Ethically and legally appropriate
This joint research project featured as a topic during the conference on 27 September 2017 which marked the Center for Big Data Statistics (CBDS)’s first anniversary. Project leader Bob van den Berg of CBS and UM researcher Johan van Soest gave a presentation on this subject. ‘The more datasets are linked, the more information is generated. A major question for both organisations is how to ensure safe handling of big data during this linking process,’ Van den Berg explains. CBS collects and manages all data with extreme care. The outcome of research must never be traceable to individual persons. Van den Berg: ‘Having said that, the emergence of big data and the opportunities it harnesses constantly give rise to new potential issues. Is working together as research institutions and combining our data an ethically and legally sound thing to do? Would we be able to collaborate without sharing datasets? How does one safeguard the privacy of research participants? What about in the future? Being a statistical authority, CBS considers clear answers to these questions to be of vital importance.’
Exchanging of data in a technically safe manner is not terribly complicated, says Van den Berg. ‘We have found a good solution for that. The challenge is minimising exchange of data and organising the process in a way that is legally, ethically and socially responsible.’ Van Soest: ‘From the point of view of privacy, we want to share as little detail as possible at the level of individual patients. On the other hand, research makes the availability of specific details compelling. How do we find the right balance?’ CBS and Maastricht University are addressing this question on the ground using the Maastricht Studie health survey as a test case. This survey is conducted among the population of South Limburg on prevention, causes and treatment of diabetes. The study entails monitoring of local residents over the course of five years using physical examinations and questionnaires. Van Soest: ‘This yields enormous amounts of data. To extract even more information, we want to link this to CBS’ data; CBS has data available such as lifestyle, income, family situation and home environment of these participants.’
‘The challenge is minimising exchange of data and organising the process in a way that is legally, ethically and socially responsible.’
Trusted third party
One way of achieving this is having a so-called trusted third party bring together the university’s data and CBS’ data. This third party is trusted by both sides and acts as a host where data are stored and encrypted temporarily for analysis, and who subsequently provides the health researchers with the results only of that analysis. Van den Berg: ‘This will be the first method to look into for the moment. What is required for this approach, what needs to be communicated to the survey participants before we can proceed? Is this socially acceptable? In a follow-up phase, we will also be looking at other methods, e.g. the university using CBS data at a higher level of detail rather than linking data directly.’ Van Soest: ‘This would be easier in terms of privacy, but whether it produces the desired details in order to address the specific research question remains to be seen.’
Meanwhile, a proof of concept for the chosen method is ready which has been agreed upon by legal experts from both CBS and Maastricht University. As soon as the Maastricht Studie participants have approved linking of the data, Van den Berg and Van Soest can go ahead. Their expectations are high. Van den Berg: ‘There are numerous medical studies for which linking of data is the desirable approach. But it applies equally to other research fields, for instance energy consumption or travel behaviour. In due time, our results will be applicable to a wide range of areas. CBS’ ambition is to take on the role of trusted third party in the future, as a separate unit operating within CBS, as well as for outside parties looking to collaborate. Against the legal background of the Statistics Netherlands Act and given our know-how and experience, we are the best placed party to do the job.’
At the CBDS, Statistics Netherlands bundles its expertise in the field of big data technologies and methods with those of more than 40 partners from the public and private sectors, science and education. The research project described above recently obtained a large amount of subsidy, which will finance the research for the coming two years.