Graph View

ActorsStandardsResourcesGovernance

LogoLogo

OMOP CDM — Observational Medical Outcomes Partnership Common Data Model

Overview

The OMOP Common Data Model (CDM) is an open standard for structuring observational health data, such as EHR records, insurance claims, and registries, into a common format. It enables federated network studies: each site runs the same analytical code locally and shares only aggregate results, without exposing raw patient data. OHDSI now maintains it, building on the original OMOP initiative, a public-private partnership involving the FDA (Food and Drug Administration), academic institutions, and pharmaceutical companies. The TEHDAS joint action recommended OMOP CDM as the common data model for secondary use under the EHDS, though the Regulation text does not name it explicitly.

Interoperability

OMOP CDM and HL7 FHIR interoperability is being developed jointly by OHDSI and the HL7 Vulcan Accelerator. A FHIR-to-OMOP transformation specification is under active development as of 2025. Converting data into the model is a build step that reads from source systems, reshapes records, and writes them into CDM tables. OMOP-to-FHIR tooling exists, but a fully standardised bidirectional mapping is not yet complete. OMOP CDM also maps to CDISC SDTM for clinical trial data integration, with conversion guidance maintained by the OHDSI Clinical Trials Working Group.

CDM Structure

The OMOP CDM organises clinical data into domain tables. For each domain, OHDSI’s Vocabulary team designates a fixed set of Standard vocabularies. Their concepts can be used directly. Any other vocabulary is Non-standard and must be mapped into a Standard one before use in cross-site analysis. A domain can have more than one Standard vocabulary at once, so data already coded in any of them can be kept as-is. The OHDSI Standardized Vocabularies span more than 100 vocabularies in total. The table below covers those most relevant to neuroscience data.

Standard vocabularies

DomainTableContentsStandard vocabulary
PersonPERSONDemographics, one row per patientNot coded
Observation PeriodOBSERVATION_PERIODPeriods of clinical data availabilityNot coded
VisitVISIT_OCCURRENCEHospital stays, outpatient visits, EDNot coded
ConditionCONDITION_OCCURRENCEDiagnosesSNOMED CT
DrugDRUG_EXPOSUREPrescriptions and administrationsRxNorm, CVX (vaccines)
ProcedurePROCEDURE_OCCURRENCEClinical proceduresSNOMED CT, CPT4 (US), HCPCS (US), ICD-10-PCS (US)
MeasurementMEASUREMENTLab results and vitalsLOINC
ObservationOBSERVATIONClinical findings not fitting other domainsSNOMED CT, LOINC
NoteNOTEUnstructured clinical notesNot coded
DeathDEATHDate and cause of deathNot coded
SpecimenSPECIMENBiological samplesNot coded

Supplementary vocabularies

UCUM (Unified Code for Units of Measure) standardises units of measure across domains, in Measurement, Drug Exposure, and elsewhere. ATC (Anatomical Therapeutic Chemical Classification) provides drug-class groupings for the Drug domain, mapped to RxNorm Ingredient concepts within the OMOP vocabulary.

National and non-standard vocabularies

National procedure vocabularies such as CCAM (France) or OPCS-4 (UK) do not hold Standard status. Like source codes from ICD-10 and MedDRA, they are mapped into a Standard vocabulary, typically SNOMED CT. This mapping happens via Athena, which also covers ICD-11, MeSH, HPO, MONDO, ORDO, and NDC. MedDRA concepts largely lack a full “maps to” equivalence to OMOP standard concepts, so adverse events coded in MedDRA typically require manual mapping.

Extensions

Ratified extensions

The Oncology CDM Extension represents cancer diagnoses using ICD-O-3 histology and topography codes, mapped to precoordinated SNOMED CT concepts. It is targeted for CDM v6.1 and not yet part of the core released model. A separate vocabulary extension, OMOP Genomic, provides a canonical reference list of somatic genomic variants clinically relevant to cancer. It consolidates public cancer variant knowledgebases into the OMOP Standardized Vocabularies, rather than into the table structure itself. As of 2022 it covered more than 95,000 variants across 575 cancer genes.

Local and custom extensions

Individual OMOP CDM implementations frequently build their own custom tables for data types the core model does not cover. The model is explicitly designed to be extended locally. The All of Us Research Program is one example. Its electronic health record, survey, and physical measurement data are standardised to core OMOP CDM tables. Its wearable device and genomic data are held in separate, program-specific custom tables outside the model. Such local extensions are common practice across OHDSI implementations, but unlike the ratified extensions above, they are not themselves standardised across sites.

Connections

  • governedBy: OHDSI
  • relatedTo: EHDS (de facto secondary-use data model for HealthData@EU, not explicitly named in the Regulation text)
  • recommendedBy: TEHDAS
  • implements: SNOMED CT
  • implements: LOINC
  • implements: ICD-10
  • implements: RxNorm
  • implements: CPT4
  • implements: HCPCS
  • relatedTo: ATC (drug-class classification, mapped to RxNorm Ingredient concepts within the OMOP vocabulary)
  • relatedTo: ICD-O-3 (mapped via the ratified Oncology CDM Extension, targeted for CDM v6.1 and not yet part of the core released model)
  • relatedTo: MedDRA (source terminology for adverse events, incomplete mapping to OMOP standard concepts)
  • relatedTo: CCAM (French Non-standard procedure vocabulary, mapped into SNOMED CT rather than holding Standard status itself)

Resources

Graph View

ActorsStandardsResourcesGovernance

Tags

Backlinks