Network requirements
A strategic priority for DARWIN EU® is to expand the scope of the network for generation of new evidence supporting the benefit-risk decision making by the EU Regulatory Network. In collaboration with EMA and the Scientific Panel, the Coordination Centre shall seek to identify new data partners and data sources and integrate new infrastructure and techniques into the network of data partners.
The data partners enable DARWIN EU to use their data in its scientific studies, and provide analysis results to the DARWIN EU Coordination Centre, in accordance with data protection rules. They can be part of the data network directly or through a Data Permit Authority, as shown in the figure below.
DARWIN EU is expected to bring on board a minimum of ten data sources every year.
The following requirements are defined for candidate data sources:
- Data sources collecting health data routinely and representative of the different types of real-world data in terms of data elements, setting (primary & secondary care), population and origin (e.g. electronic health care records, claims)
- Data sources which collectively provide a broad geographical coverage
- Data sources containing patient-level data with a unique patient identifier linking all records relating to a given patient
- Medicines prescribed or dispensed identifiable with quantities (e.g. doses, package size) and dates allowing to calculate cumulative doses and duration of use and linked to individual but unidentifiable patients
- Clinical events formally coded, with accurate dates and linked to individual but unidentifiable patients
- Data already converted or planned to be converted into a common data model
The current Data Sources were selected after consultation with the DARWIN EU Advisory Board and following the prioritisation criteria below:
- Continuous data collection (at least annual data updates)
- A lag time of less than six months in data availability for analysis and capture of health outcomes and medicines prescribing or dispensing
- Data converted into the Observational Medical Outcomes Partnership (OMOP) common data model