Facilitating phenotype transfer using a common data model.

Area of Science:

• Genomic research
• Biomedical informatics
• Clinical data standardization

Background:

• Implementing clinical phenotypes across research networks is labor-intensive and prone to errors.
• Standardizing data through a common data model can streamline this complex process.

Purpose of the Study:

• To evaluate the feasibility and efficiency of implementing the Observational Medical Outcomes Partnership (OMOP) Common Data Model across Electronic Medical Records and Genomics (eMERGE) network sites.
• To compare the implementation time and accuracy of phenotypes using the OMOP Common Data Model versus traditional methods.

Main Methods:

• The eMERGE network sites adopted the OMOP Common Data Model for their electronic health record (EHR)-linked DNA biobanks.
• Two existing eMERGE phenotypes were converted to OMOP format and implemented across the network.
• Challenges with laboratory data encoding were identified.

Main Results:

• Phenotype implementation using the OMOP Common Data Model was feasible across sites, though laboratory data presented challenges.
• OMOP phenotype execution took less than a day, a significant improvement over the weeks required for manual implementation.
• Agreement between OMOP and original eMERGE phenotypes varied (43%-100%), highlighting issues in original definitions and data.
• Standardized OMOP queries revealed inconsistencies in original phenotype implementations.

Conclusions:

• A common data model like OMOP can drastically accelerate phenotype implementation, offering a net benefit as more phenotypes are added.
• Sharing standardized phenotype code and logic via a common data model mitigates human error and reinterpretation issues inherent in narrative definitions.