OHDSI Home | Forums | Wiki | Github

ACHI Procedure Codes to SNOMED?

Hi,

Hope everyone doing well!
We have started to map the ACHI Procedure codes to SNOMED but there are 6000+ codes to map, we are just wondering anyone has mapped ACHI Procedure codes to SNOMED. We would like to know if anybody else is working with this process or already has done and could help with this mapping process.

Thanks

Hi @DinujaS , great hearing that you try to tackle ACHI. I guess you are doing this within the OHDSI Australia community and you are in touch with @nicolepratt ? (or is it @Nicole_Pratt ?).
Procedure code mapping is one of the toughest as they very often primarily reflect billing information rather than clinical data and combine multiple pieces of clinical data into one code… You might need to decide if the target concept is a procedure, a drug exposure or a device or maybe a combination (AND) of some. If you come across an OR, you have to decide for one of them or take the step to make that particular concept a standard if there is no proper target to be found. Also, many of the codes have only been created to full empty slots in a matrix and are rarely used. First step therefore is to look at your real word data and get frequencies so that you know which ones will give you the best value when mapped. Also, as most of them are built hierarchically, it might be worthwhile to map the less specific ones first and fall back to those mappings (uphill mapping) if you cannot find a proper mapping for the more specific ones. Usagi can help a lot in the mapping process, but very often has to give up with the complex syntax of procedure codes…
I would hope that you can initiate a collaboration and divide the work a little between multiple Australian institutions (and / or pool frequencies from multiple institutions giving you better ideas about priorities).
Feel free to reach out with more questions.
~Mik

Dear @mik ,

Thank you for your response. Yes I am working with @Nicole_Pratt. We have started to get the frequencies of the current used ACHI codes in our datasets and have started to do manual mappings which are not properly matching with standard concept codes via Usagi. Hope we are in the correct path. If We come up with any questions let you know. Thanks in advance.

Kind regards,
Dinuja S’Wickrama

t