OHDSI Home | Forums | Wiki | Github

Missing Cohort Domain for person?

Developers
#1

All,

According to the cohort_definition docs for CDM v5, we should be using the domain concept id from the DOMAIN table for cohort_definition.subject_concept_id. For cohorts of persons, there appears to be no domain available for conceptually mapping these cohorts (using Vocabulary5.0-20141013).

It seems that the Domains are a mix of table and column specific. Should every clinical table be represented as a Domain?

Bill

#2

@wstephens:

You have a knack for putting the finger into the wound. :smile:

In reality, the Domain Concepts will be one of the following three:

concept_id concept_name
2 Gender - we could use Race and Ethnicity instead
33 Provider specialty
8 Visit

Let’s come up with a better way.

#3

I agree with Bill and support (second) his proposal to have concepts for every clinical table. In the cohort table - I need to refer to either provider or patient. Yet - there is no concept for person table.

I posted a while age almost exactly this issue to the forum.

#4

I’m now running into this problem for the CDM_IPCI ETL . . . my recommendation is you just add a CONCEPT_ID to the VOCABULARY_ID = “DOMAIN” called PERSON. Practically everything else is in there!

Why I need it:
In IPCI there are two types of time periods, the time for which you have records for someone, and the time at which you are confident you have a complete dataset (IPCI Research Ready Time). We are defining the time period where there is data in the OBSERVATION_PERIOD table and the time when the data is “Research Read” in the COHORT table. I want to define that the SUBJECT_ID in my COHORT table is referring to a PERSON_ID not a VISIT_OCCURRENCE_ID or PROVIDER.

#5

In addition I also need a CARE_SITE domain. I want to use the FACT_RELATIONSHIP table to relate CARE_SITE to PROCEDURE_OCCURENCE.

#6

In case anyone is wondering the following DOMAIN concepts were created:
56-Person
57-Care site

t