it seems the data is not released into the public domain (as it would be the case for other ICD10 sources) but seems to come with a license cost? That is still possible, but would require us to make it license restricted in Athena and get agreement from CIHI to process this data free of charge for OHDSI purposes.
CIHI claims it is derived from the International (original) ICD10 version. I was casually looking at a couple of codes from the free coding standards document and felt that it might be inspired here and there from ICD10-CM, but wouldn’t see a clear pattern right away. Your best bet is probably to go look for complete matches between ICD10-CA and ICD10 first, then for others look for uphill mapping or equivalents in ICD10-CM and for all of those find the “Maps To” relationships to standard concepts to build your own mappings. I am however a bit confused by the code that you presented in your example: 80000 or 80003 do not seem to be valid ICD10 style codes, even if omitting the separating dot. It does look more like ICD9 coding which again is also unlikely as “800.00” would point at closed fracture of skull. And “Neoplasm, benign” would probably rather be a chapter in ICD10. So, try to find the actual codes in your source that will match concept codes in OHDSI ICD vocabularies and then start some nifty scripts collecting matching concepts and their relationships from your downloaded Athena content. What is left over I would check for frequency of occurence in the source data and go for some additional manual mapping for the most frequent ones.

t