If you are using fact_relationship table today, could you post what you are using it for? I am curious and maybe the whole community would like to compare notes / lessons learned ?
I also would like to run a formal informatics study on OHDSI fact linking and do a formal survey of sites/experts willing to have a qualitative interview about use of fact_relationship. (message me or email me or reply here)
Winning use cases can be turned into guidance in official spec or official conventions. Formal functions in OHDSI tools (e.g., Atlas) can be pursued if indeed standardized (e.g., mother-child linkage).
Also, if you will be at the symposium, and see me, stop me and let’s talk various uses of it as of 2026.
@Eugene_Paulenkovich and I are presenting “Data Link Odyssey: Methods for Linking Related Records in OMOP CDM v5.4” as a poster presentation at the upcoming OHDSI Global Symposium. As the name implies, our poster discusses the different methods for linking records in the OMOP CDM along with the pros and cons of each method.
At Fred Hutch Cancer Center we are currently using the fact_relationship table to map hierarchy in care sites, using relationship_concept_id of 46233689 (“Care Site contains Care Site”) and 46233688 (“Care Site is part of Care Site”). As we continue our CDM mapping, I expect we may use the table for other mapping.
We also use fact_relationship to capture family relationships, however instead of using ‘Child to Parent Measurement’ we use any descendant of 4216170 - Immediate family member to allow other relationships as well, e.g. ‘Child’, ‘Mother’, ‘Father’. Relationship direction is not explicit, so we interpret these as ‘Child (of)’, ‘Mother (of)’, etc.