Validation of a common data model for active safety surveillance research

TLDR

Systematic analysis of observational medical databases for active safety surveillance is impeded by variation in data models, coding systems, and the difficulty analysts face in understanding robust clinical data models, which also often lack the computational support needed for large‑scale systematic analysis across many interventions and outcomes. The authors propose translating data from idiosyncratic models into a common data model (CDM) to improve analyst comprehension and enable large‑scale systematic analysis while faithfully representing the source database. They instantiated the Observational Medical Outcomes Partnership (OMOP) CDM in a relational database, loaded data from ten observational healthcare databases, and developed and executed a comprehensive set of analytic methods to evaluate performance. The OMOP CDM accurately represented data from all ten databases using standardized terminologies, and the analytic methods performed sufficiently well to support active safety surveillance.

Abstract

Systematic analysis of observational medical databases for active safety surveillance is hindered by the variation in data models and coding systems. Data analysts often find robust clinical data models difficult to understand and ill suited to support their analytic approaches. Further, some models do not facilitate the computations required for systematic analysis across many interventions and outcomes for large datasets. Translating the data from these idiosyncratic data models to a common data model (CDM) could facilitate both the analysts' understanding and the suitability for large-scale systematic analysis. In addition to facilitating analysis, a suitable CDM has to faithfully represent the source observational database. Before beginning to use the Observational Medical Outcomes Partnership (OMOP) CDM and a related dictionary of standardized terminologies for a study of large-scale systematic active safety surveillance, the authors validated the model's suitability for this use by example.To validate the OMOP CDM, the model was instantiated into a relational database, data from 10 different observational healthcare databases were loaded into separate instances, a comprehensive array of analytic methods that operate on the data model was created, and these methods were executed against the databases to measure performance.There was acceptable representation of the data from 10 observational databases in the OMOP CDM using the standardized terminologies selected, and a range of analytic methods was developed and executed with sufficient performance to be useful for active safety surveillance.