We are excited to announce that the first release of Fred Hutch’s OMOP data model is now live! OMOP is a standardized data model for observational health research. It allows researchers at Fred Hutch to connect to an international network of researchers and observational health databases.

Summary

This is the first major release of the Fred Hutch OMOP CDM based on Fred Hutch Epic Clarity patient data within Clinical and Research Data System (CARDS).

This release is the first full build of the core OMOP tables. It enables scientific project teams at Fred Hutch and will provides data for the Cancer AI Alliance (CAIA).

The following tables are now ready for use:

  • person
  • location
  • death
  • visit_occurrence
  • condition_occurrence
  • procedure_occurrence
  • measurement
  • drug_exposure
  • observation_period
  • cdm_source

This release uses the 20260227 vocabulary version.

User Impacts

Users will now be able to see data in the curated_omop catalog on CARDS.

Inclusion Criteria

We implemented basic inclusion criteria for the OMOP database so that persons with low-quality clinical data were excluded. Our inclusion criteria are as follows:

  • Must have a “U Number” (MRN)
  • Must have at least one visit
  • Not a test or fake patient
  • Must have a date of birth (and not in the future)

Technical Notes

  • DQX data quality monitoring was implemented for proactive monitoring of critical data issues
  • Declarative Automation Bundles (DABs) and CI/CD actions were implemented for deploying jobs in prod and test environments

Current Limitations

Integrating data from clinical source systems will be an ongoing process. If there are data missing that you would like to see incorporated, please email us at analytics@fredhutch.org. The following are notes on known limitations that we will be addressing in future development cycles.

  • drug_exposure: Immunizations and anesthesia records have not been incorporated

  • measurement: The current OMOP version incorporates patient vitals from encounters, but vitals from Epic’s “flowsheets” have not been incorporated. These capture more frequent monitoring of vitals during inpatient hospitalizations, but will need to be manually mapped to OMOP standard concepts.

  • procedure, condition: We may not be seeing a complete picture of procedures and conditions from hospital/professional billing due to restrictions on how financial data is exchanged between FHCC and UWM. However, it is likely that these events are adequately captured by EHR orders and other Epic Clarity tables.

  • Data completeness from Epic Clarity: Drug and procedure records may be less complete (compared to other tables) in pre-2021 years (Epic go-live was in March 2021). A full assessment of data completeness and quality will be conducted with the OHDSI Achilles and DQD packages.