Fact Relationship

Epic equivalent: No direct equivalent — implicit relationships made explicit

The fact_relationship table captures relationships between records across different OMOP tables. In EHR systems, these connections are usually implied through timing, proximity, or clinical context. OMOP makes them explicit — linking a biopsy procedure to its pathology finding, or a pregnancy diagnosis to a delivery event.

Think of it as the glue for non-obvious cross-domain connections.

Field Reference

Field reference (click to expand)

OMOP Field	What It Captures
domain_concept_id_1	Domain of the first record (condition, procedure, measurement, etc.)
fact_id_1	The record ID in the first domain's table
domain_concept_id_2	Domain of the second record
fact_id_2	The record ID in the second domain's table
relationship_concept_id	The nature of the link (e.g., "Has associated finding")

Example relationship concept:

Concept ID	Meaning
44818770	"Has associated finding" — links a procedure to a justifying diagnosis

What to Watch For

Common pitfalls

Rarely populated

Most ETL pipelines don't fill this table by default. Emory does not currently populate it. See Known Issues for updates.

Soft joins — no foreign keys

Relationships are concept-driven, not enforced by referential integrity. Validate carefully.

Requires understanding the relationship vocabulary

You need to know the relationship_concept_id to interpret what the link means.

Research Patterns

Question	Tables Involved
Link pathology results to biopsy procedures	fact_relationship between measurement and procedure_occurrence
Diagnoses that justified a specific surgery	fact_relationship linking condition_occurrence to procedure_occurrence
Composite blood pressure from systolic + diastolic	measurement + fact_relationship ("is component of")
Pregnancy episode from related diagnoses + procedures	fact_relationship with temporal link concepts
Conditions tied to device insertions	fact_relationship between condition_occurrence and device_exposure