Observational Medical Outcomes Partnership (OMOP) common data model (CDM)

  • Updated

All of Us utilizes the Observational Medical Outcomes Partnership (OMOP) common data model (CDM) to organize and standardize raw data for collaborative research and large-scale analytics. Visit the links below to learn more about this national project and the teams that support it.

Please review our All of Us Research Program Data Access Framework for further information.

Common reasons counts change between CDR releases

Several factors can contribute to differences in participant counts, condition counts, or other cohort results across All of Us CDR releases:

1. Changes in participant withdrawal status

Each CDR has its own data cutoff date, and participant consent or withdrawal status can change between releases. Participants who have withdrawn consent before a CDR's cutoff date are excluded from that CDR—even if they appeared in an earlier release.

2. Data cleaning and quality improvements

Each CDR incorporates ongoing improvements to data validation and quality control. These may include:

  • Improved data quality checks
  • Updated validation rules
  • Resolution of source data inconsistencies
  • Removal of records with invalid or unresolvable values
  • Privacy-related filtering, where appropriate

As these processes evolve, some records included in an earlier CDR may be corrected, excluded, or represented differently in a later release.

3. OMOP vocabulary updates and concept changes

The All of Us Research Program uses the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) and standardized vocabularies maintained by the Observational Health Data Sciences and Informatics (OHDSI) community.

Each CDR is built using a specific OMOP vocabulary version, which often reflects a vocabulary release from roughly 6–12 months earlier due to the CDR development, validation, and release cycle.

 As vocabularies evolve, concepts may be:

  • reassigned to different domains (e.g., from Condition to Observation),
  • deprecated and replaced,
  • remapped to different standard concepts,
  • updated in descendant/hierarchy relationships, or
  • refined to reflect improved clinical semantics.

These changes can alter how cohort criteria are interpreted and applied, which may increase or decrease counts, even when using the same query.

For details, consult the OMOP vocabulary release notes corresponding to the vocabulary version used in each CDR.

4. Changes in contributing EHR site data and Extract, Transform, Load (ETL) processing

The set of contributing healthcare organizations and their submitted data may change over time. Differences may result from:

  • updated source data submitted from a site,
  • improvements to ETL processes,
  • corrections to historical source data, or
  • changes in source system implementations

Any of these changes can affect participant counts and clinical record counts between CDR releases.

5. Rare data submission or data curation issues

Less commonly, discrepancies may reflect issues such as:

  • unexpected problems in data submitted by a site,
  • data transmission or submission failures, or
  • curation issues identified after release.

When reported, these are evaluated case by case and, when appropriate, addressed in subsequent releases.

Summary

Differences in participant or condition counts across CDR releases are expected and often reflect updates to the underlying data, standardized vocabularies, or curation processes rather than missing or incorrect data.

Because OMOP vocabularies are maintained by the OHDSI community, interpretation of changes in mapping, hierarchies, and domain assignments is governed by the OMOP vocabulary releases—not the All of Us Research Program.

OMOP Versioning Notes

There have been updates to the OMOP ETL version that may have an impact on the different All of Us CDRs. The following site details changes for version v20260227, released in February 2026. Updates to concept domains (ex: condition to observation) may account for changes in numbers between different CDRs. 

Was this article helpful?

5 out of 5 found this helpful

Have more questions? Submit a request

Comments

0 comments

Article is closed for comments.