Accessing transaction records for linkage in longitudinal cohort studies: how can researchers safeguard participants?

Scoop.it ShareThis

News

gender pay gapTransaction records provide longitudinal researchers with a novel source of data with which to link with cohort data, giving potential for new insights into lifestyle choices and health outcomes. But how willing are the public to share details of their banking and loyalty card records with researchers, revealing not only details about their weekly shop, but also bank balance, salary and other potentially sensitive information?

In order to generate evidence on public opinion in this domain, 20 participants from Children of the 90s were invited to join a series of focus groups exploring their thoughts on sharing transaction records for linkage into the Children of the 90s databank.

Kate Shiells, a Research Associate from the University of Bristol, explains what was learnt from these conversations, and how findings from the study could provide researchers with a range of safeguards for recruiting and involving participants in similar projects.

How will my data be used?

Initially, we found that the majority of participants were unsure of how their transaction records could be useful for research and wanted to understand how this could benefit future generations. Attitudes evolved over the focus groups as participants learnt more about the application of transactional data in health research. In particular, participants appeared more willing to share their transaction records for research with clear benefits for society, emphasising the influence of prosocial behaviour in donating data. This indicates the importance of providing participants with information that highlights the impact of the proposed study.

Questions also arose around the potential accuracy of the data derived from transaction records, given that much of what you buy may be consumed or used by someone else. Therefore, we recommend that any issues associated with accuracy should be addressed by researchers during consenting procedures. Furthermore, participants were curious who their data would be shared with. Sharing data with commercial companies is not permitted within the Children of the 90s framework. However, due to the nature of the type of data under discussion, a common concern amongst participants was whether there could be an impact on their credit history or insurance premiums should their data be shared commercially, and so reassurance that this will not take place may be required.

Can I trust researchers with my data?

A second common theme running through the conversations was trust. Having been part of the Children of the 90s cohort since birth, the participants of our study already had an understanding of the importance of various aspects of research practice and were confident in the standards upheld by Children of the 90s. For instance, they were aware of how pseudonymisation was used to protect identities. However, some were wary about whether external researchers could be trusted not to misuse data. Participants alluded to Cambridge Analytica as an example of how personal data can be harvested and used to manipulate individuals. Losing control of data when it is linked and where it might ‘end up’ was also a cause of concern.

If participants are to feel confident in sharing their transaction records for longitudinal cohort studies, they will require information about the legitimacy and trustworthiness of third-party data users, which could be achieved by providing participants with clear information about study objectives, as well as details covering all aspects of research integrity, such as anonymity, confidentiality, and encryption procedures.

How much control will I have over my data?

Participants spoke about the need for control over their data and choice about how it is shared. Firstly, some wanted choice over the field of research for which they share their data, such as research into different diseases or mental health. There were also ongoing conversations about what constitutes sensitive information across the focus groups, with varying levels of sensitivity associated with different data types. This led participants to stipulate that they should have a choice over the types of data they donate, with categories on the consent form covering data revealing purchased medicines, salary or gambling. Participants felt that data revealing details about third parties associated with their transactions could also be classed as sensitive, as the third party would not have consented to have their data used in research. Likewise, any information relating to children was agreed to be sensitive and should not be shared. A distinction was also made between online and instore purchases, with participants explaining how sometimes purchases made online can be associated with privacy preserving behaviours.

A third key element of maintaining control over data was related to consent mechanisms. Although concerns were raised over both prospective and retrospective data collection, retrospective data collection was seemingly preferable for most participants, as they felt less vulnerable knowing what transactional data had been generated. However, fears around prospective data collection were largely linked to a misunderstanding around the concept of live data, mainly its associations with tracking. We suggest that participants are provided with the choice to share prospective or retrospective data, with associated risks explained. As regards to forms of consent, participants were largely happy with an initial opt-in consent to extract their transaction records for storage in the Children of the 90s databank, with an ongoing option to opt-out for the subsequent reuse of data and a yearly check of consent agreed to be most appropriate.

We hope that our findings have contributed initial evidence towards generating an ethical framework for accessing transactional datasets for use in longitudinal population studies. Such frameworks are imperative for ensuring that researchers follow robust ethical and privacy preserving measures during the integration of novel datasets with longitudinal cohort data. Further research with the general public exploring attitudes towards the donation of transaction records for research will take place in 2020 as part of the Data Donation project, led by Dr Anya Skatova, Turing and Vice Chancellor’s Fellow at the University of Bristol, and lead author on this paper.

Citations

Skatova A, Shiells K, Boyd A.  Attitudes towards transactional data donation and linkage in a longitudinal population study: evidence from the Avon Longitudinal Study of Parents and Children [version 1; peer review: awaiting peer review]. Wellcome Open Res 2019, 4:192 (https://doi.org/10.12688/wellcomeopenres.15557.1)

Skatova A, Goulding J (2019) Psychology of personal data donation. PLoS ONE, 14:11: e0224240. (https://doi.org/10.1371/journal.pone.0224240)

Wellcome’s Longitudinal Population Studies Working Group (2017) Longitudinal Population Studies Strategy. (https://wellcome.ac.uk/sites/default/files/longitudinal-population-studies-strategy_0.pdf)


Kate Shiells is a Research Associate at the University of Bristol.