BACKGROUND: The IMI ConcePTION project was initiated to developan ecosystem for the generation of robust evidence on medicinessafety in pregnancy and breastfeeding based on pre-clinical andhuman data. Analysis of distributed and heterogeneous health datarequires standardized tools that run against a common data model(CDM). In the past 15 years many tools for distributed analytics,including CDMs, have been tested in Europe and beyond. We capital-ized on these learnings and propose a CDM to address the needs ofthe ConcePTION ecosystem.
OBJECTIVES: To develop a CDM that captures the richness of availableEuropean health data sources based on a process that optimisestransparency in mapping of content with separate syntactic andsemantic harmonization, to enable evaluation of medicines use, effec-tiveness and safety during pregnancy and breastfeeding.
METHODS: Data access providers (DAPs) involved in the ConcePTIONproject were selected on the basis of the data source they have accessto and experience in the field. Comprehensive qualitative interviewslasting 120 min were conducted between DAPs and pairs of inter-viewers. Data from the interview answer forms were extracted in astandard manner and synthesised to i) develop a framework fordescribing the data sources, ii) support the development of the CDM and iii) support the design of the Extract Transform Load (ETL) proce-dure to the ConcePTION CDM.
RESULTS: Twenty DAPs were interviewed, with access to heterogeneousbut complementary data (e.g. hospital discharge records, dispensed pre-scription records, birth and mortality registries, congenital anomaly reg-istries). The CDM we proposed based on the evaluation of theinterviews is divided into 4 sections: 1) Routine healthcare data (datagenerated during routine healthcare), 2) Surveillance data (data gener-ated out of routine healthcare, e.g. cohorts, ad-hoc registries), 3) Curatedtables (derived and summarized information of interest, e.g. birth date,mother-child linkage) and 4) Metadata (information about the datasource), without prior semantic harmonization of the vocabularies used.
CONCLUSIONS: The analysis of data source descriptions and interviewswith DAPs in the ConcePTION project demonstrated the unique fea-tures of European data sources and led to the development of theConcePTION CDM. This will allow rapid implementation of transpar-ent distributed analytics and will next be implemented in studies ofmedicines use, effectiveness and safety during pregnancy andbreastfeeding, and in the general population.
