Condition Occurrence

Epic equivalent: Clarity diagnosis table / Encounter diagnoses / Problem list

The condition_occurrence table captures diagnoses and problems recorded for a patient — billing diagnoses, problem list entries, discharge diagnoses, and other structured condition data. Each row is a single recorded condition event, often tied to a visit.

In Epic, you'd query the diagnosis table or encounter diagnoses. In OMOP, all of these consolidate here with standardized SNOMED codes (mapped from ICD-10, ICD-9, etc.).

Epic-to-OMOP Field Mapping

Field reference (click to expand)

OMOP Field	Epic Equivalent	What It Captures
condition_occurrence_id	Condition record ID	Unique identifier for the condition event
person_id	Patient ID / MRN	Links to the patient
condition_concept_id	ICD/SNOMED standard concept	Standardized condition (e.g., "Type 2 Diabetes Mellitus"). Always use this for analysis
condition_start_date	Date of diagnosis	When the condition was recorded or began
condition_end_date	Resolution / stop date	When the condition resolved (often null in EHRs)
condition_type_concept_id	Source context	Origin: problem list, billing claim, EHR note, etc.
condition_source_value	ICD-10 / ICD-9 / local code	The source code as it appeared in Epic
condition_source_concept_id	Source code concept	Maps the source value to its OMOP concept ID
visit_occurrence_id	Linked encounter	Visit during which the diagnosis was recorded (may be null for ongoing problems)
provider_id	Recording provider	Who documented the diagnosis

What to Watch For

Common pitfalls

Chronic conditions repeat

A patient with diabetes may have hundreds of condition_occurrence rows — one for every visit where it was recorded. Use condition_era if you want the continuous period.

Missing end dates

Most EHRs don't capture when a condition resolves. NULL in condition_end_date does not mean the condition is still active — it means we don't know.

Use concept_id, not source_value

condition_source_value contains raw ICD codes. For standardized cross-site analysis, always use condition_concept_id (SNOMED).

Research Patterns

Question	Tables Involved
Patients diagnosed with diabetes in the past year	condition_concept_id (diabetes descendants) + condition_start_date
Time from cancer diagnosis to first treatment	condition_occurrence (cancer) + drug_exposure or procedure_occurrence
COVID-19 diagnoses during inpatient stays	condition_concept_id + visit_occurrence_id filtered by inpatient
Diagnoses more common in ER visits	condition_occurrence + visit_occurrence with ER filter
ICU comorbidity burden	condition_occurrence + visit_detail filtered for ICU