OBJECTIVES: The Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) provides a structure for organizing and standardizing patient data coming from disparate data sources. OMOP-CDM has never been implemented on French data. This study describes the conversion and evaluation of the French Disease Analyzer database (DA) to the OMOP-CDM model. The converted database (DA-OMOP) was validated by replicating a published study performed with an equivalent EMR database (Longitudinal Patient Database, LPD).

METHODS: DA database (QuintilesIMS, EMR France) is an electronic medical record database of 1200 General Practitioners in France. DA data were mapped into CDM following the OMOP V.5 specification and converted into different domains using correspondence tables to convert native language into common language. Accuracy and completeness of the information were evaluated. Conversion was tested with the replication of a published study regarding antithrombotic treatments in patient with Atrial Fibrillation (AF) conducted with LPD (QuintilesIMS, EMR France). Results obtained with DA-OMOP were compared with those obtained with DA native and LPD

RESULTS: In total 99% of disease codes, exams, procedures and devices codes and 88% of drug codes were converted. Missing codes (12%) were related to homeopathy and OTC treatments. In DA-OMOP, mean age of AF patients was 74.4 years, (versus 74.6 in DA and 74.6 in LPD), 58.4% were men (versus 58.4% in DA and 59.5% in LPD) and 81.3% had a CHADs score ≥1 (versus 81.6% in DA and 83.1% in LPD). Additionally, 51.9% of thromboembolic high risk patients (45% in LPD) were not treated or inadequately treated according to ESC guidelines and 62.5% of patients with a score CHADs=0 (against 66.4% in LPD) were over-treated.

CONCLUSIONS: Agreement between DA-OMOP results and those of LPD indicates a satisfying conversion and provides a robust validation for DA-OMOP. This will enable the integration of French data set into the international OMOP network.