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

3 LITERATURE REVIEW: EXPLORING BENEFIT-SHARING

3.1 This chapter considers key issues related to benefit-sharing in the context of data sharing for research and planning purposes. It describes key findings from a desk-based literature review examining key concepts and methods of benefit-sharing.

3.2 The evidence base for this chapter has been developed primarily from a literature review of relevant legal and sociological materials, including published peer-reviewed papers written in English, the grey literature (including official reports, primary and secondary legislation) and relevant books.

3.3 Scotland and, more generally, the wider-UK have developed an impressive international reputation for conducting research on electronic records. Such records can be used for diverse purposes, including administrative planning, quality assurance and for conducting health, social and economic research. The increasing number of initiatives dedicated to electronic-record based research demonstrates the perceived value of such data.

3.4 Alongside the potential benefits of using these records are associated risks. For example, matters become complicated where the outcomes of the research can lead to commercial profit, particularly where research conducted within the private sector is based on data which emerges from the public sector. The question of how such benefits can and should be shared between stakeholders (including citizens and those who store and use the data) must be considered.

3.5 This chapter builds on previous interdisciplinary work carried out in the field by: conducting a cross-sectoral survey of relevant benefit-sharing models, drawing on lessons learned, and making recommendations about how issues around 'benefit-sharing' might be approached in the context of data sharing.

3.6 The chapter details the results of a literature review, conducted in order to survey the extent of proliferation of (real and theoretical) benefit-sharing models in the field of data sharing, and to uncover evidence of any evaluations of their effectiveness that have been conducted to date. It provides a synthesis of common features of viable benefit-sharing models, and includes a commentary on these features and their potential impact on the design and delivery of benefit-sharing.

3.7 More specifically, the chapter considers how a working definition of benefit-sharing might be constructed, noting that it includes but is not restricted to wealth, health and social benefits. It also offers key exemplar areas of benefit-sharing. The report focuses on known pressure points of concern which previous research has highlighted (including the relevance of the public/private nature of the funding and use of data resources).

3.8 This chapter is organised in three parts:

Firstly, it introduces the concept of benefit-sharing and its origins in order to set the context of the discussion on how this concept relates to the research setting. The literature review considers the concept of benefit-sharing as has been discussed in other sectors (of most relevance for this research, in the context of biobanks). However, it has to date remained relatively under-explored in relation to sharing electronic patient records and regardless of context, little concrete evidence exists around what benefit-sharing models look like in practice.

Secondly, it offers an overview of actual and theoretical/conceptual benefit-sharing models which have been advanced, considering their key features, strengths and limitations in the context of how they might be translated to the data-sharing context.

Thirdly, it includes a discussion of the key considerations which must be factored in to any benefit-sharing approaches which might be adopted. The chapter concludes by offering key considerations for the Scottish Government moving forward, and by suggesting key themes for exploration in the deliberative work.

Introduction to Benefit-sharing

3.9 There is increasing recognition of the potential value that can arise out of research based on large-scale datasets. Much of the data used in research originates from the public sector, and the benefits that can arise from research are diverse, including economical and societal gains such as improved planning and wellbeing. Ensuring that we are making the most that we can out of this type of research is becoming a priority. For example, within the health sector, the Academy of Medical Sciences (2011: 57) has stated that it is 'crucial' that data is accessible across the population in order to maximise the potential benefits of health research. Indeed, the recent Caldicott 2 Report (2013: 21) has introduced a new principle which explicitly stresses to those responsible for granting access to patient data that 'the duty to share information can be as important as the duty to protect confidentiality'.

3.10 The value of such research stretches far beyond the health sector too. In addition to the Scotland-wide Data Linkage Framework aimed at maximising the potential of statistical data, the Administrative Data Taskforce (2012) has recently recommended the establishment of four UK-wide Administrative Data Research Centres in order to better facilitate research based on administrative data.

3.11 An important issue related to research is benefit-sharing. Benefit-sharing is a concept which, broadly stated, implies that the benefits (and burdens) of research should be 'universally shared' in a fair and equitable manner (Sheremata and Knoppers 2007: 160).

3.12 Key legal instruments and guidance which have called for commitments to benefit-sharing were initially developed within International Law (Sheremata and Knoppers 2003) in order to address 'how to distribute rights to exploit and benefit from natural resources' (Hayden 2007: 734). It has been argued that having a fair agreement of benefit-sharing in place can avoid exploitation of individuals (and their resources) (e.g. the Convention on Biological Diversity 1992, Bonn Guidelines 2002 and Human Genome Organisation 1996).

3.13 Additional rationales in favour of benefit-sharing have been advanced within the literature; it is argued that sharing benefits arising out of collective resources addresses notions of justice and fairness. Trust is also seen as a rationale for benefit-sharing (Laurie and Hunter 2004, Knoppers 2000). It is also thought that benefit-sharing itself can lead to 'increased medical interventions to improve health [and] increased knowledge about health' which can be made available to citizens, can lead to job creation (particularly in pharmaceuticals, and research more generally), and to profits for commercially oriented research companies (Schroeder and Gefenas 2012: 20).

3.14 When we turn, however, to consider what benefit-sharing actually means in practical terms, things become more complex. First, the concept of benefit-sharing is constantly evolving (Knoppers 2000) because different benefits and burdens can arise depending on the type of research under discussion. For example, the potential benefits and risks associated with clinical testing of medications on humans will be different to those from research related to using information from public sector datasets.

3.15 Second, 'benefit-sharing' can mean different things to different people. Some definitions are particularly broad: one definition describes it as sharing the benefits of whatever results from using 'biological resources, community knowledge, technologies, innovations or practices' (Organisation of African Unity: 2000). As a technical term, benefit-sharing 'describes an exchange between those who grant access to a particular resource and those who provide compensation or rewards for its use' (Schroeder 2007: 205). This technical definition also illustrates how broadly benefit-sharing can be defined, and the importance of articulating not only what 'benefit' means, but also who stands to receive and provide benefits.

3.16 We consider working definitions of benefit-sharing in section 4 below, but when reading this chapter, it is important to keep in mind that BS is a broad and evolving concept and arriving at consensus on what constitutes a fair agreement is not a straightforward undertaking, let alone arriving at a consensus on how different contributors might benefit and in what way.

3.17 In the data-sharing context, the key question which arises is how to ensure that the benefits (and burdens) of research are fairly and equitably shared between the users and the contributors/managers of the data. This is particularly important where private sector organisations stand to benefit from the use of personal information contributed by citizens (and particularly where benefit is in the form of commercial profit) and maintained by the public sector. On the other hand, unrealistic expectations can be made about benefits, particularly when considering the extent to which companies actually profit from data-based research - not every research study will result in 'positive' outcomes, let alone considerable, if any, financial gain.

3.18 Although little discussion has taken place around benefit-sharing in the specific context of sharing electronic records for research, some lessons can be learned from previous approaches and experiences in other sectors. Benefit-sharing has been discussed in other settings including forestry partnerships, corporate social responsibility and biobanks. As we will see, discussion tends to be very aspirational and there is a lack of guidance on how benefit-sharing looks in practice.

3.19 Benefit-sharing raises many important questions which must be considered before determining which approaches might be best suited for this particular context. First and foremost, we must consider which benefits will be identified and offered. Further questions then arise including (but not exclusively) the following:

• who decides what constitutes a benefit?
• how would benefits be distributed?
• would benefit-sharing take place according to the number of people who contribute to the project, how much effort participation involves, and/or some other factor? (Millum 2012)
• how will contributions be weighed? : For example, will contributions be made by individuals about whom the data is based, those who collect the data, those (organisations) that manage the data resource, the scientists who analyse the data, 'the institutions that trained the scientists, and the distribution of benefits?
  

3.20 Benefit-sharing approaches which have been adopted and considered in the case of human genetic databases or biobanks may offer the most in terms of guidance around how it should be approached when sharing electronic patient records. However, given the novel nature of constructing benefit-sharing mechanisms in this particular setting, additional (and potentially separate) concerns unique to data-linkage may arise. For example, individuals may view sharing genetic information in a different light to sharing non-genetic information.

Arguments against benefit-sharing

3.21 The aim of this review is to explore the issues involved with benefit-sharing rather than to assess whether benefit-sharing is in itself an appropriate mechanism. However, before proceeding, it is worth noting that there is some scepticism and criticism of benefit-sharing. For example, Chadwick and Berg (2001: 321) note that where an emphasis on benefit-sharing is made, this may be with the intent to 'buy people off' rather than to strive for distributive justice.

3.22 For others (particularly when considering human genetic research), the outcomes of research should be considered as benefits in themselves. Some consider that altruism should be the guiding principle for contributors to research (Schroeder 2007, Kadri 2007). An altruistic approach implies that we act in the interest of others without expecting personal benefits.

3.23 Some have viewed benefit-sharing as a way of trying to legitimise commercialisation and profit, arguing that financial or other benefits should not compete against 'overarching values like human health and quality of life' (Kadri 2007: 5). Barr et al (2006:110) also warn of the risks of inadequate arrangements for sharing of benefits, with communities signing away their rights. This can happen where information sharing and consultations are inadequate. This flags up the importance for the Scottish Government in ensuring that all stakeholders are involved in consultations and that appropriate information sharing procedures are put in place. It is also important to remember that benefit-sharing does not negate the need for informed consent to research where the law requires this (Chadwick and Berg 2001).

3.24 Another overarching concern is that benefit-sharing can give rise to elite capture - that is, where resources which are meant to benefit the larger population are actually only used by a few individuals or groups in power. Again, including all stakeholders in the consultation process in a meaningful way, as well as ensuring careful and transparent oversight of research can serve to lower risks of elite capture.

3.25 All of the risks mentioned above are not recounted in order to imply that benefit-sharing should be avoided, but rather, they serve to stress that we should be aware of the potential failures which can arise where inappropriate models or approaches to benefit-sharing are adopted. Given the potential importance of the role of benefit-sharing in research and the increasing number of initiatives dedicated to maximising the potential benefits of research arising out of data-linkage, it is paramount that questions around benefit-sharing are explored now in a robust, open, transparent and inclusive manner.

3.26 Governments can play a positive role in facilitating benefit-sharing agreements where they have in place mechanisms and policies for fostering partnership, dialogue and negotiation. There is a real opportunity for the Scottish Government to shape an effective and efficient approach to benefit-sharing which reflects the concerns of the public.

Benefit-sharing models

3.27 This section outlines some benefit-sharing models discussed within the literature. There is a particular (but not exclusive) focus on models discussed in the context of large-scale DNA databases/biobanks because those discussions are most likely to raise issues which are also relevant to the use of large-scale electronic datasets for research and planning purposes.

3.28 The review below includes an overview of key elements and structures incorporated (or proposed for inclusion) within each model, and comments on the associated strengths and limitations of the models. Special consideration is given to how well each model (or its key elements) could be transplanted to the data-sharing setting.

3.29 The literature review incorporated sources including published peer-reviewed papers written in English, the grey literature (including official reports, primary and secondary legislation) and relevant books.

3.30 The methods used in this literature review are described in chapter 2. Bibliographic searching focussed on the last 10 years (although some sources which extended beyond this timeframe were of relevance), including international literature. Key words identified for searching included: benefit-sharing, profit share, access, commercialisation, and evaluation. Reference lists were reviewed in order to ascertain further articles not identified through searches. The titles and abstracts of the resultant literature were reviewed in order to distinguish relevant and less relevant articles. In the end 26 journal articles were included in addition to reports, a doctoral thesis, primary and secondary legislation, and websites. Academic colleagues with experience on the concept of benefit-sharing were also consulted, and sources which they recommended were included.

3.31 In order to identify, assess and note the most useful elements and considerations contained in different models (and the wider literature), a bespoke template was constructed and each source was tested against the template. The criteria included reflect key considerations which must be factored-in to benefit-sharing approaches. The criteria are:

• the model being discussed (and context) and where applicable, the type of data being shared
• key features of the model
• any working-definitions of benefit-sharing which were offered
• examples of benefit to include wealth, health, social, cultural and others;
• nuances between public/private access to/uses of data
• points of concern (including the relevance of: the public/private nature of funding of data resources; purposes for which data are used; and commercial objectives in access and use)
• a note of or discussion on net contributors/beneficiaries in the model;
• whether or not public dialogue was developed in the construction of the model and if so, to what extent and
• any additional comments/observations

3.32 The analysis begins with an overview of concrete models which have been implemented in practice, beginning with national and subsequently international examples. These practical examples serve to help the reader develop a more concrete idea of how benefit-sharing has been implemented in practice, and to indicate the challenges and limitations of BS models.

Actual models of benefit-sharing

Generation Scotland

3.33 Generation Scotland (GS) is a collection of human biological samples from over 30,000 participants, which are made available for medical research. It involves a partnership between Scottish University Medical Schools, NHS Scotland and research participants. It receives core funding from the Scottish Government and through collaborative research (Generation Scotland 2013).

3.34 GS initially had 2 oversight bodies:

• a Scientific Committee which consists of members of all participating institutions including doctors, scientists and project managers. The committee has overall decision-making authority for overseeing access to and uses of GS resources.
• the Advisory Board was appointed by the Scottish Executive. It consisted of experts in genetics, law and ethics and would oversee the work of the Scientific Committee advising on conformance with the Governance framework, ensuring that Generation Scotland resources are used in the best interests of the public and advising on ethical issues that may arise during the study for genetic and health related research.

3.35 In its policy on access and publications, GS stressed that participants or their heirs would not receive payment for any profits made from use of their samples, nor would individual patient contributions be acknowledged. Participants would not receive direct benefits apart from basic health advice (feedback to the patient and, with permission, their GP, on basic clinical information, for example, height, weight, BMI, blood pressure, serum cholesterol, glucose, and smoking status).

3.36 In terms of sharing any income that would arise out of projects, the GS policy set out that any net profit (once patents and legal costs were paid) from commercial exploitation would be divided between the parties as follows: 25% to the University parties; 25% to the NHS parties; 25% to the party/parties which own the rights to the commercial projects, and 25% to Generation Scotland.

3.37 GS had a sustained programme of public engagement which explored public reaction to a wide range of issues relevant to the use of genetics in healthcare. Focus groups indicated that they would prefer an independent public body to govern the database; different review committees (scientific and ethical) were suggested in order to maintain trust. Feedback from specialists also stressed the need for 'independent management and governance and for openness and accountability' (Haddow et al 2008: 146).

3.38 GS offers a helpful example of how a BS model might be constructed. It offered explicit expressions on how benefits would be shared (and in order to manage expectations, clarification about the limitations of feedback that participants would receive about their health). It had two oversight bodies to provide for independence and to monitor accesses to, and uses of, participant information. However, there was no lay-representation on either the Scientific Committee or the Advisory Board and, despite engaging in the consultation process, GS has been criticised for not involving donors or publics in how the resource functions.

Newfoundland and Labrador Model (Pullman and Latus: 2003)

3.39 This model was developed in response to increased interest in carrying out research on the inhabitants of the Canadian province of Newfoundland and Labrador. The province was particularly attractive for genetic researchers given the relative homogeneity of the population. A benefit-sharing protocol was developed; this assesses individual genetic studies on a case by case basis.

3.40 The model had two oversight bodies:

• a Standing Committee on Human Genetic Research (SCHGR) that would operate alongside but at arm's-length to;
• the Provincial Health Research Ethics Board (PHREB), in order to ensure appropriate benefit-sharing arrangements were made.

3.41 A benefit-sharing proposal with supporting rationale must be submitted to the SCHGR which would either approve this or make necessary changes. In order to preserve independence, the PHREB would make its decision independently, however approval would necessitate a satisfactory report from the SCHGR.

3.42 The establishment of a PHREB helps to maintain local control over access to health records and to ensure its own research infrastructure is developed and expanded. However, as with Generation Scotland, the Canadian model has been criticised for not providing the actual donors with control over how resources which they have helped to create are used. Winckoff (2008:11) argues that by excluding the donors from the decision-making process, 'expert decision-making becomes less legitimate'.

UK Biobank

3.43 UK Biobank is a major national health resource and a registered charity. It recruited 500,000 people aged between 40-69 years from across the UK. Donors have undergone measures, provided blood, urine and saliva samples for future analysis, provided detailed information about themselves and agreed to have their health followed. Commercial companies are allowed to profit from their research provided the proposal that they submit falls within UK Biobank purposes and is compliant with its scientific and ethical requirements.

3.44 The Ethics and Governance Framework (EGF) sets up standards for UK Biobank to adhere to, and to ensure that samples and data from the biobank are only used for research that is scientifically and ethically approved (UK Biobank). The UK Biobank Board of Directors have final decision-making authority on access and use of samples (or these may be delegated to access committees). An independent Ethics and Governance Council (EGC) is responsible for ensuring that UK Biobank operates in accordance with the EGF and that it acts in accordance with the participant and public interests. Selection of members of the EGC is in accordance with the Nolan Principles on Public Life (1995).

3.45 UK Biobank has been commended for the extensive public engagement programme that it has conducted; however, it has also been criticised as being biased, ignoring public concerns, and for having excluded important issues such as health strategies and the science behind the project (Goddard 2004, Wallace 2005, Winickoff 2008).

Iceland Health Sector database and deCODE

3.46 Iceland authorised deCODE Genetics, a private for-profit company, to access publicly available health records of the entire Icelandic population, including genetic and genealogical data (Greely 2000). DeCODE created a health database and was granted an exclusive license which gave it the right to use the database for 12 years for a license fee. The sub-license agreement provided that all Icelanders could access any products that resulted from the research (free of charge for the patent term). Additionally, deCODE agreed to pay 6% of profits annually to the Ministry of Health (Arnason 2007).

3.47 DeCODE was met with significant controversy - this may also be to do with the involvement of a private company and the presumed consent and privacy problems that were associated with the organisation. In terms of benefit-sharing, it has been argued that the Icelandic model exploited the Icelandic population and that they were not fully aware of the importance or extent of what was being given away, and the fact that the directors of deCODE were responsible to the stakeholders rather than the population at large was ethically troublesome (McInnis 1999). Further, the offer of free drugs to the Icelandic population was perceived, by some, as a harm (to health) rather than a benefit in that new drugs posed more risk (than older well-established drugs) due to the lack of knowledge on side effects (Arnason 2007).

Estonian Genome Project (EGP)

3.48 This is a public-private partnership developed by scientists under the not-for-profit Estonian Genome Foundation. The Human Genome Regulation Act Estonia 2001 provides that the database can only be used for scientific research, medical treatment, public health research and statistical purposes. In order to facilitate the delivery of pharmaceutical and health related products, a for-profit US entity, EGeen, was granted commercial access to all data emerging from the EGP; however, the Estonian Genome Foundation holds a financial stake in the company. The EGP also created an advisory ethics committee.

3.49 Unlike some other models, under the EGP, all information generated from the initiative is fed back to practitioners 'in the hope of providing better healthcare (Knoppers 2003: 100). However, public-private endeavours can also be problematic, giving rise to different expectations around the outcomes (and thus benefits) of a project. Some (private) funders may be more concerned with immediate (financial) benefit rather than long-term population benefits (Metspalu 2004).

Convention on Biological Diversity (CBD)

3.50 This is a regulatory instrument rather than a benefit-sharing model, but it has formed the basis for many benefit-sharing models and it is worthwhile noting the principles that underpin it. The CBD stresses that stakeholder engagement is essential in order to ensure that access and benefit-sharing arrangements are adequately developed and implemented. It demands that member states set up national authorities that are responsible for granting access to genetic resources. Responsibilities of such authorities can include: negotiating terms for access e.g. requiring prior informed consent; monitoring benefit-sharing agreements and their enforcement; and providing mechanisms for ensuring effective stakeholder participation.

3.51 The principles in the CBD (1992) (and related Bonn Guidelines) are flexible enough to be extended to the human genetic setting (Sheremeta and Knoppers 2007).

Theoretical models of benefit-sharing

3.52 This part of the chapter looks at some conceptual approaches to benefit-sharing that have often been proposed in an attempt to tend to the failures and limitations of previous benefit-sharing models which have been implemented. One of the key challenges with considering theoretical models is that they tend to be abstract and aspirational in nature. This makes it difficult to identify in sufficient detail how such ideas can be translated from theory into practice.

3.53 Despite the challenges that come with considering conceptual models, there is still value in discussing proposed ideas for benefit-sharing models. In particular, models propose solutions for problems previously encountered in practice. Rather than look at specific models, this section considers the common features which have emerged within the literature on theoretical models of benefit-sharing.

Collaborative arrangements

3.54 Features - Collaborative arrangements attempt to provide positive social, economic and environmental outcomes for local communities by engaging with these communities effectively. There arrangements can include contract-based models and community-based partnerships (Lewis, Freeman and Borreill 2008). Partnerships and collaborative agreements are set up between many stakeholders, including participants, those responsible for managing the resource, and, potentially, governments, research agencies, non-governmental organisations and other relevant groups.

3.55 Example - The World Bank Agricultural and Rural Development Department (Barr et al 2007: 33) carried out extensive research in order to understand how forest partnerships could be forged between local communities, forestry industries and other key actors. The research suggested that the most important factors to be included in any collaborative arrangements are:

• legal validity
• full bargaining (identifying each issue and interests in these issues)
• mutual respect
• common expectations about the project and what it means to keep agreements
• self-determination - each individual should feel like they are making a choice
• trust
• practicality
• verifiability
• communication and
• recognition of, and sensitivity to, past history

3.56 Amongst the factors outlined above, four factors were almost universally important according to the study: trust, practicality, mutual respect and communication (ibid. 33).

Contract-based models

3.57 Features - Contract-based models include legally valid contracts (expressed in plain English). The contracts include points on ensuring that the terms of the BS are kept. In particular, contracts can include information on how disputes will be handled, details on practicalities, communication, verification and incentives, and consideration of common issues that have previously led to disagreements.

3.58 Comments - Entering a written document such as a contract can have the effect of emphasising to all parties concerned that they are making a 'true commitment' to upholding their responsibilities (Barr et al 2006: 23) This type of formal document can also allow all stakeholders to consider the different risks involved and the roles and responsibilities of each of the stakeholders. A contract can also serve the purpose of informing potential outside investors/other parties of the agreement.

3.59 One risk with contract-based models is that they can be open to abuse and 'elite capture' (ibid.: 40), that is, where resources which are meant to benefit the larger population are actually only used by a few individuals or groups in power. An example of elite capture might be where a contract does not include any provisions for a company which profits financially from using the resources to contribute a percentage of its profit back into the resource.

Community-based partnerships

3.60 Features - Community-based partnerships serve to make stakeholder engagement part of the 'core business strategy' (ibid.: 7). These would normally be in the form of a legally valid (but not overly complicated) agreement; again, it would ideal if this was set out in plain English. The agreement would include appropriate and effective community consultation in order to identify community needs and build trust. A community-based partnership would also serve to increase transparency about how resources are managed, setting measurable goals and reporting on progress. A workable governance structure that would allow further development to respond to unexpected trends/event can be incorporated.

3.61 Comments - Community-based partnerships can offer the opportunity for issues to be debated in a fair and open manner. Such partnerships openly acknowledge that there is room for disagreement and make sure that fair and practical governance structures are in place. Expectations can be managed by defining roles and responsibilities from the outset.

Participatory Governance Shareholder Model

3.62 Features - This is a theoretical model for benefit-sharing (based on a corporate shareholder model) which has been advanced by Winickoff, who argues that community consultation alone is not enough. He argues that legitimate group representation is a key component to benefit-sharing, and that this can be achieved through partnership governance. Partnership governance views donors as shareholders (rather than stakeholders) and implements 'control rights at the level of the research participation collective' (Winickoff 2008: 12).

3.63 Winickoff's model would grant a participant group a legal stake in the decision-making process and a formal governance role. Using UK Biobank as an example, he proposes that when providing consent, individuals can volunteer to participate in a donor association. This association would then elect leaders/representatives to subsequently sit on the UK Biobank Board of Directors and on the Ethics and Governance Committee.

3.64 Comments - As we have noted above, many models have been criticised for excluding participants from making decisions about how resources which they have helped to create are used. Giving participants a formal role in the governance structure can offer an element of control which, it has been argued, is important for participants (Winickoff 2008; Haddow et al 2008).

3.65 Although on its face, a shareholder model appears to offer control to participants, it faces significant operational challenges, and it is questionable whether the idea of having shareholders is appropriate when data resources are managed by a registered charity or a not-for-profit company. For example, Hunter argues that the idea of shareholders goes against the notion of a company like UK Biobank and asks whether donor representation would best serve the aims and objectives of the company 'which is aimed at benefiting the wider community or society' (Hunter 2011: 232).

Interim Summary

3.66 Given the scarcity of research which has been carried out on benefit-sharing in the context of sharing personal information, and the relative novelty of secondary data usage for research, no perfect model has been advanced to date and certainly none for the specific purposes of sharing electronic health records for research. However, the literature does offer us some potential ideas for key elements that can be included in a benefit-sharing model.

3.67 Most of the models included two independent bodies within their governance framework: (1) one body with ultimate authority for approving access to resources, and (2) one (normally independent) body which also scrutinised applications (such as ethics committees) but which appeared independent and geared towards scrutinizing applications on behalf of the public (interest), i.e. to ensure appropriate benefit-sharing mechanisms were in place.

3.68 The independent scrutiny bodies were also employed with an oversight and monitoring function, to ensure that benefit-sharing measures were enforced. In a study around UK Biobank, survey participants showed strong support in favour of the Ethics and Governance Council having an advisory role to UK Biobank on matters around access to the resources (Webster et al 2008).

3.69 Some models have been praised in the literature for having included an element of public consultation or dialogue in their approaches (often in line with CBD emphasis on stakeholder engagement), but this has often stopped short of actual inclusion of lay members/donor representatives in any of the decision-making/oversight bodies.

3.70 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. Such arrangements may, however, raise important practical considerations. For example, how many representatives should be included? The risk of having too many representatives may arise. Some representatives might have their own vested interests (especially when representatives volunteer themselves rather than being voted for by all participants). There is also the problem of determining where representatives would sit and the powers they might have (or lack) - would they sit on a board and, if so, would this be purely advisory or one with real weight in decision-making?

3.71 It appears common practice to have a formal document in place. These documents set out standards and procedures adopted by organisations or managers of resources. These can be in the form of framework documents, legal contracts or access policies and protocols. These documents are often explicit in communicating which benefits will not be shared with participants, and procedures for third party access (and the distribution of benefits resulting from such research); however, some organisations have been accused of lacking openness and transparency in these matters.

Discussion

3.72 This section outlines key findings from the literature review through discussion of key considerations which should be factored in to assessing or selecting an appropriate benefit-sharing model. In particular this section highlights issues which were identified after applying the template (discussed at 3.5 above) to the concrete and theoretical benefit-sharing models discussed in the previous section.

3.73 As we have already noted, the literature included in the review discusses benefit-sharing in different contexts but not in the specific context of sharing electronic records. For this reason we must extrapolate findings and apply these to the data-sharing setting where relevant.

Working definition of 'benefits'

3.74 The literature shows that 'benefits' can be construed widely rather than in a restricted sense. Benefit can be imaginative and, in any benefit-sharing model, it is important to ask what potential benefits are being offered and excluded. It is also crucial to understand who is providing (or withholding) benefits, and to whom these benefits are provided (or withheld from). It is also important to clarify how benefits would be shared, and the circumstances for benefit-sharing to take place.

3.75 Key international provisions on benefit-sharing tend to define benefit in a broad and inclusive sense. For example, the UN refers to the right of everyone to enjoy the benefits of scientific progress (The United Nations International Convention on Economic, Social and Cultural Rights 1976). The HUGO Statement on Benefit-Sharing (2000) differentiates between benefit and a financial profit. In situations where benefit is restricted to a percentage of profit, how BS may be conceptualised is limited, particularly given that only a small proportion of research studies will lead to significant commercialization (Sheremata and Knoppers 2007: 173).

3.76 As Sheremata and Knoppers note (2007: 161), 'What constitutes a benefit for a particular individual or community will depend on the circumstance, the needs, the values, and the cultural priorities and expectations of that individual or that community'.

3.77 Due to the fact that benefits can be defined so broadly, diverse examples of benefit (not restricted to monetary benefit) are offered and there is evidence in the literature that publics do not necessarily expect immediate benefit. For example, in Generation Scotland consultations, participants considered benefits to include the scope for earlier diagnosis, better quality of life, cures for cancer and mental health, and more research in general (Haddow et al 2008).

3.78 How benefits are defined is important, i.e. whether these are agreed before research commences without room for negotiation, or are linked to the outcome of the research. It has been argued that merely including participants in consultations could actually be a benefit in itself (HUGO 2000).The Convention on Biological Diversity (1992) states that informed consent processes must include: information about benefit-sharing arrangements in place, about the agreement in place between research partners, and have specific provisions which address benefit-sharing and information about benefit-sharing arrangements in place including the details about agreements which have been made between different research partners.

Examples of benefit

3.79 Kadri (2007: 13) notes that often it is not about 'handing out benefits' but, rather, enabling access to products and technologies that have resulted from the research. Some population-scale database organisations specifically stress that they will not provide feedback or direct benefit to donors/participants, emphasising that the fruits of research will be directed towards future generations (e.g. UK Biobank). A survey conducted in Australia found that participants strongly agreed that treatments resulting from research should be accessible to donors (Nicol and Critchley 2009).

3.80 In some cases, however, feedback/health benefits are offered. For example, Generation Scotland offered (limited) feedback regarding health statistics to individuals and with permission, GPs. In Iceland, deCode offered access to health products resulting from research for the duration of the patent, however, for some (including a previous owner of the organisation), provision of new drugs was a harm to the health of patients, in that these carried more risks due to the lack of knowledge of adverse effects (Arnason 2007).

3.81 Where monetary benefits are offered, these might include 'payment of access fees, royalties, license fees, research funding, joint ventures of intellectual property rights, sharing of information, research collaboration, contribution in education, technology transfer, capacity building, social recognition and joint ownership of intellectual property rights (Sheremeta and Knoppers 2007: 161).

3.82 Benefits can also be identified as research facilities and jobs (Pullman and Latus 2003), as well as research cooperation and training of local researchers, i.e. capacity building.

3.83 The HUGO Ethics Committee also includes improved environment and increased human respect within benefit and suggests that, at the very least, participants should receive a 'thank you' and information on the overall outcome of the project (presented in a way that is straight-forward to understand). 'A thank-you does not trivialize benefit-sharing. It is a sign of respect for persons and their basic intelligence and altruism' (HUGO 2000).

3.84 The notion of solidarity was also encountered in the literature. Solidarity implies a realistic approach about benefits, acceptance that trade-offs will be made and that endeavours such as those involving data sharing/research can be packaged in a manner that speak to collective overall gain. This solidarity-based approach appears to be a more European construct and one which would appeal in the Scottish context.

3.85 Individual, community, national or international levels of benefit-sharing exist. The following table developed by Kadri (2007: 17) outlines key (non-exhaustive) benefits at different levels.

Table 1: Types of Benefits (Kadri 2007)

	Health benefits	Commercial/economic benefits	Scientific benefits
Individual level	Designer drugs and other individual aspects of "personalised medicine"	Profits to the investors	Non-instrumental knowledge: development of science and gaining of new information as a value in itself, regardless of whether it is useful to humans
Communal level		Non-medical benefits to communities; capability-building
National, state level	Efficient health care services, policy planning, etc.	Development of biotech and related sectors, new jobs etc; capability-building
Global level	Eradication of diseases, etc.

Public/private nuances

3.86 One important consideration is whether attitudes to benefit-sharing differ depending on whether there is involvement from the public or private sector. One survey conducted in Australia on biobank participants revealed that participants thought that the biobanks should receive payment from those accessing the resources, and especially when commercial companies were accessing the resources (Nicol and Critchley 2009).

3.87 On the other hand, there is a risk of overstating any difference between public and private sector research, for example, research that might begin in the public sector may eventually lead to products that are sold commercially in the private sector (Pullman and Latus: 2003). For this reason, some argue that communities should be wary of signing away economic benefits simply because a research initiative starts out in the public sector.

3.88 Some have stressed that as governmental and non-profit organisations also partially sponsor some research, these are less likely to be motivated by profit-potential and 'attention should be paid to the differing capabilities' of organisations to provide benefits (Kadri 2007: 8).

3.89 Sheremeta and Knoppers (2007: 167) have suggested that some biobank-related concerns may be specific to the private realm, in particular there may be concerns 'about the ability of commercial biobanks to adequately protect biobank participants and to return any meaningful information'.

3.90 In the case of deCODE, it could be argued that the private nature of the endeavour was, to some extent, harmful to research (beyond that carried out by the organisation): 'deCODE has been given exclusive rights to the medical information contained in the database. No one can use that information for scientific research without permission from deCODE, and it is explicitly stated that a permission will be given only if the proposed research is not perceived to conflict with deCODE's business interests.' (Arnason 2007: 14).

Pressure points of concern

False expectations/promises

3.91 Expectations on the benefits of research can be unrealistic, and, similarly, perceived risks associated with research can also be exaggerated, particularly in the context of human genetics (Kadri 2005). For example, promises of access to treatments and products arising through database use during the patent period can be empty given that it is not possible to foresee how many, if any, drugs will result directly from the DNA data used (Greely 2000).

3.92 In addition to the difficulty in predicting what types of medicines and therapies might arise out of research, it is also important to note the inherent difficulty in predicting what commercial benefit might accrue from studies. Things are further complicated when forced to consider whether or not these would be long or medium term benefits and, if so, how they might be shared.

3.93 Despite the inclusion of explicit statements in access policies stating that feedback will not be provided, there may still be an expectation from participants that they will receive some form of feedback. There may also be an expectation that if the individual is identified as being at risk, the researchers will contact them. On the other hand, not all individuals see feedback as a positive benefit and it may give rise to unnecessary anxiety (Melas et al 2010).

3.94 It is worthwhile considering whether unrealistic expectations of benefit and risk might also be associated with research using large-scale databases which give rise to even more unpredictability in terms of potential research outcomes. It is crucial that any benefit-sharing agreements offer explicit and clear information around benefit-sharing so as not to raise unrealistic or false expectations.

Profit and commercialisation

3.95 Concerns around exploitation have been made where for-profit companies stand to make significant commercial profits as a result of accessing resources. This is particularly concerning where commercial sectors do not offer adequate compensation for using the resources. There is also a fear that commercialisation will make research outcomes inaccessible, especially where these are patented and the wider scientific community is excluded from the knowledge which was generated (Chadwick and Berg 2001, Kadri 2007).

3.96 Whilst the potential negative consequences of commercialisation should be taken into account, it is also important to understand that different studies will have very different potentials for generating economic benefit; many studies do not lead to discoveries or products that can be patented and marketed. Further, many projects are expensive and time-consuming to carry out, especially epidemiological studies - for which large-scale databases could be used (Pullman and Latus 2003). Public consultation on benefit-sharing has indicated that the public feels that where commercial organisations are involved in using resources, the biobanks should also receive a share (Nicol and Critchley 2009). Others also raise the point that commercial/private interests are to some extent a 'necessary evil' in the research setting (Haddow et al 2007: 276).

Ownership and access

3.97 Ownership has been raised as an important issue in some sectors (particularly where the resources involve community land) but is more problematic when discussing electronic records and particularly where this is stored in large-scale databases. Some have questioned whether a property-ownership model could be adopted in the benefit-sharing setting (Kadri 2007); however, it is difficult to see how this would be implemented in practice, particularly where databases would contain population-level datasets. In the context of biobanks, one approach has been to consider the managers of biobanks as custodians or stewards rather than owners of the data. In Scotland, ISD NHS NSS acts as custodian for a vast range of datasets.

3.98 With regard to accessing information, various factors must be considered. For example, whether or not users should be charged for access. One study found that charging fees for access was considered acceptable by participants, so long as the practical considerations of how profit would be shared had been discussed (Webster et al 2008).

3.99 Another important consideration around access is based on who can access the information. In the Generation Scotland consultation, participants suggested that access to information they regarded as private should only be granted to medical personnel, academics or research scientists who had obligations of confidentiality. In the case of UK Biobank, some participants raised concerns over international researcher access to the resources (Webster et al 2008). Further, access by insurance companies was flagged as inappropriate and potentially leading to non-participation. Concerns were also raised over access by employers and banks. Involvement of pharmaceutical companies was also generally perceived negatively, for some it was considered a necessary evil, as were commercial/private interests.

3.100 This suggests that there is not a one-size-fits-all approach to how resources should be accessed and that there are sensitivities around certain groups accessing information, particularly depending upon whether they might stand to profit significantly. It also suggests that the public might be more comfortable when their information is used by individuals/groups that are bound by obligations of confidentiality.

Contributors/beneficiaries

3.101 Benefit-sharing pertains to the distribution of benefits but also, often implicitly, to the burdens arising from research and beneficiaries are not limited to those who partake in studies, but also 'those from the same community or the wider population where the research is undertaken' (Kadri 2007: 3). In the Universal Declaration on Bioethics and Human Rights (2005), UNESCO stresses that benefits derived should be shared with 'society as a whole and the international community' (Article 15).

3.102 It is important to acknowledge that how discussions around BS (and justifying BS) take place, and who is involved in these discussions will exclude or include specific communities. For example, if we consider benefit-sharing as compensation for voluntary risk, then it is only fair that benefits are restricted to those who have participated. If, however, we consider our information common property, then the benefits are to be extended across humanity.

Involvement in public dialogue

3.103 Whilst the joint literature review on citizen empowerment will look more closely at methods to engage the public, here we consider approaches which have been taken specifically in the context of constructing benefit-sharing models.

3.104 Involvement in public dialogue did take place in actual benefit-sharing models (e.g. Generation Scotland and UK Biobank) however this dialogue was/is not always sustained throughout the course of the data use. This is problematic because the very nature of large scale research (and research more generally) dictates that the uses of data can change over time, as can the involvement of different (public and private) stakeholders. Public consultation and dialogue is also criticised as being overly tokenistic.

3.105 Upstream consultation, i.e. where the public is consulted before benefit-sharing models are in place has also gained support in the literature (Haddow et al 2008), and one study indicated that participants were highly in favour of knowing what benefit-sharing arrangements were in place before deciding to participate in a project (Nicol and Critchley 2006). It is important to note, however, the difference between approving to uses on behalf of the wider community and providing consent for use of information where this is required (the community cannot consent on behalf of individuals).

3.106 An interesting nuance noted by Hunter is that HUGO places the onus on researchers to consult with the public, but other approaches have traditionally placed this responsibility with those managing biobanks as opposed to those who will be using the resource, because managers of biobanks are perceived to be in a relationship of trust with the participants (Hunter 2011). It has also been argued that engagement of the scientific community in planning and conducting genetic research has been inadequate (Berg and Chadwick 2001). It is worthwhile considering whether including researchers in the consultation process could serve to help engender a relationship of trust with the public.

Solidarity

3.107 In addition to notions of justice, fairness and equity, some authors have considered whether the ethical principle of solidarity should be appealed to when discussing benefits (e.g. Kadri 2007, Prainsack and Buyx 2011). This approach would imply that benefits arising from research would be shared based on principles of social justice. The notion of solidarity is increasingly being discussed within bioethics literature, it is 'linked to ideas about how societies functions, and about how and where the boundary between individual, familial, communal and societal spheres of responsibilities should be drawn' (Prainsack and Buyx 2011: 1). More generally, solidarity emphasises the relationships between members of a community and responsibilities that might arise out of that relationship. A solidarity-based approach in the context of data sharing and research might imply that the overall aims of research should be to assist others rather than to prioritise financial gain.

3.108 The HUGO Statement on Benefit-sharing (2000) includes references to the principles of solidarity; HUGO insists that affluent nations and commercial entities 'foster health for all humanity'. This broadens the scope of benefit to humanity rather than restricting it in a group, and it does not necessitate direct BS. For example, Kadri (2007: 14) notes that 'fostering health' could include increasing the research focus on diseases that are prevalent in poorer countries, so that the whole of humanity can benefit from the technology and know-how'. This might be as an alternative to increasing the complexity of informed consent and the amount of information that is given to participants (Chadwick and Berg 2001).

3.109 One problem with basing benefit-sharing on the solidarity principle is that tensions will arise between the principle and taking a compensatory approach to benefit-sharing. This would be particularly significant in the context of data-linkage as 'the larger the number of participants in research, the more significant the concept of solidarity becomes in benefit-sharing' (Kadri 2007: 13). Further, benefit-sharing 'in the form of personal gains may undermine the solidarity notion and, thus, decrease enrolment of those who participate based on altruistic motives (Melas et al 2010: 94). Broadly used, altruism refers to 'acts and practices of people to benefit others without: (a) being obliged to do so due to a legal norm; and (b) without receiving, or even expecting, anything in return' (Prainsack and Buyx 2011: xiii).

Summary and Implications

3.110 It is clear from the literature review that there is a lack of evidence of benefit-sharing models that move beyond the rhetoric (in general, and more specifically in relation to data sharing) and of actually delivering effective models and solutions in practice. To attempt to offer a prescriptive model of benefit-sharing would be both premature and, arguably, lack legitimacy without prior public consultation.