OHDSI Home | Forums | Wiki | Github

COHORT and COHORT_DEFINITION tables in Vocabulary schema

Hello

The Vocabulary schema has table COHORT and COHORT_DEFINITION. The COHORT_DEFINITION table is empty but the COHORT table is not. Additionally, data in vocabulary.COHORT table doesn’t match the cohorts in ATLAS.

My understanding is that ATLAS writes back to the results schema. Tables in cdm and vocabulary schema are all read-only to ATLAS. I checked table results.cohort and data in that table matches well with cohorts presented on ATLAS.

I am not sure how the vocabulary.COHORT table got populated and how those two tables in vocabulary schema should be used.

Please advise.

Thanks
Jack

It’s actually the ‘cdm schema’ not the vocabulary schema; the vocabulary schema is just a sub-schema of the cdm schema containing the vocabulary related tables (domain, concept, etc)

But, you are correct that the cdm schema contains a cohort and cohort_definition table. These were originally intended to be written to, however people typically make their CDM schemas read-only, and so a separate ‘results’ schema was introduce to store analytical results (including cohort generations).

If you have content in your cdm schema’s cohort table, then you have a misconfiguration on your WebAPI: the ‘results daimon’ should specify your results schema, as shown in the setup documenation.

Note, you can view the schemas that are set up for a source in the ‘configuration’ left nav of Atlas.

Thanks, Chris.

here is the schema set up shown in the Atlas. Does it look correct?

I noticed that the vocabulary.cohort table was not updated when new cohorts were created in Atlas. If the current setup is correct as shown above, it could be a misconfig in the past in WebAPI and was corrected.

so basically the COHORT and COHORT_DEFINATION tables in cdm schema are at no use and shouldn’t be populated, correct?

Thanks
Jack

Correct. They may be populated for some site-specific purpose, but Atlas doesn’t leverage those tables.

Does it mean that individual sites are free to use those tables to define cohorts for other than a network study? I am defining a data model for a prospective study and I am thinking to use COHORT_DEFINITION and COHORT tables to split my full patient set into mutually exclusive subsets. Is there anything in the definitions of these tables that would prevent such use of these tables?

I think if you want custom tables for your specific use, go ahead and create them. I’m not sure the cdm’s COHORT table was built with your use-case in mind.

t