Public Acceptability of Data Sharing Between the Public, Private and Third Sectors for Research Purposes

EXECUTIVE SUMMARY

Introduction

In 2012 the Scottish Government commissioned research to explore the public acceptability of cross-sectoral data linkage for research and statistical purposes to inform the ongoing development of a Scotland-wide Data Linkage Framework. The research indicated that the public was, in principle, broadly supportive of data linkage, particularly for health research, and of the overall objectives of the Data Linkage Framework. However, this support was conditional and a range of ambivalences and concerns were also expressed: there was significant unease about the private sector having access to public sector data and, more specifically, about the scope for commercial gain arising from data linkage (Davidson et al, 2012).

In Joined up data for better decisions: A strategy for improving data access and analysis, the Scottish Government acknowledged these concerns and outlined a commitment to continue working with members of the public and other stakeholders to explore fully the appropriateness, concerns, benefits and risks of private sector involvement in the use and analysis of data collected and held by public bodies.

Accordingly, in spring 2013 the Government commissioned Ipsos MORI Scotland, along with Dr Mhairi Aitken, Professor Sarah Cunningham-Burley, Professor Graeme Laurie, Dr Claudia Pagliari and Nayha Sethi from the University of Edinburgh, to conduct research to enhance understanding of sensitivities around data sharing between the public, private and third^[1] sectors for statistical and research purposes.

Specific objectives of the research were to establish: whether and how attitudes and sensitivities varied depending on the sectors and specific type of organisation involved in data sharing, the type of data being shared and the planned uses of the data; whether the public think about 'public benefit' differently in relation to the private and third sector's use of personal data compared to the public sector's; what methods of benefit-sharing are most acceptable to the public; and what methods could be most effective in empowering citizens in decision making about how their data are used.

The study comprised three elements: a desk-based literature review of international benefit-sharing models arising from the value of data sharing; a similar review of different methods that have been used to empower citizens in decision making about how their data are used; and a series of deliberative events with members of the public. The literature reviews were undertaken before the deliberative research so as to inform the range of issues and examples to be put to participants at the events.

Review of the literature on benefit-sharing models

Benefit-sharing is a concept that implies that the broadly envisioned advantages or returns from research should be distributed in a fair and equitable manner. Key legal instruments which have called for commitments to benefit-sharing were initially developed within International Law in order to address the distribution of rights to exploit and benefit from natural resources. The concept has begun to be applied in the research context, particularly in relation to biobanking. Benefit-sharing appeals to wider concepts of justice and fairness and is also invoked in the context of promoting trust. Benefit-sharing can mean different things to different people and any broad definition belies complexity in more practical terms; it is an evolving concept and in the context of data sharing, practical application is only just beginning. There have also been some criticisms of benefit-sharing, both as a concept (research outcomes should be considered benefits in themselves and shared de facto and research participation should be based on altruism) and in application (where models of benefit-sharing may legitimise commercialisation with limited attempts at distributive justice).

Six examples of benefit-sharing models in use were elicited from the literature reviewed, three of which were used as examples in the deliberative workshops. The models varied from clearly defined dispersal of net profits across different parties (Generation Scotland), through oversight of benefit-sharing protocols (Newfoundland and Labrador Model) to ensuring better health care through feeding back findings to practitioners (Estonian Genome Project).

Common features have been identified in the practical and conceptual models within the literature. Firstly, collaborative arrangements which attempt to provide positive social, economic and environmental outcomes for local communities by engaging with these communities. Secondly, contract-based models which ensure that the terms of the agreed benefit-sharing are kept, for example, by including points about dispute resolution, verification and incentives. Thirdly, community based partnerships which serve to make stakeholder engagement a core component, usually based on a legally valid agreement around community consultation. Fourthly, a participatory governance shareholder model argues that community consultation is not enough and partnership governance is required whereby donors/participants are treated as shareholders rather than stakeholders.

Despite the scarcity of research and evaluation on benefit-sharing in the context of sharing personal information and sharing data for research and statistical purposes, the literature review does offer some potential ideas for key elements that can be included in a benefit-sharing model. Most of the models included two independent bodies within their governance framework: one with ultimate authority for approving access to resources, and the second, which also scrutinised applications but which appeared independent and geared towards examining applications to ensure appropriate benefit-sharing mechanisms were in place. The independent scrutiny bodies were also employed with an oversight and monitoring function, to ensure that benefit-sharing measures were enforced. Some models have been praised in the literature for having included an element of public consultation or dialogue in their approaches but this has often stopped short of actual inclusion of lay members/donor representatives in any of the decision-making/oversight bodies.

Models which include direct participant representation may offer meaningful inclusion of participants, and provide them with some control in the decision-making process around what happens to collective resources, but issues remain about representation and interests.

Review of the literature on empowering citizens in decision making

Promoting public dialogue is a key component of the policy-making process; benefits include generating an understanding of public attitudes, stimulating greater public participation in civic life, and engendering trust. However, the extent to which public engagement is an empowering and meaningful practice remains a matter of debate. Public engagement comprises a range of goals and purposes; different underpinning orientations may reflect normative, instrumental or substantive positions - normative in the sense that public engagement should be part of our democracy; instrumental in terms of building trust in government; and substantive through including citizens in decision making for improved social outcomes.

Public engagement also comprises a range of activities which can map onto these diverse positions. Although there are different typologies available, these can be assembled into three broad, yet overlapping, arenas of practice: awareness raising, which promotes dissemination of information to apprise and educate; consultation, which seeks to elicit information from the public in order to inform decisions; and empowerment, which promotes citizen control through active participation. Awareness raising can promote public acceptance and includes methods such as a media campaigns or leaflets. Consultation can support the development of socially acceptable policy and may be achieved through surveys or qualitative research. Empowerment can enhance democracy through capacity building and may involve dialogic methods such as citizens' juries or participatory appraisal.

The literature review identified examples of awareness raising, consultation and empowerment and in many cases the studies reviewed can be seen to have aimed at and/or achieved more than one of these goals. Most, however, involved consultation and/or awareness raising. Empowerment was promoted in 12 of the 51 studies reviewed; this was achieved in combination with consultation and awareness raising, demonstrating that activities that inform are important across the range of public engagement activities. These studies can be viewed as empowering participants through devolving control over the engagement processes and outcomes, the development of new skills, greater understanding of issues under discussion and increased confidence to participate. Very few studies explicitly evaluated the engagement process but there was some evidence that participants appreciated the opportunity to take part in these activities and found the experience rewarding. Evaluation against clear criteria is required to assess both the process and outcomes of participatory engagement.

Deliberative research findings

Sensitivities around data sharing between the public, private and third sectors

As in the 2012 research, there was significant concern about the potential for shared personal data to be hacked or otherwise obtained by unauthorised individuals or groups. This concern was largely informed by past, high profile cases of public officials leaving laptops, data sticks or hard copy documentation in public places. When reminded that data shared for research purposes would be anonymised, most participants appeared to feel reassured that this would mitigate against personal details falling into the wrong hands, though a significant minority were more sceptical.

Still, there was near universal acceptance of public bodies - including the Scottish Government, the NHS, local authorities and the police - having access to anonymised personal data from other organisations (whether public, private or third sector) for research purposes. There were two main factors underpinning this support. Firstly, there was a commonly held view that public sector organisations were concerned with delivering public benefits or promoting the "public good" and that any research they undertook would be similarly oriented. Secondly, and despite concerns about public officials misplacing or losing data, there was an assumption that public bodies had more stringent data protection and security procedures in place than other types of organisation - or at least were more accountable to the public than those other organisations when a breach occurred.

Private sector involvement in data sharing was a more contentious issue and there was strong spontaneous opposition to data being used by the private sector for the sole purpose of profit maximisation. This is not to say that participants were entirely opposed to private sector organisations accessing data, or that no level of profit was acceptable. Rather, the consensus was that private sector access to personal data should only be granted where this is likely to result in some form of public benefit.

Public benefit tended to be conceived of primarily in terms of improvements to local services, local areas or public health, rather than individual-level or direct financial benefits. However, there was some unprompted suggestion that the private sector should be required to pay for access to data and/or to share any profits resulting from research with the relevant data owner(s), so as to generate funds that can be reinvested towards the public good.

Third sector access to data was seen as more acceptable than private sector access but somewhat less so than public sector access. There was a widely held assumption that many third sector organisations, like public bodies, are concerned with promoting the public good. At the same time, there was a perception of the third sector as something of an unknown entity, which prompted feelings of unease about its organisations having access to data. There was a tendency for participants to draw a distinction between charities on the one hand, and pressure groups on the other, with the former generally trusted more than the latter. It was commonly suggested that pressure groups had agendas to advance that may or may not reflect the interests of the general public

When participants were asked about the relative acceptability of different data types being linked and shared for research purposes, they tended to begin by saying that it depended on who would be accessing the data and for what purpose, reinforcing the centrality of trust as a determinant of views. Still, they did distinguish between different data types, with three types in particular provoking considerable discussion and debate across the events: postcode data; sexual orientation data; and commercial data.

There was widespread concern that a focus on postcode data in research could result in areas being negatively labelled - for example as "deprived", "rough" or low-achieving - and, subsequently, in residents of those areas experiencing stigma or discriminatory treatment. Additionally, there was a view of postcode information as potentially disclosive. This view was most common among people in the more rural locations of Oban and Galashiels several of whom contended that individuals could "quite easily" be identified within a dataset focusing on a small geographic area from a combination of their postcode and other data types.

Data on sexual orientation was most commonly a focus for LGBT participants; most of whom expressed ambivalence on the matter of such information. On the one hand they emphasised a need for better, more accurate data on sexual orientation to increase understanding of LGBT issues and inform improved service provision for these groups. On the other hand, there was trepidation about the potential for such data to fall into the wrong hands and be misused, particularly in the event of its being 'de-anoymised', rendering individuals potentially identifiable.

With regard to commercial data, there was general opposition to the idea of banking and other financial information being shared with any third party organisation, whether public, private or third sector. This reflected a view that financial data, even in anonymised form, is very private and should remain so. A similar level of opposition was expressed over the sharing of data held by internet providers and social media companies, with this data too regarded as too personal to be shared.

Benefit-sharing

Benefit-sharing was perceived to be important and necessary, although it was clear that participants thought of benefits differently in relation to the public and third sectors' use of data compared to private sector use. While they perceived the realisation of benefits as an inevitable goal of research conducted by public and some third sector organisations, the prevailing view was that the main goal of private sector research was ultimately to generate profit and, as a consequence, their access to data should be more strictly controlled. Reflecting these views, benefit-sharing models were commonly seen as more relevant in the case of data sharing involving the private sector compared to the public and third sectors.

Benefit-sharing models were mainly conceived of in terms of who should benefit from data sharing and how they should benefit, with discussions focusing mainly on data subjects and data users.

In relation to data subjects, a clear distinction was drawn between research requiring their proactive participation and research that draws on routinely collected administrative or statistical data. With regard to the former, the consensus was that data subjects should benefit directly, with suggestions for direct benefits ranging from financial incentives to advice and access to services. With regards to routinely collected data, the prevailing view was that individual data subjects should not necessarily benefit directly and, instead, society in general, or the specific population to which the data relates, should be the main beneficiaries. 'Societal benefits' were conceived of in terms of primary benefits - outcomes which arose directly from research, such as improved services - and secondary benefits - broader and longer-term benefits, such as better public health.

Discussions pertaining to data users focused primarily on private companies. Despite participants' initial aversion to the idea of companies profiting from research using shared data, the discussion of benefit-sharing led to the emergence of more nuanced views. There was general recognition that profits provided an incentive for private companies to invest in and conduct research, which, in turn, contributed to wider economic benefits. However, participants remained concerned about the potential for private companies to make excessive profits. Consequently, they felt it important that benefit-sharing models incorporated clear provisions to curb the level of profits made from research. Specific suggestions included provisions to ensure that private companies: pay to use data; share profits; provide affordable products and services; or reinvest profits in local communities.

Aside from data subjects and data users, participants identified other countries and future generations as groups that might become beneficiaries of research. There was a view that recognising these groups in benefit-sharing models, particularly in cases where it is difficult to identify an immediate benefit to data subjects or wider society, would go some way towards demonstrating that the research is in the public interest.

Empowering citizens in decision making

There was unanimous agreement that public involvement in decision making on data sharing, including the development of benefit-sharing models, was important and appropriate. In terms of specific forms of involvement, there was a stronger appetite for transparency, feedback and consultation than for more active forms of involvement such as agenda-setting and representation. There seemed to be two main reasons underlying this.

First, while participants felt that the public would be able to contribute to more general discussions about how their data should be used, there was a view that most members of the public did not have the requisite knowledge and expertise to contribute to more specific decisions about the types of research that should be carried out, and other similarly complex issues. Second, it was commonly felt that most people were either unwilling or unable (due to family and work commitments) to devote time to acting as public representatives.

Reflecting participants' views on potential forms of public involvement in decision making, three methods of involvement were commonly mentioned.

First, there was strong support for the idea of setting up an oversight body, comprising a range of stakeholders, to oversee data sharing. Participants felt that public interests should be represented in the oversight body in an indirect way - there was a specific suggestion for a third party organisation that would regularly consult with the public and feed back to the oversight body accordingly. There was a preference for this consultation to inform general principles around data sharing (including, rules on what information is made available, who should be able to access it and governance arrangements) rather than to solicit views on specific requests for access to data.

Second, the internet was frequently mentioned as a potentially useful and cost-effective way of engaging the public in decision making around data sharing. A popular suggestion was for a website that could act as a 'one-stop shop' for information on data sharing.

Third, participants, mindful of the limitations of a solely online approach (specifically concerning those without internet access), often felt any such approach should be supplemented with an initial television-based public awareness-raising campaign in the style of a party political broadcast or a public health campaign.