On 18 June, the Royal Statistical Society's Emerging Applications and Social Statistics sections held a meeting about the ethics of research with pre-existing data, ie where specific, individual, informed consent was not available. It followed an RSS-sponsored ESRC National Centre for Research Methods event earlier in the year on Data Policy, which found statistical ethics emergent in the face of rapidly changing technology. This meeting was held in Manchester University, jointly with NCRM, to move the discussion out of London, and to come closer to some of the solutions like Connected Health Cities. Common themes were presented across different governance structures, legal entities and research activities, not least the public benefit basis, transparency and the demands on researchers.
Madeleine Murtagh of Newcastle University, who chairs METADAC (Managing Ethico-social, Technical and Administrative issues in Data ACcess), spoke first about governance, legislation, ethics, and some examples of challenges arising. Participant consent for large longitudinal studies is often broad in terms of general research use, but this means it has to respond to changing times. The potential for use of genetic data and data linkage meant socially sensitive research had been ruled out implicitly and would need revisiting. Plain language summaries were an important part of the transparency principle and public engagement is a critical component to understanding and maintaining the trust which is hard won, but so easily lost in a way that would be catastrophic for a large study.
Simon Whitworth of the UK Statistics Authority followed with a description of the processes for using existing data for statistics in government and now for research. The National Statistician’s Data Access Advisory Committee (NS-DEC) has evolved over more than three years and sees expanding demand with the provisions of the Digital Economy Act. Government statisticians have been challenged to specify the public benefit of better statistics - what are the better decisions which will arise?
Mary Tully of Manchester University and Connected Health Cities described the citizen jury process taken to engage with members of the public about health record sharing. In general, the public has positive attitudes which are conditional on providing a public benefit, however, some are strongly opposed. Directly attributable patient health benefits receive strong support but commercial proposals with health benefits much less.
Tony Calland of the Health Research Authority who chairs the Confidentiality Advisory Group (CAG) explained the role of CAG its legal basis and experiences. His experience as a GP led him to support the idea of care.data which would allow follow up to patients discharged from hospital. But he also appreciated the personal and privileged nature of information shared with GPs, which gave a duty of confidence. The Common Law duty of confidence from which CAG considers what can reasonably be expected also requires that there was a right to opt out. This is a lower bar than GDPR, which may mean that processes will be debated in the coming months.
Peter Elias of Warwick University described some of the recommendations of an ESRC supported OECD report on emerging data for social and economic research. He outlined the challenges of consent, including that seeking consent can bias your sample, and also offered some problems for the future. The separation between health data and other data is at odds with social reality in which health relates to most other aspects of life and society but it is currently excluded from research provisions. The role of data controllers and the legal basis they have for processing is extensive but the extent to which this confers consent for specific processing needs reflection.
The RSS is considering how to respond to this changing landscape, including setting up a new group on Data Ethics, to consider data stewardship. This will complement the existing interest of the Data Science Section in the ethics of AI, fairness of algorithms, and business ethics of data science practice. Suitable expertise within bodies responsible is patchy and thinly spread so RSS members are encouraged to engage with local and wider initiatives.