Optimising cohort data in Europe

ways (Devriendt et al., 2021) while on the legal plane, the development of a General Data Protection Regulation (GDPR)-compliant model for the curation and access to cohort data requires extensive resources. Such differences in access arrangements make the re-use of existing data unnecessarily complex and expensive in terms of time and resources as many bureaucratic hurdles must be overcome before access to data can be granted, therewith compromising the exploitation of cohort data. y y A lack of functionality to share cohort data with different infrastructure types, which leads to management problems. Collaboration parameters between infrastructures (and thus the research institutions curating them) should be well defined. However, this is difficult to achieve because we still do not have a generic mechanism for interoperability of projects and methodologies. Moreover, data interoperability (both at individual level and at metadata level) focuses on the types of data, data flows and core standards that are the most likely to facilitate data sharing and data access. The aim is to achieve FAIR data on a consistent basis, that is, data that is findable, accessible, interoperable and reusable. Also, sharing information amongst systems relies on the adoption of interoperability standards (reference information models/templates and terminologies) (Do et al., 2011). For example, the clinical research and the health care domains each use different standards as “models of use”. CDISC standards are widely used in the clinical research domain, while in health care, the most widely used content and messaging standards are by Health Level 7 International (HL7). Additionally, the prominent terminology systems used are different (e.g. WHO Drug Dictionary [WHODD] or Medical Dictionary for Regulatory Activities [MedDRA] in clinical research versus Systematized Nomenclature of Medicine-Clinical Terms [SNOMED CT] or International Classification of Disease [ICD-10] codes in health care). The Food and Drug Administration (FDA) highlighted the interoperability problem between clinical research and health care in “Use of electronic health record data in clinical investigations”, a guidance document issued in July 2018, and intended to assist study sponsors, clinical investigators, contract research organisations, and institutional review boards (IRBs), in the use of electronic health record (EHR) data for FDA-regulated clinical investigations. In summary, there are barriers at many levels which hinder effective data interoperability. For example, the structure in which the source data are stored may be heterogeneous and unclear across cohorts (e.g. data sources are not always tracked, the rationale for the definition of variables is lacking), which can result in a loss of interoperability. Data sharing may be compromised by contradictory data access and data transfer procedures across participating institutions providing source data. The same applies even more for data from different cohort studies. The integration of different cohort studies into a European cohort repository may be prevented by the heterogeneity in used measures and instruments to assess a certain construct, as well as the use of different or ambiguous terminology. Finally, interoperability is further impeded through incompatibility due to the use of specific software for different cohort projects and their respective infrastructures.