OHDSI Home | Forums | Wiki | Github

CDM_SOURCE table usage and conventions

I am replying in forum to an email discussion.

question was: (from Claire)

I want to follow-up on our discussion about adding a value to the cdm_version column in the CDM_SOURCE table. The way I understand it is that there should only be one row in this table so if I put a blank row it could cause problems with any current builders and current conventions.

Erica suggested putting the version as a default on the column so that when the table is populated it will default to the version I specify if the builder doesn’t fill it.

My answer is

I would rather have a rule that says that only one row is allowed in cdm_source.
What if I have 2 rows in there and one says cdm version 5.2 and the other row says 5.3 - so what is the version Heel should “plan for”?

Even better – delete some columns from the table (e.g., cdm version) and put it into METADATA

(if you insist on multiple sources capture)


Shouldn’t this be somewhere in a THEMIS group?

Looking at how METADATA table work have been developing, I have been questioning the need for the CDM_SOURCE table at all. Most of the information in CDM_SOURCE is metadata.

CDM version, sources, vocabs are perfect examples of metadata. We are using those value quite extensively in ARACHNE today and have been discussing on using those in other platforms such as ATLAS/WebAPI, ACHILLES to determine feature compatibility. The problem is that in most of the database CDM_SOURCE is not or partially populated.