Home     Research topics     Activities     People     Results     Contact

Landscaping

The project started with an inventory of biobanks and relevant databanks. Based on this inventory we prioritised databanks that were linked to biobanks. Criteria for prioritization included completeness, the availability of previous validation studies, feasibility and legal compliance for linkage to biobanks. This inventory has been made publically available.

Catalogue

Biolink NL developed two types of catalogues. The first contains key parameters of the biobanks and databanks (level 3 catalogue*). This catalogue will be made externally searchable through the internet, in collaboration with the UMC Groningen. A second and more detailed catalogue was built, in which a few variables on subject level from biobanks and databanks are available and where the data are linked (“level 5 catalogue”). In other words, this catalogue should help investigators to check the overlapping population between biobanks and databanks for their research project. Linkage of data in the level 5 catalogue is based on probabilistic record linkage algorithms.

This second type of catalogue is by far the most demanding in terms of technique, knowledge, governance, security and privacy. This catalogue will not be externally searchable, but can only be used as an internal tool for data managers who handle requests for information from investigators. These data managers have knowledge about the data, are trained in the use of the catalogue and are aware of the relevant privacy and security aspects.

This catalogue was set up in cooperation with the Foundation Mondriaan Health Research Data. Mondriaan has already developed a level 5 catalogue embedded in Privacy Enhancing Techniques and Procedures such as encryption and a Trusted Third Party. It is anticipated that modules developed by the UMC Groningen will be integrated in this catalogue to further structure the data.

1. *Explanation on the catalogue levels can be found under ‘research topics’.
   

Record Linkage methodology

The methodology of record linkage, the validity, reliability and impact of record linkage for research questions in terms of methodology are investigated in a simulation study followed by a number of demonstration projects. A variety of methods can be employed to link different data sources and these methods vary in complexity. Deterministic approach is certainly useful for certain applications; this method is relatively simple in terms of complexity, efficiency, and accuracy. However, it cannot easily handle errors in the linkage variables. Probabilistic methods are more complex and have the potential to identify more links as they can deal with errors in the variables. The theoretical and practical underpinning of the various methods for record linkage were studied together with the relative strengths of each approach.

Another important aspect is to investigate whether the developed methodologies can also be used in a setting where Privacy Enhancing Techniques such as encryption or a Trusted Third Party may impose practical constraints. For these methodological studies, we started off using data from several demonstration projects that represent different types of biobanks. In these demonstration projects an optimal methodology of record linkage was developed for different situations (e.g. biobanks with and without informed consent for record linkage, biobanks and registries with varying availability of personal identifiers), and for linkage with multiple registries at the same time. It was assessed which methods perform well in different scenario’s  with different linkage variables, cohort characteristics (size, coverage of the population, representation of specific diseases (high/medium/low incidence)).

Validation of linked records

Linked records can be used to cross-validate e.g. outcomes or questionnaires; repair missing data or enrich subject profiles. Criteria will be defined for validation studies, and selected projects will be executed to study the validity of linked medical record data, in order to assess the added value of enriching biobanks with other data for scientific research. We performed a literature review of validation studies for various (medical) data linkage projects and their results. Based on the linkages within our demonstration projects, we showed how to use and cross-validate linked data that came from various sources. We validated both the linkage procedures and the content of retrieved information.

Ethical, legal and social issues (ELSI)

Goal is to develop the ethical-legal framework for linkage of biobanks and registries, data requests, data sharing, and governance of linked biobank/registry data. Both the biobanks and the registries concerned are governed by the statutory (legislative) legal framework in general and self-regulating instruments (Code Goed Gebruik and Code Goed Gedrag). Some biobanks and/or some registries may be governed by a specific statutory legal framework (e.g. CBS-registries). Some forms of linking or linking for specific purposes, or linking specific data to specific registries may be governed by a specific statutory framework as well (collectively: the Regulatory Framework). In addition, the proposed linking will be subject to consent of (or waivers of consent, where appropriate), and to any associated restrictions or commitments to participants (the Consent Framework).

Surveys on the above mentioned frameworks of biobanks, registries but also ethical committees were conducted. We explored whether it is possible to create a competent body to oversee access to the joint data sets that will result of the linkage of biobanks and registries. The data (and sample, where appropriate) access policies of the biobanks and registries in the Netherlands will be investigated. From the results of this survey we sought to extract and draft common minimum standards to act as the point of reference for the conditions (ownership, IP, publications, feedback of both data and results to emitting biobank/registry) to which the linkages envisioned by Biolink NL will be subject.

We subjected the thus built legal framework to a reality check by applying it to the linkage projects conducted by Biolink NL. An overall ingredient of the above steps was to monitor possibilities to link the infrastructure at the European level as envisioned by BBMRI EU, by adhering as much as possible and as much as required to any standards developed within BBMRI EU. It is to be expected that the ELSI activities have a transcending purpose for BBMRI. ELSI activities will be tuned between BBMRI rainbow projects.

Project deliverables

Biolink NL aims to realise the following concrete deliverables:

1. National infrastructure for linkage of biobanks to registries (including technical, governance, and ethical/legal aspects).

2. Implementation of linkage of selected biobanks to specific registries.

3. A web based Portal including procedures for linkage and a central catalogue of linked biobanks/registries that can be searched for specific patient populations. The catalogue should contain minimal harmonized dataset and also informed consent information.

4. Assessment of validity and utility of medical data from various registers to determine medical outcomes for biobanks.

5. 5.Best practice guidelines for data sharing and record linkage, covering methodological, ethical and legal aspects (disseminated on portal).