OHDSI Home | Forums | Wiki | Github

Recent changes in SNOMED CT relationships?

Hi @Christian_Reich @gregk @Dymshyts

About 25% of ICD-9-CM and ICD-10-CM mappings (condition codes only) to SNOMED CT have changed in the recent vocabulary release (however, some inaccurate mappings exist. I’ll post the list on github later).

Are we using a new algorithm to map source codes to SNOMED CT? And, how often the mappings may change in future? These changes may have significant impact on the results.

Thanks,
Hamed

Hi Hamed, – what ‘Vocabulary Release’ are you referring to? Which entity released it?

All –

What tool is used to do the mapping?

Is there a document describing the CLM (Content Lifecycle Management) process for vocabulary version control, mapping, and configuration management?

@abedtash_hamed:

This is the result of a big revision to improve the quality, all manual. Unfortunately, there is no other way: Even if you had a way to reliably identify equivalent description strings it would not be enough, because you also have to compare the descendants to make sure the semantic meaning is equivalent.

And hell no, we will not keep doing that. :smile: It’s bad enough that the CMS is making new ICD10CM babies like the rabbits: a year ago we had 70k, and now the number is almost 110k. These all have to be mapped. I hope the provider community is going to push back against that craziness.

But please provide your list.

Thanks @Christian_Reich. I agree with you text processing is not enough to map the codes to standard concepts. Manual review is the gold standard. I’ll post the list.

That’s great we are not making changes so often :smile:

@Ron The OMOP vocabulary is maintained by OHDSI community, available http://athena.ohdsi.org . The vocabulary I analyzed is dated 12/22/2017. You may find more information, here: https://github.com/OHDSI/Vocabulary-v5.0

@Ron:

It’s the OHDSI Vocabulary group, which essentially is the Odysseus team and myself. We get a lot of support from the community through contributions like this one.

No tool. Because we are not developing terminologies:

  1. We “steal” them, ie. we incorporate existing vocabularies from external sources (mostly public, some private).
  2. For the mappings and other relationships: We also try to rely as much as possible on external sources, but for these we need more exceptions.

Exceptions from rule 1): RxNorm Extension and the various internal OMOP vocabularies, like domains, concept classes, etc.
Exceptions from rule 2): The ICDs to SNOMED, MedDRA to SNOMED, various procedure Concepts to SNOMED (even though that needs a big revision), and various drug vocabularies to RxNorm and RxNorm Extension.

We are currently not mapping Measurements, Devices, Observations and Specimen, because either there are no widely accepted generic terminologies for these to begin with, or because it is unwieldy and the use cases supporting it deprioritize these (e.g. Devices).

Documentation: Only in-code documentation on Github. Nothing for the faint-hearted.

1 Like

@Christian_Reich I just submitted the issue on github #148

t