Optimising cohort data in Europe

Characteristics

Pooled analysis Meta-analysis

Federated analysis

Sharing and pooling data from different cohort studies is not always possible due to ethical, legal and consent restrictions

Non-harmonised data and too different study designs can entail big heterogeneities

It entails complex infrastructure and the development of specific statistical packages is still novel

Main disadvantage

Equivalent results when model assumptions and estimation methods are the same across studies − In all the types of analysis, it is necessary to check for potential different effects of confounding and exposure variables between studies. − The meta-analysis approach cannot be used to perform advanced statistical analyses such as those used in machine learning which are currently being used in personalised medicine. − The use of federated analyses is increasingly being considered in recent initiatives to overcome technical, ethical and legal issues, especially in cohorts that include omics data.

Comparison of results

Other remarks

2.3. Interoperability Interoperability, as defined by the Healthcare Information and Management Systems Society (HIMSS), "is the ability of different information systems, devices or applications to connect, in a coordinated manner, within and across organisational boundaries to access, exchange and cooperatively use data amongst stakeholders, with the goal of optimising the health of individuals and populations." In general terms, interoperability refers to the ability and the potential of information systems to share data, which depends, in particular, on the design and content of data structure, that is, the way data is organised. The interoperability of data infrastructures is the most important challenge for the optimisation of cohort data exploitation in Europe. This is in line with the concerns of the data intensive research community, where the lack of interoperability remains a major obstacle to the integration of cohort studies (Gaudet-Blavignac et al., 2021). Some of the main difficulties regarding data infrastructure interoperability include: y y A significant heterogeneity in data infrastructures among cohort studies. This prevents the establishment of a shared, harmonised repository of cohorts. y y The tendency from countries, institutions and research infrastructures to create their own data access procedures, resulting in heterogeneous (and sometimes contradictory) access arrangements on the methodological and legal plane. On the methodological plane, research institutions assign roles for data access in different