OHDSI Home | Forums | Wiki | Github

Using ETL Data to Derive an ICD-10-CM to SNOMED CT Map

Developers
#1

I am creating a ICD-10-CM map to SNOMED (Yes, I am aware that this will not always yield a 1-1 mapping)

I have been told by a high level informaticist that there are data files referenced somewhere in the OHDSI scope that can be used to create such a map between ICD-10 and SNOMED. This does not have to be a 1-1 map. However, the data set found on

https://www.nlm.nih.gov/research/umls/mapping_projects/snomedct_to_icd10cm.html

can yield up to 1800 SNOMED values per ICD10. This suggests that this is not an optimal dataset.

What specific files could support finding all these relationships?

RKW

#2

@keithwelch:

OHDSI provides a mapping from ICD10CM to SNOMED for free through the Standardized Vocabularies. You can read all about things in the Book of OHDSI, but in a nutshell:

  • Create a OMOP database using the DDL available in the Github, but you really only need the tables CONCEPT and CONCEPT_RELATIONSHIP.
  • Download the Vocabularies from Athena and load at least these two tables.
  • Extract the mapping:

select c1.concept_code as ICD10CM_code, c2.concept_code as SNOMED_code
from concept c1
join concept_relationship on concept_id_1=c1.concept_id
join concept c2 on concept_id_2=c2.concept_id
where c1.vocabulary_id=‘ICD10CM’ and c2.vocabulary_id=‘SNOMED’
and relationship_id=‘Maps to’

and voilà.

1 Like
t